TRANSDUCERS FOR INSTRUMENTATION

Prof. Ankur Gupta

Department of Electrical and Electronics Engineering
IIT Delhi
 INDEX

[Link] TOPICS [Link]

Week 1
1 LECTURE 1 3
2 LECTURE 2 14
3 LECTURE 3 25
Week 2
4 LECTURE 4 33
5 LECTURE 5 45
6 LECTURE 6 55
Week 3
7 LECTURE 7 67
8 LECTURE 8 78
9 LECTURE 9 86
Week 4
10 LECTURE 10 95
11 LECTURE 11 106
12 LECTURE 12 116
Week 5
13 LECTURE 13 123
14 LECTURE 14 132
15 LECTURE 15 140
Week 6
16 LECTURE 16 148
17 LECTURE 17 157
18 LECTURE 18 166
Week 7

1
19 LECTURE 19 174

20 LECTURE 20 182
21 LECTURE 21 189
Week 8
22 LECTURE 22 196
23 LECTURE 23 202
24 LECTURE 24 210
Week 9
25 LECTURE 25 216
26 LECTURE 26 223
27 LECTURE 27 231

Week 10
28 LECTURE 28 238
29 LECTURE 29 245
30 LECTURE 30 251
Week 11
31 LECTURE 31 259
32 LECTURE 32 266
33 LECTURE 33 274
Week 12
34 LECTURE 34 282
35 LECTURE 35 290
36 LECTURE 36 299

2
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 01
Introduction to the Course

Hello, welcome to the course Transducers and Instrumentation. Today's outline is that
we will discuss the introduction to transducers, Transduction principles, and the
classification of transducers. So, the transducer is a device that converts one form of
energy into a corresponding signal. So, the incoming energy can be of any form. Let us
say it is a mechanical, a thermal, it is a electromagnetic, optical, and chemical. The input
form of energy can be anything and Transducer generates the equivalent output signal.
This transducer take the form of a sensor or an actuator. So, basically transducer
depending upon the application it takes the form of sensors or an actuator.

So, mostly we deal with the sensors because the sensor is a device That converts a
physical parameter to an electrical output. So, the input form of energy can be anything, it
can be mechanical, optical, chemical it can be anything, but the output what it generates
is a electrical signal that is called a sensor. And why we are more interested in the sensor
is because the electrical signals are easy to process. We have electronics, we have
computers, we have ah so many microcontrollers etc. Where we can easily process these
electrical signals and generate the meaningful data. So, when we say transducers most of
the time we are dealing with the sensors Means the input can be anything, it is an
environmental parameter, but the output is an electrical signal. The sensor is thermometer
for example, we have a thermometer we are talking about here the digital thermometer
which measures the the temperature of environment and gives the reading in the form of
electrical output. Which can be a digital display it gives the temperature on the display
like 100 degree Fahrenheit, 102 degree Fahrenheit depending upon the upon the
parameter. So, the sensor is a device which converts physical parameter into a electrical
signal.

It is a device that detects or measures a signal or stimuli from the outside environment
and acquires information from the real world. What we live in the real world and most of
the parameters are analogue in nature and this sensor sense those parameters and generate
an equivalent electrical output. So, this is sensor the next part is the actuator. So, the
actuator is kind of dual of sensor. Actuator converts an electrical signal into a physical
output. So, the actuator accepts a electrical signal generated by some electronic circuit or
electrical circuit and it generates a physical output in the form of rotation, in the form of
pressure or in the form of temperature it can be anything. For example, a heater. So,
heater is what it accepts electrical signal when we power it up from our socket the

3
electrical signal or electrical power comes in and it generates a thermal output. The
temperature of heater goes up or goes down depending upon the signal. So, heater is an
example of actuator where it converts an electrical signal into a physical output.

This is a device that generates a signal or a stimulus. So, in pictorial view you can see we
have a real world which is analogue in nature and the sensor accepts the these physical
parameters. For example, temperature it can be pressure, it can be a chemical output. So,
the sensor accept these environmental parameters and convert that into a electrical signal
which is further processed by the system an electrical system or an electronic system. The
actuator does a positive of that it accepts these electrical signals and convert that into a
physical parameter. For example, the heater we considered here which converts this
electrical signal into temperature. So, all together this is called transducer and depending
upon the application what we are applying here we call them as a sensor or a actuator and
the combined both of the them combinedly called transducers.

So, in this course, we will mostly talk about the sensors because sensors are important
because it converts all the physical parameters into electrical signal and as we discussed
electrical signals are easy to process using our computer or electronics. So, now we
discuss various transduction principles. So, the sensors we make they use multiple
physical properties of of materials and convert and use those properties to convert out
electrical to convert physical parameters into electrical signals. So, various transduction
principles are based on the input and output for example, the input is mechanical and the
output can be a mechanical or it can be a thermal, it can be electrical, it can be magnetic,
it can be radiant and it can be chemical and and others as well. So, input can be
mechanical and output can be anything or we change the input the input let us say we talk
about the thermal input and we convert that into mechanical or thermal electrical.

So, these sensors use various properties which convert this thermal input to magnetic
output or mechanical output. Similarly, we have electrical input coming in and we can
convert it into mechanical. So, that time it takes the form of actuator where we take the
electrical and convert it into mechanical. So, this row is kind of actuator where the input
is electrical and the output is mechanical or thermal or magnetic or other forms.
However, if the output is electrical and the input is anything else then this is sensor. So,
this row corresponds to sensor where we have output in the form of electrical. So, let us
discuss all these transduction principles in little more detail. We have here an acoustic
sensor which is mechanical to mechanical. So, in this diagram we are converting a
mechanical energy into a corresponding mechanical signal. So, you can see a diaphragm
here which is connected to this pressure bellows and here the pressure port is there. So,
what actually it is happening here we have this diaphragm which is very sensitive to
pressure change. So, any pressure change here let us say the pressure is increasing this
diaphragm will be pushed down which will be detected by this pressure bellows. If this

4
diaphragm goes down this pressure bellows will be squeezed or they will be compressed.

Depending upon that when this pressure port you will get a pressure difference. So, this
is kind of converting mechanical energy into a mechanical energy. This is one example of
transduction principle where we are converting mechanical to mechanical itself. The next
is thermal to mechanical we have we are converting thermal energy into mechanical
energy. So, the example of this is thermal flow meter. What we are doing here this is this
is the actual photo of the of the of the transducer and the working principle is here where
we have a heating element which is shown in red here this is your heating element which
is generating heat and this will be uniformly distributed. This will be initially uniformly
distributed. However, if there is some gas flowing in this region gas or liquid flow in this
region this heat signature will be something like this shown in this yellow curve because
the incoming gas flow is at different temperature it can be colder than this temperature.

So, this because of this heat flow the thermal profile of this heating element will be
similar to this colored waveform as shown here. So, because of this different color profile
we can measure how much gas is flowing into the instrument. If the gas is low the the gas
flow is low the profile will be very uniform and if the gas flow is very high this profile
will be further deviated from from left side. So, this is an example of thermal to
mechanical converter. Next is mechanical to electrical conversion. So, mechanical to
electrical means we have a mechanical energy and we are generating corresponding
electrical signal. So, this is actually a sensor. Here the example is acoustic sensors.
Acoustic sensors we have these pressure change at the input and convert it into
mechanical output into electrical output. So, this is a acoustic sensor which is generated
by some some company. So, this acoustic sensor if this has a kind of a diaphragm at the
input if the pressure changes then there is a Op-Amp circuitry connected to it and that
Op-Amp circuit will convert this pressure change into a corresponding electrical output.
So, this is acoustic sensor this is one commercially available acoustic sensor.

Another example is this delay line sensor. What this delay line sensor does this is made
on a piezoelectric substrate. Piezoelectric substrate we will discuss in more detail in ah
further lectures, but piezoelectric substrate generates corresponding electrical signal from
an incoming mechanical signal. So, this piezoelectric substrate is there and on top of this
substrate we make these in IDT structures or inter digital electrodes IDT's we call it this
is these are like a comb structure where we have port where terminal number 1 and 2 and
we apply signal on on these IDT structure. These IDT structures generate equivalent
mechanical output from this electrical signal which travels into this delay line sensor
from left to right and at the output on the right side these IDT's pick up this signal and
convert back into an electrical signal which is further processed by electronic circuits.

So, this is the basic working principle of delay line sensors where we are converting

5
electrical to mechanical and then mechanical to electrical and based on the material we
put here this IDT structures does sensing. All these sensors all these mechanism we will
discuss later this is just the introduction. Next is thermal to mechanical and the very good
example is our normal thermometer or the mercury thermometer. So, thermal to
mechanical energy conversion is there where this thermometer is exposed to the
temperature and there is a mercury inside the metal mercury is there which takes this
thermal energy from ambient and because of the thermal expansion of of metal because
of the temperature there is a change in the level of mercury. So, if we have a low
temperature the temperature is less you see the level is less and if temperature goes high
this mercury or some other material there is a elongation. So, this is a physical change in
the material which is mechanical in nature and this mechanical output is produced by the
thermal input the temperature. So, this is thermal to mechanical conversion this is one
type of transduction principle.

Similarly, we have thermal to electrical conversion where we can convert this thermal
energy into a electrical output and again this is a sensor because the output is electrical in
nature. The example of thermal to electrical is thermocouples. These thermocouples are
pair of metals where we have metal 1 or we can write type A and we have type B.

So, we have two types of metals the wires made of these two different metals they are
joined together and we make two junction one we put at the reference level or generally
we call cold junction and the other junction is our sensing arm or sensing junction which
is used to measure the temperature. This thermocouple senses the temperature at the
sensing node with reference to the cold junction or the reference junction we call it cold
junction as well as the reference junction. So, this temperature measurement is always
relative to the reference junction and depending upon the temperature difference between
these two junctions there is a electrical potential build up between these two nodes node 1
and node 2 which you can measure using normal multimeter or some other electrical
instruments. So, this is thermocouple is an example of thermal to electrical conversion in
say is a thermocouple sensor and this is the actual photograph which is available in the
open market. Another transduction principle is thermal to optical where we are
converting this thermal energy into a corresponding output or into a corresponding
optical signal.

We have this kind of arrangement where if optical fiber is going through all this liquid
and this liquid is at different temperature. If the light is passing through a optical fiber
and the temperature is changed. So, multiple properties of these optical waves actually
changes, and we will see that in optical sensors that these change in the optical properties
can be detected using your analog interferometer or various other instruments. So, this is
a thermal to optical converter. This is one form of transduction. Another transduction is
electrical to mechanical where we are converting electrical signal into a mechanical

6
signal. So, here the input is electrical, and output is mechanical or it can be anything. So,
here we are talking about the actuator because the actuator takes input as a electrical
signal.

So, one example of this kind of actuator is piezoelectric sensors where we have some
materials which show piezoelectricity. Piezoelectricity is when these special materials are
exposed to some mechanical input. For example, you change the size of these material by
applying pressure. They generate equivalent output voltage and reversibly if you apply
electrical output electrical input to them then there is a mechanical change in the in the
size of these materials. So, for example, we have piezoelectric material here which is
here. For example, one piezoelectric material is SiO2 or we say quartz. When we apply a
electrical signal across this then there is a physical change. This is expanding in this
direction or compressing. So, this physical change is corresponding to how much signal
we apply at the electrical signal we apply across this quartz crystal. This is called
piezoelectric effect and corresponding sensor is piezoelectric sensor and this is the actual
sensor this is how it is seen in the market. Another transduction principle is electrical to
thermal where we are converting electrical energy into thermal energy.

One example we have already seen which is our heater in a normal filament where we
apply electrical power or electrical signal and correspondingly heat is generated. The
other example is this Peltier sensors. These Peltier sensors they are made up made up of
thermo thermocouples we have two different metals. When we apply a electrical signal
between these two terminals this is positive this is negative. When we apply electrical
signal across this depending upon the material we choose in both the metals one side of
the this plate becomes hot and other side of this Peltier effect Peltier kit becomes cold.
So, this is called Peltier effect and this is the Peltier kit we generally use for temperature
measurement or temperature control also because the effect is reciprocal.

So, this Peltier kit converts this electrical signal into a thermal output. The next
transduction principle is electrical to electrical where we have an input signal as a
electrical signal and the output is also generated in the form of electrical signal. Example
is some charge controller device where we are controlling the charge very much used in
solar panels where we the solar panels are generating electrical charge from the from the
sunlight and this electrical charge is again controlled by these devices. These are charge
control devices and example this is the example of electrical-to-electrical transduction.
Example 2 electrical which is again one another important transduction principle because
the output is electrical it means it is a sensor the input is magnetic, but the output is in the
electrical form.

One example is hall sensors. Hall sensor is it can measure the current flow in current
flow in the circuit as well as it can measure the magnetic field applied across the sensor.
So, the basic structure of this hall sensor is we have a conductive medium where the

7
current can pass we apply a potential difference across it from a DC supply and
simultaneously we apply this magnetic field. This is the magnetic field which is the input
to this sensor. Depending upon this magnetic field these charge carriers will feel a
Lorentz force they will the positive side will go for example, this side and the magnetic
and the negative side will go to the other side. If they accumulate positive on one side and
negative on the other side essentially there is a potential build up across it and this
potential difference is depending upon the applied magnetic field.

If we have high magnetic field applied then the hall voltage will be higher. So, this is
example of magnetic to electrical conversion. Another example is chemical to electrical
where we are converting chemical measurement to an electrical output this is again a
sensor. So, here we have this IDT like structure we have two electrodes this is let us say
one this is two this is a comb like structure we have this first comb and this is the second
and this is a IDT structure or call it inter digitated electrodes and these electrodes we dip
into the liquid or the chemical and depending upon the conductivity of chemical the
resistance between this electrode and this electrode changes because the liquid is filled in
between. So, depending upon the conductivity of that solution electrolyte solution that
conductivity we measure or we can measure the resistance which is proportional to the
conductivity of the electrolyte.

So, this is chemical to electrical sensor. So, we have discussed now multiple types of
transduction principles where we apply a input and the output is generated in the form of
electrical or other measurements, but we all considered where the input is only one, but
some sensors for example, the hall sensors we just discussed it can have more than one
one inputs. So, for example, in the hall sensor we have electrical and magnetic. So, this
hall sensor can have two inputs electrical and magnetic. So, it is not possible to kind of
classify all the sensors into a one definition. So, in the case of more than one input
sometime it is the application aspects or the measurement type is given more important
than input type because the input we have more than one.

So, for example, we have mechanical energy as the input and measurement can be
length, area, volume, pressure, force, acceleration, torque these are all mechanical in
nature. Another form of energy can be thermal where we have temperature, heat flow,
entropy, state of matter all these measurements are there, but the essential the energy
input is thermal in nature. We have electrical energy where we can measure charge,
current, voltage, resistance and there are so many parameters we can measure, but the
form of energy is electrical. We can have magnetic energy for example, the magnetic
field intensity, flux density, permeability magnetic movements and so forth. Another
form of energy is radiant where we can have radiation and the measurement can be
intensity, phase, refractive index, reflectance, transmittance, absorbance, wavelength,
polarization and so on.

8
Similarly, we can have even the chemical energy also as an input and the measurement
can be the concentration, the composition of liquid, oxidation, reduction potential,
reaction rate, pH is also chemical in nature and others. So, these are various
measurements which we can measure, but the type of energy are different. Now we come
to the classification of these transducers. So, it is difficult to classify all the sensors under
one criteria because there are so many types of sensors some are mechanical input, some
are electrical input, some are optical input. So, it is very difficult to classify all the
sensors under a one criteria and that is why there are different criteria's are adopted
worldwide for the classification of sensor.

One classification is based on the transduction principle which we will discuss there are
multiple types of transduction principles. Second is the primary input quantity in means
the measure and what it measures. The third is material and the technology used in the
sensor. The fourth is applications based what application actually we are solving using
this sensor or actuator and the property type the what is the property we are using in the
sensor. So, these are basically few types of classification though all the sensors cannot be
classified in these five as well we can have multiple classification criteria's as well, but
we will we are going to discuss these five.

So, the first one is the transduction principle based on the type of transduction, type of
conversion of energy from one type to another type depending upon the transduction
principle we can have classification of these sensors. Transduction principles based on
these physical, chemical or biological effects. So, transduction principle can be a physical
type where we have thermo electric, photo electric, photo magnetic, magneto electric,
electromagnetic, thermo elastic, electrostatic, thermo magnetic, thermo optic, photo
electric and others. So, all these effects these are physical physical properties of
materials. The other transduction principle is chemical in nature where we can have a
chemical transformation the chemical used in the sensor it changes the property.

So, physical transformation, electrochemical process, spectroscopy and others these are
all chemical type of transduction. We can have biological transduction as well where the
compound we use in the sensor it changes its property when it when it binds with some
biological components. So, here biochemical transformation is there physical
transformation effect on test organs spectroscopy and others these are all biological in
nature. So, these are transduction based principle we can classify the sensors. Other
classification is the measure end base measure end is the input parameter which is
coming from the real world we are measuring the temperature then the temperature is the
measure end and we are converting into electrical signal.

So, classification can be based on the measure end measure end type. For example, we
have a caustic sensor where the measure end can be a wave amplitude we can rather
measure the phase, the polarization, the spectrum or the wave velocity and other

9
parameters. We can use any of the this measure end and the sensors will be the caustic
sensors. Biological type where the stimuli is biological for example, the biomass type
concentration state and others these are the all the measure ends we can use. If the stimuli
is chemical then we can have multiple components for example, identities concentration
states and others.

Stimuli can be electric where we can have multiple type of measure ends one is charge,
the current flowing, the potential build up, voltage, electric field and in these we can have
amplitude, phase, polarization, spectrum and conductivity, permittivity and various other
parameters. If the stimuli is magnetic we have magnetic field again we can measure the
amplitude, phase, polarization, spectrum, magnetic flux, permeability and other measure
ends. If the stimuli is optical we can have wave amplitude, phase, polarization, spectrum,
wave velocity, refractive index, emissivity, reflectivity, absorption others. We can have
multiple types of measure end. If the stimuli is mechanical we can again have multiple
measure ends position which can be linear as well as angular, acceleration, force, stress,
pressure, strain, mass, density, movement, torque, speed, rate of mass, transport, shape,
roughness these are all mechanical parameters and these can be used as a measure end.

Another stimuli is radiation where the radiation type can be a measure end what kind of
radiation is there is infrared, it is ultraviolet, it is in the visible region. So, that can be a
measure end type energy, intensity and other measure end type. If the stimuli is thermal
we can have thermal flux, we can have specific heat, thermal conductivity and others. So,
these are all classification based on the measure end type. So, we can have the material
based sensor, material based classification where the sensor fabrication material is used
for the classification.

So, where we can have basically two type one is the inorganic type and the second is the
organic type. Inorganic type sensors are for example, the material used is a conductor just
like a metal or it is a semiconductor for example, silicon based sensors or some biological
substance these are all fall in inorganic type of material, inorganic type of sensors. The
other sensor type is organic where we can have insulators, we can have liquid gas or
plasma also these are also used as a sensor and other materials. So, this is basically the
classification based on the material used for sensor fabrication. Some classification we
can do based on the property type, the property we what we use for these transduction.

So, for example, we have these properties on x axis and the technology used on this y
axis for example, flow we can measure and we can have differential pressure, positional
displacement, vortex, thermal mass, electromagnetic, corollarys, electro ultrasonic,
anemometer, open channel these are all for flowmetry. We can have level where we have
mechanical, magnetic, differential pressure, thermal, displacement, vibrating rod,
magnetostrictive, ultrasonic, radio frequency, capacitance type and microwave radar or

10
nuclear type. Another property is the temperature where we can have filled in systems or
the RTDs, RTDs are we can have filled in systems, we can have RTDs, we can have
thermistors, we can have ICs thermocouples, inductive, inductively coupled or radiation.
Another property is the pressure where we can have elastic, liquid based, manometers,
inductive LVDTs, piezoelectric, electronic fiber, optic, MEMS or vacuum. Proximity and
displacement we have potentiometric inductive LVDTs, the capacitive, magnetic,
photoelectric, magnetostrictive or ultrasonic.

Another property is acceleration where we can have accelerometers or gyroscopes, these

are very much used nowadays as a MEMS device in all the smart electronics. We can
have image for example, we can use CMOS, we can use CCDs or the charge coupled
devices, we can have the acceleration. Another is gas and chemical where we can have
chemical bead, electrochemical, thermal conductance, paramagnetic ionization, infrared
and semiconductor. Another property is biosensors where we have electrochemical, light
addressable potentiometer, LEP we call in short surface plasmon resonators or resonant
mirrors and others as well for example, mass, force, load, humidity, moisture and
viscosity. So, these are the property based classification we where we use the property of
of sensing and we classify all these sensors and transducers based on this property type.

Apart from this we have multiple emerging type of technologies and then we called
emerging sensors where all these because of the advancement of microelectronics area,
we have multiple sensing technologies coming in. And some of the example is given here
where we have these image sensor technology which is CMOS based. So, we have image
sensors and they are fabricated using CMOS process and they use multiple type of
technology actually the signal processing technologies are different. Based on that we can
classify and they are used pretty much in traffic and security surveillance, blind spot
detection as auto sensors for example, the robots in inside the robots we use them, video
conferencing, consumer electronics, biometrics and PC imaging these are these sensors
are very much used. Next we have these motion detectors or the motion detector
technology I they use IR ultrasonic or microwave radars.

These are very much used in defense application for example, the obstruction detection,
autos or the robots or even for example, the the detection of drones that is also part of
radar engine, security detection, toilet surveillance, toilet activation, kiosk, videograms
and simulations, light activation there are multiple applications where these sensors are
widely used. The third is biosensors or the biosensor technology which is electrochemical
in nature. These biosensors are very much used in water testing, food testing for example,
in food testing we do contamination detection using these biosensors typically we can
calculate how if the food is edible or not if these fruits which comes from outside outside
the country we can judge the quality of those fruits and other edible products whether
they are edible or they are not edible using these biosensors. Medical care device is

11
another very big area where we use these biosensors for example, the sugar test kit which
is a biosensor where the drop of blood we put on the strip and it gives the sugar level
inside the blood is widely used for a diabetic patient and this is a biosensor. There are
multiple type of biosensors available in the market, sugar testing is one another one thing
there can be some other sensors for example, the radiation sensor which has which is also
a biosensor.

Medical warfare or the agent detection this is also also a area where biosensors are
widely used. Accelerometer technologies this the acceleration measurement and the
technology we used here is the MEMS devices. MEMS is a short form of micro electro
mechanical system, micro electro mechanical system. So, these are mechanical systems
which are generally big in nature, but we fabricate these mechanical systems using
CMOS process or fabrication process and the size is so small that we start calling them
micro electro mechanical systems. These mechanical systems detects the mechanical
input and convert that into an electrical output.

So, very good example of this is we have a gyroscope in inside the laptops or inside the
mobile phones which detects the orientation of mobile phone. For example, we have
mobile phone in the in the portrait mode then our screen becomes portrait if we rotate the
mobile phone then the screen automatically changes the orientation. So, that is done
using these MEMS devices because there is a hanging pendulum inside which detects in
which direction we are putting our mobile based on that the screen is changed. This is
one application of these MEMS devices. The vehicle dynamic system is also very
important where we have these airbags in our cars and these airbags are connected to a
pendulum inside it is a kind of pendulum it is a mechanical device is a very small inside.

So, we can call it as MEMS device. Whenever there is a accident or there is a collision
this pendulum detects that jerk in the in the system and because of this this change in the
mechanical system these airbags gets open and the passengers are saved. So, one example
is airbag and these are also used in patient monitoring including pacemakers etcetera
pacemakers are again life saving devices where these all emerging sensors are used. So,
these are some classification based on the emerging technologies where these sensors are
are fabricated in mass scale using fabrication technology CMOS based or other
fabrication technologies. So, these are all the classifications. However, it is not possible
to classify all that sensors in the world into these definitions and we can have some other
classification as well which can be active or passive.

So, this another classification can be based upon the power or energy requirement of
these sensors. So, for example, we have a sensor and this does not need any electricity or
external bias or external power to convert the signal into electrical signal then this sensor
is called a passive sensor because it does not need any extra power supply. If a sensor it
needs any external power supply or some external energy to do the transduction to

12
convert the input into the output then this sensor is called the active sensor because this
needs external power supply. So, some sensors they require power supply these are called
active sensors. Active sensors means you have some electronic circuits inside which is
detecting some change in the parameter and generating output.

So, all these electronic circuits they need some external power supply. So, these are all
active sensors this is one classification where we need power supply. The passive sensors
are something like your piezoelectric substance for example, your quartz or SiO2 when
we apply a physical change or we compress it this SiO2 crystal compress it or expand it,
it automatically generates a potential across it. This system does not need any external
power supply we just press the system and it generates an equivalent electrical output and
this concept is reciprocal. So, this kind of system does not need any external power
supply and we call it a passive system or passive sensor.

Another example of active and passive sensor is the mercury thermometer. We have this
old mercury thermometer where this mercury is filled in a glass tube and when we expose
this whole tube to the temperature this thermometer when we let us say connect it to our
body it takes the thermal energy from our body and because of this thermal energy that
the mercury inside it changes its volume it expands when we heat up the things it
expands. So, the mercury filled inside it expands inside that tube and we have multiple
markings on that based on that expansion we can measure how much is the temperature.
So, this thermometer the mercury filled thermometers they are passive in nature because
we need not to have any external battery external power source or any other kind of input
biasing to this thermometer this is passive thermometer. However, the new thermometers
which we have nowadays they are digital in nature they show the number as a
temperature reading. So, these thermometers they have some kind of thermocouples
inside which measures the temperature but this electrical this signal generated by this
thermocouple this is further processed using some electronic circuitry for example, this
kind of electronic circuitry and this electronic circuitry gives its output on LCD or the
screen we have in these thermal sensors or this thermometer.

This electronic circuit it uses external biasing or some external power source where some
voltage is applied here let us say 5 volt typically and this circuit performs only when this
voltage supply is there. So, all these thermometers we have digital thermometers they
have a battery inside which is powering all these electronic circuit and based on that there
is a sensing by these digital thermometer. So, these digital thermometers they are active
sensors. So, we can classify based on these two properties as well where we say passive
sensors where we do not need any power supply and we have an active sensors where we
need an external power source to do the processing. So, we discussed multiple types of
classification of transducers and we discussed the introduction to transducers.

Thank you.

13
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 02
Introduction to Transducers, Sensors, and Actuators.

Hello, welcome to the course Transducers for Instrumentation. Today we are going to
discuss some characteristics of sensors. These characteristics are some parameters for
which we develop the sensors. So, for example, in a particular sensor we have this much
of resolution, this much of linearity, range, what are all the parameters for which a sensor
is need to be designed. So, these are some characteristic of a sensor which we are going to
discuss. So, in normal environment the data comes in and this data is not a pure data it is
full of noise and it keep changing.

So, these sensors what we are going to design these So, these sensors what we are going to
design these are required to be appropriately characterized. So, because the data is noisy
so, we need to take care of this noise and the sensor need to filter out the data from it. So,
the term here is this characterization. So, this characterization is done in terms of certain
parameters and characteristics of sensor.

So, there are certain parameters for which this sensor is need to be characterized. And these
sensors are like measurement systems they have two types of characteristics one is the
static and one is the dynamic. Before discussing these static and dynamic characteristics
we have a term in hand which is characterization. So, what is a characterization? So, the
device characterization is the process of quantifying the electrical behavior and
performance of electronic device. , It may be another device but we are particular about
electronic system.

So, we are discussing electronic devices. So, this characterization is the process of
quantifying how we can quantify the behavior of a device such as transistors, diodes,
sensors etcetera. So, this involves measuring the various electrical parameters such as gain
of an amplifier, stability and various other terms this involves measurement of these
electrical parameters and these frequency response repeated measurements it allows
engineers to understand how the device operates under different operating conditions.
Because the sensor is not going to work throughout its life at a particular fixed voltage or
fixed parameter only. This parameter is going to be changed and because of that the output
of the sensor is going to be changed.

So, we need to characterize this sensor so that we know upfront what is the input parameters
and based on that what output is going to be there. This is also done to know any
performance limitations or issues with the sensor upfront. So, if we have let us say sensor

14
has a limited range so that can be quantified using characterization process. In
characterization process what generally we do we apply a full range of input parameters
and check the output of the sensor. So, this gives us the idea of how the sensor is going to
perform in real world.

So, this is called characterization. So, what we do we characterize the sensors or

transducers characterize the sensors or transducers in terms of certain parameters and these
parameters or these characteristics are of two types one is the static one and one is the
dynamic one. So, as the name says static characteristics we have the characteristic of sensor
which does not evolve or does not change with time. So, these characteristics are like sensor
the sensitivity, the sensitivity is something how much is the change in the output per unit
change in the input so that is called sensitivity. Resolution for example, the smallest amount
of change in the input that can be detected by the sensor so that is called resolution.

Linearity of the sensor for example, how much the sensor output deviates from a standard
straight line output. Drift is one another property. So, there is a drift in the output does not
evolve or does not change with time. So, these characteristics are like sensor the sensitivity,
the sensitivity is something how much is the change in the output per unit change in the
input so that is called sensitivity. Resolution for example, the smallest amount of change
in the input that can be detected by the sensor so that is called resolution.

Linearity of the sensor for example, how much the sensor output deviates from a standard
straight line output. Drift is one another property. So, there is a drift in the output of a
sensor with the input it kind of drifts from a from its regular output by a certain amount so
that is called drifting. And there are special cases in that there are some zero drift full scale
drift there are other types of drift. Range we characterize the sensor for for range as well
range is the upper limit minus the lower limit of the output or the input Repeatability this
is also something does not change with time.

So, if we apply a repeated input at the sensor how much the output deviates from its
previous input. So, we apply let us say a similar input 10 times all 10 times the output will
not be same there will be a certain difference at the output for all the 10 inputs this is called
the repeatability issue. So, the sensors has this repeatability issues. Reproducibility is
another characteristic which is static characteristic repeatability over long time lapses
between the measurements. So, these are some static characteristics which we are going to
discuss in detail.

The other sensor characteristics are dynamic in nature dynamic means the output of the
sensor actually drifts or it changes with time. So, these are called dynamic characteristics.
So, these are first is rise time. So, for example, we have a sensor when we apply a step
input for example, to this and the output changes like this graph it shoots up and then it
goes down and it settles down to certain final value. So, there is a rise time involved rise

15
time is how much time the output of a sensor takes in reaching from 10 percent of the final
value to the 90 percent of the final value.

So, for example, we have 1 which is kind of normalized output. So, 0.1 to 0.9. So, this
much and this much. So, this time difference what this sensor takes from reaching from 0.1
to 0.9 this is called the rise time of a sensor. Delay time is how much time the output takes
from reaching from 0 to 50 percent of its final value or the steady state value. So, steady
state value as we have seen this is 1.

So, how much time the output takes from 0 to 50 percent and here it is shown as TD this is
the delay time or the time delay of the sensor. The peak time peak time is how much time
the output takes starting from 0 to 50 percent starting from 0 and to reach the peak value
of its output. So, peak value is this value which is MP here and how much time the sensor
takes is 0 to TP. So, TP is the peak time this is the time it takes in reaching from 0 to the
peak value. The settling time, settling time is the time taken by the output to reach its steady
state value though this is the dynamic characteristic and it takes very long time in settling
to exactly value of 1.

So, we define a criteria that when the output reaches 1 percent or 5 percent that depends on
the application. So, when the time time taken by the sensor to settle down to within 1
percent of the steady state value. So, this steady state value is 1 percent and when the output
reaches to 1 percent of this value then we say this is the settling time and typically we can
see is plotted here T S which is typically 1 percent of the steady state value. The percentage
overshoot is something how much the output overshoots above the steady state value. The
steady state value is 1 and how much in percentage this output overshoots above this value
of 1.

So, this is again MP the peak the this this is TP which is peak time and in peak time the
output approaches the percentage overshoot. So, this is the percentage overshoot how much
the output overshoots above the steady state value. Steady state error is deviation of the
actual SSV from the desired value. So, how much the steady state value is different or is
deviated from the desired value these are the some dynamic characteristics. So, let us
discuss these static characteristics first in detail.

So, these are some of the static characteristics which does not change with time and these
are depends these depends mostly on the sensor the quality of sensor the material of the
sensor and these characteristics are these characteristics are inherent to the sensor. The first
one is accuracy. So, the accuracy is specified by actually inaccuracy or usually error in the
measurement and the formula for error is given like this epsilon A which is the error in
percentage is equal to x m minus x t divided by x t into 100, where x m is the measured
value of the sensor value and x t is the true value of the measurement. So, the error actually
is how much is the difference between a actual value of a parameter and the measured value

16
of a parameter. For example, we have a stick which is 1 meter in length this is the actual
value of the true value of this object which is 1 meter, but when we measure it using a tape
we measure out to be like this is 101 centimeter or 1.01 meter. The actual value of the
object is x t which is 1 meter only, but x m the measured value is 1.01 and because of this
measurement we have a error in measurement which is given by x m minus x t which is
1.01 minus 1. So, this is the error which we have in measurement and we multiply it by
100 so that it comes in percentage. So, this error is actually the difference between actual
measurement and the measured value and which is normalized to the true value of the
measurement.

So, this is called epsilon a sometime this error is often expressed for a full scale output as
well. What we mean by this full scale output is we have a sensor or a meter this has a
particular range let us say from 0 to 100. So, this sensor is going to measure only from 0 to
100 which is the full scale or the sensor. So, we define this error as a full scale output means
how much is the error in percentage for a complete full scale reading. So, we define this
by epsilon f s o is equal to x m minus x t which is the measured value and the true value,
but this is now divided by the full scale x f s o which is the full scale output of the sensor
and we multiply again with 100 so that it is in percentage.

So, this error is often expressed in terms of the full scale output. So, we have two ways of
expressing this error which is one the accuracy as a percentage of full scale percentage f s
or percentage of reading which is the first case where we normalize it with the true value
of measurement. There are difference between these two expressions of error. Let us take
it using an example if we have an instrument where the accuracy specified as percentage f
s or the full scale if we specify the error as a full scale output then the error will have a
fixed value no matter where the flow is in the flow range. So, here we are talking about a
flow meter which is going to measure the flow of liquid let us say water.

So, this is the diagram of a flow meter that where the liquid goes in goes out and this has
a particular range from let us say 0 to 100 or something. If we show the error we as a
percentage of full scale then the error will have a fixed value. We calculate this value at a
full scale and this error remains fixed no matter how much liquid is flowing into this flow
meter. So, for an example this instrument calibrated for a flow of 100 litre per minute with
a stated accuracy of 1 percent of full scale. It means when a flow of 100 litre is there through
this sensor the error will be 1 percent of 100 litre which is 1 litre per minute.

So, this error is now fixed because we are expressing this in terms of full scale. No matter
now how much liquid is flowing into this sensor the error is plus minus 1 litre per minute
this is the error which is produced by this flow meter. Now, if instead of 100 litre per minute
now my flow is only 50 litre per minute, but the error remains same 1 litre per minute. How
much is the error in percentage now? 1 litre in 50 litre per minute which is 2 percent this is
now higher than 1 percent what we have shown for a full scale reading. Going further if

17
the flow reduce further to 10 litre per minute this error remains same 1 litre per minute then
the error is 10 percent which is quite huge error.

So, if we express the error in terms of full scale reading then the magnitude of error remains
same and if the input reduces or the output reduces then this error becomes significant
compared to the full scale output. However, if we express this error in terms of percentage
of reading then the error will this always remains percentage of what actually the flow is.
So, for example, if we have if we express this error in terms of percentage R d then 100
litre per minute has a percentage of 1 percent it means 1 litre, but when the flow reduces to
10 litre the error is only 1 percent of 10 litre. So, this is the error which is given by the error
which is quite less error compared to what we have shown in full scalar case. So, these are
the two ways of expressing the error as in terms of full scale output or in terms of the actual
magnitude. The next static characteristics is precision. So, the precision is how far the
measured quantity is reproducible as also how close it to its true value. So, the next quantity
is precision. So, the precision is up to how far the measured quantity is reproducible. So,
we have output which is y as a percentage of full scale and the input is x which is in
percentage.

Now, we sweep the input from 0 to x m which is the complete value. If we do this same
measurement multiple times there is going to be the difference in the output. So, in one
case when we sweep from 0 to x m the output traverse this curve let us say this is 1. If
second time we apply the same input from 0 to x m the output is this second graph. So,
there is a certain difference between these two output and this is called the repeatability.

It means that out the sensor is not perfectly repeatable and there is a there exist a certain
change or the certain difference in the output for the same input. So, this term repeatability
is close to precision which is the difference in output at a given value of input when
obtained in two consecutive measurements. Next, we come to the difference between
accuracy and precision. How what is the difference between an accuracy and a precision?
Accuracy is the agreement between the measured value and the true value. It shows that
what we measured how much it is close to the true value of the object true value of the
parameter.

However, the precision is if we repeat the experiment multiple times how closely the output
is confined all the output should be in a very close range that is called precision. In a graph
if we plot we on the y axis we have this probability density function and on the x axis we
have this value which is where we have this true value of output this is the true value or
the reference value. But, when we apply multiple measurements and most of the time we
end up at this value which is let us say our mean value of measurement and the difference
between this actual value or the reference value and this measured value this is our
accuracy. However, we are doing this measurement multiple times and let us say all these
100 for example, 100 measurements all these measurements are lying here follows a

18
Gaussian curve and the peak exists in between here. How confined these output are in a
particular range this is specified by the precision.

So, this is the accuracy how much is the difference between actual and measured and how
much output is confined to a narrow range that is called the precision. So, accuracy
indicates proximity to the true value however, the precision to the repeatability or
reproducibility of the measurement. This is the difference between the accuracy and
precision. Accuracy is the degree to which the measured value is close to the correct value,
measured value is close to the correct value. However, the precision is it describes how
close the measured values are to each other in the data set it need not to be the accurate one
or it need not to be close to the true value.

Usually accuracy involves single factor measurement you just apply one input and you can
measure the accuracy of sensor. However, for to measure the precision you have to repeat
the experiment multiple times and get multiple numbers of output and see how close they
are confined. For a measurement to be accurate it should be precise. However, the
measurement can be precise without being accurate means the accurate measurement is the
measurement when it is precise and accurate. However, it measurement can be precise to
certain other value which is different than the true value so, means it may not be accurate.

Let us take an example of some bullets we fire let us say we fire 5 bullets from a gun and
one knows the exact number of bullets fired, but multiple attempts are made to obtain the
precise result. The bullets hitting the bird's eye are precise and those hitting close to the
bird's eye are accurate. So, this is the goal so, let us say we have this as a goal we want to
hit on this which is the true value of measurement. We are hitting the bullets and bullets
are not hitting true value but they are hitting somewhere here, here which is different than
true value it means this is not accurate. However, if we see all these marks they are not
even confined to very close to each other it means this is not even precise.

However, this case all these hit marks they are very much in a very close range though they
are not at the true value it means this is precise, precise means the output is very much
confined, but not accurate which is not equal to the true value of measurement. In this case
if we see the third case when we have some hit marks are there on the true value it means
the measurement is accurate, but it is not precise again because we see there is a distribution
of these hitting marks. This case, case number 4 where we have all these marks are right at
the true value which means the measurement is accurate and all the output or all the marks
are very much confined to each other it means this is precise as well. So, this is precise and
accurate. So, this is the desired property of a sensor or a transducer, it need to be accurate
and it need to be precise.

So, this is a example of a case where the system is precise, but it is not accurate. Let us talk
about a refrigerator which has a thermometer inside. So, this is it measures the temperature

19
and displays out the reading and we have taken 10 readings which is 39.1, 39.4, 39.5 and
so on 39.1 all these values we have taken from this sensor. However, the actual temperature
inside the refrigerator is 37. So, we can see all the output readings these are all the output
of a sensor these are all confined to 39.1, 39.4 they are very much close to each other it
means this is precise the sensor which is used in the refrigerator is precise, but it is not very
accurate because the difference between the actual and the measured value is 2 degree
which is 39 let us say 39.0 minus 37. So, it around 2 degree centigrade is the error which
is produced by the measurement of sensor. So, this case is we have precise sensor, but
which is not accurate sensor. This is another example of accuracy and precision. So, there
are let us say 2 sets of data from measurement on 2 different transducers let us say set A
and set B. Set A measurement is 32.56, 32.55 and so on set B is 15.38, 15.37 and we need
to find out which one is more precise. So, we subtract the lowest data point from the highest
one. So, for set A we have 32.56 minus 32.48 which is 0.08 and for set B we have 15.38
minus 15.32 which is 0.6. So, this difference it shows the range of output from the highest
one.

So, this is the maximum minus minimum. So, this shows the complete range of output how
much the data has taken up The higher the range of output more less precise the sensor is.
It means the sample B or the set B has the lowest range and so is the more precise. So, this
transducer 2 or transducer B is more precise compared to A. next static characteristic is
resolution. So, the resolution is defined as the smallest increment change in the output
which can produce a detectable output of the sensor.

This is often expressed as percentage of measured range. So, the formula of resolution is
R max as a percentage equal to 100 into delta x m divided by m R. How do we calculate is
m R or the measured range is equal to x max minus x min which is the maximum value of
x minus the minimum value of x and for a detectable output delta y if the minimum change
required is delta x then R max is 100 into delta x m divided by m R. Let us take an example
for this resolution measurement. We have a measurement system which can take
measurement across a plus minus 10 volt range. It means that the range is plus minus 10
volt. This is 0, this is 10 volt and this is minus 10 volt. So, this is 0 and this measurement
is being done using a 16 bit A to D converter, 16 bit analog to digital converter. So, the
number of levels we will have in this A to D converter will be 2 to the power 16. These
will be the number of levels produced by A to D converter. So, 2 to the power 6 equal to
65536. These will be the number of levels produced by A to D converter and how much
this will correspond to in the voltage minus 10 to plus 10 means 20 volt. This full range is
20 volt minus 10 to 10. So, the resolution is given by the resolution of the system will be
this 20 volt divided by 2 to the power 16 which comes out to be around 305 micro volt. So,
the resolution of this complete system using this 16 bit A to D converter will be 305 micro
volt. This much will be the output generated by this A to D converter. Next characteristic
is the threshold. So, the threshold is at the 0 value condition of the measurement the input

20
quantity is at it is 0 level. How much is how much the smallest input change that produces
a detectable output is called the threshold. For example, we have a MOSFET transistor the
input is 0 the gate voltage let us say is 0 how much gate voltage we need to apply so that
the MOSFET turns on. So, that is called the threshold voltage of a MOSFET. Another
parameter is sensitivity which is very important for sensors. The sensitivity is the ratio of
incremental output to incremental input. So, we can write S equal to delta y which is the
output divided by delta x which is the input. So, this is output divided by input.

Usually we express it in normalized form. So, we can write delta y upon delta x divided by
y by x how much is the actual the the base value of output and input we normalize the
sensitivity using their base values. In other words the sensitivity is for a unit change in the
input how much output is being produced by the sensor. So, the sensitivity is the term
which we want to have more and more for a unit input if the sensor can produce more
output that will be good or the sensitivity will be very high. If sensitivity or the out level
changes with time temperature and or other parameters without any change in the input
level that is called drift in the system. Sometime the output of a sensor changes with other
parameters as well for example, temperature and that is called a drift and which leads to
most of the time which leads to the instability in the system. The example is let us consider
a pressure sensor that has a measurement range of 0 to 100 psi and an output range of 0 to
5 volt. Its sensitivity is 0.05 volt per psi. The sensitivity of an instrument says nothing about
the quality of an instrument.

This is typically established by the instrument's accuracy and precision. Next parameter is
selectivity. The output of a sensor let us say we have a sensor which we design for a
particular parameter. This sensor is not only sensitive to only that parameter, but it is
sensitive to multiple other parameters for example, temperature. If the temperature changes
the output of the sensor also going to be changed. So, the output of a sensor may change
when afflicted by environmental parameters or even some other variables and this may
appear as an unwanted signal. So, the sensor is then said to be non-selective. When the
sensor starts sensing other parameters as well then the sensor is called a non-selective
sensor and we can write this partial sensitivity to other parameters as Sjk where k is some
other input and j is some other output and Sjk equal to delta yk divided by delta xj means
change in input x produces a change in the output y where the output y may not be a desired
output from delta x. And a selectivity matrix would be obtained Sjk and the jth entry and
obviously, an ideally selective system will have only diagonal entries. So, this we will see
in the next example. So, for example, we have a gas sensor and we make this gas sensor to
detect oxygen, carbon dioxide, CO, NH3, VO and other VOCs. So, the input quantities are
these 5 and based on based on that the output will be generated.

So, for example, if we flow the oxygen and the sensor oxygen output according to oxygen
will be produced if we flow CO in the system the CO output corresponding to CO will be
produced. However, the sensors they have partial sensitivity to others as well. For example,

21
we have we flow oxygen in the system there will be a output generated by O2 according to
O2, but there will be partial sensitivity to CO as well which should not be there. Similarly,
if we let us say we flow NH3 in the system. So, NH3 should be detected, but along with
this VOC is also be going to be detected. So, this is undesired which is because of the non-
selectivity of the sensor. If a sensor is very selective then we will have only the diagonal
element means oxygen will produce output according to oxygen here CO2 will produce
output according to CO2 only CO will produce output to CO only and NH3 to NH3 and
VOC to VOC to we we will have only these diagonal elements in the selectivity matrix for
a selective sensor. Next parameter is non-linearity when the output of a system deviates
from a perfectly linear output then it is called a non-linear system and this non-linearity
can be specified in two different ways. First is let us say we have in this graph we have this
output produced. So, we can fit a straight line in this output and the difference between this
fitted line and the actual data this is the error produced by measurement.

So, this is one expression of non-linearity where we fit the data fit a line in the data and
express the error. Second is we just join the first point and the last point using a straight
line and we measure this error between the this fitted line and the actual data. So, this is
based on joining the end points of the scale and the maximum non-linearity in the first
method this is always lesser than the second method. So, this method will have lesser error
compared to this method. A consequence of non-linearity is distortion. So, when the output
of a sensor deviates from a straight line output then it produce distortion which is defined
as the deviation from an expected output. So, let us understand what is non-linearity. So,
consider a linear system where we have this this is a completely linear system and we apply
two frequencies to input let us say frequency 1, f 1 and frequency 2 which is f 2. We are
talking here in terms of frequency domain and we have this frequency system which is
spectrum f 1 and f 2 these two frequency we apply at the input. This linear system will only
produce output at these two frequencies means the output will be amplified that the level
here will be different let us say at input it is A and the output it will be A 1, but the
frequencies will be same f 1 and f 2 only.

So, no harmonic generation in linear system. However, if we have a non-linear system here
we apply only two frequencies, but because of the non-linearity the system will generate
harmonics. Harmonics are integer multiples of the input frequencies. So, at the output we
get f 1 and f 2 and we get harmonic generation harmonics as well. So, we get f 1 and f 2
and the second order harmonics will be f 1 plus f 2 and f f 2 minus f 1 these are the second
order where we have two terms added or subtracted together and this f 2 plus f 1 will be
somewhere here and f 2 minus f 1 will be here and f 2 plus f 1 will be somewhere here.
Third order harmonic will be where three terms are involved here two terms of f 1 minus
one term of f 2 and two terms of f 2 minus one terms of f 1. So, this is third order harmonic
which appears somewhere here. So, this is a non-linear system where the system generates
harmonics of the input frequencies and these harmonics are generated due to the non-

22
linearity of the system. These type of non-linearity or these harmonic generations these are
y very important for high frequenc applications. Next characteristic is hysteresis. So,
hysteresis is the difference in the output of the sensor for a given input x when x reaches
its value in upscale and downscale direction means we have a input range let us say from
x min to x max we when we apply from x min to x max the output follows this path, but
when we come back from x max to x min the output follows a different path. The output
does not go to the same path it follows a different path while coming back and for going
up this is the path. So, these two paths are different ideally it should be same, but if the
system has hysteresis or the memory effect we call it then these outputs are different and
there is exist an area in between. So, this is called hysteresis in the system this is the case
of analog where we have a in analog measurement hysteresis look like this. However, in
the digital case it look like this when input is changed from low to high it follows this path,
but when coming back from high to low value this follows different path. So, this is a
digital case where a Schmitt trigger produce this kind of hysteresis loop.

Next parameter is the impedance. input output impedance of the sensor or a transducer is
the characteristic which need to be considered when we start connecting more than one
systems in cascade. Let us say we have one system and we want to connect it to the other
system that time we need to take care of the how much is the input impedance and output
impedance of these systems and these causes significant restrictions in interfacing these
blocks. For example, here in this case we have this voltage source which is 100 milli volt
in magnitude and it has a let us say internal impedance of 1 ohm and we want to connect it
to an amplifier which is shown here this is amplifier. This amplifier has an input impedance
of let us say 100 ohm. In isolation condition this voltage source will produce a 100 milli
volt output here when it is in isolation it is not connected to any other block. But as soon
as we connect this voltage source to an amplifier which is 100 ohm in input impedance the
voltage at the input will not be 100 milli volt this will be distributed between this ZS and
Z1. The net voltage will not be 100 milli volt now it will be approximately 100 divided by
100 plus 1 into 100 milli volt. This is your Z in divided by Z in plus ZS into 100 milli volt.
This is your Z in divided by Z in plus ZS into 100. So this 100 milli volt will be divided
using resistor divider across this path and this voltage is now lesser than 100 milli volt
because this quantity is lesser than 1. That will be typically some 99 milli volt or so. So
because of the interconnection of these two blocks now the output produced by this voltage
source is reduced as soon as we connect these blocks together. So this input and output
impedance need to be taken care of if we have a very less impedance at the input of this
amplifier this voltage is further going to be reduced and accordingly your system gain will
be limited. So these input and output impedances are some static characteristic which need
to be considered. Next is isolation and grounding. So isolation is necessary to eliminate or
at least reduce undesired electrical magnetic or electromagnetic and mechanical coupling.

23
So when we have these sensors and we connect it to some electronics then that electronics
may be working at different voltage and the sensor may be working at different voltage for
example, 5 volt, but the processing unit which is let us say computer or some electronics
which is working at 12 volt. So this sensor should not be exposed to this 12 volt because
sensor is at 5 volt if 12 volt comes in the sensor can be damaged. So there should be an
isolation between the sensor and electronics however, the data should flow from sensor to
electronics it means the AC signal the data should flow, but the DC voltage should be
blocked. So we use some kind of decoupling capacitors between these two blocks so that
the DC voltage are isolated however, the AC signal can flow. Similarly grounding is also
necessary to establish a common node between these different parts of the system with
respect to which the potential of any point in the system remains constant.

We need to have a common ground with reference to that we define all the voltages of
other nodes. So these are some static characteristics of the sensors which we use for
characterization of these sensors. These are some other examples for example, the length
of a rectangular box is 1.2 meter, but if it was measured with a tape and the length was
measured at 1.22 what is the accuracy of the system. So similar kind of experiment similar
kind of example we considered earlier as well. This is the actual value or the true value Xt
and 1.22 this is the Xm and so we know the formula error equal to Xm minus Xt divided
by Xt into 100. So we put the values and we get the results around 1.67 percent is the error
in this measurement hence the accuracy is 100 minus 1.67 which is 98.33 percent. Another
example is let us say we have a measuring tape and it can measure with an accuracy of 99.8
percent only. What is the possible range of length which can be obtained by using this
measurement tape to measure a cloth of length 2 meter. So this 2 meter length is the value
of Xt and we are given accuracy which is 99.8 percent so the error will be 99.8 percent 100
minus 99.8 percent which is 0.2 percent. Now the new measurement or the measured value
using this measuring tape will be actual value 2 meter plus minus the error which is 2 plus
minus 0.004. So the maximum value of measurement will be 2.004 and the minimum value
will be 2 meter minus 0.004 which is 1.996.

So this is all we have for today.

Thank you.

24
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 3
Dynamic Characteristics of Transducer

Hello, welcome to Transducers for Instrumentation. Today we will discuss the Dynamic
Characteristics of Sensor, what are the parameters which we use for characterization which
are dynamic in nature unlike the static characteristics which are fixed with time. For
example, the accuracy of a measurement which is a static characteristics whatever the error
give sensor gives today if we measure tomorrow the error will be the same. So, it does not
change with time. However, some characteristics are there of a sensor which is changing
with time the the performance of a sensor for example, the output it the how how much is
the output response with respect to time. So, that is the dynamic characteristic of a sensor.
So, the dynamic response it involves the determination of transfer function of the complete
system, the frequency response how the system is going to respond to a certain frequency
the impulse response if we apply a sharp impulse at the input if we apply step response at
the input and then evaluation of time dependent outputs. So, if we apply these functions
which are time dependent functions if we apply this function at the input what will be the
output response of a sensor. So, these are the dynamic characteristics. For determining
these dynamic characteristics different inputs are given to the sensor and the response
characteristics are studied how the sensor is responding to these inputs.

With step input the specification in terms of the time constant of the sensor are generally
made and generally the sensor is a single time constant device it has a particular time
constant and accordingly it produce the output. Though this topic is largely covered in
control system we will just look at some basic characteristics which are dynamic in nature.
So, we discuss a second order system which is a very common type of system which we
consider in control system which we have in this example we have a mechanical system
where we have this mechanical block with mass m and this is connected to this is spring
with a spring constant k and also connected to a damper where the damping coefficient is
b and this block is here on a friction less support where it is free to move this direction and
the other side is connected to a fixed wall. Now if we write the force equation of on this on
this block minus F b minus F k this is because of the damper and the spring respectively
equal to minus b d x by d t minus a k x equal to the force on this block which is m into a
m into d square x over d t square. So, this is the total force which is acting on this
mechanical block and if we just rearrange this equation the equation comes out to be m d
square x over d t square plus b d x by d t plus k x into equal to 0.

Now we can see this is a second order equation which is in with respect to time d square x
by d t square this x is the displacement what we produce and this displacement is with

25
respect to time. So, this is a second order differential equation. So, this is how we make a
model or a equation of this mechanical block. Similarly, if we consider a electrical system
the same kind of equations we can make. So, we have here a capacitor which is connected
in parallel to a inductor which is in series with a resistor. So, this in this has R L and C and
on the left we have the source connected to it. Now voltage across the resistance will be I
into R I is the current flowing through this resistors into R. So, V R equal to I into R voltage
across the inductor this is V L which is L di by d t where L is the inductance and I L is the
current flowing through this inductor which is with respect to time. Current through a
capacitor is given by C d V C by d t where C is the capacitance of this capacitor and V C
is the voltage across the capacitor build up and divide by the d t. Again, we can write the
equation across it and d d square I L by d t square plus R by L d L d I L by d t plus 1 upon
L C I L equal to 1 upon L C I S this you can easily write using Kirchhoff current law. And
again, if you see here this is a second order system. So, here if we see if we have a
mechanical system that also we can model and that is a second order system and if we have
electrical system in hand that also we can model and apart from the parameters where we
have R and L in electrical system and in mechanical system, we have M and K etcetera.
Apart from these constant parameters the equations are same. So, we can apply the same
theory to both kind of systems. Before solving these equations actually, we use Laplace
transform to solve this kind of equations.

Why do we use it? Because the Laplace transform converts a calculus problem for example,
these linear differential equations to an algebra problem. So, instead of solving the
differentiation and integration we apply Laplace transform and by doing that this becomes
a algebra equation where we can just add multiply divide and subtract this these formulas
we can apply. So, how to use this Laplace transform? What we do? We take the Laplace
transform of this differential equation after taking this Laplace transform it becomes a
algebraic problem. So, we solve this algebra problem and after solving we again take
inverse Laplace transform convert it back to into the time domain. So, one example here is
given we take Laplace transform of this time domain function and convert it into if into a
F s which is in Laplace domain and again, we after solving we take inverse Laplace of this
F s convert it back to the time domain. So, this is how we apply the Laplace transform to
these equations and the standard form of this second order system when we apply the
Laplace transform is this equation which is very general equation G s equal to omega n
square divided by s square plus 2 zeta omega n s plus omega n square where zeta is the
damping ratio and omega n is the undamped natural frequency of the system. This
conversion into Laplace domain and other these calculations are basically covered in
control system which we are not going to discuss completely here, but we start from this
point where we convert it into Laplace domain and then we start solving the the the
question. So, for example, we have this kind of system where we have gain in the system
and we have a transfer function in Laplace domain. So, suppose loop transfer function we
can write the output divide by the input which is the transfer function and we can write like

26
this and this is this comes out to be second order system. This second order system have
transfer function of the following form B upon s square plus As plus B we can use this or
we can use the old one which is omega n square upon s square plus 2 zeta omega n s plus
omega n square.

This also we can use and this transfer function may be multiplied by a constant which can
be here as well this which affects the exact constant of time domain signal, but not its form.
Now, we have this equation which is second order equation and we call it the characteristic
equation. This equation is called characteristic equation and based upon the roots of this
quadratic equation of this characteristic equation the system can be classified in one of the
4 forms which are shown here. So, if we have the poles or the roots of this characteristic
equation which both are real it means there is no imaginary part to these roots then the
system will be over damped. And the system responds if we see in the time domain it look
like this over damped response. This is the final value which is steady state value or we
can say and the system approaches this steady state value in a very sluggish way this is
called over damped. And the characteristic of over damped system is the roots of this
characteristic equation both are real. Next case may be when all these these 2 roots are
complex they have real part as well as the imaginary part then the system will be under
damped. Under damped system is shown this is a under damped system where the system
shoots above the steady state value and then tries to settle it down. It means the damping
in the system is not enough then there is a overshoot in the system. And the characteristic
is when these both the roots are complex. If we have only the 2 imaginary roots of this
characteristic equation means there is no real part to that then the system will be undamped
means there is no damping in the system at all. So, if we apply a certain input to the system
the system goes into the oscillation mode. This is the undamped case where there is no
damper involved. For example, if we are talking about the mechanical system the first
example the block is sitting on frictionless surface and damper is not there only spring is
attached we disturb this equilibrium we pull the mass and leave it this system will be in
oscillation state where there is no force available to stop the system to stop this vibration.

Friction is not there and the damper is also not there that is case of an undamped system
and the root of this characteristic equation in that case will come out to be both imaginary.
Another case is the critically damped where both the roots are repeated real and both roots
are same and they are real. In that case the system achieves the steady state value fast and
settles down there. So, this is the critically damped case where system achieves steady state
value and the both of the roots are repeated real. So, here we have example for example,
we have this as a system where characteristic equation is S square plus 9 S plus 9. This is
the characteristic equation if we take out take the root of this characteristic equation those
roots if we plot on real and imaginary axis those roots fall here and here. So, both roots are
real with no imaginary part imaginary part is 0 j omega. So, the system response will be
something like this which is a over damped case. Another case is when we have

27
characteristic equation S square plus 2 S plus 9 this is the characteristic equation if we take
out the root of this equation and plot it on real and imaginary axis both root comes like this
both roots are real as well as imaginary part. So, in that case the system is under damped
and system will have overshoots and undershoot.

So, this is called overshoot and this is called undershoot and this is in between we have
steady state value. The third case is we have this type of equation where S square plus 9
where the zeta is 0 and in this case if we plot it plot the root of this equation on real and
imaginary axis they comes out here where both are imaginary with without real part. In
that case the system is undamped and it goes into the oscillatory mode if we apply a step
input to this. So, this is the oscillation around this steady state value because there is no
force existing there which is opposing this motion. The fourth case is the critically damped
where we have equation for example, S square plus 6 S plus 9 this is the characteristic
equation, and we have if we calculate the root and plot it on real and imaginary axis both
roots are real and they are same if we plot it plot the output of this system that comes into
the critically damped. The system achieves the output at very fast rate compared to the over
damped case and reaches the steady state value. So, these are the four cases of output
response which depends on the root of this characteristic equations. So, this is one way of
characterizing the output. However, we can characterize the response of second order
system using two parameters omega n and zeta instead of roots. So, instead of calculating
the root of that characteristic equation we can define these two parameters which are omega
n and zeta based on that we can classify the system.

Omega n is the natural frequency of the system. So, this is the frequency of oscillation
without damping if we remove the damper from the sensor then the frequency of oscillation
is this natural frequency. So, for example, in an RLC circuit if we remove the resistor from
the system because resistance is the element which is opposing this which are kind of
dissipating this energy and stopping this oscillation. If we remove the resistor then the
frequency of oscillation will be the natural frequency. The damping ratio zeta this measures
the amount of damping like how much damper or how much damping force is there in the
system. So, damping ratio defines that and the typical range of zeta is between 0 and 1.
Damping ratio zeta is defined as exponential decay frequency divided by the natural
frequency and the exponential decay frequency zeta d is the real axis component of poles
of a critically damped or undamped system. So, we discussed we can have a quadratic
equation and we can solve for the roots or we can compare using standard characteristic
equation which is S square plus 2 zeta omega n S plus omega n square. This is standard
characteristic equation we can compare using this and find out the value of zeta and omega
n. These zeta and omega n are more meaningful compared to finding the roots. When we
start designing these are the more handy parameters. So, earlier we calculated the roots for
example, for a undamped system we had roots like this, but if we calculate zeta, zeta is 0
for this undamped system. Similarly, for under damped system we calculated two roots

28
which are two complex for these two complex system zeta comes out to be between 0 and
1. For repeated roots where we have both the roots are real and it repeated in case of
critically damped zeta comes out to be equal to 1. And for under damped case where these
both the roots are real zeta is greater than 1.

So, depending upon the value of zeta we can specify whether the system is undamped. If
zeta is between 0 and 1 we can have under damped. If zeta is equal to 1 then we have
critically damped and if the zeta is greater than 1 we have over damped. So, we can
characterize the system using the value of zeta instead of finding out the root of this
characteristic equation. So, here are some examples which we can consider. Let us say we
have this first example where we have characteristic equation is S square plus 8 S plus 12.
If we compare it using our standard equation where we have S square plus 2 zeta omega n
S plus omega n square we can see omega n square is 12. So, we can calculate omega n is
equal to under root 12 and 2 zeta omega n is equal to 8. So, we can calculate zeta is 8 upon
2 into omega n which is root 12. So, this is how you can calculate omega n and zeta. And
based on the value of zeta you can specify which type of network it is. So, in first case zeta
comes out to be 1.155 which is a over damped system. Similarly you can calculate for this
this example S square S square plus 8 S plus 16 you can compare it with characteristic
equation here and find out the value of zeta and zeta comes out to be 1 in this case which
is critically damped. For for the case C we have S square plus 8 S plus 20 you can compare
again with characteristic equation and zeta is 0.8 which is an under damped case. So, this
is how using these characteristic equation and looking at the transfer function of the system
we can we can calculate or we can know the system is over damped under damped and
based on that we can calculate how the system is going to respond if we apply a step over
input to the system. So, math expression for ah YT for ah under damped case let us say we
consider the under damped case where damper is not very strong means the zeta falls
between 0 and 1. This is the transfer function is Laplace domain this is Laplace domain
and to convert it back into time domain we take inverse Laplace and the output in the time
domain is given by this equation this is time domain equation is because there is a time t
variable here present here. So, this is a time domain equation and we can now plot it using
ah computer or we can plot it by varying this t. The output in time domain of the system
looks like this.

So, we have this this is a time domain the time axis and this is the magnitude and the output
is follows this this blue curve. So, this is the response of an under damped system this is
the plot of this time domain equation. Few parameters we can see in this the output start
from 0 and reaches a peak value which is Y max or the peak value in certain time which is
let us say t p and then there is a decay in the magnitude because of the damping. Now we
can define multiple parameters for this system based on the output characteristic. The first
one is the peak time. So, the peak time is the time required to reach the first peak in the
output when the output starts ramping it reaches a peak it comes down and then it oscillates

29
and reaches the steady state value the first peak the time it takes to reach that that first peak
is the peak time the t p we call it and in this graph if we see this is the peak value and this
is the time which it takes to reach from 0 to this peak value which is t p. Percentage
overshoot so, this overshoot is always represented in terms of percentage this overshoot is
represented in percentage which is relative to the steady state value. So, percentage
overshoot means how much percentage it goes above the steady state value. So, percentage
overshoot or percentage os is the amount that the response exceeds the final value at t p
final value is your steady state value steady state or SSV we call it in short. Settling time
so, settling time is the time required for the oscillations to die down.

So, as we can see the oscillations take a lot of time in completely dying out. So, we define
a range when the oscillations are within 2 percent of the final value 2 percent or 5 percent
that depends on application. So, for example, we define when the output remains within 2
percent of the final value then that is the time taken to reach that that 2 percent within 2
percent value is your settling time. Rise time means when the output is rising how much
time it takes reaching from 1 from 10 percent to the 90 percent of final output. Final output
means the steady state value the how much time the system output takes reaching from 10
percent to 90 percent of steady state value that is the rise time. So, these are some
parameters which we can calculate and the formula for these all these parameters are given
here. If we have this let us say characteristic equation S square plus 2 zeta omega n S plus
omega n square then the peak time is T p which is given by this equation peak value is
given by this expression percentage overshoot is this and the settling time is 3 upon zeta
omega n S or 4 upon z depending upon we specify 2 percent or we ah stay with the 5
percent value. So, depending upon this the time the settling time equation is that. We can
see one thing all these parameters depends only the omega n and zeta. So, these 2
parameters are very much important in in describing the system. Some properties are the
percentage overshoot depends on zeta, but not omega n. So, the how much system
overshoots with respect to the steady state value that depends only on zeta the damping
factor it does not depend on the natural frequency of the system. From second orders
transfer function analytic expression of delay and rise times are hard to obtain if we have
only the second order transfer function ah it was very difficult to obtain. Time constant is
1 upon zeta omega n which indicates the convergence speed. For zeta greater than 1 we
cannot define peak time and peak value because for zeta greater than 1 this is over damped
case where there is no overshoot of the output hence we cannot define the peak time, peak
value and percentage overshoot.

So, let us consider some numerical example where we have the transfer function of the
system is given by C s upon R s and which is 100 upon s square plus 20 s plus 100. So,
find out the nature of the system means it is over damped, under damped or critically
damped. So, this is the transfer function given to us and we need to compare it with the
characteristic equation. So, by comparing this you can see omega n square is 100. So, the

30
omega n comes out to be 10 and we compare 2 zeta omega n equal to 20 this is 2 zeta
omega n equal to 20 we put value of omega n as 10. So, zeta comes out to be 1. So, when
the zeta comes out to be 1 it means the system is critically damped. So, another example
when a second order system is subjected to a unit step input the value of zeta is 0.5 and
omega n is 6 radian per second determine the rise time, peak time, settling time and peak
overshoot. So, for the peak time we have T p equal to pi upon omega n under root 1 minus
zeta square we put the value of zeta and omega n which these 2 parameters are given we
can apply omega n here and zeta here which is T p comes out to be 0.6 second. So, this is
the peak time the system takes in reaching the its first peak. Settling time is 4 upon zeta
omega n a we apply zeta and omega n here and we get settling time is 1.33 seconds.
Maximum overshoot in the system is e to power minus pi into pi into zeta divide by e to
the power under root 1 minus zeta square into 100.

So, M p comes out to be 16 percent. So, this 16 percent is the maximum overshoot above
steady state value. So, we have steady state value at this value and the maximum overshoot
is 16 percent this is 16 percent above the steady state value. Another example is let us say
we have this second order system where the equation is S square plus 2 S plus 4. We again
compare it with characteristic equation S square plus 2 zeta omega n S plus omega n square
and we calculate omega n is omega n equal to 2 and zeta equal to 0.5. Once we know the
value of zeta n and omega zeta n omega n then we can calculate omega d which is the
damping frequency equal to omega n under root 1 minus zeta square this one. The peak
time is T p equal to pi upon omega d which is 1.82 seconds. Settling time we can calculate
3 upon zeta omega n S which is 3 second and the maximum overshoot is 16.3 percent. So,
these are the some examples how do we define a system and we know how the system is
going to response to a unit step input and how the output will look like in time domain
what will be the maximum overshoot what will be the peak time, settling time and other
parameters. We can use this these parameters and design the system itself. We use these
parameters let us say we need a system where the peak time should be x and the let us say
the settling time should be y if these parameters are known we can even design the system.
So, one example is given here we have this mechanical system which is a mass which is
connected to again this this spring and a damper the transfer function we can write in
Laplace domain like this 1 upon j divided by S square plus d by j into S plus k upon j. We
compare it again with characteristic equation S square plus 2 zeta omega n S plus omega n
square.

We know the value of zeta we know the value of omega n because we know the parameters
of the values of peak time settling time overshoots from that we can calculate omega n and
zeta and from this transfer function which we developed for this mechanical system. We
compare it and we calculate omega n is equal to under root k upon j and 2 upon zeta is 2
zeta omega n equal to d upon j. We know the value of omega n and zeta. So, we can
calculate the value of k and j which is the system parameters k is the spring constant j is

31
this mass and d is the damping. So, these values we can decide based on how much peak
time or settling time we want in the system. So, the specifications of let us say 20 percent
overshoot allows us to calculate zeta is this and the specification of T S equal to 2 allows
us to calculate zeta omega n and we can easily calculate the value of d and the value of j.
So, these are some examples which shows that the system how the system performs when
we apply a unit step input to the system. These are all the characteristics which are time
dependent. So, these are dynamic characteristics of the system.

This is all for today.

Thank you.

32
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 4
Errors in Measurement and Instrumentation, Propagation of Errors.

Hello, welcome to the course Transducer for Instrumentation. In today's lecture, we will
discuss about the errors in measurement and the propagation of error. Errors or I would
like to call it uncertainty because with word error most of the students confuse it about
the mistake or the blunder they have performed in some measurement which is not. The
error is actually an inevitable part of a measurement you cannot actually remove the error
completely, you can only suppress it to certain extent by using some very advanced
instruments or taking very good care, but it is not possible to entirely eliminate the error.
So, error is not something which a student or a scientist is performing during the
measurement it is a part of the system it depends on the system characteristics as well.
So, the error does not mean a mistake or blunder, error is in a scientific measurement it
means an inevitable uncertainty it is a kind of randomness in a measurement that exists in
all the measurements no matter what measurements we are performing it is a mechanical
measurement or it is a electrical measurement or any other optical or any other
measurement.

The error will be inevitable it cannot be removed. So, error cannot be eliminated by being
very careful you can be 100% careful during all your measurements, but still the certain
amount of error will be there in all the measurements it can be less or it can be big
depending upon multiple factors which we are going to discuss, but it cannot be made 0.
Some sources of error are intrinsic to the process for example, we use a multimeter or
some voltmeter to measure certain parameter that voltmeter itself has some problem
inside. So, some errors can be depending on some intrinsic parameters of the instrument
and cannot be removed entirely. However, some errors are there which you can remove
and all the errors can only be minimized so that they are very less or negligible compared
to the measured quantities. So, for example, we measure for example, we have a 1 meter
rod and we want to measure its length using a inch tape. So, depending upon the quality
of our tape the measurement can be 1 meter plus few millimeter or 1 meter minus few
millimeter. So, this few millimeter extra is it comparable to the original length of the rod
so that actually dictates whether we can leave out this error or we need to improve on our
system. So, error can be only be minimized so that they are very less or negligible
compared to the base value. For example, these are the following general classifications
of error for consumer purposes we tolerate around 5 to 10 percent of errors in
measurements. For example, we have a fridge in our kitchen and the fridge temperature is
let us say 30 degree inside. So, plus minus 5 or 10 percent error in the temperature

33
reading is acceptable for consumer goods. So, we are not very much interested in
knowing whether it is 30 degree or 30.1 degree whether it is 29 or 31 it is acceptable for
consumer purposes.

However, for engineering purposes for example, we are building a dam now 1 percent is
acceptable we cannot accept 5 or 10 percent error in measurements while we are making
it a dam or a bridge. So, the amount of error acceptable is lesser in engineering purposes
because the criticality of measurements is more. The third is for scientific purposes when
we have a scientific experiment going on in the lab and we are interested in very much
precise measurements. So, that time we cannot tolerate even 1 percent of the error in our
measurement that time we go typically for 0.1 percent error in our scientific
measurement.

Though these numbers are not very fixed that depends upon the application for example,
some engineering applications can be so demanding that you need to have 0.1 percent
error and some scientific measurements can be accepting like 1 percent of error. So, but
in a nutshell these are the some numbers what we can accept in a measurement. So, the
classification of error so we can classify those errors what generally occurs in all the
measurements. We can have one is the gross error, the second is the systematic error and
the third one is random or residual or accidental [Link], these are the three types of
errors we can classify and further these systematic errors we can further classify into
instrumental error and then environmental and observational. So, the systematic error we
can further divide into three parts instrumental, environmental and observational. Let us
discuss each of them in detail. The gross error is caused by human mistakes mainly when
let us say you have some measurement to be performed and the human being who is
measuring this experiment or the measurement he performs certain error which is not the
fault of the instrument or any other equipment is purely the human way of measuring if
that has an error that is called the gross error. Let us take the example of this mass which
is hanging on this screen on this spring. So, we have a spring here and we have this mass
which is hanging on this spring depending upon the mass there will be an extension in the
spring and we want to measure how much this end goes down because of the mass. So,
this end goes down and this is the stable state this is the state. We want to measure how
much this spring is elongated using this tape measuring tape we have here a measuring
tape. Now if the observer or the person who is measuring this is right perpendicular to
this axis you see this there is axis y axis if the observer is exactly perpendicular to this to
this axis then the measurement is accurate which is around 12. So, this is the exact
measurement, but let us say the observer is at a angle to this axis let us say case 1 where
the observer is a little bit higher what he will see typically this arrow which is connected
to this spring this arrow is not at 12, but it is slightly below this 12.

So, this is 12.3 if the observer is at a certain height compared to this needle. In case 2 let
us say the observer is below this level then observer will see this arrow is not right at 12,

34
but it is actually above than 12. So, in this case the reading will be typically 11.8 or
something. So, here the instrument is same in instrument is not moving anywhere there is
no error in the instrument itself, but depending upon the location of observer the readings
are different where 12 is the exact reading which is the best case here this is the best case
where the reading is accurate and these are the two cases where observer is not at 90
degree then the reading is different that is purely depending upon how a person is
measuring this experiment. However, there is a certain gap here if we see closely this
needle which is connected to this spring and there is a gap between this needle and this
inch tape this gap actually dictates how much the error will be if the observer is
performing an error. So, if this distance is more than the error will be higher if the
distance is less the error will be lesser. So, instruments are designed in such a way that
these kinds of error be minimized, but they are not exactly 0 there will be some error in
the measurement due to this gap here between the pointer and the scale there will be an
error generated in the measurement. This gross error is a human error in reading,
recording and calculating measurement results this is purely based on the human errors.
The experimenter or the observer may grossly misread the scale. So, this also happens in
the laboratory for example, the reading of the thermometer is 21.5 degree, but while
noting down the temperature reading accidentally the observer writes 31.5 instead of 21.5
is just this digit is mistakenly taken which is not correct and because of that there is a
huge error from 21 to 31 it is almost 50 percent of the error.

So, we call it the gross error the magnitude of this is not controlled by the instrument.
Another kind of errors can be like they may transpose the reading while recording like
reading 25.8 degree centigrade and record as 28.5. So, exact reading is 25.8, but when
they are actually recording in computer or in the notebook these digits are misplaced. The
record at the recording time the observer records 28.5 instead of 25.8. So, these are the
errors made by human beings and these are called the gross errors. These gross errors
these are caused by human mistakes in reading the instruments or using the instruments
depending upon what kind of instrument it is this error can be higher or lower. If these
are mechanical systems where because of the bad condition of instruments these errors
can be more if the instrument is designed such a way that the mistake should be minimal
then these errors are minimized. These errors may also occur due to incorrect adjustment
of the instrument. For example, in the last slide we saw that the needle was connected to
the spring, but that needle was very far from the tape. So, in that case the error will be
higher if we bring this needle very close to the tape no matter then how where the
observer is that error can be minimized. So, this can also occur due to incorrect
adjustment of the instrument and of course the computational mistakes are always there
25.8 can be written at 28.5. So, that is always there well a human is recording the values.
These errors cannot be treated mathematically because the origin of these errors are not
mathematical in nature while doing a mistake a mistake can be of any form and we
cannot actually treat these errors mathematically.

35
They are only can be treated by repeated measurements should repeat the experiments
multiple times. So, probability of having an error of the of the same magnitude at the
same point will be lesser. So, we perform multiple measurements and take the average
values of that. So, these cannot be eliminated, but can be minimized by repetitive
measurements. These happen because of the improper use of an instrument and this error
can be minimized by taking proper care in reading and recording measurements. Though
during experiments most of the people pay attention and do proper care in reading and
recording, but these errors can can occur and which leads to the gross error. In general
indicating instruments change ambient conditions to some extent when connected to a
complete circuit. All several readings at least 3 we say, but the more is good we perform
the same experiments multiple times and take the readings and we average them out. So,
let us say in one experiment a human error is there that will show up there if all the
readings are at one place and suddenly one reading only is misplaced or having higher
deviation from the other readings. So, that will show up and that can be corrected using
these multiple measurements typically 3, but more than 3 is good and these must be taken
to minimize the effect of ambient condition changes. The other type of error is systematic
error. Systematic error is due to the shortcomings of the instruments. So, instrument may
have problems inside. So, because of that there is certain error generation in the
measurement that is called systematic error. These shortcomings can be like defective or
worn parts, aging effects or the effect of environment on the instrument.

For example, we are measuring a voltage using a voltmeter. This voltmeter it has multiple
circuits inside those circuits age after a certain period 1 or 2 year the performance of that
circuit actually degrades and because of that there can be an error generation by this
voltmeter in measuring the voltage. So, this can be one example. Another is the voltmeter
let us say the voltmeter is analog which actually the needle is moving on the scale. These
moving parts have wear and tear over the time. So, if you use continuously this voltmeter
of 1 year, 2 year, 5 year after certain time there will be wear and tear in the system and it
will start showing us some drift with the actual voltage. So, these are some systematic
errors which happen because of the instruments because the instruments has some defects
inside and these systematic errors are of three types the instrumental error, the
environmental error and the observational error. These are the systematic errors. So, the
first is instrumental error. These errors are inherent to the instrument while measuring
instrument because of their mechanical structure or electrical structure whatever that
instrument is. These are inherent to that instrument. We took the example of voltmeter it
has internal circuit which degrades with time and it shows a drift from the actual voltage.
So, this is inherent to this error which is generated by this voltmeter is inherent to that
instrument to this voltmeter. So, this is a instrumental type of error. Another error can be
because of the friction in the bearings of various moving parts, irregular spring tensions
stretching of springs etcetera.

36
There can be multiple sources of the systematic error or this instrumental error and this
type of error can be avoided by selecting a suitable instrument for the particular
measurement application. For example, we want to measure precisely we may either
switch to the new voltmeter or we service this voltmeter find out what are all the errors
and we correct this voltmeter for those errors and we then use this particular
measurement system. We can apply correction factor that is also one way for example,
the voltmeter is costly and not easy to repair. Then these systematic errors we know how
much error is being produced. For example, because of the aging of this voltmeter 1 volt
actual 1 volt of signal is measured at 1.1 volt by this voltmeter. So, we always know that
actual voltage is 1 volt and voltmeter is giving 1.1 volt it means 0.1 is the error which is
generated by this voltmeter and this 0.1 volt of error can be subtracted from all the
subsequent measurements. So, this is one way of putting a correction factor by
determining this instrumental error. The third option is to calibrate the instrument.
Calibration of a instrument is when we have a known set of input and we generate the
output by the instrument and we manually tweak this measurement system or this
voltmeter to adjust the output to the desired output. So, we know the input output is
known and we tweak the faulty output to the correct output by doing this external kind of
adjustment this is called the calibration of the system. The second is environmental error
due to external conditions affecting the measurement including surrounding area
conditions such as change in temperature, humidity, barometer, pressure etcetera. So,
temperature is a very very temperature is a surrounding parameter which impacts almost
all the circuits.

If the temperature rises the performance of electronic circuits generally goes bad. So,
every 10 degree change in the temperature every 10 degree rise in the temperature
typically change that current of devices by a certain amount. This amount is practically
known by our measurements. So, if the temperature is changing in the surrounding this
causes a systematic error in the measurement of instrument. If there is a significant
change in the temperature then our error will be more and this error need to be corrected
using look up table or some other approach. To avoid this kind of error we use air
conditioner with wherever it is possible. We want to maintain the external temperature at
the same point no matter what outside temperature is we want to keep all the temperature
in the laboratory typically to room temperature which is 27 degree. So, this can be used
by using air conditioner. Sealing certain component in the instrument this is one way of
protecting these internal circuitries from interference from temperature change from
humidity and so many other environmental factors. We kind of put epoxy around these
components and put inside the instrument so that these parameters does not impact the
performance of these circuits

We sometime use magnetic shields magnetic shield is used to cut down the interference
because of the RF signals. If we have a mobile phone or a some radiating antennas

37
nearby these antennas generate EM waves which interfere with the electronic circuits. So,
we put a magnetic shield to prevent this interference and to reduce the error. So, this is
second types of error which is environmental error. The third is observational error. So,
this is introduced by the observer the human being who is taking this measurement and
this is most common is parallax error which we discussed earlier that is spring has a
needle connected and which is moving on this tape. So, that is kind of parallax error and
estimation error there is another kind of error while reading the scale. So, while reading a
human being can make the mistake these are most common type of observational error.
For example, an observer who tend to hold his head too far to the left while reading the
position of the needle on the scale. So, this was a observational error. So, these are the
third type of systematic error. The next type of error is random error. As the name says
these are random in nature. So, due to these are due to unknown causes we do not know
exactly what cause this kind of errors, but these errors exist in the measurements.

These occur when all systematic errors has accounted. So, for example, we put a
temperature sensor outside the room and the outside the environmental temperature is let
us say 30 degree C, but when we actually measure this the output of this sensor this never
comes 30.0 it always comes with some error. There is a we see the signal generated by
the sensor this is always fluctuating a little bit, because there are certain unknown factors
which are affecting the performance of the sensor. Because of that there is a error
generated in the measurement of temperature. So, these are called random errors and the
source is unknown. There can be more than one sources for example, humidity for
example, the radiation from the sun that also impacts the performance. The other thing
can be like surroundings, the shielding there are so many factors which can contribute to
cause this random error. And this is generally accumulation of small effects. There are
multiple parameters which are affecting this error and this random error is generally the
accumulation of all those effects required at high degree of accuracy.

This can be avoided by increasing number of readings. We increase the number of

readings and then we average it out. So, that is one way of reducing this type of error or
we use statistical means to obtain best approximation of true value. So, we perform
certain standard deviation or the Gaussian distribution kind of analysis on the data to get
the desired output which is estimating the error how much the error is possible, but it is
not possible to completely remove that. All these errors we will discuss in detail. So, the
difference between these random and systematic error is these random errors results from
random effects in the measurement. The magnitude and sign of a random error changes
from measurement to measurement. So, this random error can be positive for one
measurement and can be negative for another measurement. The magnitude as well as
sign can also change and measurements cannot be corrected for random errors. A
measurement if it has random errors these measurements cannot be corrected for these
type of errors. However, a systematic error results from systematic effects which we

38
know what kind of parameters can impact my accuracy and the magnitude and sign of
systematic error is constant from measurement to measurement.

For example, we measure a length and the error is plus 2 for example, 100 centimeter
plus 2 centimeter this plus 2 centimeter will always be in the positive direction. For
example, this will always be 100 plus 1 centimeter and 100 plus 3 centimeter it will never
be 100 minus 1 centimeter the systematic error. The magnitude and the sign of this error
will be constant and these kind of errors can be corrected because we know how much
the error is generated by the systematic sources. So, we know these values and this whole
of the measurement can be corrected for the systematic errors. So, as we discussed last
time that we discussed accuracy and precision where we see this graph we have
probability density function here and on the x axis we have the values. Our true value is
here which is we are calling the reference value. Let us say this is the value of actual
parameter we are measuring let us say the temperature the temperature is exactly at 30
degree. So, this is the true value or the reference value, but when we measure it using our
sensor this 30 degree is not a 30 degree, but it is somewhere like 32 degree or 35 degree.
So, there is a difference in the measurement between the true value and the actual value
we get from the sensor. This difference we called accuracy which is 35 minus 30 this is
the error and when we measure this temperature multiple times this follows some sort of
a Gaussian kind of distribution. Sometime when we measure it the error comes in 35.1
degree the next measurement it may come 34.9 which is left side of this bar again 35.2
sometime on the right side or on the left 34.8 degree. So, if we perform multiple
measurements this is going to follow a Gaussian kind of distribution and this is called the
precision.

The precision means the difference in the multiple readings should be smaller. So, the
narrower this graph is the more precise the instrument is this is what we discussed last
time. So, now we define accuracy and precision in terms of error. So, this is the case that
what we discussed where we have precise and accurate measurement this is precise and
accurate where the measurement is precise right at the true value and accurate as well
here. If we talk about in the random and systematic errors the random error is also small
and the systematic error is also small in this [Link] we talk about this case B where we
have precise, but not accurate precise, but not accurate means the all the values what we
measure they are very tightly bound no measurement is overshooting or undershooting
compared to the rest of the values. So, they are all in a very tight space and that we call
the random error is small. So, here we see this random errors which are generated in the
instrument these are small it means these are precise the instrument is precise when the
random error is small. However this is not accurate because the true value is right at the
center and all the values are different than this it means there is a systematic error
existing which is large here. So, this error is systematic error which is large and this is
corresponding to not accurate. So, the accuracy you can say is caused by systematic error

39
and precision is controlled by the random error. Other two cases are like random error is
large and systematic error is small it means this is not precise, but accurate and this fourth
case where this random error is also large and the systematic error is also large it means
this is neither accurate and nor precise this is not accurate and not precise. So, these
random and systematic errors actually define whether the instrument is accurate and
precise or not. So, now let us have a look about the Gaussian distribution of these errors.
When we perform these multiple measurements on a single instrument all the outputs
generated by this instrument they form a sort of Gaussian distribution means none of the
output will be very far from the true value.

Probability of the outputs will fall in and around to the true value and there will be some
outputs which can be little bit far and that depending upon the distance from the true
value their probability of finding at that place will be lesser. For example, we have this
Gaussian distribution here which is the true value here which is let us say 1 and we are
performing these experiments let us say 100 times then most of our measurements for
example, here we take case of 68 percent these 68 percent cases will be falling within this
range plus minus 1. So, if we are taking total 100 measurements 68 experiments will
show a value which are closer to this between this and this only some 100 minus 68 will
be this side or this side will be away from this region. So, the Gaussian distribution is
when we have Gaussian type of distribution of these error or the uncertainty we start
defining the error as x plus minus sigma where the x is the base value and the sigma is
the variance where we call sigma as variance and between this plus minus 1 sigma let us
say we have x plus minus 1 sigma in this range this is sigma this is x, x plus minus 1
sigma in this range 68 percent of the values will be falling. If we take sigma x plus 2
sigma the range increases from here to here and we cover more values for this range. So,
we expect the true but unknown value of x to be an interval given by x minus sigma and
x plus sigma. So, majority of the outputs will be between x minus sigma and x plus sigma
between this range our 68 percent of the values will be falling if the error follows a
Gaussian distribution. So, for example we have a line here which we have drawn let us
say here and I ask you what is typically the length of this line let us say in centimeter. If I
ask 100 students to tell me the length of this line most of them will say let us say 6
centimeter but most of the students say 6 centimeter less number of students will say it is
like 5 centimeter or 5 to 4 centimeter or 6 to 7 centimeter. Very less students will say that
this is a 1 meter long but there can be 1 or 2 outliers who will say it look like a 1 meter
long. So, this is kind of a output we are generating where we have these errors which are
in the measurement. So, the area value is let us say 6 centimeter but we have variance in
the measurement. So, the area under the normal curve between this line and this line is
this much 0.68 and depending upon the number of the number n we always write mu plus
minus 1 sigma means this is the value mu and this is sigma between plus minus 1 sigma
we have 67 percent or probability of exceeding mu plus minus n sigma is 0.5. If we have
mu plus minus 1 sigma around 32 percent only will be the outliers 68 percent will be all

40
within this curve. If I extend this n to 2 it means I am talking about mu plus minus 2
sigma. If I increase my range of acceptance then my error is then my range is from mu
plus minus 2 sigma and I cover this whole of the graph and only 5 percent is the outliers
5 percent outliers. If I take n equal to 3 then only 3 percent outliers will be there in my
measurement. So, if I keep on increasing my sigma values then I am covering more and
more of outliers into my measurement.

So, if the variance or the sigma magnitude is less then mu plus 2 sigma will also be small
and in 2 sigma I am covering almost all the measurements except 5 percent only 5
percent will be outlier to this and 95 percent will be covered in the measurement. So, it is
very important to have a variance which is less. It is very unlikely that a measurement
taken at random from Gaussian pdf will be more than plus minus 3 sigma because it is
just the number is only 0.3 percent. So, most of the measurements will be covered if we
consider mu plus minus 3 sigma range only 0.3 percent will be the outliers. Then we
calculate the most probable value of these in the measurements. So, most probable value
is it is also known as the arithmetic mean or the average value which is nothing but the
summation of all the values divided by the number of values and the MPV is the sum of
all the measurements divided by total number of measurements. Standard deviation sigma
also known as standard error or the variance which is sigma square equal to sigma M
minus MPW MPV 2 square divided by N minus 1 where M minus MPV is referred to as
the residual M is the actual value minus MPV is the most probable value which we
calculated last slide. The sigma is computed by taking the square root of this above
equation.

So, this is sigma square we take sigma as under root of this. So, this is our variance. Let
us take an example of this kind of measurement. So, we have a distance which is
measured repeatedly in the field and we got these 4 values one is 31.459 meter, 31.458
meter, 31.460 meter and 31.457 meters. Now we want to compute the most probable
value of the measurement and the standard error for this data. So, this is how we calculate
these are all those 4 values 1, 2, 3, 4. So, first we calculate the MPV or the M bar which
is the sum of all these value divided by the number of readings. So, we have this reading
1 plus 2 plus 3 plus 4 the sum is 125.834 divided by 4 is equal to this is the MPV value
31.459. Then we calculate M minus M bar which is the individual residual this reading
31.459 minus MPV which is 0 this minus MPV this third reading minus MPV and the
fourth reading minus MPV and then we take the square here in this graph in this column.
Then the standard error or the variance we can calculate using this M minus M bar whole
square divided by N minus 1 which comes out to be 0.001 meter. So, the most probable
value of the measurement is 31.459 and the value that is most likely to occur when you
perform multiple measurements on the same field this is the value which most probably
will come as a output. This value represents the peak value on the normal distribution
curve. So, what distribution curve we saw there this is the peak value here of all the

41
measurement let us say our 4 measurements over somewhere here and this is the peak
value which is 31.459 this is the peak value. The standard error is 0.001 so it means 68%
of the values would be expected to lie between this MPV which is 31.459 plus minus
0.001 this is the variance or let us say this is 31.459 plus 0.001 and this is this value is
31.459 minus 0.001 between these values 68% of our mean measurements would fall. If
we measure the same field 100 times 68 measurements will fall in this range of length
these values are computed using the MPV and the standard error. So, this is how we
measure the errors. Now the next important point is the propagation of error when we do
these measurements we definitely come up with some error in all the measurements.

Then these errors actually propagate to the final output for example, we have some
intermediate steps of measuring for example, we have voltage measurement how do we
do a voltage measurement we do a current measurement and a resistance measurement I
and R and then we multiply this I and R to calculate V the Ohms law V equal to IR. So,
measurement of I has its own error because this is a separate measurement and
measurement of R is also a separate measurement so it also has its own error. So I has an
error R has its own error these both errors will actually contribute to the error in voltage
which is I into R. So, these errors generated by us at the first place they actually
propagate throughout the system. So, the propagation of error is if we have X and Y
which are independent random errors delta X and delta Y so X is given something like X
plus minus delta X where X is the base value and delta X is error and Y is also given Y
plus minus delta Y where delta Y is the error. Then the error in addition of these two
parameters if we have third parameter Z which is addition of X and Y X plus Y then the
overall error will be given by del Z equal to under root X del X square plus del Y square
means the square root of under the square root variance in X square plus square of
variance in Y that will be the total variance in Z. This is the one rule for means
propagation of error. The second is the same if we have X is given as X plus delta X Y
plus delta Y and now the output Z is like X into Y X is multiplied by Y this is exactly
same as V equal to IR the Ohms law. We have error in I we have error in R then how
much will be the error in V. So, the formula for calculation of this type of error is del Z
upon Z where del Z is the error in final output this Z is the base value of final output
equal to under root del X upon X to whole square plus del Y upon Y to whole square.

This is the formula if we have two variables which are in multiplication earlier they are
here they were in addition here they are in multiplication and one more important thing is
it is X plus Y and X minus Y also calls same error X plus Y and X minus Y both are
same for error calculation. The third rule is if Z is some function of X then del Z or the
variance of Z is derivative of FX which is F dash X into variation of variance of X which
is del X. So, this is third rule and the fourth rule is if we have more than one parameter
which can cause error for example, there is an error in current measurement and there is a
error in the resistance measurement then how the error will be propagated. This is given

42
by this formula sigma Q square with Q is the function here Q is a function of X and Y
where this X and Y can be correlated or they can be uncorrelated. So, this is the
generalized formula where we have variance of X square into partial derivative of Q with
respect to X square plus sigma Y square into partial derivative of Q with respect to Y
square plus 2 sigma XY dQ by dx into dQ by dy. Most of the time during these
measurements these two parameters which are causing error they are most of the time
uncorrelated. Uncorrelated means if one parameter is causing some error this generation
of error does not impact the error generated by the second parameter. So, they are
uncorrelated. So, in this uncorrelated case this sigma XY becomes 0 or this whole term
goes away. So, we have the overall formula for uncorrelated cases this. We will see some
examples here for this. For example, we have to find the volume of a certain cube you
measure its side as this. So, the side is given which is let us say s of this cube side is s
which is given 2 which is the base value of the side 0.02 which is the error in the
measurement or the variance.

We need to find the error in the volume. So, volume is given by s cube s into s into s. So,
that we can apply this rule number 3 the various variance in V will be equal to the
derivative of s3. So, del V will be d by ds of s cube into delta s. So, this is what we
calculate 3s square into delta s which comes out 0.24. So, this will be the variance of
volume or the error caused in the volume if we get an error of 0.02 in the side
measurement the total 0.24 will be the error in the volume measurement because volume
is not directly measured volume is calculated from the side measurement. So, the actual
volume will be 8 plus minus 0.2 we can neglect this 4 here as well. Another example is
suppose we want to measure 3 numbers as follows x is 200 plus minus 2 this is the error
in x or this is we can call it delta x this is delta y which is error in y and this is del z
which is error in z.

Now, the expression for the output Q is x upon y minus z. So, this is the function x has its
own error y has its own error z has its own error. Now, we want to calculate how much
overall error will be there in Q. So, let us say d equal to y minus z this denominator we
write separately y minus z and in the addition and subtraction the error is delta x square
plus delta y square and the root of that. So, this comes out to be 10 plus minus 2 under
root 2 this root 2 is because of that under root del x square plus del y square the rule
number 1. So, the error in the denominator is 10 plus minus 3 which is del d I can call it
then the overall error in Q will be x upon d and now we have this in x upon y. So, this is
the formula for calculating the error and it comes out to be 20 plus minus 6. So, this is the
overall error in Q which is 6 if we have individual errors as 2, 2 and 2. Another example
is let us say we want to calculate a power in a electric circuit. So, power is given as P
equal to I square R and I and R has its own individual errors I is 1 plus minus 0.1 which
is let us say del I can call it which is the error in current measurement and resistance also
has its own error 10 plus minus 1 ohm. So, this 1 ohm is the error in the resistance

43
measurement. Now the base value of power will be 10 which is I square R the base value
is I of I is 1, 1 square into R the base value of R is 10 so this is 10. So, the base value of
power is 10 watts. Now we want to calculate the error produced in power if we have
individual errors. So assuming I and R are uncorrelated means the error in I does not
produce any extra error in R and while measuring R error in R does not produce any extra
error in I.

In that case that sigma xy terms goes away which is 0. Now the formula is sigma P
square the variance of power equal to sigma I square into del P by del I to whole square
plus sigma R square del P by del R to whole square at R equal to 10. The partial
derivative of P with respect to I which comes out to I R plus sigma R square into I square
the whole square. We can now put the values in this formula comes out to be 5 watt
square which is sigma square. So, the sigma or the variance of the error in power is under
root 5 typically 2.238 will come so we neglect all these decimal values and the final
power will be 10 plus minus 2 watt. So, the base value of power is 10 with a 2 watt of
error or inaccuracy. So it means when we say power value we take 10 plus minus 2 watt
all of our 100 measurements 68 percent will be in this range 10 plus minus 2 watt. If we
perform 100 measurements 68 will be between 10 plus 2 which is 12 and 10 minus 2
which is 8. So, in this range if we perform 100 measurements on this instrument 68 times
this instrument will show the number between 8 and 10 because the N we are talking here
is 1. So, this is the some example of the propagation of error this is what we have for
today.

Thank you.

44
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 05
Thermal Sensors: Equilibrium and Predictive measurements

Hello, welcome to the course Transducers for Instrumentation. Today we are going to
discuss a new topic the Thermal Sensors. As the name says thermal sensors means we are
trying to measure the temperature of a certain body. The temperature of a body is the
measure of how much thermal energy a body has. So, while measuring the thermal energy
or the temperature we need to discuss certain points. The very first thing is whenever we
measure the temperature of a body which we assume is in equilibrium the first body is in
equilibrium and as soon as we connect our sensor to measure the temperature we are
disturbing that equilibrium.

So, body was in equilibrium let us say at 50 degree centigrade it was uniform throughout
the time as soon as we connect a sensor to measure its temperature no matter how small
the that temperature sensor is this is going to disturb that equilibrium. How it happens? Let
us say we have a thermal body. So, we have this ok. So, we have a body which is at certain
temperature let us say TB this temperature is in equilibrium at time equal to 0.

As soon as we connect a second body to this , this is in close contact with this body this is
a point of contact. This is the point of contact as soon as we make this connection there is
a certain amount of thermal energy which transfers from the body to the temperature
sensor. This is our second body which is a sensor and initial let us say temperature sensor
is TS. So, because of the thermal difference TB minus TS there is going to be a transfer of
thermal energy. So, this measurement actually requires certain amount of energy which
transfers from body to sensor.

No matter how small it is that depends on what is the temperature difference and what is
the thermal capacity of the second body as well as the first body. So, it depends on multiple
factors, but as soon as we connect two things together which are at different temperature
there is going to be a energy exchange no matter how small it is, it is going to disturb the
equilibrium. So, this sensor is going to warm up or cool down depending upon the
temperature difference. If the TB is lesser than TS then the sensor temperature will go
down and if temperature of body is greater than the temperature of sensor then the sensor
temperature will go up. So, this rise or fall in the sensor temperature is dependent on the
heat exchange whether heat is flowing from the body to sensor or from sensor to the body.

So, there is a heat exchange involved which is disturbing our equilibrium. Now how small
this sensor is no matter how small it is, it is going to disturb the equilibrium because this is

45
let us say very big body with thermal capacity very high this sensor is a small, but having
a thermal capacity which is very less, but still there will be a thermal exchange no matter
how small this sensor is and it is going to disturb the equilibrium. And this happens in all
the modes of conduction. For example, here we have a metallic body here and a sensor
which is physically in contact. So, we have conduction form of heat transfer, we can have
a convection where kind of air is flowing from higher to lower temperature.

We can have this equilibrium disturbance even in radiative sensor. For example, a body is
radiating and there is a sensor which is far apart which is taking the measurement of this
radiating body even in that case the equilibrium will be disturbed because at time t equal
to 0 or at the starting of measurement we assume that the body is in equilibrium with its
surrounding. So, if a body is radiative let us say we have a body here which is radiating,
this is right now in equilibrium with the surroundings as soon as we bring the temperature
closer to that we bring the temperature sensor closer to this body. This is let us say our
probe of temperature sensor as soon as we bring it closer to this body, this is going to
disturb the equilibrium because the current environment is not equal to the environment
which we had earlier. The surrounding is disturbed so, the equilibrium is also disturbed.

So, this equilibrium disturbance this happens even in the radiative sensors and this happens
in conduction and convection already. So, this means whatever we do whatever
measurement we are doing for the temperature sensor there is inherently there is an error
this is inevitable which we discussed even in error propagations. This small amount of error
or big amount of error depending upon amount of heat exchange this error we cannot
eliminate completely we cannot make it 0. So, we have to minimize this error in
temperature measurement, and this is the engineering task how to minimize this error
during the temperature measurement and we apply multiple techniques to do this. So, now
depending upon the heat exchange we have two methods of measuring the temperature.

First is the equilibrium-based temperature measurement and the other technique is

predictive temperature measurement. So, we have equilibrium and the second is predictive.
In equilibrium method a thermal equilibrium is reached between the sensor and the body
and we wait enough amount of time so, that this equilibrium is reached and the temperature
readings are stable. It means the heat exchange between the body and the sensor is
negligible it may not be 0 because this is exponential function and tte he steady state value
will come only in infinite time. Practically there is a exchange of heat which is negligible
after let us say 5 or 6 time period.

So, we wait enough amount of time so, that this equilibrium is reached a the measurements.
So, in equilibrium measurement in this technique we wait long enough for equilibrium to
achieve. So, we have to wait for a long amount of time depending upon the RC time
constant of thermal network. So, this is the disadvantage of equilibrium-based technique

46
we have to wait long amount of time to perform the measurement. However, the reading
what we get in equilibrium technique will be more accurate.

So, this is the disadvantage and advantage is this is more accurate. So, in equilibrium based
we have to wait for a longer amount of time if time constant is really high, but the accuracy
of the reading will be much better. The other technique to overcome this problem of time
is predictive based measurement. So, in predictive method the equilibrium is not reached.
So, we do not wait enough for the equilibrium to reach we just make the contact of sensor
with the body and depending upon the how fast the energy is being transferred from body
to sensor the temperature rise is there.

So, looking at the temperature rise and doing some mathematical modeling we can predict
what will be the steady state value of this graph because we know that the the charging of
this the rising of this temperature is an exponential function and in how much time and
how what will be the steady state value. So, we do some sort of a prediction to measure the
temperature the however, the system is still not in equilibrium the temperature has not
reached the sensor is still the reading is being updated, but we measure this graph for a
small amount of for a small range and based on this range we predict what will be the
steady state value. So, timing wise the predictive method is FAST which is an advantage
of this technique. However, this prediction is not very good because we are taking data
only for a small amount of range and because of that the extrapolation is not very accurate.
So, this predictive method not very accurate, this is a disadvantage.

In equilibrium this is advantage that this technique is more accurate, but it takes long
amount of time which is a disadvantage. So, these are the two techniques we apply in
temperature measurement. So, now let us discuss how these techniques affects the accuracy
of temperature measurement. So, for that we need to first make a model of our temperature
sensor, the whole system we where we have body the sensor connected to it. So, we want
to make a model of this whole system.

So, for example, we have a body which is at temperature TB, this is our object. This is the
object we want to measure the temperature. Now we connect a sensor to it which is
generally small compared to the size of the body or the object. So, this is our sensor, this
sensor temperature we can denote it by TS, the object is TB. And generally this sensor is
always connected to some external electronic circuitry or external world.

So, we will have a cable or something which is connecting this sensor to the outside
electronics or we call it an environment and this environment is at temperature T0. And
this environment temperature we consider it as a boundary condition that this temperature
does not change, remain fixed with whatever the system happens within the system. This
T0 temperature will always remain same. Now this is the complete system where we have
just connected this sensor to the object and trying to measure the temperature this TB and

47
the initial temperature of the sensor is TS. So, now, we can make an equivalent model of
this whole system.

So, we have this object TB in between we have a sensor which is at temperature TS. Now
this temperature sensor has its own thermal capacity. So, what is the thermal capacity?
Thermal capacity is the amount of heat to be supplied to an object to produce a unit change
in the temperature. So, we have a object to raise a temperature by 1 degree centigrade of
this object how much energy we need to give it to this object. So, that is the thermal
capacitance of this object.

It is simply the product of mass into the specific heat of this of this sensor. So, we have this
sensor which is here. Now this has some thermal capacity which we can represent using a
capacitor which is connected to ground and this capacitor is analogous to electrical
capacitor where in electrical capacitor we store charge here in this thermal capacitor we
store thermal energy. So, this this is the thermal capacity of sensor. Now this sensor is
connected to this body using a certain resistance.

Let us say this resistance is R1 this is the thermal resistance. So, thermal resistance is
analogous to electrical resistance where the electrical resistance is kind of stopping the
current from flowing when we apply a voltage across it the higher the resistance the more
it stops the current to flow. The same function applies here we have this thermal resistance
R1 which is the resistance between the body and the sensor temperature and it depends on
multiple factor first of all is the surface of contact. So, when we are making this contact of
this sensor with this body let us say we have this sensor and this is connected to this body.
Now we can see only a very small portion of the sensor is connected to the body not not
all the surface of sensor is connected to body.

It means the thermal energy can flow only through that contact area which is which is in
contact with both the object and the sensor. So, the cross section through which the energy
can transfer is very less. So, accordingly the resistance is very high. So, this is analogous
to electrical terminology. So, this is the thermal resistance which is between this object and
the sensor.

This is our sensor and we have this node here. Now this sensor is also connected to a
external environment. Let us say we call it thermal resistance R2 and here this is connected
to environment which is fixed. The temperature of environment we consider it the fixed at
T0 degree. So, we may be represent this like a rigid wall sort of thing. So, we have this
environment which is fixed. This temperature sensor the temperature of this temperature
sensor is rising when we are connecting it to body. Now this rising temperature is also
different than the temperature of environment. The environment temperature is T0 and
sensor is Ts. So, there is a difference between these two temperature.

48
Whenever there is a difference in the temperature the heat energy will flow. The heat will
flow from higher temperature to lower temperature. So, assuming that sensor temperature
is high and it is being rising this sensor is going to lose some energy from the sensor to the
environment. And how fast that energy will be transferred from sensor to environment that
depends on the contact resistance of this sensor and the environment which is R2. So, now
we have two thermal resistances one between the sensor and the object which depends on
the contact area how well connected this sensor is with the object R1.

And we have second thermal resistance R2 how isolated this sensor is from the
environment whether it is losing some energy to the environment or it is not losing. So, we
have these two thermal resistances after the movement of coupling when we connect this
sensor physically to the object that is the movement of coupling. So, after this movement
of coupling or after the object temperature actually changes the sensor temperature is TS
and the sensor is losing energy from TS to T0 because these are different temperature.
Almost never these temperatures are equal because the amount of heat energy which is
raising the temperature of sensor this also depends the difference between the temperature.
So, the higher the difference the more energy will flow and raise the temperature, but when
both the bodies come very close into the same temperature the amount of heat transfer will
be lesser because it depends on T1 minus T2.

So, lesser energy flow means the rate of change of temperature will be less. So, this follows
some sort of exponential behavior where we have this is time and this is the temperature
of sensor. Initially because the TB is very higher than TS the rate of change will be very
fast, but as the time passes and TS is now very close to TB then the rising temperature of
the sensor will be very less and if we see here there will be certain difference because this
is this is an exponential graph and it will take infinite amount of time in reaching
equilibrium means when both the temperature becomes same. So, this is not possible to
achieve this equilibrium in theoretical ah time limits we practically assume after certain
point after long after waiting for so long the system again becomes into equilibrium where
there is a heat flow between TB minus TS which is charging this thermal capacitance. So,
there is a heat flow from TB to TS there is a heat flowing from here to here and the same
amount of heat is actually transferred from sensor to the environment when the both of
these rates are equal it means the system is in equilibrium and sensor reading is stable.

So, that is that condition we call it the equilibrium case when this whole system again
reaches into a thermal equilibrium. Now, when both of these rates are equal then we can
write the equation for them or maybe we can write here itself when both of these rates are
equal, we can write TB minus TS divide by R1 the difference between the temperature of
object and sensor divide by the thermal resistance R1 this rate is equal to TB minus T
naught divide by R1 plus R2. So, we are saying here this rate TB minus TS divide by R1
this rate is equal to the complete rate from object to the environment it means the sensor is
not storing further energy in itself and sensor has come into the equilibrium. So, the rate is

49
TB minus T naught divide by R1 plus R2 the total resistance in the path. So, this is the
equation we can write if both of these rates are equal and we can now rearrange this
equation to get TS equal to TB minus TB minus T naught into R1 upon R2.

So, this is the equation or we can further solve this and this comes out TS equal to TB
minus delta T into R1 upon R2. So, this is the equation we get for our system where the
sensor temperature TS is equal to TB the object temperature minus certain term which is
delta T R1 upon R2. So, this delta T will always be there we cannot control it much which
is the difference between the body and the environment which we do not have any control
environment temperature is fixed and the body temperature is what we are going to
measure. So, this is what it is. So, this delta T we do not have much of a control.

For accurate measurement what we need TS equal to TB when the sensor temperature is
equal to the object temperature that time the sensor will give the accurate reading. Our
desired condition is TS equal to TB and what we have this equation is TS equal to TB
minus delta T into R1 upon R2. So, now we can note two points here this R1 which is the
contact resistance the thermal contact resistance between the object and the sensor. So, this
temperature this thermal resistance need to be as small as possible.

So, we have R1 should be as small as possible. What do we mean by this statement R1

should be as small as possible. That resistance between the object and the sensor should be
as small as possible means there should be an efficient heat flow the thermal energy flow
from the object to the sensor. It should not be limited by the contact area of between these
two objects and other factors as well. So, the sensor should have a very good contact with
this body the whole of the surface of the object should be in very close contact with the
with the object then the resistance R1 will be very less. Even we can put some artificially
epoxy or some other materials to increase this contact area to decrease this R1 the contact
this thermal resistance.

So, we can decrease this R1 to make this term smaller delta T into R1 upon R2. We either
decrease this R1 or second thing we can do is R2 should be as high as possible. R2 should
be as high as possible. What do we mean the resistance the thermal resistance between the
sensor and the environment it should be very high. It means there should not be any heat
flow from the sensor to the environment.

The sensor should not lose any amount of thermal energy to the environment. It means it
should be very much isolated the sensor should be very much isolated from the
environment then we have R2 should be very high. So, when we make these R1 should be
very less and R2 should be very high this second term is going to decrease and we will
have TS as close to TB as possible by applying this R1 and R2 conditions. But these R1
and R2 are finite values these we cannot make this 0 R1 and R2 no matter how small R1 is
or how high R2 is this term delta T into R1 upon R2 will always be a finite number and

50
that will give us some error in the measurement and TS will never be equal to TB. So,
based on these two analysis what can we do during our measurement to make this error as
small as possible.

So, one way of doing R1 as small as possible is you embed this the sensor into the object
itself. So, we try to have more and more contact area between common between these
object and the sensor. So, one way is let us say we have this object what we considered
earlier which is a temperature TB this is object to measure. And we somehow dig a cavity
inside this. So, we have this object now we dig a cavity in this and we put the sensor deep
inside this object.

So, this is our sensor and we put some epoxy as well here and this is connected to the
outside world which is at T0. So, this is our sensor now which is buried deep inside the
object. So, that we can see now this sensor is covered almost from every side by this object
it means the heat can be exchanged from all the sides decreasing its thermal resistance R1.
So, this is one way of one way of making R1 as small as possible. So, this is this is one
technique of making R1 as small as possible.

We can do in multiple techniques this is engineering task actually now to reduce this
contact resistance. We can put some epoxy next to this next to this sensor. So, that because
of the surface roughness if the contact is not proper. So, we put epoxy in between. So, that
there is a close contact between these object and sensor.

Second we can do is we kind of put a spiral over here for example, this sensor is connected
to a external circuitry using this wire. This wire actually we put multiple turns here inside
this object only. So, that because this sensor is losing thermal energy through this cable
only because this is the connection between sensor and outside world. We are increasing
the length of this cable here inside this. So, that the resistance R2 becomes a little bit higher
and sensor does not lose much energy to the outside world.

So, this is one way of making R1 smaller. Next technique we can apply where we can make
R2 as large as possible. So, we are making R2 large. So, how we can do that? Let us say
we have this is the body or the object at temperature TB and we connect our sensor to this.
Now we do not want this sensor to lose any energy to the environment, but the problem is
environment temperature is fixed which is at T0 we cannot do much about it and the sensor
temperature is rising. So, there is a difference always between the sensor temperature and
the environment whenever there is a temperature difference heat is going to flow.

Whatever we put in in between these objects everything has certain amount of thermal
conductivity higher or lower that depends on material, but it is always there it is finite. So,
the sensor is always going to lose some energy to environment because of this temperature
difference. So, now we can apply a technique which can be used to reduce this error. The
problem here is the temperature difference the sensor temperature is different than

51
environment temperature. So, if we can somehow give a artificial environment to this
sensor where this local environment the temperature of this local environment is same as
sensor temperature then we do not have any temperature difference between sensor and
local environment and then there will be no heat loss or the heat energy will not transfer
from sensor to an that local environment.

So, this we can do like this for example, we have this sensor connected through this cable.
Now, we put an artificial cavity around this let us say this is artificial cavity this cup sort
of thing which is inside this there is some space and sensor is inside that cavity. So, let us
say we have this cavity now this cable comes out from this cavity and this cavity
temperature is also being tracked through other cable which goes to some instrumentation
amplifier also. Let us say we have amplifier coming in which is plus and here it is minus
and then it is some sort of driver circuit which is controlling this. So, we have this cavity
which is containing this sensor the temperature of this cavity also can be controlled using
this driver circuit.

This driver is controlling the temperature of this cavity and here we have this comparator
plus and minus plus comes from the sensor and minus comes from the cavity. Now this
comparator is tracking the sensor temperature and bringing the cavity temperature this is
cavity. So, this comparator tracking the sensor temperature and adjusting the cavity
temperature so that they are matched whatever is the sensor temperature at that point the
cavity will be heated up or cooled down so that there is no temperature difference. So,
when there is no temperature difference between this sensor and this local cavity so there
is no heat exchange between these two objects so this heat exchange is not there. So, in
that way we can reduce this heat exchange between sensor and the environment and we
bring down that error by the error factor delta T R2 by R1 upon R2 to the minimum
possible.

So, this is one way of making R2 large. This thermal shield which is here which we have
made artificially this is usually made by metals or some aluminium which the metal should
have very good conductivity so that we can adjust the temperature of cavity as soon as the
temperature of sensor is changed. So, these are the techniques where we can minimize this
error R1 upon R2 into delta T by applying certain techniques. Now let us consider the
dynamic case. The dynamic case is when you connect the sensor to the object and the
temperature of sensor start rising and it has not reached equilibrium yet. So, let us consider
the dynamic case where the temperature is still changing with time.

So, now first we need to make some assumptions to simplify our calculations and these
assumptions we will relax later on as we go along. So, the first assumption I am making
here is the thermal resistance between the sensor and the environment is very large means
the R2 the amount of R2 is very high the sensor is not losing any energy to the environment.

52
So, this is the first assumption I am making. The thermal resistance between sensor and
environment is infinite.

R2 tends to infinity. So, this is the first assumption I am making to solve my system that
the R2 is infinite the sensor is not losing energy to environment. The second assumption I
am making here is the object temperature does not change when I connect the sensor to
this. It means the size or the thermal capacity of The object is very very high compared to
the sensor. The sensor thermal capacity is very small or the sensor is very small in size
compared to the object. So, this is the second assumption object temperature does not
change after sensor is attached.

So, these are the two assumption I am making before proceeding to solve the dynamic case.
Now, I can write the heat transfer. This I can write del Q equal to alpha into TB minus TS
into del T. So, this is the heat transfer equation and the same thing I can write in terms of
the thermal capacity of the object as M into C into del T. So, these are the two equations
for heat transfer where this C is the specific heat and M is the effective mass.

So, when I put both of these equations equal to each other then we have alpha TB minus
TS into del T equal to MC del T or we can define a time constant which is tau T equal to
MC divided by alpha or M into C into R1. So, these are all the constants this M C and alpha
these are all constant we can put all of these constants together and call it with a different
name which is the thermal time constant of the system and this is very much analogous to
the RC time constant we have in electrical system. So, this thermal time constant exactly
the calculations are same just like the RC network within for example, in one thermal time
constant temperature reaches almost 63 percent of the initial gradient. So, this time constant
is exactly same as electrical time constant we have. So, now coming back to this equation
we have alpha TB minus TS into del T equal to MC into dT we rearrange the terms and
putting this time constant we have del T divided by TB minus TS equal to del T divided
by tau T which is the time constant or you can write the final equation which is TS equal
to TB minus delta T into e to the power minus T upon tau T.

So, this is the complete equation of the system assuming we have made two assumptions
after those assumptions we have this is the equation for the system. Now we can plot this
equation the x axis is time and this is very very much similar to the RC charging equation,
but on the x axis we have time and the y axis we have temperature here instead of the
potential and the final the temperature of the body is TB which the sensor is going to
achieve in infinite time which is TB and the actual rise in the sensor temperature is
something like this exponential. This will achieve in infinite time, but here now we can
define the thermal constants which is tau T. So, this graph says that sensor temperature
achieves almost 63 percent or 0.63 of TB in one time period. In one time period sensor
temperature will rise to 63 percent of the initial temperature difference and of course, in
second in two time period it will reach certain amount and in third in fourth in a almost 5

53
or 6 time constant it will practically achieve the temperature of object this is the
temperature change of the sensor. Now we study a more realistic case when we start
removing those assumptions what we made earlier we made two assumptions. So, first now
we are going to remove let us remove the first assumption which is that the resistance
between the sensor and the environment is not infinite it is has certain finite value it means
instead of achieving a final value of TB the sensor is never going to achieve that final value
TB because there will always be a heat exchange between the sensor and the environment
which will keep this temperature always lesser than TB. So, after removing first assumption
carbon tends to infinite this is the condition we are removing it means the sensor steady
state value will not be the object temperature TB, but somehow to a lower value than TB.
So, let us say this is TB which is the body temperature now because sensor is losing energy
to environment it will reach to a steady state temperature to a different temperature which
is not exactly TB, but let us we call it TB1 where TB is not equal to TB1 both the
temperatures are unequal.

So, there is there is a difference between the final value of the sensor. So, this happens
because of the finite value of R1. Now let us we remove the second assumptions as well
which we made earlier. So, after removing second assumption which is the object is not
very big and the sensor is not very small. So, if they are comparable or at least not negligible
then as soon as we connect the sensor to the object, object has certain thermal energy stored
within as soon as this thermal energy transfers to sensor that there will be a certain amount
of dip in the temperature of object because certain amount of energy is transferred to some
to the sensor.

So, if we plot the response after removing this assumption. So, this is TB and because of
the first assumption we have different value TB1 the final value of sensor. Now as soon as
we connect the sensor there will be a certain dip in the characteristic of sensor voltage
change and this is because of the finite mass of object. So, instead of having a smooth curve
now we have a dip in the sensor temperature because of the finite mass of object or the
sensor size is comparable to the object. So, here we considered the dynamic case where
when we connect the sensor to the object what will be the time response of the system.

So, this is all for today.

Thank you.

54
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 06
Thermal Sensors: Thermocouples

Hello, welcome to the course Transducers for Instrumentation. Last time we were
discussing about the thermal sensors. These are the sensors which measures the thermal
energy of an object in form of temperature and we saw that we have two kinds of
techniques to measure one is predictive based technique and one is the equilibrium based
and we saw some advantage and disadvantage of both the techniques. So, now we will
discuss today about different type of thermal sensors. So, as we discussed the temperature
is the measure of average kinetic energy of the molecules of a gas, liquid or solid. So, we
have an object which has its thermal energy stored in it.

So, the average kinetic energy of the molecules that is what we measure using these
thermal sensors and a very well known thermal sensor is the mercury thermometer as
almost everyone has seen this. This is nothing but a glass tube in which we fill up the
mercury which is a liquid metal and we expose this to the temperature and as soon as that
we connect it to the heated body because of the thermal energy exchange the temperature
rise in the mercury metal this expands and because when it expands it moves up in that
glass tube and we have different marks on this through which we measure the
temperature. So, it is a very old technique of having of making these thermometer this is
nothing but the glass tube we fill it with liquid metal like mercury and when we expose
this whole thing to temperature which is T b the body temperature which is higher than
the sensor temperature or the mercury T b is higher than T s then this mercury expands
and we have different demarcation different marks on this glass tube. This is how we
measure the temperature using this mercury thermometer, but it is not suitable for
multiple applications.

For example, the very first thing is there is a glass tube involved which is very fragile
many a times when in some rough environment you cannot use this kind of thermometer
it is subject to break down mercury is also not a very good element to use in these
mercury thermometer it is not very good for health. So, it has multiple limitations where
we cannot use this mercury thermometer. So, we have multiple different kinds of thermal
sensors available one is the thermocouple, second is resistance thermometer or RTD we
call it in short, third is the thermistors and the fourth is some novel semiconductor based
sensors. So, these are the type of sensors which we are going to discuss. The first one is
thermocouple as the name says thermocouple means it has one pair of metals different
metals metal 1 and metal 2 we join them together these two metals are in the form of wire

55
we take these wires twist them at their end we make this kind of couple and that is what
we call it thermocouple.

These are the sensors composed of two different materials here I want to put material not
metal. So, it is not that the concept what we are using here is not limited only to metals it
can be two different materials one may be the silicon one may be some other material
need not to be metal only it works in semiconductors as well. So, the basic architecture of
thermocouple is shown above we have two metals one is this wire which is metal 1 and
this bottom wire which is metal 2 and we join them together here physically they are
connected here and this is our sensing arm and the other side we have these two wires
here we can connect some multimeter or so voltmeter let us say this voltmeter is going to
measure the voltage developed across it. So, when we have this kind of assembly and we
subject this assembly to two different temperatures one is the reference temperature we
call it cold reference temperature and one end is at some hot sensing arm. So, when there
is a temperature difference between this end and this end of these thermocouple then a
voltage will be build up across these two materials this voltage is linearly proportional to
the temperature difference between cold arm and the hot arm.

So, this voltage is generated when there is a temperature gradient between the hot
element and the cold reference junction. This change of voltage is linearly proportional to
the change in temperature for a quite wide range the voltage developed is very linear to
the temperature difference in the hot and cold junction. And the amount of the magnitude
voltage or EMF generated is proportional to or is dependent on the metals used. What
materials or metals we use that depends the magnitude of EMF depends on those
materials. These thermocouples have multiple applications for example, plastic injection
molding machineries, food processing equipments, dicing, semiconductor processing,
heating elements, heating treatment, medical equipments, industrial heat treating,
package equipments there are multiple applications where we use these thermocouples
currently.

And some advantage of and disadvantage of thermocouples are the advantage of

thermocouples are these are very simple in nature we just need to have 2 wires join them
together in this is what a sensor we have made. So, this is very simple in structure very
rugged high temperature operation is just limited by the melting point of materials. So,
we can subject to higher temperature, but the metals what we use that should not be
deteriorated by this temperature this is what the only limitation otherwise this
thermocouple is has a very high temperature range. It is a low cost it is no resistance led
wire problem this we will discuss in some later slides point resistance testing. So, when
we make this thermocouple we twist both the metals wire together we make a sensing
node.

56
 So, this is a kind of point we have which we use to measure the temperature. So, it is a
point measurement sensing this has quite fast response to temperature changes if there is
a temperature change at hot junction it detects quite fast. The disadvantage of
thermocouples are these are least stable and least repeatable. So, if you do the same
measurement using the same equipment 100 times the readings are not very reliable and
repeatable it has low sensitivity. So, if the changes are very small in temperature.

So, you have it is difficult to measure the output generated voltage very accurately
extension wire problem is there which must be of same thermocouple type. So, this
extension wire problem is there, but if you use the same kind of wire you can eliminate
that up to a point. Another problem is wire may pick up radiation this is kind of a noise
they can pick up because these wires are very long in in size they can start picking up RF
radiation from nearby sources and this noise comes as a error in the measurement. It has
not very good accuracy of the temperature measurement. So, these are some advantage
and disadvantage of thermocouples.

The next is the resistance thermometer or we call it RTD's the resistance temperature
detector or in RTD in short. So, these are the sensors which are typically made of a single
pure metal. So, we have a metal and we measure the temperature using the property of
the property of measurement is the resistance of a metal changes with the temperature. If
we have a metal and its electrical resistance is let us say R0 the electrical resistance are
equal to R0 plus alpha into delta T. So, this is the resistance of any metal the resistance
depends on the base value of resistance and it also depends on the temperature difference
with a proportionality constant.

So, the electrical resistance of a metal is a function of change in the temperature with a
base value. This electrical resistance is function of temperature the most accurate
resistance thermometers are the one that use metal that have a very linear relationship.
So, in this expression we have just written only one term alpha into delta T, but there may
be some other second order terms let us say beta delta T square and so on, but we are
limiting our discussion only to the very first term delta T. So, if we want to choose a
metal which gives which should give a very linear response we want to have a metal that
does not have this delta T and other high order dependencies. So, we carefully choose the
metal which has a linear relationship with temperature one such metal is the platinum.

So, if we use a platinum as a metal and we make a sensor using this that gives you must
very much linear relationship and the characteristic of this RTD is you have this on the y
axis you have resistance on the x axis you have temperature and the characteristic is quite
linear. And we see when we are increasing the temperature there is a increase in the
resistance. So, this is R equal to R naught plus alpha T when we increase the temperature
delta T there is a increase in the resistance of this wire and the construction of these
RTDs are very simple you have just a tube on which you can bound this lead wire let us

57
say we take platinum. So, we take wire of platinum and we just wound it on this tube like
this on this core and this comes out with two terminal one and here. So, we have two
terminals now we measure the electrical resistance here using a multimeter and we
expose this whole assembly to the temperature let us say the body temperature and here
this is our sensor which is at temperature T s.

So, we expose this whole body to the measuring temperature what we want to measure
and we measure the electrical resistance of this platinum wire which is a linear function
of change in the temperature. So, these RTDs has multiple applications for example, air
conditioning and refrigeration servicing is one application, furnace servicing, food
service processing, medical research and textile production these are some applications
they are used at other places as well. We have some advantage and disadvantage of these
RTDs for example, these RTDs are most stable over the time the they are quite stable for
a very long time they are very much accurate they are most repeatable temperature
measurements very resistance to contamination this is one property which sometime is
very important because let us say we have a we have a sensor it should not react with the
environment. For example, we have a gas chamber and we want to measure the
temperature of this gas chamber from inside we do not want to put a sensor which can
react with this gas. So, if we use a platinum as a wire platinum is very inert material it
does not react with the gas or any other liquid which is inside.

So, we can easily put this sensor inside that chamber. On the other hand if we have some
other metals let us say copper or some other metal which can easily react with even with
water or even with some other liquids. So, they contaminate the whole chamber if we put
this kind of sensor in the chamber they contaminate everything. So, that time we want a
sensor using materials such that it does not react with environment. So, these sensors or
these RTDs are very resistant to contamination corrosion of the RTD elements the
disadvantages are the high cost.

High cost means the platinum is very expensive metal and we want to have a very long
wire of this platinum wire to make a sensor and to wound it on a on a core. So, the cost
goes up they have a very high cost of metal and they have very slow response to the
change in the temperature. Low sensitivity to small temperature changes they still cannot
measure a very small change in the temperature. Sensitive to vibration strains the
platinum element wire. Decalibration if used beyond sensor temperature readings and this
is also again somewhat fragile because the core core is also sometime made up of quartz
or glass.

So, it is somewhat fragile. The third is the thermistors. So, the thermistor is a specific
type of resistance thermometer they are also a special type of RTD, but in RTD we have
R equal to R naught plus alpha T means there is a positive temperature coefficient here
this plus sign and this alpha is plus. So, the resistance of a metal always increases with

58
temperature increase the temperature is rising. So, resistance of metal will always
increase.

However, in thermistor we have a different. However, in thermistor we have this

differently. Thermistor can have positive temperature coefficient as well as negative
temperature coefficients NTC or the negative temperature coefficients. So, the
relationship here is not very linear. In the RTD case we saw the output response which is
very linear with temperature.

However, in thermistor it is not very linear it is some other curve depending upon the
material used and this makes it different than the RTD. However, we are still measuring
the electrical resistance just like the RTD's. The temperature range in which thermistors
can be used is small compared to the normal RTD's. So, these thermistors are useful for a
very short range of temperatures in that range only their sensitivity is very good. Let us
say we have this is the typical input output characteristic of a thermistor we have
electrical resistance on y axis temperature on x axis and we see this type of graph.

Now, the sensitivity is the slope of this curve. So, if we have let us say a linear graph like
RTD the slope is not very high. So, sensitivity is less, but here if we see this exponential
graph if we take the sensitivity or the slope which is very high, but it is for a small range
of temperature. This characteristic may be deviating at some point to some positive
temperature coefficient just like this. So, the thermistors are useful for a small range of
temperature.

We will discuss all these thermocouples, the RTD's and thermistor in detail after studying
these advantage and disadvantage. Some applications of thermistors are they are most are
seen in medical equipments because the most of the time in medical equipments the
temperature range is very small. Thermistors are also used for engine coolants, oil and
temperature measurement in the transportation industry. So, multiple places they are
used. Advantage of thermistors are they are highly sensitive to small temperature changes
because of their NTC or the negative temperature coefficient.

Temperature measurement becomes more stable with use and copper or nickel extension
wires can be used. At disadvantage of these thermistors are limited temperature range.
We do not have a very wide temperature range for thermistors. They are still somewhat
fragile some initial accuracy drift is there. Decalibration if used beyond the sensors
temperature ratings and lack of standard for [Link] is also one bottle neck of
thermistors and the fourth ones are semiconductor based sensors. So, these sensors are
made of semiconductors for example, silicon. So, these are entirely different structure
compared to these conventional thermal sensors. So, this we are all these four sensors we
are going to discuss now in detail. So, before jumping to actual structure of these thermal
sensor let us brush up some fundamentals of heat transfer.

59
So, we have basically three mechanisms of heat exchange which is first is conduction,
second is convection and third is the radiation. So, these are the three modes of heat
exchange and conduction is the heat transfer primarily through solids and it can also
occur with flutes as well. So, when there are two bodies in close contact to each other
then there is a heat transfer between one to another and that is the conduction mode and
the rate of heat exchange is let us say Q conduction equal to K into A into difference in
temperature which is delta T here T hot minus T cold divided by L where Q is the heat
flow, K is the thermal conductivity, T is temperature, L is the length of material how long
the material is and A is the cross section area of the heat flow. So, this is the formula for
the heat flow how much heat is going to flow when they are in close contact and there is
a contact resistance when two materials join to form a conduction path. There is a
resistance at the point where the two materials join this occurs because of the two
surfaces are rarely completely flat when we have two materials join them together the
surface when they are making contact this is not very clean or very kind of flat.

So, there is always a gap between these two surfaces because of that the contact area is
not the actual physical area which is in contact. So, because of this voids of air there are
some contact resistance on the contact. So, let us take one example for this a rectangular
block has an area of 400 mm square a engineer would like to use it to cool his heat source
that is producing a total of 50 watt with a specification of 90 degrees. The heat source is
placed in a chamber with an air temperature of 65 degrees. So, the environment
temperature is 65 if the length is 10 mm what is the minimum conductivity of the
material in order to cool the heat source to specification used in the block.

So, we know the equation Q conduction equal to k a T hot minus T cold divided by L and
these are the values which are given to us from here this is T cold this is T hot and this is
the area and the length is given here which is 10 mm. So, we can put all these values
together here it will come out to be something like this. So, we can solve for the
conductivity k which is the last here in this example. So, this is one example of
conduction based heat transfer. The next method is convection, convection is the heat
transfer from a surface to a fluid and some common fluids include air and water.

So, when we have a heated body placed in the air through the convection mode it can lose
the heat to the air and the formula for the convection based heat transfer is Q convection
equal to h c into a into this temperature difference which is delta T where h here is the
convection heat transfer coefficient. So, this is the different coefficients from the earlier
one. This mode of heat transfer is used mostly to cool down the heated body for example,
we have fin like structures placed next to the heated body. So, that the contact area
between the metal and the fluid increases the air can flow between these fins and these
fins lose the energy which take the heat energy from object and lose it into the air
keeping the object cool down. This is the way of removing the heat from the heated
object.

60
The third mode is the radiation, the heat transfer between surfaces via electromagnetic
waves. So, when a body is having certain thermal energy it has the temperature which is
higher than the ambient it loses its energy the heat energy in form of electromagnetic
waves as well. So, this is third mode of radiation all matter is at a non zero temperature
emits electromagnetic waves whatever the temperature is that all these object emits these
electromagnetic waves including gases and liquids. So, typically this mode of heat
transfer can be more complex than conduction and convection, but at very high
temperature this is the mode which dominates to radiate the energy because if we see the
equation for Q radiation equal to h r into A into T source minus T surrounding and we put
all these values together the actual equation is Q radiation equal to epsilon C sigma A T
source to power 4 minus T surrounding to the power 4. So, here you see the rate of heat
transfer is dependent on fourth power of temperature it means at low temperatures this
mode does not dominate, but as soon as the temperature start rising this mode starts
dominating at high temperatures and very high temperatures T to the power 4 is the term
which dominates compared to the linear order of convection and conduction.

So, this radiation is the most common mode of heat transfer at elevated temperature. So,
now coming back to the thermocouples, so to understand thermocouple we investigate
how the current actually flows in an isothermal conductor. So, we have this block and
inside this we have free electrons all these free electrons which are free to move in any
direction. So, the there are certain equation for example, current density J equal to E into
N into V E is the electron charge N is the electron density and V is the drift velocity. So,
based on that you have certain current density in this material.

Similarly, you can write the heat flux as well JQ equal to QN into V the Q is the heat per
charge instead of having electrical charge now we have heat per charge in into N into V
or we can also write pi into N into V or we can also write pi into J where pi is the Peltier
coefficients. So, what the thermocouple is we have as we discussed we have in
thermocouple 2 different materials material 1 and material 2 and material 3 and material
4. So, we have 2 different materials we join them together at 1 junction and on the other
junction we measure the EMF generated between these 2 and this junction we need to put
it at cold junction or we can call it the reference junction typically 0 degree using R 1 and
R 2. So, this is the ice or so we can put this junction into the ice which is at 0 degree C
and we can use this as a sensing arm or sensing junction. This junction we put to the in
the in the close contact of body temperature and we the voltage build up here is linearly
proportional to this temperature difference between 0 degree and the sensing junction.

This effect was observed by Thomas John Seebeck discovered in 1821 and this is this
effect is called the Seebeck effect which is the temperature difference generates the
voltage. How does it generates the voltage? A pictorial view is something like this we
have this object where 1 junction is cold junction and this other junction is hot junction
and inside this block we have all these electrons which carrying charges as well as they

61
are also carrying heat or the thermal energy. So, when these electrons are close to hot
junction they pick up this kinetic energy from this high temperature when they pick up
this kinetic energy they move fast. So, when they move fast they spend less time towards
the hot junction and when they move towards the cold junction they lose this kinetic
energy because this junction is cold it takes away heat from these carriers. So, this slows
down here this slow down here and they spend more time in cold junction.

So, if we can see statistically more electrons will be confined in this cold junction
because they are spending more time here compared to the hot junction where the
electrons are less. Let us say we have only 10 electrons here and 100 electrons here this
side and these electrons are carrying the charge as well they have negatively charged. So,
we can see there is a potential build up delta V because of this mismatch in the in this
charge carrier. So, this is the basic principle of this Seebeck effect how a temperature
change from hot to cold is generating the voltage difference. So, this charge diffusion
under a temperature gradient and built in potential resisting diffusion this was observed
by Seebeck and we write Seebeck coefficient S is equal to minus delta V upon delta T
where delta V is the change in the voltage V hot minus V cold divided by delta T which
is the change in the temperature T hot minus T cold and this is called the Seebeck
coefficient.

The opposite of this effect is Peltier effect. The Peltier effect says we have same
thermocouple, and we make these junctions this is the material 1 and this is material 2. If
we apply a current or the voltage from outside here, we are applying external electrical
stimuli here then this junction and this junction will have temperature difference or the
temperature difference or the temperature gradient will can be produced using the
electrical signals and this is called the Peltier effect. So, Peltier discovered this in 1834 an
electrical current creates a cooling or heating effect at the junction depending upon the
direction of current flow where the current you are flowing. So, this is kind of reverse of
Seebeck effect observed by Peltier and the Peltier effect is like this you have two
different metals let us say this is 1 and this is 2 when you connect external voltage source
the current flows like this and it generates a heated surface on one side which is hot
junction and this junction becomes cold. So, we can control this external voltage or the
current flow in this assembly and we can controlled how the hot and cold junction are
different in temperature this delta T we can easily control using this delta V.

The Peltier effect this is called the Peltier effect and the Q Peltier is equal to pi 1 minus pi
2 into J this is reversible with current direction where Q is the heat absorbed or released
pi is the Peltier coefficient these are the Peltier coefficients and J is the current flowing
through the system. The other effect is the Thomson effect the Thomson effect says the
amount of heat released or absorbed by an object it depends on how much current is
flowing through that object. So, let us say we have this material and there is a current
flowing inside this how much heat is being absorbed from environment how much heat is

62
absorbed or released to the environment it depends on how much current is actually
flowing into the material this is called Thomson effect this was invented by William
Thomson. And the Thomson effect if we see pictorially we have this object where the
current is flowing current is nothing but the mobile charges which are flowing from one
end to other end they are carrying charges and they are carrying the thermal energy as
well because in the form of kinetic energy. Now this area is exposed from environment so
amount of heat absorbed here it depends on how many charge carriers are picking up this
energy from environment and when they pick up this thermal energy they go to other
side.

So, if we have higher current for higher current you have large number of electrons large
number of electrons present here which are ready to pick up this thermal energy from
environment and move away. So, you have large number of electrons means the large
thermal energy they pick up if they the number of electrons are lesser the lesser energy
will be absorbed. So, this is what the Thomson effect is the heat absorbed or released
depends on how much current is flowing into the object. And Thomson coefficient tau is
equal to 1 upon i dQ by dx upon dT upon dx and there is a Kelvin relationship which
connects all these thermal coefficient Thomson coefficient to the Peltier coefficient and
Seabag coefficient. So now we are coming back to sensors the thermal sensors these
thermocouples consist of a pair of dissimilar wires as we discussed joined at one end.

This connecting point is known as a measuring junction only the junction itself does not
produce the thermocouple small signal voltage. This is the when we make this
thermocouple by twisting two wires we make it sensing junction. This junction itself is
not creating the output voltage in fact the whole length of wire is creating the output
voltage where the temperature difference is between the hot junction and the cold
junction. But entire length is contributing to the linear EMF build up. The actual
thermoelectric effect is an extended and continuous that is distributed along the entire
length of the thermocouple.

So, we have let us say this is our thermocouple which has material 1 and material 2 this
junction is hot junction and this is cold this is hot junction. Now if we just think about
only one metal let us say metal 1 the thermal profile when we subject this metal to the
temperature gradient as we saw the hot junction is let us say at some potential and the
cold junction is at different potential because as we discussed the electrons spend less
time in hot junction and spend more time in cold junction there is a potential build up. So,
this metal or let us say metal 1 this has this field or the electric potential profile this Y
axis electric potential. So, hot junction is at different potential compared to the cold
junction this is your delta V because of this is your the delta T the temperature gradient.

So, this profile is for the material 1. Now we have material 2 which has different profile.
So, this is material 1 now if we have material 2 this will have different this is material 2

63
this is material 1 let us say. So, these profiles are different it means at this junction we are
forcing the potential to be same. So, the other end will develop different potential because
of this delta V let us say we have delta V1 and for material 2 we will have delta V2
material 2 will have delta V2. So, this and because of this delta V1 and delta V2 there
will be an effective delta V the potential build up across these two terminals.

So, if we have two same metals let us say we take two same metals and we join them
together make a thermocouple it will not be possible to have EMF build up why because
metal 1 will have this thermal profile and the metal 2 which is same as metal 1 will also
have the same material profile. So, the difference between these two profiles will be 0.
So, this delta V will be 0 if we have two same metals that is why we in the thermocouple
we need to have two different metals join together then only it will produce a EMF which
is linearly proportional to the temperature. All thermo electric activity takes place in the
center zone we see this is the center zone which is creating this voltage difference
otherwise at the junction the voltage is same the junction the voltage is same and at the
cold junction the voltage is same this is the entire length in between this hot and cold
junction which is creating this voltage change which is the center zone of decreasing
temperature. So, this is the actual working principle how we use this thermocouples in
practice we have this hot junction and this is material 1 and material 2 these are bring
together to a cold junction or the reference temperature typically 0 degree using ice or so.

So, this is filled with ice which is which keep this junction at 0 degree. Now, we connect
both of these wires to volt meter and this is the hot junction and this is the cold junction.
Now, we connect both of these wires to volt meter and this is the hot junction and this is
the cold junction. Now, we connect both which are connecting this volt meter to these 2
junctions this is material 1 and material 2 which are different to each other and we
connect copper wires to connect volt meter to this end. Now, this volt meter will read a
different it change in the voltage depending upon how much is the difference between the
cold junction and the hot junction.

So, this delta V is proportional to the delta T this is the actual measurement setup of
thermocouples. Materials for thermocouple what we need to be careful about the melting
point of thermocouple material should be higher than the measuring temperature. So, let
us say we use 2 different metals the limitation of this thermocouple is it cannot measure a
temperature which is higher than the melting point of these metals otherwise these metals
will melt down and the wire will kind of damaged. The dissimilar materials on joining
should be able to produce large EMF. So, if we choose a metal carefully we need to have
a metal 1 and metal 2 which are very different in the electrical profile when they are
subjected to these temperature gradients.

If both the metals we choose are of same potential profile then the delta V will be very
small and it will be very difficult to measure the output voltage. So, we need to carefully

64
choose all the these 2 different metals so that they produce large EMF at the output.
Temperature is determined indirectly through calibration of EMF with temperature. So,
for as possible the linear variation of EMF with temperature is desired. We want to have
a linear EMF or the delta V is proportional linearly proportional to delta T this is what we
expect from the thermocouples.

These thermocouple materials should be resistant to atmospheres in furnaces. So, these

metals which we choose for making these thermocouple these metals should not react
with the surroundings or the chamber where we want to put this thermocouple as a
thermal sensor otherwise they can contaminate the chamber itself. They should be
chemically inert as well. There are multiple materials are available for thermocouples for
example, type B we use metals as platinum 6 percent rhodium and platinum 30 percent
rhodium these are 2 different metals we use if we use B type thermocouple. Similarly, we
have E type thermocouple J type thermocouple K depending upon the industry standards
we have multiple types of thermocouples where we use different metals depending upon
our goal of measurement.

So, we have R type S type T type. So, all these types they have their specific metals to
choose and there are certain characteristics are given here. For example, C back
coefficient is given degree centigrade standard wire error is there or ohm per double foot
20 AWG. So, this is the dimension of wire the cross section. So, depending upon the type
of thermocouple the material for the metal 1 and metal 2 are fixed. Let us say for J type
we have material 1 as iron and the material 2 as Constantin.

And for example, we take J type thermocouple then we have this is the table which is
giving the delta V which is the voltage build up proportional to the temperature how
much temperature we change and how much voltage is generated by this J type
thermocouple this is given in this table which are easily available in the market when we
buy that is a J type thermocouple from market we can always have this table downloaded
from various sources. So, this table says when there is a change from minus 200 to minus
90 it means 10 degree change how much is the voltage change and this is 200 then this is
201 this is 202 we have here 10 degrees. So, here this is 200 this is 199 this is 198 197.
So, from this is the 10 degree step and for each 10 degree we have this row which is 10 in
nature. So, we can read this table to measure the voltage for 1 degree centigrade rise in
temperature.

So, we see the example let us say we take a J type thermocouple whose reference
junction is at 0 degree c. So, all these measurements are done the table in the previous
slide this is all reference to 0 degree this produce a EMF voltage of 4.115 milli volt what
is the temperature at the measuring junction. So, 4.115 is the voltage measured by
voltmeter now we want to calculate how much is the temperature of heated junction.

65
 This produce a EMF voltage of 4.115 milli volt what is the temperature at the measuring
junction. So, 4.115 is the voltage measured by voltmeter now we want to calculate how
much is the temperature of heated junction. So, for a J type thermocouple we seen the
table earlier we simply need to find the temperature from the table which would generate
a EMF of 4.115. So, if we go back to this same table and we try to calculate this 4.115
milli volt potential that come out to be between 78 and 79 degree. So, 78 degrees
somewhere this is 70 this is 80 and here it is 7.

So, our 4.115 milli volt is in between this 4.079 and 4.113 which is which corresponds to
8 and 9 and here 70 degree. So, the temperature is around 78 degree or 79 degree or we
can even extrapolate as well to calculate the exact temperature. So, this is how we read
these the tables for given for this thermocouples which are standard in J type
thermocouple. Another example is example number 2, in example 2 we have a J type
thermocouple whose reference junction is at 21 degree C. So, instead of having 0 degree
C we generally we keep a reference temperature now the reference temperature itself is
higher than 0 degree by 21 degree C 21 degree centigrade it is higher than 0 degree.

Now it is producing a EMF voltage of 2.878 milli volt. Now we want to calculate what is
the temperature of hot sensing junction. So, we discussed that the voltage developed is
linearly proportional to the temperature gradient and the EMF generated from 0 to T 0
degree to T we can split it up between 0 to 21 and then 21 to the temperature T because
the relationship is linear. Now our cold junction is at 21 degree.

So, from this cold junction to this temperature we are known that EMF is 2.878 2.878 and
for because this is J type thermocouple. So, we can calculate from the table that how
much is the EMF for 0 to 21 degree that we can calculate from the table and we can
calculate the effective EMF which is 3.54 and we can again go back to the table and
calculate what is the temperature for this effective EMF which is 75.7 degree centigrade.
So, these are some examples of thermocouples which you can use and they shows how
you can calculate the values of thermocouple using the truth table for that type of
thermocouples.

So, this is all for today.

Thank you.

66
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 07
Thermal Sensors: RTDs and Thermistors

Hello, welcome to the course Transducers for Instrumentation. Last lecture we discussed
a special type of temperature sensor which is thermocouple in which we had two
materials to in fact, two different materials which join them together and the voltage
build up across the cold junction is linearly proportional to the temperature gradient. So,
there we are using a multimeter or a voltmeter to measure the voltage directly from the
thermocouples. Now, today we are going to discuss another type of thermal sensor which
is RTD or we call it the Resistance Temperature Detectors in or RTD in short. So,
Humphrey Davy showed that the resistivity of a metal shows a proportional temperature
dependence means the electrical resistivity or R is equal to R naught which is the base
resistance of that material plus alpha into delta T. So, the electrical resistance of a
material is proportional to the temperature gradient which we apply and if there is a
change in the temperature of this material there is a proportional change in the electrical
resistance and R naught is the base resistance which is a fixed value and William
Simmons they demonstrated that the use of platinum as a element in a resistance
thermometer.

So, he used the platinum wire as an element to make RTD. Why platinum is specially
used because it can withstand high temperatures you can use the platinum wire to
measure very high values of temperatures it can withstand without melting and still
maintaining excellent stability even at those high temperature. As a noble metal it shows
limited susceptibility to contamination this platinum you can put inside a chamber which
is filled of some gas or some ah reactive liquid this platinum does not react with the
environment or that particular gas or liquid and it does not contaminate the fluid inside.
So, platinum is a very noble material to use as a RTD though this design produces a very
stable element the thermal contact is very poor.

So, the construction of a resistance temperature detector or RTD is we take a wire of

platinum and we just wound it on a glass rod or something like this we take we take a
glass ah glass rod and we wound a platinum wire on this rod and this is what our RTD is.
This produce a very stable element the thermal contact is very poor. So, when we make
this kind of sensor how do we have a proper contact of this assembly with a heated body
let us say which is circular in nature. So, a very small portion of this assembly let us say
this portion only gets in contact with the heated body. It limits the thermal contact area
which gives us different kind of errors in the measurement.

67
So, this is the limitation of this kind of design which results in a slow thermal response
time the fragility of the structure also limits it multiple laboratory environments. So, this
type of assembly it can give you RTD sensor, but it is very limited to multiple laboratory
situations. So, now we switch to thin film RTDs or the metal film RTDs. So, the concept
is same we have electrical resistance which is proportional to delta T the electrical
resistance is proportional to the temperature. To make a electrical resistance we need to
have a two terminal element and in between there is a certain length of this material we
want to use it as a electrical resistor.

Instead of having a long wire now we deposit a very thin film of the material on a
substrate typically let us say platinum or glass metal glass serif film is deposited or
screened onto a small flat ceramic substrate typically silicon. So, let us say we have one
substrate on top of it we deposit a very thin film may be a few micron thick only and then
we pattern it using photolithography there are certain fabrication steps involved in this we
etch out the unwanted area unwanted film what we deposited and we pattern a serpentine
like structure which is our resistance now. So, we deposit a thin film and using
photolithography we make this kind of structure. Now we have two terminals here one is
this is terminal 1 this is terminal 2 and this entire length is my resistor I can connect these
two terminals to my multimeter and measure the resistance of this wire. Now the size of
these RTDs which we make using these thin film this is very small.

So, this film RTD of a substantial reduction in assembly time because this can be done in
a very scalable manner. So, it reduce the assembly time and has further advantage of
increased resistance for a given size because the thickness of film is very small it means
that to have a higher value of resistance we can go further thin down the thin film and we
save the material. Now even if we are using the platinum which is very costly material
the amount of platinum I am using in depositing this thin film is very small. So, I can go
for very expensive metals and still make the sensors very cheap. So, this is one advantage
of this kind of fabrication.

Due to this manufacturing technology the device size itself is small which means it can
respond quickly to small change in temperature. This we discussed earlier in lecture if we
have a very small sensor the thermal capacity of the sensor is very small. So, by making a
sensor very small it can rise to that body temperature in a very small time. It just need a
very small amount of thermal energy and it can pick up the temperature exactly to the
sensing the body the heated body and it can give the response very fast. So, the response
time of RTDs when they are made using thin film that thermal the response time is very
small.

These are some metals and their resistivity in ohm per cm or the circular mill foot. Let us
for example, we have gold, silver, copper, platinum, tungsten, nickel and these are the
typical resistivities. So, depending upon the application where we want to have the value

68
of resistance which is fixed by application we can choose any of the metal from this
table. This we can note the all metal produce a positive change in resistance for a positive
change in temperature. When the temperature is rising the resistance of these metals this
always rises.

So, this R is actually R naught plus alpha delta T where this term is positive with
temperature rise the resistance actually goes high. Though in RTDs gold and silver are
rarely used as RTD elements because of their lower resistivity. It is not because of the
cost of these metals they are not used because of the low resistivity because the very low
value of resistance it is very difficult to measure those resistances. We need a certain
resistivity to make resistances and the resistance values typically in ohm to kilo ohm
range is easy to measure comparable to 0.1 ohm or few milli ohm of resistivity resistance
this is very difficult to measure.

So, gold and silver are rarely used. Tungsten has a very relatively high resistivity, but this
is reserved for very high temperature applications because it is extremely brittle this
metal is extremely brittle and difficult to work. So, it has very high resistivity, but we use
this metal only for very high temperature measurements. Copper is also used occasionally
as an RTD element it is low resistivity forces the elements to be longer because if the
resistivity is lower we need to have a longer wire to have certain resistance in ohm or kilo
ohm range. Otherwise, for small size sensor the resistance of the copper will be in milli
ohm will be very difficult to measure, but its linearity and very low cost makes it an
economical alternative.

Copper is still used because of the good linearity and very low cost its upper temperature
limit is only 120 degree C after that the copper metal actually does not respond very well
to temperature gradients. The most common RTDs are made up of platinum nickel or
nickel alloys these are the materials which commonly used for RTD fabrication though
the these metals are costly, but because we are using very thin film of these materials the
cost of the material comes down and we can still use these expensive metals. The
economical nickel derivative wires are used over a limited temperature range they are
quite non-linear and tend to drift with time for measurement integrity platinum is the
obvious choice. So, now we come to how do we do the measurement using these RTDs
we have these RTDs which are fabricated using thin film they are very small in size.
Now, how do we take measurement of temperature using these RTDs.

So, the common values of resistance for a platinum RTD range from 10 ohm to several
1000 ohms these are the typical resistance values of RTD. 10 to let us say 50 kilo ohm
standard temperature coefficient of platinum wire is alpha which is 0.00385 it means for
a 1 degree change in the temperature the resistance of platinum wire is 0.00385 ohm. For
a 100 ohm wire this corresponds to 0.00385 ohm. For a 100 ohm wire this corresponds to

69
0.00385 ohm per degree c at 0 degree c. Now, we take this resistance value which is
0.00385 ohm and we have to calculate the resistance value of the platinum wire.

So, this resistance value is 0.00385 ohm and at 0 degree c. Now, we take this RTD which
is fabricated using platinum wires or the platinum metal this is our RTD and here we are
exposing it to temperature TS or the TB which is body temperature and here is my
instrument connected which is measuring this change in the electrical resistance. Now,
this instrument which may be a multimeter this meter is connected to this RTD through
these lead wires these extension wires these are called lead wires which also has its own
resistance. Let us say 5 ohm is the resistance of this one such lead wire.

So, 5 ohm here and 5 ohm here. So, total is exactly 100 ohm of RTD plus 2 into 5 ohm
because of the lead wire. So, actually we have a RTD of 100 ohm, but effectively when
we join this using lead wires the total resistance is not 100 ohm which is now 100 plus 10
ohm which is 110 ohm. So, because of these extra lead wires now my meter is not
reading 100 ohm, but it is actually reading 110 ohm. So, this measurement wires leads to
the sensor may be several ohms or even tens of ohms depending upon the value of these
lead wires this reading can be even more degraded.

So, a 10 ohm lead impedance implies 10 over 0.385 almost 26 degree centigrade error in
the measurement these lead wires can produce. So, this is the problem when we have a
sensor we want to connect it to a measuring unit using some lead wires these lead wires
also has electrical resistance. So, one classical method of this error is using the bridge the
wheat stone bridge we have this is the bridge we have let us say R 1, R 3, R 2, R 4 and
here this is RTD which is acting as R 2. Now, this is the conventional way of removing
this error this is the conventional way where this RTD is exposed to that body
temperature which is from outside and all the other resistance is are not exposed.

This RTD is made of platinum it will give a linear change in the resistance when subject
to this TB and this we can detect using this bridge, but when we expose this RTD to some
temperature we need to keep it separate from the other assemblies or the other resistances
because these resistance also has some temperature dependency. So, still we connect this
RTD which is far away from the whole assembly this is whole assembly we do not want
to expose to different temperature only this RTD is taken away using these lead wires and
this area is exposed to the temperature which is TB which is some higher temperature.
Now, because of this extra lead wire we have lead wires we still have some resistances of
these lead wires. So, this is the extra lead wires which we have here which causes the
error in the measurement. So, what we do we take this kind of assembly where we have
this RTD these are the lead wires and we take measurement at this point only and we
connect extra lead wire here on the other side of this measurement to compensate for this
extra lead wire.

70
Let us say this is our lead wire, our lead so these are also our lead which we connect on
the other arm of bead stone bridge to remove this error and we take measurement
between these two points. So, this is four lead connection for resistance connection and
we put dummy lead wires in the measurement to reduce the error, but it increases some
cost because of this extra wires and we can completely correct for this resistance error
using this bead stone bridge. So, this is what actually we are doing in the this bead stone
bridge example we are doing a four point measurement instead of having a two point
measurement. So, let us discuss what this four point resistance measurement is or
sometime we call it the Kelvin connection. So, this type of four point measurement is
done for precise resistance measurement.

So, up to now we know two probe method and the drawback of this two probe method is
error due to contact resistance. So, we have for example, this is the resistance which we
are trying to measure this is a resistance and how do we measure the resistance we apply
a certain known voltage and we measure the current and then resistance is nothing but V
upon I we apply a known voltage and we measure the current. So, this is the voltage V
which we apply here accordingly there will be a current flowing into this which is I let us
say it is flowing from here to here and this will produce a voltage between these two
points which you can measure using voltmeter this is the voltage you are applying here
this is the known voltage V e let us say. So, the problem here is this resistance is when
material of this resistance which is in connection with these lead wires which is used to
connect these instruments to this resistance whenever there is a junction here all the
junction has some contact resistance. So, let us say we take metal one and metal two we
join them together it is not like they do not have a contact resistance and this contact
resistance is very much different than the resistance of material one and resistance of
material two it is not same.

So, the contact resistance depends on the materials and this is always there. So, if we see
this setup microscopically we have a resistance here which we are going to measure but
we have certain contact resistance here and we have at this junction wherever there is a
junction we have these contact resistance let us say RC and this is in series with this
resistance or Rl the load resistance. So, actual resistance here is Rl plus two times RC.
So, this two times RC is an extra resistance coming from the contact resistance which
produces the error when I am connecting a voltage source voltage across it there is a
current flowing this current depends on not only the Rl but as on the RC as well because
RC also in series with Rl. So, this current is proportional to Rl plus two times RC which
is the error the contact resistance and this inaccurate value of current i is giving me the
error in the resistance the R calculation from this formula this is this has error because of
this extra term twice RC.

So, the contact resistance in this assembly is giving me error but now the contact
resistance is something which we cannot remove whatever material we choose material

71
one and material two they will always have some contact resistance. So, this we cannot
remove but the what Kelvin connection is this contact resistance is giving me extra
voltage drop or it is affecting the current only because if there is a current flowing inside
this if we do not have a current flowing into the contact resistance then it does not impact
the value of voltage. So, this is the typical assembly of four point Kelvin connection we
have the resistance here where we connect the voltage from outside here in the from the
external two probes which are carrying all the currents the current is flowing from this
side to this side and we have contact resistance here RC and some current is flowing
which we can monitor from here. Now, voltmeter instead of connecting at across these
two probes we have separate probe for measuring this potential and this potential is a
high impedance potentiometer the high current need not to flow in this in this setup this
current is not very high it is very small in magnitude. So, if we have contact resistance
here as well but there is no current flowing into these RC's these are not going to impact
any voltage difference and the actual voltage we are measuring between this and this.

So, between these two probes we are actually measuring the resistance and which we can
again scale to the full length of this resistance value. So, this is the four point or the four
probe method which is also called Kelvin resistance we have two probes for supplying
current and again two probe for measuring the voltage or we call it the inner probes. So,
these inner probes does not draw any current hence the contact resistance does not impact
measurements. So, this RC is always there contact resistance between two metals these
are always there but the current is not flowing so it does not impact the voltage reading.
So, this is the Kelvin connection and this is the Kelvin connection we generally use in
RC.

So, RTD measurements so we have this RTD element we excite this externally and
across this we measure the voltage Vout and we measure the resistance of these RTDs. So,
this is four terminal lead resistance correction and voltmeter we use with the high input
impedance and no current flows in the voltage connections this current is very negligible
or we can write 0 because of this there is no problem of contact resistance. Now
resistance to temperature conversion this RTD is a more linear device than the
thermocouple but it still requires some curve fitting the CVD equation or this Schoenleer-
Wern-Duisen equation is used to approximate the RTD curve which is given here RT
equal to R naught plus R naught alpha T minus delta T upon 100 minus 1. So, this is the
complete equation where we have RT which is resistance at temperature T R naught is
the resistance at 0 degree C which is a fixed value alpha is the temperature coefficient for
that particular metal delta is typically 1.49 for platinum and beta is typically 0 for T
greater than 0 or it is 0.11 for T less than 0. The exact values of coefficient alpha beta and
delta are determined by testing the RTD at four different temperature and solving the
resulting equation. These coefficients alpha beta and delta these are typically measured
when we do some calibration of this developed RTD. Calibration is something what we

72
when we fabricate a sensor then we apply certain input known inputs to the sensor and
the output is known. So, we calibrate the output of the sensor and find out all the
coefficients alpha beta and delta. So, after this we come to the third type of thermal
sensor which is thermistors.

A thermistor is a semiconductor made by mixture of metallic oxides such as oxides of

manganese, nickel, cobalt, copper and uranium. So, this thermistor unlike the RTD it is a
mixture. However, in RTD we discussed a pure metal which has resistance proportional
to temperature. Thermistor is not a pure metal it is a mixture of multiple mixtures and
depending upon what you mix you can have more than two materials to mix with
depending upon the mixture and the and it contains output characteristic of thermistor
changes. Thermistors may have negative temperature coefficients means the electrical
equation for resistance can be R equal to R naught minus alpha T.

This can be a negative temperature coefficients or we call it NTC in short. So, this is the
typical characteristic of thermistor where we can have negative temperature coefficients.
It means if the temperature is rising the resistance of thermistor can go down. So, this
temperature goes up but the resistance of thermistor can go down unlike the RTD where
it always goes high. This resistance versus temperature graph we can plot RTD's and
thermistors on the same graph.

We see RTD is always increasing linearly in the temperature of the thermistor and it is
increasing in the temperature of the thermistor. So, this is the temperature of the
thermistor and this is the temperature of the thermistor. For thermistor graph we see RTD
is always increasing linear graph where the temperature always rises with temperature
with this resistance is always rising with respect to the temperature. For thermistor we
can have multiple curves depending upon the mixture we use we can have different
different characteristic. For example, we have this graph for a particular thermistor one
where we have positive temperature coefficient for most of the temperature range and we
have negative temperature coefficient for a small range.

This can be our thermistor two where we have mostly the negative temperature
coefficient depending upon the mixture we use. So, we have positive temperature
coefficients as well as negative temperature coefficients in the same thermistor. For
example, this thermistor in this temperature range this is NTC the negative temperature
coefficient and from here to this range this shows a positive temperature coefficient. So,
the same thermistor for different range of temperature it can show a PTC or an NTC.
However, the very unique characteristic of thermistor is if we see the slope here of this
graph the RTD slope is very linear and the slope is not very high, but if we see this
thermistor graph here in this range from 40 to 50 degree the graph is very sharply rising
in this temperature range only.

73
So, in this temperature range of the graph is very sharp it means the slope of this graph is
very high and the slope of the graph is the sensitivity of a sensor. It means the sensor will
produce a very sharp output for a very small change in the input at the input measurement
which is temperature. So, it is 1 degree temperature rise will give a very high output or
change in the resistance. So, for a very short range of temperature the thermistor output is
very high and the sensitivity is very high. So, these thermistors are very much useful
when you have a very narrow band narrow range of temperature, but you want to
precisely measure the temperature.

The thermistor model is given by this equation which is basic equation log natural S
equal to A naught plus A 1 upon T plus A 2 upon T square where S is the measured
resistance and the simple model to first order approximation is log S equal to A plus beta
M upon T or you can solve it to have this equation which is the temperature. Beta M is
the metallic characteristics, T naught is the reference temperature and S naught is the
reference temperature. A higher value of beta is desired to have a high sensitivity if the
beta is increasing then the graph actually this has more and more slope. Sensitivity is
given by S equal to minus beta upon T square which is the equation for sensitivity and
the modified this model is T equal to A plus B log natural R upon R naught. These are the
equations which we use to model these thermistors though we know that the thermistor is
a non-linear device.

So, we do some sort of curve fitting to have this exact match and we use these models to
accurately model the thermistors. The manufacturing process of thermistor is very
different than other sensors. The thermistors for thermistor fabrication we start with two
or more semiconductor powders we have two or more different semiconductor powders
this we mix together with a binder. Binder is something like something kind of a glue sort
of thing we put all these mixtures in this binder to form a slurry. Small drop of the slurry
are formed over the lead wires dried and put in a sintering furnace.

So, for example, we make a mixture of two different two or more different semiconductor
powders make it like a paste. Now, we take two lead wires this is one this is two lead
wires we dip this whole these two wires into this mixture and take it out. So, when we
take it out a small amount of slurry will be on the end of this thing joining these both of
these two lead wires together. This droplet will be formed here and this is our sensor we
take it out and then we dry it and put into a sintering furnace. During sintering the metal
metallic oxide shrunk into the lead wires and formed the electrical connection.

So, this is the electrical connection which is formed during sintering. This bead we call
this droplet the bead this bead is then isolated using some glass or some coating we put
on it. So, that it does not react with the environment. So, these are sealed by coating them
with glass and then this glass coating improves the stability by eliminating water
absorption this should not be exposed to the humidity or water in the air. The resistance

74
of the thermistor may vary from a few ohm to several kilo ohm that depends on the
mixture we use.

The thermistor size is very small when we dip these lead wires and take it out the amount
of slurry which making a droplet is very small. So, when we have a very small sensor
then we get two advantages one is the very small time constant and ease of access to
small surface. We have a very small sensor. So, we can have a very fast and very fast
time constant. So, we can have a very fast response of this sensor and we can even
measure the temperature of a very small object as well.

So, these are the two advantage by having a small size sensor. A silicon with varying
amount of boron impurities can have either a positive or negative temperature coefficient
of resistance. So, if we have silicon which is very commonly used semiconductor if we
have boron impurities changing upon the quantity of this boron in the silicon this whole
assembly can show a positive as well as negative temperature coefficient for different
values of boron. Germanium doped with arsenic gallium is used for cryogenic
temperatures. So, this is another thermistor. The linearity of thermistors is a interesting
property a thermistor is a non-linear device unlike RTD or thermocouple those two
temperature sensors are linear in nature the output voltage generated in thermocouple is
linear to temperature.

Similarly, the resistance change in RTD is linear to temperature change. The thermistor
here is a non-linear device if we change 1 degree centigrade of temperature the output
depends on where we are on that curve. If the sensitivity is very high in that range it will
show a very high response very sharp change in the resistance if the curve is flatter in that
particular temperature range the output will not be so high. So, this is a non-linear device.
However, in certain applications one resistance can be added in parallel to thermistor to
improve the linearity.

So, having a high sensitivity means we have a non-linearity in the device. So, sometime
we can sacrifice some sensitivity of the sensor, but we want to improve on the linearity.
So, what we do we add a external resistance in series with this thermistor and that extra
resistance is linear in nature. So, we can somehow linearize the total output of this whole
assembly. The sensitivity of this sensor is very thermistor is very high therefore, the
change in resistance resulting from a temperature change is much higher.

Then the small change in resistance of the lead wire. So, if we have a lead wire
connecting this thermistor the resistance change in thermistor will be much higher
compared to change in the lead wire. So, we do not have much of a lead wire problem
here. Hence, this error resulting from lead wire can usually small enough to be neglected
for thermistor. So, let us take one example of thermistors we have a 10 kilo ohm NTC
thermistors this is a 10 kilo ohm NTC thermistor which is negative temperature

75
coefficients. So, we have a temperature change with beta equal to 3500 determine the
thermistor resistance at T equal to 30 and 50 degree also determine the relative sensitivity
in ohm per degree centigrade room temperature is 25 degree.

So, answer is minus 210 ohm per degree centigrade. And we know the formula from
earlier slide the sensitivity is equal to minus beta upon T square. So, these are all the
thermal sensors we discussed which are conventional three sensors like thermocouple
RTDs and thermistors. When we make a complete system using these thermistors or
these thermal sensors these network or these systems there is a analogy between electrical
and thermal networks that we can see. For example, we have this thermal network which
is nothing but a electronic PCB we have a board here on top of this we have a
semiconductor IC and this semiconductor IC is doing some processing and we are
feeding it some voltage from outside this IC is generating some heat this heat is need to
be removed from this IC.

So, that it can work properly. So, what we put we put on top of this a heat sink which
look like this which is a aluminum heat sink aluminum heat sink this heat sink has its
own thermal resistance. So, we have this assembly and this heat sink to improve the
contact area we put some glue or epoxy in between and we have this ceramic casing of IC
which is sitting here this is IC and on the bottom we have PCB. So, this is a thermal
network. However, we can make electrical kind of network using the thermal resistances
where we have this junction temperature instead of having the junction voltage we can
now have a junction temperature which is here which is generating all this heat junction
and this heat is flowing to surroundings which is air which is here and which is here
through the board. So, this temperature can flow from the top through this heat sink or
from the bottom through this PCB.

So, we have two paths from this junction to the air which is this is path one and this is
path two and this is path two this is heated junction. So, this junction is connected to the
case of this IC and which has its thermal resistance which is R package this case is
connected to the sink using some interface material which can be the epoxy o r glue
which has its own thermal resistance. So, this case is connected to sink through that
thermal resistance and then this sink is connected to the air which is flowing in the fins
through this air which is flowing through this R heat sink which is the thermal resistance
of heat sink. So, this heat is flowing from T junction to T air through this thermal
resistance network this is one path second path is this thermal junction is connected to
this board junction and this board junction is connected to air using intermediate these
resistances this can be bond wires or solder balls and this resistance may be some FR4
substrate. Which has its own thermal resistance so we can see a thermal network is
exactly behaving like a electrical network we have a node a high voltage junction here.

76
So, instead of high voltage we have a high t emperature junction here and the
environment which is the air or the fixed temperature which we can consider as a
reference node or we can assume let us say this is 0. So, if we assume this 0 degree and
this is 25 degrees C the heat is going to flow from this node to 0 degree with here this is
also 0 degree. So, heat will flow this side and this side through this thermal resistances.
So, we see that though it is a thermal network the calculation of this thermal network is
very similar to the electrical networks how we solve these using Kirchhoff's law. So,
there is a correlation between these electrical and thermal parameters for example, we
have in electrical systems we have V equal to I into R the voltage equal to current into the
resistance which is our Ohm's law for thermal network we have delta T equal to P into
RTH which is delta T is the temperature difference instead of the potential difference we
have temperature difference instead of current we have power dissipation in what and
instead of electrical resistances we have thermal resistances.

So, all these electrical parameters has some similar parameters in thermal networks for
example, voltage difference is analogous to the temperature difference electrical
capacitance is analogous to thermal capacitance. So, these thermal networks can also be
solved just like an electrical network.

So, this is all for today.

Thank you.

77
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 08
Thermal Sensors: Electrical vs Thermal Networks

Welcome to the course transducers for instrumentation. We are discussing thermal

sensors and last lectures we are we discussed about electrical versus thermal parameters.
So there is a correlation between the electrical parameters as well as the thermal
parameters and we discussed that for example we have certain parameters in electrical
systems like voltage, current, voltage difference. So the same type of parameters are exist
in thermal parameters as well. For example the temperature difference, power dissipation,
thermal resistance etc. So we are going to discuss today about the electrical versus
thermal parameters. So we will discuss electrical versus thermal. So in electrical system
we have like voltage difference. Similarly in the thermal system we have temperature
difference. In electrical system we have current.

In thermal system we have the dissipated power. Similarly we have electrical resistance.
And in thermal we have the thermal resistance. So there is a corollary between all these
electrical parameters and thermal parameters and we have multiple laws in electrical
system to solve electrical network. Similarly the same type of laws are applicable to
thermal network as well and we can solve the thermal network using those laws.

We just need to replace voltage difference by temperature difference and electrical

resistance by the thermal resistance. So similarly we can solve the thermal networks. For
example we have Ohm's law in electrical system. We have a resistance. We apply a
potential difference across this.

It is a positive and negative and there is a current flow in this resistor which is I.
Similarly if we think about the thermal network we have a thermal resistance which is
our TH the thermal resistance and we apply a temperature difference across this
resistance which is delta T then the power dissipation will be there which is p dot. So this
is very similar to electrical Ohm's law and we can write delta T is equal to pd into RTH
the thermal resistance. For electrical network we can write V is equal to iR. So both of
these examples are similar and we can apply the electrical laws to the thermal network as
well.

So let's talk about electrical and thermal resistances. We have a material here. The
thickness of this material is D and we apply a potential difference here across it which is
V and the resistance of this is given by R equal to D upon A into I. Similarly we have a
material in thermal networks and we apply a delta T across this and the power

78
dissipation. So flowing through this material is pd and we can write RTH is equal to T
upon A into lambda th where this psi is electrical conductivity and here the lambda is
thermal conductivity.

Here R which is our normal electrical resistance and here RTH is thermal resistance. So
this is the electrical versus thermal resistances. Both behave in a similar way except
instead of voltage difference we apply the temperature difference and instead of the
current flow which has heat flow flowing through the material. Similarly we have
electrical and thermal capacitances which are analogous to each other. In electrical
system we have electrical capacitance which is generally denoted by C and if we and if
we apply a voltage across this capacitor then there is a charge Q which is stored in this
capacitor.

Similarly in thermal networks we have C thermal which is the thermal capacitance and
we apply a delta T across this element and there will be pdT stored within this thermal
capacitance and we can write C is Q upon V or we can write I into T upon V. In thermal
we can write the thermal is equal to E upon delta T or pdT upon delta T where C is the
capacitance where the unit is farads. For the thermal capacitance the unit is joules per
degree centigrade. So these are the difference in the units but both behave almost similar
to each other. Using these elements for example thermal networks can comprise of
thermal resistance and thermal capacitance very similar to electrical networks which can
have electrical resistance and electrical capacitances.

We can make networks and solve using similar types of electrical laws. Now we discuss
some electrical RC networks and see how we solve those electrical networks and we
apply similar knowledge to solve the thermal RC networks. So let us move to electrical
RC networks. So when we discuss electrical RC networks the very first topic of
discussion is the RC delay because whenever we have an element which is resistance and
it is connected to a capacitor there exists a delay which is RC delay or the time constant is
RC and this delay is inherent to the network and let us discuss this in terms of some
examples. We have a system and we are sending a pulse through this system.

So we are sending a train of pulses through this system which has R and C components
joined together and at the near end or where we are transmitting these pulses we are
sending this very perfect kind of pulses. You can see that rising and falling edge are very
steep and it is almost a square kind of pulses. When we send this kind of signal at near
end, so this is our near end pulses, we are sending it very perfect signals but when we
receive these signals at the far end they don't exist like this when we receive at the far end
they look something like having some exponents. For example, they will look like this.
These are the waveforms at far end.

79
So we can see at near end we are sending very perfect signals but at the receiving end the
signal is distorted. This happens because of the capacitive effect. Because charging is
responsible for this kind of effect. So we can say that every node in the system, in an
electrical system, every node has a certain capacitance with respect to ground and when
we apply a transient signal this node need to be charged or discharged based on the signal
applied. This charging of each and every node it takes time because we have finite
amount of current which is provided by the input and this node need to be charged up to a
certain value to respond to input signal.

So this charging and discharging takes time and when we apply a fast transient the
system does not respond immediately to the input. So this is why we have this far end
signal which is very much distorted compared to the input signal which we apply. Now
let's say we discuss the RC delay of the system. We have a input node where we have a
resistance connected to it which is R and there is a node which is output node. This is Vin
and we have this capacitance which is connected to ground.

This is the simple RC delay model circuit and we need to analyze the behavior of this RC
delay of this circuit. So when we apply a input at the Vin and we are assuming two cases
when the output is zero and when the output is high when we have Vout we are plotting
this Vout with respect to time and we are assuming that initially V at time t e qual to zero
Vout is the voltage at the output node which is a low voltage. Now we apply a high input
at Vin and my circuit respond to this. My output voltage rises like this and it saturates to
certain value which is Vin. This is the case when Vin is higher than Vout at t equal to
zero. The other case is when the output voltage is already high and we are bringing it
down by applying a low input at the input voltage. This is our Vout at t equal to zero.
This is the value of Vin. In both of these graphs we can see the output voltage Vout tries
to achieve the value of Vin whatever it is. If it is higher than Vout it increases and if it is
lower than Vout it actually decreases. And the charging is exponential. You see there is
not a sudden kind of charging. There is an exponential behavior of Vout with respect to
time which we can see. And theoretically Vout approaches Vin in infinite time because it
is an exponential curve. So it never actually reaches the final value but it achieves a
certain value within some reasonable time. So now we can write our KVL and KCL for
this network. This is input which is applied. This is R. This is V. This node is Vout. And
this is Vin. This is R. This is Vout. Now we can write KCL which is off current law at
node out. The current into out from the left is Vin minus Vout divided by R. Which is this
current. And we can write the similar expression for the capacitor. Which is C dV by dt.
And because both of these currents are equal we can equate them. So we write Vin minus
Vout divided by R equal to E dVout by dt. Or we can write dVout by dVout divided by dt
is equal to one upon Rc Vin minus Vout.

So this is the equation we come up with and if we solve it the solution of this equation is
Vout which is a function of time is equal to Vin plus Vout at t equal to zero. Minus Vin

80
into e to the power minus t upon Rc. This is the solution of this equation and we can see
now the output voltage here which is a function of time this is a exponentially rising
equation where e is to the power of minus t upon Rc. So we can further rearrange it. Vout
as a function of t is equal to Vin.

Vin minus e to the power minus t upon tau plus Vout at t equal to zero e to the power
minus t upon tau. So in this equation we can see two things one is I have replaced Rc
with a constant tau because R and C are circuit parameters which we are not going to
change. So R and C combinedly we start writing with tau and which has a very important
function in these equations and we call it the time constant. This Rc has a unit of time we
can see t upon tau. So both t and tau have the same unit.

So this tau or Rc we call it time constant. The other thing we need to note is our Vout is a
function of Vin plus Vout at t equal to zero. It means the initial charge which is already
stored on the capacitor that is going to dictate my starting condition. So how much time it
takes to reach to certain value that not only depends on the input but also depends on
what is the previous state of the capacitor which is Vout at t equal to zero. Now we can
again plot these charging and discharging graphs. This is for charging. So in charging
case the capacitor is initially charged to Vout at t equal to zero and then we apply a Vin
which is let's say here and the output voltage rises exponentially something like this and
in one time constant which is tau. In tau time the output actually achieves the 63 percent
of the total pull speed. We can write this value is 0.63 times Vin plus 0.37 times Vout at t
equal to zero. In one time period if we assume that the capacitor was initially charged
was not charged this does not have any charge then Vout is zero it means one time period
the output will rise to 63 percent of the final value which is here. We can do this math on
this equation and we can find out the output voltage will be 63 percent of the input
voltage. This is for the charging case. For the discharging case when we have Vin is
lesser than Vout. The initial charge of the capacitor is Vout at t equal to zero.

We apply a Vin which is of lower value and the capacitor discharges from this Vout to
Vin exponentially. And it achieves the final value in infinite time because it's a
exponential graph. Similarly just like the charging case here also we have a time constant
which is tau and in one tau or one time period the discharging we can write is 0.63 times
Vin plus 0.37 times Vout at t equal to zero. So both charging and discharging case we can
say if the initial charge on the capacitor is zero then output reaches 63 percent of the
input voltage in one time period. So 63 percent of the transition completes in one time
period. So let's take an example for this. We have a RC network. And the input we apply
a pulse input which is Vin plus minus. The value of R is one kilo ohm and the value of C
is one kilo farad. We are assuming that Vin is zero for a very long time. It means the
capacitor in this network is fully discharged. We can say the Vin is zero for a long time.
Or we can say the capacitance is fully discharged.

81
Now what happens? This is the pulse we apply at input which is Vin and this is with
respect to time. The output will respond to this input and will start rising. This is the
output charging. And from the previous discussion we know that the output charges to 63
percent approximately in one time period. How much is the time period that we can
calculate from here? The tau or R into C which is the time period. The time constant is
equal to one kilo ohm multiplied by one picofarad which comes out one nanosecond. It
means that the output charging which is this black curve in one nano-second it will reach
63 percent of the input value. And the input value we are applying here is 10 volt. This is
a step of 10 volt. So that this value at one time period is 0.63 multiplied by 10 which is
6.3 volt. This value will be achieved in one time period. So this is how the RC charging
time constant actually works. This charging and discharging of the internal nodes actually
distorts the signal because as we saw in the very first slide we have a very perfect input
applied at the input. But when the signal reaches at the far end the signal actually
distorted that happening because of the charging and discharging of internal nodes. So
this phenomena actually gives rise to phenomena something called pulse distortion. If we
apply the pulses to this electrical network these pulses will be distorted by when they
reaches at the far end. So let us discuss what is a pulse distortion.

So first I apply an input which is Vin here. This is the input I apply and the output will
rise. This is the input and the output rises like this which is exponential curve. This is my
output. However the output will reach the input in infinite time theoretically because this
is exponential curve. If we wait long enough for practical purposes we can say the output
reaches input in few time constants. But if my input frequency is high for example I have
the same input but now instead of remaining high for a long time it comes back. This is
my input then my output will be something like this and instead of continue reaching the
steady state value immediately it will start following the input which is now discharging.
So we can see there is a difference between the input and output. So there is a difference
between these two values. Let us name it Vin and this is time. So there exists a voltage
difference and that is because the output did not have enough time to reach a steady state
value because the input is already being pulled down to zero. So the output will start
following the input and there exists difference in the final value and output never
achieves actually the final value. It remains lesser than that. Another thing happens if we
increase the input again it is like a pulse then the output again will start increasing and
again will start decreasing at this point.

So we can see the output is actually not rail to rail. The ground or the top supply the
output is always in between the value is not reaching the final value or the steady state
value because of the high frequency of the input pulses. If input is low frequency then the
output will have enough time just in first case. If the input is low frequency it means the
output will comfortably reach the steady state value and it will reach the highest value
which is let us say VDD. But for high frequency signals the output will not reach higher

82
or lower value completely and this gives rise to the difference in the voltage between
input and output. Now let us discuss three cases where we can have pulse distortion and
how this pulse distortion is affecting the shape of the pulses with respect to the frequency.
So we have pulse distortion. This is input pulse positive and negative. This is R. This is C
and this is ground. This is output node. And here we see three different cases. This is V
out with respect to time. This is pulse width which is 0.1 times RC. The pulse width or
the pulse duration is 10 percent of the RC time constant.

The second case we take for pulse width of equal to RC. Here we have pulse width equal
to 1 time RC or equal to 1 time constant. And the third case we take where the pulse
width is long enough. The pulse width here is let us say 10 times RC. So now we have
three cases where the input frequency is very high. It means the pulse width is very small
or equal to 0.1 RC. One frequency is moderate which is pulse width of equal to RC and
one frequency is very low frequency where the pulse width of the signal is 10 times RC.
So let us plot the inputs for a pulse width of 0.1 RC will be a very short pulse. This is the
input pulse. For pulse width of 1 RC, the pulse width will be bit longer compared to the
first one. And this is the input pulse. And for 10 RC, pulse width will be very long. So in
case of pulse width equal to 10 RC, let us consider the low frequency case which is easy
to understand. The output has enough time to charge and discharge. It can reach let us say
high value and again to the low value.

So we can see the pulse is very good shape. This is the output. We can see the pulse is
good. In case of pulse width equal to 1 RC, the frequency is moderate. It is not very low
but not very high as well. Now we see the charging of this output, that output node, the
characteristic remains same. It will take 63 percent, it will rise to 63 percent of this full
value. This will be like 63 percent. And by reaching this, it will take time T which is
equal to the pulse width. When it reaches 63 percent, the input will be again coming back
to ground. That means this will again come down here and it will start discharging. Now
we can see the output pulse here is not as good as in the case of pulse width equal to 10
RC.

The pulse here is very good but here the pulse has not even reached the top value, already
reaches to 63 percent of its value. And then again goes down. Now the third case which is
pulse width equal to 0.1 RC which is a very high frequency which we are applying to this
RC network. The pulse width is very small and the network, the output node, it still take
0.1 time period to reach 63 percent and our time period, the pulse width is not even one
time constant. So it just start increasing and by the time it reaches this point, the input
again make a transition to ground and the output again follows this. So the output node,
the output voltage will be like this and we can see for a input of let us say this top value is
5 volt across all the pulses, the output does not even rise to typically 1 volt or so. So for a
pulse width of 10 RC, the output reaches comfortably to higher value. Pulse width equal
to RC does not reach to maximum value, it only reaches to 63 percent.

83
But for a very high pulse, for a very high frequency pulse width equal to 0.1 RC, output
does not even reaches to midpoint of the voltage which is not good because this small
change in the output will not be detected comfortably with the next stage. So this is the
effect of frequency. This is the effect of frequency on the circuit performance. The output
node does not even rises to its higher value and that is happening not because of the
circuit parameters but because of the input frequency which is high enough and circuit is
not able to respond to that input signal. This is happening because of the inherent time
delay which the time is taken by the internal nodes to charge and discharge.

We have a RC network where R is 2.5 kilo ohm and C is 1 nano Farad. If voltage pulse is
applied. The voltage pulse has a width of 5 microsecond and a height of 4 volt. It is
applied at the input and we need to sketch the output of the circuit. Now this is the input
where we apply a voltage which is initial value is 0 and the height is 4 volt and the time
period is the width of this pulse is 5 microsecond.

This time is 5 microseconds. This pulse is applied at the input. We are assuming that the
capacitor initially is discharged. It does not have any charge initially with respect to
ground and we need to plot the output. So this can be solved using our RC charging
equation. So the first the output voltage will increase to approach the 4 volt because when
the input is transitioning from 0 to 4 volt the output will try to reach this 4 volt and the
charging is governed by that RC time constant and when after 5 microsecond when the
input is transitioning back again to 0 the output will follow this the input and again goes
back to ground.

We can plot the output versus input. It will look like this. We have this V out with respect
to time and the input is 0 at 0 and the pulse is applied at T equal to 0 which is at 4 volt
and it goes down to 0 at 5 microsecond. So this is 5 microsecond and this is 4 volt. Now
the input will rise, the output will rise. It reaches to its maximum value whatever is there
at 5 microsecond and again goes down following the input and the characteristic of this V
out we can find out using the RC charging equation. For charging we can write V out is a
function of time which is 4 minus 4 e to the power minus T upon tau which is 2.5
microsecond tau is R into C which comes out 2.5 microsecond minus T upon 2.5
microsecond. This is for the charging where the time is less than equal to 5 microsecond
and for the discharging case the initial value on the capacitor will be this value. It's not
the 0 because the capacitor is initially charged now when it is starting going down the
initial value is not 0 but this value and we can write the output is 3.44 e to power minus T
minus 5 microsecond upon 2.5 microsecond. This is for T greater than 5 microsecond. So
this is the equation of charging and this is the equation of discharging when the output is
following the input. So this is how we calculate the RC delays of electrical networks.
Next we will see some examples for the thermal networks and we apply the similar kind
of concept for the thermal networks where we have thermal resistances and thermal

84
capacitances instead of the electrical resistance and electrical capacitance which we just
had here. So next we will discuss the thermal networks.

That's all for today.

Thank you.

85
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 9
Thermal Sensors: Thermal RC Networks

Hello, ah welcome to the course Transducers for Instrumentation. Last lecture we were
discussing about the electrical and thermal networks and we saw that the thermal
networks can also be solved just like the electrical networks. For example, the resistances
are replaced by the thermal resistance, the current flow is replaced by heat flow, the
temperature difference is corresponding to the potential difference. So, the similar kind of
laws just like the Ohm's law they are applicable to the thermal network as well. So, today
we will discuss some thermal resistances and how we can connect them in series or
parallel and how do we solve them. So, we discuss thermal resistances in parallel,
thermal resistance which are in parallel. So, for example, we have these two materials are
material A and material B with some thermal conductivity of kA and kB and we want to
investigate the thermal flow when these materials are connected in parallel. So, for
example, we have surface here which is surface of A and which is this is surface of B. So,
the first assumption we are making is across the surface of this material A, this surface
the temperature remains same. It means the temperature of the surface at this point is
same across the surface.

So, this is the assumption just like a point contact in electrical networks. So, similarly we
have the same assumption in thermal networks. So, this point has a temperature which is
same across the surface. Similarly, for point for material B this point has the same
temperature across the surface. So, this is the assumption we are making. Similarly, on
the other surface the far surface we have this point where the temperature is same across
the surface and we assume that there is no thermal conduction between this surface. So,
there is no thermal power dissipation is not there across this. So, given these assumptions
now we have temperature difference which we apply across this assembly and we want to
investigate how the heat flows. So, let us say we have a temperature which is T1 or Ts1
whatever we call it. So, this is the temperature we apply at the near end and at the far end
let us say the temperature is Ts2.

Now, we can write the resistance of each one of this material is for example, that R
conductive for a material A is equal to L upon kA which is the conductivity multiplied by
the area. So, here the L is this and the area is cross section area of this material. So, this is
the R conduction for material A. Similarly, for R conduction B will be L upon kB into A
the area of for example, the area of material A is A subscript A and the area of material B
is A subscript B. So, these are the thermal resistances of these material A and B and
because they are in connected in parallel we have the overall resistance which is 1 upon R

86
total equal to 1 upon R A plus 1 upon R B or we can say this comes out 1 upon L upon
kA into A A plus 1 upon L upon kB into A B. So, this is the total resistance which is R
total of this whole assembly and this is very similar to the parallel connection of electrical
resistances. Now, we can calculate the heat flow across this whole assembly when we
apply this temperature difference which is TS1 on the on the near side and TS2 is on the
far side. So, we can calculate the heat flow which is equal to the temperature difference
which is TS1 minus TS2. This is the temperature difference just like your potential
difference divided by the total resistance which is R total. So, this is the total heat flow
which is equal to the difference in the temperature divided by the total thermal resistance.

This is very much equal to the electrical current which is flowing through the material
that is equal to the voltage difference divided by the resistance. So, this is the parallel
connection of these thermal resistances though we can more generalize it by considering
that one source will be supplying this heat energy to this assembly. So, let us consider a
little bit more generalized case. We have this material assembly. This is material A, this
is material B with thermal conductivity KA and KB respectively and these are the points
at the surface. Now, we assume that there is a source of heat energy which is here at the
temperature T1. This is supplying some heat which is going to both of the materials one
here from this point to this point and one from here. This mode of the heat transport is
convection because there is no direct contact between this point and this point. Similarly,
on the far side we have a receiving end which is at temperature T2 and this heat is
flowing from this point to the far end. This is also the convection mode and inside the
material the mode of heat transport is conduction. Now, we can write the resistance
similarly to the previous slide where we have 1 over R total equal to 1 upon RA plus 1
upon RB the resistance of the path A and the resistance of path B. Now, the path A has 3
subsections. One is this path where the heat is flowing from T1 to material A through the
convection mode and this is the second section where the mode of transport is conduction
and heat is flowing from this point to this point and the third section is where the heat is
again flowing through convection from the material A to the point T2. So, we can write
the overall resistance of the upper path like RA the thermal resistance through path A is
equal to 1 upon H1 into AA. Here the H this H is the coefficient for convection.

So, 1 upon H1 into AA which is the area of this surface AA plus the second section
which is conduction which we can write L upon KA into AA very similar to the last slide
and then again the third section which is again convection and we can write 1 upon H2
into AA. So, this is the overall thermal resistance through the path A and similarly we can
write for the path B which is again 3 section this is first section this is second section and
this is third section and we can write for RB is equal to 1 upon H1 AB plus for the second
section which is conduction that is L upon KB into area of B plus the third section which
is again convection 1 upon H2 into area of B. So, these are the respective values of RA
and RB and we can calculate the total resistance 1 upon R total is equal to 1 upon H1 AA

87
plus L upon KA into AA plus 1 upon H2 into area of A plus 1 upon 1 upon H1 into area
of B plus L upon KB into area of B plus 1 upon H2 into area of B. So, this is how we can
write the total resistance of material of two materials when they are in connected in
parallel and we can also write this equivalent model of this whole assembly that we can
write something like this. We have the first point which is T1 here similar to this we have
T1 and then we have the first resistance which is in the form of convection and the value
of this is 1 upon H1 into AA. The second element in this is because of the conduction
mode and that is L upon KA into area of A and the third element which is again because
of the convection to this temperature T2. The value of this is 1 upon H2 into area of A.
So, this is the model for the upper path the lower path is also similar to this. The value of
this we can write similarly 1 upon H1 into area of B this is L upon KB into area of B and
this is 1 upon H2 into area of A. So, this network is very similar to the electrical network
and we can apply any temperature difference T1 and T2 and calculate how much the heat
will flow from the point T1 to the point T2.

We can apply all kind of Ohm's law to calculate all this heat flow from T1 to T2. So, we
can write the heat flow for this assembly which is again similar to the last slide which is
T1 minus T2 divided by R total. R total is the whole combination of this model. So, this
is the thermal resistance connection in parallel. Now, we consider little complicated case
where we have multiple materials joined together in terms of series and parallel and see
how we can calculate the heat flow in that whole assembly. So, we calculate thermal
resistance of series parallel network. So, for example, we have assembly where we have
different materials. This is material for example A. Then we have material B and we have
material C and another material is connected like this which is material D. And
correspondingly we have the thermal conductivity of these materials which is Ka, Kb, Kc
and Kd. And the length of these material is for example LA and LB. LB is equal to LC
and this is LD. So, this is the whole assembly of materials and we want to calculate how
much heat will be flowing through this assembly if we apply a certain temperature
difference across this assembly. So, let us say we apply a temperature difference which is
T1 this side and this side is T2. Whole area of this is A. So, now we need to make an
assumption here. Similarly in the first slide we had assumption that at the point of contact
the temperature is same across the surface. So, similar assumption we need to make here
that there is no temperature difference perpendicular to the x axis. For example, the heat
is flowing in the x direction. The heat flow is this side and there is no heat flow
perpendicular to this.

Means the surfaces which are normal to x direction they are isothermal. The heat will not
be flowing from material B to material C. The heat will be flowing only from material A
to material B and material A to material C and then back from material B to material D
and material C to material D. So, this is the assumption we need to make. This is the
assumption that surfaces which are normal to x direction are isothermal. So, they are at

88
the same temperature of each other and there is no heat flow in the perpendicular to x
direction only the heat is flowing from T1 to T2 in x direction only. So, now we have we
can treat them like a pore resistance where we have this material A which has certain
resistance. Then we have two resistances corresponded to material B and material C and
then back to material D. And the value of these resistances we can calculate just like for
the conduction case which is L upon L A upon K A into area of A and this is for material
B which is L B divided by K B into area of B which is actually half of the area of A but
we can write for generality area of B this is material D and we can write L B L C divided
by K A K C into area of C and this is material D we can write L of D divided by K D into
area of D and the temperature applied is T1 this side and T2 this side. So, we can solve
for this series parallel connection which is R A plus R B parallel R C in series with R D
that will be the total resistance and the heat flow we can calculate T1 minus T2 divided
by the total resistance.

The heat flow for this network will be T1 minus T2 divided by the R total and R total we
can calculate of this thermal resistance this is R total. So, this is how we can solve a
thermal network very similar to the electrical networks. We need to discuss one more
point of the thermal resistance which is equal to the contact resistance of electrical
network. So, similarly in thermal networks we have thermal contact resistance very
similar to the electrical contact resistance. So, this is the thermal contact resistance. So,
when we have two materials that is a material A and we join this with material B this is
material A this is material B and there is a heat flow from the left to the right. Now when
these two materials are joined together they behave differently at the surface is where
they are having a contact with each other that is called the contact resistance which is
different than their internal their intrinsic resistances. So, the reason for this is for
example, we have this surface which is in contact of the surface of material A is in
contact with surface of material B. When we join them together we assume that the whole
surface is in contact with each other, but in practicality this is not the perfect contact. If
we zoom in at this point if we zoom at this point very closely and we see the surface is
there is actually not in very much contact. If we assume that this should be the point of
contact or the physical contact actually the material A let us take the material A with blue
color the material A surface is something like this because of the surface roughness
material A is like this and the material B the surface is like this. So, we see that there is
no proper contact between material A and material B because of the surface roughness
they are not very much in close contact. So, now if the heat is flowing from material A to
material B that heat will have two paths one is the path where they are in contact here
material A is in contact with material B. So, heat will have different resistance which is
let us say R1 this resistance will be different than this resistance where these materials are
not in contact this blue curve is very far away than the green curve and we have the
different resistance R2 and we can calculate whole of the resistances of this which are
different different values because of the surface roughness and this gives rise to the

89
contact resistance or the thermal contact resistance. Thermal resistance arise due to
surface roughness and this is again just like a electrical contact resistance this thermal
contact resistance is also a problem in proper heat conduction and that is why if we want
to have a good heat conduction from material A to material B generally we try to put
some kind of epoxy on the surface of this.

So, that the amount of surface which is in contact with each other that increases and this
contact resistance actually goes down. So, we sometime put some epoxy on the thermal
conductive epoxy to reduce this contact resistance. So, this thermal contact resistance is
very similar to the electrical contact resistance where we apply some 4 point Kelvin
probe method to get rid of the contact resistance similarly, we have thermal contact
resistance in thermal networks. So, next we discuss the thermal RC networks we will this
consider some practical cases where we have multiple assemblies and we try to
investigate how much the heat will be flowing from the source to the ambient
temperature and we will we will consider some real applications where we can apply
these thermal networks and solve for the thermal power dissipation. So, we consider the
case of a IC package where we have a semiconductor IC which is silicon based IC and
we have some epoxy connected to it and then some heat sink connected to it the transistor
on the silicon chip will be generating some heat and how that heat will be transferred to
the ambient temperature. So, let us say we have a semiconductor which is silicon and on
this we have P plus device made of P plus material P plus doping on the silicon below the
silicon we have some dye attach and below this we have some heat sink. So, we have
now 4 different materials here the first material is the P plus or P plus well we can
assume it is like a resistance which is generating heat when we apply a voltage across it
and this P well has certain thermal resistance which is Rth let us say this is 1 and certain
capacitance the thermal capacitance let us say Cth1. Similarly, we have another material
which is silicon and Rth2 is the thermal resistance of silicon and Cth2 is the thermal
capacitance of silicon. The third material here is this dye attach and thermal resistance is
let us say Rth3 and Cth3 is the thermal capacitance and after this we have the lead frame
or after that we will have heat sink. So, now we have this whole assembly which is
practical assembly in any electronic circuit.

Now, we need to make a model of this whole assembly so that we can solve it for power
dissipation. So, we can make a model some similar to this is the delta T which is the
temperature difference being generated by this by this P plus this is T junction then we
have a thermal capacitance connected to it which is Cth1 and Rth1 then this is connected
to silicon which is Cth2 and Rth2. Then again the dye attach is connected to it which is
Cth3 and Rth3. So, this is the whole assembly till this dye attach and further a heat sink is
connected to this let us take the different color here this heat sink is connected which is
dissipating this heat to outside environment which can be modeled just like a thermal
capacitor and the thermal resistance which is in low in magnitude.

90
So, this is heat sink. Now, when the junction that or the device inside this chip they are
generating heat that heat need to be flow outside to the environment because the junction
temperature has its limits the junction temperature should not exceed a certain value. So,
that the device can perform its operation satisfactorily. So, we need to have this whole
network efficient enough so that the heat generated inside or this delta T the heat can be
transferred from this junction to the outside environment. Now, typically in data sheets a
value of Rth is given for DC case this the value of Cth the thermal capacitance is not
given generally because we assume that in steady state condition or where the heat has is
flowing in steady state condition there is no transient there this capacitance can be
omitted. So, this thermal capacitance is exactly like our electrical capacitance where in
the case of DC currents the capacitance can be removed similarly for the steady state
condition of heat flow these thermal capacitances can also be omitted and we can easily
calculate how much current will be flowing just based on the thermal resistance which is
here in series Rth1 plus Rth2 plus Rth3. So, in data sheets typically Cth value is not
included. Because, in case of steady state this can be omitted. So, the steady state
condition for thermal RC network is very easy to solve we just need to remove all the
thermal capacitances and just the thermal resistances will be added in series or parallel
whatever is the case and we can calculate how much heat will be flowing based on that
we can calculate we can calculate how much junction temperature will rise and this rise
need to be within the limits of semiconductor devices specified by the data sheet. So, next
so this is the thermal RC network for DC conditions or steady state conditions. Let us
calculate one example a switch has its resistance which is rds on to 24 milli ohm this is
the electrical resistance and the current flowing through this switch is 5 amp i is 5 amp.

So, this is electrical switch 24 milli ohm and the current is 5 amp. The maximum thermal
resistance Rth maximum Rth is equal to 55 degree centigrade per watt ambient
temperature which is T ambient or Ta is 85 degree centigrade we need to find the
junction temperature. So now first we can see this we have a electrical resistance which is
dissipating some power then we need to calculate how much power it is burning that
power will be in terms of heat and that heat will be flowing from this resistor to the
ambient temperature to the ambient and the thermal resistance of this resistance is 55
degree centigrade per watt and ambient temperature is given 25 how much junction will
temperature will rise. So, first we calculate how much power is being generated. So, the
power dissipation is equal to I square R 5 square multiplied by 24 milli ohm this comes
out to be 0.6 watt. So, this 0.6 watt of power is being generated by this electrical
resistance and this power because this is burned inside this resistance this is in terms of
heat this heat is going to flow outside depending upon the junction temperature difference
and the ambient temperature. So the thermal resistance is given is as 55 degree centigrade
per watt means for a heat flow of 1 watt junction temperature will rise 55 degree
centigrade. So, the total junction temperature will be equal to the ambient temperature
plus the temperature rise because of the heat flow. So, we can write T A which is the

91
ambient temperature plus the thermal resistance which is 55 degree centigrade per watt
into the power dissipated which is 0.6 watt. So, this temperature is at T 5 plus at T 3 it
comes out to be 118 degree centigrade. So, this is T junction means the junction
temperature will rise to a steady state value of 118 degree centigrade when the ambient
temperature is 85 degree based on and this temperature rise is due to the electrical power
dissipation which is being happening inside the resistance. So, this is the simple case of
DC condition where we are assuming that there is no transient response and the heat is
flowing continuously from the resistance to the ambient temperature. Next we consider a
more generalized case of transient response where the heat is actually being generated
through pulses it is not a steady state condition where the source is at a particular
temperature and the heat is flowing steady state.

Now the source which is a electrical device that is being turned on and off and that will
generate some transient response in the heat flow. So, that we consider in transient
response. So, if we try to solve the same model which we just saw here this is the model
of this thermal network which is very similar to electrical network and we try to solve
with this series and parallel connection of resistance and capacitance we will get
expression very similar to what we have in electrical networks and that total impedance
now we call it Zin in terms of omega which is the angular frequency. So, the model will
look something like this which is 1 upon 1 1 upon j omega C Thevenin plus 1 upon R
thermal 1 plus 1 upon j omega C thermal 2 is C thermal 1 plus 1 upon R thermal 2 and
plus this will continue till the last element it depends on how many number of elements
we have we will get this kind of expression. Now looking at this expression we can say
that this model is fairly complex and this is difficult to solve we have in practical cases
we have multiple thermal resistances and thermal capacitances. So, the number of terms
in this expression will be very high. So, we do not want to solve this kind of model for
our practical network.

So, what we have in real practice is a thermal graph and this thermal graph is generated
by industries based on that they do extensive characterization of their assemblies and they
come up with this thermal graph using this thermal graph we can apply a pulse and
calculate how much the thermal resistance will be there for that particular assembly and
this thermal resistance is particular to that thermal assembly it is not usable by any other
assembly. So, this thermal graph looks something like this. So, this is a thermal graph
which is generated by the industries by extensive characterization. So, what this thermal
graph says on the y axis we have this junction thermal resistance and on the x axis we
have the t pulse which is the time period of the pulse. Now in the case of periodic event
when we have pulse applied to the network let us say this pulse rises at this point and fall
at this point. So, this time between rising and falling this is the time period time which is t
pulse compared to the total time period which is t and the duty cycle of this pulse is given
by d which is equal to t pulse upon t. So, in this particular case we have duty cycle of 50

92
percent because half of the time the pulse is on and half of the time the pulse is off. So,
this thermal graph has multiple tracks on this. This for example, this is the case for single
pulse where we just apply one pulse it turns on and turns off and it remains there there is
no extra pulses. So, in the case of single pulse we take this graph if the duty cycle of our
pulse which is applied here is 1 percent then we take this graph 1 percent if we have 10
percent we take this graph and if the duty cycle is 50 percent we take this graph.

So, this is how we take this thermal graph and we calculate how much the thermal
resistance of our assembly will be. So, using this graph we can calculate our thermal
resistance instead of solving this complex model. So, instead of this complex model we
go for thermal graph. Let us take one example of transient temperature. In this example
we have a certain assembly and we apply a single pulse of 400 watt to this semiconductor
package for 200 microseconds. So, let me draw this pulse. So, we have a pulse which is
this pulse is 400 watt in magnitude and the pulse is 200 microsecond. This is time this is
200 microsecond and this is power. So, this is the pulse which we apply to a certain
assembly for which we are given the thermal graph and we need to calculate how much
junction temperature will rise for that particular assembly. And the ambient temperature
generally it is 25 degree centigrade or the T ambient is 25 degree centigrade. Now we
need to calculate the junction temperature. So, how to solve this? We are given that the
power dissipation is 400 watt because the pulse the power of this pulse is 400 watt. The
thermal resistance now we need to calculate that for this assembly how much will be the
thermal resistance. Remember if we actually solve the package for this thermal resistance
we may have multiple thermal resistances. For example, the silicon will have its own
thermal resistance ceramic package dyes bonds every element has will have its own
thermal resistance. So, instead of using that we go for a thermal graph and in thermal
graph we need to know what is the duty cycle.

So, here it is a single pulse it is a single pulse we need to calculate the thermal resistance
of this for 200 microsecond. So, let us go to the thermal this is the thermal graph and for
single pulse we need to take this graph which is for this single case and we can see for
200 microsecond here it is around 200 microsecond the value will be around 0.083
degree centigrade per watt. So, this we can calculate 0.083 through this thermal graph
and we take this back to our calculation. The thermal resistance is equal to 0.083 degree
centigrade per watt. Now the total junction temperature will be equal to T ambient plus
the temperature rise which is the power dissipated into the thermal resistance which is
equal to 25 degree plus 400 into 0.083 this comes out to be 58 degree centigrade. So, the
junction temperature Tj will be 85 degree centigrade and if we plot this rise in the
temperature it will look something like this we have this time axis and this is the junction
temperature which will rise when there is a power dissipated. This will be originally at 25
degree centigrade this is 25 and this will rise to its peak value and then again fell down
and this peak value will be 58 degree centigrade.

93
So, this is how we can calculate the junction temperature of the whole assembly. So, this
example is in case of single pulse where we have applied just only one rising and one
falling edge. Let us take one more example where instead of one single pulse we apply
the train of pulses to this assembly and calculate how much the temperature will rise. For
comparison we will take the time period as same as 200 microsecond, but now we will
take the power which is very less compared to the last slide. Now we take 1.44 watt only
as the peak power. So, this is the pd, this is time and we apply. We apply these pulses
which is in 1.44 watt of magnitude and the time is t pulse is 200 microsecond and duty
cycle is 50 percent. The ambient temperature is same as 25 degree centigrade. Now we
want to calculate how much junction temperature will rise if we apply this train of pulses.
So, the power dissipation is same as given to us. The power dissipation is given to us as
1.44 watt. We need to calculate the thermal resistance. So, this thermal resistance we
need to calculate through the thermal graph. Our duty cycle is 50 percent and the time t
pulse is still same as 200 microsecond. So, we go back to that thermal graph. Here for the
duty cycle of 50 percent we need to take this graph which is shown here. This is for 50
percent duty cycle and for the same 200 microsecond, this is approximately 200
microsecond. The junction temperature, the thermal resistance Zth comes out to be 23
degree centigrade per watt. This is approximately 23. So, this reading we can take from
thermal graph and put it back here which is 23.

Now we can calculate the junction temperature is equal to ambient temperature plus the
temperature rise which is Pd into Zth thermal resistance. This is 25 degree plus 1.44 watt
into 23 degree centigrade per watt. This comes out to be 58 degree centigrade. This is the
junction temperature. So, this is the case in terms of periodic pulses where we have this
periodic pulse applied to the assembly and this is the temperature rise. If we plot this
junction temperature with respect to time, the graph will look something like this. We
have initial temperature which is 25 degree and the temperature will rise when the pulse
is applied. Then it will go back, then again it will rise, then it will go back, then it will go
rise, then it will go back. So, it will achieve a peak temperature which is given by 58
degree centigrade. So, in steady state condition when we keep on applying these pulses,
the temperature will rise to 58 degree centigrade. One thing to note here is in the last slide
we applied 400 watt of power, but the pulse was a single pulse. So, the temperature rise
was also 58 degree, but if we apply a train of pulses even if the magnitude is very less
which is 1.44 watt compared to the 400 watt earlier, the temperature rise is still same
because these periodic pulses will keep on raising the temperature of the assembly and
causes a more rise in the temperature which is not beneficial for semiconductor devices.
So, this is how we solve all these thermal networks. It is very similar to the electrical
networks.

So, this is all for today.

Thank you.

94
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 10
Thermal Sensors: Silicon Based

Hello, welcome to the course Transducers for Instrumentation. Last lecture we discussed
about sub-conventional thermal sensors for example, thermistors, RTDs, thermocouples,
they are all like conventional temperature sensors which we are using from quite some
time and then we discussed some thermal RC networks. Now today we will discuss some
silicon based sensors which we can actually integrate with all silicon devices and make
them miniature. The kind of motivation for using these silicon based thermal sensors is
silicon based technologies are very much matured and we can scale them down to save
the cost and they take very less amount of area compared to other thermal sensors. So we
want to integrate these thermal sensors along with other devices like MOSFETs and
when we can integrate them, we can join them with some signal processing circuits and
we can directly take the output in form of some digital data which we can process further.
So going with the silicon based devices or silicon based sensors is very much
advantageous in terms of the complexity, the amount of cost it involves to fabricate one
sensor.

Though the cost of single sensor can be higher but if we scale them and we can make
them silicon compatible and we can MOS production, we can produce them in mass then
the cost of these devices drastically comes down. So today we are going to discuss some
thermal sensors which are silicon based or which are compatible with silicon processing
technology. So we discussed today thermal sensors which are now silicon based. So we
are going to discuss silicon based.

So we need to have some property of silicon which is proportional to the temperature. We

need to find out a property y of material which we can use it as a transduction principle
and that property is changing almost linearly proportional to the temperature. So if we
characterize the silicon, we know that the thermal diffusivity of silicon is very much
proportional to the temperature. So before this actually these integrated sensors are
widely used because of their small size, low cost and ease of fabrication and the output is
digital. Traditionally we used BJTs or the bipolar junction transistors where the behavior
of these BJTs are very much proportional to the temperature and BJTs are intrinsically
sensitive to the process spread.

When we use these BJTs and if there is a process variation because of some process
variation or the temperature variation or some other fluctuations in the manufacturing
process, the performance of BJT is very much sensitive to these change in the process. So

95
they are not very much immune to these process variations and that is the problem with
BJT in using BJT as a temperature sensor. Though this can be improved by trimming
individual devices, for example we make a BJT using a very controlled process but it is
still sensitive to fabrication process. We can trim it down individually, we can reduce the
number of fingers and so many other things but the cost of trimming individual device is
very much high. So we don't want to do that individual processing.

So BJT in that sense is not very much advantageous because it is very much sensitive.
And the other thing about BJT is this BJT suffers from the exponential increase in the
leakage current with respect to temperature. So BJT we use earlier as a temperature
sensitive device and this is the case with the BJT. But the problem with BJT is this is
intrinsically sensitive to the process spread. So BJT is very much sensitive to the process
spread.

What we mean by this process spread is when we are manufacturing these BJTs, let us
say we have a process spread. So BJT is very much sensitive to the process spread. We
have let us say emitter, this is base and this is collector and we are doing this doping here
in the emitter and collector. So the performance of this BJT is very sensitive to this
doping. If there is a 5 or 10 percent variation in this doping, then the current is very much
sensitive to the process spread.

Actually it is very much sensitive to this change in the doping. So this is the problem with
BJT having a wider process spread compared to the other devices. As we said this can be
improved by trimming individual device. We can have certain techniques to reduce the
current if the doping is high but that individual trimming is very much costly in terms of
time as well as money. So individual trimming is possible.

This can be improved by trimming individual devices but increase cost. So this is the
problem with BJT. And the other problem of BJT is the exponential increase in the
leakage current when the temperature increases. So the leakage current of BJT is very
much sensitive to the process spread. So the increase in the leakage current exponentially
increases.

So we see that BJT increases exponentially and we use BJT as a device to sense the
temperature and we form some sort of band gap reference circuits to detect this change.
For example we have a typical band gap reference circuit which looks something like
this. This is R1. This is R2. This is R3. This is R4. This is R5. This is R6. This is R7.
This is R8. This is R10. This is R11. This is R12. This is R13. This is R14. This is R15.
This is R18. This is output node and here we have this is the positive connection and this
is the negative connection. This is our BJT T2 and this is BJT T1. This is R3. This is R4
and if there is a current flowing in this arm which is IC1 and this is IC2, this is our BJT
T2. This is our BJT T2. Then we can write the delta VBE, the change in the base to

96
emitter voltage is equal to KT upon Q where k is the Boltzmann constant, T is
temperature and Q is the charge into log natural IC1 upon IC2. This is our BJT T2.

This is our BJT T2. This is our BJT T2. So, here we can see this change in the voltage
delta VBE which is sensed by this comparator. This change in the voltage is proportional
to the absolute temperature T and if temperature is changing, then we have a large change
in the base to emitter voltage. So, in this way this circuit is actually detecting the
temperature change, but the problem as we discussed here is the process spread. Our BJT
is very much sensitive to the process spread.

If there is a slight fluctuation in terms of doping in terms of the area or the geometrical
lengths, the parameters, then this change very much affects these IC1 and IC2 and
because this is a delta VBE is now proportional to IC1 and IC2 as well. So, this is the
process spread and this circuit is very sensitive to this process spread or this change in the
current and this can be tuned, but the cost of tuning is very high and of course, the
leakage current which we have not considered here in this formula, but the leakage
current also increases exponentially with the temperature. So, these are the problems that
exist with the BJT devices when we use BJT as a temperature sensing device. So, let us
discuss some other type of temperature sensor where we do not use BJTs, we use some
inherent property of silicon to detect the temperature. For that actually we need to find
out a property of silicon or a property of material which changes with the temperature.

So, for silicon we know that the thermal diffusivity of silicon is a very good parameter
which actually has a temperature dependence. So, we have a thermal diffusivity of silicon
and this is the temperature of silicon. So, this is the temperature of silicon and this is the
temperature of silicon has a temperature dependence which is 1 over T to the power 1.8.
So, this is the relation of temperature with the thermal diffusivity of silicon.

So, this is the thermal diffusivity in case of pure silicon and because the silicon we use in
IC fabrication process is very much pure. So, the thermal diffusivity actually shows this
kind of relation in real life because the silicon is very much pure. Now, the question is
what the thermal diffusivity is? So, thermal diffusivity is the rate of heat flow in the
silicon is the thermal diffusivity. For example, we have silicon material the one side is at
temperature T 1, other side it has temperature T 2, then the heat will flow based on the
difference of temperature T 1 minus T 2. The rate of this heat flow this is called thermal
diffusivity and this rate of flow depends on the actual temperature of silicon.

Let us say the silicon slab is at temperature T. So, the heat flow from T 1 to T 2 depends
on the actual temperature of silicon which is T. So, the thermal diffusivity of silicon here
is proportional to 1 upon T to the power 1.8. So, this is the relation experimentally
measured for pure silicon and thermal diffusivity as we discussed let us say we have a
slab of silicon. This slab is at temperature T and on one side I have a supply temperature

97
T 1, this is T 1, the other side temperature is T 2. So, the heat flow is proportional to T 1
minus T 2, but the thermal diffusivity the rate of this how much fast this heat will flow
from T 1 to T 2 that depends on 1 upon T to the power 1.8. So, this is the relation
experimentally known and now we want to use this information to make a thermal sensor
and because this is all silicon base this is very much integrable. So, this is the relation
experimentally known and now we want to use this information to make a thermal sensor
and because this is all silicon base this is very much integrable.

So, this d or we call it the thermal diffusivity which is d here, this d can be determined by
the energy of the silicon. So, this is the energy of the silicon. So, this is the energy of the
silicon by measuring characteristic of ETF or the energy of the silicon. So, this is the
electro thermal filter ETF is called electro thermal filter. So, the sensor which we can
make will something look like this. We have a piece of silicon and this is the silicon. So,
this is bulk silicon. In the silicon we grow a thermal oxide on top. Here, we have a layer
of oxide and in this silicon on top of this silicon we have a layer of oxide. In this we have
oxide, we fabricate two elements one is the heater and one is the temperature sensor.

So, we have a heater here, this is a heater. So, this is the heater and apart from this let us
say we have 100 micron distance, we fabricate another structure which is a temperature
sensor. So, this is the temperature sensor. This is temperature sensor. Now, this distance
is very much precise. That we can decide based on our layout and design technique. This
100 micron we can fix it as per the geometry of this sensor. So, now we have this silicon
on top we have oxide and we have two elements which is heater and the other element is
temperature sensor. So, this is the temperature sensor. We have a filter which is a
temperature sensor. This is the temperature sensor. Now we have this heater and the other
element is temperature sensor. Now, this silicon has a temperature of T. Let us assume
this the temperature of this slab is T. Now, this heater this can be excited electrically
which we can apply a electrical pulse and this is the temperature it will generate the heat
accordingly. So, we have let us say the input is a pulse something like that.

So, this is the input which we are applying to the heater. Now, when the input is high the
heater is going to generate heat and when the input goes low the heater is going to shut
down. So, if we see the heater is going to generate the heat pulses as per the input
applied. Input is a pulse. So, the output of heater will also be pulsating in nature, but that
will be heat way heat pulses. So, this is the heat pulse. So, this is the heat pulse. There
will be heat pulses generated here by this heater. Now, this heater is surrounded by all
silicon. This is this heater is residing in the bulk silicon. So, the part which is very close
to this heater of this silicon is going to heat up and the other the temperature of this bulk
silicon is at different temperature which is T.

It means the heat these heat pulses generated by this heater are going to diffuse now in all
the direction. So, this heat pulses which are being generated here they will be kind of

98
diffusing in all the directions. They will go in all the direction in bulk silicon. They will
not go on the top because the oxide is not a very good conductor of heat. So, these heat
pulses are going to diffuse now in silicon in all the direction. So, these heat pulses which
are generated by this heater they are now going here, here, here, here and this direction as
well. Apart from this heater 100 micrometer which is the precise distance we have placed
a thermal sensor or the temperature sensor. We will discuss what that kind of temperature
sensor is that in the next page, but let us say we have a temperature sensor which is
placed 100 micrometer apart from this heater. This temperature sensor is going to sense
now these pulses. So, this heat pulses which are transmitted by this heater they travel
from heater all the way through temperature sensor.

They appear here and this temperature sensor will produce its output based on those heat
pulses and the output of this temperature sensor will be electrical in nature. So, at the
output here I get a electrical signal which is proportional to the heat pulses received by
the temperature sensor. So, at the output what we are going to receive is like heat pulses
something like this. They will not be as sharp as we have applied at the input because the
whole bulk silicon is not going to be applied at the input. So, this heat pulses will act like
a electro thermal filter. This mass of this silicon will act like a filter and it will damp
down all the high frequency component. Only the low frequency component will pass
through. These heat pulses are now received by this sensor and generated these electrical
pulses. So, now we have two signals. One is the input signal which is very accurate
digital signal which we have applied at the heater and the second is the output.

This is the electrical signal generated by temperature sensor. Now, these heat pulses
which are being generated by heater and going all the way to temperature sensor they will
take certain amount of time. So, this is the output and the output time. These heat pulses
from heater to sensor they will take certain amount of time because there is a distance of
100 micrometer in between these two objects and the time taken will be proportional to
the speed of these heat pulses. So, this time taken will be equal to 100 micrometer which
is the distance between these two objects divided by the effective velocity of these heat
pulses.

This velocity of heat pulses is nothing but the thermal diffusivity. How fast the heat can
diffuse in the silicon from one point to other. So, this velocity is proportional to the
thermal diffusivity which in turn proportional to my absolute temperature. It means if my
bulk silicon this whole slab if this whole slab is at higher temperature my thermal
diffusivity will be lower means the heat pulses will take more time in going from heater
to sensor if the temperature is high. If the temperature is low then the thermal diffusivity
will be high as per this relation and it will take less time going from heater to the sensor
because the distance is fixed which is 100 micrometer here in this case.

99
So, because of this velocity or thermal diffusivity dependence on temperature now my
this formula the time taken is actually temperature dependent. My temperature of bulk
silicon affects my time taken. So, now what we can do we take both of these input and
output let us say this is my input which is my pulse which I apply at the heater and this is
my output which I generate with using this temperature sensor. This generated output will
be slightly delayed because pulses will take time in travelling from heater to sensor. So,
my output will look something like this and there is a phase difference between input and
output.

If we see closely the rising age of input is at this point and the rising age of output at this
point and this is the phase difference. This difference is between input and output. This
phase difference is actually now proportional to the absolute temperature of silicon. So,
input we already know because we are applying this input through some digital
electronics and output is generated by this sensor and this is a electrical output which
goes to some electronic circuitry. There we can compare both of these input and output
and figure out how much is the this how much is the phase difference between these two
waveforms.

This phase difference is proportional to the thermal diffusivity or in turn the temperature
of the bulk silicon. So, by this measurement of phase difference we can find out how
much is the temperature of this bulk silicon. So, this is how this acts like a temperature
sensor which is fully silicon based sensor because we are using all the silicon material
only and this phase difference is proportional to my thermal diffusivity and in turn the
temperature. The output of this we can write which is phi is the phase difference is equal
to minus R where R is the distance between heater and the temperature sensor under root
pi the frequency of the signal which we apply divide by D which is the thermal
diffusivity of silicon. So, this is how the phase actually is dependent on the thermal
diffusivity and in turn the temperature of the whole silicon.

So, this is the silicon based temperature sensor and this is the cross sectional view of
silicon based sensor. So, this is the cross section view where we take the silicon and we
take a cross section of this and we see the device from on the surface through this cross
section. Now, let us discuss some more properties of this thermal sensor which the first is
this thermal diffusivity is very weak function of process variation. As we discussed in
first slide the BJT actually suffers from process variation if there is a slight process
change or there is a fluctuation in the fabrication the performance of BJT or the current of
the BJT increases very drastically. But here the thermal diffusivity this property of silicon
it does not depend very much on the doping.

We take this thermal diffusivity of pure silicon and if there is a slight amount of doping is
there present in the thermal in the silicon then the thermal diffusivity is not very much
related to that process variation or that the small amount of doping. So, we can say this

100
thermal diffusivity is weakly sensitive to process doping. So, let us see. So, this thermal
diffusivity is very weakly sensitive to doping fluctuation what the doping generally used
in CMOS process which is typically 10 to power 14 to 10 to power 20 or 21. In this
amount of doping the thermal diffusivity is very much very least sensitive to this kind of
doping fluctuation.

It means from wafer to wafer or lot to lot process variation will not impact the
performance of the thermal sensor if we make thermal diffusivity waste temperature
sensor. The other point is these ETFs they do not require trimming these ETF or electro
thermal filters they do not require trimming. The thermal diffusivity D is very well
defined very well defined for IC grade silicon. When we take pure silicon which in case
we use for IC fabrication the silicon wafer this D the thermal diffusivity is very well
defined and very well experimented. So, it does not change with wafer to wafer because
the IC grade silicon is very much pure.

So, this ETF or this structure the thermal sensor it does not need individual trimming
which is very cost effective. So, now let us see the top view of this structure how it look
like. So, we have a heater there for example we make a heater something like this. So,
this is a heater with two terminals where we apply electrical input. On the periphery of
this heater at a certain distance let us say 100 degrees.

We have the temperature sensor which we are going to place now around this. So, these
are the thermopiles or thermocouples. Which we are placing around this heater and if you
see this heater is kind of equidistance to all the thermopiles. All this distance from this
heater to these thermopiles these are constant which is like 100 micrometer which we
have discussed last time. So, this these are thermopiles or the thermocouples where we
need two metals as we discussed. In thermocouple we need two different materials to
make this thermocouple. And this thermocouple is connected in series with other and this
is connected in series with the other thermocouple. So, this is how these thermopiles are
connected. So, we have here one thermocouple this is one thermocouple where we have
two different materials. So, these thermopiles or thermocouple is made of n plus diffusion
layer. This heater is made of n plus diffusion and this thermopiles are made of p plus
diffusion and this connection is made up of aluminum.

So, we can see this p plus diffusion and aluminum they form a thermocouple which is
going to sense the temperature change. This is my sensing edge which is close to the
heater. Now if we see closely we have placed all these thermocouples equidistance from
the heater and we are joining all of them in series because the first end of goes to the
second end and second thermocouple and the first end of the second thermocouple goes
to the second end of the third one and so on. So, all these thermocouples are in series all
thermocouples are in series to increase the output voltage and they are placed

101
equidistance. So, that when the heat is generated by this heater this is going to diffuse in
all the direction.

So, we need to make sure when these heat pulses are reaching at the thermocouple it
reaches at the same time to all the thermocouple. We have 8 thermocouple here. So, this
heat pulse which is generated at the center this need to reach at the same time to all the
thermocouples. So, we have placed this sensing edge such that this has same distance
from this heater which is fixed like 100 micrometer and now this aluminum and P plus
diffusion they are different materials.

So, they will act like a thermocouple. In thermocouple we need two different materials.
So, here we have P plus diffusion and aluminum the pair of this P plus and aluminum
makes a thermocouple and now we use 8 thermocouples connect them in series to get
more output voltage. So, we apply this input here at the heater and we get the output in
terms of electrical pulses generated by this thermocouples. These thermocouples hot
junctions are located roughly circular contour or the equidistance from the heater. We can
write it down the hot junctions of thermocouples is roughly circulated equidistance from
heater.

So, when a heat pulse is generated from this heater and reaches this thermocouple this
edge. So, this heat pulse reaches at the same time to all these thermocouples and these are
placed roughly in circular fashion. So, that the distance is same. Now, when this heat
pulse reaches this edge let us say this is the heat pulse which is reaching this edge this
thermocouple this sensing end becomes hot junction and the other end which is far away
we assume it is far away compared to this hot junction this act like a reference junction
and this reference junction is far away from this heater which we can assume let us say at
some fixed temperature T. So, this reference node and this hot junction the difference the
temperature difference between this hot junction and the reference junction this gives rise
to the thermocouple voltage which is added up in series by all these thermocouples and
generates the output at the top end.

So, in this way this thermal sensor works and it senses the temperature of the silicon slab
and gives the output as electrical signal. So, now let us discuss what is the output of this
structure this thermal sensor with respect to the temperature measurement or when we
increase the temperature how much is the response of this thermal sensor. So, at constant
frequency this output of ETF is near linear function or we can say T to power 0.9 if we
remember from last slide the diffusivity is proportional to T to the power 1.8 inversely
proportional and because these comes in the under root. So, the at the effective phase
output pi is actually is almost a near linear it is not perfectly linear it is not T to power 1.
In fact, if this comes out at T to power 0.9, but it is very much kind of very close to
linear. So, phase output of this is near linear with temperature. It means this structure can
be used very well as a temperature sensor because the output of this sensor is very close

102
to linear. So, this is the output of this sensor x axis is the temperature which we are
changing and on the y axis we have the phase difference and the output of this structure is
something like this. So, this is not perfectly linear because the dependence is only T to
power 0.9 it is not T to power 1 and this is the output of this structure which is silicon
based sensor. The phase difference is actually a near linear function of temperature is
almost in nearly almost a linear function linear can be something like this, but it is more
or less linear.

So, this is the case when we use bulk silicon. This is the behavior of phase difference
with respect to temperature when we use a silicon wafer which is a bulk silicon wafer
then this is the this is the behavior if we use another kind of wafer for example, we have
SOI wafers or silicon on insulator wafer then the behavior of the structure will be
something like this. So, this is the case for SOI wafer. So, we can see that the
performance of this temperature sensor is almost similar for SOI wafer as well as bulk
silicon, but in case of SOI wafers the amount of phase difference is little higher that is
because the SOI wafers we have a another oxide layer which is below the silicon oxide.
So, when we are generating heat pulses the heat does not diffuse through the oxide layer
and we have on the top we have oxide layer and the bottom we have oxide layer. So, this
SOI device is the silicon which is trapped between all these between these two oxide
layers the heating of these SOI devices are much higher compared to the bulk devices
that is why we see a difference in the phase difference here for SOI wafers as well as the
bulk silicon.

So, SOI wafers we get more phase difference compared to the bulk silicon. These SOI
devices are also comes in two subsets one is the partially depleted SOI or we call it
PDSOI and the other one is FDSOI we call it fully depleted SOI. So, we have two terms
here for SOI wafers we have PDSOI we call it partially depleted SOI and we have fully
depleted SOI. So, we have two terms here for SOI wafers we have fully depleted SOI and
we have another FDSOI is called fully depleted SOI. In case of PDSOI which is partially
depleted SOI this amount of silicon is more on top of oxide.

So, the behavior for PDSOI is very much similar to the bulk silicon. So, the bulk silicon
is PDSOI is very much similar to the bulk silicon, but in case of FDSOI the fully depleted
SOI the silicon layer thickness is very small and the heat cannot be effectively removed
from the top or the bottom because of the oxide layers. So, fully depleted SOI gives you
more phase difference compared to the bulk silicon and this is the difference between
PDSOI and FDSOI. So, now let us discuss about the doping sensitivity of this structure.
So, we have two terms here so these ETF or this electro thermal filters these are electrical
devices where we are measuring the electrical performance of these thermocouples. They
are not very much sensitive to the process fluctuations for example, doping fluctuation or
little bit of geometrical fluctuations they are not very much sensitive to those kind of
fabrication based changes.

103
So, these ETFs are much less sensitive to the process fluctuations. So, this ETFs are
much less sensitive to spread in doping concentration. So, doping concentration if there is
a change it does not impact much the performance of ETF though if the doping levels are
very high then it can contribute to the performance change in ETF because in that case
the performance change in ETF is very high. So, doping concentration if there is a change
it does not impact much the performance of ETF though if the doping levels are very high
then it can contribute to the performance change in ETF because in that case if the doping
is really high then the second order effects will also come into picture. For example, the
thermal diffusivity of silicon which we have assumed 1 upon T to power 1.8 that is in
case of pure silicon, but if the doping levels of silicon is very high then we can no longer
assume that silicon is in pure state.

Then this thermal diffusivity factor changes and because of that the performance of ETF
will actually change. So, the high levels of doping can contribute to the performance
change in ETF can contribute to the performance change in ETF. So, this is the high level
of doping. We can consider case where the experiments are performed using N plus plus
implant which is very high implant and normally used to reduce the resistivity of silicon
well.

So, there is experiment which was conducted. In this experiment we use N plus plus
implant plus means it is very high doping. This is normally used for the high level of
doping and normally used to reduce the resistivity of silicon. So, in this experiment N
plus implant was used which is very high doping. It was doped into the silicon and we
measured the output of this structure and the result is this highly doped structures they
give they have significant batch to batch variation when we measure the performance of
the silicon.

So, this is the high level of doping. So, this is the high level of doping. So, this is the high
level of doping. So, this have significant batch to batch variation when we compare the
performance of these different batches when we use very high doped implants then there
is a significant error in terms of the output. So, these highly doped ETFs was found to
have significant batch to batch variation. So, for example, we have the same structure as
we discussed earlier. We have this silicon and this is SiO2 substrate. So, we have this is
SOI wafer. So, we have the same structure as we discussed earlier. So, we have this
silicon and this is SiO2 substrate. So, we have this is SOI wafer. So, we have oxide layer
below the silicon. We have silicon here and then we have again silicon oxide and we
have this heater and this temperature sensor which goes out.

This is again oxide layer, SiO2 and now we do N plus implant here. This is a heavy
implant we put in this device. Now, we cannot assume that this silicon is in pure state and
the d is not now proportional to T to power 1.8 because this formula is for the pure
silicon and now we have this silicon which is very much or very heavily doped with N

104
plus. So, if we use very high doping then we have batch to batch variation and if we see
the output of these ETFs with respect to temperature this is on the x axis we have
temperature here on the y axis we have phase which is pi. If we plot this is let us say our
normal SOI ETF without then if we plot this on the same graph if you plot the
performance of SOI ETF with N plus plus doping it will look something like this.

There is a significant change with respect to the doping this is the performance of SOI
ETF with N plus implant. So, this is the performance change of this ETF if we have a
very high doping placed in the silicon device with though which is very unlikely because
when we make this silicon devices we allocate a separate space for this wafers and we do
not put this kind of heavy dopings when we are fabricating this structures. So, this is
unlikely, but for to understand what is the effect of doping this is beneficial to analyze
these cases when we have doping. So, in a reasonable amount of doping which is 10 to
power 14 to 10 to power 20 this does not impact very much the performance of these
ETFs and this kind of sensor which is completely silicon based this silicon based sensor
can be used as a temperature sensor this is for fully integrable with silicon devices all the
process steps are compatible with CMOS fabrication and this is the performance of these
ETFs which are fairly linear having a linear performance linear output with respect to the
input measurement is very advantages and very beneficial for all the sensors and here we
see this silicon based sensor is the output actually is very much linear with the input
measurement which is temperature. So, today we discussed the thermal sensors which is
completely silicon based or silicon integrable devices.

So, this is all for today.

Thank you.

105
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 11
Thermal Sensors: Nano-Materials Based

Hello, ah welcome to the course Transducers for Instrumentation. Last lecture we

discussed some thermal sensors which are silicon compatible means they follow the
fabrication process of silicon and we can integrate them with other semiconductor
devices such as MOSFETs and BJTs. So, ah from scaling point of view those sensors are
beneficial we can scale them as per our requirement and we can batch process them
which decrease the cost. So, the last lecture we discussed those silicon based sensors.
Today we are going to talk about some thermal sensors which are not silicon compatible
they are like emerging technologies. For example we will first talk about the carbon
nanotube based sensors we can make a thermal sensors using carbon nanotube and let us
see how we can make a thermal sensor using these CNTs.

So, for carbon nanotubes we know that the temperature coefficient of CNT that
temperature dependence resistivity of carbon nanotubes that is a proportional to the
temperature increase. So, we can use this property of CNT to make thermal sensors. So
let us talk about carbon nanotube based thermal sensors. CNT is the abbreviation for
carbon nanotube.

So now we know by experiments that carbon nanotubes or CNTs show a temperature

dependence and therefore, can be used as a thermal sensor. So by the experiment we
come to know that temperature coefficient of resistance the temperature coefficient of
resistance or we call TCR in short this is approximately negative 0.15 percentage per k.
This is the measured value of temperature coefficient of resistance which is TCR for
CNTs this is minus 0.15 percent per degree.

So, it means that the resistance of carbon nanotubes it changes 0.15 percent per degree
change in the temperature and we can see that this is negative. So here we see in the
number is negative it means the resistance decreases with increase in the temperature. So
we can say this is a NTC or the negative temperature coefficient NTC which stands for
negative temperature coefficient is very similar to the thermistors we have seen earlier.
They also have the negative temperature coefficient for a certain range, and they are very
beneficial in that particular NTC range.

So, the CNTs they also have a negative temperature coefficient which is minus 0.15
percent per degree and this number is very competitive with other materials just like
some other thermocouples or some other sensors. So we can use the CNT as a

106
temperature sensor. This number is very competitive with other materials. The next
which is very important thing about these CNTs is these are all nanomaterials and when
we make a sensor using nanomaterials these sensors tends to be very sensitive for any,
any sensing because for when we go for the nanostructures the surface to volume ratio
actually increases when we decrease the size of nanoparticles.

So here in the case of CNTs these are also very fine nanoparticles. So the surface to
volume ratio actually increases when we go for finer and finer particles. So this is another
important point about these nanomaterials. These nanomaterials could lead to new type of
extreme miniaturized sensors. With very high sensitivity.

Of course, ah some other benefits are also there. For example, the low power
consumption and fast response time. So as we discussed earlier as well if the size of the
sensor is very small the response time actually improves. We have a very fast response
time of sensor if the size of sensor is small because this needs very less amount of energy
transfer from hot-to-hot body to this sensor to come to the same temperature of the body.
So when we go for nanomaterial based sensors the fast response time we get as a benefit.

Other benefit is low power consumption because the sensor actually takes less amount of
power because the size itself is very small and the more important thing is the high
sensitivity. We have here a very high sensitivity in case of nanomaterials because in case
of nanomaterials the surface to volume ratio actually improves. So we can consider these
nanoparticle as a kind of sphere. So we can write the surface the formula for the surface
and formula for the volume and we can take the division and we come up with this
relation. So we have surface to volume ratio for a sphere is surface is 4 pi R square and
the volume is 4 by 3 pi R cube which comes out 3 upon R.

So, the surface to volume ratio for these nanoparticle comes out to be 3 upon R and if we
start reducing the size of our nanoparticles for example if we apply the limit here where
the limit R tends to 0 we see this term almost approaches to infinity. So, this surface to
volume ratio actually improves very much if we have finer and finer particles. So, when
we have more area available for all these nanoparticles to react with the surrounding it
means the sensitivity of these materials goes very high when we decrease their sizes. So,
we can see here that surface area surface to volume ratio actually effective surface area
actually improves, and we have more and more sensitivity if we go for smaller and
smaller radius on the size of the sphere. So, this is one benefit we get when we go for
nanomaterials such as CMTs.

The main bottleneck about the development of these nanomaterials is the integration of
these materials with the other micro fabrication for example silicon based fabrication
which is actually dominates all the fabrication process and these new nanomaterials they
are not very much compatible with the silicon process. So, we cannot batch process them

107
along with the silicon devices. So, this is the limitation of these CNT based or some other
fancy nanomaterial-based sensors. So, the limitation we can write of these nanomaterials
sensors. Integration of nanostructures with microfabrication.

So, these newer and newer materials we cannot batch process them with the silicon
devices or in the silicon fabs because these materials when we use them along with
silicon they act as a impurity for the silicon devices and the performance of those silicon
devices goes bad. So, we cannot have the same fabrication process where we can use
silicon as well as these CNT or some other nanomaterials. So, we need to fabricate them
separately in a very different fabrication line and then we connect them with electronics
extra electronics through discrete components. So, this is what we do to make a complete
sensor using nanomaterials. So, let's see the fabrication process of these CNT based
thermal sensors.

So in these CNT based sensors we are using the property of CNT that the resistance of a
CNT changes with the temperature. So effectively we want to measure the resistance of
CNT and to measure the resistance we need two points, two input and output port where
we can apply electrical signal a known voltage and we will measure the current ratio V
and I gives us the resistance. So we need two points along a CNT or carbon nanotube
where we can apply a known electrical voltage. To do that what we do we start with a
with a substrate we have let's say this substrate first we do the pattern a certain catalyst
here for example this red is the catalyst. This is the first step where we pattern catalyst.

In the next step we grow CNT so we have this substrate where we have put these catalyst.
Now the CNT will grow only on where the we have put the catalyst. So for example CNT
will grow something like this on the CNT on these pattern catalyst. This step is grow
CNT. The third step we actually join them we put a glue in this in this structure for
something called condensed liquid.

This will look something like this we have wafer with pattern catalyst on top of this we
have grown these CNTs and now we pour some sort of a liquid on this on this structure.
So when we pour the liquid on this structure this whole assembly will come to that come
in contact with this that liquid that liquid will fill up here in this structure. So this liquid is
some sort of a glue when we put this this liquid on this structure this whole of these
whole of these CNTs will kind of stick to each other all other CNTs and then we can heat
it up or we can evaporate this liquid which will come out from here and the CNT which
are stick together they will remain in contact with each other. So after this third step
when we pour this condensed liquid or glue on this assembly the fourth step is we
evaporate this liquid. So the structure will become like this.

The CNTs will be joined to each other and now we can see that there is a electrical
contact made here at this point where these are all joined. So we have connected two

108
CNTs more than two CNTs with each other and now this makes a electrical contact. So in
this way we can join two CNTs together because the problem with single CNTs we on
the bottom we have a contact with CNT but on the top when we are growing these CNTs
there is no way to form a second connection on the CNT and for resistance measurement
we will need at least two terminals to apply a potential difference. So when the CNT is
growing after the growth there is no way of putting a one more contact on top of the
CNT. So for this problem we are growing two or more than two CNTs together and join
them on the top so that they make electrical contact and now two CNTs which are
connected on the top these two CNTs has two contact pad at the bottom where we have
put the catalyst.

For example if we can number them together let's say we have one, two, three, four, five.
So similarly we have one, two, three, four, five. Now we have these pad numbers pad
number one and two so between pad number one and pad number two we can apply a
potential electric potential and the current will flow between these two CNTs this CNT
and that CNT and it will give us a resistance drop across this CNT and we know the
thermal resistance of CNT is minus 0.15 accordingly we measure the temperature of
temperature through this assembly. So this makes a kind of CNT based temperature
sensor.

So let's discuss the fabrication process of CNT in more detail. So the first step is we start
with a silicon wafer and we pattern and we put a 1 nanometer thick film of iron or nickel
catalyst. So this layer of iron or nickel catalyst is patterned on this silicon wafer using
photolithography. So we have this silicon wafer on top of this we put a nickel or iron
layer which is very thin layer and then in the next step we pattern it to make pads. In this
step now we do photolithography and make the pattern this is the pattern we make.

In the second step now in the second step we grow the CNT which is vertically aligned
CNTs and we grow those CNTs on these patterned structures which is iron and nickel
which act as a catalyst. So when we start growing these CNTs these CNT will start
growing on these pads only because these act as a catalyst. So as a second step we grow
CNTs. CNTs by thermal CFT CVD or we can say chemical vapor deposition. So in the
next step what we have this is our silicon wafer where we have patterned our pads and
now CNT will grow on these pads.

This is one CNT wire and this is another CNT. We can name it this is silicon this is CNT
and this is catalyst. In the next step we put the solvent. It is such as acetone. So in this
step we take this condensed liquid which is which can be acetone or some other glue.
This acetone is poured on this structure on this forest of CNTs when we pour this acetone
all the CNTs will stick to each other and due to the capillary rise thus the solvent is
actually drawn into this forest of CNT independently. So if we draw this step now we
have this wafer. These are our patterned catalyst and these CNTs are now grown. Now

109
when we pour the liquid here this acetone is poured here. These two wires of CNTs will
come into contact of each other these wires and in the next step when we evaporate this
CNT this acetone this liquid will evaporate and the structure this CNT which are now
gelled together and made the physical contact now they will remain in that contact after
that after this evaporation of acetone.

Now during infiltration and evaporation the CNTs with each within each structure
densifies and each structure which is shaped individually by the forces resulting from the
capillary action. The final structure when we done this evaporation of this acetone the
final structure will look something like this. We have this silicon wafer on top we have
this pattern catalyst and these CNTs will kind of connect to each other on the top. These
are basically two CNTs which are connected on the top and on the bottom we have two
separate pads let's name this pad 1 and pad 2 and we can apply the potential difference
across these pads 1 and 2 so that we can measure the resistance of this whole assembly.
So when we apply a potential difference across 1 and 2 the current will flow from pad 1
to pad 2 and accordingly we can measure the resistance of this whole assembly R equal to
V upon I where we apply a known voltage and measure the current and the resistance is
calculated.

So this is how we can make a CNT based thermal sensor where we are measuring the
resistance of these CNTs by applying a electrical signal. Now this is the fabrication
process what we follow in the lab to make this kind of CNT based thermal sensor. The
actual for to improve the yield of our process we actually don't do only with two pads we
actually do multiple pads at a time because this is a natural process when we put the
acetone and evaporate it, it's not guaranteed that two CNTs from pad 1 and pad 2 they
will come in contact with each other. So it's not 100% guaranteed that it will happen so
we increase the chances by putting more number of pads and whichever gives us the
resistance the close loop formation between one pad and another pad we can take that
assembly that pad 1 and 2 into account. So in actual process we put multiple pads for
example this is one pad, this is second pad, this is third, this is fourth, fifth and sixth.

So this is the top view of wafer where we are putting these all these six pads let's name it
as 1, 2, 3, 4, 5, 6 and we pattern the catalyst on this, this is the catalyst pattern and on top
of this catalyst we grow the CNTs, these CNTs there will be multiple CNTs growth on a
single pattern. So there will be so many CNTs coming up here and here it will look like a
forest of CNT and when we put acetone here and evaporate that acetone these all CNTs
will stick to each other and will give rise to the electrical connection between them. So
actual structure of this CNT look something like that these are all mixed up or juggled up
with each other. So now we see let's say between pad 1 and pad 2 we can apply a
potential difference and the current will flow from this pad to this CNT and then we will
come back here, this is how the current will flow and we can measure the resistance of
this whole CNT structure is R equal to V upon I. And now we know that this resistance

110
of CNTs this is this has a temperature coefficient or TCR which is minus 0.15 it means
the resistance of this whole assembly will decrease with the temperature change. If the
temperature goes high the resistance of this structure will decrease by 0.15% with every
increase in the degree increase in the temperature. So now if we plot the measured value
of this resistance for this structure it will look something like this. On the x axis we plot
temperature, let's say this is 0, this is 20, 40, 60.

And on the y axis we have the resistance which is in ohm and the measured data let's say
the axis is something like 1100, 1200, 1300, 1400 and so on. And the measured value of
this resistance will have a negative temperature coefficient so the resistance will decrease
with increase in temperature. So the measured value will look like something like this. If
we measure across multiple paths the data the measured data will look something like
this. So now here we have the negative temperature coefficient, the resistance is actually
decreasing and we have multiple values because this we can use any number of paths
with any for example the resistance between pad 1 and pad 6 will be different resistance
between pad 1 and pad 5 will be different between pad 1 and pad 3 will be the different
resistance because this is all not a very controlled kind of connection of CNTs this is all
natural CNTs maybe between pad 1 and pad 3 there are 100 CNTs which are joined
together but between pad 1 and pad 4 maybe 1000 CNTs have are joined together so the
resistance effective resistance of between pad 1 and pad 4 is less compared to earlier.

So because of that there will be a difference when we measure between different paths so
that's why we see multiple lines here. The resistance actually goes down with temperature
with every degree centigrade. One more point to notice here is that the response is very
linear. We can see that resistance change is with temperature is very much linear means
we can very easily make a sensor using these CNTs response will be linear. Linear is
always a good output if we have coming from a sensor we always desire to have a linear
output with respect to the measurement or the input quantity.

So here we have a quite linear output with respect to the temperature change. So the only
two electrodes are enough only two electrodes are enough for measurement the above has
six electrodes just to increase the yield. So this is how we can make a thermal sensor
using carbon nanotubes where we have seen how to join these CNTs together we saw that
when we grow the CNTs very difficult to make a second connection on top of CNT. So
what we do we grow two CNTs together and we join them on the top using this
condensed liquid or acetone we put the acetone so that they kind of mix up or juggle up
with each other then we evaporate this acetone so that they will remain in contact and
when we make a contact between two CNTs on the bottom they have different pads. So
we can measure the resistance between these two pads electrically and we see the
response the resistance is changing with temperature.

111
So this is how we can make a temperature sensor using CNTs. Now we make another
temperature sensor using different material which is something called graphene.
Graphene is a very popular material from at least last 10-15 years and people are making
many kind of sensors using graphene. For graphene as well we know that the resistance
of the graphene or the graphene flakes is actually changes with temperature and that also
has a negative temperature coefficient. It means the resistance of graphene actually
decreases when we increase the temperature.

So this we now we see how can we make a temperature sensor using graphene. So to start
the fabrication of graphene based thermal sensor we again start with the substrate which
is kind of a polyamide is used as a substrate for graphene because this has a strong
adhesion with the other metals and this the thermal capacity of the polyamide which the
thermal the temperature which this can handle is the same as the thermal capacity of the
polyamide which the thermal the temperature which this can handle is high compared to
other substrate. So for graphene based thermal sensor fabrication we start with the
polyamide as a substrate. This polyamide or in short we call it PI or PL whatever we call
it. This polyamide is been increasingly used as substrate due to these two reasons.

The first is this has strong adhesion to metal coatings which provides a high degree of
strain delocalization. So because of this strong adhesion to metal coatings we have a very
high degree of strain delocalization. There is no strain actually developed on the substrate
because of this coating the other coatings. The second point why we use polyamide for
graphene-based fabrication is the wide temperature range. This can sustain a temperature
of typically minus 270 degrees to 400 degrees.

It is a vast temperature range which polyamide can handle and this is ideal for many
deposition techniques. Such as such as sputtering and E beam evaporation. So, we have
these two benefits of polyamide why we use this polyamide for graphene based thermal
sensor or where we want to fabricate a graphene based structure we start with polyamide.
Polyamide is flexible as well so we can use the polyamide substrate for wearable
electronics where this the substrate need to be flexible which can bend along with the
body movements. So polyamide is good in that sense as well we can use it for wearable
electronics.

So now why do we want to use graphene? So we have certain properties of graphene

which are very good and attractive to make a sensor. So let's discuss some properties of
graphene. The first one is the excellent mobility of charge carriers. So the graphene has
very good mobility of charge carriers. The second is the high electrical conductivity
which is because of the high mobility of course.

So we have a high electrical conductivity. The third property of graphene which is very
interesting is the high optical transmittance. And the fourth property of graphene is high

112
mechanical stability. So these are the few properties of graphene where we have a very
high electron mobility of in graphene and because of this high mobility we have very
high electrical conductivity of graphene. We have very high optical transmittance that the
light can actually pass through the material. And the fourth is the high mechanical
stability that the material is actually very much stable against the mechanical stress.

So these are the properties which make graphene a good candidate for making some kind
of sensors. Now we are looking at making the thermal sensor. So as we discussed the
graphene has a temperature coefficient or the TCR thermal coefficient of resistance
which is proportional to temperature and which actually has a NTC or the negative
temperature coefficient. Let's see how it look like for the graphene. Let's plot the
resistance versus temperature for graphene. On the x axis we have temperature and on the
y axis we have the resistance in ohm. That temperature is in degree C and the resistance,
this is zero. The resistance will look something like this. This has a NTC or the negative
temperature coefficient.

This is for graphene or graphene flakes. Sometime instead of using graphene flakes we
use the other material which is made up of graphene only which is called strontium
graphene oxide. We call it SRGO or the strontium graphene oxide. And if we plot the
resistance versus temperature for strontium graphene oxide this also has a negative
temperature coefficient. It looks something like this. We have the NTC or strontium
graphene oxide, SRGO where we have this NTC is more negative compared to the
graphene flakes.

So these are the two response of the resistance of graphene flakes and strontium graphene
oxide with respect to temperature and both has negative temperature coefficient. The
resistance actually decreases with each degree increase in the temperature. So now we
can make a thermal sensor using them. So the fabrication of fabrication process of this
graphene based thermal sensor is. So first we start with a silicon substrate and on top of
the silicon substrate we put a polyimide layer.

So let's say polyimide is let's take a yellow color for polyimide. So this is polyimide layer
and this is on top of silicon substrate which is on the bottom of the silicon substrate. We
have this silicon on top of this we have put a polyimide layer. This is polyimide and this
is silicon. Now the next step is we define the electrodes on this structure using
photolithography. We define electrodes using photolithography. And the structure will
look like. So we have silicon on the bottom and on top we pattern these pads or
electrodes. So these are the two electrodes we pattern the rest is the photoresist. This is
the pad we have patterned. And rest is the photoresist. This is the electrodes. This is the
positive photoresist or PPR we can call it. Now the next step is to sputter copper on top of
this so that these electrodes are conductive. On the bottom we still have silicon. Now
these pads will look something different because now we put copper on top of it. So let's

113
take the metal as blue. Now these electrodes we have sputtered the copper to make them
electrically conductive. This is first electrode and this is second electrode. This is the step
where we sputter copper. This is followed by lift off process. So now we have electrodes
which are patterned on this polyimide substrate. And now we have put copper on top of
these electrodes. Now these electrodes are conductive. We can use these electrodes as
electrode 1 and electrode 2 and we can apply a potential difference across it. But still
these electrodes are far apart. There is no interconnection between these electrodes. So
now the next step is we take the solution where this solution contains this SRGO or the
strontium graphene oxide which is dissolved in this. We take this solution and pour on
these electrodes where these are in close proximity. Here these two electrodes are very
much close. So we pour this liquid on top of this and evaporate this liquid.

So when we evaporate this liquid, the flakes of these SRGO or the graphene flakes, these
flakes will stick there and make a conductive path between these two electrodes and then
we can apply the potential difference. So the next step what we have is drope casting of
SRGO or strontium graphene oxide. We take this liquid in a dropper and we pour this on
this area. In this area we pour this liquid which is the solution of this strontium,
germanium oxide and then we evaporate this liquid. The flakes of SRGO will remain
there and make an electrical contact between these two pads. So our structure will look
something like this. This is first electrode and this is second electrode. And when we
have dropped this, drop casting of this SRGO and evaporated the liquid, there will be a
fine flakes of strontium graphene oxide that will stay there and make electrical contact
between pad 1 and pad 2 and this black dot, this is our graphene. So now we have the
complete structure on polyamide. These two pads are connected through the graphene
and this graphene has a negative temperature coefficient which the resistance will
decrease with the temperature and we can measure the temperature or sense the
temperature by measuring the electrical resistance, how much the electrical resistance
changes. Based on that we can measure how much is the temperature is changed outside.
So this is the complete structure though this is still on the silicon wafer and silicon wafer
is not flexible. So we can now remove the back silicon wafer which act just like a handle
wafer just for our convenience or to give the mechanical strength to polyamide. So now
we can remove this silicon wafer from below of this polymide sheet and this polymide
sheet will come out. Now, we will remove all of polyimide tape from the silicon
substrate.

And then finally we have a polymide sheet only with this electrical structure. We have
this polymide sheet. These are metal pads which is made up of copper. Let's say this is
pad one and pad two. This is pad one and pad two and in between this pad one and pad
two we have a drop of graphene which is there acting as a temperature sensor. Now we
apply an electrical voltage between this pad one and pad two and we can measure the
resistance and this resistance is changing with the temperature.

114
So this is the complete assembly of graphene-based temperature sensor and now this
temperature sensor is on the polymide sheet which is very flexible in nature. So we can
use this polymide sheet or this thermal sensor as a variable electronics because This
substrate this polymide can bend with the physical movement of the body. So unlike
silicon, which is kind of has a mechanical strength, and it cannot be bent, This polymide
sheet is very much flexible and can be used as a variable electronics sensor as a thermal
sensor for variable applications. So these are some thermal sensors which are based on
nanomaterials. They are not compatible with the silicon process but as a new
nanomaterials, we can make some temperature sensors using these nanomaterials itself.

So, this is all for today.

Thank you.

115
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 12
Thermal Sensors: MEMS Based

Hello, welcome to the course transducers for instrumentation. Last lectures we discussed
some thermal sensors which are made up of silicon and last lecture we discussed a thermal
sensor which is based on the nano materials such as graphene. Today in today's lecture we
will discuss some thermal sensors using MEMS device. The MEMS is the micro electro
mechanical systems. These devices, these are made up of materials which for example
some metals or metal oxides and the property of these materials is such that some properties
like mechanical properties of these materials change with the temperature. So we want to
use these properties and make a thermal sensors which gives us the temperature in terms
of the mechanical property change.

So many a properties for example the stiffness or the Young's modulus these are all
mechanical properties which are greatly influenced by the temperature. So we want to use
such materials which shows a corresponding change in the mechanical properties with
temperature and such a material which is widely used is vanadium oxide which we
generally use for micro fabrication and making MEMS devices and today we will see such
a sensor which we can use for temperature sense. So we discussed today some thermal
sensors which are MEMS based. So the full form of MEMS is micro electro mechanical
systems. So we use a material which is called let us say vanadium oxide. Here for this
material mechanical properties are temperature dependent. So we use vanadium oxide for
sensor fabrication because along with the mechanical properties we need to take care of
certain other parameters so that the material is compatible with the fabrication process. For
example for almost all the materials the mechanical properties are very much temperature
dependent but those materials are not very much compatible with the fabrication process
we follow to make these MEMS devices. These MEMS devices are very small in nature.
These are for example micro word is there micro electro mechanical system. These are
mechanical systems which we excite using some electrical stimuli. So we fabricate these
devices along with our normal CMOS kind of process. So the material which we choose to
fabricate these MEMS thermal sensor this material should be compatible with the
fabrication process we follow. So that's why we use vanadium oxide which is very much
compatible.

Now we take this material and we want to make a resonator because the mechanical
properties are very much used to make a resonator and the temperature change of a
mechanical resonator this can be determined from the change in the resonant frequency.
So we have here something called change in resonant frequency. So we have a MEMS

116
device where this device is resonating at a particular frequency and if the temperature is
changing then there is a corresponding increase or decrease in the quality factor of this and
because of that we get a change in the resonant frequency. So this change in the resonant
frequency with temperature we can detect and this gives you the corresponding output in
terms of temperature. So we follow now the fabrication process of such a MEMS device.
So for that we start with this silicon wafer. So we have a silicon wafer here. This is silicon
wafer and the thickness is typically 200 to 300 micron. This is step one. Now we take this
silicon wafer and we grow the oxide both side of this using a dry oxidation or wet oxidation
whatever preferable. And second step we this is step B where we put the oxide both the
side. Let's say this is our oxide and on top of this silicon oxide we put a thin layer of gold.
This is layer of gold which is typically 100 nanometer and on top of this gold layer we put
this vanadium oxide. This is our vanadium oxide or VOX we call it in short and the
thickness is typically 400 to 500 nanometer. So this is the second step where we take this
silicon and we put the oxide both the sides.

We just put a uniform layer of oxide both the side. On top of it we put a layer of gold very
thin layer of gold to make a electrical contact and on top of this gold layer we put the
vanadium oxide which is our material where the mechanical properties are dependent on
temperature. So we take this as a second step now we in the next step we etch this pattern
on top. So we have our silicon the oxide on top we have gold and we have vanadium oxide
which we pattern as per our dimensional requirement. So this is our step C where we pattern
VOC pattern vanadium oxide. In the next step we take this silicon and from the back side
actually we remove oxide and micro machine it. We kind of etch the silicon from the
bottom so our oxide looks something like this. This is our oxide. So in this step D we
pattern the silicon. So what we do from the back side of the wafer we start etching the
silicon. So first there is a oxide layer we etch that silicon oxide layer and start etching the
wafer from the back side. So it is called the photolithography from the from the back side
of the silicon wafer and we etch out all the silicon so that we reach oxide till oxide on the
top. So one requirement of this kind of fabrication process is the silicon wafer which we
start with that should be thinner than our normal silicon wafers. We cannot have a very
thick silicon wafer to do this processing because in the next in the kind of last step we need
to we need to etch whole of the silicon and reach on the top. So if the silicon wafer is too
thick it will be difficult to etch all the way till to the top.

So we start with the silicon wafer where the thickness is not very high. It is typically 200
to 300 micrometer. So in step D where we pattern this silicon we pattern from the back
side and we pattern this and silicon till here. So this silicon is also removed in this pattern
of silicon. So this part of silicon also gets removed and what we get as a final device is
something like this. We have silicon here. On top of this we have oxide and then we have
gold which is used for the contact and on top of this we have this V O X layer or vanadium
oxide layer and this is other side of silicon. So this is our final structure. So in this final

117
structure we see that this vanadium oxide this is a thin layer of vanadium oxide material
which is joined on the one side of this layer which is in contact with other device other
stack of layers and one side is actually hanging in the air. So if we see this is like a cantilever
or you can say micro cantilever which one end of this micro cantilever is fixed on the wafer
and the other end of this micro cantilever is free to move.

Now because of the material properties of this micro cantilever this micro cantilever is now
going to resonate along with its axis up and down. So this micro cantilever will resonate at
its resonant frequency. This resonant frequency depends on the material properties. We
have vanadium oxide the material properties are different so resonant frequency will be
different. If this micro cantilever is based on copper material then the resonant frequency
will be [Link] now this micro cantilever V O X based micro cantilever is now free to
move in up and down direction and the frequency is fixed based on the vanadium oxide.
Now this resonant frequency is a strong function of temperature because the mechanical
properties of material are temperature dependent and these mechanical properties are
giving rise to this resonant frequency. So let us discuss little bit about what is the resonant
frequency and how it change and how we measure them. So let us discuss something called
series resonant circuit. When we have a series resonant circuit something like a inductor or
capacitor and a resistor connected in series they have a particular resonant frequency and
a factor something called Q or the quality factor. This quality factor of resonant circuit is a
measure of how good the quality is for this resonant circuit. For example if the quality
factor is very good we have a sharp detectable change in the magnitude of this resonant
circuit. So we have the quality factor or we call it Q it is a measure of quality of resonance.
So a high value of Q or high value of this quality factor it means a narrow bandwidth of
the circuit when we excite this electrical circuit because the Q is higher it means we have
a very sharp response of this resonant circuit at that particular frequency where the quality
factor is very high. If we have a very sharp change in the kind of impedance of this resonant
circuit it is very easy to detect by electrical means and it is desirable for making a sensor.

So this high value of quality factor is desirable to make a sensor any kind of sensor in fact
not only the temperature sensor. So we have a higher value of this Q this corresponds to a
narrow bandwidth which is desirable in many applications. So for a definition Q is the ratio
of power stored to the power dissipated Q is the ratio of power stored to power dissipated.
Or we can write the Q or the quality factor is power stored in the circuit divided by the
power dissipated. So the quality factor or Q of a resonant circuit is the ratio of power stored
how much power is stored electrically in that resonant circuit divided by the power
dissipated by this electrical circuit. So to improve Q either we need to increase the power
stored in the circuit which is generally fixed and the second term is the power dissipated
which means how much power is being dissipated at the resonance. So this value actually
changes when we change the impedance of this circuit. For example beyond resonance the
impedance actually increases and below the resonance also the impedance increases at

118
particularly the resonance the impedance actually decreases and the current increases. So
because of this sharp change in the impedance in at that particular resonant the quality
factor is very high because the power dissipated is very less. So we have the definition of
Q which is power stored divided by the power dissipated. Now a series resonant circuit
looks like a resistance at a resonant frequency. So series resonant circuit looks like a
resistance at resonant frequency means the XL the impedance of inductor is equal to XC
which is the impedance because of the capacitor. So at resonance the impedance of
capacitor actually nullifies the effect of inductor and because of this null we get the
impedance as a purely resistive value because XL and XC cancels out each other and we
get a net impedance as the resistance of this series resonant circuit. So XL equal to XC at
resonance. So at resonance when this XL and XC cancels out we get a current which is at
maximum at resonance.

So the current is maximum at resonance and hence the impedance is at minimum. If we

plot the graph of current versus frequency it will look something like this. This is the
frequency axis on X. Let's say for example this is 100 hertz and this is 1 kilohertz. On the
Y axis we have current which is in milliamp. So if we plot this current versus frequency
graph for a series resonant circuit the graph looks something like it peaks up at certain point
and the base goes down at beyond this frequency. This frequency is the resonant frequency
and this is the resonance point. This is resonant frequency. Below this resonant frequency
the effect of series resonant circuit is dominated by a capacitor because the impedance of
capacitor dominates at low frequency and above this resonant frequency the effect of
inductor dominates because XL or the j omega L the effect of this inductor dominates at
high frequency. So below this resonance point this area is dominated by the capacitor and
above this frequency these frequency ranges these are dominated by the capacitive nature
of series resonant circuit and at this particular resonant point these capacitor and these
inductor values they cancels out each other and the net impedance looking into this series
resonant circuit is totally resistive. So this XL and XC cancels out leaving only the
resistance part means the impedance overall decreases and because of this decrease in the
impedance we see a sharp increase in the current here and this sharp increase in the current
can be easily detected by electronic circuits which are placed alongside this sensor. So here
we have this resonant point and the corresponding resonant frequency. So we can note
down few points for example at resonance this circuit looks purely resistive. Below
resonance it looks capacitive. And above resonance this looks inductive.

So here we see at resonant frequency the circuit behavior actually changes from the
capacitive to the inductive behavior and sharp at resonance the circuit is purely resistive it
means there is no inductive or capacitive effect these effects can be easily detected by
external circuits. Now these resonant current peaks can also be changed by adding some
external resistances. For example this resonant peaks can be changed by varying the series
resistor. And if we change this resistor which actually change the Q or the quality factor.

119
And when we change the quality factor of a certain resonant circuit it means the peak of
this resonant circuit actually changes and depending upon Q increases or decreases this
peak comes down and the curve actually flattens out.

It does not remain same the peak also varies and the broadness of this curve also varies and
depending upon the Q we have different different shapes of this current graph. So this also
affect the broadness of curve. So we can say that a low resistance high Q circuit this has a
narrow bandwidth and if we have a circuit where we have a high resistance or low Q there
we will have a wider bandwidth. So if we plot these characteristic for a series resonant
circuit we get something like this on the x axis we have frequency and on the y axis we
have again the current and now we can plot multiple graphs for multiple Q. For example
this is one graph where we have a high Q we can have another graph where for example
let's say this Q is 10 we can plot another graph where we have Q is limited so this graph is
let's say for Q equal to 5 we can have some other graph this graph is for Q equal to 2 and
let's say one more graph we have this is for Q equal to 1.

So if we see in this graph when the quality factor increases we see there is a sharp change
from the capacitive nature to the inductive nature of the circuit which is easy to detect
compared to the graph where quality factor is just one there is a gradual change from
capacitive to inductive nature and it is very difficult to change very difficult to detect where
the circuit is changing from capacitive to inductive. So having a high quality factor is very
good for a sensor we can very sharply and very easily detect where the circuit is changing
its behavior from capacitive to inductive. So we see here that when the quality factor goes
down the peak goes down the peak value is somewhere here which goes down as the quality
factor decreases also the bandwidth of this circuit also changes because the bandwidth we
measure for a particular graph which is the point where this graph reaches 70% of its peak
value. For example this is the peak value for let's say we are talking about Q equal to 10
this is the peak value and the bandwidth of this graph or this graph for Q equal to 10 is the
point where this graph reaches 70% of its peak value. For example this is let's say the 70%
of if this is 100% this is 70% if we assume this is total 100% up to here.

So when the value the peak value decreases to 70% of its peak value this difference in the
frequency is the bandwidth of the circuit this is the bandwidth that this f1-f2 we can call it
let's say frequency 1 and this point is frequency 2 this bandwidth is f2-f1 the difference in
the frequency where the magnitude reaches 70% of its peak value. We can write it the
bandwidth is equal to fc over Q where fc is the resonant frequency divided by the quality
factor and we measure the bandwidth or we can call it delta f between 70.7% amplitude
points of series resonant circuit. So here we see when we have a series resonant circuit it
changes the behavior from capacitive to inductive and how sharp it changes the behavior
that depends on the quality factor of this circuit. So for example we have a very high quality
factor circuit where the Q is very high then we have a very noticeable change in the

120
behavior of the circuit we have this peak value which is very high so it is easy to detect and
make a sensor which is very much sensitive to this change in the peak value.

However if we have a quality factor which is very less then it is very difficult to detect that
change this change can be little bit on the left on the right which is difficult to detect by
electronic circuit so we want a higher and higher Q. So the same quality factor we will
measure now for our fabricated device which we just did it two slides back we made this
MEMS device which is based on vanadium oxide material where this micro cantilever is
hanging in the air fixed at the one end and it is resonating at its resonant frequency. So
based on this resonant frequency we can measure how much is the temperature change
because the resonant frequency is a very strong function of the temperature or the
mechanical properties are dependent very much on the temperature. So now we see the
behavior of this MEMS circuit so we have that MEMS that vanadium oxide based sensor.
If we plot the resonant frequency versus temperature on the X axis we have temperature
and on the Y axis we have now the resonant frequency.

Let's say this is origin this typical value is 40 kilohertz this is 20 kilohertz this is 30 kilohertz
just for an example and the temperature is changing from 0 to 100 degrees. This frequency
is the resonant frequency is now going to change and for the fabricated device which is
based on vanadium oxide the graph looks something like this which is fairly linear graph
with change in the temperature. So now we see the resonant frequency which means that
this vanadium oxide micro cantilever is actually fixed at one end and it is swinging at the
other end with certain resonant frequency. This resonant frequency is now decreasing with
increase in the temperature for this vanadium oxide based thermal sensor and if we plot the
quality factor of this MEMS device which will look something like this. On the X axis we
have temperature again between 0 and 100 and on the Y axis now we have quality factor
this is 500, 400, 300. If we plot the quality factor of this micro cantilever this looks
something like this. The quality factor has this kind of behavior. So now we see the resonant
frequency is a function of temperature in fact it is a very strong function of temperature
and the behavior of this resonant frequency is fairly linear with temperature. To make a
sensor the desirable property of a sensor is the output should be linear compared to the
measurement or the input quantity and here we see the resonant frequency is fairly linear
with the change in the temperature. Linear output is very much desirable to make a sensor
of very good quality.

Now this quality factor actually varies from let's say 400 to 600. Of course it depends on
the dimensional properties of micro cantilever and this exhibits no significant change or
significant dependence in 20 degree to 80 degree centigrade range. So these are the some
measured value in a particular vanadium oxide cantilever it's not that these are very fixed
these values can be changed if you change the geometrical properties of your sensor these
are one these are one measured value for a particular thermal sensor. The next point is there
is a linear shift in the resonant frequency with the temperature. So as we can see in the first

121
graph there is a linear shift in the resonant frequency with temperature change. Which can
be used for temperature sensing applications. So today we saw a MEMS based temperature
sensor where we are having this MEMS device or the micro electromechanical system
where we fabricated a micro cantilever fixed end and which is resonating at a particular
resonant frequency because of the thermal vibrations. Now this resonant frequency is a
very strong function of temperature and we see that the resonant frequency is very much
linearly dependent on the temperature change. So there is a one degree change in the
temperature gives a linear change in the frequency change the resonant frequency change
which is very much desirable function of a thermal sensor or in fact any kind of sensor.

So this is all for today.

Thank you.

122
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 13
Optical Sensors: Introduction and Classification

Hello, ah welcome to the course Transducers for Instrumentation. Today we will discuss
about the optical sensors, these are the sensors where we use light as a sensing
mechanism. There are multiple parameters associated with the light for example
intensity, phase, polarization, and many other parameters. We can use any of these
parameters and make a sensor which can measure our measurement for example velocity
or ah refractive index or any multiple types of measurements we can measure using these
optical sensors. These optical sensors primarily use light as a sensing mechanism. So if
we see we have in optical sensors we have a optical source.

So today we are discussing optical sensors and in this optical sensors we first have an
optical source. This is the source which generates this light or optical wave. So light
actually originates from this optical source and then goes on something called fiber. We
have a fiber here which is carrying this optical wave or light and after this we have a
measurement zone.

This is the zone where measurement is taking place. We have certain operand which is
applied here on this optical wave and the measurement is happening here. So this fiber
we can call it the feed fiber which is going from optical source to measurement zone.
After this measurement zone there is again one more fiber which goes from this
measurement zone to the optical detector. This optical fiber we can call it the return fiber.

So in a optical sensor we have primarily these three zones. One is the optical source
where the wave is actually originated and goes into the optical fiber. This optical fiber
then goes to the measurement zone where we apply this measurement. It can be a
pressure, it can be refractive index, it can be pressure, it can be any of the measurement
and this is the zone where some of the property of this optical wave changes based on the
measurement and after that we have a return fiber which carries this altered wave back to
the optical detector. So this return fiber it can carry the wave back to the original
destination itself where the optical source is sitting the same location or it can be a far
end destination or far end location where our optical detector is sitting.

It may be the same place or it may be the other place but primarily we have these three
units in a optical sensor optical source, the optical detector which detects this light and a
measurement zone where the measurement is taking place. So now we have multiple
parameters of a optical wave or light. Light is a kind of optical wave. We have multiple

123
parameters which we can choose which we can alter based on the measurement. For
example we have intensity, phase, polarization, etc.

So we have light may change in any of these optical parameters. Second is phase. Third is
polarization. Fourth is wavelength. And five is the spectral distribution. So, we have
these optical parameters for example intensity. So, when the light originates from a
source It has a particular intensity, and then it goes into a measurement zone where the
intensity of this light can be altered. For example we have certain different material in
between. So based on the refractive index relative to the first one for example fiber based
on these refractive index the intensity of light can be modulated and then the light at the
far end it has different intensity than the first one. So difference between these intensities
gives us the idea of which material was in between.

So in that way we can sense what that material is because all the materials have different
refractive indices and based on that the intensity modulation will be different. So
intensity can be one of the parameters. Second may be phase polarization how much
polarized the light is wavelength or we can say the frequency. This can also be changed
and can be used as a sensing element. The spectral distribution is also one important
parameter.

So when we send a optical wave we do not actually send a particular frequency we

actually send a band of frequencies. So that band propagates in an optical fiber and based
on the media in between this optical distribution. Some of the frequency can get absorbed
at that particular different materials. So based on the missing frequency at far end we can
identify which material is there in between which can absorb that particular wavelength.
So these are the few optical parameters which we can use to make a optical sensor.

So we can make multiple kind of optical sensors and some of the measurement for these
optical sensors can be optical sensor measurements. That means we can make a optical
sensor which we can measure temperature, pressure, we can measure flow of a liquid as
well using optical sensors. The magnetic field can also be measured. In fact we can
measure some other proper other measurement as well. For example the chemical species
which is the chemical present in the measurement zone.

What sensor we can make using optical sensors? Radiation, pH, pH of a liquid, humidity,
strain, velocity. Electric fields. In fact we can measure the caustic fields as well. So these
are some of the measurements which you can measure using optical sensors. For example
the temperature.

There is a parameter which in fact affects almost all the sensors. So let us say we have a
optical sensor where we are using optical fiber of length let us say 1 kilometer because
the optical fibers are very long in length and if the temperature changes the kind of length
of this optical fiber also changes there is a expansion in the optical fiber because of the

124
temperature. Because of the expansion of this optical fiber there is a change in the return
path of light from source to the back to the source or to the destination that time actually
changes because of the difference in the in the length of the fiber. So that we can detect
as a phase difference between the input and output and by the phase difference we can
measure how much is the length change in the optical fiber that change in the length is
corresponding to the change in the temperature. So in that way there are many
measurements which we can measure using optical sensors.

For example liquid levels, displacements, radiation, humidity, even electric fields. So
these are some measurements. Let us talk about some classification of these optical
sensors first. So the first classification can be based on the modulation and demodulation
process. In this we use multiple parameters for example intensity can be modulated, the
phase, the frequency, polarization etcetera.

This is first classification. Second classification can be based on their application. For
example we want to measure some physical quantity for example pressure. So this can be
one of the measurement or some chemical composition we want to measure using optical
sensor and maybe some biomedical applications. The third classification can be based on
the type of sensing.

So there can be two types of sensing. First is the extrinsic and the second is intrinsic type.
So we can classify optical sensors basically based on three types. One is the modulation
and demodulation. We are modulating some of the parameters for example intensity or
phase.

Second classification can be based on the application. Let us say we want to measure the
pressure. So we can say it is a pressure sensor which is using optical means. And the third
classification can be based on the intrinsic and extrinsic type which will come to this
point later. So the first one is based upon the modulation and demodulation.

So in this type of classification let us say we have an optical wave which we can
represent by simple equation E cos omega t plus phi. So we have let us say a equation E
which is the amplitude and this is a function of time let us say cos omega t plus theta
which is again a function of time. This E can also be a function of time. So this is a
simple equation of a optical wave and we can change any of this parameter. For example
intensity based sensor is the one Where we change the amplitude of this wave.

So intensity based sensor we change the amplitude E. So for example if we plot this
optical wave in three dimension let us say this is one direction, the other direction let us
take different color this is another direction let us say. So we have a optical wave which
is nothing but a electromagnetic wave propagating in this direction. So we have electric
and magnetic field vectors.

125
This is electric field vector. And then we can plot the magnetic field vector along this. So
that magnetic field vector is perpendicular to electric field vector. It will look something
like this. So these electric field and magnetic field vectors they are perpendicular to each
other and both are perpendicular to the direction of propagation of this wave. So this
wave is propagating in this along this black arrow which is let us say in the x direction
then in the y direction we have electric field vector and in the z direction we have
magnetic field vector.

So this wave has all these parameters for example it has the amplitude, it has a time
varying cos factor, this electric field is changing based on the time how when the time
progresses it is just like a wave. So we can change any of these parameters for example
intensity sensor uses the amplitude variation. We can choose another parameter for
example frequency based sensor. So we can choose omega which is a function of time.
This omega is the angular frequency which can be changed.

The third is the phase modulating sensing. This is theta and we can have a polarization
modulation. So we can choose any of these parameters for example the amplitude, the
frequency, the phase or even the polarization because we know that light is actually
polarized in certain direction. This polarization we can change and make use of the
properties of polarization to make a optical sensor. So this is the classification based on
the modulation and demodulation techniques.

This classification is based on the application. So here we can have multiple types of
application for example the chemical sensor, the temperature sensor or the strain sensor,
biomedical sensors, electrical and magnetic sensors, and the magnetic sensors. Others can
be rotation sensor for example, pressure sensor and displacement or position sensor. So
this is the classification based on the application. What application we want to serve to
use this sensor? For example chemical sensor is the sensor where we want to investigate
which chemical is there in certain cavity that we can measure using optical sensor.

We will see in some other in some next lectures we can make this optical sensor which
can detect which material it is. So for example, in sensor or temperature sensor we want
to measure the strain or the temperature. So based on these applications we can also
classify these type of optical sensors. The third one is based on intrinsic and extrinsic
one. The third classification is based on the intrinsic and extrinsic type.

So in this type of classification we have two types of sensor. First is we have a light
source which is generating the light or optical wave. This optical wave goes into optical
fiber and in the measurement zone or in between we have this measurement zone. For
example, we want to make a pressure sensor here and measurement zone looks something
like this. It has two bodies one on the top one on the bottom and this has these kind of

126
structures and pressure is applied in this Y direction and on the far end we have this light
detector.

So in this type of assembly we see light source is at one end and light detector is at other
end. In between there is a there is running a optical fiber which is carrying this optical
wave. In the measurement zone we have a this kind of assembly on top and bottom both
from both the sides we apply this pressure. When we apply this pressure these both
bodies will come into close to each other and they will start bending this optical fiber. So
if when we apply this pressure on this a assembly our optical fiber will be bend it will
come something like this.

This optical fiber will bend zigzag and based on this bending of optical fiber the intensity
or multiple parameters of optical wave will change. So this is called intrinsic type
because the optical wave does not leave its media which is this optical fiber from the
source to the destination. It is a single optical fiber it is not cut in between or there is no
material in between inserted in this optical fiber. So it is called intrinsic type and this
optical fiber remains uncut from the source to the destination. So this is called intrinsic
type where this light travels inside the optical fiber.

The other type of sensor is extrinsic type where we cut this optical fiber and insert a
special measurement zone and that measurement zone actually modulate the property of
light. So in extrinsic type we have this light source. This generates the optical wave then
this goes on a optical fiber let's say I am showing the optical fiber with certain width.
This optical fiber is now cut here at this point and we insert a special measurement zone
here. So we have let's say a special measurement zone we will talk about this zone later.

This measurement zone let's say has a length L and this is a different media which is
coupled to this optical fiber and after this measurement zone we have another optical
fiber which is coupled at the output of this media and goes to the light detector. So in this
extrinsic type of sensor we have two wires two fibers one is which is connected from
source to the measurement zone which carries the input light and after that measurement
zone we have a different fiber which is coupled between the measurement zone and the
light detector. This carries the return wave or return optical wave based on the difference
between the input in at the input and at the output there are certain parameters are
changed based on the difference in these parameters at input and output we can detect
what is actually changed because of this measurement zone and based on the property or
shift in that particular parameter we can detect what that stimulus is or what that
measurement is. So we have here this measurement zone which is actually inserted in
between these two fibers.

And a separate zone is inserted in between. So, we can see there is a clear discrimination
between the intrinsic type of optical sensor and extrinsic type of optical sensor. If the

127
wire is not cut from the transmitter to the receiver then it is intrinsic type and if the wire
is cut in between and inserted certain different measurement zone then it is a extrinsic
type of optical sensor. So, there are certain advantage and disadvantage of these intrinsic
and extrinsic type. For example, we have the extrinsic one here and this side we have
intrinsic type. So, if we talk about the extrinsic type these are easy to multiplex.

However, the intrinsic type these are tougher to multiplex. So, what do we mean by
multiplexing? We have one fiber coming in and we want to send this optical wave to
three or four different type of different receivers. So, we need to multiplex or we need to
join all these three four wires to the single transmitter fiber. This is difficult in intrinsic
type because we are not cutting this optical fiber. So, there is no way we can attach
multiple units to this optic this intrinsic type optical sensor.

However, in extrinsic type we are cutting this optical fiber. So, now we can put some
multiplexers optical multiplexers after this fiber and we can connect multiple units of
receivers at this point. So, this is easy to multiplex in case of extrinsic type intrinsic type
is difficult to multiplex. The next one is the connection problem because we are cutting
this optical fiber. So, we have we may have multiple connection problems in connecting
different receivers.

However, in intrinsic type we have a very reliable connection because we are not cutting
the optical fiber. This is easier to use this is more elaborate kind of signal mode
demodulation. Some of the applications of these extrinsic type is the temperature,
pressure, liquid level and flow. These are some applications for extrinsic type intrinsic
type optical sensor. We have typical applications like rotation, acceleration, strain,
acoustic pressure, and vibration is also one of the measuring where we apply these
intrinsic type sensor.

So, these optical sensors though we use it in many applications they have certain
advantages compared to other sensors. For example, some optical sensors has these
particular advantages and the very first one is the electrical isolation. So, when we talk
about optical sensors or optical devices the very good advantage of these optical devices
is electrical isolation. When we make electrical devices when we have certain electric
circuit on one end and another electric circuit which is very close by connected to each
other. The one state change of the first electric circuit actually has some interference with
the second circuit as well.

There is a change in the one so it affects the output of the second as well. So, there is a
electrical interference because of the parasitic coupling between the first and the second.
Another example is for example we have two wires running in parallel. So, one state
change in first of the wire it actually couples to the second wire as well because there is a
certain parasitic capacitance in between these two wires. So, this electrical interference is

128
a big problem at very high speed circuits where we use a very small signal to send from
transmitter to receiver.

If there is a electrical interference with this wire from any other place or any other wire
running around then this signal is impacted by that interference this is a big problem in
electrical circuits. Optical circuits however they are not as sensitive as electrical circuit
because the parasitic capacitance or inductance they do not cause any coupling or optical
coupling between these two wires. So, they are used for electrical isolation very one very
good example of this is let us say we have small signal circuit and one end and on the
other side we have a very heavy load for example a big motor is running on the other end
a small microcontroller is controlling this motor. Now if this electrical motor need to shut
off we suddenly turn it off then because of the parasitic inductance there is a very high
voltage build up at this the input of the second stage the motor and this high voltage can
impact this microcontroller which is running at very low voltages for example pi volt or
so. This motor however is running at 250 volt or even beyond and this there is a spike of
voltage at the out at because of this sudden shutdown of motor it can go up to 1000 of
volts.

Now this high voltage has some parasitic capacitance connected to the input side and this
high voltage will easily influence the output of this microcontroller. So, this is always
there this parasitic capacitance or parasitic inductance these are always there in electrical
circuits. If we want to get rid of this application we can use some optical devices optical
devices like we have a microcontroller now the signal of this microcontroller goes to the
motor as a optical signal not as a electrical signal. So, there is a transmitter like for
example LED, LED is there which sends the signal and this optical signal goes to this
motor which is running at 1000 volt this has a optical receiver now this optical signal
goes from LED to receiver. Now there is no way coming back a optical light which can
be generated by electrical receiver cannot generate this light.

So, there is a optical isolation there is a electrical isolation between the input side and the
output side. So, these optical sensors these are very much used as a electrical isolation.
The other advantage is compact and light though these are subjective for example a
optical sensor can also be having a significant weight and size. The third is amenable to
multiplexing. So, as we discussed in intrinsic and extrinsic type we can multiplex the
output of these sensors to multiple receivers.

For example a sensor is sending its data in terms of optical wave which is generated by
this LED and this LED can now be coupled to many of the receivers which can
simultaneously receive this data. So, multiplexing of these optical sensors is very easy
compared to other type of sensors. The other is extremely high bandwidth capability.
This is also very important parameter about the optical waves the bandwidth available to
us in terms of optical wave is very high compared to the electrical bandwidth of signals.

129
The bandwidth is directly proportional to how much past we can send our data and how
much more data we can send.

So, that depends on the bandwidth available to us. In optical waves we have enormous
amount of bandwidth available to us which we can use to transmit our signal. So,
compared to the electrical signals we can send more amount of data at much faster speed
if we use optical sensors or the optical wave as a transmitting media. The other is high
sensitivity and high dynamic range. So, these optical sensors are very highly sensitive
because the quality factor of these sensors is very high we can very easily detect the peak
or the crest in these type of sensor. These are very highly sensitive and the dynamic range
of these sensors is also very high.

The other advantage of these optical sensors is the remote sensing. This is also a very
good advantage of these optical devices the remote sensing. For example, in optical
sensors we use optical fibers which are very thin in dimension and they run for very long
distances. For example, from one country to another we have optical fiber network this
optical fiber can carry the optical wave from one country to other country. In fact, very
long distances thousands of kilometers without degrading in the intensity. If in on the
other hand if we use electrical signals these electrical signals degrades with distance
because the electrical resistance of these wires when we run them for very long distances
the electrical resistance is very high.

Then the electrical signal we send this degrades or this dissipates in just like a power
dissipation and the signal actually degrades with the distance. However, in case of optical
sensors we can send these optical waves through these optical fibers to a very very long
distances thousands of kilometers without even using a repeater or without even losing
the signal intensity. So, remote sensing is very good application of these optical devices
where we can send these signal or this data from one end to the other end at very far
distances. The other advantage is the ability to embed in unfriendly environment or the
harsh environments.

So, here with harsh environment we mean. The environment which is full of noise and
interference though the optical fibers they run the optical wave actually is transmitted
through these optical fibers by in total internal reflection and they do not lose out to the
environment. So, whatever is the environment it is noisy or it has interference it does not
affect the quality of optical wave which is actually being transmitted inside this optical
fiber. So, we can employ this optical devices in harsh environments as well. The next
advantage is distributed and array sensors.

Covering extensive structures. And geographical locations. So here what we mean by the
distributed and array sensors, we can do a sensing in distributed nature for a particular
very long or very big structure. For example, we have a water dam which is very big in

130
structure. It's a 1 kilometer long kind of structure and with very massive amount of
concrete it is built on. And we need to be aware that how much is the strength of this dam
and whether there is a degradation in the strength of this dam. One way of doing it to put
multiple sensors, in fact thousands of sensors along the length of this dam.

Even then if we have let us say one sensor every 1 meter apart, there may be a crack or
damage in between these sensors. For example, there are two sensors 2 meter 1 meter
apart, there is a crack in between. So, this will not be detected by any of this sensor. So,
this is the problem of how many sensors we can put to monitor the strength of this dam.
We can do this by these optical sensors very easily which by because they can use for
distributed kind of sensing.

Distributed sensing is when we have a optical fiber which is running along the full length
of dam. So, this there is a optical fiber which is running from one end to other end
without any break. So, if there is any crack happening in between of the dam wherever it
is, this will be always detected by this optical fiber because this is a continuous fiber from
start to end. There is no 1 meter gap in between two positions. So, wherever there will be
a gap, there is a if there is a weakening of structure that can be easily detected by these
optical devices.

So, this is called distributed sensing where we are sensing not at a particular point, but
along a complete length or complete area we are sensing. So, today we discussed the
optical sensors, t heir classification, what are the different type of measurements. We saw
some advantage and disadvantage of sensors of these optical sensors. Some of them is
electrical isolation, high sensitivity, remote sensing and distributed sensing.

So, this is all for today.

Thank you.

131
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 14
Optical Sensors: Signal Propagation

Hello, welcome to the course Transducers for Instrumentation. Last lecture we were
discussing about the optical sensors, we saw the basic building blocks of a optical sensor
which are three mainly. One is the optical source where we generate these optical waves
or the light. The second element is the optical fiber which carries this light from the
source to the destination and in between we have the sensing done in the measurement
zone. This is the second element, and the third element is the optical detector where we
receive this optical wave and compare it with the generated wave and we compare how
the property of this wave is changed. Based on this change in the property we detect what
is the measurement in between.

So these are the three elements what we discussed and then we discussed about the
classification of optical sensors. So today we will discuss a little bit more about all these
basic building blocks and some basic understanding about the optical waves. So we have
the first object which is the optical source. So optical source is the element where we
generate these optical waves or the light.

So mainly we have heard about the LEDs or the light emitting diodes and the second one
is the laser. So these are the few devices by which we generate these optical waves, these
elements, these LEDs they receive the electrical signal from external in the electrical
circuits and generates a corresponding light wave or optical wave which is then coupled
to the optical fiber. So these devices or LEDs they work on a principle that the optical
wave it propagates in form of some quantas, quantas of energy. So these optical waves
are nothing but that quanta of energy which transmits at the velocity of light and these
quantas are called photons. So these photons are generated by these LEDs or lasers and
they are coupled to optical fiber where they propagate.

So we have a certain structure of these LEDs. So for example, the first point is these
optical sources operate on the quantas, these optical sources operate on the idea that
electromagnetic energy can appear in discrete amounts which is known as quanta. These
quantas are called photons. So when this energy is radiated in terms of optical light in
fact this consists of the quantas which is finite amount of energy and they are discrete in
nature. It is not a continuous band of frequency, the quanta or this energy is fixed with
this photon.

132
It is not a continuous band, it is discrete in nature. So the amount of energy one photon is
carrying that is discrete in nature. These energies depend on the frequency of the photon.
So the energy of these photons varies directly with the frequency. And some typical
optical sources are. For example, LEDs or the light emitting diodes. We call them LEDs
and the second one is the lasers. So these are two types of optical sources which are used
in semiconductor industry to generate these optical waves. The first one is light emitting
diodes where this emits the light and the second one is the lasers which also emits light
which is slightly different. The mode of operation of laser is slightly different than LEDs.

We will discuss that in detail. So these are the two types of optical sources we use and
typical construction of LED is something like that. We have a anode contact. On top of
this we have a insulating layer. This is anode. On top of this we have gallium arsenide
layer which is doped p-type. So we can say this is a p-type gallium arsenide layer. On top
of this layer we have let us say n-type L-Gan layer, L-gas layer. This layer is n-type
aluminum gallium arsenide. This is the n-type layer and on top of this we have cathode
contact. This is the cathode contact. So this is a typical structure of a LED or light
emitting diode. And in this structure we have an anode and cathode and in between we
have layers of aluminum gallium arsenide and gallium arsenide. These layers actually
generate these optical photons when we excite this layer with electrical signal when we
bias a voltage between this cathode and anode. And when these photons are made by this
layer they are actually coupled directly to the optical fiber. For example this is the optical
fiber. These photons will be generated at this layer between L-gas and gas layer and these
photons will be traveling this way and then they will travel within this optical fiber. Let
us say this photon this generates here and that travels this way. So all these photons are
generated between this p-type gallium arsenide layer and n-type aluminum gallium
arsenide layer. Here we have used this special material which is gallium arsenide and
aluminum gallium arsenide. We have not used silicon here because silicon is not suitable
for optical devices. We typically use gallium arsenide devices because the gallium
arsenide device or the gallium arsenide material is a direct band gap semiconductor. For
optical device to generate these photons we need a material which has direct band gap
semiconductor which is a direct band gap semiconductor that only can generate these
photons efficiently. However, silicon is a indirect band gap semiconductor. If we plot
briefly about the band gap of silicon so the maxima of valence band and the minima of
conduction band is not exactly for the at the same K point which is the momentum axis.
So this and this, this is the band gap for silicon. This is a indirect band gap material so
when a electron jumps from this minima to the maxima this has to take two path one is
from here to somewhere in between here and then it has to jump from here which is a
inefficient process. However, for the gallium arsenide this is a direct band gap
semiconductor and these both maxima and minima both are at the same K axis for the
same K value. So electron can directly jump from maxima to minima and can generate
photons in this process. However, in silicon we generate here from here to here this

133
energy is lost in terms of phonons which is not a optical particle it is actually dissipated
in terms of heat in the crystal.

So silicon if we use silicon to make a optical devices and we bias the silicon with
electrical signal instead of generating the optical photons actually silicon start heating up
because of the generation of these phonons. So silicon is not very much suitable for
optoelectronic applications we use something called gallium arsenide material which is a
direct band gap semiconductor. So these are the some optical sources where we have
LED's the light emitting diodes and the lasers we will discuss in detail later on in other
lecture but this is typical structure of a LED. Next is the optical detectors. So optical
detector is the element which receives this light these photons these are captured within
this device and corresponding to these photons this optical detector generates a electrical
signal according to how many electron how many photons are actually received at the
interface and these optical detectors mainly we use P-I-N diodes which is called P type
intrinsic and N type.

So it is a short form of P-I-N positive intrinsic and negative diodes P-I-N diodes. So the
most common optical detectors used in fiber optic systems is P-I-N diodes. These diodes
operate in reverse bias mode. Next is the P-I-N diodes which is positive intrinsic and
negative this is the structure of these P-I-N diodes we always apply a reverse bias voltage
across these devices so that we have a very high electric field built up inside this device.
So when a photon comes this has a particular energy when this photon hits this area of
very high electric field it generates this electron hole pair corresponding to its energy how
much energy it is carrying and these electron hole pairs are generated which is a electrical
signal these are the charged particles they generate this electrical signal across the load
which is connected outside to this P-I-N diode.

So in this way we generate a electron hole pair which is electrical signal from this optical
signal which is the photon. So this is how we generate these electrical signal and these
devices they operate in the avalanche mode. Avalanche is something when a single
photon comes in that hits the interface and generate a electron hole pair. This electron
hole pair again gains some energy from the external applied electric field which is very
high. When they start receiving this energy and they become energetic enough to knock
one more electron hole pair and in this way because of this chain reaction one electron
hole pair again generates two more electron hole pair, two electron hole pair make one
three like four electron hole pair.

So in this way there is a chain reaction and suddenly we have we see so many electron
hole pairs generated in the device. So a single photon knocks one electron hole pair but
eventually generates so many electron hole pair this is called the avalanche effect. So
these all these pin diodes or P-I-N diodes they work on the avalanche effect to generate

134
electrical charges. As a photo detector P-I-N diodes takes advantage of its wide depletion
region. In which electrons can create electron hole pair.

And for these devices the junction capacitances should be low to allow for fast switching.
So these devices these photo detectors they generate electrical signal from the photons
and the junction capacitances which is the parasitic capacitance of any of the electrical
device. This junction capacitance or parasitic capacitance should be low enough so that
these devices can operate at very high speed. We want higher and higher speed from
these devices. So we intelligently design these devices so that the junction capacitances
or the stray capacitances between two different materials these capacitance should be as
low as possible.

And the cross section structure of these LED's are like this these P-I-N junctions are like
this. We have first a cathode. This is cathode. On top of this we have n type layer. On top
of this we have intrinsic region which is little wider than these doped layers. This is
intrinsic region. And on top of this intrinsic region we have p type layer. And on top of
this we have second contact. These are our anodes. Now we apply a very high electric
field in the reverse bias condition across this device between the anode and cathode. So
when we apply a very high voltage across these two terminals a very high electric field is
developed in the device and that electric field will be confined more into the intrinsic
region which is already having no charged particles within inside.

So all this high electric field appears across the intrinsic region which is shown here.
Now if the photon comes in that strikes this intrinsic region where we do not have any
charged particle and because of the energy of photon it generates electrical charged
particle electron hole pair which then again start receiving energy from this high electric
field and knock down one more electron hole pair. And this process continues and within
a fraction of time this single photon leads to thousands of electron hole pair and that
electron hole pair gives us the electrical signal which we can take across anode and
cathode. So these photons or this light they come here and strike this intrinsic region and
generate electron hole pair and these electron hole pair further creates two more and then
they again create few more.

So this is avalanche effect. So this is how these optical detectors work based on the
avalanche effect. The third element of optical sensors are the optical fiber. So optical
fiber is nothing but a kind of wave guide for optical signals for these lights. So this light
when we couple this light to the optical fiber this light always stays within this optical
fiber it does not come out because of the total internal reflection. This light propagates
within this optical fiber from one distance to another and this distance is very long.

This light can travel even thousands of kilometers without even degradation in its
parameters. So the structure of a optical fiber is something like this. So first we have

135
optical fiber which is nothing but a wave guide for light. An optical fiber is essentially a
wave guide for light. The optical fiber consists of two main components one is the core
and second one is the cladding. This is the second component that surrounds the core.
And the index of refraction of the cladding less than that of the core which causes these
rays of light. Leaving the core to be refracted back into the core. So we have for example
if we see the cross section of a optical fiber it looks something like we have a core. This
is the core of this optical fiber and around this we have a cladding which is let us say this.
This is cladding and above this we have the jacket. This is for the protection of these
optical fiber. So this is the cross section cut of this optical fiber and we have the cross
section. This is how this optical fiber looks like. And this we have this cladding and then
we have this jacket for protection. So, this on the left side this is a cross section of fiber.
And when we send these optical photons into this optical fiber they all travel along the
length of this optical fiber through total internal reflection. These photons they never goes
out from this core to the cladding and they always remain within itself and they can travel
in this way to a very long distances. So this is how a optical fiber look like. These optical
fibers are made up of either some glass or plastic and they can be made in a very long
distances.

Their dimension is very thin but the length of these optical fiber is very long. So, these
optical fibers are made from thin strand of either glass or plastic. We can use any one of
it. So, they must be enclosed in some protective jackets. These strands these are very thin
and they run for kilometers and they have very mechanical strength. So, what we do on
top of core and cladding there is one more layer which is the jacket or the protective
jacket which keeps these optical fiber from breaking because of the mechanical shocks.
So, we have this protective jacket. Often these fibers we do not rely actually on the single
fiber because of some unexpected damage happen in between. So generally we use 2, 3
or more optical fibers in the same protective jacket or even with other protective jacket
but we use 3, 4 or 5 optical fibers or more than 5 just to increase on the redundancy if
there is a breakage of optical fiber some of the fiber at least one of the optical fiber will
be running fine. So often we use more than one optical fiber just in case one or more
optical fiber breaks.

So, this is due to mechanical shock or some other factors. Another important property
about using optical devices is we can make a full duplex system using these optical
fibers. What full duplex system is in the same fiber we can send one signal from one
direction and the another signal from the reverse direction. This is not possible in
electrical wire transfer because the voltage from one signal if we send one signal from the
left the voltage is there along this electrical wire. If we send another signal from the right
side this voltage is going to interference with the first signal and we these signals actually
corrupt each other. However in case of optical fibers we can send one optical wave from
left to right and another optical wave from right to left. So we can use the same hardware

136
to communicate both the sides. This is called full duplex system and it is easy to make in
optical devices. So we can make it is easier to make full duplex system. So, using optical
fibers or using these optical fibers.

So, these are some properties about the optical fiber system which we are going to use for
optical sensors. Now let us discuss more about how a optical wave actually transmits or it
propagates through the optical fiber. So these optical fibers work on the principle of total
internal reflection. So, let us see. So, all these transmission of these optical waves in the
optical fiber that works on the principle of total internal reflection. It means when this
optical wave hits the interface between core and the cladding this does not propagate to
the cladding and it remains in the core itself if the angle of incidence is greater than the
critical angle. So when this wave is traveling in the core it does not refract to the cladding
and it keep on bouncing back in the fiber itself and it can travel a long distance without
even getting out to the cladding. So this is called total internal reflection and the angle of
refraction is governed by the Snell's law. So, this angle of interface between these two
medias is governed by the Snell's law. And this law says N1 into sin theta 1 is equal to
N2 sin theta 2 where this Snell's law says N1 sin theta 1 equal to N2 sin theta 2 this N1
and N2 are the refractive indices of both the medias N1 is the refractive indices of media
1 N2 is the refractive indices of media 2 sin theta 1 the theta is the angle between the
perpendicular to the interface and the angle of incidence at what angle the wave is
actually coming and theta 2 is the angle in media 2. So we can plot this for example, if
we have two medias let us say this is media 1 the with refractive index media N1 and this
is media 2 where refractive indices is media N2 and this is the interface between these
two media this is the perpendicular axis to this and a wave is actually coming from media
1 to media 2. So this is the angle which is theta 1 and this is the angle which is called
theta 2 and the relation in between these theta 1 theta 2 N1 N2 is governed by this Snell's
law. This is the case when we have eta 1 or N1 is greater than N2 or N2. So where here
we have this wave which is coming from media 1 with the angle theta 1 and goes into
media 2 where the angle is theta 2. We can see here the theta 2 is greater than theta 1
because we have N1 greater than N2. So this is a typical case when wave is incident at a
particular angle theta 1. Now if we keep on increasing the theta 1 at some point this theta
2 angle will becomes 90 degree. It means when the wave is incident on this interface this
wave does not go into N2 or the another media or we can call this is cladding and this is
core. Let us say N1 is our core of optical fiber and N2 is the cladding. So now a wave is
coming from the core at angle theta 1 and theta 1 is such that theta 2 becomes 90 degree.

In that case the wave does not go into the cladding, but it remains at the interface. So we
can draw that particular case. This is theta 1, this is theta 2, this is the interface and the
wave comes from media 1 and it remains this theta 2 becomes 90 degree. This is 90
degree. So this is a particular case when the theta 1 is such that the theta 2 becomes 90
degree and in this case we call it the critical angle. Theta 1 is called the critical angle. So

137
if the incident wave from material 1 has an angle which is greater than critical angle this
wave does not go into the cladding, but remains itself in the core only the core of optical
fiber and if we further increase this theta 1 this incident wave angle then instead of
having on the surface itself this wave actually bounce back into the core itself. So we can
plot that third case where this angle this is N 1, this is N 2 and now the incident angle is
more than theta 1. In this case this wave actually comes back into the cladding into the
core itself. Let us say this is the angle theta where theta is greater than theta 1. Let us say
we call it theta critical. This is our theta critical angle and now our theta 1 is greater than
theta critical. So this wave actually does not go into the cladding it comes back into the
core itself. So this is called the total internal reflection. In this case where the incident
angle is more than the critical angle it is called total internal reflection. By this
phenomena this wave can travel a long distance whenever it is hitting the core and the
cladding interface it is bouncing back into the core itself. So the whole of this optical
wave remains within the core itself it does not go into the cladding or does not lose out
the photons outside into the cladding. In this by this phenomena this wave can travel even
very long distances even kilometers as well. So this is the phenomena which is used in
the optical fiber transmission. The next important thing is the numerical aperture. When
we couple these optical devices for example the optical source when we couple this
optical source into optical fiber these waves need to be incident on the optical fiber at a
particular angle not greater than a certain angle which is governed by numerical aperture.

When we couple these optical source with optical fiber there is a cone shape of structure
where this angle actually is very critical so that all the optical wave is coupled into the
optical fiber and it does not bounce back or get outside. So we have something called
numerical aperture. This is closely related to the critical angle. For example, we have a
optical fiber here which has core and cladding in itself. So now let us extend this axis and
we see there is this cone sort of area.

If the light is incident within this angle then only this light will be coupled to the fiber. If
the light comes away from this or the angle of incidence is more than this then this light
is not coupled to the optical fiber. So this critical angle is governed by numerical aperture
and this is given by angle of acceptance. This angle let us say theta this is the angle of
acceptance. This is twice the angle of given by the numerical aperture. This theta is 2
times sin inverse numerical aperture. So this angle of acceptance is double of this angle
what this cone actually is making at this horizontal axis. This N1 and N2 are the
refractive indices of both the media N1 and first and second. In these optical we can also
make switches in optical devices and these switches are unlike the electrical switches in
electrical switches we have relays and transistors. Here we can use make use of prism
which is optical device and we can make switches using these prisms.

So we can make optical switches. So one way to make switches for example we have 3
fibers coming in and we want to switch the input from one of this to the other one at the

138
output we have this one fiber. We can physically move this fiber up and down. We can
align first we align with this so the signal goes here then we can move this whole fiber
up. We make into this position then the signal is coupled from this let us say we number
it this is 1, this is 2, this is 3. So we can couple the input from either from fiber 1 or fiber
2 by moving this fiber up and down. This is one way of doing it however we can use the
prism which is like this we have let us say these 3 fibers let us say this is input fiber and
this is output fiber 1 and this is output fiber 2. Now in this assembly input fiber is
carrying this optical wave which need to be switched between output 1 or output 2. So we
can make use of prism a prism something like this. So in first case we have prism placed
like this so now our optical wave goes here this bounce back here and then goes to output
1. So this is first case when the output is switched to When the input is coupled to output
1. In another case we can move this prism to the down physically move it in the
downward direction and let us say this is the second case where we move this prism
something like this and now the wave is coupled to this prism goes down and coupled to
output 2. So this is case 1 and this is case 2. So in case 1 we see if we keep the prism on
the upward direction then input is coupled to output 1 if we move the prism down the
input is coupled to the output 2. So, in this way we can make these optical switches
though there is a physical motion involved moving of the prism is not very much accurate
way of steering the signal from one fiber to other fiber. Electrical ways of switching the
signal from one to other is easy and more reliable compared to physical movement of
prism.

The problem with physical movement is the wear and tear along with the time and the
reliability of how accurately you can place or you can place this prism so that the input is
coupled reliably to the output. So, this is the problem with using these optical switches,
but these are available to use. Another point we need to notice we need to use lenses.
These are necessary with these approach to avoid excessive heat excessive loss of light.
So, we can use this prism as optical switch, but the problem is the reliable connection so
for that to keep the loss of light under control we need to further use the lenses so that we
can direct this incoming light at a very particular point and reflect it back and couple it to
the output. So today we see the optical sources optical detectors and the basic working
principle of how the light propagates into the optical fiber.

So, this is all for today.

Thank you.

139
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 15
Optical Sensors: Interferometric based

Hello, ah welcome to the course Transducers for Instrumentation. Last lecture we

discussed about the basics of optical sensors and some optical sources and optical
detectors and how the wave actually travels in inside the optical fiber. Today we will
discuss some real optical sensors which we use to to sense certain measurements. So we
will basically discuss three types of sensors. One is the first one is the interferometric
sensors. The second one is the distributed sensors and the third one is the Bragg grating
based sensors.

So first we have interferometric sensor which is a point sensor. It means the the
measurement which we are doing is is limited to a confined zone. We have a specially
designed a measurement zone where the measurement is taking place with this sensor and
the data comes back to transmitter of the receiver. The second sensor which we are going
to discuss is distributed sensor. As we discussed earlier if we have a large structure and
we want to monitor its performance or want to sense certain parameter along the entire
length then we need to have a distributed sensor. It is not it cannot be done by using a
point sensor or we cannot use multiple sensors to monitor the entire length of certain
structures. In that case we need a distributed sensor. The third one is Bragg grating based
sensor. This sensor which we can use for the distributed or for the point sensing as well.
So these three sensors we are going to discuss. So the first one is interferometric sensor.
These sensors provide a very good level of sensitivity. And some of the applications of
these sensors are. Temperature sensor. Wind sensor. Pressure sensor. And we can
measure even the refractive index of liquids as well. So these are some applications of
interferometric sensors where we can apply these sensors to make the measurement. The
good point about interferometric sensor is they have a very good level of sensitivity.
These interferometric sensors are all again we have four kinds of sensors in these
interferometric sensor which are. The first one is Fabry-Perot sensor. The second is
McZandie sensor. Or we can call it MZ sensor. The first one we can call it FP sensor.

The third one is Michelson sensor. And the fourth one is Sagnac sensor. So these are the
four optical sensors which we are going to cover in these interferometric sensors. So let's
first discuss about the Fabry-Perot sensor. This is Fabry-Perot sensor. Or we can call in
short FBI sensor. So this is generally composed of two parallel plate reflecting surfaces.
Which are separated by a certain distance. And the interference occurs. Due to multiple
super positions. Of both reflected and transmitted beams. So in this FBI sensor we have

140
two interfaces. And the wave travels. Let's say wave is traveling in X direction. And then
we have two interfaces. Or two discontinuity in the path of this optical wave. This wave
is going to reflect at both of the interfaces. Some part will be reflected back at first
interface. Some will be transmitted. And some part of this transmitted wave will again get
reflected at the second interface. So when we look at the origin we get two waves. One is
the reflected at interface one. And the second one is reflected at interface two. And there
will be a time difference between these two waves. Because this second wave need to
travel a extra distance of separation. What this the separation is between these two
discontinuities. So accordingly there will be a phase difference between these two
incoming between these two reflected waves. And based on that phase difference we can
determine what is the time taken by the wave in between these interfaces. So let's say we
have a optical fiber here. This is cladding and inside this we have core. And the wave is
traveling in this optical fiber. Now we have first discontinuity here. Let's say this is my
discontinuity here. And my second discontinuity is at this point. After this we have again
this optical fiber. And the distance between these two interfaces is L which is a known
number. Now my wave is coming in from the left inside this optical fiber. The first
reflection will take place at this interface. Some of the wave will go back into the core
which will be reflected back at interface one. This is interface one. This wave let's say we
call it reflected one. Some part of this wave will be transmitted into this medium. And it
will be reflected at interface two. This is interface two. And this wave will go back here.

And we receive another wave which is reflected at interface two. Let's say call it R2. So
here we can see the wave R1 it traveled less distance compared to R2 because R2 has
traveled twice the length L which is the separation between the interface one and
interface two. So if we plot this R1 and R2 on the same scale this R2 will be time delayed
compared to R1. This will give us the phase difference between these R1 and R2. Now
this phase difference depends upon how much time the optical wave takes within this
cavity which is of length L. Now this cavity it may be filled by air or it may be filled by
some other liquid. We can pour some liquid into this cavity and because of that different
refractive index of that chemical or whatever we pour it the velocity of wave will be
different and accordingly the phase difference between R1 and R2 that will be different.
So based on the how much the phase difference we get we can detect what material is
poured into this empty cavity. So this is the working principle of this Fabry-Perot
interferometer. This is the case where we have this is extrinsic type it means the fiber is
cut in between and we have made a artificial cavity here and we can pour liquid here. We
can pour chemicals here and based on the difference of refractive index we get different
different phase difference between R1 and R2. This is the case of extrinsic type. We have
cut the fiber here. The another case we can make intrinsic type where we are not cutting
the fiber we have a single fiber only. So this is a single fiber and now within this fiber
itself we make discontinuity at this point and at this point. This we can do by laser
scribing and or many other methods. So this case is intrinsic type where we are not

141
cutting the optical fiber and here when we are sending this optical wave here some of the
part will be reflected at first interface and some part will be reflected at interface 2 and at
the input side we get two waves one is R1 and R2. So this is how if FBI sensor look like
we can have extrinsic type as well where we have a separate cavity where we can pour
the liquid. So this can work as a refractive index meter we can detect the material based
on the refractive index of that material.

The other is intrinsic type where we are not cutting the optical fiber and this can also be
used as a temperature sensor for example because when the temperature increases this
length which is a fixed number here we know because when we laser scribe it we know
this number very precisely this L if the temperature changes this length of this optical
fiber changes and based on this this change in the L R1 and R2 will get different phase
differences because R2 is the one which has to travel this length two times however R1 is
fixed in the in return path. So R2 always changes with change in the L which is a
function of temperature. So this intrinsic type sense intrinsic type FBI sensor we can use
as a temperature sensor. The phase difference of this FBI sensor is given by the delta FBI
is equal to 2 pi by lambda into eta into 2 L. So this is the phase difference we get in case
of FBI sensor where this lambda is the wavelength of incident light. This eta is the
reflective index of cavity material and L is the distance the physical length of cavity. So
this is the formula for getting how much phase shift we get we can see this depends on
three parameters one is the lambda what frequency or what wavelength we choose for the
optical optical wave light wave. So that is generally fixed by experiments we generally
choose a single optical source and based on that our wavelength is fixed. So this is
generally a fixed number the second parameter is the refractive index of cavity material.
So what we put in this cavity the refractive index of this material changes the phase
difference.

So according to this Ri of this cavity material we get the phase difference delta and the
other property other parameter where this phase shift depends is the length which is the
physical length of cavity which is generally fixed unless case of temperature sensor this
length of the cavity is generally fixed and no number when we make this sensor we
generally know how much is the length of this cavity. So this is again a constant number
so our phase shift depends on how much is the Ri of cavity material and we can make
some kind of chemical sensor using FBI sensor. To measure the refractive index of
liquids because the measure end measure end is the chemical actually in this case so
because the measure end can access the cavity very easily. So we have a cavity here in
FBI sensor and we can pour our liquid in that and based on the refractive index of this
material what we pour in we get different different phase shift. So if we plot the graph
between the refractive index and phase difference it will look like something like this.
First we plot the intensity normalized intensity on y axis and the round type on the round
trip delay this is nothing but the phase difference. So we see based on the different

142
materials we get different different peaks here. Let's say this is one peak we detect this is
for let's say ethanol the other material can be like this where we have acetone. Some other
case we can have like this where the liquid we pour in the cavity is water. So based on the
different different liquids what we pour in the cavity we get different different peaks in
this in this graph which is intensity versus the round trip delay and if we plot this
measuring refractive index which is our actual measurement this is measured refractive
index of liquid versus what we actually know that liquid let's say refractive index this
kind of sensor performs pretty well and we can see kind of a very linear fit with the
measured data. Let's say this is the measured data and this is the blue line is kind of the
labeled RI so we get a very good fit between the measured and the labeled RI and this
FBI sensor can be used as a refractive index meter of liquids. So this is FBI sensor. Next
we discuss the max and the sensor or the MZ interferometer. We can call it MZ sensor or
MZI sensor. These are commonly used in diverse sensing applications because of their
flexible configuration.

These are commonly used in diverse sensing applications because of their flexible
configuration. These MZI sensors have very flexible configuration and they have
basically two arms. One arm is considered as a reference arm which we do not change
and the other arm is called the sensing arm where we apply the measure end and the
difference between the parameters of these two waves one is traveling in the reference
arm one is traveling in the sensing arm. So based on the difference between these two
waves we figure out what measure end is applied at MZI sensor. So this sensor this MZI
sensor has two different arms. Which are the reference arms and the sensing arm. Here
this reference arm is kept isolated. From external variation. And only sensing arm is
exposed to this variation. So here we have one fiber which is coming in and we use some
optical coupler to split the wave in two direction.

This splitting is done by 3 dB coupler. So here in this diagram we have a wave which is
coming in from x direction and we use a 3 dB coupler which is which splits this wave this
single incoming wave into two different path of course the intensity will be half because
half of the wave come into the upper arm which is let us say called reference arm and the
half of the wave will go into the lower arm which is sensing arm. Now the reference arm
we do not expose it to the measure end only the sensing arm is the one where we apply
the measure end for example temperature which is the sensing arm is exposed to and
reference arm is kept isolated. So because this is the single wave which was split at the
input the sensing arm the parameter of this optical wave will change based on the
measure end and at the destination where we get both of these waves one wave will be
different than the other and because we send the single wave at the input we know by the
difference how much is the difference between the parameters based on that difference in
the parameter we can detect what is the measure end which we have applied on the
sensing arm. So we have here this is reference arm where which we kept which we keep

143
isolated and this is the sensing arm and the wave is actually travelling here as well as here
and here we apply the measure end. These waves one will be impacted by measure end
and the other will be isolated when they reach here at this far end we again use a 3 dB
coupler here and this 3 dB coupler is kind of reciprocal it can split as well and it can
again combined also.

So this 3 dB coupler again combine both the waves back and here at the output we get a
single wave. Now we have two waves in this one is let's say R1 which is from the
reference arm and one is the R2 which is from the sensing arm. This R1 is from the
reference and this R2 is from the sensing arm and based on the difference in these two
waves we can detect what is the measure end is applied for example the temperature is
applied then how much is the temperature is applied that we can measure using the phase
difference between these two reference wave and the sensing wave. So this is how this
MZI sensor works. Now there are many ways of creating the interference it in between
the core and the cladding of this optical fiber there are multiple ways by which we can
make the interference in the optical waves.

So we can discuss some of them for example we have one fiber inside this we have core
and the cladding. So let's say this is core and this black is cladding. If we make certain
discontinuities for example we put some grating here in the in the core so if in this fiber
we have introduced certain discontinuities which is the fringes here whenever this optical
wave sees a discontinuity it's tend it tends to this kind of refractive at that interface. So
this wave let's say this optical wave which is traveling here this will be refracted into the
cladding some part of this will be reflected into reflected into cladding and some part will
be remain in core and they will travel separately in core and cladding. However this
fringes they work in the reciprocal way when these waves reach here they can again be
combined by these fringes for these discontinuities and we get a wave here at the output
which is come which is a combination of the wave which is traveling in core.

Let's say this is our core plus our cladding. Some part of this optical wave was traveling
in cladding so that will also be recombined back at interface two and interface one is
splitting this wave into core and cladding. So this is one way of making interference the
second can be we have optical fiber and some section of this optical fiber is misaligned
with rest of it. Inside this we have this core. So we can see here some part of this optical
fiber is misaligned and we have discontinuity and let's say this is discontinuity one and
this is discontinuity two. These kind of discontinuity also act like a refraction at this
point.

So some part will be traveling in the core and some part will be refracted into the
cladding and again at the interface two this wave will be combined back here and we get
this our core plus our cladding. So any discontinuity in the path of light it causes the

144
optical wave to reflect to some other media. This is second case we can have multiple
type of cases for example in case number three we use certain grating we can make
certain grating inside this optical fibers. So here we change the material of this optical
fiber here and here. So now if the wave is traveling this side this wave will be splitted
here by this discontinuity some part will go into core and some will be in cladding and
again by the second discontinuity which is here this will be combined back into the core
and we get our core plus our cladding. So these are some of the waves by which we can
create the interference in optical waves. Some of them are preferred because for example
the grating which are easy to make using laser sources we can we can very precisely
make these gratings into the optical fiber and we can create these interferences. So this is
a good way of making these splitting these two waves from a single wave. So coming
back to the MZI sensor we can see a particular case where this MZI sensor we can use for
temperature measurement. So in this MZI sensor we take a optical fiber and we create
multiple gratings at the input side and some at the output side with a known distance l in
between.

This is d and this optical fiber we feed using a light source and on the other side we have
a detector. Now this is the assembly of MZI sensor we have a fiber where we have
precisely put this grating at the input side and at the output side with a distance which is
known and which is fixed l. Now we expose this assembly to the temperature which we
want to measure these gratings what we have put we have seen in last slide we these
gratings can actually split the wave into core and the cladding. So the wave coming in at
the left this actually gets split into core and cladding and again combined back at the
output between the core and cladding is combined and goes to the optical detector. Now
there is a difference between the temperature expansion of core and cladding material and
that difference gives rise to the difference in the phase between these two waves which is
one is travelling in the core one is travelling in the cladding and the interference fringes
are formed to the core mode the interference fringes are formed. And the phase change is
given by K into delta n effective and the l which is the distance between these two
fringes. So this is the formula to calculate the phase difference and if we plot this
transmission with respect to the wavelength on the y axis we have transmission on the x
axis we have the wavelength this looks something like this let's say this is one wave if we
change the temperature this transmission coefficient actually changes. So we have
different, different these different, different minimas for different temperatures if we are
changing the temperature here and if we plot this wavelength as a function of temperature
which is our actual measurement on y axis we plot the wavelength and x axis we put the
temperature we can see a fairly linear graph where this wavelength is linearly changing
with rise in the temperature this graph is very linear. So this is a typical output for MZI
sensor where we can see the output which is the function wavelength which is very much
linear to the temperature change a linear change is always desirable in making the sensor.
So this is MZI sensor which we have used as a temperature sensor here this length

145
actually is changing with temperature there is a elongation in the or the expansion in the
of this optical fiber length because of the temperature expansion and this length this
change in the length on the expansion in the length is giving rise to the difference in the
wavelength.

So this is MZI sensor. Or MI sensor. In MI sensor this is very much similar to the
previous sensor in previous sensor we have two arms one is the reference arm one is the
sensing arm and of this reference arm is kept constant and sensing arm is exposed to
measurement at the output we again combine these sensing and reference arm to get a
common output if we see closely this structure is very symmetrical from the input we
split up and again we combine to get the output if we cut this structure in between and put
a mirror is a physical mirror then this wave will be again reflected back to the origin and
because these waves can travel in both the directions there will be no kind of problem in
cutting this structure into half. So this MI sensor or Michelson interferometer is very
similar to this MZI sensor. The basic concept is same. Which is interference between the
beams in two arms.

But each beam is reflected at the end of each arm. So, in the last in the MZI sensor we
have this optical fiber and we use a 3 dB coupler to split the wave into two and one we
call the reference arm the second is the sensing arm this is 3 dB coupler this wave comes
from the left now instead of completing these arms we put some sort of mirror here this is
actual mirror where we are reflecting the light this light goes here and again reflects back
from here the same thing in the sensing arm it goes and reflect back here and 3 dB
coupler is also reciprocal so it can again combine these reflected waves and send it back
to the origin where we can detect it using optical detector. So now we see we need not to
complete this whole assembly and we can we need not to use another 3 dB coupler at the
output we just kind of reflect all the waves back one is the reference arm which we again
keep isolated so this is isolated and this is the sensing arm which where we apply
measurement and except this change in the structure all the phenomena of this MI sensor
is very close to the MZI sensor we can use this sensor in a similar way as MZI sensor.
The fourth sensor is the Segnik sensor or Segnik interferometer. So, these Segnik sensors
these are very much of interest nowadays because of their robust configuration easy
fabrication and we can use in multiple environments so these Segnik interferometer
sensors are of very much interest nowadays. The basic structure of these SI sensor is
something like we have a optical fiber which is coming in here and then we have a 3 dB
coupler which splits this wave which is coming from this fiber into two part and this fiber
both the end are connected to a single fiber in this way this is a circular loop of the fiber
and again we have a fiber going out this is our input and this is our output this is 3 dB
coupler.

146
Now from the input let us say one wave is coming here this is splitted by this 3 dB
coupler half of the wave will go in this direction half of the wave will go in this direction.
So, we can see in this assembly half this wave is split by this 3 dB coupler half wave will
go in this direction it means it is moving clockwise and the other wave is moving
anticlockwise so both these both of these waves they are originated from same wave, but
now they are traveling in opposite direction one in clockwise and one in anticlockwise.
So, right now the path difference in both the waves is same so there will be no phase
difference at the output now what we do we take this assembly and rotate it this whole
assembly is now being rotated let us say this is the rotation of this whole assembly. Now
what will happen the wave number one for example, this is the wave number one when
the rotation is in opposite direction this wave will take less time in reaching from one end
of this fiber to the other end because by the time it reaches the other end the assembly
moves in this the opposite direction so the end point actually comes nearer. However, this
wave number two it takes longer time in reaching from start to end because by the time
this wave reaches at the far end this assembly rotates in the same direction it means the
end point shifts further away from this wave and it takes more time compared to the wave
one which is taking less time.

So because of this difference in the in reaching the end point there is a phase difference in
between these two waves wave number one and wave number two so these are originated
from the same wave so we can compare the phase difference of wave one and wave two
and see how much is the speed of rotation of the assembly. So this the faster we rotate the
more will be the phase difference between wave one and wave two. Due to simple
structure and robustness. And send into opposite direction one return on return to the
point of entry the two light waves allowed to exit this ring and they undergo interference.
And the phase difference delta pi is approximated as eight pi upon lambda into omega. A
to A where this A is the area and omega is the frequency of rotation. So A is the area of
this loop and omega is the frequency of rotation. This phase difference here delta pi is a
function of this omega or how fast the assembly is being rotated and if we plot this phase
shift with angular velocity. On the y axis we have phase shift and x axis we have angular
velocity. The graph looks something like this which is the phase shift with angular
velocity and we can see that phase shift is actually increasing with angular velocity
because this delta pi is proportional to w the or the omega the angular velocity. So today
we discussed interferometric sensors which are basically four types and we discussed
each four of them here.

So this is all for today.

Thank you.

147
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 16
Optical Sensors: Distributed

Hello, welcome to the course transducers for instrumentation. In the last lecture we
discussed some optical sensors which are based on the interferometric properties of
optical waves means the light which was travelling inside the fiber that had some sort of
interference with the return light or with some measurement which is affecting this
optical waves. Today we will discuss about some distributed sensors. Distributed sensors
they are also optical sensors but they differ in the way they work. For example we have a
very large structure and we want to measure temperature or the strain across all the length
for example we have a kilometer long dam and we are interested in the physical strength
of this dam because the amount of water it is holding is too much. So we need to be
concerned about the strength how much strength this dam has and there is no kind of
degradation in the structure.

So we are interested in the entire length not at a particular point. So it is a one kilometer
length one kilometer long dam. We don't want to kind of put thousand sensors every one
meter apart and even though we do that there is a possibility that the crack can appear in
between that one meter. So it is very difficult to do point measurement for large
structures and that is why we use some sort of distributed sensors which are very easy to
make using optical fiber.

So today we are going to discuss distributed sensors distributed fiber optic sensors. And
as we discussed these distributed sensors is offers the ability to measure temperature or
strain at thousands of points. Along a single fiber. And these are very beneficial for
monitoring of large structures such as dams, tunnels, pipelines etc. So all these structures
are very much large in the length and this sensing systems these sensing systems are
based on brilliant on Raman scattering. So, these distributed fiber optic sensing this offers
the ability to measure continuously at across the entire length we can measure the
temperature or the strain for a very large length such as dams, tunnels and pipelines etc.
The property we use in these distributed kind of optical sensor is we shine a very high
intensity light into the optical fiber and this optical fiber is very long in length perhaps
kilometers or even long. We shine a very high intensity of light into this optical fiber and
there is a reflection at every point into the optical fiber because this light is very intense
there is a certain amount of light which reflects back across and across the entire length
of the optical fiber and this reflected wave by the total time from where we send the light
and we receive the light we can estimate at what point this light is being reflected
depending upon this time duration we can estimate at what point this particular wave is

148
reflecting that and this is how we do the measurement. The property is we shine a very
high intensity light into the fiber and there is a reflection at across all the length of optical
fiber. So the property is if an intense light at a known wavelength is shown into a fiber a
very small amount of it is scattered back from every location along the optical fiber itself.

So when we are shining a intense light into the optical fiber across its entire length there
is a scattering of light and that light comes back to us which we can measure using
optical detector. Now this optical wave or the light is made up of photons which has
particular wavelength but we have already written here the known wavelength we send
with a very high intensity. So the wavelength we know wavelength means the frequency
or the energy of photons we know upfront what photon we are sending in when these
photons are reflected back from the optical fiber then if the optical fiber at that particular
point let's say in the middle of optical fiber there is a zone where the temperature is high
it means we have more thermal energy there concentrated in towards the center of this
optical fiber when this photon is reflected back from that point this photon receives
energy from this temperature which is higher temperature or the thermal energy is higher.
So this thermal energy gives rise in the energy of photon when it is reflecting back so
what we receive the reflected wave is higher in the energy or higher in the frequency
means the wavelength goes down. So we are sending a known wavelength photon into
the fiber and we are getting a photon which is higher in the energy it means this photon
has received some energy somewhere in the optical fiber.

So depending upon the difference of energy we can find out how much is the temperature
at that point into the optical fiber and that point we can calculate based on how much is
the round delay of this photon when we are sending and when we are receiving so there is
a phase difference between input and what we receive as the output so there is a phase
difference so that tells us how far that point is and a difference in energy gives us how
much is the temperature rise. So this is how we do the measurement. So this is the
property we use and based on these properties when this photon receives energy from
temperature we have two cases so this the Raman interaction this is called Raman
interaction. So the first is the stoke we call it stoke this is the outcome when the material
absorbs energy the material or the optical fiber absorbs energy and emitted photon has a
lower energy and the second point is anti-stoke the second outcome is anti-stoke where
the material loses energy. And the photon we receive has a higher energy than what we
send.

So these are the two cases when we send a known wavelength of light into the optical
fiber the returned photon may have higher energy than what we send it means it has
received energy from thermal energy in between and the other case is the received photon
has less energy it means it has lost some energy somewhere in the middle of optical fiber
means some zone has the temperature which is lesser than the normal temperature it
means the thermal energy is lost there to that low temperature. So these are the two cases

149
where we have stokes and anti-stokes when we shine this intense light. Now these the
structure of this distributed optical fiber is very simple we have a very long optical fiber
for example, this is distributed optical fiber sensor. So we have a very long optical fiber it
may have turn in between and these distances are very long typically let us say one
twenty meter loop this is twenty meter loop and let us say this is hundred meter loop. So
this is the optical fiber which we are using now we shine a high intensity light so we use
some sort of a lens here to couple all the light into this optical fiber.

So this is the light let us say which comes from here and this is directed into this optical
fiber using this lens. This light is actually sent by laser or something which is very high
intensity in nature so we have another lens which is coupling this light to this laser we
have a laser source. So laser gives you very high intensity of light which goes into this
lens and this intense light is coupled to this optical fiber which is very long in length. So
as we discussed from each and every point of this optical fiber the photons will be
reflected back from this entire length which is very much big let us say kilometers from
every point these photons are reflecting back and now we need to read these electrons
read these photons means we want to measure how much is the energy of these photons.
So we can intercept this path in between where this lens is sending this high intensity
light but as well as receiving these small amount of photons which are reflected back we
intercept this path in between we are using some sort of beam splitter here let us say this
is beam splitter put at 45 degree so this reflected light will be moved towards this let us
say here and we have some sort of a grating at this point this light again goes to a lens
which couples it to some APD or avalanche photo diode this is the detector which we will
discuss next this light is coupled to some sort of APD this is avalanche photo diode which
in fact generates an electrical signal this is generally fed to an amplifier which is all
electrical amplifier so from this onward all the signals are electrical in nature we have
amplifier and then we have certain processing unit and then we have this beam.

So this is the typical structure of distributed fiber optic sensor we are shining a light into
the optical fiber using a laser which is very high intensity light source this light is coupled
to this optical fiber using lens so when this light goes in this is reflected back across the
entire length of optical fiber so photons are coming back as well so we put in between
this beam splitter which is so this beam splitter is splitting this incoming and outgoing
optical wave and sending these received photons towards the avalanche photo diode or
APD this APD is a semiconductor device which detects the photon coming in and
generates the equivalent electrical output this electrical output is generally very small in
magnitude so we prefer to have a amplifier before processing this signal so this is a
electrical amplifier we put after a lunch photo diode this amplifies this amplifier
amplifies the signal then we can use some microcontroller sorry our computers to process
this data this amplified signal and we can put it on the display so this is how we measure
how much is the photon energy compared to what we sent using the lasers so typical

150
stoke and anti-stoke wavelengths are let us say anti-stoke wavelength is typically 1450
nanometer again it depends on what you are sending through this laser and typical stoke
signal is 1650 nanometer based on these frequencies we classify the Raman filter which
we use in this distributed optical sensors so next we discuss the Bragg grating based
sensor which is which is a different phenomena of measuring a measure and using the
optical properties of material Bragg grating based sensor is we put a grating in the optical
fiber and depending upon the grating what we inscribe on these optical fiber this optical
fiber absorbs or reflect certain wavelengths of optical light and depending upon what is
the grating size and grating distance that wavelength is different based on this spacing of
these Bragg's grating we can detect what wavelength is going to come back or going to
be transmitted and this this spacing between the grating is the function of temperature so
this is what this is how you can make a temperature sensor using Bragg's grating based
optical sensors. So we have Bragg's grating based fiber optic sensor. So this fiber optic
fiber Bragg grating or in short we sometime call it FVG is a type of distributed sensor. In
this FVG sensor a short segment of optical fiber reflects be passing by. So at the
transmitted end what we receive we receive all the wavelengths except the one which is
reflected back by the grating.

So you can see here in this graph we are missing this particular wavelength here. This is
let us say lambda dash and on the receiving end where on the input side where we are
sending this wavelengths we get this reflected wavelength. This is power and this is the
reflected one reflected wavelengths and there we do not have other wavelengths only this
lambda dash. So this is the reflected wavelength which we receive at the input side and
these gratings these are very simple to make on the optical fiber. We photo inscribed
these gratings on the optical fiber during the manufacturing process itself.

We can precisely inscribe these gratings and we know the distance very precisely into
these gratings. These gratings are simple sensing elements which can be photo inscribed.
So this is the property which we call fiber drag grating based sensor. This is the property
we want to use now to make a actual sensor. So what we know is we put a grating and
based on the spacing of this grating a certain wavelength is reflected back from this
grating and this wavelength is corresponding to how much spacing we have in between
this grating.

So now using this property now we want to make a sensor. Therefore we are making here
a drag grating based sensor. So this particular wavelength is reflected by this grating
which has a constant gradient spacing in between. Now this spacing we can subject to the
measurement. For example we want to measure the temperature. So when we expose this
grating to the outside temperature let's say that temperature is higher there will be a
thermal expansion in the optical fiber based on that if there is a thermal expansion the
spacing between these grating changes. So earlier let's say this is 100 nanometer when the
grating is subject to some temperature which is higher then this 100 nanometer becomes

151
110 nanometer. So this difference between the earlier spacing and the new spacing the
wavelength which is reflected back that wavelength also changes. Earlier let's say we are
getting 600 nanometer wavelength reflected now that wavelength will be 650 nanometer.
So that reflected wavelength also changes based on the spacing the difference in the
spacing of these gratings which is proportional to the outside temperature.

So this is the property we want to use to make a temperature sensor using brag grating
based optical sensors. So the construction is again very simple. In this FDG sensor
instead of using a laser we generally use a broadband source. We use broadband source
because we need to send a wide spectrum into the optical fiber. So this is the optical fiber
and here we have let's say 3 dB coupler.

So what 3 dB coupler does it splits the waveform. This is the optical fiber here where we
have this grating and we apply the measurement here. This is the measurement zone. This
measurement can be anything it can be temperature, it can be pressure or force which
changes the spacing between this grating. This is the grating. be passing by. So at the
transmitted end what we receive we receive all the wavelengths except the one which is
reflected back by the grating. So you can see here in this graph we are missing this
particular wavelength here. This is let us say lambda dash and on the receiving end where
on the input side where we are sending this wavelengths we get this reflected wavelength.
This is power and this is the reflected one reflected wavelengths and there we do not have
other wavelengths only this lambda dash. So this is the reflected wavelength which we
receive at the input side and these gratings these are very simple to make on the optical
fiber. We photo inscribed these gratings on the optical fiber during the manufacturing
process itself. We can precisely inscribe these gratings and we know the distance very
precisely into these gratings. These gratings are simple sensing elements which can be
photo inscribed. So this is the property which we call fiber drag grating based sensor.
This is the property we want to use now to make a actual sensor. So what we know is we
put a grating and based on the spacing of this grating a certain wavelength is reflected
back from this grating and this wavelength is corresponding to how much spacing we
have in between this grating. So now using this property now we want to make a sensor.
Therefore we are making here a drag grating based sensor. So this particular wavelength
is reflected by this grating which has a constant gradient spacing in between. Now this
spacing we can subject to the measurement. For example we want to measure the
temperature. So when we expose this grating to the outside temperature let's say that
temperature is higher there will be a thermal expansion in the optical fiber based on that
if there is a thermal expansion the spacing between these grating changes. So earlier let's
say this is 100 nanometer when the grating is subject to some temperature which is higher
then this 100 nanometer becomes 110 nanometer. So this difference between the earlier
spacing and the new spacing the wavelength which is reflected back that wavelength also
changes.

152
Earlier let's say we are getting 600 nanometer wavelength reflected now that wavelength
will be 650 nanometer. So that reflected wavelength also changes based on the spacing
the difference in the spacing of these gratings which is proportional to the outside
temperature. So this is the property we want to use to make a temperature sensor using
brag grating based optical sensors. So the construction is again very simple. In this FDG
sensor instead of using a laser we generally use a broadband source. We use broadband
source because we need to send a wide spectrum into the optical fiber. So this is the
optical fiber and here we have let's say 3 dB coupler. So what 3 dB coupler does it splits
the waveform. This is the optical fiber here where we have this grating and we apply the
measurement here. This is the measurement zone. This measurement can be anything it
can be temperature, it can be pressure or force which changes the spacing between this
grating. This is the grating. Now this grating will reflect some of the wave which is going
this way but some wavelength will be reflected back by this grating. That wave will reach
here at 3 dB coupler and this 3 dB coupler will change its path to our detection system.
This is optical detector. This can measure which wavelength is reflected back by this
sensor and now if we plot the input and output. So we send a input where we have all the
wavelengths and at the receiving end at the optical detector we receive the input to
optical detector. Let's say before application of temperature the wavelength reflected is
lambda dash and when we apply the temperature then there is a change in the grating
spacing because of that this wavelength will shift and the new wavelength is let's say
lambda double dash. So there is a difference between the reflected wavelength and this
difference in the wavelength is dependent on the temperature what we apply in the
measurement zone. This is the received signal by optical photo detector.

This is the input and if we measure the optical wave at the far end, at far end it will look
something like this. We have this lambda and this is power and at the far end we will
receive all the wavelength except this lambda double dash. This particular wavelength is
lambda dash and this is lambda double dash. So this is how the Bragg's grating based
sensor works. This grating is very particular to a wavelength depending upon its spacing
and if the measure end can change the spacing of this grating then the wavelength
reflected by this grating is different which is proportional to the measure end. So measure
end can be temperature which can thermally expand the optical fiber or it can be force
where we are applying a force which can pull this fiber optic sensor and there is a change
in the distance between the grating. It can be pressure, it can be multiple other
measurements which we can use to make a sensing using of this fiber grating based
sensors. Now if we plot the relation between the reflected wavelength and the actual
temperature. So this is the temperature and on y axis we have the wavelength shift. This
Bragg's grating based sensor gives quite a linear response with the temperature and we
can see the wavelength reflected is increasing because when the temperature is increasing
that is because when the temperature is rising there is a thermal expansion in fiber the
spacing is increasing. It means the grating is now sensitive to the wavelength which are

153
more in length. So that is why we see for the higher temperature the wavelength reflected
is higher compared to the lower temperature. So this is the typical output of a Bragg
grating based thermal sensor. So this is all about the optical fiber where we discussed
interferometric based sensors, distributed sensors and Bragg grating based sensor. Now
let us discuss about the source the optical source which is the integral part of this optical
fiber optical sensors and the detectors which is also again as important as the optical
source.

So before jumping into the optical sources or optical detectors let us review quickly about
some semiconductor physics which is involved in working of these optical devices. So let
us brush up some of the device physics. So we know that there are three types of material
we have one is the insulator and if we plot the band gap of these insulators we have
certain conduction band and we have certain valence band. This is conduction band and
this is valence band and the energy difference between this conduction band and the
valence band which we denote by EG or the band gap.

This is high. This is typically 6 electron volt or even higher. So if we see the band gap of
insulator we have conduction band and valence band and the energy difference between
these two bands are very much higher if typically 6 electron volt or even higher. So it is
not possible for an electron to jump from valence band to the conduction band and if
there is no free charge carrier in the conduction band it means the conduction will not
happen. If we apply electric field across this material then there is no mobile charge
carriers in the conduction band which can conduct. So even if we apply a high electric
field there is no current flowing into the material and this material is called insulator. The
second case is we have metal. In metal we have this conduction band and valence band
they are typically overlapped. This is conduction band and this is valence band and we
have some area which is overlapped. This is valence band. So in metal we have these
both of these bands are overlapped to each other it means there is no energy difference
between valence band and conduction band and the electrons in the valence band can go
to the conduction band and they are freely available to move if we apply a electric field.
So for a metal if we apply a electric field or the potential electric potential then these
charge carriers which are electrons they are free to move and they conduct current. So we
when we apply a voltage across a metal it conducts current. So these are two typical cases
where insulator the band gap is very high; electrons cannot jump at all. In metal we have
no energy band gap it means the electrons are always available no matter what voltage
we apply the electrons are available for conduction. The third case which is a special case
is semiconductor. Semi means half something like that. So it is a semiconductor it means
it can conduct or it cannot conduct depending upon external influences. So we have here
conduction band and valence band which are not overlapped. So this is a conduction band
and this is valence band and the energy gap is there which is which we call band gap and
this band gap now is the order of one electron volt. Typically the band gap between the

154
conduction band and the valence band the band gap is typically one electron volt for
semiconductors. Now there is a band gap it means electrons refer to be in the valence
band they are all available in the valence band but if they acquire some amount of energy
from outside it may be thermal energy it may be the energy they acquire from high
electric field.

The energy which is typically in the order of one electron volt if a electron achieves this
much of energy it can jump from valence to conduction band and when the electron goes
into conduction band this is now available for electrical conduction. If we apply electric
potential across semiconductor then this electron can move and it can provide the current.
However if the electron does not receive enough energy of the order of one electron volt
all these electrons are in the valence band it means the semiconductor whatever voltage
we apply it cannot conduct. So depending upon the external factors this material can
conduct it can become a metal or like a metal or it can become an insulator close to
insulator it can conduct the current or it cannot conduct the current based on the external
factors. So this is called semiconductor and typically the band gap of this semiconductor
is typically one electron volt. So these semiconductors are of great interest and we make
all these optical detectors and optical sources using these semiconductors and even all the
electronic devices we use semiconductor material a very known very common material
which is semiconductor is silicon we use. So silicon is a semiconductor material very
heavily used in electronic circuits. However for optical devices we do not use silicon we
use some other materials we will come to that point. So here we have semiconductor
where the conduction and valence bands are separated by certain band gap. If we talk
about these semiconductor only at T equal to 0 means the temperature is 0 Kelvin so
there is no thermal energy.

So there is no thermal energy available to electrons from outside. So if we plot the

conduction and valence band the conduction band is typically like this and the valence
band is typically like this. At T equal to 0 all the electrons are in the valence band and
conduction band is completely empty it has available state which electron can fill up but
currently all these states are empty. So I am drawing it by open symbol which means that
these states are available to occupy but they are empty at T equal to 0. When we increase
the temperature, let us say at T greater than 0 means now we have thermal energy.

Some of these electrons which are very on the top of this valence band they have high
probability that they will receive this energy from thermal energy and this electron will
jump from maxima of valence band to the minima of conduction band because this is the
least amount of energy they need to jump and this electron will jump from here to here
this electron can also jump. So some of these electrons which are already high in the
energy they can jump from valence band to the conduction band. So this is EB and this is
EC energy of conduction band some of these electrons will jump to conduction band at T
greater than 0 kelvin when the thermal energy is available to all these electrons. Then we

155
have these electrons in the conduction band now if we apply a external electric field these
electrons. let us say this is the case when now we have these electrons available in
conduction band these electrons are here and rest of the electrons are still in valence
band.

Now if we apply a electric field this is with external bias at T greater than 0 kelvin. So
when we apply the electric field these electrons will see a potential difference or electric
field in let us say x direction and they will shift or they will move in a particular direction
for example these electrons will see a drift because of this electric field. This shape is not
symmetrical now some of the electrons will be towards the left because of the external
bias what we apply. And based on the electrons are there in the conduction band based
upon the curvature of these bands these conduction band the electron mass is different
which is called the effective mass of electron and this effective mass of charged particle
depends on the curvature of this conduction and valence band. So the relation between
this the effective mass and the conduction and the curvature is one over m0 which is the
mass of charged particle is equal to one over h square into d square e by dk square.

So this is the relation between the effective mass and the curvature. Curvature is d square
e by dk square is the curvature of this conduction or valence band and m0 is the effective
mass of electron. So we discussed today the Bragg grating based sensor and some of the
device physics what we need to understand the optical sources and optical sensor. Next
lecture we will continue from the device physics and We will see how the optical source
actually works.

So this is all for today.

Thank you.

156
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 17
Optical Sensors: PIN photo Detectors

Hello, welcome to the course Transducers for Instrumentation. Last lecture we were
discussing about ah the photo detectors ah these are semi semiconductor devices and
before discussing the photo detectors, we were discussing some of the device physics
which is involved in fabricating and understanding the functionality of these photo
detectors. So, we discussed earlier that ah based on the band gap of a material, we can
classify them into ah three kinds of material, one is metal, then we have a insulators and
then we have semiconductors where the band gap is not that high ah like a insulator or it
is not even overlapping just like metal. So, we have a decent band gap between
conduction band and valence band and based on that we can classify these
semiconductors. These semiconductors then we discussed the energy bands, we have at
temperature T equal to 0, we do not have any mobile electrons present in conduction
band. So, there is no kind of conduction in semiconductors for ah time for temperature T
equal to 0, but when we increase the temperature electrons acquire enough energy to
move to the conduction band and then they can contribute to conduction.

Then we discussed about the effective mass of a charged particle which actually depends
upon the shape of band gap. So, the how the band is actually the shape of this band gap
dictates how much will be the effective mass of that charged particle in that particular
energy band. So, let us discuss that energy band ah let us derive how the energy band
shape dictates the effective mass of a charged particle. So, the particle effective mass let
us derive the formula for this. So, let us say we have a E K diagram or the energy
momentum diagram, this axis is E axis and this is K or E the momentum and we have a
certain band something like this, let us assume that this is parabolic. So, for this band we
can write the energy of the charged particle E is equal to half m naught into V square.
This is the kinetic energy of this charged particle m naught is the effective mass and V is
the velocity. This I can write like E square upon 2 m where this E is the momentum
which is which can be written h into K or this E square upon 2 m I can write h square K
square divided by 2 m naught m naught is h h is the plant constant here. So, I can write
del E by del K the derivative of energy with respect to K is twice h square K divided by
m naught or h into P divided by m naught or I can write even h into m naught V power m
naught.

Now the velocity I can write 1 over h into del E by del K. So, here we can see del E by
del K is the derivative of this band with respect to K. So, del E by del K is nothing, but
the slope of this graph of this band with respect to K. So, I can write now here del square

157
E by del K square the second derivative of this band with respect to K is nothing, but del
by del K of h square K divided by m naught or it comes out h square upon m naught or I
can again rearrange it the mass or I would say the effective mass m naught of this we
have 1 over h square into del square E by del K square. So, this is the relation of effective
mass with respect to the shape or the second derivative of this E K diagram. So, here we
can see the effective mass m naught is inversely proportional to the curvature of this
band. So, here this is 1 over m naught in fact, 1 over m naught is equal to half 1 over K
square d square E over d K square. So, we can see the m naught which is the effective
mass is inversely proportional to del square E by del K square which is actually the
second derivative of this E K diagram or the or the this band. The second derivative
means the curvature how fast this slope is changing with respect to K. So, it means the
more the curvature of this band actually dictates how much will be the effective mass of a
particular charge carrier.

This does not depend on it is not a fixed value it changes the mass of this charged particle
it changes where this a charged particle is in that particular band. So, if a band is kind of
flat the mass will be different if the band is very sharp the mass of that same charged
particle will be different. So, this is the relation which is we can write in the text. The
second derivative of E versus K is inversely proportional to the effective mass of free
electrons. So, the mass of a charged particle is inversely proportional to the curvature or
the second derivative of the band with respect to K. Now, based on the band gap how the
band are located in a particular material we can classify these materials in two types one
is the direct band gap semiconductor and one is the indirect band gap semiconductor.
And we will see that why for optical devices to make a optical detector optical sources
why do we need a direct band gap material compared to an indirect band gap material.
So, we have two types of materials based on the band gaps we have direct and indirect
band gap materials. So, let us first discuss a direct band gap material we plot E versus K
this is etcetera and we have P like this and the bands are something like this. This is the
balance band and we have the conduction band which looks something like this is the
conduction band this is the band gap. Band gap is the difference in the energy of these
two bands at the maximum of balance band and the conduct the minimum of conduction
band. So, we have some charged particle let us say we say these are the charged particles
here and here and there are some charged particles in the conduction band this is direct
band. So, here if we see we see a conduction band where there is a minima and we see a
balance band where there is a maximum and both of these minima and maxima they exist
at the same value of K at the same value of momentum axis. It means these electrons
which are there this black these black electrons we have in conduction band and balance
band. These electrons can jump to for example, from balance band to conduction band
just by acquiring certain amount of energy they can acquire some energy from external
means and they jump from conduction to balance to conduction or conduction to balance
and in this process they need not to change their momentum because the momentum is

158
same for both minima and maximum. So, they can easily jump from balance to
conduction and conduction to balance. So, to jump from balance to conduction it means
go up they just acquire some energy and they can jump up in the reverse process they can
jump from conduction to balance and they can emit a photon they have to lose this
amount of energy the equivalent to E g the band gap they need to lose this much of
energy to jump from conduction to balance and this energy will be released in terms of
photon and the energy of that photon will be equal to the band gap of material which is E
g in this case. So, in this process we see the electron or the charge carrier they need not to
change their momentum it is a easy process just is the losing or acquiring some energy in
terms of photons. So, this process is easy or I would say efficient to jump from up and
down from one band to another this is called direct band gap material. So, this kind of
material is used for optical devices because they need not to change their momentum it is
a efficient process the material l can easily absorb the photon or it can easily emit the
photon.

So, this is a direct band gap material which is used for optical devices the other case is
the indirect band gap material which we will discuss now. So, in indirect band gap
material we have this k axis and we have this E axis and the band look like something
this is the valence band and the conduction band look something like this. So, here if we
see this conduction band minima this is the minima point this does not exist for the same
value of k where the maximum of valence band is. So, we have electrons in the valence
band which exist here and in the minima of conduction band the electrons exist here and
we have the band gap which is Eg which is the difference between the maxima and the
minima this value. So, here if we see we have a conduction band where the minima is at
different k value than the maxima of valence band means this electron which is in the
valence band if it has to jump to conduction band it has to do two process in that jumping
from valence to conduction the first is it has to acquire certain amount of energy
equivalent to the Eg the band gap.

So, that it can jump that much of energy barrier and it has to change its momentum the
momentum is kind of E is equal M into V it has to change its momentum to jump from
valence to conduction. So, it has to do two process and the same it is a kind of in the
reverse process when it jump to conduction to valence it has to lose certain amount of
energy it has to lose that photon with equivalent to some band gap and it has to change its
momentum which is a physical quantity we call it the phonon. So, one is the photon and
one is the phonon. So, photon is the amount of energy which is released as a optic as a
kind of optical energy in terms of photon and phonon is the amount of energy which is
acquired or given to the lattice. So, here if we see the conduction band this electron now
if it has to jump from this conduction to valence first it has to change its momentum and
come to the same momentum point.

159
So, this electron first it goes to the same value by releasing a phonon here this emits a
phonon and then jumping from this place to the valence band it emits a photon. The
phonon is something it change its momentum and release this energy to the lattice and
when it energy is burnt in the lattice the lattice heats up. So, this process of releasing a
photon is not as efficient or as good compared to the direct band gap materials. So, we
call this as indirect band gap semiconductor this process of releasing a photon is not that
efficient. So, we do not use this kind of material in making optical devices.

So, silicon is one example of this indirect band gap material where silicon if we bias it
does not in fact, efficiently emit photon, but it release phonon phonon releasing a phonon
means the energy is released to the lattice it means the device heats up. The lattice of the
material starts acquiring this energy from those biases and phonons will be generated
instead of photons. So, silicon will not be as effective or it is not usable for optical
devices. So, these materials are indirect band gap materials. Indirect band gap materials
and these direct band gap materials are good for optical devices.

So, now let us discuss about the photo detectors and how do we make these photo
detector devices these are made of semiconductor devices which actually receives a
photon and gives its response as the electrical signal. So, photo detector is a device which
converts optical signal into electrical signal. And this is the fundamental element of
optical receivers. So, these photo detectors are the device which converts the optical
energy which is photon, and it converts it in equivalent electrical signal which is current
or voltage. So, these devices when we employ these kind of device in circuitry most of
the time these devices are followed by some amplifier because the output generated by
these optical devices these are very small in magnitude because the photon receive and
converted into the electrical electrical output those numbers are small.

So, these optical receivers or the optical devices these are followed by an amplifier and
that amplifier is again followed by some single signal conditioning how do we process
this data. So, some of the requirements of these photo detectors are like we have some
requirement for these. The first one is they should be smaller in size. We can write
compatible physical dimensions means they should be small in size. The second one is
low noise and high gain.

The third is the fast response. It means they should have a high bandwidth. The fourth is
insensitive to temperature variation. And the fifth one is long operating life and low cost.
So, we have certain requirements from these optical detectors. First is the compatible
physical dimensions and should be small size because we need to couple these devices
with the optical fibers and optical fibers are very the diameter of these optical fibers are
very small these are very thin strands. So, the smaller size of these optical detector is
good for the operation. The second is the low noise and high gain. So, the device should
not generate much of a noise and the gain of these devices should be high. So, that we

160
can detect even a small amount of photon as well. So, that the gain of these devices
should be high. The third is the fast response time because the if we can have a fast
response time means that we can have a very high bandwidth and we can we can transfer
the signal at much higher rate. So, we should have a fast response time. The fourth is
insensitive to temperature variation which is in fact, a requirement for almost all the
sensors we have. And the fifth one is the long operating life and low cost because the
fabrication of these devices are very expensive. So, the fabrication cost of these devices
should be should be should we should bring it to lower value. So, that we can deploy
them in a large number of devices. So, we can have now photo detectors. And in these
photo detectors in fact, the photo diodes meet most of the requirements. Hence, widely
used as photo detectors. So, we use photo diodes as photo detectors because they meet
most of the requirements and there are two types of photo diodes we use. The first one is
the PIN photo diode or we can say the positive intrinsic negative or in short we call it
PIN. So, the these are type of photo diodes where we have no internal gain and robust
these are robust detectors. And the second one is the avalanche photo diode. Or in short
we call it APD avalanche photo diode.

In this we have it is an advanced version of this PIN where we can have with internal
gain. Let us say M this is due to the cell multiplication process. So, we have two types of
photo detectors or the photo diodes. One is the PIN photo diode where this diode does not
have any internal gain. It means one photon is received then one amount of electrical
signal will be generated. And the other is the avalanche photo diode this photo diode has
internal gain. So, if a one photon is received it multiplies the response within itself. So,
that we can get M times the response of what a single photon can generate. So, these are
two types one is PIN the the name itself confirm the structure positive intrinsic and
negative which which we going to discuss next. And the second is the avalanche photo
diode where we have the internal gain. Most of both of these devices need to be biased in
the reverse bias. We apply a high electric field across these devices in the reverse bias
case. So, we have these photo diodes are sufficiently reversed biased during normal
operation means no current flow without illumination. And the intrinsic region is fully
depleted of cables. So, these photo diodes are always sufficiently reversed biased we
apply the voltage such that this diode is always biased in the reverse bias.

And in the reverse bias we do not have a conduction a current flowing into the into the
device. And the intrinsic region which is in between this this P and N type for example,
positive intrinsic and negative this intrinsic layer is completely depleted of any charge
carrier. So, there is no current flow in the device we always bias in the reverse bias. So,
let us now let us now discuss how a P I N diode look like. This is P I N photo diode. So,
the P I N photo diode look like we have a diode here and it has mainly three regions. The
first one is P type this is P type dope and this is N type dope and in between we have
intrinsic region. And I have purposely made it bigger because the reception of signal is

161
going to happen in this intrinsic region only. So, this is our photo diode and we bias it in
the reverse bias. So, this terminal P type we apply to negative of the battery. So, this is
negative connected to P type and the positive is connected to N type. So, this whole
device is connected in reverse bias and we receive signal electrical signal across these
two points. We have here. So, we have here a resistance connected between these two
terminals. So, across this resistance we receive electrical signal this is our load resistance
R L and we receive signal across this load resistance which is our output. So, in this
structure if we see we have P I N three regions of the device this photo diodes P type
intrinsic and N type. So, intrinsic region is anyway depleted of charge carriers we do not
have any doping in this intrinsic region. So, no charge no freely moving charge carriers
here. This intrinsic region is exposed to the incoming photons. For example, we align this
intrinsic region to the optical fiber which is bringing these photons or the light energy
from other kind of side. These photons hit this intrinsic region and we have biased whole
of this here this photo diode in reverse bias it means the whole of the electric field from
the battery is applied across this intrinsic region. This intrinsic region is having a very
high electric field build up because of this reverse bias applied. So, this intrinsic region
which is applied to a very high electric field within itself now it is exposed to the photons
the photon comes in. So, what happens let us see we receive a photon here this photon
comes in here which is now being absorbed by this intrinsic region absorbed means the
energy of this photon is now released to the material what this energy will do this energy
will knock down electron and hole in itself which generates a electron hole pair. So, this
photon here it gets annihilated and generate electron hole pair one is the negative its
electron and one is photon sorry the hole.

So, it generates electron hole pair by releasing the energy to the the material and now this
electron hole pair they see a high electric field across this intrinsic region. So, they move
in in that electric field direction electron moves to the right and the hole moves to the left
and this electron hole pair is generated when these electron hole pair complete their
journey from here when they are generated and reach reach the terminal they develop a
potential across this load resistance because these are charge carriers when they move the
current will flow and that this current will be flowing into this load resistance which we
can measure here as a output. So, the output we receive across this load resistance as
electrical signal. So, this is how the current is actually generated in these photo diodes we
can write it down here the high electric field present in the depletion region causes photo
generated carriers to separate to be collected this gives rise to the current in the external
circuit which is known as photo correct. So, we apply this reverse bias and the photon
comes in release the energy and this release energy knocks down electron hole pair these
electron hole pair because of the applied reverse bias they move accordingly and when
this charge pair carriers move in the circuit they produce a current which is detected by
our load resistance and we sense this as a output of this devices.

162
Now, this electrical current we see is proportional to how many photons we receive if we
receive one photon only one electron hole pair is generated which gives rise to a small
current if we get 100 photons. So, it means 100 electron hole pairs will be generated and
it gives rise to more current. So, the same amount of current is detected by the load
resistance as a output. So, if we see the band diagram of this structure. We have a
conduction band and a valance band in p side. And we have a conduction band and
valance band in the n side and in the intrinsic region because we apply a high electric
field there is a slope of this conduction and valance band. So, this is how these
conduction band and valance band look like. This is p region this is n region and this is
our intrinsic region and when the electron hole pair is generated let us say we generate a
electron here and a hole here they move in to opposite direction electron comes down to
downward and the hole which look which act like a bubble it goes up when a photon is
actually getting this position. Let us say we have a photon which is having some energy
eg and the photon energy which is actually h into mu the frequency of this photon. So,
here we see the photon energy should be equal or more than the band gap of this material
because electron hole pair which is generated now this electron need to jump to
conduction band.

So, the amount of energy which is required to knock down electron hole pair is at least
equal to the band gap of the material. So, here eg is the band gap of this material what we
choose on. So, different different materials use different detect different kind of light. So,
for example, one material the band gap is close to the blue light it is a good detector for
the blue light. However, a different material where the band gap is more close to the red
light that material will be good for detection of red light. So, this is how we choose
different materials to detect different colors of light. So, this is the band gap of this is the
band energy band diagram of photo diode. Now, let us discuss we have some optical
power some optical power is is received by this device. Now the optical power absorbed
in in the depletion region. Can be written as P as a function of x is equal to P naught into
1 minus e to power minus alpha s which is a function of lambda into x. Here P naught is
the incident power. And x is the distance and this alpha which is in fact a function of
lambda is the absorption coefficient. So, let us assume P naught amount of energy the
optical energy is incident on this material how much of the energy will be absorbed by
this material this optical energy. So, when the energy falls the P naught energy falls into
the material right at the interface all the energy is not absorbed by the material. Some of
the energy some of the photons goes well within the material and they get absorbed at at
certain distance away from the surface. So, when this energy hits the material it goes little
bit into the material itself some photon goes into the material and goes and get
annihilated after a certain distance which is x and the relation between this x the distance
and the amount of energy absorbed is given by this P equal to P naught which is the
incident energy into 1 minus e to power minus alpha into x.

163
This alpha is the absorption coefficient how efficiently the material absorbs this optical
power or the photons. And this absorption coefficient is a material property different
materials have different different absorption coefficient. And this alpha is also a function
of the lambda which is the wavelength of optical energy what type of optical energy is
incident on the material. So, accordingly this absorption coefficient is also different. So,
this is the relation between absorbed energy and the incident energy. Next is this
absorption coefficient alpha strongly depends on wavelength. And the lambda of the
optical energy is given by we can write in micrometer. So, this formula directly give the
value in micrometer 1.24 upon the band gap energy which is in electron volt. So, if we
write the energy of the band gap energy in electron volt this formula directly give the
value of lambda in micrometer. The next is if we take the reflectivity also into account.
The absorbed power is given by 1 minus R f into P is equal to which is function of x is
equal to P not 1 minus to power minus alpha into x into 1 minus R f. So, here when these
photons are incident on the material not all the photons goes into the material because
some of the part or some part of these photons will get reflected back from this material.
And if we consider this reflectivity. So, let us say some portion of these photons are
reflected back how much goes into the material is 1 minus the reflectivity. So, 1 minus
reflectivity this is this varies from this 0 to 1. So, 1 minus R f is the amount of photons
the fraction of photons which actually goes into the material and then the absorption of
these photons which go inside the material they get annihilated in the intrinsic region
based on this formula P x equal to P naught 1 minus e to power minus alpha into x. So, if
we consider this reflectivity as well. So, we have to put the factor 1 minus R f also into
account. So, the next is the photo current produced by this absorption. So, when these
photons get annihilated in the intrinsic region they give rise to the photo current resulting
from absorption I p the photo current is equal to Q upon h into mu into P naught 1 minus
e to power minus alpha into x into 1 minus R f where this Q is the charge on an electron h
is the plank constant and mu is the frequency of photon.

So, the photo current generated is I p and if let us say this has certain efficiency as well
not all the photons give rise to a electron hole pair then we can have a quantum efficiency
eta which is equal to number of electron hole pair generated divided by the number of
incident photons or we can write it eta equal to photo current divided by Q divided by P
naught the incident photon energy divided by h mu. So, this is the quantum efficiency
formula and the next is the responsivity. We can write it by R is equal to I p the photo
current divided by the incident optical power P naught, or we can write eta P upon h into
mu. So, this is the responsivity formula where how much current is generated by P
naught amount of incident photons. So, this is the responsivity and quantum efficiency is
because not all the photons give rise to electron hole pair. So, how many electron hole
pairs are generated or unit amount of photons incident on the material. So, that is that
ratio is called the efficiency of the material how efficiently we can generate electrical
signal from the optical signal. So, this is the quantum efficiency. So, today we discussed

164
photodiodes and basically in the photodiodes we have discussed the P i n photodiode
where we have P intrinsic and n type region, and this photodiode does not have any
internal gain the amount of photons it receives the output electrical output is proportional
to the number of photons we receive at the input.

So, this is all for today.

Thank you.

165
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 18
Optical Sensors: APD Photodiodes and LED Sources

Hello, welcome to the course Transducers for Instrumentation. Last lecture we discussed
some optical detectors mainly the photodiodes or PIN diode and we saw that the structure
of this photo detector is P type intrinsic and N type there are three regions and the
intrinsic region is ah has very high electric field because we bias this photo diode in
always in reverse bias. So, intrinsic region has very high electric field and a photon
comes in and it generates electron hole pair and that electron hole pair gives rise to the
electrical ah voltage or the current which we sense across a load resistor as a electrical
output. So, this we saw and then we discussed some of the equations which ah govern
how many photons will be absorbed in a particular distance when when we shine a light
on a material because not all the photons are absorbed on surface itself they actually
penetrate the material they go well inside and there is a kind of exponential relation of
absorption with the distance. So, this we discussed last lecture today we are going to
discuss the avalanche photo diode which is we call APD. Unlike PIN diode where the
electrical output is kind of say equal to or the ah kind of corresponding to number of
photons received by the device because one electron one photon give rise to one electron
hole pair and that give rise to one unit of ah charge outside. Unlike PIN in avalanche
photo diode if a one ah photon comes in then we have a very high electric field in the
intrinsic region. So, that the the generated electron hole pair they acquire energy from this
very high electric field and they acquire so much of energy that they that the same
electron can itself knock down one more electron hole pair. So, this is kind of a chain
reaction one electron hole pair give rise to two electron hole pair and these two electron
hole pair give rise to four electron hole pair. So, this is a chain reaction and suddenly
within ah ah fraction of ah seconds we get so many electron hole pair that a reasonable
amount of current flows in the outside load.

So, a single photon give rise to significant current. So, we have some sort of a gain in the
avalanche photo diode. So, we are going to discuss today avalanche photo diode. So, this
is avalanche photo diode or we call it APD in short and the structure and electric field
profile of this APD is something like this. We have the similar region for example, we
have this N plus here and then we have a P region here and this is intrinsic region and this
side we have P plus region and all these photo detectors are always biased in the reverse
bias mode. So, we have a external voltage which is connected in reverse bias means the
positive terminal is connected to N plus and the negative terminal is connected to P plus
through a load resistance across which we will measure the output voltage. So, this is our

166
R load this is my electrical output. So, this is the structure of APD or avalanche photo
diode and we have intrinsic region which is connected to P type this side and which is
further connected to N plus. If we plot the electric field profile of the same structure this
looks something like this. So, we have a very high electric field building in this vicinity
where N plus and P are joined together because this is a very heavily doped region.

So, the electric field will be very high in this is electric field with respect to x which is the
distance across this device and the electric field profile look something like this. We have
a peak at this point and then there is a constant field profile because intrinsic region the
potential will be dropped linearly and then we have like this. So, this is the electric field
profile if we apply a certain voltage across this APD and we see here there is a peaking of
electric field happening at this interface N plus and P and we call it avalanche region and
this is our depletion region. And the physics of working of this device is exactly same
where we have intrinsic region where photon will strike and it will knock down electron
hole pair. Now this electron hole pair will move towards the terminal where this charge
carrier see a avalanche region. Avalanche region is a short region where the electric field
is very high as we can see in this field profile because of this very high electric field these
charge carriers will start gaining energy when they are moving through this region. So,
they will acquire enough energy so that this electron or hole they can knock down one
more electron hole pair by releasing that energy to to the lattice. This generated electron
hole pair will further move according to the electric field and start gaining energy from
electric field and knock down again one more electron hole pair. So, this is a chain
reaction one photon give rise to one EH pair or electron hole pair. This one EH pair give
rise to two EH pair two gives four and four gives eighteen and this is like a geometrical
series.

So, we get at the end a very high current in the output load which is the R load. So, one
single photon gives rise to a significant current in the output load. So, we can say that
there is a gain in the device where this gain is multiplied to the number of photon
received we get let us say hundred photons, but that hundred photon give rise to one
million kind of electron hole pair. So, this is a gain in the in terms of the signal. So, this is
the structure of APD. Now, let us write down few important points. So, the APD has an
internal gain M which is obtained by having a electric field in fact, high electric field that
energizes photo generated electrons. The second point is these electrons. Ionize the other
bound electrons in the valence band upon colliding with them which is known as impact
ionization. Then these newly generated electrons and holes are also accelerated by the
high electric field and gain energy to cause further impact ionization.

This phenomena is actually called the avalanche effect and hence the name of this device
which is avalanche photo diode. This phenomena is called avalanche effect. So, we see
that this avalanche photo diode generates multiple electron hole pair even with a strike of
a single photon. It means there is a gain which we can say M times it depends on the

167
device geometry and it depends on how much reverse bias we applied, how much is the
electric field profile. So, depends on all those things this gain actually depends on those
profiles, but this APD has a internal gain and one photon gives rise to M number of
electron hole pair and accordingly the current which is proportional to the number of
charge gain. So, this is the avalanche photo diode which is second type of photo diode
which we use for these optical detection. Let us see some of the numerical problems what
for the equations we just discussed last lecture. Let us have some of the examples based
on that. So, we have example here. So, we need to find out the maximum value of the
energy band gap energy gap or we call it band gap which a semiconductor used as a
photo detector can have if we want to make for yellow light for example when we have
yellow light it means we are we know the lambda for yellow light which is 600
nanometer. So, this is the example. So, let us see what is the solution. So, here we want to
make a photo detector which is sensitive to yellow light it means it should detect when
the yellow light is coming and for the yellow light it is universal value that yellow light
has a lambda of 600 nanometer. So, when yellow light will come the photons the yellow
light carries they have a certain energy which is equal to which is equivalent to the 600
nanometer wavelength. So, we need to find out that much energy what is the energy of
those photons and if the energy of these photons are more than the band gap of the
material more than or equal to the band gap of the material then the material can absorb
these photons and generate equivalent amount of electrical signal.

So, we need to find out how much is the energy of this photon and our band gap of the
material should be equal to or equal to or lower than this the energy of these photons. So,
for the limiting case we take band gap as equal to the energy of photons. So, to calculate
energy of photons we know E the energy of photon is equal to h c upon lambda where h
is this plane constant c is the velocity of light and lambda is the wavelength of light. So,
we can put the values here h is 6.626 into 10 to power minus 34 multiplied by c which is
3 into 10 to power minus 34 and divide by lambda which is 600 nanometer. So, we need
to convert it into meter 10 to power minus 9 and the value comes out as 3.3 into 10 to
power minus 19 joule or you can express it in electron volt which is more relevant for the
band gap calculation it comes out 2.07 electron volt. So, here we see that if we use a
material which has a band gap of 2.07 electron volt it means it can absorb the photon of
yellow light and it can produce a electrical signal. So, this material which has band gap
2.07 electron volt it can act as a good photo detector for yellow light. So, this is how we
calculate the band gap of material which material should be used for which kind of light.
Let us take the second example this is one. So, a photo detector whose area is given let us
say 5 into 10 to power minus 2 centimeter per centimeter square is irradiated with yellow
light whose intensity is 2 milli watt per centimeter square.

So, now we assume that each photo each photon generates 1 electron hole pair. In short
we can write electron hole pair as PHP then we need to calculate the number of electron

168
hole pairs generated per second. So, here is example we have a photo detector whose area
is given 5 into 10 to power minus 2 centimeter square and we irradiated this area using
yellow light for which we already know the energy of the photon and intensity of this
light is 2 milli watt per centimeter square. So, when we shine this area with this much of
intensity of yellow light how many electron hole pairs are being generated per second.
So, this is what we need to find out. So, here the assumption is that each photon
generated gives rise to 1 electron hole pair means the efficiency or the quantum
efficiency of these photons are 100 percent all the photon give rise to a electron hole pair.
Here the assumption is efficiency or the quantum efficiency is 100 percent. So, this is the
assumption we are making then we can calculate the number of electron hole pair
generated per second is equal to the number of incident photon per second. So, the
number of number of incident photons per second is equal to I the photo current divided
by the charge of electron which is equal to the p node the incident power of photons
divided by h nu or h c by lambda which is the energy of a single photon. So, these two
quantities are equal because we are assuming that the quantum efficiency is 100 percent.

Now, we can put these values together where I p h over e is number of electrons per
second. So, we can calculate from this equation that p node upon h nu is equal to 2 into
10 to power minus 3 multiplied by 5 into 10 to power minus 2 what divide by h into nu
which we have calculated in the earlier example this value which is 3.3 into 10 to power
minus 19 and the value comes in 3.02 into 10 to power 14 per second. So, in this example
we see if we irradiate this photo detector with intensity of 2 milliwatt per centimeter
square with yellow light then how many number of electron hole pairs will be generated
assuming that each photon is giving rise to giving rise to electron hole pair. So, 3 into 10
to power 14 electron hole pairs will be generated in the material or in the photo detector
and this will come out as a electrical signal across the load resistance. So, these are some
of the examples which we can try based on the equations we discussed for photo diodes.
So, we discussed two photo photo detectors one is p i n photo diode and the avalanche
photo diode. Next we are going to discuss the source the optical source which can
generate the optical signal or the optical photons from the electrical signal we apply
electrical on the device and it generates the optical photons which we can coupled to the
optical fiber and optical fiber carries these photons to the long distance. So, let us discuss
the photo source. So, photo source or optical source is nothing but a light emitting diode.
So, we have LED we call it in short. So, this is a diode which is forward biased and it
gives rise to the optical photons and we call it the light emitting diode and the structure is
also very similar to a normal LED. So, we discussed LED the light emitting diode. A
light emitting diode or we call LED in short is a device which converts optical energy to
light energy or to optical energy and these devices are quite inexpensive robust and have
long life.

169
These devices can be easily modulated that means switch on and off at high speeds. The
third is this can couple enough output power over a small area to couple to fibers. So, we
have this LED or the light emitting diodes which is a device which converts electrical
energy into the optical photons the optical energy and these LEDs has certain properties.
First is these are very inexpensive in nowadays with the progress of microelectronics
these LED devices are very inexpensive. These are quite robust because these are solid
state devices. So, they are quite robust in harsh environment as well and have very long
life because there is no moving part. They can be easily modulated because the size of
these LEDs are very small and junction capacitance is also very less. So, we can
modulate or we can turn them on and off at very high speeds which is desirable to send
these optical signals and we can couple enough output power from these LEDs because
the area where the light is being emitted that region is very small. So, we can easily
couple that to optical fiber efficiently. So, that all the photons which are generated by
LED they couple to the fiber and they go inside the fiber. So, these are some
characteristic of LED as a optical source. Then we have two types of phenomena's one is
called photoluminescence and one is the electroluminescence. So, these are two different
phenomena's occur to generate these photons. Let us discuss them as well. So, the first
one is the photoluminescence in this the electron hole pairs these excess electron hole
pairs are created by photon absorption.

The photon is emitted by the combination. So, this process is called photoluminescence.
And the other phenomena is electroluminescence. Where this excitation of these excess
carriers is a result of an electric current caused by an applied electric field. And this
happens for example in LEDs. So, we see here two phenomena's one is
photoluminescence. In photoluminescence a electron hole pair combines and they get
annihilated give and releasing the extra the amount of energy. This energy is released in
terms of a photon. So, photon get generated by annihilation of electron hole pair. So, this
is photoluminescence. In electroluminescence a current is passing in the device and this
current gives rise to the this the energy is actually released by the from this current
because the current is the flow of electrons. These electrons acquire energy from the
applied electric field and this extra energy is released in terms of photon. So, the photon
generated by in this process is called electroluminescence. So, these are the two
phenomena's and electroluminescence actually happens in LEDs or the light emitting
diode. So, let us see how this happens in terms of the energy band diagram.

When a junction is made of p-type and n-type material the Fermi levels on both the sides
are not aligned. Because, the Fermi level for a system equilibrium should be uniform the
band bending takes place. So, now we have two material one is p-type and one is n-type.
And remember these material this material is not silicon material this is a direct band gap
material because as we discussed last lecture for optical full photon generation it we need
a direct band gap material. So, this is a direct band gap material for example, gallium

170
arsenide or so. So, now we have p-type material and n-type material when we join them
together in equilibrium the Fermi level should be uniform. So, but for individual material
the Fermi level is not aligned. So, when we join them together the band bending takes
place and we can draw the band diagram for this structure. We have this first side where
we have band something like this and on the other side we have like this. Let us say this
is p-type where this Fermi level is close to the valence band. So, this is our E b and this is
E c the conduction band energy and this is our Fermi level and our intrinsic Fermi level
lies somewhere in the middle this is our E i the intrinsic Fermi level. Now when we join
them together these Fermi levels actually in the in equilibrium this Fermi level should be
flat. So, the band bending takes place which looks something like this and here like this.
So, we have the band bending and this region is formed as a intrinsic region which is
depleted of charge carriers and sometimes we call it the depletion region as well because
this region is depleted of charge carriers. And in now in this we have certain holes which
are majority in p-type and we have some electrons let us say this one they are in majority
in n-type region and the energy band gap is let us say E g which is uniform in both the
sides E g is the band gap between this level and this level this is E c this is E b.

So, now we have drawn the energy band diagram of LED in equilibrium where we are
not applying any bias to this device when we are not applying any bias it means the
Fermi energy the Fermi level remains flat and because in p in the p-side Fermi level is
close to the valence band and in the n-type Fermi level is close to the conduction band. If
these level need to be aligned it means the band bending will take place that is as shown
here and when we join them together there is a depletion region formed this region is
kind of depleted of mobile carrier just like our normal diode this is what happens in our
diode the depletion region forms the same thing happening here the depletion region is
formed and the band bending whole of the band bending is taking place in the depletion
region. When the band bending is taking place it means we have a certain electric field
available in this region. So, depletion region is the region where all the electric whatever
voltage is applied that whole voltage will appear across the depletion region where the
band bending is taking place if there is no there are the bands are flat it means there is no
potential applied in that particular region. So, when we apply a voltage bias across this
device this band bending will be further high and that causes a high electric field this
electric field give rise gives the energy to the moving charge carriers.

So, this is what happens when we apply the electric field. So, the LED works in the
forward biased case. The carriers diffuse to the other side of the junction where they can
combine quickly. So, we can draw the same energy band diagram in when we apply a
voltage. This is EC this is EV the valance band and now we apply a voltage across this
device this is a forward bias voltage. So, when we apply a forward bias voltage these
electrons these electrons will start moving towards P type and these holes which are on
the P side they will start moving towards an N type this is the intense equation for the

171
depletion region. So, now we have this device and we apply a forward biased across this
device. Now, electron and hole will see a field which is pushing them towards the other
side. Now, when this electron from N side where this is in the access or the majority
carrier when it moves to the P side where this is a minority carrier and it can recombine
there because there are so many holes available. So, when this electron and holes will
recombine they will release the extra amount of energy in terms of a photon and this
photon will come out as a optical photon and this works as a LED or the light emitting
diode. So, this device now these electrons can recombine and give rise to a photon all
these recombination will give rise to a photon. So, this is how this LED works the
recombination of electrons and holes also takes place non radial non radiatively. So, we
see here that these electron and holes are recombining and emitting the extra amount of
energy as a photon, but sometime this release of the energy is non radiatively it means
this does not lead to a photon, but it is given as a kind of phonon to the lattice and the
device actually heats up instead of emitting a photon. So, this phenomena actually
reduces the output of LEDs. Hence there is no 100 percent output and some of the energy
is actually wasted in terms of heat in the device.

So, let us consider some equations for all these devices. The fraction of electrons that are
injected into the depletion layer which results in photons getting produced is called the
internal quantum efficiency or eta of LED. So, if n is the number of electrons injected
into the depletion layer every second the power output of device is given by P is equal to
eta which is the internal quantum efficiency multiplied by the number of electron injected
into the depletion layer and multiplied by the energy of single photon which is h nu. So,
this is the power of power output of this device or we can also write it some like eta I into
h nu upon E, E is the electronic charge. Now if the energy of photon is measured in
electron volt the current in milli amp the above expression becomes E the power in milli
watt is equal to eta E gamma into I. So, these are the formulas to calculate the output
power of a device when the number of electrons which are going into the depletion layer
is known eta is known the quantum internal quantum efficiency and the energy of a
single photon is known. So, let us have one example based on this. A gallium arsenide
LED radiates at 900 nanometer. If the forward current in LED is 20 milli amp calculate
the output power of the circuit. Assuming the power output assuming a quantum
efficiency of 2 percent. So, in this example is given that we have a gallium arsenide based
LED and this LED radiates at 900 nanometer. So, 900 nanometer is the wavelength of
photon it means we can calculate the energy of a single photon how much is the energy it
of a single photon it radiates. If the forward current in LED is 20 milli amp it means 20
milli amp current is flowing by this flow we can calculate how many electrons are
actually injected into depletion layer per unit time per unit per second. So, if the current is
given 20 milli amp we know number of electrons going there and the quantum internal
quantum efficiency is given as 2 percent which is a reasonable number for this internal
quantum efficiencies it is in the range of this only. So, now let us find out the power of

172
this device. The energy of the photon first we calculate in electron volt is h c upon e into
lambda this we can put down the numbers for blanks constant which is 6.62 into 10 to
power 34 into 3 into 10 to power 8 divide by 1.6 into 10 to power minus 19 multiply by 9
into 10 to power minus 7 which is the wavelength of this photon. This number comes out
to be around 1.38 electron volt. So, this is the energy of a single photon.

Now we can calculate the power output P is equal to 0.02 which is the efficiency internal
quantum efficiency multiply by the energy of a single photon which is 1.38 multiply by
the current which is 20 milli amp here. So, the power output of the device comes out to
be 0.55 milli watt. So, this is the power of this gallium arsenide LED. So, this is how we
calculate the power of any LED which is in terms of energy of single photon and how
much current it is flowing. So, today we saw one optical detector which is avalanche
photo diode then we discussed optical source which is LED and we saw the band diagram
how a LED generates the photo photons and these are the some examples to calculate
how many photons or how many how much the power of this LEDs.

So, this is all for today.

Thank you.

173
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 19
Optical Sensors: Heterojunctions and LASERs

Hello, welcome to the course Transducers for Instrumentation. Last lecture we discussed
some optical detectors just like PIN diode and avalanche photo diode and we discussed
one type of optical source which is LED the light emitting diode which is very widely
used for these optical power generation. The next important light source is the laser
which we are going to discuss today. The laser is a short form of light amplification by
stimulated emission of radiation. So, the word stimulated is a keyword here which we are
going to discuss today how the light actually emission happening in laser which is
different than the normal LED. Before discussing the lasers we need to understand a little
bit about some device physics the properties of some material. So, we have some sort of
homojunctions and heterojunctions when we make the junctions in the semiconductor
device. So we have two types of junctions homojunction means when we form a junction
and the material on both side of the junction is the same material then we call it
homojunction and when the material is different on one side of junction is different than
other side of junction then it is called heterojunction. For example when we make a
silicon based diode we say it is a p-type material and we take n-type material join them
together then a junction is formed which is called depletion region. If we see both the
sides though it is p-type and n-type on the other side the material the fundamental
material the base material is same which is silicon.

Silicon is p-type dope on one side and n-type dope on the other side, so it is called
homojunction. However, if we take two different materials and we make a junction so
one side it is let us say gallium arsenide and one side is aluminium gallium arsenide or
some different materials some composition of the material is different then we call it a
heterojunction and both these junctions have different properties. So, we are going to
discuss today what is homojunction and heterojunction and then we will discuss the
lasers which is the optical source. So, we have a homojunction both the p-type and the n-
type regions are of the same semiconductor material. So, in homojunction both materials
are same. In heterojunction the junction layers are made of different semiconductor
materials. So, if we draw energy band diagram for these heterojunctions one side the
band gap is larger for example we have let us say this kind of band where this band gap is
Eg1. On the other side the band gap is smaller this is let us say Eg2. So now we have two
different material one side is Eg1 where we have high band gap and other side we have
Eg2 where this band gap is smaller. When we join these two materials now the band need
to be continuous because these are energy bands energy distribution is continuous it

174
cannot be discrete. So if we join these bands they are going to join like this and this. So
here now we see there is a sharp change in the energy bands and if we see here we are
making some sort of a notch here or the well we call it. If we also draw the Fermi level
for this p-type and n-type because Fermi level is close to p-type here this is the Fermi
level or EF this is p-side let us say we have positive charges here and we have electrons
here. This is EC and this is EV. Now in this energy band diagram if we see carefully this
Fermi level is flat and when we are joining these two different materials together in
conduction band there is a notch this notch or the well we call it.

The tip of the lowest or the minima of this well is coming below the Fermi level the EF
this dotted blue line which is Fermi level. The minima of this conduction band or at this
point at this well is coming below the Fermi level it means this region or this notch or this
well will always be filled with the electrons because this is below the Fermi level so the
probability of finding electrons here is very high. So all these this well will be always
filled with the electrons and because these electrons are in the conduction band so this on
the top this is a conduction band. These conduction band electrons are free to move so we
always have some electrons in the device at this point this well quantum well we can call
it. These electrons are always available for the conduction. So without doing anything
without applying a bias we just added two different materials and because of the energy
band the difference in the band gap we are creating a notch a well where electrons are
freely available for conduction. So now if we apply a electric field in let us say
perpendicular to direction of this board this board these electrons will be free to move as
per the applied voltage. So this happens in case of heterojunctions because both the
materials are different the different band gaps and we get these free electrons sometime
we call it the 2D electron gas because these electrons are available there is like kind of
liquid or you can say the gas they can move freely if we apply electric field. So these
charge carriers are confined in a space away from the doping impurities. Hence improved
mobility or the high electron mobility.

So the difference between these heterojunction where we are getting these electrons this
is different process than how we create electrons or the mobile charge carriers in normal
MOSFET. The difference is in MOSFET we do the doping in the channel so that we get
these extra electron the charge carriers these are given by these impurities boron or
phosphorus depending upon p-type or n-type. But the problem in doing that is when we
put impurities the mobility of electron decreases because there is a kind of imperfection
in the lattice when we put these impurities and these charge carriers need to travel in this
channel so the mobility decreases. However in heterojunction we are not putting any
impurity here we are not doping the material at exactly at this quantum well. These
electrons are here because of this discontinue because of the difference in the band gap.
So these electrons are created by that and these are away from the impurities so they are
seeing a perfectly fine crystal and their mobility is very high and this is the basis of

175
something called HEMT or the high electron mobility transistors. These are very high
devices used for high speed applications HEMTs. So this is the phenomena that give
leads to the HEMT devices which is this two dimensional electron gas. So now we know
that gallium arsenide is a direct band gap material. Let's say we have gallium arsenide
this is a direct band gap material and it emits in infrared and for infrared typically the
wavelength is 872 900 nanometer.

This is one material and the other material is let's say gallium phosphide which we know
is around 2.2 electron volt. So here we have now two material one is gallium arsenide
which is a direct band gap material and the band gap is such that it emits in the infrared
region and another material is gallium phosphide which is a indirect band gap. So these
materials are also the mixture for example gallium arsenide material is nothing but
gallium when it is mixed with arsenic then it is gallium arsenide and which is a direct
band gap and when we mix gallium with phosphide or the phosphorous then it becomes
gallium phosphide which is a indirect band gap material and band gap is 2.2 electron volt.
Now this arsenic and phosphorous both are in the group 3 so they belong to the same
group in the periodic table. So this arsenic and phosphorous they can also be mixed
together. For example we can have something called gallium arsenide phosphide where
we mix gallium with certain amount of arsenic and certain amount of phosphorous. So
when all these three materials are mixed together we call it gallium arsenium phosphide
GAASP with certain fraction of arsenic and remaining fraction of phosphorous. Then the
effective band gap of this material which is now the newer material made up of these
three material. The effective band gap is determined by how much is the ratio of arsenic
and phosphide in this complete material. So for example we have both these arsenic and
phosphide belong to same group in periodic table it is possible to make alloy. Having
composition of gallium arsenic the component of arsenic is 1 minus x. X and
phosphorous is so we are taking x part of phosphorous and 1 minus x part of arsenic and
mixing it with gallium so that we are making a new component which is gallium arsenide
phosphide. Now the lattice constant of these gallium arsenide and gallium phosphide are
quite comparable or almost equal so we can mix them effectively.

So the lattice constants of gallium arsenide and gallium phosphide are nearly equal for x
value less than 0.5 and 0.45. Radiation in red and infrared, red and infrared range.
However if we increase the composition of phosphorous means x is greater than 0.45 and
alloy becomes indirect band gap. So this is a process of even converting a material or
making a different material and tuning the band gap of this new material. For example
gallium arsenide has a fixed band gap of that is a direct band gap and which is fixed.
Gallium phosphide is again a different material with a fixed band gap of 2.2 electron volt.
Now we can play around with the composition of phosphorous and arsenic and we can
tune this band gap depending upon how much phosphorous we have in this material and
how much arsenic we have. So depending upon that we can tune the band gap of

176
materials and as we saw for the composition of phosphorous if it is less than 45 percent
the alloy remains a band direct band gap material it emits in red and infrared range. If we
increase the amount of phosphorous more than 45 percent or x value is more than 0.45
the alloy becomes an indirect material. So we can see even how continuously this kind of
proportion changes the band diagram.

So let us say we have this graph where on the x axis we have mole fraction which is x
value. This is 0 and this is 1 and in between we have let us say this is 0.5, 0.2, 0.4, 0.6,
this is 0.8 and 1 and on the y axis we have energy band for the band gap starts from
typically let us say 1.4 to 2.6 and it is a linear graph. So now we have two graph one is
for the direct band gap which starts from let us say 1.4 and it goes like this. This is for the
direct band gap and the other graph is for indirect band gap which comes like this. This is
indirect band gap. So here we have two graphs with one is for the direct band gap, one is
for the indirect band gap. These are the values of the conduction band and this is the
difference between the value of conduction band and the top peak of the valence band. So
we see as the mole fraction is changing from 0 to 1 these both of these graphs are
changing and whatever is the minimum value that will be the band gap of the material
overall. So we see below 0.45 the red curve is below the green curve it means the
material will remain a direct band gap material for x less than 0.45 and higher than 0.45
the green curve is below the red curve it means that material will become a indirect band
gap material. So depending upon the mole fraction of how much phosphorous we add in
the mixture the band gap of material changes. So even if we reduce the mole fraction to
0.

2 we can see the band gap is actually reducing compared to the band gap at 0.45 mole
fraction. So this is how we can tune the band gap of material within a limit. So this is
how we change the material and changing the material band gap means that the light the
kind of which light it will emit that also changes. So these indirect band gap materials are
generally inefficient as light emitting diodes.

So now let us discuss the optical source which is laser. Laser is an acronym so the full
form of this is light amplification by stimulated emission of radiation. So laser is a device
and that laser is an acronym actually so the full form of laser is light amplification by
stimulated emission of radiation. So we are producing a light we are amplifying certain
photons the photon energy by stimulated emission of radiation. Stimulated is a keyword
here we are stimulating the emitted photons and they are generating further photons by
stimulation.

So we discussed earlier LEDs and photo detectors so this is the difference between the
photo detector for example if we when we discussed photo detector we had their energy
band diagram such that this is the conduction band and this is the valence band AC EV
and in photo detector we receive a photon from outside. The photon comes in h nu and it

177
excites an electron from the valence band to the conduction band. So this is what happens
in photo detectors. In LEDs or the light emitting diodes we discussed we have conduction
band and valence band AC EV and there is a hopping of electron from conduction band
to valence band and in this process this electron is actually losing energy and this energy
comes out in terms of a photon h nu. So this is a photon which is being generated by this
the hopping of this electron from conduction to valence band.

So this is kind of a reverse process happening in photo detector so this is what LED. Now
let us discuss the third case which is lasers what we have in laser is a conduction band
and valence band. In laser we have so many electrons available here to jump down the
primary requirement here is we have so many electrons here which are ready to jump to
some conduction to valence band. So when they will jump from conduction to valence
band they are going to lose energy in terms of photon. Now we have a photon which is
there which is generated elsewhere now this photon is travelling in the lattice and there
are so many more electrons available which are about to jump from conduction to
valence means they are about to generate these photons.

So our original photons actually stimulates or affects these further photons so that the
emitted photon the new photon which is generated from hopping this photon is coherent
with the originally available photon so this is what it is called stimulated emission. So we
in this process we have a photon which comes in h nu and at the same time one electron
is jumping from conduction to valence this is about to lose energy in terms of photon.
Now this photon is actually reacting with this electron such that in the process it
generates two photons now this is the newly generated photon and this is the original one
which is on the left now it generates two photons they are coherent means the all their
properties are same. So this process is called stimulated emission this is very particular to
the lasers. The main or the necessary condition for this to happen is there should be so
many electrons available in the conduction band so that when a photon comes in at the
approximately that the same time one electron is there which is jumping from conduction
to valence band.

So it means we need to have lot of electrons available in conduction band ready to jump
in it means we need to have a very high concentration of electrons in conduction band or
we can say the population inversion. So we need to have a population inversion which is
the necessary condition for lasers to work. So this is the difference between LED and
lasers. So this is the process happens in lasers. Now we saw that there is a something
called population inversion which is a necessary condition for lasing action.

So let us discuss what this population inversion is. So in a semiconductor population

inversion can be obtained by means of high carrier injection which results in electrons
and holes in the same special region. So now if we look at the band diagram for this
material we have let us say the conduction band and we have the valence band. For

178
example this is our conduction band this is the valence band. Now in normal case we
have so many electrons available in the valence band.

So this is the normal case where we have conduction band which is completely empty
and the valence band which is completely filled now. Now we inject a high current so
that many electrons has energy high enough so that they can jump to the conduction
band. So remember in the first case in the earlier case some of the electrons only have
energy to jump to conduction band and they were giving rise to conduction. Here in this
case we have a very high carrier concentration which are jumping to conduction band or
this called a population inversion where the complete layer a top layer of valence band
completely all the electrons jump to the conduction band and remains in the conduction
band. It means the conduction band is completely filled on the top however a valence
band is empty in the bottom. So if we apply a population inversion here after population
inversion the bands look something like this. This is valence band conduction band and
this layer all the electrons jump to the top and leaving behind a completely empty layer in
the valence band. So if we can see whole of this area this whole population actually have
migrated to the top which is here. So this is called the population inversion where all the
electrons in the valence band they have jumped to the conduction band leaving behind a
complete layer of empty states and all the states in the conduction band are now filled
with electrons. So we see here now we have so many electrons in the conduction band
and these conduction band electrons they can lose this extra energy anytime and they can
jump back into the valence band anytime and give rise to the photon.

So this is called the population inversion. Now the incident photon which is coming from
the left if the incident photon with energy h nu this is the energy of photon if this energy
is greater than Eg the band gap of material but it is lesser than Ec minus Ev the difference
between conduction and valence band. If the photon energy is such that is between Eg
and Ec minus Ev this cannot be So to understand this let us draw this diagram again we
have a conduction band and we have valence band and we have population inversion it
means we have many electrons in the conduction band all these states are filled and in the
valence band all the top layer is empty this whole layer is actually empty there is no
electron in the valence band the top layer of valence band this is no electron. Now when
photon is coming from the left now this photon it cannot be absorbed if the energy of this
photon is higher than Eg but lesser than Ec minus Ev because to absorb this photon a
electron need to be present here somewhere so that it can jump from the valence band to
conduction band and absorb this photon this is not possible because there is no electron
available here electrons are further down in the valence band so they cannot absorb they
need more energy compared to what energy photon has to jump to valence band. So this
photon will not be absorbed by the material however at the same time so many electrons
are there which are ready to jump to jump down to valence band emitting a photon. So in
this process this photon actually interferes with the energy emission of electrons and

179
generate to one more photon because of this hopping these both the photons are coherent
or the same in all the characteristic how they interact it to each other that is beyond the
scope of this course but this photon actually reacts or interfere with the hopping of
electron and generate another photon.

So let us say we have now this electron which is jumping down to empty state here and in
this process we are generating two photons this is h nu and this is another photon which
is h nu. When these two photons are generated again they are propagating in the same
channel now other electrons are ready to jump for example these electron this electron is
also ready to jump from top to bottom to this empty state and this also reacts with these
incoming photons so let us say these two electrons are jumping down emitting a photon
and two photons are coming from the left which interferes with this emission of photon
and at the end we get four photons of exactly same characteristics. So this process is
called amplification by stimulated emission and the necessary condition for this to occur
is the condition for stimulated emission another population inversion is the energy of
incoming photon should be greater than eg the band gap of material but it should be
lesser than the difference between ec – eb. This is the necessary condition for laser
action. So this is how the laser device works in this we have a population inversion where
all the electrons in the valence band they jump to the conduction band because of the very
high current flowing or very high and very high charge is provided to this lattice.

Now these electrons on the conduction band they are ready to jump at any time they are
continuously jumping now if a photon comes in where the photon energy is higher than
the band gap so it can be absorbed but it will not be absorbed because there is no electron
available right on top of valence band which can take this amount of energy and jump to
conduction band because there is no electron available in the valence band and there is no
empty state in the conduction band where it can jump. So this photon will not be
absorbed on in fact this photon will not be absorbed but at the same time if electron is
jumping down emitting a photon this incoming photon will react with interfere with this
release of energy and that resulting photons are identical in nature. So this one photon
give rise to two photons which is a amplification one photon becomes two and both are
same and these two photons further react with two jumping down electrons and create
four electrons which are again of the same properties and there is further amplification
from two to four. So this is called the amplification by stimulated emission because here
the photon is interacting with a jumping down electron and releasing the same kind of
photons so we call it amplification by stimulated emission. So this is how a laser device
works and the condition for stimulated emission is h nu then the energy of photon should
be higher than the band gap but it should be lower than the band gap.

So this is the difference between the conduction band and the valence band and at this
point this is Efc and let us say we call it this Efv so this is in fact Efc minus Efv. How
many electrons are jumping from top to bottom that dictates what is the value of Efc and

180
Efv. So this is the condition for stimulated emission. Next we discuss a little bit about the
wave guides we can make these wave guides where light can travel in those wave guides
on chip and light will travel through total internal reflection and these are these can be
made using micro electronics processes. So we can make something called wave guides
to carry photons from one place to other and making of this is very simple.

If we have a p-n junction this side is let us say p-type this is n side and this is the junction
or depletion region. Now we see this region which is depletion region this is this does not
have any moving charge carrier available here it means the property of this material is
different than p side and n side. So if we measure the refractive index of this material
along this direction if we measure the refractive index the refractive index will come out
like this. So this is refractive index in this region the refractive index will be different
because this material is does not have any charge particles. In this region this refractive
index is different. So if there is a change in the refractive index it means we can compare
it with the optical fiber where we have core which is of different refractive index and we
have cladding which is of different refractive index and now if we shine a light in this
structure along this direction more of the photons will be confined in the center region.
We shine a light in this structure many of these photons will be confined in this region.
These are the photons confined in this depletion region and this can act as a waveguide
all these photons will be travelling from one side to another within this structure itself. So
we can say the electron hole concentration in the depletion layer modifies the refractive
index.

Creating a waveguide which helps to confine the photons. So here we have a structure
where we have PNN type material when we join them together they form a depletion
layer and because this depletion layer is different than both of the materials because it
does not have free charged particles. So there is a slight change in the refractive index of
this depletion region compared to both the sides which is very similar to our optical fiber
where we have a core which has different refractive index and on the cladding which is
surrounding this which has different refractive index. So just like our optical fiber carries
these photons or confined all the photons in the core itself the same structure if we shine
the light here the photons will be confined in the depletion region because of this
difference in the refractive index. So today we discussed the lasers and before that we
discussed homojunction and heterojunctions and we saw for the lasing action we need a
population inversion and we saw the necessary condition for lasing action and then we
saw the waveguides where we can confine these photons instead of optical fiber we can
use waveguides which is easy to make on chip when you are making these using
microelectronics fabrication process it is easy to make waveguides on the wafer.

So, this is all for today.

Thank you.

181
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 20
Acoustic Sensors: Introduction

Hello, welcome to the course Transducers for Instrumentation. Today we will discuss
about acoustic sensors. These are special type of sensors where we use acoustic waves
and acoustic sensitive materials to make a sensor. So acoustic sensors primarily use
longitudinal waves or the mechanical waves as a media. We use the properties of these
waves and make a acoustic sensor. So let us discuss first a little bit about what are these
waves, the mechanical waves or the acoustic wave what we use in these sensors and what
are their typical properties. So, we have acoustic waves, so today we are discussing
acoustic sensors and in this we have something called mechanical waves or longitudinal
mechanical waves. So in these type of waves the oscillations occur in the direction of
wave propagation. The oscillations occur in the direction of wave propagation. So for
example we have wave which is travelling in x direction and this is the media.

So let us say time t equal to 0, this wave is at the extreme corner where we can see that
this spacing is different. So it is, this is because of this difference in the spacing compared
to the normal condition we have a perturbation here and at t equal to let us say t0 this
perturbation moves to x direction, so at time t equal to 0 this distance moves somewhere
in the middle for example here is the this perturbation. So we have a medium or this
media which is composed of atoms or so and when the wave hit from the x direction from
the left then there is a oscillation of these atoms of the media. These oscillations or this
disturbance this is in the direction of wave propagation. So wave is travelling in the x
direction and the atoms of the media they are also oscillating in the x direction only.

So at t equal to 0 this oscillation is at the extreme left and then at after sometime t equal
to t0 this movement actually shift in the x direction only. So the direction is same. So this
is what we call longitudinal mechanical waves and our sound waves what we speak so
these sound waves are type of mechanical waves only. So we have sound waves these are
created by alternate compression and expansion of solids, liquids or even gases. So sound
waves are created by alternate compression and expansion of solid, liquid or gases. And
these sound waves are longitudinal mechanical waves. So between 20 hertz and 20
kilohertz. So here we have sound waves which are created by alternate compression and
expansion within the solid, liquids and gases. So as we speak the sound actually travels
from the transmitter to the receiver because we have air which is having full of atoms.
When we speak this sound is created by us and then it moves through the movement of
these atoms in the air. So air has these atoms and when we speak in this direction for
example in towards the front so these atoms move these atoms which are present in the

182
air these gases these atoms move in the direction of wave propagation. So we have sound
waves and the typical frequency of sound wave is 20 hertz to 20 kilohertz which a human
can can hear. So these are type of longitudinal mechanical waves. So now we have some
properties of sound waves. The sound the speed of sound depends on the media to which
the waves are passing sound is often a fundamental property of material. Hence the speed
of sound changes with ambient temperature. So the speed of sound changes with ambient
conditions. There are so many ambient conditions can have one of them is temperature.
So the speed of sound changes with ambient conditions. For example speed of sound is
around 343 meter per second in air.

This is 1482 meter per second in water and it is much higher 5960 meter per second in
solid like steel. So here we have the speed of sound which is a property of sound this is
often a property of material in which this wave is traveling. So when the speed when the
sound is traveling in the air the speed is 343 meter per second. However when it is
traveling in the steel when we hit the steel on one end through longitudinal waves the
wave travel in the steel with the speed of 5960 meter per second which is much higher
than the speed in air. So these are the fundamental properties of material itself what kind
of material you choose based on that we have the speed of sound different in that
material. So these are the materials we have these materials most of the time they are not
linear they are not the response to this speed is not very linear is it most of most of the
materials we have they are non-linear in nature and when the speed is sensitive to that
non-linearity we generate some sort of harmonics in the in the waves. So we have
something called non-linear mediums. And the speed of sound is also slightly sensitive to
the sound amplitude. Which means that there are non-linear propagation effects. Such as
the harmonics the production of harmonics. And tunes. So we have a non-linear medium
if the sound is traveling in the non-linear medium the speed actually is not only
dependent on the material property but it also depends on how much is the amplitude of
the wave. If we have a very high magnitude then the speed is slightly different then the in
that case we have a smaller amplitude. So smaller amplitude may have a different speed
than compared to a higher amplitude signal and that is because of the non-linearity in the
material. So because of these non-linearities we generate some sort of harmonics. The
harmonic of a wave is a component frequency of the signal that is an integer multiple of
fundamental frequency.

We apply a frequency which is fundamental frequency to this non-linear medium then

because of the non-linearity in the material the harmonics will be generated. Harmonics
are nothing but different frequencies which are integer multiple of the fundamental
frequency. For example we apply a 10 kilohertz signal then the harmonics generated will
be 10 kilohertz and then 20 kilohertz, 30 kilohertz it is an integer multiple of 10 kilohertz,
10 kilohertz multiplied by 1 multiplied by 2 multiplied by 3 and so on. So like that the
harmonics will be generated which are integer multiple of fundamental frequency. So if

183
let us say we apply a frequency to this non-linear medium and the fundamental frequency
is let us say 440 hertz then the second harmonic will be 2F or we can call it 2 times F this
is let us say 1 time F which is integer multiple of 440. So this is 440 multiplied by 2, 880
hertz this is called second harmonic or sometime we call it the first overtone and then we
get a third harmonic which is 3 times F, 3 times of 440 means 1320 hertz. This is third
harmonic and the second overtone and if we have the another frequency which is fourth
harmonic or 4F that is 1760 or we call it the fourth harmonic.

So we apply a signal of 440 hertz here to this non-linear medium and we get the integer
multiples of the applied frequency which is 880, 1320 and 760 hertz which is we call
second harmonic, third harmonic and fourth harmonic and so on. We get multiple we get
different harmonics even to a very high value as well. The generation of these harmonics
depends on how non-linear the material is if the material is very highly non-linear then
the amplitude of these harmonics are different it is much higher and if the material is very
linear it does not have much of a non-linearity then the amplitude of these harmonics will
be lesser. So we want a linear medium so that the amount of harmonic generation is less
and we do not lose our energy of the wave in terms of the harmonic generation.
Harmonics are something which is undesirable or some sort of noise we can we can see
them as a noise. So these are the harmonics generated. Now there is one more term
associated the pitch of the sound. So the pitch of a sound is determined by the frequency
of the sound. For example a low pitch sound is thunder or gunshots like thunders.
Medium is for example telephone bell and a for example high pitches like small bells. So
the pitch of a sound which is another parameter for the sound waves the pitch of a sound
is determined by how much it is for example the low pitch sound is a sound which is
generated in the thunders or the gunshots. Medium pitch sound is telephone bell or
something and the high pitch sound is like very metallic bells when they ring this is a
high pitch sound which they produce. So these are the type of pitch we have in the sound
waves. Now we make sensors using these acoustic waves the properties of acoustic
waves and some of the applications of these acoustic waves are found very in very wide
variety. So the typical applications for example in military we have multiple applications
for example chemical substance detector. And bio substance detectors in automotive parts
and engine performance control in industrial and environment. We have water quality,
workplace monitoring. Some applications involve food industry as well where we use
them in process control. And food freshness we can also check with them. And we use
acoustic sensors even in the medical fields as well. For example in the emergency room
diagnostics and drug discovery and deployment. So these are some typical applications
where these acoustic sensors are used military, automotive industry and environment,
food industry and medical. In military also these acoustic sensors are very much used
because for example the underwater surveillance.

184
The RF waves they do not propagate into the ocean or into the water. These RF waves
cannot be used within the under the sea or under the water. So for underwater typically
we use the caustic waves and hence we use the caustic sensors for these underwater
surveillance which is a field which is very much linked to the military for the defense
applications. So these are some typical applications where we use acoustic sensors and let
us see what are some typical advantages of acoustic sensors here. So the advantages of
acoustic sensor is the first one is these acoustic wave sensors are very versatile devices.
They are competitively priced. get damaged very early, they are quite sensitive, they are
decently sensitive enough and inherently or intrinsically reliable for their operation. They
can be used as passively and wirelessly. So these are the sensors which do not require a
external bias circuitry to perform the function. So for example, a diaphragm based
acoustic sensor when the wave hit on this diaphragm, this creates a pressure difference.
So this is a passive device which does not need external bias or external source to power
it up. The pressure will be generated by this diaphragm and this pressure how do we
change that is the subsequent step, but this acoustic sensor does not need a voltage source
or any external supply. So they can be used as passively and even they can be connected
wirelessly. And some other application include these measuring force, acceleration and
displacement and flow etc. So we have for example, we have two items for example,
microphone which is acoustic sensor for air waves in the audible range. And second term
we have hydrophone which is again a acoustic sensor, but for liquid and both of these are
pressure sensors. So we have two terms one is microphone this is the acoustic sensor, but
this is for the air waves in the audible range. And second we have a hydrophone which is
used within the liquid in the underwater sensing or so. This hydrophone is the working
principle is same, but it is used for the liquid or underwater. So this is also a caustic
sensor and both of these sensors are pressure sensor they measure the pressure generated
by these acoustic waves and this is converted to electrical signal.

These microphone and hydrophones they have a moving diaphragm. And a displacement
transducer. That converts its motion into electrical signal. And these microphones and
hydrophones they have some parameters on which they differ for example, the
sensitivity, directional characteristics. Frequency bandwidth. And dynamic range. So
these microphone and hydrophone they both have a moving diaphragm. So there is a
diaphragm there in these acoustic sensors. When a wave hit this diaphragm this
diaphragm actually moves or shake depending upon the waves what kind of wave it is. So
this diaphragm responds to this wave and it moves and there is a transducer attached to it
which measures how much this diaphragm is moved and based on that we can calculate
how much what is the kind of amplitude and the frequency of the incoming acoustic
wave. So this is how a caustic wave is detected by microphone or a hydrophone. So this
signal is then converted to electrical signal and these microphone and hydrophones they
differ in terms of their sensitivity, how much sensitive a hydrophone is. Some of the
hydrophones or microphones they are directional. For example, we want to send the wave

185
in this direction and we want to receive only we want to detect only from this direction
than they have. Then we use directional hydrophones they are particular only in one or
desired direction.

Some hydrophones are omnidirectional they measure any waves coming from any
direction. So some hydrophones are directional they have directional characteristics.
Frequency bandwidth how much the frequency bandwidth they have and the dynamic
rate based on these factors these hydrophones and microphones they differ. So now these
hydrophones or microphones they generally have diaphragm one is moving and one is
generally fixed. So for example, we have for example, let us take example of electrical
system which is capacitor. So in capacitor we have two plates one is positively charged
when we apply a bias across it and one is negatively charged. Now the area is A and we
have applied a voltage which is V and the distance is D between these two plates. This is
also we can use as a acoustic sensor or a hydrophone. For example, if the first plate of
this positive plate of this capacitor actually moves then the capacitance of this capacitor
changes because the capacitance of this is C equal to epsilon naught K A upon D. So this
is the formula for the capacitance of this capacitor when a plate moves in x direction
upon application of this acoustic wave this distance changes and based on that if the
distance is decreasing the capacitance will increase.

So there is a change in the capacitance of this capacitor if this plate is moving and this
plate we can assume it is a diaphragm which is moving with respect to a fixed diaphragm
or fixed plate then if this diaphragm is moving based on that there is a capacitance change
and this capacitance we can measure using electrical circuits when we apply a voltage
across these two plates. So this is one example where we can use this as a hydrophone or
a pressure sensor. Now a capacitor microphone entirely converts the capacitance between
plates into an electrical voltage. So this capacitor actually converts the distance which is
the distance moved by this diaphragm this converts linearly into the equivalent electrical
voltage and there are some circuits which we can use to calculate how much is the
capacitance change of this capacitor. So few circuit example a microphone for example
we can use this kind of circuit where we have a resistance on the bottom we connect this
capacitor and below there is a ground there this is our output and this is the supply
voltage let us say positive V this is our capacitor which is actually moving diaphragm so
the capacitor is actually a variable capacitor so this is our microphone and when the
sound wave coming from the left so this is the sound wave which hits this capacitor from
the left there is a change in the capacitor so when the capacitor is changing its voltage
across it then this output voltage which is detected on top of this capacitor this voltage
output also changes.

So if we assume the incoming wave is sinusoidal so at the output node V out we get a
electrical signal which is sinusoid in nature. So this is a simple example of a circuit where
this resistance is let us say R or we can have a more improved circuit which is using

186
active devices for example we have a MOSFET and on top we have a resistor this is out
and the capacitor is connected here this is a variable capacitor the voltage applies plus V
as a bias this resistance is let us say R this is MOSFET M. So in this circuit if a acoustic
wave hits from the left the voltage across this capacitor changes and this is this voltage
change which is happening because of this diaphragm movement this voltage is applied
at the gate of MOSFET. Now this work as a amplifier so here we have a MOSFET which
is connected on to the resistance on top with a supply bias this active device act as a
amplifier. So we have a small signal generated by the diaphragm which is now boosted
by this MOSFET by this amplifier and at the output node V out we get a much amplified
version of signal which is generated by diaphragm. So this is a active device or the
amplified version of the input signal. So here we have a non amplified version and here
we have a amplified signal. So these are some circuits which we can use pressure sensor
working and how to detect the output of a pressure sensor. So now to discuss acoustic
sensors how do we make these acoustic sensors we need to discuss first there is a special
term which is called piezoelectricity or the piezoelectric effect. So there is a effect in
some material that if we compress those material or expand those materials means we
change the shape of those materials so they develop a electric potential across it.

So this is the fundamental properties of some materials not all the materials have this
kind of property. If we have a piezoelectric material which shows this kind of property
we take a crystal of this piezoelectric material and we compress it then it generates a
voltage across it plus voltage on the top and minus let us say on the bottom and if we
expand this material this polarity of this voltage reverses on the top it comes negative and
on the top on the bottom it comes positive. So this effect this phenomena is called
piezoelectricity. So acoustic devices or acoustic sensors use piezoelectric material to
generate acoustic waves. And this piezoelectricity refers to the production of electrical
charges by imposition of mechanical stress. So this piezoelectricity refers to the
production of electrical charges in a material if we apply a mechanical stress on this
material. We compress it or expand it means we are changing the structure of this
material accordingly the voltage will be developed on the one side of on the one and on
the top and the bottom side of this material. So this is called piezoelectricity and in a
diagram if we show it we have let us say a piezoelectric material and we apply a force
from top and bottom. So we apply a force this side then it develops electrical charges
across it for example positive is coming this way and negative is this way. So we get a
net potential difference which we can measure using some multimeter or so a measuring
instrument if it has a if we connect this multimeter to this side. So we have a material
which is piezoelectric. So here we have this piezoelectric material and we are applying a
force from the top and the bottom. So this piezoelectric material is generating some
charges positive on the right side and negative on the left side. So this charge we can
measure using some measurement device for example multimeter or some high end
equipment. So this piece this property is shown by certain materials only which are called

187
piezoelectric material and this effect is called piezoelectricity. So this is the effect which
is very important because this effect is used in acoustic sensors to generate acoustic
waves.

So this is all for today.

Thank you.

188
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 21
Acoustic Sensors: Piezoelectric Materials

Hello, welcome to the course transducers for instrumentation. Last lecture we discussed
about acoustic sensors these are the sensors which use the acoustic waves for
measurements and acoustic wave as we know these are mechanical waves and there is a
physical motion involved of the media, the media is composed of let us say the atoms and
there is a physical motion of these atoms which carry this kind of waves which are
acoustic waves. In that we saw some practical applications where all these acoustic
sensors are useful and then we started discussing about the piezoelectricity and the
piezoelectric effect. So, in piezoelectric effect what we discussed last lecture that there
are certain materials if you apply a mechanical stress across those materials they develop
a potential difference across it and this develop potential is proportional to how much
stress we apply. So, here in this slide we see there is a piezoelectric material and we are
applying the force from top and bottom and this piezoelectric material is developing
some electric charge which we can measure through some measuring instruments let us
say multimeter or some voltmeter. So, this property is called piezoelectric effect and the
materials are called piezoelectric materials.

These materials are very much useful for making these acoustic devices or acoustic
sensors and we use this property primarily in this kind of devices. So, this we discussed
last lecture. Now let us see how this property is actually shown by this piezoelectric
material how a mechanical stress give rise to a electric property which is potential. So,
we have let us say we consider silicon oxide which is a piezoelectric material and the
silicon oxide material if we see the configuration of the crystal is something like this.

Let us say we have a crystal in uniform state when we are not applying any force. So, at
that point we have silicon and oxygen atoms in this structure. This is let us say silicon
and we have oxygen atom. This is bounded to another silicon atom and we have another
oxygen atom and we have again an another silicon atom. So, this is let us say the
structure of silicon oxide crystal and this silicon is let us say positively charged and
oxygen has negatively charged and this is how they are connected. Silicon is positively
charged and oxygen is negative and they are giving rise to silicon oxide crystal. So, this is
the state when we do not apply any mechanical stress means no mechanical stress. Now
at this material we apply a mechanical stress for example, we apply a compressive stress
we are applying a force which is compressing this crystal lattice from both the sides. Let
us say from left and right we are applying some force. So, the crystal will become
something like in x direction it will squeeze and in y direction it will kind of come out it

189
will expand. So, silicon atoms which are positively charged they will go up and negative
oxygen atoms they will kind of compress. So, this will be a typical structure if we apply a
physical stress in x direction. So, here we see if we look at this complete material the
positive silicon atom this has come out in y direction relatively what it was earlier with
no mechanical stress and the negative which is this oxygen atom on the bottom this has
come down in y direction compared to what it was earlier. So, there is a net potential
build up by in this inside this material the positive has gone in the positive y direction and
negative oxygen atoms has gone in the negative y direction. So, if we see the effectively
there is a net charge positive charge build up in the top portion of this material and
negative charge build up in then in the bottom in the lower portion of this material.

So, effectively a potential positive potential will build up in positive x direction positive y
direction and in negative charge will build up in negative y direction. So, this is the case
when compressive mechanical stress is applied or we can say the positive mechanical
stress is compressing. So, when we are compressing this material there is a positive
charge on the top and negative charge on the bottom. Now let us see the reverse case
when instead of compressing the material we are trying to expand the material we are
applying a expanding force from both the sides. So, what happens to the charge? So, this
is case C where the force is expanding this material in x direction. So, what will happen
the positive charge will go inside which is silicon atom and negative charge will be
dominating compared to the normal case. So, this is what happen when we apply a
expanded force and if we see this configuration compared to the case A where no
mechanical stress is applied relatively the upper portion of this material is relatively
negatively charged because earlier in the neutral case positive was on the very top and
negative was compensating that positive charge and there was no charge because of that.
When we apply this expanding force this positive comes down and negative both the
atoms goes up. So, it means there is a net negative charge build up on the top of the
material. So, now we have a negative charge build up overall and positive on the bottom.
So, if we see now in case A when there was no mechanical stress applied there was no
overall charge on the material when we apply mechanical compressive strength stress on
case B top side is positively charged and when we apply a case in case C a expanding
force the top layer is negatively charged. So, that depending upon its compressive or
expanded force the top layer is changing its polarity from positive to negative. So, it is
changing the polarity of voltage build up. So, this is case C where the mechanical stress is
applied which is expanded. So, this is the process happens in the materials which shows
this piezoelectric property based on the applied force they develop this electric potential.
So, next these piezoelectric devices let us write down their properties these piezoelectric
devices can operate at fairly high frequencies. So, let us write down their properties.
These piezoelectric devices can operate at quite high frequencies and these are often used
in ultrasonic sensors. These piezoelectric microphones can be used as piezoelectric
actuators. And, usually these the same device actually can be used in either mode.

190
So, these piezoelectric materials these materials can operate at fairly high frequencies. Of
course, not in microwave regime or so, but they can operate at fairly high frequencies.
Often these are used in ultrasonic sensors in making these ultrasonic kind of sensors even
in underwater monitoring. These piezoelectric microphones or these piezoelectric devices
they can be used in either mode you can use that as a receiver as well as a transmitter. So,
this is because the piezoelectric property is reversible means when we are applying a
physical stress on the material it develops voltage or in reverse case if we apply a time
changing electric field across that across this material it can generate this mechanical
stress or mechanical expansion and contraction. So, this property is reversible we can
either apply electrical signal and generate a mechanical wave or we can apply a
mechanical wave on these materials and they develop a electric potential across it. So,
this property is reversible and some of the materials which show this property one we
saw is silicon oxide or we call it quartz crystal. The other material is lithium niobate
which is very common material for these acoustic sensors and the other one is lithium
tantalate that is also used as a caustic material or piezoelectric material. So, the common
materials used for acoustic sensors are the first one is quartz. Or we call it silicon oxide.
The second one is lithium niobate O3 and the other material is lithium tantalate. So, these
are few materials which we can use as a piezoelectric material to fabricate acoustic
sensor. Let us discuss a little bit about the acoustic waves which we use generally for
these sensors. These acoustic waves are described by the mood of their propagation.
Acoustic wave devices are described by the mood of wave propagation.

Through the piezoelectric substrate. And the first one to this is longitudinal waves. So,
these acoustic wave devices actually we characterize them in terms of mood in what
mode this wave is propagating through those materials and the first one to consider the
mode of wave is the first is longitudinal waves. This is the wave this is the longitudinal
waves are the waves when a wave is travelling in a material and the atoms of the material
are oscillating in the direction of wave propagation. So, that is a longitudinal wave.

In the longitudinal direction. Or the direction of wave propagation. So, let us say the
wave is travelling in the x direction. And the atoms of the material they are oscillating in
the same direction. So, these atoms they are actually oscillating in the same direction. So,
when this is happening these atoms sometime they come close to each other and
sometime they go far away than each other it means there is a compression and expansion
happening in the material because of that this wave travels the atoms are actually
transferring this energy from one atom to other and this is how this wave propagates in
the material. This can be generated in solid as well as liquid and even gaseous. So, these
in this kind of wave the particle density fluctuates and they can be generated in solids as
well as liquid. So, these are longitudinal waves. The next is transverse or shear waves.

These are the waves when the wave travel in x direction the atoms of the material they do
not oscillate in x direction, but they oscillate in perpendicular to the wave direction. So,

191
the example of this wave is when we throw a stone in the water. So, the wave generated
in between and it expands towards all the direction in that case the water particles they
are not moving in the direction of wave the direction of wave is tangentially outward
from the center of impact and the atoms are moving in perpendicular direction up and
down which is perpendicular to wave direction. So, in this transverse or shear wave the
particle oscillates at a right angle or transverse to the direction of propagation. So, in
transverse wave this motion is happening in the perpendicular direction compared to the
wave motion the wave direction.

This kind of mode travelling of these waves they require an acoustically solid material
compared to the other cases. So, this is not very effectively done in case of for example,
liquids or gases it is not very effective way of wave propagation. This mode requires an
acoustically solid material. So, in transverse waves we have a direction of wave
propagation and the atoms oscillate perpendicular to this motion. So, this is how this
wave actually travels. The third one which we are going to discuss now is the surface
waves. The surface wave is something which is different than both the two in
longitudinal we have motion in direction of wave propagation, in other we have a motion
perpendicular to the wave propagation. In this surface wave this motion is neither in the
direction and nor a completely tangential or 90 degree to the wave propagation. This is
some sort of a elliptical motion when these atoms are not even in the direction not
completely in the right angle they are the motion is in a certain angle and they oscillate at
certain angle compared to the wave propagation. So, that is a shear wave or the surface
wave we call it. So, the third one is the surface waves. This surface wave combine both
the longitudinal and transverse motion to create a elliptical kind of orbit. These surface
waves combine both a longitudinal and transverse wave to create an elliptical orbit
motion. So, in this case it is neither a longitudinal wave nor a transverse wave this is
somewhere in between and combination of this longitudinal and transverse is such that it
takes a elliptical sort of motion. So, this is elliptical orbit and the major axis of this ellipse
is perpendicular to the surface of the solid.

So, this is a solid so the major axis of this ellipse is perpendicular to this surface. The
major axis of the ellipse is perpendicular to the surface of the solid. And these surface
waves are generated. When a longitudinal wave generates a surface near the something
we call it second critical angle. And they travel at a velocity between 0.87 and 0. 0.87 and
0.95 depending upon the material of a shear wave. So, here we have the surface waves
which are neither longitudinal nor traverse they are somewhere in between and the
motion of these particles are elliptical and the major axis of this ellipse is perpendicular
to the surface. Now how these surface waves are generated? So, these are generated when
a longitudinal wave hits the surface of piezoelectric material at second critical angle. So,
this second critical angle we will come to know in the next slide what is a second critical
angle. But when this longitudinal wave hits the piezoelectric material at second critical

192
angle then these surface waves are generated and their velocity is somewhere in between
0.87 and 0.95 of the shear wave they are little bit slower than the shear wave. So, when so
this surface wave if we draw the cartoon for this let us say this is the wave propagation
and now the motion of atoms are neither in the direction nor perpendicular it is some kind
of ellipse. This kind of motion is kind of rotating motion with a it is not a complete circle
it is a ellipse kind of motion. This is how these surface waves are formed and this is how
they propagate. These surface waves they travel on the surface of the material.

So, we have a material piezoelectric material and these are the waves which travel on the
surface of this material and these surface waves are very sensitive to any surface defects.
For example we have a material and on the surface there is some abnormality or
something other some other material is there these surface waves are very much sensitive
to that impurity. So, if when they travel they are generally reflected or their properties
changes based on what is there on the surface. So, these surface waves are very sensitive
to surface defects and this is one of the application of these surface waves that if we want
to check the purity of the surface whether it is very clean or there is no impurity on that
we can use this surface wave we can send this surface wave and if there is a change in the
characteristic we know that there is some impurity or something is there on the surface.
Moreover these surface waves can even bend along the material let us say the material is
not a straight slab it is a kind of a angle in that.

So, these surface waves they can even bend in the as per the material shape. So, in that
way also these are very useful for checking the surface defects. Therefore, these waves
are useful because they are sensitive to surface defects. So, these are very sensitive to
surface defects. So, next we just considered that these surface waves are generated by
longitudinal waves only and in a special case when this longitudinal wave hit the surface
at second critical angle. Now let us see what the second critical angle is. So, we know
that when a wave hits a material then we have a boundary between the first material and
the second material let us say in this case first material is air or something the second
material is this piezoelectric material and this longitudinal wave is hitting the interface
between material 1 and material 2. So, when it hits the surface there will be some part
which will be reflected back in the same material and some part will be refracted into the
other material just like our optical wave propagation. Longitudinal waves when they hit
the another material because of the refractive index difference some part is reflected
some part is refracted and that process is governed by Snell's law in optical. So, the same
thing happens here as well when this longitudinal waves hits the surface some part is
reflected and some part is refracted and this angle of reflection and refraction that is
determined by Snell's law.

So, let us draw that Snell's law first. So, what is a second critical angle let us say this
what we are defining. So, in normal case when a longitudinal wave is hitting this surface
this is the perpendicular direction and longitudinal wave is hitting here and the velocity is

193
let us say VL1, V is the velocity, L is for the longitudinal waves, L1 is the medium 1 this
is medium 1 let us say and this is medium 2. So, VL1 is the velocity in medium 1 of
longitudinal wave this is hitting the surface at an angle theta 1. So, some part will be
reflected back that will be at theta 1 itself VL1 let us say dash this angle will be theta 1
and some part will be going to the material 2 which is refracted part and that is let us say
here. This angle will be theta 2 and the velocity in this medium will be VL2 velocity of
longitudinal wave in material 2. And we can write the Snell's law as sin theta 1 upon VL1
is equal to sin theta 2 upon VL2. So, we can write that VL1 is longitudinal wave velocity
in material 1 and VL2 is longitudinal wave velocity in material 2. So, this is the case
when a longitudinal wave is hitting on the surface of piezoelectric material some part is
reflected and some part is refractive and the Snell's law is given by sin theta 1 theta is the
angle of incidence sin theta 1 upon VL1 is equal to sin theta 2 upon VL2. So, this is a
normal case of like optics but what happens in case of which is different than these for
these acoustic materials when a longitudinal wave hits the surface when it goes into the
piezoelectric material it does not only generate VL2 the longitudinal wave but it also
generates a shear it a different wave which is at a different angle than theta 2 and one
longitudinal wave generate two waves inside this piezoelectric material. So, the actual
case what happens which is different than optics is we can draw that here we get two
waves in the material 2.

So, when a longitudinal wave generates the interface form of the energy can cause
particle movement in transverse direction. That means we have interface let us say this is
the interface this is material 1 this is material 2 and the wave is hitting at this point which
is longitudinal wave VL1 at angle theta 1 some part will be VL1 the longitudinal wave at
theta 2 and some part some energy will be going into a different kind of wave which is
shear wave here at an angle theta 4. So, now the Snell's law applies even in this case as
well and the velocity for example, so this wave is Vs2 velocity of shear wave in material
2 and we can write the Snell's law here sin theta 1 upon VL1 is equal to sin theta 2 upon
VL2 equal to sin theta 3 upon Vs1. So, some part will be reflected as well this is VL1
dash and Vs1 this angle is theta 3 is equal to sin theta 4 upon Vs4. So, here what we have
when a longitudinal wave is hitting the surface of this piezoelectric material some part of
this energy goes into the shear wave and some part will be longitudinal wave. Now we
have two waves in material 2 which are a different angle theta 2 and theta 4 and theta 2 is
generally higher than theta 4. Now in optics what we call a critical angle that angle is
when the diffractive wave is at 90 degree for example, this longitudinal wave is coming
at theta 1. Now we start increasing the theta 1 it means the theta 2 and theta 4 both will
start increasing and at some point this theta 1 is such that the theta 2 becomes 90 degree
means the refracted in the material the refracted longitudinal wave reaches on the surface
is at a 90 degree then that perpendicular. So, that value of theta 1 is called the critical
angle normally, but when this theta 2 becomes 90 degree theta 4 is still lesser than theta 2
that shear wave is still going inside the material this longitudinal wave has come on the

194
surface, but this shear wave is still going in deep into the material because theta 4 is still
smaller than 90 degree. So, we do not stop this theta 1 just at that critical angle we further
move it we further increase it.

So, that this theta 4 also becomes 90 degree when this theta 4 also becomes 90 degree
that angle that theta 1 is called now second critical angle and this is the angle when we
get this surface wave generated on the material. So, that value of theta 1 where this shear
wave this theta 4 also becomes 90 degree that angle that theta 1 is called the second
critical angle and this is the angle when we hit the surface by longitudinal wave a surface
wave is generated on the material and this surface wave will now travel on the surface of
this piezoelectric material. So, this value is called second critical angle. Let us write
down the value values what we used here. V s 1 is the shear wave velocity in material 1
and V s 2 is the shear wave velocity in material 2. Now, let us write down what we just
discussed the first critical angle. So, the angle of refraction. So, the longitudinal wave is
90 degree and all of the energy from the refracted longitudinal wave is now converted to
surface. So, this is the first critical angle. Following longitudinal wave and there is also
an incident angle that makes the angle of reflection of the shear wave 90 degree. This is
called second critical angle. We can write here second critical angle. This is called second
critical angle and beyond this second critical angle only the surface wave will be
generated. So, we saw that there is a first critical angle when longitudinal wave hits the
material at first critical angle longitudinal wave the refracted one it comes on the surface
and if we further increase this theta 1 beyond this first critical angle then there is a angle
at which the shear wave also comes on the surface and that gives rise to the surface wave
which travels on the surface of this material. So, it is important to hit the material at theta
1 which is at the second critical angle or more. So, that we can generate the surface wave
and this surface wave will travel on piezoelectric material and the surface this will travel
on the surface itself and it is very sensitive to any surface defects or if there is a another
material which is put on the surface this surface wave is very much sensitive to that
material.

So, this is all for today.

Thank you.

195
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 22
Acoustic Sensors: Surface Acoustic Wave (SAW) type

Hello, welcome to the course Transducers for Instrumentation. Last lectures we were
discussing about the acoustic waves, how the acoustic wave travels, what are their
general properties and the modes of their travel, how they travel in the material or on the
surface of the material. So, today we are going to discuss about the sensors, how do we
make sensors out of these properties of materials and we will be using these piezoelectric
materials for making these acoustic sensors. So, depending upon the mode of propagation
of these waves, these acoustic waves we can classify these sensors into two types, one is
the surface acoustic wave sensor or we call it SAW sensor as say WSAW and the second
is the BAW sensor or bulk acoustic wave sensor where these waves travel within the bulk
of the material. So, in surface acoustic wave the wave acoustic wave travels actually on
the surface of these piezoelectric material and in BAW sensor the wave travels in the
material from one end to another into the bulk of this piezoelectric material and we call it
BAW sensor. So, we have two types of sensors for these acoustic sensors. These acoustic
wave devices are described by the mode of wave propagation. True or on a piezoelectric
substrate. And as we saw we have two types of sensors, one is the surface acoustic wave
and the other is the surface acoustic wave. The surface of the wave propagates on the sub
surface of the substrate on the surface of the substrate it is known as SAW sensor.

And the second is BAW sensor or bulk acoustic wave sensor. If the wave propagates
through the substrate. It is known as BAW sensor. So, we have two types of acoustic
sensors, one is SAW sensor where the wave travels on the top of substrate it does not go
beneath the material and the second is BAW sensor where this wave actually travels into
the substrate from one end to another. So, let us discuss SAW sensor first. So, SAW
sensor is we take a piezoelectric material which is must for these acoustic device and we
generate the acoustic wave on surface of this piezoelectric material and this wave will
travel from one end to another on this piezoelectric material and in between whatever is
the sensing zone is that sensing zone is in between the transmitter and the receiver. So,
transmitter is the one which generate these acoustic wave from some electrical wave and
this acoustic wave travels on that this piezoelectric material in between there is a sensing
happening which is called the sensing zone and then at the receiver end we have some
structure which convert this acoustic wave into an electrical signal back. So, that we can
measure this electrical response which is generated by the sensing zone. So, let us
consider this circumnus acoustic wave sensors first. So, first we need to generate this
acoustic wave and that we generate this acoustic wave and that we generate using some

196
structures called IDT or inter digital transducers. We call them IDT in short. So, this
consists of two interlocking comb type shapes. They are called metallic coatings. They
are applied to a piezoelectric substrate. These IDTs are used to convert electrical signal
into acoustic waves.

They are used to convert electrical signal into surface acoustic waves. And we can by
changing the length width position of these IDTs. So, this is the position and thickness of
the IDTs. The performance of sensor can be optimized. So, in saw sensors the surface
acoustic wave sensor we need to generate a acoustic wave on the surface of the material.
So, how to generate this surface wave? We put a IDT structure which is inter digital
transducer. This IDT structure is nothing but a comb like structure. The comb like we
have multiple fins from one side and the multiple fins from other side and they are kind
of interlocked to each other without touching the different electrodes. So, this consists of
interlocking comb type structure which is called IDT structure and we apply a electrical
signal to this IDT structure and this structure generates the surface wave on the surface of
piezoelectric material. So, this has two electrodes one from the left one from the right and
they are interlocked and we generate we apply the electrical signal on these two
electrodes and accordingly surface wave will be generated.

The performance of this surface wave or the sensor in general this can be optimized by
adjusting the length, width and the position of this IDT structure. So, let us see how this
IDT structure look like. So, we have a IDT structure which is placed on top of
piezoelectric substrate and generally this is photolithographically printed. So, let us say
we use gold as a material. This is one comb structure and the second one is in between
this interlocked. So, this another terminal is. Something like this. So, we can even put
some color here. This is first terminal. So, this is the second terminal. This is terminal
number one. And this is terminal number two. So, this is terminal number three. So, this
is the second terminal. So, this is the IDT structure. We have comb like structures from
both the sides one from the top one from the bottom and they are like interlicked without
touching each other. These two act like a electrical poles and we apply the electrical
signal which is let us say sinusoid wave across these two terminals. Now the spacing of
this IDT structure determines for which frequency this IDT structure is the optimized and
convert to acoustic wave effectively. This spacing is the spacing between these two IDT
structures as well as the spacing in between these two IDT structure as well as the
thickness of each and every fin. So let us say we have this distance as in terms of lambda.

So this distance is lambda by four which is the thickness of this fin and the spacing
between these fins for example this and this. This is the spacing between these two fins
overall. This is also lambda by four and the overall from this point to this point is lambda.
So depending upon the thickness of the fin and the spacing between these two fins which
we can control as lambda by four. Lambda is the wavelength of applied electrical signal
which will be converted into acoustic wave. So the frequency of acoustic wave will be

197
the same as the frequency applied as a electrical signal. The electrical signal is let us say
coming at one kilohertz accordingly there will be a wavelength. So we design this IDT
for that particular one- kilohertz electrical signal and we choose this lambda accordingly.
Based on this lambda by four spacings of this IDT structure this is effective for one
kilohertz signal if lambda is accordingly for one kilohertz. So for different different
frequency we need to optimize this IDT structure to get maximum conversion of
electrical to acoustic wave.

So this is one IDT structure and this is also reciprocal in the work because if we apply an
electrical signal between electrode one and electrode two, this IDT will generate acoustic
wave on the surface. On the far end, when we want to convert this acoustic back to
electrical The same IDT structure can be used. So when the acoustic wave will hit this
structure a potential will be developed between electrode one and electrode two this will
be measured using electrical equipments this voltage is the output of IDT structure. So
this IDT structure the same structure can convert electrical to acoustic wave and acoustic
wave back to electrical. So to make a saw sensor we need to use two units of this IDT
structure on top of a piezoelectric material. So practically we have a surface acoustic
wave sensor like this. So we have a piezoelectric substrate this is our piezoelectric
material. On top of this piezoelectric substrate we have this IDT structure. This is port
one where we are applying electrical signal and then this port one will generate acoustic
waves those waves will travel in this direction and when they reach at the far end we
have the same IDT structure placed here to convert it back into electrical signal.

This is port two. And this center zone is our sensing zone. So here in this we see we have
a piezoelectric substrate on port one we put a IDT structure and port two also is the same
IDT structure with same width and length and all the configuration. And when we apply
electrical signal on port one there will be a surface acoustic wave generated by this IDT
structure which will propagate in both the directions. So on the left we see a boundary so
this is not going into that direction on the right side this wave will propagate. In between
we have a sensing zone where we will apply different material or we do sensing for
example we want to check what material it is so we put that material on this sensing zone
accordingly the characteristic of surface wave changes. And this surface acoustic wave
we sense it using port two which is again the same IDT structure and it will generate
acoustic wave back to electrical wave. This acoustic wave then this electrical wave
generated by port two will be compared to the electrical wave we applied on port one and
difference will give us what was there in the sensing zone. So this is how these surface
acoustic wave or the surface acoustic wave sensors, saw sensor works. So these saw
devices are manufactured using the same photolithography process what we use for
integrated circuits. So these are the acoustic wave devices are manufactured using the
same photolithography process that integrated circuits use.

198
The only difference is there is no junction actually exists in acoustic wave devices. The
only difference is that no junction exists or more precisely no semiconductor junction
exists in acoustic wave devices. And typically aluminum is deposited in pattern to make
these IDT arrays. So these acoustic wave we manufacture them using the same process
what we use for our IC fabrication the silicon based semiconductor fabrication process
the same process we use. The only difference in case of acoustic wave devices is there is
no semiconductor junction actually we need to make.

We just need to make we just need to deposit a layer of metal on piezoelectric substrate
and we need to pattern it photolithographically. We need not to go for the ion
implantation or diffusion or other kind of process to make junctions because we do not
need a junction. We just simply need IDT structure so we first deposit the metal on
surface of piezoelectric material and then photolithographically we pattern the IDT
structure on this metal using the masks and the dimensions of these IDT structures are not
very small as compared to integrated circuits what we use in those IC fabrication. So a
kind of crude mask will also be good enough to pattern these IDT structures. So let us see
how a fabrication process or the manufacturing of these acoustic devices looks like. So
first, we start with the substrate. So we have a substrate which is our piezoelectric
substrate and on top of this we deposit a layer of metal let's say aluminum or gold or
anything. This is a thin layer of metal. This is metal frame. And in the second step we use
photolithography.

So this is our substrate. On top of this we have metal. On top of this metal we put
something like photoresist which is a light sensitive material. So this is photoresist. This
is the metal and this is substrate. Now we use mask which we use in integrated circuits.

So we have a mask here. The certain portion of this mask is opaque and certain portion is
kind of transparent. So we have this mask. The IDT structure is patterned on this mask.
This is the pattern which is put on this mask. This depends on what design you have. So
when you put a mask on top of it and you shine a UV light on the top. This is UV light.
The light will pass when this mask is transparent leaving a pattern on the photoresist and
photoresist will harden and accordingly the pattern will come on the substrate on the
photoresist and the metal. So when we use this mask we are transferring the pattern on
the mask to the photoresist and photoresist will harden where the pattern is formed. Then
we use chemicals to remove this photoresist and the same pattern will be transferred to
the metal. So in part C as a step C we have this pattern transform to onto the photoresist.
This is our substrate and then we have this metal. And on top of this photoresist is
patterned now as per our mask. This is substrate and this is photoresist.

This is step C. Now we use chemical to wash away the extra metal which is exposed now
to the outside environment and some of the metal which is shielded by this photoresist
that metal will not go away by the chemical treatment. So when we do this chemical

199
treatment now our pattern is transferred on the metal. So in this step D we have this
substrate and our metal is patterned now. This is the metal or IDT structure. So as we can
see we start with substrate put the metal on top of it by deposition. Then we use
photolithography using the mask we transfer the pattern on the mask to the photoresist
and that photoresist will be hardened in that area as well. Then we do the chemical
treatment we do etching of this photoresist which is not hardened that will go away and
then we etch the metal which is now the portion of that metal which is exposed to
environment that will be washed away. Then we remove the portion of metal which is
isolated using this photoresist that will remain there and later on we remove this hardened
photoresist as well. So at the last we get a substrate where on top of it only the metal is
there which is already patterned so we get this IDT structure.

Now let us see how this IDT structure works. So we have this IDT structure one side.
This is port one. We sometime call it short device and we connect electrical signal to this.
This IDT will generate acoustic waves and on that far side as well we have the same IDT
structure. Yeah, and in the middle, we have this sensing zone where we will sense. Now
this surface acoustic wave will be generated using this first IDT. This electrical signal
which is here shown by this red element. This electrical signal is applied on this short
device and acoustic wave will generate. What acoustic wave is? The atoms of this
piezoelectric material which is the substrate here those atoms will oscillate. So we will
have let us say these are the atoms. These atoms when acoustic wave is applied here.
These atoms will oscillate back and forth up and down and will propagate this acoustic
wave from one side to another. So this is the sensing zone where this acoustic wave is
travelling and at the receiving end this port two this acoustic wave will be converted back
into electrical signal which we measure using some volt meter or we can apply a load
across it. It can be a volt meter or some electrical instrument to measure this output
voltage. So this is a typical saw sensor with bias circuit.

Some points about this saw devices are the displacements. They decay exponentially
away from the surface so that most of the wave energy almost 90 or 95% is confined
within a depth equal to one wavelength. The width of the electrodes usually equal to the
width of IDT gaps. The breadth of inter electrode gaps. Giving the maximum conversion
of maximum conversion of electrical to mechanical energy. So, here we have this saw
sensor IDT1 converts into the electrical signal into acoustic wave. Accordingly,
according to this acoustic wave the atoms of isoelectric material they oscillate top and
down and this happens in the sensing zone if there is any other material in this sensing
zone this motion of these atoms actually changes compared to the we do not have that
material which actually changes the velocity of acoustic waves. So accordingly port two
will generate electrical signal which is lagging or leading behind compared to the original
signal and this change we can detect using this saw device. Some property of this saw
device is this surface wave which is travelling on top most of the energy is confined only

200
on the surface and this decays exponentially within one wavelength of the signal. So
almost 95% of the energy remains on the surface of the material.

So this is saw device and usually when we design these IDT structures the gap between
the two comb-like structures the gap between them is equal to generally the width of the
electrode so that the conversion efficiency is much higher. If we keep that gap equal to
the width of electrode which is lambda by 4 as we saw last slide. If we keep them lambda
by 4 lambda by 4 so the conversion efficiency of the electrical signal to acoustic wave is
much higher for a IDT structure.

So, that is all for today.

Thank you.

201
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 23
Acoustic Sensors: Bulk Acoustic Wave (BAW) type

Hello, welcome to the course Transducers for Instrumentation. Last lecture we discussed
about the acoustic sensors and we saw there are two types of acoustic sensors, one is the
surface acoustic wave sensor where the acoustic wave travels on the surface of
piezoelectric material and the second is the bulk acoustic wave sensor where the wave
actually goes within the material, the piezoelectric material from one end to another end.
And we discussed a configuration how do we fabricate a saw sensor, there are some
photolithographic steps and then we saw there are two IDT structures which we pattern
on a piezoelectric material, there is no junction formation, so we do not need any
diffusion or ion implantation step. We just need to pattern two IDT structures on top of
piezoelectric material. Now IDT structures are same, so these are reciprocal they can
even convert a electrical signal to acoustic wave and acoustic wave back to electrical
signal at far end or at the receiving end. So these IDT structures which are generating this
signal from electrical to acoustic wave, this acoustic wave is generated perpendicular to
the fringe of this com structure or IDT structure.

So we have IDT structure where pins are in perpendicular direction, so the acoustic wave
will be generated in horizontal direction perpendicular to this fringe . So now this
acoustic wave this will be generated at IDT structure and it will propagate in X plus X
direction and in negative X direction as well. So this acoustic wave which is generated by
this IDT structure this will propagate to the right and it will propagate to the left itself if
there is a piezoelectric material there. So if we do not have a piezoelectric material on left
side the wave will propagate to right side or the positive X direction.

Now earlier we saw we have a sensor where we have one IDT on the left and this is
generating acoustic wave, acoustic wave is travelling to the right where it is converted
back to electrical signal by same IDT structure. And in between we have a measurement
zone, this measurement zone is there because of the piezoelectric material we have a
certain velocity of acoustic wave. So this depends on what piezoelectric material we
choose. Based on this material this velocity is fixed it means when we apply a electrical
signal on the left we generate a electrical signal back from the right IDT and there is a
phase difference because of the time delay between these two IDT structure. Sometime
we call that delay line sensor the same saw acoustic wave sensor is sometime called delay
line sensor because we are measuring the delay in the in the acoustic wave going from
left to the right.

202
And this delay is fixed with the distance between IDT structure which is known to us and
the material what we choose. So that material is also known to us. So while fabricating
this sensor we know how much is going to be the delay of this delay line sensor or
acoustic wave sensor. Now when we do sensing using this delay line sensor, so we have
in between this delay this measurement zone if there is any impurity on on that surface
because of that the velocity of a caustic wave will change because of that impurity. So if
the velocity is changing the time taken from left to right is changing it means the delay
will change according to the impurity present on the measurement zone.

So that actual delay which was let us say delta or let us say t earlier the delay with
application of some impurity that time delay will change from t to t plus delta some delta
error function will be there which is proportional to what impurity it is and what size of
the impurity it is. So this is how this delay line sensor is used as a sensor as a caustic
sensor that the delay of this caustic wave sensor is proportional to what is the measure
and what is actually applied on the measurement zone. So that we discussed last time
delay line sensor. Today we are going to discuss a little advanced version of that delay
line sensor which is resonator or saw resonator we call it surface acoustic wave resonator.
So we discuss this saw resonator or we call it one port saw resonator.

Only one IDT placed at the center of the substrate. Is used for both input and output
transduction. So we have structure something like this. This is piezoelectric material. And
in this we fabricate one IDT structure at the center.

So we have let us say this IDT structure. This is one terminal and this is second terminal.
So these are the two terminals of IDT structure and we call it input and output IDT. So
here we have this piezoelectric material. So we have this piezoelectric material. So we
have this piezoelectric material. So we fabricate only one IDT structure. This will do both
the function as input and output. So when we apply a electrical signal here it is going to
generate a acoustic wave and if an acoustic wave hits it this that signal will be converted
back into electrical signal and this input and output ID input and output electrical wave
this we can do in time multiplexing manner for example for a fraction of second we just
apply input and this IDT will generate acoustic wave. This acoustic wave generated by
this IDT structure will travel in both the direction positive x axis and minus x axis
because this IDT is same the symmetrical along this y axis.

So we have this acoustic wave which is going to be generated and will travel in positive x
direction as well as in negative x direction. So these are the acoustic wave generated.
Now both the side acoustic wave will travel if we put somehow a reflector on both the
edges for example we have a reflector present. The reflector is nothing but this strip of
metals only they are out of phase of this frequency so that they start reflecting this wave.
So these are the reflectors. So when we place these reflectors on both the sides these
acoustic wave will travel from IDT to the reflector and at the reflector they will bounce

203
back convert their direction will change they will come back to IDT and the time taken
by the acoustic wave will be the distance which is 2 times the distance from IDT to
reflector because there is a complete cycle IDT to reflector and then reflector back to
IDT. So 2 times the distance between IDT to reflector divide by the velocity of acoustic
wave and that acoustic wave velocity is known to us because of the material what we
choose. So in this way this reflector actually bounce back these acoustic wave which
comes again to the same IDT structure we generated it and at this time these acoustic
wave will be converted to electrical signal. So the same IDT structure now will convert
this acoustic to electrical and give us as a output. When we see these two applied signal
these two signals will be delayed in the time and that time what we just discussed it will
be twice the distance between IDT and reflector divided by the velocity. So these two
signals will be displaced by that time delay and this is called the SAW resonator this is
kind of a resonator where this acoustic wave is bouncing back to the same IDT. So this
kind of structure is used for making of sensor and there are two types of these SAW
resonator depending on the application the one is the direct and one is the indirect. Direct
SAW resonator is something where we apply the electrical signal directly on this
structure. Indirect is something we transmit the signal and it receives the signal and
applied to this device. So we have two types of these surface acoustic wave sensors.

There are two types of wave sensors of surface acoustic wave sensor. The one is the
direct and second one is indirect. So let's discuss them in brief the direct source sensor. In
this type of direct source sensor the measure end affects the propagation of SAW in the
sensor. The measure end for example temperature, pressure, strain etc, affects the
propagation of SAW. SAW in the sensor in attenuation and delay. So the example for this
is the torque sensor. So in torque sensor we have a shaft on which we are applying a
torque. So this is a shaft and we are applying a torque on this let's say like this and we
want to measure how much is the torque is being applied on this shaft. So we can place a
SAW sensor and we can kind of paste it on the shaft itself. So we have these IDT
structures which we place on the shaft itself one let's say in this direction and one is
perpendicular to the z-axis this direction. So we can paste these SAW sensors right on the
shaft itself. Now when we apply a torque on this shaft because of this application of this
torque there is a deformation in the shaft and because the SAW sensor is sticking to this
shaft there will be a deformation in the SAW sensor itself. Now think of it as a delay line
sensor let's say we have pasted a delay line sensor on the shaft in case of no torque
applied in the normal neutral condition this delay line sensor has a time delay of T.

When we apply the torque now because of the deformation let's say there is a
compression in the SAW sensor because the SAW sensor is pasted on this shaft because
of this compression the distance from the left IDT to the right IDT this distance is
reduced. So accordingly the time taken from first from the source from the source to the
receiver that time delay will reduce now because of the because the distance is now lesser

204
because of the compression. So this we will detect it by a change in the time delay earlier
let's say the time delay was T now the new time delay will be T minus delta T. So this
delta T which is the now the difference between the delays this is proportional to how
much torque is applied. If we apply a high torque it means there will be a very high
compression and the size of this SAW sensor will reduce means the distance from source
to destination will further reduce and accordingly there will be this time delay the
difference in the time delay will be proportional to how much torque is being applied on
the shaft.

So this is how we can use these SAW sensors as a torque sensor and we see here we
pasted one SAW sensor on the shaft. If we want to have like perpendicular in which
direction the torque is being applied that we can measure by placing two SAW sensor one
in the let's say one direction and one is the in the perpendicular direction. So we by
analyzing both the signals we can know in which direction the torque is up is being
applied. So this is what the direct SAW sensor is where the measure end which is torque
here this measure end is directly influencing the speed of SAW the surface acoustic wave
it is directly influenced by the measure end. So when we apply this torque there is a direct
change in the velocity or the how much time it takes from source to the destination. So
we call it the direct SAW sensor and this SAW device is rigidly mounted to this shaft. So
this is the shaft and SAW device is rigidly mounted on a flat spot of shaft. The torque
applied will stress the sensor. If we use let's say both the sensors we use two sensors for
practical applications. Two SAW sensors are used such that their central line are at right
angle.

So this is how this direct SAW sensor works we apply the torque which causes a
compression in the SAW device and because of this compression there is a change in the
delay of this acoustic wave from source to destination. This is direct SAW sensor now the
indirect SAW sensor is let's say we have a SAW sensor which is far away which
transmits a signal which is detected by some antenna and applied on this SAW sensor and
there will be a reflected acoustic wave which is according to what that sensor is. So the
application of this kind of SAW sensor is for example in identification or let's say we
have a spacecraft flying there and we want to check whether this spacecraft is ours or it's
from someone else. So we make a SAW sensor place it on the aircraft with the antenna
when we send a signal the reflected wave will be as per that SAW device what we have
placed already. This SAW device may have multiple reflectors so accordingly how many
reflectors we used in that SAW device reflected wave will have the signature of those
reflectors and we can know by those signatures whether it is our own aircraft or someone
else. So let's discuss what it indirect SAW sensor is and what is the use of this in terms of
identification of the objects. So number two is indirect SAW sensors. The information in
this case about the measurement is gathered by another sensor. So what happens in

205
indirect SAW sensor let's say we have a base station. On this we have a radar or
something which sends the signal.

So this is our base station and this sends the signal outside and we have a object. This is
an unknown object for example which has this SAW sensor. This is our SAW sensor with
antennas and in this we have some reflectors for example these are reflectors. So these
are the reflectors for example here we have one reflector, here we have three reflectors,
here we have two reflectors and this is our antenna and this base station is actually
transmitting a pulse towards this unknown object. So the reflected wave now will be
having the signature of these reflectors and the reflected wave will be having one peak
because of this one reflector then it will have three peaks and then it will have two peaks.
So this will be our reflected wave and this is our transmitted wave. So here we see we
have a base station which is trying to read what object it is whether it is our own object or
whether it is someone else object. So one dedicated SAW sensor is allotted to each and
every device when the base station sends this transmitted wave towards this object. This
wave this is a single frequency which will be detected by these antennas and this electric
this will be generated to electrical signal and applied to the SAW sensor. Now this
electrical wave will travel as a acoustic wave in SAW device based on the number of
reflectors there will be a reflected wave.

So if we have one reflector it means it will be reflected there once some part will be
reflected some part will be reflected in the same direction. Then we have three reflectors
it means it will be reflected at three instances. So it will generate three peaks in the
reflected signal and further we have two reflectors there those two reflector will give rise
to two peaks. So the reflected wave we receive that one peak three peak and two peak
that is according to how many reflectors we have put. So we can we can kind of put the
coding in terms of number of reflectors and the readout signal when we get this reflected
wave at back at the base station by the counting the number of peaks we can know what
is the code of this SAW sensor and this code can be unique to each and every object. So
this is how these indirect SAW sensor works and if we see the actual working of SAW
here we have a SAW sensor there. This is piezoelectric material and on top of this we
have antennas. These are antennas and these antennas convert the signal to electrical
signal which is applied to these IDT structures. These are the IDT structures. So when we
send the signal this signal is going towards this SAW device.

This will be picked up by antennas and will be converted to here into a SAW of the
surface acoustic wave. Then in the same line we have some reflectors placed. This is one
reflector. Then we have three reflectors and then we have two reflectors. So the reflected
SAW wave which will again be converted to the electrical signal will come out like this.
One peak because of the first reflector, then three peaks because of three reflectors and
then two peaks because of these two reflectors. So this will be the wave coming out of the
SAW sensor or we can say the RF response. And this we can call it ID tag using SAW.

206
This is the sample ID tag using SAW device. And this is the process of ID tag
identification.

So this is the second type of SAW sensor which is an indirect SAW sensor. We apply the
signal and the signal will be read out by a separate sensor which is here in this case this
radar is this will gather that information and will decode the information. SAW sensor is
used as a media to gather that information. So here it is acting as an indirect SAW sensor.
Now we discussed SAW sensor which is the surface acoustic wave sensor. In all these
cases the wave is actually traveling on the surface of piezoelectric material. Let us
discuss something called BAW sensor or the bulk acoustic wave sensor where the wave
will actually penetrate into the material into it pass through the material not on the
surface but into the bulk. That is why we call it the bulk acoustic wave sensor. In these
sensors a piezoelectric film is sandwiched So in BAW sensor we have a piezoelectric
material. This is a sheet of piezoelectric material. On both the side we put metal film
which will act like an electrode. So this is the first metal sheet on top and the second
metal sheet is on the bottom. This will act as a second electrode. When we apply a
electrical signal across these terminals let us say this I apply a signal then there will be an
acoustic wave generated in between these two electrodes and it will be propagating
within this piezoelectric material something like this. So this is the acoustic wave and this
is now propagating within the material this is piezoelectric material and the equivalent
circuit model of this consists of a fixed structure capacitance in parallel with frequency
dependent electromechanical resonance circuit.

So the electrical equivalent model of this will look something like a capacitor which is
fixed capacitor because of this material and a electromechanical resonance circuit which
is in parallel to this. It has its own capacitance, resistance and small amount of
inductance. Let us say I call it C0 and this is C1, R1 and L1. So here we have this
piezoelectric material on top and bottom of this material we place these electrodes these
are metal sheets and we apply our electrical signal across these two metal sheets. So
when we apply this time varying signal on these on this film a wave will be generated
within the material this piezoelectric material and this will will propagate from top to
bottom let us say from one electrode to another electrode this is within the bulk. If we
make a electrical equivalent model of this structure we have a C0 capacitance which is
because of the fixed capacitance structure this assembly is going to now store certain
amount of charge because this is a kind of a capacitor we have two plates and we have
some dielectric in between. So we get a C0 because of this and in parallel to this we have
a electromechanical resonant circuit because this is kind of resonating circuit which give
rise to a C1, R1 and L1. So this is a resonant circuit. So when we get a resonant circuit it
means this is going to store some amount of energy within itself and we have to talk
about something called quality factor of this structure. So we need to discuss now the
these resonators has certain Q or the quality factor.

207
So these barysonators stores the maximum acoustic energy within the structure achieving
a high electrical quality factor. We represent it with Q. So this going to store some energy
and the material is going to be chosen for this BOS sensors must optimize both electrical
and electrical systems. And mechanical properties. So we have this resonator in fact, and
this resonant circuit. whenever we have a resonant circuit we have an associated quality
factor with this how much energy is stored within the structure that defines how much is
the quality factor is and the material we use for fabrication of these BOS sensor that must
be carefully chosen because it involves electrical as well as the mechanical properties of
the material. So we have a quality factor associated with these devices. Now let us see
these bulk acoustic waves are again of two types the one we call it the film bulk acoustic
wave and the other one is this solidly mounted acoustic wave. So in the film bulk
acoustic wave we have a thin film which is containing all the energy stored in the solidly
mounted bulk acoustic wave this film is actually resting on different materials and the
some amount of energy is actually lost to those materials itself. So that will change the
quality factor to a degraded this mechanical energy to go to the substrate.

However this is shielding. This shielding is not as effective as we have in FBAS sensor
where this material is actually hanging in the air. So there is no effectively transfer of
mechanical energy through the air. But here we have reflector stack which can little bit
prevent this loss of energy but still there is some amount of energy which will be
transferred from piezo material to the substrate. So if we think about this structure, this
structure is easy to fabricate because we need not to have that etching from the
underneath of this structure. We just need to deposit multiple layers. Deposition of these
layers is very easy in fabrication process. We can just put one metal on top of another
metal. So the fabrication of this structure is easy but the quality factor of this structure is
not as good as the FBAS sensor. So we have the pros and cons of this structure is the low
quality factor but it is easy to fabricate.

The low quality factor but it is easy to fabricate. So, these are some pros and cons of the
solidly mounted structure and this film bulk acoustic and solidly mounted bulk acoustic
these are all types of BAW sensor and today we discussed SAW sensor where we have
delay line sensors. There are again two types one is direct and one is indirect based upon
the application and here we discussed the BAW devices or the BAW sensor where
acoustic wave is travelling within the bulk and we saw again two types of BAW sensor.
One is the film bulk acoustic wave the quality factor we get is very good but the
fabrication of these structures are difficult. On the other hand these solidly mounted
structures these are easy to fabricate we do not need to go for these under etching kind of
techniques but the quality factor here is not that good as compared to the film bulk
acoustic. So this is all for today. Thank you. this mechanical energy to go to the substrate.
However this is shielding. This shielding is not as effective as we have in FBAS sensor
where this material is actually hanging in the air. So there is no effectively transfer of

208
mechanical energy through the air. But here we have reflector stack which can little bit
prevent this loss of energy but still there is some amount of energy which will be
transferred from piezo material to the substrate. So if we think about this structure, this
structure is easy to fabricate because we need not to have that etching from the
underneath of this structure. We just need to deposit multiple layers. Deposition of these
layers is very easy in fabrication process. We can just put one metal on top of another
metal. So the fabrication of this structure is easy but the quality factor of this structure is
not as good as the FBAS sensor.

So we have the pros and cons of this structure is the low quality factor but it is easy to
fabricate. The low quality factor but it is easy to fabricate. So, these are some pros and
cons of the solidly mounted structure and this film bulk acoustic and solidly mounted
bulk acoustic these are all types of BAW sensor and today we discussed SAW sensor
where we have delay line sensors. There are again two types one is direct and one is
indirect based upon the application and here we discussed the BAW devices or the BAW
sensor where acoustic wave is travelling within the bulk and we saw again two types of
BAW sensor. One is the film bulk acoustic wave the quality factor we get is very good
but the fabrication of these structures are difficult. On the other hand these solidly
mounted structures these are easy to fabricate we do not need to go for these under
etching kind of techniques but the quality factor here is not that good as compared to the
film bulk acoustic.

So, this is all for today.

Thank you.

209
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture – 24
Magnetic Sensors: Magneto-elastic type

Hello, welcome to the course transducers for instrumentation. Today we are going to talk
about the magnetic sensors. Magnetic sensors are the sensors where we where we use the
magnetic properties of a material to sense a measurement. So magnetic properties is we
can classify broadly in two terms one is when a change in the magnetic property give rise
to change in the size or the dimension of the material and the second is when we apply a
magnetic field and the magnetic property give rise to a electrical change in the material
property. So there are two ways we can use a magnetic material to make a magnetic
sensor. So the first one is the magneto-elastic property where the dimension of a the
sample or the magnetic material is changing when we apply a magnetic field. So today
we are starting magnetic sensors and in this we use magnetic properties of material to
make a sensor for sensing of measurement. So the first one in this is magneto-elastic
properties or magneto-elastic effect. These are the effects when there is a change in the
sample dimensions. Q2 applied magnetic field.

So this is magneto-elastic effect and the reverse of this effect is elasto-magnetic effect
when there is a change in magnetic susceptibility. Due to applied mechanical stress. So,
here we have two effects magneto-elastic effect when we apply a magnetic field and the
sample dimension the length and width of the sample actually changes because of the
application of this magnetic field that is called magneto-elastic effect and the reverse of
this is elasto-magnetic effect when we when we apply a mechanical stress when we
compress the material by applying force or we expand it because of the change in the
physical dimension of the material it develops certain change in the magnetic properties
such as the magnetic magnetic susceptibility. So this effect is called elasto-magnetic
effect. Let us see how this happens actually in the materials. So in the materials we have
generally the dipoles magnetic dipoles inside the materials and this happens due to the
alignment of these dipoles when they are aligned then the dimension actually changes of
these material and when they are in the random fashion there is a different length. So let
us see in by a cartoon how this happens actually. So for example we have a certain
material which is magnetic material here and inside this we have magnetic dipoles. So
when these magnetic dipoles are not aligned so let us say this is one dipole and the
direction is like this we have second dipole. So right now they are randomly arranged but
if we apply a magnetic field and all these dipoles within this material they align to that
magnetic field then they look something like this.

210
This is first dipole, this is second dipole. So if we see on the bottom when we apply a
magnetic field all these dipoles are aligned in a line and they take let us say a distance of
L which is the dimension along x axis. This is the length f of this material L of this
material but when we do not have a magnetic field all these dipoles without magnetic
field they are random in nature they are aligned randomly. Because of that if we see there
is a difference in the distance how much area they occupy or how much length they
occupy in x direction we see they occupy somewhat smaller length than L and this is let
us say the error or the extra length which is caused because of the application of magnetic
field. So here H is 0 and here is H is applied which is a vector quantity. So here we see
because of the application of magnetic field all these dipoles are aligned and when they
get aligned they take up more length compared to when they were taking without the
magnetic field. So because of this change in the length of this dimension the x direction
of this material changes it increases when we apply a magnetic field. So this effect is
called magnetostriction when the length of a material changes with application of a
magnetic field. So we have a effect something called magnetostriction. This is the
property of ferromagnetic material. Which cause them to expand or contract in response
to an applied magnetic field. So here we have a property which is called
magnetostriction.

This property actually because of this property a ferromagnetic material changes its
dimension because of the applied magnetic field and how does it change that we have
shown in this figure when the dipoles which are there in the ferromagnetic material when
they get aligned they take up a different length and when they are not aligned they take
up it something smaller length and because of this delta the change in the length this is
this property is called magnetostriction. Now magnetization when we apply a magnetic
field there are two types of materials one material is they because of the application of
magnetic field there is a change in the magnetic property of that material and the second
type of materials is when we apply magnetic field there is a change in the electrical
properties of those material for example the resistance when we apply a magnetic field
there is a resistance change of those materials. So there are two types of materials based
on the what is the output of those materials. So magnetization changes or produces
effects which are mechanical or electrical in nature. Then the first one is magnetic elastic
field sensors. These are the sensors when there is a change in the Young's modulus with
magnetization. So there is a change in Young's modulus with magnetization. This is also
called delta y effect. And these sensors are often termed as acoustic delay line
components. So acoustic delay line components or ADLC. So this is first type of sensors.
The next we can make torque or force sensor using magnetic materials. So these are
torque or force sensors.

The torsion is produced using a ferromagnetic rod carrying current when subject to a
longitudinal field. So these are torque or force sensor. The third one which we are going

211
to discuss in this in these lectures are magnetoresistive sensors. When there is a change in
the electrical resistance. When subject to a magnetic field. So these are magnetoresistive
sensors. And the fourth one is the Hall sensors or the Hall effect sensors or we can also
call them magneto galvanic sensors or we call them magneto galvanic sensors. In these
sensors there is a crystal carrying a current. Produces a transverse voltage. When subject
to a magnetic field. So these are the four types of sensors where we have the first one is
magneto elastic. Then there is a change in the Young's modulus of the material when we
apply a magnetic field. So this is called magneto elastic field sensor. This we are going to
discuss. The second one is torque and force sensor when we apply a torsion in the
ferromagnetic material.

Then there is a kind of change in the property of material. This is torque and force sensor.
The third one is magnetoresistive sensor when the electrical resistance of the material
changes with the applied magnetic field that is magnetoresistive sensors. And the fourth
one is the Hall effect sensors or we call them magneto galvanic sensors. There is a
material which is carrying a current and we apply a magnetic field on this crystal or on
this substrate. There is a transverse voltage which develops in the material. So this is
based on the Hall effect and we call it the Hall effect sensor or the Hall sensor. So let us
discuss magnetic field, magnetic field sensors in detail. And this delta y or the change in
the Young modulus is an effect of the outcome of magnetostriction. So, a demagnetized
ferromagnetic material. When undergoes a mechanical stress. It develops two types of
stresses. So, when we have a ferromagnetic material and we expose it to a mechanical
stress, there are two types of stress actually developed within this material. One is the
normal stress which is developed in all the materials and the second one is this is special
stress which happens because of the magnetoelastic strain because the dipoles are
actually aligning because of that there is a extra stress which is developed inside the
material which is special only for the ferromagnetic material. So when we expose this
material to magnetic stress, there are two types of strain which is actually being
developed. One is the normal one which happens in all the material and the second one is
special to magnetic materials because of this alignment of dipoles. So we have two
strains developed. The first one is the plane mechanical elastic strain. Let us say we call it
epsilon s and the second one is the magnetoelastic strain epsilon m which is the result of
reorientation of those dipoles. Reorientation of those magnetic dipoles or magnetic
domains by the applied stress and let us call the applied stress as SA.

So we can write the overall Young's modulus for this magnetized material. So overall
Young's modulus of the demagnetized material. So the overall Young's modulus of this
material is given by capital Y which is the Young's modulus is equal to SA the applied
stress divided by the total strain which is developed and the total strain is epsilon s plus
epsilon m. So the Young's modulus here is different than the normal material which is SA
over epsilon s. For the magnetic material this is SA upon stress strain developed because

212
of the plane mechanical stress and an additional strain which is epsilon m. So the Young's
modulus is now a function of this special strain as well. This delta Y effect occurs in
nickel iron based crystalline alloys. And in some amorphous alloys. Such as this alloy
where the iron is let's say 40% nickel is 38. So this delta Y effect occurs in nickel iron
based crystalline alloys and some other amorphous alloys. For making magnetic field
sensors and one typical construction is the delay line sensors. Delay line sensor is very
much similar to what we discussed in acoustic sensors. In delay line sensor we have a
substrate which in our case now is a magnetic material and on top of it we put IDT
structures. So this is a typical delay line sensor where in this case we have a magnetic
material and we apply a electric signal on the left and we measure at the right side using
some instrument voltmeter or so and there is a wave which is propagating on this
material. Now when we apply a magnetic field on this magnetic material there is a
change in the length because of the magnetostriction.

This length of this magnetic material will change and if the length of this magnetic
material will change the delay reaching from the input to the output will change. So this
is called delay line sensor and this delay which is produced by because of the change in
the length. This change in the length is proportional to the applied magnetic field. So we
apply a magnetic field higher magnetic field then the elongation will be more and it will
produce more delay in the measurements. So in this way we can measure the applied
magnetic field using the delay line sensor.

So this is our delay line sensor. The sound velocity or in fact sound is a mechanical wave
so the velocity of this mechanical wave is given by V is equal to under the root Young's
modulus Y over rho which is the density of the material where this rho is the density of
the material. And the change in the velocity delta V is given by delta V over V is equal to
delta Y the change in the Young's modulus divided by the original Young's modulus. in
V of about 10% can be obtained. So here we see a delay line sensor where we use the
magnetic material as a substrate. When we apply the magnetic field there is a change in
the dimension change in the length of this magnetic material. Accordingly there is a
change in the Young's modulus which is the property of material and the sound velocity
or the velocity of mechanical waves is given by V equal to under root Y upon rho. Rho is
the density of the material which we assume remain fixed or remains constant. Then delta
V upon V delta V is the percentage change in velocity is equal to under root delta Y upon
Y which is the change in the Young's modulus of the material. And typically in the alloys
what we mentioned here typically a change of 10% we can achieve easily at room
temperature when we apply these magnetic fields.

So this is the delay line sensor. Let us now discuss the force or the torque or the pressure
sensors. Torque or force sensor using magnetic materials and the very common example
is the yoke coil sensor is used for torque measurement. So in this yoke coil sensor we
have a U shaped magnetic pole. magnetic pole pieces are mounted in a crossed fashion.

213
facing the sharp surface to provide narrow gap. One of these U branches used as primary.
with two coils P1 and P2 excited by NAC supply. And the other U branch or the second
U branch is mounted at right angle. So two secondary coils let us say S1 and S2. So in
this torque or force sensor we use a yoke coil based sensor. In yoke coil sensor what we
have we have two U shaped coils one U shaped coil it has two coils actually one on one
side and one on the other side. These two P1 or P2 they act as a primary coil we excite
both of them using some AC signal and we have a second U shaped branch which is
placed perpendicular to this primary. So primary let us say it aligns in the Y direction
then secondary is aligned in the X direction. Secondary also has two coils one S1 one S2.
So this secondary coil will pick up the flux generated by the primaries P1 and P2. So it
looks something like this. This is one primary and on this primary we have two coils
which is bounded on one side here and one on the other side here and we excite them
using external electrical signal. These are connected here. This is the primary and we let
us name it P1 and P2 and we have a secondary U shaped curve which is let us say
perpendicular to this. This is secondary winding. This will pick up the flux generated by
the primary which we can measure using some voltmeter or so and these secondary
windings are let us say S1 and S2 and these are S1 and S2 are perpendicular to P1 and P2.

This assembly is called yoke coil and we put it in very close proximity with the moving
shaft. This is the shaft which is applied some torque in this direction. So now if we see all
these primary and secondary coils and plot the flux generated by primary and received by
secondary, these are our primaries let us say P1 and P2. When we do not apply any
torque to the shaft, there is a uniform field distribution generated by this P1 and P2. And
here we have our secondary placed here and here perpendicular to P1 and P2. So in this
diagram, this is the flux diagram we see P1 and P2 are generating the flux in the shaft.
Shaft is coated with some magnetic material. And P1 and P2 when they are generating
this flux, this flux is uniform and it reaches S1 and S2 at the same time. P1 generating
flux and that also reaches at the same time when the flux generated by P2 reaches the
secondary. So because of the symmetry we can see S1 and S2 receives the flux at the
same time generated by P1 as well as P2. This is the condition which happens when there
is no torque applied on the shaft. There is no kind of field or the change in the magnetic
properties of material of the shaft or something is coated on the shaft. When there is a
torque applied because of the magnetostriction, when we apply a mechanical stress on the
shaft, the magnetic property of this material, the shaft that it changes and because of the
change in the magnetic field or the magnetic properties, the flux what we have shown
here which is uniform for time being, this uniformity will disturb and because of the
application of torque, if we apply more torque, the more change will be there in the
magnetic property and more disturbance will be there in this flux.

So if we draw the same diagram, when we apply the flux, it will look slightly different, it
will not be uniform and this non-uniformity which will be picked up by S1 and S2.

214
Accordingly, we can see that how much is the flux change compared to the normal
condition that change in the flux or change in the secondary voltage that is proportional
to the torque applied on the shaft. So if we plot the flux in case of applied strain, so this is
the one without mechanical stress or without torque. And now we plot flux with torque.
These are P1 and P2 primaries. This is the neutral axis where we have these secondaries
S1 and S2. Now because of the change in the magnetic property of material, this flux let's
say linked more towards S1 compared to S2. So we see there is a more flux picked up by
S1 compared to S2 and vice versa. So this change in S1 and S2 voltage because this flux
will lead to some potential buildup across secondary windings S1 and S2. This magnitude
of this secondary voltage gives us the reading of how much the torque is applied actually
on the shaft. One point to note here in case of yoke coils because we are relying on the
magnetic property change of the shaft. So the yoke coil which we are using here, this
need to be placed in very much proximity of the shaft so that the change in the magnetic
property of the shaft affects significantly on the flux linkage of primary and secondary.
So we need to have a very close closely placed yoke coil with the shaft for accurate
measurements.

So depending upon the material the sensitivity of this transducer system changes. For
example, a more branch type design sensitivity varies from 1 milli volt per Newton meter
to about 2.5 milli volt per Newton meter when material is C15 steel or chrome MOB
steel. In general, the torque sensitivity increases with increasing hardness of
ferromagnetic material. Other parameters governing the sensitivity are excitation
frequency, etc. And these can be chosen based on the following basis. For example,
transient requirements, shaft speed, shaft material, and shaft speed. So, in this yoke coil
based sensor, depending upon the material, the sensitivity actually changes and for C15
steel, the sensitivity actually varies from 1 milli volt per Newton meter to 2.5 milli volt
per Newton meter. And this generally the sensitivity decreases for the hardness of
ferromagnetic material. The more the harder the material is, the sensitivity goes down.
And there are multiple parameters which we can use for improving the sensitivity. For
example, excitation frequency or the frequency at which this AC signal is applied, that
can be used as one parameter. And we choose all these factors based on multiple
requirements. For example, transient requirements, shaft speed, how much is the shaft
speed, shaft material, and the output signals. These are some parameters on which we
decide which frequency and what parameters we need to use to make this torque sensor.

So, this is all for today.

Thank you.

215
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 25
Magnetic Sensors: Magneto-resistive type

Hello, welcome to the course transducers for instrumentation. Last lecture we started the
magnetic sensors, and we discussed some property for example, magneto-elastic property
and discussed a sensor based on that property. Magneto-elastic is the property when we
apply a magnetic field and the dimension of the specimen or the material changes
according to applied magnetic field. This property is reversible so if there is a change in
the dimension accordingly there is a magnetic field change or the magnetic flux changed
passing through that material and we discussed a torque sensor using yaw coil based
torque sensor and we measure that if there is a torque on a shaft the dimension of the
shaft actually changes because of this applied high torque and because of this change in
the dimension there is a flux linkage which is different than in the neutral condition when
we do not apply a torque. So when we apply a torque versus when we do not apply a
torque there is a change in the dimension and accordingly the flux changes. So that was a
torque sensor. Though the problem with measuring these change in the dimension is for
example, we are applying a torque on the shaft and there is a 1 micrometer kind of small
change is there in the dimension. So sometime it is difficult to detect that small
magnitude of change in the dimension. So we go for a different kind of property
something called magneto resistive property. Magneto resistive property is something
when we apply a magnetic field on a material and the resistance of this material changes.

So this property is magneto resistive property and the materials which show these
properties are magneto resistive materials and we can make sensors using this property.
The easy thing about the change in the resistance is this can be electrically measured. We
can apply a certain bias across the material and how much current is flowing that gives us
the value of resistance and if there is a change in the resistance even if it is a small
change in the resistance that can be easily captured by the electronics what we use for
biasing. So we go for these magneto resistive based sensors. So today we discuss
magneto resistive sensors on this property of a material changes with an applied magnetic
field is known as magnetoresistance. So there is a change in the electrical resistance of
material when we apply the magnetic field. However, the materials when we apply these
magnetic field they also have some amount of magnetostriction. Magnetostriction means
there is a change in the dimension. So when we apply a time varying magnetic field such
as in transformers for example when this magnetic field is changing with time
accordingly let us say sinusoidal wave there is a change in the dimension of these
material and that change in the dimension is also oscillating because this change is

216
because of that magnetic field and the magnetic field itself is time varying. So the
dimension which is changing this is also time varying because of this change in the
dimension and this is periodic change.

So magnetic this material this is oscillating this is sometime it is increasing and sometime
it is decreasing based on the magnetic field which is sinusoid it increases and decreases
sometime. So this material because of continuous compression and expansion this
material is also like oscillating it is shaking. So because of that because this is a physical
change in the dimension this material loses energy in terms of this mechanical waves.
This material is sitting on some other surface is in the proximity of some other materials.
This magnetic material now starts losing energy to outside environment because of
changes in this physical dimension and this is a loss happening because of this oscillation
that when we apply magnetic field we do not want the energy to be wasted to outside
environment from this material but because of magnetostriction there is a change there is
a transfer of energy this mechanical energy to outside environment. So we have
something called magnetostriction which we discussed earlier as well. The phenomena in
a material changes physical dimension by an applied magnetic field is called
magnetostriction and the ferromagnetic materials lose energy to heat from the physical
expansion. And contraction of the material caused by the magnetic field. This is
particularly important for power distribution applications. Power transfer efficiency.

Is a critical factor. So we discussed that there is always a contraction and expansion in the
material when we apply a time-varying magnetic field, and because of this the material
actually loses energy in terms of heat to outside environment. It shakes, it vibrates and
there is a kind of mechanical transfer of energy from this material ferromagnetic material
to the environment. For power distribution systems and these application there is a
critical factor of power transfer efficiency how much power we transfer using these
materials and these materials are very much used in power distribution system for
example the transformer. So electrical transformer is nothing but there are two windings
one is primary one is secondary and we convert a high voltage from the primary to a low
voltage in the secondary.

In the transformer we use a coil to kind of link the flux from primary to secondary and
this core this is made up of magnetic material which is used for proper linking and this
material is chosen so that there is a maximum flux linkage from primary to secondary. It
means that the magnetic properties of this core should be changing very highly and
instantaneously with the applied magnetic field or the electrical field on the primary.
Now if the material this core material also has certain amount of magnetostriction means
the size the dimension of this core also change with applied signal applied signal at the
primary, then there is a continuous compression and expansion and this core loses this
energy to outside environment in terms of mechanical vibrations. So very common
example of this is when we go for go near to a very big transformer on the street or on

217
somewhere then we see a humming sort of noise that is because of these magnetostriction
because when this transformer is continuously converting this electrical energy from one
high voltage to low voltage, then because of this frequency of electrical power supply
there is a continuous contraction and expansion in the core, and that gives rise to that
humming sort of sound, and this is nothing but the losing of energy to the outside
environment. So for power distribution system this is a very critical factor and we want to
have such type of material when there is no magnetostriction, or the amount of
magnetostriction is very less so the core does not lose energy to the outside environment
in terms of heat or vibration.

So one such material is the called nickel-iron alloy which we use for making the
transformer cores. So we have a nickel iron alloy which is nothing but nickel 81 and iron
19 was found to have essentially zero magnetostriction and widely adopted as
transformer core material. Allium alloy is the name given to this material. For
approximately 80-20 nickel iron alloy. So we have a alloy which is almost 81% nickel
and 19% iron.

So this composition of these metals which makes this alloy this is found to have almost
zero magnetostriction it means when we apply a magnetic field, it does not change any
dimension it does not create any kind of oscillation if we apply a time-varying magnetic
field. So this alloy is very much useful for making these transformer cores because it
generally does not lose any energy to outside environment in terms of mechanical or heat.
So this is very much used this nickel-iron alloy for transformer cores, and it is a name
given to this alloy which is perm alloy which is roughly 80 and 20% of nickel and iron.
So this is a magnetic material which does not have any magnetostriction. Now these
magnetic materials what we are going to discuss there are two types one is the hard
magnetic material and one is the soft magnetic material.

Hard magnetic material is what we have to apply a very high magnetic field to polarize
them to convert them into a permanent magnet and soft material are such materials we
need a very small amount of magnetic field and they those materials get magnetized. So
according to this we have two types of magnetic materials one is the soft type and one is
the hard type. So hard and soft magnetic materials. Though these terms are kind of poorly
defined. A soft magnetic material are easily modified by an external magnetic field.

Soft materials may even be affected by the internal field from the device. These soft
magnetic materials typically have a relative permeability of greater than 1 typically 10 or
higher. So these soft magnetic materials they can be easily modified by applying a small
amount of magnetic field and even that small magnetic field can be even generated within
the device itself due to some external factors and typically these magnetic materials have
a relative permeability in the order of 10 or even higher may be 1000 value. These are the
soft magnetic materials because they are easy to be magnetized by the small magnetic

218
field. They can be characterized by essentially. being immune to the effects of internal or
external magnetic field. And the relative permeability of hard magnetic material are
nearly unity. So these are hard magnetic materials. They are very difficult to get
magnetized and they are very immune to even internal or externally applied magnetic
field and the relative permeability of these hard magnetic materials are almost close to
unity. Now, if we want to make a permanent magnet, for example in the speaker phones
or we need a permanent magnet, which is always there.

So it is good to make these permanent magnets using the hard materials or hard magnetic
materials because once you get them magnetized, it is difficult to get them magnetized,
but once you magnetize them demagnetization is also not easy. So you magnetize them
once, and they retain this magnetization for almost for a very long time. So permanent
magnets they are made up of hard magnetic fields or hard magnetic materials. On the
other hand if you have some sort of a transformer action where you are changing the
magnetic property using a time varying magnetic field, and you want to change it very
rapidly. Those materials need to be soft materials so that it is easy to magnetize and
demagnetize them.

So we have permanent magnets which should be made using the hard magnetic materials
and for example the transformer code that should be soft magnetic materials. We have
permanent magnets are typically hard magnetic materials and per this permalloy is
considered a soft magnetic material. So we saw that we have two types of materials
magnetic materials hard type and soft type is easy to magnetize one compared to the
other one. Now we will see how to make a sensor using these magnetic materials. The
sensor is something now on which we will apply a magnetic field from outside and we
will measure how much is the magnetic field apply on these materials.

So the property of these materials will change as per the applied magnetic field. Now let
us see how do we make these sensors. So the first thing is we use these materials but we
want to have a very small size of sensor. So typically we go for the deposition of these
materials on a certain substrate for example let us say silicon we have a silicon substrate
and on top of them we deposit these magnetic materials a very thin layer using the metal
deposition and then we will kind of do photolithography to etch out the extra material and
we pattern these magnetic materials on the silicon that is our sensor. Now, we apply a
magnetic field on this patterned magnetic material and see how much the change in the
property.

For example the resistance of this material how much is change. So let us see how do we
make this sensor. So the process is the film deposition. So we begin with a substrate on
which or a base on which to create the magnetic field. This substrate is sometimes glass
or ceramic. But if the permealloy to be integrated. With an active device active device or
circuit. Such as amplifier the substrate need to be silicon. The substrate must be smooth

219
because the film is very thin. The permalloy is deposited for example, by sputtering on
substrate. For many magnetic films the deposition is performed with a magnetic field. To
create a preferred magnetic direction. Often referred to as easy access. Alternatively. The
preferred direction may be created after deposition also.

By heating the film. In the presence of external field. So when we make the sensor we
begin with a substrate on which we deposit this material magnetic material to make a
sensor and this substrate is sometimes glass or ceramic. But sometime we want a active
circuit as well for example amplifier because the signal generated by these materials are
very small. So we want to have immediately an amplifier stage to boost this signal
immediately. So this active circuit sometime we want it on the chip itself because if we
use a discrete amplifier the signal will fade away when reaching from the magnetic
material to the amplifier.

So sometime we want the circuit or the amplifier right very close to this magnetic field.
So if we want to integrate this magnetic field along with the active circuit then we need to
go for the silicon only because on silicon only we can make all these integrated circuits
the amplifier. And the substrate must be smooth because the film we use of this magnetic
material is very thin. So the substrate must be very smooth so that we can lay down a
layer which is very thin. And this permalloy is deposited generally as a thin magnetic
material often by sputtering there are multiple process but sputtering is generally used for
the permalloy deposition on the substrate.

When we deposit these materials on the substrate for example by sputtering so sputtering
is nothing but this material actually we bombard this material and this material actually
kind of go out in the inside that chamber and stick to our substrate. While this material is
sticking to the substrate at that point itself we apply a magnetic field in the chamber so
that when this material is actually depositing on the subset that if that point itself it is
magnetically kind of aligned all the domains they are already aligned so we get a easy
access which is our preferred access in which we apply the magnetic field accordingly the
material will be deposited. So by default there is a magnetization of this material while
deposition itself. So this is what we do when we deposit these materials on substrate. This
is sometime difficult because applying a magnetic field while doing the sputtering this
can be a little bit challenging sometimes.

So this can also be done after deposition of the film what we can do we apply a high
magnetic field and we heat up the material this thin film we will do a little bit of heating
so that all the atoms or all the domains magnetic domains magnetic dipoles they are they
have this kind of this thermal energy to move there and we apply a magnetic field so that
they get aligned to that magnetic field and then cool it down so that they remain there so
this magnetic material actually get magnetized even after the deposition by heating it up
using the applied external magnetic field. So this is how we deposit these materials on a

220
silicon or any other substrate which is a very thin film. So now we have these magnetic
field measurement and let us first discuss about this measurement along the hard axis
with a magnetic field applied perpendicular to the mechanical length. perpendicular to the
mechanical length in the hard axis direction the response of the device is shown below.
So for example we have this graph on the x axis we have the angle which is let us say
given in 0 degree 0 10 20 30 and so on.

This axis is let us say given in worsted and on the y axis we have delta R by R in
percentage so let us say this is at 0% this is minus 0.5 minus 1 minus 1.5 minus 2.0 and
the graph actually look something like this at this 0. So this is the graph of delta R by R
with applied magnetic field when this magnetic field is changing this angle from let us
say minus 20 to plus 20 this is the response of the magnetic material. So the resistance
decreases with applied field the resistance is maximum at 0 applied field. The maximum
change is approximately 2.5%. And this change is film thickness dependent.

Up to a bulk level of around 3%. And thicker film typically has a higher delta R. So, in
this case we make this sensor using this magnetic material and we measure along the hard
axis. The hard axis is what this is the axis which is perpendicular to the physical length.
So for example we fabricated that the sensor which has a magnetic material deposited in
x direction. So the dimension perpendicular to this for example y dimension this is the
hard axis for this sensor.

Now we apply the magnetic field along the hard axis. So this is not along the mechanical
length this is perpendicular to the mechanical length and we are applying a magnetic field
on this material. Now if we change this magnetic material let us say for minus 20 to 20
for example we change the dimension the direction in negative it was like in let us say in
the negative y direction in when it is 20 it is in the positive y direction. So when this
magnetic field is applied on along the hard axis there is a change in the resistance of this
magnetic material and this change is typically shown with this graph. This is maximum at
zero applied magnetic field and when we apply the magnetic field in either of the
direction there is a change in the electrical resistance it decreases by approximately 2% if
we apply a magnetic field even in the positive or as well as negative the direction the
change is same we see the bell shaped kind of curve which is symmetrical along the zero
axis.

So it shows the direction of the magnetic field does not matter but there is a change in the
magnetic field when we apply along the hard axis and the magnitude of this is typically
2% which is good enough for detection using the electrical amplifiers or so we can detect
this 2% change in the magnet in the resistance using our electronic circuits. So we have
this maximum applied field at the zero and decreases to 2.5% this change is also the film
thickness is dependent if we deposit a thicker film it typically gives a higher change in
the change in the delta R by R so this change is also the thickness dependent how much

221
thick we have deposited the this magnetic material. So now the magnetic field
measurement along the easy axis along easy axis. So when a field is applied along the
mechanical length then it is along the easy axis.

So, along the mechanical length in the easy axis direction. The figure below shows the
change in the resistance along the easy axis. So, we have this is the zero. And this is 10,
20, 30 and this is in the reverse direction minus 10, minus 20, minus 30 and the y axis is
delta R by R. And the change is a little complicated than the hard case while going higher
this is the zero point while going higher it shows this kind of behavior and then goes like
this. This is while going this direction while coming back let's take the different color
because this shows different path while coming back. So while coming back this comes
like this and decreases here and it goes like this. So here we see when we change the
direction of magnetic field while going on towards negative to positive this dip or the
decrease in the delta R by R it happens at let's say 15 value. But while coming back from
positive to negative this does not happen at the same point but now it happens on the
between minus 20 or minus 30 or so. So this has different path for going high and going
low this is we call the hysteresis it means that the device shows the hysteresis effect and
this kind of behavior of this graph sometime we call it the chirp.

So this is how the magnetic field of these sensor changes we have we fabricate these the
strip the very thin film of these magnetic material on the substrate. And when we apply a
magnetic field the resistance of these which we can measure using our biasing circuitry
we apply a constant bias across it and measure the current flowing through this. Now the
effect when we apply this magnetic field this depends on in which direction we are
applying the magnetic field if it is along the hard axis then we have shown that bell
shaped curve that is the change in the resistance and if we apply the magnetic field in the
easy axis or on or the along the mechanical length then the resistance change shows this
kind of typical behavior and we call this something called chirp. So this is how the
magnetic sensor works more specially specifically the magneto resistive sensor works
where the change in the resistance of the magnetic material when we apply this external
magnetic field.

So, this is all for today.

Thank you.

222
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 26
Magnetic Sensors: AMR and GMR

Hello, welcome to Transducers for Instrumentation. Last lecture we discussed the

magnetic sensors where the magnetic sensors are made of magneto resistive materials.
Magneto resistance is the property when the resistance of a material changes when we
apply a magnetic field from outside or it is developed inside. So, the electrical resistance
of the material changes and this electrical resistance is easy to measure we apply a certain
known voltage across the material and measure how much is the current flowing. So,
accordingly R equal to V upon I we can calculate the resistance of the material. Now, ah
today we will discuss how do we deposit these materials on some substrate.

We talked about a little bit about substrate and when we deposit these materials on
substrate. So, today we discuss how do we decide that how to deposit and how much the
material thickness we need to deposit on the substrate and accordingly how much
resistance we are going to get for to make a magnetic sensor. So, let us discuss something
called sheet resistance. So, when we deposit these materials on substrate we deposit them
in the form of thin sheet we sputter the material or there are many ways to deposit the
material, but the material is generally a thin sheet on the substrate and then we pattern it
as per our requirement.

So, how do we calculate the resistance of these sheets? So, that is called the sheet
resistance of the material and that is what we are going to discuss. So, we have something
called sheet resistance. So, for this actually we know that all the materials have certain
resistivity which is called rho this is the resistivity and this is the material property the
unit is ohm meter. So, this rho is the resistivity and the property of any material and the
resistance accordingly the electrical resistance r is given by rho L upon A which is the
cross section area or we can say T into W. So, cross section area we are assuming that
this is like a rectangle.

So, one side is W and other side is T. So, this is the formula of electrical resistance R
equal to rho L upon A, A is T into W and we can have certain deposited metal. For
example, we deposit something like this so, let us say this is the sheet of metal we deposit
on substrate length is this and W is this and T is this thickness from here to here. So, this
cross section we have one side is W which is the width of this sheet and T is the thickness
of the sheet and accordingly the electrical resistance is rho L upon T into W. In this we
can see that this thickness of this material is generally fixed because when we deposit this

223
material on a substrate we generally know how much is the thickness we are going to
deposit.

So, this T is generally a constant it is it can be varied but it is most of the time, discrete in
nature. For example, we go for 100 nanometer thickness 200 nanometer thickness 150
nanometers thickness. So, generally this T is priorly known to us and rho the resistivity
this depends on the material what we choose. So, this is also known to us when we
deposit a material we know what is the resistivity of this material. So, this rho and T both
are known to us and they are like a constant.

So, this rho upon T we can combine together and give it a different name something
called rho square. So, we write it rho square and L upon W. So, this rho square is the
sheet resistance and the unit is ohm per square and this is square is a dimensionless unit.
And once we have this formula we can just count the number of squares to get the total
resistance. So, R is equal to R square multiplied by the number of squares. So, here we
have this we deposit this metal with length L and width W and the thickness is already
known T and the resistivity also known. So, we write the formula for electrical resistance
as R equal to R square which is the combination of rho and T. So, R square into L upon
W. So, L is the length of the metal we deposit and width W is the width. If we assume L
is equal to W means the length is equal to the width then we are ended up with a square.

If we look from the top this is just a square where the length is equal to the width. So, if
length is equal to width it means we are talking about a square and L upon W is equal to
1 it means the electrical resistance is equal to the sheet resistance which is a constant
number for a given material and the thickness. So, now when we are depositing we know
the sheet resistance rho upon T. So, we just need to multiply by how many squares we are
adding up in series to get the total resistance. For example, we are adding up 10 squares a
square can be of any length the length as long as the length is equal to width the
resistance is going to be same. So, if we are adding 10 squares 1, 2, 3, 4 like that then if
we are adding 10 squares the total resistance of this pattern is equal to 10 times the sheet
resistance. So, this is very easy calculation to get the desired resistance when we fabricate
these sheets on the substrate. So, this is one example where we have this length is L, W
and T if we want to add two or more sheets two or more of these blocks together we can
also do that. For example, if we if we make like this, this is one slab and then we add one
more with it. So, the net net resistance from this side r equal to let us say r1 by 2 where
this r1 is the resistance of this sheet. So, if this sheet has resistance of r1 then if we add
one block in parallel then the overall resistance will be r1 by 2. So, this is how we can
add these blocks together to make a desirable resistance which we can use as a magnetic
sensor. So, we have certain sheet resistance known to us for different different materials.
For example, if we talk about some typical sheet resistances in let us say 180 nanometer
technology node. These are the technology node to use for microelectronics fabrication.

224
So, let us say for 180 nanometers process we are talking about and this side is the layer
which layer we use and here is the sheet resistance. And if we talk about let us say
diffusion the silicided one then the sheet resistance is typically 3 to 10. If diffusion
without silicide no silicide the resistance is typically 50 to 200 the sheet resistance
actually increases. Similarly, if we talk about polysilicon this is silicided one this is again
3 to 10 and the polysilicon with no silicide then the resistance varies from 50 to let us say
almost 400. And then again we have certain metals metal layers. So, metal layer one as
like sheet resistance 0.08 typically metal two is 0.05 metal three is again 0.05 ohm per
square metal four is typically 0.08 metal five is typically 0.02 ohm per square and metal
six let us say is roughly 0.02 ohm per square. So, if we see we have multiple layers in
microelectronics fabrication which we can use as a to make a resistor. The first one is the
diffusion when we diffuse certain area this also act as a resistor and there are two types in
this one is silicided one one is the non silicided one. Silicided one is the one where we
put some more metal and make some sort of alloy in that region.

So, that is the silicided one. So, accordingly the sheet resistance decreases around 3 to 10
and if we do not have this silicidation then the roughly the sheet resistance is higher little
bit than 50 to 200. Similarly, with polysilicon polysilicon is generally used as a gate of
mosfet and this also has two types one is the silicided one where again we make sort of a
alloy. So, the sheet resistance is lower 3 to 10, but if we do not do silicidation then the
resistance is roughly 50 to 400. So, if we want to have a resistance in our the requirement
of our resistance is higher for example, we want to make 1 kilo ohm of resistance. We
want to go for non silicided polysilicon or non silicided diffusion because the sheet
resistance itself is very close to 1 kilo ohm for example, 200 or 400. So, we will need to
add less number of squares to get 1 kilo ohm because 1 square is already give us giving
us 400 ohm. So, we just need to add 2 and half square 2 square and then half. So, that the
total resistance comes out 400 plus 400 plus 200 all together 1 kilo ohm. So, if we want a
resistance with higher value, we go for non silicided materials non silicided diffusion and
polysilicon.

However, if we want to have a resistance with lesser value for example, 0.1 ohm or 0.2
ohm then we would like to go for let us say metal 1 metal 2. These are the metal layers
when which we use for fabrication. So, we have a substrate silicon substrate on top of this
we use metal 1 for interconnection then metal 2 then metal 3 4 5 something like that and
when we are going higher the thickness of these metal layers increases. So, because this
thickness or the T of this of these layers are increasing the sheet resistance R square is
nothing, but rho upon T. So, if the T increases the thickness increases the sheet resistance
goes down this we can see in metal 1 as well as metal 6 metal 1 has higher sheet
resistance 0.08 and metal 6 has lower sheet resistance 0.02 that is because the thickness
of the material is increasing in metal 5 and 6. May be the material is still same for
example, copper or something, but because of the deposition thickness which is higher in

225
metal 6 the sheet resistance of metal 6 is lower. So, if we want to go for smaller and
smaller resistance value we want to go for these metal 1 2 whichever is appropriate for
our calculation. So, this is how we decide which one which material to choose for making
this sheet this appropriate resistance. So, now we see how do we fabricate and we
measure this sheet resistance. So, the fabrication process we discussed last lecture that we
take substrate and deposit the metal on this. So, for example, we have just deposited some
material with certain thickness on a substrate.

So, this is our deposited material now. Now we need to measure the its resistance which
we do using 4 probe method. So, 4 probe method is something like this. This is our
deposited substrate deposited magnetic material. Now, we put 4 probes on it. So, this is
pad 1 and this side we put pad 2. These are the 2 pads to apply the voltage and then we
have 2 more pads here which are generally to measure the voltage. Let us call these pads
A and B. So, this is called Kelvin connected anisotropic magneto resistive sensor. Or we
call AMR in short. So, in this sensor we fabricate we deposit certain material magnetic
material on the substrate which is shown here in gray color. This deposited material now
we put 4 pads on top of it to measure the electrical resistance. However, we just need
only 2 pads for normal measurement we apply a certain known voltage across these 2
pads and calculate how much current is going flowing through accordingly R equal to V
upon I we can calculate the resistance. But generally the resistance change in these type
of sensor is very small. So, for very precise measurement in fact, whenever we want to
measure a resistance very precisely instead of going for 2 probe measurement, we go for
4 probe measurement which we are going to discuss in a minute. This 4 probe
measurement is also called Kelvin connection and that is why this whole sensor is called
Kelvin connected anisotropic magneto resistive sensor AMR sensor which is Kelvin
connected. So, this easy axis of the material is in line with the mechanical length. So, this
mechanical length is in X direction which is the easy axis of this material now which we
discussed in the last lecture as well. So, easy axis of the material is in line with the
mechanical length of the resistor the material for example, material is Permalloy changes
its resistance and we measure the change in the resistance per resistance. So, we measure
delta R upon R the change in the resistance divided by the total resistance is measured.

So, that the base value of resistance does not impact the calculations. So, that the base
value of resistance does not impact calculations. Now this above figure shows picture
how do we measure this resistance. So, this use 4 point Kelvin contacts to minimize the
effect of contact resistance between the Permalloy and connecting metals. So, we have
this metal deposited and we put 4 probes on it the easy axis of the material is in line with
the mechanical length and now if we apply a magnetic field on this assembly the
resistance is going to change. We may generally measure delta R upon R instead of just
measuring the change in the resistance which is delta R we measure delta R upon R
which is the actual value of this resistance. So, that the base value does not impact our

226
calculation we are just focused on how much change is there in per unit resistance. So,
for example, percentage change in the resistance is what we are looking for. So, for that
actually we measure delta R upon R for the measurement. Now this figure shows the how
we do the measurement which is basically 4 point measurement and this is called the
Kelvin connection.

So, instead of using just 2 pads we are putting 4 pads here and by the way why do we use
4 pads because whenever we have a material for example, here permalloy which is
deposited here. Now we need to apply a certain voltage across this material permalloy.
So, that we do using some external probes these probes are made up of some material for
example, copper. So, when we connect these copper probes on the permalloy which is a
different material nickel iron alloy. So, as soon as we connect this these different
materials together there is a contact resistance which comes into picture.

So, we know that resistance of copper stand alone that is known to us we know the
resistance of permalloy that is also known to us, but when these 2 different material joins
that resistance the contact resistance is depending on which metals we choose and this is
different for all the different combination of materials. So, when this contact resistance
comes into picture this is also creates some extra resistance in the assembly which is not
required or which is kind of extra to our system which should not be there. So, to deal
with this contact resistance we go for Kelvin connection or this 4 point measurement. Let
us see a little bit how do we do this 4 point Kelvin measurement. So, why Kelvin 4 probe
method used? So, drawbacks of 2 probe method. So, is the error due to contact resistance.
So, let us say we have this material which is deposited here. Now we apply a certain
voltage and we have another source. So, this is the assembly this is point x and this is
point y and this length is L. This is the current measured and this is the voltage measured.
So, here we have this permaloy which is deposited I want to measure the resistance of
this permalloy. So, what I do I apply a potential across these x and y point and the
distance between them in length L. So, this potential is applied through this battery which
is E here and now I measure 2 parameters one is the current flowing into this loop and
one is the voltage measured across these point x and y. So, this voltage and the current
flowing into this circuit it depends on how much is the resistance between x and y of this
permaloy. In fact, right starting from these 2 points which is let us say I start calling x
dash and y dash. So, between x dash and y dash whatever is the resistance that will
dictate how much current will flow and how much voltage will build up.

Now between x dash and y dash there are 3 resistances one is the resistance because of
the permaloy which let us say I call by this red resistance symbol. This is the resistance of
permaloy which is known to us because we know which material we use and we know
the resistivity. But at point x and at point y where this copper metal which is shown in
blue when this copper metal is contacting this permalloy that time this contact resistance
is not known to us how much it is that we cannot measure. So, I represent this resistance

227
let us say by this green symbol resistance here and resistance here. So, now if we see
overall between x dash and y dash we have 3 resistances one red and two green these
resistances because of the contact resistance. And these contact resistance are always
there because of this 2 different materials. So, we want to have only this red resistance
which is due to permalloy, but these 2 resistances are always there and this is an error in
our measurement because of this contact resistor. So, this is the problem with going with
2 probe method because the current need to flow in through this through these materials
and these resistance are always there. So, they are going to give us a certain voltage
across them and that is an error to our measurement. So, we go for a 4 point measurement
where we put 4 probes one for measuring of voltage and one for the current.

So, current is in the total loop and voltage where we measure the voltage we do not
actually flow any current into that. So, that the contribution by contact resistance is 0 if
there is no current flowing into this contact resistance they are not going to provide any
error in the measurement. So, this is what we do in 4 point measurement let us see how
that happens. So, Kelvin 4 probe method produces 2 probes for supplying current and 2
probes for measuring voltage. So, in Kelvin connection we have like this. The first 2
probes 1 and 2 they are used for applying current. And the other 2 probes let us say A and
B they are used for applying current. So, they are used for measuring the voltage only.
So, this is how we do the measurement by 2 probes we apply this current which is
independent to the measurement of voltage and these 2 inner probes A and B they do not
carry any current because the impedance of voltmeter is very high. So, there is no current
effectively flowing in A and B probe.

So, whatever is the contact resistance in probe A and B that does not impact the reading
of voltmeter. So, we can precisely measure the potential across these 2 probes A and B
and to calculate resistance we simply need to divide the this voltage divide by the total
current which gives us the value of R. So, this is a 4 point Kelvin measurement and this
inner probes does not draw any current. So, let us write this as well inner probes does not
draw any current does not impact the measurement. So, this is what we have shown the
anisotopic magneto resistive sensor where we deposit this permaloy and we apply a
magnetic field and the resistance of this material changes that resistance we can measure
directly using 2 probe or 4 probe.

So, we go for 4 probe measurement because these give us the more accurate result
compared to the 2 probe measurement. So, this is AMR sensor the magnetic AMR
sensor. Let us discuss now the GMR sensor which is called the giant magneto resistive
sensor. So, the principle behind this GMR sensor is if we have 2 magnetic materials
which are polarized in one direction then if there is a layer in between which is carrying
current this current or this the resistance of this material is actually dependent on what is
the polarity of the other 2 layers. So, let us discuss this GMR sensor the giant magneto
resistive sensor.

228
So, in this type of material for example, we have 2 layers which are having one other
layer in between. Let us say this is first layer in between we have a sandwich layer. So,
let us say this is the first layer in between. So, let us say this is the first layer in between.
So, this is the assembly of GMR sensor where we have 2 different magnetic materials
shown by this red let us say this is A and the other one is green which is B now.

So, these 2 magnetic layers they are polarized in opposite directions red is polarized
towards the towards the right and the green is polarized in let us say in the opposite
direction. And if there is a current flowing in the middle layer which is this blue layer we
are applying a potential across this through this battery and there is a current flowing this
current get scattered by this opposite polarized magnetic materials and the current sees
more scattering and the resistance seems to be higher in this case where these materials
are oppositely polarized. So, 2 different ferromagnetic materials sandwiched by thin
conductive layer and the mobility of electrons with the parallel spin is higher than those
with anti parallel spin. And these magnetic layers are typically 4 to 6 nanometer and this
conductive layer is typically 3 to 5 nanometers. So, here we have these 2 ferromagnetic
materials and one conductive layer is sandwiched between them. Now the mobility of
these charge carriers in the conductive layer this blue is the conductive layer. The
electron mobility in conductive layer depends on how the other 2 layers are polarized if
they are in opposite direction then the mobility of electrons is lower and if they are
polarized in the same direction then the mobility of these electrons are higher. So, this
can be easily detected using the external circuitry and we can check whether the both the
materials are same are polarized in the same direction or the reverse direction. So,
depending upon this we can find multiple applications of this kind of sensor or devices
one application of these is the hard disk, where we have this magnetic material coated on
the disk and the head which reads the data on this hard disk that is kind of GMR sensor.

Other application is making logic gates using these materials. So, for example, A and B
layers if they are similarly in the same direction they are polarized that can be a one state
if they are oppositely polarized that can be treated as a sec the second state. So, we can
use these kinds of GMR sensors as the logic devices as well. So, this is the case shown
here where the the both the metals both the materials are polarized in opposite direction
and in the same direction, we can have the higher mobility. So, we can plot that as well.
So, this is the lower mobility case where we have this scattering and lower mobility.

In another case if both are polarized in same direction. So, this is the case shown here
where the both are polarized in opposite direction this green is also in the same direction
this is A this is B and this is conductive layer and we apply a voltage across it. So,
because of these two layers polarized in the same direction we have lower scattering and
higher mobility. So, this is what the thing is GMR sensor or giant magneto resistive
sensor and this is how it works. The typical applications are in the magnetic hard drives
which we use in computers.

229
There we have a disk which is coated with these magnetic materials when we write on
these magnetic disk the the head actually polarize the magnetic material in a certain
direction and while during reading head reaches to this location and check whether this is
polarized in the same direction or in the opposite direction if it is in the same direction
the current of in this head increases if this is polarized in the reverse direction the current
in the head [Link], this is how we do reading and writing in typical computer hard
drives. This is the one typical application of GMR sensor. So, today we discuss two
magnetic sensors one is the AMR sensor and one is the GMR sensor.

So, this is all for today.

Thank you.

230
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 27
Magnetic Sensors: Hall Measurements

Hello, welcome to the course transducers for instrumentation and we were discussing
about the magnetic sensors or magnetoelectric sensor. So last lecture we discussed about
AMR and GMR sensor the GaN magneto resistive sensors. Now today we are going to
discuss about the Hall sensor which is another form of magnetic sensor as well. So the
Hall sensor we can classify as a electrical sensor and as well as a magnetic sensor
because the effect we measure using the Hall sensor is magnetic effect we primarily
measure. So this Hall effect sensor is basically we apply a voltage across a semiconductor
or a material where the electrons are flowing and we apply a magnetic field as well and
based on the magnetic field these charge carriers see a force applied on that which is the
Lorentz range force and based on that the output is generated. So this Hall effect sensor is
very much used in measuring the rotational speed. So for example we have a rotating disk
and we put a Hall effect sensor nearby to this rotating disk and we measure how much
rotation per minute is going to be there on that rotating disk. So we discussed today the
Hall effect sensor. So let us say we have a rotating disk and this has multiple fractions
and these are nothing but the north and south poles of a magnet alternating and this disk
is rotating along this axis. So we place now a Hall sensor very next to this disk.

This is Hall sensor and we can measure how much rotation this disk has. So primarily we
have a Hall sensor or Hall device in this. This device is actually the sensing element and
this converts periodically applied magnetic field into voltage. So we have this Hall device
which is sensing element and it is converting the magnetic flux which is linked by these
permanent magnet placed on this disk. This flux is converted into a voltage by this Hall
device or Hall sensor. Now this developed voltage is very small in magnitude. The
voltage is few millivolts only. So this very small magnitude is not suitable to process
using the extra electronics. So first we have to amplify it. So most of the time when we
employ a Hall device or a Hall sensor this is followed by a amplifier which is kind of a
linear amplifier. We can make use of BJT or MOSFET for making this amplifier and this
amplifier is used to boost this signal which is in millivolt to a range of let us say volt. So
this Hall device is now followed by a linear amplifier. The output of these Hall devices
goes to a linear amplifier and the purpose of this linear amplifier is amplify the Hall
voltage. And for this purpose we can use a very well known common emitter amplifier
configuration. So now we have a Hall device which is sensing this magnetic flux and
converting it to the voltage which is in few millivolts and then followed by a linear
amplifier which is amplifying this few millivolt into volt kind of range.

231
But the problem is when the signal comes from the Hall device that signal comes along
with the noise as well which is always there in these circuits. So this noise also gets
amplified by linear amplifier because the amplifier cannot differentiate between a signal
and a noise. So whatever is the input to linear amplifier that is going to be amplified. So
signal is also amplified but the noise also get amplified by this linear amplifier. So most
of the time this linear amplifier is also followed by some filter which is can be a low pass
filter or depending upon the application. So this linear amplifier is followed by a filter or
sometime we use Schmitt triggers because Schmitt trigger also can be used as a filter. So
this is followed by a Schmitt trigger. So we have the next stage of linear amplifier is the
Schmitt trigger. And this is basically nothing but the comparator circuit. This creates
basically digital pulse based on the amplified Hall voltage. So this is the complete
system. Where we have a Hall device which converts flux to a small magnitude of
voltage. This is followed by amplifier which converts this or boost the signal from few
millivolt to volt and that we use a Schmitt trigger. Schmitt trigger is nothing but kind of a
comparator. Let us say we assume that the reference value is 0.

So anything below 0 will be treated as 0 and when anything any voltage above 0 volt will
be treated as 1 volt. So this Schmitt trigger converts the data or converts the input
whatever is generated by linear amplifier into a digital form which is having 0 and 1 for a
kind of simplicity. This signal can be easily processed by further electronics. And this
Schmitt trigger also act as a filter. We will see when we discuss this Schmitt trigger
today. So let us now discuss individual blocks in detail. So first is the Hall device or the
Hall sensor. Hall effect sensor. So this sensor is made up of for example P type
semiconductor and has 4 electrodes. So this is the Hall effect sensor. A DC voltage is
applied between the left and the right electrode. So let us see how this Hall sensor look
like. So we have a semiconductor material. This is our semiconductor material where we
have 4 terminals. 2 are left and right. This is left and this is right. These are the terminals
going out and we apply a voltage across these two terminals. So let us say this is V
external for the DC supply. And other two terminals are let us say top and bottom. Let us
say this side is we call top and this side we call B or bottom. So these two terminals we
take out and we measure the voltage across these two terminals. This is nothing but the
Hall voltage. And on top of this we have a magnet which is applying the magnetic field
on this material. So we have a magnet. This magnet has north and south poles.

So this is let us say north pole, this is south pole and this has a magnetic flux linked with
it. So let us say this is like this. So this is the flux which is linked now to this material as
well semiconductor material. And in this material we have free moving electrons and
holes. So let us say electrons are moving from right to left. So these electrons will see a
force applied on them because of this magnetic field which is called Lorentz force. And
these electrons will start accumulating towards the B terminal as shown here. And holes
on the opposite because of the opposite charge they will see similar force but in opposite

232
direction. So they will accumulate towards the top. So here in this setup we have a
semiconductor, p-type semiconductor slab where we are applying a DC voltage and
because of that application of voltage there is a current flowing in the semiconductor and
this current of course is made up of electrons and holes, the charge carriers. Now when
we have this magnet on top of it because of this there is a applied magnetic field on this
semiconductor. When a charged particle moves in a magnetic field that experience a
force that is called QvB sin theta or the Lorentz force we call it. Because of this motion
of charged particle this experience a force and because the electrons are negatively
charged they see this force in let us say towards B terminal and the holes which are
positively charged they see the opposite the same force is the opposite direction and they
accumulate towards the top. So when there is a accumulation of or the separation of these
charged particles positive towards the top negative towards the bottom then because of
this accumulation there is a voltage build up across this semiconductor slab between T
and V the top and bottom terminal and this we measure using a voltmeter and this voltage
developed is called the hall voltage. So this is how this hall sensor works.

The more the magnetic field means if the magnet is really close to the semiconductor the
force magnitude is more it means the more hall voltage will be generated and this also
depends on what is the velocity of these electrons and holes in the semiconductor. So
here we apply this DC voltage between top between the left and right and we measure
this hall voltage on the top and bottom. So because of this the free electrons move within
the electric field through the hall material or the semiconductor material move within the
electric field through the semiconductor material. Now when magnet comes closer to hall
plate the magnetic field B creates the Lorentz force. So when magnet comes closer to the
hall plate the magnetic field B creates Lorentz force. FB and the formula for this FB is
minus Q V B sin theta where this Q is the normal charge on electron V is the velocity of
electron and B is the magnetic field. So the magnitude of Lorentz force is minus Q V B
sin theta, Q is the electronic charge which is 1.6 into 10 to power minus 19, B is the
velocity of electron how fast the electron is moving into this magnetic field and B is the
magnetic field the value of magnetic field how much is the magnetic field applied. Sin
theta is the theta is the angle between the velocity and the magnetic field and most of the
time because this magnet is very perpendicular to the motion of electrons so theta
becomes 90 degree so sin theta becomes 1 so we can neglect sin theta from this equation
for time being. So the Lorentz force which is applicable on these electrons as minus Q V
B and depending upon whether it is electron or hole the polarity of this force will be
opposite in nature.

So because of this these electrons will accumulate towards bottom and holes will
accumulate towards the top we can even draw this by that arrows these holes are attracted
towards the top. So this is how the volt hole voltage is being generated in these hole
sensor. Now we can write the hole voltage build up in this sensor BH is equal to I into B

233
upon ne into d and where this I is the DC current through the hole device. B is the applied
magnetic field. E is the electron charge. C is the thickness of the material or the
semiconductor. N is the charge carrier density. So this is the output hole voltage VH
equal to IB upon nED. I is the DC current which is flowing into the semiconductor which
we can measure very easily using some emitter. So all these parameters are something
which we can measure very easily in a setup. B is the applied magnetic field. E is the
electronic charge which is a universal constant. D is the thickness of material how thick
the semiconductor material is and for precise applications precise measurement we try to
keep this D very well below very very small thickness and N is the charge carrier density
of the material. So this is the output voltage generated by this hole sensor. Now let us
think about the sensitivity of this hole sensor.

So let us write down the sensitivity. So we have this graph for hole voltage VH. The hole
voltage developed if we change the magnetic field in this direction. So when magnetic
field is positive we have let us say hole voltage build up like this and in negative when
the direction of magnetic field is reversed the hole voltage generated is in the reverse
direction. So this is a typical generation of hole voltage versus magnetic field. So if we
see the slope of this graph is nothing but the sensitivity. The slope of this graph is the
sensitivity and sensitivity we can write S A equal to VH hole voltage developed per unit
applied magnetic field. So this we can write I upon N E D. So this is the sensitivity of
hole sensor and usually the thickness is in micrometer let us say if we can write the
typical values of D which is in the order of 250 micrometer and the current in these
sensors are typically of the order of 10 micro amps. So these are the typical values of D
and I and sensitivity is given by I upon N E D. So how do we increase the sensitivity of a
hole sensor either we increase the current I so if we increase the current which means we
have to apply high voltage and get more current it means the power dissipation will be
more in the holes effect that is not as wide advisable.

So I also has certain limits so beyond that those limits we cannot increase the value of I
or we can reduce the thickness D so that sensitivity goes high but that also has limitations
so how thin we can go in that semiconductor plate or semiconductor film that also has
physical limitations we cannot go much much more thinner materials because of many
other issues. So D also has certain limits so the sensitivity is dictated by the current and
the thickness of the material. So this is the hole sensor which we apply a magnetic field
and the hole voltage VH is developed VH is given by IB upon N E D. So this voltage
now is applied or it is it goes to the next step which is the amplifier unit linear amplifier
which amplify the small amount of hole voltage into a measurable volt kind of range. So
let us discuss that linear voltage amplifier. These linear voltage amplifiers are nothing but
the common emitter amplifiers which we which is very well known configuration. So we
can draw a common emitter amplifier. So if we talk about the BJT version then common
emitter amplifier looks similar to this. This is an NPN transistor. This has collector

234
resistance and emitter resistance and on the base side we use mostly voltage divider bias.
This is a simple configuration of a NPN amplifier linear amplifier. We have typically a
voltage of 5 volt at the supply and this is the Rc the collector resistance. This is the
voltage divider bias so Rb1 or R1 simply and R2 are the voltage divider biases. This is Re
the emitter resistance and at the emitter we have a capacitor as well which is called Ce or
the emitter capacitance and we use Cb and C load. This is RL the load resistance and this
is typically the signal unit which is the preceding unit which is generating the small
signal and applying it at the Cb and this voltage generated by this V signal. This actually
gets amplified at the output node which is V0. So this is a simple structure of linear
amplifier or common emitter amplifier. We will not discuss how this amplifier actually
works. This can be referred to some second year basic electronics course. What the
amplifier actually does this takes the signal from V sig which is applied at the capacitor at
the base.

This small signal is amplified by this NPN transistor which has a gain let us say beta and
the output is taken at the collector end and the collector is having a resistance Rc and at
the emitter we have a emitter resistance which is needed for the stability purpose and
because stability is needed only for the DC condition for AC or when we are talking
about the signal it should not be there. So we connect a parallel capacitance CE which
bypass this emitter resistance in case of the signal analysis and we get the boosted output
at the Vout. So this is how a common emitter amplifier works. We will not go much into
the detail of how it works. So this amplifier now takes the signal of millivolt range and
convert it back into convert it into a volt kind of range which is now amplified to
typically 1 or 2 volt or so. Now this amplified signal is applied to the Schmitt trigger. So
the configuration of Schmitt trigger is very simple. We generally make it using Op-Amp
where this is positive, this is negative terminal, negatively connected to ground and the
positive terminal has a feedback. These are R1 and R2 and this is the output, this is 0.
Now this is the configuration of a Schmitt trigger which act as a filter and it converts the
data into a digital form.

So let us talk about what actually is the problem is and why do we want to employ a
Schmitt trigger to rectify that problem. So the problem is when we get a data from hall
device that is full of noise. The signal is not very smooth it has little kind of spikes. Then
we feed this to the linear amplifier which further amplifies everything the data as well as
the noise. So now the noise also get amplified. Now we need to convert this data into
digital form so that we can process this data. But because of that noise it is not easy to
convert that into digital form. Let us see how it is not possible and what we should do. So
for example if we do not have a Schmitt trigger we just use a inverter to convert the data
into digital and this is the threshold of an inverter. Let us say this is a fixed crossing or
fixed threshold and our signal is something like noisy. So this is our signal which is a
sinusoid but full of noise. Now if we see at this point if we have a single threshold just

235
like an inverter, inverter works like if anything above this it will be detected as 1,
anything below this it will be detected as 0. So now if we see closely here we are having
0 here but as soon as this threshold is crossed at this point we start receiving 1. But we
see again this signal again goes below this threshold voltage so again it becomes 0 here
and after this 0 this signal again goes up because of this noise. We see now a 1 here so we
see this ideally there should be only 1 transition from 0 to 1 because the signal is this
large signal and which transits from 0 to 1 only once.

But because of this noise in this section we get an additional 1 and 0. So if we convert it
into digital this is like 0 then we get a small 1 which again goes back to 0 and then again
1 which continues here. So this is a undesirable thing which is like a high frequency
transition or spike we can say which is undesirable and this happened because of the
noise. The same thing happens actually here as well this 1 is there and then when signal
goes below this threshold black line this becomes 0 again because of the noise this
becomes 1 here and again becomes 0 here. So same thing we have here this goes 0 and
then again for a short while it becomes 1 and continues to be 0. So these are the things
which are undesirable these transitions are actually happening because of the noise in the
signal. It is not signal is crossing only once this threshold but because of the small
amount of noise this noise is converted into a digital data which is a high frequency noise
and we do not have any means of removing it because the signal has these digital
transitions. So our further electronics will treat it as a signal only. So this is the problem
with noise when it get enhanced and we cannot deal with it using a single threshold
which is just a black line. So instead of having this a single threshold we employ a
Schmitt trigger which has 2 threshold.

It means there is a hysteresis in the graph and for going up there is a separate threshold
and for going back or going low is a separate threshold and we have 2 thresholds now
instead of 1 black line and these 2 thresholds will filter out this noise. So let us see the
response of a Schmitt trigger. The Schmitt trigger is something like the response is this is
V out and here this is V in and the response of Schmitt trigger is something like while
going high this goes like this and at some point it triggers and goes to high voltage like
this. So this is the threshold voltage while going high or we call it Vth threshold voltage
going towards high while coming back let us draw it in a different color. So while going
back when we decrease the input the threshold is different. So this is the path while going
high to low and this threshold is called Vtl the threshold voltage while going low and the
midpoint is that 0 voltage or the origin because we have connected this terminal to 0 volt.
We can apply certain voltage here so that we can shift also this entire graph. For example
if instead of 0 we apply certain voltage if Vr let us say this Vr is 0 here if Vr is not 0 then
we can draw it differently. This is Vr here this Vr is 0 and this is Vtl Vth and this is Vtl.
So now because of these two threshold voltage we can plot the same graph here. Now
instead of this single threshold now we have two thresholds. This is Vth and this is Vtl

236
and we plot the we apply the same signal full of noise. So now the high will be detected
at this point not at this point. So even if this signal is going low here for a short while
because of the noise the low will not be detected until the signal is reached Vtl. So this
small amount of noise will not be detected as a signal and if we plot the output this is 0
here, this is 1 here only after this and 0 here and this 0 actually is detected till this point.
So 0 will be up to here till the signal crosses Vth and here this will become 1 and then
this 0 will be detected when the signal goes below Vtl so this point.

So here this is 0, this is 1 and this is 0 which is the perfect case what the signal should be
and we can remove all these noises what is actually inserted by the amplifier or generated
by the sensor itself. So Schmidt trigger is actually used to filter out these noises which are
inherently there in the system. Now if we see carefully that the signal the output signal is
little delayed compared to the earlier case that delay we can live with it, it is already is
always there in the system because of we have employ the Schmidt trigger in the circuit.
So this is how a Schmidt trigger works. Now let us see a small example of linear
amplifiers and we can have it as a numerical problem. So one example, so a common
emitter amplifier circuit. As a load resistance of 1.2 kilo ohm and a supply voltage of 12
volt. For the transistor to be switch on completely. Assume collector to emitter voltage or
BCE equal to 0. And the value of emitter resistance Re if it has a voltage drop of 1 volt.
Cross it. Calculate value of other circuit resistance as well. Beta for the NPN transistor is
given as 100. So this is a numerical example which we can solve and the solution of this
is given below. So we have this NPN transistor.

So the collector resistance or RC will come out as 1.2 kilo ohm. This is 12 volt. The
value of R1 will come as 20 kilo ohm and R2 is around 3.6 kilo ohm. The emitter
resistance come out to be 220 ohm. BCE we assumed 0 as given and here at this point we
take the output Pout. These are the typical values of resistances what comes out after
calculation. So we can do all these maps at some different time. The IC current come out
like 4.58 milli ohm. So this is a small example for linear amplifiers. This is particularly
covered in some semiconductor device courses like electronic devices and circuits
basically is a P-TAC second year course, complete course. So we will not discuss much
about how do we solve these circuits but this is typical circuit for linear amplifier. So
today we discuss the Hall effect sensor and basically after the Hall effect sensor we also
discuss the linear amplifier and the Schmitt trigger which is basically used as a filter for
the Hall effect devices.

So, this is all for today.

Thank you.

237
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 28
Radiation Sensors: Gas Filled Detectors

Hello, welcome to the course Transducers for Instrumentation. Today we will discuss
about the radiation sensors. So, as the name says radiation sensors, we are going to
measure the radiation which is happening from outside environment and we want to
measure how much radiation is received using some sensors. So, as we know the
radiation is nothing but the electromagnetic wave which is coming in and as the
electromagnetic wave is nothing but the stream of photons. So, we have photons which
are high in energy. So, when these photons strike or interact with some other matter, they
produce certain kind of effects on that matter.

For example, it ionizes certain gases or it kind of give out some electron hole pairs in
semiconductor. So, these kind of effects actually happens when a high energy photon
strikes with certain media and this is what we want to measure how much ionization or
how much how many photons are received in a particular cavity or in a particular volume
and we want to convert that quantity into a measurable electrical signal. So, this is what
the radiation sensor actually does. So, we are going to start today the radiation sensor. So,
radiation sensor, there are three types of detectors that are most commonly used. So,
before that radiation sensor is nothing but the sensors used to measure radiation. And
there are three types of detectors which are mainly used. So, we have the radiation sensor,
the radiation sensor, the radiation sensor, the radiation sensor. The first one is the gassed
field detectors. The second is scintillators. And the third one is the solid state detectors.
And these gas field radiators detectors are further of three types. The first one is
ionization chamber. The second is proportional counters. And the third one is the geyser
counter or geyser-muller counter. In short, we call it GN as well. So, here we have
radiation detectors which are of three types. One is the gas field detectors, the other is
scintillators and the third is the solid state detectors which are composed of for example,
some semiconductors like silicon based or some organic material based detectors. Those
are those falling the semiconductor or the solid state detectors.

The first one is gas field detectors. As the name says, there is a gas field in a cavity and
we measure based on the property change of this gas. So, these gas field detectors are
further of divided into three types. One is the ionization chamber. The second is
proportional counters. And the third is the GM detector or GM counter or the geyser
detector. So, these are of three types gas field detectors. Basically, all these three work on
the similar kind of principles but based on the applications and the voltage applied they
differ in the functionality. So, we have gas field detectors which we are going to discuss

238
first now. So, the first is gas field detectors and as the name says when the gas in the
detector comes in contact with radiation it reacts with the gas becoming ionized. And the
resulting charge is 0. So, here we have gas field detectors where we have a cavity and in
that cavity we place a certain gas along with two electrodes cathode and anode and we
apply a voltage difference across these these two electrodes. When ionized ionization
radiation comes in means the high energy photon comes in it reacts with the gas it reacts
with what the gas is filled in that cavity and when it reacts this ionizes the gas ionizes
means the it creates the charges in the gas filled in the cavity and because we have
applied a potential difference this charge will be accumulated on these cathode and anode
and that charge how much charge is developed by this ionization this can be measured
using some electrical instrument like a emitter or voltmeter. So, that charge we can
measure which is proportional to how much radiation is coming and reacting with the
gas. So, this is how this gas field detector works and as we said there are three types one
is the ionization chamber the second one is the proportional counters and the third one is
the gm counters.

So, there are three types let us discuss the first one in detail. So, this works on the same
principle, but this operates at a low voltage. So, this is anode and the measure
measurement from the primary ions only. So, let us see the structure of these ionization
chamber. We have a cavity where we fill a gas. So, let us say this is a cylindrical cavity.
This is filled with a certain amount of gas and in this we place two electrodes let us say
this is the electrode at the center and there is one electrode on the edge we call it anode
and this is cathode and inside the cavity we have filled gas. Inside these electrodes are
connected to the variable DC voltage source. So, we have two terminals. The other
terminal we connect to anode through a meter or current meter or so, so that we can
measure how much charge is actually developed.

So, this is a typical structure of this gas filled detector. We have this cavity filled with the
gas and there is a DC voltage source where we apply a known DC voltage across anode
and cathode. Now, when the energetic particle comes from outside and passes through
this cavity that is going to knock down the gas particles because of this its high energy
breaks that these kind of particles the gas atoms. It ionizes the gas and when the gas gets
ionized because we are applying it certain DC voltage across anode and cathode they will
move towards these electrodes they will be based on the polarity if it is positive it goes to
the negative side. So, when this gas gets polarized one, we assume that one photon give
rise to one kind of pair of electron hole though it is not electron hole it is actually the gas
which dissociates which ionizes and one unit of charge in the gas.

This charged particle now reaches an anode which we can measure using this electrical
current how much electrical current is flowing through that kind of corresponds to how
much of the photons are received in the cavity. So, we have this let us say some high
energy photon comes in and this knocks down the gas particles ionizes the gas. This is

239
high energy photon or the radiation. So, this is how these gas chambers actually work and
they registers a measurement from primary ions primary ions means the how many
photons are received those are the primary ions which when they convert it into the
ionized gas that is called the primary ions. So, let us say 100 photons come in and they
give rise to 100 charged particles in this cavity. So, those 100 are called the primary ions
and this setup registers only that because those 100 charged particles will reach to anode
and cathode and give rise to current which is proportional to 100 photons. So, that is why
it registers a measurement from primary ions. Next is because it registers only from the
primary ions the measurement that the detector records is directly proportional to the
number of ions or pairs created. So, this the measurement shows that the detector records
is directly proportional to the number of ion pairs created. This is particularly useful as a
measure of absorbed dose over time.

They are also valuable for the measurement of high energy gamma rays. So, we have
these ionization chamber where the measurement is being done only from the primary
ions. So, the number of photons received they convert into the ion pairs and which is the
measurement is going to be done. So, because of this we can say this measures the
absolute number of photons and how much photons are received based on that our output
reading is there. So, this kind of measurement is useful for a longer time. For example,
we want to measure for in let us say 1 hour how much of radiation is being received. So,
this kind of measurement can be done using an ionization chamber. So, we because the
charge developed is directly proportional to how many photons are received. So, we can
use this as a measurement for absorbed dose how much of dose is actually absorbed by
the gas over a period of time. So, we take let us say 1 hour or 1 day or a finite amount of
time.

In that finite amount of time how many photons or how many energetic particles are
received and absorbed by the gas. So, this is what the measurement is and they are also
valuable for high energy gamma rays that also ionizes the gas. So, for gamma ray
detection this is also used these ionization chamber. So, these are the good points about
ionization chambers. However, these chambers are not able to differentiate between
different types of radiation. For example, one is the gamma radiation one is the x rays
radiation both are ionization radiation. But when the rays come in and they ionize the gas
we cannot detect by ionization chamber whether it is because of the gamma rays or
because of the x rays. So, thus that detection is not possible using ionization chambers.
So, we have ion chambers that are unable to discriminate between different types of
radiation. Meaning, they cannot be used for spectroscopes. And they are also widely used
in laboratories. To establish different standards. For calibration. And sometime they are
little expensive than other kind of detectors. So, here these ion chambers they are not able
to discriminate between different kind of radiation as we say it cannot differentiate
whether the ionization is happening because of the x rays or because of the gamma rays.

240
So, if it cannot discriminate between different type of radiation we cannot use it for
spectroscopy. Spectroscopy is something where we detect what kind of radiations are
coming in. So, this kind of ionization chambers are not suitable for the spectroscopy
measurements. They are widely used in laboratories because we measure the absolute
number of photons how many photons are coming in which are being absorbed by the
gas. So, they are very much used in reference standards to make to calibrate the
instruments how many photons are actually received.

So, for calibration of those instruments these ion chambers are very much used. And
depending upon the size if we want these kind of detectors in a bigger size then the cost
actually goes up. So, they are little expensive than other kind of detectors. So, this is the
first type of ionization chamber. Let us discuss the second type. The second type is
proportional counter. So, proportional counters are also gas filled detectors and the
configuration of these detectors is similar to what we just saw in the last slide. We have a
cavity which is filled with the gas and we apply we have two electrodes in that cavity
where we apply that the voltage. So, the proportional chamber differ from that is in
proportional chamber we apply a little higher voltage in anode and cathode. So, that when
a gas ionizes these ionized particles they further give rise to something called avalanche,
but this avalanche is not a complete avalanche it is kind of a discrete avalanche.

We do not apply so high voltage. So, that it goes purely into avalanche region. We apply
the voltage only in certain range where the avalanche happens, but it does not happen
very high. So, it is kind of a discrete avalanche we call and when a high energy photon
comes in this is absorbed by the gas it creates charged particle. This charged particle start
gaining energy from this high electric field or the high voltage what we apply here it
acquires that and that energy and convert a avalanche which is limited in nature or we can
say discrete avalanche. So, we have these proportional counters which operate at a
slightly higher voltage expected such that discrete avalanches are generated. Each ion
pair produces a single avalanche. So, that an output current pulse is generated which is
proportional. To the energy deposited by the radiation. This is in the proportional
counting region. So, here in proportional counter we apply a little higher voltage.

So, that these discrete avalanche comes into picture they get generated and each ion pair
produces single avalanche. So, that an output current pulse is generated because of this
avalanche because avalanche will suddenly release certain amount of charges which
appear to be pulse at the output we measure in the electrical ah signal. So, this is this
voltage is chosen. So, that the entire counter or the detector work in proportional counting
region it means the avalanche are generated only in the discrete nature. Now, if we have
radiation which is coming in we can differentiate between the radiations because whether
it is a gamma ray where it is a kind of x ray radiation because these radiations are differ
in terms of energy how much how much is the energy of these charged particles for the
photon coming in. So, a high energy photon will give rise to a more stronger discrete

241
avalanche and a lesser energy will give something like a lesser ah kind of avalanche. So,
this counter proportional counter we can use for certain amount of spectroscopy which
we cannot do in the earlier case where it was only measuring the amount of radiation not
able to differentiate. Proportional counter however we can differentiate between the
different types of radiation as well. measure energy of radiation and provide
spectrographic information differentiate between alpha and beta particles alpha and beta
particles. And the other advantage is the large area detectors can also be constructed.

So, in this type of detectors for example if we see the gas filled cavity where we have this
gas filled in and we have the electrodes there. So, for example this is the one electrode
which is let us say anode and in this gas-filled cavity this is the complete cavity full of
gas. A high energy ray is coming like this. So, it ionizes the gas for example let us say
this is the ionization is being happening here. So, this charged particle come this way and
this is the avalanche region. This avalanche will give rise to multiple charged particles
something like this one charged particle give rise to multiple charged particles. So, this is
one avalanche let us say second it is ionizing here. So, this charged particle comes here
and then start ionizing here 1 to 2, 2 to 4 and something like that. So, this is one
avalanche and this is let us say the third avalanche which is happening here. This is let us
say one avalanche 2 and 3. So, here in this figure we can see that a high energy wave
high energy electromagnetic wave is coming in this red direction this that is high energy
ionization.

So, this high energy particle is coming in this direction and the gas which is filled in this
cavity that get ionized after ionization this charged particle move towards the anode, but
it does not give rise to avalanche till the point it avalanche it reaches the avalanche region
and the avalanche is happening only in this avalanche region which is enough only for a
discrete avalanche is not like complete avalanche effect is a controlled avalanche effect
and we call it let us say the discrete avalanche happening and here we see 1, 2, 3 these are
based on the how many charged particles are created by ionization particle. So, this is
how this proportional counter works. The disadvantage of this kind of proportional
counter is these anode wires are little bit delicate and they lose efficiency in long term use
of these proportional counter. So, the disadvantage are that anode wires are delicate and
can lose some efficiency in detectors due to deposition. So, this is second type of counter
which is proportional counter now we are measuring how much of charge is actually
received in the cavity and which is and we are generating a controlled avalanche by
applying a slightly higher voltage the voltage is not very high. So, that the complete
avalanche is not happening the avalanche is happening only in a small region which is
proportional to how many charged particles are actually generated by ionization particles.

So, this is proportional counter the third one is the GM counter or geyser counter we say
in that geyser counter we apply much higher voltage compared to the proportional
counter. So, that structure of GM counter or the geyser counter is exactly same only the

242
difference is the voltage we apply if we apply a very high voltage the generated charged
particle immediately creates a avalanche a sea of ah charged particles and this avalanche
is generated throughout the gas and we cannot detect how many ionization particles are
received because even a single ionization particle can create this avalanche because of
very high applied electric field or the voltage.

So, that kind of measurement is used to detect for example single photon kind of detector
if we want to measure whether a photon is coming or not some radiation is there or not
we can use those GM counter if a single even a single photon comes in that ionizes the
gas immediately we see a avalanche of charged particles and immediately we see a very
high value of current flowing into these anode and cathode. So, these GM counters are
used as a single photon detectors. So, let us discuss those GM counter or sometime called
geyser molar or simply the geyser counter. So, they work on higher voltage. So, this
voltage is selected such that. Each ion pair creates an avalanche. But, by the emission of
UV photons.

Multiple avalanche are created. Along the anode wire. And at the gas volume. So, this
ionizes from as little as single ion pair event. So, if we see the cavity view just like we
saw last slide. We have this anode which is there in the gas filled cavity. And a high
energy wave comes in. And even a single event occurs of ionization. Let us say this is the
single event which occurs in the gas cavity. This single event give rise to the ionize the
avalanche which is happening from 1 to 2, 2 to 4 and 4 to 8 something like this. And in
between because of this very high applied field in the cavity, this also create other
avalanche as well. So, let us say this particle which gains sufficiently high energy, it
creates its own avalanche separate from this first avalanche. So, everywhere these
avalanche are happening. These can also create some avalanche. So, at the end we get
multiple avalanche happening in the cavity and that happens because we have applied a
higher voltage between anode and cathode. This high voltage creates a very high electric
field in the gas region. And these charged particles they accelerate because of this high
electric field. And they gain energy from this high electric field when they acquire
enough energy to knock down other gas particles as well.

So, they start their own avalanche as well. So, one even a one photon which is received
from ionization particles, a single photon can give rise to multiple avalanche and in
instant whole gas is getting ionized because of a single photon which is received. So, this
kind of counter is used to detect whether a radiation is there or not because even a single
photon comes in this gazer counter immediately ionizes all the gas and we receive its
very strong pulse in the output the current pulse because of this so many charge. So, we
the current pulse produced by the ionizing events. Which can generate which can derive
the count rate or radiation and those. The advantage of these gm counters are they are
cheap and very robust devices. The disadvantage is it cannot measure the amount the
energy of radiation. So, again there is no spectroscopic information. Hence, no

243
spectroscopic information. So, this is the gm counter where we have a instant avalanche
happening even if it charge a single charge particle comes in the whole chamber of gas is
actually ionized because of the avalanche. The advantages are they are very cheap and
robust devices, but the problem is we cannot detect between different kind of radiation
because any of the charge particle if it is coming from gamma ray or x ray or from any of
the radii or any of the ionizing radiation a single rate ionization is creating a avalanche
and after that we cannot detect whether it was this avalanche was generated by a gamma
ray photon or a x ray photon.

So, this is the problem that we do not have any spectroscopic information from these gm
counters. So, these are this is the third type of gas filled detectors which is the gm
counter. If we classify based on this applied electric field or applied voltage then we have
these three kind of radiation detectors which is gas filled. If we want to draw a graph
based on the ion pair charge collection. So, let us say we have this graph where on the x
axis we have the voltage applied and on the y axis we have rate of charge collected.

So, in the first region where the voltage applied is less let us say this is the lower region
where we apply lower voltage and this region is where we apply a higher voltage. So, the
first one is the ion chamber region. So, this is the region where this ion chamber region
detector works. In medium applied voltage there is the proportional counting region and
when the applied voltage is very high then we have a geyser region where we get sudden
avalanche. So, this graph shows when we increase the voltage the first lower voltage is
ion chamber region where the rate of charge collected is kind of constant and we cannot
differentiate between the radiation this is ion chamber radiation.

Then we have a linearly increasing graph with the voltage and rate of charge actually
collected is proportional to how much voltage we apply. This is the region for
proportional region where we can detect whether which kind of radiation is coming into
the gas cavity. The third is the higher voltage which is the geyser region or GM counter
works. There again the instant avalanche comes in and we do not have any spectroscopic
information we cannot detect whether the radiation is of this type of that or that type. But
in GM counter even a single photon we can detect very easily which shows that whether
radiation is present or not present. So, these are the three types of gas field detectors
which we use for radiation detectors.

So, that is all for today.

Thank you.

244
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 29
Radiation Sensors: Scintillators and Solid-State Detectors

Hello, welcome to the course Transducers for Instrumentation and we are discussing the
radiation sensors or radiation detectors. In this lecture we discussed about gas filled
detectors where we discussed that if we have 7a chamber and we fill the gas inside this
chamber and the radiation is there. So because of that radiation the gas ionizes and the
number of ions which are generated in the gas is proportional to somehow the radiation
what is received by this gas and based on the applied potential difference we had three
kinds of these gas filled detectors. So today we are going to discuss about the different
type of detectors which is scintillation detectors and these kind of detectors work on the
principle that some materials if we apply a radiation on those certain materials they emit
light, the light is actually immersed out of those materials and this process or this
phenomena is called fluorescence or photo phosphorescence. So, this because of this
phenomena when radiation is coming to the substance and it generates a light this light
can be measured by photo detectors or so. The amount of light generated is now
proportional to how much radiation is received by this material.

So, this kind of process is used in these scintillation detectors. So today we are going to
discuss these type of detectors. So we have scintillation detectors. So, some substances
emit light when radiation falls on them. The emission process is called fluorescence. Or
phosphorescence. And this light output can be used as a measure of incident radiation.
This is the basic principle of these scintillation detectors. So with the material which
generate this light this is coupled to some optical detector which measures how much
intensity of the light is received and basically this intensity is very much proportional to
intensity of the radiation received by this material. So now we have three different kinds
of process by which this light is actually emitted. The first one is fluorescence where the
prompt emission of light is there when we have a material, and we shine the radiation on
this immediately the light actually is generated by this material which we can detect. So
this is called the fluorescence. The next is the delayed fluorescence where we shine the
radiation and there is a delay in the emission of this light. So that is the second one. The
third one is phosphorescence where the intensity or the wavelengths of light is lower than
the fluorescence case. So this is third type of process by which a material generates these
light. So we have three types of process. The first one is fluorescence. In this process a
prompt emission is there, prompt emission of visible radiation from a substance
following its excitation.

245
So, this is the first process. The second process is the delayed fluorescence. The delayed
emission of visible radiation from a substance following its radiation or excitation. And
the third one is phosphorescence. In this the emission of longer wavelength is there
compared to fluorescence and this is lighter than fluorescence. So this is the third type of
process. The third type of process is the delayed fluorescence. So this is the third type of
process. So this is the third type of process. So this is the third type of process. So this is
the third type of process. We have three types of oscillators. So let us look at the kind of
scintillators. So in this type of scintillators we have a cavity where this material is put in.
For example, this is the cavity where a material is put in. This material show the property
of phosphorescence or fluorescence. Now this is the cavity which is called scintillators.
And the radiation comes in from outside. For example, this is the radiation coming in and
it hits this material. This is gamma ray photon. So when this radiation hits this material, a
light photon is generated. This is a light photon. The wavelength of this photon is in the
visible range. Now this cavity is coupled to another cavity generally. This cavity is used
for multiplying of this effect.

For example, one photon comes in from the left. Then this cavity is electron multiplier or
simply we can call it the multiplier wherein we have a transparent section where this
photon can go in. This is the region where this photon can go in. And then this photon
actually generates photoelectron. Here you have a magnetic shield across it. This is the
magnetic shield. And this interface is typically a photocathode. And the other side is the
anode. So when this light photon comes in this cavity which is a multiplier cavity, this get
bounced back by multiple electrodes and give rise to photoelectrons. And these
photoelectrons which are now charged particles, these can be detected by this cathode
and anode which we place in this cavity. So the number of electrons received by this
cathode and anode give rise to a current in the external circuit which is a proportional
measure of how many light photons are there in this multiplier cavity. And that is also
proportional to the number of gamma ray photons we detected because the material is
generating kind of proportional number of photons from these gamma ray photons. So
this is the typical structure of a scintillator. This is a typical scintillator. So now let us
discuss some important properties of a good scintillation detector.

So, the very first property of a good scintillation detector is high scintillation efficiency.
The second is the light yield should be proportional to the deposited energy or the
radiation. Light yield should be proportional to the deposited energy. Next is the detector
material should be transparent to the wavelengths of its own emission. The detector
material of its own wavelength. The next point which is important is the decay time of
the induced luminescence should be shorter. should be short. And the other point is it
should be possible to make them in large size and desired shape. So, in large size and
desired shape. So, these are few important properties of good scintillation detector. The
very first is high scintillation efficiency means when we shine the radiation on the

246
material the proportional number of optical photons should be generated. So that is the
scintillation efficiency and for this detector to work good we should have a high
scintillation efficiency of that material. The next is the light yield should be proportional
to the deposited energy. The number of visible photons generated should be somehow
proportional to the number of gamma rays photons received by the material. So that we
can make a count of how much of the radiation is being detected by the material.

So, the light yield should be proportional to the deposited energy. The third is the
detector material should be transparent to the wavelength of its own emission. So we
have a material when we shine the radiation on it. After accepting this radiation this
material is going to generate of the visible photon the optical photon and this material
should be transparent for this particular wavelength of photon. Otherwise, what will
happen if it is not transparent this generated photon will be annihilated within the
material itself and it will not come out which we can detect using our other circuitry. So,
the detector material should be transparent to the wavelength of its own emission what it
generates. So, it should be transparent to that emission. The next is the decay time of the
induced luminescence should be short that is another important point, and it should be
possible to make this kind of detectors in large size and desired shape so that we can have
multiple applications of these kind of scintillation detectors. So, these are some important
points for this. Now let us see some more important properties of good scintillation
detectors. So, the refractive index should be near to that of glass means typically 1.5 or so
to permit efficient coupling. With the photo multiplier tube. The next is the high density
and high atomic number. And it should have good temperature stability and mechanical
properties as well. It should have good resolution and ease of operation. So, these are few
more important properties of these scintillation detectors. The first one is the refractive
index should be close to the glass because glass is almost transparent to the visible light
and if this refractive index is close to glass or 1.5 we can couple our photo multiplier tube
easily with the scintillators and the photons which are visible photons they can now travel
from scintillation to the photo multiplier tube without any problem. So refractive index
should be close to 1.5. The good temperature stability and mechanical properties of
course the scintillation detectors should be stable for any temperature change and
mechanical properties should be good. It should have good resolution means if we get a
little amount of radiation on the material we should have a detectable change in the
output so that we can easily measure how much is the radiation being detected and ease
of operation is the whole detector should be easy to operate. For example we apply a
certain instrument and it directly gives the reading in how much of the radiation is being
received. So this is second type of radiation detector where we have scintillation detector.
Now let us discuss the third type of radiation detector which are solid state detectors
where we use these semiconductor based detector for example CMOS based detectors.
These CMOS devices they also we can use them to detect the detector these radiation. So
let us discuss these solid-state detectors. So, in solid state detectors MOS structure or the

247
metal oxide semiconductor structure are useful are useful because of their superior
sensitivity. As well as excellent compatibility. With the existing microelectronics
technology.

So, these solid state detectors are different than other gas field detectors and scintillators.
In these type of detector we use semiconductor devices and one such semiconductor
device is MOS device or we call metal oxide semiconductor structure which is a fairly
known structure in microelectronics. This structure is useful because of their superior
sensitivity it gives a very high sensitivity as well as the excellent compatibility with the
existing microelectronics technology. So by the microelectronics technology we know
that these MOS structure can be integrated in a very large number and also we can attach
the active circuits which is used for processing the data right on the with the device itself
on the same substrate. So we have a substrate where we use a MOSFET as a radiation
detector. Alongside of this detector as well we can place a active circuit. So if the
distances are short there is less chance of noise produced in the system. So these metal
oxide semiconductor structures they give very excellent sensitivity and very good
compatibility with MOS technology. So let us discuss what a MOS structure is and how it
detects the radiation. So a typical MOS structure is something we have let us say we see
the cross section of a MOSFET. This is let us say n-type silicon and in this we have
source and drain region which is p-type in this case and we have a thick gate oxide on top
of it which act like a gate for this MOSFET. So let us say this is that thick gate oxide.
This terminal is source and this terminal is drain and this terminal is gate terminal. Now
this is a typical MOS structure and if there is a radiation which is falling on this structure
let us say the radiation is coming from somewhere.

This is radiation. What this radiation is going to do it is going to create electron hole pair
in the oxide which is a insulator. This oxide which is a insulator means it does not
provide conduction to charge particles, but because of this radiation there is going to be
generation of electron hole pair. Let us say this positive this side negative separated to
other side. So because of this radiation now this radiation is creating electron hole pair in
the oxide. Now the oxide does not conduct the electricity. So whatever is the charges
which are generated by this radiation they are trapped in the gate oxide. So gate oxide
which was earlier neutral without the radiation when we shine the radiation it start
containing a fixed amount of charge within itself. Now as we know the operation of this
MOSFET structure the VT or the threshold voltage of this MOSFET is strongly
dependent on the gate oxide parameters how thick the gate oxide is what material the gate
oxide is whether it has certain charge trapped in it. So when there is a radiation and the
charge is produced by this radiation which is trapped now in the oxide so the threshold
voltage of this MOSFET changes compared to the non-radiated case. This change in the
threshold voltage can be easily detected by IV measurement or the current versus voltage
measurement that will show a shift in the threshold voltage and that threshold voltage

248
shift is directly proportional to how many charge particles are there trapped in the oxide
layer.

And these number of charge particles which are trapped in the oxide they are also
proportional to how much dose of radiation and how much energetic the radiation is
based on that radiation property these charge particles are proportional to that. So we can
use this MOSFET structure as a radiation sensor where this gate oxide is the primary
sensor where we are trapping the charge based on the radiation and this trapping or
retrapping of the charge it changes the threshold voltage of a complete MOSFET and we
measure IV characteristic and measure how much shift is there this how much shift is
now proportional to the amount of radiation received by this MOSFET. So we have now
the influence of radiation. MOSFET on MOS characteristic depends on both the dose and
the parameters of device structure including the oxide thickness. And the effect of
radiation on CV characteristic or IV characteristic can be seen. On CV characteristics is
seen as a flat band voltage shift towards negative gate voltages then exposed to gamma
rays or to other ionization or to other ionizing radiation. And this is due to the formation
of positively charged trap centers. This happens due to the formation of positively
charged trap centers in the oxide or silicon SiO2 interface. So as we discussed we use this
MOSFET device for radiation detection and radiation is creating these charged particles
in the oxide and because the oxide is non-conductive it is an insulator it does not allow
these generated charges to escape to some other place and these charges are trapped in
the gate oxide only.

Because of this trapping of these charged particles now the threshold voltage of this
MOSFET is changed and to measure a threshold voltage shift in a MOSFET either we
can use IV characteristic the current versus voltage characteristic or we can do a CV
characteristic where we plot the capacitance gate capacitance versus the applied voltage
and this CV characteristic the capacitance voltage characteristic this shows the shift in the
kind of negative gate voltages and this shift is proportional to how much is the threshold
voltage change. So this shift is proportional to the amount of radiation detected by the
MOSFET. So this is the basic structure of solid state detectors this is how they work.
Now let's see what is the effect of multiple parameters for example the oxide thickness
which is the very critical parameter now for these detectors. Let's see how the thickness
of these oxide changes the performance of these solid state detectors.

So the dependence on oxide thickness. So what happens in the structure the radiation
induced charges a shift in the position of flat band voltage of CV curve along the voltage
axis. So if we plot a typical change in the flat band voltage versus the thickness. So let's
say on the Y axis I plot change in flat band voltage and on the X axis I have thickness of
gate oxide typically in nanometer this is 0 and typically 100 nanometer 200 and 300 and
so on and this is let's say 0.4, 0.8 like that. We can have different graphs for different
radiation. For example one graph is like this. This is for typically 2 gray of radiation. The

249
other graph can be something like this which is your 16 gray of radiation. The third graph
can be like 64 gray and like this. So this is let's say 100 gray or so. So this is though the
numbers are not important the trend is there when we receive the more radiation there is a
more change in the flat band voltage and this can be easily detected by calculating how
much charge is there in the gate oxide with respect to thickness. So if we plot the other
graph where we plot the charge versus thickness and because we know the charge which
is Q is equal to epsilon naught A upon B into delta B. So this is nothing but the
capacitance of gate epsilon naught kA upon D.

And Q equal to C this is this full thing is C and delta V. So the Q is proportional to now
all these factors and if I plot Q on the Y axis in coulomb and X axis in thickness again the
axis is same as the first graph. The two gray kind of graph looks something like this and
there is a significant change in the total charge which we can easily detect using electrical
multimeter and some other instruments. So these are some typical trend in the gate oxide
charge with respect to the thickness of the gate oxide. So we can see that there is a
measurable change in the charge when we change the thickness of gate oxide and
according to the radiation received which is 2 gray or 16 gray or 64 gray we get different
different amount of charge trapped in the gate oxide.

Attributed to trapping of holes generated by radiation. And the electrons generated during
ionization enter into the metal contact or substrate leaving behind a hole in the oxide. So
these are the dependency of the change in the flat band and the charge trapped in the
oxide with respect to the thickness of gate oxide which is SiO2 and we have significant
change in this flat band voltage if we change the thickness of the MOSFET and we see
for 2 gray and 10 gray and 100 gray kind of radiation that this change is very noticeable.
The shift, which is actually in the negative direction, so this shift is generally attributed to
the trapping of holes because depending upon the shift whether it is in the right direction
or the left direction in the graph we can estimate whether it is because of the electrons or
because of the holes. In this particular case of this radiation measurement the shift in the
graph always suggests that this is because of the trapping of holes in the oxide. So the
theory is that the electrons which is generated because when the radiation hit the oxide it
generates equal amount of electron and hole there.

So, the hole get trapped in the oxide and the electron which is generated this disappear
either into the metal contacts we have connected to the gate or it is annihilated in the
substrate and leaving behind a hole which is trapped in the oxide because of this trapping
there is a shift in the threshold voltage or the change in the flat band voltage MOSFET.
So today we discussed the scintillation detectors and the solid-state detectors which has
superior sensitivity compared to the earlier detectors. So, this is all for today.

Thank you.

250
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 30
Smart Sensors: Introduction and Architecture

Hello, welcome to the course Transducers for Instrumentation. Today we are going to
discuss about the smart sensors. The smart sensor is not a very different sensor. In fact,
the main sensing element in the smart sensors is one of the sensors what we have already
discussed. For example, optical sensor, magnetic sensor or thermal sensor. All these
sensors are there as a sensing element in the smart sensors. The only difference between
those sensors and the smart sensor is we put some amount of intelligence in the system
itself so that when we are sensing the environment using these traditional sensors and we
get the data from this sensor output, that data itself is processed within that sensor itself.
So the sensor has a one part which is the traditional sensing and the data generated by this
traditional sensing is processed using some extra electronics attached to it. So the whole
system now it is called the smart sensor. It has certain intelligence, it can have memory, it
can have some algorithms which can process this data and the refined data or the
processed data is sent for further processing or for further visualization. So this is called
smart sensor. So we have smart sensor here. So this smart sensor is a device that takes
input from the physical environment. And uses built-in computer resources. To perform
redefined functions. So smart sensor is a device which takes the input from environment
using some traditional sensor. Let us say we are talking about a thermal sensor now
which is a smart thermal sensor. So the basic element of sensing is still that thermal
sensor what we have discussed earlier. This traditional sensor is going to measure how
much is the thermal energy and based on that it is going to convert this information into
an electrical signal. Now this electrical signal is further processed by external circuitry
which is inbuilt in this smart sensor.

This extra circuit is going to compute on this data which is electrical data coming in. This
electrical data will be processed using this electronic circuitry to perform predefined
function. For example, it can have multiple stages. For example, the data comes with
some noise that need to be cancelled out. The amount of signal is very weak so it need to
be amplified and then this data will be converted into a digital domain because the
processing of data in digital domain is very easy. So we perform something called analog
to digital conversion and these A to D are there in this smart sensor itself. They convert
this data from analog to digital and then this data, this digital data is further processed
using algorithms. This algorithm is also part of extra electronics which we put alongside
the sensor. So this whole system which can process the data also and transfer it to some
other location, this whole system is called the smart sensor. So this sensor, the smart

251
sensor is also crucial and integrated element, an integral element of something called
Internet of Things. Or sometime it is called IoT in short. These IoT technologies makes it
possible to do this or about those things over the internet. Or a similar sensor network. So
these smart sensors, these are very crucial and integral part of IoT or something called
Internet of Things which is a very new, very recent concept of putting sensor nodes. So
the explanation of IoT is for example we have a very large field and we want to measure
the moisture of the soil using some sensors.

So one way is we put multiple traditional sensors like humidity sensor or soil moisture
sensor each and every point in that large field, every 10 meter or so. So the number of
sensor we need to put will be like 100 or even more sensors we need at every interval.
These sensors will generate their own data. Each sensor will have this moisture to
electrical conversion and this electrical data need to be processed. So now it has multiple
requirements. For example the very first is the soil moisture does not change very rapidly.
It takes time in changing. So we are not worried very much about the real time or the
instantaneous data. So what is the moisture now? It's going to remain almost in the
similar range for quite some time, let's say one hour or two hours. So we do not need this
data to be transmitted every time. So the sensors which are hundreds of sensors which we
need to put in that field, they need not to be sending data all the time for processing.
Sometimes we know the data that remains in the same range for over a certain amount of
time. So this is one crucial aspect of these sensors and smart sensors. So we put the
electronic circuitry in such a way that the data which it is not measured continuously, it is
measured in after certain interval of time. And the IoT technology makes it possible to
transmit data from or about these things over the internet. So this hundreds of sensors
which we have put together it's called sensor network and the sensors, individual sensor
is called sensor node. This sensor node can transmit this data, this moisture or humidity
data directly to the internet which we can see in our remote location or it can send its data
to neighboring sensor node so that this data hops on from one node to another one and it
reaches its destination. So there are two ways of sending this data by these smart sensors
over the internet directly or to similar sensor nodes or sensor networks. So this is another
way of transmitting this data. So next we have low power and low cost mobile
processors.

Typically provide computation resource, IoT environment and a sensor that just sends its
data. For remote processing is not considered a smart sensor. So as we have seen this
example where we have a large field and we put hundreds of sensors or sensor nodes in
that field to measure the moisture of soil. Now because we need to put large number of
sensor nodes in a given field, so there are two constraints which comes up. The very first
is the low cost. These sensor nodes in the individual sensors need to be very low in cost
so that we can deploy in hundreds of numbers. The next is the low power. So when we
are putting these sensors which are smart sensors now and extra electronics is also put

252
along with them for doing all these processing. This extra electronics as well as the
sensor now they require power supply to perform the processing in that function. This
power supply is generally in the form of built in battery. We need to provide a certain
amount of battery in the sensor node itself because if we put hundreds of sensors in a
field there is no way we can power them individually from a power socket. So they need
to be self sufficient in power. So the power supply should be adequate for this sensor to
run for a long time. It is not even possible to replace the battery or charge the battery
because we have hundreds of sensor in the field. So low power and low cost sensors need
to be there and generally we put a microprocessor or a very small processor which is
good enough to process the signal generated by these sensors and this is the processor
which provide computational resource in the IoT environment.

And a sensor which just receives the environment which sends the environment and
convert it into electrical domain and just send it to the remote location processing. It does
not process any data within inside. It is not called smart sensor. Smart sensor is the one
which accumulates the data and process the data using built in algorithms and then
transfer it to the remote location. That is a smart sensor. If it does not process any data
within itself then this is not considered as a smart sensor. So let us look at the block
diagram of a smart sensor. So in the all the sensor smart sensors we have a traditional
sensor which is the sensing unit. This is the traditional sensor which is sensing the
environment and generating the electrical signal. This signal actually received by signal
conditioning unit. Then the bus is sent to another unit which is analog to digital
conversion. After this the data is sent for application of algorithm. This application of
algorithm phase can have simultaneous input from the memory which is the inbuilt
memory in the smart sensor and the user interface. So once this algorithm application is
over then the data is sent to communication or transceiver. So this is a typical block
diagram of a smart sensor. Of course the number of blocks can change depending upon
the application but a general smart sensor contains these kind of blocks. The very first is
the sensing unit which is our traditional sensor. For example the moisture sensor in this
case what the example we have taken. So the moisture sensor is nothing but which is
converting the moisture or the humidity or the water content into a electrical signal. Now
this electrical signal goes from sensing unit to signal conditioning.

Signal conditioning is the unit which process this signal so that it is appropriate to apply
or to transfer to electronics because the signal generally we receive from any sensor is not
in the desired format or is not kind of suitable to directly transfer it to electronics. For
example the noise, the generated signal generally has noise and we need to filter it out
before sending it for processing. Another important point is the amplification. If the
signal generated by the sensor is very small in magnitude it will not be easily processed.
So signal conditioning unit does all these functions. It amplifies, it filters out noise and do
other things as well and make the signal appropriate to apply for other process, other

253
blocks which is further electronics. When the signal conditioning sends the data to A to D
converters or analog to digital converter, this unit converts this analog signal into digital
signal because most of the time when the signal is generated from traditional sensors the
signal is always analog. Most of the time it is analog in nature and processing analog
signal is difficult compared to digital signal. We have all these digital processors with us.
So we can process a digital data very easily compared to analog.

So it is required to convert this analog signal into a digital signal using this analog to
digital converter. Once the data is converted into digital this is sent to this application
algorithm block where this data, this digital data now we process it using built-in
algorithms. This algorithm and all other previous data is stored generally in memory. So
we have the second block attached to this block which is memory which is in direct
communication to this block. It can receive and send the data to memory and using these
algorithms the processing of data is happening. At the same time some smart sensors can
have a display as well which is a user interface. If I want to intervene with these smart
sensor processing I also have an option to change the things by myself or by manual
intervention. So for that there is a user interface. It may or may not be present depending
upon the case what we want to make. But there can be a user interface to control this
algorithm which is applied on this data. So once algorithm is applied and the data is
processed it can be in the multiple stages, may not be a single block, may be multiple
blocks. Once the data is ready then we need to send it to remote location. It may be
within the sensor network or it may be over the internet. That is done using this
communication transceiver. Transceiver is nothing but transfer transmitter and receiver.

So this block can send the signal as well as receive the signal from outside world and this
is the communication block which is used to transfer the data from sensor node to sensor
network or to internet. So this is the complete block diagram of a smart sensor. However
depending upon the application few of the blocks cannot be there or few extra blocks can
be there in this block diagram. So we have here. The first is the sensor and actuator
hardware. The second is the conditioning circuitry. Then we have microcontroller and we
do multiple processing in this. Let us say signal filtering in this stage can also be done. So
these are the few building blocks of a smart sensor. Now let us see the some advantage
and disadvantage of a smart sensor. So the very first advantage of smart sensors is energy
efficiency and sustainability. Second is access to data in different environments. In some
difficult environment. High performance. Next advantage is the built-in analytics and
processing. This also offers self calibration and self assessments and it can also detect
erroneous data.

So these are some advantages of smart sensors where the energy efficiency and
sustainability. How the smart sensors are energy efficient because as we discussed in the
example the soil moisture it does not change very much with time. Only in the longer
interval it changes. It does not change very frequently. Maybe in one minute or two

254
minute it does not change. Whatever is the soil moisture is there it remains in that range
only for quite long time. So if we put a normal sensor without any intelligence this sensor
will keep measuring the environment the moisture and keep sending the data to outside
world which is very energy hungry. If we put a smart sensor however the smart sensor
knows that this sensing need to be done in every one hour or every ten hour. So for one
hour or ten hour this sensor node will go into an ideal mode which will not be burning the
power from the battery. So in that way the smart sensors are energy efficient. Next is the
access to data in difficult environment. Some environments are there has some harsh
environments where it is not possible to measure this data using traditional sensors. So
we use these advanced smart sensors to measure those environments. High performance
these sensors can send this data very much efficiently and compute on the data. Built in
analytics and processing. So we have extra hardware and extra electronics
microprocessors in place. These processors will keep on applying some analytics on the
data and processing on this data and after the processed data will be sent to the remote
location for visualization. So this is the advantage of smart sensor. Self calibration and
self assessment.

This is also very good advantage of smart sensor. The self calibration is something all the
sensors all sensor what we put in the environment they deviate from normal behavior
with time. So if it is a normal sensor if there is a drag in the sensor output that we will not
be able to detect at the far end when we receive the data. However these smart sensors
they have built in algorithm and they can keep checking the data when the data is
continuously drifting towards one direction then they can detect it as a problem of
calibration and they can calibrate the sensor itself so that the sensor detects the accurate
reading without any drift or drag. So this is another advantage of smart sensors and it can
also detect erroneous data. So for any reason if the conventional sensor is failing it is not
giving accurate readings and the readings are way off than its normal value. This can be
easily detected within the node itself when the data is coming from sensor and processor
is detecting whether this data is not within that range. So instead of sending this data this
erroneous data to remote location this data will be corrected within itself or the sensor
will be recalibrated so that it will start receiving the meaningful data. So this is the
advantage that sensor nodes can detect these erroneous data. So let's see some of the
disadvantage now of these smart sensors. These sensor nodes are susceptible to
tampering and hacking. The higher cost because of extra electronics we need to put here.
Silence, battery lifetime, expertise. And meeting requirement for IoT development. So
these are some disadvantages of smart sensors. The very first disadvantage of smart
sensor is they are susceptible to tampering and hacking because we have extra electronics
in place which is continuously monitoring the data and this processing is happening
because of this microprocessors we put which has certain memory and we put algorithms
in that which is processing this data.

255
Now with intelligent systems these sensor nodes are susceptible to hacking. One can
place or it is possible to change the algorithm itself so that this sensor node will start
sending different data than what it should be sending. So this is susceptible to tampering
and hacking. Another disadvantage is the higher cost because these microprocessors what
we put there along with the built-in battery and some other electronic circuitry this
increase the cost of overall system which is a very crucial factor for large scale
application because in a field let us say we want to put 100 sensor nodes even a small
increase in the cost of one sensor node that get multiplied by 100 number of sensors so
the overall cost increases very much. This is the disadvantage. Maintenance sometime is
difficult for these smart sensors though it is case to case basis whether the sensor needs
maintenance or not but this is also one of the disadvantage. Battery lifetime is also one
disadvantage because we are putting extra amount of electronics which is also driven by
the battery itself. So here in this case battery lifetime reduces depending upon the extra
electronics and transfer of the data and other factors. So battery lifetime is shorter which
is a disadvantage for smart sensors.

Expertise is also a problem. So these sensors need human expertise of putting algorithms
and connecting it interfacing it to the traditional sensor sending data. So all these
expertise is needed which is a kind of a maybe a bottleneck for deployment and meeting
requirement for IoT developments. So IoT is nothing but arrangement of the sensor node
in a particular fashion and sending this data over the internet. So to do this all these
sensor nodes need to be following certain rules or regulations imposed by IoT system. So
meeting those specification is also sometimes challenging for these smart sensors. So
these are some advantage and disadvantages of these sensor nodes or the smart sensors.
Now all these smart sensors as we discussed have traditional sensing which is then just
like our thermal sensor or optical sensor in a normal way which converts let us say
temperature to electrical signal. Now this electrical signal is processed using some extra
electronics or some processors. So there are two kinds of processors which are very
widely used in the world. The very first one is the Arduino which is a very low cost
microcontroller and the other one is the Raspberry Pi which is a more powerful version of
microprocessor which can do multiple computation. So these are two types of
microcontroller and microprocessor used for tensor node development. So these are
shown here. These are the images of Arduino and Raspberry. So Arduino on the left
which is this blue color PCB. This has certain ICs placed inside and this Arduino is built
up by this Arduino team and Raspberry Pi is on the left on the right side where we have a
strong microprocessor built in and it can do multiple kind of applications together.

So let us discuss what is the difference between these Arduino and Raspberry. So one
side we write Arduino and the other is Raspberry or Raspberry Pi it is called. So Arduino
is a development circuit board. It is based on microcontrollers. It does not have operating
system. It only does only one task at a time. If we want to add some extra features then

256
we need to use some extra Arduino shields. It can be programmed in CE or CPP. CE
language or C++ language. The logic level, the voltage is 5 volt for this case. It needs
additional Arduino shields to connect to internet. And there is no wireless connectivity on
the board. This has open source hardware and software. And this is not very costly. So
these are some points about the Arduino. Arduino circuit board is nothing but a
development circuit board. This comes as a development circuit board and this is based
on the microcontroller. It is not the microprocessor. It is a microcontroller. It means the
data comes in and it will be processed using this microcontroller and the output will be
generated in real time.

So this is a microcontroller based board and there is no operating system inside the
Arduino. Whatever the algorithm or the data we put that we put in a memory and there is
no further updation of this algorithm by itself or using the external data. It is just the
algorithm with what we developed and put in the Arduino that is going to process now all
the data from sensor. So there is no operating system in Arduino and it usually does only
one task at a time. So the signal comes in and it will be processed using Arduino and the
output signal will be generated. There is no multiple algorithms which will be running at
a time to process the data. It is only one algorithm which will be running in sequence.
Usually Arduino needs extra hardware to add certain other features. For example, we
want another feature of like display or something or to kind of transfer the data using
Bluetooth. For that we need extra hardware to put on Arduino. Standalone it is not there.
Arduino can be programmed in C language or C++ language. The logic level is 5 volt
means the voltage needed to turn on the Arduino and all the input output signal they all 5
volt in the electrical voltage. So this is very much compatible to very legacy kind of
sensors which still work in 5 volt domain. And it needs additional hardware to connect to
internet.

So there is no Wi-Fi or internet transceiver or chip already in place in Arduino. So it

cannot send the data to us to the internet. We need extra hardware for Arduino to do this
function. No wireless connectivity also on the board. So there is no Wi-Fi chip or
Bluetooth chip on the board which can send the data. Anything we need we need to put a
separate hardware on top of Arduino which is generally available in the market. This is a
open source hardware and software structure. It means anyone can update this hardware
or upgrade this software based on the requirement and there is no restriction by the
manufacturer. So this is a open source hardware and generally Arduino is not very costly
which is very suitable for these sensor node application. So these are some facts about the
Arduino. Let us discuss the second very popular microprocessor now which is Raspberry
Pi. So here we have Raspberry Pi which is a single board computer. This is based on a
microprocessor and it has a operating system which is Raspberry Pi OS which is a Linux
based operating system. It can perform multiple tasks simultaneously. Hardware can be
attached or extra features can be programmed in multiple language in Python, C++ etc.

257
The logic level is 3.3 volt. Can connect to internet using Wi-Fi or Ethernet. Wi-Fi and
Bluetooth on the board. It is a not open source system and it has a operating system
which is Raspberry Pi OS and which is Linux based OS. The hardware can be attached
for extra features though Raspberry Pi has lot of features on the board itself but for any
extra feature we can always put a separate hardware and connect it to Raspberry Pi. This
can be programmed in multiple language for example Python, C, C++ or even Scratch.
There are multiple programming language which can support Raspberry Pi. The logic
voltage level is 3.3 volt which is different than Arduino. So Raspberry Pi work on a lower
voltage it means the power consumption of Raspberry Pi can be limited lesser than
Arduino but Raspberry Pi has so many functionality on the board and the speed of
operation is higher hence the Raspberry Pi overall consumes more power generally than
Arduino. Raspberry can connect to internet using Wi-Fi or Ethernet. It already has this
Wi-Fi and Bluetooth chip on board. Raspberry Pi is not an open source platform means
we do not have functionality of we do not have flexibility of changing the Raspberry Pi
board whatever it comes from the manufacturer we have to use in the same way. And the
Raspberry Pi is generally expensive compared to Arduino it is like 3 or even depending
upon the version of the board it can be 5 to 10 times expensive than Arduino. So these are
the basic points about Arduino and Raspberry Pi. These are the microcontrollers and
microprocessor generally used for making smart sensor nodes.

So, this is all for today.

Thank you.

258
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 31
Smart Sensors: Microcontroller based.

Hello, welcome to the course Transducers for Instrumentation. We are discussing smart
sensors and last lecture we discussed that smart sensor is nothing but the conventional
sensor which is now interfaced with external electronic circuitry, microprocessor and
microcontrollers which actually processing this data generated by conventional sensor
and the processing is done at locally itself making it as a sensor node and this data which
is already processed is sent to remote location to some other sensor mesh or over the
internet. This smart sensor now they have this external circuitry which is based on
microcontrollers or microprocessor. There are two types and we discussed one
microcontroller which is very much used is Arduino and the microprocessor which
widely used is the Raspberry Pi or R Pi we call it. So today we are going to discuss this
microcontroller based smart sensors and we use Arduino generally for this application.
So Arduino is a open source hardware it means you can modify the hardware itself as per
your own requirements and the software which is used to program these Arduinos we
have to implement certain algorithm or so that is also a open source you can modify you
can change it you can use it any form to suit your applications.

So Arduino is a open source platform most of the time used with for a low cost sensors
because the cost of Arduino is very less compared to microcontroller based hardware
such as R Pi and there are plenty of interfaces board available for example in the Arduino
itself we do not have much of a functionality but external hardware's are available which
you can use as per your requirement. For example you need a pressure sensor so these
pressure sensors are available as a standalone unit. You can buy it and plug it with the
Arduino to measure the pressure. Generally other types of hardware's are available for
these Arduino boards. It is an open source and very flexible easy to use hardware and the
software programming is also not very complex this can sense the environment by
making use of the external sensor circuitry that may be a conventional sensor or may be a
ready made hardware available in the market and generally on off light or there are many
kind of motor and actuator hardware's you can make using Arduino.

So let us discuss today what we can how we can use a Arduino board to make a smart
sensor. So we discussed microcontroller based smart sensor. Arduino has become the
standard microcontroller. And researchers to its ease of use, low cost and plenty of
interface boards. This is an open-source electronics. Flexible, easy to use hardware and
software. This Arduino can sense the environment by receiving input from a variety of
sensors. It can affect its surroundings. By controlling light, motor or power. Or other

259
actuators. This microcontroller on the board is programmed using, number one is the
Arduino programming language. And based on the wiring. So here we have this
microcontroller Arduino. This is an open source electronics which can sense the
environment by receiving the inputs from a variety of sensors. You can attach a
temperature sensor, pressure sensor or many other kind of sensors which are readily
available in market or you can develop these sensors in house and you can hardware's
you can make using Arduino. So let us discuss today what we can how we can use a
Arduino board to make a smart sensor. So we discuss microcontroller based smart sensor.
Arduino has become the standard microcontroller. And researchers to its ease of use, low
cost and plenty of interface boards. This is an open source electronics.

Flexible, easy to use hardware and software. This Arduino can sense the environment by
receiving input from a variety of sensors. It can affect its surroundings. By controlling
light, motor or power. Or other actuators. This microcontroller on the board is
programmed using, number one is the Arduino programming language. And based on the
wiring. So here we have this microcontroller Arduino. This is an open source electronics
which can sense the environment by receiving the inputs from a variety of sensors. You
can attach a temperature sensor, pressure sensor or many other kind of sensors which are
readily available in market or you can develop these sensors in house and you can
connect directly to these microcontroller which is very easy to program and interface
with the sensor. It can affects its surrounding by controlling certain input output pins
which you can use to actuate the other hardware. And this microcontroller you can
program it in two parts. One is the hardware circuitry which pin you connect to which
sensor. This is one way of making a complete circuit and then based on this wiring you
can program the microcontroller which is sitting in this Arduino microcontroller. There is
a microcontroller placed on this board that can be programmed using a programming
language which is open source again for this Arduino and the programs generally we
write for Arduino they are called sketches. So these are the basic features of this
microcontroller. Let's discuss how is the power compatibility of these Arduino boards. So
the power pins are as follows. So we have let's say a Arduino board. In this Arduino
board generally we have a USB connector.

We have a power connector as well. And then we have multiple pins on the side. And
some of these pins are 3.3 volt and some are 5 volt. So some of the pins are 3.3 volt
output. Some pins are 5 volt output. And some pins are input signals. So here we have
this Arduino board which has a USB connector where you can connect a USB cable
coming from your computer. Directly you can connect this Arduino to your computer
using this USB cable and the power will be drawn from this USB connector itself which
is 5 volt power supply. So the Arduino board runs on 5 volt power supply and all the
other pins are 5 volt logic level. Some pins are 3.3 volt output. This 3.3 volt is generated
on board in Arduino using external power circuitry. This 3.3 volt output you can use as a

260
power supply to a very small sensors if the need arises otherwise all the pins and the logic
levels in Arduino are 5 volt logic level. So the 3.3 volt signal which is coming in this
need to be step up to 5 volt so that it is compatible with Arduino board. So all the
hardware circuitry which we need to use for Arduino should be 5 volt range. So the V in
supply, the input voltage to the Arduino. Vold when it is using an external power source.

We can supply voltage. Through this pin. The next is 5 volt in the regulated power supply
to power. The microcontroller and other components. This can come either from V in via
regulator or from the power supply or supplied by USB or another regulated 5 volt
supply. The 3.3 volt in a 3.3 volt supply. generated by on board chip. The maximum
current drawn is 50 milli amp and GND is the ground pins. So these are some details
about the Arduino hardware where we have multiple pins and the Arduino as a
microcontroller works at 5 volt supplies. Though it has a regulator on chip which can
convert 5 volt to 3.3 volt and this 3.3 volt supply is available on board and can be used to
interface some of the sensors. However, the maximum drawn current is only 50 milli amp
because the regulator is limited by this current only. So we cannot draw a high current
from this pin or we cannot attach a heavy load directly to this pin which is running at 3.3
volt. If the need arises to connect a heavy load we have to use certain buffer circuitries
which boost up this signal.

So this 3.3 volt supply is all present and ground is the universal ground which should be
connected to whole of the external circuitries and 5 volt supply is always available on this
Arduino. It can be derived directly from the USB which we are connecting it to connect
to the computer or from a power connector which we connect through a adapter or it can
be directly given as a 5 volt to this volt. So these are some basics of these power pins of
Arduino. Let us look at a little bit about the basic programming of Arduino. How do we
do this programming for Arduino? So the basic programming. We need to declare all the
variables at the top. Generally, we have two loops, one is the setup and one is the loop.
These are the two statements which are there in all Arduino sketches. That are a part of
every Arduino sketch.

The first one is the setup. It is called once when the sketch starts. It is a place to do setup
task like setting the pin mode or initializing libraries. For example, this is how a setup a
sketch starts. We define a void setup and this loop starts and then we write our statements
and then this loop closes. So this is how a sketch starts. The second function is the loop
function. And the important part of every sketch. And we need to include both functions
in the sketch. Even if we do not need them for anything. This is how this void loop
function is. We write void loop function. It starts and then we write our statements and
we close the bracket. So here we have two basic functions. One is the setup, one is the
loop. Both are important and we need to include them in our program or sketch. Even if
we do not use them but we still need to use both of them setup and loop.

261
The difference between setup and loop is the setup is called only once and this is the
program where the sketch will start and end and within this function all the programming
will happen. Now if we need a repetitive measurement of programming that is done in the
loop function. This function is written separately. This loop function is for example we
have a sensor and this is a temperature sensor. We want to measure the temperature every
one minute. So setup is the program which is setting up all the parameters. For example
the pin numbers with which the sensor is connected. The libraries whatever we use for
math or for sensing. So that is done in the setup itself. Then every second, every minute
we need to take the measurement of temperature. This is a repetitive process. So this
measurement is actually done in the loop statement. Every one minute setup will call the
loop function and when the loop function is called it will perform the measurement of
temperature from a sensor and gives it reading back to the setup function where it will
calculate all the output and then wait for one minute or so and again call the loop
function. So setup function is the one which is called only once and the program execute
in there and the loop function is a repetitive process which we need to perform
repetitively in the program. So there are two main functions.

First is the variables. We have certain variables in the programming language to perform
the task. So let us discuss variables. You can use variables in a similar way as they are
used in math or physics or even any programming language. As they are used in math or
physics. Or in any other programming language. All variables have to be declared before
they are used. And optionally set an initial value. Means initializing the variables. So the
example is. int LED pin equal to 13. The other example is. integer var equal to value. So
we have certain variables in the programming language. And these variables for Arduino
it can be in a similar way that they are used in math or physics or even any other
programming language. Similar way we can use in Arduino. All variables need to be
declared first before we use them in the program. However this is optional to put a initial
value or initializing the variable. So for example we have int LED pin equal to 13. So
here we are initializing a variable which is LED pin. This is the variable name. And the
type of variable is given here int which is integer for short. So LED pin is a variable
which is of type integer. And the value we are assigning to this is equal to 13. Means
LED pin wherever we will call in the program will represent a value of 13 assigned to it.
In another example we have a var which is the variable name which is type integer again.

And the value is equal to val the value. This val we can declare later on or by successive
measurements it can come up. So here we are not initializing the variable by its value but
we have to declare the variable and its type which is integer here. So the basic
programming is the basic programming includes multiple functions such as the digital
write function includes a value on a pin. The possible values are low or 0 volt or high
which is 5 volt. For example, digital write 13, high and digital write 11. So here we have
the basic programming which has a function like digital write. Digital write is a function

262
which outputs the value to a particular pin. We have to specify the pin number in digital
write as well as the value. The values are possible of two types only. One is the low
which is 0 volt and the other one is high which is 5 volt because the Arduino runs at 5
volt so all the digital high is 5 volt. Now the example of this digital write is digital write
13, high means digital write will enable this pin 13 which is the pin number of in that
Arduino board. The pin number 13 will be applied a digital high which is 5 volt so the
pin number 13 will have a voltage of 5 volt. In the other example digital write 11, low the
pin number 11 will be pulled down to 0 volt because this is specified as low voltage. The
next is the delay function. The delay causes the Arduino to wait for the specified number
of milliseconds. Before continuing on the next line. So, for example. Delay 1000. This
statement creates a delay of 1 second. So we have the another function which is very
useful is the delay function. This function whenever it is called it delays the programming
of Arduino for that number whatever is written in inside this braces. The example is delay
1000 means whatever is the number given in this brackets that number of milliseconds
the program halts it waits for these many milliseconds before going to the next line.

So delay 1000 means 1000 milliseconds it will wait to go to the next line. 1000
milliseconds means is the 1 second time. So after 1 second the next line will be executed.
So these are some basic functions we will not cover all the programming how to do
Arduino programming and all the sensors connected to it that is a separate course by
itself. So we will just focus on basic understanding of Arduino and how it works and
what is the basic functionalities. There are lots of other functions as well which are
equally important for Arduino. So other useful functions for example the pin mode. This
set a pin as input or output. Digital write. That is a digital pin high or low. The third is
digital read. This read digital pins state. Next is analog read. This reads an analog pin.
Analog write. This write an analog PWM value. The other is delay. This waits for an
amount of time in milliseconds. The other function is millis which gets the current time.
So there are some other functions which are useful for making a Arduino sketch. The one
is the pin mode. This sets the pin as input or output. So as we know Arduino has multiple
pins which can be configured as input or output. We can use the same pin to receive the
signal as well as to send the signal. So pin mode is the function which is used to set the
pin as input or output. If we set the pin as input the external voltage whatever is applied
to that pin that will be detected by Arduino. If we set the pin as output then the value will
be written by Arduino on that pin and the external circuit is going to receive that value.

So pin mode is the function to assign the value whether it is the pin will be used as a
input pin or the output pin. Digital write is the function which is used to enable that pin
high or low. So when we write digital write high it will apply a 5 volt signal and when we
write digital write low it means 0 volt will be applied through that pin. Similarly digital
read is a function which is used to read the value from that pin if that pin is assigned as a
receiver or the input type then the external circuit is applying a voltage on this pin and

263
Arduino is going to read whether this is a 0 volt applied from the external circuit or it is a
5 volt applied on that pin. So these are digital write and digital read means the value can
be only 0 or 5 Volt. Similarly we have analog read. Analog read is the function which is
used to read the analog value because the sensor which is generally connected to the
Arduino they give the output as a analog signal. Analog signal not necessarily need to
have 0 or only 5 volt value. The value or the voltage can be in between as well so that is a
analog signal. Analog read is the function used to detect how much is the voltage at the
pin supplied by the sensor and this voltage will be used for further calculations or the
programming.

So analog read is to use to read the analog signal from that pin. Similarly analog write is
the function which is used to apply analog value to the pin. Delay as we just discussed
last slide. Delay is a function which waits for the program for these many milliseconds
whatever we specified in that bracket. For example, if we specified delay 10 the program
is going to halt or to wait for 10 milliseconds before going to the next line. Delay 1000
means the 1000 milliseconds or the value is 1 second. The program will wait for 1 second
before going to the next level. Millie's is the function which is used to get the current
value or the current time in the program. So these are some basic functions which are
used for programming in the Arduino sketches. Let us see a program how a program in
Arduino look like and how do we make a small programming Arduino sketch or Arduino
IDE.

So we take a very small example of traffic light control. So in this program we first
initialize some variables which is int. LED red is 13. integer LED green is equal to 11
and integer type LED yellow is equal to 12. So these are the three variables we are
defining LED red is equal to 13. These are nothing but the pin numbers what we are
going to use. The pin number 13 will be connected to the red light. Pin number 11 will be
connected to the green light and pin number 12 will be connected to the yellow light. So
we have these three lights for traffic light control. Now we write the setup loop on the
setup function void setup and in this we declare pin mode LED red, output because we
need to control all the pins from Arduino so all the three pins will be in the output mode.
And here we close the bracket. So this is the setup statement in the program. Before this
we have defined three variables LED red which is assigned the value 13, LED green
which is integer type variable assigned to value 11 and LED yellow which is again an
integer type initialized to value 12. Now we need to have two functions in the Arduino
sketch. One is the setup function and one is the loop function. Setup function is the
function which we generally used only for once to set up all the constraints in the
program. Here the setup the constraints we are making pin mode LED is output. So the
pin mode what the pin the pin number 13 is used as output pin for throughout this
program pin number 11 which is LED green this is also output type pin and yellow which
is connected to the yellow pin this is pin is also a output type. All the three pins are of

264
output type. Next is the loop statement. And we write in the same program void loop and
we write digital write LED green void comma high. Then we put a delay of 5000
milliseconds it means 5 seconds. Then we write digital write LED green low and digital
write LED green low. LED yellow as high. Then we put a delay of further 5000
milliseconds and again write digital write LED yellow as low and digital write LED red
as high.

Then we further put delay of 5000. And then we write digital write LED red as low. And
then we close the brackets. So this is the void loop which will be executed repeatedly and
what we are doing here the first statement is digital write LED green high. So the green
light the pin which is associated to this green light will be turned high it means the 5 volt
will be applied to this pin and this 5 volt will be applied for 5 second because the next
statement is delay 5000 milliseconds. So 5000 millisecond means the 5 second delay
program is going to wait to execute next instruction. So for 5 second the green light will
be at 5 volt or green light will be high. After 5 second digital write LED green low. So
the light will be turned low the green light the voltage applied will be 0 now and in the
next statement immediately the LED yellow high. So yellow pin the pin which is
connected to the yellow light will be applied a 5 volt signal and further this program is
going to halt for again 5 second before executing the next instruction. So the yellow light
will be turned on for 5 second. Next after 5 second the yellow light will be turned low
and the red light digital write LED red comma high the red light will be turned on high
and the program again will wait for 5 seconds. So the red light will be turned on for 5
second and then the red light will be turned low and this loop will execute itself forever.
So if we see from outside first the green light will turn on after 5 second it will turn off
the yellow light will turn on after 5 second the yellow light will turn on turn off and the
red light will turn on and again it will be remain red light for 5 second it will turn off after
this and again green light will turn on. So this is the basic structure of a Arduino program
and some basic functions we discussed which are used to make this Arduino sketch. Of
course this is not enough to make all the Arduino sketches and all functional programs
which are used for actual programming of Arduino, other sources can be used to learn the
Arduino programming.

This is all for today.

Thank you.

265
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 32
Smart Sensors: Microcontroller based

Hello, welcome to the course Transducers for Instrumentation. We are discussing about
smart sensors and we saw that smart sensor is nothing but a conventional sensor and
some extra electronics which is in the kind of microprocessor or a microcontroller. This
extra electronics is used to process the data generated by this conventional sensor locally
itself so that we extract out the meaningful data from the measured, the raw data and this
meaningful data is now next sent to the mesh or to the internet whatever is the destination
maybe. So this smart sensor we discussed in the last lecture we have a microcontroller
which is Arduino most of the time. This microcontroller can be used to receive the data
from a conventional sensor and this Arduino has multiple pins which can be configured
as input and output. So input pins if you specify a pin as a input that pin is expected to
receive data from the sensor and provide it to the microcontroller.

If you specify a pin as output pin that pin is supposed to send out some signal to control
the external circuitry that may be a sensor or maybe some interface circuitry which is
controlling the sensor itself. So, we saw microcontroller earlier and we saw that
microcontroller especially the Arduino one is very less expensive and easy to use. It is a
open source platform where you can apply your own changes to the hardware and
software itself and you can modify as per your requirement. So that is the advantage of
open-source platform where you are free to choose or free to change the hardware as per
your requirement and the Arduino is low cost and very limited in functionality by itself
but there are multiple hardware shields available in the market which can enhance the
performance of Arduino. For example there is no on-chip communication interface on
Arduino, wireless communication is not available on the Arduino board itself. So external
Bluetooth or Wi-Fi modules are available you can buy from the market and insert it into
the slot of Arduino then Arduino can send and receive data wirelessly to internet.
Similarly other hardware's for example different sensors or different relays, controllers all
these things are available for Arduino to choose as per your requirement they are easily
available very low cost and you can build up your system based on the Arduino the core
or the heart of the project is Arduino and on top of that you can put multiple shields to
achieve your specifications. So that is the case of microcontroller based smart sensor
where we have Arduino which is low cost and for large number of deployments we prefer
a open source platform which is easy to use and less expensive. The next kind of smart
sensor which use a microprocessor based approach these smart sensors again the sensing

266
unit is the conventional sensor only which is generating this raw data from the
measurement.

Now we have a microprocessor which is sitting there locally to this micro this sensor
which is processing this data and finding out the meaningful data and transferring it to the
internet or some other destination. Very famous microprocessor based hardware is the
raspberry pi or we call r pi in short. Raspberry pi is not an open source platform you
cannot change the hardware of raspberry pi by yourself and this is not that much low cost
as Arduino is this is little bit expensive platform which is a computer is a computer in
itself. Raspberry pi is limited in the performance compared to our normal
microprocessors but it can be used as a standalone microprocessor you can make a small
computer also out of this raspberry pi. So this raspberry pi is a microprocessor which has
almost all the functionality what our normal processor has in our laptop or our desktop
PC's or even most of the time in our tablets as well.

So this microprocessor based raspberry pi as we discussed is not an open source platform

we are limited by changing the hardware of this board itself and the software. However
raspberry pi is also very much popular amongst the students and researchers because this
gives us flexibility to easily make a prototype once you have a sensor made in your lab
you can easily interface it with raspberry pi and make a complete system using this smart
sensor. So let us discuss little bit about what this raspberry pi is and what is the hardware
specifications of this. So today we are discussing microprocessor based smart sensors and
we will refer raspberry pi or sometime we call r pi in short. So we take some specification
of this raspberry pi and we are referring to raspberry pi 3 model B.

So we have this as a system on a chip which is a Broadcom chip this is manufactured by

Broadcom company. There is a CPU inside which is ARM Cortex 1.2 GHz is the
frequency of operation and this is manufactured by ARM company. RAM or the random
access memory is around 1 GB which is LPDDR 2. LPDDR is the short for low power
dual data rate RAM which is 900 MHz. There is a storage on the board in the form of
micro SD card. So you can have a micro SD card installed in the board. It is also
available on the board. This communication is also there which is 2.4 GHz 802.11n
wireless standard. There is video output from the card which is HDMI format. GPIO.
There is a port 0 40 pin header. There are multiple ports available on the board. For
example we have HDMI. Camera interface etc. And the cost of this board is generally
expensive compared to Arduino based micro controller. So these are some of the features
of Raspberry Pi or R Pi which is a microprocessor based platform. It has a SOC which is
Broadcom manufactured this is the company which manufactures this IC. CPU is
manufactured by ARM cortex. ARM is the company which manufactures these CPUs.
The frequency of operation is 1.2 GHz. This frequency is little less compared to what our
laptop generally has.

267
Laptop and computers they generally have like 3.5 GHz or even more. So the frequency
of operation is 1.2 GHz of this Raspberry Pi. We are considering Raspberry Pi 3. The
GPU is again manufactured by Broadcom company. The RAM or the Random Access
Memory which is installed in this microprocessor or Raspberry Pi. This is generally 1
GHz and this is LPDDR2 which is low power DDR dual data rate. 2 is the version of this
kind of memory. It operates at 900 MHz. Storage is in the form of micro SD card. So as
such there is no memory available on the Raspberry Pi but there is a slot available to put
a micro SD card. You can put micro SD card as per your requirements for example 16
GB or 32 GB. So this Raspberry Pi supports micro SD card slot and this micro SD card
act as a memory for all the operations. So if you want to install some software or install
OS so all that operation will be performed on micro SD card. Micro SD card will be
providing all the memory into the Raspberry Pi. Ethernet this is available on board itself.
There is a particular slot for Ethernet. You can connect the Ethernet cable directly in that
port. Wireless functionality is also available in RPi which is 2.4 GHz wireless standard.
Video output is available on RPi which is in the form of HDMI. So we have a HDMI slot
available on RPi. We can directly connect HDMI cable to this port and the display can be
taken on a monitor on a TV or other display devices. So HDMI port is there. GPIOs
which is general purpose IOs.

So this RPi has 40 pin header. We have 40 pins are available on Raspberry Pi and these
pins can be configured as input and output similar to the Arduino what we have the
command to set a pin as a input or output. Similarly in different syntax but these GPIOs
or these are also configurable as input and output in Raspberry Pi as well. So there is a 40
pin header available. There are multiple ports. For example, HDMI is there for display.
3.5 mm audio video jack which is for the putting the microphone the headphone or so.
USB 2 interface is there. We can put our USB devices directly to the board and that can
be detected just like our normal computer. Ethernet camera interface is also there. We can
buy a camera from the market and directly connect that camera to the Raspberry Pi. Then
that camera will be detected by Raspberry Pi. So these are all the functionality the
hardware available on the board itself when we purchase a RPi from the market. So all
these functionalities comes along on the board which are not generally there in case of
Arduino.

So the cost of this Raspberry Pi board is expensive compared to Arduino because of so

much of functionality and performance higher performance compared to the Arduino. So
these are some hardware details. Let us discuss the interface a little bit. So in Raspberry
Pi we have 17 GPIO pins. Out of those foot tip in header we have 17 GPIO pins
available. Most of them have alternative functions. Two pins for UART, two for I2C and
six for SPI. All 17 pins can be GPIO. They can be assigned as input as well as output.

They all support interrupts. There is internal pull-ups and pull-downs for each pin. And
I2C as this I2C particular pin have on board pull-ups. These pins are 3.3 volt, not 5 volt

268
like Arduino. These are connected directly to inside chips. So, because this board is
supporting 3.3 volt only. So sending a 5 volt signal to a pin may damage the R Pi. And
the maximum permitted current draw from a 3.3 volt pin is around 50 milliamp. So these
are some specification about the Raspberry Pi. There are 17 GPIO pins out of this foot tip
in header. These 17 GPIO pins most have alternate functions. So these can be used as a
GPIO as well or they can be used directly for other application. For example,
communication with the other chips and there are multiple standards of communication.
For example, UART or I2C or SPI. So all these pins, 17 pins they can be used standalone
as a input output pin or they can be used for this particular applications directly. So out of
these all these 17 pins can be set as a GPIO just like the Arduino. In Arduino we can set a
pin as input and the output. Here also we can set these pins as input or output and all of
these pins support interrupts. Interrupt is something when a system is running a particular
a computation is running and a hardware sends a signal to change its course. For
example, there is a change in the temperature outside. So the hardware which is the
conventional sensor it will change its reading. So this will be detected as a interrupt by
the microprocessor and this interrupt will be handled in the normal execution of the
program. The program will detect this interrupt and based on its programming how to
handle this interrupt that execution will go on. So all these 17 pins on these GPIO pins
they supports all interrupts at their end and internal pull ups and pull down for each pin.

So pull up and pull down is a extra hardware connected to the pin in case of high
impedance. For example, there is nothing connected at the pin outside. There's no 0 or 1
for example, no 0 voltage or 3.3 volt is applied on the pin. Then this pin is not forced to
any particular voltage it can be anywhere between 0 and 3.3 volt. So to avoid this
ambiguity of what the voltage is we generally connect a pull up or pull down circuitry
based on our application. We want a pull up or a pull down. So this is a weak kind of pull
up circuit which pulls the node to higher voltage just in case there is no voltage connected
by external circuitry or internal circuitry. So pull up and pull down define the voltage of a
pin in case this pin is not driven by any other hardware. So all these pull up and pull
down circuitries are generally available on GPIOs to give them a specified value because
not defining a voltage is very harmful for a receiver because if the input is in high
impedance the receiver will detect it as a high impedance and there will be a constant
burning of power at the receiver stage.

So we generally provide a pull up or a pull down circuitry to each and every pin. So this
is all available there in Raspberry Pi. I2C pin has onboard pull ups. So I2C is a special
kind of standard where we transfer the data in this particular format which is I2C or Inter-
chip Integrated Circuit which is the kind of full form of I2C. This is a special
communication scenario where we have only pull down circuits available on the
hardware and the pull up is generally provided by the weak pull ups which we use for
pull up and pull down. So a dedicated hardware to pull down is available but for pull up

269
we generally use inbuilt pull ups on the ICs which define the node to high voltage and
when we want the node to bring down to 0 then we activate our regular hardware which
we put as a I2C hardware. So this I2C pin has onboard pull ups. Pins are 3.3 volt not 5
volt like Arduino. So the Arduino the voltage level was 5 volt and all the input output
pins were 0 or 5 volt range. Here in case of Raspberry Pi the pins are 0 or 3.3 volt. These
are not 5 volt compatible. So 3.3 volt is the maximum voltage which you can apply to
these pins of a Pi. They are connected directly to inside chips.

These pins are generally routed directly to the core chips which are working at 3.3 volt
and sending a 5 volt to a pin may damage the R Pi by itself. So if we apply a 5 volt
instead of 3.3 volt and because these pins are directly connected to the chip inside this 5
volt is enough to overstress the circuitry which is placed inside the chip and that can get
damaged because of this 5 volt supply. So, we cannot connect a 5 volt sensor or any 5
volt hardware directly to a Raspberry Pi. We have to down convert this from 5 volt to 3.3
volt. There are multiple hardwares available in the market which you can directly use. On
one side you apply a 5 volt signal and the other side you get a 3.3 volt signal directly
from those hardwares. So these are all available for R Pi and the maximum current which
you can draw from a pin is limited to 50 milliamp. We cannot put a very heavy load
directly with the GPIOs because the maximum current capability is 50 milliamp. If we
need higher current, for example we want to run a motor using a Raspberry Pi. Motor
takes let us say 100 milliamp of current. So we need to put a buffer which provides this
much of current of 100 milliamp so that this motor turns on and this buffer generally take
very less current from a Raspberry Pi and generate the required current from DC power
source or the external supplies. So the maximum current is limited to 50 milliamp. So the
next step to use R Pi is installing and setting up the operating system for Raspberry Pi
which is nothing but called the Raspbian. Which is an operating system for R Pi. So as
we discussed the Raspberry or R Pi this is exactly like our normal laptop or desktop
computer. This has almost all the functionality which our laptop or computer has. Similar
to the operating systems what our desktop computers has Raspberry also works on the
operating system unlike Arduino. Arduino there is no operating system and we simulate
we kind of execute our program in our normal PC and when it works we put this program
or this algorithm directly into the Arduino which keep on keep on iterating that program.

There is no operating system in Arduino. Here in Raspberry Pi we have a dedicated

operating system which is called Raspbian which act as an operating system. Operating
system here is needed because the operating system is the interface which connects the
hardware to the software what we are going to make. For example we want to make a
program for traffic light signals. So this program is directly not executable on the
hardware so operating system provides an interface where this program will be converted
back into a compatible language which hardware can understand. So operating system is
needed for R Pi unlike Arduino and this operating system is called Raspbian. This

270
operating system is available from the company directly and we can download this
directly from the official site. So, to do this go to official raspberry pi download site and
download the Raspbian. We need to write the OS. It can be used by R Pi. We need to
boot the R Pi for the first time and perform the first-time configuration setup.

This includes typically partition, password, file system etc. Then we can connect to Wi-Fi
network. So, this is a brief instruction on how to start raspberry pi. So once we have the R
Pi hardware with us we can go to the official R Pi site to download the version of
Raspbian whatever is the latest one available on internet which we can download. We
save this file this operating system file directly to the SD card which is there in the R Pi
so that it can be used by raspberry pi. Then we need to boot the R Pi for the first time and
perform the first configuration setup for example partition how the partition we need to
make in the micro SD card, password setting or the file system what kind of file system
we need to use in that micro SD card and then we can connect to Wi-Fi network. Of
course these instructions are not complete because discussing the use of raspberry pi and
how to use raspberry pi to make a smart sensor that is not the scope of this work. This is
just for the knowledge of the viewer that we can use a raspberry pi to make a smart
sensor. So, once we setup the R Pi or the raspberry pi with operating system the next step
is to download the python development environment. So generally, we use python as a
language to program the raspberry pi. Python is freely available on internet we can use
any Linux based operating system and the python generally comes with this distribution.

Python is a language which is used for programming these raspberry pi. So I strongly
encourage the viewer to learn python small programs how to make a small program in
python and execute it using python and how to put this program into a raspberry pi to
perform a certain operation. So the next step of this is python tool installation. Python
development tool installation which typically includes libraries, install integrated
development environment. Or we call IDE and we have command line terminal. So, once
we setup R pi with Raspbian which is the operating system then we setup the python
environment in the raspberry operating system so that we can use python as a language
for programming this R pi module.

Python development tool actually contains some libraries which are needed for the
execution of program. Integrated development environment which is a dedicated IDE or
the environment which we use to write the program and execute it and we have a
command line terminal as well so that we can directly write the command and send them
to the hardware. So, python development environment is needed to program the raspberry
pi. So let us discuss little bit about the python programming language. So, python is very
easy computer language which is very the most many of the syntax are common to other
programming languages. So, we will discuss very briefly about the python. So, control
flow and data types. So, in python once a variable is given a value its type is
automatically set and will not change unless we reassign two different types. Once a

271
variable is given value its type is set. And will not change until reassigned. And a
variable can hold value of any type. Type is specified by value and not by the variable.
For example, we have numerical type. integer which is 32-bit precision. And A equal to 4
is a typical assignment for integer type. The other is long which is unlimited precision.
We can assign for example A equal to any value. So, this is a long type. The next is the
float type which is the floating-point number. A equal to let us say 6.9758 whatever the
value is. And we have complex type which is complex number. A equal to 5, 6. So this A
is now 5 plus 6i. So, in python we have a variable which is given a value. So as soon as
we assign a value to a variable its type is set by that assignment, and it does not change
unless we reassign it to a different type.

For example, we have here the integer type which is 32-bit precision and when we have a
variable for example A equal to 4 or A equal to 5. So, this assignment of A to 4 this
dictates the type of variable A. Now by A equal to 4 python knows that the variable A is
assigned a value of 4 and the type of this variable is an integer. Similarly we have another
type is long where we assign a very long number to a variable A equal to 1000 and so on.
So then the python knows that this variable is of long type. Similarly float is the floating
point number when we have a number which is having a decimal places. So, A equal to
6.9758 when this number is assigned to a variable the python knows we need not to
explicitly declare the variable type. Python knows by assigning this number to variable A
that the variable A is of floating type in nature. The next is the complex number where if
we assign A equal to 5, 6 then python knows that variable A is of complex type.

So we need not to assign when we declare the variable. However, the type is fixed now
we want if we want to use the same variable as a instead of integer now we want to use as
a float so we need to reassign this variable as floating type. So this is the difference. So
these are the some numerical type. Next the first assignment to a variable actually created
the variables type do not need to be declared. Python figures out by its own. Therefore,
the assignment is equal to and comparison is. Double equal to and for numbers plus
minus multiply divide etcetera. For logical are words for example and or not etcetera. So
here we have these variables which need not to be defined initially and whatever is the
assignment to those variable python knows by itself that this is of the type which is
assigned to this variable. The next the assignment is equal to when we assign a value to a
variable we put A equal to 4 but when we are conditionally checking whether these two
are equal then we write double equal to which is very common practice in all the
programming languages and the python follows the same similar kind of syntax as other
programming languages. For example, in other programming languages well when we
add two numbers or multiply two numbers we use plus or star sign for numbers and for
logical operators when we are performing and operation or not operation then we
generally write and in words A and D and in words so that the python knows that this is a
logical operation on these two variables. So this is how the python works this discussion

272
is not complete python language is very big kind of very kind of involved language and I
advise to study the python language from some other materials this course is not to teach
the python language but we will discuss only very basics of python language how to use
the python language and raspberry pi to do a smart sensor to make a smart sensor and to
perform certain operations. So let us see one example based on python and we stick back
again to that traffic light signal what we just saw in the Arduino case. So here we are
using raspberry pi program in python. So in this case we have a raspberry pi hardware
which has multiple GPIO pins and I choose let us say few pins to make this street light
controller. This is for example the GND pin or the ground pin which I connect to the
universal ground of external circuit.

I connect to the red LED GPIO gate. I connect to green LED and GPIO 7. I connect to
the green LED. And GPIO 21 I am keeping for the external control like a button push
button also. Now python program for such arrangement is something like this. traffic
lights buzzer from time import sleep. Buzzer is equal to buzzer at pin 15 button is equal
to button at pin 21 and lights at pin 23 is equal to traffic lights at 25, 8, 7. button wait for
pass button wait for press then buzzer on then light dot green. On then sleep for one
second then light yellow. On then sleep for one then light dot red. On then again sleep for
one and then lights off and then buzzer off. So this is a small example of python code
which can be written to control this traffic light. This has a push button which we press
and when we press this button this while loop is executed. What this while loop is doing?
This is turning on these red, green and yellow lights LED lights within a delay of one
second because we are putting a sleep of one second. So all these functions these are
actually imported from certain libraries and if we see the second line for example from
time import sleep. So sleep is a very well defined function which python knows very
well. Whenever we call this function sleep python knows that the program need to be
halted for that many seconds whatever is the argument written in the sleep function sleep
bracket one then it means for one second the program need to be halted or stayed idle for
one second. This function is not readily available in all the programs whatever function
you need that is available in particular libraries for example math functions are generally
available in maths. Sleep function is available in time library those libraries we need to
import in our program so that python knows these functions and perform the computation
accordingly. So, this is a example of python program of course this is a very small
program for real sensor development or a particular smart sensor we need to expertise on
the python programming itself as well as the hardware connection the electronic circuitry
that also that particular knowledge need to be acquired from other sources as well to
make a complete smart sensor.

So, this is all for today.

Thank you.

273
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 33
Data Acquisition: Signal Conditioning

Hello, welcome to the course Transducers for Instrumentation. We discussed some

conventional sensors and we discussed that there are certain smart sensors which are
nothing but the conventional sensor which is interfaced with extra electronics for example
microcontroller or a microprocessor which locally process the signal generated by these
conventional sensors and this meaningful data which is already processed by these
controllers. This is now sent to the base station or over the internet so that we do not send
kind of erroneous data or the garbage data generated by sensor because the sensor is
sometime may overshoot or may provide a data which is not meaningful so we need to
process that data as soon as possible so that we can find out only the relevant data out of
it. So we have a for example in smart sensors we have a conventional sensor which is
generating this data and on the other hand we have these microcontrollers or
microprocessors which are processing this data. Now the data which is generated by these
sensors this is not most often adequate to apply directly to these microcontrollers because
for example the generated data by a sensor is typically in a very small range for example
few millivolt of signal only is generated by these conventional sensors. However, the
microcontrollers they work in 3.3 volt or 5 volt range as we saw in smart sensors so they
expect a data which varies from 0 volt to 3.3 volt or 5 volt depending upon the
microcontroller. So the expected range is 0 to let us say 3.3 volt but that signal generated
by a conventional sensor is in few millivolts only.

So the very first step to do to make this data adequate to a microcontroller is to amplify it.
So the very first thing which we can think of is to amplify this data from few millivolt to
few volt kind of a range. So this is one step. There are certain other steps as well which we
have to do to make this data appropriate or kind of suitable to apply to other electronics.
So this stage where we are modifying this data generated by sensor making it suitable to
apply to microcontroller. These stages are called signal conditioning and signal acquisition.
So signal conditioning is the process where we process the data generated by these smart
sensors make it suitable to interface or to give it to further electronics. So we have a signal
conditioning. So signal conditioning is the operation formed on the signal to convert it to
a form suitable for interfacing. With other elements in the system. There are many factors
actually which may prevent a signal used by one device or circuit from being usable by
another device or circuit requiring an intermediate rate to bridge the gap. So we have signal
conditioning which is the operation which we perform on the data and this data is generated
by some conventional sensor and this operation involves certain steps so that this data

274
becomes usable to another electronics or another circuitry. And there are multiple factors
in fact that what we saw is one thing that the data is of very small range few millivolt and
the electronics which is processing it, it expects the data in 3.3 volt range. So this is one
aspect where we need to amplify the data to make it suitable.

There are other things as well for example noise is there in the data that we need to filter
out and some other aspects. So all these factors are there which prevent this signal which
is produced by this sensor from being usable by another circuit or devices which actually
processing this data. So this unusability is there because of certain aspects and there are
multiple factors involved in this and to process these factors we need extra hardware or
extra circuitry to perform these functions on the data and this is called the signal
conditioning of the sensor. Now this kind of bridging operation is called signal
conditioning. And there are certain examples of this process. For example, the range and
offset correction. Convergence. Convergence, for example current to voltage conversion.
And filtering. So, here we have signal conditioning hardware which process this data
generated by conventional sensor and make it suitable so that it can be processed efficiently
with the microcontrollers or microprocessors.

So this signal conditioning hardware is separate from conventional sensors as well as from
the microcontrollers or microprocessors. This hardware is called signal conditioning
hardware. This hardware acts some sort of a bridge between a conventional sensor and the
controller. This bridging what the function is providing this is called signal conditioning
and there are certain example. For example one is the range and offset correction range we
just discussed. If the sensor is producing a data which is varying only in few millivolt let
us say 0 to 10 millivolt of range which is very small range and the data generated is
confined only in this range. Now this range is not suitable to process by these
microcontrollers. So we need to amplify this range from 0 to 10 millivolt we convert it to
0 to 3.3 volt. So this is a range conversion. Similarly, if we have certain offset in the sensor
data that offset also can be corrected using this signal conditioning hardware. This is one
function which need to be performed before processing the data. The second is the
convergence. For example, our sensor is a current sensor and our microcontrollers which
are nothing but voltage devices that measure the voltage at the pin. So, we need to convert
this current data into the appropriate voltage data.

So, this kind of conversion is also happening in signal conditioning block. The third is the
filtering. Filtering is removing the unwanted signals or which we refer noise. All the
sensors when they generate data that data is most of the time analog in nature along with
certain other frequencies which are unwanted by us. So, this unwanted frequencies or
unwanted data is like a noise to us and this need to be removed from the data before
processing the data using our microcontrollers. So, this kind of filtering stage is also a part
of signal conditioning block. So, these are the some of the functions which we do using
signal conditioning hardware. Let us see how a signal conditioning block look like in the

275
full block diagram. So, this is a typical block diagram of a sensor where we have first a
transducer or simply a sensor. Now this transducer exchange the data it has both input and
output communication with unit something called signal conditioning unit. So, we have a
signal conditioning unit. This is signal conditioning unit. This has multiple parts for
example one function of this signal conditioning unit is to provide excitation to the
transducer. Excitation means it provides necessary bias conditions and the signals. The
other block is a bridge completion.

Then we have next stage amplification. This is generally followed by a filtering unit. So,
this is signal conditioning unit. This process the data and send it to the next unit which is
data acquisition unit. So, here we have a transducer which is providing the data this can be
a normal conventional sensor for example a thermal sensor which is detecting the
temperature and providing its input to the signal conditioner. This signal is very small in
magnitude 0 to 10 milli volt or so and this is having certain amount of noise as well as
unwanted frequencies. Now this data this controller or this smarts this conventional sensor
is attached to the signal conditioning unit. The purpose of signal conditioning unit is to
provide necessary excitation to the sensor for example the sensor is active sensor which
requires external DC supply and certain signals as well when we want to activate the sensor
or when we want to deactivate the sensor. These kind of signals can be generated using the
signal conditioning unit. So, it has a separate portion of excitation which is necessary to
excite the sensor and the data which is generated by the sensor this comes to the bridge
completion. So, we can have some sort of bridge completion for example the Wheatstone
bridge we use for the digestive measurement. So, this kind of bridge completion is provided
by signal conditioning unit. Now this signal which is still a very small signal 0 to 10 milli
volt this is applied to amplifier which can be our Op-Amp based amplifier or BJT based
amplifier this must be a linear amplifier. This amplifier is now going to amplify the data
from few milli volt to volt range 0 to 3 volt or 5 volt as per our requirement. So, this
amplification is a linear function which is applied to the signal whatever is the signal
coming in if it has noise or it does not have noise it does not matter everything whatever is
received by amplifier is going to be amplified.

So, at the amplifier output we get the amplified signal as well as the amplified noise as well
because amplifier cannot distinguish between a signal and a noise it amplifies whatever
comes as a input. So, after amplification stage generally we have a amplified data with
amplified noise. Now this amplified noise which is unwanted, and we need to filter it out
before further processing. So, this filtering is done by filter stage these filters are generally
the next stage of amplification because as soon as we amplify the signal noise also get
amplified and we need to suppress the noise immediately after that. So almost everywhere
amplification stage is generally followed by a filtering stage here in case most of the time
low pass filters. So, this stage filter out the noise and the amplified signal now amplified
by this amplifier is now sent to the next unit which is data acquisition unit. Now this data

276
acquisition unit has its own certain role before sending its data to the microcontroller. Let
us discuss this data acquisition. So, this data acquisition primarily has a sample and hold
circuit. This stage is then followed by the analog to digital converter. And this stage is
further followed by the multiplexing and data transmission. This is multiplexer and after
this stage the data is sent for further stages for processing, filtering or display. So, here
once this data which is already filtered out the noise this data is sent to data acquisition
unit. This data acquisition unit again has basically three parts. The first one is sample and
hold circuitry.

This is used to sample the data at certain instances and hold the data during that period so
that this data can be applied to analog to digital converter. The next stage analog to digital
converter is this has multiple levels and based on the level of analog signal a code is
generated by this analog to digital converter. This code is a digital code, for example
011101 this kind of code. This code is now easy to process using microcontrollers because
microcontrollers are digital, and they process the digital data efficiently. So, this analog to
digital converter gives out that code to the next unit which may or may not be multiplexer.
For example, we have multiple sensors connected to a single controller then in that case
we need multiplexer so that one sensor is connected to controller at a time and the data is
processed only for that particular sensor. So, we may or may not need multiplexer in all
the smart sensors depending upon our number of sensors attached to microcontrollers. So,
this is the data acquisition unit which after this unit we get a data which is digital in nature
0 and 1 0 for example low voltage or high voltage and this digital data is now suitable for
further processing or we can record it directly to some hardware or we can display it in the
some LCD or this kind of display. So, the data goes to the next stage. So, this is a complete
bridging happening in the smart sensors when we have transducer on one end which is
conventional sensor and on the extreme right we have a processing or recording or display
which is nothing but our microcontrollers and in between we need these two types of
hardware which is one is signal conditioner and one is data acquisition unit.

These units we need separately to make the data suitable for processing. So, we are going
to discuss signal conditioning now in little detail now. So, the signal conditioning unit it
provides external excitation and grounding. It completes the circuit for some time we called
bridging. It linearizes the data. It filters out the noise amplification. It also provides
isolation. It isolates one part of system electrically from other and typically the input is in
milli volt and the output is in volts. So, these are the some of the functions performed by
this signal conditioning unit. It provides external excitation and grounding. If sensor need
a 5 volt supply then this signal conditioning unit provides a 5 volt supplied with limited
current so that the sensor can act efficiently. It completes the circuit whatever is the
bridging or whatever external hardware circuit is needed for sensor to work that is provided
by signal conditioning unit. It linearizes the data. It filters out the unwanted noise.
Amplification is generally done in signal conditioning unit so that the input is in milli volts,

277
but the output is in volts. And this also isolates one part of the system to another part of the
system electrically so that they do not interfere with each other. For example, the sensor
which is producing data in milli volt that stage might be working at very low voltages. If
we connect directly this stage to microcontroller which is running at 5 volt this 5 volt can
interfere or can damage this sensor directly. So, signal conditioning unit provides an
isolation so that the sensor can act efficiently at low voltages and microcontroller sees a
signal which is 0 to 5 volt. So, these are some of the functions of signal conditioning unit.
As we discussed the first stage is the bridging stage where we put necessary hardware for
example Wheatstone bridge or a linear resistive divider or any kind of electrical network
which is particular to that sensor. So bridging is that function. Now after bridging is the
amplification stage which is generally there because the signal generated as we see in milli
volt and we need in kind of volt range. So, amplification is the stage where we need to put
in the signal conditioning unit.

So, we see the amplification stage. So, the ADC generally requires high levels of voltage.
Typically, 0 volt to 5 volt range. So low voltage sensor output may need amplification prior
to conversion. Now for this step we can use an op amp circuit with gain. May be inverting
or non-inverting. If signal need to be DC shifted, we can use an op amp level shift circuit.
So, one configuration of op amp based amplifier is something like this. We have amplifier
which is made up of op amp. Op amp has two inputs one is plus one is minus and one is
the output. Now this op amp on one side this is connected to the sensor using a resistor. So
this is V sensor. This is R. This generally has a feedback resistor. This is feedback resistor
RF and on the positive side we attach a voltage which is V ref and this is V amp. So this is
the basic configuration of a amplifier using an op amp. In this we have a sensor which is
connected to the negative terminal of op amp using a resistor R and in feedback we have a
resistor RF. So, this unit act as a amplifier. Now on the positive side we can directly connect
it to ground so that we do not shift the level of the signal. If we put a voltage source V ref
in between so this means then we can shift the entire voltage by this V ref voltage if it is
0.1 volt so we can provide a offset of 0.1 volt using this V ref in the output signal and
output is collected at the output node which is V0 here with respect to ground. So, this is a
simple stage simple op amp based amplifier. We can have op amp based amplifier or we
can have VJT or MOSFET based amplifiers so that can be used as per the requirements.
This is amplifier stage. So now this amplifier does not know whether the signal is with
noise or without noise. Whatever comes in at the V sensor node that will be amplified using
this circuit and the V out will have a amplified output along with the noise if noise is there
that noise also gets amplified. This amplified noise is the problem because we cannot send
this noise also to the controllers otherwise, they also cannot detect this whether it is a signal,
or it is a noise.

So, this stage is immediately followed by a filtering stage where we filter out this noise.
So, the next stage is the filtering stage. So, these filters may be classified as digital filter or

278
analog filter. So, the digital filter is. These filters are implemented using a digital computer
or special purpose digital hardware. And the next is the analog filters. They may be
classified as either passive or active. And usually implemented with R, L and C
components and operational amplifiers. So, in this stage where we have this filter these
filters can be classified first as digital or analog filters where the digital filters are
implemented using our special purpose hardware or a digital computer. In this stage we
convert the data into digital form, and we have certain algorithms which can filter out the
noise out of the data out of this digital data using some algorithms and these algorithm run
on our digital computers or our normal computers or we can use special purpose hardware
which are specified in filtering out this digital data. So, there are digital filters available
which we can use but for that we need to convert the data first into the digital form. This
is first classification the second is the analog filters where we have these analog filters
which are made up of electrical components like resistor, capacitor and inductors and these
components along with we can use an Op-Amp or operational amplifier. So, these analog
filters are again classified into two types one is the passive filters, one is the active filter.
Passive filter means we do not need external power supply to act as a filter. For example,
a resistor, inductor and capacitor, the combination of all these component only does not
need an external power supply and they can act as a filter without any power supply so
these are passive filters. However, if we put a Op-Amp as well along with these filters these
Op-Amp these are semiconductor devices they need external power supply external DC
power supply to act as a Op-Amp so in that case the filter is called a active filter. So analog
filters are of two types passive filter and active filter. Now let us see how a filter actually
look like and how it works. Typically, a signal after amplification must be low pass filtered
to remove high frequencies generated by amplifiers. So, for example we have a signal
which is amplified by amplifier and when we filter these plot this graph which is on the x
axis is frequency and on the y axis we have gain. So now our signal is which is getting
amplified now we apply a filter, and the filter has certain characteristics for example it has
a pass band and a stop band.

Similarly for a filter we have a pass band and a stop band this portion where the gain is
high of a filter this is called pass band and where the gain of a filter is very low is called
the stop band and this portion where the gain is transitioning from high to low this is called
transition zone and there is a very low value of gain which is called the noise floor. This
red is our frequency response of our filter. So, this is a typical amplification or the
frequency response of a low pass filter. We are discussing here the low pass filter which
the name itself says low pass means the low frequencies which are applied to this filter
they will pass through, and the high frequencies will be cut down. So, we are applying a
low pass function to our data because the noise we are assuming is high frequency which
are generated by amplifier. So, this high frequency noise we want to filter out using this
filter. Now this filter has certain filter response which we can see for the low frequencies
which are close to origin. These low frequencies filter has a high gain it means these low

279
frequencies will come and they will smoothly pass through this filter because of this high
gain. However, this filter response is very less or the gain is very small. It means if the high
frequency comes in at the filter those frequencies will be suppressed because of very low
gain of this filter and these high frequencies will not pass through this filter. So, this act as
a low pass filter high cut filter high frequencies will be cut down. So, this we can see by
this red filter response high gain on low frequencies and very low gain on the high
frequencies. This very low gain is not actually 0 is a certain value which is not actually 0
it means the high frequencies which will come they will not be eliminated because the gain
is not 0 gain is very low value but still a certain value. So these low frequencies these high
frequencies will be suppressed to a very low value which is generally called the noise floor
of a system that very low values generally we consider as a noise floor and we do not
operate our systems very close to this noise floor. These are very small signals which are
generally considered as noise floor.

So, the response of a filter should be very close to noise floor in the stop band and where
we have a high gain of filter this is called pass band. This pass band and stop band the
transition does not happen very sharply. So, because these are realistic components RLMC
they do not have a very sharp transition from high frequency to low frequency. There is a
transition smooth transition from pass band to stop band that is called the transition zone
and we generally do not want to operate our sensors in this transition zone because there is
no certainty of the gain in this transition zone. So, we want to have our signals very much
close to the pass band and the noise is all in the stop band region. So, this is a typical
filtering stage of a signal conditioning unit. So let us see some realistic filters. So, there are
four types of filters generally used. The first one we just saw is a low pass filter where we
have frequency axis and the gain. Ideally we expect a flat transition zone and the transition
is very sharp and in the stop band we want a absolute zero gain. So, this is the desired
characteristic of a low pass filter. But with actual circuitry we get this kind of characteristic.
This is the practical or real filter which has certain transition zone and transition does not
happen abruptly. This is an ideal case for a low pass filter. Similarly, we have something
called a high pass filter where the low frequencies will be cut down only the high
frequencies will be passing through.

The desired response is like this. This is ideal and the actual response is like this. This is
high pass filter. The third we have a band pass filter where only a band of frequencies will
be passing through. All low frequency or high frequencies will be rejected. So for a band
pass filter we have a desired response like this which is the ideal case. However practical
case this filter response is like this. There will be some of the transition region where the
frequencies will be still passing through. So, this is real response of a band pass filter. And
the fourth one is the band reject filter. This is the opposite of band pass filter. In band reject
filter certain frequencies only will be rejected everything else below that frequency or
above that frequency will be passing through. So, in band reject filter we have a desired

280
response like this. For a band of frequency, the response is zero everywhere this is high.
So, this is the ideal case. In the practical application the filters have this kind of response.
This is the real response of band reject filter or band pass filter or band stop filter. So, these
are four types of filters generally used for filtering. Low pass filter is passing low
frequencies rejecting higher. High pass filter is removing low frequencies passing high
frequencies. Band pass filter means we have a certain band only where we are interested
in. Below this frequency or higher than this frequencies all will be rejected that is band
pass filter and band reject filter is we want to reject certain band where these confined
frequencies we do not want these frequencies to come over. So, in that band the gain is
zero except or the below these frequencies or higher these frequencies those frequencies
will be passing through. So that is band stop filter. These are the four types of filter
generally used in signal conditioning unit. So, this is the third block of signal conditioning
unit.

That is all for today.

Thank you.

281
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 34
Smart Sensors: Data Acquisition: Analog to Digital Conversion

Hello, welcome to the course Transducers for Instrumentation. Last lecture we were
discussing about the signal conditioning unit, and we discussed that for a smart sensor we
have a conventional sensor and in extra electronics which process the signal to bridge the
gap between these two we need external hardware or the extra hardware which provide
necessary conditions so that this the data generated by the sensor is suitable to apply to
these microcontrollers or microprocessors for processing. So in that bridging circuit we
have mainly two components, one is the signal conditioning unit we last discussed in the
last lecture. The first is the signal conditioning unit where we have bridge completion
amplification and filtering unit and the second is data acquisition unit where we saw we
have three modules in that sample and hold circuit, analog to digital converter and
multiplexer if you needed. So, in last lecture signal conditioning unit we saw that we
have bridge completion and then we have amplification. For amplification we use any
kind of linear amplifier which boost the signal generated by the transducer or a sensor.

This signal is basically in milli volt range and amplifier amplifies it to a volt range
typically up to 3.3 volt or 5 volt depending upon the microcontroller. After this
amplification because noise also get amplified, we apply a filter and we saw four types of
filter low pass filter, high pass filter, band pass filter and band reject filter or band stop
filter. So, depending upon our requirements we can filter out the unnecessary noise from
the signal and we need to find out what is our signal frequency in and around of that
signal frequency we need a filter which has a pass band located in that range and the rest
of the frequencies should be eliminated.

So that was the work of a filter in signal conditioning unit. So, after a signal conditioning
unit we get a signal which is now amplified in a volt range in typically 1 to 2 volt of
range, and which is free of noise now. But the signal achieved at this point after signal
conditioning, this is still an analog signal. Analog means it is continuous in time. So, this
signal is difficult to process using our microcontrollers or microprocessors because these
hardwares these controllers they can process the data which is in digital form more
efficiently.

So our analog signal now it is amplified but still not suitable to apply directly to a
microcontroller or a microprocessor. We most of the time convert this analog signal into
a digital signal and that part is done using data acquisition unit. In data acquisition we
have primarily these three components one is the sample and hold circuit. What this

282
circuit does it samples the analog signal in time and then hold the data for a limited
amount of time so that during this time even if the analog signal is changing the output of
a sample and hold is stuck only to that value wherever we have take the sample and now
this sample will be converted into a digital form using analog to digital converter. Sample
and hold we use because the next stage analog to digital converter it takes amount of time
to convert this data into a digital form.

So, during that time this data need to be stable it should not be changing with time.
However, the analog signal which is coming after signal conditioning unit that is
continuously changing. So, we need a sample and hold circuit which samples the data fast
enough and sample this data and hold it for certain time. So this circuit is called sample
and hold circuit. Then this is followed by analog to digital converter where a
corresponding code will be generated from this hold signal which is depending upon the
level of this data or the voltage of this data a particular code will be generated that
corresponds to the value of analog signal and this data or this code is in digital form
011101 typically like that.

This data now is ready to be processed using any microcontroller or microprocessor. So

this is the complete flow of making a smart sensor from the conventional sensor to the
microcontroller or microprocessor. So, we today we are going to discuss the data
acquisition part which is this block. So we discuss the data acquisition and in this we
discuss basically first two things one is the analog signal. These analog signals that vary
continuously throughout a defined range.

So these are the analog signals that vary continuously throughout a defined range. These
are representative of noise. Many physical quantities, for example temperature and
velocity or pressure. This represents generally a voltage or a current level. However, on
the other hand we have digital signals.

These are the signals that take on specific values only. These types of signals are required
for digital logic operations. And these are also representative of physical quantities by a
series of binary numbers. So this is the data acquisition part. So here we have two types
of signals.

One is the analog signal and the other is the digital signal. Analog signals that are
continuous throughout the range which assumes all the values in between for example if
analog signal is changing from 0 to 5 volt all the intermediate voltage levels are being
occupied by this analog signal at some point voltage will be 1.1-volt, 1.2 volt, 1.23 volt
all the values are allowed for an analog signal.

So, this is continuous in time. This represents many physical quantities, for example we
have a temperature sensor. So based on the temperature it generates a output voltage or a
current it can be any of them. So this voltage it produce let us say 1 milli volt. This

283
corresponds to the physical quantity which is the temperature here and when this value
changes it means the physical quantity is changing the temperature is changing because
of that this signal generated is changing.

So, this is an analog type of signal which is produced mostly by sensors. On the other
hand, we have a digital signal that signals take on only specific values. For example, a
digital signal can have only 0 volt or it can have a 5 volt if we are talking about 5 volt
compatibility. So 0 volt is called 0 most of the time and 5 volt is called 1 the binary logic
level 1. So this is not like the voltage is 1 volt.

Voltage can be anything, but the high voltage is represented by a binary 1. So, if we say
binary value 0101 it means the signal is changing from 0 to 5 volt then 0 to 5 volts like
that. So, these voltage levels are defined by the microcontroller we use. If let us say we
are using a Arduino based microcontroller which is 5 volt input output so this high level
is 5 volt. If we are using a raspberry pi which is 3.3-volt microprocessor based hardware
so then our high level should be 3.3 volt should not be 5 volt. So this high voltage should
be compatible to our next electronics which is going to follow this stage of data
acquisition. Now these 0 and 10 and 1 which is now binary numbers they also represent
the physical quantities. For example a temperature is there which is detected by our
temperature sensor.

Interestingly it generates 1 milli volt of signal which is analog in nature but if we convert
this signal into a corresponding binary code let us say that comes out to be 001. If the
temperature increases so that the output voltage also increases this sensor which this
sensor produces. This voltage increase will give the corresponding increase in the binary
number so this 001 becomes 100 or so. So this binary number is also a representative of
physical quantities however the form of representation is different. Earlier it was a analog
signal now we have a digital signal means the code the binary code where all the low and
high values are only specified values 0 or 5 volt and the code which is 101011 this code
represents the physical quantity not the actual value which is 5 volt.

So 5 volt does not represent the physical quantity but the code actually represents the
physical quantity this is the difference between the analog signal processing and our
digital signal processing. Now we have certain advantages of digital representation. So,
the first one is the values are limited to specific discrete segments. So, the second one is
the values are limited to specific discrete segments. Next, it is not subject to the
distortions just like analog signals.

These are easy to reproduce. And these are easy to copy or stored. So, these are some of
the advantages of digital representation. The very first one is the values are limited to
specific discrete segments only. So, we need not to deal with all the continuous values of

284
analog signal we are only handling the discrete values of voltages and based on that we
are processing our signal.

So, we are losing some accuracy in some sense but this is now very easy to implement
certain algorithms or do this digital processing. So, these values are very limited so we
can handle them very easily. The second the advantage of this binary representation or
the digital signal is these are not subject to distortion or just like in analog signal. For
example, we have analog signal which is continuous in range if there is a certain
distortion or some noise comes in that will distort the signal which is continuous now and
we do not have any mean to detect whether this is a spike in the signal or this spike is
caused by some external noise. So, we cannot detect later on whether signal is has certain
noise or the signal is like this itself.

So, this is the problem with analog signals however in digital signal we know that the
voltage levels are only 0 or 5 volt anything in between is we can kind of discard that
voltage. So, signal can assume only 0 volt or 5 volt no voltage is allowed in between. So,
when we receive the signal and we find out that voltage is not actually 5 volt it is 4.5 volt
slightly lesser than 5 volt. So, we can safely say that this has happened because of some
noise and we need to restore it back to 5 volt.

So, at the later stage we can easily restore this error what is caused by some external
noise or some other factors. So, these digital signals these are not subject to distortion or
error just like in analog signals. So, these are easy to reproduce. So, when we transmit
these data this digital signals because of some loss or because of some noise if the signal
magnitude goes down instead of 5 volt we receive the data as 4.5 volt then we can put a
inverter or some external circuitry which can restore this data back to 5 volt.

So, this 4.5 volt becomes again 5 volt without losing any information. So, this is easy to
reproduce. The next is easy to copy or stored. So, we have multiple medias to store
digital data for example flash drive or hard disk drive. So, these all the devices work on
storing on the digital data which is in the form of 0 or 1 or we can say the 0 volt or the
high voltage.

So, all these medias are compatible with the digital data. So, these are some advantages
of digital representation. Now let us discuss about how we convert this analog signal into
a digital signal. So, the very first block is sample and hold circuit.

So, we discuss sample and hold. So, a sample is an instantaneous measurement of an

analog voltage. So, sample and the sampling frequency is the number of samples taken
per unit time. So, for example we have a data which is analog data now coming in x axis
is time and y axis we have voltage input and we have a data which is something like this.
So, this is an analog signal. So, what we do we have a kind of sampler which take the
sample of this data at certain instances only.

285
So, this is time and this is the sampler voltage and now we have a sampling frequency
applied to this like this and the actual hardware for this is very simple. So, we have a
switch connected here and then a capacitor. This is the output Vout and this is the voltage
Vi which is generated by some sensor or some external circuitry. This is Vi and this point
and Vs actually controls this switch. So, we see Vs is nothing but a train of pulses which
has a certain frequency it is called sampling frequency and when this frequency is applied
on this switch this is the switch when there is a high voltage applied this switch will turn
close this whole thing will be conducting and when there is a 0 volt applied on Vs this
switch will be open.

So, when this switch is closed whatever is the voltage at Vi that will pass through this
switch and reach V0, V0 becomes Vi. So, if we plot the Vout graph as well in the same
graph. So, these are the instances when the switch is going to be closed and the path will
be conducted and the switch will remain on till we apply this pulse and as soon as the
pulse goes down the switch turns off. So, this green dotted line shows where the switch is
going to be turned on and V output becomes Vi. So, if we plot Vi in this graph when the
switch turns on here V output becomes Vi and the V output will follow Vi if the Vi is
changing.

So, we can plot like this the Vi V output is Vi is changing very similar to Vi here Vi is
going up so V output also going up. This is V output this is the time axis. Now when the
switch this sampling frequency or Vs goes low at that point whatever is the voltage
remained at V0 that will be kept V0 because the capacitor is there which is going to hold
this voltage across this capacitor and Vout is going to hold the voltage whatever is
applied by Vi and the duration for which this output will be hold that is the time till the
next pulse arrive which is the sampling frequency. So, if we see this V output graph this
is going to hold this value till the next pulse comes in which is the next green line and
when this comes again V output is going to follow V input which is something like this
and when that pulse goes back again this again goes down it remains same and the pulse
goes down. So, this voltage will be retained at constant value.

So, this is called sampling by using this pulse train and when the pulse is not there the
voltage is actually hold that is why the hold circuit is called sample and hold circuit. So,
if we plot the V output for the whole pulse of train it will look something like this. So,
this is a typical signal which is going to be produced by this sample and hold and we can
see this does not represent exactly the analog signal but it is very similar to analog signal.
So, if we see the analog signal first goes up and then goes down and again goes up we
can see this whole sample and hold circuit after this we also have this orange color output
which is also again increasing first then decreasing and then increasing it again though
now the values are not kind of like continuous in analog signal we take the sample of
analog signal for a certain amount of time and then hold this measured value for certain

286
amount of time which is called hold time. So, we have now two things one is sample time
and one is the hold time and the sampling frequency which is defined by our requirement.

So, if we keep sampling frequency high enough then we can reproduce this analog signal
using this sample and hold circuit if we have very high sampling frequency our output
signal will be very close to the analog signal if we have a low sampling frequency then
we lose out much of a information in sample and hold. So, this is how a sample and holds
work and this is a very simple circuit where we have a capacitor which is used to store
this data or used to hold the value of voltage and this switch it can be a mosfet switch
where we connect source and drain to one VI and V output and the gate is controlled by
VS signal which turns this mosfet on and off and this is a very simple circuit of sample
and hold. Now the accuracy of this digital representation. It depends on sampling
frequency and quantization. Where this quantization is? The number of bits used to
represent an analog voltage as a digital number.

And the resolution is the analog step size. For example, we have certain key parameters
such as full scale voltage range which is R number of bits which is B number of
quantization level which is 2 to the power B and we have quantization width which is Q
equal to R upon 2 to the power B. So, in last slide we discussed sample and hold circuit
which produce the discrete samples of a analog signal. So, these samples are now discrete
in time and discrete in the voltage levels because we take a sample and we hold the
sample for a certain amount of time so that this value is not changed now. Now this
sample and hold circuit produce this output which now need to be converted into a digital
signal. This is done by analog to digital converters or A to B we call it in short.

And this conversion now it actually depends on the sampling frequency and quantization.
So, sampling frequency we saw in last slide when we take the sample that depends on
how many samples we collect in unit time that is the sampling frequency and if we
collect more samples it means we are very much close to the original data. If we collect
less samples it means we are quite away from the actual data and we introduce certain
amount of errors in our conversion. So, our digital representation the accuracy of that
depends on the sampling frequency how fast we sample and what are the levels we
consider for quantization.

So, for example we have a full range of 0 to 5 volt. So, the signal can assume any voltage
between 0 to 5 volt if it is an analog signal. But when we convert it into digital we define
certain levels the signal can assume only those levels and if the signal falls anywhere in
between those levels this will be truncated to the higher value or the lower value
depending upon us. So, any signal voltage for example, let us say we have a 5 volt range
0 to 5 volt and we define only 5 levels 1 volt, 2 volt, 3 volt, 5 volt, 4 volt and 5 volt. Any
signal cannot assume any voltage between 2 and 3 volt.

287
So, if my sampled signal comes out 2.7 volt then we truncate it to 3 volt. So, that this
does signal does not fall in between this voltage. So, once I fix my quantization levels my
signal can only assume only those quantization levels it cannot be in between them. So,
this quantization levels depends on how many bits I use for representation. So, for
example I have a range of 5 volt I want to put levels of like 8 levels in this full 5 volt
range then I need 3 bits to represent all these levels because 2 to the power 3 is equal to 8.

If I want to split this whole range into 8 discrete levels I need 3 bits of 3 bits of binary
data to represent all these levels. So, 1 sample will correspond to 3 bits of binary data.
Now if I have a 5 volt range I just split into 8 levels then my signal cannot assume
anything in between it will be truncated to the higher or lower value then the error is
more. If I increase more quantization levels I if I want to have let us say 16 levels then
my reproduced signal is very close to my analog signal. So, my accuracy is better, but to
represent these 16 levels I need to have now 4 bits.

So, for each sample the data which will be converted to that will have 4 bits of binary
data. So, my size of data actually increases if I go for more quantization, but my accuracy
improves. So, there are certain key parameters to convert this data now to a digital signal.
So, we have this full scale range.

So, let us see a graph where it can be represented better. This is time axis and on the y
axis we have these levels which are in voltage. This complete voltage range is called R
from 0 to maximum voltage whatever is allowed. Now I have multiple levels let us say
this is 1 level, this is 2 level, this is 3 level, this is 4 level, this is 5, this is 6, this is 7. So,
we have total now 8 levels. The first one is this which I can say 0 0 0 then I can write 0 0
1, 0 1 0 and 0 1 1.

This is 1 0 0, 1 0 1, 1 1 0 and 1 1 1. So, these are all 8 levels corresponding to 3 bits. So,
here we have B equal to 3 bits and the number of levels equal to 2 to power B which is
equal to 2 to the power 3 or 8. So, we have first level, second level, third, fourth, five, six,
seven and eight. So, these are the 8 levels. Now if a signal comes in for example, sample
and hold circuit generates a signal which comes out like this, any arbitrary signal.

Now if you want to convert it into a binary signal it can be easily coded. So at this point
the value of the signal is close to this 0 signal. So here the code is 0 0 0 then here 0 0 1
then here it is 0 1 1 then it is 1 0 0 then it is 0 1 1 here it is 1 1 0 then this is 0 1 1 here it
is 0 1 0 and 0 0 1 and back to 0 0 0. So if we want to represent this signal in terms of
binary data we can easily say the signal can be represented as 0 0 0 then 0 0 1, 0 1 1, 1 0
0, 0 1 1, 1 1 0, 0 1 1, 0 1 0, 0 0 1 and 0 0 0. So this signal which is generated by the
sample and hold circuit can now represented by this stream of binary data which is in the
form of 0 and 1 and this 0 and 1 is logical 0 and 1 it can be in the voltage range it can be
0 or 5 volt where this 0 represent a 0 volt and 1 represent a 5 volt or 3.3 volt. This voltage

288
level depends on the microcontroller or microprocessor we are using for Arduino or that
controller this 1 volt 1 logic level is 5 volt and for raspberry pi this 1 logic level is 3.3
volt. So this is how we can convert a signal which is generated by a sensor which was of
very small value first we do a signal conditioning on that and then by data acquisition or
this external hardware which converts first it to sample and hold and then we use ADC to
convert it into a digital stream which is shown here. So this signal which is arbitrary
signa l now is converted into a digital form.

So, this is all for today.

Thank you.

289
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 35
Smart Sensors: RF Sensing: Basics of EM Fields, Antennas and RFID Tags

Hello, welcome to the course Transducers for Instrumentation. Today we will discuss
about the RF sensors or the RF sensing. These types of sensors are very much different
than other conventional type of sensors because in these type of sensors we use RF waves
or the radio frequency waves for the detection and the properties of RF waves are very
much different than what we have seen in the entire course about thermal sensors or
optical sensors. So now we use radio frequency waves for the detection and these type of
sensors for example RF sensors or in particular RFID tags, the radio frequency
identification tags. These are very much used to identify the objects or the people or any
other object. These are used to identify, or we can track even the movement of the object
using these RFID tags.

So, in, when we say we are discussing RF sensors we are particularly interested in RFID
tags how they work and what is the principle behind those type of sensors. So today we
are going to discuss RFID tags or RF sensors. So, RF sensing, here we are discussing
mainly RFID tags. Which are used for identification.

So, RFID tag or nothing but the radio frequency identification tag. This describes the
RFID tags. These are RFID tags or RFID tags. To automatically identify people or
objects. So, these RFID tags are nothing but the technologies that use radio waves to
automatically detect people or object.

Here the word automatically is a keyword here. These RFID tags we can use for detection
automatically. However, there is another type of technology which is for example
barcode kind of tags where we see a barcode on multiple products and those barcodes can
also be used for identification but the problem with barcode is that process is slow and
that is line of sight only. So, we have a receiver, we have to align with the barcode then
only that barcode data will be read by the reader. However in RFID case the receiver can
be a little bit non-aligned because this is not line of sight.

So anywhere RFID tag is in proximity of the receiver the data can be read by the receiver.
So that is the main difference between RFID based tags or the barcode based tags where
RFID tags use RF waves for data transfer or data reading by the receiver. This RFID
system primarily has mainly three parts. The first one is the RFID tag where the data is
stored the identification key is stored that RFID tag is one part then we have some sort of

290
antenna in this RFID tag as well as in the receiver. These antennas use RF waves for
communication.

So, whatever is the data stored in the RFID tag that is transmitted using these antenna so
that is the second part and the third part is a transceiver unit which actually read this data
and connect to a central server or some data storage unit where all the identification keys
are stored along with the object identification or the information. So, these are primarily
three modules in a RFID system. So, we have three parts to a RFID system. The first one
is the RFID tag which is actually programmed with the information. The second part is
the transceiver unit.

This has a decoder to interpret the data. Then the third one is the antenna which is equally
important for the system to work. So this antenna provides a mean of communication and
energy to communicate RFID tags. The third part is the field of the antenna and detects
the activation signal from the antenna causing the RFID tag to transmit the information.
The third part is the transceiver unit.

So, here we have three parts in RFID system. The first one is the RFID tag which is the
tag or the card which we carry or the object or the people carry with them. So, this RFID
tags contains a certain information about that card maybe a unique number and all the
different cards are given different numbers. So this is the information which is stored in
the RFID tag. This particular number or we can say the information the specific object
information this is buried in this RFID tag.

Then the second part is the transceiver. Transceiver is the base unit which reads this data.
So, whenever this tag comes in the close proximity of the receiver then receiver is the
unit or the transceiver is the unit which read this particular key which is stored in RFID
tag and detects it communicates with a computer where this computer contains all the
data of all these keys along with the information of the object. For example, this
particular key belongs to this person. So, this data is stored in a particular computer and
transceiver communicate with the computer and find out who is the object or which who
is the person who is carrying this RFID tag.

So, this is second module. The third module which is very important is the antenna. So,
antenna is the unit which transmits the energy in terms of RF waves. So, when there is a
communication we use RF waves or the radio frequency waves for this communication
and this antenna is a mean of communication and energy to communicate with RFID tag.
So, this RFID tags there are two types of RFID tags we will discuss that.

These RFID tags may or may not contain a power source inside. So, this RFID tags to
submit to transfer the data to receiver it needs certain energy. So, this antenna used for a
communication as well as transfer of energy from the receiver to RFID tag and RFID tag
will use this energy to transmit the data back to the receiver. So, this antenna is used for

291
communication as well as it provides energy to communicate with the RFID tags. So this
is one function of antenna.

The second is when this RFID tag when they passes through the close proximity of an
antenna, antenna has a certain area where it can radiate the energy. So as soon as the
RFID tag passes through that area the energy will be given to RFID tags and RFID tag
detects this energy using a inbuilt antenna which may be a smaller antenna and this
energy is now used to communicate this information stored in the RFID tag to the
receiver. So, this energy is sent back to receiver with a particular information which is
stored in the RFID tag. So, this is how the receiver detects whether which kind of
particular RFID tag is moving around in that radiation zone or close proximity to the
antenna. So, these are the three main systems of RFID, main modules of RFID system.

Let's discuss antenna in a little bit detail which is the key part of these RFID tags. So let's
see how an antenna works. So for example we have a electrical system which is giving
electrical signal which is time varying. So this is let's say transmitter. This system sends
out electrical signal on a piece of metal which is our antenna.

So now this is our antenna and the signal comes like this. So let's say blue is the electrical
signal and when this electrical signal comes to this piece of metal which is antenna this
start radiating and let's say this radiation is shown by this red graph. So, these are the RF
waves which are generated using the electrical signal. So, this is how the electromagnetic
field is generated over the RFA. And this part is called the antenna which causes this
conversion of electrical signal into RF signal or the electrical energy to an RF energy or
astromagnetic energy.

This part is antenna and this is the transmitting antenna. There are certain rules which
governs the radiation pattern of an antenna and there are certain rules for the length of the
antenna which antenna is more efficient to converge this what kind of energy into a
electromagnetic energy. So that is covered in the separate course antennas. So, for time
being we just discussed that there is a piece of metal which is used to radiate electrical
energy into the electromagnetic wave. On the receiving part so these waves can actually
travel in the vacuum and when they reach the other side of the far end so these are the
waves which are travelling in vacuum.

And when they come in contact with the receiving antenna these waves again converge
back and generate electrical signal which is equivalent to the wave received. So this is the
receiving part. These waves when they arrive on the antenna they lead to electrical signal
which is shown here in blue and this wave can be detected the external electronic circuit.
This is a similar antenna but now this antenna is acting as the receiver. So, this is how
these two antennas work transmitting and receiving antenna can be of the same size and

292
shape because antenna same antenna can be received for the transmission as well as
reception.

So in for example in the transmitting antenna we connect the antenna to a transmitter

which is electrical transmitter which produces a time varying signal or the sinusoid wave
for example. This sinusoidal wave electrical signal when this is applied to this antenna it
produces a movement of charge along the length of the antenna for example from the
middle to the towards the edge of this antenna. In this process when the charge is moving
this produce a electromagnetic field which goes out of the structure in the vacuum and
which creates this EM waves. So, this time varying electrical signal produces a time
varying electromagnetic field which propagate further towards the right direction and this
is how the electromagnetic waves are produced using this antenna. When these waves
they travel in the vacuum they can travel in air or vacuum.

So, when they hits the similar kind of antenna it may be the same size as the transmitting
antenna. So at the receiving end when these waves hits these receiving antenna they again
produce the similar electromagnetic field or the same electrical signal in the antenna what
was being used to generate them. This electrical signal is produced at the receiving
antenna and now this electrical signal which flows out in terms of a electrical signal for
example sinusoid wave. This sinusoid wave can be read by the receiver which may be a
simple oscilloscope or may be some other measuring instrument. This signal will be
electrical which can now be further processed using external electronic circuit or the
computer.

So, this is how the antenna works the both antennas are same but one work as a
transmitter and one work as a receiver in this unit. So, this is how an antenna works. Now
we have two types of tags what we just discussed. A tag or RFID tag contains certain
information within inside it may be a number may be a bar code may be a code or
something. This RFID tag is of two types one is the passive tag which does not have a
power source inside.

It just has that information buried into a layer and an antenna which is attached to this
circuitry. This does not have any battery or the power source to transmit this information
outside. This is called passive RFID tags. The other type of RFID tags are active RFID
tags where we have a battery placed inside and this battery will provide enough power to
the antenna so that this information can be sent outside without getting energy or without
a requirement of energy from the outside one. So, these are two types of RFID tags
passive and active RFID tags.

So, there are two types of RFID tags. The first one is passive RFID tags. So, these are the
tags which does not have any internal power supply. And generally the power is supplied
by the reader itself. And there are certain merits and demerits of these passive RFID tags.

293
The very first one is the cost. This is less expensive. The size is very small. The power is
supplied by the reader itself. Distance is almost none. The read distance is typically a few
feet.

Personal life typically up to 20 years. And the memory size is around 16 KB or so. So we
have passive RFID tags which do not have any internal power supply and the power is
supplied by the reader itself. Whoever is the device which is trying to read the RFID tag
that has to supply the power to RFID tag. Certain advantage and disadvantage of passive
RFID tag is the cost. The cost is of these RFID tag which is passive.

So, the cost is very less. These are very less expensive compared to the active RFID tags
because we just need a small antenna which receives the RF waves and send it back. Size
is also very small. We do not need any extra circuitry or electronic components and the
battery. So, the size of these RFID tags are very small even in a few mm we can make
these RFID tags.

The power is supplied by the reader itself. So there is no internal power supply available.
Reader has to provide all the power and power is needed because these RFID tags need to
send the information back. So, to sending this information this RFID tag needs certain
power. So, this power is provided by the reader itself. Maintenance is almost none
because there is no component or the circuitry which is decaying or which is degrading
with time.

So, there is almost none maintenance of these RFID tags except normal wear and tear. So
the read distance is typically a few feet because as we increase the distance the strength
of RF signal goes down. So to receive a particular amount of energy for transmission this
RFID tag need to be very close to the antenna or the receiver unit. So the read distance is
limited to a few feet that is the drawback of passive RFID tags. The life is typically 15 to
20 years though there is no component which degrades with time.

But due to normal wear and tear the life is typically 15 to 20 years. And the memory is
around 16 KB or less because we do not have complicated electronic circuitry in within
the RFID tags. So the memory is also very limited. We cannot store much of information
in the RFID tags. So generally the information which is stored in RFID is typically a
number a sequence of numbers that is called key and this key is matched at the receiver
side with the complete information of the object.

So, all the information is stored in the receiver unit and the RFID tag just contain a
particular key or the sequence of numbers. So the memory of passive RFID tags are
limited. This is also a limitation of passive tags. So these are some advantage and
disadvantage.

294
Now let us see how the RFID works. So in RFID tags which is particularly passive type
we have a RFID reader. This reader actually produces a electromagnetic field. Let us say
these are the electromagnetic field generated by the receiving unit. As the distance goes
high the field the strength of the field goes down. So this RFID reader is creating this
electromagnetic field.

When we have a RFID tag which is placed close to this field this RFID tag contains a
special structure which can pick up this RF energy from the electromagnetic field. In this
region magnetic field is affected by the RFID tag data. This RFID tag has a particular
digital number stored which can be in form of binary tag. So this tag ID is a unique
number given to this RFID tag and when we place this RFID tag in the electromagnetic
field this magnetic field is affected by this tag ID or the data stored on this RFID tag.
This whole process is called near field sensing for RFID tags and this happens typically
less than 100 megahertz frequency.

So, this is how this passive RFID tag works. So, in this system we have a RFID reader.
This reader has a antenna or the system which produce this alternating EM field. So this
RFID reader has a certain area where the strength of this electromagnetic field is
sufficient enough. When we have a RFID tag this RFID tag is stored with a tag ID or the
sequence of number which will be in the binary form let us say 011011 something like
that. This tag ID is buried in the RFID tag and this RFID tag may contain some sort of
spiral structure which can be some sort of a coil.

When we place this coil in a alternating electromagnetic field the flux RF flux will be
linked to this coil and this coil will pick up the energy from the electromagnetic field.
Now this energy will provide energy to RFID tag to send the data tag ID to the receiver
unit or the RFID reader unit. So, this is how the data will be transmitted from RFID tag to
the reader. Reader will then connect to the computer and figure out which RFID tag or
which tag ID test based on that the identification will be done. This will be called the near
field sensing for RFID tags because the distance from the receiver and the RFID tag is
not very high it is limited to few feet only.

So, we use for near field sensing and the frequency of operation is typically less than 100
megahertz. We use high very low frequencies here because we need to transmit sufficient
amount of energy from the reader to RFID tags and we use less than 100 megahertz or so
kind of frequencies to excite the coils so that the energy is coupled efficiently to RFID
tags. So, this is how the passive RFID tag works. Let us see the active RFID tags. So
these active RFID tags they have their own internal power source.

And this source is typically in the form of a replaceable battery. One can have a
replaceable battery and there are certain advantage and disadvantage of these active tags.
The very first one is the cost. So, they are more expensive than passive RFID tags.

295
The size is larger. The power is provided by the battery. So this is battery replacement
required. The read distance is up to hundreds of feet. The life is typically few years, one
to three years depending upon the battery. And the memory can be high typically 512 kV
or more. So, RFID tags are the tags where we have an internal power source right in the
tags itself and this power source can be in terms of the replaceable battery with the button
cell or the coin cell.

This battery provides the energy required for transmission on receiving of the data. Now
the cost of RFID tags are higher because we need extra circuitry as well as the power
source on the tag itself. So these are more expensive compared to the passive RFID tags.
The size is also larger because these extra circuitry or the hardware need to be placed on
the tag itself.

So, because of that the size goes up. The power is now provided by the battery only
which is the advantage now in case of active RFID tags because in the passive RFID tags
the energy received from electromagnetic field is very small. So this small energy only
need to be used to transmit the data. However, in active RFID tag we have a dedicated
battery to send the data. So we have a power source available within the tag itself.

So, this is advantage for RFID tags. Maintenance is required here because these batteries
they run out of charge every now and then. So whenever the battery is discharged we
need to replace the battery or charge the battery. So little maintenance is required in the
active RFID tags. The read distance however is increased because we have a power
source on the tag itself.

This power can be used to send the data. So the data can be sent over a longer distance.
So, we have up to hundreds of feet we can send this data typically. The life of these RFID
tags are limited by the battery because they use battery the power source to send the data.
If the battery is drained, then these RFID tags will not work. So typically depending upon
the life of battery 1 to 3 years we need a replacement battery.

So, the typical life depends on the battery. We can replace the battery every 3 years so
they can continue for a longer time. Memory now we can have even larger memories
compared to passive RFID tags because we have now power source and we have a
dedicated electronics placed on the tag itself. This extra electronics can have certain
amount of memory placed inside the tag itself. We can use complex ID types or we can
even put certain amount of information also about the particular object in the RFID tag
itself. However, in the passive RFID tags we were just allowed to store a sequence of
numbers which is a unique number and this unique number was matched at the receiver
side to find out the details.

296
However, in active RFID tags we can put little bit of information in the RFID tag itself
and we need not to do the processing at the later stage. So these are some advantages of
active RFID tags. Let us see how these active RFID tags works.

So let us see how it works. The active RFID tags. This is also called the far field sensing.
Which is typically more than 100 MHz frequency. So in far field sensing or active RFID
tags we have again an RFID reader. This reader has a certain antenna attached to it.

This is a simple dipole antenna. This dipole antenna converts the electrical signal into
electromagnetic field and this electromagnetic field goes out in terms of RF waves. Now
we have a RFID tag which is placed up to few hundreds of feet. Now this RFID tag use
internal power to send the information which is stored in this RFID tag for example this
information. Now this information will be modulated with high frequencies so that it can
be transmitted through this antenna.

This is modulated signal comes out of antenna which can be typically like this. So this
RF signal will come out from this transmitting antenna and this signal will be sent
towards the reader which will be going to detect this signal. So now we have a RFID
reader which is sending and receiving a RF signal through these antennas and the
working of antenna we just saw in the few slides back. Then we have a RFID tag which
is in proximity of this RFID reader and the proximity we say it is hundreds of feet which
is quite a long distance. This RFID tag is using a battery inside along with certain
hardware electronics and antenna is attached to it.

So, this RFID tag has certain information this RFID or the key or the tag ID. This
information is now mixed with high frequencies so that this frequency can be easily
transmitted using this antenna placed on RFID tag. We need to mix it with high
frequencies so that the size of antenna comes down because we know that the length of
antenna is proportional to the wavelength of the signal. So the longer is the wavelength
the longer we need the antenna. So to keep a small size of RFID tag we need to have a
very high frequency so that the wavelength of RF signal is smaller. So this RFID key is
mixed with high frequencies and applied to the antenna that antenna will transmit this
electrical signal in terms of RF wave and this RF wave will reach the reader it can travel
a certain distance because it is powerful enough using the internal battery source.

This wave will be received by the RFID reader and will be converted into electrical
signal back and goes to the computer and the computer does all the processing. So this is
how the active RFID tags works or we call it the far field sensing and the frequency we
generally use is typically high frequencies because the size of antenna need to be small
enough so that we can efficiently transmit the RF signal without increasing the length of
antenna. So typically more than 100 megahertz. This is far field sensing operating at
greater than 100 megahertz.

297
Now let's discuss some of the frequencies what we used with RFID tags. The first one is
the low frequency or LF. These are typically 125 kilohertz. These have shorter read
length and rate but these are less sensitive to interference.

The next is the high frequency or HF. These are typically 13.5 megahertz. The third is the
ultra high frequency or UHF we call it. These are typically 860 to 930 megahertz. They
have faster data rate and the fourth is the microwave where we have the frequency range
is 2.45 gigahertz.

These have highest data read among these but the range is limited of typically Q feet. So,
in these RFID tags we typically use certain frequencies. For example the low frequencies
or LF we call it. The frequency used is 125 kilohertz. The benefit of using low
frequencies these are very less sensitive to interference if there is a another signal nearby.

These low frequencies are not very much distorted. The problem is the short read length
and the rate we cannot read the data very fast using these low frequencies. Then we go to
little higher frequency which is called HF or the high frequencies. The frequency is
typically 13.5 megahertz. Here we have a higher data rate than the LF the low
frequencies. Above this we have UHF or ultra high frequencies where the frequency
range is typically 860 to 930 megahertz and we have a faster data rate in UHF and the
fastest is the microwave. These are the fastest RFID tags where we use microwave
frequencies of 2.45 gigahertz frequency. We have highest data rate in this kind of RFID
tags, but the range of these RFID tags are typically limited.

So, up to a few feet only we can detect the RFID tags if we are using the microwave
signals. So, these are some of the frequencies we used for RFID tags in the system RFID
systems and this is how the RFID tags works for identification. These RFID tags are very
much used in industry for example for identification of objects for identification of
animals and these are also used in commercial places like malls and shops for
identification of a particular object and this RFID tag can be conveniently placed on the
object which because these RFID tags are very small in size they can be very comfortably
placed on the object and we can track all the objects in the complete warehouse or
industry.

So, these are this is the RFID tags and this is all for today.

Thank you.

298
 Transducers For Instrumentation
Prof. Ankur Gupta
Centre for Applied Research in Electronics (CARE)
Indian Institute of Technology, Delhi
Lecture - 36
Mechanical and Thermal Issues

Hello, welcome to the course Transducers for Instrumentation. Today we will discuss some
mechanical and thermal aspects of sensor design. These are the topics which are most of
the time not given enough considerations while designing our sensors and systems. For
example, thermal issues and the associated physics to that. We all know that electronic
circuits or electronic devices. They are all very sensitive to thermal variation or the
temperature variation. If the temperature of a device changes or it increases, then the
performance of this device goes down and accordingly the speed of operation and all other
reliability issues comes into picture.

So, these are the aspects when we need to consider when the temperature of a system goes
high and we need to do certain corrections in our design or we involve certain other type
of things, for example, heat sinks in our design to incorporate these kind of issues. These
thermal issues are always there in all the system design and day by day they are again
becoming more and more challenging because in the earlier technology the heat dissipation
was not a very big issue but going forward to lower and lower technology nodes the heat
dissipation by the device itself is so high that the total heat dissipation within a unit area is
going high because of so many devices are actually fabricated within a single chip. So the
heat dissipation is going high accordingly the temperature will go high so we need to
remove this heat energy from the chip so that the chip can work in its limited range of
temperature and removing this heat from a very confined area to the outside environment
is very critical and is going to be more and more challenging day by day and today we are
going to discuss what are the challenges about removing this heat from the IC and how we
can improve on the heat removal of these systems. So today we are going to discuss some
thermal aspects or the mechanical and thermal aspects of system design. So for example if
we focus on thermal aspects so the physics of semiconductors, heat sinks, is an important
aspect of semiconductor device cooling. In earlier days a simple metallic piece was good
enough. to carry away the excessive heat. However, the continuous more demand and more
demanding interest in the heat to be vacated on the electronic devices brought up the newer
types of heat sinks.

So, for example, specially designed forced convection. along the short fins and guided
directed air flux down the longitudinal fins. So we need to understand the heat sinks and
how heat sink actually perform the remove the heat from the electronic devices and give it
to the outside environment. So we need to understand the physics of heat sink which is an
important aspect of semiconductor device cooling which is a very important topic for

299
reliability point of view. Earlier days however the heat generated by the electronic devices
was not very high. So, a simple piece of metal was good enough if we take a small piece
of metal and put it on the electronic device or the chip. So that was good enough to remove
the heat from the IC because the power density of that semiconductor chip was not very
high. So, a simple metallic piece can dissipate that much of heat. However, as the
technology is growing we want more and more devices to be fabricated on a unit area. So
each device is dissipating some heat.

So overall in a unit area the number of devices are high. Accordingly the power density
which is the power dissipated within that area is also very high compared to the earlier
technologies. So, this heat need to be removed quickly otherwise the performance of the
device which are working there in that area that performance goes down because of the
elevation in temperature. So, day by day this challenge is becoming more and more critical.
So, there are continuous demand to how to evacuate this heat from the electronic devices
and there are newer and newer type of heat sinks are coming up.

For example, there is a specially designed forced convection. So earlier we just had a
metallic sheet which can dissipate the heat but now we have started using forced convection
along the short fins. So, these fins has these heat sinks has small fins and we force the air
through those short fins to cool down or to dissipate the energy to outside environment or
guided directed air flux down the longitudinal fins. So, this is also another type of heat
sinks. So let us discuss a little bit about these heat sinks.

These are the heat sinks nothing but the piece of metal and these metal has little bit of fins
placed on top of it so that it increases the overall area of these heat sinks. So, something
like this. This is just a piece of metal, and these are fins. So, this is conventional heat sink.
The second one is the forced convection with short fins. In this type of design, we have let
us say a complete PCB and on this PCB there is a heat sink which has very short fins and
that is in circular in nature and in between we can place a fan which can force the air
through these fins. So, in this type of arrangement this air will be forced through these fins
and by convection this heat will be removed from these fins. This is forced convection and
let us say the third one is the guided air flux. So in this type of heat sink application, we
have a PCB where all the electronic circuits are there and we need to remove the heat from
them. So, the heat sink is on the one side, and we place a fan on one side and this whole
assembly is now placed such that the air is actually guided in this direction.

This air flux is actually guided through the fins of this heat sink, and this is how the heat is
being removed from the system. So, this is the third type where we have guided air flux
down the longitudinal fins, and this is the third type. So, there are certain considerations
which we need to see while we are designing these heat sinks for better removal of heat
from the electronic circuits. So, these are like thermal conductivity and cooling surface
area must be compared with material and manufacturing cost. Most heat sinks are made

300
from aluminum because of its low thermal resistance light weight and low cost. Cover is
also sometime used for heat sinks although lower in thermal resistance. Then aluminum its
cost and higher weight. So aluminum is less desirable for certain applications. So we have
heat sink which need to be designed. So before designing we need to consider certain
aspects for example the thermal conductivity of the material.

What material we are using for making these heat sinks that material should be chosen
carefully and the thermal conductivity and the how much cooling surface area is that need
to be compared before choosing a heat sink or before designing a heat sink and that also
should be considered along with the material and the manufacturing cost of the heat sink.
For example, most of the heat sinks nowadays are made up of aluminum which is a very
light metal which is very light in weight and it has low thermal resistance which is good
for heat sink application and the cost is also very low for aluminum based heat sinks.
However sometimes we use copper also to make these heat sinks where we have lower
thermal resistance just aluminum but the cost of the copper for the same amount of heat
sink is high and its weight is also higher. Weight is also a critical factor because these heat
sink which need to be placed on top of the package which is placed on a PCB. So if the
weight is higher and because of the thermal vibration this extra weight can cause some
problems.

So aluminum is lightweight so that is more preferable compared to copper heat sinks which
are heavier in weight and the cost is also high for the copper, so this is less desirable in
certain applications. Based on the construction of heat sinks we can have multiple type of
heat sinks which are like this. So based upon geometrical structure we have stamped heat
sink. These are made from a single sheet of metal worked and bent to desired thermal
properties. The second is extruded heat sink. These are cost effective and provide good
thermal performance. The third is bonded fins. These are made by bonding the fins
separated from sheet metal. And the fourth one is the folded fin heat sink. In this type of
heat sink folding the fins over themselves provide a large surface.

So, we have some geometrical structure based on that we can classify some heat sinks. The
very first or the very easy one is the stamped heat sinks where we take a sheet of metal and
we just cut the shape of heat sink using some die and we cut it out and bend this structure
like a heat sink so that can act as a heat sink but the area is limited because whatever is the
area of that sheet that is going to be the area of heat sink to dissipate the energy. This is
very easy to make but the performance is not very high then we have extruded heat sinks
which is more cost effective and this has very good thermal performance. Most of the time
with heat sinks we have certain extra structures which come out of heat sink. These extra
structures increase the surface area of heat sink very dramatically.

So, for example bonded fin heat sink we have a piece of metal and on top of it we bond
certain fin like structure a wall sort of structure very short fins we bond them on the surface.

301
These fins has little volume but the surface area of these fins are very high so that they are
in contact with so much of air and the air will take away the excessive heat from these fins.
The fins are used to increase the surface area of heat sink and there are two types one is the
bonded fin where we take the fin and bond on the piece of metal and this is called bonded
fin heat sink. The other one is the folded fin heat sink where we have these fins and we
fold again these fins again so that a single fin has now double the area of the earlier one.
In this way we can increase the area the surface area of total heat sink and this can
effectively evacuate the heat from the devices the electronic devices.

So let us see some geometrical and thermal design constraints. So, for example, we have a
system where these heat sinks are encapsulated in this closed assembly this is a closed
assembly where the air is flowing from the left to carry away the excessive heat. This is
the direction of air flow and we are looking at the side view. In this assembly we have a
package electronic package which is generating this high amount of heat and we need to
place a heat sink on it. So, if we look at the side view the heat sink will look like this we
will see a single fin here this is the fin and from side view we will see only one fin and the
air is flowing from left to right. However, if we see the front view then we have the system
where this chip is placed and from the side view we will see the heat sink like this. And the
air is flowing going into the board from the front and this in the side view we see only a
single fin however this heat sink has five fins this increases the total area which is in contact
with air. So, this is the whole area which is now in contact with air and air can take away
the excessive heat from all of this surface. So, because of this fins we can see that the area
is very high compared to the volume of metal. So all of this area which is shaded blue here
this is all in contact with air and air can take away the heat from this heat sink.

So, these are some of the design constraints if the air is flowing from left to right the fins
of heat sink need to be aligned in such a way that more amount of air flows through the
heat sink through the fins. So, the fins are fins should be aligned with the air flow direction.
So, this is how we put heat sink in a system. We cannot put a heat sink where the air flow
is from left to right and heat the fins are aligned in the perpendicular direction. So, these
are some of the design constraints while we design a complete system for example a heat
sensitive systems they need to be designed perfectly so that the air flow pattern is aligned
with the direction of fins.

So in forced convection where we are forcing air through the fins the critical parameters
for heat sink design include upstream air temperature which is let us say ta and the velocity.
The cross-section area details such as width and height also play important role in heat sink
design. So, here we see that the heat sink is actually very well aligned with the direction of
air flow so that maximum amount of air is in contact with fin while the air is flowing. To
design these heat sink in the forced convection case forced convection is the case where
we put a local fan very close to the heat sink and this fan is pushing the air through the fins.
So, in this kind of arrangement where we have forced air flow the critical parameters to

302
design the heat sinks these are the incoming air temperature what is the temperature of air
which comes in which is generally the room temperature.

So, the temperature of environment or the temperature of air which is coming in that is a
critical parameter as well as the velocity by which the air is flowing. So that depends on
the performance of fan what fan we are putting if that fan is very strong that can push more
air so the velocity of air will be very high and it can take more amount of heat from the
fins. So, temperature of air and the velocity these are critical parameters while designing
these heat sinks in forced convection case as well as along with this the casing dimensions
for example the cross section area of the casing where we are putting the heat sink that also
plays an important role while designing these heat sinks. So there are certain simulators
available where you can plug in all these details and find out the air flow pattern in the
assembly accordingly we can design the heat sink for that application. So let us see what
are the critical parameters to design the heat sinks.

The important factor for the air flow is the maximum volume the heat sink can occupy
which is nothing, but the width of heat sink multiplied by the length of heat sink multiplied
by the height of heat sink. The design must accommodate the component specifications
such as power dissipation and the maximum junction temperature. Jmax. Also play a
critical role in the design. And another critical factor is the overall cost of thermal solution.
So, these are some of the factors which we considered during the heat sink design the very
first one is how much is the volume available or how much is the space available in the
assembly to put the heat sink which is nothing but the maximum width, maximum length,
and the maximum height of heat sink. So this is the volume which our heat sink can occupy
to remove the heat efficiently. The design while we are designing these heat sinks these
also accommodate the component specifications. For example, the chip set which is sitting
beneath the heat sink what is the specification of that chip set how much is the power it
dissipates for example this is a 1 watt of power which is generated in the chip set or it is 2
watt or it is 5 watt depending upon that power dissipation we need to design the heat sink.
If the power dissipated more we need bigger and bigger heat sinks to remove the heat and
we also should know the junction temperature the maximum junction temperature which
is allowed in that chip set.

Most of these devices the electronic devices the performance degrades with the
temperature. So, there is a certain junction temperature which is specified for those devices
above those temperature we cannot operate that chip set. So that critical junction
temperature or the maximum junction temperature we should know so that we need to keep
the temperature of chip set well below that maximum junction temperature. So in that way
we need to design the heat sink. The weight of heat sink is also important because these
chip sets these are placed on the board using a very small pins or small balls under solder
balls. So, if the weight of heat sink is very high and we put this heavy weight heat sink on
top of chip set this can damage the pins or the chip set. So the weight of the heat sink is

303
also very important while designing and the overall cost of the heating solution is also a
key parameter while designing these thermal solutions. The overall cost includes the heat
sink as well as we put extra fans or motors so to for the forced convection of air. So that
also comes in the thermal solution cost. So that also we need to consider while designing
these heat sinks. So, for example we have a heat sink. So, this is a typical heat sink. And
there are certain parameters, for example this is the width of heat sink in the perpendicular
direction this is the length of heat sink, and this is distance is b from here to here and the
air is flown in this direction inside the fins. The spacing between the fins is s and this the
fin thickness is let us say delta. And this heat sink is now placed on a casing or the chip set.
This chip set is at a temperature T case or T casing and there is a resistance in between this
T case and heat sink which is let us say R internal.

This is the temperature of base of heat sink that is a T base. So, these are some of the
parameters we have assumed for this heat sink. To select a proper heat sink we should
consider following things. To do this it is necessary to know T case. Also, the power
dissipation which is Q of module. This is the first point we need to measure this is the
second and the third one is we need to know the thermal resistance at the casing and heat
sink interface. Which is R internal. So, when we know all these three parameters we can
calculate the heat sink temperature. So, maximum allowed temperature at the surface of
heat sink can be calculated. E base is equal to T case minus Q mode into R in. So, this is
the formula to calculate the temperature at the surface of heat sink. So here we see a heat
sink which has certain fins the width is W length is L and this heat sink is placed on top of
a casing or module. This module is generating the heat which is the in terms of heat
dissipation the power dissipation. This we need to know to design a proper heat sink and
for example this is Q which is the power dissipation dissipated by this module. And the
temperature of this module is let us say T case which is easy to measure.

Now we need to calculate the R in which is the thermal resistance between this T case this
the chip set and the heat sink there is a certain amount of epoxy or paste is generally put
based on that there is a certain thermal resistance that is R in that we need to know and T
case we know by measurement. So, if we know all these parameters, we can easily calculate
the T base which is the temperature at the base of heat sink. This is the maximum allowable
temperature at the heat sink. We need to put this temperature always below T base which
is our critical parameter. So, we need to design our heat sink in such a way that this
temperature T base is always remain the temperature actual temperature at the heat sink
always remain lesser than the T base temperature.

So, heat sinks can be designed by using computer modeling and other softwares we can
design these heat sinks, and this is a very simple structure but we can play around with the
B with S with delta there are many parameters which we can play around to calculate the
thermal resistance of heat sink and accordingly we can design a proper heat sink.

304
So, this is all for today.

Thank you.

305
 THIS BOOK
IS NOT FOR
SALE
NOR COMMERCIAL USE

(044) 2257 5905/08

[Link]
[Link]