CIRCUITS CLASS NOTES

Professor Jau-Horng Chen

Department of Engineering Science and Ocean Engineering
National Taiwan University
Part I Introduction and DC Analysis .................................................................................................................................. 4
Chapter 1 Introduction .......................................................................................................................................................... 4
1.1 Construction of Circuit Models............................................................................................................................... 4
1.2 Engineering Notation ................................................................................................................................................. 5
1.3 Voltage, Current, and Power ................................................................................................................................... 6
1.4 DC and AC ........................................................................................................................................................................ 6
1.5 Circuit Elements ........................................................................................................................................................... 7
1.6 KCL and KVL................................................................................................................................................................. 12
1.7 Earth and Ground ...................................................................................................................................................... 15
Chapter 2 Resistors and Resistive Circuits .................................................................................................................. 16
2.1 Ohm’s Law and the Resistor .................................................................................................................................. 16
2.2 Series and Parallel ..................................................................................................................................................... 17
2.3 Real-World Resistors ............................................................................................................................................... 21
2.4 Norton and Thévenin Equivalent Circuits ....................................................................................................... 25
2.5 Node-Voltage Method .............................................................................................................................................. 29
2.6 Simplification of Circuit Analysis ........................................................................................................................ 30
Chapter 3 Operational Amplifiers ................................................................................................................................... 34
3.1 Operational Amplifier Basics ................................................................................................................................ 34
3.2 Negative Feedback and Virtual Short ................................................................................................................ 37
3.3 Non-Inverting and Inverting Amplifiers .......................................................................................................... 41
3.4 Arithmetic Operations using Operational Amplifiers ................................................................................. 43
3.5 Comparators and Positive Feedback ................................................................................................................. 48
3.6 Data Conversion ......................................................................................................................................................... 51
Part II Transient Analysis ........................................................................................................................................................ 58
Chapter 4 Capacitors and Inductors............................................................................................................................... 58
4.1 Capacitors ..................................................................................................................................................................... 58
4.2 Inductors ....................................................................................................................................................................... 62
4.3 Real-World Capacitors and Inductors ............................................................................................................... 66
Chapter 5 Capacitive and Inductive Circuits............................................................................................................... 69
5.1 First-Order Circuits ................................................................................................................................................... 69
5.2 Second-Order and Higher-Order Circuits ........................................................................................................ 75
5.3 OPA Integrators and Differentiators .................................................................................................................. 84
Part III AC Analysis..................................................................................................................................................................... 87
Chapter 6 Laplace Transform and Sinusoidal Steady State .................................................................................. 87
6.1 Sinusoidal Signals and Fourier Series ............................................................................................................... 88
6.2 First-Order Circuits with Sinusoidal Inputs.................................................................................................... 89
6.3 Laplace Transform for Circuit Analysis ............................................................................................................ 91
6.4 Sinusoidal Steady State and Phasor Transform ............................................................................................ 93
Chapter 7 Impedance and Frequency-Domain Circuit Analysis ......................................................................... 97

2
 7.1 Resistors, Capacitors, and Inductors in the Frequency Domain............................................................. 97
7.2 Impedance and Admittance ................................................................................................................................... 99
7.3 Series and Parallel in the Frequency Domain ............................................................................................. 100
7.4 Sinusoidal Steady-State Analysis Using Impedance and Transfer Function .................................. 102
Chapter 8 Frequency-Selective Circuits..................................................................................................................... 106
8.1 Filter Basics ............................................................................................................................................................... 106
8.2 Passive Filters........................................................................................................................................................... 108
8.3 Active Filters ............................................................................................................................................................. 116
8.4 High-Order Filters .................................................................................................................................................. 121

3
 Part I Introduction and DC Analysis
In this part, a brief introduction to this course is first given. Theories and techniques for analyzing
direct current (DC) circuits are then introduced, with the most important being the Kirchhoff’s circuit
laws. Various passive circuit elements and operational amplifiers then are discussed in detail.

CHAPTER 1 INTRODUCTION

This chapter will introduce the world of circuits and electronics under the assumption that freshman
“General Physics” and “Calculus” have already been completed.

Circuits are normally categorized into lumped circuits and distributed circuits. In this course, only
lumped circuits will be discussed. Distributed circuits are discussed in the Electromagnetics II course
(not provided by our department). For lumped circuits, the physical dimensions of the circuits are far
less than the wavelength of the signal such that the voltage across and the current through the
conductor connecting the circuit elements can be considered as identical.

Electronic circuits are normally categorized into discrete circuits and integrated circuits (ICs). In this
course, both types of electronic circuits are covered. Discrete circuits are normally defined as circuits
that use discrete elements, such as resistors, capacitors, inductors, and transistors to construct a
circuit on a circuit board. ICs normally use semiconductor technology to construct electronic circuits
with all resistors, capacitors, inductors, and transistors fabricated on a semiconductor substrate.
Discrete circuits are circuits that are integrated on a board and are sometimes called hybrid IC.
Whereas integrated circuits fabricated on a monolithic semiconductor substrate are called monolithic
IC. A more detailed introduction on both discrete and integrated circuits is given next semester when
semiconductor circuit elements are introduced.

1.1 CONSTRUCTION OF CIRCUIT MODELS

The first thing to do before analyzing a circuit is to construct a circuit model. For example, the
flashlight shown in Figure 1-1 has a light bulb, a switch, a couple batteries, a few wires, and finally a
plastic casing. Therefore a circuit model can be constructed as shown in Figure 1-2 using circuit
elements. The circuit model in Figure 1-2 assumes the wires are ideal conductors and the plastic
casing has nothing to do with the overall circuit. A more realistic model might consider the batteries
having an internal resistance 𝑅𝑏𝑎𝑡𝑡𝑒𝑟𝑦 , the wires are not ideal having a combined resistance 𝑅𝑤𝑖𝑟𝑒 and
self-inductance 𝐿𝑤𝑖𝑟𝑒 as shown in Figure 1-3. A competent engineer has the ability to construct the
simplest circuit model that gives satisfactory analysis results. A more complex circuit model with
more circuit elements is not always better. Otherwise, one might have to have models that consider
the atomic interaction within the wire inside the flashlight.

4
 Figure 1-1 Photo of a typical flashlight.

Figure 1-2 Circuit model of a flashlight.

Figure 1-3 A more realistic circuit model of a flashlight.

1.2 ENGINEERING NOTATION

Electrical engineers usually prefer using engineering notation over scientific notation to represent
very large or small numbers. Engineering notation uses prefixes that are separated by three orders of
magnitude. Table 1.1 shows the standard prefixes used in engineering notation. One should be aware
that the engineering notation used here is different from the engineering notation used in computers,
where the magnitude prefixes are separated by 210 or 1024 instead of 1000.

5
 Table 1.1 Standard prefixes used in engineering notation

Magnitude Prefixes Abbreviation Magnitude

tera- T 1012
giga- G 109
mega- M 106
kilo- K 10³
milli- m 10-3
micro- μ 10-6
nano- n 10-9
pico- p 10-12
femto- f 10-15

1.3 VOLTAGE, CURRENT, AND POWER

The movement of electrical charges creates current. Current is defined as the amount of electric
charge movement over a specific amount of time, which can be written as
1.1)
𝑑𝑞
𝑖= ,
𝑑𝑡

where 𝑖 is current in amperes (A), 𝑞 is charge in coulombs (C), and 𝑡 is time in seconds. Voltage is
defined as the amount of energy required to separate a certain amount of charges and can be written
as
1.2)
𝑑𝐸
𝑣= ,
𝑑𝑞

where 𝑣 is voltage in volts (V) and 𝐸 is energy in joules (J). Power is defined by the amount of energy
consumed over a specific amount of time. Therefore,
1.3)
𝑑𝐸 𝑑𝐸 𝑑𝑞
𝑃= = ⋅ = 𝑣 ⋅ 𝑖,
𝑑𝑡 𝑑𝑞 𝑑𝑡

where 𝑃 is power in watts (W). Since voltage and current may be functions of time, power may also be
a function of time. When power calculated using the previous equation is positive, it is said that power
is dissipated. Conversely, when the power calculated is negative, it is said that the power is generated.

1.4 DC AND AC

DC stands for direct current and AC stands for alternating current. Direct current means that the
current does not change its polarity and the current is always flowing in one direction. A more

6
commonly used definition for DC these days is constant current or voltage. A mathematical definition
for DC voltage is
1.4)
𝑣(𝑡) = 𝑉𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

and
1.5)
𝑖(𝑡) = 𝐼𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

for DC current, where 𝑉𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 is a constant voltage and 𝐼𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 is a constant current that do not
change with time.

Alternating current means the current alternates polarity. The most common AC voltage or current
waveform is the sine wave such that
1.6)
𝑣(𝑡) = 𝑉𝑚 sin(2𝜋𝑓𝑡),

where 𝑉𝑚 is the magnitude or amplitude of the voltage signal in volts (V) and 𝑓 is the frequency in
hertz (Hz). AC voltage or current does not have to be sine waves, but the average has to always be
zero.

1.5 CIRCUIT ELEMENTS

Circuit elements are normally categorized into active elements and passive elements. However, the
definition of active and passive elements is rather vague. One popular definition of an active element is
“an element that generates power”. Therefore, batteries and generators are considered active
elements. The definition for a passive element is “an element that does not generate power”. Under
this definition, the element may dissipate power or store energy. Resistors, diodes, and transistors are
all circuit elements that dissipate power. Capacitors and inductors are circuit elements that store
energy and are also called energy storage devices, which will be discussed in Chapter 4.

Circuit elements have various number of terminals. The terminals may be connected to one or more
circuit devices through wires. Resistors, capacitors, inductors, batteries, and diodes are two-terminal
devices. Transistors are devices that have three or four terminals. An integrated circuit may have
numerous terminals.

1.5.1 I DEAL S OURCES AND D EPENDENT S OURCES

Sources can be divided into two categories namely ideal sources and dependent sources. Ideal sources
can be further divided into ideal voltage sources and ideal current sources. An ideal voltage source, as
shown in Figure 1-4(a), maintains its voltage regardless of the current and load. The voltage does not
have to be constant and can be a function of time. Similarly, an ideal current source, as shown in
Figure 1-4(b), maintains its current regardless of the voltage and load. Both the ideal voltage source
7
and ideal current source are non-existent. A realistic voltage source such as a battery has internal
resistance and is normally modelled as an ideal voltage source in series with a small resistor as shown
in Figure 1-4(c). A realistic current source is normally constructed with transistors and is normally
modelled as an ideal current source in parallel with a large resistor as shown in Figure 1-4(d).

(a) ideal voltage source (b) ideal current source

(c) realistic voltage source (d) realistic current source

Figure 1-4 Ideal and realistic sources.

A dependent source is a voltage or current source with its value dependent on a voltage or current
elsewhere. There are four types of dependent sources. The first is a voltage-controlled voltage source
(VCVS) as shown in Figure 1-5(a). A VCVS can be defined by
1.7)
𝑣𝑜 = 𝐴𝑣 𝑣𝑖 ,

where 𝐴𝑣 is a dimensionless factor called voltage gain. 𝐴𝑣 can be a function of the input voltage 𝑣𝑖 ,
temperature, or time and might not be a constant. An ideal voltage amplifier is basically a VCVS with a
constant voltage gain. An operational amplifier (OPA) is a widely seen voltage amplifier. The second is
a current-controlled current source (CCCS) as shown in Figure 1-5(b). A CCCS can be defined by
1.8)
𝑖𝑜 = 𝐴𝑐 𝑖𝑖 ,

where 𝐴𝑐 is a dimensionless factor called current gain. 𝐴𝑐 might be a function and not a constant. An
ideal current amplifier is basically a CCCS with a constage current gain. A bipolar junction transistor
(BJT) is a widely seen current amplifier. However, the current gain of a BJT is a nonlinear function of
input current, temperature, and time. The third is a voltage-controlled current source (VCCS) as
shown in Figure 1-5 (c). A VCCS can be defined by
1.9)
𝑖𝑜 = 𝐺𝑚 𝑣𝑖 ,

where 𝐺𝑚 is called transconductance with dimension of ampere/volt (A/V). 𝐺𝑚 can be a function and

8
might not be constant. An ideal transconductance amplifier is basically a VCCS with a constant
transconductance. A field-effect transistor (FET) is a widely seen transconductance amplifier.
However, the transconductane gain of a FET is a function of input voltage, temperature, and time. The
fourth is a current-controlled voltage source (CCVS) as shown in Figure 1-5 (d). A CCVS can be defined
by
1.10)
𝑣𝑜 = 𝑅𝑚 𝑖𝑖 ,

where 𝑅𝑚 is called transresistance with dimension of volt/ampere (V/A). 𝑅𝑚 can be a function and
not a constant. An ideal transresistance amplifier is basically a CCVS with a constant transresistance.
The CCVS is the most rarely seen among the four types of dependent sources.

(a) VCVS (b) CCCS

(c) VCCS (d) CCVS

Figure 1-5 Schematic symbols of dependent sources.

1.5.2 R ESISTORS , C APACITORS , I NDUCTORS , AND T RANSFORMERS

The ideal resistor as shown in Figure 1-6(a) is a two-terminal device that is governed by the Ohm’s
Law
1.11)
𝑣 = 𝑖𝑅,

where 𝑅 is resistance in ohms (Ω). A resistor always dissipates power and generate heat when the
voltage across or the current through it is not zero. The power dissipated in a resistor can be written
as
1.12)
𝑣2
𝑃 =𝑣∙𝑖 = = 𝑖 2 𝑅.
𝑅

The process of dissipating electric energy through resistance to generate heat is called ohmic heating.

The ideal capacitor as shown in Figure 1-6 (b) is a two-terminal device that is governed by the
equation

9
1.13)
𝑞 = 𝐶𝑣,

where 𝐶 is capacitance in farads (F). The previous equation can also be written as a differential
equation
1.14)
𝑑𝑞 𝑑𝑣
=𝐶 .
𝑑𝑡 𝑑𝑡

From the definition of current in Eq. 1.1, it can be concluded that

1.15)
𝑑𝑞 𝑑𝑣
𝑖= =𝐶 .
𝑑𝑡 𝑑𝑡

An ideal capacitor neither dissipates nor generates power. It is an energy-storage device and will be
discussed in detail in Chapter 4.

The ideal inductor as shown in Figure 1-6(c) is a two-terminal device that is governed by the equation
1.16)
Φ = 𝐿𝑖,

where Φ is magnetic flux in webers (Wb) and 𝐿 is inductance in henrys (H). The previous equation
can also be written as a differential equation
1.17)
𝑑Φ 𝑑𝑖
=𝐿 .
𝑑𝑡 𝑑𝑡

From Faraday’s law, it is known that

1.18)
𝑑Φ
𝑣= ,
𝑑𝑡

which easily leads to

1.19)
𝑑Φ 𝑑𝑖
𝑣= =𝐿 .
𝑑𝑡 𝑑𝑡

An ideal inductor neither dissipates nor generates power. It is an energy-storage device like a
capacitor and will also be discussed in more detail in Chapter 5.

The ideal transformer as shown in Figure 1-6(d) is a four-terminal device that is governed by
1.20)
𝑑Φ
𝑣1 = 𝑁1
𝑑𝑡
10
and
1.21)
𝑑Φ
𝑣2 = 𝑁2 ,
𝑑𝑡

where 𝑁1 , 𝑁2 and 𝑣1 , 𝑣2 are the number of turns and voltages in the primary and secondary parts of
the transformer. Therefore, the voltage ratio between the secondary and primary part of the
transformer can be written as
1.22)
𝑣2 𝑁2
= .
𝑣1 𝑁1

If 𝑁2 is larger than 𝑁1 , then the transformer is a step-up transformer when the output voltage (𝑣2 ) is
greater than the input voltage (𝑣1 ). Conversely, the transformer is a step-down transformer when the
output voltage (𝑣2 ) is less than the input voltage (𝑣1 ).

(a) resistor (b) capacitor (c) inductor (d) transformer

Figure 1-6 Schematic symbols of various circuit elements.

1.5.3 D IODES

A diode is a two terminal device. The two terminals are called the cathode and anode as shown in
Figure 1-7. Ideally, when a diode is forward biased, it is a short circuit. Whereas, when a diode is
reverse biased, it is an open circuit. The Shockley diode equation

1.23)
𝑣𝐷
𝑖𝐷 = 𝐼𝐷0 (𝑒 𝑛𝑉𝑇 − 1)

is used to model a diode both under forward bias and reverse bias, where 𝐼𝐷0 is the saturation current
of the diode, 𝑉𝑇 is the thermal voltage of around 25 mV at room temperature, 𝑛 is called ideality factor,
which is between 1 and 2 for diodes made of silicon. It can be clearly seen that Eq. 1.23 is a nonlinear
function unlike the aforementioned circuit elements. Another widely used nonlinear circuit element is
the transistor which will be discussed in another class.

Figure 1-7 Circuit symbol of the diode.

11
 1.6 KCL AND KVL

The Kirchhoff’s circuit laws are named after German physicist Gustav Kirchhoff. All analyses of circuits
are based on these two laws and all circuits must obey these two laws.

1.6.1 N ODES AND B RANCHES

Before dealing with the Kirchhoff laws, nodes and branches must first be defined. A node is defined as
a point where two or more circuit elements join. For circuits with ideal wires connecting various
circuit elements, the wires can be a node or part of a node. For the circuit in Figure 1-8, a is a node
connecting the 20mA current source and 25kΩ resistor. However, b and c is actually a single node (bc)
since they are connected with a single wire and has no circuit elements in between. Similarly, d and e
is a single node (de) and f, g, h, i, and j is also a single node (fghij). Therefore, the circuit in Figure 1-8
has a total number of four nodes.

Figure 1-8 Example circuit for nodes and branches.

A path is a connection between any two nodes. A branch is a path between two nodes that does not
pass through any other nodes. A closed path or loop is a path that starts and ends at the same node
while passing through each node only once. In the circuit shown in Figure 1-8, a-b, b-i, c-h, c-d, d-g, e-f,
and a-j are all branches and the circuit has a total of seven branches. The circuit has a total of ten
closed paths or loops, namely a-b-i-j-a, a-b-c-h-i-j-a,a-b-c-d-g-h-i-j-a, a-b-c-d-e-f-g-h-i-j-a, b-c-h-i-b,
b-c-d-g-h-i-b, b-c-d-e-f-g-h-i-b, c-d-g-h-c,c-d-e-f-g-h-c, and d-e-f-g-d. For a very simple circuit like
this, it is quite obvious that nodes are easier to find than loops.

1.6.2 K IRCHHOFF ’ S C URRENT L AW

The Kirchhoff’s current law (KCL) is stated as the sum of all currents flowing into or out of a circuit
node is zero, which can be written in the mathematic form of
1.24)
Σ𝑖 = 0.

A variation of KCL is that the sum of all currents flowing into a circuit node equals to the sum of all
currents flowing out of a circuit node, which can be written as
1.25)
Σ𝑖𝑖𝑛 = Σ𝑖𝑜𝑢𝑡 .

12
The circuit shown in Figure 1-9 is the same circuit as Figure 1-8 with additional labels for nodes and
branch currents. For node ○
1 , the sum of all currents flowing into it can be written as 𝑖 + (−𝑖 ) = 0,
1 2

which is in the form of Eq. 1.24. By reorganizing the previous results, the form of Eq. 1.25 can be
obtained, which is 𝑖1 = 𝑖2 . For node ○
2 , KCL in the form of Eq. 1.25 can be easily obtained as

𝑖2 + 𝑖4 = 𝑖3 + 𝑖5 . For node ○
3 , KCL in the form of Eq. 1.25 can be easily obtained as 𝑖5 = 𝑖6 + 𝑖7 .

Figure 1-9 Example circuit for KCL.

The circuit in Figure 1-10(a) consists of two ideal current sources. The current flowing into and out of
a node is obviously unequal and disobeys KCL. Therefore, the circuit is impossible to exist. However,
ideal current sources are actually non-existent. If the two current sources are realistic current sources
with 1MΩ of parallel resistances as shown in Figure 1-10(b), the circuit would obey KCL and the
resistances would each have 0.5A of current flowing through them.

(a) Ideal current sources (b) Realistic current sources

Figure 1-10 KCL of current sources.

1.6.3 K IRCHHOFF ’ S V OLTAGE L AW

The Kirchhoff’s voltage law (KVL) is stated as the sum of all branch voltages in a closed path or loop is
zero, which can be written in the mathematic form of
1.26)
Σ𝑣 = 0

Another way of expressing KVL is that the sums of all branch voltages of two paths sharing the same
starting and ending node that can form a closed loop are equal, which can be written as
1.27)
Σ𝑣𝑝𝑎𝑡ℎ1 = Σ𝑣𝑝𝑎𝑡ℎ2 .

The voltage difference between two nodes ○

1 and ○
2 is defined as

13
1.28)
𝑣12 = 𝑣1 − 𝑣2 .

For a loop ○
1 -○
2 -○
3 -○
1 , the sum of all branch voltage can be written as

1.29)
Σ𝑣 = 𝑣12 + 𝑣23 + 𝑣31 = (𝑣1 − 𝑣2 ) + (𝑣2 − 𝑣3 ) + (𝑣3 − 𝑣1 ),

which obviously is zero. Similar to KCL, KVL must always be obeyed. For the circuit in Figure 1-11(a),
the voltage differences between the two paths are different and therefore disobey KVL, which makes it
impossible to exist. If the ideal voltages sources are replaced with realistic ones each with 1Ω of series
resistance, KVL would be followed, and the resistors each have a 0.5V voltage drop.

(a) Ideal voltage sources (b) Realistic voltage sources

Figure 1-11 KVL of voltage sources.

EX. 1.1
With both KCL and KVL introduced, we can now try to solve the example circuit in Figure 1-9. Assume
we are interested in finding 𝑖1 and 𝑖7 . We can start by finding the KCL equations for nodes ○
1 ,○
2 , and

○
3 .

𝑖1 = 𝑖2 … (𝑎)
{𝑖2 + 𝑖4 = 𝑖3 + 𝑖5 ⇒ 𝑖2 + 20mA = 𝑖3 + 𝑖5 … (𝑏)
𝑖5 = 𝑖6 + 𝑖7 ⇒ 𝑖5 = 10mA + 𝑖7 … (𝑐)
There is no need to find the KCL equation for node ○
4 , since it is actually the sum of the previous three

equations. For a circuit with 𝑁 nodes, only 𝑁 − 1 simultaneous KCL equations can be found. Since we
have five unknown variables and only three simultaneous KCL equations, we need to find two more
KVL equations to solve all of the variables.
2 − 𝑖2 ∙ 25kΩ − 𝑖3 ∙ 10kΩ = 0 … (𝑑)
{
2 − 𝑖2 ∙ 25kΩ − i5 ∙ 5kΩ − i7 ∙ 20kΩ = 0… (𝑒)
By substituting (c) into (b) and (e),
𝑖2 = 𝑖3 + 𝑖7 − 10mA … (𝑓)
{𝑖2 ∙ 25kΩ + 𝑖3 ∙ 10kΩ = 2 … (𝑔)
𝑖2 ∙ 25kΩ + i7 ∙ 25kΩ = −48 … (ℎ)
By substituting (f) into (g) and (h),
𝑖 ∙ 35kΩ + 𝑖7 ∙ 25kΩ = 252 … (𝑖)
{3
𝑖3 ∙ 25kΩ + 𝑖7 ∙ 50kΩ = 202 … (𝑗)
Therefore,
252 × 2 − 202
𝑖3 = = 6.71mA
45kΩ
and
14
 𝑖7 = 0.68mA.
Finally,
𝑖1 = 𝑖2 = 𝑖3 + 𝑖7 − 10mA = 6.71mA + 0.68mA − 10mA = −2.61mA.

1.7 EARTH AND GROUND

Voltage, or more specifically relative voltage, is defined as the potential difference between two nodes
or points. Theoretically, the reference node can be arbitrary chosen. However, the earth is a preferred
choice. The absolute voltage of a node is defined as the voltage difference between the node and the
earth. In circuit analysis, the earth or ground is considered an ideal reference voltage such that it
remains zero no matter how much current it sinks or sources. Since the ground connection to earth is
never ideal, a voltage difference always exists. The circuit symbol of ground is shown in Figure 1-12.

Figure 1-12 Circuit symbol of ground.

The reference point of a circuit is often called “ground” even if no physical connection exists between
them. For instance, a portable electronic device that is battery operated, the negative terminal of the
battery is connected to a large electrically conductive surface and is referred to as ground. For safety
reason and noise reduction, appliances are often not connected directly to the outlet. Instead, a
transformer is used to isolate the main portion of the circuit and the power source. The primary side
of the transformer is connected to the power source while the load or main portion of the circuit is
connected to the secondary side. A large electrically conductive surface connected to the secondary
side is referred to as ground or zero potential.

15
 CHAPTER 2 RESISTORS AND RESISTIVE CIRCUITS

This chapter will introduce resistors and resistive circuits. Two important concepts, equivalent
resistance and equivalent circuit, will be presented and discussed in detail. These two concepts will be
used extensively throughout this course.

2.1 OHM’S LAW AND THE RESISTOR

Ohm’s law is named after German physicist Goerg Ohm. It states that the current through a conductor
or resistor is directly proportional to the voltage. Therefore, we define
2.1)
𝑣
𝑅=
𝑖

as resistance in ohms (Ω) or V/A. We can also define

2.2)
𝑖 1
𝐺= =
𝑣 𝑅

as conductance in siemens (S) or A/V. The circuit symbol of a resistor is shown in Figure 2-1.

Figure 2-1 Circuit symbol of a resistor.

The resistance of a uniform material is defined as

2.3)
𝐿
𝑅=𝜌 ,
𝐴

where 𝜌 is resistivity in ohm-meter (Ω ∙ m), 𝐿 is length in meters (m), and 𝐴 is area in square meters
(m2 ). The reciprocal of resistivity is conductivity
2.4)
1
𝜎= ,
𝜌

which has a unit of siemens per meter (S/m). The conductivity and resistivity of some commonly seen
materials are summarized in Table 2.1.

16
 Table 2.1 Conductivity and resistivity of common materials

Category Material Conductivity Resistivity

Conductor Silver 6.2 × 107 1.6 × 10−8

Copper 5.9 × 107 1.7 × 10−8
Gold 4.4 × 107 2.3 × 10−8
Aluminum 3.7 × 107 2.8 × 10−8
Semiconductor Germanium 2.2 0.46
Silicon −4 2300
4.3 × 10
(Semi-insulator) Gallium Arsenide −6 106
10
Insulator Rubber 10−12 1012

2.2 SERIES AND PARALLEL

2.2.1 R ESISTORS IN S ERIES

When resistors are in series as shown in Figure 2-2, the currents flowing all resistors are identical
because of KCL.

Figure 2-2 Resistors in series.

Therefore, the total voltage across all the resistors can be written as
2.5)
𝑣 = 𝑣1 + 𝑣2 + ⋯ + 𝑣𝑘
= 𝑖1 𝑅1 + 𝑖2 𝑅2 + ⋯ + 𝑖𝑘 𝑅𝑘
= 𝑖𝑅1 + 𝑖𝑅2 + ⋯ + 𝑖𝑅𝑘 .

The equivalent resistance of resistors in series can then be defined as

17
2.6)
𝑣
𝑅𝑒𝑞 =
𝑖
= 𝑅1 + 𝑅2 + ⋯ + 𝑅𝑘
𝑘

= ∑ 𝑅𝑛 .
n=1

2.2.2 R ESISTORS IN P ARALLEL

When resistors are in parallel as shown in Figure 2-3, the voltages across all resistors are identical
because of KVL. Therefore,
2.7)
𝑣 = 𝑣1 = 𝑣2 = ⋯ = 𝑣𝑘 .

From KCL,
2.8)
𝑖 = 𝑖1 + 𝑖2 + ⋯ + 𝑖𝑘
𝑣1 𝑣2 𝑣𝑘
= + + ⋯+
𝑅1 𝑅2 𝑅𝑘
𝑣 𝑣 𝑣
= + + ⋯+ .
𝑅1 𝑅2 𝑅𝑘

Therefore, the equivalent resistance of resistors in parallel can then be defined as

2.9)
𝑣
𝑅𝑒𝑞 =
𝑖
𝑣
= 𝑣 𝑣 𝑣
𝑅1 + 𝑅2 + ⋯ + 𝑅𝑘
1
=
1 1 1
𝑅1 + 𝑅2 + ⋯ + 𝑅𝑘
1
= .
𝑘 1
Σn=1 𝑅 𝑛

From Eqs. 2.2 and 2.9, the equivalent conductance can then be expressed as
2.10)
𝑘

𝐺𝑒𝑞 = ∑ 𝐺𝑛 .
n=1

18
 Figure 2-3 Resistors in parallel.

The operator “∥” is often used for resistors in parallel. For example, if 𝑅𝑎 and 𝑅𝑏 are in parallel, the
equivalent resistance is
2.11)
𝑅𝑒𝑞 = 𝑅𝑎 ∥ 𝑅𝑏
1
=
1 1
𝑅𝑎 + 𝑅𝑏
𝑅𝑎 𝑅𝑏
= .
𝑅𝑎 + 𝑅𝑏

If 𝑅𝑎 , 𝑅𝑏 , and 𝑅𝑐 are in parallel, the equivalent resistance is

2.12)
𝑅𝑒𝑞 = 𝑅𝑎 ∥ 𝑅𝑏 ∥ 𝑅𝑐
1
=
1 1 1
𝑅𝑎 + 𝑅𝑏 + 𝑅𝑐
𝑅𝑎 𝑅𝑏 𝑅𝑐
= .
𝑅𝑎 𝑅𝑏 + 𝑅𝑏 𝑅𝑐 + 𝑅𝑐 𝑅𝑎

2.2.3 V OLTAGE D IVIDER AND C URRENT D IVIDER

The voltage divider as shown in Figure 2-4 is a simple circuit that can generate an arbitrary voltage
difference between zero and the input voltage. From KVL,
2.13)
𝑣 = 𝑣1 + 𝑣2 = 𝑖𝑅1 + 𝑖𝑅2 .

Therefore,
2.14)
𝑅1
𝑣1 = 𝑖𝑅1 = 𝑣
𝑅1 + 𝑅2

and
2.15)
𝑅2
𝑣2 = 𝑖𝑅2 = 𝑣.
𝑅1 + 𝑅2

19
By changing the ratio of 𝑅1 and 𝑅2 , the voltages 𝑣1 and 𝑣2 can be changed.

Figure 2-4 Resistive divider circuit.

The current divider is shown in Figure 2-5. Similar to a voltage divider, the current divider can
generate an arbitrary current between zero and the input current. From KCL,
2.16)
𝑣 𝑣 𝑅1 + 𝑅2
𝑖 = 𝑖1 + 𝑖2 = + = 𝑣.
𝑅1 𝑅2 𝑅1 𝑅2

Therefore,
2.17)
𝑣 𝑅1 𝑅2 1 𝑅2
𝑖1 = =( )( )𝑖 = 𝑖
𝑅1 𝑅1 + 𝑅2 𝑅1 𝑅1 + 𝑅2

and
2.18)
𝑣 𝑅1 𝑅2 1 𝑅1
𝑖2 = =( )( )𝑖 = 𝑖.
𝑅2 𝑅1 + 𝑅2 𝑅2 𝑅1 + 𝑅2

By changing the ratio of 𝑅1 and 𝑅2 , the currents 𝑖1 and 𝑖2 can be changed.

Figure 2-5 Current divider circuit.

2.2.4 E QUIVALENT R ESISTANCE

Equivalent resistance is a very important concept and tool for analyzing circuits, especially analog
electronic circuits. For a multistage circuit, the latter stage is often only a load to the prior stage. If the
latter stage can be treated as an equivalent resistance during analysis of the prior stage, calculations
would be a lot easier than analyzing the whole circuit. In reality, ICs are always designed in such a
fashion with a design involving tens or hundreds of design engineers.

The simplest way to find the equivalent circuit of a resistive network is to use series and parallel.
20
However, not all circuit can be applied. The circuits in Figure 2-6(a) and Figure 2-6(b) are two
examples. In Figure 2-6(a), resistors are configured neither in series nor in parallel. For the circuit in
Figure 2-6(b), a dependent source coexists with a resistor, which also cannot be simplified with series
or parallel. A general method for finding the equivalent resistance is to use a test source as in Figure
2-7. Either a current source or a voltage source can be used and the equivalent resistance can be
calculated as
2.19)
𝑣𝑡𝑒𝑠𝑡
𝑅𝑒𝑞 = .
𝑖𝑡𝑒𝑠𝑡

(a) (b)
Figure 2-6 Equivalent resistances that cannot be found using series and parallel calculations.

(a) Usage of test voltage (b) Usage of test current

Figure 2-7 Calculation of equivalent resistance using test sources.

2.3 REAL-WORLD RESISTORS

Resistors can be made by almost any material. The most commonly seen resistors are introduced here.

2.3.1 C ARBON F ILM R ESISTORS

Carbon film resistors are the most commonly seen resistors in the laboratory. Carbon film resistors
are made by wrapping a thin carbon film around an insulating core. These resistors are axial-lead
devices. The technology that uses axial-lead devices is called through-hole technology. Because of the
long leads in axial-lead devices, the devices normally have more parasitic inductances. The resistor
value of carbon film resistors can be identified using the color code on the resistor. The resistors have
three color bands close together on one side and one further away on the other side as shown in

21
Figure 2-8. The lone color band on one side represents the precision or tolerance of the resistor. The
other three on the other side represents the resistor value. The numbers represented by the color
codes are summarized in Table 2.2. Color codes 𝒂 and 𝒃 represents the significant figures and color
code 𝒄 represents the multiplier such that
2.20)
𝑅 = (10𝒂 + 𝒃) × 10𝒄 .

For the resistor shown in Figure 2-8, 𝒂 is 1 represented by the brown band and both 𝒃 and 𝒄 are 0
represented by the black bands. Therefore, the resistor is a 10Ω resistor. The gold band on the far
right shows that this resistor has a ±5% tolerance.

Figure 2-8 A 10 ohm carbon film resistor.

Table 2.2 Resistor color codes

Color Number Multiplier Tolerance

Black 0 100
Brown 1 101 ±1%
Red 2 102
Orange 3 103
Yellow 4 104
Green 5 105 ±0.5%
Blue 6 106
Purple 7 107
Grey 8 108
White 9 109
Silver 10-2 ±10%
Gold 10-1 ±5%

2.3.2 T HICK F ILM R ESISTORS

Thick film resistors are the most widely used resistors in the industry now. The resistors are
manufactured by depositing a thick film of resistive paste onto an insulating substrate. The resistors
are in the form of surface mount devices (SMD) that can be soldered with a reflow oven. Table 2.3
summarizes the dimension of widely seen SMD packages. Thick file resistors with package sizes larger
than 0402 usually have their resistance value printed on the resistor as shown in Figure 2-9. Typical
tolerances range from 0.5% to 5%.

22
 Table 2.3 Dimensions of SMD packages

Package Imperial dimensions Metric dimensions

0201 20mil x 10mil 0.51mm x 0.25mm

0402 40mil x 20mil 1.02mm x 0.51mm
0603 60mil x 30mil 1.52mm x 0.76mm
0805 80mil x 50mil 2.03mm x 1.27mm
1206 120mil x 60mil 3.04mm x 1.52mm
1210 120mil x 100mil 3.04mm x 2.54mm

Figure 2-9 An 0603 220Ω resistor and an 0402 180 Ω resistor.

2.3.3 T HIN F ILM R ESISTORS

Thin film resistors are sometimes called metal film resistors. This kind of resistor is made by
depositing a thin film of metallic material onto an insulating substrate. The resistance value can be
tightly controlled and typical tolerances range from 0.01% to 0.5%. Thin film resistors are very
expensive and are normally used only in applications requiring very high precision.

2.3.4 W IRE W OUND R ESISTORS

Wire wound resistors are made by winding a metal wire, usually nickel and chromium alloy, to make a
resistor. Wire wound resistors often have a ceramic housing that look like cement and are often called
cement resistors even though it does not contain any cement. Wire wound resistors are used in high
power applications.

23
 Figure 2-10 Cement resistors.

2.3.5 V ARIABLE R ESISTORS

A two-terminal variable resistor is basically the same as a three-terminal potentiometer except that
the variable resistor only uses two terminals of the potentiometer. The circuit symbol of a variable
resistor is shown in Figure 2-11(a) and the circuit symbol of a potentiometer is shown in Figure
2-11(b). For a potentiometer, the resistance between terminals a and c is constant. When the shaft or
screw of the potentiometer is rotated, the resistance between a and b, 𝑅𝑎𝑏 , and the resistance between
b and c, 𝑅𝑏𝑐 , are changed. A variable resistor is basically taking either 𝑅𝑎𝑏 or 𝑅𝑏𝑐 for use. The
potentiometer is a voltage divider by itself. Therefore, by rotating the shaft or screw, the voltage of
terminal b can be any voltage between the voltages of a and c. The potentiometer is widely used in
joysticks. The voltage of terminal b is used to tell the position or angle of the joystick. A variable
resistor is sometimes called a trimmer, which can be used to trim the performance of a circuit.

(a) Variable resistor (b) Potentiometer

Figure 2-11 Circuit symbols of variable resistor and potentiometer.

Figure 2-12 A typical potentiometer.

24
 2.4 NORTON AND THÉVENIN EQUIVALENT CIRCUITS

Similar to equivalent resistance, equivalent circuit is also a very important concept and tool for
analyzing circuits, especially analog electronic circuits. Unlike equivalent resistance that can only
model passive elements like a resistive network. The Norton and Thévenin equivalent circuits
discussed here have an additional element or term to model the independent sources inside a circuit.
Two circuits are equivalent circuits when the voltages and currents are always identical when
connected to the same load.

2.4.1 N ORTON E QUIVALENT C IRCUIT

The Norton equivalent circuit was developed by American engineer Edward Lawry Norton in the
1920s. Norton’s theorem states that any linear circuit can be expressed by an equivalent current
source (𝑖𝑠𝑐 ) and an equivalent resistance (𝑅𝑒𝑞 ) in parallel. If the circuits in Figure 2-13(a) and Figure
2-13(b) are equivalent, the voltage across and the current through 𝑅𝐿 should be identical. When 𝑅𝐿 is
short circuit, the current through 𝑅𝐿 equals to 𝑖𝑠𝑐 in Figure 2-13(b). When 𝑅𝐿 is open circuit, the
voltage across 𝑅𝐿 is 𝑣𝑜𝑐 = 𝑖𝑠𝑐 𝑅𝑒𝑞 . Therefore, the Norton equivalent circuit can be derived by
measuring the open-circuit voltage and the short-circuit current.

(a) unknown circuit (b) Norton equivalent circuit (c) Thévenin equivalent circuit
Figure 2-13 Thévenin and Norton equivalent circuits

2.4.2 T HÉVENIN E QUIVALENT C IRCUIT

The Thévenin equivalent circuit was developed by French engineer Léon Charles Thévenin.
Thévenin’s theorem states that any linear circuit can be expressed by an equivalent voltage source
(𝑣𝑜𝑐 ) and an equivalent resistance (𝑅𝑒𝑞 ) in series. If the circuits in Figure 2-13(a) and Figure 2-13(c)
are equivalent, the voltage across and the current through 𝑅𝐿 should be identical. When 𝑅𝐿 is short
circuit, the current through 𝑅𝐿 equals to 𝑖𝑠𝑐 = 𝑣𝑜𝑐 /𝑅𝑒𝑞 in Figure 2-13 (c). When 𝑅𝐿 is open circuit, the
voltage across 𝑅𝐿 equals to 𝑣𝑜𝑐 . Therefore, the Thévenin equivalent circuit can also be derived by
measuring the open-circuit voltage and the short-circuit current. Moreover, it might be obvious that
the Thévenin equivalent circuit and the Norton equivalent circuit are equivalent circuits themselves.
From the previous results, the V-I relationship of an arbitrary linear circuit is shown in Figure 2-14.

25
 Figure 2-14 V-I relationship of a linear circuit.

EX. 2.1
Assume the following two circuits are covered and cannot be seen. Using what you have learnt, how
can the two circuits be distinguished from each other?

The circuit on the left is actually the Thévenin equivalent circuit of the circuit on the right. No matter
what load is connected, the voltages and currents will be the same for the two circuits.

However, without any load connected, the 1.5V battery provides no current, while the 3V battery will
provide 10mA of current. As a result, the two 150Ω resistors will dissipate power and generate heat.
Therefore, the circuit on the right will be hotter than the one on the left.

The goal of this example is to show the concept that equivalent circuits are not the same circuits. Even
though from the outside or load point of view, the two circuits behave identically, internally they may
operate completely different.

2.4.3 S OURCE T RANSFORMATION

Source transformation is a way to simplify circuit analysis using Norton and Thévenin equivalent
circuits. Since Figure 2-13(b) and Figure 2-13(c) are equivalent circuits, they can be interchanged to
simplify circuit analysis. The easiest way to explain source transformation is through an example.

26
EX. 2.2

We will continue to use the example circuit in Figure 1-9. Assume we are interested in finding 𝑖1
and 𝑖7 .

We first perform source transformation to find 𝑖1 .

27
Therefore,
2 − 85.72
𝑖1 = ≈ −2.60mA.
25kΩ + 7.143kΩ

Similarly for 𝑖7 ,

Using current division,

12.14kΩ
𝑖7 = 1.81mA ∙ ≈ 0.68mA
12.14kΩ + 20kΩ

28
 2.5 NODE-VOLTAGE METHOD

The node-voltage method is a systematic method of writing equations for circuit analysis. The node-
voltage uses essential nodes for KCL analysis. An essential node is a node where three or more
elements join. The node-voltage method uses the following four steps for circuit analysis:
1. Find essential nodes.
2. Use the essential node with the most branches as reference (not required, but recommended).
3. Define node voltages.
4. Use KCL on all essential nodes other than the reference.

EX. 2.3

We will continue to use the example circuit in Figure 1-9. Assume we are interested in finding 𝑖1
and 𝑖7 .
1. The essential nodes are ○
2,○
3 , and ○
4.

2. We pick node ○
4 as the reference node.

3. Define the node voltage at ○

2 as 𝑉2 and node voltage at ○
3 as 𝑉3 .

4. Write the KCL equations for nodes ○

2 and ○
3
2 − 𝑉2 𝑉2 𝑉2 − 𝑉3
+ 20mA = +
{ 25kΩ 10kΩ 5kΩ
𝑉2 − 𝑉3 𝑉3
= 10mA +
5kΩ 20kΩ
4 − 2𝑉2 + 1000 = 5𝑉2 + 10𝑉2 − 10𝑉3
⇒{
4𝑉2 − 4𝑉3 = 200 + 𝑉3
17𝑉2 − 10𝑉3 = 1004
⇒{
4𝑉2 − 5𝑉3 = 200
1004 − 2 × 200 604
𝑉2 = = ≈ 67.1V
17 − 2 × 4 9
⇒ 604
4× − 200 616
𝑉3 = 9 = ≈ 13.7V
{ 5 45
2 − 𝑉2
⇒ 𝑖1 = 𝑖2 = ≈ −2.60mA
25kΩ
𝑉3
⇒ i7 = ≈ 0.68mA
20kΩ

29
 2.6 SIMPLIFICATION OF CIRCUIT ANALYSIS

2.6.1 S UPERPOSITION

One way of simplifying circuit analysis is through the use of superposition. The superposition theorem
states that for a linear circuit, the total response of the circuit is the sum of the individual responses of
the circuit. A linear circuit is a circuit that contains only linear circuit elements such as resistors,
independent sources, and linear dependent sources. When calculating individual responses, all
sources other than the one of interest should be turned off. A turned-off voltage source is a voltage
source with a constant 0V of voltage across it, which is essentially a short circuit. A turned-off current
source is a current source with a constant 0A of current through it, which is essentially an open circuit.

EX. 2.4
We will continue to use the example circuit in Figure 1-9. Assume we are interested in finding 𝑖1
and 𝑖7 .

Let’s first consider only the 2V voltage source. Note that turning off the two current sources results in
open circuits.

2
𝑖1,2V = = 0.0622mA
25kΩ + [10kΩ ∥ (5kΩ + 20kΩ)]
Since, 𝑖1,2V = 𝑖2,2V and 𝑖5,2V = 𝑖7,2V current division can be used such that
10kΩ
𝑖7,2V = 𝑖1,2V ∙ = 0.0178mA.
10kΩ + (5kΩ + 20kΩ)

Second, consider only the 20mA current source. Note that turning off the 2V voltage source results in a
short circuit.

30
Using current division, the current that flows out of the 20mA current source can be divided into two
parts 𝑖3,20mA and (𝑖5,20mA − 𝑖2,20mA ), where (𝑖5,20mA − 𝑖2,20mA ) = (𝑖7,20mA − 𝑖1,20mA ) can be derived as
10kΩ
−𝑖1,20mA + 𝑖7,20mA = 20mA ∙ = 8.889mA.
10kΩ + [25kΩ ∥ (5kΩ + 20kΩ)]
Since both −𝑖1,20mA and 𝑖7,20mA flow through 25kΩ thus
−𝑖1,20mA = 𝑖7,20mA ,
−𝑖1,20mA = 𝑖7,20mA = 4.445mA.
Finally, consider only the 10mA current source.

Using current division,

(25kΩ ∥ 10kΩ) + 5kΩ
𝑖7,10mA = −10mA ∙ = −3.778mA
(25kΩ ∥ 10kΩ) + 5kΩ + 20kΩ

10kΩ
𝑖1,10mA = 𝑖5,10mA ∙
25kΩ + 10kΩ
20kΩ 10kΩ
= 10mA ∙ ∙
(25kΩ ∥ 10kΩ) + 5kΩ + 20kΩ 25kΩ + 10kΩ
= 1.778mA

Therefore, using superposition

𝑖1 = 𝑖1,2V + 𝑖1,20mA + 𝑖1,10mA = 0.0622 − 4.445 + 1.778 = −2.605mA
and
𝑖7 = 𝑖7,2V + 𝑖7,20mA + 𝑖7,10mA = 0.0178 + 4.445 − 3.778 = 0.685mA.

Using superposition, another method can be used to find the equivalent resistance 𝑅𝑒𝑞 of Thévenin
and Norton equivalent circuits. By turning off all sources in an unknown circuit, the equivalent

31
resistance 𝑅𝑒𝑞 of Thévenin and Norton equivalent circuits would be the same as the equivalent
resistance 𝑅𝑒𝑞 of the unknown circuit.

2.6.2 Δ- TO -Y AND Y- TO -Δ T RANSFORMATIONS

Simplification of circuits can often be achieved by using series and parallel. However, it is not always
possible to simplify a circuit with just series and parallel. It is always possible to revert back to using
KCL and KVL, but it would be desirable if an easier method can be used and one would not have to
solve a huge set of simultaneous equations. The Δ-to-Y and Y-to-Δ transformations discussed here are
such methods.

Figure 2-15 Δ and Y resistive networks.

Using the concept of superposition, nodes a, b, and c have separate voltage sources 𝑣𝑎 , 𝑣𝑏 , and 𝑣𝑐
connected to it. The two circuits are equivalent if all currents 𝑖𝑎 , 𝑖𝑏 , and 𝑖𝑐 flowing out of the voltage
sources are the same for both circuits. Therefore, the following set of equations can be written as

2.21)
𝑣𝑎 𝑣𝑎 𝑣𝑎 1 1 𝑅2 + 𝑅3
+ = 𝑖𝑎 = ⇒ + =
𝑅𝑏 𝑅𝑐 𝑅 𝑅3 𝑅𝑏 𝑅𝑐 𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1
𝑅1 + 𝑅 2+ 𝑅
2 3
𝑣𝑏 𝑣𝑏 𝑣𝑏 1 1 𝑅1 + 𝑅3
+ = 𝑖𝑏 = ⇒ + =
𝑅𝑎 𝑅𝑐 𝑅 𝑅
𝑅2 + 1 3 𝑅𝑎 𝑅𝑐 𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1 ,
𝑅1 + 𝑅3
𝑣𝑐 𝑣𝑐 𝑣𝑐 1 1 𝑅1 + 𝑅2
+ = 𝑖𝑐 = ⇒ + =
𝑅𝑎 𝑅𝑏 𝑅 𝑅2 𝑅𝑎 𝑅𝑏 𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1
{ 𝑅3 + 𝑅 1+ 𝑅
1 2

then
2.22)
1 1 1 𝑅1 + 𝑅2 + 𝑅3
+ + = .
𝑅𝑎 𝑅𝑏 𝑅𝑐 𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1

By subtracting the three results from Eq. (2.16) from Eq. (2.17) the Y-to-Δ transformation can be
derived such that

32
2.23)
𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1
𝑅𝑎 =
𝑅1
𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1
𝑅𝑏 = .
𝑅2
𝑅1 𝑅2 + 𝑅2 𝑅3 + 𝑅3 𝑅1
𝑅𝑐 =
{ 𝑅3

From Eq. (2.18), it can be derived that

2.24)
𝑅𝑎 𝑅2
=
𝑅𝑏 𝑅1
𝑅𝑏 𝑅3
=
𝑅𝑐 𝑅2
𝑅𝑐 𝑅1
=
{𝑅𝑎 𝑅3

and by some simple manipulation back into Eq. (2.16), the Δ-to-Y transformation can be derived as
2.25)
𝑅𝑏 𝑅𝑐
𝑅1 =
𝑅𝑎 + 𝑅𝑏 + 𝑅𝑐
𝑅𝑐 𝑅𝑎
𝑅2 = .
𝑅𝑎 + 𝑅𝑏 + 𝑅𝑐
𝑅𝑎 𝑅𝑏
𝑅3 =
{ 𝑅𝑎 + 𝑅𝑏 + 𝑅𝑐

33
 CHAPTER 3 OPERATIONAL AMPLIFIERS

In this chapter, the operational amplifier (OP amp or OPA) will be discussed and is the basis of all
analog circuits. The most widely known OPA is the μA741 by Fairchild Semiconductor, which was
released in 1961 and still used to this day. The operational amplifier got its name from its ability to
perform various mathematic operations. During the early days of IC development, there was a
competition between analog computers and digital computers. With the improvement of
semiconductor technology, the speed of digital circuits significantly improved and virtually all
computation related circuits are implemented with digital circuits now. However, analog circuits still
has its irreplaceable position in power-related and microwave-related areas. The interface circuits
between analog and digital circuits or data converters are also considered to be analog circuits.

3.1 OPERATIONAL AMPLIFIER BASICS

The simplest OPA as shown in Figure 3-1(a) has three terminals. The non-inverting input (𝑣𝐼,𝑝 ),
inverting input (𝑣𝐼,𝑛 ), and output (𝑣𝑂 ). An OPA with only one output is called a single-ended OPA.
Whereas an OPA with two outputs that are differential is called a fully-differential OPA as shown in
Figure 3-1(b). The OPA is basically a VCVS where the relationship between the output and input can
be written as
3.1)
𝑣𝑂 = 𝐴𝑣 ∙ (𝑣𝐼,𝑝 − 𝑣𝐼,𝑛 ) = 𝐴𝑣 𝑣𝐼 .

(a) Single-ended (b) Fully-differential

Figure 3-1 Ideal OPAs.

For a fully differential OPA, the output voltage is the difference between the non-inverting output
(𝑣𝑂,𝑝 ) and the inverting output (𝑣𝑂,𝑛 ). The relationship between the output and input can be written
as
3.2)
𝑣𝑂 = 𝑣𝑂,𝑝 − 𝑣𝑂,𝑛 = 𝐴𝑣 ∙ (𝑣𝐼,𝑝 − 𝑣𝐼,𝑛 ) = 𝐴𝑣 𝑣𝐼 .

The previous equation defines only the difference between the non-inverting output and the inverting
output. However, the actual voltages are not defined. Therefore, fully-differential OPAs have an
additional output common-mode voltage input (𝑣𝑂,𝑐𝑚 ). The fully-differential OPA has internal
common-mode feedback (CMFB) circuitry that sets the common-mode voltage of 𝑣𝑂,𝑝 and 𝑣𝑂,𝑛 to
𝑣𝑂,𝑐𝑚 such that

34
3.3)
𝑣𝑂,𝑝 + 𝑣𝑂,𝑛
𝑣𝑂,𝑐𝑚 = .
2

Therefore, both
3.4)
𝐴𝑣
𝑣𝑂,𝑝 = 𝑣𝑂,𝑐𝑚 + ∙ (𝑣𝐼,𝑝 − 𝑣𝐼,𝑛 )
2

and
3.5)
𝐴𝑣
𝑣𝑂,𝑛 = 𝑣𝑂,𝑐𝑚 − ∙ (𝑣𝐼,𝑝 − 𝑣𝐼,𝑛 )
2

become well-defined values. It is desirable for an OPA to have its gain as high as possible. For an ideal
OPA, it is assumed that the gain is infinity. Figure 3-2(a) shows the equivalent circuit of an ideal
single-ended OPA. Figure 3-2(b) show the equivalent circuit of an ideal fully-differential OPA.

(a) Single-ended (b) Fully-differential

Figure 3-2 Equivalent circuits of OPAs.

A more realistic OPA has two additional terminals the positive power supply (𝑉𝑃𝑆+ ) and negative
power supply (𝑉𝑃𝑆− ) as shown in Figure 3-3. Since the output voltage is limited by the power supply,
3.6)
𝑣𝑂,𝑚𝑎𝑥 = max(𝑣𝑂 ) ≤ 𝑉𝑃𝑆+

and
3.7)
𝑣𝑂,𝑚𝑖𝑛 = min(𝑣𝑂 ) ≥ 𝑉𝑃𝑆− .

35
 (a) Single-ended (b) Fully-differential
Figure 3-3 OPAs with power supply terminals.

The voltage transfer characteristics (VTC), or input-output relationship, is shown in Figure 3-4. When
the input to the OPA is small such that the output voltage 𝑣𝑂 is between 𝑣𝑂,𝑚𝑎𝑥 and 𝑣𝑂,𝑚𝑖𝑛 , the OPA is
operating in the linear region and follows Eqs. 3.1 and 3.2. When the input is too large, the OPA
saturates and operates in the saturation region and has an output of 𝑣𝑂,𝑚𝑎𝑥 or 𝑣𝑂,𝑚𝑖𝑛 . A special case
where 𝑣𝑂,𝑚𝑎𝑥 = 𝑉𝑃𝑆+ and 𝑣𝑂,𝑚𝑖𝑛 = 𝑉𝑃𝑆− is called a rail-to-rail OPA, since the power supply of an
integrated circuit is sometimes referred as rail. An even more realistic model of OPAs considers the
input resistance (𝑅𝑖𝑛 ) and output resistance (𝑅𝑜𝑢𝑡 ) as shown in Figure 3-5. For an ideal OPA, the input
resistance is infinity and output resistance is zero such that no voltage division will exist at the input
and output of the OPA. Consider the case in Figure 3-6 where the input voltage source has a non-ideal
source resistance 𝑅𝑆 and the output load is a finite resistance 𝑅𝐿 , the overall gain can be expressed as
3.8)
𝑣𝑂 𝑅𝑖𝑛 𝑅𝐿
= 𝐴𝑣 ,
𝑣𝑆 𝑅𝑆 + 𝑅𝑖𝑛 𝑅𝑜𝑢𝑡 + 𝑅𝐿

which has voltage division at both the input and output if the OPA is not ideal.

Figure 3-4 VTC not considering (red) and considering (black) power supply terminals.

36
 (a) Single-ended (b) Fully-differential
Figure 3-5 Equivalent circuit of OPAs considering power supply and input/output resistances.

Figure 3-6 Realistic OPA model with external source and load resistances.

3.2 NEGATIVE FEEDBACK AND VIRTUAL SHORT

Negative feedback is a condition when the output of the OPA is connected back to the inverting input
of the OPA through a circuit network. Since the gain of the OPA is large, the only way for the output
voltage to be finite is when the voltage of the inverting input equals to the voltage of the non-inverting
input. Since the two voltages are always equal like a short circuit, this condition is called virtual short.
Unlike conventional short circuit, there is always no current flowing between the two inputs. In reality,
feedback is actually the most complex part of analog circuit design since careless design may lead to
circuit instability. More detailed analysis of feedback is discussed in Part IV of this course and the
Automatic Control class. In this chapter, all negative feedbacks are assumed to be stable.

Consider the negative feedback circuit in Figure 3-7, the relationship between the input voltage (𝑣𝑆 )
and output voltage (𝑣𝑂 ) is to be determined. We will first attempt to solve this problem with just KCL
and KVL. The input resistance of the OPA is large and assumed to be infinity. Thus the input current
into the non-inverting input (𝑖𝐼,𝑝 ) can be assumed to be zero, such that

37
3.9)
𝑣𝐼,𝑝 = 𝑣𝑆 − 𝑖𝐼,𝑝 𝑅𝑆 = 𝑣𝑆 .

Using Eq. 3.1,

3.10)
𝑣𝑂 = 𝐴𝑣 ∙ (𝑣𝐼,𝑝 − 𝑣𝐼,𝑛 ) = 𝐴𝑣 ∙ (𝑣𝑆 − 𝑣𝑂 ).

Therefore,
3.11)
𝐴𝑣
𝑣𝑂 = 𝑣 .
𝐴𝑣 + 1 𝑆

As 𝐴𝑣 approaches infinity,
3.12)
𝑣𝑂 = 𝑣𝑆 .

For an OPA with voltage gain of 100, the error to the previous calculation is only 1%. For uA741, the
typical gain is 200000, which leads to an error of 0.0005% that is negligible. Note that Eq. 3.12 does
not have the terms of the source resistance 𝑅𝑆 and the load resistance 𝑅𝐿 , which is actually a special
characteristic of the OPA. Now consider the circuit with the virtual short concept previously
introduced. Since the output is connected back to the inverting input, negative feedback condition is
fulfilled. Therefore,
3.13)
𝑣𝑂 = 𝑣𝐼,𝑛 = 𝑣𝐼,𝑝 = 𝑣𝑆 − 𝑖𝐼,𝑝 𝑅𝑆 = 𝑣𝑆 ,

which is a lot easier to analyze than the first method.

The circuit in Figure 3-7 is called the unity-gain buffer. It is widely used in analog circuits since the
output voltage equals to the source voltage, regardless of the source resistance 𝑅𝑆 or load
resistance 𝑅𝐿 . Without the unity-gain buffer, the source resistance and load resistance will form a
voltage divider such that
3.14)
𝑅𝐿
𝑣𝑂 = 𝑣 .
𝑅𝑆 + 𝑅𝐿 𝑆

A signal source such as a sensor with a large source resistance and a measurement instrument with a
small input (load) resistance such as a standard 50Ω or 75Ω system may pose a significant problem
without the use of the unity-gain buffer.

38
 Figure 3-7 Unity-gain buffer circuit.

EX. 3.1
Assume an OPA has an internal input resistance of 100 k, output resistance of 100 , and voltage
gain of 1000 V/V, What is the input and output resistances when it is configured as a unity-gain buffer
with 𝑅𝑆 = 50 Ω and 𝑅𝐿 = 1 kΩ?

First, start off by drawing the equivalent circuit including the internal input and output resistances.

The input resistance can be found by using a test source in front of the OPA. By performing KCL at the
39
output node
𝐴𝑣 𝑣𝐼 − 𝑣𝑂 𝑣𝑂
+ 𝑖𝑡𝑒𝑠𝑡 =
𝑅𝑜𝑢𝑡 𝑅𝐿
1000 × 𝑖𝑡𝑒𝑠𝑡 × 100k − 𝑣𝑂 𝑣𝑂
+ 𝑖𝑡𝑒𝑠𝑡 =
100 1k
100M × 𝑖𝑡𝑒𝑠𝑡
⇒ 𝑣𝑂 ≈ .
1.1
By using KVL,
𝑣𝑡𝑒𝑠𝑡 = 100k × 𝑖𝑡𝑒𝑠𝑡 + 𝑣𝑂 = 91009k × 𝑖𝑡𝑒𝑠𝑡
𝑣𝑡𝑒𝑠𝑡
⇒ 𝑅𝑖 = = 91009 kΩ ≈ 91 MΩ.
𝑖𝑡𝑒𝑠𝑡
It can be seen that the input resistance of the unity-gain buffer is approximately higher than the OPA
by the voltage gain of 1000 V/V.

The output resistance is basically the Thevenin or Norton resistance of the OPA. Since the OPA
includes a VCVS, the resistance must be found by turning off all ideal sources and using a test source.
100𝑘
𝑣𝐼 = − 𝑣 ≈ −𝑣𝑡𝑒𝑠𝑡
100𝑘 + 50 𝑡𝑒𝑠𝑡
𝑣𝑡𝑒𝑠𝑡 − 𝐴𝑣 𝑣𝐼 1001𝑣𝑡𝑒𝑠𝑡
𝑖𝑡𝑒𝑠𝑡 = =
100 100
𝑣𝑡𝑒𝑠𝑡 100
⇒ 𝑅𝑜 = = ≈ 0.1 Ω
𝑖𝑡𝑒𝑠𝑡 1001

EX. 3.2
For the following amplifier, assume with 𝑅𝑆 = 50 Ω and 𝑅𝐿 = 1 kΩ. Please plot the VTC curve and find
𝑣𝑂 if 𝑣𝑆 = 6 sin(2𝜋̇ ∙ 1000𝑡).

When −5 V ≤ 𝑣𝑂 ≤ 5 V, virtual short exists and 𝑣𝑂 = 𝑣𝐼,𝑝 = 𝑣𝑆 .

40
When 𝑣𝐼,𝑝 = 𝑣𝑆 > 5 V, the OPA is positively saturated and 𝑣𝑂 = 5 V.
When 𝑣𝐼,𝑝 = 𝑣𝑆 < −5 V, the OPA is negatively saturated and 𝑣𝑂 = −5 V.

3.3 NON-INVERTING AND INVERTING AMPLIFIERS

The inverting and non-inverting amplifiers are the basis of arithmetic operation using OPAs and are
introduced in this section. In the next section, the two types of amplifiers are used to perform various
kinds of arithmetic operations.

3.3.1 N ON -I NVERTING A MPLIFIERS

OPAs can be configured as non-inverting amplifiers and inverting amplifiers. The unity-gain buffer
discussed in 4.2 has a gain of +1 and is a special case for non-inverting amplifiers made of OPAs. The
general form of the non-inverting amplifier is shown in Figure 3-8. Since the output is connected back
to the inverting input through a resistive divider made of 𝑅1 and 𝑅2 , negative feedback is fulfilled.
Therefore, virtual short happens between 𝑣𝐼,𝑝 and 𝑣𝐼,𝑛 . Since the input resistance of the OPA is large,

41
the current through 𝑅𝑆 is negligible such that
3.15)
𝑅2
𝑣𝑆 = 𝑣𝐼,𝑝 = 𝑣𝐼,𝑛 = 𝑣 .
𝑅1 + 𝑅2 𝑂

Therefore, the voltage gain of the non-inverting amplifier can be written as

3.16)
𝑣𝑂 𝑅1 + 𝑅2 𝑅1
𝐴𝑣 = = =1+ ≥ 1.
𝑣𝑆 𝑅2 𝑅2

For the special case of unity-gain buffer, 𝑅1 is short circuit or zero and 𝑅2 is open circuit or infinity
such that the gain is the minimum achievable gain for non-inverting amplifiers, which is 1.

Figure 3-8 The general form of a non-inverting amplifier.

3.3.2 I NVERTING A MPLIFIERS

The general form of the inverting amplifier is shown in Figure 3-9. Since the output is connected back
to the inverting input through a resistor 𝑅2 , negative feedback is fulfilled. Therefore, virtual short
happens between 𝑣𝐼,𝑝 and 𝑣𝐼,𝑛 , which equal to 0. Using KCL,
3.17)
𝑣𝑆 − 0 0 − 𝑣𝑂
=
𝑅𝑆 + 𝑅2 𝑅1

Therefore, the voltage gain of the inverting amplifier can be written as

3.18)
𝑣𝑂 𝑅1
𝐴𝑣 = =− < 0.
𝑣𝑆 𝑅𝑆 + 𝑅2

Unlike the non-inverting amplifier, the gain of the inverting amplifier is dependent on the non-ideal
source resistance. If 𝑅1 is chosen such that 𝑅𝑆 is comparably negligible, the gain can be simplified as
3.19)
𝑣𝑂 𝑅1
𝐴𝑣 = =− .
𝑣𝑆 𝑅2

42
 Figure 3-9 The general form of an inverting amplifier

3.4 ARITHMETIC OPERATIONS USING OPERATIONAL AMPLIFIERS

OPAs can be configured to perform various arithmetic operations. In this section, addition, subtraction,
multiplication, and division are introduced. Multiplication and division are actually performed
through the exponential of logarithmic sum and difference. Therefore, the exponential and logarithmic
operations are also introduced. Differentiation and integration require the use of capacitors and
inductors and will be introduced in Chapter 5.

3.4.1 A DDITION

The summing amplifier is shown in Figure 3-10. Since the output is connected back to the inverting
input through a resistor 𝑅𝐹 , negative feedback is fulfilled. Therefore, virtual short happens between
𝑣𝐼,𝑝 and 𝑣𝐼,𝑛 , which equal to 0. Using KCL,
3.20)
𝑣𝐴 − 0 𝑣𝐵 − 0 0 − 𝑣𝑂
+ = .
𝑅𝐴 𝑅𝐵 𝑅𝐹

Therefore, the voltage gain of the summing amplifier can be written as

3.21)
𝑅𝐹 𝑅𝐹
𝑣𝑂 = − ( 𝑣𝐴 + 𝑣 ),
𝑅𝐴 𝑅𝐵 𝐵

which is in the form of weighted addition. The weighting factors can be determined by choosing the
correct resistance values. For a special case where 𝑅𝐹 = 𝑅𝐴 = 𝑅𝐵 ,
3.22)
𝑣𝑂 = −(𝑣𝐴 + 𝑣𝐵 ),

which is the sum of two inputs inverted. An inverting amplifier with gain of -1 can be cascaded to
remove the minus sign if needed.

43
 Figure 3-10 The OPA summing amplifier.

3.4.2 S UBTRACTION

The difference amplifier is shown in Figure 3-11. Since the output is connected back to the inverting
input through a resistor 𝑅𝐵 , negative feedback is fulfilled. Therefore, virtual short happens between
𝑣𝐼,𝑝 and 𝑣𝐼,𝑛 . Using KCL on the inverting input,
3.23)
𝑣𝐴 − 𝑣𝐼,𝑛 𝑣𝐼,𝑛 − 𝑣𝑂 𝑅𝐵 𝑣𝐴 + 𝑅𝐴 𝑣𝑂
= ⇒ 𝑣𝐼,𝑛 = .
𝑅𝐴 𝑅𝐵 𝑅𝐴 + 𝑅𝐵

Using voltage division at the non-inverting input,

3.24)
𝑅𝐷 𝑅𝐵 𝑣𝐴 + 𝑅𝐴 𝑣𝑂
𝑣𝐵 = 𝑣𝐼,𝑝 = 𝑣𝐼,𝑛 = .
𝑅𝐶 + 𝑅𝐷 𝑅𝐴 + 𝑅𝐵

Therefore,
3.25)
𝑅𝐴 + 𝑅𝐵 𝑅𝐷 𝑅𝐵
𝑣𝑂 = 𝑣𝐵 − 𝑣 ,
𝑅𝐴 𝑅𝐶 + 𝑅𝐷 𝑅𝐴 𝐴

which is weighted subtraction. For a special case where 𝑅𝐴 = 𝑅𝐵 = 𝑅𝐶 = 𝑅𝐷 ,

3.26)
𝑣𝑂 = 𝑣𝐵 − 𝑣𝐴 ,

which is the difference of the two inputs.

44
 Figure 3-11 The OPA difference amplifier.

EX. 3.3
For the following circuit, please express 𝑣𝑂 in terms of 𝑣𝐴 and 𝑣𝐵 .

Since the OPA is operating under negative feedback, virtual short exists and thus
𝑣𝐼,𝑛 = 𝑣𝐼,𝑝 = 𝑣𝐵 .
Using KCL
𝑣𝐴 − 𝑣𝐼,𝑛 𝑣𝐼,𝑛 − 𝑣𝑂
=
𝑅𝐴 𝑅𝐵
⇒ 𝑅𝐵 𝑣𝐴 − 𝑅𝐵 𝑣𝐵 = 𝑅𝐴 𝑣𝐵 − 𝑅𝐴 𝑣𝑂
⇒ 𝑅𝐴 𝑣𝑂 = (𝑅𝐴 + 𝑅𝐵 )𝑣𝐵 − 𝑅𝐵 𝑣𝐴
𝑅𝐴 + 𝑅𝐵 𝑅𝐵
⇒ 𝑣𝑂 = 𝑣𝐵 − 𝑣
𝑅𝐴 𝑅𝐴 𝐴

The output is basically the superposition of a non-inverting amplifier and an inverting amplifier. Since
the OPA is a linear circuit element, superposition is valid to use.

EX. 3.4
For the previous example, assume 𝑅𝐴 = 1 kΩ, 𝑅𝐵 = 5 kΩ, 𝑅𝐿 = 10 kΩ, 𝑉𝑃𝑆+ = 5 V, 𝑉𝑃𝑆− = −5 V, and
𝑣𝐴 = 0.5 sin(2𝜋 ∙ 1000𝑡) V. What is the range of 𝑣𝐵 such that the output sine wave is not clipped.

45
From the previous example
5+1 5
𝑣𝑂 = 𝑣𝐵 − 𝑣𝐴 = 6𝑣𝐵 − 2.5sin(2𝜋 ∙ 1000𝑡)
1 1

2.5 V 2.5 V
⇒− ≤ 𝑣𝐵 ≤
6 6
5 5
𝑜𝑟 − V ≤ 𝑣𝐵 ≤ V
12 12

3.4.3 M ULTIPLICATION AND D IVISION

The logarithmic amplifier is shown in Figure 3-12. Since the output is connected back to the inverting
input through a diode 𝐷1 , negative feedback is fulfilled. Therefore, virtual short happens between 𝑣𝐼,𝑝
and 𝑣𝐼,𝑛 , which equal 0. Using KCL on the inverting input,
3.27)
𝑣𝑆 − 0 0−𝑣𝑂
= 𝑖𝐷1 ≈ 𝐼𝐷0 𝑒 𝑛𝑉𝑇 .
𝑅𝐴

Therefore,
3.28)
𝑣𝑆
𝑣𝑂 = −𝑛𝑉𝑇 ln ( ),
𝑅𝐴 𝐼𝐷0

which is the logarithmic of the input signal.

Figure 3-12 The OPA logarithmic amplifier.

The exponential amplifier is shown in Figure 3-13. Since the output is connected back to the inverting
input through a resistor 𝑅𝐴 , negative feedback is fulfilled. Therefore, virtual short happens between
𝑣𝐼,𝑝 and 𝑣𝐼,𝑛 , which equal 0. Using KCL on the inverting input,
3.29)
𝑣𝑆 −0 0 − 𝑣𝑂
𝑖𝐷1 ≈ 𝐼𝐷0 𝑒 𝑛𝑉𝑇 = .
𝑅𝐴

Therefore,
46
3.30)
𝑣𝑆
𝑣𝑂 = −𝑅𝐴 𝐼𝐷0 𝑒 𝑛𝑉𝑇 ,

which is the exponential of the input signal.

Figure 3-13 The OPA exponential amplifier.

Multiplication can be performed by adding two logarithmic outputs and then taking the exponential of
it, such that
3.31)
𝑒 (ln 𝑣𝐴 +ln 𝑣𝐵 ) = 𝑒 ln(𝑣𝐴 𝑣𝐵 ) = 𝑣𝐴 𝑣𝐵 .

The actual circuit implementation is shown in Figure 3-14. The two logarithmic outputs are first
calculated such that
3.32)
𝑣𝐴
𝑣𝑂𝐴 = −𝑛𝑉𝑇 ln ( )
𝑅𝐴 𝐼𝐷0

and
3.33)
𝑣𝐵
𝑣𝑂𝐵 = −𝑛𝑉𝑇 ln ( )
𝑅𝐵 𝐼𝐷0

The output of the summing amplifier is

3.34)
𝑣𝑂𝐶 = −(𝑣𝑂𝐴 + 𝑣𝑂𝐵 )

𝑣𝐴 𝑣𝐵
= 𝑛𝑉𝑇 [ln ( ) + ln ( )]
𝑅𝐴 𝐼𝐷0 𝑅𝐵 𝐼𝐷0

𝑣𝐴 𝑣𝐵
= 𝑛𝑉𝑇 ln ( 2 )
𝑅𝐴 𝑅𝐵 𝐼𝐷0

Then the output can be derived as

47
3.35)
𝑣𝑂𝐶
𝑣𝑂𝑈𝑇 = −𝑅𝐷 𝐼𝐷0 𝑒 𝑛𝑉𝑇

𝑣 𝑣
𝑛𝑉𝑇 ln( 𝐴 𝐵2 )
𝑅𝐴 𝑅𝐵 𝐼𝐷0
= −𝑅𝐷 𝐼𝐷0 𝑒 𝑛𝑉𝑇

𝑅𝐷
=− 𝑣 𝑣
𝑅𝐴 𝑅𝐵 𝐼𝐷0 𝐴 𝐵

Figure 3-14 The OPA multiplier.

Division can be easily performed by subtracting two logarithmic outputs and then taking the
exponential of it such that
3.36)
𝑣 𝑣𝐴
ln( 𝐴 )
𝑒 (ln 𝑣𝐴 −ln 𝑣𝐵 ) = 𝑒 𝑣𝐵 = .
𝑣𝐵

Implementation of this can be done by replacing the summing amplifier in Figure 3-14 with a
difference amplifier shown in Figure 3-11.

3.5 COMPARATORS AND POSITIVE FEEDBACK

From Figure 3-4, an OPA with maximum and minimum outputs can be considered as having a binary
output when operating under open loop (without feedback). Mathematically, unless the two inputs of
the OPA are equal, the output will be at one of the saturated state. In reality, with the existence of
noise, the chance of having an output in the linear region is virtually zero.

48
3.5.1 C OMPARATORS

The OPA can operate as a non-inverting comparator as shown in Figure 3-15(a) if the non-inverting
input is connected to the input signal and the inverting input is connected to the reference voltage
(𝑉𝑅𝐸𝐹 ) to be compared with. When the input is greater than the reference voltage, the output will
be 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑎𝑥 . When the input is less than the reference voltage, the output will be 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑖𝑛 .
For simplicity, the comparator is often drawn as in Figure 3-15(b), which looks exactly like an ideal
OPA. The difference is that the comparator does not have any negative feedback connections. Also, a
comparator without power supply connections is assumed to have outputs of the two supply rails. To
limit the output voltage, a voltage clipping circuit to be discussed next semester can be added.

(a) Comparator with power supply connections (b) Comparator without power supply connections
Figure 3-15 Open-loop OPA used as a comparator.

3.5.2 H YSTERIC C OMPARATORS U SING P OSITIVE F EEDBACK

With the existence of noise or interference, the comparator might toggle unexpectedly. As a result, the
hysteric comparators are used to mitigate this problem. The hysteresis is designed large enough to not
respond to noise and interference. An example showing the difference between a conventional
comparator and a hysteric comparator is shown in Figure 3-16. The two comparator outputs are
slightly shifted to better show the difference. It is clear that the hysteretic comparator is more
immune to noise. However, at the expense of a slower response since the signal must be greater than
the hysteresis for it to respond.

Figure 3-16 Comparator output of a noisy signal (－input signal, －conventional, －hysteric, －hysteresis).

49
Figure 3-17(a) shows the schematic of a non-inverting hysteric comparator circuit. The output is fed
back to the non-inverting input of the OPA through 𝑅2. Therefore, the OPA is configured as positive
feedback instead of negative feedback. Moreover, the virtual short assumption is not valid. The non-
inverting input can be derived as
3.37)
𝑅2 𝑅1
𝑣𝐼,𝑝 = 𝑣𝐼𝑁 + 𝑣𝑂𝑈𝑇 .
𝑅1 + 𝑅2 𝑅1 + 𝑅2

If the both the input and output are initially at their low state 𝑣𝐼𝑁 = 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑖𝑛 , 𝑣𝐼,𝑝 < 0 = 𝑣𝐼,𝑛 and
the output is at a stable state. As the input increases, the output will change its state to 𝑣𝑂𝑈𝑇 =
𝑣𝑂,𝑚𝑎𝑥 when 𝑣𝐼,𝑝 > 𝑣𝐼,𝑛 = 0, which is when
3.38)
𝑅1
𝑣𝐼𝑁 = −𝑣𝑂,𝑚𝑖𝑛 .
𝑅2

The input voltage when the output changes its state is called the tripping voltage. As the input voltage
further increases, the output will retain its high state. As the input decreases from 𝑣𝐼𝑁 = 𝑣𝑂𝑈𝑇 =
𝑣𝑂,𝑚𝑎𝑥 , The output will change its state to 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑖𝑛 when 𝑣𝐼,𝑝 < 𝑣𝐼,𝑛 = 0, which is when
3.39)
𝑅1
𝑣𝐼𝑁 = −𝑣𝑂,𝑚𝑎𝑥 .
𝑅2

The VTC of the non-inverting hysteric comparator is shown in Figure 3-17(b).

(a) Schematic (b) VTC

Figure 3-17 A non-inverting hysteretic comparator.

Figure 3-18(a) shows the schematic of an inverting hysteric comparator circuit. The output is fed back
to the non-inverting input of the OPA through 𝑅2 . Therefore, the OPA is configured as positive
feedback. The non-inverting input can be derived as
3.40)
𝑅1
𝑣𝐼,𝑝 = 𝑣𝑂𝑈𝑇 .
𝑅1 + 𝑅2

50
If the input is initially at 𝑣𝐼𝑁 = 𝑣𝐼,𝑛 = 𝑣𝑂,𝑚𝑖𝑛 and output at 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑎𝑥 , 𝑣𝐼,𝑝 > 0 > 𝑣𝐼,𝑛 thus this is a
stable state. As the input increases, the output will change its state to 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑖𝑛 when
3.41)
𝑅1
𝑣𝐼𝑁 = 𝑣𝐼,𝑛 = 𝑣𝐼,𝑝 = 𝑣𝑂,𝑚𝑎𝑥 .
𝑅1 + 𝑅2

Further increase in the input voltage will not change the output. As the input decreases from 𝑣𝐼𝑁 =
𝑣𝐼,𝑛 = 𝑣𝑂,𝑚𝑎𝑥 , the output will change its state from 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑖𝑛 to 𝑣𝑂𝑈𝑇 = 𝑣𝑂,𝑚𝑎𝑥 when the input
voltage drops to
3.42)
𝑅1
𝑣𝐼𝑁 = 𝑣𝐼,𝑛 = 𝑣𝐼,𝑝 = 𝑣𝑂,𝑚𝑖𝑛 .
𝑅1 + 𝑅2

Further decrease in the input voltage will not change the output.

(a) Schematic (b) VTC

Figure 3-18 An inverting hysteretic comparator.

The non-inverting hysteric comparator is often called the Schmitt trigger and the inverting hysteric
comparator is sometimes called the inverting Schmitt trigger. Their circuit symbols are shown below
in Figure 3-19.

(a) Schmitt trigger (b) Inverting Schmitt trigger

Figure 3-19 Circuit symbols of Schmitt trigger and inverting Schmitt trigger.

3.6 DATA CONVERSION

Data conversion is the process of converting analog to digital and digital to analog. With the
technological advancement of digital ICs, computation is almost always performed digitally. The use of

51
digital technologies, specifically microcontrollers or microprocessors, provides benefits both in speed
and cost. However, circuits interacting with human senses, environmental sensors, power related
circuits, and high-frequency communication circuits are not able to be implemented digitally. The
most common way to implement circuits that process analog signals is shown in Figure 3-20. An
analog-to-digital converter (ADC) is required to convert the analog input signal to digital for the
microprocessor to process. After processing, the digital output signal is sent to a digital-to-analog
converter (DAC) to convert the processed signal back to analog.

Figure 3-20 Typical signal processing utilizing a digital signal processor or microcontroller.

3.6.1 A NALOG - TO -D IGITAL C ONVERTERS

The analog-to-digital converter (ADC) performs quantization and sampling of analog signals. The
quantization process is basically rounding and truncating data, and therefore creates error. Such error
is called quantization error and has noise-like behavior. Such noise is called quantization noise and is
considered white noise since it has no correlation with the original signal. Sampling is the process of
converting a continuous-time signal into a discrete-time signal. Figure 3-21 shows an example of
sampling and quantization. It can be seen that higher resolution or more bits results in less
quantization error. Both quantization and sampling will be discussed in detail in the course “Signals
and Systems”.

Figure 3-21 Example of sampling and quantization.

Quantization in ADC is made possible by comparators, making it possibly the most important circuit
block inside an ADC. The comparator itself is actually a 1-bit quantizer since it has two states. A 1-bit
52
ADC circuit built with a comparator is shown in Figure 3-22(a). For a input voltage 𝑣𝐼𝑁 between 0
and 𝑉𝐹𝑆 , 𝑣𝑂𝑈𝑇 will give a logic low 𝑉𝐿 when 𝑣𝐼𝑁 < 𝑉𝐹𝑆 /2 and will give a logic high 𝑉𝐻 when 𝑣𝐼𝑁 > 𝑉𝐹𝑆 /2.
The input-output relation is shown in Figure 3-22(b).

(a) Circuit (b) Input-output relation

Figure 3-22 A 1-bit ADC.

To increase the resolution of the ADC, more comparators can be used to distinguish more levels.
Figure 3-23(a) shows a 3-bit ADC using this architecture. The analog input and digital output is
summarized in Table 3.1. The outputs 𝑣𝑂𝑈𝑇𝐴 to 𝑣𝑂𝑈𝑇𝐺 are in thermometer code and can be converted
to binary code using the digital encoder circuit shown in Figure 3-23(a). This type of ADC is called the
flash ADC for its speed. The input-output relation is shown in the top of Figure 3-23(b). By comparing
the output with an ideal curve, the quantization error can be calculated and shown on the bottom of
Figure 3-23(b). For this circuit the error is between 0 and +1LSB.

(a) Circuit (b) Input-output relation and error

Figure 3-23 A 3-bit flash ADC.

53
 Table 3.1 Digital Outputs of a 3-Bit Flash ADC

𝒗𝑶𝑼𝑻𝑨 𝒗𝑶𝑼𝑻𝑩 𝒗𝑶𝑼𝑻𝑪 𝒗𝑶𝑼𝑻𝑫 𝒗𝑶𝑼𝑻𝑬 𝒗𝑶𝑼𝑻𝑭 𝒗𝑶𝑼𝑻𝑮 Binary Output

𝑣𝐼𝑁 > 0.875𝑉𝐹𝑆 1 1 1 1 1 1 1 B2B1B0 =111
0.875𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.750𝑉𝐹𝑆 0 1 1 1 1 1 1 B2B1B0 =110
0.750𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.675𝑉𝐹𝑆 0 0 1 1 1 1 1 B2B1B0 =101
0675𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.500𝑉𝐹𝑆 0 0 0 1 1 1 1 B2B1B0 =100
0.500𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.375𝑉𝐹𝑆 0 0 0 0 1 1 1 B2B1B0 =011
0.375𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.250𝑉𝐹𝑆 0 0 0 0 0 1 1 B2B1B0 =010
0.250𝑉𝐹𝑆 > 𝑣𝐼𝑁 > 0.125𝑉𝐹𝑆 0 0 0 0 0 0 1 B2B1B0 =001
0.125𝑉𝐹𝑆 > 𝑣𝐼𝑁 0 0 0 0 0 0 0 B2B1B0 =000

(a) Circuit (b) Input-output relation and error

Figure 3-24 A 3-bit flash ADC with shifted resistor ladder.

To limit the quantization error, the resistor ladder in Figure 3-23(a) can be slightly modified as in
Figure 3-24(a) such that the quantization error is limited to ±0.5LSB except the maximum level. The
input-output relation and quantization error are shown in Figure 3-24(b).

The flash ADC is only one of the many popular ADC architectures. Other architectures include the
pipelined ADC, successive approximation register (SAR) ADC, and sigma-delta (Σ-Δ) ADC. These ADC
architectures are summarized and compared in Table 3.2. The other architectures are out of the scope
of this course. Those who are interested can take courses in “Mixed-Signal ICs”.
54
 Table 3.2 Comparison of Popular ADC Architectures

Architectures Resolution Speed Cost

Flash ~8 bits Very High High
Pipelined 10~16 bits High High
SAR 10~24 bits Medium Low
Σ-Δ 24~32 bits Low Low

3.6.2 D IGITAL - TO -A NALOG C ONVERTERS

The digital-to-analog converter (DAC) performs the opposite of the ADC. It converts sampled
quantized data back to an analog continuous signal. An example is shown in Figure 3-25. The DAC
output is somewhat distorted and this phenomenon is called aliasing. A low-pass filter to be discussed
in Chapter 8 is needed to solve this problem. This particular low-pass filter is called an anti-aliasing
filter. More detail analysis of this will be introduced in the “Signals and Systems” class.

Figure 3-25 Example of a DAC output.

The most straightforward way to implement a DAC is the binary-weighted DAC architecture using
weighted resistors, which is basically a summing amplifier. A 4-bit binary-weighted DAC is shown in
Figure 3-26. The output voltage can be simply derived as
3.43)

𝐵0 𝐵1 𝐵2 𝐵3
𝑣𝑂 = −𝑅𝐹 ( + + + )
𝑅 𝑅 𝑅 𝑅
20 21 22 23
𝑅𝐹 0
=− (2 𝐵0 + 21 𝐵1 + 22 𝐵2 + 23 𝐵3 ).
𝑅

55
𝐵0 to 𝐵3 are the binary input codes. If digital inputs are implemented in CMOS, a logic high voltage
would be 𝑉𝐷𝐷 and a logic low voltage is 0 or ground. If reverse polarity is not desired, an inverting
amplifier can be cascaded at the output of the DAC. In reality, binary-weighted DACs are rarely used
because the ratio between the largest and smallest resistors is 2𝑁 , where 𝑁 is the DAC resolution. The
R-2R ladder DAC architecture only has two values of resistors 𝑅 and 2𝑅, easing the implementation
and has better resistor matching. A 4-bit R-2R ladder DAC is shown in Figure 3-27. The output voltage
can be derived using Thévenin equivalent circuits such that
3.44)
𝐵0
2 + 𝐵1 + 𝐵
2 2
+ 𝐵3
𝑣𝑂 = 2
2
= (𝐵0 /24 + 𝐵1 /23 + 𝐵2 /22 + 𝐵3 /21 ).

Figure 3-26 A 4-bit binary-weighted DAC using weighted resistors.

Figure 3-27 A 4-bit R-2R ladder DAC.

Binary-weighted DACs and R-2R DACs are only two of the many DAC architectures. Other popular

56
architectures include the current-steering DACs and the unit-element DACs. Each DAC architecture
has its pros and cons. Therefore, DACs are often implemented with multiple architectures to have
optimized performance.

57
 PART II TRANSIENT ANALYSIS
In this part, a new variable, time (𝑡), will be introduced in circuit analysis. For purely resistive circuits,
adding a time variable does not change the analysis method. The major difference is when the circuit
includes capacitors and/or inductors. With capacitors and/or inductors, solving the circuit will
involve solving a linear ordinary differential equation (ODE).

CHAPTER 4 CAPACITORS AND INDUCTORS

Resistors discussed in the previous chapter are elements that dissipate energy, while sources are
elements that provide energy unless they are being charged. Capacitors and inductors discussed in
this chapter are elements that store energy and exhibit memory effect, which make them possible to
be used as memories in computers.

Circuits with capacitors and inductors not only have responses that change with time, but also require
differential equations to solve. This chapter introduces the capacitor and inductor elements. Circuits
involving capacitors and inductors will be discussed in the next chapter.

4.1 CAPACITORS

The ideal capacitor is a two-terminal device that is governed by the equation

4.1)
𝑞 = 𝐶𝑣,

where 𝐶 is capacitance in farads (F). The previous equation can also be written as a differential
equation
4.2)
𝑑𝑞 𝑑𝑣
=𝐶 .
𝑑𝑡 𝑑𝑡

From the definition of current in Eq. 1.1, it can be concluded that

4.3)
𝑑𝑞 𝑑𝑣
𝑖= =𝐶 .
𝑑𝑡 𝑑𝑡

The circuit symbol of a capacitor is shown in Figure 4-1. If the voltage across the capacitor is constant
or DC, the current through the capacitor would be zero. The capacitor is therefore open circuit to DC
voltages.

58
 Figure 4-1 Circuit symbol of a capacitor.

The differential equation in Eq. 4.3 can also be written in the integral form
4.4)

∫ 𝑖 𝑑𝑡 = ∫ 𝐶 𝑑𝑣.

At 𝑡 = 𝜏
4.5)
𝜏
∫ 𝑖 𝑑𝑡 = 𝐶 ∫ 𝑑𝑣 = 𝐶𝑣(𝜏)
−∞

then
4.6)
1 𝜏 1 𝜏
𝑣(𝜏) = ∫ 𝑖 𝑑𝑡 = 𝑣(𝜏0 ) + ∫ 𝑖 𝑑𝑡,
𝐶 −∞ 𝐶 𝜏0

Where 𝑣(𝜏0 ) term shows the circuit’s present time voltage depends on the prior time voltage, which is
essentially memory effect. If the current 𝑖 is zero, the voltage or charge will be retained. This is the
basis of the dynamic random access memory (DRAM). Figure 4-2 shows a basic DRAM
implementation. The control line selects which capacitor to read or write to and the data is sent in or
read out from the data lines. Since actual capacitors have leakage currents, the cells have to be
refreshed every once in a while to retain the charges.

Figure 4-2 Implementation of DRAM.

The instantaneous power of a capacitor can be calculated by

59
4.7)
𝑑𝑣
𝑝 = 𝑣𝑖 = 𝐶𝑣 .
𝑑𝑡

Therefore, the energy stored by the capacitor can be calculated by integrating the previous equation
4.8)
1
𝐸 = ∫ 𝑝 𝑑𝑡 = ∫ 𝐶𝑣 𝑑𝑣 = 𝐶𝑣 2 ,
2

if the initial voltage is zero.

4.1.1 P ARALLEL P LATE C APACITORS

The simplest form of capacitor is the parallel plate capacitor as shown in Figure 4-3. The capacitance
can be calculated by
4.9)
𝐴 𝐴
𝐶=𝜖 = 𝜖0 𝜖𝑟 ,
𝑑 𝑑

where 𝜖0 =8.854(pF/m) is the free space permittivity, 𝜖𝑟 is the relative permittivity of the dielectric, 𝐴
is the area of the parallel plates in square meters (m2 ), and 𝑑 is the distance between the plates in
meters (m).

Figure 4-3 A parallel place capacitor.

4.1.2 C APACITORS IN P ARALLEL

When capacitors are in parallel as shown in Figure 4-4, the voltages through all capacitors are
identical because of KVL such that
4.10)
𝑣 = 𝑣1 = 𝑣2 = ⋯ = 𝑣𝑘 .

By using KCL, the total current through all the capacitors can be written as
4.11)
𝑖 = 𝑖1 + 𝑖2 + ⋯ + 𝑖𝑘
𝑑𝑣1 𝑑𝑣2 𝑑𝑣𝑘
= 𝐶1 + 𝐶2 + ⋯ + 𝐶𝑘
𝑑𝑡 𝑑𝑡 𝑑𝑡
60
 𝑑𝑣 𝑑𝑣 𝑑𝑣
= 𝐶1 + 𝐶2 + ⋯ + 𝐶𝑘
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝑣
= (𝐶1 + 𝐶2 + ⋯ + 𝐶𝑘 )
𝑑𝑡

Therefore, the equivalent capacitance of capacitors in parallel can then be defined as

4.12)
𝐶𝑒𝑞 = 𝐶1 + 𝐶2 + ⋯ + 𝐶𝑘
𝑘

= ∑ 𝐶𝑛 .
n=1

Figure 4-4 Capacitors in parallel.

4.1.3 C APACITORS IN S ERIES

When capacitors are in series as shown in Figure 4-5, the current through all capacitors are equal
because of KCL such that
4.13)
𝑖 = 𝑖1 = 𝑖2 = ⋯ = 𝑖𝑘 .

Because of KVL, the total voltage across all capacitors can be written as
4.14)
𝑣 = 𝑣1 + 𝑣2 + ⋯ + 𝑣𝑘
1 𝜏 1 𝜏 1 𝜏
= ∫ 𝑖1 𝑑𝑡 + ∫ 𝑖2 𝑑𝑡 + ⋯ + ∫ 𝑖𝑘 𝑑𝑡
𝐶1 −∞ 𝐶2 −∞ 𝐶𝑘 −∞
1 𝜏 1 𝜏 1 𝜏
= ∫ 𝑖 𝑑𝑡 + ∫ 𝑖 𝑑𝑡 + ⋯ + ∫ 𝑖 𝑑𝑡
𝐶1 −∞ 𝐶2 −∞ 𝐶𝑘 −∞
𝜏
1 1 1
=( + + ⋯ + ) ∫ 𝑖 𝑑𝑡
𝐶1 𝐶2 𝐶𝑘 −∞

Therefore, the equivalent capacitance of capacitors in series can then be defined as

4.15)
1
𝐶𝑒𝑞 =
1 1 1
𝐶1 + 𝐶2 + ⋯ + 𝐶𝑘
1
= .
1
∑𝑘n=1
𝐶𝑛
61
 Figure 4-5 Capacitors in series.

4.2 INDUCTORS

The ideal inductor is a two-terminal device that is governed by the equation

4.16)
Φ = 𝐿𝑖,

where Φ is magnetic flux in weber (Wb) and 𝐿 is inductance in henrys (H). The previous equation can
also be written as a differential equation
4.17)
𝑑Φ 𝑑𝑖
=𝐿 .
𝑑𝑡 𝑑𝑡

From Faraday’s law, we know that

4.18)
𝑑Φ
𝑣= .
𝑑𝑡

It can then be concluded that

4.19)
𝑑Φ 𝑑𝑖
𝑣= =𝐿 .
𝑑𝑡 𝑑𝑡

The circuit symbol of the inductor is shown in Figure 4-6. If the current through the inductor is
constant or DC, the voltage across the inductor would be zero. The inductor is therefore short circuit
to DC voltages.

62
 Figure 4-6 Circuit symbol of an inductor.

The differential equation in Eq. 4.19 can also be written in the integral form
4.20)

∫ 𝑣 𝑑𝑡 = ∫ 𝐿 𝑑𝑖.

At 𝑡 = 𝜏
4.21)
𝜏
∫ 𝑣 𝑑𝑡 = 𝐿 ∫ 𝑑𝑖 = 𝐿𝑖(𝜏)
−∞

then
4.22)
1 𝜏 1 𝜏
𝑖(𝜏) = ∫ 𝑣 𝑑𝑡 = 𝑖(𝜏0 ) + ∫ 𝑣 𝑑𝑡,
𝐿 −∞ 𝐿 𝜏0

where 𝑖(𝜏0 ) term shows the circuit’s present-time current depends on the prior-time current, which is
essentially memory effect. Memory based on the inductor was actually used twenty years earlier than
the capacitor-based DRAM. This memory was called the magnetic-core memory or core memory,
which was made obsolete after the cheaper, easier to manufacture and higher-density DRAM was
introduced by Intel in the 1970s.

The instantaneous power of an inductor can be calculated by

4.23)
𝑑𝑖
𝑝 = 𝑣𝑖 = 𝐿𝑖 .
𝑑𝑡

Therefore, the energy stored by the inductor can be calculated by integrating the previous equation
4.24)
1
𝐸 = ∫ 𝑝 𝑑𝑡 = ∫ 𝐿𝑖 𝑑𝑖 = 𝐿𝑖 2 ,
2

if the initial current is zero.

4.2.1 S OLENOID I NDUCTORS

The simplest form of the inductor is the solenoid inductor as shown in Figure 4-7. The inductance can
be calculated by

63
4.25)
𝑁2𝐴 𝑁2𝐴
𝐿=𝜇 = 𝜇0 𝜇𝑟 ,
𝑙 𝑙

where 𝜇0 = 4π × 10−7 (H⁄m) = 1.26 (μH⁄m) is the free space permeability, 𝜇𝑟 is the relative
permeability of the core, 𝐴 is the area of the core in square meters (m2 ), and 𝑙 is the length of the
windings in meters (m).

Figure 4-7 A solenoid inductor.

4.2.2 I NDUCTORS IN S ERIES

When inductors are in series as shown in Figure 4-8, the currents through all inductors are identical
because of KCL such that
4.26)
𝑖 = 𝑖1 = 𝑖2 = ⋯ = 𝑖𝑘 .

By using KVL, the total voltage across all the inductors can be written as
4.27)
𝑣 = 𝑣1 + 𝑣2 + ⋯ + 𝑣𝑘
𝑑𝑖1 𝑑𝑖2 𝑑𝑖𝑘
= 𝐿1 + 𝐿2 + ⋯ + 𝐿𝑘
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝑖 𝑑𝑖 𝑑𝑖
= 𝐿1 + 𝐿2 + ⋯ + 𝐿𝑘
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝑖
= (𝐿1 + 𝐿2 + ⋯ + 𝐿𝑘 )
𝑑𝑡

Therefore, the equivalent inductance of inductors in series can then be defined as

4.28)
𝐿𝑒𝑞 = 𝐿1 + 𝐿2 + ⋯ + 𝐿𝑘
𝑘

= ∑ 𝐿𝑛 .
n=1

64
 Figure 4-8 Inductors in series.

4.2.3 I NDUCTORS IN P ARALLEL

When inductors are in parallel as shown in Figure 4-9, the voltage across all inductors are equal
because of KVL such that
4.29)
𝑣 = 𝑣1 = 𝑣2 = ⋯ = 𝑣𝑘 .

Because of KCL, the total current all inductors can be written as

4.30)
𝑖 = 𝑖1 + 𝑖2 + ⋯ + 𝑖𝑘
1 𝜏 1 𝜏 1 𝜏
= ∫ 𝑣1 𝑑𝑡 + ∫ 𝑣2 𝑑𝑡 + ⋯ + ∫ 𝑣𝑘 𝑑𝑡
𝐿1 −∞ 𝐿2 −∞ 𝐿𝑘 −∞
1 𝜏 1 𝜏 1 𝜏
= ∫ 𝑣 𝑑𝑡 + ∫ 𝑣 𝑑𝑡 + ⋯ + ∫ 𝑣 𝑑𝑡
𝐿1 −∞ 𝐿2 −∞ 𝐿𝑘 −∞
𝜏
1 1 1
= ( + + ⋯ + ) ∫ 𝑣 𝑑𝑡
𝐿1 𝐿2 𝐿𝑘 −∞

Therefore, the equivalent inductance of inductors in parallel can then be defined as

4.31)
1
𝐿𝑒𝑞 =
1 1 1
𝐿1 + 𝐿2 + ⋯ + 𝐿𝑘
1
= .
1
∑𝑘n=1
𝐿𝑛

65
 Figure 4-9 Inductors in parallel.

4.3 REAL-WORLD CAPACITORS AND INDUCTORS

Real-world capacitors and inductors have non-ideal characteristics. Different types of capacitors and
inductors are introduced and discussed here.

4.3.1 R EAL -W ORLD C APACITORS

Real world capacitors have non-idealities and their non-ideal characteristics depend on the family of
capacitors they belong to. The model widely used for real-world capacitors is shown in Figure 4-10.
Other than the capacitance of the capacitor of interest, the model has three additional elements. The
parallel resistance 𝑅𝑝 models the resistance of the dielectric since it is not a perfect insulator. The
series resistance 𝑅𝑠 models the conductive loss caused by the wires or leads of the capacitor. The
series inductance 𝐿𝑠 models the self-inductance of the wires and leads of the capacitor.

Figure 4-10 Circuit model of real-world capacitors.

The two most popular capacitor families are the ceramic capacitors and the electrolyte capacitors. The
ceramic capacitor uses ceramic as the dielectric. Its characteristics are the closest to an ideal capacitor
among all capacitor families. Figure 4-11(a) shows ceramic capacitors with radial leads and Figure
4-11(b) shows a SMD ceramic capacitor. The dielectric leakage current of a ceramic capacitor is small
meaning the dielectric is close to a perfect insulator and thus 𝑅𝑝 → ∞. The connections to external
circuit elements are short and low loss hence 𝑅𝑠 → 0. For ceramic capacitors with leads, the series
resistance is larger. The series inductance of SMD a ceramic capacitor is small, since the electrical
connection is short. For radial-lead ceramic capacitors, the parasitic inductance is large because of the
long leads. Leads are normally cut short after soldering to reduce the series inductance and resistance.
The major drawback of ceramic capacitors is the capacitance values are small compared to electrolytic
capacitors.

66
 (a) Radial-lead ceramic capacitors (b) SMD ceramic capacitors
Figure 4-11 Ceramic capacitors.

The two most popular electrolytic capacitors are the aluminum electrolyte and the tantalum
electrolytic capacitors. The major drawback of electrolytic capacitors is that electrolytic capacitors are
polarized. Using the electrolytic capacitors backwards may cause it to explode. The circuit symbol for
a polarized capacitor is shown in Figure 4-12(a). A polarization mark is sometimes added to the
symbol to show the anode of the capacitor as shown in Figure 4-12(b).

Figure 4-12 Circuit symbol of polarized capacitor.

Figure 4-13(a) shows an aluminum electrolytic capacitor. This type of capacitors is known for its large
value, higher operating voltage, and low price. It is widely used for high power applications. This type
of capacitor has vent marks on top of the can, which will burst open to release gas and liquid when
improperly used. Without those vent marks, the capacitor will simply explode in an uncontrolled
fashion. Aluminum electrolytic capacitors are known to have larger leakage and thus lower 𝑅𝑝 . Both
the series resistance and inductance are larger because of the leads and the internal connections.
Moreover, the electrolytes inside the capacitor will degrade from aging, and may blow up after years
of use.

(a) Radial-lead aluminum elctrolytic capacitor (b) SMD tantalum capacitor

Figure 4-13 Electrolytic capacitors.

Figure 4-13(b) shows an SMD tantalum electrolytic capacitor. Even though the tantalum capacitor is
67
an electrolytic capacitor, the electrolyte used is solid instead of the liquid form in aluminum
electrolytic capacitors. The solid electrolyte degrades significantly slower than liquid electrolytes. The
performance of tantalum capacitors is between the two previously mentioned capacitors and
tantalum capacitors are usually used in medium power applications.

4.3.2 R EAL -W ORLD I NDUCTORS

Like capacitors, real-world inductors have non-idealities. The model widely used for real-world
inductors is shown in Figure 4-14. Other than the inductance of the inductor of interest, the model has
three additional elements. The series resistance 𝑅𝑠 models the conductive loss in the wires of the
windings. The parallel resistance 𝑅𝑝 models the resistance of the insulating material between the
wires since it is not a perfect insulator. The parallel capacitance 𝐶𝑝 models the capacitance between
the wires where the insulating material is the dielectric material.

Figure 4-14 Circuit model of real-world inductors.

Real-world inductors are categorized into air-core inductors and ferromagnetic-core inductors. Air-
core inductors may be inductors that do not have any cores at all or uses cores that are not
ferromagnetic materials such as glass or plastic. Air-core inductors do not saturate when used under
high currents, but are limited in inductance values. They are also more suitable for high-frequency
applications because ferromagnetic cores have significant high-frequency core losses. Ferromagnetic
cores have relative permeability (𝜇𝑟 ) typically in the thousands and thus the inductance can be
several thousand times larger than air-core inductors with similar sizes. Losses in inductors are very
complex because it not only have electrical losses, but also have magnetic losses. The model shown in
Figure 4-14 is often not adequate. Since both capacitors and inductors are energy storage devices, in
applications where both are interchangeable, capacitors are normally chosen over inductors.

(a) Solenoid air-core inductor (b) Toroid ferrite-core inductor

Figure 4-15 Real-world inductors.

68
 CHAPTER 5 CAPACITIVE AND INDUCTIVE CIRCUITS

Solving circuits involving capacitors and inductors require solving differential equations. The order
number of the circuit is defined as the order number of the ordinary differential equation (ODE) to be
solved. Generally, the total number of inductors and capacitors is the order number of the circuit and
ODE, but is not always true.

5.1 FIRST-ORDER CIRCUITS

First-order circuits are circuits that can be solved by a first-order differential equation. The circuit
may have either one capacitor or one inductor.

5.1.1 F IRST -O RDER RC C IRCUITS

In an RC circuit, the capacitor current is

5.1)
𝑑𝑣𝐶
𝑖𝐶 = 𝐶 .
𝑑𝑡

The capacitor voltage 𝑣𝐶 is always continuous; otherwise the capacitor current 𝑖𝐶 would be infinity. If
𝑑𝑣𝐶
no disturbance occurs in the circuit, 𝑖𝐶 = 0 = 𝐶 𝑑𝑡
and the capacitor voltage 𝑣𝐶 would remain
constant. Assume a circuit is disturbed at 𝑡 = 0, then at 𝑡 = 0− , 𝑖𝐶 = 0 and 𝑣𝐶 is constant. Similarly, as
𝑡 approaches infinity, 𝑖𝐶 = 0 and 𝑣𝐶 is constant. Therefore, the capacitor C can be considered as open
circuit at those two time points.

Figure 5-1 First-order RC circuit.

For the RC circuit shown in Figure 5-1, the switch turns off at 𝑡 = 0. Because of KCL,
5.2)
𝑑𝑣𝐶 𝑣𝐶
𝑖𝐶 + 𝑖𝑅 = 0 ⇒ 𝐶 + = 0.
𝑑𝑡 𝑅

By separating the variables

5.3)
𝑑𝑣𝐶 𝑑𝑡
=− .
𝑣𝐶 𝑅𝐶
69
By integrating both sides from 𝑡 = 0 to 𝑡 = 𝑡0 ,
5.4)
𝑣𝐶 (𝑡0 ) 𝑡0
𝑑𝑣𝐶 𝑑𝑡
∫ =∫ − .
𝑣𝐶 (0) 𝑣𝐶 0 𝑅𝐶
𝑣𝐶 (𝑡0 ) 𝑡0
⇒ ln[𝑣𝐶 (𝑡0 )] − ln[𝑣𝐶 (0)] = ln =−
𝑣𝐶 (0) 𝑅𝐶
𝑣𝐶 (𝑡0 ) 𝑡0 𝑡0
⇒ = 𝑒 −𝑅𝐶 ⇒ 𝑣𝐶 (𝑡0 ) = 𝑣𝐶 (0)𝑒 −𝑅𝐶
𝑣𝐶 (0)

The capacitor voltage 𝑣𝐶 is continuous; therefore, at 𝑡 = 0

5.5)
𝑣𝐶 (0− ) = 𝑣𝐶 (0+ ) = 𝑉0 .

Thus for 𝑡0 ≥ 0,
5.6)
𝑡0
𝑣𝐶 (𝑡0 ) = 𝑉0 𝑒 −𝑅𝐶 .

Note, for 𝑡0 < 0, the capacitor is connected to an ideal voltage source, and the capacitor voltage 𝑣𝐶 =
𝑉0 . The capacitor current 𝑖𝐶 can be found by Eq. 5.1. For 𝑡0 < 0, the capacitor voltage is constant and
the capacitor current remains at 0. For 𝑡0 > 0, the capacitor current can be found as
5.7)
𝑡0
𝑑 (𝑉0 𝑒 −𝑅𝐶 ) 𝑉0 − 𝑡0
𝑖𝐶 (𝑡0 ) = 𝐶 =− 𝑒 𝑅𝐶 .
𝑑𝑡0 𝑅

Note that
5.8)
𝑉0
𝑖𝐶 (0− ) = 0 ≠ − = 𝑖𝐶 (0+ )
𝑅

and thus the capacitor current is not continuous. The resistor current, capacitor current, and capacitor
voltage are plotted in Figure 5-2.

Figure 5-2 Voltage and current curves of first-order RC circuit in Figure 5-1.
70
The differential equation in Eq. 5.2 is a homogeneous equation. The response calculated from a
homogeneous equation is called a natural response. From a circuit point of view, a natural response is
a naturally decaying response of a circuit without any external voltage or current sources.

Figure 5-3 First-order RC circuit with step input.

For the RC circuit shown in Figure 5-3, the voltage source has a step change at 𝑡 = 0. The differential
equation for solving the capacitor voltage can be written as
5.9)
𝑉𝑆 − 𝑣𝐶 𝑑𝑣𝐶
𝑖𝑅 = 𝑖𝐶 ⇒ =𝐶 .
𝑅 𝑑𝑡

The previous equation is a non-homogeneous differential equation because of the additional voltage
source. By separating the variables
5.10)
𝑑𝑣𝐶 𝑑𝑡
=− .
𝑣𝐶 − 𝑉𝑆 𝑅𝐶

By integrating both sides from 𝑡 = 0 to 𝑡 = 𝑡0 ,

5.11)
𝑣𝐶 (𝑡0 ) 𝑡0
𝑑𝑣𝐶 𝑑𝑡
∫ =∫ − .
𝑣𝐶 (0) 𝑣𝐶 − 𝑉𝑆 0 𝑅𝐶
𝑣𝐶 (𝑡0 ) − 𝑉𝑆 𝑡0
⇒ ln[𝑣𝐶 (𝑡0 ) − 𝑉𝑆 ] − ln[𝑣𝐶 (0) − 𝑉𝑆 ] = ln =−
𝑣𝐶 (0) − 𝑉𝑆 𝑅𝐶
𝑣𝐶 (𝑡0 ) − 𝑉𝑆 𝑡0
⇒ = 𝑒 −𝑅𝐶
𝑣𝐶 (0) − 𝑉𝑆
𝑡0
⇒ 𝑣𝐶 (𝑡0 ) = 𝑉𝑆 + [𝑣𝐶 (0) − 𝑉𝑆 ]𝑒 −𝑅𝐶
𝑡0
= 𝑉𝑆 + (𝑉0 − 𝑉𝑆 )𝑒 −𝑅𝐶 .

This response from a step input voltage is called step response. By replacing 𝑉𝑆 with 0, the previous
result will be the same as the natural response. For first-order RC circuits, the time constant is defined
as
5.12)
𝜏 = 𝑅𝐶.
71
A longer time constant means it takes longer time for the circuit to reach steady state. A larger
resistance or capacitance would lead to longer time constants.

For natural response and step response of first-order RC circuits, there exists a generic form of
5.13)
𝑡
𝑣𝐶 = 𝐴𝑒 −𝜏 + 𝐵.

By finding the boundary condition at 𝑡 → ∞,

5.14)
lim 𝑣𝐶 (𝑡) = 𝐵.
𝑡→∞

Then by finding the boundary condition at 𝑡 = 0,

5.15)
𝑣𝐶 (0) = 𝐴 + 𝐵 = 𝐴 + lim 𝑣𝐶 (𝑡) ⇒ 𝐴 = 𝑣𝐶 (0) − lim 𝑣𝐶 (𝑡)
𝑡→∞ 𝑡→∞

and thus both unknowns can be solved. The time constant of an arbitrary first-order RC circuit can be
found by calculating the equivalent resistance seen by the capacitance. Therefore, the generic form of
the capacitor voltage can be written as
5.16)
𝑡
−
𝑅𝑒𝑞 𝐶
𝑣𝐶 (𝑡) = lim 𝑣𝐶 (𝑡) + [𝑣𝐶 (0) − lim 𝑣𝐶 (𝑡)] 𝑒 .
𝑡→∞ 𝑡→∞

The latter part of Eq. 5.16 that decays exponentially with time is called the transient response or
transient output. The first part that remains constant is called the steady-state response or steady-
state output.

EX. 5.1
For the following circuit, please find 𝑣𝐶 (𝑡) and 𝑖𝐶 (𝑡).

First start by finding the time constant

𝜏 = 𝑅𝑒𝑞 𝐶 = (𝑅1 ∥ 𝑅2 )𝐶.

Since
𝑣𝑆 (0− ) = 0,
𝑣𝐶 (0− ) = 0 = 𝑣𝐶 (0+ ).

At 𝑡 → ∞, C becomes open circuit and

72
 𝑉0 𝑅2
𝑣𝐶 (𝑡) = .
𝑅1 + 𝑅2
Therefore,
𝑉0 𝑅2 𝑡
−
𝑣𝐶 (𝑡) = (1 − 𝑒 (𝑅1 ∥𝑅2 )𝐶 ) 𝑢(𝑡).
𝑅1 + 𝑅2
Since
𝑑𝑣𝐶 (𝑡)
𝑖𝐶 (𝑡) = 𝐶 ,
𝑑𝑡
𝑖𝐶 (𝑡) = 0 for t < 0 and
𝑡
𝑉0 𝑅2 𝑅1 + 𝑅2 −(𝑅 ∥𝑅 V0 − 𝑡
𝑖𝐶 (𝑡) = 𝐶 ( 𝑒 1 2 )𝐶 ) = 𝑒 (𝑅1 ∥𝑅2)𝐶
𝑅1 + 𝑅2 𝑅1 𝑅2 𝐶 𝑅1
for t > 0.
Therefore,
𝑉0 − 𝑡
𝑖𝐶 (𝑡) = [ 𝑒 (𝑅1 ∥𝑅2 )𝐶 ] 𝑢(𝑡)
𝑅1

5.1.2 F IRST -O RDER RL C IRCUITS

In an RL circuit, the inductor voltage is

5.17)
𝑑𝑖𝐿
𝑣𝐿 = 𝐿 .
𝑑𝑡

The inductor current 𝑖𝐿 is always continuous; otherwise the inductor voltage 𝑣𝐿 would be infinity. If
𝑑𝑖𝐿
no disturbance occurs in the circuit, 𝑣𝐿 = 0 = 𝐿 𝑑𝑡
and the inductor current 𝑖𝐿 would remain constant.
Assume a circuit is disturbed at 𝑡 = 0, then at 𝑡 < 0, 𝑣𝐿 = 0 and 𝑖𝐿 is constant. Similarly, as 𝑡
approaches infinity, 𝑣𝐿 = 0 and 𝑖𝐿 is constant. Therefore, the inductor L can be considered as short
circuit at those two time points.

Figure 5-4 First-order RL circuit.

For the RL circuit shown in Figure 5-4, the switch turns off at 𝑡 = 0. Because of KVL,
5.18)
𝑑𝑖𝐿
𝑣𝐿 = 𝑣𝑅 ⇒ 𝐿 = 𝑖𝑅 𝑅 = −𝑖𝐿 𝑅.
𝑑𝑡

By separating the variables

73
5.19)
𝑑𝑖𝐿 𝑑𝑡
=− .
𝑖𝐿 𝐿 ⁄𝑅

By integrating both sides from 𝑡 = 0 to 𝑡 = 𝑡0 , the natural response can be calculated as

5.20)
𝑡 𝑡0
𝑉0 −𝐿/𝑅
− 0
𝑖𝐿 (𝑡0 ) = 𝑖𝐿 (0)𝑒 𝐿/𝑅 = 𝑒 .
𝑅𝑆

The step response of a first-order RL circuit can be found to be similar to that of a first-order RC
circuit. For first-order RL circuits, the time constant is defined as
5.21)
𝐿
𝜏= .
𝑅

Like first-order RC circuits, there exists a generic form for the natural response and step response of
first-order RL circuits
5.22)
𝑡
𝑖𝐿 = 𝐴𝑒 −𝜏 + 𝐵.

By finding the boundary condition at 𝑡 → ∞,

5.23)
lim 𝑖𝐿 (𝑡) = 𝐵.
𝑡→∞

Then by finding the boundary condition at 𝑡 = 0,

5.24)
𝑖𝐿 (0) = 𝐴 + 𝐵 = 𝐴 + lim 𝑖𝐿 (𝑡) ⇒ 𝐴 = 𝑖𝐿 (0) − lim 𝑖𝐿 (𝑡)
𝑡→∞ 𝑡→∞

and thus both unknowns can be solved. The time constant of an arbitrary first-order RL circuit can be
found by calculating the equivalent resistance seen by the inductance. Therefore, the generic form of
the inductor current can be written as
5.25)
𝑡
−
𝐿/𝑅𝑒𝑞
𝑖𝐿 (𝑡) = lim 𝑖𝐿 (𝑡) + [𝑖𝐿 (0) − lim 𝑖𝐿 (𝑡)] 𝑒 .
𝑡→∞ 𝑡→∞

74
EX. 5.2
Please find 𝑖(𝑡) for the following circuit.

First assume 𝑡 ′ = 𝑡 − 2 s.
If 𝑡 ′ < 0 ⇒ 𝑣𝑥 = 1 A × 5 Ω = 5 V ⇒ 3𝑣𝑥 = 15 V.
If 𝑡 ′ > 0 ⇒ 𝑣𝑥 = 0 ⇒ 3𝑣𝑥 = 0.
The circuit can then be simplified as the following figure, where 𝑣𝑠 = 15 for 𝑡 ′ < 0 and 𝑣𝑠 = 0
elsewhere.

At 𝑡 ′ = 0,
15 V
𝑖(𝑡 ′ ) = = 1.5 A.
10 Ω
When 𝑡′ → ∞ ,
𝑖(𝑡 ′ ) = 0.
The time constant
𝐿 10 mH
𝜏= = = 1 ms.
𝑅 10 Ω
Therefore, for 𝑡 ′ > 0
𝑡′ ′
𝑖(𝑡 ′ ) = 1.5𝑒 −1 ms = 1.5𝑒 −1000𝑡
and finally for 𝑡 > 2 s
𝑖(𝑡) = 1.5𝑒 −1000(𝑡−2) (A).
For 𝑡 < 2 s
15 V
𝑖(𝑡) = = 1.5 A.
10 Ω

5.2 SECOND-ORDER AND HIGHER-ORDER CIRCUITS

Second-order circuits require solving second-order ODEs that are still solvable but require more effort
than first-order circuits. To solve circuits with even higher order, computers are normally used.

5.2.1 S ECOND -O RDER RLC CIRCUIT

Figure 5-5 shows a series RLC circuit. From KCL,

75
5.26)
𝑖𝐿 (𝑡) = 𝑖𝐶 (𝑡) = 𝑖𝑅 (𝑡) = 𝑖(𝑡)

Using KVL,
5.27)
𝑣𝑆 (𝑡) = 𝑣𝐿 (𝑡) + 𝑣𝐶 (𝑡) + 𝑣𝑅 (𝑡)
𝑑𝑖(𝑡)
=𝐿 + 𝑣𝐶 (𝑡) + 𝑅𝑖(𝑡)
𝑑𝑡
𝑑2 𝑣𝐶 (𝑡) 𝑑𝑣𝐶 (𝑡)
= 𝐿𝐶 2
+ 𝑅𝐶 + 𝑣𝐶 (𝑡)
𝑑𝑡 𝑑𝑡

Figure 5-5 Second-order series RLC circuit.

If 𝑣𝑆 (𝑡) = 𝑉0 𝑢(𝑡), for 𝑡 > 0. The characteristic equation is

5.28)
𝐿𝐶𝑥 2 + 𝑅𝐶𝑥 + 1 = 0

with roots of
5.29)

𝑅 𝑅 2 1
√
𝑥=− ± ( ) −
2𝐿 2𝐿 𝐿𝐶

If 𝑅 2 𝐶 > 4𝐿, there are two distinct real roots

5.30)

𝑅 𝑅 2 1 𝑅 𝑅 2 1
√
𝑥1 𝑎𝑛𝑑 𝑥2 = − − ( ) − 𝑎𝑛𝑑 − √
+ ( ) −
2𝐿 2𝐿 𝐿𝐶 2𝐿 2𝐿 𝐿𝐶

and the response will be in the form of

5.31)
𝑣𝐶 (𝑡) = 𝑉𝑓 + 𝐾1 𝑒 𝑥1 𝑡 + 𝐾2 𝑒 𝑥2 𝑡

and the response is overdamped. At 𝑡 → ∞, 𝑣𝐶 (𝑡) = 𝑉𝑓 = 𝑉0 . At 𝑡 = 0+ , 𝑣𝐶 (𝑡) = 0 thus

5.32)
𝐾1 + 𝐾2 = −𝑉0

76
 𝑑𝑣𝐶 (𝑡)
and 𝑖(0) = 𝐶 𝑑𝑡 𝑡=0
| = 0, since the inductor current is continuous, then

5.33)

𝑑𝑣𝐶 (𝑡) 𝑅 𝑅 2 1 𝑅 𝑅 2 1
𝐶 | √ √
= 0 = 𝐾1 𝑥1 + 𝐾2 𝑥2 = 𝐾1 (− − ( ) − ) + 𝐾2 (− + ( ) − ).
𝑑𝑡 𝑡=0 2𝐿 2𝐿 𝐿𝐶 2𝐿 2𝐿 𝐿𝐶

Therefore,
5.34)
2
𝑅 √ 𝑅 2 𝑅 𝑅 2 1 2
1 𝐾1 [ + √( ) − ]
2𝐿 + (2𝐿) − 𝐿𝐶 2𝐿 2𝐿 𝐿𝐶 2
𝑅 𝑅 1
𝐾2 = 𝐾1 = = −𝐾1 𝐿𝐶 [ + √( ) − ] .
𝑅 𝑅 2
1 1 2𝐿 2𝐿 𝐿𝐶
− 2𝐿 + √(2𝐿) − 𝐿𝐶 − 𝐿𝐶

and
5.35)
𝑅 𝑅 2 1 2 𝑅 𝑅 2 1
−( +√( ) − )𝑡
2𝐿 2𝐿 𝐿𝐶 𝑅 𝑅 2 1 −( −√( ) − )𝑡
𝑉0 𝑒 − 𝐿𝐶 [2𝐿 + √(2𝐿) − 𝐿𝐶 ] 𝑉0 𝑒 2𝐿 2𝐿 𝐿𝐶

𝑣𝐶 (𝑡) = 𝑉0 + 2 .
𝑅 𝑅 2 1
𝐿𝐶 [2𝐿 + √(2𝐿) − 𝐿𝐶 ] − 1

If 𝑅 2 𝐶 = 4𝐿, the two roots are real and equal

5.36)
𝑅
𝑥=−
2𝐿

and the response will be in the form of

5.37)
𝑅 𝑅
𝑣𝐶 (𝑡) = 𝑉𝑓 + 𝐾1 𝑒 𝑥𝑡 + 𝐾2 𝑡𝑒 𝑥𝑡 = 𝑉𝑓 + 𝐾1 𝑒 −2𝐿𝑡 + 𝐾2 𝑡𝑒 −2𝐿𝑡

and the response is critically damped. At 𝑡 = 0+ , 𝑣𝐶 (𝑡) = 0 thus 𝐾1 = −𝑉𝑓 = −𝑉0 and 𝑖(0) =
𝑑𝑣𝐶 (𝑡)
𝐶 𝑑𝑡 𝑡=0
| = 0 then

5.38)
𝑑𝑣𝐶 (𝑡)
𝐶 | = 0 = (𝐾1 𝑥𝑒 𝑥𝑡 + 𝐾2 𝑒 𝑥𝑡 + 𝐾2 𝑡𝑥𝑒 𝑥𝑡 )|𝑡=0 = 𝐾1 𝑥 + 𝐾2 = 0
𝑑𝑡 𝑡=0
𝑅
⇒ (−𝑉0 ) (− ) + 𝐾2 = 0
2𝐿
𝑅𝑉0
⇒ K2 = −
2𝐿

Therefore,

77
5.39)
𝑅 𝑅𝑉0 − 𝑅 𝑡
𝑣𝐶 (𝑡) = 𝑉0 − 𝑉0 𝑒 −2𝐿𝑡 − 𝑡𝑒 2𝐿 .
2𝐿

If 𝑅 2 𝐶 < 4𝐿, the two roots are complex conjugates and the response will be the same as Eq. 5.35 and
the response is underdamped. Since the two roots are complex conjugates, the output can also be
written in the form of
5.40)
𝑣𝐶 (𝑡) = 𝑉𝑓 + 𝐴𝑒 −𝛼𝑡 cos(𝜔𝑑 𝑡 + 𝜙)
𝐴 −𝛼𝑡 𝑖𝜔 𝑡 𝑖𝜙
= 𝑉𝑓 + 𝑒 (𝑒 𝑑 𝑒 + 𝑒 −𝑖𝜔𝑑 𝑡 𝑒 −𝑖𝜙 )
2
𝐴 𝐴
= 𝑉𝑓 + (cos 𝜙 + 𝑖 sin 𝜙 )𝑒 −𝛼𝑡+𝑖𝜔𝑑 𝑡 + (cos 𝜙 − 𝑖 sin 𝜙 )𝑒 −𝛼𝑡−𝑖𝜔𝑑 𝑡 .
2 2

Thus
5.41)
𝑅
𝛼=
2𝐿

and
5.42)

𝑅 2 1 1 𝑅 2
𝜔𝑑 = √− [( ) − ] = √ − ( ) .
2𝐿 𝐿𝐶 𝐿𝐶 2𝐿

𝑑𝑣𝐶 (𝑡)
At 𝑡 = 0+ , 𝑣𝐶 (𝑡) = 0 thus 𝐴 cos 𝜙 = −𝑉𝑓 = −𝑉0 and 𝑖(0) = 𝐶 | = 0 then
𝑑𝑡 𝑡=0

5.43)
𝑑𝑣𝐶 (𝑡)
𝐶 | = 0 = [−𝐴𝑒 −𝛼𝑡 𝜔𝑑 sin(𝜔𝑑 𝑡 + 𝜙) + 𝐴𝛼𝑒 −𝛼𝑡 cos(𝜔𝑑 𝑡 + 𝜙)]|𝑡=0
𝑑𝑡 𝑡=0

1 𝑅 2 𝑅
⇒ −𝐴√ − ( ) sin 𝜙 + 𝐴 cos 𝜙 = 0
𝐿𝐶 2𝐿 2𝐿

𝑅 1 𝑅 2
⇒ cos 𝜙 = √ − ( ) sin 𝜙
2𝐿 𝐿𝐶 2𝐿
𝑅
⇒ tan 𝜙 = 2𝐿
2
√ 1 −(𝑅)
𝐿𝐶 2𝐿

𝑅 2 1
𝑉0 (2𝐿)
⇒𝐴=− = −𝑉0 √1 + = −𝑉0 √ 𝐿𝐶
cos 𝜙 1 𝑅 2 1 𝑅 2
𝐿𝐶 − (2𝐿 ) 𝐿𝐶 − ( 2𝐿)

78
Finally, the capacitor voltage can be written as
5.44)
1 𝑅
𝐿𝐶 −
𝑅
𝑡 1 𝑅 2 2𝐿
𝑣𝐶 (𝑡) = 𝑉0 − 𝑉0 √ 𝑒 2𝐿 cos √ − ( ) 𝑡 + tan−1
.
1 𝑅 2 𝐿𝐶 2𝐿 1 𝑅 2
𝐿𝐶 − (2𝐿) √ −( )
[ 𝐿𝐶 2𝐿 ]

The capacitor voltage is an exponentially decaying sinusoidal signal. An example is shown in Figure
5-6. Underdamped circuit creates a signal peaking called an overshoot. Overshoot in a circuit must be
taken seriously since this transient signal may destroy sensitive components because of overvoltage.
The decaying sinusoid is a phenomenon called ringing. The time required for the ringing to die out is
called settling time. It is desirable to have a short settling time since it may degrade performance from
waiting for signals to settle.

Figure 5-6 Example of an underdamped circuit response.

5.2.2 S ECOND -O RDER RCRC CIRCUIT

Figure 5-7 Second-order RCRC circuit.

For the circuit shown in Figure 5-7, there are two capacitors and two resistors. From the introduction
of this chapter, this should be a second-order circuit. However, the differential equation has to be
derived for validation. By performing KCL on the node shared by 𝑅1 , 𝑅2 , and 𝐶1 such that
5.45)
𝑣𝑆 (𝑡) − 𝑣𝐶1 (𝑡) 𝐶1 𝑑𝑣𝐶1 (𝑡) 𝑣𝐶1 (𝑡) − 𝑣𝐶2 (𝑡)
= +
𝑅1 𝑑𝑡 𝑅2
79
 𝑣𝑆 (𝑡) 𝐶1 𝑑𝑣𝐶1 (𝑡) 𝑣𝐶1 (𝑡) 𝑣𝐶2 (𝑡)
⇒ = + − .
𝑅1 𝑑𝑡 𝑅1 ∥ 𝑅2 𝑅2

Next, perform KCL on the node shared by 𝑅2 and 𝐶2 such that

5.46)
𝑣𝐶1 (𝑡) − 𝑣𝐶2 (𝑡) 𝐶2 𝑑𝑣𝐶2 (𝑡)
=
𝑅2 𝑑𝑡
𝑅2 𝐶2 𝑑𝑣𝐶2 (𝑡)
⇒ 𝑣𝐶1 (𝑡) = 𝑣𝐶2 (𝑡) + .
𝑑𝑡

By substituting Eq. 5.46 into Eq. 5.45,

5.47)
𝑣𝑆 (𝑡) 𝐶1 𝑑𝑣𝐶2 (𝑡) 𝐶1 𝑅2 𝐶2 𝑑2 𝑣𝐶2 (𝑡) 𝑣𝐶2 (𝑡) 𝑅2 𝐶2 𝑑𝑣𝐶2 (𝑡) 𝑣𝐶2 (𝑡)
= + + + ∙ −
𝑅1 𝑑𝑡 𝑑𝑡 2 𝑅1 ∥ 𝑅2 𝑅1 ∥ 𝑅2 𝑑𝑡 𝑅2
𝑑2 𝑣𝐶2 (𝑡) 1 + 𝑅2 𝐶2 𝑑𝑣𝐶2 (𝑡) 𝑣𝐶2 (𝑡)
= 𝐶1 𝑅2 𝐶2 2
+ (𝐶1 + ) +
𝑑𝑡 𝑅1 ∥ 𝑅2 𝑑𝑡 𝑅1
𝑑2 𝑣𝐶2 (𝑡) 𝑅1 + 𝑅2 𝑑𝑣𝐶2 (𝑡)
⇒ 𝑣𝑆 (𝑡) = 𝑅1 𝐶1 𝑅2 𝐶2 2
+ [𝑅1 𝐶1 + (𝑅1 + 𝑅2 )𝐶2 + ] + 𝑣𝐶2 (𝑡),
𝑑𝑡 𝑅2 𝑑𝑡

which is a second-order differential equation. Solving this equation is the same as solving the RLC
circuit in 5.2.1.
EX. 5.3
Please find 𝑣𝑜𝑢𝑡 (𝑡) for the following circuit, where 𝑅1 = 5 kΩ,. 𝑅1 = 10 kΩ, 𝐶1 = 10 μF, and 𝐶2 = 1 μF.

Even though this circuit has two capacitors, the circuit is not a second-order circuit. After the switch is
opened, the top part of the circuit and bottom of the circuit has only one connection. Because of KCL,
the current flowing through that connection must be zero. It is more obvious if the circuit is redrawn
as the following circuit.

80
It can be seen that the circuit is actually composed of two independent first-order circuits.
For the top circuit,
5
𝑣𝐶1 (𝑡) = 18 × =6V
5 + 10
when 𝑡 = 0.
As 𝑡 → ∞, 𝑣𝐶1 (𝑡) = 0.
The time constant 𝜏1 = 𝑅1 𝐶1 = 5 kΩ × 10 μF = 50 ms,
and thus
𝑣𝐶1 (𝑡) = 6𝑒 −20𝑡 V.

For the bottom circuit,

10
𝑣𝐶2 (𝑡) = 18 × = 12 V
5 + 10
when 𝑡 = 0.
As 𝑡 → ∞, 𝑣𝐶2 (𝑡) = 0.
The time constant 𝜏2 = 𝑅2 𝐶2 = 10 kΩ × 1 μF = 10 ms,
and thus
𝑣𝐶2 (𝑡) = 12𝑒 −100𝑡 V.

Finally,
𝑣𝑜𝑢𝑡 (𝑡) = 𝑣𝐶1 (𝑡) + 𝑣𝐶2 (𝑡) = 6𝑒 −20𝑡 + 12𝑒 −100𝑡 V.
for t>0 and 𝑣𝑜𝑢𝑡 (𝑡) = 18𝑉 for t<0.

5.2.3 H IGHER -O RDER CIRCUITS

Figure 5-8 shows a third-order low pass filter circuit. For 𝐿1 ,

5.48)
𝑑𝑖𝐿1 (𝑡)
𝑣𝑆 (𝑡) − 𝑣𝐶 (𝑡) = 𝐿1 .
𝑑𝑡

For C,
5.49)
𝑑𝑣𝐶 (𝑡)
𝑖𝐶 (𝑡) = 𝐶 .
𝑑𝑡

For 𝐿2 ,
5.50)
𝑑𝑖𝐿2 (𝑡)
𝑣𝐶 (𝑡) − 𝑣O (𝑡) = 𝐿2 .
𝑑𝑡

Also,
5.51)
𝑣𝑂 (𝑡) = 𝑅𝑖𝐿2 (𝑡)

81
 𝐿2 𝑑𝑣𝑂 (𝑡)
⇒ 𝑣𝐶 (𝑡) = 𝑣𝑂 (𝑡) +
𝑅 𝑑𝑡

By performing KCL on the node shared by 𝐿1 , 𝐿2 , and 𝐶 such that

5.52)
𝑖𝐿1 (𝑡) = 𝑖𝐶 (𝑡) + 𝑖𝐿2 (𝑡)
𝑑𝑖𝐿1 (𝑡) 𝑑𝑖𝐶 (𝑡) 𝑑𝑖𝐿2 (𝑡)
⇒ = +
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑣𝑆 (𝑡) − 𝑣𝐶 (𝑡) 𝑑2 𝑣𝐶 (𝑡) 𝑣𝐶 (𝑡) − 𝑣O (𝑡)
⇒ =𝐶 +
𝐿1 𝑑𝑡 2 𝐿2
𝑣𝑆 (𝑡) 𝑑2 𝑣𝐶 (𝑡) 𝑣𝐶 (𝑡) 𝑣𝐶 (𝑡) 𝑣O (𝑡)
⇒ =𝐶 + + −
𝐿1 𝑑𝑡 2 𝐿1 𝐿2 𝐿2
𝑣𝑆 (𝑡) 𝐶𝐿2 𝑑3 𝑣𝑂 (𝑡) 𝐶𝑑2 𝑣𝑂 (𝑡) 𝐿1 + 𝐿2 𝑑𝑣𝑂 (𝑡) 𝑣𝑂 (𝑡)
⇒ = + + ∙ +
𝐿1 𝑅 𝑑𝑡 3 𝑑𝑡 2 𝑅𝐿1 𝑑𝑡 𝐿1
𝐶𝐿1 𝐿2 𝑑3 𝑣𝑂 (𝑡) 𝐶𝐿1 𝑑2 𝑣𝑂 (𝑡) 𝐿1 + 𝐿2 𝑑𝑣𝑂 (𝑡)
⇒ 𝑣𝑆 (𝑡) = + + + 𝑣𝑂 (𝑡),
𝑅 𝑑𝑡 3 𝑑𝑡 2 𝑅 𝑑𝑡

which is a third-order differential equation. We will leave solving higher-order differential equations
to the “Engineering Mathematics” course.

Figure 5-8 A third-order low-pass filter.

Figure 5-9 shows a circuit with three capacitors, two resistors, and an OPA. The circuit can be solved
by first performing KCL on the node shared by 𝑅1 , 𝐶1 , 𝐶2 , and 𝐶3 such that
5.53)
𝑑[𝑣𝑆 (𝑡) − 𝑣1 (𝑡)] 𝑑𝑣1 (𝑡) 𝑣1 (𝑡) 𝑑[𝑣1 (𝑡) − 𝑣𝑂 (𝑡)]
𝐶1 = 𝐶2 + + 𝐶3 .
𝑑𝑡 𝑑𝑡 𝑅1 𝑑𝑡

Next perform KCL on the node shared by 𝐶2 and 𝑅2 such that

5.54)
𝑑𝑣1 (𝑡) 𝑣𝑂 (𝑡)
𝐶2 =− .
𝑑𝑡 𝑅2

By substituting Eq. 5.54 into Eq. 5.53,

5.55)
𝑑𝑣𝑆 (𝑡) 𝑑2 𝑣1 (𝑡) 𝑑𝑣1 (𝑡) 𝑣1 (𝑡)
𝐶1 = 𝑅2 𝐶2 𝐶3 2
+ (𝐶1 + 𝐶2 + 𝐶3 ) + ,
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑅1

82
which is a second-order differential equation with three capacitors. The order of a circuit is usually
the total number of capacitors and inductors, but there are exceptions.

Figure 5-9 Second-order high-pass filter circuit.

EX. 5.4
Please find v(t) of the following circuit.

First only consider the switch that switches first at 𝑡 = 0,

𝑣(𝑡 = 0) = 0
and
10 kΩ
𝑣(𝑡 → ∞) = (10 mA × ) × 20 kΩ = 33.33 V.
50 kΩ + 10 kΩ
The equivalent resistance seen by the capacitor is
𝑅𝑒𝑞 = 40 kΩ ∥ 20 kΩ = 13.33 kΩ
and thus the time constant
𝜏 = 𝑅𝑒𝑞 𝐶 = 133.3 μs.
The capacitor voltage can then be derived as
𝑣(𝑡) = 33.33 − 33.33𝑒 −7500𝑡 V.
Next consider the switch that switch at 𝑡 = 0.2 ms,
𝑣(𝑡 = 0.2 ms) = 25.89 V.
′
Assume 𝑡 = 𝑡 − 0.2 ms,
then 𝑣(𝑡’ = 0) = 25.89 V.
As 𝑡 ′ → ∞, both the 10 mA current source and the 10 V voltage source contribute to the capacitor
voltage. Using superposition,
𝑣(𝑡 ′ ) = 𝑣10mA (𝑡 ′ ) + 𝑣10V (𝑡 ′ ),
where

83
 10
𝑣10mA (𝑡 ′ ) = (10 mA × ) × (20 kΩ ∥ 10 kΩ) = 14.29 V
20 + 10 + 20 ∥ 10 + 10
and
20 ∥ 40
𝑣10V (𝑡 ′ ) = 10 × = 5.71 V.
20 ∥ 40 + 10
Thus,
𝑣( 𝑡 ′ → ∞) = 14.29 + 5.71 = 20 V.
The equivalent resistance seen by the capacitor is
𝑅𝑒𝑞 = 40 ∥ 20 ∥ 10 = 5.714 kΩ
and thus the time constant
𝜏 = 𝑅𝑒𝑞 𝐶 = 57.14 μs.
Finally,
′
𝑣(𝑡’) = 20 + 5.89𝑒 −17500𝑡 V
and
𝑣(𝑡) = 20 + 5.89𝑒 −17500(𝑡−0.2 ms) V.
To summarize all three cases
0 V, if 𝑡 < 0
𝑣(𝑡) = {33.33 − 33.33𝑒 −7500𝑡 V, if 0 < 𝑡 < 0.2 ms
20 + 5.89𝑒 −17500(𝑡−0.2 ms) V, if 𝑡 > 0.2 ms

5.3 OPA INTEGRATORS AND DIFFERENTIATORS

OPAs not only can perform elementary arithmetic operations shown in Chapter 3, but also perform
integration and differentiation. Figure 5-10 shows an OPA integrator. Because of virtual short, the
noninverting input voltage 𝑣𝑖,𝑛 = 0. The input-output relation can be derived as
5.56)
𝑣𝑆 (𝑡) − 0 𝑑𝑣𝐶 (𝑡) 𝑑[0 − 𝑣𝑂 (𝑡)]
= 𝑖𝐶 (𝑡) = 𝐶 =𝐶
𝑅 𝑑𝑡 𝑑𝑡
1 𝑡
⇒ 𝑣𝑂 (𝑡) = − ∫ 𝑣 (𝑡)𝑑𝑡,
𝑅𝐶 −∞ 𝑆

which is the integration of the input signal 𝑣𝑆 (𝑡).

Figure 5-10 OPA integrator.

84
Figure 5-11 shows an OPA differentiator. Because of virtual short, the noninverting input voltage
𝑣𝑖,𝑛 = 0 . The input-output relation can be derived as
5.57)
𝑑[𝑣𝑆 (𝑡) − 0] 0 − 𝑣𝑂 (𝑡)
𝐶 = 𝑖𝐶 (𝑡) =
𝑑𝑡 𝑅
𝑑𝑣𝑆 (𝑡)
⇒ 𝑣𝑂 (𝑡) = −𝑅𝐶 ,
𝑑𝑡

which is the differentiation of the input signal 𝑣𝑆 (𝑡).

Figure 5-11 OPA differentiator.

EX. 5.5
The following circuit is a signal generation circuit. The first OPA is configured as an integrator. The
second OPA is configured as an inverting hysteretic comparator. The third OPA is configured as an
inverting comparator. Assume all OPAs are ideal and have supplies of ±10 V, please design this circuit
such that 𝑣1 is a 1 kHz triangular wave with peak-to-peak voltage of 5 V.

The peak-to-peak voltage of 𝑣1 is determined by the hysteresis of the inverting hysteretic comparator.
From Chapter 4, the output transitions are at

85
 𝑅1
𝑣𝐼𝑁 = ±𝑣𝑂𝑈𝑇 .
𝑅1 + 𝑅2
For a 5 Vp−p output,
𝑅1 2.5 1
= =
𝑅1 + 𝑅2 10 4
and thus
𝑅2 = 3𝑅1 .

The frequency of the triangular wave is determined by the values 𝑅 and 𝐶. The period of a 1 kHz signal
is 1 ms. If the triangular wave is symmetric, the wave spends 0.5 ms rising and 0.5 ms falling. The
output of the OPA integrator can be written as
1 𝑡
𝑣1 (𝑡) = − ∫ 𝑣 (𝜏)𝑑𝜏.
𝑅𝐶 −∞ 3
Assume 𝑣3 switches from +10 𝑉 to −10 𝑉 at 𝑡 = 𝑡0 , the output voltage of 𝑣1 for 𝑡 > 𝑡0 can be written
as
1 𝑡 1 𝑡0 10(𝑡 − 𝑡0 )
𝑣1 (𝑡) = − ∫ 𝑣3 (𝜏)𝑑𝜏 − ∫ 𝑣3 (𝜏)𝑑𝜏 = − 2.5.
𝑅𝐶 𝑡0 𝑅𝐶 −∞ 𝑅𝐶
To have a period of 1 ms, the output 𝑣1 must rise from −2.5 V to 2.5 V in exactly 0.5 ms. Therefore,
10 × 0.5 ms
𝑣1 (0.5 ms + 𝑡0 ) = 2.5 = − 2.5
𝑅𝐶
and thus
𝑅𝐶 = 1 ms.

Let 𝑅 = 𝑅1 = 1 kΩ, then 𝑅2 = 3 kΩ and 𝐶 = 1 μF.

86
 PART III AC ANALYSIS
In this part, AC analysis using sinusoidal steady-state analysis and Laplace transform will be
introduced. Transient analysis was introduced in the previous part, which is circuit analysis in the
time domain. AC analysis will perform circuit analysis in the frequency domain. The major benefit of
performing analysis in the frequency domain is that instead of solving a set of differential equations, a
set of linear equations are to be solved which is a lot easier. However, AC analysis has its limitations
and will be discussed in detail.

CHAPTER 6 LAPLACE TRANSFORM AND SINUSOIDAL STEADY STATE

In Part II, only dc and pulsed inputs were used as stimulus for first-order circuits. In this chapter,
sinusoidal inputs will be discussed. It will be shown that analyzing sinusoidal signal in the frequency
domain will significantly simplify the calculations.

The Laplace transform, Fourier transform, and phasor transform are used in this chapter for circuit
analysis. However, the definitions of Laplace transform and Fourier transform are slightly different in
the Engineering Mathematics course and the Signals and Systems course. The Laplace transform in
Engineering Mathematics is defined as
6.1)
∞
𝑋(𝑠) = ℒ{𝑥(𝑡)} = ∫ 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡.
0

Whereas, the Laplace transform in Signals and Systems is defined as

6.2)
∞
𝑋(𝑠) = ℒ{𝑥(𝑡)} = ∫ 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡.
−∞

The definition above is sometimes called the bilateral Laplace transform, which is sometimes denoted
as
6.3)
∞
𝑋(𝑠) = ℬ{𝑥(𝑡)} = ∫ 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡.
−∞

When the signal 𝑥(𝑡) is multiplied by the unit step function, 𝑢(𝑡), or if 𝑥(𝑡) = 0 for 𝑡 < 0, the
definitions of the Laplace transform in the two courses would be the same. The Fourier transform and
the inverse Fourier transform in Engineering Mathematics are defined as
6.4)
∞
1
𝑋(𝜔) = ℱ{𝑥(𝑡)} = ∫ 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡
√2𝜋 −∞
∞ .
−1 {𝑋(𝜔)}
1 𝑗𝜔𝑡
𝑥(𝑡) = ℱ = ∫ 𝑋(𝜔)𝑒 𝑑𝜔
{ √2𝜋 −∞

87
Whereas, the Fourier transform and the inverse Fourier transform in Signals and Systems are defined
as
6.5)
∞
𝑋(𝜔) = ℱ{𝑥(𝑡)} = ∫ 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡
−∞
.
−1 {𝑋(𝜔)}
1 ∞
𝑥(𝑡) = ℱ = ∫ 𝑋(𝜔)𝑒 𝑗𝜔𝑡 𝑑𝜔
{ 2𝜋 −∞

The Fourier transform in Signals and Systems is a special case of the Bilateral Laplace transform
defined in that course, where 𝜎 = 0 for the complex frequency
6.6)
𝑠 = 𝜎 + 𝑗𝜔.

The definition of the Laplace transform used in this course is identical to that defined in the
Engineering Mathematics course. The definitions of the Fourier transform and inverse Fourier
transform used in this course are identical to that used in the Signals and Systems course.

6.1 SINUSOIDAL SIGNALS AND FOURIER SERIES

A typical sinusoidal signal is shown in Figure 6-1. The sinusoidal signal can be expressed in the form
of
6.7)
𝑥(𝑡) = 𝑋𝑜𝑓𝑓 + 𝑋𝑚 sin(2𝜋𝑓𝑡 + 𝜙),

where 𝑋𝑜𝑓𝑓 is the offset, 𝑋𝑚 is the amplitude or magnitude, 𝑓 is the frequency in Hertz (Hz), and 𝜙 is
the initial phase in radians. The period of the sinusoidal signal is defined as
6.8)
1
𝑇=
𝑓

in seconds (s).

Figure 6-1 A typical sinusoidal signal.

A periodic signal is defined as

88
6.9)
𝑥(𝑡) = 𝑥(𝑡 + 𝑛𝑇),

where 𝑛 is an arbitrary integer and 𝑇 is the period. A periodic signal can be express as a Fourier series
such that
6.10)
∞
𝑎0 2𝑛𝜋 2𝑛𝜋
𝑥(𝑡) = + ∑ [𝑎𝑛 cos ( 𝑡) + 𝑏𝑛 sin ( 𝑡)],
2 𝑇 𝑇
𝑛=1

where
6.11)
2 2𝑛𝜋
𝑎𝑛 = ∫ 𝑥(𝑡) cos ( 𝑡) 𝑑𝑡
𝑇 𝑇 𝑇

and
6.12)
2 2𝑛𝜋
𝑏𝑛 = ∫ 𝑥(𝑡) sin ( 𝑡) 𝑑𝑡.
𝑇 𝑇 𝑇

As a result, if a periodic signal is used as the input of a linear circuit, superposition can be applied such
that the overall response is the sum of the individual responses of each sinusoidal component. Since
this applies to arbitrary periodic signals, solving the sinusoidal response of circuits is of much interest
in circuit analysis.

6.2 FIRST-ORDER CIRCUITS WITH SINUSOIDAL INPUTS

In Chapter 5, constant voltage or current were used as the forcing function of differential equations.
Other forcing functions have been discussed in the Engineering Mathematics course. In this chapter,
sinusoidal forcing functions are discussed. The input of the circuit in Figure 5-3 can be an arbitrary
sinusoidal signal as shown in Figure 6-2. The first-order differential equation of the RC circuit can be
written as
6.13)
𝑑𝑣𝐶 (𝑡) 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙) 𝑢(𝑡) − 𝑣𝐶 (𝑡) 𝑑𝑣𝐶 (𝑡) 𝑣𝐶 (𝑡) 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙) 𝑢(𝑡)
𝐶 = ⇒𝐶 + = .
𝑑𝑡 𝑅 𝑑𝑡 𝑅 𝑅

The capacitor voltage will have the form of

6.14)
𝑡
𝑣𝐶 (𝑡) = [𝑋𝑒 −𝑅𝐶 + 𝑌 sin(𝜔0 𝑡) + 𝑍 cos(𝜔0 𝑡)] 𝑢(𝑡).

For 𝑡 < 0, 𝑣𝐶 (𝑡) = 0. By substituting Eq. 6.14 into Eq. 6.13,

89
6.15)
𝑋 𝑡 𝑋 𝑡 𝑌 𝑍
[− e−𝑅𝐶 + 𝜔0 𝑌𝐶 cos(𝜔0 𝑡) − 𝜔0 𝑍𝐶 sin(𝜔0 𝑡)] + [ 𝑒 −𝑅𝐶 + sin(𝜔0 𝑡) + cos(𝜔0 𝑡)]
𝑅 𝑅 𝑅 𝑅
𝑉𝑆 sin(𝜔0 𝑡 + 𝜙)
=
𝑅

for 𝑡 ≥ 0. Since
6.16)
𝑉𝑆 sin(𝜔0 𝑡 + 𝜙) 𝑉𝑆 𝑉𝑆
= sin(𝜔0 𝑡) cos(𝜙) + cos(𝜔0 𝑡) sin(𝜙),
𝑅 𝑅 𝑅

it can be concluded that

6.17)
𝜔0 𝑌𝑅𝐶 + 𝑍 = 𝑉𝑆 sin(𝜙)
{
−𝜔0 𝑍𝑅𝐶 + 𝑌 = 𝑉𝑆 cos(𝜙)
𝜔0 𝑅𝐶 sin(𝜙) + cos(𝜙)
𝑌= 𝑉𝑆
1 + 𝜔02 𝑅2 𝐶 2
⇒ .
sin(𝜙) − 𝜔0 𝑅𝐶 cos(𝜙)
𝑍= 𝑉𝑆
{ 1 + 𝜔02 𝑅2 𝐶 2

Let
6.18)
𝜃 = tan−1 𝜔0 𝑅𝐶

such that
6.19)
𝜔0 𝑅𝐶
sin(𝜃) =
√1 + 𝜔02 𝑅2 𝐶 2
,
1
cos(𝜃) =
{ √1 + 𝜔02 𝑅2 𝐶 2

then
6.20)
sin(𝜃) sin(𝜙) + cos(𝜃) cos(𝜙) 𝑉𝑆 cos(𝜙 − 𝜃)
𝑌= 𝑉𝑆 =
√1 + 𝜔02 𝑅2 𝐶 2 √1 + 𝜔02 𝑅2 𝐶 2
cos(θ) sin(𝜙) − sin(𝜃) cos(𝜙) 𝑉𝑆 sin(𝜙 − 𝜃)
𝑍= 𝑉𝑆 =
{ √1 + 𝜔02 𝑅 2 𝐶 2 √1 + 𝜔02 𝑅 2 𝐶 2

and thus
6.21)
𝑡 𝑉𝑆 cos(𝜙 − 𝜃) 𝑉𝑆 sin(𝜙 − 𝜃)
𝑣𝐶 = [𝑋𝑒 −𝑅𝐶 + sin(𝜔0 𝑡) + cos(𝜔0 𝑡)] 𝑢(t)
√1 + 𝜔02 𝑅2 𝐶 2 √1 + 𝜔02 𝑅2 𝐶 2

90
 𝑡 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙 − 𝜃)
= [𝑋𝑒 −𝑅𝐶 + ] 𝑢(𝑡).
√1 + 𝜔02 𝑅2 𝐶 2

Since the capacitor voltage is zero at 𝑡 = 0,

6.22)
𝑉𝑆 sin(𝜙 − 𝜃)
𝑋= − .
√1 + 𝜔02 𝑅2 𝐶 2

Therefore, the capacitor voltage can be finally solved as

6.23)
𝑉𝑆 𝑡 𝑉𝑆
𝑣𝐶 (𝑡) = [− sin(𝜙 − 𝜃) 𝑒 −𝑅𝐶 + sin(𝜔0 𝑡 + 𝜙 − 𝜃)] 𝑢(𝑡).
√1 + 𝜔02 𝑅2 𝐶 2 √1 + 𝜔02 𝑅2 𝐶 2

𝑡
The first part of the solution decays with time (𝑒 −𝑅𝐶 ) and is called the transient response of the circuit.
The second part of the solution is a constant sinusoidal output and is called the steady-state response
of the circuit. One interesting characteristics of the steady-state response is that the output is a
sinusoidal signal that has the same frequency as the input. Only the phase and amplitude change. This
is an important characteristic of a linear time-invariant (LTI) system, which will be discussed in detail
in the Signals and Systems course.

Figure 6-2 A first-order RC circuit with sinusoidal input.

6.3 LAPLACE TRANSFORM FOR CIRCUIT ANALYSIS

Laplace transform can also be used to solve the previous problem. By performing Laplace transform
on both sides of Eq. 6.13.
6.24)
𝑑𝑣𝐶 𝑣𝐶 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙)𝑢(𝑡)
ℒ {𝐶 + } = ℒ{ }
𝑑𝑡 𝑅 𝑅
𝑉𝐶 (𝑠) 𝑉𝑆 𝜔0 𝑠
𝑠𝐶𝑉𝐶 (𝑠) − 𝐶𝑣𝐶 (0) + = [ 2 2 cos(𝜙) + 2 sin(𝜙)].
𝑅 𝑅 𝑠 + ω0 𝑠 + ω20

Since 𝑣𝐶 (0) = 0,
6.25)
𝑉𝑆 𝑠 sin(𝜙) + 𝜔0 cos(𝜙)
𝑉𝐶 (𝑠) =
1 + 𝑠𝑅𝐶 𝑠 2 + 𝜔02
91
 𝑉𝑆 [𝜔0 𝑅𝐶 cos(𝜙) − sin(𝜙)] 1
= +
(1 + 𝜔02 𝑅2 𝐶 2 ) 𝑠+
1
𝑅𝐶
𝑉𝑆 [𝜔0 𝑅𝐶 sin(𝜙) + cos(𝜙)] 𝜔0 𝑉𝑆 [sin(𝜙) − 𝜔0 𝑅𝐶 cos(𝜙)] 𝑠
2 2 2 2+ 2 2 2 .
1 + 𝜔0 𝑅 𝐶 2
𝑠 + 𝜔0 1 + 𝜔0 𝑅 𝐶 𝑠 + 𝜔02
2

By performing inverse Laplace transform,

6.26)
𝑣𝐶 (𝑡) = ℒ −1 {𝑉𝐶 (𝑠)}
𝑉𝑆 [𝜔0 𝑅𝐶 cos(𝜙) − sin(𝜙)] − 𝑡
= 𝑒 𝑅𝐶 𝑢(𝑡) +
(1 + 𝜔02 𝑅2 𝐶 2 )
𝑉𝑆 [𝜔0 𝑅𝐶 sin(𝜙) + cos(𝜙)] 𝑉𝑆 [sin(𝜙) − 𝜔0 𝑅𝐶 cos(𝜙)]
2 2 2 sin(𝜔0 𝑡) 𝑢(𝑡) + cos(𝜔0 𝑡) 𝑢(t)
1 + 𝜔0 𝑅 𝐶 1 + 𝜔02 𝑅2 𝐶 2
𝑉𝑆 sin(𝜙 − 𝜃) 𝑡 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙 − 𝜃)
= [− 𝑒 −𝑅𝐶 + ] 𝑢(𝑡),
√1 + 𝜔02 𝑅2 𝐶 2 √1 + 𝜔02 𝑅2 𝐶 2

which is identical to Eq. 6.23. The usage of Laplace transform appears to simplify the process of
finding the final time-domain response.

As shown previously, Laplace transform can be used to find the overall response of a circuit with a
sinusoidal input. However, it has already been shown in the Engineering Mathematics course that it
can be applied to all kinds of inputs. One special input is the Dirac delta function or the unit impulse
function 𝛿(𝑡). The output of a circuit or system is called the impulse response when the input is 𝛿(𝑡),
which is normally denoted as ℎ(𝑡). If the impulse response is known, the output response of an
arbitrary input signal 𝑥(𝑡) can be found to be
6.27)
∞
𝑦(𝑡) = 𝑥(𝑡) ∗ ℎ(𝑡) = ∫ 𝑥(𝜏)ℎ(𝑡 − 𝜏)𝑑𝜏,
−∞

which is the convolution between the input signal and the impulse response. With this in mind, the
reason why solving sinusoidal response is of much interest in circuit analysis can be explained. If 𝑥(𝑡)
is periodic, it can be expressed in the Fourier series form such that
6.28)
𝑦(𝑡) = 𝑥(𝑡) ∗ ℎ(𝑡)
∞
𝑎0 2𝑛𝜋 2𝑛𝜋
= { + ∑ [𝑎𝑛 cos ( 𝑡) + 𝑏𝑛 sin ( 𝑡)]} ∗ ℎ(𝑡)
2 𝑇 𝑇
𝑛=1
∞ ∞
𝑎0 2𝑛𝜋 2𝑛𝜋
= ∫ { + ∑ [𝑎𝑛 cos ( 𝜏) + 𝑏𝑛 sin ( 𝜏)]} ℎ(𝑡 − 𝜏)𝑑𝜏 ,
−∞ 2 𝑇
𝑛=1
𝑇

∞ ∞ ∞
𝑎0 2𝑛𝜋 2𝑛𝜋
= ∫ ℎ(𝑡 − 𝜏)𝑑𝜏 + ∑ ∫ [𝑎𝑛 cos ( 𝜏) + 𝑏𝑛 sin ( 𝜏)] ℎ(𝑡 − 𝜏)𝑑𝜏,
2 −∞ −∞ 𝑇 𝑇
𝑛=1

which is the sum of the individual sinusoidal response.

92
 6.4 SINUSOIDAL STEADY STATE AND PHASOR TRANSFORM

Under sinusoidal steady state, it is assumed that the time has passed long enough such that the
transient response has decayed to zero. Laplace transform is not defined for 𝑡 < 0 and thus likely have
a discontinuity at 𝑡 = 0. Fourier transform is defined for all 𝑡 and appears to be more suitable for
calculation under sinusoidal steady state.

Figure 6-3 A first-order RC circuit with sinusoidal input under sinusoidal steady state.

If the circuit in Figure 6-3 is under sinusoidal steady state, Eq. 6.13 can be rewritten as
6.29)
𝑑𝑣𝐶 𝑣𝐶 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙)
𝐶 + = .
𝑑𝑡 𝑅 𝑅

By performing Fourier transform on both sides,

6.30)
𝑑𝑣𝐶 𝑣𝐶 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙)
ℱ {𝐶 + }= ℱ{ }
𝑑𝑡 𝑅 𝑅
𝑉𝑆 𝑒 𝑗(𝜔0 𝑡+𝜙) − 𝑒 −𝑗(𝜔0 𝑡+𝜙)
= ℱ{ }
𝑅 2𝑗
𝑉𝐶 (𝜔) 𝑉𝑆 𝑗𝜙
⇒ 𝑗𝜔𝐶𝑉𝐶 (𝜔) + = [𝑒 2𝜋𝛿(𝜔 − 𝜔0 ) − 𝑒 −𝑗𝜙 2𝜋𝛿(𝜔 + 𝜔0 )]
𝑅 2𝑗𝑅

The frequency-domain representation of𝑣𝐶 (𝑡) can then be derived as

6.31)
𝑉𝑆 𝑒 𝑗𝜙 2𝜋𝛿(𝜔 − 𝜔0 ) − 𝑒 −𝑗𝜙 2𝜋𝛿(𝜔 + 𝜔0 )
𝑉𝐶 (𝜔) =
1 + 𝑗𝜔𝑅𝐶 2𝑗
𝑉𝑆 𝑒 𝑗𝜙 2𝜋𝛿(𝜔 − 𝜔0 ) 𝑉𝑆 𝑒 −𝑗𝜙 2𝜋𝛿(𝜔 + 𝜔0 )
= −
1 + 𝑗𝜔0 𝑅𝐶 2𝑗 1 − 𝑗𝜔0 𝑅𝐶 2𝑗
𝑉𝑆 𝑒 𝑗𝜙 (1 − 𝑗𝜔0 𝑅𝐶) 2𝜋𝛿(𝜔 − 𝜔0 ) 𝑉𝑆 𝑒 −𝑗𝜙 (1 + 𝑗𝜔0 𝑅𝐶) 2𝜋𝛿(𝜔 + 𝜔0 )
= −
1 + 𝜔02 𝑅2 𝐶 2 2𝑗 1 + 𝜔02 𝑅2 𝐶 2 2𝑗
𝑉𝑆 𝑒 𝑗𝜙 [cos(θ) − 𝑗sin(𝜃)] 2𝜋𝛿(𝜔 − 𝜔0 ) 𝑉𝑆 𝑒 −𝑗𝜙 [cos(θ) + 𝑗sin(𝜃)] 2𝜋𝛿(𝜔 + 𝜔0 )
= −
√1 + 𝜔02 𝑅2 𝐶 2 2𝑗 √1 + 𝜔02 𝑅2 𝐶 2 2𝑗

𝑉𝑆 𝑒 𝑗(𝜙−𝜃) 2𝜋𝛿(𝜔 − 𝜔0 ) − 𝑒 −𝑗(𝜙−𝜃) 2𝜋𝛿(𝜔 + 𝜔0 )

= [ ].
√1 + 𝜔02 𝑅 2 𝐶 2 2𝑗

93
By performing inverse Fourier transform,
6.32)
𝑣𝐶 (𝑡) = ℱ −1 {𝑉𝐶 (𝜔)}
𝑉𝑆 𝑒 𝑗(𝜙−𝜃) 2𝜋𝛿(𝜔 − 𝜔0 ) − 𝑒 −𝑗(𝜙−𝜃) 2𝜋𝛿(𝜔 + 𝜔0 )
= ℱ −1 { [ ]}
√1 + 𝜔02 𝑅2 𝐶 2 2𝑗
𝑉𝑆
= sin(𝜔0 𝑡 + 𝜙 − 𝜃),
√1 + 𝜔02 𝑅2 𝐶 2

which is same as the steady-state response derived in Eq. 6.26. It is quite obvious that using Fourier
transform to find the sinusoidal steady state response is not much simpler than solving with Laplace
transform and then removing the transient response.

A simpler method for analyzing a circuit under sinusoidal steady state is the phasor transform. Phasor
is defined as a complex number that carries the amplitude and phase information of a sinusoidal
function. Using Euler’s formula, a sinusoidal signal can be rewritten as
6.33)
𝑥(𝑡) = 𝑋𝑚 sin(𝜔𝑡 + 𝜙) = ℐ𝓂{𝑋𝑚 𝑒 𝑗(𝜔𝑡+𝜙) } = ℐ𝓂{𝑋𝑚 𝑒 𝑗𝜙 𝑒 𝑗𝜔𝑡 } = ℐ𝓂{𝑿𝑒 𝑗𝜔𝑡 },

where
6.34)
𝑿 = 𝑋𝑚 ∠𝜙 = 𝑋𝑚 𝑒 𝑗𝜙

is the phasor of the sinusoidal signal 𝑥(𝑡). We can then define the phasor transform as
6.35)
𝑿 = 𝒫{𝑥(𝑡)} = 𝒫{𝑋𝑚 sin(𝜔𝑡 + 𝜙)} = 𝑋𝑚 𝑒 𝑗𝜙 .

Similarly, the inverse phasor transform can be defined as

6.36)
𝑥(𝑡) = 𝒫 −1 {𝑿} = 𝒫 −1 {𝑋𝑚 𝑒 𝑗𝜙 } = ℐ𝓂{𝑋𝑚 𝑒 𝑗𝜙 𝑒 𝑗𝜔𝑡 } = 𝑋𝑚 sin(𝜔𝑡 + 𝜙).

Phasor itself can be a function of frequency such that

6.37)
𝑿 = 𝑿(𝜔) = 𝑿(𝑗𝜔) = 𝑋𝑚 (𝜔)∠𝜙(𝜔) = 𝑋𝑚 (𝜔)𝑒 𝑗𝜙(𝜔) .

Moreover, phasor transform can be considered as a special case for Fourier transform that is only
valid for sinusoids. The biggest benefit of using Laplace transform, Fourier transform, or phasor
transform to convert a signal to the frequency domain is its simplicity in dealing with derivatives.
Assume differentiation is performed on 6.33,
6.38)
𝑑𝑥(𝑡)
= 𝜔𝑋𝑚 cos(𝜔𝑡 + 𝜙)
𝑑𝑡

94
 𝜋
= 𝜔𝑋𝑚 sin ( − 𝜔𝑡 − 𝜙)
2
𝜋
= −𝜔𝑋𝑚 sin (𝜔𝑡 + 𝜙 − )
2
𝜋
𝑗(𝜔𝑡+𝜙− )
= ℐ𝓂 {−𝜔𝑋𝑚 𝑒 2 }

𝑗𝜋
= ℐ𝓂 {−𝑒 − 2 𝜔𝑋𝑚 𝑒 𝑗𝜙 𝑒 𝑗𝜔𝑡 }

= ℐ𝓂{𝑗𝜔𝑿𝑒 𝑗𝜔𝑡 }.

Therefore,
6.39)
𝑑𝑥(𝑡)
𝒫{ } = 𝑗𝜔𝑿.
𝑑𝑡

Similarly,
6.40)
𝑑2 𝑥(𝑡)
𝒫{ } = (𝑗𝜔)2 𝑿 = −𝜔2 𝑿.
𝑑𝑡 2

Since the phasor transform is a special case of Fourier transform that is only valid for sinusoids, their
symbols in the frequency domain are often interchangeable such as
6.41)
𝑿 = 𝑿(𝜔) = 𝑿(𝑗𝜔) = 𝑋(𝜔) = 𝑋(𝑗𝜔) = 𝑋(𝑠).

𝑋(𝑠) is valid for the s-plane where 𝑠 = 𝜎 + 𝑗𝜔. However, for simplicity it is also used interchangeably
for sinusoids such that 𝑠 = 𝑗𝜔.

With the phasor transform defined, we can perform phasor transform on 6.29 such that
6.42)
𝑑𝑣𝐶 𝑣𝐶 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙)
𝒫 {𝐶 + } = 𝒫{ }
𝑑𝑡 𝑅 𝑅
𝑉𝐶 (𝜔) 𝑉𝐼 (𝜔)
𝑗𝜔𝐶𝑉𝐶 (𝜔) + = ,
𝑅 𝑅

where
6.43)
𝑣𝑖 (𝑡) = 𝑉𝑆 sin(𝜔0 𝑡 + 𝜙).

Therefore,
6.44)
𝑉𝐼 (𝜔)
𝑉𝐶 (𝜔) = .
1 + 𝑗𝜔𝑅𝐶

95
By performing inverse phasor transform on both sides leads to
6.45)
𝑉𝐼 (𝜔)
𝑣𝐶 (𝑡) = 𝒫 −1 {𝑉𝐶 (𝜔)} = 𝒫 −1 { }
1 + 𝑗𝜔𝑅𝐶
−1 𝜔𝑅𝐶
−1
𝑉𝐼 (𝜔)𝑒 −𝑗 tan
=𝒫 { }
√1 + 𝜔02 𝑅2 𝐶 2
𝑉𝑆
= sin(𝜔0 𝑡 + 𝜙 − 𝜃),
√1 + 𝜔02 𝑅2 𝐶 2

which is the same as the previous derivation using Laplace transform and Fourier transform. It is
apparent that analysis with phasor transfor is a low simpler.

96
 CHAPTER 7 IMPEDANCE AND FREQUENCY-DOMAIN CIRCUIT ANALYSIS

In the previous chapter, it was shown that the phasor transform is suitable for converting a sinusoidal
steady-state signal into the frequency domain. After calculation in the frequency domain, the inverse
phasor transform can be used to convert back to time domain to find the time-domain solution. In this
chapter, the characteristics of various circuit elements are derived in the frequency domain.

7.1 RESISTORS, CAPACITORS, AND INDUCTORS IN THE FREQUENCY DOMAIN

The frequency-domain representation of a time-domain voltage signal can be found by performing

phasor transform such that
7.1)
𝑽 = 𝑉(𝜔) = 𝑉(𝑗𝜔) = 𝒫{𝑣(𝑡)}.

Similarly, for a time-domain current signal, the frequency-domain representation is

7.2)
𝑰 = 𝐼(𝜔) = 𝐼(𝑗𝜔) = 𝒫{𝑖(𝑡)}.

With voltage and current defined in the frequency domain, the characteristics of resistors, capacitors,
and inductors in the frequency domain can then be derived.

7.1.1 R ESISTORS

In the time domain, the relationship between the voltage across and the current through a resistor is
defined by Ohm’s law
7.3)
𝑣(𝑡)
𝑅= ,
𝑖(𝑡)

which can be rearranged as

7.4)
𝑣(𝑡) = 𝑅 ∙ 𝑖(𝑡).

By performing Phasor transform on both sides,

7.5)
𝒫{𝑣(𝑡)} = 𝒫{𝑅 ∙ 𝑖(𝑡)}
𝒫{𝑣(𝑡)} = 𝑅 ∙ 𝒫{𝑖(𝑡)}
𝑉(𝑗𝜔) = 𝑅 ∙ 𝐼(𝑗𝜔).

The phase of the voltage and the phase of the current are the same as shown in Figure 7-1. Under this
condition, it is said that the voltage and current are in phase.

97
 Figure 7-1 Sinusoidal voltage and current waveform of a resistor.

7.1.2 C APACITORS

The time-domain characteristics of a capacitor is governed by

7.6)
𝑑𝑣(𝑡)
𝑖(𝑡) = 𝐶 .
𝑑𝑡

By performing Phasor transform on both sides,

7.7)
𝑑𝑣(𝑡)
𝒫{𝑖(𝑡)} = 𝒫 {𝐶 }
𝑑𝑡
𝑑𝑣(𝑡)
𝒫{𝑖(𝑡)} = 𝐶 ∙ 𝒫 { }
𝑑𝑡
𝐼(𝑗𝜔) = 𝑗𝜔𝐶𝑉(𝑗𝜔)

π
The phase of the voltage lags the phase of the current by 90° or as shown in Figure 7-2.
2

Figure 7-2 Sinusoidal voltage and current waveform of a capacitor.

7.1.3 I NDUCTORS

The time-domain characteristics of an inductor is governed by

7.8)
𝑑𝑖(𝑡)
𝑣(𝑡) = 𝐿 .
𝑑𝑡

By performing Phasor transform on both sides,

7.9)
𝑑𝑖(𝑡)
𝒫{𝑣(𝑡)} = 𝒫 {𝐿 }
𝑑𝑡

98
 𝑑𝑖(𝑡)
𝒫{𝑣(𝑡)} = 𝐿 ∙ 𝒫 { }
𝑑𝑡
𝑉(𝑗𝜔) = 𝑗𝜔𝐿𝐼(𝑗𝜔).

π
The phase of the voltage leads the phase of the current by 90° or 2 as shown in Figure 7-3.

Figure 7-3 Sinusoidal voltage and current waveform of an inductor.

7.2 IMPEDANCE AND ADMITTANCE

Impedance is defined as the ratio of the voltage and current in the frequency domain
7.10)
𝑉(𝑗𝜔)
𝑍(𝑗𝜔) = ,
𝐼(𝑗𝜔)

where 𝑍 is in ohms (Ω). The impedance is a complex function of 𝑗𝜔 and can be written as
7.11)
𝑍(𝑗𝜔) = ℛℯ{𝑍(𝑗𝜔)} + 𝑗ℐ𝓂{𝑍(𝑗𝜔)}
= 𝑅(𝑗𝜔) + 𝑗𝑋(𝑗𝜔),

where 𝑅 is the real part called resistance and 𝑋 is the imaginary part called reactance. Similar to the
relationship between resistance and conductance, the reciprocal of impedance is called admittance
defined as
7.12)
1 𝐼(𝑗𝜔)
𝑌(𝑗𝜔) = =
𝑍(𝑗𝜔) 𝑉(𝑗𝜔)

in siemens (S). The admittance is also a complex function of 𝑗𝜔 and can be written as
7.13)
𝑌(𝑗𝜔) = ℛℯ{𝑌(𝑗𝜔)} + 𝑗ℐ𝓂{𝑌(𝑗𝜔)}
= 𝐺(𝑗𝜔) + 𝑗𝐵(𝑗𝜔),

where 𝐺 is the real part called conductance and 𝐵 is the imaginary part called susceptance.

The main benefit of using impedance for circuit analysis in the frequency domain is that high-order
differential equations in the time domain are simplified into linear equations in the frequency domain.
To solve the final time-domain solution, inverse Fourier transform can be performed on the

99
frequency-domain solution, which would be lot easier than solving a high-order differential equation
in the time domain. More often, the frequency-domain solution is sufficient to grasp the characteristics
of a circuit and converting back to time-domain would be unnecessary. Frequency-selective circuits or
filters to be discussed in the next chapter are such kind of circuits.

7.3 SERIES AND PARALLEL IN THE FREQUENCY DOMAIN

Calculations in the frequency domain or calculations using impedances simplify differential equations
into linear or polynomial equations. Using impedance representations for capacitors and inductors
would make calculations the same as calculating with resistors. It will be shown that all circuit
analysis techniques discussed in Chapter 2 apply to impedance calculations.

7.3.1 I MPEDANCES IN S ERIES

When impedances are in series as shown in Figure 7-4, the currents through all impedances are
identical because of KCL. Therefore, the total voltage across all the impedances can be written as
7.14)
𝑉(𝑗𝜔) = 𝑉1 (𝑗𝜔) + 𝑉2 (𝑗𝜔) + ⋯ + 𝑉𝑘 (𝑗𝜔)
= 𝐼1 (𝑗𝜔)𝑍1 (𝑗𝜔) + 𝐼2 (𝑗𝜔)𝑍2 (𝑗𝜔) + ⋯ + 𝐼𝑘 (𝑗𝜔)𝑍𝑘 (𝑗𝜔)
= 𝐼(𝑗𝜔)𝑍1 (𝑗𝜔) + 𝐼(𝑗𝜔)𝑍2 (𝑗𝜔) + ⋯ + 𝐼(𝑗𝜔)𝑍𝑘 (𝑗𝜔).

The equivalent impedance of impedances in series can then be defined as

7.15)
𝑉(𝑗𝜔)
𝑍𝑒𝑞 (𝑗𝜔) =
𝐼(𝑗𝜔)
= 𝑍1 (𝑗𝜔) + 𝑍2 (𝑗𝜔) + ⋯ + 𝑍𝑘 (𝑗𝜔)
𝑘

= ∑ 𝑍𝑛 (𝑗𝜔),
n=1

which is the same as resistors in series.

Figure 7-4 Impedances in series.

100
7.3.2 I MPEDANCES IN P ARALLEL

When impedances are in parallel as shown in Figure 7-5, the voltages across all impedances are
identical because of KVL. Therefore,
7.16)
𝑉(𝑗𝜔) = 𝑉1 (𝑗𝜔) = 𝑉2 (𝑗𝜔) = ⋯ = 𝑉𝑘 (𝑗𝜔).

From KCL,
7.17)
𝐼(𝑗𝜔) = 𝐼1 (𝑗𝜔) + 𝐼2 (𝑗𝜔) + ⋯ + 𝐼𝑘 (𝑗𝜔)
𝑉1 (𝑗𝜔) 𝑉2 (𝑗𝜔) 𝑉𝑘 (𝑗𝜔)
= + + ⋯+
𝑍1 (𝑗𝜔) 𝑍2 (𝑗𝜔) 𝑍𝑘 (𝑗𝜔)
𝑉(𝑗𝜔) 𝑉(𝑗𝜔) 𝑉(𝑗𝜔)
= + + ⋯+ .
𝑍1 (𝑗𝜔) 𝑍2 (𝑗𝜔) 𝑍𝑘 (𝑗𝜔)

Therefore, the equivalent impedance of impedances in parallel can then be defined as

7.18)
𝑉(𝑗𝜔)
𝑍𝑒𝑞 (𝑗𝜔) =
𝐼(𝑗𝜔)
𝑉(𝑗𝜔)
=
𝑉(𝑗𝜔) 𝑉(𝑗𝜔) 𝑉(𝑗𝜔)
+ + ⋯+
𝑍1 (𝑗𝜔) 𝑍2 (𝑗𝜔) 𝑍𝑘 (𝑗𝜔)
1 1
= = .
1 1 1 𝑘 1
+ + ⋯+ Σn=1
𝑍1 (𝑗𝜔) 𝑍2 (𝑗𝜔) 𝑍𝑘 (𝑗𝜔) 𝑍𝑛 (𝑗𝜔)

The equivalent admittance can be express as

7.19)
𝑘

𝑌𝑒𝑞 = ∑ 𝑌𝑛 .
n=1

Figure 7-5 Impedances in parallel.

As for resistors, the operator “∥” is also used for impedances in parallel. For example if 𝑍𝑎 and 𝑍𝑏 are
in parallel, the equivalent impedance is
7.20)
𝑍𝑒𝑞 (𝑗𝜔) = 𝑍𝑎 (𝑗𝜔) ∥ 𝑍𝑏 (𝑗𝜔)

101
 1
=
1 1
+
𝑍𝑎 (𝑗𝜔) 𝑍𝑏 (𝑗𝜔)
𝑍𝑎 (𝑗𝜔)𝑍𝑏 (𝑗𝜔)
= .
𝑍𝑎 (𝑗𝜔) + 𝑍𝑏 (𝑗𝜔)

7.3.3 E QUIVALENT I MPEDANCE

Similar to resistances, the simplest way to find the equivalent circuit of an impedance network is to
use series and parallel. However, not all circuit can be applied. A general method for finding the
equivalent impedance is to use a test source as in Figure 2-7. Either a current source or a voltage
source can be used and the equivalent impedance can be calculated as
7.21)
𝑉𝑡𝑒𝑠𝑡 (𝑗𝜔)
𝑍𝑒𝑞 (𝑗𝜔) = ,
𝐼𝑡𝑒𝑠𝑡 (𝑗𝜔)

which is the same as calculating equivalent resistances. In fact, superposition, voltage divider, current
divider, and Δ-to-Y conversion can all be applied to impedance calculation.

(a) Usage of test voltage (b) Usage of test current

Figure 7-6 Calculation of equivalent impedance using test sources.

7.4 SINUSOIDAL STEADY-STATE ANALYSIS USING IMPEDANCE AND TRANSFER FUNCTION

In the previous chapter, phasor transform were used for analyzing an RC circuit under sinusoidal
steady state. Using impedance for circuit analysis, the whole task can be significantly simplified. The
capacitor voltage in the frequency domain is the result of voltage division of the input sinusoidal
signal and can be written as
7.22)
1
𝑗𝜔𝐶
𝑉𝐶 (𝑗𝜔) = 𝑉𝐼 (𝑗𝜔) .
1
𝑅 + 𝑗𝜔𝐶

We define
7.23)

102
 1
𝑉𝐶 (𝑗𝜔) 𝑗𝜔𝐶 1
𝐻(𝑗𝜔) = = = ,
𝑉𝐼 (𝑗𝜔) 𝑅 + 1 1 + 𝑗𝜔𝑅𝐶
𝑗𝜔𝐶

where 𝐻(𝑗𝜔) is the transfer function of the circuit. The frequency-domain capacitor voltage can be
expressed as
7.24)
𝑉𝐶 (𝑗𝜔) = 𝐻(𝑗𝜔)𝑉𝐼 (𝑗𝜔) = |𝐻(𝑗𝜔)|𝑒 𝑗∠𝐻(𝑗𝜔) 𝑉𝐼 (𝑗𝜔).

The time-domain capacitor voltage can then be found by performing inverse phasor transform on
𝑉𝐶 (𝑗𝜔) such that
7.25)
𝑣𝐶 (𝑡) = 𝒫 −1 {𝑉𝐶 (𝑗𝜔)} = 𝒫 −1 {|𝐻(𝑗𝜔)|𝑒 𝑗∠𝐻(𝑗𝜔) 𝑉𝐼 (𝑗𝜔)}
= |𝐻(𝑗𝜔)|𝒫 −1 {𝑒 𝑗∠𝐻(𝑗𝜔) 𝑉𝐼 (𝑗𝜔)}
= |𝐻(𝑗𝜔)|𝑒 𝑗∠𝐻(𝑗𝜔0 ) 𝑣𝑖 (𝑡)
= |𝐻(𝑗𝜔)|𝑉𝑆 sin(𝜔𝑡 + 𝜙 + ∠𝐻(𝑗𝜔)).

The output voltage is the input voltage multiplied by a complex gain. The magnitude of the complex
gain is the magnitude of the transfer function and the phase shift relative to the input signal is the
argument of the transfer function. If the input is purely sinusoidal such as 𝑣𝑖 (𝑡) = 𝑉𝑆 sin(𝜔𝑡 + 𝜙), the
capacitor voltage can be written as
7.26)
𝑉𝑆
𝑣𝐶 (𝑡) = sin(𝜔𝑡 + 𝜙 − 𝜃),
√1 + 𝜔 2 𝑅2 𝐶 2

where
7.27)
𝜃 = −∠𝐻(𝑗𝜔) = tan−1 𝜔𝑅𝐶,

which is the same result as Eq. 6.32. It is apparent that using impedance and transfer function for
sinusoidal steady-state calculation is significantly simpler than solving the differential equation either
directly or using Laplace transform.

To demonstrate the use of transfer function when solving a circuit with a more complex input, let’s
assume a square wave is used as the input to the circuit in Figure 6-3. Assume the square has a
frequency of 𝜔 and an amplitude of 𝑉𝑆 , which can be expressed in Fourier series as
7.28)
∞
4𝑉𝑆 sin(𝑛𝜔𝑡)
𝑣𝑖 (𝑡) = ∑ .
𝜋 𝑛
𝑛=1,3,5…

Using the results in 7.25, the output capacitor voltage can be expressed as

103
7.29)
∞
4𝑉𝑆 1
𝑣𝐶 (𝑡) = ∑ |𝐻(𝑗𝑛𝜔)|𝑉𝑆 sin(𝑛𝜔𝑡 + ∠𝐻(𝑗𝑛𝜔))
𝜋 𝑛
𝑛=1,3,5…
∞
4𝑉𝑆 1 𝑉𝑆
= ∑ sin(𝑛𝜔𝑡 − tan−1 𝑛𝜔𝑅𝐶).
𝜋 𝑛 √1 + 𝑛2 𝜔 2 𝑅2 𝐶 2
𝑛=1,3,5…

It can be seen that the different frequency components of the square wave have different gains and
phase shifts. The higher frequency components have smaller gains and larger phase shifts. The
variation in gain and phase shift is called frequency response that will be discussed in detail in the
next chapter. In 7.29, the lowest frequency component of the Fourier series where n=1 is called the
main tone of a periodic signal and components where n>1 are called the harmonics.

To better understand the previously discussed material a sinusoidal signal with an amplitude of 1 and
𝜔 = 4 is plotted in the time domain and frequency domain below.

Figure 7-7 Time-domain and Frequency-domain plots of a sinusoidal signal.

Similarly, a square-wave signal with an amplitude of 1 and 𝜔 = 4 is plotted in the time domain and
frequency domain below. It can be seen that a square wave consists of a main tone with harmonics
that become smaller as the frequency increases.

104
 Figure 7-8 Time-domain and Frequency-domain plots of a square-wave signal.

Assuming 𝑅𝐶 = 0.05, the input square wave (blue) and the output capacitor voltage (red) are
compared in the time domain and frequency domain below. The frequency response of the RC circuit
is plotted in green for comparison.

Figure 7-9 Time-domain and Frequency-domain plots of a square-wave signal as the input to an RC circuit .

105
 CHAPTER 8 FREQUENCY-SELECTIVE CIRCUITS

Frequency-selective circuits are also called filters. These circuits have the ability to let certain
frequency components pass through and attenuate or block other unwanted frequency components.
The frequency band that can pass through a filter is called the passband. The frequency band that is
blocked or attenuated is called the stopband. Depending on the circuit elements that are used to
construct the filters, filters can be categorized into passive filters and active filters. A passive filter uses
only passive components such as R, L, and C. The gain of a passive filter is always less than or equal to
one. An active filter uses components such as operational amplifiers that can provide gain in addition
to its frequency-selection characteristics.

8.1 FILTER BASICS

Depending on the frequency response, filters are normally categorized into low-pass filters, high-pass
filters, band-pass filters, and band-stop filters. Frequency response is the measure of transfer function
evaluated with respect to frequency or angular frequency (𝜔). Frequency response is normally
expressed in the form of a Bode plot, where the magnitude and phase are plotted with respect to
frequency in log scale. The magnitude is normally expressed in decibel (dB) defined as
8.1)
𝐴𝑑𝐵 = 20 ∙ log(𝐴) (dB).

8.1.1 L OW -P ASS F ILTERS

Low-pass filters (LPFs) are filters that allow low-frequency components to pass through and block
high-frequency components. The magnitude response of an ideal low-pass filter is shown in Figure
8-1(a). The transition frequency between passband and stopband is called the cutoff frequency (𝜔𝑐 ).
The sharp transition between stopband and passband in the ideal low-pass filter is impossible to
implement in real life. A real low-pass filter would have a magnitude response like the one shown in
Figure 8-1(b), where the transition is more gradual and the passband may have ripples. The cutoff
frequency is normally defined as the frequency where the magnitude is 3-dB lower than the maximum
magnitude and thus often denoted as 𝜔3𝑑𝐵 and called the 3-dB cutoff frequency.

(a) Ideal low-pass filter (b) Realistic low-pass filter

Figure 8-1 Example magnitude responses of low-pass filters.

8.1.2 H IGH -P ASS F ILTERS

106
High-pass filters (HPFs) are filters that are the inverse of low-pass filters, which allow high-frequency
components to pass through and block low-frequency components. The magnitude response of an
ideal high-pass filter is shown in Figure 8-2(a). A real high-pass filter would have a magnitude
response like the one shown in Figure 8-2(b).

(a) Ideal high-pass filter (b) Realistic high-pass filter

Figure 8-2 Example magnitude responses of high-pass filters.

8.1.3 B AND -P ASS F ILTERS

Band-pass filters (BPFs) are filters that allow a specific finite frequency range to be passed while
blocking both low-frequency and high-frequency components. The magnitude response of an ideal
band-pass filter is shown in Figure 8-3(a). There are two transition frequencies between passband
and stopband 𝜔𝑐1 and 𝜔𝑐2 . A real band-pass filter would have a magnitude response like the one
shown in Figure 8-3(b), where the transition is more gradual and the passband may have ripples. The
cutoff frequencies are normally defined as the frequencies where the magnitude is 3-dB lower than
the maximum.

(a) Ideal band-pass filter (b) Realistic band-pass filter

Figure 8-3 Example magnitude responses of low-pass filters.

8.1.4 B AND -S TOP F ILTERS

Band-stop filters (BSFs), sometimes called band-reject filters, are filters that are the inverse of band-
pass filters. Band-stop filters block a specific finite frequency range while allowing both low-frequency
and high-frequency components to pass through. The magnitude response of an ideal band-stop filter
is shown in Figure 8-4(a). A real band-stop filter would have a magnitude response like the one shown
in Figure 8-4(b), where the transition is more gradual and the passband may have ripples.

107
 (a) Ideal band-stop filter (b) Realistic band-stop filter
Figure 8-4 Example magnitude responses of band-stop filters.

8.2 PASSIVE FILTERS

Passive filters are constructed by only passive elements such as R, L, and C. Whenever possible,
inductors are replaced by capacitors because capacitors are cheaper and closer to ideal ones.
Inductors are made of windings. The higher the inductance, the more the windings there are and thus
higher internal resistance and capacitance, which lead to more loss and lower operable frequencies.
When analyzing filters, frequency-domain analysis using the transfer function is usually utilized. The
filter transfer function 𝐻(𝑗𝜔) is only defined for the 𝑗𝜔 axis on the 𝑠-plane. However, 𝐻(𝑠) is often
used by assuming 𝑠 = 𝑗𝜔 and will be used in this chapter.

8.2.1 F IRST -O RDER L OW -P ASS F ILTERS

First-order low-pass filters can be in the form of RC or RL. An RC low-pass filter is shown in Figure
8-5(a).The transfer function can be easily calculated as
8.2)
1
𝑉𝑜𝑢𝑡 (𝑠) 𝑠𝐶 1
𝐻(𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 1
𝑅 + 𝑠𝐶 1 + 𝑠𝑅𝐶

For low frequencies where 𝑠 → 0, 𝐻(𝑠) → 1. For high-frequencies, 𝑠 → 𝑗∞, 𝐻(𝑠) → 0. The circuit is
indeed a low-pass filter. To find the 3-dB cutoff frequency,
8.3)
1 1
|𝐻(𝑗𝜔3𝑑𝐵 )| = | | = 20 log(1) 𝑑𝐵 − 3𝑑𝐵 = −3𝑑𝐵 = .
1 + 𝑗𝜔3𝑑𝐵 𝑅𝐶 √2

Thus,
8.4)
𝜔3𝑑𝐵 𝑅𝐶 = 1

and
8.5)
1 1
𝜔3𝑑𝐵 = 𝜔𝑐 = = .
𝑅𝐶 𝜏

108
Similarly, for the RL low-pass filter shown in Figure 8-5(b). The transfer function can be easily
calculated as
8.6)
𝑉𝑜𝑢𝑡 (𝑠) 𝑅 1
𝐻(𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 1 + 𝑠𝐿/𝑅

(a) RC-low-pass filter (b) RL low-pass filter

Figure 8-5 First-order passive low-pass filters.

The 3-dB cutoff frequency can be found to be

8.7)
𝑅 1
𝜔3𝑑𝐵 = 𝜔𝑐 = = .
𝐿 𝜏

The transfer function of the first-order low-pass filter can be expressed in the generic form of
8.8)
1
𝐻(𝑠) = .
1 + 𝑠/𝜔𝑐

The magnitude response can be calculated as

8.9)

1 1
|𝐻(𝑗𝜔)| = | |= .
𝑗𝜔 𝜔 2
1+𝜔 √1 + ( )
𝑐 𝜔𝑐

The phase response can be calculated as

8.10)

1 𝑗𝜔 𝜔
𝜙(𝜔) = Arg(𝐻(𝑗𝜔)) = Arg ( ) = Arg(1) − Arg (1 + ) = − tan−1 ( ).
𝑗𝜔 𝜔𝑐 𝜔𝑐
1+𝜔
𝑐

The transfer function has a pole located at

8.11)
𝑝 = −𝜔𝑐 .

Assume a transfer function can be expressed as

109
8.12)
𝑁(𝑠)
𝐻(𝑠) = ,
𝐷(𝑠)

where both the numerator 𝑁(𝑠) and the denominator 𝐷(𝑠) can both be expressed as polynomial
functions of 𝑠. The pole is defined as the solution to 𝐷(𝑠) = 0. Similarly, zero is defined as the solution
to 𝑁(𝑠) = 0. In this case, there is no zero in the first-order low-pass filter. Figure 8-6 shows the
frequency response or Bode plot of the first-order low-pass filter. The Bode plot is basically plotting
the magnitude and phase responses separately versus frequency. The magnitude is normally plotted
in dB and the frequency is normally plotted in log scale.

Figure 8-6 Frequency response of a first-order low-pass filter.

8.2.2 F IRST -O RDER H IGH -P ASS F ILTERS

First-order high-pass filters can be in the form of RC or RL. An RC high-pass filter is shown in Figure
8-7(a) The transfer function can be easily calculated as
8.13)
𝑉𝑜𝑢𝑡 (𝑠) 𝑅 𝑠𝑅𝐶
𝐻(𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 1
𝑅 + 𝑠𝐶 1 + 𝑠𝑅𝐶

For low frequencies where 𝑠 → 0, 𝐻(𝑠) → 0. For high-frequencies, 𝑠 → 𝑗∞, 𝐻(𝑠) → 1. The circuit is
indeed a high-pass filter. To find the 3-dB cutoff frequency,
8.14)
𝑗𝜔3𝑑𝐵 𝑅𝐶 1
|𝐻(𝑗𝜔3𝑑𝐵 )| = | | = −3𝑑𝐵 = .
1 + 𝑗𝜔3𝑑𝐵 𝑅𝐶 √2

Thus,
8.15)
𝜔3𝑑𝐵 𝑅𝐶 = 1

and

110
8.16)
1 1
𝜔3𝑑𝐵 = 𝜔𝑐 = = .
𝑅𝐶 𝜏

Similarly, for the RL high-pass filter shown in Figure 8-7(b). The transfer function can be easily
calculated as
8.17)
𝑉𝑜𝑢𝑡 (𝑠) 𝑠𝐿 𝑠𝐿/𝑅
𝐻(𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 1 + 𝑠𝐿/𝑅

(a) RC-high-pass filter (b) RL high-pass filter

Figure 8-7 First-order passive high-pass filters.

The 3-dB cutoff frequency can be found to be

8.18)
𝑅 1
𝜔3𝑑𝐵 = 𝜔𝑐 = = .
𝐿 𝜏

The transfer function of the first-order high-pass filter can be expressed in the generic form of
8.19)
𝑠/𝜔𝑐
𝐻(𝑠) = .
1 + 𝑠/𝜔𝑐

The magnitude response can be calculated as

8.20)
𝑗𝜔 𝜔
𝜔𝑐 𝜔𝑐
|𝐻(𝑗𝜔)| = | |= .
𝑗𝜔 𝜔 2
1+𝜔 √1 + ( )
𝑐 𝜔 𝑐

The phase response can be calculated as

8.21)
𝑗𝜔
𝜔𝑐 𝜋 𝑗𝜔 𝜋 𝜔
𝜙(𝜔) = Arg(𝐻(𝑗𝜔)) = Arg ( ) = − Arg (1 + ) = − tan−1 ( ).
𝑗𝜔 2 𝜔𝑐 2 𝜔𝑐
1+𝜔
𝑐

111
The transfer function has a pole located at
8.22)
𝑝 = −𝜔𝑐

and a zero located at

8.23)
𝑧 = 0.

Figure 8-8 shows the frequency response or Bode plot of the first-order high-pass filter.

Figure 8-8 Frequency response of a first-order high-pass filter.

8.2.3 S ECOND -O RDER RLC F ILTERS

The RLC series resonator is shown in Figure 8-9. This series resonator can be a low-pass filter, high-
pass filter, band-pass filter, or a band-stop filter depending how the output is configured. The low-pass
output is the output across the capacitor 𝐶 such that
8.24)
1 1
𝑉𝑙𝑝 (𝑠) 𝑠𝐶 𝐿𝐶
𝐻𝑙𝑝 (𝑠) = = = ,
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 + 1 𝑠 2 +𝑠𝑅+ 1
𝑠𝐶 𝐿 𝐿𝐶

which makes it a second-order low-pass filter. The high-pass output is the output across the inductor
𝐿 such that
8.25)
𝑉ℎ𝑝 (𝑠) 𝑠𝐿 𝑠2
𝐻ℎ𝑝 (𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 + 1 𝑅
𝑠2 + 𝑠 +
1
𝑠𝐶 𝐿 𝐿𝐶

which makes it a second-order high-pass filter. The second-order low-pass filter and high-pass filter
are basic building blocks of high-order filters and will be discussed in detail in 8.4.

112
 Figure 8-9 RLC series resonator.

Unlike low-pass and high-pass filters that can be implemented with first-order circuits, band-pass and
band-stop filters must be implemented with at least a second-order circuit. Actually, implementations
of band-pass and band-stop filters must be of even orders, which will also be discussed in 8.4. The
band-pass output is the output across the capacitor 𝑅 such that
8.26)
𝑅
𝑉𝑏𝑝 (𝑠) 𝑅 𝑠𝐿
𝐻𝑏𝑝 (𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 + 1 𝑅
𝑠2 + 𝑠 +
1
𝑠𝐶 𝐿 𝐿𝐶

The center frequency is defined as the frequency |𝐻𝑏𝑝 (𝑠)| is at its maximum, which can be found by
differentiating |𝐻𝑏𝑝 (𝑠)| such that

8.27)
𝑅
𝜕 𝜕 𝑠
|𝐻 (𝑠)| = | 𝐿 |
𝜕𝜔 𝑏𝑝 𝜕𝜔 𝑠 2 + 𝑠 𝑅 + 1
𝐿 𝐿𝐶
𝑅
𝜕 𝜔𝐿
=
𝜕𝜔 | 1 − 𝜔 2 + 𝑗𝜔 𝑅 |
𝐿𝐶 𝐿
𝑅
𝜕 𝜔𝐿
=
𝜕𝜔 2 2
√( 1 − 𝜔 2 ) + (𝜔 𝑅 )
𝐿𝐶 𝐿
1
2 2 2 2 −
𝑅√ 1 2 ) + (𝜔 𝑅 ) − 𝜔𝑅 [( 1 − 𝜔 2 ) + (𝜔 𝑅 ) ]
2 𝑅2 2
(
𝐿 𝐿𝐶 − 𝜔 𝐿 2𝐿 𝐿𝐶 𝐿 [4𝜔3 + 2 ( 2 − 𝐿𝐶 ) 𝜔]
𝐿
= 2 = 0.
1 𝑅 2
( − 𝜔 2 ) + (𝜔 )
𝐿𝐶 𝐿

Assume the denominator is not zero,

8.28)
1
2 2 −
𝑅 1 𝑅 2 𝜔𝑅 1 𝑅 2 2 𝑅2 2
√( − 𝜔 2 ) + (𝜔 ) = 2
[( − 𝜔 ) + (𝜔 ) ] [4𝜔3 + 2 ( − ) 𝜔]
𝐿 𝐿𝐶 𝐿 2𝐿 𝐿𝐶 𝐿 𝐿2 𝐿𝐶

113
 2
1 𝑅 2 𝜔 𝑅2 2
⇒ ( − 𝜔2 ) + (𝜔 ) = [4𝜔3 + 2 ( 2 − ) 𝜔]
𝐿𝐶 𝐿 2 𝐿 𝐿𝐶

2
1 𝑅 2 𝑅2 2
⇒ ( − 𝜔2 ) + (𝜔 ) = 2𝜔4 + ( 2 − ) 𝜔2
𝐿𝐶 𝐿 𝐿 𝐿𝐶
1 2
⇒ 𝜔4 = ( )
𝐿𝐶
1
⇒𝜔= .
√𝐿𝐶

The center frequency is usually denoted as 𝜔0 . At the center frequency,

8.29)
𝐻𝑏𝑝 (𝑠) = 𝐻𝑏𝑝 (𝑗𝜔0 ) = 1.

Therefore, the 3-dB cutoff frequencies can be found at

8.30)
𝑅 𝑅
1 𝑗𝜔3𝑑𝐵 𝜔3𝑑𝐵
|𝐻𝑏𝑝 (𝑗𝜔3𝑑𝐵 )| = =| 𝐿 |= 𝐿
√2 1 𝑅 2 2
√( 1 − 𝜔3𝑑𝐵 2 ) + (𝜔3𝑑𝐵 𝑅 )
2
𝐿𝐶 − 𝜔3𝑑𝐵 + 𝑗𝜔3𝑑𝐵 𝐿 𝐿𝐶 𝐿
2
1 𝑅 2
⇒ ( − 𝜔3𝑑𝐵 2 ) = (𝜔3𝑑𝐵 )
𝐿𝐶 𝐿
𝑅 1
⇒ 𝜔3𝑑𝐵 2 ± 𝜔3𝑑𝐵 − =0
𝐿 𝐿𝐶
𝑅 𝑅2 4 𝑅 𝑅2 4
∓ 𝐿 ± √ 2 + 𝐿𝐶 ∓ 𝐿 + √ 2 + 𝐿𝐶
𝐿 𝐿
⇒ 𝜔3𝑑𝐵 = = ,
2 2

since 𝜔3𝑑𝐵 > 0. For a band-pass filter , there are two 3-dB cutoff frequencies. The bandwidth of a
bandpass filter is defined as
8.31)
𝑅
𝐵𝑊 = 𝛽 = 𝜔3𝑑𝐵,𝑢𝑝𝑝𝑒𝑟 − 𝜔3𝑑𝐵,𝑙𝑜𝑤𝑒𝑟 = .
𝐿

Therefore, the transfer function can be rewritten in to

8.32)
𝑉𝑏𝑝 (𝑠) 𝛽𝑠
𝐻𝑏𝑝 (𝑠) = = 2 .
𝑉𝑖𝑛 (𝑠) 𝑠 + 𝛽𝑠 + 𝜔02

By comparing the coefficients of the transfer function, the bandwidth and center frequency can be
found directly. The quality factor of a band-pass filter is defined as
8.33)

114
 𝜔0
𝑄= .
𝛽

Therefore, the transfer function of a band-pass filter can also be expressed as

8.34)
𝜔0
𝑉𝑏𝑝 (𝑠) 𝑄 𝑠
𝐻𝑏𝑝 (𝑠) = = .
𝑉𝑖𝑛 (𝑠) 𝑠 2 + 𝜔0 𝑠 + 𝜔2
𝑄 0

The band-stop output is the output across the capacitor 𝐶 and inductor 𝐿 such that
8.35)
1 1
𝑉𝑏𝑠 (𝑠) 𝑠𝐿 + 𝑠𝐶 𝑠 2 + 𝐿𝐶
𝐻𝑏𝑠 (𝑠) = = = .
𝑉𝑖𝑛 (𝑠) 𝑅 + 𝑠𝐿 + 1 𝑠 2 +𝑠𝑅 + 1
𝑠𝐶 𝐿 𝐿𝐶

The center frequency, cut-off frequencies, and bandwidth calculations are similar to the band-pass
filter and are omitted here. The transfer function of the band-stop filter can be expressed as
8.36)
𝑉𝑏𝑠 (𝑠) 𝑠 2 + 𝜔02 𝑠 2 + 𝜔02
𝐻𝑏𝑠 (𝑠) = = 2 = .
𝑉𝑖𝑛 (𝑠) 𝑠 + 𝛽𝑠 + 𝜔02 𝑠 2 + 𝜔0 𝑠 + 𝜔2
𝑄 0

EX. 8.1
Please find the transfer function of the following second-order passive low-pass filter.

1
𝑉𝑜𝑢𝑡 (𝑠) 𝑅 ∥ 𝑠𝐶
𝐻(𝑠) = =
𝑉𝑖𝑛 (𝑠) 1
𝑠𝐿 + 𝑅 ∥ 𝑠𝐶
𝑅/𝑠𝐶
𝑅 + 1/𝑠𝐶
=
𝑅/𝑠𝐶
𝑠𝐿 +
𝑅 + 1/𝑠𝐶
𝑅
= 1 + 𝑠𝑅𝐶
𝑅
𝑠𝐿 +
1 + 𝑠𝑅𝐶
1/𝐿𝐶
= 2
𝑠 + 𝑠/𝑅𝐶 + 1/𝐿𝐶
In a second-order band-pass filter, 𝑄 is the ratio between the center frequency and bandwidth.
115
Whereas, a larger 𝑄 in a second-order low-pass filter leads to a larger peaking in the frequency
response. The peaking mechanism can be seen clearly in Figure 8-10. In the previous example,
increasing 𝑅 can lead to a larger 𝑄. The peaking can also be explained by the location of the poles in
the transfer function. By increasing 𝑅 or 𝑄, the pair of poles moves closer to the 𝑗𝜔–axis. When 𝑅
approaches infinity, the poles would lie on the 𝑗𝜔–axis and the gain would approach infinity at 𝜔 =
1/√𝐿𝐶.

Figure 8-10 Second-order low-pass filter responses under various quality factors.

8.3 ACTIVE FILTERS

The outputs of the passive filters introduced previously are always smaller than the inputs. Active
filters have the benefit of not only frequency selectivity, but also provide gain if necessary. Another
benefit of active filters is that second-order and high-order filters can be implemented without the use
of inductors. The active filters shown in this chapter will use only capacitors. Even though it is
possible to use inductors to implement active filters using the methods discussed in this chapter, it
makes no sense cost-wise and performance-wise.

8.3.1 F IRST -O RDER A CTIVE L OW -P ASS F ILTERS

The first-order active low-pass filter is shown in Figure 8-11. The transfer function can be written as
8.37)
𝑉𝑜𝑢𝑡 (𝑠)
𝐻𝑙𝑝 (𝑠) =
𝑉𝑖𝑛 (𝑠)
1
𝑅2 ∥ (𝑠𝐶 )
=−
𝑅1
𝑅2
=− 𝑠𝐶
1
𝑅1 (𝑅2 + )
𝑠𝐶

116
 𝑅2 1
= (− )( ).
𝑅1 1 + 𝑠𝑅2 𝐶

Similar to the passive RC filter, this filter has a 3-dB cutoff frequency of
8.38)
1
𝜔3𝑑𝐵 = .
𝑅2 𝐶

However, the active filter provides a gain of

8.39)
𝑅2
𝐴=− .
𝑅1

Figure 8-11 First-order active low-pass filter.

8.3.2 F IRST -O RDER A CTIVE H IGH -P ASS F ILTERS

The first-order active high-pass filter is shown in Figure 8-12. The transfer function can be written as
8.40)
𝑉𝑜𝑢𝑡 (𝑠) 𝑅2 𝑅2 𝑠𝑅1 𝐶
𝐻ℎ𝑝 (𝑠) = =− = (− ) ( ).
𝑉𝑖𝑛 (𝑠) 1 𝑅1 1 + 𝑠𝑅1 𝐶
𝑅1 + (𝑠𝐶 )

Similar to the passive RC filter, this filter has a 3-dB cutoff frequency of
8.41)
1
𝜔3𝑑𝐵 = .
𝑅1 𝐶

However, the active filter provides a gain of

8.42)
𝑅2
𝐴=− .
𝑅1

117
 Figure 8-12 First-order active high-pass filter.

8.3.3 S ECOND -O RDER A CTIVE L OW -P ASS F ILTERS

The inverting second-order active low-pass filter is shown in Figure 8-13. The transfer function can be
found to be
8.43)
1
𝑉𝑜𝑢𝑡 (𝑠) 𝑅1 𝑅2 𝐶1 𝐶2
𝐻𝑙𝑝 (𝑠) = =− .
𝑉𝑖𝑛 (𝑠) 2 1 1 1 1
𝑠 + 𝑠 (𝑅 𝐶 + 𝑅 𝐶 + 𝑅 𝐶 ) + 𝑅 𝑅 𝐶 𝐶
1 1 2 1 3 1 2 3 1 2

Figure 8-13 Inverting second-order active low-pass filter.

For convenience when design high-order filters with second-order low-pass filters, the second-order
low-pass filter also has a generic form similar to the second-order band-pass filter shown previously.
The generic second-order low-pass filter transfer function can be written as
8.44)
𝑉𝑜𝑢𝑡 (𝑠) 𝐴𝜔02
𝐻𝑙𝑝 (𝑠) = = 𝜔 ,
𝑉𝑖𝑛 (𝑠) 𝑠 2 + 𝑄0 𝑠 + 𝜔02

where 𝐴 is the gain of the low-pass filter. The gain of the previous low-pass filter can be found to be
8.45)

118
 1
𝑅1 𝑅2 𝐶1 𝐶2 𝑅3
𝐴 = 𝐻𝑙𝑝 (0) = − =− .
1 𝑅1
𝑅2 𝑅3 𝐶1 𝐶2

EX. 8.2
Please find the transfer function of the following second-order active low-pass filter.

Applying KCL at node a

𝑉𝑖𝑛 (𝑠) − 𝑉𝑎 (𝑠)
= 𝑠𝐶2 [𝑉𝑎 (𝑠) − 𝑉𝑜𝑢𝑡 (𝑠)]
𝑅1
⇒ 𝑉𝑖𝑛 (𝑠) − 𝑉𝑎 (𝑠) = 𝑠𝑅1 𝐶2 𝑉𝑎 (𝑠) − 𝑠𝑅1 𝐶2 𝑉𝑜𝑢𝑡 (𝑠)
𝑉𝑖𝑛 (𝑠) + 𝑠𝑅1 𝐶2 𝑉𝑜𝑢𝑡 (𝑠)
⇒ 𝑉𝑎 (𝑠) =
1 + 𝑠𝑅1 𝐶2
Applying KCL at node c
𝑉𝑏 (𝑠) − 𝑉𝑐 (𝑠)
= 𝑠𝐶1 𝑉𝑐 (𝑠)
𝑅2
⇒ 𝑉𝑏 (𝑠) = (1 + 𝑠𝑅2 𝐶1 )𝑉𝑐 (𝑠)
Because of virtual short
𝑉𝑎 (𝑠) = 𝑉𝑏 (𝑠)
and
𝑉𝑐 (𝑠) = 𝑉𝑜𝑢𝑡 (𝑠),
Thus
𝑉𝑖𝑛 (𝑠) + 𝑠𝑅1 𝐶2 𝑉𝑜𝑢𝑡 (𝑠)
= (1 + 𝑠𝑅2 𝐶1 )𝑉𝑜𝑢𝑡 (𝑠)
1 + 𝑠𝑅1 𝐶2
⇒ 𝑉𝑖𝑛 (𝑠) = (1 + 𝑠𝑅2 𝐶1 + 𝑠 2 𝑅1 𝑅2 𝐶1 𝐶2 )𝑉𝑜𝑢𝑡 (𝑠)
1
𝑉𝑜𝑢𝑡 (𝑠) 𝑅1 𝑅2 𝐶1 𝐶2
⇒ 𝐻(𝑠) = = .
𝑉𝑖𝑛 (𝑠) 𝑠 1
𝑠2 + +
𝑅1 𝐶2 𝑅1 𝑅2 𝐶1 𝐶2

8.3.4 S ECOND -O RDER A CTIVE H IGH -P ASS F ILTERS

The inverting second-order active high-pass filter is shown in Figure 8-14. The transfer function can
be found to be
8.46)
𝑉𝑜𝑢𝑡 (𝑠) 𝐺𝑠 2
𝐻ℎ𝑝 (𝑠) = =− .
𝑉𝑖𝑛 (𝑠) 2 2𝐺 + 1 𝐺
𝑠 +𝑠 𝑅 𝐶 +
2 𝑅1 𝑅2 𝐶 2
119
 Figure 8-14 Inverting second-order active high-pass filter.

For convenience when design high-order filters, the second-order high-pass filter also has a generic
form. The generic second-order high-pass filter transfer function can be written as
8.47)
𝑉𝑜𝑢𝑡 (𝑠) 𝐴𝑠 2
𝐻ℎ𝑝 (𝑠) = = 𝜔 ,
𝑉𝑖𝑛 (𝑠) 𝑠 2 + 0 𝑠 + 𝜔02
𝑄

where 𝐴 is the gain of the high-pass filter. The gain of the previous high-pass filter can be found to be
8.48)
𝐴 = 𝐻ℎ𝑝 (𝑗∞) = 𝐺.

8.3.5 S ECOND -O RDER A CTIVE B AND -P ASS F ILTERS

The inverting second-order active band-pass filter is shown in Figure 8-15. The transfer function can
be found to be
8.49)
1
𝑉𝑜𝑢𝑡 (𝑠) 𝑅3 𝐶 𝑠
𝐻𝑏𝑝 (𝑠) = =− .
𝑉𝑖𝑛 (𝑠) 2 1
𝑠2 + 𝑠 𝑅 𝐶 + 2
2 (𝑅1 ∥ 𝑅3 )𝑅2 𝐶

To include the effect of gain, the generic second-order band-pass filter transfer function can be
modified as
8.50)
𝜔
𝑉𝑜𝑢𝑡 (𝑠) 𝐴 𝑄0 𝑠
𝐻𝑏𝑝 (𝑠) = = 𝜔 ,
𝑉𝑖𝑛 (𝑠) 𝑠 2 + 𝑄0 𝑠 + 𝜔02

where 𝐴 is the gain of the band-pass filter.

120
 Figure 8-15 Inverting second-order active band-pass filter.

8.4 HIGH-ORDER FILTERS

High-order filters are designed with generic transfer functions or filter tables where the cutoff
frequency or center frequency is 1 rad/s. These filters are called normalized filters where the
resistance is also normalized to 1Ω. De-normalization is used to change the resistance and frequency
to the desired values. Assume a normalized filter is designed with component values of 𝑅𝑛 , 𝐿𝑛 , and 𝐶𝑛 ,
the de-normalized component values can be found to be
8.51)
𝑅𝑑 = 𝑅𝑠𝑐𝑎𝑙𝑒 𝑅𝑛
𝐶𝑛
𝐶𝑑 =
𝑅𝑠𝑐𝑎𝑙𝑒 𝜔𝑑 ,
𝑅𝑠𝑐𝑎𝑙𝑒 𝐿𝑛
𝐿𝑑 =
{ 𝜔𝑑

where 𝑅𝑠𝑐𝑎𝑙𝑒 is the resistance scaling factor and 𝜔𝑑 is the desired design frequency. The poles and
zeros after de-normalizing are
8.52)
𝑝𝑑 = 𝜔𝑑 𝑝𝑛
{𝑧 = 𝜔 𝑧 .
𝑑 𝑑 𝑛

8.4.1 B UTTERWORTH L OW -P ASS F ILTERS

The Butterworth filter is the easiest to design high-order filter, because all the poles lie on the unit
circle. The Butterworth filter is completely flat in the passband, which also gives it a name of
“maximally-flat filter”. The pole locations for an 𝑁-th order normalized Butterworth low-pass filter are
at
8.53)
(2𝑘 − 1)𝜋 (2𝑘 − 1)𝜋
𝑝𝑘𝑛 = − sin [ ] + 𝑗 cos [ ],
2𝑁 2𝑁

where 𝑘 = 1 … 𝑁. The poles are evenly spaced on the unit circuit of the left half of the 𝑠-plane. The
121
transfer function can be easily found as
8.54)
𝑁
1
𝐻𝑙𝑝𝑛 (𝑠𝑛 ) = ∏ .
𝑠𝑛 − 𝑝𝑘𝑛
𝑘=1

The magnitude response can be found as

8.55)
1
|𝐻𝑙𝑝𝑛 (𝑗𝜔𝑛 )| = .
√1 + 𝜔𝑛2𝑁

Example magnitude responses of various filter orders are compared in Figure 8-16.

Figure 8-16 Normalized Butterworth low-pass filter responses with various orders.

Using a 1-MHz second-order Butterworth filter with gain of 2 as an example, the pole locations are at
8.56)
√2 √2
𝑝1𝑛 , 𝑝2𝑛 = − ±𝑗 ,
2 2

thus the transfer function can be found as

8.57)
1 1
𝐻𝑙𝑝𝑛 (𝑠𝑛 ) = = 2 .
(𝑠𝑛 − 𝑝1𝑛 )(𝑠𝑛 − 𝑝2𝑛 ) 𝑠𝑛 + √2𝑠𝑛 + 1

Using the second-order low-pass filter in Figure 8-13, the gain can be calculated as
8.58)
1
𝑅1 𝑅2 𝐶1 𝐶2 𝑅3
𝐴=− =− = −2.
1 𝑅1
𝑅2 𝑅3 𝐶1 𝐶2

122
The gain is specified to be 2, if polarity is important, a non-inverting architecture should be used or an
inverting amplifier can be cascaded. The cutoff frequency for the normalized low pass filter is 1 rad/s
and thus
8.59)
1 = 𝑅2 𝑅3 𝐶1 𝐶2

and
8.60)
𝜔0𝑛 1 1 1 1
= = √2 = ( + + ).
𝑄 𝑄 𝑅1 𝐶1 𝑅2 𝐶1 𝑅3 𝐶1

From above, there are five unknowns and only three equations, which suggest the number of possible
solutions are unlimited. For simplicity, we can set 𝑅1 = 1 Ω, 𝑅2 = 2 Ω, and 𝑅3 = 2 Ω. 𝐶1 can then be
calculated from Eq. 8.60 as √2 F. From Eq. 8.59, 𝐶2 can be calculated as 1/4√2 F. Since the de-
normalized resistance is not specified, an arbitrary scaling factor can be used. A scaling factor of 1000
is used here. The frequency scaling factor 𝜔𝑑 = 2𝜋 × 1 MHz = 2𝜋 × 106 . Finally, the de-normalized
component values can be found as
8.61)
𝑅1𝑑 = 1 Ω × 1000 = 1 kΩ
{𝑅2𝑑 = 2 Ω × 1000 = 2 kΩ
𝑅3𝑑 = 2 Ω × 1000 = 2 kΩ

and
8.62)

√2
𝐶1𝑑 = ≈ 225 pF
1000 × 2𝜋 × 106 .
1/4√2
𝐶2𝑑 = ≈ 28.1 pF
{ 1000 × 2𝜋 × 106

8.4.2 C HEBYSHEV L OW -P ASS F ILTERS

The Chebyshev filter has ripples in the passband. However, it has a very steep transition region
between the passband and stopband making its frequency selectivity great. The pole locations for an
𝑁-th order normalized Chebyshev low-pass filter are at
8.63)
(2𝑘 − 1)𝜋 1 1 (2𝑘 − 1)𝜋 1 1
𝑝𝑘𝑛 = − sin [ ] sinh [ 𝑠𝑖𝑛ℎ−1 ( )] + 𝑗 cos [ ] cosh [ 𝑠𝑖𝑛ℎ−1 ( )] ,
2𝑁 𝑁 𝜖 2𝑁 𝑁 𝜖

where 𝑘 = 1 … 𝑁 and 𝜖 is the ripple factor. The magnitude response can be found as
8.64)
1
|𝐻𝑙𝑝𝑛 (𝑗𝜔𝑛 )| =
√1 + 𝜖 2 cos2[𝑁 cos−1(𝜔𝑛 )]

123
for 𝜔𝑛 ≤ 1 and
8.65)
1
|𝐻𝑙𝑝𝑛 (𝑗𝜔𝑛 )| =
√1 + 𝜖 2 cosh2[𝑁 cosh−1(𝜔𝑛 )]

for 𝜔𝑛 > 1. The ripple is 3 dB when 𝜖 = 1 and 1 dB when 𝜖 = 0.509. Example magnitude responses of
various filter orders with 1-dB ripples are compared in Figure 8-17.

Figure 8-17 Normalized Chebyshev low-pass filter responses with various orders.

8.4.3 B ESSEL L OW -P ASS F ILTERS

The magnitude response of the Bessel filter is terrible. Not only does it have a gradually decreasing
pass band, but also a long transition region. The main benefit of using a Bessel filter is its superior
phase response. The phase response of the Bessel filter is linear with frequency and thus it is also
named the “linear-phase filter”. The delay of a signal can be calculated as
8.66)
𝑑𝜙(𝜔)
𝜏(𝜔) = − .
𝑑𝜔

For a linear phase filter, the delay is constant. However, for other filters the delay is frequency
dependent, which creates a phenomenon called dispersion. For wide bandwidth communication
applications, it is important for all of the information to arrive at the same time, and thus the Bessel
filter is often used. With the advancement of digital technology, finite-impulse response (FIR) digital
filters, which are linear in phase, are usually used now.

The transfer function of a low-pass Bessel filter is defined as

8.67)
𝐵𝑁 (0)
𝐻𝑙𝑝 (𝑠𝑛 ) = ,
𝐵𝑁 (𝑠𝑛 )

124
where
8.68)
𝐵0 (𝑠) = 1
{ 𝐵1 (𝑠) = 1 + 𝑠 .
2
𝐵𝑚 (𝑠) = (2𝑚 − 1)𝐵𝑚−1 (𝑠) + 𝑠 𝐵𝑚−2 (𝑠)

Unlike Butterworth and Chebyshev filters, the transfer function of Bessel filters is not normalized to
1 rad/s and thus additional scaling is necessary.

(a) Un-normalized Bessel low-pass filter (b) Normalized Bessel low-pass filter
Figure 8-18 Magnitude response of various order Bessel low-pass filters.

Figure 8-19 Phase response of various order Bessel low-pass filter.

The magnitude responses of 5th-order Butterworth, Chebyshev, Bessel filter along with five first-order
RC filter cascaded are compared in Figure 8-20. It can be clearly seen that the Chebyshev filter has the
best frequency selectivity and the cascaded RC filter with five real poles have the worst. Filter
performance can be improved by placing poles as conjugate pairs in the left half of the s-plane.

125
 Figure 8-20 Magnitude response comparison of various 5 th -order low-pass filters.

8.4.4 H IGH -O RDER H IGH -P ASS AND B AND -P ASS F ILTERS

In this chapter, a lot of effort was put into the design of low-pass filters. The main reason is that high-
pass and band-pass filters can be designed by performing variable substitution to low-pass filters. The
transfer function of a high-pass filter can be found by substituting 𝑠𝑛 with 1/𝑠𝑛 such that
8.69)
𝐻ℎ𝑝𝑛 (𝑠𝑛 ) = 𝐻𝑙𝑝𝑛 (1/𝑠𝑛 ).

The high-pass filter has the same filter order as the low-pass filter. The transfer function of a band-
pass filter can be found by substituting 𝑠𝑛 with 𝑄𝑏𝑝 (𝑠𝑛 + 1/𝑠𝑛 ) such that
8.70)
𝐻𝑏𝑝𝑛 (𝑠𝑛 ) = 𝐻𝑙𝑝𝑛 (𝑄𝑏𝑝 (𝑠𝑛 + 1/𝑠𝑛 )).

The band-pass filter order is twice the order of the low-pass filter.

EX. 8.3
Design a fourth-order Buttworth band-pass filter with a center frequency of 1 MHz, bandwidth of 100
kHz, and passband gain of 1.

First start with a normalized second-order Butterworth low-pass filter, which has a transfer function
of
1
𝐻𝑙𝑝𝑛 (𝑠𝑛 ) = 2
𝑠𝑛 + √2𝑠𝑛 + 1
Use variable substitution to convert the transfer function into a band-pass filter
1
𝐻𝑏𝑝𝑛 (𝑠𝑛 ) = 𝐻𝑙𝑝𝑛 (𝑄𝑏𝑝 (𝑠𝑛 + ))
𝑠𝑛

126
 1
= 2
1 1
10 (𝑠𝑛 + 𝑠 ) + 10√2 (𝑠𝑛 + 𝑠 ) + 1
𝑛 𝑛
0.01𝑠𝑛2
= 2
[𝑠𝑛 + 0.0682𝑠𝑛 + 0.9318][𝑠𝑛2 + 0.0732𝑠𝑛 + 1.0734]
The above transfer function can be implemented with two-stages of second-order active band-pass
filters. The component values can then be frequency scaled to 1 MHz.

127