Systems

Broadly speaking, a system is anything that

responds when stimulated or excited
The systems most commonly analyzed by
engineers are artificial systems designed and
built by humans
Engineering system analysis is the
application of mathematical methods to the
design and analysis of systems

Description of Systems

M. J. Roberts - All Rights Reserved.

Edited by Dr. Robert Akl

1

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

System Examples

Feedback Systems
In a feedback system the response of the system is fed back

and combined with the excitation is such a way as to optimize

the response in some desired sense. Examples of feedback

systems are

1. Temperature control in a house using a thermostat

2. Water level control in the tank of a flush toilet.

3. Pouring a glass of lemonade to the top of the glass without


overflowing.

4. A refrigerator ice maker which keeps the bin full of ice


but does not make extra ice.

5. Driving a car.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Systems

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

A Multiple-Input, MultipleOutput System Block Diagram

Systems have inputs and outputs

Systems accept excitations or input signals
at their inputs and produce responses or
output signals at their outputs
Systems are often usefully represented by
block diagrams
A single-input, single-output system block diagram

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Block Diagram Symbols

Block Diagram Symbols

Three common block diagram symbols for an amplifier (we will

use the last one).

Block diagram symbol for an integrator

Three common block diagram symbols for a summing junction

(we will use the first one).

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

An Electrical Circuit Viewed as a

System

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Zero-State Response of an RC
Lowpass Filter to a Step Excitation

An RC lowpass filter is a simple electrical system.

If an RC lowpass filter with an initially uncharged capacitor is

It is excited by a voltage vin ( t ) and responds with a

excited by a step of voltage v in t = Au t its response is

voltage vout ( t ) . It can be viewed or modeled as a

()

) ()

()

()

v out t = A 1 et / RC u t . This response is

single-input, single-output system

called the zero - state response of this system

because there was initially no energy stored in
the system. (It was in its zero-energy state.)
If the excitation is doubled, the zero-state
response also doubles.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Zero-Input Response of an RC
Lowpass Filter

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

10

Homogeneity
In a homogeneous system, multiplying the
excitation by any constant (including complex
constants), multiplies the zero-state response by the
same constant.

If an RC lowpass filter has an initial charge on the capacitor

of V0 volts and no excitation is applied to the system its zero -

()

input response is v out t = V0 et / RC , t > 0.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

11

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

12

Homogeneity

Homogeneity
Let y ( t ) = exp ( x ( t )) . Is this system homogeneous?
Let x ( t ) = g ( t ) . Then y ( t ) = exp ( g ( t )) .
Let x ( t ) = K g ( t ) . Then y ( t ) = exp ( K g ( t )) = exp ( g ( t ))

K y ( t ) = K exp ( g ( t )) y ( t ) K y ( t ) , Not homogeneous

To test a system for homogeneity use this logical

process. Apply a signal g ( t ) as the excitation x1 ( t )
and find the zero-state response y1 ( t ) . Then apply

the signal K g ( t ) as x 2 ( t ) where K is a constant and

find the zero-state response y 2 ( t ) . If y 2 ( t ) = K y1 ( t )

homogeneous.

If g(t)
y1 ( t ) and K g(t)
K y1 ( t )
H

H is Homogeneous
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

() ()
Let x ( t ) = g ( t ) . Then y ( t ) = g ( t ) + 2.
Let x ( t ) = K g ( t ) . Then y ( t ) = K g ( t ) + 2
K y ( t ) = K g ( t ) + 2K y ( t ) K y ( t ) , Not homogeneous
Let y t = x t + 2. Is this system homogeneous?

for any arbitrary g ( t ) and K, then the system is

13

Time Invariance

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

14

Time Invariance
Let y ( t ) = exp ( x ( t )) . Is this system time invariant?
Let x ( t ) = g ( t ) . Then y ( t ) = exp ( g ( t )) .
Let x ( t ) = g ( t t ) . Then y ( t ) = exp ( g ( t t ))
y ( t t ) = exp ( g ( t t )) y ( t ) = y ( t t ) , Time Invariant

If an excitation causes a zero-state response and

delaying the excitation simply delays the zerostate response by the same amount of time,
regardless of the amount of delay, the system is
time invariant.

() ( )
() ()
() ( )
Let x ( t ) = g ( t t ) . Then y ( t ) = g ( t / 2 t )
y ( t t ) = g (( t t ) / 2 ) y ( t ) y ( t t ) , Time Variant
Let y t = x t / 2 . Is this system time invariant?
Let x1 t = g t . Then y1 t = g t / 2 .
2

If g(t) H y1 ( t ) and g(t t 0 ) H y1 ( t t 0 ) H is Time Invariant

This test must succeed for any g and any t 0 .

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

15

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Additivity

16

Additivity

If one excitation causes a zero-state response and another excitation

causes another zero-state response and if, for any arbitrary
excitations, the sum of the two excitations causes a zero-state
response that is the sum of the
two zero-state responses, the
system is said to be additive.

( ) ( ( ))
Let x ( t ) = g ( t ) . Then y ( t ) = u ( g ( t )) .
Let x ( t ) = h ( t ) . Then y ( t ) = u ( h ( t )) .
Let x ( t ) = g ( t ) + h ( t ) . Then y ( t ) = u ( g ( t ) + h ( t ))
y ( t ) + y ( t ) = u ( g ( t )) + u ( h ( t )) u ( g ( t ) + h ( t )) . Not additive.
(For example, at time t = 3, if g ( 3) = 4 and h ( 3) = 2,
y ( 3) + y ( 3) = u ( 4 ) + u ( 2 ) = 1+ 1 = 2. But y ( 3) = u ( 4 + 6 ) = 1.)
Let y t = u x t . Is this system additive?
1

If g(t) H y1 ( t ) and h(t) H y 2 ( t )

and g ( t ) + h ( t ) H y1 ( t ) + y 2 ( t ) H is Additive
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

17

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

18

Linearity and LTI Systems

Linearity and LTI Systems

If a system is both homogeneous and additive

it is linear.
If a system is both linear and time-invariant it
is called an LTI system
Some systems which are non-linear can be
accurately approximated for analytical
purposes by a linear system for small
excitations
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

In an LTI system, the analysis of the effect of an excitation

on a system can be found by breaking the excitation into pieces,
finding the responses to those pieces individually and then adding

()
( )
()
) ( )
( )
0.25rect ( t / 4 ) or into any other convenient sum of functions that
equals x ( t ) .
those responses. Let x t = rect t / 4 . We could break x t

into u t + 2 u t 2 or we could break it into 0.75rect t / 4 +

19

Stability

Any system for which the zero-state response

occurs only during or after the time in which
the excitation is applied is called a causal
system.
Strictly speaking, all real physical systems are
causal

21

Memory

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

22

Static Non-Linearity
Many real systems are non-linear because the
relationship between excitation amplitude and
response amplitude is non-linear

If a systems zero-state response at any arbitrary

time depends only on the excitation at that same
time and not on the excitation or response at any
other time it is called a static system and is said to
have no memory. All static systems are causal.
A system whose zero-state response at some
arbitrary time depends on anything other than
the excitation at that same time is called a dynamic
system and is said to have memory
Any system containing an integrator has memory
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

20

Causality

Any system for which the response is

bounded for any arbitrary bounded excitation,
is called a bounded-input-bounded-output
(BIBO) stable system
A continuous-time LTI system described by a
differential equation is stable if the
eigenvalues of the solution of the equation all
have negative real parts

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

V-I Diagram for a

Linear Resistor

23

V-I Diagram for a Diode

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

24

Static Non-Linearity

Invertibility

For an analog multiplier, if the two excitations are the same

single excitation signal, the response signal is the square of
that single excitation signal and doubling the excitation
would cause the response to increase by a factor of 4
Such a system is not homogeneous and therefore not linear

A system is said to be invertible if unique

excitations produce unique zero-state responses. In
other words, if a system is invertible, knowledge of
the zero-state response is sufficient to determine the
excitation

This full-wave rectifier is

a non-invertible system

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

25

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Dynamics of Second-Order
Systems

Dynamics of Second-Order
Systems

The eigenfunction of an LTI system is the complex exponential.

The eigenvalues of a second-order system are either both real or
occur in a complex-conjugate pair. The general solution form is
a sum of two complex exponentials and a constant. For example,
the capacitor voltage in a series RLC circuit excited by a voltage
step of height A is

()

v out t = K1e

+ 2 2 t
n

+ K2e

2 2 t
n

26

+ 2 2 t
n

The solution form K1e

2 2 t
n

+ K 2 e

applies

to all second-order LTI systems. It can also be written as

+ j c )t
j c )t
K e(
+ K e(
where = 1 2 and
1

is the critical radian frequency and = / n is the

+A

damping ratio.

where = R / 2L and n2 = 1 / LC. is the damping factor and

n is the natural radian frequency.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

27

28

Discrete-Time Systems

Complex Sinusoid Excitation

With the increase in speed and decrease in

cost of digital system components, discretetime systems have experienced, and are still
experiencing, rapid growth in modern
engineering system design
Discrete-time systems are usually described
by difference equations

Any LTI system excited by a complex sinusoid responds with

another complex sinusoid of the same frequency but generally
a different magnitude and phase. In the case of the RLC circuit
if the excitation is vin ( t ) = Ae j 2 f0t the response is vout ( t ) = Be j 2 f0t
where A and B are, in general, complex. B can be found by
substituting the solution form into the differential equation and
finding the particular solution. In the RLC circuit
A
B=
( j2 f0 )2 LC + j2 f0 RC + 1

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

29

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

30

Block Diagram Symbols

Discrete-Time Systems
In a discrete-time system events occur at points in time but not

between those points. The most important example is a digital

computer. Significant events occur at the end of each clock

cycle and nothing of significance (to the computer user) happens

between those points in time.



Discrete-time systems can be described by difference (not

differential) equations. Let a discrete-time system generate an

excitation signal y[n] where n is the number of discrete-time

intervals that have elapsed since some beginning time n = 0.

Then, for example a simple discrete-time system might be

described by

y [ n ] = 1.97 y [ n 1] y [ n 2 ]

The block diagram symbols for a summing junction and an

amplifier are the same for discrete-time systems as they are

for continuous-time systems.

Block diagram symbol for a delay

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

31

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

Discrete-Time Systems

32

Discrete-Time Systems
y [ n ] = 1.97 y [ n 1] y [ n 2 ]

The equation
y [ n ] = 1.97 y [ n 1] y [ n 2 ]

We could solve this equation by iteration using a computer.

Initial Conditions

says in words

yn = 1 ; yn1 = 0 ;
while 1,

The signal value at any time n is 1.97 times the signal value at the

yn2 = yn1 ; yn1 = yn ; yn = 1.97 * yn1 - yn2 ;

previous time [ n 1] minus the signal value at the time before that

end

[ n 2 ].

If we know the signal value at any two times, we can compute its
value at all other (discrete) times. This is quite similar to a

We could also describe the system

with a block diagram.

second-order differential equation for which knowledge of two

independent initial conditions allows us to find the solution for all
time and the solution methods are very similar.
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

33

Discrete-Time Systems

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

34

Solving Difference Equations

y [ n ] = 1.97 y [ n 1] y [ n 2 ]

On the previous two slides we found the solution to

y [ n ] = 1.97 y [ n 1] y [ n 2 ]

With the initial conditions y [1] = 1 and y [ 0 ] = 0 the (zero-

by iteration as a sequence of numbers for y [ n ]. We can also solve

input) response is y [ n ].

linear, constant-coefficient ordinary difference equations with techniques

that are very similar to those used to solve linear, constant-coefficient
ordinary differential equations. The eigenfunction of this type of equation
is the complex exponential z n . As a first example let the equation be
2 y [ n ] y [ n 1] = 0. The homogeneous solution of this equation is
then y h [ n ] = Kz n . Substituting that into the equation we get

2Kz n Kz n 1 = 0. This is the characteristic equation. Dividing

through by Kz n 1 we get 2z 1 = 0 and the solution is z = 1 / 2.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

35

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

36

Solving Difference Equations

Solving Difference Equations

The solution of inhomogeneous equations is also similar to
differential equation techniques. Let 5y [ n ] 3y [ n 1] = (1 / 3) with
n

an initial condition of y [ 0 ] = 1. The characteristic equation is

The eigenvalue for the equation 2 y [ n ] y [ n 1] = 0 is then z = 1 / 2

5z 3 = 0. The eigenvalue is 3 / 5 and the homogeneous solution is

and the homogeneous solution is y h [ n ] = K (1 / 2 ) . Since the equation

n

y h [ n ] = K ( 3 / 5 ) . The particular solution is a linear combination of

n

is homogeneous, the homogeneous solution is also the total solution.

the forcing function (1 / 3) and all its unique differences. The first
n

To find K we need an initial condition. Let it be y [ 0 ] = 3. Then

backward difference of (1 / 3) is (1 / 3) (1 / 3)
n

y [ 0 ] = K (1 / 2 ) = K = 3 and y [ n ] = 3(1 / 2 ) .
0

n1

, which can be written

as 2 (1 / 3) . This is just the same function but with a different

n

multiplying constant. So the first difference of an exponential is also

an exponential. Therefore the only functional form we need for the
particular solution is K p (1 / 3) .
n

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

37

Solving Difference Equations

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

38

A System

Substituting the particular solution form into the difference equation

we get 5K p (1 / 3) 3K p (1 / 3)
n

Kp =

(1 / 3)

5 (1 / 3) 3(1 / 3)
n

n 1

n 1

= (1 / 3) . Solving,
n

1/ 3
= 1 / 4. Then the total
5 (1 / 3) 3

If the excitation x[n] is the unit sequence, the zero-state

response is

solution is y [ n ] = K ( 3 / 5 ) (1 / 4 ) (1 / 3) . Applying initial

n

n
y [ n ] = 5 4 ( 4 / 5 ) u [ n ]

conditions y [ 0 ] = K ( 3 / 5 ) (1 / 4 ) (1 / 3) = K 1 / 4 = 1
0

Therefore K = 3 / 4 and the total solution is

y [ n ] = ( 3 / 4 ) ( 3 / 5 ) (1 / 4 ) (1 / 3) .
n

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

39

A System

40

System Properties

The properties of discrete-time systems have

the same meaning as they do in continuoustime systems

If the excitation is doubled, the zero-state response doubles.

If two signals are added to form the excitation, the zero-state

response is the sum of the zero-state responses to those two

signals. If the excitation is delayed by some time, the zero-state

response is delayed by the same time. This system is linear and

time invariant.

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

41

M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

42

Eigenfunctions of LTI Systems

The eigenfunction of an LTI system is the
complex exponential
The eigenvalues are either real or, if
complex, occur in complex conjugate pairs
Any LTI system excited by a complex
sinusoid responds with another complex
sinusoid of the same frequency, but generally
a different amplitude and phase
All these statements are true of both
continuous-time and discrete-time systems
M. J. Roberts - All Rights Reserved. Edited by Dr. Robert Akl

43