EMS507 Control System

Analysis and Design

Week 5 – Simple linear systems

Yunpeng Zhu
Simple linear systems
© Copyright 2023 Yunpeng Zhu. All Rights Reserved
Edition: v1.1
Table of Contents
1 The first order system .............................................................................- 1 -
1.1 Concept .................................................................................................- 1 -
1.2 Examples ...............................................................................................- 1 -
1.3 Responses of the first order system ......................................................- 2 -
2 The second order system ........................................................................- 8 -
2.1 Impulse responses of the second order system ....................................- 8 -
2.2 Unit step responses of the second order system ................................ - 12 -
2.3 Time Domain Specifications ................................................................ - 15 -
3 Summary ............................................................................................... - 19 -

i
1 The first order system
1.1 Concept

In control theories, the order of the system is defined by its characterization

equation.
Obviously, a proper first order transfer function is written as
K
G(s) 
Ts  1
where K and T are real numbers.
 Proper means the order of the numerator is smaller than that of the
denominator;
 First order means the order of the characterization equation is 1.

1.2 Examples

Discussion: Do you remember any first order systems we have learned?

 The RC circuits:

Applying Kirchhoff’s Second Law gives:

Vi  VR  VC

1 dVC
C
With VR  iR and VC  i dt  i  C ,
dt
dVC
VR  VC  iR  VC  RC  VC  Vi
dt

-1-
 The water tank:

d(AH ) dH dH
Q  Qo  A A   gH  Q
dt dt dt
1.3 Responses of the first order system

Quiz 1.1: How to get the first order transfer function using a close loop control?

1
K
G ( s )  K Ts 
1 Ts  1
1
Ts
Without loss generality, in the study of first order systems, we can take K  1 ,
thus
1
G(s) 
Ts  1
The whole system can be represented as
1
Y ( s)  U ( s)
Ts  1
(a) Impulse response
Consider the unit impulse signal as an input to the first order system,

-2-
 u (t )   (t )  U ( s)  1

Substitute U ( s )  1 in

1
Y ( s)  U ( s)
Ts  1
yields
1
Y (s) 
Ts  1
Rearrange the above equation in head-1 standard form:
1 1
Y (s) 
T s  T 1
Laplace transform pair:
1
e  t , t  0 Re( s)  
s 

1 T1 t
y (t )  e , t  0
T

The unit impulse response y(t) is an exponential decaying signal for positive
values of ‘t’ and it is zero for negative values of ‘t’.
Quiz 1.2: Plot the impulse response of the system
5 y(t )  2 y(t )  u (t )
-3-
 Conduct Laplace transform on both sides of the equation:
5sY ( s)  2Y ( s)  U ( s)

The transfer function of the system is

Y ( s) 1 1 1
G(s)   
U ( s ) 5s  2 5 s  2
5
Subject to impulse input, the system response is

1 1 1  52 t
Y ( s )  G ( s )U ( s )   y (t )  e , t  0
5s 2 5
5

(b) Unit step response

Consider the unit impulse signal as an input to the first order system,

1
u (t )  us (t )  U ( s) 
s

1
Substitute U ( s)  in
s
1
Y ( s)  U ( s)
Ts  1
yields
1 1
Y ( s) 
Ts  1 s
Rearrange the above equation in head-1 standard form:
-4-
 1 1 A B A(Ts  1)  Bs
Y (s)    
s Ts  1 s Ts  1 s (Ts  1)
Let A(Ts  1)  Bs  1, there is

A 1

 AT  B  T  B  0  B  T
Therefore:
1 T 1 1
Y ( s)    
s Ts  1 s s  T 1
Laplace transform pair:
1
us (t ) Re( s)  0
s
1
e  t , t  0 Re( s)  
s 
1
 t
y (t )  1  e T
,t 0
1
 t
The transient response: ytr (t )  e T
,t 0

The steady state response: yss (t )  1, t  0

The value of the unit step response y(t) is zero at t = 0 and for all negative
values of t. It is gradually increasing from zero value and finally reaches to one
in steady state. The steady state value depends on the magnitude of the input.
Quiz 1.3: Plot the unit step response of the system
3 y(t )  y(t )  2u (t )
-5-
 Conduct Laplace transform on both sides of the equation:
3sY ( s)  Y ( s)  2U ( s)

The transfer function of the system is

Y ( s) 2
G( s)  
U ( s) 3s  1
Subject to unit step input, the system response is
1
2 1 1 1  t
Y ( s )  G ( s )U ( s )   2(  )  y (t )  2  2e 3 , t  0
3s  1 s s s 1
3

(c) Unit ramp response

Consider the unit ramp signal as an input to the first order system,

1
u (t )  tus (t )  U ( s) 
s2

1
Substitute U ( s)  in
s2
1
Y ( s)  U ( s)
Ts  1
yields
1 1
Y ( s) 
Ts  1 s 2

-6-
Rearrange the above equation using partial fractions:

A B C Bs (Ts  1)  A(Ts  1)  Cs 2 1
Y (s)  2     2
s s Ts  1 s (Ts  1)
2
s (Ts  1)

Let Bs (Ts  1)  A(Ts  1)  Cs 2  ( BT  C )s 2  ( B  AT )s  A  1 , there is

 BT  C  0 C  T
2

 
 B  AT  0   B  T
A 1 A 1
 
Therefore:

1 T T2 1 T T
Y (s)  2    2 
s s Ts  1 s s s 1
T
Laplace transform pair:
1
us (t ) Re( s)  0
s
1
e  t , t  0 Re( s)  
s 
1
t, t  0 Re( s)  0
s2
1
 t
y (t )  t  T  Te T
,t 0
1
 t
The transient response: ytr (t )  Te T
,t 0

The steady state response: yss (t )  t  T , t  0

-7-
The unit ramp response, y(t) follows the unit ramp input signal for all positive
values of t. But, there is a deviation of T units from the input signal.

2 The second order system

2.1 Impulse responses of the second order system

We first consider a mass-spring-damper system below.

k
m
y(t) u(t)

The differential equation can be written as

my  cy  ky  u

This is a very typical second order system. In follows, we will focus on

investigating the properties of this system.
 Consider c  0
If c  0 , there is no damping, the differential equation becomes
my  ky  u

At the zero initial conditions, take Laplace transform on both sides of the
equation, yields

ms 2Y ( s )  kY ( s )  U ( s )

The transfer function is

1
G( s) 
ms  k
2

Assuming the input is unit impulse, there is

k
1 1 m  y (t )  1 sin( k t ), t  0
Y (s)  2 
ms  k mk s 2  k mk m
m
-8-
according to the Laplace Transform pair:
Sine function 
sin(t )  us (t ) or sin(t ), t  0 s  2
2

This indicates that when you give the system an impulse, the system will
k
oscillate under the frequency   (rad/s), this frequency is known as the
m
natural frequency. Natural frequency is the intrinsic property of a system.
 Consider c  0
The differential equation
my  cy  ky  u

can be normalized as (Normal form)

c k 1 1
y y  y  u  y  2n y  n 2 y  u
m m m m

k c c
where n  is the natural frequency; 2n     is the
m m 2mn
damping factor.

k
The frequency n  is also written as n  2 f n , where f n is the physical
m
frequency with the unit Hz.
Why we normalize the second order system in this way?

If   1 , let u  u m , the second order system becomes

y  2n y  n 2 y  u

Taking Laplace transform on both sides:

s 2Y ( s )  2n sY ( s )  n 2Y ( s )  U ( s )

The transfer function becomes:

1
G (s) 
s 2  2n s  n 2

-9-
 2n  4( 2  1)n 2
The system has two complex poles: s1 , s2  .
2
s1 , s2    jd

with   n and d  n 1   2 .

We know that a complex transfer function can be decomposed by partial

fractions, then the inverse Laplace Transform can be applied to each fraction
to find the time responses.

In the unit impulse response of the system, real poles pairs e  t dominant
transient responses, while imaginary poles pairs sin(dt ) dominant steady
state oscillations.
The imaginary part of the poles d above is known as the damp natural
frequency as it dominants the oscillation of the second order system.

If   1 , the second order system becomes

y  2n y  n 2 y  u

Taking Laplace transform on both sides:

s 2Y ( s )  2n sY ( s )  n 2Y ( s )  U ( s )

The transfer function becomes:

1 1
G ( s)  
s  2n s  n
2 2
( s  n ) 2

The system has two real and equal poles: s1  s2  n .

- 10 -
 If   1 , The transfer function of the second order system becomes
1
G (s) 
s 2  2n s  n 2

2n  4( 2  1)n 2

The system has two real poles: s1 , s2  .
2
The impulse response is therefore non-oscillation.

Summarize:
1. The second order system wants to oscillate

2. The oscillating frequency is the (damped) natural frequency

k
n  or d  n 1   2
m
3. The damping aims to reduce this oscillation.

- 11 -
4. A summary table
 0 Undamped Imaginary poles Non-stop oscillation
0   1 Under damped Complex poles Decayed oscillation
 1 Critically damped Equal real poles No oscillation
 1 Over damped Real poles No oscillation

2.2 Unit step responses of the second order system

Now we can say the typical second order transfer function can be written as
1
G (s) 
s  2n s  n 2
2

BUT in many cases, we want to know the system responses under the unit step
input.
For a damped second order system subject to unit step input

y  2n y  n 2 y  us

the steady state response y (t ) is a constant and therefore

dx d2 x
lim  0 , lim 2  0
t  dt t  dt

Substitute these into the differential equation and hence:

us
y  2n y  n 2 y  us  n 2 y  us or y 
n 2

- 12 -
This implies that under the unit step input, the steady state output is 1 n 2 .
Usually we prefer the output is 1 for the convenience of system analysis.
Therefore, we often rewrite the normalized second order system as

y  2n y  n 2 y  n 2u

and the standard transfer function is

n 2
G ( s)  2
s  2n s  n 2

Note that both

1 n 2
G (s)  and G ( s )  2
s 2  2n s  n 2 s  2n s  n 2

are correct transfer functions of a second order system, but the second one is
normalized considering its steady state response under the unit step input is 1.
Quiz 2.1: How to get the normalized second order transfer function using a
close loop control?

n 2
s ( s  2n ) n 2
G (s)   2
n 2 s  2n s  n 2
1
s ( s  2n )

The whole system can be represented as

n 2
Y (s)  2 U (s)
s  2n s  n 2

Now we will discuss the responses of the second order system under the unit
step input:

n 2 1
Y (s)  2
s  2n s  n s
2

- 13 -
The time domain response can be reconstructed by using the inverse Laplace
Transform. Similar to the impulse response, we can get the Undamped, Under
damped, Critically damped, and Over damped responses below

 0 Undamped y (t )  1  cos(nt ), t  0
0   1 Under damped e  t
y (t )  1  [d sin(d t )   cos(d t )],
d
  n , t  0
 1 Critically damped y (t )  1  e nt  nte nt , t  0

 1 Over damped 1
y (t )  1  e s1t
2(    1)(   1)
2 2

,t  0
1
 e s2 t

2(    1)(   1)
2 2

The derivations of the above results are complex, there is no need to

remember them. But the following plots are important:

- 14 -
2.3 Time Domain Specifications

- 15 -
All the time domain specifications of an underdamped second order system
are represented in this figure.
The response up to the settling time ts is known as transient response and the
response after the settling time is known as steady state response.
We will talk about this figure step by step.
(a) Delay time
It is the time required for the response to reach half of its final value from the
zero instant. It is denoted by td.
The step response of the normalized second order system will end at 1.
Therefore, at t  td , the value of the step response will be 0.5. This delay time
td is
1  0.7
td 
n
(b) Rise time
It is the time required for the response to rise from 0% to 100% of its final
value. This is applicable for the under-damped systems. For the over-damped
systems, consider the duration from 10% to 90% of the final value. Rise time is
denoted by tr. The rise time tr is
 
tr  , cos( )  
d
(c) Peak time
It is the time required for the response to reach the peak value for the first
time. It is denoted by tp. At t  t p , the first derivate of the response is zero.

The peak time tp and the damped frequency d are inversely proportional to
each other:

tp  (about half period)
d

- 16 -
(d) Overshoot
Peak overshoot Mp is defined as the deviation of the response at peak time
from the final value of response. It is also called the maximum overshoot.
Mathematically, we can write it as
M p  y (t p )  y ()

where y (t p ) is the peak value of the response; y () is the final (steady state)
value of the response.

Substituting t p  in to the response of the system, yields
d


1 2
Mp e

Percentage of peak overshoot % M p can be calculated by using this formula.



Mp 1 2
%M p   100%  e  100%
y ( )

From the above equation, we can conclude that the percentage of peak
overshoot % M p will decrease if the damping ratio  increases.

(e) Settling time

It is the time required for the response to reach the steady state and stay
within the specified tolerance bands around the final value. In general, the
tolerance bands are 2% and 5%. The settling time is denoted by ts.
The settling time for 5% tolerance band is:
3
ts   3
n
The settling time for 2% tolerance band is:
4
ts   4
n
where   1 n  1  is the time constant.

- 17 -
 Bothe the settling time t s and the constant  are inversely proportional to
the damping ratio 
 Bothe the settling time t s and the constant  are independent of the
system gain. That means even the system gain changes, the settling time
t s and the constant  will never change.

Quiz 2.2: Find the time domain specifications of a control system having the
closed loop transfer function
4
G(s) 
s  2s  4
2

when the unit step signal is applied as an input to this control system.
We know that the standard form of the transfer function of the second order
closed loop control system as

n 2
G ( s)  2
s  2n s  n 2
By equating these two transfer functions, we will get the undamped natural
frequency n as
n  2 rad/s
And the damping ratio  as
  0.5
The damped frequency d can therefore be achieved as

d  n 1   2  2 1  0.52  1.732 rad/s

Substitute,  value in the following relation:

cos( )      cos 1 (0.5) 
3
Substitute the above necessary values in the formula of each time domain
specification. The following table shows the formulae of time domain
specifications.

- 18 -
 Time domain Substitution of
formula Final value
specification values in formula
1  0.7 1  0.7  0.5
Delay time td  td  td  0.675 s
n 2

  
tr  
Rise time 3 tr  1.207 s
d tr 
1.732
 
Peak time tp  tp  t p  1.813 s
d 1.732
 0.5
 
% Over shoot 1 2
%M p  e 10.52 % M p  16.32%
%M p  e

Settling time for 2% 4 4

ts  ts  ts  4 s
tolerance band n 0.5  2

3 Summary
Unit step response is the most important case:

- 19 -