Chapter 3

Time-Domain Analysis of
Control System
 Chapter 3
Time-Domain Analysis of the linear systems
 Introduction
 First-order system analysis
 Second-order system analysis
 High-order system analysis
 Stability and Algebraic criteria
 Analysis of stable error
 3.1 Introduction
3.1.1 Basic and macroscopically requirements
to design a control system.
1) The system must be stable (stability)
—First requirement.
2) The control should be accurate (accuracy).
3) The response should be quick-acting (rapidity).
3.1.2 Basic assumption conditions for analyzing a
linear control system
The response of a system could be :

transient steady-state
portion portion

3
3.1.3 Typical test input signal
1. why to research the test 2. Which types of the
input signal?
test input signal ?
 The actual input signal of
 Step input signal;
the system is multifarious,
normally a standard test  Ramp input signal;
input signal should be  Parabolic input
chosen→to analyze the
system performance. signal;
 Allow the designer to  Pulse input signal;

compare several designs  Sinusoidal input

project. signal;
 Many input signals of the
control systems are similar
to the test signals.
4
 3. Typical test input signal
1) Step input signal 2) Ramp input signal
r(t)  At t 0 r (t )
A t 0 r ( t ) 
r (t )  0 t0 tg  A
0 t 0 A
2 α
R(s)=A/s
R( s )  A / s 0 t
0 t
A=1 — unity step input A=1— unity ramp function
Application : test signal of Application :
the constant-input systems some input-tracking systems
Such as: Such as:
switch turn on; relay close train uniformity speed mo-
tion; elevator uniformity
step signal is also called speed raise.
the position signal. Ramp signal is also called
the velocity signal.
5
3) Parabolic input signal 4) Pulse input signal
r (t ) r (t )
1 2 A 0 t 
 At
r ( t )  2
t 0  
r ( t ) 
 0 t0 0 t
t , t  0
 0ε
R( s )  A / s 2 t
lim r ( t ) A1  ( t ) 0
A = 1 — unity parabolic 0

signal Unity impulse. r (t )

Application : R(s)  L ( t ) 1 A  (t )

same as the Ramp signal t

Application:
such as:
Servo-control system some impulse experiments
Such as:
Parabolic signal is also called Impactive disturbance
the acceleration signal.

6
5) Sinusoidal signal

 Asin t t 0

r ( t ) 
 0 t 0


A
R( s ) 
s2  2

Application: Frequency domain analysis of control systems

such as: Communication system, Radar system …

7
3.1.4 Relationship between impulse response and other
responses
1. For the typical input signals
Theorem:
t
If : r1 ( t )  c1 ( t ) dr1 ( t )
and r2 ( t )  or r2 ( t ) r1 ( t )dt
dt
r2 ( t )  c2 ( t ) 0
t
dc1 ( t )
Then : c2 ( t ) 
dt
0
or c 2 (t) 
 c1 (t)dt
d
proof: r2(t)  r1(t)  R2(s)  sR1(s)
dt
C 2 ( s ) G ( s ) R2 ( s ) G ( s ) sR1 ( s ) d
c2 ( t )  c1 ( t )
 s[G ( s ) R1 ( s )]  sC1 ( s ) dt
as above : t
t
r2 ( t ) 
r (t )dt
0
1 

c 2 ( t )  c1 ( t )
0
8
If: g(t) →impulse response of a linear system.
g(t) = L-1[ G(s)]
h(t) → step response of a linear system.
ct(t) → ramp response of a linear system.
ctt (t) → parabolic response of a linear system
then:

9
3.1.5 The transient performance specifications of
a control system
 For different system, the research aim is different.
For example, the tracking servo control systems
are different from the constant-regulating systems.
 performance measure — in terms of the step
responses of the systems.

A typical unity step response of a control system is

shown in following figure

10
Performance specification definition:

A
Overshoot σ% = 100%
A B

Setting time ts

Peak time tp B

Risetime tr
Rise
Time tr

11
1) Rise time tr : usually ,it is defined as the time required for the step
response to reach the final value first time.
c( t ) t t c() ( first )
r
c()  the final value of t he time re sponse
(for the under damped systems)

c( t ) t t 0.9c( ) ( first )
t r t 2  t 1 2

c( t ) t t 0.1c( ) ( first )
1
(for the over damped systems)

2) Peak time tp : is the time to the maximum of the first

overshoot.
dc( t )
0 ( first )
dt t t p
(for the under damped systems)
12
3) Percent overshoot σp : is the maximum amount the system
overshoots its final value divided by its final value.
c ( t p )  c ( ) Mp
 p%  100%  100%
c ( ) c ( )
c( t p )  the peak value of the time response
(For the under damped systems)
4) Setting time ts : the time required for the step response to
settle to within a certain tolerance. The tolerance  5%or 2%
c() 2%

(c(t )  c() t t   or
s
c() 5%

σ% → smoothness of the response.
tr 、 tp 、 → rapidity of the response.
ts → rapidity of the transient process of the response.
13
h(t)

ts

tr

t
h(t)
σ%= A 100%
A B

tp ts t
tr
 3. 2 Time Response of First-Order System
1. What is the first-order system?
The system could be described by a first-order
differential equation—the first-order system .

Differential dc(t )
equation T  c(t ) r (t ) Where T is called
dt
C (s) 1 the time constant
Generalized (s)  
Transfer function
R( s) Ts  1
1
Block diagram TS  1

Example:
16
2 . The step response of the first-order system
1
r (t ) 1(t ) R(s) 
s
1 1 1 T
C ( s )  ( s ) R ( s )    
Ts  1 s s Ts  1
t

c(t ) 1  e T

1) the magnitude of ourput may be calculated from

t T , c(t ) 1  e  1 0.632 63.2%
t 3T , c(t ) 0.95 95%
t 4T , c(t ) 0.98 98%
so
3T  0.05
t s 
4T  0.02
 1 Slope 1/T
2) the initial value of the slope is
T
t
dc(t ) 1 T 1 1
t 0  e t 0 
dt T T
0.632
A
Which means that the response
would reach its final value in T
seconds if it had continued to
T
increase at the initial rate, so that
the parameter T is called the
constant of the first-order system
and has unit of seconds.

3 . As the same way, we can get the other typical test signals
response of the first-order system
 The typical response of the first-order system

r(t) R(s) C(s)= s) R(s) §3.2.3

c(t) 一阶系统的典型响应 Response

t) 1

1(t)

t
4. The first-order feedback systems
suppose the first-order
feedback system is shown in
right Fig.

C (s) KG ( s ) K
the closed - loop tansfer function is  
R ( s ) 1  KG ( S ) Ts  K  1
K 1 K /T
if the input is unit step, then C(S)  . 
Ts  K  1 s s[ s  ( K  1) / T ]
K K 1
 .s  .
K 1 K  1 s  ( K  1) / T
K 1
K  t
hence the time response c(t )  (1  e T
)
K 1
Suppose the T 1, K 1, we get c(t ) 0.5(1  e  2t )
We note:
①The steady-state output of the
feedback system is 0.5
②The time constant of the
feedback system is 0.5,so the
feedback system response is faster
than that of the open-loop system.

however, if set K 10, the output becomes

c (t ) 0.909(1  e  11t )
①The steady-state output of the feedback system is reduced to
less than 10%
②The close-loop time is 0.09,resulting over 10 times the response
speed of the open-loop system.
Note the following conclusion:
①using feedback possible to improve the dynamic
response ,meanwhile, this process may also have
some detrimental effects such as the appearance of
an error between the input and steady-state output.
②using a large value of gain not only does the
system response become very fast ， but also the
error becomes smaller. But in practical case the gain
cannot be increased indefinitely without signal
saturation occurring, causing the system to become
nonlinear.
Example: the first-order system is shown as
fig.1, now change it to a feedback system
shown as fig.2, but you need to reduce the
setting time as 0.1 times as the open-loop
response and keep the gain invariable, how Fig.1
to set the K0 and KH ?

The close-loop system Fig.2

should satisfy:
10 K O 10 K O
K G(S ) 10 K O 1  10 K H
(s)  O  0 .2 s  1  
1  K H G ( s ) 1  10 K H 0.2 s  1  10 K H 0. 2
s 1
0 .2 s  1 1  10 K H
 10 K O
 1  10 K  K * 10
 0 .2 H  K O 10
 
1  10 K H
T * 0.02  K H 0.9
 3.3 Time response of the second-order system
1. What is the second-order system?
The system could be described by a second-order differential
equation.

2. Normalized second-order systems and its transfer function

R (s ) k C (s ) R(s )  n2 C (s )
 s(Ts  1) s 2  2 n s   n2

Here: k 1
n  
T 2 kT
natural undamped damping ratio
oscillatory frequency
24
 3. step response of the typical second-order systems
C (s)  n2
 2 
R( s ) s  2 n s   n
2

2
The characteristic equation is D( s )  s 2  2 n s   n

We can get the characteristic roots: s1, 2   n  n  2  1

There are four cases depending on the value of 
1) 0<ξ<1.
0< Under damping case
Im
s1
s1, 2   n  j n 1   2
n  d  n 1   2
    j d  cos 1  Re

Let   n  d  n 1   2   n S-plane

2  1
s2  s1
1
 tg cos 1 
 Poles distribution
(Damped angle ) 25
 n2 1
C(s)  2 2

s  2 n s   n s
1 s  2 n
  c(t)
s ( s   n  j d )( s   n  j d )
1 s   n  n
  [  ]
s ( s   n )   d
2 2
( s   n )   d
2 2 1
   n t  n   n t 
c ( t ) 1   e cos  d t  e sin  d t 
  d  t
  
  n t 
1  e cos  d t  sin  d t  Unit step response
 1   2 
  It’s a decayed
1 oscillatory curve
1  e   n t sin( d t   )
1  2
• The system oscillation at frequency  d ,which is called the
damped oscillatory frequency and just given by the imaginary
part of the pole;
• The oscillation is exponentially decayed and the decayed rate is
controlled by the constant n ,which is called as damping
factor and just given by the real part of the pole. 26
The next fig. shows the step response as a function of the
dimensionless time  n t for various values of damping ratio
2) ξ>1. over damping case
Im
Re
s1   n   n  2  1 S2 S1

s2   n   n  2  1 Poles
distribution

Taking inverse Laplace transform

c(t)
1

t
Unit step
response
 3) ξ=1. critical damping case
Im
s1   n   n  2  1 Re

s2   n   n  2  1

Poles distribution
s1 s2   n

c( t )  L 1 C ( s )
1
  2
n 1 c(t)
L  2   1
 s  2 n s   n s 
2


1 1 1 n 
L    2 
 s s  n (s  n )  t
1  (1   n t )e   n t Unit step response
t 0 29
4) ξ≤0. Zero or negative damping case( system is unstable )

 0  1  0  1
Poles distribution:
s1,2  n  j n   1 2 s1,2 n  n 1   2
s1 s2  j n
Im Im Im

Re Re Re

c(t ) 
Unit step response: c(t )
 2  1 ) nt  2  1 ) n t
1 nt 1 e (  e( 
1  e sin( d t   ) 1 
c(t ) 1  cos  n t 1  2 2
(
2  1   1 2

  1 2
)

c(t)
1

t
30
4. The transient performance analysis of the second-order system
1
for the 0    1 case c(t ) 1  e  nt sin( d t   )
1  2
(t 0)

① Rise time tr
make : c(t ) t t 1   cos 1 
r tr 
d  d  n 1   2
getting: sin(ωd tr  θ) 0
analysis :   or  n  poles are away from the real axis  t r  .

② Peak time tp

dc(t) tp 
Make : t t p 0 d
dt
analysis : similar to tr

31
 ③ Setting time ts ④ maximum overshoot σp

Make : c(t)-c()  Δ c() c ( t )  c ( )

t t s  p% 
c ( ) t t p



1  2
 p % e 100%

analysis :   poles are away from

analysis :  and  n  poles are away
imaginary axis and close to the real axis
from the i maginary a xis  t s  .
  p%  .

Note: 2
 0.707 (  )   p % 4.3%
2
— optimal damping factor , second-order optimal syste m .

32
3.4 Time response of the higher-order systems
b0 s m  b1s m 1    bm 1
Assume : C ( s )  n n 1
R( s ) Make : r(t) 1(t)  R(s) 
a0 s  a1s    an s
m
k  ( s  zi )
k ( s  z1 )  ( s  zm ) 1 i 1 1
then : C ( s )    q r

( s  s1 )( s  s2 )  ( s  sn ) s 2 s
 ( s  s j )  ( s  2 k k   k )
j 1 k 1

bm
an q A j r Bk s  Ck q  2r n
   2
s j 1 ( s  s j ) k 1 s  2 k   k n m
Here : z i and s j  the zeros and real poles of the transfer func tion.
s k    k  k  j k 1   k2  the complex poles of the transfer func tion.
A j , Bk , C k  the constant of the partial fraction expansion of C ( s ).
We can get the unit step response( for the stable system) :
bm q s jt r C  B  
k k k  kk t
c(t )    Aje   k e sin( k 1   k2 )t
an j 1 k 1 
k 1   2
k
33
 Discuss the unit step response:
q r
b s jt C k  Bk  k  k   k  k t
We have:
c( t )  m 
an  A je   2
e sin( k 1   k2 )t
j 1 k 1  k 1   k

1) for the stable system must all poles are in the left s - plane.

2) The relevant transient portion of the response is exponentially

declined for the real pole of the left s - plane.

3) The relevant transient portion of the response is a declined

oscillation for the complex pole of the left s - plane.

4 ) The farther from the imaginary axis the pole of the left s - plane is,
the faster the relevant transient portion of the response is declined.
34
 5) The zeros of the system only affect the coefficient magnitude of
the transient portions.

6) The poles, which are the most close to the imaginary axis in the
left s - plane, are of the most influence to the transient response.

Dominant poles :
the poles  most close to the imaginary axis in the left s - plane.
Im
s3 s1
| ReS 3 | 5 Re
S 1,2     j d   n s2

• there isn’t finite zeros located relatively near the pole(s). 5 n  n
• in the left s-plan
• all other poles are located sufficient far from the
imaginary axis more than 5 times 35
 We can use the dominant poles to approximate a nth-order
system to a second-order or a first-order system.

◆ Reduction to Lower Order Systems

Example: A two-order System
C 100
( s) 
R ( s  2)( s  10)

There are two forms: -10 -2

C 100
Evans form ➡ ( s) 
R ( s  2)( s  10)
C 5
Bode form ➡ ( s) 
R (1  s / 2)(1  s / 10)
1. For the Evans form, if
C 100 100
 then Css lim 50
R s2 s 0 s  2

2. For the bode form, if

C 5 5
( s)  then Css lim 5
s  0 (1  s / 2)
R 1 s / 2
Conclusion: the Bode form must be
used in order to correctly obtain the
transfer function of the reduced-order
system.
Consider a third order system in Bode form:
C 1

R (1  s )( s 2 /  n2  2s / n  1)
If
Im

ωd

-ζωn Re
the real pole may -1/τ
be neglected