Lecture Notes

Introduction to Control Systems

Instructor: Assoc. Prof. Dr. Huynh Thai Hoang

Department of Automatic Control
Faculty of Electrical & Electronics Engineering
Ho Chi Minh City University of Technology
Email: hthoang@[Link]

© H. T. Hoang - HCMUT 1
 Chapter 5

DESIGN OF CONTINUOUS
CONTROL SYSTEMS

© H. T. Hoang - HCMUT 2
 Content
 Introduction
 Effect of controllers on system performance
 Control systems design using the root locus method
 Control systems design in the frequency domain
 Design of PID controllers
 Control systems design in state-space
 Design of state estimators

© H. T. Hoang - HCMUT 3
Introduction

© H. T. Hoang - HCMUT 4
 Introduction to design process

 Design is a process of adding/configuring hardware as well as

software in a system so that the new system satisfies the
desired specifications.

© H. T. Hoang - HCMUT 5
 Series compensator
 The controller is connected in series with the plant.

R(s) Y(s)
+ GC(s) G(s)

 Controllers: phase lead, phase lag, lead-lag compensator, P,

PD, PI, PID,…

 Design method: root locus, frequency response

© H. T. Hoang - HCMUT 6
 State feedback control
 All the states of the system are fed back to calculate the control
rule.

r(t) u(t) x(t) y(t)

+
 x (t )  Ax (t )  Bu (t ) C

 State feedback controller: u (t )  r (t )  Kx (t )

K  k1 k2  kn 

 Design method: pole placement, LQR,…

© H. T. Hoang - HCMUT 7
Effects of controller on system
performance

© H. T. Hoang - HCMUT 8
 Effects of the addition of poles
 The addition of a pole (in the left-half s-plane) to the open-
loop transfer function has the effect of pushing the root locus
to the right, tending to lower the system’s relative stability and
to slow down the settling of the response.
Im s Im s Im s

Re s Re s Re s

K K K
G (s)  G(s)  G (s) 
s  p1 ( s  p1 )( s  p2 ) ( s  p1 )( s  p2 )( s  p3)
© H. T. Hoang - HCMUT 9
 Effects of the addition of zeros

 The addition of a zero (in the left-half s-plane) to the open-

loop transfer function has the effect of pulling the root locus to
the left, tending to make the system more stable and to speed
up the settling of the response.
Im s Im s Im s

Re s Re s Re s

K K ( s  z1 )( s  z2 )
G (s)  G(s) 
( s  p1 )( s  p2 )( s  p3) ( s  p1 )( s  p2 )( s  p3)
© H. T. Hoang - HCMUT 10
 Effects of lead compensators
 Transfer function:
1  Ts
GC ( s )  K C (  1)
1  Ts
 Frequency response:
1  Tj
GC ( j )  K C
1  Tj
 Characteristics of the Bode plots:
1    1 
 max  sin  
  1
1
 max 
T 
L ( max )  20 lg K C  10 lg 
 The lead compensators improve
the transient response (POT, ts,..)
© H. T. Hoang - HCMUT 11
 Lead compensator implementation
 Lead compensator transfer function:
U ( s ) R2 R4 1  R1C1s 1  Ts
  KC (  1  R1C1  R2C2 )
E ( s ) R1 R3 1  R2C2 s 1  Ts

E(s)
U(s)

© H. T. Hoang - HCMUT 12
 Effects of lag compensators
 Transfer function:
1  Ts
GC ( s )  K C (  1)
1  Ts
 Frequency response:
1  Tj
GC ( j )  K C
1  Tj
 Characteristics of the Bode plots:
1    1 
 min  sin  
   1
1
min 
T 
L (min )  20 lg K C  10 lg 
 The lag compensators
reduce the steady-state error.
© H. T. Hoang - HCMUT 13
 Lag compensator implementation
 Lag compensator transfer function:
U ( s ) R2 R4 1  R1C1s 1  Ts
  KC (  1  R1C1  R2C2 )
E ( s ) R1 R3 1  R2C2 s 1  Ts

E(s)
U(s)

© H. T. Hoang - HCMUT 14
 Effects of lead-lag compensators
 1  1T1s   1   2T2 s 
 Transfer function: GC ( s)  KC     (1  1, 2  1)
 1  T1s   1  T2 s 
 Bode diagram

 The lead-lag compensators improve transient response and

reduces the steady-state error.
© H. T. Hoang - HCMUT 15
 Lead-lag compensator implementation
 Lead-lag compensator transfer function:
U ( s ) R6 R4 [(R1  R3 )C1s  1]( R2C2 s  1)

E ( s ) R5 R3 ( R1C1s  1)[(R2  R4 )C2 s  1]

E(s)
U(s)

© H. T. Hoang - HCMUT 16
 Effects of proportional controller (P)

 Transfer function: GC ( s)  K P
 Increasing proportional gain leads to decreasing steady-state
error, however, the system become less stable, and the POT
increases.
y(t)
 Ex: response of a
proportional control
system whose
plant has the
transfer function
below:
10
G ( s) 
( s  2)( s  3)
© H. T. Hoang - HCMUT 17
 Effects of proportional derivative controller (PD)

 Transfer function:  Bode diagram

GC ( s )  K P  K D s  K P (1  TD s )

 The PD controller is a
special case of phase lead
compensator, the
maximum phase lead is
max=900 at the frequency
max=+.
 The PD controller speed up
the response of the system,
however it also makes the
system more sensitive to
high frequency noise.
© H. T. Hoang - HCMUT 18
 Effects of proportional derivative controller (PD)
 Note: The larger the derivative constant, the faster the
response of the system.
y(t)

unompensated

© H. T. Hoang - HCMUT 19
 PD controller implementation
 PD controller transfer function:

U ( s ) R2 R4
 (1  R1C1s )  K P  K D s
E ( s ) R1 R3

E(s)
U(s)

© H. T. Hoang - HCMUT 20
 Effects of proportional integral controller (PI)
 Transfer function:  Bode diagram
KI 1
GC ( s )  K P   K P (1  )
s TI s
 The PI controller is a
special case of phase lag
compensator, the minimum
phase lag is min= 900 at
the frequency min=0.

 PI controllers eliminate
steady state error to step
input, however it can
increase POT and settling
time.

© H. T. Hoang - HCMUT 21
 Effects of proportional integral controller (PI)
 Note: The larger the integral constant, the larger the POT
of response of the system.
y(t)

uncompensated

© H. T. Hoang - HCMUT 22
 PI controller implementation
 PI controller transfer function:
U ( s ) R2 R4 R2C2 s  1 KI
  KP 
E ( s ) R1 R3 R2C2 s s

E(s)
U(s)

© H. T. Hoang - HCMUT 23
 Effects of proportional integral controller (PID)
 Transfer function:
 Bode diagram
KI
GC ( s )  K P   KDs
s
1
 GC ( s )  K P (1   TD s )
TI s
 1 
 GC ( s )  K P 1  1  TD 2 s 
 TI 1s 

 Effects of PID controllers:

 speed up response of
the system
 Eliminate steady-state
error to step input.

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 24

 PID controller implementation
 PID controller transfer function:
U ( s ) R2 R4 ( R1C1s  1)( R2C2 s  1)

E ( s ) R1 R3 R2C2 s

E(s)
U(s)

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 25

Comparison of PI, PD and PID controllers

y(t)

Uncompensated

© H. T. Hoang - HCMUT 26
 Control systems design
using the root locus method

© H. T. Hoang - HCMUT 27
 Procedure for designing lead compensator using the root locus

s  (1 / T )
Lead compensator: GC ( s )  K C (  1)
s  (1 / T )
*
 Step 1: Determine the dominant poles 1, 2 from desired
s
transient response specification:
Overshoot (POT) 
    s1*, 2  n  jn 1   2
Settling time ts n
 Step 2: Determine the deficiency angle so that the dominant
poles s1*, 2 lie on the root locus of the compensated system:
n m
 *  180 0   arg( s1*  pi )   arg( s1*  z i )
i 1 i 1

where pi and zi are poles & zeros of G(s) before compensation.

 *  180 0   angle from pi to s1*   angle from zi to s1*
© H. T. Hoang - HCMUT 28
 Procedure for designing lead compensator using the root locus

 Step 3: Determine the pole & zero of the lead compensator

*
Draw 2 arbitrarily rays starting from the dominant pole s1
such that the angle between the two rays equal to *. The
intersection between the two rays and the real axis are the
positions of the pole and the zero of the lead compensator.
Two methods often used for drawing the rays:
 Bisector method
 Pole elimination method

 Step 4: Calculate the gain KC using the formula:

GC ( s )G ( s ) s s*  1
1

© H. T. Hoang - HCMUT 29
 Example of designing a lead compensator using RL

R(s) 50 Y(s)
+ GC(s)
s ( s  5)

 Objective: design the compensator GC(s) so that the

response of the compensated system satisfies: POT<20%;
ts< 0,5sec (2% criterion).

 Solution:
 Because the design objective is to improve the transient
response, we need to design a lead compensator:
s  (1 / T )
GC ( s )  K C (  1)
s  (1 / T )
© H. T. Hoang - HCMUT 30
 Example of designing a lead compensator using RL (cont’)

 Step 1: Determine the dominant poles:

   
POT  exp    0.2    ln 0.2  1,6    0,45
 1 2  1 2
 
Chose   0,707

4 4
t qñ   0,5  n   n  11,4
 n 0,5  
Chose  n  15
The dominant poles are:
s1*, 2   n  j n 1   2  0,707  15  j15 1  0,707 2

s1*, 2  10,5  j10,5

© H. T. Hoang - HCMUT 31
 Example of designing a lead compensator using RL (cont’)

 Step 2: Determine the deficiency angle:

Method 1:
 *  1800  arg[(10,5  j10,5)  0]  arg[(10,5  j10,5)  (5)]
0   10,5   10,5 
 180  arctan   arctan 
   10,5    5,5 
 1800  (135  117,6)
Im s
   72,6
* 0
s*
j10,5
Method 2:
 *  1800  ( 1   2 )
 1800  (1350  117,60 ) 2 1 Re s
O
   72,6
* 0 10,5 5

© H. T. Hoang - HCMUT 32
 Example of designing a lead compensator using RL (cont’)

 Step 3: Determine the pole and the zero of the compensator

(bisector method) Im s
x s* P
j10,5

B A C 1 Re s
O
10,5 5

 OPˆ x  *   OPˆ x  * 
sin    sin   
 2 2  2 2
OB  OP  28,12 OC  OP  8,0
 OPˆ x  
*
 OPˆ x  
*
sin    sin   
 2 2  2 2
s8
 GC ( s )  K C
s  28
© H. T. Hoang - HCMUT 33
 Example of designing a lead compensator using RL (cont’)

 Step 4: Determine the gain of the compensator:

GC ( s )G ( s ) s  s*  1

 10,5  j10,5  8 50
 KC . 1
 10,5  j10,5  28 (10,5  j10,5)(10,5  j10,5  5)

10,79  50
 KC 1
20,41  15  11,85
 K C  6,7
 Conclusion: The transfer function of the lead compensator is:
s8
GC ( s )  6,7
s  28
© H. T. Hoang - HCMUT 34
Root locus of the system

Root locus of the Root locus of the

uncompensated system compensated system

© H. T. Hoang - HCMUT 35
 Transient response of the system

y(t)

uncompensated
compensated

Transient response of the system

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 36

 Exercise

R(s) Y(s)
+ GC(s) G(s)

20
G(s) =
s( s + 3)
 Objective: Design the phase lead compensator GC(s) so that
the compensated system has a pair of complex poles with
  0.8 and n  4
Calculate the POT and ts before and after compensation.
Simulate the system to verify the design.

© H. T. Hoang - HCMUT 37
 Exercise

R(s) Y(s)
+ GC(s) G(s)

5
G(s) =
s( s + 2)( s + 8)
 Objective: Design the phase lead compensator GC(s) so that
the compensated system has a pair of complex poles with
  0.8 and n  3
 Solution:

© H. T. Hoang - HCMUT 38
 Procedure for designing lag compensator using the root locus

s  (1 / T )
Lag compensator: GC ( s)  KC (   1)
s  (1 / T )
 Step 1: Determine  to meet the steady-state error requirement:
KP KV Ka
 * or  * or  *
KP KV Ka

1 *
 Step 2: Chose the zero of the lag compensator:  Re( s1, 2 )
T

1 1
 Step 3: Calculate the pole of the compensator:  .
T T

 Step 4: Calculate KC satisfying the condition: GC ( s)G ( s) s  s*  1

1, 2

© H. T. Hoang - HCMUT 39
 Example of designing a lag compensator using RL

R(s) 10 Y(s)
+ GC(s)
s( s  3)( s  4)

 Objective: design the compensator GC(s) so that the

compensated system satisfies the following performances:
steady state error to ramp input is 0,02 and transient
response of the compensated system is nearly unchanged.

 Solution:
 The compensator to be design is a lag compensator:
s  (1 / T )
GC ( s)  KC (   1)
s  (1 / T )
© H. T. Hoang - HCMUT 40
 Example of designing a lag compensator using RL (cont’)

 Step 1: Determine 
The velocity constant of uncompensated system :
10
KV  lim sG ( s)  lim s  0.83
s 0 s 0 s ( s  3)( s  4)

The desired velocity constant:

* 1 1
KV  *   50
exl 0,02
KV 0.83
Then:  * 
KV 50

  0,017

© H. T. Hoang - HCMUT 41
 Example of designing a lag compensator using RL (cont’)

 Step 2: Chose the zero of the lag compensator

The pole of the uncompensated system:
10 s1, 2  1  j
1  G (s)  0  1 0  
s ( s  3)( s  4) s3  5
 The dominant poles of the uncompensated system: s1, 2  1  j
1 1
Chose:  Res1   1   0,1
T T
 Step 3: Calculate the pole of the compensator:
1 1 1
  (0,017)(0,1)   0,0017
T T T
s  0,1
 GC ( s)  K C
s  0,0017
© H. T. Hoang - HCMUT 42
 Example of designing a lag compensator using RL (cont’)

 Step 4: Determine the gain of the compensator

GC ( s )G ( s ) s s*  1

s  0,1 10
 KC . 1
s  0,0017 s( s  3)( s  4) s  1 j

( 1  j  0,1) 10
 KC . 1
( 1  j  0,0017) ( 1  j )( 1  j  3)( 1  j  4)

KC  1,0042  1

s  0,1
 GC ( s ) 
s  0,0017

© H. T. Hoang - HCMUT 43
Root locus of the system

Root locus of the Root locus of the

uncompensated system compensated system
© H. T. Hoang - HCMUT 44
Transient response of the system

y(t)

uncompensated
compensated

Transient response of the system

© H. T. Hoang - HCMUT 45
 Exercise

R(s) Y(s)
+ GC(s) G(s)

12
G(s) =
s( s + 2)( s + 8)
 Objective: Design the phase lag compensator GC(s) so that
the system after compensation has steady error to ramp input
satisfying ess  0.05
 .
Solution:

© H. T. Hoang - HCMUT 46
Exercise

© H. T. Hoang - HCMUT 47
 Procedure for designing lead lag compensator using the RL

The compensator to be designed

GC ( s)  GC1 ( s)GC 2 ( s)

phase phase
lead lag
 Step 1: Design the lead compensator GC1(s) to satisfy the
transient response performances.

 Step 2: Let G1(s)= G (s). GC1(s)

Design the lag compensator GC2(s) in series with G1(s) to
satisfy the steady-state performances (and not to degrade the
transient response obtained after phase lead compensating)

© H. T. Hoang - HCMUT 48
 Example of designing a lead lag compensator using RL

R(s) 4 Y(s)
+ GC(s)
s( s  0.5)

 Objective: design the compensator GC(s) so that the

compensated system has the dominant poles with  = 0.5,
n =5 (rad/sec) and the velocity constant KV =80.

 Solution
 The compensator to be designed is a lead lag compensator
because the design objective is to improve the transient
response and to reduce the steady-state error.
GC ( s)  GC1 ( s)GC 2 ( s)

© H. T. Hoang - HCMUT 49
 Example of designing a lead lag compensator using RL (cont’)

 Step 1: Design the lead compensator GC1(s)

The dominant poles:
s1*, 2  n  jn 1   2  0,5  5  j 5 1  0,52

s1*, 2  2,5  j 4,33

The deficiency angle:

 *  180 0  ( 1   2 )
 180 0  (120 0  1150 )
 *  550

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 50

 Example of designing a lead lag compensator using RL (cont’)

Chose the zero of the lead compensator so that it eliminates

the pole at –0.5 of G(s) (pole elimination method)
1
 0,5
T1
OA  0,5
sin APˆ B sin 550
AB  PA  4.76 0
 4.5
sin PAB sin 60
B A 1
 OA  AB  5
T1
–1/T1 –1/T1
s  0,5
GC1 ( s )  K C1
s5

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 51

 Example of designing a lead lag compensator using RL (cont’)

Calculate KC1: GC1 ( s )G ( s ) s  s*  1

s  0,5 4
K C1 . 1
s  5 s ( s  0,5) s  2,5  j 4,33

K C1  6,25

s  0,5
 GC1 ( s )  6,25
s5

The lead-compensated open-loop system:

25
G1 ( s )  GC1 ( s )G ( s ) 
s ( s  5)

28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 52

 Example of designing a lead lag compensator using RL (cont’)

 Step 2: Design the lag compensator GC2(s)

1
s
T2
GC 2 ( s )  K C 2
1
s
T2
 Determine :
25
KV  lim sG1 ( s )  lim s 5
s 0 s  0 s ( s  5)

KV*  80

KV 5 1
  *  
KV 80 16

© H. T. Hoang - HCMUT 53
 Example of designing a lead lag compensator using RL (cont’)

 Determine the zero of the lag compensator:

1
 Re( s * )  Re(2,5  j 4,33)  2,5
T2
1
Chose:  0,16
T2

 Calculate the pole of the lag compensator:

1 1 1
 .  .(0,16)
T2 T2 16
1
  0.01
T2

© H. T. Hoang - HCMUT 54
 Example of designing a lead lag compensator using RL (cont’)

 Calculate KC2 using the gain condition: GC 2 ( s )G1 ( s ) s  s*  1

 GC 2 (s) s s G1 (s) s s   1
 *
 *

 2,5  j 4,33  0,16

 KC 2 1
 2,5  j 4,33  0,01
 K C 2  1.01

The transfer function of the lag compensator:

( s  0,16)
GC 2 ( s )  1,01
( s  0,01)

( s  0,5)( s  0,16)
Final result: GC ( s )  GC1 ( s )GC 2 ( s )  6,31
( s  5)( s  0,01)
© H. T. Hoang - HCMUT 55
 Exercise

R(s) 10 Y(s)
+ GC(s)
s ( s  2)

 Objective: design the compensator GC(s) so that the

compensated system has the POT < 4%, ts<1sec and
ess=0.02 (ramp input).

© H. T. Hoang - HCMUT 56
Control system design in
frequency domain

© H. T. Hoang - HCMUT 57
 Procedure for designing lead compensators in frequency domain

Ts  1
The lead compensator: GC ( s )  KC (  1)
Ts  1
 Step 1: Determine KC to meet the steady-state error requirement:
K C  K P* / K P or K C  KV* / KV or K C  K a* / K a
 Step 2: Let G1(s)=KCG(s). Plot the Bode diagram of G1(s)
 Step 3: Determine the gain crossover frequency of G1(s):
L1 (C )  0 or G1 ( j C )  1
 Step 4: Determine the phase margin of G1(s) (phase margin
of uncompensated system): M  180  1 ( C )
 Step 5: Determine the necessary phase lead angle to be
added to the system:   M *  M  
max
M * is the desired phase margin,   50  200
© H. T. Hoang - HCMUT 58
 Procedure for designing lead compensators in frequency domain

1  sin  max
 Step 6: Calculate  :  
1  sin  max
 Step 7: Determine the new gain crossover frequency (of
the compensated open-loop system) using the conditions:
L1(C* )  10lg  or G1 ( jC* )  1 / 

1
 Step 8: Calculate the time constant T: T 
C* 
 Step 9: Check if the compensated system satisfies the gain
margin? If not, repeat the design procedure from step 5.
 Note: It is possible to determine C (step 3), M (step 4) and
*C (step 7) by using Bode diagram instead of using analytic
calculation.
© H. T. Hoang - HCMUT 59
 Design lead compensator in frequency domain - Example

R(s)
GC(s) 4 Y(s)
+
s( s  2)

 Objective: Design the compensator GC(s) so that the

compensated system satisfies the performances:
KV*  20; M *  50 0 ; GM *  10dB
 Solution:
 The transfer function of the lead compensator to be designed:
1  Ts
GC ( s )  K C (  1)
1  Ts

© H. T. Hoang - HCMUT 60
 Design lead compensator in frequency domain – Example (cont’)

 Step 1: Determine KC
The velocity constant of the uncompensated system:
4
KV  lim sG ( s )  lim s 2
s 0 s 0 s ( s  2)
The desired velocity constant: KV*  20

 KV* 20 
KC   K C  10
KV 2
4
 Step 2: Denote G1 ( s )  K C G ( s )  10.
s ( s  2)
20
 G1 ( s ) 
s (0,5s  1)
Draw the Bode diagram of G1(s)
© H. T. Hoang - HCMUT 61
Design lead compensator in frequency domain – Example (cont’)

-20dB/dec
26

-40dB/dec

2 c=6

M

-160

© H. T. Hoang - HCMUT 62
 Design lead compensator in frequency domain – Example (cont’)

 Step 3: The gain crossover frequency of G1(s)

According to the Bode diagram: C  6 (rad/sec)

 Step 4: The phase margin of G1(s)

According to the Bode diagram:
1 (C )  1600
 M  180  1 (C )  200

 Step 5: The necessary phase lead angle to be added:

 max  M *  M   (chose =7)
  max  500  200  70

  max  37 0

© H. T. Hoang - HCMUT 63
 Design lead compensator in frequency domain – Example (cont’)

 Step 6: Calculate 
1  sin  max 1  sin 37 0
    4
1  sin  max 1  sin 37 0
 Step 7: Determine the new gain crossover frequency using
Bode plot L (  )  10 lg   10 lg 4  6dB
1 C

The abscissa of the intersection between Bode magnitude

diagram and the horizontal line with ordinate of 6dB is the
new gain crossover frequency. According to the plot (in slide
54), we have:
C  9 (rad/sec)
 Step 8: Calculate T
1 1
T   T  0,056  T  0,224
C  (9)( 4 )
© H. T. Hoang - HCMUT 64
Design lead compensator in frequency domain – Example (cont’)

-20dB/dec
-40dB/dec
+20dB/dec
-20dB/dec
-6
-40dB/dec

-40dB/dec

1/T=4.5 c=6  ’c=9 1/T=18

M *
M

-160

© H. T. Hoang - HCMUT 65
 Design lead compensator in frequency domain – Example (cont’)

 Step 9: Check the gain margin of the compensated system

According to the compensated Bode diagram, GM* = +, then
the compensated system fulfills the design requirements.

 Conclusion: The designed lead compensator is:

1  0,224 s
GC ( s )  10
1  0,056s

© H. T. Hoang - HCMUT 66
 Design lead compensator in frequency domain – Example 2

R(s) Y(s)
+ GC(s) G(s)

20e-0.01s
G(s) =
s( s + 4)

 Objective: Design the compensator GC(s) so that the

compensated system has: FM * ³ 600 ; GM *  10dB and
steady-state error to unit ramp input ess*  0.05;

 Solution:

© H. T. Hoang - HCMUT 67
 Design lead compensator in frequency domain – Example 3

R(s) Y(s)
+ GC(s) G(s)

16e 0.005 s
G(s) 
( s  2)( s  5)
 Objective: Design the compensator GC(s) so that the
compensated system has:  M *  600 , GM *  10dB and
steady-state error to unit step input ess*  0.05;

 Solution:

© H. T. Hoang - HCMUT 68
 Procedure for designing lag compensators in frequency domain

Ts  1
The lag compensator: GC ( s )  K C (  1)
Ts  1
 Step 1: Determine KC to meet the steady-state error requirement:
K C  K P* / K P or K C  KV* / KV or K C  K a* / K a
 Step 2: Let G1(s)=KCG(s). Plot the Bode diagram of G1(s)
 Step 3: Determine the new gain crossover frequency C*
satisfying the following condition:
1 (C* )  180  M *  
*
M is the desired phase margin,   50  20 0
 Step 4: Calculate  using the condition:
1
L1 ( )  20lg 
* or G1 ( j ) 
*
C C

© H. T. Hoang - HCMUT 69
 Procedure for designing lag compensators in frequency domain

 Step 5 : Chose the zero of the lag compensator so that:

1
 C*  T
T
 Step 6: Calculate the time constant T:
1 1 
 T
T T
 Step 7: Check if the compensated system satisfies the gain
margin? If not, repeat the design procedure from step 3.

  (
Note: It is possible to determine 1 C , C (step 3), L1 (C* )
*
) *

(step 4) by using Bode diagram instead of using analytic

calculation.

© H. T. Hoang - HCMUT 70
 Design lag compensator in frequency domain – Example

R(s)
GC(s) 1 Y(s)
+
 s( s  1)(0.5s  1)

 Objective: design the lag compensator GC(s) so that that

compensated system satisfies the following performances:
KV*  5; M *  400 ; GM *  10dB
 Solution
 The transfer function of the lag compensator to be designed:
1  Ts (  1)
GC ( s )  K C
1  Ts

© H. T. Hoang - HCMUT 71
 Design lag compensator in frequency domain – Example (cont’)

 Step 1: Determine KC
The velocity constant of the uncompensated system:
1
KV  lim sG ( s )  lim s 1
s 0 s 0 s( s  1)(0.5s  1)
The desired velocity constant: KV*  5
KV*
 KC  5
KV
 Step 2: Denote G1 ( s )  K C G ( s )
5
 G1 ( s ) 
s ( s  1)(0.5s  1)
Draw the Bode diagram of G1(s)
© H. T. Hoang - HCMUT 72
Design lag compensator in frequency domain – Example (cont’)

-20dB/dec

14 -40dB/dec

-60dB/dec

1 2

© H. T. Hoang - HCMUT 73
 Design lag compensator in frequency domain – Example (cont’)

 Step 3: Determine the new gain crossover frequency:

1 (C )  1800  M *  
 1 (C )  1800  400  50
 1 ( C )  1350
According to the Bode diagram:  C  0.5 (rad/sec)
 Step 4: Calculate  using the condition:
L1 ( C )  20 lg 
According the Bode diagram:L1 ( C )  18 (dB)
 18  20 lg   lg   0,9    10 0,9
   0,126
© H. T. Hoang - HCMUT 74
 Design lag compensator in frequency domain – Example (cont’)

 Step 5: Chose the zero of the lag compensator:

1
 C  0.5
T
1
Chose  0.05  T  20
T
 Step 6: Calculate the time constant T
1 1
  0,126  0,05  0,0063  T  159
T T
 Step 7: It can be verified in the Bode diagram that the
compensated system satisfies the gain margin requirement.
(20 s  1)
Conclusion GC ( s )  5
(159 s  1)
© H. T. Hoang - HCMUT 75
 Design lag compensator in frequency domain – Example (cont’)

-20dB/dec

L1(’c)
14 -40dB/dec GM*

L’(’)
-60dB/dec

0.0067 0.05 ’c=0.5 1’ 2

-135

© H. T. Hoang - HCMUT 76
 Design lag compensator in frequency domain – Example 2

R(s) Y(s)
+ GC(s) G(s)

20e-0.2 s
G(s) =
s( s + 4)

 Objective: Design the compensator GC(s) so that the

compensated system has: FM * ³ 600 ; GM *  10dB and
steady-state error to unit ramp input ess*  0.05;

 Solution:

© H. T. Hoang - HCMUT 77
 Design lag compensator in frequency domain – Example 3

R(s) Y(s)
+ GC(s) G(s)

16e 0.01s
G(s) 
( s  2)( s  5) 2
 Objective: Design the compensator GC(s) so that the
compensated system has:  M *  600 , GM *  10dB and
steady-state error to unit step input ess*  0.05;

 Solution:

© H. T. Hoang - HCMUT 78
Comparison of phase lead and phase lag compensator

(Dorf and Bishop (2008), Modern control system –p.729)

© H. T. Hoang - HCMUT 79
Design of PID controllers

© H. T. Hoang - HCMUT 80
 Zeigler  Nichols method 1
 Determine the PID parameters based on the step response of
the open-loop system.

u(t) y(t)
Plant
y(t)

T1 T2
© H. T. Hoang - HCMUT 81
Zeigler  Nichols method 1 (cont’)

R(s) Y(s)
+ PID Plant

 1 
PID controller: GC ( s )  K P 1   TD s 
 TI s 

Controller KP TI TD
P T2/(T1K)  0
PI 0.9T2/(T1K) 3.3T1 0
PID 1.2T2/(T1K) 2T1 0.5T1

© H. T. Hoang - HCMUT 82
 Zeigler  Nichols method 1 – Example
 Problem: Design a PID y(t)
controller to control a furnace
providing the open-loop
characteristic of the furnace 150
obtained from a experiment
beside. t (min)
K  150
8 24
T1  8 min  480 sec
T2  24 min  1440 sec
T2 1440  1 
K P  1.2  1.2  0.024 GPID ( s )  0.0241   240s 
T1K 480  150  960s 
TI  2T1  2  480  960 sec

TD  0.5T1  0.5  480  240 sec

© H. T. Hoang - HCMUT 83
 Zeigler  Nichols method 2
 Determine the PID parameters based on the response of the
closed-loop system at the stability boundary.

+ KKcr Plant

y(t)

Tcr
t

© H. T. Hoang - HCMUT 84
Zeigler  Nichols method 2 (cont’)

R(s) Y(s)
+ PID Plant

 1 
PID controller: GC ( s )  K P 1   TD s 
 TI s 

Controller KP TI TD
P 0.5Kcr  0
PI 0.45Kcr 0.83Tcr 0
PID 0.6Kcr 0.5Tcr 0.125Tcr

© H. T. Hoang - HCMUT 85
 Zeigler  Nichols method 2 – Example
 Problem: Design a PID controller to control the angle position
of a DC motor, providing that by experiment the critical gain of
the system is 20 and the critical cycle is T= 1 sec.
 Solution:
 According to the given data:
K cr  20
Tcr  1 sec
 Applying Zeigler – Nichols method 2:
K P  0.6 K cr  0.6  20  12
 1 
GPID ( s )  121   0.5s 
TI  0.5Tcr  0.5  1  0.5 sec  0.125s 
TD  0.125Tcr  0.125  1  0.125 sec

© H. T. Hoang - HCMUT 86
 Analytical method for designing PID controller

 Step 1: Establish equation(s) representing the relationship

between the controller to be designed and the desired
performances.

 Step 2: Solve the equation(s) obtained in step 1 for the

parameter(s) of the controller.

© H. T. Hoang - HCMUT 87
 Analytical method for designing PID controller

 Example: Design PID controller so that the control system

satisfies the following requirements:
 Closed-loop complex poles with =0.5 and n=8.
 Velocity constant KV = 100.
R(s) 100 Y(s)
+ GPID(s)
s 2  10s  100

 Solution: The transfer function of the PID controller to be

designed
KI
GC ( s )  K P   KDs
s
© H. T. Hoang - HCMUT 88
 Analytical method for designing PID controller (cont’)

 Velocity constant of the controlled system:

 K  100 
KV  lim sGC ( s )G ( s )  lim s K P  I  K D s  2 
s 0 s 0  s  s  10 s  100 
 KV  K I
According to the design requirement: KV = 100
 K I  100

 The characteristic equation of the controlled system:

 KI  100 
1  KP   K D s  2 0
 s  s  10 s  100 
 s 3  (10  100 K D ) s 2  (100  100 K P ) s  100 K I  0 (1)

© H. T. Hoang - HCMUT 89
 Analytical method for designing PID controller (cont’)

 The desired characteristic equation:

( s  a )( s 2  2 n s   n2 )  0
 ( s  a )( s 2  8s  64)  0
 s 3  (a  8) s 2  (8a  64) s  64a  0 (2)

 Balancing the coefficients of the equations (1) and (2), we have:

10  100 K D  a  8 a  156.25

 
100  100 K P  8a  64   K P  12,14
100 K  64a  K  1,54
 I  D

100
Conclusion: GC ( s )  12,64   1,54 s
s
© H. T. Hoang - HCMUT 90
 Exercise

 Design PI controller so that the control system satisfies the

following requirements:
 Closed-loop complex poles with =0.6 and n=8.

R(s) 100 Y(s)

+ GPI(s)
s 2  10s  100

 Solution: The transfer function of the PID controller to be

designed
KI
GC ( s )  K P   KDs
s
© H. T. Hoang - HCMUT 91
 Manual tuning of PID controllers

 Effect of increasing a parameter of PID controller independently

on closed-loop performance:

Steady-
Para- Settling
Rise time POT state Stability
meter time
error
Small
KP Decrease Increase Decrease Degrade
change
KI Decrease Increase Increase Eliminate Degrade
Minor Improve if
KD Decrease Decrease No effect
change KD small

© H. T. Hoang - HCMUT 92
 Manual tuning of PID controllers (cont.)
A procedure for manual tuning of PID controllers:
1. Set KI and KD to 0, gradually increase KP to the critical gain
Kcr (i.e. the gain makes the closed-loop system oscilate)
2. Set KP Kcr /2
3. Gradually increase KI until the steady-state error is
eliminated in a sufficient time for the process (Note that too
much KI will cause instability).
4. Increase KD if needed to reduce POT and settling time (Note
that too much KD will cause excessive response and
overshoot)
5. Fine tuning of Kp, Ki and Kd to meet the desired
perfomances

© H. T. Hoang - HCMUT 93
Control systems design in state-space
using pole placement method

© H. T. Hoang - HCMUT 94
 Controllability
 x (t )  Ax(t )  Bu (t )
 Consider a system: 
 y (t )  Cx(t )
 The system is complete state controllable if there exists an
unconstrained control law u(t) that can drive the system from
an initial state x(t0) to a arbitrarily final state x(tf) in a finite
time interval t0  t tf . Qualitatively, the system is state
controllable if each state variable can be influenced by the
input.

y(t)

Signal flow graph of an incomplete state controllable system

© H. T. Hoang - HCMUT 95
 Controllability condition

 System:
 x (t )  Ax(t )  Bu (t )

 y (t )  Cx (t )
 Controllability matrix

C  [ B AB A2 B  An1 B]

 The necessary and sufficient condition for the controllability is:

rank ( C )  n

 Note: we use the term “controllable” instead of “complete

state controllable” for short.

© H. T. Hoang - HCMUT 96
 Controllability – Example
 x (t )  Ax(t )  Bu (t )
 Consider a system 
 y (t )  Cx(t )
where:  0 1 5 
A  B  C  1 3
 2  3  2
Evaluate the controllability of the system.

 Solution: Controllability matrix:

5 2 
C  B AB   C  
 2  16 
 Because:
det( C )  84  rank ( C )  2
 The system is controllable

© H. T. Hoang - HCMUT 97
 State feedback control
r(t) u(t) x(t) y(t)
+
 x (t )  Ax (t )  Bu (t ) C

 Consider a system described by the state equations:

 x (t )  Ax(t )  Bu (t )

 y (t )  Cx(t )
 The state feedback controller: u (t )  r (t )  Kx (t )
 The state equations of the closed-loop system:
 x (t )  [ A  BK ] x(t )  Br (t )

 y(t )  Cx(t )
© H. T. Hoang - HCMUT 98
 Pole placement method
If the system is controllable, then it is possible to determine
the feedback gain K so that the closed-loop system has the
poles at any location.
 Step 1: Write the characteristic equation of the closed-loop
system
det[ sI  A  BK ]  0 (1)

 Step 2: Write the desired characteristic equation:

n
 ( s  pi )  0 (2)
i 1
pi , (i  1, n) are the desired poles
 Step 3: Balance the coefficients of the equations (1) and (2),
we can find the state feedback gain K.
© H. T. Hoang - HCMUT 99
 Pole placement method – Example
 Problem: Given a system described by the state-state
equation:
 x (t )  Ax(t )  Bu (t )

 y (t )  Cx(t )

0 1 0 0 

A 0 0 1  B   3  C  0 0 1
   
 4  7  3 1
 Determine the state feedback controller u (t )  r (t )  Kx(t )
so that the closed-loop system has complex poles with
  0,6;n  10 and the third pole at 20.

© H. T. Hoang - HCMUT 100

 Pole placement method – Example (cont’)
 Solution
 The characteristic equation of the closed-loop system:
det[ sI  A  BK ]  0
 1 0 0  0 1 0  0  
  
 det s 0 1 0   0 0 1   3k1 k2 k3   0
     
 0 0 1  4  7  3 1 
       
 s 3  (3  3k 2  k3 ) s 2  (7  3k1  10k 2  21k3 ) s  (4  10k1  12k3 )  0 (1)

 The desired characteristic equation:

( s  20)( s 2  2 n s   n2 )  0
 s 3  32 s 2  340 s  2000  0 (2)

© H. T. Hoang - HCMUT 101

 Pole placement method – Example (cont’)

 Balance the coefficients of the equations (1) and (2), we have:

3  3k 2  k3  32

7  3k1  10k 2  21k3  340
4  10k  12k  2000
 1 2

 Solve the above set of equations, we have:

k1  220,578

k 2  3,839
k  17,482
 3

 Conclusion: K  220,578 3,839 17,482 

© H. T. Hoang - HCMUT 102

 Exercise
 Problem: Given a system described by the state-state
equation:
 x (t )  Ax(t )  Bu (t )

 y (t )  Cx(t )
0 1 0
A  B  C  1 0
 3 1 2

 Verify that the system is controllable

 Determine the state feedback controller u (t )  r (t )  Kx(t )
so that the closed-loop system has complex poles with
  0.8, n  4

© H. T. Hoang - HCMUT 103

Exercise

© H. T. Hoang - HCMUT 104

Exercise

© H. T. Hoang - HCMUT 105

Exercise P11.16 (Modern Control Systems)

© H. T. Hoang - HCMUT 106

Exercise P11.14 (Modern Control Systems)

© H. T. Hoang - HCMUT 107

Design of state estimators

© H. T. Hoang - HCMUT 108

 The concept of state estimation
 To be able to implement state feedback control system, it is
required to measure all the states of the system.
 However, in some applications, we can only measure the
output, but cannot measure the states of the system.
 The problem is to estimate the states of the system from the
output measurement.
 State estimator (or state observer)

© H. T. Hoang - HCMUT 109

 Observability

 x (t )  Ax(t )  Bu (t )
 Consider a system: 
 y (t )  Cx(t )
 The system is complete state observable if given the control
law u(t) and the output signal y(t) in a finite time interval
t0  t tf , it is possible to determine the initial states x(t0).
Qualitatively, the system is state observable if all state variable
x(t) influences the output y(t).

y(t)

Signal flow graph of an incomplete state observable system

© H. T. Hoang - HCMUT 110
 Observability condition

 System  x (t )  Ax(t )  Bu (t )

 y (t )  Cx(t )
It is necessary to estimate the state xˆ (t ) from mathematical
model of the system and the input-output data.

 Observability matrix:  C 
 CA 
 
O   CA 
2

 
  
CAn 1 

 The necessary and sufficient condition for the observability is:

rank (O )  n
© H. T. Hoang - HCMUT 111
 Observability – Example
 x (t )  Ax(t )  Bu (t )
 Consider the system 
 y (t )  Cx(t )
0 1 1 
where: A  B  C  1 3
 2  3  2
Evaluate the observability of the system.
 Solution: Observability matrix:
C  1 3
O    O  
CA   6  8 
 Because det(O )  10  rank (O )  2
 The system is observable

© H. T. Hoang - HCMUT 112

 State estimator
r(t) u(t) x(t) y(t)
x (t )  Ax (t )  Bu (t )
 C

+
L 
xˆ (t )
B ++
+  C
yˆ (t )

 xˆ (t )  Axˆ (t )  Bu (t )  L( y (t )  yˆ (t ))
 State estimator: 
 yˆ (t )  Cxˆ (t )
L  [l1 l2  ln ]
T
where:
28 November 2022 © H. T. Hoàng - [Link]/~hthoang/ 113
 Design of state estimators
 Requirements:
 The state estimator must be stable, estimation error
should approach to zero.
 Dynamic response of the state estimator should be fast
enough in comparison with the dynamic response of the
control loop.
 It is required to chose L satisfying:
 All the roots of the equation det( sI  A  LC )  0 locates
in the half-left s-plane.
 The roots of the equation det( sI  A  LC )  0 are further
from the imaginary axis than the roots of the equation
det( sI  A  BK )  0
 Depending on the design of L, we have different state estimator:
 Luenberger state observer
 Kalman filter
© H. T. Hoang - HCMUT 114
 Procedure for designing the Luenberger state observer

 Step 1: Write the characteristic equation of the state observer

det[ sI  A  LC ]  0 (1)
 Step 1: Write the desired characteristic equation:
n
(2)
 ( s  pi )  0
i 1
pi , (i  1, n) are the desired poles of the state estimator
 Step 3: Balance the coefficients of the characteristic
equations (1) and (2), we can find the gain L.

© H. T. Hoang - HCMUT 115

 Design of state estimators – Example
 Problem: Given a system described by the state equation:
 x (t )  Ax(t )  Bu(t )

 y (t )  Cx(t )

0 1 0 0 

A 0 0 1  B  3 C  1 0 0
   
 4  7  3 1
 Assuming that the states of the system cannot be directly
measured. Design the Luenberger state estimator so that the
poles of the state estimator lying at 20, 20 and 50.

© H. T. Hoang - HCMUT 116

 Design of state estimators – Example (cont’)
 Solution
 The characteristic equation of the Luenberger state estimator:
det[ sI  A  LC ]  0
 1 0 0  0 1 0   l1  
       
 det s 0 1 0   0 0 1   l2 1 0 0  0
 0 0 1  4  7  3 l  
      3 
 s 3  (l1  3) s 2  (3l1  l2  7) s  (7l1  5l2  l3  4)  0 (1)

 The desired characteristic equation:

( s  20) 2 ( s  50)  0
 s 3  90 s 2  2400s  20000  0 (2)

© H. T. Hoang - HCMUT 117

 Design of state estimators – Example (cont’)

 Balancing the coefficients of the equ. (1) and (2) leads to:
l1  3  90

3l1  l2  7  2400
7l  3l  l  4  20000
 1 2 3

 Solve the above set of equations, we have:

l1  87

l2  2132
l  12991
3

Conclusion L  87 2132 12991

T


© H. T. Hoang - HCMUT 118

End of Chapter 5

© H. T. Hoang - HCMUT 119