Lecture Notes

Introduction of Control Systems

Course coordinator: Assoc. Prof. Dr. Huynh Thai Hoang

Department of Automatic Control (DAC)
Faculty of Electrical & Electronics Engineering
Ho Chi Minh City University of Technology
Email: hthoang@hcmut.edu.vn

Instructor: Lecturers of Dept

Dept. of Automatic Control

10 January 2016 © H. T. Hoang - www4.hcmut.edu.vn/~hthoang/ 1

 Chapter
p 3

SYSTEM STABILITY ANALYSIS

10 January 2016 © H. T. Hoàng - ÐHBK TPHCM 2

 Content
 Stability
St bilit conceptt
 Algebraic stability criteria
 Necessary
N condition
diti
 Routh’s criterion
 Hurwitz’s criterion
 Root locus method
 Root locus definition
 Rules for drawing root loci
 Stability analysis using root locus
 Frequency response analysis
 Frequency response
 Bode criterion
 Nyquist’s stability criterion
10 January 2016 © H. T. Hoàng - ÐHBK TPHCM 3
 Stability concept

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 4

 BIBO stability
 A system
t i defined
is d fi d tot beb BIBO stablet bl if every b
bounded
d d input
i t
to the system results in a bounded output over the time
interval [t0,+∞)
+∞) for all initial times t0.
u(t) y(t)
System

y(t) y(t) y(t)

Stable system System at Unstable

stability boundary system
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 5
 Poles and zeros
 Consider a system described by the transfer function (TF):
Y ( s ) b0 s m  b1s m 1    bm 1s  bm
G(s)  
U ( s ) a0 s n  a1s n 1    an 1s  an

 Denote: A( s )  a0 s n  a1s n1    an1s  an (TF’s denominator)

B ( s )  b0 s m  b1s m1    bm1s  bm (TF

(TF’ numerator)

 Poles: are the roots of the denominator of the transfer

f
function,
i i the
i.e. h roots off the
h equation
i A(s)
A( ) = 0.
0 Since
Si A( ) is
A(s) i
of order n, the system has n poles denoted as pi , i =1,2,…n.
 Zeros: are the roots of the numerator of the transfer function,
i.e. the roots of the equation B(s) = 0. Since B(s) is of order m,
the system has m zeros denoted as zi, i =1,2,…m.
=1 2 m

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 6

 Pole – zero plot
 Pole – zero plot is a graph which represents the position of
poles and zeros in the complex s-plane.

Pole
Zero

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 7

 Stability analysis in the complex plane

 The stability of a system depends on the location of its poles.

 If all the poles of the system lie in the left-half s-plane then the
system is stable.
 If any of the poles of the system lie in the right-half
right half s-plane
s plane
then the system is unstable.
 If some of the poles of the system lie in the imaginary axis
and the others lie in the left-half s-plane then the system is at
the stability boundary.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 8

 Characteristic equation

 Characteristic equation: is the equation A(s) = 0

 Characteristic polynomial: is the denominator A(s)

 Note:

Feedback systems Systems described

R(s) Y(s) byy state equations
q
+_ G(s)
 x (t )  Ax(t )  Bu (t )

H(s)  y (t )  Cx(t )

Characteristic equation Characteristic equation

1  G(s) H (s)  0 det sI  A  0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 9

 Algebraic stability criteria

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 10

 Necessary condition

 The necessary condition for a linear system to be stable is

that all the coefficients of the characteristic equation of the
system
t mustt be
b positive.
iti
 Example: Consider the systems which have the characteristic
equations:
ti
s 3  3s 2  2 s  1  0 Unstable
s 4  2 s 2  5s  3  0 Unstable
s 4  4 s 3  5s 2  2 s  1  0 Cannot conclude about the stability

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 11

 Routh’s stability criterion
R l for
Rules f forming
f i the
th Routh
R th table
t bl
 Consider a linear system whose characteristic function is:
a0 s n  a1s n1    an1s  an  0
 To analyze the system stability using Routh
Routh’s s criterion, it is
necessary to form the Routh table according to the rules:
 The Routh table has n+1 rows.
 The 1st row consists of the even-indexed coefficients.
 The 2nd row consists of the odd-indexed coefficients.
 The element at row ith column jth (i  3) is calculated as:
cij  ci 2, j 1   i .ci 1, j 1
ci 2,1
with i 
ci 1,1
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 12
 Routh’s stability criterion
Routh table

cij  ci 2, j 1   i .ci 1, j 1

ci 2,1
i 
ci 1,1
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 13
 Routh’s stability criterion
R
Routh’s
h’ criterion
i i statement
 The necessary and sufficient condition for a system to be
stable is that all the coefficients of the characteristic equation
are positive and all terms in the first column of the Routh table
have positive signs.
signs
 The number of sign changes in the first column of the Routh
table is equal
eq al the number
n mber of roots lying
l ing in the right-half
right half s-
s
plane.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 14

 Routh’s stability criterion – Example 1
 Analyze the stability of the system which have the following
characteristic equation: s 4  4 s 3  5s 2  2 s  1  0
 S l ti
Solution: R th table
Routh t bl

 Conclusion: The system

y is stable because all the terms in the
first column are positive.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 15
 Routh’s stability criterion – Example 2

 Analyze the system described by the following block diagram:

R(s)
( ) Y(s) 50
+_ G( )
G(s) G ( s) 
s ( s  3)( s 2  s  5)
1
H (s) 
H(s) s2

 Solution: The characteristic equation of the system:

1  G ( s ).
1 ) H (s)  0
50 1
 1 . 0
s ( s  3)( s  s  5) ( s  2)
2

 s ( s  3)( s 2  s  5)( s  2)  50  0
 s 5  6 s 4  16 s 3  31s 2  30 s  50  0
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 16
 Routh’s stability criterion – Example 2 (cont’)
 R h table
Routh bl

 Conclusion: The system is unstable because the terms in the

first column change their signs two times. The characteristic
equation has two roots with positive real parts.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 17
 Routh’s stability criterion – Example 3
 Find the condition of K for the following system to be stable.

R(s) Y(s)
( )
+ G(s) K
G(s) 
s ( s 2  s  1)( s  2)

 Solution: The characteristic equation of the system is:

1  G(s)  0
K
 1 0
s ( s  s  1)( s  2)
2

 s 4  3s 3  3s 2  2 s  K  0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 18

 Routh’s stability criterion – Example 3 (cont’)
 R th table
Routh t bl

 The necessary & sufficient condition for the system to be stable:

 9
2 K 0 14
 7  0K 
 K  0 9

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 19

 Routh’s stability criterion – Special case #1

 If a first-column term in any row is zero, but the remaining

terms in that row are not zero or there is no remaining term,
then the zero term is replaced by a very small positive
number  and the rest rows of the Routh table is calculated as
the normal case.
case

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 20

 Routh’s stability criterion – Example 4
 A l
Analyze th stability
the t bilit off the
th system
t whose
h characteristic
h t i ti equation
ti is:
i
s 4  2 s 3  4 s 2  8s  3  0
 S l ti
Solution: R th table
Routh t bl

 Conclusion: Because the terms in the first column change

their signs
g two times,, the system
y is unstable and it has two
poles lying in the right-half complex plane.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 21
 Routh’s stability criterion – Special case #2
 If all the coefficients in any row are zero:
 Forming an auxiliary polynomial with coefficients of the last
row above
b th “all-zero-term
the “ ll t row”,
” denote
d t the
th auxiliary
ili
polynomial as A0(s).
 Replace the “all all-zero-term
zero term row
row” by another row whose
elements are the coefficients of the derivative dA0(s)/ds.
 Then continue to calculate the Routh table as the normal
case.

 Note: The roots of A0(s) are also the roots of characteristic

equation.
q

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 22

 Routh’s stability criterion – Example 5

 Analyze the stability of the system whose characteristic equation is:

s  4 s  8s  8s  7 s  4  0
5 4 3 2

 Solution: Routh table

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 23

 Routh’s stability criterion – Example 5 (cont’)

 The auxiliary polynomial:

dA0 ( s )
A0 ( s )  4 s  4
2
  8s  0
ds
 The roots of the auxiliary polynomial (are also the roots the
characteristic equation):
A ( s)  4s 2  4  0  sj
0

 Conclusion:
 All the terms in the first column are positive  characteristic
equation has no root lying in the right-half s-plane.
 The characteristic equation has two roots lying in the imaginary
axis.
i
 The number of roots lying in the left-half s-plane is 5 – 2 = 3.
Th system
The t iis att th
the stability
t bilit boundary.
b d

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 24

 Hurwitz’s stability criterion
R l for
Rules f forming
f i the
h Hurwitz
H i matrix
i
 Given a system whose characteristic equation is:
a0 s n  a1s n1    an1s  an  0
 To analyze
y the system
y stabilityy using
g Hurwitz’s criterion,, it is
necessary to form the Hurwitz matrix according to the rules:
 The Hurwitz matrix is a square matrix of order nn.
 The diagonal consists of the coefficients a1 to an.
 The odd row of the Hurwitz matrix consists of the odd-indexed
coefficients of the characteristic polynomial; the indexes
increase on the right and decrease on the left of the diagonal.
 The even row of the Hurwitz matrix consists of the even even-
indexed coefficients of the characteristic polynomial; the
indexes increase on the right
g and decrease on the left of the
diagonal .
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 25
 Hurwitz’s stability criterion
H
Hurwitz
i matrix
i

 a1 a3 a5 0
a7 
a a a4 0
a6 
 0 2 
 0 a1 a3 a5  0 
 
 0 a0 a2 a4  0 
    
 
0     an 

Hurwitz’s criterion statement

 The necessary and
Th d sufficient
ffi i t condition
diti f the
for th system
t t be
to b
stable is that all the determinants of the principal sub-matrices
of the Hurwitz matrix are positive.
positive

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 26

 Hurwitz’s stability criterion – Example 1
A l
Analyze th stability
the t bilit off the
th system
t whose
h characteristic
h t i ti equation
ti is:
i
s 3  4s 2  3s  2  0
 S l ti
Solution:  a1 a3 0  4 2 0
Hurwitz matrix: a a2 0   1 3 0 
 0   
 0 a1 a3  0 4 2

The
e dete a ts 1  a1  4
determinants:
a1 a3 4 2
2    4  3  1  2  10
a0 a2 1 3
a1 a3 0
a1 a3 4 2
 3  a0 a2 0  a3  2  2  10  20
a0 a2 1 3
0 a1 a3
 Conclusion: The system
y is stable because all the determinants
are positive.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 27
 Hurwitz’s stability criterion – Some corollaries
 A 2nd order
d system t i stable
is t bl if th
the coefficients
ffi i t off th
the
characteristic polynomial satisfy the conditions:
ai  0, i  0,2
 A 3rd order system is stable if the coefficients of the
characteristic polynomial satisfy the conditions:
ai  0, i  0,3

a1a2  a0a3  0
 A 4th order system is stable if the coefficients of the
characteristic polynomial satisfy the conditions:
ai  0, i  0,4

a1a2  a0a3  0
a a a  a a 2  a 2 a  0
 1 2 3 0 3 1 4

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 28

 The root locus method

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 29

 The concept of root locus (RL)
 Example:
E l Plot
Pl t off allll the
th roots
t off the
th following
f ll i characteristic
h t i ti
equation when K changes from 0  + .
s 2  4s  K  0

 Definition: Root locus is the set of all the roots of the

q
characteristic equation of a system
y when a real p
parameter
changing from 0  +.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 30
 Magnitude and phase condition of the root locus

 IIn order
d to
t applyl the
th rules
l forf construction
t ti off the
th roott locus,
l
first we have to equivalently transform the characteristic
equation to standard form:
N (s) (1)
1 K 0
D(s
(s)
where K is the changing parameter.
N ((ss )
Denote: G0 ( s )  K
D( s)
A
Assume th t G0(s)
that ( ) has
h n poles
l pi and
d m zeros zi.
(1)  1  G0 ( s )  0
 G0 ( s )  1 magnitude condition
 
G0 ( s )  ( 2l  1) phase
h condition
di i
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 31
 Rules for construction of the root locus
 Rule 1: The number of branches of a root locus = the order
of the characteristic equation = number of poles of G0(s) = n.

 Rule 2:
 For K = 0: the root locus begins at the poles of G0(s).
(s)
 As K goes to + : m branches of the root locus end at m
zeros of G0(s), the nm remaining branches go to infinity
approaching the asymptote defined by the rule 5 & rule 6.

 Rule 3: The root locus is symmetric with respect to the real

axis.

 Rule 4: A point on the real axis belongs to the root locus if the
total number of poles and zeros of G0(s) to its right is odd.
odd

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 32

 Rules for construction of the root locus (cont’)
 Rule 5: The angles between the asymptotes and the real axis
are calculated by:
( 2l  1)
 (l  0,1,2, )
nm
 Rule 6: The intersection between the asymptotes and the real
axis is a point A defined by:
n m

 pole   zero  p zi i (pi & zi are poles

OA   i 1 i 1
and zeros of G0(s) )
nm nm

 Rule 7: Breakaway / break-in points (or break points for

short), if any, are located in the real axis and are satisfied the
equation: dK
0
ds
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 33
 Rules for construction of the root locus (cont’)
 Rule 8: The intersections of the root locus with the imaginary
axis can be determined by using the Routh-Hurwitz criteria or
b substituting
by s j into the characteristic equation.
s bstit ting s=j eq ation

 Rule 9: The departure angle of the root locus from a pole pj

(of multiplicity 1) is given by:
m n
 j  1800   arg( p j  zi )   arg( p j  pi )
i 1 i 1
i j

The geometric form of the above formula is:

j = 1800 + (angle
l ffrom the
th zero zi (i=1..m)
(i 1 ) tto the
th pole
l pj )
 (angle from the poles pi (i=1..m, i≠j) to the pole pj )

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 34

 The root locus method – Example 1
 Sketch the root locus of the following system when K=0+.

R(s) Y(s)
+ G(s) K
G ( s) 
s ( s  2)( s  3)

 S l ti
Solution:
 The characteristic equation of the system:

1  G ( s)  0 
K
1 0 (1)
s ( s  2)( s  3)
 Poles: p1  0 p2  2 p3  3
 Zeros: none
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 35
 The root locus method – Example 1 (cont’)
 The asymptotes: 
1  (l  0)
3
(2l  1) (2l  1) 
   2   (l  -1)
nm 30 3
3   (l  1)

OA 
 pole   zero [0  ( 2)  ( 3)]  0
 
5
nm 3 0 3
 The break points:
(1)  K   s ( s  2)( s  3)  ( s 3  5s 2  6 s )
dK
  (3s 2  10 s  6)
ds
dK  s1  2.549 ( rejected)
Then 0  
ds  s2  0.785
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 36
 The root locus method – Example 1 (cont’)
 Th intersections
The i i off the
h root locus
l with
i h the
h imaginary
i i axis:
i
Method 1: Using the Hurwitz’s criterion
(1)  s  5s  6 s  K  0 (2)
3 2

Stability condition:
K  0  K  0  0  K  30  K cr  30
 
a1a2  a0 a3  0 5  6  1  K  0
Substitute Kcr = 30 into the equation (2) and solve the
equation, we have the intersections of the root locus with the
imaginary axis. s  5
1
s  5s  6s  30  0 
  s2  j 6
3 2

s   j 6
3
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 37
 The root locus method – Example 1 (cont’)

 The intersections of the root locus with the imaginary axis:

Method 2:
(1)  s 3  5s 2  6s  K  0 (2)

Substitute s=j into the equation (2):

 j  3
 5 j   6 j   K  0
2
  j  5  6 j  K  0
3 2

  0

  j 3  6 j  0 K  0
  
  2
K 0    6
 5

 K  30

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 38

 The root locus method – Example 1 (cont’)

Im s

j 6

Re s
3 2 0

j 6

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 39

 The root locus method – Example 2
 Sketch the root locus of the system below when K=0+.

R(s) Y(s)
+ G(s) K
G (s) 
s ( s 2  8s  20)

 Solution:
 The characteristic equation of the system:
K
1  G ( s)  0  1 0 (1)
s ( s  8s  20)
2

 Poles: p1  0 p2,3  4  j 2
 Zeros: none
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 40
 The root locus method – Example 2 (cont’)

The asymptotes: 
 1  (l  0)
3
(2l  1) (2l  1) 
   2   (l  -1)
nm 30 3
3   (l  1)

OA 
 pole   zero [0  ( 4  j 2)  ( 4  j 2)]  (0)
 
8
nm 3 0 3
 The break points:
p
(1)  K  ( s 3  8s 2  20s)

dK
 (3s 2  16 s  20)
ds
dK s1  3.331  (2 break
K points
0
Then 0  
s2  2.00 s ( s accepted)
8s  20)
2
ds
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 41
 The root locus method – Example 2 (cont’)
 The intersections of the root locus with the imaginary axis:
(1)  s 3  8s 2  20s  K  0 (2)

Substitute s=j into the equation (2):

( j )3  8( j ) 2  20( j )  K  0

  j 3  8 2  20 j  K  0

  0

 8 2  K  0 K  0
  
   3
 20  0    20
 K
 K  160 1  0
s ( s  8s  20)
2

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 42

 The root locus method – Example 2 (cont’)

 The departure angle of the root locus from the pole p2

 2  1800  [arg( p2  p1 )  arg( p2  p3 )]

 1800  arg[(4  j 2)  0]  arg[(4  j 2)  (4  j 2)]
 1 2  
 180  tg 
0
  90
 4 
 1800  153.5  90
 2  63.50
m n
 j  180   arg( p j  zi )   arg( p j  pi )
0

i 1 i 1
i j

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 43

 The root locus method – Example 2 (cont’)

Im s
j 20

+j2
j2
63.50

Re s
4 2 0

j2

 j 20

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 44

 The root locus method – Example 3
 Sketch the root locus of the system below when K=0+.

R(s)
( ) Y(s)
( )
+ G( )
G(s) K ( s  1)
G ( s) 
s ( s  3)( s 2  8s  20)

 Solution:
 The characteristic equation of the system:
K ( s  1)
1  G(s)  0  1 0 (1)
s ( s  3)( s  8s  20)
2

 Poles: p1  0 p2  3 p3, 4  4  j 2
 Zeros: z1  1
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 45
 The root locus method – Example 3 (cont’)
 Th asymptotes:
The

1  (l  0)
3
(2l  1) (2l  1) 
   2   (l  -1)
nm 4 1 3
3   (l  1)

OA 
 pole   zero [0  ( 3)  ( 4  j 2)  ( 4  j 2)]  ( 1)
 
10
nm 4 1 3
 The break points:
s ( s  3)( s 2  8s  20) dK 3s 4  26s 3  77 s 2  88s  60
(1)  K    
( s  1) ds ( s  1) 2

dK  s1, 2  3,67  j1,05 K ( s (rejected)

Then 0   1
1)
0
ds s3, 4  0,66  sj(0s.97 2 (rejected)
 3)( s  8s  20)
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 46
 The root locus method – Example 3 (cont’)
 Th intersections
The i i off the
h root locus
l with
i h the
h imaginary
i i axis:
i
(1)  s 4  11s 3  44s 2  (60  K ) s  K  0 (2)
Substitute s=j into the equation (2):
 4  11 j 3  44 2  (60  K ) j  K  0
  0

K  0
 4  44 2  K  0   5,893
  
  3
 (60  K )  0 
 11
 K  322
   j1,314 (rejected)
 K,7( s  1)
1K  61 0
s ( s  3)( s  8s  20)
2

 the intersections are: s   j 5,893 Critical gain: K cr  322

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 47
 The root locus method – Example 3 (cont’)

 The departure angle of the root locus from the pole p3

3  180  1  (  2  3   4 )
 180  146,3  (153,4  116,6  90)
3  33.7 0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 48

 The root locus method – Example 3 (cont’)
Im s
+j5,893

+j2
33.70

3 1 2 Re s
4 3 1 0
4
j2

5 893
j5,893

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 49

 The root locus method – Example 4
 Given the system below:

R( )
R(s) Y(s) 10
+ GC(s) G(s) G (s)  2
( s  9 s  3)
KI
GC ( s )  K P 
s
 For KI = 2.7, sketch the root locus of the system when
KI=0+, note that dKP / ds=0 has 3 roots at 3,  3, 1.5.

 For KP =270, KI = 2.7, the system is stable or not?

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 50

 The root locus method – Example 4 (cont’)

 Solution:
 The characteristic equation
q of the system:
y
1  GC ( s )G ( s )  0

 1   KP 
2.7  10 
 2 0
 s  s  9 s  3 
10 K P s
 1 0 (1)
( s  9)( s  3)
2

 Poles: p1  9 p2   j 3 p3   j 3

 Zeros: z1  0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 51

 The root locus method – Example 4 (cont’)
 The asymptotes:

(2l  1) (2l  1)  /2 (l  0)

  
nm 3 1  /2 (l  1)

OA 
 pole   zero [ 9  ( j

3 )  (  j 3 )]  (0)

9
nm 3 1 2
 The break points:
 s1  3
dK P 
0   s2  3
ds s  1.5 (rejected)
3
The root locus has two break points at the same location 3

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 52

 The root locus method – Example 4 (cont’)
 The departure angle of the root locus from the pole p2
 2  1800  arg(
g( p2  z1 )  [[arg(
g( p2  p1 )  arg(
g( p2  p3 )]

 1800  arg( j 3  0)  [arg( j 3  (9))  arg( j 3  ( j 3 ))]

 1  3  
 180  90  tg 
0
  90
 9 
 2  1690

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 53

 The root locus method – Example 4 (cont’)

 For KI =2.7 the root locus

is located completely in
the left-half
f f s-plane when
KP =0+, so the system
is stable when KI =2.7,
=2 7
KP =270.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 54

 Frequency domain analysis

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 55

 Frequency response

 Observe the response of a linear system at steady state

when the input is a sinusoidal signal.
signal

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 56

 Frequency response definition
 It can be observed that, for
f linear system, iff the input is a
sinusoidal signal then the output signal at steady-state is also
a sinusoidal signal with the same frequency as the input,
input but
different amplitude and phase.
u (t)=Umsin (j) y (t)=Ymsin (j+)
+ )
G(s)
U (j) Y (j)

 Definition: Frequency response of a system is the ratio

between the steady-state
y output
p and the sinusoidal input.
p
Y ( j )
Frequency response 
U ( j )

It is proven that: Frequency response  G ( s ) s j  G ( j )

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 57

 Magnitude response and phase response
 l G(j) is
IIn general, i a complex
l ffunction
ti and d it can b
be
represented in algebraic form or polar form.
) e j ( )
G ( j )  P( )  jQ( )  M ( ).
where:
M ( )  G ( j )  P 2 ( )  Q 2 ( ) Magnitude response

) 
1  Q (
 ( )  G ( j )  tg   Phase response
 P ( ) 

 Physical meaning of frequency response:

 The magnitude
g response
p p
provides information about the ggain
of the system with respect to frequency .
 The phase response provides information about the phase
shift between the output & the input with respect to frequency
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 58
 Graphical representation of frequency response
 Bode diagram: is a graph of the frequency response of a
linear system versus frequency plotted with a log-frequency
axis Bode diagram consists of two plots:
axis.
 Bode magnitude plot expresses the magnitude response
gain L() versus frequency  .

L( )  20 lg M ( ) [dB]

 Bode phase plot expresses the phase response ()

y .
versus frequency
q

 Nyquist
yq plot: is a g
p graph
p in p
polar coordinates in which the
gain and phase of a frequency response G(j) are plotted
when  changing from 0+.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 59

 Graphical representation of frequency response (cont’)

Bode
B d di
diagram Nyquist
N i t plot
l t

Gain margin

Gain margin

Phase margin

Phase margin

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 60

 Frequency response of basic factor
Proportional gain

 Transfer function: G ( s )  K
 Frequency response: G ( j )  K
 Magnitude response: M ( )  K  L( )  20 lg K
 Phase response:  ( )  0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 61

 Frequency response of basic factor (cont.)
Proportional gain

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 62
 Frequency response of basic factor (cont.)
Integral factor

1
 Transfer function: G ( s ) 
s
1 1
 Frequency response: G ( j )  j
j 
1
 Magnitude response: M ( )   L( )  20 lg 

 Phase response:  ( )  900

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 63

 Frequency response of basic factor (cont.)
Integral factor

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 64
 Frequency response of basic factor (cont
(cont.)
.)
Derivative factor

 Transfer function: G(s)  s

 Frequency response: G ( j )  j

 Magnitude response: M ( )    L( )  20 lg 

 Phase response:  ( )  900

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 65

 Frequency response of basic factor (cont.)
D i i factor
Derivative f

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 66
 Frequency response of basic factor (cont
(cont.)
.)
Fi t d lag
First-order l factor
f t
1
 Transfer function: G( s) 
T 1
Ts
1
 Frequencyy response: G ( j ) 
Tj  1
1
 Magnitude response: M ( ) 
1  T 2 2

 L ( )  20 lg 1  T 2 2

 Phase response:  ( )  tg 1 (T )

 Approximation of the Bode diagram by asymptotes:
  1 / T : the asymptote lies on the horizontal axis
   1 / T : the asymptote has the slope of 20dB/dec
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 67
 Frequency response of basic factor (cont.)
Fi t d lag
First-order l factor
f t

corner freq
frequency
enc
corner frequency

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 68
 Frequency response of basic factor (cont.)
Fi t d lead
First-order l d factor
f t

 Transfer function: G ( s)  Ts  1
 Frequency response: G ( j )  Tj  1

 Magnitude response: M ( )  1  T 2 2

 L ( )  20 lg 1  T 2 2
 Phase response:  ( )  tg 1 (T )
 Approximation of the Bode diagram by asymptotes:
   1 / T : the asymptote lies on the horizontal axis
  1 / T : the asymptote has the slope of +20dB/dec
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 69
 Frequency response of basic factor (cont.)
Fi t d lead
First-order l d factor
f t

corner frequencyy

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 70
 Frequency response of basic factor (cont
(cont.)
.)
Second-order oscilating factor
1
 Transfer function: G (s)  2 2 (0    1)
T s  2Ts  1

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 71

 Frequency response of basic factor (cont.)
Second-order oscilating factor
1
 Frequency response: G ( j ) 
 T 2 2  2Tj  1
1
 Magnitude response: M ( ) 
(1  T 2 2 ) 2  4 2T 2 2

 L( )  20 lg (1  T 2 2 ) 2  4 2T 2 2
2T 
1 
 Phase response:  ( )  tg  2 2
1 T  
 Approximation
pp of the Bode diagram
g byy asymptotes:
y p
  1 / T : the asymptote lies on the horizontal axis
  1 / T : the asymptote has the slope of 40dB/dec

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 72

 Frequency response of basic factor (cont.)
S
Second-order
d d oscilating
il ti factor
f t
Corner frequency

Bode diagram N
Nyquist
i t plot
l t
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 73
 Frequency response of basic factor (cont
(cont.)
.)
Time delay factor

 T
Transfer
f function:
f ti G ( s)  e Ts

 Freq enc response:

Frequency response G ( j )  e Tj

 Magnitude
M it d response: M (  )  1  L ( )  0
 Phase response:  ( )  T

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 74

 Frequency response of basic factor (cont.)
Time delay factor

Bode diagram Nyquist plot

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 75
 Frequency response of control systems
 Consider
C id a control
t l system
t which
hi h has
h th transfer
the t f function
f ti
G(s). Suppose that G(s) consists of basis factors in series:
l
G ( s )   Gi ( s )
i 1
l
 Frequency response: G ( j )   Gi ( j )
i 1
l l
 Magnitude response: M ( )   M i ( )  L( )   Li ( )
i 1 i 1
l
 Phase response:
p  ( )   i ( )
i 1

 The Bode diagram

g of a system
y consisting
g of basic factors in series
equals to the sum of the Bode diagram of the basic factors.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 76
 Approximation of Bode diagram
 S
Suppose that the TF off the system is off the form:
f
G ( s )  Ks G1 ( s )G2 ( s )G3 ( s )
(>0: the system has ideal derivative factor(s)
<0: the system has ideal integral factor(s))

 Step 1: Determine all the corner frequencies i =1/Ti , and sort

g order 1 <2 < 3 …
them in ascending

 Step 2: The approximated Bode diagram passes through the

point A having the coordinates:
  0

 L( )  20 lg
l K    20 lg
l 0
where 0 is a frequency satisfying 0 < 1 . If 1 > 1 then it is
possible to chose 0 =1.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 77
 Approximation of Bode diagram (cont’)
 Step 3: Through point A, draw an asymptote with the slope:
S
 ( 20 dB/dec  ) if G(s) has  ideal integral factors
dB/d  ) if G(s)
 (+ 20 dB/dec h  ideal
G( ) has id l dderivative
i ti ffactors
t
The asymptote extends to the next corner frequency.

 Step 4: At the corner frequency i =1/Ti , the slope of the

asymptote is added with:
 (20dB/dec  i) if Gi(s) is a first-order lag factor (multiple i)
 (+20dB/dec  i) if Gi(s) is a first-order lead factor (multiple i)
 ( 40dB/dec  i) if Gi(s) is a 2nd order oscillating factor (multiple i)
(40dB/dec
 (+40dB/dec  i) if Gi(s) is a 2nd order lead factor (multiple i)
The asymptote extends to the next corner frequency
frequency.

 Step 5: Repeat the step 4 until the asymptote at the last

corner frequency
f i plotted.
is l tt d
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 78
 Approximation of Bode diagram – Example 1
 Plot the Bode diagram using asymptotes:
100(0,1s  1)
G (s) 
s (0,01s  1)
Based on the Bode diagram, determine the gain cross
frequency of the system.
system

 Solution:
 Corner frequencies:
1 1 1 1
1   ( d/ ) 2 
 10 (rad/sec)   100 (rad/sec)
( d/ )
T1 0,1 T2 0,01
 The Bode diagram pass the point A at the coordinate:
  1

 L( )  20 lg
l K  20 lg
l 100  40
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 79
 Approximation of Bode diagram – Example 1 (cont’)

L(), dB

A
40
20dB/dec

0dB/dec
20
20dB/dec
0 -1
1 0 1 2 3 lg
10-1 100 101 102 c 

 In the Bode diagram, the gain crossover frequency is 103

rad/sec.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 80
 Example 2 – Bode diagram to transfer function

 Determine the transfer function of the system which has the

approximation Bode diagram as below:

L(), dB

60 0dB/dec
54
D E
A
40 20dB/dec
B C
26
20 0dB/dec

0
-1 0 1 1.301 2 lg
g1 g2 g3

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 81

 Example 2 – Bode diagram to transfer function (cont’)

54  26
 The slope of segment CD:  40 (dB/dec)
2  1.301
 The corner frequencies:
40  26
lg  g1  0   0 .7   g1  100.7  5 (rad/sec)
20
lg  g 2  1.301   g 2  101.301  20 (rad/sec)
lg  g 3  2   g 3  102  100 (rad/sec)
K (T1s  1)(T2 s  1) 2
 The transfer function has the form: G ( s ) 
s (T3s  1) 2
20 lg K  40  K  100
1 1 1 1 1 1
T1    0.2 T2    0.05 T3    0.01
 g1 5  g 2 20  g 3 100
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 82
 Crossover frequency
 Gain crossover frequency(c): is the frequency where the
amplitude of the frequency response is 1 (or 0 dB).
M (c )  1  L(c )  0

 Phase crossover frequency (): is the frequency where

phase shift of the frequency response is equal to 1800 (or
equal to  radian).
 ( )  1800   ( )   rad

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 83

 Stability margin
 i (GM):
G i margin
Gain (GM)
1
GM   GM   L( ) [[dB]]
M ( )

Physical meaning: The gain margin is the amount of positive

gain at the phase crossover frequency required to bring the
system to the stability boundary.

 Phase margin (M)

M  1800   (c )
Physical meaning: The phase margin is the amount of
additional phase lag at the gain crossover frequency required to
bring the system to the stability boundary.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 84

 Nyquist stability criterion

 Consider a unity feedback

C f system shown below, suppose
that we know the Nyquist plot of the open loop system G(s),
the problem is to determine the stability of the closed-loop
closed loop
system Gcl(s).
R(s) Y(s)
( )
+ G(s)

 Nyquist criterion: The closed-loop system Gcl(s) is stable if

and only if the Nyquist plot of the open-loop system G(s)
encircles the critical point (1, 1 j0) l/2 times in the
counterclockwise direction when  changes from 0 to + (l is
the number of p
poles of G(s)
( ) lying
y g in the right-half
g s-plane).
p )

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 85

 Nyquist stability criterion – Example 1
 Consider
C id an unity
i negative
i f db k system, whose
feedback h open-
loop system G(s) is stable and has the Nyquist plots below
(three cases).
cases) Analyze the stability of the closed-loop system.
system

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 86

 Nyquist stability criterion – Example 1
 Solution
The number of poles of G(s) lying in the right-half s-plane is 0
b
because G( ) is
G(s) i stable.
t bl Th
Then according
di t the
to th Nyquist
N i t
criterion, the closed-loop system is stable if the Nyquist plot
G(j) does not encircle the critical point (1,
1 j0)

 Case : G(j) does not encircle (1,

1 j0)
 the close-loop system is stable.
 Case : G(j) pass (1,
1 j0)
 the close-loop system is at the stability
boundary;
 Case : G(j) encircles (1, j0)
 the close-loop system is unstable.
unstable

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 87

 Nyquist stability criterion – Example 2
 Analyze
A l th stability
the t bilit off a unity
it negative
ti f db k system
feedback t
whose open loop transfer function is:
K
G ( s) 
s (T1s  1)(T2 s  1)(T3 s  1)
 Solution:

 Nyquist
yq p
plot: Depending
p g
on the values of T1, T2,
T4 and K, the Nyquist
plot
l off G(s)
G( ) could
ld b
be one
of the three curves 1, 2
or 3
3.

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 88

 Nyquist stability criterion – Example 2 (cont’)

The number of poles of G(s) lying in the right-half s-plane is 0

because G(s) is stable. Then according to the Nyquist
criterion,
it i th closed-loop
the l dl system
t i stable
is t bl if the
th Nyquist
N i t plot
l t
G(j) does not encircle the critical point (1, j0)

 Case : G(j) does not encircle (1, j0)

 the
th close-loop
l l system
t i stable.
is t bl
 Case : G(j) pass (1, j0)
 the
th close-loop
l l system
t i att the
is th stability
t bilit boundary;
b d
 Case : G(j) encircles (1, j0)
 the
h close-loop
l l system is
i unstable.
bl

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 89

 Nyquist stability criterion – Example 3
Given an unstable open-loop systems which have the Nyquist
plot as below. In which cases the closed-loop system is stable?

Stable Unstable
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 90
 Nyquist stability criterion – Example 3 (cont’)

Given an unstable open-loop systems which have the Nyquist

plot as below. In which cases the closed-loop system is stable?

Unstable

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 91

 Nyquist stability criterion – Example 3 (cont’)

Given an unstable open-loop systems which have the Nyquist

plot as below. In which cases the closed-loop system is stable?

Stable Unstable

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 92

 Nyquist stability criterion – Example 4
 Gi
Given a open-loop
l system
t which
hi h has
h the
th transfer
t f function:
f ti
K
G( s)  (K>0 T>0,
(K>0, T>0 n>2)
(Ts  1) n
Find the condition of K and T for the unity negative feedback
closed-loop system to be stable.
 Solution:
 Frequency response of the open-loop system:
K
G ( j ) 
(Tj  1) n
K
M ( ) 
 T   1
 Magnitude: n
2 2

 Phase:  ( )  ntg 1 (T )

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 93
 Nyquist stability criterion – Example 4 (cont’)

 Nyquist plot:

 Stability condition: the Nyquist plot of G(j) does not encircle

point (1,j0).
the critical p ,j ) Accordingg to the Nyquist
yq plot,, this
p
requires:
M ( )  1

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 94

 Nyquist stability criterion – Example 4 (cont’)

We have:  ( )   ntg (T )  

1


  
 tg (T ) 
1
 (T )  tg  
n n
1  
   tg  
T n

Then: M ( )  1 K
  n
1
   
2 
 T 2 tg    1 
1
  T  n  
 
n   
    
 K   tg    1 
2
  n  
 
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 95
 Bode criterion
 Consider a unity feedback system, suppose that we know the
Nyquist plot of the open loop system G(s), the problem is to
determine the stability of the closed-loop
closed loop system Gcl(s).
(s)
R(s) Y(s)
+ G(s)

 Bode criterion: The closed-loop system Gcl(s) is stable if the

gain margin and phase margin of open-loop
open loop system G(s) are
positive.
GM  0
  The closed - loop system is stable
 M  0

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 96

 Bode criterion – Example
 Consider a unity negative feedback system whose open-loopopen loop
system has the Bode diagram as below. Determine the gain
margin,
g ,p phase margin
g of the open-loop
p p system.
y Is the closed-
loop system stable or not?
Bode diagram:
c  5
   2
L(( )
GM L(  )  35dB
 ( c )  2700
GM  35dB

180
M  180 0  (270 0 )  900
M Because GM<0 and
(C)
 C M<0, the closed-loop
system is unstable.
10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 97
 Remark on the frequency domain analysis

 If the closed-loop system as below, the Nyquist and Bode

criteria can also be applied and in this case the open-loop
open loop
system is G(s)H(s).

R(s)
(s) E(s) Y(s)
()
+_ G(s)

Yfb(s)
()
H(s)

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 98

 End of chapter 3

10 January 2016 © H. T. Hoàng - www4.hcmut.edu.vn/~hthoang/ 99