Stability of Linear Control System

Unstable system is useless. Amongst the important specifications

of the controller design is that the overall system must be stable.
This subject will discuss the stability of LTIV SISO systems.

Bounded-Input Bounded-Output (BIBO) Stability

With zero initial conditions, the system is said to be BIBO stable,
or simply stable, if its output y  t  is bounded for a bounded input.

y t   0 u  t    g   d
If u t   M , then
y t   N

where M and N is    ve  

Zero-Input Stability
The system is said to be zero-input stable or stable, if the
zero-input response subject to the finite initial conditions
reached zero as t approaches infinity.
In mathematic:
lim y  t   0 if u  t   0 for t  t 0
t 

dan y  t0   M  

As y t  approaching 0 as t approaching  then it is also known

as asymptotic stability.
All the stability conditions are the same, hence it will be referred
as stability.
Method of Determine Stability
Stability of a system is solely based on the characteristic
equation of the system. The system is stable if all poles are
located in left hand side of the imaginary axis of s-plane.
If one of the poles is located or placed on the imaginary
axis, then the system is said to marginally stable or marginally
unstable.

Example:
20
G s  
 s  1 s  2 s  3
Poles at s = -1, -2, -3; therefore stable

20 s  1
G s  

 s  1 s 2  2s  2 

unstable due to a pole at s  1.

20 s  1
G s  

 s  2 s 2  4 
Poles at s  2,  j 2
Marginally unstable due to multiple poles at s   j2

10
G s  
 s 2  4 2  s  10
Marginally unstable due to multiple poles at s   j2

20 s  1
G s  
s  s  2  s  4 
Poles at s  0,2,4

 Marginally stable if the pole is purposely place at s  0

Computer program such as MATLAB, PROGRAM CC,

CONTROL W etc. could be used to find the roots of
characteristic eqn. and determine system stability.
However, some time in design the roots are unknown or
there are some variable parameters embedded in the
characteristic equation of the system. These 3 methods are
normally been used to determine the system stability without
solving the roots of characteristic equation:
1. Routh-Hurwitz criterion
2. Nyquist criterion
3. Bode Diagram

Routh-Hurwitz Criterion
For a transfer function,
G s  

K bm s m  bm 1s m 1    b1s1  b0 
an s n  an 1s n 1    a1s1  a0

where n  m

Routh table is:

sn : an an-2 an-4 … 0
sn-1 : an-1 an-3 an-5 … 0
sn-2 : c1 c2 c3 … 0
sn-3 : d1 d2 d3 … 0
   
s1 : g1 0
 s0 : h1 0

where

a n 1 a n  2  a n a n  3 a n 1 a n  4  a n a n  5
c1  ; c2 
a n 1 a n 1

an 1a n  6  a n a n  7
c3 
a n 1

c1an 3  c2 an 1 c a c a
d1  ; d 2  1 n 5 3 n 1
c1 b1

Routh- Hurwitz Criterion

1. If all elements in the first column have the same sign, the
system is stable and all poles are on LHS of s-plane.
2. The number of changes of signs is equal to the number of
unstable poles or poles on RHS of s-plane. System is unstable.
3. If the first element of a row is zero, change it with a small    ve
' '.
4. If all elements of a row are zero, poles are on the imaginary
real axis or on LHS of the imaginary axis of s-plane.

Example 1:
If the characteristic equation of a system is,
s 5  s 4  6s 3  5s 2  12s  20  0

Determine the stability of the system.

 Routh’s Table
s5 : 1 6 12 0
s4 : 1 5 20 0
s3 : 1 8 0
2
s : 13 20 0
s1 : 124
13
0
s0 : 20 0

There are 2 sign changes in the first column, thus the system has
2 poles on RHS of s-plane. Therefore, the system is unstable.

Example 2
s 5  s 4  5s 3  5s 2  12s  10  0

Routh’s Table
s5 : 1 5 12 0
4
s : 1 5 10 0
s3 :  2 0
s2 : 5 2  

10 0
1  2 10  2
s: 
0
0
s : 10 0

There are 2 sign changes in the first column, thus the system has
2 poles on RHS of s-plane. Therefore, the system is unstable.

Example 3
s 3  3s 2  2s  5  0

Routh’s Table:
s3 : 1 2 0
2
s : 3 5 0
1
s: 13 0
0
s : 5 0
There is no sign change in the first column, hence the system
stable.

Example 4
Determine the range of k so that the system will remain stable.

r t   1 y t 
 k s s  1 s  2

Characteristic equation,
1  GH  0
k
1  s  s  1 s  2  k
s s  1 s  2
 s 3  3s 2  2s  k  0

Routh’s Table
s3 : 1 2 0
s2 : 3 K 0
6K
s1 : 0
3
s0 : K 0

The system will remain stable if there is no sign change in the

first column, or
6K
0 and K 0
3
6 K

3 3
 K 6
Hence, overall 0K 6 is the range of K for the system to remain
stable.

Example 5
Determine the range of k for the following system to remain
stable:
 2 s  3
 k
 s 3  3s 2  2s

2k  s  3
Char. Eqn.  1  GH  1  0
s  3s 2  2 s
3

s 3  3s 2   2k  2  s  6k  0

Routh’s Table,
s3 : 1 2k  2 0
2
s : 3 6k 0
1
s: 2 0
0
s : 6k 0

 system stable when k 0

Example 6
Determine the ranges of K p and K D so that the system will be
stable.

r t   1 y t 
 K p  KDs
 s 3  3s 2  2s

K p  KDs
Char. Eqn.  1  GH  1  0
s 3  3s 2  2 s
s 3  3s 2   2  K D  s  K p  0
Routh’s Table,
s3 : 1 2  KD 0
s2 : 3 Kp 0
6  3K D  K p
s1 : 0
3
s0 : Kp 0

So the system will be stable if,

Kp  0

6  3K D  K p
0 or K D  K p 3  2
3
e.g . if K p  6 then KD  0

Example 6
Determine the range of k so that the system is boundary stable
and determine the oscillation frequency of the system if the char.
Eqn. of the system is:
s 4  25s 3  15s 2  20s  k  0

Routh’s Table,
s4 : 1 15 k 0
3
s : 25 20 0
2
s : 14.2 k 0
1
s : 20  1.76k 0
0
s : k 0

Hence, the system stable for k  0 and 20  1.76k  0

ie. 0  k  11 .36

Therefore, the system is boundary stable when k  11 .36

Auxiliary Eqn. A s   14.2 s 2  11 .36  0

 s 2  0.8

s   j 0.894
so the oscillation freq. is 0.894 rad / sec

Example 7:
Determine the range of k for the system to be stable, Char. Eqn.
is
s 4  ks 3  2s 2   k  1 s  10  0

Routh’s Table:
s4 : 1 2 10 0
s3 : k k 1 0
s2 : k 1
k
10 0
s1 : 9k 2 1 0
k
0
s : 10 0

k must satisfies,
k 0 
k  1
k 1 

 9k 2  1  0
1
or k 2  
9

1
So the system stable when k  0 and k  1 . The final condition k2  
9
cannot be satisfied as k2 is always   ve . Therefore, the system is
unstable.