STABILITY OF CONTROL SYSTEMS

Natural response describes the way the system dissipates or acquires energy. The form or nature of this
response is dependent only on the system to be controlled, and not on input. On the other hand, the form of
the force response is dependent on input signal.

Total response

Forced response + Natural response

+
Particular solution Homogeneous solution

Figure 1: Components of the total output response

x  t   1   1.105e0.6t cos  2.166t  0.277  

Page 1 of 19
Figure 2: Output response of an under-damped second order system with a step input

For a control system to be useful, the natural response must satisfy one of the following two conditions:

1. Eventually approach zero when time tends to infinity, thus leaving only the forced response to
influence the response of the system (Figure 1), or
2. Oscillates with a Simple Harmonic Motion with its natural frequency (Figure 2)

Figure 3: Output response of an undamped second order system with a step input
If the natural response of a system grows without bound rather than diminishing to zero or oscillating, the
natural response becomes so much greater than the force response, then the system is no longer controlled.
This condition is called INSTABILITY. This condition leads to self-destruction of the physical device if limiting

Page 2 of 19
stops are not part of the design. A time response on UNSTABLE system will show a transient response
without bound and without any evidence of a steady state response.

Figure 4: Output response of unstable system

A STABLE SYSTEM produces a bounded output for a given bounded input.

Types of systems based on Stability

Systems can be classified based on stability as follows:

a. Absolutely Stable systems

If the system is stable for all the range of the system component values, then it is known as absolutely stable
system. The open loop control system is absolutely stable if the poles of the open loop transfer function are
located in left side of the imaginary axis in S-plane.

Similarly, the closed loop control system is absolutely stable if all poles of the closed loop transfer function
are located in left side of the imaginary axis in S-plane.

b. Conditionally stable systems

If the system is stable for a certain range of system component values, then it is known as conditionally stable
system

Page 3 of 19
 c. Marginally stable system

If the system is stable by producing an output signal with a constant amplitude and constant frequency of
oscillations for bounded input, then it is known as marginally stable system. The open loop control system is
imaginally stable if any two poles of the open loop transfer function, are located on the imaginary axis in S-
plane (Figure 2). Similarly, the closed loop control system is absolutely stable if all poles of the closed loop
transfer function are located in left side of the imaginary axis in s-plane.

Control systems must be designed to be stable, thus their natural response must decay to zero as the time
approaches infinity or the system must oscillate.

In other words, if the natural response decays to zero as time approaches infinity, the transient response will
also die out, thus leaving only the forced response and steady state error characteristics can be designed.

Analytical conditions for stability

A single-input, single output system transfer function (Open-loop transfer function) ban be written

C  s  b0 s m  b1s m 1  b2 s m  2  ...  bm
T s   ,m n (1)
R  s  a0 s n  a1s n 1  a2 s n  2  ...  an

The denominator a0 s n  a1s n 1  a2 s n  2  ...  an of the transfer function T  s  is called CHARACTERISTIC

POLYNOMIAL and its roots are called POLES of the transfer function.

The CHARACTERISTIC EQUATION is formed by equating the characteristic polynomial to zero.

The characteristic equation is

a0 s n  a1s n 1  a2 s n  2  ...  an  0, a0 0 (2)

The stability of the closed-loop system can be determined by the poles of the closed of the closed loop
system, that is by the ROOTS OF THE CHARACTERISTIC EQUATION.

It can be shown and it is known that the nature of the time response of a system is related to the location of
the ROOTS of the characteristic equation in s-plane.

Page 4 of 19
Page 5 of 19
Non-zero second order systems

Canonical form of the transfer function is

n 2
T s  2
s  2n s  n 2

Table 1: Summary of response behavior of non-zero second order systems

Nature of Dam Characteristic Roots of the Location of Time response

damping ping equation characteristic the poles in s-
ratio equation plane
Undamped  0 s1   jn
s 2  n 2
(Non- s2   jn
damped)

Under  1 s 2  2n s  n 2 s1  n  jn 1   2

damped s1  n  jd

s2  n  jn 1   2

s2  n  jd

Critically  1 s 2  2n s  n 2 s1  s 2  s  n

damped

Over-  1 s 2  2n s  n 2 s1  n  n  2  1

damped

s2  n  n  2  1

Page 6 of 19
Conclusion

For all cases of non-zero second order systems, the roots (poles)of the characteristic equation are located
at the left-hand side of the imaginary axis jd or at the negative side of the real axis  . It can be concluded
that, with a unit step input, all those systems are stable.

 The undamped oscillates continuously with a constant amplitude and a constant frequency
 The under-damped oscillates and exponentially diminishes the amplitude of oscillation to zero when
the time approaches to infinity.
 Critically damped will die rapidly without oscillating to leave the forced response to control the system
 The over-damped dies quicker than the critically damped
and also leaves the forced response to act alone.

Example of an unstable second order system

The closed loop transfer function of a control system is given

by

9s  147
T s 
s 2  6s  144

Study the stability of the system if the system is run with a

unit step input.

Solution

The characteristic equation is

6  j 540
s1   3  j11.62
2 Figure 5: Location of poles in s-plane for
unstable second order system
6  j 540
s2   3  j11.62
2

The roots of the characteristic equation are located the right side of the Imaginary axis. Therefore, the
system is not stable.

Page 7 of 19
The time response will be found taking inverse Laplace transform of the output signal C(s).

The output is

1 9s  147
C  s  T  s 
s s  s 2  6s  144 

The time response is found by taking inverse Laplace transform of C(s)

 9 s  147 
c  t   L1  
 s  s 2  6 s  144  
 

Using partial fractions method

9s  147 A Bs  D
C s    2
s  s  6s  144  s s  6s  144
2

As 2  6 As  144 A  Bs 2  Ds
C s 
s  s 2  6s  144 

147
144 A  147  A   1.021
144

A  B  0  B  1.021

D  6 A  9

D  9  6 A  9  6 1.021  2.874

The output can be written

1.021 1.021s  2.874 1.021 1.021s  2.874

C s   2   2
s s  6s  144 s s  6s  144

 1.021 
L1    1.021
 s 

The denominator can be written as follows:

Page 8 of 19
s 2  6s  144   s  3  135
2

Therefore,

1.021s  2.874 1.021s  2.874


s 2  6s  144  s  3  135
2

1.021s  2.874 1.021s 2.874

 
 s  3  135  s  3  135  s  3  135
2 2 2

Adding and subtracting 3 from the numerator and rearranging

1.021s s 33 s 3 3
 1.021  1.021  1.021
 s  3  135  s  3  135  s  3  135  s  3  135
2 2 2 2

Using the shift theorem for cosine function

 s 3 
L1 1.021   1.021e3t cos 135t
  s  3  135 
2


Using the shift theorem for the sinusoidal function

3 3 135

 s  3  135 135  s  3  135
2 2
 

 3  1.021 3 1  135 
L1 1.021   L    0.264e3t sin 135t
  s  3  135 
2
135   s  3  135 
2
  

 1.021s 
L1    1.021e3t cos 135t  0.264e3t sin 135t
  s  3  135 
2

The time response

c t 1.021  e3t 1.021cos11.62t  0.511sin11.62t 

Page 9 of 19
The amplitude

X  1.0212  0.5112  1.304

The phase angle

0.511
  tan 1  tan 1 0.5  26.587
1.021

26.587  3.14
  0.464 rad
180

The time response

c  t   1.021 1.304e3t cos 11.62t  0.464

Plotting Matlab program

>> t=55:0.01:60;

>> x=1.021-1.304*exp(3*t).*cos(11.62*t-0.464);

>> plot(t,x),grid

>>

Figure 6: Time response for unstable system

Page 10 of 19
 STABILITY BY ROOT LOCUS TECHNIQUE

Introduction

ROOT LOCUS is a graphical technique that gives a QUALITATIVE DESCRIPTION of a control system ‘s
performance. It is also a powerful QUANTITATIVE TOOL that yield more information. Even if root locus can
be used to evaluate the system’s parameters of FIRST, and SECOND ORDER SYSTEMS, its real power
lies in its ability to provide solution for SYSTEMS OF HIGHER ORDER THAN TWO.

For example, under the right conditions, a fourth order system parameters can be designed to yield a given
percentage overshoot and settling time. The root locus can be used describe QUALITATIVELY the
parameters of the system as various parameters are changed.

For example, the effect of varying the GAIN upon percentage overshoot, settling time and peak time can be
vividly displayed. Beside transient response, the root locus also gives a graphical representation of A
SYSTEM’S STABILITY

Figure 7:Root locus technique

Page 11 of 19
The canonical representation of a closed loop control system is shown by Figure 8

Figure 8: Canonical (generic) form of a closed loop control system

The Forward or Open loop transfer function

NG s  Numerator G  s 
G s   (3)
DG s  Denominator G  s 

The feedback transfer function

N H s Numerator H  s 
H s   (4)
DH  s  Denominator H  s 

If the feedback transfer function H  s   1 , the feedback transfer function is called a UNIT OR UNITARY

FEEDBACK transfer function.

Remarks:

1. The poles of an open-loop transfer function G  s  are easy to find; they are known by inspection and

do not change with the changes of the system gain K

2. The poles of the closed loop transfer function are more difficulty to find.

Page 12 of 19
The gain K is any positive number that is equal to zero or greater than zero  K  0 . Figure 9 shows a closed

loop control system with a gain K in cascade with the system G(s).

Figure 9: A closed loop control system with a gain K

The equivalent system is shown by Figure 10.

Figure 10: A system with a gain

The closed loop transfer function is

NG s 
K
KG  s  DG s 
T s  
1  KG  s  H  s  NG s  N H  s 
1 K
DG s  DH  s 

Page 13 of 19
 NG s 
K
DG s  KNG s   DG  s  DH  s 
T s  
DG s  DH  s   KNG  s  N H  s  DG s   DG s  DH  s   KNG  s  N H  s  
DG s  DH  s 

KNG s   DH  s 
T s  (5)
DG s  DH  s   KNG s  N H  s 

Observations

1. The ZEROS of the closed loop transfer function T(s) consist of the zeros of the open-loop transfer
function G(s) and the poles of the feedback transfer function H(s)
2. The POLES of T(s) are not immediately known without factoring the denominator, and they are
function of the gain K.

Since the system transient response and stability are dependent upon the POLES of the closed loop
transfer function T(s), we have no knowledge of the system performance unless we factor the
denominator for specific values of the gain K.

Example

H  s  1

The closed-loop transfer function is

Page 14 of 19
 K  s  3
G s s  s  1 s  2  s  4 
T s  
1 G  s K  s  3
1
s  s  1 s  2  s  4 

K  s  3
s  s  1 s  2  s  4 
T s 
s  s  1 s  2  s  4   K  s  3
s  s  1 s  2  s  4 

K  s  3 s  s  1 s  2  s  4 
T s  
s  s  1 s  2  s  4  s  s  1 s  2  s  4   K  s  3

Finally, the closed loop transfer function is given by

K  s  3
T s 
s  s  1 s  2  s  4   K  s  3

The characteristic equation of the closed-loop transfer function is obtained by equating its denominator to
zero

s  s  1 s  2 s  4  K  s  3  0

Definition

The root locus is a graphical representation of the closed-loop poles (roots of the characteristic equation) as
the gain K is varied from zero to infinity  0  K    .

The root locus shows the changes in transient response as the gain K, varies from zero to positive infinity

 0    .
K

>> s=tf('s');

>> num=[1 3];

>> den1=[1 0];

>> den2=[1 1];

Page 15 of 19
>> den3=[1 2];

>> den4=[1 4];

>> den34=conv(den3,den4);

>> den=conv(den12,den34);

>> G=tf(num,den)

G=

s+3

--------------------------

s^4 + 7 s^3 + 14 s^2 + 8 s

Continuous-time transfer function.

>> rlocus(G)

Page 16 of 19
Here is an example of how the function rlocus works:

Consider a randomly generated stable Transfer Function Model of the form

G(s)=num(s)/den(s). The command rlocus(tf(num,den))can also be used for construction of root locus of
G(s)

>> num=[0 -0.1022 0.0316 0.1934 -0.1795 0.1620];

>> den=[1.0000 6.2190 50.6538 222.7866 359.5180 162.7478];

>> rlocus(tf(num,den))

Page 17 of 19
Page 18 of 19
Page 19 of 19