You are on page 1of 19

Chapter 6: Stability

3 requirements of a control system study:


1. Transient response Done

2. Steady-state response Done


3. Stability This chapter

Definitions: (Applicable to LTI system only)


In terms of natural response:
(1) A system is stable only if the natural response approaches zero as time
approaches infinity.
(2) A system is unstable if the natural response approaches infinity as time
approaches infinity.
(3) A system is marginally stable if the natural response neither decays
nor grows but remain constant or oscillates.
1
In terms of total response: (Also called BIBO definition)
(1) A system is stable if every bounded input yields a bounded output.
(2) A system is unstable if any bounded input yields an unbounded output.

Stability in terms of Pole locations:



excluding
1. Stable system: has closed-loop transfer function with imaginary axis
poles only in the left-half-plane (LHP) pole

Only LHP poles  stable system
2. unstable system:
a) Any RHP pole  unstable system jω

b) Also more than 1 pole on the jω-axis lead to the sum >1
of responses of the form: 
At n cost   , n  0, 1, .....
This also results in instability. 2
3. Marginally stable system: One pole is on the imaginary jω
axis and other poles (if any) are on the LHP. =1

It is not always a simple matter to determine whether a F/B control system is


stable.

Although we know the poles of the forward TF in the above figure, but we do
not know the pole locations of the equivalent closed-loop TF shown below.

So, finding pole locations without factoring the denominator is not easy.
3
Consider the system’s closed-loop TF to be of the form:
ns  ns 
Ge  T  
d s  an s n  an1s n1  .........  a1s  a0

Under certain conditions, we can draw some conclusion about the stability of
the system:

Sufficient Condition for a system to be unstable:


1. All coefficients an, an-1, …. a0 are not of the same sign
OR
2. Any coefficient is zero (BIBO unstable)  can be at best marginally stable

Ques: How many poles are on the RHP, LHP and jω-axis?

Routh-Hurwitz Criterion for Stability can answer to this question.


Ans:
Step 1: Get the closed-loop TF and generate Routh table.

4
ns  ns 
T 
d s  a4 s  a3 s  a2 s 2  a1s  a0
4 3

Initial table layout:

Now fill the remaining fields as shown below:

5
Step 2: Apply the following rule. This is called, “Routh-Hurwitz” Criterion.

“The number of roots of the polynomial d(s) that are in the RHP is
equal to the number of sign changes in the first coefficient column of
the table”

Example 6.1: Make the Routh table for the system shown below:

Sol.:
Step 1: find the equivalent closed loop TF and generate Routh table.
Equivalent TF is shown here:

G
This is found by using the relation: T
1 G
here, H=1

Here, d s   s 3  10s 2  31s  1030 6


The Routh Table: here, d s   s 3  10s 2  31s  1030

10 1 1030 103 0 0

 72 0 0

 103 0 0

Step 2: Apply the “Routh-Hurwitz” Criterion.

There are two sign changes in the first column. So there are 2 poles in the
RHP.
So, the system is unstable.

7
Special Cases:
1. Zero in the first column (not the entire row is zero)
Problem: Division by 0
Solution: Replace 0 by a very small number () and compute the table
fields in standard way.
Example 6.2: Determine the stability of the closed-loop transfer function

Sol.: here, d s   s5  2s 4  3s3  6s 2  5s  3


Routh Table-

1 3 1 5 1 0
7
0  0
2 6
 
2 3
 
2 0
2 2 2 2

0 0

3 0 0 8
Determine sign of the 1st column for positive and negative .

 
 
 

 

 

 

Observation: 2 sign changes for both cases.

Result: Two poles are on the RHP

9
Special Cases:
2. Entire row is zero (AZR: All-Zero Row)
Example 6.4: Determine the number of right-half-plane poles in the closed-loop
transfer function

Sol.: Here, d s   s5  7s 4  6s3  42s 2  8s  56


Routh Table- If an AZR appears, create an
auxiliary polynomial with the
previous row. Start with the power
7 1 42 6 56 8 of ‘s’ in the index column and
include alternate powers thereafter.
0 4 1 0 12 3 0 00

3 8 0 Here, auxiliary polynomial:


1 Ps   s 4  6s 2  8
0 0
3 dP
 4s 3  12s  0
8 0 0 ds
Now, replace AZR with these
coefficients (4, 12 and 0) 10
Observation: All entries in the in the first column are positive, so no sign
changes.
Result: There are no RHP pole.

Now question, ‘Is it possible to predict the no. of poles on the jω-axis from
Routh table? ’
Yes. Here is how it is.

“An AZR can occur only if a purely even polynomial is a factor of the original
polynomial.”
Example of even poly: s 4  5s 2  8
A Roots are symmetrical & real

Even poly has roots B roots are symmetrical &


only symmetrically imaginary
placed about origin
C roots are symmetrical &
complex. So qudrantal.
11
 The row previous to the AZR contain this even polynomial.
 Everything from the row containing the even poly downward is a test of
only the even poly.
 Rows above the even poly is a test for the other poles.

How to check for no. of jω-axis poles?


- Check if there is an AZR
NO  there are no jω-axis poles
Yes 
no of jω-axis poles = order of the even poly – 2  (no. of sign
changes in the
even poly part of
the table)

12
Example 6.5: For the transfer function

tell how many poles are in the right half-plane, in the left half-plane, and on the
jω-axis.

The auxiliary polynomial which is an even polynomial, is

So,
13
Location of the poles of the even poly (4th-order)

No sign change in the first column. So no pole on the RHP.


Thus, no. of jω-axis poles = order of the even poly – 2  (no. of sign changes in the
even poly part of the table)
 40
4
Location of other poles (4th-order)

Two sign change in the first column. So two poles are on the RHP. The remaining two
poles are on the LHP.
14
Summary of pole locations:

Results: The system is unstable.

15
Example 6.9: Stability Design via Routh-Hurwitz
Problem: Find the range of gain, K, for the system of Figure below to be stable,
unstable, and marginally stable. Assume K > 0.

Sol.: The closed-loop TF,


The Routh Table:
s3 1 77

s2 18 K
1 77 1 0
1386  K
0
18 K 18 0
s1   
18 18 18

s0 K 0

Since K is assumed positive, all elements in the first column are always positive,
except the s1 row. It can be positive, zero or negative depending on the value of
16
K.
(1) Stable: No sign change in the 1st column.
1386  K K  1386
0 Thus,
18

So, the system is stable for 0  K  1386


(2) Unstable: There will be at least a single sign change. Since, K>0,
1386  K
0 Thus, K  1386
18

In this case, two sign changes. So, 2 RHP and 1 LHP.


So, the system is unstable for K  1386

(3) Marginally stable: There should be an AZR. It may signify jω-axis poles.
1386  K Thus, K  1386
0
18

The auxiliary even polynomial: Ps   18s  1386


2

dPs 
Differentiating w.r.t. s, we have,  36s  0
ds

Now, replace the AZR row with the coefficients of the differentiated poly.
17
So, now the Routh Table for K=1386:

0 0
0
Location of the poles of the even poly (2nd-order)

No sign change in the first column. So no pole on the RHP.


Thus, no. of jω-axis poles = order of the even poly – 2  (no. of sign changes
in the even poly part of
 20 the table)
2

Location of other poles (1st-order)

No sign change. So, no poles are on RHP. 1 pole is on LHP.


18
For K=1386, 2 poles lie on jω-axis and 2 poles are on the LHP. So, the system is
marginally stable.

19

You might also like