School of Engineering

5405ELE Instrumentation and

Control Engineering
Dr Qian Zhang
[email protected]
Room 504A, Byrom Street
Week 5
Stability and Routh-Hurwitz
Criterion
Stability
• A general form of Transfer Functions (ratio of LT of output to LT of input):
Y(s) b m s m  b m 1s m 1  ...  b1 s  b 0
G(s)  
U(s) a n s n  a n 1s n 1  ...  a 1s  a 0
• The denominator polynomial equation is the Characteristic Equation (CE),
i.e.
A(s)  a n s n  a n 1s n 1  ...  a 1s  a 0  0

• The roots of CE, p1, p2, …, pn, are the Poles.

• When is a system stable?

 All poles are in the left half of the s-plane. If one or more poles are in
the right half plane, the system is unstable.
Unit Impulse Response of 1st Order System

y(t) a>0
u(t)
1 a<0
U(s) Y(s) 0 t
0 t 1
(s  a)
G(s)

1 1
Y(s)  U(s)  s-plane s=σ+jω
s a s a jω
From Laplace transforms of impulse
function U(s) = 1.
y(t)=eat 0 as t→∞ if a<0 x x
y(t)=eat ∞ as t→∞ if a>0 0 σ
a<0 a>0
STABLE UNSTABLE
Roots of s - a = 0 are ‘poles’.
Unit Impulse Response of 2nd Order System

Roots (poles) are given from: s2 + as + b = 0 u(t)

 a a2  4 b
s  a1 , a 2  0 t U(s) 1 Y(s)
y(t)
2 s 2  as  b
Case 1: a2 > 4b → two real poles, s = a1, a2 G(s)
and G(s) is:
1 s-plane s=σ+jω
G s =
(s − a1 )(s − a2 ) y(t) jω
y(t)
1 A1 A2
Y s =G s U s = = + 0 t
s − a1 s − a2 s − a1 s − a2 0 t

y t = A1 ea1 t + A2 ea2 t x x x x
σ
a1 a2 0 a1 a2

y(t) → 0 as t → ∞ if both a1 and a2 < 0 STABLE UNSTABLE

y(t) → ∞ as t → ∞ if both a1 and a2 > 0

For a stable solution the poles need to be on left-hand side of the complex plane (LHP)
 Unit Impulse Response of 2nd Order System
 a  j 4 b a 2
Case 2: a2 < 4b → complex conjugate pair s  s1 , s 2 
2
1
In this case G(s) can be written as: G(s)  where ξ  1
s 2  2 ξω n s  ω 2n

where ξ is the damping ratio and ωn is the undamped natural

frequency, and the characteristic equation is: s-plane s=σ+jω
s 2  2 ξω n s  ω 2n  0 y(t) jω y(t)

Poles are 
s1 , s 2  ω n  ξ  j 1  ξ 2  0
x
t 0
x
t

1 0 σ
Y(s)  U(s) G(s)  2 x
s  2 ξω n s  ω 2n x
-ξωn < 0, poles have -ξωn > 0, poles have
From Laplace transform table: -ve real parts +ve real parts
1 ‘STABLE’ ‘UNSTABLE’
y(t)  e  ξω n t sin(ω n 1  ξ 2 t)
ωn 1  ξ2
y(t) is an exponentially decaying sinusoidal if ξ >0, or an exponentially expanding sinusoid if ξ<0
 Unit Impulse Response of nth Order System
• The poles are given by the ‘n’ roots of:
A(s)  a n s n  a n 1s n 1  ...  a 1s  a 0  0
which is a mixture of real poles and complex conjugate pairs, giving (neglecting any repeated
roots):
Ai Ai
Y(s)    2
i s α i i s  2 ξ ω
i ni s  ω 2
ni

From Laplace transform

y(t)   A i e   Ai e
a i t  ξ i ω ni t
sin(ω n i 1  ξ i2 t)
i i

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

0 t 0 t 0 t

STABLE MARGINALLY UNSTABLE

STABLE
Location of Poles and Stability
s-plane
s=σ+jω
jω
y(t)
y(t)

t
0 0 t
0
x x
x

x x x
σ
0
x
x
x
y(t) y(t)
1 y(t)

0 0 t
t 0 t
‘STABLE’ ‘UNSTABLE’

Location of Poles and Stability

Location of Poles and Stability: Examples
Location of Poles and Stability: Examples
3. Example 1 under unity negative feedback with a proportional controller gain K. What
values of K are required for stability of the closed loop system?
Routh-Hurwitz Stability Criterion - Polynomial Testing
• Routh’s criteria is a method for assessing stability without finding roots.
• The criteria can be used to determine the range of gain for which the closed-loop
system is stable.

Given the characteristic equation:

Q(s)  a n s n  a n 1s n 1  a n  2 s n 2  ...  a 1s  a 0  0
determine the number of unstable poles (i.e. roots of Q(s) = 0 with +ve real
parts

Notes
(i) If a0 is zero divide by s to obtain Q(s) in the right form.
(ii) Ensure that an is positive; if not multiply the characteristic
equation by -1.
Routh-Hurwitz Stability Criterion - Polynomial Testing
Forming the Routh array
The coefficients of the characteristic equation are first arranged in the pattern shown to form the
first two rows of the array. These coefficients are then used to evaluate the rest of the entries.

sn an an-2 an-4 an-6 …

an an2 an an4
sn-1 an-1 an-3 an-5 an-7 …
a n1 a n3 a n1a n 2  a n a n3 a n1 a n5 a n1a n 4  a n a n5
C1    C2   
sn-2 C1 C2 C3 … where
a n1 a n1 a n1 a n1
a n1 a n3 a n1 a n5
sn-3 d1 d2 …
C1 C2 C1a n3  a n1C 2 C C 3 C1a n5  a n1C 3
: : d1    d2   1 
C1 C1 C1 C1
: :
s1 j1
s0 k1 etc etc

• The number of roots with +ve real parts is equal to the number of changes in sign of the
• coefficients in the first column.
• The system is stable if all terms in the first column are positive.
Routh-Hurwitz Stability Criterion - Testing Stability
Example 4: Q(s) = s5 + s4 + 3s3 + 9s2 + 16s + 10 = 0

s5 1 3 16 1 3 1 16
1 9 1×9−3×1
𝑐 = − 1 10 = − 1 × 10 − 16 × 1 = 6
𝑐1 = − =− = −6 2
s4 1 9 10 1 1 1 1
s3 -6 6 1 9 1 10
1 × 6 − 9 × (−6) −6 0 = 1 × 0 − 10 × (−6) = 10
s2 10 10 𝑑1 = − −6 6 = = 10 𝑑2 = −
−6 6 −6 6
s1 12 −6 6
10 10 (−6) × 10 − 6 × (10)
s0 10 𝑒1 = − =− = 12
10 6
10 10
12 0 10 × 0 − 10 × 12
𝑓1 = − =− = 10
12 12

Actual roots (poles) are: 1+2j

There are two left hand column sign changes. Hence, 1-2j
the polynomial has two right hand plane roots and the -1+j
system is unstable -1-j
-1
Routh-Hurwitz Stability Criterion – Finding Parameter Range
What is the maximum value of the adjustable gain K if the system
Example 5:
with the given characteristic equation is to remain stable?
Q(s) = s3+ 2s2 + 4s + K

s3 1 4 1 4
𝑐1 = − 2 𝐾 = −1 × 𝐾 − 4 × 2 = 8 − 𝐾
s2 2 K 2 2 2

s1 (8-K)/2 0 2 𝐾
(8 − 𝐾)/2 0 2 × 0 − 𝐾 × (8 − 𝐾)/2
𝑑1 = − =− =𝐾
s0 K (8 − 𝐾)/2 (8 − 𝐾)/2

8−𝐾
For a stable system: 𝑐1 = >0⇒𝐾<8 𝑑1 = 𝐾 > 0
2
Thus, 0<𝐾<8
Routh-Hurwitz Stability Criterion – Finding Parameter Range
Example 6: What is the maximum value of adjustable gain K if the system is to remain stable?
+ K 10
R(s)  s s  s  10
2 Y(s)
-
1
s2

The closed-loop characteristic equation is:

K 10 1
1  2  0
s s  s  10 s  2
s ( s  2)( s 2  s  10)  10 K  0
Q ( s )  s 4  3s 3  12 s 2  20 s  10 K  0

Employ Routh’s criteria

Routh-Hurwitz Stability Criterion – Finding Parameter Range
Example 6:
1 12
Q(s) = s4 + 3s3 + 12s2 + 20s + 10K = 0 1 × 20 − 12 × 3 16
𝑐1 = − 3 20 = − =
3 3 3
1 10𝐾
3 0 1 × 0 − 10𝐾 × 3
s4 1 12 10K 𝑐2 = − =− = 10𝐾
3 3
s3 3 20 0 3 20
16 16
16/3 10K 10𝐾 3 × 10𝐾 − 20 × 3 320 3
s2 𝑑1 = − 3 =− =( − 30𝐾) ×
16 16 3 16
2 3
s1 (106  30 K ) 3 3
3 16

s0 10K 16/3 10𝐾

𝑑1 0 16/3 × 0 − 10𝐾 × 𝑑1
𝑒1 = − =− = 10𝐾
𝑑1 𝑑1
The stability range for K is:
2
(106  30 K )  0 so that K  3.5556
3
and K  0
Routh-Hurwitz Stability Criterion – Finding Parameter Range
Example 6:
Homework: Use MATLAB and/or Simulink simulations (as follows)
to investigate the stability of the system