5.

Stability of Control Systems

• A system is stable if finite input produces a
bounded or finite output.
• For example for every step input applied to a
system the out put must be finite.
• When it is subjected to an impulse input the
output dies away to zero as time goes infinity.
• If the output tends to infinity as t approaches
infinity, then the system is unstable.
• If the output does not die out to zero or
increase to infinity but tends to some finite but
non-zero value then the system is said to be
critically or marginally stable.
Chapter 5 1
Poles and Zeros
• The closed-loop transfer function G(s), generally
be represented by:
k ( s m  am1s m1  am2 s m2  ....  a1s  ao
G(s) 
( s n  bn1s n1  bn2 s n2  ....  b1s  bo
Equivalenty
k ( s  z1 )( s  z2 )....( s  zm )
G(s) 
( s  p1 )( s  p2 )....( s  pn )
Where: z1, z2, ….,zm are zeros
• values of s for which G(s) = 0
p1, p2, ….,pn are poles
• values of s for which G(s) = 
k - Gain of the system
Chapter 5 2
• Generally poles and zeros can be real or
complex. s    j
Examples:
s 1 s 1
a) 2   poles are +2 and +2 ; zero +1
s  4 s  4 ( s  2) 2

1 1  1  4 1  j 4  1
b) 2 ; s1,2    0.5  j 0.87
s  s 1 2 2 jω
2
Pole-zero plots 1

• Poles are marked with „x‟ σ

-2 -1 1 2
and zeros with „o‟ -1
-2

Chapter 5 3
Examples: 1 1
1) G ( s )   o ( s )   ( s)
s2 s2 i
1
if i ( s )  1  o ( s )   o (t )  e 2t
s2
1 1
2) G ( s )   o ( s )   ( s)
s2 s2 i
1
if i ( s )  1  o ( s )   o ( t )  e 2 t
s2
1
3) G ( s )   o ( s )  G ( s )i ( s )
[ s  ( 2  j )][ s  ( 2  j )]
if i ( s )  1  o ( s )  G ( s)  o (t )  e 2t sin t
jω jω jω
2 2 2
1 1 1

σ σ σ
-2 -1 1 2 -2 -1 1 2 -2 -1 1 2
-1 -1 -1
Fig.1 -2 Fig.2 -2 Fig.3 -2
Chapter 5 4
• In general when an impulse is applied to a system
the output is in the form of the summation of a
number of exponential terms.
• The following is possible to happen.
– If just one of the exponential terms is of an
exponential growing type, the system is unstable
• (i.e if any one of the poles has a +ve real part)
– If all the poles are on the left side of the pole-zero
plot or s-plane, then the system is stable.
– If just one pole is in the right hand side it is
unstable.
– If one or more poles lie on the vertical axis of a
pole-zero plot, then it is Critically stable.
Chapter 5 5
Summary 1

Figure: Pole location corresponding to three values of ς

Chapter 5 6
 Summary 2

Figure: Time functions associated with points in the S-plane

(LHP, left half-plane; RHP, right half-plane)
Chapter 5 7
5.1. The Routh-Hurwitz stability criterion
• The roots of the characteristic equations of a
system may not be well easily obtained if it is in
the form
ansn + an-1sn-1+an-2sn-2+ ----+a1s+ao, n>3
Routh-Hurwitz Criterion
1. Inspect the coefficients
– If all are +ve and none of them is zero then the
system can be “Stable”.
– If any coefficient is negative then the system is
definitely “Unstable”.
– If any coefficient is zero then the system is most
probably “Critically Stable”.
Chapter 5 8
• Examples:
 s3+2s2+ 3s + 1  can be stable
 s3 -2s2+ 3s + 1  unstable
 s3 +2s2+ 3s  at the best, critically stable
• For those that could be stable, the coefficients are
written in order called Routh Array.
sn an an-2 an-4 …..

Sn-1 an-1 an-3 an-5 …..

• Further rows are calculated from elements in the

two rows immediately above. Successive rows
are calculated until only zeros appear.
Chapter 5 9
 s n an an-2 an-4 an-6
sn-1 an-1 an-3 an-5 an-7
sn-2 b1 b2 b3 …
sn-3 c1 c2 c3 ….
. .
. .
s2 x1 x2 x3
s1 y1 y2
s 0 z1

Chapter 5 10
• If all the elements in first column of the array are
+ve, all the roots have negative real part hence a
stable system.
• If there are any negative elements, the number of
sign change is equal to the number of roots with
+ve real parts, that is Unstable system.

Examples…………..

• Note: Routh criterion gives only the number of

roots in the right half of the s-plane. It gives no
information in regard to the values of the roots &
does not distinguish between real & complex
roots.
Chapter 5 11
Special cases
Difficulty 1:
• When the 1st term in any row of the Routh-array
is zero while rest of the row has at least one non-
zero term.
Methods
1. Substitute a small +ve number ε for the zero & proceed to
evaluate the rest of the Routh-array. Then examine the
signs of the 1st column by letting ε →0.
2. Modify the original characteristic equation by replacing s
by 1/z. Apply the Routh‟s test on the modified equation
interms of z. The number of z-roots with +ve real parts
are the same as the number of s-roots with +ve real parts.
This method works in most but not all cases.
Example……………
Chapter 5 12
Difficulty 2:
• When all the elements in any one row of the Routh-
array are zero.
• This condition indicates that there are symmetrically
located roots in the s-plane (pair of real roots with
opposite signs and/or pair of conjugate roots on the
imaginary axis and/or complex conjugate roots
forming quadrates in the s-plane)
• The polynomial whose coefficients are the elements of
the row just above the row of zeros in the R - array is
called an auxiliary polynomial. This polynomial gives
the number and location of root pairs of the
characteristic equation which are symmetrically
located in the s-plane. The order of the auxiliary
polynomial is always even.
Chapter 5 13
Method
• Replace the row of zeros in the R-array by a raw
of coefficients of the polynomial generated by
taking the 1st derivative of the auxiliary
polynomial.

Examples……………

Chapter 5 14
5.2. Root Locus Technique
• Possibility of unstable operation is inherent in all
feedback control systems, because of the very
nature of feedback systems.
• Hence, the determination of stability of a system
is necessary but not sufficient, because a stable
system with low damping is still undesirable.
• Therefore, we should proceed to determine the
relative stability of a stable system after we make
sure the absolute stability.

Chapter 5 15
• Relative stability in terms of:
– peak overshoot
– Settling time,
• It is directly related with the location of the
closed loop poles of a stable system.
Design problem
• Achieving the desired performance of a system
by adjusting the location of its closed-loop poles
in the s-plane.
• This will be possible by varying one or more
system parameters.

Chapter 5 16
Possible solution
• Routh's criterion does not help to solve this.
• Classical technique (factorizing & determining
roots of the characteristic equation) is laborious
for higher degrees greater than 2.
• Root Locus technique (by W.R. Evans) is the best
solution.
• The technique Provides graphical method of
plotting the locus of the roots in the s-plane as a
given system parameter is varied over the
complete range of values (i.e. 0 to ∞)

Chapter 5 17
• Roots corresponding to a particular value of the
system parameter or the value of the system
parameter for a desired root location can be
determined from the locus. And it:
is the best way to analyze a system‟s dynamic
response

also provides a measure of sensitivity of roots

to the variation in the parameter

is applicable for single as well as multiple

loop systems

Chapter 5 18
Root Locus
• Consider the following simple 2nd Order system.
R(s) +
∑
k C(s)

- s( s  a )

C ( s) k
  2
R( s) s  as  k Characteristic Equation

a  a
2

s1,2       k   always stable for +ve a and k
2  2  
• But the dynamic behavior is controlled by the
magnitudes of a & k.
Chapter 5 19
• Consider variable gain k (common practice)
while a is held constant.
1. 0 ≤ k < a2/4 → real & distinct roots. Specially,
When k = 0 → s1 = 0 and s2 = -a and this are
called the open loop poles.

2. k = a2/4 → real and equal roots, s1=s2=-a/2.

3. a2/4 < k <  → complex conjugate roots

with unvarying real part (i.e. -a/2)

Chapter 5 20
Root Locus Plot
∞
↑ • Roots lying on the –ve
k jω
k> a2/4
real axis corresponds to
s an over-damped system,
k=0 θ2 -a/2 θ1 (i.e. 0≤ k <a2/4).
-a σ • k = a2/4 → critically
k = a2/4 k=0 damped system.
k
↓
Break-away
Point
• k > a2/4 → Under
k > a2/4 ∞ damped system.

• Roots move along the line σ = -a/2.

• The above locus is plotted from direct solution of the
characteristic equation and is highly tedious for
higher order greater than 2.
Chapter 5 21
Evans simplified system
• Characteristic equation →1 + P(s) = 0
– where P(s) = G(s)H(s)
– G(s)H(s) is termed as Open loop Transfer Function
(in block diagram terminology) and Loop
Transmittance (in signal flow graph terminology)

• 1+ P(s) = 0 → P(s) = -1

 P( s)  1 and
P( s )  180o (2q  1); q  0,1, 2,...
• These are the two Evan‟s Conditions.
Chapter 5 22
• A plot of the points in the complex plane that
satisfy this angle criterion is the Root Locus. And
a gain corresponding to a root (point on the root
locus) can be determined from the magnitude
criterion.

Example:

Chapter 5 23
Construction of Root loci
• 1 + G(s)H(s) =0
• G(s)H(s) is generally known in the factored form
as it is obtained by modeling the TF of individual
components comprising the system
m
  z s  1
i
 1  G ( s ) H ( s )  1  P( s)  1  k ' i 1  0,
  p 
n

j
s 1
j 1
m
  s  zi 
or 1  P(s)  1  k i 1 0
n
 s  p j 
Chapter 5 j 1 24
 n m m n
Where, k= k '  p j  zi  k '  z  p
i j
j 1 i 1 i 1 j 1

• k and k‟ are the open loop gains.

• -zi=-1/ zi (i=1,2,3,…,m) are the zeros of P(s).
• -pi=-1/ pj (j=1,2,3,…,n) are the poles of P(s).
• Note: In all physically real systems n>m,
– i.e. number of poles of P(s) > number of zero of P(s)

Chapter 5 25
• The Evans condition will then be:
m
 s  zi
Magnitude criterion: k i 1  1 and
n
 s  pj
j 1
m n
Angle criterion:  ( s  zi )   ( s  p j )  (2 q  1)180o ; q  1,2,3,...
i 1 j 1

• Follow a trial and error procedure to satisfy the

angle criterion for a point on the root locus.
• After determining many loci points in this
manner, draw a smooth curve through these pts.
Chapter 5 26
• The value of k for particular location of so, from
the magnitude criterion.
n
 so  p j
j 1
k  m
 so  zi
i 1
Products of phasor lengths from so to open loop poles
k 
Products of phasor lengths from so to open loop zeros

• This is a tedious procedure.

• For quick approximate sketch of root locus the
following construction rule could be used.
Chapter 5 27
Construction Rules
• Provides a guide for selection of a trial point such
that more accurate root locus can be obtained by
few trials.
Rule 1:
• The root locus is symmetrical about the real axis.
– Proof: Roots are either real or complex conjugates or
combination of both.
Rule 2:
• As k increases form 0 to ∞, each branch of the
root locus originates from an open-loop pole
with k=0 and terminates either on an open-loop
zero or on infinity with k=∞.

Chapter 5 28
• The number of branches terminating at infinity
equals the number of open-loop poles minus
number of zeros.

– Prove:

Rule 3:
• A point on the real axis lies on the locus if the
number of open-loop poles plus zeros on the
real axis to the right of this point is odd.

– Prove:

Chapter 5 29
Rule 4:
• The (n-m) branches of the root locus which tend
to infinity, do so along straight line asymptotes
whose angles are given by:
 2q  1 180
A  ; q  0,1,2,...,(n  m  1)
nm
– Prove:
Rule 5:
• The asymptotes cross the real axis at a point
known as centroid, determined by the relation:
(sum of real parts of poles-sum of real parts of zeros
(number of poles-number of zeros)
─ Example………..
Chapter 5 30
Rule 6:
• The breakaway points of the root locus are the
solutions of dk/ds=0.
– Defn.: Breakaway points are points at which
multiple roots of the characteristic equation
occur.
Note:
i. Since the characteristic equation can have real
as well as complex multiple roots, its root locus
can have real as well as complex breakaway
points, but because of conjugate symmetry of
the root loci, the breakaway points must either
be on the real axis or occur in complex
conjugate pairs.
Chapter 5 31
ii. All the roots of equation dk/ds=0 are not
breakaway points. The actual breakaway points
are roots at which the root locus angle criterion
is met.

Breakaway directions of Root locus Branches

• The root locus branches must approach or leave
the breakaway point on the real axis at an angle
of +180o/r, where r is the number of root locus
branches approaching or leaving the points.
– Example………

Chapter 5 32
Rule 7:
• The angle of departure from an open-loop pole
is given by:
p = + 180o(2q+1)+, q= 0,1,2, …
– where  is the net angle contribution, at this pole,
of all other open-loop poles and zeros.
• Similarly, the angle of arrival at an open-loop
zero is given by:
z = + 180o(2q+1)-, q= 0,1,2, …
– where  means the net angle contribution at the
zero under consideration of all other open-loop
poles & zeros.
Chapter 5 33
• The angle of departure from a real open-loop
pole or the angle of arrival at a real open-loop
zero is always 00 or 1800. Hence, no need to
calculate these angles for real pole and zero, but
need for complex poles & zeros.

Rule 8:
• The intersection of root locus branches with the
imaginary axis can be determined by use of the
Routh criterion.
-Illustration………
-Example………
Chapter 5 34
5.3. Systems with Transportation lag
• For the feedback system (which are assumed to
be linear, lumped parameter models) the time
taken for energy transmission is negligible, i.e. the
output begins to appear immediately on
application of input.
• For transmission systems like channel–lines, pipes,
belt conveyors definite time elapses after
application of the input before the output
becomes effective.
• This type of pure time-lag is known as
transportation lag or dead time.
Chapter 5 35
• Note: Linear lumped parameter model is not
valid under such situations unless the pure time
lag is negligible compared to other lags in the
system
• Consider a component whose output is the same
as the input but delayed by T sec

co (t )  ci (t  T )
 Co ( s )  Ci ( s )e  sT
Co ( s )
  e  sT , non minimum phase function
Ci ( s )

Chapter 5 36
Example:
• Consider the following still plate thickness
controlling system
– Reference input: voltage signal corresponding to the
desired thickness.
– Feedback: thickness gauge provides a feedback voltage
signal proportional to actual plate thickness.
– Error voltage: actuates the motor which positions the
rolls.
Reference + Amplifier Motor
Voltage ∑
-
Feedback Thickness Gears
voltage Gauge Controlled
signal Roll

Fixed Roll
Chapter 5 37
• The open-loop transfer function of the system is
given by: k1
G( s) H ( s) 
s( m s  1)
• Here, transportation lag is ignored.
• The reality is that a finite time must elapse before
a change in thickness of steel plates between rolls
reaches the point of measurement at the gauge.
• Hence, assume the delay time is T sec.

k1e  sT ke  sT
G( s) H ( s)  
s( m s  1) s( s   )
k1 1
where k  and  =
Chapter 5
m m 38
• For a very small transportation lag,
e sT  (1  sT ) Taylor's Approximation
k ( s  1 T )
 G( s) H ( s) 
s( s   )
• Hence, the characteristic equation will be:
k (s  1 T )
 1  G( s) H ( s)  1   1  P( s )  0
s( s   )
 P( s )  1
• Now, the angle criterion is modified as:
P( s)  (2q)180o , q  0,1,2,...
• This leads to the following modification in
construction rules of 3,4 and 7.
Chapter 5 39
i. Replace “odd number” by “even number” for
nr+mr values to locate the on-locus and not-on-
locus.
ii. Replace “180o(2q+1)” by “180o(2q)”.
• The root locus plot for the above equation
taking α=2 and T=1sec is shown below.
jω

-0.732
k=0 k=0 k=∞ k→∞
-2 -1 1 σ

Chapter 5 40
 End of Chapter Five

Chapter 5 41