Professional Documents
Culture Documents
B. S. Manke - Linear Control Systems
B. S. Manke - Linear Control Systems
CONTROL SYSTEMS
with MATLAB Applications
B.S. Manke
Formerly, Prof. of Electrical Engineering,
Maulana Azad National Institute of Technology
Bhopal (M.P.)
KHANNA PUBLISHERS
Operational Office
B-35/9, G.T. Karnal Road, Industrial Area,
(Near Telephone Exchange), Delhi-110033
Phones : 011-27224179 • Mob. 09811541460
email : contactus@khannapublishers.in
24
Published by :
Romesh Chander Khanna & Vineet Khanna
for KHANNA PUBLISHERS
2-B, Nath Market, Nai Sarak
Delhi- 110 006 (India)
Website : www.khannapublishers.in
To all readers of our books, from yourself if being defrauded by pirates to prevent, please make sure that
there is an Hologram on the cover of our books with the below specifications. If you find any book without
Hologram and Description, please mail us at contactus@khannapublishers.in
Thanking you
2 Channel Effect
“K” , “P”
25
To my grandsons
Aadhit, Abhyudit
and granddaughters
Ira, Nitya
Preface to the First Edition
This textbook has been written to explain the basic principles of Linear Control Systems
and an effort is made to present the subject in a simple and sequential manner to enable the
students to acquire a good grasp of fundamentals of the subject.
The text presented covers the course content of the subject Linear Control Systems of
Indian Universities and is meant for pre-final/final year students of electrical, electronics and
mechanical engineering.
The material given in this book has been thoroughly class tested by the author while
teaching the subject of control systems at undergraduate level for the past several years.
This book is divided in 9 chapters. The first four chapters give the basic concepts of the
subject from the view point of control system representation. Chapter 5 presents the
modelling of control systems and the respective mathematical models derived therein. The
time response and steady state analysis is given in Chapter 6. Necessary derivations have been
derived from the first principles. The stability analysis is described in Chapter 7. The methods
of ascertaining stability: Routh-Hurwitz criterion, Nyquist criterion, Bode plot and root locus
plot have been explained step by step in a simplified manner to make the explanation easily
understandable. The compensation methods and introduction to state space analysis is
described in chapters 8 and 9 respectively.
Suitable illustrative examples as well as solved examples have been incorporated in the
text to make the subject clear and interesting. A list of references is given at the end.
Selective unsolved problems have been included at the end of each chapter to help the
student to judge himself whether he has gained sufficient workable knowledge of basic
principles involved. Answers to odd numbered problems being given in Appendix I.
The salient feature of this book is the inclusion of objective type multiple choice questions
given in Appendix II covering the entire text which would be of great help for the students
preparing for competitive examinations.
The author hopes that this book will serve the purpose of introducing basic principles of
Linear Control Systems to undergraduate students for whom it is written.
The author would welcome any comments and suggestions to further improve the
usefulness of this book.
The author acknowledges his indebtedness to Miss Saroj Rangnekar, Asst. Prof. in Elect.
Engg., Maulana Azad College of Technology, Bhopal who thoroughly checked the manuscript
and made useful suggestions.
The text written in the book deals with the concepts of feed-back control theory. The first
five chapters stress on the fundamental concepts regarding representation and modelling of a
control system. The subsequent chapters deal with the time response analysis, stability
analysis, compensation method, state variable approach, and sampled data/discrete data
systems.
Each chapter contains solved examples to support the theory developed. Unsolved
problems have been included as an exercise.
The addition of analytical approach to derive the equation in s-plane for the root locus
plot, can be used to calibrate the complex part of the root locus.
The root locus approach towards the design of compensators have been included in the
text.
The answers to graphical solutions may slightly deviate due to graphical errors.
The chapter on computer solutions to control problems gives the use of MATLAB*
software. The examples on various topics in the text have been solved using MATLAB software.
This verifies the answers obtained using analytical solution.
Appendices given at the end of the book include :
Appendix I : Answers to Selected Problems
Appendix II : A Set of Objective Questions
Appendix III : Short Answer Type Questions
Appendix IV : List of Key Formulae, Charts and Calculation Tables
Any constructive suggestion for further improvement of the book will be appreciated.
The author acknowledges help given by Dr. D.M. Deshpande, ex. Prof. M.A.N.I.T. Bhopal for
discussions with him towards the revision of this edition.
The author is thankful to Shri Kratu Khanna and staff of the Khanna Publishers, New Delhi
for their role in bringing out this revised edition of this book.
(vi)
Contents
1. INTRODUCTION 1 15
1.1 An Example of Control Action ... 2
1.2 Open-Loop Control System ... 2
1.3 Closed-Loop Control System ... 2
1.4 Use of Laplace Transformation in Control Systems ... 3
1.5 Laplace Transform ... 3
1.5.1 Derivation of Laplace Transform ... 4
1.5.2 Basic Laplace Transform Theorems ... 6
1.6 Solved Examples ... 7
Problems ... 15
2. TRANSFER FUNCTION 16 24
2.1 Poles and Zeros of a Transfer Function ... 17
2.2 Transfer Function and its Relationship with Impulse Response ... 19
2.3 Procedure for Determining the Transfer Function of a Control System ... 19
2.4 Solved Examples ... 20
Problems ... 24
3. BLOCK DIAGRAMS 25 61
3.1 Block Diagram Reduction ... 26
3.2 Solved Examples ... 32
Problems ... 58
(vii)
(viii)
Chapter Pages
Chapter Pages
Chapter Pages
Chapter Pages
Chapter Pages
Contains highly graded solved problems selected from question papers of various
Technical Universities, Competitive Examinations and self developed variety of
questions for clearly and deeper understanding of matter/topics covered in text.
References 769
Index 770 772
Introduction
T
his chapter deals with basic ideas about the open-loop and closed-loop control systems. The
differential equations describe the dynamic operation of control systems. The Laplace transform
transforms the differential equation into an algebraic equation, the solution is obtained in the transform
domain. The time domain solution is determined by taking the inverse Laplace transform.
CONTENTS
CONTROL SYSTEM
1
2 LINEAR CONTROL SYSTEMS
Input Output
Control
r c
Action
Open-Loop Closed-Loop
σ
ω
z
L ≥
∞
f (t) ⋅ e − st dt
0
L
L
L L L
∞ ∞ 1
L e at ⋅ e − st dt = e( a − s) t dt =
0 0 (s − a)
1
∴ L
(s − a)
L
LM 1 OP
1
N (s − a) Q
(s − a)
L
∴
1
L
(s − 0)
LM 1OP
1
L
N sQ
s
L
ω
ω
ω
∴
ω
L
3
(s − jω)
ω ω
ω
1 s + jω
∴ L ω
ω = 2
(s − jω) (s + ω 2 )
s
L ω
(s + ω 2 )
2
LM OP
ω
L
ω
(s 2 + ω 2 )
N (s Q
LM OP
s
L ω
2
+ ω2)
N (s + ω )Q
ω
L
ω
2 2
α
ω
α
ω
∴
α
ω
INTRODUCTION 5
1
α
ω
1
L =
3
s − (− α + jω) (s + α) − jω
α ω α ω
ω
α
1 (s + α) + jω
∴ L ω
ω =
(s + α) − jω (s + α) 2 + ω 2
α
( s + α)
L ω
(s + α) 2 + ω 2
LM (s + α) OP
α
ω
L
ω
N ( s + α) + ω Q
(s + α) 2 + ω 2
LM ω OP
α
L ω
2 2
N ( s + α) + ω Q
α
L
ω
2 2
∴
1
L
(s − 1)
3
t2 t3
+ 1+ t +
+ ...
∠ 2 ∠3
1 1 1 1 1
+ 2 + 3 + 4 + ...
(s − 1) s s s s
1
s
1 − sT
e
s
1
s2
2 1
t
2 s3
∠
+
1
s+a
1
s−a
6 LINEAR CONTROL SYSTEMS
1
(s + a)2
1
( s − a)2
∠
+
+
ω
ω
s2 + ω 2
s
ω
s2 + ω 2
α
ω ω
( s + α )2 + ω 2
α ω (s + α)
( s + α )2 + ω 2
α
α
s2 − α 2
s
α
s2 − α 2
∴
LM t OP
1 1
L
L 2
N ∠2 Q
s s
L t OP
2
1
L
L M
s2
N ∠n Q
LM ∠n OP
n
1
∠n
L
Ns Q
n+1
s sn + 1
L n+1
df (t)
L
dt
d 2 f (t)
L ′
dt 2
d 3 f (t)
L ′ ″
dt 3
INTRODUCTION 7
z
df (t) d 2 f (t)
′ ″
LM OP
dt dt 2
zz
MN PQ
−
LM OP
+
f (t)
zzz
L +
MN PQ
−
−
LM OP
+ +
z zz zzz
L f (t) + +
MN PQ
−
− −
+ + +
L f (t) + + +
f (t), f (t), f (t)
L
d
L − F (s)
ds
Example 1.6.1.
1 s+6
2
s (s + 1) s (s + 4s + 3)
1 s+ 2
s 2 + 4s + 8 s 2 + 4s + 6
5 + +
+ + +
2
s (s + 4s + 5)
1
Solution.
s (s + 1)
8 LINEAR CONTROL SYSTEMS
k1 k
+ 2
s s+1
1 1
LM 1 − 1 OP
∴ −
s s+1
N s s + 1Q
L L
1 1
L L − L−1
s s+1
∴
Ans.
s+6
s (s 2 + 4 s + 3)
s+6
∴
s (s + 1) (s + 3)
+ +
+
+
1
2
1 1 1 1
∴ 2⋅ − 2.5 ⋅
LM2 ⋅ 1 − 2.5 ⋅ 1 + 1 ⋅ 1 OP
+ ⋅
s s+1 2 s+3
N s s + 1 2 s + 3Q
L L
1 1 1 1
L L−1 2 ⋅ − L−1 2.5 ⋅ + L−1 ⋅
s s+1 2 s+3
∴ Ans.
1
2
s + 4s + 8
∴
+ +
L L
+ +
L L ⋅
+ +
INTRODUCTION 9
1 −2 t
∴ e sin (2t) Ans.
2
s+2
s2 + 4s + 6
[(s + 2) 2 + ( 2 ) 2 ]
s+2
∴
(s + 2) 2 + ( 2 ) 2
s+2
L L
(s + 2) 2 + ( 2 ) 2
∴ 2 Ans.
5
s (s 2 + 4 s + 5)
5 k1 k s + k3
+ 22
2
s (s + 4 s + 5) s s + 4s + 5
1 s+4
∴ − 2
s s + 4s + 5
1 s+4
∴ −
s [(s + 2) 2 + (1) 2 ]
1 s+2 1
− −2
LM 1 − s + 2 OP
2 2
s [(s + 2) + (1) ] [(s + 2) 2 + (1) 2 ]
N s (s + 2) + (1) Q
1
L L −2
2 2 2 2
(s + 2) + (1)
1 s+2 1
L L −1 − L −1 − L −1 2
2 2
s [(s + 2) + (1) ] [(s + 2) + (1) 2 ]
2
∴
Ans.
s2 + 2s + 3
s + 6 s 2 + 12s + 8
3
s 2 + 2s + 3
∴
(s + 2) 3
k1 k2 k3
+ +
2
(s + 2) (s + 2) (s + 2) 3
10 LINEAR CONTROL SYSTEMS
1 2 3
∴ − +
LM 1 − 2 OP
2
s + 2 (s + 2) (s + 2) 3
N s + 2 (s + 2) Q
3
L L −1 +
2
(s + 2) 3
1 2 3
L L −1 − L −1 + L −1
2
s+2 (s + 2) (s + 2) 3
3
∴ 2
Example 1.6.2.
dx
2 + 8x
LM OP LM OP
dt
N Q N Q
Solution.
dx dx
L 2 + 8x L
+ L [8 x] L
L 2
dt dt
10
∴
s
10
s
2s + 5
s (s + 4)
k1 k
+ 2
s s+4
1.25 0.75
∴ +
s s+4
Ans.
Example 1.6.3.
d2x dx
+2 + 2x
′
LM OP + L L
2
dt dt
OP + L
MN PQ MN
Solution.
Q
L
′
INTRODUCTION 11
1
2
(s + 2s + 2)
1
∴
LM 1 OP
[(s + 1) 2 + (1) 2 ]
N (s + 1) + (1) Q
L L −1 2 2
∴
Ans.
Example 1.6.4.
LM d y OP + L L5 dy O + L [6 y]
Solution.
N dt Q MN dt PQ
2
L L
2
∴
′
12
(s − 1)
′
12
(s − 1)
12
(s − 1)
6s + 6
(s − 1)(s 2 + 5s + 6)
6s + 6
∴
(s − 1)(s + 2)(s + 3)
k1 k k
+ 2 + 3
s−1 s+2 s+3
1 2 3
LM 1 + 2 − 3 OP
∴ + −
s−1 s+2 s+3
N s − 1 s + 2 s + 3Q
L L−1
12 LINEAR CONTROL SYSTEMS
1 2 3
L L−1 + L−1 − L−1
s−1 s+2 s+3
∴ Ans.
Example 1.6.5.
⋅
+
Ω
µ
100
Solution.
∴
s
1 × 10 −6 s
100
⋅
6 −6
[(2 × 10 × 1 × 10 s) + 1] s
⋅
= µ Ans.
→∞
×
×
+
×
−
Example 1.6.6.
Solution.
di(t)
dt
3
∴
E E
∴ ∴
s s
1 E
⋅
R + Ls s
100
s 2 + 2s + 50
→ ∞
Solution.
1 1 100
∴ ∴ ⋅ 2
s s (s + 2 s + 50)
1 100 100
lim s =
Ans.
2
s→0 s (s + 2 s + 50) 50
Example 1.6.8.
+ +
Solution.
1 1 s
∴ ∴ ⋅
s2 s 2 (s 2 + 6 s + 25)
1 s
lim lim s
s→0 s→0 s 2 (s 2 + 6 s + 25)
1
Ans.
→ + + 25
Example 1.6.9. ′ ″
4s + 1
s(s 2 + 2)
Solution.
∴ Ans.
14 LINEAR CONTROL SYSTEMS
+
+
+
=
=
+
→∞ →∞ →∞
+ +
∴
′ Ans.
FG 1 − 8 IJ
″ lim ″ lim L ″
H sK
t→0 s→∞
FG s + 2 IJ FG 1 + 2 IJ
H sK H sK
(s − 8) (s − 8)
lim s . = lim lim
2
s→∞ (s + 2) s→∞ s→∞
2
″ Ans.
4
s(s + 2)
=
→
→∞
4 4
lim s lim
s→0 [ s (s + 2)] s → 0 (s + 2)
4
s (s + 2)
k1 k
+ 2
s s+2
2 2
∴ −
s s+2
LM 2 − 2 OP
INTRODUCTION 15
N s s + 2Q
LM 2 − 2 OP
L L
N s s + 2Q
L L
∴
PROBLEMS
+
+ + +
+
+
+ +
+
+
1.2.
+ +
′
+ +
′
θ
+ θ +
θ θ′ θ
∫θ
+
1.3.
+ +
1.4.
+ α +
+ α + ω + +
1.5.
+ α +
α£
+ α + ω + +
2
Transfer Function
n this chapter, the transfer function derivation of a control system is explained. The concepts of poles-
Izeros of the transfer function are discussed. Having given a physical system the procedure for obtaining
CONTENTS
TRANSFER FUNCTION
For any control system there exists an input termed as excitation or cause (denoted as r)
which operates through a transfer operation termed as transfer function (denoted as g) and
as c). Thus the cause and effect relationship between the output and input is related to each
other through a transfer function. The relationship between the output and input is
The arrow in the diagram indicates the direction of control action thus the block
The transfer function is expressed as the ratio of output quantity to input quantity.
Therefore,
c
g = (2.1)
r
16
TRANSFER FUNCTION 17
As the cause and effect approach to control systems is general, it is not necessary that
the output be of the same category as that of input. For example in the case of an electric
motor the input is an electrical quantity and the output being a mechanical one.
With reference to control system wherein all mathematical functions are expressed by
their corresponding Laplace transforms, therefore, the transfer function is also expressed as
control system it is assumed that all the initial conditions concerning the differential
L c (t)
G(s) = ...(2.3)
L r (t)
“The transfer function of a control system is defined as the ratio of the Laplace
transform of the output variable to the Laplace transform of the input variable assuming all
The transfer function of a linear control system can be expressed in the form of a quotient of
A( s) a0 s n + a1 s n – 1 + a2 s n – 2 + ... + an
G(s) = = ...(2.4)
B( s) b0 s m + b1s m – 1 + b2 s m – 2 + ... + bm
The numerator and the denominator can be factored into n and m terms respectively,
with such a factorisation the above expression for the transfer function can be expressed as,
A( s) K ( s – s1 ) (s – s2 ) ... (s – sn )
G(s) = = ...(2.5)
B( s) ( s – sa ) (s – sb ) ... ( s – sm )
a0
where K = is known as the gain factor of the transfer function.
b0
In the transfer function expression (2.5), if the numerator is equated to zero, then n
roots of the equation are s , s , s , ..., s and where as equating the denominator to zero, the
1 2 3 n
Poles of the transfer function : In the transfer function expression (2.5), if s is put
equal to s , s , ..., s it is noted that the value of the transfer function is infinite, hence s , s ,
a b m a b
Zeros of the transfer function : In the transfer function expression, if s is put equal
to s , s , ..., s it is noted that the value of the transfer function is zero, hence s , s , ..., s are
1 2 n 1 2 n
The poles s , s , ..., s or the zeros s , s , ..., s are either real or complex and the
a b m 1 2 n
It is possible that either poles or zeros may coincide. Such poles or zeros are called
multiple poles or multiple zeros, otherwise non-coinciding poles or zeros are called simple
The multiple poles/zeros occur due to the presence of repetitive factor in the
The graphical symbol for a pole is X and for a zero O. The said symbols are used when
poles and zeros are to be shown on a real and imaginary axes (s-plane).
The poles and zeros mentioned above are the finite ones. From equation (2.5) it is
observed that :
1. If n > m, then the value of the transfer function is found to be infinity for s = ∞.
Hence it is concluded that there exists a pole of the transfer function at infinity
Therefore, in addition to finite poles and zeros of a transfer function, if poles and zeros
located at zero and infinity are considered, then for a rational function the total number of
(s + 2) (s + 4)
G (s) =
s(s + 3) (s + 5) (s + 2 – j 4) (s + 2 + j 4)
For the above transfer function, the poles are at (1) s = 0, (2) s = – 3, (3) s = (– 2 + j4),
a b c
As the number of zeros should be equal to number of poles, the remaining three zeros
are located at s = ∞.
Τhe pole-zero locations are plotted s-plane as shown in Fig. 2.1.1.
+jw
x j4
–s –5 –4 –3 –2 +s
- poles
- zeros
x j4
–jw
Fig. 2.1.1. Poles-zeros in s-plane.
TRANSFER FUNCTION 19
In the transfer function expression of a control system the highest power of s in the
numerator A (s) is either equal or less than that of the denominator B (s). *The transfer
function of a system is completely specified in terms of its poles and zeros and the gain factor.
In a control system the output is related to the input by a transfer function as defined earlier,
i.e.
C (s)
= G(s)
R(s)
or C(s) = R(s) G(s) ...(2.6)
The output time response can be determined by taking inverse Laplace transform of
relation (2.6). If the input is specified as unit impulse at t = 0, then R(s) = 1 and the
C(s) = 1 . G(s)
L – 1
C(s) = L –1
G(s)
The inverse Laplace transform of G(s) is, therefore, called the impulse response of a
system, or the transfer function of a system is the Laplace transform of its impulse response.
The output impulse response of a system in time domain is thus solely determined by its
transfer function. The transfer function of a system depends on its elements, assuming
initial conditions as zero and is independent of the input function. It will be shown in the
later chapters that the transient performance of a control system is influenced only by the
CONTROL SYSTEM
The following steps give a procedure for determining the transfer function of a control
system :
2. Take the Laplace transform of the system equations, assuming initial conditions
as zero.
4. Take the ratio of the Laplace transform of the output and the Laplace transform of
the input.
*For the purpose of control system analysis, only finite poles/zeros and poles and zeros at the origin are
considered.
20 LINEAR CONTROL SYSTEMS
U
U
G(s) = .
U
U
Plot the poles and zeros in s-plane and determine the value of the transfer function at
s = 2.
1 (s 2 + 4) (1 + 2.5 s)
Solution. G(s) =
2 (s 2 + 2)(1 + 0.5s)
The poles are determined from the equation
(s
2
+ 2) (1 + 0.5s) = 0
+jw
1 +j2
s = ± j , s = – = – 2 + j Ö2
0.5
The zeros are determined from the equation.
–s –2 –4 +s
2
(s + 4) (1 + 2.5s) = 0 – j Ö2
Therefore, the zeros are, –j2
s = ± j = ± j 2,
–jw
1 Fig. 2.4.1. Pole zero configuration for
s = – = – 0.4
2.5 example 2.4.1.
2
1 (2 + 4) (1 + 2.5 × 2) 1 8×6
G(s) = . 2 = . = 2. Ans.
2 (2 + 2)(1 + 0.5 × 2) 2 6×2
(Fig. 2.4.2).
The value of the transfer function at s = 1 is found to be 3.2. Determine the transfer
+ jw
–s –4 –3 –2 O +s
– jw
function consists of one term in the numerator and three terms in the denominator.
K (s + 3)
G(s) =
s (s + 2) (s + 4)
It is given that at s = 1, the value of G (s) is 3.2
K (1 + 3) K ×4
∴ G(1) = 3.2 = =
1(1 + 2) (1 + 4) 1× 3 × 5
3.2 × 1 × 3 × 5
∴ K = = 12
4
12 (s + 3)
∴ G(s) = Ans.
s (s + 2) (s + 4)
Example 2.4.3. Find the transfer function of the network given below (Fig. 2.4.3).
ei eo
C
Input Output
Fig. 2.4.3. Network for example 2.4.3. Fig. 2.4.4. R.C. network.
1
e
i
= Ri +
1
C z i dt ...(1)
and e
o
=
C z i dt
Assuming initial conditions as zero and taking Laplace transform on both sides of
...(2)
1 I (s)
E (s) = RI(s) + . ...(3)
i
C s
1 I (s)
and E (s) = . ...(4)
o
C s
or E (s) =
LM
R+
1
I (s)
OP ...(5)
i
N Cs Q
1
and E (s) = I (s) ...(6)
o
Cs
From equations (5) and (6) the transfer function is obtained below,
+
U
'
U
Q %U '
U
Q
1
= or = Ans.
'
U
K
LM 4 +
OP +
U
'
U
K
RCs + 1
N %U Q
22 LINEAR CONTROL SYSTEMS
F θ F θ
T = ,
+ H ...(1)
FV FV
Assuming initial conditions as zero and taking the Laplace transform on both sides of
θ
U
= Ans.
6
U U
,U + H
Example 2.4.5. A torque T is applied to a load through a gear having a gear ratio
1
N /N (Fig. 2.4.5).
2 1
If the moment of inertia and coefficient of viscous friction of the load is J and f
2 2
respectively, determine the transfer function relating the applied torque T and the angular
1
shift θ 1
at the primary shaft side.
T1 q1 N1 Primary
Gear
Secondary J2 : F2
T2 q2
Gear
N2
Fig. 2.4.5. Gear system for example 2.4.5.
Solution. The effect of the applied torque T through the gear results in application of
1
torque T to the load and the torques T and T are related as follows :
2 1 2
6 0
=
6 0
F θ Fθ
T
2
= ,
+ H
...(1)
FV FV
6 0
∵
=
6 0
0
∴ T
2
= 6
...(2)
0
Substituting for T in equation (1) from equation (2) the following equation is
2
obtained :
0 F θ F θ
6
= ,
+ H
...(3)
0 FV FV
TRANSFER FUNCTION 23
Again, since N
1
θ 1
= N
2
θ 2
0
or θ 2
=
θ
...(4)
0
F θ 0 F θ
=
...(5)
FV 0 FV
F θ 0 F θ
=
...(6)
FV 0 FV
θ
F
F θ
Substituting for and in equation (3) from equations (5) and (6) the
FV FV
0 0 F θ 0 F θ
6
= ,
+ H
...(7)
0 0 FV 0 FV
T = ,
FG 0
IJ
F
θ
+ H
FG IJ0
F θ
...(8)
1
H 0 K FV
H K 0 FV
put J = ,
FG 0
IJ
...(9)
1eq
H 0 K
and f = H
FG 0
IJ
...(10)
1eq
H 0 K
The terms J1eq and f1eq are known as the moment of inertia and coefficient of viscous
friction of the secondary side referred to the primary side of the gear. Substituting for
,
FG IJ
0
and H
FG IJ
0
from expressions (9) and (10) in equation (8) the following equation
H K
0
H K0
relating T
1
and θ 1
is obtained.
F θ F θ
T
1
= ,
GS
+ HGS
...(11)
FV FV
Assuming initial conditions as zero and taking Laplace transform on both sides of
equation (11)
T
1
(s) = J
1eq
s
2
θ 1
(s) + f
1eq
s θ 1
(s) ...(12)
or T
1
(s) = s(J
1eq
s + f
1eq
) θ 1
(s) ...(13)
θ
U
= Ans.
6
U
U
,
GS
U + H
GS
24 LINEAR CONTROL SYSTEMS
PROBLEMS
U +
U +
G(s) =
U
U +
U + U +
Determine the poles and zeros and show the pole-zero configuration in s-plane.
-
U +
G(s) =
U
U +
U +
U + U +
Determine the poles and zeros of the transfer function. Plot the pole-zero configuration in s-
plane.
–t
g(t) = e (1 – cos 2t)
2.4. A force F when applied to a mass of M results in a displacement of X, obtain the transfer
2.5. A series circuit consisting of a resistance R and the inductance of L is connected to a d.c.
supply of E volts. Determine the transfer function relating the current flowing in the circuit
2.6. Determine the transfer function of the network shown in Fig. 2P-1 relating E (s) and E (s).
o i
2.7. Derive the transfer function of the system shown in Fig. 2P-2, K being the amplifier gain.
R1 R2
ei C1 K C2 eo
Block Diagrams
CONTENTS
R(s) + E(s) C(s)
G1(s) G2(s)
–
25
26 LINEAR CONTROL SYSTEMS
R(s)A C1(s)
G1(s)
(a) (b)
BLOCK DIAGRAMS 27
+ –
X(s) Y(s)
(a)
– +
Y(s) X(s)
(b)
G1(s)
+
R(s) + C(s)
G2(s)
+
R(s) C(s)
G3(s) G1(s) + G2(s) + G3(s)
(a) (b)
28 LINEAR CONTROL SYSTEMS
R(s) A B C(s)
G(s)
R(s) A B C(s)
G(s)
1 R(s)
R(s) G(s)
R(s) G(s)
(a) (b)
)
*
C(s) C(s)
G(s)
C(s)
G(s)
(a) (b)
*
)
X(s)
(a)
B
R(s) A R(s)G(s) + C(s) = [R(s) G(s) ± X(s) G(s)]
G(s)
±
X(s)G(s)
G(s)
X(s)
(b)
)
*
BLOCK DIAGRAMS 29
B
R(s) A R(s) G(s) + C(s) = [R(s) G(s) ± X(s)]
G(s)
±
X(s)
(a)
1 B 1
A R(s) ± X(s) C(s) = [R(s) + X(s) ] G(s)
R(s) + G(s) G(s)
G(s)
= [R(s) G(s) ± X(s)]
± 1
X(s)
G(s)
1/G(s)
X(s)
(b)
Fig. 3.1.10. Summing point at position B shifted to position A.
11. Shifting of a take off point from a position before a summing point to a position
after the summing point is shown in Fig. 3.1.11 (a) and (b).
X(s)
R(s)
(a)
±
+
X(s) R(s)
–
+
(b) X(s)
12. Shifting of a take off point from a position after a summing point to a position
before the summing point is shown in Fig. 3.1.12 (a) and (b).
(a)
30 LINEAR CONTROL SYSTEMS
X(s)
+ C(s) = [R(s) ± X(s)]
X(s)
(b)
*
)
(a1) (a2)
=
±
m
=
BLOCK DIAGRAMS 31
R(s) + E(s) C(s)
G(s)
± R(s) G(s) C(s)
1 m G(s)
(b1) (b2)
m
=
m
=
±
±
m
=
=
>
4 I
BLOCK DIAGRAMS 33
G3
+
R + + + C
G1 G2
– –
H1
G3
+
R + + + C
G1 G2
– –
G2H1
R + + C
G1 G2+ G3
– –
G2H1
R + + C
G1 G2+ G3
– –
G2H1
G2 + G3
R + G2 + G3 C
G1
1 + (G2 + G3)
–
G2H1
G2 + G3
+
+
+ +
+
+ + +
+
+ + +
G2
+
R + C
G1 G3
+
–
H1
G2
+
R + + C
G1 G3
–
G1H1
G2
+
R + + C
G1 G3
–
G1H1
R G1 + G2 + G3 C
–
G1H1
R + C
(G1 + G2)G3
–
G1 + G2
+
+ +
+
+
+
R G1G3 + G2G3 C
1 + G1G3H1
Ans. (ii)
H2
R + – C
G1 G2
–
G5 G3
+ +
H1 G6 G4
+ +
45 78
36 LINEAR CONTROL SYSTEMS
6 9
+ +
H2
R + –
G1 G2 C
–
G5/G2
+
H1 G6 G3+G4
+
=
>
R + G1G2 C
1+G1G2H2
–
G5H1+G2G6H1(G3+G4)
G2
A
A
+
+
+
+ ×
+
+
+
+
+
BLOCK DIAGRAMS 37
+
+ =
G3
+
R + G1 G2 C
– +
H1 H2
+
+
G3/G1
+ R + G1
G1G2+G3 C
R + C
G1 G2 – G1
+
–
H1 H2
H1 H2 +
+ +
+
R + G1G2+G3 C
G1
– G1 R + C
G1G2+G3
–
G1
H1
G1G2+G3 G1H1+G3H2+G1G2H2
+ G1G2+G3
H2
+
?
> @
?
+
+
+
+
38 LINEAR CONTROL SYSTEMS
+
+
+
+
+
×
+
+
+
+
+
H3
+ –
R + + C
G1 G2 G3
– –
H1 H2
H3/G3
+ – +
R G1 + G2 G3 C
– –
H1 H2
H3/G3
–
R + + G3 C
G1 G2
– 1+G3H2
H1
>
=
– + –
R + G2G3 C R + G1
G2G3 C
G1
+ 1+G3H2 – 1+G3H2
–
H1 H1
H3/G1G3
– G1 G2G3
R C
+ 1+G1H1 1+G3H2
+
+
+
+
×
+
+
+
+
+
+
+
R + C
G1 G2
+
+ –
G3 H1
R + + C
G1 G2
+ –
G3 H1
According to Fig. 3.2.6 (b) the overall transfer function is determined below
Fig. 3.2.6 (b) Block diagram reduction of system in F ig. 3.2.6 (a).
%
) + ) ) )) +) )
=
=
Ans.
4 +) *
+) *
Example 3.2.7. Determine the overall transfer function relating C and R for the
G3
+
R + + + C
G1 G2
– –
H2
H1
Solution. Shift the take off point after block G to a position before block G .
2 2
G3
+
R + + + C
G1 G2
– –
H2
G2H1
Shift the take off point before block G to a position after block G (Rule 12).
2 1
G3
+
R + + + C
G1 G2
– –
+
G2H1 H2
–
Fig. 3.2.7 (b) Block diagram reduction of system in F ig. 3.2.7 (a).
BLOCK DIAGRAMS 41
Shift summing point before block G to a position after block G .
2 2
G3
+
R + + + C
G1 G2
– –
G2
+
G2H1 H2
–
Fig. 3.2.7 (c) Block diagram reduction of system in Fig. 3.2.7 (b).
G3
+
R + + + C
G1 G2
– –
G2
+
G2H1 H2
–
Fig. 3.2.7 (d) Block diagram reduction of system in F ig. 3.2.7 (c).
Shift the take off point after block H to a position before block H .
2 2
G3
+
R + + + C
G1 G2
– –
G2H2
+
G2H1 H2
–
Fig. 3.2.7 (e) Block diagram reduction of system in Fig. 3.2.7 (d).
The block diagram shown in Fig. 3.2.7 (e) can be redrawn as shown in Fig. 3.2.7 (f).
G3
+
R + + + + C
G1 G2
+ – –
G2H1
G2H2
G2H1H2
R + + 1 C
G1 G2 + G3
1 + G2 H 2
+ –
G2H1
G2H1H2
Fig. 3.2.7 (g) Block diagram reduction of system in F ig. 3.2.7 (f).
R + 1 1 C
G2 + G3
1 + G1G2H1 1 + G2H2
+
G2H1H2
Fig. 3.2.7 (h) Block diagram reduction of system in F ig. 3.2.7 (g).
)
)
+)
% +)
) *
+)
*
=
4 )
Ō
)
+)
) * *
+)
) *
+)
*
)) +) )
= Ans.
+)
*
+) )
* −)
) ) * *
Example 3.2.8. Determine the overall transfer function for the block diagram
(Fig. 3.2.8).
R + + C
G1 G3 G4
– –
H1
+
G5 H2
–
G2
Solution. (i) Shift the summing placed before block G towards left of block G .
5 5
R + G4 C
G1 G3
1 + G4H1
–
+ +
G5 H2
–
G2 G5
R + x + G3G4
G1 C
1 + G4H1
– –
G2 – G5 G5H2
G – G (– G ) = G + G G
1 2 5 1 2 5
) )
(ii) Eliminating the summing point located before block :
+)
*
) )
% +) * ) )
=
=
Z
+
) )
×) *
+)
*
+)
) ) *
+)
*
The block diagram takes the form as shown in Fig. 3.2.8 (c).
R G3G4 C
G1 + G2G5
1 + G4H1 + G3G4G5H2
% ) )
∴ = (G
1
+ G G ) ×
2 5
4 +)
*
+)
) ) *
% )) ) +) ) ) )
or = Ans.
4 +)
*
+) ) ) *
Example 3.2.9. Determine the transfer function C/R for the system block diagram
G5
R + + + + C
G1 G2 G3 G4
– – – +
H1 H2
H3
Solution. (i) Eliminate the summing point placed before block G , the equivalence is
1
G /(1 + G H ).
1 1 1
(ii) Shift the take-off point located before block G towards right side of block G .
3 3
G5/G3
G1 +
R + G2
+
G3 G4 C
1 + G1H1 +
– –
H2
H3
) ) ) +)
(i) The blocks G /G
5 3
and G
4
are in parallel, the equivalence is +) =
.
) )
(ii) Eliminate the summing point placed before block G , the equivalence is G /(1 +
3 3
G H ).
3 2
R + G1 G3 G5 + G3G4 C
G2
– 1 + G1H1 1 + G3H2 G3
H3
)
)
)
+)
)
The blocks ) and
+)
*
+)
*
)
) )
) +) ) ) ) )
+) ) ) )
×) ×
×
=
+)
*
+)
*
)
+)
*
+)
*
+)
) * *
R + G1G2G5 + G1G2G3G4 C
– 1 + G1H1 + G3H2 + G1G3H1H2
H3
) ) )
+) ) ) )
% +)
+) * +) ) * *
*
=
4 ) ) ) + ) ) ) )
+ ×
*
+ ) * + ) * +) ) * *
BLOCK DIAGRAMS 45
Simplifying,
% )) ) +) ) ) )
= Ans.
4 +)
*
+)
*
+) ) )
*
+) ) ) )
*
+) ) ) * *
Example 3.2.10. Obtain the overall transfer function for a control system having the
G3
–
R + + + C
G1 G2 G4
– – +
+
H2 H1
+
H3
R + + + – C
G1 G2 G3 + G4
– – +
H2 H1
+
H3
(i) Shift the take-off point placed before block (G + G ) towards right side of block
3 4
(G + G ).
3 4
(ii) Eliminate the summing point located before block G , the equivalence is G /(1 + G ).
2 2 2
R + + G2 C
G1 G3 + G4
1 + G2
– –
1
G3 + G 4
+
G
H21 H1
+
H3
) * +) * +
*
+ =
)
+)
) +)
) )
) +)
×
) + ) =
+)
+)
R + + G2(G3+G4) C
G1
1+G2
– –
G3H1+G4H1+1
H2
G3+G4
H3
) *
+) * +
) * *
+) *
*
+ *
H × = , thus [Fig. 3.2.10 (d)]
2
)
+)
) +)
+ x+ G2(G3+G4)
R C
G1
– 1+G2
–
G3H1H2+G4H1H2+H2
G3+G4
H3
)
)
+)
%
+)
∴ =
Z )
) +) ) * * +) * * + *
+
×
+)
) +)
% ) )
+)
)
Simplifying, = , the block diagram takes
Z +) +)
*
+)
) * *
+)
) * *
R + G1(G2G3+G2G4) C
– 1+G2+G2H2+G2G3H1H2+G2G4H1H2
H3
)
) )
+)
)
% +) +)
*
+)
) * *
+)
) * *
=
4 )
) ) +) )
+
× *
+) +)
*
+)
) * *
+)
) * *
% ) ) )
+)
) )
Simplifying, = Ans.
4 +) +)
) * *
+)
) * *
+)
) ) *
+)
) ) *
Example 3.2.11. Reduce the system shown below to a single block representation.
G2
R + + + +
G1 G3 G4 C
– – –
H1
H2
H3
Solution. Split the summing point before block G [Fig. 3.2.11 (a)]
4
(ii) Shift the summing point placed after block G towards left side of block G .
3 3
G2
R + + + +
G1 G3 G4 C
+
– – –
H1
H2
H3
G2/G3
R + + + + G4 C
G1 G3
1+G4H1
– –
H2
H3
)
(ii) Combine block G and as they are in cascade, the equivalence is
3
+)
*
G G /(1 + G H ).
3 4 4 1
G2/G3
+ +
R + G3G4 C
G1
– + x – 1+G4H1
H2
H3
G2/G3
+
R + G3G4 C
G1
– + 1+G4H1+G3G4H2
H3
) )
% +) * ) )
∴ =
=
Z ) ) +) * +) ) *
+
× *
+)
*
Shift summing point placed before block G towards right side of block G [Fig.
1 1
3.2.11 (e)]
G2/G3
+ y G3G4
R G1 + C
+ 1+G4H1+G3G4H2
–
G1H3
+ + G3G4
R G1 C
– + 1+G4H1+G3G4H2
G1H3
) )
% +)
*
+)
) *
∴ =
[ ) )
+
×) *
+)
*
+)
) *
% ) )
or =
[ +)
*
+)
) *
+)
) ) *
) ) ) +)
The blocks G
1
and G /G
2 3
are in parallel, the equivalence is G
1
+
=
) )
G2/G3
+ G3G4
R G1 C
+ 1+G4H1+G3G4H2+G1G3G4H3
R G1G3+G2 G3G4 C
G3 1+G4H1+G3G4H2+G1G3G4H3
% ) ) +) ) )
and =
×
4 )
+)
*
+)
) *
+)
) ) *
% ) ) )
+) )
or = Ans.
4 +)
*
+)
)
* +)
) ) *
Example 3.2.12. Obtain the overall transfer function for the block diagram (Fig.
3.2.12).
G2
R + + + +
G1 G3 G4 C
– – – +
H1 H2
H3
Solution. (i) Shift the take-off point placed before G towards right side of block G .
1 1
(ii) Shift the summing point located after block G towards left side.
3
G2/G1
+ + + +
R G1 G3 G4 C
– – – +
H1 H2
H3/G3
G2/G1
G1 + + +
R G3 G4 C
1+G1H1 – +
–
H2
H3/G3
Fig. 3.2.12 (b) Block diagram reduction for F ig. 3.2.12 (a).
Eliminate the summing point located before block G , the equivalence is G /(1 +
3 3
G2/G1
G1 + G3 +
R G4 C
1+G1H1 1+G3H2 +
–
H3/G3
Fig. 3.2.12 (c) Block diagram reduction for Fig. 3.2.12 (b).
G2/G1
R G1 + G3G4 +
C
1+G1H1 1+G3H2 +
–
H3/G3
Fig. 3.2.12 (d) Block diagram reduction for Fig. 3.2.12 (c).
Shift the summing point located after block G G /(1 + G H ) towards left side to get
3 4 3 2
G2 1+G3H2
´
G1 G3G4
G1 + + G3G4
R C
1+G1H1 + 1+G3H2
–
H3/G3
G2 1+G3H2
´
G1 G3G4
G1 + + G3G4
R C
1+G1H1 + × 1+G3H2
–
H3/G3
+
+
+
=
+
=
+ +
+ ×
+
R G1 G1G3G4+G2(1+G3H2) G3G4 C
1+G1H1 G1G3G4 1+G3H2+G4H3
+ +
× ×
+
+
+
+ +
Ans.
+
+
+
+ +
Example 3.2.13.
G2
+ + + +
R G1 G3 G4 C
– – –
H1 H2
Solution.
G2
+ + + +
R G1 G3 G4 C
– – –
H1H2
H2
H1H2
G2
+ + + + +
R G1 G3 G4 C
– – – –
H1H2
H2
H1H2
R + + G4 C
G1+G2 G3
– 1+G4H1H2
–
H2
H1H2
+
BLOCK DIAGRAMS 53
R + x + G3G4 C
G1+G2
– 1+G4H1H2
–
H2
H1H2
+
∴
+ ×
+
+ +
R + G3G4 C
G1+G2
1+G4H1H2+G3G4H2
–
H1H2
+
+
+
+
+
+
∴
+
+ ×
+
+
+
Ans.
+
+
+ +
Example 3.2.14.
G2 D
+ +
R + + + C
G1 G3 G4
– –
H2
H1
Solution.
G2 D
+ +
R + + + C
G1 G3 G4
– –
H2
H1
D
+
R + G3 + C
G1 + G2 G4
1 + G3H2
–
H1
+
D
+
R + (G1 + G2)G3 + C
G4
1 + G3H2
–
H1
R + (G1 + G2)G3 Cr
G4
1 + G3H2
–
H1
+
+
+
+
+
+
Ans.
+
+
+
D
+
G3 + Cd
G1 + G2 G4
1 + G3H2
–H1
+ Cd
G4
–
G3
G1 + G2 H1
1 + G3H2
+
= Ans.
+ + + +
+
+
+
+
+
+
+
+ Ans.
+
+
+
56 LINEAR CONTROL SYSTEMS
Example 3.2.15.
D1 H2
+ –
R + + + C
G1 G2
– –
H3
+
H1
+
D2
Fig.3.2.15.
Solution.
D1 D1
+ - G +
R + + 2 C R + + + G2 C C
G1 G1
– 1+G2H2 1+G2H2
– – –
H3 H3
+ +
H1 H1
+ +
D2 D2
Fig. 3.2.15 =
Fig. 3.2.15 >
=
+
D1
+ G2 Cr
R + + G2 C C R +
G1 G1
1+G2H2+G2H3 – 1+G2H2+G2H3
–
H1
H1
>
?
+ +
∴
+ ×
+
+
BLOCK DIAGRAMS 57
+ + +
+G G2 C d2
D1 + G2 C d1 1 1+G2H2+G2H3
1+G2H2+G2H3
+ +
–H1
+
G1 H1
D2
@
A
+ +
− −
+
+
+
+ +
D2 + G2 C d2 D2 + –G1G2H1 C d2
–G1H1
+ 1+G2H2+G2H3 + 1+G2H2+G2H3
B
C
−
+
+
−
+ +
−
− ×
+ +
− + + +
58 LINEAR CONTROL SYSTEMS
+
+
+
+
+
+
+
−
+ Ans.
+
+ +
PROBLEMS
3.1.
G3
+
R + + C
G1 G2
–
H1
3.2.
R + + + C
G1 G2 G3 G4
– – –
H2
H1
3.3.
G3
+
Input + + + Output
G1 G2
– –
H2
H1
3.4.
R1 + C1
G1
–
H1
H2
–
R2 + C2
G2
3.5.
G5
G3
+ +
R + + + + C
G1 G2 G4
– –
H1
H2
3.6.
G3
+
H1
–
+
R + + C
G1 G2
–
H2
3.7.
G4
+
R + + + C
G1 G2 G3
– –
H1
3.8.
G2
D
+
R + + + + + + C
G1 G3
– – –
H1
H2
H3
3.9.
G2 R2
+
R1 + + + C
G1 G3
–
H1
3.10.
R2
+
R1 + + C
G1 G3
–
+
H1 H2
+
Fig. 3P-10.
BLOCK DIAGRAMS 61
3.11.
D
G2
+ +
R + G1 + G3 + G4 C
- -
H1
H2
Fig. 3P-11
3.12.
G2
+
R + G1 + + G3 + C
- – –
H1
H2
Fig. 3P-12
3.13.
G3
R + + + +
G1 G2 C
– –
H2
H1
Fig. 3P-13
"
CONTENTS
N =N
62
SIGNAL FLOW GRAPHS 63
x0 + x1
a 1
+
N =N >N
b
x0 a
x1
Input Output
node node
N =N >N
Solution.
Ans.
−
−
x1 a x2 b x3 d x4 e x5 g x6
c f
− +
+
Ans.
− + +
SIGNAL FLOW GRAPHS 65
x1 a x2 b x3 d x4 f x5 g x6
c e
Ans.
− +
− +
x1 a x2 x3 d x4
Ans.
d
a
e f h
x1 x2
g
−
−
SIGNAL FLOW GRAPHS 67
+
− −
− + −
− + +
−
+
−
Ans.
− +
+
∆
∑
=
∆
∆
∆
£¤¢
68 LINEAR CONTROL SYSTEMS
G2
R
+
+
G1 C
+
–
H1
G2
1 2
¥
G1
R C
–H1
=
∆
∆ +∆ + +
= Ans.
∆ +
+ +
+
G2
+
R + + C
G1
–
H1
SIGNAL FLOW GRAPHS 69
Solution.
=
∆
∆ +∆ +
+
= = Ans.
∆ +
+
R + C
G1 G3
Input Output
+
G2 H1
+ –
70 LINEAR CONTROL SYSTEMS
and P = G G
2 2 3
L = – G H
1 3 1
forward paths, therefore, path factors Fig. 4.4.1 (a) Signal flow graph of Fig. 4.4.1.
are
∆ 1
= 1 and ∆ 2
= 1
∆ = 1 – (L 1
) = 1 – (– G H ) = 1 + G H .
3 1 3 1
Applying Mason’s gain formula the closed-loop transfer function is determined below
% 2 ∆ +2∆ ) ) +) ) ) ) +) )
=
=
=
Ans.
4 ∆ +)
*
+)
*
Example 4.4.2. Represent the following set of equations by a signal flow graph and
x = ax + fx ; x = bx + ex
2 1 2 3 2 4
x = cx + hx ; x = dx + gx .
4 3 5 5 4 2
transmittances are joined by lines to draw signal flow graph as shown in Fig. 4.4.2.
P = abcd and P = ag
1 2
L = f, L = ce and L = dh
1 2 3
The loop L = ce is not touching the forward path P , therefore, the path factor
2 2
associated with P
2
is ∆ 2
= 1 – ce.
= 1 – f – ce – dh + fce + fdh
SIGNAL FLOW GRAPHS 71
∆ +∆ +
−
=
∆ −
− − +
+
+
−
Ans.
−
− − +
+
c
–
R + + + C
a b d
+ +
i
j
+ +
+
e f h
–
g
Solution.
=
Ans.
∆
Ans.
72 LINEAR CONTROL SYSTEMS
∆ Ans.
∆
Ans.
G3
+
R + + + C
G1 G2
– –
H1
Solution.
G3
–1
R G1 G2 1 C
–H1
∆
∆
∆ +∆ +
=
∆ − − − −
SIGNAL FLOW GRAPHS 73
+
Ans.
+
+ +
G3
+
R + + + C
G1 G2
– –
H2
H1
Solution.
G3
R G1 1 G2 1
C
–H1 –H2
∆ Ans.
∆
∆
¦
+
Ans.
+ + −
74 LINEAR CONTROL SYSTEMS
Example 4.4.6.
D
G2
+ +
R + + + C
G1 G3 G4
– –
H1
H2
Solution.
G2 D
R G3 1 G4 C
G1
–H1
–H2
G2
R 1 G3 1 G4 C
G1
–H1
–H2
∆
∆
∆ +∆ +
= Ans.
∆ +
+
+
G2
–
R + + G1 + G3 G4 + C
– – –
H1
Solution.
– G2
R G1 1 G3 G4 C
–1
–1 –1
H1
=
∆
∆
∆ +∆ + −
= Ans.
∆ + + +
76 LINEAR CONTROL SYSTEMS
(Fig. 4.4.8).
G3
H1
– +
R + + C
G1 G2 G4
+
– +
G5
H2
Draw the signal flow graph to determine the overall transfer function.
Solution. The signal flow graph for the given block diagram is shown in Fig. 4.4.8 (a).
L = G (– H ), L = G G . 1 . (– H ) and L = G G (– H ).
1 4 1 2 2 4 2 3 2 5 2
∆ = 1 – (L 1
+ L
2
+ L )
3
= 1 – {G (– H ) + G G 1(– H ) + G G (– H )}
4 1 2 4 2 2 5 2
= 1 + G H + G G H + G G H
4 1 2 4 2 2 5 2
G3
P = G G G . 1
1 1 2 4
– H1
All loops are touching the path P , therefore,
1
R G1 G2 G4 1 C
Path factor : ∆ = 1
1
G5
P = G G G . 1
2 1 2 5
P , therefore,
3
∆ 3
= (1 + L ) = {1 + (– G H )} = 1 + G H
1 4 1 4 1
% 2
∆ +
2
∆ +
2
∆
=
4 ∆
) ) )
+) )
)
+ )
+ )
*
=
+)
* +)
) * +) ) *
% ) ) )
+) )
)
+ )
+ ) *
or = . Ans.
4 +)
* +)
) * +) ) *
SIGNAL FLOW GRAPHS 77
G5
Solution.
G5
+
R + + + C
G1 G2 G3 G4
– –
H1 H2
G5/G3
R G2 + +
G1 + G3 G4 C
1 + G2H1 –
H2
G5/G3
R G2 + C
G1 + G3 G4
1 + G2H1 +
–
H2
+ +
+
+
+
¦
% ) ) + )
+ ) * )
= )
×
×
×)
4 )
+) *
+)
*
% ) ) ) )
+ ) ) )
+ ) *
or = . Ans.
4 +)
*
+) * +)) * *
Example 4.4.10. Draw the signal flow graph for a control system whose block
G5
+
R G1 + G2 + G3 + G4 C
– –
H1
H2
Solution. The signal flow graph is drawn as shown in Fig. 4.4.10 (a).
L = G (– H ) and L = G (– H )
1 2 1 2 3 2
∆ = 1 – (L + L )
1 2
G5
= 1 – {G (– H ) + G (– H )}
2 1 3 2
= 1 + G H
2 1
+ G H
3 2 R G1 1 G2 G3 G4 C
The forward paths are :
P
1
= G . 1 . G G G
2 3 4
– H1 – H2
All loops are touching the forward path P , therefore,
1 Fig. 4.4.10 (a) Signal flow graph for
P = G G G
2 1 5 4
= {1 – G (– H )} = 1 + G H
2 1 2 1
% 2
∆ +2∆
=
4 ∆
) ) ) )
+ ) ) )
+ )
*
=
+)
* +) *
% ) ) ) )
+) ) )
+)
*
or = . Ans.
4 +) * +)
*
PROBLEM
4.1. Draw signal flow graphs for problems 3.1 to 3.13 (Chapter 3) and verify answers using Mason’s
gain formula.
5
control system has to be mathematically modeled as per the physical laws governing its elements. In
A this chapter, transformed domain mathematical models have been developed. The approach being
the model determination through transformed equation and development of corresponding block
diagram. The rules of block diagram reduction are applied to obtain the model of the control system. The
solved examples are given for the application of concepts explained in the text.
CONTENTS
elements arranged in a planned manner. Now, hereunder the procedure for modelling a
control system based on transfer function approach is described. Transfer functions of some
elements commonly used in control systems have been derived and examples have been
given to use the derived results for obtaining the mathematical model of a control system.
79
80 LINEAR CONTROL SYSTEMS
V (s) = R I(s) + R I(s) + I(s)
i 1 2
%U
or V (s) =
LM
4
+4
%U + OP +
U ...(5.1)
i
N %U Q
and V (s) = R I(s) + +
U
o 2
%U
4 %U +
or V (s) = +
U ...(5.2)
o
%U
8
U
Q
4 %U
+
=
8
U
K =
4 +4
%U + ?
8
U
Q
+ U6
or = ...(5.3)
8
U
K
+ U6
LM 4
OP
%U
V (s) =
i MM
PP +
U + 4 +
U
4 +
N
%U Q
=
4
+4
+ 4 4 %U?
or V (s) = +
U ...(5.4)
i
4 %U +
and V (s) = R I(s) ...(5.5)
o 2
MODELLING A CONTROL SYSTEM 81
Therefore, the transfer function is given by
8
U
Q
4
4 %U
+
=
8
U
K
4 +4
FG 4 4
%U +
IJ
H 4 +4
K
8
U
Q
α
+ U6
or = ...(5.6)
8
U
K
+ U6
4
where α=
4 +4
T = R C
1 1
4 4
and T = %
2
4 +4
in Fig. 5.2.1.
of K is, Newton/meter). The restoring force f for the Fig. 5.2.1. Translational spring-mass
k
damper. The damping force varies in direct proportion to the velocity. The coefficient of
Application of force f (t) to the mass results in a displacement x (meters). The equation
of motion for the system is obtained by applying Newton’s second law of translational
motion.
F Z FZ
or /
= − H – Kx + f (t) ...(5.7)
FV FV
↓ ↓ ↓
Damping Spring Applied
force
82 LINEAR CONTROL SYSTEMS
F Z FZ
/
+ H + -Z = f (t) ...(5.8)
FV FV
2
Ms X(s) + fs X(s) + KX(s) = F(s) ...(5.9)
If X(s) is specified as the output and F(s) as input then the transfer function of the
:
U
= ...(5.10)
(
U /U + HU +-
Force-voltage analogy
as follows :
.
FV
FK
+ 4K +
%
z
As the current is the rate of flow of electric charge.
K FV = e(t) ...(5.11)
Fig. 5.2.2. Electrical circuit for force
voltage analogy.
FS
∴ i = ...(5.12)
FV
F S FS
.
+ 4 + S = e(t) ...(5.13)
FV FV %
It is observed that equation (5.8) is analogous to Eq. (5.13) and following analogies are
established :
F S FS
As the quantities . 4 S G
V are
FV FV %
F Z FZ
voltages and / H -Z H
V forces, therefore,
FV FV
aforesaid analogy is called force-voltage analogy. Fig. 5.2.3. Electric circuit for force current
analogy.
Force-current analogy
The nodal equation as obtained for the circuit shown in Fig. 5.2.3 is analogous to
spring-mass damper system described by the equation (5.8). The said analogy is established
below.
MODELLING A CONTROL SYSTEM 83
According to nodal analysis
i + i + i = i(t) ...(5.14)
L C R
The common voltage across the three parallel elements of the circuit (Fig. 5.2.3) is
i
L
=
z GFV ...(5.15)
i = G ...(5.16)
R
4
FG
i = % ...(5.17)
C
FV
Substituting relations (5.15), (5.16) and (5.17) in equation (5.14). The equation given
below is obtained :
.
z G FV +
4
G +%
FV
= i(t) ...(5.18)
F ψ
e = ...(5.19)
FV
F ψ F ψ
ψ+ +%
= i(t) ...(5.20)
. 4 FV FV
F ψ F ψ
%
+ + ψ = i(t) ...(5.21)
FV 4 FV .
F ψ F ψ F Z FZ
As the quantities %
ψ K
V are currents and / H -Z H
V
FV 4 FV . FV FV
On the basis of force current analogy the equation for translational mechanical
mechanical circuit diagram simplifies the procedure for writing system equations.
shown in Fig. 5.2.4. It is required to set up equation for the system on application of force f (t)
are considered as nodes. These lower ends and the mass have
diagram for the system the lower end of the mass is also
The construction of the mechanical circuit is done Fig. 5.2.4. Mechanical spring-
From the mechanical circuit diagram of Fig. 5.2.5 it is noted that the forces developed
F Z
Inertia force = /
FV
FZ
Viscous friction force = H
FV
According to nodal analysis the sum of the forces mentioned above is equal to applied
force f (t). Thus the equation for the system relating displacement x and the applied force f (t)
is given below :
F Z FZ
/
+ H + -Z = f (t) ...(5.22)
FV FV
2
Ms X(s) + fsX(s) + KX(s) = F(s) ...(5.23)
If X(s) is specified as the output and F(s) as input, then the transfer function of the
:
U
= ...(5.24)
(
U /U + HU + -
2
The moment of inertia of the system is J (unit of J is kg-m ). The inertia torque
developed in the system is J α, where α is the angular acceleration produced due to applied
MODELLING A CONTROL SYSTEM 85
torque T (unit of T is Nm). The viscous damping is offered by the damper. The damping
torque varies in direct proportion to the angular velocity of the shaft. The coefficient of
The equation for the system is obtained by applying Newton’s second law of rotational
motion, therefore,
or J α = – fω + T ...(5.25)
F θ
and α=
...(5.27)
FV
Substituting relations (5.26) and (5.27), in equation (5.25) the system equation given
below is obtained
F θ F θ
,
= − H +6
FV FV
F θ F θ
,
+ H = T ...(5.28)
FV FV
Assuming initial conditions as zero, taking Laplace transform on both sides of the
Js
2
θ(s) + fs θ(s) = T (s)
If θ(s) is specified as the output and T(s) as input the transfer function relating θ(s)
and T(s) is given below :
θ
U
= ...(5.29)
6
U ,U + HU
In order to determine the system transfer function relating shaft speed and torque
U θ
U U
= ...(5.30)
6
U ,U + HU
follows:
F θ
Inertia torque = ,
FV
F θ
Viscous friction torque = H
FV
According to nodal analysis the sum of two torques mentioned above is equal to
applied torque T. Thus the equation for the system relating angular shaft shift θ and applied
torque T is given below
F θ F θ
T = ,
+ H ...(5.32)
FV FV
Hydraulic actuator. Hydraulic actuator shown in Fig. 5.3.1 controls large displacement
power at the shaft of main piston. The input power requirement is very small and provided
by the displacement of the valve piston. The linear motion of the valve piston controls the
When the input force through the displacement x is applied to the valve piston, the
valve piston moves to the right. Piston valve ports (2) and (3) are thus uncovered and oil at
higher pressure from port (2) enters to the left side of the main piston through main port A.
At the same time the main port B is connected to the sump through port (3). The pressure P
1
MODELLING A CONTROL SYSTEM 87
on the left side of the main piston is higher than the
From Fig. 5.3.2, it is observed that the oil flow q is a function of valve displacement x
q = f (x, P ) ...(5.33)
L
obtained
∂S ∂S
dq = FZ + F2 ...(5.34)
.
∂Z ∂2 .
Measuring x, P and q from initial zero values and assuming partial derivatives as
L
constant, then on integration of equation (5.34), the equation given below is obtained
q =
FG ∂ IJ + FG ∂ IJ
S
Z
S
2 ...(5.35)
H∂ K H∂ K
Z 2
.
.
In equation (5.35)
put K =
FG ∂ IJ
S
and K =
FG − ∂ IJ S
1
H∂ K
Z
2
H ∂ K 2
.
Therefore, q = K x – K P ...(5.36)
1 2 L
neglecting leakage then the rate of flow of oil into main cylinder is given by
F[
q = # ...(5.37)
FV
where A is the area of the piston. Substituting equation (5.37) in equation (5.36) following
equation is obtained :
F[
# = K x – K P ...(5.38)
1 1 2 L
FV
P =
FG - Z − #
F[ IJ ...(5.39)
L
- H
FV K
The force developed by the piston is AP , therefore, multiplying equation (5.39) by A
L
F =
# FG - Z − #
F[ IJ ...(5.40)
- H
FV K
88 LINEAR CONTROL SYSTEMS
The force F is applied to load having a mass M and coefficient of viscous friction at the
load is f, therefore,
F [ F[
F = /
+ H ...(5.41)
FV FV
/
F
[
+ H
F[
=
# FG - Z − #
F[ IJ ...(5.42)
FV
FV - H
FV K
Assuming initial conditions as zero, taking Laplace transform on both sides of
2
#
Ms Y(s) + fs Y(s) = [K X(s) – As Y(s)] ...(5.43)
1
-
From equation (5.43), the transfer function of hydraulic actuator relating Y(s) and X(s)
is obtained below
;
U #- -
= ...(5.44)
:
U U= /U +
H + # ?
;
U -
or = ...(5.45)
:
U U
+ U6
#- /
where K = and T =
-
H + #
H + #
F [ F[
/
+ H + -[ = AP
i
...(5.49)
FV FV
From equation (5.50) the transfer function relating the output Y(s) and input P (s) is
i
;
U #
= ...(5.51)
2
U
K /U + HU +-
Heat transfer system. A thermal system used for heating flow of water is shown in
Fig. 5.5.1. An electric heating element is provided in the storage tank to heat the flow of
tank is insulated from surrounding atmosphere to reduce heat loss. The thermal resistance
The energy input to the system is q (joule/sec) and obtained from the electric heater.
As the specific heat of the thermal insulation is very low there is negligible heat storage in
the insulation. However, there is a heat loss from the water to surrounding atmosphere.
It is assumed that the water in the tank is thoroughly mixed during the heating
If the thermal capacity is C (joule/°C) then the rate of heat flow for the water in the
tank is given by
F θ
q = % ...(5.52)
i
FV
The rate of heat flow from the water to the surrounding atmosphere is
θ −θQ
q = ...(5.53)
t
4
q = q + q ...(5.54)
i t
Substituting for q and q in equation (5.54) from Eqs. (5.53) and (5.52) following
t i
equation is obtained :
F θ θ −θ
q = %
Q
+ Q
FV 4
F θ θ θ
or q = %
Q
+ Q
− ...(5.55)
FV 4 4
90 LINEAR CONTROL SYSTEMS
i.e. θ, therefore, the differential equation for the system is rewritten below neglecting the
In order to find the transfer function relating input heat energy q and the water
temperature θ o
as output, take Laplace transform the both sides of the equation (5.56)
Q(s) = Cs θ o
(s) + θ Q
U
4
From the above equation, the transfer function of the system relating θ o
(s) and Q(s) is
given by
θ Q
U 4
= ...(5.57)
3
U
4%U +
1. Field controlled D.C. motor. Transfer function of a field controlled d.c. motor
relating angular shift in the shaft and the input field voltage is derived hereunder.
The circuit diagram of a field controlled d.c. motor used for control systems is shown in
Fig. 5.6.1.
ia (const.)
Rf Ra La
Input Lf
Vf Jm f
m
Output qm
The input voltage V is applied to the field winding which has a resistance and
f
kept constant and thus the motor shaft is controlled by the input voltage V . The field current
f
i produces a flux in the machine which in turn produces a torque at the motor shaft. The
f
moment of inertia and the coefficient of viscous friction at the motor shaft are J and f
m m
Since the armature current is kept constant, the relationship between the developed
T
M
∝i f
...(5.58)
or T = K i
M f f
FK H
V = R i + .H V (s) = R I (s) + sL I (s) ...(5.59)
f f f f f f f f
FV
F θ F θ
T
M
= ,
O
O
+ H
O
O
T
M
(s) = s J
2
m
θ m
(s) + sf
m
θ m
(s) ...(5.61)
FV FV
F ω
T
M
= ,
O
O
+ H
O
ω O T
M
(s) = sJ
m
ω m
(s) + f
m
ω m
(s) ...(5.62)
FV
From equations (5.59) to (5.61), block diagram relating the output θ (s) and the input
m
Fig. 5.6.2 (a) Block diagram of a field controlled Fig. 5.6.2 (b) Block diagram reduction
θ O
U -
H
= ...(5.63)
8 H
U U
4
H
+ U. H
U,
O
+ H
O
From equations (5.59), (5.60) and (5.62), the block diagram as shown in Fig. 5.6.3 (a) is
The block diagram given in Fig. 5.6.3 (a) is reduced and shown in Fig. 5.6.3 (b).
Fig. 5.6.3 (a) Block diagram of a field controlled Fig. 5.6.3 (b) Block diagram reduction
ω O
U -
H
= ...(5.63 a)
8 H
U
4
H
+ U. H
U,
O
+ H
O
d.c. motor relating angular shift in the shaft and the input armature voltage is derived here
under.
The circuit diagram of an armature controlled d.c. motor used for control systems is
The input voltage V is applied to the armature which has a resistance of R and
a a
inductance of L . The field current i supplied to the field winding is kept constant and thus
a f
the armature input voltage controls the motor shaft output. The moment of inertia and the
coefficient of viscous friction at the motor shaft being J and f respectively. The angular
m m
As the field current is kept constant the relationship between the developed motor
T
M
∝i a
or T = K i ...(5.64)
M T a
The applied input voltage V is opposed by the back e.m.f. e developed in the
a b
Fθ
e
b
∝ω m
∵ ωm
=
O
FV
F θ
∴ e
b
= K
b
O
...(5.65)
FV
FK
C
V – e = R i + L V (s) – E (s) = R I (s) + sL I (s) ...(5.66)
a b a a a a b a a a a
FV
F θ O
e
b
= K
b
E (s) = sK
b b
θ m
(s) ...(5.67)
FV
or e
b
= K
b
ω m
E (s) = K
b b
ω m
(s) ...(5.67a)
F θ F θ
T
M
= ,
O
O
+ H
O
O
T
M
(s) = s J
2
m
θ m
(s) + s f
m
θ m
(s) ...(5.69)
FV FV
F ω
T
M
= ,
O
O
+ HO ω O
T
M
(s) = sJ
m
ω m
(s) + f
m
ω m
(s) ...(5.70)
FV
From equations (5.66) to (5.69) the block diagram relating the output θ m
(s) and the
sKb
Fig. 5.6.4 (a) Block diagram of armature controlled d.c. motor relating θm
(s) and V (s).
a
The block diagram given in Fig. 5.6.4 (a) is reduced and shown in Fig. 5.6.4 (b).
Va(s) + KT qm(s)
s(Ra+ sLa) (sJm + fm)
–
sKb
From equations (5.66), (5.67), (5.68) and (5.70) the block diagram as shown in Fig.
Kb
Fig. 5.6.5 (a) Block diagram of armature controlled d.c. motor relating ω m
(s) and V (s).
a
The block diagram given in Fig 5.6.5 (a) is reduced and shown in Fig. 5.6.5 (b).
Va(s) + KT wm(s)
(Ra+ sLa) (sJm + fm)
–
Kb
In the block diagram shown in Fig. 5.6.4 (b), the forward path transfer function is
-
6
G(s) =
U
4C + U. C
U, O + HO
θ O
U )
U
=
8C
U + )
U *
U
-
6
θ O
U U
4
C
+ U. C
U,
O
+ H
O
or =
8C
U -
+ 6
U-
D
U
4
C
+ U. C
U,
O
+ H
O
94 LINEAR CONTROL SYSTEMS
θ O
U -
6
Simplifying, =
8C
U U
4C + U. C
U, O + HO +
U- 6
- D
If the armature inductance is neglected, then the modified transfer function is given
by,
θ O
U -
6
= ...(5.71)
8C
U U4C
U, O + HO + U- 6
-D
θ O
U -
6
or =
8C
U U
U4C , O +4 C
HO + - 6 - D
LM -
6
OP
θ O
U N
4
C
H
O
+- 6
-
D
Q
or =
8C
U
U
LM U4C , O
+
OP
N 4
C
H
O
+- 6
-
D Q
θ O
U -
O
or = ...(5.72)
8C
U U
+ U6 O
-
6
where K =
m
4C H O + - 6 - D
4 ,
C O
and T =
m
4C H O + - 6 - D
The block diagram representing the above transfer function shown in Fig. 5.6.6.
Va(s) Km qm(s)
s(1+ sTm)
Since ω m
(s) = s θ (s)
m
Fig. 5.6.7 (a) Block diagram of armature controlled Fig. 5.6.7 (b) Simplified block diagram of
relating ω m
(s) and V (s).
a
ω O
U -
O
= ...(5.73)
8C
U
+ U6 O
MODELLING A CONTROL SYSTEM 95
Relation between torque constant K and back e.m.f. constant K
T b
T
M
ω m
= e i
b a
Since T
M
= K
T
i
a
and e
b
= K
b
ω
m
.
K
T
i ω
a m
= K
b
ω m
i .
a
Hence, K
T
= K .
b
3. Two phase a.c. servomotor. Transfer function of a two phase a.c. servomotor
relating the angular shift in the shaft and the input control voltage is derived hereunder.
The circuit diagram of a two phase a.c. servomotor is shown in Fig. 5.6.8.
A two phase a.c. servomotor is a two phase induction motor having drag cup type rotor
construction. The control voltage V (t) is applied to the control winding and a fixed voltage
c
having a phase difference of 90° w.r.t. control winding voltage is applied to the reference
winding. The control voltage results in the development of the motor torque T . The torque
M
Fig. 5.6.8. Two phase a.c. servomotor. Fig. 5.6.9. Torque speed characteristics of a
The moment of inertia and the coefficient of viscous friction at the motor shaft are
respectively J
m
and f . The angular shift resulting in the motor shaft being
m
θ m
and the
From the torque-speed characteristics the dynamic equation relating the motor torque
T
M
= m ω m
+ KV
c
...(5.74)
If T
0
is the stalling torque and ω 0
is the no load speed. The parameters m and K are
determined in terms of T
0
and ω 0
as follows:
(i) When the speed is zero, the torque is T (stalling torque) and this stalling torque is
0
Therefore, T = KV
0 c
6
or K = expressed in Nm/V
8E
6
m = – expressed in Nm/(rad/sec) is a negative number.
ω
96 LINEAR CONTROL SYSTEMS
F θ
ω
O
Since, =
m
FV
F θ
T
M
= O
O
+ -8
E ...(5.75)
FV
F θ
T
M
= O
O
+ -8
E T
M
(s) = sm θ m
(s) + KV (s)
c
...(5.76)
FV
F θ F θ
T
M
= ,
O
O
+ H
O
O
T
M
2
(s) = s J
m
θ m
(s) + sf
m
θ m
(s) ...(5.77)
FV FV
In equations (5.76) and (5.77) the block diagram relating the output θ m
(s) and the
The block diagram given in Fig. 5.6.10 (a) is reduced in following steps and shown in
Fig. 5.6.10 (a) Block diagram of two phase a.c. Fig. 5.6.10 (b) Block diagram reduction
servomotor relating θ m
(s) and V (s).
0
of F ig. 5.6.10 (a).
K
Vc(s) (fm – m) qm(s)
Vc(s) Km qm(s)
sJm
s 1 + (f – m) s(1+ sTm)
m
Fig. 5.6.10 (c) Block diagram reduction of Fig. 5.6.10 (d) Simplified block diagram of two
According to Fig. 5.6.10 (d) the transfer function of a two phase servomotor relating
the output θ
m
(s) and the input V (s) is given by the relation
c
θ O
U -
O
= ...(5.78)
8E
U U
+ U6 O
-
where K =
m
H
O
− O
is termed as motor gain constant.
MODELLING A CONTROL SYSTEM 97
,
O
and T = is termed as motor time constant.
m
HO − O
Since, ω (s) = s
m
θ m
(s) the block diagram relating ω m
(s) and V (s) is shown in Fig. 5.6.11
c
Fig. 5.6.11 (a) Block diagram of two phase a.c. Fig. 5.6.11 (b) Simplified block diagram
servomotor relating ω m
(s) and V (s).
c
of an a.c. two phase motor relating
ω m
(s) and V (s).
c
The transfer function relating the output ω (s) and the input V (s) is thus
m c
ω O
U -
O
= ...(5.79)
8E
U
+ U6 O
5.7 GENERATORS
1. (a) Separately excited d.c. generator. The transfer function of a separately excited d.c.
generator relating the input field winding voltage V and the induced voltage developed
f
The circuit diagram is shown in Fig. 5.7.1. The resistance and the inductance of the
field winding are respectively R and L . The generator being driven at a constant speed and
f f
is operating in the linear portion of the magnetising characteristics. The current i flowing in
f
the field winding produces a flux in the generator which in turn induces a proportional
voltage at the armature terminals. The relationship between the induced voltage e and the
Rf if
e
Vf Lf G e D if
Input Output
De
Kg =
D if
if
Fig. 5.7.1. Separately excited d.c. generator. Fig. 5.7.2. Magnetisation characteristics.
e ∝i f
or e = K i ...(5.80)
g f
equations are
FK H
V
f
= 4H KH + .H V (s) = R I (s) + sL I (s)
f f f f f
...(5.81)
FV
From equations (5.81) and (5.82) the block diagram relating the output E(s) and the
input V (s) is drawn and shown in Fig. 5.7.3 (a) and simplified as shown in Fig. 5.7.3 (b).
f
Fig. 5.7.3 (a) Block diagram of separately Fig. 5.7.3 (b) Simplified block diagram of
'
U -
I
= ...(5.83)
8
H
U
4
H
+ U. H
Transfer function of a separately excited d.c. generator relating the input field winding
In this case the output of the d.c. generator is connected to load having a resistance
R . The resistance of the armature being R and the armature inductance is considered
L a
equations are
8 8
U
V V
I = I (s) = ...(5.85)
L L
4. 4.
From equations (5.84) and (5.85) the block diagram relating V (s) and E(s) is drawn
t
input field winding voltage V (s), the two block diagrams shown in Figs. 5.7.3 (b) and 5.7.5
f
are joined as shown in Fig. 5.7.6 which is reduced in steps as shown in Fig. 5.7.7 (a) to (c).
E(s) + Vt(s)
1
–
Ra IL(s)
IL(s) 1
Ra
RL
Ra IL(s) 1
Ra
RL
Fig. 5.7.6. Block diagram of F ig. 5.7.3 (b) and Fig. 5.7.5 connected.
Vf (s) Kg 1 Vt(s)
E(s)
(Rf + sLf ) Ra
1+
RL
Vf (s) Kg R L Vt(s)
(Rf + sLf ) (RL + Ra)
Fig. 5.7.7 (c) Block diagram of separately excited d.c. generator relating V (s) and V (s).
t f
According to Fig. 5.7.7 (c) the transfer function relating the output V (s) and the input
t
8
U - 4
I .
V
= ...(5.86)
8 H
U
4
H
+ U. H
4. + 4C
output voltage and the input field winding voltage is derived hereunder.
100 LINEAR CONTROL SYSTEMS
Fig. 5.7.8. The input voltage e applied to the main field winding having a resistance and an
f
inductance of R and L respectively and the output voltage e is developed across the brushes
f f
c and d.
Rq iq
Lq
b
Rf Lf if f2
c d e1
f1
ef a
Input
e, Output
which in turn developes an e.m.f. e across the brushes a and b. The resistance and the
1
inductance of the armature winding across the brushes a-b are respectively R and L .
q q
e
1
∝i f
or e = K i ...(5.87)
1 1 f
The e.m.f. e developed across the brushes a-b is short circuited on itself. The short
1
e ∝i q
or e = K i ...(5.88)
2 q
equations are
FK H
e = R i + L E (s) = R I (s) + sL I (s) ...(5.89)
f f f f f f f f f
FV
FKS
e = R i + L E (s) = E I (s) + sL I (s) ...(5.91)
1 q q q 1 q q q q
FV
From equations (5.89) to (5.92) the block diagram relating the output E(s) and the
input E (s) is drawn and shown in Fig. 5.7.9 (a) and reduced as shown in Fig. 5.7.9 (b).
f
MODELLING A CONTROL SYSTEM 101
Hence, the transfer function relating the output E(s) and the input E (s) is given by,
f
'
U - -
= ...(5.93)
'
H
U
4
H
+ U. H
4
S
+ U. S
V/(rad/sec).
E(s) = K
tg
ω(s) ...(5.95)
'
U Fig. 5.7.11. Block diagram of d.c.
= K ...(5.96)
ω
U tg tachogenerator.
induction generator.
Output
The tachogenerator is coupled to a shaft, of which w winding
speed is to be measured. The reference winding is supplied Input
by a reference voltage and the output voltage e is induced Fig. 5.7.12. A.C.
The amplitude (or the r.m.s. value) and the phase of the output voltage depends on the
direction of rotation. The output voltage e is thus related to the shaft speed as follows :
e ∝ω
or e = K
tg
ω ...(5.97)
equation is,
E(s) = K
tg
ω(s) ...(5.98)
w(s) E(s)
Ktg
The block diagram relating the output E(s) and the input ω(s) is drawn and shown in
Fig. 5.7.13. and the transfer function is given by,
'
U
= K ...(5.99)
tg
ω
U
An a.c. tachogenerator can also be used on d.c. systems by using a phase sensitive
rectifier.
Error detectors are used to measure (sense) the error signal in control systems. The error is
the difference between the actual magnitude of the output and the magnitude of the desired
output. The magnitude of the desired output is held constant as a reference and also termed
as input. Thus any deviation from the desired output (i.e. input or reference) results in an
error signal.
In a way error detectors also perform the function of a transducer, i.e. to convert error
signal obtained in one form to another form which is needed for the control action. For
example in the control for the position of a shaft the error is the angular difference in the
positions of the input (reference) shaft and the output shaft and is expressed in radian or
degree. A potentiometer error detector measures this error and converts the same into a
proportional voltage.
+
– e=Ke(qr – qo)
qo Output
qr
The input shaft is coupled to the potentiometer marked A and is held fixed at a desired
e ∝ (θ r
– θ o
)
or e = K (
e
θ r
– θ o
) ...(5.100)
where K known as the gain of the error detector, and expressed is V/rad.
e
or e = K
e
θe
...(5.101)
where θ e
θ
= (
r
– θ o
).
E(s) = K [
e
θ (s) – θ
r o
(s)] ...(5.102)
or E(s) = K
e
θ (s)
e
...(5.103)
The block diagram relating the output E(s) and the input θ (s) drawn and shown in
r
Fig. 5.8.2.
qo(s)
Fig. 5.8.2. Block diagram of potentiometer error detector.
'
U
= K ...(5.104)
e
θ G
U
'
U
or = K (5.105)
e
θ T
U −θ Q
U
2. Synchro (selsyn) error detector. Fig. 5.8.3 shows synchro transmitter and
synchro control transmitter arrangement for converting angular position difference into a
The rotor of the control transformer is coupled to the output shaft of the control
θ e
= ( θ
r
– θ o
) between the positions of the input (reference) and the output shafts.
explained below :
e (t) = E
i m
sin (2 π ft) ...(5.106)
voltage induced by transformer section across the stator winding 1n is given by,
e
1n
= K E
m
sin (2 π ft) ...(5.107)
where K is a constant of proportionality. As the stator windings 2n and 3n are 240° and 120°
apart (angle measured as positive in anti-clockwise direction) w.r.t. the winding 1n, the
e
2n
= K E
m
π ft) cos (240°)
sin (2
Equations (5.107), (5.108) and (5.109) indicate that three voltages e , e and e are
1n 2n 3n
Now, if the rotor of the synchro transmitter shifts in anti-clockwise direction through
e
2n
= K E
m
sin (2π ft) cos (240° – θ) ...(5.111)
Equations (5.110), (5.111) and (5.112) reveal that magnitudes of the voltages e , e
1n 2n
and e
3n
vary sinusoidally w.r.t. θ as shown in Fig. 5.8.4.
MODELLING A CONTROL SYSTEM 105
Fig. 5.8.4. Terminal voltage across stator of synchro transmitter w.r.t. rotor position (radian).
The three voltages e , e and e are connected consecutively to three stator windings
1n 2n 3n
of the control transformer and produce a resultant flux in the air gap of the same, which in
turn induces a voltage across the rotor winding of the control transformer. The magnitude of
phase for ( θ r
– θo
) = 270° and finally zero for ( θ r
– θ o
) =
Therefore, the magnitude of the output induced voltage e developed across the rotor of
e = K
s
sin ( θ r
– θ o
) ...(5.113)
It is to be noted that the frequency of the voltage e is the same as that of supply and
sin ( θ r
– θ ) ≅ (θ
o r
– θo
)
Hence, e = K (
s
θ r
– θ o
)
∴ e = K
s
θ e
...(5.113)
K is expressed in V/rad.
s
The above equation indicates that angular error is converted into a proportional
E(s) = K
s
θ e
(s) ...(5.114)
The block diagram representation of synchro error detector is shown in Fig. 5.8.6.
'
U '
U
= K , or = K ...(5.115)
s s
θ G
U = θ T
U −θ Q
U?
It is to be noted that the phase of e (the output voltage) reverses at points a and b as
Example 5.9.1. Find the transfer function relating displacements y and x for the
Solution. Let force f (t) be applied at point A. The mechanical circuit diagram for the
From the mechanical circuit diagram the equations for the system are
F[ F
H
+ -[ + H
[ − Z = 0 ...(1)
FV FV
F
H (x – y) = f (t) ...(2)
FV
Assuming initial conditions as zero and taking Laplace transform on both sides of
x
A
f1
y
y f1
x
A
f2 K
f2 K f(t)
B
B
Fig. 5.9.1. F ig. for example 5.9.1. Fig. 5.9.1 (a) Mechanical circuit diagram for example 5.9.1.
;
U H U 6 U
=
=
Ans.
:
U =
H + H U + -?
6 U +
H
H
+ H
where T = and T =
1 2
- -
Example 5.9.2. Draw the mechanical circuit diagram for the system shown in
f(t) x1 x2
K1 f3 K2
M1 M2
f1 f2
Fig. 5.9.2. System for example 5.9.2.
X1 X2
f3
f(t) K1 M1 f1 f2 M 2 K2
F Z FZ F
At node (1) K x
1 1
+ /
+ H
+ H
Z
−Z
= f (t) ...(1)
FV FV FV
F Z FZ F
At node (2) K x
2 2
+ /
+ H
+ H
Z
−Z
= 0 ...(2)
FV FV FV
equations (1) and (2), transformed equations for the system are written below :
2
M s X (s) + (f + f ) sX (s) + K X (s) – f sX (s) = F(s) ...(3)
1 1 1 3 1 1 1 3 2
2
– f sX (s) + M s X (s) + (f + f ) sX (s) + K X (s) = 0 ...(4)
3 1 2 2 2 3 2 2 2
Ans.
Example 5.9.3. Obtain a mathematical model for the mechanical system shown in
Fig. 5.9.3.
Solution. The mechanical circuit diagram for the circuit is drawn in Fig. 5.9.3 (a).
On the basis of force-current analogy the equations for the system are written below :
FZ F Z F
Z −Z
H
+ -
Z
+ /
+ -
Z
−Z
+ H
= 0 ...(1)
FV FV FV
F Z F
Z −Z
/
+ -
Z
−Z
+ H
= f (t) ...(2)
FV FV
equations (1) and (2) and rearranging terms, following equations representing the system
are obtained :
LM /
U +
H + H U +
- + - −
H U +- OP L :
U OP LM OP(
U
PQ MN
= Ans.
MN −
H
U +-
/U
+ H
U +-
:
U
Q N Q
MODELLING A CONTROL SYSTEM 109
X1 K2 X2
f1
M1 f2 M2
K1 f(t)
Example 5.9.4. The diagram given in Fig. 5.9.4 represents a closed-loop control
system for regulating the speed of a field controlled d.c. motor. Determine the value of the
Amplifier ia (const.)
+ Tacho.Gen.
0.1 V/rpm
–
Solution. As K the motor field constant is given in rpm/V; the resistance and
inductance of the motor field winding are not needed for calculations.
Amplifier Motor
Vr + Vf Output
2 10
– N = 100 rpm
0.1
Tacho. gen.
Fig. 5.9.4 (a) Block diagram of F ig. 5.9.4.
0 ×
= =
8 + × ×
T
∴ V
r
= N × [ ∵ N = 100 rpm.]
V = 100 × = 15 V. Ans.
r
110 LINEAR CONTROL SYSTEMS
Example 5.9.5. The scheme given in Fig. 5.9.5 represents a voltage regulator,
Amplifier Generator
+ Rf 600 W
Vr – e 2 V/V V1 G Vt = 250 V
Lf 400 W
if
kg = 500 V/A
Solution. The fraction of the output voltage compared with the reference voltage is
400/(400 + 600) = 0.4. The generator field inductance need not be considered as the system
block diagram for the steady state condition is to be drawn. The block diagram considering
Amplifier Generator
Vr + e Vf 1 1 if Vt
2 = Kg = 500
Ra 100
–
0.4
8
× ×
V
= =
8T
+× × ×
∴ V
r
= V
t
× [ ∵ V
t
= 250 V]
V
r
= × = 125 V. Ans.
If the generator terminal voltage is to be kept constant within 2% of the full load value
E = V + i R ...(1)
i t L a
V = i R ...(2)
t L L
8 K 4 K 4
V
=
. .
= . .
'K 8V +K .
4. K. 4. +K .
4C
8 4
V .
or = ...(3)
'K 4. + 4C
'
K
g
=
K
= = 300 V/A
KH
Er + e if Ei RL Vt
KA Kg
Ref. – RL + R a
0.1
The block diagram under steady state condition for the system is developed as shown
8
V
=
given
'
K
=
Therefore, from equation (3)
4 8
.
=
V
= = 0.88
4. + 4C 'K
Substituting the above values in Fig. 5.9.6 (a), the block diagram shown in
Er + e Vt
KA 300 0.88
–
0.1
Fig. 5.9.6 (b) Block diagram of Fig. 5.9.6 (a) redrawn with numerical values.
The transfer function relating the error e and the reference voltage E is given by
r
G
=
LM 4GH
G
=
OP
'
T
+ -
#
× × × N 4 + )* Q
G
or = ...(4)
'
T
+ - #
112 LINEAR CONTROL SYSTEMS
e =
× ×
Substituting equations (5) and (6) in equation (4) following relation is obtained.
= ...(7)
+ - #
Example 5.9.7. The field winding of a separately excited d.c. generator has an
below :
Field current i in A 0 2 4 6 8 10
f
Determine the transfer function of the generator relating the open circuit terminal
Solution. The circuit diagram and the magnetisation characteristics of the generator
200
e
Rf if
150
Vf G e 100 De
Dif
Lf
50
2 4 6
Fig. 5.9.7 (a) Circuit diagram for example 5.9.7. Fig. 5.9.7 (b) Magnetisation characteristics
∆G
The generator field constant is K = = = 25 V/A
g
∆K H
MODELLING A CONTROL SYSTEM 113
The system equations are V (s) = R I (s) + sL I (s) ...(1)
f f f f f
Eliminating I (s) from equations (1) and (2) the transfer function relating E(s) and
f
'
U -
I
= ...(3)
8
U
H
4
H
+ U.
H
'
U
= = Ans.
8
U
H
+ U× + U
Example 5.9.8. From the diagram of an open-loop control system represented below
(Fig. 5.9.8).
(b) Determine the transfer function relating angular velocity of the load and the input
voltage V .
f
(c) If the system is to be made a closed-loop one, suggest a method and incorporate the
if (const.)
Rf Ra La
Kg
Vf Lf eg eb JL
w
Jm
R
f
= 1000 Ω, L
f
= 100 H, K
g
= 1000 V/A
–4 2
K = 1.2 V/(rad/sec), J = 0.3 × 10 kg-m ,
b L
–4 2
J = 0.2 × 10 kg-m .
m
Neglect friction.
Solution. (a) The generator, armature controlled motor and load are connected in
cascade. The block diagrams for these elements are drawn as per procedure explained earlier
in this chapter and the block diagram relating the output ω(s) and the input V
f
(s) is
Substituting the values of system parameters the block diagram shown in Fig.
Kb
1.2
Fig. 5.9.8 (b) Values of system parameters substituted in F ig. 5.9.8 (a).
Using block diagram reduction technique the block diagram of Fig. 5.9.8 (b) is reduced
1.2
(b) From the block diagram of Fig. 5.9.8 (d) the transfer function relating ω(s) and
ω
U ×
=
8
U
H
+ U
U
+ U + ×
×
=
. Ans.
+ U
U + U + ×
ω UU
×
=
8
H
+ ×
+ × + ×
ω
UU
= =
8
H
∴ V
f
= 1.2 ω ss
.
MODELLING A CONTROL SYSTEM 115
In order to convert the system into a closed-loop one, a tachogenerator feedback is
used. The transfer function of the feedback block will have a gain of 1.2 V/(rad/sec). The block
1.2
Fig. 5.9.8 (e) Open-loop system of Fig. 5.9.8 (d) converted into closed-loop system.
Example 5.9.9. A torque controller uses a split field motor for position control. The
error detector gain is 10 V per radian error and the amplifier transconductance is 100 mA/
–4
V. The motor torque constant is 5 × 10 Nm/mA. The moment of inertia and the coefficient
–5 2 –4
of viscous friction at the motor shaft are respectively 1.25 × 10 kg-m and 5 × 10 Nm/
(rad/sec). The motor is coupled to the load through a gear having a ratio 20 : 1.
Draw the block diagram of the system showing the transfer function of each block and
determine the overall transfer function relating the output and input.
Solution. The system described in this example is a closed-loop control system for
position control. The positional output is compared with the reference input by an error
detector. The error thus obtained is amplified in the form of appropriate signal and applied
to the motor for producing controlling torque. The output of the motor is coupled to the
The block diagram of the system based on the aforesaid sequence is developed and
–4
K = 10 V/rad, K = 100 mA/V, K = 5 × 10 Nm/mA
e A T
–5 2 –4
J = 1.25 × 10 kg-m , f = 5 × 10 Nm/(rad/sec),
0
=
0
Substituting the above values in Fig. 5.9.9 the block diagram given in Fig. 5.9.9 (a) is
obtained.
R(s) + 1 1 C(s)
10 100 5 × 10–4 20
– 1.25 × 10–5s2 + 10–4s
The block diagram of Fig. 5.9.9 (a) is reduced as shown in Fig. 5.9.9 (b) and further
From the block diagram (Fig. 5.9.9 c), the transfer function of the system is
determined below :
%
U
U + U
= = Ans.
4
U
+
×
U + U +
U + U
resistance from a 24 d.c. supply. The motor takes a current of 5 A on stalling and the
The motor runs at 1000 rpm taking a current of 1 A. The moment of inertia and
–3 2 –3
coefficient of viscous friction are 4 × 10 kg-m and 1.5 × 10 Nm/(rad/sec) respectively.
Solution. T = K i
stall T stall
∴ 0.915 = K
T
× 5
∴ K
T
= 0.183 Nm/A
8
i
stall
= ∴ 5 =
4 4
C C
∴ R
a
= = 4.8 ohm
e = V – i R = 24 – 1 × 4.8 = 19.2 V
b a a
As e
b
= K
b
ω and ω= π×
π ×
∴ 19.2 = -
D
×
or K = 0.183 V/(rad/sec)
b
MODELLING A CONTROL SYSTEM 117
If the armature inductance is neglected, the block diagram of an armature controlled
V(s) + KT q(s)
sRa(sJm + fm)
–
sKb
Fig. 5.9.10. Block diagram of an armature controlled d.c. motor neglecting armature inductance.
From Fig. 5.9.10 the transfer function of the motor is determined below :
-
6
θ
U U4C
U,O + HO
=
8
U -
+ 6
U-
D
U4C
U,O + HO
θ
U -
6
or =
8
U U
U4 ,
C O
+4 C
H
O
+- 6
-
D
θ
U
=
8
U U
U × × − + × × − + ×
= = . Ans.
U
U + U
U +
Example 5.9.11. Fig. 5.9.11 shows a positional control system for controlling the
position of a shaft. The potentiometer error detector measures the deviation of the output
of which is fed to an armature controlled d.c. motor. The motor shaft is coupled to the load
–5
Motor torque constant K = 10 × 10 Nm/A
T
–5
Motor back e.m.f. constant K = 10 × 10 V/(rad/sec)
b
–5 2
J = 10 × 10 kg-m
m
–5
f = 10 × 10 Nm/(rad/sec)
m
Draw the block diagram and determine the overall transfer function relating the
+
θr – θc
Ra N1 = 5
e KA M
θc L
o
if (const.) a
d
N2 = 50
Solution. The error detector, amplifier, motor and gear are arranged in the block
qr(s) + + KT N1 qc(s)
Ke KA
sRa(Jeqs + feq) N2
– –
sKb
Fig. 5.9.11 (a) Block diagram of control system for example 5.9.11.
Substituting system parameter values the block diagram of Fig. 5.9.11 (a) is redrawn
qr(s) + + 10 × 10 –5 5 qc(s)
2 10
–5 –5 50
– – s × 0.2(10 × 10 s + 50 × 10 )
5 × 10 × 10–5
Fig. 5.9.11 (b) System parameter values substituted in F ig. 5.9.11 (a).
The block diagram of Fig. 5.9.11 (b) is reduced in following steps and shown in Fig.
qr(s) + + 5 1 qc(s)
20
– s2 + 5s 10
–
s × 10 × 10–5
qr(s) + 5 1 qc(s)
20 2
s + 5.0005s 10
–
qr(s) + 10 qc(s)
2
– s + 5.0005s
From Fig. 5.9.11 (e) the overall transfer function is determined below
θ E
U U
+ U
= = Ans.
θ
U
U
+ U +
T
+
×
U + U
Example 5.9.12. For the closed-loop control system shown in Fig. 5.9.12. Draw the
+ ia (const.)
θr θc
–
N1 = 10
Rf
e KA Lf M θc
JL, fL
N2 = 100
Fig. 5.9.12. Closed-loop control system for example 5.9.12.
2
Moment of inertia for load J = 3.0 kg-m .
L
0
2
J = J + J = 0.02 + 3.0 = 0.05 kg-m
eq m L
0
0
f = f + f = 0.03 + 4.5 = 0.075 Nm/(rad/sec)
eq m L
0
The error detector, amplifier, motor gear and load are connected in cascade. The block
diagrams for these elements are drawn as per the results obtained in earlier sections and the
qr(s) + Kf 1 N1 qc(s)
Ke KA
(Rf + sLf) (Jeqs2 + feqs) N2
–
Substituting the numerical values the block diagram of Fig. 5.9.12 is redrawn as
Fig. 5.9.12 (b) Block diagram of Fig. 5.9.12 (a) with numerical values substituted.
The block diagram in Fig. 5.9.12 (b) is reduced as shown in Fig. 5.9.12 (c).
The block diagram of Fig. 5.9.12 (c) is simplified and shown in Fig. 5.9.12 (d).
θ E
U U
+ U
+ U
=
θ
U
T
+ ×
U
+ U
+ U
=
= U
+ U
+ U + ?
=
U + U + U +
θ E
U
or = Ans.
θ T
U
U
+ U + U +
Example 5.9.13. A closed-loop speed control system uses a two-phase a.c. motor. The
error signal is amplified by an amplifier the output of which is connected to the motor. The
The amplifier gain is 20 V/V and tachogenerator used in the feedback path has a gain
constant K = 0.2 V/(rad/sec). The moment of inertia and coefficient of viscous friction
tg
–4 2 –4
referred to motor shaft side are respectively 1.5 × 10 kg-m and 1 × 10 Nm/(rad/sec).
MODELLING A CONTROL SYSTEM 121
The motor has a stall torque of 0.12 Nm at 100 V input and the torque decreases by
Solution. The characteristics of the two phase motor is drawn in Fig. 5.9.13 (a).
0.12
T(Nm)
0.6
0 2000 N(rpm)
Fig. 5.9.13 (a) Torque-speed characteristics of two-phase motor of example 5.9.13.
From the torque speed characteristics [Fig. 5.9.13 (a)] the constants m and K of the
− –4
m = = – 3 × 10 Nm/(rad/sec)
π
×
–4
K = = 12 × 10 Nm/V
The block diagram for the system is developed and shown in Fig. 5.9.13 (b).
Ktg
Substituting numerical values, the block diagram of Fig. 5.9.13 (b) is shown in Fig.
5.9.13 (c).
0.2
Fig. 5.9.13 (c) Block diagram of 5.9.13 (b) with numerical values.
122 LINEAR CONTROL SYSTEMS
The block diagram of Fig. 5.9.13 (c) is reduced and shown in Fig. 5.9.13 (d).
0.2
The block diagram of Fig. 5.9.13 (d) is simplified and shown in Fig. 5.9.13 (e).
R(s) + 24 w(s)
1.5s + 4
–
0.2
From the block diagram shown in Fig. 5.9.13 (e) the transfer function of the system is
determined below
ω
U
U +
= = . Ans.
4
U
U +
+ ×
U +
Example 5.9.14. A separately excited d.c. motor is supplied at its field winding by a
constant voltage and the armature is connected to a voltage of 500 V. The resistance and the
inductance of the armature are respectively 0.75 ohm and 5 mH. The motor and the load
2
has an equivalent moment of inertia of 50 kg-m . The friction is negligible. The speed of the
Draw the block diagram representation of the motor and determine the transfer
function of the motor relating the load torque and the motor speed.
π0 π ×
ω m
= = = 52.35 rad/sec
In this example as the load torque is also applied, the accelerating torque T for the
A
load is the difference between the developed motor torque T and the load torque T , hence
M L
T = T – T
A M L
or T = T + T ...(1)
M A L
8
U
C
− '
U
D
∴ I (s) = ...(3)
a 4
C
+ U. C
MODELLING A CONTROL SYSTEM 123
∵ T
M
(s) = K
T
I (s)
a
...(4)
Substituting for I (s) in equation (4) from equation (3) following relation is obtained
a
-
6
T (s) = [V (s) – E (s)] × ...(5)
M s b
4
C
+ U.
C
T
M
(s) = Js ωm
(s)
T
M
(s) = Js ω m
(s) + T (s)
L
or ω m
(s) = [T
M
(s) – T (s)]
L
...(6)
,U
E (s) = K
b b
ω m
(s) ...(7)
Using equations (5), (6) and (7) the block diagram as shown in Fig. 5.9.14 (a) is
obtained.
TL(s)
Va(s) – Eb(s)
–
Va(s) + KT Tm(s) + 1 wm(s)
Ra + sLa Js
–
Eb(s)
Kb
There are two inputs V (s) and T (s). If V (s) is held at a fixed value and the only the
a L a
effect of T (s) is considered, then in such a case the effect of V (s) is taken as zero and the
L a
block diagram given in Fig. 5.9.14 (a) takes the form as shown in Fig. 5.9.14 (b).
–TL(s) + 1 wm(s)
Js
–
KT Kb
Ra + sLa
Fig. 5.9.14 (b) Block diagram of F ig. 5.9.14 (a) considering V (s) = 0.
a
124 LINEAR CONTROL SYSTEMS
ω
U
O ,U
=
− 6
U -
6
-
D
.
+
,U
4
C
+ U. C
ω O
U
4
C
+ U.
C
or = Ō ...(8)
6
.
U = ,U
4
C
+ U. C
+ -
6
-
D
?
As the speed is 52.35 rad/sec at 500 V (effect of armature drop is neglected), therefore,
K = = 9.55 V/(rad/sec)
b
K
T
= K
b
∴ K
T
= 9.55 Nm/A
ω O
U
+ × −
U
=
−
6
.
U = U
+ × U + × ?
−
+ × U
+ U
= – = –
U + U +
U
+ U +
+ U
= – Ans.
U + U +
PROBLEMS
5.1. Find the transfer function relating displacements y and x for the mechanical system shown in
Fig. 5P-1.
f1
f2
K
y x
Fig. 5P-1. Mechanical system for problem 5.1.
5.2. On the basis of force-current analogy write the equations for the system given in Fig. 5P-2.
loop control system for speed control using a field Fig. 5P-3. Mechanical system for
controlled d.c. motor. Motor rpm to field current ratio is problem 5.3.
ia (const.)
+ if
e KA M
– 100 rpm
Vr
+ Tacho gen.
–
Fig. 5P-4. Diagram for problem 5.5.
5.6. A closed-loop position control system has a controller gain constant K Nm/radian error, the
T
2
moment of inertia and the coefficient of viscous friction are respectively J kg-m and f N-m/
5.7. The moment of inertia J and the coefficient of viscous friction f for a field controlled d.c.
m m
–4 2 –4
servo-motor are respectively 5 × 10 kg-m and 12.5 × 10 Nm/(rad/sec). The motor torque
constant K being 2.5 Nm/A. Draw the block diagram and determine the transfer function
f
relating the angular speed of the shaft and the field current.
5.8. An armature controlled d.c. motor has an armature resistance of 0.37 ohm. The moment of
–6 2
inertia is 2.5 × 10 kg-m . A back e.m.f. of 2.09 V is generated per 100 rpm of the motor speed.
The torque constant of the motor is 0.2 Nm/A. Determine the transfer function of the motor
5.9. A control system is shown in Fig. 5P-5. The armature controlled d.c. motor is connected to a d.c.
generator. Determine the transfer function of the system relating the output angular velocity
and the input field voltage V . The parameters of the system are given bleow:
f
Rf Ra
Vf Lf eg eb J
w
if
(const.)
GENERATOR MOTOR LOAD
Fig. 5P-5. Control system for problem 5.9.
–4 2
Combined inertia of motor and load J = 0.5 × 10 kg-m
Friction is negligible, R
f
= 1000 Ω, L f
= 20 H . R
a
= 50 ohm
L is negligible.
a
–4
T = (8 V – 0.3 N) × 10 Nm
M m
where N is speed in rpm and V is the input voltage. The motor is used through a gear ratio of
m
10 : 1 for position control. The gain of the error detector is 0.1 V/degree. The amplifier gain is
25 V/V. Determine the transfer function of the system if the coefficient of viscous friction and
–4
the moment of inertia referred to motor shaft side are repectively 50 × 10 Nm/(rad/sec) and
–4 2
80 × 10 kg-m .
5.11. Fig. 5P-6 given below gives a block diagram for speed regulator system. The accelerating
torque is the difference between the torque developed by the controller and load torque T .
L
The controller gain is 0.00102 Nm/rad and the tachogenerator constant is 0.191 V/(rad/sec).
TL(s)
–
Vr(s) + Tc(s) + TA(s) 1 w(s)
KT
Js
–
Ktg
The no load speed is adjusted to 1000 rpm. The moment of inertia is J and the friction is
negligible.
5.12. A separately excited d.c. motor is directly coupled to a load. The combined moment of inertia
of the load and the motor is J and friction is negligible. The armature resistance is R and
a
Derive the transfer function relating the motor speed and the applied armature voltage and
load torque as independent variables. Also obtain an expression for steady state speed.
6
of Control Systems
his chapter deals with the time response analysis of feedback control systems. The test signals are
T specified as unit step, unit ramp, unit parabolic and unit impulse. The time response is divided in two
The transient response of a second order control system can be underdamped, critically damped, over-
damped depending on the value of damping ratio. The reduction in damping ratio tends towards
underdamped response and in the absence of damping sustained oscillations are noted.
The expressions for the time response for a specified input have been derived explaining each step.
The transient response specifications for a second order control system subjected to a unit step
function are maximum overshoot, under shoot, natural and damped frequency of oscillations, rise time,
The higher order control system has the degree of s in the characteristic equation as more than 2.
However, the higher order control system can be approximated to second order as per the s-plane
location of the roots (poles of closed-loop transfer function) of the characteristic equation.
At the end of the transient response of higher order control systems, the concept of dominant poles is
explained.
The improvement in the time response is achieved by applying control actions as described in the text.
CONTENTS
127
128 LINEAR CONTROL SYSTEMS
Introduction. In the earlier chapter the process of modelling a control system has been
described, now hereunder time response analysis of control systems is dealt with. As the
input to a control system cannot be assessed before hand, therefore, an input test signal is
Specified input is one to which a control system is frequently subjected to. Once a
control system responds satisfactorily to a test input signal, its time response to actual input
In the initial part of time response of a control system transients appear and during the post
transient part steady state is achieved. Theoretically steady state means a state of the
output of a control system as the time approaches infinity after initiation of the input. In
actual practice the steady state period and transient period is identified in terms of time
constants of a control system. Thus, the time response of a control system is divided in two
parts namely :
Typical time response of a control system for a specified input test signal is shown in
Fig. 6.1.1.
Output
Steady State
Error
Input
c(t)
r(t) = 1
t
t Transient Steady
State
(a) Input test signal (b) Output response
Fig. 6.1.1. Typical time response of a control system.
The transient part of time the response reveals the nature of response (i.e. oscillatory
The steady state part of time response reveals the accuracy of a control system. Steady
state error is observed if the actual output does not exactly match with the input.
Specified input test signals applied for time response analysis of a control system are
described below :
(1) Step function. Step function is described as sudden application of input signal as
0 t
Fig. 6.2.1. Step function/displacement function. Fig. 6.2.2. Ramp function/velocity function.
If R = 1 unit, the step function is called unit step function and the corresponding
Laplace transform is
L1 = .
U
If R = 1, then r(t) = t and the ramp function is called unit ramp function and the
Lt=
.
U
r(t)
r(t) = Rt2/2, for t ³ 0
r(t) = 0 , for t < 0
Slope Rt
0 t
Fig. 6.2.3. Parabolic function/acceleration function.
V
If R = 1, then r(t) = and the parabolic function is called unit parabolic function and
r(t)
L
F I
V
GH JK
=
U
.
It is a well known fact that step function is a first time derivative of ramp function and
F
Unit impulse function = (unit step function)
FV
Hence, the Laplace transform of unit impulse function is derived from the Laplace
F
L (unit impulse function) = L (unit step function)
FV
A first order control system is one wherein highest power of s in the denominator of its
transfer function equals 1. Thus a first order control system is expressed by a transfer
%
U
= ...(6.1)
4
U U6 +
6.3.1. Time response of a first order control system subjected to unit step input
function
The output for the system is expressed as
C(s) = R(s) ...(6.2)
U6 +
As the input is a unit step function
r(t) = 1 and R(s) =
U
C(s) = ...(6.3)
U U6 +
Breaking R.H.S. of Eq. (6.3) into partial fractions
C(s) = Ō
U U6 +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 131
or C(s) = Ō ...(6.4)
U
U +
6
L – 1
C(s) = L – 1
Ō
U
U +
6
– t/T
c(t) = 1 – e ...(6.5)
– t/T
The error is given by e(t) = r(t) – c(t) = 1 – (1 – e )
– t/T
= e ...(6.6)
ŌV 6
The steady state error is e = NKO G = 0
ss
V →∞
The graphical representation of the time response shown in Fig. 6.3.2 indicates that
the response is exponential type and the steady state value is 1 unit. As the time increases
the disparity between the output and input approaches to nil, hence, the steady state error is
zero.
0.02
c(t) 0.98 1
0.63
O T 4T t ¥
Transient Steady State
Fig. 6.3.2. T ime response of a first order control system subjected to unit step function.
6.3.2. Demarcation between the transient part and steady state part of the time
response in terms of time constant
The magnitude of the output time response equals 0.632 unit (63.2%) in time T (one time
constant) and approximately 0.98 unit (98%) in 4T (four times the time constant). The final
The time 4T is the basis of demarcation between the transient part and the steady
state part of the time response of a control system. When the actual output reaches within
2% of the desired value it is said that steady state has reached. The time 4T is termed as
The measure of time from the initiation of the input signal upto the settling time of a
control system is the transient period and beyond settling time steady state appears.
Effect of time constant on the speed of the time response of a control system:
A decrease in the value of time constant of a control system indicates that the steady state
reaches earlier. Therefore, lower the time constant faster is the time response of a control
Fig. 6.3.4 T im
6.3.3. Time response of a first order control system subjected to unit ramp input
function
The output for the system is expressed as
C(s) = R(s) ...(6.7)
U6 +
As the input is a unit ramp function
r(t) = t and R(s) =
U
C(s) = ...(6.8)
U
U6 +
Breaking R.H.S. of Eq. (6.8) into partial fractions
Ō U6 6
C(s) = +
U
U6 +
C(s) =
Ō 6 +6 ...(6.9)
U U
U +
6
F I
L – 1
C(s) = L – 1
GG
Ō 6
+6
JJ
GH U U
U +
6
JK ess = T
∴ c(t) = (t – T + Te
–t/T
) ...(6.10) c(t)
T
M G
TI GIN
– t/T
The error is given as e(t) = r(t) – c(t) = t – (t – T + Te )
E
G
LA
– t/T
= (T – Te ) ...(6.11)
– t/T
The steady state error is e
ss
= NKO (T – Te ) = T 0
V →∞ T 2T 3T t
The time response in relation to Eq. (6.10) is shown in Fig. 6.3.4. T ime response of first
velocity matches with the input velocity but lags behind the input by time T and a positional
error of T unit exists in the system. It is also observed that lower the time constant lesser is
6.3.4. Time response of a first order control system subjected to unit impulse input
function
The output for the system is expressed as
C(s) = R(s) ...(6.12)
U6 +
As the input to the system is a unit impulse function, its Laplace transform is 1, i.e.
C(s) = 1 ...(6.12 a)
U6 +
Taking inverse Laplace transform on both sides of Eq. (6.12 a)
L – 1
C(s) = L– 1
U6 +
or L – 1
C(s) = L– 1
6
U + 1
6 T
c(t)
∴ c(t) = e
– t/T
...(6.13)
6
corresponding time response of a first order control system order control system subjected to
subjected to these input functions is given in table below : unit impulse input function.
– t/T
Unit ramp c(t) = t – T + Te Differentiate
U
– t/T
Unit step c(t) = 1 – e
U
– t/T
Unit impulse 1 c(t) = e Integrate
6
It is observed that step function is first derivative of a ramp function and impulse
function is first derivative of a step function. From the derived time response expression it is
concluded that the output time response also follows the same sequence as that of input
functions.
134 LINEAR CONTROL SYSTEMS
A second order control system is one wherein the highest power of s in the denominator of its
A general expression for the transfer function of a second order control system is given
by
%
U ω
= P
...(6.14)
4
U U
+ ζω P
U +ω
6.4.1. Time response of a second order control system subjected to unit step input
function
The output for the system is given by
ω
P
C(s) = R(s) ...(6.15)
U
+ ζω P
U +ω
r(t) = 1 and R(s) =
U
ω
P
C(s) = ...(6.16)
U U
+ ζω P
U +ω
P
) as [(s + ζω )
n
2
+ ω
P
(1 – ζ 2
)] and
U + ζω P
C(s) = Ō ...(6.17)
U =
U + ζω P
+ ω
Ō ζ
P
?
put ω d
= ω
P
Ō ζ
+ ζω
U
P
C(s) = Ō . ...(6.18)
U + ζω + ω
U
P F
U + ζω ζω ω
P P F
C(s) = Ō Ō ...(6.18 a)
U
U + ζω + ω P
F
ω F
U + ζω + ω
P
F
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 135
Taking inverse Laplace transform on both sides of Eq. (6.18 a)
+ ζω U ζω ω
L – 1
C(s) = L – 1
Ō
P
Ō
P
F
...(6.19)
U + ζω + ω ω
U + ζω + ω
U
P F F P F
ζω V
ζω ζω V
∴ c(t) = 1 – G
Ō P
EQU ω F
V Ō
P
G
Ō P
UKP ω F
V ...(6.20)
ω F
Since ω d
= ω P
Ō ζ
Eq. (6.20) is rewritten as below
Ō ζω P
V
G
c(t) = 1 – [
Ō ζ
cos ω d
t + ζ sin ω d
t] ...(6.21)
Ō ζ
put sin φ= Ō ζ
, ∴ cos φ=ζ
Ō ζω P
V
G
∴ c(t) = 1 – (sin φ cos ω d
t + cos φ sin ω d
t)
Ō ζ
Ō ζω P
V
G
or c(t) = 1 – sin (ω d
t + φ) ...(6.22)
Ō ζ
Ō ζ
Since ω d
= ω n
Ō ζ
and φ = tan – 1
ζ
ζω
G
Ō
P
V
Ō
Ō ζ
c(t) = 1 – UKP
ωP
Ō ζ V + VCP
...(6.23)
Ō ζ ζ
ζω
G
Ō
P
V
Ō
Ō ζ
or e(t) = UKP ω P
Ō ζ V + VCP ...(6.24)
Ō ζ ζ
ζω
G
Ō
P
V
Ō
Ō ζ
The steady state error e
ss
= NKO UKP ω P
Ō ζ V + VCP
V →∞
Ō ζ ζ
The time response expression given by Eq. (6.23) indicates that for values of ζ < 1 the
response presents exponentially decaying oscillations having a frequency ω n
Ō ζ
and
The term ω n
is called natural frequency of oscillations. The term ω d
= ωn
Ō ζ
is
called damped frequency of oscillations. The term ζ affects the damping and called damping
136 LINEAR CONTROL SYSTEMS
The time response of a second order control system given by Eq. (6.23) is influenced by
its damping ratio ( ζ). The cases for the values of damping ratio as (a) ζ < 1 (b) ζ = 0 (c) ζ = 1
(d) ζ > 1 are considered below :
(a) The time response and the error in relation to Eqs. (6.23) and (6.24) for ζ < 1 are
1
T=
zwn
0.02 unit
c(t)
1 unit
0 t
(a)
e(t)
0
t
(b)
Fig. 6.4.2. T ime response and error of a second order control system
The response settles within 2% of the desired value (1 unit) after damping out the
(b) An expression for the time response of a second order system having ζ = 0 when
or c(t) = 1 – sin ( ω n
t + tan
– 1
∞)
π
or c(t) = 1 – sin
ω P
V +
or c(t) = (1 – cos ω n
t) ...(6.25)
The time response in relation to Eq. (6.25) is plotted in Fig. 6.4.3 which indicates
(c) An expression for the time response of a second order control system having ζ= 1
c(t)
1 unit
o t
Fig. 6.4.3. T ime response of a second order control system ( ζ = 0, undamped case)
subjected to unit step input functions.
Taking limits of Eq. (6.23) as the value of ζ approaches 1, the equation derived below
is obtained
ζω
G
Ō
P
V
Ō
Ō ζ
c(t) = NKO Ō UKP ω P
Ō ζ V + VCP
ζ →
Ō ζ ζ
Ō ζ
Ō ζω P
V
G
∴ c(t) = NKO Ō UKP
ω P
Ō ζ V + φ
ζ →
Ō ζ
Ō ζω P
V
G
or c(t) = NKO Ō =UKP
ω P
Ō ζ
V cos φ+ cos ( ω n
Ō ζ
t) sin φ]}
ζ →
Ō ζ
Ō ζ
Ō ζω V
P
G
c(t) = NKO Ō =UKP
ω P
Ō ζ V ζ + EQU
ω P
Ō ζ V Ō ζ ...(6.26)
ζ →
Ō ζ
Ō ζω P
V
G
c(t) = NKO Ō
ω P
Ō ζ V + Ō ζ
ζ →
Ō ζ
Ō ζω P
V
G
= NKO Ō Ō ζ
ω P
V +
ζ →
Ō ζ
ω
or c(t) = [1 – G
Ō
P
V
(1 + ω n
t)] ...(6.29)
138 LINEAR CONTROL SYSTEMS
The time response in relation to Eq. (6.29) is plotted in Fig. 6.4.4. The response is
(d) An expression for the time response of a second order control system having ζ>1
when subjected to unit step input function is derived hereunder:
ω
P
C(s) = R(s) ...(6.30)
U
+ ζω P
U +ω
As the input is a unit step function r(t) = 1 and R(s) = 1/s, therefore, substituting in
Eq. (6.30)
ω
P
C(s) = ...(6.31)
U U
+ ζω P
U +ω
ω
P
C(s) =
U
U + ζω P
Ō ω
ζ
P
Ō
ω
P
or C(s) = ...(6.32)
U =U +
ζ + ζ
Ō ω P
? =U +
ζ Ō ζ
Ō ω P
?
C(s) = Ō
U ζ
Ō
ζ Ō ζ
Ō = U +
ζ Ō ζ
Ō ω P
?
+ ...(6.33)
ζ
Ō
ζ+ ζ
Ō = U +
ζ + ζ
Ō ω P
?
Ō
ζŌ ζ Ō ω P
V Ō
ζ+ ζ Ō ω P
V
G G
c(t) = 1 – + ...(6.34)
ζ Ō
ζ Ō ζ Ō ζ Ō
ζ+ ζ Ō
The time response in relation to Eq. (6.34) is plotted in Fig. 6.4.5. The response is
called overdamped.
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 139
Eq. (6.34) indicates that the time response has two time constants one slower [T = 1/
1
( ζ– ζ
Ō ) ωn
] and the other faster [T
2
= 1/( ζ+ ζ
Ō ) ω n
]. For values of ζ comparatively
greater than 1, the effect of the faster time constant on the time response can be neglected
T = .
ζ Ō ζ
Ō ω P
In relation to Eqs. (6.23), (6.25) and Eq. (6.34), the time response of a second order
control system subjected to unit step input function is plotted in Fig. 6.4.6.
0.5
0.8
z=
z=
0.02 Unit
c(t)
1
z=
0.98 Unit 1 Unit
2
z=
0 t
Fig. 6.4.6. Comparison of unit step input time response of a
ζω
G
Ō
P
V
Ō
Ō ζ
E
V =Ō UKP
ω P
Ō ζ V + VCP
Ō ζ
ζ
response of the system. When ζ = 1 the actual damping equals ω . This actual damping when
n
ζ = 1 is known as critical damping. At the value of critical damping the oscillations in the
The ratio of the actual damping to critical damping is known as the damping ratio as
#EVWCN FCORKPI ζω
= =ζ P
%TKVKECN FCORKPI ω P
system is given by
%
U ω
= P
...(6.35 a)
4
U U
+ ζω P
U +ω
The denominator of the above expression is equated to zero and following equation is
formed
s
2
+ 2 ζω s +
n
ω
P
= 0 ...(6.36)
s , s
1 2
= – ζω ± jω n n
Ō ζ
The roots s , s are also the poles of transfer function expressed by relation (6.35 a). A
1 2
study of roots s , s gives a prediction about the nature of time response. The real part of the
1 2
roots represents the damping and the imaginary part represents damped frequency of
s
2
+ 2 ζω n
s + ω
P
= 0
The location of roots of the characteristic equation for various values of ζ (keeping ω n
fixed) and the corresponding time response for a second order control system is shown in
Fig. 6.4.7.
c(t) z= + Imj
0
c(t)
t
z<
t 1
w
wn 1 – z2
co
n
s
–1
c(t) z=
1
z
=
q
t – Re z>1 zwn + Re
c(t)
t
– Imj
Fig. 6.4.7. Location of roots of the characteristic equation and corresponding time response.
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 141
From Fig. 6.4.7 it is inferred that the change over from underdamped to overdamped
response takes place at ζ = 1. The value of ζ from the location of roots is calculated as
ζ = cos θ.
6.4.4. Transient Response Specifications of Second Order Control System
The time response of an underdamped control system exhibits damped oscillations prior to
reaching steady state. The specifications pertaining to time response during transient part
Max. over-shoot 1
T=
ζωn
2%
c(t)
100%
0 tr tp ts t
The rise time is the time needed for the response to reach from 10% to 90% or 0 to
100% of the desired value of the output at the very first instant. Usually 0-100% basis is used
The rise time (0 to 100%) for underdamped second order control system is calculated
below. As expression for the time response of a second order control system is given by
Ō ζω P
V
G
c(t) = 1 – UKP =
ω P
Ō ζ V + φ?
Ō ζ
At the first instant when time response reaches 100% of the desired value, i.e. c(t) = 1,
Ō ζω P
V
T
G
1 = 1 – UKP =
ω P
Ō ζ V
T
+ φ?
Ō ζ
Ō ζω P
V
T
G
or UKP =
ω P
Ō ζ V
T
+ φ? =
Ō ζ
Ō ζω P
V
T
G
As is finite
Ō ζ
∴ sin =
ω P
Ō ζ V
T
+ φ? = 0
=
ω P
Ō ζ V
T
+ φ? = π
142 LINEAR CONTROL SYSTEMS
πŌφ
∴ t
r
= ...(6.37)
ω P
Ō ζ
Ō ζ
where φ = tan – 1
ζ
response. At some instances of time the output exceeds its desired value while at some other
instances the output is lower than its desired value. The maximum positive deviation of the
output with respect to its desired value is known as maximum overshoot and denoted as M .
p
M = c(t) – 1
p max
E
V Ō
OCZ
% M = × 100
p
The time needed to reach the maximum overshoot is called peak time and denoted as
tp. It is observed that Mp occurs when the slope of the time response curve after the initiation
of the input signal is zero. Therefore, if the time response expression is differentiated and
equated to zero, the peak time tp can be determined and by substituting the value of tp in the
Ō ζω P
V
G
c(t) = 1 – UKP =
ω P
Ō ζ V + φ?
Ō ζ
Ō ζω P
V
FE
V G
∴ = Ō ω P
Ō ζ EQU =
ω P
Ō ζ V + φ?
FV
Ō ζ
Ō ζω
Ō ζω G
P
V
–
P
UKP =
ω P
Ō ζ V + φ?
Ō ζ
FE
V
Put = 0
FV
Ō ζω P
V
G
∴ ]Ō ω P
Ō ζ
EQU =
ω P
Ō ζ
V + φ? + ζ ω P
UKP =
ω P
Ō ζ
V + φ?_ = 0
Ō ζ
Ō ζω P
V
G
In the above equation, since is finite
Ō ζ
∴ ω n
Ō ζ
. cos =
ω P
Ō ζ
V + φ? = ζω n
sin =
ωP
Ō ζ
V + φ?
Ō ζ
∴ tan =
ω P
Ō ζ
V + φ? =
ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 143
Ō ζ
since, φ = tan –1
, substituting in above equation
ζ
Ō
Ō ζ Ō ζ
tan
ω P
Ō ζ V + VCP = ...(6.38)
ζ ζ
ω P
Ō ζ t = n π ...(6.39)
where n = 0, 1, 2...
(6.39), hence,
π
t = ...(6.40)
p
ω P
Ō ζ
c(t) is determined by putting t = t in the time response expression, therefore,
max p
Ō ζω P
V
R
G
c(t)
max
= 1 – . sin =
ω P
Ō ζ V
R
+ φ?
Ō ζ
π
Ō ζω P
ω Ō ζ
G
P
π
= 1 – UKP
ω P
Ō ζ + φ
Ō ζ
ω P
Ō ζ
Ō ζπ Ō ζπ
Ō ζ Ō ζ
G G
= 1 – . sin ( π + φ) = 1 – . (– sin φ)
Ō ζ Ō ζ
Ō ζ
∵ φ = tan –1
, therefore, sin φ= Ō ζ
hence,
ζ
Ō ζπ
Ō ζ
G
c(t)
max
= 1 + Ō ζ
Ō ζ
ζπ
Ō
Ō ζ
or c(t) = 1 + G ...(6.41)
max
M = c(t) – 1
p max
ζπ
Ō
Ō ζ
∴ M
p
=
+ G
– 1
Ō ζπ
Ō ζ
or M = G ...(6.42)
p
ζπ
Ō
Ō ζ
%M = G × 100 ...(6.43)
p
144 LINEAR CONTROL SYSTEMS
n = 2 in Eq. 6.39.
100%
A graph relating M
p
and ζ is plotted in
Fig. 6.4.9.
% Mp
magnitude of the oscillations present in the
t = 4 ...(6.44)
s
ζω P
Note: On 5% basis the settling time for a second order control system is approximately three
t = 3
s
ζω P
6.4.5. Time response of a second order control system subjected to unit ramp input
function
As the input is a unit ramp function
r(t) = t and R(s) =
U
ω
P
C(s) = ...(6.45)
U
U
+ ζω P
U +ω
P
) as (s + ζω n
)
2
+ ω
P
(1 – ζ2
) and
+ ζω
U
P
C(s) = Ō
U =
U + ζω + ω ζ
U
Ō ?
P P
put ω
d
= ω n
Ō ζ
U + ζω
∴ C(s) = Ō
P
U
U =
U + ζω + ωP
F
?
ζω
P
or C(s) = Ō Ō
U =
U + ζω + ω + ζω +ω
U ? =
U ?
P F P F
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 145
ζω P
ω
P
or C(s) = Ō Ō
U
ω
P
U =
U + ζω P
+ω
F
? =
U + ζω P
+ω
F
?
ζ + ζ ω
U
P
or C(s) = Ō Ō Ō
U
ω UP
U + ζ ω + ω
P
F
U + ζω P
+ω
ζ ζ + ζω
U
ζ
Ō
or C(s) = Ō Ō + P
U
ω P
U ω P
U + ζ ω + ω =
U + ζω
P
F P
+ω
F
?
ζ ζ + ζω
U
ζ Ō
ω
or C(s) = Ō +
P
+ F
U
ω P
U ω P
=
U + ζ ω +ω P
F
? ω F
=
U + ζω +ω P
F
?
...(6.46)
ζ ζ ζω V
ζ Ō
ζω
c(t) = t – + G
Ō
P
. cos ω t + G
Ō
P
V
. sin ω t
ω ω
P P
d
ω F
d
ζ ζ Ō ζω
ζ Ō Ō ζω
or c(t) = t – + G
P
V
=EQU Ō ζ V? + G
P
V
UKP =
ω
P
Ō ζ V?
ω ω
P P ω Ō ζ
Ō ζω
ζ G
P
V
or c(t) = t – + = ζ Ō ζ EQU ωP
Ō ζ V
ω Pω Ō ζ
+ (2 ζ 2
– 1) sin
ω P
Ō ζ
V? ...(6.47)
put 2 ζ Ō ζ
= sin φ,
∴ (2 ζ 2
– 1) = cos φ, hence
Ō ζω
ζ G
P
V
Ō ζω
ζ G
P
V
or c(t) = t – + UKP =
ω P
Ō ζ V + φ? ...(6.48)
ω Pω Ō ζ
ζ
Ō ζ
where φ = tan –1
ζ
Ō
Ō ζ
It can also be shown that φ = 2 tan –1
ζ
Ō ζω
ζ G
P
V
∴ e(t) = t – V Ō + UKP =
ω P
Ō ζ V + φ?
ω Pω Ō ζ
P
146 LINEAR CONTROL SYSTEMS
Ō ζω
ζ
G
P
V
= Ō UKP =
ω P
Ō ζ V + φ? ...(6.49)
ω P ω Ō ζ
As t →∞ the second term in Eq. (6.49) approaches zero, therefore, the steady state
error is given by
ζ
e = ...(6.50)
ss
ω P
2 ζ/ω n
unit and the output response lags by a
time of 2 ζ /ω .
n
increasing ω .
n
After the disappearance of Fig. 6.4.10. The response of a second order control
6.4.6. Time response of a second order control system subjected to impulse input
function
As the input is a unit impulse function, therefore, R(s) = 1 and the output of a second
ω
P
C(s) = ...(6.51)
U
+ ζ ω P
U +ω
P
) as [(s + ζωn
)
2
+ ω
P
(1 – ζ 2
)]
ω
P
C(s) =
=
U + ζω P
+ω
P
Ō ζ
?
ω
ω P
Ō ζ
or C(s) =
P
...(6.52)
ω P
Ō ζ
=
U + ζω P
+
ω P
Ō ζ
?
ω Ō ζω V
c(t) =
P
G
P
UKP =
ω P
Ō ζ V? ...(6.53)
Ō ζ
From Eq. (6.53) the time response for the following three values of ζ is determined and
plotted in Fig. 6.4.11.
(a) ζ < 1
ω Ō ζω V
c(t) =
P
G
P
UKP =
ω P
Ō ζ V ? ...(6.54)
Ō ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 147
ζ<1
ζ=1
ζ>1
c(t)
Unit
0 Impulse
t
0 t
(a) (b)
Fig. 6.4.11. T ime response of a second control system subjected to unit impulse input function.
The time response given by above equation presents decaying exponential oscillations
(b) ζ = 1
ω Ō ζω V
c (t) = NKO
P
G
P
UKP =
ω P
Ō ζ V?
ζ →
Ō ζ
ω Ō ζω V
=
P
G
P
ω
P
Ō ζ V
Ō ζ
ω
∴ c(t) = t ω P
G
Ō
P
V
...(6.55)
(c) ζ>1
ω Ō ζω V
c(t) =
P
G
P
UKP =
L ω P
ζ Ō V ?
L ζ Ō
ω Ō ζω V
=
P
G
P
L UKPJ =
ω P
ζ Ō V ?
L ζ Ō
+ω
ζ Ō V Ō
ω
ζ Ō V
ω P Ō ζω P
V
G
P
Ō G
P
= G
ζ Ō
ω P Ō
ζ Ō
ζ
Ō ω P
V Ō
ζ+
ζ
Ō ω P
V
or c(t) = =G Ō G ? (6.56)
ζ Ō
A third order control system is one wherein the highest power of s in the denominator of its
transfer function equals 3. Thus a third order control system is expressed by a transfer
%
U -
= ...(6.57)
4
U U
+ CU
+ DU + E
3 2 2
(s + as + bs + c) = (s + d) (s + b s + c )
1 1
The constant d, b and c can be determined in terms of a, b and c. Thus the transfer
1 1
148 LINEAR CONTROL SYSTEMS
function of a third order control system given by Eq. (6.57) can also be expressed as
%
U -
= ...(6.58)
4
U
U + F
U
+DU+
E
The denominator of expression Eq. (6.58) is composed of product of a first order term
The unit step time response of a third order control system given by expression
-
C(s) = ...(6.59)
U
U + F
U
+DU+
E
The R.H.S. of equation can be expanded using partial fraction technique and the
- - - U + -
C(s) = + +
U
U + F
U
+ D U
+ E
- - - +U+ -
and c(t) = L –1
+
+
...(6.60)
U
U + F
U
+DU+ E
Aforesaid procedure for obtaining the time response is adopted for the illustrative
Illustrative example : (a) Obtain an expression for unit step time response of a
%
U
=
4
U U
+ U + U +
3 2
The denominator f (s) = (s + 12.3s + 3.74s + 1.68) equals zero at s = –12, therefore,
(s + 12) is its first order factor and its quadratic factor is obtained by dividing f (s) by (s + 12),
2
thus the quadratic factor is (s + 0.3s + 0.14), therefore,
%
U
=
4
U
U +
U + U +
%
U =
U
U +
U + U +
- - - U + -
C(s) = + +
U
U +
U
+ U +
The coefficients can be determined as
–4
K = 1, K = – 9.47 × 10 , K = – 0.99 and K = – 0.31
1 2 3 4
× Ō
U +
∴ C(s) = Ō Ō
U + + U +
U
U
2 2 2
The term (s + 0.3s + 0.14) can be expressed as [(s + 0.15) + (0.34) ] hence,
× Ō
+
U
C(s) = Ō Ō
U + =
U + +
U ?
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 149
× Ō
+
U
= Ō Ō
U + =
U + +
U ?
× Ō
U + +
= Ō Ō
U
U + =
U + +
?
× Ō
U + ×
= Ō Ō Ō
U + + +
+ +
U =
U ? =
U ?
× Ō
U +
= Ō Ō Ō
U
U + =
U +
+
? =
U +
+
?
–4 –12t –0.15t
c(t) = [1 – 9.47 × 10 e – 0.99 e cos (0.34 t)
–0.15t
– 0.47 e sin (0.34 t)] Ans. ...(6.61)
Illustrative Example : (b) Obtain an expression for a unit step time response of a
%
U
=
4
U U
+ U +
U +
3 2
The denominator f (s) = (s + 1.5s + 0.5s + 0.168) equals zero at s = 1.2, therefore,
(s + 1.2) is its first order factor and its quadratic factor is obtained by dividing f (s) by (s +
2
1.2), thus the quadratic factor is (s + 0.3s + 0.14), therefore,
%
U
=
4
U
U +
U + U +
C(s) =
U
U +
U + U +
- - - U + -
C(s) = + +
U
U +
U
+ U +
The coefficient can be determined as
+ U
= Ō Ō
U
U + =
U + +
?
U + +
= Ō Ō
U
U + =
U +
+
?
150 LINEAR CONTROL SYSTEMS
+
U ×
= Ō Ō Ō
U
U + =
U + +
? =
U + +
?
U +
= Ō Ō Ō
U
U + =
U +
+
? =
U +
+
?
...(6.62)
Illustrative Example : (c) Obtain an expression for unit step time response of a
%
U
=
4
U U
+ U +
U +
3 2
The denominator f (s) = (s + 0.42s + 0.17s + 0.0168) equals zero at s = – 0.12,
therefore, (s + 0.12) is its first order factor and its quadratic factor is obtained by dividing
2
f (s) by (s + 0.12), thus the quadratic factor is (s + 0.3s + 0.14),
%
U
∴ =
4
U
U +
U
+ U +
As the input is R(s) = 1/s.
∴ C(s) =
U
U +
U
+ U +
Expanding R.H.S. of above expression into partial fractions
- - - U + -
C(s) =
+
+
U
U +
U
+ U +
2 2 2
The term (s + 0.3s + 0.14) can be expressed as [(s + 0.15) + (0.34) ], hence
U Ō
= Ō +
U
U + =
U +
+
?
U + Ō
= Ō +
U
U + =
U +
+
?
+
U
= Ō + × Ō
U
U + =
U + +
? =
U + +
?
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 151
Taking inverse Laplace transform on both sides of the above equation
...(6.63)
6.5.1. Effect of first order term time constant on time response of third order control
system
The respective first order term time constants for time response expressions (6.61), (6.62)
(a) T = sec (b) T = sec (c) T = sec
a b c
(a), (b) and (c) are s = – 12, s = – 1.2 and s = – 0.12. The time constant and the damped
a b c
frequency of oscillations in the relation to quadratic term are 1/0.15 sec and 0.34 rad/sec
a
c(t)
Sa Sb Sc
– Re – 12 – 1.2 – 0.12
K = 1.68 K = 0.168 K = 0.0168 b 1 unit
c
×
(– 0.15 – j0.34) 0 t
– Imj
Fig. 6.5.1 Poles of illustrative examples Fig. 6.5.2. Time response of illustrative
(a), (b) and (c) plotted in s-plane. examples (a), (b) and (c).
A comparative analysis of expressions (6.61), (6.62) and (6.63) and study of Fig. 6.5.2
reveal that the time response of a third order control system is governed by (i) first order
The first order term time constant has a great influence on the time response of third
order control system. More the value of the time constant in the first order term greater is its
effect and vice-versa. The relative location (in s-plane) of the first order pole with respect to
quadratic (second order) pole decides the nature of the time response of a third order control
system.
If the first order pole is nearer to the imaginary axis in s-plane then the response is
overdamped and if the quadratic (second order) complex poles are nearer to imaginary axis
It is concluded that for a system having many poles, the nearness of the poles towards
imaginary axis in s-plane dominates the nature of the time response, hence the poles (poles
are also roots, of the characteristic equation) of a closed-loop transfer function nearer to
imaginary axis are called dominant poles (or dominant roots of the characteristic equation).
152 LINEAR CONTROL SYSTEMS
%
U -
= ...(6.64)
4
U U
P
+ C U
P Ō
+ C U
P Ō
+ + C
P
In general any higher order polynomial can be factored as a product of first order
%
U -
= ...(6.65)
4
U
U + R
U
+ R
U
+ R
U
+ ζω P
U +ω
where p , p , p
1 2 3
... and (– ζω ±jω n n
Ō ζ
) are the poles. In view of expression (6.65) the
output is given by
-
C(s) = R(s) ...(6.66)
U + R
U
+ R
U
+ ζω P
U +ω
In order to obtain the time response from (6.66) when subjected to a specified input
function, the R.H.S. of (6.66) is expanded using partial fraction technique and inverse
The procedure for obtaining the time response of a third order control system has
magnitude of the output w.r.t. time and such a response is called unstable one. The specified
%
U
=
4
U U
+ U + U +
3 2
The denominator f (s) = s + 1.5s + 0.5s + 3 equals zero at, s = – 2, therefore, (s + 2) is
its first order factor and its quadratic factor is obtained by dividing f (s) by (s + 2), thus the
2
quadratic factor is (s – 0.5s + 1.5), therefore,
%
U
=
4
U
U +
U − U +
%
U =
U
U +
U − U +
- - - U + -
%
U = + +
U
U +
U
Ō U +
The coefficients can be determined as
+
U
= Ō Ō
U
U + =
U Ō +
?
U Ō +
= Ō Ō
U
U + =
U Ō
+
?
U Ō ×
= Ō Ō Ō
U
U + =
U Ō
+
? =
U Ō
+
?
U Ō
= Ō Ō Ō
U
U + =
U Ō
+
? =
U Ō
+
?
– 2t + 0.25t + 0.25t
c(t) = [1 – 0.23e – 0.83e cos (1.2t) – (0.477) e sin (1.2t)] ...(6.67)
The nature of the time response given by Eq. (6.67) and poles are plotted in Fig. 6.6.1.
From the graph (Fig. 6.6.1) it is observed that although the input is finite, the output
tends to infinity as time approaches to infinity and, therefore, the system is called unstable.
The system having order higher than 2 are sometimes liable to be unstable.
c(t)
1 unit
t
(a)
Fig. 6.6.1 T ime response and pole locations of a third order unstable system.
A desirable feature of a control system is the faithful following of its input by the output.
However, if the actual output of a control system during steady state deviates from the
reference input (i.e. desired output), the system is said to possess a steady state error.
As the steady state error is an index of accuracy of a control system, therefore, the
The steady state performance of a control system is assessed by the magnitude of the
steady state error possessed by the system and the system input specified as either step or
ramp or parabolic.
The magnitude of the steady state error in a closed-loop control system depends on its
open-loop transfer function, i.e. G(s) H(s) of the system. The classification of open-loop
transfer function and the feedback path transfer function of a control system is known as
open-loop transfer function. In general the open-loop transfer function of a system is given
by
-
+ U6
C
+ U6
D
G(s) H(s) = ...(6.68)
0
U
+ U6
+ U6
Ō Ō CTG VJG \GTQU
6
C
6
D
...(6.69)
Ō Ō CTG VJG RQNGU
6 6
The number of poles at the origin possessed by an open-loop transfer function is the
H(s)
'
U
= ...(6.70)
4
U + )
U *
U Fig. 6.7.1. Block diagram of a closed-loop
control system.
or E(s) = R(s)
+ )
U *
U
Applying final value theorem, the steady state error can be determined as follows:
e
ss
= NKO U'
U = NKO U4
U ...(6.71)
U → U → + )
U *
U
From equation (6.71) it is noted that the steady state error possessed by a closed-loop
control system depends on the input and the open-loop transfer function.
The actual output of a control system may be in any physical form, it is called as
“position” or displacement. The first derivative of the actual output is called “velocity” and
The table given below explains the meaning of “position”, “velocity” and “acceleration”
Position e i θ
de di dθ
Velocity
dt dt dt
d2e d2i d 2θ
Acceleration
dt 2 dt2 dt 2
put Kp = lim G(s) H(s), Kp is called static positional error coefficient, therefore
s→0
ess = 1 ...(6.73)
1 + Kp
(ii) Static velocity error coefficient :
Static velocity error coefficient is associated with unit ramp input applied to a closed-
loop control system and is determined below :
1
ess = lim R(s)
s→0 1 + G(s) H (s)
As the input is R(s) = 1/s2
ess = lim s 12 1
s→0 s 1 + G ( s) H (s)
= lim 1 1
= lim
s → 0 s + s G(s) H ( s) s → 0 s G( s) H ( s)
put Kv = lim s G(s) H(s), Kv is called static velocity error coefficient, therefore,
s→0
→ +
+
→
=
→ +
→
→
+
+
+
+
→
+
+
→ +
+
∴
∴
+
+
+
+
+
+
→ → +
+
∴
+
+
+
+
→
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 157
+
+
→ +
+
∴
+
+
+
+
→
+
+
→ +
+
∞
+
+
+
+
→
+
+
→ +
+
∴
+
+
+
+
→
+
+
→ +
+
∴
∴
158 LINEAR CONTROL SYSTEMS
+
+
+
+
→
+
+
→ +
+
∴ ∞
+
+
+
+
→
+
+
→ +
+
∴ ∞
+
+
+
+
→
+
+
→ +
+
∞ ∞
+
∞ ∞
∞ ∞
→ ∞
=
+
=
+
′
″
+
160 LINEAR CONTROL SYSTEMS
″
″
+
+
∴
+
τ τ
− τ
τ
∫ ∞
τ
τ L
τ τ
τ
τ τ
τ τ ′
″ ′″
τ
τ τ
τ
− τ′
+
″
−
′″
+
τ
∫
→ ∞
τ τ
τ
− τ ′
+ ″
− ′″
+
τ
→∞
→∞ ∫
∞ ∞ τ
τ τ
τ +
τ
∫
τ τ− ∫
′
∫
″
τ
∞ ∞ ∞ τ
τ − ′
τ τ
τ + ″
τ
∫
τ ∫
∫
τ
∞
∫
τ τ
∞
∫ τ τ
τ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 161
∞ τ
∫
τ τ
M
∞ τ
∫
τ τ
∴
′ ″ ′″
∞
− τ
∫
τ τ
∞ ∞
− τ
τ=
→
→ ∫
τ τ= ∫
τ
∴
→
∞
∫ τ τ
−
τ
τ
∞
τ τ
τ =
→
∫
∴
→
→
M
→
+
+ + +
≤
+ + +
162 LINEAR CONTROL SYSTEMS
∴
′ ″
∞
¡
∫
∞
¡
∫
∞
¡ ¢
¢
∫
∞
¡ ¢
¢
∫
→∞
∞
¡
£
ω
+ ζω +ω
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 163
ζ ¤
¡
ω
∴
+ ζω +ω
ω
∴
−
+ ζω +ω
+ ζω
+ ζω
+ω
+ ζω
L
+ ζω
+ω
+ ζω
ζω
L +
+ ζω
+ω
−ζ
+ ζω
+ω
−ζ
+ ζω ζ ω −ζ
L +
+ ζω
+ω −ζ
+ ζω
+ω −ζ
−ζ
ζ
−ζω
∴ ω
−ζ + ω
−ζ
−ζ
−ζω
ζ
∴ ω
−ζ + ω
−ζ
−ζ
ζ
ω −ζ + ω − ζ
−ζ
+ ω −ζ ζ ω −ζ ζ − ω −ζ
− ζω
+ +
−ζ −ζ
− ζ
ζ
− ζω
− ζ
+ ω
−ζ + ω
− ζ
−ζ
−ζ
− ζ
+ ζω
+
− ζ
+ ζω
−ζ
+ ζω
+ ω
−ζ
ζ ω −ζ
+
+ ζω
+ ζω − ζ
−ζ
164 LINEAR CONTROL SYSTEMS
∞
¡
∫
¡
∞
∫
=
→∞ ∫
L
→∞ ∫
=
→ ∫
=
→ →
∞
¥ ¡
∫
∴
¡
→
Q
− ζ
+ ζω
∴ ¡ +
→ − ζ
+ ζω
−ζ
+ ζω
+ ω
−ζ
ζ ω −ζ
+
+ ζω
+ ω − ζ
−ζ
− ζ
ζω ζ ω −ζ
+ +
−ζ
ζω
−ζ
ω
ω
−ζ
ζ − ζ
ζ
+ +
ω ζ − ζ
−ζ
+ ζ − ζ ζ +
− ζ
ζ +
ζ + ζ
ζ
ω ζ − ζ
ω ζ − ζ
ω
ω
∴ ω ¡
ζ +
ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 165
¡
ω
ζ
ISE
2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
ζ
−ζω ω
−ζ + ω
−ζ
−ζ
∞
¡
∫
L
−
166 LINEAR CONTROL SYSTEMS
− ζ
ζ
− ζω
+ ω −ζ + ω −ζ
− ζ
−ζ
−ζ
− ζ
ζ
− ζω
∴ + ω
−ζ + ω
−ζ
− ζ
−ζ
−ζ
− ζ
+ ζω
+
− ζ
+ ζω
−ζ
+ ζω
+ ω
−ζ
ζ ω −ζ
+
+ ζω
+ ω − ζ
−ζ
− − ζ
+ ζω
− ω −ζ
∴ −
− ζ
+ ζω
− ζ
¦ + ζω
+ ω − ζ
§
ζ + ζω
−
¦ + ζω
+ ω −ζ
§
− ζ
+ ζω
− ω −ζ
∴ +
−ζ
+ ζω
−ζ
¦ + ζω
+ ω −ζ
§
ζ + ζω
+
¦ + ζω
+ ω
−ζ
§
∞
¡
∫
=
→
−
− ζ
+ ζω
− ω −ζ
∴ ¡ +
→ −ζ
+ ζω
−ζ
¦ + ζω
+ ω −ζ
§
ζ + ζ ω
+
¦ + ζω
+ ω
−ζ
§
− ζ
ζ −
ζ
+ +
ζ ω −ζ
−ζ
ω ω
− ζ
ζ −
ζ
+ + ζ +
ω ζ
−ζ
−ζ
ω ζ
¡ ζ +
ω ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 167
¡
ω
ζ
Fig. 6.7.3.
ζ
ζ
¡
∴
ζ + =
ζ −
ζ
ζ ω ζ ζ
ω
∴ ζ −
ζ
ζ
ω ζ ζ
ω
ζ + +
ζ ×
6.8 SENSITIVITY
∆
∆
Q
∴
∆
¦∆
§
∆
¦∆
§
R(s) + C(s)
G(s)
Input Output
Input Output –
R(s) C(s)
G(s)
H(s)
¡
∆
=
+
+ ∆
∆
+ ¦
+ ∆
§
+ ∆
∆
+
+ ¦ ∆
§
∴
∆
≅
+
∆
∆
≅
+
+
+
Q
+
∆
∴
∆
≅
+
+
∆
∆
≅
+
¡
∆
¢
¢¨¨
©
©
∂
=
∂
6.8.3. Sensitivity of overall transfer function M(s) with respect to forward path
transfer function G(s)
∂
=
∂
∂
=
∂
Open-loop control system.
∂
∂
∂
∂
Closed-loop control system.
+
∂
¦ +
§ − ¦
§
=
∂
¦ +
§
∂
=
∂
¦ +
§
∂
= =
∂
¦ +
§
+
=
+
6.8.4. Sensitivity of overall transfer function M(s) with respect to feedback path
transfer function
∂
=
∂
∂
=
∂
+
∂
¦
§
=
∂
¦ +
§
=
+
ω
R(s) + 25 C(s)
s(s + 2)
–
0.25
Solution.
+
= = =
+
+ +
+ ×
+
ω
−ω + ω
∴ =
−ω + ω +
ω
− + × ¢ − + ¢
∴ = = Ans.
+
− + × +
172 LINEAR CONTROL SYSTEMS
×
+
= = =
+
+ +
+ ×
+
ω
ω
∴ =−
−ω + ω +
∴ =− = Ans.
− + × × +
£
+
R(s) + K C(s)
1 + st
–
+
+ +
+
+
+
+ +
+
+
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 173
The time constant in the case of open-loop system is T and that in closed-loop case is
T/(1 + Kb). Since K > 1 and b being positive T/(1 + Kb) < T, indicating improvement in time
An automatic control system is used to maintain its output within desirable limits by means
of a control action. Any deviation of the output from the reference input is detected by an
error detector. The error thus detected is used as an actuating signal for control action
through a controller. The proportional control action and types of control actions used for
improving transient and steady state response of a control system are described hereunder.
proportional to the error signal. The error signal being the difference between the reference
input signal and the feedback signal obtained from the output.
For the system considered as shown in Fig. 6.9.1. The actuating signal is proportional
to the error signal, therefore, the system is called proportional control system.
It is desirable that the control system be underdamped for the point of view of quick
increasing forward path gain of the system. The increase in forward path gain reduces the
steady state error, but at the same time maximum overshoot is increased. For satisfactory
Without sacrificing the steady state accuracy, the maximum overshoot can be reduced
added with derivative of the error signal. Therefore, the actuating signal for derivative con-
FG
V
e (t) = e(t) + 6F ...(6.131)
a
FV
where T is a constant.
d
174 LINEAR CONTROL SYSTEMS
sTd
+
R(s) + E(s) + wn2 C(s)
1
– Ea(s) s(s + 2zwn)
ω
+
+ ζω
ω
+ +
+ ζω
+ ω
+ ζ ω + ω +ω
ζω ω ω
ζω + ω
ζ′
ω
ω
ζ′ ζ +
+
ω
+ ζ′ω + ω
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 175
ω
+ ζω +ω
ω
+ ζω
+
+ ω
+
+ ζω
+ ζω
+ ζω + ω
+ω
+ ζω
∴
+ ζω + ω
+ω
+ ζω
∴
→ + ζω + ω +ω
ζ
ω
ω
R(s) + 16 C(s)
2
s + 1.6s
–
Fig. 6.9.3.
Solution.
+
+ +
+
+
ζω ω
ζω
ζ =
ω ×
ω
ζ′ ζ +
ω
∴ ∴ Ans.
R(s) + 16 C(s)
1 + 0.3s
s2 + 1.6s
–
Fig. 6.9.4.
−ζ −
π −
− π −
−
ζ
=
ω −ζ −
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 177
π −
−
Ans.
×
π π π
= =
Ans.
×
ω −ζ −
ζπ π
− −
−
× = ×
−ζ
Ans.
+
+ +
+
∴ = +
+ + + + + +
+
+ +
+ + + +
+
∴ − +
+ + + +
+
− +
+ + − + + −
+
− +
+ + + +
+
− +
+ + + +
+
− + ⋅
+ + + +
+
− + ⋅
+ + + +
\ − −
+
−
Ans.
\
\
−
− π −
−
=
Ans.
\
ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 179
Ki
s
+ Ea(s)
R(s) E(s) n2 C(s)
1
+ + s(s + 2n)
–
Ê ˆ ω
Á + ˜
Ë ¯ + ζ ω
Ê ˆ ω
+ Á + ˜ ◊
Ë ¯ + ζ ω
+ ω
+ ζω + ω
+ ω
ζω ω ω
¡
ζ ω ¢
¡ ζω £
+ ω
+ ζω
180 LINEAR CONTROL SYSTEMS
+
+ ω
+ ⋅
+ ζ ω
+ ζω
+ ζω + ω + ω
ζω
¡
+ ζω
+ ζω + ω + ω
3
+ ζω
\
⋅
+ ζω + ω + ω
+ ζω
⋅ ⋅
Æ Æ + ζω + ω + ω
ζ
ω
ζ
¤
ω
ζ
¤
∞
ω
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 181
6.9.4. Proportional Plus Derivative Plus Integral Control (PID Control)
∫
+ +
ω
+ ζω + ω
ω
+ ⋅
+ ζω + ω
ω
+ ζω + ω + ω
182 LINEAR CONTROL SYSTEMS
sKt
R(s) n2Ea(s) C(s)
+ s2(s + 2n + 2n Kt)s
–
ζω ω ω
ζω + ω
ζ′
ω
ω
ζ′ ζ +
⋅
ω
+ ζ ω + ω
+
ω
+ ⋅
+ ζω + ω
+ ζω + ω
+ ζω + ω + ω
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 183
+ ζω + ω
\ ⋅
+ ζω + ω + ω
Æ
+ ζω + ω
⋅ ⋅
Æ + ζω + ω + ω
ζ
+
ω
Solution.
ζω ω
ζ ω
ζ =
ω ◊
ω
ζ′ ζ +
ω
\ Ans.
184 LINEAR CONTROL SYSTEMS
−ζ −
π −
− ⋅ π −
−
ζ
=
ω −ζ −
π −
−
Ans.
◊
π π π
= =
Ans.
−
ω −ζ ◊
ζπ π
− −
−
−ζ
◊ = ◊
Ans.
ζ ◊
= Ans.
ω
ω
+ ζω + ω + ω
ζ ω
\
+ ◊ ◊ + ◊ +
+ +
ζ′ ω ω
\ ζ′ =
ω ◊
− ζ′ −
π −
− π −
−
ζ′
=
ω − ζ′ −
π −
− π −
=
Ans.
π π
=
Ans.
ω − ζ′ −
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 185
ζ′ π π
− −
− π
Ans.
− ζ′
◊ =
ζ ◊
+
= Ans.
ω
ζ
+
s + 1.2
+
s + 0.8
Fig. 6.10.1.
+ +
R(s) + 4 C(s)
+ ⋅ + [s(s + 4) + 4]
+ +
Ans. s+2
+ +
\ ⋅
+ +
+ +
¡
= + +
+ + + +
\ ⋅ − + ⋅
+ +
L L − +
+ +
\
Ans.
+
+ +
= = =
+ + +
+ + +
+
=
+
+
\
+
+
= +
+ + +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 187
∴
+ +
+
∴
+
+
= + +
+ + +
∴
+
+ +
+
+
= =
+ +
+
+
∴
+ +
∴
+ +
= + +
+ + + +
188 LINEAR CONTROL SYSTEMS
∴
+
+ +
+ C(s)
R(s) + + 1
4
+ s(s + 2)
– –
0.5
+
+ + +
= =
+
+ +
+
+ +
R(s) + + 1 C(s)
s+4
s(s + 2)
– –
0.5
R(s) + (s + 4) C(s)
(s2 + 2.5 + 2)
–
+
∴
+ +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 189
+
∴
+ +
+ +
= +
∴
+ + + +
∴
+ + + +
+ + +
+ + +
+ + + +
+
+ + + +
+
+
+ + + +
+ +
+
+ + + +
+
+ + × + +
R(s) + 20 C(s)
(s + 1) (s + 5)
+ +
= = =
+ +
+ +
+
+ +
ζω
∴ ζ =
ω ×
ω ω
ζ =
π π
=
ω (
ζ
)
ζπ π
× =
ζ
π π
=
ω
ζ
π π
=
ω
ζω
ω
× = ×
π ζω π ×
θ
r(s) + 1 o(s)
KT
Js2 + fs
–
θ +
= =
θ
+
+ +
+
θ
=
θ + +
θ
=
θ + +
∴ θ θ
+ +
π π
θ =
π
∴ θ
π
∴ θ
+ +
ζ ω
∴ ζ =
ω ×
ζ ω
π
ζ
θ
ω
ζ +
ζ ζ
×
π
θ
+
π
θ
+
+
= =
+ + +
+
+
∴ ω
ζ ω
∴
ζ = =
ω
ζ π
ζ π
− −
×
−ζ
−ζ
∴
ζ π ζ π
− −
−ζ −ζ
∴
−ζ
ζ
π
ζ
ζ ζ
∴
ζ
ζ
ζ π
ζ π
− −
×
−ζ
×
−ζ
∴
ζ
ζ
ζ ζ
ζ
= ×
ζ
ζ
= = =
ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 193
ω ω
ω
ω
+
× ×
ω +
ω
+ × ×
+
ω
ω
∴ ω
+
ω +
π
ω × π
∴ π
ω
∴ π
ω π +
+
+
∴ π
ω
− +
194 LINEAR CONTROL SYSTEMS
ω π
ω π
ω ω = π
→ → +
π ×
π
ω +
=
ω +
+ × ×
+
+
∴ ω
ω
+
ω π
+
∴ π
ω
+
+ π ×
π
ω ω = π = =
→ → +
π
×
π
θ
=
+ +
θ
θ
+ +
θ θ
→
∴ θ
→ + +
∴ θ
→ + +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 195
∴
θ
θ
ζπ
−
×
−ζ
ζπ ζπ
−
×
−ζ
−ζ
∴
∴
ζπ
−ζ
ζπ
×
−ζ
−ζ π
∴ ∴
ζ
ζ
π
ω −ζ
π
∴ ∴ ω
ω −
+ +
∴ ω
∴
∴
ζω
∴
×
π
= ×
= ×
× = × = ×
Error
Input Load Gear
Detector Motor Output
R(s) 1 N1 C(s)
Ke Km
+ Js2 + fs N2
–
+
+
+
+ +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 197
ζω
∴ ζ =
ω ×
sTd
+
R(s) + 2.58 C(s)
1
+ 0.25s2 + 1.0s
–
+
+
+ +
+
+
+
+
+
∴ ω
ζω
198 LINEAR CONTROL SYSTEMS
sKt
Fig. 6.10.7.
θ θ
×
θ +
×
θ
+
+
θ ×
θ + × +
r(s) + 8E (s)4
×a10 o(s)
[250s2 + (8 × 104Kt + f)s]
×
θ + × +
θ ×
+
+ × +
θ ×
θ + × + + ×
θ
θ
× +
+ +
× +
+
+ ∴ ω
× +
ζω
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 199
ζ
∴ ζω
× π π
=
+
+ R(s) 25 C(s)
+
s(s + 5)
+
+
Fig. 6.10.8.
+ +
ζω
∴ ζω
ζ =
ω ×
ω ω −ζ = −
ζπ π
− −
−
× ×
−ζ
200 LINEAR CONTROL SYSTEMS
ζ ×
=
ω
sKt
+
+
+
+ +
R(s) + E25
a(s) C(s)
2
[s + (25Kt + 5)s]
+ +
+
ζπ π
− −
−
× = ×
−ζ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 201
R(s) + K C(s)
s(s + 6)
–
0.2
+
+
+
+ +
∴ ω
ζω
ζ
∴
× ×
∴ ω
×
=
ζω ×
π π
=
ω −ζ −
ζπ π
− −
−
× = ×
−ζ
202 LINEAR CONTROL SYSTEMS
TL(s)
R(s) + 1 C(s)
6
+ 0.15s2 + 0.9s
¡
¡
R(s) + 1 C(s)
6 2
0.15s + 0.9s
+
=
+ ×
+
+ + + +
∴
ω ζω
∴
ζ
×
ζπ π
= × =
ζ
×
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 203
ζ ×
=
ω
¡
–TL(s) + 1 C(s)
2
0.15s + 0.9s
+
=
¡
+
+
=
+ +
¡
¡
+ +
¡ ¡
∴
+ +
∴ =
→ + +
β
β
ω ζ
+
∴
+ β +
+
204 LINEAR CONTROL SYSTEMS
β
β
∴
ζω ω
¡
ω
ζω
β
3
+
∴
3
∴
+ × +
∴ =
= = =
× + +
+
+ + +
=
+ ×
3
+ +
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 205
+ +
= =
+
+
+
+ ×
+ +
+ +
+
+
∴ =
+ +
+
ω
ζω
∴
ζ =
ω ×
ζω
×
ζπ
π
× =
ζ
= ×
×
¢
+ +
¢ = = ∞
→ → + +
+
206 LINEAR CONTROL SYSTEMS
=
+ + ∞
+ +
¢ = =
→ → + +
+
×= ×
+ +
¢ = =
→ → + + +
= × =
× ∞
ζω ω
¢
= = = ∞
→ → + ζω + ω
= = =
→ → + ζω + ω
ω
= = =
→ → + ζω + ω
+ +
= ¢ = ∞
→ → + +
+ +
= ¢ = ∞
→ → + +
+ +
= ¢ = ∞
→ → + +
β
α β
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 207
+ β
= ∴
=
+
+ α + β +
β αβ
+ β
=
+ α
=
+
∴ =
+ β
+
+ α
+ α
∴
+ α+ β
3
+ α
= =
→ → + α+ β
α
β
+
= = ∴
→ + +
+
=
→ →
+ +
+ +
+
+ +
=
→ + +
+
= = ∴
→ + +
+
=
→ →
+ +
+ +
+
+
+ +
=
→ + +
= ∴
′
′′
¥
+
+
¥
+
=
+ +
+
=
+ +
¦
+
+ +
∴
′
″
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 209
′′′
+
£
£
′ £ £
″ ′″ £ ″″
+ +
= = = =
+
+ + +
+
+
+ +
∴ ∴
+ +
+
= ∴
→ → +
+
=
→ →
+
+ +
+
+
→ +
+
+ +
→ +
+ § +
¨
+
→ +
+ +
+
=
→ + → +
∴
= =
→ → → +
+
+
=
→ + → +
∴
=
→ →
210 LINEAR CONTROL SYSTEMS
→ +
+ − × +
−
→ +
− +
− =
→
+ →
+
∴
¢
Er + Eo
G
–
= = =
+ ¢ +
×
∴ =
∴ =
¤ = = =
+ ¢ +
×
3
¤ = ∴ =
©
∴
∆
∆
∆
∴ ∆
× =
ω
+ ζ ω ω
=
ω + ζ ω + ¢ ω
+ ¢
+ ζ ω
∂
¤ =
∂
3
ω
+ ζω + ¢
ω
∂ + ζ ω + ¢
ω ω
− ω¢ ω ω + ζω
= =
∂
+ ζ ω + ¢
ω
+ ζ ω ¢+
ω
ω + ζ ω ω
∴ ¤
+ ζ ω + ¢
ω + ζ ω +
¢
ω
+ ζω
¢ ω
¤ = ¤ =+
+ ζ ω + ¢ ω + ζ ω
¢
∂ ¢
¤ =
¢
∂¢
3
ω
+ ζ ω + ¢
ω
212 LINEAR CONTROL SYSTEMS
+ ζ ω + ¢ ω
∂
ω −
ω
ω
∴ =−
∂¢ + ζ ω + ¢
ω
+
ζ ω ¢ +
ω
¢
ω ω
∴ ¤ =−
¢
+ ζ ω + ¢
ω + ζω ¢ + ω
¢ ω
+ ζ ω + ¢
ω
PROBLEMS
R(s) + + 1 C(s)
s+1
s+2
– –
1
s+2
1
s(s + 1)
R(s) + + 2 C(s)
s + 0.1
s(s + 0.5)
– –
0.5
1 1
R(s) 1 4 s s+2 1 C(s)
– 0.75
¤
θ
TIME RESPONSE ANALYSIS OF CONTROL SYSTEMS 213
ω
=
+ ζω +ω
+
+
£
¤
££
¢
ª
+
+ β + +
β
=
+ +
INPUT 2
TL(s)
INPUT 1 OUTPUT
R(s) + 1 C(s)
K
+ Js2 + fs
+ +
+ + + +
£¢
+ +
« π
+
=
+ +
+
¢
+ +
¬
+
+
6.20. Calculate the reference voltage E for the output voltage E = 250 V, given that the forward
r o
path gain K = 200 for the open-loop system shown in Fig. 6.P-6.
Er K Eo
1 + sT1
(b) Calculate the reference voltage E and the data in (a) if the system is made closed-loop with
r
6.21. Determine the sensitivity of the overall gain for the system shown in Fig. 6.P-7 and calculate
the value of K such that the sensitivity is 0.2 under steady state conditions.
R(s) + 9 C(s)
K
s2 + 4s + 9
0.2
6.22. The block diagram shown in Fig. 6.P-8 shows a speed control system. Compare the time
response
Er(s) KT N1 w(s)
K
+ 1 + sT1 N2
–
sKt
CONTENTS
• Stability in Terms of Characteristic Equation of a Control System
• To Determine the Number of Roots having Positive Real Parts for a Polynomial
• Hurwitz Determinants of a Polynomial
• Routh-Hurwitz Criterion
• Solved Examples : Routh-Hurwitz Criterion
• Nyquist Criterion
• Gain Margin and Phase Margin
• Relative Stability from Nyquist Plot
• Gain Phase Plot
• Closed-loop Frequency Response of a Unity Feedback Control System from Nyquist Plot
• Constant Magnitude Loci : M-Circles
• Constant Phase Angle Loci : N-Circles
• Closed-loop Frequency Response of Control System from M and N-Circles
• Gain Adjustment Using M-Circle
• Nichols Chart
• Cutoff Frequency and Bandwidth
• Solved Examples : Nyquist Criterion
• Inverse Nyquist Plot
216
STABILITY ANALYSIS OF CONTROL SYSTEMS 217
• I/M Circles
• Gain Adjustment using 1/M Circle
• Bode Plot
• Solved Examples : Bode Plot
• Bode Plot for Time Delay Element e–sT
• Correlation Between Transient Response and Frequency Response
• Root Locus
• Salient Features of Root Locus Plot
• The Procedure for Plotting Root Locus
• Solved Examples : Root Locus
• Root Contours
Introduction. In the chapter on time response of control system, the time solution of
system equations were obtained. The time solutions thus obtained give the variation of the
system output with time as independent variable. The time solution can be classified into
two categories, the one wherein the output tending to a finite value and the other wherein
the output approaching towards infinite value with the advancement of time. The former
case is termed as stable system while the latter is termed as unstable system.
An illustration to explain stability and instability is shown in Fig. 7.0.1.
C1(t)
r(t)
C2(t)
r(t)
c(t)
– Re + Re
t
– Imj
(a) Stable
+ Imj
c(t)
– Re + Re
t
– Imj
(b) Sustained oscillations (marginally stable)
+ Imj
c(t)
– Re + Re
t
– Imj
(c) Unstable
Fig. 7.1.1. Time response and location of roots of characteristic equation.
In general the characteristic equation 1 + G(s) H(s) = 0 can be expressed in polynomial
form as follows :
a0sn + a1sn 1 + a2sn 2 + ... + an = 0 ...(7.1)
Only the sign of the real part of the roots of the polynomial (7.1) is to be ascertained
which in turns indicates absolute stability or instability.
− a1 ± a12 − 4 a0 a2
s1, s2 =
2a0
It is observed that for the real parts of both the roots to be negative, the coefficients a0,
a1 and a2 should be either positive or of same sign. Thus if all the coefficients of a first order
and second order polynomial are either positive or of same sign then it is concluded that the
roots of the polynomial does not have a positive real part and a control system whose charac-
teristic equation is expressed by such a polynomial is stable.
(c) However, the requirement that all positive coefficient or same sign coefficient is not
sufficient for ascertaining the stability of a control system whose characteristic equation has
an order higher than 2.
A clarification of the above statement is supported by an example of a third order
characteristic equation given below :
s3 + s2 + 5s + 125 = 0 ...(7.4)
On factorization Eq. (7.4) is expressed as
(s + 5) (s2 4s + 25) = 0 ...(7.5)
The three roots of Eq. (7.5) are
s1 = 5
s2, s3 = (+ 2 ± j4.5)
Although all the coefficients of the third order Eq. (7.4) are positive the equation has
distinctly two roots s2 and s3 having positive real parts which makes the system unstable.
Consider another third order characteristic equation given below :
s3 + 4s2 + 6s + 4 = 0 ...(7.6)
On factorization Eq. (7.6) is expressed as
(s + 2) (s2 + 2s + 2) = 0 ...(7.7)
The three roots of Eq. (7.7) are
s1 = 2
s2, s3 = ( 1 ± j1).
It is observed that none of the roots of Eq. 7.6 have positive real parts, therefore, the
system is stable.
Hence, for determining stability of higher order systems, in addition to aforesaid
necessary condition, a sufficient condition is further needed and this sufficient condition is
given by Hurwitz determinants as given below.
a1 a6 − a0 × 0
b5 = = a6
a1
222 LINEAR CONTROL SYSTEMS
b1 a3 − a1 b3 c1 b3 − b1 c3
c1 = , d1 = etc.
b1 c1
From the array (7.11) so formed, it is observed that the elements of the first column
are related to Hurwitz determinants [for example b1 = (D2/D1)]. The number of changes of
sign in the first column elements of Routh array gives the number of positive real part roots
of the polynomial. Therefore, for a stable system there should be no change of sign in the first
column of Routh array formed from the coefficients of the characteristic equation expressed
in polynomial form.
While applying Routh-Hurwitz criterion it is imperative that no powers of s in the
characteristic equation be absent. Any absence of such powers indicate the presence of at
least one positive real part root and confirms system instability by inspection.
However, if the characteristic equation contains either only odd or even powers of s,
this indicates that the roots have no real parts and possess only imaginary parts and there-
fore the system has sustained oscillations in its output response.
Following difficulties are faced while applying Routh-Hurwitz criterion :
1. The element in one of the first columns of Routh array is zero.
2. All the elements in one of the rows of Routh array are zero.
Above cases are considered in examples 7.5.4 and 7.5.5, which follows :
Example 7.5.1. A closed-loop control system has the characteristic equation given by,
s3 + 4.5s2 + 3.5s + 1.5 = 0. Investigate the stability using Routh-Hurwitz criterion.
s3 1 3.5
s2 4.5 1.5
4.5 × 3.5 − 1 × 1.5
s1 3.16 = 0
4.5
3.16 × 1.5 − 4.5 × 0
s0 1.5 = 0
3.16
In the first column of the Routh array there is no change in sign of the elements, hence
there is no root of the characteristic equation with positive real part and the system is stable.
Example 7.5.2. Apply Routh criterion to determine the stability of the system having
the characteristic equation as
s3 + 4 × 102 s2 + 5 × 104 s + 2 × 106 = 0.
Solution. The calculations in the Routh table are simplified by substituting s = αp,
(α = 102) in the characteristic equation, thus the characteristic equation takes the following
form :
(102 p)3 + 4 × 102 (102 p)2 + 5 × 104 (102 p) + 2 × 106 = 0
or p3 + 4p2 + 5p + 2 = 0
STABILITY ANALYSIS OF CONTROL SYSTEMS 223
The Routh table is formed below :
p3 1 5
p2 4 2
p1 4.5 0
p0 2 0
All the first column elements are of the same sign, therefore, the system is stable.
Example 7.5.3. Determine the stability of a system whose overall transfer function is
given below
C (s) 2s + 5
=
R( s) s 5 + 1.5s 4 + 2s 3 + 4s 2 + 5s + 10
If the system is found unstable, how many roots it has with positive real part ?
s5 1 2 5
s4 1.5 4 10
Sign change
s3 0.66 1.66 0
Sign change
s2 0.227 10 0
s1 27.4 0
s0 10 0 0
As there are two changes in sign of the elements in the first column of Routh table, it
is concluded that the system has two roots with positive real part and as such the system is
unstable.
Example 7.5.4. Determine the stability of a closed-loop control system whose charac-
teristic equation is
s5 + s4 + 2s3 + 2s2 + 11s + 10 = 0.
while forming the Routh array as above the third element in the first column is zero and thus
the Routh criterion fails at this stage. The difficulty is solved if zero in the third row of the
first column is replaced by a symbol ε and Routh array is formed as follows :
s5 1 2 11
s4 1 2 10
s3 ε 1 0
Sign change
⎛ ε × 2 − 1×1⎞
s2 lim ⎜ ⎟ = ∞ 10 0
ε→0 ⎝ ε ⎠
Sign change
⎛ 2
10 ε ⎞
s1 lim ⎜ 1 − ⎟ =1 0 0
ε→0 ⎜ 2ε − 1 ⎟⎠
⎝
s0 10
The limits of fourth and fifth element in the first column as ε → 0 from positive side
are ∞ and + 1 respectively, indicating two sign changes, therefore, the system is unstable
and the number of roots with positive real part of the characteristic equation is 2. Ans.
The presence of a zero in the first column of the Routh s tabulation leads to following
conclusions :
(i) equal real roots with opposite signs. As one of the roots is +ve, the system is un-
stable as indicated by the sign change in the first column.
(ii) pair of conjugate root on imaginary axis. This gives marginal stability provided
there is no sign change in the first column.
In the both cases as per procedure 0 is replaced by ε.
The conclusions as above are verified below :
(i) F(s) = (s 1) (s + 1)2 (s + 2) : two equal roots with opposite sign s = 1, s = + 1
or F(s) = s4 + 3s3 + s2 3s + 2
The Routh s tabulation is formed below :
s4 1 1 2
s3 3 3
s2 2 2
s1 0 (ε) + as ε → 0
s0 2
There is one sign change in the first column ; the system being unstable.
(ii) F(s) = (s + 1) (s + 2) (s2 + 1) : pair of conjugate root on imaginary axis s = ± j 1
or F(s) = s4 + 3s3 + 3s2 + 2
The Routh s tabulation is formed below :
s4 1 3 2
s3 3 3
s2 2 2
s1 0 (ε) + as ε → 0
s0 1
STABILITY ANALYSIS OF CONTROL SYSTEMS 225
There is no sign change in the first column and noting that the polynomial F (s) has a
pair of conjugate root on imaginary axis, the system is marginally stable.
The statements mentioned above are verified as per the examples given below :
F(s) = (s2 + 1) (s + 1) (s + 2) (s + 3) : pair of conjugate root on imaginary axis s = ± j 1
= s5 + 6s4 + 6s3 + 12s2 + 5s + 6
s5 1 6 5
s4 6 12 6 Aux. equation A(s) = 6s2 + 6
s3 4 4 dA (s) = 12s
ds
dA (s)
s2 6 6 * coeff. of
ds
s1 0(12)* 0
s0 6
There is no sign change in the first column, the system is marginally stable.
F(s) = (s2 + 2)2 (s + 1) (s + 2) : pair of conjugate root on imaginary axis of multiplicity
order two i.e. s = ± j 2 , s = ± j 2
F (s) = s6 + 3s5 + 6s4 + 12s3 + 12s2 + 12s + 8
s6 1 6 12 8
s5 3 12 12 (1) A (s) = 2s4 + 8s2 + 8
dA ( s) dA (s)
s4 2 8 8 = 8s2 + 16s *coeff. of
ds ds
s3 0(8)* 0(16)* 0 (2) A (s) = 4s2 + 8
dA (s) dA (s)
s2 4 8 = 8 ** coeff. of
ds ds
s1 0(8)** 0
s0 8
There is no change of sign in the first column of Routh s tabulation but existence of
two rows having zero elements make the system unstable.
Example 7.5.6. Using Routh criterion investigate the stability of a unity feedback
control system whose open-loop transfer function is given by
− sT
G(s) = 2 e
s(s + 2)
e− sT
= ...(1)
(s2 + 2s + e− sT )
STABILITY ANALYSIS OF CONTROL SYSTEMS 227
The characteristic equation is
s2 + 2s + e sT = 0 ...(2)
⎛ s2 T 2 ⎞
Since = ⎜⎜ 1 − sT +
e sT + ⋅ ⋅ ⋅ ⎟⎟ ...(3)
⎝ 2! ⎠
Considering only first two terms of series (3), the characteristic equation can be
written as
s2 + 2s + (1 sT) = 0
or s2 + (2 T) s + 1 = 0 ...(4)
The Rough array for equation (4) is formed below :
s2 1 1
s1 (2 T) 0
s0 1 0
For the system to be stable the sign of the first element in the Routh table should be
positive, therefore, the condition for the stability of the system is given by
(2 T) > 0
∴ T < 2. Ans.
Example 7.5.7. (a) Using Routh-Hurwitz criterion determine the relation between K
and T so that unity feedback control system whose open-loop transfer function given below
is stable
K
G(s) =
s[s(s + 10) + T]
Example 7.5.7. (b) Determine the value of K such that the roots of the characteristic
equation given below lie to the left of line s = 1
s3 + 10s2 + 18s + K = 0.
Solution. Put s = s1 1
∴ (s1 1)3 + 10 (s1 1)2 + 18(s1 1) + K = 0
or s13 + 7 s12 + s1 + (K 9) = 0
The Routh s tabulation is given below :
s13 1 1
s12 7 (K 9)
7 − ( K − 9)
s11
7
0
s1 K 9
K < 16 and K > 9,
K to have values between 9 and 16. Ans.
Example 7.5.8. The open-loop transfer function of a unity feedback control system is
given by
K
G(s) =
s(sT1 + 1) (sT2 + 1)
Applying Routh-Hurwitz criterion determine the value of K in term of T1 and T2 for
the system to be stable.
Example 1. Let G(s) H(s) = (s + 1.5) and s being transversed in s-plane along a line
through points sa = (1 + j1) and sb = (2.8 + j0.5) as shown in Fig. 7.6.1.
Solution.
Imj Imj
j2 j2
sa(1 + j1) G(sa)H(sa)
sb(2.8 + j0.5)
j1 G(sb)H(sb)
Mb j1 Ma Mb
Ma fa fb fa
fb
1.5 0 1 2 3 Re 1 2 3 4 5 Re
s-plane 2.5
4.3
(a) (b) G(s)H(s)-plane
Fig. 7.6.1. Mapping from s-plane to G(s)H(s)-plane for G(s)H(s) = (s + 1.5).
The values of G(sa) H(sa) and G(sb) H(sb) are calculated below :
G(sa) H(sa) = (sa + 1.5)
= [(1 + j1) + 1.5] ...(1)
230 LINEAR CONTROL SYSTEMS
= (2.5 + j1)
G(sb) H(sb) = (sb + 1.5) = [(2.8 + j0.5) + 1.5]
= (4.3 + j0.5) ...(2)
In Fig. 7.6.1 (a) the zero of the transfer function is shown at location s = 1.5 in
s-plane. A phasor Ma is drawn from the point s = 1.5 to the point sa. The magnitude Ma and
phase φa of this phasor gives the value of the G(s) H(s) at sa in polar form. Similarly Mb and
gives the value of G(s) H(s) at sb. The values obtained in equations (1) and (2) are plotted in
G(s) H(s)-plane.
From Fig. 7.6.1 (b) it is observed that the magnitude and the phase of the transfer
function G(s) H(s) = (s + 1.5) at a point in G(s) H(s)-plane is given by the magnitude and the
phase of a phasor drawn from the origin of G(s) H(s)-plane.
The change in phase of G(s) H(s) as s is varied from sa to sb is (φb φa).
Example 2. Another example is given below wherein change in the phase of G(s) H(s)
is observed when the specified path of traversing s in s-plane being a closed one. The said
closed path is traversed in clockwise direction. The transfer function under consideration
has two zeros, the one lies inside the closed path and the other outside the closed path.
Z2 Imj
M2
f2 sa
sb G(sa)H(sa) G(sb)H(sb)
M1
f1
sd Z1
sc G(sd)H(sd)
G(sc)H(sc)
O s-plane Re
G(s)H(s)-plane
(a) (b)
Fig. 7.6.2. Mapping from s-plane to G(s)H(s)-plane for G(s)H(s) = (s z1) (s z2).
*The transfer function G(s) H(s) considered has the number of poles greater than the number of zeros.
232 LINEAR CONTROL SYSTEMS
Imj
Imj
sa
G(sd)H(sd)
sd G(sa)H(sa) ∆ Aug = 2π (P Z)
P1 2π = 2π (1 Z)
Re ∴ Z=2
– 2p Re
sc sb
G(sc)H(sc) G(sb)H(sb)
Imj
Imj
G(sd)H(sd)
sd
sa G(sc)H(sc)
P2 P1
∆ Aug = 2π (P Z)
2π = 2π (2 Z)
Re Re
2p ∴ Z=1
sc
sb G(sb)H(sb) G(sa)H(sa)
Imj Imj
sa )
sb (s b
)b H G(sa)H(sa)
s
SA G(
sd ∆ Aug = 2π (P Z)
× P1 0 = 2π (P Z)
sc
Re ∴ Z=0
× P2 Re G(sd)H(sd)
P = 0 (given) G(sc)H(sc)
DAUG = 0
Imj
Imj
sd
sc G(sa)H(sa)
P2 P1 ∆ Aug = 2π (P Z)
sb sa Re Re 4π = 2π (2 Z)
G(sd) ∴ Z=0
G(sb)H(sb) H(sd)
G(sc)H(sc)
P = 2 (given)
DAUG = + 4p
*If G(s)H(s) has a pole on imaginary axis then such poles are excluded from the path and the modified
path around such a pole being a semi-circle in the RHS of s-plane. The radius of the semi-circle being
approaching to zero.
234 LINEAR CONTROL SYSTEMS
Z+ is also the number of roots of the characteristic equation 1 + G(s) H(s) = 0 with positive real part.
STABILITY ANALYSIS OF CONTROL SYSTEMS 235
j0T is neglected in comparison with 1
K
∴ G(j0) H(j0) = = j∞ = ∞ ∠ 90°
j0
(b) Put ω = + ∞ in Eq. (7.24)
K
∴ G(j∞) H(j∞) =
j ∞ ( j ∞T + 1)
1 is neglected in comparison with j∞.
K
∴ G(j∞) H(j∞) = 0= 0 ∠ 180°
j 2 ∞T
It is thus observed that as ω increases from ω = + 0 to ω = + ∞ the plase angle of G(jω)
H(jω) changes from 90° to 180° and the respective magnitudes are ∞ and 0 .
The nature of the plot is shown in Fig. 7.6.5. The plot for values ω = 0 to ω = ∞ is
shown by dotted lines. The arrows marked on the plot indicate the direction of increasing ω.
Imj Imj
w=–0
w=+0
w=–¥ r
q
– 1 + j0 w=+¥ Re Re
w=–0
G(s)H(s)-plane
w=+0
K
Fig. 7.6.5. Nyquist plot for G(s) H(s) = for Fig. 7.6.6. Variation of s from s = 0
s(sT + 1) to s = + 0 in s-plane.
variation ω = + 0 to ω = + ∞ and ω = ∞ to ω = 0.
The equation of the phasor along the semi-circular arc shown in Fig. 7.6.6 is given by
(s 0) = re jθ
s = re jθ ...(7.27)
substituting for s in Eq. (7.26) from (7.27) following relation is obtained
K K
G (s) H(s) = n jnθ
= n e− jnθ ...(7.28)
s→0 r e j
In s-plane
At s = j0, θ = π/2
and at s = + j0, θ = + π/2
Therefore, change in θ as s is varied in s-plane from s = j0 to s = + j0 is (+ π/2) ( π/2)
= + π.
The corresponding change in phase of G(s) H(s) in G(s) H(s) plane is determined
below :
K − jnθ
Since G ( s) H(s) = e
r→0 rn
and at s= j0, θ = π/2
⎛ π⎞
K − jn⎜ − ⎟
∴ G( j0) H( j0) = lim n e ⎝ 2 ⎠
r→0 r
jnπ
K
= lim e 2 ...(7.29)
r→0 rn
Similarly at s = + j0, θ = π/2
⎛π⎞
K − jn⎜ ⎟
⎝2⎠
∴ G(+ j0) H(+ j0) = lim n e
r→0 r
− jnπ
K
= lim e 2 ...(7.30)
r → 0 rn
From Eqs. (7.29) and (7.30) it is noted that the corresponding change in the phase of
G(s) H(s) in G(s) H(s)-plane as s is varied in s-plane from s = j0 to s = + j0 as per direction
(anticlockwise) shown in Fig. 7.6.6 is given by
⎛ nπ ⎞ ⎛ nπ ⎞
⎜− ⎟ − ⎜+ ⎟ = nπ (clockwise direction) ...(7.31)
⎝ 2 ⎠ ⎝ 2 ⎠
It is, therefore, concluded that, if in the s-plane s changes from s = j0 to s = + j0 by
π radian (anti-clockwise) then the corresponding change in phase of G(s) H(s) in G(s) H(s)-
phase is nπ radian (clockwise) and the magnitude of G(s) H(s) during this phase change is
infinite. n denotes the type of the transfer function or the number of poles of G(s) H(s) at
the origin.
The closing angle of Nyquist plot from s = j0 to s = j0 is given in the following table
according to type of G(s) H(s) :
STABILITY ANALYSIS OF CONTROL SYSTEMS 237
0 0 0 The points ω = 0 ω = + 0
are coincident
1 1 π ∞
2 2 2π ∞
3 3 3π ∞
, ,
, ,
n n nπ ∞
Note. From Nyquist plot the stability of a closed-loop system as well as that of open-loop sys-
tem can be determined. If the origin of G(s) H(s)-plane is considered as (0, 0) the open-loop stability is
obtained and consideration of the point ( 1 + j0) as origin gives the stability of closed-loop system.
Examples given below illustrate the application of Nyquist criterion to determine the
stability of closed-loop control systems.
K
Solution. G(s) H(s) =
s (sT + 1)
Put s = jω
K
∴ G(jω) H(jω) = ...(1)
jω ( jωT + 1)
Rationalizing Eq. (1) and separating into real and imaginary parts following equation
is obtained
KT jK
G(jω) H(jω) = − − ...(2)
(ω2T 2 + 1) ω (ω2T 2 + 1)
From Eq. (2) it is observed that as ω increases from ω = + 0 to ω = + ∞ both the real part
and the imaginary part lie in the third quadrant of G(s) H(s)-plane.
At ω = + 0, ∠ G( + j0) H( + j0) = 90° and the magnitude approaches infinity.
At ω = + ∞, ∠ G( + j∞) H( + j∞) = 180° and the magnitude approaches zero.
The Nyquist plot as ω varied from ω = + 0 to ω = + ∞ is shown in Fig. 7.6.7. The plot for
ω = ∞ to ω = 0 is shown by dotted lines.
As the system is type 1 the plot is closed from ω = 0 to ω = + 0 through an angle of
π (clockwise) with an infinite radius. The arrowheads shown in Fig. 7.6.7 are in the direc-
tion of increasing ω.
As the point ( 1 + j0) is not encircled by the plot, therefore,
N=0
238 LINEAR CONTROL SYSTEMS
Imj Imj
w=–0
+ j¥
+ j0
w=–¥
w=+¥
Re – 1 + j0 Re
– j0
– j¥
w=+0
s-plane G(s)H(s)-plane
K
Fig. 7.6.7. Nyquist plot for G(s) H(s) = .
s(sT + 1)
The number of zeros (roots) of the characteristic equation with positive real part is
determined by using relation*
N = (P+ Z+)
because, P+ = 0 ∴ Z+ = 0
Hence, the number of zeros (roots of the characteristics equation) with positive real
part is nil and the closed-loop system is stable.
R(s) + K C(s)
(sT – 1)
–
1
s
*The value of N is determined by imagining a phasor whose one end rests on the point ( 1 + j0), the
other end of this phasor is allowed to trace the Nyquist plot starting at ω = +0 through ω = +∞, ω = ∞,
ω = 0 and finally ending at ω = +0. The net counter clockwise rotations of this phasor gives the value
of N. A clockwise rotation of this phasor is noted by a negative number of N.
STABILITY ANALYSIS OF CONTROL SYSTEMS 239
KT jK
G(jω) H(jω) = − 2
+
2 2 2
...(2)
(ω T + 1) ω (ω T + 1)
From Eq. (2) it is observed that as ω increases from ω = + 0 to ω = + ∞ the real part and
the imaginary part lie in the second quadrant of G(s) H(s)-plane.
At ω = + 0, ∠G( + j0) H( + j0) = 90° and the magnitude approaches infinity.
At ω = + ∞, ∠G(j∞) H( j∞) = 180° and the magnitude approaches zero.
The Nyquist plot as ω is varied from ω = + 0 to ω = + ∞ is shown in Fig. 7.6.9. The plot
for the values of ω from ω = ∞ to ω = 0 is shown by dotted lines.
Imj
w=+0
Imj
+ j¥
+ j0
w=+¥
0 Re – 1 + j0 w=–¥ Re
– j0
– j¥
w=–0
G(s)H(s)-plane
K
Solution. G(s) H(s) =
s (sT1 + 1) (sT2 + 1)
Put s = jω
K
∴ G(jω) H(jω) =
jω ( jωT1 + 1) ( jω T2 + 1)
− jK [(− jωT1 + 1) (− jω T2 + 1)]
=
ω (ω2T12 + 1) (ω2 T22 + 1)
K [ − ω (T1 + T2 ) − j (1 − ω2 T1 T2 )]
= ...(1)
ω (ω2T12 + 1) (ω2 T22 + 1)
Separating equation (1) into real and imaginary parts following equation is obtained :
K (T1 + T2 ) jK (1 − ω2T1T2 )
G(jω) H(jω) = ...(2)
(ω T1 + 1) (ω2T22 + 1)
2 2
ω (ω2T12 + 1) (ω2 T22 + 1)
At, ω=+0
G(+ j0) H(j + 0) = K(T1 + T2) j∞
π
= ∞ ∠ ...(3)
2
From Eq. (1) it is observed the Nyquist plot crosses the negative real axis and the
value of ω at which this occurs is determined by equating the imaginary part of Eq. (2) to zero
therefore,
K (1 − ω2T1T2 )
=0 ...(4)
ω (ω2T12 + 1) (ω2 T22 + 1)
The solution of Eq. (4) is
ω=± 1
T1T2
1
+ j¥ w=+
ÖT1T2
+ j0
w=+¥
0 Re w=–¥ Re
– j0
1
w=– – kT1T2
– j¥ ÖT1T2
T1 + T2
s-plane
w=+0
G(s)H(s)-plane
Fig. 7.6.10. Nyquist plot for G(s) H(s) = K/s(sT1 + 1)(sT2 + 1).
the point ( 1 + j0) is not encircled by the Nyquist plot and the closed-loop system is found to
be stable (Note that P+ = 0).
K T1T2
However, if >1
(T1 + T2 )
the point ( 1 + j0) is encircled twice in clockwise direction indicating that there are two roots
of the characteristics equation with positive real part and the closed-loop system is unstable.
The stability condition is derived below :
K T1T2 ⎛ 1 1 ⎞
For stability ≤ 1 or K ≤ ⎜ + ⎟
(T1 + T2 ) T
⎝ 1 T 2 ⎠
Alternatively, general shape of the Nyquist plot (Fig. 7.6.10) in relation to open-loop
transfer function
K
G(jω) H(jω) =
jω ( jωT1 + 1) ( jωT2 + 1)
can be drawn using following method.
K
G(jω) H(jω) =
2
ω ω T12 + 12 ω2T22 + 11
and ∠G(jω) H(jω) = 90° tan 1 ωT1 tan 1 ωT2
ω (T1 + T2 ) 1
= 90° tan 1 for ω <
1 − ω2T1T2 T1T2
1
ω (T1 + T2 ) 1
= 90° tan 180° for ω >
1 − ω2T1T2 T1T2
(1) As ω → 0
G(j0) H(j0) → ∞
G(j0) H(j0) → 90°
242 LINEAR CONTROL SYSTEMS
1
(2) At ω =
T1T2
1
(T1 + T2 ) / T1T2
= 90° tan
0
= 90° tan 1 ∞=
90° = 180° 90°
(3) As ω → ∞
G(j∞) H(j∞) → 0
and ∠G(j∞) H(j∞) → ( 90° 0°) 180° = 270°
The Nyquist plot crosses the negative real axis of G(s) H(s)-plane at
1 ⎛ 1 ⎞ ⎛ 1 ⎞ K T1T2
and G j ⎜⎜
ω= ⎟H⎜ j
⎟ ⎜
⎟ =−
⎟
T1T2 ⎝ T1T2 ⎠ ⎝ T1T2 ⎠ T1 + T2
As per above derivations general shape of the Nyquist plot is shown in Fig. 7.6.10 and
closed-loop stability conditions mentioned above can be verified.
* The plot of G(jω) H(jω) for values of ω from ω = +0 to ω = + ∞ is also known as polar plot.
STABILITY ANALYSIS OF CONTROL SYSTEMS 243
Imj c(t)
w=¥
– 1 + j0 Re t
w
(a)
Imj c(t)
w=¥
Re t
– 1 + j0
w
(b)
Imj c(t)
w=¥
– 1 + j0 Re t
w
(c)
Fig. 7.7.1. Nyquist plot and corresponding step response.
The Nyquist plot shown in Fig. 7.7.1 (b) is for the value of K which brings the closed-
loop system on the verge of instability. At this value of K, the plot passes through the point
( 1 + j0). The point ( 1 + j0) is known as critical point. Unit circle Imj
For smaller values of K than above mentioned value
Phase
the system is stable. crossover
Nearer the Nyquist plot to the critical point w2
( 1 + j0) the system tends towards more oscillations
in its output. This nearness of the Nyquist plot to the w=+¥
critical point ( 1 + j0) determines the relative stability Re
–1 + j0 f
of a control system expressed in terms of gain margin
and phase margin.
The Nyquist plot for a stable system is shown in Gain
crossover
Fig. 7.7.2. The plot crosses the negative real axis at ω
w1
= ω2 and the corresponding magnitude G(jω2)
H(jω2) is less than 1. If the gain K of the system is w
further increased so that the plot passes through the Fig. 7.7.2. Gain margin and phase margin
point ( 1 + j0), the system starts tending towards from Nyquist plot.
244 LINEAR CONTROL SYSTEMS
instability. The magnitude of G(jω) H(jω) equals to 1 at this point. Thus, it is concluded
that if the gain is increased by a factor 1/G(jω2) H(jω2) the closed-loop system becomes
unstable. Thus there exists a margin for increasing the gain of the system. The frequency ω2
at which the Nyquist plot crosses the negative real axis is termed as phase crossover
frequency.
The gain margin is a factor by which the gain of a stable system is allowed to increase
before the system reaches instability.
In decibel the gain margin is given by
1
G.M. = 20 log10 db ...(7.32)
G( jω2 ) H ( jω2 )
For stable systems G(jω2) H(jω2) < 1, the gain margin in decibel is positive.
For marginally stable systems G(jω2) H(jω2) = 1, the gain margin in decibel is zero.
For unstable systems G(jω2) H(jω2) > 1, the gain margin in decibel is negative and
the gain is to be reduced to make the system stable.
Phase Margin. From Fig. 7.7.2 is observed that the unit circle (circle with centre at
origin and radius unity) crosses the Nyquist plot at ω = ω1. If the Nyquist plot is rotated
through an angle φ in the clockwise direction then the plot passes through the point ( 1 + j0)
and the closed-loop system becomes unstable. The rotation of the plot mentioned above is
achieved by changing the phase of G(jω) H(jω). The angle φ is known as the phase margin of
the system.
The phase margin of a stable system is the amount of additional phase lag required to
bring the system to the point of instability.
Fig. 7.7.2 gives the Nyquist plot for a stable system and the phase margin φ is positive.
The phase of the open-loop transfer function at ω = ω1 is ∠G(jω1) H(jω1).
The phase margin is given by
P.M. = 180° + ∠G(jω1) H(jω1) ...(7.33)
∠G(jω1) H(jω1) being measured clockwise.
The frequency ω1 at which the Nyquist plot crosses the unit circle is known as gain
cross over frequency. For a stable system the phase margin is positive.
Unit
circle
w=¥ w=+¥
fA w=–¥ Re
–1 + j0 fB Re
c b a 0
A
w
w w=+0
B
Fig. 7.8.1. Nyquist plot for two systems having same Fig. 7.8.2. Nyquist plot of conditionally stable system.
gain margin but different phase margin.
However, if the point ( 1 + j0) lies between ab the point ( 1 + j0) is encircled once in
clockwise direction and then once again in the anti-clockwise direction resulting in net
encirclements of the point ( 1 + j0) as nil indicating that the system is stable.
The system discussed above is stable only for a particular range of K any decrease or
increase in the value of K makes the system unstable. Such systems are called conditionally
stable systems.
The concepts of gain margin and phase margin are not applicable to conditionally
stable systems.
Gain
Imj crossover Imj
Unit circle Unit circle
Phase
crossover w1
w2 w=¥ w2 w=¥
–1 Re –1 Re
Gain
crossover
Phase
crossover
w
w
G(jw)H(jw)-plane G(jw)H(jw)-plane
(a) (b)
Fig. 7.9.1 (a) Polar plot for stable system. (b) Polar plot for unstable system.
+ +
[G(jw)H(jw)] db
[G(jw)H(jw)] db
w w
PM + w2
w1 GM –
0 db w1 0 db
GM +
w2
PM –
– –
–270° –180° –90° –270° –180° –90°
ÐG(jw)H(jw)-plane ÐG(jw)H(jw)-plane
(a) (b)
Fig. 7.9.2 (a) Gain phase plot for stable system. (b) Gain phase plot for unstable system.
|1
C( jω) G( jω) OB
M(ω) = ...(7.35)
+G
)|
∴ = = →
(jw
R( jω) 1 + G( jω)
(jw
AB
|G
)|
B
→
C( jω) ∠ OB ∠α
and ∠ = = = ∠(α θ) ...(7.36) w
R( jω) → ∠θ
∠ AB Fig. 7.10.1. Nyquist plot (polar plot)
As per Eqs. (7.35) and (7.36) the equation given below for transfer function 7.34.
is obtained
C ( jω)
= M ejφ ...(7.37)
R( jω)
where φ = (α θ)
The magnitude of C(jω)/R (jω) is given by M and the phase by φ. By taking measure-
→ →
ments of phasors OB, AB and angles θ, α at different values of ω on the Nyquist plot the
closed-loop frequency response can be obtained. The closed-loop frequency response consists
of two parts (i) a graph between C(jω) / R(jω) and ω (ii) a graph between ∠C(jω) / R(jω)
and ω.
The loci of constant magnitude and constant phase angle of closed-loop transfer func-
tion C(jω) / R(jω) can be drawn in a complex G(jω)-phase and from these loci the closed-loop
frequency response can be obtained.
∵ M = C( jω) = G( jω)
R( jω) 1 + G( jω)
x + jy
∴ M= ...(7.38)
1 + x + jy
x 2 + y2
∴ M = ...(7.39)
(1 + x)2 + y2
On squaring both sides of Eq. (7.39) and simplifying, following equation is obtained :
(1 M2) x2 2M2 x + (1 M2) y2 = M2 ...(7.40)
2M 2 M2
or x2 x + y 2 = ...(7.40 a)
(1 − M 2 ) (1 − M 2 )
248 LINEAR CONTROL SYSTEMS
2
⎛ M2 ⎞
Add ⎜⎜ 2 ⎟⎟ to both sides of Eq. (7.40 a)
⎝1 − M ⎠
2 2
2M 2 ⎛ M2 ⎞ M2 ⎛ M2 ⎞
∴ x2 x + ⎜ ⎟
⎜ (1 − M 2 ) ⎟ + y 2 =
+ ⎜ ⎟⎟ ...(7.41)
(1 − M 2 ) ⎝ ⎠ (1 − M 2 ) ⎜⎝ 1 − M 2 ⎠
2 2
⎛ M2 ⎞ ⎛ M ⎞
or ⎜⎜ x − ⎟ + y2 = ⎜⎜ ⎟⎟ ...(7.42)
⎝ 1 − M 2 ⎟⎠ ⎝1 − M
2
⎠
For different values of M, Eq. (7.42) represents a family of circles with centre at x =
(M2 / 1 M2), y = 0 and radius as (M / 1 M2). On a particular circle the value of M (magni-
tude of closed-loop transfer function) is constant, therefore, these circles are called M-circles.
The centres and radii of M-circles for different values of M are given in the following
table and M-circles are drawn in Fig. 7.11.1 :
M2 M
M Centre x = , y=0 Radius r =
1 M2 1 M2
M = 1.6
M=1 Y
M = 0.5
M=3
⎛ y⎞ ⎛ y ⎞
φ = tan 1 tan 1
⎜ ⎟ ⎜ ⎟
⎝ x⎠ ⎝1 + x ⎠
⎡ ⎛ y⎞ ⎛ y ⎞⎤
or tan φ = tan ⎢ tan − 1 ⎜ ⎟ − tan − 1 ⎜ ⎟⎥
⎣ ⎝x⎠ ⎝ 1 + x ⎠⎦
⎡ ⎛ y ⎞⎤ ⎡ ⎛ y ⎞⎤ y y
tan ⎢ tan − 1 ⎜ ⎟ ⎥ − tan ⎢ tan − 1 ⎜ ⎟⎥ −
⎣ ⎝ x ⎠⎦ ⎣ ⎝ 1 + x ⎠⎦ = x 1+ x
=
⎡ ⎛ y ⎞⎤ ⎡ ⎛ y ⎞⎤ y y
1 + tan ⎢ tan − 1 ⎜ ⎟ ⎥ ⋅ tan ⎢ tan − 1 ⎜ 1+ ⋅
⎟⎥ x (1 + x)
⎣ ⎝ x ⎠⎦ ⎣ ⎝ 1 + x ⎠⎦
y
or tan φ = ...(7.44)
x + x + y2
2
y
or x2 + x + y 2 =0 ...(7.45)
N
⎛1 1 ⎞
Add ⎜ + ⎟ on both sides of Eq. (7.45)
⎝ 4 4N 2 ⎠
⎛ 2 1⎞ ⎛ 2 y 1 ⎞ ⎛1 1 ⎞
∴ ⎜x + x + ⎟ + ⎜y − + 2 ⎟
= ⎜ + ⎟
⎝ 4⎠ ⎝ N 4N ⎠ ⎝ 4 4N 2 ⎠
2 2
⎛ 1⎞ ⎛ 1 ⎞ ⎛1 1 ⎞
or ⎜x + ⎟ + ⎜ y− ⎟ = ⎜ + ⎟ ...(7.46)
⎝ 2⎠ ⎝ 2N ⎠ ⎝ 4 4N 2 ⎠
For different values of N Eq. (7.46) represents a family of circles with centre at
1 1 1 1
x= , y= and radius as + 2
⋅
2 2N 4 4N
On a particular circle, the value of N or the value of phase angle of the closed-loop
transfer function is constant, therefore, these circles are called N-circles.
The centre and the radii for N-circles are given in the following table and N-circles are
drawn in Fig. 7.12.1.
250 LINEAR CONTROL SYSTEMS
1 1 1 1
φ N = tan φ Centre x = ,y= Radius R = +
2 2N 4 4N 2
90° ∞ 0 0.5
60° 1.732 0.289 0.577
50° 1.19 0.42 0.656
30° 0.577 0.866 1.0
10° 0.176 2.84 2.88
0° 0 ∞ ∞
+ 10° 0.176 2.84 2.88
+ 30° 0.577 0.866 1.0
+ 50° 0.19 0.42 0.656
+ 60° 1.732 0.289 0.577
+ 90° ∞ 0 0.5
Y
2
30°
60° 1
90°
120° X
–2 –120° 1 2
–90°
–60° –1
–30°
–2
22
30°
w=
w=
M = 1.6 w= 20
M = 0.5
M=2 14
30
90°
w = 15
w=
w=¥ 150° w=¥
Re w = 12 –150° Re
w = 10 –90°
5
w=5 w=
–1
4 –60°
G(jw) G(jw) w=
w=4 –30°
1
w=
w w
Fig. 7.13.1 (a)
M-Circle Imj
M2
B –1 1 – M2
A y 0 Re
1–
M
90°
M
2
P
y = sin–1(1/M)
1
G1(jω) =
jω (0.1 ω + 1) ( j0.2 ω + 1)
where G1(jω) = G(jω) for K = 1
Rationalizing G1(jω) and separating into real and imaginary part
0.3 j (1 − 0.02 ω2 )
G1 (jω) = −
(0.01 ω2 + 1) (0.04 ω2 + 1) ω(0.01 ω2 + 1) (0.04 ω2 + 1)
The polar plot of G1(jω) is plotted in Fig. 7.14.2 as per calculations given in the follow-
ing table :
ω rad/sec 2 3 4 7.07 10 ∞
Re G1 (jω) 0.248 0.2 0.158 0.1 0.03 0
Imj G1 (jω) j0.38 j0.184 j0.022 j0 + j0.001 + j0
1
⎛ 1 ⎞
ψ = sin ⎜ ⎟
⎝ Mr ⎠
1 ⎛1⎞
ψ = sin ⎜ ⎟
⎝2⎠
ψ = 30°
The line at an angle of ψ = 30° is drawn as shown in Fig. 7.14.2 and a circle tangential
to the polar plot as well as the line is drawn. The trangential point of the line and the
circle is marked P and the perpendicular from the point P intersects the negative real
axis at A = 0.22.
Imj
Mr = 2
0.1
w = 10 Re
–0.5 –0.4 –0.3 –0.22 w=¥
A 0
1
y = sin–1
4 –0.1 Mr
w=
P 1
= sin–1 = 30°
w=3 –0.2 2
–0.3
w
w=2 –0.4
Fig. 7.14.2. Gain adjustment for G(jω) = K/jω( j0.1ω + 1)( j0.2ω + 1).
If the circle drawn in Fig. 7.14.2 be the desired M circle corresponding to Mr = 2, then
the point A should be ( 1 + j0). This is possible if the scale is changed by a factor 1/ OA and
hence the value of K to obtain the desired resonant peak Mr = 2 is :
1 1
K= = = 4.54
OA 0.22
254 LINEAR CONTROL SYSTEMS
G3 f4 M = 1.5
s
ReG(jw)
M=3 G4 f1 f2 f3 f4 Re
Critical point –1 + j0
–180°
–270° –90°
Fig. 7.15.1. M-circle transformation to Nichols chart. Fig. 7.15.2. M-circles in gain-phase plane.
Constant M and constant N circles in the Nichols chart are deformed into squashed
circles. The complete Nichols chart extends for the phase angle of G(jω) from 0 to 360° but
the region of ∠Gj(ω) generally used for analysis of systems is between 90° and 270°. These
curves repeat after every 180° interval. If the open-loop transfer function of the unity feed-
back system G(s) is expressed as
G(s) = G(s) ejθ = G(s) [cos θ + j sin θ]
and the closed-loop transfer function
G(s)
M(s) =
1 + G(s)
Substituting (s = jω) in the above equations the frequency functions are
G(jω) = G(jω) [cos θ + j sin θ]
G( jω)
and M(jω) = Mejφ =
1 + G( jω)
Eliminating G(jω) from above two equations
G( jω)
M=
2
G( jω) + 2 G( jω) cos θ + 1
STABILITY ANALYSIS OF CONTROL SYSTEMS 255
0.25 db
28 db
–2°
0.5 db
24 db
–5°
1 db 20 db
b
0 db
5d
16 db
–0.
2 db
–1
12 db
db
3 db
0°
–1
4 db
–20° 8 db
6 db
db
db
–2
4 db
12 db –3 b
–4 d
b
–5 d 0 db
b
–6 d
–9 db – 4 db
–40°
–12 db – 8 db
–30°
–50°
– 12 db
–60°
–70°
–80°
–100°
–200°
–18 db – 16 db
– 2 40 °
–220°
– 20 db
–180°
–90°
–120°
–140°
–160°
–24 db
– 24 db
– 28 db
– 240°
– 220°
– 200°
– 180°
– 160°
– 140°
– 120°
– 100°
– 80°
– 60°
– 40°
– 20°
0°
1 sin θ
and φ = tan
| G( jω) | + cos θ
These equations define the plots in Nichols chart shown in Fig. 7.15.3.
0.25 db
28 db
–2°
0.5 db
24 db
–5°
1 db w = 0.2 20 db
– 0 db
db
.5
16 db
2 db
w = 0.3 –0
–1
12 db
db
3 db w = 0.5
0°
–1
4 db
–20° 8 db
6 db
db
db
–2
4 db
12 db w= –3 b
1 .0 –4 d b
5d
–
b 0 db
–6 d
w
–9 db – 4 db
=
1.
5
w
–40°
– 8 db
=
–12 db
2.
0
–30°
w = 3.0
–50°
– 12 db
–60°
–70°
–80°
–100°
–200°
–18 db – 16 db
w = 4.0
–240°
–220°
– 20 db
–180°
–90°
–120°
–140°
–160°
–24 db
– 24 db
– 28 db
– 240°
– 220°
– 200°
– 180°
– 160°
– 140°
– 120°
– 100°
– 80°
– 60°
– 40°
– 20°
0°
Fig. 7.15.4. Gain phase plot superimposed on Nichols chart for illustrative example.
From the data available in above table, the closed-loop frequency response of the
system is plotted as shown in Fig. 7.15.5 (a) and (b). Ans.
db
10
|C(jw)/R (j w )|
0
0.2 0.5 1 1.5 2 3 4 w
–5
– 10
– 15
–20
(a)
– 80°
– 120°
– 160°
– 200°
(b)
Fig. 7.15.5 (a) and (b) closed-loop frequency response for illustrative example.
frequency and the frequency range 0 to ωc is called the bandwidth of the system. The
response peak Mr occurs at resonance frequency ωr.
Mr
Gain (db)
3db
w
w wr wc (Log Scale)
BANDWIDTH
2
−2ωn2 (2ζ 2 − 1) + 4ωn4 (2ζ 2 − 1)2 + 4ωn4
ωc =
2
= ωn2 (1 2ζ2) + ωn2 (2ζ 2 − 1)2 + 1
{ 2 4 2
= ωn2 (1 − 2ζ ) + 4ζ − 4ζ + 2 }
STABILITY ANALYSIS OF CONTROL SYSTEMS 259
∴ ωc = ωn (1 − 2ζ 2 + 4ζ 4 − 4ζ 2 + 2)1 / 2
The negative sign under quadratic solution as well as while taking square root is not
taken into consideration because, ωc is a real and positive quantity.
Hence, the bandwidth of a second order system having non-zero magnitude at ω = 0 is
given by
B.W. = ωn (1 − 2ζ 2 + 4ζ 4 − 4ζ 2 + 2)1 / 2
The capacity of a feedback control system to reproduce input signal is indicated by its
bandwidth. Frequencies greater than the cut off frequency are attenuated, where as,
frequencies lower than the cut off frequency are transmitted.
A system with large bandwidth has lower rise time.
1 1
ω
and ∠ G(jω) H(jω) = tan ω tan
2
1 3ω
= tan for ω < 2
2 − ω2
1
3ω
= tan 180° for ω > 2
2 − ω2
(1) As ω → 0
G(j0) H(j0) → 25
and ∠ G(j0) H(j0) → 0°
(2) At ω = 2
50
G( j 2) H ( j 2) = = 11.78
2 2 2 2
2 +1 2 +2
1
3 2 1
and ∠ G( j 2) H ( j 2) = tan = tan ∞= 90°
2−2
260 LINEAR CONTROL SYSTEMS
(3) As ω → ∞
G(j∞) H(j∞) → 0
and ∠ G(jω) H(jω) → 180°
As per above calculations, general shape of the Nyquist plot is shown in Fig. 7.17.1.
As the given open-loop transfer function is type O, hence, the points ω = + 0 and ω = 0
are coincident.
Imj
Imj
w=+¥
w=+0
w=–¥ w=–0
Re 1 + j0 w=+¥ w=+0 Re
w=–0
w=–¥
Example 7.17.2. Using Nyquist criterion investigate the closed-loop stability of the
system whose open-loop transfer function is given below :
K( s + 1)
G(s) H(s) = ⋅
( s + 0.5 ) ( s − 2 )
Consider (i) K = 1.25, (ii) K = 2.5.
Also determine the limiting value of K for stability.
K ω2 + 12
G(jω) H(jω) =
ω2 + 0.52 ω2 + 22
STABILITY ANALYSIS OF CONTROL SYSTEMS 261
1 1 ω 1 ω
and ∠ G(jω) H(jω) = tan ω tan tan 180°*
0.5 −2
= tan 1 ω tan 1 2ω + tan 1 0.5ω 180°
1 1.5 ω 1
= tan tan 2ω 180°
1 − 0.5 ω2
⎧ 1.5 ω ⎫
⎪⎪ 1 − 0.5 ω2 − 2ω ⎪⎪
= tan 1
⎨ 1.5 ω ⎬ − 180°
⎪1 + × 2ω ⎪
⎩⎪ 1 − 0.5ω2 ⎭⎪
1 ω (ω2 − 0.5)
= tan − 180°
1 + 2.5ω2
(1) As ω → 0
G(j0) H(j0) → K
and ∠G(j0) H(j0) → 180°
(2) As ω = 0.5
K (0.5)2 + 12
G( j 0.5) H ( j 0.5) = = 0.66 K
(0.5)2 + (0.5)2 0.52 + 22
⎛ 1 ω ⎞
* The real part of term 1 / (jω 2) is ive. Hence the corresponding phase angle is ⎜ tan 180° ⎟
⎝ 2 ⎠
262 LINEAR CONTROL SYSTEMS
Example 7.17.3. The open-loop transfer function of a unity feedback control system is
given below :
(s + 0.25)
G(s) =
s 2 (s + 1) (s + 0.5)
Determine the closed-loop stability by applying Nyquist criterion.
(s + 0.25)
Solution. G(s) = 2
s (s + 1) (s + 0.5)
Put s = jω
( jω + 0.25)
G(jω) = 2
( jω) ( jω + 1) ( jω + 0.5)
ω2 + 0.252
G(jω) =
ω2 ω2 + 12 ω2 + 0.52
1
ω 1 ω 1 ω
and ∠ G(jω) = 180° + tan tan tan
0.25 1 0.5
= 180° + tan 1 4ω tan 1ω tan 1 2ω
3ω
= 180° + tan 1 4ω tan 1
1 − 2ω2
3ω
4ω −
= 180° + tan 1 1 − 2ω2
4ω ⋅ 3ω
1+
1 − 2ω2
1 ω (1 − 8ω2 )
= 180° + tan
1 + 10ω2
(1) As ω → 0
G(j0) → ∞
and ∠ G(j0) → 180°
(2) At ω = 1 / 8 = 0.125
0.1252 + 0.252
G( j 0.125) = = 5.3
0.1252 0.1252 + 12 0.1252 + 0.52
STABILITY ANALYSIS OF CONTROL SYSTEMS 263
K (1 + sT1 )
Solution. G(s) H(s) =
s2 (1 + sT2 )
Put s = jω
K (1 + jωT2 )
∴ G( jω) H( jω) =
( jω)2 (1 + jωT2 )
K 12 + ω2T12
G( jω) H( jω) =
ω2 12 + ω2T22
and ∠ G(jω) H(jω) = 180° + tan 1 ωT1 tan 1 ωT2
1
ω(T1 − T2 )
= 180° + tan
1 + ω2T1T2
(a) T1 > T2
(1) As ω → 0
G( j0) H( j0) → ∞
∠ G( j0) H( j0) → 180° + 0°
264 LINEAR CONTROL SYSTEMS
(2) As ω → ∞
G( j∞) H( j∞) → 0
∠ G( j∞) H( j∞) → 180° + 0°
For any value of 0 < ω < ∞, the mag- Imj
nitude G( jω) H( jω) is finite and since
T1 > T2, ∠ G( jω) H( jω) is placed in the 3rd
quadrant of G(s) H(s)-plane. Therefore,
for ω = 0 to ω = ∞ general shape of the
Nyquist plot is shown in Fig. 7.17.4 (a).
As the open-loop transfer function
w=–0 –1+ j0 w=–¥
is of type 2, the Nyquist plot is completed
w=+¥ Re
from ω = 0 to ω = + 0. Through an angle w=+0
2π ( 360°) as shown in Fig. 7.17.4 (a).
It is noted that the number of en-
circlements of the critical point ( 1 + j0)
G(s)H(s)-plane
the Nyquist plot is nil,
∴ N=0
P+ = 0
Fig. 7.17.4 (a) Nyquist plot T1 > T2 .
N = P+ Z+
∴ 0 = 0 Z+
∴ Z+ = 0
The closed-loop system is stable.
(b) T1 < T2
K 12 + ω2T12
G(jω) H(jω) =
ω2 12 + ω2T22
and ∠ G(jω) H(jω) = 180° tan 1 ωT1 + tan 1 ωT2
1
ω (T2 − T1 )
= 180° tan
1 + ω2T1T2
∵ T1 < T2
ω (T2 − T1 )
∠ G(jω) H(jω) = 180° tan 1
1 + ω2T1T2
(1) As ω → 0
G(j0) H(j0) → ∞
∠ G(j0) H(j0) → 180° 0°
(2) As ω → ∞
G(j∞) H(j∞) → 0
∠ G(j∞) H(j∞) → 180° 0°
Thus, for any value 0 < ω < ∞, the magnitude G(jω) H(jω) is finite and since T1 < T2,
∠ G(jω) H(jω) is placed in the second quadrant of G(s) H(s)-plane.
General shape of the Nyquist plot is shown in Fig. 7.17.4 (b). As the open-loop transfer
function is of type 2, the plot is completed from ω = 0 to ω = + 0 through an angle
2π ( 360°) as shown in Fig. 7.17.4 (b).
STABILITY ANALYSIS OF CONTROL SYSTEMS 265
Imj
w=+0
–1 + j0 w=+¥
w=–¥ Re
w=–0
G(s)H(s)-plane
(2) The Nyquist plot crosses the negative real axis at phase crossover frequency ω2.
Therefore, ∠ G(jω2) H(jω2) = 180°
∴ 1 0.08 ω22 = 0
1
or ω2 = rad./sec.
0.08
2.5 K
G(jω2) H(jω2) =
ω2 (0.4ω2 ) + 1 (0.2ω2 )2 + 1
2
2.5 K
=
1 0.16 0.04
+1 +1
0.08 0.08 0.08
2.5 2.5 K 2.5 K K
= = = =
1 4.5 / 0.08 56.25 3
3 1.5
0.08 Imj
Thus, the Nyquist plot crosses the negative w=–0
K
real axis of G(s) H(s)-plane at ⋅
3
(3) As ω → ∞
G(j∞) H(j∞) → 0
and ∠ G(j∞) H(j∞) → 270°
The open-loop transfer function is of type 1, K
w=¥
3
therefore, completion of the Nyquist plot for the
w=–¥ Re
variation from ω = 0 to ω = + 0 is effected through
an angle π ( 180°) as shown in Fig. 7.17.5. + 1 r/s w=–
Ö0.08
1
G.M. db = 20 log10 = 20 log10 3
( K / 3) K
= 20 log10 3 20 log10 K. Ans.
(b) K = 3 G(s)/H(s)-Plane
w=+0
G.M. = 20 log10 3 20 log10 3 = 0 db.
Since, G.M. = 0 db. Therefore P.M. = 0°. Ans. Fig. 7.17.5. Nyquist plot for Example 7.17.5.
15 (1 + 4 s)
Solution. G(s) H(s) = 2
(s + 100) (1 + 0.25 s)
Put s = jω
15 (1 + j 4ω)
G(jω) H(jω) = ...(1)
(100 − ω2 ) (1 + jω0.25)
STABILITY ANALYSIS OF CONTROL SYSTEMS 267
Rationalizing Eq. (1) and separating into real and s = + j¥ Imj
imaginary part following equation is obtained :
s = + j10
15 (1 + ω2 )
G(jω) H(jω) =
(100 − ω2 ) (1 + 0.625ω2 ) s = + j0
j 56.25 ω
+ ...(2) s = – j0 Re
(100 − ω2 ) (1 + 0.0625ω2 )
The open-loop transfer function has poles on imagi- s = – j10
nary axis
s = – j¥ s-plane
(s = − 100 = ± j 10),
Fig. 7.17.6 (a) Variation s in s-plane.
these poles are avoided from the Nyquist path in s-plane
as shown in Fig. 7.17.6 (a).
The real and imaginary parts of G(jω) H(jω) for some values of ω are calculated as
given in the following table :
ω 0 9 10 11 ∞
Re G(jω) H(jω) + 0.15 + 10.68 +∞ 10.18 0
The real part as well as imaginary part of G(jω) H(jω) are both positive for values of ω
just less than 10 and, therefore, at these frequencies the phase of G(jω) H(jω) is given by
⎡ + 56.25 × ω ⎤
∠G ( jω) H ( jω) = tan 1 ⎢ 2 ⎥
ω < 10 ⎣ + 15 (1 + ω ) ⎦
⎡ 56.25 × 10 ⎤
⎢ 1
= tan 2 ⎥ = 20.4°
⎣ 15 (1 + 10 ) ⎦
For values of ω just greater than 10 the real part and imaginary part of G(jω) H(jω)
are both negative and at these values of frequency the phase of G(jω) H(jω) is given by
⎡ − 56.25 × ω ⎤
∠G( jω) H ( jω) = tan 1 ⎢ 3 ⎥
ω → > 10 ⎣ − 15 (1 + ω ) ⎦
⎡ − 56.25 × 10 ⎤
= tan 1
⎢ 2 ⎥ = 159.6°
⎣ − 15 (1 + 10 ) ⎦
Thus at ω = 10 the phase of G(jω) H(jω) suddenly changes through 159.6 (+ 20.4) =
180° and the magnitude being infinite. Therefore as per above analysis and calculated
values of G(jω) H(jω) at various of ω, the Nyquist plot takes the shape as shown in
Fig. 7.7.6 (b). As the type of the system is 1 the points ω = 0 and ω = + 0 are coincident.
The point ( 1 + j0) is not encircled by the Nyquist plot and P+ = 0
∴ N = P+ Z+
∴ 0 = 0 Z+ ∴ Z+ = 0
As Z+ = 0, the number of roots of the characteristics equation of the closed-loop system
having positive real part is nil indicating that the system is stable.
268 LINEAR CONTROL SYSTEMS
Imj
w ³ –10
w £ 10
°
2 0.4
w=+0
– 1 + j0 w=–0 Re
0.15
9 .6°
– 15 w £ –10
w ³ 10
Example 7.17.7. Draw the Nyquist plot for the open-loop transfer function given
below and comment on the closed-loop stability.
ω 0 0.816 1 2 2 ∞
Imj
w=–0
w=+2
w=+
– 0.816
w=+¥
w=–¥ Re
– 1 + j0
w=–2
w=+0
or 2 = (0 Z+)
∴ Z+ = 2
–0
Thus, the closed-loop system has 2 roots w=
having positive real part and the system is un-
stable. Fig. 7.17.8. Nyquist plot for example 7.17.8.
Solution. Put s = jω
32
∴ G(jω) H(jω) =
jω ( jω + 6)3
32
G(jω) H(jω) =
ω ( ω2 + 6)3
put ω = 2
32 32 32
∴ G(j 2 ) H(j 2 ) = 3
= 3
= =1
2(2 + 6) 2( 8) 2 8 8 8
∴ gain crossover frequency ω1 = 2 rad./sec. Ans.
ω
∠ G(jω) H(jω) = 90° 3 tan 1
6
At the phase crossover frequency ω2,
∠ G(jω2) H(jω2) = 180°.
ω2
∴ 180° = 90 tan 1
6
1 ω2
or tan = 30°
6
ω2
∴ = tan 30°
6
1
∴ ω2 = ( 6 ) tan 30° = 6× = 2 rad./sec.
3
STABILITY ANALYSIS OF CONTROL SYSTEMS 271
∵ ω1 = ω2
G.M. = 0 db. Ans.
and P.M. = 0°
Since, ω1 = ω2, the system is marginally stable.
Example 7.17.10. Determine the value of K for a unity feedback control system
having open-loop transfer function
K
G(s) H(s) = , such that
s (s + 2) (s + 4)
(i) Gain margin = 20 db, and (ii) Phase margin = 60°.
6Kω jK (− ω2 + 8)
= −
ω (ω2 + 4) (ω2 + 16) ω (ω2 + 4) (ω2 + 16)
At the point of intersection of the Nyquist plot with the real axis the imaginary part of
the above expression is zero. Therefore, the phase crossover frequency ω2 is determined
below :
K (− ω22 + 8)
− =0
ω22 (ω22 + 4) (ω22 + 16)
∴ ω2 = 8 rad/sec
The intersection of the Nyquist plot with the real axis occurs at ω = ω2 = 8
6K 8 K
∴ G(j 8 ) H(j 8 ) = 2 2
=−
8( 8 + 4) ( 8 + 16) 48
To have a gain margin of 20 db i.e. a = 0.1
K
= 0.1
48
∴ K = 4.8. Ans.
K
(ii) G(jω) H(jω) =
jω ( jω + 2) ( jω + 4)
272 LINEAR CONTROL SYSTEMS
⎛ ω⎞ ⎛ ω⎞
1
∴ ∠G(jω) H(jω) = ⎜ ⎟ tan 1 ⎜ ⎟
90° tan
⎝2⎠ ⎝4⎠
If the gain crossover frequency is ω1, the phase margin is :
Phase margin = ∠G(jω1) H(jω1) + 180°
1⎛
1 ⎛ ω1 ⎞ ω1 ⎞
or 30° = tan ⎜ ⎟ + tan ⎜ ⎟
⎝ 2 ⎠ ⎝ 4 ⎠
ω1 ω1 6ω1
+
6ω1
or tan 30° = 2 4 = 8 2 =
1−
ω1 ω1 ω 8 − ω12
⋅ 1− 1
2 4 8
1 6ω1
or =
3 8 − ω12
or ω12 + (6 3 )ω1 8=0
Example 7.17.11. Sketch Nyquist plot for a unity feedback control system having
given
Ks 3
G(s) =
( s + 1) ( s + 2 )
and determine the stability condition.
STABILITY ANALYSIS OF CONTROL SYSTEMS 273
Solution. Put s = jω
K ( jω)3 − jK ω3
∴ G(jω) = or G(jω) =
( jω + 1) ( jω + 2) (2 − ω2 ) + j 3ω
− jK ω3 [(2 − ω3 ) − j3ω]
or G(jω) =
(2 − ω2 )2 + (3ω)2
− 3 K ω4 jK ω3 (2 − ω2 )
or G(jω) = 2 2 2
−
(2 − ω ) + (3ω) (2 − ω2 )2 + (3ω)2
(a) G( jω) : Re (G(j0) → 0 Imj (Gj0) → + j0
ω→0
(b) The intersection of G(jω) plot with ive real axis is obtained by equating imaginary
part of G(jω) to zero and solving for ω. Therefore,
Kω2 (2 − ω2 )
=0
(2 − ω2 )2 + (3ω)2
or (2 ω2) = 0
The intersection with ive real axis occurs at :
ω = ± 2 rad/sec
The intersection is obtained by substituting ω = 2 in real part of G(jω), i.e.
4
− 3 K ( 2) − 3K × 2 × 2
G(j 2 ) = =
2 2
(2 − 2 ) + (3 2) 2 (2 2)2 + 3 × 3 × 2
2
or G(j 2 ) = K
3
(c) G( jω) : Re G(j∞) → 3K
ω→0
Imj G(j∞) → + j ∞
As per steps (a), (b) and (c), the general shape of the Nyquist plot is drawn as shown in
Fig. 7.17.9.
K ( jω)3 K ( jω)3
G( jω) = = = K jω
ω→0 ( jω + 1) ( jω + 2) ( jω) ( jω)
or G( jω) = K j∞ = ∞ ∠90°
ω→0
The closing of Nyquist plot from ω = ∞ to ω = ∞ is done as explained below :
In s-plane the RHS region is closed from s = j∞ to s j∞ through a semicircle of infinite
radius in the clockwise direction, hence the corresponding points in G(s)-plane i.e. G(j∞) to
G( j∞) are closed through a semicircle of infinite radius to account for effective numerator
term i.e. K j∞ of G(jω) as ω → ∞.
The completed Nyquist plot is shown in Fig. 7.17.9.
It is given that, the number of poles of G(s) having positive real part is nil i.e. P+ = 0.
The encirclements of critical point ( 1 + j0) are determined below :
274 LINEAR CONTROL SYSTEMS
Imj
w=+¥
Imj
+j¥
+j0 w = Ö2
– 1 + j0
w=–0
Re w=+0 Re
–j0 – 2 K
3
–j¥
s-plane
w=–¥
3
(1) If K <
2
The critical point ( 1 + j0) lies outside the Nyquist plot, hence N = 0
∴ N = P+ Z+
or 0 = 0 Z+
∴ Z+ = Nil
The system is stable
3
(2) If K >
2
The critical point ( 1 + j0) and encircled twice in the clockwise direction by the
Nyquist plot, hence N = 2.
∴ N = P+ Z+
or 2 = 0 + Z+ ∴ Z+ = 2
3
The system is unstable. For stability K ≤ . Ans.
2
Example 7.17.12. Sketch the Nyquist plot and examine the closed-loop stability of a
control system having open-loop transfer function given below :
Ks ( 1 + 2s )
G(s) H(s) = ⋅
s 3 + 4s + 8
Solution. For the application of Nyquist plot it is required to know the number of
poles (P+) of G(s) H(s) which have positive real part. For the transfer function :
STABILITY ANALYSIS OF CONTROL SYSTEMS 275
Ks (1 + 2s)
G(s) H(s) =
s3 + 4 s + 8
P+ can be determined by applying Routh-Hurwitz criterion to following equation :
s3 + 4s + 8 = 0
s3 1 4
s2 (0) ε 8
4ε − 8
s1
ε
s0 8
4ε − 8
As ε → 0 the sign of → hence, there are two sign changes in the first column of
ε
Routh s table and as such P+ = 2.
Put s = jω in the expression for G(s) H(s), thus
K jω (1 + j 2ω)
G(jω) H(jω) =
( jω)3 + j 4ω + 8
K ( jω − 2ω2 ) K ( jω − 2ω2 )
or G(jω) H(jω) = =
− jω3 + 4 jω + 8 8 + jω (4 − ω2 )
K ( jω − 2ω2 ) [8 jω (4 − ω2 )]
or G(jω) H(jω) =
[8 + jω (4 − ω2 )][8 − jω (4 − ω2 )]
− K ω2 (ω2 + 12) + jωK (8 + 10ω2 − 2ω4 )
=
64 + ω2 (4 − ω2 )2
Separating into real and imaginary part :
2 2 2 4
G(jω) H(jω) = − K ω (ω + 12) + jωK (8 + 10ω − 2ω )
64 + ω2 (4 − ω2 )2 64 + ω2 (4 − ω2 )2
(a) G( jω) H ( jω) : Re G(j0) → 0
ω→0
Imj G(j0) → j0 Imj
(b) The intersection of the Nyquist plot with
ive real axis occurs at a value of ω given by solving
ω K (8 + 10ω2 − 2ω4 )
=0 w= –0
64 + ω2 (4 − ω2 )2 – 1 + j0 w=0
or 2ω4 + 10ω2 + 8 = 0 w=¥ Re
w=–¥
or (ω2)2 + 5 (ω2) + 4 = 0 – 1.25 K
At ω = + j2.38 rad/sec
− K (2.38)2 (2.382 + 12) − K (5.7) (5.7 + 12)
G(j2.38) = 2 2 2
=
64 + (2.38) (4 − 2.38 ) 64 + (5.7) (4 − 5.7)2
100.9 K
= = 1.25 K
80.4
The intersection of the Nyquist plot with ive real axis is thus, 1.25 K.
(c) G( jω) H ( jω) : Re G(j∞) → 0
ω→∞
Imj G(j∞) → j0
From the data determined above the Nyquist plot is drawn as shown in Fig. 7.17.10.
(1) It is noted that, for 1.25 K > 1, the critical point ( 1 + j0) is encircled twice in the
anticlockwise direction by the Nyquist plot, hence N = 2, P+ = 2.
∴ N = P+ Z+
or 2 = 2 Z+ ∴ Z+ = Nil
and the closed-loop system is stable. Ans.
(2) However, for 1.25 K < 1
N=0
P+ = 2 ∴ N = P+ Z+
or 0 = 2 Z+ ∴ Z+ = 2
and the closed-loop system is unstable. Ans.
The condition for stability is, thus :
1.25 K ≥ 1 or K ≥ 0.8. Ans.
Example 7.17.13. Determine the critical value of K for the stability of a unity feed-
back control system whose open-loop transfer function is given by
Ke− 0.5s
G(s) = ⋅
( s + 1)
Use Nyquist plot method.
Ke− 0.5 s
Solution. G(s) = Imj
(s + 1)
Put s = jω
w = 3.67 r/s
Ke− 0.5 jω
∴ G(jω) = w=+0
( jω + 1) w=+¥ Re
∵ e j0.5ω =1
K K K
∴ = 3.68
jω + 1 2
ω +1 2
Fig. 7.17.11. Nyquist plot (from ω = + 0
and ∠ G(jω) = tan 1ω 0.5 ω to ω = + ∞) for example 7.17.13.
= (tan 1 ω + 0.5 ω)
STABILITY ANALYSIS OF CONTROL SYSTEMS 277
(1) As ω → 0
G(jω) → K
and ∠ G(jω) → 0°
(2) As ω → ∞
G(jω) → 0
and ∠ G(jω) → ∞
As ω varies from ω = 0 to ω = ∞, G(jω) decreases and ∠ G(jω) changes continuously
from 0° to ∞. The Nyquist plot takes the shape of a spiral intersecting real and imaginary
axis many times. The plot for the variation from ω = 0 to ω = ∞ is shown in Fig. 7.17.11.
(3) The intersection points of the plot with the negative real axis of G(s) H(s)-plane are
determined using the relation given below :
(tan 1 ω + 0.5 ω) = 180 (2k + 1)
k = 0, 1, 2, ...
For ω > 0, at the first instant, the said intersection is given by considering k = 0. The
frequency at this intersection point is the phase crossover frequency ω2. Thus,
(tan 1 ω2 0.5 ω2) = 180°
Using trial and error, this equation is satisfied at ω2 = 3.67 rad./sec.
For stability G(jω2) < 1.
Ke− j 0.5 ω2
G(jω2) =
jω2 + 1
∵ e− j 0.5 ω2 =1
K
∴ G(jω2) =
ω22 + 12
ω2 = 3.67 rad./sec.
K K
∴ G(j 3.67) = =
(0.367) + 12 2 3.68
For stability, the critical point ( 1 + j0) should not be encircled by the Nyquist plot.
K
∴ <1
3.68
or K < 3.68. Ans.
Solution. Put s = jω
3ω 1
= 11.46ω 90° tan
; (ω < 2 )
2 − ω2
Intersection of the Nyquist plot with the negative real axis of G(s) H(s)-plane is deter-
mined using following relation
∠ G(jω) H(jω) = 180 (2k + 1)
k = 0, 1, 2, ...
At the first instant the Nyquist plot intersects the negative real axis of G(s) H(s)-plane
for k = 0 and the frequency at the intersection point is ω2. Therefore,
∠ G(jω2) H(jω2) = 180°
1 3ω2
∴ 11.46 ω2 90° tan = 180°
2 − ω22
3ω2
1
or 11.46 ω2 tan = 90°
2 − ω22
The equation above is satisfied at ω2 = 1.114 rad./sec. (phase crossover frequency)
K
G(jω2) H(jω2) =
jω2 ( jω2 + 1) ( jω2 + 2)
(Note: e j0.2ω = 1)
K
=
ω2 ω22 + 12 ω22 + 22
ω2 = 1.114 rad./sec.
∴ G(j1.114) H(j1.114) = K K
=
1.114 1.114 + 1 2 2 2
1.114 + 2 2 3.86
The general shape of the Nyquist plot for ω = 0 to ω = ∞ is drawn as shown in
Fig. 7.17.12.
Imj
w=¥
Re
w = 1.114 r/s
–K 3
3.86
G(s)H(s)-plane
w=0
Example 7.17.15. Determine the value of T for stability limit for a unity feedback
control system having open-loop transfer function as
sT ⋅ 10
G(s) = e .
s ( 0.1 s + 1)
10
∴ G(jω1) = =1
jω1 ( j0.1ω + 1)
10
or =1
ω1 (0.1ω1 )2 + 12
∴ 0.01 ω14 + ω12 − 100 = 0
− 1 ± 12 − 4 × 0.01 (− 100)
Solving ω12 =
2 × 0.01
= ( 0.5 ± 1.115) × 100
Considering +ive value
ω12 = 61.5
∴ ω1= 61.5 = 7.84 rad./sec.
Incorporating the time delay element e sT, the magnitude of G1(jω) is not affected but
the phase angle is modified as below :
∠ G1 (j ω1) = 90° tan 1 0.1 ω1
= 90° tan 1 0.1 × 7.84
= 90° 38.09° = 128.09°
P.M. = 180° + ∠ G1(jω1)
= 180° 128.09° = 51.81°
If the phase angle ( ω1T × 180/π) provided by the time delay element equals 51.81°,
then the limiting value of T for stability is determined as below :
− ω1T × 180
= 51.81°
π
280 LINEAR CONTROL SYSTEMS
− 7.84T × 180
or = 51.81°
π
∴ T = 0.115 sec. Ans.
1
Solution. (a) Let G1(s) =
s (0.5 s + 1) (0.2 s + 1)
Put s = jω
1
∴ G1(jω) =
j ω ( j 0.5 ω + 1) ( j 0.2 ω + 1)
To determine gain crossover frequency ω1 put G(jω1) = 1
1
∴ =1
jω1 ( j 0.5 ω1 + 1) ( j 0.2 ω1 + 1)
1
∴ =1
ω1 (0.5 ω1 )2 + 12 (0.2 ω1 )2 + 12
1
or =1
ω12 (0.25ω12 + 1) (0.04ω12 + 1)
∴ ω12 (0.25ω12 + 1) (0.04ω12 + 1) 1=0
put ω12 = x
∴ x (0.25 x + 1) (0.04 x + 1) 1 = 0
Using trial and error method above equation is nearly satisfied at x = 0.8062 and
acceptable being +ive.
∴ ω1 = 0.8062 = 0.897 rad./sec.
Phase margin = 180° ∠ G1(jω1)
= 180° tan 1 0.5ω1 tan 1 0.2ω1
= 180° tan 1 0.5 × 0.897 tan 1 0.2 × 0.897
= 180° 24.156° 10.17° = 55.6°. Ans.
The closed-loop system is stable with a phase margin 55.6°.
(b) To make the system marginally stable, the phase angle to be provided by the time
delay element is 55.6°.
ω1T × 180
∴ = 55.6
π
55.6 × π
∴ T=
ω1 × 180
ω1 = 0.897 rad./sec.
55.6 × π
∴ T= = 1.08 sec. Ans.
0.897 × 180
Example 7.17.17. The open-loop transfer function of a unity feedback control system
is given below :
10Ke− sT
G(s) =
s ( s + 1) ( s + 7 )
(a) For K = 1, determine T such that the system is marginally stable.
(b) For T = 1 sec., determine the limiting value of K such that the system is marginally
stable.
ω1T × 180
∴ 141.13 = 180
π
∵ ω1 = 1 rad./sec.
1 × T × 180
∴ 141.13 = 180
π
(180 − 141.13) π
∴ T= = 0.678 sec. Ans.
1 × 180
(b) T = 1 sec.
10 Ke− s.1
∴ G(s) =
s (s + 1) (s + 7)
Put s = jω
10 Ke− jω
∴ G(jω) =
jω ( jω + 1) ( jω + 7)
ω × 180 1 ω 1
ω
∠ G(jω) = 90° tan tan
π 1 7
ω
ω+
ω × 180 1 7
= 90° tan ω
π 1 − ω⋅
7
ω × 180 1
8ω ;
= 90° tan ω< 7
π 7 − ω2
At the phase crossover frequency ω2, ∠ G(jω2) = 180°
ω2 × 180 1
8ω2
∴ 90° tan = 180°
π 7 − ω22
8ω2 1
or = 57.3 ω2 tan= 90°
7 − ω22
Solving by trial-error method : ω2 = 0.79 rad./sec.
10 K
G(jω2) =
jω2 ( jω2 + 1) ( jω2 + 7)
10 K
=
ω2 ω22 + 12 ω22 + 72
ω2 = 0.79 rad./sec.
10 K K
∴ G(j0.792) = =
0.79 (0.79) + 1 2 2
(0.79) + 72 2 0.708
K
For stability < 1; K < 0.708
0.708
For marginally stability : K = 0.708. Ans.
STABILITY ANALYSIS OF CONTROL SYSTEMS 283
Example 7.17.18. The open-loop transfer function of a unity feedback control system
is given by
G(s) = K
s ( 0.5 s + 1) ( 0.25 s + 1)
assuming K = 1 sketch the gain phase plot. Determine the value of K such that
(a) G.M. = 8 db and (b) P.M. = 20°.
30
K=1
∴ For G.M. = 8 db : 0
8 db
8.6 db
15 db
GM
On comparing (4) and (3), it is noted that the zeros (roots of the characteristic equation
1
z′1, z′2, ... z′m) of the function 1 + G(s) H(s) and 1 + are same.
G(s) H (s)
Therefore, as per Nyquist s method, presence of the zeros of 1 + G(s) H(s) in a specified
region of s-plane can be identified from the plot of 1/[G(s) H(s)].
The plot of 1/[G(s) H(s)] is known as inverse Nyquist plot wherein the independent
variable s is varied on the Nyquist contour.
The stability criterion for the function 1/[G(s) H(s)] is the same as that for [G(s) H(s)].
The inverse Nyquist plot method is applied to following examples to determine closed-
loop stability.
Example 7.17.19. The open-loop transfer function of a unity feedback control system
is given below :
K
G(s) =
s ( sT + 1)
Sketch the inverse Nyquist plot and determine the closed-loop stability.
1 s (sT + 1)
Solution. =
G (s) K
The independent variable s is varied in s-plane as per Nyquist path from s = j0 to
s = j∞, from s = j∞ to s = j∞ through R → ∞ by a semicircle to cover the entire RH-side, from
s = j∞ to s = j0 and from s = j0 to s = + j0 along a semicircle of radius δ → 0 such that any
pole of 1/[G(s) H(s)] at the origin of s-plane is excluded.
STABILITY ANALYSIS OF CONTROL SYSTEMS 285
Put s = jω
1 jω ( jωT + 1)
∴ =
G( jω) K
1 ω ω2 T 2 + 12
=
G( jω) K
and ∠1/G(jω) = 90° + tan 1 ωT
(1) As ω → 0
1
→0
G ( j 0)
∠ 1/G(j0) → 90°
(2) As ω → ∞
1
→∞
G( j∞)
∠ 1/G(j∞) → 180°
Imj
Imj
s=j+¥
R ¥
s = j0 w= ¥ w=0
–1+j0
s 0 Re w =– ¥ w=0 Re
s=j–0
s=j–¥
1
– plane
G(s)
s-plane
π π
In the s-plane θ changes from to rad. as ω changes from ∞ to ∞. Thus, the
2 2
π ⎛π⎞
change in θ is − − ⎜ ⎟ = π rad. The corresponding change in ∠ 1/G(jω) is 2 ( π) = 2π rad.
2 ⎝2⎠
Hence, the inverse Nyquist plot is completed by a circular path of radius → ∞ through
an angle 2π rad. ( 360°) as shown in Fig. 7.17.15.
P+ : Number of poles 1/G(s) having positive real part = Nil.
As per the Nyquist plot, the critical point ( 1 + j0) is not encircled.
286 LINEAR CONTROL SYSTEMS
∴ N=0
N = P+ Z+
∴ 0 = 0 Z+
Z+ = 0. The closed-loop system is stable. Ans.
Example 7.17.20. Apply Nyquist criterion to determine the stability of a unity feed-
back control system having given :
K
G(s) = .
s ( sT − 1)
1 s (sT − 1)
Solution. =
G (s) K
Put s = jω
1 jω ( jωT − 1)
∴ =
G ( jω) K
1 ω ω2T 2 + 12
=
G( jω) K
ωT 1
and ∠ 1/G( jω) = 90° + tan 180°
−1
= 90° tan 1 ωT 180°
(Note : The real part of term (jωT 1) is ive. Therefore, Imj
1 ωT
the phase angle is tan 180°)
−1
(1) As ω → 0 w=–¥
–1 + j0 w = –0
1 w=0 Re
→0 w=¥
G ( j 0)
∠ 1/G(j0) → 90° 1
– plane
G(s)
(2) As ω → ∞
1
→∞ Fig. 7.17.16. Inverse Nyquist plot for
G ( j∞)
example 7.17.20.
∠ 1/G(j∞) → 180°
(3) Since, the number of poles of 1/G(s) at the origin of s-plane is nil. Therefore, points
corresponding to ω = 0 and ω = 0 are coincident in 1/G(s)-plane.
1
(4) → ∞ ej2θ
G ( jω) ω→∞
π π
In the s-plane θ changes from to rad. as ω varied from ∞ to ∞. Thus, the
2 2
π ⎛π⎞
change in θ in s-plane is ⎜ ⎟ = π rad. ( 180°).
2 ⎝2⎠
STABILITY ANALYSIS OF CONTROL SYSTEMS 287
The corresponding change in ∠ 1/G(jω) is 2 ( π) = 2π rad ( 360°). Hence, the inverse
Nyquist plot is completed by a circular path of radius → ∞ through an angle 2π rad. ( 360°)
as shown in Fig. 7.17.16.
P+ : Number of poles of 1/G(s) having positive real part = Nil
As per Nyquist plot, the critical point ( 1 + j0) is encircled twice in the clockwise
direction.
∴ N= 2
N = P+ Z+
∴ 2 = 0 Z+
Z+ = 2. The closed-loop system is unstable.
C( jω) 1 y
or ∠ = tan
R( jω) x +1
288 LINEAR CONTROL SYSTEMS
∴ φ= tan 1 y
x +1
The equation above is of constant phase angle.
1
The constant circles and constant phase angle loci are shown in Fig. 7.17.18.
M
1
The point of intersection of 1/G(jω) plot and circle/constant phase angle locus give
M
1 1
the value of /φ at the corresponding frequency ω. The circle tangential to 1/G(jω) plot
M M
1
gives the value of and the frequency at the tangential point is ωr.
Mr
– f = 90°
Imj
– f = 135° – f = 45°
Imj
w – f = 180°
1 – = 0° Re
plot –1
G(jw)
x + jy
1 circle
M
Re
– f = 225° – f = 315°
– f = 270°
Fig. 7.17.17. 1/G( jω) plot. Fig. 7.17.18. 1/M circles and constant phase angle loci.
1
1
(2) Draw the line OAB such that φ = sin ⋅
Mr
STABILITY ANALYSIS OF CONTROL SYSTEMS 289
(3) Draw a circle such that it is Imj
tangent to both 1/G(jω) plot and the line w3
w2
OAB. The centre of this circle is located w1
B
at C. wr A
(4) If the circle drawn is to be 1
w
1 Mr
circle, then its centre must be ( 1 0 Re
Mr C –1 1
f = sin
Mr
+ j0). Therefore, distance OC = 1, for 1
Circle
this is to be possible, the new gain K′ = Mr
oc. Thus, the original gain K is to be
K′
changed by a factor ⋅
K
Fig. 7.17.19. Gain adjustment using 1/M circle.
1
⎡ 2ζωn ω ⎤
tan ⎢ 2 2 ⎥
...(7.50)
⎣⎢ (ωn − ω ) ⎦⎥
The Bode plot is a graph obtained from Eqs. (7.49) and (7.50) consisting of two parts as
follows :
(i) Magnitude of G(jω) H(jω) in decibel, [i.e. 20 log10 G(jω) H(jω) ] versus log10 ω.
(ii) Phase angle ∠G(jω) H(jω) versus log10 ω.
For plotting magnitude versus log10 ω plot each term of Eq. (7.49) is considered sepa-
rately and graphs are drawn. To obtain final plot the contributions due to each term are
added as described below :
0.1 1 10 100 w
60 db + 180°
Phase Angle
+ 90°
Magnitude
40 db
0°
20 log10 K
20 db – 90°
– 180°
0 db
0.1 1 10 100 w
(a) (b)
Fig. 7.18.1. Bode plot for gain term K.
1
7.18.3. Graphs for the term
(jω)N
The magnitude of the term 1/(jω)N in decibel is given by
1
20 log10 = 20 N log10 ω ...(7.52)
( jω) N
1
The phase angle is given by ∠ = 90° N ...(7.53)
( jω) N
STABILITY ANALYSIS OF CONTROL SYSTEMS 291
In view of Eqs. (7.52) and (7.53) the graphs are shown in Fig. 7.18.2. The graph for the
magnitude versus log10 ω is a straight line having a slope of 20 N db/decade.
+ 60 db
– 60 db/d
+ 40 db
– 40 db/d
Magnitude + 20 db – 20 db/d
0 db
0.1 1 10 100 w
– 20 db N=1
– 40 db N=2
N=3
0.1 1 10 100 w
0°
Phase Angle
– 90° N=1
– 180° N=2
– 270° N=3
*On logarithmic scale a decade means a ratio of 10. Sometimes the term octave is used to represent a
ratio of 2 on logarithmic scale. The relation between octave and decade is obtained below
log10 (10) 1
Octave = = decade
log10 (2) 0.301
∴ ± 20 N db/decade = ± 20 N × 0.30 = ± 6 N db/octave.
292 LINEAR CONTROL SYSTEMS
20 log10 ω + 20 log10 ⎛⎜ 1 ⎞⎟
= ...(7.59)
⎝T ⎠
In view of Eqs. (7.58) and (7.59) graph for case (i) lies on 0 db axis whereas for case (ii)
the graph has a slope of 20 db/decade. These two graphs intersect on 0 db axis at a point
determined by equating R.H.S. of Eq. (7.59) to zero.
∴ 0= 20 log10 ω + 20 log10 ⎛⎜ 1 ⎞⎟
⎝T ⎠
1
or 20 log10 ω = 20 log10 ⎛⎜ 1 ⎞⎟ , ∴ ω=
⎝T ⎠ T
1
Hence, the two graphs intersect on 0 db axis at ω = . The graphs for cases (i) and (ii)
T
are shown in Fig. 7.18.4 (a).
The phase angle for the term 1/(1 + jωT) is given by
φ= ⎛ ωT ⎞
tan 1
⎜ ⎟
⎝ 1 ⎠
(i) At very low frequencies ωT is very small
∴ φ ≅ tan 1 (0)
or φ ≅ 0°
1
(ii) At ω=
T
⎛1 ⎞
⎜ ⋅T ⎟ =
φ= tan 1 tan 1 (1)
⎝ T ⎠
or φ = 45°
(iii) At very high frequencies ω T is very large
∴ φ ≅ tan 1 (∞)
or φ ≅ 90°
The graph for phase angle is shown in Fig. 7.18.4 (b).
The magnitude versus log10 ω graphs shown in Figs. 7.18.3 and 7.18.4 for the terms
(1 + jωT) and 1/(1 + jωT) are asymptotic plots. A comparison of asymptotic plot and exact plot
is shown in Fig. 7.18.5. The exact plot is shown dotted. The exact and asymptotic plots match
each other at frequencies lower and higher than ω = 1/T. The frequency ω = 1/T is called
corner frequency. The maximum error between the exact and asymptotic plot occurs at
corner frequency, this error is calculated below :
From asymptotic plot it is observed that the magnitude at ω = 1/T is 0 db.
From exact plot the magnitude in decibel for the term (1 + jωT) at ω = 1/T is
⎛ 1 ⎞ 1
20 log10 ⎜ 1 + j ⋅ T ⎟ = 20 log10 1 + 2 ⋅ T 2 = 20 log10 2 = 3 db.
⎝ T ⎠ T
294 LINEAR CONTROL SYSTEMS
1 1
w= w= w
0.1 1 T 10 100 w 0.1 1 T 10 100
0 db 0°
Phase Angle
–2
– 20 db 0d – 45°
Magnitude
b/d
– 40 db – 90°
– 60 db
(a) (b)
Fig. 7.18.4. Bode plot for the term 1/(1 + jωT).
60 db
40 db
)
+ j wT b/d
(1 0d
20 db +2
Magnitude
w
0 db
0.1 1 10 100
1 –2
– 20 db w= 0d
T b/d
(1 1
– 40 db +j
wT
)
– 60 db
Fig. 7.18.5. Exact and asymptotic (approximate) bode plots for the terms (1 + jωT) and 1/(1 + jωT).
Thus the maximum error due to asymptotic plot of the term (1 + jωT) is 3 db. Similarly
the maximum error for the term 1/(1 + jωT) is 3 db. An error of ± 3 db is permissible.
For multiple first order terms such as (1 + jωT)±N, the magnitude in decibel is given by
= ± 20 N log10 2 = ± 3 N db.
STABILITY ANALYSIS OF CONTROL SYSTEMS 295
Thus for transfer function where multiple first order terms are involved the error
between the asymptotic plot and exact plot is appreciable (± 3 N db) and the correction to the
magnitude has to be applied at corner frequency occurring due to such multiple terms.
Further, if corner frequencies are very
near to each other the error between the as-
ymptotic plot and exact plot is appreciable and
correction to the asymptotic plot have also to be
applied in such cases.
The corrections are applied by referring
to the error curve for the term (1 + jω)± 1 shown
in Fig. 7.18.6. The ± 3 db point is located at ωT
= 1 and ± 1 db points are located at 0.5 ωT and
2ωT respectively, the points thus obtained are
joined by a smooth curve. Fig. 7.18.6. Error curve for term (1 + jωT)± 1.
In the case of multiple first order terms the phase angle is also multiplied by ±N,
therefore, at ω =/T
∠(1 + jωT)±N = ± 45 N°
and at high frequencies (i.e. ω T >> 1)
∠ (1 + jωT)±N = ± 90 N°
K
20 log10 G(jω) H(jω) = 20 log10 = 20 log10 K 20 log10 (1)
( jω)°
= 20 log10 K ...(7.61)
296 LINEAR CONTROL SYSTEMS
zero. 40 db
0d
b /d
∴ 0 = 20 log10 K 20 log10 ω
20 db
or 20 log10 K = 20 log10 ω
∴ ω = K. 0 db
0.1 1 10 w = K 100 w
Hence, the graph intersects 0 db axis at
ω = K. The graph for Eq. (7.62) is shown in Fig. Fig. 7.18.8. Initial part of Bode plot
7.18.8. for type 1 system.
(iii) Type 2 system (N = 2). For a type 2 system the equation for the initial part of the
Bode plot is
N 20 N db/decade ω = K1/N
In the case of zeros of the open-loop transfer function G(s) H(s) located at the origin of
s-plane :
K s N (sT1 + 1) (sT2 + 1) ...
G(s) H(s) =
(sTa + 1) (sTb + 1) ...
N : the number of zeros of G(s) H(s) located at the origin of s-plane.
For the frequencies lower than the lowest corner frequency the sinusoidal form of
open-loop transfer function takes the form as below :
K ( jω) N
G(jω) H(jω) =
1
It can be shown that the initial slope of the Bode plot is +20 Ndb/decade intersecting
1
0-db axis at ω = ⋅
( K )1 / N
7.18.7. Determination of static error coefficients from initial slope of Bode plot
In sinusoidal form static error coefficients Kp, Kv and Ka described in section 6.7.1 can be
written as
For type 0 system : Kp = Lim G(jω) H(jω) = K ...(7.64)
ω→0
In view of Eqs. (7.64), (7.65), (7.66) and initial slope of Bode plot for type 0, type 1, and
type 2 system respectively. Kp, Kv and Ka can be determined as follows :
Kp equals the value of K as determined from the initial part of log magnitude plot of a
type 0 system.
Kv is given by the value of ω at which the initial slope ( 20 db/decade) of log magni-
tude plot of a type 1 system intersects with 0 db axis.
298 LINEAR CONTROL SYSTEMS
ω2n 1
20 log10 = 20 log10
[(ω2n − ω2 ) + j 2ζ ωn ω] ⎛ ω 2 ⎞ ⎛ ω ⎞
⎜⎜ 1 − 2 ⎟⎟ + j 2 ζ ⎜ ⎟
⎝ ωn ⎠ ⎝ ωn ⎠
⎡ 2 2⎤
1 ⎢ ⎛ ω2 ⎞ ⎛ ω ⎞ ⎥
= 20 log10 = 20 log10 ⎢ ⎜⎜ 1 − 2 ⎟⎟ + ⎜ 2ζ ⎟ ...(7.67)
2 2 ⎝ ωn ⎠ ⎝ ωn ⎠ ⎥
⎛ ω2 ⎞ ⎛ ω ⎞ ⎣⎢ ⎦⎥
⎜⎜ 1 − 2 ⎟⎟ + ⎜ 2ζ ⎟
⎝ ωn ⎠ ⎝ ωn ⎠
ω2 ω
The terms and in Eq. (7.67) can be neglected, therefore,
ω2n ωn
ω2n
20 log10 20 log10 1
[(ω2n − ω2 ) + j 2ζ ωn ω]
20 log10 (1) = 0 db ...(7.68)
(ii) ω >> ωn (very high frequencies than ωn)
2 2
⎛ ω ⎞ ⎛ ω2 ⎞
The terms 1 and ⎜ 2ζ ⎟ can be neglected in comparison with ⎜⎜ 2 ⎟⎟ in Eq. (7.67),
⎝ ωn ⎠ ⎝ ωn ⎠
therefore,
ω2n
20 log10
[(ω2n − ω2 ) + j 2ζ ωn ω]
2
⎛ ω2 ⎞
20 log10 ⎛⎜ ω ⎞⎟
2
20 log10 ⎜⎜ − 2 ⎟⎟
⎝ ωn ⎠ ⎜ ω2 ⎟
⎝ n⎠
40 log10 ω + 40 log10 ωn ...(7.69)
From Eqs. (7.68) and (7.69) it is observed that the magnitude in decibel versus log 10 ω
graph for case (i) lies on 0 db axis and for case (ii) graph has a slope of 40 db/decade.
STABILITY ANALYSIS OF CONTROL SYSTEMS 299
The intersection of the two graphs on 0 db axis is given by equating Eq. (7.69) to zero,
∴ 0 = 40 log10 ω + 40 log10 ωn
or 40 log10 ω = 40 log10 ωn ∴ ω = ωn
Hence, the two graphs intersect on 0 db axis at ω = ωn and the corner frequency is ωn
(ωn is also natural frequency of oscillation for the second order term).
The graphs mentioned above are approximated asymptotic plots. However, the mag-
nitude near about frequency ωn depends on the value of damping ratio ζ. The maximum
magnitude is determined below :
The magnitude of the second order term is given by
ωn2 1
2
= ...(7.70)
2
[(ωn − ω ) + j 2ζ ωn ω] ⎛
2 2
ω ⎞ ⎛ 2
ω ⎞
⎜⎜ 1 − 2 ⎟⎟ + ⎜ 2ζ ⎟
⎝ ωn ⎠ ⎝ ωn ⎠
The frequency at which the maximum value of expression (7.70) is attained, is called
resonant frequency and denoted by ωr. The resonant frequency ωr is determined by taking
first derivative of Eq. (7.70) w.r.t. ω, equating to zero and solving for ω.
d ⎡ ω2n ⎤ ⎡ 1 ⎤
d ⎢ ⎥
∴ ⎢ 2 2 ⎥ =
dω ⎣⎢[(ωn − ω ) + j 2ζ ωn ω] ⎦⎥ dω ⎢ (1 − ω / ωn ) + [2ζ (ω / ωn )]2
2 2 ⎥
⎣ ⎦
−1
d ⎡ 1 ⎤ d ⎡⎢ ⎛ 2ω2 ω4 2 ⎤
2 ω ⎞⎥
= = ⎜⎜ 1 − 2 + 4 + 4ζ ⎟
dω ⎢ ⎛ 2 ⎥ dω ⎢ ω2n ⎟⎠ ⎥
⎢ 1 − 2ω + ω + 4ζ 2 ω ⎞ ⎥
2 4
⎣ ⎝ ωn ωn
⎦
⎜
⎢ ⎜ 2 2 ⎟
2 ⎟⎥
⎣ ⎝ ω n ω n ωn ⎠ ⎦
3
⎡ ω3 ζ2 ω ⎤
−
1 ⎡⎛ 2ω2 ω4 ζ 2 ω2 ⎞ ⎤ 2 ω
= ⎢⎜⎜ 1 − 2 + 4 + 4 ⎟⎥ ⎢0 − 4 2 + 4 4 + 8 2 ⎥
2 ⎣⎢⎝ ωn ωn ω2n ⎟⎠ ⎦⎥ ⎣ ωn ωn ωn ⎦
⎡ ω ω3 2 ω ⎤
⎢ − 4 2 + 4 4 + 8ζ ⎥
1 ⎣ ωn ωn ω2n ⎦
= − 3/2
...(7.71)
2 ⎡⎛
2ω2 ω4 ζ 2 ω2 ⎞ ⎤
⎢⎜⎜ 1 − 2 + 4 + 4 2 ⎟⎟ ⎥
⎢⎣⎝ ωn ωn ωn ⎠ ⎥⎦
Equating the numerator of the R.H.S. of Eq. (7.71) to zero, the following equation is
obtained :
⎡ ω ω3 2 ω ⎤ 4ω ⎛ ω2 2⎞
⎢ − 4 2 + 4 4 + 8ζ ⎥ = 0, or ⎜ − 1 + 2 + 2ζ ⎟⎟ = 0
2 ⎜
⎣ ωn ωn ω2n ⎦ ωn ⎝ ωn ⎠
⎛ 4ω ⎞ ⎛ ω2 2⎞
As ⎜⎜ ω2 ⎟⎟ ≠ 0, ∴ ⎜⎜ − 1 + 2 + 2ζ ⎟⎟ = 0
⎝ n⎠ ⎝ ωn ⎠
∴ ω = ωn 1 − 2ζ 2
Hence, the resonant frequency is given by
ωr = ωn 1 − 2ζ 2 ...(7.72)
300 LINEAR CONTROL SYSTEMS
The expression (7.72) for resonant frequency ωr holds good only for ζ ≤ 0.707. At reso-
nant frequency ωr, the magnitude of the quadratic term ω2n /[(ω2n − ω2 ) + j 2ζωn ω] attains
maximum value and this maximum value is known as resonant peak denoted by Mr. The
resonant peak Mr is calculated below in terms of damping ratio ζ.
1
Mr =
[(1 − ω2r / ω2n )2 + (2ζ ωr / ωn )2 ]
Since ωr = ωn 1 − 2ζ 2 for ζ < 0.707
1
∴ Mr =
2 2
⎡ ω2n (1 − 2ζ 2 ) ⎤ ⎡ 1 − 2ζ 2 ⎤
⎢1 − 2 ⎥ + ⎢ 2ζ ωn ⎥
⎣ ωn ⎦ ⎢ ωn ⎥
⎣ ⎦
= 1
[1 − (1 − 2ζ 2 )]2 + [2ζ 1 − 2ζ 2 ]2
1 1
= =
(4ζ 2 − 4ζ 4 ) 4ζ 2 (1 − ζ 2 )
1
or Mr = ...(7.73)
2ζ (1 − ζ 2 )
⎡ 1 ⎤
In decibel, Mr (db) = 20 log10 ⎢ ⎥
⎢ 2ζ 1 − ζ 2 ⎥
⎣ ⎦
The phase angle is given by
⎡ ω ⎤
⎢ 2ζ ⎥
ω2n ⎢ ω ⎥
∠ 2 = − tan − 1 ⎢ n
[(ωn − ω2 ) + j 2ζωn ω] ⎛ ω 2 ⎞⎥
⎢ ⎜1 − ⎟⎟ ⎥
⎢⎣ ⎜⎝ ω2n ⎠ ⎥⎦
At corner frequency i.e., at ω = ωn
⎡ ω ⎤
⎢ 2ζ n ⎥
ω2n ⎢ ωn ⎥
∠ 2 = − tan − 1 ⎢
[(ωn − ω2 ) + j 2ζωn ω] ⎛ 2 ⎞⎥
⎢ ⎜ 1 − ωn ⎟⎟ ⎥
⎢⎣ ⎜⎝ ω2n ⎠ ⎥⎦
= tan 1 (∞) = 90° ...(7.74)
At resonant frequency, i.e., at ω = ωn 1 − 2ζ 2
⎡ 2ζ ω 1 − 2ζ 2 ⎤
⎢ n ⎥
ω2n ⎛ 1 − 2ζ 2 ⎞
⎢ ωn ⎥ 1⎜ ⎟
∠ = − tan − 1 ⎢ = tan
2 ⎞⎥ ⎜ ⎟
[(ω2n − ω2 ) + j 2ζωn ω] ⎛
⎢ 1 − ω2 ⎜ 1 − 2 ζ ⎥ ⎝
ζ
⎠
n⎜ ⎟⎟
⎢ ⎝ ωn ⎠ ⎥⎦
2
⎣
STABILITY ANALYSIS OF CONTROL SYSTEMS 301
⎡ ζ ⎤
= ⎢
90° + sin 1 ⎥ ...(7.75)
⎢ 1 − ζ2 ⎥
⎣ ⎦
In view of mathematical derivations obtained above the Bode plot for the quadratic
term ω2n /[(ω2n − ω2 ) + j 2ζωn ω] is drawn in Fig. 7.18.10.
Fig. 7.18.10. Bode plot for quadratic term ωn2/[(ωn2 ω2) + j2ζωnω].
7.18.9. Procedure for drawing Bode plot and determination of gain margin, phase margin and
stability
In section 7.7 the stability in terms of gain margin and phase margin as obtained from the
Nyquist plot has been discussed. The gain margin and phase margin can also be obtained
from the Bode plot of the open-loop transfer function and the stability of closed-loop system
can be determined.
The two illustrative examples given below describe the procedure for constructing the
Bode plot and determination of stability in terms of gain margin and phase margin there-
from.
302 LINEAR CONTROL SYSTEMS
Illustrative Example 1. Construct Bode plot for the system whose open-loop trans-
fer function is given below and determine (a) the gain margin (b) the phase margin and
(c) the closed-loop stability.
4
G(s) H(s) = ⋅
s ( 1 + 0.5s ) ( 1 + 0.08s )
4
Solution. G(s) H(s) =
s (1 + 0.5s) (1 + 0.08 s)
Put s = jω
4
G(jω) H(jω) =
jω (1 + j0.5 ω) (1 + j0.08 ω)
1. The corner frequencies are
1 1
ω= = 2 rad/sec and ω = = 12.5 rad/sec
0.5 0.08
2. The starting frequency of the Bode plot is taken as lower than the lowest corner
frequency. As the lowest corner frequency is 2 rad/sec the starting frequency is taken as say
1 rad/sec.
3. As the transfer function is type 1 the initial part of the Bode plot is 20 db/decade
and the intersection of the initial part of the plot with 0 db axis occurs at ω = K, i.e., at ω = 4.
The initial slope continues till the lowest corner frequency ω = 2 rad/sec is reached.
4. The corner frequency ω = 2 rad/sec is due to the denominator term 1/(1 + j0.5ω),
therefore, the slope of the Bode plot after this frequency changes by 20 db/decade. As the
slope prior to corner frequency ω = 2 rad/sec is 20 db/decade, the slope of the plot after ω =
2 rad/sec is [ 20 + ( 20)] db/decade = 40 db/decade and this slope continues till the next
corner frequency ω = 12.5 rad/sec is reached.
The corner frequency ω = 12.5 rad/sec is due to the denominator term 1/(1 + j0.08 ω),
therefore, the slope of the Bode plot beyond this frequency changes by 20 db/decade. As the
slope prior to corner frequency ω = 12.5 rad/sec is 40 db/decade, the slope of the plot after
ω = 12.5 rad/sec is [ 40 + ( 20)] db/decade = 60 db/decade and this slope continues for
frequencies greater than ω = 12.5 rad/sec.
5. ∠G(jω) H(jω) for frequencies between ω = 1 rad/sec to ω = 100 rad/sec is calculated
as below :
∠G(jω) H(jω) = 90° tan 1 (0.5ω) tan 1 (0.08ω)
ω (rad/sec) 1 2 8 10 20 50
∠G(jω) H(jω)° 121 144 198 207 234 252
The Bode plot G(jω) H(jω) db and ∠G(jω) H(jω) versus ω (log scale) is drawn and
shown in Fig. 7.18.11.
(a) The gain in db at phase crossover frequency is the gain margin. If gain is negative
the gain margin is positive. In Fig. 7.18.11 the phase crossover frequency is 5 rad/sec and
G (j5) H(j5) = 12 db hence, the gain margin is
G.M. = 12 db
STABILITY ANALYSIS OF CONTROL SYSTEMS 303
db
20 – 120°
PM + 27° Phase
– 20 d
ÐG(jw)H(jw)|
b/d crossover
|G(jw)H(jw)|
10 – 150°
5
0 – 180°
1 2.5 4 10 20 50 100
–4 w¢ r/s
– 10 0d – 210°
Gain b/d
– 20 crossover – 240°
ÐG(jw)H(jw)
– 30 GM + 12 db – 270°
–
60
db
[G(jw)H(jw)] /d
4
Fig. 7.18.11. Bode plot for G(s) H(s) = .
s(1 + 0.5s) (1 + 0.08 s)
Solution. 1. The corner frequencies in this example are same as those in the previous
example, i.e.
1 1
ω=
= 2 rad/sec and ω = = 12.5 rad/sec
0.5 0.08
2. The transfer function is type 1, therefore, the initial slope of the plot is 20 db/
decade intersecting 0 db axis at ω = K, i.e., ω = 30. The initial slope 20 db/decade continues
till the lowest corner frequency ω = 2 rad/sec is reached. The slope of the plot between corner
frequencies ω = 2 rad/sec and ω = 12.5 rad/sec is [ 20 + ( 20)] db/decade = 40 db/decade
and the slope of the plot beyond ω = 12.5 rad/sec is [ 40 + ( 20)] db/decade = 60 db/decade.
3. ∠G(jω) H(jω) plot is same as that in the previous example.
The Bode plot G(jω) H(jω) db and ∠G(jω) H(jω) versus ω (log scale) is drawn and
shown in Fig. 7.18.12.
In Fig. 7.18.12 the gain at phase crossover frequencies ω = 5 rad/sec is G(j5) H(j5) = 9
db hence, the gain margin is
G.M. = 9 db
304 LINEAR CONTROL SYSTEMS
db
– 20 d
30 b/d
GM = – 9 db Gain
crossover – 120°
20 –4
0d
b/d – 150°
|G(jw)H(jw)|
Ð G(jw)H(jw)
10
7.1
0 – 180°
1 2 4 5 6 10 12.5 20 30 40 50 100
w r/s
– 10 |G(jw)H(jw)| – 210°
Fig. 7.18.12. Bode plot for G(s) H(s) = 30/s(1 + 0.5s)(1 + 0.08s).
Solution. Put s = jω
2e− j 0.5ω
∴ G(jω) =
jω ( j 0.5 ω + 1) ( j 0.125 ω + 1)
(a) without time delay element,
2
G(jω) =
jω ( j 0.5 ω + 1) ( j 0.125 ω + 1)
The corner frequencies are
1 ⎫
ω= = 2 rad./sec. ⎪
0.5 ⎪
⎬ due to denominator term (zero)
1
ω= = 8 rad./sec.⎪
0.125 ⎪⎭
306 LINEAR CONTROL SYSTEMS
As the type of the system is 1. The initial slope is 20 db/decade intersecting 0 db-axis
at ω = K = 2.
As the lower corner frequency = 2 rad./sec. and higher corner frequency = 8 rad./sec.,
the starting frequency for the Bode plot is selected as 1 rad./sec. and ending frequency as 10
rad./sec.
∠G(jω) = 90° tan 1 0.5ω tan 1 0.125ω
ω rad./sec. 2.0 4.0 6.0
∠ G(jω)° 149 180 198
As per data determined above, the Bode plot is drawn as shown in Fig. 7.18.14.
90°
20 db
P.M. 180°–149° = 31° 120°
|G(jw)|–db P.M. 180°–177.5° = 2.35°
150°
– 20 d Ðw/o e–j0.5w G.M. = 11.5 db
b/d
3.4 180°
0 db 1.0 2 4 6 8 10
w r/s 210°
G.M. = 7db -40 d
b/d
Ðwith e–j05w 240°
–20 db –
60
db 270°
/d
300°
–40 db
Fig. 7.18.14. Bode plot for example 3.
G.M. = 11.5 db ⎫
It is noted that, ⎬ Ans.
P.M. = 31° ⎭
(b) ∵ e j 0.5 ω = 1
The magnitude (db) plot is not affected.
The ∠G(jω) with time delay element is given by
∠ G(jω) = ∠ e j 0.5 ω 90° tan 1 0.5 ω tan 1 0.125 ω
− 0.5 ω × 180
= 90° tan 1 0.5 ω tan 1 0.125 ω
π
= 28.65 ω 90° tan 1 0.5 ω tan 1 0.125 ω
ω rad./sec. 2.0 3.4 4.0
∠ G(jω)° 177.5 180° 294.6
G.M. = 7.0 db ⎫
It is noted that, ⎬ Ans.
P.M. = 2.35° ⎭
7.18.11. Minimum phase, non-minimum phase and all pass transfer function
Transfer functions considered in previous sections have no poles and zeros in the RHS of
s-plane are called minimum phase transfer functions. Transfer functions having at least one
pole or zero in the RHS of s-plane are called non-minimum phase transfer functions.
STABILITY ANALYSIS OF CONTROL SYSTEMS 307
1 + sT1 1 − sT1
Let G1(s) = and G2(s) =
1 + sT2 1 + sT2
In sinusoidal form above transfer functions are written as
1 + jωT1
G1(jω) = ...(7.76)
1 + jωT2
1 − jωT1
G2(jω) = ...(7.77)
1 + jωT2
The transfer function given by Eq. (7.76) is a minimum phase transfer function and
that given by Eq. (7.77) is non-minimum phase transfer function. The pole-zero
configurations and Bode plots for such transfer functions are shown in Figs. 7.18.15 and
7.18.16 respectively.
Imj Imj
–1 –1 Re –1 1 Re
T2 T1 T2 T1
ÐG(jw) ÐG(jw)
|G(jw)|
|G(jw)|
0° 0°
ÐG(jw)
ÐG(jw)
G(jw) – 90° |G(jw)| – 90°
0 – 180° 0 – 180°
db w= 1 w= 1 db 1 1
T1 T2 T1 T2
w w
1 − sT
G(s) =
1 + sT
In sinusoidal form above transfer function is written as
1 − jωT
G(jω) = ...(7.78)
1 + jωT
The pole-zero configuration and Bode plot for all pass transfer function is shown in
Fig. 7.18.17.
ÐG(jw)
|G(jw)| db
Imj 0°
ÐG(jw)
|G(jw)| – 90°
Re 0 – 180°
–1 1 db w
T T
(a) (b)
Fig. 7.18.17 (a) Pole-zero configuration (b) Bode plot for all pass transfer function.
The magnitude plot lies on 0 db axis indicating that the actual gain for the frequencies
is 1, thus the transfer function passes all frequencies.
The phase angle for all pass transfer function is given by
1 − jωT 1 1
∠ = tan ( ωT) tan (ωT)
1 + jωT
= 2 tan 1 (ωT) ...(7.79)
The phase angle for all pass transfer functions given by Eq. (7.79) varies from 0° to
180° as the frequency is varied from ω = 0 to ω = ∞.
The Bode plot analysis is suitable only for minimum phase transfer functions whereas
Nyquist plot method is suitable for both non-minimum phase and minimum phase transfer
functions.
Example 7.19.1. Sketch the Bode plot for the open-loop transfer function for the unity
feedback system given below and assess stability
50
G(s) = ⋅
( s + 1) ( s + 2 )
db
40 0°
–2
0d – 45°
b /d
|G(jw)|
20 K 27.9 db – 90°
20 log10 PM = 23°
ÐG(jw)
|G(jw)|
ÐG(jw) – 135°
0 – 180°
0.1 0.2 0.4 1 2 4.5 10 20 40 100
w – 225°
–
40
Gain – 270°
–20
db
crossover
/d
Example 7.19.2. Sketch the asymptotic Bode plot for the transfer function given
below
2 ( s + 0.25 )
G(s) H(s) = 2
s ( s + 1) ( s + 0.5 )
310 LINEAR CONTROL SYSTEMS
ÐG(jw) H(jw)
– 20
|G(jw)H(jw)|
d b /d
20 –4 crossover – 150°
GM = – 20 db 0d
b/d
1.24
0 0.1 0.2 0.25 – 180°
0.4 1 2 4 5 10
w
– 20 –6 – 210°
0d
Phase b /d
– 40 crossover PM = – 35° – 240°
– 60 ÐG(jw)H(jw) – 270°
|G(jw)H(jw)|
Fig. 7.19.2. Bode plot for example 7.19.2.
(d) At gain crossover the phase angle is 215°, therefore, the phase margin as
P.M. = 180 + ( 125) = 35°. Ans.
As both the gain margin and phase margin are negative the system is unstable.
Example 7.19.3. Draw the Bode plot for the transfer function given below :
48 ( s + 10 )
G(s) H(s) =
s ( s + 20 ) ( s 2 + 2.4s + 16 )
Apply correction to the magnitude plot for the quadratic term and comment on the
stability.
db
40 Gain
crossover
30 – 90°
Phase
crossover
20 – 120°
PM = 75°
– 150°
ÐG(jw)H(jw)
10
|G(jw)H(jw)|
5
0 – 180°
1.5 – 4 10 20 50 100
20 db w
– 10 /d G.M. 5 db – 210°
–
60
– 20 db – 240°
/d
– 30 –4 – 270°
0d ÐG(jw)H(jw)
b /d
–
60
– 40
db
|G(jw)H(jw)|
/d
– 50
1
ω= = 20 rad/sec is due to denominator first order term, the slope of the magni-
0.05
tude plot after ω = 20 rad/sec is ( 40 20) = 60 db/decade.
The damping ratio is calculated below.
Since 2ζ ωn = 2.4 and ωn = 4
2.4 2.4
∴ ζ= = = 0.3
2ωn 2 × 4
The exact magnitude at ω = 4 rad/sec is determined analytically as follows :
1.5( j0.1 × 4 + 1) × 16
20 log10 G(j4) H(j4) = 20 log10
j 4 ( j0.05 × 4 + 1) [ j 2.4 × 4 + (16 − 42 )]
1
⎛ 2.4ω ⎞ 1 1
∠G(jω) H(jω) = tan (0.1ω) 90° ⎜ tan
2⎟
;ω<4 (0.05ω) tan
⎝ 16 − ω ⎠
2.4ω
∠G(jω) H(jω) = tan 1 (0.1ω) 90° tan 1 (0.05ω) 180° tan 1 ;ω>4
16 − ω2
The phase angle for the frequency range of the Bode plot is calculated and given
below :
ω rad/sec 1 4 5 10 20 50
∠ G(jω) H(jω)° 96.25 169.5 204.3 235.6 244.6 256.8
The gain margin is, G.M. = 5 db
The phase margin is, P.M. = 180° + ( 105°) = + 75°
As both the gain margin and phase margin are positive the system is stable. Ans.
STABILITY ANALYSIS OF CONTROL SYSTEMS 313
Example 7.19.4. Determine the value of gain K for the open-loop transfer function
given below so that (a) the gain margin is 15 db and (b) phase margin is 60°.
K
G(jω) H(jω) =
jω( j0.1 ω + 1) ( jω + 1)
40
db
– 90°
P.M. = 60°
20 – 120°
ÐG(jw)H(jw)
– 20
d b /d ÐG(jw)H(jw)
|G(jw)H(jw)|
– 150°
3.16
0 5
– 180°
0.1 0.2 0.4 1 2 10 50 100
–4 G.M. w r/s – 210°
0.58 0d = 15 db
b /d
– 20 – 240°
|G(jw)H(jw)| – 270°
– 40
–
60
K=2
db
/d
K=1
K = 0.58
Fig. 7.19.4. Bode plot for example 7.19.4.
The slope of the magnitude of the plot after ω = 1 rad/sec is ( 20 20) = 40 db/decade
and after ω = 10 rad/sec is ( 40 20) = 60 db/decade.
The phase angle is given by
∠G(jω) H(jω) = 90° tan 1 (0.1ω) tan 1 (ω)
The phase angle for the frequency range between ω = 0.1 rad/sec to 100 rad/sec is
calculated and given below :
ω rad/sec 0.1 1 5 10 50
∠G(jω) H(jω)° 96.3 140.7 195.6 219.7 257.4
The phase crossover is at ω = 3.16 rad/sec (Fig. 7.19.4), therefore, to have a gain
margin of 15 db the gain at ω = 3.16 rad/sec (phase crossover) should be 15 db which is
obtained by shifting the magnitude plot upwards such that the gain at ω = 3.16 rad/sec is
15 db. The initial slope of this shifted plot intersects the 0 db axis at ω = 2 rad/sec, hence,
the value of gain K of the system to have a gain margin of 15 db is
K = 2. Ans.
314 LINEAR CONTROL SYSTEMS
The phase margin of 60° is marked on the Bode plot (Fig. 7.19.4) at ω = 0.58 rad/sec. In
order to have a phase margin of 60° the gain crossover should be at ω = 0.58 rad/sec which is
obtained by shifting the magnitude plot downwards such that the gain at ω = 0.58 rad/sec
(gain crossover) is 0 db. The initial slope of this shifted plot intersects the 0 db axis at ω =
0.58 rad/sec, hence the value of gain K of the system to have a phase margin of 60° is
K = 0.58. Ans.
Example 7.19.5. Determine the open-loop transfer function from the Bode plot shown
in Fig. 7.19.5.
db
–2
0d
b/d
24.1
–4
0d
b /d
w2 w3 w rad/Sec
0 w4
1.0 w1 8 10 100
– 20
12.05 db/d
–4
20.05 0d
b/d
Solution. The initial slope of the plot is 20 db/decade and its intersection with the
0 db axis is located as ω4, hence the system is type 1 and K = ω4.
At ω1 the slope of the plot changes by 20 db/decade, hence the corresponding term of
the transfer function is 1/(sT1 + 1),
1
where T1 = ⋅
ω1
At ω2 the slope of the plot changes by + 20 db/decade, hence the corresponding term of
the transfer function is (sT2 + 1),
1
where T2 = ⋅
ω2
At ω3 the slope of the plot changes by 20 db/decade, hence the corresponding term of
the transfer function is 1/(sT3 + 1),
1
where T3 = ⋅
ω3
The calculations are given below :
1. log10 ⎛⎜ 8 ⎞⎟ = 24.1
⎝ ω1 ⎠ 40
STABILITY ANALYSIS OF CONTROL SYSTEMS 315
8 24.1 ⎞
∴ = antilog10 ⎛⎜ ⎟ =4 ∴ ω1 = 2 rad/sec.
ω1 ⎝ 40 ⎠
1 1
and T1 = = = 0.5 sec.
ω1 2
It is noted from Eqs. (7.80) and (7.81) that Mr as well as Mp occur for values of ζ as 0 ≤
ζ ≤ 0.707.
The phase margin of a second order system can be determined in terms of ζ as follows :
The open-loop transfer function of a unity feedback second order control system is
given by
ω2n
G(s) = ...(7.82)
s (s + 2ζ ωn )
In sinusoidal form Eq. (7.82) can be written as
ω2
G(jω) =
jω ( jω + 2ζ ωn )
ω2n
= ...(7.83)
− ω2 + j 2ζ ωn ω
The magnitude of Eq. (7.83) is
ω2n
G(jω) = ...(7.84)
[(− ω2 )2 + (2ζ ωn )2 ]1 / 2
At gain cross-over frequency ω1 the magnitude G(jω) is unity. Hence,
[(− ω12 )2 + (2ζ ωnω1 )2 ]1 / 2 = ω2n ...(7.85)
or ω14 + 4ζ 2
ω2nω12 − ω2n =0
1
3
0.8
2 0.6
Mr
Mp
0.4
1
0.2
90°
Phase Margin, fm
60°
30°
0°
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Fig. 7.20.2. Phase margin φm as a function of ζ.
Usually frequency response forms the basis of design of a control system. However, it
is not convenient and simple to correlate transient and frequency response of higher order
control systems as in the case of second order systems. The correlation between the transient
and frequency response for higher order systems depends on the location of closed-loop poles
(roots of the characteristic equation) in s-plane. The transient response and frequency
response of a higher order system is dominated by the dominant poles (poles nearer to
imaginary axis in s-plane) and the mathematical correlation established for second order
318 LINEAR CONTROL SYSTEMS
ζπ ζπ
− −
1 − ζ2 1 − ζ2
Solution. %Mp = e × 100 ∴ 26 = e × 100
ζπ
+
1 − ζ2 100 ζπ
or e = ∴ = 1.34
26 1 − ζ2
2
⎛ ζπ ⎞
or 1 ζ2 = ⎜ ⎟ or 1 ζ2 = 5.49 ζ2
⎝ 1.34 ⎠
1 1
∴ ζ2 = ∴ ζ= = 0.39
1 + 5.49 6.49
Resonant frequency :
ωr = ωn 1 − 2ζ 2
∴ 8 = ωn 1 − 2ζ 2 = ωn 1 − 2 × 0.392 = ωn 0.699
8
∴ ωn =
= 9.6 rad/sec
0.699
The overall transfer function is
K
C (s) G (s) s (sT + 1)
M (s) = = =
R (s) 1 + G (s) H (s) 1 + K
⋅1
s (sT + 1)
K K /T
==
s T + s + K s2 + 1 s + K
2
T T
The characteristic equation is
1 K
s2 + s+ =0
T T
On comparing above equation with s2 + 2ζ ωns + ω2n = 0 (characteristic equation of a
second order system)
1
2ζ ωn =
T
STABILITY ANALYSIS OF CONTROL SYSTEMS 319
1 1
∴ T= = = 0.13. Ans.
2ζ ωn 2 × 0.39 × 9.6
K
and ωn2 =
T
∴ K = ωn2 T = (9.6)2 (0.13) = 11.9. Ans.
1 1
Resonant peak : Mr = = = 1.39. Ans.
2ζ 1 − ζ 2 2 × 0.39 1 − 0.392
Gain crossover frequency
ω1 = ωn (4ζ 4 + 1)1 / 2 − 2ζ 2 = 8.69 (4 × 0.394 + 1)1 / 2 − 2 × 0.392
Ks
Example 7.21.1. Let G(s) H(s) = ⋅
( s + 1)
–1 – 0.5 0
K=0 K=1 K=¥ Re
No. of open-loop poles P = 1
No. of open-loop zeros Z = 1
– j0.5
No. of root locus branches N = 1
Ks
Fig. 7.21.2. Root locus for G(s) H(s) = ⋅
(s + 1)
K
Example 7.21.2. Let G(s) =
s ( s + 4)
and H(s) = 1.
Solution. The characteristic equation of the system is
1 + G(s) H(s) = 0
K
∴ 1+ ⋅1 = 0
s (s + 4)
or s2 + 4s + K = 0
The two roots of above equation are
s1, s2 = 2± 4−K
For various values of K the roots are
K 0 2 4 6 ∞
s1 0 1 2* 2+ j 2 2 + j∞
s2 4 3 2* 2 j 2 2 j∞
Imj
j3
K=¥ j2
K=6 j1
–4 –3 –2 –1 K=0 Re
K=0 K=2
K=6 – j1
No. of open-loop poles P = 2
K=4
K=¥ – j2 No. of open-loop zeros Z = 0
No. of root locus branches N = 2
– j3
Following conclusions are drawn from the study of root locus plotted in Fig. 7.21.3.
1. For the values of K between K = 0 and K = 4 the two roots are real negative indicat-
ing overdamped time response.
2. For K > 4 the two roots are complex with negative real part indicating that the time
response is underdamped, i.e. exponentially decaying oscillations.
3. Whatever may be the value (+ ve) of K there is no chance for the real part of the
roots to become positive, hence the system is inherently stable.
K
Example 7.21.3. Let G(s) = ⋅
s ( s + 1) ( s + 3 )
Imj
K=¥
j2
K = 12
j1
K = 0.631
–3 –1
– j1
No. of open-loop poles P = 3
No. of open-loop zeros Z = 0
– K = 12
– j2 No. of root locus branches N = 3
K=¥
Fig. 7.21.4. Root locus for G(s) H(s) = K/s(s + 1)(s + 3).
Example 7.21.4. A typical root locus plot for a system having 5 open-loop poles and 1
open-loop zero is shown in Fig. 7.21.5.
Solution.
Angle of Intersection
Departure Point With
Imj Axis
K=¥ K=¥
K=0
I n As
3
te ym
Asymptotes
rs
ec pto
tio te
n
pt
5 1
.
K=¥ K=¥
K=0
Fig. 7.21.5. Typical root locus plot for a system having Number of open-loop poles P = 5 and
Number of open-loop zero Z = 1. No. of root locus branches N = 5.
STABILITY ANALYSIS OF CONTROL SYSTEMS 323
(2k + 1) × 180°
where k = 0, 1, 2, ... upto (P Z) 1
P−Z
7. Intersection of asymptotes on real axis. The asymptotes intersect at a point x
on the real axis given by
Σ Poles − Σ Zeros
x=
P−Z
8. Intersection points with imaginary axis. The value of K and the point at which
the root locus branch crosses the imaginary axis is determined by applying Routh criterion
to the characteristic equation. The roots at the intersection point are imaginary.
9. The angle of departure from complex pole. The angle of departure of the root
locus from a complex pole is given by
φd = 180° (φp φz)
where φp is the sum of all the angles subtended by remaining poles.
and φz is the sum of all the angles subtended by zeros.
The angle of departure is tangent to the root locus at the complex pole.
10. The angle of arrival at complex zeros. The angle of arrival of the root locus at
a complex zero is given by
φa = 180° (φz φp)
where φz is the sum of all the angles subtended by remaining zeros.
and φp is the sum of all the angles subtended by poles.
The angle of arrival is tangent to the root locus at the complex zero.
The steps given above are sufficient to plot the root locus for a given open-loop transfer
function. The calculation of the value of K at a specified point on the root locus, determina-
tion of damping ratio ζ, assessment of time response and stability etc. are illustrated in the
following solved examples.
Example 7.24.1. A unity feedback control system has an open-loop transfer function
K
G(s) =
s ( s + 4)
Draw the root locus and determine the value of K, if the damping ratio ζ is to be 0.707.
dK
∴ = (2s + 4)
ds
dK
Put =0 ∴ (2s + 4) = 0
ds
∴ s = 2 is a breakaway point.
6. The angle of asymptotes is given by
(2k + 1) 180°
where k = 0 and 1
P−Z
The two asymptotes are at an angle of
(2 × 0 + 1) 180°
(i) = 90°
2−0
(2 × 1 + 1) 180°
(ii) = 270°
2−0
7. The asymptotes intersect on real axis at a point given by
Σ Poles − Σ Zeros (0 − 4) − 0
x= = =−2
P−Z 2
With the data calculated in steps (1) to (7) the root locus is plotted in Fig. 7.24.1.
Since ζ = 0.707
and cos θ = ζ (Refer Fig. 6.4.7)
∴ cos θ = 0.707
θ = 45°
From the origin a line at an angle θ = 45° is drawn as shown in Fig. 7.24.1 which
intersects the root locus plot at s = ( 2 + j2). As the point s = 2 + j12 lies on the root locus the
following equation (magnitude condition) should be satisfied at s = ( 2 + j2)
G(s) H(s) = 1
K K=¥ Imj
or =1
s (s + 4)
(– 2 + j2) K = 8 j2
K
∴ =1 90°
(− 2 + j 2) [(− 2 + j 2) + 4] – 4 270° q
K=0 –2 K=0 Re
K
=1 –1
(− 2 + j 2) (2 + j 2) q = cos z
–1 K=¥ – j2
= cos (0.707)
= 45°
K
=1 Fig. 7.24.1. Root locus for G (s) H (s)
(2 + 2 ) (22 + 22 )
2 2
= K/s(s + 4).
K
=1 or K = 8. Ans.
8 8
326 LINEAR CONTROL SYSTEMS
Example 7.24.2. Sketch the root locus for the open-loop transfer function of a unity
feedback control system given below and determine
(i) The value of K for ζ = 0.5 (ii) the value of K for marginal stability (iii) the value of
K at s = 4 and obtain the closed-loop transfer function for K = 1.66.
K
G(s) = .
s ( s + 1) ( s + 3 )
K=¥
j2
j1.732
K = 12
– 0.35 + j0.6
j1
K=¥
180° 60°
K = 12 K=0 q K=0
–4 –3 –1.33 – 1 – 0.45 0 Re
300°
– j1
–1
q = cos z
–1 – j1.732
= cos (0.5) – j2 K =12
= 60°
K=¥
From the origin a line at an angle θ = 60° is drawn as shown in Fig. 7.24.2 which
intersects the root locus plot at s = ( 0.35 + j0.6). As the point s = ( 0.35 + j0.6) lies on the
root locus the following equation is satisfied.
K
G(s) H(s) =1 or =1
s (s + 1) (s + 3)
Put s = ( 0.35 + j0.6)
K
∴ =1
(− 0.35 + j 0.6) [( − 0.35 + j0.6) + 1] [( − 0.35 + j 0.6) + 3]
K
∴ =1 or K = 1.66. Ans.
1.66
(iii) The third root of the ch. eqn. is located on the s-plane negative real axis beyond
s = 3. For K = 1.66, this root can be determined using root locus magnitude condition G(s)
H(s) = 1. Since, H(s) = 1 (unity feedback)
∴ G(s) = 1
K
∴ =1
s ( s + 1) ( s + 3)
Put K = 1.66
1.66
∴ =1
s ( s + 1) ( s + 3)
The equation above is satisfied (to a fair approximation) at s = 3.231.
The overall transfer function is given by :
C (s) G (s) K /[ s (s + 1) (s + 3)] K
= = =
R (s) 1 + G (s) 1 + { K /[ s (s + 1) (s + 3)]} × 1 s (s + 1) (s + 3) + K
C (s) 1.66
Put K = 1.66 ∴ =
R (s) s (s + 1) (s + 3) + 1.66
The characteristic equation n is s (s + 1) (s + 3) + 1.66 = 0. As per root locus plot, the
three roots are s1, s2 = ( 0.35 ± j 0.6), s3 = 3.231.
C (s) 1.66
∴ =
R (s) (s + 3.231) (s + 0.35 + j 0.6) (s + 0.35 − j 0.6)
C (s) 1.66
or = 2
. Ans.
R (s) (s + 3.231) (s + 0.7 s + 0.4825)
1 ⎛3⎞
φp1 = 180° tan ⎜ ⎟ = 123.6°
⎝2⎠
1 ⎛3⎞
φp2 = tan ⎜ ⎟ = 36.86°
⎝4⎠
φp3 = 90°
330 LINEAR CONTROL SYSTEMS
F(s)
F = (– 4) = + 6 10
5
–5 – 4.5 –4 –3 –2 –1
0
–5
F = (– 4.2) = 0
– 10
F = (– 4.5) = – 12 – 15
– 20
– 25
– 30
– 35
F = (–5)= – 45 – 40
Imj
(– 2, + j3)
j3
j2
3
j1
f p2 f p1
–6 –5 –4 –3 –2 –1 0 Re
4 2 – j1
– j2
f p3
– j3
(– 2, – j3)
Fig. 7.24.4. Pole-zero configuration for G(s) H(s) = K/s(s + 6)(s2 + 4s + 13).
Imj
K=¥
K=¥
(–2, + j3)
– 70.4° K = 229.4
K=0 + j2.98
135° – j1
45°
K=0 K=0
–6 –4.2 –2.5 0 Re
225°
315°
– j1
K=0 – j2.98
+ 70.4° K = 229.4
Fig. 7.24.5. Root locus plot for G(s) H(s) = K/s(s + 6)(s2 + 4s + 13).
From the root locus plot (Fig. 7.24.5) it is noted that for K > 229.4 the two roots have
positive real part, therefore, the system is stable if,
K < 229.4. Ans.
Example 7.24.4. Sketch the root locus plot for the system having open-loop transfer
function is given by
K
G(s) H(s) = ⋅
s (s + 4 ) (s2 + 4s + 13)
(2k + 1) × 180°
The four angles of asymptotes are given by P − Z , k = 0, 1, 2, and 3
(2 × 0 + 1) + 180° (2 × 1 + 1) × 180°
(i) = 45° (ii) = 135°
4 −0 4−0
(2 × 2 + 1) × 180° (2 × 3 + 1) × 180°
(iii) = 225° (iv) = 315°
4−0 4−0
The asymptotes intersect on the real axis at
Σ Poles − Σ Zeros (0 − 4 − 2 + j3 − 2 − j3) − (0) − 8
x= = = = 2
P − Z 4 −0 4
The pole-zero configuration is shown in Fig. 7.24.6.
The angle of departure from the complex pole ( 2 + j3) is given by (Refer Fig. 7.24.6)
φ( 2 + j3) = 180° (φp1 + φp2 + φp3)
1 ⎛3⎞
φp1 = 180° tan ⎜ ⎟
⎝2⎠
1 ⎛3⎞
φp2 = tan ⎜ ⎟
⎝2⎠
φp3 = 90°
⎡ −1 ⎛3⎞ −1 ⎛3⎞ ⎤
∴ φ( 2 + j3) = 180° ⎢180° − tan ⎜⎝ 2 ⎟⎠ + tan ⎜⎝ 2 ⎟⎠ + 90° ⎥ = 90°
⎣ ⎦
Imj
K=¥
K=¥
(–2, + j3)
Imj
K=0
(– 2, + j3) (–2, + j1.58)
j3
j 2.56
j2 K = 148.6
135°
3
45°
j1
K=0 K=0
f p2 f p1
–4 –2 –1 0 Re
–4 –3 –2 –1 0 Re 225°
315°
2 2 – j1
– j 2.56
– j2
K = 148.6
f p3
(–2, – j1.58) K=0
– j3
(–2, + j3)
(– 2, – j3)
K=¥ K=¥
Fig. 7.24.6. Pole-zero configuration for Fig. 7.24.7. Root locus plot for
G(s) H(s) = K/s(s + 4)(s2 + 4s + 13). G(s) H(s) = K/s(s + 4)(s2 + 4s + 13).
STABILITY ANALYSIS OF CONTROL SYSTEMS 333
and the angle of departure from the complex pole ( 2 j3) is
φ( 2 j3) = + 90°.
The root locus plot takes the shape as shown in Fig. 7.24.7. There are three break-
away points. One of the breakaway point is identified at s = 2, the other two breakaway
points are complex and can be obtained as follows :
The characteristic equation is
s(s + 4) (s2 + 4s + 13) + K = 0
or K = (s4 + 8s3 + 29s2 + 52s)
dK
∴ = ( 4s3 + 24s2 + 58s + 52)
ds
dK
Put =0 ∴ (4s3 + 24s2 + 58s + 52) = 0
ds
The breakaway points are given by the roots of the above equation, one of the root is
identified as s = 2 (at s = 2 above equation is satisfied), therefore,
(4 s3 + 24 s2 + 58 s + 52)
= (4s2 + 16s + 26)
(s + 2)
and thus (s + 2) (4s2 + 16s + 26) = 0
The two complex breakaway points are given by the roots of the equation (4s2 + 16s +
26) = 0, therefore the complex breakaway points are
− 16 ± (162 − 4 × 4 × 26)
s= = ( 2 ± j1.58)
2×4
The intersection of the root locus plot with the imaginary axis is determined by apply-
ing Routh criterion to the characteristic equation
s4 + 8s3 + 29s2 + 52s + K = 0
The Routh array is arranged below
s4 1 29 K
s3 8 52
s2 22.5 K
s1 (52 0.35 K)
s0 K
For stability (52 0.35 K) ≥ 0
∴ K ≤ 148.6
The value of s at the intersection point with the imaginary axis where K = 148.6 is
given by solving the auxiliary equation for s2 term.
∴ 22.5s2 + K = 0 or 22.5s2 + 148.6 = 0
148.6
∴ s=± j = ± j2.56
22.5
The root locus plot is sketched in Fig. 7.24.7.
334 LINEAR CONTROL SYSTEMS
K (s + 1)2
Solution. G(s) H(s) =
(s + 2)2
Let any point in the s-plane s = (σ + jω) lie on the root locus
K (σ + j ω + 1)2
∴ G(σ + jω) H(σ + jω) =
(σ + j ω + 2)2
ω ω
and ∠ G(σ + jω) H(σ + jω) = 2 tan 1 2 tan 1
σ+1 σ+2
For the point to lie on root locus, then as per the phase angle condition.
∠ G(σ + jω) H(σ + jω) = ± (2k + 1) 180° ; k = 0, 1, 2, ...
Consider k = 0
∴ ∠ G(σ + jω) H(σ + jω) = 180°
1
ω 1
ω
∴ 2 tan 2 tan = 180°
σ+1 σ+2
⎧ ⎛ ω ω ⎞ ⎫
⎪ ⎜σ + 1 − σ + 2⎟ ⎪
⎪ ⎝ ⎠ ⎪ =
or 2 ⎨ tan −1 ⎬ 180°
⎪ ω2 ⎪
1+
⎪⎩ (σ + 1) (σ + 2) ⎪⎭
ω ω
−
σ+1 σ+ 2
or tan −1 = 90°
ω2
1+
(σ + 1) (σ + 2)
Taking tangent on both sides,
ω ω
−
σ + 1 (σ + 1) (σ + 2)
= tan ( 90°)
ω2
1+
(σ + 1) (σ + 2)
ω ω
−
σ+1 σ+2
or =∞
ω2
1+
(σ + 1) (σ + 2)
ω2
∴ 1+ =0
(σ + 1) (σ + 2)
or (σ + 1) (σ + 2) + ω2 = 0
STABILITY ANALYSIS OF CONTROL SYSTEMS 335
or σ2 + 3σ + 2 + ω2 = 0
or σ2 +
3σ + 2.25 2.25 + 2 + ω2 = 0
or (σ + 1.5)2 + (ω 0)2 = (0.5)2
The equation derived above is of a circle in s-plane with centre ( 1.5, j0) and radius
0.5. Ans.
Example 7.24.6. For the open-loop transfer function given below. Determine the
value of K at s = 2 and comment on the stability and time response of the system
K (s + 1)
G(s) H(s) = 2
⋅
(s + 0.4s + 0.4)
(0.4 2 − 4 × 1 × 0.4)
− 0.4 ±
s1, s2 = = ( 0.2 ± j0.6)
2×1
The no. of open-loop zero Z = 1 at (s = 1).
As P > Z, the number of root locus branches N = P = 2. On the real axis the root locus
exists between s = 1 (zero) and s = ∞.
The angle of asymptote is given by
(2k + 1) × 180°
where k = 0
P−Z
Therefore, the angle of asymptote is
(2 × 0 + 1) × 180°
= 180°
2−1
The characteristic equation is
(s2 + 0.4s + 0.4) + K (s + 1) = 0
(s2 + 0.4 s + 0.4)
or K=
(s + 1)
dK ⎡ (s + 1) (2s + 0.4) − ( s2 + 0.4 s + 0.4) × 1 ⎤ (s2 + 2s)
∴ = ⎢ ⎥ =
ds ⎣⎢ (s + 1)2 ⎦⎥ (s + 1)
In order to determine breakaway point put (dK/ds = 0) and solve for s, therefore,
(s2 + 2s)
=0
(s + 1)
or s(s + 2) = 0
∴ s = 0 and s = 2
As the point s = 0 does not lie on the root locus and such is not a breakaway point. The
point s = 2 lies on the locus and is a breakaway point (between two zeros the root locus
breaks away).
336 LINEAR CONTROL SYSTEMS
Example 7.24.7. The transfer function of a unity feedback control system is given by:
K
G(s) =
s (s + 2) (s + 4)
Determine :
(a) The value of K to have 40% overshoot for unit step input.
(b) The value of K for sustained oscillations in output.
(c) The value of Kv corresponding to value of K as obtained in (a).
(d) The value of settling time ts.
Solution.
1. The root loci start (K = 0) at s = 0, s = 2 and s = 4.
2. As there is no open-loop zero. The root loci terminate (K = ∞) at infinity.
3. P = 3, Z = 0 (P > Z). Therefore, number of branches of the root locus N = 3.
4. The root locus branches on the real axis exists between s = 0 and s = 2 and beyond s = 4.
5. Between open-loop poles s = 0 and s = 2, there exists a breakaway point as calcu-
lated below :
The characteristic equation is
s (s + 2) (s + 4) + K = 0
dK
or K= (s3 + 6s2 + 8s) ∴ = (3s2 + 12s + 8)
ds
STABILITY ANALYSIS OF CONTROL SYSTEMS 337
Imj
∞
K
j 2.8
s = –0.6 + j1.75 K = 48
K = 15.74 j1.75
j1
θ=7 3.7°
∞ K K=0 K= 0 K=0
0
– 0.6
–4 –2 Re
– 0.85
– j1
K =48
– j 2.8
K
∞
Fig. 7.24.9. Root locus plot for example 7.24.7.
dK
Put =0 ∴ 3s2 + 12s + 8 = 0 or s2 + 4s + 2.66 = 0
ds
− 4 ± 42 − 4 × 1 × 2.66
− 4 ± 2.3
∴ s= = = 3.15, 0.85.
2 2
As the breakaway point has to lie between s = 0 and s = 2, the breakaway point is
s= 0.85.
(2k + 1)
6. The angle of asymptotes is given by × 180° where, k = 0, 1, 2
P−Z
The three asymptotes are at an angle of
(2 × 0 + 1) (2 × 1 + 1) (2 × 2 + 1)
(i) × 180° = 60°, (ii) × 180° = 180° and (iii) × 180° = 300°
3−0 3−0 3−0
7. The three asymptotes intersect on the real axis at
Σ Poles − Σ Zeros (0 − 2 − 4) − (0)
x= =
P−Z 3−0
or x = 2.
8. The value of K for marginal stability is determined by applying Routh-Hurwitz
criterion to the characteristic equation.
s(s + 2) (s + 4) + K = 0 or s3 + 6s2 + 8s + K = 0
The Routh array is tabulated below :
s3 1 8
s 2 6 K
48 − K
s1
6
s0 K
338 LINEAR CONTROL SYSTEMS
The value of K at which the root locus plot crosses the imaginary axis is determined by
equating the first term in s1 row to zero, therefore,
48 − K
=0 ∴ K = 48
6
At K = 48. The root locus plot intersects imaginary axis and the value of s at the
intersection point is determined by solving auxiliary equation formed from the s2 terms in
Routh array.
∴ 6s2 + K = 0 or 6s2 + 48 = 0
or s2 = 48/6 or s = ± j 8 = ± j2.8
As per the data calculated above the root locus plot is drawn as shown in Fig. 7.24.9.
(a) The maximum overshoot given is 40%
−ζπ
40 1 − ζ2
∴ =e where, ζ = damping ratio
100
ζπ ζπ
or loge (0.4) = or 0.916 =
2
1 ζ 1 ζ2
2
ζ2 ⎛ + 0.916 ⎞
or 2
=⎜ ⎟ = 0.0846
1−ζ ⎝ π ⎠
or ζ2 = 0.0846 0.0846 ζ2 or ζ2 = 0.078 ∴ ζ = 0.28
θ = cos 1 ζ = cos 1 0.28 = 73.7°
A line passing through origin and making an angle of 73.7° with ve real axis is
drawn on the root locus plot. This line intersects root locus plot at s = 0.6 + j1.75.
The value of K at s = ( 0.6 + j1.75) is calculated below, which is the corresponding
value for 40% overshoot.
The point s1 = 0.6 + j1.75 lies on the root locus. Therefore,
K
1 =
s1 (s1 + 2) (s1 + 4)
K
∴ 1 =
(0.6 + j1.75) (− 0.6 + j1.75 + 2) (− 0.6 + j1.75 + 4)
K
=
(0.6 + j1.75) (1.4 + j1.75) (3.4 + j1.75)
K K
= =
2 2 2
0.6 + 1.75 1.4 + j1.75 2 2
3.4 + 1.75 2 1.84 × 2.24 × 3.82
∴ K = 1.84 × 2.24 × 3.82 or K = 15.74. Ans.
(ii) The root locus plot crosses the imaginary axis at K = 48, hence, for sustained
oscillations K = 48. Ans.
K
(c) Kv = lim s G(s) H(s) = lim s
s→0 s→0 s (s + 2) (s + 4)
∵ K = 15.74 as determined in (a)
15.74 15.74
Kv = lim = = 9.96. Ans.
s→0 (s + 2) (s + 4) 2 × 4
STABILITY ANALYSIS OF CONTROL SYSTEMS 339
(d) s= 0.6 + j1.75 ∴ ζωn = 0.6
4 4
Settling time ts = = = 6.66 sec. Ans.
ζωn 0.6
Example 7.24.8. Draw the root locus plot for a system having open-loop transfer
K
function as : G(s) H(s) =
s (s + 1) (s + 3)
Determine : (a) Gain margin and (b) Phase margin, for K = 6.
Imj
¥
K 0 Re
– 0.45
K =12
– j 1.732
K
¥
K
∴ 1+ =0
s (s + 1) (s + 3)
or s (s + 1) (s + 3) + K = 0 or K= (s3 + 4s2 + 3s)
∴ dK/ds = (3s2 + 8s + 3)
put dK/ds = 0 ∴ 3s2 + 8s + 3 = 0
− 8 ± 82 − 4 × 3 × 3
or s= or s = 1.33 ± 0.88 = 0.45, 2.21
2×3
As the breakaway point has to lie between s = 0 and s = 1, the breakaway point is
s = 0.45.
8. The value of K for marginal stability is determined by applying Routh criterion to
the characteristic equation. Therefore,
s (s + 1) (s + 3) + K = 0 or s3 + 4s2 + 3s + K = 0
The Routh array is tabulated below :
s3 1 3
s2 4 K
12 − K
s1
4
s0 K
For marginal stability
12 K = 0 or K = 12
The intersection of root locus plot with imaginary axis is given by solving the equation,
4s2 + K = 0
or 4s2 + 12 = 0 or s = ± j1.732.
From the data determined above, the root locus plot is drawn as shown in Fig. 7.24.10.
On the root locus plot at s = a the value K = 6 is marked (by trial and error).
This is obtained by drawing lengths oa, ba and ca from open-loop poles at s = 0, s = 1
and s = 3 respectively, such that,
K K
1 = or 1 = or K = 1.21 × 1.6 × 3.1
oa ba ca 1.21 × 1.6 × 3.1
∴ K=6
At a , s = ( 0.2 ± j1.23)
(a) Gain Margin : The desired value is K = 6, and the value for marginal stability as
obtained from root locus plot is K = 12.
K (marginal stability) 12
∴ Gain Margin = = =2
K (desired) 6
or G.M. db = 20 log10 (2) = 6 db. Ans.
(b) Phase Margin : To determine phase margin for a given value K = 6, locate a point
jω1 on the imaginary axis of the root locus plot such that G(jω1) H(jω1) = 1, thus, ω1 is the
gain crossover frequency. The point jω1 is located as below
K
G(s) H(s) = put s = jω
s (s + 1) (s + 3)
STABILITY ANALYSIS OF CONTROL SYSTEMS 341
K
∴ G(jω) H(jω) =
jω ( jω + 1) ( jω + 3)
At the gain crossover frequency ω1, G(jω1) H(jω1) = 1.
To determine the gain crossover frequency, put G(jω1) H(jω1) = 1 and the given
value K = 6
K 6
∴ 1= or 1=
jω1 ( jω1 + 1) ( jω1 + 3) ω1 ω12 + 1 ω12 + 9
or ω12 (ω12 + 1) (ω12 + 9) = 36
Above equation is nearly satisfied at ω1 = 1.2 rad/sec.
Hence, ω1 = 1.2 rad/sec. is the gain crossover frequency.
Phase margin = 180° + ∠G(jω1) H(jω1) = 180° + (− θ p1 − θ p2 − φ p3 )
⎛ 1.2 1.2 ⎞
= 180° + ⎜ − 90° − tan −1 − tan −1 ⎟
⎝ 1 3 ⎠
= 180° (90° + 50.2° + 21.8°) = 180° (162°) = 18°
Hence, P.M. = 18°. Ans.
*Note : θ p , θ p , and θ p are taken as ve angles as they are contributed by the poles of G(s) H(s).
1 2 3
Example 7.24.9. Sketch the root locus plot for the open-loop transfer function given
below :
K (s 2 + 4)
G(s) H(s) =
s (s + 2)
Calculate the value of K at (a) breakaway point (b) s = 0.69 + j0.9.
Solution. 1. The root locus branches start (K = 0) from open-loop poles s = 0 and s = 2.
2. The root locus branches terminate (K = ∞) on open-loop zeros s = + j2 and s = j2.
3. P = Z = 2, hence, the number of root locus branches N = 2.
4. The root locus branches on the real axis appear between s = 0 and s = 2.
5. The breakaway point is determined below :
The characteristic equation is
K ( s2 + 4)
1 + G(s) H(s) = 0 ∴ 1+ =0
s (s + 2)
or s(s + 2) + K (s2 + 4) = 0
s(s + 2) s2 + 2s
∴ K= =
(s2 + 4) s2 + 4
dK [(s2 + 4) (2s + 2) − (s2 + 2s) (2s)] 2s2 + 8 s + 8 2 s2 − 8 s − 8
= = =
ds (s2 + 4)2 (s2 + 4)2 (s2 + 4)
dK
Put =0
ds
∴ 2s2 8s 8 = 0 or s2 4s 4=0
342 LINEAR CONTROL SYSTEMS
Imj
+ 4 ± 42 − 4 × 1 × (− 4)
∴ s= 225°
2
j2
= + 2 ± 2 2 = 4.82, 0.82.
¥
®
As the point s = 0.82 lies on the root lo-
K
cus, therefore, s = 0.82 is a breakaway point. K = 0.464 j1
6. The root locus branches for K → ∞ does s = – 0.69 + j0.9
not terminate on zeros at infinity, therefore, as-
ymptotes does not appear on the root locus plot.
7. As the root locus branches terminate (K 0 Re
–2 – 0.82
= ∞) on zeros s = ± j2 located on imaginary axis, it K=0 K = 0.2 K=0
does not cross the same.
8. The angle of arrival of root locus
branches at complex zeros (s = ± j2) is calculated – j1
below :
θa(+ j2) = 180° (θz θp)
K
®
– j2
= 180° [θ z2 (θ p1 + θ p2 )]
¥
–225°
⎧ ⎡ −1 ⎛ 2 ⎞ ⎤ ⎫
= 180° ⎨90° − ⎢90° + tan ⎜ ⎟ ⎥ ⎬
⎩ ⎣ ⎝ 2 ⎠⎦ ⎭ Fig. 7.24.11. Root locus plot for example 7.24.9.
= 180° [90° (90° + 45°)] = 225°
θa( j2) = 180° (θz θp) = 180° [θ z1 (θ p1 + θ p2 )]
⎧ ⎡ −1 ⎛ 2 ⎞ ⎤ ⎫
= 180°
⎨270° − ⎢270° tan ⎜ ⎟ ⎥ ⎬
⎩ ⎣ ⎝ 3 ⎠⎦ ⎭
= 180° [270° (270° 45°)] = 135° or 225°
With the data calculated in steps (1) to (8) the root locus plot is drawn as shown in
Fig. 7.24.12.
(a) The value of K at breakaway point s = 0.82 is calculated below :
K (s2 + 4)
G(s) H(s) =
s (s + 2)
put s = 0.82 and G(s) H(s) = 1
∴ 1 = K [(− 0.82)2 + 4]
(− 0.82) (− 0.82 + 2)
K (4.67) (0.82)(1.18)
or 1 = or K= = 0.2. Ans.
(0.82) (1.18) 4.67
(b) The value of K at s = ( 0.69 + j0.9) is calculated below :
K (s2 + 4)
G(s) H(s) = put s = ( 0.69 + j0.9) and G(s) H(s) = 1
s (s + 2)
K 14.93 K × 3.86
= =
1.28 × 2.5 1.13 × 1.581
1.13 × 1.581
∴ K= = 0.464. Ans.
3.86
Example 7.24.10. (a) A control system is shown in Fig. 7.24.12.
Rf if
Kf N1 N4 q
ve KA vf Lf L
JL
qm fL
N3
N2
qr qc
N5
2
4 ⎛ 1 1⎞ 4
= 0.1 × 10 + 1000 × 10
⎜ × ⎟
⎝ 10 5 ⎠
= 0.1 × 10 4 + 0.4 × 10 4 = 0.5 × 10 4 kg-m2
The equivalent viscous friction coeff. ref. to motor shaft side is determined below :
2
⎛N N ⎞
feq = fm + fL ⎜ 1 × 3 ⎟
⎝ N2 N4 ⎠
2
⎛ 1 1⎞
4 4
= 0.25 × 10 ⎜ + 1875 × 10
× ⎟
⎝ 10 5 ⎠
= 0.25 × 10 4 + 0.75 × 10 4 = 1 × 10 4 Nm/rad/sec.
(a) The block diagram representation is shown in Fig. 7.24.13 (a) and (b).
The take-off point before block N3/N4 in Fig. 7.24.13 (a) is shifted after the same block
as shown in Fig. 7.24.13 (b).
(a)
(b)
Fig. 7.24.13 (a) and (b) Block diagram for example 7.24.10.
On substituting given numerical values the block diagram of Fig. 7.24.13 (b) is re-
drawn and shown in Fig. 7.24.14 (a) and (b). Ans.
(b) The forward path transfer function is determined below :
0.05 K A × 2 × 1 1
G(s) = ×
(10 + 2s) (0.5 × 10 −4 s2 + 1 × 10 −4 s) 50
0.1 K A
=
−4
50 × 10 s (2s + 10) (0.5s + 1)
20 K A 20 K A
= =
s (2s + 10) (0.5s + 1) 2 × 0.5 s (s + 2) (s + 5)
or G(s) = 20 K A
, ∵ H(s) = 0.5
s (s + 2) (s + 5)
The open-loop transfer function is :
20 K A × 0.5 10 K A
G(s) H(s) = or G(s) H(s) =
s (s + 2) (s + 5) s (s + 2) (s + 5)
STABILITY ANALYSIS OF CONTROL SYSTEMS 345
(a)
(b)
Fig. 7.24.14 (a) and (b) Block diagram with numerical values for example 7.24.10.
put 10KA = K
K
∴ G(s) H(s) = ⋅ Ans.
s (s + 2) (s + 5)
The root locus plot for the open-loop transfer function given above is drawn in
Fig. 7.24.15 as per the procedure given below :
1. The root locus branches start (K = 0) from the open-loop poles s = 0, s = 2 and s = 5.
2. As there are no finite open-loop zeros, the root locus branches terminate (K = ∞) at
infinity.
3. P = 3, Z = 0, therefore, the number of root locus branches N = 3.
4. On real axis, the root locus appears between open-loop poles s = 0 and s = 2 and
beyond s = 5.
5. Between open-loop poles s = 0 and s = 2, there exists a breakaway point as calcu-
lated below :
The characteristic equation is given by :
1 + G(s) H(s) = 0
K
∴ 1+ =0 or s3 + 7s2 + 10s + K = 0
s (s + 2) (s + 5)
∴ K = (s3 + 7s2 + 10s)
dK dK
= (3s2 + 14s + 10) put =0
ds ds
∴ 3s2 + 14s + 10 = 0
or 2
s + 4.66s + 3.33 = 0
Imj
¥
K
j 3.16
K = 70
j2
s = – 0.7 + j1.3
K = 12.2 j1.3
¥ K K= 0 – 2.33 K= 0 q =60° K= 0
–5 –4 –2 –0.7 0 Re
s = – 0.88 – j 1.3
K = 4.06
–j2
K = 70
– j 3.16
¥
Fig. 7.24.15(a). Root locus plot for example 7.24.10.
(2k + 1)
6. The angle of asymptotes is given by × 180°, where K = 0, 1, 2
P−Z
(2 × 0 + 1) (2 × 1 + 1) (2 × 2 + 1)
(i) × 180° = 60°, (ii) × 180° = 180° and (iii) × 180° = 300°.
3−0 3−0 3−0
7. The asymptotes intersect on real axis at :
ΣP − ΣZ (0 − 2 − 5) − (0)
x= = or x = 2.33
P−Z 3−0
8. The value of K for marginal stability is determined by applying Routh criterion to
the characteristic equation :
s (s + 2) (s + 5) + K = 0 or s3 + 7s2 + 10s + K = 0
The Routh array is tabulated below :
s3 1 10
s2 7 K
70 − K
s1
7
s0 K
The value of K at which root locus crosses imaginary axis is given by :
70 − K
=0 or at K = 70
7
STABILITY ANALYSIS OF CONTROL SYSTEMS 347
The value of s at the crossing point of root locus and imaginary axis is calculated
below :
7s2 + K = 0 or 7s2 + 70 = 0 or s = ± j 10 = ± j3.16
(c) (i) At the breakaway point s = 0.88, the damping ratio is ζ = 1. The value of K at
s= 0.88 is determined below :
K
1 = put s= 0.88
s (s + 2) (s + 5)
K K
∴ 1 = =
(− 0.88) (− 0.88 + 2) (− 0.88 + 5) (− 0.88) (1.12) (4.12)
∴ K = (0.88) × (1.12) × (4.12) or K = 4.06
∵ 10KA = K
K 4.06
∴ KA = = = 0.406. Ans.
10 10
(ii) A line at an angle of θ = 60° with real axis corresponds to damping ratio ζ = 0.5
(cos 1 ζ = cos 1 0.5 = 60°) and cuts the root locus at s = ( 0.7 + j1.3). The value of K at
s = ( 0.7 ± j1.3) is determined below :
K
1 = put s = ( 0.7 + j1.3)
s (s + 2) (s + 5)
K
∴ 1 =
(− 0.7 + j1.3) (− 0.7 + j1.3 + 2) (− 0.7 + j1.3 + 5)
K
=
(− 0.7 + j1.3) (1.3 + j1.3) (4.3 + j1.3)
K
=
(0.49 + 1.69) (1.69 + 1.69) (18.49 + 1.69)
K K K
= = =
2.18 3.38 20.18 148.7 12.2
∴ K = 12.2 ∵ K = 10 KA
K 12.2
∴ KA = = = 1.22. Ans.
10 10
Example 7.24.10 (b). The open-loop transfer function of a non unity feedback control
system is given below:
K
G(s) H(s) =
s(s + 2)(s + 4)
The damping ratio ξ = 0.5. The variation of K is from 0 to ∞.
Determine intersection point of the damping ratio line with the complex part of the
root locus by analytical method.
Determine the value of dominant roots of the characteristic equation and the value of
K at the aforesaid intersection point, the value of the third real root.
348 LINEAR CONTROL SYSTEMS
or K = s( s + 2)(s + 4) s=
2
+ j
2
3 3
=
FG 2 + j 2 IJ FG 2 + j 2 + 2IJ FG 2 + j 2
+4
IJ
H 3 3K H 3 3 K H 3 3 K
=
FG 2 + j 2 IJ FG 4 + j 2 IJ FG 10 + j 2 IJ
H 3 3K H3 3K H 3 3K
= FG 4 + 4 IJ FG 16 + 4 IJ FG 100 + 4 IJ
H 9 3K H 9 3K H 9 3K
K = 8.29
K 8.29
∴ Simplifying, G(s) H(s) = ∴ G(s) H(s) =
s(s + 2)(s + 4) s(s + 2)(s + 4)
350 LINEAR CONTROL SYSTEMS
The third real root lies beyond s = 4 on the s-plane ve real axis. The third real root
satisfies the magnitude condition of root locus
8.29
∴ =1
s(s + 2)(s + 4)
The above equation is satisfied at s = 0.466 to a fair approximation. Hence the third
real root is located at s = 4.66. Ans.
The closed-loop poles i.e. roots of the characteristic equation 1 + G(s) H(s) = 0 obtained
2 2
by the analytical solution are s1 = 4.666, s2, s3 = ± j .
3 3
s0 0.1 K
The value of K at which the root locus plot crosses the imaginary axis is determined by
equating the first term in s1 row to zero, therefore,
( 1.25 K2 + 0.63 K 0.066) = 0
− 0.63 + 0.632 − 4 × (− 1.25) × (− 0.066)
∴ K=
2 × (− 1.25)
− 0.372 − 0.888
K= , = 0.148, 0.352.
− 2.5 − 2.5
The root locus crosses the imaginary axis at K = 0.148 and K = 0.352.
For finding the crossing points the auxiliary equation formed from s2 terms is solved,
by putting K = 0.148 and K = 0.352 respectively, therefore,
352 LINEAR CONTROL SYSTEMS
K=¥
(–0.5, + j0.59)
j 0.56
K=0 K = 0.352
– 55.8°
j 0.4
Breakaway
Point j 0.2
K = 0.148
180° 60° Breakaway
K=¥ K = ¥K = 0 Point
–0.5 –0.37 –0.23 –0.1 0 0.08 0.2 Re
300° K=0
–j0.2
K = 0.148
–j0.4
+ 55.8°
K=0 –j0.56
(–0.5, – j0.59) K = 0.352
K=¥
− 1 ± 12 − 4 × 1
open-loop zero : s = 0, open-loop poles : s = = 0.5 ± j 0.866
2
The number of root contour branches = 2
The starting point (α = 0) of root contours is at s = 0.5 ± j 0.866
The terminating point (α → ∞) of root contours is at s = 0 and α → ∞
The angle of asymptotes :
(2k + 1)
∠ Asymptotes = × 180° ; k = 0
P−Z
(2 × 0 + 1)
= × 180° = 180°
2 −1
The root contour is present on entire ve real axis.
The characteristic equation is :
αs
1+ =0
s2 + s + 1
s2 + s + 1
or α= −
s
dα ⎪⎧ s (2 s + 1) − (s2 + s + 1) × 1 ⎪⎫ ⎪⎧ s2 − 1 ⎪⎫
∴ = ⎨ ⎬ = ⎨ 2 ⎬
ds s2
⎩⎪ ⎭⎪ ⎩⎪ s ⎭⎪
The breakaway point is determined using dα = 0
ds
∴ s2 1 = 0 or s=±1
The breakaway point is identified at s = 1 as it lies on root counter branch.
The angles of departure from complex poles s = 0.5 ± j 0.866 are determined below :
θd = 180° {(θ p2 − φ z )}
( − 0.5 + j 0.0866)
⎧ ⎛ −1 0.5 ⎞ ⎫
= 180° ⎨90° − ⎜ 90° + tan ⎟ ⎬ = 180° {90° (90° + 30°)} = 210°
⎩ ⎝ 0.866 ⎠ ⎭
STABILITY ANALYSIS OF CONTROL SYSTEMS 355
θd = 180° {(φ p1 − φ z )}
( − 0.5 − j 0.0866)
⎧ ⎛ −1 0.5 ⎞ ⎫
= 180°
⎨ 90° − ⎜ − 90° tan ⎟⎬
⎩ ⎝ 0.866 ⎠ ⎭
= 180° { 90° ( 90° 30°)} = 150°
As per data calculated above the root contour plot is drawn and shown in Fig. 7.25.2.
–0.5 + j0.866 jw
210°
a=0
¥¬a a=1
–1
a=¥ s
Breakaway
point
150°
a=0
–0.5 – j0.866
α (−1)
∴ 1= 2
(−1) + (− 1) + 1
∴ α = 1. Ans.
Example 7.25.2. The block diagram of a unity feedback control system is shown in
figure given below :
R(s) + 5s + b 1 C(s)
– s s(s + 4)
⎧ ⎛ 1 2 ⎞⎫ 1 2
φ( 2 + j 1) = 180° (θ p2 + θ p3 ) = 180° ⎨(+ 90°) + ⎜ 90° + tan ⎟ ⎬ = tan = 63.43°
⎩ ⎝ 1 ⎠⎭ 1
The angle of departure from the pole ( 2 j1) is calculated below :
φ( 2 j 1) = 180° (θ p1 + θ p3 )
⎧ ⎛ −1 2 ⎞ ⎫
= 180° ⎨(− 90°) + ⎜ − 90° − tan ⎟ ⎬ = 296.57° = (360° 296.57°) = 63.43°.
⎩ ⎝ 1 ⎠⎭
jw b®¥
+ jÖ5
–2 + j1 s = (–0.66 + j0.88)
b=0 – 63.43°
–1
q = cos z
¥¬b = 53.1°
0 b=0 s
–1.66 –1.33
–1.01
b=0 63.43°
–2 –j1
–jÖ5
b®¥
Fig. 7.25.4. Root contours for example 7.25.2.
358 LINEAR CONTROL SYSTEMS
As per the data calculated above the root contour is plotted as shown in Fig. 7.25.4.
The damping ratio line drawn at an angle
θ = cos 1 ζ = cos 1 0.6 = 53.1°
intersects root contour plot at s = ( 0.66 + j 0.88). As the point s = ( 0.66 + j 0.88) lies on the
root contour.
Therefore, G1 (s) H1 (s) s = ( 0.66 + j 0.88) = 1
β
∴ 1 =
(− 0.66 + j 0.88 ) {(− 0.66 + j 0.88)2 + 4 (− 0.66 + j 0.88) + 5}
β
or 1= ∴ β = 11.7. Ans.
1.1 (1.21 + 4 × 1.11 + 5)
PROBLEMS
7.1. Determine the stability of the systems whose characteristic equations are given
below :
(a) s4 + 10s3 + 30s2 + 100s + 25 = 0 (b) s4 + 8s3 + 36s2 + 80s + 20 = 0
(c) s5 + 13s4 + 54s3 + 82s2 + 170s + 330 = 0
(d) s5 + 4s4 + s3 + 4s2 s + 6 = 0 (e) s4 + s3 + 5s2 + 4s + 4 = 0.
7.2. For the characteristic equations given below determine the limiting value of K for
stability
(a) s3 + 4Ks2 + (K + 4) s + 5K = 0 (b) s4 + 7s3 + 10s2 + K = 0
(c) s4 + 4s3 + 6s2 + (K + 5) s + 2K = 0.
7.3. Determine the number of roots having positive real part of the characteristic
equation whose block diagram is shown in Fig. 7-P-1.
R(s) + 26 C(s)
2
s(s + 2) (s + 2s + 2)
–
(s + 1)
7.4. Use Routh criterion to investigate the stability of a control system having open-loop
transfer function given below
K (s + 2)
G(s) H(s) = .
s (s + 3) (s2 + 2s + 3)
7.5. Determine the value of K for the following equation such that the roots lie towards
the left of line s = 1 in s-plane.
s3 + 3s2 + 3s + K = 0.
STABILITY ANALYSIS OF CONTROL SYSTEMS 359
Ks2
(b) Sketch Nyquist plot having given, G(s) H(s) = and examine closed-
3
s + 4s + 4
loop stability in terms of parameter K.
7.10. For the open-loop transfer function given below sketch the Nyquist plot when (a) K =
1.5 (b) K = 2.5 and comment on the closed-loop stability.
K (s + 4)
G(s) H(s) = ⋅
s (s − 2)
7.11. The characteristic equation of a unity feedback system is given by, 0.5 Ts3 + (0.5 + T)
s2 + s + 10 = 0
(a) Determine the open-loop transfer function
(b) Apply Nyquist criterion to the open-loop transfer function to determine the value
of T such that the system is stable.
7.12. The open-loop transfer function of a feedback control system is given by, G(s) H(s) =
K
2
s + s −2
Plot the Nyquist plot and show that the closed-loop system is stable if K ≥ 2.
(Hint : Factor the denominator to find P+).
360 LINEAR CONTROL SYSTEMS
7.13. Plot the gain-phase plot from the data given below :
Determine :
(a) G.M. and P.M.
(b) the factor by which the gain is to be multiplied to have G.M. = 10 db
(c) the factor by which the gain is to be multiplied to have P.M. = 30°.
7.14. The gain frequency measurements for a control system are given below. Sketch the
gain phase plot and determine the gain margin, phase margin
ω rad/sec 1 2 4 8 10 20
G(jω) 4.8 2.3 0.9 0.26 0.16 0.026
∠G(jω) 106° 123° 141° 186° 198° 229°
Fig. 7-P-3 (a) (i) Logarithmic plot for example 7.18 (a) (i).
STABILITY ANALYSIS OF CONTROL SYSTEMS 361
48 db
–4
0d
b/d
–2
8 db 0d
b /d w2
0 w1
0.1 0.6 w
–7.5 db
–4
0d
b/d
Fig. 7-P-3 (a) (ii) Logarithmic plot for example 7.18 (a) (ii).
(b) The magnitude (db)-frequency (rad/sec) plot given in Fig. 7-P-3b. Determine the
open-loop transfer function and sketch the phase angle frequency plot. Obtain
the gain margin and phase margin and determine from the Bode plot the value of
gain such that the system is marginally stable.
db
20 – 20 db/d
G(jw) H(jw)
0
1 2 6 10 25 100
–4 w
0d
b /d
–7.5 db
–6
0d
b/d
Fig. 7-P-3 (b) Magnitude (db)-frequency (rad/sec) plot for problem 7.18 (b).
7.19. Sketch the root locus plot for each of the following open-loop transfer function by
determining :
(i) Number of root locus branches
(ii) Angle of asymptotes
(iii) Intersection points of asymptotes on real axis
K K
(a) G(s) H(s) = (b) G(s) H(s) =
s (s + 2.5) 2
s (s + 4)
K (s + 1) K (s + 4)
(c) G(s) H(s) = (d) G(s) H(s) = 2 ⋅
s2 (s + 4) s (s + 1)
7.20. The transfer functions of a feedback control system are given below
K K
(a) G(s) = and H(s) = 1 (b) G(s) H(s) =
s (s + 1) (s + 3) (s + 1) (s + 10) (s + 30)
Sketch the root locus plot as K is varied from 0 to ∞ and determine :
(i) the value of K such the system becomes unstable.
(ii) the value of K such that the system is critically damped.
362 LINEAR CONTROL SYSTEMS
7.21. Sketch the root locus plots for the following open-loop transfer functions and
determine the value of K such that the damping ratio is 0.707.
K (s + 20) K (s + 30)
(a) G(s) H(s) = (b) G(s) H(s) = ⋅
s (s + 1) (s + 10) (s + 30) (s + 1) (s + 10) (s + 20)
7.22. The block diagram of a control system is given in Fig. 7-P-4. Plot the root locus as a
function of parameter K and comment on the stability.
(s + 1)
CONTENTS
• Phase-Lead Compensation
• Phase-Lag Compensation
• Phase-Lag-Lead Compensation
• Concluding Remarks
• Feedback Compensation
• Root Locus Approach
• Solved Examples
363
364 LINEAR CONTROL SYSTEMS
H(s)
Imj Imj
K=¥
¥
K=¥ K= K=0
K=¥
K=0 K=0 K=0 Re K=0 K=0 Re
K= Zero
¥ added K=¥
R1
eo
ei R2
Imj
db
180° Eo(jw)/Ei(jw)
Eo (jw)
Ei (jw)
fm 0°
1/T wm 1/aT
X 0 db 180°
1 1 Re – 20 log10a w
d
–
aT
–
0 db/
T +2
Fig. 8.1.2. Pole-zero configuration and Bode plot for phase-lead network.
The two corner frequencies are
1
ω= lower corner frequency
T
1
ω= upper corner frequency.
αT
The maximum phase-lead φm occurs at mid-frequency ωm between upper and lower
corner frequencies.
LM
log10 ωm= 1 log 10 1 + log 10 1 FG IJ FG IJ OP
∴
2 T N αT H K H KQ
366 LINEAR CONTROL SYSTEMS
1
∴ ωm = ...(8.3)
αT
The phase angle ∠Eo(jω)/Ei(jω) can be calculated as
Eo ( jω) 1 1
∠ = tan (ωT ) tan (ωαT )
Ei ( jω)
1
At ω = ωm = the phase angle is φm,
αT
φm = tan 1 FG 1 . T IJ tan FG 1 1
αT
IJ
∴
H αT K H αT K
or φm = tan 1
FG 1 IJ tan ( α ) 1
H αK
1
α
α
∴ tan φm =
1
1+ . α
α
1 α
or tan φm = ...(8.4)
2 α
1 α
and sin φm = ...(8.5)
1+α
From pole-zero configuration of phase-lead network (Fig. 8.1.2) it is observed that the
zero is nearer to origin as compared to pole, hence the effect of zero is dominant, therefore,
the phase-lead network, when introduced in cascade with forward-path of a transfer
function, the phase shift is increased.
The Bode plot of phase-lead network (Fig. 8.1.1) reveals that the lead network allows
to pass high frequencies and low frequencies are attenuated.
As the gain is reduced at low frequencies additional gain is needed in the system to
account for the reduction in gain.
where β>1 ei R2
eo
R + R2
β= 1 C
R2
and T = R2 C
Fig. 8.2.1. Phase-lag network.
COMPENSATION OF CONTROL SYSTEMS 367
The transfer function given by Eq. (8.6) can be expressed in sinusoidal form as
Eo ( jω) 1 + jωT
= ...(8.7)
Ei ( jω) 1 + jωβT
The pole-zero configuration and Bode plot for transfer function (8.7) is shown in
Fig. 8.2.2.
db
Imj
180°
Eo (jw)
Eo(jw)/Ei(jw)
Ei (jw)
fm
0°
1 1
bT wm w
T –180°
O 0 db
1 1 Re –2 – 20 log10 b
– – 0d
T bT b/d
Fig. 8.2.2. Pole-zero configuration and Bode plot for phase-lag network.
The two corner frequencies are
1
ω= upper corner frequency
T
1
ω= lower corner frequency
βT
The maximum phase-lag φm occurs at mid frequency ωm between upper and lower
corner frequencies.
LM
log10 ωm = 1 log 10 1 + log 10 1
FG IJ FG IJ OP
∴
2 MN
βT T H K H K PQ
1
∴ ωm = ...(8.8)
βT
The phase angle ∠Eo(jω)/Ei(jω) can be calculated as
Eo ( jω) 1 1
∠ = tan (ωT ) tan (ωβT )
E i ( jω )
1
At ω = ωm = , the phase angle is φm
βT
1 β
∴ tan φm = ...(8.9)
2 β
1 β
and sin φm = ...(8.10)
1+ β
Note. As β > 1, (1 β)/(1 + β) is negative indicating that φm is negative which means
that the phase angle is lagging.
From the pole-zero configuration of phase-lag network (Fig. 8.2.2) it is observed that
the pole is nearer to origin compared to zero, hence the effect pole is dominant, therefore, the
368 LINEAR CONTROL SYSTEMS
phase-lag network when introduced in cascade with forward-path of a transfer function, the
phase shift is reduced. The Bode plot of phase-lag network (Fig. 8.2.2) reveals that the lag
network allows to pass low frequencies and high frequencies are attenuated.
O O
1 1 1 1 0 Re
– – –
aT1 T1 T2 bT2
Lead Lag
db
Eo(jw)/Ei(jw)
180°
Eo (jw)
Ei (jw)
0°
1/bT2 1/T2 1/T1 1/ aT1
0 db – 180°
– w
20
d b/d
db 20
/d +
a < 1, b > 1, ab = 1, T1 < T2
Bode Plot
Fig. 8.3.2. Pole zero configuration and Bode plot phase-lag-lead network.
COMPENSATION OF CONTROL SYSTEMS 369
The transfer function given by (8.11) can be expressed in sinusoidal form as
Eo ( jω) (1 + jωT1 ) (1 + jωT2 )
= ...(8.12)
Ei ( jω) (1 + jωαT1 ) (1 + jωβT2 )
The pole zero configuration and Bode plot for the transfer function (8.12) is shown in
Fig. 8.3.2.
R(s) + K C(s)
s(1 + sT)
–
R(s) + + K C(s)
s(1 + sT)
– –
skt
R(s) + K C(s)
s(1 + sT)
–
1 + sKt
K=¥
Imj
K C(t)
K=0 q K=0
1 0 Re
–
T
r(t)
cos q = x1
Kt = 0
0 t
K=¥
Fig. 8.5.4. Root locus plot and time response of uncompensated system (Kt = 0).
COMPENSATION OF CONTROL SYSTEMS 371
(2) Kt < T
The poles of the transfer function (8.17) are at s = 0 and s = ( 1 /T), the zero at s =
( 1/Kt) such that Kt < T.
Fig. 8.5.5 shows the root locus plot and the corresponding time response of transfer
function (8.17) for Kt < T.
Imj
Kt < T
cos q = x2
C(t)
r(t)
O x2 > x 1 t
Fig. 8.5.5. Root locus plot and time response of compensated system (Kt < T).
From Fig. 8.5.5 it is noted that for the same value of K as in case (1). The damping is
reduced.
(3) Kt > T
The poles of the transfer function (8.17) are at s = 0 and s = ( 1/T). The zero is at
s = ( 1/Kt) such that Kt > T.
Fig. 8.5.6 shows that root locus plot and the corresponding time response of transfer
function (8.17) for Kt > T. As the roots have no imaginary part, the closed-loop time response
is overdamped.
The analysis given above reveals that the time response of a control system can be
adjusted by using tachogenerator feedback.
372 LINEAR CONTROL SYSTEMS
Imj
C(t)
r(t)
t
Fig. 8.5.6. Root locus plot and time response of compensated system (Kt > T).
Example 8.6.1. The open-loop transfer function of a unity feedback control system is
K
given by, G(s) =
s(1 + 0.2s)
Design a suitable compensator such that the system will have Kv = 10 and P.M.= 50º.
K
Solution. K v = lim sG(s) = lim s . =K
s→0 s→0 s(1 + 0.2s)
∴ K = Kv = 10
The open-loop transfer function is thus
10
G(s) =
s(1 + 0.2s)
The Bode plot for the above transfer function is plotted in Fig. 8.6.1. The frequency
range selected is ω = 1 rad/sec to ω = 10 rad/sec. The phase angle is given by
∠G(jω) = 90º tan 1 (0.2ω)
Therefore,
ω (rad/sec) 1 5 10 50
∠G(jω)º 101.3 135 153.3 177.3
COMPENSATION OF CONTROL SYSTEMS 373
From the Bode plot (Fig. 8.6.1) it is noted that uncompensated system is stable having
a phase margin of 30º and the gain crossover occurs at ω = 6.75 rad/sec.
45°
fm
0°
ÐGc(s)
20 db
–2 – 45°
0d
b/d
ÐG(s)
ÐG¢(s) – 90°
Phase angle
10 db Gc(s)
Mag. db
– 135°
30°
50°
0 db P.M.
– 180°
5.5
1 2 5 10 13.7 20 50 100
8.75
w
– 4 db – 225°
–
– 10 db
40
|G¢(s)|
db
/d
|G(s)|
1 α ωm ωm 8.7
The upper corner frequency = = = = = 13.7 rad/sec.
αT α α 0.4
The transfer function of the phase lead compensation network with amplifier gain
(1/α) = 2.5 is
FG1 + jω 1 IJ
Gc(jω) =
1 + jωT H 5.5 K (1 + j 0.18ω)
1 + jωαT
=
FG1 + jω × 0.4 × 1 IJ = (1 + j0.072ω) . Ans.
H 5.5 K
The open-loop transfer function of the compensated system is
(1 + 0.18s) 10
G′(s) = Gc(s) G(s) = . . Ans.
(1 + 0.072s) s(1 + 0.2s)
The Bode plot for the compensated system is shown in Fig. 8.6.1 and the phase margin
of the compensated system is 50º.
Example 8.6.2 Design a phase lead compensation network for a system having open-
loop transfer function given below such that the system will have acceleration error
coefficient Ka = 100 sec 2 and suitable phase margin for stable operation
K
G(s) H(s) = 2
.
s (1 + 0.05s)
K
Solution. K a = lim s2 G(s) H (s) = lim s2 . 2
=K
s→ 0 s→ 0 s (1 + 0.05s)
∴ K = Ka = 100
The open-loop transfer function is thus
100
G(s)H(s) = 2
s (1 + 0.05s)
The Bode plot for the above transfer function is plotted in Fig. 8.6.2. The frequency
range selected is ω = 1 rad/sec to ω = 100 rad/sec. The phase angle is given by
∠G(jω) H(jω) = 180º tan 1 (0.05ω)
Therefore,
ω (rad/sec) 1 10 20 50 100
∠G(jω)H(jω)º 182.8 206.5 225 248.2 258.6
From the Bode plot (Fig. 8.6.2) it is noted that the P.M. is 26º and the system is
unstable. A suitable phase margin varying between 20º to 50º is assumed, thus
(a) Assuming P.M. = 20º the required phase lead is
φm = 20º ( 26º) + Tolerance
As the system is type 2, the phase angle decreases rapidly hence a higher tolerance
say 10º is considered
∴ φm = 20º ( 26º) + 10º = 56º
COMPENSATION OF CONTROL SYSTEMS 375
40 db
ÐGc(s) 90°
30 db fm= 56°
–4 0°
0d
20 db b/d
Gc(s)
10 db – 90°
Phase Angle
ÐG¢(s) P.M. 20°
Mag. db
10 w
0 db – 180°
1 2 5 16.5 20 54 100
ÐG(s)H(s) – 26°
– 10 db
– 10.3 db – 270°
– 20 db G¢(s)
– 360°
–
G(s)H(s)
60
db
/d
Fig. 8.6.2. Bode plot for example 8.6.2 (φm = 56°).
(b) Assuming P.M. = 50º the required phase lead is
φm = 50º ( 26º) + 10º = 86º
In case (a) the phase lead contributed is 56º whereas in case (b) the phase lead is 86º,
whenever the phase lead to be contributed by compensation network exceeds 60º then it is
preferred to use two networks in cascade each contributing phase lead of φm/2.
The compensation networks for cases (a) and (b) are designed below :
Case (a) φm = 56º
1 sin 56º 1 0.829 0.17
α= = = = 0.093.
1 + sin 56º 1 + 0.829 1.829
The maximum phase-lead 56º, is contributed at mid-frequency ωm of the phase-lead
network. Allowing an amplification of 1/α the magnitude of phase-lead compensation
network transfer function at ω = ωm is determined below :
1
∵ ωm =
αT
1
1+ .T
1 + jω m T αT 1
∴ Gc ( jω m ) = = =
1 + j ω m αT 1 α
1+ . αT
αT
1 1
In decibel, 20 log 10 Gc ( jω m ) = 20 log 10 = 20 log 10 = 10.3 db.
α 0.093
From the Bode plot (Fig. 8.6.2) it is noted that for uncompensated system at a gain of
10.3 db the frequency is 16.5 rad/sec. The new gain crossover frequency for the
compensated system is selected as ω = ωm = 16.5 rad/sec.
The two corner frequencies for the phase-lead network are
1
The lower corner frequency = = α ω m = 0.093 × 16.5 = 5 rad/sec
T
376 LINEAR CONTROL SYSTEMS
1 α ωm 16.5
The upper corner frequency = = = = 54 rad/sec
αT α 0.093
The transfer function of the phase-lead network with amplifier gain (1/α) = (1/0.093) =
15.75 is
FG1 + jω 1 IJ
Gc(jω) =
(1 + jωT ) H 5K (1 + j 0.2ω)
Ans.
(1 + jωαT )
=
FG1 + jω × 0.093 × 1 IJ = (1 + j0.0186ω)
H 5K
The open-loop transfer function of the compensated system is
(1 + 0.2s) 100
G′(s) = Gc(s)G(s) H(s) = . 2 Ans.
(1 + 0.0186s) s (1 + 0.05s)
The Bode plot for the compensated system is shown in Fig. 8.6.2. The P.M. of the
compensated system is 20º and the system is stable.
Case (b). As the phase lead required is 86º (> 60º) the same is provided by two
identical phase-lead networks each contributing a phase lead of 43º, therefore,
1 sin 43º 1 0.681
α= = = 0.19
1 + sin 43º 1 + 0.681
The phase lead φm = 43º due to each phase-lead network is contributed by each
compensation network at mid-frequency ωm of the phase lead network. Allowing an
amplification of 1/α for each compensation network, the magnitude of each phase-lead
compensation network is determined below :
∵ ω = 1
m
αT
1
1+ j .T
1 + jω m T αT 1
∴ Gc ( jω m ) = = =
1 + jω m αT 1 α
1+ j . αT
αT
1 1
In decibel, 20 log10 Gc ( jω m ) = 20 log 10 = 20 log 10 = 7.2 db
α 0.19
The total magnitude due to two identical phase-lead networks is 7.2 + 7.2 = 14.4 db.
From Bode plot (Fig. 8.6.3) it is observed that for uncompensated system, at a gain of
14.4 db the frequency is 21 rad/sec. The new gain crossover frequency for the compensated
system is selected as ω = ωm = 21 rad/sec.
The two corner frequencies for the phase-lead networks are
1
The lower corner frequency = = α ω m = 019 . × 21 = 9.15 rad/sec
T
1 αω m ω m 21
The upper corner frequency = = = = 48.2 rad/sec
αT α α 0.19
The transfer function of the phase-lead network with amplifier gain (1/α)(1/α) =
(1/0.19)(1/0.19) = 27.7 is
(1 + jωT )(1 + jωT ) (1 + jωT ) 2
Gc(jω) = =
(1 + jωαT )(1 + jωαT ) (1 + jωαT ) 2
COMPENSATION OF CONTROL SYSTEMS 377
40 db ÐGc(s)
90°
fm= 86°
30 db
– 0°
40
20 db db
/d Gc(s)
– 90°
10 db
Phase Angle
ÐG¢(s) w
Mag. db
–
60
– 20 db G¢(s)
db
/d
G(s)H(s)
FG 1 + jω 1 IJ 2
=
H 9.15 K
=
(1 + j 0.11ω) 2
. Ans.
FG 1 + jω × 0.19 × 1 IJ 2
(1 + j 0.02ω) 2
H 9.15 K
The open-loop transfer function of the compensated system is
(1 + 0.11 s) 2 100
G′(s) = Gc(s)G(s)H(s) = . Ans. 2
. 2
(1 + 0.02s) s (1 + 0.05s)
The Bode plot for the compensated system is shown in Fig. 8.6.3. The phase margin of
the compensated system is 50º and the system is stable.
Example 8.6.3. The open-loop transfer function of a unity feedback control system is
given by
K
G(s) =
s(1 + 0.5s) (1 + 0.2s)
It is desired that
(i) for a unit ramp input the steady state error of the output position be less than 0.125
degrees/(degree/second)
(ii) the P.M. ≥ 30º and (iii) the G.M. ≥ 10 db
Design a suitable compensation network.
The Bode plot for the above transfer function is plotted in Fig. 8.6.4. The frequency
range selected is ω = 0.1 rad/sec to ω = 10 rad/sec. The phase angle is given by
∠G(jω) = 90º tan 1 (0.5ω) tan 1 (0.2ω)
Therefore,
ω(rad/sec) 0.1 1 2 5 10
∠G(jω)º 94 127.1 156.8 203.1 232
From the Bode plot (Fig. 8.6.4) it is observed that the gain crossover and phase
crossover frequencies are very near to each other as such the system is nearly marginally
stable, to make the system relatively more stable, i.e. with a desired P.M. ≥ 30º, the gain
crossover frequency must be shifted to a lower value and this possible by using a phase-lag
compensation network.
– 20
db/d
40 db ÐGc(s) – 0°
fm= 35°
30 db – 45 °
G(s)
ÐG(s) G¢(s)
– 90°
20 db
ÐG¢(s) w
Phase Angle
–4 – 135°
0d w
Mag. db
10 db P.M.
30° b /d
0 db 2 5 – 180°
0.3 0.4 0.1 0.18 0.5 1 1.8 8 10
12db – 225°
– 10 db – 13 db
Gc(s)
–6
0
– 270°
db
– 20 db
/d
FG 1 + jω 1 IJ
Gc(jω) =
(1 + jωT ) H 0.18 K (1 + j 5.55ω)
(1 + jωβT )
=
FG 1 + jω × 4.46 × 1 IJ = (1 + j24.7ω)
H 0.18 K
The open-loop transfer function of the compensated system is
(1 + 5.55s) 8
G′(s) = Gc(s) G(s) = .
(1 + 24.7s) s (1 + 0.5s) (1 + 0.2s)
The Bode plot for the compensated system is shown in Fig. 8.6.4 and P.M. = 30º,
G.M. = 12 db is obtained.
Example 8.6.4. The open-loop transfer function of a unity feedback control system is
given by
K
G(s) =
s (s + 2)
The system is to have 25% maximum overshoot and peak time 1.0 second. Determine
the value of K and tachometer feedback constant Kt.
R(s) + K C(s)
s(s + 2)
–
R(s) + + K C(s)
s(s + 2)
– –
sKt
tp = 1 sec (given)
π
∵ tp =
ωn 1 ζ2
π
∴ 1= ∴ ωn 1 ζ2 = π
2
ωn 1 ζ
π π
or ωn = = = 3.42 rad/sec
2
1 ζ 1 (0.4) 2
The characteristic equation for the compensated system is
s2 + (2 + KKt) s + K = 0
From the characteristic equation it is noted that
ωn = K and 2 ζωn = (2 + KKt)
∴ 3.42 = K
or K = (3.42)2 = 11.7. Ans.
2ζωn = (2 + KKt)
∴ 2 × 0.4 × 3.42 = (2 + 11.7 Kt) or Kt = 0.065. Ans.
a1 < b1 , a 2 > b2
KK c
The gain KcK and K are determined as above and Kc = .
K
2. Method no. 2
Place the compensator zero s = a coinciding with one of the real open-loop pole of the
given control system.
1
ω 1
ω
∴ θc = tan tan
σ + p1 b+σ
Having given, the values of θc, sd = σ + jω and p1, the value of compensator pole b can
be determined. As the open-loop pole is cancelled by the compensator zero placed at p1 the
order of the characteristic equation remains unchanged.
Method no. 3
Place the zero a of the compensator just below the desired closed-loop pole sd as
shown in Fig. 8.7.5.
Example 8.7.1. The forward path transfer function of a unity feedback control
system is given below:
K
G(s) =
s( s + 9)
COMPENSATION OF CONTROL SYSTEMS 385
Solution.
1. The desired closed-loop pole sd
ξπ
1 ξ2
%Mp = e
ξπ
9.4 1 ξ2
∴ = e
100
From the above relation the value of the damping ratio is calculated as ξ = 0.6. The
desired closed-loop is given by
sd = ξωn + jωn 1 ξ2
∵ ξ = 0.6 and ωn = 12 rad./sec.
∴ sd = 0.6 × 12 + j 12 1 0.6 2
or sd = 7.2 + j 9.6
2. Uncompensated root locus plot
As per the root locus plot procedure the root locus for the open-loop transfer function
K
G(s) =
s(s + 9)
is shown in Fig. 8.7.6.
K ¥
jw
sd = –7.2 + j 9.6
K=0 q K=0
–9 sb = –4.5 0 s
K
or θc + = 180°
( 7.2 + j 9.6) ( 7.2 + j 9.6 + 9)
FG
tan 1
9.6
+ tan 1
9.6 IJ
or θc
H 7.2 1.8
= 180°
K
∴ θc [(180° 53.13°) + 79.37°] = 180°
Simplifying, θc = 26.24°
4. To determine the parameters a and b of phase-lead compensator.
Graphical method:
The desired closed-loop pole (sd = 7.2 + j 9.6) is shown in Fig. 8.5.7 at P.
∠OPA = ∠POM
OD FG IJ 7.2
∴ ∠OPA = 90° + tan 1
PD H K
= 90° + tan 1
9.6
= 126.86°
sd = –7.2 + j 9.6 jw
P
A j 9.6
a
= 63.43°
2
qc
= 13.12°
2 q c qc
2
2
a = 126.86°
–b q1 q2
–a
B C E D O s
–7.2
KK c (s + 9.5)
∴ =1
s(s + 9)(s + 15.2) s = sd = 7.2 + j 9.6
Example 8.7.2. Determine the transfer function of a cascade compensator for a unity
feedback control system having the following open-loop transfer function,
K
G(s) =
(s + 2)(s + 4)
Following specifications are to be satisfied:
(i) %Mp = 16.3%
(ii) Reduction in setting time 60%
(iii) The order of the characteristic equation to remain the same.
∴ K = (s + 2)(s + 4) s = s1 = 3+ j3 3
= | ( 3 + j 3 3 + 2) ( 3 + j 3 3 + 4) |
= 28
2. Compensated case
4
ts = , ∵ ξωn = 3 (root locus breaks away)
1
ξω n1
4
∴ ts = sec.
13
ts = 0.6 ts (given)
2 1
4
∴ ts = 0.6 × = 0.8 sec.
2
3
4 4 4
Also, ts = , ∴ ωn = = = 10 rad./sec.
2
ξω n2 2 ξts2 0.5 × 0.8
Hence, the desired closed-loop pole sd for the compensated case is
sd = ξωn + jωn 1 ξ2
2 2
= 0.5 × 10 + j 10 1 0.5 2
Simplifying, sd = 5 + j 8.66
The desired closed-loop pole sd = 5 + j 8.66 does not lie on the root locus plot (Fig.
8.7.8). Hence, the phase nagle θc contributed by the phase lead compensator is determined
below, by applying root locus angle condition at sd = 5 + j 8.66
K
i.e. θc + = 180°
(s + 2)(s + 4) s = sd = 5 + j 8.66
K
∴ θc + = 180°
( 5 + j 8.66 + 2)( 5 + j 8.66 + 4)
LMFG 180° 1 8.66 IJ FG 1 8.66 IJ OP =
∴ θc
NH tan
3 K H
+ 180° tan
1 KQ 180°
K c (s + a)
Gc(s) = , a<b
(s + b)
As the order of the characteristic is to remain unchanged, the zero of the compensator
s= a is placed at the open-loop pole s = 2.
The phase lead compensator takes the form
K c (s + 2)
Gc(s) =
(s + b)
The angle θc contributed by the lead compensator at sd = 5 + j 8.66 is given by
K c (s + 2)
θc =
(s + b) s = sd = 5 + j 8.66
∵ θc = 25.7° (determined earlier)
K c (s + 2)
25.7° =
(s + b) s = sd = 5 + j 8.66
( 5 + j 8.66 + 2) 3 + j 8.66
∴ 25.7° = =
(b 5) + j 8.66 (b 5) + j 8.66
FG
= 180° tan 1
8.66
tan 1
IJ8.66
H 3 Kb 5
8.66
= (180° 70.89°) tan 1
b 5
Simplifying,
8.66
1
tan = 83.41°
b 5
8.66
∴ = tan 83.41° = 8.65
b 5
Simplifying, b=6
Thus, the transfer function of phase lead compensator is
K c (s + 2)
Gc(s) =
(s + 6)
The open-loop transfer function using phase-lead compensator is given below:
KK c (s + 2) KK c
G(s) Gc(s) = =
(s + 2)(s + 4)(s + 6) (s + 4)(s + 6)
The desired closed-loop sd = 5 + j 8.66 lies on the compensated root locus. Therefore,
the root locus magnitude condition is to be satisfied at sd = 5 + j 8.66.
∴ G(s) Gc (s) s = sd = 5 + j 8.66
=1
Therefore,
KK c
=1
(s + 4)(s + 6) s = sd = 5 + j 8.66
Example 8.7.3. The open-loop transfer function of a unity feedback control system
operating with damping ratio ξ = 0.5 is given below:
K
G(s) =
s(s + 2)(s + 5)
The settling time is 5.63 sec.
Design a suitable compensator to reduce the settling time by a factor of 0.6.
sd = ξωn + jωn 1 ξ2
2 2
K
or θc + = 180°
( 1.18 + j 2.05)( 1.18 + j 2.05 + 2)( 1.18 + j 2.05 + 5)
θc +
LM FG 180° 2.05
1 IJ
tan 1
2.05
tan 1
2.05
=
OP 180°
N H tan
K Q
∴
1.18 0.82 3.82
or θc + [ (180° tan 1 1.73) tan 1 2.5 tan 1 0.536] = 180°
COMPENSATION OF CONTROL SYSTEMS 393
∴ θc + [ (180° 59.97°) 68.19° 28.19°] = 180°
Simplifying, θc = 36.41°
2. Phase Lead Compensator
The transfer function of a phase lead compensator is given below:
K c (s + a)
Gc(s) = ,a<b
(s + b)
The phase lead compensator provides a lead angle θc at sd = 1.18 + j2.05.
K c (s + a)
∴ θc =
(s + b) s = sd = 1.18 + j 2.05
∵ θc = 36.41° (calculated earlier)
K c (s + a)
∴ 36.41° =
(s + b) s = sd = 1.18 + j 2.05
Select a = 2 such that the open-loop pole at s = 2 is cancelled and the order of the
characteristic equation remains unchanged.
K c (s + 2)
∴ 36.41° =
(s + b) s = sd = 1.18 + j 2.05
( 1.18 + j 2.05 + 2)
=
(b 1.18) + j 2.05
2.05 2.05
= tan 1 tan 1
0.82 b 1.18
2.05
∴ 36.41° = 68.19° tan 1
b 1.18
Rearranging,
12.05
tan = 31.78°
b
1.18
2.05
∴ tan 31.78° =
b 1.18
2.05
∴ 0.619 =
b 1.18
Solving above equation,
b = 4.49
Compensated gain KKc
The value of the compensated gain KKc at sd = 1.18 + j 2.05 is determined using root
locus magnitude condition to the compensated open-loop transfer function.
The compensated transfer function is
KK c (s + 2)
G(s) Gc(s) =
s(s + 2)(s + 5)(s + 4.99)
KK c
Simplifying, G(s) Gc(s) =
s(s + 5)(s + 4.99)
394 LINEAR CONTROL SYSTEMS
KK c
∴ =1
s(s + 5)(s + 4.99) s = sd = 1.18 + j 2.05
Example 8.7.4. Design a compensator for a unity feedback control system having
open-loop transfer function as
K
G(s) =
s(s + 8)
The required specifications are:
Damping ratio ξ = 0.5 and the steady state error due to unit ramp input is reduced by
a factor of 6.25.
396 LINEAR CONTROL SYSTEMS
wn
=8
Simplifying,
K = (s2 + 8s)
The break away point is obtained by relation K=0 q = 60° K=0
dK 0 Re
=0 –8
ds sb = –4
dK
∴ = (2s + 8)
ds Fig. 8.7.10. Root locus plot for Example 8.7.4.
or (2s + 8) = 0
8
∴ s= = 4
2
Hence, the break away point on the root is located at
σb = 4
Angle of asymptotes are given by
(2k + 1)
∠Asym = × 180° k = 0.1
p z
(2 × 0 + 1)
= × 180° = 90°
2 0
(2 × 1 + 1)
and = × 180° = 270° = 90°
2 0
2. Lag Compensator
σb = ξωn
∴ ξωn = 4
∵ ξ = 0.5 (given)
4
ωn = = 8 rad./sec.
0.5
The desired closed-loop pole is given by
sd = ξωn + jωn 1 ξ2
= 4+j8 1 0.5 2
Simplifying, sd = 4 + j 4 3
As per given data the steady state error is to be reduced by a factor of 6.25.
COMPENSATION OF CONTROL SYSTEMS 397
K vn
∴ = 6.25
K vo
The transfer function of a lag compensator is given below,
(s + a)
Gc(s) = , a>b
(s + b)
K vn a
Also =
K v0 b
a
and = 6.25
b
select b = 0.05, ∴ a = 0.05 × 6.25 = 0.3125
(s + 0.3125)
∴ Gc(s) =
(s + 0.05)
The angle contributed to the root locus (uncompensated) due to use of lag compensator
at s = sd = 4 + j 4 3 (desired closed-loop pole) is determined below,
s + 0.3125
∠Gc(s) =
s + 0.05 s = sd = 4+ j4 3
4 + j 4 3 + 0.3125 3.68 + j 4 3
= =
4 + j 4 3 + 0.05 3.95 + j 4 3
F 4 3 I F 180° 4 3 I
GH
= 180° tan 1
3.68 JK GH tan 1
3.95 JK
= (180° 62.02) (180° 60.25)
= 1.77°
As the angle 1.77° falls within a range 0° to 3°. Hence, the lag compensator design
is accepted.
The phase lag compensated open-loop transfer function is given below:
K (s + 0.3125)
G(s) Gc(s) =
s(s + 8)(s + 0.05)
The value of K (uncompensated gain at (s = sd = 4 + j4 3 ) for a second order
characteristic equation is K = ωn2 = 8 × 8 = 64
64(s + 0.3125)
∴ G(s) Gc(s) = . Ans.
s(s + 8)(s + 0.05)
Example 8.7.5. The open-loop transfer function of a unity feedback control system is
given below:
K
G(s) =
(s + 1)(s + 6)
The damping ratio = 0.6.
Design a suitable compensator such that the steady state error coeff is Kp = 41.37 for a
unit step input.
398 LINEAR CONTROL SYSTEMS
k®¥
locus plot for G(s) = K/(s + 1)(s + 6) is drawn in
Fig. 8.7.11.
θ = cos 1 ξ = cos 1 0.6 = 53.15° sd = –3.5 + j 4.65
Let, the damping ratio line intersect the root
locus say at sd = σ + jω.
The equation for the damping ratio line is
q
derived below.
0 Re
The break away point can be calculated as –6 sb = –3.5 –1
K=0 K=0
σ = 3.5 as per root locus plot procedure.
Fig. 8.7.11. Root locus for Example 8.7.5
ω
∴ = tan θ
σ
( ve slope) is the equation for the damping ratio line.
∵ σ = 3.5
ω
∴ = tan 53.15°
3.5
= 1.33
∴ ω = ( 3.5) ( 1.33) = 4.65
Hence, the desired closed-loop pole is
sd = 3.5 + j 4.65
The value of gain K at s = sd = 3.5 + j 4.65 is determined using magnitude condition
of root locus i.e.
G(s) s = sd = 3.5 + j 4.65
=1
K
∴ =1
(s + 1)(s + 6) s = sd = 3.5 + j 4.65
2. Compensated case
Kvn = 19.8 sec 1 (Given)
The transfer function of a lag compensator is given below
(s + a)
Gc(s) = , a>b
(s + b)
a K vn
∵ = (mentioned earlier)
b K v0
a 19.8
∴ = =6
b 3.25
Select parameter b = 0.05
∴ parameter a = 6 × 0.05 = 0.3
(s + 0.3)
and Gc(s) =
(s + 0.05)
The angle contribution by the lag compensator at the given closed-loop pole s = 2+
j2 2.5 is determined below,
(s + 0.3)
∠Gc(s) =
(s + 0.05) s = sd = 2 + j 2 2.5
2 + j 2 2.5 + 0.15
=
2 + j 2 2.5 + 0.025
1.85 + j 3.16
=
1.975 + j 3.16
FG
= 180° tan 1
3.16 IJ FG
180° tan 1
3.16 IJ
H 1.85 K H 1.975 K
= 1.66°
which is in the range of 0° to 3°. Hence, the design is accepted. The compensated open-loop
transfer function is,
16.25(s + 0.15)
G(s) Gc(s) = . Ans.
s(s + 5)(s + 0.025)
COMPENSATION OF CONTROL SYSTEMS 401
Example 8.7.7. The forward path transfer function of a unity feedback control
system is given below:
K
G(s) =
s(s + 2)(s + 8)
Design a lag compensator such that
(a) Maximum over shoot 16.4% for a unit step input.
(b) The steady state error 0.16 rad. for unit ramp input.
Solution. The general shape of the root locus plot is for the open-loop transfer
function
K
G(s) =
s(s + 2)(s + 8)
is shown in Fig. 8.7.12.
1 ω 1 ω 1 ω
Rearranging, tan + tan = 180° tan
σ+2 σ+8 σ
F ω + ω I
Simplifying, tan 1
GG σ + 2 σ + 8 JJ = 180° tan ω 1
GG 1 ω 2
JJ σ
H (σ + 2)(σ + 8) K
or tan 1
LM ω(2σ + 10) OP = 180° tan ω 1
N σ + 10σ + 16 ω Q
2
1
2
σ
∵ tan 180° = 0
ω (2σ + 10) ω
∴ 2 2
=
σ + 10σ + 16 ω σ
Rearranging the equation for the complex part of the root locus relating ω and σ in s-
plane is obtained below,
ω2 = 3σ2 + 20σ+ 16
The break away point on the root locus is obtained by substituting ω = 0.
∴ Solving
3σ2 + 20σ + 16 = 0
The acceptable solution is
σ = 0.545
Hence, σb = 0.545
Equation for the damping ratio line
The damping ratio line makes angle θ with the s-plane real axis as shown in Fig.
8.7.12, where θ = cos 1 ξ
The value of the damping ratio ξ is determined using relation below,
ξπ
1 ξ2
Mp = e
∵ %Mp = 16.4%
ξπ
16.4 1 ξ2
∴ =e
100
On solving above equation,
ξ = 0.5
∴ θ = cos 1 0.5 = 60°
and tan θ = tan 60° = 3
Thus, the damping ratio line has a slope as under,
ω
= tan θ ( ve slope)
σ
Therefore, the equation for the damping ratio line is given below,
ω= 3 (σ)
COMPENSATION OF CONTROL SYSTEMS 403
As already derived the equation for the complex part of root locus is
ω2 = 3σ2 + 20σ + 16 = 0
Substitute ω = 3 (σ) in the above equation
∴ 3σ2 = 3σ2 + 20σ + 16 = 0
∴ σ = 0.8 and ω = 3 (σ) = 3 ( 0.8) = 0.8 3
Therefore, the intersection point is at
sd = 0.8 + j 0.8 3
which is the closed-loop pole (uncompensated).
Determination of the value of K at sd = 0.8 + j0.8 3
The value of K at s = sd is determined using root locus magnitude condition i.e.
K
=1
s(s + 2)(s + 8) s = sd = 0.8 + j 0.8 3
K
∴ =1
s(s + 2)(s + 8) s = sd = 0.8 + j 0.8 3
a 6.25
∴ = = 4.66
b 1.34
404 LINEAR CONTROL SYSTEMS
Example 8.7.8. The open-loop transfer function of a unity feedback control system is
given below:
6
G(s) =
s( s + 4)
Design a lag-lead compensator such that the closed-loop desired pole is placed as
sd = 3 ± j 1.5 2
and the velocity error coefficient. Kv = 15 sec 1.
6
Solution. G(s) =
s(s + 4)
K v0 = lim s G(s)
s→0
uncompensated
s6 1
= lim = 1.5 sec
s→0 s(s + 4)
Lead Compensator
The desired closed-loop pole sd = 3 + j1.5 2 does not lie on the uncompensated root
locus. The angle θc to be contributed by the lead compensator is determined below using
angle condition of root locus at s = sd = 3 + j 1.5 2
6
θc + = 180
s(s + 4) s = sd = 3 + j 1.5 2
6
∴ θc + = 180°
( 3 + j 1.5 2 )( 3 + j 1.5 2 + 4)
6
or θc + = 180°
( 3 + j 1.5 2 )(1 + j 1.5 2 )
LM F 180° 1.5 2 I 1.5 2 OP
MN GH JK
1 1
∴ θc + tan tan = 180°
3 1 PQ
∴ θc 180° + 35.26° 64.75 = 180°
Simplifying, θc = 29.49°
The transfer function of a lead compensator is given by,
K c (s + a1 )
G (s) = , a1 < b 1
lead (s + b1 )
Select, a1 = 4 to cancel the open-loop pole s = 4
K c ( s + 4)
∴ Gc (s) =
lead (s + b1 )
K c ( 3 + j 1.5 2 + 4)
or θc =
(b1 3) + j 1.5 2
406 LINEAR CONTROL SYSTEMS
1
1.5 2
or 29.49° = 64.75 tan
(b1 3)
Simplifying,
1 1.5 2
tan = 35.26
(b1 3)
1.5 2
∴ = tan 35.26 = 0.7069
b1 3
∴ b1 = 6
K = 6 (given)
6 K c (s + 4) 6Kc
∴ G(s) Gc (s) = =
lead s(s + 4)(s + 6) s(s + 6)
The value of 6Kc is determined applying root locus magnitude condition at s = sd = ( 3
+ j 1.5 2 )
∴ G(s) Gc (s) s = sd = 3 + j 1.5 2 =1
lead
6K c
∴ =1
s(s + 6) s = sd = 3 + j 1.5 2
= | ( 3 + j 1.5 2 ) ( 3 + j1.5 2 + 6) |
= | ( 3 + j 1.5 2 ) (3 + j 1.5 2 ) |
13.5
= 13.5, Kc = = 2.25
6
13.5
∴ G(s) Gc (s) =
lead s(s + 6)
The output signal of G(s) Gc (s) is the input to lag compensator.
lead
∴ Kv0 = lim s G(s) Gc(s)
s→0
s × 13.5 1
= lim = 2.25 sec
s(s + 6)
s→0
Gc (s) =
FG s + a IJ ,
2
a2 < b 2
lead Hs+b K 2
COMPENSATION OF CONTROL SYSTEMS 407
a2 K vn
=
b2 K v0
K vn 15
∵ = = 6.66
K v0 2.25
a2
∴ = 6.66
b2
Select b2 = 0.02, therefore, a2 = 0.02 × 6.66 = 0.133
(s + 0.133)
∴ Gc (s) =
lag (s + 0.02)
The angle contributed by the lag compensator at s = sd = 3 + j 1.5 2 is calculated as
(s + 0.133)
∠Gc (s) =
lag (s + 0.02) s = sd = 3 + j 1.5 2
( 3 + j 1.5 2 + 0.133)
=
( 3 + j 1.5 2 + 0.02)
( 2.867 + j 1.5 2)
=
( 2.98 + j 1.5 2 )
F 1.5 2 I F 180° 1.5 2 I
GH
= 180° tan 1
2.867 JK GH tan 1
2.98 JK
= 1.05°
As the contributed angle is within the range 0° to 3°, the design is accepted.
6 × 2.25(s + 0.133)
G(s) Gc (s) Gc (s) = . Ans.
lead lag s(s + 6)(s + 0.02)
Example 8.7.9. The unity feedback control system has open-loop transfer function as
below:
72.5
G(s) =
s(s + 3)(s + 12)
Design a compensator such that the closed-loop dominant (desired) pole is located at
s= 2.5 + j2.5 3 and the velocity error coefficient. for a unit ramp input is Kv = 18 sec 1.
Solution. The general shape of the root locus is shown in Fig. 8.7.13. The break away
point on the s-plane real axis is determined below.
The characteristic equation is given by
1 + G(s) = 0
K
or 1+ = 0, where K = 72.5 (given)
s(s + 3)(s + 12)
∴ K = [s(s + 3)(s + 12)]
= [s3 + 15s2 + 36s]
dK
put = 0, ∴ 3s2 + 15s + 36 = 0
ds
408 LINEAR CONTROL SYSTEMS
Imj
K®¥
sd = –2.5 + j 2.5Ö3
K = 72.5
sb = –1.4
72.5
or θc + = 180°
( 2.5 + j 2.5 3)( 2.5 + j 2.5 3 + 3)( 2.5 + j 2.5 3 + 12)
72.5
∴ θc + = 180°
( 2.5 + j 2.5 3)(0.5 + j 2.5 3)(9.5 + j 2.5 3)
LM F 180° 2.5 3 I F tan 2.5 3 I F tan 2.5 3 I OP =
MN GH JK GH JK GH JK PQ
1 1 1
or θc + tan 180°
2.5 0.5 9.5
∴ θc + [ (180° tan 1 3 ) (tan 1 5 3 ) (tan 1 0.263 3 )] = 180°
∴ θc + [ (180° 60°) 83.41° 24.46°] = 180°
Simplifying, θc = 47.87°
Lead Compensator
The transfer function of a lead compensator is given by,
K c (s + a1 )
Gc (s) = , a 1 < b1
lead (s + b1 )
Select, a1 = 3, to cancel open loop pole at s = 3
K c (s + 3)
∴ Gc (s) =
lead (s + b1 )
K c (s + 3)
∵ θc =
(s + b1 ) s = sd = 2.5 + j 2.5 3
COMPENSATION OF CONTROL SYSTEMS 409
K c ( 2.5 + j 2.5 3 + 3)
=
(b1 2.5) + ( j 2.5 3 )
∵ θc = 47.87 (calculated earlier)
K c (0.5 + j 2.5 3)
∴ 47.87° =
(b1 2.5) + j 2.5 3
1
2.5 3
= 83.41° tan
b1 2.5
1 2.5 3
∴ tan = (83.41° 47.87°) = 35.54°
b1 2.5
2.5 3
∴ = tan 35.54°
b1 2.5
2.5 3
or = 0.714
b1 2.5
Simplifying, b1 = 8.56
K c (s + 3) 72.5 K c (s + 3)
Thus, Gc (s) = and G(s) Gc (s) =
lead (s + 8.56) lead s(s + 3)(s + 2)(s + 8.56)
72.5 K c
or G(s) Gc (s) =
lead s(s + 12)(s + 8.56)
Since, sd = 2.5 + j 2.5 2 lies of lead compensated root locus, the magnitude condition
of the root locus is applicable,
∴ G (s)Gc (s) s = sd = 2.5 + j 2.5 3
=1
72.5 K c
∴ =1
s(s + 12)(s + 8.56) s = sd = 2.5 + j 2.5 3
72.5 × 5.36
= lim s
s→0 s(s + 12)(s + 8.56)
∴ Kv0 = 3.78
Lag Compensator
The transfer function of a lag compensator is given below:
(s + a2 )
Gc (s) = , a 2 > b2
lag (s + b2 )
a2 K vn 18
= = = 4.76
b2 K v0 3.78
Select b2 = 0.03, ∴ a2 = 0.03 × 4.76 = 0.1428
(s + 0.1428)
∴ Gc (s) =
lag (s + 0.03)
The angle contributed by the lag compensator at s = sd = 2.5 + j 2.5 3 is determined
below
(s + 0.1428)
∠Gc(s) =
(s + 0.03) s = sd = 2.5 + j 2.5 3
= ( 2.357 + j 4.33)
( 2.47 + j 4.33)
FG
= 180° tan 1
4.33 IJ FG
4.33 IJ
H 2.357
180°
K H
2.47 K
= (180° 61.43°) (180° 60.29°)
= 1.14°
As the angle contributed by the lag compensator is 1.14°, lies in the range (0° to 3°),
the design is accepted
72.5 × 5.36(s + 0.1428)
∴ G(s) Gc (s) Gc (s) = . Ans.
lead lag s(s + 12)(s + 8.56)(s + 0.03)
Example 8.7.10. Design a lag-lead compensator for a control system, the open-loop
transfer function (unity feedback) is given as,
3.57
G(s) =
s(s + 1)(s + 5 )
The damping ratio ξ = 0.5, the settling time ts = 5 sec. and the velocity error coeff. Kv =
5 sec 1 for a unit ramp input.
Solution. The general shape of the root locus for open-loop transfer function
K
G(s) =
s(s + 1)(s + 5)
COMPENSATION OF CONTROL SYSTEMS 411
Imj
K®¥
sd = –0.8 + j 0.8Ö3
K = 3.57
sb = 0.47
Fig. 8.7.14. Root locus for Example 8.7.10.
is shown in Fig. 8.7.14. The break away point can be determined as σb = 0.478. The
damping ratio line makes an angle θ with the ve real axis of s-plane.
where θ = cos 1 ξ = cos 1 0.5 = 60°
The damping ratio line intersects the root locus such that K = 3.57 (given)
The desired settling time ts = 5 sec.
4 4 4
∵ ts = , ∴ 5= , thus ξωn = = 0.8
ξω n ξω n 5
0.8
Further, ξωn = 0.8 and ξ = 0.5, ∴ ωn = = 1.6 rad./sec.
0.5
The desired closed-loop pole sd is given by the following relation,
sd = ξωn + jωn 1 ξ2
3.57
Therefore, θc + = 180°
s(s + 1)(s + 5) s = sd = 0.8 + j 0.8 3
3.57
Thus, θc + = 180°
( 0.8 + j 0.8 3)( 0.8 + j 0.8 3 + 1)( 0.8 + j 0.8 3 + 5)
LM F 180° 0.8 3 I F tan 0.8 3 I F tan 0.8 3 I OP =
MN GH JK GH JK GH JK PQ
1 1
∴ θc + 180°
0.8 0.2 4.2
or θc + [ (180° 60°) (81.78°) (18.21)°] = 180°
412 LINEAR CONTROL SYSTEMS
Simplifying, θc 40°
Lead Compensator
The transfer function of a lead compensator is given by,
K c (s + a1 )
Gc (s) = , a 1 < b1
lead (s + b1 )
Select the value of a1 such that it is placed on the s-plane negative real axis just below
the real part of desired closed-loop pole sd. Therefore, a1 = 0.8 and the lead compensator
transfer function takes the form as
K c (s + 0.8)
Gc (s) = , 0.8 < b1
lead (s + b1 )
The angle contributed by the lead compensator at the desired closed-loop pole sd =
0.8 + j 0.8 3 is determined using following relation,
K c (s + 0.8)
θc =
(s + b1 ) s = sd = 0.8 + j 0.8 3
1
0.8 3
∴ 40° = 90° tan
b1 0.8
1
0.8 3
Rearranging, 50° = tan
b1 0.8
0.8 3
∴ tan 50° =
b1 0.8
0.8 3
∴ 1.191 =
b1 0.8
Solving above equation,
b1 = 1.96
The lead compensated open-loop transfer function is written below,
3.57 K c (s + 0.8)
G(s) Gc (s) =
lead (s + 1)(s + 5)(s + 1.96)
The desired closed-loop pole sd = 0.8 + j 0.8 3 , lies on the lead compensator root
locus. Hence, the value of gain (gain factor) Kc is determined applying root locus magnitude
condition to above transfer function at s = sd = 0.8 + j 0.8 3 .
3.57 K c (s + 0.8)
∴ =1
s(s + 1)(s + 5)(s + 1.96) s = sd = 0.8 + j 0.8 3
COMPENSATION OF CONTROL SYSTEMS 413
Rearranging,
s(s + 1)(s + 5)(s + 1.96)
3.57 Kc =
(s + 0.8) s = sd = 0.8 + j 0.8 3
(− 0.8 + j 0.8 3 )(− 0.8 + j 0.8 3 + 1)(− 0.8 + j 0.8 3 + 5)(− 0.8 + j 0.8 3 + 1.96)
=
(− 0.8 + j 0.8 3 + 0.8)
Lag Compensator
The transfer function of a lag compensator is given by,
(s + a2 )
Gc (s) = , a2 > b2
lag (s + b2 )
K vn a
also = 2
K v0 b2
∵ Kvn = 5 sec 1 (given), Kv0 = 1.047 sec 1 (calculated earlier)
a2 5
∴ = = 4.77
b2 1.047
Select, b2 = 0.025, ∴ a2 = 0.025 × 4.77 = 0.119
(s + 0.119)
Thus, Gc (s) =
lag (s + 0.025)
The angle θ contributed by the lag compensator at sd = 0.8 + j 0.8 3 is determined
below,
(s + 0.119)
θ=
(s + 0.025) s = sd = 0.8 + 0.8 3
PROBLEMS
K
8.1. For the open-loop transfer function, G(s) H(s) = determine the value of
s(s + 0.5s)
gain K such that the system will have a phase margin of at least 45º.
8.2. Design a lead compensator such that the system having following open-loop transfer
function will have a phase margin of about 40º.
5
G(s) H(s) = .
(1 + s) (1 + 0.5s) 2
8.3. A unity feedback control system has following open-loop transfer function, G(s) =
1
. Design a suitable compensator such that (i) Kp = 3 (ii) G.M. ≥ 15 db
(1 + 0.5 s) 3
(iii) P.M. ≥ 30º.
8.4. Following specifications :
(i) Kv ≥ 4 (ii) G.M. ≥ 6 db (iii) P.M. ≥ 20º refer to a system whose open-loop transfer
K
function is G(s) H(s) =
s(1 + 0.2 s) 2
Design a phase-lead compensation network to satisfy the above specifications.
8.5. The system shown in Fig. 8P-1 is to be compensated by tachometer feedback such
that the maximum overshoot is limited to 30%. Determine the value of tachometer
constant Kt.
R(s) + 3 C(s)
s2
–
R(s) + 9 C(s)
s(s + 1)
–
(a)
R(s) + 9 C(s)
s(s + 1)
–
sKt + 1
(b)
Fig. 8P-2. System for problem 8.6.
8.7. The open-loop transfer function of a unity feedback control system is given below,
K
G(s) =
s(s + 10)
The max. overshoot is 24.3% (unit step input), maintaining the same damping ratio,
design a compensator such that the settling time is reduced by a factor of 2.
8.8. A unity feedback control system having open-loop transfer function as
K
G(s) =
s(s + 6)
operates at a daming ratio ξ = 0.6.
It is desired to have a settling time ts = 1.0 sec. Design a suitable compensator.
8.9. Design a lead compensator such that a unity feedback control system having
following transfer function,
K
G(s) = 2
s +1
becomes stable with damping ratio ξ = 0.5 and settling time ts = 2.5 sec. The
uncompensated system operates at closed-loop poles s = ± j 5 .
8.10. The open-loop transfer function (unity feedback) is given below
K
G(s) =
s(s + 4)
The damping ratio is ξ = 0.4. It is specified that the velocity error coefficient for unit
ramp input is Kv = 25 sec 1. Design a lag compensator.
8.11. The forward path transfer function of a unity feedback control system is given by,
K
G(s) =
s(s + 4)(s + 8)
Design a lag compensator such that the maximum overshoot is ≤ 16.3° and steady
state velocity error coefficient to a ramp input is Kv = 12 sec 1.
416 LINEAR CONTROL SYSTEMS
CONTENTS
• State Space Representation
• The Concept of State
• State Space Representation of Systems
• Block Diagram for State Equation
• Transfer Function Decomposition
• Solution of State Equation
• Transfer Matrix
• Controllability
• Observability
• Solved Examples
• State Variable Feedback
Introduction. The analysis of control systems carried out in earlier chapters is based on
transfer functions and graphical approach such as root locus, Bode and Nyquist plot. The
input output relations are in the form of transfer function. In this approach of analysis initial
conditions are considered zero meaning that the system is initially at rest and the time
solution obtained is in a general form due to input only. However, for multiple-input
multiple-output and systems initially not at rest, the transfer function approach for analysis
is inadequate and not convenient.
417
418 LINEAR CONTROL SYSTEMS
The use of state space approach for the analysis of control systems enables to
overcome the shortcomings of transfer function approach. The state space approach to be
described hereunder can be conveniently used for analysis of control system with basic
knowledge of matrix algebra.
The analysis of a control system using state space approach carried out in time
domain by representing a system in the form of first order differential equations by selecting
suitable state variables wherein first order derivative terms are arranged on left hand side
and right hand side terms are free from derivatives. The variables in such a form of rep-
resentation are known as state variables. In state space approach multiple input- multiple-
output system equations can be arranged in matrix form which facilitates their solution.
d2x(t) . dx(t) .
= x2 = x2 = x2
u(t) 1 + + dt2 dt x(t) = x1
a
– –
b
a
c
a
Fig. 9.1.1. Block diagram representation of differential Eq. (9.2).
Therefore, the dynamic equations describing the system equations can be written as
x1 = x2 ...(9.3)
c b 1
and x 2 =
x x + u ( t) ...(9.4)
a 1 a 2 a
In the matrix form Eqs. (9.3) and (9.4) can be written as
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 419
⎡ 0 1 ⎤ ⎡0 ⎤
⎡ x1 ⎤ ⎢ ⎥ ⎡ x1 ⎤ ⎢ ⎥
⎢ x ⎥ ⎢ c
= b ⎥ ⎢ ⎥ ⎢ 1 ⎥ u(t)
+ ...(9.5)
⎣ 2⎦ ⎢ ⎣ x2 ⎦
⎣ a a ⎦⎥ ⎣⎢ a ⎦⎥
The representation given by Eq. (9.5) is known as state space representation of the
system given by Eq. (9.1).
0)
At any instant of time t > t0 the state of the
x(t
system is given by the values of x1(t) and x2(t), x (t ) t = t1
2 1
therefore, x1(t) and x2(t) are called state variables. x(t 1)
Fig. 9.2.1 shows a plot of x1(t) and x2(t) on co-
ordinate axes. The starting point of the plot is at
x1(t0) x1(t1) x(t)
instant t = t0 and as time increases the plot traces a
path. Fig. 9.2.1. Path traced by state vector.
At t = t0 the state of system is given by state
vector x(t0) and t = t1 by state vector x(t1). At any instant of time t > t0 the state vector is
related to state variables as
→ → →
x(t) = x1 (t) + x2 (t)
Thus the state of a system at any instant of time t > t0 is determined by a state vector
x(t). The state vector x(t) is the sum of state variables. The path traced by a state vector with
the increase of time is known as state trajectory.
The analysis given above can be generalized for nth order system where the state of
nth order system in matrix form can be written as
⎡ x1 ⎤ ⎡ u1 ⎤
⎢x ⎥ ⎢u ⎥
⎢ 2⎥ ⎢ 2⎥
⎢ x3 ⎥ ⎢ u3 ⎥
x = ⎢ ⎥ n × 1 ; and the input vector in matrix form can be written as u = ⎢ ⎥ m×1
⎢ : ⎥ ⎢ : ⎥
⎢ : ⎥ ⎢ : ⎥
⎢ ⎥ ⎢ ⎥
⎣⎢ xn ⎦⎥ ⎣⎢um ⎦⎥
The state vector thus can be represented in n-dimensional space known as state space.
For a second order system the state space reduces to a state plane.
As a consequence of the analysis of the above illustrative example the definitions
concerning state space approach are given below :
(1) State Variables. The smallest set of variables which determine the state of a
dynamic system are called state variables.
It is not necessary that the state variables be physically measurable or observable
quantities.
420 LINEAR CONTROL SYSTEMS
(2) State. The state of a system is the smallest set of state variables such that the
knowledge of these variables at t = t0 (initial condition) together with the inputs completely
determines the behaviour of the system for any time t > t0.
(3) State Vector. If n state variables are necessary to determine the behaviour of a
given system, the variables can be considered as n components of a vector called state vector .
(4) State Space. The n dimensional state variables are elements of n dimensional
space called state space .
The terms y(t), y (t) ... yn 1(t) can be considered as a set of n state variables.
Let y = x1
y = x2 ...(9.6a)
:
y n 1 = xn
The set of Eqs. (9.6a) can be written as
x1 = x2
x 2 = x3
:
x n 1 = xn (9.7)
x n = anx1 an 1 x2 ... a1 x n 1 +u
A general representation of set of Eqn. (9.7) can be expressed in the form of a state
equation as
x = Ax + Bu
⎡ x1 ⎤ ⎡ 0 1 .. 0 0⎤ ⎡0 ⎤
⎢x ⎥ ⎢ 0 0 .. 1 0 ⎥⎥ ⎢0 ⎥
⎢ 2⎥ ⎢ ⎢ ⎥
⎢ : ⎥ ⎢ 0 0⎥ ⎢0 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
where x=⎢ : ⎥;A=⎢ 0 0 .. 0 1 ⎥ ; B = ⎢:⎥ ...(9.8)
⎢ : ⎥ ⎢ : : .. : : ⎥ ⎢:⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ : ⎥ ⎢ : : .. : : ⎥ ⎢:⎥
⎢⎢ x ⎥⎥ ⎢⎢ a an .. 0 a1 ⎥⎦⎥ ⎢⎢1 ⎥⎥
⎣ n⎦ ⎣ n 1 ⎣ ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 421
where x = state vector (n × 1) u = control (scalar)
A = matrix (n × n) B = matrix (n × 1)
The output equation is given by
⎡ x1 ⎤
⎢x ⎥
⎢ 2⎥
y = [1 0 0 .. 0] ⎢ : ⎥ ...(9.9)
⎢ ⎥
⎢ : ⎥
⎢⎣ xn ⎥⎦
or y = Cx ....(9.10)
where C = [1 0 0 ... 0], y = output (scalar), C = matrix (1 × n)
The output equation is algebraic and state equation is a set of first order differential
equations.
(2) State space representation for v1 i v2
electrical network. An electrical network is
shown in Fig. 9.3.1. If the initial values of the R1 L
current i(0), the capacitor voltages v1(0) and v2(0)
are known, then the dynamic behaviour of e C1 C2 R2
electrical network can be completely defined for
t > 0.
Thus the state of the network for t > 0 is
completely determined by selecting v1(t), v2(t) and Fig. 9.3.1. Electrical network.
i(t) as state variables.
The equations describing dynamic behaviour of the system are
v1 e dv
+ C1 1 + i = 0 ...(9.11)
R1 dt
dv v
C1 2 + 2 i = 0 ...(9.12)
dt R2
di
L + v2 v1 = 0 ...(9.13)
dt
Rearranging and simplifying, following equations are obtained :
dv1 v1 i e
= + ...(9.14)
dt R1C1 C1 R1C1
dv2 v2 i
= + ...(9.15)
dt R2C2 C2
di v1 v2
= ...(9.16)
dt L L
As v1, v2 and i are selected as state variables, therefore,
x1 = v 1
x2 = v 2
and x3 = i
In view of above set of equations state space representation of the system is obtained
below :
422 LINEAR CONTROL SYSTEMS
1 1 1
x1 = x x3 + e
R1C1 1 C1 R1C1
1 1
x 2 = x + x
R2C2 2 C2 3
1 1
x3 = x x
L 1 L 2
In vector-matrix form state space representation is given below :
⎡ 1 1 ⎤
⎢ 0
R1C1 C1 ⎥ ⎡ 1 ⎤
⎡ x1 ⎤ ⎢ ⎥ ⎡ x1 ⎤ ⎢RC ⎥
⎢ ⎥ ⎢ 1 1 ⎥ ⎢ ⎥ ⎢ 1 1⎥ e ...(9.17)
⎢ x2 ⎥ = ⎢ 0
R2C2 C2 ⎥⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥
⎢⎣ x3 ⎥⎦ ⎢ ⎢⎣ x3 ⎥⎦ ⎢ 0 ⎥
⎢ 1 1 ⎥ ⎣⎢ ⎦⎥
⎢ 0 ⎥
⎣⎢ L L ⎦⎥
(3) State space representation for transfer function. The transfer function of a
control system can be represented in the form of a state block diagram wherein the variable
appearing at the output of each integrator is considered as state variable and state space
representation obtained therefrom.
Consider the following transfer function
C( s) 1
= 2 ...(9.18)
R(s) s + a1 s + a2
The corresponding differential equation for the above transfer function is
d2 c dc
2
+ a1 + a2 c = r
dt dt
To draw the state block diagram the equation above is rearranged below :
d2 c dc
2
= a2 c a1 +r
dt dt
The state block diagram is shown . . .
r + + x2 x2 = x1 c = x1
in Fig. 9.3.2.
– –
The output of last integrator gives d 2c dc
the output c. The output of each dt 2 dt
integrator is assigned a state variable in a1
ascending order from right to left as
a2
shown in Fig. 9.3.2.
As per the block diagram, the
Fig. 9.3.2. State block diagram for
state equations for the system are : d2c/dt2 + a1(dc/dt) + a2c = r.
x1 = x2
x 2 = a2x1 a1x2 + r
The output is denoted by y. Therefore,
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 423
y = x1
As per above equations, the state model in vector-matrix form is given below :
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ a r ...(1)
a1 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦
+
⎣ 2⎦ ⎣ 2
⎡x ⎤
y = [1 0] ⎢ 1 ⎥
⎣ x2 ⎦
On reversing the order of state variables i.e. assigning the integrator output by a state
variable in descending order from right to left. Denoting x2 in place of x1 and x1 for x2, the
state equations are
x1 = a1 x1 a2 x2 + r
x2 = x1
and y = x2
The state model in vector-matrix form is given below :
⎡ x1 ⎤ ⎡ a1 a2 ⎤ ⎡ x1 ⎤ ⎡1 ⎤
⎢ ⎥ = ⎢ r ...(2)
0 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣0 ⎥⎦
+
⎣ x2 ⎦ ⎣ 1
⎡x ⎤
y = [0 1] ⎢ 1 ⎥
⎣ x2 ⎦
The state model (2) can be obtained from (1) by applying similarity transformation
x = Px
⎡0 1 ⎤ 1 ⎡0 1 ⎤
where P = ⎢ ⎥;P ⎢1 0 ⎥
⎣1 0 ⎦ ⎣ ⎦
x = Ax + B u
substitute x = Px
∴ P x = A P x + B u
on pre multiplying by P 1
x = P 1 A Px + P 1B u
⎡ 0 1 ⎤ ⎡0 ⎤
A= ⎢ and B= ⎢ ⎥
⎣ − a2 − a1 ⎥⎦ ⎣1 ⎦
⎡ x1 ⎤ ⎡0 1 ⎤ ⎡ 0 1 ⎤ ⎡0 1 ⎤ ⎡ x1 ⎤ ⎡0 1 ⎤ ⎡0 ⎤
∴ ⎢ ⎥ = ⎢ ⎥⎢ ⎢ 1 0 ⎥ ⎢ x ⎥ + ⎢ 1 0 ⎥ ⎢1 ⎥ r
x
⎣ 2⎦ ⎣1 0 ⎦ ⎣ − a2 − a1 ⎥⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦⎣ ⎦
⎡ x1 ⎤ ⎡ − a1 − a2 ⎤ ⎡ x1 ⎤ ⎡1 ⎤
or ⎢ ⎥ = ⎢ ⎢ x ⎥ + ⎢0 ⎥ r
x
⎣ 2⎦ ⎣ 1 0 ⎥⎦ ⎣ 2⎦ ⎣ ⎦
⎡x ⎤
y = [1 0] P ⎢ 1 ⎥
⎣ x2 ⎦
424 LINEAR CONTROL SYSTEMS
⎡0 1 ⎤ ⎡ x1 ⎤
= [1 0] ⎢ ⎥⎢ ⎥
⎣1 0 ⎦ ⎣ x2 ⎦
⎡ x1 ⎤
∴ y = [0 1] ⎢ ⎥
⎣ x2 ⎦
d 2c + a1dc dc + a
r + dt 2 dt dt
1
+ c y
. x2 .
x2 + x1 x1
+
–a1
–a2
On obtaining a state model, the system matrix in four different forms can be found to be :
⎡ 0 1 0 0 ⎤
⎢ 0 0 1 0 ⎥⎥
(1) A= ⎢
⎢ 0 0 0 1 ⎥
⎢ ⎥
⎣⎢ − a4 − a3 − a2 − a1 ⎦⎥
The system matrix A given above contains upper elements of the diagonal as unity
and the last row consists of negative coefficients of the characteristic polynomial. All other
elements are zero.
⎡ − a1 − a2 − a3 − a4 ⎤
⎢ 1 0 0 0 ⎥⎥
(2) A= ⎢
⎢ 0 1 0 0 ⎥
⎢ ⎥
⎢⎣ 0 0 1 0 ⎥⎦
The system matrix A given above contains lower elements of the diagonal as unity and
the first row consists of negative coefficients of the characteristic polynomial. All other
elements are zero.
For the case (1) and (2), the system matrix A is known to be in Controllable Canonical
form. Also known as companion form 1.
⎡ − a1 1 0 0⎤
⎢− a 0 1 0 ⎥⎥
(3) A= ⎢ 2
⎢ − a3 0 0 1⎥
⎢ ⎥
⎣⎢ − a4 0 0 0 ⎦⎥
The system matrix A given above contains upper elements of the diagonal as unity
and the first column contains negative coefficients of the characteristic polynomial. All other
elements are zero.
⎡0 0 0 − a4 ⎤
⎢1 0 0 − a3 ⎥⎥
(4) A= ⎢
⎢0 1 0 − a2 ⎥
⎢ ⎥
⎢⎣0 0 1 − a1 ⎥⎦
The system matrix A given above contains lower elements of the diagonal as unity and
the last column consists of negative coefficients of the characteristic polynomial. All other
elements are zero.
For the case (3) and (4), the system matrix A is known to be in Observable Canonical
form. Also known as companion form 2.
426 LINEAR CONTROL SYSTEMS
On observation of state models mentioned above, it is concluded that these models are
not same, though have been derived from the same transfer function. Thus, it is concluded
that for a system the state model is not unique.
Application of Signal Flow graph for the determination of state model from a
given transfer function.
The overall transfer function C(s)/R(s) can be obtained using signal flow graph
technique. In this respect the Mason s gain formula is given below :
k
C ( s) ∑ k = 1 Pk ∆ k
= ...(9.20)
R( s) ∆
The term refers to output C(s) and R(s) input. The notations Pk (forward path gain), ∆k
(path factor) and ∆ (graph determinant) have already been explained in chapter 3.
The signal flow graph in relation to state variable analysis is also termed as state
diagram. The signal flow graph for a given transfer function can be drawn in several ways.
However, for convenience the case of all loops touching each other as well as to forward path
is considered. This reduces the Mason s gain formula as below :
C(s) Sum of forward path of gains
=
R(s) 1 − Sum of individual loop gains
For the transfer function
C(s) 1
= 2
R(s) s + a1 s + a2
The signal flow graph is developed below :
On multiplying the numerator and denominator by s 2, above transfer function is
rewritten as
C(s) s −2
=
R(s) 1 − (− a1 s −1 − a2 s −2 )
The transfer function takes the form of Mason s gain formula wherein,
P = s 2 (forward path gain)
L1 = a1s 1 (loop gain).
L2 = a2s 2 (loop gain).
With the information as above, the signal flow graph is drawn in Fig. 9.3.4.
In the state signal flow graph, the output of . .
x2 s–1 x2 1 x1 s–1 x1
each integrator (s 1) is denoted by a state variable in r c
ascending order from right to left. The output is
denoted as y and input r. –a1
The state equations are
–a2
x1 = x2
Fig. 9.3.4. State signal flow graph
x2 = a 2 x1 a 1 x2 + r
(controllable canonical form).
and the output equation
y = x1
In vector-matrix form the state model is
⎡ x1 ⎤ ⎡ 0 1 ⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ − a r
− a1 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦
+
⎣ 2⎦ ⎣ 2
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 427
⎡ x1 ⎤
y = [1 0] ⎢ ⎥
⎣ x2 ⎦
On reversing the order of state variables i.e. replacing x1 and x2 and x2 by x1 , the
state model takes the form given below :
⎡ x1 ⎤ ⎡ − a1 − a2 ⎤ ⎡ x1 ⎤ ⎡1 ⎤
⎢ ⎥ = ⎢ 1 0 ⎥⎦ ⎢ x ⎥ + ⎢0 ⎥ r
⎣ x2 ⎦ ⎣ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
y = [0 1] ⎢ ⎥
⎣ x2 ⎦
For the transfer function
C(s) s −2
=
R(s) 1 − (− a1 s −1 − a2 −2 )
the state signal flow graph is drawn in a different way as shown in Fig. 9.3.5.
. .
The output of each integrator is assigned a r x2 s–1 x2 1 x1 s–1 x1 y
state variable as shown in Fig. 9.3.5. The output
and input are y and r respectively. The state
equation are –a1
x1 = a 1 x1 + x2
–a2
x2 = a 2 x2 + r
and the output equation Fig. 9.3.5. State signal flow graph
(observable canonical form).
y = x1
In vector-matrix form the state model is given below :
⎡ x1 ⎤ ⎡ − a1 1 ⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ − a 0 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦
+ r
⎣ 2⎦ ⎣ 2
⎡x ⎤
y = [1 0] ⎢ 1 ⎥
⎣ x2 ⎦
On reversing the order of state variables in the state signal flow graph i.e. replacing x1
by x2 and x2 by x1 , the state model takes the form :
⎡ x1 ⎤ ⎡0 − a2 ⎤ ⎡ x1 ⎤ ⎡1 ⎤
⎢ ⎥ = ⎢ ⎥⎢ ⎥+⎢ ⎥r
⎣ x2 ⎦ ⎣1 − a1 ⎦ ⎣ x2 ⎦ ⎣0 ⎦
⎡x ⎤
y = [0 1] ⎢ 1 ⎥
⎣ x2 ⎦
D
Du
Bu + x x Cx + +
u
B C y
+
Ax
A
Fig. 9.4.1. Block diagram for state Eqs. (9.21) and (9.22) representing
multiple-input multiple-output system.
In most of single-input-single-output (SISO) systems, the output consists of linear
combination of state variables, as such D is a null matrix. Therefore, the state model of a
SISO system is given by
x = Ax + Bu ...(9.23)
y = Cx ...(9.24)
where x = state vector (n × 1), u = control (scalar)
y = output (scalar), A = matrix (n × n)
B = matrix (n × 1), C = matrix (1 × n)
The block diagram for the SISO system based on Eqs. (9.23) and (9.24) is shown in
Fig. 9.4.2.
Bu +
u X X y
B C
+
Ax
A
d2 y dy
2
+2 + 5y = u ...(9.26)
dt dt
To obtain state model for the system governed by the above differential equation, the
block diagram indicating integration process is shown in Fig. 9.5.1.
The output of each integrator is dy
d2y = x1 = x2
assigned a state variable as shown in 2 = x2 dt
dt x1 = y
Fig. 9.5.1 and the state equations are u + +
written below : + +
x1 = x2
–2
x2 = 5x1 2x2 + u
and the output is y = x2 –5
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ 5 ⎢ x ⎥ + ⎢1 ⎥ u ...(9.27)
⎣ 2⎦ ⎣ 2 ⎥⎦ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
y = [1 0 ] ⎢ ⎥
⎣ x2 ⎦
The transfer function decomposition based on above example is carried out using
following methods.
x3 x2 = x3 x1 = x2
u + + + x1 = y
+ + +
–9
–26
–24
x1 = x2
x2 = x3
x3 = 24x1 26x2 9x3 + u
and the output is y = x1
⎡ x1 ⎤ ⎡ 0 1 0⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ 0 0 1⎥⎥ ⎢ x ⎥ + ⎢0 ⎥ u
or ⎢ 2⎥ ⎢ ⎢ 2⎥ ⎢ ⎥ ...(9.29)
⎢⎣ x3 ⎥⎦ ⎢⎣ 24 26 9 ⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣1 ⎥⎦
⎡ x1 ⎤
y = [1 0 0] ⎢ x2 ⎥
⎢ ⎥
⎢⎣ x3 ⎥⎦
The transfer function of a system involving a differential term in its input is given
below :
Y (s) 2s + 1
= ...(9.30)
U (s) s2 + 2s + 3
The transfer function (9.30) is split into two parts as follows :
Y (s) 1 2s
= 2 + 2 ...(9.31)
U (s) s + 2s + 3 s + 2s + 3
The block diagram for the term
Y (s) 1
= ...(9.31a)
U (s) s2 + 2s + 3
is drawn as shown in Fig. 9.5.3.
The second part of the transfer function (9.31) i.e. 2s/(s2 + 2s + 3) is the twice multiple
of the first differential of the first part, which is available as the input to the last integrator
as shown in Fig. 9.5.3. Therefore, the block diagram in relation to transfer function (9.30) is
drawn as shown in Fig. 9.5.4.
x2 x1 = x2
u + + x1 = y
+ +
–2
–3
x1 = x2 2
x2 2x2
u + + x1 + x1 + 2x2 = y
+
+ +
–2
–3
x3 x3 x2 x2 x1 x1
u + + +
+ + +
–2 –3 –4
⎡ x1 ⎤ ⎡ 4 1 0⎤ ⎡ x1 ⎤ ⎡0 ⎤
The state model : ⎢⎢ x 2 ⎥⎥ = ⎢⎢ 0 3 1⎥⎥ ⎢ ⎥ ⎢ ⎥
⎢ x2 ⎥ + ⎢0 ⎥ u ...(9.34)
⎢⎣ x3 ⎥⎦ ⎢⎣ 0 0 2⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣1 ⎥⎦
⎡ x1 ⎤
y = [1 0 0] ⎢⎢ x2 ⎥⎥
⎢⎣ x3 ⎥⎦
The coefficients of matrix A of a state equation, obtained as per cascade decomposition
of the transfer function give a triangular matrix known as pole-zero form.
432 LINEAR CONTROL SYSTEMS
1 1
x1 x1 x – x2 + x3 = y
2 1 2
u + 1 + +
2
+ + +
–2
x2
+ x2
–1
+
–3
x3 x3
+ 1
2
+
–4
⎡ x1 ⎤
⎡1 1⎤ ⎢ ⎥
y= ⎢ 1 x2
⎣2 2 ⎥⎦ ⎢ ⎥
⎢⎣ x3 ⎥⎦
The coefficient matrix A obtained as per parallel decomposition of a transfer function
is a diagonal one and called simply canonical or diagonal or partial fraction expansion form.
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 433
9.5.4 Parallel decomposition of a transfer function having repeated roots of its characteristic
equation
The three methods given earlier for obtaining the state model of a transfer function,
considered distinct roots of its characteristic equation. For the case of repeated roots the
procedure for obtaining the state model of a given transfer function is given below :
Let, the transfer function be
Y (s) 2 s2 + 8 s + 7
= ...(9.38)
U (s) (s + 2)2 (s + 1)
On spliting R.H.S. of the transfer function (9.38) into partial fractions, the following
form of the transfer function is obtained.
Y (s) 1 1 1
= + + ...(9.39)
U (s) (s + 1) ( s + 2) (s + 2)2
The block diagram based on integration process similar to that of parallel
decomposition case is drawn in Fig. 9.5.7.
x1
u + x1 + +
x1 + x2 + x3 = y
+ + +
–1
x2 x3
+ x2 + x3
+
+ +
–2 –2
⎡ x1 ⎤
y = [1 1 1] ⎢ x2 ⎥
⎢ ⎥
⎣⎢ x3 ⎦⎥
The matrix A of the state equation (9.40) is called the Jordan s canonical form.
It is noted that the transfer functions given by (9.28), (9.33) and (9.36) are same but
written in different form. The corresponding state models are given by state equations
(9.29), (9.34) and (9.37) respectively.
434 LINEAR CONTROL SYSTEMS
–9
–26
–24
Fig. 9.5.8. State signal flow graph : Y(s)/U(s) = s 3/1 ( 9s 1 26s 2 24s 3).
x1 = x2
x2 = x3
x3 = 24x1 26x2 9x3 + u
The output equation is y = x1
The equations given above are arranged in vector-matrix form to obtain the state
model as below :
⎡ x1 ⎤ ⎡ 0 1 0⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ 0 0 1⎥⎥ ⎢ x ⎥ + ⎢0 ⎥ u
⎢ 2⎥ ⎢ ⎢ 2⎥ ⎢ ⎥
⎣⎢ x3 ⎦⎥ ⎢⎣ 24 26 9 ⎦⎥ ⎣⎢ x3 ⎦⎥ ⎢⎣1 ⎦⎥
⎡ x1 ⎤
y = [1 0 0] ⎢ x2 ⎥
⎢ ⎥
⎢⎣ x3 ⎥⎦
Y (s) 2s + 1
(b) =
U (s) s2 + 2s + 3
On multiplying the numerator and denominator by s 2 the transfer function given
above is rewritten below :
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 435
Y (s) 2s 1 + s 2 2s 1
+s 2
= =
U (s) 1 + 2s 1 + 3s 2
1 ( 2s 1
3s 2 )
The state signal flow graph for the above transfer function is drawn as shown in
Fig. 9.5.9.
The output of each integrator (s 1) is assigned a state variable as shown in Fig. 9.5.9.
The state equations are obtained below :
x1 = x2
x2 = x1 2x2 + u
. .
1 x2 s–1 x2 1 x1 s–1 x1 1
u y
–2
–3
Fig. 9.5.9. State signal flow graph : Y(s)/U(s) = (2s 1 + s 2)/[1 ( 2s 1 3s 2)].
–2 –3 –4
Assigning a state variable to each integrator (s 1) in Fig. 9.5.10, the state equations
are
x1 = 4x1 + x2
x2 = 3x2 + x3
x3 =
2x3 +u
The output equation is y = x1
The equations given above are arranged in vector-matrix form to obtain the state
model as below :
⎡ x1 ⎤ ⎡ 4 1 0⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ 0 3 1⎥⎥ ⎢ x ⎥ + ⎢0 ⎥ u
⎢ 2⎥ ⎢ ⎢ 2⎥ ⎢ ⎥
⎢⎣ x 3 ⎥⎦ ⎢⎣ 0 0 2⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣1 ⎥⎦
⎡ x1 ⎤
⎢ ⎥
y = [1 0 0] ⎢ x2 ⎥
⎢⎣ x3 ⎥⎦
(3) Parallel Decomposition
Y (s) 1
The transfer function =
U (s) (s + 2)(s + 3)(s + 4)
is factored as
Y (s) 12 1 12
= + +
U (s) s + 2 s + 3 s + 4
The transfer function given above is rewritten as below :
1 1 1 1
s 1 s
Y (s) 2 1s 2
= + +
U (s) 1 ( 2s 1 ) 1 ( 3s 1 ) 1 ( 4s 1 )
The signal flow graph for the transfer function is drawn in Fig. 9.5.11 where in each
integrator (s 1) is assigned a state variable.
. 1
1 x1 s–1 x1 2
u y
–2
1 –1
.
1 x2 s–1 x2
1
–3 2
s–1
.
x3 x3
–4
⎡ x1 ⎤
⎢ ⎥
y = [1 1 1] ⎢ x2 ⎥
⎢⎣ x3 ⎥⎦
⎛ A2t ⎞
∵ eAt = ⎜ I + At + + ... ⎟⎟ ...(9.48)
⎜ 2!
⎝ ⎠
substitute Eq. (9.48) in (9.47)
dx(t)
∴ = AeAt x(0) ...(9.48a)
dt
The state Eq. (9.43) is
dx(t)
= Ax(t)
dt
Therefore, equating Eqs. (9.48) and (9.43) following relation is obtained,
Ax(t) = AeAt x(0)
or x(t) = eAt x(0) is the solution of state equation {dx (t)/dt} = Ax(t)
The initial state x(0) at t = 0, is shifted to a state x(t) at time t. This matrix exponential
eAt carries out the transition of the state during time t, eAt is termed as state transition
matrix and denoted as φ(t).
Solution of non-homogeneous state equation
The non-homogeneous state equation has initial condition x(0) along with the input
u(t) as given below
dx(t)
= Ax(t) + Bu(t) ...(9.49)
dt
The solution of Eq. (9.49) is obtained by rearranging the same as below :
dx(t)
Ax(t) = Bu(t)
dt
On premultiplying both sides by e At
At { At
e x (t) Ax(t)} = e Bu(t)
At At x(t) At
or e x (t) Ae =e Bu(t) ...(9.50)
d
∵ {e At x(t)} = e At x (t) Ae At x(t) ...(9.51)
dt
substitute Eq. (9.51) in (9.50) to obtain following relation
d At x(t)} At Bu(t)
{e =e ...(9.52)
dt
On integrating Eq. (9.52) between time limits 0 to t,
t
∫
At Aτ
e x(t) = x(0) + e Bu (τ) dτ
0
τ
∫
A( t τ)
or x(t) = eAt x(0) + e Bu (τ) dτ ...(9.52a)
0
∵ φ(t) = eAt
τ
∴ x(t) = φ(t) x(0) + ∫ 0
φ(t τ) Bu (τ) dτ ...(9.53)
The solution of non-homogeneous state Eq. (9.49) is given by Eq. (9.53). The first term
on the right hand side of the solution is the natural response (zero input response) where as
the second term forced response (zero state response).
440 LINEAR CONTROL SYSTEMS
The Eq. (9.53) is a state transition equation wherein the initial condition occurs at t = 0.
dx(t)
The state equation is given by, = Ax(t) + Bu(t) ...(9.54)
dt
The solution of the above linear time invariant homogeneous equation can be obtained
by Laplace transform method. Taking Laplace transform of the Eq. (9.54) on both sides,
s X(s) x(0) = AX(s) + BU(s) ...(9.55)
where x(0) is the initial state at t = 0.
On rearranging Eq. (9.55),
X(s) = (s I A) 1 x(0) + (s I A) 1 BU(s)
put φ(s) = (sI A) 1
∴ X(s) = φ(s) x(0) + φ(s) BU(s) ...(9.56)
1 1 1
where φ(t) = L φ(s) = L (s I A) is state transition matrix.
Taking inverse Laplace transform on both sides of equation (9.56)
L 1 X(s) = L 1 φ(s) x(0) + L 1 φ(s) BU(s)
or x(t) = φ(t) x(0) + L 1 φ(s) BU (s)
t
using convolution integral, x(t) = φ(t) x(0) + ∫ 0
φ(t τ) BU (τ) dτ. ...(9.57)
The equation (9.57) gives the solution of state equation (9.41) using Laplace transform
method.
⎡ 1 4⎤
The state transition matrix eAt for A = ⎢ is determined on substituting
⎣ 2 5 ⎥⎦
values of A, A2, A3, ... in the Eq. (9.58)
⎡ 1 4⎤
A= ⎢
⎣ 2 5 ⎥⎦
A2 = AA
⎡ 1 4⎤ ⎡ 1 4 ⎤ ⎡ − 7 − 16 ⎤
= ⎢
⎣ − 2 − 5⎥⎦ ⎢ − 2 − 5⎥ = ⎢ 8
⎣ ⎦ ⎣ 17 ⎥⎦
An = A A A ... up to n
⎡1 0 ⎤ ⎡ 1 4⎤ ⎡ − 7 − 16 ⎤ t2
∴ eAt = ⎢ ⎥ + ⎢ ⎥ t+⎢ + ...
⎣0 1 ⎦ ⎣ − 2 − 5⎦ ⎣ 8 17 ⎥⎦ 2
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 441
⎡ 7 2 16 2 ⎤
⎢1 + t − 2 t + ... 4 t − 2 t + ... ⎥
= ⎢ ⎥
⎢ − 2t + 8 t 2 + ... 1 − 5t + 17 t 2 ...⎥
⎢⎣ 2 2 ⎥⎦
t2 t3
∵ e t=1 t+ − + ...
2! 3!
3t 9t 2 27t3
and e =1 3t + − + ...
2! 3!
It can be verified that
7 2
1+t t + ... = 2e t e 3t
2
16 2
4t t + ... = 2e t 2e 3t
2
8
2t + t2 ... = e t + e 3t
2
17 2
1 5t + t ... = e t + 2e 3t
2
⎡ 2e− t − e−3 t 2e− t − 2e−3 t ⎤
∴ φ(t) = eAt = ⎢ ⎥
⎢⎣ − e− t + e−3 t − e− t + 2e−3t ⎥⎦
(2) Laplace transform method
The Laplace transform of the state transition matrix is given by
φ(s) = (s I A) 1
φ ... (9.59)
The state transition matrix φ(t) is determined by taking inverse Laplace transform on
both sides of equation (9.59).
∴ φ(t) = L–1(s I A) 1
φ
⎡ 1 4⎤
The state transition matrix φ(t) for A = ⎢ is determined using Laplace
⎣− 2 5 ⎥⎦
transform method as below :
⎡ s 0⎤ ⎡ 1 4⎤ ⎡s − 1 − 4 ⎤
(s I A) = ⎢ ⎥ ⎢ − 2 − 5⎥ = ⎢ 2
⎣0 s ⎦ ⎣ ⎦ ⎣ s + 5 ⎥⎦
1
Adj (s I A )
(s I A) =
sI−A
⎡s + 5 4 ⎤ ⎡s + 5 4 ⎤ ⎡s + 5 4 ⎤
⎢ − 2 s − 1⎥ ⎢ − 2 s − 1⎥ ⎢ − 2 s − 1⎥
⎣ ⎦ ⎣ ⎦
= = = ⎣ ⎦
⎡s − 1 − 4 ⎤ 2
s + 4s + 3 (s + 1)(s + 3)
⎢ 2 s 5 ⎥
⎣ + ⎦
442 LINEAR CONTROL SYSTEMS
⎡ s+5 4 ⎤
⎢ (s + 1)(s + 3) (s + 1)(s + 3) ⎥
= ⎢ ⎥
⎢ −2 s −1 ⎥
⎢ (s + 1)(s + 3) (s + 1)(s + 3) ⎥⎦
⎣
⎡ 2 −1 2 −2 ⎤
⎢s + 1 + s + 3 +
s + 1 s + 3⎥
= ⎢ ⎥
⎢ −1 1 −1 2 ⎥
⎢s + 1 + s + 3 s + 1 s + 3 ⎥⎦
+
⎣
The state transition matrix is
φ(t) = L 1 (s I A) 1
⎡ 2 −1 2 −2 ⎤
⎢s + 1 + s + 3 +
s + 1 s + 3⎥
=L 1 ⎢ ⎥
⎢ −1 1 −1 2 ⎥
⎢s + 1 + s + 3 s + 1 s + 3 ⎥⎦
+
⎣
⎡ 2e t − e−3t 2e− t − 2e−3t ⎤
or φ(t) = ⎢
φ ⎥
⎢⎣ − e− t + e−3t − e− t + 2e−3t ⎥⎦
(3) Cayley-Hamilton Theorem
The Cayley-Hamilton theorem is useful in matrix analysis and applied for the
determination of state transition matrix eAt.
The theorem states that every square matrix satisfies its own characteristic
equation.
The characteristic equation of a matrix A of order n is given by
λI A =0 ...(9.60)
which can be expressed in polynomial form of λ as below :
f (λ) = λn + a1 λn 1 + a2 λn 2 + ... + an = 0 ...(9.61)
As per Cayley-Hamilton theorem matrix A satisfies characteristic Eq. (9.61), thus
f (A) = An + a1An 1 + a2An 2 + ... + an I = 0 ...(9.62)
where, I is the identity matrix of order n.
The theorem is verified for the matrix A as given below :
⎡ 1 4⎤
A= ⎢ ⎥
⎣ − 2 − 5⎦
The characteristic equation is
λI A =0
⎡λ 0 ⎤ ⎡ 1 4⎤
∴ ⎢0 λ ⎥ − ⎢ − 2 − 5⎥ = 0
⎣ ⎦ ⎣ ⎦
λ −1 −4
or =0
2 λ+5
∴ (λ 1)(λ + 5) (2) ( 4) = 0 or λ2 + 4λ + 3 = 0
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 443
∴ f (λ) = λ2 + 4λ + 3
∵ λ2 + 4λ + 3 = 0
∴ f (λ) = 0
f (A) = A2 + 4A + 3I
⎡ 1 4⎤
∵ A= ⎢ ⎥
⎣ − 2 − 5⎦
⎡ 1 4⎤ ⎡ 1 4⎤ ⎡ 1 4⎤ ⎡1 0 ⎤
∴ f (A) = ⎢ ⎥ ⎢ ⎥ +4 ⎢ ⎥ +3 ⎢ ⎥
⎣ − 2 − 5⎦ ⎣ − 2 − 5⎦ ⎣ − 2 − 5⎦ ⎣0 1 ⎦
⎡ − 7 − 16 ⎤ ⎡ 4 16 ⎤ ⎡3 0 ⎤
= ⎢ ⎥ + ⎢ ⎥ + ⎢ ⎥
⎣ 8 17 ⎦ ⎣ − 8 − 20 ⎦ ⎣0 3 ⎦
⎡− 3 0⎤ ⎡3 0 ⎤
= ⎢ ⎥ + ⎢ ⎥ =0
⎣ 0 − 3⎦ ⎣0 3 ⎦
or f (A) = 0
Hence, Cayley-Hamilton theorem is verified.
Application of Cayley-Hamilton theorem for determination of state transition matrix
F(A) is function of matrix A such that the degree of F(A) is higher than or of the order
of A and if the characteristic equation of matrix A is given as
f (λ) = λn + a1λn 1 + a2λn 2 + ... + an = 0
then, according to Cayley-Hamilton theorem
f (A) = 0
As already stated the degree of F(A) is higher than the order of A, the same can be
expressed as
F(A) = D(A) f (A) + R(A)
Since, f (A) = 0
∴ F (A) = R(A)
R(A) can be written in polynomial form as
F(A) = R(A) = α0 I + α1A + α2A2 + ... + αn 1 An 1
n 1
= ∑
k=0
αk A k ... (9.63)
k
and also, eλt = ∑
k=0
α kλ k ...(9.66)
444 LINEAR CONTROL SYSTEMS
1 1
α1 = ⎛⎜ e− t − e−3t ⎞⎟
⎝2 2 ⎠
∵ eAt = α0 I + α1A
⎛3 1 ⎞ ⎡ 1 0 ⎤ ⎛ 1 − t 1 −3 t ⎞ ⎡ 1 4 ⎤
∴ eAt = ⎜ e− t − e−3 t ⎟ ⎢ ⎥ + ⎜ e − 2 e ⎟ ⎢− 2
⎝2 2 ⎠ ⎣0 1 ⎦ ⎝ 2 ⎠⎣ 5 ⎥⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 445
⎡ 3 − t e−3t ⎤ ⎡ e− t e−3 t ⎤
⎢ e − 0 ⎥ ⎢ − 2e− t − 2e−3 t ⎥
2 2 ⎢ 2 2
= ⎢ ⎥ ⎥
⎢ 3 −t e−3t ⎥ ⎢ −t −3 t 5 −t 5 ⎥
⎢ 0 e − ⎥ ⎢⎣ e + e − e − e−3 t ⎥
⎣ 2 2 ⎦ 2 2 ⎦
⎡ 2e t − e 3 t 2 e− t − 2e−3 t ⎤
or eAt = ⎢ − t −3 t
⎥
⎣⎢ e + e − e− t + 2 e−3 t ⎦⎥
(4) Determination of state transition matrix using matrix diagonalization
The state transition matrix eAt is conveniently determined when A matrix is given in
the form of a diagonal matrix. This is verified as below :
The diagonal form of matrix A is given by
⎡ λ1 0 ⎤
A= ⎢ ⎥ ...(9.71)
⎣ 0 λ2 ⎦
The state transition matrix eAt can be expressed in the form of a series as
A 2 t 2 A 3 t3
eAt = I + At + + + ... ...(9.72)
2! 3!
On substituting the value of A in (9.72) from (9.71)
⎡ λ12 2 ⎤ ⎡ λ 3 t3 ⎤
⎢ t 0 ⎥ ⎢ 0 ⎥
⎡1 0 ⎤ ⎡λ 0⎤ ⎢ 2! ⎥ + ⎢ 3! ⎥ + ...
eAt = ⎢ ⎥ + ⎢ 1 +
⎣0 1 ⎦ ⎣0 λ 2 ⎥⎦ ⎢ λ12 2 ⎥ ⎢ λ 3 t3 ⎥
⎢ 0 t ⎥ ⎢ 0 ⎥
⎣ 2! ⎦ ⎣ 3! ⎦
⎡ eλ1t 0 ⎤
or eAt = ⎢ λ2t
⎥
⎣⎢ 0 e ⎦⎥
Thus, the state transition matrix of a diagonal matrix having diagonal elements
λi (i = 1, 2, 3, ...) is also a diagonal one with its diagonal elements as exponential of
λit (i = 1, 2, 3, ...)
In case, the matrix is not in diagonal form, the same may be transformed into diagonal
form using a similarity transformation as described in the texts of matrix algebra. A non-
diagonal matrix A with distinct eigen values is considered for the transformation.
The homogeneous state equation is given below
x = Ax ...(9.73)
The variable x is transformed using following similarity transformation
x = Mz ...(9.74)
M is known as modal matrix of A
∴ x = M z ...(9.75)
Substituting in Eq. (9.73) from (9.74) and (9.75) the transformed equation Eq. (9.73) is
M z = AMz
Premultiplying on both sides by M 1
z = M 1 AMz
446 LINEAR CONTROL SYSTEMS
⎡ m11 ⎤
where x1 = ⎢ ⎥
⎣ m21 ⎦
⎧⎪ ⎡1 0 ⎤ ⎡ 1 4 ⎤ ⎫⎪ ⎡ m11 ⎤
∴ ⎨(− 1) ⎢ ⎥ −⎢ ⎥⎬ ⎢ ⎥ =0
⎩⎪ ⎣0 1 ⎦ ⎣ − 2 − 5⎦ ⎭⎪ ⎣ m21 ⎦
⎡ − 2 − 4 ⎤ ⎡ m11 ⎤
or =0
⎢ 2
⎣ 4 ⎥⎦ ⎢⎣ m21 ⎥⎦
∴ 2m11 4m21 = 0
and 2m11 + 4m21 = 0
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 447
∴ m11 = 2m21
select m11 = 1
∴ m21 = 0.5
⎡m ⎤ ⎡ 1 ⎤
Thus, x1 = ⎢ 11 ⎥ = ⎢ ⎥
⎣ m21 ⎦ ⎣ − 0.5⎦
(ii) The eigen vector x2 of matrix A associated with eigen value λ1 = 3 is obtained by
solving the following equation
[λ2 I A] x2 = 0
⎡m ⎤
where x2 = ⎢ 11 ⎥
⎣ m22 ⎦
⎧⎪ ⎡1 0 ⎤ ⎡ 1 4 ⎤ ⎫⎪ ⎡ m12 ⎤
∴ ⎨(−3) ⎢ ⎥−⎢ ⎥⎬ ⎢ ⎥ =0
⎩⎪ ⎣0 1 ⎦ ⎣ −2 −5 ⎦ ⎭⎪ ⎣ m22 ⎦
⎡ − 4 − 4 ⎤ ⎡ m12 ⎤
or =0
⎢2
⎣ 2 ⎥⎦ ⎢⎣m22 ⎥⎦
∴ 4m12 4m22 = 0
and 2m12 + 2m22 = 0
∴ m12 = m22
∴ select m12 = 1
m22 = 1
⎡m ⎤ ⎡1⎤
Thus, x2 = ⎢ 12 ⎥ = ⎢ − 1⎥
⎣ m22 ⎦ ⎣ ⎦
The modal matrix M is given by
M = [x1 : x2]
⎡ 1 1⎤
or M= ⎢ ⎥
⎣ − 0.5 − 1⎦
1⎡ 2 2⎤
and M = ⎢ ⎥
⎣ − 1 − 2⎦
Λ=M 1 AM
⎡ 2 2⎤ ⎡ 1 4⎤ ⎡ 1 1⎤ ⎡ 1 0⎤
or Λ= ⎢ ⎥ ⎢ − 2 − 5 ⎥ ⎢ − 0.5 − 1⎥ = ⎢ 0 − 3 ⎥
⎣ − 1 − 2 ⎦⎣ ⎦⎣ ⎦ ⎣ ⎦
⎡ e− t 0 ⎤
∴ eΛt = ⎢ ⎥
⎢⎣ 0 e−3 t ⎥⎦
and eAt = MeΛt M 1
⎡ 1 1⎤ ⎡ e− t 0 ⎤⎡ 2 2⎤
= ⎢ ⎥ ⎢ ⎥
−3 t ⎢ − 1 − 1 ⎥
⎣ − 0.5 − 1⎦ ⎢⎣ 0 e ⎥⎦ ⎣ ⎦
448 LINEAR CONTROL SYSTEMS
⎡ e− t e −3 t ⎤ ⎡ 2 2⎤
= ⎢ −t
⎥
−3 t ⎢ − 1 − 2 ⎥
⎣⎢ − 0.5e − e ⎦⎥ ⎣ ⎦
⎡ 2e− t e−3 t 2e− t 2e−3t ⎤
or eAt = ⎢ − t ⎥
⎢⎣ e + e−3t −t −3 t
e + 2e ⎥⎦
The matrices A and Λ = M 1 AM are similar and have the same characteristic
equation.
The representation of matrix A in phase variable form is given below :
⎡ 0 1 0 0 ... 0 ⎤
⎢ 0 0 1 0 ... 0 ⎥⎥
⎢
Α= ⎢ 0 0 0 1 ... 0 ⎥
⎢ ⎥
⎢ ⎥
⎢− a − a − a − a − an ⎥⎦
⎣ 1 2 3 4
The phase variable form of matrix A has a unit element above its diagonal and
negative coefficient elements in the last row. All other elements being zero.
The Modal matrix with reference to phase variable form of matrix A is denoted as P
and given by
⎡ 1 1 1 ... 1 ⎤
⎢ λ λ2 λ3 ... λ n ⎥⎥
⎢ 1
P = ⎢⎢ λ1 λ32 ... λ n2 ⎥
2
λ 22
⎥
⎢ ⎥
⎢ n −1 n −1 n −1 n −1 ⎥
⎢⎣λ1 λ2 λ3 ... λ n ⎥⎦
where λ1, λ2, λ3, ... λn are distinct eigen values of matrix A, which is given in phase variable
form.
The matrix Λ = P 1 AP is a diagonal one. The matrix P is known as Vandermonde's
matrix.
The matrices A and P 1 AP are similar and have the same characteristic equation.
The procedure given below explains the use of matrix diagonalization method to
determine state transition matrix for a given phase variable form of matrix A.
⎡ 0 1⎤
Let A= ⎢ ⎥
⎣ − 6 − 5 ⎦
It is given that matrix A is in phase variable form, hence Vandermonde's matrix P is
used for diagonalization.
The characteristic equation is given by
λI A =0
⎡λ 0 ⎤ ⎡ 0 1⎤
∴ ⎢
0 ⎥ − ⎢ − 6 − 5⎥ = 0
⎣ λ ⎦ ⎣ ⎦
⎡λ −1 ⎤
or ⎢6 λ + 5⎥ = 0
⎣ ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 449
∴ λ(λ + 5) (6)( 1) = 0
or λ2 + 5λ + 6 = 0 or (λ + 2)(λ + 3) = 0
The eigen values are λ1 = 2 and λ2 = 3
∴ Vandermonde s matrix
⎡1 1⎤
P= ⎢
⎣ λ1 λ 2 ⎥⎦
⎡ 1 1⎤
or P= ⎢ ⎥
⎣ − 2 − 3⎦
1 ⎡ 3 1⎤
and P = ⎢ ⎥
⎣ − 2 − 1⎦
∴ Λ = P 1 AP
⎡ 3 1⎤ ⎡ 0 1⎤ ⎡ 1 1⎤
= ⎢ ⎥ ⎢ − 6 − 5⎥ ⎢ − 2 − 3⎥
⎣ − 2 − 1 ⎦⎣ ⎦⎣ ⎦
⎡ − 6 − 2⎤ ⎡ 1 1⎤
= ⎢
⎣ 6 3 ⎦ ⎣ − 2 − 3 ⎥⎦
⎥ ⎢
⎡− 2 0⎤
∵ Λ= ⎢ ⎥
⎣ 0 − 3⎦
⎡ e−2 t 0 ⎤
∴ eΛt = ⎢ −3 t
⎥
⎢⎣ 0 e ⎥⎦
and eAt = P eΛt P 1
−2 t
⎡ 1 1⎤ ⎡ e 0 ⎤ ⎡ 3 1⎤
or eAt = ⎢ ⎥ ⎢ ⎥⎢ ⎥
⎣ − 2 − 3 ⎦ ⎣⎢ 0 e ⎦⎥ ⎣ −2 −1⎦
−2 t
⎡ e−2 t − e−3 t ⎤ ⎡ 3 1⎤
= ⎢ ⎥
−3 t ⎢ ⎥
⎢⎣ − 2e−2 t − 3e ⎥⎦ ⎣ −2 −1⎦
⎡ 3e−2 t − 2e−3 t −2 t
−e − e
−3 t ⎤
= ⎢ ⎥
⎢⎣ − 6 e−2 t + 6 e−3 t − 2e−2 t + 3e−3 t ⎥⎦
Invariance of eigen values under similarity transformation
The similarity transformation
x = Px
P is a non-singular matrix used for transforming the system matrix A into a desired
form. The determinant sI A on such transformation remains unchanged as shown
below :
The system matrix on transformation becomes P 1 AP and the determinant is
sI P 1 AP which can be written as :
= sI P 1 AP
= sP 1P P 1 AP
450 LINEAR CONTROL SYSTEMS
= P 1(sI A)P
= P1 (sI A) P
= P1 P (sI A)
= P 1P sI A
= sI A
Since the eigen values can be determined by using sI A = 0, therefore the
invariance of eigen values under similarity transformation is established.
∴ φ 1(t) = φ( t)
A(t A(t
t ) t ) A(t t )
(3) φ(t2 t1) φ(t1 t0) = e 2 1 e 1 0 = e 2 0
∴ φ φ(t2 t1) φ(t1 t0) = φ(t2 t0)
(4) φ(t)]k = eAt eAt eAt ... = eAkt
[φ upto k terms = eAkt
∴ φ(t)]k = φ(kt)
[φ k = positive integer
φ (t) = d
(5) eAt = AeAt
dt
∴ φ (t) = A φ(t)
G(s) = Y(s)
U(s)
Therefore, Y(s) = G(s) U(s) ...(9.86)
In expanded form Eq. 9.70 can be written as
Y1 (s) ⎡ G11 (s) G12 (s) ... G1m (s) ⎤ ⎡ U1 (s) ⎤
Y2 (s) ⎢ G (s) G (s) .... G (s) ⎥ ⎢ ⎥
⎢ 21 22 2m ⎥ ⎢ U2 (s) ⎥
: ⎢ : : : ⎥ ⎢ : ⎥
= ⎢ ⎥ ⎢ ⎥ ...(9.87)
: ⎢ : : : ⎥ ⎢ : ⎥
: ⎢ : : : ⎥ ⎢ : ⎥
⎢ ⎥ ⎢ ⎥
Yp (s) ⎣⎢G p1 (s) G p2 (s) ... G pm (s) ⎥⎦ ⎢⎣U m (s) ⎥⎦
The transfer function for multiple-input multiple-output system is given below
G(s) = C(sI A) 1 B + D ...(9.88)
9.8 CONTROLLABILITY
The controllability is in relation to transfer of a system from one state to another by
appropriate input controls in a finite time. Sometimes such a control is not possible and this
can be verified by using controllability test matrix. The concept of controllability and
derivation of controllability test matrix are given below :
Initial Initial
cond. cond.
x2(t0) x1(t0)
x2 x2 x1 x1
u + + +
+ + +
–2 2
The block diagram representation of state Eqs. (9.89) and (9.90) is shown in Fig. 9.8.1.
x 1 = 2x1 + x2 +u ...(9.89)
x 2 = 2x2 ...(9.90)
In the block diagram representation (Fig. 9.8.1), it is noted that the state x1 is
controllable whereas the state x2 is independent of the input control u. Hence, the input has
no control over the state x2 and as such it is not possible to operate the state x2 from its initial
state x2(t0) to the final state x2 (tf) in a finite time period (tf t0), using input control u. Thus,
the system is state uncontrollable. The term controllability is also referred to as state
controllability.
A system is said to controllable, if it is possible to have an input u to transfer the
system from any given initial state x(t0) to any given final state x(tf) over a specified interval
of time (tf t0).
As stated above, the dependence of state variables on the input gives the concept of
controllability. Consider the following state equations.
x = Ax + Bu ...(9.91)
y = Cx + Du ...(9.92)
where x = velocity vector (n × 1) x2
x = state vector (n × 1) x(n)
u = input vector (m × 1) x(0)
x(2)
y = output vector (p × 1)
A = matrix (n × n) x1
x(1)
B = matrix (n × m)
C = matrix (p × n)
Fig. 9.8.2. Control of state vector.
D = matrix (p × m)
The initial state of a system having matrix A (2 × 2) can be represented as x(0) and
final as x(n) as shown in Fig. 9.8.2. The condition under which an input control can drive the
system from initial state x(0) to final state x(n), concerns controllability. The mathematical
relations given below explain the transfer process of a control system from initial state x(0)
to final state x(n),
x(1) = Ax(0) + Bu(0) ...(9.93)
x(2) = Ax(1) + Bu(1) = A {Ax(0) + Bu(0)} + Bu(1)
x(3) = Ax(2) + Bu(2) = A{Ax(1) + Bu(1)} + Bu(2)
:
:
x(n) = An x(0) + An 1 Bu(0) + An 2 Bu(1) + ... + Bu(n 1)
⎡ u(n − 1) ⎤
⎢ u(n − 2)⎥
⎢ ⎥
⎢ : ⎥
Thus, x(n) An x(0) = [B : AB : A2B : ... An 1 B]
⎢ ⎥ ...(9.94)
⎢ : ⎥
⎢ u(1) ⎥
⎢ ⎥
⎢⎣ u(0) ⎥⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 453
For the system described by equations (9.91) and (9.92), L.H.S. of the equation (9.94)
gives initial and final state. For the system to be completely state controllable, it is necessary
and sufficient that the n × nm matrix given below has a rank of n.
U = [B : AB : A2B : ... : An 1B] ...(9.95)
The matrix U is known as controllability test matrix, which is n × nm
The controllability condition of a system depends on the coefficient matrices A and B,
and it is said that the pair A, B is controllable indicating that the rank of the test matrix is n.
However, if controllability matrix U is n × n i.e. square matrix, then the condition for
state controllability is U ≠ 0 i.e. matrix be non singular.
⎡ x1 ⎤ ⎡ 1 1⎤ ⎡ x1 ⎤ ⎡1 ⎤
Solution. ⎢ x ⎥ = ⎢0 −1⎥ ⎢ x ⎥ + ⎢0 ⎥ u
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡1 ⎤ ⎡ 1 1⎤
∴ B= ⎢ ⎥ and A= ⎢ ⎥
⎣0 ⎦ ⎣ 0 −1 ⎦
⎡ 1 1 ⎤ ⎡1 ⎤ ⎡1 ⎤
AB = ⎢ ⎥ ⎢ ⎥ = ⎢ ⎥
⎣ 0 −1 ⎦ ⎣ 0 ⎦ ⎣0 ⎦
⎡1 1 ⎤
U = [B : AB] = ⎢ ⎥
⎣0 0 ⎦
U = (1 × 0 1 × 0) = 0
The test matrix U is found to be singular, hence the rank of U is not equal to n(n = 2)
and the system is uncontrollable. Ans.
⎡1 ⎤ ⎡− 2 0⎤
Solution. B = ⎢ ⎥ and A = ⎢ ⎥
⎣1 ⎦ ⎣ 0 − 1⎦
⎡− 2 0 ⎤ ⎡1⎤ ⎡ 2 ⎤
AB = ⎢ ⎥⎢ ⎥ = ⎢ ⎥
⎣ 0 − 1⎦ ⎣1⎦ ⎣ 1 ⎦
⎡1 2⎤
U = [B : AB] = ⎢
1 ⎥⎦
∴
⎣1
U = {1 × ( 1) 2 × ( 2)} = 1
The test matrix U is found to nonsingular, hence the rank of the test matrix U is equal
to n (n = 2) and the system is controllable.
454 LINEAR CONTROL SYSTEMS
9.9 OBSERVABILITY
The method of determining the state of a system by observing its output concerns
observability. Sometimes it is desirable to get information about the state variables from the
measurements of the output and input. In case, it is not possible to obtain any one of the
states from the measurements of the output, the state is said to be unobservable and the
system as a whole is unobservable. The condition under which the state of a system is
measured from the output is explained below :
Initial Initial
cond. cond.
x2(t0) x1(t0)
x2 x2 x1 x1
u + +
+ +
–1 –2
⎡ y(0) ⎤ ⎡ C ⎤
⎢ y (1) ⎥ ⎢ CA ⎥
⎢ ⎥ ⎢ ⎥
⎢ y(2) ⎥ ⎢ CA 2 ⎥
or ⎢ ⎥ = ⎢ ⎥ x(0)
⎢ : ⎥ ⎢ : ⎥
⎢ : ⎥ ⎢ ⎥
⎢ ⎥ ⎢ : ⎥
⎣⎢ y(n − 1) ⎦⎥ ⎢⎢CA n − 1 ⎥⎥
⎣ ⎦
−1
⎡ C ⎤ ⎡ y(0) ⎤
⎢ CA ⎥ ⎢ y (1) ⎥
⎢ ⎥ ⎢ ⎥
⎢ CA 2 ⎥ ⎢ y(2) ⎥
∴ ⎢ ⎥ ⎢ ⎥ = x(0) ...(9.100)
⎢ : ⎥ ⎢ : ⎥
⎢ : ⎥ ⎢ : ⎥
⎢ ⎥ ⎢ ⎥
⎢⎢CA n − 1 ⎥⎥
⎣ ⎦ ⎣⎢ y(n − 1) ⎦⎥
For unique determination of state x(0), the matrix (9.100) must have inverse and
therefore
⎡ C ⎤
⎢ CA ⎥
⎢ ⎥
⎢ CA 2 ⎥
⎢ ⎥ ...(9.101)
⎢ : ⎥
⎢ : ⎥
⎢ ⎥
⎢⎢CA n − 1 ⎥⎥
⎣ ⎦
the matrix (9.101) is to have a rank of n. However, if the matrix (9.101) is n × n i.e. square
matrix then for establishing observability, it should be nonsingular.
The matrix (9.101) is known as observability test matrix which can be written into an
n × np row matrix form and denoted as V being given below:
V = [CT: ATCT : (AT)2 CT : ... : (AT)n 1 CT] ...(9.102)
If the observability condition is satisfied i.e. observability test matrix V be of rank n,
then it is said that C, A is observable pair.
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
Solution. ⎢ x ⎥ = ⎢ − 2 − 3 ⎥ ⎢ x ⎥ + ⎢1 ⎥ u
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
y = [1 1] ⎢ ⎥
⎣ x2 ⎦
456 LINEAR CONTROL SYSTEMS
⎡ 0 1⎤
A = ⎢ − 2 − 3 ⎥ , C = [1 1]
⎣ ⎦
⎡0 − 2 ⎤ ⎡1 ⎤
AT = ⎢ T
⎥,C = ⎢1 ⎥
⎣ 1 − 3⎦ ⎣ ⎦
⎡0 − 2 ⎤ ⎡1⎤ ⎡ − 2⎤
ATC T = ⎢ ⎥ ⎢ ⎥ = ⎢ ⎥
⎣ 1 − 3 ⎦ ⎣1⎦ ⎣ − 2⎦
The observability test matrix is
⎡1 − 2 ⎤
V = [CT : ATCT] = ⎢1 − 2 ⎥
⎣ ⎦
V = {1 × ( 2) 1 × ( 2)} = 0
The observability test matrix V is singular. The rank of matrix V is not equal to n(n = 2).
Therefore, the system is unobservable. Ans.
⎡. ⎤
⎢ x1 ⎥ ⎡ − 2 1⎤ ⎡ x1 ⎤ ⎡1⎤
(b) ⎢. ⎥ = ⎢ ⎥ ⎢ ⎥ + ⎢1⎥ u
⎣ 0 1⎦ ⎣ x2 ⎦ ⎣ ⎦
⎣⎢ x2 ⎦⎥
⎡x ⎤
y = [1 1] ⎢ 1 ⎥
⎣ x2 ⎦
⎡− 2 1⎤
A= ⎢ ⎥ , C = [1 1]
⎣ 0 − 1⎦
⎡− 2 0⎤ ⎡1⎤
AT = ⎢ ⎥ , CT = ⎢1⎥
⎣ 1 − 1⎦ ⎣ ⎦
⎡− 2 0 ⎤ ⎡1⎤ ⎡ − 2⎤
AT C T = ⎢ ⎥ ⎢ ⎥ = ⎢ ⎥
⎣ 1 − 1⎦ ⎣1⎦ ⎣0⎦
The observability test matrix is
⎡1 − 2 ⎤
V = [CT : ATCT] = ⎢
⎣1 0 ⎥⎦
V = {1 × 0 1 × ( 2)} = 2
The observability test matrix V is nonsingular. The rank of the matrix V is equal to n
(n = 2). Therefore, the system is observable. Ans.
Example 9.10.1. For the electrical network shown in Fig. 9.10.1, determine the state
model. Consider i1, i2 and vc as state variables. The output variable are i1 and i2.
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 457
R2 L2 L1 R1
e2 i2 C vc i1 e1
Example 9.10.2. Determine the state model for the electrical circuit shown in
Fig. 9.10.2. Select state and output variables.
R1 R3
i1 + i2
R2 i2
v
e C
i1
L
dv R1i1 v
= − ...(7)
dt C( R1 + R3 ) C ( R1 + R3 )
Put i1 = x1 and v = x2, therefore,
( R1 R2 + R2 R3 + R3 R1 ) x1 R1 x2 R3 e
x 1 = + + ...(8)
L( R1 + R3 ) L( R1 + R3 ) L( R1 + R3 )
R1 x1 x2 e
x 2 = − + ...(9)
C( R1 + R3 ) C( R1 + R3 ) C( R1 + R3 )
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 459
Select (i1 + i2) as output variable, therefore,
y = (i1 + i2) ...(10)
The state model is given below :
⎡ ( R1 R2 + R2 R3 + R3 R1 ) R1 ⎤ ⎡ R3 ⎤
⎢ L( R + R ) 0 ⎥ ⎡ e⎤
⎡ x1 ⎤ ⎢ L( R1 + R3 ) L( R1 + R3 ) ⎥ ⎡ x1 ⎤ 1 3
⎢ x ⎥ = ⎢ ⎥ ⎢ ⎥+ ⎢ ⎥⎢ ⎥
⎣ 2⎦ ⎢ R1 1 ⎥ ⎣ x2 ⎦ ⎢ 1 ⎥ ⎣ e⎦
⎢ 0
C ( R1 + R3 ) ⎥⎦ ⎢ C( R1 + R3 ) ⎥⎦
−
⎣ C ( R1 + R3 ) ⎣
⎡ x1 ⎤
and y = [1 1] ⎢ ⎥ Ans.
⎣ x2 ⎦
Example 9.10.3. Obtain the state equations for the field controlled D.C. motor shown
in Fig. 9.10.3
Rf
Kf ia (const)
Lf qm, Jm, fm
ef if
Solution. The differential equations for the field control D.C. motor are given below :
dif
ef = Rf if + Lf ...(1)
dt
Torque developed : Te = Kf if
2
d θm dθ m
Load torque : Tm = J m + fm
dt
2 dt
Load torque equals developed torque. Therefore,
2
d θm dθ m
K f i f = Jm + fm ...(2)
dt
2 dt
On rearranging Eqs. (1) and (2) following equations are obtained :
dif Rf if ef
= + ...(3)
dt Lf Lf
d 2 θm fm dθm K f if
= . + ...(4)
dt 2 Jm dt Jm
dθ m
Let, if = x1, θm = x2 and = x3 be state variables and y = θm be the output variable.
dt
Therefore, state equations are as written below :
460 LINEAR CONTROL SYSTEMS
Rf ef
x1 = x1 +
Lf Lf
x 2 = x3
Kf fm
x 3 = x1 x and y = θm
Jm Jm 3
⎡ Rf ⎤
⎢ 0 0 ⎥ ⎡1 ⎤
⎡ x1 ⎤ ⎢ Lf ⎥ ⎡ x1 ⎤ ⎢ ⎥
⎢ ⎥ ⎢ ⎢ ⎥ ⎢ Lf
or ⎢ x2 ⎥ = ⎢ 0 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0
⎥e
⎥ ⎥ f
⎢⎣ x 3 ⎥⎦ ⎢ K f fm ⎥ ⎢⎣ x3 ⎥⎦ ⎢ ⎥
⎢ J 0 ⎣⎢ 0 ⎦⎥
⎢⎣ m Jm ⎥⎥⎦
⎡ x1 ⎤
⎢ ⎥
y = [0 1 0] ⎢ x2 ⎥ Ans.
⎢⎣ x3 ⎥⎦
Example 9.10.4. Draw the state block diagram for the transfer function given below
and obtain state equations.
C(s) 1 C(s) 2s + 1
(a) = and (b) = .
R(s) (s + 1) (s + 3) R(s) s 2 + 2
Solution. (a) The transfer function is composed of two terms 1/(s + 1) and 1/(s + 3).
The state block diagram for individual terms is drawn and then joined as shown in Fig.
9.10.4.
r(t) x2 x2 x1 x1
+
+ c(t) = y
– –
1 3
1 1
s+1 s+3
Fig. 9.10.4. State block diagram for C(s)/R(s) = 1/(s + 1)(s + 3).
In Fig. 9.10.4 the output of each integrator is designated by a state variable and thus
the state equations are
x1 = 3x1 + x2
x 2 = x2 + r(t) and y = x1
In matrix form the state equations are
⎡ x1 ⎤ ⎡ 3 1⎤ ⎡ x1 ⎤ ⎡0 ⎤ ⎡ x1 ⎤
⎢ x ⎥ = ⎢ 0 1⎥⎦ ⎢ x ⎥ + ⎢1 ⎥ r(t) and y = [1 0] ⎢ ⎥ Ans.
⎣ 2⎦ ⎣ ⎣ 2⎦ ⎣ ⎦ ⎣ x2 ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 461
(b) The transfer function can be re- Input + x2 x1 = x2 x1
written as
Output
C(s) 1 2s –
= +
R( s) s2 + 2 s2 + 2 2
The state diagram for the term + 2) 1/(s2
Fig. 9.10.5. State block diagram for term 1/(s2 + 2).
is shown in Fig. 9.10.5.
The derivative signal s/(s2 + 2) is available before the last integrator in Fig. 9.10.5,
therefore, the state block diagram for transfer function (2s + 1)/(s2 + 1) takes the form as
shown in Fig. 9.10.6.
r(t) x2 +
+ + c(t)
x2 = x1 x1 y = x1 + 2x2
–
Fig. 9.10.6. State block diagram for C(s)/R(s) = (2s + 1)/(s2 + 1).
2
d y
3dy du
Solution. + 4y =+ + 3u
dt dt 2 dt
Assuming zero initial conditions and taking Laplace transform on both sides of the
above equation :
Y (s) s+3
Y(s) (s2 + 3s + 4) = U(s) (s + 3) or = 2
U (s) s + 3s + 4
The state block diagram as per above transfer function is shown as drawn in
Fig. 9.10.7.
462 LINEAR CONTROL SYSTEMS
1
x1 = x2
x2 x1 +
u + + +
3 y
+ +
–3
–4
The output of each integrator is assigned a state variable as shown in Fig. 9.10.7.
Therefore, the state equations are
x1 = x2
x 2 =
4x1 3x2 + u
The output equation is given by
y = 3x1 + x2
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤ ⎡ x1 ⎤
or ⎢ x ⎥ = ⎢ 4 ⎢ x ⎥ + ⎢1 ⎥ u and y = [3 1] ⎢ ⎥ . Ans.
⎣ 2⎦ ⎣ 3 ⎥⎦ ⎣ 2⎦ ⎣ ⎦ ⎣ x2 ⎦
Y (s) s 2 + 3s + 2
= 3
U (s) s + 9s 2 + 26s + 24
Determine the state model. Use direct decomposition method.
2
Y (s) s + 3s + 2
Solution. =
U (s) s3 + 9s2 + 26 s + 24
Y (s) 2 3s s2
or = 3 + 3 + 3
U (s) s + 9 s + 26 s + 24 s + 9 s + 26 s + 24 s + 9 s + 26 s + 24
2 2 2
2
The state block diagram for the term is drawn as shown in
s3 + 9 s2 + 26 s + 24
Fig. 9.10.8 (a).
x3 x2 = x3 x1 = x2
u + + + x1
2
+ + +
–9
–26
–24
Fig. 9.10.8 (a) State block diagram for the term 2/(s3 + 9s2 + 26s + 24).
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 463
The output of each integrator is assigned a state variable as shown in Fig. 9.10.8 (a).
1
1. At the output of integrator no. 1 from RHS, the term is available.
s3 + 9 s2 + 26 s + 24
s
2. At the output of integrator no. 2 from RHS, the term 3 2
is available.
s + 9 s + 26 s + 24
s2
3. At the output of integrator no. 3 from RHS, the term 3 2
is available.
s + 9 s + 26 s + 24
2
Y ( s) s + 3s + 2
In view of points (1), (2) and (3) the state block diagram for = is
U ( s) s3 + 9 s2 + 26 s + 24
drawn as shown in Fig. 9.10.8 (b).
3
x2 = x3 x1 = x2
+ + + x3 x1 + + +
u
2 y
+
+ + + 2x1 + 3x2 + x3 = y
–9
–26
–24
Fig. 9.10.8 (b) State block diagram for Y(s)/U(s) = (s2 + 3s + 2)/(s3 + 9s2 + 26s + 24).
As per the state block diagram Fig. [9.10.8 (b)], the state equations are written below :
x1 = x2
x 2 = x3
x 3 = 24x1 26x2 9x3 + u
and y = 2x1 + 3x2 + x3
⎡ x1 ⎤ ⎡ 0 1 0⎤ ⎡ x1 ⎤
⎢ ⎥ ⎢ ⎢ ⎥ ⎡0 ⎤
or ⎢ x2 ⎥ = ⎢ 0 0 1⎥⎥ ⎢ x2 ⎥ + ⎢1 ⎥ u
⎢⎣ x 3 ⎥⎦ ⎢⎣ 24 26 9 ⎥⎦ ⎢⎣ x3 ⎥⎦ ⎣ ⎦
⎡ x1 ⎤
⎢ ⎥
and y = [2 3 1] ⎢ x2 ⎥ Ans.
⎢⎣ x3 ⎥⎦
x1
x1 x2 x2
+ x1 +
+
u
3
+
+ +
–1 –2
Fig. 9.10.9. State block diagram for Y(s)/U(s) = (s + 3)/(s + 1)(s + 2).
The output of each integrator in Fig. 9.10.9 is assigned a state variable as shown.
As per state block diagram, the state equations are formed as below :
x1 = x1 + u ...(1)
x 2 = 3x1 + x1 2x2 ...(2)
substituting for x1 in Eq. (2) from (1) following equations are obtained
x1 = x1 + u
x 2 = 2x1 2x2 + u
The output equation is : y = x2
⎡ x1 ⎤ ⎡ 1 0⎤ ⎡ x1 ⎤ ⎡1⎤
The state model is : ⎢ ⎥ = ⎢ ⎢ x ⎥ + ⎢1⎥ u
⎣ x 2 ⎦ ⎣ 2 2⎥⎦ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
and y = [0 1] ⎢ ⎥ Ans.
⎣ x2 ⎦
Y (s) (s + 3)
Example 9.10.8. The transfer function of a system is given by = .
U (s) (s + 1) (s + 2)
Obtain the state model in canonical form using parallel decomposition method.
Y (s) (s + 3)
Solution. =
U (s) (s + 1) (s + 2)
(s + 3) K1 K2
= +
(s + 1) (s + 2) (s + 1) (s + 2)
The coefficients K1 and K2 can be determined as : K1 = 2 and K2 = 1.
Y (s) 2 1
Therefore, = +
U (s) (s + 1) (s + 2)
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 465
The state block diagram for above transfer function is drawn as shown in Fig. 9.10.10.
x1
u + x1 +
2
2x1 – x2 = y
+ +
–1
+ x2 x2
–1
+
–2
The output of each integrator is assigned a state variable as shown in Fig. 9.10.10 and
the state equations are written below :
x1 = x1 + u
x 2 = 2x2 + u and y = 2x1 x2
⎡ x1 ⎤ ⎡ 1 0⎤ ⎡ x1 ⎤ ⎡1⎤
or ⎢ x ⎥ = ⎢ 0 ⎢ x ⎥ + ⎢1⎥ u
⎣ 2⎦ ⎣ 2 ⎥⎦ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
and y = [2 1] ⎢ ⎥ Ans.
⎣ x2 ⎦
Example 9.10.9. Obtain the state space equations for the transfer function given as,
10(s + 1)
G(s) = 2
.
(s + 4) (s + 2)
10 (s + 1) K1 K2 K3
Solution. Let, + = +
(s + 5) (s + 2) ( s + 5) 2( s + 2) (s + 2)2
The coefficients can be determined as :
40 40 10
K1 = , K2 = , and K3 =
9 9 3
40 40 10
Therefore, G(s) = 9 + 9 + 3
(s + 5) ( s + 2) (s + 2)2
The state block diagram for the transfer function given above is constructed as shown
1
in Fig. 9.10.11. It is to be noted that the term 2
is obtained in the state block diagram
(s + 2)
1
by cascading two blocks of the term .
(s + 2)
466 LINEAR CONTROL SYSTEMS
x1
u + x1
– 40
+ +
9 y
+ + +
–5 40
9
+ x2 x2 + x3 x3
– 10
9
+ +
–2 –2
Fig. 9.10.11. State block diagram for G(s) = ( 40/9)/(s + 5) + (40/9)/(s + 2) + ( 10/3)/(s + 2)2.
The output of each integrator is assigned a state variable and the state equations
w.r.t. Fig. 9.10.11 are obtained as below :
x1 = 5x1 + u
x 2 = 2x2 + u
x 3 = x2 2x3
40 40 10
and y= x + x x
9 1 9 2 3 3
⎡ x1 ⎤ ⎡ 5 0 0⎤ ⎡ x1 ⎤ ⎡1 ⎤
⎢ ⎥ ⎢ ⎢ ⎥ ⎢ ⎥
or ⎢ x2 ⎥ = ⎢ 0 2 0 ⎥⎥ ⎢ x2 ⎥ + ⎢1 ⎥ u
⎢⎣ x 3 ⎥⎦ ⎢⎣ 0 1 2⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣0 ⎥⎦
⎡ x1 ⎤
40 40 10 ⎤ ⎢ ⎥
and y = ⎡⎢ x Ans.
⎣ 9 9 3 ⎥⎦ ⎢ 2 ⎥
⎢⎣ x3 ⎥⎦
The state block diagram for the above mentioned two parts is drawn and shown in
Fig. 9.10.12. The state equations formed as per Fig. 9.10.12 are given below :
x1 = x2 ...(1)
x 2 = 5x1 2x2 + x 3 + 2x3 ...(2)
x 3 = 3x3 + u ...(3)
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 467
1 1
x3 x1 = x2
+ x2 x1 +
u + x3 + +
2 4
+ + 4x1 + x2 = y
+ + +
3 –2
–5
Fig. 9.10.12. State block diagram for G(s) = (s + 2)(s + 4)/(s + 3)(s2 + 2s + 5).
⎡ x1 ⎤
⎢ ⎥
and y = [4 1 0] ⎢ x2 ⎥ Ans.
⎢⎣ x3 ⎥⎦
Example 9.10.11. Obtain the time response of the system given below :
x = Ax
⎡ 0 1⎤ T
where A= ⎢ ⎥ ; given x(0) = [1 1]
⎣ 2 0⎦
⎡ x1 ⎤
and y = [1 1] ⎢ ⎥ .
⎣ x2 ⎦
⎡ s 0⎤ ⎡ 0 1⎤ ⎡ s 1⎤
(sI A) = ⎢ ⎥ ⎢ 2 0 ⎥ = ⎢2
⎣0 s ⎦ ⎣ ⎦ ⎣ s ⎥⎦
468 LINEAR CONTROL SYSTEMS
1
⎡s
1 1⎤
(sI A)
= ⎢ ⎥
⎣2 s ⎦
(sI A) 1 is determined below :
⎡ s 2⎤
1. Arrange for minor of (sI A), i.e. ⎢ ⎥
⎣ 1 s⎦
⎡s 2⎤
2. Change to cofactor, i.e. ⎢ ⎥
⎣1 s ⎦
⎡ s 1⎤
3. Interchange rows and columns, i.e. ⎢ ⎥
⎣ 2 s⎦
⎡ s 1⎤
∴ Adj. (sI A) = ⎢ ⎥
⎣ 2 s⎦
⎡ s 1⎤ ⎡ s 1⎤
⎢ s ⎥⎦⎢ 2 s ⎥⎦
Adj (sI A) ⎣ 2 ⎣
4. (sI A) 1 = = = 2
sI A s 1 (s + 2)
2 s
⎡ s 1 ⎤
⎢ s2 + 2 s2 + 2 ⎥
= ⎢ ⎥
⎢ 2 s ⎥
⎢ 2 ⎥
⎣ s +2 s2 + 2 ⎦
∵ φ(s) = (sI A) 1
⎡ s 1 ⎤
⎢ s2 + 2 s + 2⎥
2
∴ φ(s) = ⎢ ⎥
⎢ 2 s ⎥
⎢ 2 2 ⎥
⎣ s + 2 s + 2⎦
∴ The state transition matrix φ(t) is
⎡ s 1 ⎤
⎢ s2 + 2 s + 2⎥
2
φ(t) = L 1 φ(s) = L 1 ⎢ ⎥
⎢ 2 s ⎥
⎢ 2 ⎥
⎣ s +2 s2 + 2 ⎦
⎡ 1
cos 2t sin 2t ⎤ ⎡ 1
⎢ 2 ⎥ cos 2t sin 2t ⎤
= ⎢ ⎥=⎢ 2 ⎥
⎢ 2 ⎢ ⎥
sin 2t cos 2t ⎥ ⎢⎣ 2 sin 2t cos 2t ⎥⎦
⎣⎢ 2 ⎦⎥
x(t) = φ(t) x(0)
⎡ 1
⎡ x1 (t) ⎤ ⎢ cos 2t sin 2t ⎤ ⎡1⎤
2 ⎥
∴ ⎢ x ( t) ⎥ ⎢
=
⎥ ⎢1⎥
⎣ 2 ⎦ ⎢
⎣ 2 sin 2t cos 2t ⎥⎦ ⎣ ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 469
1
∴ x1(t) = cos 2t + sin 2t
2
x2(t) = 2 sin 2t + cos 2t. Ans.
y = x1 x2
3
∴ y= sin 2t. Ans.
2
⎡ 0 1⎤ ⎡0 ⎤
x = ⎢ x + ⎢ ⎥ u(t)
⎣ 2 1⎥⎦ ⎣ 1⎦
where x1(0) = 1, x2(0) = 0 and u(t) = 1.
∵ φ(s) = (sI A) 1
⎡ s+1 1 ⎤
⎢ 2 2 ⎥
s +s+3 s + s + 3⎥
∴ φ(s) = ⎢
⎢ 2 s ⎥
⎢ 2 2 ⎥
⎣⎢ s + s + 3 s + s + 3 ⎦⎥
The state transition matrix φ(t) is
φ(t) = L 1 φ(s)
⎡ s+1 1 ⎤
⎢ 2 2 ⎥
s +s+3 s + s + 3⎥
=L 1 ⎢
⎢ 2 s ⎥
⎢ 2 2 ⎥
⎣⎢ s + s + 3 s + s + 3 ⎦⎥
x(t) = L 1 φ(s) x(0)
⎡ s+1 1 ⎤ ⎡ s+1 ⎤
⎢ 2 2 ⎥ ⎢ 2 ⎥
1 ⎢ s +s+3 s + s + 3 ⎥ ⎡1 ⎤ 1 ⎢ s + s + 3 ⎥
=L
⎥ ⎢⎣0⎥⎦
=L
⎢ 2 s ⎢ 2 ⎥
⎢ 2 ⎥ ⎢ 2 ⎥
⎣⎢ s + s + 3 s2 + s + 3 ⎦⎥ ⎣⎢ s + s + 3 ⎦⎥
⎡ (s + 0.5) + 0.5 ⎤
⎢ 2 2 ⎥
1 ⎢ (s + 0.5) + (1.65) ⎥
=L
⎢ 2 1.65 ⎥
⎢ 1.65 . 2 2⎥
⎢⎣ (s + 0.5) + (1.65) ⎥⎦
⎡ (s + 0.5) 0.5 1.65 ⎤
⎢ + . 2⎥
2
(s + 0.5) + (1.65) 2 1.65 2
(s + 0.5) + (1.65) ⎥
=L 1 ⎢
⎢ 1.65 ⎥
⎢ 1.21 . 2 2 ⎥
⎢⎣ (s + 0.5) + (1.65) ⎥⎦
Taking inverse Laplace transform, the zero input response is
⎡(e 0.5 t . cos 1.65 t + 0.3 e 0.5 t
. sin 1.65 t) ⎤
x(t) = ⎢ ⎥
0.5 t
⎣⎢ 1.21 e . sin 1.65 t ⎦⎥
(ii) Zero state response is given by
x(t) = L 1 φ(s) BU (s)
⎡ s+1 1 ⎤ ⎡ 1 ⎤
⎢ 2 2 ⎥ ⎢ s(s2 + s + 3) ⎥
⎢ s +s+3 s + s + 3 ⎥ ⎡0 ⎤ 1 1 ⎢ ⎥
=L 1
⎥ ⎢⎣1 ⎥⎦ s
=L
⎢ 2 s ⎢1 s ⎥
⎢ 2 ⎥ ⎢s . 2 ⎥
⎣⎢ s + s + 3 s2 + s + 3 ⎦⎥ ⎣ ( s + s + 3) ⎦
⎡ 1 ⎤ ⎡ 11 3 ⎤
⎢ s(s2 + s + 3) ⎥ ⎢ 3 s (s2 + s + 3) ⎥
=L 1 ⎢ ⎥=L 1⎢ ⎥
⎢ 1 ⎥ ⎢ 1 1.65 ⎥
⎢ 2 ⎥ ⎢ 1.65 . 2 2⎥
⎣ (s + s + 3) ⎦ ⎣ (s + 0.5) + (1.65) ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 471
Taking inverse Laplace transform, the zero state response is
⎡ ⎧ ⎤
⎛ 0.28 ⎞⎟ ⎫⎪⎥
2
⎢ 1 ⎪1 e
0.5 t 1
⎨ . sin ⎜ 1.65t + tan 1
⎬
⎢3 ⎜ 0.28 ⎟ ⎪⎥
x(t) = ⎢ ⎪⎩ 1 0.282 ⎝ ⎠ ⎭⎥
⎢ 1 ⎥
⎢ . e 0.5 t . sin (1.65 t) ⎥
⎣ 1.65 ⎦
⎡0.33 {1 1.04 e 0.5 t . sin (1.65 t + tan 1 3.42)}⎤
or x(t) = ⎢ ⎥
⎢⎣0.6 e 0.5 t . sin 1.65 t ⎥⎦
The total response
x(t) = Zero input response + Zero state response
0.5 t 0.5 t
⎡ x1 (t) ⎤ ⎡(e . cos 1.65 t + 0.3 e . sin 1.65 t) ⎤
∴ ⎢ x ( t) ⎥ = ⎢ ⎥
0.5 t
⎣ 2 ⎦ ⎢⎣ 1.21 e . sin 1.65 t ⎦⎥
⎡0.33 {1 1.04 e 0.5 t . sin (1.65 t + tan 1 3.42)}⎤
+ ⎢ ⎥
⎢⎣0.6 e 0.5 t . sin 1.65 t ⎥⎦
∴ x1(t) = 0.33 + e 0.5 t 1
[cos 1.65t + 0.3 sin 1.65t] 0.34 sin (1.65t + tan 3.42)]
x2(t) = 0.61 e 0.5t . sin 1.65t. Ans.
Example 9.10.13. Obtain the state transition matrix in the form eAt and determine
the time response for the system, x = Ax
⎡ 0 1⎤
where A = ⎢ ⎥ and x1(0) = 1, x2(0) = 1.
⎣ 2 0⎦
At 2 At 3
Solution. φ(t) = eAt = I + At + + + .....
2! 3!
⎡ 0 1⎤
A= ⎢ ⎥
⎣ 2 0⎦
⎡ 0 1⎤ ⎡ 0 1 ⎤ ⎡ 2 0⎤
A2 = ⎢ ⎥⎢ ⎥=⎢
⎣ 2 0⎦ ⎣ 2 0⎦ ⎣ 0 2 ⎥⎦
⎡ 2 0 ⎤ ⎡ 0 1⎤ ⎡ 0 2⎤
A3 = ⎢
2 ⎥⎦ ⎢⎣ 2 0 ⎥⎦ ⎢⎣4 0 ⎥⎦
=
⎣ 0
A 2 t2 A 3 t3
φ(t) = I + At + + + ...
2! 3!
⎡ 2t 2 ⎤ ⎡ 2t 3 ⎤
0 ⎥ ⎢ 0 ⎥
⎡1 0 ⎤ ⎡ 0 t ⎤ ⎢⎢ 1 × 2 ⎥+⎢ 1 × 2 × 3⎥
= ⎢ ⎥ +⎢ ⎥+ + ...
⎣ 0 1 ⎦ ⎣ 2t 0 ⎦ ⎢ 2t ⎥ ⎢ 4t 3 ⎥
2
⎢ 0 ⎥ ⎢ 0 ⎥
⎣⎢ 1 × 2 ⎦⎥ ⎣⎢ 1 × 2 × 3 ⎦⎥
472 LINEAR CONTROL SYSTEMS
⎡ 2 3 ⎤
2t 2t
⎢ 1 2
+ ... t
2×3
+ ...⎥
= ⎢ ⎥
⎢ 4t
3
2t
2 ⎥
⎢ 2t + ... 1 + ... ⎥
⎣⎢ 2×3 2 ⎦⎥
⎡ ⎛ 2 t2 ⎞⎛ 2 23 t3 ⎞ ⎤
⎢ ⎜1 2 + ... ⎟ ⎜ t + ... ⎟ ⎥
⎢ ⎝ 2 ⎠ ⎜⎝ 2 2 2×3 ⎟
⎠ ⎥
= ⎢ ⎥
⎢ ⎛ 3 t3 ⎞⎛ 2 t2 ⎞⎥
⎢ ⎜ 2 2t 2 2
2×3
+ ... ⎟ ⎜ 1 2
2
+ ... ⎟ ⎥
⎣⎢ ⎝ ⎠⎝ ⎠ ⎦⎥
⎡ cos 2t 1
sin 2t ⎤
= ⎢⎢ 2 ⎥
⎥
⎣⎢ 2 sin 2t cos 2t ⎦⎥
x(t) = (t) x(0)
φ
⎡ 1
⎡ x1 (t) ⎤ ⎢ cos 2t sin 2t ⎤ ⎡1⎤
2 ⎥
∴ ⎢ x ( t) ⎥ ⎢
=
⎥ ⎢1⎥
⎣ 2 ⎦ ⎢
⎣ 2 sin 2t cos 2t ⎥⎦ ⎣ ⎦
1
x1(t) = cos 2t + sin 2t
2
∴ x2(t) = 2 sin 2t + cos 2t . Ans.
⎡ 1 4⎤
Solution. A= ⎢
⎣ 2 5 ⎥⎦
The matrix A given above is a 2 × 2 general matrix as such eigen vector approach is
considered for its diagonalization.
The eigen values of λ1 and λ2 matrix of A are determined by solving following
equation :
λI A = 0
⎧⎪ ⎡ λ 0 ⎤ ⎡ 1 4 ⎤ ⎫⎪ ⎡λ 1 4 ⎤
A = ⎨⎢ ⎥ ⎢ 2 ⎬=
5 ⎥⎦ ⎭⎪ ⎢ 2 λ + 5⎥⎦
λI
⎩⎪ ⎣ 0 λ ⎦ ⎣ ⎣
∴ λI A = (λ 1) (λ + 5) (2) ( 4)
= λ2 + 4λ + 3 = (λ + 1) (λ + 3)
put (λ + 1) (λ + 3) = 0
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 473
Thus, two distinct eigen values are :
λ1 = 1 and λ2 = 3
Determination of eigen vectors of matrix A is given below :
Eigen vectors of matrix A are determined by solving the following equation :
{λiI A} xi = 0
where, λi and xi are respectively the eigen value and corresponding eigen vector of matrix A.
(i) The eigen vector x1 associated with eigen values λ1 is determined by solving the
equation given below :
{λ1I A} x1 = 0
⎡ 1 4⎤
∵ λ1 = 1 and A= ⎢
⎣ 2 5 ⎥⎦
⎪⎧ ⎡1 0 ⎤ ⎡ 1 4 ⎤ ⎪⎫
⎨( 1) ⎢ ⎥ ⎢ 2 ⎬ x =0
5 ⎥⎦ ⎭⎪ 1
∴
⎩⎪ ⎣0 1 ⎦ ⎣
⎡m ⎤ ⎡ 2 4 ⎤ ⎡ m11 ⎤
Let, x1 = ⎢ 11 ⎥ , ∴ ⎢ =0
⎣ m21 ⎦ ⎣ 2 4 ⎥⎦ ⎢⎣ m21 ⎥⎦
∴ 2m11 4m21 = 0
or 2m11 + 4m21 = 0
put m11 = 1 ∴ m21 = 0.5
⎡m ⎤ ⎡ 1 ⎤
Therefore, x1 = ⎢ 11 ⎥ = ⎢ ⎥
⎣ m21 ⎦ ⎣ 0.5⎦
(ii) The eigen vector x2 associated with eigen value λ2 is determined by solving the
equation given below :
{λ2I A} x2 = 0
⎡ 1 4⎤
∵ λ2 = 3 and A= ⎢
⎣ 2 5 ⎥⎦
⎪⎧ ⎡1 0 ⎤ ⎡ 1 4 ⎤ ⎪⎫
⎨( 3) ⎢ ⎥ ⎢ 2 ⎬ x =0
5⎥⎦ ⎭⎪ 2
∴
⎩⎪ ⎣0 1 ⎦ ⎣
⎡m ⎤ ⎡ 4 4 ⎤ ⎡ m12 ⎤
Let, x2 = ⎢ 12 ⎥ , ∴ ⎢ =0
⎣ m22 ⎦ ⎣ 2 2 ⎥⎦ ⎢⎣ m22 ⎥⎦
∴ 4m12 4m22 = 0
or 2m12 + 2m22 = 0
put m12 = 1 ∴ m22 = 1
⎡ m ⎤ ⎡ 1⎤
Therefore, x2 = ⎢ 12 ⎥ = ⎢ ⎥
⎣ m22 ⎦ ⎣ 1⎦
Thus, the modal matrix is given by :
⎡m m12 ⎤ ⎡ 1 1⎤
M = ⎢ 11 ⎥ =⎢
⎣ m21 m22 ⎦ ⎣ 0.5 1⎥⎦
474 LINEAR CONTROL SYSTEMS
1 ⎡ 2 2⎤ ⎡ 1 4⎤ ⎡ 1 1⎤ ⎡ 1 0⎤
M AM = ⎢
2 ⎥⎦ ⎢⎣ 2 ⎥ ⎢ 1⎥⎦ ⎢⎣ 0 3 ⎥⎦
=
⎣ 1 5 ⎦ ⎣ 0.5
∵ x = Mz
1 ⎡ 2 2 ⎤ ⎡1 ⎤ ⎡ 2 ⎤
z(0) = M x(0) = ⎢
2 ⎥⎦ ⎢⎣0 ⎥⎦ ⎢⎣ 1⎥⎦
=
⎣ 1
1
⎡ 2 2 ⎤ ⎡0 ⎤ ⎡ 2 ⎤
M B= ⎢ ⎥ ⎢1 ⎥ = ⎢ 2 ⎥
⎣ 1 2 ⎦⎣ ⎦ ⎣ ⎦
1 1
substituting the values of M AM, z(0) and M B in equation for Z(s), the equation given
below is obtained :
1 1
⎧⎪ ⎡ s 0 ⎤ ⎡ 1 0 ⎤ ⎫⎪ ⎡ 2 ⎤ ⎧⎪ ⎡ s 0 ⎤ ⎡ 1 0 ⎤ ⎫⎪ ⎡ 2⎤ 1
Z(s) = ⎨ ⎢ ⎥ ⎢ 0 ⎬ ⎢ 1⎥ + ⎨ ⎢0 s ⎥ ⎬
⎪⎩ ⎣0 s ⎦ ⎣ 3⎥⎦ ⎪⎭ ⎣ ⎦ ⎪⎩ ⎣ ⎦
⎢ 0
⎣ 3⎥⎦ ⎪⎭ ⎢ 2⎥ s
⎣ ⎦
1 1 ⎡ 2 ⎤
⎡s + 1 0 ⎤ ⎡ 2 ⎤ ⎡s + 1 0 ⎤ ⎢ s ⎥
= ⎢ ⎢ 2⎥
⎣ 0 s + 3⎥⎦ ⎢ 1⎥ + ⎢ 0
⎣ ⎦ ⎣ s + 3 ⎥⎦ ⎢ ⎥
⎣ s⎦
⎡ 1 0 ⎤ ⎡ 1 0 ⎤ ⎡ 2 ⎤
⎢s + 1 ⎥ ⎡ 2 ⎤ ⎢s + 1 ⎥ ⎢ s ⎥
= ⎢ ⎥ + ⎢ ⎥ ⎢ 2⎥
⎢ 0 1 ⎥ ⎢⎣ 1 ⎥⎦ ⎢ 1 ⎥ ⎢ ⎥
0
⎢⎣ s + 3 ⎥⎦ ⎢⎣ s + 3 ⎥⎦ ⎣ s⎦
⎡ 2 ⎤ ⎡ 2 ⎤ ⎡ 2 ⎤ ⎡ 2 2 ⎤
⎢ s + 1 ⎥ ⎢ s(s + 1) ⎥ ⎢ s + 1 ⎥ ⎢ s s + 1 ⎥
= ⎢ ⎥+⎢ ⎥=⎢ ⎥+⎢ ⎥
⎢ 1 ⎥ ⎢ 2 ⎥ ⎢ 1 ⎥ ⎢ 2 2 ⎥
+
⎢⎣ s + 3 ⎥⎦ ⎢⎣ s(s + 3) ⎥⎦ ⎢⎣ s + 3 ⎥⎦ ⎢⎣ 3s s + 3 ⎥⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 475
⎡2 ⎤
⎢s ⎥
or Z(s) = ⎢ ⎥
⎢ 2 1 ⎥
⎣⎢ 3s 3 (s + 3) ⎦⎥
∵ X(s) = MZ(s)
⎡2 ⎤ ⎡2 2 1 ⎤
⎡ 1 1 ⎤ ⎢s ⎥ ⎢ s 3s 3 (s + 3) ⎥
X(s) = ⎢ ⎢ ⎥=⎢ ⎥
1⎥⎦ ⎢ 2
∴
⎣ 0.5 1 ⎥ ⎢ 1+ 2 + 1 ⎥
⎢⎣ 3s 3 (s + 3) ⎥⎦ ⎢⎣ s 3s 3 (s + 3) ⎥⎦
⎡2 2 1 ⎤ ⎡4 1 ⎤
⎢ s 3s 3 (s + 3) ⎥ 1
⎢ 3s 3 (s + 3) ⎥
1
x=L ⎢
∴ ⎥=L ⎢ ⎥
⎢ 1+ 2 + 1 ⎥ ⎢ 1 + 1 ⎥
⎢⎣ s 3s 3 (s + 3) ⎥⎦ ⎢⎣ 3s 3 (s + 3) ⎥⎦
⎡ 4 1 e 3t ⎤
⎡ x1 ⎤ ⎢ 3 3 ⎥
∴ ⎢x ⎥ = ⎢ 1 1 ⎥ Ans.
⎣ 2⎦ ⎢ + e ⎥ 3 t
⎣ 3 3 ⎦
T
⎡1 ⎤ ⎡ x ⎤
The output equation is : y = ⎢ ⎥ ⎢ 1 ⎥
⎣0 ⎦ ⎣ x2 ⎦
⎡ x1 ⎤
or y = [1 0] ⎢ ⎥ ∴ y = x1
⎣ x2 ⎦
4 1 3t
or y= e Ans.
3 3
Alternative solution
As already determined,
⎡ 1 1⎤ 1 ⎡ 2 2⎤
M= ⎢ ;M = ⎢
⎣ 0.5 1⎥⎦ ⎣ 1 2⎥⎦
1 ⎡ 2 2⎤ ⎡ 1 4⎤ ⎡ 1 1⎤
Λ =M AM = ⎢
2 ⎥⎦ ⎢⎣ 2 ⎥ ⎢ 1⎥⎦
∴
⎣ 1 5 ⎦ ⎣ 0.5
⎡ 1 0⎤
= ⎢
⎣ 0 3 ⎥⎦
⎡e t 0 ⎤
∴ e∧t = ⎢ ⎥
3t
⎢⎣ 0 e ⎥⎦
eAt = Me∧t M 1
⎡ 1 1⎤ ⎡e t 0 ⎤⎡ 2 2⎤
= ⎢ ⎢ ⎥
⎣ 0.5 1⎥⎦ ⎢⎣ 0 3t ⎢
e ⎥⎦ ⎣ 1 2 ⎥⎦
476 LINEAR CONTROL SYSTEMS
t t
⎡ 1 1 ⎤ ⎡ 2e 2e ⎤
= ⎢ ⎢ ⎥
⎣ 0.5 1⎥⎦ ⎢ e 3t 3t
⎣ 2e ⎦⎥
⎡ 2e t e 3 t 2e
t
2e
3t ⎤
or eAt = ⎢ ⎥
⎢⎣ e t + e 3 t e
t
+ 2e
3t
⎥⎦
t
x(t) = φ(t) x(0) +
∫ 0
φ (t τ) Bu (τ) dτ
⎡1 ⎤ ⎡0 ⎤
∵ φ(t) = eAt ; x(0) = ⎢ ⎥ and B = ⎢1 ⎥
⎣0 ⎦ ⎣ ⎦
⎡ 2e t e 3t 2e
t
2e ⎤3t
⎡1 ⎤
∴ x(t) = ⎢ ⎥ ⎢0 ⎥
⎢⎣ e t + e 3 t e
t 3t
+ 2e ⎥⎦ ⎣ ⎦
⎡ 2e (t τ) 3 (t τ) (t τ) 3 (t τ) ⎤ ⎡0 ⎤
t e 2e 2e
+ ∫ 0
⎢
⎢⎣ e (t τ)
+e
3 (t τ)
e
(t τ)
+ 2e
3 (t τ)
⎥ ⎢ ⎥ dτ
⎥⎦ ⎣1 ⎦
⎡ (2e t 3t (t τ) 3 (t τ)
e )⎤ t ⎡ 2e 2e ⎤
= ⎢
⎢⎣( e t 3t
⎥+
e ) ⎥⎦
∫ 0
⎢
⎢⎣ e (t τ)
+ 2e
3 (t τ)
⎥ dτ
⎥⎦
⎡ (2e t 3t ⎡ 2e t . eτ 2 e 3 t . e3 τ ⎤
e )⎤ t
∴ = ⎢
⎣⎢( e
t 3t
⎥+
+ e ) ⎦⎥
∫ 0
⎢
t 3t
⎣⎢ e . e + 2e . e ⎦⎥
τ 3τ
⎥ dτ
t
⎡ t 2 3t 3τ ⎤
⎡ (2e t e 3 t ) ⎤ ⎢ 2e . e e .e
τ
3 ⎥
= ⎢ ⎥+⎢ ⎥
t 3t 2
⎢⎣( e + e ) ⎥⎦ ⎢ e t . eτ + e 3 t . e3τ ⎥
⎣ 3 ⎦0
⎧⎡ t t 2 3t 3t ⎤ ⎡ 2e t . e0 2 e 3t . e0 ⎤ ⎫
⎡ 2e t e 3 t ⎤ ⎪ ⎢ 2 e . e 3
e .e
⎥ ⎢ 3 ⎥⎪
= ⎢ ⎥ + ⎨⎢ ⎥ ⎢ ⎥⎬
t 3t 2 ⎢ e t . e0 + 2 e 3t . e0 ⎥ ⎪
⎣⎢ e + e ⎦⎥ ⎪ ⎢ e t . et + e 3t . e3t ⎥
⎩⎣ 3 ⎦ ⎣ 3 ⎦⎭
⎧⎡ 2 ⎤ ⎡ 2e t 2 e 3t ⎤ ⎫
⎡ 2e t e 3 t ⎤ ⎪ ⎢ 2 3 ⎥ ⎢ 3 ⎥⎪
= ⎢ ⎥ + ⎨⎢ ⎥ ⎢ ⎥ ⎬
⎢⎣ e t + e 3t ⎥⎦ ⎪ ⎢ 1 + 2 ⎥ ⎢ e t + 2 e 3t ⎥ ⎪
⎩⎣ 3⎦ ⎣ 3 ⎦⎭
⎡ 4 1 e 3t ⎤
⎡ x1 ⎤ ⎢ 3 3 ⎥
or ⎢x ⎥ = ⎢ 1 1 ⎥ Ans.
⎣ 2⎦ ⎢ + e 3t ⎥
⎣ 3 3 ⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 477
The output equation is
⎡x ⎤
y = [1 0] ⎢ 1 ⎥
⎣ x2 ⎦
∴ y = x1
4 1 3t
or y= e Ans.
3 3
⎧⎪ ⎡ λ 0 ⎤ ⎡ 0 1⎤ ⎫⎪ ⎡λ 1 ⎤
[λI A] = ⎨ ⎢ ⎥ ⎢ 6 ⎬=
⎪⎩ ⎣ 0 λ ⎦ ⎣ 5 ⎥⎦ ⎪⎭ ⎢6 λ + 5⎥
⎣ ⎦
∴ λI A = λ (λ + 5) ( 1) (6)
= λ2 + 5λ + 6 = (λ + 2) (λ + 3)
put (λ + 2) (λ + 3) = 0
Thus, the two distinct eigen values of matrix A are :
λ1 = 2 and λ2 = 3
The Vandermonde s matrix is given by
⎡1 1⎤ ⎡ 1 1⎤
P= ⎢
λ 2 ⎥⎦ ⎢⎣ 2 3⎥⎦
=
⎣ λ1
The inverse of matrix can be determined as :
⎡ 1 1⎤
Adj ⎢
⎣ 2 3⎥⎦ ⎡ 3 1⎤
P 1= =⎢
1 1 ⎣ 2 1⎥⎦
2 3
1 ⎡ 3 1⎤ ⎡ 0 1⎤ ⎡ 1 1⎤ ⎡ 2 0⎤
P AP = ⎢
1⎥⎦ ⎢⎣ 6 5 ⎥⎦ ⎢⎣ 2 3 ⎥⎦ ⎢⎣ 0 3 ⎥⎦
=
⎣ 2
∵ x = Pz
1 ⎡ 3 1⎤ ⎡0 ⎤ ⎡ 3 ⎤
∴ z(0) = P x(0) = ⎢
1⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣ 2 ⎥⎦
=
⎣ 2
⎡ 3 1
1⎤ ⎡0 ⎤ ⎡ 1⎤
P
B= ⎢
1⎥⎦ ⎢⎣1 ⎥⎦ ⎢⎣ 1⎥⎦
=
⎣ 2
Substituting the values of P 1 AP, z (0) and P 1 B in equation for Z(s), the following
equation is obtained :
1 1
⎪⎧ ⎡ s 0 ⎤ ⎡ 2 0 ⎤ ⎪⎫ ⎡ 3 ⎤ ⎪⎧ ⎡ s 0 ⎤ ⎡ 2 0 ⎤ ⎪⎫ ⎡ 1⎤
Z(s) = ⎨ ⎢ ⎥ ⎢ 0 ⎬ ⎢ 2 ⎥ + ⎨ ⎢0 s ⎥ ⎬ ⎢ 1⎥ U(s)
⎩⎪ ⎣0 s ⎦ ⎣ 3⎥⎦ ⎭⎪ ⎣ ⎦ ⎪⎩ ⎣ ⎦
⎢ 0
⎣ 3⎥⎦ ⎭⎪ ⎣ ⎦
1 1
⎡s + 2 0 ⎤ ⎡ 3⎤ ⎡ s + 2 0 ⎤ ⎡ 1 ⎤1
= ⎢
⎣ 0 s + 3 ⎥⎦ ⎢ 2⎥ + ⎢ 0
⎣ ⎦ ⎣ s + 3 ⎥⎦ ⎢ 1⎥ s
⎣ ⎦
⎡ 1 0 ⎤ ⎡ 1 0 ⎤⎡ 1 ⎤
⎢s + 2 ⎥ ⎡ 3 ⎤ ⎢s + 2 ⎥⎢ s ⎥
= ⎢ ⎥ +⎢ ⎥⎢
⎢ 0 1 ⎥ ⎢⎣ 2 ⎥⎦ ⎢ 1 ⎥ ⎢ 1 ⎥⎥
0
⎢⎣ s + 3 ⎥⎦ ⎢⎣ s + 3 ⎥⎦ ⎣ s ⎦
⎡ 3 ⎤ ⎡ 1 ⎤ ⎡ 3 ⎤ ⎡ 1 1 ⎤
⎢ s + 2 ⎥ ⎢ s(s + 2) ⎥ ⎢ s + 2 ⎥ ⎢ 2s 2(s + 2) ⎥
= ⎢ ⎥+⎢ ⎥=⎢ ⎥+⎢ ⎥
⎢ 2 ⎥ ⎢ 1 ⎥ ⎢ 2 ⎥ ⎢ 1 1 ⎥
+
⎢⎣ s + 3 ⎥⎦ ⎢⎣ s(s + 3) ⎥⎦ ⎢⎣ s + 3 ⎥⎦ ⎢⎣ 3s 3(s + 3) ⎥⎦
⎡ 3 + 1 1 ⎤
⎢ s + 2 2s 2(s + 2) ⎥
= ⎢ ⎥
⎢ 2 1 1 ⎥
+
⎢⎣ s + 3 3s 3(s + 3) ⎥⎦
∵ X(s) = PZ(s)
⎡ 3 + 1 1 ⎤
⎡ 1 1⎤ ⎢ s + 2 2s 2(s + 2) ⎥
∴ X(s) = ⎢ ⎢ ⎥
⎣ 2 3⎥⎦ ⎢ 2 1 1 ⎥
+
⎢⎣ s + 3 3s 3(s + 3) ⎥⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 479
⎡ 3 + 1 1 2 1
+
1 ⎤ ⎡1 + 5 5 ⎤
⎢ s + 2 2s 2(s + 2) s + 3 3s 3(s + 3) ⎥ ⎢ 6 s 2(s + 2) 3(s + 3) ⎥
∴ = ⎢ ⎥=⎢ ⎥
⎢ 6 1 1 6 1 1 ⎥ ⎢ 5 5 ⎥
+ + + +
⎢⎣ s+2 s s + 2 s + 3 s s + 3 ⎥⎦ ⎢⎣ s + 2 s + 3 ⎥⎦
⎡1 + 5 5 ⎤
⎢ 6 s 2(s + 2) 3(s + 3) ⎥
x=L 1 ⎢ ⎥
⎢ 5 5 ⎥
+
⎢⎣ s + 2 s + 3 ⎥⎦
⎡ 1 5 2t 5 e 3t ⎤
⎡ x1 ⎤ ⎢ + e ⎥
∴ ⎢x ⎥ = ⎢6 2 3
⎥
Ans.
⎣ 2 ⎦ ⎢ 5e 2 t + 5e 3 t ⎥⎦
⎣
⎡x ⎤
y = [6 1] ⎢ 1 ⎥ = 6x1 + x2
⎣ x2 ⎦
= 6 ⎛⎜ 1 + 5 e 2t 5
e⎞ + ( 5e 2t + 5e
⎟
3t 3t)
⎝6 2 3 ⎠
= (1 + 15e 2t 10e 3t 5e 2t + 5e 3t)
or y = (1 + 10e 2t 5e 3t) Ans.
Alternative solution
As already determined,
⎡ 1 1⎤ 1 ⎡ 3 1⎤
P= ⎢ ;P = ⎢
⎣ 2 3 ⎥⎦ ⎣ 2 1⎥⎦
1 ⎡ 3 1⎤ ⎡ 0 1⎤ ⎡ 1 1⎤
∴ Λ=P AP = ⎢
⎣ 2 1⎥⎦ ⎢⎣ 6 5 ⎥⎦ ⎢⎣ 2 3 ⎥⎦
⎡ 2 0⎤
= ⎢
⎣ 0 3 ⎥⎦
⎡ e 2t 0 ⎤
∴ e∧t = ⎢ ⎥
3t
⎢⎣ 0 e ⎥⎦
eAt = P e∧t P 1
⎡ 1 1⎤ ⎡ e 2t 0 ⎤⎡ 3 1⎤
= ⎢ ⎢ ⎥
⎣ 2 3 ⎥⎦ ⎢⎣ 0 3t ⎢
e ⎥⎦ ⎣ 2 1⎥⎦
⎡ 1 1⎤ ⎡ 3 e 2 t e 2t ⎤
= ⎢ ⎢ ⎥
⎣ 2 3 ⎥⎦ ⎢ 2 e 3 t 3t
e ⎦⎥
⎣
480 LINEAR CONTROL SYSTEMS
⎡ 3e 2 t 2e 3t e 2t
e 3t ⎤
= ⎢ ⎥
2t 3t 2t 3t
⎣⎢ 6 e + 6e 2e + 3e ⎦⎥
t
x(t) = φ(t) x(0) +
∫ 0
φ (t τ) Bu (τ) dτ
⎡1 ⎤ ⎡0 ⎤
∵ φ(t) = eAt ; x(0) = ⎢ ⎥ and B = ⎢1 ⎥
⎣0 ⎦ ⎣ ⎦
⎡ 3e 2t 3t 2t ⎤3t
2e e e ⎡1 ⎤
x(t) = ⎢ ⎥ ⎢0 ⎥
2t 3t 2t 3t
⎢⎣ 6 e + 6e 2e + 3e ⎥⎦ ⎣ ⎦
t ⎡ 3e 2( t τ)
2e
3( t τ)
e
2( t τ)
e
3( t τ) ⎤ ⎡0 ⎤
+ ∫ 0
⎢
⎢⎣ 6 e 2( t τ)
+ 6e
3( t τ)
2e
2( t τ)
+ 3e
3( t
⎥
τ) ⎢1 ⎥
⎥⎦ ⎣ ⎦
dτ
⎡ 3e 2 t 2e 3t ⎤ t ⎡ e 2( t τ)
e 3( t τ) ⎤
= ⎢ ⎥+
⎢⎣ 6 e 2 t + 6 e 3 t ⎥⎦ ∫ 0
⎢
⎢⎣ 2e 2( t τ)
+ 3e
3( t τ)
⎥ dτ
⎥⎦
⎡ 3e 2 t 2e 3t ⎤ t ⎡ e 2 t . e2 τ e 3 t . e3 τ ⎤
= ⎢ ⎥+
⎢⎣ 6 e 2 t + 6 e 3 t ⎥⎦ ∫ 0
⎢ ⎥ dτ
⎢⎣ 2e 2 t . e2τ + 3e 3 t . e3 τ ⎥⎦
t
⎡ 1 2 t . e2 τ 1 e 3 t . e3 τ ⎤
⎡ 3e 2 t 2e 3 t ⎤ ⎢ 2 e 3 ⎥
= ⎢ ⎥+⎢ ⎥
+ 6 e ) ⎦⎥ ⎢ 2 e 2t . e2 τ + 3 e 3 t . e3 τ ⎥
2t 3t
⎣⎢( 6 e
⎣ 2 3 ⎦0
⎡ 3e 2t 2e 3t ⎤ ⎡⎢⎛⎜ 1 e0 1 e0 ⎞⎟ ⎛1 e
⎜
2t 1
⎞⎤
⎟⎥e
3t
= ⎢ ⎥ + ⎝2 3 ⎠ ⎝2 3
⎠
⎢⎣ 6 e 2t + 6 e 3t ⎥⎦ ⎢⎢ 0 0 2t 3t
⎥
⎣ ( e +e ) ( e + e ) ⎦⎥
⎡ 3e 2t 2e 3t ⎤ ⎡⎢⎛⎜ 1 1 1 e 2t + 1 e 3t ⎞⎟ ⎤⎥
= ⎢ ⎥ + ⎝2 3 2 3 ⎠
⎢⎣ 6 e 2t + 6 e 3t ⎥⎦ ⎢⎢ 2t 3t
⎥
⎣ ( 1+1+ e e ) ⎥⎦
⎡ 1 5 2t 5 e 3t ⎤
⎡ x1 ⎤ ⎢ + e ⎥
or ⎢x ⎥ ⎢6 2
= 3
⎥
Ans.
⎣ 2⎦ ⎢ 5e 2 t
+ 5e 3 t
⎣ ⎦⎥
⎡ x1 ⎤
y = [6 1] ⎢ ⎥ = 6x1 + x2
⎣ x2 ⎦
1 5 5
= 6 ⎛⎜ + e ⎞ + ( 5e
2t 3t
e ⎟ 2t + 5e 3t)
⎝6 2 3 ⎠
= 1 + 10e 2t 5e 3t Ans.
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 481
Example 9.10.16. The state block diagram of a control system is shown below (Fig.
9.10.13). Write state equations and determine transfer matrix.
x1
+ + x1
2
+
–2 +
u x1 + x2 = y
4 +
+ x2
+ x2
+
–5
Solution. As per state block diagram (Fig. 9.10.13), state equations are :
x1 = x1 2x2 + 2u
x 2 = 4x1 5x2 + u
The output equation is : y = x1 + x2
⎡ x1 ⎤ ⎡1 − 2⎤ ⎡ x1 ⎤ ⎡2 ⎤
∴ ⎢ x ⎥ = ⎢4 − 5 ⎥ ⎢ x ⎥ + ⎢1 ⎥ u
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡x ⎤
and y = [1 1] ⎢ 1 ⎥ Ans.
⎣ x2 ⎦
The coefficient matrices are :
⎡1 − 2⎤ ⎡2⎤
A= ⎢ ⎥ , B = ⎢ ⎥ and C = [1 1]
⎣4 − 5⎦ ⎣1 ⎦
⎧⎪ ⎡ s 0 ⎤ ⎡1 − 2⎤ ⎫⎪ ⎡(s − 1) 2 ⎤
{sI A} = ⎨ ⎢ ⎥ −⎢ ⎥ ⎬=⎢
⎩⎪ ⎣0 s ⎦ ⎣4 − 5 ⎦ ⎪⎭ ⎣ − 4 (s + 5)⎥⎦
⎡s + 5 − 2 ⎤
⎢ s − 1⎥⎦
Adj (sI − A) ⎣ 4
(sI A) 1 = = 2
sI − A s + 4s + 3
⎡s + 5 − 2 ⎤ ⎡ s+5 2 ⎤
⎢ 4 s 1 ⎥ ⎢ (s + 1)(s + 3) −
(s + 1)(s + 3) ⎥
A) 1 = ⎣ ⎦ =⎢
−
or (sI ⎥
(s + 1)(s + 3) ⎢ 4 s −1 ⎥
⎢ (s + 1)(s + 3) (s + 1)(s + 3) ⎥⎦
⎣
482 LINEAR CONTROL SYSTEMS
Example 9.10.17. Determine the transfer matrix from the data given below :
⎡− 3 1⎤ ⎡1⎤
A= ⎢ ⎥ , B = ⎢ ⎥ , C = [1 1] and D = 0.
⎣ 0 − 1⎦ ⎣1⎦
⎡s + 1 1 ⎤
⎢ 0 s + 3⎥⎦
adj (sI − A ) ⎣
(sI A) 1 = =
sI − A (s + 1)(s + 3)
⎡ s+1 1 ⎤ ⎡ 1 1 ⎤
⎢ (s + 1)(s + 3) (s + 1)(s + 3) ⎥ ⎢ s + 3 (s + 1)(s + 3) ⎥
= ⎢ ⎥=⎢ ⎥
⎢ s+3 ⎥ ⎢ 0 1 ⎥
⎢ 0
⎣ (s + 1)(s + 3) ⎥⎦ ⎢⎣ s+1 ⎥⎦
⎡ 1 1 ⎤
⎢s + 3 (s + 1)(s + 3) ⎥ ⎡1⎤
∴ G(s) = [1 1] ⎢ ⎥ ⎢1⎥ , D = 0 (given)
⎢ 0 1 ⎥ ⎣ ⎦
⎢⎣ s+1 ⎥⎦
⎡ 1 + 1 ⎤
⎢ s + 3 (s + 1)(s + 3) ⎥ 1 1 1
= [1 1] ⎢ ⎥= + +
⎢ 1 ⎥ s + 3 ( s + 1)( s + 3) s +1
⎢⎣ s + 1 ⎥⎦
(s + 1) + 1 + ( s + 3)
=
( s + 1)(s + 3)
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 483
2s + 5
or G(s) = Ans.
(s + 1)(s + 3)
Example 9.10.18. Determine the transfer matrix for the system given below and
draw the block diagram
⎡ x 1 ⎤ ⎡ 0 3 ⎤ ⎡ x1 ⎤ ⎡1 1⎤ ⎡ 2 1 ⎤ ⎡ x1 ⎤
⎢ x ⎥ = ⎢ − 2 − 5 ⎥ ⎢ x ⎥ + ⎢1 1⎥ u(t) and y = ⎢ ⎥⎢ ⎥.
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦ ⎣ 1 0 ⎦ ⎣ x2 ⎦
⎡ s+5 3 ⎤
⎡ 2 1 ⎢
⎤ (s + 2)(s + 3) (s + 2)(s + 3) ⎥ ⎡1 1⎤
∴ G(s) = ⎢
1 0 ⎥⎢ ⎥ ⎢1 1⎥
⎣ ⎦ ⎢− 2 s ⎥ ⎣ ⎦
⎢⎣ (s + 2)(s + 3) (s + 2)(s + 3) ⎥⎦
⎡ (2s + 8) (s + 6) ⎤
⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥ ⎡1 1⎤
= ⎢ ⎥ ⎢1 1⎥
⎢ (s + 5) 3 ⎥ ⎣ ⎦
⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥⎦
⎣
⎡ (3s + 14) (3s + 14) ⎤
⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥
= ⎢ ⎥ Ans.
⎢ (s + 8) (s + 8) ⎥
⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥⎦
⎣
Y(s) = G(s) U(s)
⎡ (3s + 14) (3s + 14) ⎤
⎡ Y1 (s) ⎤ ⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥ ⎡U1 (s) ⎤
∴ ⎢ Y (s) ⎥ = ⎢ (s + 8) (s + 8) ⎥
⎥ ⎢U (s) ⎥
⎣ 2 ⎦ ⎢ ⎣ 2 ⎦
⎢ (s + 2)(s + 3) (s + 2)(s + 3) ⎥⎦
⎣
(3 s + 14) (3s + 14)
∴ Y1(s) = U (s) + U (s)
(s + 2)( s + 3) 1 ( s + 2)(s + 3) 2
( s + 8) (s + 8)
Y2(s) = U (s) + U (s)
(s + 2)( s + 3) 1 ( s + 2)(s + 3) 2
The block diagram for the system is shown in Fig. 9.10.14.
(3s + 14)
(s + 2) (s + 3)
(s + 8)
(s + 2) (s + 3)
U2(s) + y2(s)
(s + 8) +
(s + 2) (s + 3)
Example 9.10.19. The state equations of a control system are given below : Examine
for complete state controllability.
x 1 = − 1 x1 + 1 u and x 2 = − 1 x2 + 1 u .
T1 T1 T2 T2
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 485
⎡− 1 0 ⎤ ⎡1⎤
⎡ x1 ⎤
⎢ T1 ⎥ ⎡ 1 ⎤ ⎢ T1 ⎥
x
Solution. ⎢ x ⎥ = ⎢ ⎥ ⎢x ⎥ + ⎢ 1 ⎥ u
⎣ 2⎦ ⎢ 0 1⎥ ⎣ 2⎦ ⎢ ⎥
−
⎢⎣ T2 ⎥⎦ ⎢⎣ T2 ⎥⎦
⎡1⎤ ⎡− 1 0 ⎤
⎢ T1 ⎥ ⎢ T ⎥
B= ⎢ ⎥ and A= ⎢ 1 ⎥
⎢1⎥ ⎢ 0 −
1⎥
⎢⎣ T2 ⎥⎦ ⎢⎣ T2 ⎥⎦
⎡− 1 ⎡ 1 ⎤ ⎡− 1 ⎤
0 ⎤
⎢ T1 ⎢ T1 ⎥ ⎢ T12 ⎥
⎥
AB = ⎢ ⎢ ⎥=⎢ ⎥ ⎥
⎢ 0 ⎢ 1 ⎥ ⎢− 1 ⎥
−
1 ⎥
⎢⎣ ⎢⎣ T2 ⎥⎦ ⎢⎢ T 2 ⎥⎥
T2 ⎥⎦
⎣ 2 ⎦
1 1 ⎛ 1 ⎞ 1 1
or − − ⎜− ⎟ ≠0 or − ≠0
T1T2
2 T2 ⎜ T 2 ⎟ T2 T12
T1T22
⎝ 1 ⎠
1 ⎛ 1 1 ⎞ 1 ⎛ T2 − T1 ⎞
or ≠0 or ≠0
T1T2 ⎜⎝ T1 T2 ⎟⎠ T1T2 ⎜⎝ T1T2 ⎟⎠
−
1
or (T2 − T1 ) ≠ 0
(T1T2 )2
∴ T1 ≠ T2 Ans.
⎡ x1 ⎤ ⎡0 1 0⎤ ⎡ x1 ⎤ ⎡0 1⎤
Solution. ⎢ ⎥ = ⎢
1⎥⎥
⎢ ⎥ ⎢ ⎥ ⎡ u1 ⎤
⎢ x2 ⎥ ⎢0 0 ⎢ x2 ⎥ + ⎢0 0 ⎥ ⎢u ⎥
⎢⎣ x 3 ⎥⎦ ⎣ 2⎦
⎣⎢0 − 2 − 3⎦⎥ ⎣⎢ x3 ⎦⎥ ⎣⎢ 1 1⎦⎥
⎡ 0 1⎤ ⎡0 1 0⎤
B = ⎢⎢0 0 ⎥⎥ and A = ⎢0 0 1⎥⎥
⎢
⎢⎣ 1 1⎥⎦ ⎢⎣0 − 2 − 3⎥⎦
⎡0 1 0⎤ ⎡ 0 1⎤ ⎡ 0 0⎤
⎢
AB = ⎢0 0 1⎥⎥ ⎢0 0 ⎥ = ⎢ 1 1⎥⎥
⎢ ⎥ ⎢
⎢⎣0 − 2 − 3⎥⎦ ⎢⎣ 1 1⎥⎦ ⎢⎣ − 3 − 3 ⎥⎦
⎡0 1 0⎤ ⎡ 0 0⎤ ⎡ 1 1⎤
A2B ⎢
= ⎢0 0 1⎥⎥ ⎢ 1 1⎥ = ⎢ − 3 − 3⎥⎥
⎥ ⎢
⎢
⎢⎣0 − 2 − 3⎥⎦ ⎢⎣ − 3 − 3⎥⎦ ⎢⎣ − 7 7 ⎥⎦
Controllability test matrix is given by :
U = [B : AB : A2B]
⎡0 1 0 0 1 1⎤
∴ U = ⎢0 0 1 1 − 3 − 3 ⎥⎥
⎢
⎢⎣ 1 1 − 3 − 3 7 7 ⎥⎦
Collect the columns from U to form a square 3 × 3 to form non-zero determinant
matrix as given below :
⎡0 1 0⎤
⎢0 0 0 1 0 1 0 0
∴
⎢ 1⎥⎥ = 0 − (1) +0
1 −3 1 −3 1 1
⎣⎢ 1 1 − 3⎦⎥
= (1) {0 × ( 3) (1) (1)} = 1
⎡0 1 0⎤
As ⎢0 0 1⎥⎥ ≠ 0
⎢
⎣⎢ 1 1 − 3⎦⎥
Therefore, the rank of U is equal to 3 (n = 3), hence the system is completely state
controllable. Ans.
⎡ 0 1 0⎤ ⎡ 1⎤ ⎡10 ⎤
A = ⎢⎢ 0 0 1⎥⎥ , B = ⎢0 ⎥ and CT = ⎢⎢ 5 ⎥⎥ .
⎢ ⎥
⎣⎢ − 6 − 11 − 6 ⎦⎥ ⎢⎣ 1⎥⎦ ⎢⎣ 1 ⎥⎦
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 487
Solution. (a) Controllability test :
⎡1 ⎤ ⎡ 0 1 0⎤
B = ⎢0 ⎥ , A =⎢ 0 0 1⎥⎥
⎢ ⎥ ⎢
⎣⎢1 ⎦⎥ ⎢⎣ − 6 − 11 − 6 ⎥⎦
⎡ 0 1 0 ⎤ ⎡1 ⎤ ⎡ 0 ⎤
⎢
AB = ⎢ 0 0 1⎥⎥ ⎢⎢0 ⎥⎥ = ⎢⎢ 1⎥⎥
⎢⎣ − 6 − 11 − 6 ⎦⎥ ⎣⎢1 ⎦⎥ ⎣⎢ − 12⎦⎥
⎡ 0 1 0⎤ ⎡ 0⎤ ⎡ 0⎤
A 2B ⎢
= ⎢ 0 0 1⎥⎥ ⎢ 1⎥ = ⎢ − 12 ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ − 6 − 11 − 6 ⎥⎦ ⎢⎣ − 12 ⎥⎦ ⎢⎣ 61⎥⎦
Controllability test matrix is given by :
U = [B : AB : A2B]
⎡1 0 0⎤
∴ U = 0 1 − 12⎥
⎢
⎢ ⎥
⎣⎢ 1 12 61⎦⎥
1 − 12 0 − 12 0 1
U =1 +0 +0
12 61 1 61 1 12
∴ U ≠0
Therefore, the rank of U is equal to its order i.e. 3(n = 3) indicating that the system is
controllable. Ans.
(b) Observability test :
⎡10 ⎤
⎡ 0 1 0⎤
CT = ⎢⎢ 5 ⎥⎥ , A =
⎢ 0 0 1⎥⎥
⎢
⎣⎢ 1 ⎦⎥
⎢⎣ − 6 − 11 − 6 ⎥⎦
⎡0 0 − 6 ⎤
⎢ ⎥
A = ⎢ 1 0 − 11⎥
T
⎢⎣0 1 − 6 ⎦⎥
⎡0 0 − 6 ⎤
⎡10 ⎤ ⎡ − 6 ⎤
A TC T = ⎢⎢ 1 0 − 11⎥⎥
⎢ 5 ⎥ = ⎢ − 1⎥
⎢ ⎥ ⎢ ⎥
⎣⎢0 1 − 6 ⎦⎥
⎣⎢ 1 ⎦⎥ ⎢⎣ − 1 ⎦⎥
⎡0 0 − 6 ⎤ ⎡ − 6 ⎤ ⎡6 ⎤
⎢ ⎥⎢ ⎥ ⎢ ⎥
(A ) C = ⎢ 1 0 − 11⎥ ⎢ − 1 ⎥ = ⎢5 ⎥
T 2 T
⎢⎣0 1 − 6 ⎥⎦ ⎢⎣ − 1 ⎥⎦ ⎢⎣5 ⎥⎦
Observability test matrix is given by
V = [CT : ATCT : (AT)2 CT]
⎡10 − 6 6 ⎤
∴ V = ⎢ 5 − 1 5⎥
⎢ ⎥
⎣⎢ 1 − 1 5 ⎦⎥
488 LINEAR CONTROL SYSTEMS
−1 5 5 5 5 −1
V = 10 − (− 6) +6
−1 5 1 5 1 −1
= 10 [( 1) 5 5 ( 1)] + 6 ( 5 × 5 5 × 1) + 6 [(5)( 1) ( 1)(1)]
= 10 × 0 + 6 × 20 + 6 × ( 4) = 96
Observability matrix V ≠ 0, hence its rank i.e. 3 equals order i.e. 3, therefore system is
completely observable.
1
x2 = x3 x1 = x2
x3 x1 +
u + + +
2
+ 2x1 + x2 = y
+ + +
–9
–26
–24
Fig. 9.10.15. State block diagram for Y(s)/U(s) = (s + 2)/(s3 + 9s2 + 26s + 24).
⎡ 0 1 0 ⎤ ⎡ 0 ⎤ ⎡ 1⎤
A2B = ⎢ 1⎥⎥ ⎢⎢ 1⎥⎥ = ⎢⎢ − 9 ⎥⎥
⎢ 0 0
⎢⎣ − 24 − 26 − 9 ⎥⎦ ⎢⎣ − 9 ⎥⎦ ⎢⎣ 55 ⎥⎦
The controllability test matrix is given by :
U = [B : AB : A2B]
⎡0 0 1⎤
∴ ⎢
U = ⎢0 1 − 9 ⎥⎥
⎢⎣ 1 − 9 55⎥⎦
1 −9 0 −9 0 1
U = 0 −0 +1
− 9 55 1 55 1 −9
= 0 + 0 + 1 [0 × ( 9) (1) × (1)] = 1
∴ U ≠0
Thus, the rank of U is 3 and order is also 3. The system is completely controllable.
Ans.
(b) Observability test :
⎡0 0 − 24 ⎤ ⎡2⎤
AT = ⎢⎢1 0 − 26 ⎥⎥ , CT = ⎢⎢1 ⎥⎥
⎢⎣0 1 − 9 ⎥⎦ ⎢⎣0 ⎥⎦
The observability test matrix is given by :
V = [CT : ATCT : (AT)2 CT]
⎡2 0 − 24 ⎤
∴ V = ⎢⎢1 2 − 26 ⎥⎥
⎢⎣0 1 − 7 ⎥⎦
2 − 26 1 − 26 1 2
V = 2 −0 + (− 24)
1 −7 0 −7 0 1
= 2[2 × ( 7) (1) × ( 26)] + 0 [1 × ( 7) ( 26)
× (0)] 24 (1 × 1 + 2 × 0)
= 24 + 0 24 = 0
V = 0,
Therefore its rank ≠ 3, hence, the system is not observable.
490 LINEAR CONTROL SYSTEMS
–5 –4
Fig. 9.11.2. State signal flow graph : Y(s)/U(s) = s 1/[1 ( 5s 1)] . s 1/[1 ( 4s 1)].
⎡0 1⎤
⎢1 − 5⎥ ≠ 0
⎣ ⎦
Thus, the matrix A is controllable. Hence, the state variable feedback can be applied.
⎡− 4 1⎤ ⎡0 ⎤
A BK = ⎢ ⎥ − ⎢ ⎥ [k1 k2]
⎣ 0 − 5 ⎦ ⎣1 ⎦
⎡− 4 1⎤ ⎡ 0 0⎤
= ⎢ ⎥−⎢
⎣ 0 − 5⎦ ⎣ k1 k2 ⎥⎦
⎡−4 1 ⎤
= ⎢ ⎥
⎣ 1
− k (5 + k )
2 ⎦
(1) The desired characteristic equation is
(s + 1 + j2) (s + 1 j2) = 0
or s2 + 2s + 5 = 0 ...(1)
492 LINEAR CONTROL SYSTEMS
⎡ s 0⎤ ⎡ − 4 0 ⎤
⎢0 s ⎥ − ⎢ − k =0
− (5 + k2 ) ⎥⎦
∴
⎣ ⎦ ⎣ 1
s+4 −1
or =0
k1 s + (5 + k2 )
∴ (s + 4)(s + 5 + k2) ( 1) (k1) = 0
or s2 + (9 + k2)s + (20 + 4k2 + k1) = 0 ...(2)
Comparing coefficients in (1) and (2)
9 + k2 = 2
and 20 + 4k2 + k1 = 5
∴ k1 = 13
k2 = 7
K = [ k1 k2 ] = [13 7] Ans.
–k1= –13
–k2 = 7
1 1 s–1 1 s–1 1
r . . y
u x2 x2 x1 x1
–5 –4
Fig. 9.11.3. State variable feedback for Y(s)/U(s) = 1/(s + 5)(s + 4).
or Y ( s) 1
=
U ( s) s2 + 9 s + 20
. .
1 x2 s–1 x2 1 x1 s–1 x1 1
u y
–9 –20
Fig. 9.11.4. State signal flow graph : Y(s)/U(s) = s 2/[1 ( 9s 1 20s 2)].
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
∴ ⎢ x ⎥ = ⎢ ⎥ ⎢ x ⎥ + ⎢1 ⎥ u
⎣ 2⎦ ⎣ − 20 − 9 ⎦ ⎣ 2⎦ ⎣ ⎦
0 1⎤
The system matrix A = ⎡⎢ ⎥ is in controllable canonical form. Thus, the state
⎣ − 20 − 9 ⎦
variable feedback can be applied. The state equation with state variable feedback is
x = (A BK) x + Br
⎡ 0 1⎤ ⎡0 ⎤
(A BK) = ⎢ − 20 − 9 ⎥ − ⎢1 ⎥ [ k1 k2 ]
⎣ ⎦ ⎣ ⎦
⎡ 0 1⎤ ⎡ 0 0⎤
= ⎢ ⎥ − ⎢k
⎣ − 20 − 9 ⎦ ⎣ 1 k2 ⎥⎦
⎡ 0 1 ⎤
= ⎢ ⎥
⎣ − (20 + k1 ) − (9 + k2 ) ⎦
(1) The desired characteristic equation is
s2 + 2s + 5 = 0 ...(3)
(2) Since, the matrix (A BK) is in controllable canonical form the characteristic
equation (with state variable feedback can be written by inspection as
s2 + (9 + k2) s + (20 + k1) = 0 ...(4)
–k1 = 15
–k2 = 7
r 1 1 1 1 y
. .
u x1 s–1 x2 x2 s–1 x1
–9 –20
PROBLEMS
9.1. Obtain state space representation for the electrical network shown in Fig. 9-P-1.
Select the current i and voltage across the capacity vc as state variables.
1 mH 1000 W
10 V i 0.001 mf vc
9.2. Write state equations for the electrical network given below (Fig. 9P-2). Select v1, v2
as state variables and i as output variable.
1Ω V1 1Ω V2
i
e 1F 0.5 F
9.3. In Fig. 9P-3, consider i1, i2 and voltage across the capacitor as state variables, and
also i1 and i2 as output variables. Obtain the state model.
R1 L1 L2
e i1 C i2 R2
9.4. Obtain state space representation of the electrical netwrok shown in Fig. 9P-4.
C L
e i1 R1 C i2 R2
9.5. Represent the equations for the armature controlled D.C. motor (Fig. 9P-5) in state
dθ
space form. Consider ia, θ and as state variables and θ as output variable.
dt
Rf La Ra
if (const)
Lf ief b va
KT
qm, Jm, fm
9.6. A mechanical system is shown in Fig. 9P-6 choose x(t) and x (t) as state variables and
represent system equations in state space form.
x(t)
f(t) K
M
9.7. Write state space equations for the system shown in Fig. 9P-7.
f1
K1
M1
K2 y1(t)
M2
f(t) y2(t)
9.8. Obtain state model for the transfer functions given below :
C ( s) (s + 2) C(s) 1
(a) R(s) = ( s + 1)( s + 3) (b) R(s) = (s + 1)(s + 3)
Y (s) s2 + s + 2
U (s) s + 9 s2 + 26 s + 24 . Determine the state model.
= 3
496 LINEAR CONTROL SYSTEMS
9.10. For the differential equation given below, write state equations.
2
d y 4 dy du
+ + 2y = 3 +u
dt
2 dt dt
Y (s) 6s
9.11. Consider the system given by = obtain a state variable
U (s) s3 + 6s2 + 11s + 6
representation for the system.
9.12. Use parallel decomposition method to determine the state model for the transfer
function given below :
s2 + 6 s2 + 8
G(s) =
(s + 3)(s2 + 2s + 2)
9.13. A system is represented by the equations given below
⎡ x1 ⎤ ⎡ 0 ⎡ x1 ⎤ ⎡0⎤
1⎤ T
⎢ x ⎥ = ⎢ ⎢ x ⎥ + ⎢1 ⎥ u . Given x(0) = [1 1]
⎥
⎣ 2⎦ ⎣ − 4 − 5⎦
⎣ 2⎦ ⎣ ⎦
Determine (a) state transition matrix φ(t), (b) Zero input response, (c) Zero state
response for u = 1 and (d) Total response.
9.14. Use Laplace transform method to determine the time response given that :
⎡ x1 ⎤ ⎡ − 3 1 ⎤ ⎡ x1 ⎤ ⎡1 ⎤
(a) ⎢ x ⎥ = ⎢ − 4 0 ⎥ ⎢ x ⎥ + ⎢0 ⎥ u(t)
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡ x1 ⎤
x1(0) = 0 ; x2(0) = 0 and y = [1 1] ⎢ ⎥
⎣ x2 ⎦
⎡ x1 ⎤ ⎡ − 4 1 ⎤ ⎡ x1 ⎤ ⎡ 2⎤
(b) ⎢ x ⎥ = ⎢ − 3 0 ⎥ ⎢ x ⎥ + ⎢0 ⎥ u(t)
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡x ⎤
x1(0) = 0 ; x2(0) = 0 and y = [1 − 1] ⎢ 1 ⎥ .
⎣ x2 ⎦
9.15. Show that the state transition matrix (STM) for the state equations
⎡ −3t −t −3 t ⎤
⎡ x1 ⎤ e −e
⎡− 3 1⎤ ⎡ x1 ⎤ ⎢e ⎥.
⎢ x ⎥ = ⎢ ⎢x ⎥ is 2
⎣ 2⎦ ⎣ 0 − 1⎥⎦ ⎣ 2⎦
⎢ ⎥
⎢⎣ 0 −e
−t ⎥⎦
9.16. Determine the state transition matrix and the output time response, given that
⎡ x1 ⎤ ⎡ 0 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
⎢ x ⎥ = ⎢ ⎥ ⎢ x ⎥ + ⎢5 ⎥ u(t)
⎣ 2⎦ ⎣ − 2 − 3 ⎦ ⎣ 2⎦ ⎣ ⎦
x1(0) = 1; x2(0) = 0
INTRODUCTION TO STATE SPACE ANALYSIS OF CONTROL SYSTEMS 497
⎡ 0 1⎤ ⎡ x1 ⎤
and y= ⎢ ⎥ ⎢x ⎥ .
⎣ − 2 − 3 ⎦ ⎣ 2⎦
9.17. Represent the following differential equation in state variable form
2
d y dy
+3 + 2 y = u(t)
dt
2 dt
9.23. Write state equations for the state block diagram shown in Fig. 9P-8 and determine
the transfer matrix G(s).
x1
+ + x1
2
+
–2 +
x1 + 2x2 = y
u x2
4 +
+ x2 x2
+
2
+
–5
⎡ − 1 − 2 − 1⎤ ⎡2⎤
A = ⎢ 0 −1 1⎥⎥ , B = ⎢1 ⎥ , C = [1 0 1] and D = 0.
⎢ ⎢ ⎥
⎣⎢ 1 0 − 1⎦⎥ ⎣⎢1 ⎦⎥
9.27. Show that the system having following coefficients matrices :
⎡2 0 0 ⎤
A = ⎢0 2 0 ⎥ and C = [2 2 2]
⎢ ⎥
⎣⎢0 3 1 ⎥⎦
⎡1 2 1⎤
is unobservable and if C = ⎢ ⎥ observable.
⎣1 1 1⎦
10
Sampled Data Control Systems
T he analysis of sampled-data control systems is given in this chapter. The sampled-data control
system forms the basis of understanding discrete time control system. The analysis of sampled-
data control system is carried out in z-domain using z-transform. The chapter gives explanation of
sampling process and derivation of z-transform of sampled functions. The hold-circuit, pulse transfer
function, time response and stability, with regards to sampled-data control systems is dealt in this
chapter.
CONTENTS
• Sampler
• Sampling Process
• Laplace Transform of Sampled Function
• z-Transform
• z-Transform of Some Useful Functions
• Inverse z-Transform
• Hold Circuit
• Reconstruction of Signal Minimum Sampling Frequency
• Pulse Transfer Function (z-Transfer Function)
• Stability Analysis of Sampled-Data Control Systems
• Solved Examples
Introduction. The analysis of linear control systems is based on the fact that the signals at
various points in the system are continuous with respect to time. However, in some applica-
tions it is convenient to use any one or more control signals at discrete time intervals of time,
for example in some industrial process control applications the signal is available only in
sampled data form as a sequence of pulses. The control systems using one or more signals at
discrete time intervals are known as sampled data control systems.
In a sampled-data control system the signal at any one or more places is sampled and
appears in the form of a pulse train at periodic intervals. The control signal is thus ampli-
tude sampled train of pulses and the control systems employing such signals are known as
sampled data control systems.
499
500 LINEAR CONTROL SYSTEMS
Fig. 10.0.1 shows a sampled-data control system. The continuous error signal e(t) is
sampled at intervals of time T with the help of a sampler. The output of the controlled
process is continuous one with respect to time because a hold circuit preceds the controlled
process block.
e(t)
* Process
r(t) + e(t) Hold
to be
c(t)
Input T Circuit Output
– Sampler Controlled
*
e(t)
e(t)
t t
0 T 2T 3T 4T 5T 6T
Error Signal Sample Error Signal
Fig. 10.0.1. Sampled-data control system.
Digital Output
Process
Input + Digital D/A
to be
Continuous
Computer Converter
– Controlled
A/D
Converter
10.1 SAMPLER
The basic element of a sampled-data control system is a sampler which samples the continu-
ous signal into a sequence pulses appearing at regular interval of time.
Fig. 10.1.1 (a) shows a switch being used as a sampler. The input to the switch is a
continuous signal denoted f (t) as shown in Fig. 10.1.1 (b). The switch is closed for a very
short duration of time say ∆t and then remains open for some duration of time. This opera-
tion is repeated with a time period T. The time T is known as a sampling time and the switch
output appears only for the closing duration (∆t) of the switch. The signal f (t) is thus sampled
at a regular interval of time as shown in Fig. 10.1.1 (c).
f(t) T *
f(t)
(a) Sampler
f(t)
*
f(t)
t
Dt T
2T
3T
*
f(t)
t
T
2T
3T
The sampled signal is denoted as f *(t) meaning that f *(t) is obtained after sampling
the input signal f (t) at a regular interval of time.
The actuating signal for a sampled-data system is f *(t) which is in the form of a
modulated pulse train.
Ideal Sampler. The signal modulated pulse train having pulse width ∆t is shown
in Fig. 10.1.1 (c). The pulse width of an ideal sampler approaches to zero and therefore the
output f *(t) of an ideal sampler is the input signal modulated impulse train as shown in
Fig. 10.1.1 (d).
However, if the pulse width ∆t is very small as compared to the dominant time
constant of the system as well as to the sampling time T then to a fair approximation the
pulse train can be considered as an impulse train. This approximation simplifies the
analysis of sampled-data control systems.
The sampled output denoted by f *(t) is known as starred function of f (t). Fig 10.1.2
shows a function f(t) being sampled twice. It is assumed that the two samplers operate in
synchronism i.e., their sampling time and the instants of operation are same.
f(t) *
f(t) [*f(t)]* = f*(t)
T T
Sampler 1 Sampler 2
Fig. 10.1.2. Function f(t) sampled twice.
As the two samplers operate in synchronism the output of the second sampler is also
f *(t) therefore, [f *(t)]* = f *(t)
It is thus concluded that starring a starred function results in starred function itself.
f*(t)
f(t)
t 0 T 2T 3T 4T t
f(t) f*(t)
dT(t)
Input Output
dT(t)
0 T 2T 3T 4T t
The above equation gives the sampler output which is a weighted train of pulses. At
the sampling instants (t = kT) the sampler output f *(t) appears in the form of pulses as
shown in Fig. 10.2.1.
and the Laplace transform of a unit impulse train is e skT. Therefore the Laplace transform
of Eq. (10.4) is
∞
F*(s) = ∑ f (kT ) e
k=0
skT ...(10.5)
δ
T(t) = δ(t kT) is a unit impulse train wherein the unit impulse occurs at sampling instants t = kT:
where k = 0, 1, 2, 3 ... The unit impulse train is the first time derivative of sampled unit step function
u(t kT) occurring at t = kT. Therefore,
d 1 skT
LδT (t) = Lδ(t kT) = L [u(t kT)] = sLu(t kT) = s e
dt s
or LδT (t) = e skT
504 LINEAR CONTROL SYSTEMS
f *(t) is starred function and F*(s) is known as Laplace transform of starred function or
starred transform.
Consider f (t) as unit step function, therefore f (kT) = 1 and the Laplace transform of
the starred unit step function f *(t) is thus
∞
F*(1) = ∑1 .e
k=0
− skT
; where, k = 0, 1, 2, 3...
10.4 z-TRANSFORM
The Laplace transform of starred f *(t) known as impulse train Laplace transform is given by
∞
F*(s) = ∑ f (kT ) e
k= 0
skT
...(10.7)
Substituting z = esT
1
i.e. s= loge z in Eq. (10.7)
T
and considering z > 1 following relation is obtained
∞
F*(s) = ∑ f (kT ) z
k=0
k
; where k = 0, 1, 2, 3...
1
s= . loge z
T
The relation given above is known as z-transform of the function f (t) and denoted as
Z[f (t)] or F(z), therefore
∞
Z[f (t)] = F(z) = ∑ f (kT ) z
k= 0
−k
...(10.8)
where k = 0, 1, 2, 3...
In series form F(z) given by Eq. (10.8) can be written as
F(z) = f (0)z0 + f (T) z 1 + f (2T) z 2 + f (3T) z 3 + ... + f (kT) z k ...(10.9)
On comparing (10.9) and (10.6) it is noted that the values of f (t) at sampling instants
kT are given by the coefficients of z k terms in (10.9). f (0), f (T), f (2T), ... are the values of
function f (t) at instants k = 0, 1, 2... respectively.
Properties of z-Transform. Z[f (t)] or F(z) denotes z-transform of f (t). Some proper-
ties of z-transform are listed below :
SAMPLED DATA CONTROL SYSTEMS 505
1. Addition : Z[f1(t) + f2(t)] = Z[f1(t)] + Z[f2(t)] = F1(z) + F2(z)
2. Substraction : Z[f1(t) f2(t)] = Z[f1(t)] Z[f2(t)] = F1(z) F2(z)
3 Multiplication by constant : Z[a f (t)] = a Z[f (t)] = a F(z)
4. Delayed function: Z[f (t kT)] = z k Z[f (t)] = z k F(z)
d d
5. Multiplication by t : Z[t f (t)] = Tz Z[f (t)] = Tz F(z);
dz dz
where T is sampling time.
Z ⎡⎢ 1 f (t)⎤⎥ = 1 Z[ f (t)] 1 F ( z)
6. Division by t :
⎣t ⎦ T ∫ z
dz = −
T ∫ z
dz
z = esT
∞
or Z[1] =
k=0
∑1. z k
k = 0, 1, 2, 3...
In series form above expression can be written as
Z[1] = 1+ z 1 + z 2 + z 3...
506 LINEAR CONTROL SYSTEMS
Fig. 10.5.1 (a) shows continuous unit step function f (t) = u(t) and Fig. 10.5.1 (b) gives
u(t) being sampled periodically with time period T.
f(kt)
f(t) = u(t) =u(kT)
1
1
0 T 2T 3T 4T t
t
Fig. 10.5.1. Unit-step function and its sampled form.
Fig. 10.5.1 (b) can also be expressed in number
sequence form f (k) = u (k) where k being the sampling u(k)
number. The unit step sequence is shown in Fig. 10.5.2
which gives the value of the sampled unit step function at
the sampling sequence number k = 0, 1, 2, 3...
If the sampling time is T, the corresponding sam- 0 1 2 3 4
pling time at the sampling number k = 0, 1, 2, 3... is given k
by t = 0, T, 2T, 3T... respectively. Fig. 10.5.2. Unit step sequence.
The unit step sequence can also be written as, u(k)
= (1)k meaning that the unit step function u(t) being sampled periodically with period T at
the sampling number k = 0, 1, 2, 3...
The coefficients of the series
Z[1] = 1 . z0 + 1 . z 1 + 1 . z 2 + 1 . z 3 + ... 1 . z k
give the values of the unit step function at the sampling number k = 0, 1, 2, 3... respectively
as shown in Fig. 10.5.2.
As the z-transform of u(t) is (z/z 1) the inverse z-transform of (z/z 1) yields u(t) at
the sampling instants only. The inverse z-transform* is denoted as Z 1 [F(z)].
The z-transform of unit step function u(t) is (z/z 1), therefore, in number sequence
form
⎡ z ⎤
Z 1⎢ ⎥ = (1)
k
⎣ z − 1⎦
For which the corresponding continuous time function u(t) = 1.
(2) z-transform of unit ramp function f (t) = r(t) = t :
f (t) = t
t≥0
Put t = kT ∴ f (kT) = kT
k = 0, 1, 2, 3...
*The inverse z-transform is not unique as two functions may have different values between two
sampling instants but same value at the sampling instants.
SAMPLED DATA CONTROL SYSTEMS 507
∞
∵ F*(s) = ∑ f (kT ) e
k=0
skT
z = esT
In series form above expression can be written as :
Z[t] = 0Tz0 + Tz 1 + 2Tz 2 + 3Tz 3 + ...
or Z[t] = Tz 1 + 2Tz 2 + 3Tz 3 + ...
Multiplying both sides by (z 1)
(z 1) Z[t] = (z 1)(Tz 1 + 2Tz 2 + 3Tz 3 + ...)
= T(1 + z 1 + z 2 + z 3 + ...)
In summation form above expression can be written as
T zT
(z 1) Z[t] = =
1 − z −1 ( z − 1)
zT
∴ Z[t] =
( z − 1)2
⎡ zT ⎤
and Z 1 ⎢ 2⎥ = t
⎢⎣ ( z − 1) ⎥⎦
⎡ zT ⎤ 0 1 2 3 4
corresponding inverse z-transform is Z =k 1
⎢ 2⎥ k
⎢⎣ ( z − 1) ⎥⎦
Fig. 10.5.3. Sampled sequence of unit
for which the corresponding time function is f (t) = t. ramp function r(t) = t.
(3) z-transform of f (t) = e at
f (t) = e at t ≥ 0
put t = kT ∴ f (kT) = e akT
k = 0, 1, 2, 3, ...
∞
∵ F * (s) = ∑ f (kT ) e
k=0
− skT
− akT
put f (kT ) = e and z = esT,
508 LINEAR CONTROL SYSTEMS
∞
∴ F(z) = F*(s) = ∑e
k= 0
akT
z k
z = esT
∞
or Z[e at] = ∑e
k= 0
akT
z k
at
⎛ z ⎞
As the z-transform of function e is ⎜ ⎟⎟ therefore, in number sequence form
⎜ z − e− aT
⎝ ⎠
the corresponding inverse z-transform is
⎡ z ⎤
Z 1⎢ − aT ⎥
=e akT = (e aT)k
⎣⎢ z − e ⎦⎥ e–akT
Substituting c = e aT
⎡ ⎤
Z 1⎢ z = (c)k
⎣z − c ⎥⎦ 0 1 2 3 4
kT
Some commonly used z-transforms are given
in Table 10.5 Fig. 10.5.4. Sample sequence of f(t) = e at.
Table 10.5
1 δ(k) discrete 1,
unit impulse
z 1
z−1 (1)k u(t) = 1
s
zT 1
2 k r(t) = t
( z − 1) s2
z( z + 1)T 1
t2
2 (k)2 s3 f (t) =
( z − 1) 2
z
=
z 1
z − c (c)k f (t) = e at
z − e − aT s+a
zTe− aT zcT 1
= k(c)k f (t) = te at
( z − e− aT )2 ( z − c)2 (s + a)
2
z sin ωT ω
2 sin kωT f (t) = sin ωt
z − 2 z cos ωT + 1 s + ω2
2
z( z − cos ωT ) s
cos kωT 2 2 f (t) = cos ωt
2
z − 2 z cos ωT + 1 s +ω
− at
z(e sin ωT ) aT)k
( − a + jω )T
(e sin kωT
[z − e ][ z − e( − a − jω )T
]
zc sin ωT ω
= = (c)k sin kωT 2 2 f (t) = e at sin ωt
2
z − z(2c cos ωT ) + c 2 (s + a) + ω
z( z − e− aT cos ωT ) aT)k
(e cos kωT
[ z − e( − a + jω)T
][ z − e( − a − jω)T
]
z( z − c cos ωT ) s+a
= = (c)k cos kωT f (t) = e at cos ωt
2
z − z(2c cos ωT ) + c 2
(s + a)2 + ω2
510 LINEAR CONTROL SYSTEMS
Power Series Method. Function F(z) can be expressed as a ratio of two polynomials
in z as follows :
a0 zn + a1 zn − 1 + a2 zn − 2 + ... + an
F(z) = m m −1 m−2
...(10.10)
b0 z + b1 z + b2 z + ... + bm
If n ≤ m then, dividing the numerator by denominator the above function F(z) can be
written as :
F(z) = f0z0 + f1z 1 + f2z 2 + f3z 3 + ... ...(10.11)
On comparing the coefficients of the power series Eq. (10.11) and z-transform expres-
sion Eq. (10.9), it is concluded that the values of f*(t) at the sampling instants are given by
the coefficients of the power series (10.11).
However, the power series method gives the value of the corresponding function of
f *(t) at the sampling instants i.e. f (0), f (T), f (2T), f (3t), ... whereas the partial fraction
expansion method gives an expression for f *(t) in a closed form.
Relation between z-transform and Impulse Train Laplace Transform
As the z-transform is obtained by putting z = esT in the impulse train Laplace trans-
form F*(s). Therefore, F*(s) is obtained by substituting esT for z in z-transform expression.
z
For example the z-transform, F(z) = − sT
z−e
has the following impulse train Laplace transform
sT
e 1
F * (s) = sT − aT
or F * (s) = aT − sT
.
e −e 1 e e
Illustrative Example 10.6.1. Find the inverse z-transform of the function given
below, given that the sampling time T = 1 sec.
F(z) = 0.632z .
z 2 − 1.368z + 0.368
F(z) = 0.632 z
2
z − 1.368 z + 0.368
F ( z) 0.632
∴ = 2
z z − 1.368 z + 0.368
The denominator can be factored as
(z 1)(z 0.368)
F ( z) 0.632
∴ =
z ( z −1)( z − 0.368)
Expanding R.H.S. into partial fractions
F ( z) K1 K2
= +
z z − 1 z − 0.368
SAMPLED DATA CONTROL SYSTEMS 511
The coefficients can be determined as
K1 = 1 and K2 = 1
∴ F ( z) = 1 1
−
z z − 1 z − 0.368
Since e 1 = 0.368 and T = 1 sec (given)
F ( z)
∴ = 1 1 or F(z) = z
−
z
z z−1 z − e −T z − 1 z − e− T
From z-transform table
⎡ z ⎤ ⎡ z ⎤
Z 1⎢ ⎥ =1 and Z 1 ⎢ −T ⎥ = (e )*
t
⎣ z − 1⎦ ⎣⎢ z − e ⎥⎦
∴ f *(t) = (1 e t)*
Substituting t = kT; k = 0, 1, 2, 3...
∞
f *(t) can be written as : f (kT) = ∑ (1 − e
k=0
− kt
) δ(t kT) Ans.
0.949z 1 0.317z 2
0.949z 1 ∓ 1.298z 2 ± 0.349z 3
0.981z 2 0.349z 3
0.981z 2 ∓ 1.342z 3 ± 0.361z 4
0.993z 3 0.361z 4
512 LINEAR CONTROL SYSTEMS
0.632 z
Hence F(z) = 2
z − 1.368 z + 0.368
= 0z0 + 0.632z 1 + 0.864z 2 + 0.949z 3 + 0.981z 4 + 0.993z 5 + ...
The first few terms of the sampled function f *(t) at the sampling instants k = 0, 1, 2, 3,
4, 5 are given by the F(z) series therefore,
f (0) = 0, f (1) = 0.632, f (2) = 0.864, f (3) = 0.949, f(4) = 0.981,
f (5) = 0.993
It is noted that the sampled sequence i.e. f *(t) can be obtained by two methods men-
tioned above. However, the power series method requires long division to get more terms
whereas number sequence approach using z-transform table appears to be more convenient.
Moreover, if it is not convenient to factorize the denominator of function F(z) then the power
series can be obtained by long division to get the value of f *(t) at the sampling instants.
The graphical representation of the sampled sequence f (k) = (1)k (0.368)k is shown in
Fig. 10.6.1.
f(k)
0.632
0.864
0.949
0.981
0.993
0 1 2 3 4 5 k
Fig. 10.6.1. f(k) = (1)k (0.368)k.
The value of f (k) as k approaches higher values is obtained by using final value
theorem as follows:
0.632 z
f (∞) = lim (1 z 1) F(z) = lim (1 z 1)
z→1 z→1 ( z − 1)( z − 0.368)
⎛ z − 1⎞ ⎡ 0.632 z ⎤ 0.632 0.632
= lim ⎜ ⎟ ⎢ ( z − 1)( z − 0.368) ⎥ = zlim =
z→1 ⎝ z ⎠ ⎣ ⎦ → 1 ( z 0.368) 0.632
or f (∞) = 1
Difference equations
The analysis of sampled-data control system can be carried out in terms of difference
equations because the input to the system is in the form of a signal occurring at regular
intervals (sampling instants) of time. The analytical results obtained by this method are
equivalent to those obtained by z-transform analysis, because in the latter case, the values of
the function only at the sampling instants are considered.
In terms of z-transform, the output function C(z) is given by
G( z)
C(z) = R(z)
1 + G( z) H ( z)
The pulse transfer function G(z)/[1 + G(z) H(z)] can be expressed as a power series in
inverse powers of z.
SAMPLED DATA CONTROL SYSTEMS 513
⎛ a1 a2 ⎞
∴ C(z) = ⎜ a0 + z + 2 + ...... ⎟ R(z)
⎝ z ⎠
or 1 2
C(z) = (a0 + a1z + a2z + ......) R(z)
The time function at the sampling instants of the above equation can be written as
follows :
c(kT) = r(kT) + a1r(kT 1) + r(kT 2) + ....
In number sequence form
c(k) = a0r(k) + a1r(k 1) + a2r(k 2) + ...
As per above equation the output of a second order sampled-data control system at
any given sampling instant is given by the input at the same instant, at one sampling instant
earlier, at two sampling instant earlier.
∴ c(k + 2) + a1c(k + 1) + a0c(k) = b0r(k) + b1r(k + 1)
is the difference equation representation of a second order sampled-data control system.
The difference equation representation of a third order sampled-data control system
is given by
c(k + 3) + a2c(k + 2) + a1c(k + 1) + a0c(k)
= b0r(k) + b1r(k + 1) + b2r(k + 2)
The z-transform is related to difference equation in a similar manner as the Laplace
transform to differential equation. The Laplace transform converts linear differential equa-
tion with constant coefficients into algebraic equation in terms of s . Similarly, z-transform
converts difference equation with constant coefficients into algebraic equation in terms of
z.
z-transform of x(k) is denoted as X(z).
z-transform of x(k + 1) is determined below :
∞ ∞
Z[x(k + 1)] = ∑ x(k + 1) z
k=0
k = ∑ x(k) z
k =1
k+1
⎡ ∞ ⎤
=z ⎢
⎣⎢
∑ x(k)z
0
−1
− x(0) ⎥
⎦⎥
∞
∵ ∑ x(k) z
0
−1
= x(z)
Solution.
c(k + 2) = 5c(k + 1) + 6c(k) + u(k)
put k=0
c(0 + 2) = 5c(0 + 1) 6c(0) + u(0)
or c(2) = 5c(1) 6 c(0) + u(0) = 5 × 1 6×0+1=6
put k=1
c(1 + 2) = 5c(1 + 1) 6c(1) + u(1)
or c(3) = 5c(2) 6c(1) + u(1) = 5 × 6 6 × 1 + 1 = 25
put k=2
c(2 + 2) = 5c(2 + 1) 6c(2) + u(2)
or c(4) = 5c(3) 6c(2) + u(2) = 5 × 25 6 × 6 + 1 = 90
k = 0, 1, 2, 3, 4 ...
∴ c(k) = 0, 1, 6, 25, 90, ... Ans.
The difference equation given above is solved below by z-transform method.
c(k + 2) 5c (k + 1) + 6 c(k) = u(k)
Taking z-transform on both sides,
Z[c (k + 2) 5Z [c (k + 1)] + 6Z [c (k)] = Z [u (k)]
z
∴ [z2 C(z) z2c(0) zc(1)] 5 [zC(z) zc (0)] + 6C(z) =
z−1
given that c(0) = 0 and c (1) = 1
z
∴ [z2 C (z) z2 × 0 z × 1] 5 [zC(z) z × 0] + 6C(z) =
z−1
z2
Rearranging, C(z) =
( z − 1) ( z2 − 5 z + 6)
C( z) z K1 K2 K3
or = = + +
z ( z − 1) ( z − 2) ( z − 3) z −1 z − 2 z −3
1 3
The coefficients K1, K2 and K3 can be determined as K1 = , K2 = 2 and K3 =
2 2
C ( z) 1 1 1 3 1
∴ = ⋅ − 2. + .
z 2 z−1 z−2 2 z−3
z 2z 3z
or C (z) = − +
2 ( z − 1) ( z − 2) 2 ( z − 3)
Taking inverse z-transform on both sides
c(k) = 1 (1)k − 2 (2) k + 3 (3) k
2 2
SAMPLED DATA CONTROL SYSTEMS 515
for k = 0, 1, 2, 3, 4 ...
∴
c(k) = 0, 1, 6, 25, 90 ... Ans.
It is noted that the initial values given in the problem automatically appear in the
generated sequence.
C*(t)
f*(t)
f(t)
O t 0 T 2T 3T 4T 0 T 2T 3T 4T
kT kT
Fig. 10.7.1. Sampled-data control system : Input and output.
The original signal is reconstructed from the sampled signal by using a hold circuit.
The hold circuit brings smoothness in the sampled output. At the sampling instants the hold
signal and the original signal have the same value. The use of hold circuit enables to hold the
signal between two consecutive sampling instants at the preceded value till the next
sampling instant is reached.
Zero-order Hold. The hold circuit mentioned above holds the output signal at a fixed
level between two consecutive sampling instants such that the slope of the hold circuit out-
put signal is zero or in other words hold signal is the zeroth derivative of an impulse. Such a
holding device is called zero-order hold (ZOH). The ZOH device is also known as box-car
generator. The input and output of a zero-order hold are shown in Fig. 10.7.2.
Transfer Function of Zero-order Hold Circuit. Fig. 10.7.3 shows a unit impulse
input given to a zero-order hold circuit which holds the input signal for a duration T and
therefore, the output appears to be a unit step function till duration T.
As the output of ZOH is a unit step function appearing upto time T, the same can be
written as
1
h(t) = u(t) u(t T) ∵ Lu(t) =
s
1 − sT 1 1
and Lu(t T) =
⋅e ∴ Lh(t) = − ⋅ e− sT
s s s
therefore, the Laplace transform of the output of ZOH is
1 1 − sT
H(s) = − ⋅e
s s
516 LINEAR CONTROL SYSTEMS
f(t)
fh(t)
f(kt)
0 t
0 T 2T 3T 4T 5T 6T kT
(c) Sampled signal (d) ZOH output
Fig. 10.7.2. Hold circuit.
h(t)
d(t)
1 1
O T O T
t t
Input Output
ZOH
1 − e − j ωT
∴ Gh0(jω) =
jω
which can also be written as :
(1 − e− jωT ) 2e− jωT / 2 (e jωT / 2 − e− jωT / 2 )
Gh0(jω) = 2e jωT / 2 .e− jωT / 2 = ⋅
j2 ω ω j2
2 e − j ωT / 2 ⎛ ωT ⎞ Te
− j ωT / 2
⎛ ωT ⎞
= sin ⎜ ⎟= sin ⎜ ⎟
ω ⎝ 2 ⎠ ω T ⎝ 2 ⎠
2
T sin (ωT / 2) − jωT / 2
or Gh0(jω) = ⋅e
ωT / 2
If the sampling frequency is ωs, then the sampling time T is given by
2π
T=
ωs
⎛ ω × 2π ⎞
T sin ⎜ ⎟
⎝ 2ωs ⎠ − j ⎛ ω × 2π ⎞
∴ Gh0(jω) = ⋅e ⎜ ⎟
ω × 2π ⎝ 2 ωs ⎠
2 ωs
⎛ πω ⎞ ⎛ πω ⎞
T sin ⎜ ⎟ ⎛ πω ⎞ sin ⎜ ⎟
⎝ ωs ⎠ j
− ⎜ω ⎟ ⎝ ωs ⎠
or Gh0(jω) = ⋅ e ⎝ s ⎠ which gives Gh0(jω) = T πω
πω
ωs ωs
⎛ πω ⎞
and ∠Gh0(jω) = ⎜ ⎟+θ
⎝ ωs ⎠
⎛ πω ⎞ ⎛ πω ⎞
where θ = 0, for sin ⎜ ⎟ > 0, and θ = π, for sin ⎜ ⎟ <0
⎝ ωs ⎠ ⎝ ωs ⎠
In view of above derivations gain-frequency and phase frequency graphs for ZOH
circuit are shown in Fig. 10.7.4.
T 0 ws 2ws 3ws w
|Gho(jw)|
ÐGho(jw)
–90°
F*(jw)
F*(jw) ws < 2w1
ws > 2w1
– ws ws
2 2 – ws ws w1
2 2
Fig. 10.8.2. Sampler output frequency Fig. 10.8.3. Sampler output frequency
spectrum : ωs > 2ω1. spectrum : ωs < 2ω1.
A comparison of Figs. 10.8.2 and 10.8.3 shows that for the case ωs > 2ω1, the primary
(original) frequency components of the input signal are isolated from the complementary
frequency components in the sampler output and thus the original signal is reconstructed.
From Fig. 10.7.4, it is noted that the zero-order hold circuit has a characteristic of low-pass
filter and therefore, if a ZOH circuit is placed after the sampler, the high frequency (comple-
mentary) components are blocked and at the output of ZOH the original signal appears.
In case, if ωs < 2ω1, then in addition to primary frequency components, the sampler
output contains complementary frequency components also and these two components
overlap each other in the original frequency band from ω1 to ω1. In this case the sampler
output amplitude does not match with the original input signal and thus the output appears
distorted. Therefore, if ZOH circuit is placed after the sampler, then ZOH will not
reconstruct the original signal as desired.
In view of the above, if the original signal is to be recovered from the sampler then it is
to be ensured that the sampling frequency should be greater than twice the highest
frequency contained in the input signal i.e. ωs ≥ 2ω1.
The above result is known as sampling theorem.
SAMPLED DATA CONTROL SYSTEMS 519
Fig. 10.9.1. Sampled-data control system : Fig. 10.9.2. Sampled-data control system :
Block diagram in time domain. Block diagram in s-domain.
From Fig. 10.9.3 it is noted that, the input to the Fig. 10.9.3. Input sampling.
transfer function G(s) is sampled i.e. R*(s). Therefore,
the output is considered sampled one and thus
[C(s)]* = [G(s) R*(s)]*
or C*(s) = G*(s) [R*(s)]* ...(10.12)
As starring a starred function results in starred function itself, therefore
C*(s) = G* (s) R*(s) ...(10.13)
In z-transform form the equation (10.13) can be written as
C(z) = G(z) R(z)
C( z)
or = G(z) ...(10.14)
R( z)
The pulse transfer function (z-transfer function) of the sampled-data control system
shown in Fig. 10.9.3 is G(z).
The z-domain block diagram is shown in Fig. 10.9.4.
S1 S2
R(z) C(z) R(s) R* (s) C1(s) C1*(s) C(s)
G(z) G1(s) G1(s)
T T
Fig. 10.9.4. Sampled-data control system : Fig. 10.9.5. Two blocks in cascade each using input sampler.
Block diagram in z-domain.
S1
R(z) C(z) R(s) R* (s) C(s)
G1(z) G2(z) G1(s) G2(s)
T
Fig. 10.9.6. Block diagram representation Fig. 10.9.7. Two blocks in cascade using
of pulse transfer function (10.19). sampler before first block.
(b) The blocks in cascade using sampler in the input of first block :
Fig. 10.9.7 shows two blocks in cascade such that the input given to the first block is
sampled.
As per block diagram reduction procedure R(s) R* (s) C(s)
G1(s)G2(s)
Fig. 10.9.7. is redrawn as shown in Fig. 10.9.8. T
It is noted from Fig. 10.9.8 that
Fig. 10.9.8. Fig. 10.9.7 redrawn.
C(s) = G1(s) G2(s) R*(s)
...(10.20)
starring Eq. (10.20), C*(s) = [G1(s) G2(s) R*(s)]* = G1 G2* (s) [R*(s)]*
or C*(s) = G1 G2* (s) R*(s) ...(10.21)
The terms G1 G2* (s) means that the starring is done after combining G1(s) and G2(s) as
per block diagram reduction rules.
In terms of z-transform Eq. (10.21) can be written as
C(z) = G1G2(z) R(z) ...(10.22)
SAMPLED DATA CONTROL SYSTEMS 521
The pulse transfer function is
C( z)
= G1G2(z) ...(10.23)
R( z)
The block diagram representation of Eq. (10.23) is shown in Fig. 10.9.9.
R(z) C(z)
G1G2(z)
Illustrative Example. 10.9.1 Determine the pulse transfer function for the system
represented by the block diagram shown in Figs. 10.9.10 (i) and 10.9.10 (ii).
R(s) *
R(s) C(s)
1 1
G1(s) = G2(s) =
T s+1 s+2
(i)
R(s) *
R(s) C(s)
1 1
G1(s) = G2(s) =
T s+1 s+2
T
(ii)
Fig. 10.9.10. Illustrative Example.
1 1 1 1
Solution. (i) G1G2(s) = ⋅ = −
s +1 s + 2 s +1 s +2
In terms of z-transform
z z z(e−T − e−2T )
G1G2(z) = − =
z − e− T z − e−2T ( z − e−T ) ( z − e−2T )
The pulse transfer function is
C( z)
= G1G2(z)
R( z)
C( z) z(e−T − e−2T )
or = ...(10.24)
R( z) ( z − e−T ) ( z − e−2T )
(ii) In terms of z-transform
z z
G1(z) = and G2(z) =
z−e −T
z − e−2T
The pulse transfer function is given by
C( z) z2
= G1(z) G2(z) =
R( z) ( z − e−T ) ( z − e−2T )
C( z) z2
or = ...(10.25)
R( z) ( z − e ) ( z − e−2T )
− T
522 LINEAR CONTROL SYSTEMS
Illustrative Example 10.9.2. Determine the z-transform of the output for the
sampled-data system shown in Fig. 10.9.13 consider input function to be a unit step.
R(s) C(s) *
C(s)
1
Input s+1 T Output
C(z) = z z z( z − e−T ) − z( z − 1)
− −T
=
z−1 z− e ( z − 1) ( z − e− T )
or C(z) = z( z − e− T ) Ans.
( z − 1) ( z − e − T )
SAMPLED DATA CONTROL SYSTEMS 523
R* (s)
∴ E*(s) = ...(10.32)
1 + GH * (s)
Starring Eq. (10.29)
C*(s) = [G(s) E*(s)]* = G* (s) [E*(s)]*
or C*(s) = G*(s) E*(s) ...(10.33)
Substituting Eq. (10.32) in (10.33)
R* (s)
C*(s) = G*(s) .
1 + GH * (s)
C * ( s) G* (s)
or = ...(10.34)
R* ( s) 1 + GH * (s)
In terms of z-transform relation (10.34) can be written as the required pulse transfer
function
C( z) G( z)
= ...(10.35)
R( z) 1 + GH ( z)
The block diagram representation of the pulse transfer function (10.35) is shown in
Fig. 10.9.15.
(b) Error and feedback sampling. Fig. 10.9.16 shows a closed-loop control system
wherein the error signal as well as feedback signal being sampled such that the samplers are
synchronized.
*
R(s) + E(s) E(s)
G(s)
C(s)
T
R(z) G(z) C(z) –
1 + GH(z) *
C(s)
H(s)
T
Fig. 10.9.15. Block diagram in z-domain : Fig. 10.9.16. Error and feedback sampling.
Error sampling.
524 LINEAR CONTROL SYSTEMS
R(s) *
R(s) E(s) C(s)
G(s)
R(z) G(z) C(z) T + –
1 + GH(z) H(z)
H(s)
Fig. 10.9.17. Block diagram in z-domain : Fig. 10.9.18. Closed-loop system : Input sampling.
Error and feedback sampling.
(c) Closed-loop system with input sampling. Fig. 10.9.18 shows a closed-loop control
system wherein the input signal R(s) is sampled.
From Fig. 10.9.18 it is noted that
C(s) = E(s) G(s) ...(10.43)
and *
E(s) = R (s) C(s) H(s) ...(10.44)
Substituting for E(s) from Eq. (10.44) in Eq. (10.43)
C(s) = [R*(s) C(s) H(s)] G(s) = R*(s) G(s) C(s) H(s) G(s)
or C(s) [1 + G(s) H(s)] = G(s) R*(s)
SAMPLED DATA CONTROL SYSTEMS 525
G(s)
∴ C(s) = R*(s) ...(10.45)
1 + G(s) H (s)
Starring Eq. (10.45)
*
C*(s) = ⎡⎢ G( s) ⎤ ⎡ G
R* (s)⎥ = ⎢
⎤ *
(s)* ⎥ [R (s)]
*
⎣ 1 + G ( s) H ( s) ⎦ ⎣ 1 + GH ⎦
⎡ G ⎤
= ⎢ (s)* ⎥ R*(s) ...(10.46)
⎣ 1 + GH ⎦
The term ⎡⎢ G (s)* ⎤⎥ means that the starring is done after combining G(s) and H(s)
⎣ 1 + GH ⎦
as per block diagram reduction rules.
In terms of z-transform the relation (10.46) can be written as
⎡ G ⎤
C(z) = ⎢ ( z) ⎥ R(z)
⎣ 1 + GH ⎦
and thus the required pulse transfer function is
C( z) G
= ( z) ...(10.47)
R( z) 1 + GH
The block diagram representation of the pulse transfer function is shown in Fig.
10.9.19.
R(z) G C(z)
(z)
1 + GH
Im
R(s) E(s) *
E(s) C(s)
G(s) STABLE UNSTABLE Re
+ – T REGION REGION
H(s)
Fig. 10.10.1. Sampled data feedback control Fig. 10.10.2. Stability region in s-plane.
system with error sampling.
The overall transfer function given by Eq. (10.48) can be expressed in z-transform as
C( z) G( z)
= ...(10.50)
R( z) 1 + GH ( z)
The characteristic equation of z-transfer function (10.50) is
1 + GH(z) = 0 ...(10.51)
In terms of z-transform form of the characteristic equation i.e. 1 + GH(z) = 0, the
stability region in z-plane corresponding to s-plane is located by mapping from s-plane to
z-plane.
Mapping from s-plane to z-plane
As the z-transform is obtained by substituting z = esT in the corresponding impulse
train Laplace transform, therefore the relation between the variables s and z is
z = esT
With reference to location of the roots of the characteristic equation is s-domain the
imaginary (jω) axis in the s-plane divides stable and unstable regions and the corresponding
regions in z-domain can be obtained by putting s = ± jω in Eq. (10.52) and plotting the values
of z thus obtained in another complex plane called z-plane. Therefore,
z = e± jωT ...(10.53)
or z = (cos ωT ± j sin ωT) ...(10.54)
From the Eq. (10.54) the magnitude
z =1 ...(10.55)
and phase angle ∠z = ± ωT ...(10.56)
In view of the Eqs. (10.55) and (10.56) the corresponding variation of the dependent
variable z in the z-plane as the independent variable ω varied along the imaginary axis in
s-plane is given by a circle of unit radius centred at the origin of z-plane shown in Fig.
10.10.3.
The L.H.S. of s-plane is mapped as the inside and R.H.S. outside of the unit circle in
z-plane which is varified below :
(a) Let s = α ± jω ; a point in the L.H.S. of s-plane. The corresponding point in z-plane
is given by
z = e( α ± jω)T = e αT (cos ωT ± j sin ωT)
∴ z = e αT
and ∠z = ± ωT
SAMPLED DATA CONTROL SYSTEMS 527
Unstable
Im
Stable
Stable Unstable
Re Re
region region
Region
Region
s-plane z-plane
Fig. 10.10.3. Mapping from s-plane to z-plane.
As α the real part of the point under consideration lies in the L.H.S. s-plane and T
being positive
∴ z <1
Hence, the point ( α ± jω) with negative real part located in s-plane lies inside the
unit circle when mapped into z-plane.
(b) Let s = α ± jω ; a point in the R.H.S. of s-plane.
The corresponding point in z-plane is given by
z = e(α ± jω)T = eαT (cos ωT ± j sin ωT)
∴ z = eαT
and ∠z = ± ωT
As α the real part of the point under consideration lies in the R.H.S. of s-plane and T
being positive
∴ z >1
Hence, the point (α ± jω) with positive real part located in s-plane lies outside the unit
circle mapped into z-plane.
In view of the above analysis it is concluded that for the sampled data system to be
stable all the roots of the z-transformed characteristic equation 1 + GH(z) = 0 should be
located inside the unit circle centred at the origin in z-plane and in case, any root is located
outside the unit circle centred at the origin in z-plane makes the system unstable.
If the roots of the characteristic equation 1 + GH(z) = 0 are located inside the unit
circle centred at the origin in z-plane then the time response of a sampled data system
decays with the sampling number k = 0, 1, 2, 3, ... indicating stability while any root located
outside the unit circle centred at the origin in z-plane leads to growing response with k = 0, 1,
2, 3 ... and thus the system is unstable.
If the characteristic equation is of higher degree then it is not simple to determine the
roots. However, the roots of the characteristic equation 1 + GH(z) = 0 which are located
inside the unit circle centered at the origin in z-plane can be identified without factoring the
characteristic equation. This is done by changing the variable from z to w . This change of
variable maps the interior of the z-plane unit circle into left-half of complex w-plane while
the exterior of the z-plane unit circle is mapped into right-half of w-plane and therefore the
Routh-Hurwitz criterion can be applied to determine the stability. The change of variable
from z to w is known as bilinear transformation.
The example given below illustrates the determination of the stability by modifying
Routh-Hurwitz test to suit sampled data control system.
528 LINEAR CONTROL SYSTEMS
Illustrative Example 10.10.1. Determine the pulse transfer function and stability
of the sampled-data control system shown in Fig. 10.10.4 for sampling time (a) T = 0.5 sec.
(b) T = 1 sec.
R(s) E(s) *
E(s) 5 C(s)
+ –
s(s + 0.5)
5 K K2
Solution. G(s) = = 1 +
s (s + 0.5) s s + 0.5
The coefficients can be determined as
K1 = 10 and K2 = 10
10 10
∴ G(s) = −
s s + 0.5
⎡ z z ⎤ 10 z (1 − e−0.5T )
Taking z-transform, G(z) = 10 ⎢ ( z − 1) − ⎥ =
⎣ ( z − e−0.5T ) ⎦ z2 − z (1 − e−0.5T ) + e−0.5T
5
Since, G(s) = and H(s) = 1
s(s + 0.5)
5 5 10 10
∴ G(s) H(s) = ×1= = −
s(s + 0.5) s(s + 0.5) s s + 0.5
⎡ z z ⎤ 10 z (1 − e−0.5T )
Taking z-transform, GH(z) = 10 ⎢ − −0.5T ⎥ =
⎣ ( z − 1) ( z − e ) ⎦ z2 − z (1 + e−0.5T ) + e −0.5T
The pulse transfer function for the sampled-data control system shown in Fig. 10.10.4
is given by
C( z) G( z)
=
R( z) 1 + GH ( z)
Substituting for the values of G(z) and GH(z) determined above, the pulse transfer
function is
10 z (1 − e−0.5T )
C( z) z − z (1 + e−0.5T ) + e−0.5T
2
=
R( z) 10 z (1 − e−0.5T )
1+
z2 − z (1 + e−0.5T ) + e−0.5T
C( z) 10 z (1 − e−0.5T )
or = 2 Ans.
R( z) z − z (11 e−0.5T − 9) + e−0.5T
The characteristic equation is, z2 z (11 e 0.5T 9) + e 0.5T =0
SAMPLED DATA CONTROL SYSTEMS 529
(a) As the sampling time is T = 0.5 sec the characteristic equation becomes
z2 z (11 e 0.5 × 0.5 9) + e 0.5 × 0.5 = 0
or z2 + 0.42z + 0.78 = 0
Applying bilinear transformation from z-plane to w-plane i.e. putting z = (1 + w)/
(1 w) gives
2
⎛1 + w⎞ ⎛1 + w⎞
⎜ ⎟ + 0.42 ⎜ ⎟ + 0.78 = 0
⎝1 − w⎠ ⎝1 − w⎠
or 1.36w2 + 0.44w + 2.2 = 0
The Routh array for the above equation is formed below
w2 1.36 2.2
w1 0.44 0
w0 2.2
There is no change of sign in the first column of the Routh array, hence, all the roots of
the characteristic equation lie within the unit circle centred at the origin of z-plane, there-
fore, the system is stable. Ans.
(b) As the sampling time T = 1 sec the characteristic equation is
z2 z (11 e 0.5 × 1 9) + e 0.5 × 1 = 0
or z2 + 2.33z + 0.606 = 0
Applying bilinear transformation from z-plane w-plane i.e. putting z = (1 + w)/(1 w)
gives
⎛1 + w⎞ ⎛1 + w⎞
⎜ ⎟ + 2.33 ⎜ ⎟ + 0.606 = 0
⎝1 − w⎠ ⎝1 − w⎠
or 0.724w2 + 0.788w + 3.93 = 0
The Routh array for the above equation is formed below
w2 0.724 3.93
w1 0.788 0
w0 3.93
There is one change of sign in the first column of the Routh array, hence, one root of
the characteristic equation is located outside the unit circle in z-plane, therefore, the system
is unstable. Ans.
The corresponding continuous-time system given by the above illustative example is
of second order and as such absolutely stable. Whereas, if the system is made sampled data,
it becomes unstable for the sampling time T = 1 sec. However, decreasing the sampling time
to T = 0.5 sec, the system is table.
Thus, it is concluded that a sampled-data control system behaves more like a continu-
ous time system when the sampling time is reduced (sampling frequency increased).
Jury s stability test :
The identification of the roots of a characteristic polynomial F(z) = 1 + G(z) H(z) within
the unit circle for determining system stability of a sampled-data control system is done by
Jury s stability criterion. This criterion is an algebraic one. Consider the following nth order
polynomial in z
530 LINEAR CONTROL SYSTEMS
4 bn 1 bn 2 bn 3 ... bk 1 ... b1 b0
5 c0 c1 c2 ... cn k ... cn 2
6 cn 2 cn 3 bn 4 ... ck 2 ... c0
2n 6 p0 p1 p2 p3
2n 5 p3 p2 p1 p0
2n 4 q0 q1 q2
2n 3 q2 q1 q0
In the array mentioned above, the elements of 2nd, 4th, 6th ...etc. upto (2n 4) rows
are respectively the elements of 1st, 3rd, 5th...etc. upto (2n 5) rows in reverse order. The
elements of 3rd, 5th ... etc. upto (2n 3) rows are determined by following second order
determinants.
a0 an − k
bk = for 3rd row, k = 0, 1, ... (n 1)
an ak
b0 bn − 1 − k
ck = for 5th row, k = 0, 1, ... (n 2)
bn − 1 bk
c0 cn − 2 − k
dk = for 7th row, k = 0, 1, ... (n 3)
cn − 2 ck
SAMPLED DATA CONTROL SYSTEMS 531
p3 p0
q2 =
p0 p3
p3 p1
q1 = for (2n 3) row
p0 p2
p3 p2
q0 = p0 p1
Sufficient conditions for the roots of the characteristic polynomial F (z) to lie within a
unit circle in z-plane are given by
a0 < a n
b 0 > bn 1
c0 > cn 2
p 3 > p0
p2 > q0 with (n 1) constraints
Row z0 z1 z2 z3 z4
1 1 5 12 8 2
2 2 8 12 5 1
3 b0 b1 b2 b3
4 b3 b2 b1 b0
5 c0 c1 c2
n = 4, a0 = 1, a1 = 2, a2 = 12, a3 = 8 and a4 = 2
a0 an − k
bk =
an ak
a0 a4 1 2
k = 0; b0 = = = 1×1 2× 2= 3
a4 a0 2 1
a0 a3 1 8
k = 1; b1 = = = 1×5 2× 8= 11
a4 a1 2 5
532 LINEAR CONTROL SYSTEMS
a0 a2 1 8
k = 2; b2 = = = 1×8 2× 8= 8
a4 a2 2 8
a0 a1 1 5
k = 3; b3 = = = 1×8 2× 5= 2
a4 a3 2 8
b0 bn − 1 − k
ck =
bn − 1 bk
b0 b4 − 1 − 0 b0 b3 −3 −2
k = 0; c0 = = =
b4 − 1 b0 b3 b0 −2 −3
= 3× 3 ( 2× 2) = 5
b0 b4 − 1 − 1 b0 b2 −3 −8
k = 1; c1 = = =
b4 − 1 b1 b3 b1 − 2 − 11
=( 3× 11) ( 2× 8) = 17
b0 b4 − 1 − 2 b0 b1 − 3 − 11
k = 2; c2 = = =
b4 − 1 b2 b3 b2 −2 −8
= ( 3 × 8) ( 2 × 11) = 2
Sufficient conditions for stability are
a0 < a4 or 1 < 2 satisfied
b0 > b 3 or 3 > 2 satisfied
c0 > c2 or 5 > 2 satisfied
The required necessary and sufficient conditions are satisfied, hence, the system is
stable. Ans.
Example 10.11.1. Find the impulse train Laplace transform (starred transform)
F* (s) for the following :
at z
(a) f (t) = u(t) (b) f (t) = e (c) F(z) =
z − e − aT
∞
= ∑ 1. e − skT
f*(t) = 1
k=0
–akT
k=0
f*(t) = e
∞
= ∑e
k=0
− akT
e− skT
z
From z-transform table Z[u(t)] =
z −1
z
and Z[e 2t] =
z − e−2T
2t] z z z2 − ze−2T − z2 + z
∴ Z[u(t) e = − =
z − 1 z − e−2T ( z − 1) ( z − e−2T )
534 LINEAR CONTROL SYSTEMS
z (1 − e−2T )
= Ans.
( z − 1) ( z − e−2T )
(b) f (t) = (1 e 2T)
z 5t]
z
From z-transform table, Z[1] = and Z[e =
z −1 z − e−5T
z
5t]
z
∴ Z[1 e −=
z − 1 z − e−5T
as the sampling time T = 0.2 sec.
5t]
z z z z
∴ Z[1 e = − −5 × 0.2
= −
z −1 z − e z − 1 z − 0.368
z ( z − 0.368) − z ( z − 1) 0.632 z
= = 2 Ans.
( z − 1) ( z − 0.368) z − 1.368 z + 0.368
∞
∴ Z[eat] = ∑e
k=0
akT
z− k = 1 + e+ aT z 1 + e + 2aT z 2 + e + 3aT z 3 + ...
∞
∴ Z[f (t)] = ∑ sin kωT. z
k=0
−k
e jωkT − e− jωkT
∵ sin kωT =
2j
∴ Z[f (t)] = Z [sin ωt]
1 ⎡ ⎤
∞ ∞
= ⎢ ∑
2 j ⎢k = 0
e jωkT z− k − ∑ e− jωkT z− k ⎥
⎥⎦
⎣ k=0
SAMPLED DATA CONTROL SYSTEMS 535
The summation of the above series is
1 ⎡ 1 1 ⎤ 1 ⎡ z z ⎤
Z[sin ωt] = ⎢ − ⎥= −
2 j ⎣⎢ 1 − e jωT z−1 1 − e− jωT z−1 ⎦⎥ 2 j ⎢⎣ z − e jωT z − e− jωT ⎥⎦
e j ωT − e − j ω T 1
= z. ⋅
2j ⎛e j ωT
+ e − j ωT ⎞
z2 − 2 z ⎜⎜ ⎟⎟ + 1
⎝ 2 ⎠
z sin ωT
= 2
Ans.
z − 2 z cos ωT + 1
(c) f (t) = e αt cos ωT ∴ f (kT) = e αkT cos ωkT
∞
∵ Z[f (t)] = ∑ f (kT )z
k=0
−k
; k = 0, 1, 2, 3...
e jωkT + e− jωkT
∵ cos kωT =
2
1⎡ ⎤
∞ ∞
∴ Z[e αkT cos ωkT] = ⎢ ∑
2 ⎢k = 0
e−αkT e jωkT z− k + ∑e −αkT
e− jωkT z− k ⎥
⎥⎦
⎣ k=0
1⎡ 1 1 ⎤
Z[e αt cos ωt] = 2 ⎢ −αT jωT −1
+ −αT − jωT −1 ⎥
⎣1 − e e z 1−e e z ⎦
z ( z − e−αT cos ωT )
= Ans.
z2 − 2 ze−αT cos ωT + e−2αT
536 LINEAR CONTROL SYSTEMS
1
Example 10.11.4. Determine z-transform of the function F(s) = 2
and
s + 2s + 2
assume sampling time T = 1 sec.
1 1
Solution. F(s) = 2
=
s + 2s + 2 [ s + (1 − j1)][ s + (1 + j1)]
K1 K2
= +
s + (1 − j1) s + (1 + j1)
The coefficients can be determined as
1 1
K1 = and K2 =
j2 j2
1 ⎡ 1 1 ⎤
∴ F(s) = ⎢ − ⎥
j 2 ⎣ s + (1 − j1) s + (1 + j1) ⎦
Taking z-transform
1 ⎡ z 1 ⎤
F(z) = ⎢ − (1 + j1)T ⎥
− (1 − j1)T
−
j2 ⎣ z − e z−e ⎦
put T = 1 sec (given)
1 z [ z − e (1+ j1) ] − z [ z − e (1− j1) ] 1 z2 − ze (1+ j1) z2 + ze− (1− j1)
= ⋅ = ⋅
j2 [ z − e (1− j1) ][ z − e− (1+ j1) ] j2 [ z − e− (1+ j1) ][ z − e− (1+ j1) ]
3 2 e 0.2 s 3 ⎛K K2 ⎞
Solution. F(s) = + = 2 + 2e−0.2 s ⎜ 1 +
s2 + 9 s (s + 2) s + (3) 2
⎝ s s + 2 ⎟⎠
The coefficients K1 and K2 can be determined as
1 1
K1 = and K2 =
2 2
SAMPLED DATA CONTROL SYSTEMS 537
3 ⎡1 1 1 1 ⎤
∴ F(s) = + 2e−0.2 s ⎢ ⋅ − ⋅ ⎥
s2 + (3)2 ⎣ 2 s 2 ( s + 2) ⎦
The term e 0.2s
can be written as
e 0.2s
= e sT(0.2/T), where T is the sampling period.
put z=e sT ∴ Z[e 0.2s] = z 0.2/T
From z-transform table
⎡ 3 ⎤ z sin 3T
Z⎢ 2 2⎥
= 2
⎢⎣ s + (3) ⎥⎦ z − 2 z cos3T + 1
⎡1 ⎤ z ⎡ 1 ⎤ z
Z⎢ ⎥ = and Z⎢ ⎥= −2T
⎣s⎦ z − 1 ⎣ s + 2⎦ z − e
⎡ 3 ⎛ 1 1 1 1 ⎞⎤
∴ F(z) = Z ⎢ 2 2
+ 2e−0.2 s ⎜ ⋅ − ⋅ ⎟⎥
⎣ s + (3) ⎝ 2 s 2 s + 2 ⎠⎦
z sin 3T ⎡1 z 1 z ⎤
= 2
+ 2 z−0.2 / T ⎢ ⋅ − ⋅ −2T ⎥
z − 2 z cos 3T + 1 ⎣ 2 ( z − 1) 2 ( z − e ) ⎦
since T = 0.05 sec
z sin (3 × 0.05) ⎡1 z 1 z ⎤
∴ F(z) =
2
+ 2 z−0.2 / 0.5 ⎢ ⋅ − ⋅ −2 × 0.05 ⎥
z − 2 z cos (3 × 0.05) + 1 ⎣ 2 ( z − 1) 2 ( z − e )⎦
z sin 0.15 ⎡1 z 1 z ⎤
=
2
+ 2 z −4 ⎢ ⋅ − ⋅ −0.1 ⎥
z − 2 z cos 0.15 + 1 ⎣ 2 ( z − 1) 2 ( z − e ) ⎦
0.149 z ⎡ 1 z ⎤
= 2
+ z−4 ⋅ z ⎢ − ⎥
z − 2 × 0.989 z + 1 ⎣ ( z − 1) ( z − 0.905) ⎦
0.149 z z−3 ( z − 0.905 − z + 1)
= +
z2 − 1.978 z + 1 ( z − 1) ( z − 0.905)
0.5z
Example 10.11.6. Find inverse z-transform of the function F(z) = .
(z − 1) 2
1 1
⎡ 0.5 z ⎤
Hence, f (t) = Z [F(z)] = Z ⎢ 2⎥
= 0.5t
⎣⎢ ( z − 1) ⎦⎥
The function f (t) = 0.5t is sampled and T = 0.5 sec. Therefore in terms of sample
number sequence
f (kT) = kT = 0.5k
The sampled unit ramp function mentioned above with, T = 0.5 sec is shown in Fig.
10.11.3. f (kT)
0 1 2 3 k (sample no.)
0 0.5 1.0 1.5 kT (time) sec.
Fig. 10.11.3. Function f(t) = t sampled with T = 0.5 sec.
Example 10.11.7. Find the inverse z-transform and number sequence for the
function
0.632z
F(z) = 2
z 1.368z + 0.368
given that sampling time (a) T = 1 sec (b) T = 0.2 sec.
0.632 z
Solution. (a) F(z) =
2
z − 1.368 z + 0.368
The denominator can be factorized as (z 1) (z 0.368)
0.632 z
∴ F(z) =
( z − 1) ( z − 0.368)
F ( z) 0.632
or =
z ( z − 1) ( z − 0.368)
0.632 K1 K2
Let = +
( z − 1) ( z − 0.368) z − 1 z − 0.368
The coefficients can be determined as
K1 = 1, K2 = 1
F ( z) 1 1
∴ = −
z z − 1 z − 0.368
z z
or F(z) = −
z − 1 z − 0.368
It is noted that e 1 = 0.368
Sampling time T = 1 sec (given)
SAMPLED DATA CONTROL SYSTEMS 539
z z
∴ F(z) =
−
z − 1 z − e−T
∴ f (t) = Z 1 [F(z)] = (1 e t)
∴ f (kT) = (1 e kT)
or f (k) = (1 e k) Ans.
In terms of number sequence form Z 1 F(z) can be written as
f (k) = (1)k (0.368)k ; T = 1 sec. Ans.
The function f(k) is plotted as shown in Fig. 10.11.4.
f (k T)
0.632
0.864
0.949
0.981
0.993
0 1 2 3 4 5 k (sample no.)
0 1 2 3 4 5 kT (sample no.)
Fig. 10.11.4. f (k) = (1)k (0.368)k; T = 1 sec.
z z
(b) F(z) = −
z − 1 z − 0.368
It is noted that e 1 = 0.368 and sampling time T = 0.2 sec. (given)
∴ e 5T = 0.368
z z
Hence F(z) = 5T
z 1 z e
∴ f (t) = Z 1 [F(z)] = (1 e 5T)
∴ f (kT) = (1 e 5kT)
In number sequence form
f (k) = (1)k (0.368)k ; T = 0.2 sec.
f (k) = (1)k (0.368)k Ans.
f(k T)
0.864
0.949
0.981
0.993
Example 10.11.8. Find inverse z-transform in number sequence form for the function
z
F(z) = ⋅
z 2 − z + 0.5
540 LINEAR CONTROL SYSTEMS
z
Solution. F(z) = 2
z − z + 0.5
The denominator (z2 z + 0.5) can be factorized as
(z2 z + 0.5) = (z 0.5 + j0.5) (z 0.5 j0.5)
F ( z) 1
∴ =
z ( z − 0.5 + j0.5) ( z − 0.5 − j0.5)
1 K1 K2
Let = +
( z − 0.5 + j 0.5) ( z − 0.5 − j0.5) ( z − 0.5 + j 0.5) ( z − 0.5 − j0.5)
The coefficients can be determine as
K1 = j1 and K2 = j1
F ( z) j1 j1
∴ = −
z z − 0.5 + j0.5 z − 0.5 − j0.5
jz jz
or F(z) = −
z − 0.5 + j0.5 z − 0.5 − j0.5
⎡ jz jz ⎤
∴ f (k) = Z 1 ⎢ j − (0.5 − j 0.5) − j − (0.5 + j0.5) ⎥
⎣ ⎦
= j(0.5 j0.5)k (0.5 + j0.5)k]
j(π/4))k
= j(0.5 2 e (0.5 2 ej(π/4))k] = j(0.5 2 )k [e j(π/4)k ej(π/4)k]
⎡⎛ π π ⎞ ⎛ π π ⎞⎤
= j(0.5 2 )k ⎢⎜ cos k − j sin k ⎟ − ⎜ cos k + j sin k ⎟ ⎥
⎣⎝ 4 4 ⎠ ⎝ 4 4 ⎠⎦
⎛ π ⎞
= j(0.5 2 )k ⎜ − j 2 sin k⎟
⎝ 4 ⎠
π
or f (k) = 2(0.5 2 )k sin k Ans.
4
Example 10.11.9. Find the pulse transfer function for the error sampled system
shown in Fig. 10.11.6.
R(s) E(s) *
E(s) 1 C(s)
+ – T = 1 sec. s(s + 1)
G(s) = 1
s(s + 1)
1 K K2
Let = 1 +
s(s + 1) s s+1
SAMPLED DATA CONTROL SYSTEMS 541
The coefficients can be determined as
K 1 = 1 K2 = 1
1 1
∴ G(s) = −
s s+1
z z
Taking z-transform, G(z) = −
z − 1 z − e−T
Since, T = 1 sec.
∴ G(z) = z z z z 0.632 z
− −1
= − =
z−1 z− e z − 1 z − 0.368 ( z − 1) ( z − 0.368)
(2) Since, H(s) = 1
1 1 1
∴ GH(z) = × 1 or GH(s) = −
s (s + 1) s s +1
Taking z-transform
z z
GH(z) = −
z − 1 z − e−T
Since, T = 1 sec
z z z z
∴ GH(s) = − = −
z − 1 z − e−1 z − 1 z − 0.368
0.632 z
or GH(z) =
( z − 1) ( z − 0.368)
(3) The pulse transfer function for the error sampling is given by
0.632 z
C( z) G( z) ( z − 1) ( z − 0.368)
= =
R( z) 1 + GH ( z) 1 + 0.632 z
( z − 1) ( z − 0.368)
Example 10.11.10. Obtain the pulse transfer function for the system shown in
Fig. 10.11.7.
R(s) E(s) *
E(s) 1 C(s)
ZOH
+ – T = 1 sec. s(s + 2)
Example 10.11.11. Determine the unit step time response for the pulse transfer
function
C( z) z
= .
R( z) 2
z − z + 0.5
C( z) z
Solution. = 2
R( z) z − z + 0.5
The term (z2 z + 0.5) can be factorized as follows :
Let (z2 z + 0.5) = 0
the two roots of the above equation are
z1 = (0.5 j0.5) and z2 = (0.5 + j0.5)
∴ (z2 z + 0.5) = (z 0.5 + j0.5) (z 0.5 j0.5)
C( z) z
∴ =
R( z) ( z − 0.5 + j0.5) ( z − 0.5 − j0.5)
Since the input is unit step function i.e. r(t) = 1, therefore,
z
R(z) =
z−1
z z
∴ C(z) = ⋅
z − 1 ( z − 0.5 + j 0.5) ( z − 0.5 − j 0.5)
C( z) z
or =
z ( z − 1) ( z − 0.5 + j 0.5) ( z − 0.5 − j0.5)
K1 K2 K3
= + +
( z − 1) ( z − 0.5 + j0.5) ( z − 0.5 − j0.5)
The coefficients can be determined as
K1 = 2, K2 = 1 and K3 = 1
C( z) 2 1 1
∴ = − −
z z − 1 z − 0.5 + j 0.5 z − 0.5 − j0.5
2z z z
or C(z) = − −
z − 1 z − 0.5 + j0.5 z − 0.5 − j0.5
⎡ 2z z z ⎤
∴ c(k) = Z 1
⎢ − − ⎥
⎣ z − 1 z − 0.5 + j 0.5 z − 0.5 − j 0.5 ⎦
= 2(1)k (0.5 j0.5)k (0.5 + j0.5)k
= 2(1)k (0.5 2 e j(π/4))k (0.5 2 e j(π/4))k
⎛ π π ⎞ ⎛ π π ⎞
= 2(1)k (0.5 2 )k ⎜ cos k − j sin k⎟ − (0.5 2)k ⎜ cos k − j sin k⎟
⎝ 4 4 ⎠ ⎝ 4 4 ⎠
π
or c(k) = 2(1)k 2(0.5 2 )k cos k Ans.
4
544 LINEAR CONTROL SYSTEMS
PROBLEMS
z2
10.3. Find inverse z-transform, F(z) = .
( z − 1) ( z2 − 1.6 z + 1.64)
10.4. (a) Determine pulse transfer function for the sampled-data control system shown in
Fig. 10-P-1.
R *
R 2 C1 *
C1 1 C
T s+1 T s+2
R 1 2 C
s+2 s+1
T = 1 sec.
R(s) + 1 C
*
C(s)
2
T = 1 sec. s + 2s + 2 T = 1 sec.
–
10.7. Obtain z-transfer function for the error sampled system shown in Fig. 10-P-5 and
deter-mine time response if the input is unit step. Given that T = 1 sec and G(s) =
1/[(s + 1) (s + 2)].
R(s) + *
E(s) C(s)
G(s)
T
–
Fig. 10.P-5. Error sampled system for problem 10.7.
10.8. Determine the unit step time response for the sampled-data control system shown
in Fig. 10-P-6. Given that G(s) = 1/[s(s + 1)] ; H(s) = (s + 4) and T = 1 sec.
R(s) E(s) *
E(s) C(s)
G(s)
+ – T
H(s)
ZOH
R(s) + E(s) *
E(s) 2K C(s)
T s(s + 2)
–
Alternatively in the MATLAB command window type sisotool (sys), then next
window is opened by default with root locus editor and Bode plot editor. The root lacus plot
and Bode plot are simultaneously displayed on the computer screen. The plots can be
modified as per the constraints given in the problem and by left clicking the mouse on the
given plot either gain can be changed by dragging the plot up and down or the poles/zeros
simple or complex can be added/deleted.
MATLAB functions used for the solution of examples.
iztrans(F, k) Find inverse Z transform. Find f (KT) given F(Z) i.e. find f (nT)
laplace(X) Finds L[x(t)]
pretty(x) Pretty prints x
sym(v) Convert v to symbolic object
syms x y z Declare x, y and z to be symbolic objects
ztrans(f) Find Z transform of f (nT)
SOLUTION OF PROBLEMS USING COMPUTER 549
The comment % gives more readibility for the programme. The sentence followed by
the comment % is non executable statement and no code is generated in the MATLAB.
The notation semicolon (;) after the MATLAB command suppresses output on the
terminal as well as print out.
If the semicolon (;) is not used the output is displayed either on the terminal or print
out.
It facilitates the beginner not to use semicolon (;) after the command because it gives
the visual display of the data.
Running MATLAB
M files or script files store a sequence of commands. MATLAB is capable of executing
the M file. A program can be written and saved in ASCII format in a file. The file has m
extension in directory. Another type of M file is a function file. The function file has a name
and it follows the work function.
Solution. The numerator and denominator polynomials is s for the given transfer
function are as
p1 = 8s2 + 56s + 96
q1 = s4 + 4s3 + 9s2 + 10s
The numerator is assigned a variable name num and the coefficients of s are arranged
in descending order, same procedure is followed for denominator polynomial. The transfer
function given variable name q1 and further commands are followed sequentially as below :
The pole-zero map is obtained as shown in Fig. 11.1. From the pole-zero map of the
transfer function, poles and zeros are indicated as below :
poles : s = 0 ; s = 2 ; s = 1 + j1 ; s = 1 j1
zeros : s = 3 ; s = 4.
550 LINEAR CONTROL SYSTEMS
2.5
2 x
1.5
1
Imaginary Axis
0.5
0 x x
–0.5
–1
–1.5
–2 x
–2.5
–4 –3.5 –3 –2.5 –2 –1.5 –1 –0.5 0
Real Axis
Fig. 11.1
Example 11.2. A control system is shown in Fig. 11.2. Determine the transfer function.
R(s) + + 4 C(s)
s(s + 4)
– –
+
s + 1.2
+
s + 0.8
Fig. 11.2
Solution. The numerator and denominator of the transfer function in the forward
path of block diagram are found in the first step. The numerator polynomial is considered as
n1 and denominator polynomial as d1. The MATLAB function printsyn is used to display the
transfer function in the forward path. Similarly the transfer function in the second and third
feedback path of the block diagram are displayed. In the successive step the closed-loop
transfer function is obtained using the cloop function. The sign in the paranthesis for the
cloop indicates the ive feedback. Next the parallel function is used to obtain the resultant
transfer function in the feedback path. Finally feedback function is used to obtain the
transfer function as
C(s) 4
=
R(s) s2 + 12s + 12
SOLUTION OF PROBLEMS USING COMPUTER 551
Following commands are used to obtain, the transfer function
num/den =
2s + 2
1
> > [n6, d6] = feedback (n4, d4, n5, d5, – 1)
n6 =
0 0 0 0 4
d6 =
0 0 1 12 12
> > printsys (n6, d6)
num/den =
4
s^2 + 12s + 12
Example 11.3. The figure 11.3 represents a control system, obtain closed-loop
transfer function.
+
R(s) + + 1 C(s)
4 s(s+2)
– – +
0.5
Fig. 11.3
Solution. The MATLAB script for obtaining the closed-loop transfer function is as
follows :
The transfer function for the blocks in the toward path are obtained. The function
Example 11.4. Determine the damping ratio, damped frequency of oscillatory roots
and % overshoot for a unit step response given that
C(s) 1
=
E(s) s(1 + 0.5s) (1 + 0.2s)
The system is unity feedback type.
The closed-loop transfer function for the given condition is determined as below :
1
C(s) s(1 + 0.5s) (1 + 0.2s)
=
E(s) 1 + 1
s(1 + 0.5s) (1 + 0.2s)
554 LINEAR CONTROL SYSTEMS
or = 10
2
(s + 5.5) (s + 1.5s + 1.8)
or C(s) 10 .
=
R(s) s 3 + 7s 2 + 10s + 10
The time response is shown in Fig. 11.4 and following results are noted.
Step Response
1.4
1.2
1.6
1
Amplitude
0.8
0.63
0.6
0.4
0.2
0
0 1 2 3 4 5 ts 6 7 8
tp Time (sec)
T = 1.4 sec. 5.6 sec.
2.72 sec.
Fig. 11.4
SOLUTION OF PROBLEMS USING COMPUTER 555
% Mp = 16%, ξ = 0.5 (as per calculation)
Example 11.5. The open-loop transfer function of a unity feedback control system is
given by
K(s + 2)
G(s) =
s + βs 2 + 4s + 1
3
Determine the value of K and β such that the closed-loop response to unit step function
has ωd = 4 rad/sec.
The value of K and β are arbitrarily chosen and the step response for the transfer
function is displayed. Then by trial and error method new values are selected and same
procedure is repeated till the desired step response having ωd = 4 rad/sec. is obtained.
The step response obtained as per above procedure is shown in Fig. 11.5.
The time for half cycle is noted as 0.78 sec. Hence, the frequency of oscillation is
determined as
(1/2)
ωd = 2π × = 4 rad/sec.
0.78
556 LINEAR CONTROL SYSTEMS
Step Response
From : r
1.4
1.3
1.2
0.8
Amplitude
0.6
To : y
0.4
0.2
0
0 2 4 6 8 10 12
0.78 Time (sec.)
sec
Fig. 11.5. Time response for example 11.5.
Example 11.6. A unity feedback control system has forward path transfer function
given by
s+ 2
G(s) =
s(s + 1)
Determine (a) the time response for unit step response. (b) the rise time (c) the
maximum overshoot.
The rise time is 1.12 sec. and overshoot (%) is 6.7 as shown in Fig. 11.6.
Step Response
1.4
System:sys
1.2 Peak amplitude: 1.067
Overshoot (%): 6.7
At time: 2.37
1 System:sys
Rise Time: 1.12
0.8
Amplitude
0.6
0.4
0.2
0
1 2 2.37 3 4 5 6
Time (sec)
Fig. 11.6
Example 11.7. Determine the time response for the following transfer function
C(s) s+ 2 .
=
R(s) s 2 + 2s + 2
den =
1 2 2
> > sys5 = tf (num, den)
Transfer function :
s + 2
s^2 + 2 s + 2
> > load ltiexamples
> > ltiview
The unit step response is given in Fig. 11.7. From Fig. 11.7, the following results are
obtained
Mp = 6.7%, tp = 2.37 seconds.
1.4
System: sys 5
1.2 Peak amplitude: 1.067
Overshoot (%): 6.7
At time: 2.37
0.8
Amplitude
0.6
0.4
0.2
0
1 2 2.37 3 4 5 6
Time (sec)
Fig. 11.7
Example 11.8. The block diagram of a unity feedback control system is shown in
figure 11.8.
Find Mp , tp the time period of oscillations and the number of cycles completed before
reaching the steady state.
R(s) + 20 C(s)
– (s + 1) (s + 5)
Fig. 11.8
SOLUTION OF PROBLEMS USING COMPUTER 559
Solution. Commands are given below :
From the step response as shown in Fig. 11.9 tp = 0.78 seconds. The period of
oscillation is 2 seconds, ts = 1.19 seconds
C(t)max=0.87
Step Response
0.9
0.8
0.7
0.6
0.5
Amplitude
0.4
0.3
0.2
0.1
tp = ts =
0.78s 1.19s
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2.0
Time (sec)
Fig. 11.9
560 LINEAR CONTROL SYSTEMS
Example 11.9. Determine the time response for a given transfer function
C(s) 4s + 4
=
R(s) s 2 + 2s + 5
for a step input and find %Mp and tp.
Step Response
1.6
1.4
System: sys4
Peak amplitude: 1.53
1.2 Overshoot (%): 91.2
At time: 0.773
1
Amplitude
0.8
0.6
0.4
0.2
0
0 0.773 1 2 3 4 5 6
Time (sec)
Fig. 11.10
SOLUTION OF PROBLEMS USING COMPUTER 561
The transfer function is created using the tf function and variable name sys4 is
assigned. Using the step function, the step response is displayed in Fig. 11.10. The mouse is
right clicked in one corner of window in Fig. 11.10 then peak response is clicked with left
button of mouse. The peak response point is clicked, % overshoot is 91.2 and tp = 0.773
second.
Example 11.10. A unity feedback control system is shown in Fig. 11.11 given below.
By using derivative control the damping ratio is to be made 0.8. Determine the value of Td
and compare, the rise time, peak time and maximum overshoot for the system
(a) without derivative control
(b) with derivative control
R(s) + 16 C(s)
– s2 + 1.6s
Fig. 11.11
16
s^2 + 1.6s + 16
Step Response
1.6
1.52
1.4
1.2
1
Amplitude
0.9
0.8
0.6
0.4
0.2
0.1
0
0 0.8 1 2 3 4 5 6 7
0.45
tp Time (sec.)
tr
Fig. 11.12. Time response for example 11.10 (without derivative control).
From Fig. 11.12 following results are obtained :
tr = 0.45 sec. (time needed to reach output response from 10% to 90%)
tp = 0.8 sec.
Mp = 52.6%
(b) The derivative control is shown in following block diagram.
R(s) + 16 C(s)
1 + sTd
2
– s + 1.6s
Fig. 11.13
The transfer function with derivative control is determined as below :
The time response for unit step input is shown in Fig. 11.14.
Step Response
1.4
1.2
1.12
1
0.9
Amplitude
0.8
0.6
0.4
0.2
0.1
0
0.2 0.4 0.6 0.65 0.8 1 1.2 1.4 1.6 1.8
tp
tr = 0.32 Time (sec)
Fig. 11.14. Time response for example 11.10 (with derivative control).
From Fig. 11.14. following results are obtained :
tr = 0.32 sec (time needed to reach output response from 10% to 90%)
tp = 0.65 sec
Mp = 12%.
564 LINEAR CONTROL SYSTEMS
Example 11.11. Determine an expression for a unit step response for the control
system shown in figure given below :
R(s) + + 1 C(s)
s+1
– – (s + 2)
1
(s + 2)
1
s(s + 1)
Fig.11.15
Solution. The MATLAB script to get the unit step response is as follows :
>> syms s t
n1 = [1 1];
d1 = [0 1];
printsys (n1, d1)
n2 = [0 1];
d2 = [1 2];
printsys (n2, d2)
n3 = [0 1];
d3 = [1 1 0];
printsys (n3, d3)
n4 = [0 1];
d4 = [1 2];
printsys (n4, d4)
[n5, d5] = feedback (n1, d1, n2, d2, –1);
printsys (n5, d5)
[n6, d6] = feedback (n5, d5, n3, d3, –1);
printsys (n6, d6)
n7 = [0 1];
d7 = [1 2];
[n8, d8] = series (n6, d6, n7, d7);
printsys (n8, d8)
s^4 + 4 *s^3 + 5*s^2 + 2*s
C =
s*(2*s^4 + 10*s^3 + 18*s^2 + 14*S + 4)
c = ilaplace (c)
c =
1/2*exp (-t)
>>
The transfer functions in the forward and feedback path are obtained by feeding the
coefficients in the row vectors corresponding to numerator and denominator polynomial
in each case. The feedback function for inner and outer loop is non unity e.g. 1/(s + 2) and
SOLUTION OF PROBLEMS USING COMPUTER 565
1/s(s + 1) respectively. Hence, MATLAB function feedback is used to obtain the closed-loop
transfer function for both the loops. The MATLAB function series is subsequently used for
obtaining the resultant transfer function as the two blocks are connected in cascade.
The answer for the unit step response is
c(t) = 0.5e t.
Example 11.12. A unity feedback control system has its open-loop transfer function
given by G(s) = (4s + 1)/4s2. Determine an expression for the time response when the system
is subjected to
(a) unit impulse input function.
(b) unit step input function.
The open-loop transfer function is assigned a variable name sys1. Next the closed-
loop transfer function of a ive feedback is obtained. The closed-loop transfer function is
given by the variable name F. For a unit impulse input R(s) = 1, ilaplace command of the
command summary table is used to obtain the function f or c(t). The answer is as follows :
1 0.5t 0.5t
c(t) = te +e
4
1
The same procedure is used further but the input is unit step i.e., R(s) =. Hence the
s
function of which the inverse Laplace transform is to be found is modified. The function c(t)
in time domain is obtained as
c(t) = 1 + 0.5 te 0.5t e 0.5t.
Example 11.13. The impulse response of a system is given by g(t) = e t (1 cos 2t).
Determine the transfer function of the system.
>> syms t
>> f = exp (–t) * (1 – cos (2*t))
f =
exp (-t) * (1 – cos (2*t))
>> F = laplace (f)
F =
1/(1 + s) – (1 + s)/((1 + s)^2 + 4)
>>
The first command is symst constructs a symbolic object for time variable t. The
Laplace transform of f [timefunction f(t)] is thus obtained using command Laplace (f).
The transfer function obtained is
1 s+1
G(s) = − Ans.
s + 1 (s + 1)2 + 4
Example 11.14. The block diagram of a unit feedback control system is shown in
Fig. 11.16.
R(s) + 20 C(s)
– (s + 1) (s + 5)
Fig. 11.16
Solution. Determine the characteristic equation of the system, ωn, ξ, ωd, tp, %Mp, the
time at which the first understood occurs and the time period of oscillations and the number
of cycles completed before reaching the steady state.
SOLUTION OF PROBLEMS USING COMPUTER 567
1.5708
>> oscillations = (wd/2* pi))* (4/(zeta*wn)) % number of
oscillation before steady state
oscillations-
0.8488
>>
Example 11.15. The open-loop transfer function of unity feedback control system is
given by
G(s) = 25 .
s(s + 5)
568 LINEAR CONTROL SYSTEMS
n1 = 25 ; d1 = [1 5 0] ;
[n2, d2] = cloop (n1, d1, –1) ;
sys2 = tf(n2, d2)
wn = (25)
zeta = 5/(2* wn)
wd = wn* sqrt (1 – zeta^2)
mp = exp ((- zeta* pi)/sqrt (1 – zeta^2))* 100
ess = (2* zeta)/wn
% increase of damping ratio from 0.5 to 0.75
% with tacho feedback
zetan = 0.75
wn = sqrt (25)
kt = (2* zetan* wn – 5)/25
mp = exp ((– zetan* pi)/sqrt(1 – zetan^2))* 100
Transfer function :
25
s^2 + 5s + 25
wn =
5
zeta =
0.5000
wd =
4.3301
mp =
16.3034
ess =
0.2000
zetan =
0.7500
wn =
5
kt =
0.1000
mp =
2.8375
>>
SOLUTION OF PROBLEMS USING COMPUTER 569
For the given open-loop transfer function, from the results,
ωn = 5 rad/sec., ωd = 4.3301 rad./sec., damping ratio = 0.5, % Mp = 16.3034, ess = 0.2
With damping ratio = 0.75 and tachometer feedback Kt = 0.1, % Mp = 2.8375.
Example 11.16. The open-loop transfer function of a unity feedback control system is
given by
10
G(s) = 2
s + 6s + 10
Find (a) Kp (b) Kv (c) Ka.
Solution. The MATLAB script of obtain the values of Kp, Kv and Ka is written below :
numg=
10
deng=
1 6 10
Transfer function:
10
s^2 + 6 s + 10
kp =
1
Ess=
0.5000
570 LINEAR CONTROL SYSTEMS
numsg=
10 0
densg=
1 6 10
Transfer function :
10 s
s^2 + 6 s + 10
kv =
0
Warning: Divide by zero.
ess=
Inf
nums2g =
10 0 0
dens2g =
1 6 10
Transfer function:
10 s^2
s^2 + 6 s + 10
ka =
0
Warning: Divide by zero.
ess =
Inf
>>
The open-loop transfer function is assigned a variable G. The command d.c. gain (G)
finds the dc gain for G(s) i.e., s = 0 and as per the formula Kp = lim G(s), it is the value of
s→0
Kp, Kp = 1 is obtained. For finding Kv, as per theory Kv = lim sG(s), hence sG(s) is obtained
s→0
and d.c. gain (sG) given Kv = 0 and using lim s2G(s), Ka = 0 is calculated. This solution uses
s→0
same commands to obtain steady error in each case for step, ramp and parabolic inputs
accordingly.
Example 11.17. Determine the position, velocity and acceleration error coefficients
for a system given by
100(s + 2)(s + 40)
G(s)H(s) = .
s 3 (s 2 + 4s + 200)
Solution. The MATLAB script for finding the position, velocity and acceleration error
coefficients is as follows.
SOLUTION OF PROBLEMS USING COMPUTER 571
Kp = ∞, Kv = ∞ and Ka = ∞.
Example 11.18. The open-loop transfer function of a control system is given below
2(s 2 + 3s + 20)
G(s)H(s) =
s(s + 2)(s 2 + 4s + 10)
Determine the static error coefficients and steady state error for the input given as :
(a) 5 (b) 4t (c) 4t2/2.
572 LINEAR CONTROL SYSTEMS
Solution. The MATLAB script for obtaining the static error coefficients and a steady
state error is as follows :
ka =
0
Warning : Divide by zero.
ess =
Inf
>>
Example 11.19. A closed-loop control system when subjected to a unit step input has
an expression for time response given by
c(t) = 0.5 + 1.25e t 1.75e 12t
Determine the overall transfer function.
Solution. The MATLAB script is written below :
>> syms s t
>> c = 0.5+1.25*exp (–3*t)–1.75*exp (12*t)
c =
1/2+5/4*exp (–3*t)–7/4*exp (-12*t)
>> C = laplace (c)
C =
1/2/s+5/4/(s+3)–7/4/(s+12)
>>
The first command after execution constructs the symbolic objects for s and t. Next,
the output c(t) is fed. The MATLAB function laplace finds the Laplace transform of c(t). The
output c(t) of the closed-loop transfer function is for a specified reference step input r(t) and
1
R(s) = .
s
The Laplace transform of the output c(t) as obtained from MATLAB application is
⎛1/ 2 5 / 4 7 / 14 ⎞
C(s) = ⎜
+ 12 ⎟⎠
+ −
⎝ s s + 3 s
1
∵ R(s) =
s
C ( s) ⎛1/ 2 5 / 4 7 / 14 ⎞
∴ = s⎜
s + 3 s + 12 ⎟⎠
+ −
R(s) ⎝ s
C ( s) 17.25(s + 1.043)
or = Ans.
R( s) (s + 3)(s + 12)
Example 11.20. A unit step input is applied to a unity feedback control system whose
open-loop transfer function is given by
K
G(s) =
s(sT + 1)
Determine the values of K and T given that maximum overshoot is 25%. Calculate the
gain crossover frequency and phase margin.
574 LINEAR CONTROL SYSTEMS
Solution. In the given transfer function the numerical values of K and T are to be
chosen and the corresponding time response plot is obtained in order to get the Mp = 25%.
The same procedure is to be followed number of times by trial and error and ultimately the
final value of Mp = 25% is obtained from the displayed transient response plot. The
commands to be followed sequentially are as below :
After execution of this command the corresponding time response plot is selected from
the menu of loop response after clicking the loop (step) response. The time response is
highlighted with the mouse and in the window, time response plot is displayed as shown in
Fig. 11.17.
Step Response
From : r
1.4
System: Closed-loop: r to y System: Closed-loop: r to y
I/O: r to y I/O: r to y
1.2
Time (sec): 0.365 Time (sec): 1.47
Amplitude: 1.25 Amplitude: 0.996
1
Amplitude
0.8
To : y
0.6
0.4
System: Closed-loop: r to y
0.2 I/O: r to y
Time (sec): 0.077
Amplitude: 0.216
0
0 0.5 1 1.5
Time (sec)
Fig. 11.17. Unit step response for example 11.20.
The value of K = 11.9 and T = 0.13 gives 20% maximum overshoot.
The same command i.e. >> sisotool (sys10) is to be executed and instead of step
response the Bode plot is highlighted and OK button is clicked for the display box. The Bode
plot is shown in Fig. 11.18 and phase margin of 43.3° at a given crossover frequency ω = 8.16
rad/sec is noted.
SOLUTION OF PROBLEMS USING COMPUTER 575
Open-loop Bode Editor (C)
50
40
30
20
10
–0
Gain
crossover frequency
–10 8.16 rad/sec.
–20
–30
G.M. : Inf
–40 Freq. Inf
Stable loop
–50
–90
–135
P.M.
Example 11.21. A unity feedback control system has its forward path transfer
function as
210
G(s) =
s(s + 2)(s 2 + 12s + 6)
Give a MATLAB programme to determine
(a) The roots of the characteristic equation
(b) The Bode plot for open-loop transfer function.
576 LINEAR CONTROL SYSTEMS
Pole-zero Map
1
0.8
0.6
0.4
0.2
Imag axis
–0.2
–0.4
–0.6
–0.8
–1
35 30 25 20 15 10 5 0
Real axis
Fig. 11.19
SOLUTION OF PROBLEMS USING COMPUTER 577
For the given open-loop transfer function G(s), the closed-loop transfer function is
calculated first. The characteristic equation is
s4 + 44s3 + 390s2 + 192s + 210 = 0
The roots of the characteristic equation are obtained using MATLAB function roots
and the roots are
31.9898, 11.5578, 0.2262 + 0.7181i and 0.2262 0.7181i
The roots are shown in Fig. 11.19.
The Bode plot is obtained using function sisotool and shown in Fig. 11.20.
Open-loop Bode Editor (C)
50
–50
–100
–150
G.M.: 18.1 dB
Freq: 2.09 rad/sec
Stable loop
–200
–90
–180
–270
Example 11.22. Sketch the Bode plot for the transfer function given below :
2(s + 0.25)
G(s)H(s) = 2
s (s + 1)(s + 0.5)
578 LINEAR CONTROL SYSTEMS
Solution. The given transfer function G(s)H(s) is modelled as sys1 using following
commands.
Bode Diagram
Gm = –20.561 dB (at 0.35355 rad/sec.), Pm = –36.665 deg (at 1.117 rad/sec)
100
Gain
50
crossover frequency
Magnitude (dB)
w2 = 1.117 rad/sec
20
G.M.
0
– 50
– 10
– 135
– 180
Phase (deg.)
Phase P.M.
crossover frequency
w2 = 0.35353 rad/sec
– 225
– 270
10–2 10–1 100 101
Frequency (rad/sec)
Fig. 11.21. Bode plot for example 11.22.
SOLUTION OF PROBLEMS USING COMPUTER 579
The function margin sys1 computes and plots the gain margin and phase margin
directly as displayed in Fig. 11.21
The phase crossover frequency = 0.35355 rad/sec
The gain crossover frequency = 1.117 rad/sec
The gain margin = 20.561 db
The phase margin = 36.665 deg.
As both margins are negative, hence the system is unstable.
>> num = 4
num =
4
>> den = [.04 .58 1 0]
den =
0.0400 0.5800 1.0000 0
>> sys = tf (num, den)
Transfer function :
4
0.04s ^ 3 + 0.58s ^ 2 + s
>> load ltiexamples
>> ltiview
margin (sys)
The gain margin and phase margin are positive, as shown in Fig. 11.22 the system is
stable.
580 LINEAR CONTROL SYSTEMS
Bode Diagram
Gm = 11.186 dB (at 5 rad/sec.), Pm = 27.826 deg (at 2.4696 rad/sec)
50
0
Magnitude (db)
Gain G.M.
crossover frequency
w1 = 11.186 rad/sec.
–50
–100
–90
–135 Phase
crossover frequency
Phase (deg.)
w2 = 5 rad/sec.
PM
–180
–225
–270
–1 0 1 2
10 10 10 10
Frequency (rad/sec.)
Fig. 11.22. Bode plot for example 11.23.
>> num = 50
num =
50
>> den = [1 3 2]
den =
1 3 2
>> sys 3 = tf (num, den)
Transfer function :
50
s∧ 2 + 3s + 2
>> load ltiexamples
>> ltiview
SOLUTION OF PROBLEMS USING COMPUTER 581
The transfer function model sys3 is obtained. The lti examples are loaded. The lti
view browser is opened next. Under file menu, sys3 is imported. Nyquist plot is chosen
(Fig. 11.23). The closed-loop stability for the given open-loop transfer function as indicated is
stable.
Nyquist Diagram
15
10
System: sys3
5 Phase margin (deg): 24.4
Imaginary Axis
–5
–10
–15
–1 0 5 10 15 20 25
Real Axis
Fig. 11.23. Nyquist plot example 11.24.
Example 11.25. Draw the Nyquist plot for the open-loop transfer function given
below and comment on closed-loop stability
2.2
G(s)H(s) = .
s(s 2 + 2s + 2)(s + 1)
Nyquist Diagram
30
System : sys 4
Real : – 2.2
Imag : 28
20 Freq (rad/sec) : – 0.0393
System : sys 4
Phase margin (deg.) 0.531
10
Imaginary Axis
– 10
System : sys 4
– 20 Real : – 2.2
Imag : – 29.5
Freq (rad/sec) : 0.0374
– 30
–2 – 1.5 –1 – 0.5 0
Real Axis
Fig. 11.24. Nyquist plot for example 11.25.
The transfer function is created by selecting the proper values of coefficients of
numerator and denominator polynomials in s. The lti examples are loaded in the next step.
To load the model sys4 into the lti viewer under file menu, as a first step import is selected
and the desired model sys4 is chosen in the lti browser. After importing the model in the
1ti viewer, the plot type i.e. Nyquist is highlighted. The Nyquist plot is obtained as shown in
Fig. 11.24.
The phase margin obtained is 0.531° which is very small ; as such the system can be
considered as marginally stable.
Solution.
(i) Following MATLAB Commands are given
>> n1 = [10 – 10]
n1 =
10 – 10
>> d1 = [1 5 6]
d1 =
1 5 6
>> sys 5 = tf (n1, d1)
Transfer function:
10s - 10
s ^ 2 + 5s + 6
>> load ltiexamples
>> ltiview
SOLUTION OF PROBLEMS USING COMPUTER 583
The given transfer function is obtained using tf command. After execution of
ltiview, the litviewer window is opened. In the litviewer, from the file menu click the
button on import. In the ltibrowser window select the system to import e.g. sys5 and click
the OK button. In the litviewer by default the step response is displayed. Then in the right
hand corner right click the mouse and select plot type. Then select Nyquist from the given
menu. The Nyquist diagram is obtained as shown in Fig. 11.25.
Nyquist Diagram
2
1.5
1
System: sys5
Real: –1.67
Imaginary Axis
0.5
Imag: 0 w=–¥
Freq. (rad/sec): 0
0
–1 + j0 w=¥
–0.5
–1
–1.5
–2
–2 –1.5 –1 –0.5 0 0.5 1 1.5 2
Real Axis
Fig. 11.25
At point a, the plot crosses the ive real axis (ω = 0). The critical point ( 1 + j0) is
encircled once in clockwise direction
∴ N= 1
The number of open-loop poles is nil
∴ P+ = 0
∴ N = P+ Z+
or 1 = 0 Z+
∴ Z+ = 1
The closed-loop system is unstable.
(ii) Following MATLAB commands are given
>> n1 = [4 –4]
n1 =
4 – 4
>> sys10 = tf (n1, d1)
Transfer function:
4s - 4
s^2 + 5s + 6
>> load ltiexamples
>> ltiview
584 LINEAR CONTROL SYSTEMS
As per procedure given in (ii) The Nyquist plot is obtained as shown in Fig. 11.26.
The critical point ( 1 + j0) is not encircled by the Nyquist plot the closed-loop system is
stable.
Nyquist Diagram
0.8
0.6
0.4
System: sys10
Real: –1.667
Imaginary Axis
0.2
w=0 Imag: 0 w=–¥
Freq. (rad/sec): 0
0
w=¥
w = –0
– 0.2
–1 + j0
– 0.4
– 0.6
– 0.8
– 0.8 – 0.6 – 0.4 – 0.2 0 0.2 0.4 0.6 0.8
Real Axis
Fig. 11.26
Example 11.27. Sketch the root locus plot for the open-loop transfer function given
below :
K(s 2 + 4)
G(s)H(s) =
s(s + 2)
Calculate the value of K at (a) breakaway point and (b) s = 0.69 + j0.88.
Solution.
The following commands are given
The transfer function is obtained with K = 1 using the tf function with p and q vari-
ables. The p and q variables correspond to the coefficients in the numerator and denominator
polynomials respectively. After execution of rlocus function, the root locus is displayed. The
rlocfind command when executed open, the graphic window displaying the root locus and
the cursor. The cursor is placed on the breakaway point s = 0.81 and then K = 0.2072 is
displayed.
After execution of rlocfind command and placing the cursor at s = 0.69 + 0.88i
K = 0.4575, is obtained as shown in Fig. 11.27.
Root Locus
2
1.5
1 s = – 0.69 + 0.88 i
Imaginary Axis
0.5
0
s = – 0.81
–0.5
–1
–1.5
–2
Example 11.28. The open-loop transfer function of a unity feedback system is given
below :
K
G(s) =
s(s + 2)(s 2 + 2s + 2)
Plot the root locus to determine the value of gain K (i) at the breakaway point (ii) for
marginal stability.
Following commands are given to obtain the root locus plot as shown in Fig. 11.28.
>> p = 1
p = 1
>> q = [1 4 6 4 0]
q =
1 4 6 4 0
>> sys 5 = tf (p, q)
Transfer function:
1
s^4 + 4s^3 + 6s^2 + 4s
>> rlocus (sys 5)
Root Locus
2 ¥ ¥
K
1.5 K
K=0
1
Imaginary Axis
0.5
0
K=0 K=0
– 0.5
–1
K=0
– 1.5
K K
–2 ¥ ¥
K
2
=1
(− 1)(− 1 + 2) [(− 1) + 2(− 1) + 2]
K
or =1 ∴ K=1
− 1(1)(1)
SOLUTION OF PROBLEMS USING COMPUTER 587
(ii) The root locus plot crosses the imaginary axis at s = j1, for marginal stability at
s = j1
K =1
( j1)( j1 + 2) [(− j1)2 + 2(− j1) + 2]
K
or =1
( j1)( j1 + 2)( j 2 + 1)
K
or =1 ∴ K = 5.
2 2
1 1 +2 22 + 12
Example 11.29. Sketch the root locus plot for the open-loop transfer function
K
G(s)H(s) =
s(s + 1)(s + 3)
Determine (i) the value of K at s = 4
(ii) the frequency of sustained oscillations.
Following commands are given to get the transfer function assigned by variable sys 3.
>> p = 1
p = 1
>> q = [1 4 3 0]
q =
1 4 3 0
>> sys 3 = tf (p, q)
Transfer function:
1
s^3 + 4s^2 + 3s
>> rlocus (p, q)
Root Locus
6
System: sys
4 Gain: 12
Pole: 1.72i
2 Frequency
Imaginary Axis
–8 –6 –4 –2 (rad/sec): 1.72
0
System: sys
Gain: 12
–2 Pole: – 4
Damping: 1
–4
–6
–8 –6 –4 –2 0 2
Real Axis
Fig. 11.29
(i) The value of gain K at s = 4 is obtained below
K
=1
− 4(− 4 + 1)(− 4 + 3)
K
or =1 ∴ K = 12 at s = 4
− 4 (− 3)(− 1)
(ii) The root locus plot crosses the imaginary axis at s = j 1.72, hence the frequency of
sustained oscillations is ω = 1.72 rad./sec.
Example 11.30. Sketch the root locus plot for the system when the open-loop transfer
function is given by
K
G(s) = 2
.
s(s + 4)(s + 4s + 13)
>> p = 1
p = 1
>> q = [1 8 29 52 0]
q =
1 8 29 52 0
>> sys4 = tf (p, q)
Transfer function:
1
s^4 + 8s^3 + 29s^2 + 52s
>> rlocus (p, q)
SOLUTION OF PROBLEMS USING COMPUTER 589
The root locus plot is shown in Fig. 11.30.
Setting the cursor at the selected location on the root locus plot, the desired informa-
tion with reference to system performance can be obtained.
Root Locus
8 ¥ ¥
K K
6
4 K=0
Imaginary Axis
2
– 10 –8 –6 –4 –2 0
0
K=0 K=0
–2
–4 K=0
–6 K
K
¥
–8 ¥
– 10 –8 –6 –4 –2 0
Real Axis
Fig. 11.30
Example 11.31. The open-loop transfer function of a unity feedback control system is
given by
K
G(s) =
s(1 + 0.2s)
Design a lead compensator such that the system will have Kv = 10 and P.M. = 57°.
K
Solution. Kv = lim s G(s) = lim s .
s→0 s→0 s (1 + 0.2s)
K
∴ 10 = lim s .
s→0 s (1 + 0.2s)
or K = 10
and in order to satisfy steady state error requirement
G(s) = 10
s (1 + 0.2s)
The following MATLAB commands are given to plot the Bode plot for uncompensated
system.
>> n = 10; d = [0.2 1 0];
sys 6 = tf (n, d)
Transfer function:
10
0.2s^2 + s
>> sisotool (sys 6)
590 LINEAR CONTROL SYSTEMS
In the graphical window of sisotool, the Bode plot for the open-loop transfer
function is shown as in Fig. 11.31. The phase margin of 38.7° is displayed.
The phase margin is less than the desired value. The transfer function of the lead
compensator is chosen by trial and error by assigning various values of zeros and poles for
the network. The reader has the flexibility of choosing real or complex values.
The gain crossover frequency displayed is 6.24 rad./sec. Selecting zero of the lead
compensating network say at ω = 5.5 rad/sec. and pole at ω = 13.8 rad/sec.
10
Magnitude (db)
– 10
– 20
– 30
– 40
G.M. : Inf
– 50 Freq : Inf
Stable loop
– 60
– 90°
Phase (deg.)
– 135°
PM : 38.7°
Freq : 6.24 rad/sec
– 180° 2
100 2 3 4 5 6 7 8 9 101 20 30 40 50 60 10
Frequency (rad/sec.)
Fig. 11.31. Bode plot for uncompensated system.
The transfer function of the lead compensator is
j ω + 5.5
Gc (jω) =
jω + 13.8
5.5( j 0.018 ω + 1)
or Gc (jω) =
13.8 ( j 0.072 ω + 1)
0.398 ( j 0.18 ω + 1)
or Gc (jω) =
( j 0.072 ω + 1)
SOLUTION OF PROBLEMS USING COMPUTER 591
Open-loop Bode Editor (C)
20
10
Magnitude (db)
– 10
– 20
G.M. : Inf
– 30 Freq : Inf
Stable loop
– 40
– 90
Phase (deg.)
– 135
PM : 57.2°
Freq : 8.04 rad/sec
– 180 1 2
100 2 3 4 5 6 7 8 9 10 20 30 40 50 10
Frequency (rad/sec.)
Fig. 11.32
Accounting for attenuation factor of 0.398 by increasing gain by 1/0.398. The transfer
function of lead compensator can be modified as
( j 0.18 ω + 1)
Gc(jω) =
( j 0.072 ω + 1)
The transfer function of the compensated system is
( j 0.18 ω + 1) × 10
Gc(jω) G(jω) =
( j 0.072 ω + 1) jω(1 + j 0.2 ω)
(0.18 s + 1) × 10
or Gc(s) G(s) =
( j 0.072 s + 1) s(1 + 0.2s)
The Bode plot for the above compensated transfer function is displayed in Fig. 11.32
and the phase margin in noted as 57.2° Ans.
Example 11.32. Use Nichols chart to obtain the closed-loop frequency response
C(jω)/R(jω) for a unity feedback control system having open-loop transfer function given
below :
20
G(s) =
s(s + 2)(s + 5)
592 LINEAR CONTROL SYSTEMS
>> n1 = 20
20
n1 =
>> d1 = [1 7 10 0]
d1 =
1 7 10 0
>> sys 3 = tf (n1, d1)
Transfer function:
20
s ^ 3 + 7s ^ 2 + 10s
>> load ltiexamples
>> ltiview
The given transfer function is assigned the variable name sys3. It is obtained by
executing three commands. The lti examples are loaded. After execution of MATLAB
command lti view, lti view window is opened. In this window variable name sys3 is
highlighted by clicking the mouse button, subsequently OK buttor in pressed. By default the
step response of the given transfer function is displayed. At the bottom of extreme right hand
corner the plot type Nichols is clicked. The Nichols chart is then obtained as shown in
Fig. 11.33.
The grid is further clicked. The gain phase and frequency is displayed at the points a,
b, c and d as shown in Fig. 11.33. After preparing the table for the selected points, the
closed-loop response can be obtained.
Ans.
Example 11.33. Obtain the state model for the transfer function given below :
C(s) (s + 2)
= .
R(s) (s + 1)(s + 3)
Dp=D
40 System : sys 3 Cp=C*P
p = [0
Bp=inv(P)*B
System : sys 3 3 dB a
b –3 dB
Phase (deg): – 178
Ap=inv(P)*A*P
1
0]
– 40 –40 dB
d System : sys 3
Phase (deg): – 239
Gain (dB) : – 40.5
– 60 –60 dB
– 80 –80 dB
– 100 –100 dB
– 120 –120 dB
– 140 –140 dB
– 360 – 315 – 270 – 225 – 180 – 135 – 90 – 45 0
Open-loop phase (deg)
P =
0 1
1 0
Ap =
0 1
–3 –4
Bp =
0
1
Cp =
2 1
Dp =
0
A = P =
–9 –26 –24 0 0 1
1 0 0 0 1 0
0 1 0 1 0 0
B = Ap =
1 0 1 0
0 0 0 1
0 –24 –26 –9
C = Bp =
1 1 2 0
D = 0
0 1
Cp =
2 1 1
Dp =
0
Example 11.35. Obtain the state model for the following transfer function by direct
decomposition.
Y(s) 1
= .
U(s) s 2 + 2s + 5
P = [0 1; 1 0]
% Phase variable form
Ap = inv(P)*A*P
Bp = inv(P)*B
Cp = C*P
Dp = D
A = P =
– 2 – 5 0 1
1 0 1 0
B = Ap =
1 0 1
0 – 5 – 2
C = Bp =
0 1 0
D = 1
0 Cp =
1 0
Dp =
0
>>
Example 11.36. Write MATLAB script to determine the state transition matrix for
⎡ 1 4⎤
A= ⎢ ⎥.
⎣− 2 − 5 ⎦
Solution. The MATLAB script for determination of the state transition matrix is as
follows :
⎡ x1 ⎤ ⎡1 1⎤ ⎡ x1 ⎤ ⎡0 ⎤
Solution. ⎢ x ⎥ = ⎢0 − 1⎥ ⎢ x ⎥ + ⎢1 ⎥ u
⎣ 2⎦ ⎣ ⎦ ⎣ 2⎦ ⎣ ⎦
⎡1 1⎤
Hence, the system matrix A = ⎢ ⎥
⎣0 − 1 ⎦
⎡0 ⎤
B= ⎢ ⎥
⎣1 ⎦
The MATLAB script is as follows :
>> A = [1 1; 0 –1]
A =
1 1
0 –1
>> B = [1; 0]
B =
1
0
>> P = ctrb(A, B)
P =
1 1
0 0
>> RANK(P)
ans =
1
>>
Example 11.38. Obtain the transfer function from the following data
⎡ −5 1 ⎤ ⎡ 1⎤
A= ⎢ ⎥ , B = ⎢ ⎥ , C = [2 1] , D = 0.
⎣ −6 0 ⎦ ⎣ 2⎦
The ss2tf command converts state variable form to transfer function. The transfer
function is
Y(s) 4s + 8
= 2 .
U(s) s + 5s + 6
>> A = [0 1 ; –6 –5]
A =
0 1
–6 –5
>> B = [0 ; 1]
B =
0
1
>> C = [2 1]
C =
2 1
>> D = [0]
D =
0
>> [num, den] = ss2tf (A,B,C,D)
num =
0 1.0000 2.0000
den =
1.0000 5.0000 6.0000
>> printsys (num, den)
num/den =
s+2
s^2 + 5s + 6
The matrices A, B and C are arranged first, the first command displays the matrix A.
The ss2tf converts the state variable form to transfer function form. The transfer
function obtained is
Y (s) s+2
= 2 . Ans.
U (s) s + 5s + 6
Example 11.40. Obtain the pulse transfer function for the system shown in figure
below :
Fig. 11.34
1 − e− st
Gho(s) =
s
600 LINEAR CONTROL SYSTEMS
1 t 1 4t 3t 3 3t sin
1.1. (i) f (t) = e e (ii) f (t) = e cos 2t e 2t
3 3 2
(iii) f (t) = cos t cos 2t (iv) f (t) = e 0.5t (1 0.25t)
2 t 1 3t
(v) f (t) = e + e
3 3
C(s) 1 2+ 3 (2 + 3) t (2 3) t
1.3. = 2 ; e(t) = [e +e ]
R(s) s + 4 s + 1 2 3
1.5. (i) F(∞) = 0 (ii) F(∞) = 0.4.
Chapter 2
Eo (s) K
2.7. = .
Ei (s) (sR1C1 + 1)(sR2 C2 + 1)
Chapter 3
C G1G2 + G3
3.1. =
R 1 + G2 H1
G1G2 + G1G3
3.3. Overall transfer function =
1 + G1G2 H1 + G2 H2 + G3 H2
690
APPENDIX I : ANSWERS TO TYPICAL PROBLEMS 691
G4H2(G2 + G3)
G5 + G4(G2 + G3)
Chapter 4
Chapter 5
f2
s
Y (s) K
5.1. =
X (s) ( f1 + f2 ) s + 1
K
5.3. M1 x1 + f1 ( x1 x 2 ) + K2x1 = f (t)
Chapter 6
C(s) 4 0.998t 1
6.3. = c(t) 1.33 [1 1.22e sin (1.41t + tan 1.41)]
R(s) s2 + 2s + 3
6.5. ζ = 0.59, ωn = 1.69 rad/sec 6.7. Mp = 27.5%, ts = 2 sec.
C(s) 17.25 (s + 1.043) e−2 t 1
6.9. = 6.11. c(t) = 1 − − sin (t + 45°)
R(s) (s + 3) (s + 12) 2 2
692 LINEAR CONTROL SYSTEMS
Chapter 7
7.1. (a) stable, (b) stable, (c) unstable, (d) unstable and (e) limitedly stable
7.3. Unstable, 2 roots with positive real parts
7.5. K < 69
7.7. (a) stable, (b) unstable, (c) unstable (d) stable ; G.M. 6.84 db and (e) Marginally stable.
7.9. (a) stable, if K < 6, (b) K > 1
Chapter 8
8.1. K = 2
1 1
8.3. β = 2.99, = 2 rad/sec, = 0.668 rad/sec.
T βT
8.5. Kt = 0.4.
APPENDIX I : ANSWERS TO TYPICAL PROBLEMS 693
Chapter 9
9.1. x1 = i, x2 = vc
⎡ x1 ⎤ ⎡106 103 ⎤ ⎡ x1 ⎤ ⎡103 ⎤
⎢⎣ x 2 ⎥⎦ = ⎢ 9 ⎥ 10
0 ⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣ 0 ⎥⎦
+
⎣10
⎡ R1 1 ⎤
⎢− L 0 − ⎡1 ⎤
L1 ⎥
⎡ x1 ⎤ ⎢ 1 ⎥ ⎡ x1 ⎤ ⎢ L1 ⎥
⎢ R2 1 ⎥ ⎢ ⎥ ⎢ ⎥ ⎡ y1 ⎤ ⎡1 0 ⎤ ⎡ x1 ⎤
⎢
9.3. x 2 =⎥ 0 x2 + 0 e,
L2 ⎥ ⎢ x ⎥ ⎢⎢ 0 ⎥⎥ ⎢⎣ y2 ⎥⎦ ⎢⎣0 1 ⎥⎦ ⎢⎣ x2 ⎥⎦
− =
⎢ x ⎥ ⎢ L2
⎣ ⎦ 3 ⎢ 1 ⎥ ⎣ 3 ⎦
1 ⎢ ⎥
⎢ 0 ⎥ ⎣ ⎦
⎢⎣ C C ⎥⎦
⎡ Ra Rb ⎤
⎢− L 0 −
⎡ x1 ⎤ La ⎥ ⎡ x1 ⎤ ⎡ 1 ⎤ ⎡ x1 ⎤
⎢ a ⎥
9.5. ⎢ x 2 ⎥ = ⎢ 0 0 1 ⎥ ⎢ x2 ⎥ + ⎢ La ⎥ ea, y = [0 1 0] ⎢ x2 ⎥
⎢ x ⎥ ⎢ ⎥ ⎢x ⎥
⎣ 3⎦ ⎢ Kb f ⎥ ⎢x ⎥ 0 ⎣ 3⎦
⎢ J 0 − ⎥⎣ 3⎦ ⎣ ⎦
⎣ J ⎦
9.7. x1 = y1, x3 = y2 x2 = y1 , x4 = y 2
⎡ 0 1 0 0⎤
⎢⎛ K K ⎞ f K2 ⎥ ⎡ 0 ⎤
⎡ x1 ⎤ ⎢⎜ − 1 − 2 ⎟ − 0 ⎥ ⎡ x1 ⎤ ⎢ 0 ⎥
⎢ x 2 ⎥ ⎢⎝ M1 M2 ⎠ M1 M2 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ f (t)
⎢ x ⎥ = ⎢ 0 0 0
⎢ ⎥
1 ⎥ ⎢ x3 ⎥ ⎢ 1 ⎥
⎢ 3⎥ ⎢ ⎥ ⎢x ⎥ ⎢ ⎥
⎢⎣ x 4 ⎥⎦ ⎢ K2 K ⎣ 4⎦ ⎢M ⎥
0 − 2 0⎥ ⎣ 2⎦
⎣⎢ M2 M2 ⎥⎦
⎡ 0 1 0⎤ ⎡0 ⎤
9.9. A = ⎢ 0 0 1⎥ , B = ⎢0 ⎥ , C = [2 1 1]
⎢ − 24 − 26 − 9 ⎥ ⎢1 ⎥
⎣ ⎦ ⎣ ⎦
⎡ 0 1 0⎤ ⎡0 ⎤
9.11. A = ⎢ 0 0 1⎥ , B = ⎢0 ⎥ , C = [1 6 0]
⎢ − 6 ⎦⎥ ⎢1 ⎥
⎣− 6 − 11 ⎣ ⎦
⎡ ⎛ 4 − t 1 −4 t ⎞ ⎛ 1 − t 1 −4 t ⎞ ⎤
⎢ ⎜⎝ 3 e − 3 e ⎟⎠ ⎜⎝ e − e ⎟⎠ ⎥
3 3
9.13. φ(t) = ⎢ ⎥
⎢ ⎛ − 4 e− t + 1 e−4 t ⎞ ⎛ 1 − t 4 −4 t ⎞ ⎥
⎢⎣ ⎜⎝ 3 ⎟⎠ ⎜⎝ − e + e ⎟⎠ ⎥
3 3 3 ⎦
⎡ 5 − t 2 −4 t ⎤ ⎡ 1 −t 1 −4 t ⎤
e − e ⎥ ⎢1 − 3 e + 12 e ⎥
⎡ x1 (t) ⎤ = ⎢ 3 3 , ⎡ x1 (t) ⎤
= ⎢
⎢ 5 8 −4 t ⎥ ⎢⎣ x2 (t) ⎥⎦ 1 − t 1 −4 t ⎥
⎣⎢ x2 (t) ⎦⎥ ⎢ − e −t
+ e ⎥ ⎢ e − e ⎥
( ZSR )
( ZIR)
⎣⎢ 3 3 ⎦⎥ ⎢⎣ 3 3 ⎦⎥
⎡ 4 −t 7 −4 t ⎤
⎡ x1 (t) ⎤ ⎢1 + 3 e − 12 e ⎥
Total responce ⎢ = ⎢ ⎥
( ZIR + ZSR) ⎣ x2 (t)⎥⎦ 4 7
⎢ − e− t + e−4 t ⎥
⎣⎢ 3 3 ⎦⎥
694 LINEAR CONTROL SYSTEMS
Chapter 10
z (z − 2 + T ) z ( z − cos ωT )
10.1. (a) 2
(b) 2
( z − 1) z − 2 z cos ωT + 1
695
696 LINEAR CONTROL SYSTEMS
1
7. For input u(t) = 2 the steady output for the transfer function G(s) = 2
is
s + 1.5s + 4
(a) 2 (b) 0.5
(c) 1.5 (d) 1
8. The forward-path gain of a control system is lmj
2.5 and the pole-zero configuration of the
overall transfer function is shown in Fig. Q.
8. The overall transfer function is
2.5( s + 1)
(a) × × ×
s( s + 2) ( s + 3) –3 –2 –1 0 Re
2.5( s + 2)
(b)
s( s + 1)( s + 3)
2.5( s + 3)
(c)
s( s + 1)( s + 2) Fig. Q. 8
s+3
(d)
2.5( s + 1)( s + 2)
9. The inverse Laplace transform of 1.5/s (s + 1) is
(a) (1 e 1.5t) (b) 1.5 (1 e 1.5t)
(c) 1.5 (1 e t) (d) (1.5 + e 1.5t)
10. The transfer function is defined as
(a) the ratio of Laplace transform of output to Laplace transform of input considering
initial condition as zero
(b) the ratio of Laplace transform of input to Laplace transform of output considering
initial conditions as zero
(c) the ratio of input to output
(d) the ratio of output to input
11. The error detector element in a control system gives
(a) the sum of the reference signal and feedback signal
(b) the sum of the reference signal and error signal
(c) the difference of the reference signal and feedback signal
(d) the difference of the reference signal and output signal
12. Unit impulse response of a system in Laplace transform form gives
(a) transfer function (b) system gain
(c) unit step function (d) unit ramp function
13. A system is represented by the differential equation
d2 x dx
M +F + Kx = u(t)
dt
2 dt
The transfer function relating X(s) and U(s) is
APPENDIX II : OBJECTIVE TYPE QUESTIONS 697
M F
(a) 2
(b) 2
Ms + Fs + K Ms + Fs + K
K 1
(c) 2 (d) 2
Ms + Fs + K Ms + Fs + K
d 2 c(t) 5dc(t)
14. The equation governing a control system is given by + + 4c(t) = 3r(t)
dt
2 dt
The transfer function for the system is
C(s) 3 C(s) 5
(a) = (b) =
R(s) (s + 1)(s + 4) R(s) (s + 1)(s + 4)
C ( s) 1 C(s) 4
(c) = (d) =
R( s ) s 2 + 3 s + 4 R(s) s2 + 3s + 5
15. The transfer function relating output I(s).... and input E(s) in a series RLC circuit is
given by
Cs Cs
(a) 2
(b) 2
(s LC + RCs + LC) (s LC + RCs + 1)
1 s Cs
(c) . (d)
C (s2 LC + RCs + 1) 2
(s + LCs + R)
16. The transfer function relating θ(s) and E(s) for the figure given below
N1
E(s) 1
R f + sL f K
J, f
q
N2
is
K ( N1 ) K
(a) (b)
s( Rf + sLf )( Js + f ) ( N 2 ) 2 2
s( Rf + sLf )[ J ( N1 / N2 ) s + f ( N1 / N 2 ) ]
K ( N2 ) K
(c) (d)
2
s(sRf + sLf )( Js + f )( N1 ) s( Rf + sLf )[ J ( N1 / N2 )s + f ( N1 / N2 )]
19. The generator constant Kg for a generator driven at 1000 RPM is 50 V/A; if the
generator is made to run at 1500 RPM the new value of the generator constant will
be
1
(a) 75 V/A (b) 33 2
V/A
(c) 150 V/A (d) 100 V/A
20. The transfer function relating E and R for the block diagram given below :
G2
E +
R + + C
G1
–
is
1 1
(a) (b)
1 + (G1 + G2 ) H 1 + G1G2 H
G1G2 G1G2
(c) (d)
1 + (G1 + G2 ) H 1 + G1G2 H
21. Equivalent block diagram for the figure given below :
R +
G C
–
R + C R + C
is G G
– –
(a) (b)
R G
+ C R G
+ C
+ +
1/G G
(c) (d)
APPENDIX II : OBJECTIVE TYPE QUESTIONS 699
22. Indicate the correct equivalent of the block diagram given below :
G1
G2
G1 G1 G1 G1G2
G1/G2 G1 + G2 G1G2
G2
+
R + + C
G1
– – +
is
G1 + G2 G1G2
(a) (b)
1 + G1G2 + G1 H + G2 H 1 + G1 + G2 + G1 H + G2 H
G1G2 G1 + G2
(c) (d)
1 + G1G2 + G1 H + G2 H 1 + G1 + G2 + G1 H + G2 H
24. The transfer function for the network given below :
is
1 R
(a) (b) RCs + 1
RCs + 1 ei C eo
1 R
(c) (d) s+1
R C
s+1
C
25. C/R for the block diagram shown below :
H2
R + + – C
G1
– –
H1
H3
is
G1 G1
(a) (b)
1 + G1 H1 + G1 H2 + G1 H3 1 + G1 H1G1 H2 + G1 H3
700 LINEAR CONTROL SYSTEMS
1 1
(c) (d)
G1 (G1 H1 + G1 H2 + G1 H3 ) G1 (G1 H1G1 H2 + G1 H3 )
26. The block diagram given below :
G2
R + + C
G1 G3
–
is equivalent to
–
r(t) + + + c(t)
4
– –
5 10
G2
+
R +
G1
+ C
28. –
H1
The signal flow graph for the system whose block diagram shown above will be
G2
G2
R C R C
(a) (b)
G1 1 G1
–H1 –H1
APPENDIX II : OBJECTIVE TYPE QUESTIONS 701
G2 G2
R C R C
(c) (d)
G1 1 G1 1
H1 H1
29. The number of closed-loops in the signal flow graph shown below :
R C
is
(a) 6 (b) 8
(c) 7 (d) 5
30. The number of forward paths in the signal flow graph shown below :
R C
is
(a) 1 (b) 2
(c) 3 (d) 5
31. The graph determinant for the signal flow graph shown below :
f g
R a b c d C
e h
is
(a) ∆ = f ce dh (b) ∆ = 1 + f + ce + dh + fce + fdh
(c) ∆ = 1 f ce dh (c) ∆ = 1 f ce dh + fce + fdh
32. The overall transfer function for the signal flow graph given below :
H1
R G1 G2 1
C
–H2
702 LINEAR CONTROL SYSTEMS
is
G1G2 G1G2
(a) (b)
1 − G1G2 H1 + G2 H2 1 + G1G2 H1 + G2 H2
G1G2 G1G2
(c) (d)
1 − G1G2 H1 − G2 H2 1 + G1G2 H1 + G2 H1
33. Figure given below represents unit step response; the time constant of the system is
(a) 0.005 sec (b) 2 sec
(c) 0.02 sec (d) 0.5 sec
0.02 c(t)
c(t)
t
0.02 sec
34. The steady value of the output on application of unit step input to transfer function
C(s) 2.5 , will be
=
R(s) s2 + 2.5s + 10
1
(a) (b) 1
2.5
(c) 0.25 (d) 2.5
100
35. A closed-loop control system has G(s) = and H(s) = (s + 1)
s2
The steady state output for a unit step input is
(a) 1.0 (b) 1/100
(c) 1/10 (d) 10
C(s) 16
36. For the system =
R(s) s2 + 8 s + 16
The nature of the time response will be
(a) overdamped (b) underdamped
(c) critically damped (d) none of these
C(s) 25
37. For the system = the damping factor ζ and damped frequency of
R(s) s2 + 6 s + 25
oscillations ωd will be
(a) ζ = 0.6, ωd = 5 (b) ζ = 0.4, ωd = 6
(c) ζ = 0.5, ωd = 3 (d) ζ = 0.3, ωd = 5
38. Which one of the following transfer function will have greatest maximum overshoot
2 16
(a) 2
(b) 2
s + 2s + 9 s + 2s + 16
APPENDIX II : OBJECTIVE TYPE QUESTIONS 703
25 36
(c) (d)
s2 + 2s + 25 2
s + 2s + 36
39. The location of the roots of a second order system is given in figure below, the system
has an actual damping
jw
3 + j4
3 – j4
(a) 3 (b) 5
(c) 4 (d) 2
40. The expression given below is the impulse response of a feedback control system
c(t) = e 0.6t sin 0.8t
The damping ratio and natural frequency of oscillations are
(a) 0.6, 1 rad/sec (b) 1, 0.6 rad/sec
(c) 0.8, 0.6 rad/sec (d) 1, 0.8 rad/sec
C ( s) 800
41. The transfer function of a control system is given by = can
R( s) s( s + 1)( s + 20)( s + 25)
be approximated as
C ( s) 3.2
(a) C (s) = 40 (b) =
R( s) s( s + 1)( s + 25) R( s) s( s + 1)( s + 20)
C(s) 1.6 C(s) 80
(c) = (d) =
R(s) s(s + 1) R(s) (s + 20)(s + 25)
42. The closed-loop poles for a system having open-loop transfer function G(s) H(s)
= K/s (s + 2) at K = 2 are
(a) 1 ± j1 (b) 1 ± j1
(c) 2 ± j1 (d) 2 ± j1
43. To a fair approximation the unit step time response for the system
C(s) 4
=
R(s) (s + 100)(s2 + 2s + 4)
is
0.5t ⎛ 1 0.75 ⎞
(a) e sin ⎜ 2 0.75t + tan
⎝ 0.5 ⎟⎠
(b) 1 e −2 t ⎛ 1 0.75 ⎞
sin ⎜ 2 0.75t + tan
0.75 ⎝ 0.5 ⎟⎠
704 LINEAR CONTROL SYSTEMS
⎛ 0.75 ⎞
(c) e−2 t sin ⎜ 2 0.75t + tan 1
⎝ 0.5 ⎟⎠
0.5 t
e ⎛ 0.75 ⎞
(d) 1 sin ⎜ 2 0.75t + tan 1
0.75 ⎝ 0.5 ⎟⎠
44. The overall transfer function of a system is given below :
C(s) 2.5(s + 2)
=
R(s) 5(s + 1)(s + 3)
The corresponding differential equation is
(a) c + 3c + 4 c = 0 (b) c + 4 c + 2c = r + r
d2 x dx
47. A control system is defined by the mathematical relationship +6 + 5x = 12u(t)
dt
2 dt
The response of the system as t → ∞ is
(a) x = 6 (b) x = 2
(c) x = 2.4 (d) x = 0.2
4
48. A unity feedback control system having G(s) = is subjected to a unit step
s (s + 2.4)
input. The expression for e(t) is given by
1 ⎞ −1.2 t −1 4 ⎞
(a) e(t) = ⎛⎜ ⎟e sin ⎛⎜ 1.6 t + tan ⎟
⎝ 0.8 ⎠ ⎝ 3⎠
1 ⎞ −1.2 t −1 4 ⎞
(b) e(t) = ⎛⎜ − ⎟e sin ⎛⎜ 1.6t + tan ⎟
⎝ 0.8 ⎠ ⎝ 3⎠
1 ⎞ −1.6 t −1 4 ⎞
(c) e(t) = ⎛⎜ ⎟e sin ⎛⎜ 1.2t + tan ⎟
⎝ 0.6 ⎠ ⎝ 3⎠
1 ⎞ −1.6 t −1 4 ⎞
(d) e(t) = ⎛⎜ − ⎟e sin ⎛⎜ 1.6 t + tan ⎟
⎝ 0.6 ⎠ ⎝ 3⎠
APPENDIX II : OBJECTIVE TYPE QUESTIONS 705
49. The impulse response of the system
C ( s) 8
= is
R( s) s( s + 2)( s + 4)
(a) c(t) = 2 e 2t + e 4t (b) c(t) = 1 + 2e 2t 4e 4t
2t
(c) c(t) = 1 2e + e 4t (d) c(t) = 2 + e 2t 4e 4t
50. The damping ratio of a second order control system is increased from 0.3 to 0.6, the
corresponding change in the % overshoot is
(a) 37.08% to 9.47% (b) 9.47% to 37.08%
(c) 60% to 30% (d) 20% to 50%
51. For a second order system given that ωn = 4 rad/sec and ζ = 0.8. The settling and peak
times are
(a) 1.3 s, 1.25 s (b) 1.25 s, 1.3 s
(c) 1.3 s, 2.5 s (c) 2.5 s, 3.1 s
52. The unit step response of a feedback control system is given by c(t) = 1 1.25 e 3t sin
(4t + 53.1°). The closed-loop transfer function is
C(s) 2.5
(a) = (b) C(s) = 2.5
R(s) s2 + 6 s + 2.5 R(s) s2 + 5s + 25
C(s) 25
(c) = (d) C(s) = 25
R(s) s2 + 10 s + 25 R(s) s2 + 6 s + 25
53. Insertion of negative feedback in a control system affects :
(a) the transient response to vanish uniformly
(b) the transient response to decay very fast
(c) no change in transient response jw
0 1 2 t (sec)
1 s
(c) lim (d) lim
G(s) H (s)
s→0 s → 0 G( s) H ( s)
66. The steady state error for a type 2 system subjected to unit ramp input is
(a) 1 (b) ∞
1
(c) (d) 0
K
67. Finite steady state error
(a) varies inversely with K (b) varies directly with K
(c) is independent of K (d) none of the above
K
68. The type of the transfer function, G(s) H(s) = is
s3 + 2 s 2 + 3 s
(a) 1 (b) 2
(c) 3 (d) 4
100
69. A unity feedback system having transfer function, G(s) = is subjected to
s(0.1s +1)
unit ramp input. The steady state error will be
(a) 1 (b) 0
(c) 0.01 (d) 0.05
70. With the increase of type of a system the steady state error for a particular input
function
(a) increases (b) decreases
(c) remains changed (d) first increases then decreases
71. A type 1 system is subjected to acceleration input signal leads to steady state error as
(a) 1 (b) K
1
(c) (d) ∞
K
72. Derivative feedback control
(a) increases rise time (b) increases overshoot
(c) decreases steady state error (d) does not affect the steady error
73. Which one of the following transfer function exhibit least steady state error for ramp
input
9 16
(a) 2
(b)
s + 2s + 9 s2 + 2s + 16
708 LINEAR CONTROL SYSTEMS
25 36
(c) 2
(d) 2
s + 2s + 25 s + 2s + 36
74. The time response of a system is given by
⎡ e−1.2 t
⎛ 1 4⎞ ⎤
c(t) = 1 − ⎢⎣sin ⎝ 1.6t + tan 3 ⎠ ⎥⎦
0.8
characteristic equation of the system is
(a) s2 + s + 0.8 = 0 (b) s2 + 2.4s + 4 = 0
(c) s2 + 4s + 2.4 = 0 (d) s2 + s + 4 = 0
75. Which of the following complex roots relate to greatest value of ωn
(a) s1, s2 = 1 ± j1 (b) s1, s2 = 1 ± j2
(c) s1, s2 = 4 ± j3 (d) s1, s2 = 4 ± j2
76. Which one of the following complex roots relate to least value of ωd
(a) s1, s2 = 2 ± j2 (b) s1, s2 = 4 ± j3
(c) s1, s2 = 1 ± j4 (d) s1, s2 = 0.5 ± j2.5
77. ωd and settling time are related as
4 1 − ζ2 ζ
(a) ωd = ts . (b) ωd =
ζ ts 4 1 − ζ 2
2
1 4 1−ζ tsζ
(c) ωd = . (d) ωd =
ts ζ 2
4 1−ζ
78. The roots of a closed-loop characteristic equation of the system shown below :
R(s) + 24 C(s)
– s + 10
1/s
are :
(a) 4, 10 (b) 4, 6
(c) 4, + 6 (d) 4, + 10
79. For a unity gain open-loop pole zero configuration is shown below :
jω
– |+j|
–3 σ
– |+j|
1 + Kts
c(t)
(a) (b) c(t)
t t
c(t) c(t)
(c) (d)
t t
95. The stability of a system
(a) decreases as the type of system increases
(b) increases as the type of system increases
(c) does not change as the type of system increases
(d) none of the above
96. A second order characteristic equation having all the coefficients of same sign
represents a system which is
(a) inherently stable (b) marginally stable
(c) inherently unstable (d) conditionally stable
APPENDIX II : OBJECTIVE TYPE QUESTIONS 711
97. The Routh-Hurwitz criterion gives
(a) relative stability (b) absolute stability
(c) gain margin (d) phase margin
98. The Routh array for the characteristic equation of a system is given below
s3 T1T2 1
(T1 + T2 ) K
s2
(T1 + T2 ) − KT1T2
s1 (T1 + T2 )
0
s K
the system is stable if
1 1 1 1
(a) K < + (b) K = +
T1 T2 T1 T2
1 1
(c) K > + (d) K = T1 + T2
T1 T2
99. For the Routh array given below :
s3 1 3
4 K
s2
(12 − K)
s1
4
s0 K
the system is unstable if
(a) K > 12 (b) K = 12
(c) K > 39 (d) K < 3
100. The location of the closed-loop conjugate pair of pole on jω axis indicates that the
system is
(a) stable (b) unstable
(c) marginally stable (d) critically stable
101. A closed-loop control system has a pair of complex pole of multiplicity 2 located on the
imaginary axis of the s-plane. This indicates
(a) stability (b) unstability
(c) absolute stability (d) relative stability
102. As the type of a open-loop transfer function increases the stability
(a) increases (b) decreases
(c) does not change (d) none of (a), (b) and (c)
103. A second order control system with ξ < 0 is always
(a) stable (b) marginally stable
(c) unstable (d) inherently stable
104. A second order control system with ξ = 0 is always
(a) marginally stable (b) stable
(c) unstable (d) none of these
712 LINEAR CONTROL SYSTEMS
is given by +0
(a) N = (Z P) s
(b) N = (P Z) –0
(c) N = 2 (P Z) –¥
(d) N = 2 (Z P)
111. The number of open-loop poles having positive real Imj
part is 1 and the Nyquist plot is drawn in figure :
The value of K for the closed-loop system to be K/2
w=+0 w=–¥
stable is
w= –0 w = + ¥ Re
(a) K > 1 (b) K < 1
(c) K < 0.5 (d) K = 0.5
112. In G(s) H(s)-plane the Nyquist plot from ω = 0 to ω = + 0 is closed through an angle
(a) nπ (b) 2nπ
nπ
(c) (d) nπ
2
where n is the number of open-loop poles at the origin.
APPENDIX II : OBJECTIVE TYPE QUESTIONS 713
113. The Nyquist plot for a control system is shown in figure : Imj
the type of the system is w = +0
(a) 0 (b) 1
(c) 2 (d) 3
114. In problem no. 100, if the number of poles having w=+¥
w=–¥ Re
positive real parts is nil, then the number is – +j0
encirclements of the critical point is
–0
(a) 2 (b) 2 w=
(c) 2 (d) 3
115. The origin for the investigation of closed-loop stability
in relation to Nyquist criterion is
(a) 1 + j0 (b) + 1 + j0
(c) E0^ + ^j1 (d) 0 j1
116. The Nyquist plot for a system is having the number of open-loop positive poles as 1 is
shown below :
Imj
w=+0 w=–¥
r p
S w=–0 q w=+¥ Re
w=–0 w=+¥
w=+0 w=–¥ Re
is
(a) 0 (b) 1
(c) 2 (d) 4
118. The open-loop transfer function of a unity feedback control system is given as,
1000
G(s) = the phase crossover frequency ω2 is
s(s + 10)2
(a) 1000 rad/sec (b) 500 rad/sec
(c) 10 rad/sec (d) 5 rad/sec
714 LINEAR CONTROL SYSTEMS
119. While increasing the value of gain factor K the system becomes :
(a) less stable (b) more stable
(c) unstable (d) absolutely stable
120. Nyquist criterion enables to determine
(a) absolute stability (b) relative stability
(c) both (d) either (a) or (b)
jw
121. The Nyquist plot is shown in figure, what will be the type of the
system w=0
(a) 0 (b) 1
(c) 2 (d) 3 s
122. At gain factor K = 0, the roots of the characteristic equation are
placed at
(a) open-loop poles (b) open-loop zeros
(c) both (a) and (b) (d) none of these position
123. The order of the differential equation governing a
system having its polar plot is shown in figure :
is
w=¥
(a) 3 (b) 2 s
(c) 1 (d) 0 w=0
124. The polar plot for a transfer function is shown in figure : Imj
An addition of a zero will change the plot as
w=+¥
Re
Imj Imj w
w
w=+¥
Re
(a) w=+¥ Re (b)
Imj
Imj
w
w=+¥ w=+¥
(c) Re
(d) Re
w
APPENDIX II : OBJECTIVE TYPE QUESTIONS 715
125. The gain margin for a system whose Nyquist plot drawn below :
Imj
w=–0
0.5
w=+¥
w=–¥ Re
–1 + j0
w=+0
is
(a) 0.5 db (b) 1.5 db
(c) 5 db (d) 6 db
126. The phase margin in relation to following polar plot is
Imj
j1
2
w=+¥
–1 Re
Ö3
2
w=+¥
–1 + j0 Re
o t
o t
716 LINEAR CONTROL SYSTEMS
o t o t
Imj
128. The Nyquist plot is shown in figure. The open-loop
transfer function of the feedback control system is
s+2
given below, G(s) H(s) = 2
. The number of + ve w=¥ w=0
s (s − 4) 0
w=–¥
Re
– 1 + j0 w=–0
real part roots of the characteristic equation is
(a) nil (b) 3
(c) 1 (d) 2
129. The open-loop transfer function of a feedback control system is given below, G(s) H(s)
e − Ts
= . ∠ G(jω) H(jω) = 0 occurs at
s( s + 2)
(a) ω = 2 tan ωT (b) ω = 2 cot ωT
cot ω tan ωT
(c) ω = (d) ω =
2T 2T
Imj
130. The order and type of the system having its Nyquist
plot shown in figure are
(a) 3, 2 0 w=–0
(b) 3, 0 w=0 Re
(c) 2, 1
(d) 2, 0
131. The open-loop transfer function of a feedback control system is given below, G(s) H(s)
K p>0
= ; and p ≠ q. For stability
(s + p)(s + q) q > 0
pq
(a) 0 < K < (b) 0 < K < pq(p + q)
p+ q
p p+q
(c) 0 < K < q (p + q) (d) 0 < K < Imj
pq
132. With reference to polar plot shown in figure. The gain w=¥
margin is –3 –1 0 Re
(a) 9.55 db
(b) 9.55 db
(c) 5.95 db
(d) 5.95 db w=0
APPENDIX II : OBJECTIVE TYPE QUESTIONS 717
133. The part of Nyquist plot for the open-loop Imj
transfer function of a feedback control
system is shown in figure. The gain and
phase margins are, 0 w=0
w=¥ 160° Re
(a) 6 db, 120°
– 0.5 1
(b) 2 db, 60°
(c) 6 db, 30°
(d) 6 db, 90°
134. The shape of the Nyquist plot is shown in Imj
figure. It is known that the number of
open-loop poles having + ve real part is
two. For the system to be stable, the range – 2K
of positive values of K being w = –¥ w=0
0 w=¥ w=–0 Re
(a) K > 0.5
(b) K < 0.5
(c) K = 5
(d) K > 2
135. A part of the Nyquist plot is shown in figure, Imj
the corresponding transfer function is
(a) K
2 0 Re
s (sT1 + 1)
w=–¥
K w=–0
(b)
s(sT1 + 1)
K
(c)
s(sT1 + 1)(sT2 + 1)
K
(d) 3
s (sT1 + 1)
Imj
136. Consider the Nyquist plot shown in figure,
which one of the following transfer
functions represent this plot
0 w=0
1 w=¥ Re
(a) 1
(s + 2)2 4
1
(b) 2
s + 3s + 2
1
(c)
(s + 2)3
1
(d)
(s + 2)
137. A unit feedback control system has the following open-loop transfer function
1
G(s) = . The phase margin is
(s + 2)2
718 LINEAR CONTROL SYSTEMS
π
(a) π rad. (b) rad.
3
π
(c) − rad. (d) π rad.
2 2
138. The shape of gain-phase plot corresponding to Nyquist plot given below :
Imj
w=–0
w=+¥
–1 + j0 w=–¥ Re
w=+0
is obtained as
w=+0 w
w
|G(jw)| db
|G(jw)| db
(a) 0 (b) 0
w w
|G(jw)| db
|G(jw)| db
(c) 0 (d) 0
– 210°
– 180°
– 150°
ÐG(jw)
(d) 2 db, 30°, 10 rad/sec, 10 rad/sec
Imj
142. As observed from the polar plot is shown in figure
(a) G.M. is + ve, P.M. is + ve w=+¥
(b) G.M. is ve, P.M. is ve –1 + j0 Re
(c) G.M. is ve, P.M. is + ve
(d) G.M. is + ve, P.M. is ve
w
143. The initial slope of the Bode plot for a transfer function having no poles at the origin is
(a) 10 db/decade (b) 0 db/decade
(c) 10 db/decade (d) 24 db/decade
144. The initial slope of the Bode plot for a type 2 system intersects 0 db axis at
(a) ω = 0 (b) ω = K
(c) ω = K (d) ω = K2
145. The error at the corner frequency due to the term (1 + jωT)± N is
(a) ± 5 N db (b) ± 3 db
(c) ± 6 db (d) ± 3 db
146. For a stable system
(a) G.M. and P.M. both are positive
(b) G.M. and P.M. both are negative
(c) G.M. is positive and P.M. is negative
(d) G.M. is negative and P.M. is positive
147. The gain for constructing the Bode plot in relation to transfer function
20
G(s) H(s) = is
s( s + 2)
(a) 20 (b) 10
(c) 2 (d) 40
148. The gain for drawing the Bode plot for the transfer function
320 (s + 2)
G(s) = 2
s(s + 1)(s + 8 s + 64)
(a) 10 (b) 5
(c) 16 (d) 160
720 LINEAR CONTROL SYSTEMS
– 6 db/oct.
60 db
– 6 db/oct.
0 db
1 2.5 5 12.5 ω
161. The phase crossover frequency for the open-loop transfer function
20
G(s) H(s) = , is
(s + 1)(s + 12 )(s + 13 )
(a) 1.0 rad/sec (b) 2.0 rad/sec
(c) 2 rad/sec (d) 3 rad/sec
162. The Bode plot of a feedback control system has a constant slope of 20 db/decade for
all frequencies and a fixed phase angle 90°. The phase margin is
(a) 90° (b) 0°
(c) 90° (d) 180°
163. The Bode plot shown in figure
is for |G(jw)| db
(a) pure integration –2
0d
b/d
(b) pure differentiation ec.
K
164. For the transfer, G(s) H(s) = . Asymptotic Bode plot is shown below :
⎛1 ⎞
⎜⎝ s + 1⎟⎠
b
|M| db
20 log10 K
0 b 10 b log10 w
ÐG(jw) |G(jw)|
20 db – 90°
– 180°
1/T 10/T 100/T
is for a
(a) non-minimum phase transfer function
(b) minimum phase transfer function
(c) all pass transfer function
(d) type 2 transfer function
724 LINEAR CONTROL SYSTEMS
–2
0d
db b /d
–2
0d
b /d
0.1 0.2 1 4 5 ω
–
40
db
db /d
– 20
db/d
2 5 10 – w
40
db
/d
is
25(1 + 0.5s) 5(1 + 0.5s)
(a) 2
(b)
s (1 + 0.1s) s2 (1 + 0.1s)
25(1 + 0.s) 25(1 + 2s)
(c) (d)
s(1 + 0.1s) s2 (1 + 10 s)
181. For the Bode plot shown below
–2
0d
20 db b /d |G(jw)|
90°
10 db
ÐG(jw)
0 db 180°
w
–4
0d
b/d
188. As K approaches infinity the root loci are asymptotic to lines which make an angle
with real axis given by
2kπ (2 k + 1)π
(a) (b)
P−Z P−Z
(k + 1)π (2k + 1) π
(c) (d) .
P−Z P−Z 2
189. With reference to following characteristic equation of a feedback control system. The
centroid of the root locus plot
s3 + 2s2 + Ks + K = 0
(a) 0.5 (b) 0.5
(c) 1 (d) 1
190. The intersection of the asymptotes of root locus w.r.t. following open-loop transfer
K ( s + 1)
function G(s) H(s) = is given by
2
s( s −1)( s + 5s + 6)
(a) s = + 1 (b) s = 1
(c) s = 5 (d) s = 6
726 LINEAR CONTROL SYSTEMS
(a) (b)
(c) (d)
192. Which of the following RL plot indicates a stable system for all values of K.
(a) (b)
(c) (d)
193. The intersection point of RL branches with s-phase imaginary axis is determined
using
(a) Bode plot (b) Routh-Hurwitz array
G ( s) G(s)
(d) = 1 and ∠ = (2k + 1) π
1 + G ( s) H ( s) 1 + G(s) H (s)
204. If the root locus branches cross the imaginary axis the system becomes
(a) overdamped (b) critically damped
(c) stable (d) unstable
205. If the root locus lies only on the negative real axis then the time response is
(a) overdamed (b) underdamped
(c) oscillatory (d) sustained oscillations
206. The intersection of the root locus branch with the imaginary axis is determined by
using equation
(a) G(s) H(s) = 0 (b) 1 G(s) H(s) = 0
(c) 1 + G(s) H(s) = 0 (d) [1 + G(s) H(s)] 1 = 0
207. The asymptotes corresponding to root locus plot for the open-loop transfer function of
K
a unity feedback control system G(s) = . Cross the imaginary axis at
s( s + 5)
(a) ω = ± 2 (b) ω = 0
(c) ω = ± 5 (d) never cross.
208. The root locus plot is shown below in the figure.
Imj
¥ –1
K
0 K=0 Re
K=0
APPENDIX II : OBJECTIVE TYPE QUESTIONS 729
The corresponding open-loop transfer function is
K K
(a) (b)
2
s (s + 1) (s + 1)(s + 2)
K K
(c) 3
(d) 2
(s + 1) (s + 1)(s + 2)
209. The open-loop transfer function of a unity feedback control system is given by
K (s + 2)
G(s) = . The root locus lies on the real axis between
s(s + 4)(s + 5)
(a) s = 2 and s = 4 (b) s = 4 and s = 5
s = 5 and s → ∞ s = 0 and s = 2
(c) s = 0 and s = 2 (d) s = 4 and s = 5
210. The open-loop transfer function of a feedback control system is given by G(s) H(s) =
K
2 ; 0 < K < ∞ . The number of breakaway points on the root locus
s (s + 4)(s + 4 s + 13)
plot is /are
(a) Nil (b) one
(c) two (d) three
211. The angle of departure of the root locus from the complex open-loop pole ( 1 j1),
215. If the root locus branches do not cross the imaginary axes the system is
(a) conditionally stable (b) inherently stable
(c) maginally stable (d) likely to be unstable
216. The value of K at which the root locus crosses the imaginary axis makes the system
(a) stable (b) underdamped
(c) marginally stable (d) unstable
217. The number of root loci branches for the following open-loop transfer function
K (s + 2)(s + 4)
G(s) H(s) = , is
s (s + 1)(s + 3)(s2 + 5s + 6)
(a) 5 (b) 4
(c) 3 (d) 2
218. The transfer function of an integral compensator is given by
1 1
(a) (b)
s s2
K
(c) (d) Ks
s
219. While incorporating lead-compensation network the gain factor of the amplifier is
1
(a) α (b)
α
1
(c) α2 (d)
α2
220. The lag-compensation has a
(a) zero nearer to the origin (b) pole nearer to the origin
(c) pole at the origin (d) zero at the origin
221. The lag-compensator
(a) decreases the bandwidth (b) increases the bandwidth
(c) does not affect the bandwidth (d) increases steady state error
222. The open-loop transfer function of a system is
10(1 + 0.25 sec)
G(s) H(s) =
s (1 + 0.6 s)
the phase shift as ω → 0 and ω → ∞ will be
(a) 0° and 90° (b) 180° and + 180°
(c) + 90° and 90° (d) 0° and 0°
223. The transfer function given below
1 + R2C2 s
Gc (s) = , represents
1 + ( R1 + R2 ) C2 s
(a) lead-network (b) lag-network
(c) lag-lead network (d) feedback network
224. The phase lag-lead network shifts the phase of a control signal in order that the
output phase
APPENDIX II : OBJECTIVE TYPE QUESTIONS 731
(a) lags at low frequencies and leads at high frequencies
(b) leads at low frequencies and lags at high frequencies
(c) becomes independent of frequencies
(d) leads at all frequencies
225. A lead compensation needed for a closed-loop system has the following transfer
function
⎛ 1 ⎞
K ⎜ 1 + s⎟
⎝ a ⎠
Gc(s) = , for such a network
⎛ 1 ⎞
⎜⎝ 1 + s⎟
b ⎠
(a) a < b (b) b < a
(c) a > Kb (d) b < Ka
226. The lead compensation network is considered to be
(a) high pass filter (b) low pass filter
(c) equalizer (d) none of the above
227. The transfer function of a lead compensation network is given below :
α(1 + sT )
Gc (s) =
(1 + sαT )
The mid-corner frequency is given by
1 1
(a) ωm = (b) ωm =
αT T a
1
(c) ωm = αT (d) ωm =
α T
228. The maximum phase lead is contributed by a phase lead network at
(a) mid corner frequency (b) lower corner frequency
(c) upper corner frequency (d) phase corner frequency
229. Which one of the following gives the transfer function of a phase-lag compensation
network?
1 + sαT 1 + sαT
(a) ;α<1 (b) ;α>1
1 + sT 1 + sT
1 + sαT 1 − sαT
(c) ;α=1 (d) ;α<1
1 + sT 1 + sT
230. The analysis of multiple-input multiple-output system is conveniently studied by
(a) state-space approach (b) root locus approach
(c) characteristic equation approach (d) Nichols chart
231. The state transition matrix φ(t) is given by
(a) [sI] [A] (b) {[sI] [A]} 1
(c) L 1{[sI] [A]} 1 (d) L 1{[sI] [A]}
232. In exponential series form the state transition matrix is
(a) e [A]t (b) e[A]t
(c) e [A] (d) e [A]
732 LINEAR CONTROL SYSTEMS
.
233. For the state model [ x ] = [A][x] + [B]u, y = [c] [x] the transfer matrix is
(a) [C]{[sI] [A]} 1[B] (b) [C] {[sI] [A]} [B]
(c) [C] [B] {[sI] [A]} 1 (d) [C] [B] {[sI] [A]}.
⎡1 ⎤
234. For second order system with an initial condition of ⎢ ⎥ without any external input,
⎣ 2⎦
the state transition matrix for the system is given below :
⎡ e−2t 0 ⎤
φ(t) = ⎢ ⎥
⎢⎣0 e− t ⎥⎦
The state of the system at the end of one second is
⎡0.135⎤ ⎡0.135 ⎤
(a) ⎢ ⎥ (b) ⎢ ⎥
⎣0.735⎦ ⎣0.368 ⎦
⎡0.271⎤ ⎡0.135⎤
(c) ⎢ ⎥ (d) ⎢ ⎥
⎣0.735⎦ ⎣ 1.47 ⎦
235. The differential equation governing the dynamics of a separately excited d.c. motor is
given below :
d2θ
dθ K 2 K
J + 2
+f
θ= Va
dt dt L L
The above equation is arranged in the following form
⎡ d2θ ⎤
⎢ 2⎥ ⎡ dθ ⎤
⎢ dt ⎥ = [ A ] ⎢ dt ⎥ + [ B]Va
⎢ dθ ⎥ ⎢ ⎥
⎢ ⎥ ⎣⎢ θ ⎦⎥
⎣ dt ⎦
The value of coefficient matrix [A] is given by
⎡f K2 ⎤ ⎡ K2 f⎤
(a) ⎢ J ⎥ (b) ⎢ LJ ⎥
− −
⎢ J ⎥ ⎢ J⎥
⎢⎣ 1 0 ⎥⎦ ⎢⎣ 0 1 ⎥⎦
⎡ 0 1 ⎤ ⎡ 1 0 ⎤
(c) ⎢ K 2 f⎥
⎥ (d) ⎢
f K2 ⎥
⎥
⎢− − ⎢ − −
⎣⎢ LJ J ⎦⎥ ⎢⎣ J LJ ⎥⎦
236. With reference to state space representation of the equation,
d 2 y(t) 2dy(t)
2
+ + 3y(t) = u(t). The matrices A and B are,
dt dt
⎡ 0 1⎤ ⎡0 ⎤ ⎡ − 2 − 3⎤ ⎡0 ⎤
(a) A = ⎢ ⎥ ;B= ⎢ ⎥ (b) A = ⎢ ⎥ ;B = ⎢ ⎥
⎣ − 3 − 2⎦ ⎣1 ⎦ ⎣ 0 1⎦ ⎣1 ⎦
⎡0 − 2 ⎤ ⎡1 ⎤ ⎡− 2 0⎤ ⎡1 ⎤
(c) A = ⎢ ⎥ ;B=⎢ ⎥ (d) A = ⎢ ⎥ ;B=⎢ ⎥
⎣1 − 3 ⎦ ⎣0 ⎦ ⎣ 1 − 3⎦ ⎣0 ⎦
APPENDIX II : OBJECTIVE TYPE QUESTIONS 733
⎡i ⎤
x ⎡− 1 0⎤ ⎡ x1 ⎤ ⎡0 ⎤ ⎡ x1 ⎤
237. A control system is represented by ⎢ 1 ⎥ = ⎢ u and y = [1 1]
⎢ i ⎥ ⎣ 0 − 3⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦ ⎢x ⎥
+
⎣ 2⎦
⎢⎣ x2 ⎥⎦
The system transfer function is
s 1
(a) (b)
s+1 s+3
1 s
(c) (d) 2
s+1 s + 4s + 3
⎡i ⎤
x ⎡− 1 0 ⎤ ⎡ x1 ⎤ ⎡0⎤
238. The state model is given below, ⎢ 1 ⎥ = ⎢ u. If the input is unit step,
⎢ i ⎥ ⎣ 0 − 2⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦
+
⎢⎣ x2 ⎥⎦
the time response is
(a) undamped (b) underdamped
(c) critically damped (d) overdamped
i ⎡0 2⎤
249. An atonomous state euaqtion is [ x ] = [A] [x] where, [A] = ⎢ ⎥ . The polers are
⎣8 0 ⎦
located at
(a) j4, j4 (b) 4, 2
(c) j2, j2 (d) 8, 4
240. The characteristic equation for a system having state equation
⎡i ⎤
⎢ x1 ⎥ = ⎡ 1 4 ⎤ ⎡ x1 ⎤ ⎡0⎤
u is
⎢ i ⎥ ⎢⎣ − 2 − 5⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1 ⎥⎦
+
⎣⎢ x2 ⎦⎥
(a) s2 + 5s + 8 = 0 (b) s2 + 4s + 3 = 0
(c) s2 + 4s + 5 = 0 (d) s2 + 4s + 2 = 0
⎡i ⎤
x ⎡ 2 0 ⎤ ⎡ x1 ⎤ ⎡1⎤
241. For the system ⎢ 1 ⎥ = ⎢ r(t)
⎢ i ⎥ ⎣0 4 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣1⎥⎦
+
⎢⎣ x2 ⎥⎦
⎡ x1 ⎤
y = [4 0] ⎢ ⎥
⎣ x2 ⎦
where r(t) is a unit impulse function. Having initial conditions as nil, the output y is
determined as
(a) y(t) = 2e2t (b) y(t) = 4e2t
(c) y(t) = 2e4t (d) y(t) = 4e4t
242. For the homogenous state equation
⎡i ⎤
⎢ x1 ⎥ = ⎡ − 3 1⎤ ⎡ x1 ⎤
⎢ i ⎥ ⎢⎣ 0 − 2⎥⎦ ⎢⎣ x2 ⎥⎦
⎢⎣ x2 ⎥⎦
734 LINEAR CONTROL SYSTEMS
⎡ x1 (0)⎤ ⎡ 2⎤
with ⎢x (0)⎥⎦ ⎢⎣ 1⎥⎦
=
⎣ 2
The steady state value of x is
⎡0 ⎤ ⎡ − 2⎤
(a) ⎢ ⎥ (b) ⎢ ⎥
⎣0 ⎦ ⎣ − 1⎦
⎡ − 1⎤ ⎡0 ⎤
(c) ⎢ 3⎥ (d) ⎢ ⎥
⎣ ⎦ ⎣1 ⎦
243. For the matrix
⎡ 1 4⎤
[A] = ⎢
⎣ − 2 − 5⎥⎦
The eigen values are
(a) λ1 = 1 ; λ2 = 3 (b) λ1 = 1 ; λ2 = 4
(c) λ1 = 2 ; λ2 = 4 (d) λ1 = 1 ; λ2 = 2
⎡ 0 1 0⎤
244. [A] = ⎢ 1⎥⎥
⎢ 0 0
⎣⎢ − 6 − 11 − 6 ⎦⎥
⎡0 1 0 ⎤ ⎡ 1 1 1⎤
⎢ ⎥ ⎢ − 1 − 2 − 3⎥
(a) [P] = ⎢1 2 3⎥ (b) [P] = ⎢ ⎥
⎣⎢1 4 9⎥⎦ ⎢⎣ 1 4 9⎥⎦
⎡ 1 4 9⎤ ⎡− 1 0 0⎤
⎢ ⎥ ⎢ 0 −2 0⎥⎥
(c) [P] = ⎢ − 1 − 2 − 3⎥ (d) [P] = ⎢
⎢⎣ 1 4 9⎥⎦ ⎣⎢ 0 0 − 3⎦⎥
⎡1 0 ⎤
245. For a given state space representation A = ⎢ ⎥ . The state transition matrix is
⎣0 4 ⎦
⎡ e− t 0 ⎤ ⎡0 e− t ⎤
(a) φ(t) = ⎢ ⎥ (b) φ(t) = ⎢ ⎥
⎣⎢0 e−4 t ⎦⎥ ⎢⎣ e
−4 t
0 ⎥⎦
⎡ e−4 t 0 ⎤ ⎡0 e−4 t ⎤
(c) (t) =
φ ⎢ ⎥ (d) (t) = ⎢
φ ⎥
⎢⎣0 e− t ⎥⎦ ⎢⎣ e
−t
0 ⎥⎦
246. A system is described as, x = Ax + Bu. The controllability test matrix is given by
(a) ... : A2B : AB : B (b) B : AB : A2B : ...
2
(c) B : BA : BA : ... (d) ... BA2 : BA : B
247. In z-plane the unit circle corresponds to
(a) imaginary axis of s-plane (b) + real axis of s-plane
(c) real axis of s-plane (d) origin of s-plane
248. On transformation using z = esT, a point ( 1 + j2) in the s-plane is placed in z-plane
(a) inside the unit circle (b) outside the unit circle
(c) on the unit circle (d) cannot be placed in the z-plane
APPENDIX II : OBJECTIVE TYPE QUESTIONS 735
249. The faithful reconstruction of a signal on account of sampling is obtained if
(a) ωs = ω1 (b) ωs ≥ 2ω1
(c) ωs ≤ ω1 (d) ωs ≤ 2ω1
Answers
205. (a) 206. (c) 207. (d) 208. (a) 209. (b) 210. (d)
211. (a) 212. (a) 213. (c) 214. (a) 215. (b) 216. (c)
217. (a) 218. (c) 219. (b) 220. (b) 221. (a) 222. (a)
223. (b) 224. (a) 225. (a) 226. (a) 227. (b) 228. (a)
229. (a) 230. (a) 231. (c) 232. (a) 233. (a) 234. (a)
235. (c) 236. (a) 237. (c) 238. (d) 239. (a) 240. (b)
241. (b) 242. (a) 243. (a) 244. (b) 245. (a) 246. (b)
247. (a) 248. (a) 249. (b)
III
Short Answer Questions
One compulsory question having 10 or 20 parts usually appear in question papers. The
said compulsory question is invariably short answer type. An effort is made to make the
readers aware of this fact and as such short answer type questions are framed
hereunder. These type of questions will be of help towards viva-voce examinations and
job related interviews.
Feedback path
737
738 LINEAR CONTROL SYSTEMS
B(s)
The ratio = G(s) H(s) is known as open-loop transfer function.
E(s)
Q. 11. What is a block diagram ?
Ans. The symbolic representation of a transfer function is represented by a block diagram.
The complete control system can be represented by interconnecting block diagrams
of components as per the requirements.
Q. 12. Define non-touching loops.
Ans. The non-touching loops in a signal flow graph are those which do not share a common
node.
Q. 13. Write Mason s gain formula.
k
∑P
k=1
k ∆k
Ans. Overall gain (transmittance) =
∆
where P = Number of forward paths
Pk = Forward path transmittance of the kth path.
Graph determinant
∆ = 1 sum loop gains (transmittances)
+ sum of products of all possible pairs of non-touching loop gains
sum of products all possible triplets of non-touching loop gains
+ ........
∆k = Path factor associated with the kth path is equal to the graph
determinant of the signal flow graph which remains after
erasing the kth path.
Q. 14. What is the use of gear train in a control system ?
Ans. In a control system the load demands high torque at low speed while the motor
supplies low torque at high speed. In such a case a gear train is used having a gear
ratio (N2/N1) > 1. N1 is the gear on the motor side and N2 on the load side.
Q. 15. Show that in mks system KT = Kb.
Ans. The mechanical power output of the motor = TMωm
The electrical power input to the motor = ebia
Neglecting armature and rotational losses
TMωm = ebia
∵ TM = KTia and eb = Kbωm
∴ KT iaωm = Kbωmia, or KT = Kb.
Q. 16. Show how the transfer function of a motor is modified if a gear train is
attached to it.
Gear
Input θm(s) Output
Ans. Motor transfer N2
Vm(s) function N1 θo(s)
N2
= Gear ratio.
N1
740 LINEAR CONTROL SYSTEMS
Q. 17. State the assumption made while obtaining the transfer function of an
armature controlled d.c. motor.
Ans. The armature inductance La is neglected.
Q. 18. Write the transfer function of an armature controlled d.c. motor relating
angular displacement with applied voltage Va.
θ m (s) KT
Ans. = 2
Va (s) ( Ra + sLa ) (s J + sf ) + sK T K b
Neglect La in comparison with Ra
θ m (s) KT
∴ = .
Va ( s) sRa (sJ + f ) + sK T K b
Q. 19. Write equivalent moment of inertia and coefficient of friction with
reference motor side. The ratio is 20.
=JG
F N I = J FG 1 IJ2 2
H N JK H 20 K
1
Ans. Jeq
2
F N I = f FG 1 IJ .
=fG
2 2
H N JK H 20 K
1
feq
2
Q. 20. What is a techogenerator ?
Ans. The electromechanical device which generates the output voltage proportional to
shaft speed is called a tachogenerator.
Q. 21. Name the test signals used in time response analysis.
Ans. The commonly used test signals are unit (1) impulse (2) unit step (3) unit ramp
(4) unit acceleration.
Q. 22. What is meant by time response of a control system ?
Ans. The time response of a control means as to how the output of a system varies in
accordance with time when a specified test signal is applied.
Q. 23. Distinguish between the static response and dynamic response.
Ans. The static response corresponds to steady state part and dynamic response refers to
transient part of the time response of control system.
Q. 24. How the response of control system is determined ?
Ans. The time response denoted as c(t) gives the variation of output of a control system as
a function time on application of a specified input signal.
The closed-loop transfer function is given by
C(s) G(s)
=
R(s) 1 + G(s) H (s)
R(s) G(s)
∴ C(s) =
1 + G(s) H (s)
Taking inverse Laplace transform on both sides,
1 R(s) G(s)
C(t) = L .
1 + G(s) H (s)
APPENDIX III : SHORT ANSWER QUESTIONS 741
Q. 25. What is meant by characteristic equation ?
Ans. The characteristic equation is obtained by equating the denominator polynomial of
the overall transfer function C(s)/R(s) to zero.
For a control system,
C(s) G(s)
=
R(s) 1 + G(s) H (s)
The characteristic equation is
1 + G(s) H(s) = 0.
Q. 26. The open-loop transfer function is given below :
K
G(s) H(s) =
s(s + 2)
Write the characteristic equation.
Ans. The characteristic equation is given by
1 + G(s) H(s) = 0
K
∴ 1+ =0
s(s + 2)
or s2 + 2s + K = 0.
Q. 27. Draw the typical unit step time response of underdamped second order
control system.
Ans.
Mp
c(t) 0.02
tp ts t
∴ s2 + 4s + 6 = 0
− 4 ± 42 − 4 × 1 × 6
s1 , s2 = = 2 ± j 2.
2
Q. 30. Define settling time in relation to time response.
Ans. The settling time is the time required to settle the time response within a tolerance
band of steady state value. Usually the tollerance band is either 2% or 5%.
ts = 4T on 2% basis
ts = 3T on 5% basis
where T = time constant.
Q. 31. Define damping ratio.
Ans. The damping ratio ζ is defined as the ratio actual damping (ζωn) to critical
damping (1 × ωn).
Q. 32. Write the relationship between the % maximum overshoot (Mp) and
damping ratio ζ.
− ζπ
1 − ζ2
Ans. % Mp = e × 100, ζ < 1
The decrease in the value of damping ratio increases the maximum overshoot. For
ζ = 0, % Mp = 100%.
Q. 33. Write the expression for peak time (tp) in terms of ωn and ζ.
π
Ans. tp = .
ω n 1 − ζ2
Q. 34. Sketch the impulse response corresponding to closed-loop transfer
function poles shown below.
Ans.
jw c(t)
O s O t
×
Q. 35. A second order control system has a damping ratio ζ = 0.8 and natural
frequency of oscillation ωn = 12 rad./sec. Determine the damped frequency
of oscillation.
C(s) G(s)
Ans. =
R(s) 1 + G(s)
16
=
s2 + 6s + 16
ωn = 16 = 4 rad./sec.
2ζωn = 6
∴ 2ζ × 4 = 6, ζ = 0.75
ζ < 1, the response is underdamped oscillatory.
Q. 37. Mention the nature of transient response of second order control system
for different types of damping.
Ans. (1) Oscillatory (sustained oscillations) for ζ = 0
(2) Damped oscillatory (underdamped) for ζ < 1
(3) Exponentially rising (critical damping) for ζ = 1
(4) Exponentially rising (overdamped) for ζ > 1. The rise time is more as compared to
critical damping.
Q. 38. The closed-loop transfer function is given below :
C(s) 25
= 2
R(s) s + 4s + 25
Determine ωn and ζ.
Ans. The characteristic equation is
s2 + 4s + 25 = 0
ωn = 25 = 5 rad./sec.
2ζωn = 4
∴ 2ζ × 5 = 4
ζ = 0.4.
Q. 39. If the damping ratio is increased, how will it affect the settling time ?
4
Ans. The settling time ts = .
ζω n
The settling decreases as the damping ratio is increased (keeping constant ωn).
Q. 40. What is the value of damping ratio for an overdamped response?
Ans. ζ > 1.
Q. 41. What is the frequency of oscillation for underdamped system ?
Ans. ω d = ω n 1 − ζ2 , ζ < 1.
Q. 42. The characteristic equation of a system is given below :
(s + 2)(s + 3)(s2 + 2s + 2) = 0
Determine the dominant roots.
744 LINEAR CONTROL SYSTEMS
1 − 0.82
π − tan −1 π − tan −1 0.75 π − 0.643
= 0.8 = = = 0.346 sec.
12 1 − 0.82 12 × 0.6 7.2
Q. 47. What is a dominant pole pair ?
Ans. The closed-loop L.H.S. complex pole pair nearer to s-plane imaginary axis is known
as dominant pole pair. These poles dominate the time response while the other
L.H.S. poles far away from the s-plane imaginary axis have less effect on the time
response.
k
Q. 48. What is the order of the system whose transfer function is G(s) = ?
s 2 (s + 2)
Ans. The total number of poles of the transfer function is 3, hence the order of the
system is 3.
Q. 49. Distinguish between the type and order of a system transfer function.
Ans. (i) The number of poles at the origin gives the type of a transfer function.
(ii) The total number of poles give the order of a transfer function.
APPENDIX III : SHORT ANSWER QUESTIONS 745
Q. 50. What is static positional error coefficient for a type 0 system ?
Ans. The static positional error coefficient is associated with unit step input and given by
Kp = K
where K is gain of the system.
1
The steady state error is ess = .
1 + Kp
Q. 51. What is the steady state error of a type 1 system subjected to a parabolic
input ?
Ans. The transfer function of a type 1 system is
K (1 + sTa ) ...
G(s) H(s) =
s(1 + sT1) ...
The acceleration error coefficient is
Ka = lim s 2 G(s) H (s)
s→0
K (1 + sTa ) ...
= lim s2 . =0
s→0 s(1 + sT1) ...
1 1
lss =
= = ∞.
Ka 0
Q. 52. A second order control system has damping ratio ζ = 0.6 and natural
frequency of oscillations ωn = 10 rad./sec. Determine the resonant peak and
resonant frequency.
1 1
Ans. Resonant peak : Mr = = = 1.041
2
2ζ 1 − ζ 2 × 0.6 1 − 0.62
In polar form
w=¥ K
K ωT O Re
G( jω) = ∠ − tan −1 f = – tan–1 wT w=0
2
ω T +1 2 1 K w
Öw2T 2 + 1
At ω = 0, G( j0) = K
∠G( j0) = 0° G(jw)-plane
748 LINEAR CONTROL SYSTEMS
At ω = ∞, G( j∞) = 0
∠G( j∞) = 90°.
Imj
Q. 75. Show the shape of the polar plot for
1 w=¥
G(s) H(s) = .
s(sT1 + 1)(sT2 + 1) O Re
1
Ans. G( jω) H( jω) =
jω( jωT1 + 1) ( jωT2 + 1) w
G(jw) H(jw)-plane
As ω → 0, G( j0) H( j0) → ∞
∠G( j0) H( j0) → 90° w=0
As ω → ∞, G( j∞) H( j∞) → 0
∠G( j∞) H( j∞) → 270°.
Q. 76. Draw the polar plot for the transfer function
K
G( jω) = . Imj
jω
K w=¥
Ans. G( jω) = O Re
ω
∠G( jω) = 90° w G(jw)-plot
As ω → 0, G( j0) → ∞ w=0
∠G( j0) → 90°
As ω → ∞, G( j∞) → 0
∠G( j∞) → 90.
K
Q. 77. Sketch the polar plot of G(s) = .
s2
5
Ans. G( jω) =
( jω )2
Imj
5
G( jω) =
ω2 w=0 w w=¥
∠G( jω) = 180° O Re
As ω → 0, G( j0) → ∞ G(jw)-plane
∠G( j0) → 180°
As ω → ∞, G( j∞) → 0
∠G( j∞) → 180°.
Q. 78. Draw the shape of polar plot of type 0 and 3rd order transfer function
having no zeros. Imj
Ans. The 0 and 3rd order transfer function is given below :
K
G(s) = w=¥ K
( sT1 + 1)( sT2 + 1)( sT3 + 1) O w=0 Re
K w
G( jω) =
( jωT1 + 1)( jωT2 + 1)( jωT3 + 1) G( jw)-plane
APPENDIX III : SHORT ANSWER QUESTIONS 749
K
G( jω) =
2
ω T12 + 1 ω T22 + 1 ω 2T32 + 1
2
1 w=¥
G( jω) =
2 2 O Re
ω ω T +1
1 ωT w
∠G( jω) = 90° tan
1 G( jw)-plane
As ω → 0, G( j0) → ∞ w=0
∠G( j0) → 90°
As ω → ∞, G( j∞) → 0
∠G( j∞) → 180°.
Q. 80. What is meant by frequency response of a control system ?
Ans. The steady state response of a system to sinusoidal input is known as frequency
response. G( jω) and ∠G( jω) as a function of frequency ω give the plot of frequency
response.
Q. 81. State Nyquist criterion.
Ans. The expression for the Nyquist criterion is given below :
N=P Z
where N = Number of counter clockwise encirclements of the point ( 1 + j0) by the
Nyquist plot.
Z = Number of zeroes of the characteristics polynomial [1 + G(s) H(s)] or the
roots of the characteristic equation 1 + G(s) H(s) = 0 having +ive real
part.
P = Number of poles of the open-loop transfer function G(s) H(s) having
+ real part i.e., poles located in R.H.S. of s-plane.
For stability Z=0
∴ N = P.
Q. 82. What is Nyquist plot ?
Ans. The plot of open-loop sinusoidal form of transfer function G(s) H(s) in a complex
plane corresponding to Nyquist contour traced in clockwise direction in s-plane is
called Nyquist plot.
750 LINEAR CONTROL SYSTEMS
KT1 T2
Q. 88. The gain at the phase crossover frequency is , determine the gain
T1 + T2
margin.
1
Ans. G.M. = 20 log10 db
G( jω 2 ) H ( jω 2 )
T1 + T2
= 20 log10 db.
KT1T2
Q. 89. Define phase margin.
Ans. The phase margin (φm ) is defined as the additional phase lag required at the gain
crossover frequency ω1 to make the system marginally stable. Thus,
φm = 180° + ∠G( jω1) H( jω1).
Q. 90. What is the advantage of frequency response analysis ?
Ans. The frequency response plot of an open-loop transfer function gives the assessment
of relative stability.
APPENDIX III : SHORT ANSWER QUESTIONS 751
Q. 91. Write the advantages of Bode plot.
Ans. 1. The method of sketching the Bode plot is simple.
2. The multiplication of magnitudes of terms is converted to addition.
3. From the experimental Bode plot the transfer function can be determined.
4. The low frequency and high frequency behaviour of the transfer function can be
determined.
Q. 92. What is Bode plot ?
Ans. The Bode plot is a frequency response plot of the open-loop transfer function
expressed in sinusoidal form. It consists of two parts as below :
1. Transfer function magnitude in db versus log10 ω.
2. Phase angle of the transfer function versus log10 ω.
Q. 93. Show that 20 db/decade = 6 db/octave.
Ans. 1 decade = log10 10 = 1
1
2
1 octave = log10 = 0.301
1
∴ 20 db/decade = 20 × 0.301 db/octave = 6 db/octave.
Q. 94. What is the maximum error due to asymptotic magnitude plot for the term
ωT) ?
(1 + jω
1
Ans. The maximum error is 3 db occuring at the corner frequency ω = .
T
Q. 95. Find the corner frequencies for the transfer function
K
G(s) H(s) = .
s (s + 1)(s + 4)
Ans. The corner frequencies due to zeroes are
(i) ω = 1 rad./sec.
(ii) ω = 4 rad./sec.
Q. 96. Determine the corner frequencies and initial slope of the Bode plot for the
open-loop transfer function given below :
20 (s + 0.25)
G(s) H(s) = .
(s + 0.5) (s + 10)
Ans. The corner frequencies are
(i) ω = 0.25 rad./sec due to zero
(ii) ω = 0.5 rad./sec due to pole
(iii) ω = 10 rad./sec due to pole
The type of the given transfer function is 0, therefore, the initial slope is 0 db. At the
frequencies lower than the lowest corner frequency the Bode plot magnitude starts
at 20 log10 K and
20 × 0.25
K= = 1.
0.5 × 10
Q. 97. How G.M. and P.M. can be determined from the Bode plot ?
Ans. G.M. = G( jω2) H( jω2) db, where ω2 is the phase crossover frequency.
P.M. = 180° + ∠G( jω1) H( jω1), where ω1 is the gain crossover frequency.
752 LINEAR CONTROL SYSTEMS
8
Q. 98. Draw the Body plot for G(s) = .
s
Ans. (1) Type 1 transfer function, the initial slope of the magnitude plot is 20 db/decade
intersecting 0 db axis at ω = K = 8 rad/sec.
(2) ∠G( jω) = 90°.
20 db
– 20
|G(jw)| db/d
ec
0 db
1 2 4 8 10 r/s w
1 2 4 8 10 r/s w
0°
ÐG(jw)
– 90°
9
Q. 99. Sketch the magnitude (db) plot for G(s) = .
s2
Ans. Type 2 transfer function, the initial slope of the magnitude plot is 40 db/decade
intersecting 0 db axis at ω = K = 9 = 3 rad./sec.
15 db
20 db
– 40
|G(jw)| db/d
ec
0 db
1 2 3 4 8 10 r/s w
LM x OP LM 0
1 1 0 OP LM x OP LM0OP
1
Ans. MM x PP = MM 0
2 0 1 PP MM x PP + MM0PP r (t)
2
N x Q N− 8
3 −6 −4 Q N x Q N2Q
3
LM x OP
1
y= 1 0 0 MM x PP .
2
Nx Q
3
758 LINEAR CONTROL SYSTEMS
LM 1 1 1 OP
......
......
P= M PP
λ 1 λ2 λ3
MM− λ: 2
1 − λ22 − λ23 ......
PQ
N : :
where, λ1, λ2, λ3, ...... are eigen values of the given matrix A.
Q. 138. What is understood by pulse transfer function ?
C ( z)
Ans. The function G(z) = is known as pulse transfer function.
R( z)
C(z) is the z-transform of the output function and R(z) is the z-transform of the input
function.
Q. 139. Give relation between s-domain and z-domain.
Ans. z = esT.
Q. 140. Write z-transform of unit step function.
z
Ans. Z[1] = .
z−1
Q. 141. What is hold circuits ?
Ans. The hold circuit enables to hold the signal between two consecutive sampling
instants at the preceding value till the next sampling instant is reached.
APPENDIX III : SHORT ANSWER QUESTIONS 759
Q. 142. Sketch the stable region in z-plane.
Ans. The inside region of a unit circle centred at the origin of z-plane is the stable region
for discrete-data (sampled data) control systems.
Imj
Unit circle
Stable
O Re
Region
z-plane
π φ 1
1 ξ2
Rise time : tr = ; φ = tan
ωn 1 ξ2 ξ
π
Peak time : tp =
2
ωn 1 ξ
4
Settling time : ts = (2% basis)
ξωn
3
ts = (5% basis)
ξωn
760
APPENDIX IV : KEY FORMULAE, TABLES, CHARTS 761
ξπ
2
1 ξ
% Max. overshoot : Mp = e × 100%
ξ = 0 (no damping)
c(t) = (1 cos ωnt)
ξ = 1 (critical damping)
ωn t
c(t) = {1 e (1 + ωnt)}
ξ < 1 (over damped)
(ξ ξ2 1) ωn t (ξ ξ2 1) ωn t
e e
c(t) = +
2 ξ2 1 (ξ ξ2 1) 2 ξ2 1 (ξ + ξ2 1)
Characteristic equation : s2 + 2ξωns + ωn2 = 0
Root of the characteristic equation : s = ξωn ± jωn 1 ξ2
Response to impulse input
ξ<1
ωn
c(t) =
1 ξ 2
e ξ ωnt
{
sin (ω n 1 ξ2 ) t }
ξ=1
c(t) = t (ω 2n e ωn t )
ξ>1
c(t) =
ωn ⎡ e (ξ ξ2 1) ωn t
e
( ξ + ξ2 1) ωn t ⎤
⎢ ⎥⎦
2 ξ 2
1 ⎣
1
Steady state error ess =
1 + Kp
e
lss =
Kv
e
lss =
Ka
E(s) 1
Error transfer function =
R(s) 1 + G(s) H (s)
C(s) (1 + T α) ω 2n
Transfer function with derivative control : = 2
R(s) s + (2 ξω n + ω 2nTα ) s + ω 2n
C(s) (s + K i ) ω 2n
Transfer function with integral control : = 3
R(s) s + 2 ξω n s2 + ω 2n s + K i ω 2n
C(s) ω 2n
Transfer function with derivative feedback control : = 2
R(s) s + (2 ξω n + ω 2n K t ) s + ω 2n
762 LINEAR CONTROL SYSTEMS
1
Gain Margin : G.M.db = 20 log10
G ( jω2 ) H ( jω2 )
Phase Margin : P.M. = 180° + ∠G( jω1) H( jω1)
The initial slope of the Bode plot depends on the type of open-loop transfer function
G( jω) H( jω). The table below give the slope and corresponding type.
0.01
0.01 ± 0.00043 0 ± 0.00043
T
0.1
0.1 ± 0.043 0 ± 0.043
T
0.5
0.5 ± 1.0 0 ± 1.0
T
0.76
0.76 ± 2.0 0 ± 2.0
T
1
1.0 ± 3.0 0 ± 3.0
T
1.31
1.31 + 4.3 ± 2.3 ± 2.0
T
2
2.0 + 7.0 ± 6.0 ± 1.0
T
10
10 ± 20.043 ± 20 ± 0.043
T
100
100 ± 40.00043 ± 40 ± 0.00043
T
Semilog scale
The Bode plot consists of G(jω) H(jω) in decible versus log10 ω. A unit change in log10 ω
corresponds to one decade (ratio of 101) variation of ω i.e. ..., ω = 1 rad/sec to ω = 10 rad/sec,
ω = 10 rad/sec to ω = 100 rad/sec, .......
This change, as independent variable and G( jω) H( jω) db as dependent variable.
The semilog scale can be plotted on plain paper. Select some scale say 1 decade = 6
cms. The selection of scale is arbitrary and to be adjusted according to the size of the paper.
The calibration is given below :
⎛ 4ω ⎞
4ω log10 ⎜ ⎟ = 0.602 0.602 × 6 = 3.6 cm
⎝ ω ⎠
⎛ 5ω ⎞
5ω log10 ⎜ ⎟ = 0.699 0.699 × 6 = 4.19 cm
⎝ ω ⎠
⎛ 6ω ⎞
6ω log 10 ⎜ ⎟ = 0.7782 0.7782 × 6 = 4.66 cm
⎝ ω ⎠
⎛ 7ω ⎞
7ω log10 ⎜ ⎟ = 0.8451 0.8451 × 6 = 5.07 cm
⎝ ω ⎠
⎛ 8ω ⎞
8ω log 10 ⎜ ⎟ = 0.9031 0.9031 × 6 = 35.72 cm
⎝ ω ⎠
⎛ 9ω ⎞
9ω log 10 ⎜ ⎟ = 0.9542 0.9542 × 6 = 5.41 cm
⎝ ω ⎠
⎛ 10ω ⎞
10 ω log10 ⎜ ⎟ = 1.0 1 × 6 = 6 cm
⎝ ω ⎠
60 db
40 db
20 db
0 db
1 2 3 4 5 6 7 8 9 10 20 30 40 100ω
1.8 cm
4.19 cm
6 cm
log10 ω scale
Ratio 1.7 1.6 1.5 1.4 1.3 1.2 1.0 0.9 0.8
Decibel 20 18 15 12 10 9 8 7 6 5
Ratio 10 7.94 5.62 3.99 3.16 2.82 2.51 2.25 2.0 1.78
Ratio 1.59 1.41 1.26 1.12 1.06 1.0 0.945 0.894 0.795 0.707 0.63
Decibel 5 6 7 8 9 10 12 15 18 20
Ratio 0.562 0.5 0.447 0.398 0.354 0.316 0.251 0.178 0.125 0.10
APPENDIX IV : KEY FORMULAE, TABLES, CHARTS 767
0.25 db
– 28 db
2 o
0.5 db
24 db
– 5o
1 db 20 db
db
0 db
16 db
0.5
2 db –
10 o
12 db
db
3 db
–1
4 db –2
0 o
8 db
6 db
b
2d
db 4 db
3
–
12 db – db
–4 db
–5 b 0 db
d
–6
– 4 db
– 9 db
– 40
– 8 db
– 12 db
o
– 30
– 50 – 12 db
– 60 o
o
– 70
o
– 100
– 80
– 200
– 18 db – 16 db
o
o
o
– 20 db
– 240
– 220
– 140
– 90
o
– 160
– 120
– 180
– 24 db
o
o
o
o
– 24 db
– 28 db
o
o
– 240
– 220
– 200
– 180
– 160
– 140
– 120
– 100
– 80
– 60
– 40
– 20
o
0
Nichols Chart
Eo (s) α (1 + sT )
1. Lead network transfer function : = ; α<1
Ei (s) (1 + sαT )
1
Mid-frequency : ωm =
( α )T
1 (1 α)
Maximum phase load : φm = sin
(1 + α)
768 LINEAR CONTROL SYSTEMS
Eo (s) 1 + sT
2. Lead network transfer function : = ; β>1
Ei (s) 1 + sβT
1
Mid-frequency : ωm =
( β )T
1 (1 β)
Maximum phase lag : φm = sin
(1 + β)
State model : x = Ax + Bu
(MIMO) y = Cx + Du
State model : x = Ax + Bu
(SISO) y = Cx + Du
State transition matrix : φ(t) = L 1 (s I A) 1
φ(t) = eAt
Time solution :
x(t) = φ(t) x(0) + L 1 φ(s) Bu(s)
t
or x(t) = φ(t) x(0) + ∫ 0 φ (t τ) Bu(τ) dτ
Y(s) 1B
Transfer matrix : = C (sI A) +D
U(s)
Characteristic polynominal : sI A =0
Controllability test matrix :
U = B : AB : A2B : ...... : An 1B
769
Index
C Critical damping, 138, 139
A
Critically damped system, 137
Absolute stability, 218, 712 Cascade compensation, 364 Cut-off frequency, 257
A.C. servomotor, 95 Cayley-Hamilton theorem, 442
A.C. tachogenerator, 101 application, 443 D
Acceleration error coeff, 156 Characteristic equation, 140, 741 Damped frequency, 136
Acceleration function, 129 Classification of transfer Damping factor, 136
Actual damping, 135, 136 function, 154
Damping ratio, 135, 742
Actuating signal, 173 Closed-loop control system, 2,
Decade of frequency, 291
169, 488
All pass transfer function, 306 Decaying oscillations, 151
Closed-loop transfer function :
Amplidyne generator, 99 Decibel, 766
equivalence, 30
Angle of arrival from complex D.C. technogenerator, 101
Close-loop frequency response,
zero, 324 Derivative control, 173, 176
250
Angle of departure from complex Derivative feed-back control, 181
Comparator, 2
pole, 324 Diagonalization of matrix, 410,
Compensation network, 364 368
Argument principle, 750 758
Compensation network root
Armature controlled D.C. motor, 93 Difference equations, 512
locous approach 380
Asymptotes : Bode plot, 294 Digital control system, 500
Compensation, 363
Asymptotes : Root locus, 321, 322 Displacement function, 129
phase lag, 366
Asymptotic bode plot, 752 Dominant pole pair, 744
phase lead, 364
phase-lag-lead, 368 Dominant poles, 151
B
Conditionally stable system, 245
Back e.m.f. constant, 92 Constant phase angle loci, 287
E
Bandwidth, 258 Control actions, 173 Eigen value, 444, 445
Bilinear transformation, 527 derivative, 173 Eigen vector, 446, 447
Block diagram, 25 example of, 2 Electrical network, 80
closed-loop, 30 proportional, 173 Equivalent moment of inertia, 23
of control system, 17 Control system, 1, 734 Equivalent viscous friction, 23
of transfer system, 16 block diagram of, 17 Error coeff. acceleration, 155
reduction, 25 controllers in, 745 Error coeff. position, 155
Bode plot, 289, 294, 751 determination, 740 Error coeff. velocity, 155
initial slope, 295 determining transfer Error detector, 102
magnitude error table, 763 function, 19 potentiometer, 102
procedure for drawing, 301 Controllability, 451 synchro (selsyn), 103
solved examples, 308 Controllable canonical form, 426 Error signal, 2
time delay, 304 Corner frequency, 293, 752 sampling, 523
transfer functions, 289 Correlation between transient Error transfer function, 31
Breakaway point, 323 and frequency response, 315 Excitation (Cause), 16
770
INDEX 771
F Integral-of-time-multiplied- N
absolute error, 162
Feedback compensation, 369 Integral-of-time-multiplied- Natural frequency of oscillations,
Feedback path transfer function, square error, 162 135
30 Inverse Laplace transform, 3 N-circles, 249, 719
Feedback, 2 Inverse Nyquist plot, 284 Nichol s chart, 254, 732
Field controlled D.C. motor, 90 Inverse z-transform, 508 Non-minimum phase transfer
Final value theorem, 7 partial expansion, 509 function, 306
Force current analogy, 82 power series, 510 Number sequence, 511
Force voltage analogy, 82 Nyquist criterion application,
Forward path transfunction, 31 J 231
Forward path transmittance, 63 Nyquist criterion, 229
Frequency response, 315 Jordan s canonical form, 433 Nyquist plot e st, 241
analysis, 762 Jury s stability test, 529 relative stability, 244
solved examples, 259
G L
Lag lead compensator design 404
O
Gain adj : 1/M circle, 288
Gain adj : M-circle, 289 Laplace transform pairs, 5 Observability, 454
Gain cross-over frequency, 243 Laplace transform theorems, 6 Observable canonical form, 427
Gain margin from Bode plot, 301 Laplace transform, 3, 441 Octave of frequency, 291
Gain margin, 242, 244, 715 derivation of, 4 Open-loop control system, 2, 169
Gain phase plot, 245 impulse train, 504 Open-loop transfer function, 154
Gain term K, sampled function, 503 Oscillatory (overdamped)
graphs, 290 use in control systems, 3 response, 135
Gear, 22 Linear control system, 1 Output node, 63
Generalized error coeffs, 159 Loop transmittance, 63, 64 Output signal, 2
Generators, 97 Overall transfer function, 30
excited d.c., 97 M Overdamped response, 138
Generator constant, 97 Overshoot, 142
1/M circles, 287
Graph determinant, 66 Mapping from s-plane to
G(s)H(s)-plane, 229
P
H Mapping from s-plane to Parabolic (acceleration) function,
z-plane, 526 129
Heat transfer system, 89
Mason s gain formula, 67, 704 Pass transfer function, 306
Higher order system, 22
MATLAB, 500 Path factor, 67
Hold circuits, 515
Maximum overshoot, 142 Peak time, 142
Hurwitz determinant, 220
Maximum phase transfer Performance indices, 162
Hydraulic actuator, 86
function, 279 Phase crossover frequency, 272
Hydraulic system, 86
M-Circles, 247, 718 Phase margin bode plot, 301
Mechanical circuit diagram, 83 Phase margin, 244, 715
I Mechanical systems, 81 Phase variables, 757
Impulse function, 129 rotational, 82 Phase-lag compensation, 80, 366
Impulse response, 146 translational, 81 Phase lead compensator design
Impulse train, 504 Minimum phase function, 306 381
Infinite series, 440 Minimum sampling frequency, Phase lag compensator design
Initial value theorem, 6 518 394
Input node, 63 Modal matrix, 445, 447 Phase-lag network, 79
Input test signals, 128 Motor gain constant, 94 Phase-lead network, 364
Integral absolute error, 162 Motor time constant, 94 PI control, 178
Integral control, 178 Multiple poles, 18 PID control, 181
772 LINEAR CONTROL SYSTEMS
Pneumatic actuating valve, 88 Separately excited D.C. second order system (ramp
Pneumatic system, 88 generator, 97 input), 144
Polar plot, 747 Servomotors, 90 second order system (step
Poles : complex, multiple, simple, two phase a.c., 95 input), 134
17 Settling time, 144 third order system, 147
in s-plane, 18 first order system, 131 Transfer function decompo-
of transfer function, 140 second order system, 144 sition, 428
Positional error coeff., 155 Signal flow graph, 62 cascade, 431
Proportional control, 173 Stability definition, 218 direct, 429
Pulse transfer function, 519, 758 Stability: sampled-data, 490 parallel, 432
from block diagrams, 519 Stable system, 217, 244 Transfer function, 16
Starred function, 503 block diagram of, 16
R Starred transform, 504 poles and zeros of, 17
State equation: solution, 438 impulse response, 19
Ramp (velocity) function, 129 of non-homogeneous, 439 Transfer matrix, 450
Relative stability, 218, 244, 747 State signal flow graph, 391 Transient response, 315
Resonant frequency, 299, 753 State space representation, 420, specifications, 128
Resonant peak, 300 421, 422 Translational spring-mass
Response (effect), 16 State transition matrix, 438 damper system, 81
Rise time, 141 determination of, 441 Transmittance, 63, 64
Root contours, 351 Laplace transform method, Two phase A.C. servomotor, 95
Root locus, 319 441
angle of arrival, 324 properties, 450 U
angle of asymptote, 323 State variables, 429
angle of departure, 324 Undamped (sustained oscilla-
feedback, 490
asymptote intersection, tion) response, 136
State, 420
324 Unit parabolic function, 129
concept, 419
breakaway point, 323 Unity feedback, 31
space, 420
imaginary axis Unstable system, 217
vector, 420
intersection, 324 Static error coeffs., 155
plotting procedure, 322 Steady state error, 153, 158
V
salient features, 323 Steady state response, 128 Vander Monde s matrix, 449, 758
solved examples, 324 Step (displacement) function, Velocity error coeff., 155
starting point, 323 128 Velocity function, 129
terminating points, 323 Summing point, 27
Rotational mechanical system, Synchro error detector, 17
84
Z
Routh-Hurwitz criterion, 221 Zero input response, 439
T
solved examples, 222 Zero order hold, 515
Rules for drawing signal flow Take-off point, 26 transfer function of, 515
graph, 74 Techogenerator constant, 101 Zeros of transfer function, 17
Thermal system, 89 Zero state response, 439
S Time delay element z-transform, 508
Bode plot, 304 inverse, 508
Sampled-data control system,
Nyquist plot, 240 of useful functions, 505
500
Time response : properties, 504
Sampler, 501
first order system (ramp z-transfer function, 519
ideal, 502
input), 132
Sampling process, 502
first order system (step
Sampling theorem, 518 input), 130
Sensitivity, 168 higher order system, 152