Chapter-5 “DYNAMIC PERFORMANCE”

Time-Response of Closed Loop

feedback Control Systems.

5-1 Introduction

How will the output from a system change when there is an input to it? For
example, if we consider a mercury-in-glass thermometer as a system with an
input of temperature and an output of the level of the mercury in the glass
capillary, how will the level change when the thermometer is suddenly
immersed in hot water, i.e. given a step input? As a matter of experience we
know that the level will increases as a result of the input and take a certain
amount of time before it reaches its steady state value.

In Chapter 3 the methods of describing a system by means of a transfer function

was introduced; the transfer function is the output divided by the input when
both are written as s-functions. In this chapter we consider how we can use
transfer functions to determine how the output of a system will change with time
for particular inputs. This chapter is concerned with a method we can use to
answer this question and, in general, determine how the output of systems
changes when there is a change in input.

5-2 Inputs signals

In analyzing and designing control systems, we must have a basis of comparison
of performance of various control systems. This comparison can be achieved by
applying predefined input test signals and comparing the corresponding output

1
(response) of various control systems and knowing the capability of the system
to cope with the actual input signal. The commonly used test input signals are
those of impulse functions, step functions, ramp functions, acceleration
functions, sinusoidal functions. With these test signals, mathematical and
experimental analysis of control systems can be carried out easily since the
signals are very simple functions of time.

5.2.1 Impulse input

 (t )
We have the input existing for just a very brief time

Input
before dropping back to zero. A unit impulse input
r t    t  which starts at a time t = 0 and rises to the Time

value 1 has a Laplace transform as R s  1.

Unit Impulse

5.2.2 Step input (displacement function)

1
We have the input suddenly being switched to a constant
Input

value at some particular time. A step input r t   R u t 

Time
which starts at a time t=0 and rises to the constant value R
Unit step input
has a Laplace transform R s  R / S

Note; R measured in meter or radians so, called displacement function.

5.2.3 Ramp input (velocity function)

We have the input starting at some time and then  (t )

increasing at a constant rate. A unit ramp input which
Input

Time
ramp 2
starts at time t = 0 and rises by 1 each second has a Laplace transform of the
input as an s-function

R
r t   R t u t  , R s  
S2

Note; R measured in m/sec or rad/sec so, called velocity function.

5.2.4 Acceleration function (parabolic input)

Used as a test signal for control systems which their

inputs increased in quadratic form with respect to time

Input
and written in the following form

R 2 R Time
r t   t u t  , R s   parapolic inputs
2 S3

Where; R measured in m/s2 or rad/ s2 so, called acceleration function.

Note that, In s-domain, the velocity function is the integration of displacement

function and the acceleration function is the integration of the velocity function.

5.2.5 Sinusoidal function

Used as a test signal for control systems which their actual inputs function can
be represented as a sum of sin-waves with different amplitude and frequencies.
And we shall study the response due to this function in details in chapter-7.

r t   R sint .

2 1
Where, the angular frequency    2f , and frequency f 
 

3
5.3 Time response of higher order systems (Determining output)
Any feedback control system can be R(s) E(s) C(s)
+ G(s)
represented as the block diagram shown in -
B(s)
Fig. (5.1). The response of the system can
H(s)
be written as:
Fig.(5.1) Feedback control system

G s 
C s   R s  5.1
1 H s G s 

To find the time response Ct  for a certain input signal R t  , you must obtain
the inverse of Laplace transform for the equation 5.1, and this can be done by
using the partial fraction method See appendix-I. To apply this method you
must put the numerator and dominator in the following general form for the
transfer function of higher ordered systems:

C s  K  s  1 s  z s 
r

 5.2
R s   im1 2 n s  Pi 

Where,

K, The constant feed forward gain.

zs , The zeros of the overall transfer function of the control system. s=1:r.

Pi , The poles of the overall transfer function of the control system. i=1:m+2n.

Where, m and n defined the number of the roots as you will see next.

At zeros, the value of the transfer function goes to zero, while at the poles the
value of transfer function is undefined so, we must take an attention to the poles
which is defined as roots of the characteristic equation.

4
 1 G s H s  0 5.3

In general case, the root of characteristic equation 5.3 may be pure real roots,
pure imaginary roots, complex roots (real part and imaginary part), complex
conjugate roots or a combination of the pre-mentioned roots.

Let us define the poles of the characteristic equation as

The pure real roots are:  1 ,  2 , ………,  m

The pure imaginary roots are:  J1 ,  J2 ,………,  Jn

The complex roots are:  1  J1 ,  2  J2 ,………,  n  Jn

Their complex conjugate roots are:  1  J1 ,   2  J2 ,…,  k  Jn

Thus we can resolve the characteristic equation to,

1 G s  H s    im 1 s  i   nk 1 s   k  J k  s   k  J k 

1 G s  H s    im1 s  i   nk 1 s   k  J k  5.4

From equation 5.4 it is clear that the total number of roots is (m+2N).

Equation 5.2 can be rewritten as

C s  K  sr  1 s  z s 
 5.5
R s   im 1 s  i   nk 1 s   k  J k 

Thus the system response for a unit step displacement input r t   u t  which
1
has a Laplace transform R s   can be written as.
s
Ksr  1 s  z s 
C s   5.6
s  im 1 s  i   nK  1 s   k   J k  s   k   J k 

5
Equation 5.6 can be written in partial fraction form as:

1 m
Ai n
Bk S  D k
C(s) 
s
 
i 1 s   i
 
k 1 s   k   k
2 2
5.7

The time response can be obtained by using the inverse Laplace Transform of
equation 5.7 with the help of using Laplace Table as,

m n
C t   1   A i e  i t
  E k e  k t cos k t  k  5.8
i 1 k 1

The time response consists of the algebraic sum of three terms; the first term
represents the steady state response corresponding to the input function which is
the same as the input, while the other two terms represent the Transient
 i t
response. The exponential functions e is either to be exponentially decaying
function (Converging) or exponentially increasing function (diverging). The
diverge and converge depend on the sign of the power of the exponential
function. The third term is oscillatory decaying or oscillatory diverging function.
For these reasons the Transient response is responsible on the system Stability
while the steady state response is responsible on the system Accuracy as
discussed later.

Fig. (5.2) Effect of closed loop poles position in the s-plane on the

6
5.4 Time response of first order systems
A first order system Shown in Fig. (5.3.a) can be modeled in time domain by
differential equation of the form:

dy  t 
  y t   x t  5.9
dt

R(s) E(s) 1 C(s)

+
x - τs

K
k(x-y) Fig.(5.3.b) closed loop Block
y
diagram of first order system

c cy R(s) C(s)

1
τs  1

Fig (5.3.a). First order Fig.(5.3.c) Simplified

system block diagram

The corresponding model in s-domain for zero initial conditions can be written as:

 s Y( s )  Y( s )  X( s ) 5.10

Otput Ys  C s  1
  5.11
Input Xs  R s  s 1

Where, , Is defined as the time constant depend on the physical properties of

the control system (damping factor c and stiffness constant k) for Example the
first order system shown in Fig. (5.3.a) has a differential equation of the form:

c y  ky  kx

The Laplace transform can be written as

cs Y(s)  kY(s)  k X(s)

7
Can be reduced to the form of equation 5.11 as

Otput Ys  C s  k 1 1
    5.12
Input Xs  R s  cs  k  c  s  1
 s  1
k

Thus for the given system the time constant   c which represent the ratio
k

between the damping coefficient to the stiffness factor which are the physical
properties of the given system. the closed loop feedback control system shown
in fig.(5.3.b) while the closed loop transfer function shown in fig.(5.3.c).

In the following subsections we shall find the time response of the first order
control system due to a unit step, unit ramp and unit impulse inputs.

5.4.1 Unit step response of L.T.I, first order system.

When a first-order system is subject to a unit impulse input r t   u t  then

1
R s   , And the output transform Cs is obtained from equation 5.12 as
s

1 1
C s   
s s  1

Expanding C(s) into partial fraction as

A B
C s  
s s  1

This leads to A=1 and B=-τ

1  1 1
C s      5.13
s s  1 s s  1 

8
Taking the inverse Laplace of equation 5.13 yields to the time response as;

c t   1  e  1  t For t≥0 5.14

Plot of equation 5.14 is shown in fig. (5.4), the main important points on the
curve are stated as follows:

At t=0,  c  t   0 , and at t=τ,  c  t   1  e 1  0.632 , this give indication to

the response characteristic for smaller time constant we can reach a faster
response so the time constant which is related directly to the system properties
can be used as a measure to the speed of response.

C(t)

Fig (5.4) step response of first order system

Another important characteristic of the equation 5.14 can be interpreted by

taking the response derivatives as:

c  t    1   e 1  t 5.15

At t=0,  c t   1 

At t  ,  c t   0
9
The above two equations give an indication to the steady state error where the
slope of the tangent line to the exponential curve is decreasing monotonically
from 1  at t=0 to 0 at t   .

Error e t   r t   c t 

C   1 , r t   1

Thus, the steady state error is

e ss  r t   c    0

5.4.2 Unit ramp response of L.T.I, first order systems

The Laplace transform of a unit-ramp input signal is:

1
R s  
s2

Thus the output can be written and fracture as;

1 1  2
C s   2    5.16
s s  1 s 2 s s 1

Taking the inverse Laplace transforms of equation 5.16 yields,

C t   t     e  1  t For t ≥0 5.17

The error signal e(t) is then.

e t   r t   c t    1   e  1  t 

10
At t  ,  e  1  t  0 , and thus the error signal approach to e    , and
this is mean that the smaller the constant time  , the smaller steady state error
for unit ramp input. The response is shown in fig. (5.5).

C(t)

Fig. (5.5) unit ramp response of first order system.

5.4.3 Unit-Impulse response of L.T.I, first order system.

When a first-order system is subject to a unit impulse input r t    t  then,

R s  1

The output transform C s is obtained from equation 5.12 as:

1
C s 
s  1

11
 1 1
C s    5.18
 s  1 

Taking the inverse Laplace of equation 5.18 yields to the time response as;

c t   1  e  1  t For t≥0 5.19

Plot of equation 5.19 is shown in Fig. (5.6), the main important points on the
curve are stated as follows:

C(t)

1
𝜏

Fig. (5.5) unit step time response of first order system

1
At t=0,  c  t  


1 1
At t=τ,  c t   e


This give an indication to the response characteristic for smaller time constant
we can reach a faster response so the time constant which is related directly to
the system properties can be used as a measure to the speed of response.

12
An Important property of linear time invariant systems can be discovered as
follows:

Unit-Ramp response c t   t     e  1  t For t≥0

Unit-step response c t   1  e  1  t For t≥0

Unit-Impulse response c t   1  e  1  t For t≥0

It is clear that the Unit-impulse response is the derivatives of the step response,
and the step response is the derivative of the ramp response. Also, the ramp
response is the integral of the step response and the step response is the integral
of the impulse response. The constants of the integration can be determined
from the zero output initial conditions.

Example

A circuit has a resistance R in series with a capacitance C. The differential

equation relating the input v and output vc, (the voltage across the capacitor), is:
dvc
v  RC  vc
dt

Determine the output of the system when there is a 3 volt impulse input.

As a function of s the differential equation becomes:

Vc (s) 1
V(s)  RC sV(s)  Vc (s) Hence, G(s)  
V(s) RC s  1

The output for 3volt input is

3
t
1 RC 3  RC
Vc (s )  3   e
RC s  1 1 RC
s
RC
13
5.5 Time Response of second order systems
The servo positioner motor mechanism or the mass-spring- damper mechanism
is considered as examples of second order system. Their differential governing
equation can be derived as:
R(s) E(s) k C(s)
+
- s( m s  c )

Fig.(5.7.b) Block diagram of

second order system

k [ u  y ] R(s) C(s)
 2n
s 2  2 n s   2n

c y Fig.(5.7.c) Simplified block

diagram

  c y  k( u  y )
my
my  c y  ky  u(t ) 5.20

The Laplace transform is.

ms 2 Y(s )  c sY(s )  kY(s )  kU(s )

ms 2
 c s  k Y(s )  kU(s )

Output Y(s) K
 
Input U(s) ms2  c s  k 

The closed loop transfer function is.

Y(s) Km
 2

U(s) s  c m s  k m   (5.21)

14
The general form of transfer function that describes the dynamic behavior of the
second order control system can be written as;

C(s) 2n
 5.22
R (s) s 2  2n s  2n

K: spring stiffness, m: mass and c: damping coefficient which represent the

system properties. Comparing equation 5.21&22 you can write

k
2n  5.23
m

c
2n  5.24
m

n , Is the natural angular frequency at which the system oscillates.

 , Is the damping ratio.

n and  they represent the system properties in implicit form.

In the next subsections we shall discus the response of the second order systems
due to an input of unit step signal and unit Impulse signal. The denominator of
closed loop transfer function is in quadratic form is called the characteristic
equation of the closed loop system. The poles (roots) of this equation can be
found as follows:

s 2  2n s  2n  0 5.25

2
 2  n  2  n 
s1, 2      n
2

2  2 

s1, 2    n  n  2  1 5.26

15
The poles of the characteristic equations s1 and s2, depend on the value of the
damping ratio  the four types of roots shall be discussed as follows:

I.   0  s1, 2   n  1  s1, 2   in

The two poles are pure complex conjugate (No Damping)

II. 0    1  s1, 2   n  in 1   2

The two poles are complex conjugate (Under damping)

III.   1  s1, 2   n

The two poles are pure real and equal (critical-damping)

IV.   1  s1, 2    n  n  2  1

The two poles are pure real not equal (over damping)

Fig.(5.8) shows the effect of the roots of the characteristic equation on the
damping of second ordered control system.

c(t) Under damping

Critical damping
Over damping

time
Fig.(5.8) Effect of the characteristic equation poles

on the damping of the second-order system.

16
5.5.1 Unit-step response of second order systems
When a second-order system is subject to a unit-step input, i.e. R(s)=1/s then the
output transform is:

1 2n
C(s )   2
s s  2n s  2n

The time response c(t) is obtained using partial fraction method taking into
consideration the three cases of the characteristic equation poles

Under-damped case ( 0    1)

For this case the roots of the characteristic are defined as.

s1   n  id , & s2   n  id

The damped natural frequency is defined as

d   n 1   2 5.27

The partial fraction of the output for this case is written as

A Bs  D
C(s )   2
s s  2n s  2n

Using fraction equality we can obtain the factors A, B, and D as:

A=1, B=-1, & D  2n Thus.

1 s  2n
C(s )   2
s s  2n s  2n

1 s  2n
C(s )  
s s1  n  id s 2  n  id 

17
 1 s  2n
C(s )  
s s1  n 2  id 2

The partial fraction form that is suitable for inverse Laplace transform table is;

1 s  n n
C(s)    5.28
s s1  n 2  d 2 s1  n 2  d 2

Using equation 5.28 one can rearrange the third term as

1 s  n  d
C(s)    
s s1  n   d
2 2
1  2 s1  n   d
2 2

Hence, the inverse Laplace transform of above equation is obtained as


c( t )  1  e  n t cos d t  e  n t sin d t
1  2

  
c( t )  1  e  n t  cos d t  sin  t  5.29
 1  2 d

 


The two harmonic functions cos d t  sin d t can be reduced to a
1  2
single harmonic function with new amplitude A and phase angle shift as


cos d t  sin d t  A sin  cos d t  A cos  sin d t  A sind t  
1  2
Where,
Imaginary

 1
1  2

A sin  =1 A cos  
1  2 

 Real

18
 1 1  2
A & tan   5.30
1  2 

c( t )  1  e   n t A sind t   5.31

Thus the time response of second order system is

 1   
 sin d t  tan1 1  
2
c( t )  1  e   n t   t≥0 5.32
 1  2    
   

The transient response as in equation 5.32 is oscillatory decay sine wave with
frequency d which varies with damping ratio  as sown in fig. (5.9).

As shown in fig (5.7.b) the error signal can be obtained as

et   r( t )  c( t )

 1   
 sin  d t  tan 1 1  
2
e n t   5.33
 1  2    
   

The error signal is damped sinusoidal oscillation. At steady state, or at t=∞, no

error exists between the input and the output ess=0.

c(t)

time

19
Undamped case (   0 )

The time response of second order for the un-damped case is obtained by
substituting   0 at equation 5.32

c( t )  1  cosn t t≥0 5.34

Oscillatory non decaying cosine wave with natural angular frequency of the
system thus the system become semi-stable.

Critical damped case (   1)

The two roots are pure real s1   n , & s 2   n and have equal value.

1 2n
C(s)   2
s s  2n s  2n

The partial fraction of the output for this case is written as

1 2n A B C
C(s)     
s s  n 2
s s  n 2 s  n 2

You can get the constants A=1, B=-1, and C  n

1 1 n
C(s )   
s s  n  s  n 2

The time response of second order system is obtained using Inverse Laplace as

c t   1  e   n t   n t e   n t

c t   1  e   n t 1  n t  t≥0 5.35

20
This response shown in fig.(5.9) at   1.

Over damped case (   1)

The two poles are real and not equal

2n
C s  
s s  s1  s  s 2 

Where;


s1  n    2  1  & 
s 2  n    2  1 
The inverse Laplace transform found as

 e  s1 t e  s 2 t
n 
c t   1     t≥0 5.36
2   1  s1
2 s2 

The response include two decaying exponential function. One of them is

decreasing faster than the other. The faster decaying term corresponding to the
smallest time constant. This mean that one of the real roots are closer to the
imaginary axis in s-domain. If we neglect this fast decay root say s1, then the
response can be written as;

 e s 2 t
n 
c t   1     5.37
2 2  1  s2 

If   we can write the response as

c t   1  e  s 2 t 5.38

This response is the same as first order system as we neglect one pole of the
characteristic equation. The response is shown in fig.(5.9) at   2 .

21
5.5.2 Impulse response of second order systems

When a second-order system is subject to a unit-impulse input, i.e. R(s)=1 then the
output transform is:

2n
C(s )  2
s  2n s  2n

The time response c(t) is obtained using partial fraction method taking into
consideration the three cases of the characteristic equation poles

Case of under damping ( 0    1)

c( t ) 
n
1  2

e n t sin n 1  2 t  t≥0 5.39

Case of critical damping (   1 )

c( t )  2n te  n t t≥0 5.40

Case of over damping (   1)

n  2 1 ) n t n  2 1 ) n t
c( t )  e (    e (  
2 2  1 2 2  1
,t≥0 5.41

Note that we can obtain the time

response c(t) by differentiating the
corresponding unit step response. The
unit impulse response is always positive
or zero c(t) ≥ 0.

22
Example

What will be the state of damping of a system having the following transfer
function and subjected to a unit step input?

1
G (s ) 
s 2  8 s  16

solution

For a unit step input R(s) = 1/s and so the output is given by:

1 1 1
C(s )   2 
s s  8 s  16 ss  4(s  4)

The roots of s2 + 8s + 16 are s1 = s2 = -4. Both the roots are real and equal and so
the system is critically damped.

Example

A system has an output c(t) related to the input r(t) by the differential equation:

c( t )  5 c ( t )  6 c( t )  r( t )

What will be the output from the system when it is subject to a unit step input?
Initially both the output and input are zero.

Solution

We can write the Laplace transform of the system as

(s 2  5s  6) C( t )  R (s )

The transfer function is thus:

C(s ) 1
G (s )   2
R (s ) s  5 s  6
23
The roots of s 2  5 s  6 are s1 = -3, and s2= -4. Both the roots are real and
negative, so the system is over damped.

For unit step input he output is given by

1 1 1
C(s)   2 
s s  5 s  6 ss  3(s  2)

Using partial fraction method you can factorize the output as

1 1 1
C(s )   
6s 3s  3 2s  2

Using inverse of Laplace lead to the time response as

c( t )  0.17  0.33e 3 t  0.5e 2 t

Problems

1. A system has a transfer function of 5/(s + 3). What will be its output as a
function of time when subject to

(a) A unit step input of 1 volt,

(b) A unit impulse input of 1 V?

2. A system has a transfer function of 2/(s + 1). What will be its output as a
function of time when subject to
(a) A step input of 3 volt,
(b) An impulse input of 3 volt?
3. A system has a transfer function of 1/(s + 2). What will be its output as a
function of time when subject to

(a) A step input of 4 Volt,

24
 (b) A ramp input unit impulse of 1 Volt/s?

4. Systems have a transfer functions of:

8s  3s  8 8s  1

G (s )  , G (s ) 
s  2s  4 s  22
What will be the output from each system when it is subject to

(a) A unit impulse input?

(b) A unit step input?
5. The input x and output y of a system are described by the differential
equation:

d2y dy
 4  3y  x
dt 2 dt

If initially the input and output are zero, what will be the output when
there is a unit step input?

6. A system has a transfer function of 100/(s2+s+100). What will be its

natural frequency n and its damping ratio  ?
7. A system has a transfer function of 10/(s2+4s+9). Is the system under-
damped, critically damped or over-damped?
8. A system has a transfer function of 3/(s2+6s+9). Is the system under-
damped, critically damped or over-damped?
9. A system has a forward path transfer function of 10/(s+3) and a negative
feedback loop with transfer function 5. What is the time constant of the
resulting first-order system?

25
 Transient response specification
of closed loop feedback control
system.
1- Introduction
As discussed before when a system is subject to, say, a unit step input it may
give an output which eventually settles down to some steady state response. The
response that it gives before settling down to this steady state is called its
transient response. This section is about the parameters used to specify the
transient response of systems and whether the transients lead to stable or
unstable systems.

For example, if we have a spring system (Figure) and suddenly

K
apply a load to it, it has a transient response which results in it
taking some time to reach its steady state value and also it is m
likely to overshoot the steady state value before it finally
settles down to the steady state value. What factors can we
1
change with the spring system to get it to respond more
Input

quickly to an input and also to minimum the

overshooting? These are questions that are often posed for Time
control systems. As another illustration, consider a control
Amplitude

system used with an automatic machine to position a overshooting

workpiece before some machining operation, we need to steady

state
know how fast the system will respond to an input signal and
Time
position the item in the required position and will the

26
system be like the spring system when a load is applied to it and overshooting of
the required position occur. Overshooting is undesirable in such a situation and
so, if it occurred in such a control system, we would need to consider what steps
can be taken to eliminate it. Parameters are used as a way of specifying how fast
a system will respond to an input and how quickly it will settle down to its
steady state value.

With the above spring system, the result of applying a load is that, after
some oscillations with ever decreasing amplitude, the transients die away, and
the system settles down to a steady state value. The system is said to be stable.
If, however, the oscillations had continued with ever increasing amplitude, then
no steady state value would have been reached and the system would be
unstable. This section takes a brief look at the conditions for stability of systems.

2 Transient Response specification for first order system

The time response of the first order system is given by equation 5-14

c t   1  e  1  t For t≥0

the parameters used to specify the transient performance are of first order system
are defined as follows:

Delay time

The delay time td is the time required for the output response to reach 50% of its
steady state value. Thus, since 1 is the final value, the time taken to reach 50%
of this value is given by:

50%  1  e  1  td

1 / 2  e  1  td

27
 1 / 2  e  1  td

t d   ln 2 (1)

Rise time

The rise time tr is the time required for the output to rise from 10% to 90% of its
steady state value. Note that the specification is not always in terms of 10% to
90%, sometimes it is 0% to 100%. Since 1 is the steady state value then the time
taken to reach 10% of that value is:

10%  1  e  1  t10

9 / 10  e 1  t10

10 / 9  e 1  t10

ln10  ln 9  t10 / 

Also,

90%  1  e 1  t90

1/ 10  e 1  t90

10  e 1  t90

ln 10  t 90 / 

tr= t90-t10=τ ln9 (2)

28
Example 1

Determine the delay time and the rise time for a first order system with the
3
transfer function: G(s) 
2s  1

solution

The transfer function indicate that the steady state gain is 3 and the time constant
is 2sec. thus

t d   ln 2 =2ln2=1.4 sec

t r  2 ln 9  4.4 sec

Example 2

A mercury-in-glass thermometer acts as a first order system with an input of

temperature and an output of the mercury position against a scale. The
thermometer is initially at 0o centigrade and is then suddenly placed in water at
100oc After 80 s the thermometer reads 98% Determine:

(a) The time constant, (b) the delay time, (c) the rise time.

Solution

(a) For such a system the output θ is related to the input by the equation:


t  100 1  e  1  t 
Hence: 
98 100 1  e  1  80 
0.2  e  1  80

The time constant τ=49.7 sec.

29
b) The delay time td is

t d   ln 2 =49.7 ln2=34.5 sec

c) Rise time

t r   ln 9  49.7 ln 9  109.2 sec

3 Transient Response specification for second order system

As discussed before the form the output takes for a unit step input depends on
the system properties  , n . The corresponding outcomes being under damped,
critically damped and over damped

For the under damped oscillations of a system we can calculate the damping
frequency as: d  n 1   2 . The output corresponding to step response (k)
given by:

   
c( t )  k 1  e n t  cos d t  sin d t  3
  1  2 
 

The system parameters defined as follows:

Delay time, td: is the time is the time required for the response to reach the half
the final value the very first time.

Rise time, tr:

The rise time tr is the time taken for the response c(t) to rise from 0 to the
steady-state value 1. Thus:
Imaginary
n 1   2

e   n t r
1  1 sin d t r   
1 2 
n Real
Since, e   t  0
n r

30
 Mp
1.0

Amplitude

0.02
005
Allowable Tolerance
0.5

td
Time
tr

tp

ts

Fig Unit step response curve

Then sin d t r    0

Thus, d t r    


tr  (4)
d

We can thus reduce the rise time by increasing the damped natural frequency,
this value being determined by the undamped natural angular frequency n and
the damping ratio  .

Peak time, tp:

The peak time tp is the time taken for the response to rise from 0 to the first peak
value. As shown in Fig at the peak the tangent to the response c(t) is horizontal
i.e:

dc( t ) d    
 1  e n t  cos d t  sin d t 
dt dt   1  2 
  

31
dc( t )     d 
 n e n t  cos d t  sin d t   e n t   d sin d t  cos d t 
dt  1  2   1  2 
   

d n 1   2
Note that,   n , thus the cosine terms in the above equation
1  2 1  2

cancel each other thus.

dc( t )   
 n e n t  sin d t   d e n t sin d t
dt  1  2 
 

dc( t )  n  2 
  d  e n t sin d t
dt  1  2 
 

dc( t ) n
 e n t sin d t
dt 1  2

dc( t ) n  t
 e n p sin d t p  0
dt t t p 1  2

The above equation yield to the following condition

sin d t p  0 , so,

d t p  0, , 2, 3,

Since the peak time corresponds to the first peak overshoot, d t p   , This is the

time for the oscillating response to complete one half-cycle, i.e.  thus:


tp  5
d

32
Maximum overshoot Mp;

The overshoot is the maximum amount by which the response overshoots the
steady state value and is thus the amplitude of the first peak. The overshoot is
often written as a percentage of the steady state value. The steady state value is
when t tends to infinity and thus css=c(∞)=1. so Mp can be obtained as

M p  c t p   c    c t p   1

   
M p  1  e n t  cos d t  sin d t  1
  1  2 
  t t p  
d

 n   

   n 
 cos    sin    e d  e n
n
d 12 1 2
M p  e e
 1  2 
 

The maximum percent overshoot is

 
12
Mp  e  100% (6)

Note that the overshoot does not depend on the natural

frequency of the system but only on the damping ratio  .
1
As the damping ratio approaches 1 so the percentage
Mp
.

overshoot approaches zero as shown in figure.

0 1 
Settling time ts;

The settling time ts is used as a measure of the time taken for the oscillations to
die away. It is the time taken for the response to fall within and remain within
some specified percentage of the steady-state value (see Fig. 1 ). Thus for the

33
2% criteria settling time, the amplitude of the oscillation should fall to be less
than 2% of css. We have

 1 
c( t )  1  e n t   sin t  
 1  2  d
 

   
c( t )  k 1  e n t  cos d t  sin d t 
  1  2 
 

With Css=1, d  n 1   2 , the amplitude of oscillation is [c(t)-css] when c(t) is a

maximum value. The maximum value occurs when d t  , 2, 3,  and thus we
have

cos d t  1 and sin d t  0

Maximum amplitude of oscillation is

cmax  c( t )  css  en t  1

We are looking for the absolute of amplitude so

cmax   e  n t  1  e  n t

For the 2% criteria settling time, the settling time ts will occur when the
maximum amplitude is 2% of css i.e. 0.02 css. Thus

e  n ts  0.02

Taking logarithms gives

ln e  n ts  ln 0.02  3.9

 t  4 , thus the settling time is

n s

34
 4
t 
s 
for 2% tolerance criteria. (7)
n

3
t 
s 
for 5% tolerance Criteria. (8)
n

In designing a system the following are the typical points that are considered:

For a rapid response, i.e. small rise time, the natural frequency must be large.
Fig.(2) shows the types of response obtained to a unit step input to system
having the same damping factor of 0.2 but different natural angular frequencies.
The response time with the natural angular frequency of 10 rad/s, damped
frequency 9.7 rad/s, is much higher than that with a natural angular frequency of
1 rad/s, damped frequency 0.97 rad/s.

Fig.2 Unit step response of 2nd order system has the same damping ratio but
different natural frequency.

The damping factor is typically in the range 0.4 to 0.8 since smaller values
give an excessive overshoot and many oscillations before the system settles
down. Larger values render the system sluggish since they increase the
response time. Though, in some systems where no overshoot can be

35
 tolerated, a high value of damping factor may have to be used. Figure 4.4
shows the effect on the response of a second order system of a change of
damping factor when the natural angular frequency remains unchanged.

Fig. 3 Unit step response of 2nd order system have the same natural frequency
but different damping ratio.

Example 3

Consider the system shown in fig, where

R(s) E(s) k C(s)
+
  0.6 and   0.5 rad / sec . What is the rise s( m s  c)
n -
time t r , peak time t p , maximum overshoot

M p and settling time t s when the system is subjected to a unit step input.

Solution

Given system properties   0.6 , and n  0.5 rad / sec

Imaginary

1
1  2

d  n 1   2  4 rad / sec


1 2  Real
  tan 1  0.93 rad


36
  3.14  0.94
Rise time tr    0.55 sec .
d 4

 3.14
Peak time tp    0.785 sec
d 4

 
33.14

12
Maximum overshoot Mp  e 100%  e 4
 0.095  9.5%

4 4
Settling time t 
s 
  1.33 sec 2% criteria.
3
n

3 3
t 
s 
 1 sec 5% Criteria.
3
n

   wn
 
1 2
Mp  e  e wd
 0.095  9.5%

Example 4

For the system shown determine the values of gain k and velocity feedback
constant km so that the maximum overshoot in the unit step response is 0.2 and
the peak time is 1sec. obtain the rise time and settling time.

Assume B=1 Nm/rad/sec and J=1 kgm2.

R(s) 1 C(s)
+ k
+ Js  B S
- -

Km

37
Using the block diagram reduction the given control system can be reduced to a

single block with total transfer function as:

k
C s  G s  s s 1
 
R s  1  G s  H s  1 k
1 k ms 
s s 1

k k
  2
s  s  1 k  1  k m s  s   1  k m  s  k

C s  k
 2
R s  S  1 kk m  s  k

Comparing the above equation with the general form of second order system
yields;

2n  k , 2  n  1  k k m

The unknowns k, km can find by knowing the system properties  , and n as;

 
12
Mp  e  0.2


 1.61    0.456
1 2


tp   d  3.14 rad
d

d
d  n 1   2  n   3.53 rad / sec
1 2

Thus. k  2n  12.5

38
 2  n  1
2 n 1  k k m  km   0.178
k

 d
tr  , and   tan 1  1.1 rad
d n

3.14 1.1
tr   0.65 sec
3.14

4
t s tol 2%    2.42 sec
 n
2

3
t s  Tol 5%    1.26 sec
n

Example

Figure shows a mechanical vibratory system. When 2 Ib of force (step input) is

applied to system, the mass oscillates as shown. Determine the system constants
c, m, and k. from the corresponding response curve. The displacement y is
measured from the equilibrium position.

K
f(t)
y(t)
0.1
m
0.0095ft

y
c
t

Solution

The equation of motion is:

m y  c y  ky  f t 

39
 f (t )
y  2  ny  n2 y 
m

Y(s) 1/ m 1 k/m 1 2n

 2    
F(s) s  2 n s  2n k s 2  2 n s  2n k s 2  2 n s  2n

2
As; f ( t )  2Ib then F s  
s

2 2n
Ys    2
k s s  2 n s  2n

From the response curve y ss  0.1 ft thus by using the final value theorem the
constant k can be calculated as.

2
y ss  y     lims 0 sY( s )   0.1  k  20 Ib / Ft
k

From the response curve t p  2 , thus the damping frequency d can be

calculated as.

  
tp   2 d   1.57 rad / s
d d 2

The response shows that the maximum overshoot M p  0.0095 ft , thus the

damping ratio  can be obtained as.

  
  
1  2 1 2 1 2
Mp  e  0.0095  e  ln 0.0095  ln e


 4.656   
1  2

  0.6

40
The natural frequency n obtained as

d  n 1   2

d 1.57
n    1.96 rad / sec
1  2 1 0.6 2

Just the system properties n and  defined, all the system constant can be
obtained from the properties relations

k c
2n  & 2n 
m m

Thus ;

k
m  5.25 slug  166 Ib
2n

C  2  n m  12.21 Ib /( ft / sec)

3 Steady state error in unity feedback control systems

A unity feedback control system is R(s) E(s) C(s)
+ G(s)
shown in Fig.(4),where the feedback ‐
transfer function is unity H(s)=1. The error B(s)
H(s)
signal E(s) is represented by the difference
Fig.4. feedback control system
between the input signal R(s) and the output
signal C(s). The purpose of the controller is to reduce this error until the error
vanishes. In some cases the control system cannot reduce error signal E(s) to
zero. This will cause a steady state error ess which represent the offset between
the input and the output signals as shown in Fig(5). This error occurred due to
hysical properties of the system such as static friction,

41
backlash, and amplifier drift, as well as aging or deterioration. In this section
however, we shall investigate a type of steady state error that is caused by the
incapability of the system to follow particular types of inputs as step to improve
the control system.

ess
Output

offset
Time

Fig.(6.5) steady state error -unit impulse input

4 Steady state error

The closed loop transfer function for the unity feedback control system shown in
Fig.(4) can be written as

C(s) G (s )
 R(s)
+
E(s)
G(s)
C(s)
R (s) 1  H(s)G(s) ‐
B(s)
E s   R s   B s  H(s)

 R s   H s C s 

 R s   H s E(s) G s , thus.

E(s)  H s E(s) G s   R s 

E(s)1  H s G s   R s 

R s 
E s   (9)
1 H s G s 

Equation (9) stated that the system error depends on the input type R(s) and the
system properties G(s) and H(s)

42
By using the final value theorem the error can be calculated as

e ss  lim t   c t   lim s  0 s E s 

s R s 
ess  lim s  0 (10)
1 H s G s 

For unity feedback control system H(s)=1 the steady state error is

s R s 
ess  lims  0 (11)
1 G s 

Thus its clear from the above equation that the steady state error depends on the
type of input function R(s) and the system properties (characteristics) G(s).

4.1 Steady state error of the system with step displacement input

(Position error constant Kp)

System steady state error due to step-input signal r t   R  t  Where, R is the

R
displacement in (m) and Laplace transform is R s  
s

R
s
s R s  s  R
e ss  lim s  0  lim s  0
1  G s  1  G s  1  lim s  0 G s 

If we define positional static error constant

K p  lim s  0 G s  (12)

Then, the static steady state error can be written as.

R
ess  (13)
1 k p

43
4.2 Steady state error of the system with step velocity input

(Velocity error constant Kv)

System steady state error due to step-input signal r t   R t which have a

R
Laplace transform of R s   where, R measured in (m/sec)
s2

Using Equation (11) the steady state error can be written as.

R
s
s R s  s2  R R
ess  lim s  0  lim s  0 
1  G s  1  G s  lims  0 s  lims  0 sG  s  kv

If we define velocity error constant

K v  lim s  0 sG s  (14)

Then, the steady state error can be written as.

R
ess  (15)
kv

4.2 Steady state error of the system with step acceleration input

(Acceleration error constant Ka)

1 2
System steady state error due to step-input signal r t   R t which have a
2
R
Laplace transform of R s  
s3

where,

R: measured with the acceleration unit (m/sec)

Using Equation (11) the steady state error can be written as.

44
 R
s
s R s  s3  R
e ss  lim s  0  lim s  0
1  G s  1  G s  lims  0 s  lims  0 s 2 G  s 
2

If we define velocity error constant

K a  lims  0 s 2G  s  (16)

Then, the steady state error can be written as.

R
ess  (17)
ka

5 System characteristic and their effects on the steady state

error

We shall study effect of the system characteristic on the steady state error for
different input test signals (i.e. on the coefficients of the static error). The system
characteristic is represented by the open loop transfer function H(s)G(s) as
shown in Fig. (4). For a unit feedback control system H(s)=1 then the open loop
transfer function represented by G(s) that can be written as.

k 1 T1 s  1 T2 s  ............... 1 Tms 

G s  
s N 1 Ta s  1 Tb s  .............. 1 Tn s 

Where.

k, T1, T2, T3, …….., Tm and Ta, Tb, …………., Tn are constants and named as;

k: gain constant of the control system.

T: Time constant of the control system.

SN: poles at the origin.

45
N: the power of free S which defines the system type as.

For N=0 type- zero system.

N=1 type- one system.

N=2 type- two system and so on.

5.1 Type zero system

Where the transfer function in this case is

k 1 T1 s  1 T2 s  .........

G s  
s 0 1 Ta s  1 Tb s  ..........

For step displacement input

The position static error coefficient kp is.

R R
k p  lim s  0 G s   k  ess    constant
1 k p 1 K

For velocity input

The velocity static error coefficient kv is.

R R
k v  lim s  0 s G s   0  ess   
kv 0

For acceleration input

The acceleration static error coefficient ka is.

k a  lim s  0 s 2 G s   0
R R
 ess   
ka 0

We can conclude that performance of system type ZERO is very bad relative to
the steady state error

46
5.2 Type one system

Where the transfer function in this case is

k 1  T1 s  1  T2 s  ...........
G s  
s1 1 Ta s  1 Tb s  ...........

For step displacement input

The position static error coefficient kp is.

R R
k p  lims  0 G s     ess   0
1 k p 1 

For step velocity input

The velocity static error coefficient kv is.

R R
k v  lim s  0 s G s   k  e ss    constant
kv k

For step acceleration input

The acceleration static error coefficient ka is.

R R
k a  lim s  0 s 2 G s   0  e ss   
ka 0

5.3 Type-2 systems or higher

Where the transfer function in this case is

k 1 T1 s  1 T2 s  ...........

G s  
s 2 s  Ta s  1 Tb s  ............

For step displacement input

47
 The position static error coefficient kp is.

R R
k p  lim s  0 G s     e ss   0
1 k p 1 

For step velocity input

The velocity static error coefficient kv is.

R R
k v  lim s  0 s G s     ess   0
kv 

For step acceleration input

The acceleration static error coefficient ka is.

R R
k a  lim s  0 s 2 G s   k  ess    constant
ka k

Type-3 systems or higher

Where the transfer function in this case is

k 1  T1 s  1  T2 s  ...........
G s  
s 3 s  Ta s  1  Tb s  ............

For step displacement input

The position static error coefficient kp is.

R R
k p  lim s  0 G s     e ss   0
1 k p 1 

48
 For step velocity input (ramp)

The velocity static error coefficient kv is.

R R
k v  lim s  0 s G s     e ss   0
kv 

For step acceleration input (quadratic)

The acceleration static error coefficient ka is.

R R
k a  lim s  0 s 2 G s     ess   0
ka 

This is mean that type-3-system is the best one for steady state errors but in
other word as the power of the denominator increase the less steady state error
achieved which in turn leads to a good accuracy but effect on the stability of
the system as will be discussed later.

Table 6-1 summarizes the steady-state errors for type 0, type1 and type 2
systems when are subjected to various inputs

Step- input Ramp input Acceleration input

r t   R r t   R t 1
r t   R t2
2

Type- 0 system R
1 k
 

Type-1 system R
0 k


Type- 2 system R
0 0 k

49
Example 6

Determine the position, velocity, and acceleration error coefficient for the
following feedback control system with unity feedback. The open loop transfer
function is given by

100
G s  
1 0.1s  2  4s 
Solution

100 100 / 2 50
G s    
1 0.1s  2  4s  1 0.1s  1 2s  1 0.1s  1 2s 
The above open loop transfer function is zero-type-system. thus, by using
equations

1 1
k p  50  e ss    0.02
1  50 51

kv  0  e ss  

ka  0  e ss  

Example 7

Determine the position, velocity, and acceleration error coefficient for the
following feedback control system with unity feedback. The open loop transfer
function is given by

k1
G s   .
s s  5 s  2
2

50
 0.1 k1
G s  
Solution s 1  0.2s  1  0.5s 
2

kp    e ss  0

kv    e ss  0

1 10
k a  0 .1 k 1  e ss  
0.1 k1 k 1

Problems
1) Determine the delay time and the rise time for the following first order
systems:
a) G(s)= 1/ (4s + 1).
b) G(s)=5/ (s + 1).
c) G(s) =2/ (s + 3).
2) A first order system has a time constant of 30s. What will be its delay
time and rise time when subject to a unit step input?
3) A first order system when subject to a unit step input rises to 90% of its
steady state value in 20 s. Determine its time constant, delay time and rise
time?
4) Determine the natural angular frequency, the damping ratio, the rise time,
percentage overshoot and 2% settling time for systems with the following
transfer functions:
a) 100/(s2+4s+100).
b) 49/(s2+4s+49).
5) Determine the natural angular frequency, the damping factor, the rise
time, percentage overshoot and 2% settling time for a system where the
output y is related to the input x by the differential equation:

51
 d 2 c( t ) d c( t )
2
5  16 c( t )  16 r( t )
dt dt

6) For the feedback system shown in Figure, what gain K should be used to
give a rise time of 2s?

+ 1
k
- ss  1

7) Are the systems with the following transfer functions stable?

1 1 1
, , and
s 2  2s  1 s 2  2s  10 s  1s  3
8) Figure shows a feedback control system +
K G(s)
with unity feedback. Will the system be -

stable when

a) K=1, G(s) = 1/ [s(s + 1)],

b) K=3, G(s) = 1/ [s + 4)(s- 1)],
c) K=5, G(s) = 1/ [s + 4)(s- 1)]?
9) State if the following systems are stable, the relationship between input
r(t) and output c(t) being described by the differential equations:
d 2 c( t ) d c( t )
a) 2
3  2 c( t )  r( t )
dt dt
d 2 c( t ) d c( t )
b)   6 c( t )  r( t )
dt 2 dt

52
10) Determine the position, velocity, and acceleration error coefficients
for the following feedback control systems with unity feedback. The
open loop transfer functions are given by:
50
a) G (s ) 
(1  0.1s)(1  2s)
k
b) G (s ) 
s(1  0.1s)(1  0.5s )
k
c) G (s ) 
s (s  4s  200)
2 2

k (1  2s )(1  4s)
d) G (s ) 
s 2 (s 2  2s  10)
11) Determine also the steady-state errors for a unit step input, a unit ramp
input t and an acceleration input t2/2.
12) A servo mechanism is being designed to keep the antenna of tracking
radar pointed at a flying target. The servo system must be able to
follow a target traveling a straight-line course with speed up to 600
m.p.h. with a maximum permissible error of 0.01 degree. The shortest
distance from the antenna to the target is 1000 ft. Determine the value
of the velocity error coefficient kv to satisfy these specifications.
13) Find the impulse response for the system with the following transfer
0.4s  2.4
function G(s)  , plot the poles in s-domain.
0.2s2  0.4s  1
14) Find the steady state response for the system whose output is given
10
by: C(s) 
s(s  1)
15) Consider the system shown below. Determine the value of k such that
the damping ratio   0.5 Then obtain the rise time, peak time,
maximum overshoot and settling time in the unit step response.

53
 R(s) + + 16 1 C(s)

- s  0.8 s
-

16) Consider the unit-step response of a unity feedback control system whose
1
open loop transfer function is: G(s)  . Obtain the rise time, peak
s(s  1)

time, maximum overshoot, and settling time. (Use 2% criterion & 5%

criterion)
17) For the unit-impulse input, find the time response for the system with the
C(s) 0.4s  2.4
following transfer function,  . plot the poles in s-
R(s) 0.2s 2  0.4s  1

domain, and free hand plot the response in time domain.

18) Figure below is a block diagram of space-vehicle attitude – control
system. Assuming the time constant is T of the controller to be 3 sec and
2
the ratio of torque to inertia K J to be rad 2 / sec 2 , find the damping ratio
9
of the system.

C(s)
R(s) + 1
K(Ts+1)
_ Js 2
space
vehicle

19) Determine the values of K, and k of the closed loop system shown in
figure-3 so that the maximum overshoot in unit-step response is 25% and the
peak time is 2 sec. Assume R(s) + + C(s)
K J 1
_ _ s
that J=1 kg-m2. s
k

54
Basic control action and their effect on the
system response. (Methods to improve the system
performance)

1 Introduction

Controllers are components which basically have an input of the error signal,
i.e. until the difference between the required value signal and the feedback
signal, and an output of a signal to modify the system output.

The simplest form of controller is an on--off device which switches on some

correcting device when there is an error and switches it off when the error
ceases. However, such a method of control has limitations and often more
sophisticated controllers are used. While there are many ways a controller could
be designed to react to an error signal, a form of controller which can give
satisfactory control in a wide number of situations is the three-term or PID
controller. The term control mode is used for the type of response a controller
gives to an error signal and the three basic modes that are used are proportional
(P), integral (I) and derivative (D); the three-term controller is a combination of
all three modes.

The control problem can be stated as: “The control action signal u(t) will be
such that the controlled output c(t) will be equal to the reference input r (t) for all
values of time.

In practice, or as discussed in the previous sections there will always be transient

errors, but the transient period should be kept as small as possible. It is usually
possible to design the controller so that steady state errors are minimized, or
ideally, eliminated that is the reason why we use controllers.

45
 r(t) + e(t) u(t) c(t)
Amplifier Actuator Plant
set point -

Automatic Controller

Sensor

Fig.(1) Closed loop control system

A generalized closed-loop control system is shown in Fig.(1). With the

following terminology:

r(t): Reference input set point

e(t): Actuating error signal

u(t): Control action signal

2 Proportional control action. (P-controller)

The controller produces a control action u(t) that is proportional to the error e(t)
fig.(2), i.e. Controller output is

u( t )  e( t )  u ( t )  k p e( t )
controller
output

U( s )
U(s )  k p E (s )   kp time
E(s )
set point
Kp: The proportional gain constant.
error
variable

r(t) + e(t) u(t)

KP
- Fig.(2) controller output

proportional Controller

46
3 Effect of adding proportional controller on the performance
of feedback control systems.

Let us consider the second ordered system fig.(3) as an example for illustration.
When adding a proportional controller to its closed loop the new closed loop
control system can be drawn and simplified as:

R(s) E(s) +  2n C(s)

+ 1
‐ + s(s  2n )

Kp

Fig.(3) Block diagram of second order

system with proportional controller

R(s) 1  k 
p
2
n
C(s)
s2  2 s  1  k 
n p
2
n

Simplified block diagram

Comparing the total transfer function G(s) before and after adding proportional
gain kp is added, we conclude that:

The change in system properties happened only for the natural frequency of the
system. The new natural frequency can be written as:

n\  1  k p n
2 2
(1)

Thus the proportional controller is equivalent to amplifier with adjustable gain

for the natural frequency of the control system. Increasing Kp to minimize the
steady state error will, however, result in an increase in the natural frequency
n\ of the oscillatory output signal and a reduction in the damping ratio  \ , as

47
indicated by equations (1,23, &24), hence an increase in the overshoot Mp.
There is thus a problem with just using the proportional mode of control in that
to minimize the offset means increasing the overshoot and so a compromise has
to be reached.

4 Derivative Control Action. (D-controller)

The controller produces a control action that is proportional to the rate at which
de( t )
the error is changing fig.(4), i.e. controller output is
dt

de( t ) de ( t )
u( t )   u( t )  k d
dt dt

controller
output
U (s )
U(s )  k d sE (s )   kd s
E(s )
time
Kd: The derivative gain constant. set point

error
variable

r(t) + e(t) u(t)

Kd S
‐

derivative Controller Fig.(7.4) controller output

5 Effect of adding derivative controller on the performance of

feedback control systems.

When a derivative controller is added to the second order closed loop control
system, the block diagram can be drawn and simplified as;

48
 R(s) E(s) +  2n C(s)
+ 1
‐ + s(s  2n )

Kd S

Fig.(5) Block diagram of second order

system with derivative controller

R(s) 1  k d s 2n C(s)

 
s 2  2 n  k d  2n s   2n

Simplified block diagram

Comparing the total transfer function G(s) before and after adding derivative
gain kd, is added we conclude that:

The derivative controller improves the damping ratio 2 \ n  2n  k d 2n

can be written as:

1
 \    k d n  \  
2

Thus any change of the derivative gain leads to an increase in the damping
factor  \ , which in turn reduces the max overshoot M p that is good for the

system stability. Also you can see from the nominator that the derivative control
action added new zero 1  k d s  for the system which increase its stability.

Example;

A linear proportional servomechanism has an equivalent moment of inertia of

5kgm2 and damping coefficient of 58Nms/rad, and its motor give a Torque of
350Nm/rad of error. It was observed that the steady state error due to ramp input

49
is relatively large. To reduce this error to one tenth of its original value, the
motor constant is changed and a derivative control is added to provide the
original degree of damping. Find the new constant of the motor and the time
constant of the derivative control.

Solution:

J  7.5 kgm2 , C  58 Nms / rad ,

K  E(s)  T  350 E(s)  K  350 Nm

I. The original system:

 r (s) + E(s) 1  c (s)

k
Js  Cs
2
-
Motor
Load

The closed loop transfer function

k k
J s  Cs
2
2n
Ms    J 
k c k s 2  2n s  2n
1 2 s  s
2

J s  Cs J J

k c
The system properties are: 2n  , & 2n  .
J J

c c
Thus the damping ratio is:  
2Jn 2 Jk

k
The feed forward transfer function is: G s s   , thus For ramp input
Js  Cs
2

t   Rt the steady state error can be written as:

50
 K R R
k v  lim s  0 s G s   ,  ess   (I)
C kv K C

II. The Modified system

R(s) + E(s) + C(s)

1
1 K \
- + Js  Cs
2

KdS

The closed loop transfer function

k\
1  k ds  n\
2

M s  
\ J 
 c  k \ k d  k \ s 2  2 \ n\ s  n\ 2
s  
2
s 
 J  J

2 k\ c  k\kd
The new system properties are: n\  , & 2 \ n\  .
J J

c  k\kd
Thus the damping factor is:  
\

2 Jk \

The feed forward transfer function is: G s s  

1  k ds k \ , thus For ramp input
Js 2  Cs
t   Rt the steady state error can be written as:

K\ R R
k \v  lim s  0 s G s   ,  ess   (II)
C k v\ K \ C

ess\ 1 K
The first condition is: e  1 10 ess 
\
ss   \  K  3500 Nm / rad
\

ess 10 K

51
 C  K\kd C
The second condition is:    \

2 JK \ 2 JK

C  K \ 

Thus, the time constant of the derivative controller is; k d   1
K \  k 


Kd 
58
3500
 10 1  0.0358 sec
6 Integral Control Action. (I-controller) or (reset controller).

The controller produces a control

action that is proportional to the
integral of the error signal e(t)
controller
output

with time fig.(6), i.e. controller

output is

t t
u( t )   e( t )  u( t )  k i  e( t ) set point
error
variable

t 0 t 0

ki U( s ) k i
U(s )  E(s )  
s E( s ) s
Fig.(6) Integral controller output proportional to
the area under error‐time curve
Ki: The integral gain constant.

R(s) + E(s) ki U(s)

- s

Integral Controller

The integral of the error with time is the total area under the error-time graph up
to the time concerned and thus can have a value even when the error has
changed back to zero.

52
7 Effect of adding Integral controller on the performance of
feedback control systems.

When an integral controller is added to the second order closed loop control
system, the block diagram can be drawn and simplified as.

R(s) E(s) +  2n C(s)

+ 1
- + s(s 2 n )

ki
s

Fig.(7.7) Block diagram of second order

system with integral controller

R(s) C(s)
s  k i 2n
s 3  2 n s 2  2n s  k i  2n

Simplified block diagram

Comparing the total transfer function G(s) before and after adding integral gain
ki, is added we conclude that:

The integral controller increases the control system from type-2 to type-3 hence
it improve the accuracy of the system (ess). Thus the system may be stable or
unstable according to the value of constant integral gain ki. so to check the
system stability we calculate the critical constant integral gain kc by using Routh
stability test and if k i  k c then the system is unstable otherwise k i  k c the
system become stable.

53
8 Rate feedback controller

This type of controller can be achieved by adding differentiator sensor in the

feedback loop of the control system, so the controller output is the derivative of
the system output as:

d d
u( t )  e( t )  u( t )  k v e( t )
dt dt

U(s )
U (s )  k v s E (s )   k vs
E(s )

kv: The feedback rate gain constant.(velocity gain constant).

R(s) E(s) +  2n C(s)

+
‐ ‐ s(s 2n )

Kv s

Fig.(8) Block diagram of second order

system with rate feedback controller

R(s)  2n C(s)
 
s 2  2 n  k v  2n s   2n

Simplified block diagram

Comparing the total transfer function G(s) before and after adding feedback rate
controller we conclude that:

The change in the system properties happened only for the damping ratio  , the

new damping ratio  \ can be written as:

 \  2n  k v 2n

54
So, the rate feedback control action has the same effect as adding derivative
controller.

The basic three mode controller can be combined together to improve the system
performance as follows:

PD-controller: proportional plus derivative controller.

PI-Controller: proportional plus integral controller.

PID-controller: proportional plus integral plus derivative controller.

9 PD-controller

The output control action signal of PD- Proportional

controller shown in fig.(9) can be obtained as kp

E(s) + U(s)
de( t ) Derivative
u ( t )  k p e( t )  k d , +
dt kd s

Assume that the constant k d  k p  d , thus. Fig.(9).PD‐controller

de( t )
u ( t )  k p e( t )  k p  d ,
dt

Taking the Laplace transform and rearrange the equation yield.

U( s )
U(s )  k p 1   d s  E(s )   k p 1   d s 
E(s )

R(s) E(s) U(s)

+ k P 1   d s 
-

PD-Controller

Where,  d is the derivative time constant.

55
Note that the constants k p and  d are adjustable for tuning the system.

The addition of a derivative mode to a proportional controller modifies its

response to inputs. Fig.(10) shows the response to a ramp input. A PD controller
provides an element to the response which is largest when the rate of change of
the error is greatest and diminishes as it becomes smaller. Thus with a step
input, the controller output
rises faster when we apply the
P‐only
step input signal than with just
proportional control. With PD
time time
control, the output rises more
rapidly towards the steady-
P+D
state value and the overshoot D‐only
is reduced. Because the
derivative mode reduces time time
system oscillations, we can Fig.(10) PD controller output to a ramp input
increase the proportional gain
element to higher values than would be feasible with just the proportional mode
without the oscillations becoming too great a problem and so reduce the steady
state error. The derivative mode is never used alone because it is not capable of
maintaining a control signal under steady error conditions. It is always used with
the proportional mode and often additionally with the integral mode.

Proportional
10 PI-Controller
kp
E(s) + U(s)
The output control action signal of PI-controller
Derivative
shown in fig(11) can be obtained as +
ki/s

Fig.(11) PI‐controller

56
 t
u ( t )  k p e( t )  k i  e( t )dt ,
0

Assume that the constant k i  k p  i , thus.

t
u ( t )  k p e( t )  k p i   e( t )dt ,
0

Taking the Laplace transform and rearrange the equation yield.

 1  U(s )  1 
U(s )  k p 1   E(s )   k p 1  
  s
i  E(s )   s
i 

Where,  i is the integral time constant

R(s) + E(s)  1  U(s)

k p 1  
-  is 

PI-Controller

Note that the constants k p and  i are adjustable to make the controller perform

its job with required performance and maintain a good stability.

i : adjust the integral control action.

k p : affect both the proportional and integral parts.

The addition of an integral element to a proportional controller modifies the

response Fig.(11) removing the offset and giving a steady-state value the same
as the input set value. This elimination of offset is because the integral mode
gives a controlling response which is proportional to the area under the

57
error-time graph up to the
current point and so can give
a controller output signal
P‐only
even when the error has
become zero. With a PI
time time
controller, we can still have
a controller output with zero P+I
I-only
error and so there is no need
for offset. PI control tends to
time time
be used with systems where
load disturbances occur Fig.(11) PI controller output to a step input

frequently.

11 PID-Controller

The output control action signal of PID- Proportional

controller shown in fig.(12) can be obtained as kp

Derivative
t
E(s) + U(s)
de( t ) kds
u( t )  k p e( t )  k i  e( t )dt  k d +
0
dt Integral
ki/s
Assume that the constant k i  k p  i , thus.
Fig.(12).PID‐controller
t
u ( t )  k p e( t )  k p i   e( t )dt  k p  d
de( t )
,
0
dt

Taking the Laplace transform and rearrange the equation yield;

 1  U(s )  1 
U(s )  k p 1    d s  E(s )   k p 1    d s 
 is  E(s )  i s 

58
Note that the constants k p i and d are adjustable for tuning the system.

E(s)
R(s) +  1  U(s)
k p 1    i s 
-  is 

PID-Controller

The addition of the proportional mode to the integral mode removes the offset
and gives a steady-state value the same as the input set value; the addition of the
derivative mode enables a system to rise more rapidly to the steady-state value.

Example

The figure below shows a block diagram of a position control system

incorporating integral control and velocity feedback control. Determine the
open-loop transfer function of the system, If J=2 kgm2, C=28 Nms/rad and
k=450 Nm/rad, Find the value of kt which will raise the damping ratio to its
critical value, when there is no integral control. Also find the value of T at
which the system becomes just unstable and the frequency of oscillations.

 r (s) 1  c (s)
+ 1 + k +
+ - Js  Cs
2

-
1
k ts
Ts
Integral Controller Velocity feedback

1
G1  1 
Ts

59
 1
J s  Cs
2
1
G2   2
ks
1 2 t J s  C  k t  s
Js  Cs

k 1  T s 
G  K G1 G 2 
T s J s  C  k t 
2

When there is no integral control

k
G  kG 2 
J s  C  k t  s
2

k/J 2n
G  2
 C  kt  s  2  n s
s2   s
 J 

Thus the system properties and damping ratio can obtain as

k C  kt
2n  , 2  n 
J J

k C kt
for  1 , 2 
J J

k t  2 k J  C  2 450  2  28  32 Nm s / r

The solution for time constant for student by using Routh

stability conditions

Example

A linear servo mechanism has an equivalent moment of inertia of j=10-5kgm2

and a damping coefficient C=10-4Nm/(rad/sec). the system is stabilized by
means of acceleration feedback, so that its motor exerts a torque given by

60
 Tm  k 1 e  k 2c  Nm.

Where e  r  c  is the angular positional error in radians between the

reference input  r and the output c , k1=0.004 and k 2c is the additional
feedback signal. Draw the block diagram of the system and find:

a) The value of k2 so that the system will critically be damped.

b) The steady state error for a constant angular velocity r  1.26rad / sec .

What is the significance of this acceleration feedback? Explain how the system
is stabilized using this kind of feedback.

r + + Tm
1 c
k1
‐ + Js 2  Cs

k2s2

The closed loop transfer function can be obtained as Series transfer function
reduced to

K1
G1 s  
J s  Cs
2

K1
J s  Cs 2
The first minor loop eliminated to M1 s 
K1 K 2 s 2
1 2
J s C s

K1
M1 s  
J  K1 K 2 s 2  C s
Thus feed forward transfer function

61
 K1 /  J  K1 K 2  2n
G s  Equivalent to 2 , thus, the new
s 
2 C
s s  2 n s
J  K1 K 2
improved system properties are defined as:

C K1
2  \ n\  , n\ 
J  K1 K 2 J  K1 K 2

For critical damping  \  1

K1 C
2 
J  K1 K 2 J  K1 K 2

C2
4 K1 
J  K1 K 2

10  10    250 sec 2

1  C2  1  5 4 2 
K2   J   
K1  4 K1  0.004  4  0.004 


The steady state error

R
e ss  , R  1.26
Kv
K1 0.004
K v  lim s  0 s G s     40
C 10  4

1.26
e ss   0.0315
40

C C
2  \ n\  , 2  n 
J  K1 K 2 J

So,  \   Thus the system is stabilized by increasing the damping ratio.

62
Problems

1. A position control system has a moment of inertia of J=1000 (kg m2)

while the controller gain k=1100 (Nm/min) of misalignment as shown in
fig. If the system is critically damped by means of output rate feedback,
determine the steady state positional error due to an input speed of 6
(rev/min).If a derivative control is added to the system incorporating a
time constant Td=0.1(s) and the output rate feedback is adjusted to give
critical damping, determine the new steady state error.

 r (s) + 1  c (s)
k
Js  Cs
2
-
Controller
Load

2. Figure shows a control system designed to control the level of water in the
container to a constant level. It uses a proportional controller with Kp
equal to 10. The valve gives a flow rate of 10 m3/h per percent of
controller output, its flow rate being proportional to the controller input. If
the controller output is initially set to 50% what will be the outflow from
the container? If the outflow increases to 600 m3/h, what will be the new
controller output to maintain the water level constant?

3. A control system uses a proportional controller to control a system with a

transfer function of K and unity feedback. What will be the steady state
error if the proportional controller has a gain Kp of 10 and K=0.3 and a
step input of 4 units is applied?

63
4. A control system uses a proportional controller to control a system with a
transfer function of K and unity feedback. What should the gain Kp of the
controller be to give an offset error of 0.01 unit if K =0.1 and there is a
step input of 4 units to the system?
5. Figure shows a liquid level control system and its representation by a
block diagram. Determine the way controller
the output will vary with time if + 0 .2
the controller is (a) proportional - s  0.5
only with a proportional gain of 2,
(b) integral only with an integral gain of 2.
6. A closed loop control system has a PID controller with transfer function
Kp+(Ki/s)+Kds and is cascaded with a process having a transfer function
of 10/(s+5)(s+10). If the system has unity feedback, what is the transfer
function of the closed-loop system?

64