CHAPTER 7

Time Response Analysis of Linear

Systems

Learning objectives

• Use of appropriate mathematical tool in obtaining the time-response solution of a given system
model (transfer function or differential equation) for a specified input signal.

• Distinguish transient and steady state response mathematically as well as graphically.

• Correlate analytical results with the physical behaviour of the system.

7.1 Introduction
In preceding chapters we have discussed in detail about the various mathematical representation
for the dynamics of physical systems and eventually obtained its block diagram representation too.
The transfer function representation of a complex interconnected system expresses the input-output
relationship of the system under study in time domain or in frequency domain depending upon
whether the input signal applied to the system is time domain signal or a frequency domain signal.
In this chapter we will discuss the time-response analysis of linear systems represented by re-
spective transfer function model between the designated input and the output port. In general time
response means, when a linear system is subjected to a time-varying signal it produces an output/re-
sponse (response may be current, voltage, displacement or vibration depending upon the system
under consideration) which varies with time.
16 Time Response Analysis of Linear Systems

The variation of the response with time (or time response) of any linear system (or feedback
control system) has two distinctive parts in it:

i. Transient part-this part is called transient response.

ii. Steady-state part-this part is called steady state response.

Transient response of a system lasts for a very short duration of time less than or equal to the time-
constant of the system, whereas the steady state response comes into effect when transient response
ceases and it lasts for larger duration of time. Both the responses are significant for understanding
the behaviour and performance of the system.
The mathematical tools required for carrying out the time-response analysis of any physical system
from a given transfer function model are Laplace transform, inverse Laplace transform and Heaviside
partial fraction method. A brief outline of the mathematical treatment for finding time-domain
response of a system (represented by a transfer function) is shown in Fig. 7.1

Figure 7.1: Systematic mathematical work flow in obtaining time-domain response

7.1 Introduction 17

7.1.1 Illustration of mathematical work flow

Let us consider a simple electrical circuit (example of an LTI system) as shown in Fig. 7.2, and write
its dynamical equation. Let us assume v(t) as input and i(t) as the response due to input. Now,
using the physical law governing the given circuit (i.e. KVL) we can write:

Figure 7.2: Series R-L circuit

di(t)
v(t) = Ri(t) + L (7.1)
dt

Taking Laplace transform of (8.1) under zero initial condition on both the sides:

V (s) = RI(s) + LsI(s)

I(s) 1/L
= (7.2)
V (s) (s + R/L)

Assuming input signal to be v(t) = u(t) (i.e. unit step input) we can write I(s) as,

1/L
I(s) = × V (s)
(s + R/L)
1/L 1
I(s) = × (7.3)
(s + R/L) s

Carrying out the partial fraction of (8.3) we can write,

A B
I(s) = + (7.4)
s (s + R/L)

Finding the residues A and B by Heaviside method:

1/L
A = s× |s=0
s(s + R/L)
1
A =
R
18 Time Response Analysis of Linear Systems

and,

1/L
B = (s + R/L) × |s=−R/L
s(s + R/L)
1
B = −
R

Hence, I(s) can be written as,

1/R 1/R
I(s) = − (7.5)
s (s + R/L)

Now taking inverse Laplace transform of I(s) in (8.7) we can get time response as,

1 1
i(t) = − e(−R/L)t (7.6)
R R

The above time-domain response is shown in Fig. 7.3. The first term of the response refers to the
steady state value of the response (dependent on the type of input) which is obtained as:

Figure 7.3: Time-domain response of series R-L circuit

µ ¶
1 1
lim i(t) = iss = lim + e(−R/L)t
t→∞ t→∞ R R
1
iss =
R

The second term of the response refers to the transient response of the system (exponential function
in this case) and it lasts for a short duration (some fraction of seconds) and is dependent on the
7.1 Introduction 19

system poles, not on the the input signal.

7.1.2 Input signals used for time-response analysis

For practical control systems the use of actual input signal may not be known as a priori and such
actual signals if known cannot be expressed by simple mathematical equations, thus complicating
the entire analysis. As, the transient response is dependent upon the pole of the system and not
on the type of input, so simple time-varying signals can be used for the purpose of analysis which
characterizes certain attributes of the actual signals.
Some of the attributes of actual signal may be, sudden shock - characterized by impulse signal,
sudden change - step input, constant velocity - ramp input, constant acceleration - parabolic input.
Thus we can analyze the transient (time) response of linear system (or linear control system) with
any one of these input signals for understanding its behaviour. In view of the above discussion, such
input signals are referred as standard test signals.

1. Step input: Signal whose value changes from one level to another level in zero time. Mathemat-
ically it is expressed as,

r(t) = au(t)

where, r(t)=input signal and a=magnitude of the signal, and

u(t) = 1, t > 0
= 0, t < 0

Laplace transform of r(t) is:

a
R(s) =
s

The graph for step-input signal is shown in Fig. 7.4. This signal can be regarded as DC bias
in the system.

2. Ramp input: This signal increases linearly from zero with time. Mathematically it is expressed
as,

r(t) = ar(t)

where, r(t)=input signal and a=slope, and

r(t) = t, t > 0
= 0, t < 0
20 Time Response Analysis of Linear Systems

Figure 7.4: Unit-step input

Laplace transform of r(t) is:

a
R(s) =
s2

The graph for ramp-input signal is shown in Fig. 7.5. This signal can be generated by integrating
the step input signal.

Figure 7.5: Unit-ramp input

3. Parabolic input: This signal increases as a square of time from zero. Mathematically, it is
expressed as,

r(t) = ap(t)
7.1 Introduction 21

where, r(t)=input signal and a=coefficient, and

t2
p(t) = >0
2
= 0, t < 0

Laplace transform of r(t) is:

a
R(s) =
s3

The graph for parabolic-input signal is shown in Fig. 7.6. This signal can be generated by
integrating the ramp input signal.

Figure 7.6: Unit-parabolic input

4. Impulse input: This signal is defined only at t = 0 and it is zero at other values of time. It is
also called as δ-function. Mathematically, it is expressed as

r(t) = aδ(t)

where, r(t)=input signal and a=strength of the impulse, and

δ(t) = 1, t = 0
= 0, otherwise

Laplace transform of r(t) is:

R(s) = a

The graph for impulse-input signal is shown in Fig. 7.7. This signal can be generated by
22 Time Response Analysis of Linear Systems

differentiating the step input signal. In practice a pure impulse signal as shown in Fig. 7.7
cannot be generated, it is usually a pulse signal of very small width.

Figure 7.7: Unit-impulse input

7.2 Time Response of First-Order Control System

C(s)
Order of the control system is judged from its closed-loop transfer function R(s) by counting the
number of finite poles. A first-order closed-loop control system has only one closed-loop pole. Let us
consider a unity feedback control system shown in Fig. 7.8.

Figure 7.8: First order closed-loop system

The closed-loop transfer function of the above system is,

C(s) 1
= (7.7)
R(s) 1 + τs

Now let us examine the time response behaviour of the system (8.10) subjected to unit-step input.
7.2 Time Response of First-Order Control System 23

Unit-Step response: If the above system is subjected to unit-step input, then the transform-domain
response (C(s)) in (8.10) can be expressed as,

1
C(s) = × R(s) (7.8)
1 + τs

where, R(s) = 1s . The substitution of R(s) in (7.8) gives,

1 1
C(s) = × (7.9)
1 + τs s

Carrying out the partial fraction of (7.9) we can get,

1 1
C(s) = − 1 (7.10)
s s+ τ

Taking inverse Laplace transform of (7.10) on the both sides yields,

c(t) = 1 − e−t/τ (7.11)

The nature of the time-response of a first-order control system due to step input is shown in Fig.
7.9. Following time-response specifications of a first-order control system can be defined from the

Figure 7.9: Time-response of first order system

response curve,

1 Time Constant: The time at which the response reaches 63.2% of the final steady-state value.
It is denoted by τ and measured in seconds. The time constant can be evaluated from the
24 Time Response Analysis of Linear Systems

response curve by finding the slope of the response at t = 0, which is given as,

dc(t) d n o
|t=0 = 1 − e−t/τ |t=0
dt dt
1
=
τ

So, the time-constant of the system is reciprocal of the slope, i.e., τ Seconds. Time constant of
the system is an indication of the response speed of the system (how fast or slow the response
reaches certain percentage of the system). Larger time constant is an indication of sluggish
system and smaller time constant indicates that the system is faster. We know that series R-C
and R-L circuits are examples of first-order system, and their corresponding time-constants are
L
RC and R respectively. Thus, it is obvious that they are functions of physical parameters of
the system and it can be adjusted by varying the physical parameters of the system as desired.

2 Settling time: This is time required by the response to reach and stay thereafter within a specified
band called tolerance band (which is some small percentage of the final value). This is usually
2 to 4 % of the final value.

3 Steady-State Error: The difference between the reference input(desired output) and the actual
response (or regulated output) at a steady state (when t → ∞) is called steady state error.
Mathematically for the first order system subjected to step input it is expressed as,

e(t) = r(t) − c(t)

t
= 1 − (1 − e− τ )
t
= e− τ

So, steady state error is given by:

t
ess = lim e− τ = 0
t→∞

Hint:
The standard test signals can be derived from one another either by integrating or differentiating
original signal from one another.
For LTI system, as the input signals hold the relationship depicted in Fig. 7.10, the responses
due the inputs (ramp, step and impulse) also holds the same relationship as indicated in Fig. 7.11.
It must be noted that the response due to the integral of the original signal can be obtained by
integrating the response of the original signal and by determining the integrating constants from zero
output initial conditions.
Self Practice Question:
Find the time response of the first order system subjected to (i) Unit ramp input and (ii) Unit impulse
input. Also find the steady state error.
7.3 Time Response of a Second-Order Control System 25

Figure 7.10: Relationship between standard test signals

Figure 7.11: Relationship between responses due to standard test signals

7.3 Time Response of a Second-Order Control System

Let us consider a second-order unity feedback control system as shown in Fig. 7.12 below, with
2
ωn
prototype second-order forward path transfer function given as G(s) = s(s+2ζωn ) . The closed loop
26 Time Response Analysis of Linear Systems

Figure 7.12: Closed Loop second order system

transfer function for the system is given as,

C(s) ωn2
= (7.12)
R(s) s2 + 2ζωn s + ωn2

where, ωn = natural frequency of oscillation (rad/sec) and ζ = damping ratio of the system. The
characteristic equation of the system is determined by equating the denominator polynomial of the
overall transfer function to zero, which is expressed as,

s2 + 2ζωn s + ωn2 = 0 (7.13)

Upon solving the characteristic equation (7.13) we will get two characteristic roots and they are also
called the closed-loop poles of the transfer function. The nature of these roots tells us the behaviour
of the system. The roots of the characteristic equation (7.13) is given by,
p
−2ζωn ± 4ζ 2 ωn2 − 4ωn2
s1,2 =
p2
= −ζωn ± ωn ζ 2 − 1
p
= −ζωn ± jωn 1 − ζ 2
p
where, ωn 1 − ζ 2 = ωd is called the damped natural frequency of oscillation of the system (rad/sec).
Thus the roots are s1,2 = −ζωn ± jωd . The location of these complex conjugate roots in an s − plane
is shown in Fig. 7.13 below. For finding the response of closed-loop second-order system subjected
to unit-step input, we consider r(t) = 1, ∀ t ≥ 0 (∀ - for all) which upon taking the Laplace transform
gives R(s) = 1s . Thus the transfer function in (7.12) can be written as,

ωn2
C(s) = × R(s)
s2 + 2ζωn s + ωn2
ωn2
= (7.14)
s(s + 2ζωn s + ωn2 )
2
7.3 Time Response of a Second-Order Control System 27

Figure 7.13: Locus of complex conjugate root pair in an s-plane

Partial fraction of (7.14) gives,

1 s + 2ζωn
C(s) = − 2
s s + 2ζωn s + ωn2
1 s + 2ζωn 1 s + ζωn ζωn
= − = − −
s (s + ζωn )2 + ωd2 s (s + ζωn )2 + ωd2 (s + ζωn )2 + ωd2
1 s + ζωn ζωn ωd
= − − × (7.15)
s (s + ζωn )2 + ωd2 ωd (s + ζωn )2 + ωd2

Taking Inverse Laplace transform of (7.15) yields,

ζωn −ζωn t
c(t) = 1 − e−ζωn t cos ωd t − e sin ωd t (7.16)
ωd

Substituting the value of ωd in (7.16) equation gives,

e−ζωn t p
c(t) = 1− p { 1 − ζ 2 cos ωd t + ζ sin ωd t} (7.17)
1 − ζ2

From the Fig. 7.13 one can further express c(t) as,

e−ζωn t
c(t) = 1− p {sin θ cos ωd t + cos θ sin ωd t}
1 − ζ2
e−ζωn t
= 1− p {sin(ωd t + θ)} (7.18)
1 − ζ2
28 Time Response Analysis of Linear Systems

The above expression of c(t) is valid for 0 < ζ < 1, as for these values of ζ one can get complex
conjugate pair of roots. This response is referred as an under damped response and is shown in Fig.
7.14 Following points may be noted about the nature of this response:

Figure 7.14: Under damped response of second order system

i. The response represents exponentially decaying oscillations with frequency of oscillation determined
by ωd .

1
ii. The time constant of the exponentially decaying envelope is determined by ζωn .

iii. It is obvious from the expression of c(t) for second order system that it is influenced by damping
ratio ζ.

Further, we will discuss the effect of different values of ζ (damping ratio) in the nature of the response.

7.3.1 Nature of response for different values of ζ

Case I - The response of the second order system obtained from the expression of c(t) (under
damped case, i.e., 0 < ζ < 1) by substituting ζ = 0 in (7.18) is,

π
c(t) = 1 − sin{ωn t + }
2
= 1 − cos ωn t (7.19)

Such response for second-order system is called sustained oscillation/undamped response. One
can observe from the response graph in Fig. 7.15 that there are no decay in the amplitude of
7.3 Time Response of a Second-Order Control System 29

the response with time. This is due to the fact that for ζ = 0, the roots of second-order control
system (i.e., s1,2 ) the nature of the roots are imaginary.

Figure 7.15: Sustained oscillatory response of second order system

Case II - When ζ = 1, the response of the second order control system can be computed by
obtaining the nature of the roots s1,2 first and then finding the expression of c(t) from the
expression of C(s) discussed earlier.

s1,2 = −ωn + j0

The nature of the roots are found to be real and equal. Next, rewriting the expression of C(s),
we get,

ωn2
C(s) = (7.20)
s(s + ωn )2

Carrying out the partial fraction of (7.20) we get,

1 1 ωn
C(s) = − − (7.21)
s s + ωn (s + ωn )2

Taking inverse laplace transform of (7.21) gives,

c(t) = 1 − e−ωn t (1 + ωn t) (7.22)

The steady state value of this response can be found out from (7.22)using the following expres-
sion given below,

css = lim c(t) = 1 (7.23)

t→∞
30 Time Response Analysis of Linear Systems

Such a response for second order system is called critically damped response. One can observe
from the graph in Fig. 7.16 that there are no oscillations in the response. This is due to the
fact that for ζ = 1, the nature of the roots are real and equal.

Figure 7.16: Critically damped response of second-order system

Case III - When ζ > 1, the response of the second order control system can be computed by
obtaining the nature of the roots s1,2 first and then finding the expression of c(t) from the
expression of C(s) discussed earlier.
p
s1,2 = −ωn ± ωn ζ2 − 1

The nature of the roots are found to be real and distinct. Next, rewriting the expression of
C(s), we get,

ωn2
C(s) = p p (7.24)
s(s + ζωn + ωn ζ 2 − 1)(s + ζωn − ωn ζ 2 − 1)

Carrying out the partial fraction of (7.24 we get,

1 1
C(s) = − p p p
s 2 ζ − 1(ζ − ζ − 1)[s + (ζ − ζ 2 − 1)ωn ]
2 2

1
+ p p p (7.25)
2 ζ 2 − 1(ζ − ζ 2 − 1)[s + (ζ + ζ 2 − 1)ωn ]

Taking inverse laplace transform of (7.25) equation yields,

√ √ 2
ζ 2 −1)ωn t
−e(ζ− −e(ζ+ ζ −1)ωn t
c(t) = 1− p p + p p (7.26)
2 ζ 2 − 1(ζ − ζ 2 − 1 2 ζ 2 − 1(ζ + ζ 2 − 1
7.3 Time Response of a Second-Order Control System 31

The steady state value of this response can be found out using the expression,

css = lim c(t) = 1 (7.27)

t→∞

Such a response for second order system is called over damped response. The response is slower
than critically damped response and have no oscillations as the roots are real. The response
is shown in the Fig. 7.17. Due to the presence of two distinct real roots, one can observe two
time constants in (7.26),

Figure 7.17: Over damped response of second order system

1
τ1 = p
(ζ − ζ 2 − 1)ωn

and,

1
τ2 = p
(ζ + ζ 2 − 1)ωn

For a given ζ > 1, τ1 > τ2 , the response due to τ1 will be slower than that of τ2 . Consequently,
the response will be dominated by the slower time constant due to physical constraints of the
system.

7.3.2 Location of characteristic roots in ‘s-plane’ and its corresponding

time-response

Fig. 7.18 depicts the location of roots of a second order system in a s-pane for different ζ values and
their corresponding time-domain response.
32 Time Response Analysis of Linear Systems

Figure 7.18: Location of roots in s-plane & corresponding time-response

7.3.3 Unit-step under-damped response specifications of a second order

system
The transient (time domain) response for a second-order system subjected to unit-step input with
ζ < 1 (damping ratio) is called an under-damped response as shown in Fig. 7.14 above. We present
a detailed response curve in Fig. 7.19 for an under-damped second-order system to highlight some
of the response specifications. The response specifications for a system helps us to understand and
estimate the nature/behaviour of the system. Some of the significant response specifications are
defined below.

1. Peak Overshoot: The maximum positive deviation of the time-response from its desired (or
steady state) value is known as peak overshoot. It is designated as Mp . This specification is
expressed in percentage.

c(tp ) − c(∞)
%Mp = × 100
c(∞)

For unit-step input, css = c(∞)=1, so

c(tp ) − 1
%Mp = × 100
1
7.3 Time Response of a Second-Order Control System 33

Figure 7.19: Under-damped response

2. Delay time: It is the time taken by the response to reach 50% of the steady state (final) value
in its first attempt. It is denoted by td .

3. Rise time: It is the time taken by the response to rise from 0 to 100% of the final value. It is
denoted by tr .

4. Peak time: It is the time taken to reach the maximum value of the response for the first time.
It is denoted by tp . This is the time corresponding to the peak overshoot in the response curve.

5. Settling time: It is the time taken by the response to reach and stay within a specified
tolerance band (usually 2 to 5%) of its final value. The limit on the tolerance band is decided
by the user as per their desired application. It is denoted by ts .

6. Steady-state error: Difference between actual and desired value of the response at time,
t → ∞ is referred as steady-state error. It is denoted as ess .

ess = lim {r(t) − c(t)}

t→∞
34 Time Response Analysis of Linear Systems

7.3.4 Determination of response specifications

1 Determination of tp : We know that the transient response of an under-damped second order

system subjected to unit-step input is given by:

e−ζωn t
c(t) = 1− p {sin(ωd t + θ)}
1 − ζ2

for c(t) to be maximum, the condition is,

d
c(t) = 0
dt

So, differentiating c(t) with respect to ’t‘ , equating it to 0 and upon algebraic manipulation
one can get,

sin(ωd t) = 0
ωd t = n π, where, n = 0, 1, 2, ....

Since, the peak time corresponds to first overshoot in the response. The value of n = 1, thus
one can write i.e.,

ωd tp = π
π
tp =
ωd
p
where, ωd = ωn 1 − ζ2

2 Determination of Mp : For a unit-step input to the second order system, the peak overshoot as
per definition cited above can be expressed as,

Mp = c(tp ) − 1
e−ζωn tp
Mp = p {sin(ωd tp + θ)}
1 − ζ2

substituting the value of tp in the above equation one can get,

e−ζπ
%Mp = p × 100
1 − ζ2

3 Determination of tr : For under-damped response shown in the Fig. 7.19, the value of the
7.3 Time Response of a Second-Order Control System 35

response at that time is given by,

c(tr ) = 1
e−ζωn tr
1 = 1− p {sin(ωd tr + θ)} (7.28)
1 − ζ2

Upon solving (7.28) we get,

sin(ωd tr + θ) = 0
nπ − θ
tr = (7.29)
ωd

Now, n = 1 as this value is attained for the first time, thus the value is,

π−θ
tr = (7.30)
ωd
p
where, θ = cos−1 ζ and ωd = ωn 1 − ζ 2.

4 Determination of ts : For finding the settling time we need to consider the tolerance band as per
our desire or requirement first. Considering exponentially decaying envelope for 5% tolerance
band as shown in Fig. 7.19 we can write,

e−ζωn ts
1+ p = 1.05
1 − ζ2
e−ζωn ts
p = 0.05
1 − ζ2

1
In the above expression ζωn = τ , the time constant of the exponential decay. Hence, from the
above equation for ζ = 0.69 ts can be written as,
p
ln(0.05 1 − ζ 2 )
ts =
ζωn
ts ≈ 3τ

For 2% tolerance band, following the same procedure as above we get ts ≈ 4τ .

Self Practice Question:

Find the transient response of second-order system subjected to (i) unit-ramp and (ii) unit-impulse
input.

7.3.5 Solved Examples

S7.1: For the feedback control system shown in Fig. 7.20 below, determine the transfer function and
derive an expression relating the output time if the input is a step having magnitude of 2 units.
36 Time Response Analysis of Linear Systems

Figure 7.20: Block diagram

Soln.: Using block diagram reduction technique we can obtain a simplified block diagram rep-
resentation of closed loop system as shown in the Fig. 7.21 below. Next, finding overall transfer
function of the system,

Figure 7.21: Simplified block diagram

7.3 Time Response of a Second-Order Control System 37

C(s) 4
=
R(s) s2 + 8s + 12
4
=
s(s + 2) + 6(s + 2)
4
=
(s + 2)(s + 6)

2
Given that, R(s) = s so, the transform domain output equation becomes,

2 4
C(s) = ×
s (s + 2)(s + 6)
8
=
s(s + 2)(s + 6)

Partial fraction of the above equation gives,

A B C
C(s) = + +
s s+2 s+6

Finding the residues for the distinct roots as follows,

8 2
A = s× |s=0 =
s(s + 2)(s + 6) 3
8
B = s+2× |s=−2 = −1
s(s + 2)(s + 6)
8 −1
C = s+6× |s=−6 =
s(s + 2)(s + 6) 3

In view of the above partial fraction the C(s) is written as,

2 1 1
C(s) = − −
3s s + 2 3(s + 6)

Taking Inverse Laplace transform of the above equation one can get,

2 1
c(t) = − e−2t − e−6t
3 3

S7.2: A unity feedback control system has its open loop transfer function given by,

4s + 1
G(s) =
4s2

Determine the time-domain response when the system is subjected to input signals,

1. Unit-step input

2. Unit-impulse input

Soln.: Hints only,

38 Time Response Analysis of Linear Systems

C(s) s+0.25
1. For Unit-step input: First find overall transfer function from the given G(s), i.e, R(s) = (s+0.5)2 ,
next find out C(s) expression by substituting the value of R(s) = 1s , than carry out the partial
fraction and finally find the inverse Laplace transform of C(s). The time-response for this
system is c(t) = 1 − e−0.5t (2 − 0.5t).

2. For Unit-impulse input: The time response may be obtained by the similar procedure outlined
above or alternatively, one can simply find the derivative of c(t) expression of the unit-step
response. The reason is unit-impulse input can be obtained by differentiating the unit-step
input, so for the LTI system response can be obtained by simply differentiating the other
response, if their input are differentially related.

S7.3: The overall transfer function of a unity feedback control system is given by,

C(s) 20
=
R(s) s2 + 6s + 25

Determine the following, (i) natural frequency of oscillation (ωn ) (ii) damping ratio (ζ) (iii) damped
frequency of oscillation (ωd ) (iv) Peak overshoot (%Mp ) (v) Peak time (tp ) (vi) time at which the
first undershoot (means negative peak of the oscillation shown in Fig. 7.19) occurs (vii) time period
of oscillation and (ix) number of cycles completed before the response reaches the steady state.
Soln.: From the given closed-loop transfer function, we can have the characteristic equation by
C(s)
equating D(s) = 0 (where D(s) is the denominator polynomial of R(s) ),

s2 + 6s + 25 = 0

From the above characteristic equation, following quantities may be evaluated -

ωn2 = 25, ωn = 5rad/sec

2ζωn = 6, ζ = 0.6
ωd = 4 rad/sec

Now, evaluating tp ,

π
tp = p = 0.78 Sec
ωn 1 − ζ 2

− √ ζπ
Mp = e 1−ζ 2 = 9.4%

Time at which first undershoot occurs,

2π
t = p = 1.56Sec
ωn 1 − ζ 2
7.3 Time Response of a Second-Order Control System 39

Period of oscillation,

2π
T = = 1.56Sec
ωd

As steady state is reached in time 4τ for 2% tolerance band, so the number of oscillation completed
before reaching the steady state is given by,

4 ωd
= × = 0.84 cycle
ζωn 2π

S7.4: The maximum overshoot for a unity feedback control system having its forward path transfer
function as,

K
G(s) =
s(sT + 1)

is to be reduced from 60% to 20%. The system is subjected to unit-step input. Determine the factor
by which K should be reduced to achieve aforesaid reduction.

Soln.: Finding the closed-loop transfer function,

C(s) K/T
=
R(s) s2 + s/T + K/T

The characteristic equation is given by,

s K
s2 + + = 0
T T r
K 1 1
ωn = , 2ζωn = , ζ = √
T T 2 KT

Let, K1 be the forward path gain when Mp1 = 60% and the corresponding damping ratio is ζ1 and
similarly K2 be the forward path gain when Mp2 = 20% whose corresponding damping ratio is ζ2 .

−√1
ζ π
2
1−ζ1
Mp1 = e × 100

Upon solving we get the solution of ζ1 = 0.158. Similarly, for the second case one can calculate
ζ2 = 0.456. Assuming ‘T ’ to be constant, one can write,

1 1
ζ1 = ×√
2 K1 T
1 1
ζ2 = ×√
2 K2 T
√
ζ1 √K2 T K2 1
So, ζ2 = K1 T
, therefore, K1 = 0.12. Thus, K2 = 8 × K1 .

S7.5: By what factor the amplifier gain must be varied for an unity feedback control system with
40 Time Response Analysis of Linear Systems

an open loop transfer function given by,

K
G(s) =
s(1 + sτ )

where, T and K are constant having positive values. If (i) the peak overshoot of a unit step response
of the system has to be reduced from 25% to 10% and (ii) damping ratio has to increased from 0.2
to 0.707.

Soln.: Finding the closed-loop transfer function,

K
C(s) τ
= s
R(s) s2 + +K
τ τ
K/τ
=
s2 + s/τ + K/τ

From the above equation one can write the following,

p
ωn = K/τ , 2 ζ ωn = 1/τ

1
ζ =
2ωn τ

From this relation, one can express ζ in terms of K as follows,

1
ζ = √
2 Kτ
(7.31)

Hence the ratio between two ζ can be written in terms of forward path gains,
r
ζ1 K2
=
ζ2 K1

(i) Let damping ratio and forward path gain be ζ1 and K = K1 when Mp = 25% and similarly for
Mp = 10%, zeta2 and K2 be the corresponding damping ratio and the forward path gains respectively.
Now, finding ζ1 and ζ2 knowing the values of corresponding ζ’s.

√−πζ1 2
1−ζ1
Mp1 = e , ζ1 = 0.404
√−πζ2
2
1−ζ2
Mp2 = e , ζ2 = 0.5912

Now, we can write,

µ ¶2
K2 0.5912
= = 2.141
K1 0.404
7.3 Time Response of a Second-Order Control System 41

The amplifier gain K2 must be increased by an amount 2.141 times than that of K1 for reducing the
above mentioned peak overshoot.
(ii) From the relationship,
r
ζ1 K2
=
ζ2 K1

one may calculate that amplifier gain is to be reduced by 12 times to achieve the desired increase in
the damping ratio.
S7.6: Measurements conducted on a servomechanism show that the system response to be

c(t) = 1 + 0.667e−50t − 1.667e−20t

The system is subjected to unit-step input. Obtain the following,

1. Expression for the closed-loop transfer function in pole-zero form.

2. Identify the pole locations and forward path gain of the system.

3. Determine ωn and ζ of the system and identify the nature of the response.

Soln.: (i) Taking the Laplace transform of the c(t) expression given in the problem under zero
initial condition,

1 0.667 1.667 1
C(s) = + − , R(s) =
s s + 50 s + 20 s
C(s) 1000 1
= ×
R(s) (s + 50)(s + 20) s
(7.32)

(ii) From the expression of closed-loop transfer function one can find the pole locations as s =
−50, − 20 and forward path gain as 1000.
(iii) The characteristic equation of the system is,

s2 + 70s + 1000 = 0

Thus, comparing with the characteristic equation of a standard second order system s2 + 2ζωn s + ωn2
one can write,

ωn2 = 1000
ωn = 31.627 rad/sec
42 Time Response Analysis of Linear Systems

and

2ζωn = 70
70
ζ = = 1.1066
2 × 31.627

The value of the ζ indicates that the system has an over damped system response. The root or
the closed-loop pole that dominates the response is s = −20 has it will have higher time constant
compared to the pole located at s = −50.
S7.7: The closed-loop control system shown in Fig. 7.22 consists of error rate control. If the
0
damping ratio ζ is to be kept at 0.6, then determine the value of K . Compare the values of (i)
settling time for 2% tolerance band (ii) maximum overshoot and (iii) steady state error when the
input is unit-ramp for the system with and without error rate control and comment on the effect of
error rate control.

Figure 7.22: Block diagram

Soln.: With Error-rate control:

0
25(1 + sK )
G(s) =
s(s + 5)

The closed-loop transfer function is given by,

0
θo (s) 25 + 25sK
=
θi (s) s + s(5 + 25K 0 ) + 25
2

Comparing the denominator polynomial of the above equation with the standard characteristic equa-
tion of a second-order system one can write the following,

ωn = 5 rad/sec
0
2 ζ ωn = 5 + 25K
0 2×ζ ×5−5
K = = 0.04
25
7.3 Time Response of a Second-Order Control System 43

4 4
ts = = = 1.33 sec
ζωn 0.6 × 5n
√−ζπ
Mp = e 1−ζ2
= 5.26%
1 1 1
ess = = = = 0.2
Kv lims→0 sG(s) 5

Without Error-rate control:

25
G(s) =
s(s + 5)

The closed-loop transfer function is given by,

θi (s) 25
=
θo (s) s2 + 5s + 25

Comparing the denominator polynomial of the above equation with the standard characteristic equa-
tion of a second-order system one can write the following,

ωn = 5 rad/sec
2 ζ ωn = 5
5
ζ = = 0.5
2 × ωn

4 4
ts = = = 1.6 secs
ζωn 0.5 × 5
√−ζπ
Mp = e = 51.99%
1−ζ2

1 1 1
ess = = = = 0.2
Kv lims→0G(s) 5

Comments: Without error rate control ts and Mp are increased while the steady-state error of the
system is not affected compared to the system without this control.
S7.8: Fig. 7.23 below represents a schematic arrangement of an A.C position control system.
Following are the transfer functions of various associated sub-components that constitutes the control
system.

i Motor transfer function is

θ(s) Km
= , where Km = 20 rad/s/V &Tm = 0.2 s
Vc (s) s(Tm s + 1)

ii Gear ratio’s are,

θ˙f θ˙o 1
= 1, =
θ˙o ˙
θm 50
44 Time Response Analysis of Linear Systems

iii Synchro is used as an error detector which has a sensitivity Ks = 1 V olt/degree.

iv DC amplifier has a gain of Kdc V /V .

v Modulator gain is KM = 2 Vrms /Vdc .

vi AC amplifier has a gain Kac = 30 V /V .

The output is given to the control phase winding of the servo motor. Draw the block diagram
representation of the given system. Now, given two operational scenario for this control system,

1. Determine the value of Kdc such that the steady state error in the position is less than 3 degrees.
For this value of Kdc determine the ζ and settling time of the system for response to settle within
2% tolerance band.

2. If derivative term is added into the system, such that the modulator output which eventually
feeds input to the servomotor is given by:

de(t)
eA = Kdc e(t) + Kx
dt

Determine the value of Kx such that the damping ratio is improved to 0.6. Does this improve-
ment affect the steady-state error as in part (1)? Also calculate the modified ts and compare it
with the result in part (1).

Figure 7.23: Block diagram

7.3 Time Response of a Second-Order Control System 45

Soln: Will be produced soon.

S7.9: For the circuit shown in Fig. 7.25 below, find the value of the resistor such that 30%
overshoot appears across the voltage vc (t), when C = 2µF and L = 1H. Given that vin (t) is a
unit-step voltage.

Figure 7.24: R-C series circuit

Soln: Finding the expression for the voltage across capacitor ‘vc (t)’ as it is treated as a response
in this problem. Writing KVL equation for the circuit shown in Fig. 7.25,
Z t
di(t) 1
Ri(t) + L + i(t)dt = vin (t) (7.33)
dt C 0

Taking Laplace transform on both the sides of (7.33),

1 1
I(s){R + Ls + } = (7.34)
Cs s

Simplifying (7.34) we can write,

1
L
I(s) = Rs 1
(7.35)
s2 + L + LC

Writing the expression for Vc (S) in terms of I(s),

I(s)
Vc (s) =
Cs
1
LC
Vc (s) = Rs 1
(7.36)
s(s2 + L + LC )

The dynamics of the voltage across C can be analyzed from the denominator polynomial of Vc (s) i.e.,

Rs 1
s2 + + = 0 (7.37)
L LC

standard second-order characteristic equation is given by,

s2 + 2ζωn s + ωn2 = 0 (7.38)

46 Time Response Analysis of Linear Systems

Comparing (7.37) & (7.39) and further substituting the values of L & C we get,

1
ωn2 =
LC
ωn = 707.107rad/sec (7.39)

again,

R
2ζωn =
L
R
ζ = (7.40)
1414.214

Now, as per the question the system has 30% overshoot,

√ −ζπ 2
0.3 = e (1−ζ )
(7.41)

Solving (7.41) we get,

R = 506.13Ω (7.42)

S7.10: Given a mechanical rotational system shown in Fig. ?? below, find the value of J and D
to yield 10% overshoot and settling time ts = 1sec in an output angular displacement θ for a unit
step input in torque.

Figure 7.25: Rotational system

Soln:The transfer of the above system has been found out by the method described in Chapter 3
of this book,

1
J
G(s) = D K
s2 + Js + J

The characteristic equation of a standard second order system is given by,

s2 + 2ζωn s + ωn2 = 0
7.4 Steady-State Error 47

Comparing the coefficients of both the above equations we get,

p K D
ωn = ( ) & 2ζωn =
J J
4
In view of (7.43 and the given value of ts = 1sec we can find following values as ts = ζωn ,

D
= 8 (7.43)
J

From (7.43) & (7.43) we can compute ζ,

D
2ζωn = = 8
J
8 p J
ζ = = 4 ( ) (7.44)
2ωn K

Now considering 10% overshoot for the system given in the question,

√ −ζπ 2
0.1 = e (1−ζ )
(7.45)

Solving (7.45) we get,

ζ = 0.5912
p J
0.5912 = 4 ( )
K
J
= 0.022
K
J = 5.5 × 10−3 Kg − m2 , where, K = 0.25 N − m/rad (7.46)

Again, (7.43) we get,

D = 8 × 5.5 × 10−3
D = 0.044N − m − s/rad (7.47)

7.4 Steady-State Error

If the actual output of the system during steady state deviates from the reference input (desired
output), then the closed-loop control system is said to possess steady state error. It is an index of
accuracy in the transient response of the system. Steady state performance of the system is assessed
by the magnitude of the steady-state error.
Steady-state error is the difference between the reference input (desired output) and the actual
output(regulated output) of a feedback control system after the steady state has been reached, hence
it is assumed through out the discussion in the chapter that the system is stable. This assumption is
48 Time Response Analysis of Linear Systems

obvious as for a stable system the natural response tends to zero as time tends to infinity, whereas for
unstable system the natural response cannot reach steady-state value rather it becomes unbounded.
In practice the system stability must be checked before assessing the steady-state error of the system.
However, a detailed discussion on system stability will be dealt in the succeeding chapter.

7.4.1 Steady-state error for Unity Feedback System

For a closed loop control system shown in Fig. 7.26, we will deduce error transfer function by
considering error signal as output (E(s)) and input signal (R(s)).

Figure 7.26: Closed-loop control system

C(s) G(s)
=
R(s) 1 + G(s)H(s)
C(s) = G(s)E(s)
E(s) 1
=
R(s) 1 + G(s)H(s)

Now using Final value Theorem, let us find the value of steady state error ess ,

ess = lim e(t) = lim SE(s)

t→∞ s→0
R(s)
ess = lim S
s→0 1 + G(s)H(s)

From the above equation it is clear that ess depends upon R(s) and G(s). For a given system if R(s)
changes, the steady-state error also changes.
The illustration for the above fact is provided below by considering a first-order unity feedback
1
control system with G(s) = τs .

1 Steady-state error due to unit-step input: The transient response is given by,

t
c(t) = 1 − e− τ
7.4 Steady-State Error 49

Error for the system at time ‘t’ is given by,

t
e(t) = r(t) − c(t) = e− τ

So, the steady state error will be,

t
ess = lim e− τ = 0
t→∞

The response is illustrated in Fig. 7.27

Figure 7.27: Response & ess curve for unit-step input

2 Steady-state error due to unit-ramp input: The transient response for same system to ramp-input
is given by,

t
c(t) = t − τ (1 − e− τ )

Error for the system at time ‘t’ is given by,

t
e(t) = r(t) − c(t) = τ (1 − e− τ )

So, the steady state error will be,

t
ess = lim τ (1 − e− τ ) = τ
t→∞
50 Time Response Analysis of Linear Systems

The response is illustrated in Fig. 7.28

Figure 7.28: Response & ess curve for unit-ramp input

This shows that for a given system, ess is dependent on R(s) (input of the system).

In the next section, we will present different classification of feedback control system and carry
out detailed analysis of steady state error of different types of feedback control system highlighting
the effect of R(s) and G(s)H(s) (open-loop control system for non unity feedback control system) or
G(s) (open-loop control system for unity feedback control system) on ess .

7.4.2 Types of Feedback Control System

The classification of feedback control system is based on the presence of number of integrators in the
open-loop transfer function of the system G(s)H(s) (for non unity feedback control system) or G(s)
(for unity feedback control system). For identifying the presence of the number of integrators in a
closed-loop control system, G(s)H(s) or G(s) should be expressed in time-constant format instead of
pole-zero format as represented below,

K(1 + sTa (1 + sTb ).......

G(s)H(s) =
sn (1 + sT1 )(1 + sT2 ).....

where, K = forward path gain, − T1a , − T1b ...... are the zeros, − T11 , − T12 ..... are the poles and n is the
number of poles at the origin (or equivalently the number of integrators).
7.4 Steady-State Error 51

Now, depending upon the value of n in the above open-loop transfer function one can have three
different types of feedback control system,

1. For n = 0, Type-0 feedback control system, indicates presence of no integrator in G(s)H(s).

2. For n = 1, Type-1 feedback control system, indicates presence of 1 integrator in G(s)H(s).

3. For n = 2, Type-2 feedback control system, indicates presence of 2 integrators in G(s)H(s).

The justification for considering the value of n = 0,1,2 is that all the control system have the capability
k
of responding with zero steady state error to all polynomial inputs of the form r(t) = ( tk! )u(t), where
u(t) = 1, ∀t ≥ 0. These are treated as standard test signals for performance measurement of the
linear system. For k = 0, it is a step input; for k = 1, its is a ramp input and for k = 2, it is a
parabolic input which corresponds to position, velocity and acceleration inputs, respectively for any
physical problems.

7.4.3 Steady-State Error Constants

Steady state error constants are the measure of steady state error of a given system for a specified
input. As steady state error is calculated during steady state period, the error is also called as static
error. Static error constants helps in deciding how steady state error can be reduced or eliminated.
As discussed in section 7.4.1, ess is expressed as -

ess = lim e(t) = lim SE(s)

t→∞ s→0
R(s)
ess = lim S
s→0 1 + G(s)H(s)

In this section, we will derive the expressions of steady state error (static error) constants from
the above equation for three different inputs, such as position (unit-step), velocity (unit-ramp) and
acceleration (unit-parabolic). These error constants are function of the open-loop transfer function
G(s)H(s) of the system. One can infer that, ess is affected by changing input for a given system, or
for a given input, ess tends to be different for different types of feedback system.

[Link] Static error constants & ess due to different inputs

Here, we compute the static error constants and the steady state error due to different inputs applied
to the system.

i. Positional error constant: It is evaluated for unit-step input, i.e, R(s) = 1s . Substituting the
value of R(s) in the expression of ess , we get,

1
ess = lim
s→0 1 + G(s)H(s)
1
ess =
1 + lims→0 G(s)H(s)
52 Time Response Analysis of Linear Systems

where, lims→0 G(s)H(s) = Kp . Thus one can write ess in terms of Kp as,

1
ess =
1 + Kp

1
ii. Velocity error constant: Velocity error constant is defined for unit ramp input, i.e, R(s) = s2 .
Substituting the value of R(s) in the expression of ess one can get,

1
ess = lim
s→0 s(1 + G(s)H(s))
1
ess =
lims→0 s(G(s)H(s))

where, lims→0 s(G(s)H(s)) = Kv . Thus one can write ess in terms of Kv as,

1
ess =
Kv

iii. Acceleration error constant: Acceleration error constant is defined for unit parabolic input,
1
i.e, R(s) = s3 . Substituting the value of R(s) in the expression of ess one can get,

1
ess = lim
s→0 s2 (1 + G(s)H(s))
1
ess =
lims→0 s2 (G(s)H(s))

where, lims→0 s2 (G(s)H(s)) = Ka . Thus one can write ess in terms of Ka as,

1
ess =
Ka

[Link] Static error constants & ess due to different types of systems

Here, we compute the static error constants and the ess by considering different types of feedback
control systems subjected to different inputs.

i. Type-0 System: For type-0 system there is no integrator present in the expression of G(s)H(s),
K(1+sTa )(1+sTb )...
i.e, G(s)H(s) = (1+sT1 )(1+sT2 ).... , thus one can have following error constants and correspond-
ing ess .
1
a. Kp = lims→0 G(s)H(s), hence Kp = K, ess = 1+K .
1
b. Kv = lims→0 s(G(s)H(s)), hence Kv = 0, ess = Kv = ∞.
2 1
c. Ka = lims→0 s (G(s)H(s)), hence Ka = 0, ess = Ka = ∞.

ii. Type-1 System: For type-1 system there is one integrator present in the expression of G(s)H(s),
K(1+sTa )(1+sTb )...
i.e, G(s)H(s) = s(1+sT1 )(1+sT2 ).... , thus one can have following error constants and correspond-
ing ess .
7.4 Steady-State Error 53

K 1
a. Kp = lims→0 G(s)H(s), hence Kp = 0 = ∞, ess = 1+∞ = 0.
1
b. Kv = lims→0 s(G(s)H(s)), hence Kv = K, ess = K.
1
c. Ka = lims→0 s2 (G(s)H(s)), hence Ka = 0, ess = 0 = ∞.

iii. Type-2 System: For type-2 system there are two integrators present in the expression of
K(1+sTa )(1+sTb )...
G(s)H(s), i.e, G(s)H(s) = s2 (1+sT1 )(1+sT2 ).... , thus one can have following error constants and
corresponding ess .
K 1
a. Kp = lims→0 G(s)H(s), hence Kp = 0 = ∞, ess = 1+∞ = 0.
K 1
b. Kv = lims→0 s(G(s)H(s)), hence Kv = 0 = ∞, ess = K = 0.
c. Ka = lims→0 s2 (G(s)H(s)), hence Ka = K, ess = ∞.

We will summarize the result of this sub-section in the form of table for clarity in interpreting the
results.

Table 7.1: Steady-state errors for different inputs & systems

Steady state error
Input type Type-0 Type-1 Type-2
1
Unit-Step 1+Kp 0 0
1
Unit-Ramp ∞ Kv 0
1
Unit-Parabolic ∞ ∞ Ka
Kp = lims→0 G(s)H(s) Kv = lims→0 sG(s)H(s) Ka = lims→0 s2 G(s)H(s)

7.4.4 Solved examples

S7.11: A unity feedback control system has an open-loop transfer function given below. Find the
static error constants due to unit-step, ramp and parabolic inputs.

25
G(s) =
(1 + 0.2s)(1 + 5s)

Soln:

1. Positional error constant:

Kp = lim G(s)
s→0
25
Kp = lim = 25
s→0 (1 + 0.2s)(1 + 5s)

2. Velocity error constant:

Kv = lim sG(s)
s→0
25 s
Kv = lim = 0
s→0 (1 + 0.2s)(1 + 5s)
54 Time Response Analysis of Linear Systems

3. Acceleration error constant:

Ka = lim s2 G(s)
s→0
25 s2
Ka = lim = 0
s→0 (1 + 0.2s)(1 + 5s)

S7.12: Determine the error coefficients and static error for the system described by the following
transfer functions.

1
G(s) =
s(s + 1)(s + 10)
H(s) = (s + 2)

Soln: Finding the open-loop transfer function,

(s + 2)
G(s)H(s) =
s(s + 1)(s + 10)

The system is identified as Type-1 system from the open-loop transfer function. Now for this type of
system using standard formulae we find the error coefficients first,

1. Positional error constant:

Kp = lim G(s)H(s)
s→0
(s + 2)
Kp = lim = ∞
s→0 s(s + 1)(s + 10)

2. Velocity error constant:

Kv = lim sG(s)H(s)
s→0
s(s + 2) 1
Kv = lim =
s→0 s(s + 1)(s + 10) 5

3. Acceleration error constant:

Ka = lim s2 G(s)H(s)
s→0
s2 (s + 2)
Ka = lim = 0
s→0 s(s + 1)(s + 10)

Now, finding the steady state error using the error constants computed above,
1
1. Unit-step input: ess = 1+Kp = 0

1
2. Unit-ramp input: ess = Kv = 5
7.4 Steady-State Error 55

1
3. Unit-parabolic input: ess = Ka = ∞

S7.13: For a unity feedback system the open-loop transfer function is given by

K
G(s) =
s(s + 5)(s + 25)

Determine the static error coefficient and also the minimum value of K for which the steady state
error is less than 0.5 for unit-ramp input.
Soln: It is a Type-1 system, the error constants are as follows,

1. Positional error constant:

K
Kp = lim G(s) = lim = ∞
s→0 s→0 s(s + 5)(s + 25)

2. Velocity error constant:

sK K
Kv = lim sG(s) = lim =
s→0 s→0 s(s + 5)(s + 25) 125

3.

s2 K
Ka = lim s2 G(s) = lim = 0
s→0 s→0 s(s + 5)(s + 25)

Given that, ess < 0.5, thus for finding the value of K for unit-ramp input one can write,

1
< 0.5
Kv
Kv > 2
K
> 2
125
K > 250

The minimum value of K should be 250.

S7.14: The transfer function of a D.C motor for position control is given by,

Θ(s) K
P (s) = = rad/V olt
V (s) s[(Js + b)(Ls + R) + K 2 ]

Where the values of L is neglected, J = 3.2284 × 10−6 Kgm2 , b = 3.5077 × 10−6 N ms, K = 0.0274 &
R = 4Ω. The system is made closed-loop with unity feedback and cascading a proportional controller
of gain KP = 10−1 with the D.C. motor. Evaluate the static error constants and its corresponding
steady error.
Soln: The closed-loop block diagram arrangement for the system in the question is shown in the
Fig. 7.29 below. As the value of inductance of the motor armature is neglected i.e., L = 0, the motor
56 Time Response Analysis of Linear Systems

Figure 7.29: D.C motor position control with proportional controller

transfer function becomes,

Θ(s) K
P (s) = = rad/V olt (7.48)
V (s) s[(Js + b)(R) + K 2 ]

Substituting the values of J, b, K, KP & R, the open-loop transfer function of the motor control
system becomes,

0.0274
G(s)H(s) = 10−1 ×
s(12.9136 × 10−6 s + 14.1059 × 10−6 )
0.0274 × 10−1 1
G(s)H(s) = −6
× (7.49)
14.1059 × 10 s(1 + 0.9155s)

From (7.49) it is clear that the system is Type-1 system. Finding the static error coefficients,

Kp = lims→0 G(s)H(s)
Kp = ∞

Kv = lims→0 sG(s)H(s)
Kv = 1.94245 × 102
Kv = 194.245

Ka = lims→0 s2 G(s)H(s)
Ka = 1.94245 × 102
Ka = 0

Now, computing steady state errors, Steady-state error due to step input can be found from Kp ,

1
ess = = 0
1 + Kp
7.4 Steady-State Error 57

Steady-state error due to ramp input can be found from Kv ,

1
ess = = 5.148 × 10−3
Kv

Steady-state error due to parabolic input can be found from Ka ,

1
ess = = ∞
Ka

7.4.5 Steady-State Error for Nonunity Feedback System

Usually feedback control system do not have unity gain feedback path rather consists of some pure
gain or dynamics due to the use of compensator or due to physical model of the system. But a
non unity feedback control system can always be converted into an equivalent unity feedback control
system using block diagram reduction technique.
The method of transforming a nonunity feedback control system into unity feedback control system
will be briefly discussed here.
Let us consider a nonunity feedback control system with feedback path transfer function H(s) as
shown in Fig. 7.30.

Figure 7.30: Nonunity feedback control system

In Fig. 7.30, error is not the difference between input R(s) and output C(s), thus we refer the
signal at the input of the plant (after the summing junction) as actuating signal Ea (s). Now, we are
to show that E(s) = R(s) − C(s) which is the case for unity feedback control system. To accomplish
this task, use block diagram reduction rules in Fig. 7.30 to equivalently convert this block into an
unity feedback control system. The conversion steps are illustrated below.
From Fig. 7.30, one can write the expression for Ea (s) as,

Ea (s) = R(s) − C(s)H(s)

58 Time Response Analysis of Linear Systems

Further one can write this above equation equivalently as,

Ea (s) = R(s) − C(s) [(H(s) − 1) + 1]

The block diagram representation of the above equation is shown in Fig. 7.31. Using simple manip-

Figure 7.31: Block diagram reduction step 1

ulation one can redraw the block diagram shown in Fig. 7.32. Further one can write the expression

Figure 7.32: Block diagram reduction step 2

7.4 Steady-State Error 59

of E(s) in terms of Ea (s) now as,

Ea (s) = R(s) − C(s) [(H(s) − 1) + 1]

= R(s) − C(s) − C(s) [H(s) − 1]
Ea (s) = E(s) − C(s) [H(s) − 1] , ∵ E(s) = R(s) − C(s)

The last expression can be realized as shown in Fig. 7.33 shown below, the block diagram reduction

Figure 7.33: Block diagram reduction step 3

of Fig. 7.33 gives an equivalent unity feedback control system as shown in Fig. 7.34 with the forward
path transfer function as,

G(s)
Ge (s) =
1 + G(s)H(s) − G(s)

After this equivalent transformation, one can apply the steady state error analysis as discussed in

Figure 7.34: Block diagram reduction step 4

Section 7.4.1.
60 Time Response Analysis of Linear Systems

7.4.6 Solved Examples

S7.15: For a nonunity feedback control system shown in Fig. 7.35, determine the system type and
steady-state error for unit-step input. Assume that the input and output are same.
Soln: It seems from the open-loop transfer function G(s)H(s) that the system type is 1, but this

Figure 7.35: Unity feedback control system

inference is not correct due to the presence of the feedback path dynamics H(s) as indicated in Fig.
7.35. The implication of this assertion is that R(s) 6= C(s), thus to express E(s) explicitly as the
difference between R(s) and C(s), we transform the nonunity feedback system into an equivalent
feedback systems as discussed above earlier.

50
G(s) =
s(s + 20)
1
H(s) =
s+2

The equivalent forward path transfer function,

G(s)
Ge (s) =
1 + G(s)H(s) − G(s)
50(s + 2)
=
s + 22s2 − 10s − 50
3

The system type is ’0‘, since there is no pure integrator in the Ge (s). Now, finding Kp ,

Kp = lim Ge (s) = −2
s→0
1
e(∞) = = −1
1 + Kp

Negative value of e(∞) means that, output is larger than that of input steps.
7.5 Error series & Dynamic error coefficients 61

7.5 Error series & Dynamic error coefficients

The static error coefficients (Kp , Kv , Ka ) computed above for different types of system reveals that,
the value of steady-state error is non-zero, or zero, or infinity. Thus the inherent difficulty with the
static error coefficient is that error cannot be computed at any instant of time other than at t → ∞.
Thus, variation of error with time can be computed by finding dynamic error coefficients and hence
express error as a function of time, which is referred as error series.
Let us consider a unity feedback control system as shown in Fig 7.36. The error transfer function
for the system is,

Figure 7.36: Unity feedback control system

E(s) 1
= = We (s)
R(s) 1 + G(s)
Z ∞
We (s) = we (t)e−sτ dτ
0

Again, one can express,

E(s) = We (s)R(s)

Now, using convolution integral theorem, one can find the inverse Laplace transform of E(s),
Z t
e(t) = we (τ )r(t − τ )dτ
−∞
Z t
= we (τ )r(t − τ )dτ, f or only t ≥ 0
0

Expanding r(t − τ ) using Taylor’s series one can get,

τ2
r(t − τ ) = r(t) − τ ṙ(t) + r̈(t) − .......
2!
62 Time Response Analysis of Linear Systems

Substituting the value of r(t − τ ) in the expression of e(t) gives,

Z t
τ2
e(t) = we (τ ){r(t) − τ ṙ(t) + r̈(t) − ......}dτ
0 2!
Z t Z t Z t
τ2
= r(t) we (τ )dτ − ṙ(t) τ we (τ )dτ + r̈(t) we (τ )dτ − .......
0 0 0 2!

Now, one can write e(t) as,

e(t) = r(t)C0 − ṙ(t)C1 + r̈(t)C2 − .......

where,
Z t
C0 = we (τ )dτ = lim We (s)
0 s→0
Z t
d
C1 = − τ we (τ )dτ = lim We (s)
0 s→0 ds
Z t 2
τ d2
C2 = we (τ )τ = lim 2 We (s)
0 2! s→0 ds

Similarly by induction one can easily write,

dn
Cn = lim We (s), n = 0, 1, 2...
s→0 dsn

Hence, C0 , C1 , C2 , .....Cn are the dynamic error coefficients present in the error series e(t).

7.5.1 Relation between Static & dynamic error coefficients

Finding the Laplace transform for the error series e(t) deduced above, we get following expression,

E(s) = (C0 + sC1 + s2 C2 + ......)R(s)

The above expression is valid for Type-0 system in the transform domain as the denominator of
G(s)H(s) does not contain any integrator. Now, for unit-step input R(s) = 1s , the steady state error
can be calculated as,

ess = lim sE(s) = C0

s→0

Comparing this ess with the ess of Type-0 system obtained earlier for unit-step input gives us the
relation,

1
C0 =
1 + Kp
7.5 Error series & Dynamic error coefficients 63

For Type-1 system, as the G(s)H(s) expression has one integrator in the denominator polynomial,
so the expression of E(s) will be of the form,

E(s) = (sC1 + s2 C2 + ......)R(s)

Now, for unit-ramp input the steady state error for Type-1 system is obtained as,

ess = lim sE(s) = C1

s→0

Comparing this ess with the ess of Type-1 system obtained earlier for unit-ramp input gives us the
relation,

1
C1 = , & C0 = 0
Kv

For Type-2 system, as the G(s)H(s) expression has two integrators in the denominator polynomial,
so the expression of E(s) will be of the form,

E(s) = (s2 C2 + ......)R(s)

Now, for unit-parabolic input the steady state error for Type-2 system is obtained as,

ess = lim sE(s) = C2

s→0

Comparing this ess with the ess of Type-2 system obtained earlier for unit-ramp input gives us the
relation,

1
C2 = , C1 = 0, & C0 = 0
Ka

7.5.2 Solved examples

S7.16: The error transfer function for a closed-loop control system is given by

5s + 1
We (s) =
5s + K + 1

Find the generalized error coefficient for the system. Write down the error series for at least three
terms for the following inputs,

1. r(t) = u(t)

2. r(t) = t. u(t)
t2
3. r(t) = 2 u(t)
£ α3 2
¤
4. r(t) = α1 + α2 t + 2 t + e−α4 t u(t)
64 Time Response Analysis of Linear Systems

Soln: We know that generalized or dynamic error series is expressed as,

e(t) = C0 r(t) + C1 ṙ(t) + C2 r̈(t) + . . .

As only three terms need to be considered in the series, so three coefficients need to be computed
(C0 , C1 , C2 ) which is shown below,

1
C0 = lim We (s) =
s→0 K +1
· ¸
d
C1 = lim We (s)
s→0 ds

5K
=
(K + 1)2
· 2 ¸
d
C2 = lim We (s)
s→0 ds2

−50 × K
=
(K + 1)3

Thus the generalized error series for the given inputs are as follows,

1 When input is r(t) = u(t), then the error series as mentioned above will have one term as derivative
terms in this case vanishes and it is given by,

1
e(t) =
K +1

2 When input is r(t) = t u(t), then the error series as mentioned above will have two terms as second
derivative terms in this case vanishes and it is given by,

1 5K
e(t) = t+
K +1 (1 + K)2

t2
3 When input is r(t) = 2! u(t), then the error series as mentioned above will have all the terms and
it is given by,

1 t2 5K 50K
e(t) = + t−
K +1 2 (1 + K)2 (1 + K)3

£ α3 2
¤
4 When input is r(t) = α1 + α2 t + 2 t + e−α4 t u(t), then the error series will have terms due to
first three types of input of r(t). As the exponential term decays to zero at steady state hence
7.6 Sensitivity 65

it is not included in the error series. Thus we have the following,

µ ¶
α3 t2
r(t) = α1 + α2 t +
2
ṙ(t) = α2 + α3 t
r̈(t) = α3

Hence,
· ¸
1 α3 t2 5K 50K
e(t) = α1 + α2 t + + [α2 + α3 t] − α3
(1 + K) 2 (1 + K)2 (1 + K)3

More problems will be added in due course of production of the book.

7.6 Sensitivity

Due to changing environmental conditions and aging of the control system components, the param-
eters of a control system tends to vary. The variation in the system parameters may affect the
working and performance of the designed control system, thus it is desirable to have minimal effect
on the system performance with the variation in system parameters. One such measure in control
theory is sensitivity. It is defined as the ratio of proportional change of a control variable (say M) to
proportional change in system parameter (say G). Mathematically, it can be described as,

M ∆M/M
SG =
∆G/G

For small incremental changes one can write it in terms of partial derivatives,

M ∂M/M
SG =
∂G/G

7.6.1 Feedback characteristics of a closed loop control system

Effects of feedback in an automatic control system is explained here using the sensitivity measure.
To realize the relative effect of feedback we also consider an open-loop control system. Fig. 7.37 (a)
and (b) shows an open and closed-loop control system, respectively. In Fig. 7.37 (b), D(s) represents
disturbance input to the system. Now let us analyze the sensitivity with respect to different variations
in the system parameters.

I. Sensitivity of T (s) (C(s)/R(s)) with respect to G(s) variation:

66 Time Response Analysis of Linear Systems

Figure 7.37: (a)Open loop and (b) Closed loop control system

i) For an open loop control system shown in Fig. 7.37 (a), one can write,

C(s)
T (s) = = G(s)
R(s)
C(s) = G(s)R(s)

If G(s) slightly changes to G(s) + ∆G(s), then corresponding change in output will be,

C(s) + ∆C(s) = [G(s) + ∆G(s)] × R(s)

C(s) + ∆C(s) = C(s) + ∆G(s)R(s)

Thus, change in system output due to change in G(s) for open loop system is,

∆C(s) = ∆G(s) × R(s) (7.50)

Computing the sensitivity as per the definition provided above,

T ∂T G
SG = ×
∂G T
= 1, as T = G in this case
7.6 Sensitivity 67

ii) For closed loop control (with D(s) = 0) shown in Fig. 7.37 (b), one may write it as,

G(s) C(s)
T (s) = =
1 + G(s)H(s) R(s)

Slight change in G(s) leads to G(s) + ∆G(s) which correspondingly affects the system
output C(s) as,

G(s) + ∆G(s)
C(s) + ∆C(s) = × R(s)
1 + G(s)H(s) + ∆G(s)H(s)

Again ∆G(s) << G(s), thus ∆G(s)H(s) can be neglected compared to the term
G(s)H(s). This leads to an expression,

G(s) ∆G(s)
C(s) + ∆C(s) = × R(s) + × R(s)
1 + G(s)H(s) 1 + G(s)H(s)
∆G(s)
C(s) + ∆C(s) = C(s) + × R(s)
1 + G(s)H(s)

Hence, the change in system output in this case will be,

∆G(s)
∆C(s) = × R(s)
1 + G(s)H(s)

Computing the sensitivity,

T ∂T G
SG = ×
∂G T
(1 + GH) − GH G
= 2
×
(1 + GH) G/(1 + GH)
1
=
1 + GH

Thus, after comparing both the sensitivities it is clear that, in case of closed-loop system
1
it is 1+GH times less than that of open-loop system, thus indicating that variation of G(s)
affects C(s) lesser in case of closed-loop system.

II. Sensitivity of T (s) with respect to H(s) variation: As there is no feedback element in case
T
of open-loop control system, so in this case sensitivity (SH ) will be compared with respect to
T
the previous sensitivity (SG ) for the closed loop control system.

G(s)
T (s) =
1 + G(s)H(s)
68 Time Response Analysis of Linear Systems

T
Sensitivity SH for the closed loop system is written as,

T ∂T H(s)
SH = ×
∂H T (s)
GH
= −
1 + GH
T T
Comparing the sensitivities SG and SH , one can infer that closed loop system is more sensitive
T
to variation in H(s) compared to G(s), as H(s) → ∞, SH → 1.

III. Effect of feedback on system dynamics: Consider a first-order open-loop control system
C(s) K 1
shown in Fig. 7.37 (a), with R(s) = s+T1 , so the time constant of the system is τ = T1 and
DC gain= TK1 . Now when the system is made closed loop with H(s) = 1 and D(s) = 0 as shown
C(s) K
in Fig. 7.37 (b) then R(s) = s+(K+T1 ) . In consequence, the time constant of the closed loop
1
system reduces to τc = K+T 1
compared to τ of open loop system, thus the response for closed
K
loop system is enhanced. Also the DC gain in this case is reduced to K+T1 compared to open
loop system. Thus, it indicates that feedback effects the system dynamics by changing the pole
locations and steady state performance is affected too.

7.6.2 Solved problem

S7.17: A D.C Generator terminal voltage V0 is regulated using automatic voltage regulator (AVR)
whose forward path gain is 100 and the feedback path gain is 0.8. Determine the following,

1. The reference voltage Vr for keeping V0 at 300 Volts.

2. If the forward path gain is reduced by 10%, then what is the % change in the terminal voltage
of the generator.

If the system is made open-loop, then compute the results in (1) and (2).
Soln: AVR of a D.C generator can be represented by a simple closed-loop control system as shown
in Fig. 7.38. For closed-loop system:

V0 G 100
= =
Vr 1 + GH 81
81
Vr = × V0 = 243V olts
100

Now let us compute the percentage change in terminal voltage V0 in both open-loop & closed-loop
system when G(s) is reduced by 10%.

1. For Open-loop system (without feedback gain H):

V0
= G = 100 V olts
Vr
V0
Vr = = 3 V olts
G
7.7 Higher-order system response 69

Figure 7.38: Closed-loop AVR system

2. For closed loop system: Given that the forward path gain is reduced by 10%, hence Gnew =90.
The sensitivity of the overall gain T with respect to G is expressed as,

T G ∂T
SG = ×
T ∂G
1 1
= =
1 + GH 73
T % change in T
SG =
%changeinG
1 % change in T
=
73 10
10
% change in T = = 0.136%
73
% change in V0 = % change in T = 0.136% (7.51)

For open-loop system: Change in terminal voltage ∆V0 and change in forward path gain
∆G are related as,

∆V0 = ∆G × Vr
∆V0
∆V0 = × 100
V0
30
= × 100 = 10%
300

7.7 Higher-order system response

So far in this chapter we have discussed about the time-domain performance evaluation of a second-
order system in terms overshoot, settling time, damping ratio and few other time-domain specifica-
tions. It must be emphasized here that, if a system order is more than two or a system has zeros in it,
70 Time Response Analysis of Linear Systems

then the time-domain performance evaluation of the system is not so straight forward as in the case
of standard second-order system. In this context, we would present briefly the validity & conditions
under which higher-order system dynamic response can be approximated by a second-order system.

7.7.1 Pole-Zero cancelation

Let us consider a system having transfer function of the form,

10»(s »+»»
2)
G(s) = »
» (7.52)
(s
» »
+ 2) (s 2 + 7s + 12)

Now due to the exact location of pole and zero in (7.52) there will be cancelation of pole and zeros
in the transfer function. This phenomenon is called pole-zero cancelation. Due to cancelation a
third-order system (higher order) response can now be approximated by a second-order system. Thus
effect of pole-zero cancelation can be utilized for approximating the dynamics of higher order system.

7.7.2 Zeros in the Proximity of a Pole

Always it may not be possible to have exact cancelation of poles & zeros, rather their location may
be very close-by. In such case by evaluating the residues of the poles we may justify the pole-zero
cancelation effect or alternatively the approximation of higher-order system dynamics by lower-order.
Let us illustrate this fact using two distinct cases.

Case I Let us consider step-response (in s-domain) of a system given by,

32(s + 3)
C(s) = (7.53)
s(s + 2.5)(s + 4)(s + 6)

Carrying out the partial fraction of (7.53),

A B C D
C(s) = + + +
s s + 2.5 s + 4 s + 6
1.6 1.216 2.664 2.286
C(s) = − − + (7.54)
s s + 2.5 s+4 s+6

Here, the pole located at s = −2.5 which is close to zero at s = −3 has residue value of 1.216
which is of the same order as that of the residues of other poles in this system. Thus, in this
case a second-order approximation of the system is not really possible or valid. In other words,
the pole-zero cancelation effect cannot be realized in this system though the poles & zeros are
close by.

Case II Let us consider another step-response (in s-domain) of a system given by,

32(s + 3)
C(s) = (7.55)
s(s + 3.02)(s + 4)(s + 6)
7.8 MATLAB based analytical problems 71

Carrying out the partial fraction of (7.55),

A B C D
C(s) = + + +
s s + 3.02 s + 4 s + 6
1.32 0.073 4.08 2.68
C(s) = + − + (7.56)
s s + 2.5 s + 4 s + 6

Location of pole at s = −3.02 which is close to zero at s = −3 has residue value of 0.073.
The residue value for pole at s = −3.02 is much below the values of other residues, indicating
that, magnitude of the response due to this pole is far below other poles and hence it can be
neglected. In other words, the pole-zero cancelation effect between pole at s = −3.02 and a
zero at s = −3 can be realized in this system.
Consequently, a second-order approximation of the system is possible or valid in this case and
is represented as,

1.32 4.08 2.68

C(s) u − + (7.57)
s s+4 s+6

7.8 MATLAB based analytical problems

In this section, we will discuss various MATLAB programs for carrying out the analysis computa-
tionally that have been discussed in the preceding sections of this chapter. This section will use some
of the basic MATLAB as well as Control System Toolbox of MATLAB functions.
Example 1: Given a standard transfer function of a second order system,

ωn2
G(s) =
s2 + 2ζωn s + ωn2

Display the result for given values of ωn = 6 and ζ = 0.6 in the form of given G(s).
Program:

1 %Example 1 Chapter 7
2 clear all
3 clc
4 omega n=6;%natural frequency of oscillation of the system
5 zeta=0.6;%%Damping ratio of the system
6 num=[0 0 omega nˆ2];%Coefficients of the given numerator polynomial
7 den=[1 2*zeta*omega n omega nˆ2];%Coefficients of the given denominator polynomial
8 sys=tf(num,den)%Creates a continuous−time transfer function

Display:

1 Transfer function:
2 36
72 Time Response Analysis of Linear Systems

3 −−−−−−−−−−−−−−−−
4 sˆ2 + 7.2 s + 36

Note: To know in details about the function ‘tf ’, type help tf in the MATLAB command
window. This command is available with CONTROL SYSTEM TOOLBOX of MATLAB.
Example 2: Obtain the unit-step response of the system whose forward path transfer function
is given by,

36
G(s) =
s2 + 7.2s + 36

Assume that the system has unity feedback.

Analysis:

C(s) G(s)
=
R(s) 1 + G(s)

If,

N (s)
G(s) =
D(s)
C(s) N (s)
=
R(s) D(s) + N (s)

Program:

1 %Example 2 Chapter 7
2 clear all;clc;
3 num=[0 0 36];
4 den=[1 7.2 36];
5 den1=num+den;
6 step(num,den1)%Step response command
7 grid on
8 title('Closed loop response of unity feedback control system')
9 xlabel('t(secs)')%Label x−axis
10 ylabel('Response')%Label y−axis

Display: The unit-step response plotted by the MATLAB command ‘step’ is shown in the Fig.7.39
below. Note: For details of function ‘step’ command type ‘help step’ in the MATLAB command
window.
Example 3: Given ωn = 6rad/sec and for different values of the damping ratios ζ = 0, 0.1, 0.5, 0.7, 1, 1.5,
obtain the unit-step response. Analyze the effect of ζ on the unit-step response of the system repre-
sented by its closed-loop transfer function as, also plot the closed loop poles.

C(s) ωn2
=
R(s) s2 + 2ζωn s + ωn2
7.8 MATLAB based analytical problems 73

Closed loop response of unity feedback control system

0.7

0.6

0.5
Response

0.4

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
t(secs) (sec)

Figure 7.39: Step response of a closed loop system

Program:

1 %Example 3 Chapter 7
2 clear all;clc;
3 omegan=6;
4 t=0:0.1:10;%Sweeping/Varying time from 0 to 5 secs in steps of 0.1 sec
5 [row col]=size(t);
6 zeta=[0 0.1 0.5 0.7 1 1.5];%Storing various values of zeta in the form of vector
7 for k=1:6%Selection of k is such as to run the loop 6 times for 6 different values of zeta
8 num=[0 0 omeganˆ2];%Declaring the coefficients of N(s) of C(s)/R(s)
9 den=[1 2*zeta(k)*omegan omeganˆ2];%Declaring the coefficients of D(s) of C(s)/R(s)
10 sys=tf(num,den);
11 [c(row:col,k),t]=step(sys,t);%Uses the user−supplied time vector 't' for simulation.
12 ple(:,k)=pole(sys);
13 end
14 plot(t,c)
15 disp(' zeta=0 zeta=0.1 zeta=0.5 zeta=0.7
zeta=1 zeta=1.5')
16 ple;
17 disp(ple)
18 grid on
19 title('Unit step response for different values of zeta')
20 xlabel('t(secs)')%Label x−axis
21 ylabel('Response')%Label y−axis
22 text(1.5,1.911,'\zeta=0')
23 text(1.6,1.385,'\zeta=0.1')
74 Time Response Analysis of Linear Systems

24 text(0.5,1.124,'\zeta=0.5')
25 text(0.5,0.9653,'\zeta=0.7')
26 text(0.6,0.8743,'\zeta=1')
27 text(0.4,0.5322,'\zeta=1.5')
28 figure(2)
29 for j=1:12;
30 plot(real(ple(j)),imag(ple(j)),'x')
31 title('Pole−zero plot')
32 xlabel('Real part')%Label x−axis
33 ylabel('Imaginary part')%Label y−axis
34 grid on
35 hold on
36 end

Display: The poles of the system for different values of ζ can be displayed in tabular fashion as
shown below in the command window. The pole-zero plot for the system is shown in the Fig.7.40.
The unit-step response for different values of ζ are shown in Fig.7.41 below.
Analysis: By looking at the curves following conclusions may be drawn,

zeta=0 zeta=0.1 zeta=0.5 zeta=0.7 zeta=1 zeta=1.5

0 + 6.0000i -0.6000 + 5.9699i -3.0000 + 5.1962i -4.2000 + 4.2849i -6.0000 + 0.0000i -15.7082
0 - 6.0000i -0.6000 - 5.9699i -3.0000 - 5.1962i -4.2000 - 4.2849i -6.0000 - 0.0000i -2.2918

a) For ζ = 0, the response is a sustained oscillations, refer Fig. 7.41. As the closed-loop poles are
imaginary pairs as shown in Fig. 7.40 (marked with brown color).

b) For ζ < 1 (i.e., ζ = 0.1, ζ = 0.5, ζ = 0.7) the response is underdamped as shown in Fig.7.41. The
reason is that the closed loop poles are complex conjugate pairs (having both real & imaginary
part) as shown in Fig.7.40 (marked with blue color).

c) For ζ = 1 the response is critically damped as shown in Fig.7.41, as both the poles are real and
equal as shown in Fig.7.40 (marked with green color).

d) For ζ > 1 (i.e, ζ = 1.5) the response is overdamped as shown in Fig.7.41. The reason is that
the closed-loop poles are real & distinct with one pole far away from the imaginary axis while
another one is near to the imaginary axis. The response is due to the closed pole which is
nearer to the imaginary axis as it has higher time-constant and is considered to be dominant
pole (marked with red color).

Example 4: Compute the following time-domain specifications for a unity feedback second order
control system represented by an open-loop transfer function,

25
G(s)H(s) =
s2 + 6s

where, H(s) = 1. Obtain the following, (i) Rise time (ii) Peak-time (iii) Max. overshoot and (iv)
Settling-time.
7.8 MATLAB based analytical problems 75

Pole−zero plot
6

2
Imaginary part

−2

−4

−6
−16 −14 −12 −10 −8 −6 −4 −2 0
Real part

Figure 7.40: Pole-zero plot

Solution: We will present two programs following two approaches, (a) analytically and (b) graphi-
cally.
25 N (s)
Program (a): Analytical solution: Given G(s)H(s) = s2 +6s = D(s) , thus closed-loop transfer func-
C(s) N (s)
tion is R(s) = D(s)+N (s) .
Program (a): Graphical solution: Computationally the graphical solution is obtained by plotting
the step response of the system. The information of the simulation program have been used here
computing the various time-response specifications.

1 clear all
2 clc
3 num=[0 0 25];%N(s) of G(s)H(s)
4 den=[1 6 0];%D(s) of G(s) H(s)
5 den1=num+den;%D(s)of C(s)/R(s), closed loop transfer function
6 omegan=sqrt(num(:,3));%Natural frequency of oscillation
7 zeta=den1(:,2)/(2*omegan);%Damping ratio
8 omegad=omegan*sqrt(1−(zeta)ˆ2);%damped frequency of oscillation
9 tau=1/(zeta*omegan);%Time constant or real part of the pole
10 theta=acos(zeta);
11 %%%Analytical Approach%%%%%%%%
12 tpa=pi/omegad;%Peak time
76 Time Response Analysis of Linear Systems

Unit step response for different values of zeta

2
ζ=0
1.8

1.6

1.4 ζ=0.1

1.2
ζ=0.5
Response

1
ζ=0.7
ζ=1
0.8

0.6
ζ=1.5

0.4

0.2

0
0 1 2 3 4 5 6 7 8 9 10
t(secs)

Figure 7.41: Step response of a closed loop system

13 tra=(pi−theta)/omegad;%Rise time
14 ts1a=3*tau;%For 5% tolerance band
15 ts2a=4*tau;%For 2% tolerance band
16 Mp=exp(−zeta*pi/(sqrt(1−zetaˆ2)))*100;%Peak overshoot
17 rslta=[tpa tra ts1a ts2a Mp];%Analytical results stored as an array
18 disp('Analytical values:')
19 disp(' tpa tra ts1a ts2a Mp')
20 disp(rslta)%Display of analytical results
21 %%%%%Graphical Approach%%%%%
22 t=0:0.02:10;%Simulation time length
23 [row col]=size(t);
24 [z,y,t]=step(num,den1,t);%stores the values of response and time do not plots the response
25 plot(t,z)%plots the step response
26 grid on
27 title('Unit step response')
28 xlabel('t(sec)')
29 ylabel('Response')
30 p=1;
31 while z(p)<1
32 p=p+1;
33 end
34 tr=p*0.02;%Rise time
7.8 MATLAB based analytical problems 77

35 [zmax,tp]=max(z);
36 tp=(tp−1)*0.02;%Peak time
37 Mpg=(zmax−1)*100;%Percentage peak overshoot
38 k=col;
39 while z(k)>0.98 && z(k)<1.02%Considering 2%tolerance band
40 k=k−1;
41 end
42 ts=k*0.02;%Settling time
43 rsltg=[tp tr ts Mpg];%Graphical results stored as an array
44 disp('Graphical values:')
45 disp(' tp tr ts Mpg')
46 disp(rsltg)%Display of graphical results

Unit step response

1.4

1.2

1
Response

0.8

0.6

0.4

0.2

0
0 1 2 3 4 5 6 7 8 9 10
t(sec)

Figure 7.42: Step response for Example 4

1 Analytical values:
2 tpa tra ts1a ts2a Mp
3 0.6587 0.3932 2.0000 2.6667 37.2326
4

5 Grahical values:
6 tp tr ts Mpg
7 0.6600 0.4200 2.2600 37.2318
78 Time Response Analysis of Linear Systems

Result/Display: The difference in the result due to two approaches is mainly due the step size
considered in the graphical simulation. The result is displayed below for both the approaches. The
step response is shown in Fig.7.42.
Example 5: The open-loop transfer function of a unity feedback control system is given as,

K(s + 10)
G(s)H(s) =
s(s + 1)(s + 20)(s + 50)

Compute the static error constants as well as the steady state error for the given type of feedback
control system.
Program:

1 %Example 5
2 clear all
3 clc
4 %%%%%%%%'Position coefficient Kp'%%%%%%%
5 K=input('Enter desired value of system gain K,K= ');
6 numg=K*[1 10]; %Define numerator of the system's transfer function
7 deng=poly([0 −1 −20 −50]); %Define denominator of the system's transfer function
8 G=tf(numg,deng); %Form G(s) LTI object
9 Kp=dcgain(G); %Calculate and display the value of Kp
10 %%%%%%%%%'Velocity coefficient Kv'%%%%%%%%%%%%
11 numgs=conv([1 0],numg); %Define numerator of sG(s)
12 dengs=deng; %Define denominator of sG(s)
13 Gs=tf(numgs,dengs) ; %Form sG(s) LTI object
14 Gs=minreal(Gs); %Cancel common s in numerator and denominator
15 Kv=dcgain(Gs); %Calculate and display Kv
16 %%%%%%%%'Acceleration coefficient Ka'%%%%%%%%%%
17 dengsa=deng; %Define denominator of sˆ2G(s)
18 numgsa=conv([1 0 0],numg); %Define numerator of sˆ2G(s)
19 Gsa=tf(numgsa,dengsa); %Form sˆ2G(s) LTI object
20 Ka=dcgain(Gsa); %Calculate and display acceleration coefficient Ka
21 static const=[K Kp Kv Ka];
22 disp(' K Kp Kv Ka')
23 static const;
24 disp(static const)
25 %%%%%%%STEADY−STATE ERROR Computation%%%%%%%%%%%%%
26 %%%%%%'Step input'%%%%%%
27 esss=1/(1+Kp); %Calculate and display steady state error for step input
28 %%%%%%%%'Ramp input'%%%%%%%
29 essr=1/Kv; %Calculate and display steady state error for ramp input
30 %%%%%%%%'Parabolic input'%%%%%%%%%%%
31 essp=1/Ka; %Calculate steady state error for parabolic input
32 ess=[esss essr essp];
33 disp(' esss essr essp')
34 ess;
35 disp(ess)
7.8 MATLAB based analytical problems 79

Result/Display: The results for the static error constants and the steady state error will be
displayed in the command window as shown below. Values of the static constants and the steady
state errors obtained for the given Type 1 system conforms with that of the result presented in Table
1.1.

1 Enter desired value of system gain K,K= 20

2 K Kp Kv Ka
3 20.0000 Inf 0.2000 0
4

5 esss essr essp

6 0 5.0000 Inf

Example 6: Obtain the unit-ramp and unit-impulse response for the closed-loop control system
represented by the transfer function given below,

C(s) 5
=
R(s) s2 + 2s + 5

Program:

1 %Example 6, Chapter 4
2 clear all
3 clc
4 %%%%%%%%%%Unit Impuse response%%%%%%%%%
5 num=[0 0 5];%N(s) of G(s)
6 den=[1 2 5];%D(s) of G(s)
7 figure(1)
8 sys=tf(num,den);
9 impulse(sys);
10 grid;
11 title('Unit Impulse Response of a Second order system')
12 %%%%%%%%Unit Ramp Response%%%%%%
13 num1=[0 0 0 5];%N(s) of G(s)
14 den1=[1 2 5 0];%D(s) of G(s)
15 sys1=tf(num1,den1);
16 t=0:0.1:5;
17 c=step(sys1,t);
18 figure(2)
19 plot(t,c,'o',t,t,'−')
20 grid;
21 title('Unit Ramp Response of a Second order system')

Display: The unit-impulse and unit-ramp response is shown in the Fig. 7.43 and 7.44 respectively.
80 Time Response Analysis of Linear Systems

Unit Impulse Response of a Second order system

1.4

1.2

0.8
Amplitude

0.6

0.4

0.2

−0.2

−0.4
0 1 2 3 4 5 6
Time (sec)

Figure 7.43: Unit impulse response for Example 6

Unit Ramp Response of a Second order system

5

4.5

3.5

2.5

1.5

0.5

0
0 1 2 3 4 5

Figure 7.44: Unit ramp response for Example 6

Discussion:

a There is no ramp command in the MATLAB , thus we have used step command to simulate the
response.

b For a given G(s), the method adopted is that, to divide G(s) by s and the modified transfer
7.9 Unsolved Problems 81

function is passed as an argument in the ’step‘ command.

C(s) 1 G(s)
c Mathematically, it can be illustrated as, R(s) = G(s) and R(s) = s2 . Thus c(s) = s × 1s .

7.9 Unsolved Problems

U7.1: A block diagram of position servomechanism using armature controlled D.C servo motor is
used to control the position of the mechanical load in accordance with the position of the reference
shaft. The moment of inertia of the motor armature referred to the load shaft is 300 Kg − m2 and
the motor torque at the load is 8 × 104 N − m/rad of error. The viscous damping coefficient referred
to the load shaft is 6 × 103 N − m/(rad/sec). The transmission of torque from motor shaft to the
output shaft is through gear trains of gear ratio n = 2. Determine the following,

Figure 7.45: Block diagram

1. Find the time-domain response for the system when it is subjected to a step input of 2 units,
find the damped frequency of oscillations, rise time, peak overshoot and peak time.

2. Find the steady state error when the command signal is constant angular velocity of 1 r.p.m.

3. Find the steady state error when steady torque of 1000 N − m is applied at the load shaft.

U7.2: The gain constant of a controller is 0.8Nm/rad. The controller is used for unity feedback
control system for controlling the position of a load having moment of inertia J = 10 × 10−4 Kg − m2
and viscous friction 3 × 10−2 N m/(rad/sec). Obtain an expression relating the time and error angle.
The input shaft being fed with a velocity of 2.5 r.p.m.
82 Time Response Analysis of Linear Systems

U7.3:The forward path transfer function of a unity feedback control system is given by,

K(s + 2)
G(s) =
s3 + βs2 + 4s + 1

Determine the values of K and β such that sustained oscillations have a frequency of 10 rad/sec.
U7.4: For a feedback control system shown in the Fig. 7.46 below, with following specifications,

a Maximum overshoot for a unit-step function applied to input no. 1 is 25%.

b Steady state error for a unit parabolic input applied to input no. 1 is 0.3 rad.

c A torque of 2 Nm is suddenly applied to the system with input no. 1 being held fixed.

Compute the values of the controller gain K, the moment of inertia J and the coefficient of viscous
friction f .

Figure 7.46: Block diagram for problem U1.4

U7.5: Determine Kp , Kv and Ka and the corresponding steady state error for the following types
of inputs,

a r(t) = 2 u(t)

b r(t) = 2 t u(t)
t2
c r(t) = 2 u(t)

Computation are to be done for two different unity feedback control systems represented by the
following open-loop control system,
K
i G(s) = s2 +3s+100

K(1+0.1s)(1+10s)
ii G(s) = s2 (s2 +3s+100)

U7.6: For the closed-loop system represented by a block diagram in the Fig. 7.47 prove the
following facts,

a The change in the system parameter α and K there is no reduction or increase in sensitivity of
steady-state error subjected to ramp-input.
7.9 Unsolved Problems 83

b Increase in K reduces the sensitivity of a closed-loop transfer function to change in the parameter
α.

Figure 7.47: Block diagram for problem U1.6

U7.7: Find the sensitivity of the steady state error to the change in K for the system shown in
the Fig. 7.48 below.

Figure 7.48: Block diagram for problem U1.7

U7.8: A unity feedback control system has its open-loop transfer function given by,

s+1
G(s) =
4s2

Determine an expression for the time-response when the system is subjected to following input func-
tions:

a Unit-impulse input.

b Unit-ramp input.

U7.9: For the system shown in Fig. 7.49 find the following,

a The system type

b Static error constant for input 100 t u(t).

c Steady state error for the input in (b).

84 Time Response Analysis of Linear Systems

d Steady state value of the actuating signal.

Figure 7.49: Block diagram for problem U1.9

7.10 Chapter Summary

• For practical control systems the use of actual input signal may not be known as a priori
and such actual signals if known cannot be expressed by simple mathematical equations, thus
complicating the entire analysis. For analysis of LTI system the standard test signals used are
sudden shock - characterized by impulse signal, sudden change - step input, constant velocity -
ramp input, constant acceleration - parabolic input.

• A first-order closed-loop control system has only one closed-loop pole. Its transfer function is,

C(s) 1
=
R(s) 1 + τs

• The response of the FOS due to step-input is given by,

c(t) = 1 − e−t/τ

• Standard transfer function of a second order control system is given by,

C(s) ωn2
=
R(s) s2 + 2ζωn s + ωn2

• Nature of the unit-step response for a second order system is influenced by the value of ζ
(damping ratio),

– 0 < ζ < 1 - underdamped response

– ζ = 1 - critically damped response
– ζ > 1 - over damped response

• The time-response specifications for an underdamped response are,

7.10 Chapter Summary 85

– Peak Overshoot: The maximum positive deviation of the time-response from its de-
sired (or steady state) value is known as peak overshoot. It is designated as Mp . This
specification is expressed in percentage.
– Delay time: Time required for the response to reach 50% of the steady state (final) value
in the first attempt. It is denoted by td .
– Rise time: Time required for the response to rise from 0 to 100% of the final value. It is
denoted by tr .
– Peak time: Time required to reach the maximum value of the response for the first time.
It is denoted by tp , this is the time corresponding to the peak overshoot in the response
curve.
– Settling time: Time required for the response to reach and stay within specified tolerance
band (usually 2 to 5%) of its final value. The limit on the tolerance band is decided by
the user as per their desired application. It is denoted by ts .
– Steady state error: Difference between actual and desired value of the response at time
(t → ∞) is referred as steady state error. It is denoted as ess .

• If the actual output of the system during steady state deviates from the reference input (desired
output) then the closed-loop control system is said to possess steady state error. The error
transfer function for a closed loop control system is,

E(s) 1
=
R(s) 1 + G(s)H(s)

• Types of Feedback Control System are,

1. For n = 0, Type-0 feedback control system, indicates presence of no integrator in G(s)H(s).

2. For n = 1, Type-1 feedback control system, indicates presence of 1 integrator in G(s)H(s).
3. For n = 2, Type-2 feedback control system, indicates presence of 2 integrators in G(s)H(s).

• Steady State Error Constants are,

– Positional error constant: It is evaluated for unit-step input, lims→0 G(s)H(s) = Kp

– Velocity error constant: Velocity error constant is defined for unit ramp input, lims→0 s(G(s)H(s)) =
Kv
– Acceleration error constant: Acceleration error constant is defined for unit parabolic
input, lims→0 s2 (G(s)H(s)) = Ka

• Static error constants & ess due to different types of systems:

Type-0 System:
1
– Kp = lims→0 G(s)H(s), hence Kp = K, ess = 1+K .
86 Time Response Analysis of Linear Systems

1
– Kv = lims→0 s(G(s)H(s)), hence Kv = 0, ess = Kv = ∞.
1
– Ka = lims→0 s2 (G(s)H(s)), hence Ka = 0, ess = Ka = ∞.

Type-1 System:

K 1
– Kp =lims→0 G(s)H(s), hence Kp = 0 = ∞, ess = 1+∞ = 0.
1
– Kv = lims→0 s(G(s)H(s)), hence Kv = K, ess = K s.
1
– Ka = lims→0 s2 (G(s)H(s)), hence Ka = 0, ess = 0 = ∞.

Type-2 System:

K 1
– Kp = lims→0 G(s)H(s), hence Kp = 0 = ∞, ess = 1+∞ = 0.
K 1
– Kv = lims→0 s(G(s)H(s)), hence Kv = 0 = ∞, ess = K = 0.
1
– Ka = lims→0 s2 (G(s)H(s)), hence Ka = K, ess = K.

• Steady State Error for Nonunity Feedback System: Usually feedback control system do
not have unity gain feedback path rather consists of some pure gain or dynamics due to the
use of compensator or due to physical model of the system. But a non unity feedback control
system can always be converted into an equivalent unity feedback control system using block
diagram reduction technique.

• Error series & Dynamic error coefficients:

e(t) = r(t)C0 − ṙ(t)C1 + r̈(t)C2 − .......

where, r(t) is the input signal, ṙ(t), r̈(t) are the first & second derivative of the input signal and
Z t Z t
d
C0 = we (τ )dτ = lim We (s), C1 = − τ we (τ )dτ = lim We (s)
0 s→0 0 s→0 ds
Z t
τ2 d2
C2 = we (τ )τ = lim 2 We (s)
0 2! s→0 ds

7.11 Multiple Choice Questions

M1. Unit-step response of system is given by

5 5
c(t) = + 5t − e−2t
2 2

The transfer function of the system is,

10(s+1) 5 10(s+1)
(i) s(s+2) (ii) 2 s(s+2)
5(s+1) 10
(iii) s(s+2) (iv) s(s+2)
7.11 Multiple Choice Questions 87

M2. Match the following,

Location of closed-loop poles

i. Simple pole on negative real-axis

ii. A pair of conjugate imaginary poles

iii. A pair of complex conjugate poles

iv. A simple pole on positive real-axis

Time-response

a. Sustained oscillation

b. Exponentially rising response

c. Exponentially decaying response

d. Underdamped response

M3. In practice the value of ζ for a second order system is

(i) ζ = 1 (ii) ζ > 1
(iii) ζ = 1 (iv) 0.4 < ζ < 0.7

M4. The influence of ζ > 1 on the response of a second order system is characterized by
(i) Two distinct real roots of characteristic equation lying on left-half of s-plane
(ii) A pair of complex conjugate pair of root with positive real parts
(iii) Two real and equal roots lying on left-half of s-plane
(iv) None of the above

M5. If the roots of the characteristic equation are s = −ζωn ± jωd , then damped frequency of
oscillation of the system is,
p p
(i) ωd = ωn 1 − ζ 2 (ii) ωd = ωn (1 − ζ)2
p
(iii) ωd = ωn (iv) ωd = ωn ζ 2 − 1

10
M6. The damping ratio of a system characterized by its closed-loop transfer function T (s) = (s+2)(s+5)
is
(i) 1.107 (ii) 0.707
(iii) 1 (iv) 0.8

M7. A second order system having characteristic equation s2 + 19.6s + 196 = 0. The time-constant
of the underdamped response is,
(i) 0.2 sec (ii) 0.5 sec
88 Time Response Analysis of Linear Systems

(iii) 0.102 sec (iv) Response does not decay exponentially

M8. Unit-step response of a second order system has peak overshoot of 16.3%, what is teh value
of ζ for the system,
(i) 1 (ii) 0.5
(iii) 1.12 (iv) None of the above

M9. The time-response of a second order system is c(t) = 1 − 1.281e−5t sin(6.245t + 0.8957). The
value of ζ for the system is,
(i) 0.625 (ii) 0.707
(iii) 1 (iv) > 1

M10. Match the following,

i. Positional error constant

ii. Velocity error constant

iii. Acceleration error constant

a. lims→0 s2 G(s)

b. lims→0 G(s)

c. lims→0 sG(s)

5s+K
M11. A system having G(s) = s2 has a unity feedback. If the steady-state error is 0.05, then the
value of K is,
(i) 20 (ii) 25
(iii) 250 (iv) 200

M12. A Type-1 system is subjected to unit-parabolic input, its corresponding Kv & ess are,
1
(i) ∞ & 0 (ii) Constant & Kv
(iii) 0 & ∞ (iv) Constant & ∞

50
M13. A unity feedback control system with G(s) = s(s+10) is subjected to an input r(t) = 1 + 2t, its
dynamic error series is,
(i) e(t) = 0.2r(t) + 0.2ṙ(t) − 0.02r̈(t) (ii) 0.4
(iii) 0.36 (iv) 0.4t

M14. A Type-1 unity feedback control system has characteristic equation s2 + 6s + 25 = 0 and
K(s+α)
G(s) = (s+β) . If Kv = 10 for this system then what will be the corresponding values of K, α and β
7.11 Multiple Choice Questions 89

(i) 3,10,2 (ii) 3,7.14,2

(iii) 3.5,7.14,2.5 (iv) 1,7,-2.5

M15. A system is characterized by open loop gain K = 500, poles are located at s = 0and − 15. It is
subjected to unity feedback with unit-step signal at the input. The value of settling time and %Mp
are respectively,
(i) 0.53 and 32.13 (ii) 0.5 and 20
(iii) 0.32 and 50 (iv) 0.53 and 13

M16. If the system has tp = 0.25sec, ts = 1.25sec and DC gain of 0.9, then the second order
transfer function for the system is,
169 15
(i) s2 +6.5s+169 (ii) s2 +5s+150
152.1
(iii) s2 +6.5s+169 (iv) None of the above

M17. The time at which successive maxima and minima recurs for a unit-step underdamped re-
sponse forms,
(i) A.P. series (ii) G.P series
(iii) Combination of (i) and (ii) (iv) Do not follow any relationship

K
M18. Given G(s) = s(1+τ s) , if the peak over shoot for the system is reduced from 75% to 25%,
then by how much % will the gain be reduced,
(i) 5% (ii) 50%
(iii) 5.08% (iv) gain will not change

M19. Steady-state performance of a system can be improved by,

(i) Increasing the type of the system (ii) Decreasing the type of the system
(iii) Adding integrator to feed-forward of the system (iv) Both (i) and (iii)

t
M20. If the unit-step response of a system is c(t) = 1 − e− T , then its unit-impulse response will
be,
1 − Tt
(i) Same as response of unit-step (ii) c(t) = Te
t t
(iii) c(t) = t − T e− T (iv) c(t) = −e− T

M21. If the negative real part of the closed loop poles are away from the imaginary-axis, then
the time response will have,
(i) Large time constant (ii) Small time constant
(iii) No effect on time constant (iv) Large ζ

M22. Relative dominance of closed loop poles is assessed by,

90 Time Response Analysis of Linear Systems

(i) Ratio of real parts exceeding by 5 times (ii) No zeros located near by the poles
(iii) Ratio of real parts exceeding by at least 10 times (iv) Both (i) and (ii)

M23. The settling time for an underdamped response within 1% tolerance band is,
3 4
(i) ζωn (ii) ζωn
4.6
(iii) ζωn (iv) None of the above

M24. Presence of open loop zero in the left half of s-plane in addition to stable open loop poles,
subjected to unit-step input has the following affect on the response,
(i) Overshoot increases (ii) Overshoot decreases
(iii) ζ decreases (iv) Response is not affected