Page |1

Control system
A control system consists of processes or systems (known as plants) assembled for
the purpose of obtaining a desired output with desired performance, given a specified
input.
The figure shows a control system in its simplest form.
The input to the control system is the desired output
in a suitable form.
As an example consider the case of an elevator. When
a person presses the tenth-floor button, the elevator 10

rises to the tenth floor with a speed and floor levelling

accuracy designed for the passenger comfort. The
push of the tenth floor button is the input that
represents the desired output.
This input is shown as a step function in the figure. The performance of the elevator
is also shown as repsonse curve along with the input. Two major measures of
performance are i) the transient response and ii) the steady state error. In the
elevator example, the passenger comfort and passenger patience are dependent on its
transient response. If the response is too fast, passenger comfort is compromised; if
too slow, passenger patience is compromised. The steady state response is also another
important performance parameter. If the elevator does not stop properly at the tenth
floor level, the safety and convenience of the passenger would be sacrificed.
The control system is built for four primary reasons:
1. Power amplification 2. Remote control 3. Conveniene of input form
4. Compensation for disturbances
For example, a radar antenna, positioned by the low power rotation of a knob at the
input, requires a large amount of power for its output rotation. A control system can
produce the needed power amplification, or power gain.
Control systems are also useful in remote or dangerous locations. For example, a
remote-controlled robot arm can be used to pick up material in a radioactive
environment.
Control systems can also be used to provide convenience by changing the form of the
input. For example, in a temperature control system, the input is a position on a
thermostat. The out put is heat. A convenient position input yields a desired heat
output.
Another advantage of control system is the ability to compensate for disturbances. If
the response of a system changes from its commanded input value due to disturbances
or noise, the system must be able to detect the change and correct itself. The input
will not change to make the correction. The system itself should measure the deviation
and do the correction accordingly.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

Open-Loop Systems
The figure shows an open-loop system. It
consists of an input transducer, which
converts the form of input to that used by
the controller.
The controller drives the plant. The input is given as the reference or the desired output. The
actual output is also called controlled variable. Other signals, such as disturbances, are added to
the controller and process outputs via summing junctions. For example, the plant can be a furnace,
where the output variable is temperature. The controller consists of fuel valves and the electrical
system that operates the valves. Another example is the bread toaster. The disadvantage of
open-loop system is its inability to correct for the disturbances.
Closed loop systems
The disadvantage of the open-loop system is
overcome by the closed-loop system. The input
transducer converts the form of input to the
form used by the controller.
The output transducer, or sensor, measures the output rsponse and converts it into the form
used by the controller. For example, if the controller uses electrical signals to operate the valves
of a temperature control system, the input position and the output temperature are converted
to electrical signals. The input position can be converted to a voltage by a potentiometer, and the
output temperature can be converted to a voltage by a thermistor, a device whose electrical
resistance changes with temperature.
The first summing junction algebraically adds the signal from the input to the signal from the
output, which arrives via the feedback path, the return path from the output to the summing
junction. Here, in the figure the out put signal is subtracted from the input signal. The result is
generally called the actuating signal or error signal.
The closed-loop system compensates for
disturbances by measuring the output
response, feeding that measurement back
through a feed back path, and comparing that
response to that input at the summing
junction.

If there is any difference between the two responses, the system drives the plant, via the
actuating signal to make a correction. If there is no difference, the system does not drive the
plant, since the plant’s response is already the desired response.
➢ Closed-loop systems have the advantage of greater accuracy than open-loop systems.
➢ They are less sensitive to noise, disturbances, and changes in the environment.
➢ Transient response and steady-state error can be controlled more efficiently closed-loop
system often by a simple adjustment of gain in the loop and sometimes by redesigning the
controller. The redesign is known as compensating the system and to the resulting hardwae
as a compensator.
➢ Closed-loop systems are more complex and expensive

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
Analysis & design
Analysis is the process by which a system’s performance is determined.
A control system is dynamic: It responds to an input by undergoing a transient response before
reaching a steady-state response. Three major objectives in the analysis and design of systems
are producing the desired transient response, reducing the stready state error, and achieving
stability. Design is the process by which a system’s performance is created or changed.
Let us consider an example of the disk drive of a computer. Since
reading and writing cannot take place until the head stops
(transient response), the speed of the read/write head’s
movement from one track on the disk to another influences the
overall speed of the computer. The read/write head should be
positioned over the commanded track without error (steady
state response)

The Design Process

➢ Step 1: Transform requirements into a physical system
➢ Step 2: Draw a functional block diagram
➢ Step 3: Create a schematic
➢ Step 4: Develop a mathematical model (Block Diagram)
➢ Step 5: Reduce the block diagram
➢ Step 6: Analyse and design
Impulse 𝛿(𝑡) Transient
𝛿(𝑡) = ∞, 0− < 𝑡 < 0+ response
= 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒 modelling

Step 𝑢(𝑡) Transient

𝑢(𝑡) = 1, 𝑓𝑜𝑟 𝑡 > 0 response,
= 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒 steady
state error
Ramp 𝑡𝑢(𝑡) steady
Standard test input signals are used, 𝑡𝑢(𝑡) = 𝑡, 𝑓𝑜𝑟 𝑡 ≥ 0 state error
both analytically and during testing, to = 0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
verify the design.
Parabola 1 2
𝑡 𝑢(𝑡) steady
2
state error
1 2 1
𝑡 𝑢(𝑡) = 𝑡 2 , 𝑓𝑜𝑟 𝑡 ≥ 0
2 2

Sinusoid sin 𝜔𝑡 Transient

response,
steady
state error

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
Let us consider a few examples how to convert physical systems to block diagrams.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

Human beings experience a fever (increase in body

temperature) associated with illness. It is the outcome
of governed by the thermostatic controller of the body.
The increased activity of the hostile micro-organisms
triggers the central nervous system to command an
increas in the set point (i.e, reference input) of the
thermostat, because it happens that the immune
response of the body can be strengthened at a higher
temprature.

Exercise 1
A temperature control system operates by sensing the difference between the thermostat
setting and the temperature and then opening a fuel valve an amount proportional to this
difference. Draw a functional closed-loop block diagram indetifying the input and output
transducers, the controller, and the plant.
Exercise 2
An aircaft’s attitude varies in roll, pitch, and yaw as defined in
figure. Draw a functional block diagram for a closed-loop
system that stabilizes the roll as follows: The system
measures the actual roll angle with a gyro and compares the
actual roll angle with the desired roll angle. The ailerons
respond to the roll angle error by undergoing an angular
deflection. The aircraft responds to this angular deflection,
producing a roll angle rate. Identify the input and output
transducers, the controller, and the plant.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
MATHEMATICAL MODELLING OF CONTROL SYSTEMS
The control theory that we deal with here is the linear control theory, where the systems that
are under consideration are Dynamic, Linear, and Time Invariant systems. A dynamic system , if
the cause and effect are proportional, it is called linear. In a linear system, the principle of
superposition applies. That is, the response to several inputs can be calculated by treating one
input at a time and adding the results. Such systems will be governed by linear differential
equations. A differential equation is linear if the coefficicents are constants (time invariant) or
functions only of the independent variable (time varying).
Transfer Function
The transfer function of a linear, time-invariant system is defined as the ratio of the Laplace
tansform of the outpit (Response function) to the Laplace transform of the input (Driving
function) under the assumption that all initial conditions are zero.
ℒ[𝑜𝑢𝑡𝑝𝑢𝑡]
𝑇𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛 = 𝐺(𝑠) = { }
ℒ[𝑖𝑛𝑝𝑢𝑡] 𝑧𝑒𝑟𝑜 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠
➢ The transfer function of a system is a mathematical model which is an operational method
of expressing the differentail equation that relates the output variable to the input
variable.
➢ The transfer function is a property of a system itself, independent of the magnitude and
nature of the input or driving function.
➢ TF does not provide any information concerning the physical structure of the system. The
transfer functions of many physically different systems can be identical.
➢ If the transfer function of a system is known, the output response can be studied for
several inputs for understanding the nature of the system.
➢ If the transfer function of a system is unknown, it may be established experimentally by
introducing known inputs and studying the output of the system. Once established, the
transfer function gives a full description of the dynamic characteristics of the system.
A revisit to Laplace Transform
∞

ℒ[𝑓(𝑡)] = 𝐹(𝑠) = ∫ 𝑓(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡

0
∞
𝑓(𝑡) = 𝐴, 𝑓𝑜𝑟 𝑡 ≥ 0 ∞
𝑒 −𝑠𝑡 𝐴
ℒ[𝑓(𝑡)] = ∫ 𝐴 𝑒 −𝑠𝑡 𝑑𝑡 = 𝐴 [ ] =
−𝑠 0 𝑠
0

Step function If 𝐴 = 1, it is unit step ℒ[𝑢(𝑡)] =

1
𝑠
function 𝑢(𝑡)
∞ ∞
𝑓(𝑡) = 𝐴𝑡, 𝑓𝑜𝑟 𝑡 ≥ 0 −𝑠𝑡
𝑒 −𝑠𝑡
∞
𝑒 −𝑠𝑡 𝐴 𝑒 −𝑠𝑡 𝐴
∞
ℒ[𝑓(𝑡)] = ∫ 𝐴𝑡 𝑒 𝑑𝑡 = 𝐴 𝑡 [ ] −𝐴∫ 1. 𝑑𝑡 = [ ] = 2
−𝑠 0 −𝑠 𝑠 −𝑠 0 𝑠
Ramp function 0 0

𝑛]
𝑛!
ℒ[𝑡 =
𝑠 𝑛+1
∞
Harmonic 𝑓(𝑡) = sin 𝜔𝑡 𝑒 𝑖𝜔𝑡
−𝑒 −𝑖𝜔𝑡
𝑖 ∞ ∞
ℒ[sin 𝜔𝑡] = ∫ 𝑒 −𝑠𝑡 𝑑𝑡 = − {∫ 𝑒 −(𝑠−𝑖𝜔)𝑡 𝑑𝑡 − ∫ 𝑒 −(𝑠+𝑖𝜔)𝑡 𝑑𝑡}
function 0
2𝑖 2 0 0
∞ ∞
𝑖 𝑒 −(𝑠−𝑖𝜔)𝑡 𝑒 −(𝑠+𝑖𝜔)𝑡 𝑖 1 1
=− {[ ] − [ ] } = − {[ ]−[ ]}
2 −(𝑠 − 𝑖𝜔) 0 −(𝑠 + 𝑖𝜔) 0 2 (𝑠 − 𝑖𝜔) (𝑠 + 𝑖𝜔)
𝑖 (𝑠 + 𝑖𝜔) − (𝑠 − 𝑖𝜔) 𝑖 2𝑖𝜔 𝜔
=− [ ]=− { 2 }= 2
2 𝑠2 + 𝜔2 2 𝑠 + 𝜔2 𝑠 + 𝜔2
𝑠
𝑓(𝑡) = cos 𝜔𝑡 ℒ[cos 𝜔𝑡] = 2
𝑠 + 𝜔2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

∞
Translated f(t)u(t) f(t-𝜏)u(t-𝜏)
ℒ[𝑓(𝑡 − 𝛼)] = ∫ 𝑓(𝑡 − 𝜏) 𝑢(𝑡 − 𝜏) 𝑒 −𝑠𝑡 𝑑𝑡
function
0
𝑝𝑢𝑡 𝑡 − 𝜏 = 𝛼 ≫ 𝑑𝑡 = 𝑑𝛼
∞ ∞

∫ 𝑓(𝑡 − 𝜏) 𝑢(𝑡 − 𝜏) 𝑒 −𝑠𝑡 𝑑𝑡 = ∫ 𝑓(𝛼) 𝑢(𝛼) 𝑒 −𝑠(𝛼+𝜏) 𝑑𝛼

0 0
∞

= 𝑒 −𝑠𝜏 ∫ 𝑓(𝛼) 𝑢(𝛼) 𝑒 −𝑠𝛼 𝑑𝛼 = 𝑒 −𝑠𝜏 𝐹(𝑠)

0
∞
Pulse function 𝑓(𝑡) = 𝐴, 𝑓𝑜𝑟 0 ≤ 𝑡 ≤ 𝜏 ∞

= 𝐴 𝑢(𝑡) − 𝐴 𝑢(𝑡 − 𝜏) ℒ[𝑓(𝑡)] = ∫ 𝐴. 𝑢(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 − ∫ 𝐴 𝑢(𝑡 − 𝜏) 𝑒 −𝑠𝑡 𝑑𝑡

0
0
𝐴
𝐴 𝐴 𝐴
𝜏 = − 𝑒 −𝑠𝜏 = (1 − 𝑒 −𝑠𝜏 )
𝑠 𝑠 𝑠
𝐴 The strength of the pulse is the area of rectangle.
𝜏
𝜏 𝐴
ℒ[𝑓(𝑡)] = (1 − 𝑒 −𝑠𝜏 )
𝜏𝑠
Impulse 𝐴 𝐴 𝐴𝑠𝑒 −𝑠𝜏
𝑔(𝑡) = lim : 0 ≤ 𝑡 ≤ 𝜏 ℒ[𝑔(𝑡)] = lim { (1 − 𝑒 −𝑠𝜏 )} = lim { }=𝐴
𝜏→0 𝜏 𝜏→0 𝜏𝑠 𝜏→0 𝑠
𝐴
𝐴𝑠 𝜏 → 0, 𝑡ℎ𝑒 ℎ𝑒𝑖𝑔ℎ𝑡 →∞ If 𝐴 = 1, it is unit impulse function (dirac delta function 𝛿(𝑡))
𝜏
The area= A ℒ[𝛿(𝑡)] = 1
𝐴 𝐴
𝜏 𝜏
𝜏
∞ ∞
Multiplication 𝑒 −𝑎𝑡 𝑓(𝑡)
ℒ[𝑒 −𝑎𝑡 𝑓(𝑡)] = ∫ 𝑒 −𝑎𝑡 𝑓(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡 = ∫ 𝑓(𝑡) 𝑒 −(𝑠+𝑎)𝑡 𝑑𝑡 = 𝐹(𝑠 + 𝑎)
by 𝑒 −𝑎𝑡 0 0
∞
Exponential 𝑓(𝑡) = 𝑒 −𝑎𝑡 1
ℒ[𝑓(𝑡)] = ∫ 𝑒 −𝑎𝑡 𝑢(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡 =
function 0
𝑠+𝑎
∞
Real 𝑑
ℒ [ 𝑓(𝑡)] 𝑒−𝑠𝑡
∞ ∞
𝑒−𝑠𝑡
𝑑𝑡 ∫ 𝑓(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡 = [𝑓(𝑡) ] − ∫ 𝑓′(𝑡) 𝑑𝑡
differential 0
−𝑠 0 0 −𝑠

theorem 𝑓(0) 1
𝐹(𝑠) = + ℒ[𝑓′(𝑡)] ≫≫ ℒ[𝑓 ′ (𝑡)] = 𝑠𝐹(𝑠) − 𝑓(0)
𝑠 𝑠
𝑑𝑛 𝑑𝑛
ℒ [ 𝑛 𝑓(𝑡)] ℒ[ 𝑓(𝑡)] = 𝑠 𝑛 𝐹(𝑠) − 𝑠 𝑛−1 𝑓(0) − 𝑠 𝑛−2 𝑓 ′ (0) − ⋯
𝑑𝑡 𝑑𝑡 𝑛
∞
Real ℒ [∫ 𝑓(𝑡)𝑑𝑡] 𝑒−𝑠𝑡
∞ ∞
𝑒−𝑠𝑡
∫ [∫ 𝑓(𝑡)𝑑𝑡] 𝑒 −𝑠𝑡 𝑑𝑡 = [∫ 𝑓(𝑡)𝑑𝑡 ] − ∫ 𝑓(𝑡) 𝑑𝑡
Integration 0
−𝑠 0 0 −𝑠

theorem 𝑓(𝑡)𝑑𝑡 1 𝐹(𝑠) 𝑓 −1 (0)

ℒ [∫ 𝑓(𝑡)𝑑𝑡] = [∫ ] + 𝐹(𝑠) = +
𝑠 𝑠 𝑠 𝑠
𝑡=0

Final value theorem

If 𝑓(𝑡) and 𝑓’(𝑡) are Laplace transformable, if 𝐹(𝑠) is the Laplace transform of 𝑓(𝑡), and
lim 𝑓(𝑡) exists, then lim 𝑓(𝑡) = lim 𝑠𝐹(𝑠)
t→∞ t→∞ s→0
∞
𝑑
ℒ[𝑓 ′ (𝑡)] = 𝑠𝐹(𝑠) − 𝑓(0) ≪≫ ∫[ 𝑓(𝑡)] 𝑒 −𝑠𝑡 𝑑𝑡 = 𝑠𝐹(𝑠) − 𝑓(0)
𝑑𝑡
0
∞
𝑑
lim ∫ [ 𝑓(𝑡)] 𝑒−𝑠𝑡 𝑑𝑡 = lim {𝑠𝐹(𝑠) − 𝑓(0)}
s→0 𝑑𝑡 s→0
0

lim 𝑒−𝑠𝑡 → 1 ≪≫ [𝑓(𝑡)]∞

0 = lim {𝑠𝐹(𝑠)} − 𝑓(0)
s→0 s→0
𝑓(∞) − 𝑓(0) = lim {𝑠𝐹(𝑠)} − 𝑓(0)
s→0
lim 𝑓(𝑡) = lim 𝑠𝐹(𝑠)
t→∞ s→0

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

Initial value theorem

If 𝑓(𝑡) and 𝑓’(𝑡) are both Laplace transformable, if 𝐹(𝑠) is the Laplace transform of 𝑓(𝑡), and
lim 𝑠𝐹(𝑠) exists, then 𝑓(0+ ) = lim 𝑠𝐹(𝑠)
s→∞ s→∞
∞
𝑑
ℒ[𝑓 ′ (𝑡)] = 𝑠𝐹(𝑠) − 𝑓(0+ ) ≪≫ ∫[ 𝑓(𝑡)] 𝑒 −𝑠𝑡 𝑑𝑡 = 𝑠𝐹(𝑠) − 𝑓(0+ )
𝑑𝑡
0
∞
𝑑
lim ∫ [ 𝑓(𝑡)] 𝑒 −𝑠𝑡 𝑑𝑡 = lim {𝑠𝐹(𝑠) − 𝑓(0+ )}
s→∞ 𝑑𝑡 s→∞
0+
lim 𝑒−𝑠𝑡 → 0 ≪≫ 0 = lim {𝑠𝐹(𝑠)} − 𝑓(0+ )
s→∞ s→∞
𝑓(0+ ) = lim {𝑠𝐹(𝑠)}
s→∞

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1

Example 1
𝑑2 𝑦 𝑑𝑦
Consider a system represented by the differential equation 𝑑𝑡 2
+ 4 𝑑𝑡 + 3𝑦 = 2 𝑟(𝑡); where
𝑑𝑦
the initial conditions are 𝑦(0) = 1, (0) = 0, 𝑎𝑛𝑑 𝑟(𝑡) = 1, 𝑡 ≥ 0. What is the output 𝑦(𝑡)?
𝑑𝑡
What is the steady state value of the output?
𝑑2 𝑦 𝑑𝑦
2
+4 + 3𝑦 = 2
𝑑𝑡 𝑑𝑡
Taking Laplace Transform on both sides,
1
{𝑠 2 𝑌(𝑠) − 𝑠𝑦(0) − 𝑦̇ (0)} + 4{𝑠𝑌(𝑠) − 𝑦(0)} + 3 𝑌(𝑠) = 2 ( )
𝑠
2
2 2 𝑠+4
{𝑠 + 4𝑠 + 3} 𝑌(𝑠) = + 𝑠 + 4 ≪≫ 𝑌(𝑠) = +
𝑠 𝑠(𝑠 2 + 4𝑠 + 3) 𝑠 2 + 4𝑠 + 3
2 + 𝑠(𝑠 + 4) 𝑠 2 + 4𝑠 + 2
𝑌(𝑠) = =
𝑠(𝑠 2 + 4𝑠 + 3) 𝑠(𝑠 + 1)(𝑠 + 3)
Using the partial fraction
𝑠 2 + 4𝑠 + 2 𝐴 𝐵 𝐶
= + +
𝑠(𝑠 + 1)(𝑠 + 3) 𝑠 𝑠 + 1 𝑠 + 3
𝑠 2 + 4𝑠 + 2 = 𝐴(𝑠 + 1)(𝑠 + 3) + 𝐵𝑠(𝑠 + 3) + 𝐶𝑠(𝑠 + 1)
2
𝑃𝑢𝑡 𝑠 = 0 ≫≫ 2 = 𝐴 × 1 × 3 ≫≫ 𝐴 =
3
1
𝑃𝑢𝑡 𝑠 = −1 ≫≫ 1 − 4 + 2 = 𝐵 × (−1) × 2 ≫≫ 𝐵 =
2
−1
𝑃𝑢𝑡 𝑠 = −3 ≫≫ 9 − 12 + 2 = 𝐶 × (−3) × (−2) ≫≫ 𝐶 =
6
2 1 1 1 1 1
𝑌(𝑠) = ( ) + ( ) −( )
3 𝑠 2 𝑠+1 6 𝑠+3
Taking inverse Laplace Transform
2 1 −𝑡 1 −3𝑡
𝑦(𝑡) = + 𝑒 − 𝑒
3 2 6
The steady state value of the output
2 1 1 2
𝑦𝑠𝑠 = lim 𝑦(𝑡) = lim { + 𝑒 −𝑡 − 𝑒 −3𝑡 } =
𝑡→∞ 𝑡→∞ 3 2 6 3
We can also use the final value theorem,
2 1 1 1 1 1 2
𝑦𝑠𝑠 = lim 𝑦(𝑡) = lim 𝑠𝑌(𝑠) = lim 𝑠 {( ) + ( ) −( ) }=
𝑡→∞ 𝑠→0 𝑠→0 3 𝑠 2 𝑠+1 6 𝑠+3 3

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

Example 2
Find the transfer function 𝐺(𝑠) = 𝐸𝐸𝑜 (𝑠) of the R-L-C circuit shown in
(𝑠)
𝑖

figure
𝑒𝑖 = 𝑒𝐿 + 𝑒𝑅 + 𝑒𝐶 We know that
𝑑𝑖 1 𝑑𝑞 𝑑𝑖 𝑑2 𝑞 𝑑𝑞
𝑒𝐿 = 𝐿 ≪≫ 𝑒𝑅 = 𝑅 𝑖 ≪≫ 𝑒𝐶 = ∫ 𝑖 𝑑𝑡 𝑖= ≪≫ 𝑑𝑡 = ≪≫ ∫ 𝑖 𝑑𝑡 = ∫ ( ) 𝑑𝑡 = 𝑞
𝑑𝑡 𝐶 𝑑𝑡 𝑑𝑡2 𝑑𝑡
𝑑𝑖 1 𝑑𝑞 1 𝑑2 𝑞
Input voltage, 𝑒𝑖 = 𝐿 𝑑𝑡
+ 𝑅 𝑖 + 𝐶 ∫ 𝑖 𝑑𝑡 …………(1) (1) → 𝑒𝑖 = 𝐿 + 𝑞 +𝑅
1
𝑑𝑡 𝐶 𝑑𝑡2
Output voltage, 𝑒𝑜 = 𝑒𝐶 = 𝐶 ∫ 𝑖 𝑑𝑡 ……………….(2) 1
(2) → 𝑒𝑜 = 𝑒𝐶 = 𝑞
𝐶
Take Laplace Transforms on both sides of eqns Take Laplace Transforms on both sides of
(1) and (2) and keeping all initial conditions equal eqns (1) and (2) and keeping all initial
to zero conditions equal to zero
𝐼(𝑠) 𝐿𝐶𝑠 2 + 𝑅𝐶𝑠 + 1 1
𝐸𝑖 (𝑠) = 𝐿 𝑠𝐼(𝑠) + 𝑅 𝐼(𝑠) +
𝐶𝑠
=(
𝐶𝑠
) 𝐼(𝑠) 𝐸𝑖 (𝑠) = (𝐿𝑠 2 + 𝑅𝑠 + ) 𝑄(𝑠)
𝐶
𝐼(𝑠) 1
𝐸𝑜 (𝑠) = 𝐸𝑜 (𝑠) = 𝑄(𝑠)
𝐶𝑠 𝐶
The transfer function of the system, 1 𝑄(𝑠)
𝐸𝑜 (𝑠) 𝐶 1
𝐼(𝑠) 𝐺(𝑠) = = =
𝐸𝑜 (𝑠) 𝐶𝑠 1 𝐸𝑖 (𝑠) 2 1
(𝐿𝑠 + 𝑅𝑠 + ) 𝑄(𝑠)
𝐿𝐶𝑠 2 + 𝑅𝐶𝑠 + 1
𝐺(𝑠) = = = 𝐶
𝐸 (𝑠)
𝑖 𝐿𝐶𝑠2 + 𝑅𝐶𝑠 + 1 𝐿𝐶𝑠2 + 𝑅𝐶𝑠 + 1
( ) 𝐼(𝑠)
𝐶𝑠

Example 3
Find the transfer function 𝐺(𝑠) = of the electrical system
𝑉𝑐 (𝑠)
𝑉 (𝑠)
shown
𝑑(𝑖1 − 𝑖2 ) 1 𝑑(𝑖1 − 𝑖2 ) 1
𝑣(𝑡) = 𝑖1 𝑅1 + 𝐿 𝑖2 𝑅2 + ∫ 𝑖2 𝑑𝑡 − 𝐿 =0 𝑣𝑐 = ∫ 𝑖 𝑑𝑡
𝑑𝑡 𝐶 𝑑𝑡 𝐶 2
Taking Laplace Transform on both sides with zero initial conditions
𝑉(𝑠) = 𝐼1 (𝑠)𝑅1 + 𝐿𝑠{𝐼1 (𝑠) − 𝐼2 (𝑠)} 𝐼2 (𝑠) 𝐼2 (𝑠)
𝐼2 (𝑠)𝑅2 + − 𝐿𝑠{𝐼1 (𝑠) − 𝐼2 (𝑠)} = 0 𝑉𝑐 (𝑠) =
𝐶𝑠 𝐶𝑠
1
𝐿𝑠 𝐼1 (𝑠) = 𝐼2 (𝑠) {𝐿𝑠+𝑅2 + }
𝐶𝑠
1
𝐿𝑠+𝑅2 + 𝐶𝑠
𝐼 (𝑠) = 1 𝐼 (𝑠) 2
𝐿𝑠
𝐿𝐶𝑠2 +𝑅2 𝐶𝑠 + 1
𝐼1 (𝑠) = 𝐼2 (𝑠)
𝐿𝐶𝑠2
𝐿𝐶𝑠2 +𝑅2 𝐶𝑠 + 1
𝑉(𝑠) = 𝐼1 (𝑠){𝑅1 + 𝐿𝑠} − 𝐿𝑠𝐼2 (𝑠) = { } {𝑅1 + 𝐿𝑠}𝐼2 (𝑠) − 𝐿𝑠 𝐼2 (𝑠)
𝐿𝐶𝑠2
𝐿𝐶𝑠2 +𝑅2 𝐶𝑠 + 1
= [{ } {𝑅1 + 𝐿𝑠} − 𝐿𝑠] 𝐼2 (𝑠)
𝐿𝐶𝑠2
1
𝑉𝑐 (𝑠) 𝐶𝑠
𝐺(𝑠) = =
𝑉(𝑠) 𝐿𝐶𝑠2 +𝑅2 𝐶𝑠 + 1
{ 2 } {𝑅1 + 𝐿𝑠} − 𝐿𝑠
𝐿𝐶𝑠

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

Example 4 𝑘 𝑥(𝑡)
𝑚 𝑓(𝑡)
𝑋(𝑠)
Find the transfer function 𝐺(𝑠) = 𝐹(𝑠)of the mechanical system shown.
𝑥, 𝑥̈
Consider the freebody diagram of the system when the mass moves a
𝑚𝑔
distance 𝑥 to right of the equilibrium position. 𝑘𝑥 𝑓(𝑡)
The spring force, 𝐹𝑠 = 𝑘𝑥, where x is the relative displacement 𝑅
between the ends of the spring.
The equation of motion of the system is 𝑚𝑥̈ = Σ𝐹𝑥
𝑚𝑥̈ = 𝑓(𝑡) − 𝑘𝑥 ≫≫ 𝑚𝑥̈ + 𝑘𝑥 = 𝑓(𝑡)
Taking Laplace transforms on both sides and assuming zero initial conditions,
𝑋(𝑠) 1
𝑚𝑠 2 𝑋(𝑠) + 𝑘𝑋(𝑠) = 𝐹(𝑠) ≫≫ 𝑋(𝑠){𝑚𝑠 2 + 𝑘} = 𝐹(𝑠) ≫≫ 𝐺(𝑠) = = 2
𝐹(𝑠) 𝑚𝑠 + 𝑘
Example 5
𝑋(𝑠)
Find the transfer function 𝐺(𝑠) = 𝐹(𝑠)of the mechanical system 𝑘
𝑚
𝑥(𝑡)
𝑓(𝑡)
represented by the spring-mass-damper shown. 𝑏

The damping force, 𝐹𝑑 = 𝑏𝑥̇ , where 𝑏 is the damping coefficient and 𝑥̇ 𝑚𝑔

𝑥, 𝑥̈

is the relative velocity between the ends of the damper. 𝑘𝑥

𝑏𝑥̇ 𝑓(𝑡)
𝑅
The equation of motion of the system is 𝑚𝑥̈ = Σ𝐹𝑥
𝑚𝑥̈ = 𝑓(𝑡) − 𝑘𝑥 − 𝑏𝑥̇ ≫≫ 𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥 = 𝑓(𝑡)
Taking Laplace transforms on both sides and assuming zero initial conditions,
𝑋(𝑠) 1
𝑚𝑠 2 𝑋(𝑠) + 𝑏𝑠𝑋(𝑠) + 𝑘𝑋(𝑠) = 𝐹(𝑠) ≪≫ 𝑋(𝑠){𝑚𝑠 2 + 𝑏𝑠 + 𝑘} = 𝐹(𝑠) ≫ 𝐺(𝑠) = =
𝐹(𝑠) 𝑚𝑠 2 + 𝑏𝑠 + 𝑘
Example 6
𝑥2 (𝑡) 𝑥1 (𝑡)
𝑋2 (𝑠)
Find the transfer function 𝐺(𝑠) = 𝐹(𝑠)
of the mechanical 𝑘
𝑘
𝑚 𝑚 𝑓(𝑡)
system represented by the spring-mass-damper shown. 𝑏

The equations of motion of the system 𝑥2 , 𝑥2̈ 𝑥1 , 𝑥1̈

𝑚𝑔
𝑚𝑥̈1 = 𝑓(𝑡) − 𝑘(𝑥1 − 𝑥2 ) − 𝑏(𝑥1̇ − 𝑥2̇ ) 𝑚𝑔
𝑘(𝑥1 − 𝑥2 )
𝑚𝑥̈1 + 𝑏(𝑥1̇ − 𝑥2̇ ) + 𝑘(𝑥1 − 𝑥2 ) = 𝑓(𝑡)
𝑘𝑥2 𝑓(𝑡)
𝑏(𝑥1̇ − 𝑥2̇ )
𝑚𝑥̈ 2 = 𝑘(𝑥1 − 𝑥2 ) + 𝑏(𝑥1̇ − 𝑥2̇ ) − 𝑘𝑥2 𝑅 𝑅
𝑚𝑥̈ 2 − 𝑏(𝑥1̇ − 𝑥2̇ ) − 𝑘(𝑥1 − 𝑥2 ) + 𝑘𝑥2 = 0
Taking Laplace transforms on both sides of these equations and assuming zero initial
conditions,
𝑚𝑠 2 𝑋1 (𝑠) + 𝑏𝑠{𝑋1 (𝑠) − 𝑋2 (𝑠)} + 𝑘{𝑋1 (𝑠) − 𝑋2 (𝑠)} = 𝐹(𝑠)
𝑋1 (𝑠){𝑚𝑠 2 + 𝑏𝑠 + 𝑘} − 𝑋2 (𝑠){𝑏𝑠 + 𝑘} = 𝐹(𝑠)
𝑚𝑠 2 𝑋2 (𝑠) − 𝑏𝑠{𝑋1 (𝑠) − 𝑋2 (𝑠)} − 𝑘{𝑋1 (𝑠) − 𝑋2 (𝑠)} + 𝑘𝑋2 (𝑠) = 0
𝑚𝑠 2 + 𝑏𝑠 + 2𝑘
𝑋2 (𝑠){𝑚𝑠 2 + 𝑏𝑠 + 2𝑘} − 𝑋1 (𝑠){𝑏𝑠 + 𝑘} = 0 ≫≫ 𝑋1 (𝑠) = { } 𝑋2 (𝑠)
𝑏𝑠 + 𝑘
𝑚𝑠 2 + 𝑏𝑠 + 2𝑘
𝑋2 (𝑠) { } {𝑚𝑠 2 + 𝑏𝑠 + 𝑘} − 𝑋2 (𝑠){𝑏𝑠 + 𝑘} = 𝐹(𝑠)
𝑏𝑠 + 𝑘
𝑋2 (𝑠) 𝑏𝑠 + 𝑘
𝐺(𝑠) = =
𝐹(𝑠) (𝑚𝑠2 + 𝑏𝑠 + 2𝑘)(𝑚𝑠2 + 𝑏𝑠 + 𝑘) − (𝑏𝑠 + 𝑘)2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

Exercise 1
𝐶(𝑠)
Find the transfer function, 𝐺(𝑠) = 𝑅(𝑠), corresponding to the differential equation
𝑑3𝑐 𝑑2𝑐 𝑑𝑐 𝑑2 𝑟 𝑑𝑟
𝑑𝑡 3
+ 3 𝑑𝑡 2 + 7 𝑑𝑡 + 5𝑐 = 𝑑𝑡 2 + 4 𝑑𝑡 + 3𝑟
Exercise 2
2𝑠+1
Find the differential equation corresponding to the transfer function 𝐺(𝑠) =
𝑠2 +6𝑠+2

Exercise 3
𝑠
Find the ramp response for a system whose transfer function is 𝐺(𝑠) =
(𝑠+4)(𝑠+8)

Exercise 4
𝐼2 (𝑠)
Find the transfer function 𝐺(𝑠) = 𝑉 (𝑠)
of the electrical system
shown
Example 5
𝑥2 (𝑡) 𝑥1 (𝑡)
𝑋1 (𝑠)
Find the transfer function 𝐺(𝑠) = of the mechanical 𝑘
𝑘
𝐹(𝑠)
𝑚 𝑚 𝑓(𝑡)
system represented by the spring-mass-damper shown. 𝑏

Example 6
𝑥2 (𝑡) 𝑥1 (𝑡)
𝑋2 (𝑠)
Find the transfer function 𝐺(𝑠) = 𝐹(𝑠)
of the mechanical 𝑘
𝑘
𝑚 𝑚 𝑓(𝑡)
system represented by the spring-mass-damper shown. 𝑏 𝑏

Example 7
Obtain the transfer function of a dc motor i) armature controlled ii) field controlled
Example 8
Obtain the partial fraction expansion of the following function using MATLAB/SCILAB:
10(𝑠 + 2)(𝑠 + 4)
𝐹(𝑠) =
(𝑠 + 1)(𝑠 + 3)(𝑠 + 5)2
Then, obtain the inverse Laplace transform of F(s).
Example 9
Obtain the partial fraction expansion of the following function using MATLAB/SCILAB:
(𝑠 + 1)
𝐹(𝑠) =
𝑠(𝑠 2 + 𝑠 + 1)
Then, obtain the inverse Laplace transform of F(s).

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
Electrical-Mechanical Analogous systems
Series 𝑅 − 𝐿 − 𝐶 circuit Mechanical System Parallel 𝑅 − 𝐿 − 𝐶 circuit
𝑖

𝑘 𝑥(𝑡) 𝑣
𝑣 𝑚 𝑓(𝑡)
𝑑q
𝑖= 𝑏
𝑑𝑡 𝑑𝜙
𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥 = 𝑓(𝑡) 𝑣=
𝑑𝑖 1 𝑑𝑡
𝐿 + 𝑅 𝑖 + ∫ 𝑖 𝑑𝑡 = 𝑣 𝑣 𝑑𝑣 1
𝑑𝑡 𝐶 +𝐶 + + ∫ 𝑣 𝑑𝑡 = 𝑖
2
𝑑 𝑞 𝑑𝑞 𝑞 𝑑2 𝑥 𝑑𝑥 𝑅 𝑑𝑡 𝐿
𝐿 2 +𝑅 + =𝑣 𝑚 2 +𝑏 + 𝑘𝑥 = 𝑓(𝑡) 2
𝑑 𝜙 1 𝑑𝜙 𝜙
𝑑𝑡 𝑑𝑡 𝐶 𝑑𝑡 𝑑𝑡 𝐶 2+ + =𝑖
𝑑𝑡 𝑅 𝑑𝑡 𝐿
𝑉𝑜𝑙𝑡𝑎𝑔𝑒 𝐹𝑜𝑟𝑐𝑒 𝐶𝑢𝑟𝑟𝑒𝑛𝑡
𝐶ℎ𝑎𝑟𝑔𝑒 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝐹𝑙𝑢𝑥
𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑉𝑜𝑙𝑡𝑎𝑔𝑒
𝐼𝑛𝑑𝑢𝑐𝑡𝑎𝑛𝑐𝑒 𝑀𝑎𝑠𝑠 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒
𝑅𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝐷𝑎𝑚𝑝𝑖𝑛𝑔 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 1/𝑅𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒
1 𝑆𝑝𝑟𝑖𝑛𝑔 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 1/𝐼𝑛𝑑𝑢𝑐𝑡𝑎𝑛𝑐𝑒
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒

Example 1
𝑥2 (𝑡) 𝑥1 (𝑡)
Obtain the force-voltage and force-current analogous 𝑘2
𝑘1
systems for the mechanical system shown. 𝑚2 𝑚1 𝑓(𝑡)
𝑏1

The governing differential equations for the mechanical 𝑥2 , 𝑥2̈ 𝑥1 , 𝑥1̈

system are 𝑚2 𝑔 𝑚1 𝑔
𝑘1 (𝑥1 − 𝑥2 )
𝑚1 𝑥̈1 + 𝑏1 (𝑥1̇ − 𝑥2̇ ) + 𝑘1 (𝑥1 − 𝑥2 ) = 𝑓(𝑡)
𝑘2 𝑥2 𝑓(𝑡)
𝑚2 𝑥̈ 2 − 𝑏1 (𝑥1̇ − 𝑥2̇ ) − 𝑘1 (𝑥1 − 𝑥2 ) + 𝑘2 𝑥2 = 0 𝑏1(𝑥1̇ − 𝑥2̇ )
𝑅 𝑅
𝐹 𝑣 1
𝐿1 𝑞̈ 1 + 𝑅1 (𝑞1̇ − 𝑞2̇ ) + (𝑞 − 𝑞2 ) = 𝑣
𝑥 𝑞 𝐶1 1
𝑥̇ 𝑖 𝑑𝑖1 1
𝐿1 + 𝑅1 (𝑖1 − 𝑖2 ) + ∫(𝑖1 − 𝑖2 ) 𝑑𝑡 = 𝑣
𝑚 𝐿 𝑑𝑡 𝐶1
1 1
𝑏 𝑅 𝐿2 𝑞̈ 2 − 𝑅1 (𝑞1̇ − 𝑞2̇ ) − (𝑞1 − 𝑞2 ) − 𝑞2 = 0
𝐶1 𝐶2
𝑘 1
𝑑𝑖2 1 1
𝐶 𝐿2 − 𝑅1 (𝑖1 − 𝑖2 ) − ∫(𝑖1 − 𝑖2 )𝑑𝑡 + ∫ 𝑖2 𝑑𝑡 = 0
𝑑𝑡 𝐶1 𝐶2
𝐹 𝑖 1 1
𝐶1 𝜙̈1 + (𝜙1̇ − 𝜙̇2 ) + (𝜙1 − 𝜙2 ) = 𝑖
𝑥 𝜙 𝑅1 𝐿1
𝑥̇ 𝑣 𝑑𝑣1 1 1
𝐶1 + (𝑣1 − 𝑣2 ) + ∫(𝑣1 − 𝑣2 ) 𝑑𝑡 = 𝑖
𝑚 𝐶 𝑑𝑡 𝑅1 𝐿1
1 1 1 1
𝑏 𝐶2 𝜙̈2 − (𝜙1̇ − 𝜙2̇ ) − (𝜙1 − 𝜙2 ) + 𝜙2 = 0
𝑅 𝑅1 𝐿1 𝐿2
𝑘
1 𝑑𝑣2 1 1 1
𝐿 𝐶2 + ∫ 𝑣2 𝑑𝑡 = (𝑣1 − 𝑣2 ) + ∫(𝑣1 − 𝑣2 ) 𝑑𝑡
𝑑𝑡 𝐿2 𝑅1 𝐿1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

Exercise 1
𝑥2 (𝑡) 𝑥1 (𝑡)
Obtain the force-voltage and force-current 𝑘2
𝑘1
analogous systems for the mechanical system 𝑚2 𝑚1 𝑓(𝑡)
𝑏1
shown. 𝑏2

Exercise 2 𝑥2 (𝑡) 𝑥1 (𝑡)

𝑥3 (𝑡)
Obtain the force-voltage and force-current 𝑘2
𝑘3 𝑘1
analogous systems for the mechanical system 𝑚3 𝑚2 𝑚1 𝑓(𝑡)
𝑏1
shown. 𝑏2

Exercise 3
Obtain the equivalent mechanical system

Exercise 4 𝑘3
Obtain the force-voltage and force-current 𝑘4 𝑥3 (𝑡)
𝑚3
analogous systems for the mechanical system
shown. 𝑘2
𝑏2

𝑚2
𝑥2 (𝑡)

𝑏1 𝑘1

𝑚1
𝑥1 (𝑡)

𝑓(𝑡)

Exercise 5 𝑘3
Obtain the force-voltage and force-current 𝑘4 𝑥3 (𝑡)
𝑚3
analogous systems for the mechanical system 𝑏4
shown. 𝑏2
𝑘2

𝑚2
𝑥2 (𝑡)

𝑏1 𝑘1

𝑚1
𝑥1 (𝑡)

𝑓(𝑡)
Exercise 6 𝑘3
𝑏3
Obtain the force-voltage and force-current 𝑘1 𝑥2 (𝑡)
𝑚2
analogous systems for the mechanical system 𝑏1
shown. 𝑏2
𝑘2

𝑚1
𝑥1 (𝑡)

𝑓(𝑡)

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
Block Diagram Reduction
1. Blocks in
series
𝑋2 = 𝐺1 𝑋1 ≪≫ 𝑋3 = 𝐺2 𝑋2
𝑋3 = 𝐺1 𝐺2 𝑋1
2. Parallel
blocks

3. Moving a
summing
point
ahead of a
block 𝐶(𝑠) = {𝑅(𝑠) ∓ 𝑋(𝑠)}𝐺(𝑠)
𝐶(𝑠) = {𝑅(𝑠)𝐺(𝑠) ∓ 𝑋(𝑠)𝐺(𝑠)}
4. Moving a
summing
point
behind of
𝐶(𝑠) = 𝑅(𝑠)𝐺(𝑠) ∓ 𝑋(𝑠)
a block

1
𝐶(𝑠) = {𝑅(𝑠) ∓ 𝑋(𝑠) } 𝐺(𝑠)
𝐺(𝑠)
5. Moving a
pick off
point
1
ahead of a 𝑋2 = 𝐺𝑋1 ≪≫ 𝑋1 = 𝑋2
𝐺
block
6. Moving a
pick off
point 𝑋2 = 𝐺𝑋1
behind a
block
7. Reducing a 𝑌
𝑌 = 𝑋1 − 𝑍 ≫ 𝑍 = 𝐻𝑋2 ≫
feedback 𝑍
𝑋2 = 𝐺𝑌 = 𝐺(𝑋1 − 𝑍) = 𝐺(𝑋1 − 𝐻𝑋2 )
loop
𝐺
𝑋2 = ( )𝑋
1 ∓ 𝐺𝐻 2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

Example 1
Reduce the given block diagram and
find the overall transfer function of
the system.

Moving the pickoff point ahead of the Combining the series blocks 𝐺3 and 𝐺4
block 𝐺4

Eliminating the feedback loop Combining the series blocks

Eliminating the feedback loop Combining the series blocks

Eliminating the feedback loop

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

Example 2
Determine the closed loop transfer function 𝐻3
𝐶(𝑠) 𝑅(𝑠) + − +
𝐺1 𝐺2 𝐺3 𝐶(𝑠)
𝑇(𝑠) = 𝑅(𝑠) + −
−
𝐻1 𝐻2

Moving the pickoff point ahead of block 𝐺3 Combining the series blocks 𝐺2 and 𝐺3
𝐻3/𝐺3
𝐻3/𝐺3
𝑅(𝑠) + − +
+ 𝐺1 𝐺2𝐺3 𝐶(𝑠)
𝑅(𝑠) + − + −
+ 𝐺1 𝐺2 𝐺3 𝐶(𝑠) −
− −
𝐻1 𝐻2
𝐻1 𝐻2

Eliminating the feedback loop Moving athe pickoff point ahead of the
𝐻3/𝐺3 block 𝐻3/𝐺3
𝑅(𝑠) + − 𝐺2 𝐺3 𝑅(𝑠) + − 𝐺2 𝐺3
+ 𝐺1 𝐶(𝑠) + 𝐺1 𝐶(𝑠)
− 1 + 𝐺2 𝐺3 𝐻2 − 1 + 𝐺2 𝐺3 𝐻2
𝐻1
𝐻1 𝐺2 𝐺3
1 + 𝐺2 𝐺3 𝐻2

Combining the series block Eliminating the feedback loop

𝐻3/𝐺3 𝐺1 𝐺2 𝐺3
𝑅(𝑠) + 𝐶(𝑠)
𝑅(𝑠) + − 𝐺1 𝐺2 𝐺3 1 + 𝐺1 𝐺2 𝐻3 + 𝐺2 𝐺3 𝐻2
+ 𝐶(𝑠) −
− 1 + 𝐺2 𝐺3 𝐻2
𝐻1 𝐻1
𝐺2 𝐺3 𝐺2 𝐺3
1 + 𝐺2 𝐺3 𝐻2 1 + 𝐺2 𝐺3 𝐻2

Eliminating the feed-back loop

𝑅(𝑠) 𝐺1 𝐺2 𝐺3 𝐶(𝑠)
1 + 𝐺1 𝐺2 𝐻3 + 𝐺2 𝐺3 𝐻2 + 𝐺1 𝐻1 + 𝐺1 𝐺2 𝐺3 𝐻1 𝐻2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

Exercise 1
𝑌(𝑠)
Determie the closed loop transfer function 𝑇(𝑠) = 𝑅(𝑠)

Exercise 2
𝐶(𝑠)
Determie the closed loop transfer function 𝑇(𝑠) =
𝑅(𝑠)

Exercise 3
𝐶(𝑠)
Determie the closed loop transfer function 𝑇(𝑠) = 𝑅(𝑠)

Exercise 4
𝐶(𝑠)
Determie the closed loop transfer function 𝑇(𝑠) = 𝑅(𝑠)

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
Signal Flow Graph
Signal flow graphs are alternative to the block diagrams. Block diagrams consist of blocks
(systems represented by transfer functions), signals, summing points, and pick off points. But,
signal flow graph consists of branches and nodes. Branches represent the sytems (transfer
functions) and nodes represent the signals.
As we are dealing with linear systems, the input-output relation for
each block will be linear. 𝑅(𝑠) 𝐶(𝑠)
𝐺(𝑠)
𝐶(𝑠) = 𝐺(𝑠)𝑅(𝑠)
For the complete system, we will get a number of such linear
equations. 𝑅(𝑠)

Signal flow graphs are graphical representation of linear equations. The 𝐺(𝑠) 𝐶(𝑠)
above equation is represented as a signal flow graph as shown.
𝑛𝑜𝑑𝑒 𝑏𝑟𝑎𝑛𝑐ℎ

The input and output points or junctions are called nodes. All branches leaving a node will pass
on the nodal signal signal to the output node all of each branch unidirectionally. The summation
of all variables entering a node is equal to the node variable. A path is a branch or a continuous
sequence of brances from one node to another. A loop is a closed path that originates from a
node and terminates on the same node, with no node being met twice along the path. Two loops
are said to be nontouching if they do not have a common node.
Example 1 𝐻3
𝑅(𝑠) + − +
Consider the block diagram representation of a + 𝐺1
−
𝐺2 𝐺3 𝐶(𝑠)
−
control system shown. Draw the signal flow graph. 𝐻1 𝐻2
Let 𝑎, 𝑏, 𝑐, … … represent the output signals from 𝐻3
each functional block. Let us write the linear input- 𝑅(𝑠) + 𝑎 −𝑏 𝑓
𝐺1 𝑐
+ 𝑑 𝐺2 𝑒 𝐺3
𝐶(𝑠)
+ −
output relations for each of these blocks. −
𝑔 𝐻1 ℎ 𝐻2

𝑎 = 𝑅(𝑠) − 𝑔 𝑔 = 𝑐. 𝐻1 −1 𝐻3
𝑓
𝑏 =𝑎−𝑓 𝑓 = 𝑒. 𝐻3 𝐺3 𝐶(𝑠)
𝑅(𝑠) 𝑎 𝑏 𝑐 𝑑 𝐺2 𝑒
1 1 𝐺1 1
𝑐 = 𝑏. 𝐺1
𝑑 =𝑐−ℎ ℎ = 𝐶(𝑠). 𝐻2
𝐻2
𝑒 = 𝑑. 𝐺2 −1 𝐻1 −1
𝑔 ℎ
𝐶(𝑠) = 𝑒. 𝐺2
In signal flow graph, these signals are represented by nodes. Branches represent the transfer
functions of the blocks. The values written on the branches like (1, −1, 𝐺1 , 𝐻1 𝑒𝑡𝑐) are known as
branch gains. The signal flow graph has one path from the input node to the output node.
𝑃1 ⟹ 𝑅(𝑠) − 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝐶(𝑠) The path gain 𝑃1 is the product of all branch gains in the
path. 𝑃1 = 1 × 1 × 𝐺1 × 1 × 𝐺2 × 𝐺3 = 𝐺1 𝐺2 𝐺3
There are three loops:
𝐿1 ⟹ 𝑎 − 𝑏 − 𝑐 − 𝑔 − 𝑎 The loop gain is the product of all branch gains in the loop.
𝐿2 ⟹ 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝑏 𝐿1 = 1 × 𝐺1 × 𝐻1 × −1 = −𝐺1 𝐻1
𝐿3 ⟹ 𝑑 − 𝑒 − 𝐶(𝑠) − ℎ − 𝑑 𝐿2 = 𝐺1 × 1 × 𝐺2 × 𝐻3 × −1 = −𝐺1 𝐺2 𝐻3
𝐿3 = 𝐺2 × 𝐺3 × 𝐻2 × −1 = −𝐺2 𝐺3 𝐻2
Since there are common nodes, loops 𝐿1 and 𝐿2 are touching each other like loops 𝐿2 and 𝐿3 . The
loops 𝐿1 and 𝐿3 are nontouching.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
The signal flow graph is a pictorial representation of the linear algebraic expressions showing
the interdepencies of variables.
Mason’s signal flow gain formula
Consider the following set of simultaneous algebraic expressions 𝑎11
𝑟1 𝑏1
𝑏1 𝑟1 +𝑎11 𝑥1 + 𝑎12 𝑥2 = 𝑥1 𝑥1
𝑏2 𝑟2 +𝑎21 𝑥1 + 𝑎22 𝑥2 = 𝑥2 𝑎21 𝑎12
𝑏2
The input variables are 𝑟1 and 𝑟2 , and the output variables are 𝑥1 and 𝑥2 . 𝑥2
𝑟2
The signal flow graph for the above equations is shown in the figure. 𝑎22

There are two paths each from the input side to each output
𝑃111 (𝑝𝑎𝑡ℎ 1 𝑓𝑟𝑜𝑚 𝑖𝑛𝑝𝑢𝑡 1 𝑡𝑜 𝑜𝑢𝑡𝑝𝑢𝑡 1) ⟹ 𝑟1 → 𝑥1 𝑃112 (𝑝𝑎𝑡ℎ 1 𝑓𝑟𝑜𝑚 𝑖𝑛𝑝𝑢𝑡 1 𝑡𝑜 𝑜𝑢𝑡𝑝𝑢𝑡 2) ⟹ 𝑟1 → 𝑥1 → 𝑥2
Path gain, 𝑃111 = 𝑏1 Path gain, 𝑃112 = 1. 𝑎21 = 𝑏1 𝑎21
𝑃 221 (𝑝𝑎𝑡ℎ 2 𝑓𝑟𝑜𝑚 𝑖𝑛𝑝𝑢𝑡 2 𝑡𝑜 𝑜𝑢𝑡𝑝𝑢𝑡 1) ⟹ 𝑟2 → 𝑥2 → 𝑥1 𝑃 222 (𝑝𝑎𝑡ℎ 2 𝑓𝑟𝑜𝑚 𝑖𝑛𝑝𝑢𝑡 2 𝑡𝑜 𝑜𝑢𝑡𝑝𝑢𝑡 2) ⟹ 𝑟2 → 𝑥2
Path gain, 𝑃 221 = 𝑏2 . 𝑎12 = 𝑏2 𝑎12 Path gain, 𝑃 222 = 𝑏2

There are three loops in the signal flow graph with loop gains, 𝐿1 = 𝑎11 , 𝐿2 = 𝑎22 , 𝑎𝑛𝑑 𝐿3 = 𝑎12 𝑎21 .
Of these, loops 𝐿1 and 𝐿2 are not touching each other.
The equations are re-written as
𝑏1 𝑟1 = (1 − 𝑎11 )𝑥1 + (−𝑎12 )𝑥2 (1 − 𝑎11 ) (−𝑎12 ) 𝑥1 𝑏 𝑟
[ ] { } = { 1 1}
(−𝑎21 ) (1 − 𝑎22 ) 𝑥2 𝑏2 𝑟2
𝑏2 𝑟2 = (−𝑎21 )𝑥1 + (1 − 𝑎22 )𝑥2
𝑏1 𝑟1 −𝑎12 (1 − 𝑎11 ) 𝑏1 𝑟1
| | | |
𝑏2 𝑟2 (1 − 𝑎22 ) (−𝑎21 ) 𝑏2 𝑟2
𝑥1 = 𝑥2 =
(1 − 𝑎11 ) −𝑎12 (1 − 𝑎11 ) −𝑎12
| | | |
(−𝑎21 ) (1 − 𝑎22 ) (−𝑎21 ) (1 − 𝑎22 )
𝑏1(1 − 𝑎22 )𝑟1 + 𝑏2 𝑎12 𝑟2 𝑏1 . (1 − 𝑎22 ) (𝑏2 𝑎12 ). 1 𝑏1(𝑎21 )𝑟1 + 𝑏2 (1 − 𝑎11 )𝑟2 (𝑏1 𝑎21 ). 1 𝑏2 (1 − 𝑎11 )
𝑥1 = = 𝑟1 + 𝑟2 𝑥2 = = 𝑟1 + 𝑟2
1 − (𝑎11 + 𝑎22 + 𝑎12𝑎21 ) + 𝑎11 𝑎22 ∆ ∆ 1 − (𝑎11 + 𝑎22 + 𝑎12 𝑎21 ) + 𝑎11𝑎22 ∆ ∆

The denominator ∆ is the determinant given by

∆= 1 − (𝑠𝑢𝑚 𝑜𝑓 𝑎𝑙𝑙 𝑙𝑜𝑜𝑝 𝑔𝑎𝑖𝑛𝑠) + (𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑜𝑓 𝑎𝑙𝑙 𝑐𝑜𝑚𝑏𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑠 𝑜𝑓 𝑡𝑤𝑜 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 𝑙𝑜𝑜𝑝𝑠)
𝑁

∆= 1 − ∑ 𝐿𝑖 + ∑ 𝐿𝑖 𝐿𝑗 − ∑ 𝐿𝑖 𝐿𝑗 𝐿𝑘 + ⋯ … ..
𝑖=1 𝑖,𝑗 𝑖,𝑗,𝑘
𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔

The numerator for 𝑥1 with the input 𝑟1 is 𝑏1 times (1 − 𝑎22 ) which is the product of the gain of
path 1 from input 1 to output 1 (𝑃11
1 )
and the value of ∆ after eliminating all the loops touching
that path (∆11 ). The numerator for 𝑥1 with the input 𝑟2 is (𝑏2 𝑎12 ) times (1) which is the product
1

of the gain of path 2 from input 2 to output 1 (𝑃21

2 )and
the value of ∆ after eliminating all the
loops touching that path (∆221 ).
Similarly, the numerator for 𝑥2 with the input 𝑟1 is (𝑏1 𝑎21 ) times (1) which is the product of the
gain of path 1 from input 1 to output 2 (𝑃12
1 )
and the value of ∆ after eliminating all the loops
touching that path (∆12 ). The numerator for 𝑥2 with the input 𝑟2 is 𝑏2 times (1 − 𝑎11 ) which is the
1

product of the gain of path 2 from input 2 to output 2 (𝑃22 2 )and

the value of ∆ after eliminating
all the loops touching that path (∆22 ).
2

1
𝑃11 . ∆111 𝑟1 + 𝑃21
2
. ∆221 𝑟2 1
𝑃12 . ∆112 𝑟1 + 𝑃22
2
. ∆222 𝑟2
𝑥1 = ≪≫ 𝑥2 =
∆ ∆
If we have a single input R(s)-single output C(s) system with different paths from input to output
𝑃1 ∆1 𝑅(𝑠) + 𝑃2 ∆2 𝑅(𝑠) + ⋯ 𝑃1 ∆1 + 𝑃2 ∆2 + ⋯
𝐶(𝑠) = =( ) 𝑅(𝑠)
∆ ∆
The transfer function of the system is given by,
𝐶(𝑠) 𝑃1 ∆1 + 𝑃2 ∆2 + ⋯ ∑𝑘 𝑃𝑘 ∆𝑘
𝐺(𝑠) = =( )=
𝑅(𝑠) ∆ ∆

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
This is known as Mason’s Signal flow gain formula.

Example 2 𝐻3

Consider the block diagram representation of a 𝑅(𝑠) +

+
−
𝐺1 +
−
𝐺2 𝐺3 𝐶(𝑠)
−
control system shown. Draw the signal flow graph and 𝐻1 𝐻2
obtain the overall transfer function of the system.
Let 𝑎, 𝑏, 𝑐, … … represent the output signals from each 𝑓 𝐻3
functional block. Let us write the linear input-output 𝑅(𝑠) + 𝑎 −𝑏
𝐺1 𝑐
+ 𝑑 𝐺2 𝑒 𝐺3
𝐶(𝑠)
+ −
relations for each of these blocks. −
𝑔 𝐻1 ℎ 𝐻2

𝑎 = 𝑅(𝑠) − 𝑔 𝑔 = 𝑐. 𝐻1 −1 𝐻3
𝑓
𝑏 =𝑎−𝑓 𝑓 = 𝑒. 𝐻3 𝐺3 𝐶(𝑠)
𝑅(𝑠) 𝑎 𝑏 𝑐 𝑑 𝐺2 𝑒
1 1 𝐺1 1
𝑐 = 𝑏. 𝐺1
𝑑 =𝑐−ℎ ℎ = 𝐶(𝑠). 𝐻2
𝐻2
𝑒 = 𝑑. 𝐺2 −1 𝐻1 −1
𝑔 ℎ
𝐶(𝑠) = 𝑒. 𝐺2
The signal flow graph has one path from the input node to the output node.
𝑃1 ⟹ 𝑅(𝑠) − 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝐶(𝑠) The path gain 𝑃1 is the product of all branch gains in the
path. 𝑃1 = 1 × 1 × 𝐺1 × 1 × 𝐺2 × 𝐺3 = 𝐺1 𝐺2 𝐺3
There are three loops:
𝐿1 ⟹ 𝑎 − 𝑏 − 𝑐 − 𝑔 − 𝑎 The loop gain is the product of all branch gains in the loop.
𝐿2 ⟹ 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝑏 𝐿1 = 1 × 𝐺1 × 𝐻1 × −1 = −𝐺1 𝐻1
𝐿3 ⟹ 𝑑 − 𝑒 − 𝐶(𝑠) − ℎ − 𝑑 𝐿2 = 𝐺1 × 1 × 𝐺2 × 𝐻3 × −1 = −𝐺1 𝐺2 𝐻3
𝐿3 = 𝐺2 × 𝐺3 × 𝐻2 × −1 = −𝐺2 𝐺3 𝐻2
Since there are common nodes, loops 𝐿1 and 𝐿2 are touching each other like loops 𝐿2 and 𝐿3 . The
loops 𝐿1 and 𝐿3 are nontouching.
𝑁

∆= 1 − ∑ 𝐿𝑖 + ∑ 𝐿𝑖 𝐿𝑗 − ∑ 𝐿𝑖 𝐿𝑗 𝐿𝑘 + ⋯ … ..
𝑖=1 𝑖,𝑗 𝑖,𝑗,𝑘
𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔

∆= 1 − (𝐿1 + 𝐿2 + 𝐿3 ) + (𝐿1 𝐿3 ) = 1 + 𝐺1 𝐻1 + 𝐺1 𝐺2 𝐻3 + 𝐺2 𝐺3 𝐻2 + 𝐺1 𝐺2 𝐺3 𝐻1 𝐻2
∆1 is the value of ∆ after eliminating all the loops touching the path 1 (All loops are touching
𝑃1 ).
∆1 = 1
𝐶(𝑠) 𝑃1 ∆1 𝐺1 𝐺2 𝐺3
𝐺(𝑠) = = =
𝑅(𝑠) ∆ 1 + 𝐺1 𝐻1 + 𝐺1 𝐺2 𝐻3 + 𝐺2 𝐺3 𝐻2 + 𝐺1 𝐺2 𝐺3 𝐻1 𝐻2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

Example 3 ℎ
𝑎 𝑏 𝑐 𝑒 𝑓
Draw the signal flow graph and find the overall
𝑑
𝑔
transfer function of the system.
𝑖

𝑎 = 𝑅(𝑠) − 𝑖 𝑖 = 𝑌(𝑠). 𝐻3 𝐻2
−1 ℎ
𝑏 = 𝑎𝐺1 𝑒 𝑓 𝐺4 𝑌(𝑠)
𝑅(𝑠) 𝑎 𝑏 𝑐 𝐺2 𝑑 1 𝐺3
1 𝐺1 1
𝑐 =𝑏−ℎ ℎ = 𝑓. 𝐻2
𝑑 = 𝑐. 𝐺2 1 𝐻1
−1 𝑖
𝑒 =𝑑+𝑔 𝑔 = 𝑌(𝑠). 𝐻1 𝑔

𝑓 = 𝑒𝐺3 𝐻3

𝑌(𝑠) = 𝑓. 𝐺4
The signal flow graph has one path from the input node to the output node.
𝑃1 ⟹ 𝑅(𝑠) − 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝑌(𝑠) The path gain 𝑃1 is the product of all branch gains in the
path. 𝑃1 = 1 × 𝐺1 × 1 × 𝐺2 × 1 × 𝐺3 × 𝐺4 = 𝐺1 𝐺2 𝐺3 𝐺4
There are three loops:
𝐿1 ⟹ 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝑌(𝑠) − 𝑖 The loop gain is the product of all branch gains in the loop.
−𝑎
𝐿1 = 𝐺1 × 1 × 𝐺2 × 1 × 𝐺3 × 𝐺4 × 𝐻3 × −1 = −𝐺1 𝐺2 𝐺3 𝐺4 𝐻3
𝐿2 ⟹ 𝑐 − 𝑑 − 𝑒 − 𝑓 − ℎ − 𝑐
𝐿2 = 𝐺2 × 1 × 𝐺3 × 𝐻2 × −1 = −𝐺2 𝐺3 𝐻2
𝐿3 ⟹ 𝑒 − 𝑓 − 𝑌(𝑠) − 𝑔 − 𝑒
𝐿3 = 𝐺3 × 𝐺4 × 𝐻1 × 1 = +𝐺3 𝐺4 𝐻1
Since there are common nodes, loops 𝐿1 and 𝐿2 are touching each other like loops 𝐿2 and 𝐿3 , and
loops 𝐿1 and 𝐿3 .There are no nontouching loops.
𝑁

∆= 1 − ∑ 𝐿𝑖 + ∑ 𝐿𝑖 𝐿𝑗 − ∑ 𝐿𝑖 𝐿𝑗 𝐿𝑘 + ⋯ … ..
𝑖=1 𝑖,𝑗 𝑖,𝑗,𝑘
𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔

∆= 1 − (𝐿1 + 𝐿2 + 𝐿3 ) = 1 + 𝐺1 𝐺2 𝐺3 𝐺4 𝐻3 +𝐺2 𝐺3 𝐻2 − 𝐺3 𝐺4 𝐻1
∆1 is the value of ∆ after eliminating all the loops touching the path 1 (All loops are touching
𝑃1 ).
∆1 = 1
𝐶(𝑠) 𝑃1 ∆1 𝐺1 𝐺2 𝐺3 𝐺4
𝐺(𝑠) = = =
𝑅(𝑠) ∆ 1 − 𝐺3 𝐺4 𝐻1 +𝐺2 𝐺3 𝐻2 + 𝐺1 𝐺2 𝐺3 𝐺4 𝐻3

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

Example 4 𝐺4
𝑖
+
Draw the signal flow graph and find 𝑅(𝑠) + 𝑎
𝐺1
𝑏 + 𝑐 𝐺2 𝑑 𝐺3 𝑒 𝑓 𝐺5
𝐶(𝑠)
− +
the overall transfer function of the −
𝑔 𝐻
system. 1
ℎ 𝐻2
𝑎 = 𝑅(𝑠) − ℎ ℎ = 𝐶(𝑠). 𝐻2 𝐺4 1
𝑖
𝑏 = 𝑎𝐺1 𝑒 𝑓 𝐺5 𝐶(𝑠)
𝑅(𝑠) 𝑎 𝑏 𝑐 𝐺2 𝑑 𝐺3 1
1 𝐺1 1
𝑐 =𝑏−𝑔 𝑔 = 𝑑. 𝐻1
𝑑 = 𝑐. 𝐺2 𝐻1
−1
𝑔
−1 ℎ
𝑒 = 𝑑𝐺3
𝐻2
𝑓 = 𝑒+𝑖 𝑖 = 𝑎𝐺4
𝐶(𝑠) = 𝑓. 𝐺5
The signal flow graph has two paths from the input node to the output node.
𝑃1 ⟹ 𝑅(𝑠) − 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝐶(𝑠) Path gains: 𝑃1 = 1 × 𝐺1 × 1 × 𝐺2 × 1 × 𝐺3 × 𝐺5 = 𝐺1 𝐺2 𝐺3 𝐺5
𝑃2 ⟹ 𝑅(𝑠) − 𝑎 − 𝑖 − 𝑓 − 𝐶(𝑠) 𝑃2 = 1 × 𝐺4 × 1 × 𝐺5 = 𝐺4 𝐺5
There are three loops:
𝐿1 ⟹ 𝑐 − 𝑑 − 𝑔 − 𝑐 Loop gains: 𝐿1 = 𝐺2 × 𝐻1 × −1 = −𝐺2 𝐻1
𝐿2 ⟹ 𝑎 − 𝑏 − 𝑐 − 𝑑 − 𝑒 − 𝑓 − 𝐶(𝑠) − ℎ − 𝑎 𝐿2 = 𝐺1 × 1 × 𝐺2 × 1 × 𝐺3 × 𝐺5 × 𝐻2 × −1 = −𝐺1 𝐺2 𝐺3 𝐺5 𝐻2
𝐿3 ⟹ 𝑎 − 𝑖 − 𝑓 − 𝐶(𝑠) − ℎ − 𝑎 𝐿3 = 𝐺4 × 1 × 𝐺5 × 𝐻2 × −1 = −𝐺4 𝐺5 𝐻2
Since there are common nodes, loops 𝐿1 and 𝐿2 are touching each other like loops 𝐿2 and 𝐿3 , and
loops 𝐿1 and 𝐿3 . Loops 𝐿1 and 𝐿3 are nontouching loops.
𝑁

∆= 1 − ∑ 𝐿𝑖 + ∑ 𝐿𝑖 𝐿𝑗 − ∑ 𝐿𝑖 𝐿𝑗 𝐿𝑘 + ⋯ … ..
𝑖=1 𝑖,𝑗 𝑖,𝑗,𝑘
𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔

∆= 1 − (𝐿1 + 𝐿2 + 𝐿3 ) + (𝐿1 𝐿3 ) = 1 + 𝐺2 𝐻1 +𝐺1 𝐺2 𝐺3 𝐺5 𝐻2 + 𝐺4 𝐺5 𝐻2 + 𝐺2 𝐺4 𝐺5 𝐻1 𝐻2

∆1 is the value of ∆ after eliminating all the loops touching the path 1 (All loops are touching
𝑃1 ).
∆1 = 1
∆2 is the value of ∆ after eliminating all the loops touching the path 2 (loops 𝐿2 and 𝐿3 are
touching 𝑃2 ).
∆2 = 1 − 𝐿1 = 1 + 𝐺2 𝐻1
𝐶(𝑠) 𝑃1 ∆1 + 𝑃2 ∆2 𝐺1 𝐺2 𝐺3 𝐺5 + 𝐺4 𝐺5 (1 + 𝐺2 𝐻1 )
𝐺(𝑠) = = =
𝑅(𝑠) ∆ 1 + 𝐺2 𝐻1 +𝐺1 𝐺2 𝐺3 𝐺5 𝐻2 + 𝐺4 𝐺5 𝐻2 + 𝐺2 𝐺4 𝐺5 𝐻1 𝐻2

𝐺4 𝐺5 + 𝐺1 𝐺2 𝐺3 𝐺5 + 𝐺2 𝐺4 𝐺5 𝐻1
𝐺(𝑠) =
1 + 𝐺2 𝐻1 +𝐺4 𝐺5 𝐻2 + 𝐺1 𝐺2 𝐺3 𝐺5 𝐻2 + 𝐺2 𝐺4 𝐺5 𝐻1 𝐻2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6

Exercise
𝑌(𝑠)
Draw the signal flow graph and determie the closed loop transfer function 𝑇(𝑠) = 𝑅(𝑠)
1. 2.

3. 4.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
POSITION CONTROL SERVO SYSTEM
Consider the servo system shown in Figure. In control engineering, a servo system is an automatic
device that uses error-sensing negative feedback to correct the action of the mechanism. The
operation of the system is explained below.
The error measuring device is the pair of potentiometers which convert the input and output
position in to proportional voltages. The desired output position is given as the input to the first
potentiometer. The electric potential of the arm is proportional to the angular position r of the
arm. The output shaft position determines the angular position c of the arm of the output
potentiometer. The difference between the input angular position r and the output angular
position c is the error signal e
𝑒= 𝑟– 𝑐
𝑒𝑟 is the input voltage corresponding to the position r and 𝑒𝑐 is the output voltage corresponding
to output position c.
𝑒𝑟 = 𝐾0 𝑟 𝑎𝑛𝑑 𝑒𝑐 = 𝐾0 𝑐
where 𝐾0 is a proportionality constant.
Therefore, the error voltage is
𝑒𝑣 = 𝑒𝑟 − 𝑒𝑐
This error voltage is applied to the amplifier with gain 𝐾1 . The output voltage of the amplifier
𝐾1 𝑒𝑣 = 𝑒𝑎
is applied to the armature circuit of the dc motor. A fixed voltage is applied to the field winding.
When there is an error, the motor rotates so that the error becomes zero.
If the field current is constant, the torque developed by the motor is proportional to the
armature current 𝑖𝑎
𝑇 = 𝐾2 𝑖𝑎 , where 𝐾2 is the motor constant.

When the armature is rotating, a voltage proportional to the product of the flux and angular
velocity is induced in the armature. For a constant flux (constant field voltage), the induced
voltage (or the back emf)
𝑑𝜃
𝑒𝑏 = 𝐾3 𝑑𝑡 , 𝐾3 is the back emf constant of the motor and 𝜃 is the angular displacement of the
motor shaft. The speed of the armature-controlled dc servomotor is determined by the
armature voltage 𝑒𝑎 . The differential equation for the armature circuit is
𝑑𝑖𝑎
𝑒𝑎 = 𝐿𝑎 + 𝑅𝑎 𝑖𝑎 + 𝑒𝑏
𝑑𝑡
𝑑𝑖𝑎 𝑑𝜃
𝑒𝑎 = 𝐿𝑎 + 𝑅𝑎 𝑖𝑎 + 𝐾3 … … … … … … … … (𝑎)
𝑑𝑡 𝑑𝑡

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
The torque developed by the motor is 𝑇 = 𝐾2 𝑖𝑎
The torque equilibrium (Eqn of rotation)
𝐽0 𝜃̈ = 𝑇 − 𝑏0 𝜃̇ 𝑏0 𝜃̇
𝐽0 𝜃̈ + 𝑏0 𝜃̇ = 𝑇 = 𝐾2 𝑖𝑎 … … … … … . (𝑏) 𝜃̈
𝐽𝑜
Where 𝐽0 is the mass moment of inertia of the motor, load, and gear train
with respect to the motor shaft axis and 𝑏0 is the viscous friction
coefficient of the combination of the motor, load, and gear train referred 𝑇
to the motor axis.
Taking Laplace Transforms of equations (a) and (b)
𝐸𝑎 (𝑠) = 𝐿𝑎 𝑠 𝐼𝑎 (𝑠) + 𝑅𝑎 𝐼𝑎 (𝑠) + 𝐾3 𝑠 𝜃(𝑠) … … … … … … … … (𝑐)
𝐽0 𝑠 2 𝜃(𝑠) + 𝑏0 𝑠𝜃(𝑠) = 𝐾2 𝐼𝑎 (𝑠) … … … … . (𝑑)
( 𝐽0 𝑠2 +𝑏0 𝑠) 𝜃(𝑠)
From (d) 𝐼𝑎 (𝑠) = 𝐾2
which when substituted in to equation (c) gives
( 𝐽0 𝑠 2 + 𝑏0 𝑠) 𝜃(𝑠)
𝐸𝑎 (𝑠) = (𝐿𝑎 𝑠 + 𝑅𝑎 ) + 𝐾3 𝑠 𝜃(𝑠) … … … … … … … … (𝑒)
𝐾2
𝜃(𝑠) 𝐾2 𝐾2
= 2
= … … … … … … … … (𝑓)
𝐸𝑎 (𝑠) ( 𝐽0 𝑠 + 𝑏0 𝑠)(𝐿𝑎 𝑠 + 𝑅𝑎 ) + 𝐾2 𝐾3 𝑠 𝑠{( 𝐽0 𝑠 + 𝑏0 )(𝐿𝑎 𝑠 + 𝑅𝑎 ) + 𝐾2 𝐾3 }
If the gear ratio of the gear train is n (i.e., for each revolution of the motor shaft, the output
shaft rotates n times), then
𝐶(𝑠) = 𝑛 𝜃(𝑠) … … … … . . (𝑔)
We have 𝑒𝑣 = 𝑒𝑟 − 𝑒𝑐 = 𝐾0 𝑟 − 𝐾0 𝑐. i.e.,
𝐸𝑣 (𝑠) = 𝐸𝑟 (𝑠) − 𝐸𝑐 (𝑠) = 𝐾0 {𝑅(𝑠) − 𝐶(𝑠)} = 𝐾0 𝐸(𝑠)…………………..(h)

R(s)
𝐾2
θ (s) C(s)
+ 𝐸𝑣 (s) 𝐸𝑎 (𝑠)
𝐾0 𝐾1 𝑠{( 𝐽0 𝑠+𝑏0 )(𝐿𝑎 𝑠+𝑅𝑎 )+𝐾2 𝐾3 } n
-
𝐾0

If both the potentiometer constants are the same, we can combine them as shown below and
make a unity feed back system.

R(s) + 𝐸 (s) 𝐸𝑎 (𝑠) θ (s) C(s)

𝐸𝑣 (s) 𝐾1
𝐾2
n
𝐾0 𝑠{( 𝐽0 𝑠+𝑏0 )(𝐿𝑎 𝑠+𝑅𝑎 )+𝐾2 𝐾3 }
-

If we look at the forward path, the series blocks can be combined with the transffer function
𝐶(𝑠) 𝐾0 𝐾1 𝐾2 𝑛
𝐺(𝑠) = =
𝐸(𝑠) 𝑠{( 𝐽0 𝑠 + 𝑏0 )(𝐿𝑎 𝑠 + 𝑅𝑎 ) + 𝐾2 𝐾3 }

R(s) + 𝐸 (𝑠) 𝐾0 𝐾1 𝐾2 𝑛 C(s)

𝑠{( 𝐽0 𝑠+𝑏0 )(𝐿𝑎 𝑠+𝑅𝑎 )+𝐾2 𝐾3 }
-

If 𝐿𝑎 is very small
𝐾0 𝐾1 𝐾2 𝑛 𝐾0 𝐾1 𝐾2 𝑛
𝐾0 𝐾1 𝐾2 𝑛 𝑅𝑎 𝑅𝑎
𝐺(𝑠) = = =
𝑠{( 𝐽0 𝑠 + 𝑏0 )𝑅𝑎 + 𝐾2 𝐾3 } 𝐾2 𝐾3 𝐾2 𝐾3
𝑠 {( 𝐽0 𝑠 + 𝑏0 ) + 𝑅 } 𝐽0 𝑠 2 + (𝑏0 + 𝑅 )𝑠
𝑎 𝑎

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
𝐾2 𝐾3 1
If we multiply 𝐽0 and 𝑏0 + 𝑅𝑎
by 𝑛2
, then the inertia and damping are referred to the output
shaft axis.
𝐾2 𝐾3 𝐾0 𝐾1 𝐾2 𝑛
𝐽0 𝑏0 + 𝑅 𝑅𝑎
𝑎
𝐽= 2 , 𝐵= 2
, 𝐾=
𝑛 𝑛 𝑛2
𝐾
𝐺(𝑠) = 2
𝐽𝑠 + 𝐵𝑠

R(s) + 𝐸 (𝑠) C(s)

𝐺(𝑠)
-

The overall transfer function,

𝐾
𝐺 𝐽𝑠 2 + 𝐵𝑠 𝐾
𝑇(𝑠) = = = 2
1 + 𝐺𝐻 1 + 𝐾 𝐽𝑠 + 𝐵𝑠 + 𝐾
𝐽𝑠 2 + 𝐵𝑠
Reference
1. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
TIME RESPONSE ANALYSIS
The transfer function 𝐺(𝑠) of any system will be a ratio with numerator and denominator
appearing as polynomials in s. Some examples are given in table 1.
Table 1. Examples of transfer function
2 2𝑠 𝑠+1 𝑠+2 2𝑠 2 + 1
𝐺(𝑠) = 𝐺(𝑠) = 𝐺(𝑠) = 𝐺(𝑠) = 𝐺(𝑠) =
𝑠+2 𝑠+2 𝑠+5 (𝑠 + 5)(𝑠 + 1) 4𝑠 3 + 𝑠 2 + 5𝑠 + 2
𝑝(𝑠)
In general, the transfer function can be written as 𝐺(𝑠) = 𝐾 where 𝐾 is the gain. The degree
𝑞(𝑠)
of 𝑝(𝑠) will be always less than the degree of 𝑞(𝑠). The equation 𝑞(𝑠) = 0 is known as the
characteristic equation, and the roots of the characteristic equation 𝑞(𝑠) = 0 are known as the
poles. The roots of the equation 𝑝(𝑠) = 0 are known as zeroes of the system. The poles and
zeroes of transfer functions given in Table 1 are shown below in Table 2.
Table 2. Poles and zeroes
Zero No zero 𝑠=0 𝑠 = −1 𝑠 = −2 𝑠 = 0 ± 𝑗0.707
Pole 𝑠 = −2 𝑠 = −2 𝑠 = −5 𝑠 = −1, −5 𝑠 = −0.3842, 0.0671 ± 𝑗1.1389
The complex zeroes and poles appear as conjugate pairs.
The poles and zeroes can be represented in a complex
plane (s is a complex variable 𝜎 + 𝑗𝜔). The poles are
represented by ‘x’ zeroes by ‘o’. The pole-zero map of
the last transfer function in Table 1 is shown in the
figure .

#Exercise 1
Plot in Matlab/scilab, the Pole-Zero map of the transfer function
2𝑠 2 +3𝑠+5
𝐺(𝑠) =
𝑠(4𝑠 3 +𝑠 2 +5𝑠+2)

The poles and zeroes influence the transient and steady state responses of the system. We can
𝜋(𝑠+𝑧 )
always represent the transfer function as 𝐺(𝑠) = 𝐾 𝜋(𝑠+𝑝𝑖 ) where 𝐾 is the gain (constant),
𝑖
where 𝐾 is the gain (constant), 𝑧𝑖 represent the zeroes, 𝑝i represents the poles, and 𝜋 stands for
2𝑠 2 +1
the product. Hence the transfer function 𝐺(𝑠) = can be written as
4𝑠 3 +𝑠 2 +5𝑠+2
(𝑠+𝑗0.707)(𝑠−𝑗0.707)
𝐺(𝑠) = 0.5 .
(𝑠+0.3842)(𝑠−0.0671+𝑗1.1389)(𝑠−0.0671−𝑗1.1389)
The value of 𝐾 = 0.5, zeroes 𝑧𝑖 = ±𝑗0.707 and poles 𝑝𝑖 = −0.3842, 0.0671 ± 𝑗1.1389

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
I order systems
Systems which are governed by first order
𝑋(𝑠) 𝑌(𝑠)
differential equations are known as first order 𝐺(𝑠) =
𝑎
systems. 𝑠+𝑎

𝑌(𝑠) 𝑎
Consider a first order system represented by the transfer function 𝐺(𝑠) = = 𝑠+𝑎
𝑋(𝑠)
There is one pole at 𝑠 = −𝑎, and no zeroes for the system. The transfer function can also be
1
where 𝑇 = 𝑎.
1
written as 𝐺(𝑠) =
1+𝑇𝑠

Step Response
1
For a unit step input, 𝑥(𝑡) = 1, 𝑋(𝑠) = 𝑠
1 𝑎 𝐴 𝐵
𝑌(𝑠) = ( )= + ≪≪≫≫ 𝑎 = 𝐴(𝑠 + 𝑎) + 𝐵𝑠
𝑠 𝑠+𝑎 𝑠 𝑠+𝑎
𝑤ℎ𝑒𝑛 𝑠 = 0 → 𝐴 = 1 ≪≫ 𝑤ℎ𝑒𝑛 𝑠 = −𝑎 → 𝐵 = −1
1 1
𝑌(𝑠) = −
𝑠 𝑠+𝑎
The response 𝑦(𝑡) is obtained by taking the inverse Laplace Transform.
𝑡
𝑦(𝑡) = 1 − 𝑒 −𝑎𝑡 = 1 − 𝑒 −𝑇
The error 𝑒(𝑡) = 𝐷𝑒𝑠𝑖𝑟𝑒𝑑 𝑜𝑢𝑡𝑝𝑢𝑡 − 𝑎𝑐𝑡𝑢𝑎𝑙 𝑜𝑢𝑡𝑝𝑢𝑡=𝑥(𝑡) − 𝑦(𝑡) = 1 − (1 − 𝑒 −𝑎𝑡 ) = 𝑒 −𝑎𝑡
, 𝑦(𝑡) = 1 − 𝑒 −1 = 0.632.
1
When 𝑡 =
𝑎
The value of the time taken by the system to reach 63.2% of the final output is known as the
time constant (𝑇)of the system which represents the speed of response of the system. The
smaller the time constant of the system the faster the response of the system.
The steady state (final) value of the response is obtained by
1 𝑎
Lim 𝑦(𝑡) = 1 or lim 𝑠𝑌(𝑠) = lim 𝑠 𝑠 (𝑠+𝑎) = 1
𝑡→∞ 𝑠→0 𝑠→0
The steady state error is Lim 𝑒(𝑡) = 𝑒 −𝑎𝑡
=0
𝑡→∞

For a particular case of 𝑎 = 2 (𝑖𝑒. , 𝐺(𝑠) = 𝑠+2 = 1+0.5𝑠 ),

2 1

the step response 𝑦(𝑡) is plotted against time 𝑡 and is

shown in figure. The blue line is the input (desired
output) and the red is the actual output. The time
constant of the system is 𝑇 = 0.5 𝑠.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

# Exercise 2
Using Matlab/scilab, plot in one figure the step response of the first order system by taking
𝑎 = 1, 3, 𝑎𝑛𝑑 5 and compare the time constants.
# Exercise 3
𝑏
Obtain the expression for the step response of 𝐺(𝑠) = 𝑠+𝑎. Compare the step responses of
1 2 3
𝐺(𝑠) = 𝑠+2, 𝐺(𝑠) = 𝑠+2, 𝐺(𝑠) = 𝑠+2 by plotting them in the same figure using Matlab/Scilab

Ramp response
1
For a unit ramp, 𝑥(𝑡) = 𝑡, 𝑋(𝑠) = 𝑠2
1 𝑎 𝐴 𝐵 𝐶
𝑌(𝑠) = 2 . = + 2+ ≪≫ 𝑎 = 𝐴 𝑠(𝑠 + 𝑎) + 𝐵(𝑠 + 𝑎) + 𝐶𝑠 2
𝑠 𝑠+𝑎 𝑠 𝑠 𝑠+𝑎
1
𝑠 = 0 → 𝐵 = 1 ≪≫ 𝑠 = −𝑎 → 𝐶 =
𝑎
1
Equating coefficients of 𝑠 2 ≫≫ 0 = 𝐴 + 𝐶 ≫≫ 𝐴 = −𝐶 = − 𝑎
−1 1 1 1 1
𝑌(𝑠) = ( 𝑎 ) 𝑠 + 𝑠2 + (𝑎) 𝑠+𝑎
𝑡
−1 1 1
∴ 𝑦(𝑡) = ( 𝑎 ) + 𝑡 + (𝑎) 𝑒 −𝑎𝑡 = 𝑡 − 𝑎 (1 − 𝑒 −𝑎𝑡 ) = 𝑡 − 𝑇(1 − 𝑒 −𝑇 )
The unit ramp response of a first order system 𝐺(𝑠) =
2
is shown in figure. Red line is the desired output(input)
𝑠+2
and blue line is the actual output.
As 𝑡 takes a large value, the term (1 − 𝑒 −𝑎𝑡 ) → 1. Then
1
𝑦(𝑡) = 𝑡 − . That is, at steady state, the actual output will
𝑎
1
be always less than the desired output by (𝑎).

1 1
The error, 𝑒(𝑡) = 𝑥(𝑡) − 𝑦(𝑡) = 𝑡 − [𝑡 − (1 − 𝑒 −𝑎𝑡 )] = (1 − 𝑒 −𝑎𝑡 )
𝑎 𝑎
1
At steady state (as 𝑡 → ∞), the error 𝑒(𝑡) =
𝑎

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
Impulse Response
For a unit impulse, 𝑥(𝑡) = 𝛿(𝑡), 𝑋(𝑠) = 1
𝑎
𝑌(𝑠) = 1. ( )
𝑠+𝑎
1 𝑡
𝑦(𝑡) = 𝑎𝑒 −𝑎𝑡 = 𝑒 −𝑇
𝑇
As 𝑡 → ∞, the response 𝑦(𝑡) = 0
The unit impulse response of a first order system
2
𝐺(𝑠) = is shown in figure. The input acts only
𝑠+2
for a short duration of time and as time progresses,
the response exponentially decays to zero.

#Exercise 4
Using Matlab/Scilab, plot the impulse responses of the following first order systems in the
same graph.
1
1. 𝐺(𝑠) = 𝑠+1
5
2. 𝐺(𝑠) = 𝑠+5
𝑎
Now for a first order system with transfer function 𝐺(𝑠) = ,
𝑠+𝑎

Ramp response 𝑦(𝑡) = 𝑡 − 1𝑎 (1 − 𝑒−𝑎𝑡 )

𝑑
Step response 𝑦(𝑡) = 1 − 𝑒−𝑎𝑡 = (𝑟𝑎𝑚𝑝 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒)
𝑑𝑡
𝑑
Impulse response 𝑦(𝑡) = 𝑎𝑒−𝑎𝑡 = (𝑠𝑡𝑒𝑝 𝑟𝑒𝑠𝑝𝑜𝑛𝑠𝑒)
𝑑𝑡

#Exercise 5
125
A first order system has a transfer function 𝐺(𝑠) = 100𝑠+125. What is the time constant of
the system? Determine the output of the system after 𝑡 = 0.5 𝑠 when it is subjected to a
unit step, unit ramp, and unit impulse inputs. What is the steady state error for unit ramp
input?

#Exercise 6
A first order system has a unit step response given by 𝑦(𝑡) = 1 − 𝑒 −0.75𝑡 . What will be its
responses to unit ramp and unit impulse inputs after 1 𝑠. What is the transfer function of
the system?

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
II order systems
Consider the second order mechanical system represented by
spring, mass, and dashpot. We had derived the differential
equation governing the dynamics of the system as
𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥 = 𝐹(𝑡)…………………………..(1)
This equation is a second order, ordinary, non-homogeneous,
linear differential equation with constant coefficients. Since it Fig 5. Second
is non-homogeneous equation, the solution will have two parts- a order mechanical system
homogeneous solution (from the corresponding homogeneous
part) and a particular solution (depending upon the nature of the
particular type of input F(t)). The homogeneous solution is known
as the natural response and the particular solution is known as
the forced response.
Consider the corresponding homogeneous equation
𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥 = 0 …………………………..(2)
Since the equation is linear, any exponential function will be a possible solution. Assuming a
solution
𝑥(𝑡) = 𝐴𝑒 𝑝𝑡 ……………………………………….(3) where A is an arbitrary constant which can be evaluated
from the initial conditions and p is a constant but not arbitrary. Substituting (3) in (2), we get
𝑚𝑝2 𝐴𝑒 𝑝𝑡 + 𝑏𝑝𝐴𝑒 𝑝𝑡 + 𝑘𝐴𝑒 𝑝𝑡 = 0
𝐴𝑒 𝑝𝑡 (𝑚𝑝2 + 𝑏𝑝 + 𝑘) = 0,
𝐴 or 𝑒 𝑝𝑡 can not be equal to zero as it gives 𝑥(𝑡) = 0 for all values of time which is a trivial solution.
Hence for (3) to be a solution of (2), the condition is
(𝑚𝑝2 + 𝑏𝑝 + 𝑘) = 0 …………………………..(4)
This is known as the characteristic equation, because the solution of it is going to give some
particular characteristics of the system (which is p in this case).
−𝑏±√𝑏2 −4𝑘𝑚
𝑝= 2𝑚
…………………………………..(5)
We have two roots for p. Hence 𝑥1 (𝑡) = 𝐴1 𝑒 𝑝1 𝑡 and 𝑥2 (𝑡) = 𝐴2 𝑒 𝑝2 𝑡 are possible solutions of (2).
Since the equation (2) is of second order, the solution should have two arbitrary constants.
Because of the fact that the equation is linear, we can apply the principle of superpostion and
any linear combination of the two solutions will be the possible solution.
Hence the homogeneous solution is
𝑥(𝑡) = 𝐴1 𝑒 𝑝1 𝑡 + 𝐴2 𝑒 𝑝2 𝑡 …………………………..(6) if the roots p1 and p2 are distinct (real or complex)
𝑥(𝑡) = (𝐴1 + 𝐴2 𝑡)𝑒 𝑝𝑡 ……………………………….(7) if the roots are real and equal. (Refer your S1 Calculus)
Depending on the value of the discriminant √𝑏 2 − 4𝑘𝑚 of (5), we have three cases

Case(i) 𝑏 2 > 4𝑘𝑚

In this case, the roots are real and distinct. The effect of damping force (b) is more predominant
than the spring force (k) and inertia force(m). This system is called an overdamped system.
Case(ii) 𝑏 2 < 4𝑘𝑚
In this case, the roots are complex conjugates. The damping effect is less predominant in this
system, which is called an underdamped system.

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Case(iii) 𝑏 2 = 4𝑘𝑚
Here, the roots are real and equal. This is called a critically damped system. Of the three cases,
this is the only equality expression. We can find out the critical damping coefficient of the
system as
𝑘 𝑘
𝑏𝑐 = √4𝑘𝑚 = 2𝑚√𝑚 = 2𝑚𝜔𝑛 …………………..(8) where 𝜔𝑛 = √𝑚 is the undamped natural frequency
of the system .
So, for every given system, we can calculate the critical damping coefficient associated with that
system because it involves only the system properties 𝒎 and 𝝎𝒏 (or 𝒎 𝑎nd 𝒌).But the actual
damping of the system can be different.
For example; consider a mass 5 kg which is suspended with a spring of stiffness 1000 N/m. The
critical damping coefficient of the system can be calculated as

𝑏𝑐 = √4𝑘𝑚 = √4 × 1000 × 5 = 141.4 𝑁/𝑚𝑠-1 (𝜔𝑛 = √

1000
= 14.14 𝑟𝑎𝑑/𝑠, 𝑏𝑐 = 2x5x14.14 = 141.4 𝑁/𝑚𝑠-1)
5

Suppose, it is allowed to vibrate in three media viz. air, water, and oil. Then, the actual damping
present in the system will be different in these three cases (depending on their viscosity) which
may not be equal to the critical damping we have calculated above. The actual damping can be
more, equal, or less than the critical damping. Hence, we can always represent the actual damping
present in a given system as a factor of the critical damping coefficient of that system (which
can be evaluated from the system properties)
𝑏 = 𝜁 𝑏𝑐 where 𝜁(zeta) is known as the damping ratio.
𝑎𝑐𝑡𝑢𝑎𝑙 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑦𝑠𝑡𝑒𝑚
𝜁=
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑐𝑜𝑒𝑓𝑓𝑖𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑦𝑠𝑡𝑒𝑚
Therefore, the actual damping coefficient of the system
𝑏 = 𝜁 𝑏𝑐 = 2𝑚𝜁𝜔𝑛 ………………………………(9)
If 𝜁=1, the system is critically damped.
If 𝜁>1, the system is overdamped.
If 𝜁<1, the system is underdamped.

Now, (5) can be written as

−𝑏 √𝑏2 −4𝑘𝑚 −𝑏 𝑏 2 𝑘
𝑝1,2 = ± = ± √( ) − ( ) = −𝜁𝜔𝑛 ± √(𝜁𝜔𝑛 )2 − (𝜔𝑛 )2 = −𝜁𝜔𝑛 ± 𝜔𝑛 √𝜁 2 − 1 …(10)
2𝑚 2𝑚 2𝑚 2𝑚 𝑚

If 𝜁 > 1,the roots p1 and p2 are real and distinct. Then the homogenous solution (6) can be written
as
(−𝜁𝜔𝑛 +𝜔𝑛 √𝜁 2 −1)𝑡 (−𝜁𝜔𝑛 −𝜔𝑛 √𝜁 2 −1)𝑡
𝒙(𝑡) = 𝐴1 𝑒 + 𝐴2 𝑒
𝑥(𝑡) =𝑒 −𝜁𝜔𝑛 𝑡 (𝐴1 𝑒 (𝜔𝑛 √𝜁 ………………………(11)
2 −1)𝑡 2 −1)𝑡
+ 𝐴2 𝑒 (−𝜔𝑛 √𝜁
If 𝜁 = 1,the roots p1 and p2 are real and equal (𝑝1,2 = −𝜔𝑛 ). Then the homogenous solution (7) can
be written as
𝑥(𝑡) = (𝐴1 + 𝐴2 𝑡)𝑒 −𝜔𝑛 𝑡 ………………………………………………….(12)
If 𝜁 < 1,the roots p1 and p2 are complex conjugates.
𝑝1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 √1 − 𝜁 2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑑 where damped frequency 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2. Then the
homogenous solution (6) can be written as

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

𝑥(𝑡) = 𝐴1 𝑒 (−𝜁𝜔𝑛+𝑗𝜔𝑑 )𝑡 + 𝐴2 𝑒 (−𝜁𝜔𝑛−𝑗𝜔𝑑 )𝑡 =𝑒 −𝜁𝜔𝑛𝑡 (𝐴1 𝑒 𝑗𝜔𝑑 𝑡 + 𝐴2 𝑒 −𝑗𝜔𝑑 𝑡 )…………(13a)
=𝑒 −𝜁𝜔𝑛 𝑡 {𝐴1 (𝑐𝑜𝑠𝜔𝑑 𝑡 + 𝑗 𝑠𝑖𝑛𝜔𝑑 𝑡 . ) + 𝐴2 (𝑐𝑜𝑠𝜔𝑑 𝑡 − 𝑗 𝑠𝑖𝑛𝜔𝑑 𝑡 . )}
=𝑒 −𝜁𝜔𝑛 𝑡 {(𝐴1 + 𝐴2 ) 𝑐𝑜𝑠 𝜔𝑑 𝑡 + 𝑗(𝐴1 − 𝐴2 ) 𝑠𝑖𝑛𝜔𝑑 𝑡}
=𝑒 −𝜁𝜔𝑛 𝑡 (𝐶1 𝑐𝑜𝑠 𝜔𝑑 𝑡 + 𝐶2 𝑠𝑖𝑛𝜔𝑑 𝑡)………….(13b) C1 and C2 are the two new arbitrary constants
=𝑒 −𝜁𝜔𝑛 𝑡 (𝑋0 𝑠𝑖𝑛𝜃 𝑐𝑜𝑠 𝜔𝑑 𝑡 + 𝑋0 𝑐𝑜𝑠 𝜃 𝑠𝑖𝑛𝜔𝑑 𝑡) where 𝐶1 = 𝑋0 𝑠𝑖𝑛𝜃 and 𝐶2 = 𝑋0 𝑐𝑜𝑠𝜃 so that
𝐶
𝑋0 = √𝐶12 + 𝐶22 and 𝜃 = tan−1 𝐶1
2

i.e, 𝑥(𝑡) = 𝑋0 𝑒 −𝜁𝜔𝑛 𝑡

sin(𝜔𝑑 𝑡 + 𝜃)………………(14) where 𝑋0 and 𝜃 are the new arbitrary constants.
The response of an underdamped system can be represented by any one of the three equations
(13a), (13b), or (14).

Now we have obtained the natural response (homogeneous solution) of a second order system
if the system is overdamped (𝜁 > 1)
2 −1)𝑡 2 −1)𝑡
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛𝑡 (𝐴1 𝑒 (𝜔𝑛√𝜁 + 𝐴2 𝑒 (−𝜔𝑛√𝜁
𝑥(𝑡) = (𝐴1 + 𝐴2 𝑡)𝑒 −𝜔𝑛𝑡 if the system is critically damped (𝜁 = 1)
𝑥(𝑡) = 𝑋0 𝑒 −𝜁𝜔𝑛𝑡 sin(𝜔𝑑 𝑡 + 𝜃) if the system is underdamped 𝜁 < 1)

The values of the arbitrary constants have to be determined from the initial conditions.
If we look at the first expression above for the overdamped system, irrespective of the values
of the arbitrary constants as 𝑡 → ∞,
𝒆−𝜻𝝎𝒏𝒕 → 0, 𝟐 −𝟏)𝒕
𝒆(𝝎𝒏 √𝜻 → ∞, 𝒆(−𝝎𝒏 √𝜻
𝟐 −𝟏)𝒕
→0
hence 𝑥(𝑡) → 0

In second expression of critical damped system as 𝑡 → ∞,

(𝐴1 + 𝐴2 𝑡) → ∞, 𝑒 −𝜔𝑛 𝑡 → 0, hence 𝑥(𝑡) → 0

In both the above cases, the system will not oscillate. Once the vibration is initiated by giving
some initial conditions, the system comes to equilibrium without oscillation. The time taken by
the system to come to equilibrium is smaller for critically damped system than an overdamped.

For the under damped system, the response is a sine curve with amplitude 𝑿𝟎 𝒆−𝜻𝝎𝒏 𝒕 and circular
frequency 𝜔𝑑 . The period of oscillation 𝑇 = 𝜔
2𝜋
. The amplitude of oscillation 𝑿𝟎 𝒆−𝜻𝝎𝒏𝒕 decays
𝑑
exponentially to zero as 𝑡 → ∞.
Only an underdamped sytem will vibrate/oscillate when it is disturbed from equilibrium.

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

The homogeneous solutions for the three different cases are plotted in the figure below. The
arbitrary constants are assumed as 𝐴1 = 1, 𝐴2 = 0, 𝑋0 = 1, and 𝜃 = 900 and 𝜔𝑛 = 5

period

We can see that irrespctive of the values of damping factor, the natural response vanishes after
some time.

The particular solution is obtained by considering a particular input 𝐹(𝑡). The response that we
get is known as the forced response which will exist as long as 𝐹(𝑡) exists. We will consider the
forced responses for step and impulse inputs later. The complete solution (total response) will
be the sum of natural response(homogeneous solution) and forced response (particular solution)

#Exercise 1
Find the values of the arbitrary constants in the expressions for the homogeneous responses
of overdamped (𝜁 = 1.7), critically damped (𝜁 = 1), and underdamped (𝜁 = 0.7) with undamped
natural frequency (𝜔𝑛 = 5 𝑟𝑎𝑑/𝑠) for the following set of initial conditions
(i) 𝑥(0) = 1, 𝑥̇ (0) = 0 (ii) 𝑥(0) = 0, 𝑥̇ (0) = 1 (iii) 𝑥(0) = 1, 𝑥̇ (0) = 1
Plot the responses using Matlab/Scilab/Excel
(Hint: For substituting the velocity initial condition, you need to differentiate the expression
for x(t) with respect to time.)

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Let us again consider the roots of the characteristic equation given in (10)
𝑝1,2 = −𝜁𝜔𝑛 ± 𝜔𝑛 √𝜁 2 − 1
Let us vary the value of 𝜁 keeping 𝜔𝑛 as constant (= 5 𝑟𝑎𝑑/𝑠)
𝜁 −𝜁𝜔𝑛 𝜔𝑛 √𝜁 2 − 1 𝑝1 = −𝜁𝜔𝑛 + 𝜔𝑛 √𝜁 2 − 1 𝑝2 = −𝜁𝜔𝑛 − 𝜔𝑛 √𝜁 2 − 1
0 0 j5.00 +j5.0 -j5.0
0.2 -1.0 j4.89 -1.0+j4.89 -1.0-j4.89
0.4 -2.0 j4.58 -2.0+j4.58 -2.0-j4.58
0.6 -3.0 j4.00 -3.0+j4.00 -3.0-j4.00
0.8 -4.0 j3.00 -4.0+j3.00 -4.0-j3.00
1.0 -5.0 0 -5.0 -5.0
1.2 -6.0 3.31 -2.69 -9.31
1.4 -7.0 4.89 -2.11 -11.89
1.6 -8.0 6.25 -1.75 -14.25
Look at the values of 𝑝1 and 𝑝2 in the above table. It gives the values of the roots of the
characteristic equation when 𝜁 is varied from 0. The values of the roots are plotted in complex
plane in figure below. The horizontal axis is the real axis and vertical one is the imaginary axis.
When 𝜁 = 0 (undamped), the roots are ±𝑗5 which are on the imaginary axis. As we increase the
value of 𝜁 𝑡𝑜 0.2 (underdamped), the roots are complex conjugates with negative real part and
shift to the left side of the imaginary axis. The values of root p1 are marked in red dots and p2
are shown in blue dots. When 𝜁 =1 (critically damped), both the roots are equal at −5. When 𝜁
increases above 1(overdamped), both the roots are negative real numbers. One moves to a large
negative value and the other approaching to zero. Please note that the value of 𝑝1 will never be
equal to zero however high the value of 𝜁 may be, because the negative part −𝜁𝜔𝑛 will be always
greater than the positive part 𝜔𝑛 √𝜁 2 − 1 . If we continuously change the parameter 𝜁 and plot
the roots of the charcteristic equation, we will get the locus of the roots of the characteristic
equation which is known as the ROOT LOCUS. We will discuss the root locus technique in detail
later.
6

0
-16 -14 -12 -10 -8 -6 -4 -2 0

-2

-4

-6

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

If you have understood this far, let us approach this in another way.
The governing equation of a second order dynamic mechanical system is
𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥 = 𝐹(𝑡).
Taking Laplace transform on both sides
𝑚{𝑠 2 𝑋(𝑠) − 𝑠𝑥(0) − 𝑥̇ (0)} + 𝑏{𝑠𝑋(𝑠) − 𝑥(0)} + 𝑘𝑋(𝑠) = 𝐹(𝑠)
(𝑚𝑠 2 + 𝑏𝑠 + 𝑘)𝑋(𝑠) − {(𝑚𝑠 + 𝑏)𝑥(0) + 𝑚𝑥̇ (0)} = 𝐹(𝑠) ……………………….(15)
(𝑚𝑠 2 + 𝑏𝑠 + 𝑘)𝑋(𝑠) = {(𝑚𝑠 + 𝑏)𝑥(0) + 𝑚𝑥̇ (0)} + 𝐹(𝑠)
{(𝑚𝑠+𝑏)𝑥(0)+𝑚𝑥̇ (0)}+𝐹(𝑠)
𝑋(𝑠) =
𝑚𝑠2 +𝑏𝑠+𝑘
{(𝑚𝑠+𝑏)𝑥(0)+𝑚𝑥̇ (0)} 𝐹(𝑠)
= 𝑚𝑠2 +𝑏𝑠+𝑘
+ 𝑚𝑠2 +𝑏𝑠+𝑘 …………………………………(16)
The first term on RHS will give the natural response (homogeneous solution) and the second term
will give the forced response (particular solution)

If we know the initial conditions and 𝐹(𝑠) {which is the Laplace Transform of the input 𝐹(𝑡) }, we
can find out the response 𝑥(𝑡) by taking the inverse Laplace Transform of (16)

Assume that no input acts on the system [𝐹(𝑠) = 0] and initial conditions 𝑥(0) = 1, 𝑎𝑛𝑑 𝑥̇ (0) = 0
{(𝑚𝑠 + 𝑏)𝑥(0) + 𝑚𝑥̇ (0)}
𝑋(𝑠) = … … … … … … … … … … . (17)
𝑚𝑠 2 + 𝑏𝑠 + 𝑘
𝑏
{(𝑚𝑠+𝑏).1+𝑚.0} 𝑠+ 𝑠+2𝜁𝜔𝑛
= (Refer Eq.(8) and (9))
𝑚
= 𝑏 𝑘 =
𝑚𝑠 2 +𝑏𝑠+𝑘 𝑠2 + 𝑠+ 𝑠2 +2𝜁𝜔 𝑛 𝑠+𝜔𝑛
2
𝑚 𝑚

𝑥(𝑡) can be calculated by taking the invers transform. Let us write 𝑋(𝑠) as partial fraction
𝑠+2𝜁𝜔𝑛 𝑠+2𝜁𝜔𝑛
𝑋(𝑠) = (𝑠+𝜁𝜔 2 2 2
= ……………… (18)
𝑛 ) −(𝜁𝜔𝑛 ) +𝜔𝑛 [𝑠+𝜁𝜔𝑛 ]2 −[𝜔𝑛 √𝜁 2 −1]2

Let us assume that the system is critically damped (𝜁 = 1), The equation (18) reduces to
𝑠 + 2𝜔 𝑛 𝐴 𝐵
𝑋(𝑠) = 2 =𝑠+𝜔 + 2 (𝑃𝑎𝑟𝑡𝑖𝑎𝑙 𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛)
(𝑠 + 𝜔𝑛 ) 𝑛 (𝑠 + 𝜔𝑛 )
𝑠 + 2𝜔𝑛 = 𝐴(𝑠 + 𝜔𝑛 ) + 𝐵………………………………………………….(19a)
By letting 𝑠 = −𝜔𝑛 , we get 𝐵 = 𝜔𝑛
Equating constants on both sides of (19a), we can write 2𝜔𝑛 = 𝐴𝜔𝑛 + 𝐵
We get 𝐴 = 1
1 𝜔𝑛
𝑋(𝑠) = +
𝑠 + 𝜔𝑛 (𝑠 + 𝜔𝑛 )2
Taking inverse Laplace Transform
𝑥(𝑡) = 𝑒 −𝜔𝑛 𝑡 + 𝜔𝑛 (𝑒 −𝜔𝑛 𝑡 ) 𝑡 = 𝑒 −𝜔𝑛 𝑡 (1 + 𝜔𝑛 𝑡) … … … … … … … … . (19𝑏)
This is exactly same as (12) with 𝐴1 = 1, 𝐴2 = 𝜔𝑛
(You will be getting the same expression as (19b) when you do #Exercise 1 for the critically
damped system with the same initial conditions. Verify!!!!!!)

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Let us consider an overdamped system (𝜁 > 1). The equation (18) is written as (for simplification)
𝑠+2𝑎
𝑋(𝑠) = where 𝑎 = 𝜁𝜔𝑛 and 𝑏 = 𝜔𝑛 √𝜁 2 − 1
(𝑠+𝑎)2 −𝑏2
𝑠+2𝑎 𝐴 𝐵
𝑋(𝑠) = {(𝑠+𝑎)+𝑏} {(𝑠+𝑎)−𝑏}
= 𝑠+𝑎+𝑏 + 𝑠+𝑎−𝑏

𝑠 + 2𝑎 = 𝐴(𝑠 + 𝑎 − 𝑏) + 𝐵(𝑠 + 𝑎 + 𝑏) . Let us put 𝑠 = 𝑏 − 𝑎 and 𝑠 = −𝑎 − 𝑏 to evaluate 𝐴 and 𝐵

𝑎+𝑏
𝑏 − 𝑎 + 2𝑎 = 𝐴(𝑏 − 𝑎 + 𝑎 − 𝑏) + 𝐵(𝑏 − 𝑎 + 𝑎 + 𝑏) ≫ 𝐵= 2𝑏
𝑎−𝑏
−𝑎 − 𝑏 + 2𝑎 = 𝐴(−𝑎 − 𝑏 + 𝑎 − 𝑏) + 𝐵(−𝑎 − 𝑏 + 𝑎 + 𝑏) ≫ 𝐴=− 2𝑏
𝑎−𝑏 𝑎+𝑏
− 1 𝑎+𝑏 𝑎−𝑏
𝑋(𝑠) = 2𝑏 + 2𝑏 = { − }
𝑠 + 𝑎 + 𝑏 𝑠 + 𝑎 − 𝑏 2𝑏 𝑠 + (𝑎 − 𝑏) 𝑠 + (𝑎 + 𝑏)
Taking inverse Laplace Transform
1
𝑥(𝑡) = {(𝑎 + 𝑏) 𝑒 −(𝑎−𝑏)𝑡 − (𝑎 − 𝑏) 𝑒 −(𝑎+𝑏)𝑡 }
2𝑏
1 −(𝜁𝜔𝑛 −𝜔𝑛 √𝜁 2 −1))𝑡 −(𝜁𝜔𝑛 +𝜔𝑛 √𝜁 2 −1))𝑡
𝑥(𝑡) = {(𝜁𝜔𝑛 + 𝜔𝑛 √𝜁 2 − 1) 𝑒 − (𝜁𝜔𝑛 − 𝜔𝑛 √𝜁 2 − 1) 𝑒 }
2𝜔𝑛 √𝜁 2 − 1
𝑒 −(𝜁𝜔𝑛 )𝑡 (𝜔𝑛 √𝜁 2 −1))𝑡 −(𝜔𝑛 √𝜁 2 −1))𝑡
𝑥(𝑡) = {(𝜁𝜔𝑛 + 𝜔𝑛 √𝜁 2 − 1) 𝑒 − (𝜁𝜔𝑛 − 𝜔𝑛 √𝜁 2 − 1) 𝑒 }
2𝜔𝑛 √𝜁 2 − 1
𝜁+√𝜁 2 −1) √𝜁 2 −1)𝑡 𝜁−√𝜁 2 −1) √𝜁 2 −1)𝑡
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛𝑡 { 𝑒 (𝜔𝑛 − 𝑒 −(𝜔𝑛 } …………………..(20)
2√𝜁 2 −1 2√𝜁 2 −1

Compare this with equation (11) and your #Exercise 1 problem of overdamped system response
with same initial conditions. Verify the value of A1 and A2 !!!!!

Consider now an underdamped system with 𝜁 < 1, the equation (18) is written as
𝑠+2𝜁𝜔𝑛 𝑠+2𝜁𝜔𝑛 (𝑠+𝜁𝜔𝑛 )+𝜁𝜔𝑛
𝑋(𝑠) = = 2 = [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2
[𝑠+𝜁𝜔𝑛 ]2 −[𝜔𝑛 √𝜁 2 −1]2 [𝑠+𝜁𝜔𝑛 ]2 −[𝑗𝜔𝑛 √1−𝜁 2 ]

Recall that 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2 and j2=(-1)

(𝑠+𝜁𝜔𝑛 ) 𝜁𝜔𝑛 (𝑠+𝜁𝜔𝑛 ) 𝜁𝜔𝑛 𝜔𝑑
𝑋(𝑠) = + = + ( )
[𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2 𝜔𝑑 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2

Taking inverse Laplace Transform (revisit the laplace transforms of sin wt and cos wt)
𝜁𝜔𝑛 𝜁
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛𝑡 cos 𝜔𝑑 𝑡 + (
𝜔𝑑
) 𝑒 −𝜁 𝜔𝑛𝑡 sin 𝜔𝑑 𝑡 = 𝑒 −𝜁𝜔𝑛𝑡 (cos 𝜔𝑑 𝑡 + ( ) sin 𝜔𝑑 𝑡)
√1−𝜁 2
𝑒 −𝜁𝜔𝑛 𝑡 𝑒 −𝜁𝜔𝑛𝑡
= (√1 − 𝜁 2 cos 𝜔𝑑 𝑡 + 𝜁 sin 𝜔𝑑 𝑡 ) = (𝑋0 sin 𝜃 cos 𝜔𝑑 𝑡 + 𝑋0 cos 𝜃 sin 𝜔𝑑 𝑡)
√1−𝜁 2 √1−𝜁 2
√1−𝜁2
where 𝑋0 sin 𝜃 = √1 − 𝜁2 𝑎𝑛𝑑 𝑋0 cos 𝜃 = 𝜁 so that 𝑋0 2=(1 − 𝜁2 ) + 𝜁2 = 1 and tan 𝜃 =
𝜁
1
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛 𝑡
sin(𝜔𝑑 𝑡 + 𝜃) …………………………………(21)
√1−𝜁2

Compare this with equation (14) and your #Exercise 1 problem of underdamped system response
with same initial conditions. Verify the value of 𝑋0 and 𝜃 !!!!!
#Exercise 2.
Starting from Equation (17), evaluate the natural responses of over, under, and critically
damped second order systems for the following initial conditions. Verify your answer with
#Ex. 1. (i) 𝑥(0) = 0, 𝑥̇ (0) = 1 (ii) 𝑥(0) = 1, 𝑥̇ (0) = 1

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

If we set all initial conditions equal to zero, (16) reduces to
𝐹(𝑠)
𝑋(𝑠) = .
𝑚𝑠2 +𝑏𝑠+𝑘
The transfer function of the system is given by (Recall the definition of Transfer Function)
𝑘
𝑋(𝑠) 1 1 1 𝑘 1 1 𝑚
𝐺(𝑠) = = 𝑚𝑠2 +𝑏𝑠+𝑘 = 𝑏 𝑘 = (𝑘) (𝑚) 𝑏 𝑘 = (𝑘) 𝑏 𝑘
𝑌(𝑠) 𝑚(𝑠2 + 𝑠+ ) (𝑠2 + 𝑠+ ) (𝑠2 + 𝑠+ )
𝑚 𝑚 𝑚 𝑚 𝑚 𝑚
1 𝜔𝑛 2
𝐺(𝑠) = ( ) 2
𝑘 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛 2
𝑝(𝑠)
This is now of the form 𝐺(𝑠) = 𝐾 𝑞(𝑠).
𝜔𝑛 2
The standard form of a second order system is considered as 𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
2

The characteristic equation is 𝑞(𝑠) = 0, and the roots of the characteristic equation are called
poles of the system
𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛 2 = 0 ……………………………………………………..(22)
−2𝜁𝜔𝑛 ±√(2𝜁𝜔𝑛 )2 −4𝜔𝑛 2
𝑠= 2
= −𝜁𝜔𝑛 ± 𝜔𝑛 √𝜁 2 − 1 …………….(23) [Same as equation (10)]

Example 1
25
The transfer function of a second order system is 𝐺(𝑠) = Find the natural frequency
𝑠2 +5𝑠+25
and damping ratio of the system.
𝜔𝑛 2
Comparing the given transfer function with the standard form
𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
2

𝜔𝑛 2 = 25, and 2𝜁𝜔𝑛 = 5

5
Hence 𝜔𝑛 = 5 rad/s, and 𝜁 = = 0.5
2𝜔𝑛

#Exercise 3
Given below are the natural frequency and damping ratio of a second order system respectively.
What is system transfer function? Where is the location of the poles?
i) 3 rad/s and 0.25 ii) 5 rad/s and 1 iii) 2 rad/s and 1.25 iv) 4 rad/s and 0

#Exercise 4
Find the natural frequency and damping ratio of the second oerder systems given below
24 15
i) 𝐺(𝑠) = 2 ii) 𝐺(𝑠) = (𝑠+3)(2𝑠+5)
3𝑠 +8𝑠+24
#Exercise 5
The poles of three second order systems are given below. Obtain their transfer functions,
damping ratios and natural frequencies.
i) -1, -5 ii) -3,-3 iii) -2±j5 iv) ±j5

Having learned the natural responses of a second order system both by solving the differential
equation in the conventional way and with Laplace transform, let us look at the forced responses
using the Laplace Transform technique

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Impulse Response of Second Order Systems
𝜔𝑛 2
Let us consider a second order system represented by the transfer function 𝐺(𝑠) =
𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
2

𝑋(𝑠) 𝜔 2
𝐺(𝑠) = 𝐹(𝑠) = 𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔 2
𝑛 𝑛
𝜔 2
𝑋(𝑠) = 𝐹(𝑠) 𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔 2
𝑛 𝑛

For a unit impulse 𝐹(𝑡) = 𝛿(𝑡), 𝐹(𝑠) = 1

𝜔𝑛 2 𝜔𝑛 2 𝜔𝑛 2
𝑋(𝑠) = = = ………………………..(24)
𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2 (𝒔+𝜻𝝎𝒏 )𝟐 −(𝜻𝝎𝒏 )𝟐 +𝝎𝒏 𝟐 [𝒔+𝜻𝝎𝒏 ] −[𝝎𝒏 √𝜻𝟐 −𝟏]𝟐
𝟐

Critically damped system (𝜁 = 1)

The equation (24) reduces as follows.
𝜔𝑛 2 𝐴 𝐵
𝑋(𝑠) = 2 = + 2 (𝑃𝑎𝑟𝑡𝑖𝑎𝑙 𝐹𝑟𝑎𝑐𝑡𝑖𝑜𝑛)
(𝑠 + 𝜔𝑛 ) 𝑠 + 𝜔 𝑛 (𝑠 + 𝜔𝑛 )
𝜔𝑛 2 = 𝐴(𝑠 + 𝜔𝑛 ) + 𝐵………………………………………………….(24a)
By letting 𝑠 = −𝜔𝑛 , we get 𝐵 = 𝜔𝑛 2
Equating constants on both sides of (24a), we can write 𝜔𝑛 2 = 𝐴𝜔𝑛 + 𝐵
We get 𝐴 = 0
𝜔𝑛 2
𝑋(𝑠) = 2
(𝑠 + 𝜔𝑛 )
Taking inverse Laplace Transform
𝑥(𝑡) = 𝜔𝑛 2 𝑡 𝑒 −𝜔𝑛 𝑡 ……………………………………………………(25)
Compare this equation with the natural response obtained for a critically damped second order
system with the initial conditions 𝑥(0) = 0, 𝑥̇ (0) = 1 (#Exercise 1 and 2)

Overdamped system (𝜁 > 1)

The equation (24) is written as (for simplification)
𝜔 2
𝑋(𝑠) = (𝑠+𝑎)𝑛2 −𝑏2 where 𝑎 = 𝜁𝜔𝑛 and 𝑏 = 𝜔𝑛 √𝜁 2 − 1
𝜔 2 𝐴 𝐵
𝑋(𝑠) = {(𝑠+𝑎)+𝑏}𝑛{(𝑠+𝑎)−𝑏} = 𝑠+𝑎+𝑏 + 𝑠+𝑎−𝑏

𝜔𝑛 2 = 𝐴(𝑠 + 𝑎 − 𝑏) + 𝐵(𝑠 + 𝑎 + 𝑏) . Let us put 𝑠 = 𝑏 − 𝑎 and 𝑠 = −𝑎 − 𝑏 to evaluate 𝐴 and 𝐵

𝜔𝑛 2
𝜔𝑛 2 = 𝐴(𝑏 − 𝑎 + 𝑎 − 𝑏) + 𝐵(𝑏 − 𝑎 + 𝑎 + 𝑏) ≫ 𝐵= 2𝑏
𝜔𝑛 2
𝜔𝑛 2 = 𝐴(−𝑎 − 𝑏 + 𝑎 − 𝑏) + 𝐵(−𝑎 − 𝑏 + 𝑎 + 𝑏) ≫ 𝐴=−
2𝑏
2 2
−𝜔2𝑏
𝑛 𝜔𝑛 2
𝑋(𝑠) =
𝑠+𝑎+𝑏
+ 2𝑏
𝑠+𝑎−𝑏
= 𝜔2𝑏𝑛 {𝑠+(𝑎−𝑏)
1 1
− 𝑠+(𝑎+𝑏)}
Taking inverse Laplace Transform
𝜔𝑛 2 −(𝑎−𝑏)𝑡
𝑥(𝑡) = {𝑒 − 𝑒 −(𝑎+𝑏)𝑡 }
2𝑏
𝜔𝑛 2 −(𝜁𝜔𝑛 −𝜔𝑛 √𝜁 2 −1))𝑡 −(𝜁𝜔𝑛 +𝜔𝑛 √𝜁 2 −1))𝑡
𝑥(𝑡) = {𝑒 − 𝑒 }
2
2𝜔𝑛 √𝜁 − 1
𝜔𝑛 𝑒 −(𝜁𝜔𝑛)𝑡 (𝜔𝑛 √𝜁 2 −1))𝑡 −(𝜔𝑛 √𝜁 2 −1))𝑡
𝑥(𝑡) = {𝑒 − 𝑒 }
2√𝜁 2 −1
𝜔𝑛 (𝜔𝑛 √𝜁 2 −1)𝑡 𝜔𝑛 −(𝜔𝑛 √𝜁 2 −1)𝑡
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛 𝑡 { 𝑒 − 𝑒 } …………………..(26)
2√𝜁 2 −1 2√𝜁 2 −1

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Compare this equation with the natural response obtained for a overdamped second order system
with the initial conditions 𝑥(0) = 0, 𝑥̇ (0) = 1 (#Exercise 1 and 2)
Underdamped system (𝜁 < 1)
The equation (24) is written as
𝝎𝒏 𝟐 𝝎𝒏 𝟐 𝝎𝒏 𝟐
𝑋(𝑠) = = 2 =
[𝑠+𝜁𝜔𝑛 ]2 −[𝜔𝑛 √𝜁 2 −1]2 [𝑠+𝜁𝜔𝑛 ]2 −[𝑗𝜔𝑛 √1−𝜁 2 ] [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2

Recall that 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2 and j2=(-1)

𝝎 𝟐 𝜔𝑑
𝑋(𝑠) = ( 𝜔𝒏 ) [𝑠+𝜁𝜔 2 +[𝜔 2
𝑑 𝑛] 𝑑]

Taking inverse Laplace Transform (revisit the laplace transforms of sin wt)
𝝎 𝟐
𝑥(𝑡) = ( 𝜔𝒏 ) 𝑒 −𝜁𝜔𝑛𝑡 sin 𝜔𝑑 𝑡
𝑑
𝝎𝒏
𝑥(𝑡) = 𝑒 −𝜁𝜔𝑛𝑡 sin 𝜔𝑑 𝑡 …………………………………(27)
√1−𝜁2

Compare this equation with the natural response obtained for an underdamped second order
system with the initial conditions 𝑥(0) = 0, 𝑥̇ (0) = 1 (#Exercise 1 and 2)
The impulse responses of a second order
system with 𝜔𝑛 =5 rad/s and 𝜁 values of 0.5, 1,
and 1.5 are shown in Figure. The behavior of
the system when an impulse is given as the
input (forced response/particular solution) is
similar to its natural response (homogeneous
solution) with the initial conditions 𝑥(0) = 0,
𝑥̇ (0) = 1. This is understandable because the
impulse acts only for a very short duration
initially and thereafter the system behaves
naturally as there is no input/forcing term.

#Exercise 6
Given below are the natural frequency and damping ratio of a second order system respectively.
Plot the impulse responses using Matlab/Scilab.
ii) 3 rad/s and 0.25 ii) 5 rad/s and 1 iii) 2 rad/s and 1.25 iv) 4 rad/s and 0

#Exercise 7
The transfer functions of second order system are given below. Plot the impulse responses
using Matlab/Scilab
24 15
i) 𝐺(𝑠) = ii) 𝐺(𝑠) = (𝑠+3)(2𝑠+5)
3𝑠2 +8𝑠+24
#Exercise 8
The poles of three second order systems are given below. Plot the impulse responses using
Matlab/Scilab
i) -1, -5 ii) -3,-3 iii) -2±j5 iv) ±j5

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Step Response of Second Order Systems
𝜔𝑛 2
Consider a second order sytem with transfer function 𝐺(𝑠) =
𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
𝑋(𝑠) 𝜔𝑛 2
𝐺(𝑠) = =
𝐹(𝑠) 𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
𝜔 2
𝑋(𝑠) = 𝐹(𝑠) 𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔 2
𝑛 𝑛

For a unit step input 𝐹(𝑡) = 1, 𝐹(𝑠) = 1/𝑠

1 𝜔𝑛 2 1 𝜔𝑛 2 1 𝜔𝑛 2
𝑋(𝑠) = ( ) =( ) =( ) ………….(27a)
𝑠 𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
2 𝑠 (𝒔+𝜻𝝎𝒏 ) −(𝜻𝝎𝒏 )𝟐 +𝝎𝒏 𝟐
𝟐 𝑠 [𝒔+𝜻𝝎𝒏 ] −[𝝎𝒏 √𝜻𝟐 −𝟏]𝟐
𝟐

Critically damped system (𝜁 = 1)

The equation (27a) becomes
1 𝜔𝑛 2 𝐴 𝐵 𝐶
𝑋(𝑠) = ( ) = + + ………………………………(28)
𝑠 (𝑠+𝜔𝑛 )2 𝑠 𝑠+𝜔𝑛 (𝑠+𝜔𝑛 )2

𝜔𝑛 2 = 𝐴(𝑠 + 𝜔𝑛 )2 + 𝐵𝑠(𝑠 + 𝜔𝑛 ) + 𝐶𝑠
Substituting 𝑠 = 0, we wet 𝐴 = 1
When 𝑠 = −𝜔𝑛 ≫≫ 𝑪 = −𝝎𝒏
Equating coefficients of s2, 0 = 𝐴 + 𝐵 ≫≫ 𝐵 = −1
1 1 𝝎𝒏
Therefore 𝑋(𝑠) = 𝑠
− 𝑠+𝝎 − 2
𝒏 (𝑠+𝝎𝒏 )

Taking inverse Laplace Transform

𝑥(𝑡) = 1 − 𝑒 −𝜔𝑛 𝑡 − 𝜔𝑛 𝑡𝑒 −𝜔𝑛 𝑡
𝑥(𝑡) = 1 − (1 + 𝜔𝑛 𝑡)𝑒 −𝜔𝑛 𝑡 ……………………………………………...(29)
The response curve starts from origin (when 𝑡 = 0) and reaches 1 as t→ ∞
#Exercise 9
Plot equation (29) in Matlab/Scilab/Excel for different values of 𝜔𝑛 =1, 3, and 5

Overdamped system (𝜁 > 1)

We have equation (27)
1 𝜔𝑛 2
𝑋(𝑠) = (𝑠 )
[𝑠+𝜁𝜔𝑛 ]2 −[𝜔𝑛 √𝜁 2 −1]2
1 𝜔 2
𝑋(𝑠) = (𝑠 ) (𝑠+𝑎)𝑛2 −𝑏2 where 𝑎 = 𝜁𝜔𝑛 and 𝑏 = 𝜔𝑛 √𝜁 2 − 1
𝜔 2 𝐴 𝐵 𝐶
𝑋(𝑠) = 𝑠 {(𝑠+𝑎)+𝑏}𝑛 {(𝑠+𝑎)−𝑏} = 𝑠+𝑎+𝑏 + 𝑠+𝑎−𝑏 + 𝑠

𝜔𝑛 2 = 𝐴 𝑠(𝑠 + 𝑎 − 𝑏) + 𝐵 𝑠(𝑠 + 𝑎 + 𝑏) + 𝐶(𝑠 + 𝑎 + 𝑏)(𝑠 + 𝑎 − 𝑏) .

Let us put 𝑠 = 𝑏 − 𝑎, 𝑠 = −𝑎 − 𝑏, and 𝑠 = 0 to evaluate 𝐵, 𝐴, and 𝐶 respectively.
𝜔 2
𝜔𝑛 2 = 𝐵(𝑏 − 𝑎)(𝑏 − 𝑎 + 𝑎 + 𝑏) 𝑛
≫ 𝐵 = − 2𝑏(𝑎−𝑏)
𝜔 2
𝜔𝑛 2 = 𝐴(−𝑎 − 𝑏)(−𝑎 − 𝑏 + 𝑎 − 𝑏) 𝑛
≫ 𝐴 = 2𝑏(𝑎+𝑏)
𝝎𝒏 𝟐
𝜔𝑛 2 = 𝐶(𝑎 + 𝑏)(𝑎 − 𝑏) ≫ 𝐶= =1 ( note that 𝑎2 − 𝑏 2 = 𝝎𝒏 𝟐 )
𝑎2 −𝑏2
𝑛 𝜔 2 1 𝑛 𝜔 2 1 1
𝑋(𝑠) = (2𝑏(𝑎+𝑏) ) 𝑠+𝑎+𝑏 − (2𝑏(𝑎−𝑏) ) 𝑠+𝑎−𝑏 + 𝑠

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Taking inverse Laplace Transform
𝑒 −(𝑎+𝑏)𝑡 𝑒 −(𝑎−𝑏)𝑡
𝑥(𝑡) = {𝜔𝑛 2 − 𝜔𝑛 2 + 1}
2𝑏(𝑎+𝑏) 2𝑏(𝑎−𝑏)
𝜔𝑛 2 (𝑎−𝑏)𝑒 −(𝑎+𝑏)𝑡 −𝜔𝑛 2 (𝑎+𝑏)𝑒 −(𝑎−𝑏)𝑡
=1+ 2𝑏(𝑎 2 −𝑏2 )
𝜔𝑛 2 (𝑎−𝑏)𝑒 −(𝑎+𝑏)𝑡 −𝜔𝑛 2 (𝑎+𝑏)𝑒 −(𝑎−𝑏)𝑡
=1+
2𝑏𝜔𝑛 2
(𝑎−𝑏)𝑒 −(𝑎+𝑏)𝑡 −(𝑎+𝑏)𝑒 −(𝑎−𝑏)𝑡 𝑒 −𝑎𝑡
=1+ 2𝑏
= 1+( 2𝑏
) {(𝑎 − 𝑏)𝑒 −𝑏𝑡 − (𝑎 + 𝑏)𝑒 𝑏𝑡 }
𝑒 −𝑎𝑡
=1−( 2𝑏
) {(𝑎 + 𝑏)𝑒 𝑏𝑡 − (𝑎 − 𝑏)𝑒 −𝑏𝑡 }

𝑒 −𝜁𝜔𝑛 𝑡 2 −1 2 −1
𝑥(𝑡) = 1 − [(𝜁𝜔𝑛 + 𝜔𝑛 √𝜁 2 − 1 )𝑒 𝜔𝑛 √𝜁 𝑡
− (𝜁𝜔𝑛 − 𝜔𝑛 √𝜁 2 − 1 )𝑒 −𝜔𝑛 √𝜁 𝑡
]
2𝜔𝑛 √𝜁 2 −1
𝜁+√𝜁 2 −1 2 −1 𝜁−√𝜁 2 −1 2 −1
𝑥(𝑡) = 1 − 𝑒 −𝜁𝜔𝑛 𝑡 {( ) 𝑒 𝜔𝑛 √𝜁 𝑡
−( ) 𝑒 −𝜔𝑛 √𝜁 𝑡
}………………..(30)
2√𝜁 2 −1 2√𝜁 2 −1

#Exercise 10
Plot equation (30) in Matlab/Scilab/Excel with 𝜻 =2 for different values of 𝜔𝑛 = 1, 3, and 5

Underdamped system (𝜁 < 1)

Let us start from equation (27)
1 𝜔𝑛 2 1 𝝎𝒏 𝟐 1 𝝎𝒏 𝟐
𝑋(𝑠) = ( ) =( ) =( )
𝑠 [𝒔+𝜻𝝎𝒏 ]𝟐 −[𝝎𝒏 √𝜻𝟐 −𝟏]𝟐 𝑠 [𝑠+𝜁𝜔 ]2 −[𝑗𝜔 √1−𝜁 2 ]2 𝑠 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2
𝑛 𝑛

Recall that 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2 and 𝑗 2 = (−1)

1 𝜔𝑛 2 𝐴 𝐵𝑠+𝐶
𝑋(𝑠) = ( ) = + (since the numerator can not be factorised,it is kept as such)
𝑠 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2 𝑠 𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
𝜔𝑛 2 = 𝐴(𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛 2 ) + (𝐵𝑠 + 𝐶)𝑠
When 𝑠 = 0, 𝐴 = 1
Equating coefficients of s2, 0 = 𝐴 + 𝐵 ≫≫ 𝐵 = −1
Equating coefficients of s, 0 = 𝐴2𝜁𝜔𝑛 + 𝐶 ≫≫ 𝐶 = −2𝜁𝜔𝑛
1 𝑠+2𝜁𝜔𝑛 1 𝑠+2𝜁𝜔𝑛
𝑋(𝑠) = − 𝑠2 +2𝜁𝜔 2
=𝑠− [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2
𝑠 𝑛 𝑠+𝜔𝑛
1 𝑠+𝜁𝜔𝑛 𝜁𝜔𝑛
= 𝑠 − [𝑠+𝜁𝜔 2 2 − [𝑠+𝜁𝜔 2 +[𝜔 2
𝑛 ] +[𝜔𝑑 ] 𝑛] 𝑑]
1 𝑠+𝜁𝜔𝑛 𝜁𝜔 𝜔𝑑
= 𝑠 − [𝑠+𝜁𝜔 2 2 − ( 𝜔 𝑛) [𝑠+𝜁𝜔 2 +[𝜔 2
𝑛 ] +[𝜔𝑑 ] 𝑑 𝑛] 𝑑]
1 𝑠+𝜁𝜔𝑛 𝜁 𝜔𝑑
𝑋(𝑠) = − [𝑠+𝜁𝜔 2 2 −( )
𝑠 𝑛 ] +[𝜔𝑑 ] √1−𝜁 2 [𝑠+𝜁𝜔𝑛 ]2 +[𝜔𝑑 ]2

Taking inverse Laplace Transform (revisit the laplace transforms of 𝑠𝑖𝑛 𝜔𝑡 and 𝑐𝑜𝑠 𝜔𝑡)
𝜁 𝜁
𝑥(𝑡) = 1 − 𝑒 −𝜁𝜔𝑛 𝑡 cos 𝜔𝑑 𝑡 − ( ) 𝑒 −𝜁𝜔𝑛 𝑡 sin 𝜔𝑑 𝑡 = 1 − 𝑒 −𝜁𝜔𝑛 𝑡 {cos 𝜔𝑑 𝑡 + ( ) sin 𝜔𝑑 𝑡}
√1−𝜁 2 √1−𝜁 2
𝑒 −𝜁𝜔𝑛𝑡
= 1− {√1 − 𝜁 2 cos 𝜔𝑑 𝑡 + 𝜁 sin 𝜔𝑑 𝑡}
√1−𝜁 2
𝑒−𝜁𝜔𝑛𝑡
=1 − (𝑋0 sin 𝜃 cos 𝜔𝑑 𝑡 + 𝑋0 cos 𝜃 sin 𝜔𝑑 𝑡) where 𝑋0 sin 𝜃 = √1 − 𝜁2 𝑎𝑛𝑑 𝑋0 cos 𝜃 = 𝜁
√1−𝜁 2

√1−𝜁2
𝑋0 2=(1 − 𝜁 2 ) + 𝜁 2 = 1 and tan 𝜃 =
𝜁
1 −𝜁 𝜔𝑛 𝑡
𝑥(𝑡) = 1 − 𝑒 sin(𝜔𝑑 𝑡 + 𝜃) …………………………………(31)
√1−𝜁2

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

#Exercise 11
Plot equation (31) in Matlab/Scilab/Excel with 𝜻 =0.2 for different values of 𝜔𝑛 =1, 3, and 5.
Note that the value of θ should be in radian.
The unit step responses of these systems are shown in figure below

2𝜋
𝑇= .
𝜔𝑑

Looking at the plots of equations (29), (30), and (31), we can understand that the responses of
critically damped and overdamped systems reach the desired output without oscillation as t→ ∞,
where as, an underdamped system response will have oscillations before attaining the final value.
2𝜋
The frequency (circular frequency) of oscillations is 𝜔𝑑 rad/s with period 𝑇= . These
𝜔𝑑
oscillations die out after some time and the response reaches a steady state.

#Exercise 12
Using Matlab/Scilab, plot the step responses of a system with 𝜻 =0.1 for different values of
𝜔𝑛 =1, 3, and 5. Repeat it with 𝜔𝑛 =5 for different values of 𝜻 =0.1, 0.5, 0.8, 1, and 1.5. Examine
the plots and draw your inferences.
We can see that the transient reponse and steady state response depends on the values of the
poles of the second order transfer function −𝜻𝝎𝒏 ± 𝝎𝒏 √𝜻𝟐 − 𝟏 given in Equaion(23). It is already
explained that if the system is overdamped, the poles are negative real numbers given by
−𝜻𝝎𝒏 − 𝝎𝒏 √𝜻𝟐 − 𝟏 and −𝜻𝝎𝒏 + 𝝎𝒏 √𝜻𝟐 − 𝟏
If it is critically damped, we have a double pole on the real axis at −𝜻𝝎𝒏. For underdamped
system, the poles are complex conjugate with negative real part given by −𝜻𝝎𝒏 ± 𝒋𝝎𝒏 √𝟏 − 𝜻𝟐
25
Let us consider a transfer function G(s) = 𝑠2 +2𝑠+25 . The poles are obtained from the
characteristic equation 𝑠 2 + 2𝑠 + 25 = 0, the roots of which are 𝑠 = −1 ± 𝑗4.9. (We can also find
the poles by knowing the values of 𝜁 = 0.2 𝑎𝑛𝑑 𝜔𝑛 = 5 from the transfer function sothat
−𝜁𝜔𝑛 = −1 and 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2 = 4.9)

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

We can plot these poles on the s-plane which shown in figure.
2
The distance 𝑂𝑃1 = √(−𝜁𝜔𝑛 )2 + (𝜔𝑑 )2 = √(𝜁𝜔𝑛 )2 + (𝜔𝑛 √1 − 𝜁 2 ) = √𝜁 2 𝜔𝑛 2 + 𝜔𝑛 2 (1 − 𝜁 2 ) = 𝜔𝑛

The angle θ can be found out using any of the following relations
𝜁𝜔𝑛
tan 𝜃 =
𝜔𝑛 √1−𝜁 2
=
√1−𝜁 2
..(32a) cos 𝜃 = 𝜔𝑛
= 𝜁 …...(32b) sin 𝜃 =
𝜔𝑛 √1−𝜁 2
= √1 − 𝜁 2 ..(32c)
𝜁𝜔𝑛 𝜁 𝜔𝑛

Where ever be the pole location, the distance between the pole and the origin is always 𝜔𝑛 . If
we fix 𝜔𝑛 and change the value of 𝜁, the pole location moves along the circular path as shown by
the blue * in fig. 10 b.
If we fix 𝜁 and change the value of 𝜔𝑛 , the location of the pole moves along the line inclined at
an angle 𝜃 = cos −1 𝜁 as shown by the brown in fig.10 b.

P1 +𝑗𝜔𝑑 = 𝑗4.9

(𝜃
−𝜁𝜔𝑛 = −1
O

−𝑗𝜔𝑑 = −𝑗4.9

P2

25
The pole-zero plot of G(s) = 𝑠2 +2𝑠+25 b. movement of the pole location with 𝜁,𝜔𝑛

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

You might have observed the influence of 𝜁 𝑎𝑛𝑑 𝜔𝑛 (or rather the location of the poles) on the
transient response while doing #Excersise 16. The figures below shows the influence of damping
factor and natural frequency on the step response. What are your inferences?

(b)
(a)

(c) (d)
Comparison of the step responses for different values of damping factor and frequency
Please remember that the type of transient respons is determined by the closed loop poles while
the shape of the transient response is primarily determined by the closed loop zeroes.
We have already seen that depending on the location of the poles, the transient response can be
undamped, underdamped, critically damped, or overdamped. Now see the influence of location of
zero on the transient response.
Look at fig. 12. They represent the unit
step responses of
25
i) 𝐺(𝑠) = 𝑠2+5𝑠+25 (No zero)
25𝑠
ii) 𝐺(𝑠) = 𝑠2+5𝑠+25 (zero at z=0)
25(𝑠+1)
iii) 𝐺(𝑠) = 𝑠2+5𝑠+25 (zero at z= -1)

iv) 𝐺(𝑠) = 𝑠2+5𝑠+25 ((zero at z= -4)

25(𝑠+4)

Poles remain the same.

Fig.12

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

The transient behavior of an underdamped system can be specified by rise time, peak time,
settling time, and maximum percentage overshoot.
Rise time (tr): It is the time taken by the system to reach the desired output for the first time.
When 𝑡 = 𝑡𝑟 , 𝑥(𝑡) = 1. Therefore, Equation (31) can be written as
1
1= 1− 𝑒 −𝜁𝜔𝑛 𝑡𝑟 sin(𝜔𝑑 𝑡𝑟 + 𝜃)
√1−𝜁 2

1
𝑒 −𝜁𝜔𝑛𝑡𝑟 sin(𝜔𝑑 𝑡𝑟 + 𝜃) = 0
√1−𝜁2

sin(𝜔𝑑 𝑡𝑟 + 𝜃) = 0
𝜔𝑑 𝑡𝑟 + 𝜃 = 0, 𝜋, 2𝜋 … … (If we consider 0, we get negative time. Hence we take 𝜋)
𝜋−𝜃
𝑡𝑟 = ……………………………….(33a) where 𝜃 = cos−1 𝜁 and 𝝎𝒅 = 𝝎𝒏 √𝟏 − 𝜻𝟐
𝜔𝑑

Peak Time (tp): It is the time taken by the system to reach the maximum (peak) value of the out
𝑑
put. When x(t) is maximum, 𝑑𝑡
𝑥(𝑡) = 0. Differentiating Equation (31) with repect to time

𝑑 1 1
𝑑𝑡
𝑥(𝑡) = − 𝑒−𝜁𝜔𝑛𝑡𝑝 cos(𝜔𝑑 𝑡𝑝 + 𝜃) 𝜔𝑑 + sin(𝜔𝑑 𝑡𝑝 + 𝜃) {− (−𝜁𝜔𝑛 )𝑒−𝜁𝜔𝑛𝑡𝑝 } = 0
√1−𝜁 2 √1−𝜁2

cos(𝜔𝑑 𝑡𝑝 + 𝜃) 𝜔𝑑 = sin(𝜔𝑑𝑡𝑝 + 𝜃)𝜁 𝜔𝑛

sin(𝜔𝑑 𝑡𝑝 +𝜃) 𝜔 √1−𝜁 2
= 𝜁𝜔𝑑 = = tan 𝜃 (from 32a)
cos(𝜔𝑑 𝑡𝑝 +𝜃) 𝑛 𝜁

tan(𝜔𝑑 𝑡𝑝 + 𝜃) = tan 𝜃 = tan(𝜋 + 𝜃)

𝜔𝑑 𝑡𝑝 + 𝜃 = 𝜋 + 𝜃
𝜋
𝑡𝑝 = ………………………………………………(33b)
𝜔𝑑
Maximum overshoot: It is the maximum value reached by the system response above the
desired output. When 𝑡 = 𝑡𝑝 , 𝑥(𝑡𝑝 ) is maximum
1 1 −𝜁 𝜔𝑛
𝜋 𝜋
𝑥(𝑡𝑝 ) = 1 − (𝑒 −𝜁𝜔𝑛𝑡𝑝 ) sin(𝜔𝑑 𝑡𝑝 + 𝜃) = 1 − (𝑒 𝜔𝑑 ) sin(𝜔 + 𝜃)
𝑑 𝜔𝑑
√ 1 − 𝜁2 √ 1 − 𝜁2
𝜋 𝜋𝜁
−𝜁𝜔𝑛 −
1 𝝎𝒏 √𝟏−𝜻𝟐 1 √𝟏−𝜻𝟐
=1− (𝑒 ) sin(𝜋 + 𝜃 ) = 1 + (𝑒 ) sin 𝜃
√1−𝜁2 √1−𝜁2

𝜋𝜁 𝜋𝜁
− −
1 √𝟏−𝜻𝟐 √𝟏−𝜻𝟐
=1+ (𝑒 ) √1 − 𝜁 2 = 1 + 𝑒 (from 32 c)
√1−𝜁2

𝜋𝜁 𝜋𝜁
− −
√𝟏−𝜻𝟐 √𝟏−𝜻𝟐
Maximum overshoot = 𝑥(𝑡𝑝 ) − 𝑑𝑒𝑠𝑖𝑟𝑒𝑑 𝑜𝑢𝑡𝑝𝑢𝑡 = (1 + 𝑒 )−1= 𝑒

𝜋𝜁
−
√𝟏−𝜻𝟐
Max.overshoot= 𝑒 ………………………..(33c) (it is independent of 𝜔𝑛 )

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Settling time (ts): The amplitude of the underdamped response will be always between the
1
. The
1
envelopes 1 ± 𝑒−𝜁𝜔𝑛𝑡as shown in figure 11.Time constant of these envelopes is
√1−𝜁2
𝜁𝜔𝑛

speed of response depends on the value of this time constant. The settling time is defined as the
time required for the system to settle within a certain percentage (usually ±2%) of the desired
output. This ±2% band is shown by the green line in fig.11. Settling time is the time taken by the
system after which the output completely remains within the band. This happens approximately
when 𝑒−𝜁𝜔𝑛𝑡𝑠 < 0.02
4
Or 𝑡𝑠 ≅ ………………………………..(33d)
𝜁𝜔 𝑛
Having learnt the response of second order systems, we can appreaciate the influence of poles
on the system response. The poles which are far away from the origin will not have much influence
on the transient response. The complex conjugate poles near the origin ogf the s-plane
relative to the other poles dominate the transient response and are called dominant poles. If we
have a real pole and complex poles, if the ratio of real pole to the real part of the complex pole
exceeds 5, then the complex poles are considered as dominant. Even if the system is of higher
order, if there are dominant poles the system can be reduced to second order.

1
1+ 𝑒−𝜁𝜔𝑛𝑡
√1 − 𝜁 2

1
1− 𝑒−𝜁𝜔𝑛𝑡 𝑠𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑡𝑖𝑚𝑒
√1 − 𝜁 2
𝑡𝑠

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Example 2
25
For a negative unity feed back system with forward transfer function 𝐺(𝑠) = 𝑠(𝑠+7) find
the rise time, peak time, 2% settling time, and maximum percentage overshoot.
25
Given 𝐺(𝑠) = 𝑠(𝑠+7) , 𝐻(𝑠) = 1
25
𝑌(𝑠) 𝐺(𝑠) 𝑠(𝑠+7) 25
Transfer fuunction of the system= = 𝑋(𝑠) = 1+𝐺(𝑠)𝐻(𝑠) = 25 = 𝑠2+7𝑠+25
1+𝑠(𝑠+7)
𝜔𝑛 2
Comparing this with the standard form
𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2
2

𝜔𝑛 2 = 25, and 2𝜁𝜔𝑛 = 7

7
Hence 𝜔𝑛 = 5 rad/s, and 𝜁 = = 0.7 (The system is underdamped)
2𝜔 𝑛
𝜋−𝜃
Rise time 𝑡𝑟 = where 𝜃 = cos−1 𝜁 = 45.60 = 0.79 𝑟𝑎𝑑
𝜔𝑑
𝝎𝒅 = 𝝎𝒏 √𝟏 − 𝜻𝟐 = 𝟑. 𝟓𝟕 𝒓𝒂𝒅/𝒔
𝜋−0.79 𝜋
𝑡𝑟 = 3.57
= 0.659 𝑠 𝑡𝑝 = = 0.88 𝑠
𝜔𝑑
4 𝜋𝜁
𝑡𝑠 ≅ 𝜁𝜔 = 1.14 𝑠 −
√𝟏−𝜻𝟐
% max overshoot =𝑒
𝑛
𝑥100 = 4.5%

Example 3
A second order control system has the following specifications. Find the required values of
the natural frequency and damping ratio of the system. What will be its rise time & peak
time?
1. Maximum percantage overshoot permitted is 5%.
2. 2% settling time must be less than 4𝑠

Given % overshoot =5% (𝜔𝑑 = 𝜔𝑛 √1 − 𝜁2 = 1.05𝑟𝑎𝑑/𝑠)

(𝜃 = cos−1 𝜁 = 46.40 = 0.809 𝑟𝑎𝑑)
𝜋𝜁
− Settling time ts =4
√𝟏−𝜻𝟐
= 0.05 ≫≫ −
𝜋𝜁
𝑒 √𝟏−𝜻𝟐
= ln 0.05 = −2.995 4
=4 ≫≫ 𝝎𝒏 = 𝟏. 𝟒𝟓 𝒓𝒂𝒅/𝒔
𝜁𝜔𝑛
2 2 2
𝜋 𝜁 = 8.974(1 − 𝜁 ) ≫≫ 𝜻 = 𝟎. 𝟔𝟗
𝜋 𝜋−𝜃 𝜋−0.809
Peak time 𝑡𝑝 = = 2.99 𝑠 Rise time 𝑡𝑟 = = =
𝜔𝑑 𝜔𝑑 1.05
2.22 𝑠

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

#Exercise 13
𝐾
The open loop transfer function of a unity negative feed back system is 𝐺(𝑠) = . A
𝑠(𝑠+2)
system response to a unit step input is specified as follows.
Peak time 𝑡𝑝 = 1.1𝑠,
Maximum overshoot= 5%.
Determine whether both the specifications can be met simultaneously.

#Excercise 14
𝐾
A unity negative feedback control system has the plant 𝐺(𝑠) = 𝑠(𝑠+ . Determine the
√2𝐾)
percentage overshoot and 2% settling time for unit step input. For what range of K, the
settling time is less than 1 s.

#Exercise 15
500
A closed loop control system has a transfer function 𝑇(𝑠) = . Plot its step
(𝑠+10)(𝑠2 +10𝑠+50)
response using Matlab/Scilab. Then consider the relatively dominant complex poles and plot
its step response. Compare the results.

#Excercise 16
The control system for positioning the head of a cd drive has the closed loop transfer
0.313 (𝑠+0.8)
function 𝑇(𝑠) = . Plot using Matlab/Scilab, the poles and zeroes and the step
(𝑠+0.25)(𝑠2 +0.3𝑠+1)
response of the system. What are the time response specifications like rise time, peak time,
and overshoot. Now look at the dominance of the complex poles. If we consider the
dominannt poles only, how the tansient response specifications change?

Control Engineering…………………………………………………………………………Dr. Biju N, Professor, SOE, CUSAT

Page |1
STEADY STATE ERROR OF FEEDBACK CONTROL SYSTEMS
One of the fundamental reasons for using feedback is the improvement in the reduction of
steady stare error of the system. The steady state error is the error after the transient
response has decayed (i.e., the value of error when t→ ∞. Let us compare the steady state errors
of an open loop and closed loop system.
Consider the open loop system shown in figure.
The input is 𝑋(𝑠) and the output is 𝑌(𝑠). The system X(s) Y(s)
G(s)
transfer function is 𝐺(𝑠).
𝑌(𝑠) = 𝐺(𝑠). 𝑋(𝑠)

The diffrence between the desired output (input) and the actual output is known as the error
𝑒(𝑡) = 𝑥(𝑡) – 𝑦(𝑡)
𝐸𝑜 (𝑠) = 𝑋(𝑠) − 𝑌(𝑠) = 𝑋(𝑠) – 𝐺(𝑠)𝑋(𝑠) (the suffix ‘o’ denotes open loop and ‘c’ closed loop)
𝐸𝑜 (𝑠) = [1 − 𝐺(𝑠)] 𝑋(𝑠) …………………………… (1)
For a unit step input, 𝑋(𝑠) = 1/𝑠
1
The steady state error = lim 𝑒(𝑡) = lim 𝑠 𝐸0 (𝑠) = lim 𝑠 [1 − G(s)] 𝑠 = 1 − 𝐺(0) ……………. (2)
𝑡→∞ 𝑠→0 𝑠→0
Consider the closed loop system
X(s) Ec(s) Y(s)
𝑍(𝑠) = 𝑌(𝑠) 𝐻(𝑠) and 𝑌(𝑠) = 𝐸𝑐 (𝑠). 𝐺(𝑠)𝐸𝑐(𝑠) = G(s)
𝑋(𝑠) – 𝑍(𝑠) = 𝑋(𝑠) – 𝑌(𝑠)𝐻(𝑠) = 𝑋(𝑠) − 𝐸𝑐 (𝑠)𝐺(𝑠)𝐻(𝑠) +
-
𝐸𝑐 (𝑠) {1 + 𝐺(𝑠)𝐻(𝑠)} = 𝑋(𝑠)
1 Z(s) H(s)
𝐸𝑐 (𝑠) = 𝑋(𝑠) …………………..(3)
1+𝐺(𝑠)𝐻(𝑠)
Assuming a unity feed back, 𝐻(𝑠) = 1
1
𝐸𝑐 (𝑠) = 𝑋(𝑠) ……………………………(4)
1+𝐺(𝑠)
The steady state error for a unit step input (𝑋(𝑠) = 1/𝑠)
1 1 1
= lim 𝑒(𝑡) = lim 𝑠 𝐸𝑐 (𝑠) = lim 𝑠 [1+𝐺(𝑠)] 𝑠 = 1+𝐺(0) ………………………………………………..(5)
𝑡→∞ 𝑠→0 𝑠→0
The value of 𝐺(𝑠) when 𝑠 = 0 is often called the dc gain and normally is greter than 1. Therefore,
comparing equations (2) and (3), it can be seen that the steady state error of a closed loop
system is considerably less than the open loop system.

We can now determine the steady state error of the system for three standard test inputs
1
𝐸(𝑠) = 𝑋(𝑠)
1+𝐺(𝑠)𝐻(𝑠)

Remember 𝐺(𝑠) stands for the open loop transfer function (transfer function of the
forward path)

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
In order to explain how these test signals are used, let us assume a position control system,
where the output position follows the input commanded position.
Step inputs represent constant position and thus are useful in determining the ability of
the control system to position itself with respect to a stationary target. An antenna position
control is an example of a system that can be tested for accuracy using step inputs.
Ramp inputs represent constant-velocity inputs to a position control system by their linearly
increasing amplitude. These waveforms can be used to test a system’s ability to follow a linearly
increasing input or, equivalently, to track a constant velocity target. For example, a position
control system that tracks a satellite that moves across the sky at a constant angular velocity.
Parabolic inputs, whose second derivatives are constant, represent constant acceleration inputs
to position control systems and can be used to represent accelerating targets, such as a missile.

STEP INPUT (constant)

The input is a step of magnitude 𝐴, 𝑥(𝑡) = 𝐴
𝐴
𝑋(𝑠) = 𝑠
1 1 𝐴
𝐸(𝑠) = 𝑋(𝑠) = 1+𝐺(𝑠)𝐻(𝑠) . 𝑠
1+𝐺(𝑠)𝐻(𝑆)
1 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim 𝑠. 𝐸(𝑠) = lim 𝑠. [1+𝐺(𝑠)𝐻(𝑠) ] . 𝑠 = 1+lim 𝐺(𝑠)𝐻(𝑠) = 1+𝐾 where 𝐾𝑝 = lim 𝐺(𝑠)𝐻(𝑠) and is
𝑠→0 𝑠→0 𝑝 𝑠→0
𝑠→0
known as position error constant

∏𝑀
𝑖=1(𝑠+𝑧𝑖 )
The general form of the forward transfer function is 𝐺(𝑠) = 𝐾 where ∏ denotes the
∏𝑃
𝑘=1(𝑠+𝑝𝑘)
product of factors. If there are 𝑁 poles at the origin, then we can decompose the denominator
as 𝑠 𝑁 ∏𝑄
𝑘=1(𝑠 + 𝑝𝑘 ) so that 𝑁 + 𝑄 = 𝑃.
∏𝑀 (𝑠+2)(𝑠+1)
𝐺(𝑠) = 𝐾 𝑄
𝑖=1(𝑠+𝑧𝑖 ) for example 𝐺(𝑠) = 100 𝑠3 (𝑠+3)(𝑠+10)
𝑠𝑁 ∏𝑘=1(𝑠+𝑝𝑘 )

The number of s in the denominator indicates number of integrations. The number of inegrations
𝑁 is often called type number.
∏𝑀 (𝑠+1)
For a type zero system, 𝑁 = 0 , 𝐺(𝑠) = 𝐾 𝑖=1(𝑠+𝑧𝑖 ) for example 𝐺(𝑠) = 100 (𝑠+3)(𝑠+10)
∏𝑄
𝑘=1(𝑠+𝑝𝑘 )

𝐴
𝑒𝑠𝑠 = 1+𝐾 where 𝐾𝑝 = lim 𝐺( 𝑠)𝐻(𝑠).
𝑝 𝑠→0
∏𝑀
𝑖=1(𝑠+𝑧𝑖 )
If 𝑁 ≥ 1, 𝐺(𝑠) = 𝐾 𝑄 , Kp = lim 𝐺( 𝑠)𝐻(𝑠). = ∞
𝑠𝑁 ∏𝑘=1(𝑠+𝑝𝑘 ) 𝑠→0
𝐴
and steady stae error, 𝑒𝑠𝑠 = =0
1+𝐾𝑝

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
RAMP INPUT (linear)
Input is a ramp with slope A, 𝑥(𝑡) = 𝐴𝑡
𝑋(𝑠) = 𝐴/𝑠 2
1 𝐴 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim 𝑠. 𝐸(𝑠) = lim 𝑠. . 2 = lim = =
𝑠→0 𝑠→0 1 + 𝐺(𝑠)𝐻(𝑠) 𝑠 𝑠→0 𝑠 + 𝑠𝐺(𝑠)𝐻(𝑠) lim 𝑠𝐺(𝑠)𝐻(𝑠) 𝐾𝑣
𝑠→0
Where 𝐾𝑣 = lim sG(s)H(s) and is known as velocity error constant
𝑠→0
∏𝑀
𝑖=1(𝑠+𝑧𝑖 )
If 𝐺(𝑠) is type 0 (no s term in the denominator), 𝐺(𝑠) = 𝐾
∏𝑄
𝑘=1(𝑠+𝑝𝑘 )

𝐾𝑣 = lim sG(s)H(s) = 0 ≫≫ 𝑒𝑠𝑠 = ∞ . A type 0 system can not track a ramp input.
𝑠→0
∏𝑀
𝑖=1(𝑠+𝑧𝑖 )
If 𝐺(𝑠) is type 1 (one s in the denominator), 𝐺(𝑠) = 𝐾
𝑠 ∏𝑄
𝑘=1
(𝑠+𝑝𝑘 )
𝐴
𝐾𝑣 = lim sG(s)H(s) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ≫≫ ≫≫ then we have a constant steady state error 𝑒𝑠𝑠 =
𝑠→0 𝐾𝑣
∏𝑀
𝑖=1(𝑠+𝑧𝑖 )
If 𝐺(𝑠) is type 2 or higher (2 or more s in the denominator), 𝐺(𝑠) = 𝐾 𝑄
𝑠𝑁 ∏𝑘=1(𝑠+𝑝𝑘 )

𝐾𝑣 = lim sG(s)H(s) = ∞ then, 𝑒𝑠𝑠 = 0

𝑠→0

ACCELERATION INPUT (Parabolic)

1 A
If the input is 𝑥(𝑡) = 𝐴𝑡 2 , then 𝑋(𝑠) =
2 𝑠3
1 𝐴 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim 𝑠. 𝐸(𝑠) = lim 𝑠. . 3 = lim 2 2
= 2
=
𝑠→0 𝑠→0 1 + 𝐺(𝑠)𝐻(𝑠) 𝑠 𝑠→0 𝑠 + 𝑠 𝐺(𝑠)𝐻(𝑆) lim 𝑠 𝐺(𝑠)𝐻(𝑠) 𝐾𝑎
𝑠→0
Where 𝐾𝑎 = lim s G(s)H(s) and is known as acceleration error constant
2
𝑠→0

For type 0 𝑎𝑛𝑑 1 systems, 𝐾𝑎 = lim s 2 G(s)H(s) = 0 or 𝑒𝑠𝑠 = ∞

𝑠→0
𝐴
For type 2 systems, 𝐾𝑎 = lim s G(s)H(s) = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡, 𝑒𝑠𝑠 is constant and is equal to
2
𝑠→0 𝐾𝑎
For type 3 and above, 𝐾𝑎 = lim s G(s)H(s) = ∞ , 𝑒𝑠𝑠 = 0
2
𝑠→0
We can summarise this as follows
Type Number 0 1 2 3
Input

Step, 𝑥(𝑡) = 𝐴, 𝒆𝒔𝒔 =

𝑨 0 0 0
𝑋(𝑠) = 𝐴/𝑠 𝟏 + 𝑲𝑷

Ramp, 𝑥(𝑡) = 𝐴𝑡, ∞

𝒆𝒔𝒔 =
𝑨 0 0
𝑋(𝑠) = 𝐴/𝑠2 𝑲𝒗

Parabolic, ∞ ∞
𝒆𝒔𝒔 =
𝑨 0
1 𝑲𝒂
𝑥(𝑡) = 𝐴𝑡 2
2
𝑋(𝑠) = 𝐴/𝑠3

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
The error constants, 𝐾𝑝 , 𝐾𝑣 , 𝑎𝑛𝑑 𝐾𝑎 , of a control system describe the ability of a system to
reduce or eliminate the steady state error. They are used as numerical measures of steady state
performance. The system designer determines the errror constants for a given system and
explores the methods of increasing the error constants (thereby minimising the steady state
error) while maintaing an acceptable transient response.

We can reduce the feedback system shown in the figure in to a single block of transfer function
𝐺(𝑠)
𝑇(𝑠) = 1+𝐺(𝑠)𝐻(𝑠) with input 𝑋(𝑠) and output 𝑌(𝑠). Then 𝑌(𝑠) can be written as
𝑌(𝑠) = 𝑇(𝑠). 𝑋(𝑠)
The error 𝐸(𝑠) = 𝑑𝑒𝑠𝑖𝑟𝑒𝑑 𝑜𝑢𝑡𝑝𝑢𝑡 – 𝑎𝑐𝑡𝑢𝑎𝑙 𝑜𝑢𝑡𝑝𝑢𝑡 = 𝑋(𝑠) – 𝑌(𝑠) = 𝑋(𝑠) {1 − 𝑇(𝑠)}
The steady state error, ess = lim 𝑠 𝐸(𝑠) = lim 𝑠 𝑋(𝑠){1 − 𝑇(𝑠)} (Final value theorem)
𝑠→0 𝑠→0
This can also be used to find out the steady state error.
Example 1
A severely disabled person could use a mobile Y(s), actual
robot as an assisting device. The steering control X(s), desired heading angle
system for such a robot can be represented by heading angle

the block diagram. The steering controller 𝐺1 (𝑠) = G1(s) G2(s)

+
𝐾
𝐾1 + 𝑠2 . The vehicle dynamics is represented by -
𝐾
𝐺2 (𝑠) = . Determine the steady state error
1+𝑇𝑠
for a step and ramp input.
The open loop transfer function of the given system
𝐾2 𝐾 𝑠𝐾1 +𝐾2
𝐺(𝑠) = 𝐺1 𝐺2 =(𝐾1 + )( ) = 𝐾. . Look at the denominator, this is type 1 system
𝑠 1+𝑇𝑠 𝑠 (1+𝑇𝑠)
𝐻(𝑠) = 1
The position error constant for a step input 𝑥(𝑡) = 𝐴,
𝑠𝐾 +𝐾
You can also verify the result using
𝐾𝑝 =lim 𝐺( 𝑠)𝐻(𝑠). = lim 𝐾. 𝑠 (1+𝑇𝑠)
1 2
= ∞. 𝑒𝑠𝑠 = lim 𝑠 𝑋(𝑠) {1 − 𝑇(𝑠)}
𝑠→0 𝑠→0 𝑠→0
𝐴
The steady state error 𝑒𝑠𝑠 = 1+𝐾 = 0
𝑝

If the input is ramp, 𝑥(𝑡) = 𝐴𝑡

𝑠𝐾1 +𝐾2
The velocity error constant 𝐾𝑣 = lim 𝑠𝐺( 𝑠)𝐻(𝑠) = lim 𝑠 𝐾. .
𝑠→0 𝑠→0 𝑠 (1+𝑇𝑠)
𝑠𝐾1 +𝐾2 𝐾
𝐾𝑣 = lim 𝑠 ( ) (1+𝑇𝑠) = 𝐾. 𝐾2
𝑠→0 𝑠
𝐴 𝐴
The steady state error, 𝑒𝑠𝑠 = =
𝐾𝑣 𝐾.𝐾2

Example 2
What is the steady state error to a constant
reference input 𝑟(𝑡) = 3𝑐𝑚. 1(𝑡) for the
following feedback positioning system?
10 3
𝐺(𝑠) = 𝑠+1 , 𝑅(𝑠) = 𝑠 , 𝐻(𝑠) = 1
𝐴 10
Steady state error for a step input 𝑒𝑠𝑠 = where Kp =lim 𝐺( 𝑠)𝐻(𝑠). = lim = 10
1+𝐾𝑝 𝑠→0 𝑠→0 𝑠+1
𝐴 3 𝑐𝑚
𝑒𝑠𝑠 = 1+𝐾 = 1+10 = 0.27 𝑐𝑚 You can also verify the result using
𝑝
𝑒𝑠𝑠 = lim 𝑠 𝑅(𝑠) {1 − 𝑇(𝑠)}
𝑠→0

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

Example 3
Design the controller 𝐷(𝑠) for an error of
0.05 for a unit ramp input

Looking at the transfer function of the plant, we can understand that it is 𝑡𝑦𝑝𝑒 0. For getting a
constant steady state error for a unit ramp, the forward path transfer function should be 𝑡𝑦𝑝𝑒 1.
That is there should be one integrator in the forward path.
𝐾
Assuming the simplest case, 𝐷(𝑠) =
𝑠
𝐾 (𝑠+2)
𝐺(𝑠) = 𝑠(𝑠+1)(𝑠+5)
, 𝐻(𝑠) = 1, 𝑅(𝑠) = 1/𝑠2 (𝐴 = 1), 𝑒𝑠𝑠 = 0.05
𝐴 𝐾 (𝑠+2) 2𝐾
𝑒𝑠𝑠 = 𝐾 where 𝐾𝑣 = lim 𝑠𝐺( 𝑠)𝐻(𝑠) = lim 𝑠 𝑠(𝑠+1)(𝑠+5) =
𝑣 𝑠→0 𝑠→0 5
1
i.e., 0.05 = 2𝐾 which gives 𝐾 = 50.
5

Example 4
Determine the controller gain 𝐾 to limit the
steady state error of the system to 0.02 for
a constant reference input

10 𝐾
Open loop transfer function 𝐺(𝑠) = 𝑠+5 You can also verify the result using
2 1
𝐻(𝑠) = , For a unit step input 𝑅(𝑠) = , 𝐴=1 𝑒𝑠𝑠 = lim 𝑠 𝑅(𝑠) {1 − 𝑇(𝑠)}
𝑠+2 𝑠 𝑠→0
𝑒𝑠𝑠 = 0.02
𝐴 10𝐾 2
𝑒𝑠𝑠 = 1+𝐾 where 𝐾𝑝 = lim 𝐺( 𝑠)𝐻(𝑠). = lim (𝑠+5) (𝑠+2) = 2𝐾
𝑝 𝑠→0 𝑠→0
1
i.e., 0.02 = 1+2𝐾
1 + 2𝐾 = 50 which will give thr Controller gain 𝐾 = 24.5

Exercise 19
10(𝑠+4)
19.1 A system with unity negative feed back has transfer function G(s) =𝑠(𝑠+1)(𝑠+2)(𝑠+5)
Determine the steady state error for a unit step and unit ramp input.
𝐾
19.2 A feedback system is shown in figure. 𝐺(𝑠) = 𝑠(𝑠+2) and
(𝑠+3) X(s) + G(s) Y(s)
𝐻(𝑠) = . Determine the steady state error for a unit step Gp(s) +
(𝑠+0.1)
-
when 𝐾 = 0.4 and 𝐺𝑝 (𝑠) = 1. Select an appropriate value for
H(s)
𝐺𝑝 (𝑠) sothat the steady state error becomes zero for a unit
step input.
19.3 A thermometer requires 1 𝑚𝑖𝑛 to indicate 98% of the response to a step input. Assming
it as a first order system, determine the time constant. If the thermometer is placed in a
bath whose temperature is changing linearly at a rate of 10 𝑑𝑒𝑔/𝑚𝑖𝑛, how much error will it
show?

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
DISTURBANCE SIGNALS IN A FEEDBACK CONTROL SYSTEM
One of the important effect of feed back in a control system is the control and partial
elimination of the effect of disturbance signals. A disturbance signal is an unwanted input signal
that affects the system’s output.
Consider a closed loop system.
D(s)
Let the plant is rpresented by
G2(s) and the controller by G1(s). +
X(s) + G2(s) Y(s)
A disturbance 𝐷(𝑠) acts on the G1(s) +
-
system. Now, we have two inputs
H(s)
in to the system.

Since we are dealing with only linear systems, we can apply the principle of superposition. We
will consider only one input at a time and find the corresponding output. The total output when
both inputs are present will be equal to the sum of the individual outputs.
Let us assume 𝐷(𝑠) = 0. The input is only 𝑋(𝑠).
𝐺 𝐺
The block diagram can be reduced to a single block of transfer function 𝑇1 (𝑠) = 1+𝐺1 𝐺2 𝐻
1 2
𝑌1 (𝑠) 𝐺1 𝐺2 𝐺1 𝐺2
= from which 𝑌1 (𝑠) = 𝑋(𝑠) ( )
𝑋(𝑠) 1+𝐺1 𝐺2 𝐻 1+𝐺1 𝐺2 𝐻

Now consider that 𝑋(𝑠) = 0. The input is only 𝐷(𝑠). Now the blocks in feedback path becomes
(−𝐻) and (𝐺1 )
𝐺
𝑇2 (𝑠) = 1+𝐺 2𝐺
1 2𝐻
𝑌2 (𝑠) 𝐺2 𝐺2
= from which 𝑌2 (𝑠) = 𝐷(𝑠) ( )
𝐷(𝑠) 1+𝐺1 𝐺2 𝐻 1+𝐺1 𝐺2 𝐻
When both inputs are present, 𝑌(𝑠) = 𝑌1 (𝑠) + 𝑌2 (𝑠)
𝐺1 𝐺2 2 𝐺 2 𝐺
𝑌(𝑠) = 𝑋(𝑠) ( ) + 𝐷(𝑠) ( ) = ( ) {𝐺1 𝑋(𝑠) + 𝐷(𝑠)}
1+𝐺1 𝐺2 𝐻 1+𝐺 𝐺 𝐻1 2 1+𝐺 𝐺 𝐻 1 2
Consider the case |𝐺1 𝐻| ≫ 1 and |𝐺1 𝐺2 𝐻| ≫ 1 (Modulii of the corresponding complex variable)
𝐺2 𝐺2 1
𝑌2 (𝑠) = 𝐷(𝑠) ( ) ≅ 𝐷(𝑠) 𝐺 ≅ 𝐷(𝑠) ≅0 (As |𝐺1 𝐻| ≫ 1, 1
𝐺1 𝐻
≅ 0)
1+𝐺1 𝐺2 𝐻 1 𝐺2 𝐻 𝐺1 𝐻
That is, the effect of the disturbance is suppressed. This is an advantage of feedback systems.
𝐺1 𝐺2 𝐺 𝐺 1
Now, 𝑌1 (𝑠) = 𝑋(𝑠) ( ) ≅ 𝑋(𝑠) 1 2 ≅ 𝑋(𝑠)
1+𝐺1 𝐺2 𝐻 𝐺 𝐺 𝐻 1 2 𝐻
𝑌1 (𝑠)
This means that if |𝐺1 𝐺2 𝐻| ≫ 1, then the closed loop transfer function becomes
𝑋(𝑠)
independent of 𝐺1 (𝑠) 𝑎𝑛𝑑 𝐺2 (𝑠) and becomes inversely proportional to 𝐻(𝑠). This is another
advantage of closed loop system. If we have a unity feed back system where 𝐻(𝑠) = 1, then the
actual ouput becomes equal to the input (which is the desired output).

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
Exercise 20
20.1
The figure shows a closed loop system with reference input R(s) and disturbance input D(s).
Obtain the expression for the output C(s) when both are present.
D(s)

+
R(s) + G2(s) C(s)
G1(s) +
-

20.2
Obtain the transfer function Y(s)/X(s) and Y(s)/D(s) for the system shown below.

D(s)

X(s) + Y(s)
+ + G2 G3
- Gc G1 +
-
H1

H2

20.3
Obtain the expression for steady state error for the system shown if both R(s) and D(s) are
present.

D(s)

+ C(s)
R(s) +
G1 G2
+
-

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
AUTOMATIC CONTROLLERS
Industrial controllers may be classified according to their control actions as
i) On-off controllers
ii) Proportional controllers
iii) Integral controllers
iv) Proportional-plus-Integral controllers
v) Proportional-plus-Differrential controllers
vi) Proportional-plus-Integral-plus-Differential Cntrollers
Most of the industrial controllers use electricity or pressurised fluids like oil or air as power
sources. Accordingly, controllers may also be classified as electronic controllers, hydraulic
controllers or pneumatic controllers. The choice of the controller is based on the nature of plant
and the operating conditions, safety, cost, reliability, availability, accuracy, weight, size
On-off controllers
+ e U1
In on-off control (two position control) system, the u
actuating element has only two fixed positions, - U2
commonly on and off. Let the actuating error signal be
e(t) and the output signal from the controller be u(t).
In two position control, the signal u(t) remains at either maximum or minimum value, depending
on whether the actuating error signal is positive or negative.
𝑢(𝑡) = 𝑈1, for 𝑒(𝑡) > 0
= 𝑈2, for 𝑒(𝑡) < 0
Where 𝑈1, and 𝑈2, are constants. The minimum value 𝑈2 is either zero or - 𝑈1. On-off controllers
are generally electrical devices, and an electric solenoid operated valve is widely used in such
controllers.
Proportional cotrollers e
+ KP
u
For a controller with proportional control action, the -
relationship between the output of the controller 𝑢(𝑡)
and the actuating error signal 𝑒(𝑡) is
𝑢(𝑡) = 𝐾p 𝑒(𝑡) or
𝑈(𝑠)
𝐸(𝑠)
= 𝐾𝑃 where 𝐾 p is called proportional gain. (Please
note that this is different from the position error
constant 𝐾 p)
What ever be the actual mechanism may be or whatever the form of operating power may be,
the proportional controller is essentially an amplifier with adjustable gain.
Integral controllers + e 𝐾𝑖
u
In a controller with integral control action, the value of the -
𝑠

controller output 𝑢(𝑡) is changed at a rate proportional to the

𝑑
actuating error signal 𝑒(𝑡), 𝑑𝑡
𝑢(𝑡) = 𝐾𝑖 𝑒(𝑡)

𝑢(𝑡) = 𝐾𝑖 ∫ 𝑒(𝑡) 𝑑𝑡
𝑈(𝑠) 𝐾𝑖
The transfer function of the inegral controller 𝐸(𝑠)
= 𝑠

Where 𝐾i is an adjustable constant

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2

Proportional-plus-Integral (PI)controller
The control action of a PI controller is defined by + e 𝐾𝑖
𝐾𝑝 + u
- 𝑠
𝑢(𝑡) = 𝐾𝑃 𝑒(𝑡) + 𝐾𝑖 ∫ 𝑒(𝑡) 𝑑𝑡
𝑈(𝑠) 𝐾𝑖 1
= 𝐾𝑝 + = 𝐾𝑝 (1 + )
𝐸(𝑠) 𝑠 𝑇𝑖 𝑠
Where
𝐾𝑃
= 𝐾𝑖 and 𝑇𝑖 is known as integral time. Fig. 4.4
𝑇𝑖

Proportional-plus-derivative (PD) controller

+ e 𝐾𝑝 + 𝐾𝑑 𝑠 u
The control action of a PD controller is defined by
𝑑 -
𝑢(𝑡) = 𝐾𝑃 𝑒(𝑡) + 𝐾𝑑 𝑒(𝑡)
𝑑𝑡
𝑈(𝑠)
= 𝐾𝑃 + 𝐾𝑑 𝑠 = 𝐾𝑃 (1 + 𝑇𝑑 𝑠)
𝐸(𝑠)
Fig. 4.5
Where 𝐾𝑃 𝑇𝑑 = 𝐾𝑑 and 𝑇𝑑 is called derivative time.

Proportional-plus-integral-plus-derivative (PID) control

PID control action is a combination of proportional
+ e 𝐾𝑖 u
control action, inegral control action, and derivative 𝐾𝑝 + + 𝐾𝑑 𝑠
- 𝑠
control action
𝑑
𝑢(𝑡) = 𝐾𝑃 𝑒(𝑡) + 𝐾𝑖 ∫ 𝑒(𝑡) 𝑑𝑡 +𝐾𝑑 𝑒(𝑡)
𝑑𝑡
𝑈(𝑠) 𝐾𝑖 1
= 𝐾𝑝 + +𝐾𝑑 𝑠 = 𝐾𝑝 (1 + +𝑇 𝑠) Fig. 4.6
𝐸(𝑠) 𝑠 𝑇𝑖 𝑠 𝑑

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
We can get these control actions electronically, pneumatically, or hydraulically. Let us discuss
how we can get these control actions electronically using operational amplifiers (op amps). You
are requested to refer the text book Modern Control Systems by K Ogata for pneumatic and
hydraulic control actions.
Operational Amplifiers (op amps) are used to amplify signals in sensor circuits. They are also
frequently used in filters used for compensation purposes.
The figure shows an op amp. Usually the ground is chosen as -
0 𝑉 and the input voltages 𝑒1 and 𝑒2 are measured relative
e1 -
to the ground. The input voltage 𝑒1 is inverted at the minus
terminal and 𝑒2 is not inverted. The total input to the
e2 + e0
amplifier is 𝑒2 − 𝑒1 . Therefore, the output voltage
𝑒0 = 𝐾(𝑒2 − 𝑒1 ), where 𝐾 is the differetial gain (voltage
gain).
The magnitude of 𝐾 is approximately of the order of 105 for dc signals and low frequency (less
than 10 𝐻𝑧) ac signals. The value of 𝐾 decreases with increase in frequecny and it is
approximately equal to 1 for frequencies above 1 𝑀𝐻𝑧. Since op amps amplifies the difference
in voltages, it is also called differential amplifier. Since the gain of the op amp is very high, it is
necessary to have negative feed back from the output to input to make the amplifier stable. The
feed back is made from the output to the negative input so that the feedback becomes negative.
In an ideal op amp, no current flows in to the input terminals and the output voltage is not
affected by the load connected to the output terminal. In otherwords, the input impedence of
op amp is infinity and the output impedence is zero. In actual case, this may not be exactly true.
Inverting amplifier
Let us consider an op amp circuit as shown. The R2
positive terminal is connected to the gound. Let e’ be
i2 -
the potential at the branching point for i2.
i1 R1
-
Then, e’
ei + e0
′ ′
𝑒𝑖 − 𝑒 𝑒 − 𝑒𝑜
𝑖1 = , 𝑖2 =
𝑅1 𝑅2

Since for an ideal op amp, negligible current flows in to the amplifier, 𝑖1 = 𝑖2

𝑒𝑖 − 𝑒 ′ 𝑒 ′ − 𝑒𝑜
= … … … … … … … . (1)
𝑅1 𝑅2
For op amps, 𝑒𝑜 = 𝐾(𝑒2 − 𝑒1 ) = 𝐾(0 − 𝑒 ′ ) = −𝐾𝑒 ′
Since 𝐾 is usually very large, 𝑒 ′ should be very small (close to zero) so that their product gives
a reasonable value for 𝑒𝑜 . Hence equation (1) becomes
𝑒𝑖 −𝑒𝑜
=
𝑅1 𝑅2
𝑅2
𝑒𝑜 = − 𝑒𝑖 … … … . . (2)
𝑅1
𝑅
The output is negative. Thus the circuit shown is an inverting amplifier with gain (𝑅2 ).
1

If 𝑅1 = 𝑅2 , then the op amp circuit shown acts as an inverter.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

Noninverting amplifier
R2
Look at the circuit given in figure. Here the
i2 -
negative terminal is connected to the ground. i1 R1
The input voltage 𝑒𝑖 is given to the positive -
terminal with respect to the ground. e’
𝑖1 =
0 − 𝑒′
, 𝑖2 =
𝑒 ′ − 𝑒𝑜 ei + e0
𝑅1 𝑅2
′
0−𝑒 𝑒 ′ − 𝑒𝑜
𝑖1 = 𝑖2 ≫≫≫ =
𝑅1 𝑅2
−𝑅2 𝑒 ′ = 𝑅1 (𝑒 ′ − 𝑒𝑜 ) ≫≫≫ (𝑅1 + 𝑅2 )𝑒 ′ = 𝑅1 𝑒𝑜 The circuit can be also drawn as shown in
𝑅1 figure -
𝑒′ = 𝑒
(𝑅1 + 𝑅2 ) 𝑜
But, 𝑒𝑜 = 𝐾(𝑒2 − 𝑒1 ) +
𝑅1
∴ 𝑒𝑜 = 𝐾(𝑒𝑖 − 𝑒 ′ ) = 𝐾 (𝑒𝑖 − R2
𝑒 )
𝑅1 + 𝑅2 𝑜 ei -
i2 e0
𝑒𝑜 + 𝐾
𝑅1
𝑒 = 𝐾𝑒𝑖 e ’
(𝑅1 + 𝑅2 ) 𝑜 R1
i1
1 𝑅1
𝑒𝑜 ( + ) = 𝑒𝑖
𝐾 𝑅1 + 𝑅2

1
Since 𝐾 is very large, 𝐾 can be neglected. Hence,
𝑅1 + 𝑅2 𝑅2
𝑒𝑜 = ( ) 𝑒𝑖 = (1 + ) 𝑒𝑖 … … … … … (3)
𝑅1 𝑅1
𝑅
Now the circuit acts as a noninverting amplifier with gain (1 + 𝑅2 ).
1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5
Similar to the case of electrical circuits discussed above, we can also apply the impedence
approach to the op amp circits to obtain their transfer functions
Consider the figure shown. 𝑍1 and 𝑍2 are the Z2
impedences connected. We can write
i2 -
𝑒𝑖 − 𝑒 ′ 𝑒 ′ − 𝑒𝑜 i1 Z1
𝑖1 = , 𝑖2 = -
𝑍1 𝑍2
Hence we can write 𝑒𝑜 = − 𝑍2 𝑒𝑖
𝑍 e’
ei + e0
1
𝑍2 (𝑠)
𝐸𝑜 (𝑠) = − 𝐸 (𝑠) … … … … … . (4)
𝑍1 (𝑠) 𝑖
If we have only a resister R connected across a voltage 𝑉, then 𝑉 = 𝑖𝑅 R
Taking Laplace Transform, 𝑉(𝑠) = 𝐼(𝑠) 𝑅 = 𝐼(𝑠) 𝑍(𝑠) where 𝑍(𝑠) = 𝑅……….(5a) v
1
If we have only a capacitor C connected across a voltge, 𝑉 = 𝐶 ∫ 𝑖 𝑑𝑡 C
1 𝐼(𝑠) 1
Taking Laplace transform, 𝑉(𝑠) = = 𝐼(𝑠)𝑍(𝑠) where 𝑍(𝑠) = ………….(5b) v
v
𝐶 𝑠 𝐶𝑠
1
If we have a resister and a capacitor in series 𝑉 = 𝑉𝑅 + 𝑉𝐶 = 𝑖𝑅 + 𝐶 ∫ 𝑖 𝑑𝑡
1 𝐼(𝑠) 𝑅𝐶 𝑠+1
Taking Laplace transform, 𝑉𝑜 (𝑠) = 𝑅𝐼(𝑠) + 𝐶 𝑠 = 𝐶𝑠 𝐼(𝑠) = 𝐼(𝑠)𝑍(𝑠) R C
1+𝑅𝐶 𝑠
Where 𝑍(𝑠) = 𝐶𝑠 ………………….(5c) v
If we have a resistor and a capacitor in parallel across a voltage V, then we
𝑉 𝑑𝑣
can write 𝐼 = I𝑅 + I𝐶 = +𝐶 R
𝑅 𝑑𝑡

Taaking Laplace Transform, 𝐼(𝑠) =

𝑉(𝑠)
+ 𝐶𝑠 𝑉(𝑠) = (
1+𝑅𝐶 𝑠
) 𝑉(𝑠) C
𝑅 𝑅
𝑅
𝑉(𝑠) = (1+𝑅𝐶 𝑠) 𝐼(𝑠) = 𝐼(𝑠)𝑍(𝑠) where 𝑍(𝑠) = 1+𝑅𝐶 𝑠 …………………(5d)
𝑅
v

Proportional Controller
Consider the op amp circuit shown in figure where two op amps are coneected in series. Let
𝑒 be the output voltage of op amp 1, 𝑒𝑖 the input voltage, and 𝑒𝑜 the output voltage.
Z2
R2 Z4
Z1 -
R1 Z3 R4
- -
R3
-
ei + e + e0

𝑅2 𝑅4
From equation (2), we can 𝑒 = − 𝑒, and 𝑒𝑜 = − 𝑒
𝑅1 𝑖 𝑅3
𝑅4 𝑅2 𝑅4 𝑅2
𝑒𝑜 = − ( ) (− ) 𝑒𝑖 = 𝑒
𝑅3 𝑅1 𝑅3 𝑅1 𝑖
𝐸0 (𝑠) 𝑅 𝑅
Taking Laplace transform 𝐸𝑖 (𝑠)
= 𝑅4 𝑅2 = 𝐾𝑝 , the adjustable constant
3 1

We get a proportional controller.

𝐸0 (𝑠) 𝑍 (𝑠) 𝑍 (𝑠) 𝑅 𝑅
If we use the impedence approach, 𝐸𝑖 (𝑠)
= 𝑍4 (𝑠) 𝑍2 = 𝑅4 𝑅2 = 𝐾𝑝 where 𝑍1(𝑠) = 𝑅1, 𝑍2(𝑠) = 𝑅2,
3 1(𝑠) 3 1

𝑍3(𝑠) = 𝑅3 , 𝑍4(𝑠) = 𝑅4. (Refer eq. 5a-d)

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6

Integral Controller
Consider the op amp circuit shown. The resistance R2 is replaced by a capacitance C2. Let us
1
consider the impedence approach. 𝑍1(𝑠) = 𝑅1, 𝑍2(𝑠) = 𝐶 𝑠, 𝑍3(𝑠) = 𝑅3 , 𝑍4(𝑠) = 𝑅4 (Refer eq.
2
5a-d)

Z2
C2 Z4
Z1 -
R1 Z3 R4
- -
R3
-
ei + e + e0

Therefore from equation 4,

𝑍2 (𝑠) 𝑍4 (𝑠)
𝐸(𝑠) = − 𝐸𝑖 (𝑠), 𝐸𝑜 (𝑠) = − 𝐸(𝑠)
𝑍1 (𝑠) 𝑍3 (𝑠)
𝑍4 (𝑠) 𝑍2 (𝑠) 𝑅4 1 1
𝐸𝑜 (𝑠) = − (− )𝐸𝑖 (𝑠) = 𝐸 (𝑠)
𝑍3 (𝑠) 𝑍1 (𝑠) 𝑅3 𝑅1 𝐶2 𝑠 𝑖
𝑅4
𝐸0 (𝑠) 𝑅4 1 1 𝑅3 𝑅1 𝐶2 𝐾𝑖
= =( )=
𝐸𝑖 (𝑠) 𝑅3 𝑅1 𝐶2 𝑠 𝑠 𝑠

Proportional-plus-Integral Controller
Consider the op amp circuit shown. The resistance R2 is in series with a capacitance C2. Let
1+𝑅2 𝐶2 𝑠
us consider the impedence approach. 𝑍1(𝑠) = 𝑅1, 𝑍2(𝑠) = , 𝑍3(𝑠) = 𝑅3 , 𝑍4(𝑠) = 𝑅 4
𝐶2 𝑠
(Refer Eq. 5a-d)
Z2
R2 C2 Z4
Z1
R1 Z3 R4
- -
R3
-
ei + e + e0

Therefore from equation 4,

𝑍2 (𝑠) 𝑍4 (𝑠)
𝐸(𝑠) = − 𝐸 (𝑠), 𝐸𝑜 (𝑠) = − 𝐸(𝑠)
𝑍1 (𝑠) 𝑖 𝑍3 (𝑠)
𝑍4 (𝑠) 𝑍2 (𝑠) 𝑅4 1 + 𝑅2 𝐶2 𝑠
𝐸𝑜 (𝑠) = − (− )𝐸 (𝑠) =
)( ) 𝐸𝑖 (𝑠)
(
𝑍3 (𝑠) 𝑍1 (𝑠) 𝑖
𝑅3 𝑅1 𝐶2 𝑠
𝑅4
𝐸0 (𝑠) 𝑅4 1 + 𝑅2 𝐶2 𝑠 𝑅2 𝑅4 𝑅3 𝑅1 𝐶2 𝐾𝑖
= ( )( )=( + ) = 𝐾𝑃 +
𝐸𝑖 (𝑠) 𝑅3 𝑅1 𝐶2 𝑠 𝑅1 𝑅3 𝑠 𝑠

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7

Proprtional-plus-Derivative controller
Consider the op amp circuit shown. The resistance R1 is in series with a capacitance C1. Let
𝑅
us consider the impedence approach. 𝑍1(𝑠) = 1+𝑅 1𝐶 𝑠, 𝑍2(𝑠) = 𝑅2 , 𝑍3(𝑠) = 𝑅3 , 𝑍4(𝑠) = 𝑅4.
1 1
(Refer Eq. 5a-d)

Z2 Z4
Z1 C1 R2
-
R1 Z3 R4
- -
R3
-
ei + e + e0

Therefore from equation 4,

𝑍2 (𝑠) 𝑍4 (𝑠)
𝐸(𝑠) = − 𝐸 (𝑠), 𝐸𝑜 (𝑠) = − 𝐸(𝑠)
𝑍1 (𝑠) 𝑖 𝑍3 (𝑠)

𝑍4 (𝑠) 𝑍2 (𝑠) 𝑅4 𝑅2
𝐸𝑜 (𝑠) = − (− )𝐸𝑖 (𝑠) = ( ) ( ) 𝐸𝑖 (𝑠)
𝑍3 (𝑠) 𝑍1 (𝑠) 𝑅3 𝑅1
1 + 𝑅 1 𝐶1 𝑠
𝐸0 (𝑠) 𝑅4 1 + 𝑅1 𝐶1 𝑠 𝑅2 𝑅4 𝑅2 𝑅4 𝐶1
= ( ) 𝑅2 ( )=( + 𝑠) = 𝐾𝑃 + 𝐾𝑑 𝑠
𝐸𝑖 (𝑠) 𝑅3 𝑅1 𝑅1 𝑅3 𝑅3
Proprtional-plus-Integral-plus-Derivative controller
𝑅 1+𝑅2 𝐶2 𝑠
𝑍1(𝑠) = 1+𝑅 1𝐶 𝑠, 𝑍2(𝑠) = , 𝑍3(𝑠) = 𝑅3 , 𝑍4(𝑠) = 𝑅4 (Refer Eq. 5a-d)
1 1 𝐶2 𝑠

Z2 Z4
Z1 C1 R2 C2
-
R1 Z3 R4
- -
R3
-
ei + e + e0

Therefore from equation 4,

𝑍 (𝑠) 𝑍 (𝑠)
𝐸(𝑠) = − 𝑍2 (𝑠) 𝐸𝑖 (𝑠), 𝐸𝑜 (𝑠) = − 𝑍4 (𝑠) 𝐸(𝑠)
1 3
1+𝑅2 𝐶2 𝑠
𝑍4 (𝑠) 𝑍 (𝑠) 𝑅4 𝐶 𝑠
𝐸𝑜 (𝑠) = − (− 2 (𝑠))𝐸𝑖 (𝑠) = ( ) ( 𝑅21 ) 𝐸𝑖 (𝑠)
𝑍3 (𝑠) 𝑍1 𝑅3
1+𝑅1 𝐶1 𝑠

𝐸0 (𝑠) 𝑅4 1 1 𝑅4 1 𝑅2 𝐶1
= ( )( + 𝑅2 ) ( + 𝐶1 𝑠) = ( ) ( + + + 𝑅2 𝐶1 𝑠)
𝐸𝑖 (𝑠) 𝑅3 𝐶2 𝑠 𝑅1 𝑅3 𝑅1 𝐶2 𝑠 𝑅1 𝐶2
𝑅4
𝑅 𝑅 𝐶 𝑅 𝐶 𝑅3 𝑅4 𝑅2 𝐶1 𝐾𝑖
= (𝑅4 ) (𝑅2 + 𝐶1 ) + 1 𝑠2 + 𝑅3
𝑠 = 𝐾𝑃 + 𝑠
+ 𝐾𝑑 𝑠
3 1 2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
MODELLING IN STATE SPACE
Complex systems may have multiple inputs and multiple outputs and may be time varying. Because
of the necessity of meeting increasingly stringent requirements on the performance of control
systems, a new approach has been developed in the sixties. The advent of computers also helped
its development. This is known as the modern control theory
The difference between the modern control theory and the conventional (classical) control
theory is that the former is applicable to multiple-input-multiple-output systems which may be
linear or nonlinear, time invariant or time varying, while the latter is applicable only to linear,
time invariant, single-input-single-output systems. Also the modern control theory is essentially
a time-domain approach, while the classical control theory is a complex frequency-domain
approach.
The new approach is based on the concept of state.
State: The state of a dynamic system is the smallest set of variables called state variables such
that the knowledge of these variables at 𝑡 = 𝑡0 together with the knowledge of the input for 𝑡 ≥
𝑡0 completely define the behaviour of the system for any time 𝑡 ≥ 𝑡0 .
State variables: The state variables of a dynamic system are the smallest set of variables that
determines the state of the dynamic system. If 𝑛 vaiables 𝑥 1, 𝑥 2, ………….., 𝑥 n are needed to
completely describe the behavior of a dynamic system, then such 𝑛 vaiables are a set of state
variables. Note that state variables need not generally be physically measurable or observable
quantities.
State vector: If 𝑛 state vaiables are needed to completely define the behavior of a given system,
then these n state variables can be considered as the n component of vector 𝒙. Such a vector
is called state vector.
𝒙 = [ 𝑥1, 𝑥2, … … … … . . , 𝑥n]
State space: The n-dimensional space whose coordinate axes consist of the 𝑥 1 axis, 𝑥 2 axis,
………….. 𝑥 n axis, where 𝑥 1, 𝑥 2, ………….., 𝑥 n are state variables, is called a state space.
In state space analysis we are concerned with three types of variables: input variables, output
variables, and state variables. Also remember that the state space representation of a given
system is not unique. But the number of state variables will be the same in different state space
models of a given sysytem.
Consider the second order system shown. We know the differential
equation governing this system is 𝑚𝑦̈ + 𝑏𝑦̇ + 𝑘𝑦 = 𝑢. It is a second u(t)
order differential equation. This means that the system involves two y(t)
integrators. Let us define the state variables 𝑥1 (𝑡), and 𝑥2 (𝑡)
𝑥1 (𝑡) = 𝑦(𝑡)
𝑥2 (𝑡) = 𝑦̇ (𝑡)
Then we obtain 𝑥2 = 𝑥̇ 1
𝑏 𝑘 1
Then we can re-write the governing equation as 𝑦̈ = − 𝑚 𝑦̇ − 𝑚 𝑦 + 𝑚 𝑢
𝑘 𝑏 1
𝑥̇ 2 = − 𝑥1 − 𝑥2 + 𝑢
𝑚 𝑚 𝑚
The output is 𝑦 = 𝑥1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
Now we have the state equations
𝑥̇ 1 = 𝑥2
𝑘 𝑏 1
𝑥̇ 2 = − 𝑥1 − 𝑥2 + 𝑢
𝑚 𝑚 𝑚
The second order differential equation is converted in to two first order equations.
We can write these equations in the vector-matrix form as
0 1 𝑥1 0
𝑥̇ 1 𝑘 𝑏 1
[ ] [ = ] [𝑥 ] [ ] 𝑢
+
𝑥̇ 2 − − 2
𝑚 𝑚 𝑚
The output equation is
𝑥1
𝑦 = [1 0] [𝑥 ]
2
This can be written in the standard form as
[𝒙̇ ] = [𝑨][𝒙] + [𝑩]𝒖
𝒚 = [𝑪][𝒙] + [𝑫]𝒖
0 1 0 𝑥
where [𝑨] = [ 𝑘
− 𝑚 − 𝑚],
𝑏 [𝑩] = [ 1 ], [𝑪] = [1 0], [𝑫] = 0, [𝒙] = [𝑥1 ] is the state vector, 𝑢 is
𝑚 2

the input, and 𝑦 is the output.

We can represent the state equations as block diagram shown below.

u 1 + 𝑥̇ 2 𝑥2
‫׬‬.
𝑥1 = 𝑦
‫׬‬.
𝑚
-
+ 𝑏
𝑚
+
𝑘
𝑚

Example 1
Obtain the state space respresentation of the sysytem whose transfer function is
𝑌(𝑠) 24
𝑅(𝑠)
= 𝑠3 +9𝑠2 +26𝑠+24

From the transfer function, we can write 𝑌(𝑠)(𝑠 3 + 9𝑠 2 + 26𝑠 + 24) = 24 𝑅(𝑠). The differential
equation corresponding to this is 𝑦⃛ + 9𝑦̈ + 26𝑦̇ + 24𝑦 = 24 𝑟
𝑦⃛ = −9𝑦̈ − 26𝑦̇ − 24𝑦 + 24 𝑟
The equation is of order three. Hence, there should be three state variables. Let us define the
state variables as 𝑥1 , 𝑥2 , 𝑥3 sothat
𝑦 = 𝑥1
𝑥̇ 1 = 𝑥2 = 𝑦̇
𝑥̇ 2 = 𝑥3 = 𝑦̈
𝑥̇ 3 = 𝑦⃛ = −9𝑥3 − 26𝑥2 − 24𝑥1 + 24 𝑟
That is,
𝑥̇ 1 = 𝑥2
𝑥̇ 2 = 𝑥3
𝑥̇ 3 = −24𝑥1 − 26𝑥2 − 9𝑥3 + 24 𝑟
And the output is 𝑦 = 𝑥1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
The third order differential equation is reduced to three first order differential equations.
These equations can be represented in a block diagram as shown below.

We can write this in matrix form as

𝑥̇ 1 0 1 0 𝑥1 0
[𝑥̇ 2 ] = [ 0 0 1 ] [ 𝑥2 ] + [ 0 ] 𝑟
𝑥̇ 3 −24 −26 −9 𝑥3 24
𝑥1
Output is 𝑦 = [1 0 0] [ 𝑥2 ]
𝑥3
This can be written in the standard form as
[𝒙̇ ] = [𝑨][𝒙] + [𝑩]𝒓
𝒚 = [𝑪][𝒙] + [𝑫]𝒓
0 1 0 0 𝑥1
where [𝑨] = [ 0 0 1 ], [𝑩] = [ 0 ], [𝑪] = [1 0 0], [𝑫] = 0, [𝒙] = [ 2 ] is the state
𝑥
−24 −26 −9 24 𝑥3
vector, r is the input, and y is the output.
Example 2
Obtain the state space representation for the b
system shown.

The equations of motion can be written as

𝑥̈ 1 𝑥̈ 2
𝑚1 𝑥̈ 1 = −𝑏𝑥̇ 1 + 𝑘(𝑥2 − 𝑥1 ) 𝑘(𝑥2 − 𝑥1 )
𝑚2 𝑥̈ 2 = 𝑓(𝑡) − 𝑘(𝑥2 − 𝑥1 ) 𝑚1 𝑚2 𝑓(𝑡)
𝑚1 𝑥̈ 1 + 𝑏𝑥̇ 1 + 𝑘𝑥1 − 𝑘𝑥2 = 0 𝑏𝑥̇ 1
𝑚2 𝑥̈ 2 + 𝑘𝑥2 − 𝑘𝑥1 = 𝑓(𝑡)
Now we have two second order differential In the matrix form
equations. Let us assume four state 0
1
0
𝑘 𝑘 0
variables(𝑦1 , 𝑦2 , 𝑦3 , 𝑦4 ) two each for each 𝑦̇1 −
𝑚1 −
𝑏
𝑚1
0 𝑦1
0
𝑦̇ 0 𝑦
equation. 𝑦1 = 𝑥1 and 𝑦3 = 𝑥2 [ 2] = 0
𝑦̇ 3
𝑚1 0 1 [ 2] + 0 𝑓
𝑦
𝑘 0 𝑘 0 𝑦3 1
𝑦̇1 = 𝑦2 𝑦̇4 0 − 4
[𝑚2 ]
𝑚2 𝑚2
𝑘 𝑏 𝑘 [ ]
𝑦̇ 2 = − 𝑦1 − 𝑦2 + 𝑦
𝑚1 𝑚1 𝑚1 3 [𝒚̇ ] = [𝑨][𝒚] + [𝑩]𝒇
𝑦̇ 3 = 𝑦4 𝑦1
𝑥 1 0 0 0 𝑦2
the outputs, [𝑥 ] = [
𝑘 𝑘 1 1
][ ]
𝑦̇4 = 𝑦1 − 𝑦3 + 𝑓(𝑡) 2 0 0 1 0 𝑦3
𝑚2 𝑚2 𝑚2 𝑦4

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

Example 3
Obtain the state space respresentation of the system
𝐶(𝑠) 𝑠2 +7𝑠+2
whose transfer function is 𝐺(𝑠) = 𝑅(𝑠) = 𝑠3 +9𝑠2 +26𝑠+24

This problem is different from the earlier ones since the numerator has a polynomial in s instead
of a constant.
Separate the system in to two cascaded blocks
as shown. The first block contains the
denominator and the second one the
numerator.
Let us find out the state space representation of the first block.
𝑌(𝑠) 1
= 3
𝑅(𝑠) 𝑠 + 9𝑠 2 + 26𝑠 + 24
(𝑠 3 + 9𝑠 2 + 26𝑠 + 24) 𝑌(𝑠) = 𝑅(𝑠). By taking the inverse transform, the governing equation is
𝑦⃛ + 9𝑦̈ + 26𝑦̇ + 24𝑦 = 𝑟
𝑦⃛ = −24𝑦 − 26𝑦̇ − 9𝑦̈ + 𝑟
We can consider the state variables as 𝑥1 , 𝑥2 , 𝑥3 where 𝑥1 = 𝑦, 𝑥2 = 𝑥1̇ and 𝑥3 = 𝑥2̇ . Then,
𝑥3̇ = −24𝑥1 − 26𝑥2 − 9𝑥3 + 𝑟
The state equations are
𝑥1̇ = 𝑥2 𝑥1̇ 0 1 0 𝑥1 0 The output is 𝑦 = 𝑥1
𝑥2̇ = 𝑥3 [𝑥2̇ ] = [ 0 0 1 ] [𝑥2 ] + [0] 𝑟 𝑥1
𝑥3̇ −24 −26 −9 𝑥3 1 𝑦 = [1 0 0] [𝑥2 ]
𝑥3̇ = −24𝑥1 − 26𝑥2 − 9𝑥3 + 𝑟
𝑥3
This can be represented by the block
diagram as shown. Now considering the
second block, we can write
𝐶(𝑠) = (𝑠 + 7𝑠 + 2) 𝑌(𝑠). That is,
2

𝑦̈ + 7𝑦̇ + 2𝑦 = 𝑐 .
𝑥3 + 7𝑥2 + 2𝑥1 = 𝑐

The final output is

𝑥1
𝑥
𝑐 = [ 2 7 1] [ 2 ]
𝑥3
The figure shows the block diagram
representation of the complete state
space model.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

Exercise 21
21.1
Obtain the state space representation of the
electrical network shown. The output is 𝑣0 (𝑡)

21.2
Obtain the state space representation of the system with transfer function
𝐶(𝑠) 2𝑠+1
𝐺(𝑠) = 𝑅(𝑠)
= 𝑠2 +7𝑠+9

21.3
Obtain the state space representation of the
system shown.

Correlation Between Transfer Functions and State Space Equations

Let us see how to derive the transfer function of a single-input-single -outputsystem from the
𝑌(𝑠)
state space equations. Consider a system whose transfer function is given by 𝐺(𝑠) = 𝑈(𝑠). The
state space equations of this system can be written as
[𝒙̇ ] = [𝑨][𝒙] + [𝑩]𝒖
𝒚 = [𝑪][𝒙] + 𝐷𝒖
Where [𝒙] is the state vector, 𝒖 is the input, and 𝒚 is the output.
Taking the Laplace Transforms of the above equations and setting all intial conditions to zero
𝒔[𝑿](𝒔) = [𝑨][𝑿](𝒔) + [𝑩]𝑼(𝒔)
𝒀(𝒔) = [𝑪][𝑿](𝒔) + 𝐷𝑼(𝒔)
[𝑿](𝒔) {𝒔𝑰 − [𝑨]} = [𝑩]𝑼(𝒔) ≫≫≫ [𝑿](𝒔) = {𝒔𝑰 − [𝑨]}−𝟏 [𝑩]𝑼(𝒔)
𝒀(𝒔) = [𝑪]{𝒔𝑰 − [𝑨]}−𝟏 [𝑩]𝑼(𝒔) + 𝐷𝑼(𝒔) ≫≫≫ 𝒀(𝒔) = {[𝑪](𝒔𝑰 − [𝑨])−𝟏 [𝑩] + 𝐷}𝑼(𝒔)
𝑌(𝑠)
𝐺(𝑠) = = [𝑪](𝒔𝑰 − [𝑨])−𝟏 [𝑩] + 𝐷
𝑈(𝑠)
Let us take the example of the spring mass system considered earlier. The state space
representation of the system is
0 1 𝑥 0
𝑥̇ 𝑏 ] [ 1] + [ 1 ] 𝑢
[ 1] = [ 𝑘 𝑥2
𝑥̇ 2 − −
𝑚 𝑚 𝑚
𝑥1
The output equation is 𝑦 = [1 0] [𝑥 ]
2
0 1 0 𝑥1
where[𝑨] = [− 𝑘 − 𝑏 ], [𝑩] = [ 1 ], [𝑪] = [1 0], [𝑫] = 0, [𝒙] = [𝑥 ] is the state vector, u is the
𝑚 𝑚 𝑚 2

input, and y is the output. For simplicity, let us assume some numerical values for m,b, and k.
0 1 0
m=2, b=3, k=5. Hence, [𝑨] = [− 5 − 3], [𝑩] = [ 1 ],
2 2 2

𝑠 0 0 1 −1 0
𝐺(𝑠) = [𝑪](𝒔𝑰 − [𝑨])−𝟏 [𝑩] + 𝐷 = [1 0] {[ ] − [− 5 − 2]} [ 2 ]
3 1
0 𝑠 2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6
3
𝑠+2 1 0
[1 0] [ ] [1] 0
𝑠 −1 −1 0 5 [𝑠 + 3 1] [1] 1
− 𝑠 2 2 1
= [1 0] {[5 3]} [1] = 2 = 2 = 2 = 2
𝑠+ 3 5 3 5 3 5
2 2 2 𝑠2 + 2 𝑠 + 2 𝑠2 + 2 𝑠 + 2 𝑠 2 + 2 𝑠 + 2 2𝑠 + 3𝑠 + 5
1 1
𝐺(𝑠) = 2 = 2
2𝑠 + 3𝑠 + 5 𝑚𝑠 + 𝑏𝑠 + 𝑘

Exercise 22
22.1
Convert the state and output equation to a transfer function
𝑥̇ −4 −1.5 𝑥1 2
[ 1] = [ ] [𝑥 ] + [ ] 𝑢
𝑥̇ 2 4 0 2 0
𝑥1
𝑦 = [1.5 0.625] [𝑥 ]
2
22.2
Obtain the state space representation of the system defined by
𝑥1̇ −1 1 0 𝑥1 0 𝑥1
[𝑥2̇ ] = [ 0 −1 1 ] [𝑥2 ] + [0] 𝑟 and 𝑦 = [1 0 0] [𝑥2 ]
𝑥3̇ 0 0 −2 𝑥3 1 𝑥3
22.3
Using Matlab/Scilab script, convert the given transfer function to state space
representation
2𝑠 2 + 8𝑠 + 6
𝐺(𝑠) =
𝑠 3 + 8𝑠 2 + 16𝑠 + 6
22.4
Using Matlab/Scilab script obtain the unit impulse and unit step response of the following
system
𝑥̇ 0 1 𝑥1 0
[ 1] = [ ][ ] + [ ]𝑢
𝑥̇ 2 −1 −1 𝑥2 1
𝑥1
𝑦 = [1 0] [𝑥 ] + [0]𝑢
2

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
STABILITY
The three requirements that enter in to the design of a control system are transient response,
steady state error, and stability. Of the three, stability is the most important system
specification. If a system is unstable, the transient response and steady state errors are of no
consequence.
A linear-time-invarient system is stable if every bounded input yields a bounded output.
The total respsonse of a system = Natural response + forced response
• A LTI system is stable if the natural response → 0 when time → ∞.
• A LTI system is unstable if the natural response grows unbounded when time → ∞.
• A LTI system is marginally stable if the natural response neither decays nor grows but
remain constant or oscillates as time → ∞.
If all the closed loop poles of a system are located on the left half of the s-plane, then there
will be a negative real part for the poles. Then the natural response will have an exponential
decay term 𝑒 −𝜁𝜔𝑛 𝑡 . Hence the system will be stable.
If any ofthe closed loop poles are on the right half s-plane, then the system will be unstable.Also
poles of multiplicity greater than one on the imaginary axis lead to the sum of responses
𝐴𝑡 𝑛 sin(𝜔𝑛 𝑡 + 𝜑), where n=1,2,3……… which also approaches infinity as time approaches infinity.
A system that has imaginary axis poles of multiplicity one yields pure sinusoidal oscillations as
natural response and are considered as marginally stable.
1 25
𝐺(𝑠) = 𝑠3 +8𝑠2 +19𝑠+12 𝑝𝑖 = −1, −3, −4 𝐺(𝑠) = 𝑝𝑖 = −0.5 ± 𝑗4.98
𝑠2 +𝑠+25

1 1
𝐺(𝑠) = 𝑠3 +3𝑠2 +5𝑠+3 𝑝𝑖 = −1, −1 ± 𝑗√2 𝐺(𝑠) = 𝑠3 +3𝑠2 +3𝑠+1 𝑝𝑖 = −1, −1, −1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
1 1
𝐺(𝑠) = 𝑝𝑖 = −1, +1, −1 ± 𝑗√2 𝐺(𝑠) = 𝑝𝑖 = ±𝑗√2, ±𝑗√2
𝑠4 +2𝑠3 +2𝑠2 −2𝑠−3 𝑠4 +4𝑠2 +4

1 1
𝐺(𝑠) = 𝑠6 +27𝑠4 +243𝑠2 +729 𝑝𝑖 = ±𝑗3, ±𝑗3, ±𝑗3 𝐺(𝑠) = 𝑠5 +𝑠4 +4𝑠3 +4𝑠2 +4𝑠+4 𝑝𝑖 = −1, ±𝑗√2, ±𝑗√2

1 1
𝐺(𝑠) = 𝑝𝑖 = ±𝑗√2 𝐺(𝑠) = 𝑝𝑖 = −1, ±𝑗√2
𝑠2 +2 𝑠3 +𝑠2 +2𝑠+2

1
𝐺(𝑠) = 𝑠4 +3𝑠2 +2 𝑝𝑖 = ±𝑗√2, ±𝑗1

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
Routh-Hurwitz’s Stability Criterion
The control system is stable if and only if all closed loop poles lie on the left half of the s-plane.
Most linear closed loop systems have their closed loop transfer function of the form
𝐶(𝑠) 𝑏𝑚 𝑠 𝑚 + 𝑏𝑚−1 𝑠 𝑚−1 + ⋯ + 𝑏1 𝑠 + 𝑏0
=
𝑅(𝑠) 𝑎𝑛 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0
Where a’s and b’s are constants, and 𝑚 ≤ 𝑛. Routh’s stability criterion helps us to find the number
of closed loop poles that lie in the right half s-plane without having to factorise the denominator
polynomial.
• Write the denominator polynomial in s in the form 𝑎𝑛 𝑠 𝑛 + 𝑎𝑛−1 𝑠 𝑛−1 + ⋯ + 𝑎1 𝑠 + 𝑎0 = 0
where the coefficients are real quantities and 𝑎0 ≠ 0 ; that is, possibility of any zero root
is eliminated.
• If any of the coefficients are zero or negative, there is a root or roots that are imaginary
or that have positive real parts. In such a case, the system is not stable. This is a
necessary condition, but not sufficient.
• If all coefficients are positive, arrange the coefficients of the polynomial in rows and
columns in the following pattern:
𝑠𝑛 𝑎𝑛 𝑎𝑛−2 𝑎𝑛−4 𝑎𝑛−6 …………
𝑛−1 𝑎𝑛−1 𝑎𝑛−3 𝑎𝑛−5 𝑎𝑛−7 ……
𝑠
𝑠 𝑛−2 𝑐1 𝑐2 𝑐3 𝑐4 …………..
𝑠 𝑛−3 𝑑1 𝑑2 𝑑3 𝑑4 ………….
………. ……… ………….. ……….. …………
2 𝑒1 𝑒2
𝑠
𝑠1 𝑓1
𝑠0 𝑔1

𝑎𝑛−1 .𝑎𝑛−2 −𝑎𝑛 .𝑎𝑛−3 𝑎𝑛−1 .𝑎𝑛−4 −𝑎𝑛 .𝑎𝑛−5 𝑎𝑛−1 .𝑎𝑛−6 −𝑎𝑛 .𝑎𝑛−7
where 𝑐1 = 𝑐2 = 𝑐3 =
𝑎𝑛−1 𝑎𝑛−1 𝑎𝑛−1
𝑐1 .𝑎𝑛−3 −𝑎𝑛−1 .𝑐2 𝑐1 .𝑎𝑛−5 −𝑎𝑛−1 .𝑐3 𝑐1 .𝑎𝑛−7 −𝑎𝑛−1 .𝑐4
𝑑1 = 𝑑2 = 𝑑3 =
𝑐1 𝑐1 𝑐1

This process is continued until the last row is completed. The complete array is triangular.
The Routh’s stability criterion states that the number of roots with positive real parts is equal
to the number of changes in sign of the coefficients of the first column of the Routh array.

Example 1
1
The closed loop transfer function of a system is 𝐺(𝑠) = 𝑠4 +2𝑠3 +3𝑠2 +4𝑠+5. Check for its stability.
The characteristic equation is 𝑠 4 + 2𝑠 3 + 3𝑠 2 + 4𝑠 + 5 = 0. All the coefficents are present
and all coefficients are positive. Hence, the necessary conditions for the stability of the system
are satisfied. Now we can complete the Routh array.
2×5−1×0
𝑠4 1 3 5 =5
2
𝑠 3 2 4 2 × 3 − 1 × 4
=1
2 2
𝑠 1 5
𝑠1 -6 1×4−2×5
= −6
0 1
𝑠 5
−6 × 5 − 1 × 0
=5
−6
=1
Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4

There are two sign changes in the first

column of the Routh’s array. Hence, two
roots lie in the right half s-plane. The
values of the poles can be obtained using
Matlab as −1.2878 + 0.8579𝑖,
−1.2878 − 0.8579𝑖, 0.2878 + 1.4161𝑖,
𝑎𝑛𝑑 0.2878 − 1.4161𝑖. We can see that two
roots lie on the right half s-plane. Hence
the sytem is unstable.

Example 2
Using Routh-Hurwitz criterion, determine
whether the system shown is stable or not.
𝐺 1000
The closed-loop trnafer function of the system is 𝐺(𝑠) = 1+𝐺𝐻 = 𝑠3 +10𝑠2 +31𝑠+1030
The characteristic equation is 𝑠 3 + 10𝑠 2 + 31𝑠 + 1030 = 0. All coeffients are present and all are
positive. The necessary conditions for stability are satisfied.
10 × 31 − 1 × 1030
3 = −72
𝑠 1 31 10
𝑠2 10 1030
𝑠1 -72 −72 × 1030 − 10 × 0
0 = 103
𝑠 103 −72
There are two sign changes in the first column of the Routh array. Therefore, two roots will lie
in the right half s-plane. Hence the system is unstable.
SPECIAL CASES
1. Zero in the first column of a row
If the first element of a row is zero, then a division by zero is necessary to form the
next row. To avoid this, we replace zero with a very small quantity 𝜀 and proceed. Finally,
we let 𝜀 → 0 and then see the sign changes in the first column.
Example 3
Check the stability of the system whose closed loop transfer function is given by
10
𝐺(𝑠) = 𝑠5 +2𝑠4 +3𝑠3 +6𝑠2 +5𝑠+3
The characteristic equation is 𝑠 5 + 2𝑠 4 + 3𝑠 3 + 6𝑠 2 + 5𝑠 + 3 = 0. All the coefficients are present
and all coefficients are positive. Hence, the necessary conditions for stability are satisfied.
Let us complete the Routh array to find how many roots are there in the right half s-plane.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

𝑠5 1 3 5 2×3−1×6
=0 2×5−1×3
𝑠4 2 6 3 2 2
= 3.5

𝑠3 0ε 3.5
𝑠2 6ε − 7 3 3𝜀 − 2 × 0
ε =3
𝜀
𝑠1 6ε − 24.5 − 3ε2
6𝜀 − 2 × 3.5
6ε − 7
𝑠 0 3 𝜀

6𝜀−7
×3.5−3𝜀 6𝜀−24.5−3𝜀 2
𝜀
6𝜀−7 = 6𝜀−7
𝜀
6𝜀−24.5−3𝜀2 6𝜀−7
×3− ×0
6𝜀−7 𝜀
6𝜀−24.5−3𝜀2
=3
6𝜀−7

6𝜀−7 6𝜀−24.5−3𝜀 2
As 𝜀 → 0+ , 𝜀
→a negative quantity, 6𝜀−7
→ a positive quantity.
6𝜀−7 6𝜀−24.5−3𝜀 2
As 𝜀 → 0− , 𝜀 →a positive quantity, 6𝜀−7
→ a positive quantity.
If we look at the sign changes, there are two sign changes in the first column. Hence, two
roots will lie in the right half s-plane. The system is unstable.

𝜀 → 0+ 𝜀 → 0−
𝑠5 + +
+ +
4
𝑠
𝑠 3
+ -
𝑠2 - +
𝑠1 + +
𝑠 0
+ +

2. Zeros in the entire a row

When there is an entire row of zeros, there is an even polynomial which is factor of the original
polynomial. Even polynomial will have only even powers of s. They will only have roots that are
symmetrical about the origin. This symmetry can occur under three conditions of root position.
1. The roots are symmetrical and real
2. The roots are symmetrical and imaginary
3. The roots are quadrantal
The row previous to the row of zeros contains the even
polynomial which is a factor of the original polynomial.
Everything from the row containing the even polynomial
down to the end of the Routh table is a test of only the
even polynomial

When we have a row of zeros, form the auxiliary polynomial [P(s)] using the row just above that.
𝑑 𝑑
Determine 𝑑𝑠 𝑃(𝑠). Replace the row of zeros by the coefficients of 𝑑𝑠 𝑃(𝑠) and proceed as usual.
The rows below the row of zeros will be a test of the even polynomial. Let us take an example to
demonstrate this.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6

Example 4
Check the stability of the system whose closed loop transfer function is given by
10
𝐺(𝑠) = 𝑠5 +7𝑠4 +6𝑠3 +42𝑠2 +8𝑠+56
Construct the Routh’s array
7 × 6 − 1 × 42 7 × 8 − 1 × 56
=0 =0
𝑠5 1 6 8 7 7
𝑠4 7 42 56
𝑠3 0 0
𝑠2
𝑠1
𝑠0

Now we have a row of zeros in the 𝑠 3 row. Form the auxiliary polynomial with the coefficients
of the 𝑠 4 row.
𝑃(𝑠) = 7𝑠 4 + 42𝑠 2 + 56
𝑑
𝑃(𝑠) = 28𝑠 3 + 84𝑠. Replace the row of zeros with these coefficients
𝑑𝑠
28 × 42 − 7 × 84 28 × 56 − 7 × 0
𝑠5 1 6 8 = 21 = 56
28 28
𝑠4 7 42 56

𝑠3 28 84

𝑠2 21 56 21 × 84 − 28 × 56
= 9.3
21
𝑠1 9.3

𝑠0 56
9.3 × 56 − 21 × 0
= 56
9.3

There is no sign change in the first column of the Routh array. If we look at the first column of
the table below the row of zeros, there is no sign change. Therefore, no roots of the even
polynomial is on the right half plane. Hence we have roots of the even polynomial on the imaginary
axis, and the system is marginally stable. The poles are 𝑠 = −7, ±𝑗√2, ±𝑗2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7

Example 5
Check the stability of the system whose closed loop transfer function is given by
10
𝐺(𝑠) = 𝑠5 +𝑠4 +6𝑠3 +6𝑠2 +25𝑠+25
Construct the Routh’s array
1×6−1×6 1 × 25 − 1 × 25
=0 =0
𝑠5 1 6 25 1 1
𝑠4 1 6 25
𝑠3 0 0

𝑠2
𝑠1
𝑠0

Now we have a row of zeros in the 𝑠 3 row. Form the auxiliary polynomial with the coefficients
of the 𝑠 4 row.
𝑃(𝑠) = 𝑠 4 + 6𝑠 2 + 25
𝑑
𝑃(𝑠) = 4𝑠 3 + 12𝑠. Replace the row of zeros with these coefficients
𝑑𝑠
4 × 6 − 1 × 12 4 × 25 − 1 × 0
𝑠5 1 6 25 =3 = 25
4 4
𝑠4 1 6 25
𝑠3 4 12
𝑠2 3 25 3 × 12 − 4 × 25
= −21.3
3
𝑠1 -21.3
𝑠0 25 −21.3 × 25 − 3 × 0
= 25
−21.3

There are two sign changes in the first column of the Routh array after the row of zeros. Hence,
two roots of the even polynomial will be on the right half plane. Since we have roots on the RHP,
the system is unstable. The poles are 𝑠 = −1, −1 ± 𝑗2, 1 ± 𝑗2
Example 6
Find the range of 𝐾 > 0, for which the given system is
stable, unstable, and marginally stable.

𝐾
Given that the open loop transfer function 𝐺(𝑠) = 𝑠(𝑠+7)(𝑠+11), feedback transfer function
𝐻(𝑠) = 1. The closed-loop transfer function of the system is given by
𝐾
𝐶(𝑠) 𝐺 𝑠(𝑠 + 7)(𝑠 + 11) 𝐾
𝑇(𝑠) = = = = 3
𝑅(𝑠) 1 + 𝐺𝐻 1 + 𝐾 2
𝑠 + 18𝑠 + 77𝑠 + 𝐾
𝑠(𝑠 + 7)(𝑠 + 11)
Let us construct the Routh Table.
𝑠3 1 77 𝑠3 1 77
𝑠2 18 K 𝑠2 18 1386
𝑠1 1386 − 𝐾 𝑠1 36
18
𝑠0 K 𝑠0 1386

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |8
For stable system, There shouldnot be a sign change in the first column. For 1386 > 𝐾 > 0, all
the three roots lie on the LHP.
If 𝐾 > 1386 we will have a negative sign in the first column of 𝑠1 row. Then there will be two sign
changes(from +ve to -ve to +ve) in the first column. Hence two roots will lie on the RHP plane.
The system will be unstable.
If 𝐾 = 1386, we will have a row of zeros in the 𝑠1 row.The auxiliary polynomial is
𝑑
𝑃(𝑠) = 18𝑠 2 + 1386 𝑑𝑠
𝑃(𝑠) = 36𝑠
The Routh table changes to the one as shown on the left.
Again no sign changes below the row of zeros. So, no roots of the even polynomial is on RHP.
Therefore, the symmetric roots of lie on the imaginary axis.
#Exercise 34
Determine the number of poles on the
LHP, RHP, and jω axis for the three
cases shown

#Exercise 35
A feedback control system has a characteristic equation 𝑠 3 + (1 + 𝐾)𝑠 2 + 10𝑠 + (5 + 15𝐾) = 0
The parameter K has to be positive. What is the maximum value K can assume before the
system becomes unstable? When K equals the maximum value, the system oscillates.
Determine the frequency of oscillation.
References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
ROOT LOCUS METHOD
Root locus is a graphical representation of closed loop poles as a system parameter is varied. It is
a powerful method of analysis of stability and transient response. We have learnt to design a first
order or second order system for their transient responses by choosing appropriate gain and
other system parameters. The real power of root locus technique lies in its ability to provide
solutions for higher order systems.
Consider the system shown in figure. The closed loop R(s) C(s)
transfer function G(s)
+
𝐶(𝑠) 𝐺(𝑠) -
𝑇(𝑠) = =
𝑅(𝑠) 1 + 𝐺(𝑠)𝐻(𝑠) H(s)
The characteristic equation of this system is
1 + 𝐺(𝑠)𝐻(𝑠) = 0 or 𝐺(𝑠)𝐻(𝑠) = −1 + 𝐽0
In most of the cases, 𝐺(𝑠)𝐻(𝑠) will be a polynomial in the complex variable ‘s’. Therefore, we can
write the magnitude condition as |𝐺(𝑠)𝐻(𝑆)| = 1
and the angle condition as < 𝐺(𝑠)𝐻(𝑠) = ±(2𝑘 + 1)𝜋 (odd multiples of 180°) 𝐾 = 0,1,2….
The values of s that satisfies both the angle and magnitude conditions are the roots of the
characteristic equation or the closed-loop poles.

In most cases, the forward path involves a gain

parameter 𝐾 sothat the transfer function is R(s) + C(s)
K G(s)
𝐶(𝑠) 𝐾𝐺(𝑠) -
𝑇(𝑠) = =
𝑅(𝑠) 1 + 𝐾𝐺(𝑠)𝐻(𝑠) H(s)
And the characteristic equation becomes 1 + 𝐾𝐺(𝑠)𝐻(𝑠) = 0
The locus of the roots of this characteristic equation when the parameter 𝐾 is varied from zero
to infinity is known as the root locus.
Root Locus Procedure
∏𝑚 (𝑠+𝑧 ) ∏𝑛 (𝑠+𝑝𝑗 )
Step 1: Write the charateristic equation in the form 1 + 𝐾 ∏𝑛𝑖=1(𝑠+𝑝𝑖 ) = 0 ≫≫ 𝐾 = − ∏𝑗=1
𝑚
(𝑠+𝑧 )
𝑗=1 𝑗 𝑖=1 𝑖
(𝑠+1)
For example consider 1 + 𝐾 𝑠(𝑠+2)(𝑠+4)2 =0

Step 2: Locate the poles of the closed loop transfer function

Root locus is the locus of the roots of the characteristic equation (poles) when the parameter 𝐾
varies from 0 to ∞
∏𝑛
𝑗=1(𝑠+𝑝𝑗 )
When 𝐾 = 0, − ∏𝑚
= 0 ≫≫ ∏𝑛𝑗=1(𝑠 + 𝑝𝑗 ) = 0 , the roots are simply the poles of 𝐺(𝑠)𝐻(𝑠).
𝑖=1(𝑠+𝑧𝑖 )
∏𝑛
𝑗=1(𝑠+𝑝𝑗 )
When 𝐾 = ∞, − ∏𝑚 𝑖=1(𝑠 + 𝑧𝑖 ) = 0, the roots are simply the zeros of G(s)H(s).
= ∞ ≫≫ ∏𝑚
𝑖=1(𝑠+𝑧𝑖 )

Therefore, the root locus starts from the poles and end at the zeros of G(s)H(s) as K varies from
zero to infinity. If the number of ploes are 𝑛 and number of zeros are 𝑚, usually 𝑛 will be grater
than 𝑚. Then 𝑛 − 𝑚 zeros will lie at infinity.
The poles are 𝑝𝑗 = 0, −2, −4, −4 and zeros are 𝑧𝑖 = −1. Here we have four poles (𝑛 = 4) and one
zero (𝑚 = 1). The root locus will start from the poles and end at the zeros. Three branches of
the root loci will go to the zeros at infinity (𝑛 − 𝑚 = 3)

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
Step 3: Locate the segments of the real axis that are part of the root loci. We know that the any
point on the root locus will satisfy the angle condition. The poles and root locus of the above
example is shown in the figure below.

(𝑠 + 1)
𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠 + 2)(𝑠 + 4)2
(𝑠 + 1)
< 𝐺(𝑠)𝐻(𝑠) =< =< (𝑠 + 1)− < 𝑠 − < (𝑠 + 2) − < (𝑠 + 4) − < (𝑠 + 4)
𝑠(𝑠 + 2)(𝑠 + 4)2
Let us see if the point s= -0.44+j 4.02 is a point on the root locus. Substitute this value of s in
the above expression.
< (𝑠 + 1) =< (−0.44 + 𝑗4.02) + 1 =< 0.56 + 𝑗4.02
4.02 -0.44+j4.02
= tan−1 = 82.070
0.56
< (𝑠) =< (−0.44 + 𝑗4.02) =< −0.44 + 𝑗4.02
4.02
= tan−1 = −83.750 (𝑜𝑟 96.250 )
−0.44
< (𝑠 + 2) =< (−0.44 + 𝑗4.02) + 2 =< 1.56 + 𝑗4.02
4.02 -4 -2 -1 0
= tan−1 = 68.790
1.56
< (𝑠 + 4) =< (−0.44 + 𝑗4.02) + 4 =< 3.56 + 𝑗4.02
4.02
= tan−1 = 48.470
3.56
< 𝐺(𝑠)𝐻(𝑠) = 82.07 − 96.25 − 68.79 − 48.47 − 48.47 = −180
Which is an odd multiple of 180. Hence the point (-0.44+j 4.02 ) is on the root locus.
Now let us consider a point on the real axis (say s= -0.5)
0
< (𝑠 + 1) =< (−0.5 + 𝑗0) + 1 =< 0.5 + 𝑗0 = tan−1 = 00
0.5
0
< (𝑠) =< (−0.5 + 𝑗0) = tan−1 = 1800
−0.5
0
< (𝑠 + 2) =< (−0.5 + 𝑗0) + 2 =< 1.5 + 𝑗0 = tan−1 = 00
1.5
-4 -2 -1 0
−1
0 0
< (𝑠 + 4) =< (−0.5 + 𝑗0) + 4 =< 3.5 + 𝑗0 = tan =0
3.5
< 𝐺(𝑠)𝐻(𝑠) = 0 − 180 − 0 − 0 − 0 = −180
Which is an odd multiple of 180. Hence the point (s= -0.5) is on the root locus. By this argument,
any point between pole at 0 and zero at -1 will satisfy the angle criterion. Hence the real axis
segment between s=0 and s=-1 will form part of the root loci.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
Now let us consider another point on the real axis (say s= -1.5)
0
< (𝑠 + 1) =< (−1.5 + 𝑗0) + 1 =< −0.5 + 𝑗0 = tan−1 = 1800
−0.5
0
< (𝑠) =< (−1.5 + 𝑗0) = tan−1 = 1800
−0.5 0
0 -4 -2 -1
< (𝑠 + 2) =< (−1.5 + 𝑗0) + 2 =< 0.5 + 𝑗0 = tan−1 = 00
0.5
0
< (𝑠 + 4) =< (−1.5 + 𝑗0) + 4 =< 2.5 + 𝑗0 = tan−1 = 00
2.5
< 𝐺(𝑠)𝐻(𝑠) = 180 − 180 − 0 − 0 − 0 = 0
Which is not an odd multiple of 180. Hence the point (s= -1.5) is not on the root locus. By this
argument, any point between zero at -1 and pole at -2 will not satisfy the angle criterion. Hence
the real axis segment between s=-2 and s=-1 will not form part of the root loci.
Consider another point on the real axis (say s= -2.5)
0
< (𝑠 + 1) =< (−2.5 + 𝑗0) + 1 =< −1.5 + 𝑗0 = tan−1 = 1800
−1.5
0
< (𝑠) =< (−2.5 + 𝑗0) = tan−1 = 1800
−2.5 -2
-4 -1 0
−1
0
< (𝑠 + 2) =< (−2.5 + 𝑗0) + 2 =< −0.5 + 𝑗0 = tan = 1800
−0.5
0
< (𝑠 + 4) =< (−2.5 + 𝑗0) + 4 =< 1.5 + 𝑗0 = tan−1 = 00
1.5
< 𝐺(𝑠)𝐻(𝑠) = 180 − 180 − 180 − 0 − 0 = −180
Which is an odd multiple of 180. Hence the point (s= -2.5) is on the root locus. By this argument,
any point between pole at -2 and poles at -4 will satisfy the angle criterion. Hence the real axis
segment between s=-4 and s=-2 will form part of the root loci.
Consider another point on the real axis (say s= -4.5)
0
< (𝑠 + 1) =< (−4.5 + 𝑗0) + 1 =< −3.5 + 𝑗0 = tan−1 = 1800
−3.5
0
< (𝑠) =< (−4.5 + 𝑗0) = tan−1 = 1800
−4.5
0
< (𝑠 + 2) =< (−4.5 + 𝑗0) + 2 =< −2.5 + 𝑗0 = tan−1 = 1800
−2.5 -4 -2 -1 0
0
< (𝑠 + 4) =< (−4.5 + 𝑗0) + 4 =< −0.5 + 𝑗0 = tan−1 = 1800
−0.5
< 𝐺(𝑠)𝐻(𝑠) = 180 − 180 − 180 − 180 − 180 = 3 × (−180)
Which is an odd multiple of 180. Hence the point (s= -4.5) is on the root locus. By this argument,
any point between pole at -4 and s=-∞ will satisfy the angle criterion. Hence the real axis segment
between s=-4 and s=-∞ will form part of the root loci.
The real axis segment to the left of odd number of poles and/or zeros will be part of the root
locus.
Step 4: Determine the separate loci (SL). Because the loci begin at the poles and end at the zeros,
the number of sepaate loci is equal to the number of poles.
Step 5: Root loci must be symmetrical with respect to the real axis bacause the complex roots
occur as conjugate pairs.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
Step 6: The root loci proceed to the zeros at infinity along the asymptotes centered at 𝜎𝐴 and
with angles ∅𝐴 . When the number of zeros 𝑚 is less than the number of poles 𝑛, then 𝑛 − 𝑚
sections of loci must end at infinity.
∑ 𝑝𝑜𝑙𝑒𝑠 − ∑ 𝑧𝑒𝑟𝑜𝑠
𝜎𝐴 =
𝑛−𝑚
(2𝑘 + 1)1800
∅𝐴 = ; 𝑘 = 0, 1, 2, … . . (𝑛 − 𝑚 − 1)
𝑛−𝑚
The centroid of the asymptotes can be found out from the characteristic equation
1 + 𝐾𝐺(𝑠)𝐻(𝑠) = 0. s=z1 (s-z1)(s-z2)
∏𝑚𝑖=1(𝑠 + 𝑧𝑖 )
1+𝐾 𝑛 =0
∏𝑗=1(𝑠 + 𝑝𝑗 )
[𝑠 𝑚 − (𝑧1 + 𝑧2 + ⋯ . . +𝑧𝑚 )𝑠 𝑚−1 + ⋯ … … … + (𝑧1 𝑧2 … . . 𝑧𝑚 )]
1+𝐾 =0
[𝑠 𝑛 − (𝑝1 + 𝑝2 + ⋯ . . +𝑝𝑛 )𝑠 𝑛−1 + ⋯ … … … + (𝑝1 𝑝2 … . . 𝑝𝑛 )]
[𝑠 𝑚 − (∑ 𝑧𝑒𝑜𝑠)𝑠 𝑚−1 ]
𝐴𝑠 𝑠 → ∞; 1+𝐾 =0
[𝑠 𝑛 − (∑ 𝑃𝑜𝑙𝑒𝑠)𝑠 𝑛−1 ]
𝑠 −𝑚 [𝑠 𝑚 − (∑ 𝑧𝑒𝑜𝑠)𝑠 𝑚−1 ]
1 + 𝐾 −𝑚 𝑛 =0
𝑠 [𝑠 − (∑ 𝑃𝑜𝑙𝑒𝑠)𝑠 𝑛−1 ]
[1 − 𝑠 −1 (∑ 𝑧𝑒𝑜𝑠)]
1 + 𝐾 𝑛−𝑚 =0
[𝑠 − 𝑠 𝑛−𝑚−1 (∑ 𝑃𝑜𝑙𝑒𝑠)]
The second term in the numerator is very small; we can use the binomial expansion (𝑓𝑜𝑟 𝑥 < 1)
1
1 − 𝑥 = (1 + 𝑥)−1 =
1+𝑥
1
1+𝐾 =0
[𝑠 𝑛−𝑚 − 𝑠 𝑛−𝑚−1 (∑ 𝑃𝑜𝑙𝑒𝑠)][1 + 𝑠 −1 (∑ 𝑧𝑒𝑜𝑠)]
1
1+𝐾 =0
[𝑠 𝑛−𝑚 + 𝑠 𝑛−𝑚−1 (∑ 𝑧𝑒𝑟𝑜𝑠) − 𝑠 𝑛−𝑚−1 (∑ 𝑃𝑜𝑙𝑒𝑠) − 𝑠 𝑛−𝑚−2 (∑ 𝑃𝑜𝑙𝑒𝑠) (∑ 𝑧𝑒𝑜𝑠)]
1
1+𝐾 =0
[𝑠 𝑛−𝑚 + 𝑠 𝑛−𝑚−1 (∑ 𝑧𝑒𝑟𝑜𝑠) − (∑ 𝑃𝑜𝑙𝑒𝑠)]
∏𝑚
𝑖=1(𝑠+𝑧𝑖 )
Again for large value of s, the characteristic equation 1 + 𝐾
∏𝑛
= 0 can be approximated as
𝑗=1(𝑠+𝑝𝑗 )
(𝑠−𝜎𝐴 )𝑚 1 1
1+𝐾 = 0 >> 1+𝐾 = 0 >> 1+𝐾 =0
(𝑠−𝜎𝐴 )𝑛 (𝑠−𝜎𝐴 )𝑛−𝑚 𝑠𝑛−𝑚 −𝜎𝐴 (𝑛−𝑚)𝑠𝑛−𝑚−1

Comparing the coefficients in the two equations, we can write

−𝜎𝐴 (𝑛 − 𝑚) = ∑ 𝑧𝑒𝑟𝑜𝑠 − ∑ 𝑃𝑜𝑙𝑒𝑠
∑ 𝑃𝑜𝑙𝑒𝑠 − ∑ 𝑍𝑒𝑟𝑜𝑠
𝜎𝐴 =
𝑛−𝑚
If a test point s is selected very far from the origin, < 𝐺(𝑠)𝐻(𝑠) = ±(2𝑘 + 1)𝜋
(𝑠−𝜎 )𝑚 1
< 𝐾 (𝑠−𝜎𝐴 )𝑛 = ±(2𝑘 + 1)𝜋 < 𝐾 (𝑠−𝜎 𝑛−𝑚 = ±(2𝑘 + 1)𝜋 −(𝑛 − 𝑚) < (𝑠 − 𝜎𝐴 ) = ±(2𝑘 + 1)𝜋
𝐴 𝐴)
(2𝑘 + 1)𝜋
< (𝑠 − 𝜎𝐴 ) = ±
𝑛−𝑚
∑ 𝑝𝑜𝑙𝑒𝑠 − ∑ 𝑧𝑒𝑟𝑜𝑠 (−4 − 4 − 2 − 0) − (−1) (−10) + 1
𝜎𝐴 = = = = −3
𝑛−𝑚 4−1 3
(2𝑘 + 1)1800
∅𝐴 = ; 𝑘 = 0, 1, 2
𝑛−𝑚
(2 × 0 + 1)1800 (2 × 1 + 1)1800 (2 × 2 + 1)1800
∅𝐴 = , , = 600 , 1800 , 3000
3 3 3

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5
Step 7: Determine the Break Away (Break in ) Point. At the break away or break in points there
will be multiplicity of roots. The value of s at the break away or break in point can be obtained as
follows. Write the characteristic equation in the form 𝐾 = 𝑃(𝑠). The roots of the polyomial
𝑑𝐾
𝑑𝑠
= 0 will give the break away or break in point.
𝑌(𝑠)
The characteristic equation 1 + 𝐾𝐺(𝑠)𝐻(𝑠) = 0 can be written as 1 + 𝐾 𝑋(𝑠) = 0
𝑓(𝑠) = 𝑋(𝑠) + 𝐾𝑌(𝑠) = 0
𝑑𝑓(𝑠)
This will have multiple roots at points where =0
𝑑𝑠
𝑋′(𝑠) + 𝐾𝑌′(𝑠) = 0
𝑋′(𝑠)
𝐾=−
𝑌′(𝑠)
Substituting this equation in f(s), 𝑋(𝑠)𝑌 ′ (𝑠) − 𝑌(𝑠)𝑋′(𝑠) = 0
When we solve this equation for s, we will get the location where multiple roots exists.
𝑌(𝑠) 𝑋(𝑠) 𝑑𝐾
If we write 𝐾 𝑋(𝑠) = −1 or 𝐾 = − 𝑌(𝑠). If we find 𝑑𝑠
an equate to zero
𝑑𝐾 𝑌(𝑠)𝑋 ′ (𝑠) − 𝑌 ′ (𝑠)𝑋(𝑠)
=− =0
𝑑𝑠 𝑌 2 (𝑠)
We get the same equation 𝑌 ′ (𝑠)𝑋(𝑠) − 𝑌(𝑠)𝑋 ′ (𝑠) = 0
𝑠(𝑠 + 2)(𝑠 + 4)2 𝑠 4 + 10𝑠 3 + 32𝑠 2 + 32𝑠
𝐾=− =−
𝑠+1 𝑠+1
𝑑𝐾 (𝑠 + 1)(4𝑠 3 + 30𝑠 2 + 64𝑠 + 32) − (𝑠 4 + 10𝑠 3 + 32𝑠 2 + 32𝑠)
=− =0
𝑑𝑠 (𝑠 + 1)2
3𝑠 4 + 24𝑠 3 + 62𝑠 2 + 64𝑠 + 32 = 0
The roots are −4, −2.5994, − 0.7003 ± 𝑗0.7317
The break away point is −2.5994

Step 8: Determine the points where the root loci crosses the imaginary axis. This can be found
out by use of the Routh’s stability criterion
(𝑠+1)
The characteristic equation of the given example is 1 + 𝐾 𝑠(𝑠+2)(𝑠+4)2 = 0
𝑠(𝑠 + 2)(𝑠 + 4)2 + 𝐾(𝑠 + 1) = 0
𝑠 4 + 10𝑠 3 + 32𝑠 2 + (32 + 𝐾)𝑠 + 𝐾 = 0
𝑠4 1 32 K
10 × 32 − 1 × (32 + 𝐾)
𝑠3 10 32+K 10

𝑠2 288 − 𝐾 K
10 (28.8 − 0.1𝐾) × (32 + 𝐾) − 10𝐾
1 28.8 − 0.1𝐾
𝑠 𝐾 2 − 156𝐾 − 9218
28.8 − 0.1𝐾
𝑠0 K
For K=201.7, we will have a row of zeros. Then the auxiliary polynomial will be
8.63𝑠 2 + 201.7 = 0
𝑠 = ±𝑗4.83
The root loci crosses the imaginary axis at 𝑠 = ±𝑗4.83

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6
Step 9: Determine the angle of departure from a complex pole and the angle of arrival of the
locus at a complex zero, using the angle criterion.

Step 10: Choose test points in such a way that they satisfy the angle criterion. Draw the loci
passing through these points.

Step 11: Determine the pair of dominant complex conjugate poles such that the damping ratio is
0.5.
Closed loop poles with 𝜁 = 0.5 lie on a lines passing through the origin and making angles ± cos−1 𝜁 =
± cos−1(0.5) = ±600 with the negative real axis.
The closed loop poles are −1.37 ± 𝑗2.36
The value of K that will yield these poles
can be obtained from the magnitude
condition
|𝐾𝐺(𝑠)𝐻(𝑆)| = 1
(𝑠 + 1)
𝐾| | =1
𝑠(𝑠 + 2)(𝑠 + 4)2 𝑠=−1.37+𝑗2.36
𝑠(𝑠 + 2)(𝑠 + 4)2
𝐾=| |
𝑠+1 𝑠=−1.37+𝑗2.36
|𝑠||𝑠 + 2||𝑠 + 4||𝑠 + 4|
𝐾=
|𝑠 + 1| 𝑠=−1.37+𝑗2.36

|−1.37 + 𝑗2.36||−1.37 + 𝑗2.36 + 2||−1.37 + 𝑗2.36 + 4||−1.37 + 𝑗2.36 + 4|

𝐾=
|−1.37 + 𝑗2.36 + 1|

√(−1.37)2 + (2.36)2 . √(0.63)2 + (2.36)2 . √(2.63)2 + (2.36)2 . √(2.63)2 + (2.36)2

𝐾=
√(−0.37)2 + (2.36)2
2.724 × 2.442 × 3.534 × 3.534
𝐾= = 34.8
2.389
With this value of K, the characteristic equation becomes
4 3 2 (32
𝑠 + 10𝑠 + 32𝑠 + + 𝐾)𝑠 + 𝐾 = 0
4 3 2
𝑠 + 10𝑠 + 32𝑠 + 66.8𝑠 + 34.8 = 0
The roots of this equation are −6.5466, −0.7151, − 1.37 + 𝑗2.36, − 1.37 + 𝑗 2.36

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7

*Example
𝐾
Sketch the root locus of the system if G(s)H(s) = 𝑠(𝑠3 +12𝑠2 +64𝑠+128)

Step 1:
𝐾
Write the characteristic equationin the form 1 + G(s)H(s) = 0; 1 + 𝑠(𝑠3 +12𝑠2 +64𝑠+128) = 0

Step 2:
Determine the poles and zeros of G(s)H(s). The poles are 𝑠 = 0, −4, −4 ± 𝑗4. There no zeros.
The number of poles(𝑛) = 4.
The number of zeros (𝑚) = 0.
The root loci originate from the poles of 𝐺(𝑠)𝐻(𝑠) and end at the zeros.
The number of branches of root loci proceeding to the zeros at infinity= 𝑛 − 𝑚 = 4
Step 3:
Number of separate loci is equal to the number of poles (n)
Step 4:
Locate the segments of real axis that are part of the
root loci. The real axis segment to the left of odd
number of poles and zeros will be part of the root loci.
The first pole/zero is located at the origin. So the
part of real axis from s = 0 till the next pole/zero at
s= -4 will be part of the loci. Since there are no more
poles/zeros in the real axis after s= -4, no more part
of real axis will form a part of root loci.

Step 5:
The root loci are symmetrical about the real axis
Step 6:
The root loci proceed to the zeros at infinity along
the assymptotes centered at
∑ 𝑝𝑜𝑙𝑒𝑠−∑ 𝑧𝑒𝑟𝑜𝑠 (0−4−4+𝑗4−4−𝑗4)−(0) (−12)
𝜎𝐴 = 𝑛−𝑚
= 4−0
= 4
= −3
(2𝑘+1)1800
∅𝐴 = 𝑛−𝑚
; 𝑘 = 0, 1, 2, 3
1800 3×1800 5×1800 7×1800
∅𝐴 = , , , = 450 , 1350 , 2250 , 3150
4 4 4 4

Step 7:
𝐾
1 + 𝑠(𝑠3 +12𝑠2 +64𝑠+128) = 0
Determine the break away/break in point. K = −𝑠(𝑠 3 + 12𝑠 2 + 64𝑠 + 128)
Write the characteristic equation 𝐾 = 𝑃(𝑠) 𝑑𝐾
= −(4𝑠 3 + 36𝑠 2 + 128𝑠 + 128) = 0
𝑑𝐾 𝑑𝑠
Determine the roots of 𝑑𝑠
=0 s=-3.7117 ± 𝑗2.5533, − 1.5767
We take the value 𝑠 = −1.57 for the
breakaway point

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |8

Step 8:
Determine the imaginary axis crossing by constructing the Routh table.
𝐾
The characteristic equation is 1 + 𝑠(𝑠3 +12𝑠2 +64𝑠+128) = 0
𝑠 4 + 12𝑠 3 + 64𝑠 2 + 128𝑠 + 𝐾 = 0
12 × 64 − 1 × 128)
𝑠4 1 64 K = 53.33
12
𝑠3 12 128
2 53.33 K 12 × 𝐾 − 1 × 0)
𝑠 =𝐾
12
𝑠1 53.33 × 128 − 12 × 𝐾
53.33
53.33 × 128 − 12 × 𝐾
𝑠0 K
53.33

When the root loci cross the imaginary axis, there will be a row of zeros. This is possible only
in the 𝑠1 row when 53.33 × 128 − 12 × 𝐾 = 0 or 𝐾 = 568.89.
The auxiliary polynomial is 53.33𝑠 2 + 568.89 = 0
Or 𝑠 = ±𝑗3.26. is the point where the root loci crosses the imaginary axis
Step 9:
Determination of the angle of locus departure from the
complex poles and angle of locus arrival at complex
s
zeros from the angle coriterion.
Consider a test point s very close to the complex pole
at −4 + 𝑗4. Let 𝜃1 be the angle subtented by the
vector from the pole at −4 + 𝑗4 to the test point s
𝜃2 from −4 to 𝑠 = 90
𝜃3 from −4 − 𝑗4 to 𝑠 = 90
4
𝜃4 from 0 to 𝑠 = 135( tan−1 4 = −45)
𝜃1 + 𝜃2 + 𝜃3 + 𝜃4 = 𝑜𝑑𝑑 𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑒 𝑜𝑓 1800
𝜃1 = 180 − 90 − 90 − 135 = −1350
Since the root loci is symmetrical about the real axis,
the test point s in the vicinity of the pole at −4 − 𝑗4
will have an angle contribution of +1350
Step 10:
Determine the root locations which satisfy the angle
criterion and join such points to get the locus.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |9

#Exercise 36
𝐾
Sketch the root locus of the system if 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+4)(𝑠2 +2𝑠+2). Verify your answer by
plotting it in Matlab/Scilab.
#Exercise 37
𝐾(𝑠+6)
Sketch the root locus of the system if 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+3)(𝑠2 +2𝑠+2). Verify your answer by
plotting it in Matlab/Scilab.
#Exercise 38
𝐾(𝑠+4)(𝑠+6)
Sketch the root locus of the system if 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+2)
. Verify your answer by plotting
it in Matlab/Scilab.
#Exercise 39
𝐾
Sketch the root locus of the system if 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+3)(𝑠+5)(𝑠2 +2𝑠+2). Verify your answer by
plotting it in Matlab/Scilab.
#Exercise 40
𝐾
Using Matlab/Scilab, sketch the root locus of the system whose 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+5). Observe
the changes in the root locus when it is modified by adding poles and zeros as given below
𝐾
i) 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+5)(𝑠+10)
𝐾(𝑠+1)
ii) 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+5)
𝐾(𝑠+10)
iii) 𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+5)
𝐾(𝑠+1)
iv) 𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠+5)(𝑠+10)
𝐾(𝑠+1)(𝑠+10)
v) 𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠+5)(𝑠+3)

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
DESIGN WITH ROOT LOCUS
Consider the example of an antenna
position control system as shown. The
required position is set on the input
potentiometer and the actual position is
sensed by the output potentiometer. The
block diagram is modified as shown

With the preamplifier gain 𝐾 = 64.21, let us find out the closed-loop poles and the system’s
dynamic specifications.
𝑝𝑖 = −100, −0.83 ± 𝑗1.89
Since the third pole 𝑠 = −100 is far to the left of origin, the dominant poles are −0.83 ± 𝑗1.89
𝑟𝑎𝑑 𝑟𝑎𝑑
−𝜁𝜔𝑛 = −0.83, 𝜔𝑑 = 1.89 ≫≫≫ 𝜁 = 0.402, 𝜔𝑛 = 2.06 , 𝜃 = cos −1 𝜁 = 66.3°
𝑠 𝑠
−𝜋𝜁
𝜋−𝜃 𝜋 4
𝑡𝑟 = = 1.08 𝑠, %𝑂𝑆 = 𝑒 √𝟏−𝜻𝟐 = 25%, 𝑡𝑝 = = 1.6 𝑠, 2% 𝑡𝑠 = 4.8 𝑠
𝜔𝑑 𝜔𝑑 𝜁𝜔𝑛
𝐾𝑣 = lim 𝑠𝐺(𝑠)𝐻(𝑠) = 2.5
𝑠→0
6.63𝐾
𝐾𝐺(𝑠)𝐻(𝑠) = 𝑠(𝑠+1.71)(𝑠+100)
.

The poles of 𝐺(𝑠)𝐻(𝑠) are 𝑠 = 0, −1.71, −100.

Construct the root loci. The portion of the root
loci close to the origin is shown. The third
segment of the loci will start from −100 and
extend to −∞ which is not shown in the figure.
If we draw a line at an angle 𝜃 = cos−1 𝜁, it will
cut the locus at −0.83 ± 𝑗1.89 (the dominant
closed loop poles).

The characteristic equation is 1 + 𝐾𝐺(𝑠)𝐻(𝑠) = 0

𝐾𝐺(𝑠)𝐻(𝑠) = −1
𝐾|𝐺(𝑠)𝐻(𝑠)| = 1 and < 𝐺(𝑠)𝐻(𝑠) = ±(2𝑘 + 1)1800
Since the points −0.83 ± 𝑗1.89 are on the
rootlocus, they will satisfy the angle condition.
From the magnitude condition, let us find out the
gain 𝐾.
1 1
𝐾 = |𝐺(𝑠)𝐻(𝑠)| = 6.63
−0.83±𝑗1.89 | |
𝑠(𝑠+1.71)(𝑠+100) 𝑠=−0.83+𝑗1.89

𝑠(𝑠 + 1.71)(𝑠 + 100)

=| |
6.63 𝑠=−0.83+𝑗1.89
|𝑠||𝑠+1.71||𝑠+100| |−0.83+𝑗1.89| |−0.83+𝑗1.89+1.71| |−0.83+𝑗1.89+100|
𝐾=( 6.63
) = 6.63
𝑠=−0.83+𝑗1.89
√−0.832 +1.892 √(1.71−0.83)2 +1.892 √(100−0.83)2 +1.892 √4.261 √4.3465√9838.26
= = = 64.38
6.63 6.63

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
Which is the preamplifier gain we have selected. The closed-loop gain is 6.63 × 64.38 = 426 which
is visible in the figure.
Suppose we require the following specification for the system %𝑂𝑆 = 25, 2% 𝑡𝑠 = 2 𝑠, 𝐾𝑣 = 20.
Since the %os has not been changed, the value of 𝜁 remains the same.
4
𝑡𝑠 = 2 = ≫≫≫ 𝜁𝜔𝑛 = 2 ≫≫≫ 𝜔𝑛 = 4.975 ≫≫≫ 𝜔𝑑 = 4.56
𝜁𝜔𝑛
The new closed-loop pole location is −𝜁𝜔𝑛 ± 𝑗𝜔𝑑 = −2 ± 𝑗4.56
But, we can see that root locus doesnot pass
through this point. With the given system, we
can not achieve the required design
specifications. So, we need to modify the
parameters of the plant in such a way that the
new root locus passes through the desired
point. This may not be possible always. Then
we need to add additional dynamics in to the
system so that the new root locus will pass
through the desired points. The additional
components that we add for this purpose are
known as compensators.
Additional poles and zeros are to be
introduced in the forward path for the 𝜃𝑖 (s) + 6.63𝐾 𝜃𝑜 (s)
compensation. Consider a system whose 𝐺𝑐 (𝑠) 𝑠(𝑠+1.71)(𝑠+100)

𝑠+𝑧𝑐 -
transfer function is 𝐺𝑐 (𝑠) = is added in
𝑠+𝑝𝑐
the forward path. 𝑧𝑐 is the compensator zero
and 𝑝𝑐 is the compensentaor pole.
Let us assume that 𝑧𝑐 = −2. Where should we place the compensator zero 𝑝𝑐 on the real axis?
The new pole location 𝑠 = −2 ± 𝑗4.56 should satisfy the angle condition considering the open loop
poles 𝑝𝑖 = −100, −1.71, 0, 𝑝𝑐 and the open loop zero 𝑧𝑐 = −2.
4.56
𝜃1 = −tan−1 2
= −66.320 = 113.680 (pole at s=0)
4.56
𝜃2 = −tan−1 2−1.71 = −85.850 = 94.150 (pole at s=-1.71)
∅1 = 90 (compensator zero at s=-2)
0

4.56
𝜃3 = tan−1 100−2 = 2.660 (pole at s=-100)
𝜃4 =? (compensator pole)
The angle condition is
∅1 − 𝜃1 − 𝜃2 − 𝜃3 − 𝜃4 = ±(2𝑘 + 1)180
90 − 113.68 − 94.15 − 2.66 − 𝜃4 = −180
4.56
𝜃4 = 59.50 ≫≫ 𝑥 = = 2.68
tan 59.5
Hence the compensator pole is at −4.68.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

The new forward path transfer function is

6.63𝐾(𝑠 + 2)
𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠 + 1.71)(𝑠 + 100)(𝑠 + 4.68)
The closed loop gain is 6.63𝐾 = 2580
𝐾 = 389
We have thus improved the transient response of
the system. This is known as lead compensator. In
lead compensation, the compensator zero is to the
right of the compensator pole.

The velocity error constant of the lead compensated system

6.63𝐾(𝑠 + 2) 2580 × 2
𝐾𝑣 = lim 𝑠𝐺(𝑠)𝐻(𝑠) = lim 𝑠 = = 6.44
𝑠→0 𝑠→0 𝑠(𝑠 + 1.71)(𝑠 + 100)(𝑠 + 4.68) 1.71 × 100 × 4.68
But the required 𝐾𝑣 = 20. We can notice that the lead compensator did not improve the steady
state behavior of the system. To improve the steady state behavior, we add another
compensator known as lag compensator. Now the entire compensator is known as lag-lead
compensator.Let us assume a compensator pole close to the origin, 𝑠 = −0.01.We need to improve
20
the steady state behavior by 6.44 = 3.1 times. Let us place the zero at 𝑠 = −0.031. Now the
transfer function of the lag-lead compensated system becomes
6.63𝐾(𝑠 + 2)(𝑠 + 0.031)
𝐺𝐿𝐿𝐶 (𝑠)𝐻(𝑠) =
𝑠(𝑠 + 1.71)(𝑠 + 100)(𝑠 + 4.68)(𝑠 + 0.01)
The root loci of the lag-lead compensated system is shown below. The root loci starting from
−100 and proceeding to −∞ is not shown. The root loci near the origin is enlarged and shown on
the right. The transient behavior is the same but at the same time the steady state behavior is
also improved.

Depending on where the the compensator is placed, they are also known as cascade compensator
or feedback compensator.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
Ideal Integral Controller

Original system Compensated system

➢ A pole is added at the origin ➢ Desired transient response can not

➢ The type of the system is increased. be achieved
Hence, the steady state error is
improved.
Proportional+Integral Controller

𝐾𝑖
𝐺𝑐 (𝑠) = 𝐾 +
𝑠
A zero is added because of the proportional
part in addition to the pole at the origin

➢ Steady state error is improved ➢ Transient response is unaffected.

➢ Integrator requires active element
➢ Expensive
Lag Compensator

➢ The pole is not at the origin, but close to the origin to the right of the zero
𝑧
➢ Steady state error will be improved. The new error constants will be 𝑝𝑐 times the old ones.
𝑐
➢ No effect on the transient response

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5
Proportional+ derivative controller
If the root locus does not pass through the R(s) + C(s)
𝐾(𝑠 + 𝑧𝑐 ) G(s)
desired points, it needs to be reshaped.
-

Example.

The closed-loop poles for a gain 𝐾 = 43.4 are

−7.59, −1.204 ± 𝑗2.06 . The uncompensated
dominant poles are −1.204 ± 𝑗2.06. The value of 𝜁
for this dominant poles is 0.504.
4
overshoot= 16%, settling time= = 3.32 𝑠
𝜁𝜔𝑛

Let us design a 𝑃𝐷 compensator to yield the same

overshoot and three fold reduction in settling
4
time. i.e., = 1.106 >>>> 𝜁𝜔𝑛 = 3.613. Since the
𝜁𝜔𝑛
damping factor is the same, we can locate the new
pole on the same radial line where the vertical
line from −𝜁𝜔𝑛 = −3.613 intersects.

The angle contribution of poles for the desired pole

location
6.193
𝜃1 = −tan−1 = −59.740 = 120.260 (pole at 𝑠 = 0)
3.613
6.193
𝜃2 = tan−1 4−3.613 = 86.420 (pole at 𝑠 = −4)
6.193
𝜃3 = tan−1 6−3.613 = 68.90 (pole at 𝑠 = −6)
∅1 =? ? (compensator zero at whose location yet to be
known)
The angle condition is
∅1 − 𝜃1 − 𝜃2 − 𝜃3 = ±(2𝑘 + 1)180
∅1 − 120.26 − 86.42 − 68.9 = −180
6.193
∅1 = 95.60 = −84.4 ≫≫ 𝑥 = = 0.607
tan 84.4 -3.613
Hence the compensator zero should be at
−𝜎 = −3.613 + 0.607 = −3

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6
The root locus after compensation and the transient response are shown below.

➢ Noise amplification
➢ Expensive as it requires active elements

Lead Compensator
➢ Passive element approximation ➢ Additional pole away from the origin
of PD controller, hence cheaper
𝑠+𝑧
➢ 𝐺𝑐 (𝑠) = 𝑠+𝑝𝑐 , compensator pole 𝑝𝑐 is away from
𝑐
➢ Less noise amplification origin to the left of the compensator zero 𝑧𝑐 .
➢ Choose a compensator zero some where around the open loop poles and using the angle
criterion locate the compensator pole. There will be many possibilities and choose the
most suitable one considering the static error constants, required gain to reach the
design point, and justification of the second order assumption(assumption of dominant
poles by looking at the remaing poles).

Proportional+Inegral+Derivative (PID) Controller

➢ Improves transient response ➢ Design Procedure
and steady state behavior ✓ From the requirements, find out the pole location to
➢ Uses active elements meet the transient response specifications.
𝐾𝑝 +
𝐾𝑖
, + 𝐾𝑑 𝑠 ✓ Design the 𝑃𝐷 controller
𝑠 ✓ Check the validity of the design.
✓ Design the 𝑃𝐼 controller to yield the required steady
state error value.
✓ Combine 𝑃𝐷 and 𝑃𝐼 to obtain 𝐾𝑝 , 𝐾𝑖 , 𝑎𝑛𝑑 𝐾𝑑
Lag-Lead Compensator
➢ Improves transient response ➢ Design Procedure
and steady state behavior ✓ find out the pole location to meet the transient
➢ Uses passive elements response specifications.
✓ Design the lead compensator controller
✓ Evaluate the steady state error and determine the
improvement required.
✓ Design the lag compensator

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7

#Exercise 41
Consider the system shown. Design a 𝑃𝐼𝐷
2
controller such that the peak time is 3 times
the uncompensated system with 20%
overshoot and zero stady state error.

Physical Realisation of compensators

We learnt about op-amps earlier and derived
𝑉0 (𝑠) 𝑍2 (𝑠)
=−
𝑉𝑖 (𝑠) 𝑍1 (𝑠)
By judiciously choosing 𝑍1 (𝑠) and 𝑍2 (𝑠), this circuit can be used
to implement the compensators and controllers
Function 𝑍1 (𝑠) 𝑍2 (𝑠)
𝐺𝑐 (𝑠) = −
𝑍2 (𝑠)
𝑍1 (𝑠)
Gain −
𝑅2
𝑅1

Integration 1
− 𝑅𝐶
𝑠
Differentiation −𝑅𝐶𝑠

𝐶2
PI controller 1
𝑅2 (𝑠 + 𝑅2 𝐶2 )
−
𝑅1 𝑠
PD controller −𝑅2 𝐶 (𝑠 +
1
)
𝑅1 𝐶2

PID controller 𝑅2 𝐶1
1
𝑅1 𝐶2
− [( + ) + 𝑅2 𝐶1 𝑠 + ]
𝑅1 𝐶2 𝑠

Lag compensator 1
𝐶1 𝑠 + 𝑅1 𝐶1
− [ ]
𝐶2 𝑠 + 1
𝑅2 𝐶2
where 𝑅2 𝐶2 > 𝑅1 𝐶1

Lead 1
𝐶1 𝑠 + 𝑅1 𝐶1
compensator − [
𝐶2 𝑠 + 1
]
𝑅2 𝐶2
where 𝑅2 𝐶2 < 𝑅1 𝐶1

Lag-lead compensator can be formed by

cascading the lag compensator with the lead
compensator as shown

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |8
Lag, lead, and lag-lead compensators can also be implemented with passive network.
Function Network Transfer function
𝑉𝑜 (𝑠)
𝑉𝑖 (𝑠)

Lag compensation 𝑅2 (𝑠 +
1
𝑅2 𝐶2
)
( )
𝑅1 + 𝑅2 𝑠 + 1
(𝑅1 + 𝑅2 )𝐶2

𝐶2

Lead compensation 1
(𝑠 +
𝑅1 𝐶1
)
1 1 1
𝑠+( + )
𝑅1 𝑅2 𝐶1
𝐶1

Lag-lead compensation (𝑠 +
1
𝑅1 𝐶1
) (𝑠 +
1
𝑅2 𝐶2
)
1 1 1 1
𝑠2 + 𝑠 ( + + )+
𝑅1 𝐶1 𝑅2 𝐶2 𝑅2 𝐶1 𝑅1 𝐶1 𝑅2 𝐶2

A lag-lead compensator can be realised by

cascading the lead network with the lag
network. When we use this, care should be
taken to isolate one from the other.
Otherwise, one network will load the other
and the transfer function will not be the
product of the individual transfer
functions. A possible way of achieving this
by using an op-amp for isolation.

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
FREQUENCY RESPONSE
The steady state response of a system to a sinusoidal input is known as frequency response. A
sinusoidal input to a linear sytem gives a sinusoidal out put of the same frequency. But the
amplitude and phase of the output may differ from that of input. These differences will be
functions of frequency.
Let us consider a RC network as shown. We can write,
R
1 1
𝑒𝑖 = 𝑖𝑅 + 𝐶 ∫ 𝑖 𝑑𝑡 and 𝑒𝑜 = 𝐶 ∫ 𝑖 𝑑𝑡. Taking Laplace transform
1 𝐼(𝑠) 1
𝐸𝑖 (𝑠) = 𝑅 𝐼(𝑠) + = 𝐼(𝑠) (𝑅 + ) 𝑒𝑖 𝑒𝑜
𝐶 𝑠 𝐶𝑠 C
1 𝐼(𝑠)
𝐸𝑜 (𝑠) =
𝐶 𝑠

The transfer function of the system The output 𝐸𝑜 (𝑠) = 𝐺(𝑠)𝐸𝑖 (𝑠)
1
𝐼(𝑠) Let the input voltage be 𝑒𝑖 = 𝐴 sin 𝜔𝑡
𝐸𝑜 (𝑠) 𝐶𝑠 1
𝐺(𝑠) = = = 𝐴𝜔
𝐸𝑖 (𝑠) 𝐼(𝑠) (𝑅 + 1 ) 1 + 𝑅𝐶 𝑠 𝐸𝑖 (𝑠) = 2
𝐶𝑠 𝑠 + 𝜔2
1 𝑎 𝐴𝜔 𝐵 𝑃𝑠 + 𝑄
= 𝐸𝑜 (𝑠) = = + 2
1 + 𝑇𝑠 𝑠+𝑎 𝑠 +𝜔2 2 𝑠 + 𝑎 𝑠 + 𝜔2
This is a first order system with time constant 𝐴𝑎𝜔 = 𝐵(𝑠 2 + 𝜔2 ) + (𝑃𝑠 + 𝑄)(𝑠 + 𝑎)
𝑇 = 𝑅𝐶. This can also be written as Put 𝑠 = −𝑎 ≫≫ 𝐵 =
𝐴𝑎𝜔
(𝑎 2 +𝜔2 )
1 𝑎
𝐺(𝑠) = = −𝐴𝑎𝜔
1 Equating coefficients of 𝑠 2 ≫≫ 𝑃 = (𝑎2 +𝜔2 )
𝑇(𝑠 + 𝑇) 𝑠 + 𝑎
+𝐴𝑎 2 𝜔
Equating coefficents of 𝑠 ≫≫ 𝑄 = (𝑎2 +𝜔2 )
𝐴𝑎𝜔 −𝐴𝑎𝜔 𝐴𝑎2 𝜔
𝑠+ 2 𝐴𝑎𝜔 1 𝐴𝑎𝜔 𝑠 𝐴𝑎2 𝜔
(𝑎2 2 2 2
+ 𝜔 ) (𝑎 + 𝜔 ) (𝑎 + 𝜔 2 )
𝐸𝑜 (𝑠) = + 2 2
= 2 2
[ ] − 2 2
[ 2 2
] + [ 2 ]
𝑠+𝑎 𝑠 +𝜔 (𝑎 + 𝜔 ) 𝑠 + 𝑎 (𝑎 + 𝜔 ) 𝑠 + 𝜔 (𝑎 + 𝜔 ) 𝑠 + 𝜔 2
2 2

Taking inverse Laplace transform

𝐴𝑎𝜔 −𝑎𝑡
𝐴𝑎𝜔 𝐴𝑎2
𝑒𝑜 (𝑡) = 𝑒 − cos 𝜔𝑡 + sin 𝜔𝑡
(𝑎2 + 𝜔 2 ) (𝑎2 + 𝜔 2 ) (𝑎2 + 𝜔 2 )
At steady state 𝑒 −𝑎𝑡 → 0 𝑎𝑠 𝑡 → ∞
𝐴𝑎𝜔 𝐴𝑎2
𝑒𝑜 (𝑡) = − cos 𝜔𝑡 + sin 𝜔𝑡
(𝑎2 + 𝜔 2 ) (𝑎2 + 𝜔 2 )
2 2
−𝐴𝑎𝜔 𝐴𝑎2 𝐴𝑎𝜔 𝐴𝑎2 𝜔
𝑃𝑢𝑡 = 𝑋 sin 𝜃 and = X cos 𝜃 sothat [ ] +[ ] = 𝑋 2 and tan 𝜃 = −
(𝑎2 +𝜔2 ) (𝑎2 +𝜔2 ) (𝑎2 +𝜔2 ) (𝑎2 +𝜔2 ) 𝑎

2 2
𝐴𝑎𝜔 𝐴𝑎2 𝐴𝑎
𝑋 = √[ 2
] +[ 2
] =
(𝑎 + 𝜔2 ) (𝑎 + 𝜔2 ) √(𝑎2 + 𝜔2 )
𝐴𝑎 𝐴
𝑒𝑜 (𝑡) = 𝑋 sin 𝜃 cos 𝜔𝑡 + X cos 𝜃 sin 𝜔𝑡 = 𝑋 sin(𝜔𝑡 + 𝜃) = sin(𝜔𝑡 + 𝜃) = sin(𝜔𝑡 + 𝜃)
√(𝑎2 + 𝜔2) √1 + 𝜔 2 𝑇 2
This is the frequency response of the given first order system. The magnitude of the frequency
response is √ 2 2 and phase angle 𝜃 = tan−1 (− 𝑎 ) = tan−1 (−𝜔𝑇) = −tan−1 (𝜔𝑇)
𝐴 𝜔
1+𝜔 𝑇

We can get the same results if we replace 𝑠 in G(s) by the complex frequency 𝑗𝜔
1 1 1 − 𝑗𝜔𝑇 1 𝜔𝑇
𝐺(𝑠) = 𝐺(𝑗𝜔) = 𝐺(𝑗𝜔) = 2
𝐺(𝑗𝜔) = 2
−𝑗
1 + 𝑇𝑠 1 + 𝑗𝜔𝑇 1+𝜔 𝑇2 1+𝜔 𝑇2 1 + 𝜔2 𝑇2
1
𝑀 = |𝐺(𝑗𝜔)| = ; 𝜃 =< 𝐺(𝑗𝜔) = tan−1 (−𝜔𝑇)
√1 + 𝜔2 𝑇2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
The frequency response 𝑒𝑜 (𝑡) = 𝐴. |𝐺(𝑗𝜔)| sin(𝜔𝑡+< 𝐺(𝑗𝜔))
Magnitude of the output 𝑀𝑜 = 𝑀𝑖 . 𝑀 and ∅0 = ∅𝑖 + 𝜃
𝐺(𝑗𝜔) is known as the sinusoidal transfer function. The sinusoidal transfer function, a complex function of
frequency 𝜔, is characterised by its magnitude and phase angle, with frequency as a parameter. There are
three commonly used represetations of sinusoidal transfer functions
1. Bode diagram or logarithmic plot
2. Nyquist plot or polar plot
3. Nichols plot or log magnitude vs phase plot.
BODE DIAGRAM
A Bode diagram consists of two graphs. One is the plot of the logarithm of the magnitude of a sinusoidal
transfer function against the frequency on a logarithmic scale. The other is a plot of the phase angle against
the frequency on a logarithmic scale.
The standard representation of the logarithmic magnitude of 𝐺(𝑗𝜔) is 20 log10 |𝐺(𝑗𝜔)| 𝑑𝐵. The Bode
diagram uses semi-log graph sheet: a linear scale on the y-axis to represent the magnitude in
decibels or phase angle in degrees, and a logarithmic scale on the x-axis to represent the
frequency range of interest. The logarithmic scale allows to consider the characteristics at very
low frequencies.
The main advantage in using the logarithmic plot is the relative ease of plotting the frequency
response curves. When we consider the transfer function of 𝐺(𝑠)𝐻(𝑠) 𝑜𝑟 𝐺(𝑗𝜔)𝐻(𝑗𝜔), the following
factors occur frequently.
1. Gain K
1
2. Integral term and derivative factors 𝑠
𝑠

3. First order factor (𝑠 + 𝑎) 𝑜𝑟 (1 + 𝑇𝑠) 4. Quadratic factors (𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛 2 ) or ( 𝑠 2 + 2𝜁 𝑠 + 1)

2
𝜔𝑛 𝜔𝑛
in numerator or denominator
in numerator or denominator

Gain K
Since K is a constant greater than zero,
20 log10 |𝐺(𝑗𝜔)| = 20 log10 𝐾 𝑑𝐵
which is a constant irrespective of the value of
𝜔
The magnitude plot will be a horizontal line.
< 𝐺(𝑗𝜔) = tan−1 0 = 0
The phase plot will be also a horizontal line at
zero.
The figure shows the Bode diagram for K=10.
𝐺(𝑠) = 10

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3
Integral factor (𝑠 ) (Pole at origin)
1

1 1
𝐺(𝑠) = ≫ 𝐺(𝑗𝜔) =
𝑠 𝑗𝜔
1 𝑗𝜔 −𝑗 1
|𝐺(𝑗𝜔)| = | | = | 2
|=| |=
𝑗𝜔 −𝜔 𝜔 𝜔
Magnitude plot :
1
20 log10 |𝐺(𝑗𝜔)| = 20 log10 ( ) 𝑑𝐵
𝜔
= −20 log10 (𝜔) 𝑑𝐵
Slope of the line is = −20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Phase plot
1
−
< 𝐺(𝑗𝜔) = tan −1 𝜔 = −900
0
Derivative factor (𝑠) (Zero at origin)
𝐺(𝑠) = 𝑠 ≫ 𝐺(𝑗𝜔) = 𝑗𝜔
|𝐺(𝑗𝜔)| = |𝑗𝜔| = 𝜔

Magnitude plot :
20 log10 |𝐺(𝑗𝜔)| = 20 log10 (𝜔) 𝑑𝐵
Slope of the line is = 20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒

Phase plot
𝜔
< 𝐺(𝑗𝜔) = tan−1 = +900
0

If we have 𝐺(𝑠) = 𝑠 𝑛 ≫≫ 𝐺(𝑗𝜔) = (𝑗𝜔)𝑛 ≫≫≫ |𝐺(𝑗𝜔)| = | (𝑗𝜔)𝑛 | = 𝜔𝑛

Magnitude plot : 20 log10 |𝐺(𝑗𝜔)| = 20 log10 ( 𝜔𝑛 ) = 20𝑛 log10 (𝜔) 𝑑𝐵
Slope of the line is = 20𝑛 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Phase plot: < 𝐺(𝑗𝜔) = < (𝑗𝜔)𝑛 = 𝑛 < (𝑗𝜔) = +𝑛. 900

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
𝑛 −𝑗 𝑛 −𝑗 𝑛
If we have 𝐺(𝑠) = 𝑠𝑛 ≫≫ 𝐺(𝑗𝜔) = (𝑗𝜔) = ( 𝜔 )
1 1 1
≫≫≫ |𝐺(𝑗𝜔)| = |( ) | =
𝜔 𝜔𝑛
1
Magnitude plot : 20 log10 |𝐺(𝑗𝜔)| = 20 log10 ( ) = −20𝑛 log10 (𝜔) 𝑑𝐵
𝜔𝑛

Slope of the line is = −20𝑛 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒

−𝑗 𝑛
Phase plot:
−𝑗
< 𝐺(𝑗𝜔) =< ( ) = 𝑛 < ( ) = −𝑛. 900
𝜔 𝜔

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5

First order term (𝑠 + 𝑎) 𝑜𝑟 (1 + 𝑇𝑠) in the denominator (Pole on the real axis)
𝑎 1
𝐺(𝑠) = =
𝑠+𝑎 1 + 𝑇𝑠
1 1 − 𝑗𝜔𝑇 1 𝜔𝑇
𝐺(𝑗𝜔) = = = −𝑗
1 + 𝑗𝜔𝑇 1 + 𝜔2 𝑇2 1 + 𝜔2 𝑇2 1 + 𝜔2 𝑇2
1 𝜔𝑇 1
|𝐺(𝑗𝜔)| = | 2 2
−𝑗 2 2
|=
1+𝜔 𝑇 1+𝜔 𝑇 √1 + 𝜔2 𝑇2

Magnitude plot : Phase plot

−1
20 log10 |𝐺(𝑗𝜔)| = 20 log10 (1 + 𝜔2 𝑇2 ) 2
𝑑𝐵 < 𝐺(𝑗𝜔) = tan−1 (−𝜔𝑇)
= −10 log10 (1 + 𝜔2 𝑇2 ) 𝑑𝐵 If 𝜔𝑇 ≪ 1, tan−1 (−𝜔𝑇) = 0
If 𝜔𝑇 ≪ 1, 1 + 𝜔2 𝑇 2 ≅ 1 then When 𝜔𝑇 = 1, tan−1 (−𝜔𝑇) = −450
20 log10 |𝐺(𝑗𝜔)| = 0
If 𝜔𝑇 ≫ 1, tan−1 (−𝜔𝑇) = −900
𝜔𝑇 ≫ 1, 1 + 𝜔2 𝑇 2 ≅ (𝜔𝑇)2 ,
The frequency 𝜔 = 𝑇 is known as corner frequency or break
1

then 20 log10 |𝐺(𝑗𝜔)| = −10 log10 (𝜔𝑇)2 =

frequency.
−20 log10 𝜔𝑇

When 𝜔𝑇 = 10, 20 log10 |𝐺(𝑗𝜔)| = −20

When 𝜔𝑇 = 100, 20 log10 |𝐺(𝑗𝜔)| = −40
When 𝜔𝑇 = 1000, 20 log10 |𝐺(𝑗𝜔)| = −60
Slope of the line is = −20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Example Asymptotic approximation
2 1
𝐺(𝑠) = =
𝑠+2 1 + 0.5𝑠
Corner frequency 𝜔 = 𝑇 = 0.5 = 2 𝑟𝑎𝑑/𝑠
1 1

𝑀 = −10 log10 (1 + 𝜔2 𝑇2 ), 𝜃 = −tan−1 (𝜔𝑇)

𝜔 𝜔𝑇 Mag(dB) Phase(deg)
0.01 0.005 -0.0001 -0.286
0.1 0.05 -0.011 -2.86
1 0.5 -0.969 -26.56
2 1 -3.010 -45
10 5 -14.15 -78.69
100 50 -33.98 -88.85
Corner frequency=2 rad/s
We can also draw the asymptotes as shown to
get an approximate magnitude plot.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6

First order term (𝑠 + 𝑎) 𝑜𝑟 (1 + 𝑇𝑠) in the numerator (Zero on the real axis)
𝐺(𝑠) = 𝑠 + 𝑎 = 1 + 𝑇𝑠
𝐺(𝑗𝜔) = 1 + 𝑗𝜔𝑇
|𝐺(𝑗𝜔)| = |1 + 𝑗𝜔𝑇| = √1 + 𝜔2 𝑇2
Magnitude plot : Phase plot
1
20 log10 |𝐺(𝑗𝜔)| = 20 log10 (1 + 𝜔2 𝑇2 )2
𝑑𝐵 < 𝐺(𝑗𝜔) = tan−1 (𝜔𝑇)
= 10 log10 (1 + 𝜔2 𝑇2 ) 𝑑𝐵 If 𝜔𝑇 ≪ 1, tan−1 (𝜔𝑇) = 0
If 𝜔𝑇 ≪ 1, 1 + 𝜔2 𝑇 2 ≅ 1 then 20 log10 |𝐺(𝑗𝜔)| = 0 When 𝜔𝑇 = 1, tan−1 (𝜔𝑇) = 450
𝜔𝑇 ≫ 1, 1 + 𝜔2 𝑇 2 ≅ (𝜔𝑇)2 , If 𝜔𝑇 ≫ 1, tan−1 (𝜔𝑇) = 900
then 20 log10 |𝐺(𝑗𝜔)| = 10 log10 (𝜔𝑇)2 = 20 log10 𝜔𝑇

When 𝜔𝑇 = 10, 20 log10 |𝐺(𝑗𝜔)| = 20

When 𝜔𝑇 = 100, 20 log10 |𝐺(𝑗𝜔)| = 40
When 𝜔𝑇 = 1000, 20 log10 |𝐺(𝑗𝜔)| = 60
Slope of the line is = 20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Example
𝐺(𝑠) = 𝑠 + 2 = 1 + 0.5𝑠
Corner frequency 𝜔 = 𝑇 = 0.5 = 2 𝑟𝑎𝑑/𝑠
1 1

𝑀 = 10 log10 (1 + 𝜔2 𝑇2 ), 𝜃 = tan−1 (𝜔𝑇)

𝜔 𝜔𝑇 Mag(dB) Phase(deg)
0.01 0.005 0.0001 0.286
0.1 0.05 0.011 2.86
1 0.5 0.969 26.56
2 1 3.010 45
10 5 14.15 78.69
100 50 33.98 88.85

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7

𝑠 2
A quadratic term
2𝜁
+ 𝑠 + 1 on the numerator (Complex conjugate zeros)
𝜔𝑛 2 𝜔𝑛
𝑠 2 2𝜁
𝐺(𝑠) = + 𝑠+1
𝜔𝑛 2 𝜔𝑛
2𝜔𝜁 (𝑗𝜔)2 𝜔 𝜔 2 𝜔
𝐺(𝑗𝜔) = 1 + 𝑗 + 2
= 1 + 𝑗2𝜁 − 2 = (1 − 𝑟 2 ) + 𝑗 2𝑟𝜁 𝑤ℎ𝑒𝑟𝑒 𝑟 =
𝜔𝑛 𝜔𝑛 𝜔𝑛 𝜔𝑛 𝜔𝑛

|𝐺(𝑗𝜔)| = |(1 − 𝑟2 ) + 𝑗 2𝑟𝜁 | = √(1 − 𝑟2 )2 + (2𝑟𝜁)2

Magnitude plot : Phase plot

1
2𝑟𝜁
20 log10 |𝐺(𝑗𝜔)| = 20 log10 ((1 − 𝑟 2 )2 + (2𝑟𝜁)2 )2 𝑑𝐵 < 𝐺(𝑗𝜔) = tan−1 ( )
= 10 log10 ((1 − 𝑟 2 )2 + (2𝑟𝜁) 2)
𝑑𝐵 1 − 𝑟2
If 𝑟 ≪ 1, then 20 log10 |𝐺(𝑗𝜔)| = 0 If 𝑟 ≪ 1, tan−1 (1−𝑟
2𝑟𝜁
2) ≅ 0

2𝑟𝜁
If 𝑟 ≫ 1, 20 log10 |𝐺(𝑗𝜔)| ≅ 10 log10 𝑟 4 = 40 log10 𝑟 When 𝑟 = 1, tan−1 ( ) = 900
1−𝑟 2
When 𝑟 = 10, 20 log10 |𝐺(𝑗𝜔)| = 40 If 𝑟 ≫ 1, 2𝑟𝜁
tan−1 (1−𝑟2) = 1800
When 𝑟 = 100, 20 log10 |𝐺(𝑗𝜔)| = 80
When 𝑟 = 1000, 20 log10 |𝐺(𝑗𝜔)| = 120
Slope of the line is = 40 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Example
1 2
𝐺(𝑠) = (𝑠 + 4𝑠 + 16) = (1 + 0.25𝑠 + 0.0625𝑠 2 )
16
𝜔𝑛 = 4, 𝜁 = 0.5
Magnitude plot :
20 log10 |𝐺(𝑗𝜔)| = 10 log10 [(1 − 𝑟 2 )2 + (2𝑟𝜁)2 ] 𝑑𝐵
Phase plot : < 𝐺(𝑗𝜔) = tan−1 (1−𝑟 2)
2𝑟𝜁

r 𝜔 = 4𝑟 Mag(dB) Phase(deg)

0.01 0.04 -0.00043 0.572

0.1 0.4 -0.0432 5.76

1 4 0 90
10 40 40 174.23
100 400 80 179.42

The plots can be approxiimated by

asymptotic asuumptions. But around 𝑟 = 1,
the exact shape depends on the value of
damping factor 𝜁. The above example is
solved for values of 𝜁 ranging from 0.1 to 1.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |8

𝑠 2
A quadratic term
2𝜁
+ 𝑠 + 1 on the denominator (Complex conjugate poles)
𝜔𝑛 2 𝜔𝑛
1
𝐺(𝑠) =
𝑠 2 2𝜁
+ 𝑠+1
𝜔𝑛 2 𝜔𝑛
1 1 (1 − 𝑟 2 ) − 𝑗 2𝑟𝜁 (1 − 𝑟 2 ) 2𝑟𝜁
𝐺(𝑗𝜔) = 2 = = = −𝑗
2𝜔𝜁 (𝑗𝜔) (1 − 𝑟 2 ) + 𝑗 2𝑟𝜁 (1 − 𝑟 2 )2 + ( 2𝑟𝜁)2 (1 − 𝑟 2 )2 + ( 2𝑟𝜁)2 (1 − 𝑟 2 )2 + ( 2𝑟𝜁)2
1+𝑗 + 2
𝜔𝑛 𝜔𝑛
√(1 − 𝑟2 )2 + (2𝑟𝜁)2 1
|𝐺(𝑗𝜔)| = =
(1 − 𝑟 ) + ( 2𝑟𝜁)
2 2 2
√(1 − 𝑟2 )2 + (2𝑟𝜁)2
Magnitude plot : Phase plot
−1
2𝑟𝜁
20 log10 |𝐺(𝑗𝜔)| = 20 log10 ((1 − 𝑟 2 )2 + (2𝑟𝜁)2 ) 2 𝑑𝐵 < 𝐺(𝑗𝜔) = −tan−1 ( )
= −10 log10 ((1 − 𝑟 2 )2
+ (2𝑟𝜁) 2)
𝑑𝐵 1 − 𝑟2
If 𝑟 ≪ 1, then 20 log10 |𝐺(𝑗𝜔)| = 0 If 𝑟 ≪ 1, tan−1 (− 1−𝑟
2𝑟𝜁
2) ≅ 0

2𝑟𝜁
If 𝑟 ≫ 1, 20 log10 |𝐺(𝑗𝜔)| ≅ −10 log10 𝑟 4 = −40 log10 𝑟 When 𝑟 = 1, tan−1 (− ) = −900
1−𝑟 2
When 𝑟 = 10, 20 log10 |𝐺(𝑗𝜔)| = −40 If 𝑟 ≫ 1, 2𝑟𝜁
tan−1 (− 1−𝑟2) = −1800
When 𝑟 = 100, 20 log10 |𝐺(𝑗𝜔)| = −80
When 𝑟 = 1000, 20 log10 |𝐺(𝑗𝜔)| = −120
Slope of the line is = −40 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒
Example
16 1
𝐺(𝑠) = =
(𝑠 2 + 4𝑠 + 16) 1 + 0.25𝑠 + 0.0625𝑠 2
𝜔𝑛 = 4, 𝜁 = 0.5
Magnitude plot :
20 log10 |𝐺(𝑗𝜔)| = −10 log10 [(1 − 𝑟 2 )2 +
(2𝑟𝜁)2 ] 𝑑𝐵
Phase plot : < 𝐺(𝑗𝜔) = −tan−1 (1−𝑟 2)
2𝑟𝜁

r 𝜔 = 4𝑟 Mag(dB) Phase(deg)
0.01 0.04 0.00043 -0.572
0.1 0.4 0.0432 -5.76
1 4 0 -90
10 40 -40 -174.23
100 400 -80 -179.42

The plots can be approximated by

asymptotic asuumptions. But around 𝑟 = 1,
the exact shape depends on the value of
damping factor 𝜁. The above example is
solved for values of 𝜁 ranging from 0.1 to 1.
If the damping ratio 𝜁 is zero, the magnitude
of 𝐺(𝑗𝜔) approaches ∞ as 𝑟 → 1. This is the
resonant condition. As 𝜁 varies from 0 to 0.707,
the magnitude is having a peak at a frequency
ratio 𝑟 = √1 − 2𝜁 2 . The maximum value of the
1
magnitude 𝑀𝑝 = 2
2𝜁√1−2𝜁
1
The above results are obtained by differentiating |𝐺(𝑗𝜔)| = with respect to 𝑟 and
√(1−𝑟2 )2 +(2𝑟𝜁)2
equating it to zero.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |9

Example
Draw the Bode diagram for a unity negative feedback system having a forward transfer
𝐾(𝑠+3)
function 𝐺(𝑠) = 𝑠(𝑠+1)(𝑠+2) by taking 𝐾 = 1.
3 𝑠 3 𝑗𝜔
(𝑠+3) (1+ ) (1+ )
𝐺(𝑠) = = 2 3
𝑠 : 𝐺(𝑗𝜔) = 2 3
𝑗𝜔
𝑠(𝑠+1)(𝑠+2) 𝑠(1+𝑠)(1+ ) (𝑗𝜔)(1+𝑗𝜔)(1+ )
2 2

The corner frequencies are 1, 2, 𝑎𝑛𝑑 3. We have

2. A first order term in the numerator (zero at 𝑠 = −3)
3
1. A constant in the numerator
2
3. An integrating factor (pole at the 4. A first order term in the denomiator
origin, 𝑠 = 0) ( pole at 𝑠 = −1)
5. A first order term in the denomiator (pole at 𝑠 = −2)
Magnitude plot:
3 𝑗𝜔 𝑗𝜔
(1+ ) |1.5| |1+ |
2 3 3
20 log10 |𝐺(𝑗𝜔)| = 20 log10 | 𝑗𝜔 | = 20 log10 { 𝑗𝜔 }
(𝑗𝜔)(1+𝑗𝜔)(1+ ) |𝑗𝜔||1+𝑗𝜔||1+ |
2 2
3 𝑗𝜔 𝑗𝜔
= 20 log10 | | + 20 log10 |1 + | − 20 log10 |𝑗𝜔| − 20 log10 |1 + 𝑗𝜔| − 20 log10 |1 + |
2 3 2
Phase Plot:
3 𝑗𝜔
(1+ ) 3 𝑗𝜔 𝑗𝜔
2 3
< 𝐺(𝑗𝜔) =< [ 𝑗𝜔 ] =< ( ) +< (1 + ) −< (𝑗𝜔)−< (1 + 𝑗𝜔)−< (1 + )
(𝑗𝜔)(1+𝑗𝜔)(1+ ) 2 3 2
2
1.5 (1 + 𝑗0.33𝜔) (𝑗𝜔)−1 (1 + 𝑗𝜔)−1 (1 + 𝑗0.5𝜔)−1
𝑇 = 0.33 𝑇=1 𝑇 = 0.5
Σ

Mag(dB) Phase(deg) Mag Phase Mag Phase Mag Phase Mag Phase Mag Phase
𝜔 −10 log(1 tan−1 (−𝜔𝑇)
20 log10 𝐾 tan−1 0 10 log(1 tan−1 (𝜔𝑇) −20 log(𝜔) 1
−𝜔 −10 log(1 tan−1 (−𝜔𝑇) dB deg
+ 𝜔2 𝑇 2 )
+ 𝜔2 𝑇 2 ) tan−1 + 𝜔2 𝑇 2 )
0
0.01 3.52 0 0.00005 0.19 40 -90 -0.0004 -0.573 -0.0001 -0.286 43.5 -90.
0.1 3.52 0 0.0048 1.91 20 -90 -0.043 -5.71 -0.0108 -2.86 23.5 -97
1 3.52 0 0.458 18.44 0 -90 -3.010 -45 -0.969 -26.56 0 -143
2 3.52 0 1.6 33.69 -6.02 -90 -6.989 -63.43 -3.010 -45 -11 -164
3 3.52 0 3 45 -9.54 -90 -10 -71.56 -5.118 -56.31 -18 -173
10 3.52 0 10.83 73.3 -20 -90 -20 -84.28 -14.15 -78.69 -40 -
179.7
100 3.52 0 30.45 88.28 -40 -90 -40 -89.42 -33.98 -88.85 -80 -180
1000 3.52 0 50.45 89.8 -60 -90 -60 -89.94 -53.98 --89.89 -120 -180

The plots of individual factors and that of combined transfer function are shown below.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
P a g e | 10

Example:
10(𝑠+3)
Draw the Bode diagram for the following transfer function 𝐺(𝑠) = 𝑠(𝑠+2)(𝑠2 +𝑠+2)
𝑟𝑎𝑑
The quadratic factor in the denominator gives 𝜔𝑛 = 1.414 𝑠
𝑎𝑛𝑑 𝜁 = 0.353
𝑠 𝑗𝜔
10 ×3(1+ ) 7.5(1+
)
𝐺(𝑠) = 𝑠
3
𝑠 𝑠2
: 𝐺(𝑗𝜔) = 𝑗𝜔
3
𝑗𝜔 (𝑗𝜔)2
2×2×𝑠(1+ )(1+ + ) (𝑗𝜔)(1+ )[1+ + ]
2 2 2 2 2 2

Corner frequencies are 1.414, 2, 3 𝑟𝑎𝑑/𝑠

Magnitude plot:
𝑗𝜔
7.5(1+ 3 )
20 log10 |𝐺(𝑗𝜔)| = 20 log10 | |
𝑗𝜔 𝑗𝜔 (𝑗𝜔)2
(𝑗𝜔)(1+ 2 )[1+ 2 + 2 ]
2
𝑗𝜔 𝑗𝜔 𝑗𝜔 (𝑗𝜔)
= 20 log10 |7.5| + 20 log10 |1 + | − 20 log10 |𝑗𝜔| − 20 log10 |1 + | − 20 log10 |1 + + 2 |
3 2 2
Phase Plot:
𝑗𝜔 2
7.5(1+ 3 ) 𝑗𝜔 𝑗𝜔 𝑗𝜔 (𝑗𝜔)
< 𝐺(𝑗𝜔) =< [ ] =< (7.5)+< (1 + ) −< (𝑗𝜔)−< (1 + ) −< (1 + + 2 )
𝑗𝜔 𝑗𝜔 (𝑗𝜔)2 3 2 2
(𝑗𝜔)(1+ 2 )[1+ 2 + 2
]
−1
7.5 (1 + 𝑗0.33𝜔) (𝑗𝜔)−1 (1 + 𝑗0.5𝜔)−1 2
𝑇 = 0.33 𝑇 = 0.5 1 + 𝑗𝜔 + (𝑗𝜔) Σ
2 2
( )

𝜔𝑛 = 1.414 𝑎𝑛𝑑 𝜁 = 0.353

𝜔 Mag Phase Mag Phase Mag Phase Mag Phase − tan−1 (

2𝑟𝜁 Mag Phase
−10 log (1 − 𝑟 )2
2 2 )
1 − 𝑟2
−10 log(1 tan−1 (−𝜔𝑇)
20 log10 𝐾 tan−1 0 10 log(1 tan−1 (𝜔𝑇) −20 log(𝜔) −
1 +(2𝑟𝜁) dB deg
+ 𝜔2 𝑇 2 )
+ 𝜔2 𝑇 2 ) tan−1 𝜔
0
0.01 17.5 0 0.00005 0.19 40 -90 -0.0001 -0.286 r=0.007 0.0003 -0.286 57.5 -90
0.1 17.5 0 0.0048 1.91 20 -90 -0.0108 -2.86 r=0.07 0.0326 -2.87 37.52 -93.8
1 17.5 0 0.458 18.44 0 -90 -0.969 -26.56 r=0.707 0.302 -45 17.29 -143.
1.414 17.5 0 0.871 25.24 -3.01 -90 -1.76 -35.26 r=1 3.02 -90 16.62 -190
2 17.5 0 1.6 33.69 -6.02 -90 -3.010 -45 r=1.414 -3 -135 7.08 -236
3 17.5 0 3 45 -9.54 -90 -5.118 -56.31 r=2.12 -11.6 -156.8 -5.8 -258
10 17.5 0 10.83 73.3 -20 -90 -14.15 -78.69 r=7.07 -33.8 -174.18 -39.15 -269
100 17.5 0 30.45 88.28 -40 -90 -33.98 -88.85 r=70.7 -73.8 -179.43 -100 -270
1000 17.5 0 50.45 89.8 -60 -90 -53.98 --89.89 r=707 -113.8 -179.95 -160 270

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |1
PHASE AND GAIN MARGIN
Phase margin, FM.
The phase margin is the change in open-loop phase shift required at unity gain to make the closed-
loop system unstable.
Phase margin is that amount of additional phase lag at the gain cross over frequency required to
bring the system to the verge of instability. Gain cross over frequency is the frequency at which
|𝐺(𝑗𝜔)| = 1. The phase margin 𝛾 is 1800 plus the phase angle < 𝐺(𝑗𝜔) of the open-loop transfer
function at the gain cross over frequency
𝛾 = 1800 +< 𝐺(𝑗𝜔)
For a system to be stable, the phase margin should be positive.
Gain margin, GM.
The gain margin is the change in open-loop gain, expressed in decibels (dB), required at 1800 of
phase shift to make the closed-loop system unstable.
Gain margin is the reciprocal of the magnitude |𝐺(𝑗𝜔)| at the froquency at which the phase angle
is −1800
The phase cross over frequency 𝜔1 is defined as the frequency at which the phase angle of open
loop transfer function equals −1800 .
1 1
Gain margin 𝐾𝑔 = |𝐺(𝑗𝜔 . In terms of decibels, 𝐾𝑔 𝑑𝐵 = 20 log |𝐺(𝑗𝜔 = −20 log|𝐺(𝑗𝜔1 )|
1 )| 1 )|

The gain margin in dB is positive if 𝐾𝑔 > 1 and negative if 𝐾𝑔 < 1. A positive gaian margin in dB
means the system is stable.
The figure below shows the gain margin and phase margin of a stable and unstable systems.

Gain cross over frequency

phase cross over frequency

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |2
The figures below show the Bode diagram of the examples we solved above. Identify the gain and
phase margins and comment on the stability of the systems.

-ve gain margin

Infinite gain margin

-ve phase -ve phase margin

margin

Example 1
Obtain the phase and gain margins of the
system shown for the two cases where 𝐾 = 10
and 𝐾 = 100.

The phase and gain margins for 𝐾 = 10 are Phase margin= 21°, Gain margin= 8 𝑑𝐵 Therefore,the
system gain may be increased by 8 𝑑𝐵 before the instability occurs. Increasing the gain from 𝐾 =
10 𝑡𝑜 𝐾 = 100 shifts the 0 𝑑𝐵 axis down by 20 𝑑𝐵,as shown.The phase and gain margins are Phase
margin=– 30°, Gain margin=– 12 𝑑𝐵 Thus,the system is stable for 𝐾 = 10, but unstable for 𝐾 = 100

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |3

Example 2
𝐾
For the system whose transfer function is 𝐺(𝑠) = (𝑠+5)(𝑠+20)(𝑠+50), do the following:
a. Draw the Bode plot.
b. Find the range of K for stability from your Bode plots.
c. Evaluate gain margin, phase margin, zero dB frequency, and 1800 frequency from your Bode
plots for 𝐾 = 10000.

gain margin of 19.7 dB.

Phase margin is 92.70
19.7
20 log 𝑘 = 19.7 ≫≫ 𝑘 = 10 20 = 9.66
Therefore, the gain can be increased to 𝑘 × 10000 = 96600 before the system becomes unstable.

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |4
Relation between system type and Bode plots
Consider the unity feedback control system. The
position, velocity , and acceleration error constants
describe the low frequency behavior of type 0, type
1, and type 2 systems respectively.

For a given system, only one of the static error constants in finite and significant.
The type of the system determines the slope of the log-magnitude curve at low frequencies. Hence
information concerning the magnitude of the steady state error of a control system to a given
input can be obtained from the observation of the low frequency region of the log-magnitude plot.
Position Error Constant
Let us assume that G(s) represents a type 0 system
𝐾 (1 + 𝑇𝑎 𝑠)(1 + 𝑇𝑏 𝑠) … … (1 + 𝑇𝑚 𝑠)
𝐺(𝑠) =
(1 + 𝑇1 𝑠)(1 + 𝑇2 𝑠) … … (1 + 𝑇𝑛 𝑠)
𝐾𝑝 = lim 𝐺(𝑠) = 𝐾
𝑠→0
𝐾 (1 + 𝑗𝜔𝑇𝑎 )(1 + 𝑗𝜔𝑇𝑏 ) … … (1 + 𝑗𝜔𝑇𝑚 )
𝐺(𝑗𝜔) =
(1 + 𝑗𝜔𝑇1 )(1 + 𝑗𝜔𝑇2 ) … … (1 + 𝑗𝜔𝑇𝑛 )
20 log|𝐺(𝑗𝜔)| = 20 log 𝐾 + 20 log|1 + 𝑗𝜔𝑇𝑎 | + 20 log|1 + 𝑗𝜔𝑇𝑏 | … . +20 log|1 + 𝑗𝜔𝑇𝑚 | − 20 log|1 + 𝑗𝜔𝑇1 | … … . . −20 log|1 + 𝑗𝜔𝑇𝑛 |

At very low frequencies (of the order 𝜔 →

0.01, 𝑜𝑟 0.001; 20 log|𝐺(𝑗𝜔)| → 20 log 𝐾 =
20 log 𝐾𝑝
The low frequency asymptote is a horizontal
line at 20 log10 𝐾𝑝

100(𝑠+5)
Example 3. 𝐺(𝑠) =
(𝑠+1)(𝑠+2)(𝑠+3)

100(𝑠+5)
𝐾𝑝 = lim 𝐺(𝑠) = lim = 83.3
𝑠→0 𝑠→0 (𝑠+1)(𝑠+2)(𝑠+3)
The Bode diagram of this system is shown.
From the magnitude plot, it can be seen that
the log-magnitude curve is horizontal when
the frequency is small. The log magnitude at
low frequency is apprximately equal to
20 log10 𝐾𝑝 .
i.e., 20 log10 𝐾𝑝 = 38.4 𝑑𝐵
38.4
𝐾𝑝 = 10 20 = 83.2

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |5
Velocity Error Constant
Let us assume that G(s) represents a type 1 system
𝐾 (1+𝑇𝑎 𝑠)(1+𝑇𝑏 𝑠)……(1+𝑇𝑚 𝑠)
𝐺(𝑠) = 𝑠(1+𝑇1 𝑠)(1+𝑇2 𝑠)……(1+𝑇𝑝 𝑠)
𝐾 (1+𝑇𝑎 𝑠)(1+𝑇𝑏 𝑠)……(1+𝑇𝑚 𝑠)
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim
𝑠→0 𝑠→0 (1+𝑇1 𝑠)(1+𝑇2 𝑠)……(1+𝑇𝑝 𝑠)
𝐾 (1 + 𝑗𝜔𝑇𝑎 )(1 + 𝑗𝜔𝑇𝑏 ) … … (1 + 𝑗𝜔𝑇𝑚 )
𝐺(𝑗𝜔) =
(𝑗𝜔)(1 + 𝑗𝜔𝑇1 )(1 + 𝑗𝜔𝑇2 ) … … (1 + 𝑗𝜔𝑇𝑝 )
The intersection of the initial −20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒 line or its extension
with the vertical line 𝜔 = 1 will give 20 log 𝐾𝑣
1 𝐾 (1+𝑗𝜔𝑇𝑎 )(1+𝑗𝜔𝑇𝑏 )……(1+𝑗𝜔𝑇𝑚 ) 𝐾𝑣
When 𝜔 ≪ 1, 𝐺(𝑗𝜔) = lim =
𝑗𝜔 𝑗𝜔→0 (1+𝑗𝜔𝑇1 )(1+𝑗𝜔𝑇2 )……(1+𝑗𝜔𝑇𝑝 ) 𝑗𝜔
𝐾𝑣
20 log10 |𝐺(𝑗𝜔)| = 20 log10 | | = 20 log10 𝐾𝑣 −20 log10 |𝑗𝜔| = 20 log10 𝐾𝑣 −20 log10 𝜔
𝑗𝜔
When 𝜔 = 1; 20log10 |𝐺(𝑗𝜔)|=20log10 𝐾𝑣 or 𝐾𝑣 = |𝐺(𝑗𝜔)|𝜔=1
Let the initial −20 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒 line or its extension meets the x-axis at 𝜔1 . Then 20 log10 |𝐺(𝑗𝜔)| = 0,
Hence 20 log10 𝐾𝑣 −20 log10 𝜔1 = 0; or 𝜔1 = 𝐾𝑣
10
Example 4 𝐺(𝑠) = 𝑠(𝑠+4)

10
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 𝑠(𝑠+4) = 2.5
𝑠→0 𝑠→0
10
𝐾𝑣 = |𝐺(𝑗𝜔)|𝜔=1 = | |
(𝑗𝜔)(𝑗𝜔+4)
𝜔=1

|10|
= {|(𝑗𝜔)|| }
(𝑗𝜔+4)| 𝜔=1
= |(𝑗)|||10|
(𝑗+4)|
= 10
= 2.42
1×√17

From the graph, 7.67 𝑑𝐵=20log10𝐾𝑣

7.67
𝐾𝑣 = 10 20 = 2.42
10(𝑠+5)
Example 5 𝐺(𝑠) = 𝑠(𝑠+4)(𝑠+3)

10(𝑠+5)
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 𝑠(𝑠+4)(𝑠+3) = 4.17
𝑠→0 𝑠→0
10(𝑗𝜔+5)
𝐾𝑣 = |𝐺(𝑗𝜔)|𝜔=1 = | |
(𝑗𝜔)(𝑗𝜔+4)(𝑗𝜔+3)
𝜔=1

10(𝑗𝜔+5)
= |(𝑗)|||10||(𝑗+5)|
= 1×10 26
√
={ }
(𝑗𝜔)(𝑗𝜔+4)(𝑗𝜔+3) 𝜔=1 (𝑗+4)||(𝑗+3)| √17√10
= 3.9

From the graph, 11.8 𝑑𝐵=20log10𝐾𝑣

11.8
𝐾𝑣 = 10 20 = 3.9

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |6
Acceleration Error Constant
Let us assume that G(s) represents a type 2 system
𝐾 (1+𝑇𝑎 𝑠)(1+𝑇𝑏 𝑠)……(1+𝑇𝑚 𝑠)
𝐺(𝑠) =
𝑠2 (1+𝑇1 𝑠)(1+𝑇2 𝑠)……(1+𝑇𝑝 𝑠)
𝐾 (1+𝑇𝑎 𝑠)(1+𝑇𝑏 𝑠)……(1+𝑇𝑚 𝑠)
𝐾𝑎 = lim 𝑠 2 𝐺(𝑠) = lim
𝑠→0 𝑠→0 (1+𝑇1 𝑠)(1+𝑇2𝑠)……(1+𝑇𝑝 𝑠)
𝐾 (1 + 𝑗𝜔𝑇𝑎 )(1 + 𝑗𝜔𝑇𝑏 ) … … (1 + 𝑗𝜔𝑇𝑚 )
𝐺(𝑗𝜔) =
(𝑗𝜔)2 (1 + 𝑗𝜔𝑇1 )(1 + 𝑗𝜔𝑇2 ) … … (1 + 𝑗𝜔𝑇𝑝 )
The intersection of the initial −40 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒 line or its extension with the vertical line 𝜔 = 1 will give
20 log 𝐾𝑎
1 𝐾 (1+𝑗𝜔𝑇𝑎 )(1+𝑗𝜔𝑇𝑏 )……(1+𝑗𝜔𝑇𝑚 ) 𝐾
When 𝜔 ≪ 1, 𝐺(𝑗𝜔)
𝑎
= (𝑗𝜔)2 lim = (𝑗𝜔) 2
𝑗𝜔→0 (1+𝑗𝜔𝑇1 )(1+𝑗𝜔𝑇2 )……(1+𝑗𝜔𝑇𝑝 )
𝐾𝑎
20 log10 |𝐺(𝑗𝜔)| = 20 log10 | 2
| = 20 log10 𝐾𝑎 −40 log10 |𝑗𝜔| = 20 log10 𝐾𝑎 −40 log10 𝜔
(𝑗𝜔)
When 𝜔 = 1; 20log10 |𝐺(𝑗𝜔)|=20log10 𝐾𝑎 or 𝐾𝑎 = |𝐺(𝑗𝜔)|𝜔=1
Let the initial −40 𝑑𝐵/𝑑𝑒𝑐𝑎𝑑𝑒 line or its
extension meets the x-axis at 𝜔𝑎 . Then
20 log10 |𝐺(𝑗𝜔)| = 0, Hence
20 log10 𝐾𝑎 −40 log10 𝜔𝑎 = 0; or 𝜔𝑎 = √𝐾𝑎

10(𝑠+5)
Example 6 𝐺(𝑠) =
𝑠2 (𝑠+4)(𝑠+3)

10(𝑠+5)
𝐾𝑎 = lim 𝑠 2 𝐺(𝑠) = lim 𝑠 2 𝑠2 (𝑠+4)(𝑠+3) = 4.17
𝑠→0 𝑠→0
10(𝑗𝜔+5)
𝐾𝑎 = |𝐺(𝑗𝜔)|𝜔=1 = | 2 |
(𝑗𝜔) (𝑗𝜔+4)(𝑗𝜔+3)
𝜔=1
|10||(𝑗+5)| 10√26
= = = 3.9
|(𝑗)2 ||(𝑗+4)||(𝑗+3)| 1×√17√10

From the graph, 11.7 𝑑𝐵=20log10𝐾𝑎

11.8
𝐾𝑎 = 10 20 = 3.9
It can be also seen from the plot that log
magnitude is equal to 0 dB when 𝜔𝑎 = 1.85.
Therefore, 𝐾𝑎 = 𝜔𝑎 2 = 3.4

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |7
Correlation between step transient response and open loop frequency response
The maximum overshoot in the unit step response of a standard second order system can be
exactly correlated with the resonant peak magnitude in the frequency response. Hence
essentially the same information about the system dynamics is contained in the frequency
response as in the transient response

Consider a unity negative feed back system with forward path transfer function
𝜔𝑛 2 𝐶(𝑠) 𝜔𝑛 2
𝐺(𝑠) = . The closed loop transfer function will be 𝑇(𝑠) = = . For a unit
𝑠(𝑠+2𝜁𝜔𝑛 ) 𝑅(𝑠) (𝑠 +2𝜁𝜔𝑛 𝑠+𝜔𝑛 2 )
2
1
step input, the output 𝑐(𝑡) = 1 − 𝑒−𝜁𝜔𝑛𝑡 sin(𝜔𝑑 𝑡 + 𝜃) where 𝜔𝑑 = 𝜔𝑛 √1 − 𝜁 2. The
√1−𝜁2

−
𝜋𝜁
√𝟏−𝜻𝟐
maximum overshoot for the unit step response is given by 𝑀𝑝 = 𝑒
𝜔𝑛 2
𝐺(𝑗𝜔) =
𝑗𝜔(𝑗𝜔 + 2𝜁𝜔𝑛 )
2 |𝜔𝑛 2 |
𝜔𝑛 𝜔𝑛 2
|𝐺(𝑗𝜔)| = | |= =
𝑗𝜔(𝑗𝜔 + 2𝜁𝜔𝑛 ) |𝑗𝜔||𝑗𝜔 + 2𝜁𝜔𝑛 | 𝜔 √𝜔 2 + 4𝜁 2 𝜔𝑛 2
𝜔
< 𝐺(𝑗𝜔) = 0 − 90 − tan−1
2𝜁𝜔𝑛
𝜔𝑛 2
When |𝐺(𝑗𝜔)| = 1, = 1 ≫≫ 𝜔𝑛 2 = 𝜔 √𝜔 2 + 4𝜁 2 𝜔𝑛 2 ≫≫ 𝜔𝑛 4 = 𝜔2 (𝜔2 + 4𝜁 2 𝜔𝑛 2 )
𝜔 √𝜔 +4𝜁 2 𝜔𝑛 2
2

𝜔4 + 4𝜁 2 𝜔𝑛 2 𝜔2 − 𝜔𝑛 4 = 0
−4𝜁 2 𝜔𝑛 2 ± √16𝜁 4 𝜔𝑛 4 + 4𝜔𝑛 4
𝜔2 = = 𝜔𝑛 2 (−2𝜁 2 ± √1 + 𝜁 4 )
2

𝜔 = 𝜔𝑛 √−2𝜁 2 + √1 + 𝜁 4

ie., 20 log|𝐺(𝑗𝜔)| = 0 𝑑𝐵 when 𝜔 = 𝜔𝑛 √−2𝜁 2 + √1 + 𝜁 4 and the corresponding phase angle is

𝜔𝑛 √−2𝜁 2 + √1 + 𝜁 4
−1
< 𝐺(𝑗𝜔) = −90 − tan
2𝜁𝜔𝑛
𝜔𝑛 √−2𝜁 2 +√1+𝜁 4 2𝜁
The Phase margin 𝛾 = 180+< 𝐺(𝑗𝜔) = 90 − tan−1 2𝜁𝜔𝑛
= tan−1
√−2𝜁 2 +√1+𝜁 4

The frequency 𝜔𝐵 at which the magnitude of the closed loop frequency response is 3 𝑑𝐵 below its
zero frequency ( low frequency) value is called the cut-off frequency. The closed loop system
filters out signal components whose frequencies are greater than the cut-off frequency and
transmits those signal components with frequencies lower than the cut-off frequency. The
frequency range 0 ≤ 𝜔 ≤ 𝜔𝐵 in which the magnitude of the closed loop does not drop −3 𝑑𝐵 is called
the band width of the system.
The bandwidth 𝜔𝐵 is a system’s ability to faithfully reproduce an input signal, or how well the
system can track the input sinusoid. The resonant frequency 𝜔𝑟 and −3 𝑑𝐵 bandwidth can be
related to the speed of transient response. As the bandwidth 𝜔𝐵 increases, the rise time of the
step response of the system will decrease. The overshoot to a step input can be related to 𝑀𝑝
through the damping ratio 𝜁. The bandwidth of a system 𝜔𝐵 can be approximated as 𝜔𝑛

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |8

Example 7
100
A negative unity feedback system has an open loop transfer function 𝐺(𝑠) = 𝑠(𝑠+5), determine
gain margin, phase margin, gain cross over frequency, phase cross over frquency, the velocity
error constant from the open loop frequency response.
Also determine the resonant frequency, resonant magnitude and band width from the closed
loop frequency response.
Open-loop frequency response
100 100 20
𝐺(𝑠) = ≫≫ 𝐺(𝑗𝜔) = =
𝑠(𝑠 + 5) 𝑗𝜔 𝑗𝜔
(𝑗𝜔)5 (1 + ) (𝑗𝜔) (1 + )
5 5
20 (𝑗𝜔)−1 (1 + 𝑗0.2𝜔)−1 Σ
𝑇 = 0.2

𝜔 Mag Phase Mag Phase Mag Phase Mag Phase

20 log10 𝐾 tan−1 0 −20 log(𝜔) 1 −10 log(1 + 𝜔2 𝑇 2 ) tan−1 (−𝜔𝑇) dB deg
−
tan −1 𝜔
0
0.01 26.02 0 40 -90 0 -0.11 66 -90
0.1 26.02 0 20 -90 0 -1.1 46 -91.1
1 26.02 0 0 -90 0 -11.3 26 -101.3.
5 26.02 0 -14 -90 -3 -45 9 -135
10 26.02 0 -20 -90 -7 -63.4 6 -153.4
100 26.02 0 -40 -90 -26 -87.3 -40 -177.3
1000 26.02 0 -60 -90 -46 --89.8 -80 -179.8

gain_margin = Inf

phase_margin = 28.02 deg

gain_crossover_freq = 9.39

phase_crossover_freq= Inf

At 𝜔 = 1, 20 log|𝐺(𝑗𝜔)| = 25.8 𝑑𝐵
25.8
Therefore, the position error constant 𝐾𝑣 = 10 20 = 19.5

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
Page |9

2𝜁 2𝜁 4𝜁 2
𝛾 = tan−1 ≫ = 0.639 ≫ = 0.4 ≫
√−2𝜁 2 √−2𝜁 2 −2𝜁 2 + √1 + 𝜁 4
+ √1 + 𝜁4 + √1 + 𝜁4

4𝜁 2 = −0.8𝜁 2 + 0.4√1 + 𝜁 4 ≫≫ 144𝜁 4 = (1 + 𝜁 4 ) ≫≫ 𝜁 = 0.289

−𝜋𝜁
√1−𝜁 2
%𝑂𝑆 = 𝑒 = 38.7%
1 1
The closed-loop transfer function 𝑇(𝑠) = 0.5𝑠3 +1.5𝑠2 +𝑠+1
= (𝑠+2.5214)(𝑠2 +0.4786𝑠+0.7933)

resonant_peak = 6.04 dB
resonant_freq=8.69 rad/s
bandwidth=15 rad/s

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT
P a g e | 10
Matlab code for Bode plot
clc
nump=[100];
denp=[1 5 0];
sysp=tf(nump,denp) % open loop transfer function
sys=feedback(sysp,1) % unity negative feedback loop
w=logspace(-3,3); % frequency range from 0.001 to 1000
bode(sysp,w) % open loop frequency response

[Gm,pm, wcp, wcg]=margin(sysp);

GmdB=20*log10(Gm);
gain_margin= GmdB
phase_margin=pm
gain_crossover_freq=wcg
phase_crossover_freq=wcp
hold on
bode(sys,w) % closed loop frequency response
[mag,phase,w]=bode(sys,w);

[Mp,k]=max(mag);
resonant_peak=20*log10(Mp)
resonant_freq=w(k)
n=1;
while (20*log(mag(n)))>-3; % checking log magnitude is > -3dB

n=n+1;

end
bandwidth=w(n)
mag_dB=20*log10(mag(n))

References
1. Modern Control Systems : Richard C. Dorf and Robert H. Bishop, 8/e, Addison-Wesley
2. Modern Control Engineering : Katsuhiko Ogatta, Pearson Education, LPE
3. Control Systems Engineering: Norman S. Nise, 6/e, John Wiley & Sons

Control Engineering …………………………Dr. Biju N., Professor in Mechanical Engg, SOE, CUSAT