Control Systems ELEC3114

Time Response and

Feedback Control
System Characteristics
Dr. Arash Khatamianfar

School of Electrical Engineering and Telecommunications

T2 2021
Content
• Characteristics of the step response of
• First-order systems
• Second-order systems
• Different types of step response for second- Mainly from:
order systems without zero. Nise Textbook
Chapter 4
• Overdamped,
• Critically damped, And Dorf Textbook,
• Undamped, Chapter 5.
• Underdamped.
• The general form of second-order systems
• Effects of the third pole and a zero on the
second-order system response
• Impact of disturbance and noise in a closed-
loop feedback system

Page 1
First-order step response
• The majority of dynamic systems can be modelled or approximated by
either a 1st-order or a 2nd-order LTI system (in transfer function or state
space form).
• We are interested in analysing the impact of the poles and zeros of the
system on the output response in time domain.
• In all cases, we assume that the input signal is a unit-step function.
• Let's begin with 1st-order system:

𝑌 𝑠 𝐾𝑑𝑐 𝐾
𝐺 𝑠 = = ≡
𝑅 𝑠 𝜏𝑠 + 1 𝑠 + 𝑎
𝐾
Pole of the system: 𝑠 = −𝑎 and the DC gain: 𝐺 0 = 𝐾𝑑𝑐 =
𝑎

Time constant of the 1

system: 𝜏=
𝑎

Time constant is the reciprocal of the

⟹
absolute value of the pole of the system
Page 2
First-order step response (2)
• Output response to a unit step input:

𝐾
𝑦 𝑡 = ℒ −1 𝐺 𝑠 𝑅 𝑠 = ℒ −1
𝑠(𝑠 + 𝑎)

𝐾
𝑦 𝑡 = 1 − 𝑒 −𝑎𝑡 𝑢(𝑡)
𝑎

Assuming 𝑎 > 0 (negative pole): • Poles are denoted by

a cross sign
𝑦(𝑡) • Zeros are denoted by
𝑗𝜔 a circle sign
𝐾
Left Hand Side Right Hand Side 𝑦𝑠𝑠 = 𝐾𝑑𝑐 =
𝑎
LHS RHS
0.63𝑦𝑠𝑠
𝜎
−𝑎
𝑡
1
𝜏=
𝑎
𝑠-plane

Page 3
First-order step response (3)
• Characteristics of the 1st-order step response:
𝐾
• Time Constant 𝜏 is the time it takes for the step response to 𝑦 𝑡 = 1 − 𝑒 −𝑎𝑡 𝑢(𝑡)
𝑎
rise to 63% of its final value.
𝑦(𝑡) (assuming 𝐾 = 𝑎)
• So for a first-order system, the farther the pole is from the
imaginary axis on the LHS, the faster the transient part of the
step response.

• Rise time 𝑇𝑟 is defined as the time for the response

to go from 10% to 90% of its final value.
2.2
𝑇𝑟 =
𝑎

• Settling time 𝑇𝑠 is defined as the time for the response to

reach and stay (settle) within 2% of its final value.
• This is also known as 2% settling time as in some textbooks, 5%
of final value is used to define the setting time.

4
𝑇𝑠 = = 4𝜏
𝑎

Page 4
Second-order step response
• Let’s first consider the output response of a second
order system under four different cases with no zeros.
• Case 1: Two real and distinct poles:
𝑌 𝑠 𝐾 Poles of the system: 𝑠1,2 = −𝛼, −𝛽
𝐺 𝑠 = = ⟹ 1 1
𝑅 𝑠 (𝑠 + 𝛼)(𝑠 + 𝛽) Time constants of the system: 𝜏1,2 = ,,
𝛼 𝛽
𝐾 𝐾1 𝐾2 𝐾3 𝐾
𝑦 𝑡 = ℒ −1 𝐺 𝑠 𝑅 𝑠 = ℒ −1 = ℒ −1 + + 𝐾1 =
𝛼𝛽
= 𝐾𝑑𝑐
𝑠(𝑠 + 𝛼)(𝑠 + 𝛽) 𝑠 𝑠+𝛼 𝑠+𝛽 −𝐾
𝐾2 =
𝛼(𝛽 − 𝛼)
−𝐾
⟹ 𝑦 𝑡 = 𝐾1 + 𝐾2 𝑒 −𝛼𝑡 + 𝐾3 𝑒 −𝛽𝑡 𝑢(𝑡) 𝐾3 =
𝛽(𝛼 − 𝛽)
𝑦(𝑡) (assuming 𝐾 = 𝛼𝛽 )

This response is known as

an overdamped response
−𝛽 −𝛼

𝛼, 𝛽 > 0 (negative poles) Page 5

Second-order step response (2)
Dominant pole
• An overdamped system has two time constants.
• The larger time constant has a much bigger impact on
the response.
−𝛽 −𝛼
• In other words, the exponential term in the response
with the pole closer to the origin would decay to zero
much slower, and the other exponential term with the 𝑦(𝑡) (assuming 𝐾 = 𝛼𝛽)
pole farther to the LHS would decay much faster.

• Assuming 𝛼 ≪ 𝛽 :

𝑒 −𝛼𝑡 → 0 much slower than 𝑒 −𝛽𝑡 → 0

as 𝑡 → ∞
𝑌 𝑠 𝐾
𝐺 𝑠 = =
𝑅 𝑠 (𝑠 + 𝛼)(𝑠 + 𝛽)
The poles of a system that are closer to the origin
are called dominant poles. They dominate the 𝑦 𝑡 = 𝐾1 + 𝐾2 𝑒 −𝛼𝑡 + 𝐾3 𝑒 −𝛽𝑡 𝑢(𝑡)
shape of the response because they decay to zero
more slowly than the other poles.
Page 6
Second-order step response (3)
• Case 2: Two real and repeated poles:

𝑌 𝑠 𝐾 Poles of the system: 𝑠1,2 = −𝑎, −𝑎

𝐺 𝑠 = = 2
⟹ 1
𝑅 𝑠 𝑠+𝑎 Time constant of the system: 𝜏 =
𝑎
𝐾 𝐾1 𝐾2 𝐾3 𝐾
𝑦 𝑡 = ℒ −1 𝐺 𝑠 𝑅 𝑠 = ℒ −1 2
= ℒ −1 + 2
+ 𝐾1 =
𝑎2
= 𝐾𝑑𝑐
𝑠 𝑠+𝑎 𝑠 𝑠+𝑎 𝑠+𝑎 −𝐾
𝐾2 =
𝑎
⟹ 𝑦 𝑡 = 𝐾1 + 𝐾2 𝑡𝑒 −𝑎𝑡 + 𝐾3 𝑒 −𝑎𝑡 𝑢(𝑡) −𝐾
𝐾3 = 2
𝑎

This response is known as

a critically damped
𝑦(𝑡) (assuming 𝐾 = 𝑎2 ) response

The critically damped

−𝑎 response is the fastest
response without overshoot
𝑎 > 0 (negative poles)
compared to the overdamped
response (𝛼 < 𝑎 < 𝛽 ).
Page 7
Second-order step response (4)
• Case 3: Two imaginary (conjugate) poles:

𝑌 𝑠 𝐾 Poles of the system: 𝑠1,2 = ±𝑗𝜔𝑛

𝐺 𝑠 = = 2 ⟹
𝑅 𝑠 𝑠 + 𝜔𝑛2 Time constant of the system: 𝜏 = 𝑁/𝐴

𝐾 𝐾1 𝐾2 𝑠 + 𝐾3
𝑦 𝑡 = ℒ −1 𝐺 𝑠 𝑅 𝑠 = ℒ −1 =ℒ −1 + 2 𝐾1 =
𝐾
= 𝐾𝑑𝑐
𝑠(𝑠 2 + 𝜔𝑛2 ) 𝑠 𝑠 + 𝜔𝑛2 𝜔𝑛2
−𝐾
𝐾2 = 2
𝜔𝑛
⟹ 𝑦 𝑡 = (𝐾1 + 𝐾3 cos(𝜔𝑛 𝑡)) 𝑢(𝑡) 𝐾3 = 0

𝑦(𝑡) (assuming 𝐾 = 𝜔𝑛2 )

This response is known as
an undamped response
𝑗𝜔𝑛

The natural response of the

system oscillates
−𝑗𝜔𝑛 indefinitely at the natural
frequency 𝜔𝑛 . As there is no
𝜔𝑛 > 0 (poles on 𝑗𝜔 axis) damping action, it does not
have a time constant. Page 8
Second-order step response – Standard case
• Case 4: Two complex conjugate poles:

𝑌 𝑠 𝜔𝑛2
𝐺 𝑠 = = 2
𝑅 𝑠 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
Time constant 1
Poles of the system: 𝑠1,2 = −𝜁𝜔𝑛 ± 𝜁 2 𝜔𝑛2 − 𝜔𝑛2 ⟹ of the system: 𝜏=
𝜁𝜔𝑛

Factor out −𝜔𝑛2 : 𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 1 − 𝜁 2 or 𝑠1,2 = −𝜎𝑑 ± 𝑗𝜔𝑑

• 𝜔𝑛 : Natural frequency of oscillation in the

system without damping.
• 𝜁 (zeta): Damping ratio, always a positive 0<𝜁<1
unitless constant.
• 𝜎𝑑 (sigma): Exponential damping frequency
𝜎𝑑 = 𝜁𝜔𝑛
which is the real part of the poles.
• 𝜔𝑑 : Damped frequency of oscillation which is
the imaginary part of the poles. 𝜔𝑑 = 𝜔𝑛 1 − 𝜁 2
Page 9
Second-order step response – Standard case (2)
• The output response to a unit step function:

𝜔𝑛2 𝐾1 𝐾2 𝑠 + 𝐾3
𝑦 𝑡 = ℒ −1 = ℒ −1 +
ต𝑛 𝑠 + 𝜔𝑛2 )
𝑠(𝑠 2 + 2 𝜁𝜔 𝑠 (𝑠 + 𝜎𝑑 )2 + 𝜔𝑛2 − 𝜎𝑑2
𝜎𝑑 2
𝜔𝑑

Completing the square :

𝐾1 = 1 = 𝐾𝑑𝑐
𝐾1 𝐾2 (𝑠 + 𝜎𝑑 ) 𝐾3 − 𝐾2 𝜎𝑑 𝜔𝑑
𝑦 𝑡 = ℒ −1 + + ( ) × 𝐾2 = −1
𝑠 (𝑠 + 𝜎𝑑 )2 +𝜔𝑑2 𝜔𝑑 (𝑠 + 𝜎𝑑 )2 +𝜔𝑑2 𝐾3 = −2𝜎𝑑

Substituting 𝐾1 , 𝐾2 , and 𝐾3 :
𝜎𝑑 −𝜎 𝑡
𝑦 𝑡 = (1 − 𝑒 −𝜎𝑑 𝑡 cos 𝜔𝑑 𝑡 − 𝑒 𝑑 sin(𝜔𝑑 𝑡)) 𝑢(𝑡)
𝜔𝑑

or 1
𝑦 𝑡 = (1 − 𝑒 −𝜁𝜔𝑛𝑡 cos(𝜔𝑛 1 − 𝜁 2 𝑡 + 𝜙)) 𝑢(𝑡)
1 − 𝜁2
𝜁
𝜙= tan−1
1 − 𝜁2 Page 10
Second-order step response – Standard case (3)
• The output response oscillates with the damped
frequency 𝜔𝑑 (imaginary part of complex pole pair) and 2
decays to its final value exponentially with exponential 𝐺 𝑠 = 𝜔𝑛

frequency 𝜎𝑑 (real part of complex pole pair). 𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 1 − 𝜁 2 0<𝜁<1
𝑠1,2 = −𝜎𝑑 ± 𝑗𝜔𝑑

1
This response is known as 𝑦 𝑡 =1− 𝑒 −𝜎𝑑 𝑡 cos(𝜔𝑑 𝑡 + 𝜙)
1 − 𝜁2
an underdamped response

𝑒 −𝜎𝑑 𝑡

cos(𝜔𝑑 𝑡 + 𝜙)

Page 11
Second-order step response – Standard case (4)
• As the damping ratio 𝜁 changes between zero and one, the magnitude of oscillation
changes:

𝜔𝑛2
𝐺 𝑠 = 2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

Small damping ratio 𝜁 ⟺ Lower damping

Large damping ratio 𝜁 ⟺ higher damping

Page 12
Second-order step response – Standard case (5)
• All the 2nd-order response types can be described based on
different values of the damping ratio 𝜁:
𝜔𝑛2
𝐺 𝑠 = 2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

Page 13
Characteristics/performance of a 2nd-order system
• We can characterise the performance of a second order 𝜔𝑛2
𝐺 𝑠 = 2
system from its output response by identifying or 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
measuring some important features of the response, which
are as follows: Note that in the standard 2nd-order system
• Rise time 𝑇𝑟 is defined as the time for the response to (above 𝐺 𝑠 ) the DC gain is one, so the
go from 10% to 90% of the final value. final value will exactly follow any value of
the step input (zero steady-state error).
• Peak time 𝑇𝑝 is the time required to reach the first,
or maximum, peak.
• Percent overshoot %𝑂𝑆 is the amount that
the response overshoots the steady-state
value at the peak time, expressed as a 1 + 0.02
percentage of the steady-state value.
• Settling time 𝑇𝑠 is defined as the time for the
1 − 0.02
response to reach and stay (settle) within
2% of its final value.
• Steady-state error 𝑒𝑠𝑠 is the difference
between the final value and the reference
signal in steady-state after 𝑇𝑠 .

𝑇𝑟
Rise Time
Page 14
Characteristics/performance of a 2nd-order system (2)
• Evaluation of 𝑇𝑝 : 𝜔𝑛2
𝐺 𝑠 = 2
• Set the derivative of the output equal to zero to find the time 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
where maxima (peak) and minima happen as below:
𝑑 𝜎𝑑 −𝜎 𝑡
𝑦ሶ 𝑡 = 1 − 𝑒 −𝜎𝑑 𝑡 cos 𝜔𝑑 𝑡 + 𝑒 𝑑 sin 𝜔𝑑 𝑡
𝑑𝑡 𝜔𝑑
𝜔𝑛
𝑦ሶ 𝑡 = 𝑒 −𝜁𝜔𝑛 𝑡 sin(𝜔𝑛 1 − 𝜁 2 𝑡) = 0 ⟹ 𝜔𝑛 1 − 𝜁 2 𝑡 = 𝑛𝜋
1 − 𝜁2
𝜋
For the first peak value: 𝑛 = 1 ⟹ 𝑇𝑝 =
𝜔𝑛 1 − 𝜁 2

• Evaluation of %𝑂𝑆: 𝑦𝑚𝑎𝑥 − 𝑦𝑠𝑠

• Based on the definition of %𝑂𝑆 ⟹ %𝑂𝑆 = × 100
𝑦𝑠𝑠

• Evaluate the output response at 𝑇𝑝 (𝑦𝑚𝑎𝑥 = 𝑦(𝑇𝑝 )),

and for a standard 2nd-order system with DC gain of
one 𝑦𝑠𝑠 = 1:
−𝜁𝜋
−𝜁𝜋
𝑦𝑚𝑎𝑥 = 𝑀𝑃𝑡 = 1 + 𝑒 1−𝜁 2
⟹ 1−𝜁 2
%𝑂𝑆 = 𝑒 × 100
Page 15
Characteristics/performance of a 2nd-order system (3)
• Evaluation of %𝑂𝑆 (continued): 𝜔𝑛2
𝐺 𝑠 = 2
• Note: Percent overshoot is a function of damping ratio 𝜁 only. Thus, 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
given a desired value of %𝑂𝑆 for a system, we can calculate the
desired amount of damping required to achieve that %𝑂𝑆.
−𝜁𝜋
%𝑂𝑆 = 𝑒 1−𝜁 2 × 100

− ln %𝑂𝑆/100
⟹ 𝜁=
𝜋 2 + ln2 (%𝑂𝑆/100)

• Evaluation of 𝑇𝑠 :
• Based on the definition of 𝑇𝑠 , the amplitude of the decaying
sinusoid should be equal to 0.02 of the final value at 𝑇𝑠 :
1 Approximate for all ⟹ 𝑒 −𝜁𝜔𝑛 𝑡 = 0.02
𝑒 −𝜁𝜔𝑛 𝑡 = 0.02
1 − 𝜁2 values of 0 < 𝜁 < 1:

4 1
Therefore: 𝑇𝑠 = ⟹ Time constant of an 𝑇𝑠 = 4𝜏 ⟹ 𝜏=
𝜁𝜔𝑛 underdamped system: 𝜁𝜔𝑛
Page 16
Characteristics/performance of a 2nd-order system (4)
• Evaluation of 𝑇𝑟 : 𝜔𝑛2
𝐺 𝑠 = 2
• The evaluation of 𝑇𝑟 based on its definition results in a nonlinear 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
function of 𝜁 and 𝜔𝑛 . If the normalized values of 𝑇𝑟 (i.e., 𝑇𝑟 𝜔𝑛 ) are
plotted against 𝜁, we can find an approximate linear relationship for
0.3 ≤ 𝜁 ≤ 0.8:
2.16𝜁 + 0.6

𝜔𝑛 𝑇𝑟
𝑇𝑟 =
𝜔𝑛 𝑇𝑟

• We can find 𝜁 and 𝜔𝑛 from the pole-zero location using simple

geometry.
• Recall that the poles are given as:
𝜔𝑛 circle
𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 1 − 𝜁 2

or 𝑠1,2 = −𝜎𝑑 ± 𝑗𝜔𝑑 𝜁 line

𝜁 = cos 𝜃 and 𝜔𝑛 = 𝜎𝑑2 + 𝜔𝑑2

Page 17
Characteristics/performance of a 2nd-order system (5)
• Examining the 𝜔𝑛2
𝑦(𝑡) 𝐺 𝑠 = 2
impact of the 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
movement of the
poles on the
output response:

𝑦(𝑡)

𝑦(𝑡)

Page 18
Characteristics/performance of a 2nd-order system (6)
𝜔𝑛2
• Some important points: 𝐺 𝑠 = 2
• 𝑇𝑝 , %𝑂𝑆 and 𝑇𝑟 are known as transient response characteristics 𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
of the step response of a system.
• 𝑇𝑠 and 𝑒𝑠𝑠 are known as steady-state response characteristics of
the step response of a system.
• Steady-state error 𝑒𝑠𝑠 is not limited to the step response. 𝑒𝑠𝑠 can
be calculated for any type of input (as you will learn later in this
course).
• The formulas that are obtained for 𝑇𝑝 , %𝑂𝑆 , 𝑇𝑟 and 𝑇𝑠 can be used
to determine the location of the desired closed-loop poles for
both 2nd-order and higher-order systems that we wish to behave
similarly to a 2nd- order system.
• 𝑇𝑝 and 𝑇𝑟 are used as a measure of how fast the output can first
reach its final value due to a sudden change in its input
(representing swiftness of the response).
• 𝑇𝑠 is also used as a measure of system speed since it is
approximately defined as 4𝜏, but mostly how fast the response can
settle after its transient behaviour.
• 𝑒𝑠𝑠 is mostly used in closed-loop systems as a measure of
accuracy. i.e., how closely the output can follow the reference
signal in steady-state. Page 19
Example 1
• Given the transfer function below, find 𝑇𝑝 , %𝑂𝑆, 𝑇𝑠 and 𝑇𝑟 .
• (Worked solutions of examples will be given either in lecture or uploaded later.)

100
𝐺 𝑠 =
𝑠 2 + 15𝑠 + 100

Answer:
𝑇𝑝 = 0.475 s
%𝑂𝑆 = 2.838%
𝑇𝑠 = 0.533 s
𝑇𝑟 ≅ 0.23 s.
Page 20
Example 2
• Given the pole plot shown below, find 𝜁, 𝜔𝑛 , 𝑇𝑝 , %𝑂𝑆, and 𝑇𝑠 .
• (Worked solutions of examples will be given either in lecture or uploaded later.)

Answer:
𝑇𝑝 = 0.449 s
%𝑂𝑆 = 26%
𝑇𝑠 = 1.333 s
𝜁 = 0.394
𝜔𝑛 = 7.616
Page 21
Example 3
• Given the system shown below, find 𝐽 and 𝐷 to yield 20% overshoot and a settling time
of 2 seconds for a step input of torque 𝑇(𝑡).
• (Worked solutions of examples will be given either in lecture or later on Moodle)

Answer:
𝐽 = 0.26 kg. m^2
𝐷 = 1.04 N. m. s/rad

Page 22
System response with additional poles
• Consider the unit step response of a third order system with two complex
conjugate poles and a third real pole (assume the DC gain is one):

𝑌 𝑠 𝛼𝜔𝑛2 𝜁 𝜔𝑛 ± 𝑗 𝜔𝑛 1 − 𝜁 2
𝑠1,2 = − ถ
𝐺 𝑠 = = 2 𝜎𝑑 𝜔𝑑
𝑅 𝑠 (𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 )(𝑠 + 𝛼) 𝑠3 = −𝛼

−1
𝛼𝜔𝑛2 −1
𝐴 𝐵(𝑠 + 𝜎𝑑 ) + 𝐶𝜔𝑑 𝐷
𝑦 𝑡 =ℒ = ℒ + +
𝑠 (𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 )(𝑠 + 𝛼) 𝑠 (𝑠 + 𝜎𝑑 )2 +𝜔𝑑2 𝑠+𝛼

⟹ 𝑦 𝑡 = (𝐴 + 𝑒 −𝜁𝜔𝑛 𝑡 𝐵ሖ cos 𝜔𝑑 𝑡 − 𝜙 + 𝐷𝑒 −𝛼𝑡

• Consider three cases for the location of the third pole 𝑠3 = −𝛼:

𝑠1 𝑠1 𝑠1

𝑠3 𝑠3

−𝛼 −𝛼 −𝛼 → −∞

𝑠2 𝑠2 𝑠2

Page 23
System response with additional poles (2)
• Case I: 𝛼 is relatively close to 𝜁𝜔𝑛 (the real part of
𝛼𝜔𝑛2
the complex pole pair). 𝐺 𝑠 = 2
(𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 )(𝑠 + 𝛼)
• 𝐷𝑒 −𝛼𝑡 stays much longer in the response.
• Case II: 𝛼 is much larger than 𝜁𝜔𝑛 . 𝑦 𝑡 = 𝐴 + 𝑒 −𝜁𝜔𝑛 𝑡 𝐵ሖ cos 𝜔𝑑 𝑡 − 𝜙 + 𝐷𝑒 −𝛼𝑡
• 𝐷𝑒 −𝛼𝑡 decays much quicker than the damped sinusoidal
part of the response.
𝛼 → ∞ (––)
• Case III: 𝛼 → ∞. (Ideal underdamped response)
𝛼 ≫ 𝜁𝜔𝑛 (---)
• The response is technically an ideal underdamped
response.

The response of a third-order system can be

approximated by the dominant poles of the 𝛼 close to 𝜁𝜔𝑛 (––)
2nd-order system as long as the third pole is at
least 10 times greater than the real part of the
dominant complex pole pair.

The approximated lower-order system must

have the same DC gain as the original system

Page 24
Impact of a zero on the 2nd-order system response
• Let’s investigate the impact of the zero via an example first.
• Consider 4 underdamped 2nd-order systems:

50
𝐺1 𝑠 =
(𝑠 2 + 10𝑠 + 50)

50/3(𝑠 + 3)
𝐺2 𝑠 = no zero
(𝑠 2 + 10𝑠 + 50) zero at −20
zero at −7
50/7(𝑠 + 7) zero at −3
𝐺3 𝑠 =
(𝑠 2 + 10𝑠 + 50)

50/20(𝑠 + 20)
𝐺4 𝑠 =
(𝑠 2 + 10𝑠 + 50)

As the location of the finite zero is

changed from the right side of the
dominant complex pole pair towards the left
of the 𝑠-plane, the step response
approaches the general/standard 2nd-
order system response without a zero. Page 25
Impact of a zero on the 2nd-order system response (2)
• For overdamped 2nd-order systems, adding a zero can change the step response
behaviour:
12
𝐺1 𝑠 =
(𝑠 2 + 7𝑠 + 12)

12 (𝑠 + 1)
𝐺2 𝑠 = no zero
(𝑠 2 + 7𝑠 + 12)
zero at −10
zero at −3.5
12/3.5(𝑠 + 3.5) zero at −1
𝐺3 𝑠 =
(𝑠 2 + 7𝑠 + 12)

12/10(𝑠 + 10)
𝐺4 𝑠 =
(𝑠 2 + 7𝑠 + 12)

If the finite zero is close to the

origin, the system becomes
oscillatory

Page 26
Impact of a zero on the 2nd-order system response (3)
• To see how the location of zero affects the output response of
a 2nd-order system, let’s have a look more generally at the
output response:
𝑎−𝑐 𝑎−𝑏
𝑌 𝑠 (𝑠 + 𝑎)
𝐺 𝑠 = = = 𝑏−𝑐 + 𝑐−𝑏
𝑅 𝑠 (𝑠 + 𝑏)(𝑠 + 𝑐) (𝑠 + 𝑏) (𝑠 + 𝑐)

• If the zero is far from the poles, then 𝑎 is large enough

compared to 𝑏 and 𝑐 to make the following approximation:
1 1
𝑏−𝑐 𝑐−𝑏 𝑎
⟹ 𝐺 𝑠 ≈𝑎 𝑠+𝑏
+
𝑠+𝑐
=
(𝑠 + 𝑏)(𝑠 + 𝑐)

Hence, the zero looks like a simple gain

factor and does not change the relative
amplitudes of the components of the
response.
Page 27
Minimum-phase and non-minimum phase system
• If a system has all its zeros on the left hand side of the 𝑠-plane, the system is called
minimum phase.
• If a system has at least one finite zero on the right hand side of 𝑠-plane, the system
is called non-minimum phase.

• This phenomenon happens because a zero in the system

can split the output response into two separate
components:
• a scaled version of the original output.
• a scaled version of the derivative of the output.
• With a zero on the RHP, the derivative part of the output
becomes negative, causing it to initially move in the
opposite direction of the input signal.
• An example of a physical non-minimum phase system in an
overhead crane, because as it accelerates in one direction,
its hanging load tends to move in the opposite direction due
to its inertia.

Page 28
Feedback control system characteristics
• Recall that an open-loop system operates without
feedback and directly generates the output in
response to an reference input signal.
• It is quite sensitive to input disturbances and a cascade
controller 𝐺𝑐 𝑠 should be carefully designed to achieve
the required transient and steady-state performance
• In contrast, a closed-loop negative feedback
control system uses a measurement of the output
signal and a comparison with the desired output to
generate an error signal that is used by the
controller to adjust the actuator.

Page 29
Feedback control system characteristics (2)
• Despite the cost and complexity of closed-loop control feedback systems,
there are some major advantages in using them:
• Decreased sensitivity of the system to variations in the parameters of the
process (decreased sensitivity to model uncertainty).
• Improved rejection of disturbances.
• Improved measurement noise attenuation.
• Improved reduction of the steady-state error of the system.
• Easy control and adjustment of the transient response of the system.

Page 30
Feedback control system characteristics (3)
• A standard closed-loop system can be considered to have
three inputs:
Td (s)
• Desired output signal (reference signal), 𝑅(𝑠)
• Disturbance signal 𝑇𝑑 (𝑠) R(s) + E(s) Uc (s) ++ U(s) Y(s)
Gc(s) G(s)
• Measurement noise signal 𝑁(𝑠) −
++
N (s)
• Let’s define the tracking error as:
𝐸 𝑠 =𝑅 𝑠 −𝑌 𝑠 (1)

• Find the output due to all these three inputs (for ease of
discussion, we consider a unity feedback system)
𝑌 𝑠
• Using superposition, first find 𝑅 when 𝑇𝑑 𝑠 = 𝑁 𝑠 = 0
𝑠
(1)
𝑌 𝑠 = 𝐺 𝑠 𝑈(𝑠)
ถ = 𝐺𝑐 𝑠 𝐺 𝑠 𝐸 𝑠 = 𝐺𝑐 𝑠 𝐺 𝑠 𝑅 𝑠 − 𝑌 𝑠
=𝑈𝑐 (𝑠)
𝑌 𝑠 𝐺𝑐 𝑠 𝐺 𝑠
𝑌 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 = 𝐺𝑐 𝑠 𝐺 𝑠 𝑅(𝑠) ⟹ =
𝑅 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠
𝐺𝑐 𝑠 𝐺 𝑠 is known as the loop gain.
Page 31
Feedback control system characteristics (4)
𝑌 𝑠
• Find 𝑇 when 𝑅 𝑠 = 𝑁 𝑠 = 0.
𝑑 𝑠
• Reconfigure the block diagram to have 𝑇𝑑 𝑠 as the input (the negative
feedback can be reflected into the summing junction at 𝑇𝑑 𝑠 ).
Td (s)

R(s) + E(s) Uc (s) ++ U(s) Y(s)

Gc(s) G(s)
−
++
N (s)
𝑌 𝑠 = 𝐺 𝑠 𝑈 𝑠 = 𝐺 𝑠 𝑇𝑑 𝑠 + 𝑈𝑐 𝑠
= 𝐺(𝑠)𝑇𝑑 (𝑠) − 𝐺(𝑠)𝐺𝑐 𝑠 𝑌 𝑠

𝑌 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 = 𝐺 𝑠 𝑇𝑑 (𝑠)
Td (s) + U(s) Y(s)
G(s)
𝑌 𝑠 𝐺 𝑠 −
⟹ =
𝑇𝑑 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠
Uc (s)
Gc(s)

Page 32
Feedback control system characteristics (5)
𝑌 𝑠
• Find 𝑁 when 𝑅 𝑠 = 𝑇𝑑 𝑠 = 0.
𝑠
• Reconfigure the block diagram to have 𝑁 𝑠 as the input (the
negative feedback can be reflected as a gain for 𝐺𝑐 𝑠 ).
Td (s)

R(s) + E(s) Uc (s) ++ U(s) Y(s)

Gc(s) G(s)
−
++
𝑌 𝑠 = 𝐺 𝑠 𝑈 𝑠 = 𝐺 𝑠 −𝐺𝑐 (𝑠)(𝑁 𝑠 + 𝑌 𝑠 N (s)

= −𝐺𝑐 𝑠 𝐺 𝑠 𝑁(𝑠) − 𝐺(𝑠)𝐺𝑐 𝑠 𝑌 𝑠

𝑌 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 = −𝐺𝑐 𝑠 𝐺 𝑠 𝑁(𝑠)

N (s) + Uc (s)=U(s) Y(s)

𝑌 𝑠 −𝐺𝑐 𝑠 𝐺 𝑠 − Gc(s) G(s)
⟹ = +
𝑁 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠

Page 33
Feedback control system characteristics (6)
• Adding all the outputs due to 𝑅 𝑠 , 𝑇𝑑 𝑠 , and 𝑁 𝑠 , Td (s)
we have the following:
R(s) + E(s) Uc (s) ++ U(s) Y(s)
Gc(s) G(s)
−
++
N (s)

𝐺𝑐 𝑠 𝐺 𝑠 𝐺 𝑠 −𝐺𝑐 𝑠 𝐺 𝑠
𝑌 𝑠 = 𝑅(𝑠) + 𝑇𝑑 (𝑠) + 𝑁(𝑠)
1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠

• As you can see, to reduce the influence of disturbance on the

output, the magnitude of 𝐺𝑐 𝑠 𝐺 𝑠 should be large over the
range of frequencies that characterizes the disturbance.
• So, we need to design the controller 𝐺𝑐 𝑠 such that:

𝐺 𝑠
𝐺𝑐 𝑠 𝐺 𝑠 ≫1 ⟹ ≪1
1 + 𝐺𝑐 𝑠 𝐺 𝑠

Page 34
Feedback control system characteristics (7)
• To reduce the influence of measurement Td (s)
noise on the output, the magnitude of
𝐺𝑐 𝑠 𝐺 𝑠 should be small over the range R(s) + E(s) Uc (s) ++ U(s) Y(s)
Gc(s) G(s)
of frequencies that characterizes noise. −
++
N (s)

𝐺𝑐 𝑠 𝐺 𝑠 𝐺 𝑠 −𝐺𝑐 𝑠 𝐺 𝑠
𝑌 𝑠 = 𝑅(𝑠) + 𝑇𝑑 (𝑠) + 𝑁(𝑠)
1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠

• So we need to design the controller 𝐺𝑐 𝑠 such that:

𝐺 𝑠 𝐺𝑐 𝑠
𝐺𝑐 𝑠 𝐺 𝑠 ≪1 ⟹ ≪1
1 + 𝐺𝑐 𝑠 𝐺 𝑠

Page 35
Feedback control system characteristics (8)
• At the same time, we wish the output to Td (s)

follow the reference/desired output signal

R(s) + E(s) Uc (s) ++ U(s) Y(s)
too. Which means the magnitude of Gc(s) G(s)
−
𝐺𝑐 𝑠 𝐺 𝑠 should be large over the range of ++
N (s)
frequencies that characterizes the
reference signal.
𝐺𝑐 𝑠 𝐺 𝑠 𝐺 𝑠 −𝐺𝑐 𝑠 𝐺 𝑠
• So we need to design the controller 𝐺𝑐 𝑠 𝑌 𝑠 = 𝑅 𝑠 + 𝑇𝑑 𝑠 + 𝑁(𝑠)
1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠 1 + 𝐺𝑐 𝑠 𝐺 𝑠
such that:

𝐺 𝑠 𝐺𝑐 𝑠
𝐺𝑐 𝑠 𝐺 𝑠 ≫1 ⟹ ≈1
1 + 𝐺𝑐 𝑠 𝐺 𝑠

• Clearly, wanting to make 𝐺𝑐 𝑠 large to reject We design the controller 𝐺𝑐 𝑠 for 𝐺 𝑠 such
disturbances and reduce the tracking error, and that the magnitude of loop gain
wanting to make 𝐺𝑐 𝑠 small to attenuate the 𝐺𝑐 𝑠 𝐺 𝑠 is large at low frequencies for
noise, is contradictory. disturbance rejection and small at high
• But the ranges of frequencies for disturbance and frequencies for noise attenuation.
noise are quite different.
Page 36
Questions?

Page 37