6.

Transient Response

Reference: Dorf, Richard C., and Robert H. Bishop. Modern control systems.
Pearson, 12th Edition (Chapter 5.3, 5.4, and 5.5)
Outline
• Typical test signals
• Transient response and steady-state response
• 1st order system response
• 2nd order system response
• Time response specification
• Higher-order systems

2
Time response
• Steady-state response
t approaches infinity
• Transient response
Initial state to final state
• Test signal: step input

Input

3
Specifications
Peak value of the response 𝑀𝑝 − final value of the response
• Percent overshoot: 𝑃. 𝑂. = final value of the response

• Peak time 𝑇𝑝 : time of the response to reach

its peak value 𝑀𝑝
• Rise time 𝑇𝑟 : time of the response to rise
from 10% to 90% of the input value
• Settling time 𝑇𝑠 : time of the response to
settle within 2% of the input value
• Ideally, we expect 𝑃. 𝑂. = 0, 𝑇𝑝 = 0, 𝑇𝑟 =
0, 𝑇𝑠 = 0

4
First-order systems
• 1st-order systems: denominator of TF is a first order polynomial of 𝑠
𝑌 𝑠 1
𝐺 𝑠 = = , 𝜏 = 𝑅𝐶
𝑅 𝑠 𝜏𝑠 + 1
• (Assuming zero initial conditions)
1
• Unit-step response, input 𝑅 𝑠 = 1/𝑠 𝐺(𝑠) =
𝑠𝑅𝐶 + 1

• Therefore
1 1 1 1
𝑌 𝑠 = ∙ = −
𝜏𝑠 + 1 𝑠 𝑠 𝑠 + 1/𝜏

• Finding the inverse Laplace transform

𝑦 𝑡 = 1 − 𝑒 −𝑡/𝜏 , 𝑡 ≥ 0

5
Specifications of first-order systems
• The input 𝑟 (𝑡) = 1(𝑡)
• The response
𝑦 𝑡 = 1 − 𝑒 −𝑡/𝜏 , 𝑡 ≥ 0

• At time 𝑡 = 𝜏
𝑦 𝜏 = 0.632

• Time constant: 𝜏 (Unit: sec)

• Specifications
• Overshoot 𝑃. 𝑂. = 0,
• 𝑇𝑝 = 0 Not defined
• 𝑇𝑟 = ln 0.9 − ln 0.1 𝜏 = 2.2𝜏,
• 𝑇𝑠 = −ln 0.02 𝜏 = 4𝜏

6
Second-order systems
𝐾
𝑋(𝑠) 1 F(t) 𝐵: Damping factor
𝐺 𝑠 = = 𝑀 𝐾: Spring stiffness
𝐹(𝑠) 𝑠 2 𝑀 + 𝑠𝐵 + 𝐾
𝐵
• Standard form
𝜔𝑛2
𝐺 𝑠 = 𝑔dc 2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2
• where
1
• 𝑔dc = : The DC gain
𝐾
𝐵
• 𝜁= is the damping ratio,
2 𝐾𝑀
𝐾
• 𝜔𝑛 = is the natural frequency
𝑀

• Without loss of generality, assume 𝑔dc = 1

7
Step response of second-order systems
• Step input 𝑟 𝑡 = 1 𝑡 ⇒ 𝑅 𝑠 = 1/𝑠
𝜔𝑛2 1
𝑌 𝑠 = 𝐺 𝑠 𝑅(𝑠) = 2 ∙
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 𝑠
• Poles of 𝐺 𝑠

𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 = (𝑠 − 𝑠1 )(𝑠 − 𝑠2 )

• Time response (𝑠1 ≠ 𝑠2 )

𝐴0 𝐴1 𝐴2
𝑌 𝑠 =𝐺 𝑠 𝑅 𝑠 = + +
𝑠 𝑠 − 𝑠1 𝑠 − 𝑠2
𝜔𝑛2 𝜔𝑛2 𝜔𝑛2
𝐴0 = ;𝐴 = ;𝐴 =
𝑠1 𝑠2 1 𝑠1 (𝑠1 − 𝑠2 ) 2 𝑠2 (𝑠2 − 𝑠1 )

⇒ 𝑦 𝑡 = 𝐴0 + 𝐴1 𝑒 𝑠1 𝑡 + 𝐴2 𝑒 𝑠2 𝑡
𝜔𝑛2 𝜔𝑛2 𝜔𝑛2
𝑠1 𝑡
= + 𝑒 + 𝑠 (𝑠 −𝑠 ) 𝑒 𝑠2 𝑡
𝑠1 𝑠2 𝑠1 (𝑠1 −𝑠2 ) 2 2 1
8
Case 1: overdamped systems
𝑗𝜔
• Damping is strong 𝜁 > 1
2𝜔𝑛 𝜁 2 − 1

𝜎
𝑠1 = −𝜁𝜔𝑛 + 𝜔𝑛 𝜁 2 − 1 < 0, Real number × ×
𝑠2 𝑠1
𝑠-plane
𝑠2 = −𝜁𝜔𝑛 − 𝜔𝑛 𝜁 2 − 1 < 0, Real number −𝜁𝜔𝑛

𝜔𝑛2 𝜔𝑛2 𝑠 𝑡
𝜔𝑛2
𝑦(𝑡) = + 𝑒1 + 𝑒 𝑠2 𝑡
𝑠1 𝑠2 𝑠1 (𝑠1 − 𝑠2 ) 𝑠2 (𝑠2 − 𝑠1 )

1 𝑠2 𝑠 𝑡 𝑠1 𝑠 𝑡
=1+ 𝑒1 − 𝑒2
2 𝜁 2 − 1 𝜔𝑛 𝜔𝑛

𝜔𝑛
𝑦′ 𝑡 = 𝑒 𝑠1 𝑡 − 𝑒 𝑠2 𝑡 ≥ 0
2 𝜁2 − 1
➔ monotonically increasing, no overshoot
9
Case 2: critically-damped systems
𝑗𝜔
• Damping is at critical value 𝜁 = 1
𝜔𝑛

× 𝜎
𝑠1 = −𝜁𝜔𝑛 + 𝜔𝑛 𝜁 2 − 1 = −𝜔𝑛 , Real number
𝑠1,2
𝑠-plane
𝑠2 = −𝜁𝜔𝑛 − 𝜔𝑛 𝜁 2 − 1 = −𝜔𝑛 , Real number

𝜔𝑛2 1 𝜔𝑛2 1 1 1 𝜔𝑛
𝑌 𝑠 = 2 ∙ = 2∙𝑠 = 𝑠−𝑠+𝜔 − 𝑠+𝜔
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 𝑠 𝑠 + 𝜔𝑛 𝑛 𝑛
2

𝑦 𝑡 = 1 − (1 + 𝜔𝑛 𝑡) 𝑒 −𝜔𝑛𝑡

➔ Converge to one

𝑦′ 𝑡 = 𝜔𝑛2 𝑡𝑒 −𝜔𝑛𝑡 ≥ 0

➔ monotonically increasing, no overshoot

10
Case 3: underdamped systems 𝑗𝜔
• Damping is small 0 ≤ 𝜁 < 1
𝑠1
× 𝜔𝑛
2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 − 𝑠1 𝑠 − 𝑠2 = 0 𝜔𝑛 1 − 𝜁2
𝜃 𝜎
𝜁𝜔𝑛
𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 1 − 𝜁 2
𝑠2 × 𝑠-plane
𝜔𝑛2 𝜔𝑛2 𝜔 2
𝑛
𝑦 𝑡 = + 𝑒 𝑠1 𝑡 + 𝑒 𝑠2 𝑡
𝑠1 𝑠2 𝑠1 (𝑠1 − 𝑠2 ) 𝑠2 (𝑠2 − 𝑠1 )
𝑒 −𝜁𝜔𝑛 𝑡
=1− sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃 𝜃 = cos −1 𝜁
1 − 𝜁2
➢ The real part of 𝑠1,2 determines 𝑒 −𝜁𝜔𝑛𝑡
the speed of convergence
➢ The imaginary part of 𝑠1,2 Response
determines the speed of
oscillation
➢ Most systems are in this case sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃

11
Time (sec)
Case 3.1: undamped systems 𝑗𝜔
• Damping is zero 𝜁 = 0 × 𝑠1
𝜔𝑛
𝑠1 = 𝑗𝜔𝑛 𝑠2 = −𝑗𝜔𝑛
𝜎

𝜔𝑛2 𝜔𝑛2 𝑠1 𝑡
𝜔𝑛2 𝑠-plane
𝑦 𝑡 = + 𝑒 + 𝑒 𝑠2 𝑡
𝑠1 𝑠2 𝑠1 (𝑠1 − 𝑠2 ) 𝑠2 (𝑠2 − 𝑠1 ) × 𝑠2
𝜋
= 1 − sin 𝜔𝑛 𝑡 +
2

➢ Undamped oscillation, no energy

loss
➢ Does not converge ( to the input
value), 

12
Pole locations and step response

13
Specifications of second-order systems
• Only consider the underdamped case (i.e., 0 ≤ 𝜁 < 1)

𝑒 −𝜁𝜔𝑛 𝑡
𝑦 𝑡 =1− sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃 𝜃 = cos −1 𝜁
1 − 𝜁2

• Settling time 𝑇𝑠 : time of the response to settle within 2% of the final value. Let
1
𝜏 = 𝜁𝜔 be the time constant, approximately,
𝑛

𝑡
−𝜏
𝑇
− 𝜏𝑠 4
𝑒 ≤ 0.02 ⇒ 𝑒 = 0.02 ⇒ 𝑇𝑠 ≈ 4𝜏 ≈
𝜁𝜔𝑛

14
Specifications of second-order systems
• Only consider the underdamped case (i.e., 0 ≤ 𝜁 < 1)

𝑒 −𝜁𝜔𝑛 𝑡
𝑦 𝑡 =1− sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃 𝜃 = cos −1 𝜁
1 − 𝜁2
𝑦(𝑇𝑝 )

• Peak time 𝑇𝑝 : time of the response to reach its

peak value 𝑀𝑝 ,
𝜋
𝑦 ′ (𝑡) = 0 ⇒ 𝑇𝑝 =
𝜔𝑛 1 − 𝜁 2
𝑇𝑝

• Percent Overshoot 𝑃. 𝑂.,

𝜁𝜋
−
𝑀𝑝 = 𝑦 𝑇𝑝 = 1 + 𝑒 1−𝜁 2

𝜁𝜋
𝑀𝑝 − 𝑦𝑠𝑠 −
1−𝜁 2
𝑃. 𝑂. = = 100𝑒
𝑦𝑠𝑠
15
Specifications of second-order systems
• Only consider the underdamped case (i.e., 0 ≤ 𝜁 < 1)

𝑒 −𝜁𝜔𝑛 𝑡
𝑦 𝑡 =1− sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃 𝜃 = cos −1 𝜁
1 − 𝜁2

• Peak time 𝑇𝑝 : time of the response to reach its peak value 𝑀𝑝 ,

𝜁𝜋
𝜋 −
𝑦 ′ (𝑡) = 0 ⇒ 𝑇𝑝 = 𝑀𝑝 = 𝑦 𝑇𝑝 = 1 + 𝑒 1−𝜁 2
𝜔𝑛 1 − 𝜁2

• Percent Overshoot 𝑃. 𝑂.,

𝜁𝜋
−
𝑃. 𝑂. = 100𝑒 1−𝜁 2

𝜁 0.9 0.8 0.7 0.6 0.5

𝑃. 𝑂. 0.2 1.5 4.6 9.5 16.3
16
𝜁 = 0.707 is a good balance between speed of response and the overshoot.
Specifications of second-order systems
• Only consider the underdamped case (i.e., 0 ≤ 𝜁 < 1)

𝑒 −𝜁𝜔𝑛 𝑡
𝑦 𝑡 =1− sin 1 − 𝜁 2 𝜔𝑛 𝑡 + 𝜃 𝜃 = cos −1 𝜁
1 − 𝜁2

• Rise time 𝑇𝑟 : time of the response to

reach from 10% to 90% of the input,

2.16𝜁 + 0.6
𝑇𝑟 ≈
𝜔𝑛 𝑇𝑟

when 0.3 ≤ 𝜁 ≤ 0.8

17
Effect of damping ratio and natural frequency
• Specifications for 0 ≤ 𝜁 < 1:
𝜁𝜋
4 − 𝜋
𝑇𝑠 ≈ 4𝜏 ≈ 𝑃. 𝑂. = 100𝑒 1−𝜁 2 , 𝑇 =
𝜁𝜔𝑛 𝑝
𝜔𝑛 1 − 𝜁 2
2.16𝜁 + 0.6
𝑇𝑟 ≈ when 0.3 ≤ 𝜁 ≤ 0.8
𝜔𝑛

18
Example
A DC motor is driven by input
electromagnetic torque (i.e., voltage) and 𝑅 𝐾 1 𝑌
+
assume to have a model 𝐺 𝑠 = 𝑠 . Now
1 - 𝜏𝑠 + 1 𝑠

design a controller consisting of a gain 𝐾 and

1
a first order low pass filter 𝜏𝑠+1
Determine the controller gain 𝐾 and time constant 𝜏 such that the step response has
an overshoot less than 5%, and a settling time less than 4 second.

19
Example
A DC motor is driven by input
electromagnetic torque (i.e., voltage) and 𝑅 𝐾 1 𝑌
+
assume to have a model 𝐺 𝑠 = 𝑠 . Now
1 - 𝜏𝑠 + 1 𝑠

design a controller consisting of a gain 𝐾 and

1
a first order low pass filter 𝜏𝑠+1
Determine the controller gain 𝐾 and time constant 𝜏 such that the step response has
an overshoot less than 5%, and a settling time less than 4 second

▪ Closed-loop transfer function

𝐾 𝐾
𝑠(𝜏𝑠 + 1) 𝐾 𝜏 𝜔𝑛2
𝑇 𝑠 = = 2 = = 2
𝐾 𝜏𝑠 + 𝑠 + 𝐾 2 1 𝐾 𝑠 + 2𝜁𝜔 𝑛 𝑠 + 𝜔 2
1+ 𝑠 +𝜏𝑠+ 𝜏 𝑛
𝑠(𝜏𝑠 + 1)

▪ Standard second-order system: 𝐾 1

𝜔𝑛 = ,𝜁 =2
𝜏 𝜏𝐾

20
Example
A DC motor is driven by input
electromagnetic torque (i.e., voltage) and 𝑅 𝐾 1 𝑌
+
assume to have a model 𝐺 𝑠 = 𝑠 . Now
1 - 𝜏𝑠 + 1 𝑠

design a controller consisting of a gain 𝐾 and

1
a first order low pass filter 𝜏𝑠+1
Determine the controller gain 𝐾 and time constant 𝜏 such that the step response has
an overshoot less than 5%, and a settling time less than 4 second

▪ Specifications of second-order system

4
𝑇𝑠 ≈ ≤ 4 ⇒ 𝜁𝜔𝑛 ≥ 1
𝜁𝜔𝑛
𝜁𝜋
−
𝑃. 𝑂. = 100𝑒 1−𝜁 2 ≤ 5 ⇒ 𝜁 ≥ 0.707

𝜁 0.9 0.8 0.7 0.6 0.5

𝑃. 𝑂. 0.2 1.5 4.6 9.5 16.3 21
Example
A DC motor is driven by input
electromagnetic torque (i.e., voltage) and 𝑅 𝐾 1 𝑌
+
assume to have a model 𝐺 𝑠 = 𝑠 . Now
1 - 𝜏𝑠 + 1 𝑠

design a controller consisting of a gain 𝐾 and

1
a first order low pass filter 𝜏𝑠+1
Determine the controller gain 𝐾 and time constant 𝜏 such that the step response has
an overshoot less than 5%, and a settling time less than 4 second

▪ In particularly,
𝜁𝜔𝑛 = 1 𝜁 = 0.707

▪ Controller parameter

𝐾 1
𝜔𝑛 = 𝜏= = 0.5
𝜏 ⇒ 2𝜁𝜔𝑛
1 𝐾 = 𝜏𝜔𝑛2 = 1
𝜁= 2 𝜏𝐾
22
Example (2019-2020Final Exam, Q5)

23
Third-order systems
• Consider a 3rd order system
𝜔𝑛2
𝐺(𝑠) = 2
(𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 )(𝛾𝑠 + 1)

𝑡
−
• The third pole contributes a mode 𝑒 𝛾

𝑡
1 −𝛾
• If ≥ 10 𝜁𝜔𝑛 , then the mode 𝑒
𝛾
decays much faster than 𝑒 −𝜁𝜔𝑛 𝑡 .
• In this case, the second order system
dominates the response and the roots
are called the dominant roots.

24
Computing Specifications by MATLAB

Example:

25
Computing Specifications by MATLAB
Example: 𝐺(𝑠) = 2
𝜔𝑛2
(𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 )(𝛾𝑠 + 1)

26
Practical systems
• Transient response of practical systems is very complicated
• Effect of zeros
𝜔𝑛2 (𝑠 + 𝑧)
𝐺(𝑠) = 2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

• Higher order systems

• Specifications (e.g., overshoot, settling time, etc.) are still valid to
characterize these systems, but the following relations are no longer
valid
𝜁𝜋
4 − 2.16𝜁 + 0.6
𝑇𝑠 ≈ 4𝜏 ≈ 𝑃. 𝑂. = 100𝑒 1−𝜁 2 𝑇𝑟 ≈
𝜁𝜔𝑛 𝜔𝑛

• Analyzing the system by dominant pole(s) is usually a good

approximate when others are far away
• In case of accurate specifications required, use MATLAB

27
Practical systems

28