MECE 3350U

Control Systems

Lecture 8
Transient Response

1/44
MECE 3350 - C. Rossa 1 / 44 Lecture 8
Videos in this lecture

Lecture: https://youtu.be/J8jp_3KaXLw

Exercise 35: https://youtu.be/FgjQOuxgwd0

Exercise 36: https://youtu.be/bhnWk-hTjgI

Exercise 37: https://youtu.be/aOJt3LBX1Uw

Exercise 38: https://youtu.be/VdJisStqrx8

Exercise 39: https://youtu.be/1XTYlE6O4iE

2/44
MECE 3350 - C. Rossa 2 / 44 Lecture 8
Applications

The levitation control system of the train must ensure that the train does not
touches the guide. How can we design a controller that reacts as fast as
possible with no overshoot?

3/44
MECE 3350 - C. Rossa 3 / 44 Lecture 8
Applications

The pointing control system of a space telescope is desired to achieve an

accuracy of 0.01 minute of arc. How can we limit the steady state error while
avoiding transient oscillations?

4/44
MECE 3350 - C. Rossa 4 / 44 Lecture 8
Transient response

5/44
MECE 3350 - C. Rossa 5 / 44 Lecture 8
First order systems

Consider the first order closed-loop system shown with a proportional gain k

The transfer function Y (s)/R(s) is

Y (s) 1
= τ


R(s) k
s +1

How does k influence the transient and steady state response?

To analyse the performance of the system, we need to specify a standard test

input signal.

6/44
MECE 3350 - C. Rossa 6 / 44 Lecture 8
Standard test signals

Impulse function

A t=0
δ(t) = → I(s) = A
0 t 6= 0

Step function

A t≥0 1
u(t) = → U(s) = A
0 t<0 s

Ramp function

At t≥0 1
r (t) = → R(s) = A 2
0 t<0 s

Parabolic function
2

A t2 t≥0 1
p(t) = → P(s) = A 3
0 t<0 s

7/44
MECE 3350 - C. Rossa 7 / 44 Lecture 8
Temporal response
Step response r (t) = 1
1 1 k
Y (s) = τ

→ L −1 → y (t) = 1 − e − τ t
s k
s +1
Ramp response r (t) = t
1 1 τ k
Y (s) =
→ L −1 → y (t) = t − (1 − e − τ t )
s2 τ
k
s +1 k
Effects of k for k > 0 for τ = 1.
1
y
input y
input

0 0
0 time [sec] 10 0 time [sec] 10

8/44
MECE 3350 - C. Rossa 8 / 44 Lecture 8
Temporal response - first order system

1
y
input y
input

0 0
0 time [sec] 10 0 time [sec] 10

In a first-order system:

→ k reduces the time constant of the system

→ The higher k, the faster the response

→ What is the maximum control-loop gain k ?

9/44
MECE 3350 - C. Rossa 9 / 44 Lecture 8
Time constant - first order systems
Impulse:
1 t
H(s) = → L −1 → y (t) = y (0)e − τ
sτ + 1
→ When t = τ , the response 37% (1/e) of y (0)

Step response
1 1 t
H(s) = → L −1 → y (t) = 1 − e − τ
s sτ + 1
→ When t = τ , the response 67% (1-1/e) of its steady state value

0 0
0 time [sec] time [sec]

10/44
MECE 3350 - C. Rossa 10 / 44 Lecture 8
Second-order systems

Consider now the following second order control system:

The transfer function is

Y (s) k
= 2
R(s) s + sa + k
We can rewrite the above equation in the standard formulation:

Y (s) ωn2
= 2
R(s) s + 2ζωn s + ωn2
√ √
where: ωn = k, ζ = a/(2 k).

11/44
MECE 3350 - C. Rossa 11 / 44 Lecture 8
Transient response

Y (s) ωn2
= 2
R(s) s + 2ζωn s + ωn2
√ √
where: ωn = k, ζ = a/(2 k).

How does k influence the response of the system?

→ The natural frequency ωn depends on k
→ The damping ratio ζ depends on k

The response for an unit step input when 0 < ζ < 1 is

1
h p  i
y (t) = 1 − p e −ζωn t sin ωn 1 − ζ 2 t + cos−1 ζ
1 − ζ2

12/44
MECE 3350 - C. Rossa 12 / 44 Lecture 8
Transient response - second-order systems

Y (s) ωn2
= 2
R(s) s + 2ζωn s + ωn2
1 −ζωn t
 p 
y (t) = 1 − e sin ωn 1 − ζ 2 t + cos−1 ζ (1)
β

Imaginary Axis (seconds-1 )

1

0 -3
0 time [sec] 15 -0.6 Real Axis (seconds -1 )
0

13/44
MECE 3350 - C. Rossa 13 / 44 Lecture 8
Transient response - second-order systems
Y (s) ωn2
= 2
R(s) s + 2ζωn s + ωn2
1
 p 
y (t) = 1 − p e −ζωn t sin ωn 1 − ζ 2 t + cos−1 ζ
1 − ζ2

Natural frequency for k = 1000 and k = 5000.

2 2
y
input

1 1
y

0 0
0 time [sec] 15 0 time [sec] 15

How can we evaluate the performance of the controller?

14/44
MECE 3350 - C. Rossa 14 / 44 Lecture 8
Measures of performance

→ Rise time Tr , peak time Tp , and peak value Mpt

→ Settling time Ts : y (t) within 2% of its final value
→ Percent overshoot P.O.

→ Tr and Tr characterize the swiftness of the response

→ P.O. and Ts characterize the closeness of the response to the input
15/44
MECE 3350 - C. Rossa 15 / 44 Lecture 8
Overshoot
For an unit step input, the percent overshoot is

Mpt − fv
P.O. = × 100 (2)
fv
→ Mpt is the peak value
→ fv is the magnitude of the input

Differentiating Eq (1) and setting it to zero yields the peak time as

π
Tp = p (3)
ωn 1 − ζ2
Replacing (3) into (1) gives the peak response:
√
2
Mpt = 1 + e −ζπ/ 1−ζ (4)
Thus, the percentage overshoot (fv = r (t) = 1) is
√
2
P.O. = 100e −ζπ/ 1−ζ (5)

16/44
MECE 3350 - C. Rossa 16 / 44 Lecture 8
Settling time
For an unit step input and 0 < ζ < 1, recall that
1
 p 
y (t) = 1 − p e −ζωn t sin ωn 1 − ζ 2 t + cos−1 ζ
1 − ζ2
When t = Ts , the response is within 2% of its final value, thus:

e −ζωn Ts < 0.02

or

ζωn Ts ≈ 4 (6)
therefore
4
Ts = = 4τ (7)
ζωn
where τ = 1/ζωn is the time constant.

The settling time is equal to 4 times the time constant

17/44
MECE 3350 - C. Rossa 17 / 44 Lecture 8
Settling time

The characteristic equation

s 2 + 2ζωn s + ωn2
has poles:
p
s1 = −ζωn + jωn 1 − ζ2
p
s1 = −ζωn − jωn 1 − ζ2

p
or s = −ζωn ± ωn ζ 2 − 1. Since
4
Ts ≈ (8)
ζωn

Therefore the settling time is inversely proportional to the real part of the poles.

18/44
MECE 3350 - C. Rossa 18 / 44 Lecture 8
Settling time
Consider the function
5
H=
s2 + s + 5
√ √
thus: ωn = 5 and ζ = 1/(2 5). The time constant is
1
τ = = 2 sec
ζωn
2
y
input

1
y

0
0 2 4 6 8 10
time [sec]

19/44
MECE 3350 - C. Rossa 19 / 44 Lecture 8
Poles and transient response

20/44
MECE 3350 - C. Rossa 20 / 44 Lecture 8
 Damping ratio Roots Systems response

ζ>1 Distinct real overdamped

ζ=1 Equal real damped
0<ζ<1 Complex conjugate underdamped
ζ=0 Purely imaginary undamped
ζ<0 Positive unstable

21/44
MECE 3350 - C. Rossa 21 / 44 Lecture 8
Exercise 35

A feedback system with a negative unity feedback has the loop transfer function

2(s + 8)
L(s) = C (s)G(s) = .
s(s + 4)

Determine:

→ (a) The closed-loop transfer function

→ (b) The time response for an input r (t) = A
→ (c) The percent overshoot of the response
→ (d) The steady state error

22/44
MECE 3350 - C. Rossa 22 / 44 Lecture 8
Exercise 35 - continued
(a) The closed-loop transfer function

2(s + 8)
L(s) = C (s)G(s) = .
s(s + 4)

23/44
MECE 3350 - C. Rossa 23 / 44 Lecture 8
Exercise 35 - continued
(b) The time response
A 2(s + 8)
Y (s) =
s s 2 + 6s + 16

24/44
MECE 3350 - C. Rossa 24 / 44 Lecture 8
Exercise 35 - continued
(c) The percentage overshoot
A 2(s + 8)
Y (s) =
s s 2 + 6s + 16

25/44
MECE 3350 - C. Rossa 25 / 44 Lecture 8
Exercise 35 - continued
(d) The steady state error
A 2(s + 8)
Y (s) =
s s 2 + 6s + 16

26/44
MECE 3350 - C. Rossa 26 / 44 Lecture 8
Exercise 36

Consider the following block diagram:

→ (a) Calculate the steady-state error for a ramp input

→ (b) Select k that will result in zero overshoot to a step input

27/44
MECE 3350 - C. Rossa 27 / 44 Lecture 8
Exercise 36 - continued

28/44
MECE 3350 - C. Rossa 28 / 44 Lecture 8
Exercise 36 - continued

29/44
MECE 3350 - C. Rossa 29 / 44 Lecture 8
Exercise 37

An insulin pump injection system for diabetic persons has a feedback control as
shown.

Calculate a suitable gain k so that the percent overshoot of the step response
due to the drug injection is 7%. R(s) is the desired blood sugar level and Y (s)
is the actual level. Plot the expected overshoot for different k using Matlab.

30/44
MECE 3350 - C. Rossa 30 / 44 Lecture 8
Exercise 37 - continued

31/44
MECE 3350 - C. Rossa 31 / 44 Lecture 8
Exercise 37 - continued

32/44
MECE 3350 - C. Rossa 32 / 44 Lecture 8
Exercise 38

Consider the following closed loop system

Determine:

→ (a) Determine the closed loop transfer function

→ (b) Select k so that the steady state error to a unit step input is bounded.

33/44
MECE 3350 - C. Rossa 33 / 44 Lecture 8
Exercise 38 - continued
(a) Determine the closed loop transfer function

34/44
MECE 3350 - C. Rossa 34 / 44 Lecture 8
Exercise 38 - continued
(b) Determine the steady state error to a unit ramp input

35/44
MECE 3350 - C. Rossa 35 / 44 Lecture 8
Exercise 38 - continued

(c) Determine k so that e = 0 when r (t) = 1

36/44
MECE 3350 - C. Rossa 36 / 44 Lecture 8
Exercise 39

A closed-loop system designed to orient a photovoltaic array towards the

direction of maximum solar incidence has the following structure:

If k = 20, determine:

→ (a) The time constant of the closed loop system

→ (b) The settling time to within 2% of the final value of the system to an
unit step disturbance.
37/44
MECE 3350 - C. Rossa 37 / 44 Lecture 8
Exercise 39 - continued

38/44
MECE 3350 - C. Rossa 38 / 44 Lecture 8
Exercise 39 - continued

39/44
MECE 3350 - C. Rossa 39 / 44 Lecture 8
Skills check 22 - From 2018 midterm examination

Which of the following time domain signals best describes the response of the
function H(s) given below to a unit step-type input? (4 marks/100 ).
1
H(s) =
4s 2 + s + 1

1.8 1.5 1.2

1 1

1
1

0 0 0
0 time [sec] 120 180 0 time [sec] 32 45 0 time [sec] 9 12 0 time [sec] 4 6

Answer in the last slides 40/44

MECE 3350 - C. Rossa 40 / 44 Lecture 8
Skills check 23 - From 2018 final examination

A controller is to be designed for a plant as shown.

(a) Find the transfer function Y (s)/R(s).

(b) Specify the controller gain K and the value of a so that the overall
closed-loop response to a unit step input has an overshoot of no more than
25% and a settling time of no more than 0.1 seconds.

41/44
MECE 3350 - C. Rossa 41 / 44 Lecture 8
Skills check 24 - From 2018 midterm examination

A robotic system has been designed to operate on a beating heart by ensuring

continuous contact between a surgical tool and the heart. The block diagram
of such a system is shown below where R(s) is the desired position of the
surgical tool i.e., the position of the epicardium (heat wall), and Y (s) is the
actual tool position. Any controller overshoot would result in the tool
puncturing the heart.

(a) Determine the steady-state error for an unit step input of R(s) as a
function of k.

(b) If k = 10, what is the expected overshoot ?

(c) Determine the maximum value of k that results in no overshoot.

(d) What is the settling time as a function of k ?

42/44
MECE 3350 - C. Rossa 42 / 44 Lecture 8
Answers to skills check

SC 22 - (b)

SC 23 - In (a) the transfer function is

Y (s) 100k
= 2
R(s) s + s(a + 25) + 25a + 100k
For question (b), we get k ≈ 85 and a ≈ 55.

SC 24 - (a) es s = 0 ∀k,
(b) P.O. = 35%,
(c) k < 1,
(d) Ts = 4 sec ∀k

43/44
MECE 3350 - C. Rossa 43 / 44 Lecture 8
Next class...

• Dominant poles

44/44
MECE 3350 - C. Rossa 44 / 44 Lecture 8