Automatic Control

Lecture #3 Dynamic Response

Bo-Chiuan Chen
National Taipei University of Technology

1
 Review of Laplace Transforms

• Linear time-invariant (LTI) systems

– A linear system response obeys the principle of superposition.
𝑢1 𝑡 ⇒ 𝑦1 𝑡
𝑢2 𝑡 ⇒ 𝑦2 𝑡
𝛼1 𝑢1 𝑡 + 𝛼2 𝑢2 𝑡 ⇒ 𝛼1 𝑦1 𝑡 + 𝛼2 𝑦2 𝑡

– The response of a LTI system can be expressed as the convolution of the input with
the unit impulse response of the system.
• Impulse
1
𝛿(𝑡) = lim+ 1(𝑡) − 1(𝑡 − 𝜀 )
𝜀→0 𝜀
• Convolution Integral (a special form of superposition integral)
∞
𝑦 𝑡 = න 𝑢 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 𝛿 𝑡 system ℎ 𝑡
−∞
Impulse response at t The transfer function 𝐻 𝑠 is the Laplace
or
to an input applied at  transform of the unit impulse response ℎ(𝑡).
∞
𝑌 𝑠
𝑦 𝑡 = න 𝑢 𝑡 − 𝜏 ℎ 𝜏 𝑑𝜏 =𝐻 𝑠
−∞ 𝑈 𝑠 2
Convolution

1
𝑝Δ 𝑡 = ቐΔ , 0≤𝑡≤Δ
ℎΔ (𝑡)
0, elsewhere

𝑦 𝑡 = ෍ ∆𝑢 𝑘Δ ℎΔ (𝑡 − Δ𝑘)
𝑘=0

∞
𝑦 𝑡 = න 𝑢 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
−∞

3
 Transfer Function

• The transfer function 𝐻 𝑠 is the ratio of the Laplace transform of the output of
the system to its input assuming all zero initial conditions.

𝑌 𝑠 𝑢(𝑡) 𝑦(𝑡)
= 𝐻(𝑠) 𝐻(𝑠)
𝑈 𝑠

• If the input 𝑢(𝑡) is the unit impulse 𝛿(𝑡), 𝑦(𝑡) is the unit impulse response, i.e.
𝑦 𝑡 = ℎ(𝑡).

𝐿 𝛿 𝑡 =1 𝑌 𝑠 = 𝐻(𝑠)

• The transfer function 𝐻(𝑠) is the Laplace transform of the unit impulse response
ℎ(𝑡).

4
 Review of Laplace Transforms

• Ex. 3.5 Transfer Function for an RC Circuit

From KVL,
𝑅𝑖 𝑡 + 𝑦 𝑡 = 𝑢(𝑡)
𝑑𝑦 𝑡
𝑖 𝑡 =𝐶
𝑑𝑡
𝑅𝐶 𝑦ሶ + 𝑦 = 𝑢

𝑅𝐶𝑠𝑌(𝑠) + 𝑌(𝑠) = 𝑈(𝑠)

𝑌 𝑠 1
𝐻 𝑠 = =
𝑈 𝑠 𝑅𝐶𝑠 + 1

5
 Review of Laplace Transforms
1 𝜔
• Ex. 3.6 Frequency Response 𝑀= , 𝜑 = −tan−1
𝜔2 + 𝑘2 𝑘
𝑢 = 𝐴sin 𝜔𝑡 1 𝑦 = 𝐴𝑀sin 𝜔𝑡 + 𝜑
𝐻 𝑠 =
𝑢 = 𝐴cos 𝜔𝑡 𝑠+𝑘 𝑦 = 𝐴𝑀cos 𝜔𝑡 + 𝜑
0
10
k=1;
num=1;
den=[1 k]; Magnitude -1
10
sysH=tf(num,den);
w=logspace(-2,2);
[mag,phase]=bode(sysH,w); -2
figure(1) 10
-2 -1 0 1 2
10 10 10 10 10
subplot(2,1,1)
loglog(w,squeeze(mag)); 0
ylabel('Magnitude');
phase (deg)

grid on;
subplot(2,1,2) -50
semilogx(w,squeeze(phase));
xlabel('w (rad/sec)');
ylabel('phase (deg)');
grid on; -100
-2 -1 0 1 2
10 10 10 10 10
w (rad/sec) 6
 Review of Laplace Transforms

0, 𝑡<0
• If 𝑢 𝑡 = sin 10𝑡 1 𝑡 Unit Step 1 𝑡 = ቊ
1, 𝑡≥0
10 −𝑡 1
𝑦 𝑡 = 𝑒 + sin 10𝑡 + 𝜑 = 𝑦1 𝑡 + 𝑦2 𝑡
101 101

Phase delay ≈ 𝜔𝛿𝑡

7
 Review of Laplace Transforms

• How to solve the time-domain response

8
 Review of Laplace Transforms

• Definition of Laplace Transform

∞
𝐹 𝑠 ≜ න 𝑓 𝑡 𝑒 −𝑠𝑡 𝑑𝑡
0−

• Step and Ramp Transform

𝑎 𝑏
𝑓 𝑡 = 𝑎1 𝑡 ⇒ 𝐹 𝑠 = 𝑓 𝑡 = 𝑏𝑡1 𝑡 ⇒ 𝐹 𝑠 =
𝑠 𝑠2
• Impulse Function Transform

𝑓 𝑡 =𝛿 𝑡 ⇒𝐹 𝑠 =1

• Sinusoid Transform
𝜔
𝑓 𝑡 = sin 𝜔𝑡 ⇒ 𝐹 𝑠 =
𝑠2 + 𝜔2
𝑠
𝑓 𝑡 = cos 𝜔𝑡 ⇒ 𝐹 𝑠 = 2
𝑠 + 𝜔2

9
 Properties of Laplace Transforms

• Superposition
𝐿 𝛼𝑓1 𝑡 + 𝛽𝑓2 𝑡 = 𝛼𝐹1 𝑠 + 𝛽𝐹2 𝑠
• Time Delay
∞
𝐹1 𝑠 = න 𝑓 𝑡 − 𝜆 𝑒 −𝑠𝑡 𝑑𝑡 = 𝑒 −𝑠𝜆 𝐹 𝑠
0−
• Differentiation
𝑑𝑓
𝐿 = 𝑠𝐹 𝑠 − 𝑓 0 𝐿 𝑓ሷ = 𝑠 2 𝐹 𝑠 − 𝑠𝑓 0 − 𝑓ሶ 0
𝑑𝑡
𝐿 𝑓𝑚 𝑡 = 𝑠 𝑚 𝐹 𝑠 − 𝑠 𝑚−1 𝑓 0 − 𝑠 𝑚−2 𝑓ሶ 0 − ⋯ − 𝑓 𝑚−1 0
• Integration
𝑡
1
𝐹1 𝑠 = 𝐿 න 𝑓 𝜉 𝑑𝜉 = 𝐹 𝑠
0 𝑠
• Convolution
𝐿 𝑓1 𝑡 ∗ 𝑓2 𝑡 = 𝐹1 𝑠 𝐹2 𝑠

10
 Partial-Fraction Expansion - 1

• Consider zero
𝑏1 𝑠 𝑚 + 𝑏2 𝑠 𝑚−1 + ⋯ + 𝑏𝑚+1 ς𝑚
𝑖=1 𝑠 − 𝑧𝑖
𝐹 𝑠 = =𝐾 𝑛
𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛 ς𝑖=1 𝑠 − 𝑝𝑖
𝐶1 𝐶2 𝐶𝑛 pole (modes)
= + + ⋯+
𝑠 − 𝑝1 𝑠 − 𝑝2 𝑠 − 𝑝𝑛
Characteristic Equation
𝐶𝑖 = 𝑠 − 𝑝𝑖 𝐹 𝑠 |𝑠=𝑝𝑖
𝑛

𝑓 𝑡 = ෍ 𝐶𝑖 𝑒 𝑝𝑖𝑡 1 𝑡
𝑖=1

• Ex. 3.11 Distinct Real Roots

𝑠+2 𝑠+4 𝐶1 𝐶2 𝐶3
𝑌 𝑠 = = + + 𝐶1 =? 𝐶2 =? 𝐶3 =?
𝑠 𝑠+1 𝑠+3 𝑠 𝑠+1 𝑠+3

num=conv([1 2],[1 4]);

den=conv([1 1 0],[1 3]);
Does the result agree with
[r,p,k]=residue(num,den); the hand calculation?
11
 Partial-Fraction Expansion - 2

• Repeat roots

𝐴𝑟 = 𝑠 + 𝑠𝑖 𝑟 𝐺 𝑠 |𝑠=−𝑠𝑖
𝑑
𝐴𝑟−1 = 𝑠 + 𝑠𝑖 𝑟 𝐺 𝑠 |𝑠=−𝑠𝑖
𝑑𝑠
1 𝑑2
𝐴𝑟−2 = 2
𝑠 + 𝑠𝑖 𝑟 𝐺 𝑠 |𝑠=−𝑠𝑖
2! 𝑑𝑠 How should we deal with the complex roots?
⋮
Complete the squares for the denominator.
1 𝑑 𝑟−1
𝐴1 = 𝑟−1 𝑠 + 𝑠𝑖 𝑟 𝐺 𝑠 |𝑠=−𝑠𝑖
𝑟 − 1 ! 𝑑𝑠 12
 Partial-Fraction Expansion - 3

• Example of Repeat roots

13
 Final Value Theorem

• If all poles of 𝑠𝑌(𝑠) are in the left half of the s-plane, then
lim 𝑦 𝑡 = lim𝑠𝑌 𝑠
𝑡→∞ 𝑠→0

• Ex. 3.12
3 𝑠+2
𝑌 𝑠 =
𝑠 𝑠 2 + 2𝑠 + 10

3 𝑠+2 6
𝑦 ∞ = lim𝑠𝑌 𝑠 = lim𝑠 = = 0.6
𝑠→0 𝑠→0 𝑠 𝑠 2 + 2𝑠 + 10 10
1
DC gain = lim𝑠𝐺 𝑠 = lim𝐺 𝑠
𝑠→0 𝑠 𝑠→0
3 𝑠+2
If 𝐺 𝑠 = DC gain = 0.6
𝑠 2 + 2𝑠 + 10
3
What about 𝑌 𝑠 = ?
𝑠 𝑠−2

14
 Homework #3

• Problem 3.7 (b), (g)

• Problem 3.9 (b)
• Problem 3.15
• Problem 3.17

15
 Block Diagram

• Elementary Block Diagrams

Negative
Feedback

• Block-diagram Algebra

Move a pickoff point

16
 Block Diagram

• Block-diagram Algebra (cont.)

Move a summer

Unity feedback system

17
 Block Diagram

• Ex. 3.22

2s  4
Y s 2s  4
T s 
2
 s  2
R  s  1  2s  4 s  2s  4
s2

18
 Block Diagram

• Ex. 3.23
𝑌 𝑠
𝑇 𝑠 =
𝑅 𝑠
𝐺1 𝐺2
1 − 𝐺1 𝐺3 𝐺6
= 𝐺5 +
𝐺1 𝐺2 𝐺4 𝐺2
1+1−𝐺 𝐺
1 3
𝐺1 𝐺2 𝐺5 + 𝐺1 𝐺6
=
1 − 𝐺1 𝐺3 + 𝐺1 𝐺2 𝐺4

19
 Effect of Pole Locations

• For a transfer function, zero

𝑏1 𝑠 𝑚 + 𝑏2 𝑠 𝑚−1 + ⋯ + 𝑏𝑚+1 ς𝑚𝑖=1 𝑠 − 𝑧𝑖

𝐻 𝑠 = = 𝐾
𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛 ς𝑛𝑖=1 𝑠 − 𝑝𝑖
1 pole (modes)
If 𝐻 𝑠 = ℎ 𝑡 = 𝑒 −𝜎𝑡 1 𝑡
𝑠+𝜎
Characteristic Equation
When 𝜎 > 0, the pole is located at 𝑠 < 0, and we say the impulse response (natural
response) is stable. Otherwise, it is unstable.
• Time constant
– Defined as the time
when the response
is 1/e times the
initial value.
1
𝜏=
𝜎

20
 Effect of Pole Locations

• Ex. 3.25
2s  1
H s 
s 2  3s  2
1 3
 
s 1 s  2
et  3e2t , t  0
h t   
 0, t0

“Fast poles” and “slow poles”

refer to relative rate of signal decay.

numH = [2 1];
denH = [1 3 2];
sysH = tf(numH, denH);
impulse(sysH);

21
 Effect of Pole Locations

• Impulse Responses

Neutrally stable 22
 Complex Poles

• Standard Second-order Transfer Function

𝜔𝑛2
𝐻 𝑠 = 2 𝜎 = 𝜁𝜔𝑛 , 𝜔𝑑 = 𝜔𝑛 1 − 𝜁 2
𝑠 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

𝜁 is the damping ratio

𝜔𝑛 is the undamped natural frequency
𝜔𝑛
ℎ 𝑡 = 𝑒 −𝜎𝑡 sin 𝜔𝑑 𝑡 1 𝑡
1 − 𝜁2

23
 Complex Poles

Impulse response Step response

24
 Stability

• Stability depends on whether natural response grows or decays.

25
 Oscillatory Time Response

• Ex. 3.26
2𝑠 + 1 1
𝐻 𝑠 = 2 𝜔𝑛 = 5, 𝜁 =
𝑠 + 2𝑠 + 5 5
2𝑠 + 1
=
𝑠 + 1 2 + 22
𝑠+1 1 2
=2 −
𝑠 + 1 2 + 22 2 𝑠 + 1 2 + 22

1
ℎ 𝑡 = 2𝑒 −𝑡 cos 2𝑡 − 𝑒 −𝑡 sin 2𝑡 1 𝑡
2

numH = [2 1];
denH = [1 2 5];
sysH = tf(numH, denH);
t = 0:0.1:6;
y = impulse(sysH, t);
plot(t, y);
26
 Time-domain Specifications

• Rise time 𝑡𝑟
• Settling time 𝑡𝑠
• Overshoot 𝑀𝑝
• Peak time 𝑡𝑝

27
 Time-domain Specifications
1.8
• Rise time 𝑡𝑟 ≅ What about 1st order system?
𝜔𝑛

• Overshoot and Peak Time

𝜎
𝑦 𝑡 = 1 − 𝑒 −𝜎𝑡 cos𝜔𝑑 𝑡 + sin𝜔𝑑 𝑡
𝜔𝑑

𝜎2 𝜎
=1− 𝑒 −𝜎𝑡 1 + 2 cos 𝜔𝑑 𝑡 − 𝛽 , 𝛽 = tan−1 𝜔𝑑
𝜔𝑑

𝜎2 𝜋
𝑦ሶ 𝑡 = 𝑒 −𝜎𝑡 sin𝜔𝑑 𝑡 + 𝜔𝑑 sin𝜔𝑑 𝑡 = 0 𝑡𝑝 =
𝜔𝑑 𝜔𝑑
𝜎
𝑦 𝑡𝑝 ≜ 1 + 𝑀𝑝 = 1 − 𝑒 −𝜎𝜋Τ𝜔𝑑 cos𝜋 + sin𝜋 = 1 + 𝑒 −𝜎𝜋Τ𝜔𝑑
𝜔𝑑

1−𝜁 2 ,
𝑀𝑝 = 𝑒 −𝜎𝜋Τ𝜔𝑑 = 𝑒 −𝜋𝜁 Τ 0≤𝜁<1

28
 Time-domain Specifications

• Settling Time
The value of 𝑡𝑠 when the decaying exponential reaches 1%.
4.6
𝑒 −𝜎𝑡 = 0.01 −𝜎𝑡𝑠 = −4.6 𝑡𝑠 =
𝜎
• Transient requirements in s-plane

Rise time Overshoot Settling Time

29
 Transformation of the Spec to the s-Plane

• Ex. 3.27

𝑡𝑟 ≤ 0.6 sec, 𝑀𝑝 ≤ 10%, 𝑡𝑠 ≤ 3 sec

1.8
𝜔𝑛 ≥ = 3.0 rad/sec
𝑡𝑟
𝜁 ≥ 0.6
4.6 4.6
𝜎≥ = = 1.5
𝑡𝑠 3

30
 Effects of Zeros

• The effect of zeros near poles

– In general, a zero near a pole reduces the amount of that term in the total response.

2 2 𝑠 + 1.1
𝐻1 𝑠 = 𝐻2 𝑠 =
𝑠+1 𝑠+2 1.1 𝑠 + 1 𝑠 + 2
2 2 2 0.1 0.9
= + = +
𝑠+1 𝑠+2 1.1 𝑠 + 1 𝑠 + 2
0.18 1.64
= +
𝑠+1 𝑠+2

31
 Effects of Zeros

• A second-order system with a zero

𝑠Τ𝛼 𝜁𝜔𝑛 + 1
𝐻 𝑠 = 𝛼↑ The extra zero is more far away.
𝑠Τ𝜔𝑛 2 + 2𝜁 𝑠Τ𝜔𝑛 + 1

𝜁 = 0.5

The major effect of the zero is to increase the overshoot, whereas it has very little influence
on the settling time. 32
 RHP or Non-minimum Phase Zero

• Zero in the Right-hand Plane (RHP)

Minimum-phase zero (LHP) Non-minimum-phase zero (RHP)

The only significant effect of the non-minimum phase zero is to cause the initial response
to go to the “wrong direction” and make the response somewhat sluggish.
30 −𝑠 + 6 1
Practice: 𝑌 𝑠 = 2
𝑠 + 4𝑠 + 13 𝑠
33
 Proximity of the Zero

• Ex. 3.28 Consider the second-order system with a finite zero and unity DC gain,
24 𝑠+𝑧
𝐻 𝑠 =
𝑧 𝑠+4 𝑠+6

First-order response or
second-order response?

34
 Proximity of the Complex Zeros

• Ex. 3.29 Consider the third-order feedback system with a pair of lightly damped
poles and a pair of complex zeros with the transfer function,

𝑠 + 𝛼 2 + 𝛽2
𝐻 𝑠 =
𝑠 + 1 𝑠 + 0.1 2 + 1
𝛽=1

35
 Effect of Extra Pole

• An extra pole on the standard 2nd-order system

1
𝐻 𝑠 = 2
𝑠Τ𝛼 𝜁𝜔𝑛 + 1 𝑠Τ𝜔𝑛 + 2𝜁 𝑠Τ𝜔𝑛 + 1

The major effect is to increase the rise time.

𝜁 = 0.5

36
 Effects of Pole-Zero Patterns

3.24 𝑀𝑝 = 𝑒 −𝜋𝜁 Τ 1−𝜁 2

3.29.

3.37.

37
 Design Problem

• Problem
below,

38
 Homework #4

• Problem 3.20(c)
• Problem 3.27 (hint: compare coefficients of the resulting closed-loop
characteristic equation to that of the standard 2nd-order system)
• Problem 3.35

39
 Stability

• A LTI system is said to be stable if all the roots of the transfer function
denominator polynomial have negative real parts (i.e. they are all in the left-
hand s-plane) and is unstable otherwise.
• A system is said to have bounded input-bounded output (BIBO) stability if
every bounded input results in a bounded output (regardless of what goes on
inside the system).
∞

න |ℎ 𝜏 |𝑑𝜏 < ∞ BIBO stable

−∞
• Before any cancellations of poles by zeros are made and 𝑚 ≤ 𝑛:
𝑛
𝑌 𝑠 𝑏0 𝑠 𝑚 + 𝑏1 𝑠 𝑚−1 + ⋯ + 𝑏𝑚 ς𝑚𝑖=1 𝑠 − 𝑧𝑖
𝑇 𝑠 = = = 𝐾 𝑦 𝑡 = ෍ 𝐾𝑖 𝑒 𝑝𝑖𝑡
𝑅 𝑠 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛 ς𝑛𝑖=1 𝑠 − 𝑝𝑖
𝑖=1
The system is stable iff every term goes to zero as t  
Re 𝑝𝑖 < 0
Internal stability occurs when all poles are strictly in the LHP.
What about non repeated poles on imaginary axis? (neutral stable)
What about repeated poles on imaginary axis? (unstable)
40
 Routh's Stability Criterion

• Necessary Condition
All the coefficients of the characteristic polynomial be positive.
• Necessary and sufficient condition
All the elements in the first column
of the Routh array are positive.
• If 𝑎 𝑠 = 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + 𝑎2 𝑠 𝑛−2 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛

41
 Routh's Stability Criterion

• Ex. 3.32
𝑎 𝑠 = 𝑠 6 + 4𝑠 5 + 3𝑠 4 + 2𝑠 3 + 𝑠 2 + 4𝑠 + 4

The number of RHP roots equals to the number of

sign changes in the first column.
42
 Stability vs. Parameter Range

• Ex. 3.33
Closed-loop characteristic eq.
𝑠+1
1+𝐾 =0
𝑠 𝑠−1 𝑠+6

𝑠 3 + 5𝑠 2 + 𝐾 − 6 𝑠 + 𝐾 = 0

𝐾 > 7.5

If we want Re 𝑝𝑖 < −1,

what should be done for Routh Stability analysis? (Hint: Let 𝑠 = 𝑞 − 1.)
43
 Stability vs. Two Parameter Ranges

• Ex. 3.34
Closed-loop characteristic eq.
𝐾𝐼 1
1+ 𝐾+ =0
𝑠 𝑠+1 𝑠+2

𝑠 3 + 3𝑠 2 + 2 + 𝐾 𝑠 + 𝐾𝐼 = 0

Allowable region for Stability

44
 Routh's Test for Special Case I

• Ex. W3.3 Replace the zero with a small positive constant

𝑎 𝑠 = 𝑠 5 + 3𝑠 4 + 2𝑠 3 + 6𝑠 2 + 6𝑠 + 9

Replace zero by 

6𝜀−9
9
𝜀
9𝜀2
3 − 6𝜀−9 0
9 6𝜀 − 9 6𝜀 − 9
3 − 9𝜀 ൘
𝜀 𝜀
6𝜀 − 9
lim =?
𝜀→0 𝜀
When to use l'Hôpital's rule?
9𝜀 2
lim 3 − =?
𝜀→0 6𝜀 − 9

45
 Routh's Test for Special Case II

• Ex. W3.4 Replace the zero row by coefficients of the derivative of the auxiliary
polynomial from the previous non-zero row

𝑎 𝑠 = 𝑠 5 + 5𝑠 4 + 11𝑠 3 + 23𝑠 2 + 28𝑠 + 12

𝑎1 𝑠 = 3𝑠 2 + 12 (auxiliary equation)

𝑑𝑎1 𝑠
= 6𝑠 + 0
𝑑𝑠

46
Summaries - 1

(3.91)

(3.92)

(3.93) 47
 Summaries - 2

3.10 3.58

(3.94)

3.16

48
 Summaries - 3

3.18

3.23

(3.95)

(3.96)

(3.97)

3.27 3.29 RHP zero Nonminimum phase

3.36 3.37

3.6.3
49
 Homework #5

• Problem 3.53 (a)

• Problem 3.54 (b), (d), and (e)
• Problem 3.56

50