Automatic Control System

Controller Design via Root Locus

Dr. Mahmoud Elnaggar

Electrical Power Engineering – Cairo University

Lecture Outline

• Proportional Controller

• Controllers Categorization

• Improving Steady State Performance

– Lag Compensator and PI Controller

• Improving Transient Performance

– Lead Compensator and PD Controller

• Lead-Lag Compensator and PID Controller

M. Elnaggar
Root Locus with General Controller
• Consider closed loop system as, In general,
𝑠 + 𝑧1 …
𝐺 𝑠 = 𝑘𝑔
𝑠 + 𝑝1 𝑠 + 𝑝2 …
𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
𝑠 + 𝑧ℎ1 … +
𝐶(𝑠) 𝐺(𝑠)
𝐻 𝑠 = 𝑘ℎ −
𝑠 + 𝑝ℎ1 …
𝑠 + 𝑧𝑐1 … 𝐻(𝑠)
𝐶 𝑠 = 𝑘𝑐
𝑠 + 𝑝𝑐1 …
• The closed loop system is given by,
𝐶 𝑠 𝐺 𝑠
𝐶𝑙𝑜𝑠𝑒𝑑 𝐿𝑜𝑜𝑝 𝑇𝐹 =
1 + 𝐶 𝑠 𝐺 𝑠 𝐻(𝑠)
• To use the Root Locus as an analysis and design tool, the c/cs is given as,
𝑠 + 𝑧1 𝑠 + 𝑧ℎ1 𝑠 + 𝑧𝑐1 …
1 + 𝐶 𝐺 𝑠 𝐻 𝑠 = 0 → 1 + 𝑘𝑔 𝑘ℎ 𝑘𝑐 =0
𝑠 + 𝑝1 𝑠 + 𝑝2 𝑠 + 𝑝ℎ1 𝑠 + 𝑝𝑐1 …
The Root Locus Gain
M. Elnaggar 3
Proportional Controller
• In case of proportional controller,
𝑈(𝑠)
𝐶 𝑠 = = 𝑘𝑐
𝐸(𝑠)
𝑒(𝑡)
𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
𝑢 𝑡 = 𝑘𝑐 𝑒 𝑡 𝐶 𝑠 = 𝑘𝑐 𝐺(𝑠)
+
−
• The system closed loop,
𝐶 𝑠 𝐺 𝑠 𝐻(𝑠)
𝐶𝑙𝑜𝑠𝑒𝑑 𝐿𝑜𝑜𝑝 𝑇𝐹 =
1 + 𝐶 𝑠 𝐺 𝑠 𝐻(𝑠)
𝐶/𝐶𝑆 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 ∶ 1 + 𝐶 𝐺 𝑠 𝐻 𝑠 = 0
𝑜𝑝𝑒𝑛 𝑙𝑜𝑜𝑝 𝑧𝑒𝑟𝑜𝑠
𝐶/𝐶𝑆 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 ∶ 1 + 𝑘𝑔 𝑘ℎ 𝑘𝑐 =0
𝑜𝑝𝑒𝑛 𝑙𝑜𝑜𝑝 𝑝𝑜𝑙𝑒𝑠
• The proportional controller is used to obtain,
– Given steady state performance. Increasing 𝑘𝑐 decreases the steady
state error
– A specific transient performance if the corresponding root location
located on the root locus.
M. Elnaggar 4
Proportional Controller Example
• Consider the closed loop system,
10 3
𝐺 𝑠 = 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+3
The system is controlled via proportional
controller 𝑘𝑐

• The system characteristic equation ,

1 + 𝑘𝑐 𝐺 𝑠 𝐻 𝑠 = 0
X X X
10 3
1 + 𝑘𝑐 =0
𝑠+1 𝑠+2 𝑠+3

𝑘
1+ =0
𝑠 + 1 𝑠 + 2 (𝑠 + 3)
• System poles,
𝑠 = −1 , −2 , −3

M. Elnaggar 5
Proportional Controller Example
• Real Segment : from −1 to −2 and from
− 3 to −∞ .

• Asymptotes:
𝑞 = 𝑛 −𝑚 = 3 −0= 3
−1 −2 −3
𝜎𝑎 = = −2
3

2𝑖+1 𝜋 𝜋 5𝜋
𝜃𝑎 = = , 𝜋,
3 3 3 X X X

• Break point:
𝑑𝐺𝐻 𝑑 1
= =0
𝑑𝑠 𝑑𝑠 𝑠 3 + 6 𝑠 2 + 11 𝑠 + 6
3𝑠 2 + 12𝑠 + 11 = 0
𝑠 =− −1.42, −2.577

M. Elnaggar 6
Proportional Controller Example
• Intersection with the imaginary axis,

𝑘
1+ = 0
𝑠 + 1 𝑠 + 2 (𝑠 + 3)
𝑠 3 + 6 𝑠 2 + 11 𝑠 + 6 + 𝑘 = 0

𝑆3 1 11

𝑆2 6 6+𝑘 X X
X
𝑆1 60−𝑘
6

𝑆0 6 + 𝑘

Marginal stability occurs at 𝑘 = 60 ,

Aux : 6 𝑠 2 + 66 = 0 𝑠 = ±𝐽 11 = ±𝐽3.3

M. Elnaggar 7
Proportional Controller Example
What is the range of 𝑘𝑐 for stable system ?
From Routh Table the system is stable for ,
𝑘 < 60
𝑘𝑐 𝑘𝑔 𝑘ℎ < 60 𝑘𝑐 < 2
−1.13 + 𝐽1.13 𝑘1
If exist, What is the value of 𝑘𝑐 corresponds for 5 %
O.S, What is the corresponding steady state error ?
−𝜋𝜁 X X X
𝑂. 𝑆 = 5% = 𝑒 1−𝜁 2 𝜁 = 0.7 𝜃 = 45𝑜
ς 𝐿𝑝
𝑘1 = = 3.54 𝑘𝑐 = 0.118
ς 𝐿𝑧
𝐸𝑟𝑟𝑜𝑟 𝑠𝑡𝑒𝑎𝑑𝑦 𝑠𝑡𝑎𝑡𝑒 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 ∶ 𝑘𝑝 = lim 𝑘𝑐 𝐺(𝑠) = 0.59
𝑠→0
1
𝑒𝑠𝑠 = = 0.6289 = 62.89 %
1 + 𝑘𝑝

M. Elnaggar 8
Proportional Controller Example
If exist, What is the value of 𝑘𝑐 corresponds for 5 %
O.S and 2 second settling time?
−𝜋𝜁
1−𝜁 2
−2 + 𝐽2
𝑂. 𝑆 = 5% = 𝑒 𝜁 = 0.707 𝜃= 45𝑜
4
𝑇𝑠 = 𝑅𝑒𝑎𝑙 𝑝𝑎𝑟𝑡 = 2
𝑅𝑒𝑎𝑙 𝑝𝑎𝑟𝑡
The desired operating point is not located on the root
locus. There is not 𝑘𝑐 exist to operate at it.
X X X
If exist, What is the value of 𝑘𝑐 for 2% steady state
error?
1
𝑒𝑠𝑠 = = 0.02 𝑘𝑝 = 49
1 + 𝑘𝑝
𝑘𝑝 = lim 𝑘𝑐 𝐺(𝑠) = 49 𝑘𝑐 = 9.8
𝑠→0
However, this gain results in unstable system. There
is not 𝑘𝑐 exist to achieve this low error.

M. Elnaggar 9
Controllers Categorization
• Generally, the cascaded controllers can be categorized based on the
objective of the controller. The different categories are,

– Improve transient Performance 𝑒(𝑡)

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
• Lead Compensator +
𝐶(𝑠) 𝐺(𝑠)
• PD Controller −

𝐻(𝑠)
– Improve Steady State Performance
• Lag Compensator
• PI Controller

– Improve Transient and Steady State Performances

• Lead-Lag Compensator
• PID Controller

M. Elnaggar 10
Lag Compensator
• Consider the closed loop system with unity dcgain feedback.

• The lag compensator objective is to reduce the closed loop system steady
state error. It is achievable by increasing the steady state constant.
𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
• The lag compensator is given as, +
𝐶(𝑠) 𝐺(𝑠)
−
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐 𝐻 𝑠 =1
𝑠 + 𝑃𝑐
• The steady state constant is modified by adding the compensator as
(assuming the system is type zero),

𝑘𝑝 𝑢𝑛 = lim 𝐺 𝑠
𝑠→0

𝑘𝑝 𝑐𝑜𝑚 = lim 𝐶 𝑠 𝐺 𝑠
𝑠→0

M. Elnaggar 11
Lag Compensator
• As the system is considered type zero, Then the steady state constant ,

𝑘𝑝 𝑐𝑜𝑚 = lim 𝐶 𝑠 𝐺 𝑠 = lim 𝐶 𝑠 lim 𝐺 𝑠

𝑠→0 𝑠→0 𝑠→0

𝑘𝑝 𝑐𝑜𝑚 = 𝑘𝑝 𝑢𝑛 lim 𝐶 𝑠
𝑠→0

𝑘𝑝 𝑐𝑜𝑚 𝑍𝑐
= 𝑘𝑐
𝑘𝑝 𝑢𝑛 𝑃𝑐

• In lag compensator, 𝑘𝑐 ≈ 1 . Then ,

𝑘𝑝 𝑐𝑜𝑚 𝑍𝑐
= = 𝛼 𝛼≫1 𝑍𝑐 > 𝑃𝑐
𝑘𝑝 𝑢𝑛 𝑃𝑐
• Choosing the location of the pole and zero near the origin satisfies that the
ratio between them equals to 𝛼 with minimum change on the transient
performance.

M. Elnaggar 12
Lag Compensator Summary
• Lag compensator,
𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
𝑠 + 𝑍𝑐 𝐶(𝑠) 𝐺(𝑠)
𝐶 𝑠 = 𝑘𝑐 +
𝑠 + 𝑃𝑐 −

𝑍𝑐
𝑘𝑐 ≈ 1 , = 𝛼 , 𝛼≫1 , 𝑍𝑐 > 𝑃𝑐 𝐻 𝑠 =1
𝑃𝑐

• Main desired effect,

– Increase the system gain to achieve 𝑆𝑜𝑝 = 𝜎 + 𝐽𝜔
desired steady-state error specifications.
– Introduce little effect on the path of the
root locus,
∠ 𝑠 + 𝑍𝑐 − ∠ 𝑠 + 𝑃𝑐 ≤ 2𝑜

0.035 𝜎 2 + 𝜔2 O X
𝑃𝑐 = 𝑍𝑐 𝑃𝑐
𝜔 𝛼−1

M. Elnaggar 13
Lag Compensator Design Steps
• To design a lag compensator via root
locus, 𝑒(𝑡) 𝑢(𝑡)
𝑟(𝑡) 𝑦(𝑡)
1- Determine the operating transient +
𝐶(𝑠) 𝐺(𝑠)
−
point on the uncompensated root locus,
𝑆𝑜𝑝 . 𝐻 𝑠 =1
2- Find the steady state constant of the
system.
3- Determine the gain of the controller 𝑆𝑜𝑝 = 𝜎 + 𝐽𝜔
𝛼 that is required to meet the system
steady state desired.
4- Calculate the compensator pole
location and the zero location as,

0.035 𝜎 2 + 𝜔2 𝑍𝑐
𝑃𝑐 = , = 𝛼 O X
𝜔 𝛼−1 𝑃𝑐 𝑍𝑐 𝑃𝑐

M. Elnaggar 14
PI Controller
• PI (proportional integral) controller is considered as a special condition of
the Lag compensator. Its objective is to reduce the steady state error to
zero. i.e. achieve perfect tracking for the input signal.
• Most of industrial applications use PI controller.

𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)

• Recall, Lag compensator, 𝐶(𝑠) 𝐺(𝑠)
+
−
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠 + 𝑃𝑐 𝐻 𝑠 =1
As ,
𝑘𝑝 𝑐𝑜𝑚 𝑍𝑐
= = 𝛼 → ∞
𝑘𝑝 𝑢𝑛 𝑃𝑐
𝑖𝑚𝑝𝑙𝑖𝑒𝑠 𝑡ℎ𝑎𝑡 𝑃𝑐 = 0
• PI Controller is given as,
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠

M. Elnaggar 15
PI Controller
• PI Controller is given as, 𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝐶(𝑠) 𝐺(𝑠)
−
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠 𝐻 𝑠 =1
• The control signal can be expressed as,
𝑡
𝑈(𝑠) 𝑠 + 𝑍𝑐
= 𝑘𝑐 𝑢 𝑡 = 𝑘𝑐 𝑒 𝑡 + 𝑘𝑐 𝑍𝑐 න 𝑒 𝜏 𝑑𝜏
𝐸(𝑆) 𝑠 0

Proportional Term Integral Term

• For tuning purpose :

– 𝑘𝑝𝑟𝑜𝑝 = 𝑘𝑐 (the proportional term) is responsible for the transient
performance.
– 𝑘𝑖𝑛𝑡 = 𝑘𝑐 𝑍𝑐 (the integral term) is responsible for eliminating the steady
state error and minimize the effect on the transient performance.

M. Elnaggar 16
PI Controller
• PI Controller is given as, 𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝐶(𝑠) 𝐺(𝑠)
−
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠 𝐻 𝑠 =1
𝑡
𝑢 𝑡 = 𝑘𝑐 𝑒 𝑡 + 𝑘𝑐 𝑍𝑐 න 𝑒 𝜏 𝑑𝜏
0

• PI controller advantage:
– Eliminate the error steady state by introducing an integrator in the
feedforward.

• PI controller disadvantages:
– Adding a pole at the origin deteriorates the transient performance. It is
minimized by good tuning of the integral term gain.
– Windup Phenomena

M. Elnaggar 17
Lag Compensator - Example
• Consider the closed loop system,
𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝐶(𝑠) 𝐺(𝑠)
𝑘𝑔 −
𝐺 𝑠 =
𝑠 + 1 𝑠 + 2 𝑠 + 10
𝐻(𝑠)
𝐻 𝑠 =1
The system operates at damping ratio of 0.167. It is required to obtain the
system gain and design a cascading controller 𝐶 𝑠 to reduce the steady
state error to its 10% of its value and keep the same transient performance.

• First draw the uncompensated root locus

1
– Open loop system : 𝑠+1 𝑠+2 𝑠+10

– Poles at −1 , −2 , −10
– No zeros

M. Elnaggar 18
Lag Compensator - Example
• Real Segment : from −1 to −2 and from −10 to −∞ .

• Asymptotes:
𝑞 =𝑛 −𝑚 =3 −0=3
−1 −2 −10
𝜎𝑎 = = −4.333
3

2𝑖+1 𝜋 𝜋 5𝜋
𝜃𝑎 = = , 𝜋,
3 3 3

X X X
• Break Point:
𝑑𝐺𝐻 𝑑 1
= =0
𝑑𝑠 𝑑𝑠 𝑠 3 + 13 𝑠 2 + 32 𝑠 + 20

3𝑠 2 + 26𝑠 + 32 = 0

𝑠 = −1.4853, −7.1813

M. Elnaggar 19
Lag Compensator - Example
• Intersection with the imaginary axis,
𝑘
1+ =0
𝑠 + 1 𝑠 + 2 𝑠 + 10
𝑠 3 + 13𝑠 2 + 32𝑠 + 20 + 𝑘 = 0

𝑆3 1 32

𝑆 2 13 20 + 𝑘
X X X
𝑆 1 396−𝑘
13

𝑆 0 20 + 𝑘

Marginal stability occurs at 𝑘 = 396 ,

Aux : 13 𝑠 2 + 416 = 0 𝑠 = ±𝐽 32

M. Elnaggar 20
Lag Compensator - Example
• The system operates at damping 0.167
cos 𝜃 = 𝜁 = 0.167
𝜃 = 80𝑜
𝑠𝑜𝑝
• The intersection point is,
𝑠𝑜𝑝 = −0.694 + 𝐽3.926

• The Root Locus gain can be obtained as,

ς 𝐿𝑝
𝑘 = 𝑘𝑔 𝑘𝑐 𝑘𝐻 = = 164.6 X X X
ς 𝐿𝑧
𝑘𝑔 = 164.6
• The uncompensated system steady state,
164.6
𝑘𝑝 𝑢𝑛 = lim 𝐺 𝑠 = = 8.23
𝑠→0 1 ∗ 2 ∗ 10
1
𝑒𝑠𝑠 𝑢𝑛 = = 0.1083 = 10.83%
1 + 𝑘𝑝 𝑢𝑛

M. Elnaggar 21
Lag Compensator - Example
• The compensated system steady state,
1
𝑒𝑠𝑠 𝑐𝑜𝑚 = = 0.1 ∗ 0.1083
1 + 𝑘𝑝 𝑐𝑜𝑚 𝑠𝑜𝑝 = −0.694 + 𝐽3.926
𝑘𝑝 𝑐𝑜𝑚 = 91.3

• The compensator dcgain,

𝛼 = 𝑘𝑝 𝑐𝑜𝑚 /𝑘𝑝 𝑢𝑛 = 11.09

• The zero/pole location are, X X X

0.035 𝜎 2 + 𝜔2
𝑃𝑐 = = 0.014
𝜔 𝛼−1

𝑍𝑐 = 𝛼 ∗ 𝑃𝑐 = 0.1557
ς 𝐿𝑝 𝐿𝑃𝑐
𝑘𝑐 𝑘𝑔 = ς 𝐿𝑧
𝑘𝑐 = ≈1
𝐿𝑧𝑐

M. Elnaggar 22
Lag Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−

Lag Compensated
𝐻 𝑠 =1

Uncompensated
164.6
𝐺 𝑠 =
𝑠 + 1 𝑠 + 2 𝑠 + 10

𝑠 + 0.1557
𝐶 𝑠 =
𝑠 + 0.014

M. Elnaggar 23
Lag Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−

Lag Compensated 𝐻 𝑠 =1

Uncompensated
164.6
𝐺 𝑠 =
𝑠 + 1 𝑠 + 2 𝑠 + 10

𝑠 + 0.4672
𝐶 𝑠 =
𝑠 + 0.04212

M. Elnaggar 24
Lag Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−
Lag Compensated
𝐻 𝑠 =1
Uncompensated

164.6
𝐺 𝑠 =
𝑠 + 1 𝑠 + 2 𝑠 + 10

𝑠 + 0.05191
𝐶 𝑠 =
𝑠 + 0.00468

M. Elnaggar 25
Lag Compensator - Example
• Design PI controller to eliminate the steady
state error.
• The controller, 𝑠𝑜𝑝 = −0.694 + 𝐽3.926
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠
• Let, 𝑃𝑐 = 0
• Get angle,
X X X
∠ 𝑠𝑜𝑝 + 𝑃𝑐 = 𝜃𝑃𝑐 = 100𝑜
• Then,
∠ 𝑠𝑜𝑝 + 𝑍𝑐 = 𝑍𝑃𝑐 = 100 − 2 = 98𝑜
𝑍𝑐 = 0.14
• And, 𝑘𝑐 = 𝐿𝑧𝑐 /𝐿𝑃𝑐 ≈ 1

M. Elnaggar 26
Lag Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−

𝐻 𝑠 =1

164.6
𝐺 𝑠 =
𝑠 + 1 𝑠 + 2 𝑠 + 10

𝑠 + 0.14
𝐶 𝑠 =
𝑠

M. Elnaggar 27
PD Controller
• PD (proportional derivative) controller objective is to improve the system
transient performance by reducing the system overshot or settling time. It is
achievable by changing the system root loci to pass through the required
design specifications.
𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
• PD Controller is given as, +
𝐶(𝑠) 𝐺(𝑠)
−
𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐
𝐻(𝑠)

• The control signal can be expressed as,

𝑈(𝑠) 𝑑𝑒 𝑡
= 𝑘𝑐 𝑠 + 𝑍𝑐 𝑢 𝑡 = 𝑘𝑐 𝑍𝑐 𝑒 𝑡 + 𝑘𝑐
𝐸(𝑆) 𝑑𝑡

Proportional Term Derivative Term

• where : 𝑘𝑝𝑟𝑜𝑝 = 𝑘𝑐 𝑍𝑐 , 𝑘𝑑𝑒𝑟𝑖 = 𝑘𝑐

M. Elnaggar 28
PD Controller Effect
• Considering a system with open loop transfer function,
1
𝐺 𝑠 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+5

• The uncompensated system root locus states all the

possible roots locations using proportional controller.

• To achieve a maximum of 5 % overshot, the system

damping should be (𝜁 = 0.7) X X X
– Using the uncompensated system locus, 𝑝3 𝑝2 𝑝1
the dominant poles at
𝑘𝑐 = 7.4
𝑠 = −1.26 ± 𝐽1.26
– The corresponding settling time is,
4
𝑇𝑠 = = 3.17 𝑠𝑒𝑐
1.26

M. Elnaggar 29
PD Controller Effect
• Considering a system with open loop transfer function,
1
𝐺 𝑠 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+5

• Adding a PD controller as,

𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐
𝑍𝑐 = −4
• For the same condition, 5 % overshot (𝜁 = 0.7) 𝑍𝑐
X O X X
– Using the uncompensated system locus, 𝑝3 𝑝2 𝑝1
the dominant poles at
𝑘𝑐 = 3.84
𝑠 = −1.64 ± 𝐽1.64
– The corresponding settling time is,
4
𝑇𝑠 = = 2.44 𝑠𝑒𝑐
1.64

M. Elnaggar 30
PD Controller Effect
• Considering a system with open loop transfer function,
1
𝐺 𝑠 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+5

• Adding a PD controller as,

𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐
𝑍𝑐 = −3
• For the same condition, 5 % overshot (𝜁 = 0.7) 𝑍𝑐
X O X X
– Using the uncompensated system locus, 𝑝3 𝑝2 𝑝1
the dominant poles at
𝑘𝑐 = 7.93
𝑠 = −2.12 ± 𝐽2.12
– The corresponding settling time is,
4
𝑇𝑠 = = 1.89 𝑠𝑒𝑐
2.12

M. Elnaggar 31
PD Controller Effect
• Considering a system with open loop transfer function,
1
𝐺 𝑠 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+5

• Adding a PD controller as,

𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐
𝑍𝑐 = −2
• For the same condition, 5 % overshot (𝜁 = 0.7) 𝑍𝑐
X X
O X
– Using the uncompensated system locus, 𝑝3 𝑝2 𝑝1
the dominant poles at
𝑘𝑐 = 13
𝑠 = −3 ± 𝐽3
– The corresponding settling time is,
4
𝑇𝑠 = = 1.33 𝑠𝑒𝑐
3

M. Elnaggar 32
PD Controller Effect

𝑍𝑐 = −2
𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
𝑍𝑐 = −4 𝐶(𝑠) 𝐺(𝑠)
+
−
𝑍𝑐 = −4
𝐻(𝑠)
Uncompensated

1
𝐺 𝑠 𝐻 𝑠 =
𝑠+1 𝑠+2 𝑠+5

𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐

M. Elnaggar 33
PD Controller Disadvantages
1- Impulsive Effect 𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝑃𝐷 𝐺(𝑠)
As, −

𝑑𝑒 𝑡 𝑏(𝑡)
𝑢 𝑡 = 𝑘𝑐 𝑍𝑐 𝑒 𝑡 + 𝑘𝑐 𝐻(𝑠)
𝑑𝑡
𝑑 𝑟 𝑡 −𝑏 𝑡
𝑢 𝑡 = 𝑘𝑐 𝑍𝑐 𝑟 𝑡 − 𝑏 𝑡 + 𝑘𝑐
𝑑𝑡
In case of step reference signal, the derivative of the step at time 𝑡 = 0
results in impulse effect on the control signal . i.e. 𝑢 0 = ∞

-- Solution:
It can be solved by moving the 𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
𝐺(𝑠)
controller to the feedback to have +
−
its effect on the closed loop
transient and avoid direct derivation 𝑏(𝑡)
𝑃𝐷 𝐻(𝑠)
of the input signal.

M. Elnaggar 34
PD Controller Disadvantages
2- Noise Amplification 𝑟(𝑡) 𝑒(𝑡) 𝑢(𝑡) 𝑦(𝑡)
As, +
𝑃𝐷 𝐺(𝑠)
−
𝑑 𝑟 𝑡 −𝑏 𝑡
𝑢 𝑡 = 𝑘𝑐 𝑍𝑐 𝑟 𝑡 − 𝑏 𝑡 + 𝑘𝑐 𝑏(𝑡)
𝐻(𝑠)
𝑑𝑡

The signal 𝑏 𝑡 is the measurement of the output 𝑦 𝑡 . Most of the sensors

outputs suffer from high frequency noise.
i.e. 𝑏 𝑡 = 𝑦 𝑡 + 𝐴 sin 𝑤𝑡 𝐴≪1 𝑤≫1
Although the noise amplitude is small and ineffective, but it has high
frequency. Derivation of such a signal results in amplifying the amplitude.

-- Solution: 𝐿𝑒𝑎𝑑 𝐶𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑜𝑟

Cascading the PD controller with a 𝑟(𝑡) 𝑢 𝑦(𝑡)
low pass filter helps in eliminating +
𝑃𝐷 + 𝐿𝑃𝐹 𝐺(𝑠)
−
the unwanted noise effect.
𝑎 𝑏(𝑡)
𝑐 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐 𝐻(𝑠)
𝑠+𝑎
M. Elnaggar 35
Lead Compensator
• The lead compensator objective is same as the PD controller. It is to
improve the system transient performance. It is applied without direct
derivative to avoid noise amplification.
• Lead Compensator is called “ Practical PD Controller “

• The lag compensator is given as,

𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐 𝑎𝑛𝑑 𝑍𝑐 < 𝑃𝑐
𝑠 + 𝑃𝑐
𝑆𝑑𝑒𝑠𝑖𝑒𝑑
• The compensator added angle is,
𝜃𝑐 = 𝜃𝑧 𝑐 − 𝜃𝑝 𝑐

• Choosing of the pole/zero location is used to

locate the desired point on the root locus.
𝜃𝑝 𝑐 𝜃𝑧 𝑐
X O
𝑃𝑐 𝑍𝑐
M. Elnaggar 36
Lead Compensator – Design Steps
• To design a lead compensator via root locus,
1- Determine the desired transient point with
is corresponds to the required 𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
specifications. +
𝐶(𝑠) 𝐺(𝑠)
−
2- Calculate the angle at the desired point,
𝑏(𝑡)
෍ 𝜃𝑝 − ෍ 𝜃𝑧 = ± 𝜋 𝐻(𝑠)

The angle is function of pole/zero location.

3- Let zero location 𝑍𝑐 at 𝑆𝑑𝑒𝑠𝑖𝑒𝑑
• Real part of the desired point 𝑆𝑑𝑒𝑠𝑖𝑟𝑒𝑑
• Or , Cancel slowest stable open loop
pole.
4- Calculate the compensator pole location
using the angle condition. 𝜃𝑝 𝑐 𝜃𝑧 𝑐
X O
5- Calculate the required loop gain.
𝑃𝑐 𝑍𝑐
M. Elnaggar 37
Lead Compensator - Example
• Consider the closed loop system,
𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝐶(𝑠) 𝐺(𝑠)
𝑘𝑔 −
𝐺 𝑠 = , 𝐻 𝑠 =1
𝑠 𝑠+2
𝐻(𝑠)
The system operates at damping ratio of 0.707. It is required to obtain the
system gain and design a cascading controller 𝐶 𝑠 to achieve settling time
less than 1 sec.

• First draw the uncompensated root locus

1
– Open loop system : 𝑠(𝑠+2)

– Poles at 0, −2
– No zeros

M. Elnaggar 38
Lead Compensator - Example
• Real Segment : from 0 to −2
• Asymptotes:
𝑞 = 𝑛 −𝑚 = 3 −0= 2
0 −2
𝜎𝑎 = = −1
2

2𝑖+1 𝜋 𝜋 3𝜋
𝜃𝑎 = = ,
2 2 2

• Break Point: X X
𝑝2 𝑝1
𝑑𝐺𝐻 𝑑 1
= =0
𝑑𝑠 𝑑𝑠 𝑠 2 + 2𝑠

2𝑠 + 2 = 0

𝑠 = −1

M. Elnaggar 39
Lead Compensator - Example
• The system operates at damping 0.707
cos 𝜃 = 𝜁 = 0.707 𝑠𝑑
𝜃 = 45𝑜

• The intersection point is,

𝑠𝑜𝑝 = −1 + 𝐽1
𝑠𝑜𝑝
• The Root Locus gain can be obtained as,
ς 𝐿𝑝
𝑘 = 𝑘𝑔 𝑘𝑐 𝑘𝐻 = =2 X X
ς 𝐿𝑧
𝑝2 𝑝1
𝑘𝑔 = 2
• The desired root location,

4
𝑇𝑠 = = 1 𝑠𝑒𝑐
𝜎
𝑠𝑑 = −4 + 𝐽 4

M. Elnaggar 40
Lead Compensator - Example
• The lead compensator is given by,
𝑠𝑑
𝑠 + 𝑍𝑐
𝐶 𝑠 = 𝑘𝑐
𝑠 + 𝑃𝑐
• Let, 𝑍𝑐 = −4
• Applying the angle condition, 𝑠𝑜𝑝
𝜃𝑝1 + 𝜃𝑝1 + 𝜃𝑝𝑐 − 𝜃𝑧𝑐 = 180𝑜
𝜃𝑝1 = 180 − tan−1 4/4 = 135𝑜 𝜃𝑃𝑐 𝜃𝑧𝑐
X O X X
𝜃𝑝2 = 180 − tan−1 4/2 = 116.5𝑜 𝑃𝑐 𝑍𝑐 𝑝2 𝑝1
𝜃𝑧𝑐 = 90
• So,
4
𝜃𝑝𝑐 = tan−1 = 18.5 𝑃𝑐 = 16
𝑝𝑐 − 4

M. Elnaggar 41
Lead Compensator - Example
• The gain is calculated as,
ς 𝐿𝑝
𝑘 = 𝑘𝑔 𝑘𝑐 𝑘𝐻 = = 78.6
ς 𝐿𝑧
𝑘𝑐 = 78.6/2 = 39.7
• Then the lead compensator is,
𝑠+4
𝐶 𝑠 = 39.7
𝑠 + 16
• The compensated open loop,
𝑠+4
𝐶 𝑠 𝐺 𝑠 = 78.6
𝑠(𝑠 + 2)(𝑠 + 16)
It becomes a third order system.

𝑠+2
• Remark that, another solution is, 𝐶 𝑠 = 16 𝑠+8

M. Elnaggar 42
Lead Compensator - Example

𝐶 𝑠 = 39.4 𝑠 + 4 /(𝑠 + 16) 𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−
𝑢𝑛𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑒𝑑
𝐻(𝑠)

2
𝐶 𝑠 = 16 𝑠 + 2 /(𝑠 + 8) 𝐺 𝑠 =
𝑠 𝑠+2

𝐻 𝑠 =1

M. Elnaggar 43
Lead Compensator - Example
• Same problem can be solved using PD controller. The controller is given by,
𝐶 𝑠 = 𝑘𝑐 𝑠 + 𝑍𝑐
𝑠𝑑

• Applying the angle condition,

𝜃𝑝1 + 𝜃𝑝1 − 𝜃𝑧𝑐 = 180𝑜
𝜃𝑝1 = 180 − tan−1 4/4 = 135𝑜 𝑠𝑜𝑝
𝜃𝑝2 = 180 − tan−1 4/2 = 116.5𝑜
𝜃𝑍𝑐
• So, X X X
𝑍𝑐 𝑝2 𝑝1
4
𝜃𝑍𝑐 = tan−1 = 71.5 𝑍𝑐 = 5.33
𝑍𝑐 − 4
• The gain is calculated as,
ς 𝐿𝑝
𝑘 = 𝑘𝑔 𝑘𝑐 𝑘𝐻 = =6 𝑘𝑐 = 6/2 = 3
ς 𝐿𝑧
• Then the controller is given as, 𝐶 𝑠 = 3 𝑠 + 5.33

M. Elnaggar 44
Lead Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

+
𝐶(𝑠) 𝐺(𝑠)
−
𝑢𝑛𝑐𝑜𝑚𝑝𝑒𝑛𝑠𝑎𝑡𝑒𝑑
𝐻(𝑠)

𝐶 𝑠 = 3 𝑠 + 5.33
2
𝐺 𝑠 =
𝑠 𝑠+2

𝐻 𝑠 =1

M. Elnaggar 45
Lead Compensator vs PD Controller
𝑟(𝑡) 𝑢(𝑡) 1 𝑦(𝑡)
𝐶(𝑠)
+
−
𝑠 𝑠+2

𝑛 𝑡 = 0.001 sin(100 𝑡)

𝐶 𝑠 = 3 𝑠 + 5.33
𝑠+4
𝐶 𝑠 = 39.4
𝑠 + 16

M. Elnaggar 46
Lead-Lag Compensator
• Lead-Lag compensators are used to improve both transient and steady
state performance for a given system.
• Lead compensator is used to,
– Improve the transient performance.
– There is no guarantee about the steady state performance.

• Lag compensator is used to,

– Improve the steady state performance.
– Minimize the negative effect on the transient performance.

• Designing Lead-Lag compensator depends on design Lead compensator

first to improve the transient performance and then cascade a lag
compensator to achieve the required steady state performance and keep
the desired transient specification achieved via the lead compensator.

M. Elnaggar 47
Lead-Lag Compensator – Design Steps
• To design a Lead-Lag compensator via Root Locus,
1- Analyze the uncompensated system.
2- Design a lead compensator to achieve the required transient
performance. i.e determine the pole/zero locations and the compensator
gain.
𝑘𝑐 𝑠 + 𝑧𝑐1
𝐶𝑙𝑒𝑎𝑑 𝑠 =
𝑠 + 𝑃𝑐1
3- Consider the open loop system and the Lead cascaded as a new
uncompensated system. Design lag controller for this system to achieve the
required steady state performance. i.e determine the pole/zero locations.
𝑠 + 𝑧𝑐2
𝐶𝑃𝐼 𝑠 =
𝑠 + 𝑃𝑐2
𝑟(𝑡) 𝑢 𝑦(𝑡)
4- The compensator is given as,
+
𝐿𝑎𝑔 𝐿𝑒𝑎𝑑 𝐺(𝑠)
−
𝑠 + 𝑧𝑐1 𝑠 + 𝑧𝑐2
𝐶𝑃𝐼𝐷 𝑠 = 𝑘𝑐
𝑠 + 𝑃𝑐1 𝑠 + 𝑃𝑐2 𝐻(𝑠)

M. Elnaggar 48
PID Controller
• Similar to lead-lag compensator, PID controller is the combination of PI and
PD controllers.
• PID stands for proportional integral derivative controller.
• It is the most common used controller in industry as it can achieve the
required transient performance with perfect reference tracking for most of
the applications.
𝐾𝑖𝑛𝑡
• PID controller is given as,
𝑠
𝑠 + 𝑧𝑐1 𝑠 + 𝑧𝑐2
𝐶𝑃𝐼𝐷 𝑠 = 𝑘𝑐 𝐾𝑑𝑖𝑟 𝑠
𝑠
𝑑𝑒 𝑡 𝑟(𝑡)
𝑘𝑝𝑟𝑜𝑝
𝑢 𝑦(𝑡)
𝑢 𝑡 = 𝑘𝑝𝑟𝑜𝑝 𝑒 𝑡 + 𝑘𝑑𝑒𝑟 +
𝐺(𝑠)
𝑑𝑡 −
𝑡
+𝑘𝑖𝑛𝑡 න 𝑒 𝜏 𝑑𝜏
0 𝐻(𝑠)

M. Elnaggar 49
PID Controller – Design Steps
• To design a PID controller via Root Locus,
1- Analyze the uncompensated system.
2- Design a PD controller to achieve the required transient performance.
𝐶𝑃𝐷 𝑠 = 𝑘𝑐 𝑠 + 𝑧𝑐1
3- Consider the open loop system and the PD cascaded as a new
uncompensated system. Design PI controller for this system to achieve the
required steady state performance.
𝑠 + 𝑧𝑐2
𝐶𝑃𝐼 𝑠 =
𝑠
4- For implementation, determine the proportional, derivative, and integral
gains.
𝑟(𝑡) 𝑢 𝑦(𝑡)
𝑠 + 𝑧𝑐1 𝑠 + 𝑧𝑐2 𝐶𝑃𝐼 (𝑠) 𝐶𝑃𝐷 (𝑠) 𝐺(𝑠)
𝐶𝑃𝐼𝐷 𝑠 = 𝑘𝑐 +
𝑠 −
𝑘𝑖𝑛𝑡
𝐶𝑃𝐼𝐷 𝑠 = 𝑘𝑝𝑟𝑜𝑝 + 𝑘𝑑𝑒𝑟 𝑠 + 𝐻(𝑠)
𝑠

M. Elnaggar 50
Example
• Consider the closed loop system,
𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)
+
𝐶(𝑠) 𝐺(𝑠)
𝑠+8 −
𝐺 𝑠 =
𝑠 + 3 𝑠 + 6 𝑠 + 10
𝐻(𝑠)
𝐻 𝑠 =1
• It is required to design a compensator to achieve,
- overshot of 20 %
- Peak time of 0.2 second
- zero error steady state for step input.

• First, design a PD controller to achieve the transient specifications,

𝑂. 𝑆 = 20 % 𝜁 = 0.456
𝑇𝑝 = 𝜋/𝜔𝑑 = 0.2 𝜔𝑑 = 15.7 = 𝜔𝑛 1 − 𝜁 2
Then, the required operating root location,
𝑠𝑑 = −𝜁𝜔𝑛 ± 𝐽𝜔𝑑 = −8.1 ± 𝐽 15.7
M. Elnaggar 51
Lead-Lag Compensator - Example
• The PD controller is,
𝐶𝑃𝐷 𝑠 = 𝑘𝑐 𝑠 + 𝑧𝑐1
• Applying the angle condition,
𝜃𝑝1 + 𝜃𝑝1 + 𝜃𝑝3 − 𝜃𝑧1 − 𝜃𝑧𝑐 = 180𝑜
𝜃𝑝1 = ∠ 𝑠𝑑 + 3 = 107.8𝑜
𝜃𝑝2 = ∠ 𝑠𝑑 + 6 = 97.5𝑜
𝜃𝑝3 = ∠ 𝑠𝑑 + 10 = 83.2𝑜
𝜃𝑧1 = ∠ 𝑠𝑑 + 8 = 90.36𝑜

• So,
15.7
𝜃𝑧𝑐1 = 18.160 = tan−1 𝑍𝑐1 = 55.95
𝑍𝑐1 − 8.1
• The compensator gain,
ς 𝐿𝑝 𝐿𝑝1 𝐿𝑝2 𝐿𝑝3
𝑘𝑐 = = = 5.33
ς 𝐿𝑧 𝐿𝑧1 𝐿𝑧𝑐1
M. Elnaggar 52
Lead-Lag Compensator - Example
• The PD controller is,
𝐶𝑃𝐷 𝑠 = 𝑘𝑐 𝑠 + 𝑧𝑐1 = 5.33 𝑠 + 55.95
• Second, Design a PI controller for the open loop system,
5.33 𝑠 + 8 𝑠 + 55.95
𝐶𝑃𝐷 𝑠 𝐺 𝑠 =
𝑠 + 3 𝑠 + 6 𝑠 + 10

• The PI controller is,

𝑠 + 𝑧𝑐2
𝐶𝑃𝐷 𝑠 =
𝑠 + 𝑝𝑐
where 𝑝𝑐 = 0 ,

𝜃𝑝𝑐 − 𝜃𝑧𝑐2 = 2𝑜
𝜃𝑝𝑐 = ∠ 𝑠𝑑 + 0 = 117𝑜
15.7
𝜃𝑧𝑐2 = 1150 = 180 − tan−1 𝑍𝑐1 = 0.7
8.1 − 𝑍𝑐2

M. Elnaggar 53
Lead-Lag Compensator - Example

𝑟(𝑡) 𝑢(𝑡) 𝑦(𝑡)

𝐶(𝑠) 𝐺(𝑠)
𝑃𝐼𝐷 +
−

𝑃𝐷
𝐻(𝑠)
𝑃𝑟𝑜𝑝 𝐶𝑜𝑛𝑡

𝑠+8
𝐺 𝑠 =
𝑠 + 3 𝑠 + 6 𝑠 + 10

𝐻 𝑠 =1

5.33 𝑠 + 55.95 𝑠 + 0.7

𝐶𝑃𝐼𝐷 𝑠 =
𝑠

M. Elnaggar 54
Controllers Design Summary

1- Increase system type

2- Eliminate steady state error
PI 𝑠 + 𝑧𝑐 3- Zero location near origin (LHS)
𝑘𝑐
Controller 𝑠 4- Tune zero location to minimize
effect on transient.
5- Causes wind up phenomena
Improve
Steady State

1- Reduce steady state error

𝑠 + 𝑧𝑐 2- zero and pole locate on LHS
Lag 𝑘𝑐
𝑠 + 𝑝𝑐 3- zero is greater than pole
Compensator
4- zero and pole are close to
minimize effect on transient

M. Elnaggar 55
Controllers Design Summary
1- Zero location change the root
locus shape.
2- Gain is selected based on new
PD
𝑘𝑐 𝑠 + 𝑧𝑐 root locus.
Controller
3- causes impulsive effect with step
input
4- Causes noise amplification
Improve
Transient
Performance
1- Zero location to cancel slowest
𝑠 + 𝑧𝑐 stable pole.
Lead 𝑘𝑐
𝑠 + 𝑝𝑐 2- zero is smaller than pole
Compensator
3- considered as a practical version
of PD controller to reduce noise.

M. Elnaggar 56