1

Control Engineering
Definitions ........................................................................................................................... 12
System or Plant ............................................................................................................... 12
Output: ........................................................................................................................... 12
Input: ............................................................................................................................... 12
Set point: ......................................................................................................................... 12
Goal of Control: ............................................................................................................... 12
Example: Water Tank System ...................................................................................... 12
Concept of control loop ....................................................................................................... 13
Controller: ........................................................................................................................ 13
Actuator: .......................................................................................................................... 13
Example: Airplane Actuator .......................................................................................... 13
Example: missile and helicpters: .................................................................................. 13
Example: inverted Pendulum ....................................................................................... 14
Properties of a system ........................................................................................................ 14
Statical or dynamical system: .......................................................................................... 14
Example: Modeling a DC motor ................................................................................... 14
Linearity:.......................................................................................................................... 14
Example: Non-linear system ........................................................................................ 14
Time Variant .................................................................................................................... 15
Representing systems in the Laplace domain.................................................................. 15
Solving a D.E................................................................................................................... 15
Properties of the Laplace transform ................................................................................. 16
Modeling Dynamical Systems ............................................................................................. 16
Modeling Electrical systems ............................................................................................ 16
Example: RL Circuit ..................................................................................................... 18
Example: Series RLC Circuit ........................................................................................ 18
Example: Electrical circuit 2nd order.............................................................................. 19
Example: electrical circuit 2nd order............................................................................. 20
Example: Mechanical System ...................................................................................... 23
Example: Mass-Spring Damper system ....................................................................... 23
Example (this is probably wrong!) ................................................................................ 24
Examplee: Two DoF systems....................................................................................... 25
Rotational Mechanical Systems ....................................................................................... 26
2

Example: Mass-Spring Damper system ....................................................................... 26

Example: inertia ........................................................................................................... 26
Example ....................................................................................................................... 27
Modeling Gears ............................................................................................................... 27
Example: Medeling systems with Gears ....................................................................... 27
Example: Electro-mechanical System: DC motor ......................................................... 28
Modeling of Fluid Systems .................................................................................................. 29
Valve Model..................................................................................................................... 29
Example: Double-Water Tank System ......................................................................... 30
Representation of Multiple Subsystems .............................................................................. 31
Block Diagrams: .............................................................................................................. 31
Signal flow charts: ........................................................................................................... 31
Block Diagram Simplification ........................................................................................... 31
Parallel Form ............................................................................................................ 32
Feedback Form ........................................................................................................ 32
Example Double -Tank System .................................................................................... 33
Example 2 .................................................................................................................... 33
1. .............................................................................................................................. 33
2. .............................................................................................................................. 34
3. .............................................................................................................................. 34
4. .............................................................................................................................. 34
Example: ...................................................................................................................... 35
Example: ...................................................................................................................... 35
Example: ...................................................................................................................... 36
Drawing Signal-Flow Graph for a given Block Diagram: .................................................. 36
Example dot drawing the signal flow graph: ................................................................. 37
Mason’s Rule: ..................................................................................................................... 37
Loop Gain:....................................................................................................................... 37
Forward-path Gain:.......................................................................................................... 37
Non-Touching loops: ....................................................................................................... 37
Non-Touching loop gains: ................................................................................................ 37
Example 1: ................................................................................................................... 38
Mason’s Rule Part 2: ........................................................................................................... 38
𝚫 =𝐃𝐞𝐭𝐞𝐫𝐦𝐢𝐧𝐞𝐧𝐭 𝐨𝐟 𝐭𝐡𝐞 𝐠𝐫𝐚𝐩𝐡 ....................................................................................... 38
𝐏𝐤 = 𝐭𝐡𝐞 𝐤𝐭𝐡 𝐟𝐨𝐫𝐰𝐚𝐫𝐝 𝐩𝐚𝐭𝐡 ............................................................................................ 38
3

𝚫𝐤 = 𝚫|𝐭𝐡𝐞 𝐥𝐨𝐨𝐩𝐬 𝐭𝐨𝐮𝐜𝐡𝐢𝐧𝐠 𝐏𝐤 𝐚𝐫𝐞 𝐬𝐞𝐭 𝐭𝐨 𝐳𝐞𝐫o .............................................................. 38

Example 2: ................................................................................................................... 39
Example 3 .................................................................................................................... 40
Time Response of Systems ................................................................................................ 41
DC Gain .............................................................................................................................. 41
Finding DC gain from the Transfer Function (TF) ................................................................ 42
Remarks: ......................................................................................................................... 42
Example 1: ................................................................................................................... 42
Example 2 .................................................................................................................... 43
Notes! .............................................................................................................................. 43
Why is m>n not possible? ................................................................................................ 43
What inputs do we consider............................................................................................. 44
Impulse input: ........................................................................................................... 44
Step impulse input: ................................................................................................... 44
Ramp input ............................................................................................................... 44
Step Response of a 1st order system .................................................................................. 45
Example: first order system: water tank ....................................................................... 46
Second order system .......................................................................................................... 46
Underdamped systems: ................................................................................................... 47
Measures of performance ................................................................................................ 47
Equations: ................................................................................................................ 47
The effects of 𝜁 on the step response .............................................................................. 48
Example: ...................................................................................................................... 48
Dominant Poles and Order Reduction ................................................................................. 49
Order Reduction.................................................................................................................. 49
Example: ...................................................................................................................... 49
More Examples: ........................................................................................................... 50
Pole and zero next to each other ........................................................................................ 50
System Response with additional Pole ............................................................................... 51
System Response With Non-Minimum Phase Zero (Zero in the Right Side): ...................... 51
Examples of Non-Minimum Phase Systems: ................................................................ 52
Stability of LTI Systems....................................................................................................... 52
Definition of Stability: ....................................................................................................... 52
Bounded Signal: .............................................................................................................. 52
Example of bounded signals: ....................................................................................... 52
4

Example of unbounded signals: ................................................................................... 52

How to check the stability .................................................................................................... 52
Example: ...................................................................................................................... 52
Impulse Response of second order stable and unstable systems: ...................................... 53
Roth Hurwitz Criterion: ........................................................................................................ 53
Characteristic Equation: .................................................................................................. 53
Example 1: Is the following system stable? .................................................................. 54
Example 2: For what values of k the closed loops system is stable? ............................ 54
Example 3: ................................................................................................................... 55
Special Cases for Roth Methods ......................................................................................... 55
Case 1 ............................................................................................................................. 55
Example case 1: .......................................................................................................... 55
Why we consider unity feedback? ....................................................................................... 56
Control Loop with unity feedback ..................................................................................... 56
Steady State Error (ess) ...................................................................................................... 56
Example where the input is a random signal: ............................................................... 56
Example 1: ................................................................................................................... 57
Example 2: ................................................................................................................... 57
Example 3: ess to a ramp input: ................................................................................... 58
Type of a system................................................................................................................. 58
Example: ...................................................................................................................... 58
Change in ess when increasing the order of s in R(s) ...................................................... 58
Step signal and ramp signal: ............................................................................................... 59
Example: calculate ess with a ramp input:.................................................................... 59
Example 2: ................................................................................................................... 59
Ess summary of the unity feedback case ............................................................................ 60
Example: Find the ess for step, ramp and parabolic references ................................... 61
Example: what is the minimum ess for a ramp reference ............................................. 61
The effect of disturbance..................................................................................................... 62
Internal Model Principle: Following reference. ..................................................................... 63
Method A ......................................................................................................................... 63
Example: If R(s) is step and there are 3 poles, then ess=? .......................................... 63
Example: If R(s) is step and there are 2 poles, then ess=? .......................................... 63
Example: If R(s) is ramp and there are 3 poles, then ess=? ......................................... 63
Example: If R(s) is ramp and there are 4 poles, then ess=? ......................................... 64
Method B ......................................................................................................................... 64
5

Example: If R(s) is ramp and there are 4 poles, then ess=? ......................................... 64
Control Obejctives............................................................................................................... 64
How to determine the control performance.......................................................................... 65
On-off controller .................................................................................................................. 65
On-off controller with hysteresis....................................................................................... 66
Why does an on-off controller have oscillations? ............................................................. 66
Steady State error (ess) when using a proportional controller ............................................. 66
A numerical Example: .................................................................................................. 67
System with and without internal integrator ......................................................................... 67
Example: system with integrator .................................................................................. 68
Theoretical Reason ......................................................................................................... 68
Effect of kp on the transient response (Case 1: systems with integrator) ......................... 68
Effect of kp on the transient response (Case 1: systems without integrator) .................... 69
P controller with a bias term ................................................................................................ 69
Proportional – Integral (PI) controller ................................................................................... 70
Examples of different PI controller representations ...................................................... 70
Implementation of PI controller ........................................................................................ 71
The effect of changing the integral coefficient on the response ....................................... 71
The derivative term .......................................................................................................... 72
Adding the derivative term (D-term in PID) .......................................................................... 72
Important note about the D-Term..................................................................................... 73
Summary: PID; PI; PD ........................................................................................................ 74
2 Approaches for designing a PID Controller ................................................................... 74
Zero pole Placement ....................................................................................................... 74
Example: ...................................................................................................................... 75
Example: zero pole placement ..................................................................................... 75
Example: zero pole placement ..................................................................................... 75
Example: zero pole placement ..................................................................................... 76
Pre-tune fine-tune approach ............................................................................................ 76
Zigler Nichols (Z-N) first method: .............................................................................. 76
Step system response does not have pure delay: ...........................................................................77
Zigler Nichols (Z-N) second method: ........................................................................ 77
Example: 2nd approach Z-N Method ............................................................................. 78
Root Locus (drawing the location of the closed loop) .......................................................... 79
Example: sketching the location of a closed loop pole ................................................. 79
Example: ...................................................................................................................... 80
6

Example: Generic Root Locus example ....................................................................... 80

Recap: Polar coordinates ................................................................................................ 81
Question: how to find ∠𝐺𝑠𝐻𝑠 ........................................................................................... 81
Example: can you find all the possible points that satisfy the angle criterion? .............. 82
Rules for sketching Root Locus Rules 1, 2, 3: ................................................................. 82
Examples: .................................................................................................................... 82
Rules for sketching Root Locus Rules 4 and 5: ............................................................... 82
Example: ...................................................................................................................... 83
Example: ...................................................................................................................... 83
Example: ...................................................................................................................... 83
Rules for sketching Root Locus Rule 6: ........................................................................... 84
Example: ...................................................................................................................... 84
Rules for sketching Root Locus Rule 7: ........................................................................... 84
Example: ...................................................................................................................... 85
How to find the fain and the jw-crossing axis points (with MATLAB): ............................... 85
Usage of magnitude criterion ........................................................................................... 85
Example: ...................................................................................................................... 86
Rules for sketching Root Locus Rule 888 (if n>m+2): ...................................................... 87
Example: ...................................................................................................................... 87
Design a controller using Root Locus .............................................................................. 88
Example: Finding the ROI for given design objectives:................................................. 88
Example: ...................................................................................................................... 88
Controller Design Case 1: When the root locus is already passing through ROI .............. 89
Example: ...................................................................................................................... 89
Question: what the value of is 𝑘∆?................................................................................ 89
Example: ...................................................................................................................... 89
What is 𝑘∆?.................................................................................................................. 89
Controller Design Case 2: When the root locus needs to be changed ............................. 90
Example: ...................................................................................................................... 90
Example: If the controller needs to have integrator: ..................................................... 92
Reminder: trigonometry....................................................................................................... 93
Frequency Response .......................................................................................................... 94
Example ....................................................................................................................... 94
2 methods for drawing the frequency response: .............................................................. 94
Example: ...................................................................................................................... 95
Understanding the frequency response (main property) .................................................. 95
7

Example: ...................................................................................................................... 96
New approach to find the transfer function of a system. .................................................. 96
Assymptotic approximation of Bode Diagram .................................................................. 96
Assymptotic Bode Diagram for Different terms ................................................................ 97
𝑮𝒔 = 𝒔 ...................................................................................................................... 97
𝑮𝒔 = 𝟏𝐬 .................................................................................................................... 97
𝑮𝒔 = 𝒔𝒏 .................................................................................................................... 98
𝑮𝒔 = 𝟏𝒔𝒏 .................................................................................................................. 98
𝑮𝒔 = 𝒌; 𝒌 > 𝟎 ....................................................................................................... 98
𝑮𝒔 = 𝒌; 𝒌 < 𝟎 ........................................................................................................ 99
𝑮𝒔 = 𝟏 + 𝒔𝒂 .............................................................................................................. 99
8
9
10
11
12

Definitions
System or Plant
• Any device or interconnection of elements for a desired purpose
Output:
- A physical quantity that should be controlled within a
range or at a level
Input:
- The variables that influence the output
Set point:
- The desired value for the output
- value we try to match; constant!
Goal of Control:
- To apply a proper input such that the output follows the set point

Example: Water Tank System

The goal is to control the height of water in the tank

Disturbance:
13

Concept of control loop

Controller:
- computer program (logic
Actuator:
- gets the order from the controller and changes the input

Example: Airplane Actuator

Rotation around… is called:

→ x → roll
→ y → pitch
→ z → yaw

Example: missile and helicpters:

Block diagram looks like:

we call B reference because in this case it isn’t constant

14

Example: inverted Pendulum

Properties of a system
Statical or dynamical system:
Statical: if the output at time t depends on input at time t
𝑦(𝑡) = 3 ∗ 𝑢(𝑡)  statical system
𝑑𝑢(𝑡) 𝑢(𝑡)−𝑢(𝑡−𝛿)
𝑦(𝑡) = 𝑑𝑡
= 𝛿
 dynamical system

Conclusion: any differential equation represents a dynamical system.

Example: Modeling a DC motor

𝑑𝑦(𝑡)
+ 𝛼 ∗ 𝑦(𝑡) = 𝛽 ∗ 𝑢(𝑡)
𝑑𝑡

Linearity:

A system is called linear if it satisfies the

superposition properties:
𝑢 (𝑡) → 𝑦1 (𝑡)
𝑖𝑓 { 1
𝑢2 (𝑡) → 𝑦2 (𝑡)

→ 𝛼 𝑢1 (𝑡) + 𝛽 𝑢2 (𝑡) → 𝛼 𝑦1 (𝑡) + 𝛽 𝑦2 (𝑡)

Example: Non-linear system

𝑦(𝑡) = 𝑢2 (𝑡)
𝑢 (𝑡) → 𝑦1 = 𝑢12
{ 1
𝑢2 (𝑡) → 𝑦2 = 𝑢22
→ 𝑢1 + 𝑢2 → (𝑢1 + 𝑢2 )2

• 𝑦̇ (𝑡) + 2𝑦(𝑡) = 𝑢(𝑡) →

𝑙𝑖𝑛𝑒𝑎𝑟, 𝑤𝑒 𝑐𝑎𝑛 𝑝𝑟𝑜𝑣𝑒 𝑡ℎ𝑎𝑡 𝑎 𝑙𝑖𝑛𝑒𝑎𝑟 𝐷. 𝐸. 𝑟𝑒𝑝𝑟𝑒𝑠𝑒𝑛𝑡𝑠 𝑎 𝑙𝑖𝑛𝑒𝑎𝑟 𝑠𝑦𝑠𝑡𝑒𝑚
• 𝑦̈ (𝑡) + 3𝑦(𝑡) = 𝑢(𝑡) + 𝑦(𝑡) → 𝑛𝑜𝑡 𝑙𝑖𝑛𝑒𝑎𝑟
15

Time Variant
𝑢(𝑡) → 𝑦(𝑡) → 𝑢(𝑡 − 𝜏) → 𝑦(𝑡 − 𝜏)
For a D.E. we should not have the time explicitly in the equation
𝑦̇ (𝑡) + 𝜏𝑦(𝑡) = 𝑢(𝑡) < −𝑛𝑜𝑡 𝑡𝑖𝑚𝑒 𝑖𝑛𝑣𝑎𝑟𝑖𝑎𝑛𝑡!
• In this course we only work with linear time invariant (LTI) systems
• You can show any LTI systems using a linear differential equation.

Representing systems in the Laplace domain

𝐺(𝑠) = 𝑦̇ + 2𝑦 = 3𝑢

3
𝑦(0) = 0 → 𝑦(𝑠) = 𝑢(𝑠)
𝑠+2

𝑌(𝑠) 3
𝐻(𝑠) = =
𝑈(𝑠) 𝑠 + 2

For an LTI system, the output to any arbitrary input can be calculated as follows:
𝑦(𝑡) = 𝑢(𝑡) ∗ ℎ(𝑡)
H(t) is also known as impulse response!

Solving a D.E.

𝐺(𝑠) = 𝑦̇ + 2𝑦 = 3𝑢

𝑦̇ (𝑡) + 2𝑦(𝑡) = 3𝑢(𝑡)

{
𝑦(0) = 𝑦0
In the Laplace Domain:
𝑠𝑌(𝑠) − 𝑦𝑜 + 𝑠𝑌(𝑠) = 2𝑈(𝑠)
(𝑠 + 2) 𝑌(𝑠) = 3 𝑈(𝑠) + 𝑦0
3 𝑦0
𝑌(𝑠) = 𝑈(𝑠) +
𝑠+2 𝑠+2
3 5 𝑦0
𝑌(𝑠) = ∗ +
𝑠+2 𝑠 𝑠+2

𝑦(𝑜) = 4
3 5 4
𝑌(𝑠) = ∗ +
𝑠+2 𝑠 𝑠+2
Giving:
𝑦(𝑡) = 7 ∗ 5 (1 − 𝑒 −2𝑡 ) + 4𝑒 −2𝑡 ; 𝑡>0
16

Properties of the Laplace transform

𝑓(𝑡) → 𝑓(𝑠)

• lim ┬(t → ∞) 𝑓(𝑡) = lim ┬(𝑠 → 0) 〖𝑠 𝐹(𝑠)〗 if 𝑠𝐹(𝑠) doesn’t have any pole in the
right side

• lim ┬(t → 0) 𝑓(𝑡) = lim ┬(𝑠 → ∞) 〖𝑠 𝐹(𝑠)〗

Modeling Dynamical Systems

Why modeling a system?
• Study the behavior of the system
• Prediction of the outputs
• Design of controller

 single-input-single-output (SISO) system example

Modeling Electrical systems

17

𝑽𝒐 (𝒔)
=?
𝑽𝒊 (𝒔)

𝑑𝑖(𝑡)
−𝑣𝑖 ( 𝑡) + 𝑅𝑖(𝑡) + 𝐿 =0
{ 𝑑𝑡
𝑑𝑖(𝑡)
𝑉𝑜 (𝑡) = 𝐿
𝑑𝑡
Giving
𝑣𝑖 (𝑠) = (𝑅 + 𝐿𝑠) ∗ 𝐼(𝑠)
{
𝑉𝑜 (𝑠) = 𝐿𝑠 ∗ 𝐼(𝑠)
Thus:
𝑉𝑜 (𝑠) 𝐿𝑠
=
𝑉𝑖 (𝑠) 𝐿𝑠 + 𝑅

Note: the order of the circuit is the highest s in the denominator!

18

Example: RL Circuit

Time Domain

Laplace (s)- Domain

Example: Series RLC Circuit

19

Example: Electrical circuit 2nd order

1. Start by analyzing the capacitor.

𝑑𝑉𝑐 (𝑡)
𝑖𝑐 (𝑡) = 𝑐 ∗ 𝑑𝑡

𝐼𝑐 (𝑠) = 𝐶 ∗ 𝑠 ∗ 𝑑𝑉𝑐 (𝑠) < −𝐿𝑎𝑝𝑙𝑎𝑐𝑒 𝑑𝑜𝑚𝑎𝑖𝑛

𝑰𝒄 (𝒔)
𝑽𝒄 (𝒔) =
𝑪𝒔

2. Analyze the inductor.

𝑑𝑖𝐿
𝑉𝐿 = 𝐿 ∗
𝑑𝑡
𝑽𝑳 (𝒔) = 𝑳𝒔 ∗ 𝑰𝑳 (𝒔)

3. Analyze the resistor.

𝑉(𝑡) = 𝑅 ∗ 𝐼(𝑡)
𝑽(𝒔) = 𝑹 ∗ 𝑰(𝒔)

4. Conclusion
𝐼(𝑠)
−𝑣𝐼 (𝑠) = 𝐿𝑠 ∗ 𝐼(𝑠) + 𝑅𝐼(𝑠) + =0
𝐶𝑠
1 1
{(𝑉𝐼 (𝑠) = (𝐿𝑠 + 𝑅 + @𝑉0 (𝑠) = ∗ 𝐼(𝑠))
𝐶𝑠 𝐶𝑠

𝑉0 (𝑠) 1 𝑉0 (𝑠) 1
= (=) =
𝑉𝐼 (𝑠) 𝐿𝑠 + 𝑅 + 1 𝑉𝐼 (𝑠) 𝐿𝐶𝑠 2 + 𝑅𝐶𝑠 + 1
𝐶𝑠
20

Example: electrical circuit 2nd order

From the equations in the last example:

−𝑣𝑖 + 𝐿𝑠 ∗ 𝐼1 + 𝑅1 (𝐼1 − 𝐼2 ) = 0
1
−𝑅1 (𝐼1 − 𝐼2 ) + 𝑅2 𝐼2 + 𝐼 =0
𝐶𝑠 2
1
𝑣0 (𝑠) = 𝐼
{ 𝐶𝑠 2
Thus:
1
1 𝑅1 + 𝑅2 + 𝐶𝑠
(𝑅1 + 𝑅2 + ) 𝐼2 = 𝑅1 𝐼1 (=) 𝐼1 = ∗ 𝐼2
𝐶𝑠 𝑅1
21

Modeling Mechanical System

Spring:

𝑓(𝑡) = 𝑘 𝑥(𝑡)
𝑓(𝑡) = 𝑘 (𝑥1 − 𝑥2 )

Damper:
𝑑𝑥 𝑑𝑥2
𝑓 = 𝑏 ( 𝑑𝑡1 − 𝑑𝑡
)

Mass:

𝑑2 𝑥
𝑓 = 𝑚𝛼 = 𝑚 ∗
𝑑𝑡 2
22
23

Example: Mechanical System

∑ 𝐹 = 𝑚 ∗ 𝑥̈

𝐹(𝑡) − 𝑓1 (𝑡) − 𝑓2 (𝑡) = 𝑚𝑥̈

𝑓(𝑡) − 𝑘𝑥 − 𝑏𝑥̇ = 𝑚𝑥̈

𝑓(𝑡) − 𝑘(𝑥 − 0) − 𝑏(𝑥̇ − 0) = 𝑚𝑥̈

𝑓(𝑡) = 𝑚𝑥̈ + 𝑏𝑥̇ + 𝑘𝑥

→ 𝐿𝑎𝑝𝑙𝑎𝑐𝑒 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚 → 𝐹(𝑠) = 𝑚𝑠 2 𝑋(𝑠) + 𝑏𝑠𝑋(𝑠) + 𝑘𝑋(𝑠)

𝐹(𝑠) = (𝑚𝑠 2 + 𝑏𝑠 + 𝑘 ) ∗ 𝑋(𝑠)

Giving then:

𝑋(𝑠) 1
= 2
𝐹(𝑠) 𝑚𝑠 + 𝑏𝑠 + 𝑘

Example: Mass-Spring Damper system

24

Example (this is probably wrong!)

b3

b1 b2

∑ 𝐹 = 𝑚1 ∗ 𝑥1̈ → 𝑓1 − 𝑘1 (𝑥1 − 0) − 𝑘2 (𝑥1 − 𝑥2 ) − 𝑏1 (𝑥̇ − 0) − 𝑏3 (𝑥1̇ − 𝑥2̇ ) = 𝑚1 𝑥1̈

∑ 𝐹 = 𝑚2 ∗ 𝑥2̈ → −𝑘2 (𝑥2 − 𝑥1 ) − 𝑏3 (𝑥2̇ − 0) − 𝑏2 (𝑥̇ 2 − 0) − 𝑘3 (𝑥2 − 0) = 𝑚2 𝑥2̈

𝐹1 + (𝑘2 + 𝑏3 𝑠)𝑋(𝑠) = (𝑘1 + 𝑘2 + 𝑏1 𝑠 + 𝑏3 𝑠 + 𝑚1 𝑠 2 )𝑋1 (𝑠)

1 𝑘2 + 𝑏3 𝑠
𝑋1 (𝑠) = 𝐹1 (𝑠) + 𝑋2 (𝑠)
𝑚(𝑠) 𝑚(𝑠)

Giving
𝑋1 (𝑠) = 𝐺1 (𝑠)𝐹1 (𝑠) + 𝐺2 (𝑠)𝑋2 (𝑠)
Thus:
〖(𝑘〗2 + 𝑏3 𝑠)𝑋1 (𝑠) = 〖(𝑘〗2 + 𝑘3 + 𝑏2 𝑠 + 𝑏3 𝑠 + 𝑚𝑠 2 )𝑋2 (𝑠)
Concluding:
𝑘2 + 𝑏3 𝑠
𝑋2 = ∗ 𝑋1 (𝑠)
𝑚2 𝑠 2 + 𝑏2 𝑠 + 𝑏3 𝑠 + 𝑘2 + 𝑘3

𝑘2 +𝑏3 𝑠
𝑚2 𝑠2 +𝑏2 𝑠+𝑏3 𝑠+𝑘2 +𝑘3
 we call this G_3(s)

This gives the following block diagram

25

Examplee: Two DoF systems

26

Rotational Mechanical Systems

Example: Mass-Spring Damper system

Example: inertia

∑ 𝑇𝑖 = 𝐽 𝜃̈

𝑇 − 𝑘(𝜃 − 0) − 𝐷(𝜃̇ − 0) = 𝐽 𝜃̈

𝑇 = 𝐽 𝜃̈ + 𝐷𝜃̇ + 𝑘𝜃

𝑇 = 𝐽𝑤̇ + 𝐷𝑤 + 𝑘 ∫ 𝑤

𝑇(𝑠) = (𝐽𝑠 2 + 𝐷𝑠 + 𝑘) ∗ 𝜃(𝑠)

27

Example

(ignore the numbers)

𝑇 − 𝑘2 (𝜃2 − 0) − 𝑘1 (𝜃2 − 𝜃1 ) − 𝐷2 (𝜃2̇ − 𝜃1̇ ) = 𝐽2 𝜃2̈

{
−𝐷 (𝜃 ̇ − 0) − 𝑘 (𝜃 − 𝜃 ) − 𝐷 𝜃 ̇ − (𝜃̇ ) = 𝐽 𝜃 ̈
1 1 1 1 2 2 1 2 1 2

Modeling Gears

Example: Medeling systems with Gears

28

Example: Electro-mechanical System: DC motor

Keywords: EMF; Torque-current relation; Mechanical Loading relation; electromotive force

If we apply a load to the

motor, we need to
model it, in this case
𝑇𝑙𝑜𝑎𝑑 = 0
29

Modeling of Fluid Systems

𝑑𝑣(𝑡)
{𝑞 𝑖 − 𝑞𝑜 =
𝑑𝑡
𝑣(𝑡) = 𝐴 ∗ ℎ(𝑡)

This gives:

𝑑ℎ(𝑡)
𝑞𝑖 (𝑡) − 𝑞𝑜 (𝑡) = 𝐴 ∗
𝑑𝑡

(see Valve Model subsection)

Combining the equations, we get:
̇
𝑞𝑖 − 𝑞𝑜 = 𝐴 ∗ ℎ(𝑡)
{ ℎ(𝑡)
𝑞𝑜 (𝑡) =
𝑅
ℎ(𝑡)
𝑞𝑖 (𝑡) = 𝐴ℎ̇(𝑡) +
𝑅
This gives:
𝐻(𝑠) 𝑅
=
𝑄𝑖 (𝑠) (𝐴𝑅 𝑠 + 1)

• The water tank can save energy (potential)

• The number of components that can save energy in the system is the same as the
order of the system.

Valve Model
(works like a resistance)

∝ 𝑖𝑠 𝑡ℎ𝑒 𝑠𝑖𝑔𝑛 𝒇𝒐𝒓 𝒑𝒓𝒐𝒑𝒐𝒓𝒕𝒊𝒐𝒏𝒂𝒍

Linear approximation model:

𝑃1 − 𝑃2
𝑞≈
𝑅̅
30

Example: Double-Water Tank System

Left tank:

𝑞𝑖 − 𝑞𝑜1 − 𝑞1 = 𝐴1 ∗ ℎ1̇
ℎ1 − 0
𝑞𝑜1 =
𝑅01
ℎ1 − ℎ2
𝑞1 =
{ 𝑅1

Right Tank:

𝑞1 − 𝑞𝑜2 = 𝐴2 ∗ ℎ2̇
ℎ1 − ℎ2
𝑞1 =
𝑅1
ℎ2 − 0
𝑞02 =
{ 𝑅02

Combining them, we get:

31

Representation of Multiple Subsystems

Block Diagrams:
o Represents systems with blocks, signals, summing junctions, and pickoff
points.

o
Signal flow charts:
o Represent transfer functions as lines and signals as small circular nodes.
o Summation is Implicit.

Block Diagram Simplification

𝑦 (𝑠) = 𝐺1 (𝑠)𝑅(𝑠)
{ 1
𝑦2 (𝑠) = 𝐺2 (𝑠)𝑌1 (𝑠)
𝑦2 (𝑠) = 𝐺1 (𝑠)𝐺2 (𝑠)𝑅(𝑠)
𝑮(𝒔) = 𝑮𝟏 (𝒔)𝑮𝟐 (𝒔)
32

Rules:
Parallel Form

𝑦1 = 𝐺1 ∗ 𝑅
{ 𝑦2 = 𝐺2 ∗ 𝑅
𝑦 = 𝑦1 + 𝑦2
This gives:

𝑦 = (𝐺1 + 𝐺2 ) ∗ 𝑅
Thus:
𝐺(𝑠) = 𝐺1 (𝑠) + 𝐺2 (𝑠)

Feedback Form

𝐸(𝑠) = 𝑅(𝑠) − 𝐻(𝑠)𝑌(𝑠)

{
𝑌(𝑠) = 𝐺(𝑠)𝐸(𝑠)

𝐺(𝑠)𝑅(𝑠) − 𝐺(𝑠)𝐻(𝑠)𝑌(𝑠)

𝑌(𝑠) (1 + 𝐺(𝑠)𝐻(𝑠) = 𝐺(𝑠)𝑅(𝑠)

𝑌(𝑠) 𝐺(𝑠)
=
𝑅(𝑠) 1 + 𝐺(𝑠)𝐻(𝑠)

𝐺(𝑠)
If we choose to call 1+𝐺(𝑠)𝐻(𝑠) T(s) to make it simpler:
𝐺(𝑠)
𝑇(𝑠) =
1 + 𝐺(𝑠)(−𝐻(𝑠))
33

Example Double -Tank System

𝐻2 (𝑠) 𝐺 3
= 𝐺1 (𝑠)𝑇1 (𝑠) = 𝐺1 ∗
𝑄𝑖 (𝑠) 1 − 𝐺3 𝐺2

𝐻1 (𝑠) 1 𝐺1
= 𝐺1 ∗ 𝑇2 = 𝐺1 ∗ =
𝑄𝑖 (𝑠) 1 − 𝐺2 𝐺3 1 − 𝐺2 𝐺3

Example 2

𝑌 𝐺2 𝐺3
= 𝐺1 𝑇1 = 𝐺1 ∗
𝑅 1 + 𝐺2 𝐺3 𝐻1 𝐻2

Note:
𝐹𝑜𝑟𝑤𝑎𝑟𝑑
1 + 𝐹𝑟𝑜𝑤𝑎𝑟𝑑 ∗ 𝐹𝑒𝑒𝑑𝑏𝑎𝑐𝑘

Other rules
1.
34

2.

3.

4.
35

Example:

𝑌 𝐺2 𝐺3
= 𝐺1 ∗
𝑅 1 + 𝐺2 𝐺3 (𝐻1 + 𝐻2 )

Example:

𝐺3
𝑇1 =
1 − 𝐺3 𝐻1
𝐺2 𝑇1
𝑇2 =
1 + 𝐺2 𝑇1 𝐻2

𝑌
= 𝐺1 ∗ 𝑇2
𝑅
36

Example:

𝐺3
𝑇1 =
1 + 𝐺3 𝐻3
1 𝐺2 + 1
𝑇2 = 1 + =
𝐺2 𝐺2
𝐺2
𝑇3 =
1 + 𝐺2 𝐻2
𝐺1 𝑇3
𝑇4 =
1 + 𝐺1 𝑇3 𝐻1
Thus:
𝑌
= 𝑇4 ∗ 𝑇2 ∗ 𝑇1
𝑅

Drawing Signal-Flow Graph for a given Block Diagram:

• Convert “pick‐off points” and “summation junctions” to “nodes”
• Convert the blocks to the branches.
• The negative at the summation junction should be transformed to the branches.
• The value of each node is the summation of all inputs.
• The value of all outputs from a node is the same as the value of the node
37

Example dot drawing the signal flow graph:

Mason’s Rule:
Loop Gain:
The product of all branches in a loop (starting from a node and ending at the same node not
passing a node twice).

Forward-path Gain:
The product of all gains in a path from input to the output.

Non-Touching loops:
The loops that don’t have any node in common.

Non-Touching loop gains:

The product of the loop gains from non‐touching loops, taken 2, 3, … loops at a time.
38

Example 1:

• Loop Gain:
o 𝐿1 = −1 ∗ 𝐺2 ∗ 𝐻2 ; 𝐿2 = −𝐺1 𝐺2 𝐻1 ; 𝐿3 = −𝐺3 𝐻3
• Forward-path gain:
o 𝑃1 = 1 ∗ 𝐺1 ∗ 1 ∗ 𝐺2 ∗ 1 ∗ 𝐺3 ∗ 1 ; 𝑃2 = 1 ∗ 𝐺1 ∗ 1 ∗ 1 ∗ +𝐺3 ∗ 1
• Non-Touching loops
• Non-Touching loop gains:
o 𝐿1 𝐿3 ; 𝐿2 𝐿3

Mason’s Rule Part 2:

𝑌(𝑠) ∑ 𝑃𝑘 ∗ ∆𝑘
𝐺(𝑠) = =
𝑅(𝑠) ∆

𝚫 =𝐃𝐞𝐭𝐞𝐫𝐦𝐢𝐧𝐞𝐧𝐭 𝐨𝐟 𝐭𝐡𝐞 𝐠𝐫𝐚𝐩𝐡

∆ = 1 − ∑ 𝐿𝑖 + ∑ 𝐿𝑖𝐿𝑗 − ∑ 𝐿𝑖𝐿𝑗𝐿𝑘
From previous example:
∆= 1 − (𝐿1 + 𝐿2 + 𝐿3 ) + (𝐿1 𝐿3 + 𝐿2 𝐿3 )

𝐏𝐤 = 𝐭𝐡𝐞 𝐤𝐭𝐡 𝐟𝐨𝐫𝐰𝐚𝐫𝐝 𝐩𝐚𝐭𝐡

From previous example:
𝑃1 = 𝐺1 𝐺2 𝐺3
𝑃2 = 𝐺1 𝐺3

𝚫𝐤 = 𝚫|𝐭𝐡𝐞 𝐥𝐨𝐨𝐩𝐬 𝐭𝐨𝐮𝐜𝐡𝐢𝐧𝐠 𝐏𝐤 𝐚𝐫𝐞 𝐬𝐞𝐭 𝐭𝐨 𝐳𝐞𝐫o

From previous example:
∆1 = ∆|𝐿1 =𝐿2 =𝐿3 =0 = 1
∆2 = ∆|𝐿1 =𝐿2 =𝐿3 =0 = 1

Concluding:
𝑃1 ∆1 + 𝑃2 ∆2
𝐺(𝑠) =
∆
39

Example 2:

1. Finding the loops:

𝐿1 = 𝐺2 𝐻1
𝐿2 = 𝐺4 𝐻2
{ 𝐿3 = 𝐺7 𝐻3
𝐿4 = 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6 𝐺7 𝐺8

2. Find the Non-Touching Loops

𝐿1 𝐿2 ; 𝐿3 𝐿2 ; 𝐿1 𝐿3 ; 𝐿1 𝐿2 𝐿3

3. Find forward-path

𝑃1 = 𝐺1 𝐺2 𝐺3 𝐺4 𝐺5

4. Find delta
∆= 1 − (𝐿1 + 𝐿2 + 𝐿3 + 𝐿4 ) + (𝐿1 𝐿2 + 𝐿1 𝐿3 + 𝐿3 𝐿2 ) − (𝐿1 𝐿2 𝐿3 )

5. Find Delta i
∆1 = ∆|𝐿1=𝐿2=𝐿4=0 = 1 − 𝐿3
(We equal to zero all the loops that touch the main path)

6. Find the Transfer Function

𝑌 (𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 ) ∗ (1 − 𝐿3 )
=
𝑅 1 − (𝐿1 + 𝐿2 + 𝐿3 + 𝐿4 ) + (𝐿1 𝐿2 + 𝐿1 𝐿3 + 𝐿3 𝐿2 ) − (𝐿1 𝐿2 𝐿3 )
40

Example 3

1. Find loops:

𝐿1 = 𝐺2 𝐻2
𝐿 = 𝐺3 𝐻3
{ 2
𝐿3 = 𝐺6 𝐻6
𝐿4 = 𝐺7 𝐻7

2. Non-Touching Loops
𝐿1 𝐿3
𝐿 𝐿
{ 1 4
𝐿2 𝐿3
𝐿2 𝐿4
3. Find forward path.
𝑃 = 𝐺1 ∗ 𝐺2 ∗ 𝐺3 ∗ 𝐺4
{ 1
𝑃2 = 𝐺5 ∗ 𝐺6 ∗ 𝐺7 ∗ 𝐺8

4. Find delta (gain)

∆ = 1 − (𝐿1 + 𝐿2 + 𝐿3 + 𝐿4 ) + (𝐿1 𝐿3 + 𝐿1 𝐿4 + 𝐿2 𝐿3 + 𝐿2 𝐿4 )

5. Find non-touching loop gains.

∆1 |𝐿1=𝐿2=0 = 1 − 𝐿3 − 𝐿4
∆2 |𝐿3=𝐿4=0 = 1 − 𝐿1 − 𝐿2

Note here we must separate into two because we have 2 different paths.

6. Find Transfer function.

𝑌 𝑃1 ∆1 + 𝑃2 ∆2 (𝐺1 ∗ 𝐺2 ∗ 𝐺3 ∗ 𝐺4 ∗ (1 − 𝐿3 − 𝐿4 )) + (𝐺5 ∗ 𝐺6 ∗ 𝐺7 ∗ 𝐺8 ∗ (1 − 𝐿1 − 𝐿2 ))
= =
𝑅 ∆ 1 − (𝐿1 + 𝐿2 + 𝐿3 + 𝐿4 ) + (𝐿1 𝐿3 + 𝐿1 𝐿4 + 𝐿2 𝐿3 + 𝐿2 𝐿4 )
41

Time Response of Systems

Some Definitions:
For the given transfer function

• Order of system: the maximum power of 𝑠 in the denominator (order=n for 𝐺 𝑠 )

• Poles: the roots of denominator (𝐺 𝑠 has 𝑛 poles)
• Zeros: the roots of numerator (𝐺 𝑠 has 𝑚 zeros)
• Gain (or high frequency gain):

𝒕𝒉𝒆 𝒄𝒐𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒕 𝒐𝒇 𝒉𝒊𝒈𝒉𝒔𝒕 𝒑𝒐𝒘𝒆𝒓 𝒐𝒇 𝒔 𝒊𝒏 𝒕𝒉𝒆 𝑵𝒖𝒎𝒆𝒓𝒂𝒕𝒐𝒓 𝒃𝒎

𝒌 = 𝒌𝒉 𝒇 𝒈 = = = 𝒃𝒎
𝑻𝒉𝒆 𝒄𝒐𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒕 𝒐𝒇 𝒉𝒊𝒈𝒉𝒆𝒔𝒕 𝒑𝒐𝒘𝒆𝒓 𝒐𝒇 𝒔 𝒊𝒏 𝒕𝒉𝒆 𝒅𝒆𝒏𝒐𝒎𝒊𝒏𝒂𝒕𝒐𝒓 𝟏

DC Gain

𝑦𝑠𝑠 𝑆𝑡𝑒𝑎𝑑𝑦 𝑆𝑡𝑎𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑦(𝑡)

𝑘𝑑𝑐 = =
𝑢𝑠𝑠 𝑆𝑡𝑒𝑎𝑑𝑦 𝑆𝑡𝑎𝑡𝑒 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑢(𝑡)

Note: these systems do not have DC gain! ->

42

Finding DC gain from the Transfer Function (TF)

Remarks:
• DC gain is only defined for stable systems.
• A system is stable if the step response has a steady state value.
• We can show that for unstable systems the step response doesn’t converge to a
constant value.
To find 𝒌𝒅𝒄 , assuming that the input is step we have:
1
𝑘𝑑𝑐 = 𝑦𝑠𝑠 = lim 𝑦(𝑡) = lim 𝑠 ∗ 𝑦(𝑠) = lim 𝑠 ∗ 𝑈(𝑠) ∗ 𝐺(𝑠) = lim 𝑠 ∗ ∗ 𝐺(𝑠) (=) 𝑘𝑑𝑐 = 𝐺(𝑠)
𝑡→∞ 𝑠→0 𝑠→0 𝑠→0 𝑠

1
Note: the input of a step function is 𝑈(𝑠) = 𝑠

Example 1:
6𝑠 + 9
𝐺(𝑠) =
3𝑠 2 + 12𝑠 + 9

2𝑠 + 3 2(𝑠 + (1 ∗ 5))
= =
𝑠2 + 4𝑠 + 3 (𝑠 + 3)(𝑠 + 1)

• Order:
o 2
• Poles: roots to the denominator
o P1 = -3
o P2 = -1
• Zeros: roots to nominator
o Z1 = -6
• K:
o 2
• Kdc:
9
o = 1
9
43

Example 2
2𝑠 + 2
𝐺(𝑠) =
𝑠 3 + 4𝑠 2 + 6𝑠 + 4

• Order:
o 3
• Poles: roots to the denominator
o P1 = -2
o P2 = -1 +/- j
• Zeros: roots to nominator
o Z1 = -1
• K:
o 2
• Kdc:
2
o 4
= 0.5

Notes!

• Any 𝑻𝑭 can be written as summation of 1st and 2nd order systems.

• In this course we use “n” as the number of poles, and “m” as the number of zeros.
• If 𝒎<𝒏 the system is called “strictly proper”.
• If m=𝒏 the system is called “proper”.
• 𝒎>𝒏 is not possible in real systems.

Why is m>n not possible?

7
=𝑠+
𝑠+3

𝑑𝑢(𝑡)
𝑢(𝑡) → 𝑈(𝑠) 𝑠𝑜: → 𝑆(𝑠)
𝑑𝑡
𝑑𝑢(𝑡) 𝑢(𝑡 + ∆𝑡) − 𝑢(𝑡)
=
𝑑𝑡 ∆𝑡
There is no ideal differentiator in this case
44

What inputs do we consider

Impulse input:

𝑈(𝑠) = 1

Step impulse input:

1
𝑈(𝑠) =
𝑠
The step response is the integral of the impulse response

Ramp input

1
𝑈(𝑠) = 𝑠2
The ramp response is the integral of the step response
45

Step Response of a 1st order system

𝑦(𝑡) = ?
𝑘 −𝑘
𝑘 1 𝑘 𝑘 1
𝑦(𝑠) = ∗ = = ∗ = + 𝑎
𝑎
𝑠 + 𝑎 𝑠 𝑠(𝑠 + 𝑎) 𝑠 𝑠 + 𝑎 𝑠 𝑠 + 𝑎
Thus giving:
𝑘 𝑘 −𝑎𝑡
𝑦(𝑡) = − 𝑒 ; 𝑡≫0
𝑎 𝑎
𝑘
𝑦(𝑡) = (1 − 𝑒 −𝑎𝑡 ); 𝑡≫𝑜
𝑎

Sometimes we prefer to write the 1st order system in a dc gain format:

𝑘
𝑘 𝑎 𝑘𝑑𝑐
𝐺(𝑠) = = =
𝑠 + 𝑎 1 𝑠 + 1 𝜏𝑠 + 1
𝑎
1
Being 𝜏 the time constant and 𝜏 = 𝑎
Thus, this gives:
𝑘 𝑡
𝑦(𝑡) = (1 − 𝑒 −𝜏 )
𝑠+𝑎
46

Example: first order system: water tank

𝐻(𝑠) 𝑅
=
𝑄𝑖 (𝑠) 𝑅𝐴𝑠 + 1
Being 𝑅 = 𝑘𝑑𝑐 ; and 𝑅𝐴 = 𝜏 = 𝑡𝑖𝑚𝑒 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡

Second order system

Note:
𝜁 = 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑟𝑎𝑡𝑖𝑜
𝜔𝑛 = 𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦
𝜔𝑑 = 𝑑𝑎𝑚𝑝𝑖𝑛𝑔 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦
47

Underdamped systems:

1
𝑦(𝑡) = 1 − ∗ 𝑒 −𝜎𝑡 cos(𝜔𝑑 𝑡 − 𝜃)
√1 − 𝜁2

Measures of performance

• 𝑀𝑝𝑡 = maximum value

• 𝑇𝑝 = Peak time
• 𝑇𝑟 = Rise time (from 10% to 90% of the
final grade)
• 𝑇𝑠 = settling time ( time that it takes to
get 2% of the final value and stay there)

Equations:
𝜋
𝑇𝑝 =
𝜔𝑑

4 4
𝑇𝑠 = =
𝜁 ∗ 𝜔𝑛 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑐𝑙𝑜𝑠𝑒𝑠𝑡 𝑝𝑜𝑙𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑗𝑤 𝑎𝑥𝑖𝑠

−𝜁∗𝜋
2
𝑃. 𝑂. = 100 ∗ 𝑒 √1−𝜁
48

The effects of 𝜁 on the step response

Example:
1. The following figures show the block diagram and the step response of a system:

1
𝑠 1
𝑇1 = =
1
1+ ∗𝑎 𝑠+𝑎
𝑠
1
𝑌 10𝑘 ∗ 𝑇1 ∗ 𝑠 10𝑘
= = 2
𝑅 1 + 10𝑘 + 𝑇 + 1 𝑠 + 𝑎𝑠 + 10𝑘
1 𝑠

𝑌(𝑠)
a) Obtain the closed loop transfer function 𝑇(𝑠) = 𝑅(𝑠) from the given block diagram:
𝜔𝑛2
𝑇(𝑠) =
𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2

b) Find the parameter k and a such the step response of T(s) becomes as the graph
shown
𝑃. 𝑂. = 40% → 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑡𝑎𝑏𝑙𝑒 → 𝜁 = 0.28
4 4
𝑇𝑠 = 2 = → 𝜔𝑛 = = 7.14
𝜁𝜔𝑛 2 ∗ 0.28
𝜔𝑛2 = 50
50
𝑇(𝑠) = 2
𝑠 + 4𝑠 + 50
K= 5 and a=4
49

Dominant Poles and Order Reduction

Consider the following second order system with 2 poles and no zeros:

The following figure shows the step response of G(s) together with the step response of the
following system:

Order Reduction
• Finding a system with lower order that behaves similar to the original system.
• When is order reduction possible?
o If the system has dominant pole(s)
o Dominant pole(s) are the poles very closer to the
jw‐axis (compared to the other poles) and there is
not zeros next to them.
o Very closer means at least 5‐10 times

Example:

• Eliminate the far pole (10 in this case)

• Adjust DC gain (should be the same before and after altering the far pole)
50

More Examples:

Pole and zero next to each other

If we have a pole next to a transfer function, they “cancel” each other, that’s why we keep
(s+10) instead of (s+1)
51

System Response with additional Pole

Having a zero close to the jw axis means high overshoot

System Response With Non-Minimum Phase Zero (Zero in the Right Side):
52

Examples of Non-Minimum Phase Systems:

Stability of LTI Systems

Definition of Stability:
Bounded Input Bounded Output (BIBO) Stability: A system is BIBO stable if every bounded
input yields a bounded output
Bounded Signal:
A signal 𝑢 𝑡 is called bounded if |𝑢(𝑡)| ≤ 𝑀; ∀𝑡 ≥ 0

Example of bounded signals:

• Step signal
• Sin signal

Example of unbounded signals:

• Ramp signal

How to check the stability

Using the definition: Not practical as all possible bounded inputs should be tested. Only
possible for simple systems.
Methods to check the stability: We can show that for any LTI system with transfer function
G(s), the system is stable if and only if all poles are in the left half plane (LFP).

Example:
53

Impulse Response of second order stable and unstable systems:

Roth Hurwitz Criterion:

It is a method to find the number of roots of a polynomial in the RHP.
Characteristic Equation:
Setting the denominator of the closed‐loop transfer function to zero
∆(𝑠) = 𝑎4 𝑠 4 + 𝑎3 𝑠 3 + 𝑎2 𝑠 2 + 𝑎1 𝑠 + 𝑎0 = 0

After filling the table, the number of roots in RHP is equal to the number of sign
changes in the first column.
54

Example 1: Is the following system stable?

1000
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3) 1000
𝑇(𝑠) = = 3
1000 2
𝑠 + 10𝑠 + 31𝑠 + 1030
1+
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)

There are 2 sign changes-> 2 roots in RHP -> unstable system

Example 2: For what values of k the closed loops system is stable?

𝑘
𝑌(𝑠) 𝑠(𝑠 + 2)(𝑠 + 3) 𝑘
𝑇(𝑠) = = = 3
𝑅(𝑠) 1 + 𝑘 2
𝑠 + 5𝑠 + 6𝑠 + 𝑘
𝑠(𝑠 + 2)(𝑠 + 3)

K > 30 → 2 sign changes → 2 roots in RHP

k < 0 → 1 sign change → 1 root in RHP
k = 30 → 2 roots on jw axis → will prove later
k = 0 → 1 root on jw axis → basically on origin
55

Example 3:
∆(𝑠) = 𝑠 3 + +5𝑠 2 + 6𝑠 + 30 = 0

𝑃(𝑠) = 𝑠 2 + 6
𝑑𝑃
= 0 = 2𝑠 = 0
𝑑𝑠

Note: the roots of P(s) are the roots of original

polynomial (here ∆(𝑠)) as well.

𝑃(𝑠) = 𝑠 2 + 6 = 0 → 𝑠 = ± 𝑗√6
Unstable with 2 poles on jw – axis (marginally stable)

Special Cases for Roth Methods

Case 1
• Name the row above zero row as “Auxiliary Polynomial” P(s)
• Differentiate P(s) and continue the table.
• The roots of P(s) are symmetric about the origin.

Example case 1:
∆(𝑠) = 𝑠 5 + 7𝑠 4 + 6𝑠 3 + 42𝑠 2 + 8𝑠 + 56 = 0

Notice that you can always divide a row

by a positive number.

Finding P(s) :
𝑃(𝑠) = 𝑠 4 + 6𝑠 2 + 8 = 0

𝑑𝑃
= 0 → 4𝑠 3 + 12𝑠 = 0
𝑑𝑠

How to interpret?
• P(s) is of order 4 → 4 roots symmetric about origin.
• No sign change in first column → no root in RHP
• Possible cases for having 4 roots symmetric about the origin
o in the case in red
the system is not stable
(marginally stable)
56

Why we consider unity feedback?

Control Loop with unity feedback

Difference between R(t) and y(t) is E(s)

(error)
Always make sure E(t) tends to zero

𝑦(𝑠) 𝑘(𝑠)∗𝐺(𝑠) 𝑈(𝑠) 𝑘(𝑠)

= =
𝑅(𝑠) 1+𝑘(𝑠)𝐺(𝑠) 𝑅(𝑠) 1+𝑘(𝑠)𝐺(𝑠)

𝑓𝑜𝑟𝑤𝑎𝑟𝑑
Note: 𝑓𝑜𝑟𝑚𝑢𝑙𝑎 = (𝑎𝑛𝑑 𝑖𝑡𝑠 + 𝑏𝑒𝑐𝑎𝑢𝑒 𝑡ℎ𝑒 𝑓𝑒𝑒𝑑𝑏𝑎𝑐𝑘 𝑖𝑠 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒)
1+𝑓𝑜𝑟𝑤𝑎𝑟𝑑∗𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒

Steady State Error (ess)

𝑒𝑠𝑠 = lim 𝑒(𝑡) = lim (𝑟(𝑡) − 𝑦(𝑡))

𝑡→∞ 𝑡→∞

Stability is a necessary condition for defining ess. If the closed loop is not stable, we cannot
define ess (or sometimes we say ess = ∞ for unstable systems)
On a stable system there is no steady-state error (infinity)

Example where the input is a random signal:

• Car driving 100km/h
o Position is a random signal
o What matters is not the velocity but the position
• 2 cars driving the same speed
o One is 10m behind
o Error = 10m
57

Example 1:

Example 2:

1
𝑌(𝑡)
= 𝑠 ∗2 =
2
𝑅(𝑡) 1 + (1 ∗ 2) 𝑠 + 2
𝑠

DC gain = 1 * (2/2)  stable system

Concluding: ess = 0

Note: we assume R(t) = 1

58

Example 3: ess to a ramp input:

Ess1 = 0 (converges into the reference)

Ess2 = ∞ (doesn’t have the same slope as the reference)
Ess3 = number (not determined)

Type of a system
Definition:
Type of a system is equal to “n” if and only if
1
• 𝑒𝑠𝑠 = 0 for 𝑅(𝑠) = 𝑠𝑛
𝑎𝑛𝑑
1
• 𝑒𝑠𝑠 ≠ 0 for 𝑅(𝑠) = 𝑠𝑛+1

Example:
What is the type is the following system?

Change in ess when increasing the order of s in R(s)

Goes from 0 to k to infinity if we

increase the order of s

1 1 1
• If 𝑒𝑠𝑠 𝑡𝑜 𝑅(𝑠) = is a number, then 𝑒𝑠𝑠 = ∞ to 𝑅(𝑠) = , R(s) = 𝑛+3 …
𝑠𝑛+1 𝑠𝑛+2 𝑠
1 1 1
• If 𝑒𝑠𝑠 𝑡𝑜 𝑅(𝑠) = is a number, then 𝑒𝑠𝑠 = 0 to 𝑅(𝑠) = , R(s) = 𝑛−1 …
𝑠𝑛+1 𝑠𝑛 𝑠
59

Step signal and ramp signal:

Step signal Ramp signal

Example: calculate ess with a ramp input:

𝑒𝑠𝑠 = lim 𝑒(𝑡) = lim 𝑟(𝑡) − 𝑦(𝑡) = lim(𝑠𝑅(𝑠) − 𝑠𝑌(𝑠)) = lim 𝑠𝑅(𝑠) − 𝑠𝑅(𝑠)𝑇(𝑠) = lim 𝑠 ∗ 𝑅(𝑠) ∗ (1 − 𝑇(𝑠)
𝑡→∞ 𝑡→∞ 𝑠→0 𝑠→0 𝑠→0

𝒆𝒔𝒔 = 𝐥𝐢𝐦 𝒔𝑹(𝒔) ∗ (𝟏 − 𝑻(𝒔))

𝒔→𝟎

Example 2:

Calculate ess to a step and ramp input.

1
𝑒𝑠𝑠 = lim 𝑠 ∗ 𝑅(𝑠) ∗ (1 − )
𝑠→0 𝑠2 + 𝑠 + 1
𝑠2 + 𝑠
𝑒𝑠𝑠 = lim 𝑠 ∗ 𝑅(𝑠) ∗ ( 2 )
𝑠→0 𝑠 +𝑠+1

𝑟(𝑡) = 𝑟𝑎𝑚𝑝 1 𝑠∗(𝑠+1) 1 𝑠(𝑠+1) 𝑠+1 0+1 1

{ 1 →lim 𝑠 ∗ ∗ 𝑠2 +𝑠+1 = lim 𝑠 ∗ 𝑠2 +𝑠+1 = lim 𝑠2 +𝑠+1 → 02 +0+1 = 1 = 1
𝑅(𝑠) = 𝑠→0 𝑠2 𝑠→0 𝑠→0
𝑠2

Concluding, we have a type 1

1 1 𝑠(𝑠 + 1) 𝑠+1
𝑅(𝑠) = 3
→ 𝑒𝑠𝑠 = lim 𝑠 ∗ 3 ∗ 2 = lim 2
→∞
𝑠 𝑠→0 𝑠 𝑠 + 𝑠 + 1 𝑠→0 𝑠(𝑠 + 𝑠 + 1)
60

When the closed loop block diagram looks as follows (with unity feedback), the steady state
error can be calculated by using:
Feedback is unity.
Closed loop is stable. 𝑘𝑝 = lim 𝐺(𝑠)
𝑠→0
1
𝑒𝑠𝑠 =
1 + 𝑘𝑝
For a step input
𝐺(𝑠)
𝑇(𝑠) =
1 + 𝐺(𝑠)
1
𝑒𝑠𝑠 = lim 𝑠𝑅(𝑠) ∗ (1 − 𝑇(𝑠)) = lim 𝑠 ∗ 𝑅(𝑠) ∗
𝑠→0 𝑠→0 1 + 𝐺(𝑠)

• when the reference is step → R(s) = 1/s

1 1 1
𝑒𝑠𝑠 = lim 𝑠 ∗ ∗ =
𝑠 1 + 𝐺(𝑠) 1 + 𝑘𝑝 1
𝑒𝑠𝑠 =
𝑘𝑝 = 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛 𝑒𝑟𝑟𝑜𝑟 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑘𝑣
𝑘𝑣 = lim 𝑠 ∗ 𝐺(𝑠)
𝑠→0
1 1 1 1 1
𝑅(𝑠) = 2 → 𝑒𝑠𝑠 = lim 𝑠 ∗ 2 ∗ = lim =
𝑠 𝑠→0 𝑠 1 + 𝐺(𝑠) 𝑠→0 𝑠 ∗ 𝐺(𝑠) 𝑘𝑣

• 𝑹(𝒔) = 𝟏/𝒔^𝟑
1
𝑒𝑠𝑠 = 𝑘𝑎 = lim 𝑠 2 ∗ 𝐺(𝑠)
𝑘𝑎 𝑠→0
Declaration error constant

Ess summary of the unity feedback case

Feedback is unity and control loop is stable

𝐴 𝐴
Step 𝑟(𝑡) = 𝐴 ∗ 𝑠𝑡𝑒𝑝 𝑅(𝑠) = 𝑒𝑠𝑠 = 𝐾𝑝 = lim 𝐺(𝑠)
𝑠 1 + 𝐾𝑝 𝑠→0

𝐴 𝐴
Ramp 𝑟(𝑡) = 𝐴 ∗ 𝑡 𝑅(𝑠) = 𝑒𝑠𝑠 = 𝐾𝑣 = lim 𝑠 ∗ 𝐺(𝑠)
𝑠2 𝐾𝑣 𝑠→0

𝑡2 𝐴 𝐴
Parabolic 𝑟(𝑡) = 𝐴 ∗ 𝑅(𝑠) = 𝑒𝑠𝑠 = 𝐾𝑎 = lim 𝑠 2 ∗ 𝐺(𝑠)
𝐾𝑎 𝑠→0
2 𝑠3
61

Example: Find the ess for step, ramp and parabolic references

Notice that the loop is stable!

1
• 𝑘𝑝 = lim 𝐺(𝑠) = ∞ → 𝑒𝑠𝑠 𝑠𝑡𝑒𝑝 = 1+𝑘 = 0
𝑠→0 𝑝

2∗1 2 1 3
• 𝑘𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 ∗ 𝑠(𝑠+3) = 3 → 𝑒𝑠𝑠 𝑟𝑎𝑚𝑝 = 𝑘 = 2
𝑠→0 𝑠→0 𝑣

2∗1 1
• 𝑘𝑎 = lim 𝑠 2 𝐺(𝑠) = lim 𝑠 2 ∗ 𝑠(𝑠+3) = 0 → 𝑒𝑠𝑠 𝑝𝑎𝑟𝑎𝑏𝑜𝑙𝑖𝑐 = 𝑘 = ∞
𝑠→0 𝑠→0 𝑎

Example: what is the minimum ess for a ramp reference

From the table:

𝐴
𝑒𝑠𝑠 =
1 + 𝐾𝑝
𝑘 1 𝑘
𝐾𝑣 = lim 𝑠 ∗ ∗ =
𝑠→0 𝑠 (𝑠 + 2)(𝑠 + 3) 6
Thus:
6
𝑒𝑠𝑠𝑟𝑎𝑚𝑝 =
𝑘

Checking for stability (closed loop transfer function)

𝑘
𝑌(𝑠) 𝑠(𝑠 + 2)(𝑠 + 3) 𝑘
𝑇(𝑠) = = → 𝑇(𝑠) = 3
𝑅(𝑠) 1 + 𝑘 2
𝑠 + 3𝑠 + 2𝑠 + 𝑘
𝑠(𝑠 + 2)(𝑠 + 3)

Characteristic equation: ∆(𝑠) = 𝑠 3 + 3𝑠 2 + 2𝑠 + 𝑘 = 0

We need to find if there is any root in RHP.

With this, we conclude for stability: 0 < 𝑘 < 6
And
𝑒𝑠𝑠 > 1
62

The effect of disturbance

𝑌(𝑠) = 𝑦𝑟 (𝑠) + 𝑦𝑑 (𝑠) → due to r(t) and d(t)

𝑦𝑟 (𝑠) = 𝑌(𝑠)|𝐷(𝑠)=0 and 𝑦𝑑 (𝑠) = 𝑌(𝑠)|𝑅(𝑠)=0

𝑌(𝑠) 𝐾(𝑠)𝐺(𝑠)
If D(s) = 0 → 𝑇(𝑠) = 𝑅(𝑠) = 1+𝐾(𝑠)𝐺(𝑠)

𝑌(𝑠) 𝐺(𝑠)
If R(s) = 0 → 𝑇𝑑 (𝑠) = 𝐷(𝑠) = 1+𝐾(𝑠)𝐺(𝑠)
Thus:
𝑌(𝑠) = 𝑇(𝑠) ∗ 𝑅(𝑠) + 𝑇𝑑 (𝑠) ∗ 𝐷(𝑠)
63

Internal Model Principle: Following reference.

• D(t) is called disturbance.

• The signals are affected by the disturbance.
• We want the output not to be affected by the disturbance.

Assuming the control loop is stable, we have 2 methods:

Method A
𝐷(𝑠) = 0
{ 1 → 𝑒𝑠𝑠 = 0
𝑅(𝑠) = 𝑛
𝑠

If and only if 𝐾(𝑠) ∗ 𝐺(𝑠) contains all unstable poles of R(s)

1 1
(if 𝑅(𝑠) = 𝑠𝑛 then 𝑘(𝑠) ∗ 𝐺(𝑠) 𝑠ℎ𝑜𝑢𝑙𝑑 𝑐𝑜𝑛𝑡𝑎𝑖𝑛 𝑠 )

Example: If R(s) is step and there are 3 poles, then ess=?

1
𝑅(𝑠) = = 𝑠𝑡𝑒𝑝 𝑖𝑛𝑝𝑢𝑡, 𝑡ℎ𝑒𝑛 𝑒𝑠𝑠 = ?
𝑠
𝑒𝑠𝑠 ≠ 0

Example: If R(s) is step and there are 2 poles, then ess=?

1
𝑅(𝑠) = = 𝑠𝑡𝑒𝑝 𝑖𝑛𝑝𝑢𝑡, 𝑡ℎ𝑒𝑛 𝑒𝑠𝑠 =?
𝑠
𝑒𝑠𝑠 = 0
This is a type one system

1
This happens because we have 𝑠 as k(s)
Because the controller has an integrator it means the system has a pole at 0

Example: If R(s) is ramp and there are 3 poles, then ess=?

1
𝑅(𝑠) = = 𝑟𝑎𝑚𝑝 𝑖𝑛𝑝𝑢𝑡, 𝑡ℎ𝑒𝑛 𝑒𝑠𝑠 =?
𝑠2
𝑒𝑠𝑠 ≠ 0

This system is not stable (try using the roth method to prove!)
64

Example: If R(s) is ramp and there are 4 poles, then ess=?

1
𝑅(𝑠) = = 𝑟𝑎𝑚𝑝 𝑠𝑖𝑔𝑛𝑎𝑙, 𝑡ℎ𝑒𝑛 𝑒𝑠𝑠 =?
𝑠2
𝑒𝑠𝑠 = 0

Because we have 2 integrators in the forward path

System is stable

Method B
𝑅(𝑠) = 0
{ 1 → The effect of d(s) on 𝑦𝑠𝑠 is zero if k(s) contains the poles of R(s); (in other words, if it contains 𝑠1𝑛)
𝐷(𝑠) = 𝑛
𝑠

Example: If R(s) is ramp and there are 4 poles, then ess=?

Knowing:
𝐴 3
𝐷(𝑠) = , 𝐾(𝑠) =
𝑠 𝑠

3 1
We know contains , therefore the disturbance does not affect 𝑦𝑠𝑠
𝑠 𝑠

Control Obejctives

Notice that G(s) represents actuator + plant + sensor: 𝐺(𝑠) = 𝐺𝑎 (𝑠) ∗ 𝐺𝑝 (𝑠) ∗ 𝐺𝑠 (𝑠)
Control objectives:
• The goal is to design a controller such that y(s) follows r(s) and the effect of
disturbance d(t) is rejected (as much as possible)
We are looking for:
• Stability
• Control performance (transient response, steady state response)
o P.O. ≤ 10%
o T.S ≤ 2 sec
o T.P ≤ 1 sec
• Robustness
o Guarantees the stability for the nominal part and the real part
65

How to determine the control performance

which is the best response?
- Depends on the application

Analysis:
• 2 is faster than 1 and has no overshoot, unless for very special applications, 2 is
better than 1
• 4 has many oscillations and damps slower than 3 therefore 3 is better than 4
• Between 2 and 3 depends on the application we want

On-off controller

• These controllers have constant oscillations because they cant maintain one value,
it’s either on or off

• Another issue is the frequent switching to the noise.

66

On-off controller with hysteresis

We use hysteresis to fix this issue.

Why does an on-off controller have oscillations?

• Controller only checks the sign of error. It doesn’t
consider the value of error
• To avoid that u(t) can be proportional to the error
(proportional controller)
𝑢(𝑡) = 𝑘𝑝 𝑒(𝑡)

Steady State error (ess) when using a proportional controller

When controlling the temperature of the room using P-controller the error cannot remain
zero

To show that:
• Lets assume the error is zero for a moment (meaning that the temperature has
reached 21°C for a moment)
• Since 𝑢(𝑡) = 𝑘𝑝 𝑒(𝑡) therefore u(t) becomes zero as well.
o This means that the actuator (heater) will switch off and therefore the
temperature will change (not remain zero)
67

A numerical Example:

𝑦𝑠𝑠 = ?
Lets assume 𝑦𝑠𝑠 = 30 → 𝑒𝑠𝑠 = 0 → 𝑈𝑠𝑠 = 0 → 𝑦𝑠𝑠 = 0 because we don’t get the
same value, we consider this a wrong assumption

• 𝑦𝑠𝑠 = 20 → 𝑒𝑠𝑠 = 10 → 𝑈𝑠𝑠 = 30 → 𝑦𝑠𝑠 = 90

• 𝑦𝑠𝑠 = 25 → 𝑒𝑠𝑠 = 5 → 𝑈𝑠𝑠 = 15 → 𝑦𝑠𝑠 = 45
• 𝑦𝑠𝑠 = 27 → 𝑒𝑠𝑠 = 3 → 𝑈𝑠𝑠 = 9 → 𝑦𝑠𝑠 = 27  this is the right assumption.

Finding analytically:
3
𝑌(𝑠) 3 ∗ 2𝑠 + 1 9
= = → 𝑦𝑠𝑠 = 30 ∗ 0.9 = 27
𝑅(𝑠) 1 + 3 2𝑠 + 10
2𝑠 + 1 ∗ 3

System with and without internal integrator

How does integrator affect a signal?

- If the system has an internal integrator then you can

achieve zero-state using an integrator:

The output remains constant when the inout is zero

68

Example: system with integrator

For systems with internal integrators, if the input is zero, the output remains constant

Theoretical Reason

We showed that for systems without

integrators, a P controller cannot achieve
zero ess. Here is the theoretical reason:
- According to the IMP, to have
ess = 0, the forward path should contain
one integrator (the pole of R(s)), since it is not the case here → 𝑒𝑠𝑠 ≠ 0

Effect of kp on the transient response (Case 1: systems with integrator)

We usen using P controller for a system with integrator,

• 𝑒𝑠𝑠 = 0 for all values of k (because of the integrator)
• Increasing kp increases overshoot (improves the speed as well)
• Increasing kp too much may make the system unstable
69

Effect of kp on the transient response (Case 1: systems without integrator)

What would happen if we changed the gain?

• more gain = more overshoot

• with p controller for a system with no integrator:
o 𝑒𝑠𝑠 ≠ 0
o By increasing kp, ess decreases
o By increasing kp, the overshoot increases
o By increasing kp too much, the system may become unstable

P controller with a bias term

Guessing:
B(s) = 10 because 10 * 3 = 30

This works if we know the DC gain of the system.

For practical systems, the DC gain may not be available, or it may change, therefore we
usually prefer to use an integrator in the controller (PI-Controller) that works independent of
the plant parameters.
70

Proportional – Integral (PI) controller

PI controller (in summary):

• Adds one integrator to the forward path, therefore ess drops to a high degree
• Decreases the stability margin of the control loop
• Always has 1 pole and 1 zero

Examples of different PI controller representations

These are all the same but with different representations

Ti = integrator time constant

71

Implementation of PI controller

The effect of changing the integral coefficient on the response

Increasing 1/Ti:
• System becomes more oscillatory
• Ess is zero as long as loop is stable (regardless of Ti)
• Better speed but more overshoot
• System gets faster
• Very high values for 1/Ti may cause instability of the system

If Ti = 0 we have a PI controller
72

The derivative term

(think of it like a segway)
• just because it is 85°
doesn’t mean it will turn right
• We need to account for
the speed and original
direction

Adding the derivative term (D-term in PID)

Make N big to make it further from the other poles and zeros so it doesn’t affect the
response (default: 100 = N)

By factoring kp from all terms:

This implementation is preferred in case of manual

tunning of the parameters (Simulink calls that ideal
implementation)
73

Td can improve the transient, avoiding oscillation.

Using the D term, we can reduce overshoot while keeping the speed

Important note about the D-Term

All signals are affected by noise
74

Summary: PID; PI; PD

2 Approaches for designing a PID Controller

Designing a PID controller
• Zero pole placement:
o Finding desired close loop transfer function
o Finding the controller gains (by placing the poles and zeros of the controller at
the right place) to achieve the desired closed loop
• Pre-Tune Fine Tune
o Using some methods (such as ZN1 and ZN2) to find decent values for
controller gains (Pre-tune)
o Fine tuning the controller gains

Zero pole Placement

Remember:

Standard second order system:

−𝜁𝜋
√1−𝜁 2
𝑃. 𝑂. = 100 ∗𝑒
{ 4
𝑇𝑠 =
𝜁 ∗ 𝜔𝑛
75

Example:

𝑒𝑠𝑠 − 𝑠𝑡𝑒𝑝 = 0
Design objectives: { 𝑡𝑠 ≤ 1𝑠𝑒𝑐
𝑃. 𝑂. ≤ 9%

It is possible to design a controller K(s) such that this control loop behaves like a 2nd order?
YES

1. Finding 𝜁 𝑎𝑛𝑑 𝜁𝜔𝑛

𝑒𝑠𝑠 − 𝑠𝑡𝑒𝑝 = 0
𝜁𝜔𝑛 = 4
{ 𝑡𝑠 ≤ 1𝑠𝑒𝑐 = { = 𝜔𝑛 = 6.6
𝑃. 𝑂. = 4 → 𝜁 = 0.6
𝑃. 𝑂. ≤ 9%

2. Analyze the desired control loop

2 43.5 (𝑠+2) 21.75 (𝑠+2)
𝑘(𝑠) ∗ (𝑠+2)(𝑠+8) = 𝑠(𝑠+8) = 43.5 ∗ 2𝑠
= 𝑠
Giving
1
𝑘(𝑠) = 21.75 + 43.5 ∗ 𝑠 this is a type of controller:
PI, because it has an integrator and a zero

Example: zero pole placement

What if the transfer function of the plant is:
2
𝐺(𝑠) =
𝑠(𝑠 + 2)(𝑠 + 8)
Then:
43.5 21.75(𝑠+2)
𝑘(𝑠) ∗ 𝐺(𝑠) = 𝑠+8
(=) 𝑘(𝑠) =
1
this is a PD controller
=43.5 + 21.75𝑠
43.5 = 𝑘𝑝
Being: {
21.75𝑠 = 𝑘𝑖

Example: zero pole placement

What if the transfer function of the plant is:
2
𝐺(𝑠) =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 8)

43.5 21.75 (𝑠 + 1)(𝑠 + 2)

𝑘(𝑠) ∗ 𝐺(𝑠) = (=) 𝑘(𝑠) = < −𝑃𝐼𝐷 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑙𝑒𝑟
𝑠(𝑠 + 8) 𝑠

21.75𝑠 2 + 62.25𝑠 + 43.5

=
𝑠
32.75 = 𝑘𝑑
Being {62.25 = 𝑘𝑝
43.5 = 𝑘𝑖
76

Example: zero pole placement

What if the transfer function of the plant is:
2
𝐺(𝑠) =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 4)
This system has 3 poles where none of them are at the desired location. PID controller can
only cancel 2 poles.
We cannot find a PID controller using the “zero-pole placement” approach to achieve the
desired behavior, because the system has 3 poles and we can’t find all of them.
The math behind it:

43.5
𝑘(𝑠) ∗ 𝐺(𝑠) =
𝑠(𝑠 + 8)
If we were to calculate k(s) we will see we have poles and zeros we cannot find

Pre-tune fine-tune approach

Zigler Nichols (Z-N) first method:

77

Controller Type 𝑘𝑝 𝑇𝑖 𝑇𝑑
𝑇
P ∞ 0
𝐿
𝑇 𝐿
PI 0.9 0
𝐿 0.3
𝑇
PId 1.2 2L 0.5L
𝐿
Z-N first method can only be applied to the system that have stable step response with NO
oscillation and NO overshoot.

Step system response does not have pure delay:

Zigler Nichols (Z-N) second method:

Uses the closed loop response.

1. Close the loop with a P controller.
2. Start with a small gain and increase it such that you get a pure sinusoidal response.
3. Name the gain as 𝑘𝑢 , and the period of oscillations as 𝑇𝑢

\ Parameters
Controller\ 𝑘𝑝 𝑇𝑖 𝑇𝑑

P 0.5𝑘𝑢 ∞ 0
𝑇𝑢
PI 0.45𝑘𝑢 0
1.2
𝑇𝑢
PD 0.8 𝑘𝑢 ∞
8
𝑇𝑢 𝑇𝑢
PID 0.6𝑘𝑢
2 8

Notice that the parameters of the controller are suggested

such that the overshoot of the response drops by a factor of 4.
78

Example: 2nd approach Z-N Method

Design a controller using Z-N second method:

To find 𝑘𝑢 and 𝑇𝑢 there are several methods:

1. “easiest” Roth Method:
4𝑅
𝑌(𝑠) (𝑠 + 2)3 4𝑘
= = 3
𝑅(𝑠) 1 + 4𝑘 2
𝑠 + 6𝑠 + 12𝑠 + 8 + 4𝑘
(𝑠 + 2)3
Giving:
∆𝑠 = 𝑠 3 + 6𝑠 2 + 12𝑠 + 8 + 4𝑘 = 0

Therefore:
72 − 8 − 4𝑘
= 0 (=) 𝑘 = 16 = 𝑘𝑢
6

𝑖𝑓 𝑘 = 16 → 𝑃(𝑠) = 6𝑠 2 + 72 < −𝑡ℎ𝑖𝑠 72 𝑐𝑜𝑚𝑒𝑠 𝑓𝑟𝑜𝑚 8 + 4 ∗ 16

𝑃(𝑠) = 0 → 𝑠 2 = 12 → 2 = ±𝑗√12

Note: 𝑘𝑢 = gain that makes

the system marginally
stable
79

Root Locus (drawing the location of the closed loop)

It is a method to sketch the position of closed loop poles when changing the parameters of
the system or the controller.
- why closed looped poles are important?
- Stability only depends on the closed loop poles
- Stability is a function of poles
- They affect the transient response as well
- Transient response is a function of poles and zeros

Example: sketching the location of a closed loop pole

The idea is to draw the location of the closed loop pole when changing the parameter k.

𝑘
𝑌(𝑠) 𝑠 + 1 = 𝑘
=
𝑅(𝑠) 1 + 𝑘 𝑠+1+𝑘
𝑠+1
The closed loop pole =−1 − 𝑘
𝑖𝑓 𝑘 = 0 → 𝑠 = −1
𝑖𝑓 𝑘 = 1 → 𝑠 = −2
𝑖𝑓 𝑘 = 10 → 𝑠 = −11
𝑖𝑓 𝑘 = ∞ → 𝑠 = −∞

The closest to the imaginary part the slowest the response will be.
80

Example:

What happens if we change the gain?

(𝑠 − 2)
𝑌(𝑠) 𝑘∗ 𝑘(𝑠 − 2) −3 + 2𝑘
= 𝑠+3 = →𝑠=
𝑅(𝑠) 1 + 𝑘 ∗ 𝑠 − 2 (𝑘 + 1)𝑠 + (3 − 2𝑘) 𝑘+1
𝑠+3

k 0 1 1.5 2 Infinity
s -3 -0.5 0 0.33 2

Increasing k, the system becomes slower and more

unstable.

Notice that (in this example):

• For k = 0 the CLP is equal to the OLP
• For k = infinity, the CLP is equal to the OLP
• When k>1.5, the closed loop is unstable.

Example: Generic Root Locus example

𝐾𝐺(𝑠)
𝑇(𝑠) =
1 + 𝑘𝐺(𝑠)𝐻(𝑠)

∆(𝑠) = 1 + 𝑘𝐺(𝑠)𝐻(𝑠) = 0 finding s as a function of k

1
𝐺(𝑠)𝐻(𝑆) = −𝑘 s is a complex number therefore G(s)H(s) is also a complex number

1
Notice that − 𝑘 is negative because ( k > 0 )
81

Question: How to find the angle G(s)H(s)

1
|𝐺(𝑠)𝐻(𝑠)| ∠𝐺(𝑠)𝐻(𝑠) = | | ∠180
𝑘

1
|𝐺(𝑠)𝐻(𝑠)| = | |  Magnitude criteria
𝑘
∠𝐺(𝑠)𝐻(𝑠) = 180 Angle criteria.

Note, to find the root locus we find all possible points “s” that satisfy the angle criteria

Recap: Polar coordinates

Suppose A and B are complex numbers
𝐴 = 𝑎 + 𝑏𝑗
𝐵 = 𝑐 + 𝑑𝑗
𝑏
In polar: 𝐴 = 𝑙𝑎 ∗ 𝜃𝑎 = √𝑎2 + 𝑏 2 ∗ tan−1 (𝑎)

𝑎=𝑐 𝑙 = 𝑙𝑏
When A = B →{𝑏𝑗 = 𝑑𝑗 𝑡ℎ𝑖𝑠 𝑚𝑒𝑎𝑛𝑠: { 𝑎
𝜃𝑎 = 𝜃𝑏

Question: how to find ∠𝐺(𝑠)𝐻(𝑠)

Suppose that,
𝑠+4
𝐺(𝑠)𝐻(𝑠) =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)

∠𝐺(𝑠)𝐻(𝑠) = ∑ 𝐴𝑛𝑔𝑙𝑒 𝑜𝑓 𝑧𝑒𝑟𝑜𝑠 − ∑ 𝐴𝑛𝑔𝑙𝑒 𝑜𝑓 𝑝𝑜𝑙𝑒𝑠

= 𝜙 − 𝜃1 − 𝜃2 − 𝜃3

𝑖𝑓 ∠𝐺(𝑠)𝐻(𝑠) = 180 ± 𝑛360, then the point of interest in on the root locus

Graphical approach to fund the phase of G(s)H(s) at the point of interest.

• Draw lines form “POI” to all open loop poles and zeros.
• Phase of G(s)H(s) is ∑ 𝜑𝑖 − ∑ 𝜃𝑖
82

Example: can you find all the possible points that satisfy the angle criterion?

𝐾 1
𝐺(𝑠)𝐻(𝑠) = = 𝑘∗
(𝑠 + 4)(𝑠 + 2) (𝑠 + 4)(𝑠 + 2)

Those are the points where 𝜃1 + 𝜃2 = 180

“POI” is on the root locus if the angle

G(s)H(s) = 180 + n*360

0 − (𝜃1 + 𝜃2 ) = 180 + 𝑛 ∗ 360

Rules for sketching Root Locus Rules 1, 2, 3:

1. The number of branches equals the number of open loop (loop gains) poles.
2. Root Locus is symmetric about the real axis.
3. For the real axis segment, when 𝑘 ≥ 0, the point on the real axis belongs to the root
locus if it is the left side of an odd number of zeros/poles:
a. If we add the number of poles and zeros on the right of the
selected number:
i. Odd amount = belongs to RL
ii. Even amount = doesn’t belong to RL

Examples:

Rules for sketching Root Locus Rules 4 and 5:

4. Each branch begins at an open loop pole and ends at an open loop zero or infinity.
a. N = number of open loop poles
b. M = number of open loop zeros
If n = m , all branches end at zeroes
If n>m, m branches end at zeros and (n-m) branches go to infinity

5. For the branches that end in infinity, we can draw the asymptotes:
∑ 𝑝𝑜𝑙𝑒𝑠−∑ 𝑧𝑒𝑟𝑜𝑠
a. Real axis:
𝑛−𝑚
2𝑛+1
b. Angles: 𝑛−𝑚
∗ 180
83

Example:

𝑛=2
{ →𝑛−𝑚 =1
𝑚=1
∑ 𝑃𝑜𝑙𝑒𝑠 − ∑ 𝑍𝑒𝑟𝑜𝑠 0 + (−2) − (−1)
𝜎𝜃 = = = −1
𝑛−𝑚 1
𝜃𝑎 = 180 ( 𝑓𝑟𝑜𝑚 𝑡ℎ𝑒 𝑓𝑜𝑟𝑚𝑢𝑙𝑎 𝑏𝑒𝑓𝑜𝑟𝑒)

Example:
𝐾
𝑘𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠 + 2)
𝑛=2
{ → 𝑛−𝑚 =2
𝑚=0
Asymptotes:
∑ 𝑃𝑜𝑙𝑒𝑠 − ∑ 𝑍𝑒𝑟𝑜𝑠 0 + (−2)
𝜎𝑎 = = = −1
𝑛−𝑚 2
90
𝜃={
−90
The pole will leave the real axis at -1

Example:

𝑛=3
{ → 𝑛−𝑚 =3
𝑚=0
Asymptotes:
∑ 𝑃𝑜𝑙𝑒𝑠 − ∑ 𝑍𝑒𝑟𝑜𝑠 0 + (−2 − 4)
𝜎𝑎 = = = −2
𝑛−𝑚 3
60
𝜃 = { 180
−60
Small k → poles in real axis
Increased k → poles break out (green line)
84

Rules for sketching Root Locus Rule 6:

6. Real axis break-away and break-in points

𝑑𝑘
By solving the equation 𝑑𝑠
= 0 you can find possible break-away and break-in points. Notice
𝑑𝑘
that, when finding s such that 𝑑𝑠
= 0, you need to find k as well and if k>0, then the answer
is accepted.

Example:
𝑛=3
{ 𝑡ℎ𝑒𝑛 𝑛−𝑚 =3
𝑚=0

Finding the break-away point:

∆(𝑠) = 1 + 𝐺(𝑠)𝐻(𝑠) = 0
𝑘
(=) 1 + =0
𝑠(𝑠 + 2)(𝑠 + 4)
K as a function of s:
𝑑𝑘
𝑘 − 𝑠(𝑠 + 2)(𝑠 + 4) → =0
𝑑𝑠
Therefore:
𝑠 = 0.84 → 𝑘 = 3.07
{ 1
𝑠2 = −3.15 → 𝑘 = −3.07

Rules for sketching Root Locus Rule 7:

7. The jw-axis crossing → important for finding stability
Two methods for finding jw-crossing points:
a) Routh Table
b) By finding resonance frequency
85

Example:
𝑘
𝑘𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠 + 2)(𝑠 + 4)

The point of interest is the point where

the red line intercepts the jw axis!

Finding k for marginal stability:

∆(𝑠) = 1 + 𝑘𝐺(𝑠)𝐻(𝑠) = 0 → ∆(𝑠) = 𝑠(𝑠 + 2)(𝑠 + 4) + 𝑘 = 0
∆(𝑠) = 𝑠 3 + 6𝑠 2 + 8𝑠 + 𝑘 = 0

Notice that if k>48, we will have 2 sign changes (two roots on the right side of the plane),
therefore for the jw axis crossing point we have k=48

If k = 48, → zero row → 𝑃(𝑠) = 6𝑠 2 + 𝑘 = 0 → 6𝑠 2 + 48 = 0 (=) 𝑠 = ±𝑗√8

How to find the fain and the jw-crossing axis points (with MATLAB):

You can change the gain and step response at the same time.
To make the response faster, we need more than one gain. We need to pull the right
poles to the left!

Usage of magnitude criterion

We had:
1
∆(𝑠) = 1 + 𝑘𝐺(𝑠)𝐻(𝑠) = 0 → 𝐺(𝑠)𝐻(𝑠) =
𝑘
1
|𝐺(𝑠)𝐻(𝑠)| =
𝑘
𝑘
Suppose that for the example ∆(𝑆) = 1 + =0
𝑠(𝑠+2)
You have drawn the root locus:

For what value of k the root locus is at -1+j2?

∆(𝑠) = 0 → 𝑠(𝑠 + 2) + 𝑘 = 0 → 𝑘 = −𝑠(𝑠 + 2)

86

So, we have 2 different methods of solving this:

1. Analytically:
∆(𝑠) = 0 → 𝑠(𝑠 + 2) + 𝑘 = 0
𝑘 = 𝑠(𝑠 + 2)|−1+𝑗2 = (1 − 𝑗2)(1 + 𝑗2)
= (1) + (4) = 5
2. Graphically:
Knowing that:
1
𝐺(𝑠)𝐻(𝑠) =
𝑘
1 1 ∏𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑂𝐿 𝑝𝑜𝑙𝑒𝑠
𝑘= = =
𝐺(𝑠)𝐻(𝑠) ∏ 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑧𝑒𝑟𝑜𝑠 ∗ 𝑔𝑎𝑖𝑛 ∏𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑂𝐿 𝑧𝑒𝑟𝑜𝑒𝑠 ∗ 𝑔𝑎𝑖𝑛
∏𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑝𝑜𝑙𝑒𝑠

1) Draw lines from the point of

interest to all OL zeros and poles.
2) Calculate the distances to the OL
poles:

𝑙1 ∗ 𝑙3 ∗ 𝑙4
𝑘=
𝑙2 ∗ 𝑔𝑎𝑖𝑛

Example:
1
𝐺(𝑠)𝐻(𝑠) =
𝑠(𝑠 + 2)
We have a gain of 1

Connect the point of interest to all zeros and poles and then find the
distances:
𝑙1 ∗𝑙2
𝑘= 1∗1
= 𝑙1 ∗ 𝑙2 = √12 + 22 ∗ √12 + 22 = 5

We put 1 because G(s)H(s) has no zeros.

87

Rules for sketching Root Locus Rule 888 (if n>m+2):

Rule: Summation of the closed loop poles is always a constant value
Example:
𝑘
𝑘𝐺(𝑠)𝐻(𝑠) =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)
This means:
𝑛=3
{ → 𝑤𝑒 𝑐𝑎𝑛 𝑢𝑠𝑒 𝑡ℎ𝑖𝑠 𝑟𝑢𝑙𝑒
𝑚=0

If k = 0 → summation of the closed loop poles = -1-2-3 = -6

According to the rule, for any value of k, the summation must remain equal to “-6”.
For 𝑘 = 𝑘𝑢 two of the poles are at 0 − 𝑗𝑤𝑢 and 0 + 𝑗𝑤𝑢 therefore the third pole is at -6

Point of interest is S=-6:

a) Graphical approach:

𝑙1 ∗ 𝑙2 ∗ 𝑙3 5 ∗ 4 ∗ 3
𝑘= = = 60
1∗1 1∗1

b) Analytical approach

𝑘
∆(𝑠) = 1 + 𝑘𝐺(𝑠)𝐻(𝑠) = 0 → 1 + =0
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)
𝑘 = −(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)|𝑠=−6 = 60
To find 𝑤𝑢 (if it is needed) other than the Routh table, you can divide ∆(𝑠) by (𝑠 + 6):
𝑘
∆(𝑠) = 1 + |
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3) 𝑘=60
∆(𝑠) = 𝑠 3 + 6𝑠 2 + 11𝑠 + 66 = 0
𝑠 3 + 6𝑠 2 + 11𝑠 + 66| 𝑠+6
𝑠2 +11

𝑠 3 + 6𝑠 2 + 11𝑠 + 66 = (𝑠 + 6)(𝑠 2 + 11)

𝑤𝑢 = 𝑗√11
88

Design a controller using Root Locus

The first step for designing a controller are as follows:
1) For given design objectives, we need to specify the region of interest (ROI) which is
basically an acceptable region for the closed loop poles.
2) To check whether or not the root locus of the system (with only a gain k as the
controller) enters the ROI.

Example: Finding the ROI for given design objectives:

𝑡𝑠 ≤ 2 sec 𝑎𝑛𝑑 𝑡𝑝 ≤ 1 𝑠𝑒𝑐

This means:
𝜁𝑤𝑛 ≥ 2
And
𝜋
≤ 1 → 𝑤𝑑 ≥ 𝜋
𝑤𝑑

The region of interest will be the mixture of both: Reminder:

𝑠1,2 = 𝑤𝑛 ± 𝑗𝑤𝑑
𝑤𝑑 = √1 − 𝜁 2 𝑤𝑛
The closed loop poles should be 4
Settling time: 𝑡𝑠 =
on the ROI. 𝜁𝑤𝑛
𝜋
Peak Time: 𝑡𝑝 =
𝑤𝑑
−𝜁𝜋
√1−𝜁2
P.O. = 100 𝑒

Example:
𝑡𝑠 ≤ 4 𝑠𝑒𝑐
{ 4 → 𝜁𝑤𝑛 ≥ 1
𝑡𝑠 =
𝜁𝑤𝑛

Note: 𝜁𝑤𝑛 is the distance of the CP to the jw axis

𝑃. 𝑂. ≤ 9%
This means:
𝜁 ≥ 0.6
𝜁 = cos 𝜃
89

Controller Design Case 1: When the root locus is already passing through ROI
Example:
1
𝐺(𝑠) =
𝑠(𝑠 + 2)
The goal is to design a controller such that P.O <= 9%
This means that 𝜁 ≥ 0.6 → 𝜃 ≤ 53

Notice that a gain k is enough in this example (as a controller). Also notice that k could be
selected between 0 and 𝑘∆ , 0 ≤ 𝑘 ≤ 𝑘∆ , where 𝑘∆ brings the CLP to the point ∆.

Question: what the value of is 𝑘∆ ?

Graphical approach:
To find 𝑘∆ , lets fund the point ∆ first:
Thus ∆= −1 + 1.33𝑗

Analytical approach:
𝑘
∆ (𝑠) = 1 + =0
𝑠(𝑠 + 2)
𝑘 = −𝑠(𝑠 + 2)|𝑠=−1+1.33𝑗 = 2.77
Therefore
𝑘∆ = 2.77
In control, in many cases we select the border value.
Therefore, the controller here is 𝑘(𝑠) = 2.77

Example:

Design objective: 𝑡𝑠 ≤ 4𝑠𝑒𝑐

→ 𝜁𝑤𝑛 ≥ 1

Notice that the root locus enters the ROI, therefore we are in case 1
(meaning that the controller could be only a gain).
Notice that all gains such that 0 ≤ 𝑘 ≤ 𝑘∆ are acceptable
What is 𝑘∆ ?
Method 1:
We can use the extra rule for root locus:
K=0 → sum of CLP → -2-4-8 = -14
K = 𝑘∆ → sum of CLP = -14 :
- One pole at −1 + 𝑗𝑤
- One pole at −1 − 𝑗𝑤
- One pole at P
90

How to calculate P:
(−1 + 𝑗𝑤) + (−1 − 𝑗𝑤) + 𝑃 = −14 → 𝑃 = −12
Now the question is what k when the root locus is at s = -12
𝑙1 ∗ 𝑙2 ∗ 𝑙3
𝑘∆ = = 4 ∗ 8 ∗ 10 = 320
1∗1

Method 2:
To shift the jw axis 1 unit to the left (coordinate change) and then use the Roth method:
𝑠0 = 𝑠 + 1 → 𝑠 = 𝑠0 − 1
∆(𝑠) = 1 + 𝑘(𝑠0 + 2)(𝑠0 + 4)(𝑠0 + 8) = 0
∆(𝑠0 ) = 1 + 𝑘(𝑠0 + 1)(𝑠0 + 3)(𝑠0 + 7) = 0
(=)𝑠03 + 11𝑠02 + 31𝑠0 + 21 + 𝑘 = 0

341 − 21 − 𝑘 = 0 → 𝑘 = 320

Controller Design Case 2: When the root locus needs to be changed

𝑘(𝑠−𝑧)
What are the steps to design a first order controller 𝑘(𝑠) = 𝑠−𝑝
If the integrator is not
needed:
1. Finding the desired location for the closed loop pole (point ∆) based on the design
objectives (usually a boarder points of the ROI)
2. Placing the zero of the controllers (z) at a proper point (usually under the point ∆)
3. Placing the pole of controller (p) to fulfill the angle criterion at the desired point.
4. Finding the gain of controller (k) to fulfill magnitude criterion (z) (using analytical or
graphical approaches)

Example:

Canceling the poles/zeros of the plant with the zero/pole of the controller:
Steps:
1) Cancelling the pole of systems at -3 with zero of the controllers
After cancelling the pole at -3, with the controller zero, to pass the desired place, the
controller pole should be at -8.
To find the controller gain k:
∏𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑝𝑜𝑙𝑒𝑠 𝑙1 ∗ 𝑙2
𝑘= = = 44.4
∏𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑜 𝑡ℎ𝑒 𝑧𝑒𝑟𝑜𝑠 ∗ 𝑔𝑎𝑖𝑛 1
44.4 (𝑠 + 3)
𝑘(𝑠) =
(𝑠 + 8)
Note: Never cancel the unstable pole/zero of the plant with unstable pole/zero of the
controller (if you do, the overall system becomes internally unstable!)
91

2) Placing the pole of controller to fulfill the angle criterion at the desired point.

𝑡𝑠 ≤ 1 sec → 𝜁𝑤𝑛 ≥ 4
{
𝑃. 𝑂. ≤ 9% → 𝜁 ≥ 0.6 → 𝜃 = 53

∆ = 4 ± 5.33𝑗

3) Finding the gain of controller (using analytical or graphical approaches)

If the root locus doesn’t enter the ROI we

have a case 2 and we need to ad a zero
and a pole to bring the poles to the left:

Graphical approach:

Angle criterion: ∑ 𝜃𝑖 − ∑ 𝜑𝑖 = 180

𝜃1 = 180 − 53 = 127
5.33
𝜃2 = 180 − tan−1 ( ) = 100.6
1
𝜑1 = 90
This means:
𝜃3 = 42.5
5.33
tan(𝜃3 ) =
𝑥
Therefore:
𝑥 = 5.82 𝑃 = −9.82
92

Finding k:
𝑙! ∗ 𝑙2 ∗ 𝑙3 √42 + 5.332 ∗ √12 + 5.332 ∗ √5.822 + 5.332
𝑘= = = 53.5
𝑔𝑎𝑖𝑛 ∗ 𝑙4 1 ∗ 5.33
Therefore:
𝑘(𝑠 − 7) 53.5(𝑠 + 4)
𝑘(𝑠) = =
𝑠−𝑃 𝑠 + 9.82

Example: If the controller needs to have integrator:

1
Design a controller for 𝐺(𝑠) = 𝑠+2 such that:

1
So: We conclude the new plant will be: 𝐺1 (𝑠) = 𝑠(𝑠+2) ; 𝑡𝑠 ≤ 1 sec ; 𝑃. 𝑂. ≤ 9%
𝑘(𝑠 + 𝑧)
𝑘1 (𝑠) =
𝑠+𝑃
𝑘(𝑠 + 𝑧)
𝑘(𝑠) =
𝑠(𝑠 + 𝑃)
…
93

Reminder: trigonometry
94

Frequency Response
For a system with transfer function G(s), the frequency response can be obtained by
substituting s =jw:

𝐺(𝑗𝑤) = 𝐺(𝑠)|𝑠=𝑗𝑤
Example
1 1
𝐺(𝑠) = → 𝐺(𝑗𝑤) =
𝑠+1 (𝑗𝑤 + 1)

1 ∠0 1
𝐺(𝑗𝑤) = 𝑤 = ∠ − tan−1 (𝑤)
2
√42 + 1∠ tan−1 ( 1 ) √𝑤 + 1
Plotting the frequency response:
Bode diagram (bellow):

2 methods for drawing the frequency response:

1. Polar Plot 2. Bode Plot:
95

Example:
1
𝐺(𝑠) =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 3)

1
𝐺(𝑗𝑤) =
(𝑗𝑤 + 1)(𝑗𝑤 + 2)(𝑗𝑤 + 3)
1 1 1 𝑤 𝑤
= ∗ ∗ ∗ ∠ − tan−1 (𝑤) − tan−1 ( ) − tan−1 ( )
√𝑤 2 + 1 √𝑤 2 + 4 √𝑤 2 + 9 2 3

Understanding the frequency response (main property)

For an LTI system, if the input is sinusoidal with frequency w, the output will be sinusoidal as
well, with the same frequency.
The magnitude and phase of the output depends on the “magnitude and phase of the input”
and the “magnitude and phase of the frequency response at frequency w”

𝐵𝑠𝑖𝑛 (𝑤0 𝑡 + 𝜑)

𝑢(𝑡) = 𝐴 ∗ sin (𝑤0 𝑡 + 𝜃)

Thus:
𝐵 =? 𝐵 = 𝐴 ∗ |𝐺(𝑗𝑤0 )|
{ → {
𝜑 =? 𝜑 = 𝜃 + ∠𝐺(𝑗𝑤0 )
96

Example:

1
𝐺(𝑗𝑤) =
𝑗𝑤 + 1

1
𝐺(𝑗𝑤) = ∠ − tan (𝑤)
√𝑤 2 +1

Note, from math we know:

2𝜋
𝑤=
𝑇

And
𝐵 = 𝐴 ∗ |𝐺(𝑗𝑤0 )|
𝜑 = 𝜃 + ∠𝐺(𝑗𝑤0 )
Therefore: we assume w = 1

1
|𝐺(𝑗 ∗ 1) = = 0.71
{ √2
∠𝐺(𝑗 ∗ 1) = − tan−1 (1) = −45°
Thus:
𝐵 = 𝐴 ∗ 0.71 = 0.71
𝜑 = 𝜃 + ∠𝐺(𝑗 ∗ 1) = −45°

New approach to find the transfer function of a system.

• You can apply sine wave inputs with different frequencies and find the magnitude
and phase output at each frequency.
• Next step is to find the frequency response.
• Last Step is to find the transfer function that matches the obtained frequency
response (Yet to see how).

Assymptotic approximation of Bode Diagram

For a given transfer function:
4(𝑠 + 2)
𝐺(𝑠) =
𝑠(𝑠 + 5)(𝑠 + 10)(𝑠 2 + 2𝑠 + 2)
In this course we are more interested in asymptotic approximation of Bode diagram (instead
of the exact diagram).
To plot the approximation one, the first step is to rewrite the transfer function in natural
frequency format:
4∗2 𝑠
𝑠 ∗ 10 ∗ 2 (1 + 2)
𝐺(𝑠) =
𝑠 𝑠 𝑠2
𝑠 ∗ (1 + ) ∗ (1 + 10) ∗ (1 + 𝑠 + 2 )
5
𝑠 1 1 1 1
𝐺(𝑠) = 0.08 ∗ (1 + ) ∗ ∗ 𝑠 ∗ 𝑠 ∗ 2
2 𝑠 1+ 1 + 10 1 + 𝑠 + 𝑠
5 2
97

𝑗𝑤 1 1 1 1
𝐺(𝑗𝑤) = 0.08 ∗ (1 + )∗ ∗ ∗ ∗
2 𝑗𝑤 1 + 𝑗𝑤 𝑗𝑤 𝑤2
1+
𝑠 10 1 − 2 + 𝑗𝑤
For plotting the exact Bode Diagram:
𝑤2
20 log |𝐺(𝑗𝑤) | = 10 log(0.08) + 20log (|√1 +
+ ⋯)
2
𝑗𝑤 𝑤
∠𝐺(𝑗𝑤) = ∠0.08 + ∠ (1 + ) + ⋯ = tan−1 ( )
{ 2 2

• But as said earlier, the focus is on approximate bode diagram.

• In the following we will see the approximate bode diagram for each term separately.
Later we can add the separate parts to find the overall approximate bode diagram.’

Assymptotic Bode Diagram for Different terms

𝑮(𝒔) = 𝒔

𝟏
𝑮(𝒔) =
𝐬
98

𝑮(𝒔) = 𝒔𝒏

𝟏
𝑮(𝒔) =
𝒔𝒏

𝑮(𝒔) = 𝒌; 𝒌>𝟎
99

𝑮(𝒔) = 𝒌; 𝒌<𝟎

𝒔
𝑮(𝒔) = 𝟏 +
𝒂

𝑗𝑤
𝐺(𝑗𝑤) = 1 +
𝑎
𝑤 ≪ 𝑎 → 𝑤 < 0.1𝑎
𝐺(𝑗𝑤) ≈ 1

𝑤 ≫ 𝑎 → 𝑤 > 10𝑎
𝑗𝑤
𝐺(𝑗𝑤) ≈
𝑎

𝟏
𝑮(𝒔) = 𝒔
𝒔+𝒂
100