KON 317E Control Systems

Design via Root Locus

Assoc. Prof. Dr. Tufan Kumbasar
Motivating Example
• Assume that the RL of a system is given as in the figure
and B is the desired pole
• Problem
• We can’t access B by varying K!
• Condition of angle is not satisfied
• Solutions
• Change parameters of the original system:
• Impossible or Expensive
• Add a Control/ Compensation System!

Slide 2
Controllers & Compensator: Definitions, Structures…
• Definition
• A subsystem represented as a transfer function inserted into the forward or FB path for the
purpose of improving the transient response or steady-state error.
• Main goal of the Controller
• Adds new open loop poles and zeros, thereby creating new RL that goes through the desired
closed loop pole locations.
• Two main configurations

Slide 3
Controllers & Compensator: Definitions, Structures…
• Simple control systems have two main purposes (generally) :
• Improving the transient response by changing pole locations.
• Improving the steady-state performance by changing the Type of the system.

• Proportional (P) Controller: Feed scaled error to the plant.

𝑢 𝑡 =𝐾 𝑒 𝑡 →𝐶 𝑠 =𝐾
• Integral (I) Controller: Feed integrated error
 
to the plant.
𝐾
𝑢 𝑡 =𝐾 𝑒 𝜏 𝑑𝜏 → 𝐶 𝑠 =
𝑠
 
• Derivative (D) Controller: Feed differentiated error to the plant
𝑑𝑒(𝑡)
𝑢 𝑡 =𝐾 →𝐶 𝑠 =𝐾 𝑠
𝑑𝑡

Slide 4
Controllers & Compensator: Definitions, Structures…
• Proportional-Integrator (PI) Controller
  : Feed scaled & integrated error to the plant:
𝐾 𝐾(𝑠 + 𝑎)
𝑢 𝑡 = 𝐾 𝑒 𝑡 + 𝐾 𝑒 𝜏 𝑑𝜏 → 𝐶 𝑠 = 𝐾 + ≜
𝑠 𝑠
 

Slide 5
Controllers & Compensator: Definitions, Structures…
• Proportional-Derivative (PD) Controller : Feed scaled & differentiated error to the plant:
𝑑𝑒(𝑡)
𝑢 𝑡 =𝐾 𝑒 𝑡 +𝐾 → 𝐶 𝑠 = 𝐾 + 𝐾 𝑠 ≜ 𝐾(𝑠 + 𝑏)
𝑑𝑡

Slide 6
Controllers & Compensator: Definitions, Structures…
• Proportional-Integrator -Derivative (PID) Controller : Feed scaled & integrated &
differentiated error to the  plant:
𝑑𝑒(𝑡) 𝐾 𝐾(𝑠 + 𝑎)(𝑠 + 𝑏)
𝑢 𝑡 = 𝐾 𝑒 𝑡 + 𝐾 𝑒 𝜏 𝑑𝜏 + 𝐾 → 𝐶 𝑠 = 𝐾 + + +𝐾 𝑠 ≜
𝑑𝑡 𝑠 𝑠
 

Slide 7
Improving Steady-State Error Performance
• Increasing the “Type” of the Feedforward Transfer function
• I, PI, and PID controllers might be a solution

Slide 8
Improving Steady-State Error Performance
• Remember the condition for zero steady- state errors
• Number of pure integrations in the feedforward loop
• Let us assume that it is desired to place the poles at A but
with zero steady state error for step inputs
• Check condition of angle and if satisfied
• Calculate K via condition of magnitude
• Problem-1
• Type-0 System: Finite steady-state error
• Solution-1
• Increasing the “Type” open loop Transfer function
• Integral Controller

Slide 9
Improving Steady-State Error Performance
• Problem -1
• Type-0 System: Finite steady-state error
• Solution-1
• Increasing the “Type” open loop Transfer function
• Integral Controller
• Problem-2
• A is not anymore part of the RL
• Angle contribution of the controller pole 𝜃
• Solution-2
• We need to compensate the effect of the pole by adding
zero (i.e. phase) to the system,
• PI controller!
𝐾(𝑠 + 𝑎)
𝐶 𝑠 =
𝑠
Slide 10
Improving Steady-State Error Performance
Design approach of PI Controller
𝐾(𝑠 + 𝑎)
𝐶 𝑠 =
𝑠
• Method-1 (Text-Book PI design approach) :
• Add a zero close to the pole at the origin
• Zero location can be tuned to cause RL to go through
location of previous poles
• Transient Response almost unaffected!.
• Steady State Improved.
• Method-2:
• Define a desired closed loop pole
• Use the conditions of Angle and Magnitude to tune the PI
controller

Slide 11
Improving Steady-State Error Performance
CAUTION!
• We are adding zeros to the open loop transfer function!
• Open loop Transfer function
𝑁 𝑠
𝐺 𝑠 =𝐺 𝑠 𝐺 𝑠 𝐺 𝑠 ≜
𝐷 𝑠
• Closed loop Transfer function
𝐺(𝑠) 𝑁(𝑠)
𝑇 𝑠 = =
1 + 𝐺(𝑠) 𝑁(𝑠) + 𝐷(𝑠)
• Zeros in the feedforward path are also closed loop zeros!
• Remember the effect of zeros in the dominant area!
IMPORTANT!
• The controller might assign poles to desired locations but you
might NOT satisfy time domain requirements such as Settling
Time, Overshoot, etc…
• Simulate the system to be sure all requirements have been met. Slide 12
• Redesign if simulation shows that requirements have not been met.
Example-1
It is desired to design a controller for the following open-loop TF
1
𝐺 𝑠 =
(𝑠 + 1)(𝑠 + 2)(𝑠 + 10)
Design requirements
• A closed loop pair of poles with damping ratio value of 𝜁 = 0.174
• Zero steady state error for step inputs
Let us first define 𝐶 𝑠 = 𝐾

Slide 13
Example-1

• Now, via the RL, we can calculate that the

damping ratio is 𝜁 = 0.174 for 𝐾 = 164.6
• The problem is obviously the steady state
error.
• We need a PI controller!
𝐾(𝑠 + 𝑎)
𝐶 𝑠 =
𝑠

Slide 14
Example-1
𝐾(𝑠 + 𝑎)
𝐶 𝑠 =
𝑠
• In the design of the PI, let us use Method-1,
• We simply set the zero close enough to the
pole, 𝑎 = 0.1
• Then, by using the condition of magnitude,
we can obtain 𝐾 = 158.2
• Both design requirements are satisfied now
• The damping ratio is 𝜁 = 0.174
• Zero steady state error for step inputs

Slide 15
Example-1
• What about the effect of the extra pole and
zero on the system response?
• 𝑝 = −0.0902, 𝑧 = −0.1
• They are in the dominant region but they will
cancel out their effect
• (Almost) Pole-zero cancellation
• Calculate the residue of the pole 𝑝 and
you will observe that it is almost zero
• Pole-zero cancellation
• Means that you are literally killing the
dynamics of the system
• The magnitude of the control signal will be
high!
• Might not be feasible for real-time applications

Slide 16
Example-2
It is desired to design a PI controller for the following
open-loop TF
0.25
𝐺 𝑠 =
𝑠 + 0.1
Design requirements
 Settling Time 𝑇 = 10
 Zero steady state error for step inputs
• We can calculate from
4
𝑇 = = 10 → 𝜁𝑤 = 0.4
𝜁𝑤
Thus, the real part of the pole must be −0.4.
• Let us define
𝑠 + 0.01
𝐶 𝑠 =𝐾
𝑠

Slide 17
Example-2
• Let us define
𝑠 + 0.01
𝐶 𝑠 =𝐾
𝑠
• Now, if we employ the condition of magnitude
0.25 𝑠 + 0.01 1
𝐺 𝑠 𝐶 (𝑠) = =
𝑠 + 0.1 𝑠 𝐾
𝑑 𝑑 0.3 ∗ 0.4
𝐾= = ≅ 1.231
𝑑 0.25 0.29 ∗ 0.25

• What about the other pole?

• What about the zero?

Slide 18
Improving Transient State Performance
• Shaping the Root-Locus
• D, PD, and PID controllers might be a solution
• PD controller
• Add a zero (pure derivative) to the forward path TF
𝐶 𝑠 = 𝐾(𝑠 + 𝑏)
• Design Goal
• Shorter Settling Time, Reduction of OS%...
• Design Approach
• Condition of Angle and Magnitude
• Warning: differentiation is a noisy process
• Level of noise is low, but the frequency is high compared to the signal
• Large, unwanted signals

Slide 19
Example-3
For the feedforward transfer function in a unit feedback
control system
1
𝐺 𝑠 =
𝑠(𝑠 + 4)(𝑠 + 6)
Design a PD controller to yield with dominant poles are
𝑠 , = −3.613 + 𝑗6.193.
𝐶 𝑠 = 𝐾(𝑠 + 𝑏)
Employ the condition of angle
∠𝐺(𝑠) = (2𝑖 + 1)𝜋
• The angle contribution of poles for the desired pole
location: -275.6

Slide 20
Example-3
Employ the condition of angle
∠𝐺(𝑠) = (2𝑖 + 1)𝜋
• The angle contribution of poles for the desired pole
location: -275.6
• In order to achieve -180, the angle contribution of the
placed zero should be 95.6.
• From the figure,
6.193
= tan 180 − 95.6 → 𝑏 = 3.006
3.613 − 𝑏
Employ the condition of magnitude
1
𝐺(𝑠) =
𝐾
• We calculate
𝐾 = 47.45

Slide 21
Example-4
For the transfer function in a unit feedback control system
1
𝐺 𝑠 =
𝑠(𝑠 + 2)
Design a controller to yield
 Settling Time 𝑇 = 2, Damping ratio 𝜁 = 0.707
 Zero steady state error for step inputs
• We can calculate from
4
𝑇 = = 2 → 𝜁𝑤 = 2
𝜁𝑤
We know 0.707 = cos 𝜃 → 𝜃 = 45 and thus the desired poles are
𝑠 , = −2 ± 2𝑗
• Let us first try to design, since 𝐺 𝑠 is a Type-1 system
𝐶 𝑠 =𝐾
The angle contribution of poles for the desired pole location: -
225. Thus, we do NOT have a solution.
Slide 22
Example-4
• Let us now define a PD controller
𝐶 𝑠 = 𝐾(𝑠 + 𝑎)
The zero of the PD controller must contribute with 45
degrees. Thus, we can calculate
2
= tan 45 → 𝑎 = 4
𝑎−2
Now from the condition of magnitude,
 
2∗2 2
𝐾=   =2
2 2

Slide 23
Example-5 (Alternative way)
For the feedforward transfer function in a unit feedback control system
(𝑠 + 6)
𝐺 𝑠 =
(𝑠 + 2)(𝑠 + 3)(𝑠 + 5)
Design a PD controller to yield
• Settling Time 𝑇 = 0.86, Damping ratio 𝜁 = 0.707
Let us define the characteristic equation of the closed loop transfer function
𝐾(𝑠 + 𝑎)(𝑠 + 6)
𝑝 𝑠 =1+𝐶 𝑠 𝐺 𝑠 =1+ =0
(𝑠 + 2)(𝑠 + 3)(𝑠 + 5)
𝑝 𝑠 = 𝑠 + 10 + 𝐾 𝑠 + 31 + 6𝐾 + 𝐾𝑎 𝑠 + 30 + 6𝐾𝑎 = 0
Now, we define the desired characteristic equation
𝑝 𝑠 = (𝑠 + 2𝜁𝑤 𝑠 + 𝑤 )(𝑠 + 𝑝) = 0
𝑝 𝑠 = 𝑠 + 9.28 + 𝑝 𝑠 + 43.0722 + 9.28𝑝 𝑠 + 43.0722𝑝 = 0
Note that, we had to define a 3rd order 𝑝 𝑠 since the order of 𝑝 𝑠 is 3. Here, 𝑝 represents a free
pole location

Slide 24
Example-5 (Alternative way)
Now, we equalize the coefficients of 𝑝 𝑠 and 𝑝 𝑠 ,

𝑝 𝑠 = 𝑠 + 10 + 𝐾 𝑠 + 31 + 6𝐾 + 𝐾𝑎 𝑠 + 30 + 6𝐾𝑎 = 0

𝑝 𝑠 = 𝑠 + 9.28 + 𝑝 𝑠 + 43.0722 + 9.28𝑝 𝑠 + 43.0722𝑝 = 0

We obtain the following solution

𝐾 = 4.767
𝑝 = 5.487
𝑎 = 7.2141

Slide 25
Improving Transient State and Steady State Performance
• Shaping the Root-Locus
• PID controllers might be a solution
𝐾 𝐾 𝑠 +𝐾 𝑠+𝐾 𝐾(𝑠 + 𝑎)(𝑠 + 𝑏)
𝐶 𝑠 =𝐾 + +𝐾 𝑠 = =
𝑠 𝑠 𝑠
• Design Goals
• Steady-State Error, Shorter Settling Time, Reduction of OS%...
• Design Approach
1. Improve the transient response by designing PD controller
2. Simulate the system to be sure all requirements have been met.
3. Redesign if simulation shows that requirements have not been met.
4. Improve the steady-state response by designing a PI controller for the PD+𝐺 𝑠
5. Simulate the system to be sure all requirements have been met.
6. Redesign if simulation shows that requirements have not been met.
7. Combine the PI and PD controllers to end up with the PID controller
Slide 26
Example-6
For the feedforward transfer function in a unit feedback
control system
(𝑠 + 8)
𝐺 𝑠 =
(𝑠 + 3)(𝑠 + 6)(𝑠 + 10)
Design a PID controller to yield with dominant poles are
𝑠 , = −8.13 ± 𝑗15.87.
𝐾(𝑠 + 𝑎)(𝑠 + 𝑏)
𝐶 𝑠 =
𝑠
1) Design of the PD controller 𝐶 𝑠 = 𝐾(𝑠 + 𝑎)
Employ the condition of angle
∠𝐺(𝑠) = (2𝑖 + 1)𝜋 −𝑎
• The angle contribution of poles for the desired pole
location: 198.37
• Therefore the angle of the zero must contribute with
180 − 198.37 = 18.37
Slide 27
Example-6
1) Design of the PD controller 𝐶 𝑠 = 𝐾(𝑠 + 𝑎)
Employ the condition of angle
∠𝐺(𝑠) = (2𝑖 + 1)𝜋
• The angle contribution of poles for the desired pole location: 198.37.
Therefore the angle of the zero must contribute with 180 − 198.37 = 18.37.
• We can now calculate
15.87
= tan 18.37 → 𝑎 = 55.92
𝑎 − 8.13

−𝑎
Slide 28
Example-6
2) Design of the PI controller 𝐶 𝑠 = 𝐾(𝑠 + 𝑏)/𝑠
Let us set
𝐶 𝑠 = 𝐾(𝑠 + 0.5)/𝑠
The resulting RL and from the magnitude condition 𝐾 = 4.6
3) Combine the PD and PI to define the PID
(𝑠 + 55.92 )(𝑠 + 0.5)
𝐶 𝑠 = 4.6
𝑠

Slide 29
Example-7
For the feedforward transfer function in a unit feedback control system
10(𝑠 + 12)
𝐺 𝑠 =
𝑠+4 𝑠+6
Design a PID controller to yield
 Natural Frequency 𝑤 = 4, Damping ratio 𝜁 = 1
 For ramp inputs 𝑒 = 1/3
• Let us firstly define a PID controller
𝐾 𝐾 𝑠 +𝐾 𝑠+𝐾
𝐶 𝑠 =𝐾 + +𝐾 𝑠 =
𝑠 𝑠
• For ramp inputs, we know that
𝑒 = 1 ⁄3 → 𝐾 = 3
𝐾 = lim 𝑠𝐶 𝑠 𝐺(𝑠)
→
𝐾 𝑠 + 𝐾 𝑠 + 𝐾 10(𝑠 + 12)
𝐾 = lim 𝑠
→ 𝑠 𝑠+4 𝑠+6
𝐾 = 0.6
Slide 30
Example-7
• Let us define the characteristic equation
𝐾 𝑠 + 𝐾 𝑠 + 0.6 10(𝑠 + 12)
𝑝 𝑠 =1+𝐶 𝑠 𝐺 𝑠 =1+ =0
𝑠 𝑠+4 𝑠+6
𝑝 𝑠 = (10𝐾 + 1)𝑠 + 120𝐾 + 10𝐾 + 10 𝑠 + 6 + 120𝐾 + 24 𝑠 + 72 = 0
• Now, we define the desired characteristic equation
𝑝 𝑠 = (𝑠 + 2𝜁𝑤 𝑠 + 𝑤 )(𝑎𝑠 + 𝑏) = 0
𝑝 𝑠 = 𝑎𝑠 + 8𝑎 + 𝑏 𝑠 + 16𝑎 + 𝑏 𝑠 + 16𝑏 = 0
Note that, we had to define a 3rd order 𝑝 𝑠 since the order of 𝑝 𝑠 is 3. Here, −𝑏/𝑎 represents a
free pole location
• We equalize the coefficients of 𝑝 𝑠 and 𝑝 𝑠 , and obtain the following solution
𝐾 = 0.0125
𝐾 = 0.2
𝑎 = 1.125
𝑏 = 4.5

Slide 31