MODERN CONTROL

SYSTEMS

VNU - UNIVERSITY of ENGINEERING & TECHNOLOGY

LECTURE 2: THE ROOT LOCUS METHOD

CONTENTS

> Introduction
> Root locus
▪ Basic concepts
▪ Properties
▪ Gain only controller
> Design by the root locus method
▪ Proportional-Integration (PI) controller
▪ Proportional-Derivative (PD) controller
▪ Proportional-Integration-Derivative (PID) controller
Introduction

> Performance of a control system depends on its pole locations.

▪ When adjusting one or more system parameters → pole positions changed!
𝐾𝐺(𝑠)
▪ E.g., 𝑇 𝑠 = : how to quickly determine the system’s poles as 𝐾 varies?
1+𝐾𝐺(𝑠)

▪ Need to sketch the locus of poles as 𝐾 varies without factoring denominator.

The CameraMan System

CameraMan® Presenter Camera System:

▪ Can automatically follows a subject.

Simplified block diagram for CameraMan®

▪ Two unknown parameters 𝐾1 , 𝐾2

Transfer function of the simplified model:

▪ Performance directly related to pole locations
▪ Pole locations vary with the value of 𝐾 = 𝐾1 𝐾2
CameraMan: Pole Locations
The Root Locus
> Definition: the path of closed-loop poles as the gain is varied.
▪ What information is gained from the root locus?

Complex poles for 𝐾 > 25

Multiple, real poles for 𝐾 = 25
▪ Underdamped portion of the root locus
▪ Critically damped
▪ Settling time remains the same
▪ As 𝐾 increases, %𝑂𝑆 increases, 𝑇𝑝 reduces
▪ Never crosses over into RHP → always stable

Separate real poles for 𝐾 < 25

▪ Overdamped portion of the root locus
Properties of Root Locus (1)
𝐾𝐺(𝑠)
> Consider a general control system 𝑇 𝑠 =
1+𝐾𝐺 𝑠 𝐻(𝑠)
▪ 𝑝 is a pole only if 𝐾𝐺 𝑠 𝐻(𝑠)ȁ𝑠=𝑝 = −1 = 1∠(2𝑘 + 1)𝜋.
▪ Necessary condition: if 𝑠 makes ∠𝐺 𝑠 𝐻 𝑠 = (2𝑘 + 1)𝜋, it is certainly a pole.
▪ [Appendix M] It is proved that: ∠𝐺 𝑠 𝐻 𝑠 = σ zero angles − σ pole angles.
▪ Example:

Is -2 + 𝑗3 a closed-loop pole?

-2 + 𝑗3 is not a closed-loop pole.

𝑠=-2 + 𝑗3: 56.31° + 71.57° − 90° − 108.43° = −70.55°

∠𝐺 𝑠 𝐻(𝑠) = 𝜃1 + 𝜃2 − 𝜃3 − 𝜃4

(𝑠 + 3)(𝑠 + 4)
𝐺 𝑠 𝐻(𝑠) =
(𝑠 + 1)(𝑠 + 2)
Properties of Root Locus (2)

> If 𝑝 is a closed-loop pole, then 𝐾 =

1
ฬ .
𝐺(𝑠)𝐻(𝑠) 𝑠=𝑝
1 ς pole lengths
▪ [Appendix M] It is proved that: =ς .
𝐺(𝑠)𝐻(𝑠) zero lengths

▪ Example:

2
Is -2 + 𝑗 a closed-loop pole?
2
2
𝑗
2
2
𝐿3 𝐿4 1.22 2
𝐾= = 2 = 0.33 -2 + 𝑗 is a closed-loop pole
𝐿1 𝐿2 2.12 1.22 2

2
𝑠= -2 + 𝑗 : 19.47° + 35.26° − 90° − 144.74° = −180°
2

∠𝐺 𝑠 𝐻(𝑠) = 𝜃1 + 𝜃2 − 𝜃3 − 𝜃4

(𝑠 + 3)(𝑠 + 4)
𝐺 𝑠 𝐻(𝑠) =
(𝑠 + 1)(𝑠 + 2)
Properties of Root Locus (3)
∞
∞
∞

∞ X X X X ∞

∞
∞
∞

> A transfer function can have infinite poles and zeros.

▪ If the function approaches infinity as 𝑠 → ∞, it has infinite pole(s).
▪ If the function approaches zero as 𝑠 → ∞, it has infinite zero(s).
𝐾(𝑠+3)
> Example: 𝐾𝐺 𝑠 𝐻 𝑠 =
𝑠(𝑠+1)(𝑠+2)(𝑠+4)
▪ There’re 4 finite poles at 0, −1, −2, −4 and 1 finite zero at −3.
𝐾
▪ There’re 3 zeros at infinity because lim 𝐾𝐺 𝑠 𝐻 𝑠 = = 0.
𝑠→∞ 𝑠∙𝑠∙𝑠
▪ The transfer function has an equal number of poles and zeros. Where are
the infinite zeros on the complex plane?
Simulation using Octave
% Drawing root locus for KG(s)H(s)=K(s+3)/s(s+1)(s+2)(s+4)
numg=[1 3]; %Define numerator
deng=poly([0 -1 -2 -4]); %Define denominator
G=tf(numg,deng); %Form the open-loop transfer function
rlocus(G); %Drawing the root locus
Properties of Root Locus (4)

> Five rules that help sketching a root locus:

1. #branches (the path that one pole traverses) = #closed-loop poles.
2. The root locus is symmetrical about the real axis.
3. On the real axis, the root locus exists to the left of an odd number of real-
axis, finite open-loop poles and/or finite open-loop zeros.
4. The root locus begins at the finite and infinite poles of 𝐺(𝑠)𝐻(𝑠) and ends
at the finite and infinite zeros of 𝐺(𝑠)𝐻(𝑠).
5. The root locus approaches straight lines asymptotes as the locus
approaches infinity. The asymptote equation is given as follows
σ finite poles − σ finite zeros
𝜎𝑎 = #finite poles −#finite zeros

2𝑘+1 𝜋
𝜃𝑎 = #finite poles −#finite zeros

where 𝑘 = 0, ±1, ±2, …, 𝜎𝑎 is the real-axis intercept, and 𝜃𝑎 is the angle given
in radians with respect to the positive extension of the real axis.
Applying Properties of Root Locus: Example 1

Begins at finite poles, ends

at finite zeros (Rule 4)

Two poles → 2 branches (Rule 1)

Symmetric about real axis (Rule 2)

2 real axis segments (Rule 3)

Applying Properties of Root Locus: Example 2

𝑗𝜔

Four poles → 4 branches (Rule 1) ∞

Begin at 4 finite poles, ends at 1
finite & 3 infinite zeros (Rule 4)
Symmetric about real axis (Rule 2)

∞ X X X X 𝜎

There’re 3 infinite zeros → 3 asymptotes.

−4−2−1−(−3) 4
3 real axis segments (Rule 3) ▪ 𝜎-axis intercept: 𝜎𝑎 = = − 3.
4−1
(2𝑘+1)𝜋 𝜋 5𝜋
▪ Angles: 𝜃𝑎 = 4−1
= 3
, 𝜋, 3 .

∞
Design via Gain Adjustment
> Example: Design the value of gain to yield 1.52% overshoot.
▪ Also estimate the settling time, peak time, and steady-state error.

Only valid for a system with 2 closed-loop

complex poles and no closed-loop zeros!

1.52% overshoot corresponds to 𝜁 = 0.8

The 𝜁=0.8 line crosses the locus at 3 different

values of 𝐾, which one should we pick?
Design via Gain Adjustment (cont.)
> Valid second-order assumptions:
▪ Higher-order poles are much farther into the LHP than the dominant 2nd-
order pair of poles.
▪ Closed-loop zeros are far removed from the dominant 2nd-order pole pair.
▪ Closed-loop zeros near the dominant second-order pair of poles are in close
proximity of higher-order closed-loop poles.
Drawbacks of Gain Adjustment Design

If we want to design for both %𝑂𝑆 and 𝑇𝑆

The desired transient response is represented

by point 𝐵, which is not on the root locus.

We can only obtain the 𝑇𝑆 represented by point

𝐴 via simple gain adjustment.

Counterproductive solution: replace the plant

with the one whose root locus intersects 𝐵.
Integration-only Compensator

> Adds an open-loop pole at the origin in the forward path.

▪ Increases the system type → reduces steady-state error 𝑒step ∞ to 0.
▪ Transient response fails: 𝐴 is no longer on the root locus.
Proportional-plus-Integrator (PI) Compensator
𝐾2
𝐾2 𝐾1 𝑠 + 𝐾1
𝐺𝑐 𝑠 = 𝐾1 + =
𝑠 𝑠

> Adding a pole at the origin and a zero.

▪ Drives the steady-state to zero: 𝑒step ∞ = 0.
▪ Transient response unaffected.
▪ Expensive implementation as it requires active network elements.
PI Compensator Design: Example
The uncompensated system operates with 𝜁 = 0.174
▪ 𝐾 = 164.6 → 𝐾𝑝 = 8.23 → 𝑒 ∞ = 0.108.
▪ 3rd pole is far into LHP → valid 2nd-order approx.

Effects of the PI compensator?

Far removed pole Valid pole-zero cancelation

Transient response unaffected.

Steady-state error reduces to 0.
PI Compensator Design: Example (cont.)
Proportional-plus-Derivative (PD) Compensator
Introducing a new zero to the forward path
𝐾1
𝐺𝑐 𝑠 = 𝐾1 + 𝐾2 𝑠 = 𝐾2 𝑠 +
𝐾2

> PD compensator can quicken the response:

▪ 𝐾2 must be chosen to achieve the desirable %𝑂𝑆.
𝐾1
▪ 𝐾1 must be chosen so that the compensating zero at − yields shorter 𝑇𝑠 /𝑇𝑝 .
𝐾2

> Drawbacks:
▪ requires active elements for implementation.
▪ can amplify high frequency noise.
PD Compensator Design: Example
1
𝐾𝐺𝑐 (𝑠)
(𝑠 + 1)(𝑠 + 2)(𝑠 + 5)
PD Compensator Design: Example (cont.)
PID Compensator
> A PID compensator = cascade of a PI and a PD compensators.
𝐾2
𝐺𝐶 𝑠 = 𝐾1 + + 𝐾3 𝑠 introduces 1 pole at the origin
𝑠
𝐾 𝐾
𝐾3 𝑠 2 + 1 𝑠+ 2 and 2 zeros anywhere
𝐾3 𝐾3
=
𝑠

> Two design approaches:

▪ Design for steady-state error first, followed by design for transient response.
– Normally results in overdesigned or deteriorated 𝑒𝑠𝑠 (∞).
▪ Design for transient response first, followed by design for steady-state error.
– Normally results in a slight decrease in the speed of the response.
PID Compensator: Example
Design a PID controller to achieve:
2
▪ 𝑇𝑝 is that of the uncompensated @ 20% 𝑂𝑆
3
▪ Zero steady-sate error for a step input.
Evaluate the performance of the uncompensated system

Design a PD controller to meet the transient

response specifications:
▪ 𝐼𝑚(𝑝dom ) = 15.87, 𝑅𝑒 𝑝dom = −8.13.
▪ The angular contribution required from 𝑧𝑐 is
198.37 − 180 = 18.37° → 𝑧𝑐 = 55.92.

Step 1
Step 2
 PID Compensator: Example (cont.)
Step 4
Design the PI controller for the required steady-state error
𝑠+0.5
▪ Choose 𝑧𝑐 = −0.5 → 𝐺𝑃𝐼 𝑠 = . Step 3
𝑠
4.6 𝑠+55.92 𝑠+0.5
▪ 𝐺𝑃𝐼𝐷 𝑠 = → 𝐾1 ≈ 260, 𝐾2 ≈ 129, 𝐾3 ≈ 4.6. Simulate the PD-compensated system
𝑠
▪ Redesign if needed
PID Compensator: Example (cont.)

Step 5
Simulate the PID-compensated system
▪ Redesign if needed
Summary

Key takeaways

> As a system parameter varies, root locus provides:

▪ graphical representation of the variation of system poles
▪ information about the stability & transient response
> Compensators can be designed by root locus method
▪ Reshape the root locus to go through desired pole locations
▪ Always check the validity of 2nd-order approximations
▪ Implementation: PI, PD, PID compensators
NEXT LECTURE

[Presentation] State feedback control design

> References: Richard Dorf, “Modern Control Systems”
▪ Chapter 11, Sections 1-3
▪ Whatever you find useful

> Pre-class
▪ Assignment of roles & responsibilities
▪ Collaborative research
▪ Presentation scripting

> In-class
▪ Present in turn
▪ Discussion