Ch9.

Design via Root Locus

Design Via Root Locus 1

 Learning Objectives
• Use the root locus to design cascade
compensators to improve the steady-state error
(Sections 9.1–9.2)
• Use the root locus to design cascade
compensators to improve the transient
response (Section 9.3)
• Use the root locus to design cascade
compensators to improve both the steady-state
error and the transient response (Section 9.4)
• Realize the designed compensators physically
(Section 9.6) Design Via Root Locus 2
A Quick Review of Second Order
Transient Response

Design Via Root Locus 3

Underdamped Second-Order Systems (cont.)
• We now find the relationship between the previously introduced
parameters and the pole plot.

σd is the ωd is the imaginary

magnitude of the part of the pole and is
real part of the called the damped
pole and is the frequency of
exponential oscillation
decay frequency.

Important Note:
1. the natural frequency, ωn, is the distance from the origin to the pole.

Design Via Root Locus 4

Underdamped Second-Order Systems (cont.)
• We can relate Tp and Ts to the locations of the poles.

Tp is inversely Ts is inversely
proportional proportional
to the to the real
imaginary part of the
part of the pole.
pole.

Design Via Root Locus 5

Underdamped Second-Order Systems (cont.)

When wd is fixed, When sd is fixed,

Tp does not Ts does not
change. change.

When is fixed,
%OS does not
change.

Design Via Root Locus 6

 Introduction
• We have seen that adjusting the gain factor K can improve the
system performance, e.g., the percentage overshoot.
• Question: can we improve any parameter by simply adjusting
the gain?
• For example, take a look at the root locus below. Can we
achieve better settling time while maintaining the ideal
percentage overshoot by only adjusting the gain?
The answer is No.
Changing the gain can
only move along the
root locus, and B is
not on the root locus!
better in terms
of settling time.
Design Via Root Locus 7
 Introduction (cont.)
• It is clear that in order to have closed-loop gain at point B, we
will have to modify the root locus. This can be done via
compensation in the forward path.

• Ideal compensators are the ones that use pure integration or

pure differentiation. They must be implemented using active
devices (e.g., amplifiers).
• Non-ideal compensators do not use pure integration and pure
differentiation and can be implemented using passive elements
(e.g., resistors and capacitors)
• Ideal compensators are more expensive and require additional
power sources. Non-ideal compensators are less expensive, but
Design Via Root Locus 8
they cannot yield zero steady-state error.
 Introduction (cont.)
• The compensator design essentially has three basic building blocks:
– Proportional controller: feed the error forward to the plant
– Integral controller: feed the integral of the error to the plant
– Derivative controller: feed the derivative of the error to the
plant
• The compensator or controller can be used to:
– Improve steady-state error only: PI (Proportional-plus-Integral)
controller (ideal) or lag compensator (non-ideal)
– Improve transient response only: PD (Proportional-plus-
Derivative) controller (ideal) or lead compensator (non-ideal)
– Improve both: PID (Proportional-plus-Integral-plus-Derivative)
controller (ideal) or lag-lead compensator (non-ideal)
Note: The terms lag and lead come from frequency response.
Design Via Root Locus 9
 Improving Stead-state Error Using PI
• Our goal is to improve steady-state error while maintaining roughly
the same transient response performance
• To improve the steady-state error of the system on the left, it’s
obvious that we should add a compensator pole at the origin.

Steady-state error is improved, but the

Transient response transient response could be worst: closed-
okay at closed-loop loop pole A is no longer on the root locus.
pole A

solution?

Design Via Root Locus 10

Improving Stead-state Error Using PI (cont.)
• The solution is to add a compensator zero near the newly added
compensator pole at the origin.
When the pole and zero are really
close, then the angular
contribution of the compensator
zero and compensator pole cancel
out, point A is still on the root
locus, and the system type has
been increased.

Such a compensator with a pole at

the origin and a zero close to the
pole is known as an ideal integral
compensator.
Design Via Root Locus 11
Improving Stead-state Error Using PI (cont.)
• Example: Let’s now use an example to see the effect of an ideal
integral compensator (shown in part b) over the original system
(shown in part a) while maintaining a damping ratio of 0.174.

Design Via Root Locus 12

Improving Stead-state Error Using PI (cont.)
• We will first sketch the root locus of system (a) and notice that
1. We use the radial line to locate the
closed-loop dominant poles
2. We next find the gain value K=164.6.
3. We then use the gain factor to identify
the additional closed-loop pole on the real axis

Design Via Root Locus 13

Improving Stead-state Error Using PI (cont.)
• We next sketch the root locus of system (b) with the ideal integral
compensator, and notice that
1. The root locus is changed slightly
2. A new portion of the root locus appears
between the newly added open-loop
pole and zero
3. As a result, the fourth closed-loop pole can be
Identified using the new gain value

Design Via Root Locus 14

Improving Stead-state Error Using PI (cont.)
• The system is now type 1, and its steady-state error is now zero!
• The figure below compares the output of the original system and
the compensated system.

Almost identical Much better

transient response steady-state error

Design Via Root Locus 15

Improving Stead-state Error Using PI (cont.)
• The ideal integral compensator is also known as the PI control. The
proportional and integral components can be drawn as two
separate blocks in parallel:

The value of the

zero can be adjusted
by varying K2/K1.

How to implement PI and other controllers will be discussed later

in the course. Stay tuned.
Design Via Root Locus 16
 Improving Transient Response Using PD
• We now focus on improving transient response. In general, we
would like to reduce Ts and Tp.
• The system’s desired transient response can be selected by
choosing an appropriate closed-loop pole location on the s-plane.
We have two cases:
– Case 1: if this location is on the root locus, then we only need to
modify the gain factor to reach this point.
– Case 2: if this location is not on the root locus, then we will have
to use a compensator to reshape the root locus so that the new
root locus goes through the selected closed-loop pole location.
• The second case above can be done using a proportional-plus-
derivative (PD) controller, which is also known as ideal derivative
compensation:
1. a differentiator plus a pure gain
Gc(s)=s+zc 2. adding a zero to the forward path
Design Via Root Locus 17
 Improving Transient Response Using PD
• Let’s now take a look at an example and see how the PD
controller can speed up the transient response.
• The figure below shows the uncompensated root locus:

We’d like to maintain

%overshoot and at
the same time,
reduce the settling
time.

Design Via Root Locus 18

 Improving Transient Response Using PD

Zc=2 Zc=3 Zc=4

1. In all three cases, the new closed-loop pole on the radial line has
more negative real part, which means that the settling time is now
shorter
2. In all three cases, the new closed-loop pole on the radial line has
larger imaginary part, which means that the peak time is now shorter
3. Finally, the closer the new zero to the dominant poles, the greater the
Design Via Root Locus 19
effect.
 Improving Transient Response Using PD
We now look at the time responses of the uncompensated and
compensated systems:

Reduced
settling time
and peak time
(as expected)

Percentage Reduced steady-state

overshoot error?! (Note: it happens
remains the to be this way in this
same example. In general, PD
control may not improve
steady-state error.)
Design Via Root Locus 20
 Improving Transient Response Using PD
Example: Design an ideal derivative compensator for the system below
to yield a 16% overshoot, with a threefold reduction in settling time.
Solution:
To improve the system’s performance,
we first need to understand the performance
of the uncompensated system.
At 16% overshoot, the equivalent is 0.504. We have the root locus of
the uncompensated system below:

Design Via Root Locus 21

 Improving Transient Response Using PD
Solution (cont.): It’s now clear that our new settling time is
3.32/3=1.107
and

Our next step is to find out the position of the

zero.

Recall that the sum of angles to the new closed-

loop pole from all the open-loop zeros and
poles must be odd multiples of 180 degrees.

We first find out the sum of angles without the

compensator zero:
-275.60
Design Via Root Locus 22
 Improving Transient Response Using PD
Solution (cont.): Because
We figure out the angle from the compensator zero to the new closed-
loop pole is 95.60:

Use trigonometry, we
can calculate the precise
location of the
compensator zero:

Design Via Root Locus 23

 Improving Transient Response Using PD
Solution (cont.): The root locus of the compensated system is shown
below: 1. The compensated system’s
settling time is about 1/3 of
that of the original system

2. Slightly reduced %
overshoot (due to the
effect brought by the
compensator zero)

Actual system responses

Design Via Root Locus 24
 Improving Transient Response Using PD
Implementing the PD controller: The PD controller can be implemented
using two parallel branches.

The transfer function of the compensator is:

where K1/K2 is chosen to equal the negative of the compensator zero,

and K2 is chosen to contribute to the required loop-gain value.

Design Via Root Locus 25

 PID controllers
• Proportional-plus-Integral-plus-Derivative (PID) controllers
combine both PI and PD controllers to improve both transient
response and steady-state errors

Two zeros

One pole
at origin
One zero and the pole at origin can be designed as an ideal
integral compensator; the other zero can be designed as an ideal
derivative compensator. Design Via Root Locus 26
 PID controllers (cont.)
• PID controllers can be designed using two different orders:
– Design for transient response, followed by design for steady-
state error (may decrease transient response speed after design
for steady-state error is done)
– Design for steady-state error, followed by design for transient
response (may increase or decrease steady-state error after
design for transient response is done)
• In our discussion, we will perform design for transient response first,
and then steady-state error.
• Note that PID controllers utilize active components. It is possible to
approximate the effect of PID controller using passive elements
only. For example, one can design a lag-lead compensator.

Design Via Root Locus 27

 PID controllers (cont.)
• We now use an example to show how to design a PID controller.
Example: For the system below, design a PID controller so that the
system can operate with a peak time that is two-thirds of that of the
uncompensated system at 20% overshoot and with zero steady-state
error for a step input.

Solution:
Step 1) We start from the root locus
of the uncompensated system with
20% overshoot and find out the
locations of the closed-loop poles.
We then calculate the peak time:

Design Via Root Locus 28

 PID controllers (cont.)
Solution (cont.): Step 2) In order to achieve two-thirds peak time in
the compensated system, we need to have:

To complete the PD control design,

we will need to calculate the
compensator zero next. Because the
sum of the angles from all existing
poles and zeros to the compensated
closed-loop pole is -198.370, we
conclude that the angle from the
compensator zero to the closed-loop
pole must be 18.370:

Design Via Root Locus 29

 PID controllers (cont.)
Solution (cont.): Step 3) We now design PI controller to reduce the
steady-state error. In particular, we let -0.5 be the PI compensator
zero (recall that it has to be close to the origin!).

The root locus of the PID controller:

The new gain and new
Compensator zero location of the closed-loop
of PD controller pole of the compensated
system.
Compensator
zero and pole of
of PI controller

Design Via Root Locus 30

 PID controllers (cont.)
Solution (cont.): Step 4) We now come up with the transfer function
of the PID controller as well as the values of K1, K2, and K3.

We obtain:

Design Via Root Locus 31

 PID controllers (cont.)
Solution (cont.): Step 5) Finally, we must run simulation to verify the
design (because second-order approximation no longer holds).
Reduces the steady-state error to 0, but also reduces the speed of
transient response slightly. If this is unacceptable, one can either redesign
PD or move the compensator zero of PI further away from the origin to
yield a faster response.

Improves both transient

response and steady-
state error

Design Via Root Locus 32

Physical Realization of Active Compensation
• PI, PD, and PID controllers can be implemented
using an inverting op-amp, which was discussed
in Ch.2:

Transfer function

Design Via Root Locus 33

Physical Realization of Active Compensation

Design Via Root Locus 34

Physical Realization of Active Compensation
Example: Implement the following PID control using op-amps:

Solution: we rewrite the compensator into the following form:

From the table in the previous slide, we get: .

It yields: We have four unknowns and three equations. Let

C1=0.1uF, and solve for the rest:

R1=357.65k Ohm
R2=178.891k Ohm
C1=5.59mF
Design Via Root Locus 35
PID Tuning: the Ziegler-Nichols Methods
When the mathematical model of the plant is unknown, the
parameters of PID controllers must be tuned. This process is known as
PID tuning. Ziegler and Nichols developed two methods for PID
tuning in the 1940s. The Z-N methods are rules of thumb and usually
provide a satisfactory control, instead of the optimal control.
Method 1: When the step response of the plant is S-shaped.

1
𝐺 𝐶 ( 𝑠 ) =𝐾 𝑝 + 𝐾 𝑖 +𝐾𝑑 𝑠
𝑠
Controller Kp Ki Kd
P T/L 0 0
PI 0.9T/L 0.27T/L2 0
PID 1.2T/L 0.6T/L2 0.6T
Design Via Root Locus 36
 PID Tuning: the Ziegler-Nichols Methods
(cont.)
Method 2: When the step response of the plant is similar to the
underdamped 2nd order response.

Step 1: Use a proportional controller and keep increasing Kp until it

reaches a critical value Kcr at which the output exhibits sustained
oscillations.

Step 2: Measure Pcr ,

the period of oscillation.

Design Via Root Locus 37

 PID Tuning: the Ziegler-Nichols Methods
(cont.)
Method 2: When the step response of the plant is similar to the
underdamped 2nd order response.

Step 3: Use Kcr and Pcr to determine the PID parameters:

1
𝐺 𝐶 ( 𝑠 ) =𝐾 𝑝 + 𝐾 𝑖 +𝐾𝑑 𝑠
𝑠

Controller Kp Ki Kd

P 0.5Kcr 0 0

PI 0.45Kcr 0.54Kcr/Pcr 0

PID 0.6Kcr 1.2Kcr/Pcr 0.075KcrPcr

Design Via Root Locus 38