Control System Toolbox™ 9

Getting Started Guide

How to Contact MathWorks

www.mathworks.com Web
comp.soft-sys.matlab Newsgroup
www.mathworks.com/contact_TS.html Technical Support

[email protected] Product enhancement suggestions

[email protected] Bug reports
[email protected] Documentation error reports
[email protected] Order status, license renewals, passcodes
[email protected] Sales, pricing, and general information

508-647-7000 (Phone)

508-647-7001 (Fax)

The MathWorks, Inc.

3 Apple Hill Drive
Natick, MA 01760-2098
For contact information about worldwide offices, see the MathWorks Web site.
Control System Toolbox™ Getting Started Guide
© COPYRIGHT 2000–2010 by The MathWorks, Inc.
The software described in this document is furnished under a license agreement. The software may be used
or copied only under the terms of the license agreement. No part of this manual may be photocopied or
reproduced in any form without prior written consent from The MathWorks, Inc.
FEDERAL ACQUISITION: This provision applies to all acquisitions of the Program and Documentation
by, for, or through the federal government of the United States. By accepting delivery of the Program
or Documentation, the government hereby agrees that this software or documentation qualifies as
commercial computer software or commercial computer software documentation as such terms are used
or defined in FAR 12.212, DFARS Part 227.72, and DFARS 252.227-7014. Accordingly, the terms and
conditions of this Agreement and only those rights specified in this Agreement, shall pertain to and govern
the use, modification, reproduction, release, performance, display, and disclosure of the Program and
Documentation by the federal government (or other entity acquiring for or through the federal government)
and shall supersede any conflicting contractual terms or conditions. If this License fails to meet the
government’s needs or is inconsistent in any respect with federal procurement law, the government agrees
to return the Program and Documentation, unused, to The MathWorks, Inc.

Trademarks
MATLAB and Simulink are registered trademarks of The MathWorks, Inc. See
www.mathworks.com/trademarks for a list of additional trademarks. Other product or brand
names may be trademarks or registered trademarks of their respective holders.
Patents
MathWorks products are protected by one or more U.S. patents. Please see
www.mathworks.com/patents for more information.
Revision History
November 2002 First printing New Version 5.0 (Release 12)
June 2001 Second printing Revised for Version 5.1 (Release 12.1)
July 2002 Online only Revised for Version 5.2 (Release 13)
June 2004 Online only Revised for Version 6.0 (Release 14)
March 2005 Online only Revised for Version 6.2 (Release 14SP2)
September 2005 Online only Revised for Version 6.2.1 (Release 14SP3)
March 2006 Online only Revised for Version 7.0 (Release 2006a)
September 2006 Online only Revised for Version 7.1 (Release 2006b)
March 2007 Online only Revised for Version 8.0 (Release 2007a)
September 2007 Online only Revised for Version 8.0.1 (Release 2007b)
March 2008 Online only Revised for Version 8.1 (Release 2008a)
October 2008 Third printing Revised for Version 8.2 (Release 2008b)
March 2009 Online only Revised for Version 8.3 (Release 2009a)
September 2009 Online only Revised for Version 8.4 (Release 2009b)
March 2010 Online only Revised for Version 8.5 (Release 2010a)
September 2010 Online only Revised for Version 9.0 (Release 2010b)
 Contents

Product Overview
1
Why Use This Toolbox? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-2

Using the Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

Related Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4

Building Models
2
Linear (LTI) Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
What Is a Plant? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
Linear Model Representations . . . . . . . . . . . . . . . . . . . . . . . 2-2
SISO Example: The DC Motor . . . . . . . . . . . . . . . . . . . . . . . 2-3
Building SISO Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-6
Constructing Discrete Time Systems . . . . . . . . . . . . . . . . . . 2-9
Adding Delays to Linear Models . . . . . . . . . . . . . . . . . . . . . 2-10
LTI Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-10

MIMO Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-13

MIMO Example: State-Space Model of Jet Transport
Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-13
Constructing MIMO Transfer Functions . . . . . . . . . . . . . . . 2-15
Accessing I/O Pairs in MIMO Systems . . . . . . . . . . . . . . . . 2-17

Arrays of Linear Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-18

Model Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-21

Interconnecting Linear Models . . . . . . . . . . . . . . . . . . . . . 2-23

v
 Arithmetic Operations for Interconnecting Models . . . . . . 2-23
Feedback Interconnections . . . . . . . . . . . . . . . . . . . . . . . . . . 2-24

Converting Between Continuous- and Discrete- Time

Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-26
Available Commands for Continuous/Discrete
Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-26
Available Methods for Continuous/Discrete Conversion . . 2-26
Digitizing the Discrete DC Motor Model . . . . . . . . . . . . . . . 2-26

Reducing Model Order . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-29

Model Order Reduction Commands . . . . . . . . . . . . . . . . . . . 2-29
Techniques for Reducing Model Order . . . . . . . . . . . . . . . . . 2-29
Example: Gasifier Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-30

Analyzing Models
3
Quick Start for Performing Linear Analysis Using the
LTI Viewer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-2

LTI Viewer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-7

Plot Types Available in the LTI Viewer . . . . . . . . . . . . . . . . 3-7
Example: Time and Frequency Responses of the DC
Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-8
Right-Click Menus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10
Displaying Response Characteristics on a Plot . . . . . . . . . . 3-11
Changing Plot Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14
Showing Multiple Response Types . . . . . . . . . . . . . . . . . . . . 3-16
Comparing Multiple Models . . . . . . . . . . . . . . . . . . . . . . . . . 3-17

Simulating Models with Arbitrary Inputs and Initial

Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-22
What is the Linear Simulation Tool? . . . . . . . . . . . . . . . . . . 3-22
Opening the Linear Simulation Tool . . . . . . . . . . . . . . . . . . 3-22
Working with the Linear Simulation Tool . . . . . . . . . . . . . . 3-23
Importing Input Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-26
Example: Loading Inputs from a Microsoft® Excel
Spreadsheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-28

vi Contents
 Example: Importing Inputs from the Workspace . . . . . . . . 3-29
Designing Input Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-33
Specifying Initial Conditions . . . . . . . . . . . . . . . . . . . . . . . . 3-35

Functions for Time and Frequency Response . . . . . . . . 3-37

When to Use Functions for Time and Frequency
Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-37
Time and Frequency Response Functions . . . . . . . . . . . . . . 3-37
Plotting MIMO Model Responses . . . . . . . . . . . . . . . . . . . . . 3-40
Data Markers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-42
Plotting and Comparing Multiple Systems . . . . . . . . . . . . . 3-44
Creating Custom Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-47

Designing Compensators
4
Designing PID Controllers . . . . . . . . . . . . . . . . . . . . . . . . . 4-2
Designing PID Controllers with the PID Tuner GUI . . . . . 4-2
Designing PID Controllers at the Command Line . . . . . . . 4-14

SISO Design Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15

Quick Start for Tuning Compensator Parameters Using the
SISO Design Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-15
Components of the SISO Tool . . . . . . . . . . . . . . . . . . . . . . . . 4-26
Design Options in the SISO Tool . . . . . . . . . . . . . . . . . . . . . 4-26
Opening the SISO Design Tool . . . . . . . . . . . . . . . . . . . . . . . 4-27
Using the SISO Design Task Node in the Control and
Estimation Tools Manager . . . . . . . . . . . . . . . . . . . . . . . . 4-29
Importing Models into the SISO Design Tool . . . . . . . . . . . 4-30
Feedback Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-34
Analysis Plots for Loop Responses . . . . . . . . . . . . . . . . . . . . 4-36
Using the Graphical Tuning Window . . . . . . . . . . . . . . . . . 4-42
Exporting the Compensator and Models . . . . . . . . . . . . . . . 4-49
Storing and Retrieving Intermediate Designs . . . . . . . . . . . 4-51

Bode Diagram Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-55

What Is Bode Diagram Design? . . . . . . . . . . . . . . . . . . . . . . 4-55
Example: DC Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-55
Adjusting the Compensator Gain . . . . . . . . . . . . . . . . . . . . . 4-56

vii
 Adjusting the Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-57
Adding an Integrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-59
Adding a Lead Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-63
Moving Compensator Poles and Zeros . . . . . . . . . . . . . . . . . 4-68
Adding a Notch Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-71
Modifying a Prefilter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-76

Root Locus Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-81

What Is Root Locus Design? . . . . . . . . . . . . . . . . . . . . . . . . . 4-81
Example: Electrohydraulic Servomechanism . . . . . . . . . . . 4-82
Changing the Compensator Gain . . . . . . . . . . . . . . . . . . . . . 4-88
Adding Poles and Zeros to the Compensator . . . . . . . . . . . . 4-91
Editing Compensator Pole and Zero Locations . . . . . . . . . . 4-96
Viewing Damping Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-100

Nichols Plot Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-103

What Is Nichols Plot Design? . . . . . . . . . . . . . . . . . . . . . . . . 4-103
DC Motor Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-103
Opening a Nichols Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-104
Adjusting the Compensator Gain . . . . . . . . . . . . . . . . . . . . . 4-105
Adding an Integrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-108
Adding a Lead Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-110

Automated Tuning Design . . . . . . . . . . . . . . . . . . . . . . . . . . 4-114

Supported Automated Tuning Methods . . . . . . . . . . . . . . . . 4-114
Loading and Displaying the DC Motor Example for
Automated Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-114
Applying Automated PID Tuning . . . . . . . . . . . . . . . . . . . . . 4-116

Multi-Loop Compensator Design . . . . . . . . . . . . . . . . . . . . 4-120

When to Use Multi-Loop Compensator Design . . . . . . . . . . 4-120
Workflow for Multi-Loop Compensator Design . . . . . . . . . . 4-120
Example: Position Control of a DC Motor . . . . . . . . . . . . . . 4-120

Control Design Analysis of Multiple Models . . . . . . . . . . 4-132

Multiple Models Represent System Variations . . . . . . . . . . 4-132
Control Design Analysis Using the SISO Design Tool . . . . 4-132
Specifying a Nominal Model . . . . . . . . . . . . . . . . . . . . . . . . . 4-133
Frequency Grid for Multimodel Computations . . . . . . . . . . 4-136
How to Analyze the Controller Design for Multiple
Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-137

viii Contents
Functions for Compensator Design . . . . . . . . . . . . . . . . . . 4-147
When to Use Functions for Compensator Design . . . . . . . . 4-147
Root Locus Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-147
Pole Placement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-148
Linear-Quadratic-Gaussian (LQG) Design . . . . . . . . . . . . . 4-151
Example — Designing an LQG Regulator . . . . . . . . . . . . . . 4-162
Example — Designing an LQG Servo Controller . . . . . . . . 4-165
Example — Designing an LQR Servo Controller in
Simulink . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-168

Index

ix
x Contents
 1

Product Overview

• “Why Use This Toolbox?” on page 1-2

• “Using the Documentation” on page 1-3
• “Related Products” on page 1-4
 1 Product Overview

Why Use This Toolbox?

The Control System Toolbox™ product extends the MATLAB® software to
provide functions designed specifically for control engineering.

This toolbox lets you construct and analyze linear models of dynamic systems.
Use Control System Toolbox functions to model dynamic systems as transfer
functions, in state-space form, or as arrays of frequency response data. Plot
the time and frequency responses of your system to understand how your
system behaves.

You can also use the toolbox to design and tune single-loop or multiple-loop
control systems using various classical and state-space techniques.

1-2
 Using the Documentation

Using the Documentation

If you are new to Control System Toolbox, review the Control System Toolbox
Getting Started Guide to learn how to:

• Build and manipulate linear time-invariant models of dynamical systems

• Analyze such models and plot their time and frequency responses
• Design compensators using root locus and pole placement techniques

This guide also discusses model order reduction and linear quadratic
Gaussian (LQG) control design techniques with examples.

Other documentation available for this toolbox includes:

• Release Notes — Highlights new and changed features in the latest product
release.
• Creating and Manipulating Models — In-depth information about creating,
manipulating, and analyzing linear models and linear time-invariant
(LTI) arrays. You can use LTI arrays to store collections of linear models
in one variable.
• Customization — Information about setting plot properties, including
setting preferences that persist from session to session.
• Control System Toolbox Design Case Studies — Worked examples
of control design problems, including Kalman filtering and
Linear-Quadratic-Gaussian design. The examples illustrate
design techniques in both single-input/single-output (SISO) and
multiple-input/multiple-output (MIMO) systems.
• Control System Toolbox Reliable Computations — Discussion of numerical
stability and accuracy in computation
• Using the SISO Design Tool and the LTI Viewer — Complete descriptions
of the LTI Viewer and SISO Design Tool. These tools help you analyze
systems and design SISO compensators.
• Function Reference — Describes the purpose, syntax, and options for each
function.

1-3
 1 Product Overview

Related Products
The following table summarizes MathWorks® products that extend and
complement the Control System Toolbox software. For current information
about these products and other MathWorks products, point your Web browser
to:

www.mathworks.com

Product Description
Simulink® Provides blocks for simulating
models you create and analyze
using the Control System Toolbox
software.
Fuzzy Logic Toolbox™ Complements Control System
Toolbox software with tools for
modeling complex systems based on
fuzzy logic.
Model Predictive Control Toolbox™ Helps you design and simulate model
predictive controllers for linear plant
models created using Control System
Toolbox functions.
Robust Control Toolbox™ Helps you design robust control
systems based on plant models
created using the Control System
Toolbox software. Enables you
to assess robustness based on
confidence bounds for your plant
models.
Simulink® Control Design™ Lets you use Control System Toolbox
linear control tools to analyze and
tune Simulink models.

1-4
 Related Products

Product Description
Simulink® Design Optimization™ Helps you improve controller
designs by estimating and tuning
model parameters using numerical
optimization.
“System Identification Toolbox™” Helps you estimate linear and
nonlinear mathematical models of
dynamic systems from measured
data. Use the Control System
Toolbox software to analyze and
design controllers for resulting
linear models.

1-5
 1 Product Overview

1-6
 2

Building Models

• “Linear (LTI) Models” on page 2-2

• “MIMO Models” on page 2-13
• “Arrays of Linear Models” on page 2-18
• “Model Characteristics” on page 2-21
• “Interconnecting Linear Models” on page 2-23
• “Converting Between Continuous- and Discrete- Time Systems” on page
2-26
• “Reducing Model Order” on page 2-29
 2 Building Models

Linear (LTI) Models

In this section...
“What Is a Plant?” on page 2-2
“Linear Model Representations” on page 2-2
“SISO Example: The DC Motor” on page 2-3
“Building SISO Models” on page 2-6
“Constructing Discrete Time Systems” on page 2-9
“Adding Delays to Linear Models” on page 2-10
“LTI Objects” on page 2-10

What Is a Plant?
Typically, control engineers begin by developing a mathematical description of
the dynamic system that they want to control. The system to be controlled is
called a plant. As an example of a plant, this section uses the DC motor. This
section develops the differential equations that describe the electromechanical
properties of a DC motor with an inertial load. It then shows you how to
use the Control System Toolbox functions to build linear models based on
these equations.

Linear Model Representations

You can use Control System Toolbox functions to create the following model
representations:

• State-space models (SS) of the form

dx
= Ax + Bu
dt
y = Cx + Du

where A, B, C, and D are matrices of appropriate dimensions, x is the state

vector, and u and y are the input and output vectors.
• Transfer functions (TF), for example,

2-2
 Linear (LTI) Models

s+2
H (s) = 2
s + s + 10

• Zero-pole-gain (ZPK) models, for example,

( z + 1 + j)( z + 1 − j)
H ( z) = 3
( z + 0.2)( z + 0.1)

• Frequency response data (FRD) models, which consist of sampled

measurements of a system’s frequency response. For example, you can
store experimentally collected frequency response data in an FRD model.

Note The design of FRD models is a specialized subject that this guide
does not address. See “Creating Frequency Response Data (FRD) Models”
for a discussion of this topic.

SISO Example: The DC Motor

A simple model of a DC motor driving an inertial load shows the angular rate
of the load, ω(t) , as the output and applied voltage, υapp (t) , as the input. The
ultimate goal of this example is to control the angular rate by varying the
applied voltage. This figure shows a simple model of the DC motor.

A Simple Model of a DC Motor Driving an Inertial Load

2-3
 2 Building Models

In this model, the dynamics of the motor itself are idealized; for instance,
the magnetic field is assumed to be constant. The resistance of the circuit
is denoted by R and the self-inductance of the armature by L. If you are
unfamiliar with the basics of DC motor modeling, consult any basic text on
physical modeling. With this simple model and basic laws of physics, it is
possible to develop differential equations that describe the behavior of this
electromechanical system. In this example, the relationships between electric
potential and mechanical force are Faraday’s law of induction and Ampère’s
law for the force on a conductor moving through a magnetic field.

Mathematical Derivation
The torque τ seen at the shaft of the motor is proportional to the current i
induced by the applied voltage,

τ(t) = K m i(t)

where Km, the armature constant, is related to physical properties of the

motor, such as magnetic field strength, the number of turns of wire around
the conductor coil, and so on. The back (induced) electromotive force, υemf , is
a voltage proportional to the angular rate ω seen at the shaft,

υemf (t) = K bω(t)

where Kb, the emf constant, also depends on certain physical properties
of the motor.

The mechanical part of the motor equations is derived using Newton’s law,
which states that the inertial load J times the derivative of angular rate
equals the sum of all the torques about the motor shaft. The result is this
equation,

dw
J = ∑ τi = − K f ω(t) + K m i(t)
dt

where K f ω is a linear approximation for viscous friction.

Finally, the electrical part of the motor equations can be described by

2-4
 Linear (LTI) Models

di
υapp (t) − υemf (t) = L + Ri(t)
dt

or, solving for the applied voltage and substituting for the back emf,

di
υapp (t) = L + Ri(t) + K bω(t)
dt

This sequence of equations leads to a set of two differential equations that

describe the behavior of the motor, the first for the induced current,

di R K 1
= − i(t) − b ω(t) + υapp (t)
dt L L L

and the second for the resulting angular rate,

dω 1 1
= − K f ω(t) + K m i(t)
dt J J

State-Space Equations for the DC Motor

Given the two differential equations derived in the last section, you can now
develop a state-space representation of the DC motor as a dynamic system.
The current i and the angular rate ω are the two states of the system. The
applied voltage, υapp , is the input to the system, and the angular velocity
ω is the output.

⎡ R Kb ⎤
− − ⎡1⎤
d ⎡i⎤ ⎢ L L ⎥ ⎡i⎤ ⎢ ⎥
=⎢ ⎥ ⋅ ⎢ ⎥ + L ⋅ υapp (t)
dt ⎢⎣ω⎥⎦ ⎢ K m K f ⎥ ⎣ ω⎦ ⎢ ⎥
⎢⎣ J − ⎢⎣ 0 ⎥⎦
J ⎥⎦

⎡i⎤
y(t) = [0 1] ⋅ ⎢ ⎥ + [0] ⋅ υapp (t)
⎣ ω⎦

State-Space Representation of the DC Motor Example

2-5
 2 Building Models

Building SISO Models

After you develop a set of differential equations that describe your plant, you
can construct SISO models using simple commands. The following sections
discuss

• Constructing a state-space model of the DC motor

• Converting between model representations
• Creating transfer function and zero/pole/gain models

Constructing a State-Space Model of the DC Motor

Enter the following nominal values for the various parameters of a DC motor.

R= 2.0 % Ohms
L= 0.5 % Henrys
Km = .015 % torque constant
Kb = .015 % emf constant
Kf = 0.2 % Nms
J= 0.02 % kg.m^2

Given these values, you can construct the numerical state-space

representation using the ss function.

A = [-R/L -Kb/L; Km/J -Kf/J]

B = [1/L; 0];
C = [0 1];
D = [0];
sys_dc = ss(A,B,C,D)

These commands return the following result:

a =
x1 x2
x1 -4 -0.03
x2 0.75 -10

b =
u1

2-6
 Linear (LTI) Models

x1 2
x2 0

c =
x1 x2
y1 0 1

d =
u1
y1 0

Converting Between Model Representations

Now that you have a state-space representation of the DC motor, you can
convert to other model representations, including transfer function (TF) and
zero/pole/gain (ZPK) models.

Transfer Function Representation. You can use tf to convert from the

state-space representation to the transfer function. For example, use this code
to convert to the transfer function representation of the DC motor.

sys_tf = tf(sys_dc)

Transfer function:
1.5
------------------
s^2 + 14 s + 40.02

Zero/Pole/Gain Representation. Similarly, the zpk function converts

from state-space or transfer function representations to the zero/pole/gain
format. Use this code to convert from the state-space representation to the
zero/pole/gain form for the DC motor.

sys_zpk = zpk(sys_dc)

Zero/pole/gain:
1.5

2-7
 2 Building Models

-------------------
(s+4.004) (s+9.996)

Note The state-space representation is best suited for numerical

computations. For highest accuracy, convert to state space prior to combining
models and avoid the transfer function and zero/pole/gain representations,
except for model specification and inspection. See Reliable Computations for
more information on numerical issues.

Constructing Transfer Function and Zero/Pole/Gain Models

In the DC motor example, the state-space approach produces a set of matrices
that represents the model. If you choose a different approach, you can
construct the corresponding models using tf, zpk, ss, or frd.

sys = tf(num,den) % Transfer function

sys = zpk(z,p,k) % Zero/pole/gain
sys = ss(a,b,c,d) % State-space
sys = frd(response,frequencies) % Frequency response data

For example, you can create the transfer function by specifying the numerator
and denominator with this code.

sys_tf = tf(1.5,[1 14 40.02])

Transfer function:
1.5
------------------
s^2 + 14 s + 40.02

Alternatively, if you want to create the transfer function of the DC motor

directly, use these commands.

s = tf('s');
sys_tf = 1.5/(s^2+14*s+40.02)

These commands result in this transfer function.

Transfer function:

2-8
 Linear (LTI) Models

1.5
--------------------
s^2 + 14 s + 40.02

To build the zero/pole/gain model, use this command.

sys_zpk = zpk([],[-9.996 -4.004], 1.5)

This command returns the following zero/pole/gain representation.

Zero/pole/gain:
1.5
-------------------
(s+9.996) (s+4.004)

Constructing Discrete Time Systems

The Control System Toolbox software provides full support for discrete-time
systems. You can create discrete systems in the same way that you create
analog systems; the only difference is that you must specify a sample time
period for any model you build. For example,

sys_disc = tf(1, [1 1], .01);

creates a SISO model in the transfer function format.

Transfer function:
1
-----
z + 1

Sampling time: 0.01

Adding Time Delays to Discrete-Time Models

You can add time delays to discrete-time models by specifying an input delay,
output delay, or I/O delay when building the model. The time delay must be
a nonnegative integer that represents a multiple of the sampling time. For
example,

sys_delay = tf(1, [1 1], 0.01,'ioDelay',5)

2-9
 2 Building Models

returns a system with an I/O delay of 5 s.

Transfer function:
1
z^(-5) * -----
z + 1

Sampling time: 0.01

Adding Delays to Linear Models

You can add time delays to linear models by specifying an input delay, output
delay, or I/O delay when building a model. For example, to add an I/O delay to
the DC motor, use this code.

sys_tfdelay = tf(1.5,[1 14 40.02],'ioDelay',0.05)

This command constructs the DC motor transfer function, but adds a 0.05
second delay.

Transfer function:
1.5
exp(-0.05*s) * ------------------
s^2 + 14 s + 40.02

For a complete description of adding time delays to models and closing loops
with time delays, see Time Delays.

LTI Objects
For convenience, the Control System Toolbox software uses custom data
structures called LTI objects to store model-related data. For example, the
variable sys_dc created for the DC motor example is called an SS object.
There are also TF, ZPK, and FRD objects for transfer function, zero/pole/gain,
and frequency data response models respectively. The four LTI objects
encapsulate the model data and enable you to manipulate linear systems as
single entities rather than as collections of vectors or matrices.

To see what LTI objects contain, use the get command. This code describes
the contents of sys_dc from the DC motor example.

2-10
 Linear (LTI) Models

get(sys_dc)
a: [2x2 double]
b: [2x1 double]
c: [0 1]
d: 0
e: []
StateName: {2x1 cell}
InternalDelay: [0x1 double]
Ts: 0
InputDelay: 0
OutputDelay: 0
InputName: {''}
OutputName: {''}
InputGroup: [1x1 struct]
OutputGroup: [1x1 struct]
Name: ''
Notes: {}
UserData: []

You can manipulate the data contained in LTI objects using the set command;
see the Control System Toolbox online reference pages for descriptions of
set and get.

Another convenient way to set or retrieve LTI model properties is to access

them directly using dot notation. For example, if you want to access the value
of the A matrix, instead of using get, you can type

sys_dc.a

at the MATLAB prompt. This notation returns the A matrix.

ans =

-4.0000 -0.0300
0.7500 -10.0000

Similarly, if you want to change the values of the A matrix, you can do so
directly, as this code shows.

A_new = [-4.5 -0.05; 0.8 -12.0];

2-11
 2 Building Models

sys_dc.a = A_new;

For more information on LTI properties, type ltiprops at the MATLAB

prompt. For a complete description of LTI objects, see Creating and
Manipulating Models.

2-12
 MIMO Models

MIMO Models
In this section...
“MIMO Example: State-Space Model of Jet Transport Aircraft” on page 2-13
“Constructing MIMO Transfer Functions” on page 2-15
“Accessing I/O Pairs in MIMO Systems” on page 2-17

MIMO Example: State-Space Model of Jet Transport

Aircraft
You can use the same functions that apply to SISO systems to create
multiple-input, multiple-output (MIMO) models. This example shows how
to build a MIMO model of a jet transport. Because the development of a
physical model for a jet aircraft is lengthy, only the state-space equations
are presented here. See any standard text in aviation for a more complete
discussion of the physics behind aircraft flight.

The jet model during cruise flight at MACH = 0.8 and H = 40,000 ft. is

A = [-0.0558 -0.9968 0.0802 0.0415

0.5980 -0.1150 -0.0318 0
-3.0500 0.3880 -0.4650 0
0 0.0805 1.0000 0];

B = [ 0.0073 0
-0.4750 0.0077
0.1530 0.1430
0 0];

C = [0 1 0 0
0 0 0 1];

D = [0 0
0 0];

Use the following commands to specify this state-space model as an LTI object
and attach names to the states, inputs, and outputs.

2-13
 2 Building Models

states = {'beta' 'yaw' 'roll' 'phi'};

inputs = {'rudder' 'aileron'};
outputs = {'yaw rate' 'bank angle'};

sys_mimo = ss(A,B,C,D,'statename',states,...
'inputname',inputs,...
'outputname',outputs);

You can display the LTI model by typing sys_mimo.

sys_mimo

a =
beta yaw roll phi
beta -0.0558 -0.9968 0.0802 0.0415
yaw 0.598 -0.115 -0.0318 0
roll -3.05 0.388 -0.465 0
phi 0 0.0805 1 0

b =
rudder aileron
beta 0.0073 0
yaw -0.475 0.0077
roll 0.153 0.143
phi 0 0

c =
beta yaw roll phi
yaw rate 0 1 0 0
bank angle 0 0 0 1

d =
rudder aileron
yaw rate 0 0
bank angle 0 0

Continuous-time model.

2-14
 MIMO Models

The model has two inputs and two outputs. The units are radians for beta
(sideslip angle) and phi (bank angle) and radians/sec for yaw (yaw rate) and
roll (roll rate). The rudder and aileron deflections are in degrees.

As in the SISO case, use tf to derive the transfer function representation.

tf(sys_mimo)

Transfer function from input "rudder" to output...

-0.475 s^3 - 0.2479 s^2 - 0.1187 s - 0.05633
yaw rate: ---------------------------------------------------
s^4 + 0.6358 s^3 + 0.9389 s^2 + 0.5116 s + 0.003674

0.1148 s^2 - 0.2004 s - 1.373

bank angle: ---------------------------------------------------
s^4 + 0.6358 s^3 + 0.9389 s^2 + 0.5116 s + 0.003674

Transfer function from input "aileron" to output...

0.0077 s^3 - 0.0005372 s^2 + 0.008688 s + 0.004523
yaw rate: ---------------------------------------------------
s^4 + 0.6358 s^3 + 0.9389 s^2 + 0.5116 s + 0.003674

0.1436 s^2 + 0.02737 s + 0.1104

bank angle: ---------------------------------------------------
s^4 + 0.6358 s^3 + 0.9389 s^2 + 0.5116 s + 0.003674

Constructing MIMO Transfer Functions

MIMO transfer functions are two-dimensional arrays of elementary SISO
transfer functions. There are two ways to specify MIMO transfer function
models:

• Concatenation of SISO transfer function models

• Using tf with cell array arguments

Concatenation of SISO Models

Consider the following single-input, two-output transfer function.

2-15
 2 Building Models

⎡ s −1 ⎤
⎢ s+1 ⎥
H (s) = ⎢ ⎥
⎢ s+2 ⎥
⎢⎣ s2 + 4 s + 5 ⎥⎦

You can specify H(s) by concatenation of its SISO entries. For instance,

h11 = tf([1 -1],[1 1]);

h21 = tf([1 2],[1 4 5]);

or, equivalently,

s = tf('s')
h11 = (s-1)/(s+1);
h21 = (s+2)/(s^2+4*s+5);

can be concatenated to form H(s).

H = [h11; h21]

This syntax mimics standard matrix concatenation and tends to be easier

and more readable for MIMO systems with many inputs and/or outputs. See
Model Interconnection Functions for more details on concatenation operations
for LTI systems.

Using the tf Function with Cell Arrays

Alternatively, to define MIMO transfer functions using tf, you need two cell
arrays (say, N and D) to represent the sets of numerator and denominator
polynomials, respectively. See Cell Arrays in the MATLAB documentation for
more details on cell arrays.

For example, for the rational transfer matrix H(s), the two cell arrays N and D
should contain the row-vector representations of the polynomial entries of

⎡ s −1⎤ ⎡ s+1 ⎤
N (s) = ⎢ D(s) = ⎢ 2 ⎥
⎣ s + 2 ⎥⎦ ⎣ s + 4s + 5 ⎦

You can specify this MIMO transfer matrix H(s) by typing

2-16
 MIMO Models

N = {[1 -1];[1 2]}; % Cell array for N(s)

D = {[1 1];[1 4 5]}; % Cell array for D(s)
H = tf(N,D)

These commands return the following result:

Transfer function from input to output...

s - 1
#1: -----
s + 1

s + 2
#2: -------------
s^2 + 4 s + 5

Notice that both N and D have the same dimensions as H. For a general MIMO
transfer matrix H(s), the cell array entries N{i,j} and D{i,j} should be
row-vector representations of the numerator and denominator of Hij(s), the
ijth entry of the transfer matrix H(s).

Accessing I/O Pairs in MIMO Systems

After you define a MIMO system, you can access and manipulate I/O pairs by
specifying input and output pairs of the system. For instance, if sys_mimo is a
MIMO system with two inputs and three outputs,

sys_mimo(3,1)

extracts the subsystem, mapping the first input to the third output. Row
indices select the outputs and column indices select the inputs. Similarly,

sys_mimo(3,1) = tf(1,[1 0])

redefines the transfer function between the first input and third output as
an integrator.

2-17
 2 Building Models

Arrays of Linear Models

You can specify and manipulate collections of linear models as single entities
using LTI arrays. For example, if you want to vary the Kb and Km parameters
for the DC motor and store the resulting state-space models, use this code.

K = [0.1 0.15 0.2]; % Several values for Km and Kb

A1 = [-R/L -K(1)/L; K(1)/J -Kf/J];
A2 = [-R/L -K(2)/L; K(2)/J -Kf/J];
A3 = [-R/L -K(3)/L; K(3)/J -Kf/J];
sys_lti(:,:,1)= ss(A1,B,C,D);
sys_lti(:,:,2)= ss(A2,B,C,D);
sys_lti(:,:,3)= ss(A3,B,C,D);

The number of inputs and outputs must be the same for all linear models
encapsulated by the LTI array, but the model order (number of states) can
vary from model to model within a single LTI array.

The LTI array sys_lti contains the state-space models for each value of K.
Type sys_lti to see the contents of the LTI array.

Model sys_lti(:,:,1,1)
======================

a =
x1 x2
x1 -4 -0.2
x2 5 -10
.
.
.
Model sys_lti(:,:,2,1)
======================

a =
x1 x2
x1 -4 -0.3
x2 7.5 -10
.
.

2-18
 Arrays of Linear Models

.
Model sys_lti(:,:,3,1)
======================

a =
x1 x2
x1 -4 -0.4
x2 10 -10
.
.
.
3x1 array of continuous-time state-space models.

You can manipulate the LTI array like any other object. For example,

step(sys_lti)

produces a plot containing step responses for all three state-space models.

Step Responses for an LTI Array Containing Three Models

2-19
 2 Building Models

LTI arrays are useful for performing batch analysis on an entire set of models.
For more information, see Arrays of LTI Models.

2-20
 Model Characteristics

Model Characteristics
You can use the Control System Toolbox commands to query model
characteristics such as the I/O dimensions, poles, zeros, and DC gain.
These commands apply to both continuous- and discrete-time models. Their
LTI-based syntax is summarized in the table below.

Commands to Query Model Characteristics

Command Description
size(model_name) Number of inputs and outputs
ndims(model_name) Number of dimensions
isct(model_name) Returns 1 for continuous systems
isdt(model_name) Returns 1 for discrete systems
hasdelay(model_name) True if system has delays
pole(model_name) System poles
zero(model_name) System (transmission) zeros
dcgain(model_name) DC gain
norm(model_name) System norms (H2 and L∞)
covar(model_name,W) Covariance of response to white noise

bandwidth(model_name) Frequency response bandwidth

lti/order(model_name) LTI model order

pzmap(model_name) Compute pole-zero map of LTI models

damp(model_name) Natural frequency and damping of system

poles

2-21
 2 Building Models

Commands to Query Model Characteristics (Continued)

Command Description
class(model_name) Create object or return class of object

isa(model_name) Determine whether input is object of given

class

isempty(model_name) Determine whether LTI model is empty

isproper(model_name) Determine whether LTI model is proper

issiso(model_name) Determine whether LTI model is

single-input/single-output (SISO)

lti/isstable(model_name) Determine whether system is stable

reshape(model_name) Change shape of LTI array

2-22
 Interconnecting Linear Models

Interconnecting Linear Models

In this section...
“Arithmetic Operations for Interconnecting Models” on page 2-23
“Feedback Interconnections” on page 2-24

Arithmetic Operations for Interconnecting Models

You can perform arithmetic on LTI models, such as addition, multiplication,
or concatenation. Addition performs a parallel interconnection. For example,
typing

tf(1,[1 0]) + tf([1 1],[1 2]) % 1/s + (s+1)/(s+2)

produces this transfer function.

Transfer function:
s^2 + 2 s + 2
-------------
s^2 + 2 s

Multiplication performs a series interconnection. For example, typing

2 * tf(1,[1 0])*tf([1 1],[1 2]) % 2*1/s*(s+1)/(s+2)

produces this cascaded transfer function.

Transfer function:
2 s + 2
---------
s^2 + 2 s

If the operands are models of different types, the resulting model type is
determined by precedence rules; see “Precedence and Property Inheritance”
for more information. State-space models have the highest precedence while
transfer functions have the lowest precedence. Hence the sum of a transfer
function and a state-space model is always a state-space model.

2-23
 2 Building Models

Other available operations include system inversion, transposition, and

pertransposition; see Arithmetic Operations. You can also perform matrix-like
indexing for extracting subsystems; see Extracting and Modifying Subsystems
for more information.

You can also use the series and parallel functions as substitutes for
multiplication and addition, respectively.

Equivalent Ways to Interconnect Systems

Operator Function Resulting Transfer Function

sys1 + sys2 parallel(sys1,sys2) Systems in parallel
sys1 - sys2 parallel(sys1,-sys2) Systems in parallel
sys1 * sys2 series(sys2,sys1) Cascaded systems

Feedback Interconnections
You can use the feedback and lft functions to derive closed-loop models.
For example,

sys_f = feedback(tf(1,[1 0]), tf([1 1],[1 2])

computes the closed-loop transfer function from r to y for the feedback loop
shown below. The result is

Transfer function:
s + 2
-------------
s^2 + 3 s + 1

This figure shows the interconnected system in block diagram format.

2-24
 Interconnecting Linear Models

Feedback Interconnection

You can use the lft function to create more complicated feedback structures.
This function constructs the linear fractional transformation of two systems.
See the reference page for more information.

2-25
 2 Building Models

Converting Between Continuous- and Discrete- Time

Systems
In this section...
“Available Commands for Continuous/Discrete Conversion” on page 2-26
“Available Methods for Continuous/Discrete Conversion” on page 2-26
“Digitizing the Discrete DC Motor Model” on page 2-26

Available Commands for Continuous/Discrete

Conversion
The commands c2d, d2c, and d2d perform continuous to discrete, discrete to
continuous, and discrete to discrete (resampling) conversions, respectively.

sysd = c2d(sysc,Ts) % Discretization w/ sample period Ts

sysc = d2c(sysd) % Equivalent continuous-time model
sysd1= d2d(sysd,Ts) % Resampling at the period Ts

Available Methods for Continuous/Discrete

Conversion
Various discretization/interpolation methods are available, including
zero-order hold (default), first-order hold, Tustin approximation with or
without prewarping, and matched zero-pole. For example,

sysd = c2d(sysc,Ts,'foh') % Uses first-order hold

sysc = d2c(sysd,'tustin') % Uses Tustin approximation

Digitizing the Discrete DC Motor Model

You can digitize the DC motor plant using the c2d function and selecting an
appropriate sample time. Choosing the right sample time involves many
factors, including the performance you want to achieve, the fastest time
constant in your system, and the speed at which you expect your controller
to run. For this example, choose a time constant of 0.01 second. See “SISO
Example: The DC Motor” on page 2-3 for the construction of the SS object
sys_dc.

Ts=0.01;

2-26
 Converting Between Continuous- and Discrete- Time Systems

sysd=c2d(sys_dc,Ts)

a =
x1 x2
x1 0.96079 -0.00027976
x2 0.006994 0.90484

b =
u1
x1 0.019605
x2 7.1595e-005

c =
x1 x2
y1 0 1

d =
u1
y1 0

Sampling time: 0.01

Discrete-time model.

To see the discrete-time zero-pole gain for the digital DC motor, use zpk to
convert the model.

fd=zpk(sysd)

Zero/pole/gain:
7.1595e-005 (z+0.9544)
------------------------
(z-0.9608) (z-0.9049)

Sampling time: 0.01

You can compare the step responses of sys_dc and sysd by typing

2-27
 2 Building Models

step(sys_dc,sysd)

This figure shows the result.

Note the step response match. Continuous and FOH-discretized step

responses match for models without internal delays.

2-28
 Reducing Model Order

Reducing Model Order

In this section...
“Model Order Reduction Commands” on page 2-29
“Techniques for Reducing Model Order” on page 2-29
“Example: Gasifier Model” on page 2-30

Model Order Reduction Commands

You can derive reduced-order SISO and MIMO models with the commands
shown in the following table.

Model Order Reduction

Commands
hsvd Compute Hankel singular values of LTI model
balred Reduced-order model approximation
minreal Minimal realization (pole/zero cancellation)
balreal Compute input/output balanced realization
modred State deletion in I/O balanced realization
sminreal Structurally minimal realization

Techniques for Reducing Model Order

To reduce the order of a model, you can perform any of the following actions:

• Eliminate states that are structurally disconnected from the inputs or

outputs using sminreal.
Eliminating structurally disconnected states is a good first step in model
reduction because the process is cheap and does not involve any numerical
computation.
• Compute a low-order approximation of your model using balred.
• Eliminate cancelling pole/zero pairs using minreal.

2-29
 2 Building Models

Example: Gasifier Model

This example presents a model of a gasifier, a device that converts solid
materials into gases. The original model is nonlinear.

Loading the Model

To load a linearized version of the model, type

load ltiexamples

at the MATLAB prompt; the gasifier example is stored in the variable named
gasf. If you type

size(gasf)

you get in return

State-space model with 4 outputs, 6 inputs, and 25 states.

SISO Model Order Reduction

You can reduce the order of a single I/O pair to understand how the model
reduction tools work before attempting to reduce the full MIMO model as
described in “MIMO Model Order Reduction” on page 2-34.

This example focuses on a single input/output pair of the gasifier, input 5

to output 3.

sys35 = gasf(3,5);

Before attempting model order reduction, inspect the pole and zero locations
by typing

pzmap(sys35)

Zoom in near the origin on the resulting plot using the zoom feature or by
typing

axis([-0.2 0.05 -0.2 0.2])

The following figure shows the results.

2-30
 Reducing Model Order

Pole-Zero Map of the Gasifier Model (Zoomed In)

Because the model displays near pole-zero cancellations, it is a good candidate

for model reduction.

To find a low-order reduction of this SISO model, perform the following steps:

1 Select an appropriate order for your reduced system by examining the

relative amount of energy per state using a Hankel singular value (HSV)
plot. Type the command

hsvd(sys35)

to create the HSV plot.

Changing to log scale using the right-click menu results in the following
plot.

2-31
 2 Building Models

Small Hankel singular values indicate that the associated states contribute
little to the I/O behavior. This plot shows that discarding the last 10 states
(associated with the 10 smallest Hankel singular values) has little impact
on the approximation error.

2 To remove the last 10 states and create a 15th order approximation, type

rsys35 = balred(sys35,15);

You can type size(rsys35) to see that your reduced system contains only
15 states.

3 Compare the Bode response of the full-order and reduced-order models

using the bode command:

bode(sys35,'b',rsys35,'r--')

This figure shows the result.

2-32
 Reducing Model Order

As the overlap of the curves in the figure shows, the reduced model provides
a good approximation of the original system.

You can try reducing the model order further by repeating this process and
removing more states. Reduce the gasf model to 5th, 10th, and 15th orders
all at once by typing the following command

rsys35 = balred(sys35,[5 10 15]);

Plot a bode diagram of these three reduced systems along with the full order
system by typing

bode(sys35,'b',rsys35,'r--')

2-33
 2 Building Models

Observe how the error increases as the order decreases.

MIMO Model Order Reduction

You can approximate MIMO models using the same steps as SISO models as
follows:

1 Select an appropriate order for your reduced system by examining the

relative amount of energy per state using a Hankel singular value (HSV)
plot.

Type

hsvd(gasf)

to create the HSV plot.

2-34
 Reducing Model Order

Discarding the last 8 states (associated with the 8 smallest Hankel singular
values) should have little impact on the error in the resulting 17th order
system.

2 To remove the last 8 states and create a 17th order MIMO system, type

rsys=balred(gasf,17);

You can type size(gasf) to see that your reduced system contains only
17 states.

3 To facilitate visual inspection of the approximation error, use a singular

value plot rather than a MIMO Bode plot. Type

sigma(gasf,'b',gasf-rsys,'r')

to create a singular value plot comparing the original system to the

reduction error.

2-35
 2 Building Models

The reduction error is small compared to the original system so the reduced
order model provides a good approximation of the original model.

Acknowledgment
MathWorks thanks ALSTOM® Power UK for permitting use of their gasifier
model for this example. This model was issued as part of the ALSTOM
Benchmark Challenge on Gasifier Control. For more details see Dixon, R.,
(1999), "Advanced Gasifier Control," Computing & Control Engineering
Journal, IEE, Vol. 10, No. 3, pp. 92–96.

2-36
 3

Analyzing Models

• “Quick Start for Performing Linear Analysis Using the LTI Viewer” on
page 3-2
• “LTI Viewer” on page 3-7
• “Simulating Models with Arbitrary Inputs and Initial Conditions” on page
3-22
• “Functions for Time and Frequency Response” on page 3-37
 3 Analyzing Models

Quick Start for Performing Linear Analysis Using the LTI

Viewer
In this Quick Start, you learn how to analyze the time- and frequency-domain
responses of one or more linear models using the LTI Viewer GUI.

Before you can perform the analysis, you must have already created linear
models in the MATLAB workspace. For information on how to create a model,
see Chapter 2, “Building Models”.

To perform linear analysis:

1 Open the LTI Viewer showing one or more models using the following
syntax:

ltiview(model1,model2,...,modelN)

By default, this syntax opens a step response plot of your models, as shown
in the following figure.

3-2
 Quick Start for Performing Linear Analysis Using the LTI Viewer

Click to add a legend

3-3
 3 Analyzing Models

2 Add more plots to the LTI Viewer.

a Select Edit > Plot Configurations.

b In the Plot Configurations dialog box, select the number of plots to open.

Select the
number of
plots to open

3-4
 Quick Start for Performing Linear Analysis Using the LTI Viewer

3 To view a different type of response on a plot, right-click and select a plot

type.

Right-click
to select
a plot type

3-5
 3 Analyzing Models

4 Analyze system performance. For example, you can analyze the peak
response in the Bode plot and settling time in the step response plot.
a Right-click to select performance characteristics.

b Click on the dot that appears to view the characteristic value.

Right-click to show
performance
characteristics

Click dot to
view value

3-6
 LTI Viewer

LTI Viewer
In this section...
“Plot Types Available in the LTI Viewer” on page 3-7
“Example: Time and Frequency Responses of the DC Motor” on page 3-8
“Right-Click Menus” on page 3-10
“Displaying Response Characteristics on a Plot” on page 3-11
“Changing Plot Type” on page 3-14
“Showing Multiple Response Types” on page 3-16
“Comparing Multiple Models” on page 3-17

Plot Types Available in the LTI Viewer

The LTI Viewer is a GUI for viewing and manipulating the response plots of
linear models. You can display the following plot types for linear models
using the LTI Viewer:

• Step and impulse responses

• Bode and Nyquist plots
• Nichols plots
• Singular values of the frequency response
• Pole/zero plots
• Response to a general input signal
• Unforced response starting from given initial states (only for state-space
models)

Note that time responses and pole/zero plots are not available for FRD models.

Note The LTI Viewer displays up to six different plot types simultaneously.
You can also analyze the response plots of several linear models at once.

3-7
 3 Analyzing Models

This figure shows an LTI Viewer with two response plots.

The LTI Viewer with Step and Impulse Response Plots

The next section presents an example that shows you how to import a system
into the LTI Viewer and how to customize the viewer to fit your requirements.

Example: Time and Frequency Responses of the DC

Motor
“SISO Example: The DC Motor” on page 2-3 presents a DC motor example. If
you have not yet built that example, type

load ltiexamples

at the MATLAB prompt. This loads several LTI models, including a

state-space representation of the DC motor called sys_dc.

3-8
 LTI Viewer

Opening the LTI Viewer

To open the LTI Viewer, type

ltiview

This opens an LTI Viewer with an empty step response plot window by default.

Importing Models into the LTI Viewer

To import the DC motor model, select Import under the File menu. This
opens the Import System Data dialog box, which lists all the models available
in your MATLAB workspace.

Import System Data Dialog Box with the DC Motor Model Selected

Select sys_dc from the list of available models and click OK to close the
browser. This imports the DC motor model into the LTI Viewer.

To select more than one model at a time, do the following:

• To select individual (noncontiguous) models, select one model and hold

down the Ctrl key while selecting additional models. To clear any models,
hold down the Ctrl key while you click the highlighted model names.
• To select a list of contiguous models, select the first model and hold down
the Shift key while selecting the last model you want in the list.

3-9
 3 Analyzing Models

The figure below shows the LTI Viewer with a step response for the DC
motor example.

Step Response for the DC Motor Example in the LTI Viewer

Alternatively, you can open the LTI Viewer and import the DC motor example
directly from the MATLAB prompt.

ltiview('step', sys_dc)

See the ltiview reference page for a complete list of options.

Right-Click Menus
The LTI Viewer provides a set of controls and options that you can access by
right-clicking your mouse. Once you have imported a model into the LTI
Viewer, the options you can select include

• Plot Types — Change the plot type. Available types include step, impulse,
Bode, Bode magnitude, Nichols, Nyquist, and singular values plots.

3-10
 LTI Viewer

• Systems — Select or clear any models that you included when you created
the response plot.
• Characteristics — Add information about the plot. The characteristics
available change from plot to plot. For example, Bode plots have stability
margins available, but step responses have rise time and steady-state
values available.
• Grid — Add grids to your plots.
• Normalize — Scale responses to fit the view (only available for
time-domain plot types).
• Full View — Use automatic limits to make the entire curve visible.
• Properties — Open the Property Editor.

You can use this editor to customize various attributes of your plot. See
Customizing Plot Properties and Preferences for a full description of the
Property Editor.
Alternatively, you can open the Property Editor by double-clicking in an
empty region of the response plot.

Displaying Response Characteristics on a Plot

For example, to see the rise time for the DC motor step response, right-click
and select Characteristics > Rise Time.

3-11
 3 Analyzing Models

Using Right-Click Menus to Display the Rise Time for a Step Response

By default, the rise time is defined as the amount of time it takes the step
response to go from 10% to 90% of the steady-state value. You can change this
range in the options tab of the property editor. For more information on the
property editor, see Customizing Plot Properties and Preferences.

The LTI Viewer calculates and displays the rise time for the step response.

3-12
 LTI Viewer

DC Motor Step Response with the Rise Time Displayed

To display the values of any plot characteristic marked on a plot, place

your mouse on the blue dot that marks the characteristic. This opens a
data marker with the relevant information displayed. To make the marker
persistent, left-click the blue dot.

For example, this figure shows the rise time value for the DC motor step
response.

3-13
 3 Analyzing Models

Using Your Mouse to Get the Rise Time Values

Note that you can left-click anywhere on a particular plot line to see the
response values of that plot at that point. You must either place your cursor
over the blue dot or left-click, however, if you want to see the rise time value.

For more information about data markers, see “Data Markers” on page 3-42.

Changing Plot Type

You can view other plots using the right-click menus in the LTI Viewer. For
example, if you want to see the open loop Bode plots for the DC motor model,
select Plot Type and then Bode from the right-click menu.

3-14
 LTI Viewer

Changing the Step Response to a Bode Plot

Selecting Bode changes the step response to a Bode plot for the DC motor
model.

3-15
 3 Analyzing Models

Bode Plot for the DC Motor Model

Showing Multiple Response Types

If you want to see, for example, both a step response and a Bode plot at
the same time, you have to reconfigure the LTI Viewer. To view different
response types in a single LTI Viewer, select Plot Configurations under the
Edit menu. This opens the Plot Configurations dialog box.

3-16
 LTI Viewer

Using the Plot Configurations Dialog Box to Reconfigure the LTI Viewer

You can select up to six plots in one viewer. Choose the Response type for
each plot area from the right-side menus. There are nine available plot types:

• Step
• Impulse
• Linear Simulation
• Initial Condition
• Bode (magnitude and phase)
• Bode Magnitude (only)
• Nyquist
• Nichols
• Singular Value
• Pole/Zero
• I/O pole/zero

Comparing Multiple Models

This section shows you how to import and manipulate multiple models in
one LTI Viewer. For example, if you have designed a set of compensators to
control a system, you can compare the closed-loop step responses and Bode
plots using the LTI Viewer.

3-17
 3 Analyzing Models

A sample set of closed-loop transfer function models is included (along with

some other models) in the MAT-file ltiexamples.mat. Type

load ltiexamples

to load the provided transfer functions. The three closed-loop transfer function
models, Gcl1, Gcl2, and Gcl3, are for a satellite attitude controller.

In this example, you analyze the response plots of the Gcl1 and Gcl2 transfer
functions.

Initializing the LTI Viewer with Multiple Plots

To load the two models Gcl1 and Gcl2 into the LTI Viewer, select Import
under the File menu and select the desired models in the LTI Browser. See
“Importing Models into the LTI Viewer” on page 3-9 for a description of how
to select groups of models. If necessary, you can reconfigure the viewer to
display both the step responses and the Bode plots of the two systems using
the Viewer Configuration dialog box. See “Showing Multiple Response Types”
on page 3-16 for a discussion of this feature.

Alternatively, you can open an LTI Viewer with both systems and both the
step responses and Bode plots displayed. To do this, type

ltiview({'step';'bode'},Gcl1,Gcl2)

Either approach opens the following LTI Viewer.

3-18
 LTI Viewer

Multiple Response Plots in a Single LTI Viewer

Inspecting Response Characteristics

To mark the settling time on the step responses presented in this example,
do the following:

• Right-click anywhere in the plot region of the step response plots. This
opens the right-click menu list in the plot region.
• Select Characteristics > Settling Time.

To mark the stability margins of the Bode plot in this example, right-click and
select Characteristics > Minimum Stability Margins.

Your LTI Viewer should now look like this.

3-19
 3 Analyzing Models

Multiple Plots with Response Characteristics Added

The minimum stability margins, meaning the smallest magnitude phase and
gain margins, are displayed as green and blue markers on the Bode phase
diagram. If you want to see all the gain and phase margins of a system,
right-click and select Characteristics > Minimum Stability Margins.

Toggling Model Visibility

If you have imported more than one model, you can select and clear which
models to plot in the LTI Viewer using right-click menus. For example, if you
import the following three models into the viewer, you can choose to view any
combination of the three you want.

s=tf('s');
sys1=1/(s^2+s+1);
sys2=1/(s^2+s+2);
sys3=1/(s^2+s+3);

3-20
 LTI Viewer

This figure shows how to clear the second of the three models using right-click
menu options.

Using Right-Click Menus to Select/Clear Plotted Systems

The Systems menu lists all the imported models. A system is selected if a
check mark is visible to the left of the system name.

3-21
 3 Analyzing Models

Simulating Models with Arbitrary Inputs and Initial

Conditions
In this section...
“What is the Linear Simulation Tool?” on page 3-22
“Opening the Linear Simulation Tool” on page 3-22
“Working with the Linear Simulation Tool” on page 3-23
“Importing Input Signals” on page 3-26
“Example: Loading Inputs from a Microsoft® Excel Spreadsheet” on page
3-28
“Example: Importing Inputs from the Workspace” on page 3-29
“Designing Input Signals” on page 3-33
“Specifying Initial Conditions” on page 3-35

What is the Linear Simulation Tool?

You can use the Linear Simulation Tool to simulate linear models with
arbitrary input signals and initial conditions.

The Linear Simulation Tool lets you do the following:

• Import input signals from the MATLAB workspace.

• Import input signals from a MAT-file, Microsoft® Excel® spreadsheet,
ASCII flat-file, comma-separated variable file (CSV), or text file.
• Generate arbitrary input signals in the form of a sine wave, square wave,
step function, or white noise.
• Specify initial states for state-space models.
Default initial states are zero.

Opening the Linear Simulation Tool

To open the Linear Simulation Tool, do one of the following:

3-22
 Simulating Models with Arbitrary Inputs and Initial Conditions

• In the LTI Viewer, right-click the plot area and select Plot Types >
Linear Simulation.
• Use the lsim function at the MATLAB prompt:

lsim(modelname)

• In the MATLAB Figure window, right-click a response plot and select

Input data.

Working with the Linear Simulation Tool

The Linear Simulation Tool contains two tabs, Input signals and Initial
states.

After opening the Linear Simulation Tool (as described in “Opening the Linear
Simulation Tool” on page 3-22), follow these steps to simulate your model:

1 Click the Input signals tab, if it is not displayed.

3-23
 3 Analyzing Models

2 In the Timing area, specify the simulation time vector by doing one of
the following:
• Import the time vector by clicking Import time.
• Enter the end time and the time interval in seconds. The start time
is set to 0 seconds.

3 Specify the input signal by doing one of the following:

• Click Import signal to import it from the MATLAB workspace or a file.

For more information, see “Importing Input Signals” on page 3-26.
• Click Design signal to create your own inputs. For more information,
see “Designing Input Signals” on page 3-33.

4 If you have a state-space model and want to specify initial conditions, click
the Initial states tab. By default, all initial states are set to zero.

You can either enter state values in the Initial value column, or import
values by clicking Import state vector. For more information about
entering initial states, see “Specifying Initial Conditions” on page 3-35.

3-24
 Simulating Models with Arbitrary Inputs and Initial Conditions

5 For a continuous model, select one of the following interpolation methods in

the Interpolation method list to be used by the simulation solver:
• Zero order hold
• First order hold (linear interpolation)
• Automatic (Linear Simulation Tool selects first order hold or zero order
hold automatically, based on the smoothness of the input)

Note The interpolation method is not used when simulating discrete

models.

6 Click Simulate.

3-25
 3 Analyzing Models

Importing Input Signals

You can import input signals from the MATLAB workspace after opening
the Linear Simulation Tool (see “Opening the Linear Simulation Tool” on
page 3-22). You can also import inputs from a MAT-file, Microsoft Excel
spreadsheet, ASCII flat-file, comma-separated variable file (CSV), or text file.

For information about creating your own inputs, see “Designing Input
Signals” on page 3-33. For an overview of working with the Linear Simulation
Tool, see “Working with the Linear Simulation Tool” on page 3-23.

To import one or more input signals:

1 In the Linear Simulation Tool, click the Input signals tab, if it is not
displayed.

2 Specify the simulation time in the Timing area.

3 Select one or more rows for the input channels you want to import. The
following figure shows an example with two selected channels.

3-26
 Simulating Models with Arbitrary Inputs and Initial Conditions

4 Click Import signal to open the Data Import dialog box. The following
figure shows an example of the Data Import dialog box.

3-27
 3 Analyzing Models

5 In the Import from list, select the source of the input signals. It can be
one of the following:
• Workspace
• MAT file
• XLS file
• CSV file
• ASCII file

6 Select the data you want to import. The Data Import dialog box contains
different options depending on which source format you selected.

7 Click Import.

For an example of importing input signals, see the following:

• “Example: Loading Inputs from a Microsoft® Excel Spreadsheet” on page

3-28
• “Example: Importing Inputs from the Workspace” on page 3-29

Example: Loading Inputs from a Microsoft Excel

Spreadsheet
To load inputs from a Microsoft Excel (XLS) spreadsheet:

1 In the Linear Simulation Tool, click Import signal in the Input signals
tab to open the Data Import dialog box.

2 Select XLS file in the Import from list.

3 Click Browse.

4 Select the file you want to import and click Open. This populates the Data
Import dialog box with the data from the Microsoft Excel spreadsheet.

3-28
 Simulating Models with Arbitrary Inputs and Initial Conditions

Example: Importing Inputs from the Workspace

To load an input signal from the MATLAB workspace:

1 Enter this code to open a response plot with a second-order system:

s=tf('s');
ss=(s+2)/(s^2+3*s+2);
lsim(ss,randn(100,1),1:100);

2 Right-click the plot background and select Input data.

3-29
 3 Analyzing Models

This opens the Linear Simulation Tool with default input data.

3-30
 Simulating Models with Arbitrary Inputs and Initial Conditions

3 Create an input signal for your system in the MATLAB Command Window,
such as the following:

new_signal=[-3*ones(1,20) 2*ones(1,30) 0.5*ones(1,50)]';

4 In the Linear Simulation Tool, click Import signal.

5 In the Data Import dialog box, click, Assign columns to assign the first
column of the input signal to the selected channel.

3-31
 3 Analyzing Models

6 Click Import. This imports the new signal into the Linear Simulation Tool.

3-32
 Simulating Models with Arbitrary Inputs and Initial Conditions

7 Click Simulate to see the response of your second-order system to the

imported signal.

Designing Input Signals

You can generate arbitrary input signals in the form of a sine wave, square
wave, step function, or white noise after opening the Linear Simulation Tool
(see “Opening the Linear Simulation Tool” on page 3-22).

For information about importing inputs from the MATLAB workspace or from
a file, see “Importing Input Signals” on page 3-26. For an overview of working
with the Linear Simulation Tool, see “Working with the Linear Simulation
Tool” on page 3-23.

To design one or more input signals:

1 In the Linear Simulation Tool, click the Input signals tab (if it is not
displayed).

3-33
 3 Analyzing Models

2 Specify the simulation time in the Timing area. The time interval (in
seconds) is used to evaluate the input signal you design in later steps of
this procedure.

3 Select one or more rows for the signal channels you want to design. The
following figure shows an example with two selected channels.

4 Click Design signal to open the Signal Designer dialog box. The following
figure shows an example of the Signal Designer dialog box.

3-34
 Simulating Models with Arbitrary Inputs and Initial Conditions

5 In the Signal type list, select the type of signal you want to create. It
can be one of the following:
• Sine wave
• Square wave
• Step function
• White noise

6 Specify the signal characteristics. The Signal Designer dialog box contains
different options depending on which signal type you selected.

7 Click Insert. This brings the new signal into the Linear Simulation Tool.

8 Click Simulate in the Linear Simulation Tool to view the system response.

Specifying Initial Conditions

If your system is in state-space form, you can enter or import initial states
after opening the Linear Simulation Tool (see “Opening the Linear Simulation
Tool” on page 3-22).

For an overview of working with the Linear Simulation Tool, see “Working
with the Linear Simulation Tool” on page 3-23.

You can also import initial states from the MATLAB workspace.

3-35
 3 Analyzing Models

To import one or more initial states:

1 In the Linear Simulation Tool, click the Initial states tab (if it is not
already displayed).

2 In the Selected system list, select the system for which you want to
specify initial conditions.

3 You can either enter state values in the Initial value column, or import
values from the MATLAB workspace by clicking Import state vector.
The following figure shows an example of the import window:

Note For n-states, your initial-condition vector must have n entries.

4 After specifying the initial states, click Simulate in the Linear Simulation
Tool to view the system response.

3-36
 Functions for Time and Frequency Response

Functions for Time and Frequency Response

In this section...
“When to Use Functions for Time and Frequency Response” on page 3-37
“Time and Frequency Response Functions” on page 3-37
“Plotting MIMO Model Responses” on page 3-40
“Data Markers” on page 3-42
“Plotting and Comparing Multiple Systems” on page 3-44
“Creating Custom Plots” on page 3-47

When to Use Functions for Time and Frequency

Response
You can use the LTI Viewer GUI for a wide range of applications. There
are situations, however, where you may want a more open and extensible
environment. You can use the Control System Toolbox functions for basic
time and frequency domain analysis plots used in control system engineering.
These functions apply to any kind of linear model (continuous or discrete,
SISO or MIMO, or arrays of models). You can only apply the frequency
domain analysis functions to FRD models.

Use the LTI Viewer when a GUI-driven environment is desirable. On the

other hand, use functions when you want customized plots. If you want to
include data unrelated to your models, you must use functions instead of the
LTI Viewer (which only plots model data).

The next sections discuss time and frequency response functions and how to
use these functions to create customized plots of linear model responses.

Time and Frequency Response Functions

Time responses investigate the time-domain transient behavior of linear
models for particular classes of inputs and disturbances. You can determine
such system characteristics as rise time, settling time, overshoot, and
steady-state error from the time response. You can use the Control System
Toolbox functions for step response, impulse response, initial condition

3-37
 3 Analyzing Models

response, and general linear simulations. For example, you can simulate the
response to white noise inputs using lsim and the MATLAB function randn.

In addition to time-domain analysis, you can use the Control System Toolbox
functions for frequency-domain analysis using the following standard plots:

• Bode
• Nichols
• Nyquist
• Singular value

This table lists available time and frequency response functions and their use.

Functions for Frequency and Time Response

Functions Description
bode Bode plot
evalfr Computes the frequency response at a single complex
frequency (not for FRD models)
freqresp Computes the frequency response for a set of
frequencies
gensig Input signal generator (for lsim)
impulse Impulse response plot
initial Initial condition response plot
iopzmap Pole-zero map for each I/O pair of an LTI model
lsim Simulation of response to arbitrary inputs
margin Computes and plots gain and phase margins
nichols Nichols plot
nyquist Nyquist plot
pzmap Pole-zero map
step Step response plot

3-38
 Functions for Time and Frequency Response

Functions for Frequency and Time Response (Continued)

Functions Description
hsvd Compute Hankel singular values of LTI model
bodemag Bode magnitude response of LTI models

These functions can be applied to single linear models or LTI arrays.

The functions step, impulse, and initial automatically generate an

appropriate simulation horizon for the time response plots. For example,

h = tf([1 0],[1 2 10])

impulse(h)

produces the following plot.

Impulse Response of a SISO Model

3-39
 3 Analyzing Models

Frequency-domain plots automatically generate an appropriate frequency

range as well.

Plotting MIMO Model Responses

For MIMO models, time and frequency response functions produce an array
of plots with one plot per I/O channel (or per output for initial and lsim).
For example,

h = [tf(10,[1 2 10]) , tf(1,[1 1])]

step(h)

produces the following plot.

Step Responses for a MIMO Model

The simulation horizon is automatically determined based on the model

dynamics. You can override this automatic mode by specifying a final time,

step(h,10) % Simulates from 0 to 10 seconds

3-40
 Functions for Time and Frequency Response

or a vector of evenly spaced time samples.

t = 0:0.01:10 % Time samples spaced every 0.01 second

step(h,t)

Right-Click Menus
All the time and frequency response functions provide right-click menus
that allow you to customize your plots. For more information on using the
LTI Viewer right-click menus, see “Using the Right-Click Menu in the LTI
Viewer”. This figure shows the plots from Step Responses for a MIMO Model
on page 3-40, with the right-click menu open.

Using the Right-Click Menu in a Step Response Plot

The options you can select include

3-41
 3 Analyzing Models

• Systems — Select or clear any models that you included when you created
the response plot.
• Characteristics — Add information about the plot. The characteristics
available change from plot to plot. For example, Bode plots have stability
margins available, but step responses have rise time and steady-state
values available.
• Axes Grouping — Change the grouping of your plots. Available options
are All, None, Inputs, and Outputs. You can group all the plots together,
place each in a separate plot region (none), or group the inputs or outputs
together.
• I/O Selector — Open the I/O Selector dialog box.

Use this dialog box to select/clear which inputs and outputs to plot.
• Normalize — Scale responses to fit the view (only available for
time-domain plot types).
• Full View — Use automatic limits to make the entire curve visible.
• Grid — Add grids to your plots.
• Properties — Open the Property Editor, which you can use to customize
various attributes of your plot. See Customization for a full description of
the Property Editor.
Alternatively, you can open the Property Editor by double-clicking in an
empty region of the response plot.

Data Markers
In addition to right-click menus, you can use plot data markers. These allow
you to identify key data points on your plots. This figure, using the same
plot as Step Responses for a MIMO Model on page 3-40, shows markers on
the plots.

3-42
 Functions for Time and Frequency Response

Using Plot Markers to Identify Data Points

You can move a data marker by

• Grabbing the black square located at the corner of the marker

• Dragging the marker with your mouse

The time and amplitude values will change as you move the marker. This
does not apply to markers that display plot characteristics (e.g., peak value or
rise time). In the case of plot characteristic data markers, you can view them
by placing your cursor over the dot that represents the active characteristic.
To make the data marker persistent, left-click the marker.

Note Data markers do not apply to the SISO Design Tool, which displays
data about plot characteristics in the status pane at the bottom of the SISO
Design Tool window.

3-43
 3 Analyzing Models

Right-Click Menus
Right-click on any data marker to open a property menu for the marker.

Property options for the marker include

• Alignment — Change the position of the marker. Available options are

top-right, top-left, bottom-right, and bottom-left.
• FontSize — Change the font size.
• Movable — By default, you can move data markers by clicking and
dragging. Clearing Movable forces the marker to remain at a fixed data
point.
• Delete — Remove the selected marker. Alternatively, left-click anywhere
in the empty plot region to delete all markers in the plot
• Interpolation — By default, data markers linearly interpolate between
points along the plotted curve. Select None to force the markers to snap to
nearest points along the plotted curve.

Since characteristic data markers are by definition fixed, the right-click

menus for them have fewer options.

These options work the same as they do for the full right-click menu.

Plotting and Comparing Multiple Systems

You can use the command-line response-plotting functions to plot the
response of continuous and discrete linear models on a single plot. To do

3-44
 Functions for Time and Frequency Response

so, invoke the corresponding command-line function using the list sys1,...,
sysN of models as the inputs.

stepplot(sys1,sys2,...,sysN)
impulseplot(sys1,sys2,...,sysN)
...
bodeplot(sys1,sys2,...,sysN)
nicholsplot(sys1,sys2,...,sysN)
...

All models in the argument lists of any of the response plotting functions
(except for sigma) must have the same number of inputs and outputs.
To differentiate the plots easily, you can also specify a distinctive
color/linestyle/marker for each system just as you would with the plot
command. For example,

bodeplot(sys1,'r',sys2,'y--',sys3,'gx')

plots sys1 with solid red lines, sys2 with yellow dashed lines, and sys3 with
green x markers.

You can plot responses of multiple models on the same plot. These models do
not need to be all continuous-time or all discrete-time.

Example: Comparing Continuous and Discretized Systems

The following example compares a continuous model with its zero-order-hold
discretization.

sysc = tf(1000,[1 10 1000])

sysd = c2d(sysc,0.2)
% ZOH sampled at 0.2 second

stepplot(sysc,'--',sysd,'-')
% Compare step responses

These commands produce the plot shown below.

3-45
 3 Analyzing Models

Comparison of a Continuous Model to Its Discretized Version

Use this command to compare the Bode plots of the two systems.

bodeplot(sysc,'--',sysd,'-')
% Compare Bode responses

The following plot results from the last command.

3-46
 Functions for Time and Frequency Response

Comparison of Bode Plots for a Continuous Model and Its Discretized Version

A comparison of the continuous and discretized responses reveals a drastic

undersampling of the continuous-time system. Specifically, there are hidden
oscillations in the discretized time response and aliasing conceals the
continuous-time resonance near 30 rad/sec.

Creating Custom Plots

Time and frequency response commands are useful for creating custom plots.
You can mix model response plots with other data views using response
commands together with plot, subplot, and hold.

Example: Custom Plots

For example, the following sequence of commands displays the Bode plot, step
response, pole/zero map, and some additional data in a single figure window.

h = tf([4 8.4 30.8 60],[1 4.12 17.4 30.8 60]);

subplot(221)

3-47
 3 Analyzing Models

bodeplot(h)
subplot(222)
stepplot(h)
subplot(223)
pzplot(h)
subplot(224)
plot(rand(1, 100)) % Any data can go here
title('Some noise')

Your plot should look similar to this illustration.

Example of Creating a Custom Plot

For information about plot, subplot, hold, and other options for plotting
general data, see Basic Plots and Graphs in the MATLAB Function Reference.
These documents are available in the MATLAB online help.

3-48
 Functions for Time and Frequency Response

Note Each of the plots generated by response analysis functions in Example

of Creating a Custom Plot on page 3-48 (bodeplot, stepplot, and pzplot)
has its own right-click menu (similar to those in the LTI Viewer). To activate
the right-click menus, place your mouse in the plot region and right-click. The
menu contents depend on what type of plot you have selected.

3-49
 3 Analyzing Models

3-50
 4

Designing Compensators

• “Designing PID Controllers” on page 4-2

• “SISO Design Tool” on page 4-15
• “Bode Diagram Design” on page 4-55
• “Root Locus Design” on page 4-81
• “Nichols Plot Design” on page 4-103
• “Automated Tuning Design” on page 4-114
• “Multi-Loop Compensator Design” on page 4-120
• “Control Design Analysis of Multiple Models” on page 4-132
• “Functions for Compensator Design” on page 4-147
 4 Designing Compensators

Designing PID Controllers

Control System Toolbox software gives you several tools for designing PID
controllers.

Tool When to use

• PID Tuner GUI Use to design a SISO PID controller
in the feed-forward path of
• Command-line PID tuning
single-loop, unity-feedback control
configuration.

r
+ u y
C sys
-

SISO Design Tool Use to design a SISO PID controller

in any other loop configuration.

Designing PID Controllers with the PID Tuner GUI

• “PID Tuner Overview” on page 4-2
• “PID Controller Type” on page 4-3
• “PID Controller Form” on page 4-5
• “Analyzing Your Design Using Response Plots” on page 4-5
• “Example: Designing a Controller for Fast Reference Tracking” on page 4-7

PID Tuner Overview

Use the PID Tuner GUI to interactively design a SISO PID controller in the
feed-forward path of single-loop, unity-feedback control configuration.

r
+ u y
C sys
-

4-2
 Designing PID Controllers

The PID Tuner automatically designs a controller for your plant. You specify
the controller type (P, I, PI, PD, PDF, PID, PIDF) and form (parallel or
standard). You can analyze the design using a variety of response plots, and
interactively adjust the design to meet your performance requirements.

To launch the PID Tuner, use the pidtool command:

pidtool(sys,type)

where sys is a linear model of the plant you want to control, and type is a
string indicating the controller type to design.

Alternatively, enter

pidtool(sys,Cbase)

where Cbase is a baseline controller, allowing you to compare the performance

of the designed controller to the performance of Cbase.

For more information about sys and Cbase, see the pidtool reference page.

PID Controller Type

The PID Tuner can tune up to seven types of controllers. To select the
controller type, use one of these methods:

• Provide the type argument to the launch command pidtool.

• Provide a baseline controller Cbase to the launch command pidtool. The
PID Tuner designs a controller of the same type as Cbase.
• Use the Type menu to change controller types after launching the PID
Tuner.

4-3
 4 Designing Compensators

type Entry Controller Type Continuous-time Discrete-time

input in Type Controller Formula Controller Formula
to menu (parallel form) (parallel form,
pidtool ForwardEuler
integrator formulas)
'p' P Proportional only Kp Kp
'i' I Integral only
Ki Ts
Ki
s z −1
'pi' PI Proportional and
integral Ki Ts
Kp + K p + Ki
s z −1
'pd' PD Proportional and
derivative K p + Kds z −1
K p + Kd
Ts

'pdf' PDF Proportional and

derivative with Kds 1
first-order filter on Kp + K p + Kd
Tf s + 1 Ts
derivative term Tf +
z −1
'pid' PID Proportional, integral,
and derivative Ki Ts z −1
Kp + + Kds K p + Ki + Kd
s z −1 Ts

'pidf' PIDF Proportional, integral,

and derivative with Ki Kds Ts 1
Kp + + K p + Ki + Kd
first-order filter on z −1 Ts
s Tf s + 1 Tf +
derivative term z −1

If sys is a discrete-time model with sampling time Ts, the PID Tuner designs
a discrete-time pid controller using the ForwardEuler discrete integrator
formula. To design a controller that has different discrete integrator formulas,
use one of the following methods:

4-4
 Designing PID Controllers

• Provide a discrete-time baseline controller Cbase to the launch command

pidtool. The PID Tuner designs a controller that has the same discrete
integrator formulas as Cbase.

• After launching the PID Tuner, click the button to open the PID
Tuner Settings dialog box. Select discrete integrator formulas from the
Integral Formula and Derivative Formula menus.

For more information about discrete integrator formulas, see the pid and
pidstd reference pages.

PID Controller Form

When you use the type input to launch the PID Tuner, the PID Tuner designs
a controller in parallel form. To design a controller in standard form, use
one of the following methods:

• Provide a standard-form baseline controller Cbase to the launch command

pidtool. The PID Tuner designs a controller of the same form as Cbase.
• Use the Form menu to change controller form after launching the PID
Tuner.

For more information about parallel and standard controller forms, see the
pid and pidstd reference pages.

Analyzing Your Design Using Response Plots

Select response plots from the Response menu to analyze the controller’s
performance. The PID Tuner computes the system responses assuming the
following architecture:

d1 d2
PID Plant
r + u + +
- C sys y
+ +

The following table summarizes the available response plots.

4-5
 4 Designing Compensators

Response Plotted System Description

Reference tracking Shows the closed-loop system
C ∗ sys response to a change in
(from r to y) setpoint. Use when your
1 + C ∗ sys
design specifications include
setpoint tracking.
Controller effort Shows the closed-loop
C controller output response
(from r to u) to a change in setpoint. Use
1 + C ∗ sys
when your design is limited by
practical constraints, such as
controller saturation.
Input disturbance Shows the closed-loop system
Rejection sys response to load disturbance
(from d1 to y) (a disturbance at the plant
1 + C ∗ sys
input). Use when your design
specifications include input
disturbance rejection.
Output disturbance Shows the closed-loop system
Rejection 1 response to measurement
(from d2 to y) noise.
1 + C ∗ sys
Open-loop C*sys Shows response of the
open-loop controller-plant
system. Use for
frequency-domain design.
Use when your design
specifications include
robustness criteria such
as open-loop gain margin and
phase margin.
Plant sys Shows the plant response. Use
to examine plant dynamics.

Choose time-domain (Step) or frequency-domain (Bode) responses from the

Plot menu.

4-6
 Designing PID Controllers

If you provide Cbase, check Show baseline to display the response of the
baseline controller.

Example: Designing a Controller for Fast Reference Tracking

In this example, you use the PID Tuner to design a controller for the plant:

1
sys = .
(s + 1)3

The design requirements are for the closed loop system to track a reference
input with a rise time less than 1.5 s, and settling time less than 6 s.

1 Launch the PID Tuner to design a PI controller for a first pass design.

sys = zpk([],[-1 -1 -1],1); % create the plant model

pidtool(sys,'pi')

Upon launch, the PID Tuner displays the closed-loop step response of the
system with a PI controller designed for a balance between performance
(response time) and robustness (stability margins).

4-7
 4 Designing Compensators

2 Examine the reference tracking rise time and settling time.

Right-click on the plot and select Characteristics > Rise Time to

mark the rise time as a blue dot on the plot. Select Characteristics >
Settling Time to mark the settling time. To see tool-tips with numerical
values, click each of the blue dots.

4-8
 Designing PID Controllers

The PID Tuner default PI controller provides a rise time of 2.35 s

and settling time of 10.7 s. Both results are slower than the design
requirements.

4-9
 4 Designing Compensators

Note To display the performance metrics in a table instead of in tool-tips

on the plot, click the Show parameters button . This action expands
the GUI to show performance and robustness metrics. It also displays
the controller gains.

3 Use the Response time slider to try to improve the loop performance.

Slide the slider to the right to make the loop faster. The response plot
automatically updates with the new design.

Moving the slider far enough to meet the rise time requirement of less
than 1.5 s, however, results in unacceptably long settling time. There is
also more oscillation. To achieve the faster response speed, the algorithm
must sacrifice stability.

4-10
 Designing PID Controllers

4 Change the controller type to improve the response.

Adding derivative action to the controller gives the PID Tuner more
freedom to achieve adequate phase margin with the desired response speed.

Select PIDF from the Type menu. The PID Tuner designs a new PIDF
controller. (See “PID Controller Type” on page 4-3 for more information
about available controller types.)

4-11
 4 Designing Compensators

The rise time and settling time now meet the design requirements. You can
use the Response time slider to make further adjustments to the response.
To revert to the default automated tuning result, click the button.

Note To adjust the closed-loop bandwidth and phase margin separately,

select Extended from the Design mode menu . The bandwidth is inversely
proportional to the response time.

4-12
 Designing PID Controllers

5 Analyze other system responses, if appropriate.

To analyze other system responses, use the Response menu to choose

a response plot and the Plot menu to choose time-domain (Step) or
frequency-domain (Bode) responses.

For example, to observe the closed-loop system response to disturbance at

the plant input, select Input disturbance rejection from the Response
menu.

4-13
 4 Designing Compensators

See “Analyzing Your Design Using Response Plots” on page 4-5 for more
information about available response plots.

6 Export your controller design to the MATLAB workspace.

To export your final controller design to the MATLAB workspace, click .

The PID Tuner GUI exports the controller as a
• pid controller object, if the Form is Parallel
• pidstd controller object, if the Form is Standard

For more examples of controller design using the PID Tuner, see the Control
System Toolbox demo, Designing PID for Disturbance Rejection with PID
Tuner

Designing PID Controllers at the Command Line

To design PID Controllers at the command line, use the pidtune command.
For more information and examples, see:

• pidtune reference page

• Control System Toolbox demo: Designing Cascade Control System with PI
Controllers

4-14
 SISO Design Tool

SISO Design Tool

In this section...
“Quick Start for Tuning Compensator Parameters Using the SISO Design
Tool” on page 4-15
“Components of the SISO Tool” on page 4-26
“Design Options in the SISO Tool” on page 4-26
“Opening the SISO Design Tool” on page 4-27
“Using the SISO Design Task Node in the Control and Estimation Tools
Manager” on page 4-29
“Importing Models into the SISO Design Tool” on page 4-30
“Feedback Structure” on page 4-34
“Analysis Plots for Loop Responses” on page 4-36
“Using the Graphical Tuning Window” on page 4-42
“Exporting the Compensator and Models” on page 4-49
“Storing and Retrieving Intermediate Designs” on page 4-51

Quick Start for Tuning Compensator Parameters

Using the SISO Design Tool
In this Quick Start, you get an overview of the steps for tuning compensator
parameters. Before you begin, you need:

• A model in the MATLAB workspace representing your plant

For information on how to create a model, see Chapter 2, “Building Models”.
• Design requirements for your system

To tune parameters:

1 Open the control design GUIs by typing the following in the MATLAB
Command Window:

sisotool

4-15
 4 Designing Compensators

This command opens the Control and Estimation Tools Manager and the
SISO Design Tool, as shown in the following figure.

4-16
 SISO Design Tool

2 Define the control architecture for your system.

a In the Architecture tab of the Control and Estimation Tools Manager,

click Control Architecture.
b In the Control Architecture dialog box, select the control architecture.

c Specify the sign of summing junctions as + or -.

Select
control
architecture

Specify sign of
summing junctions
as + or -

4-17
 4 Designing Compensators

3 Specify the plant model and initial guesses for the compensator elements
in the control architecture.
a In the Architecture tab of the Control and Estimation Tools Manager,
click System Data.
b In the System Data dialog box, browse for a model in the MATLAB
workspace.

Browse for
model in MATLAB
workspace

4-18
 SISO Design Tool

4 Design a compensator using automated tuning, for example PID Tuning.

a In the Automated Tuning tab of the Control and Estimation Tools

Manager, select an automated tuning method.
b Specify tuning options.

c Click Update Compensator to tune parameters.

Select
automated
tuning method

Specify
tuning options

Tune parameters

4-19
 4 Designing Compensators

5 Evaluate the system’s closed-loop performance.

a Plot the system response.

i In the Analysis Plots tab, select a plot type.

ii Select the type of response for each plot.

Select plot
type

Select the type

of response for
each plot

4-20
 SISO Design Tool

b Display specific performance characteristic for your system. Compare

values to design requirements.
iii Right-click to select performance characteristics.
iv Click on the dot that appears to view the characteristic value.

Click dot
to view value

Right-click
to select
performance
characteristics

4-21
 4 Designing Compensators

6 Refine compensator design using graphical tuning.

a Create plots for graphical tuning.

i In the Graphical Tuning tab, select the loop to tune in each plot.
ii Select plot types.

Select the loop to

tune in each plot

Select plot types

4-22
 SISO Design Tool

b In a design plot, modify the compensator by adding poles, zeros, lead,

and lag, for example. To do this, right-click to add dynamic elements
in the controller structure.

Right-click to
add dynamic
elements in
controller structure

c Add a new pole, for example, by clicking the pole location on the plot.

4-23
 4 Designing Compensators

d Modify the compensator structure by moving the poles, zeros, and other
dynamic elements directly on the plot.

Click and
drag plot

4-24
 SISO Design Tool

7 Export the compensator to the MATLAB workspace for further analysis.

a In the Control and Estimation Tools Manager, select File > Export.

b In the SISO Tool Export dialog box, select the compensator.

c Click Export to Workspace.

Select Export to
compensator workspace

4-25
 4 Designing Compensators

Components of the SISO Tool

The SISO Design Tool has the following components:

• SISO Design Task node in the Control and Estimation Tools Manager
is a user interface (UI) that facilitates the design of compensators for
single-input, single-output feedback loops through a series of interactive
tabs.
• The Graphical Tuning window is a graphical user interface (GUI) for
displaying and manipulating the Bode, root locus, and Nichols plots for the
controller currently being designed. This window is titled SISO Design
for Design Name.
• The LTI Viewer associated with the SISO Design Task. For instructions
on how to operate the LTI Viewer, see “LTI Viewer for SISO Design Task
Design Requirements”.

Design Options in the SISO Tool

The SISO Design Tool facilitates the design of compensators for single-input,
single-output feedback loops, lets you iterate on your designs and perform the
following tasks:

• Manipulate closed-loop dynamics using root locus techniques.

• Shape open-loop Bode responses.
• Add compensator poles and zeros, lead/lag networks and notch filters.
• Add and tune lead/lag networks and notch filters.
• Inspect closed-loop responses using the LTI Viewer.
• Adjust phase and gain margins.
• Perform multi-loop control design.
• Perform control design analysis for multiple models.
• Convert models between discrete and continuous time.
• Automate compensator design.

4-26
 SISO Design Tool

Opening the SISO Design Tool

You can open the SISO Design Tool using an LTI model or an array of LTI
models. This topic shows how to open the SISO Design Tool with the DC
motor example created in Chapter 2, “Building Models”

If you have not built the DC motor model, type

load ltiexamples

at the MATLAB prompt. This loads a collection of linear models, including

the DC motor. To open the SISO Design Tool and import the DC motor, type

sisotool(sys_dc)

This command opens:

• The SISO Design Task node in the Control and Estimation Tools Manager
• The Graphical Tuning window with the root locus and open-loop Bode
diagrams for the DC motor plotted by default

4-27
 4 Designing Compensators

SISO Design Task Node (Architecture Tab)

4-28
 SISO Design Tool

Graphical Tuning Window with the DC Motor Example

Using the SISO Design Task Node in the Control and

Estimation Tools Manager
The SISO Design Task node in the Control and Estimation Tools Manager
contains the following tabs for specifying controller design and behavior:

• Architecture:
- Change the feedback structure and label signals and blocks.
- Configure loops for multi-loop design by opening signals to remove the
effects of other feedback loops.
- Import LTI models or row or column arrays of LTI models into your
system.

4-29
 4 Designing Compensators

- Convert sample time of the system or switch between different sample

times to design different compensators.
- Specify nominal model and frequency grid for multimodel computations.
• Compensator Editor:
- Directly edit compensator poles, zeros, and gains.
- Add or remove compensator poles and zeros.
• Graphical Tuning:
- Configure design plots in the Graphical Tuning window.
- Use design plots to graphically manipulate system response.
• Analysis Plots:
- Configure analysis plots in the LTI Viewer.
- Use analysis plots to view the response of open- or closed-loop systems.
• Automated Tuning:
- Automatically generate compensators using optimization-based, PID,
internal model control (IMC), linear-quadratic-Gaussian (LQG), or loop
shaping methods.
- Use optimization-based methods that automatically tune the system
to satisfy design requirements (available when you have the Simulink
Design Optimization product).

Importing Models into the SISO Design Tool

If you type sisotool at the MATLAB prompt, the Control and Estimation
Tools Manager opens with the SISO Design Task node and an empty
Graphical Tuning window.

You can import an LTI model or a row or column array of LTI models by
clicking System Data in the Architecture page, as shown in the following
figure.

4-30
 SISO Design Tool

The System Data dialog box opens, as shown in the following figure.

• G is a plant modeled as an LTI model or multiple plants modeled as a row or

column array of LTI models. If H is also an array, their sizes must match.

4-31
 4 Designing Compensators

• H is the sensor modeled as an LTI model or multiple sensors modeled as

a row or column array of LTI models. If G is also an array, their sizes
must match.
• C is a controller and is an LTI model.
• F is a prefilter and is an LTI model.

The default values of G, H, C, and F are 1. See “Feedback Structure” on page

4-34 for more information.

Note If you import an array of LTI models for either G or H or both for control
design, you can use the default multimodel options or configure the options by
clicking Multimodel Configuration in the Architecture tab. For more
information, see “Control Design Analysis of Multiple Models” on page 4-132.

To import the DC motor model:

1 Select G and click Browse. The Model Import dialog box opens, as shown
in the following figure.

4-32
 SISO Design Tool

2 Select sys_dc from the Available Models list. Click Import, and then
click Close.

You can now see sys_dc in the System Data dialog box.

3 Click OK.

The Graphical Tuning window updates with the DC motor model, as shown
in the following figure.

4-33
 4 Designing Compensators

Feedback Structure
The SISO Design Tool, by default, assumes that the compensator is in the
forward path, i.e., that the feedback structure looks like this figure.

Default Feedback Structure — Compensator in the Forward Path

In this figure:

• G represents the plant

• H represents the sensor dynamics

4-34
 SISO Design Tool

• F represents the prefilter

• C represents the compensator

The default values for F, H, and C are all 1. You can see the default values in
the System Data dialog box. This means that, by default, the compensator
has unity gain.

For the DC motor, G contains the DC motor model, sys_dc.

Alternative Feedback Structures

In the Architecture page, click Control Architecture to open the Control
Architecture dialog box.

You can use the Signs and Blocks and Signals panes to change the sign of
the feedback signal into a summing junction and rename blocks and signals
in the diagram respectively. See “Block Diagram Structure Modifications”
for more details.

On any tab in the SISO Design Task node, click Show Architecture to see
the current architecture and a list of the identifiers and names associated
with the components.

4-35
 4 Designing Compensators

Analysis Plots for Loop Responses

As you design different controllers, you examine the various loop responses
for a particular design. For example, to view the closed-loop step response,
click the Analysis Plots tab. This tab lists the available responses, as shown
in the following figure.

4-36
 SISO Design Tool

Select the plot types for each plot in the Analysis Plots area. Then select the
plots to appear in the Plots list in the Contents of Plots table, as shown in
the following figure.

4-37
 4 Designing Compensators

Analysis Plots Loop Response Selection

After you select a plot, the LTI Viewer opens and displays the appropriate
response(s) opens. You can also click Show Analysis Plot to open the LTI
Viewer.

The following figure shows the resulting plot for the closed-loop step response
of the DC motor.

4-38
 SISO Design Tool

LTI Viewer Showing the Step Response for the DC Motor

The settling time of the DC motor is about 1.5 seconds, which is too slow for
many applications. Also, the plot shows a large steady-state error. You can
use Bode diagram techniques to improve the response time and steady-state
error of the DC motor step response, as described in “Bode Diagram Design”
on page 4-55.

For a row or column array of LTI models, the analysis plots show the response
of the nominal model only. To view the responses of the remaining models in
the array, right-click the plot, and select one of the following options:

• Multimodel Display > Bounds — Displays an envelope encompassing

all model responses.

4-39
 4 Designing Compensators

• Multimodel Display > Individual Responses — Displays individual

model responses.

4-40
 SISO Design Tool

The plot line in darker shade of blue is the response of the nominal model.
When you design a controller for the nominal plant, you can simultaneously
analyze the controller’s effect on the remaining model responses in the array.
For more information on analyzing control design for multiple models, see
“Control Design Analysis of Multiple Models” on page 4-132.

As you select different compensator designs, the software computes the

responses and updates the response plots in real-time in the LTI Viewer
associated with your SISO Design Task. For an array, this computation is
expensive. Therefore, real-time updates may cause delay in refreshing the
plots for:

• A large number of responses

• Responses of a large number of models

To deactivate real-time updates, unselect the Real-Time Update option.

4-41
 4 Designing Compensators

Using the Graphical Tuning Window

The SISO Design for SISO Design Task graphical tuning window is a
graphical user interface (GUI) that you use to display and manipulate the
Bode, root locus, and Nichols plots for the controller currently being designed.
You can accomplish most control design tasks using the tabs in the SISO
Design Task node in the Control and Estimation Tools Manager.

The graphical tuning window shows the plots configured in the Graphical
Tuning tab.

4-42
 SISO Design Tool

This topic describes some of the methods you can use to navigate in the
Graphical Tuning window and manipulate the window’s appearance.

Graphical Tuning Window Display

The Graphical Tuning window shows:

• Poles as x’s
• Zeros as o’s
• Gain and phase margins (by default) in the lower-left corners of the Bode
magnitude and phase plots

For the DC motor, the graphical tuning window resembles the following figure.

4-43
 4 Designing Compensators

For a row or column arrays of LTI models, the plots show the individual
responses and poles and zeros of all models in the array, as shown in the
following figure.

4-44
 SISO Design Tool

The plot line in darker shade of blue represents the response of the nominal
model. This plot also displays the response characteristics of the nominal
model.

You can change the display to view an envelope that encompasses all
individual model responses. To do so, right-click the plot, and select
Multimodel Display > Bounds. The bounds resemble the following figure.

4-45
 4 Designing Compensators

As you are designing a controller for the nominal plant, you can
simultaneously analyze the controller’s effect on the remaining models in the
array. The software computes the response for each model and plots them.
This computation is expensive and results in a slower refresh of the plots for:

• Large number of responses

• Responses for a large number of models
• Dense frequency grid

4-46
 SISO Design Tool

To speed up the plot updates, change the display to plot the envelope instead
of individual responses. For more information, see “Control Design Analysis
of Multiple Models” on page 4-132.

Changing Units on a Plot

The SISO Design Tool provides editors for setting plot options in the Graphical
Tuning window. For example, if you want to change the frequency units on
all the Bode plots created in the SISO Design Tool from rad/s to Hertz, select
SISO Tool Preferences from the Edit menu in the SISO Design Task node
on the Control and Estimation Tools Manager, as shown in the next figure.

This opens the SISO Tool Preferences dialog box.

4-47
 4 Designing Compensators

Use the options on the Units page to make the change. This unit change
persists for the entire session.

For more information about property and preference settings, see

Customization.

Right-Click Menus
The SISO Design Tool has right-click menus available in any of the plot
regions. Open the Bode magnitude menu by right-clicking in the Bode
magnitude plot. The following menu appears.

Right-Click Menu for the Bode Magnitude Plot

4-48
 SISO Design Tool

Although the menus for each plot generally contain the same options, there
are some options specific to each plot type:

• The Open-Loop Bode Editor has a Gain Target option, the Closed-Loop
Bode Editor has a Select Compensator option instead.
• For arrays of LTI models, the Multimodel Display option for the Root
Locus Editor has an option to show the poles and zeros of the nominal or all
models, instead of Bounds and Individual Responses options.

The right-click menus contain numerous features. The DC motor example

uses many of the available features; for a complete discussion of the right-click
menus, see the help for the SISO Design Tool."

Exporting the Compensator and Models

After you design the controller, you may want to save your design parameters
for future implementation. You can do this by selecting:

• File > Export in the Control and Estimation Tools Manager

• File > Export in the Graphical Tuning window.

The SISO Tool Export dialog box opens:

4-49
 4 Designing Compensators

• Select models to export area shows a list of models for the components of
your designs. For a plant or sensor modeled as row or column array of LTI
models, the components are also arrays of LTI models.
• Export As column displays variables with default names or previously
named in the System Data dialog box.
• Select design drop-down list lets you view the components of other
designs.

To export your controller to the workspace:

1 Select Compensator C in the Component column. If you want to change

the name, double-click in the cell for Compensator C.

To select multiple components, use the Shift key if they are all adjacent
and the Ctrl key if they are not.

2 Click Export to Workspace to export the compensator to the MATLAB

workspace.

3 At the MATLAB prompt, type

4-50
 SISO Design Tool

who

to view the compensator variable C.

4 To see the format in which this variable is stored, type

Clicking Export to Disk opens the Export to Disk dialog box.

You can save your models as MAT-files in any folder. The default name
for the MAT-file is the name of your original model. You can also specify a
name for the MAT-file. If you save multiple components, they are stored
in a single MAT-file.

Storing and Retrieving Intermediate Designs

You can store and retrieve intermediate compensators while you iterate on
your compensator design. To store intermediate designs, click the Design
History node or Store Design, both located on the SISO Design Task node
in the Control and Estimation Tools Manager.

Alternatively, you can select Store/Retrieve from the Designs menu in

the Graphical Tuning window. Using either method, the following Design
History page opens.

4-51
 4 Designing Compensators

If you have any intermediate designs already stored, they will appear on
the Design History page.

Click Store Design to save the current design with the default name Design;
the suffix increments when you store additional compensators without
renaming them. You can rename the design by right-clicking the name under
the node and selecting Rename.

To retrieve intermediate designs, again click the Design History node or

select Store/Retrieve from the Designs menu. From the Design History
page, select the design to retrieve, and then click Retrieve Design, as shown
next.

4-52
 SISO Design Tool

Design History Page Listing Current Designs

The Graphical Tuning window automatically reverts to the selected

compensator design.

Click any design name in the Design History to view a snapshot summary of
the design, as shown in the following figure.

4-53
 4 Designing Compensators

Design Snapshot Summary

Return to the compensator list by clicking the Design History node.

You can delete an intermediate design by selecting it from the Design

History page and clicking Delete.

4-54
 Bode Diagram Design

Bode Diagram Design

In this section...
“What Is Bode Diagram Design?” on page 4-55
“Example: DC Motor” on page 4-55
“Adjusting the Compensator Gain” on page 4-56
“Adjusting the Bandwidth” on page 4-57
“Adding an Integrator” on page 4-59
“Adding a Lead Network” on page 4-63
“Moving Compensator Poles and Zeros” on page 4-68
“Adding a Notch Filter” on page 4-71
“Modifying a Prefilter” on page 4-76

What Is Bode Diagram Design?

One technique for compensator design is to work with Bode diagrams of the
open-loop response (loop shaping).

Using Bode diagrams, you can

• Design to gain and phase margin specifications

• Adjust the bandwidth
• Add notch filters for disturbance rejection

Example: DC Motor
The following topics use the DC motor example to show how to create a
compensator using Bode diagram design techniques. From “SISO Example:
The DC Motor” on page 2-3, the transfer function of the DC motor is

Transfer function:
1.5
------------------
s^2 + 14 s + 40.02

4-55
 4 Designing Compensators

For this example, the design criteria are as follows:

• Rise time of less than 0.5 second

• Steady-state error of less than 5%
• Overshoot of less than 10%
• Gain margin greater than 20 dB
• Phase margin greater than 40 degrees

Adjusting the Compensator Gain

The LTI Viewer Showing the Step Response for the DC Motor on page 4-39,
shows that the closed-loop step response is too slow. The simplest approach to
speeding up the response is to increase the gain of the compensator.

To increase the gain:

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 Select C from the compensator selection list.

3 In the text box to the right of the equal sign in the Compensator area,
enter 38 and press Enter.

4-56
 Bode Diagram Design

Adjusting Compensator Gain on the Compensator Editor Page

The SISO Design Tool calculates the compensator gain, and Bode and root
locus graphs in the Graphical Tuning window are updated.

Alternatively, you can set the gain in the Graphical Tuning window by
grabbing the Bode magnitude line and dragging it upward. The gain and poles
change as the closed-loop set point is recomputed, and the new compensator
value is updated in the Compensator Editor page.

Adjusting the Bandwidth

Because the design requirements include a 0.5-second rise time, try setting
the gain so that the DC crossover frequency is about 3 rad/s. The rationale for
setting the bandwidth to 3 rad/s is that, to a first-order approximation, this
should correspond to about a 0.33-second time constant.

4-57
 4 Designing Compensators

To make the crossover easier to see, select Grid from the right-click menu.
This creates a grid for the Bode magnitude plot. Left-click the Bode magnitude
plot and drag the curve until you see the curve crossing over the 0 dB line (on
the y axis) at 3 rad/s. This changes both the SISO Design Tool display and
the LTI Viewer step response.

For a crossover at 3 rad/s, the compensator gain should be about 38. By

default, the Graphical Tuning window shows gain and phase margin
information in the lower-left corners of the Bode diagrams. In the Bode
magnitude plot, it also tells you if your closed-loop system is stable or unstable.

This figure shows the Graphical Tuning window.

Adjusting Bandwidth in the Graphical Tuning Window

This plot shows the associated closed-loop step response in the LTI Viewer.

4-58
 Bode Diagram Design

Closed-Loop Step Response for the DC Motor with a Compensator Gain = 38

The step response shows that the steady-state error and rise time have
improved somewhat, but you must design a more sophisticated controller
to meet all the design specifications, in particular, the steady-state error
requirement.

Adding an Integrator
One way to eliminate steady-state error is to add an integrator. To add an
integrator:

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 Right-click anywhere in the Dynamics table for the right-click menu, and
then select Add Pole/Zero > Integrator.

The following figures show this process.

4-59
 4 Designing Compensators

Adding an Integrator in the Dynamics Table

4-60
 Bode Diagram Design

Editable Integrator Parameters

Notice adding the integrator changed the crossover frequency of the system.
Readjust the compensator gain in the Compensator Editor page to bring
the crossover back to 3 rad/s; the gain should be 99.

After you have added the integrator and readjusted the compensator gain, the
Graphical Tuning window shows a red ‘x’ at the origin of the root locus plot.

4-61
 4 Designing Compensators

Integrator on the Root Locus Plot

The following figure shows the closed-loop step response.

4-62
 Bode Diagram Design

Step Response for the DC Motor with an Integrator in the Compensator

Use the right-click menu to show the peak response and rise time (listed
under the Characteristics). The step response is settling around 1, which
satisfies the steady-state error requirement. This is because the integrator
forces the system to zero steady-state error. The figure shows, however, that
the peak response is 1.3, or about 30% overshoot, and that the rise time is
roughly 0.4 second. So a compensator consisting of an integrator and a gain
is not enough to satisfy the design requirements, which require that the
overshoot be less than 10%.

Adding a Lead Network

Part of the design requirements is a gain margin of 20 dB or greater and a
phase margin of 40° or more. In the current compensator design, the gain
margin is 11.5 dB and the phase margin is 38.1°, both of which fail to meet the
design requirements. The rise time needs to be shortened while improving
the stability margins. One approach is to increase the gain to speed up the
response, but the system is already underdamped, and increasing the gain
will decrease the stability margin as well. You might try experimenting with

4-63
 4 Designing Compensators

the compensator gain to verify this. The only option left is to add dynamics to
the compensator.

One possible solution is to add a lead network to the compensator. To add

the lead network:

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 In the Dynamics table, right-click and then select Add Pole/Zero > Lead.

The following figures show the process of adding a lead network to your
controller.

Adding a Lead Network to the DC Motor Compensator on the Compensator

Editor Page

4-64
 Bode Diagram Design

Lead Network Added

Editable fields are shown in the Edit Selected Dynamics group box (right
side of page) when an item in the Dynamics table has been selected, as
shown in the following figure.

4-65
 4 Designing Compensators

For this example, change Real Zero to -7.38 and change Real Pole to -11.1.

You can also add a lead network using the Graphical Tuning window.
Right-click in the Bode graph, select Add Pole/Zero > Lead, place the ‘x’
on the plot where you want to add the lead network, and then left-click to
place it. The Compensator Editor page is updated to include the new lead
network in the Dynamics table.

Your Graphical Tuning window and LTI Viewer plots should now look similar
to these.

4-66
 Bode Diagram Design

Root Locus, Bode, and Step Response Plots for the DC Motor with a Lead
Network

4-67
 4 Designing Compensators

The Step Response plot shows that the rise time is now about 0.4 second and
peak response is 1.24 rad/s (i.e., the overshoot is about 25%). Although the rise
time meets the requirement, the overshoot is still too large, and the stability
margins are still unacceptable, so you must tune the lead parameters.

Moving Compensator Poles and Zeros

To improve the response speed, edit the selected dynamics for the lead
network in the Edit Selected Dynamics group box on the Compensator
Editor page.

1 Change the value of the lead network zero (Real Zero) to move it closer to
the left-most (slowest) pole of the DC motor plant (denoted by a blue ‘x’).

2 Change the value of the lead network pole (Real Pole) to move it to the
right. Notice how the gain margin increases (as shown in the Graphical
Tuning window) as you do this.

4-68
 Bode Diagram Design

As you tune these parameters, look at the LTI Viewer. You will see the
closed-loop step response alter with each parameter change you make.
The following figure shows the final values for a design that meets the
specifications.

Graphical Tuning Window with Final Design Parameters for the DC Motor
Compensator

The values for this final design are as follows:

• Poles at 0 and -28

• Zero at -4.3
• Gain = 84

4-69
 4 Designing Compensators

Enter these values directly in the Edit Selected Dynamics group box in the
Compensator Editor page, shown as follows (Integrator is already set to 0).

Entering Final Design Parameters on the Compensator Editor Page

The following figure shows the step response for the final compensator design.

4-70
 Bode Diagram Design

Step Response for the Final Compensator Design

In the LTI Viewer’s right-click menu, select Characteristics > Peak

Response and Characteristics > Rise Time to show the peak response and
rise time, respectively. Hover the mouse over the blue dots to show the data
markers. The step response shows that the rise time is 0.45 second, and the
peak amplitude is 1.03 rad/s, or an overshoot of 3%. These results meet the
design specifications.

Adding a Notch Filter

If you know that you have disturbances to your system at a particular
frequency, you can use a notch filter to attenuate the gain of the system at
that frequency. To add a notch filter, click the Compensator Editor tab to
open the Compensator Editor page. Right-click in the Dynamics table and
select Add Pole/Zero > Notch, as shown next.

4-71
 4 Designing Compensators

Adding a Notch Filter with the Dynamics Right-Click Menu

Default values for the filter are supplied, as shown next.

4-72
 Bode Diagram Design

Notch Filter Default Values

The following figure shows the result in the Graphical Tuning window.

4-73
 4 Designing Compensators

Notch Filter Added to the DC Motor Compensator

To see the notch filter parameters in more detail, click the Zoom In

icon on the Graphical Tuning window. In the Open-Loop Bode Editor, press
the left mouse button and drag your mouse to draw a box around the notch
filter. When you release the mouse, the Graphical Tuning window will zoom
in on the selected region.

To understand how adjusting the notch filter parameters affects the filter,
consider the notch filter transfer function.

s2 + 21n s + n2
s2 + 22n s + n2

4-74
 Bode Diagram Design

The three adjustable parameters are ξ1, ξ2, and ωn. The ratio of ξ2/ξ1 sets the
depth of the notch, and ωn is the natural frequency of the notch.

This diagram shows how moving the red ⊗ and black diamonds changes these
parameters, and hence the transfer function of the notch filter.

A Close Look at Notch Filter Parameters

In the Dynamics table on the Compensator Editor page, select the row
containing the newly added notch filter. The editable fields appear in the
Edit Selected Dynamics group box, as shown next.

4-75
 4 Designing Compensators

Editing Notch Filter Parameters

Modifying a Prefilter
You can use the SISO Design Tool to modify the prefilter in your design.
Typical prefilter applications include:

• Achieving (near) feedforward tracking to reduce load on the feedback loop

(when stability margins are poor)
• Filtering out high frequency content in the command (reference) signal to
limit overshoot or to avoid exciting resonant modes of the plant

A common prefilter is a simple lowpass filter that reduces noise in the input
signal.

4-76
 Bode Diagram Design

Open the Bode diagram for the prefilter by opening the right-click menu in
the Closed-Loop Bode Editor in the Graphical Tuning window, and then
selecting Select Compensators > F(F).

Selecting the Prefilter in the Graphical Tuning Window

For clarity, the previous figure does not show the open-loop Bode diagram for
the compensator (C). To remove the Bode diagram from the Graphical Tuning
window, go to the SISO Design Task node on the Control and Estimation
Tools Manager, click the Graphical Tuning tab, and for Plot 2, Open Loop 1,
select Plot type None.

4-77
 4 Designing Compensators

Prefilter Bode Diagram

If you haven’t imported a prefilter, the default is a unity gain. You can add
poles and zeros and adjust the gain using the same methods as you did when
designing the compensator (C) on the Compensator Editor page.

A quick way to create a lowpass roll-off filter is to add a pair of complex

poles. To do this, first click the Compensator Editor tab and change
the compensator to F. Right-click in the Dynamics table and select Add
Pole/Zero > Complex Pole. Select this line to show the editable parameters
in the Edit Selected Dynamics group box. For this example, try to place
the poles at about 50 rad/s. The following figure shows the poles added to
the prefilter Bode diagram.

4-78
 Bode Diagram Design

Adding a Complex Pair of Poles to the Prefilter Bode Diagram

By default, the damping ratio of the complex pair is 1.0, which means that
there are two real-valued poles at about -50 rad/s. The green curve, which
represents the prefilter Bode response, shows the -3 dB point for the roll-off
is at about 50 rad/s. The magenta curve, which represents the closed-loop
response from the prefilter to the plant output, shows that after the -3 dB
point, the closed-loop gain rolls off at -40 dB/decade to provide some noise
disturbance rejection.

Importing a Prefilter
As an alternative approach, you can design a prefilter using the Control
System Toolbox commands like ss or tf and then import the design directly
into the prefilter. For example, to create the lowpass filter using zpk, try

prefilt=zpk([],[-35 + 35i, -35 - 35i],1)

4-79
 4 Designing Compensators

and import prefilt by clicking System Data on the Architecture page.

This opens the System Data dialog box. Click Browse to open the Model
Import dialog box, as shown next.

Importing a Prefilter

Select prefilt from the Available Models list and click Import to import
the prefilter model. Click Close to close the Import Model dialog box. After
you have imported the prefilter model, you can modify it using the same
methods as described in this chapter for compensator design.

4-80
 Root Locus Design

Root Locus Design

In this section...
“What Is Root Locus Design?” on page 4-81
“Example: Electrohydraulic Servomechanism” on page 4-82
“Changing the Compensator Gain” on page 4-88
“Adding Poles and Zeros to the Compensator” on page 4-91
“Editing Compensator Pole and Zero Locations” on page 4-96
“Viewing Damping Ratios” on page 4-100

What Is Root Locus Design?

A common technique for meeting design criteria is root locus design. This
approach involves iterating on a design by manipulating the compensator
gain, poles, and zeros in the root locus diagram.

As system parameter k varies over a continuous range of values, the root locus
diagram shows the trajectories of the closed-loop poles of the feedback system.
Typically, the root locus method is used to tune the loop gain of a SISO control
system by specifying a designed set of closed-loop pole locations.

Consider, for example, the tracking loop

where P(s) is the plant, H(s) is the sensor dynamics, and k is a scalar gain to
be adjusted. The closed-loop poles are the roots of

q(s) = 1 + k P (s) H (s)

4-81
 4 Designing Compensators

The root locus technique consists of plotting the closed-loop pole trajectories
in the complex plane as k varies. You can use this plot to identify the gain
value associated with a desired set of closed-loop poles.

The DC motor design example focused on the Bode diagram feature of the
SISO Design Tool. Each of the design options available on the Bode diagram
side of the tool have a counterpart on the root locus side. To demonstrate
these techniques, this example presents an electrohydraulic servomechanism.

The SISO Design Tool’s root locus and Bode diagram design tools provide
complementary perspectives on the same design issues; each perspective
offers insight into the design process. Because the SISO Design Tool shows
both root locus and Bode diagrams, you can also choose to combine elements
of both perspectives in making your design decisions.

Example: Electrohydraulic Servomechanism

A simple version of an electrohydraulic servomechanism model consists of

• A push-pull amplifier (a pair of electromagnets)

• A sliding spool in a vessel of high-pressure hydraulic fluid
• Valve openings in the vessel to allow for fluid to flow
• A central chamber with a piston-driven ram to deliver force to a load
• A symmetrical fluid return vessel

This figure shows a schematic of this servomechanism.

4-82
 Root Locus Design

Electrohydraulic Servomechanism

The force on the spool is proportional to the current in the electromagnet coil.
As the spool moves, the valve opens, allowing the high-pressure hydraulic
fluid to flow through the chamber. The moving fluid forces the piston to move
in the opposite direction of the spool. Control System Dynamics, by R. N.
Clark, (Cambridge University Press, 1996) derives linearized models for the
electromagnetic amplifier, the valve spool dynamics, and the ram dynamics; it
also provides a detailed description of this type of servomechanism.

If you want to use this servomechanism for position control, you can use
the input voltage to the electromagnet to control the ram position. When
measurements of the ram position are available, you can use feedback for the
ram position control, as shown in the following figure.

Feedback Control Structure for an Electrohydraulic Servomechanism

4-83
 4 Designing Compensators

Your task is to design the compensator, C(s).

Plant Transfer Function

If you have not already done so, type

load ltiexamples

to load a collection of linear models that include Gservo, which is a linearized

plant transfer function for the electrohydraulic position control mechanism.
Typing Gservo at the MATLAB prompt opens the servomechanism (plant)
transfer function.

Gservo

Zero/pole/gain from input "Voltage" to output "Ram position":

40000000
-----------------------------
s (s+250) (s^2 + 40s + 9e004)

Design Specifications
For this example, you want to design a controller so that the step response of
the closed-loop system meets the following specifications:

• The 2% settling time is less than 0.05 second.

• The maximum overshoot is less than 5%.

The remainder of this topic discusses how to use the SISO Design Tool to
design a controller to meet these specifications.

Opening the SISO Design Tool

Open the SISO Design Tool and import the model by typing

sisotool(Gservo)

at the MATLAB prompt. This opens the SISO Design Task node in the
Control and Estimation Tools Manager and the Graphical Tuning window
with the servomechanism plant imported.

4-84
 Root Locus Design

Graphical Tuning Window Showing the Root Locus and Bode Plots for the
Electrohydraulic Servomechanism Plant

Zooming
Click the Zoom In

icon in the Graphical Tuning window. Press and hold the mouse’s left button
and drag the mouse to select a region for zooming. For this example, reduce
the root locus region to about -500 to 500 in both the x- and y-axes. This figure
illustrates the zooming in process.

4-85
 4 Designing Compensators

Zooming In on a Region in the Root Locus Plot

As in the DC motor example, click the Analysis Plots tab to set up loop
responses. Select Plot Type Step for Plot 1, then select plot 1 for Closed-Loop
r to y, shown as follows.

4-86
 Root Locus Design

Analysis Plots Loop Response Selection

For more information about the Analysis Plots page, see Analysis Plots in
"Using the SISO Design Tool and LTI Viewer."

Selecting the plot for Closed-Loop r to y opens the associated LTI Viewer.

Your LTI Viewer should look like the following figure.

4-87
 4 Designing Compensators

LTI Viewer for the Electrohydraulic Servomechanism

The step response plot shows that the rise time is on the order of 2 seconds,
which is much too slow given the system requirements. The following topics
describe how to use frequency design techniques in the SISO Design Tool
to design a compensator that meets the requirements specified in “Design
Specifications” on page 4-84.

Changing the Compensator Gain

The simplest thing to do is change the compensator gain, which by default
is unity. You can change the gain by entering the value directly in the
Compensator Editor page.

The following figure shows this procedure.

4-88
 Root Locus Design

Changing the Compensator Gain in the Root Locus Plot with the Compensator
Editor Page

Enter the compensator gain in the text box to the right of the equal sign as
shown in the previous figure. The Graphical Tuning window automatically
replots the graphs with the new gain.

Experiment with different gains and view the closed-loop response in the
associated LTI Viewer.

Alternatively, you can change the gain by grabbing the red squares on the root
locus plot in the Graphical Tuning window and moving them along the curve.

4-89
 4 Designing Compensators

Closed-Loop Response
Change the gain to 20 by editing the text box next to the equal sign on the
Compensator Editor page. Notice that the locations of the closed-loop poles
on the root locus are recalculated for the new gain.

This figure shows the associated closed-loop step response for the gain of 20.

Step Response with the Settling Time for C(s) = 20

In the LTI Viewer’s right-click menu, select Characteristics > Settling

Time to show the settling for this response. This closed-loop response does
not meet the desired settling time requirement (0.05 seconds or less) and
exhibits unwanted ringing. “Adding Poles and Zeros to the Compensator” on
page 4-91 shows how to design a compensator so that you meet the required
specifications.

4-90
 Root Locus Design

Adding Poles and Zeros to the Compensator

You may have noticed that increasing the gain makes the system
under-damped. Further increases force the system into instability, so meeting
the design requirements with only a gain in the compensator is not possible.

There are three sets of parameters that specify the compensator: poles, zeros,
and gain. After you have selected the gain, you can add poles or zeros to
the compensator.

Adding Poles to the Compensator

You can add complex poles on the Compensator Editor page. Click the
Compensator Editor tab, make sure C is selected, and then right click
in the Dynamics table. Select Add Pole/Zero > Complex Pole. Use the
Edit Selected Dynamics group box to modify pole parameters, as shown in
the following figure. For more about entering pole parameters directly, see
“Editing Compensator Pole and Zero Locations” on page 4-96.

4-91
 4 Designing Compensators

Adding a Complex Pair of Poles to the Compensator on the Compensator

Editor Page

You can also add a complex pole pair directly on the root locus plot using the
Graphical Tuning window. Right-click in the root locus plot and select Add
Pole/Zero > Complex Pole. Click in the root locus plot region where you
want to add one of the complex poles.

Complex poles added in this manner are automatically added to the

Dynamics table in the Compensator Editor page.

After you have added the complex pair of poles, the LTI Viewer response plots
change and both the root locus and Bode plots show the new poles.

This figure shows the Graphical Tuning window with the new poles added.
For clarity, you may want to zoom out further, as was done here.

4-92
 Root Locus Design

Result of Adding a Complex Pair of Poles to the Compensator

Adding Zeros to the Compensator

You can add zeros in the Dynamics table on the Compensator Editor page
or directly on the Root Locus plot in the Graphical Tuning window.

To add the zeros using the Compensator Editor page, click the
Compensator Editor tab, make sure C is selected, and then right click
in the Dynamics table. Select Add Pole/Zero > Complex Zero. Use the
Edit Selected Dynamics group box to modify zero parameters, as shown.
For more about entering zero parameters directly, see “Editing Compensator
Pole and Zero Locations” on page 4-96.

4-93
 4 Designing Compensators

Adding Complex Zeros to the Compensator on the Compensator Editor Page

You can also add complex zeros directly on the root locus plot using the
Graphical Tuning window by right-clicking in the root locus plot, selecting
Add Pole/Zero > Complex Zero, and then clicking in the root locus plot
region where you want to add one of the zeros.

Complex zeros added in this manner are automatically added to the

Dynamics table on the Compensator Editor page.

After you have added the complex zeros, the LTI Viewer response plots change
and both the root locus and Bode plots show the new zeros.

4-94
Root Locus Design

4-95
 4 Designing Compensators

Electrohydraulic Servomechanism Example with Complex Zeros Added

If your step response is unstable, lower the gain by grabbing a red box in the
right-side plane and moving it into the left-side plane. In this example, the
resulting step response is stable, but it still doesn’t meet the design criteria
since the 2% settling time is greater than 0.05 second.

As you can see, the compensator design process can involve some trial and
error. You can try dragging the compensator poles, compensator zeros, or the
closed-loop poles around the root locus until you meet the design criteria.

“Editing Compensator Pole and Zero Locations” on page 4-96, shows you
how to modify the placement of poles and zeros by specifying their numerical
values on the Compensator Editor page. It also presents a solution that
meets the design specifications for the servomechanism example.

Editing Compensator Pole and Zero Locations

A quick way to change poles and zeros is simply to grab them with your
mouse and move them around the root locus plot region. If you want to specify

4-96
 Root Locus Design

precise numerical values, however, you should use the Compensator Editor
page in the SISO Design Task node on the Control and Estimation Tools
Manager to change the gain value and the pole and zero locations of your
compensator, as shown.

Using the Compensator Editor Page to Add, Delete, and Move Compensator
Poles and Zeros

You can use the Compensator Editor page to

• Add compensator poles and zeros.

• Delete compensator poles and zeros.
• Edit the compensator gain.
• Edit the locations of compensator poles and zeros.

4-97
 4 Designing Compensators

Adding Compensator Poles and Zeros

To add compensator poles or zeros:

1 Select the compensator (in this example, C) in the list box to the left of
the equal sign.

2 Right-click in the Dynamics table for the pop-up menu.

3 From the pop-up menu, select Add Pole/Zero > Complex Pole or Add
Pole/Zero > Complex Zero.

4 Use the Edit Selected Dynamics group box to modify pole or zero
parameters.

Deleting Compensator Poles and Zeros

To delete compensator poles or zeros:

1 Select the compensator (in this example, C) in the list box to the left of
the equal sign.

2 Select the pole or zero in the Dynamics table that you want to delete.

3 Right-click and select Delete Pole/Zero from the pop-up menu.

Editing Gain, Poles, and Zeros

To edit compensator gain:

1 Select the compensator to edit in the list box to the left of the equal sign
in the Compensator area.

2 Enter the gain value in the text box to the right of the equal sign in the
Compensator area.

To edit pole and zero locations:

1 Select the pole or zero you want to edit in the Dynamics table.

2 Change current values in the Edit Selected Dynamics group box.

4-98
 Root Locus Design

For this example, edit the poles to be at −110 ± 140i and the zeros at
−70 ± 270i. Set the compensator gain to 23.3.

Your Graphical Tuning window now looks like this.

Graphical Tuning Window with the Final Values for the Electrohydraulic
Servomechanism Design Example

To see that this design meets the design requirements, look at the step
response of the closed-loop system.

4-99
 4 Designing Compensators

Closed-Loop Step Response for the Final Design of the Electrohydraulic

Servomechanism Example

The step response looks good. The settling time is less than 0.05 second, and
the overshoot is less than 5%. You have met the design specifications.

Viewing Damping Ratios

The Graphical Tuning window provides design requirements that can make it
easier to meet design specifications. If you want to place, for example, a pair
of complex poles on your diagram at a particular damping ratio, select Design
Requirements > New from the right-click menu in the root locus graph.

This opens the New Design Requirement dialog box.

4-100
 Root Locus Design

Applying damping ratio requirements to the root locus plot results in a pair of
shaded rays at the desired slope, as this figure shows.

Root Locus with 0.707 Damping Ratio Lines

4-101
 4 Designing Compensators

Try moving the complex pair of poles you added to the design so that they are
on the 0.707 damping ratio line. You can experiment with different damping
ratios to see the effect on the design.

If you want to change the damping ratio, select Design Requirements >
Edit from the right-click menu. This opens the Edit Design Requirements
dialog box.

Specify the new damping ratio requirement in this dialog box.

An alternate way to adjust a requirement is to left-click the requirement itself

to select it. Two black squares appear on the requirement when it is selected.
You can then drag it with your mouse anywhere in the plot region.

If you want to add a different set of requirements, for example, a settling time
requirement, again select Design Requirements > New from the right-click
menu to open the New Requirements dialog box and choose Settling time
from the pull-down menu. You can have multiple types of design requirements
in one plot, or more than one instance of any type.

The types of requirements available depend on which view you use for
your design. See Design Requirements for a description of all the design
requirement options available in the SISO Design Tool.

4-102
 Nichols Plot Design

Nichols Plot Design

In this section...
“What Is Nichols Plot Design?” on page 4-103
“DC Motor Example” on page 4-103
“Opening a Nichols Plot” on page 4-104
“Adjusting the Compensator Gain” on page 4-105
“Adding an Integrator” on page 4-108
“Adding a Lead Network” on page 4-110

What Is Nichols Plot Design?

An alternative method for designing compensators is to use the Nichols
plot, which combines gain and phase information in a single plot. The
combination of the two is useful when you are designing to gain and phase
margin specifications.

You can design compensators with the SISO Design Tool by using Nichols plot
techniques. This topic repeats the DC motor compensator design presented in
“Example: DC Motor” on page 4-55, only this time the focus is on Nichols plot
techniques. The design strategy, however, is the same.

1 Adjust the compensator gain to improve the rise time.

2 Add an integrator to eliminate steady-state error.

3 Add a lead network to further improve the rise time while minimizing
overshoot.

DC Motor Example
From “SISO Example: The DC Motor” on page 2-3, the transfer function of
the DC motor is

Transfer function:
1.5
------------------
s^2 + 14 s + 40.02

4-103
 4 Designing Compensators

This example uses the design criteria specified in “Design Specifications”

on page 4-84:

• Rise time of less than 0.5 second

• Steady-state error of less than 5%
• Overshoot of less than 10%
• Gain margin greater than 20 dB
• Phase margin greater than 40 degrees

Opening a Nichols Plot

To open the SISO Design Tool with a Bode diagram and a Nichols plot, use
these commands:

load ltiexamples
sisotool({'bode','nichols'},sys_dc)

The SISO Design Task node on the Control and Estimation Tools Manager
opens and the Graphical Tuning window with sys_dc opens.

4-104
 Nichols Plot Design

Graphical Tuning Window with a Bode Diagram and a Nichols Plot

Adjusting the Compensator Gain

You can adjust the compensator gain by entering a value in the Compensator
Editor page.

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 Select C from the compensator selection list.

3 In the text box to the right of the equal sign in the Compensator area,
enter the gain amount and press Enter.

4-105
 4 Designing Compensators

Adjusting Compensator Gain in the Compensator Editor Page

In this example, the new gain is 112.

You can also adjust the compensator gain in the Graphical Tuning window by
moving the Nichols curve up and down with your mouse. To do this, place
your mouse over the curve. The cursor turns into a hand. Left-click and move
the curve up to increase the gain. When you adjust the gain in this manner,
the compensator gain is automatically updated in the Compensator Editor
page.

Click the Analysis Plots tab to set the analysis plots. Select Plot Type
Step for Plot 1, and then select plot 1 for Closed-Loop r to y, as shown in
the following figure, to open a linked LTI Viewer with the closed-loop step
response from reference signal r to output signal y.

4-106
 Nichols Plot Design

Analysis Plots Loop Response Selection

4-107
 4 Designing Compensators

LTI Viewer Step Response for Compensator Gain = 112

The rise time is quite fast, about 0.15 second, but the overshoot is 18.4%
and the steady-state is about 0.82.

Adding an Integrator
One approach to eliminating the steady-state error is to add an integrator.

To add an integrator:

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 Right-click in the Dynamics table and select Add Pole/Zero >

Integrator.

This figure shows the process.

4-108
 Nichols Plot Design

Adding an Integrator in the Dynamics Table

You can also add an integrator by selecting Add Pole/Zero > Integrator
from the right-click menu in the Graphical Tuning window. When you add the
integrator in this manner, it is automatically added to the Dynamics table
on the Compensator Editor page.

Adding an integrator changes the gain margin from infinity to 10.5 dB. Since
the gain and phase margins are now both finite, the Nichols plot shows a
vertical line for the gain margin and a horizontal line for the phase margin.

The linked LTI Viewer automatically updates to show the new response.

4-109
 4 Designing Compensators

Step Response for a Compensator Consisting of a Gain and an Integrator

The steady-state value is now 1, which means the steady-state error has been
eliminated, but the overshoot is 34% and the rise time is about 0.7 second.
You must do more work to create a good design.

Adding a Lead Network

Improving the rise time requires that you increase the compensator gain, but
increasing the gain further deteriorates the gain and phase margins while
increasing the overshoot. You need to add a lead network to selectively raise
the gain about the frequency crossover. To add the lead network:

1 Click the Compensator Editor tab to open the Compensator Editor

page.

2 Right-click in the Dynamics table and select Add Pole/Zero > Lead.

This figure shows the process of adding a lead network on the Compensator
Editor page.

4-110
 Nichols Plot Design

You can also add a lead network in the Graphical Tuning window. To add
a lead network, select Add Pole/Zero > Lead from the right-click menu.
Your cursor turns into a red ‘x’. Left-click along the Nichols curve to add
the lead network. To move the lead network along the curve, left-click the
pole or zero and drag.

You can track the pole’s movement in the status bar at the bottom of the
Graphical Tuning window. The status bar tells you the current location of
the pole.

Using the Compensator Editor page, move the lead network pole to -28
and the zero to -4.3 for this design. The zero should be almost on top of the
right-most pole of the plant (the DC motor model). Adjust the compensator
gain to 84. This gives the final design.

4-111
 4 Designing Compensators

Final Nichols Plot Design for the DC Motor Compensator

The gain and phase margins are 21.9 dB and 65.7 degrees, respectively.
Inspect the closed-loop step response to see if the rise time and overshoot
requirements are met.

4-112
 Nichols Plot Design

Closed-Loop Step Response for the Final Compensator Design

As this figure shows, the rise time is 0.448 second, and the overshoot is a little
over 3%. This satisfies the rest of the design requirements.

4-113
 4 Designing Compensators

Automated Tuning Design

In this section...
“Supported Automated Tuning Methods” on page 4-114
“Loading and Displaying the DC Motor Example for Automated Tuning” on
page 4-114
“Applying Automated PID Tuning” on page 4-116

Supported Automated Tuning Methods

The SISO Design Tool simplifies the task of designing and tuning
compensators. There are five automated tuning methods in the SISO Design
Tool to help you design an initial stabilizing compensator for a SISO loop
on-the-fly or refine existing compensator design so that it satisfies a certain
user-defined design specification.

The available tuning methods are:

• Optimization-based tuning
• PID tuning
• Internal Model Control (IMC) tuning
• LQG synthesis
• Loop shaping

For a detailed discussion of these, see “Automated Tuning” in the Control

System ToolboxUser’s Guide.

Loading and Displaying the DC Motor Example for

Automated Tuning
Follow these steps to load and display the DC Motor example for automated
tuning:

1 If you have not yet built the DC Motor example, type

load ltiexamples

4-114
 Automated Tuning Design

2 To open the SISO Design Tool and import the DC motor, type

sisotool(sys_dc)

at the MATLAB prompt. This opens both the SISO Design Task node
on the Control and Estimation Tools Manager and the Graphical Tuning
window with sys_dc loaded.

3 Click the Analysis Plots tab to set the analysis plots. Select the plot type
as Step for Plot 1. Then, check the box for plot 1 to the left of Closed-Loop
r to y, as shown in the following figure, to open a linked LTI Viewer with
the closed-loop step response from reference signal r to output signal y.

4 In the LTI Viewer that appears, use the right-click menu to add rise time
and steady state values to your plot.

4-115
 4 Designing Compensators

Step Response When Compensator = 1

Note that by default, the compensator is 1 and unit negative feedback is used
(see Architecture in the Control System Toolbox User’s Guide). When a unit
step is applied to the setpoint change, the steady state value of the system
output is 0.0361, which is far from the setpoint, and its rise time is 0.589.

Applying Automated PID Tuning

1 Click the Automated Tuning tab.

2 Select PID tuning from the Design method list.

3 Leave C as the default compensator.

4-116
 Automated Tuning Design

You can choose from the following controller types:

• P
• PI
• PID

1
• PID with derivative filter
(1 + s / N )

Designing a Proportional-Only Controller

1 For the controller type, click the P option button for proportional-only
control (C = Kp).

4-117
 4 Designing Compensators

2 Select the Ziegler-Nichols open loop tuning algorithm from the

Tuning algorithm list.

3 Click the Set-Point Tracking option button.

4 Click Update Compensator. The LTI Viewer is updated with the

application of PID automated tuning with P and the compensator value is
now 205.49.

Note that the rise time is arrived at in just 0.0774 seconds, compared with
0.589 when C=1. However, the steady state value of 0.885 can still be
improved by setting the automated tuning controller type to PI, as described
in “Designing a Proportional-Integral Controller” on page 4-119.

4-118
 Automated Tuning Design

Designing a Proportional-Integral Controller

1 For the controller type, click the PI option button for proportional-integral

KI
control ( C = K p + ).
s
2 Select the Ziegler-Nichols open loop tuning algorithm from the
Tuning algorithm list.

3 Click the Set-Point Tracking option button.

4 Click Update Compensator. The LTI Viewer is updated with the

application of PID automated tuning with PI and the compensator value

(1 + 0.36 s)
is now 487.64 × .
s

Although the rise time has increased slightly (0.0876), the steady state value
is 1. Applying automated tuning using PID tuning set to PI will guarantee
zero offset.

4-119
 4 Designing Compensators

Multi-Loop Compensator Design

In this section...
“When to Use Multi-Loop Compensator Design” on page 4-120
“Workflow for Multi-Loop Compensator Design” on page 4-120
“Example: Position Control of a DC Motor” on page 4-120

When to Use Multi-Loop Compensator Design

In many applications, a single-loop design is not feasible. If you have a design
with inner loops, you can use the SISO Design Tool to design a compensator
that meets your specifications.

Workflow for Multi-Loop Compensator Design

A typical procedure is to design the innermost loop on its own. You can use the
SISO Design Tool to isolate the design on individual loops. When used this
way, the tool ignores outer loop dynamics. Once the inner loop is designed,
you can move on to the design of the outer loop compensator to achieve the
desired closed-loop behavior. “Example: Position Control of a DC Motor” on
page 4-120 shows an example of this procedure.

Example: Position Control of a DC Motor

Instead of controlling the angular rate of a DC motor, this example develops a
control law for controlling the position (angle) of the motor shaft. The block
diagram of the DC motor, as shown in the following figure, has an integrator
added as an outer loop.

4-120
 Multi-Loop Compensator Design

Block Diagram of the Position-Controlled DC Motor

The design goal for this example is the minimize the closed-loop step response
settling time while maintaining an inner loop phase margin of at least 65º
with maximum bandwidth.

For details on how to derive state-space and transfer function representations

of a DC motor, see “SISO Example: The DC Motor” on page 2-3.

Designing a multi-loop compensator for a DC motor involves the following

steps:

• “Developing a Mathematical Model of the DC Motor” on page 4-121

• “Selecting the Architecture and Importing the Model” on page 4-123
• “Designing the Inner Loop” on page 4-126
• “Tuning the Outer Loop” on page 4-128
• “Validating the Design with the LTI Viewer for SISO Design” on page 4-130

Developing a Mathematical Model of the DC Motor

These are the relevant physical constants:

R=2.0 % Ohms
L = 0.5 % Henrys
Km=0.1; Kb = 0.1 % Torque and back emf constants

4-121
 4 Designing Compensators

Kf= 0.2; % Nms

J = 0.02 % kg.m^2/s^2

First, construct a state-space model of the DC motor with one input, the
applied voltage (Va). The output is the angular rate w.

h1 = tf(Km,[L,R]); % Armature
h2 = tf(1,[J, Kf]) % Equation of motion
dcm = ss(h2) *h1; % w = h2 cascaded with h1
dcm = feedback(dcm, Kb, 1, 1);% Closes back emf loop

Adapting the Model to SISO Tool Architecture

One possible choice for your architecture is this multi-loop configuration.

Comparing this architecture to the original Block Diagram of the

Position-Controlled DC Motor on page 4-121, it is evident that the two do not
match. Using block diagram algebra, you can manipulate the original figure
into one that fits this architecture.

Position-Controlled DC Motor Rearchitected

4-122
 Multi-Loop Compensator Design

To create this representation, add an integrator to get Θ, the angular

displacement, and a pure differentiator in the inner loop’s return path. The
channel from Va to w is dcm(1), making it the appropriate channel for adding
the integrator.

G = dcm*tf(1,[1,0]) % Motor with integrator; output is theta.

C2 = tf('s') % Differentiator

Selecting the Architecture and Importing the Model

Open the SISO Design Tool by typing

sisotool

at the MATLAB prompt. Once the Controls & Estimation Tools Manager
opens, click Control Architecture on the Architecture page. Select the
multi-loop configuration with two compensators, C1 in the forward path and
C2 in the inner feedback loop — located in the lower-right corner.

Control Architecture Window

Next, import the model parameters by clicking System Data on the

Architecture tab. This opens the System Data dialog box. Set G to G from

4-123
 4 Designing Compensators

the workspace. Assume a perfect sensor and set H to 1. C1 and C2 are the
gains you will use to design a compensator. Set C1 to 1 and C2 to C2 from the
workspace. Your System Data dialog box should look like this.

Selecting SISO Design Graphical Editor Views

Once you have selected the multi-loop architecture, click the Graphical
Tuning tab. Set the plot types as follows:

1 Open-Loop 1 to "Root-Locus"

2 Open-Loop 2 to "Open-Loop Bode"

Your Graphical Tuning page should look like this.

4-124
 Multi-Loop Compensator Design

Graphical Tuning Page Set for DC Motor Multi-Loop Design

Click Show Design Plot to see the SISO Design Graphical editor.

4-125
 4 Designing Compensators

Designing the Inner Loop

You are now in a position to do the design. Start with the design of the inner
loop. To do this, go to the Architecture page and remove the effects of the
outer loop by following these steps:

1 Click Loop Configuration. This opens the Open-Loop Configuration

dialog box.

2 From the pull-down menu, select Open-Loop Output of C2.

3 Click Highlight feedback loop. This opens a figure of the control

architecture showing the loop configuration.

4-126
 Multi-Loop Compensator Design

Notice how the C1 piece of the compensator and the outer loop are grayed out.
This means that they will have no effect on the inner loop at this time.

Next, turn to the SISO Design Graphical editor. Use the Bode plot for open
loop 2 (the inner loop) and increase the gain to maximize bandwidth subject to
a 65º phase margin. This turns out to be a gain of about 16.1 for C2.

4-127
 4 Designing Compensators

Setting the Inner Loop Gain

This finishes the design of the inner loop.

Tuning the Outer Loop

The goal in designing the outer loop is to minimize the settling time. Note
that the outer loop can "see" the inner loop, so that the tuning affects the
entire system. Follow these steps:

1 Go to the Analysis Plot tab in the Controls & Estimation Tools Manager.
Select the Closed-Loop r to y check box.

2 Select Step from the Plot 1 pull-down menu. This opens the LTI Viewer
for SISO Design.

4-128
 Multi-Loop Compensator Design

3 Right-click in the step response plot and select Characteristics>Settling

Time. Your LTI Viewer should look like this.

Initial Step Response with Settling Time

The settling time is about 79 s.

Return to the SISO Design Graphical editor and increase the gain of C1 in the
root locus plot. At a gain of about 90.2, you will see the complex pair of poles
move toward a slower time constant as the third pole moves toward a faster
one. You can view the trade-off in the LTI Viewer for SISO Design. As the
gain is changed, the closed-loop step response changes.

The 90.2 gain seems to yield a good compromise between rise and settling
time.

4-129
 4 Designing Compensators

Final Gain Choice for C1

Validating the Design with the LTI Viewer for SISO Design
Turning back to the LTI Viewer for SISO Design, it is evident that the settling
time is now much lower than the original 78.9 s.

4-130
 Multi-Loop Compensator Design

With a settling time of about 0.8 s, and a phase margin of 65º in the inner
loop, the design is complete.

4-131
 4 Designing Compensators

Control Design Analysis of Multiple Models

In this section...
“Multiple Models Represent System Variations” on page 4-132
“Control Design Analysis Using the SISO Design Tool” on page 4-132
“Specifying a Nominal Model” on page 4-133
“Frequency Grid for Multimodel Computations” on page 4-136
“How to Analyze the Controller Design for Multiple Models” on page 4-137

Multiple Models Represent System Variations

Typically, the dynamics of a system are not exactly known and may vary. For
example, system dynamics can vary because of:

• Parameter value variations caused by manufacturing tolerances. For

example, the resistance value of a resistor is typically within a range about
the nominal value, 5 Ω +/– 5%.
• Operating conditions. For example, aircraft dynamics change based on
its altitude and speed.

The controller you design for such a system must satisfy design requirements
for dynamics of all models.

Control Design Analysis Using the SISO Design Tool

To design a controller for a system whose dynamics vary, you sample the
variations, create an LTI model for each sample and use the set of models to
build an array of LTI models. Then, design the controller for a representative
model from the array.

Control design analysis of multiple models in the SISO Design Tool requires
you to specify either the plant G or sensor H or both as row or column arrays
of LTI models. If both G and H are arrays, their sizes must match.

Use the SISO Design Tool to:

4-132
 Control Design Analysis of Multiple Models

1 Choose a nominal model from the array of LTI models.

2 (Optional) Specify a frequency grid for multimodel computations.

3 Design a controller for the nominal model.

4 Analyze if the controller meets the design requirements for all models in
the array.

If the controller design does not meet design requirements on all the models,
specify a different nominal model and redesign the controller. For more
information, see “How to Analyze the Controller Design for Multiple Models”
on page 4-137.

Specifying a Nominal Model

• “What Is a Nominal Model?” on page 4-133
• “How Is a Nominal Model Computed?” on page 4-133
• “Criteria for Choosing a Nominal Model” on page 4-135

What Is a Nominal Model?

The nominal model is a representative model in the array of LTI models that
you use to design the controller or for loop shaping in the SISO Design Tool.

You use the design and analysis plots to visualize and analyze the effect of the
controller on the remaining plants in the array.

How Is a Nominal Model Computed?

The following table summarizes how the software computes the nominal
model when the plant G and sensor H are arrays of LTI models in a control
architecture.

4-133
 4 Designing Compensators

Array of LTI Models Nominal Model

Both G and H • By default, computed using the first element
in both arrays.
• For a different index, computed using the
Note The sizes of both
specified index in both arrays.
arrays must match.
- The same index is also used when you
import new arrays in the SISO Design
Tool.
- If the specified index does not exist, the
nominal model index reverts to a value you
previously specified or the first element.
Only G or H Uses scalar expansion for the specified index
and G or H value.

For example, in the following default control architecture:

if

• G and H are arrays of LTI models of length 3

• Nominal model index is 2

the software uses the second element in both the arrays to compute the
nominal model:

4-134
 Control Design Analysis of Multiple Models

Nominal Model
r y
2

The nominal response from r to y is:

CGnom
T= ,
1 + CGnom H nom

where Gnom = G2, Hnom = H2 and GnomHnom is the open-loop response.

The software also computes and plots the responses showing the effect of C on
the remaining pairs of plant and sensor combinations—G1H1 and G3H3.

If only G is an array of LTI models, and the specified nominal model is 2, then
the control architecture for nominal response is:

Nominal Model
r y
2

The software also computes and plots the responses showing the effect of C for
the remaining pairs of plant indices and sensor—G1H and G3H.

Criteria for Choosing a Nominal Model

The nominal model is the model that you design a controller for. Typically,
you choose a model that:

4-135
 4 Designing Compensators

• Represents an average of the multiple models. For example, the open-loop

response of the model lies midway among the responses of all models in
the array.
• Represents a worst-case plant.
• Lies closet to the stability point.
• You want to work with for control design.

Tip You can plot and analyze the open-loop dynamics of the system on a Bode
plot to determine which model to choose as nominal.

If the controller design for the nominal model does not meet the design
requirements on the remaining plants in the array, you can specify a
different nominal model and redesign the controller. See “How to Analyze the
Controller Design for Multiple Models” on page 4-137 for more information.

Frequency Grid for Multimodel Computations

• “Algorithm for Frequency Grid Computation” on page 4-136
• “When to Specify Custom Frequency Grid” on page 4-136

Algorithm for Frequency Grid Computation

The frequency response of a system is computed at points in the frequency,
called frequency grid.

To compute the frequency grid, the software computes a logarithmic equally

spaced grid, based on the dynamics (dynamic range) of each model in the
array.

When to Specify Custom Frequency Grid

You can specify a custom frequency grid if:

• The automatic grid does not capture the system dynamics sufficiently.

4-136
 Control Design Analysis of Multiple Models

This happens because the grid is not sufficiently dense in a particular

frequency range. For example, for an underdamped systems the model
response shows sharp and tall peaks. Examine the analysis plots to verify
if these dynamics are captured in the response. If the response does not
capture these dynamics, specify a denser gridding.
• The grid computed automatically has more points in the response than
you require.
• You are interested only in a specific frequency range in the response.

Tip Specifying a less dense grid reduces the number of computations and is
less expensive computationally.

For more information, see “How to Analyze the Controller Design for Multiple
Models” on page 4-137.

How to Analyze the Controller Design for Multiple

Models
Prerequisites
An array of LTI models that model variations in the plant G, sensor H or both.
G and H must be row or column vectors of the same length.

Obtain the array using one of the following methods:

• Create multiple LTI models using the tf, ss, zpk, or frd commands. For
example:

% Specify system parameters.

m = 3;
b = 0.5;
k = 8:1:10;
T = 0.1:.05:.2;

% Create an array of LTI models to model plant (G) variations.

for ct = 1:length(k);
G(:,:,ct) = tf(1,[m,b,k(ct)]);

4-137
 4 Designing Compensators

end

% Create an array of LTI models to model sensor (H) variations.

for ct = 1:length(T);
H(:,:,ct) = tf(1,[1/T(ct), 1]);
end

In this case, G and H each contain 3 elements.

• Create an array of LTI models using the stack command. For example:

% m = 3, b = 0.5, k = 8:1:10.
% Create an array of LTI models to model plant (G) variations.
G1 = tf(1, [1 1 8]);
G2 = tf(1, [1 1 9]);
G3 = tf(1, [1 1 10]);
% Create an array of LTI models
G = stack(1, G1, G2, G3);

% Create an array of LTI models to model sensor (H) variations.

H1 = tf(1,[1/0.1, 1]);
H2 = tf(1,[1/0.15, 1]);
H3 = tf(1,[1/0.2, 1]);
% Create array of LTI models.
H=stack(1,H1,H2,H3);

• (Requires Simulink Control Design software) Perform batch linearizations

at multiple operating points. Then export the computed LTI models to
create an array of LTI models. See the Reference Tracking of a DC Motor
with Parameter Variations demo.
• (Requires Simulink Control Design and Robust Control Toolbox software)
Specify a core Simulink block to linearize to an uncertain state-space (uss
or ufrd) model. See Working with Models Containing Core Simulink or
Custom Blocks.
• (Requires Robust Control Toolbox software) Compute uncertain state-space
(uss) models from Simulink models. hen use usubs or usample to create
an array of LTI models. See Working with Models Containing Uncertain
State Space Blocks.

1 Open SISO Design Tool.

4-138
 Control Design Analysis of Multiple Models

sisotool(G,1,H)

This command opens the following window.

By default, the combination of the first plant and sensor in the arrays is
the nominal model on which you perform the control design. For more
information, see “How Is a Nominal Model Computed?” on page 4-133.

The Graphical Tuning window, which opens with the SISO Design Tool,
shows the individual responses of all models in the arrays.

4-139
 4 Designing Compensators

Tip You can view the envelope of the Bode response instead by
right-clicking the plot and selecting Multimodel Display > Bounds. See
“Using the Graphical Tuning Window” on page 4-42 for more information.

Alternatively, to view the responses of all models in the arrays:

4-140
 Control Design Analysis of Multiple Models

• Open an empty SISO Design Tool by typing sisotool and import G

and H. See “Importing Models into the SISO Design Tool” on page 4-30
for more information.
• Configure SISO Design Tool using sisoinit with the arrays of LTI
models and then open the SISO Design Tool.

2 Configure the analysis plots in the Analysis Plots tab in Control and
Estimation Tools Manager.

By default, the plots show only the nominal response.

Right-click the plot, and select Multimodel Configuration > Bounds

or Multimodel Configuration > Individual Responses to see the
individual response or envelope of all models, respectively.

4-141
 4 Designing Compensators

See “Analysis Plots for Loop Responses” on page 4-36 for more information.

3 (Optional) If you want to specify a different nominal plant, click

Multimodel Configuration in the Architecture tab.

The Multimodel Configuration Dialog window opens.

4-142
 Control Design Analysis of Multiple Models

Specify a different nominal model in Nominal Model Index field. If you

specify an index value greater than the maximum index of the arrays, the
field reverts back to a value you specified previously or 1.

Tip You can keep the Multimodel Configuration Dialog window open as
you are following the next steps.

4 (Optional) If the grid computed by Auto select is not dense enough to

capture all system dynamics, specify a different frequency grid.
a Select the User specified frequencies (rad/sec) option, and enter the
frequency grid in log scale.
b Click Apply.

For more information, see “Frequency Grid for Multimodel Computations”

on page 4-136.

5 Design the controller for the nominal plant using graphical or automated
tuning.

4-143
 4 Designing Compensators

For more information on designing a controller, see the following topics:

• “Bode Diagram Design” on page 4-55
• “Root Locus Design” on page 4-81
• “Nichols Plot Design” on page 4-103
• “Automated Tuning Design” on page 4-114

As you design the controller, use the design and analysis plots analyze to
analyze the controller’s effects on the remaining models in the array.

6 (Optional) If the controller design for the nominal model does not meet the
design requirements for the remaining models in the array:
a Specify a different nominal model in the Nominal Model Index field of
the Multimodel Configuration Dialog window.

The design and analysis plots update to show the updated nominal
model. For example, for a nominal model value of 2, the plots appear as
shown in the next figures.

4-144
Control Design Analysis of Multiple Models

4-145
 4 Designing Compensators

b Redesign the controller iteratively.

4-146
 Functions for Compensator Design

Functions for Compensator Design

In this section...
“When to Use Functions for Compensator Design” on page 4-147
“Root Locus Design” on page 4-147
“Pole Placement” on page 4-148
“Linear-Quadratic-Gaussian (LQG) Design” on page 4-151
“Example — Designing an LQG Regulator” on page 4-162
“Example — Designing an LQG Servo Controller” on page 4-165
“Example — Designing an LQR Servo Controller in Simulink” on page 4-168

When to Use Functions for Compensator Design

The term control system design refers to the process of selecting feedback
gains that meet design specifications in a closed-loop control system. Most
design methods are iterative, combining parameter selection with analysis,
simulation, and insight into the dynamics of the plant.

In addition to the SISO Design Tool, you can use functions for a broader range
of control applications, including

• Classical SISO design

• Modern MIMO design techniques such as pole placement and linear
quadratic Gaussian (LQG) methods

Root Locus Design

The following table summarizes the functions for designing compensators
using root locus design techniques.

Function Description
pzmap Pole-zero map
rlocus Evans root locus plot
sgrid
Continuous n ,  grid for root locus plots

4-147
 4 Designing Compensators

Function Description
sisotool Root Locus Design GUI
zgrid
Discrete n ,  grid for root locus plots

Pole Placement
The closed-loop pole locations have a direct impact on time response
characteristics such as rise time, settling time, and transient oscillations.
Root locus uses compensator gains to move closed-loop poles to achieve design
specifications for SISO systems. You can, however, use state-space techniques
to assign closed-loop poles. This design technique is known as pole placement,
which differs from root locus in the following ways:

• Using pole placement techniques, you can design dynamic compensators.

• Pole placement techniques are applicable to MIMO systems.

Pole placement requires a state-space model of the system (use ss to convert

other model formats to state space). In continuous time, such models are of
the form

x = Ax + Bu
y = Cx + Du

where u is the vector of control inputs, x is the state vector, and y is the vector
of measurements.

State-Feedback Gain Selection

Under state feedback u = − Kx , the closed-loop dynamics are given by

x = ( A − BK ) x

and the closed-loop poles are the eigenvalues of A-BK. Using the place
function, you can compute a gain matrix K that assigns these poles to any
desired locations in the complex plane (provided that (A,B) is controllable).

4-148
 Functions for Compensator Design

For example, for state matrices A and B, and vector p that contains the desired
locations of the closed loop poles,

K = place(A,B,p);

computes an appropriate gain matrix K.

State Estimator Design

You cannot implement the state-feedback law u = − Kx unless the full state x
is measured. However, you can construct a state estimate  such that the
law u = − K  retains similar pole assignment and closed-loop properties. You
can achieve this by designing a state estimator (or observer) of the form

 = A + Bu + L( y − C − Du)

The estimator poles are the eigenvalues of A-LC, which can be arbitrarily
assigned by proper selection of the estimator gain matrix L, provided that (C,
A) is observable. Generally, the estimator dynamics should be faster than the
controller dynamics (eigenvalues of A-BK).

Use the place function to calculate the L matrix

L = place(A',C',q).'

where A and C are the state and output matrices, and q is the vector containing
the desired closed-loop poles for the observer.

Replacing x by its estimate  in u = − Kx yields the dynamic output-feedback

compensator

 = [ A − LC − ( B − LD) K ] + Ly
u = − K

Note that the resulting closed-loop dynamics are

4-149
 4 Designing Compensators

⎡ x ⎤ ⎡ A − BK BK ⎤ ⎡ x ⎤
⎢ e ⎥ = ⎢ 0 , e = x −
⎣ ⎦ ⎣ A − LC ⎥⎦ ⎢⎣ e ⎥⎦

Hence, you actually assign all closed-loop poles by independently placing

the eigenvalues of A-BK and A-LC.

Example. Given a continuous-time state-space model

sys_pp = ss(A,B,C,D)

with seven outputs and four inputs, suppose you have designed

• A state-feedback controller gain K using inputs 1, 2, and 4 of the plant

as control inputs
• A state estimator with gain L using outputs 4, 7, and 1 of the plant as
sensors
• Input 3 of the plant as an additional known input

You can then connect the controller and estimator and form the dynamic
compensator using this code:

controls = [1,2,4];
sensors = [4,7,1];
known = [3];
regulator = reg(sys_pp,K,L,sensors,known,controls)

Pole Placement Tools

You can use functions to

• Compute gain matrices K and L that achieve the desired closed-loop pole
locations.
• Form the state estimator and dynamic compensator using these gains.

The following table summarizes the functions for pole placement.

4-150
 Functions for Compensator Design

Functions Description
acker SISO pole placement
estim Form state estimator given estimator gain
place MIMO pole placement
reg Form output-feedback compensator given state-feedback
and estimator gains

The function acker is limited to SISO systems, and you should use it only for
systems with a small number of states. The function place is a more general
and numerically robust alternative to acker.

Caution
Pole placement can be badly conditioned if you choose unrealistic pole
locations. In particular, you should avoid:

• Placing multiple poles at the same location.

• Moving poles that are weakly controllable or observable. This typically
requires high gain, which in turn makes the entire closed-loop
eigenstructure very sensitive to perturbation.

Linear-Quadratic-Gaussian (LQG) Design

Linear-quadratic-Gaussian (LQG) control is a modern state-space technique
for designing optimal dynamic regulators and servo controllers with integral
action (also known as set point trackers). This technique allows you to trade
off regulation/tracker performance and control effort, and to take into account
process disturbances and measurement noise.

To design LQG regulators and set point trackers, you perform the following
steps:

1 Construct the LQ-optimal gain.

2 Construct a Kalman filter (state estimator).

4-151
 4 Designing Compensators

3 Form the LQG design by connecting the LQ-optimal gain and the Kalman
filter.

For more information about using LQG design to create LQG regulators , see
“Linear-Quadratic-Gaussian (LQG) Design for Regulation” on page 4-152.

For more information about using LQG design to create LQG servo controllers,
see “Linear-Quadratic-Gaussian (LQG) Design of Servo Controller with
Integral Action” on page 4-157.

These topics focus on the continuous-time case. For information about

discrete-time LQG design, see the dlqr and kalman reference pages.

Linear-Quadratic-Gaussian (LQG) Design for Regulation

You can design an LQG regulator to regulate the output y around zero in
the following model.

u w v

Kalman -K Plant y
y Filter
u
LQG Regulator

The plant in this model experiences disturbances (process noise) w and is

driven by controls u. The regulator relies on the noisy measurements y to
generate these controls. The plant state and measurement equations take the
form of

x = Ax + Bu + Gw
y = Cx + Du + Hw + v

and both w and v are modeled as white noise.

4-152
 Functions for Compensator Design

Note LQG design requires a state-space model of the plant. You can use ss
to convert other model formats to state space.

To design LQG regulators, you can use the design techniques shown in the
following table.

To design an LQG regulator using... Use the following commands:

A quick, one-step design technique lqg
when the following is true:
For more information, see the lqg
reference page.
• G is an identity matrix and H = 0.
• All known (deterministic) inputs are
control inputs and all outputs are
measured.
• Integrator states are weighted
independently of states of plants and
control inputs.
• The state estimator of the Kalman
filter is based on xˆ [ n| n − 1] .
A more flexible, three-step design lqr, kalman, and lqgreg
technique that allows you to specify:
For more information, see
• Arbitrary G and H.
• “Constructing the Optimal
• Known (deterministic) inputs that State-Feedback Gain for
are not controls and/or outputs that Regulation” on page 4-154
are not measured.
• “Constructing the Kalman
• A flexible weighting scheme for State Estimator” on page 4-154
integrator states, plant states, and
• “Forming the LQG Regulator”
controls.
on page 4-156
• The state estimator of the Kalman
filter based on either xˆ [ n| n] or
xˆ [ n| n − 1] .

4-153
 4 Designing Compensators

Constructing the Optimal State-Feedback Gain for Regulation. You

construct the LQ-optimal gain from the following elements:

• State-space system matrices

• Weighting matrices Q, R, and N, which define the tradeoff between
regulation performance (how fast x(t) goes to zero) and control effort.

To construct the optimal gain, type the following command:

K= lqr(A,B,Q,R,N)

This command computes the optimal gain matrix K, for which the state
feedback law u = − Kx minimizes the following quadratic cost function for
continuous time:

∞
∫0 { x
T
J (u) = Qx + 2 xT Nu + uT Ru}dt

The software computes the gain matrix K by solving an algebraic Riccati

equation.

For information about constructing LQ-optimal gain, including the cost

function that the software minimizes for discrete time, see the lqr reference
page.

Constructing the Kalman State Estimator. You need a Kalman state

estimator for LQG regulation and servo control because you cannot implement
optimal LQ-optimal state feedback without full state measurement.

You construct the state estimate x̂ such that u = − Kxˆ remains optimal for
the output-feedback problem. You construct the Kalman state estimator gain
from the following elements:

• State-space plant model sys

• Noise covariance data, Qn, Rn, and Nn
The following figure shows the required dimensions for Qn, Rn, and Nn.
If Nn is 0, you can omit it.

4-154
 Functions for Compensator Design

Required Dimensions for Qn, Rn, and Nn

Note You construct the Kalman state estimator in the same way for both
regulation and servo control.

To construct the Kalman state estimator, type the following command:

[kest,L,P] = kalman(sys,Qn,Rn,Nn);

This command computes a Kalman state estimator, kest with the following
plant equations:

x = Ax + Bu + Gw
y = Cx + Du + Hw + v
where w and v are modeled as white noise. L is the Kalman gain and P the
covariance matrix.

The software generates this state estimate using the Kalman filter

d
xˆ = Axˆ + Bu + L( y − Cxˆ − Du)
dt
with inputs u (controls) and y (measurements). The noise covariance data

E(wwT ) = Qn , E(vvT ) = Rn , E(wvT ) = N n

determines the Kalman gain L through an algebraic Riccati equation.

4-155
 4 Designing Compensators

The Kalman filter is an optimal estimator when dealing with Gaussian white
noise. Specifically, it minimizes the asymptotic covariance

(
lim E ( x − xˆ ) ( x − xˆ )
t→∞
T
)
of the estimation error x − xˆ .

u Kalman x
y Estimator

For more information, see the kalman reference page. For a complete example
of a Kalman filter implementation, see Kalman Filtering.

Forming the LQG Regulator. To form the LQG regulator, connect the
Kalman filter kest and LQ-optimal gain K by typing the following command:

regulator = lqgreg(kest, K);

This command forms the LQG regulator shown in the following figure.

y kest -K
u
LQG Regulator

The regulator has the following state-space equations:

d
xˆ = [ A − LC − ( B − LD) K ]xˆ + Ly
dt
u = − Kxˆ

For more information on forming LQG regulators, see the lqgreg reference
page and LQG Regulation: Rolling Mill Example.

4-156
 Functions for Compensator Design

Linear-Quadratic-Gaussian (LQG) Design of Servo Controller

with Integral Action
You can design a servo controller with integral action for the following model:

v
u y w

kest y
x u Plant
-K
r-y
r
-
Integrator x
i

y
LQG Servo Controller

The servo controller you design ensures that the output y tracks the reference
command r while rejecting process disturbances w and measurement noise v.

The plant in the previous figure is subject to disturbances w and is driven by

controls u. The servo controller relies on the noisy measurements y to generate
these controls. The plant state and measurement equations are of the form

x = Ax + Bu + Gw
y = Cx + Du + Hw + v

and both w and v are modeled as white noise.

Note LQG design requires a state-space model of the plant. You can use ss
to convert other model formats to state space.

To design LQG servo controllers, you can use the design techniques shown
in the following table.

4-157
 4 Designing Compensators

To design an LQG servo controller Use the following commands:

using...
A quick, one-step design technique lqg
when the following is true:
For more information, see the lqg
reference page.
• G is an identity matrix and H = 0.
• All known (deterministic) inputs are
control inputs and all outputs are
measured.
• Integrator states are weighted
independently of states of plants and
control inputs.
• The state estimator of the Kalman
filter is based on xˆ [ n| n − 1] .
A more flexible, three-step design lqi, kalman, and lqgtrack
technique that allows you to specify:
For more information, see
• Arbitrary G and H.
• “Constructing the Optimal
• Known (deterministic) inputs that State-Feedback Gain for Servo
are not controls and/or outputs that Control” on page 4-158
are not measured.
• “Constructing the Kalman
• A flexible weighting scheme for State Estimator” on page 4-159
integrator states, plant states, and
• “Forming the LQG Servo
controls.
Control” on page 4-161
• The state estimator of the Kalman
filter based on either xˆ [ n| n] or
xˆ [ n| n − 1] .

Constructing the Optimal State-Feedback Gain for Servo Control. You

construct the LQ-optimal gain from the

• State-space plant model sys

• Weighting matrices Q, R, and N, which define the tradeoff between tracker
performance and control effort

4-158
 Functions for Compensator Design

To construct the optimal gain, type the following command:

K= lqi(sys,Q,R,N)

This command computes the optimal gain matrix K, for which the state
feedback law u = − Kz = − K[ x; xi ] minimizes the following quadratic cost
function for continuous time:

∫0 {z }
∞ T
J (u) = Qz + uT Ru + 2 zT Nu dt

The software computes the gain matrix K by solving an algebraic Riccati

equation.

For information about constructing LQ-optimal gain, including the cost

function that the software minimizes for discrete time, see the lqi reference
page.

Constructing the Kalman State Estimator. You need a Kalman state

estimator for LQG regulation and servo control because you cannot implement
LQ-optimal state feedback without full state measurement.

You construct the state estimate x̂ such that u = − Kxˆ remains optimal for
the output-feedback problem. You construct the Kalman state estimator gain
from the following elements:

• State-space plant model sys

• Noise covariance data, Qn, Rn, and Nn
The following figure shows the required dimensions for Qn, Rn, and Nn.
If Nn is 0, you can omit it.

4-159
 4 Designing Compensators

Required Dimensions for Qn, Rn, and Nn

Note You construct the Kalman state estimator in the same way for both
regulation and servo control.

To construct the Kalman state estimator, type the following command:

[kest,L,P] = kalman(sys,Qn,Rn,Nn);

This command computes a Kalman state estimator, kest with the following
plant equations:

x = Ax + Bu + Gw
y = Cx + Du + Hw + v
where w and v are modeled as white noise. L is the Kalman gain and P the
covariance matrix.

The software generates this state estimate using the Kalman filter

d
xˆ = Axˆ + Bu + L( y − Cxˆ − Du)
dt
with inputs u (controls) and y (measurements). The noise covariance data

E(wwT ) = Qn , E(vvT ) = Rn , E(wvT ) = N n

determines the Kalman gain L through an algebraic Riccati equation.

4-160
 Functions for Compensator Design

The Kalman filter is an optimal estimator when dealing with Gaussian white
noise. Specifically, it minimizes the asymptotic covariance

(
lim E ( x − xˆ ) ( x − xˆ )
t→∞
T
)
of the estimation error x − xˆ .

u Kalman x
y Estimator

For more information, see the kalman reference page. For a complete example
of a Kalman filter implementation, see Kalman Filtering.

Forming the LQG Servo Control. To form a two-degree-of-freedom LQG

servo controller, connect the Kalman filter kest and LQ-optimal gain K by
typing the following command:

servocontroller = lqgtrack(kest, K);

This command forms the LQG servo controller shown in the following figure.

u y
kest
x u
-K
r-y
r
-
Integrator x
i

y
LQG Servo Controller

The servo controller has the following state-space equations:

4-161
 4 Designing Compensators

⎡ xˆ ⎤ ⎡ A − BK x − LC + LDK x − BK i + LDK i ⎤ ⎡ xˆ ⎤ ⎡0 L ⎤ ⎡ r ⎤

⎢ ⎥=⎢ ⎥ ⎢ x ⎥ + ⎢ I − I ⎥ ⎢ y⎥
⎢⎣ x i ⎥⎦ ⎣ 0 0 ⎦⎣ i⎦ ⎣ ⎦⎣ ⎦
⎡ xˆ ⎤
u = [− K x − Ki ] ⎢ ⎥
⎣ xi ⎦

For more information on forming LQG servo controllers, including how

to form a one-degree-of-freedom LQG servo controller, see the lqgtrack
reference page.

Example — Designing an LQG Regulator

As an example of LQG design, consider the following regulation problem.

10
s+10 n
d
y
y F(s) Plant
u u+d

LQG Regulator

The goal is to regulate the plant output y around zero. The input disturbance
d is low frequency with power spectral density (PSD) concentrated below
10 rad/s. For LQG design purposes, it is modeled as white noise driving a
lowpass filter with a cutoff at 10 rad/s, shown in the following figure.

10
w (white noise) s+10 d (colored noise)

For simplicity, this noise is modeled as Gaussian white noise with variance
of 1.

The following figure shows the Bode magnitude of the shaping filter.

4-162
 Functions for Compensator Design

Bode Magnitude of the Lowpass Filter

There is some measurement noise n, with noise intensity given by

E(n2 ) = 0.01

Use the cost function

∞
∫0 (10 y
2
J (u) = + u2 ) dt

to specify the tradeoff between regulation performance and cost of control.

The following equations represent an open-loop state-space model:

x = Ax + Bu + Bd (state equations)
y = Cx + n (measurements)

where (A,B,C) is a state-space realization of 100 /(s2 + s + 100) .

4-163
 4 Designing Compensators

The following commands design the optimal LQG regulator F(s) for this
problem:

sys = ss(tf(100,[1 1 100])) % State-space plant model

% Design LQ-optimal gain K

K = lqry(sys,10,1) % u = -Kx minimizes J(u)

% Separate control input u and disturbance input d

P = sys(:,[1 1]);
% input [u;d], output y

% Design Kalman state estimator Kest.

Kest = kalman(P,1,0.01)

% Form LQG regulator = LQ gain + Kalman filter.

F = lqgreg(Kest,K)

These commands returns a state-space model F of the LQG regulator F(s).

The lqry, kalman, and lqgreg functions perform discrete-time LQG design
when you apply them to discrete plants.

To validate the design, close the loop with feedback, create and add the
lowpass filter in series with the closed-loop system, and compare the open-
and closed-loop impulse responses by using the impulse function.

% Close loop
clsys = feedback(sys,F,+1)
% Note positive feedback.

% Create the lowpass filter and add it in series with clsys.

s = tf('s');
lpf= 10/(s+10) ;
clsys_fin = lpf*clsys;

% Open- vs. closed-loop impulse responses

impulse(sys,'r--',clsys_fin,'b-')

These commands produce the following figure, which compares the open- and
closed-loop impulse responses for this example.

4-164
 Functions for Compensator Design

Comparison of Open- and Closed-Loop Impulse Response

Example — Designing an LQG Servo Controller

This example shows you how to design a servo controller for the following
system.

v
w
r y
trksys u Plant
y
LQG Servo Controller

4-165
 4 Designing Compensators

The plant has two states (x), two control inputs (u), two random inputs (w),
one output (y), measurement noise for the output (v), and the following state
and measurement equations:

x = Ax + Bu + Gw
y = Cx + Du + Hw + v
where

⎡0 1 0 ⎤ ⎡ 0 .3 1 ⎤ ⎡ −0.7 1.12⎤
A = ⎢⎢0 0 1 ⎥⎥ ⎢
B=⎢ 0 1 ⎥⎥ G = ⎢⎢ −1.17 1 ⎥⎥
⎢⎣1 0 0 ⎥⎦ ⎢⎣ −0.3 0.9 ⎥⎦ ⎢⎣ 0.14 1.5 ⎥⎦

C = [1.9 1.3 1] D = [0.53 −0.61] H = [ −1.2 −0.89]

The system has the following noise covariance data:

⎡4 2⎤
Qn = E(wwT ) = ⎢ ⎥
⎣2 1⎦
Rn = E(vvT ) = 0.7

Use the following cost function to define the tradeoff between tracker
performance and control effort:

∞⎛ ⎡1 0 ⎤ ⎞
∫0 ⎜⎝ 0.1x
T
J (u) = x + xi2 + uT ⎢ ⎥ u ⎟dt
⎣0 2 ⎦ ⎠

To design an LQG servo controller for this system:

1 Create the state space system by typing the following in the MATLAB
Command Window:

A = [0 1 0;0 0 1;1 0 0];

B = [0.3 1;0 1;-0.3 0.9];
G = [-0.7 1.12; -1.17 1; .14 1.5];
C = [1.9 1.3 1];
D = [0.53 -0.61];

4-166
 Functions for Compensator Design

H = [-1.2 -0.89];
sys = ss(A,[B G],C,[D H]);

2 Construct the optimal state-feedback gain using the given cost function
by typing the following commands:

nx = 3; %Number of states
ny = 1; %Number of outputs
Q = blkdiag(0.1*eye(nx),eye(ny));
R = [1 0;0 2];
K = lqi(ss(A,B,C,D),Q,R);

3 Construct the Kalman state estimator using the given noise covariance
data by typing the following commands:

Qn = [4 2;2 1];
Rn = 0.7;
kest = kalman(sys,Qn,Rn);

4 Connect the Kalman state estimator and the optimal state-feedback gain to
form the LQG servo controller by typing the following command:

trksys = lqgtrack(kest,K)

This command returns the following LQG servo controller:

>> trksys = lqgtrack(kest,K)

a =
x1_e x2_e x3_e xi1
x1_e -2.373 -1.062 -1.649 0.772
x2_e -3.443 -2.876 -1.335 0.6351
x3_e -1.963 -2.483 -2.043 0.4049
xi1 0 0 0 0

b =
r1 y1
x1_e 0 0.2849
x2_e 0 0.7727
x3_e 0 0.7058
xi1 1 -1

4-167
 4 Designing Compensators

c =
x1_e x2_e x3_e xi1
u1 -0.5388 -0.4173 -0.2481 0.5578
u2 -1.492 -1.388 -1.131 0.5869

d =
r1 y1
u1 0 0
u2 0 0

Input groups:
Name Channels
Setpoint 1
Measurement 2

Output groups:
Name Channels
Controls 1,2

Continuous-time model.

Example — Designing an LQR Servo Controller in

Simulink
The following figure shows a Simulink block diagram shows a tracking
problem in aircraft autopilot design. To open this diagram, type lqrpilot at
the MATLAB prompt.

4-168
 Functions for Compensator Design

Key features of this diagram to note are the following:

• The Linearized Dynamics block contains the linearized airframe.

• sf_aerodyn is an S-Function block that contains the nonlinear equations

for ( ,  ) = (0, 15 ) .

• The error signal between  and the ref is passed through an integrator.
This aids in driving the error to zero.

State-Space Equations for an Airframe

Beginning with the standard state-space equation

x = Ax + Bu

where

x = [u, v, w, p, q, r, ,  ]T

4-169
 4 Designing Compensators

The variables u, v, and w are the three velocities with respect to the body
frame, shown as follows.

Body Coordinate Frame for an Aircraft

The variables  and  are roll and pitch, and p, q, and r are the roll, pitch,
and yaw rates, respectively.

The airframe dynamics are nonlinear. The following equation shows the
nonlinear components added to the state space equation.

⎡ − g sin θ ⎤
⎢ g cos θ sin φ ⎥
⎢ ⎥
⎢ g cos θ cos φ ⎥
⎢ ⎥
⎢ 0 ⎥
x = Ax + Bu +
⎢ 0 ⎥
⎢ ⎥
⎢ 0 ⎥
⎢ q cos φ − r sin φ ⎥
⎢ ⎥
⎢⎣(q sin φ + r cos φ) ⋅ tan θ ⎥⎦

Nonlinear Component of the State-Space Equation

To see the numerical values for A and B, type

load lqrpilot
A, B

at the MATLAB prompt.

4-170
 Functions for Compensator Design

Trimming

For LQG design purposes, the nonlinear dynamics are trimmed at  = 15 and
p, q, r, and θ set to zero. Since u, v, and w do not enter into the nonlinear term

in the preceding figure, this amounts to linearizing around ( ,  ) = (0, 15 )

with all remaining states set to zero. The resulting state matrix of the
linearized model is called A15.

Problem Definition
The goal to perform a steady coordinated turn, as shown in this figure.

Aircraft Making a 60° Turn

To achieve this goal, you must design a controller that commands a steady
turn by going through a 60° roll. In addition, assume that θ, the pitch angle,
is required to stay as close to zero as possible.

Results
To calculate the LQG gain matrix, K, type

lqrdes

at the MATLAB prompt. Then, start the lqrpilot model with the nonlinear
model, sf_aerodyn, selected.

This figure shows the response of  to the 60° step command.

4-171
 4 Designing Compensators

Tracking the Roll Step Command

As you can see, the system tracks the commanded 60° roll in about 60 seconds.

Another goal was to keep θ, the pitch angle, relatively small. This figure
shows how well the LQG controller did.

Minimizing the Displacement in the Pitch Angle, Theta

Finally, this figure shows the control inputs.

4-172
 Functions for Compensator Design

Control Inputs for the LQG Tracking Problem

Try adjusting the Q and R matrices in lqrdes.m and inspecting the control
inputs and the system states, making sure to rerun lqrdes to update the LQG
gain matrix K. Through trial and error, you may improve the response time of
this design. Also, compare the linear and nonlinear designs to see the effects
of the nonlinearities on the system performance.

4-173
 4 Designing Compensators

4-174
 Index

A
Index regulators 4-156
automated tuning design 4-114 root locus 4-147
state estimator 4-149
state estimator, Kalman 4-154
B
building models
E
MIMO 2-13
SISO 2-6 estimator
gain 4-149
Kalman 4-154
C exporting compensators and models 4-49
cell array 2-16
changing
F
units on a plot 4-47
comparing final time. See also time response 3-40
models in plots 3-44 frequency response 3-41
compensator design customizing plots 3-47
multi-loop 4-120 plotting 3-44
compensators functions
editing in SISO Design Tool 4-96 time and frequency response 3-37
exporting 4-49
feedback (state estimator design) 4-149 G
concatenation
gain
model 2-16
estimator, Kalman 4-155 4-160
control design analysis of multiple models 4-132
selection 4-82
Control System Toolbox product
state-feedback 4-154
related products 1-4
Graphical Tuning window 4-26
covariance
changing units on a plot 4-47
noise 4-155 4-160
display 4-43
right-click menus 4-48
D using 4-42
data markers 3-42
denominator I
specification 2-16
input vector 2-2
design
Kalman estimator 4-154
LQG Tracker, example of 4-168 K
LQG, description 4-151 Kalman estimator
pole placement 4-148 continuous 4-154

Index-1
 Index

gain 4-155 4-160 MIMO 3-40

steady-state 4-154 model order reduction 2-29
Kalman filter. See Kalman estimator models
building 2-13
multi-loop design 4-120
L
linear models
comparing multiple models in plots 3-44 N
exporting 4-49 noise
frequency response. 3-41 covariance 4-155 4-160
model order reduction 2-29 white 3-38
time response 3-37 notch filter
LQG (linear quadratic Gaussian) method parameters 4-74
continuous LQ regulator 4-154 numerator
design. See design, LQG specification 2-16
gain, optimal state-feedback 4-154
Kalman state estimator 4-154
O
LQ-optimal gain 4-154
regulator, designing 4-164 operations on LTI models
regulator, forming 4-156 concatenation 2-16
weighting matrices 4-154 4-159 output vector 2-2
LTI Viewer overshoot 3-37
adding plot types 3-16
changing models 3-20 P
changing plot types 3-14
plotting
command line initializing 3-18
changing units 4-47
comparing multiple models 3-17
customizing plots 3-47
importing models 3-9
multiple systems 3-44
introduction to 3-8
right-click menus 3-41
models, importing 3-9
time responses 3-37
opening 3-9
pole placement 4-148
plot options
conditioning 4-151
rise time 3-11
settling time 3-19
right-click menus 3-10 R
SISO Design Task 4-26 reduced-order models 2-29
regulation 4-152 4-163
M performance 4-154
Riccati equation 4-154
markers, data 3-42
right-click menus

Index-2
 Index

LTI Viewer 3-10 compensators, editing 4-96

plot responses 3-41 editing preferences 4-47
rise time 3-37 exporting compensator and models 4-49
root locus feedback structures 4-34
compensator gain 4-88 importing models into 4-30
designing 4-147 loop responses 4-36
root locus design 4-81 Nichols plot design 4-103
functions for 4-147 opening 4-27
opening a Nichols plot 4-104
root locus design 4-81
S
storing and retrieving designs 4-51
Servo controller with integral action 4-157 Using the Graphical Tuning window 4-42
Set point tracking 4-157 SISO Tool Preferences Editor 4-47
settling time 3-37 state
SISO Design Task in the Control and Estimation estimator 4-149
Tools Manager 4-26 vector 2-2
SISO Design Task pane 4-26 state estimator design
Analysis Plots page 4-36 regulators/compensators 4-149
Automated Tuning page 4-116 state-space models 2-2
Design History page 4-52 steady state error 3-37
using 4-29
SISO Design Tool 4-26
adding a lead network 4-63 T
adding a lead network to a Nichols plot 4-110 time response
adding a notch filter 4-71 customizing plots 3-47
adding an integrator 4-59 final time 3-40
adding an integrator to the Nichols functions 3-37
plot 4-108 MIMO 3-40
adding poles and zeros 4-91 plotting 3-44
adjusting compensator gain in a Nichols time range 3-40
plot 4-105 to white noise 3-38
adjusting the compensator gain in Bode vector of time samples 3-41
design 4-56 transfer functions
applying automated PID tuning 4-116 MIMO 2-15
automated tuning design 4-114 transfer functions (TF) 2-2
Bode Diagram Design 4-55
changing the compensator gain, Bode
U
magnitude 4-56
changing the compensator gain, root units
locus 4-88 changing on a plot 4-47

Index-3
 Index

W Z
white noise 3-38 zero-pole-gain (ZPK) models 2-3
See also noise

Index-4