CISSICAL FEEDEACK CONTROL
CONTROL ENGINEERING
A Series of Reference Books and Textbooks Editor NEIL MUNRO, PH.D., D.W. Professor Applied Control Engineering University of Manchester Institute of Science and Technology Manchester, United Kingdom
1. Nonlinear Control of Electric Machinery, Damn M, Dawson, Jun Hu, and Timothy C. Buw 2. Computational Intelligence in Control Engineering, Robed E King 3, Quantitative Feedback Theory: Fundamentals and Applications, Con&tsMine H. Houpis and StevenJ. Rasmussen 4, &lfmLeaming Control of Finite Markov Chains, A. S. Poznyak, K. Najim, plnd E G&mezRamimz st Control,andFiltering for TimeDelay Systems,Magdi S. Mahmoud 6, Cla~si@l FeedbackControl:WithMATLAB, Boris J. LuneandPaul J.
EflflgM
Additional Volumesin Preparation
.CLASSICAL FEED6ACK CONTROL
With M A T ~ B
Boris 3. Lurie Paul J. Enright
Jet Propulsion Laboratory California Institute of Technology
M A R C E L
MARCEL DEKKER, INC.
D E K K E R
NEW YORK BASEL
Library of Congress CataloginginPublication Data
Lurie, B.J. Classical feedback control with MATLAB / Boris J. Lurie, PaulJ. Enright. p.cm. (Control engineering;6) ISBN 0824703707 1. Feedback control systems. I. Enright, Paul J. 11. Title. 111. Control engineering (Marcel Dekker);6. TJ2 16 .L865 2000 629.8’34~21 99087832
This book is printed on acidfree paper.
Headquarters Marcel Dekker, Inc. York, NY 10016 270 Madison Avenue, New tel: 2 126969000; fax: 2 126854540 Eastern Hemisphere Distribution Marcel Dekker AG Hutgasse 4, Postfach 8 12, CH4001 Basel, Switzerland tel: 41612618482; fax: 41612618896
”
World Wide Web http:l/www.dekker.com
The publisher offers discounts on this book when ordered in bulk quantities. For more information, write to Special SaledProfessional Marketing at the headquarters address above.
Copyright0 2000 by Marcel Dekker, Inc. Ail Rights Reserved.
Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming, and recording, or by any information storage and retrieval system, without permission in writing from the publisher. Current printing (last digit) 1 0 9 8 7 6 5 4 3 2 1
PRINTEDIN THE UNITED STATES OF AMERICA
PREFACE
CYassical Feedback Control 'describes design and implementation of highperformance feedback controllers forengineeringsystems.Thebookemphasizesthefrequencydomainapproach which is widelyusedinpracticalengineering. It presentsdesign methods for linear and nonlinear highorder controllers for singleinput, singleoutput and multiinput, multioutput, analog and digital control systems. Moderntechnologyallowsimplementation of highperformancecontrollers at a very OW cost. Conversely, several analysis tools which were previously considered an inherent part of control system courses limit the design to loworder (and therefore lowperformance) compensators. Among these are the rootlocus method, the detection of rightsided polynomial roots using the RouthHurwitz criterion, and manual calculations using the Laplace and Fourier transforms, These methods have been rendered obsolete by computers and are granted only a brief treatment in the book, making room for loop shaping, Bode integrals, structural simulation of complex systems, multiloop systems, and nonlinear controllers, all of which are essential for good design practice. In the design philosophy adopted by Classical Feedback Control, Bode integral relations play a key role. The integrals are employed to estimate the available system performance and to determine the frequency responses that maximize the disturbance rejection and the feedback bandwidth. This ability to quickly estimate the attainable performance is critical for systemlevel trades in the design of complex engineering systems, of which the controller is one of many subsystems. Only at the final design stageandonly for thefinallyselectedoption of thesystemconfiguration do the compensators need to be designed in detail, by approximation of thealreadyfound optimal frequency responses. Nonlinear dynamic compensation is employed to provide global and process stability, and improve to transient responses. The nearlyoptimal highorder compensators are then economically implemented using analog and digital technology. The first six chapters support a onesemester course in linear control. The rest of the book considers the issues of complex system simulation, robustness, global stability, andnonlinearcontrol.Throughoutthebook, MATLAB and SPICE are used for simulation and design; no preliminary experience with this software is required. The student should have some knowledge of the Laplace transform and frequency responses; the required theory is reviewed in Appendix 2. Appendix 1 is an elementary treatment of feedback control, which can be used as an introduction to the course. It is the authors' intention to make Classical Feedback Controlnot only a textbook but also a reference for students as they become engineers, enabling them to design highperformance controllers and easing the transition from school to the competitive industrial environment. The methods described in this book were used by the authors and their colleagues as the major design tools for feedback loops of aerospace and telecommunication systems. We would be grateful forany comments, corrections, and criticism our readers may take the trouble to communicate to us, via Email b.j.lurie8jpl.nasa.gov or addressed CA 91 109. to B. J. Lurie, 198/326,JPL, 4800 Oak Grove Drive, Pasadena
Acknowledgment. We thank A11a Lurie for technical editing and acknowledge the generous help of Asif Ahmed. We greatly appreciate previous discussions on many
iii
iv
Preface
control issues with Professor Isaac Horowitz, and collaboration, comments and advice of our colleagues at the Jet Propulsion Laboratory, and especially of John O'Brien, Daniel Chang, Edward Kopf (who told the authors about the jumpresonance in the attitudecontrolloop of theMariner 10 spacecraft), Drs.AlexanderAbramovichi, Thomas Bak, David Bayard (who helped edit the chapter on adaptive systems), Dimitrius Boussalis, GunShing Chen (who contributed Appendix 7), Ali Ghavimi (who contributed to Appendix A13.14), Fred Hadaegh (who coauthored several papers on which Chapter 13 is based), John Hench (who contributed the digital signal profiling function in Section 5.'1l), Kenneth Lau, Wei Min Liu, Mehran Mesbahi, Gregory Neat, Samuel Sirlin, John Spanos, andMichael Zak. Suggestions and corrections made by Professors Randolph Bird, Osita Nwokah, and Roy Smith allowed us to improve the manuscript. Dr. Jason Modisette read the manuscriptand suggested many changes and corrections. Allan Schier contributed the example of amechanicalsnakecontrolin Appendix A13.15. To all of them we extend oursincere gratitude. Boris J. Lurie Paul J. Enright
Contents
ix
Chapter 8 NIRODUCTION TO ALTERNATIVE METHODSOF ONT TROLLER DESIGN. . . . . . . . . . . . . . . . . . . . . .
8.1 8.2 8.3 8.4 8.5
QFT . . . . . . . . . . . . . . . . . . . Root locus and pole placement methods . Statespacemethodsandfullstatefeedback LQRandLQG . . . . . . . . . . . . . €2 psynthesis. and linear matrix inequalities.
. . . .
............ ............. ............. ............ ............
245 245 247 249 253 255
Chapter 9
ADAPTIVESYSTEMS . . . . . . . . . . . . . . . . . . . . . . .
9.1 9.2 9.3 9.4 9.5 9.6 9.7
Benefits of adaptationtotheplantparametervariations . . . . . . . Static and dynamic adaptation . . . . . . . . . . . . . . . . . . . . Plant transfer function identification .............. Flexible and n.p. plants . . . . . . . . . . . . . . . . . . . . . . . Disturbance and noise rejection ................. Pilot signals andditheringsystems . . . . . . . . . . . . . . . . Adaptive filters. . . . . . . . . . . . . . . . . . . . . . . . . . .
..
..
.. ..
257 257 259 259 260 261 262 264
Chapter 10 PROVISION OF GLOBAL STABILITY . . . . . . . . . . . . . 266
10.1 10.2 10.3 10.4 10.5
10.6 10.7
10.8
Nonlinearities of the actuator. feedback path.andplant . . . . . . . . . Types of selfoscillation . . . . . . . . . . . . . . . . . . . . . . Stabilityanalysis of nonlinearsystems . . . . . . . . . . . . . . . . 10.3.1 Local. linearization . . . . . . . . . . . . . . . . . . . . 10.3.2 Global stability . . . . . . . . . . . . . . . . . . . . . . Absolute stability . . . . . . . . . . . . . . . . . . . . . . . . . Popov criterion . . . . . . . . . . . . . . . . . . . . . . . . . . 10.5.1 Analogy topassivetwopoles’connection .......... 10.5.2 Differentforms of thePopovcriterion . . . . . . . . . . . . Applications of Popov criterion . . . . . . . . . . . . . . . . . . . 10.6.i Lowpass system with maximum feedback ......... 10.6.2 Bandpass system with maximum feedback . . . . . . . . . Absolutely stable systemswithnonlineardynamiccompensation . . . . 10.7.1 Nonlinear dynamic compensator .............. 10.7.2 Reductiontoequivalentsystem . . . . . . . . . . . . . . . 10.7.3 Design examples . . . . . . . . . . . . . . . . . . . . . Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
266 267 269 269 270 270 271 271 274 275 275 275 276 276 277 278 286
Chapter II DESCRIBINGFUNCTIONS . . . . . . . . . . . . . . . . . . . .
11.1
Harmonic balance . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Harmonic balance analysis . . . . . . . . . . . . . . . . 11.1.2 Harmonic balance accuracy . . . . . . . . . . . . . . . .
\
. .
289 289 289 290
CONTENTS
PnFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TOINSTRUCTORS . . . . . . . . . . . . . . . . . . . . . . . . .
iii xiii
Chapter I FEEDBACKANDSENSITIVITY . . . . . . . . . . . . . . . . . . 1
1 . 1 1 . 2 1 . 3 1 . 4
1 . 5 1 . 6
1.7
1 . 8 1 . 9 1 . 1 0 1.11 1 . 1 2 1 . 1 3
Feedback control system ....................... 1 Feedback positive and negative . . . . . . . . . . . . . . . . . . . . 3 Large feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Loopgainandphasefrequencyresponses ................ 6 1 . 4 . 1 Gain and phase responses ................... 6 1 . 4 . 2 Nyquist diagram . . . . . . . . . . . . . . . . . . . . . . . 9 1 . 4 . 3 Nicholschart . . . . . . . . . . . . . . . . . . . . . . . . . 10 Disturbance rejection . . . . . . . . . . . . . . . . . . . . . . . . 11 Example of system analysis . . . . . . . . . . . . . . . . . . . . . 12 Effect of feedback on theactuatornonlinearity . . . . . . . . . . . . . 15 Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Effect of finite plantparametervariations . . . . . . . . . . . . . . . . 18 Automaticsignallevelcontrol . . . . . . . . . . . . . . . . . . . . . . 19 Lead and PID compensators . . . . . . . . . . . . . . . . . . . . . 20 Conclusionandalookahead . . . . . . . . . . . . . . . . . . . . . 20 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Chapter 2 FEEDFORWARD. MULTILOOP. AND MINI0 SYSTEMS
2 . 1 2 . 2 2 . 3 2 . 4 2 . 5 2 . 6 2 . 7 2 . 8 2 . 9 2 . 1 0
Command feedforward . . . . . . . . . . . . . . . . . . . . . . . . Prefilterandthefeedbackpathequivalent ............. Error feedforward . . . . . . . . . . . . . . . . . . . . . . . . . . Black's feedforward method ..................... Multiloop feedback systems ..................... Local. common. and nested loops ................. Crossed loops and maidvernier loops . . . . . . . . . . . . . . . Manipulations of block diagrams and calculations of transfer functions. MINI0 feedback systems . . . . . . . . . . . . . . . . . . . . . . . Problems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 31 31 . . 33
. . 40
43 46
. . . 38
34 35 36 . 37
Chapter 3 FREQUENCY RESPONSE METHODS . . . . . . . . . . . . . . . 52
3 . 1 3 . 2 3 . 3
Conversion of timedomainrequirementstofrequencydomain 3 . 1 . 1 Approximaterelations . . . . . . . . . . . . . . . 3 . 1 . 2 Filters . . . . . . . . . . . . . . . . . . . . . . Closedloop transient response .............. Root locus ........................
. . . . . . 52 . . . . . . 52 . . . . . . 56
...... ......
58
59
V
. .. . .. . . . . . 116 4. . . . . .3 Example of a system having a loop response with a Bode step.. .1 Minimumphasefunctions . . . . . . . .4 Nonminimumphaseshift . . . . .. . . . . . . . . 74 3. . . . . . . . . . . . . . . .. . . . . . . .2 Accuracy of the loop shaping Asymptotic Bode diagram . . . . . .. 86 Problems .5 3.6 Lightly damped flexible plants.6 Bandpass systems . . . . . .63 Nyquist criterion for a system with an unstable plant. . .1 Feedback bandwidth . .. .9. . . .3. . .. . 75 3. . . . .. . ..12 3. 78 3. . .9 Contents 3. . . . . . .. . . . .. . . . . . . . . .. ..4 3. . . . . . .1 4. . . . .8 3.. . . . .2. . .. . 107 4. . . 76 3. . . . . 121 Problems .3. . . . . . . . .2. . .vi 3. . . . .. . 81 From the Nyquist diagram to the Bode diagram . . . . . .3. . . . . .6 . . . . ..2. .. . . . . . . . . . . . . . 114 4. . . . . . . .3 Integral of resistance . . . .3. 111 4. . . 67 . . . . . . .2 Integral of feedback. .... . . .9. . . . . . . . . . ... 96 4. . .9.14 Nyquist stability criterion . . 94 Optimality of the compensator design . . .. . 105 4. . .. . . . . . . . . . . . . . . .. . 106 4.. .. . . . . . . .. ... . . 100 4. . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . .2. . . . . . . . . . . . . .5 Gain integral over finite bandwidth . . . . .. . .. .. . . . . . . . .. . . .13 3. . .. . . . . . . . .... . . . . . . . . . . links . . . . . .6 3. . . . . . . . 4.. . . 74 3. . .. . .. . . .. . . . . . . . . .. . ... . . . . . . . . . . . . .. . .11 3. . . . . 120 Shaping parallel channel responses . . 61 Robustness and stability margins . . . . 94 Feedback maximization . . . . . . .4 Integral of theimaginarypart . . . . . . . ...7 Unstable plants . . . . . . . . 85 Ladder networksand parallel connectionsof m. . . . . .9. . . . . .. .10 3. . ..1 5. . . . . . . . . . . . . . . . . . . . ..2. . . .. . . . . collocated and noncollocated control .2 Sensor noise at the system output .. . . . . . . . . . . .2.5 4. .. . . . . .. .7 3. . . . . . . . . . 69 Successive loop closure stability criterion Nyquist diagramsfor the loop transfer functions with poles at the origin .1 Structural design ... .. . . . . .. . . . . ..4 4. . . . . . . .5 Bode cutoff . . 71 Bode integrals.. . . . . . . .. 113 4. . . 88 Chapter 4 SHAPING THELOOP FREQUENCY RESPONSE 4. . . . . .. . . . . . . . . . . . .. . .. 119 Coupling in MIMO systems. . . . . . . . 83 Nonminimum phase lag . . .9. . . . . . .3 Sensor noise at the actuator input . 125 Chapter 5 130 130 131 COMPENSATOR DESIGN. .. 96 4. . . . .. . . . . . . . . . . . . . . . . . .2 Bodestep . .. . . . . . . .3. .2. 108 Feedback bandwidth limitations . . . . . . . . .. . 112 4. . . . . . . 79 Phase calculations . . . .3.. . . . . . . . . . . . . . . . . .. 5. . . 97 4.5 Plant tolerances . . .. .3 . . .p.4 Reshapingthefeedbackresponse . . . . . . . . . .. . . . 110 4.. . . . .6 Phasegain relation . .. . . . . . . . . . .. . . . .2 4. . . . . . . . . . . 110 4. . . . .7 Nyquiststable systems . . . . . .. . .. 78 3. . . . . . .9. . . . .3. .
. . . . 6..2 Symmetrical regulator . . . .. . .. . . . . . ... .. Problems .. . . . . 6. . .. .. . .. . .1 Switchedcapacitor circuits . . .. Parallel connection of links ..... . .. . . .9 Active RC circuits .10. .5. 6. .10. . . . . . . .5 6.. . . . . . .. . . . . . Complex poles . ... . . . . . .. .. .. . . . . 6.. . . . . . .3 6. . .. .. . . . . . .. andpackaging . ..6.. . .1 Discrete trapezoidal integrator . . . .. . . .. . . ... . . :. ..6. . . . .. . . .. . .5 Transferfunctions with multiplepolesandzeros .. ...8 6. . .. . .7 5.. .. . . . . .. ... . ..andcomputercode . .. .. . . .. . .. . . . ... . .1. .. ... . . .. . . .. . . 6. .. ..3 Noninverting configuration . . . . .2 Signal transmission . .3 Hardware implementation .. .. .. 6. . . .. . . . . . . . . .. . . . . . .1. . . . . Loop response measurements .. . . .. . . . .. 6.. . .. . .6 5. . . . . . . . . . . . 6. . .. .. .. . . .12 Approximation of constantslope gain response . . . . . . . . . . .. .. 6.10. ...7.. . 170 170 171 172 173 174 176 178 180 180 182 183 184 184 185 186 186 187 189 190 190 192 193 193 194 196 196 200 . . . . . . 6. .. . . . . . . .10 vii 133 135 137 138 141 143 146 146 146 148 151 151 153 156 157 159 159 5. . . .. . 5.. 6. . . . .. .. . .1. . . . 6. . ... .. . . Chapter 6 ANALOGCONTROLLERIMPLEMENTATION . . . .. . . .7. . . . . . . ..2 Laplace and Tustin transforms ..3 5..4 6. .. . . . . . .. .. . ..6 Active RC filters . .4 5. . . .. . .. .noise. . .10. ..7. . . .. . . . . .1 PID compensator .3 Stabilityandtestingissues ...8 5. .2. .. . . . . .. ..10. . .2 TID compensator . .. ... . . Digital compensator design . . . . . . . .2 RCimpedance chart .. . . . . .. 5.. . . 6.. . . . . . . .. .5. . .. . . ..6 6. . . . . 6. .. 6.. .. . . . 6. .. . . . . .. . . .1.. . . . . . . . . . . . .1 Ground . . . . .4. . . . . . . . .. . Miscellaneous hardware issues . PZD tunable controller ..1 Operational amplifier . . analog or digitally controlled . . . . .10. . . . . . 6.. . . . .. ... . .. . .equations. 5.4 Blockdiagrams. . . . . 170 6. .. . 5. Command profiling . . . .. . . . . . . . .4 Opampdynamicrange.. . . . .Contents 5. .1 Bilinear transfer function . . . . .. .. .5 5.. . . . . .. . . .. . . . . . .. .2 6. . . . . .. . . .1. . Problems . . . ... . . . .. . .. . ... .. . . .. . . .. . . . . . . . 5.. .7 Nonlinear links ... . ... . . . . . .. 5. ..5 Compensator design example . .. . . 6. . . . . . ... . . 6. .. . . . .2. . . . . ..1 6. . .5. . .. .. . .2 Integrator and differentiator . . . .. . 5. .. . . . ..1 Cauer and Foster RC twopoles . . . . ... . .. . ... . . .. .. Tunablecompensator with onevariableparameter . . . . .....2 Example of compensator design . Analog compensator. .6 Aliasing and noise . . . . . . . . ...9 5. . . . . . . . .. . . . . .. .. . . .. . . .. . .7 6.. . . ... . . . . . . . .. . .. .1. .. Designanditerations intheelementvaluedomain . . . ..4. . .1. . .10.. . . . . Lead and lag links . .7 Transferfunction for thefundamental . . . ..11 5. Analog and digital controllers .. ... .. . Cascaded links . . . . . . . .. . . .. .3 Design sequence . .... . . .. . . . . . Simulation of a PID controller . .. . . . Switchedcapacitorfilters.
. Junctions of unilateral links. . . . 6 . . . . . . . 2 Electrical analogy to heat transfer . . . . . 5 Compound feedback . . . . . . 8 . . . . . . . . . 1 0 . . . . .. . . . . . . . . . . . . . . . . .. . . . . .. .. . . . . . . . . . .. . . 8 . . . . . . . 4 7 .. 4 . . . . .transformers. . . .. . 2 Ratesensors . . . . . . .. . . .. . 1 . . . . . . . . . . . . . . . .. . . . . . Linear timevariable systems . . . . . . . . . . . .. .. . . . . . . . . . 9 . . . . . . . . 2 Losslessdistributedstructures . . 6 Coulombfriction . . . 7 .. . 7 . 3 Hydraulic systems . . . . . . . . . . . . . 1 2 . . . . . 7 . . .. . . . . . . . . . . . . . 4 Noncollocated control . . . 9 .. . . . . .. 9 . . . . . . . . . .. 1 Large feedback with velocity and force sensors . . . . . . ... . . 9 7 . . . .viii Contents Chapter 7 LINEAR LINKS ANDSYSmM SIMULATION . . . . . . . . . Examples of systemmodeling . . 7 . . .. Sensor noise . . 6 . . . . . . . . . .. . . . 1 Position and angle sensors . 1 .. . 4 . 7 . . . . . . . . .. . . . . 3 Accelerometers . . . 7 . . . . 205 7 . 8 . . .. ... . . . . . Mathematicalanalogiestothefeedbacksystem . . .. . . . . . Effect of feedback on the impedance (mobility). . . . .. . . . . . Problems . 7 . . 1 Mathematical analogies . 2 . ... . . . . . 6 . . . . . . Effect of loadimpedanceonfeedback . 1 Chainconnection of twoports . . . . 7 . . . . . . .. . . . . . . .. . 3 Collocated control . 4 Seriesfeedback . . . . . .. 2 Effect of feedback on the signaltonoise ratio . 7 . .. . . . . . . . . 8 7 . . . . . . . .. . . . . . . . . . 1 . . . 8 . .. .. . . . 1 Motionsensors . . . . . 7 . . 7 . . .. . . . . 7 . 7 . . . 6 . . 9 . . 7 . . . . . . . . . . . 2 Junctionvariables . . . . . 5 7 . . . 1 0 7 . . . . .. . . . . . . . . 1 . .. . . . . Flexible structures . . . . . . . . 1 Electromechanicalanalogies . . . . . . . . . . . . . . .. . . . . . . . . .. 1 .. 6 . 1 1 7 . . . .. 7 . 3 Parallelfeedback . . 6 . . . . . . . . . . . . . . . . . . . .. . . . . . . . . 7 . . . 2 .. . .. 1 0 . . . . . . . . . . .. 7 . .. .. 7 . . 7 . . . . . . . 1 . . . . .. . . . 2 DC motors. . .. . . . . . 1 Feedbacktoparallelchannel analogy . . . . 7 .. . 3 7 . . . . .. . . 7 . 2 . .. . . . . . . . . 205 205 208 209 211 211 212 213 214 215 215 216 217 217 218 219 220 220 223 224 224 225 227 227 230 230 230 232 232 233 233 233 234 235 235 236 237 237 237 238 240 7 . . . 3 Loadingdiagram . 9 . .. . . . . . 6 7 . . 7 .. . . . . . . . . 4 . .. . . 4 Noise responses. . . 4 Piezoelements . . . . .. . . 3 Motor output mobility . . . 5 Drivers. . . 7 . . . . . . . . . . . . . 7 . . 7 . 7 . . . Flowchart for the chain connection of bidirectional twoports . . . . 1 . . .. .. . 2 Feedbacktotwopoleconnection analogy . . . . 4 . . . . . 7 7 . . . . . . ... . . . . . 2 Blackman's formula .. . .. . . .. 1 Impedance (mobility) of a lossless system . . . . . . . . . . . . . . 7 . . . . . . . . . 1 Structural design . . . . . . . . . . . . . . . 2 7 .andgears . .. . . . . . . . . . .. . Effect of the plant and actuator impedances on the plant transfer function uncertainty . . . . . . . . 4 . 7 . . . . . . . . . . . . . . . . . .. . . .. . . . 7 . 9 . . . . . .
Subharmonics.. .15 . . NDCin multiloopsystems . . . . .13. . . .. . . . . .. . . . . . . .8 11. . .. . .. . . . . . Jumpresonance . .elementarytreatment . .4 12. . . .6 Process instability .. . . . . .2 13. . . . . . . . . . . . ... . .. . . . . . . . . . . . .. . . .. . . . . . . 331 333 335 336 339 340 343 343 347 Appendix1 Al. . .13. . . . . NDC with parallel channels .. . . . . . . . . . Multiwindow control . .9 11. . . . . . . . . . . . . . . . . . . . . . . . . Chapter 12 PROCESS INSTABILITY. . . . . . Two nonlinear linksin the feedback loop . . . . . . . . . . . . . . 349 Feedbackcontrol. . . . . . . . . .. . Acquisition Timeoptimal control . . . . . . . .. .1 Odd subharmonics . . . . . . . . .. . . . . . .. . . . . . . . . . . . . . . .. . . . . . 11. . . Absolutestability of theoutputprocess . . . . . . .. . . . .. . . 11. . . . .. . . . . . . . . . . .. .. .. . . Problems.7 13. . . . . . . . . . . . . . .4. . . .. .4 11. . .1 Harmonics . . . . .9 Composite nonlinear controllers . Nonlinear links yielding phase advance for largeamplitude signals . 322 322 322 324 327 327 328 329 329 Chapter 13 MULTIWINDOWCONTROLLERS. . . .. . . . . . . . . . . . .. .. .6 11. . . .2 11. .. . .6 13. . . . . . . . . . . . . 12. . . . . . . Verification .3 11. . . . . .. . . . . .. .X Contents Describing function . .3. . . .l APPENDICES . . . . .. . . . . . . .7 11.10 11. . . . . expressions . . . . 331 13. . . .. . . . . . .2 Intermodulation. . . . Harmonicsandintermodulation .13 11. . . . . . . . . . . .. . . . .1 13. . .. . .. . . . . Nonlinear interactionbetween the local and the common feedback loops . .3. . . . . . . .. . . . . 11.2 12. . . . . . . . . . . . . . . . Negative hysteresis and Clegg Integrator . . . . . . . . . ... . . . .. . . .. . . .. . . . . . . .. . . . . . . . . .. . . . Selection order . . . . . . . . . .. . . . . . . .1 Exact formulas .5 and 13. . . NDC with a single nonlinear nondynamic link . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Windup. . .. . . . .. .. 349 Introduction . . . . . . . . . . . .11 11. . . . . . .of global stability . . Problems . . . .2 Approximate Hysteresis . Switchingbetweenhotcontrollersandtoacoldcontroller .3 12. . . . . .. . . . . . . ..1 12. . .. . . 12. . . . . . . . . . . . . . . 349 . . . .5 12. . . . . . Problems .andantiwindupcontrollers . .. 11. . .. Examples . 11. . . . . . . . . . . .. .. . . .4 13. . . . subharmonic . . .14 11. .. . . . . . . . . . . ... . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. . . . . .. .12 11. . . . . . . Describing functions for symmetrical piecelinear characteristics . . . . . . . . . . .. . . . .. . . . .4. . . . . .. . . . . . . . . . . 291 292 292 296 297 300 301 302 304 306 310 311 312 313 313 314 315 31'7 ..8 13. . . .2 Second Nonlinear dynamic compensation . . . . . . . . . . .. . . . . NDC made with local feedback . . . .. . .. . . . . .. .. . . . . . 12.. . . . .. . . . . .. . .5 11.. .. ... . . . . . . . .3 13. . .
. .. . . . .. A11. . . . . . .6 TheNyquiststabilitycriterion . . . .. .. Appendix 11 Discussions . .... . Appendix10 Phasegain relation for describing functions . . . . . .. . . A11. A1. . . . xi 349 349 350 352 353 353 354 355 355 356 357 357 358 359 359 360 360 362 362 363 365 366 367 367 371 372 372 372 373 376 379 383 384 385 386 387 387 388 388 388 389 390 390 391 391 391 392 . . . .. . . . . . . . A1. . . . . . . .4 Laplace transfer function . . .4. . . .. . .. .. . . . . . . . A1. A11... . . . . . . . . A1. . . .10 The Bodephasegainrelation . . .. . . . . .. . . . . . . .. . . . . . . . . . . . .. ..5 Polesandzeros of transferfunctions . . . A11.3 Controlsystemdesignusingfrequencyresponses . .. .. . . . . .. . . . . . . .2 Complex transfer function . . Appendix . . A l l . . . A1. . . .. . ..4.. .9 Bode integrals . . . . Appendix 4 Derivation of Bodeintegrals .. . .. . . . Appendix 6 Genericsingleloopfeedbacksystem . . . . . . . . .2. . .4 More about feedback . . . . . . . and positive real functions . . . . A1. . . .3 Why controlcannot beperfect . . . . A2. . .8 Problems . . . . . A1. . . . . . . . .. . . . . . . . . . . . . .. . . . . . . . A42 Integral of theimaginarypart . . .6 Conclusion . . . . . A2. . A1. . . . .. .1 Dynamic links . . . . . . . . . .. . A4. Appendix 5 Program forphasecalculation .. . . . . . ..2 Feedback:positiveandnegative .3 General relation .. . .. . . . . A1.. . . . . .. . .elementarytreatment . . . . .. . .. . ... . .. . . . . . . . . . .5 New words .7 Actuator’soutputimpedance . . . . . A11. A1. .. . l l Whatlimitsthefeedback? . . . . A2. . . . . . . . . A2. . . . .2. .6 Polezerocancellation. . . . . . . . .. .. . . . . . . . . . .3. . . . . .. ..4. . . . passive systems. . . A1.4 Some algebra .3 Links . . ... . . . . . . . A4.. . . . . . . . .3 Laplace transform and the $. . . . Appendix 2 Frequency responses . .. . . . . . . . Appendix 7 Effect of feedbackonmobility . .2 Loop frequency response . .. . .. . Appendix 3 Causal systems. .. . . . . . . . . . . . A11. .. . . . .. . . . . . . . . . . . . . . .. . . . . . . . ..9 Balancedbridgefeedback .. . . .. . . . .1 Feedback block diagram .. . A2. . . . . . . . . l Compensator implementation . .... . . . . .. .. . . . . . .. . . .4 Elements (links) of the feedback system . . . . . . ..5 Plant transfer function uncertainty . .. . . . . . . . . . . .. . . . . . . . A11. . . . .2. A11. . . .. . . . . . . . .2 Control accuracy limitations . . .. ... .1 Frequency responses . . .. A l l . . . . . . . . . . . . . . . . . . . . .2 Feedbackcontrol. . .. . . . . A1.4.. . .. . A2. . . . . . . . . . . .. A2.4. .... .8 Integral of feedback. . . . . . . . . .. . .1 Integral of therealpart . . . . . . .. . . ... . . . . .Contents A1. Appendix 8 Dependence of afunction on aparameter . . . . A1. . . . . .7 Timeresponses . . . . . . . .. . .2 Feedback control . . . . .. . . . . . . . . . . . . . . . . . . . . . .4. . . . . . . . . . .. . . . . .1 Selfoscillation. . .. . . . . . .plane ..dominantpoles andzeros .3. . . . . . . A2.. . . . . . . A1. . . . . . . . . . . ... . . . . . . . . ..3 Trackingsystems . . . . .. . . . . . . . . .. . A1l.5 Disturbance rejection .
. . . . . . . . . . . . . .17 A11. . . . . . and local loops . . . . . . . . . . . Describingfunction andnonlineardynamiccompensation . . . 13 A11. . . . . . . . . . . .. . . . Scanning mirror of a mapping spectrometer . . . . . . . Voltage regulator with a main. . . . . Attitudecontrol of solarpanels . . . . . . . . . . . 445 449 .12 A13. . . .10 A13. Feedbackcontroldesignprocedure . . . . . . . Examples. . .1 A13. . . . . . . . . . . . . . . . . . . . . . .15 Appendix 14 Contents Feedback maximization . . . . . . . . . . .14 A13. . . . . . . . . Appendix 13 A13. . . . . . . Telecommunication repeater . . . . . . . . . 392 394 394 394 395 395 396 396 397 398 399 399 400 402 403 404 405 407 409 410 411 411 412 417 430 431 432 441 BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . .. . .. . . . . . . . . . . . Nonminimum phase functions . . PLL computer clock with duty cycle adjustments . . . . . Telecommunicationrepeaterwith anNDC . . . . . . . .8 A13. . . . Saturn V SICflightcontrolsystem . . . . . . . . . . Industrialfurnacetemperaturecontrol . . .. . . . . . vernier. . . . . . . .12 A l l . . . .xi i A11. . . . . . . . . . . . . . . . . .16 A11. . . . . . . . . . . Design sequence . .18 A11. . . . . . . . . .4 A13. . . . . . . Bode Step toolbox. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Globalstability and absolutestability . .9 A13. . MIMO motor control having loop responses with Bode steps . . . Distributed regulators . . . . . .7 A13. . . . . . . . . . .15 A11. .6 A13. . . . .3 A13. . . . . . . . . . . . . The Bode’s book . . . . . . . . . . . . .. . . . . . . .13 A13. . . Multiloop systems .5 A13. . . . . . . . .. . . .. . . . . . . . Rocketboosternutationcontrol . . . . . Feedback maximization in multiloop systems . . . . . . . Conceptual design of an antenna attitude control . . Mechanicalsnakecontrol . . . . . NOTATION . .19 AlL20 Appendix 12. .. .2 A13. . . MIMO systems . . . . . .14 A11. . . . Pathlength control of an optical delay line . . . . . . . INDEX . . .11 A13. . . . . . . . . . . . . . . . . . . . . . . . . Attitude control of a flexible plant . . . . . . . . . .
This architecture reflects the multifaceted characterof real life design. various simulation methods (Chapter 7) followed by a brief surveyof alternative controller design methods and of adaptive systems (Chapters 8 and 9). Each consecutive layer is based on the preceding layers. Book architecture.The book presents the design techniques which the authors found most useful in designing control system for industry. utilization of the effects of feedback on impedances. a short description of the effectsof nonlinearities.theobject of control(plant) is characterized either by its measured frequency response over the range of effective! feedback orby a rather simple input/output mathematical model. modeling and simulation. These are the reasons this book starts with feedback. 2 Control system design: feedback theory and design of linear singleloop systems developed in depth (Chapters 34) followed by implementation methods (Chapter 56). Classical design does not utilize the plant’s internal variables and/or their estimates for compensation. and compensator design. and composite nonlinear controllers (Chapter 13). control plants with structural modes. four design level layers. and not with extensive plant modeling. All these controllers are highorder and nonlinear. 3 Integration of linearandnonlinearsubsystemmodelsintothesystemmodel. The appropriate loop responses are achieved by a standalone highorder compensator. Each layer considers linear and nonlinear systems. and allows illustration of the theory by real system examples without excessive idealization.2. and include a highorder linear part with a Bode step. 4 Nonlinearsystemsstudy with practicaldesignmethods(Chapters 10 and 11)$ methods of eliminationorreduction of processinstability(Chapter 12). For example. Treatment of the effects of links’ input and output impedances on the plant uncertainty in the third layeris based on the elementary feedback theory of the first layer and the effects of plant tolerances on the available feedback developed in the second layer. Design examples. and elementary simulation methods (Chapters12). The layers are the following: 1 Control system analysis: elementary linear feedback theory. roughly speaking. global stability and absolute stability are then treated more preciselyin the fourth layer. unlike the fullstate feedback approach.3. It also prepares the reader for research in the area of highperformance nonlinear controllers. The followhg are examples of highperformance controllers. introduction of absolute stability and Nyquist stable systems in the second layer is preceded by a primitive treatment of saturation effects in the first layer. This completes thefirst onesemester course in control. xiii . 0 A prototype of the controller for a retroreflector carriage and several other controllers of the Chemistry Spacecraft is described in Section 4. disturbance rejection. These controllers are chosen among those designed by the authorsof the book at the Jet Propulsion Laboratoryfor various space robotic missions. loop shaping. Phnt and compensator. Inclassicalcontrol. The material contained in this book is organized as a sequence of. feedback. telecommunication. and spaceprogram.
briefly described Example in 1. invlaplace. described detail in in Example2. Mars Global Surveyor attitude control.10.) Somare. the prerequisite to the control coyrse.13.250). Section A13.5. Section 13. More design examples are described in Appendix 13. Seqtion 5.The splane isused only for proving several theorems essential for frequency response methods. If needed. step. Section A13.6. Section 13. and only a small subset of MATLAB commands is used. Switchedcapacitorcontrollerforthe STRV spacecraftcryogeniccoolervibration rejection. poly. EE students know the frequency responses from the Signals and Systems course.y). tfazp.2. ’ (The book’s cover depicts Chemistry. set. linspace. grid. These examples can be bypassed in teaching other specialties. plot(x. conv. lp21p.1. students should be taught how to plot Bode diagrams from the block diagramspp.6.2. No preliminaryknowledge of MATLAB is assumed. highvoltage power source for Deep Space Network klystron radar. and in the feedback system design as described in this book. Inc. bilinear. impulse. Section A13. and DS1 spacecraft. (see 228. title. format. Cassini mapping spectrometer controller.9.No previous experiencewith practical applications of the Laplace transform is required. listed below in the order of theirintroduction: logspace.Example 2 in Section 7.xiv To Instructors A nonlinear digital controller for the Mars Pathfinderhighgain antenna pointing to Earth.6. Most simulation and design examples in the book use MATLAB 8 from MathWorks. Section6. but the introductory course canbegivenusingMATLABalone. EE majorsshouldalsobeshownhowtomakesimulationsinSPICE(SPICE examples givenin this book are listed in the Index). hold off.4. Some examples can use SIMULINK 8. . frequency responses for these specialties can be taught using Appendix2. If SIMULINK is used as a major design tool.10. rlocus. bode. either before or in parallel with Chapter 3. authors and described in Appendix 14 may complement the design methods described in the book. Section 13. Frequencyresponses. .2 and in Example3.Section 11. inv. lp2bp.1 and Example in 2 Section 7. Appendix 2 contains anumber of problems on the Laplace transform and frequency responses. Vibrationdamping inthemodelof aspacestellarinterferometer.andSpaceInterferometer optical delayline control. ezplot. roots. zp2tf. chemical. Cassini. Mechanical. the transforms are performed numerically with MATLAB commands step and impulse. Some simple C code examples are in given Chapter 5 . Section A13.SectionA13. Additional MATLAB functions written by the laplace. and aerospace engineering majors know frequency responses from the courses on dynamic responses.7.11.afeedbackfeedforwarddigitalattitudecontrollerforDS1 spacecraft solar panels.12. Cassini Narrow View Camera thermal control. Cassini computer clock PLL. described in detail in Section 7. in particular.1. gte linmod. inv. residue. Thedesignmethodstaughtinthisbook are based on frequency responses. hold on. Microgravity accelerometer analog feedback loop. Microwave Limb Sounder antenna attitude control of theChemistryspacecraft.
advantages and limitations of adaptive control. Secondonesemestercourse. so there is no need for conversion formulas for highorder functions. The design of highorder nonlinear controllers is covered in Chapters 1013. Design problems with mechanical plants are suitable for both ME and . Chapter 8 gives a short introduction to quantitative feedback theory and H. Digital controllers. Therefore. thus. Analog controllers are easier and cheaper to design. SIMULINK. but are far from being finalized.To Instructors xv Undergraduate course. and manufacture than digital in many applications. the output signal of a sensor is typically electrical. the engineer should beawareofthemajor concepts. when the need for practical design arises. and C make digital controller design simple. nottowastetimeontryingtoachievethe impossible. Undergraduate students need not be taught adaptive systems design since practical control systems rarely need to be designed as adaptive. troubleshoot. control and the timedomain control based on state variables. Nonetheless. and the compensators are therefore almost always electrical.atthesametime.and. These design methods have been proven very effective in practice. and sensors. and adaptive digital control. He must be able to recognize the need for suchcontrol. Further research needs to be done to advance these methods. The best way of designing a good discrete controller is to design a highorder continuous controller. the signals are analog. The digital control design can serve as a prerequisite for the following special courses on DSP. Chapter 7 describes structural designand simulation systems with drivers.and. break it properly into several links. especially when the 'course is taught to mechanicaVaerospacekhemica1 engineering majors when extra time is needed to teach frequencyresponses. The material from Appendix can 1 be used for an introductory lecture. implement. which allowsimplementation of thesummerandcompensator as analogelectrical circuits in electrical andas well mechanical/hydraulic/thermal control systems. Therefore. so there is no need for prewarping. The links of digital controllers must be loworder. The first 6 chapters. But the need for some adaptive systems does exist. or to understand the language of the specialized literature. estimation.Thesections tobebypassed are listedintheabstractsatthe beginning of each chapter. for selfstudy and as a reference for engineers who took only the first course. it shows that tailoring the output impedance of the actuator is important to reduce the plant tolerances and to increase the feedback in the outer loop. The material in Chapter 9 will enable him either to figure out how to design an adaptive system himself. also include some material better suited for a graduate course. and can be used for selfstudy or as a reference later. Chapter 6 is important for the engineers of all specialties. MATLAB.The accuracy of the Tustin transform is adequate. Analog controller implementation. Chapters 713canbeused forthesecondonesemester course (it can be a graduate course). two small tablesof formulas or one MATLAB command is all that is need$d. In particular. Very often. This material should be omitted from a onesemester course. motors. and then convert each link to a digital form. which constitute the first course in control. Problems. the chapter need not be fully covered during a onesemester course.The input of a modern mechanical actuator is most frequently electrical.
Abookletwithsolutionstothe conveyinsightsnotpresented problems is available for instructors from the publisher. .Additionalproblemsfor EE majorscanbefoundin [9].xvi To Instructors EE majors. Someproblems in thetextexplicitly.
CLASSICAL FEED6ACK CC3NTROL .
This Page Intentionally Left Blank .
Ifthefeedbackpathwerenot the forward path withthetransferfunction present. positive. The capital letters stand for the signals' Laplace transforms and also for the transfer functions of the linear links. U1. actuator and plant make up CAP. commanded elevation I Measured elevation I feedsack path. 1. 1 . U2. which is the object of the control. The notion of frequency response.1 Singleloop feedback system The feedback pathcontains some sort of sensor for the output variable and has the transfer function B. (The Nyquist stability criterion is presented in Chapter 3 . which. equal the product The return signal which goes into the summer from the feedback path is TE. 1.1 Feedback control system It is best to begin with an example.and the error E = U1 .. which is the commanded elevation angle. in this case a motor regulator (driver) and a motor. and so the system is also referred to as singleloop. The error amplified by the compensator C is applied to the actuator A. the antenna itself. and the Nichols chart are introduced. There is one input command. Fig. equals the commanded value U1. elevation anale sensor " I C A P Fig. at the output of the summer is zero.Chapter I FEEDBACK ANI) SENSITIVITY I Chapter 1 introduces the basics offeedbackcontrol. is the actual elevation of the antenna. In practice. the measured output value SU. and just one output.BU.Thepurpose of feedback is to make the output insensitive to plant parameter variations and disturbances. The compensator. which can be used as an introduction to this chapter.e. i. ) Feedback control and block diagram algebra are explained at an elementary level in Appendix 1. the Nyquist diagram. Evidently there is one feedback loop. .l(b) shows a block diagram for this control system made of cascaded elements. Laplace transfer functions are described in Appendix 2 . Ideally. links. This block diagram shows a singkinput singleoutput (SISO) system.and the output U2 would simply CAPU. the system would be referred to as openloop. The motor rotates the plant P .andlargefeedbackaredefinedanddiscussedalongwithsensitivityand disturbance rejection. Fig. Negative.l(a) depicts a servomechanism regulating the elevation of an antenna. 1. most of the time the error is nonzero but small. 1.
It is seen thatwhen the feedback is largg the error is small. The outputof the summer is so that the error where I.the PLL synchronizesthe VCO withtheinputperiodic signal. The feedback makes the error small so that the sensor voltage approximates the input voltage. so that the output signal has only a small phase difference compared with the input signal and. Example 1. Large feedback makes the phase difference (phase error) small.2. 1. A phaselocked loop(PLL) is shown in Fig. Fig. Another identical potentiometer (angle sensor) placed on the shaft of the motor produces voltage V. proportional to the shaft rotation angle.3. 1. or simply servo. A servomechanism for steering a toy car (using wires) is shown in Fig.2 Joystick control of a steering mechanism The arrangement of a motor with an angle'sensor is often called servomotor. or the refurn ratio. . and therefore the motor shaft angle tracks the joystick commanded angle. The required high reliability was achieved by using four independent parallel analog electrical circuits. The plant here is a voltagecontrolledoscillator (VCO). The phase defector combines the functions of phase sensors and input summer: its output is proportional to thephasedifferencebetweentheinputsignalandtheoutput ofthe VCO.Inotherwords. The command voltage U1is regulated by a joystick potentiometer. The system of regulating aircraft control surfaces using joysticks and servos was termed "fly by wire" when it was first introduced to replace bulky mechanical gears and cables. = T + 1 is the refurn differenceand its magnitude I F 1 is the feedback. the same frequency. The telecommunication link between the control box and the servo can certainly also be wireless. Feedback and where the product ' I ' = CAPB is called the loop transfer function. therefore. 1. Similar servos are used for animation purposes in movie production.2 Chapter Sensitivity 1.. The VCO is an ac generator whose frequency is proportional to the voltage applied to its input. Example 2..
7.Chapter 1.0. . Example 1. Example 2. and the closedloop gain coefficient 100/0. for example) preferred by engineers. the return difference F is 0.003. the feedback is positive.3. and the closedloop gain coefficient 100/1.l(b) is U.produces a negativeincrement intheoutput level when thelevelis expressed in logarithmic values (in dB. whether the feedback is positive or negative depends on the sign of the trarnsferfunction about the loop.0. Hence. We will adhere to these definitionsof “negative” and “positive” feedback (and use these terms without quotation marks) since very important theoretical developments. i. This definition was developed in the 1920s and has to do with the fact that “negative” feedbackreduces the error IEl and the output lU21. The system is saidtohave “negative” feedback when I F 1 > 1(althoughthe expression IFI is certainly positive). The feedbackissaidtobe “Positive” feedbackincreases the error and the level of the output. 1.3 Phaselockedloop Analog and digital PLLs are widely used in telecommunications (for tuning receiversandforrecoveringthecomputerclockfrom a string of digital data). F U. The return ratio T is 0.003. the = 77 is less than the feedback is negative. Feedback and Sensitivity input periodic signal 3 e ’ Phase detector error + Compensator control Fig. It is seen thatwhen T is small. so the inputtooutput transfer function of the system with the feedback loop closed. Whether the feedback is positive or negative dependson the amplitude and phase not onlyonthesignatthefeedbacksummerasisstated of thereturnratio(and sometimes in elementary treatments of feedback). and from (1.7 = 143 is greater than the openloop gain coefficient. The forwardpathgainCoefficientis100and the feedbackpath coefficient is 0. The return ratio T is . Hence.e. EF It is clear that thefeedback reduces the inputoutput signal transmission by the factor 1g. Let’s consider several numerical examples. .3. whichmakes IEl > IU1 I. commonly referred is: to as the closedloop transfer function. “positive” if IFI e 1. for synchronizingseveralmotors’angularpositions and velocities. The forwardpathgaincoefficient CAP is100 andthefeedbackpath coefficient B is .3 openloop gain coefficient.3.2 = ECAP. U = ECAP CAP 2 : ” ”. 1..2 Feedback: positive and negative The output signal in Fig.2) U1 = EF. the r e t m difference F is 1. to be studiedin Chapters 3and 4. 1.andfor manyother purposes. are based on these definitions.
3 Large feedback Multiplying the numerator and denominator formula: of(1. An imprecise actuator may be much cheaper. The return difference is 99. When the feedback is hrge. the coefficient M = 1. The forward path gain coefficient is 1000 and B = 0. and a precise sensor may also be relatively inexpensive.1.2). If B = 1. When IT I > 2.4(a). F . Fig. as shown in. the smaller the error expressed by (1.e. a system In with large feedback. and the closedloop gain coefficient is 10. u 2 " 417 1 . This feature is of fundamental importance. The return ratio istherefore100. Example 3. i. Examples are: a telescope tracking a star or a planet. and the return ratio becomes 100. an antenna on the roof of a vehicle tracking the position of a knob rotated by the operator inside the vehicle. The larger the feedback. Le. when I T I i s large.Thereturndifferenceis101.3)by B yields another meaningful A=:"U 1 T 1 "M. 1. and reasonably inexpensive system. U1 BF B where T T M==. Usingfeedbackthecheaperactuator andthesensorcanbecombinedtoform a powerful. when IT1 >> 1. precise. the return difference F = T. the affect of these uncertainties on the closedloop characteristics is small.1. i.T+l Equation (1. then the closedloop transfer function is just M and U 2 = U1.thefeedbackisnegative.1 equals the command divided by B. According to (1A). the antenna elevation angle in Fig. and a cutting tool following a probe on a modeltobe copied. 1.9.e. and the output becomes B One result of large feedback is that the closedloop transfer function depends nearly exclusively on the feedback path which can usuallv be constructed o f precise comvonents. Feedback and the feedback is always negative. If the elevation angle is required to be q. the forward path transfer function is changed to 1OO0.theoutput U 2 follows(tracks)the commanded input U1. the feedback is still negative. Such tracking sysfems are widely used.4 Chapter Sensitivity 1. since the parameters of the actuator and the plant in the forward path typically have large uncertainties. Manufacturing an actuator that is suffkiently powerful and precise to handle the plantwithoutfeedbackcanbeprohibitivelyexpensiveorimpossible. 1.4) indicates that the closedloop transfer function is the inverse o f & feedback ppth transferJEcnctionmultiplied by the coe@cient M. Example 4. then the command should be Bq. In the previous example. and the closedloop gain coefficient is 9. then IT + 1 1> 1 and the feedback is negative.
this arrangement commonly is called a vo/fage follower.1.Chapter 1. and the control system is calledregulafot). lS(b). (b) voltage follower Example 1. Introducingpositivefeedback would do just theopposite . The potentiometerwiththevoltagedivisionratio B constitutes the feedback path. Suppose that T = 100. Considerthevoltageregulator shownin Fig. Feedback and Sensitivity error 5 I (a) Fig.5. then T would become 110. the return ratio is T = 5000. This would make M = 0. (b) block diagram The amplifierinputvoltageistheerror E = U1. The command is the 5 V input voltage (when the command is constant. 12V to 30 V would be appropriate. which is reflected in the output signal.5(a) with its block diagram shownin Fig. If P were to deviate from its nominal value by +lo%. Here. . it is commonly called areference. 1.998 V. so that M = T/(T + 1) = 0. The power supply voltage isVCC.itwould increasethe variations in the closedloop inputoutput transfer function. an increase of 0. the differential amplifier with transimpedance (ratio of output current Z to input voltage E) 10AN and high input and output impedances plays the dual role of compensator and actuator.4) is lox 5000/5001 = 9. 1. andthereturnratiois T = 1OBRL. If the amplifier gain coefficient is lo4. Since the output voltage nearly equals the input voltage.1%.4(b) shows an amplifier with unity feedback.5 Voltage regulator: (a) schematic diagram. Fig. The VCC should be higher than this value. Without the feedback.9901. introduction of negative feedback in this case reduces the output signal variations 100 times. Consequently. the output voltage according to (1. The error voltage is the difference between the input and output voltages. Therefore.991. 1. The plant is theloadresistor RL. Example 3. a Hence. as in this case. Fig.4 (a) Tracking system. Example 2 . the variation of the output signal would be 10%.TE. the error voltage constitutes only lo4 of the output voltage. . Assume that the load resistor is 1ksz and the potentiometer is set to B = 0.
With the feedback.622 rad or 1. In cases(c) and (d).4. the presence of fedbacksignal TE makes IEl > IUII.425.6".by the complex numberU1= IUllcosqpl+ jlUllsinql.d) negative feedback .Tdecreases by 10% and the output voltage is lox 4500/4501 9. Insensitivityof the output This example also illustrates another feature voltage to the loading indicates that the regulator output resistance is very low. 1. therefore.e.e.1 Gain and phase responses The sum (or the difference) of sinusoidal signals of the same frequency is a sinusoidal signal withthesamefrequency. The sum of these two signals is represented by the vector (complex number) 1.e.866 +j0.002%less. itisrepresented by thevector 6Lz/4. The feedbackreducestheoutput voltage variations10%/0.only0.6 Chapter Sensitivity I.Feedback and Whentheloadresistance is reduced by lo%. The sum of these two signals is u = IU Isin(ot + cp> = (IU1Icosql+1~21cosq2)sinot + (IUllsinql+ 1~21sinq2)cosot . and the feedback at this frequency is negative.6 Examples of phasor diagrams for (a.002% = 5000 times.i.5.4(b). i.6 shows four possible phasor diagrams Ul= E + TE of the signals at the feedback summer at some frequency.925 = 1. In cases (a) and (b). IEl c IU1l.4 Loop gain and phase frequency responses 1. Example 1. i. by the complex numberU2 = I U2lcosq2 + jlU2lsinq2. I F 1 < 1 andthefeedbackat this frequency is positive. U = U1+ U2. If u2 = 6sin(ot + 7c/4). or 0. Thus. 1.) Theeffects of feedback on impedance willbe studied in detail in Chapter 7. 1. Signal 2. or 0. i. Fig.29 +j0. Example 2. it isrepresented by thevector 4 L d 6 .42 = IU2lsin(ot + R) is represented by the vector U2 = IU2ILq2.e. the vector for the sum of the signals equals the sum of the vectors for the signals.425 +j0. ReU = ReUl + ReUz and ImU= ImUl+ ImU2. The feedback dramatically alters the output impedance from very high to very low. of feedback.99778 i. Signal u1= IUllsin(ot + ql) is represented by the vector U1= IU1I&. If u1 = 4sin(u>t+ d6).e.59L 35.59L 0. withoutthefeedbacktheoutput voltage willbe 10%less. Fig.Thesummationissimplified when thesinusoidal signals are represented by vectors on a complex plane. 1. The modulus of a vector equals the signal amplitude and the phase of the vector equals the phase shift of the signal.b) positive and (c. (The same is true for thefollower showninFig.
Therefore. (s + 5). commands with substantial highfrequency content. MATLAB function conv can be used to multiply the polynomials s. Plots of the gain (and sometimes the phase) of the loop transfer function with logarithmic frequency scale are often called Bode diagrams. Feedback and 7 Replacing the Laplace variable s by j o in a Laplace transfer function.e. at 9/(2n) = 1. did develop an improved methodology of using them for feedback system design. although he. i.andtheslope of the gain response gets even Fig. by the following script: Frequency (radlsec) w = logspace(1.. 1. (18 0 h )arg'T(ja). Le.. 0 % log scale of angular % frequency w 10 ' nun = 5000.270°. where o is understood to be the frequency in rad/sec. and the loop phase shift in degrees. 2). 20 loglT(jo)l. to be explained in detail in Chapters 3 and 4.in honor of Hendrik W. 3 to 100radhec plotted over the 0 rangewiththesoftwarepackage MATLAB 0 3 from Mathworks 10" 1oo 10' Inc. the return ratio and the return difference are functions of frequency. Thefrequencyresponsesfor the loop gain in dB.7.4 Hz. The phase shift gradually changes from90" toward . The angle of Turn) is the loop phase shift which is usually expressed in degrees. 10" 1oo 10' 10 ' Theloopgainrapidlydecreases Frequency (radlsec) withfrequency. who. The loop gain is0 dB at 9 rad/sec.did not invent the diagrams. Let num T(s) == 5000 den s ( s + 5)(s + 50)  5000 s3 + 55s2 +250s +0 where num.as the frequency response. den are the numerator and denominator polynomials. bode(num. To accurately track rapidly varying commands. Plots of thegainandphaseof a transfer function vs. The transfer functionis the ratio of the signals at the link's output and input. results in the frequencydependent transfer function. the phase lag increases from 90"to 270". The loop frequency response is defined by the complex function T(jo). can be . Example 3. Bode..7 MATLAB plots of gain and phase steeper at higher frequencies. Le. one has to make the feedback large overa suffkiently wide frequency range. The plots can be drawn using angularfrequency o or f= a/(2n) in Hz. frequency are referred to generically. den. w) 8 90 8 180 2 P a 270 The plots are showninFig. this loop responses is typical for practical control systems. 1. den = [I 55 250 01. andgenerally depends on the signal frequency. The magnitude IT(ja)l expressedin decibels (dB) is referred to as the loop gain. and .Chapter Sensitivity 1.
and help conv in the MATLAB working window.b).8 by the solid line. The gain is 0 dB (Le. The loop gain decreases with increasing frequency. resultsinanoscillatoryclosedlooptransient response. feedback. Example 4.8 Chapter 1 Sensitivity Feedback and a (s + 50) in the denominator: a = [l 01. The diagram crosses the0 dB crossover frequencyfb where. ITI reduces with frequency.2 L 70°. ab = conv(a. 1. IRC1 IR=1 negative feedback. explicitly expressed as a function of jcu.3 functions used above can be obtained by typing help bode. and from the MATLAB manual. T = 10L 1 10"and F = 9L115". den= conv(ab. Conversions from one to another form of a rational function can be also done using MATLAB functions t f2 zp (transfer function to zeropole form) and zp2t f (zeropole form to transfer function). IR>>l 4 ' positive negligible feedback. The hump near the crossover frequency is a result of the positive feedback. The return ratio from Example1. the gain coefficient is1) over the.8 Typical frequency responses for 7. by definition. is T( jo) = 5000 j c u ( jco + 5)( jco + 50) At frequency cu = 2. . so the feedback is positive. b = [l 51.e. IH>l Fig.entire bandwidth of large feedback. The inputoutput closedloop system response is shown by the dotted line. This hump. and M The frequency response of the feedback I F 1 is shown by the dashed line. c = [l 501. F. T = 1L 160" and F = 0. and then becomes negkible at higkr fisquencies where F+ 1 and 20 loglF I 3 0. At frequency o = 9. 1. It can be seen that & feedback is negative (i. and should therefore be bounded. becomes Dosithe in the neighborhood OfJb. as willbedemonstratedinChapter 3. help logspace. The Bode plot of l"11 for a typical tracking system is shownin Fig. so the feedback is negative. dB 0 large feedback.. 20 loglF I > 0 ) s to a certain freauency..c) More information about the MATLA3. IT( f b ) l = 1 line at the Gain.
It can be seen that I F 1 becomes less.[270 240 210 180 150 120 901) grid It is recommended for the reader run to this program for modified transfer functions ..2 Nyquist diagram To visualize the transition from negative to positive feedback. Iml %plane 1 ' 7  ReT  0 Fig. it is helpful to look at the plot of T on the Tplane as the frequency varies from0 to .9. lO(a). 20*loglO(mag).andthegain of T. which means thefeedback is vositive there. Fig. Nyquist diagrams are commonly plotted on the logarithmic Lplane withrectangularcoordinateaxesforthephase.10(b) shows Lplane the Nyquist diagram for T = (20s +10)/(s4 + 1Os3 +20s2 + s) charted with MA'IXAB script num = 120 101. as shown in Fig. Here. 0 dB) of the Lplane. The Nyquist diagram should avoid this point by a certain margin. Notice that the critical point 1 of the Tplane maps to point (1 SO".Chapter Sensitivity 1. we use the diagram only to show the locations of the frequency bandsof negative and positive feedback in typical control systems. 1.'XTick'. den) . the distance to the diagram from the origin is distance from the1 point is IFI. ImT) or polar coordinates (IT I and arg T ) can be used. Example 1.than 1 at higkr jiiquencies.yystem's accuracyfor commands whose dominant Fourier comvonents belong to the area of negative feedback. but denrades the system's accuracyfor commands whose freauency content is in the area of positive feedback. The Nvquist diagram should not pass excessively close to the critical point1 or else the closedloop gain at this frequency be will unacceptably large.9 Nyquist diagram with (a) Cartesian and (b) polar coordinates. ITI.4. 1. 1. phase]= bode(num. 'den = [l 10 20 1 01. / r r . In practice. Either Cartesian (ReT. Feedback and 9 In general.feedback imvroves the tracking_. [mag. 1. This plot is referred to as the Nyquist diagram and is shown in Fig. plot(phase. the feedback is negative at frequencies up to fl TheNyquistdiagramis a majortoolinfeedbacksystemdesignandwillbe discussed in detail in Chapter3. 'WO') title('Lplane Nyquist diagram') set(gca. 1. andthe At each frequency.180. 0.
10 Nyquist diagrams on the Lplane. the curves indicate the tracking system20 gain log IMI .4. Feedback and and to observe the effects of the polynomial coefficient variations on the shape of the Nyquist diagram. When the Nyquist diagram for TI' is drawn on this template.11 Nicholschart . dB 15 10 5 0 5 1 0 1 5 20 Oo loo 20° 30° 40° 50° 60° TOo 80° deviation in phase from180° 90° Fig.3 Nichols chart The EIiChds chart is an Lplane template for the mapping fromTI' to M. 1.according to (lS). 1. (a) typical for a welldesigned system and (b) MATLAB generated for Example2 (a) 1. and is shown in Fig.10 Chapter Sensitivity 1. degr 50 100 270240210180150120 90 (b) Fig. 1. Lplane Nyquist diagram 100 Lplane 50 phase.1 1.
theclosedloopgainis 1.12 Disturbance sources in a feedback system The frequency response of the effectof a disturbance at the system's output can be calculated in thesame way thattheoutputfrequencyresponsetoa command is calculated:itistheopenloopeffect (DlAP. The loop gain is 15 dB. Feedback and 11 It isseenthattheclosertheNyquistdiagramapproachesthecriticalpoint of the closedloop (1 80°. 0dB).5 dB.the combined effect at the output of several different input signals is the sum of the effects of each separate signal. the closedloop gain is 6 dB.AP+ D. the Nyquist diagram should not penetrate into the area bounded by& line marked "6dB". The feedback is positive.4 = 13.e. by theFouriertransform of thistimehistorywhichgivesthe disturbance spectral denshy. is From the Nichols chart.8.. The limiting case has IMI approaching infinity.threedisturbancesources are shown. not to exceed 6 dB. for example) divided by the return difference F.5 Disturbance rejection Disturbances are signals which enter the feedback system at the inputor output of the plant or actuator.Sinceinlinearsystems. The disturbancescanbecharacterizedeither by theirtimehistory.6 = .disturbancesmightbeduetowind. InFig.4dB. as showninFig. 1. C A output I 6 I Fig. the loop phase shift is 170". indicating that the system goes unstable. Thefeedbackis15 . and therefore. The feedback is 10 . In theantennapointingcontrolsystem.P+ D.The feedback is negative.Chapter Sensitivity 1. the closedloop gain is7 dB. The loopgain is 10dB. gravity.6 dB. the higher is the peak frequency response in Fig.thegearing. Example 2. 1.1. andimperfections in themotor. The feedback is positive. ' F .12. the loop phase shift is 150".(7) = 3 dB. the loop phase shift 150". IMI is allowed to increase not more than two times. The loop gain is 1 dB.or.andthe driver. From the Nichols chart. the disturbances produce the output effect D.1. Fromthe Nichols chart. Therefore.andcauseundesirablesignalatthesystem output. the largerlMl is.in frequencydomain. i.temperaturechanges. Typically.1. Example 3. Example 1. 1. Consider several examples which make use of the Nichols chart.12. The feedback is 1 .
Thefeedback in a temperaturecontrolloop of a chamberis 188. This error can be considered and large feedback reduces the error effectively. andat 0. (or from the power amplifier 1 . Detailed calculations for a similar example will be given in the next section. Feedback and Sensitivity The effectsof the disturbances on the outvut are reduced when the feedback is negative: and increased when thefeedback is ppsitive. aerodynamic surfaces to reduce the error.. Another popular type of a system without an explicit command is the active suspension. 6 Example of system analysis We proceed now with the analysis of the simplified antenna elevation control system shown in Fig. 1. the transfer function of the plant is P(s) = 1/(Js2) As shown in Fig. which uses motors or solenoids to attenuate the vibration propagating from the base to the payload. and the disturbance rejection is the only purpose of introducing feedback. sqme slowly varyingin time and others rapidly oscillating. the temperature within the chamber changes by 6". but drops with frequency (since motors cannot mow the huge antenna rapidly). the torque 2: applied to the antenna is thesum of the torque 2:w produced by the actuatorand the disturbance wind torque It is known that for large antennas. The winds contain various frequency components. The disturbancecomponents are attenuated by the feedback accordingly.l(b). Such systems are called ~o&&J systems.Thefeedback intheantennaattitudecontrolloopis 200 atverylow frequencies. Disturbance rejectionis the major purpose for using negative feedback in most control systems. A typical example isa homing missile whichis designed to followthe target. No explicit command is given to the missile.feedbackcausesthevehicle's as a disturbance.12 Chapter 1. Since the Laplacetransform of an integrator is I/s. Example 2. Example 1.With feedback. and 5 times for the 0. To fwther reduce the higher frequency disturbances.The. the actuator transfer function(the ratio of the output torque' to the input voltage) A is = 5000/(sc le))NmN.1 H z gust components. Gusty winds disturb the orientation of a radio telescope. Assume that the elevation angle sensor hnction ia 1 Wad. the temperature within the chamber changes by only 0. 200 times for the effect of steady wind.06'. the plant is a double integrator with gain coefficient 1N. the wind torque spectral dendty is approximately proportional to I a . when the temperature outside the chamber changes. the feedback path coefficient is B = 1. the missile receives only an emor signalwhich is thedeviation from thetarget. Without feedback.1 H z thefeedbackisonly 5 . Le. 1 13. the output variable is the elevation angle. an additional feedback loop might be introduced that will adjust the position not of the entire antenna dish but of some smaller mirror in the optical path &om the antenna dish to the receiver fiontend of the transmitter). Rather. There exist systems where thereno is command atall. and the antenna is a rigid body with the moment of inertia J = 5080 kgm? The plant's input variable is torque.
05)( s + 05) s(s 5) (this simple compensator makes the system work reasonably well.SOs2 + 275s + 0.2 0 03. den. with the feedback.13 Elevation control system block diagram The spectral density of the disturbance in the antenna elevation angle is therefore proportional to 1 " 1 1 ' (s+~.1 0.Chapter 1. 1 5000s2 ' T(s) = num 50( s + 0. The closedloop transfer function M(s) = T/F = num/(nurn+ den).e.. in relative units: before the feedback was introduced.1)(s + 2) Compensator C "+ A & 1  elevation angle Js2 1 4 Fig. . w) The compensator transfer function is C(S) Fig.Large feedback must introduced be these at frequencies to reject the disturbance. 1 ~ +o.7)canbe plotted using MATLAB with: w = logspace (2. Thefrequencyresponse of (1. % frequency range % 0. SO(s + 0.1)(s+2)s2 . den = [l 2. dashed line SO( s + 0.l) .01.zS2 ~ The spectral density plot of the disturbance intheelevationangleisshown in Fig. 1.25 = : den s3(s+5)(s+l0) s5 + m 4 +50s3 ' The return differenceis F(s) = T(s)+ 1 = (num + den)/den .14.14 Spectral density of the elevation angle disturbance. bode(num. normalized to 100 dB at o = 0. solid line. sdback in dB 20 ~?.05)(s + 05) .s4 + 2 . The spectral density is largeratlowerfrequencies. 1.01 to 10 rad/sec 100 ""I. The loop transfer function is = + T ( s ) = CAP = i. 1 \ 40 num = 1. Feedback and Sensitivity 13 1 (s + 0.1.05)(s + OS) xx5000 s( s + 5) S + 10 _. although not optimally).
. 1.3. The hump on the gain response does not exceed 6 dB. Feedback and The plots of the gain and phase for the loop transfer function T (ja). w) hold on bode(g. one at a time. w) bode(num. up to 0. The labels are placed with mouse and cursor. The disturbances are greatly reduced by the feedback. den.2 Hz.4. g = num *. w) hold off % equal length of the vectors % makes the addition allowable % for T % for F % for M The plots are shown in Fig. 1. % freq range 0. T and M overlap. The plot for disturbances in the system with feedback. which satisfies the design rule mentioned in Section 1.14 Chapter Sensitivity 1. The mean square of the output error is proportional to the integral of the squared spectraldensity with linearscales of theaxes.8). Le. 10” 10” 10’ Frequency (radsec) 1 20 . T and F overlap. The feedback is large at low frequencies and is negative up 0. and in MATLAB with: for M(jo)can be made w = logspace(1/11.15 (a) Loop frequency response for the elevation control system: at lower frequencies. . is obtained by subtracting the feedback response(in dB) fiom the disturbance spectral density response. (b) Lplane Nyquist diagram The closedloop gain response20 log IMI is nearly flat up to 1.8 radlsec. The gain is peaking at 0.4 rad/sec. for F( jo).5 0. the dashed line in Fig. using MATLAB command gtext ( label ) .14. 1. bode(num.251.8 radsec. at higher frequencies.15(a).7) by(1. More precise design methods will be studied in the following chapters.1 to 10 rad/sec den = [l 15 50 0 0 01. g.cn 3 150 $ 180 210 240 10” 1oo Frequency (radlsec) 10’ (a) Fig. num = [0 0 0 50 2’7. or directly by dividing (1.Theplotsrequiredtocalculatethe reduction in the mean square error can be generated with MATLAB and the areas under . den. den.
when using small actuators.16 Output response to 1 radian step command Example 2.15(b). The transient response of the closedloop system radian the to 1 step command’ (“increase instantly the elevation angle by radian”] 1 with: is found num = [O 0 0 50 27.4 12 9 ’ 90. The response is faster.3. this is a condition for the system to become unstable. then slightly undershoots. the Nyquist diagram shifts down. step(num. the inputtooutput relationship invariably saturates when the input amplitude is large enough. If the loop gain is increased by (approximately) 20 dB. g = num + den. Similarly. However.curve“loopgainincreased”. If the compensator gain coefficient in the system with the Nyquist diagram shown in Fig. andpowerhungry. by 14 dB. or the mean square error can be directly calculated using MATLAB functions. The Gplane NyquistdiagramisshowninFig.Chapter Sensitivity 1. 1. 0. den = [l 15 5 0 0 0 01. As we already mentioned in Section 1.heavy. and the closedloop gain at this frequency therefore becomes infinite. Using the Nyquist diagram for stability analysis willbe discussed in detail in Chapter 3. and by 42 degrees fromthe right.2 .15(a) is increased 5 times. This shape of the closedloop transient response is typical and. The response generated with den = 0 .7 Effect of feedback on the actuator nonlinearity Actuators can berelativelyexpensive. and the system becomes unstable. g) grid 1.bulky.Economy requires that the. acceptable.4. the return ratio becomes1 at a certain frequency. 1. 1.The effect of the Nyquist diagram passing closer to the critical point can be seen on the closedloop transient response. but it overshoots much more and is quite oscillatory.16.e. i. The output doesn’t rise instantly by 1radian as would be ideal: it risesby 1 radian in less than 2 seconds but then overshoots by 30%. the Nyquist diagram shifts up by 14 dB and the margin from below decreases from 20 dB to 6 dB. by 40 dB from above. Feedback and 15 the responses found graphically. if the loop gain is reduced by 40 dB.actuator be as small as possible.1. about 10 seconds. commonly.O Time (sew) Fig. at some frequency the return ratio becomes 1. Saturation limits the . the Nyquist diagram shifts up by 20 dB. 6 OA The stepresponse is showninFig.8 E 0 .1.amplitudes without distortion. we have to accept the fact that these actuatorswill not be able to reproduce signals of relatively large .6 1.16. and settles to 1 radian with reasonable accuracy in.251.2*den is shownFig. The controllers must be designed such that this effect will not cause a catastrophic failure (and such a danger is quite real).We will consider these issues in Chapter 3and later. Because the output power of any actuator is limited. Example 1.nominal case.5 0. in Chapters 913.Thediagramavoidsthe critical point by significant margins: by 20 dB from below.
.l . (The slope of the line is shallow since the feedback reduces the inputoutput differential gain. the invut@ut curve appears as hard saturation when thefeedback is largg The deadzone characteristic is shown in Fig. Consider one of these products having frequency nf and amplitude Uan. c k :f . In other words.18. . signal. Therefore. The closedloop amplitude characteristic shown by thedotted line is therefore a segment of a nearly straight line.eg *. The ratio of the amplitude of a nonlinear product to the amplitude of the fundamental is thenonlinear product coefficient.3.* input (.In many actuators.1.in Fig. *** input 0 . If the forwardpathisapproximatelylinear.17(a) by the dashed line.* * * "* 9 / (a) (b) Fig.) (This gain reduction will be studied in Chapter Large feedback about the actuator changes the shape of the inputoutput characteristic.brieflydescribed in Section 7.. the input signal amplitude with the feedback is larger. the outputof bnonlinear forward path consistsof a fundamental component with amplitude U2 and additional Fourier components called nonlinear producfs.) Therefore.thenonlinearproductcan be viewed as a disturbance source added to the output as shown in Fig. the saturation soft is as shown by the solid curve.** . consider the output signal distortions caused by a small deviation of the actuator from linearity. . like pushpull class B amplifiersorhydraulicspoolvalveamplifiers. the gain coefficient of the saturation link decreases.16 Chapter Sensitivity 1. w e e d b a c k reduces the relative width of the dead zone. This feature allows the achievement of high resolution and linearity in control systems which use actuators and drivers with rather large dead zones (such actuatorsand drivers may be less expensive or consume less power from the power supply line. Large feedback reduces thedifferentialinputoutputgaincoefficient and therefore makes theinputoutput characteristicshallower as shownbythe dotted 'curve. I ~~ . 1... ~f~~~~~~ & . at anyspecified amplitude of the output signal. ..(.". loosely speaking.17(b). in a system with soft saturation. . 11 . ( . 1.and particularly. Next.1. 1.. ) . Feedback and amplitude of the output. The ideal (hart$ saturation characteristic is shown .17 Inputoutput characteristic of the actuator with (a) soft saturation and (b) deadzone When the amplitude of the input increases. then the differential feedback may remain large and the closedloop differential transfer function can be quite close to l 3 . the part of the input signal that causes no response in the output due to the dead zone. In response to a sinusoidal input with frequencyf. Relative to this input amplitude.** . If the input signal level is such that the slope (differential gain) of the saturation curve is not yet too flattened out. or the motors with mechanical gears). decreases.the ratio of the output to the input decreases. or. for the output amplitude marked by the thin dashed line.
A third harmonic of the 67 MHz signal produces an undesirable nonlinear product with frequency 200MHz thatfalls within the bandof ahigherfrequencychannel. thesystemwithoutfeedback. the disturbance. allowed developmentof longhaul multichannel telecommunication systems.This enabled an increase in the numberof telephone channels over expensive telecommunication cables. the . these methods were applied to feedback control systems to maximize the accuracy and speed of operation. 1.8 Sensitivity Sensitivityfunctionsaregenerally usedtoquantifytheundesirableeffects of some paraineters' deviations from normal of a transfer function. is reduced by thevalue of thefeedback. . The feedbacksystem sensifivity isdefined . CAPB + 1 . . Example 1. Low nonlinear distortion in the telecommunication feedback amplifiers invented by Harold Black at the Bell Laboratories in the 1920s. Example 2. control engineers.coefficient becomes 0. U.Chapter 1. For.18 Equivalent representationof nonlinear distortions in the actuator Now. When feedback of 100 is introduced over the entire fiequency bandof interest.as theratio of these changes: The smaller thesensithity. From (1. . The feedback in the amplifier is very large at lower frequencies but drops to 5 at 2oOMHz. although numerically can happen to be equal to one. 1. Feedback and Sensitivity 17 Fig. of particular interest is the sensitivity of the closedloop transfer function to the plant parameter variations. Le. Therefore. Sensitivity is not a transfer function.comparetwocases: first.)/(Uz/Ul) intheclosedloopsystemtransfer infinitesimalrelativechange function UgUl. The works of Black and the Bell Laboratories' scientists Harry Nyquist and Hendrik Bode established the basis for the fkequency domaindesignof feedback systems and feedback maxiniization. The feedback reduces the amplitude of this product5 times. In the second case. An amplifier is used to amplify signals of several TV channels. Withoutfeedback.3). Later..thenonlineardistortioncoefficient inan audio amplifier is 5%.andsecond. the nonlinear vroduct coeficientfor a closedloop_Systemis reduced by& value sffeedback atthe freauencv of the product. CAP = U . the better. dP/P causes an An infinitesimalrelativechange intheplanttransferfunction d( Uz/U.05%. the amplitude of the nonlinearproduct at thesystem'soutput.the system with feedback' and with the input signal increased so that the output signal amplitude Uz is preserved.
The massof acart whosepositionorvelocitymustbecontrolled depends on what is placed in the cart. The moment of inertia of an antenna about one axis might depend on the angle of rotation about another axis.18 then Chapter Sensitivity 1.e. Example 4. the closedloop gain changes by 0.theclosedlooptransferfunction changes by 1%. the total closedloop response uncertainty is 0.2dB. When the feedback is 10 and the plant magnitude IPI changes by lo%. 1 (CAPB+l) Therefore.Whenthe actuator and compensator implementation accuracy (together) is 0. Since the sensitivity of a largefeedback closedloop response to the feedback path is nearly 1.2dB. as was establishedby Harold Black. the product of the Example 3. Due to the feedback. Feedback and S= P d ( U 2 / U 1 ) __ P(CAPB+l)[CA(CAPB+l)C2A2PB] CAP U2 I U ~ dP (CAPB + 112 _.032 dB. Example 1. . the closedloop uncertainty willbe 0.Horowitz) can be used: . When the plant P deviates from the nominal plant Po by hp which is more than 50% of Po. the closedloop transfer function changes by only 1%. and w e e d b a c k reduces smhll variations in the output variables.01 dB. If the feedback path uncertainty is.).For example. say. When the value of thiselementchanges by 20%.9 Effect of finite plant parameter variations Plant parameters often vary widely. similar formulas describe the effects of variations in the actuator and compensator 'transfer functions. a closedloop transfer function q. payloads in a temperaturecontrolled furnace can be quite different.. 3. Example 2.mressed either in percents or in logarithmic values. 0.etc. Whentheplanttransferfunctiondependsonsomeparameter q (temperature. the total uncertainty of the forward path gain is.thisdependencecan be characterized by . 1. The chain rule can then be used to find the sensitivity of sensitivities: SX Spq. for better accuracy the Horowifzsensifivify (after IsaacM. F (1. I F 1 times.5.3 dB. The feedback is 10.10) The sensitivitv is small when the feedback is largt?. theplantgainuncertaintyis 3 dB. toi. the plant's sensitivity to one of its elementsis 0.the sensitivity Spq = [dP/P]/[dq/q]. lesser accuracy&the compensator and actuator implementation is acceptable than the accuracy ofthe'feedbackp&. Since the actuatorand compensator transfer functions enter the equation (1. pressure. The mass of a rocket changes while the propellant is beingused up. 1 S=. Sensitivity analysis is a convenient tool for calculation of closedloop response error and provides sufficient accuracy when plant parameter variationsare small.3) in the same way as P. When the feedback is10 and the plant magnitude IPI changes by 3 dB.042 dB. The feedback is 100.powersupplyvoltage.
where the plant perturbed valueP = Po+ AP. Withtheinput U 1 keptthesameforbothcasesand Uz.5.the perturbed output U2 = Ub + AU2.10 Automatic signallevel control The blockdiagram of automatic volume control in an AM receiverisshown in Fig. Thecarrier x TI[(?'+ 1)B] = 0. I Lowpass filter. detector. and the amplitude of the carrier multiplied by B is then compared with the reference voltage. thereturnratio T = 50. It can be shown that the Horowitz sensitivity is the inverse of the feedback for the perturbed plant (see Problem 8): 1 SH = 1 CA(Po +AP)B+l =. The lowpass filter B with corner frequency 0. If. When the feedback is large. 1.12) Again it is seen that large negative feedback renders the closedloop transfer function insensitive to plant parameter variations.98 V which is very close amplitude at the detector equals reference to to the desired carrier amplitude referencell? = 1V. The signal that appears at the output of the detector is the sum of the audio component and the verylowfrequency component that is proportional to the carrier.5 Hz removes the audio signal. B le I Fig. for example.11) This is the ratio of finite relative changes. and the RF carrier amplitude at themodulatorinputis O. The goal for the control system is to maintain the carrier level constant at the AM detector. B = 0. M b 1000 AM detector 11111.thenominaloutput. C = 100. (1. The error signal processed by the compensator is applied to the second input of the RF signal. Example 1.5 V. Multiplier.OOlV. F (1. multiplier. .Chapter 1 Feedback Sensitivity and 19 (1.13) 1.19 Automatic carrier level control The antenna signal is amplified and applied to the multiplier M. changing its transfer function for the the error gets small and the output carrier level approximates referencell?. in spite of variations of the signal strength at the antenna. 1.The signal from the multiplier is further amplified 1000 times and applied to the AM.19.reference = 0.
design of feedback systems with maximum available accuracy.Ofor a singleintegrator 0.since.since the feedback is limited. 1.)].15). 1. 0 implementation of controllers with analog and digital technology. Still. .[i. approximately. 0 analysis of the effects of the link’s nonlinearities. has been already chosen (it must be. the following topics need to be mastered: 0 addition of prefilters and feedforward paths to improve the closedloop responses.) The coeffkient D is . as we will learn. we present below simple design rules for two most frequently used types of the compensators. We assume q.20 Chapter 1. the compensator becomes a lead. Lead coefficient k is adjusted for the loop gain to compensators are used in the Problems 4244. structural mode. and design of nonlinear controllers.forexample. Feedback and Sensitivity Since l W signals canvary over avery large range.3%. However. and the sensor imperfection and noise atq. (If Z = 0. I (integral).by some fundamental laws. The compensator parameters are finetuned either experimentally or using a mathematical plant model and plotting computer simulated openloopand closedloop frequency responses and the closedloop stepresponse. the gain in the feedback loop can change accordingly. The coefficient (not to confuse with the plant) P = l/Plant(%).. A P D compensator transfer function C(s) = P + Zls + Dqs/(s + q) is tuned by adjusting three real coefficients: P (proportional).The feedback must be laqge even when the RF signal is the smallest acceptable (when the signal level is only slightly larger than the level of the always present noise). 0 buildingmathematicalmodels of plantsandcontrolsystemstoevaluatesystem performance. on the other hand. (The controller for the system having the Bode and Nyquist diagrams .feedback vastly improves the system performance. typically. The pole q = 4%. The reader can probably even design some control systems and also demonstrate that . The loop phase shift atCQ. k = b/[% X PZant(~..e. the system must perform well when theloop gain increases by an additional60 to 80 dB due to the increase of the signalat the antenna. in a real antenna attitude control system there also exists a feedback loop stabilizing the motor rate which is “nested” within the main loop.The zero a = 0. 1. arg T(u)b)] must be kept between 120’ and 150’ degrees (as shown in Fig. in order to design systemswith performance close to the best possible..3P/% for a doubleintegrator plant. The coefficient I S 0. The examples in Appendix 13 should start to become interesting and comprehensible. The be 0 dB at q. and. analysisanddesign of multiloopcontrolsystems . to make this chapter a short selfcontained course on servo design. and the pole b * 3%. at least 10 times lower than the frequency of the plant must not be excessive). plant and.2Pq. A lead compensator C(s) = k(s + a)/(s + b) is often used when the plant transfer function is close to a double integrator.11 Lead and €‘ID compensators Compensator designwill be discussed in Chapters 36. and D (derivative). The compensator hardwareand software implementation is described in Chapters 5 and 6.12 Conclusion and a look ahead The materialpresentedinthischapterenablessomeanalysis of singlelooplinear control systems.
(c) angular velocity of a rotating machine element.5'.13 Problems Using proper names for the actuators.35. zeros.poles: 5. (i) 1 . zeros: 3.05.12 10" 2 10 10 3.01.16 Tis equal to (a) 0. do Tothis. and the angle sensor gain coefficientB = 0. the closedloop gain coefficient is (a) 100.72.9. (c) 0.gain coefficient becomes 6000? In an antenna elevation control system. and find the error E and the command Ut to make the outputU2 = 10.1.200. (d) 5.. 400.72.5L150°. the closedloop gain coefficient is (a) 100.andreturnratio? Is this the case of positive or negative feedback? The openloop gain coefficient is.) 1.Sensitivity Feedback and 1. yaw and roll of an airplane.5000. zeros: 1. poles: 4. 100.400. (f) pitch.. What must be the command for the elevation angle to be (a) 30.7. (d) 1.72. large or negligible. and poles: (a) k = 10. For T = 99 and B equal to (a) 0. 5. (b) 0. (f) 0.5'.01. 35. (c) 3000. (f) 30 mrad? Use the MATLAB commands conv and/or zp2 tf . (e) 2. (9) 10. conversions from numbers to dB and back need be performed fast. 5 ' .5'. (c) 1. draw block diagrams fo feedback systems controlling (a) temperature and (b) pressure in a chamber. 65. and sensors. (b) 200. 100. The openloop gain coefficient is 3000. poles: 6. (d) 0.1. Chapter 21 shown in Fig. the openloop . 300. 9. the antenna moment of inertia is 430kgm'. 1. 600.1. (d) 5.(e) 2. (e) 2.3. 9. zeros: 1. the following table must be memorized: decibel number 0 1 20n 3 206 1 1. (d) 0. .8. and conclude whether the feedback is positive or negative. Is thefeedbacklarge? Howmuchwilltheclosedloopgaincoefficientchange when. (b) '15 . the feedback.thethreephase10kWmotorwithgearratio1200:l is painted green. (d) luminescence of an illuminated surface. is large. poles: 4.72'.72. (b) 0.1 Vldegree.returndifference. (c) k = 13. (c) 3. Calculate F and M. because of changes in the plant. (e) 2. (e) frequency of an oscillator. 100. calculate the closedloop transfer functions. plants. (f) 3. 3. to calculate the coefficients of thenumeratorandthedenominatorpolynomialsforthefunctionshavingthe following coefficient k. zeros: 3. (d) k = 25.8. (b) k = 20.4 1.01. (h) 10.15 must include nonlinear elements to ensure the system stability when the actuator becomes overloaded. (b) 200.(c) 3000. Whatarethefeedback. When a control system is being designed.(e) 2.65..
(d). 1. zeros: 1. Findthedenominatorroots.3.poles: 1. plot the closedloop frequency responses for (a). (b) 100/(8+ 4 s + 100). 10.72(s + 7)(s + 20)/[(s+ lO)(s+ lOO)(s + lOOO)]. find 4 s ) and M(s). zeros: 1. .72. Describe the correlation between the step timeresponses and the shapes of the frequency responses. 90. poles: 6.1. zeros: 3. .26)/[(~ + +43)(~ + 8 5 ) (+ ~ 2 5 0 ) (+ ~ 2500)l.300. 80. Describe the correlation between the slope of the gain response at higher frequencies and the curvature of the timeresponse at small times. (e) (2. (d) 100/(2 + s + 100). (f). andplottheirfrequency responses usingMATLAB.2.400. Plot the step timeresponse. 165. (c) (1Os? + 1 Os +40)/(2s4 + 20s3+ 1008 + 2000). the shapes of the frequency responses. (f) (8+ 10s + 8)/(s4+ 2s3 c 128 + 150). 12.722 + 27. poles: 5. (d) (8 + 20s + 200)/(s4+ 5s3+ 5 0 8 + 300). (9) k = 1300. (c). (c) 100/(8+ 2 s + 100).72. 300.+30s + 40)/(2s4+ 10s3 + 1008 + 900). 150. 10 Use theMATLAB command bode to plot the frequency response for the firstorder. 14 When IT1 monotonically decreases .70. Describe the correlation between the slope of the gainfrequency response and the phase shift. 35. (f) . 4.2 5 ( ~ + 2 ) (+ ~44)/[(~ + 55)(~ + 6 6 ) (+ ~7 7 ) ( + ~8800)l. 13 ForExample 2 in Section 1. (d) 2O(s + 2 ) ( ~ . (c) 1O(s + 2)(s+ 22)/[(s + 40)(s + 65)(s+ 15O)l. and thirdorder functions (a) 1O/(s + 10). 7 2+ ~5 ~0 8 + 272). secondorder. 100. 12 Use MATLAB toconvertthefunctiontoaratio of polynomialsandplotthe frequency response for the function (a) 50(s + 3)(s+ 12)/[(s+ 30)(s + 55)( s + 1OO)(s + 1OOO)].Describethecorrelationbetweenthedenominator polynomial roots. (b) 1OO/(s + 1O)2. (b) (8+ 3s + 4)/(s4 + 2s3 + 202 + 300). what happens to F and M ? 15 With the return ratio equalthe to function in Problem 12.2s + 200)/(s4 + 2 . poles: 4. . 150.in relation to frequency (as in most feedback control systems). (c) 1O O O / ( ~ + 1013. 500. (f) k = 20.22 Chapter 1.500. and the stepresponses. 11 Use MATLAB to plot the frequency response and step timeresponse for the firstand secondorder functions (a) 1O/(s + 10). Feedback and Sensitivity (e) k = 2. (h) k = 150. zeros: 3. and €3 = 1. 100. 50. (b) 60(s + 3)(s + 16)/[(s+ 33)(s + 75)(s+ 200)(s+ 2000)l. 0 Use theMATLAB command root andlor t f2 zp to calculate the poles and zeros of the function (a) (208. (b). (e). (e) 2.
Plant gain is uncertain within uncertainty in the closedloop gain? amplifier. (b) 200. The closedloop gain is 20 dB.1 dB.andpositive (regenerative)feedbackwasusedtoincreasetheamplifiergain.5. (f). The plant's sensitivity to the temperature is 0. (c) 300. Sensitivity Feedback and 23 16 With the return ratio equal to the function in Problem 12. ' 28 The feedback is (a) 10. What is the 0. (b) 200 mrad. What is the change in the closedloop gain? 27 The feedback is (a) 200. (c) (108 + 1Os +40)/(2s4+ 2s3 + 8 + 3). by how much will the output change? 29 Tubeswereexpensive in theearlydaysoftuberadioreceivers. (b) 100dB. What is the change in the closedloop gain? 24 The feedback is 80dB. (b) 100. '(c) 120 dB. The forward path consists of resonance contour tuned at the signal frequency 100kHz and an amplifier. plot the closedloop frequency responses for (a). (d). (b) 2000.amplifier with harmonic coefficient 5%.Plant gain is uncertain withinf15%. (c). What will these 26 The openloop gain coefficient is (a) 1000. (b)100. The closedloop gain coefficientis 20. Whatis this parameter after the feedback of 30 was introduced? 23 The openloop gain is (a) 80 dB. if the feedback at the frequency of the fundamental is (a) 100.000. 21 Before introduction of feedback. (d) (8 + 20s + 5)/(s4 + 5s3+ 8 + 3 ) .Chapter 1. What istheuncertainty in theclosedloopgaincoefficient? . and the return ratio is inversely proportional to the frequency.Thepositive feedback also improved the selectivity of the regenerative receiver (but narrowed the bandwidthof the received signal). B = 10. When the temperature changes by .02%? in afeedbackamplifierwithanopenloop 20 Findthethirdharmoniccoefficient harmonic coefficientof 5%. Illustrate the above using the following example. plotthe 17 Consideringfunctions in Problem12asthereturnratio. 19 What feedback needs to be introduced in an. the actuator dead zone was 5 (a) N. The forward path transfer function is 10000/(8+ 12. (c). where the angular . for the resulting harmonic coefficient to be 0.5. (c)l. the openloop gain is reduced by5%. the maximum actuator output signal was 100 m/sec.6 degrees. (d) 0. (b) (8 + 30s + 4)/(s + 2s3+ 2 8 + 3). 18 Plot Ny uist diagrams on the Lplane for the functions (a) (20j + 30s + 40)/(2s4 + s3+ 8 + 3). the openloop gain reduced by 1: dB. what is the dead zone after the feedback of 50 was introduced? 22 Before introduction of feedback.and closedloop frequency responses and the return ratio responses for (b). (c) 50. (d) 400. Because of plant parameter variations. (f). (d) 0. (e). differential gain variations constitute 25 In an variations be when 40 dB of feedback is introduced? f1. (a). (e).5s + 628*). Because of plant parameter variations.(c) 1.5dB. (e) 272. (d). (b).1 dB/degree. and B = 4.
will happen when the amplifier gain coefficient increases by 2% but the feedback path is not adjusted? 2. and negligible. find the change in thevalueoftheoutputwhentheplant changes from nominal to perturbed. the perturbed value is 50. Give numerical examples.01. find the closedloop gain frequency response of a tracking system (assuming B = 1) and plot this frequency response with logarithmic frequency scale. positive. Using Horowitz sensitivity. C. 1. done in the feedback path to keep the closedloop gain constant? What. 33 The perturbed value of the plant is 30. By how much does the feedback increase the receiver gain? What is the sensitivity at the frequency of the resonance? How do small deviationsin the amplifier's gain affect the.20. 34 Derive the expression for' the Bode sensitivity for the following links: P. 31 Prove that Horowitt sensitivity equals1IF: 32 The nominal value of the plant is 30. 37 Prove that if T is plotted upsidedown on the Nichols chart. including cases where the feedback is large (negative). The feedback path coefficient B = 0. chart in Fig. the curvilinear coordinates give20 IoglFI.theplantimplementationaccuracycanbe IF1 timesworsethanthe required accuracy for the closedloop system transfer function. . the plant sensitivity (1.5%? By 30 In the previous exam'ple. the value of the output would be 10. the nominal value is 50. Give numerical examples.24 Chapter 1. The perturbed value of the feedback is 20.Explain the effect of the feedback on the gain and the selectivity. With the nominal plant. the value of the output is 10. With the perturbed plant. Using Horowitz sensitivity. Plot the gain response of the receiver without and with regenerative feedback.output signal? What needs to be. introduce negative feedback (degenerative feedback) with B = 0. A. in your opinion. 36 From the frequency hodograph of T plotted on the Nichols. Sensitivity Feedback and velocity is expressed in kradsec.1 CAPBul dp 1+ CAPB 1 + CAPB 7 ' Therefore.0077. The nominal value of the feedback is 20.001 and thenB = 0. find the change in the value of the output when the plant changes from nominal to perturbed. What is the required implementation accuracy for these links? For example. 35 DeriveanexpressionfortheBodesensitivity of theoutputtovariations in the transfer function of the feedback path link B.10) can be derived by using (1 3 ) while keeping UI constant: S = dU2/U2 = PdU2 dP/P UzdP _ I P d UlCAPB .
72)sl.72 arcsec. (e) 2. 1. (b) w[(s + 2)s].1 8 0 O Fig. . Plot the relative disturbance spectral density without and with the feedback. (d) W[(s+ 10)sl. derivethedependenceofthe feedback and the output voltage on the load resistor RL. The return ratio is 1 OOO(s + 20)/[(s+ l)s].2mrad.4. 1.find the equivalent output impedance RLs of the voltage regulator. the system has current feedback at the output. 39 Thedisturbancespectraldensityisdescribedbythefunction(a) W[(s+ l)s]. Le. 40 Forthevoltageregulatordepicted in Fig. the output current in the load R ~ i stabilized s by the feedback. Here. Feedback and Sensitivity 25 Oo loo 20° 30° 40° 500 600 700 $00 900 phase deviation from . and the regulator output represents a current source. Therefore. B is the current sensing resistor.Chapter 1.21.1 mrad. By comparison with the known formula for the voltage at a source terminal: V = emf x RLI(RL+ Rs).' Hence. 41 Consider a current regulator with the schematic diagram shown in Fig. (c) 12 nrad. (c) W[(s+ 5)sJ. Find the value of feedback in the pointing control loop at this frequency required for reducing the disturbarlce to less than (a)0. (d) 0.05 Hz and amplitude 2 mrad in pointing of a spacecraft is caused by magnetometer boom oscillation.(e) W[(s+ 2. (b) 34 prad.20 Locus of Ton the Nichols chart 38 A periodic disturbance with frequency 0. 1. the output impedance of the regulator must be high.
Set €3 for the output current to be (a) 0.) The compensator transfer function C = 8(s + 1. 1. 42 A temperature control loop is shown in Fig. The. 5A. Plot the Lp1an. :.l)(s+ 25)].1 A. The transfer function (in 'ClkW) of the loaded furnace was measured and the experimental response was approximated by transfer function f ( s ) = SOO/[(s+ O. Explain the results.25 A.21 Current regulator. diagrams for the loop transfer function and for the inputoutput 'transfer function. measure of the temperature I Electrical thermometer I . Plot the output tirneresponse for 0.return ratio and the output impedance of the regulator. Fig.The output impedance of the amplifieris much larger thanRL+ B.66 A.23.72 A. Feedback and Sensitivity ..loop transfer function and inputoutput transfer function.01 the V step command. .26 Chapter 1. (At dc. The command is4 V.5V.the area surrounded by the "6 d B line on theNichols diagram. (b) 0. 1. Plot the output timeresponse to the step disturbance 100 W applied to the input of the furnace that represents some undesired cooling effects. thermometer transfer function is0.01 VPC.Include in theheatermodelasaturationwith threshold 2 kW. 43 A scanning mirror with an angular velocity drive is shown in Fig. (c) 0 .22.6)/(s + 0. Plot output responses to command steps of different amplitudes. The dimensionality of the signals is shown at the block joints. Find the. The heater is a voltagecontrolled power source with the gain coefficient2 kWN.22 Temperature regulator (a) Find the return ratio and the feedback at dc (in a stationary regime). 1.' . .Check whether the diagram 'enters . 1.e Nyquist diagram. (d) 0. . (b)CreateaSlMULlNKmodel. when s = 0. In each case find the. .9 ( 0 ) = 2OO0C/kW. Plot Bode. . Fig.4 A (e) 2.2). (f) 0. The reference voltage is0.
Feedback and Sensitivity 1 27 power source = Motor 4 ” angular velocity sensor  Fig. a . velocitv shaft Controller and Diver angular velocity angular measured l l i motor motor angular velocity Plant dvnamics Angular velocity sensor 1O(a.24 Angular velocity control (a) Find the... 1. a SlMUCtNK . ’. .‘ i. ’ . surrounded by.the torque toL the angle.25.function and for the inputoutput transfer function.ThemotortorquecontrolIdop is shown in Fig.30) s(s+300) volts . loop transfer function and inputoutput transfer function. ’ .01 .’ Include a saturation with threshold 2 kW in the heatermodel. I . Determine whether the diagram enters the area diagram. 1.23 Mirror drive The mirror angular velocity control loop is shown in Fig.Chapter 1. 44 A pulley with ‘torque drive is shown in Fig.1. the“6 d B line on the. At dc. is out of the feedback loop since the torque sensor is measuring the pulley torque directly at the motor. 1. the plant transfer function is 1. Plot an Lplane Nyquist diagram. The gain coefficient of the angular velocity sensor is 0. Plot Bode diagrams for the loop transfer .e.24. 1.model.Nichols Plot the output timeresponse fo’r the 1 V step command. Plot the output timeresponse to step disturbance0.Explain the results. rad/sec I Fig. The controller transfer function is C ( s )= 1O(s + 30)/[(s + 30O)sl. .V/(radlsec).Plotoutputresponses’tocommandsteps of differentamplitude.the transfer function from . control loop).1 radlsec applied to the input of the plant that represents the inaccuracy of the dc permanent magnet motor caused by switching between: the stator windings. (b) Create . . . The’ goal is to maintain the prescribed profile of the torque of the pulley so that the force in the cable lifting a load will be as ?hired. The plant dynamics.26. The motor (with driver) is a voltagecontrolled velocity source with transfer function 300 (rad/sec)N (the actuator might have an internal angular velocity. The mirror angular velocity differs from the angular velocity of the motor because of flexibility of the motor shaft and the mirror inertia. The plant transfer function (the of ratio the mirror angularvelocitytothemotorangularvelocity)is P(s) = 640000/(8 +160s + 640000).
for keeping constant the force applied to the cutting tool (there could be several possible kinematic schemes).01 the V step command.1 Nm. (d) With SlMULlNK plot position time history in response to the following command: 1 V fortheduration of .Checkwhetherthediagramentersthearea surrounded by the "6 dB" line on the Nichols diagram. the plant is a double integrator and its transfer function is 20/& plot the position time history in response to a step torque command (assume that at zero time the load is on the ground).1 m and the load mass is 2 kg. PlotanLplaneNyquistdiagram. the plant moment of inertia is0. Feedback and Sensitivity ang/e power source command _______+ = c Motor controller Motor . OV forthedur. i. Torquesensor 4 torque 0. (b) Assuming that the pulley radius is 0.28 Chapter 1. .26 Torque control (a) Find the loop transfer function and the inputoutput transfer function.e.1 V/Nm. The actuator (a motor with an appropriate driver) is a voltagecontrolled torque source with transfer function .Includeinthedrivermodelasaturationwith threshold 2 kW. Plot the torque timeresponse response to the disturbance torque as a step function. 1.02 kg x m2..2 sz rad volts Fig.30 NmN. Explain the results. . (c) CreateaSlMULlNKmodel. (e) Make a schematic drawing of torque control in a drilling rig. . Plot the torque timeresponse for 0. and1 V for the duration of the next 2 seconds.8(s + 1594s + 100).. The controller transfer function is Cys) = 0.8(s + 15) s(s+ 100) + 30 0. 1.2torque sensor Fig. The disturbance torque due to the motor imperfections is 0. (f) Make a schematic drawing of torque controlin a lathe. torque command hance Controller Plant " motor .ationof thenext 2 seconds. Plot Bode diagrams for the loop transfer function and for the inputoutput transfer function. Le. and Driver measured torque angle dynamics I .25 Torque regulationin a pulley The torque sensor gain coefficientis 0.2seconds. Plot output responses to command steps of different amplitude.
for the inputoutput transfer function. position wmmmd Controller + Driver and motor velocity angular b l/s angle Gear  position 0 .. E l command sensor position Optical .5. what is the resulting dead zone of the entire closedloop system? (b) Create a SIMULINK model. Plot Bode diagrams for the loop transfer function and. Plot the output timeresponse response to step disturbance . with gear ratio 0.28.27 A positioner with an optical position sensor Thepositionercontrolloop is shown in Fig.Plotoutputresponsestocommandsteps of different amplitude. + controller I I I Fig.03(s+ 20)/(s " ' 7 0 ) .2V/mm.Checkwhetherthediagramentersthearea surrounded by the "6 d B line on the Nichols diagram. from station to station and due to changing reception conditions. The disturbance in the angle represents the ball screw imperfections.01 V step command.1 mm. The linearized (differential) loop gainof automatic level control at the largest signal level is 1000. What is the range of the .e. Explain the results. If the gear has a dead zone of 0.1.5  mm Fig.5 mm/rad. The controller transfer function C(s) = 0. (l/s)0. PlotanLplaneNyquistdiagram. Include a saturation with threshold 3000 rpm in the driverandmotormodel.02 mm. 46 The signal at an antenna of an AM receiver varies100 times. i.The 'actuator (the driver with the motor) is a voltage controlled angular velocity source with transfer function 500 (rad/sec)N.1.0. 1.27. The plant transfer function is the ratio of the position to the velocity.28 Position control (a) Find the loop transfer function and inputoutput transfer function. 1. 0 3 ( s+ 20) rad/sec s+ 70 b 500 b l/s 0. Chapter 29 45 An xpositioner is shown in Fig. Plot the output timeresponse for the 0. A ball screw converts the motor rotational motion to translational motion of a tabre along the xaxis. The gain coefficient of the optical position sensor is 0.Sensitivity Feedback and 1.
be nearly constant.G): s . M = 0. = CAPB . the feedbackis positive but negligible (Le.1 N. and the temperature under the blanket may reach unsafe levels. 48 Due to the feedback.Po l + CAPoB 1  1 ' 1 '. from(1 . The opposite is also true (since the nominal and perturbed (current) values can be swapped). . the consumed power will. Without the thermal control. ~+CAPB =Fa Notice the peculiarity of this formula: the changes in the left side SH) (in are relative to the nominal plant value. Feedback and Sensitivity output signal carrier variations? 47 The' dependence of M on T and many other formulas of this chapter are bilinear functions. It is known that a bilinear function maps a circle (or a straight line) from the complex plane of the variable onto a circle (or a straight line) in the complex plane of the function.and it relievestheprecedingsignalsourcefromitsvoltagebeing reduced when the load is connected.CAPoB l+CAPB l+CAPoB PB CAP. while the feedbackin the right side is calculated for the perturbed plant value. 3 (b) F = 0.99. the voltage follower has high input and low output impedances.thermometer.30 Chapter 1.01. 49 The temperature will be kept safely loyv by reducing the consumed power. 21 (a) 0. 31 Giventhevalues P and Po forthenominalandtheperturbedplanttransfer functions. /I=/ = 1). Why then are the coordinate curves in the Nichols chart not circles? 48 What is the reason for using voltage followers? 49 A common electric heating pad has thermal control. What would happen if it is left under the blankets? What would happen if there were no thermal control? Answers to selected problems 1 (a)Heater. .furnacewithpayload.
orusinga feedback path with a specified transfer function. the schemes are briefly described below. The first type is referredtoas command feedforwarding andisshownin Fig. This definition reflects the importance of taking into account the large uncertainty in the signal transmission of the saturation link caused by the signal amplitude changes. coupling.. The chapter ends with a? . Fig. The methods for equivalent transformation of block diagramsare described. The transfer functionof the command feedforward path is (AP. crossed loops. with examples. nested loops.Typical MIMO systems are discussed.Therefore.introduction of multiinput multioutput (MIMO) systems. 2. Thepe. each having decisive advantages for specific applications.All of these methods enable to obtain the desired inputoutput closedloop transfer function with any arbitrary compensator transfer.4 canbeomittedfroman introductory control course. or a prefilter. 2.ANI) MIMO SYSTEMS . while the desired closedlooptransferfunctionisobtainedbyusinganappropriatefeedbackpath.Chapter 2 FEEDFORWARD.a command feedforward. the error feedforward scheme and Black’s feedforward increase disturbance rejection and reduce plant sensitivity. While the command feedforward scheme and its equivalents do not affect disturbance rejection and plant sensitivity.1 Commandfeedforward The accuracyof the system transfer function can be improved not only by the feedback but also by feeding certain signals forward and combining the signals at the load. Le. 2. 3 and 2.anddecouplingmatrices in theforwardand in thefeedbackpath. Example 1 fromSection 2. and the maidvernier loop configuration. Multiloop feedback systems are defined. the feedback.function.1 andSections 2 . Major practical types of multiloop systems are studied: local and common loops. following Bode. as those having a nonlinear (saturation) element in each loop.1 Commandfeedforwarding 31 .areseveralfeedforward schemes.thecompensatortransferfunctioncanbechosenas required to maximize the disturbance rejection. I* where Po is the nominal plant transfer function.1. MULTILOOP.I Thecommandfeedforwardschemeisequivalenttousingaprefilter. or by some cornbifizttiori of feedback and feedforward.
motors) must also be able to act fast and providehigh acceleration for the plant. and the advantages of the command feedforward decrease drastically. the transmission F times: U = ( ~ A~ + 1 = " .espe'cially over. andthesystemtransferfunctionapproaches 1.the sensitivity to the plant parameter variations. the feedback reduces. The command feedforward can substantially improve the accuracy of the output response to thecommand. Biological. certainly.threelimitationshoweveronusingthecommand feedforward: The plantmust be known pretty well. high accuracy must be available. In each path. and many other engineering systems must perform well both in slow and fast modes of operation. then Po cannot be made close to P. thefiequency bandwherethefeedback cannot be made large. (PJP) The inputoutput transfer function can also be expressed as  where Mo= TJ(To t 1). and MIMO Systems The inputoutput transfer function of the system with command feedforward is found by summingthetransmissionfunctions of parallelpaths of theinputtooutputsignal propagation. making the feedforward path gain too large would produce an excessively large signal at the inputof the actuator. when the plant transfer function does not deviate much from the known nominal plant transfer function Po. .. Feedforward. There is also a limit on the bandwidth of the command feedforward: at higher frequencies IPI typically decreases. In the slow mode. therefore. as when hitting a ball or jumping. for the transfer function to be feasible. and the actuator would become saturated. The power of the numerator polynomial of the feedforward path transfer function should be smaller than the power of the denominator. evenwhenthe feedback is not large. The accuracy of the fast components of the motion must be reasonably good but need not be as good as the accuracyof the slow mode action. Po = P and. fromthe formula) that the inputoutput transfer function is =1. and second. it follows directly from the block diagram (and. . as when surgery is performed. The same actuators (muscles. Le. if C = 0 (openloop case). when C + and both T' and TL1 vanish. For example. A plant with substantial pure delay would imply using feedforward with substantial phase advancewhich is not feasible. or the disturbance rejection. robotic. 0 The commandfeedforwarddoesnotchangethefeedbackabouttheplant.' therefore. making IPoI too small and.32 Chapter 2. If the uncertainty in the plant transfer function is large. P +T " F F F 1+ To' 1+~" This transfer function approaches 1 in two cases: first. then M = M. If the deviations of the plant P from the nominal value Po are small.P T ) Po U1 AP. Highaccuracy slow motion can be achieved with closedloop control using eyes or . Multiloop. To= T. however. There are. Example 1.
2. When the motion comprises both slow and fast components.components.4 has return ratio T and closedloop response B"T/(T+ 1 ) . and MlMO Systems 33 other position sensors. At higher frequencies (for fast motion). the actuators must be commanded directly. the compensators' gains roll down and the command feedforward paths become dominant.2 Feedforward system with fast actuator loops andslow outer loops At low frequencies (i.. 2.Chapter 2. the system inputoutput transfer function 1s To' 1 + T" approximates 1 overthe entire frequencyrange of operation. The system in Fig. Feedforward.2 Prefilter and the feedback path. since the gainof the compensator gadually decreases with frequency. The feedback configuration.e. Multiloop. of while maintaining a reasonable accuracy. For each of the controlled output variables. The actuators are equipped with wideband (fast) feedback loops using local sensors (inner loops) which make them accurate..This system has returnratio T and closedloop response RT/(T + 1). only the inner loops are closed.) are controlled by the outer feedback loops. The commander transform thegeneralinput command into commands for the individual actuators.2. the outer feedback loops control the slow components (as when adjusting the general directionof running) while for the fast. for slow motion).although muchmore accurately at lower frequencies where the feedback is large. the compensators' gains are high and the system operates closedloop. equivalent The block with transfer function R inthecommandpathprecedingthe command summer (same as the feedback summer) in Fig. To make the actuators move the plant with the maximum speed and acceleration they are capable. These twosystems are equivalent if B = 1IR. 2. The block diagram of a feedbacwfeedforward system to answer these conflicting requirementsisshown inFig. doesn't suit the fast action mode since the speed of the feedback action is bounded by the delays in the feedback loop.3is called theprefilfer.2.etc. The transition in accuracy from the slow to the fast mode is gradual. . 2. however. The plantoutputvariables(averagespeed. Fig.direction of motion.
as will be shown in Chapters 4 and 5. and it is much easier to design these .5 Error feedforward .1.5(a) describes an entirely different scheme known as error feedforwarding. . the compensator should not be compromised during the next stage of the design.4 have the closedloop transfer hnction M/Mo. 3 System with prefilter Fig. and thefeedback It might seem attractiveto integrate the prefilter.2. 2. 2.3 and 2.4. or B.1.the compensator C is defined. The system design is performed in two stages:first.2. so these linksdo need to be precise. the same as that for the system in Fig. The tolerances in the prefilter and the feedback link directly contribute to the output error.3. This goal can be achieved by a proper choice of B. R. the invutoutput transfer ficnction and can make it closer to the desired.l(b). the design of these blocks is to a large extent independent. then.3 Error feedforward Fig.or the feedforward path. or R. 2. The compensator transfer function must be chosen so as to maximize the feedback over the bandwidth of interest . Since the feedback reduces the effects of the compensator parameter variations. and may be different at different frequencies. Therefore. 2. the link into a generalized linear subsystem which can be designed using some universal performance index. and feedback path. the systems in Figs. all three systems are potentially equivalent.4 System % w i t h feedback path With R = 1/B = l/Mo. but do not change the feedback or the sensitivity. or the feedforward path (M0)'. 2 . A system having all three links: feedforward path. and MlMO Systems Fig. this is not recommended since the sensitivities and the accuracies with which the blocks should be implemented are quite different. comparedwiththesysteminFig.Feedforward. The required of the accuracy of thefeedforwardpathimplementationdependsontheaccuracy knowledge of the plant transfer function. However. 1 " 2 1~ Load Fig.34 Chapter 2 . 2. Compensator. blocks one at a time. Once designed. 2. prefilter. l. these links need not be precise. The three equivalent methods modifj. which is the implementation of a suitable nominal closedloop response. Multiloop. can always be equivalently transformed to any one of the systems in Figs.. 2.
AlllrqEB . Note that no feedback appearsin this system. 2. measured via the Bpath. Fig. the sensitivity approaches zero.aroundthesametimeheinventedthefeedbackmethod. the output summer can be made using a bridgetype signal combiner as shown in Fig. If either A M or A E or both equal 1/B. consider Black's feedforward method for sensitivity reduction. Multiloop.PB l+CA. l+CA. 2. For an electrical amplifier system with a known load.P U. (b) ideal case The inputoutput transfer function is AM +A E . The sensitivity of thesysteminputoutputtransfer function to theplantparametervariations(andtovariations in C and AI) canbe calculated as (1 . is The sensitivity of the outputto variations in A M is dU2 .6 Black's feedforward system: (a) general.4 Black's feedforward method Finally.PB + 142 9 andif A2 = 1.A2)IF. the inputoutput transfer function 1/B. Feedforward. The error is amplified by the error path and added to the system's output. so as to compensate for the initial error in the main path. the bridge prevents the output signal from the upper path from going into the feedback path. If A2 is made close to 1. (a) (b) Fig. 2. which was invented by HaroldBlack. and the lower one is the error compensation path. Practical applications of this method of sensitivity reduction are restricted by the difficulties in the designof the output summer. andMIMO Systems 35 The inputoutput transfer function is " U2  CA.6(a) depicts the method. It is more difficult to implement a mechanical system with such properties. then U 2 = Ul/B. The error signal is the difference betweenthecommand U1 andtheoutput of themainpath. 2.5(b). The upper signal path is the main one.Chapter 2.
Also. the error signalis the difference between two nearly equal signals. In some physical systems. 2. and the only difference in the resulting inputoutput transfer function of the system is ZM + ZB + ZE. . Then. butit is not common in control systems. Theremay be applications to control systems when extreme accuracy is required. If B = AM. the phase difference between the signals reaching the summers remains the same. in the case. Multiloop. 2. for signals with frequencies from hundreds of MHz to tens of GHz. the extra delay . and A E can be lowpower. 2. a different Fig. where both of the transfer functions are 1/B as in Fig. and ZE as shown in Fig.6(b). and MlMO Systems (derivation of the formulais requested in Problem11). respectively zM. These delays do not prevent tlie use of feedforward if they are properly compensated by insertion of delay link ZM + ZB in the signal path to the first summer.8. the sensitivity (2. the feedforward scheme provides redundancy: if one of the actuators. 2. both actuators are commonly stabilized by internal feedback or by some adaptive automatic gain adjustment. and delay linkZB + ZE at the outputof the main channel. and A E incorporate substantial delays.36 Chapter 2. 2. fails.2) becomes zero.8 Modifications of a linear system . In this case the output effect of the disturbanceD is also zero.on systems. as illustrated in Fig.7. In order to make AM and AE each equal to 1/B. When the gain coefficient of this actuator isA E = lIB.5 Multiloop feedback systems Linear systems can always be transformed to another configuration with number of loops.7 Black's feedforward with delay compensation This method is often employed in lowdistortion amplifiers for telecommuni cati.   Fig. . In this case the actuator made very precise. the links AM. the remaining one takes the full load and the inputoutput gain remains unchanged. and preserve these conditions in spite of variations in these actuators' parmeters. Feedforward. Such an actuator can be the error is small. zB. 2. B.
arrangement.. for small amplitude signals. with saturation links equivalently replaced by unity links.6 Local. system (a) is a singleloop system. it is required to enloop each stage by the same value of feedback.1O(b).be increased by adding inexpensive gain blocks.10 Local loops (a) and the common loop arrangement (b) If each of the three links with nominal gain coefficient k has the same tolerances. and nested loops Fig. i. Fig. In this case the closedloo gain P/(T + 1) of the singleloop system is much higher than the closedloop gain k!/(T+ 1)3 of the system with local loops. common. then to provide the same degree of accuracy.l(a) depicts local feedback loops. 2. Such systems will be studied in Chapters 413. Each stage of an amplifier has gain coefficient of 50. The total gain of the chain of these links is reduced by the productof all these feedback values. / O O ~ shown in Fig. The feedback . Fig. but this effect does need to be taken into account when designing analog compensators. This gain reduction effect is much larger in this arrangement than in the C O ~ ~ O ~ . For control systems. This definition is related to the problem of stability analysis of practical systems whose actuators are always nonlinear. only the loops which comprise nonlinear saturationtype elements are counted. wewillanalyzefeedbacksystemsinonlythelinearstateof operation. For example. Feedfoward. This is why commonloop feedback is preferred in amplification techniques where the resulting gain is important.e. 2. Example 1. 2.9 Singleloop (a) and threeloop (b) feedback systems Inthischapter. 2. and MlMO Systems 37 In accordance with Bode’s definition of physical multiloop systems. The sensitivities to each link’s parameter variations dependon the feedbackin the local loop. 2.Chapter 2. this consideration is not important since the gain can. Multiloop. in Fig. and system (b)is a threeloop system.9. 2.
2.12(b). 2. 2. An easily implemented large local feedback loop about the driver amplifiers will reduce the tolerances of the forward path to only that of the compensator. Here. as in the amplifier illustrated in Fig. The outer loop improves the accuracy of all links in the forward path. Multiloop. If opamps are used in the compensator.38 Chapter 2. Crossed dc feedback loops are frequently used in bias stabilization circuitry in amplifiers. Such loops are often formed by parasitic coupling. For example. . The choice of these variables alters the plant transfer functionP which is the ratio of the output to the input variables of the plant.12.) 2 . and the rather narrow bandwidth outer loop. Local loops are often made about links with large parameter variations. Then. The variables that are fed .backin the inner loops can be different: at the output of the driver. 2. When the common loop configuration is used. to be discussed in detail in Chapter 7. and MlMO Systems about each stage needs to be at least 10 to make the gain coefficient stable in time.ratefeedbackaboutthe motor is typically accompanied by voltage feedback about the driver. Feedforward. (These issues will be studied in more detailin Chapters 4 .and at the output of the actuator. the driver amplifier is enlooped by large feedback to make its gain accurate and stable in time.Whentheactuatoris anelectromagneticmotor. the velocity (rate) or the force. The nested loop arrangement is employed for several reasons. These loops can be nested as shown in Fig. about the plant. and actuator. primarily because the feedback bandwidth in the outer loop cannot be made arbitrarily large.large gain uncertainty and variations due to power supply voltage and temperature changes. the intermediate bandwidth loop is about the actuator (motor). the variable c& bethevoltageorthecurrent. consider the typical case of a driver implemented with an operational amplifier.1 1. including the plant.7. the gain coefficient will be much higher. 12500. force feedback about the actuator makes a rigid bodyplantadoubleintegrator whiletheratefeedbackmakestheplantasingle integrator. 7 Crossed loops and maidvernier loops Crossed feedback loops are shown in Fig. The actuator loop makes the actuatorplusdriver subsystem accurate and stable in time. The opamps without feedback have very .they must also have large local feedback.thetotalamplifier gain coefficient willbe125. . voltage n I velocity I U Fig. plant. . in a position control system.1 1 Nested feedback loops For example. when localfeedbackisused. and also to manipulate the output impedance of the driver. The wideband inner loop is about the electrical amplifier (driver).
If such an actuator is not available. due to its large inertiawhich is represented in the block diagram by the lowpasslink LP. . Multiloop.13(a). I I z Fig. the diagram is modified shown as in Fig. 2. For this purpose.13(a) and (b) show two equivalent block diagrams for the mainfvernier loop arrangement.These components. Feedforward. 2. the actuator must be fastand powerful. However. 2. then an arrangement of two complementary actuators can be employed: and the vernier acfuafor which is ordersofmagnitude faster but the main acfuafor also ordersofmagnitude less powerful. smaller in amplitude but rapidly changing. 2. 2. From the diagramin Fig. the main actuatorcannotrenderfastsignalcomponents.13(b).). voltage etc. it is apparentwhat the actuators are doing.2. are provided by the vernier actuator.12 (a) Crossed feedback loops.13 Feedback systems with the main and vernier loops . and MlMO Systems 39 Fig. (b) crossed dc loops in an amplifier For high control accuracy over a large dynamic range. Fig.Chapter. The main actuator provides most of the action applied to the plant (force.and if large changes in the output variable need notbe fast.but it is rather difficult to figure outhow to design the compensators in the main and vernier channels.
when the error is large. and the cart desaturates the voice coil.13. the lengths of the optical paths from the two primary mirrors to the summing point must be kept equal to each other.(A voice coil is an electromechanical actuator baed on a coil placed in a field of a permanent magnet. The feedback bandwidth of each subsequent loop increases. 2. The voice coil desaturates the piezoactuator. In . The small platform bearing the piezoactuator is moved by a voice coil. voice coils are widely employed in loudspeakers and hard disk drives where they position the readinglwriting heads. 2. The voice coil is the vernier for the cart. etc.8 Manipulationsofblockdiagramsandcalculationsoftransfer functions Equivalence block diagram transformations facilitate the conversions of various configurations to standard ones for the purpose of analysis.)Theslowbutpowerfulmain actuator unloads the vernier actuator from slow but large amplitude commands and disturbances.13(b) is shown only to simplifytheexplanation of how thesystemworks.) 2. but its maximum displacement (stroke) is only 50 pm. the diagram in Fig. Multiloop. The output signal of the vernier actuator is applied to the feedback summer of the main loop. Each extra loop provides an economicalway to improve the control accuracy by a few orders of magnitude. The command for the main loop is zero since the desired valueof the vernier actuator output for slowly varying signal components is zero.14(a) can be transformed into diagrams (b) and (c) by changing the node from which the signal is taken while preserving the signal value at the branch output.in linear and nonlinear modes of operation. However. The command is given to the vernier loop summer.13(a). Two examples of such a system are described in Appendix 13. there is no physical command. Due to the difference in the lqop bandwidth.40 Chapter 2.the orbiting stellar' interferometer (a high resolution optical instrument to be placed an an orbit about the earth). The voice coil is placed on a cart that can be moved on wheels along seta of rails. the vernier actuator becomes saturated and cannot compensate highfrequency disturbances in the system.) The accuracy of the voice coil control loop is lower since its feedback is limited by some mechanical structural resonances. 1 cm. and the piezoelement is the vernier for the voice coil. it is clear how the system operates. the system can be extended to a threeloop configuration. both .The position of one of these 'mirrorsis regulated by three means. the command summer in the main loop in Fig. loop couplingis rather easyto account for during the system stability analysis. The mirror is mounted on a piezoelectric actuator. By the same principle. The optical path lengths are measured with laser interferometers. For the' purpose of this 'adjustment. (Therefore. The piezoelectric actuator can be controlled with nanometer accuracy. ' Example 1. and the vernier actuator tries to reduce the error rapidly.the vernier is desaturated . and must be adjusted with nanometer accuracy. but the actuator maximum strokeis much longer. This situation is corrected . . For example.by the main loop. and MlMO Systems Now. the forward path transfer function and the feedback loop return ratioare preserved. (The control system is described in detail in Appendix 13.The entire control system is able to adjust the optical path length rapidly and very accurately. in one of the paths a variabledelayisintroduced by bouncingthelightbetweenadditionalmirrors. summer in the main loop in Fig.In this transformation. Feedforward. 2.
the following evident rules apply: (1) Transmission alongaJorward path is reduced bv the value o f & feedback in the 1004 that includes links in the p&.16 is obtained Erom the diagram in Fig. loops with unity forward paths. instead of the block diagram representations.15 by equivalencetransformations. 2.. There are two forward paths and two loops with return rations bga and cdeh. 2. Example 1. The diagram in Fig. i.Chapter 2.16 has tangent loops. the total transferfunction can be found bv superposition of the signals propagating along the paths. . n Fig. so that the transfer function is Gf abcde I 1 + bga (1 + bga)(1 + cdeh) Often. Multiloop. 2.e. systems are described by the signal flowchart exemplified in Fig. (2) When there are several parallel forward paths. 2.1 4 Feedback system equivalent transformations For the transfer function calculation.17.1 5 Block diagram of a feedback system Fig..16 Feedback system with tangent loops The diagram in Fig. According to (1. 2.e. 2. and MlMO Systems 41 (a) Fig. 2.mctions. Feedforward. The signalstakenatdifferentpoints are multiplied by additional blocks’ coefficients so that the signals at the outputs of the branches remain the same. i. summing& path transfer$.3)’ each tangent loop reduces the signal transmission by the value of feedback in the loop.
2. the inputoutput transfdunction is the forward p& transmission divided by the sum of all loop return ratios and I . the third rule can be formulated: (3) When the loops are nested.18 can be calculated as the sum of transmissions along two parallel paths.42 Chapter 2.The gain coeflicient of the graph shown in Fig.18 Flowchart Next. by converting all nested loops to the loops between the same nodes. Multiloop.1 7 System flowchart representation Example 2 . 2 Fig. and MlMO Systems Fig. 2.19 Transformation of (a) nested loops to (b) parallel loops and further to (c) a single equivalent loop . Thus.As shown in Fig.19. 2. consider nested loops. we obtain several parallel loops. Q abdeh Fig. The three rules constitute Mason's rule. The equivalent single loop has the loop transfer function equal to the sum of all the nested loop transfer functions.Feedforward. divided by the feedback in the tangent loop: (32X5)lO = 2. 2.26 1+5x6 . 2.
plantmodels. In many cases. The transferfunctionforthesystemwithnestedloopsshownin Fig. whatever the name. couplingexists but is small. The number of feedback loops does not necessarily correlate with the number of inputs and outputs. The transfer function of the plant for each loop is the plant transmission from the actuator output to the sensor input. 2. The controlled variablescouldbe. some of the. then the multiloop system is just a set of individual singleloop systems.2. and then the filtered and the unfiltered sensor signals are fedtodifferentsummingpoints. For example. + CDAPB.) On the other hand. as in Example 1 in Chapter 8. is no better than that of the standard control system configurations..thiscanbeeasilydone).theachievable performance of these control. 2. (2. singleoutputsystemissometimessplitintoseveralpathstoformposition. and these signals are combinedlinearlytoformthesignaldrivingtheactuator. different summing points. For example. velocity. singleoutput (SISO) system. + DAB. The transfer function from the. Multiloop. position command in a singleinput. Feedforward. If the equivalence of these block diagrams to the block diagrams shown in Figs.11 is CDAP DB. a multiloop system is employed to improve the Performance of a singleinput. and acceleration command. if the number of commands is 2. 2. and estimators.4 is proved(veryoften. and MIMO Systems 43 Example 3.velocity.12 are multiloop SISO systems.9 MIMO feedback systems Multiinput multioutput ( " 0 ) systems have several command inputs.someblock diagramsincludelineartimeinvariable links named predictors. it is often possible to prove the equivalence of different control schemes.3. (The performance might be inferior if the block diagrams are chosen that inherently were to limit the order of the compensators. the systems shown in Figs. 2. . 2. and several output variables are controlled simultaneously.forexample. An example of a MIMO systemisshowninFig. If the coupling transfer functions zero.4) With the block diagram manipulations. this is a 2 x 3 system.angles of different. bodiesoranglesindifferent dimensions of the same body. schemes. the sensor output is often passed through a lowpass filter to attenuate the sensor noise (as well as some components of the signal).20 where differentplant variables are regulated by separate loops. For example.11 and 2.ith actuator output to the jth sensor input (i # j ) shown by the dashed we all line is called a coupling transferfuncfion.linear filters are used to estimate the output position.1. 2.potentially equivalent block diagrams may have certain advantages from the implementation point of view. and the number of outputs is 3. and these three signals are separately feedforwarded into three . and acceleration. Very often. + 1 where D is the driver transfer function.Chapter 2.
44
Chapter 2. Feedforward, Multiloop, and
MlMO Systems
Sensor : *
matrix
f?
4
4 Sensor
~:l!j
2;
U
(b)
(a)
Fig. 2.20 2 x 2 MlMO system with loops to control nearly independent variables. The decoupling matrix canbe placed in (a) the feedback path or (b) in the forward path.
Most often,actuators 'arerelativelyexpensiveandtheirnumberinengineering systems needs to be reduced to a minimum. Therefore, as a rule, only one actuator is assigned to do a specific job (Example 3 below offers exceptions): one actuator moves the plant in one direction, the second in another, etc. Or, in the case of an electrical signal generator, one actuator varies the signal frequency, the second one the signal amplitude, the third one, the temperature of the quartz resonator, etc. Because of this, theactuatorloops intheblockdiagraminFig,2.20 are alreadytoalargeextent decoupled, Le., the terms on the main diagonal of the plant matrix (from plant actuators. to plant sensors) are substantially bigger than the offdiagonal terms. Coupling between loops canbe compensated for by using a,decoupling mafrix, whose outputs only reflect the action of an appropriate actuator. The decoupling matrix makesthefeedbackloopsindependent of eachother,simplifyingthedesign and improving the system performance. The decoupling can be done in the feedback path by , decoupling the sensor readings, orin the forward path by decoupling the signals going to actuators. Either method can make the loops independent of each other, but there is substantial differencebetween the methods: the matrix needs to be precise when placed in the feedback path, and can be less precise when placed in the forward path. A decoupling matrix for linear plants can befoundbyinvertingthematrixof known coupling transfer functions. If the coupling transfer hnctions do not contain pure delay, the decoupling matrix is causal and can be implemented with a digital or an analog computer. However, since the plant parameters are not known exactly, decoupling is never 'perfect. The following types of multiloop systems are most often encountered in practice: local actuator feedback, vernier type control with actuators differing in speed and in power, and nearly decoupled control where each of the actuators dominantly affects a specified output variable. When fast actionis of utmostimportance,complexengineeringandbiological systems are typically arranged asan aggregation of several SISO 'mechanisms with large and relatively wideband feedback in each loop and a complex precision commander producing commands to the mechanisms. When the action need not be very fast but the accuracy is of prime concern, additional slower common feedback loops are added to precisely controlthe output variables as shown in Example 1 in Section 2.1.
Example 1. The azimuth angle of the antenna in Fig. 1.l(b) might be regulated as well as the elevation, and the result would be a 2input, 2output system. The coupling between the elevation and the azimuth loops is typically small, and can be calculated
Chapter 2. Feedfoward, Multiloop, andMIMO Systems
45
and compensated which results in practically decoupling the loops.
Example 2. Spacecraft attitude controllers are commonly arranged as three separate loops for rotating the spacecraft about the x, y, and zaxes. The spacecraft inertia matrix is not symmetrical about all the axes. Therefore, the transfer function about one axis depends on the rotation angle and velocities about the other axes, and the three controllers are coupled and cannot be considered as three separateSISO systems. Good decoupling can be achieved over most of the frequency bandwidth of interest wherethespacecraftparameters are wellknownandthedecouplingmatrixtransfer functionscan beaccuratelycalculated.However,oversomefrequencyranges,for example, at the sloshmodes of the propellant in the fuel tanks, the spacecraft parameters have much larger uncertainty and,the calculated decoupling matrix is not very accurate. The uncertain coupling necessitates a reduction in the feedback in the control loops as will be discussed later,in Section 4.4. Example 3. Multipleactuators of thesametypecanbeusedtoachievethe appropriate power andlor balance. An example is the use of multiple power plants on jet transports is shown in Fig. 2.21. The output is oneof the variables defining the airplane attitude and velocity (i.e., this block diagram shows only a part of the entire control system).
I
I
Fig. 2.21 Several parallel power plants system
The arrangementprovidesredundancy, i.e.,oneengineoutcapability(OEOC). of Specialcontrolmodes(autoormanual) may be necessarytosupportthissort operation.For this' purpose,additionalfeedbackloops usingaerodynamiccontrol surfaces are applied so that a single actuator can power the plant independently in the event that the other actuators fail. The system is a rnultiloop MIMO system.
Example 4. In a TV set or in a VCR, there we several hundred feedback loops. More than 90% of the loops control electrical variables (currents, voltages), and some of the loops control image color and brightness, speed of the motors, and tension of the tape. The majority of the loops are analog, but some are digital, particularly those for tuning the receiver and for controlling the display. This, say, 300 x 300 MIMO feedback system i s conventionally designedwith frequency domain methods, as if,the loops were of 300 SISO systems. The independent, Le., as if the system were merely a combination variables to be controlled are to a large extent independent, i.e., the, diagonal terms are dominant in the 300 X 300 matrix. Only seldom is some primitive decoupling used in the forward path. The decoupling matrix is sometimes included in the feedback path to calculate the variables fed back from the sensors' readings.
46
Chapter 2. Feedforward, Multiloop, and MIMO Systems
The design of a MIMO controller as a combination of several independent loops hastheimportantadvantages of structural design. It simplifies the system testing and of modification', and troubleshooting, improves reliability, and simplifies the work in redesign.Tomeetthese goals, mostengineeringdevicesaredesignedstructurally, spite of themathematicallyattractiveidea of combinedoptimization of the entire 300 X 300 multivkiable system which, ideally, must produce at least as good or better of losing the advantages of the structural approach. performance  but at the price
2 . 1 0 Problems
1 For a tracking system(€3= 1 )with Tequal to (a)5; (b) 20; (c) 80;(d) 120;(e) 2.72 find the value of the prefilter R that makes the closedloop transfer function equal to
1.
2 Find the compensator and feedback path transfer coefficients for a system without a prefilterso that the system is equivalent to the system of Problem 1, with AP = 10 in both systems.
3 Find the compensator C and feedforward path gain coefficient FF for a system without a prefilter, withB = 1, and with AP = 10,so that the system is equivalent to the systemof Problem 1.
4 include linkB in the feedback path in the block diagram depicted in Fig. 2 . 1 .Derive an expression for the inputoutput transfer function. 5 C = 2,A = 1. Plant gain coefficient P is uncertain within the 10 to 20 range, and nominal plant gain coefficient Po = 15. Calculate the inputoutput gain coefficient ranges without and with a feedforward path. Does the feedforward affect the ratios of the maximum to the minimum inputoutput gain coefficient?
is inverselyproportionaltothefrequency.Atwhat 6 Theloopgaincoefficient frequency rangesis the benefitof using feedforward most important?
o ) ,the actuator model includes a linear gain block with gain 7 Plant P(s) = 1O/(s + l coefficient A = 10 followed by a saturation link with threshold 1, and qs) = 0.3(s + 0.35)/(s + 3 ) . The feedforward path transfer function is 0.1a(s + IO)/(s + a).' The command is sinusoidal, with possible frequencies from 0 to 10 Hz. Plot the frequency responses with MATLAB. Choose coefficient a such that the signal amplitude at the input to the saturation block will not exceed the threshold. What is the bandwidthof the feedforward?
8 Sameproblemasthepreviousone,buttheinputsignal simulations with MATLAB, find a by trial and error.
is astep 1 V. Make
I 9 find the Bode sensitivityof transfer function W .(a) W I = 100 and W2 = 2; (b) W I =SO and W2 = 50; (c) W1=9 and W2 = 10; (d) W1= 101s and W2 = 1 OO/&
+ W2 to WI if:
10 Find the Bode sensitivity to the transfer functions of the links P, Av, and AM in Fia. 2.13.(Hint Use the chain rule. First. emdov Bode sensitivitv for a singleloop
Chapter 2. Feedforward, Multiloop, and MlMO Systems
47
systemforthecompositelinkincludingthemainandvernierchannels;then, multiply the composite link Bode sensitivity by the sensitivity ofthecomposite channeltransferfunctiontovariations in onlyonechannel.)Giveanumerical example.
11 Derive the expression for*thesensitivity to variations in AM of Black’s feedforward system shownin Fig. 2.6. Compare two cases: (a) when AM= 10 and the values of the rest of the links’ gain coefficients are nominal, Le., AE = 1 and B = 0.1, and (b) whenAE deviates by3 dB from the nominal value of IO.
12 In Fig. 2.6,AE = 95, B = 0.01.Find the sensitivity of the output to variations in AM when AMis (a) 100;(b) 105; (c) 150.
,, ,
13 For the conditions described in the previous problem, and considering the maximum output signal (saturation threshold) in the main amplifier to be AM^ = 10, find themaximum output signal in the error amplifier. What is the conclusion?
, .
14 In Black’s feedforward system shown in Fig. 2.6, the erfor amplifierAE has internal feedback 6.Using the chain rule, find the sensitivity of the system’s inputoutput transfer function to the error amplifier gain variations.
15 How many loops, according to Bode’s definition, are in the systems diagrammedin Fig. .2.22?
(a)
I
Fig. 2.22 Feedback systems
16 (a) In Fig. 2.11, the driver gain coefficient changes with temperature k30% by from the nominal, the actuator changes by k15% from the nominal, and plant transfer function is uncertain within 2 dB. The loop gain in the driver loop is 30 dB, and in the actuator loop (with the driver loop closed) is 10 dB. Find the total uncertaintyin the plant loop gain.
48
Chapter Feedforward, 2. Multiloop,
and MlMO Systems is k3 dB,actuator
(b) Same problem but the driver gain coefficient uncertainty. &2 dB. uncertainty is&2 dB, and plant uncertainty is
(c) Same problem asin (b) but with loop gain in driver loop 40dB and 20dB in the actuator loop.
(d) Same problem as in (b) but with loop gain in driver loop 20 dB and 20 dB in the actuator loop. 17 Explain why a pair of actuators, one highpower and sluggish, and one lowpower and v e y fast, should typically cost less than a,single powerful and fast .&ctuator. 18 Find the inputoutput transfer function for the system shown in Fig. 2.1 0. 19 Find the inputoutput function of the system shownin Fig. 2.15. 20 For the multiloop feedback system described in Fig. 2.1 1, find the inputoutput transfer function and the sensitivities to variations in driver, actuator, and plant. 21 Deriveexpressionsforinputoutputtransferfunctionsforthesystemsshown Fig. 2.23. in
22’ (a) Calculate the decoupling matrix for the system where the sensor readings x’, y’, z’ are related to the actuator outputs x, y, z by:
= 2~ + 0.2~ + 0.3~, y’ = O.lX+ 2:1y+ O.lZ, Z‘ = 0 . 0 4 ~ + O.ly+ 1 . 9 ~ .
X’
Since the diagonal terms are dominant, X = O.~X‘, y = 0.5y’,
Z=
0.5~’.
A better approximation to x is found by substituting these first approximations to y and z.into the first equation:
X
0.5~‘ 0.05~’  0.075 z’.
Proceed with better approximations to y, z Compare these expressions with the exact solution found with MATLAB script:
A = [2 0.2 0.3; 0.1 2.1 0.1; 0.04 0.1 1.91;
inv(A)
ans =
0.5038 0.0443 0.0772 0.4795 0.0235 0.0243 0.0094
0.0215 0.5291
Draw the flowchart for the solution in the form shown in Fig. 2.24, and put the numerical values in. (See also Problem 6.10 for an analog computer implementation of the decoupling matrix using six opamps (two quad opamp I C ) and 18 resistors.)
Chapter 2. Feedforward, Multiloop, and
MlMO Systems
49
Fig. 2 . 2 3 Flowcharts
Fig. 2.24 Decoupling matrix flowchart
x', y', (b) Calculate the decoupling matrix for the system where the sensor readings
z' are related to the actuator outputs x, y, z by:
50
X'
Chapter 2. Feedforward, Multiloop, and MlMO Systems
= 2X+ y + 0.32, y' = O.lx+ 2yF 0.5z, Z '= 0.4+ ~0.5+ ~ 1.9~.
by inverting the coefficient matrix MATLAB. with (c) Sameas (b) for:
X' = 3~+ 0 . 4 +'0.3~, ~ y' = 0.3~ + 2.1y + 0.22, Z '= 0 . 0 4 ~ + 0.1y + 1.92.
(d) Same as (b) for: x'= 2x+ O.ly+ 0.12, y'= O.~X+ 3.ly 0.12, + 2 ' = 0.04~ +0 . 4+ ~ 1.92. (e) Same as (b) for:
x'= x + y  z , y'= x  y + z , Z'=  x + y + r .
(This arrangement of three piezoactuators and three load cells has been used in the spacecraft vibration isolation system described in Section 6.4.2.)
23 The frequency of a quartz oscillator depends on the crystal temperature and on the
power supply voltage (the voltage changes the capacitances of pnjunctions of the transistorthatparticipate in the resonance contour). The temperature of the environment changes from 10' C to 7 0 ' C. The power supply voltage uncertainty range is from 5 V to 6 V. The oscillator elements are placedin a small compartment ("oven") equipped with. an electrical heater and a temperature sensor. The temperature and the dc voltage are regulated by control loops. The thermal loop return ratio is The dc 600. voltage stabilizingbop return ratio is 200. The references are7 0 ' and 5 V, and the loops maintain the quartz temperature close to7 0 ' and the power supply voltage, close to5 V. For the employed quartz crystal and the transistor, the dependencies of the frequency of oscillation on the crystal temperature and on the power supply voltage are well approximated in the neighborhood of the references by linear dependencies with coefficients 1O4 Hz/% and 1 0"3 HzN. The maximum disturbances in temperature and voltage are 60' (when the environment temperatureis 10') and 1 V (whenthepowersupplyvoltage is 6 V). Fig. 2.25 shows the flowchart for calculations of the effects of the disturbances.
60%
1v
Fig. 2.25 Flowcharts representing the effects of the voltage and temperature variations on variations of the oscillator frequency
Chapter 2. Feedforward, Multiloop, and MlMO Systems
51
The loops are coupled since the dc voltage also affects the power dissipated in the transistor and, consequently, the oven temperature with the rate 20"CN. The flowchart represents a doubleinput singleoutput system. No decoupling between the control loops is required since the coupling is small and onedirectional, from the voltage to the temperature loop. (The effect of temperature on the voltage loop is negligibly small.) Calculate the total range of the frequency variationsAf due to the instability of the environment temperature and the power supply voltage. 24 Prove that, generally, when sensitivity is 0, redundancy is always provided. (Hint Use bilinear relation W = (aw + b)/(cw+ d) for the general dependence of a linear system transfer functionWon a link transfer function w.)
I
Chapter 3
FRElQrJENCY RESPONSE METHODS
Some requirements to control systems are typically expressed in frequency domain (such as disturbance rejection), while some others are most often formulated in time domain (such as rise time and overshoot). The latter need to be converted into the frequencydomainspecifications in ordertousefrequencydomaindesignmethods. Formulationsofthetimedomainrequirementsarecommonlyverysimple,andthe equivalent frequencydomain formulations are also simple. The requirements can be translated between the domains with simple approximate relations. Since most control systems are of the lowpass type, responses of standard lowpass filters are reviewed for future references. Typicalclosedloopfrequencyresponsesforhomingandtrackingsystemsare considered. TheNyquiststabilitycriterionisderivedanditsapplicationsreviewed.Stability margins are introduced and the Nyquist stability and the absolute stability discussed. The NyquistBode criterion is developed for multiloop systems’ stability analysis with successive loop closure. Feedback systems with unstable plants are analyzed with the Nyquist criterion and with the NyquistBode criterion. The effect of saturation on the system stability is briefly discussed. Static error reduction is considered for systems of the, first, the second, and the sew0 types. The notion of minimum phase (m.p.) function is introduced. The theorem is consideredof equality to zero of the integral of the feedback in dB over the frequency axis. The Bode integral of the real part of a function is applied to evaluation of impedances.TheBodeintegral of the imaginary part of afunctionisappliedto estimation of feasible changes in the loop gain response. The meaning and the significanceof the Bode general phasegain relationship are clarified, and the procedure for calculating the phase from a given gain response is explained. The problem of finding the Bode diagram from a given Nyquist diagram is considered. Nonminimum phase lag is studied. A criterion is derived for the transfer function of two m.p. parallel paths to be m.p. The use of MATLAB and SPICE is illustrated for feedback system modeling and analysis. When the book is used for a singlesemester introductory control course, Section 3 . 9 . 3can be bypassed.
3.1 Conversion of timedomain requirements to frequency domain
3.1.1 Approximaterelations
Since signals can be substituted by the sums of their sinusoidal components, and in linear links, the signal components do not interfere (Le., the superposition principle applies), linear links are fully characterized by their frequency responses. The formulas for the Fourier method are oftenderivedwiththeLaplacetransformusingcomplex a .The Laplace transform is also used to make conversions between variable s = CT +j the time domain and the frequency domain responses. For brevity, we will write W(s)
52
)/3 (3.It is shownin Fig. ontheotherhand..l(b). A stepfunction or a series of stepfunctions is usually employed as the input testsignal.1) Fig. The 3dB bandwldth is the bandwidth of a lowpass svstem up to the fieauency where the gain coeficient decreases f i times. A telecommunication antenna. The feedback response required for best disturbance rejection is commonly specified in frequency domain since the disturbances are most often characterized by their spectral density. The timedomaincharacterization.e.conversionbetweenthe frequencyandtimeresponses and specificationsiseasilyperformed by computer. If not.itisimportanttobeabletomakethe approximate conversion mentally for the purposes of creation and analysis of specifications to systems and subsystems. resolution of the tradeoffs. rise time is avproximatelv onethird of the period l/fp related to 3 dB bandwidth. Givenamathematicaldescription of alinearsystem. This can be done using the simple rules described below. i. by 3 dB.exp(atr) = 0. and the output is specified in time domain.1 (a) Frequency response and(b) time response for firstorder link a/(s + a ) In other words. This rule is employed for calculating the bandwidth required for the rise time not to be longer than prescribed. and comparison of available versions of conceptual design.0’’ diameter dish.e.1 sec which translates .at) (see Section A2. a 1. withaccuracy. The time response of such a link to step function input is 1 . The timeittakesthesignaltoriseto 0.9 is found fromthe equation 1 .exp(.2inAppendix2). Example 1.. Analytical transformationbetween the timedomain function and the Laplace transform expression can be obtained in MATLAB by functions laplace and invlaplace. Therefore. to be placed on a balloon flyingin the Venus atmosphere. Frequencyresponses are widelyemployedforcharacterizinglinksanddesign specifications. in frequency domain. Frequency Response Methods 53 evenwhenweonlymeanthefrequency response W(jo). Highorder compensators and plants are also most often characterized by their frequency responses.The time and frequency responses can be plotted with standard MATLAB plotting commands(orwithSPICEsimulation). The rate of the attitude variations of the balloon can reach S0/sec. Weassumethereaderis already familiar with using frequency responses.l(a). 3..iscommonlyapplied to systems which are required to transfer signalswithout distortions.Yet. the rise time of the antenna attitude control system must be smaller than 0. 3. needs to be pointed to Earth OS0. 3. For the firstorder lowpass transfer functiona/(s + a). the 3 dB bandwidth is the pole frequency fp = al(2n) as shown in Fig.9 to be tr ( 1/f. Appendix 2 can be of help. i. t = 0 is The line tangent to the time response at at.Chapter 3.
setfling timet. increasing n flattens the timeresponse at small times and increases the delay time. and steady state (static)error. to the timeresponse at smaller times.3 Relations between the frequency At zero time. An important correspondencealso exists between the slope of the logarithmic gainfrequencyresponse(Bodediagram) andthetimeresponse curvature. by the gain over the 0.1 to 1 Hz frequency interval. Therefore.. the rise time is still roughly . the output at the time of 1 ms. 3. Le.1 1 0 time time time Fig. Frequency Response Methods into the 3 dB closedloop bandwidth of at least 3 Hz. the crossover fiequency fb > 1. 3. 3.4 3 0 . *u 3 0 .1) was derived but higherorder. the first n timederivatives vanish for the systems with an nthorder pole at high frequencies as follows from the initial value Laplace transform theorem (see Section A2. .4(b). overshoot.54 Chapter 3 .2 Timeresponse stepfunction to input domain time and domain regions time Fig. the rise time corresponds to the operational bandwidthand the settling time corresponds to the lowerfrequency gain. 9 error 0. Particularly. It is zero when the dc gain is one. etc. the timeresponse at the timeof 1 second ismostly affected by the gain coefficient at and around 0.1). in Fig. 3.3 in Appendix 2). by the gain over the1 kHz to 10 ldlz interval. as is indicated in Fig. even though the closedloop transfer function will be not firstorder for which (3. the delay timeincreaseswiththehighfrequency asymptotic slope of the gain Bode diagram. 3. or. approximately.characterized by thefive parameters shownin Fig. and the deviation of theoutputfromthedesiredstepfunctionis commonly . according to (3. The initial and the final value Laplace transform theorems relate the gain coefficient at lowerfrequenciestothetimeresponseatlongertimes.2. For a lowpass system with relatively smooth responses. Thesecalculations of therequiredfeedbackbandwidth are sufficiently accurate for the conceptual design.2: delay timetd . of settling within the dynamic error envelope. we cam assume that the transient response at specific times is predominantly affectedby the gain coefficient over specific frequency intervals. For higherorder lowpass transfer Eunctions. However.3 Hz. the cutoff frequency of the frequency response is approximated by (3. 3. 3.the Laplace transform gives the time responses shown in Fig. The static error corresponds to the dc gain. andthegain coefficient at higher frequencies. rise timetc.4(a). Numerically.5 Hz.1) where underf understood.3. Forthegain responses with constant slopeshown in Fig. Therefore. the transient response ismore complicated. all of them required to be small.
e.6 dB/oct.3 and 3.We can trace the correspondences shownin Figs. they render veryusefulleadsforsystemanalysisandcomputeraided iterating and tuning. or. Frequency Response Methods dB 55 Et . Then. (Note that each decade contains log210 = 2. from10 to 20. Notwithstanding the relations' imprecision.) We see some correlation between the slope of the gain response and the phase response: when the slope is zero.3 octaves. . from 30 to 60. 3. 3. The slope of the gain response becomes 12 dB/oct. 40 dB per decade..5. When the plant and the loop gain responses have asymptotic slope 18 dBloct. each octave hasthe same width on the logarithmic frequency scale.s 3 3 3dB/0ct I 0 I frequency. It is an integrator since the frequency w of the VCOisproportionaltoitsinputsignalbuttheoutputvariable applied to the phase detector is the phase.5(b).thetransferfunctiondegeneratesinto 9000/s2. the phase approaches 180'. the closedloop response also has this slope since at higher frequencies the loop gain vanishes. Athighfrequencies.Chapter 3. 3.. log. Example 2. which is the same. the VCO transfer function is Ms where k is someCoefficient that characterizes the VCO gain coefficient k h . the closedloop transient response at small times will be proportional to the third power of time.The slopes of their gain responses are. The output timeresponse to the step input for the same link isshown in Fig. . Le. Example 4. Example 3. i.intoa double integrator. when the slope of the gain response approaches 40 dB/dec. the phase is zero. 12 dBloct and 18 dBtoct.5 are. for example. when w increases twice (by an octave).4 on these responses. 1 2dB/OCt c . respectively. y / n v I / time (a) (b) Fig. Therefore. the slope of the gain response is .3 is a VCO. from 20 to 40. the gain coefficient decreases twice 6 (by dB). 3. we can roughly reconstruct the timeresponse at specific times.4 Correlation between the slope of gainfrequency response and the curvatureof timedomain stepresponse From the gain and the slope of the gain response at specific frequencies. sc. Thus. 3. The octaves in Fig. 1. The frequency response for the transfer function T(s) = (s 9000 + 30)(s + 300) is plotted with MATLAB in Fig. The plant in the PLL in Fig. There exist plants which are double integrators k/s2 and triple integrators Us3.
and its phase response is less curved. the first n derivatives of the gain response of the nthorder.2 ?ime(secs) Fig. At some moment.. tobe studied in Section 3. The higher the orderof a Chebyshev or a Butterworth filter. Le.6(a) and (b) show the gain and phase responses of several lowpass filters.the sharperis the gain responses selectivity.5 (a) Frequencydomain and (b) timedomain responses for T(s)= 9OOO/[(s + 30)(s + 300)] 3 .40which is the weight function of the Bode integral. . must be the same for all these components. 3..15 0.and its phase response is also significantly curved. Lowpass filters are most often employed for attenuating highfrequency noise and disturbances outside of the filter passband.. The Bufferworthfilter has maximum flat gainfrequency response. which is the group time delay. and (b) the dependenceof the filter phase shift on frequency must be close to linear. .05 0. their responses can be better understood from their similarity to the responses of the standard lowpass filters. Frequency ResponseMethods.1 0. the slope of this dependence. filter are equal to zero at zero frequency. When the shape of the signals mustbe preserved. The filter has a lesser selectivity than the Chebyshev filter. 3. The curvature of the phase response causes different delays for sinusoidal components of different frequencies andhas a profound effect on the overshoot. 1 . and the phase shift response is more curved. The phase response ofan ideal filter with 40dB selectivity is extensively curved (it follows the response in Fig. various signal components which are not in phase at the input. Fig. 10" 10' 10 ' Frequency (radlsec) 1oo 10' 1o2 1oa Frequency (radlsec) "0 0. then: (a) the filter gain must be nearly the same for all important frequency components of the signal.Le. come all nearly in phase at the output and cause the overshoot.6). 3. 2 Filters Since most feedback control systems are of the lowpass type. The gainresponse of theChebyshev(equiripple)filterbendssharply at the corner frequency.9.56 Chapter 3.
It is easy to recognize which of the three unmarked responses corresponds to ‘the highestorder filter: the one with steepest slope of the gain highfrequency asymptote. The phase responses do not look linear since the phase shift is plotted against the frequency axis with the logarithmic scale. For the 3rdorder lowpass Butterworth filter with normalized bandwidth 1 rad/sec.to fourthorders Bessel filters are the following: 3 15 105 s2 + 3 s + 3 ’ s3 +6s2 +15s+15’ s4 + 10s3+ 45s’ + 105s + 105 ’ The gain. or lineal phase filfeq is apprQximately proportional to frequency. 4 see. The transfer functions of the second. for the 3rdorder. The overshoot of the 8thorder filter is 0.k)! k!)]. for overshoot is 16% and the settling time 60 is sec. The higher the order of a Bessel filter.E E4 0 (a) (b) Fig.Chapter 3.e. the higher is the overshoot and the lohger is the settling time. and a Bessel filters The curvature of the phase responses manifests itself in the overshoots shown in Fig.7. the better is the phase response linearity.+bo where b k = (2n . For the settling error of the *settlingtime for the firstorder normalized &esse1 filter’is 11. 3. the accuracy is 25 sec. the overshoot is 8% and the settling time to the 8thorder filter. and the smaller are the overshoot. The filter transfer function is B(s) = bnsn bn..and stepresponses for the three Bessel filters are shown in Fig. The higher the order of a Chebyshev or a Butterworth filter. 8 see.5 sec. by 1 sec).sn’ + b 0 +.. 7 Time stepresponses for the filters The phase shift of the Bessel filfer (or Thompson. 3. and the one with largest time delay at small times..35%. 3. f l % random variations of the denominator coefficients of the 8thorder Bessel filter transfer function increase the settling time up to 5 sec (i. the one with largest negative phase at higher frequencies. the rise time.. Example 2 . 3 . Example 3. 8 0 0 Fig. and the settling time. Frequency Response Methods 57 0 . . Example 1. and for the 8thorder.k)!/[2nk(n .+bk sk+.8. a Butterworth.6 Frequency responses of the (a) gain and (b) phase for the step40 dB lowpass filter and for a Chebyshev..
9 (a) Nyquist diagram.and the closedloop gain responseloglMl 20 has a hump as shown in Fig.e. 3.and stepresponses of Bessel filters of second to fourth orders 3. IT1 = 1.. 3. 3.)li(fb)l c 1. 3. the angle 180".arg Tvb) is less than 60". From the isosceles triangle shown in this picture.3) "he hump islarge when arg Tvb) is closeto 180".8 Frequency.arg T)/2]}. as aresult. At crossover frequency fb.. Commonly. Frequency Response Methods Frequency (radlsec) Time (secs) Fig. arg indicates the angle in degrees). i. its closedloop gainVMI = IT/FI becomes greater than 1 atfb . and the value of this maximum maxVMl= kf(fb)l= 1/{2sin [(180". ImT I dB1degr 10" 120 15" 180 Fig.9(a).ocb)l= b(3.9(b). (3.e.Whena tracking system has no prefilter.thefeedbackbecomespositive. .whentheNyquistdiagram approaches the point 1.58 Chapter 3. and. i.2) i n [( Tfb))/2] 180". (b) closedloop gain and phase responses The maximum of M I is commonly at a frequency somewhat smaller than fb.arg (here.2 Closedloop transient response The Nyquist diagram in the neighborhood of the critical point 1 is shown in Fig. II.
3. Commonly.e. the disturbance is not measured or observed and only 'the error is measured.therefore.tion to the target. then. or with a command feedforward path).2.e. i. it should at least in average make the closedsystem phase response linear.. However. If the phase stability margin at fb is small.1 and 2. i. the transient response of the output becomes too oscillatory.e. This will prevent mostoftheharmonica1componentsofthe signal from reaching the output in phase at any time. Homingsystems do nothavecommandsummersand. As long as the pole is in the left halfplane of s. The resulting overshoot is about 50% as shown in Fig. certainly. all poles of their transfer functions are in the left halfplaneof s. When the linksin the loop are each inherently stable. The prefilter must therefore equalize the hump of the closedloop response from the summer to.some components of the random input noise will be exponentially magnified causing the system's output to grow exponentially.10. the closedloop gain response (together with theprefilter) must approximate a Bessel filter response.1 0 Closedloop In a feedbacksystem with aprefilter(or transientresponsetostepinput with a nonunity feedback path.3. then MI = 2. The closedloop transfer function from the disturbance to the missile direction is 1/F. larg T(fb)l does not exceed 235' in such systems. or thecommand feedforward link) cannot be implemented to be exactly optimal. and there is no need to compromise the loop response. all poles of the transfer function . 3.3. 3. this doesn't create a problem.. Verification of the system stability is one of the major tasks in control system design. Then. If the prefilter (or the feedback path. Such a linear system is considered unstable and cannot perform as a control system. For the overshoot in a closedloop feedback system to be small. in order to reduce the closedloop overshoot. the transmission is infinitelybigforthesignal which is growingintime. the response of interest is that of the missile direction to the disturbance which is the changing direction to the target caused by the target motion. and the output to settle with high accuracy in a rather short time. then I F 1 decreases to 0 and iMI grows infinitely. when one of the poles has a positive real part. the frequency response of 20 log IN has a large hump.Chapter 3. do nothave prefilters. 5 4 1 0 t P m e 5 10 Fig. Frequency Response Methods 59 Example 1. the openloop and closedloop responses can be optimized independently of each other (as was shown inSections 2. When the angle of T approaches 180°. andthehumpis 6 dBhigh as shown in Fig.3 Root locus The transfer function T(s)/F(s)of a closedloop system is infinitely big for the signal components corresponding to the poles of the function.the output.and there exists an effective maneuver for the target to.2). Typically. avoid being hit. When the angle is l5Oo.. i. For a homing missile. Example 2. The homing system openloop response must be made such that the closedloop transient response be as desired.9(b). the pole spis in the right halfplane of Laplace variable s. must incorporate a broad notch. the differencebetween the missile direction and the direc. An example of such a prefilter is givenin Section 4. .
i. of T(s)/F(s) = T(s)/[T(s) + 13 crossesthe joaxis at a certain Whenapole frequency.42. the system is unstable.1 0 . rlocus(n. roots (d) %openloop poles 0. ITI = 1 and arg T = d m .05 0.The root locuswill be further discussed in Section 8. where n is an odd integer.= 0 roots(n + d) %closedloop poles ans = 1. ! 3 2 1 Real Axis 0 1 Fig. the pole at the origin moves to the left and approaches 1. andwe will continuously increasek from 0 to 1. d = [l 1 1 01. the where n closedloopastem is stable ij“I7J < 1 at allfi?quencies atwhich a. when arg T = h.1.rg T = &am is odd.3275 f 3. After that. and the openloop system is stable. the complex poles migrate to the right halfplane of s andthe system becomes unstable.2 0. this frequency becomes From here. T(s) = 1O(s + 2)/(s3+ s2 + s).2 0. Correspondingly.11 Root loci for T(S) = IO(S + 2)4s3+ t+ + 6).5000 f 0. The loci of the two complexpolesendat 0. Le.5 1 1 ’ ) . rlocus (n.5 11. The rules provide for simple loworder system analysis but the methodbecomescumbersome when appliedtohighperformancesystemswhichare highorder systems. Example 1. we place a linkwith the gain coefficientk in the loop. k) title(‘k = 1 0 . which can only happen if T(jo)at 1. ans . The root loci start at the poles of T. 2 0.2. when the loop is “gradually” closed: instead of switching the loop open or closed. The system is stable with the coefficient k up to 0. when the complex poles become purely imaginary.. n = [0 0 1 02 0 1 . a simple stability criterion follows: if an openloopflstern is stablg. There exist rules for drawing the root loci manually and for using the loci for the feedback system design.6551 0. 3 11 are plotted with: n = [lo 201. The openloop and closedloop poles can be calculated with the MATLAB commands d = [l 1 1 01. Their trajectories on the splaneare called root loci.86603. and in this case.3275 rtj3. in practice. A necessary and sufficient stability criterion based on the openloop frequency response will be derived in the next section. this function at this frequency becomes 00.Invasion of a root locus into the right halfplane of s indicates that the system becomes unstable. This criterion is convenient but it is not a necessary one.6551. .. When the loop is closed.05 0. d) hold on k = [0. hold off en . Frequency Response Methods T(s)are in the left halfplane of s.1 0. 0 5 0. ktj1.4608.1 0.5 11 5 The root loci in Fig.e.60 Chapter 3 . 3. the poles of the transfer function will change continuously from the poles of the openloop system to the poles of the closedloop system.4608i k3 : [0. some of the poles of the transfer function T(s)/F(s)can appear in the right halfIt is interesting to trace what happens plane of s.x E  As k increases and approaches infinity. d.
To find the number of zeros of F(s) in the right halfplane of s.It allows passing a judgment on whether the system is stable by observing the plot of the openloop frequency response. related to zeros outside the contour ci It is easy to notice that the vectors exercise no such revolutions. the vector s . 3.The rest of the zeros of F(s) are outside the contour.s i ) ( s . This is the condition of the closedloop system instability.. it makes the vector F(s) complete a clockwise revolution about the origin as shown in Fig. i.sk.SJSsq)(s . 3. Therefore..* are s i . The zeros of the function ( s . . S shown in Fig.12(a). measured or calculated. 3'.13 (a) Revolutions of a rational F(s) caused by (b) the trip of s aboutfunction F(s) causedby(b) s having a closed contour Q on the splane moved about a closed contour c The phase of the vectorF(s) is the sum of the phasesof its multipliers. i.s j ) ( s . Let us derive a condition for one or several of the zeros to appear in the right halfplaneof s. a revolution of a multiplier about q changes the phase of F(s) by 2n.q changes by 2n. as shown in Fig.. The most convenient among them is theNyquist criterion. sk.e. . Therefore. the number o f clockwise revolutions o f & locus F(j@ about the orinin indicates the numbers of zeros within the contour c. Frequency Response Methods 61 3. 3. I Fplane Fplane (a) Fig.4 Nyquist stability criterion Stability criteria allow verification of system stability without pole position calculations or experiments. s . 3. the transfer function T(s) does not have poles in the right halfplane of s.s ~ ). the contourc should . We will consider a singleloop feedback system which consists of lineartimeinvariable links whose transfer functions are rational with real coefficients.S j ..Sr) . While s makes a clockwise round trip about the contour Ci.q shown in Fig. j .12(b) completes a clockwise revolution. the argumentof each multiplier of the kind s .Chapter 3 . Consider a simple closed contour ci in the splane crossing neither poles nor zeros of F(s) and encompassing no poles and one zero q as shown in Fig.. 3. Therefore.13(a).e. 3.13(b).. F(s)= (s .12 (a) Revolution of a rational function Fig. Sk as Consider next a Contour c encompassing no poles and Several zeros Si.12(b). We assume that the system is stable when the loop is disconnected. the closedloop transfer function T/F can only have poles in the rig& halfplane o f s i f some of the F(s) zeros are in the right halfplane o f s. s S j ... When s makes a full trip about e..
srnal1value lumpedelements. Frequency Response Methods envelop the right halfplane. the return ratio disappears in physical systems when s is big.The diagram makes half the number of revolutions of the whole locus which reflects the existence of zero si in the first quadrant of the splane. For this particular diagram. Therefore. this particular closedloop system is unstable. Therefore. ' 2 The number of revolutions of the locus of I . it is valid as well for transcendental transfer functions since transcendental transfer functions can be closely approximated by rational functions. F(s) becomes 1 while s moves along the infiniteradius arc. 3.theNyquistcriterioncanbeappliedtoallpractical systems describedby calculated ormeasured gain and phase frequency responses. of s makes The function F(s) is rational with real coefficients. 3.] AlthoughtheNyquistcriterionhasbeenprovenhereforonlyrationaltransfer functions. Such a contour can be made of the jcuaxis and an infiniteradius arc as shown in Fig.14(a) is the mapping of the jmaxis. F($)= F(s). Thus. For instance. the locus in Fig. Therefore. the locus of F consists of two imagesymmetrical halves relating respectively to positive and negative frequencies. w is shown by the thicker curve and is The part of the locusof F drawn for positive called theNyquist diagram. The contour encompakses no poles of F since the ' + 1 are the poles of . ( @ ) poles of F = 2 shown in Fig. si and S j . complex conjugate. does not encircle the origin u f the Fdune. in the right halfplane of s. and the right halfplane s of maps into the inside of the contour on the I.14(a) about the origin gives the number of zeros of F(s) in the right halfplane of s. Fig. the transcendental pure delay function in a mediumwithdistributedparameterscanbe approximated by a rational transfer function describing a system with many. reflecting two complex conjugate zeros.plane. the number is 2. it is stable with the loop closed if and only if the Nvauist diagram for I. The Nyquist criterion follows: l f a linear mstem is stable with the feedback loop open.62 Chapter 3. . Complex conjugation I.14(b). 3. 3.14 (a) Nyquist diagram for Fand (b) the contour surrounding the right halfplane of s As mentioned before.Therefore.
16 assumes that plant parameter variations. Variations in the force and in the plant's mass change the plant gain but not the plant phase shift. is instead of the / q o origin.in gain and in phase are not correlated.and/orreducingthe loop phase lag at certain frequencies.e. where the loop gain changes up to 10.1 as shown in Fig.16(a).and (b) Lplanes The shape of stability margins shown in Fig.. 0 dB) inthe ontheLplanetypicallypassestotheleft counterclockwise direction. On the other hand. Most importantly. 1. The Stability margins guard the critical point. 3. the diagram of the critical point (180°. 3. Notice thatwhen a Nyquist diagram on the Tplane encircles the critical point (1. the system will remain stable for a certainkange of variations of the plant parameters. or.19. which means the same. deviate from the nominal values. but also how to make thesystemstable by reshaping the loop response (by changing the Fig.16 Amplitude and phase stability margins on (a) the ?.Chapter 3. 3.000 times without . 3. the Nyquist diagram tells us not only whether the system is stable or not. Consider. The unstable critical point in this case 1. Lplane I gain (a) (b) Fig. As mentioned in Chapter 1. a force actuator driving a rigid body plant.16(b). by rectangles on the Lplane in Fig. 3. for example. i. 0 dB). by a segment on the Tplane. variations of small flexibilities in the plant change the phase shift but not much the gain in the frequency region of the crossover. Loop shaping will be described in detail in Chapter 4. They are often formed as shown in Fig.14(a)).15 Nyquist diagrams compensator): by reducing the loop gain over a on Tplane specificfrequencyrange. Nyquist diagrams can be plotted also on the Gplane. 3. where the critical points are (180" f n360". 3. If the Nyquist diagram for the nominal plant does not penetrate the boundary of the stability margin. Another example is the volume control depicted in Fig. This is typical for many practical plants. remaining stable when the plant parameters.O) in the clockwise direction.15 (compare ?plane this plot with the locus in Fig. 3.5 Robustness and stability margins Practical systems are required to be not only stable but also robust. and consequently the return ratio. Frequency Response Methods 63 The Nyquist diagram is commonly drawn for T = F .
to remain stable after any set of initial conditions. i.) As illustrated in Fig. at least. and the values of the phase stabi/ifymargin y180" are typically 30' to 45".. the gain stability margin is preserved and the system is gainstabilized. The values of the lower and upper amplitude stability marginsx. and over the frequency range where the angle of T is within the interval [MOO( 1 + y).17.No limitcycle conditions. 3. 180"( 1 . and either gainstabilized or phasestabilized between these frequencies.17(b) shows the Nyquist diagram of a system which is both phase.and gainstabilized at frequencies belowf. and above&.18 Nyquist diagramsfor (a) Nyquiststable and (b) absolutely stable systems Practical systems all include nonlinear links. Frequency Response Methods any change in the loop phase shift.20(c).18(a)) at which the loop gain is larger than 1.is less suitable for practice.64 Chapter 3. The disk stability margin. 3. 3. over the frequency range where the loop gain is within the interval [x. (These values are also sufficient to guard from some nonlinear phenomena in the control loop.y)]. 3. studied in Chapter 12. and are required to be globally stable. 3. negative axis to the left Such systems are only gainstabilized (not phasestabilized) at some frequencies (such as fi andf2 in Fig. as in Fig. 3. Examples are shown in Fig. the saturation of the actuator. (These issues will be discussed in more detail in Chapters .) allowed.11. 3.e.16 are typically 6 to 10dB.17(a) and 3. conditions of periodicoscillation. 3.18(a). Tplane gain stabilized Fig. ' (a) (b) Fig. Fig.e.17 Gain and phase stabilizing in (a) the Tplane and (b) the Lplane Nyquist stability relates to a stable system whose Nyquist diagram crosses the of the point 1. the phase stability margin is preserved and the system is phasestabilked. x1 shown in Fig. i.xl]. are 9 ..
” While shrinking. 3.) One of the Nyquist criterion’s advantages is the simplicity of estimating the effects of multiplicative variationsof the loop gain coefficient.3. At these specific frequency and signal level. (UsingNyquiststabilityisacceptableandevenbeneficial if specialnonlinear dynamic links are introduced in the loop to exclude the possibility of selfoscillation. 3. then. The absolute stability notion relates to systems with a saturation link and whose Nyquist diagram is like that in Fig. AccordingtotheNyquistcriterion. For the absolute stability. Frequency Response Methods 65 When the only nonlinear link in the loop is the actuator. T is real and less than1 at two frequencies.19. Example 1. . the return signal at these frequencies is in phase and of a larger amplitude than the signal applied to the loop input. In Fig. Designing such nonlinear links is described in Chapters 10 and 11. and that increasing loop the phase delayby 60: at the frequency where IT I = 1 (the crossover frequencyfb)will also cause oscillation. 3. Fig. IT + 1 at the point 1. W m q u i s t stabiliv should be avoided in those feedback systems which have no nonlinear linksother than theactuator saturation. fi andf2. the stability margins are typically chosen as shown in Fig. Therefore.18(b).20(a) that reducing the loop gain without changing the loop phase shift will not make the system oscillate. its saturation may reduce the equivalent loop gain while retaining the loop phase shift.Chapter 3.and Lplanes Absolute stability will be studied in Chapter 10 in a more precise’ manner.The 1 < 1) if and only if T is on the unit radius disk centered feedback is positive (i. This causes the equivalent Nyquist diagram to “shrink. andwaslaterproventheoretically by H..e.19 Stability margins for a singleloop system with saturation on the T. 3.20(a)(e) arestable. i. that increasing the gain by a factor of 2 will cause oscillation.20(c). That a feedback system with this sort of T isstable was first foundexperimentallyduringdevelopment of feedback amplifiers at The Bell Laboratories.whilesystem ( f ) is unstable. it is seen from the diagram in Fig.e. loosely speaking. the diagram of a Nyquiststable system crosses the point 1.. For example.thesystems withtheNyquist diagramsshown inFig. This system is Nyquiststable. In other words. theequivalentreturnratiobecomes 1whichistheconditionofselfoscillation. not crossing the critical point while shrinking. Nyquist. 3.
Fig. (e) nonrobust system.20 Nyquist diagrams: (a) feedback is positive at frequencies at which the Nyquist diagramis on thedisk. The inventor of feedback amplifiers H.21. Another example of using these coordinates is given in Appendix 13. 3.66 Chapter 3. I Fig. (f) unstable system Example 2.26. 3. (d) robust system. The linearscaleisinconvenientfordrawingNyquistdiagramsfor practical systems. Frequency Response Methods Fig. (c) disk stability margin. A13. where the loop gain changes by several ordersof magnitude between the lowest frequencyof interest and the crossover frequency. and phase in degrees. 3. (b) Bode stability margins. as illustrated in Fig.21 Nyquist diagram with logarithmic magnitude scale . Black employed a circular coordinate system with gain in dB.
However. 3.6 Nyquist criterion for a system with an unstable plant The systemsdiscussed.e. and this concept is referred toas “guardpoint” stability margins.. (e) nonrobust system. 3. Frequency Response Methods 67 Example 3.until now were considered stable with the feedback loop disconnected. The puardDointphase margin is the phase margin at the freauency where the gain margin is zero. 3.22(e) . We would like to know what feedback will make such systems stable. (d) robust system.14(b). if highorder a compensator is optimized for closedloop performancewhileonlytheguardpointstabilitymarginsareenforced.20(e) and 3. T(s) has poles in the right halfplane of s. Now. The vertical axis corresponds 180’. i. This sort of misuse of the guardpoint stability margins may be responsible for the misconception that highorder compensators “generally” lead to nonrobust systems. 3. Because of the inescapable tradeoff between stability margins and the available feedback. This interpretation is acceptable and convenient when the order of T(s) is low. 3.22. but it does not suffice in general. consider the functionF(s) = T(s)+ 1 which may have poles and zeros within the contour in Fig. stability margins are defined as if they need only applyat discrete points instead of a solid boundary. and even when the order of T(s) is high but the response is smooth. to Fig. some physical systems are unstable without the feedback. In some of the contemporary literature. Having a pole within the contour causes the plot in (a) to rotate in the direction opposite to the revolution produced by a zero within the contour .22 Nyquist diagrams on the Lplane: (a) negative and positive feedback areas. The Lplane diagrams equivalent to the corresponding Tplane diagrams of Fig.these loop responses approach the critical point much too closely.. The guardnoint gain margin is the gain margin at the freauency where the phase margin is zero.e. 3. at fb. (b) Bode stability margins. (e) disk stability margin. i. (f) unstable system Example 4.Chapter 3.20are shown in Fig.theNyquist diagram might end up looking like Figs.
(e) Lplane Nyquist diagram for T 2 Because of the internal feedback. The Nyquist diagram for the internal loop encloses the critical point clockwise as shown in Fig.8s + 4)s2 possesses a pair of complex conjugate poleswith positive real parts. Frequency Response Methods (b). and these encirclements must be in the counterclockwise direction.23(a). the N y q a &ram must encircle the critical point the number of times eaual to the number of ovenloov poles in the first Quadrantof the splane. 0. 3.2s+. The rule follows: In order for the system to become stable when the feedback is applied. where the links characterize physical processes and relations between the variablesintheplant.0.not to scale. thermal flutter.1d(B+0.23 (a) Feedback system with an unstable plant. In the system diagrammed in Fig. Example 1.This method is routinelyappliedtotheanalysis of various unstable plants: plants with aerodynamic instability. With the . gas turbulence. the plant transfer function P(s) = 8 0 0 (s2 + 0. Consequently. (b) Nyquist diagram for the internal plant foop. (d) Nyquist diagram for TI on the Lplane. as can be judged by the negative sign of the damping coefficient in the denominator polynomial. the plant is unstable since the internal feedback path transmits the signal back to the input of the gain block in phase at the resonance frequencyo = 2 of the feedback path. 3. The corresponding Nyquist diagram encircles the critical point a number of times equal to the difference between the number of poles and the number of zeros in the first quadrant of the splane. (e) Nyquist diagram for TI.when s completes a trip about the contour.68 Chapter 3. It is cominon to describe the plant's instability as the resultof an internal feedback loop. number the of revolutions of the function locus equals the difference between the numbers of zeros and poles of F(s) within the contour. 3. and the inverted pendulum. Consider the following two examples.23(b).4) (s2 .2s+ 4) tt Plant dB 40 20 0 180 I * 90 0 90 1801 Fig. wind flutter.
3. Consider the multiloop system shown in Fig. When the amplifier is operated with the conveptional feedback path B = R1 I(R1 + R2) closed.24(b).1 as indicated in Fig. The highgainamplifiershowninFig.23(c) encircles the critical point 1in the counterclockwise direction. Frequency Response Methods 69 compensator transfer functionshown in Fig. 3. the parasitic feedback loop gain is small. and (c) Nyquist diagram for the parasitic feedback loop 3. 180') in the clockwise direction. (b) Bode diagrams for the feedback paths. the system can be analyzed without taking into account the parasitic feedback. the main loop transfer function at crossection (1) is q 0)= 8oo(s + l0)(s2 + 0. 3. 3.2s + 4) (s + 40)(s2 .25. 3. 3. 3.7 Successive loop closure stability criterion Example 1. The locus on the Lplane shown in Fig. with which the amplifier is stable.24(b)). The system is stable. that the parasitic feedback B .23(d) correspondingly encircles the critical point (0 dB.8s + 4)s2 The locus of TI(@) shown in Fig. .Chapter 3.23(e).24 System with parasitic feedback via capacitor C: (a) schematic diagram. Example 2. The Nyquist diagram for the parasitic loop is shown in Fig. the amplifier gain is much smaller (say. where the loop transfer function is foundbe to T 2 (s) = "0.0. dB 0 20 dBI Lplane Fig. Another wayof this system stability analysis is to look at the crossection (2).24(a)wouldbestable by itself. Another point of view on the problem is. and the system is stable.23(a).butsmallparasiticinputoutputcapacitance C causesselfoscillation inthe absence of the standard feedback loop via R2. 3. Assume the system is stable when all loops are System Linear Fig.25 Block diagram of a multiloop system . 3. The transfer function for the parasitic feedback path is shown in Fig.24(c). 3. = s/[s + l/(RC)] is negligible when compared with the normal path transfer function feedback path transfer function B. only 10 when B = 0. Hence.1~ +~76s3+960s2 + 1920s+ 3200 s5 + 42s4 + 84s3 + 160s2 The Nyquist diagram for this crossectionshown is in Fig. This indicates that the closedloop system is stable. 3.
27containstwolocalloopsand a common loop. Frequency Response Methods disconnected. after the fourth loop is closed.second. and third. Fig. after the second loop is closed. Thus. after the third loop is closed. it is stable with these closed if and only if the total numbers of clockwiseandcounterclockwise encirclements o f the point (1. drawing three Nyquist diagrams is required for the stability analysis. Bode formulated the Nj/@st criterion for mulfiloop systems as follows: @s When a linearsystem is stable with certain loops disconnected. and start closing the feedback loops successively which will eventually lead to the system with all the loops closed.70 Chapter 3. A series of five such Nyquist diagrams is exemplified in Fig. The systemdiagrammedinFig. 3.the critical point.drawingthe Nyquistdiagramforthecommonloop(withlocalloopsopen). the system remains stable.O) are equal to each other in a series of&quist diagrams drawn for each loop and obtained by beginning with alll o o p m e n and closing theloopg successively in any order leading to' the svstem normal configuration. The order of closing theloops andmakingtheNyquistdiagramscanbethefollowing:first. (a) Fig. and the system remains stable after the fifth loop is closed since the fifth Nyquist diagram does not encircle the critical point. 2 7 Stability analysis with (a) three and (b) two crosssectionsof the feedback loops .26. the system becomes unstable and the system transfer function possesses one pole in the first quadrant of the splane.closingthe common loop and drawing the first local loop diagram. 3. By generalizing the procedure given in the Example 1. 3. we would have to draw and to analyze all three diagrams since it is not evident than any of the three diagrams avoids encirc1ing.3. 3. of the Example 2 . In this case. the system remains unstable since the Nyquist diagram indicates no change in the difference between the numbers of poles and zeros in the right halfplane of s. the system becomes stable since the diagram encircles the critical point once in the counterclockwise direction. The system is viewed as a threeloop system in Fig.27(a). drawing the Nyquist diagram for the remaining local loop.26 that after the fEst loop is closed.26 Nyquist diagrams for the successive loop closure stability analysis It is seen in Fig. The order in which the loops are closed can be chosen at the convenience designer. 3 .
a single integrator l/(Ms). When this is the case. that the local loops are inherently stable because the phase lag in the loops is. Dependingonthetype of thesensor a rate employed . 3. it is worth starting with these loops known that the diagrams will not and not drawing the related diagrams since it is already enclose the critical point.first local loop is stable by itself. Alternatively. The frequency responses for these plant functions are shown in Fig.8 Nyquist diagrams for the loop transfer functions with poles at the origin echanical rigid diagrams a 3. whenthe . needs to be drawn and analyzed. with the cross section (1) closed. the order of closing the loops can be chosen such that it reflects the order in which the loops desaturate when the signal level gradually decreases [91. with the local loops closed. 3. however. Further. 3. or a position sensor .and doubleintegrator plants are at the origin of the splane as shown in Fig. Nevertheless. 3. dB 0 Fig.28 Mechanical rigid body plant sensor. it is convenient to start by closing the cross section (1) without drawing the related Nyquist diagram since it is known that the diagram should not encircle the critical point. The gain for the single and double integrators is infinite at zero frequency. therefore. These cross sections eliminate all feedback loops.an accelerometer. Fig. the system can be analyzed using two cross sections in the forward paths as shown Fig. a single integrator. the simplest approach might be inconsistent with the convenience of verifying global stability. or a double integrator ~/(Ms)~. Force x X 1 s " ~bllM% 11s " b1 bodyplantdriven by a force actuator. only the Nyquist diagram for the cross section (2).Chapter 3 .27(b). It remains.29(b).28 Fig.29(a). and the analysis can be performed using only two Nyquist diagrams. 3.the plant transfer function is correspondingly a constant 1/M. The poles (the singularities) for the single. Then. Frequency Response Methods 71 It is often evident.29 (a) Bode diagrams for the plant as a constant.small. For this purpose. and a double integrator and (b) the related contour on the splane . 3. only to draw the diagram for the common loop.
to calculate the angular velocitys12 and the accelerationa.30 Nyquist diagram for a stable feedback system with (a) a single integrator in the loop and (b) a double integrator in the loop The feedback control loopsare often classified as servomechanismsof Type 0 (or servo type). i. 3. and Type 2.. the small arcin the splane maps onto the Tpane as an nn: arc of infinite radius.e. if the transfer function T(s) possesses poles on the joaxis. Nyquist diagrams of Type 0 are shown in Fig.. the arc is twice longer as shown in Fig. 1 1 1 is infinitely large. 3. 3. Half of this arc becomes a part of the Nyquist diagram. and the diagram shown in Fig. The Nyquist diagram for a singleintegrator plant is shown in Fig. The properties of thesethreetypes of servomechanismscanbeunderstood by considering the feedback system shown in Fig. Frequency Response Methods The Nyquistdiagram is themapping of thecontourencompassingtheright halfplane of s.18(a) and (b).31. 3. the origin can. The plant output variable is the angle 0. Fig.30(a). and its phase changesby nn: as the phase of s changes by n: along the arc.30(b). 3. The shape.of the Nyquist diagram and the related loop frequency response at lower frequencies define the steadystate response to commands and the steadystate error reduction. in the close vicinity of the pole. For a doubleintegrator plant. This number is the number of poles of T(s) at the origin. Therefore. 3. Therefore. The contour should be chosen such that the function on the contour does not turn into infinity. Particularly. when n poles are at the origin.72 Chapter 3 . Twodifferentiators are addedaftertheplant for thesake of analysis.30(a) is of Type 1. . the contour should avoid the poles. 3. for a loop transfer function with a double pole at the origin. Type 7. 3.be avoided byan infinitesimalradius arc as shown in Fig.A double integratoris added to keep the return ratio unchanged. On this arc. Le.29(b) is of Type 2. The Nyquist diagram shown in Fig.29(b).
by an extra velocity component of moving parts of the plant or by drift in time of the values . etc. i. the dc gain coefficient (Le.32. and the disturbance in acceleration.when constant angle 6 is commanded. by disturbance torque due to wind.Chapter 3.32 Disturbances in angle. In Fig.. specific commands ul(t) need to be applied to the input. andwhenconstantaccelerationiscommanded. magneticforces. or $ 2 .In order to do this. consider the effects of disturbances entering the feedback system at different points: disturbance in position. the return signal nearly equals the command.31 (a) Feedback system block diagram and (b) the profiles of the command Assuming theerror is small. Frequency Response Methods 73 for constant accelerartion time Fig. The Type 0 system has finite loop gain at zero frequency. the error is the angle error and it is zero. and constant acceleration commands.theforwardpathgain error to the velocity and to the' acceleration s at + 0 are infinitely coefficients from the small. constant angle.of some of the plant's parameters. constant angular velocity. On theotherhand.." S S e Fig. Disturbance in position is commonly caused by misalignment of mechanical parts. and acceleration in a feedback system InaType 1 system.31(b). disturbance in velocity. Therefore. If in the commanded ul(t). the output will not track the command. 1/52. 3. velocity. but this disturbance causes a: finite change in the . Next. 3. To reduce the static error of the controlled variable. disturbance in velocity. respectively. The Type 1 system has an infinite loop gain coefficient at zero frequency. and the gain to acceleration a is zero. let us consider the problem of keeping constant one of the variables: 8. the gain coefficient at s + O)'fiom. 3. and disturbance in acceleration as shown in Fig. the steadystate summer output to the plant output (from error of the' angle 8 is smallbutfinite.the error to the controlled variable must be largg. or a. the velocity error is finite. Therefore.adisturbanceenteringatthepoint of velocityisreduced infinitely since the feedback is infinite. but the gain to velocity R is finite. The gain from the error to 8 is also infinite.theaccelerationerror is not corrected at all. and the dc gain from the errorto 8)is finite. 3.e. the velocity or the acceleration is constant. when constant velocity is commanded. the timefunctions are shown for.
employactuatorsandplants withm. an error that does not decay in time). but the lowfiequeacy phase lag in such systems approaches 2’70’ and the system is not absolutely stable. To eliminate this error.. the constant velocity disturbance causes a “hangup” error in position (Le. transferfunctions. transfer functions of stable systems have neither zeros nor poles in the right halfplane of s . A logarithmic transfer function can be presented in the form e(s) = A(s)+ B(s). Le. In some systems. ‘ ’ Example 1. and the system is still unstable. he might find out that this gain change affected the phase shift at other fiequencies. and to add extra phase lag (but not phase lead!) without affecting the gain.p. these functions are . The phasegain relations inm. if possible. Therefore. H. will Section 3. This greatly simplifies the search for the optimal design solution. of s. The Type 2 servomechanism is also referred to as a “zerovelocity error” system. Designers of controlloopsuse m. it is desirable for the transfer functions of the feedback loop links to “be m.then A becomes the real part of the function. its impedance and admittance have no zeros or poles in the right halfplane mop. Thus. although with a quite different shape of the Nyquist diagram. This extra phase lag. W.tocorrecttheshape of theNyquistdiagram.if. The Nyquist criterion uses three variables: frequency.one decided to reduce the gain at some frequencies and did so. These variables are interdependent. For such a system to remain stable after the actuator becomes overloaded.1 Minimum paiase functions Synthesis (design) of a stable feedback system using the Nyquist criterion is not quite straightforward. It will be demonstrated later in this chapter that m. The real and imaginary parts of e(@) are related. When s is replaced by ju. and phase shift.).74 Chapter 3. where A($) is the even part of the functionand B(s) is the odd part of the function. Therefore. the steadystate errors in position and in velocity are zero. Frequency Response Methods angle since the gain coefficient at dc from the angle to the velocity is infinitely small. Therefore. and there is finite reduction in the steadystate error in acceleration. It is always possible to add a constant to the gain without affecting the phase. A passive twopole is stable in the conditions of being open or shorted.p. functionsincompensatorsand. transfer functions have no zeros in the right’halfplaneof s. loop gain.9. the compensator 913.p. although not in a unique way. be considered in more detail in called nonminimalphase (nap.p. the phase delay in the feedback loop limits the available feedback. These systems have larger loop gain and better accuracy at low frequencies. As will be shown further.9 Bodeintegrals 3. must be made nonlinear. In this system. as will be discussed in Chapters 3.12.Forexample.p. Le. functions areof special interest for feedback system designers. Therefore. m.p. the gain. the phase shift. logarithmic transfer function of stable systems are analytical in the right halfplane of s. a Type 2 system should be employed.) lag..p. so that m. Functions without such phase lag are called minimum phase (map.. and B becomes the imaginary part. Bode showed that in most practical cases usingonlytwoofthem(thefrequencyandthegain)sufficesforfeedbacksystem design. the return ratio has a triple pole at zero frequency.
33 Ladder network output voltage input to current) and the transadmittance (ratio of the output current to the input voltage).33 shows a ladder passive electrical network. the minimumphase functions(meaning minimumphaselag). All of them are minimum phase functions. and vice versa.6) Now. which also have no zeros on jwaxis. so the network transfer function does not have zeros in the right halfplane of s.2 Integral of feedback Consider the function6(s) which has no singularities on the border and inside the right halfplane of s and is limited at high frequencies and therefore can be approximated for large s by the series By integrating 6(s) about a contour enclosing the right halfplane of s (see Appendix A4. when s is large. their poles and zeros are not in the righthalf plane of s.1). w e e d b a c k integral over the freauencymion where .6) we see that B1= 0.and currenttransfer functions.)dW=.Chapter 3. It is evident that a signal applied to the input of a ladder network arrives at the output unlessat least one of the series branches is open. Frequency Response Methods 75 Example 2 . by comparing this expression with (3. Hence. 3. assume that return ratio T(s)of a system that is openloop stable turns into als" for large s. 3.34. In the mostcommon case of n 2 2.9. When the feedback is negative.Therefore. the transimpedance (ratio of the Fig.Fig. and the integral of feedbackis zero: IFI > 1 and therefore In IFI > 0. Then. The outputtoinput ratios of the network can be expressed as voltage. as illustrated in Fig. Bode proved that (AA. where Q is a coefficient.Heprovedthatthephasedelayofthe minimumphase function is the smallest among the transfer functions with the same gain frequency response and is thus uniquely definedby the transfer function gain frequency response. 3. or one of the parallel branches is shorted. Bode named such functions. When the twopoles are passive. B 1 0 2n (3.and let 6 = In F.theoutputtoinputratiozeros are produced by poles of the impedances of the series twopoles and by zeros of the impedancesof the shunting twopoles. 3. 8(s)=ln(l+T)=ln1+ [ f) f =.
the Nyquist diagram should follow the stability ins' boundarv as closelymossible. 3. Therefore.5) frequencies. In practice. this distance should be kept minimal over the bandwidth of positive feedback. at higher frequencies gives B1 = 1/C.35. 1. Comparing this formula with (3.35 Twopole 2 1 ' shunted by capacitance C . the accuracy of loop shaping in the crossover area is of extreme Wortance forachieving maximum negativefeedbackoverthefunctionalfrequency bandwidth. where . Corollary 1.5' is assumed not to reduce to zero at infinite frequency and to be limited at all Z = l/(jwC).6) follows an equation called the resistance infegrak' m Fig. which is typically a kind of lowpass filter.and the chosen stability margins should not be excessive. 3. 1. as illustratedinFig. positive feedbackconcentratesnearthecrossoverfrequency fb. in other words. the larger must bethe area of positivefeedback. Then. f I theareaofnegative feedback overthefunctionaljiequency bandwidth needs"to be maximized. as shown in Fig. 3. Corollary 2. the feedback decreases the output meansquare error since the feedback becomes positive only at higher frequencies where the error components are already reduced by the plant.34 Negative and positive feedback areas Therefore. The value of the positive feedback depends on the distance from the Nyquist dia to the critical point.7 looks much smaller because this picture was drawn with a logarithmic frequency scale. Frequency Response Methods w e e d b a c k is negative is equal to the negative of the integral over the rang& positive feedback. dB . let e(@) = Z(i00) = R u a ) +jX(ja) standfortheimpedance of theparallel connection of a capacitanceC and a twopole with impedance Z'.3 Integral of resistance Next.3. if negativefeedback reduces the effect o f disturbances in certainfisq=ions. Since the positive feedback concentrated is withina few octaves near the crossover frequency.9. and from (3. Typically. negative positive feedback feedback 0 Fig.76 Chapter 3. Thepositivefeedbackareain Fig. the area of positive feedback must be maximized as well.7. or. there must exist a fiequenq region o f positive feedback where these effects are increased. The larger the negative feedbackand its frequency range.
37 Stray inductance limiting the real part of admittance The resistance integral is also useful for estimating the available performance of of mechanicalflexiblestructures. 3. and (c) with different 2 ' but the It is also seen that the maximum value of same C. becomescritical(in product in systemswherethestrayreactiveelement.38. or in the parallelor series feedback paths of such amplifiers). 2' can be implemented electrically and connected to the mechanical structure via an electromechanical transducer. (c) Chebyshev bandpass filter. micromachined mechanical system. In mechanical flexible structures where some flexible modes need to be damped. diagrammed in Fig. Fig.Importantclasses of controlandactivedamping flexible plants include active suspension systems. and large. R overthedesiredfrequencybandcanbeachieved if R equalszerooutsidethe ' a filter loaded at a matched operational band.e.2L TE ' 0 (3.8) and(3.The frequency responsesof R in Fig.36(a). Frequency Response Methods I1 Itisseenthat the area under the freauencv response of the resistance R is exclusively determined by the parallel capacitance C. does not contain a series inductance. and (d) their resistive components The similarly derived integral of the real partof the admittanceY. (b) resonance. i. 3.Chapter 3.36(d) relate to the twopolesof Fig. which can be achieved by using as 2 resistor. . 3. in the input and output circuits of wideband highfrequency amplifiers. to achieve maximum performance over a specified bandwidth. (b)..9)arewidelyappliedinradiofrequencyand of theavailablebandwidthperformance microwaveengineeringfortheevaluation C or L. The relations (3. 3. where the admittance Y' of the remaining is valid for the dual circuit shown part of the circuit does not turn to 0 at infinite frequency. Flg.37. 3. sometimes the damper can be connected only to the port where a mass or a spring limits the bandwidth of a disturbance isolation system. particular. 3. IReY(w)dm=..relativelylightweightactivelycontrolledanddampedstructuresinzero gravity environment.. The area under the curves is the same.9) in Fig. In this case.36 Twopoles made of reactive twoports loaded at resistors:~(a) lowpass.
p.B') 0Q du.12) . 3.38 Active damping of a mechanical structural mode by connecting an active damper with impedanceZ ' 3. Ad 0 Fig.9. Hence.lo).thegain. I .By (3. Frequency Response Methods 4 I M. and frequency responsesA" and A' joining at higher frequencies as shown in Fig. can be arbitrary. 3. this difference is AA. (The use of this formula will be illustratedin Section 5. thephase. In other words.5 Gain integral over finite bandwidth Still another important relation is (3. is derived in Appendix The integral can be conveniently applied to the difference between the two gain3. A. i. therefore having the same value of Am. E= (Ao"Ao') =  n. and decades. dB In (3. dB. function. and UJ.39. j (B".10) where B is the phase shift of an m.9.39 Two gain responses having where a is the difference in the phase a common highfrequency asymptote integrals. 3.e.mechanical transducer Fig. and Am are the values of the gain at respectively zero and infinite frequency. the lowfrequency gain difference is AAo . u = In ( O C ) . the difference in the areas under the phase responses.andthe frequency units are related the to natural logarithm. the available feedback is larger in Nyquiststable systems because of their larger loop phase lag.dB = 0.56~ (dec X degr) (3.Whentheunits are convertedto degrees.78 Chapter 3.5.lo). In particular. the larger the feedback.be the area of the phase lag. the larger must. the integral is taken along the frequency axis with logarithmic scale (the equation 4 ) .4 Integral of the imaginary part The relation knownas the phase integral is (3.) It follows that an increase in the loop gain in the band of operation is accompanied by an increase of the area under the frequency responseof the loop phase lag.1 1) A. ty\n" + z Spring ElectroM .
is150".12) means that i f the phase lag at frequencies o 2 1 needs to be preserved.p.e. then the loop gain responsecan be reshaped in the finctional band 5 1 as long as the area of the gain plotted against arcsino is not in Chapter4. The extent of this dependence is determined by the weight function lncoth 1~121. This formula uses natural logarithm units for the angle. 3.e. The convenience of relating the phase to the slope of the gain with this Bode formula is the reason why the gain responses drawn with logarithmic frequency scale became called Bode diagrams. the vhase shift atanv specifiedfreauencv depends on the g& w o n s e slope at all freauencies. plot(u. than the remote parts of the Bode diagram do. i. (3. b = log(coth(abs(u/2))). the slope is 6 dBloct and the phase shift is 90". proportionally.e. attenuation. the neighborhood sf u = 0. grid 5 4 3 2 1 0 3 2 1 0 1 2 3 Fig. changed. ) = . lncothdu 2 j.13) i where u = In (o/oc).40 Weight function In cothlu/2l Due to the selectivenessof the weight function. systems. For the gain response having constant slope 10 dBloct.b. i. the rule that the result (output) never precedes the cause (input) (see Appendix 3). charted in Fig.'w'). For example. the phase shift is proportional to the slope o f the gain response versusthe freauencv with logarithmic scale. For a double integrator. for single a integrator with transfer function l/s.to along the uaxis.3. contributes much more to B. The Bode phasegain relationship is equivalent to the stability of m. As seen. Since the integral in (3. 3.13) is taken from . Frequency Response Methods 79 With application to feedback systems. the phase shift.9. the &hborhood of the frequency ~ 0 (at . over the frequencies from0 to . the slope is 12 dB/oct and the phase shift is 180". i. Applications of this rule will be studied 3. and frequency. The integral can be proved with countour integration (see Appendix 4) or with the Hilbert transform.200). it is possible to find the phase shift B at a specified frequency ac using the famous Bode formula (see Appendix 4 for the proof): 1 wdA Iul B ( o .40 with the script 00 00 u = linspace(3. (3.Chapter 3. which the pkase shift is being calculated].6 Phasegain relation Given the gain frequency response A(@) of a minimum phase transfer function. It is important that this andotherBodeintegralsareapplicabletothetransfer . and to the causa/ity principle.
therefore.80 Chapter 3. (c) shows the weight function centereda at .< 1 which relates to the following lowpass response: the gain is 0 dB up to the corner frequency u)= 1. expressed in the form of the Bodediagram. the Bode diagram is shown. the slope of the Bode diagram is plotted. (c) weight function centered at a. Calculate the phase at frequency a .41 is only illustrating the structure of the phasegain relation. \ 0 Fig. accounted for separately. . or whose n.Therefore. The Bode integral allows one to exclude the phase from the three variables . 3. The relation between the remaining two variables.the system design can be based on only the gain responses which is very convenient. The area under response(d) gives the phase shift at a . in (b). additionally calculated and.41 Phase shift calculation at frequency oc:(a) Bode diagram. 3. . components are small or can be. then decreasing with unit slope. phase. The sequence of frequency responses in Fig. Frequency Response Methods functions of physical systems which are not necessarily rational.employed in the Nyquist stability criterion.p.p. . the gain and the frequency.Notice that Fig.41 illustrates the structure of the Bode formula: in (a). A l d u = 1.containsalsotheinformationaboutthephaseshift. Bode diagram methods are widely applied to practical links which are m. and frequency .gain. (b) Bode diagram slope. A practical method for the phase calculation willbe described in the next section.3. (d) shows the slope response multiplied by the weight function. (d) product of the slope and the weight" Example 1.
Corollary. approximate calculation of the phase is quite often required during the conceptual stage of the design.if the segment's slope is a dB/oct.6n dB/oct over w octaves centered at fc . or the right scale multiplied by d10.Chapter 3.Bodeprovedthatevenacrude approximation of A rendersafairly.Forthe responses typicalfor automatic control. the number of segments need not be large. . at u = 1. so that this sum as wellb w o r t i o n a l to the fieauency. 3. the phase is B ( o c ) = _I 7 In w. which is the frequency where the gain slope changes from zero 1.accuratephasefrequencyresponse. Bode diagrams can be approximated piecelinearly by segments and rays. Frequency Response Methods 81 From (3. In Fig.10 Phase calculations' Accuratecalculation of thephase lag fromthegainresponse is rarelyneededin engineering practice. and computer programs developed for calculation of the integral (3.)as . the phase responses are plotted for the gain ray that originates at fc with the slope of 6ndB/oct (dashed line). 3. we can rewrite the expression for B(w. then the left scaleof the phase should be multiplied by a/6.) = " 0 .i.p u l t e r a t low frequencies is a sum of functions. Then At smalloc and finally. However. of any lowpass filter can be approximated in a piecelinear manner. Iul Incothdu 2 where the lower integration limit corresponds to the corner frequency 0 = 1. and for small readjustments of the loop frequency responses.13).e. For these purposes.42. ihe phase o f a l o w .e. and for the segments (ramps) of the gain response with the slope of . . 2 n (3.the logarithm equals l2coJol.toi. a modified version of a graphical procedure suggested by Bode is described below.14) The phase shift is negative and proportional to the frequency.13) are used rather infrequently. B(o.). u = In(l/co. In general. and the phaseresponsesrelatedtothesecanbeadded up. each proportional to the frequency. Since and du = dln(o/mc)= a"dw. Since the gain response.
That is. looo 8 80° ‘f: 30° 20° 60° 40° 0 f c e 1 oo O0 20° 2 3 1 ’ 9 1 4 I I II 2 1 I 1 I oo 0 I 1 1 1 1 1 I I 2 I I 3 I 1 1 1 1 ’ 4 10 .42.sr goo 80° 30° 60° 50° 40° 1 40° . ‘ E : 8 L % . The total phase response issum theof the three phase responses._. 3.i 120°.5 . Frequency Response Methods E= . three segments with zero slope (no phase is related to these segments).42 (a) Ramp gain response with constant slope over woctaves centered fc.eachrelatedtoasinglesegmentora rayof thegain frequency response. and (dashed line) phase response corresponding 6 todBloct ray Example 1. . and a ray as illustrated inFig. 1 (b) 5 Fig. A loop response crosses the 0 dB line at the frequency 800Hz. 3. whilecomplying with a set of heterogeneousconstraints(such as weighted maximization of the real component over a given frequency range under the limitation in the form of a prescribed boundary for the frequency hodograph of the function). at and the gain ray starting at fc. The phaseresponseis thenobtainedasthesumofthe elementaryphaseresponses. (b) phase responses corresponding to this gain response.43. Piecelinear approximation of A(o) is particularly useful for trialanderror procedures of finding a physically realizable response for $(io) that maximizes a certain nom.In an attempt to increase the feedback at lower frequencies.82 Chapter 3.What will be the effect of this changeon the guardpoint phase margin? The effeqt can be calculated with thehelp of thechart inFig.1 . Fromthecurvemarked “1. the guardpoint phase stability margin will be reduced by 13’. The gain response is approximated bytwosegmentswithnonzero slope. it is contemplated to make the Bode diagram steeperby 6 dB/oct over an octave centered at 200 H z . Example 2. for different w.” at ‘the distance of 2 octaves from the center. 3. the phase is 13’.
p. and the lowfrequency and highfrequency asymptotic slopes. phase response and the Nyquist diagram related to a piecelinear gain response specified by the vector of corner frequencies. The inverse problem is. changing the gain response as seems reasonable. the process converges rapidly. MATLAB function BONYQAS described in Appendix 14 is based on this approximation. function. theresponsesimportantforpracticecan be found rathereasilywithaniterative procedure utilizing the Bode method of finding B(a) from A(a).5 oct/cm for more accurate calculations. 3. A piecelinear gain response can be viewed as a sum of several ray responses. phke can be calculated and a Nyquist diagram can be plotted.p. Next. . the vector of the gains at these frequencies. 5 dBlcm 0. say. Frequency Response Methods dB 83 70 60 50 40 30 20 10 0 .5 1 1 2’ 1 4 8 1 16 10 segment ’ segment 1 2dB/0Ct OdBlOCt Fig. here n is not necessarily an integer. A ray starting at a . we could calculate the related response B’ and get a physically realizable e’=A ’+ j B ’ .we would find the related phase response.) Example 3. As a rule. etc. 3. given the function B(A).43 Phase calculation for piecelinear approximation of a Bode diagram Starting with some initial response for A .11 From the Nyquist diagramto the Bode diagram From a known Bode diagram.Chapter 3.thesolutioncan befoundnumericallywithacomputerby approximating the Nyquist diagram with a highorder rational function. Although no analytical solution exists tothisproblem. and (The appropriatescalesare: 10dB/cm and 1 octkm forsketches. to find the Bode diagram. withthe slope 12n dB/oct canbeapproximatedby (s2 + aos + a : ) ” ’ . It calculates and plots the m. i.25 0.e. A ’ . the shape of the Nyquist diagram for an m. Alternatively. and the accuracy of the graphical procedure is sufficient.
3. (d) correcting the fistguess Bode diagram. rziising the expression under the sign of logarithm to the power n. . (c) plotting the Nyquist diagram.44(a). for Nyquist diagrams with some sharp angles (which are optimal for many systems).15) and (b) frequency responses for the gain and phase of (3..15) by 1lf.44 (a) Locus of the ratio in (3. The convergence is improvedwhenasharpcornerresponseis included in the set of elementary functions. \ Fig. 3. 3.15) by (f.15) . It has the peculiar property of having the gain of 0 dB over the frequencyband o 5 1. The highpass response shown in Fig.r.45 Bode diagrams for highpass and bandpass transformsof function (3. property (although not the p.e. etc. i. bandpass response shown in Fig.Frequency Response Methods The iterative process consistsof the following steps: (a) plotting a firstguess Bode diagram composed of segments and rays. the convergence of this procedureisslow. (b) calculating the related phase lag. This response is lowpass.p.15) Multiplication of the function 8 by n.44(b). For this purpose we can use the function (3.45(b) can be obtained by substitutingfin (3. preserves the function’s m. The asymptotic phase lag will be then of 15O0. . For example. and the asymptotic slope is 1OdB/oct as shown in Fig. 3. and the phase lag of n/2 for o 2 1. The frequency locusof the ratioin (3.46(a) and (b).llf) [Z]. 3.15) is plotted in Fig. (a) (b) Fig. and converges rapidlyfor smoothshaped Nyquist diagrams. for the phase lag property). the power coefficient is n = 5/3. 3. However.45(a) can be calculated by substitutingf in (3.84 Chapter 3.15) which is plotted in Fig. nn/2. 3.
This part can be connected to the rest of the diagram. 3. This lag is often called rag transport proportional to the frequency: IB. by thesameprocedureasthatusedinFig. the n. one can composeNyquistdiagramswithsharpangles. rational functions). (2)highpass. The transport lag is substantial when the feedback loop is physically long and/or the speed of thesignalpropagation in themediaislowasmighthappeninthermal. (3) constant slope and (4) a constant (notshown used as a in the picture)’producesthe responseshown in Fig.thesum of thefour responses: (1) lowpass. The transport lagof electrical signals in feedback amplifiers canbe significant when the feedback bandwidth reaches hundreds of MHz.p.47 to“glue” together the lowpass and highpass responses.12 Nonminimum phase lag Nonminimum phase lag isthelink’slaginexcess of thatgiven by theBode formula.46 (a) Nyquist diagram and (b)frequency responses for gain and phase of function (3.Frequency Response Methods 85 0 1 dB 0 Fig. lag is caused by the time z of and is the signal propagation over the media. pneumatic.15) raised to the power 5/3 Combining these frequency responses with piecelinear responses.Chapter 3. or acoustical systems. Fig. This response can be part of the Nyquist diagram for a Nyquiststable system.Forexample. we might expect the nonminimumphaselagtoappear (a) insystemswithdistributed parametersdescribed by transcendentalfunctionsand(b)insystemswhosetransfer functions have zeros in the right halfplane. Notice that the Bode formula was derived for functions with a finite number of poles and zeros (Le. 3. without zeros in the right halfplane of s . 3.47. 3. .47 (a) Example of a gain response composed of several elementary responses and(b) the response on the Lplane 3. In the systems with distributed parameters. So.I = cin.
p. The ratio 8.p. are shown. is some frequency. if the n. for Bni c 0.p.8 rad. 3. Bni 20l~i A sum of linear functionsis a linear functiori.16) For o c 0. Om.17) wheref. the function of 8. are in the right halfplane and they are the mirror imagesof the poles which are in the left halfplane of s.. lag because its phase B n = B . 8. (3.48 Poles and zerosof (a) m.B m = arg 8. 3.4)is preserved. 3.f. Since l8. ='8/8.(iO)l = lO(ja)l since the magnitudeof each multiplier( j o .48. lagB n is less than 0.48(c). 3. eachreal zero S i = q contributes n. = IB. is called pure n.p.p. i. Fig. Notice that le. of (b) n. poles and zeros are shownofann.links The ladder network diagrammed in Fig. As shown in Fig. In (b). . Frequency Response Methods The effect of the rightsided zeros Si is exemplified in Fig. 3. In (a).@) is also called all pass. shift Bni = 2 arctan(a/q) .86 Chapter 3. and of (c) the all pass 8/8. can be expressed In particular. these zeros being mirror images of some of the zerosof 8.e. generally. poles and zeros ofm. Since the zeros are either real or come in complex conjugate as pairs.49 consists of parallel and series twopoles. Therefore. itsfrequencydependencecanbeapproximated by thelinear function B.(fc)I f fc (3.p.(iO) represents the phase lag of 8 in excess of the phase lag of 0. function 8 which has some zeros in the right half plane of s.I = 1 at all e.13 Ladder networks and parallel connectionsof mop. function 8. zeros4 frequencies.p.8 rad.4q . 8.f..
transfer functions. the rightsided zeros of a general network transfer function can only result from mutual cancellation of the output signals of two or several parallel paths. of the target node.p.p.ifweusetheforcetocurrent. then the function transfer function is W W.49(b). Fig 3. becomes n. or along the same but path as.p. In other words. Similarly. for example. when the incident signal excites in different motion modes. It is therefore of interest for control system designers to have a simplemethod of detecting when a parallel connection of several paths.Hence. which is feasible even fors in the right halfplane. 3.p.9.Chapter 3 . when the signal propagates from the input to the output along different paths. a mechanical system transfer function can become n.aspreviouslymentioned inSection3. F. A r Fig.p.transferfunction zeros can only result either from infinite impedances in the series branches or from zero impedances intheshuntingbranches. W l and W2. the transfer function of a passive ladder network is always m. (b) example The transfer function for translational motion propagation in the xdirection via the mechanical system depicted in Fig.50 Mechanical ladder network Inladdernetworks.aslongasthebranchimpedancesare positive real and do not have rightsided poles and zeros. Frequency Response Methods 87 Fig.I .+ Wz is m.3.does not encompass the point . transfer function can result from parallel connection of several ladder networks. A general network transfer function can be presented as a sum (resulting from parallel connection) of several m. 3. 3.I f both W t and W 2 are stable and m. translational and torsional modes which add up in the motion..if and only if the Nyquistd i a g m for WdW.p.1.1). " . an n. each of them m. the transfer function cannot possess zeros in the right halfplan& Therefore.velocitytovoltage electromechanical analogy (described in detail in Section 7.1.p. Therefore.50 is equivalent to that of the electrical ladder networkshowninFig.49 Electrical ladder network: (a) general.p.The composite link's 1 + W2.51 shows parallel connection of two links.
(c) 45'.3.2. 3.logspace. What is the slope of the plant gain response. of a 8 The phase stability margin in a homing system is (a) 25'. 4 The plant is a rigidbody. W1+ W2 has no righthanded zeros if (1 + W2/W1) has no such zeros. in dB/oct and dB/dec? 5 The plant is a capacitor. 3. When the Nyquist criterionis used. the input signal is a current.8 to explain why the Bessel filter has nearly no .51 (a) Parallel connection oftwo links. p l o t ) . the output is an (a) acceleration (b) velocity (c) position.is required for the risetime to be less than (b) 0.printtheplotsforthegainandphase 3rdorder Butterworth filter. 3. What is the hump in dB on the closedloop response (calculate or use the Nichols plot)? 9 Use the relationship shown in Fig. Therefore. the input signal is a force. (b) closedloop system The proof is the following: The sum Wl + Wz = Wl( 1 + WdW. Wl possesses neither zeros nor poles in the right halfplane of s. The minimum phase property ofa system including more than two parallel paths can be verified with the analog of the BodeNyquist criterion for successive loop closure. the index 1 should be assigned to the path with IWdWlI to roll off at 'higher larger gain at higher frequencies in order for the ratio frequencies.). Frequency Response Methods n (4 0 Fig. (b) 35'. (d) 5 5 ' . 2 What frequency bandwidth .8. which stability can be verified with the Nyquist diagram for WdW1. 3.72 psec? (a) 5sec 3 Calculate the risetime for the syst?m? with the highestorder frequency response from those shown in Fig.88 Chapter 3. in 7 Usingsomefilterdesignsoftware. with 'linear and logarithmic frequency scales (use MATLAB commands linspace. the output is a (a) voltage (b) charge.e. 1 Derive formula (3.5 sec (c) 2 msec (d) 1 nsec (e)2. i.1.1). The latter expression is the transfer function of the'system (b). Here. in dB/oct and dB/dec? 6 PlotwithMATLABtheBesselfilterresponsesusingtransferfunctionsgiven Section 3.14 Problems Many problems on Laplace transform and frequency responses can be found at the end of Appendix 2. if 1/(1+ WdWj) has no such poles.9 and 6. What is the slope of the plant gain response. Applications for this criterion will be exemplified in Sections 5.
40501 RP 4 0 0.2s+4) EP502410 GP1. place semicolon . the return signal comes back in phase and with increased amplitude. ( 4 ) to (6) to (7) GP2 '*60 0 5 80 G P 3 7 0 0 6 1 RP2 5 0 1MEG . non inverting.23. Is the closedloop system stable? How many poles in the right halfplane does the closedloop transfer function have? 14 Plot the Bode diagram and the Lplane Nyquist diagram for the system with an unstableplantshown in Fig. zero s=10. inverting..PROBE .Chapter 3. and check the system stability.033333 .1 20 . * crossover s=20 G C 2 0 1 0 1 RC1 1 0 1MEG .52. 89 10 Why are Nyquiststable systems stable? (The gain about the loopin such systems exceeds 1. * loop closing resistor RL 7 1 1MEG . * VIN 8 0 AC 1 RIN 8 1 1 . for SPICE. hence. 12 If the jcoaxis maps onto the locus of F(io). 3. for SPICE . for SPICE RP3 6 0 1MEG RP4 7 0 1MEG for SPICE CP2 6 0 1 CP3 7 0 1 .05s+4) * plant integrators. ' ** unst_pl. Z2=(s+10) / (s+4) * plant loop. Hint: Modify the system under analysis to be' definitely stable. while. to close the loop. * compensator. The Nyquist diagram makes 2 revolutions about the critical point. after M to reduce RL .AC DEC 100 0. Z5=0. onto what area is the right halfplane of s mapped? Whatis the mapping of the zeros of F(s)? 13 The system is stable openloop. The following SPICE program analyzes the equivalent electrical schematic diagram in Fig. observing the topology of the locusof F(s).. pole s=40. .1s/(sA2+0. 3.cir for feedback loop with unstableplant . to make node. Frequency Response Methods overshoot.33333 LC 3 0 0. It is counterintuitive to suggest that such a system is stable but it is!) 11 Prove the Nyquist Criterion using the notionof continuous dependence of the roots of a polynomial on its coefficients.END .1s/(sA2+0. Changethevalueofdamping in thelocal feedback path of the unstable plant. (if using PSPICE) . and further change the coefficients gradually and continuously unt they reach their true values. closed loop 0.025 . the phase is 0. (1) nonfloating RC2 2 0 1 RC3 2 3 0.5 CP1 4 0 10 LP 4 0 0.
the xvariable being vp (7) . and the compensator transfer function C = 400(s + 5)/(s + 20). or SIMULINK. . Frequency Response Methods GP2 1 1 4F IGP3 1 1 % T RL Fig. 15 In the plant of the system shownin Fig. z is the torque andJ is the moment of inertia about the center of gravity CG. the plant aerodynamics transfer function 1000s + 2000. and the range50. I s the system stable? 16 Depending on the angle of attack of the horizontal stabilizer. To plot the Nyquist diagram. To plot the plant loop Bode diagram. change the gain block gain coefficient to 5.23. Use SPICE. the airplane in Fig. 3 . The block diagram for the pitch autopilot feedback loop is shown in Fig.53(a) can be statically unstable. (b) T = 2 5 0 ( ~ + 5)/[(~ +5 ) (+ ~ S)]. Alternatively. Is this airplane stable with the autopilot? (4 Fig. plot vdb ( 4 ) vdb ( 5 ) +2 0. 3. use MATLAB.15) is p.90 Chapter 3. plot V & (7 ) and 0. 5 3 System with aerodynamic instability in the plant 17 Show that the function (3. Consider J = 2000. MATLAB. 3.r. make the xaxis scale linear.2 *vp (7 ) . 400. 3.53(b). 3. 3. or some other analysis program.23 When using the PROBE postprocessor to plot openloop responses with convenient scales. the control surface gain coefficient of 1000. SIMULINK.52 SPICE model (defloating 1 MEG resistors not shown) of the system with an unstable plant shown in Fig. plot vdb (7) . and for plotting a Nyquist diagram for stability analysis? (a) T = 250(s + 5)/[(s + 5)(s + 5)(s + 5)]. 18 Which of the following functions can be used sa a model ofa passive physical plant. Here.
3.(3). (f) (1.youcanoftenreadthesentence:"Thefeedback in thissystem is negative.another. (2). (9) (10. 5.1 8(a). .* . 3). (b) (1. What is thephaseshiftatthefrequenciesattheendsandthecenterofthe segment (use the plot in Fig.9 dB/oct.54. (d) 3 oct wide with the slope (e) 1 oct wide with the slope 5 dB/oct. 5. 3). To copy the plots from one sheet of paper to .54 using the same magnification (either 1:1 or 1:1. 15. (h) (6 t j l . . (b) s"*' . (c) s52 .5)/[(s + 5)(s + 5)(s + 50)]? Is the feedback negative at the crossover frequency? 22 If the slope of a Bode diagram at all frequencies is (a) . (d)30 dB/dec. 0. 3. 21 What are the guardpoint gain and phase stability margins if T = 6200(s + 2. (e) (4. . (e) s~. 26 Using the Bode phasegain relation. (b) 40 dBldec. or 2. (c) 2 oct wide with the slope 20 dB/dec. 3. what is 24 The gain response can be approximated by a segment 10 dB/oct.p. what is the phase shift? 23 I f the phase shift at all frequencies is (a) 9 0 ' (b) 40' the gain response? (c) 150' (d) 210°. superimpose two the sheets on a light table (or against a window glass).5. (b) with triple integration in the loop. (d) sa. in 29 PlotphaseresponsesforpiecelinearapproximationsofcertainBodediagrams shown by the responses ( l ) . 91 19 In literature. (4) in Fig. 4. interpolate. 3. . explain why the Besseltype gain response has better phase linearity than other filters. (d) (5 k j12). (b) 1. ' ' 27 Find the expression for n.5." Does this statement need qualification? O f what should you be aware? 20 What i s the slope of the Bode diagram and the phase shift of the function: (a) sOa6 . 28 Sketch a Nyquist diagram on the Lplane related to the Bode diagram Fig.8 dB/oct. 2 ) . (c) (10. Frequency Response Methods (c) T = 2 5 0 ( ~ +5)(+ ~~ O ) ( S +~O)/(S + 500).5 oct wide with the slope 12 dB/oct. (d) T = 2 5 0 ( ~ + 5 ) (+ ~5 ) ( + ~~O)(S + 500).5). 10). 10).Chapter 3.lag at lower frequencies for the transfer function having zeros (a) (3.1). For w = 1. (c) .42)? 25 Draw a Nyquist diagram for a loop transfer function (a) having a pole at the baxis.copythe pages with Figs. 0.4.'.1).Forconvenience. (a) 3 octaves wide with the slope 6 dB/oct. 3.42 and 3. or make and use a transparent paper copy of the page with the problem.
55 shows the loop g i n responses of the thruster aititude control about the x. .+2 0 ) ] . Answers to selected problems 3 The cutoff frequency for the highest order filter is approximately 2 radlsec. and W1 and W 2 are m. b IS .92 dB 40 Chapter 3. and a third of it is the rise time . I s + W2 m.(b) Wt = 10000/[ (s. Frequency Response Methods 30 20 10 0 10 Fig. *'.W2 q + . I : IO)]. W 2 = (s. (a) W 1 = 1000/[8 s.at all freqwncies. and the period for this frequency is approximately 3sec.? * .Wtl I W2l. Approximate the responses by segments and rays. is that of the'4thorder Bessel filter. 3. 3.p.and zaxes of a spacecraft.54 Piecelinear Bode diagrams for phase calculation exercises 30 Fig. Fig. 32 In a homing system the closedloop response. Find the loop transfer function and plot the openloop Bode diagram. 3. . .if: .+ I)(s+ 5)/[(s + 200)(s + SOO)]? .55 Bode diagramIS for attitude control loops of a spacecraft . Calculate the phase shift. 4 ) ( + ~ lO)/[(s + 400)(~ + IOOO)]? (c) I. .p. . 31.
i. From Fig. Therefore. 3. gain response can be approximated piecelinear a by response as shown in Fig. 10" 1oo 10' Frequency(rad/sec) F. 24 (a) The phase shift is 110' at the center of the segment. i.it must cross the jmaxis.to make the phase lag proportional to frequency. 11 Start with a simple system with Tis) = 0. which guarantees the absence of an overshoot. . then the system must be stable. Anotherexplanationisthat. correspondingly. The responseis plotted in Fig. the lo" 1o1 increase in . the phaselagincrease over an octave should double such increase over the preceding octave. therefore.1 thatisdefinitelystable. the locus of Tnever crosses the point 1.e.. is rather close to linear because of the contribution to the phase lag at lower frequencies made by the Gin responseslope at lower Fig.ig. the curvature of the transient response also increases gradually and does not reach large values.57 Bodediagram . 9 Since the slope of the Bessel filter gain response increases gradually with frequency and does not reach large values before the gain is already very small. formula (3. It is seen that the phase response. 27 (a) 8. and within . must become 0. At the crossing frequency the closedloop gain must become infinite and F.e. 3. If thisneverhappens. 356 Bessel filter g& response frequencies. the gain response slope at lower frequencies is negligible. we see that the rise time (from 0. .the passband. Frequency(rad/sec) response should double each octave.8 where the transient response is plotted. 26 Bessel filter. For higher selectivity andphaseresponse (Butterworth.57. roughly speaking.e. Cauer).56 (not to scale). = 2[arctan(1Om) + arctan(col3)l 32 105/(s4 + 1Os3 + 4 5 8 + 105s). Start changing the .Augmentthe transfer function of Tto'polynomials of high order but with infinitesimal coefficients.the slopeofthegain . but the slope becomes very steep right after the passband ends. the phase lag is smaller atlowerfrequenciesandhigherat higher frequencies thus making the phase response curved.3. If a root of the characteristic polynomial migrates into the right halfplane of s.Methods 93 = 1 sec.Frequency Response Chapter 3.i. filters Chebyshev.coefficients while observing changes . 63' at its ends.the locus of Fduring the modifications never crosses the origin.1) gives a good result.1 to 0. in the resulting Nyquist diagram. 3.9 of the output) is about 1sec.
“most desirable.3. subject to all these limitations except the first. includes Bode a step.4. lag (analog and digital) in the feedback loop. The second part is of distributionof theavailablefeedbackoverthefunctional feedback band. In application to practical control systems.reliability. The above solution is then generalized by application of a Bode integral to reshaping the loop gain response over the functional bandwidth (Le.speed of action.4.andfortheloops where the ‘plantis unstable.17. The feedback bandwidthis limited by the sensor noise effect at the system output. The effect of resonance mode coupling on the loop shaping in MIMO systems is considered.2. for. in order to improve system performance. Increasing the compensator order from 4 to 12 and including in the 20 to 30 lines of codetothe compensatorseveralnonlinearlinkswillonlyadd controller software. When the book is used for an introductory control course. 4. thesensornoiseeffectattheactuatorinput. even at the price of making them complex. for collocated and noncollocatedcontrol. The order of the compensator tobe reasonably close to the optimal is.I Chapter 4 SHAPING T ~ ~ L O O FREQUENCY P RESPONSE The problem of optimal loop shaping encompasses two fairly independent parts that can be solved sequentially (thus making the design structural): The first partis feedback bandwidth maximization which is solved by appropriately shaping the feedback loop response at higher frequencies (in the region of crossover frequency and higher). As a rule.3. Example 1.1 Optimality of the compensator design Webster’s Collegiate Dictionary defines the word “optimal” as.solving the second part of the loop shaping problem).. typically.5 . The problemof optimal loop shaping is further described and the formulas are presented for calculation ofthemaximumavailablefeedbackoverthe specified bandwidth. The optimal shape of the loop gain response at higher frequencies.4. it is worthwhile to make them very close to optimal.. Hence.” and practical engineering views optimality as a provision for best customer satisfaction. Sections 4. 8 to 15. compensators and prefilters are many times less expensive than the plants. or several extra resistors. The second kind of the design tradeoff is that of the controller performance versus complexity. 4. The tradeoffs between these requirements are commonly quite clear and/or canberesolvedwithBodeintegrals. if 94 . Loop shaping is described for plants with flexible modes.5 can be omitted.2.3. The feedbackcontrollerdesignisincomplete without estimationof the theoretically best available performance.7. It isshown that the feedback is larger and the disturbance rejection improved in Nyquiststable systems.4. capacitors and operational amplifiers.6. It is described how to shape the responses of parallel feedback channels to avoid nonminimum phase lag while providing good frequency selection between the channels.repeatability. The Bode step i s presented in detail as a loop shaping tool for maximizing the feedback bandwidth. and disturbance rejection.4.planttolerances(includingstructural modes). and nonminimum phase.3. the controller performance means some combination of accuracy. and 4.
Phvsical system models must reflxt the uncertain& inthe system parameters. Still. Response Loop Shaping the 95 the compensator is analog .without proving that this design is the best possible. and the nonlinearities. Even when the accuracy of the system with a simple controller suffices. [9]) which was considered bymany a cornerstone of the so called modern control. which may permit replacing some initially chosen components and subsystems by simpler and cheaper ones. In physical systems the loop gain drops faster at higher frequencies than ofthat (3. the asywtotic behavior of the tranflg jknctions at higherfrequencies.Chapter 4. as we already mentioned. the system will remain operational when some of the system parameters degrade to the point that without the better controller. we should not even ask: . and not onlytheperformancespecified by acustomerrepresentative who doesn’t know in advance what kind of performance might be available. the absolute value of the return an integrator. the resulting time per operation might be reduced by.Is this design the best? If not. Example 3. the yieldof a chemical process can be raised by 2%.the sensor noise. we analyzed some design of antenna elevation angle control . This is why the compensators should be designed to provide close to optimum performance. a fighter’s maneuverability can be noticeably improved. by how much can it be improved? Normally. the design must result in a sound solution when the idealized plant is replaced by the physical plant and actuator models. Or. That is. that it cannot be applied to physical systems.and may substantially improve the system’s performance. Therefore.and. it still pays to improve the controller. and (b) what loop response achieves this performance limit.” The process of control law design that deserves tobe called optimal must provide timely information to the system engineers about the achievable control system perforrnance and the related possibilityof relieving certain requirements to the system hardware. In Chapter 1. The problem with this theory is.In the paper “When is a linear control system optimal?’ (see [1281 in the bibliography to Ref. The theory of optimal feedback loop shaping should be able to provide the answers to the questions: (a) what performance is feasible. the system fails. For example. Linear compensators are fully defined by their frequency responses. Commonly. unth additional savings on maintenance. the following definition is given: “afeedback system is optimal if and only if the absolute value o f the returndigerence is at least one at all frequencies”. since with larger margins in accuracy. however. at what cost? . and the number of pieces of the equipment at the factory can be correspondingly reduced by 5%. if the settling time of some expensive manufacturing machinery with a short repetition cycle of operation can be reduced by 20% while retaining the same accuracy.since. the optimality requirements must relate to the entire engineering system and not only to the controller in the narrow sense. The reallife controller is multiloop. 5%.etc. Meeting this condition assures dynamic optimality which is rigorously defined in the paper.) Reasonable questions for the customer to ask are: . Example 2 . (It is not. and according to Bode integral difference cannotbe “at least one at all frequencies. Or.7).andincludessignalfeedforwardandhighordernonlinear compensators. say.controlsystems are initially designed as linear withsomeidealized plant model. the cost of the compensator is small compared with the costof the antenna . the problem of optimal linear compensator design is the problem of optimal loop response shaping.
................ Nonlinear compensator design and system simulations Fig.................................. the loop gain at higher frequency should be rolled off fast to avoid instability.. The structural design flowchart for the basic feedback control system is shown in Fig.2......... .......... bearings.. ....................................................2 Feedback maximization 4.......... 4...2...........9 (andwill further discuss them in Section 7..... Also..96 Chapter Response 4.. (II Hardware i .. or feedback path) \ I . By thisresonance......... Conceptual design iterations 3 Hardware choice .. the loop responses in this example do not include the plant structural modes................... A flexible mode of a structure may beanywherebetween 30 and 40Hz.................... and to estimate the available performance without completing the lengthy compensator design.................. Can this be done? Example 4...... D . for some specific task............. and what is the value of this feedback? Can it be quickly calculated without designing the controller? 4. The feedback system therefore must be able to reject the noise over the entire bandwidth 30 to 40Hz............ 4............ and servomotors can add to the system cost)............. The system engineer asked the control designer find to the shape of the control loop response which is optimal...........theenvironmentalvibrationnoiseisaccentuatedandthe accuracy of the device under control decreases............... Loop Shaping the dish... Given the modes............................... Preliminary desilgn A1 Loopshaping at higher frequencies v A2 Shaping the loop over the functional bandwidth ............... System engineering ~ I' + / B1 Prefilter * design (or feedforward.......1 Structural design We alreadymentionedtheadvantages of structuraldesign in Section 2.............1 Feedback system design flowchart . I C Linear compensator design simulations Compensator design and system 'IC........ (although better or additional sensors..................1)... How to implement the maximum feedback over this frequency range........ t ..................................................1............................. : : Performance specifications 2 ................
by I O O response ~ shaping. achieving a. After the system engineers finally decide which hardware configuration is the best.Thecompensatorresponseisfound by subtracting the plant response from the loop response. the uncertainty of the plant parameters increases and the gain drops at higher frequencies. Inclusion of only the stages Al. As was stated in corollaries in Section 3.is therefore zero. the same happens due to the plant flexibility. Physical systems include actuators with saturation.2.theresponsesneednotbe expressed by rational. After the loop is shaped.3.maximurnarea o f negative feedback in thejknctional feedback band.Atthisstage. Since the first subproblem is to a large extent independent from the second one. of Because this and because of the sensor noise which will be studied in Sections 4. and the integral offeedback (3.Chapter 4. If a system contains no other Ir&)I . using the preliminary design results on the available control performance. by plots..in this section is the region of transition between the functional frequency band and the higherfrequency band where the feedback becomes negligible. The preliminary design addresses specification A first. Le.2 and 4.9. and B the nominal commandtooutput transfer function. This can be subdivided into A1 achieving maximum feedback bandwidth by appropriate loop shaping at higher frequencies. in thermal systems. the loop gain must sharply decrease at higher frequencies.3.The system engineers evaluate different versions of the hardware/software configurations iteratively.A2.functions. therefore. and in mechanical systems. the loop gain. The crossover region studied. only the steps A1 and A2 (and sometimesB1)need to be performed to provide the system engineers with the accurate data on the available control performance.7). the commandtooutput response is modified (if required) by adding an appropriate prefilter or command feedforward meet to specifications B. shaping the Bode diagram in the crossover r e g b (step A1 from Section 4. In otherwords. In electrical circuits. due to thermal resistances and thermal capacitances. During the conceptual design of complex engineering systems. and A2 distributing of the achieved feedback over the functional frequency range so as to exceed the worstcase specifications.3. and D the Compensator is augmented with nonlinear elements and the system is simulated. 2 Bode step In physical plants. . Shaping the Loop Response 97 The performance specifications comprise the desired responses of A the disturbance rejection and the sensitivity. Then C thecompensator and prefilterresponses are approximated by m. 2 . orby tables. rational functions and implemented as algorithms or as analog circuits. For the purpose of analysis. jkquencies atleast as ai".p.B in the iteration loop makes the loop fast. 4 . thecompensatorandprefilteraredesigned. They can be expressed by rational or transcendental functions. it makes sense tosolve themsequentially..1) is critical in achieving a maximum area of positive feedbacknear the crossover and. The slope is rather steep in decreases at higkr physicalfeedbacksystems with n 2 2. this happens due to stray capacitances and inductances. the loop gain Bode diagram can be sufficiently well approximated at these frequenciesby a line with a constant asymptotic slope of 6ndB/oct.2.
Without the step. i. reduces the feedback bandwidth). 4.2 Bode step: (a) absolute stability boundary on the Lplane. With the linear approximation (3.9s ChapterResponse 4. 4.. absolute stability is required and the stability margin boundary must be as shown in Fig. Therefore. 4. From the Bode integral of phase it follows that to maximize the feedback.6n dB/oct. The highfrequency asymptotic loop response is considered known.. It is defined by: .(fc)..The step reduces the phase lag at the crossover frequency. Because of the Bode phasegain relation. The gain monotonically decreases with increasing frequency and eventually degenerates into the highfrequency asymptote with the associated phase shift n90".The related loop phase lag is less than (1 y)l SO" until the loop gain becomes smaller than x dB. Such a diagram is shown. given the highfrequency asymptote. 4.this asymptotewith coordinates (fc . (b) piecelinear gain respbnse with related phase lag response that produce the Nyquist diagram shown in (a) which approximates the boundary The transition betweentheslope12( 1 . and B.y) dB/octandthehighfrequency asymptotic slope mustbeasshort as possible to increasetheloopselectivity:to maximize the loop gain in the functional frequency range while reducing the loop gain at higher frequencies.. the slope of the Bode diagram at these frequencies is approximately1 12( . the nonminimum phase lag at fiequencies lower thanf. the phase lag in the crossover area would be too large due to the steep highfrequency asymptote and the nonminimum phase lag.l(b). by the thin line.y) dBloct. Loop Shaping the nonlinear links.2(b). The system is phasestabilized withthemarginnot less than y 1SO" up to the frequency& where the loop gain drops to"x. is .. (a) the asymptotic slope (b) the point on.e.x) as shown in Fig. The loop gain response is piecelinearwith corner frequenciesfd and fc. the length of the step must not beexcessive.l(a). the Nyquist diagram should follow the boundary curve as closely as possible. the phase lag must be the maximum allowable. The transition is provided by the Bode step made at the gain level of xdB as shown in Fig. The nonminimum phase lagBn(fc)is assumed to be less than 1 radian which is true in welldesigned systems. (c) the nonminimum phase lag at this frequency dB degr \ loop gain 1 2(1 fidB/OCt  0 X Fig. 4. It will be shown below that this Nyquist diagram corresponds to the Bode diagram shown in Fig.but also reduces the loop selectivity (Le.17).2(b).
.14) as.. 2 n.5 rad. fclfd m 0. when n = 3 and B n ( f . This case is common but not ubiquitous. Further.2(a).andthe crossover frequencyfb = 6. maximizing the feedback in the functional frequency range. and over a . = IBn(fc)l.Chapter 4.. the Bode step frequency ratio is f d For the typical phase stability margin of 30°. The Nyquistdiagram for theBode step responsecloselyfollowsthestability boundary in Fig. in Chapter 13.2) Example 1.theBodediagrammustbemadeshallowoversomerangeinthe crossoverfrequencyregiontoensurethedesiredphasestability margin.7. Whenthespecifiedstabilitymarginsare 30’ and 10 dB. Thereare several options for where to do this: to the right offd by the Bode step. ) = 0. y = 116. and the corners of the Nyquist diagram become rounded.The phase lag related to this ray is expressed with (3. and 5. fc f The phase lag related to the asymptotic slope ray which starts atfc can be expressed with (3.e. approximately.4 kHz.14) as. Inanycase. Shaping the Loop Response 99 B. Examples of loop responses with Bode steps will be given in Sections 5. and in Appendix 13.y)180°.2)fcFd= 2. then the slope is 10 dB/oct and fd = 2fb = 12. In practice.6. 4. i. then from(4.3 so that fc 2: 30 kHz. this response is approximated by a rational transfer function. ) . f fc Consider next the “discarded” dashedline ray which is the extension of the main slope line beyond the frequencyof the beginningof the step fd. the sum of the phase contributionof the asymptotic slope and the nonminimum phase lag should equal the phase contribution of the “discarded” dashedline ray.1 1. This consideration is expressed as From this equation. i. Noise reduction requirementsand certain implementation issues may require Bode diagrams to be differently shaped in the crossover area. to the left of the crossoverfrequencyas in systems where the sensor noise is critical.e.6n IB n ( f .5. To make the loop phase lag at frequenciesf e fd equal to (1 . The loop responsewithaBodestepshouldbeemployed whenthedominant requirement is maximizing the disturbance rejection. approximately.8 kHz. (4.
should be made shallower. the loopgain response in the neighborhood of j .. The compL. However. If this overshoot exceeds the specifications.w) 0 50 " io" 1oo Frequency (radlsec) 1o1 1 0 " Frequency (radlsec) 1oo 1 o1 Fig. d = conv([l 7. 4.4 Lplans Nyquist diagram . d. The plant 1 10"s P(s) = s2 lO+s is a double integratorwith an n. bode (n.1. Example 1.3 Example of a system having a loop response with a Bode step In this section an example of the implementation of a loop response with a Bode siep is considered. This response has been used as a prototype for several practical control systems.100 Chapter 4. The output transient response a tostepdisturbance in a homing system with a Bode step and 30' to 40' phase stability margin has substantial overshoot. a prefilterorone of its equivalentscanbeintroducedtoensuregoodstepresponseswithoutreducingthe available feedback.p. For a systemwhichhasanexplicitcommandinput. Loop the frequency range nearly symmetrically situated about the crossover as in the socalled PII) controller which will be discussed in Chapter 6 .sator Eunction C(s) is a ratio of a 3rdorder to a 4thorder polynomials (the compensator design methods are described in the next chapter). this will reduce the available feedback and the disturbance rejection.4.~ + 1 0 IZ(S) lls3 +55s2+llOs+36 1 10s "=:"= ~ 34s2 ~ + 97s + 83 s2 s + 10 d ( s ) s2 10+ s s4 + 7 .200).7 34 97 83 0 01. Loop responses without any Bode step typically provide4 to 20 dB less feedback in the functional fiequency range. [1 101).2. lag. [I 101). 7 + is plotted in Fig. The loop transfer function " 1 . Shaping Response. w = logspace(1.3 Openloop frequency response 'Fig.3 with T(s) = : C(S) n = conv([ll 55 110 361. 4. 4.
or Gaussian) filter. Response Loop Shaping the 101 The crossover frequency Fig. The commandtooutput responseof the system with the prefilter should be close to the responseof a linear phase (Bessel. .810. g = n + d. 180. w) g. phase] = bode(n. 4. 55% overshoot. 4.s ' . and 13.this example and explained at length in Chapters 10. The Gplane Nyquistdiagramisplottedin [mag. grid The slopeof the loop gain at lower frequencies approaches 12 dBloct and the loop phase shift approaches 180". The overshoot can be reduced by introduction of a prefilter R(s). s2 + 20.2(a) but violates the stability marghat these frequencies. The closedloop transfer function A 4 = nlg where g = n + d is plotted in Fig. + 0.6 Closedloop stepresponse As should be expected. (The zeros are added to n to make the dimensions MATLAB can calculate9. .200).2(b) following the stability margin boundary in Fig. 2 0 ~~~ 5T U n e ( s e c S )10 15 Fig. This response provides larger feedback at lower frequencies than the response in Fig.4 with a = 1radhec. 'w' . 'wo' ) . the loop compensation must be nonlinear. p l o t (phase. Such compensation is described in the end of . To ensure stabilityand goodtransientresponses to largecommandsanddisturbances(whichsaturatethe actuator). the closedloop stepresponse plotted by step (n. 4. 4.810. 20*log10 (mag).. 11.5 with n = conv([O 0 0 l1.d.Chapter 4. 4.5 Closedloop frequency response Fig.w). bode (n.n).1. This can be achieved by using a notch prefilter R(s) = s2 + o b s +0. w = logspace(1.) of vectors n and d equal so that Frequency (racUsec) 0 1 0 l 1 oo Frequency (rad/sec) 1 o1 0 .6 hasa large.g ) and shown in Fig. 4.0. 4.
n). bode (nc.7 with nr = [l 1 0. 4. 4. w) and the closedloop response with the prefilter is plotted in Fig.811. We might therefore expect a better transient response.8 Closedloop response of the system with a notch prefilter 1 0 5 10 15 20 5 10 15 20 Time (sees) Fig. = 1. 4. ! $ e . 0 50 e 5 .102 Chapter 4. 4. gc = conv(dr. 4. In the case considered. 9).c loo Frequency (radlsec) 10’ Fig. 4.1 0 Closedloop stepresponse of the system with the prefilter composed oftwo notches . w) The phase responsein Fig.8. Response Shaping Loop the The notch is tuned to 0. bode(nr. the overshoot is only 7% on the stepresponse for the closedloop system with the prefilter plotted by step (nc. dr.811. gc.9 Closedloop stepresponse of the system with a notch prefilter Fig.r d loo Frequency (radsec) 0 a 10 10” io’ 50 101 loo radlsec Fre uenc 10’ 0 8 8 180 2 a 360 10” 0) \\ “. The prefilter response is plotted in Fig.9% with the gain at this frequency equal to 20 log( 1/2) = . the rise time and the settling time remain approximately the same. 4. 4..5.8 has a lesser curvature than that in Fig. dr = [l 2 0.6 dB. 4.3 Prefilter frequency response Fig. Indeed.gc ) and shown in Fig.8 with nc = conv(nr.
w) % closedloop response nr = [l 1 0. dr2= [l 0. d. The plant gain variations can be specified by a multiplier k.200). k). w) % prefilter notch response bode (nr.1 1 Closedloop stepresponsesof the system with notches.8 4 0. % closedloop response with notch bode(nc. [l.gc2.6 f 0. nc2 = conv(nr2. bode(nc2.w) % closedloop responsewithnotches step (nc2 . By compensating for the hump with a small. gc2. dr nc. The equalization becomes less accurate when the plant parameters deviate from the nominal. 4. d.811.6 11. 4. plot(phase. w) % openloop response [mag.5 11. 1. bode(n. Response Loop Shaping the 103 The gain response in Fig. gc2).8 still has a hump at o = 1 which contributes to the phase nonlinearity and to the overshoot. gc2 = conv(dr2.2 1 0.1.w) nr2 = [l 0.dr2.2 I 0 5 10 Time (secs) 15 Time (secs) (a) (b) Fig. dr = [l 2 0.8 and (b) for k = 2. gc2) % step response with notches For &20% variations in the plant gain coefficient. the stepresponses are shown in 10%. The overshoot remains less than 1. bode(nc2. n). n). gsn+d: bode(n.lOl). 9). phase] = bode(n. gc = conv(dr.= conv(nr. % specify k w = logspace(1.7 34 97 8 3 01. n = conv(n. nc). one can further reduce the overshoot as shown in Fig. gc).gc. % second notch response bode(nr2 .Chapter 4. 20*10glO(mag)) % Nyquist diagram grid n = conv([O 0 0 11 . 4.5 11. (a) fork = 1. Fig. g. gc2 = conv(dr2. . [l 101). nc).6 dB additional notch with nr2 = [l 0. 4.811. w) nc2 = conv(nr2.5 .2 0’ 0.2 and k = 0. gc). dr2 = [l 0.1 1. A still better and more economical equalization can be devised and implemented.10. k = 1. d = conv([l 7.6 11. w) .4 0. The effects of variations in k can be calculated as follows: n = conv([ll 55 110 361. w) step(nc2.
13 =u RJ s[s + k 2 / ( R J ) ] s(s + 9. the compensation can be made nonlinear. variations in k the overshoot remains under 20%. and a. [l Example 3. = 12 x 211. and second.027. The angle0 = [k(U .e.1 l(b).5%) variations in the coefficients transfer function and to observe that these changes do not critically affect the system performance. 1 1. still.027 Nm2.1>x) the output starts growing exponentially. It is also recommended to make simulations of the polynomials in the compensator with small (say.4) in the transfer functions for the loopand for the prefilter. ja = 0. dcl = [l 03. The torque isk( U . and therefore. A dc motor rotates a spacecraft radiometer antenna whose moment of inertia J = 0. andU is the voltage applied to the motor.7. s0 is the antenna angular velocity..5. 0 is the antenna angle. the plant's gain increasesby 8 dB from the nominal thus reducing the gain stability margin to only2 dB. Example 2.(where a 75. 4.07) * If the loop n.= with crossover frequencyfb = 12 Hz.p. This is done by. first. and 13.(s) which represent transfer functions of two parallel paths. this is not a catastrophic failure of the controller.Cl(s)] where p1 is the lowest pole of C(s) 0 7 1 can be 0). the proof (simulation) is left as a recommended student exercise. a])..&)/R where the constant (the torquetocurrent ratio) back electromotiveforce EB = ks0. With ~ 4 0 % with 20 log k = 10. splitting the compensator transfer functionC(s) into the sum of two functions C. the compensatorand driver transfer function should be C(s) fiom Example 1 multiplied by c1 U " Z " 1 __ RJ s + k 2 / ( R J ) 0.7 N d A .4.s must be replaced by da. by placing a saturation link with an appropriate threshold in front of the first path's linear link.The numerator ns and the denominator ds of the loop transfer function can be found with .077(s+9. .(s) and C. The transfer functions of the paths can be found as C1= a& +p l ) and Cz = [C(s) . the transient response becomes oscillatory as shown in Fig. The related theory and design methods are described in Chapters 10. The motor winding resistance R = 2 SZ. lag equals that in Example 1. _ I The MATLAB code for this transfer function's numerator and denominator is k = 0. From the latter two equations.EB)/R]/(s2J). The motor k = 0.Response Loop Shaping the When the plant's gain coefficient increases to 2. the MATLAB function residue or the function bointegr from the Bode Step toolbox in Appendix 14 can be used. ncl = conv(res*ja/k. res = 2. i.and with k 2 312 (i.. For finding p1 and a. To ensure that largeamplitude commands which overload the actuator will not triggerselfoscillation. To use the response in Example 1 as a prototype for a control system . a = k*k/(res*ja).e.104 Chapter. is its residue. the voltagetoangle transfer function 0 k 1 . the coefficients can be appropriately rounded. and if the loop response needs to be the same as in Example 1..07) 0 s2 k S S _. the first path being dominant at lower frequencies.andtoimprovethetransientresponsesforlargeamplitude commands.
In many practical cases the disturbances' amplitude decreases with frequency.5. 4e k e 2. 2 .4.6 < k c 1. in the linear state of operation (for small commands). (Notice that if we apply the 1p21p transform not to the entire loop but only to the compensator and retainl/s2as the plant. wb) 105 and the numerators and denominators of the two notch transfer functions of the prefilter With nr = [l 1 0. dr2sl = lp21p(nr2. The overshoothndershoot for smallamplitude stepcommands must be: for the nominal plant. the system is phasestabilized with a constant stability margin. From Example 1 it is seen that these specifications can be met when the stability are chosen. the sensitivity in the margins are 30" and 10 dB. the numbers stored with double precision. With the plant P = kP. where 0 . wb) nr2 = [l 0.811.5. for 0 . [1. If. the overshoot must * These norms on the overshoot are also applied to largeamplitude stepcommands (up to overloading the actuator 10 times). w i t h highfrequency flexible modes.andalsoassurethatmost system in production (with the plant parameters close to the nominal) have much better performance than the worst case.5 11. thenew compensator transfer function must be multiplied by of .2. ds] = lp2lp(n. [ns.4 < k c 2. d. and with hard saturation in the actuator: 0 At o > 8. format short e. for 0.1. 8e k 4 . d = conv([l 7. therise time must be less than 2 sec. If smaller stability margins frequency neighborhood offb increases and the variations in k cause larger deviations of the transient response from the nominal. [l 103). We can now write an example of a design specification for a timeinvariable controller for a plant with a prescribed nominal transfer function P. further.7 34 97 83 0 01. [nrs. the loop gain response is similar to that . The specifications on the overshoot guarantee minimum performance for the worst case ofmaximum plantparametervariations.4 Reshaping the feedback response We proceed next with step A2 from Section 4.4.6 11. Shaping the Loop Response n = conv([ll 55 110 361. dr = [1 2 0. for0 .) Example 4. No step command may trigger selfoscillation. and the loop gain should also decrease with frequency. l/03in order to gainstabilize uncertain flexible modes. [nr2s. under 5%. with shaping the loop response over the functional feedback bandwidth.2.103). the nominal loop gain coefficient should not exceed 0. although the calculations will be performed and. 2 . under 10%. Le. dr2. be under50%. wb) The results will be displayed with single precision (format shorte ) . for 0. 8e k 4 . dr.Chapter 4. 4..811.. 0 The output effect of disturbances mustbe minimized. drs] = lp2lp(nr. wb = 75. under 20%. dr2 = [l 0.
with the normalized frequency o = 1 at the upper end of the sloshmode range. I \ 4.9. 4. To find this response.12(a). The function . To matchaspecifieddisturbancerejectionresponse.13 Reshapingtheloopgainresponse constant slope response.25 . the nominal loop gain needs be to at least 16 dB.15). 4. For control system design it is commonly convenient to use as the normalized frequency the frequency at which the constantslope response gain is approximately lOdB. Therefore.5 1 f.5 Bode cutoff When the frequency components of the expected disturbances within the functional bandf 1 have the same amplitudes.theconstantslopegain response must be modified. 5 Fig. margin over the range where the modes can be. 11.125 . obtained by reshaping the Fig. 4. log.12) (preservation of the area of the loop gain) when the plot is redrawn on arcsinf scale.13 range of modes slosh shows the loop gain response to suit the problem. 4. Response Loop Shaping the shown in Fig. 4. 4.2.The responses redrawn in Fig.t 4 7 arcsin . 4. Propellant sloshing in Cassini spacecraft's causes tanks possible mode slosh a dB structural modes withgainmagnitude reshaped nominal up to 8 dB.25. Example 1. i/ .5). sc. . Numerically. the loss in feedback at lower frequencies can be estimated by application of the rule (3.12) from Section 3.106 Chapter 4.12(a) gives several examples of feasibly reshaped loop gain responses.2 and to response (3) in Fig.x. Fig. Bode made use of the function 009 defined by (3. 4.14. = 20 log 1'11 must be maximized. the loop gain within the functional band should be constant as shown in Fig.12(b)on the arcsinf scale have the same area under them over the frequency interval [O.12 Reshaped loop gain responses on (a) the logarithmic frequency scale and (b) the arcsinf scale Any frequency can be chosen to be the normalized frequency f= 1 (or o = 1). Fig. the phase response at frequencies f > 1 andthestabilitymarginscanbethesameforall showngain responses (recall (3. The value A. To provide gain 16 stabilization with 8 dB upper stability 8 . dB .
Inotherwords.5. the physically realizable bandpass transfer function can be foundby substituting s + (2nfCenterI2/S . the available loop gain A. the frequency band of functional feedback does not include dc.It is seen from the picture (and from the formulas) that this loop gain atf= 1 equalsthe value A. 4.y)dB/oct andthefeedbackbandwidthinoctavesplus1(theextra octave is that from 0. (4.fb %plane x I+functional bandwidth  . 4. A. 4. The band can be viewed as centered at some finite frequency fcenter.200] Hz.4 kHz. and the crossover frequency fb is limited by the system dynamics to 6. y = 1/6. (b) Nyquist diagram on Lplane. Response Loop Shaping the 107 hasthehighfrequency asyTptote withthe slope 2( 1 . that the constantslope response has at f= 0. From (4. dB1 degr " f degr . (a) Bode diagram. for example in vibration suppression systems.5) In the commoncase of 30"stability margin. asshown in Fig.Chapter 4.. Generally.14 Bode optimal cutoff. $3 This simple formula is quite useful for rapid estimation of the available feedback.5 to 1): A. s s 12(1 .14(a). (c) Nyquist diagram on Tplane From the triangle shown in Fig.2. 4.5) the available feedback is 60dB.Le.thefunctionalbandwidthofA. 4.y)n dB/oct. It replacesthe constantslope responsein Fig.dB feedback in the Bode optimal cufoffbecomes extended by one octave. the feedback is required to be constant over the bandwidthof [0.y)(l0g2fb + 1) . The prescribed stability margins are 30' and lOdB. is the productof the slope 12(1.2. 10 (10g2fb + 1) . Example 2.4) (4.5 1 \ (a) asymptotic slope 6n dB/oct \ (b) IC) Fig.14(a).6 Bandpass systems In some systems.
The loop response obtained with (4. each of the points being a mapping ofthe Tpiane point1 onto the Lp I TI. If.11.6) fromthelowpassBodeoptimalcutoffis showninFig. the steepness of the lowfrequency slope has only a small effect on the available feedback since the absolute bandwidth of the entire lawfrequency rolloff is rather small.16 Preservation of operational bandwidth of the bandpass transform Fig. 4. Noticethatin Fig. 4.4. (The bandwidths of the three: responses in the . 417 Bode diagrams for a wideband bandpass system When the: relative functional bandwidth is fairly wide.7 Nyquiststablesystems As mentionedinSection 4.108 Chapter 4. . and 13). A f d f dB dB 0 0 Fig. 4. 4.17 are tailored to guarantee stability of a system with a saturation link in the loop. however.is logarithmic. 4. 4. Shaping the Loop Response for s in a lowpass prototype transfer function [2].) It is seen that a higher fWneter corresponds to a smaller relative bandwidth and steeper slopesof the bandpass cutoff. the responses shown in Figs.2.2. 4.2.16. 4. mme than 2 octaves.16. This case is shown in Fig.6) as is illustrated in Fig. bop phase lag dB degr T Fig. 4. 4. two critical points for the Nyquist diagram @ avoid are shown: 180' and 180'.15 Bandpass optimal Bode cutoff: Bode diagram(a) and Nyquist diagram (b) The absolutebandwidth A .15(b). the system is tiunished with an extra dynamic nonlinear link of special design (described in Chapters 10.12. and as will bedetailed in Chapters 10 and 11. and it equals the bandwidth of the lowfrequency prototype to.15.2.17.picture do not look equal because the frequency scale . of theavailablefeedbackis aninvariant of the transform (4.
This response canbe generated by pasting together several elementary responses P I* dB degr I I %plane Fig.18 Comparison of a Nyquiststable system with an absolutely stable system (thin lines): (a) Bode diagrams.18 instead increased by using the Nyquiststable system loop response shown in Fig. the width of the lower step calculated with (4. The essential features of the response are the steep slope of . and as a result. the systemis only gainstabilized. the slope of the Bode diagram. Here.l). dB degr (4.6nl .0. x1 and x represent the upper and lower amplitude stability margins. (c) Nyquist diagrams on the Lplane Fig. At frequencies where A > xl.19 Simplified Nyquiststable loop response: (a) Bode diagram and (b) Nyquist diagram on the Lplane .6nl dBloct before the upper Bode step and the presence of two Bode steps.7) JB x (a) Fig. 4. The integral of the phase lag in this system is larger than in the absolutely stable system. 4.Chapter 4. and of the upperstep fromfp to fh with the 'similarly derived formula " fh i . of the?phasestabilizing response shown by the thin line. (b) Nyquist diagrams (not to scale). 4. Response Loop Shaping the 109 the loop phase lag. the feedback over the functional bandwidth is larger.19(a) shows a simplified response which is easier to implement (although it provides somewhat less feedback). and the available feedback can be 4.
However. At frequencies of this mode. 4.20 The Nyquist diagram should not penetrate the boundary curve specified by the properties of the nonlinear dynamic compensator.20 is specified by the features of nonlinear links in the loop (nonlinear compensators will be discussed in Chapters 1013).21 bandwidth Feedback definitions (1) The crossover frequency fb. Fig.110 Chapter 4.3 we will discuss physicalconstrains onthehighfiequency loop gain. 4. 4. 4. In the literature and in &g professional language ~f g ~ m o lengineers.20. 4..3 Feedback bandwidth limitations 4.. Inpractice. this definition . In Fig. The stability in suchsystemscan beachievedwithupperandlowerBode steps. the phase stability margin is excessive. Type 1 and Type 2 systems (recall Section 3.7) are Nyquiststable. The phase lag. the Nyquist diagram is shown with a loop on it caused by a flexible mode of the plant. the bandwidth of the loop gain exceeding 0 dB. Le.3. A certainboundarycurye A(B) exemplifiedin Fig.cannot be arbitrarily big. In accordance withthephaseintegralthisreducestheachievedfeedback. thus reducing somewhat the available feedback at lower frequencies.1 Feedbackbandwidth In Section 4.Response Loop Shaping the The larger the integral of phase.the transition between the steep lowfrequency asymptote and the crossover area is most oftenmadegradualtosimplifythecompensatortransferfunctions.butthe feedback deficit due to the loop is rather small since the mode resonance is narrow and the excess in the integral of phase is small. 4.21: dB 30 0 3 Fig. . In this book. this term m y have any of the following &ye lpterpretationsindicated in Pig. the larger is the available feedback. . first we need clarify to the &&&ions of the term feedback ~~~~~~~~~. however.
4. The hump on the response of MI is small. The frequencyf~ where IMI= 1/&.disturbance rejection. 20 logIMI = 3 dB.t No output is N. 4. This bandwidth is also called the bandwldfh of functional feedback. This frequency is the tracking system 3dB bandwidth. In such a system.24. but shallower gain response and smaller feedback at lower frequencies.4. Another wayto do this is tofindtheoutputnoisepowerby integration (on a computer or even by graphical integration) of the frequencydomain noise responses. the mean square amplitude of noise at the system " 2 . On the other hand. i.e. Larger feedback bandwidth leads to larger output noise. the overshoot response is also small. The sourceN represents the sensor noise (here.7&. this respome provides better.. Since from the noise input to the system output. the output noise increases because of the positive feedback at the crossover frequency and beyond. N is understood to be the mean square amplitude of the noise).. ConsidertheBodediagram showninFig.3& to 1.3. When the loop response is steep as shown in Fig. The output mean square error caused by thenoisecanbefoundbycomputersimulationofthe output timeresponses. This response is employed when the plant is already fairly accurate and thereis no need for large feedback at lower frequencies.Chapter 4.22 Sensor disturbance rejection. .23. In Sensor response. therefore.. This causes a substantialincrease intheoutputnoisesincethecontribution of thenoisespectral density to the mean square error is proportional to the noise bandwidth.the bandwidth of 30dB feedback).Thisloopgain response has a rather steep cutoff after& to reduce the output noise effect.g. The frequency up to which the loop gain retainsaspecifiedvalue (e. The loop response should be therefore shapedin each specific case differently to reduce the total error.. The frequency is typically from 1. the system can be viewed as a tracking system w i t h unity feedback.22.commandfeedforward is commonlyusedtoimprovetheclosedloopinputoutput response. the tradeoff is between noise effect at system's output output noise reduction and Fig. 4. ~ shaping Example 1. consider the system shown in Fig.and positive feedback near the crossover frequency should be reduced to reduce the output effect of the sensor noise. Shaping the Loop Response 111 for feedback bandwidth is accepted. The phase stability margin is in the transient large. = NTlF .2 Sensor noise at the system output Next. but smaller disturbances. 4.
the noise effect at the input of the saturation link is Fig.25 Spacecraft attitude control system using two sensors 4.. 4. The optimal frequency response for the star trackerloopisthat whichreducesthetotalnoisefromthetwosensors. 4.24 Steeper slope response Example 2.26 Noise source in a feedback system .which reduces the mean square error of the system output variable. depending on whether bright stars are available in its field of view. The drift is eliminated by a lowfrequency feedback employing a second sensor. Consider the spacecraft attitude control system in Fig. Shaping the Loop Response dB 0 Fig. the noise sourceN reflects the noise from the sensor and the noise from the preamplifier in the compensator. The system is accurate except at the lowest frequencies where the gyro drift causes attitude error.3 Sensor noise at the actuator input In the control system diagram in Fig. I tracker Star Fig. 4. Such an adaptive system is illustrated in Chapter9.3. .Le.26. the feedback path responses need to be varied to maintain the minimum of the error.When the signal at the input of the saturation link is below the saturation threshold.23 Shallow slope response Fig. The calculations can be performed in the frequency domain or by using the LQG method described in Chapter 8. a star tracker.112 Chapter 4. 4. Since the star tracker noise varies with time. 4.25 which uses a gyro as a sensor. 4.
can be particularly large. If a better amplifier and better sensors become available with half the noise mean square amplitude.26. This restricts the available feedback in the operational band by constraining ICI. 4. the noise NA is most prominent at the frequencies above&. on the feedback feasibility can be critical for loops including a substantial delay. This limitation.3. which causes a phase lag proportional to frequency. 4.26. On .p.27 Noise level at saturation link input With the typical responses of P and T shown in Fig.e. and the control systemaccuracydecreases. . the effective gain of the actuator drops. Transport delay. At these frequencies the noise N A = N/P does not depend on C . NA = NCA. 4. the actuator cannot transfer the signal. the Bode step would have to be very long. ICAPI >> 1 at lower frequencies. or else to compensate for it. When the noise overloads the actuator. With proper loop shaping. k t us consider two examples of audio systems with large transport delay.Chapter 4. the responses which are best in this sense contain Bode steps.Hence. the bandwidth is limited by the noise at the input to the actuator.4 Nonminimumphaseshift The n. Typically. Example 1. the feedback bandwidthbe can increased 1. 4. the feedback bandwidth can be increased. at the priceof a bigger noise effectat the input to the nonlinear link in Fig. and reducing ICI decreases the noise. 4. within 2 to 4 octaves It is seen from Fig. Shaping the Loop Response 113 dB 0 Fig. is attained at the price of increasing CAI to ICAI.4 times (since the mean square amplitude of the white noise is proportional to thesquare root of the noise bandwidth). In an existing system.the other hand.thenoiseeffect at theinputtothe actuator must be bounded. Maintaining the same mean square noise amplitude at the actuator input.27 that the increase of the feedback fromT to 7". at frequencies higher than &.. The noise amplitude and power increase not only because C > I C I . i. As a result. The optimalshape of theBodediagramwhichprovidesmaximumfeedback bandwidth while limiting the noise effect can be found by experimenting with computer simulation. the distortions of the signal increase. lag in the feedback loop typically should be less than 1rad at 5. but also because the noise power is proportional to the frequency bandwidth.
We might suspect that the reason is the excessive time of the signal propagation about the feedback loop.28? Probably not. The effects of this delay will be analyzed in the next chapter.29. which is typically the case of the largest plant gain. should not exceed 1 rad.000)/2n 0.frequency responsesare difficult to equalize.28 This type of real time acoustical feedback system is not feasible Fig. the feedback will be smaller in the case of the minimum . and the distance I between the speakers and the microphone being 2 m. the pure delay is (1/40. the plant responses and their variations ire commonly smooth and monotonic.Thus. In plants whose transfer functions have only real poles and zeros. 4.5 Planttolerances The plantgaintypicallydecreases withfrequency. 4. since the sound wave reflections from the room walls change the frequency responses on the way to listener's ears. noise source Y  Fig. A system for noise rejection is diagrammed in Fig. The acoustic signal propagation between the speaker and the microphone introducesnonminimum phase lag into the feedback loop.0066 = 2500rad which is 2500 times theallowablelimit. There must be a good reason. 4. the distance between the microphone and the compensating speaker shouldbe short.) Example 2.B.30. Another source ofnonminimumphaselagisthedelayintheanalogtodigital conversion when the control is digital.andthetolerances of theplant transmission function increase with frequency as illustrated by the limiting curves in Fig. since nobody does this.6 msec.it wouldbecommerciallyadvantageoustomake acoustical feedback from a microphone placed in the vicinity of one's ears. The microphone is placed at the point where it is desired to keep the noise minimum.114 Chapter 4. 4.000Hz using a feedback system like that shown in Fig. Response Loop Shaping the Example 1. Is it possible to maintain good sound quality over the entire range of audio signals up to 15.realtimefeedback inthissystemisnotpossible. At = this frequency. then frequency fd = 20kHz. When 30 dB of feedback is required up to 5 kHz.(The response canbe equalized by an adaptive system using plant identification.000008sec andthedistancebetweenthespeakerandthemicrophoneshouldbe shorter than 2. the phase lagB&) = 27c x 60. and since good quality inexpensive microphones are easilyavailable. For the frequencyfc = 15x 4 = 60 kHz. 4. To reduce the phase lag. 4. the transport delay is 6. The sound from the speaker cancels the noise.29 Soundsuppression feedback system The speed of sound being 330m/sec. Since speaker systems are expensive and their .000 x 0.6 mm. Let us check it out.3.Sincetheminimum necessary stability margins must be satisfied for the worst case. Therefore. The assembly of the microphone and the speaker is commonly mounted on a helmet. Such responses are typical for temperaturecontrol andrigidbodypositioncontrol.
andtherebylimitthe nominal available feedback. Uncertainties are also sometimes modeled by vector addition of some error response tothe transfer function(addifive uncertainty). sc. 4. 4.31 Plant gain frequency Fig. 4. plants composed of rigid bodies connected with springs and dampers. as shown in Fig. For computer design and simulation. i.3 (see Fig.havestructuralresonancescorrespondingtospecific modesofvibration.e. I dB dear 1 ain plant phase lag dB I nominal ~ 0 I 0 t log.that is multiplication of the loop transfer function by some error response (i.largerdeviationsfromthe nominal requireincreasedstabilitymarginsforthe nominalresponse. Fig. 4. 4. 4.31. like those shown in Fig.Response Loop Shaping the 115 plantgain. For some plants. The multiplicative uncertainty is typically either a constant as in Example 1 in of frequency..30. The feedback loop design is then performed for thenominalplant. This way theplantresponsetolerancesreducetheminimumguaranteed feedback. Fig.11) or a function In general. the plant uncertainty is most often modeled as mulfip/icative uncertainty. addition of some uncertaintyto the gain and phase responses).2.Largerplanttolerances .30 Boundaries of monotonic responses functions transfer plant For plants with monotonic responses.Chapter 4. 4.32 Plant structural resonance on a Bode diagram Fig. parameter uncertainty causes deviations from the nominal plant responses which are neither symmetrical nor monotonic.30.e. 4. as those shown in Fig. Stiffness and mass variations in flexible plants change the pole and zero frequenciesas . it is convenient to consider some nominal plant response.. the dependenceof the plant transfer function on its varying parameters can be complicated.33 Plant structural resonance on an Lplane Nyquist diagram Flexlcble plants. Section 4.
so the resonance loop will be kept away from the critical points180' and 540' as illustrated in Fig.32 and 4.4. Similar responses are obtained in lowloss electrical systems as such transformers and filters. The modes which need attention are the modes that are not already gainx to x1.34(b). 4. collocated and noncollocated control Some of thepolesandzeros of flexibleplants'transferfunctions are onlylightly damped. Increasing the modal damping can reduce the valueof the modal peak and notch and gainstabilize the mode. stabilized. with the damping coefficients as small as 1% and even 0. The loop gain as inthe responses of the loops withsuchplantsexhibitsharppeaksandnotches examples in Figs. 4. 4. I ' I mode 1 dB Fig.4. 4. Response Loop Shaping the shown in Fig.1%. The required phase lag can be obtained by introducing a lowpass filter in the loop (this is a better solution than adding n. the mode needs to be phasestabilized as shown in Fig. If the plant's phase uncertainty at the frequency of the flexible mode is large.34.3. lag since the filter will provide the additional benefit of attenuating modes of higher frequencies). The average loop gain at the frequencyof the mode must be no higher than (20 log Q + x) dB. The resonances typically produce loops nearly 180' wide on the LplaneNyquistdiagram as shownin Fig. those resulting in the loop gain falling within the interval from as the modes 2 and 3 shown in Fig. .34(a).p.34 Modes (a) on the loop gain response and (b) on the LplaneNyquist diagram To phasestabilize mode 3. 4.116 Chapter 4.32.33. i. A typical case of the gain stabilization of a structural mode is showninFig. The Bode step allows a steep rolloff at frequencies beyond the step.35.34. Damping of the highfrequency 'mode would allow increased feedback bandwidth. a phase lag might be added to the loop to center the mode at the phase lag of360°. 4.the modesshouldbegainorphasestabilized.e. For the closedloopsystemtobestable.6 Lightly damped flexible plants. Otherwise.. phasestabilization is not feasible and the mode must be gainstabilized. It is better to describe such plants by specified uncertaintiesin the transfer function poles and zeros. Gainstabilization of the mode reduces the feedback in the functional band.Neithermultiplicativenoradditive uncertainty conveniently characterizes these effects. 4.
. 4. d = conv(d. 4. ) . the control would be collocated. so the control is called collocated. 4.than the mass of the main body which explains why the poles are rather close to. i Fig. 1000). If the mode polezero order were reversed. and then the mode's pole and zero follow.Chapter 4. consisting of rigid bodies with masses MI.36 Mechanical plant with flexible of a highfrequency mode appendages and the sensor collocated with the actuator Next. 0 X i i s2 1 L"""".Flexible plantsare discussed in more detail in Chapter 7. . in fact. a pole follows a pole. d.w) .. the plant driving point impedance (or mobility). and its phase belongstotheinterval [go".17 dB r.e. connected with springs.35 Gainstabilization Fig. and the phase of the plant transfer function alternates between 90" and 90". 9 0 ' 1 . [l 0 5 4 1. 1.35 is noncollocated since the plant has a pole at zero frequency.36 with n = conv([l 0 101..Response Loop Shaping the 1. The motion sensor S I is collocated with the actuator and senses the motion of the first body. the transfer function from the actuator force to the sensor SIis. i. M2. Example 1.. s s2 +11 s2 +54 The masses of the appendages in this example are approximately 10 times smaller . The driving point impedance of a passive system is positive real (see Appendix 3).32and7. The gain and phase responses are plotted in Fig. If the sensors are velocity sensors.. [l 0 501 1.36. w '= logspace(0. 4. [l 0 111). d = conv([l 01. 4. the zeros. A collocated forcetovelocity translational transfer function of a body withtwoflexibleappendagesresonatingrespectivelyat3. bode (n. Theimpedancefunction of alosslessplanthas purely imaginary poles and zeros whichalternate:alongthefrequency axis. consider flexible plant model for translational motion shown in Fig. Example 2. The actuator applies a force to the first body. The plant of the control system having the loop response shown in Fig.35 radsec and negligible damping is 1 s2+10 s2+50 P ( s ) = .
on MI. on M3. the sensor location defines whether the control is collocated or noncollocated. on M2.8. is very often encountered in practice.on Mz. Fig.37. Thus. the feedback bandwidth can be widened but it is the positionof the first body that is controlled.. When the sensor is placed closer totheactuator.37 Bode diagram for a collocated mechanical plant with two flexible appendages Placing the sensor on any other body makes the control noncollocafed. Response Loop Shaping the When the sensor is a position sensor or an accelerometer. Le. 4. A Nyquist diagram for a flexible plant (Saturn V controller) is given in Appendix 13.In this case the spring connecting the bodies introduces an extra unwelcome phase lag into the loop.(However.8. The collocated and noncollocated control will be further discussed in Sections 7. but the feedback bandwidth must be reduced as shown in Fig.34(twolowerfrequencymodes)and4. a flexible appendage adds . or on M 3 in Fig. The tradeoff associatedwith where to place the sensor.a zeropole pairto the loop response as showninFigs.4. The modes do not destabilize the system since the phase lag only decreases by 180' between the addedzeropolepair. Many flexible modes are seen on this .and therefore the average gain slope and the available feedback somewhat decrease.4. A13.themodereducestheintegral of phase.6 dB/oct or6 dB/oct.26. The sensor must be placed within the power train someplace from the actuator to the tip of the tool or other object of control. While the .3 and 7. When the control is collocated. The flexibilitybetweenthebodies will introducetheerror intheposition of thetip.then we are controlling exactly the variable that needs to be controlled. i. The best results can be obtained by combining these sensors.5 (3 ' 0 50 10" 1 10' Frequency (radsec) 4 0" 10' Frequency (radsec) Fig.118 Chapter 4. 4.e.36.diagram. However. 4.or on the tip of the tool.) 50 8 .35.. an extra integrator or differentiator should be added to this transfer function which changes the slope respectively by . on MI. when the sensor is placed on the tool. Example 3.
4.an internal feedback path having a lowpass transfer hnction Bint = b/(s + a).an unstable plant can be equivalently presented as a combination of a stable forwardpath link P with internal feedback path Bint that makes the plant unstable. The loop phase lag at the mode's central frequency is 315". The plant parameter variation must not reduce this phase by lagmore than 45' or else the stability margins will be violated.7) and would make the system unstable.gainstabilization of themodesisrequiredand is implementedas shown in Fig. 4. 4. In Example 1 of Section 4. . compressibility of the air in the cylinder of an actuator creates a series "spring" between the actuator and the plant. 4.Response Loop Shaping the 119 controller was being designed. and the plant becomes unstable as seen from the Nyquist diagram in Fig. rotation of a prolate spacecraft is unstable.3. dB 1plane 0 Fig. For the purposes of analysis and design. For example. Most of the modes are gainstabilized. and the diagrams of (b) Bode and (c) Nyquist for the internal feedback Example 1. Because of this.7 Unstableplants Unstable plants are quite common. Example 5.Chapter 4. A digital controller with insufficiently high sampling frequency would cause large phase uncertainty (as will be discussed in Section 5. 4.38 (a) Plant with internal feedback. Assume that the plant is a double integrator with .14. This makes the control noncollocated and reduces the available feedback bandwidth. as shown inFig. In pneumatic systems.38(a).38(c). The control at these frequencies is analog. a largegain electronic amplifier without external feedback circuitry is often unstable. or closer to the engines where the control would be collocated and would be easier to implement.38(b). serious discussions were going on about where to place the gyros: closer to the location that needed to be better controlled (in this case the control would be noncollocated). the SaturnV launch vehicle and some airplanes are aerodynamically unstable. Consider the system diagrammed in Fig.10. It was eventually decided to play it safe and placethe gyros closer to the engines. The internal loop phase lag exceeds 180" at all frequencies. only the large mode seen in the upper right sector of the diagram is phasestabilized.38(a). the slosh modes are noncollocated. 4.5. a slug formed in the combustion chamber or a turbulence in the chamber can make a rocket unstable. Example 4. The Bode diagrams are shown in Fig.2. 4.
and P. A. Here.4 Coupling in MIMO systems As stated in examples in Section 2. Fig. 4. The gainstabilization intherangeoftheflexiblemodeisthebestchoicesincephase . a solar panel. thrusters).40 Block diagram for the coupled Because of the coupling. = Cy A. First.and will provide signals to all the sensors. A. the yactuator stable and robust. = C. In this case. This coupling may occur at any frequency within a certain frequency range defined by uncertainty in the mass and stiffness of the appendage.asrequired by theNyquistBodemultiloopstability criterion. uncertain.39 Mechanical plant a with flexible appendage loops attitude control Fig. 4. coupling is typically negligible in well designed conttol system 'where the number of the actuators is kept small. at some fkequencies the coupling can be large. the required transfer function for the compensatoriis C = (TIP .The method is especially convenientwhen Bint is small compared with BCA.However.Bini)l(AB).120 Chapter 4. Then. K(s) is the coupling transfer function. 4. When the loop is disconnected at the input to the link P. After the desired frequency response for T is specified. The attachment's flexible mode can be excitedby any of the actuators (reaction wheels. 4. and T.Response Loop Shaping the There aretwo convenient waysof analyzing and designing such systems. is 1 + (Tx/Fx)K2. However. or a magnetometer boom on a spacecraft. thisis not possible). the mainloop Nyquist diagram must encompass the critical point the in counterclockwisedirection. like an antenna. The return ratios for the controllers in x and y calculated without taking the coupling into account are T. The transfer fbnction in the ychannel between A.9. the block diagram for the coupled loops looks like that shown in Fig. The compensator Cyis then shaped properly to make the system stable (sometimes. P.the loop transfer Eunction is T = (Bint + BCA)P.4. is switched on (while the xactuator is kept on). the actuators are typically applied in mutually orthogonal directions so that the coupling between the correspondingfeedbackloops is relativelysmall.B y . Thesystemcanbedesigned with theBodeNyquistmultiloopsystemstability is disabled and T.39. The compensator can be directly designed for the unstable plant.theplantmightinclude some flexible attachment. In mechanical structures with multidimensional control. P.40. and create stability problems. as shown in Fig.B. is shaped so that the xloop is criterion as follows.
As was demonstrated in Chapter 3.p. linksW l and W2 are connected in parallel as shown in. then the total transfer function W l + W2 can become n.and yloops.p.41. xactuators produce rotation about the yaxis. Response Loop Shaping the 121 stabilization is difficult because the transfer function K2 = u2M/(s2 + 2co + q .(b) yloop while xloop closed Coupling betweenx. The effect is only profound at frequencies close to the frequency of rotation about the zaxis. Fig.Chapter 4. This compound feedback makes the loop transfer function less sensitive to plant parameter variations. 4.severalpathsareoftenconnectedin parallel. each control loop should use position (or 4*42 xypositioner velocity) and force sensors.Bodediagramsforstabilityanalysis withthesuccessiveloopclosure criterion. especially when several actuators or sensors are employed. if two stable m.p. The system analysis and design are similar to those in the case of flexible mode coupling. The frequency responses of the parallel channels should be shaped properly for the combined channel to be m.Inthiscase. Loop decoupling algorithms can then be used effectively. and yactuators produce rotation about the xaxis. If gainstabilizationcannotbe usedandthesystemneedstobephasestabilized. 4. 2 ) 2 contains double complex poles with large associated phase uncertainty.u is some coefficient. 4.41 Gain responses of the attitude control loops: (a) xloop while yloop open. ' Fig. This also reduces the variationsof the loop coupling that are caused by variations of the load and the plant parameters.43(a).5 Shaping parallel channel responses In MIMO andevenin SISO controlsystems. should be modified to make the response in the xloop shallower over the frequency range of coupling. . here. especially at the frequencyof the structural modeof the rotation. C. 4. for the x.and ycontrollers can be also caused by the effectsof rotation aboutthezaxis(even in aspacecraft without aflexibleappendage). The associated reduction of available feedback is unavoidable. The translational motions may become coupled via rotational motion due to the load asymmetry and due to yWrails structural flexibility. may look like those shown in Fig. When the number of the actuators is large (there are many thousand separate muscles in the trunkof an elephant). Design of a loop with prescribed actuator mobility will be discussed in Chapter 7. An xy positioning table is shown in Fig. 4. and makes the output mobility of the actuator dissipative and damping the plant.42.
and W 2 + WI.122 Chapter 4. if and only if the Nyquist diagramfor W1/W2. and w.When the phase difference between the channels at fi is Z outputs of the links cancel each other and therefore. &g hodograah of Wl/W7. + W2. theycanproducelargevariations in W. If the slopeof W2 is gradually changed.1 . the phase difference The between the two channelsis. 4. thin lines show the logarithmic responses of IW. the root loci for the zeros of the transfer function cross the jooaxis as shown in Fig. 4. the composite link transfer function has a pair of purely imaginary zeroskj2xfi. Since the ratioWllW2 is also stableand m. When the tolerances of the parallel channel transfer functions are not negligible. +W2 dW2  1 l+Wl/W. + W21 obtained by vector additionof the .the links' output signals. and Example 1. Thesensitivities of thesumtothe components. (b) Bode diagrams for WI. and more than IC. To constrain the sensitivities. 4. Shaping the Loop Response Fig.p. should be required not to penetrate the safetv margin around .p.does not encompass the point1. one can determine whether the' Nyquist diagram encloses the critical by point examining the Bode diagram forW11W2.43(c) and the total transfer function becomes n. The lowpass links Wl and W2are connected in parallel as shown in Fig. .43 (a) Parallel channels. As has been proven. 4.43(b). At frequency fi where the gains are equal. respectively. the sum Wl + Wz is m.p.. (c) splane root lqci for WI + W ' W2. lessthan n. ' W2 become unlimited as the ratio Wl /W2 approaches .43(a). The steep rolloff W1 and the three versions of shallower rolloff W2are shown in Fig. equal to n.
A common practical reason for using two parallel links is that one of the links (actuators.Response Loop Shaping the 123 the point . so that it can be included in the feedback loop.Giventhetransfer function of the first link.44(a) shows the responses of the lowpass linkWland the highpass link with different selectivities.45(b). and the amplitude safety margins.44(b) shows the Bode diagrams for the W ratio l /W2. and the selectivity is high.linksareimprecise(likeactuators and differentsignalpaths through the plant). excessive selectivityof the filters can make the composite n.p.W.I . Analogous to the stability margins.45(a). link W2. 4. 4. 4. then thelink parameter variations should be accounted for and sufficient safety margins introduced. 4. Combining them with frequency selection filters generates a link (actuator.ses and then approximating them with rational functions is directcalculation of thechanneltransferfunctions. the transfer function of the second link can be found directly as W 2 = 1 . t \ Fig.7. and the second link works better at higher frequencies.and W 2 responses increases.and W kind are particularlyusefulforsystemswiththemainvernieractuatorarrangement 2. character ofW1 + W2 . or sensors) works better at lower frequencies. as x and x ] .Wl. sc. W1 and (b) WdW2 In order to preserve sufficient safety margins while keeping the slope of Wl/W2 steep. However. 4. Then.Chapter 4./W2 canbeshapedasin a Nyquiststablesystem.p. log.theBodediagramfor W. 2 can be as illustrated as in Fig. 10 x f. the Bode diagram for the ratio steepens. It is seen that when the difference in the slope between W. \\ \ vv1’vv2 Fig. log.This method works well if the links are precise (as when sensors’ readings are combined). Fig. or sensor) that good is over a wide frequency range. Responses of this Fig. the related phase lag increases. described in Section An alternative to shaping the respon.44 Bode diagrams for (a) W2. the phase safefy margin is defined as yn. 4. If the.10 f. The composed link transfer function must be m. and the critical point becomes enclosed by the related Nyquist diagram. Fig.p.45 (a) Frequencyselective responses forWl and W2 and (b) Nyquiststable shape of Bode diagram for W1 / W2 which preserve m. sc.
filter: w. To improve the selectivity.r. =Y w 2 = S s+a is a highpass. log. W2 W. a y=.. the first link is chosen to be a secondorder lowpass Butterworth. and the ratio = s(s +4 2 ) (4. =1w. The phase safety margin is 90”at all frequencies. but theselectivity between the channelsis not high.10) s2 + S J ? : + l ’ dB f. 10 Fig.= 1 s2 + s & + 1 ’ The second link transfer function is then w2= 1w. If W. The margin is large. and the ratio a .  . s+a then the second channel transfer function w.46 Bode (a) and Nyquist (b) diagrams for WI. 4. sc. Example 3. Shaping Loop Response the Example 2. wi/W.is a lowpass.124 Chapter 4.
3 Sketch the phase lag response and the Nyquist diagram for the optimal Bode in cutoff Wich the Bade step is omitted. Make a sketchof the loop gain response similar to that shown in Fig. express & as a function of fet and Q. (d) the asymptotic slope is 24dB/oct. lag at 100 Hz is0 .47. 6 The sameas in Problem 4.47 Feedback bandwidth limitation due to a structural resonance 5 The samea@ in Problem 4. . (b) the asymptotic slopeis 1 2 dB/oct.numbers and should be used for initial estimates). 6rad. 5 rad. completely uncertain. lag at 100 kHz is0 .6 Problems ' 1 The slope at the crossover frequency 40 Hz is (a) 6 dB/oct (b) 9dB/oct 1. 4. the asymptote passes the 0 dB level at 5 kHz.p lag at5 kHz is 1 rad. the asymptote passes the 10 dB level at 100 Hz. 4. Digital controliep will be studied in detail in Chapter 5. the asymptote passes the 1 0dB level at 1 kHz. (The phase uncertainty is the result of using a digital controller with sampling frequency100 Hz. and the n. and the n. (e) 50dB. and assuming thatQ = 2&. and then.25 Hz if the slope (c) 1 0dB/oct (d) 12 dB/oct. Is the system stable? 4 Atthefrequencyofstructuralresonance fst = 120 Hz or higher there is a narrow resonance peak in the plant gain response and the plant phase at this frequency is. 5 rad. Using the asymptotic slope 18 dB/oct. . and Jb = 46 (which are typical . (c) 30 dB. for the peak value 1 2 0 log Q equal to: (a) 20 dB. with lowfrequency slope 10 dB/oct and asymptotic slope 1 8 d6/oct. 125 4. the asymptote passes the 10 dB level at 50 kHz. x = 10 dB. (c) the asymptotic slope is 12 dB/oct. (b) 25 dB.p. but the resonance uncertainty range starts at 85 Hz. openloop gain response Fig.p. lag at 1 kHz is 0 .Response Loop Shaping Chapter the 4. dB IQ .p. What is the feedback in dB at remains the same down to this frequency? What would be the loop phase lag if the slope is constant at all frequencies? 2 What is thelength of the Bode step and fb if (a) the asymptotic slopeis 1 8 dB/oct. but with numbers on it. (d) 40 dB. but the resonance uncertainty range starts 170 atHz. 3 0 ' and 10 dB stability margins. and the n. y = 116.) The system must be therefore gainstabilized at the frequency of the structural resonance f s t .
What is the conclusion? Explain the result by referring to the shape of the weight function in the Bode phase gain relation. (b) 2(10 .70]Hz.loop is 1 mm.30]. (d) 6 kHz.4]rad/sec to the bandwidth [40.2. the bandwidth [30. (b) 18dB/oct.(b) [0.with 3 0 ’ stability margins.2. The. 11 rad/sec to the bandwidth [30.5kHz where there might be flexible modes in the plant. (e) 2 kHz.70]Hz. 13 The actuator and plant transfer function AP is: (a) 4(10 . (d) [0. Usethe 1 rad/seccrossoverprototypedescribed in Section 4. (9 [0. and the crossover frequency isHz? 300 9 The required feedback at 10 Hz is 40 dB. 1001 Hz.126 Chapter 4. (c) [0. 1201Hz. (c)2 GHz. (d) 0 .72(10 s)(s + 6)4 (1 0 + s)(s + 1 I)]. 51 radlsec to.5 Hz if the crossover frequencyis 300 Hz and the main slope is 1 0dB/oct? (b) Whatis the maximum available feedback in dB1.70]Hz.2 Hz.3]radlsec to the bandwidth [60.000 kdsec. The system is phasestabilized at all frequencies up to&. (b) 6 Hz.8 oct and x = IOdB? 1 10 In a system with feedback bandwidth 100 Hz.70]Hz.~ ) ( ~ + 5 )~4 O2 + S ) ( S + lo)].S)(S + 2)/[8(10+ S)(S + 8)]. (e) 2. (I) 100 radkec. amplitude stability margin10 dB. Calculate what attenuationat 1. (cj[30.6 at Hz when the feedback is kept constant at frequencies below 1. (c) 3(10 S)(S + 4)/[S2(10+ S)(S + 9)]. the attenuation in the feedback loop is required to be large over 1.s)(s + 2)/[&10 + s)(s + 7)].(d) 2 to 3 GHz? the loop transfer function and the prefilter for the system with crossover 12 Find frequency: (a) 0. 6 ( 1 0 . (f) 6 MHz. (h) 4 radlsec.5 to 3 GHz. and the feedback should increase at lower frequencies with the slope 10 dB/oct. aGaAs microwave feedback amplifier.60]? 8 (a) What is the feedbackin dB at 1. (e) [O. (b) 1. and no n.6Hz.7s3 +34s2 + 9 h + 8 3 s2 s+lO ” * 14 Determine the bandpass transform from the lowpass optimal cutoff with frequency range (a) [0.with Bode optimal loop response.p. (c) 2 kHz. the system is phasestabilized. and the length of the feedback . there are two gain stages. phase stability margin 30°. (b) [0. = 1 rad? Considering this frequency as &. Findthecompensatorthatmakesthelooptransferfunctionthesameas example studiedin Section 4. What is the crossover frequency? What are the frequencies at the beginning and the end of the Bode step if the step’s length is 0. At what frequency is S. 11 In.what is the length of the Bode step? What is the available feedback over the bandwidth (a) 0 to 3 GHz. (9) 2 MHz. lag. . What is the available feedback over the bandwidth (a) [0.Response Shaping Loop the 7 The crossover frequency fb is 1 kHz.where   b in the 1 10 l ls s3 + 5 5 s 2 + l 1 0 s + 3 6 1 s+10 T(s)= C ( S ) ~ ” = s 1O+s s4 +7. .2]rad/sec to the bandwidth [50.50]. y = 1/6. 11 rad/sec to the bandwidth[30.5kHz is available in theloopresponsewithaBodestep if the asymptotic slope is chosen to be (a) 12 dB/oct. speed of signal propagation ‘is 150.3.3.
Devise a counterexample to prove that optimal shaping for the Bode diagram must be used for initial analysis as well. and the rest is as in Problem 19. 0 dB level at200 kHz. from 80 to 320 HZ.) 17 Using the phasegain chart in. calculate the phase response for a Nyquiststable system whose loop gain response is: from 0 to 10 Hz. 50 dB/oct. Loop Response 127 15 Draw a Nyquist diagram on the Tplane (not necessarily to scale. from 20 to 80 Hz.5)(s + 4) (c) 4 0)= (s + 20)(s + 50) (s + 1O)(s + 16) (d) w. Calculate the frequencies at the ends of the upper and lower Bode steps if the slope at lower frequencies is (a) 1 2 dEVoct. 4.5)(s + 5)(s 8000 " + 10) (s + 6)(s + 20)(s + 40) Is the composite linkW. 16 Initial analysis with a loworder compensator has shown that in the plant hardware configuration A. from 320 to 640Hz. 1 8 dB/oct. How to approach the design of the feedback system? 22 Two parallel links have the following transfer functions respectively: (a) & 0)= + 5)(s + 8) (s + 40)(s + 100) (s (s 1800 6)(s and W2(s) = 1000 (s + 3)(s + lo)($+ 20) ' (s + 4)(s (b) + + 10) (s + 50)(s + 125) (s + 2. largerfeedback is availablethan in configuration B.? ) 23 Two minimum phase linksin parallel made a nonminimum phase link.(s) + W ~ S m. 20 The loop gain plot crosses the versions (a) and (b). 4. later. 60dB. How would you .p. 1 0dB. he decided that configuration A should be chosen. and the asymptotic slope is (a) 1 8 dB/oct. the phase stability margin is30°.Thesystem engineer assumed (wrongly) that feedback will be larger in configuration A even when. 10 dB. (Hint: Use a plant with a fle mode. Therefore. (b) 1 8 dWoct. 18 An extra management level was added to a fourlevel management system. but only by 20 dB larger over some narrow frequency range.1 9.Fig. w How i l it affect the speed of accessing the market and adjusting the product quantities and features (make a rough estimate)? 19 In a Nyquiststable system with a response like that shown in Fig.42 (or the program from Appendix 5).($1= (s + 8 0 ) (s + 200) 4 (8) = and and and W2(s) = W2(s) = + 13)(s+ 27) ' 125 W2(s) = (s + 1. a better controllerwill be developed.and the upper and lower gain stability margins are 10dB.19 . A third link with constant gain coefficientk has been added in parallel to the two links. 1 0 dB/oct. from 640 Hz. (b) 24 dB/oct. The path from the plant output to the plant input via the regular feedback loop links is 40 dB larger than via path € ? at I most frequencies. from 10 to 20 Hz.Shaping Chapter the 4.& = 100 Hz. only to show the shape) for the Nyquiststable system in Fig. 21 The unstable plant canbe equivalently represented as a stable plant with a feedback path Bl. 3.
itis helpful for control engineers to have prepared plots like thoseshown in Fig.435.e.49. .22 kHz.I length is 3.exemplifyingtheavailabledisturbancerejectionfor two structural resonance frequencies.& = 0.4.32 log2.2) fd& = 0 . the transfer function is m.? 24 Find transfer function G(s) of a highpass in parallel to lowpass 12/[(s + 3)(s + 4)] so that the total transfer function 1. the slope can be increased to 12 dB/oct as shown by the dashed line. (a) Sincey = 116. then.18 .disturbancerejection in dB at frequency f e &/2depends on the feedback bandwidth fi.48.7 1. and fb = t42 = 0.3 oct down to 50Hz fromthecrossover. approximately as 10 log4fat).35 . the feedback 6 is x 5 = 30 dB.3 + 1) x10 = 53dB.32 log(lO00/50)= 4.2 oct). The diagramis shown in Fig. kHz.p. stability as described in Chapters 10 and 11.p.5). 4. There are 3. with any other passive source and load impedances. with some passive impedances of the signal source and the load.5 = 2.) When discussing the tradeoff between the resonance frequency and the resonance quality (Q) of the object of control and the available disturbance rejection at a meeting with mechanical designers.3 (i.p. Answers to selected problems (a) 1. from (4.according to (4. the slope is 1 0 dB/oct.25 Hz is 5 octaves below 40 Hz. find the minimum kfor the total transfer function to be m.Then. At lower frequenciesf e a/6. fst = 50 Hz and 100 Hz. 6 ~ 3 + 0. Therefore.5 3 6 Fig.128 Response Shaping Loop Chapter the 4. the step (a) The frequency & =. 4.3 = 1. a nonlinear dynamic compensator must be employed to assure global. (In this case.. I s isthis G(s) realizable? 25 Prove that if a transfer function of a linear passive twoport is m.48 Loop response Bode diagram With the slope of the Bode diagram 10 dB/oct. .theavailablefeedback is (4.
5 3 6 12 Fig.Q 129 E^ 60 50 40 ’& ‘8 e *o 2 8 4 30 10 0 .Chapter 4.751. 4.49 Dependence of disturbance rejection on the structural mode frequency and damping .Response Loop Shaping the 70 .
and computer code for the first. This is demonstrated and recommendations are formulated about how to construct a compensator by cascading the links of the firstand secondorder. Approximation of an arbitrary constantslope gain response based on asymptotic Bode diagramsis described.l(b) shows the relation between the nipple amplitudes of the gain and phase responses. A complete compensator design example is presented with the derivation of a prototype analog compensator including linear and nonlinear links. In 130 . The effect of aliasing is described and the loop response is considered for the reduction of the aliasing errors with reduced penalty to the available feedback. From the asymptotic Bode diagram the poles and zeros of the compensator transfer function are immediately evident. and it is shown how to use them to increase or reduce the slope of the compensator Bode diagram. The allocation of transfer function poles and zeros to cascaded links affects the compensator dynamic range. Lead and lag links are defined. The slope of the segments of the diagram is 6n dBloct.I Chapter 5 COMPENSATOR DESIGN Compensators. are commonly built having rational transfer functions. and it is shown that the Tustin transform is adequately accurate for practice. The asymptotic Bode diagram is used to approximate the conceptual Bode diagram for the compensator (which may include segments of any slope). the ztransform introduced. a rational approximation to the ideal transcendental response is almost always required.l(a) shows the ripples on the responses of the gain and phase caused by the Chebyshev (equiripple) approximation of a desired constantslope response by a rational transfer function. The accuracy of the constantslope implementation directly affects the value of the available feedback. Fig. The trapezoidal integrator is analyzed. The Laplace and Tustin transforms are compared. equations of compensator links.and secondorder digital compensator links. the better can be the approximation andthesmallerwillbetheripples. The higher the order of the rational function. A digital compensator can be viewed as a modification of an analog one in which the analog integrators are replaced by discrete trapezoidal integrators. where n is an integer.1 Accuracy of the loop shaping The desired frequency response of the compensator has been presented in previous chapters in terms of curves of logarithmic magnitude and phase versus logarithmic frequency. A setof normalized plots is presented for a secondorder lowpass functions having complex poles with different damping coefficients. their asymptotic responses shown. in software or hardware. 5. For implementation. 5. Compensators consisting of a parallel connection of links can be made in such a way thateachofthelinksdominatesoveracertainfrequencyrange.Fig. and computer code. and the conversions of the polynomial coefficients tabulated. AnasymptoticBodediagram is piecelinear. 5. Tustin transform zfunctions.This isa convenient wayto make steps on the loop response. equations. It is explained how to generate block diagrams. In general these curves imply transcendental functions of s. A design sequence is recommended for digital controllers.
l(c). The average slope of the gain response will therefore be reduced by (5"/180°) X 12 dBloct= 1/3 dBloct.5 dB.. and is a straight line with the slope 20n dBldec. the average phase lag must be order for themaximum phase lag not to exceed (1 reduced by the value of the phase ripple as seen in Fig. 5. relations between the ripple amplitudes (b).021ndBloct = 6ndB/oct as shown in Fig. log (a) 0 1 2 . the loss in the feedback will be 1. sc.Chapter 5. 5. The higher the order of the compensator. dB (b) ' 1 1 Fig. will force an increase in the average phase margin by 5".1 Gain and phase lag diagrams (a).5 dB and the comvensator need not beprecise. 5.2 Asymptotic Bode diagram The gain for.2. and the average slope of the gain response must be reduced correspondingly.e.75 dB as seen in Fig. and the peaktopeak ripples in the gain response are1.2 ConstantslopeBode diagrams .i. the required accuracv in the loop Rain response is tvvically not better than ~0.5. 6 d B h t Fig. 5. or 6. the smaller the ripples and the better the accuracy of the loop response.Theuse of octavesispreferred since decades do not provide the necessary resolution. a rationaltransferfunction of the Laplace variable s can be expressed as dB 6 dB/oct 0 t log. For the typical 4 octaves length of the cutoff. Thus. A 5" phase ripple amplitude.Compensator Design 131 .n9O". Generally. The corresponding gain ripple amplitude is 0. 3 4 gain ripple amplltude.1(b). transfer function s"' is the Bode plot expressed as 20n logo. The corresponding phase shift is constant at. and the ripples on the Nyquist diagram (c) Example 1. 10" peaktopeak phase ripples. 5.y)n. I (c) scale f. the designer can achieve feedback within 2 dB to 5 dB of the theoretically available using a compensator of order 8 to 15.2 dB which is marginally acceptable. Typically.
OS)( of the asymptotic Bode diagram is calculated at can be drawnas follows. sc.3(c). 1 oct/cm.3(a).& 6 dB/oct.and lowfrequency asymptotes. The gain plots (Bode diagrams) for transfer functions with respectively the single I(ia + 21~f~)/Zq. fp r. For a function having several real poles and zeros. sc.2)( ja + 0. the gain is 12dBas shown in Fig. single pole (b). A convenient scale for drawing the diagram is 10dB/cm.3 Bode diagrams for a single zero(a). First. b y its larger component. and the error of the asymptotic gain approximation (c) The responses can be approximated . 3 2 1 114 1/2 1 2 4 log. shown by the thicker lines.&l and with the single real negative pole real negative zero 2Olog 2010g 127cfp/( jo 2ltfp)l are shown in Fig. at the frequency o = '1. and less than 0. . the expression ( j a+ 2nfi) is replaced. For the calculation of the ammvtotic gain. log. Example 1.132 Chapter 5.(@)l =: 20 log k + Z: 20 logIjo . at each pole freauency..5. These are two straight' lines which have slopes OdB/oct and *6dB/oct. (c) Fig. 5. and'bends downward. the value some frequency. sc. and which intersect at the corner frequency equal to the frequency of the zero& or of the pole&.4. whether real 0. 5.1 dB at two octaves. The error is 3 dB at the corner. Compensator Design where k is a real coefficient and the zeros Szi and the poles Spj can be real or make complex conjugate pairs.3)( ja + 10) (j o +. The diagrams are drawn against the f or o frequency axis. The frequency response is calculated after replacing s by j o .by their high.r i m g w . for instance. The gain frequency response is 20 logU. The errorof such approximation is shown in Fig.(b) by thin lines. piecelinear My'f'ofic Bode diagram bends upword at each zero corner fieauency by ddB/oct. 1dB one octave away from the corner. i a .The asymptotic Bode diagram for the function L( ja)= j o + 2)( jo + 20) (jo+ 0. + 1 L t log.s ~I Z: 20 log Ijm .Spjl. It is "4 L( j a ) = j a x 2 ~ 2 0= j w x j a x 1 0 ja 4 "j ' Le. 5.
by 6 dB/oct. and0 c c 1. rd = [0. the polynomial coefficients of the numerator and denominator of thetransferfunctioncan ‘be foundfromtheknownrootsofthe numerator and denominator with rn = [0. 5.e. The Bode zeros are real and alternate along the frequency as axis plot of a rational approximation to S q is shown in Fig. 30 20 10 0 Fig.rn is an integer. num = poly(rn).3 103.. and for the purpose of finding a rational function which gain approximates a gain response defined by a plot. The polezero spacingsa and b result in an average slope of . which should equal . For the above example. denum = poly(rd) . 4 . denum) or by using the commandfreqs.i. it is6dBloct. The actual Bodediagramshownbythethinlineisobtainedbyaddingtheasymptotic error responses.Compensator Design 133 The slope of the asymptotic diagram at this frequency corresponds to the resulting by powerof ja.3 Approximation of constantslope gain response As mentioned in Chapter4. Asymptotic diagrams are widely used for conceptual design.2 0.Chapter 5.s .4 Asymptotic and actual Bode diagrams of a rational transfer function AS seen: the asymptotic diagram is fairly close to the actual Bode diagram. 5.Then. The shape of the actual Bode diagrams is commonly verified by computer simulation.= s. 5. constantslope segments are important components in shaping the loop frequency response.. and down at the poles.5(a). 5.5 2 201 .5(b).6b/(a + b)dB/oct.q dB/oct.~ where p and q are real. The irrational function sq can be approximated by a rational function whose poles and illustrated in Fig. and the Bode plot can be obtained with bode (num.theasymptoticdiagramcanbedrawn “bending” the response up at the zeros. Any constantslope function canbe decomposed into a product of a rational and an irrational functionsof s : s.
3037~~ + 0.134 Chapter 5. asimplerational an octave apart: T(s)= (s + 1 / 16)(s+ 1 / 4)(s+ 1) (~+1/32)(s+1/8)(s+l/2) 20 The gain slope is nearly constant at 3dBloct over a wide frequency range.05" of the ideal30". as the gain transitions to the flat asymptotes.~ 1o2 10" 1oo Frequency (radhec) 10' Fig. (5.462s' + 1.6 Gain and phaseresponsesof a function with alternating real poles and zeros 0.000002178.Therefore. 5. Sometimes.2342 + 1.45. Fig. while the phase remains within 0. the to 'the polynomial programs produce transfer functions which are too sensitive coefficientvariations. however.OoO10989s+0. Compensator Design dB *plane 0 (a) (4 Fig.00920'1~~ + O.7.6 showsthegainandphaseresponsesfor approximation to using three zeropole pairs spaced evenly.6.5 dB or so.02954~~ + 0.2007~~ + 0. . 5. in accordance with the Bode phasegain relation. approximation of the desired curve by the placement of the poles and zeros as described in Sections 5. At the corner frequencies.4415~~ + 2.007 dBloct by less than0. 5. 10' Example 2 .OOOooO1979 Overthefrequencyrange 0. 10'' 10" 1oo Frequency (radhec) .861~~ + 0. always resultsin a robust design. In most cases. and then implementation of the compensator by cascading loworder links as described in Section 5. compensators in control systems is typically only Curvefitting computer programs can be used for approximation of desired compensator responses._ Io .2) s6 + 2. the phase lag is nearly 45'12.5 Alternating polezero approximationto sq Example 1.0005682s+ 0. This high accuracy is rarely necessary since the required implementation accuracy for 0.On the other hand.4276~~ + 0.01 to 10Hz.thetransferfunctionsmustbecheckedover by changing the coefficients by increments which reflect the software roundoff errors and the tolerancesof the hardware implementation.g ( 1 10 0 . The38" phase at the center differs from the 45" phase of the halfintegrator due to the effects of the gain asymptotes. thegainresponsedeviatesfromtheideal 2. % . this works out well.05 dB. A ratherextreme following the example is s"" which approximation to was generated by a curvefitting program: C(s) = .
4) and (d) lag (s+l. When linear modelsof the actuator and plant are available. convergence will require For the purpose of iterative structural design. as discussedin Chapter 4. until the desired loop frequency response is achieved. 7 0 " 1oo Frequency (radlsec) 1of TO" 1oo Frequency(raasec) 1of 0 ' 1 3 CI) Q) 8 10 \ ? i 1oo Frequency (racVsec) 1o1 \ 20 101 4 1oo Frequency (racVsec) 10 ' (dl Fig. sc. and fp are the frequencies of the zero and the pole. dB 0 ~ f.. sc. no more than 5 10 iterations. ' ~ ~ . frequency responses for (c) lead (s+0.Chapter 5. log. Computer generation of frequency responses m&es the iterative design of the compensator quick and effective. respectively.7)/(s+l.Fig.7 Lead (a) and lag (b) Bode asymptotic diagrams and (thin lines) Bode diagrams. This transfer function is called lead when the zero precedes the pole. 5.e.3) wheref. f f f.4 Lead and lag links Theidealcompensatorresponseshouldbedetermined by subtractingthe known actuator/plantfrequencyresponsefromthedesiredloopresponse.7) .4)/(s+ 0. i. the zero is at a lower V. 5. Compensator Design 135 5. the iteration can be carried out using a trial compensator in series with the With some actuator and plant. experience. The simplest of these has the transfer function with a polezero pair s + 2nfz s + 2nfp (5. it is best to regard the compensator as being composed of elementary building blocks.7 shows asymptotic Bode diagrams and Bode diagrams for lead and lag Gansfer functions. log.: cfp). The transfer function is called lag when the pole frequency than the pole comes first (fp cfz).
10. 0 O I Fig.. In a system. the larger is the gain change from the lower to the higher frequencies.9. 5.excessive phase stability margin with the area a decadex degr. one can use Bode's phase integral (3.I dB 4/ A . 5. several lag links are sometimes needed for better approximation accuracy. Adding a leadlinkmakestheloopgainBodediagramlocallyshallowerthus reducing the gain at low frequenciesand locally reducing the phase lag.56a dB of feedback at lower frequencies. The current gain response is somewhat shallower. and the larger is the phase lead (or phase lag). The use of two lead links provides a closer approximation to the desired Bode diagram than could be achieved with just one lead link with a larger polezero separation. . Compensator Design The larger the polezero separation.136 Chapter 5 . Egample 1. dB lead Twith a dB I Twith two leads 0 0 lead lead 1 lead2 (b) (4 Fig.10).10 Use of the phase integral Before. 5. 5. thus increasing the loop gain at lower Erequencies as shown in Fig. 5. As in the case of lead compensation. Fig. dB \T desired gain.1 1) to estimate the available improvementin the feedback. The phase integral indicates that elimination of this excesswill yield an additional 0. and there isan . 5.8 shows the use of a compensator to change the slope 12 dB/oct of a double integrator to the desired slope of 10 dB/oct with a single lead and with two leads. (3. the desired loop gain response is a straight line with slope 10 dB/oct which corresponds to constant 150" phase lag as shown in Fig.9 Lag compensation Fig.8 Compensation of a double integrator with (a) one and (b)two lead links Lag compensation steepens locally the Bode diagram. dB = 0. deciding to introduce further compensation to improve the approximation to the desired response. sc. The trade of the excessive phase margin for larger lowfrequency loop gain can be made by introducing lag links. log.56a dec*degr f.
The magnitude of the function (5. gain.s+w. denominator of (5. Example 3. A simple lowpass filter can be obtained (5. complex poles and zeros are required.5. Compensator Design 137 5. The frequency of the complex pole must be higher than the frequency of the real pole. To implement these sharp angles on the Bode diagram. 5.4): . and to shape the loop gain over the functional frequency band.4) at the resonance where first andthelasttermscanceleachotheristhe 9uaIlfyfacfor Q = 1/(2()'where ( is the darnping coefficlenf.14. 5. dB 10 0 10 20 0. This methodwill be used to formthe Bode stepin Fig. The difference between the two logarithmic responses in Fig.25 0.and only the first term at o > coo. A thirdorder lowpass filter can be obtained by cascading a singlepole link andza link with transfer function (5..Chapter 5. 2 (4 (5.4) with ( = 0. Complex poles can be also required to compensate for the plant response.5 Complex poles Upper and lower Bode steps need to be reasonably sharp.1 1 is a notch o r a peak response. Its transfer function is the of ratio functions (5.4) at o < q Example 1. 4 Fig.5 1 2 relative frequency ~/b+. s + o ~ s =jmo andthe 4  s2 +2@o. by using transfer function Example 2. The normalized gain and phase lag frequency responses for the complex pole pair function s2 + Q " w .11 Gain and phase lag responses for a complex pole pair The asymptotic Bode diagram is calculated by retaining only the last term in the .4).4) are presented in Fig. and6 should be chosen such that the peak of the complex poles compensates the rolloff of the real pole.1 1.5. 5.
It is seen that the signals at lower frequenciesare attenuated in the first link by 54 dB. as shown in Fig. and W2 are shown in Fig. Consider the implementation of the transfer function s + 2 s+500 W(S) = w. and to c n /bat the frequency of the resonance where s =jq. 5. Assume also that there is a5 pV disturbance source at the junction of the links.13(a). dB 34 0 2 10 54 (a) (b) Fig.2. Such notches have been used in the prefilter described in Section 4. the signal drops dangerously close to the noise level.s+0: s2 +2c. 5. attention should be paid to the signal level at the link junctions so as not to impair the compensator's dyn8/llk range. Example 1. When > &. This istheamplituderange of thesignalsthelargest ofwhich isnotyet distorted by saturation in the nonlinear links. This way of making the compensator is certainly not the best since after the attenuation. . a peak response follows.. At the junction between the links. 5.When c n c c d . 5.3. a notch response results. is only2 pV so that the signal will be heavily corrupted with the noise. The widthofthenotchorthepeak depends onthechosen damping. (S)W2(s) = s+lOOo S + l O as a cascade connection of the two links. and after the amplification the noise floor will be raised. Compensator Design This function equals to 1 at zero and infinite frequencies. Such disturbances may be caused by noise or interference in the signal amplitude analog systems.13(a). and by roundoff errors in digital systems.12 Gain responses of two different implementations of the same transfer function Assume that a 1 m'v signal with various frequencies is applied to the as input shown 1Hz and at 1kHz in Fig. and then amplified in the second link by 34 dB.oos+#: Chapter 5.138 s2 +2cn0. the signal levels at differ. and the smallest is still substantially larger than the disturbance and noise meansquare amplitudes.12(a). The asymptotic gain responses forVV.6 Cascaded links When the elementary links of the compensatorare cascaded. cn 5.At 1:Hz.
Chapter 5.6 dB/oct. 5. Compensator Design Noise 5pV 139 1Hz at 2pV mV 1 . The average slope is lOdB/oct at frequencies 0. 5.13 It is better to implement the same transfer function links: W ( S ) = W3(s)W4(s) = by cascading the following s + 1 0 s+lOOO The frequency responses forW3and W4 are shown in Fig. i.5) at the end of the Bodestep and effect the desired asymptotic slope.. 6 3 ~= 1. The loop transfer function must behave as a single integrator at zero frequency. to keep in the same link the voles and the zeros which are close to each other..5. it makes the slope . When this rule is followed.12(b). signal amplitude (0. The desired width of the Bode step (4. 5. At lower frequencies. and 18 dB/oct. l l l l .buttheloopgain at frequencylOrad/secandhighermustnotexceed 35 dB. A pairof complex zeros at ocmake the corner at the beginning of the Bode step.13 Signal levels at link junctions for the responses in Fig. the rolloff at higher frequencies mustbe 18 dB/oct or steeper. The plant is a single integrator Us.lmV at 1Hz w4 ’ 1mVatlkHz Fig. The asymptoticBodediagram showninFig. 18 dBloct. i.6. Let us design the compensator as a cascade connection of several links. the link affects the slope of the total Bode . The gain and phase stability margins must be not less than lOdB and 30’. 5. When q.e. A real pole at o = 2. Example 2. The crossover frequency must be not less than 0. . then a d = 2. (a) w1 1mVat 1kH ’ .lmV 1Hz at 1mVat lkHz w2 Noise 5pV (b) ’ s + 2 s+500 ” . and the loop response must include a Bode step.15 to 2 radlsec.9 rad/sec.8 and a pair of complex poles at a frequency somewhat smaller than ocform a thirdorder lowpass filter (as explained in Example 2 in Section 5. The signal levels have a much smaller dynamic range as indicated in Fig. = 1 and x = lOdB. For this.14 iscomposed of pieceswith slopes 6 dBloct.2 mV at 1Hz)remains much larger than the The general rule is to avoid creating links with excessive attenuation or gain at Wfreauency.13(b).8. 12 dB/oct.2) is 0 . and even the smallest 5 pV noise.e.which alsosimplifiesiterative diagramoverarelativelysmallfrequencyrange adjustments of the frequency responses. 0dB/oct. oc= 3.
Compensator Design c t Fig. the corner point as seen in Fig. .8160 n = 11.0000 6. The resulting loop transfer function is T(s)=ks 1 + 0. 5. w).2000 23. 4~2.140 Chapter 5.2080 18.4 is chosen for the complex poles to compensate for the rounding effect of the real pole at o = 2.66s’ + 2 3 .or secondorder links.12s ~ The loop responses are shown in Fig. d = 1.8 ~ s6 + 6.[1 2. +218.4 s2 + 2 s + 4 s + 2.1. [mag. Damping coefficient of 0.6600 23.3960 48.8~9~1) k(1~1~1~2.8 * conv([l 0.[1 1.lS2+ ~ 4 . 6 + ~3 ~8 .15. After multiplicationof the polynomials in the numemtor and denominator: n = 4 * 2.8 s2 + 2.4 9 0 1 ) ) d = conv(conv([l 0. phase] = bode(n. the order of& w e n s a t o rt r a n s f d n c t i o n must be reasonablyh in highaerformance controllers.42 2.8~4~1)/( the coefficient k = 3.5880 38.1040 conv([l 2.14 Asymptotic Bode diagram Damping coefficient of 0. 1 + ~ 2.1 1.8. 5. 5. By using this valueof k initially. 4 + ~4 ~8 .15(a).8 s + 1. The Nyquist diagram on the Lplane plotted by W = logspace(l..1168 0 the loop transfer function becomes T(s ) = 1 + 27.5 is chosen for the complex zeros in order for the gain response to pass through. it was found thatfor the Bode diagram to pass close to the 0 dB level ato = 1. This example shows that even when the order of the plant transfer function islow.4s+ 9 s 1  The compensator is composed of four first.1125 2. d. From the condition that ato = = 1 the asymptotic loop gain coefficient 1~1 = .42].[1 2 41) 54. 1).06]. k must be increased to 4.06 stQ.4]).81.
5. and the polynomial coefficients. an n.3. 5. Lplane Nyquist diagram 40 30 20 10" 1oo 10 10' Frequency (radhec) 0 1 0 20 30 101 1oo 1o1 40 240210180150120 Frequency (radkec) (b) Fig.'wo') title('Lplane Nyquist diagram') set(gca. For example. by multiplying the function by a constant. 5. correspondingly. the highfrequency asymptotic slope can made be steeper by adding complex poles or a notch. The response can be shifted along the iirequency axis by replacing s by as (as described in Section 4.'r'. .7 Parallel connection of links The compensator may be implemented also by connecting several links in parallel as shown in Fig. 5.15 Loop frequency response (a) and the Nyquist diagram (b) It is recommended for students toplay with this response. Compensator Design plot(phase.p.16. to modify it by changing the poles and zeros (or the coefficients of the polynomials) in order to get the feeling for the sensitivity of the response to the poles. the gain response at lower frequencies can be made steeper. 0. The compensator transfer function is equal to the sum of the transfer functions of the elementary links Wl + W2 + W3. lag can be added to the plant and the Bode step made. the zeros.Chapter 5. which will also require lengthening the step. by increasing the complex pole damping coefficient the Nyquist diagram can be made more rounded.2.'XTick'. and along the gain axis. The zeros of thecompensatorresultfromthe interactions between the elementary links: the output 0 is at that valueof s at which the sum of all the links' outputs is 0.20*loglO(mag). Example 2). 180.15(b).[270 240 210 180 150 120 903) grid 141 is shown in Fig.The poles of the compensator are just thepoles of theelementarylinks. wider.
5. the plant is an integrator lls. Compensator Design I 0 Fig.142 dB Chapter 5.16 Parallel connection Fig. if required.) At the fiequency at which the link responses cross. 5.f23.17 (a) Parallel links’ gain responses (b) output signals’ addition at fi2 It is convenient to design the compensator such that each one of the parallel links dominates the response over a certain frequency band.3. Depending on the phase difference between the output signals of frequencyadjacentchannels. (Le. The crossover frequency mustbe 1 radlsec. etc.the summedsignalamplitudemaybelargerorsmaller than the amplitude of the components. at 3 2 . This way the links canibe designed and adjusted one at a time. In the system shown in Fig. 5.) the output can be found by vector addition of the links’ transfer functions. 10” 10” 10’ Frequency(radsec) 10” 1oo Frequency(radsec) 10’ (b) ” Fig.17(a).. In the filter. 5. Parallel links provide a convenient way of implementing Bode steps. the resonance is at four .18(a). as shown in Fig. as also for an option described in Chapters 1 1 and 13. (This configuration provides of placing separate nonlinear links in the parallel paths..4s + 16) with 6 = 0. 5.18 (a) Bode step implementation and ‘(b) the Bode diagram The compensator is the parallel connection of the link Cl= 4/(s2+ 4s) and the lowpass filter Cz = s/(? + 2. Example 1.
The system block diagram is shown in Fig. In this example.The current to node13 represents the force F. I force F . Example 1. CPI and CP2. In spite of a somewhat longer input file and the necessity to draw an equivalent schematic diagram.8 Simulation of a PID controller The PID confroIIerconsists of three parallel branches:Ils. We will consider an example with a double integrator plant with a flexible mode.6s + 64 s5 + 6 . The plant includes two masses.5. but the phase lag at lower frequencies is too large. the transfer function for the plant should be first found. response andthelowfrequencygainis10. the analogy is explained in detail in Section 7.SIMULJNK.18(b). D define the transfer function of the controller. but is simple and as such.and Ds.The coefficients I. P. or in the common path. 5. velocity to voltage. is shown in Fig.1.6.When MATLAB is used.19 Block diagram for a control loop a PID compensator and a double integrator plant having a flexible mode The simulationcan beperformed inMATLAB. orSPICE. 4 ~ +2 ~5 . for largelevel signals (this issue will be considered in Chapters 10 . P . mass to capacitor. 5. SIMULINK and SPICE allowthetransferfunctionderivationto beskipped(using SLMULINK block diagramsfor ladder network analysis is described in Section 7. A saturation link is commonly placed in front of the Ichannel to improve the controller performance in the nonlinear mode of operation. spring to an inductor with the inductance equal to the inverse of the spring stiffness coefficient. This can be remedied with lead links placed in C.Compensator Design 143 timesthecrossoverfrequency. is not optimal in most applications. velocity (voltage at node . quite popular. The schematic diagram for the simulation is shown in Fig. ”  “ ”  Mechanical plant Position sensor position x Fig. The position x (voltage at node 2) is calculated from the plant 13) by integration. and a dashpot (a device providing viscous friction).a spring. It is seen that the Bode step is well implemented. we will use SPICE. using SPICE has certain advantages: there isno need to generate a mathematical description of the plant if we use the common electromechanical analogies (force to current. 6 ~ +~ 64s2 The loop 2 ($1T(s)= Cl(S) + c S c . 9s2 + 29.20.and the dashpot to a resistor with the resistance equal to the inverse of the damping coefficient.1).1dB. This controller does not implement a Bode step.Chapter 5.13).1).19. 5. 5.
and 13 to ground. 8. input signal summer RSRl 1 0 1MEG . The integrators are imitated by ideal controlled current sources loaded into capacitors. . 5. Saturation links are implemented using oppositebiased diodes shunting a resistive load.20 SPICE model for the system shown in Fig. Ipath integral coefficient leakage resistor proportional coefficient differential coefficient GP *** *** 9 0 0 3 2 4 0 0 3 3 4 0 1 9 0 0 4 1 9 0 1 GD1 LD GD2 RS *** . The SPICE input file is shown below.20 *** ES 3 0 1 2 1 . . 5. as required by SPICE. 5. .MODEL DIODE D VT1 6 0 1V vT2 0 7 1v . summing resistor .001 RSAT 5 0 1K Dl 5 6 DIODE D2 7 5 DIODE . saturation in Ipath threshold = (0.19 The output currents of the three parallel paths in the compensator pass through the 1 S2 summing resistor RSUM. Compensator Design command WEST 1 I summer vrq I f vr4 I I plant I " Mechanical L"I " "  Actuator with Saturation +I model Fig. This diagram represents a simple way of simulating the performance of the block of an analog diagram in Fig. leakage resistor RSP2 2 0 1MEG .7+vTl) *GIl/GSAT *** GI1 8 0 0 5 1 C12 9 0 0 8 10 RSP8 8 0 1MEG *** . andit does not describe the implementation compensator (compensator implementation will be considered in the next chapter). 5. leakage resistor *** GSAT 5 0 0 3 0. Included in the file but not shown on the picture are the highresistance leakage resistors connecting nodes 1.19.19 in SPICE. 5. .144 Chapter 5. 2. and the voltage on the resistor represents the sum of the outputs of the parallel paths. *** PID example Figs. .
this system can be simulated using a student version of SPICE which presently allows up to 25 nodes and available free of charge fromIntusoft. . The loop gainresponseshouldbesimilarto Fig. The plots should be made for: (a) frequency response of loop gain and phase so that stability margins can be checked. END . T W 0.orPSPICEOfromMicrosim. Compensator Design 145 coefficient actuator GA1 10 0 9 0 1 *** actuator gain in RSATA 10 D3 D4 vT3 vT4 GA2 *** 100 1K 11 DIODE 12 10 DIODE 11 0 1v 0 12 1v 13 0 0 10 1 saturation force source CP1 13 0 5 RSP13 13 0 1NEG LP2 33 14 0. linear and make vp it ( 2.01 10 .21 Loop gain response that shownin Fig.inmostversions of SPICE.TRAN.vP(2). (c) transient closed loop response for different coefficients P. L) and saturation thresholds.) To plot the closedloop response.AC and enable To plot the Nyquist diagram with logarithmic scales. To plot transient response.) Runningtheprogramisrecommended as 'a dB student exercise. the transfer functions can also be specified by their poles and zeros. 5. to reduce the loop gain at the frequencies of the plant structural .The loop response for PID controllers can be augmented by additional lowpass filters and notches.AC DEC 20 0. 0 (b) closedloop frequency response.1 CP2 12 0 0.21. change the abscissa scale to .1 10 . . other graphical or postprocessor .I. V2 is position position *** VTEST 1 0 AC 1 use only when frequency responses are tested . plotvdb ( 2) . Since the number of the nodes is only 15.Chapter 5. disable with an asterisk lines VTEST and . responses of the linear links ( a sthe difference between the input and output signal levels and phases). To plottheloopresponse.02 GINT 2 0 0 13 1 CINT 2 0 1 *** V10 is force.thesummers and nonlinearitiescan also bespecified by algebraicexpressions. 5. when transient responses tested * . 0 andthenplot vdb(2).connectthe ESUM secondinputto '.(InSPICE. VPULSE and . use only when frequency responses tested ** Pulse (Vmin Vmax delay rise fall width period) * VPULSE 10 PULSE ( OV 1OV OS OS OS 500 500 ) . vp ( 2) . V13 is mass of the main body leakage resistor spring of flexible mode mass of second body losses in the flexible mode integrator to generate velocity.5 RP 14 15 0. when transient.responses tested PROBE . .
the . In digital systems. This dynamic range is equivalent 23 to bits. For a similar reason toreducetheroundingerrors.itis commonin digitalsignalprocessing(DSP) . for the purpose of reducing the required dynamic range. The resulting response has the highfrequency asymptotic slope 18 dB/oct. and the lowerfrequencies loop gain is lower than that of the responsewith the crisp Bode step by about 5 dB.10. The P coefficient is chosen such that with only this path on. bandwidth. summer. different frequency regions. When considering accuracy. either type of controller. the loop gain at the crossover frequency is “x dB..andthechoicebetweenthetwois determined mostlyby the priceof development and fabrication. 5. andofDIAand Another important issue is the compensator’s flexibilityin being adjusted for different plants For the great majority of applications. and feedback path link directly affect the output accuracy. it should be remembered that only the accuracies of the prefilter.146 Chapter 5. analog or digital. andchopperstabilizedamplifiersinside analogreferencevoltagesource.up to 1 MHz) and computational delay limit the bandwidthandthe available feedback. the prices .of a microcontroller An> converters. lower. it is advantageoustobreakthecompensatorintoseveralcascadedlinks. The integral term canbe chosen such that with only this path on. Because of this. i. The coefficient D is chosen such that with only this path of the compensator connected. The accuracy required of the compensator and alsoof command feedforward links is much be implemented with less accurate elements. three important considerationsare accuracy. feedback loops withfb > 100 kHz should be analog. Example 2.6. benchtype digital voltmeters are accurate to 7 or 8 digits. .andeaseoftroubleshooting.is. can be designed such that it willperformwell. so generally these can Analog circuitry can be made very accurate and stable in time. 4 times largerthan the crossover frequency.eachlink related to different timeconstants. For example. each such function having poles reasonably close to zeros. An additional principal drawbackof digital controllersis that the sample rate (typically.e.12 to 23 bits. ‘ a s 5 9 Analog and digital controllers While many factors may affect the choice between an analog and a digital system. Other considerations are: the type of sensors used. A triplepole lowpass filter (two complex poles and one real)with the cutoff frequency. the loop gain at lower frequencies is xdB. Compensator Design resonance and provide gain stabilization over the frequency of range the resonance.10 Digital compensator design 5.resistors. powerconsumption. approximately.1 Discrete trapezoidal integrator As is explained in Section 5.to implement a highorder transfer function as a cascade connection of firstorder and secondorder functions. and price.. is placed ’ in front of a single integrator plant. The dynamic range of a common opampis 1pV to 10V. typic*ally. MATLAB simulation is left Drawing the asymptotic Bode diagrams and making a recommended student exercise. the crossover frequency is where it must be. the accuracy and the dynamic range are frequently limited by the dynamic range of the employed An> converter which .
5.. =ITS.22 (verification of the correspondence of the equation to the block diagram is left as a Fig. nT.24 Digital integrationflowchart I (n2)TS (nl)T. let multiplication by z signify the time increase by one sample period so that ZVn1 ZUn1 = Vn 9 = U.Chapter 5. of a biquad transfer function A similar block diagram can be used for digitalimplementation of thebiquad operator. "" 1 s 8 (*3)T.. nT. The increment in v(t) for the time interval from the sampling# n 1 to the sampling # n is (n3)TS (n2)TS (nl)T.23 Trapezoidal digital integration Next. 5. 5.Compensator Design 147 The secondorder transfer function implemented using or biqffad.(5.onlytheintegration mustbeimplementedindiscretestepsperformedat sampling instants. 5. in 5.22 Feedbackimplementation student exercise). can be analogintegrators 11s with thefeedback block diagram shown Fig.. With this symbolism.8) can be rewritten as . v"l / time i trapezoidal Fig. .23. Discrete trapezoidal integration isshown in Fig. 37's . The output v(t) represents the shadowed area which approximates the integral of the input. The input analog signal u(t) is sampled at time intervals Ts. Fig.
12) s.25(a).22 but with different coefficients and with l/z replacing Us. 0 v.25(b). The origin of the splane maps onto the point (1. Since it needs at least 2 samples per period to identify a sinusoid.An equivalent but simpler flowchart can be obtained by substituting (5. while analog integrator output reaches the average value in the middle of the sampling interval. While a digital feedback system is being designed.10) into (5.11) maps the strip of width 27th of the lefthand plane of the dashdotted lines in Fig.digitalintegration delays the signal by Td2. 5. = Z"U. digital signal processing is only feasible for sinusoidal signals with fiequencies less than the Nyquist frequency fd2. (5. f ixfd2 of the splanemap onto pointsfj in the zplane. In otherwords. 5. As seen in Fig. 5. the input signal value averaged over the interval of sampling is calculated by discrete integration only at the end of the sampling interval. s =f'In z S (5. by storing a sampled value and recalling it after the sampling time The biquad (5. With such a high sampling frequency.7) can be implemented digitally by replacing integrators Us in the flowchart in Fig.10) This formula is presentedas a flowchart in Fig.7) to replace l/s.10.4. the delay Td2 can be treated as nonminimum phase lag. The Nyquist frequencies . this lag should not exceed 1rad at the crossover frequency. Since l/z signifies a delay by Ts. Therefore.comparedwithanalogintegration.2 and 4. 5.23.22 by digital'integrators.O) of the zplane. bounded by Function (5. The resultingflowchartlooks 'like theblockdiagraminFig. wemight duplicate the results of the previous section using the Laplace transform. and the points jnfsand jn& onto the point (1.2 Laplace and Tustin transforms To get an additional insight into the problem. 5.24. Compensator Design 1 T S 2 and the transfer function of the discrete trapezoidal integrator is (5. trapezoidal integration is quite accurate.2.O) of the zplane. thesampling frequency fs = l/Ts should be at least 2x/1= 6 times larger thanfb.148 VnZ Chapter 5. three important comments need to be made. we can view l/z as the Laplace transform of the delay which is z = expfs From here. and simplifjing the resulting expression to obtain a function of z as a ratio of two secondorder polynomials. onto the unit radius disk of the zplane shown in Fig. i u.1 1) . 5. As mentioned in Sections 4. Next. Operator 2l can be implemented Ts. 5.3.
s .13) (5.25(a)..e. theaccuracy of theTustintransform. fiom 2 . 1+z" q s 1z" 2 This expression is the same as (5. Within this part of the quadrant.Chapter 5.l)/(z+ 1) * Near the origin. 5. the expression for the integrator l/s is 1" . exp(2x) = (1 + x)/( 1 .10. at all frequencies below& the design uses only a small part of the first quadrant of the mapping in Fig.stillremains adequate since the required accuracy of the loop gain response implementation over this frequency range also decreases. an exponent can be approximated wellby a bilinear function. when x is small.14) From (5. 5. this approximation is known as the Tusfin transform.6. In applications to DSP..25 Mapping of (a) the zplane onto the splane with(b) function s = fs In z and with (c) Tustin transform s = 2fs(z..althoughdecreased.lfS forthereasonscitedintheendsof Sections 5. Inpracticalcontrolsystems & c 0.10.x). The Tustin transform shown in Fig. 5.'Therefore.25(a).25(c) maps the entire left halfplane of onto the unit radius disk in the zplane as shown in Fig. s=2fsz+l Le.10) signifying that the Tustin transform uses trapezoidal integration.13).1 and 5. 21 (5. At frequencies of the Bode step. 5. i. the Tustin transform is quite accurate.to 4 . Compensator Design 149 Fig..
Numerically. TheTbtin transform is used most fiequently. and bo = 0. insignificantand the method will not be discussed here.0. the following evident propertiesof the ztransform canbe'used: . is equal to the Tustin transform of z for z = 1.fs) nd = dd =: 0. the functionof s is 1. s must be expressed in kradhec. The Tustin transform can be matched to the Laplace transform at a specified frequency by prewarping.9286. fs = 100. there is not much difference between the Laplace and Tustin transforms.150 Chapter 5. rn = 210.0 .4 gives the algebraic expressions for the coefficients of the function of z. S .d. multiplying a function of s by a constant is equivalent to multiplying the function z by the same copstant.14). and either one can be used.9048  % sampling frequency in Hz Example 3. in application to compensator design.14) that z remains the same whens and fs are similarlv scaled up or down. 0.95 122 . Example 1. The function of z is 0. a transfer function of s for s = 00.9286 I z 1. The Tustin transform properties for the numeric values of the previous example are verified in the following. For the purpose of verifying the equations. and the pole is 10. the Tustin transform is also For s = . a. With Tustin transform (5. Compensator Design In fact.The same problem is solved with MATLAB by n = [l 53. respectively.9762. d = [l 101. and then appropriately scaling the gain coefficient.for z = 1. the ztransform can be also foundby mapping the poles and zeros of the function of s onto the zplane. = 0. however. the functionof s is 0. [nd. It is seen from (5.9048.0000 0.9'762 0. === +S 2005 200+5 2fs 0. 2fs 2 .1 in Section 5.9286 1. the Tustin transform is also 1 The zero of the function of s is 5. * * a transfer function of s for s = 0.dd] = bilinear(n. Table 5. is equal to the transfer function of z for z = 1.The Tustin transform of the lead s+5 s+10 with sampling frequency 100 Hz can be found by substituting s' with expression (5. for example.5. for z = 1.9048 I z Example 2 .whenfs is expressed in kHz.5. the zero and the pole of the function of z are found to be. The advantages of 'using prewarping are. For s = 0. al = 0.13).9762 0.10.
These block diagrams have forward and feedback paths. Compensator Design 15 1 . a goodoptionistouseinsteadthecommand feedforward. MATLAB. In this case. However. The Tustin transform is used to find functions of z which correspond to the secondorder functionsof s . fs 2 50fb.0. 5. a rational transfer function of the analog compensator is’ determinedandbrokenappropriatelyintosecondorderrational functions (as described in Section 5. The errors caused by the calculation errors and by the fact that the Tustin transform is only an approximation to the Laplace transform. 5 . the calculated values of the pole and the zero differ somewhat from the exact values.904762. The stability marginsare verified by increasing the loop gain and loop phase lag up to the moments when selfoscillation starts. 3 Designsequence Digital control systems can be designed as follows: Given the required feedback bandwidth fb. do notmattermuchforthedigitalfunctionsintheforwardpathofthe feedback loop. n Fig.and the system is tested or simulated on a computer (including the nonlinearities). commonly.Chapter 5. fs 2 lo&. 1 0 . The functions of z are coded in. 5. the error might not be acceptable for the links in the feedback path or in the prefilter. SIMULINK. and computer code The functions of z can be specified by block diagrams.5). 5.4 Block diagrams. The block diagrams for the firstorderand the secondorder functions are shown in Fig. zp =2fs + $ ~ E _. or by computer code.. For simulation and tuning in analog form. or SPICE can be used. equations. The sampling delay Ts/2 is imitated by introducing in the loop an allpass with polezero pair(s fs)/(s +fs) or by placing a pole at frequencyfd3.26. By approximation of the optimal response.20010 . (b) the secondorder link . 2fss 200+10 Due to the finite accuracyof the calculations. or directlyin C or in another implementation language.26 Block diagrams for (a) the firstorder link. SIMULINK).10.by equations. the sampling frequency fs is chosen. The system is simulated using a digital control software package (for example. and for better performance.
the coefficients are given in Table 5. dendl = bilinear(num.1 Coefficients for bilinear function of z a1 (pl2h + po)lm For the biquadof s P2S2 a0 (pl2fs +po)lm b0 (qo.2. the transfer function for the block diagram is found (a) to be a1 + aoz" 1+ boz" and for the block diagram (b). frequency responses of MATLAB command freqz.z +ao z2 + b. fs) where fs is the sampling frequency. Compensator Design Using Mason's rule. den. Table 5. where m =: 2fi + qo.z + bo (5. bi can be obtained by substituting (5. For the function P l S + Po 9 s + 40 which is bilinear.13) into the expressions for of s the transfer functions of s.1..z z + bo +biz" + boz2 or a2z2 +a.16) The coeflicients ai.2 Coefficients for biquad ofz As alreadymentioned. al. digital filters can be found with the For the firstorderlink in Fig. / * updating r * / . the followingC code equations can be used. 5. To verify the transform. Table 5.2fs)Im + P l S + Po s2 + 41s + 4 0 the coefficients for the biquad of z are given in Table5.z" +aoz 2 1 +ao or a. ~ l BO: . a2 +a. / * using previous valueof r to find** y = A0 * r.26(a). * * first component ofthe output * / r = x .B O * r .152 Chapter 5. with a.borenamed AO.thecoefficients of therationalfunction' of z can be calculated from the rational function of s with given numerator nun and denominator den by the MATLAB function [numd.
The cycle starts with the new sample value of the input. C1= 2. the realizable crossover frequency is lower in proportion to this delay.18) r = rl. A saturation link placed in front of Cl makes the transfer functionof the compensator dependent on the signal level.w * / y = A O * w + A l * r .6Hz.is a singlepole lowpass filter. When the signallevelisbelowthesaturationthreshold. due to limited bandwidth of the analog rate controllers for the motors (already designed). since the computer must handle not tasks. C1+ 1 and C2. the motors have 50 msec delay. For the secondorder link.Chapter 5. 153 (5. rl = x .5 + 500 + 50 c: 600 msec. there is an additional only the motor control loops but also other higher priority 500msec delay causedby four real time interrupt (RTI) delays. and they must be either static or global to keep the values stored beto used in the next cycle.e. 125 msec each. fb < 1.BO * w . the compensator transfer function is reduced to C2. When the signal is high.17) / * adding second component to ** * * the output using updated r * / The variables are recalculated each cycle.106 + ~)/(2. and C2 is alead link.5 Compensator design example A small parabolic telecommunication link antenna tracking the Earth has been placed on the Mars Pathfinder Ltinder.5/(0.0833 +s ) .i. For smallsignal amplitudes. / * updating w * / (5. First the value of the variable r that is stored in the previous cycle is used.thecompensatortransferfunction is (C1+ 1)C2. Were the delay caused only by the sampling. the crossover frequencywould befs/5 E : 1. and includes two cascaded linear links.5/600 0. the compensator function is c=(C.Bl * r. 2 : The design has been done initially in the sdomain. similarly. the following code becan used: / * using previous valuesof r. Also. However.310. Two identical brushless motors with internal analog rate feedbackloopsarticulatetheantenna intwoorthogonaldirections.Themotorsare controlled by two independent identical SISO controllers.6~62i. / * updating r * / w = r. . + 1) c. Since the total delay is not only 62.17 Hz . The variables are commonly initialized to zeros. The sampling frequency is 8 Hz. The controller is nonlinear.Compensator Design y += A1 * r.23+ s ) . C 2 =: (0. / * updating r * / y f = A2 * r.. controlledby some loop. then this variable is updated.5 msec (of sampling) but 62. The operation of such nonlinear dynamic compensators (NDCs) will be further described in Chapters 10 and 13. where C. The cycle repeats fs times per second. / * adding component tothe output * / 5.
0. After several small adjustments to the initial response were made to obtain the desired stability margins.18) correspond to the flowchart shown in Fig.e. 5. to 0. 9 (1 . Thesystemwiththeseanalogcompensators was simulated in SPICE and in. The accuracy of the coefficient 0. i.987 should be rather high since this value is subtracted from 1 at lower frequencies where z approaches 1. dB 30 2? 10 I 30 20 10 0 1. and to reduce or eliminate the overshoot.(b)* .154 Chapter 5..28 for the of case both Cl and C 2 operational. The path Cl is parallel to the path with unity gain..0.19) (5.20) The coefficients in the equations have been rounded to the required accuracy. The second expression canbe rewritten as C 2= 0 .28 Openloop'asymptoticBode diagrams for small error (upper curve) and large error (lower curve) The lead C2 provides phFe advance and partially compensates the following lags: the phase lag of up to7 RTI (for extra robustness). 5. The phase delay of the sampling was imitated byan extra pole placed at frequency fd3.987 = 0.20) should not be further rounded.. 5.013 must be accurate to 6%. 5.27. Le.987)/( 1 O. so that the number 0.15 + 0.0.9.0008. and the delay of 0. 30 tFig. the design proceeded to conversion into digital compensation. Compensator Design The asymptotic gain frequency responses of the compensators are shown in Fig.27 AsymptoticBodediagrams of compensators Fig. and for the case of Cl = 0 (lower curve). becomes 30.0.0. 5.At zero frequency.5 dB). up to 1.751~)  (5.8883 in (5.0 compensate for large time delay of up to 7 RTI. The following digital compensator equations were obtained from the analog controller functions with the Tustin transform: C1= (0.99/~) C 2 = (0. The Bode step is very long because of the necessity"t. The asymptotic loop gain frequency responses are shown in Fig.75/~).29 (a).875 sec delay.17) and (5.. Thus. Eiquations (5. the difference frequency gain coefficient to be better than 1 . MATLAB.15/~)/( 1 .8883/~)/( 1 0. C.0 0 10 20. for the accuracy of the low6% (Le.05 sec of the closed analog rate loop.
.75 #define PAR5 . ' ~.Chapter 5. doublev = 0.0.0. (5. 5. integrator .9 #define PAR4 . r = satout + PAR2 *.9 0.19).8883'~ 0. and model a of the plant (of the motor with its analog control electronics).0. 5. a delay block.30 Motor controller flowchfM The C code for the compensator follows: #define PAR1 . double satout = 0. ' global global global global global global double r = 0. 235 + 23512 . double d = 0. 'if (moterror > THRESHP) satout = THRESHP. .5/(s+ 0. * c. scaling.23) Controller.888310.681~ +=mot_ratel80 1/s.7512 lo00 +loo0 b 0. L .9912 ". Compensator Design 155 I I 0. satout = moterror.0. 8 Hz sampling Plant model.0. (moterror if < THRESHN) satout = THRESHN.0833) (s+0. r. dead zone durout moterror 2. 80Hz + sampling motposition . a scaling block that has saturation and a dead zone. double e = 0.1 . . Theblock diagram includes a saturation link in the highergain.20) The simplifiedfeedbackloopblockdiagramis shownin Fig.99 #define PAR3 . double u = 0.' i .0.15 #define PAR2 . 1464*30/(s+ 30) Fig. lowfrequency path.15 + 0.0. linear links C 1 and Cz. 5.9 { Fig.0. ' .9 1 . / * saturation * / ' ' ' .0. .106)/(s+2.29 Flowcharts corresponding to equations (5.1512 1 .7510. saturation satout + ) c* d 0.987 #define THRESHP 1000 #define THRESHN THRESHP . saturation. / * compensator C1 * / . d = PAR1 * r.30.
by An example of such a link is shown in Fi. This effect called aliasing is illustrated in Fig.5. Modulation of the highfrequency noise in the: S E I link by the sampling frequency and itsharqonics produces frequencydifference products that fall within the signal bandwidth. The motor is ratestabilized by ananalog loop with 30msec risetime. 5.31 Sampleandholdlink circuit diagram Fig. v = PARS * e. The data in the motor model is updated 10 times after each update in the controller. or directly to the plant (when the actuator is also digital).156 Chapter 5.32. adevice which samples the signal and keeps this value at its output until the time of the next sampling. bandwidth of the analog rate loop. .10 times higher thqn that of the compensator. 5. e = u + PAR4 * e.caused of by the limited. 5. Compensator Design d += PAR1 * r. This system shown in Fig. V / * lead C2 * / += PAR3 * e. It is seen that on the basis of the information sampled at discrete points.32 Aliasing The S E I link is a linear timevariable circuit.Therectangularsampleandhold integrator was used for simplicity. The capacitor charges and holds the sampled value of the signal until the next sampling. a digital motor model was employed. and for better accuracy the sampling frequency for the mo.5. 5. 6 Aliasing and noise An A/D convertercontainsa samplendhold . I ) ~  0 sampling Fig. ’ 5 . different rates are frequently used. First.31.30 is an example of a multirate system (although the controller itself is singlerate). the effectsof the highfrequency noise are added to the baseband signal at the output of the A/Dconverter.. The transfer function of z is shownin Fig. In multivariable controllers. 5.. The output of the S E I is processed digitally and then returned to the analog form by a D/A converter at the input to the actuator. and lower rates for slowly varying variables. ( S E I ) link. Le. For computer simulationsin C. u = d + moterror. two importantimplicationsforthecontrolsystemdesign follow.30 intheblock. under the block. 1 0 .30. works as . The switch samples the input signal closing for a short duration at the sampling times. it is impossible to distinguish between the lowfrequency signal with frequencyf and the highfrequency signal with frequency nfi Fromhere.The motor (plant) transfer function s is shown in Fig. The motor transfer function is therefore that ofan integrator (the angle of rotation is proportional to the time the motor is on) with an extra pole. The variable durout is the duration of time that the motor is on during the sampling period of 125msec.g.del was set to 80Hz.an amplitude modulator. faster rates for processing rapidly changing variables. and as such.
5.33 Antialiasing filter like that shown in Fig.10. the attenuation of the antialiasing filter can be . a combination of a large lowfrequency gain of the digital compensator and a small attenuation in the analog plant neartheNyquistfrequency may resultinlargeloopgainandoscillation. closedloop In feedback systems Fig.35.5 octaves than the monotonic for rejection of aliasing noise response shownby the dotted line. the gain of the digital filter for a sinusoidal signal w i t h Nyquist frequency is the sameas the dc gain. and the loop gain response which is close to the optimum looks like that shown in Fig. a highorder antialiasing lowpassfilteriscommonlyinstalled at the input to the sampleandhold link (or AD converter) of DSP systems as shown in Fig.34. The optimal response of feedback the antialiasing filter is therefore not monotonic. 5. To a certain extent. 5. the attenuation of the antialiasing filter at higher frequencies can be equalized by an increase of the gain of the digital compensator (thus making the loop gain as desired) and the introduction of a prefilter in the command path to reduce the inputoutput closedloop gain. Second. Therefore.7 Transfer function for the fundamental In this section. Fig. due to aliasing. 5. the highfrequency sensor noise N causes the output noise Nout in the functional frequency band.Inthese situations. smaller between these frequency bands.3 to 0. To reduce this error by rejectingthehighfrequency input noise.we present yet another view on the effectof digital compensation on the . dB The sensor noise is transformed into the baseband largely from the frequencies close to fs and its harmonics.Chapter 5.34 Control system with antialiasing filter The filter selectivity is limited by its the effect of the loop gain and inputoutput closedloop response. 5.35 Openloop Bodediagrams Fig. 5. Therefore. Compensator Dbsign 157 Aliasing might introduce substantial error in the An> conversion. the sampling frequency must be substantially increased.5. The noise is reduced by the antialiasing filter. Its feedbackbandwidthiswider by 0.33.
. 0 ~ Fig. When the number of samples per signal period is only two (the Nyquist frequency case). (1. 5. aslongasthesamplingfrequencyisrelativelyhigh. 5..37. where (s. and the output amplitude is sin 4.11. amplitude is 1. The input and output signals of the S/H link with 12 samples per period are plottedinFig. When. and the equivalent gain coefficient is 4 / 7 c . 5. 0 I Fig. 5.A linear digital compensator is a linear timevariable (LTV) link.isinversely proportional to the sampling frequency. approaches 0 or 180'.) gradually increases with the decrease of the sampling frequency.36. Let's define the gain coefficient in fundamentals as the ratio of the amplitude of the output signal fundamental to the amplitude of the input signal. 5. is the phaseshift between thes'hpling and the input signal.37 Effect of the phase difference between the signal and the sampling As follows from the Fourier analysis. . In Fig. The system must be made stable with sufficient margins for all possible (s. can be seen by comparing Figs. in 'particular. timedependencies of linear systems can reduce the stability margins and can make the systems oscillate.36 and 5. the amplitude of the fundamental of the output is (4h)sin (s. The uncertainty in the gain and phase (due to as the uncertaintyof (1.158 Chapter 5. It is clearly seenthatthe lag..1 dB..38 are shown the stability margin boundary and an example of a Nyquist plot on the logarithmic Nyquist plane of a welldesigned LTI system. = 90°. the gain coefficient approaches 0. the output.Compensator Design feedback loop. It is seen that the magnitude of this gain coefficient is approximately 1. Consider next the effect of introducing an S/H link in this loopatthe Nyquist frequency.37.36 Signal at the outputof a sampleandhold link Consider thecase when the signal is sinusoidal. the gain is 2. As will be shown in Section 7. When (1. Le. and the phase lag is approximately 15'. 5. the output of the S/H circuit is Ilshaped as shown in Fig. then the phase lag is go'.
Onthe other hand. Because of the relations between the phase and amplitude illustrated in Fig. 8. is usually periodical. zeros (in o. the penalty would be up to 90'. and the shape of the oscillation at the plant output is close to sinusoidal due to the plant lowpass filter properties (this will be discussed fbrther in Chapter 11). [O 0.37. z = 0.2*(x/0. 5.5 Hz. and the settling time is required to be short as as possible. diagram^ and the feedback. by usin! a hLgherorder compensator.38 Stability margin oscillation at this frequency with any possible +. (e) 2. not large. and poles (in a) 0. Assume q = 2.1 dB of gain on the fundamental.5)"2 . The resulting penalty in the transferfunction'uncertainty at Nyquist frequency available feedback is. the gainof the LTI loop at the Nyquist frequency must be below the boundary curves with maximum x .2] can be 2 ( t / ~ ). MATLAB plots the first function with ezplot('2*(3*(~/0. if it happens.( ~ . 6.n/2) + 1]/2.12 Problems 1 The crossover frequency is 100 Hz.2.Chapter 5. What are the pole and the zero frequencies? Will the new Bode diagram be more concave or more convex? What happens to the diagram if the pole and zero are interchanged? 3 Draw an asymptotic Bode diagram for the function having: (a) gain coefficient10 at o =: 0. 5.5. By how much (approximately) can the feedback be increased at (a) 10 Hz. .51) A smoothdigital commandwithsamplingfrequency fs can beexpressedwith function (Ah)( 1 + z'/+ + . boundary and the loop with x dB margin. + z . in radsec) 1. (b) 1 Hz. if. (d) 20 Hz.11 Command profiling The actuator force (or torque) timeprofile is often required not to include sharp pealcs. Compensator Design 159 Oscillation in control systems. (c) 0.where r r = .. ~or ] with q[sin(nt/z. 4.i. typically. the . 5. expressed withq [ 3 ( t / ~)~ Example 1. This is why the sampling frequency must be kept sufficiently high. if the Nyquist frequency should fall on the part of the Nyquist diagram which is phase stabilized. either by passing it through a highorder Bessel prefilteror by replacing it with a smooth timefunction.. zf. The smooth rising of the command from 0 to q over the time interval [0. and because of the extra 2. an asymptotic Bode diagram was made steeper by 6 dB/oct over the frequency interval from10 Hz to 30 Hz. 5.1 dB centered at 270' and go'. 3.Thesystemmustbephasestabilizedat all frequencies below the crossover with margin 3 0 ' . the phase peaktopeak ripples are reduced from 15 to 2 ? 2 By addition of a real pole and a real zero. Theloop gain approximates a constantslope Bode diagram in the Chebyshev sense. to rule out an Fig.. which would require reducing the slope of the Bode.5)"3)'. In these cases.e.5.position or angle stepcommand needs to be smoothed out.72Hz.
160 Chapter 5.' 10 PlotwithMATLABthenormalizedlowpassfrequencyresponsewithapairof 0. 10. t. with the damping coefficient: (a) (e) 0. (b) (s + O. 7 Draw asymptotic Bode diagrams and make the plots MATUB with for the lags: (a) (s + 15)/(s + 2). (b) slope 9 dB/oct. (d) (s + 2)/(s + 4). (c) (s + 0. (c) T(s)= 5000/[S(S + 200)( s + 6000)l. (f) slope 9 dB/oct. (j) slope 6 dB/oct. 1000.125. Usescales: 10 dB11 cm.FindtheBodediagramsfromasymptotic responses using the rule for the error: 3dB at pole. (c) (s + 5)/(s + 2.25.. frequency range 1 to 10 radlsec. and poles (inHz) 60. 0. 1600. 400. (d) gain 20 dB at f = 200.5)/(s + 2. frequency range 1 to 10 Hz.0125. 30.2). 4 Use MATLAB to make Bode ptots for the function: (a) T(s)= lOO/[s(s + 15)(s + loo)]. 1 oct/l cm.5).2).Findthelost feedback at lower frequencies. Use the MATLAB function lp21p to convert the transfer function to that having the . frequency range1 to 10 Hz. (e) (s + 2. complex poles. (d) slope 15 dB/oct. 1 dB one octave from the pole. zeros (in Hz) 15. (e) slope27 dB/oct. and poles (in a) 1. 200.. 9 If the peaking must be 8dB. (b) (s + l)/(s + 0. frequency range 1 to 10 radlsec.72). (b) gain10 dB at a = 2.99. (c) gain coefficient10 at f = 00. (9) slope 12 dB/oct. (f) (s + l)/(s + 16). frequency range 1 to 10 radlsec.20. (d) 0.l)/(s + 0.5. 200. zeros (ina)2. (h) slope 12 dB/oct. (9) (s + W(S + 4)8 Thephasestabilitymargin is excessiveby 10' overonedecade. Compensator Design 5. zeros (in Hz) 100. 5. (e) (s + 7)/(s + 2. (i) slope 18 dB/oct.10)? Find the polynomial corresponding to the peaking frequency 300 Hz. (c) 0.100. 5 Find a rational function approximation of the constant slope function: (a) slope 6 dB/oct.3. (d) (s + 814s + 4). ' 6 Draw asymptotic Bode diagrams and make the plots with MATLAB for the leads: (a) (s + 2)/(s + 15). what is the damping coefficient (use the plots in Fig. frequency range 1 to 10 radlsec.72)/(s + 21). (d) T(s)= 2 0 0 / [ & ~+ 100)(~ + 100O)l. (b) T(s)= lOOO/[s(S + 100)(~ + 500)]. frequency range1 to 10 radlsec. (b) 0. frequency range1 to 10 radlsec. (f) (s + 16)/(s + 2).and poles (in Hz) 0. (c) slope 12 dB/oct. frequency range 1 to 10 Hz. frequency range1 to 10 Hz.1 = 0 dB two octaves from the pole.5.5).
(d) 1OO(s + O.withthedampingcoefficient:(a) 0.72 kHz.Designananalog compensator composed of cascaded links.2. 18 Verify that equations (5. (f) 1OO(s + O. The amplitude stability margin must be 10 d6. & = 300 HZ. .5. (b) 104/[s (S + 3 0 ) (+ ~ IOO)].01. . 16 Using thesameplantandrequirementsas compensator composed of parallel links. 12 PlotwithMATLABthenormalizedhighpassfrequencyresponsewithapairof complexpoles. 5.3. The loop must have a Bodestepand 10 dB/octconstantslopedownto 100Hz. (e) 5000(s+ l)(s + 2)(s + IOOO)/[s(s + 20)(s + 6000)l. Design the compensator for the following plant and feedback bandwidth: (a) l/[S (S + 3 0 0 ) (+ ~ 1000)]. centered at (a) 1 rad/sec. (d) 0. 13 Plot a seriesof 5 notches with the notch amplitude 6 dB and various width. in Problem 15.singleprecisionbecamesufficienttoobtainthesameresults on all computers. are: (a) 3.. and to provide the stability margin./s.intosecondorderlinks. The loop gain response must be steep right afterfb. (d) 0. The asymptotic slope must be 1 8 dB/oct.p. Compensator Design resonance frequency 5 Hz. (b) 1OO(s + O .99 (obtain the response by dividing the lowpass transfer function 2). component of the plant is l.2. (b) 0. (e) 0.22. the asymptote crossing the 1 0 dB level at 2 kHz. (b) 10 radlsec. 20 The polesof an analog compensator. 15 The m. (e) 2. 6.99 (obtain the response by multiplying the lowpass transfer function s).It wasfoundthatasingleprecision simulation on different processors or using different compilers gave slightly differen results. designananalog 17 The feedback bandwidth is limited by the effect of the sensor noise. (c) 10 Hz. (e) 0. Explain why this might have happened.l)(s + 8)(s + 200)/[s(s + 20)(s + 6OO)J. lag of the plant is 1 rad at 2 kHz. 6 dB/oct for two octaves below fb.4.0. fb = 30 HZ. 161 11 PlotwithMATLABthenormalizedbandpassfrequencyresponsewithapair of complexpoles. (b) 0. 8. After the DSP was modified by properly breaking the function of z into secondorder multipliers.1. 14 Break the compensator function into cascaded links: (a) 5000(s + l)(s + 2)(s + IOOO)/[s(s + 20)(s + 6000)l.02. (c) 5000(s + l)(s + 2)(s + 1OOO)/[s(s + 20)(s + 6000)l. the slope of the loop gain must be only. l ) ( s + 8)(s + 200)/[s(s + 20)(s + 600)l.Chapter 5. (d)1 kHz. fb = 3 kHz. &(&) I : 1 rad.withthedampingcoefficient:(a) 0. 19 Ahighorderdigitalcompensatorwasimplemented in C withoutbreakingthe transferfunction.l)(s + 8)(s + 200)4s(s + 20)(s + 600)]. by Use the MATLAB function lp21p to convert the obtained response to that having the resonance frequency 15 Hz. (c) . (C) IO*aS (S + 3)(S + IO)]. by Use theMATLAB fuhction lp21p to convert the transfer function to that having the resonance frequency 50 Hz.and the n. in s. while doubleprecision simulation showed nearly identical results. (c) 0.p.7) follow from the diagram in Fig.
19 in Section 5.0. 25 Consider Example 2 in Section 5. In the function T(s). (a) a version with fb = 1 kHz. a Bodestep.0. (b) a version withfb = 1. 60. +' 22 For sampling frequency fs = 10 Hz. 600.4kHz. the aliasing noiseis of critical importance. The lowest structural mode with the frequency in the 100 to 150'Hzrange results from the cable'flexibility. .14) andlor MATLABcommand bilinear.0. . (b) qs)= 5(s + 2 ) / (+ ~ 3).74/2).1 500. within 3%.1200/2)/(1.600/2).0. respectively. 27 In aspacecraftscanninginterferometer.(b). and a notch at fs as in Fig. 24 Write a programin C for flz) (a). Compensator Design (b) 12.6. 16. (Hint:Eachpolecanbefoundby applying the function bilinear to the functionl/(s sp0le). With sampling frequency fs = 50 Hz. convert to C ( s )from: (a) 42)= (0. (f) ( 2 ) =: (1. (d) 1 0.)  21 Find the Tustin transforms from: (a) qs)= (s+ 3)/(s+ 2). (f) c(s) = 2(s +3)/s.l/Z). (b) @) = (0.1 mm.5. (d) qs)= 3(s + 7 ) / (+ ~20). and move the two real poles from o = 2 to the right until the guardpoint phase stability margin becomes 30'. For sampling frequency fs = 100 Hz.4/2)/(1 .1 . Assume fs = 10 kHz.39 to change the lengths of the optical paths.monotonicresponse. (e) qs)= 15(s + 8)/[(s + lOO)s].andasymptotic slope 1 2 dB/oct. (e) @r) = 0. (C) @) = (1 33. Whele willthese poles be? What will be the loop gain a at = 1 O? Are the technical . 5. 85.(c) from Problem 22.2172 + 0.21 74 + 2174/2)/(1 .O. Bode step. (c) 13.8 with (a) MATLAB and (b) SIMULINK. find the poles of the function z using of formula (5.22 + 221441 . convert toC(s) from: (d) fit) =: (0. remove the stepforming complex poles and zeros. (C) C ( S ) = 1 O(S ~)/[s(s + 4)]. the position' must be accurate within 0.272/2)/(1 0. 5.35. and the 10 dB and 3 0 ' .4444/2)/( 1 .162 Chapter 5.600/2).  23 Design a digital compensator for the analog plant P(s) = 50000(s + 200)/(s+ 300).1200 + 0. The carriage position range is 20 cm.0.0.7391/2). RemovetheBodestep. Consider: g ! n i and phase stability margins are. with the slope of the loop Bode diagram at frequencies below & approximately 1 0dB/oct. and the velocity. specifications satisfied? 26 Make simulations of the system withPID controller shown in Fig.acarnagewithretroreflectors is being moved by a motor via a cable as shown in Fig.80.11 1 1/2).
I I  analog signais B c . the command summer. As discussed in Problem 4 in Chapter 4. 100 Hz sampling " " " " " " " I I .40(a). 5. and the compensator are analog. . This 'sensor is not used for closedloop position control and is not shown in the pictures. 5. The sensor output is digital.b A . the feedback summer. (A more accurate position sensor. the prefilter. so thb control is collocated. 5. samplingfrequency is limitedto 100 Hz sincethecalculationsare performed by the flight computer on a timesharing basis with several other tasks. A I cable * "  Fig. DIA converters . The position sensor (f6bit encoder) is connected directly to the motor shaft. digital signals. the prefilter.Chapter 5.) The D/A converter is placed at the input of the motor driver. and the compensator are digital. is used to measure the exact position of the carriage for taking the science data. the control bandwidth is limited to 6 Hz. Estimateandcomparetheavailablefeedbackbandwidthandthecontrol accuracy (a) in this case and (b) when: the sensor data is read with a rather high sampling rate. Is the accuracy of the analog circuitry sufficient? Draw block diagrams.40 Block diagrams of the carriage control options The .B P r""" 1 Sensor Fig.39 Retroreflector carriage In the block diagram in Fig. Compensator Resign optical beam 163 . a laser interferometer.1 T pptical beam 2 . Consider the advantages and limitations of these two modes of the controller implementation.are placed in the command and sensor paths.
" 101 1oo IO' 10 ' Frequency (radsec) Fig. the Bode step ratio is 2.6s.43' Bode diagrams. +2. 1.1 to 2200.41 Asymptotic Bode diagrams for the lead( s + 2)/(s + 15) 14 (a) 5000(s+ l)/q (s + 2)/(s + 20). o in kradlsec The expressionin Example 1 for the return ratio. Thus. 5.2. or a = 2..l ( ~ + 2 ) ~ . T(s) = 10 s+0.+4 1  s+O.1 5(a).7 kHz. From (5.164 Chapter 5.s2 . .4s+9 s' needs to be modified: (a) I f must be scaled to change a from . 5. 5. 'we express a in krad/sec and becomes 2. The return ratio becomes .42 is similar to that shown' in Figs. Compensator Design Answers to selected problems 6 (a) The diagramis shown in Fig.8 (i. s should be replaced by sJ2.41. 5. For the scaling.2.4 1 s2 +1.42 Ideal and asymptotic Bode diagrams Fig. 5. lag is 1 rad. The asymptotic diagram shown by the solid line approximates the general shape of the idealBode diagram.35 kHz.5 oct). 5. (S + 1OOO)/(S + 6000) 15 The frequency fc must be 2 kHz since at this frequency n.2).16 (Example 1 in. .2 krad/sec. The required ideal loop response shown in Fig.p. and & = fd2 = 0. numbers. To avoid 'large.6) however with a wider Bode step.8 = 0. fd = fJ2. dB 0 log sc 10 20  Fig. Section5. 5.e.
5.)Or.21) MATLAB function conv is used to multiply the polynomials in the numerator: a = [601].7s+ 157) (5.1 1.e). The loop phase response in Fig.42.Chapter 5. b = [l 0.5 (this is already shownin Fig.41. de = conv(d. num = conv(ab. f = [l 5. The return ratio becomes S + 0.9s' + 3 0 9 .c) and in the denominator: d = [l 01. the bode command. thetwo poles at 4.88 2.88) (s + 0.5)2 9 C .22 (s + 4. 7 ~ +~ 21 1 7 . deff = conv(def.2 9 T .40 correspond to the real poles and zeros of Ti@). den = conv(deff.4)2 s + 0.4) + 0.5s + 19.883. ( s ) = 10 s + 0.f).b). the poles at the end of the Bode step must be shifted up 2.811.88 2.5 19. T 2 (s) = 601 s(s (s + 0. def = conv(de.22)(s + 5. e = [l 0. The diagramis close to the desired. Also. ( S > = 5 (s+5. If we do this. The axis must be labeled in kradlsec.4 must be somewhat increased. 5..4)2' s2 + 3.f). orinstead. Compensator Design 165 2.51.92 2. ab = conv(a. Notice that the frequency axis is erroneously labeled in radlsec since we used.wecan just add this phase lag (whichis linearly proportional to the frequency) to the phase response in Fig. wewill see that the system is stable with the desired stability margins.4 1. The lag can be modeled as described in Section 4.5s + 19.88)(s2+ 3.22 s2 + 3. 5. 8 ~ +~ 5200.22) and can be presented as three cascaded links: Ct (s) = 60.8.28s + 43.52s + 19. g = [l 10.9 times. 5.l.5)2 ' .8s The Bode diagram for this function is plotted in Fig.5 I 4.4 s2 + 5.42 does not yet include the n. 5..22 X (5. c = E 1 3. to properly scale the gain axis. The compensator transfer function is T s " 2( ) P(S) 601 + 0.40).7s + 157) (s (5.5s + 19.4) (s + 0.22)(s+ 5. for simplicity. 6 ) .223.22 ( ~ + 5 5 ) ~ s2 + 10.5)2 (s2 + 10.88)(s2+ 3. to 5.p. lag. and COrr8SpOndingly.2 T2(s)=1O s + 0.1 (s + 0.7s+ 157 1 s * " (It can be seen that the corners in the asymptotic diagram in Fig.7 1571.6 s (b) The Bode step frequency ratio must be increased from 1.4~ +~1044.22)(s+ 55)2(s2 + 10. say.42 with w = logspace (1.g) The resulting return ratio is T 2 ($1= 601s3 + 2632s2 + 13510s + 10260 s6 ' + 21.5 to 2.5 = 1.
5s + 19.6s + 64 for the plant lis. with the loop response I (s) + c.5 . d = [l 6. Fig. With the lead. This 1 approximately 1. a lead must be introduced into frequencies.19(b).3.5 times. This response must be modified to widen the Bode step.6 64 0 01. (1) We might start with the compensator from Section 5.7s + 1570 to In these expressions s is in kradsec.8 ~ 3*3 s2 4s s2 + 2. To convert the compensator. 5. . The step length looks about right.d. and increasing ci 1.3( s + 0. Compensator Design q s ) =2 s 2 + 3. w = logspace(l.44 Bode step adjustments Now. 16 Three solutions (among many possible) are given below. functions functions of s in radsec .(#) = (+ + 7.45 Loop Bode diagrams C 1 .10. 4 ~ +2 ~5 .(s) T(s)= C S s5 +6 . bode(n.33). can be done by reducing C The return ratio becomes I.8~ +37.7 having two parallel paths: Cl = 4 4 2 + 4s) and G =5 4 8 + 2.s should be replaced by d1000.9 52.4 25.Chapter 166 s2 5. to reduce the slope at lower T&) = With: 3.w) is shown in Fig.3) (s2 +4s)(s+ 1) + s 2 nl = conv(3.8 37.2 times.44.[ 1 0. 1 80 g P180 270 IO" I 210 240 loo Frequency(mdeec) Fig.4s+ 16 s s5 + 6.4~4 +2 5 . 6 ~ +~ as2 )l * This response plotted with MATLAB commands n = [10. 5.' Frequency (radsec) 8 .(if desired). 5.81. 10.4s + 16) 9sL + 29.4 + 10. 5.l). 6 ~ +~ 64s2 shown in Fig.9s+$2.' loo Frequency(radsec) 10' 10.
den) . dld2 = conv(dl.26) can be decomposed into a sum of partial fractionsasfollows(thecalculationmethodisillconditionedandrequireshigh accuracy in the initial data): nun = [601 2632 13510 102601.23) and plotted.95 radlsec.5)2 (s2 + 10.8~ +~ 46.7s+ 157 (5.d.22 (S + 5.d2).9 309. adjust the zero % (try also zero 0.4 161.4 1044. It remains now only to scale the response for the crossover frequency to be 2200 at by changings to ~(0. The plot is shown in Fig. and all the links are connected in parallel.9512200) = 42316. all other fractionsin the righthand side can be neglected.81. During this exercise.8 5200.41~ +~85.7~ +~21 1 7 . Compensator Design % during iterations.8 (5.4s + 1044.84 s6 + 7. 9 ~ +~ 64s2 (5. 8 ~ +~ 5200. pole 1. Generally. [l 0 1 ) n = conv(nl.88)(s2+ 3.r 1 r21s + r22 " P ( s ) s + 0.4s' + 32s4 + 8 6 . the function (5. adjust the pole d2 = [l 2. the coefficients in the numerators of the fractions (including those with multiple poles) can be found by adding the fractions (5.Pol.22)(s+ 5. bode(n. With MATLAB.18s + 15.24) T2 0 ) P(s) s5 601s3 + 2632s2 + 13510s + 10260 +2 1 .7s + 157) (s (5.7 2117.5)2 + r31s + r32 s2 + 10. The crossover frequency on the plotis 0.25). % during iterations. [l 11).w) T2(s)is converted to the ratio of polynomials T2W = 10. 5. Its shape is acceptable.Chapter 5.K] = residue(num. andsolvingasystem of linearequationsthatresultfromcomparingthe numerator coefficientsof s at specific powers to those in (5.45.601 P(s) or " _c + 0.5. There might be several options for such decomposition.dl) % vectors have equal length since % both polynomials are cubic w = logspace(1.75.26) The numerator of the fraction with a single pole can be found by assigning s the value of the pole and solving the resulting equation for the residue. 9 ~ +~ 309.d2) + conv(7. den = [l 21. The function can be decomposed into the of sum partial fractions T2(s) .4) (s + 0. d = 'conv(dld2.26) in which results in a ratio of polynomials in s.5) 167 dl = conv([l 4 01.l). (2) We willusethealreadyobtainedsolutiontoProblem function 9. Each such transfer function represents a link.5s + 19. Thecompensator T s " 2( ) . [Res.25) should be decomposed into a sum of transfer functions of lower order.
4) (5.7~+157' (3) In the solution to Problem 9. The polynomials canbe found as follows: n u m a r o d = [2*a (2"(a*c + b*d))] d e n s r o d = [I (2*c) (c*c + d*d)] When d is the result of calculation inaccuracy (for the double real poles) and can be neglected.3460 11.1684 0. 8 8 ) ( s2 32s + 634 ( ~ + 5 5 ) ~ s2 +10.1 + 16.1598 0.11.346 + jl1.4940 + 0.1396i 4. the compensator function " _.22)(~+55)~ +10. 5.Oe+002 0.28) num = conv([l 0.4410i 5.42 15.jl1.1396i 0.346 . a pairof complex poles appears.3430i 0. then.44103.22 s + 5.1684 + 0. [l 5.5s+ 19.jb jd) s .j444.72 s+0. instead of the double real poles 5.494 + j0.0. =p(s) s+55 (s+0.22 + 2.2042i 5.221.Compensator Design * Pol = K = [I 0.0172  Chapter 5. and neglecting the small imaginary parts ofthe doublepolesthat weknowmustbereal.5.1 + s + 0.41) dl = conv([l 0.20423.48 (s + 0 .5 19.88].98 .029s . n u m a r o d = [2*a (2*a*c)] densrod = [l (2*c) (c*c)] After making this conversion.53). [l 10. with small imaginary parts. approximately.j20.72 16.7~+157) (~~ can be presented as the product of two fractions.OOOOi0.42 s2 + 5.weobtainthecompensatortransfer function T 2 (s) =P(s) 1.1598 + 0.( c  jd) is a ratioof two polynomials (firstorder to secondorder) with real coefficients.jO.343 1.1396 s + 5.98 + j20. The compensator transfer function is T 2 (s) = P(s) + j444.4940 .494 .2200 Due to rounding errors.22)(s+55)(s2 +10. 5. T2(s) 601 P(S) + 35s + 19.[ 1 3.4) .7 1571) . den = conv(d1.27) (s+0. (5.84 '* The productof two fractions ofthe type a + jb s(c+ a .7s+157) With: .343 + s2 + 5.1396 15.168 Res = l.3460 +11.3430i 4. T2(s) 601 (s+0.84 + .88)(s2+ 3.
25758 s +5. K] = residue (num. d = 11.2973i 0.1 .414s2 ~~ +910.2072 0. 0.2973.33045 5.0151 Pol = 5.0721. sum of the two complex pole fractions of the type a + jb s(c+ a .987 189.971.3500 +11. There are many options of the compensator’s implementation.072 + 1 6 .0151 +0. 0.35. [Res.29) With MATLAB.17)’ Yz) = (16. 6 6 7 ) .414 910. 24 (a) Using (5.3889 .0. Some of them might be better suited .P o l .67 * r. b = 0.3889 + 0.3500 11.for implementing multiwindow nonlinear controllers described in Chapter 1 3 . the function (5. c = 5. r = x + 1.97 (5.j d ) is a ratio of two polynomials with real coefficients which can be found as follows: a = 0.3889.7s+ 157 which is the function of the parallel connection of three links preceded or followed by the link 601/(s+ 5.2200 K = [I The. den = [l 16.987s+ 189.( c .38 22.667 * r.42 219.7778s.jb jd) s . 4 . The C code is: y = 10 * r. 3 8 + ~ 22. prodnun = [2*a (2*(a*c + b*d))] prodden = [ l .33043.15) anci(5.5 + s2 + 10. the compensator transfer function is 0. 4 2 +219.3304. (2*c) (c*c+ d*d)] Finally.Chapter 5. y += 16. Compensator Design the second fraction is converted to the ratio of two polynomials 169 s4 s3 + 4 .5).48 17.671+IO)(z. den) Res = 0.29733.2072 0.48s ~ + 17.29) can be decomposed into a sum of partial fractions as follows (more digits are used here since the calculation method is illconditioned and sensitive to rounding errors): num = [l 4.
With a feedback circuit added. the input signals for the actuators are also electrical. including computer controlled analog compensators. Switched capacitor circuits are reviewed. leads and lags in inversion and noninversion configurations.hydraulic. Following the flatresponse range. prefilter. RC circuit design in the element value domain is described. as shown in Fig. This chapteris the last one in the introductory control course. Since the sensors' outputs and the actuators' inputs are most often analog electrical signals. the amplifiergain drops linearly with a gain coefficient very close fto ~gup until the Univ gain bandwidthfi. Opamps are usedwith largefeedbackto makethegainstableintime. and the use ofthe RGimpedance chart is explained. Tunable compensators. The dc gain of an opamp is typically 100 to 120dB. For example. PID and TID.mechanical. the unity gain bandwidth can be from 100kHz to 1GHz.andothervariables produce.generated as electrical signals. which put sf^ at 5 MHz. an integrator and a differentiator. the commands are commonly . it is convenient and economical to make the command summers and the compensators analog.1. The main building block for these circuits is the operational amplifier. Operationalamplifiercircuitsareconsidered:a summer.1. electrical output signals. Most frequently.1 Operational amplifier Industrialsensors of electrical. these signals are analog. and amplitudewindow circuits are briefly discussed.6.I Chapter 6 ANALOG CONTROLLER IMPLEMENTATION Thischapterexploresavariety of issuesconcerningdesignandjmplementation of analog electrical compensators. with an example of a bandpass tunable compensator design.Insuchcases.the compensator. Further. and also tunable compensators with one variable parameter. saturation.inanticipation of thefeedbackinducedgain reduction. 6. The availablegaintherefore must behigh.thermal. 170 . Methods of loop gain and phase measurements are outlined. the system is stable if close to lc12 up to the frequency the phaselag of the feedback path is less than 7d2 at all frequencies where the loop gain is more than unity. Depending on the type of opamp. then the opamp gain coefficient with no feedback needs to be at least 500 at 10kHz. and feedback path links can be economically and easily implemented with active RC circuits. if the gain coefficient of an opamp with feedback of 10 is required to be 50 at 10 kHz. and compensators with complex poles.1 Active RC circuits 6. command feedforward. The most important issues of analog compensators breadboarding are surveyed. In many cases. Implementations of dead zone. as a rule. The amplifier phase lag is fT. a constant slope compensator. are introduced.. Basic types of RC active filters that are employed in feedback system compensators are examined. and the feedback summer subtractstheanalog fedbacksignalfromtheanalogcommand.
2 = I1. The slope of the closedloop gain is 6 dB/oct over a wide frequency range. connects the signal source to the amplifier input.so that the output signal is Typically. Overthebandwidthwherethefeedbackislarge. 6 . VI = I& and U2 = I&.2 Inverting amplifier and inverting summer (dotted lines) The invertingconfiguration ofanopampwithfeedback circuitryis shown in Fig. where 1 1 and I 2 are the input and output currents. & = l/(sC2).Chapter 6. so that this node potential is very close to that of the ground. Twopole 2. by the dotted lines.. since at the opamp input the voltage is the feedback looperror. should not be too large since it reduces the signal at the amplifier input and increases the thermal noise whose mean square voltage at rootll temperature.The impedance& should not be too small or else the current 'in the impedance and the consumed power will be too big. according to the JohnsonNyquist formula. as is shown . 6. i. The Bode diagrams for the integrator are shown in Fig.theerrorvoltageacrossthe opamp input is very small compared with the input voltage U1 and the output voltage U2. is here.2.Therefore. 6. since the input current of the opamp itself is negligible.itistheparallel connection of the input and feedback resistances. 6. R istheresistancefaced by theinputport of theamplifier. 6. = &) and as a summer amplifier combining signals from different sources.3. on the other hand. It follows that Next. The amplifier can be used as a unity gain inverter (when 2. 1 . 6. 1 the transfer function of the inverting amplifier is The input impedance of the inverting amplifier is Z. it is very small.1 Opamp gain frequency response Fig. Analog Controller Implementation 17 1 dB I Fig. lZll and 1&1 are chosen from5 ki2 to 2 Ma.e. . The impedance Z1. 2 Integrator and differentiator An inverting integratorwith transfer function1/(R1Czs) results when the impedances in the schematic diagram in Fig.(U1/21+ UnJZm+ UJ&)&.2 are chosen to be Z1= R1. and & is the feedback path impedance.
6(a). The openandclosedloopresponsesforthedifferentiator are showninFig.6.  Fig. At higher fiequencies.4 that the efective bandwidth of the difierentiator is much smaller than that of the integrator. To provide some C can be introduced in the feedback loop in phase stability margin. The external feedback circuit for the differentiator is a lowpass filter which.The loop gain is small at very low frequencies where the impedance of the feedback capacitor is very large. 6. 3 Integrator gain responses Fig. Analog Controller Implementation 1 y  ~ o o inputoutput p gain dB 0 Fig.6. produces a double integrator in the feedback .172 dB Chapter 6. An inverting differentiator can be implemented with Zl = l/(oCl). must be connected to the ground. a lead compensator front of the opamp to reduce the gain at lower frequencies.orthe voltage follower described . the openloop gain is the product of the opamp gain coefficient and the coefficient of the voltage dividerR1/(RI + &). . dominantinthedenominator and. &. At medium and higher frequencies. the integrator is not accurate when the integration time is very long.5 Differentiator schematic diagram 6 . Therefore. R1 becomes.4. the loopgain decreases as a single integrator.4 Differentiatorgainresponses When the openloop gain is being calculated. the right endof the twopole . 6. 1 . It is seen in Fig. as shown in Fig. the integrator does not perform well at these frequencies. 3 Noninverting configuration The unitygainamplifiershowninFig. 6.Z. together with the opamp itself. 6.= R2. Therefore. This is one of the reasons why integrators but not differentiators are usually employed in analog computers.5. That is. with 90"phase stability margin. the gain coefficient about the feedback loop is " fT Rl f 4+Z2 It is large and nearly constant over the frequency range where R1 < &I =l/(oC~). 6 .loop.
K = 2 +2l . 6. keepthese resistances.Chapter 6. this charge will remain for an unpredictable time.double.e. 6.4 Opamp dynamic range. has the feedback path transmission coefficient B = 1.6 is low over the range where the feedback is largg. Hence. the largermust be the current in the transistors and consequently. . the higherfT is. Le. As a rule. Analog . The pinout is the same for the dual in line package (DIP) and surface mount (SM) package (there also exist much smaller SM packages).6 Follower (a) and noninverting amplifier (b) The fed back signal in both the inverting and noninverting amplifier configurations is proportional to the output voltage. The feedback path transfer function is B = &/(& + 22). The input voltage dc offset (the internal parasiticdc bias) is typically in the lOnV to 1mV range. this feedback loop stabilizes the output voltage. 6..orquadinonecase. neither of the input pins should be left open or . Therefore. the output impedance of the circuits in Fig. external resistances in series with the opdo increase the device noise. 6. the feedback amplifier transfer function is 1/B.ControJler Implementation 173 briefly in Section 1. While opamp circuits are tested.amp from the power supply. 6.6(b).. The standard pinouts for the doubleand the quad are shown in Fig. small.3. The choice OffT depends on the desired gain and the power consumption. It is desirable to amp input. The input noise is typically comparable with thethermalnoise of aresistor of several hundredohmsto severalkiloohms connected in series to the input. Opampscomepackagedassingle. and packaging The dynamic range of the opamp is limited from above by the output voltage swing.7. Due to the extremely high impedance of the opamp input.like RI in Figs. noise.especially intheopampimmediatelyfollowingthe feedback summer.5. i. Therefore. z 1 (a) (b) Fig.2 and 6. else the voltage on the open pin will depend on the initial charge. and from below by the input noise and drift. withthe thermal driftin the 1 nVPC to 10pVPC range. 1. thusproducingconfusionfortheexperimenter by alteringthereadings inanoften irreproducable fashion. makes the output voltage nearly independent of various disturbances and of the load impedance variations.1. 6. The schematic diagramof a more general noninverting amplifier configuration is shown in Fig. Fig. the dc current consumed by the op.6(b).
respectively. 6. 6. these time constants can be large and the capacitors become bulky.8 Typical size of capacitors mylar One quad opamp is sufficient for all the needs of a typical analog compensator. 6.9 Lag (a) and lead(b) implementation. 6. Typically. R = 2MSZ and C = 0.6 where 2 1 and Z . lead compensators are shown in Fig.5 Transfer functionswith multiple polesana zeros With the feedback amplifiers shown in Fig.8. resistance is limited by stray capacitances.1. Analog Controller Implementation cc +vcc 8 1 7 SM.1. transfer functions can be realized with multiple real poles and zeros. however. Thus. and costs an order of magnitude less than a digital microcontroller. An inverting lag compensator is shown in Fig.7 Typical doble and quad DIP and SM packages OA. (Design of these compensators will be discussed of the €ur&er in Sections 6. noninverting Inverting lead compensators are showninFig. and 9 at the end chapter. 6. order of magnitude less time than doing of the transfer The values of the RC constants correspond to the poles and the zeros functions.2.16 Hz. Typically.10(c). for example. 6. The resistances are.) . The higher the resistances are. limited. Breadboarding and testing such an analog controller takes. the smaller the capacitors can be.174 Chapter 6. The resistance in series . (a) (b) fig. like that of mylar capacitors. For lowspeed processes. occupies an order of magnitude smaller space.10(a) and (b). Fig.6. and in solutions to Problems 7. 6. 6. typically. The typical size of mylar capacitors is shown in Fig.2.8. resistor values should not exceed several megohms.2.6. an so with a digital controller. The capacitance must be stable in time. are RC twopoles.5 pF can produce a pole or a zero at a frequency of f. same pinout 1  1 C 'c DIP TOP VIEW Fig. Noninverting lag and.with the opamp input is limited by the andthefeedback requirements of keepingthethermalnoisebelowcertainlevel.itconsumes an order of magnitude less power.9(a) and (b).1427cRC) = 0.
and (b) and(c) complex poles The bridged Tcircuit in the feedback path shown in Fig. Analog Controller Implementation R2 175 R3 f f f Fig. 6. B dB Ty 01  Fig. 6.1 1 Implementation of transfer function s"'~ over the band from 1 to 100 Hz Fig.12(a) shows the bridged Tcircuif employedinfront ofanopampto implementcomplexzeros. 6.13. 6. The gain responses for the amplifier using the feedback pathB1. the .12(c). (This circuit has been used as a part of the compensator for the 100kV.12 Implementation of (a) complex zeros.12(b) allows implementation of a complex pole pair.)Thefeedbackpath B2 dominates at lower frequencies. 6. and both.10 Lead (a) and (b). The mutualcompensationofthephaseadvancedoutput signal of +upper path B1 with the phasedelayed output signal of the path B2 produces a broad notch on the Bode diagram. 6.1 1 shows the implementation of transfer function approximation (5. 6.Chapter 6. feedback pathB2. 6. and lag (c) implementation. Implementation of a complexpole pair using parallel feedback paths is also shown in Fig. At the crossing frequency. inverting Fig. are shown in Fig. The path B1 dominates at higher frequencies. As the result. 1.2) shifted to cover the band from 1 1OOHz. to sln which uses Fig.6MW precisionpowersupplyforaklystrontransmitter. the output signals of the feedback paths have nearly opposite phase so that the total feedback pathhas a pair of complex conjugate zeros.
13 Compensator with parallel feedback paths 6 . and less than 10 when Q approaches 10. (The element values of this filter become inconvenient when Q exceeds lo.) For this filter. Fig. shown inFig. The qualityfactor Q depends on the difference in phase shift between the two paths. and Q = 1 0 (R. andvaluesfortheremainingtwocanbefoundfromthe equations. For example.. It implementsthetransfer function The lowfrequency gain coefficient Ho is limited by the prescribed Q. + R2)C2 This filter is wellsuited to the implementation of lowQ complex poles like those requiredtomakeBodesteps intheexamplesinSections 5. 6.14 SallenKey lowpass filter Fig.176 Chapter 6. Ho must be chosen less than 100 when Q 2 1. 6 Active RC filters Fig. 6. The difference can be adjusted by adding series or parallel resistors to the capacitors. 6. 1 .7. Analog Controller Implementation closedlooptransferfunctionpossessesacomplexpolepair. 6. Two of the circuit elementscan be chosen. C 2 = l/[ao(Rl + Rz)Q]. and C1= ll(a2C2R4?2). Fig.1 5 Multiple feedback lowpass filter A mulfiple feedback secondorderlowpassfilter. 6. two of the element values can be chosen and the remaining three calculated from . has reducedsensitivitytocomponentparametervariations.Q. when resistor the values initially are chosen. Particularly.6 and 5.15.14 shows the schematic for a unitygain %//enKey secondorder lowpass filter with the transfer function K(s) = where a : ' s2 + Q*aos + 1 c o : 0 = . The pole frequency coo and the damping coefficient = 142Q) are prescribed. .
6. A notch (s2 + a2)/(s2 + 2&00 + a : ) can be implemented with the twinT bridge . (b) twinT notch The statevariable filtersare available as ICs.2.Chapter 6.05 68 4. This circuit enables reliable implementittion of poles and zeros with Q up to 100.7 4.3 12.The corner frequency can be changed by changingtheproduct of the capacitances. The circuit can mimic secondorder linear differential equations describing lowpass.7 0.129 0.22. changing both capacitances by some coefficient to preserve The circuit shown in Fig.5 12.17 7.This is an analogcomputerconsisting of a summer with gain adjustment resistors and two integrators. and Q. The filter response remains the same. These circuits combine the best of the analog and the digital worlds: they are easily reprogrammed.g. Harris.8 47 4.7 1.4.Qand ao. and bandrejection filters.305 1.7 0.) Flg.493 1.when all resistances are increased and all capacitances reduced by the same factor.TherealsoexistICfilters inwhich all resistors are built in and are' programmable or controlled from a computer port. andthe gain. National Semiconductor).1 gives the elements' values for the Chebyshev secondorder lowpass filter with Ho = 1 and 1 kHz cutoff frequency.13 12.diagram is similar to that shown in Fig. by changing theratio of the capacitances. with only four resistors to be added to set thecutofffrequency..1 6 (a) Statevariable filter.803 Table 6. bandpass. BurrBrown. Programsforcalculatingelementvalues andplottingfrequencyresponses for SallenKey. 5. = R2 R3 1 c2 e fo 8. (An example of using bandpass compensators with statevariable filters will be given in Section 6. and statespace filters are available from many manufacturers (e.907 3.The gain coefficientcan be changedby changing R l andthenadjustingtheratio of thecapacitancestopreservethedesired .1 Chebyshev 1 kHz lowpass filter 1 1 2 1 3 1 4 1 2 0.01 6. The block . 6.841 0.6 12.956 1.16(a) is often called asfafevariab/efi/fer. Analog Controller Implementation 177 the following equations: Table 6. andthey do not introduce the delay associated with sampling. multiple feedback. Q. MAXIM.8 100 220 4. highpass.
= i/(RC). The resulting saturation response.4%of the nominal resonance 5" lagatthefrequencytwooctavesbelowthe frequency. Example 1.17(b)shows animplementation of suchacircuit.16.17(c). and around 0.Fig.17(d). with the gain coefficient of 2. characteristicswith different dead zonesand saturation thresholds . 3V on the Zener with inverse polarity. 1 . 7 Nonlinear links A saturation link can be implemented with an opamp as shown in Fig.whileintroducingonly resonance. and the dashed line. By using different resistors and Zeners. vcc output Ioutput Fig. The 5 V saturation threshold shownin the figure results from the sum of the voltage drops across the open diodes: 4 . The resulting characteristic is not a pure dead zone . The resonance frequency a .The dotted line indicates signal transmission via the upper path.it includes saturation at the VCC level. 6.16(b).7 V on the open diode of the other Zener.17 Saturation link (a) and its characteristic (c). via the lower path. 6. 6.178 Chapter 6. Two cascaded notches with 5 = 0.01 to 0. be The potentiometer allows the adjustment of the denominator damping coefficient 5 over the range 0.6. and of the damping of the numerator (which must 0). This arrangement uses Zener diodes in the feedback path that have small differential resistance when the voltage across the diode exceeds the threshold. 6 . The variable resistors allow the adjustment of a . 6. and a dead zone link (b) and its characteristic (d) The dead zone link can be formed by summing theinput of a saturation link with its invertedoutput.17(a). Analog Controller Implementation showninFig. is shown in Fig.1 and slightly different resonance frequencies.andthe resulting responseis shown in Fig.canbeusedtorejectaplant structural resonance by at least 40 dB over the range *1.1 % underand1%overtheresonance. 6.
Analog Controller Implementation 179 can be obtained.6. The circuit to combine signals via different in Fig. (4 Fig. Sometimes it is desirable to direct signals that haveamplitudeswithinspecified windows to separate outputs for further processing.whose block diagram is shown similar way. 6.17(b) do just that.Chapter 6. 6. Fig. To an extent the circuits shown in Fig. Further examples of nonlinear dynamic links will be given in Chapters 11 and 13. Fig. depending on the signal amplitude. 6.19 Rate limiting follower. 6. 6.18 Threewindow (a) splitter and (b) combiner Nonlinear dynamic links canbe designed by combining nonlinear and linear links. rate can be limited by placing a saturation link in front of an integrator and closing a tracking feedback loop with sufficient gain coefficient k as shown in Fig.18(a) exemplifies a circuit with three windowswhichdirectstheinputsignalintothreeoutputsforfurtherprocessing. . and also to reduce the overshoot in the control system response to large commands. for two windows. (c) ramp output following step input . (b) schematic diagram.18(b). (a) block diagram. Rate limiters are often included in the command path to prevent the plant from being damaged by excessive velocity.19. can be designed in a amplitude windows. For example.
180 Chapter 6. these poles and zeros alternating along the real axis' of the splane. 6. The element values are the coefficients of a chain fraction expansion of the impedance function. and the closest to'the origin being a pole (this constitutesa part of R..Notevery function can be implemented as an impedance of an RC twopole but only one whose poles and zeros are real. den] = residue(r. In general.9. are often placed in the feedback path of The anopamptoform a compensatormakingtheloopBodediagramsteeper. .11 and many others. p = [pl p2 pi . 2 .. any RCimpedance function can be implemented in any of the ladder ('W.10. The impedance frequency response of this twopole with known elements in each branch RiCi can be easily found with SPICE.2 Design and iterations in the element value domain 6 . 6. this is the basis of collocated control discussed in Chapters 4 and 7. a series connections of R2 and C2. ml. Example 1.6. bode(num. and LCtwopoles). 2 . [num. 41' Fig. of the pole copl .21 Cauer (ladder) canonical forms of RC twopole (Notice that Foster's theorem also states that the impedance of a lossless (LC) system has alternating purely imaginary poles and zeros. 6. o z 2 = W2C2).20 Foster canonical forms of RC twopole Fig.. 6.den) .) Also. (R1+ &)/( llC1 + 1/C2). 6.I. = ~lRd(R1 + R2) Uzl ap1 = from which the element values can be easily found. . The Foster and Cauer twopoles are employed in analog compensators in Figs. the firsta series connections of R1 and Cl. 6.p). the response can be found by first q = 1/Ri and poles pi = l/(RiCi) . Foster's RC. AnyRCimpedance theoremthatgenerallyconsiders function can be implemented in either of the Foster canonical forms shown in Fig. the capacitances and resistances for the Foster form in Fig.20. RL.and then using calculating the residues r = [rl r2 ri .20(a) can be calculatedby expanding l/[sZ(s)] into a sum of elementary firstorder functions. and the second. Two parallel branches. frequencies of thetwozeros cozl and 0 .Cauer's) forms shown in Fig.. In MA'IZAB.21. Analog Controller Implementation 6. 6.I. The parallel branch form renders smaller total capacitance. 1 Cauer and Foster RC twopoles RC twopoles are widelyusedascomponentsofanalogcompensators. andthehighfrequency asymptotic valueof the impedance are expressed by the following equations: = 14RlCl)..
Thecapacitance C 1 isdominantatlower . We find the elements’ values for the Foster.p 1 ’ r2 S.OOOOOl 0. Consider the circuit shown in Fig.75 MSZ.266 MSZ. and the lowest among them is a pole.000(s + loo)(s 3.they alternate. 6.. Y(s)/s withthe Next.7895x1O4. In this case. 100. Cl= 390 pF.1 nF.2000) s( s + 500) canbeimplemented as anRCcircuitsincethepolesandthezerosarereal. The MATLAB code nun = [O. Therefore. p] = residue(num. the resistors and the capacitors are: R1= 1.Chapter 6. rather than in terms of poles and zeros.22 Example of (a) an RCtwopole and (b) of its impedance modulus response Weuseinthisexample C z << C1. den) calculates the residues: O. den = [l 2100 2000001. O.21(a) by expanding the function into the sum of two componentsof the partial fraction expansion num( s) ” + den(s) s. Analog Controller Implementation 181 Example 2 . and the poles: 2000. by comparing a terminthepartialfractionexpansionof admittance of a series connectionof a resistor and a capacitor we identify R = llr and C = r/p. While working in the laboratory. 6. Example 3.p2 + .R2 = 4. it is often more convenient to think in terms of the element values R and C. [r. C2 = 2.Impedance Z(S) = 1.000..form in Fig. Cauer forms can be of use.2105x1O4.00051.22(a). (a) (4 Fig. 6.
2 .starting with fi.for preliminary design). From this analysis it is seen that it is easy to identify the elements which need to be adjusted. i. 6. many design questions can be instantly answered. what is the series capacitance that will place the pole at 20 Hz?. it is 15 nF.e. R2 needs to be increased. Fortheleadlink in Fig.22(a)needstobe hm?ased at Specific frequencies.23. it is easy to see from the plot in Fig.C2) approximately equals . The same method is convenient to use for loop adjustments with SPICE simulation. If the source impedance is 100 kGl and the load resistance is 400m.10. 2 . Example 4. i. what isthecapacitance of the at capacitor that is shunting the 100 ld2 resistor R3. Example 3. for the compensator zero to be 200 Hz? .22(b) what needs to be done:at lower frequencies. i. At even higherfrequencies. Therefore. and the total impedance becomes R1##R2 = RIRJ(Rl + R2)... Analog Controller Implementation frequencieswhereitsimpedance X 1 = 1/(2nfi is muchhigherthan R1. Example 1.the capacitance must have 500kGl reactive impedance at 20Hz. If the source impedance is 2 0 m .182 Chapter 6. it might be desirable to reduce the values of the capacitorsand to convert the Cauer form into the parallel Foster form.. where is the pole frequency?. the capacitances’ impedances can be viewed as negligibly small. With this chart.20 kGl. it is 16 Consider next the design of the inverting compensators depicted in Fig.e. RCimpedance chart A chart for calculating the corner frequencies and for determining the values of resistors and capacitors is shown in Fig. and the computer might be on a faraway desk. 6. which is sufficient for compensator design (at least.What resistance needs to be placed in series with this capacitor tomake a poleat 1kHz?. dominant.e. the load impedance is 20 w1. C2 needs to be reduced. When the final version of the design is being implemented. 1 dB. In this way the loop response canbe adjusted in the laboratory. Beyond thefrequencyh where the impedanceX2 = 1/(27c.The reactive impedance of shunting capacitance is 1 the capacitanceat the pole frequency equals the parallel resistance of the sourceand the load which is 10 kGl. it is also possible to start with and make the iterations using the Foster form. at higher frequencies. and the pF. With some experience. Startingat frequency fi where X1 reduces below R1.C 1 needs to be reduced. where X2 = 1/(27c&C2) equals R1+ R2.10(a). the pole frequency is at the crossing of the 1 pF line H z .Since the total contour resistance is 500w1. over the interval&&I. R 1 needs to be increased. 6.7 nF. The chart accuracy is approximately lo%. and the 10 ki2 line.R2. the number of iterations is substantial. If thetotalimpedance of thecircuit inFig. . over the interval lf2&]. Example 2. 6 . Using the chart is especially convenient in the laboratory environment where the accuracy of calculation need not be high. the impedance of the small capacitor C2 becomes small enough to shunt R1 and becomes dominant.6.6. the resistor R1 becomes.
6.20 kS2.e. 6. for the compensator pole to be at 200 Hz?. I 1Hz 10 100 . resistormust provide the same transconductance of the feedback path at this frequency as the Tbranch at the pole frequency. 0 1 . it is 100k. analog or digitally controlled A multiplying . Using the converter. This method combines the best of the digital and the analogworlds:thecompensatorisanalog. Therefore. 6.7 nF. which is 5 times smaller than the 5).Chapter 6.10(b). Analog Controller Implementation . the resistance at the connection point of the shunting capacitor the isparallel connection of R3 and R4.10(c). controllable attenuator) for changing the coefficients . Design of the 6. design of thecompensators compensators in Fig.3 Analog Compensator. In this way. Example 5.24 shows the application of a P multiplying DIA converter (i. Fig. is done in no time. 6. what is the capacitance of the capacitor that is shunting the'100 ksz feedback resistorR2.withoutdigital delay. 6.01 1 183 1 kHz 10k 100k .I. yet its frequency response can be modified by software. For the lead link in Fig. When R3 = R4 = 200 ksz. What is the resistance 1 kHz? This'. it is possible to change the parameters of an analog compensator via a computer command.D/A converter is anattenuatorwithadigitallycontrolledattenuation coefficient. What is the resistance R3 that makes a zero atkHz? 1 . For the lag link in Fig.7 nF. I 1Hz 10 100 1 kHz 10k 100k Fig.2.10 is left as an exercise.. Whatisthecapacitance of the capacitor for the R2 for the pole to be at compensator zero to be at 200 Hz? ..23 RGimpedance frequency responses Example 4. transconductanceatdc(sincetheratio of thepoletothezerois R2 1 Ma.
andamplifiers. controlled digitally 8 ' Fig.Whentheswitchingfrequencyisvaried. and Bessel filters. for D/A Multiplier ~ Fig. Analog Controller Implementation and I in a. the frequency response of theswitchedcapacitorcircuit will shift onthelogarithmicfrequencyscalewithout changing its shape.6.27. The resistor is imitatedby charging a s'mall capacitor C. 6. This method of chargetransfer(chargepump)isequivalenttoaresistance R 2: l/(fsC). and switched . . 1.25 and Fig. the resistor can be replaced by a circuit that transfers the chargein small discrete steps.allpoles andzerosofthe transfer function will change by the same factor. Butterworth.24 Analog compensator.4 Switchedcapacitorfilters 6.largeresistorstendtooccupy substantial real estate. as shown in Fig*6. However.1 Switchedcapacitorcircuits Since the size of capacitors in analog circuitry is limited by economy considerations. f s times per second. Therefore. Fig. There exist ICs implementation of secondordertransferfunctions withbuiltinmultiplying converters thatare controlled froman external parallel bus. summers. 6.25 Gain regulation with 'a multiplier Analog multipliers can be used 'to change the gain coefficient of a link under the control of an analog signal. to achieve the required l&ge time constants. compensator with transfer function P + Ils. 6. the resistors must be big. and then discharging it into the load using an electronic switch. Fig. 6.26 Switchedcapacitor frequency sampling changing with integrator Fig.4.27 Change in thefrequencyresponse Switchedcapacitor activeRC circuits use switchedcapacitor integrators. log. 6.184 Chapter 6. Switchedcapacitor Chebyshev. This makes the switchedcapacitor filter easy to tune. as shown in Fig. As an alternative. 6.19. when active RC circuits are implementedassiliconICs. 9c.26 showsaswitchedcapacitorintegratorwithtransferfunction fSC14SC2)* dB 0 r.
6. A decoupling matrix with constant coefficients was calculated and implemented as an analog circuit. Therefore.Chapter 6. The frequency of the cooler operation is between 50 and 60Hz. The modes’ damping can be quite low so that the peakson the Bode diagram can beup to 40dB high.2 Example of compensator design As an example of the applicationof switchedcapacitor circuits in control compensators. experiments show that coupling from the ith piezoelement to the jth sensor is less by an order of magnitude than the coupling from the ith piezoelement to the ith sensor. Due to the system parameter variations. This disturbance rejection feedback system is a kind of homing system.whichcanbeeasilyimplementedby changing the sampling frequency for the switchedcapacitor compensator. are available from several manufacturers. the feedback of 100 cannot be implemented over a bandwidth wider than 5 Hz. 6. with zero command. with the feedback frequency response adjustable. The threeaxis sensor arrangement is shifted by 90” from the arrangement for the actuators.28. Analog Controller Implementation 185 capacitorbiquadsprogrammablefrom an externalparallelbusor by connectionto ground orto the VCC of some of the IC pins. in order to effectively reject the disturbances with frequencies in the 50 to 60 Hz range. As can be calculated. 6. the three loops can be considered fairly independent. The modes’frequencies are keptover300 Hz.3outputcontrolsystemforrejectingthevibrations of a spacecraftcamera’sfocalplanesensors. the reliable bandwidth of the feedback is only 3 Hz while the frequency ofthe cooler motorvariesby 10 Hz. Fig.28 Control loops for vibration suppressionof a cryogenic cooler for a camera focal plane The feedback bandwidth is limited by the flexible modes of the cooler armature. Thus. With the matrix. if the feedback system is implemented as a lowpass system. Therefore. the system should be the bandpass type.Thesensorsarecooled by acoldfinger connected thermally and mechanically to a cryogenic cooler as shown in Fig. althoughtheir exact valuesremain unknown. The frequency is controlled by a spacecraft computer in accordance with the temperature of the cooler. Therefore. we considera3input. the system needs to be designed as adaptive. the Bode diagram shouldbe slightly shifted along the frequency axis withoutsubstantialvariations of its shape . The cooler’s vibration is counteracted by three piezoelement actuators placed orthogonally between the cooler and its armature.4. .
and for simplicity is commonly considered tobe zero. 6. 6.29(b). a power supply for the circuitry with the schematic diagram shown in Fig. Two cascaded highQ filters have steep rolloff with phase shifts of k180". a ntype RC lowpass filter to smooth the ripple of the output voltage. is a bandpass filter with quality factor 3. first. The diagram includes a power downtransformer. was controlledfromtheflightcomputerto vary thecentralfrequency. as required.186 Chapter 6. The sampling frequency. noise. and a voltage stabilizer IC. in most cases. within the range of 50 to 60 Hz. Wlis a bandpass filter with quality factor 30. but not in all cases. 6. 6. . Still. This assumption is acceptable.as in the following. The resistance of a ground wire or a ground plane (a grounded layer of a multilayer printed circuit board) is commonly very small.29(a). a twoway diode rectifier.. 6.30.and can be performed by people without a special background in electrical engineering. These filters dominate at frequencies in the functional band (approximately 2 Hz wide) as seen in Fig. The ac current through the filter capacitors is relatively large. with shallow slopes at the crossover frequencies. To reduce the voltage variations on the analog ground.5 Miscellaneous hardware issues 6. nominally of lOkHz. The. dB 0 (a) (b) Fig. and signal interference may result from improper grounding. At the higher and the lower frequencies.29 Compensator frequency response (a) and block diagram (b) The block diagram for the compensator is shown in Fig. Signal distortion. The response is a bandpass version of a PID controller. some assorted hints and warnings are worth noting. the transformer tap wire should be placed close to the capacitors. currents flowingvia the ground plane to the central tab of the transformerwindingcreatevoltagedrops onthegroundplanethat are typicallyof several pV and can even reach several mV. Each of thesecondorderbandpassfilters was implementedusingabandpass secondorder switched capacitor filter IC. . especially through the first capacitor. and rY. Here. andshown in Fig.1 Ground Practical implementation of analog compensators is simple and forgiving.5. the lowQ channel is dominant with shallower slopes of the Bode diagram and smaller phase shift which provides the desired stability margin. 6. for example.29(a). Consider.Analog Controller Implementation The loop Bode diagrams for each of the three channelsare identical.
Any extra connection between the grounds would create paths for the ac currents to flow via the analog ground and to produce voltage drops. the actuator and the of plant a control loop are analog. The ground potential at the input to the receiver differs by Vg from the transmittergroundpotentialbecause of variousacanddccurrentsfromdifferent . In the digital partof the circuit.2 Signal transmission In some control systems. 6. the sensors are far away from the controller and need to be connected with long cables. 6. The digital currents should be prevented from flowing over the analog ground and there producing voltage drops. The resulting communication link is subject to noise and electromagnetic interference. In the analog part of the circuit. and the sensorin most cases is also analog.3 1. Fig. The ac current also flows via the power supply VCC line.5. Analog Controller Implementation 187 and second.Thelowimpedanceacconnection between the grounds via power supply lines must be eliminated. the blocking capacitors from the power lines should be connected .The digital part of the loop starts with an An> converter following the sensor. otherwise different points of the analog ground would have different potentials.30 Ground for voltage regulator Fig. 6. attention needs to be paid to maintaining proper ground configuration.Therefore.provisionsshouldbemadeforthedigital currents not to interfere with the analog input signal. This is achieved by connecting the analog and digital grounds at one point only.30. to an asymmetrical receiver using common ground as the return wire. the rectifier ground and the analog ground should be connected in a single point as shown in Fig.theimpedance of thewiresmightintroduce parasitic feedback and coupling due to imperfect ground. commonly. it isc o k o n l y sufficient to connect the VCC line to the ground by capacitors placed close to the IC that consume large pulsed current from the power supply line. Rectifier ground Fig. Since commonly the analog and digital grounds are already connected within an An> converter.31 Grounding for AID converter In systems employing digital controllers. The currents in the digital part of an A/D converter are typically several orders of magnitude larger than the lowest analog signal to bemeasured.Chapter 6. 6.32(a) shows signal transmission of the asymmetrical transmitter output voltage V. athigherfrequencies. and care should be taken to avoid parasitic coupling betweenthegrounds. lowpass filters are usually placed in the power supply line.to the proper ground points as shown in Fig. 6. 6.To do this. In particular. this point must be the only one connecting the grounds as shown in Fig. The capacitors of the filters should be placed close to the IC or else.3 1. 6.
thus suppressing the interference. 6. especially whenthe links are connected with rather long wires. Precision amplifiers with differential inputs are called insffumentafion amplifiers.thecoilimpedance is high. 6. purchased as complete ICs. Analog Controller Implementation sources flowing through the imperfect (nonzero impedance) ground wire. the coil impedance is small the since signal current flows through the two coil windings in opposite directions producing no magnetic field. bifilar series coils are employed for suppression of the interference at higher frequencies where it is rather easy to make the coil with high inductive impedance. VW") return wire/ground *  Fig. For the signal. and these voltages due to their opposite polarity cancel each other at the input to the differential receiver.188 Chapter 6. differential amplifiersare used with a separate return wire which is only connected to the ground at the transmitter output as shown in Fig. = VtVg is corrupted with the ground noise Vg.33. 1Ok Fig. 6. For theinterferingsignalaffectingonlyonewire. better yet. the received signalV. As a result. 6.32(b).32 Parasitic feedback due to common ground To avoid errors caused by imperfect ground.33Differentialamplifier Electromagnetic interferencefrom various sources in an industrial environment into the signal wires could contaminate relatively small analog signals in the control loop. The amplifier is an acceptable replacement for an instrumentation amplifierwhen the common signal component does not exceed half of the VCC (so it will not saturate the amplifiers) which is commonly the case. 6. They are commonly composed of three opamps or. As shown in Fig.34(a). Twisting the pair of wires reduces the difference in the voltages inducedby external magnetic fields in each of the wires. . forward wire return wire ~ ~. A simple differential amplifier with the gain coeficient of 2 can be built according to the schematic diagram in Fig.
6. 6. long and their inductances form a an resonance tank amplifying the signal in the parasitic feedback loop. Analog Controller Implementation 189 (a) (b) Fig. tuned.especiallyathigher frequencies and when the wires are long thus introducing stray inductances and capacitances in the feedback loop. (But. when a signal analyzer is used.36 Reducing the effect of cables Amplifiers are often connected with the rest of the equipment with long cables.thecable capacitance andtheresonances inthecablecanmaketheopampunstable.34(b). Since transformers do notconveydc. f Fig.5. 6. balancedtounbalanced transformers (baluns) can be used as shown in Fig. 6. some power amplifiers withfT of only 2 MHz can oscillate at 50MHz. 6.. The resistances cannotbe too large since the input resistor introduces noise into the circuit.and the output resistor increases the amplifier output impedance. 6. if the wires connecting the IC are too. this oscillation might manifest itself 'by only a small change in the gain and some reduction of the output power. 6.thismethodisonlysuitablefor signalsnotcontaining dc components(likesensorsignalsinfeedbacksystemsfor vibration suppression).34 Using (a) bifilar coil to reject highfrequency interference and (b) balun transformers to reject lowfrequency and mediumfrequency interference To reduce interference at lowerand medium frequencies while using asymmetrical receivers. Without oscilloscope. For example. Someopampsinthisconfiguration are onlymarginallystable.Chapter 6.3 Stability and testing issues The internal feedbackin an opamp is maximum when the opamp is used as a follower.The resistances arecommonly chosen from1 to 2 ksz.Without an oscilloscope.The instability can be eliminated by the additionof series resistorsat the input and output of the amplifier as shown in Fig. Since the cables'inputimpedancesshuntthefeedbackfortheopamp.35 commonly provides sufficient stability margins.36. at the . Placing a resistor in the feedback path as shown in Fig. An oscilloscope should be connected to the breadboard while the circuit is tested.ortroubleshooted. many possibleproblemscanbe misunderstood.35 Using a feedback resistor opamp feedback loop an follower on in the Fig.
or in the paths from the sensors.37(a).38 shows a block diagram of a feedback system witha PID COmpWlMfOrIls + P + Dqs/(s + 4). it is a good practice to leave one of the opamps as an extra available inverter for the purpose of changing the phaseof the signal from the sensor . the input. 1 PZD compensator Fig.bematched can corrupt the measurements. the scalar parameters P.37(b)isconvenient to have implemented on the board. While designing an experimental breadboard. troubleshooting. Analog Controller Implementation same time.37 (a) Breadboard with connectors and (b) oscilloscope matching termination Signal reflections from mismatched ends of a coaxial cable going to an oscilloscope should. breadboard. especially when the plant is mechanical and can be damaged by highamplitude oscillation. to be able to introduce feedback gradually and to reduce it rapidly if an oscillation starts.To eliminate the distortions. It matches the50 SZ cable and. A saturation link is commonly placed in front of the integral term (and. the IC gets hot.6. and L) are to be determined (tuned) for a specific plant.6 PID tunable controller 6 .190 Chapter 6. 6. 6. the power supply current increases. 6. flexible wire 1 * oscilloscope + Fig. one should introduce potentiometers eitherin the paths from the compensators to the actuators. While closing control loops over some physical plant.since it is quite common that the sensor output signal polarity is not knownor will be changed during experimentation. The testfixture showninFig.) On' the. the output. makes the input impedance of the probe 5 Id2 which is sufficiently highfor most troubleshootingtasks.at the priceof attenuating the measured voltage 100 times. Such potentiometers greatly improve the troubleshooting options when multiloop systems are tested.6. 6 . and some spare connectors should be The spareconnectors will be used for firmlyattached as showninFig. The third term is a bandlimited differentiation path approximatingDs for o < q. if the plant . and the output signal decreases. the cable at least at one side. to test the signals at various nodes of thecircuit. I. Here.
and with increased slope at higher frequencies. . frequencies over46. and the coefficient dB I. The Bode diagram of a typical plant is convex. the Ds termat tofd4 as shown in Fig. 6. 6.t .the P termdominatesnear fb. P is tuned for the risetime/overshoot tradeoff.and Z. the component Z/s. P can be tuned for the feedback bandwidth. c/2 Fig. D for the phase stability margin (Dterm phase advance partially compensates the of lags the highfrequency poles of P and C z ) . reduction. D for the overshoot Z is increased until it starts affecting the overshoot too much. and the Z/s term at lower frequencies up Thus.typically. monotonic.at higher frequencies. infrontoftheproportionalterm)to dB I I ~ preventwindup(see """""" 0.Chapter 6.PID compensators .39 Bode diagrams for the PID compensator Fig. The pole differentiator is typically chosen to be from 3x2xfi~ to lOx2nfpp~.40 Ranges of dominant terms in. In a system with the plant having a pole at the origin and two or more additional real poles at higherfrequencies. 6. and the compensator q in the transfer function has two real zeros corresponding to these frequencies. When a step testcommand is used for closedloop tuning. The concave response of the PZD compensator needs to be adjusted for the loop Bode diagram to have the desired slope. I I \ . 6. for disturbance rejection at lower frequencies. a lowpass filter C 2 is commonly added to attenuate the sensor noise and to gainstabilize the system at higher frequencies. y : . the distance between the corner frequencies f r p and f i D is 2 octaves or more. Analog Controller Implementation 191 is a double integrator. The componentP dominates at midrange frequencies. 6. ' " x  1: log sc Fig. If the plant response at higher frequencies is too shallow. and the component Ds. also Chapter 13). at lower fkequencies. Typically.40. t loop \ gain 12 dB/OCt single For integrator plant For double integrator plant I I P 'b .38 PID controller block diagram Fig.39 shows Bode diagrams for each path of the compensator C1and the entire compensator. B lo /c .
42. the bandlimiting pole in the differentiator is not shown in the digram). Due to this . (For simplicity.1 1. Analog Controller Implementation The PZD controller for a double integrator plant uses the term Ds to reduce the slope of the loop Bode diagram at the crossover. thefn. 6. 6.41(b) shows the components forming the output of signal the T/Dcompensator at the corner frequency fm. the performance of the controller is not optimal.41(a). 6. 6. Bodethe step responses provide larger feedback. Fig. corner of the2 7 D gain frequency response is sharper than the similar comer inPID the compensator. Several automatic tuning procedures based on tuning the transient response are successfully and widely used.p.43 Bode diagrams for T/D compensator The vector diagram in Fig.192 Chapter 6. and prefilters can reduce the overshoot. 6. Implementation of the transfer function s"' wasalreadydescribedinSection 5. The output signal amplitude is the same as the amplitudesof the components. I. 6. "_ / s d0 \ I 0 Fig. However. 6. the vector diagram at the comer frequencies fp~ is as shown in 3 dB than each of the components. When the plant has also several highfrequency poles or a large n. the zeroslope Ppath can be replaced by the tilted response path Tsln withconstantgainslope2dB/oct. 6.42 TID controller Fig. it is the industry standard for tunable controllers.41 Vector diagrams for corner frequencies for PI0 (a) andTI0 (b)compensators As shown in Fig. With the same average loop gain at higher frequencies.43.where T isascalartunableparameter. lag. Because of this. It typically provides an acceptable transient response without a prefilter or command feedforward.2 TZD compensator In the PID controller.The output is larger by Fig.6. the PZD controller may be augmented by a fourth parallel branchwith a bandlimited double differentiator. 6. Bode diagrams for the T. and D paths' gains are shown in Fig. The controller is easily tuned to provide robust and fairly good performance for a great variety of plants. In fact. The PZD controller is very popular.3 and exemplified in Fig.
Chapter 6. Analog Controller Implementation
193
and due to the slope of the Tterm gain response, the controller provides response.
a better loop
Example 1. Consider PID and TID controllers for a firstorder plant using a sensor 80 H z .Limited by the sensor noise, whose transfer function possessesa triple real pole at z .The Bode diagrams for the loop gain achieved with the crossover frequency fb is = 20H PID andITDcompensatorsareshowninFig.6.44(a).Thecontrollershavethesame coefficient D (i.e., the same gain at higher frequencies and, therefore, nearly the same level of highfrequency noise). At the typically critical frequencies of about halffb, the feedback in the I T D controller is 4dB larger than that in the PID controller (although the TID feedback is still smaller with a Bode step). than the feedback achievable dB
0
(a)
(b)
Fig. 6.44
Bode (a) and Nyquist (b) diagrams for PI0 and T/Dcontrollers
The Lplane openloop diagrams for PID and TID control are shown in Fig. 6.44(b). It is seen that the PID phase stability margin nearfb is too large, 7 0 ° ,and the phase margin is not excessive in the TIDcompensator. PID and TID controllers are both easy to tune when the plant parameters change, e.g. for a temperaturecontrol ofanindustrialfurnace with variablepayload.Whenthe controller is not supposed to be tuned for each individual plant, preference should be given to a highorder compensator witha Bode step.
6.7 Tunable compensator with one variable parameter
6.7.1 Bilinear transfer function Plantcharacteristics may depend on thevaryingenvironment. . Examples are the dependence of an aircraft’s dynamics on altitude, and the dependence of a telecommunicationcable’sattenuation on temperature.Fortheloopresponsetobe  patameter,the optimalforeachoftheintermediatevalues oftheenvironmental compensator must be variable as shown in the block diagram for the resulting adaptive system in Fig. 6.45 (and in Fig. 9.2). We assume we know how the plant response varies withthechangingenvironment(althoughnotprecisely, so thatcancellation of the effects in an openloop manner cannot be done).
194
Chapter 6. Analog Controller Implementation
I
I
'I
Fig. 6.46 Flowchart for transfer function dependent on a parameter
Fig. 6.45 Adaptive system block diagram
It would be convenient to modify the transfer coefficient (or immitance; this term' defines both impedance and admittance) W of a linear system by only changing one scalarparameter w. Here, w standseitherforthevariabletransferfunction of a unilateral link, or for immitance of a variable twopole. The function W(w) is bilinear (a ratio of linear functions) [I] and as such can be expressed as
W ( w )=
WIW(0)+ wW() w1 + w
(6.13)
'
If w representsthevariableimpedance of a twopole, w1 isthedrivingpoint impedance between the terminals to which the twopole w is connected. If w designates the transfer coefficient ofan amplifier, then Uw1 is the feedback path transmission coefficient for this amplifier. The flowchart corresponding to (6.13) is shown in Fig. 6.46. in Fig.6.46should be regulated Theresponses of thevariablecompensator smoothly. For this purpose, not all functions of the form (6.13) will do, For example, if it is desired to gradually change the slopeof a Bode diagram from 6dB to 6dl3, and we correspondingly choose W(0) tobe an. integrator, and W() a differentiator,.as shown in Fig. 6.47, and only use a gain block w in series with the differentiator, then the response will be changed from that of an integrator to that of a differentiator with w changed from 0 to a large value, but, as is seen in Fig. 6.48, the frequency responseof W in the intermediate position possesses a zero, Le., the regulation is not smooth.
0
Ef
v \
\
f log sc
Fig. 6.47 Block diagram of a regulator 6.7.2. Symmetrical regulator,
Fig. 6.48 Frequency responses of regulator
a
Smooth regulation can be obtained with Bode symmetrical regulators. Regulation is called symmetrical with respect to the nominal value w, of the variable parameter w when the maximum relative deflectionsof w from w,, up and down, cause symmetrical
Chapter 6. Analog Controller Implementation
195
(in dB) variations in W as shown in Fig.6.49, i.e., when the regulation has the following property:
(6.14)
Fig. 6.49 Regulation frequency responses of a symmetrical regulator
'
Fig. 6.50 Flowchartfora symmetrical regulator
By substituting this expression into (6.13) we have
W,
=v/Q
(6.15) (6.16)
and
The flowchart for the symmetrical regulator is shown in Fig. 6.50. The gain of the regulator 2010glq = 2O10g~W(wo)~ + 2010g
11
(6.17)
changes gradually with variations in w. When w = w,, the second component is 0. The second component retains the value but changes, the when sign w, is switched from0 to as illustrated in Fig. 6.49. The regulation is exactly symmetrical for these maximum variations of the gain response. It can be shown that in the Taylor expansion of the second component, the firstorder term dominates to the extent that the gain 2010glWI depends onw/w, nearly linearly over the regulation range more than 20 dB. The regulator can be used in the compensator for changing the loop Bode diagram, in accordance with changing system requirements, for solving the tradeoffs between the available disturbance rejection and the output noise. It can be used to compensate the effects of known plant parameter variations in adaptive systems (adaptive systems are studied in Chapter 9). Also, a nonlinear element can be used in place of the variable element w. In this case, a nonlinear dynamic link with desirable properties can be built. Suchnonlinearlinkscanbeemployed incompensatorsforenhancingthesystem performance as will be discussed in Chapters 10, 1 1, and 13.

196
Chapter 6. Analog Controller Implementation
6 . 7 . 3 Hardwareimplementation
A hardware implementation of such a regulator is shown in Fig. 6.51(a). The possible implementations of the block l/Q as lowpass and highpass RC filters are shown in Fig. 6.51(b) and (c).
The regulatorresponseswiththehighpassfilter are showninFig. 6.52, The regulator can beused to vary the compensator response when the plant response varies mostly at higher frequencies. Additional examples are given in Sections 11.7, 13.5.
0
/ i
r, log sc
Fig. 6.52 Frequency responses for the symmetrical regulator
Regulation of frequency responses can be performed also with digital FIR (finite impulseresponse)filters whichcanbemadetoapproximateanydesiredfrequency response.However,usingthefilterforgradualresponsechangesrequireschanging several coefficients which is less convenient.
6.8 Loop response measurements
plant response. The choice between the plant simulations and the actual plant 'measurements might depend on the plant size and accessibility. For example, if the plant is big or expensive (like a spacecraft or heavy machinery) or, if it is very small like a microwave feedback amp1if"ler or a micromachined mechanical device whichis difficult to access, then simulation of the plant should play amajor role in the 'feedback loop design. In this case the plant model needs to, be accurate, and substantial efforts are justified to develop this model.At the beginning of the project, when this model is not yet completed, simpler models mustbe developed for preliminary control design, to be able to answer the questions crucial for the plant design. On thelaststages of the
To design and to veri@ the loop frequency response, one has to calculate or measure the
Chapter 6. Analog Controller Implementation
197
project, when the plant is built, it can be measured and the feedback loop tested.In all cases, the compensator must be tested first, before testing the entire loop. Ontheotherhand, when theplantisinexpensive and easily measured, the compensator (analog or digital) can be designed well directly in the laboratory while working with the real plant, by measuring the loop response and adjusting the elements of analog compensators or coefficients in digital compensators. The loop response can be measured when the feedbackloop is open as in Fig. 6.53, or while the system is in the closedloop configuration. While the loop response is being tested, the loop can be broken, theoretically, at any cross section. In practice, the cross section mustbe carefully chosen such that the signal at this cross section is large enough for convenient measurements but not excessively large to make it’difficult to generate 1 of the compensator,as such a signal. Typically, the loop is broken the after first stageC shown in Fig. 6.53.
Fig. 6.53 Breaking the feedback loop
The choice of the cross section for breaking the loop also depends how on easy isit to simulate the load for the loop output. For conventional control loops with separate compensators this is not an issue, but it can be a difficult problem in some systems, especially at microwave frequencies in or magnetics. Measurements of the openloop response in the openloop configuration are not always convenient or even possible. The feedback loop transfer function measurements can be complicated by parasitic feedback loops from the loop output to the loop input (especially at higher frequencies). Also, some plants cannot used be without feedbackthey’could be unstable, selfdestructive, or dangerous to deal with. The control loops about such plants’ need to be measured in the closedloop configuration  however, perhaps with reduced feedback. Loop response measurements in the closedloop configuration require closedloop the that = system be stable. If it is not, it still might be stable when the feedback is reduced by an extra attenuator, i.e., the system is closed with Fig. 6.54 Loop response smaller feedback. Injection of the measurements with the loop closed signal into the feedback loop for the purposeof the openloop response measurements with the loop kept closed is shown in Fig. 6.54. The loop transfer function is the ratio of the signal at input 2 to the signal at input 1 (the latter is applied to the “reference” input of the analyzer). A practicalcircuitforthesignalinjectionis shownin Fig. 6.55. The variable resistor allows reducing the loop gain, and the RC lead is used to increase the test signal level at higherfrequencies  as will be discussed below. The capacitance value C = 2.lO’/j, placesthezero of thisleadatapproximately O.OSfb, andthepole at

198 approximately 4 f .
Chapter 6. Analog Controller implementation
20k 2k
t
4
2k
input 1
+
input 2
I
Fig. 6.55 Summer for injecting testsignals
into the,closedloop
an analog If compensator already includes an inverting opamp, the signal injectioncan be implementedwithafew passive elements as shown in Fig. 6.56, using the opamp as a summing amplifier. Signal analyzers typically use two types oftestsignals:sweptfrequencysinusoidal and pseudorandom with subsequent fast F,g, Signal injection using an existing inverting amplifier transform (m)and averaging* already Both methods are quite suitablefor the loop response measurements. At higher frequencies, sinusoidal excitation is easy to implement andis appropriate. At low frequencies, FFT might provide better zwuracy or shorter time of measurements. Still, lowfrequency response measurements often take, longerthandesired,especially when thetestisperformedinthefield,orsome equipment normally in operation needs tobe shut down to perform the test. Part of the problem is that when the testsignal is employed with constant amplitude of sinusoidal signals at all frequencies, or with constant spectral density of pseudorandom signals, the dynamic rangeof the signals at the output of the loopbecomes very large, since the loop gain is many tenths of dB inthefunctionalfeedbackbandand 20 dBathigher frequenciesstill of interestforthestability margintest.Then,thesignal,amplitude chosen so as not to overload the actuator at lower frequencies becomes too small at higherfrequencies, and data recovery from the noise requires multiple runs and averaging. An extra linear linkwith the gain increasing with the frequency, such as that shown in Fig. 6.57, can be introduced at the signal analyzer generator output to reduce the dynamic rangeof the loop output signal and correspondingly to reduce the required time for accurate measurements. In the absence of a signal analyzer, the loop gain response can be measured pointbypoint, at discrete frequencies, with a signal generator and a twoinput oscilloscope. The input signal of the loop is applied to one of the oscilloscopeYinputs, and the loop output is applied to the other Yinput. The phase difference between the signals can be seen on the scope screen with sufficient accuracy.
Chapter 6. Analog Controller Implementation
199
Fig. 6.57 (a) Gain response corrector to be placed in cascade with the signal source to improve the measured signaltonoise ratio, and (b) the corrector gainfrequency response
With a sweep generator and a scope, an addition of the circuitryshown in Fig. 6.58 can mechanize the loop response measurements. Here, two amplifiers (optional) increase the levels of the signals at the input and output of the feedback loop. The amplifier gain can be made to increase with frequency in order to reduce the dynamic range of the signals.
V, , Imp gain
Y, ,phase lag X , frequency
Oscilloscope
Fig. 6.58 Set for loop response measurements
The two singlesupply comparators convert I the sinusoidal signals to rectangular single1 ' 1 I 1 t polarity signals uz and u1 as illustrated in Fig. 6.59. The Exclusive OR serves as phase 2' detector. Its output signalu3 is nonzero onlywhen t u 3 onlyone of theinputsignalsisnonzeroas shown u4fl __ 6.59. Fig. in The average value u4 of signal the u3 t is proportionaltothephaselag between u2 and Fig. 659 Timediagrams for u1. The lowpass filter producing u4 can be a the phase detector signals Butterworth active RC filter. When the plant transfer function is relatively simple, the plant can be identified by its response to the step function applied to its input. This method is often employed for tuning PZD controllers for chemical processes is as discussed below. While the process is goingon openloop, the output is made steady by adjusting the command manualy. Then, a small step is applied directly to the plant input as shown inFig.6.6Q(a). This stepproduces anincrementintheplant'souput,(controlled variable) as shown in Fig. 6.60(b). From the response, the delay time td and the rate r are determined. The parameters in the . PZD controller written in the form
'
I
n

~
"
200
Chapter 6. Analog Controller Implementation
a[1 + l/(bs)]( 1 + cs)/( 1 + cs/4 are set to be: a = l/(rtd),b = 5td, c = td2, d = 4, and then furtheradjustedexperimentallyfor goodclosedloopperformance,Thisopenloop ZieglerNichols tuning procedure is not universal and does not necessarily lead to good tuning of PID controllers for all possible plants.
increment command
. I ,
C
b o m m o m m m m o o m
output
m a
0
step
time
t d
(a) Fig. 6.60 Plant transient response measurements, (a) block diagram and (b) transient response
A more general approach is to createa plant model with a transfer function having a couple of real polesand zeros, and adjust the values of the model poles and zeros w i t h Matlab to obtain the same transient response as thatof the real plant. Then, using this plant model,a compensator can be designed that provides the desired loop response.
6.9 Problems
(a) Design an amplifier with gain coefficient 25 using an opamp in the inverting configuration.Theamplifierwillbeemployedinthecompensatorforacontrol system with 1 kHz feedback bandwidth. Chooseh, and choose resistors such that the power consumption by the feedback resistor from the 12 V power supply does not exceed3 mW. (b) Do the same for the gain coefficient  100, &, = 100kHz, power not exceeding 0.2 mW. (a) Discuss the accuracy of implementation of a transfer function as an opamp with RC impedances. (b) Define the required tolerances of resistors and capacitors. Draw a schematic diagram for an opamp circuit implementing the function: (a) ax by c; (b) +ax+ by+ c; (c) ax+ by c; (d) ax + bJydt  C , (e) ax + bdy/dt c; (f) &ax  by  c)dt.

Draw a schematic 'diagram and specify the values of the circuit elements (use a lOOMz feedback resistor) for a practical opamp circuit implementing the function: (a) 0.5~ 0 . 2 ~3; (b) +2x + 8y + 0.2; (c) 3x+ 6 y + 1; (d) 4x + 4jydt + 0.3; (e) 2.72~ + 0.5dyldt + 0.1; (f) I(~x + 0 . 4~ 2 )dt.


Analog Controller Chapter 6. Implementation
201
Using 20 kQ feedback resistors in opamp summers, draw a schematic diagram for an analog implementation of decoupling matrix:
= 0.5038~'+ 0.0443~'+ 0.07722' y = 0.0235~'+ 0.4796~' 0.021 52' z =: 0.0094~' 0.0243~'+ 0.5291Z '
X

= 0 . 2 , ' + 0.05~' 0.1 Z ' y = 0.1 X' + 0 . 6 ~+ ' 0.01Z' z = 0.001X'  0.002~'  0.72'
X
~ 0.32 3x + 0 . 4 + 0.3~  2.1y + 0.22 O.O4X+ O.ly+ 1 . 9 ~
2 x + O.ly+ 0.lz 0 . 1 ~  3 . 1 ~0.12 + +0.4+ ~ 1.9~ 0.04~
x+y2 x+ y+ z xy+z
+ y + 0.5jydt + z x+y+z x  O.ljxdt+ y z
X
Choose the type of opamp for applications requiring low noise insthe bandwidth: (a) 10Hz; (b) 100 Hz; (c) 1 MHz. Choose the type of opamp to use when the output power needs to be up to 3 W and the frequency band of interest i s : (a) 0 to 100 kHz;(b) 0 to 1 MHz; (c) 100 Hz to 1 kHz; (d) 1 kHz to 10 kHz. In the compensators using the inverting opamp configurations in Fig. 6.10(a),(c), the maximum gain (at lower frequencies for the lag and at higher frequencies for the lead) must be 40 dB. Choose f f 2 = 1 MQ andfind the remaining element values for: (a) polefp= 5 Hz, zero fi = 10 Hz; (b) pole G= 12 Hz, zerofz = 20 Hz; (c) pole = 300 Hz, zero fz = 1000 Hz; (d) pole6= 15 Hz, zero fi = 40 Hz; (e) polefp= 27.2 Hz, zerofi = 120 Hz; (f) polefp= 800 Hz, zerofz = 3000 Hz. Find the elements analytically (start with resistors), or with MATLAB, or with the chart in Fig. 6.23. Use a noninverting opamp with RC feedback according to Fig. implement a lead link, assuming R 2 = 500 k C 2 and de gain 20 dB with (a) fz = 100 Hz, fp = 300 Hz; (b) fi = 50 HZ, fp = 150 HZ;
6.9(b) to
fp = 600 HZ.h = 500 HZ. to 11 Find a transfer function with three poles and two zeros approximating the gain response slope4 dB/oct over the range 3 to 30 Hz.2. and plot the gain response. fz = 1000 Hz. highfrequency asymptotic slope 8 1 (c) Design a Nyquiststable loop Bode diagram. Design a compensator with this transfer function using an opamp in the inverting configuration. (f) fz = 60 HZ. Make simulations with SPICE. i= 10 HZ. using the method illustrated in Fig. Use an initially asymptotic Bode diagram. fp = 100 HZ. stability margin10 dB. The gain at dc must be 26 dB.12(a) to produce a broad notch. and plot the gain response. (d) Butterworth 5thorder with cutoff at 30 Hz. Find filter elements. 6. 12 Use the opamp with RC feedback shownin Fig. 6. (c) Chebyshev5thorderwith0. (e) fz = 27.3. 15 Someantialiasingfilter ICs passthedcinputdirectlytotheoutput. (b) 6 = 150 HZ. (d) Butterworth 5thorder. Use the FILTER1 program from BurrBrown. then (a) simulate the response in MATLAB or SPICE and(b)adjustthepositions of polesandzeros(ifusingMATLAB)andthen calculate the element values.without contaminating the dc signal by the dc drift oftheemployedopamp. 14 Using the FILTER2 program (available from BurrBrown) to design the following multiple feedback filters: (a) a single secondorder link with 200 Hz cutoff frequency 4 =and 0. using a real pole and a real zero.4dBpeaktopeakripples in thepassbandand cutoff at 10 kHz.9(a) implement a lag link.6. and a pair of complex zeros.. simulate the filter performance. with the lower and the upper Bode steps at 10 dB and 15 dB levels respectively.7. (c) Chebyshev 5thorder with 0. (b) Butterworth 3rdorder with cutoff at 1 kHz.2 Hz. applications are such filters especially suitable? For what in Fig.4 dB peaktopeak ripples in the passband. 13 UsetheFILTER1program(availablefromBurrBrown)todesignthefollowing SallenKey filters: (a) a lowpass secondorder link with cutoff frequency of 200 Hz and 4 = 0. simulate the filter performance. Find sensitivities to the filter elements. and the highfrequency asymptote must be6dB/oct.14 16 (a) Implement a Bode step with the SallenKey lowpass filter shown with the data in Example 1 in Section 6. using SallenKey filters. assuming R 1 = 10 and dc gain 30 dB with (a) 16 = 300 Hz. (b) Do it for the loop with fb = 500 Hz. gain dB/oct.1 9(a).202 Analog Controller Chapter 6. main Bode diagram slope 10 dB/oct. (b) Butterworth 3rdorder. (d) & = 150 Hz.fp = 400 HZ. Find filter elements. or directly adjust the values of the circuit elements (if using SPICE). fp= 100 Hz. 6. (c) fp = 3 HZ.or useMATLAB simulation for second . Implementation (c) fz = 10 HZ. a noninverting opamp with 10 Use RC feedback according to Fig.
Becauseofthis. 6.1 2.23. find: 21 Using the plot (a) where is the pole of impedance of the parallel connection of 1 mF and k S 2 . approximately. at 800 Hz. (d) what capacitor needs to be connected in series with a50 k12 resistor to make a zero at800 Hz. draw the frequency response of the magnitude of the impedanceofthe RC circuit in Cauerform in Fig.withcapacitances (starting from input) 0. andR1= 50125 = 2 ki2. or SIMULINK to solve Problem 24.. i. 25 Use MATLAB Answers to selected problems 1 (a)At 1 kHztheclosedloopgaincoefficient is 25.whenthefilter is used to implement a Bode step. Therefore.the sensitivity of the modulus of the impedance to a resistance or a capacitance within . 9 0 ' 1 interval. 10%. 1 (b) whereis the zero of impedance of the series connection 10 of nF and 10 k 1 2 . 6. i. by 5%. 20 (a) Using the plot in Fig. the transfer function pole is at 200 Hz and the zero.. 6. the openloop gain coefficient must be 750.andthefeedbackmustbe 30 dB. in 24 Make SPlCE simulation for the circuit (a) Fig. Discuss how to make these simulations with MATLAB and SIMULINK.19. (c) make the plot with MATLAB after deriving the equations describing the circuit. with saturation type of characteristic) using the compensators in Fig.12(b). 22 Choose ICs for the circuit in Fig. The angleofan RC impedance is within the [0.20(b). is R = 1/(Cfs). > 750x 1 kHz = 750 kHz. 10 nF and resistances 100 ki2.Analog Controller Chapter 6. (c) Fig. 6.58. 2 (a)Whenimpedanceofaseriesfeedbackbranchchangesby +5%. I 8 Prove that the equivalent resistance of a charge pump 19 Use SPICE to simulate the circuits in Figs.6.13 . (b) Fig. (c) whatis the impedance of the series connection ofkG? 20 and 4 pF at 300 Hz.18 and 6.e. (e) what resistor needs to be connected in parallel with a 2pF capacitor to move the pole from 0 Hz to 30 Hz. A.23. 6. (b) make the plot with SPICE. 203 17 Use the FILTER1 program to study sensitivity of SallenKey filters. then build and test the device. 6. in Fig.1 2(a). (f) R and C for the series RC twopole connected in parallel to 100 kS2 feedback resistor of an opamp in inverting configuration. Show that plent of marginexistswhenusing1%resistortolerances. andofthe parallel branch. 5 ki2 (you can draw the response directly on a copy of the chart). 6. the gain coefficient increases by.1 mF. 6. 23 Simulate a control loop for pointing a spacecraft with flexible solar panels using thrusters (in pulsewidth modulation mode. R2 > 50 ki2. Implementation order links with appropriate damping.e.
wherefrom C = 400 pF. the parallel connectionof resistances R 1 and R3 is R 1 # R3 = &/31 = 16.62 Schematic diagram fora decoupling matrix implementation 8 (a) Since fi > fp. R 1 = Rd11 =J 45. c . the gain' coefficient is 100fdfi = 50. W Z . 3 (d) The summer is shown in Fig. Le. 6.. From here. Thus. the gain coefficient is 3 times larger (because the ratio of the pole to the zero frequencies 3). The capacitance can be found from the equality l/(oC) = R 3 at the frequency of the poleof the compensator transfer function which is thefrequency ofthezerooftheshuntingimpedance in thefeedbackpath.Analog Controller Implementation the circuit is less than 1. The capacitanceC = 16 nF is found from the condition 1/(27cfic) = &. 6. 0~ Le. and R3 = 477 The capacitanceisfoundfromtheconditionthatatthepolefrequency R 3+&= 1/(2nfPC). Then. 9.5 k s 1 .61.48.this is a lag link and the schematic in Fig. C =20 nF..48 = (R3 # R2 + RI)/RI. frequency.10(c) can be used. . 10 (a)Atzerofrequency. required gain coefficient10 at zero frequency. From here.1 /?I/(16.6fdfz = 9.1 + RI) =: 25 WZ (wearedesigningaforwardpathcompensator so there is noneedfor extreme accuracy of calculations). From this.61 Implementation of function ax + b x [integral of ( y ) ] c 5  I Fig. At infinite the gain coefficient at zero frequency must be 100. R3 # R2 = 848 kQ. At infinite frequency. it is 30.. .204 Chapter 6. C = 2 1 ~ 1 F. 9 (a) From the. the gain coefficient is 31. and 1 = 10 W Z .6 =: (R2+ R#R1 wherefrom R 2 = 306 W Z .1 kQ and R 3 = 16. Y Fig. 6.thegaincoefficient is 31. R 3 # RZ = 500 W Z . R R 3 = 1 Ma. At infinite frequency. is Le. 6.
7. and the capacitors. mechanical. 1 .preferably.M4. 7. by rigid bodies specified by their massesMz.2.4. thermal. the inductors are replaced by springs with spring coefficients k l . The feedback system equations are analogous to the equations describing parallel connection of two links and parallel connection of two twopoles. Control engineers deal with systems that are part electrical and part mechanical..10. motors. These analogies allow the theory developed for twopole connection to be applied to feedback systems. k3.4. Example 1.3. to simplify system modeling and to make it structural. For a shortened control course. It is seen that the topology of the diagram is conveniently preserved. collocated and noncollocated control. like drivers.1 1 can be omitted. to increase the available feedback). In thelattercase.This may be done either by theuseror. and gears. and the effect of feedback on the output noise. and 7.5. 7.generally. the conversion allows the use of programs which were initially supposed to be used for the analysis of electrical circuits. 7. 7. Pigs.theeffectof feedback on impedance are considered. etc. the actuator output impedance needs to have a specific value.i. Converting such a system to an equivalent system containing only one kind of variable and described by onlyonekindofphysicallawcanfacilitateboththe preliminary backoftheenvelope analysis and the writeup of the computer input file. The chapter ends with a brief demonstration of the specifics of linear timevariable systems.e.2. 205 . sensor noise. and current I to force F.5.1 Mathematical analogies 7 . by a computer.mostuseful are thosewhichpreservepower.they convert the product (voltagex current) into the product (velocity X force). likeSPICE. 1 Eleetro~mechanical analogies The first step in simulating a system behavior on a computer is to generate the equations describingthesystem. 7. series.Chapter 7 I LINEAR LINKS AND SYSTEM SIMULATION Two approaches are presented for modeling systems composed of electrical.1 and 7. as w be explored in Chapters 10 and12. and deriving mathematical equations for the system. There existseveralmathematicalanalogies betweenmechanicalandelectrical systems. Equivalent block diagrams are developed for chain connection of twoports. Sections 7. as universal control analysis and simulation tools. Also.2 show an example of the application of this analogy. including parallel.the.2.8.27. thermal and hydraulic elements: describing the system elements and the topology of their connections. Wewill most o€ten use the vo/ta$efove/ocity analogyrelatingvoltage U tovelocity V. and compound feedback. It is emphasized that to reduce the plant uncertainty (and therefore.Amongthem. The use of localfeedbackloopstomodifythisimpedanceand.2. Several issues are considered important for feedback control: flexible structures. 7.parametersdescribingthesystem’selements andthe topology of their connections constitute the simulation program input file.
V4 = F ~ / ( s M ~ ) F2 = V9M2. ** The term “mechanical impedance” is sometimes used to mean the mobility. Linear Links and System Simulation Fig. capacitance into mass. . 7.The * To accentuate the analogy. 7.3(a). inductance into the inverse of the stiffness coeffkient. The strut consists of a spring. The equality of the sum of currents to zero at a node correspondsto the equality of the sum of forces tozero at arigidbody(takingintoaccountD’Alembert’sforce). Kirchhoff ’s equations correspondingly are where For bodies 2 and 4.1 summarizes the analogy for translational and rotational motions. Newton’s equations are FI where + F2 + F3 = : 0. Since mobility is the ratio of relative velocity to force. Generally. the equalityof the sum of the voltages (relative potentials) about a contour to zero reflects the equality to zero of the sum of relative velocities about the contour. Table 7. a dashpot (a device providing viscous friction). and a linear motor (voice coil) connected in parallel and driven by an electrical amplifier in accordancewiththeinformationobtainedfromthe force sensor(loadcell).1 Electricalcircuit schematic diagram Fig. electrical impedance converts into mechanical mobility.206 Chapter 7. and not similar to resistors as is conventional in drawing mechanical diagrams. and sometimes the inverse of the mobility. Consider the active suspension strut diagrammed in Fig. Resistance converts into the inverse of the viscous damping coefficient. the mobility of a difficult to move andlor heavy load by a given force is small**. 7. the springs are shown similar to inductors. Example 2. Zero voltageoftheelectrical“ground”mostcommonlytranslates to zerovelocityofthe inertial reference. Similarly.2 Mechanicalcircuit schematic diagram* For nodes 2 and 4.
currenttoforce) analogy described above is . The voltagetovelocity (i.XF = 0 at a rigid body. 7. the notion of mobility is particularly applicable to the designof suspension and vibration isolation systems. the position and force the or the angle and the torque. using the analogy of a mechanicaljoint to ajoint of cascaded electrical twoports is of great help. Fig. I mobility Z = V/F I mobility 2 = ah. At the same time.XI = 0 at a rigid body. Fig. in the great majority of practical cases. XAV= 0 XAs1 = 0 caDacitance C I mass M I moment of inertiaJ I l/(stiffness coefficientk) l/(angular stiffness) inductanceL l/(angular viscous damping l/(viscous resistance R coefficient 23) coefficient) .e. 7. For these systems. using the electromechanical analogy gets more difficult (although it is still possible).ET= 0 about a contour. For complicated mechanical systems. specialized simulation programs should be used (like ADAMS 8 or SD FAST 8). The mobility must be large at frequencies where the vibration should not pass through. with the joints characterized each by only two variables.1 Voltagetovelocityanalogy 207 I c electrical I translational I rotational I relative velocity V relative angular velocity s1 voltage U (or V) current I force F torque z power P = UI power P = VF power P = lczz I impedance 2 = U . the mobility should be low enough at lower frequencies in order for the motion of the base to be conveyed to the body of interest. Linear Links System andSimulation Table 7. Thus.Chapter 7. the mobility is defined as the ratio of the difference in velocities at its ends to the force. bodies. For the suspension strut. control engineers only deal with slides or pin joints. Le..3 (a) Activesuspensionstrut and (b) its mobility frequency response For the analysis of mechanical systems with many degrees of freedom at thejoints such as balljoints.3(b) exemplifies the modulus of the mobility with and without introducing feedbackto make the isolation better..However.XU = 0 along a contour through along a contour through a a sequenceof connected sequence connected bodies. vibrations of the base should be prevented from shaking the payload. I at a node.
Inpiezoelectricandelectrostatictransducers micromachined devices). Table 7 .055Btu. and the analogy shown in Table 7. E M = 0 at arigidbody.ZI = 0 about a contour. Since radiation increases as the fourth power of the temperature difference between the bodies (or between the . 3 Voltagetotemperaturedifference analogy dectrical voltage (potential difference) U (or V) current Z power P = UZ impedance 2 = U/I at a node.Zg = 0 (not counting heat accumulation inCT) thermal capacitanceCT which is the product of the specific heat c and the mass M thermal resistanceR T = dT/dg Heat also can be transferred by radiation and convection.= 0 l/(stiffness coefficient k) mass M B I rotational torque 't angular velocityQ angle 0 power P = QT inverse of mobility Y = z XZ along a contour through a sequence of connected bodies. Zr = 0 l/(angular stiffness) moment of inertiaJ angular viscous damping viscous dam coefficient 7. The heat flow is measured in Watts or Btdsec.208 Chapter 7. XI.2 Electrical analogy to heat transfer The electrical analogy to conductive heat transferis shown in Table 7. 1kW corresponding to 1.Linear Links and System Simulation especially convenient when the system includes electromagnetic transducers like electrical motors and solenoids.2 may appear more convenient.1. ZAV= 0 at a rigid body. velocity V displacement x power P = VI.Z Z=0 capacitance C resistance R = dU/dZ thermal temperature difference AT heat flowg ATg thermal impedance & = AT/g at a body. Table 7.2 Voltagetoforceanalogy electrical voltage U (or V) current I charge q power P = U Z impedance 2 = UL at a node.3. The Fourier lawfor heat flow is analogous to Ohm's law. the applied voltage producesforce. capacitance C inductance L resistance R translational force I. inverse of mobilityY = F/Z along a contour through a sequence of connected bodies. where current creates a magnetic field which in turn (the latterareusedin produces force.
the area.the model also incorporated the driver and compensator. and Tl representabsolutetemperatures of the secondary mirror. the reservoir of volume v1 is filled with a liquid keptat pressure hl.5.1. R2 to the liquid flow (we consider the volume of the liquid in the pipe insignificant. The sum of the liquid flow at each node is 0. 232 and 231 representheatersattachedtothesecondaryandprimarymirrors. The pipes connecting the reservoir to the cylinder with current volume:v2 and pressure h2 present resistances R1. A hydraulic system is exemplified in Fig. 7. Here. The capacitances are the products of the bodies’ specific heats and masses.it can be represented by a nonlinear current source. 7. T. With this approach. G. the case.507 X lo’ X T.4 Heat transfer analysis for a spacecraft photocamera This circuit was simulated using SPICE . = 0. 7. or else the system must be considered as a system w i t h distributed parameters). and the primary mirror.5 X lo’ T X T:.4 by an equivalent electrical circuit employed to simulate the temperature control of the Cassini spacecraft narrow view photocamera.. 7. T ‘2 Fig.1 X lo’ T X T 1 4 . and the air flow. The plunger stem moves a ZL. GI = 0. The analogy is exemplified in Fig.Chapter 7. For heat transferred via convection.Thenonlinearsourcesrepresenttheheat radiation into fiee space fromthe mirrors and the case: G2 = 0. or switching it on and off. and the resistors characterize the thermoconductivitiesbetweenthebodies. 7. the frequenky and time responses can be plotted without derivationof the plant equations.3 Hydraulic systems Example 1.Thetemperatures T2. The valve controls the liquid flow by introducing extra resistance in the pipe. Linear Links and System Simulation 209 body and theenvironment).. The valve is operated by an electromagnet fed from the driver amplifier of the controller. respectively.5 Hydraulic system example . the equivalent thermal resistance depends on many parameters including the geometry. mechanical load with mobility U Hydraulic variables I i Mechanical variables Fig. Here.
and the resistance is Example 2. Still. and. An electrical analog to a hydraulic system is described in Table 7.4. Table 7. The inputoutput characteristic for the amplifier shown in Fig. .7). and dead zone is reduced many times (recall Section 1. XI= 0 capacitance C resistance R = dU/dI ~ ~~ hydraulic pressure differenceAh flow of liquid. particularly. The hydraulic and mechanical diagrams can be converted to their electrical equivalents and connected by an ideal transformer that preserves the power and provides appropriate correspondencebetweenthehydraulicandthemechanical variables. q is proportionalto h. in most applications. The valve directs the liquid under pressure to the appropriate side of the plunger. The dead zone should be sufficiently large to reliably the stop power waste when the input signal is zero. each passing a single polarity signal (which is the class B mode of operation). 7. i. In this case the dependence of q on h is nonlinear. However. By application of large feedback (not shown in Fig.6) the inputoutput characteristic is made much closer to linear. this system can be simulated with a program for electrical circuits analysis.q power P = Ahq impedance 2 = Ap/Ah at a node. and mechanical variables.6 demonstrates the analogyof (a) an electrical amplifier to (b) a hydraulic amplifier.6(c) is not quite linear. Thesummercombinestheoutputcurrents of the branches. The hydraulic amplifier input is the position of the spool valve.the flow of liquid is turbulent. electrical. A small dead zone in input voltage prevents the waste of power supply current in the quiescent mode. The power is the same to the left and to the right sides of the vertical dashed line. the zone must be small enough for the inputoutput characteristic to be close to linear. 7. volume per sec.4 Voltagetodifferenceinpressureanalogy electrical voltage (potential difference) U current I power P = UI impedance 2 = U/I at a node. 7. The electrical amplifier contains two parallel branches.e.X q = 0 volume v resistance R = dWdq is laminar. Then. andthe Whentheflowofliquidinapipe resistance R = dwdq does not depend onh.210 Chapter 7. thus providing output force at the plunger stem. Linear Links System and Simulation The dashed lines separate the areas on the drawing for hydraulic. contains a dead zone. Fig..when the inputsignalis 0.
It facilitates redesigning the system. the effects of the following link on the output variable of the previous link. since. and (e) the electrical amplifier’s inputoutput characteristic Thisprovision is especiallydifficulttoimplementinhydraulicamplifiers.dependingontheapplication)withelectrical output. Thus. For example. 7. the accuracyrequirementsforthelinks in theforwardpathcanbemuch relaxed from the required accuracy for the summer and feedback path links). Since the specifications for each of the subsystems are tailored suchas to minimizetherequirements. 7. The dynamics and linearity can be improvedby using local feedback about the actuator. The effects of loading are conveniently described using the notion of impedance. 2 . the compensators themselves are built from simpler links cascaded or connected in parallel. a small opening is insufficient to supply the liquid to the output cylinder.Structuraldesign simplifies the analysis and design.6 (a) Electrical class 6 amplifier. a power electrical amplifier (motor driver).e. each subsystem can be refined and redesigned independently of the others. 1 Structural design With structural design. Linear Links and System Simulation high press line output 21 1 input: position output: force (a) (b) (c) Fig. the inputoutput characteristic is steep for slow motion. we have to include into consideration the effects of loading. and an electrical motor moving the spool valve shaft. but if the output motion is fast. and facilitates having several people work together on the same project. The interface between the subsystems should be as simple as possible. (b) spool valve hydraulic mechanical amplifier. and theirinput is an analog or digital electrical signal.. a compensator. Between cascaded blocks in the block diagrams. It simplifies verifying system performance.The inputoutput characteristic of the spool valve amplifier is different for different speeds of the action. and troubleshooting. the interface is the simplest: a single variable serves as the output of the preceding link and the input of the following links. but the gain decreases for faster motions. These variations in the actuator nonlinearity and dynamics limit the available accuracy of a control system using such an actuator. However. i. When the output motion is slow. The subsystems are composed of smaller subsystems. etc. Such integrated actuators are commercially available. as long as the requirements for the subsystem are defined. even a small opening suffices to provide nearly full available pressure on the plunger.2 Junctions of unilateral links 7 . This feedback can be implemented using a pressure sensor(orapositionsensor.Chapter 7.itpreventsthesubsystems’overdesign(for example. making design tradeoffs. the system is composed of subsystems which are relatively large functionalpartswithrathersimpleinterconnectionsbetweenthem. . in most physical links.
or emf. when the source impedance is very large. the source is called source of current I or force P. Linear Links and System Simulation 7 . 7. Here.212 Chapter 7.8 Variables at the links' joint For mechanical systems.Thisisprobablysobecauseatamechanical junction. andFB is the brake force. the output voltage (or velocity) does not depend on the value of the load impedance and therefore is called source of voltage U (or velocity V). 7. 3 angles and 3 torques need to be considered at a ball joint.Thenotion of mobility facilitates the analysis of such systems. The voltage at the junction is and the current is E b Driver Fig.7. . is the free (unloaded) velocity.8 shows a preceding link (twoport) loaded at the input impedance of the following link. Nevertheless. 2 Junction variables Any electrical or mechanical source can be equivalently represented in Thevenin's and Norton's forms shown in Fig. E is the opencircuit voltage. the most common mechanical junctions are the simple ones. 7. mechanical driving point impedance (or mobility) is usedlessfrequently. 2 . When the internal source impedance 2. These junctions are mathematicallysimilartoconnections of electricaltwonodeports. as mentioned before. IS is the shortcircuit current. 7. in mechanical systems design. For example. there could be more than two variables to deal with. like a slide or a pin with one position (or rotation) variable and one force (or torque) variable. Fig.3 Thevenin and Norton representations of (a) electrical and (b) mechanical signal sources(for the voltagetovelocity analogy) While the notionof driving point impedance is universally employed in the design of electrical circuits. Fig. V. is small.
= Q. 3 Loading diagram Thevoltageacrosstheterminalsofthesignalsourcewithresistiveloadshownin Fig. the angular velocity is Q=QRsz. The smaller the resistance. the effect of loading can be neglected and U =: E.e. 7. IS 4 a b ’ += f iIi 0 voltage. 7. i. and also when RL is very small when RL is very large and.RsZ. therefore. Fig. Resistance Rs determines the slope of the loading lineZ = ( E ./Rs. similarly.. The output power UZ varies with the load resistance. the current so that the output voltageis small. 7.6). u 0 ! i e 0 k (a) (4 Fig. . It becomes zero is small.6) where is the free run (no load) relative angular velocity.(2) inductive motor. the output impedance of an operational amplifier is typically much lower than the load impedance due to the application of voltage feedback.Chapter 7.9(a) is U =I E . When z equals the brake forque (or sfall forque)z. the product of the angular velocity and the torque. For a rotary mechanical actuator (motor).(4) motor with velocity feedback The loading diagramshown in Fig. (7.9(b) reflects equation (7.U)/Rs.7) (7. For instance.9(b). 7. 2 . the angular velocity becomes zero. 7. is defined similarlyto the power in Fig. the steeper the line.The power. 7 .9(c) shows the loading diagram(1) for a permanent magnet motor described by linear equations (we do not call the motor “linear” since the term linear motor is (2) exemplifies a nonlinear loading reserved for translational motion motors). Linear Links and System Simulation 213 When the driver’s output impedance (mobility) 2 s is much lower than the input impedance (mobility) ZL of the following link. . The slope of this curve defines the differential output resistance of the motor. The curve diagram for a motor with a flux winding. (3) motor with torque feedback.9 Schematic diagram (a) and loading diagrams for (b) an electrical linear signal source and (c) for motors: (1) permanent magnet motor.
theoutputpowercannotbeincreased. the output mobility of the motor is very low. and its implications will be discussed in the next section. Linear Links and System Simulation The loading curve can be changed by application of feedback about the motor although. and the plant transfer function is simply 1.Largetorquefeedback maintains the torque constant as shownby the curve(3). the actuator can be viewed as a force source.3 Effect of the plantand actuator impedances on the plant transfer function uncertainty Plant parameter uncertainty affects the accuracy of the closedloop transfer function in two ways. Consider an actuator driving a plant whose mass M is uncertain. does not depend on the load. the plant velocity equals the actuator velocity. Example 1.. the plant acceleration FIM and the plant velocity are uncertain due to the uncertainty of M. First. the plant is completely unpredictable (as in the extreme case described in Alice in Wunderland ". The output mobility of the motor with torque feedback is large as can be seen from the slope of the loading line. . In contrast. within the range of the motor torque and velocity capabilities. In the hypothetical case that the plant is perfectly known (the socalled full state feedback described in Chapter S ) . . Therefore. Let's consider two extreme cases of the output mobility of the actuator shown in Fig. The control system output variable (controlled variable) is the plant velocity. no feedback control can be implemented.. infinite feedback is available. When this mobility is much larger than the load mobility. 7. on the other hand.as will be shown in Chapters 5 and 9. 7. Output impedance (mobility) is an important parameter of drivers and motors. the actuator can be viewed as a velocity source. directly: the variations in the closedloop transfer function are feedback times smaller than the plant transfer function variation. The loading line for a motor with velocity feedback is shown by the curve (4). Second. the larger is the feedback that can be implemented.10 (a) and (b). an actuator with an appropriate output mobility must be used. When. The torque does not depend on the angular velocity and. the plant transfer function uncertainty depends to a large extent on the value of the actuator output mobility. certainly.214 Chapter 7. indirectly: the smaller the plant parameter variations. The actuator output mobility can be modifiedby application of local feedback about the actuator. respectively. when the actuator is a velocity source. As will be shown in the examples below. you can balance an eel on the end of your nose"). + I=' A "b 1/M b Integrator (a) (b) Fig. '7. therefore. and when the actuator's mobility is much smaller than the load mobility. Reducing the plant transfer function uncertainty is an issue of high priority in feedback system design.10 Actuators with (a) constant force and (b) constant velocity driving a rigid body plant When the actuator is a force source.
) The actuator is placed between the 'second body and the spring. and (c) its modification Notice that in the electrical circuit diagram. Linear Links System andSimulation 215 Example 2 . The output impedance (mobility)is defined by the t y p u the sensor used. 7. (The rotational version of this system is the problem of pointingacameramountedontheendof a flexible boom attached to a spacecraft.Chapter 7. the order of the elements connected in series can be changed.11 (a) Actuator and mechanical plant.andalinearcombination of their readingsisfedback. and the position of the second body is controlled. . i. is stabilizedandmadeequal to thecommand or reference.4 Effect of feedback on the impedance (mobility) 7. If force feedbackis employed.e.thefeedbackiscalled compound. and damping can be introduced which makes smoother the interaction between various feedback loops. a force source is obtained. if this is a problem. Hence. and this force does not depend on the spring parameters. the voltage is stabilized.. a velocity source is obtained.7. What. the output mobility of the control system at the joint of the actuator and the load can be made as desired. Conversion from a mechanical to the equivalent electrical circuit might therefore simplify the analy Whilechoosingtheactuatoroutputimpedance.thefeedbacksystemoutputrepresentsacurrentsource.(b) its equivalent electrical circuit diagram. Two sensorsare employed. If the actuator is a force source.1 Large feedback with velocity and force sensors Examples already considered showed that feedback has profound a effect on impedance. The outputs of the sensors are combined and fed back.Consider the plant depicted in Fig. its output impedance is very large. Similar modifications of mechanical diagrams are not as evident. one measuring the force and the other measuring the velocity. If velocity feedback is employed. i. is measured bv the sensor. .attentionalsomustbepaidto reduction of disturbance transmission from the base to the object of control. .is floafhg (not connected to the ground). f (a) (b) (c) Fig. If both force andvelocitysensorsareused. 7. If a current sensoris employed. 7.12.e.theoutput impedance is high. With compound feedback.. A feedbackloopabout an actuatorisdepicted in Fig.notexactly 0. a velocity source actuator results in a profound resonance on the control loop response. the suspension resonance does not appear on the plant transfer function response: the second body only sees the force F. Compound feeedbackis employed in most biological andmany robotic motion control systems. Diagram (b) can be transformed into the form (c) where of the one terminals of the source E is grounded and the capacitor C.4.11. A link Z is included in the force sensor path. but very small). On the other hand. if 'the sensor measures the output voltage. and the feedback system output is a voltage source which means its internal (output) impedance is 0 (certainly.
the return ratiois T(). F = U / [ B ( Z L+ 2)] * The same expression follows from the analysis of the circuit diagram in Fig.) The formula expresses . Therefore.14(b) where the source with electromotive .13 ~ ~ point impedance i ~ i calculated system be can using feedback of system a Blackman's formula. with the feedback loop the disconnected. In electrical systems. The feedback return ratio in the system certainly depends somehow the on impedance 2. 7.9) where Zo is the impedance without the feedback..4. 7.12 (a) Feeding back a linear combination of sensor readings and (b) an equivalent circuit diagram for the junction Since the velocityV = FZL. WJ when ZL=.2 Blackman's formula A twowire connection where the current in one (direct) wire equals the negative of the current in the second (return) wire. If the feedback is large and therefore the error is small. return Fig. 7. 7. By this definition. 0. From here. the return signal isBF(& + 2). ~.13 Z L is the comprises a singledirectional twoport network and a threeport linear system. the dimensionof Z is mobility. the return ratio is T(O). Fromthiscomparison we concludethat &e compound feedback makes the actuator output imvedance (mobi1ity)xqual to the transfer function &the link 2. The system shown in Fig.216 Chapter 7. (exampleswillbeprovided later) so that Z ' = T(ZL).forceUIB and internal impedance 2 is loaded at impedance 2 L . Linear Links System andSimulation The output of the link 2 must have the dimension of velocity so that it can be summed with V. the return signal equals the command U . i.~ ~ 2 = Zo T(0)+ 1 T() + 1 (7.5. impedanceofanexternalloadthatcanbeconnectedtotheportwiththeoutput impedance 2.4.e. (The proofs are given in Appendices6 and 7. is called a port. The input impedance of a feedback ~i~~ 7.and when ZL= 00. a linear combination of the current and voltage at'thejunction can be obtained using Wheatstone bridge circuitry as will be shown in Section 7.
Here. 7. the output impedance of the amplifieris 7.Z.15(a). is ratefeedback. or some other velocity sensor. Linear Links and System Simulation 217 driving point impedancevia three quantitiesthat are muchsimplertoestimateand calculate: the impedance without the feedback.Thefeedbackmaintainsthetapevelocity constant and independentof the tape tension. The output terminalis connected to the input terminal. the input of the feedback path. Large parallel feedback canbeused to makenearlyperfectvoltagesources. three ports are connected . is y. A driver amplifier with high input and output impedances has transconductance Y (i.Suchfeedbackis Themechanicalanalogtoparallelfeedback showninFig. 7. (a) electrical and (b) mechanical In a parallel feedback circuit.. the velocity proportional to the input signal and independent of the loading conditions. Velocityfeedbackcanbeemployed. the output of the forward path. The feedback makes the actuator a velocity source. negative parallel feedback reduces the impedance.5. and the load are all connected in parallel. 7. and the feedbackin the cases when the system is simplified by setting load impedance to 0 or . 1. It usesasasensoratachometermotorwinding. (when no load is connected to the output).The signal source impedance is2s. for example.Then.inthemotorwhich propagates the tape in a. The feedback path transfer function.4.via a feedback twopole with impedance ZB. 7.10) Hence. reducing the load impedance (mobility) to 0 disconnectsthe feedback loop and makes T(0)= 0. and Blackman's formula reduces to (7. 7. the output current is YV).anoptical encoder.4.suchastheonedescribedin Chapter 1 and depicted in Fig. from the output current to the input voltage. The fed back signalis therefore proportional to the voltage across the load.14(b). Fig.4Seriesfeedback A ser!es feedback diagram is shown in Fig.Chapter 7. when voltage V is applied to the input. taperecorder.14 Parallel feedback. Here.3Parallelfeedback The parallel feedbackblock diagram is shown in Fig. Example 1.14(a). This is why parallel feedback isalso called voltage feedback.e.
a force or a torque source can be made using series feedback from a load cell or another kind of’force or torque sensor. (b) mechanical. thus maintaining the torque nearly constant and independent of the load. Example 1. ~ rotor flux winding (7.Linear Links System andSimulation in series: the load. the output of the forward path. The current is sensed by the small series resistor R3. Nearlyperfect current sources can be made using large series feedback. it increases the impedance. In mechanical systems. therefore. the short circuit current). to the load current. then the impedance (7. providing proper constant tension in the cable.218 Chapter 7. Compound feedback differs in this respect from series feedback and from parallel feedback. Increasing the load. Compound feedback stabilizes not the voltage or current at the system’s output but the system’s output impedance and the emf (and therefore. Thisis why series feedback is also called current feedback.29. The output voltage is connected to the inputof the feedback path via the voltage divider R1. R2. A circuit diagram for an[ amplifierwithcompoundfeedback at the output and series feedback at the input is shown in Fig. The effect of the flux winding can be described by the following internal feedback mechanism. Le. 7. 1. The torque sources are commonly used in pulleys. (c) series motor When series feedback is. In this circuit. 7. increases the rotor current and. When both IT(0)I and IT(=)l are much larger than 1. Blackman’s formula reduces to z= zo[T(O)+ 11 . 7.. increasing the torque.. reduces the voltage on the rotor winding.negative.depend on the forward path gain.16.e. The total signal at the input to the Bcircuit is . i.11) Fig.5 Compoundfeedback Compound feedback is the type of feedback when both T(0) and T(=) are nonzero.15 Series feedback: (a) electrical. and the input of the feedback path.12) does not . Then. compound feedback stabilizes the output mobility.4. and the free run velocity.15(c). 7. the flux winding is connected in series to the rotor as shown in Fig. the brake torque. Such a current source was depicted in Fig. increasing the load impedance to 00 disconnects the feedbackloop and makesT(*) = 0. The feedback signal is proportional to this current.so that the same current flows through these three ports. the flux winding current. therefore. This increases the voltage drop on the flux winding and. In mechanical rotational systems. In series electrical motors.
the input port of a feedback amplifier. impedance equal to 75IIZ or 50 R to match a cable. Linear Links System andSimulation 219 approximately ULR2/(R1 + R2) + ILR3.Chapter 7. we only considered the output port of the actuator. to obtain the required output mobility of the motor.16 is oftenreferredtoasanimplementation of bridge 2. and the Bcircuit input. Example 3. R2.13) . the output of a servo motor imitates a damper of desired value.5 Effect of load impedance on feedback We have already seen that the load impedance affects the feedback in the cases of series and parallel feedback. When compound feedback employs angular velocity and torque sensors. While the impedance at a specified port is being calculated. or directly from (7. The load impedance is connected to one diagonal of the bridge. 7. or an antenna. it is sufficient to implement compound feedback in the driver amplifier. which only requires several resistors. according to the rule developed when we were considering the system in Fig. This way a flexible plant this motor is driving can be damped and the control accuracy improved.2.thebetterthismethod works. it can be any other port. However. this port open or shorted. Fig. when the feedback is large. 7. the motor output mobility is proportional to the output electrical impedance of the driver amplifier . 7.6.16 Resistive compound feedback at the output of an amplifier The circuit inFig. Blackman’sformulaallowscalculation of thedrivingpointimpedanceatany specified port. and T(0) and T(=) are the return ratios with. to another diagonal.12. Then. the output impedance is R3(R1+ R2)/R2. losses in a motor are small. In highfrequency amplifiers. Therefore. Until now. respectively. bridge feedback with Wheatstone or transformer bridges is often employed to make the output filter.2. is the impedance without feedback at this port.aswillbe shown in Section 7. Generally. 7. R3 andtheoutputimpedanceoftheamplifier constitute a Wheatstone bridge.1l). The drawback of this method is the need to use velocity and torque sensors which may be relatively expensive. a Example 2. Thesmallerthemotorwindingresistance. the return ratio can be expressed as a bilinear function of ZL [9] (see also Appendix 8): (7. When power. feedback since resistors R1. Using only a rate sensor with resulting rate feedback and a driver amplifier with an appropriate output impedanceis cornmonly sufficient for most practical applications. for example.which is the most important for the control system designer.
220
Chapter 7. LinearLinks and
System Simulation
It is easy to check that when ZL= 0, then from (7.13), T(&) = T(O), and when ZL = 00, then T(ZL) = T(w), The feedback dependence on the load makes it more difficult to design systems where the loadimpedance is uncertain or couldvary,especiallywhentheplant is flexible and the resonance fiequencies are not well known.It is preferable in this case to have the feedback independentof the load. The general condition for this is
2 = zo,
(7.14)
since in this case, according to Blackman's formula, T(0)= T(=), and therefore fiom (7.13), T(&) = T(0)= T(). In order to implement this condition, a local loop about the actuator can be made.This loop provides the desired value of the output impedance , for the main loop over the plant as of the actuator, this impedance (mobility) being 2 shown in Fig. 7.17. Such a combinationof local and main feedbackis called balanced bridge feedback.
. w' .
Fig. 7.17 Balanced bridge feedback
7 , 6 Flowchart for chain connectionof bidirectional twoportq
7 . 6 . 1 Chain connection of twoports
Whentwoports are bidirectional,likeelectricalmotorsthatcanalsobeusedas dynTos, or like transformers and gears, the load €or a link depends on the load for the following link. For example, the input impedance of an electrical motor is affected by the input mobility of the gear connected to the motor, and the gear input mobility is affected by its load, A general linear twoport is described 'by a pair of linear equations relating its two input variables and two output variables, like the following equations:
I2 =:
u I ~+ dU2
I

(7.15)
U1= c I ~ + bU2 Equations (7.15) can be representedby the block diagram in Fig. 7.18. The inside of the diagram for each bidirectional twoport is representedby four unidirectional links.
Chapter 7. Linear Links and System Simulation
22 1
Fig. 7.18 Flowcharts of (a) a cascade connection of twoports, (b) a parallel impedance twoport, (c) a series impedance twoport, (d) a ladder network; (e),(f) block diagrams for the ladder network with current and voltage signal sources
222
Chapter 7.Linear Links System andSimulation
The meanings of the coefficients a, b, c, d can be understood from the boundary conditions. The coefficienta is the current gain coefficient under the condition that U2 = 0, Le., under the conditionthattheoutputportisshorted.Thecoefficient b is the reverse direction voltage gain coefficient under the condition that ZI = 0, i.e., that the left port is open. The coefficient d is the output conductance under the condition that the left port is open. The coefficient c is the input impedance under the condition that the output portis shorted.
Example 1. An electrical twoport consisting of shunting impedance 2 ; and the related flowchart are shown in Fig. 7.18(b). The equations of the twoport are
12
= Zl
 (l/Z1)U2 and U1= U2.
Example 2 .An electrical twoport consisting of series impedance& and the related flowchart are shown in Fig. 7.18(c). The equations of the twoport are
12
= I1 and
Ul = &Il
+ U2.
Example 3. A ladder electrical twoport loaded at 2, and excited by current source
Zs with internal impedance 2 s is shown in Fig. 7.18(d).
The transimpedances (the ratios of the output voltages to the input currents) for the networks consisting of 3 and of5 branches, are, correspondingly,
Forthemostfrequentlyencounterednetworkconsisting of threebranches,the numerator numtr and denominator dentr of the transimpedance are expressed in terms of the numerators and denominators ofthe branch impedancesni,;dias shown below: numtr = n1 n3dI dZd3, dentr= n2n3 + nI n3 In1n2 (when using MATLAB, an appropriate number of zeros must be added'in front of the vectors to be added). The flowchart of the network can be obtained by cascading flowcharts for parallel and series branches. The resulting block diagram is shown in Fig. 7.18(e); here,I(&) is the current flowing through impedance Z , , , and U ( Z n ) , the voltage across impedance Z ,. As is seen, the summers in the upper row represent Kirchhoff's law for the currents in the nodes, and the summers in the lower row represent Kirchhoffs law for the voltages about the elementary contours. Bode diagrams of the flowcharts transfer functions can be obtained with SIMULINK using commands described in Section 7.7. When the numerator order n of a link transfer function is higher than the order of the denominator, SIMULINK might not be able to find the solution. In this case asncan be added to the denominator with a sufficiently small coefficient a. With current and voltage summers, any circuit (not only the ladder) can be modeled with a flowchart and modified, if required, by changing only a few links.
Example 4. In the previous example, the signal source is equivalently replacedby a source consisting of emf E = ZS&, in series with impedance &. The resulting block diagram is shown in Fig. 7.18(f). A further example is given in Appendix 13.12.
Chapter 7. Linear Links and System Simulation
223
7 . 6 . 2 DC motors
An electrical dc motor is an electromechanical transducer. The current I and the voltage U characterize the electrical side of the motor model. The generated torque'2: is applied to the mechanical load whichis rotated with angular velocity I I Z . In application the motor, to equations (7.15) be can ollows: asrewritten Electrical
= kI WZm
source (driver)
U = rZ+ ksZ
(7.16)
Electrical Mechanical side I
I
side
andreflectedintheflowchart Fig. in 7.19. Here, k is the Fig. 7.19 Electrical motor flowchart electromechanical conversion coefficient, i.e., the ratio of the break torque (Le., the torque when the velocity I(z = 0) to the applied current. As follows from the first equation, Z m is the motor output mobility measured while the electrical winding is open, i.e., when the current is 0. This mobility reflects the viscous friction in the bearings and the moment of inertia of the rotor, Zm = 1IB + ~/(Js). ZLis the load mobility. The transfer function from the output of the link k in the forward path to the input of the second link k can be found using (1.3) as
,
This is the parallel connection of the load and motor mobilities. The input voltage U equals the sumof the voltage drop rI on the winding resistance r and the back elecfromofive forcekC2. (The reactive component of the winding impedance is mostfrequentlyneglected since it is typicallymuchsmallerthanthe resistance, but in some cases this reactance must also be accounted for.) If the mechanical losses, the rotor inertia, and the winding resistance are neglected, then z= k1 and U = ksZ so that '2:Q= IU, i.e., the power is converted from electrical to mechanical forms without losses. In most lowpower feedback control applications, brushless motors are used with a permanent magnet rotor. The stator windings (phases) are switched in accordance with the .information about the angle position of the rotor obtained from position sensors based on the Hall effect or from optical encoders. The encoder consists of a disk on the motor shaft with specific patterns printed on it and several photosensors placed on the stator and separated by specific angles to read the printed information. The encoders can also serve as angle sensorsfor the feedback path. The sensors are described in more detail in Section7.9.1. A permanent magnet motor can be driven by sinusoidal ac current generatedby dc to ac inverter. The phase and the frequency of the ac current are controlled by a separate feedback loop, using either arotor angle sensor or the information about the rotor angle extracted from the currents and voltages the in motor windings. The periodic dependence of thecoefficient k ontheangleofrotationcauses periodic variationsinthemotortorque. To modelthiseffectinthesystem'sblock diagram, a parallel branch can be added to the k path. This branch includes a multiplier
224
Chapter 7 .Linear Links and System Simulation
to whose second input the shaft angle is applied via the function ak sinncp, where a is the relative amplitudeof the torque variations, cp is the shaft angle, and n the number of torque ripples per rotation. A similar branch added in parallel with the path l/& in Fig. 7,19 can be used to model the holding torque in step motors. Higher powerdc motors have both rotor and stator windings. The stator winding is often referred to as a flux winding. The motor can be controlled by varying the current in either winding or in both windings.
7 . 6 . 3 Motor output mobility
Using the flowchart in Fig. 7.19, the motor output mobility Zout mot can be calculated as ; .The link l/&, which is the inverse of the the inverse of the transmission from $2 to 1 U to I . The mobility output impedanceof the driver amplifier, should be connected from is
= (zd r)fP (7.17) (Derivation of this formula is requested in Problem 21.) 'When r is relatively small and a voltage driver is used, the actuator output mobility is low and the actuator approximates a velocity source. When the plantis driven by a velocity source, the angular velocity is constant and independent on the load and friction. The effect of T can be compensated using a small sensing resistor in series with the motor. The voltage drop on this resistor is proportional to the voltage drop on r. By amplifying this voltage and applying it with proper phase to the input of the driver amplifier (i.e., by making a compensating feedback loop), an extra voltage at'the output of the driver amplifier is created which has the same amplitude and opposite phase compared to the voltage drop on T,thuscompensatingtheeffect of r, Thesensing resistor should be temperature dependent, or some additional circuitry should be added to compensatefor the temperature dependence of the winding resistance. The samecircuitcanbeanalyzed anddesignedwithBlackman'sformula by creatingadriveramplifier with the output impedance equal to ,T, thus making 2 , + Y = 0 and Zoutmot = 0. The feedbackbandwidthinthecompensatingloopislimited by thewinding inductance. The systemstabilitycanbeanalyzed with,themultiloopBodeNyquist criterion, andor with the BodeNyquist criterion for connections of twopoles [2].
zout mot
7.6.4 Piezoelements
Piezoelement actuators can be analyzed using following the equations:
F =aU+dV
I = cU +,bV
(7.18)
reflected in the flowchart shown in Fig. 7.20 Piezoactuator flowchart Fig. 7.20. The coefficient a is measured under the condition of zero output velocity,Le., when the output is clamped. It is the ratio of the clamped force to the incident voltage. The coefficient b is the reverse direction transmission coefficient from V to I while the input port is shorted.
Chapter 7. Linear Links System andSimulation
225
The Coefficient d is the inverseof the output mobility under the condition that the input port is shorted. The coefficient c is the input port electrical conductance under the condition that the output port is clamped. The coefficient a is measured under the condition of zero output velocity, Le., when the output is clamped. It is the ratio of the clamped force to the incident voltage. The coefficient b is the reverse direction transmission coefficient from V to I while the input port is shorted. The coefficient d istheinverse of theoutputmobilityunderthe condition that the input port is.shorted. The coefficient c is the input port electrical conductance underthe condition that the output port is clamped. Both the electromagnetic actuatorsand the piezoactuators possess some hysteresis due totheferromagneticandthepiezoelectricmaterialproperties.Mostoften,the hysteresisissmallandcan beneglected.Ifnot,itcanbemodeled by introducing hysteresis links (described in Chapter 10) into the elementary links in the diagrams in Figs. 7.19 and 7.20. 7.6.5 Drivers, transformers, and gears The input impedances of driver amplifiers are typically high, i.e., the amplifiers are volfageconfrolled. The amplifiers are characterized by their transconductance YT = IdU1 (measuredwithzeroimpedanceload)ortheirvoltagegaincoefficient K = ElUl (measured with no load); here, 1/1 is the voltage at the driver's input, and U,is d where z d is the output impedance the voltage at the driver's output. Evidently, = YT z of the driver.The driver amplifier flowcharts are shown in Fig. 7.21.
Fig. 7.21 Flowcharts for (a), (b) a driver amplifier with output impedance (c) a current driver, and (d) a voltage driver
a,
Example 1. In Fig. 7.21(b), the emf at the output of the driver E = UIK. The motor angularvelocitycanbecalculatedfromtheequationsobtained by cascadingthe flowcharts.for the driver and themotor in Fig. 7.19. With the motor mobility neglected or included into the load, the angular velocity S2 and its sensitivity to variations in the load mobilityZL are
a=
E 
I
dS2
z d+r 1+
k
and
ZLk2
sn =a dZL Z L
1
1 "  ZLk2.
z d
+
r
t
It is commonlydesiredthatvelocityinthevelocitycontrolledsystemsbeless dependentontheloadmobility Z L (whichincludestheuncertainfriction andload dynamics), i.e., IS,l needs to be small. The sensitivity depends on zd. Therefore, it is important to choose and implement z d properly. The sensitivity becomes small when
226
Chapter 7.Linear Links and System Simulation
lZd + rl is small (recall Example 1 from Section 7.3, the description of the effect of Y in Section 7.6.2, and the effect of the actuator impedance described in Section 7.5).
A flowchart for electrical an transformer with the turn ratio n is shown in Fig. 7.22(a), and a flowchart Fig. 7.22 Flowcharts for (a) an electrical 1:n transformer and (b) a mechanical gear for a mechanical gear with the velocity ratio n, in Fig. 7.22(b). The resistance of the primary winding being11 and of the secondary, r2, the total equivalent resistanceof the primary is Y = r1+ r2/n2.For a mechanical gear box, the equivalent viscous friction coefficient for the motor output motion is B = B1+ B2/n2. The composite flowchart of a driver and a motor with an attached gear box is a cascade connectionof the three corresponding flowcharts.
Example 2 . Thedriverhasvoltagegaincoefficient10andoutputimpedance 0.25 $2.The motor winding impedance is 20 + 0;0003s R and the motor constant is 0.1 Nm. The motor rotor's moment of inertia is 0.02 Nm2. The gear amplifies the motor torque twice, i.e., the gear ratio, the load angle to the motor angle(or, the load angular velocity to the motor angular velocity), is 1:2. The losses in the gear and bearings make B = 0.01 Nm/(rad/sec). The load's moment of inertia is 0.2 Nm2, i.e., the load mobility is a is 5/s. The model for the driver, motor, gear, and load assembly shown in Fig. 7.23 cascade connection of the models in Figs.7.21,7.19,7.22, and the load. Since the output impedance of the driver and the impedance of the motor winding are connected in series, the model can be simplified by changing the value of the link U0.25 = 4 to V(20.25 + 0.003s)and eliminating the link 20 + 0.003s.
""""""""""""""""""""".
c " " " " " " " " " " " ,
."""""""""""""""",
Fig. 7.23 Block diagram of cascade connection of driver, motor, gear, and load
Example 3 .The actuator of the previous examples applied to the plant which is not rigid: the torque (zL friction torque) is applied to the body with moment of inertia J1 (which reflects the inertia of the gear and the motor), and an antenna with moment of inertia J3 is connected to the gear via a shaft with torsional stiffness coefficient k2. Therefore, the load mobility 51s in Fig. 7.23 must be replaced by the mobility 2: which is the input impedanceof the equivalent electrical ladder ITtype network:
z=
1 1 Jls + s +1
k2
J3s
Chapter 7. Linear Links System andSimulation
227
This mobility has two zeros, k j d m , and three poles: kj,/k,( J , + J 3 )/ J, J3 and 0.The pair of purely imaginary poles will bring about a pair of complex poles in the transfer function 0L/U1 where 0 L is the angular velocityof the output of the gear. These poles will be substantially damped by the output mobility of the actuator.
7 . 6 . 6 Coulomb friction
Coulomb friction is modeled by the block shown in Fig. 7.24(a). The dependence of the friction force Fcoulomb on the velocity V is shown in Fig. 7.24(b).
V
friction (a)
Fig. 7.25 Model ofadynamic
Fig. 7.24 Coulomb friction model system friction with and characteristic
The friction model is commonly incorporated in the plant model as a feedback path shown in Fig. 7.25. In this composite plant model, F is the force applied to the plant, and the difference between this force and the Coulomb friction force is applied to the plant dynamics (the summer is also shown in Fig. 7.23). When the actuatoris not a pure force source, the inverse of the actuator output mobility can be included in the model as a feedback path in parallel with the Coulomb friction link. In Fig. 7.22, the model of a gear was shown with viscous friction B. Instead of this block (or, parallel to this block), a Coulomb friction link can be placed. The Coulomb friction link can be also placed parallel to the link l/Z,,, in Fig. 7.19.It is seen that when Zm is small, the link l/Zmwill dominateandtheeffectofCoulombfrictionwillbe negligible. It is seen therefore that the effect of Coulomb friction greatly depends on the actuator’s output mobility.
Example 1. When a rigid body is dragged over another rigid body with a rough surface by a soft spring, Coulomb friction causes oscillation (with some rearrangement of coordinates, this is the case of playing a violin). In this case the actuator’s output impedance is large, and the actuator is a nearly pure force source. However, when the source is that of velocity, evidently, no oscillation occurs.
7.7 Examples of system modeling
A system model can be described either by the system elements and the topology of their connections (as in SPICE), or by mathematical equations (as in MATLAB and many other computer languages). The use of flowcharts simplifies the organization of the mathematical description of the system. The equations expressed in the flowcharts canbeenteredintheinput file of acomputersimulationprogramviaagraphical interface (as in SIMULINK@ and some other control software packages).
Example 1. A SIMULINKlike model of a control system using a brushless dc
228
Chapter 7.Linear Links System andSimulation
permanent magnet motor is shown in Fig. 7.26.
ideal motor
e sin
t
3
Compensator +
t . sinmp
e
command
b 
+. _ .
winding constant
position feedback
+ Quantizer
Fig. 7.26 Block diagram of a control system
The system includesan input summer, a compensator with voltage output, a voltage driver as a voltage controlled source of voltage U,a brushless permanent magnet dc motor with motor constant k, a plantwith specified mobility, a friction model, a position (angle) feedback including a ‘quantizer (since an opticalencoderusedastheangle sensor),andthemodels of themotortorquevariationsandcoggingasperiodic dependencies on the output angle 9.The torque z is applied to the plant, the output angular velocity is a. The back’ electromotive force is subtracted from the driver’s output voltage U,and, divided by the motor winding resistancer, produces the winding current I . The torque z istheidealmotortorque(currentmultiplied by the motor constant) fiom whichthetorquevariations,coggingtorque,andfrictiontorqueare subtracted. SIMULINK can provide the output timeresponse to the input signal. The SIMULINK analysis tools (signal source; oscilloscopes; plotters; multiplexers to provide data for the workspace)are not shown in Fig.7.26. Their use is described in the manuals. & A Bode diaaram can be foundas follows: disconnect the feedback vath, attach invort 1 @om the Sirnulink “Connections” libra?) to the command inpgt, outport 1 to the loop outvut, and type in the MATLAB commandwindow:
[A;B,C,D]=linmod(’file_name‘);
bode(A,B,C,D,l)
The closedloop frequency response can be obtained by applying the same program to the system with reconnected feedback path. The meaning of the matricesA,B,C,D is explained in Chapter 8.
Example 2. The two methods of system description are illustrated below with an example of a vibration isolation system shown in Fig.7.2’7.A voice coil actuator with a load cell sensor is placed between two flexible bodies with mobilities 21 and,&. The vibration source is on the second body, and the actionsof the voice coil should reduce the vibrations of the first body. The load cell together with its amplifier has sensitivity 1 VN., The voicecoil is characterized by thecoilresistance r = 4L l andthe electromechanical coefficient k = 3 N/A. The coefficient’s equivalent is a lossless downof 3 (the ratioof the primary to the secondary windings). transformer with the turn ratio
I
7. then probably flowchart simulation withSIMULINK will take less time than simulation with SPICE. to a large extent. Linear Links and System Simulation 229 Fig. force to be Fig. It can be used. on the designer’s personal preferences. I 1 Load cell L C  Driver electrical side V o i d coil mechanical side Fig. the driver amplifier needs to be designed and optimized simultaneously with the compensator. All currents Body 1 andvoltagestotheleft of thevertical dashed line representphysicalelectrical variables in amperes and volts. The schematic diagram can now be codedinto the input file of SPICE. The load cell is therefore represented by an ampermeter. and the designer is a mechanical engineer. and SPICE.Chapter 7. If. the driver amplifier is already designed and should not be changed. Zout is the output impedance of the driver amplifier. 7. Here.27 Vibration isolation system case.equivalent 7. they represent forces in newtons velodities and in dsec. 7. 7. for example.29 Flowchart descriptionof a disturbance isolation feedback system The choice of themodelingmethoddependsontheavailabilityofsimulation software. for example. with SIIMULINK.29 shows a flowchart description for the same system.28 the shows F i r electrical schematic diagram.andthedesigneris an electricalengineer. in this Fig. If.28 Equivalent schematic diagram for a vibration isolation system Fig. the compensator is analog. on the problem specifics. a currentcontrolled voltage source. to the right of the line. however. on the system links. correspondingly. . the compensator C is implemented in software. and. then probably the design and simulation is easier to perform using an equivalent electrical schematic diagram.
'7.thecloseristhepole tothezeroandthesmalleristheshiftinthe asymptote. for example. by the masses of allthebodiesconnectedtogether by springs which areconsideredstiffatverylow frequencies. The angle of the mobilityis 90" where 1 is rising. As a result.31 Driving point impedance (mobility) of a lossless system. as shown in Fig. the poletozero ratio of the flexiblemode equals the square root of the ratio of the sum of the masses to the mass of the main body. If. Linear Links System andSimulation '73 Flexiblestructures 73. there could be either a pole or a zero. 7. The highfrequency asymptote is determined solely by the body to which the incident force is applied. the peaks and valleys on the frequency response of 1 2 1become smoothed. an actuator drives a massive main body from which a small additional mass is suspended on a spring. logarithmic scale: (a) having a pole at zero frequency and a zero at infinite frequency.2 Lossless distributed structures Distributed structures can be approximated by lumped element structures with a large . '7.31. Le. The lowfrequency asymptote is determined by the rigid body mode. 7. the losses in mechanical systems without special dampers can so besmall that the peaks reach 40 and even60 dB over the smoothed response of the mobility.36.8. At zero andat infinite frequencies.230 Chapter 7. a l&sless twopole impedance Fig.1 Impedance (mobility)of a lossless system The Foster fheorem states that the zeros and poles of a driving point impedance (mobility) of a lossless system are purely imaginary and alternate on the jaaxis.and 90"where IZI Figs. 7.30. Still. It isseenthatin Fig.. and a prominent suspension mode It can be calculated that each additional zeropole pair lifts the highfrequency asymptote by the square of the ratio of the pole to the zero frequencies. the highfrequency asymptote is shifted from the lowfrequency one. 4. (b) with zeros at zero frequency and infinite frequency. This ratio commonly depends on the mass participation in the flexible mode. The smaller the mass of the smallerbody. Structural damping displaces the poles and zeros to the left of the jaaxis. Allotherbodiesaredisconnected splane springs' body since the from this mobilities become very large at high frequencies. The lowfrequency asymptote and the highfrequency asymptote reflect I21 beingeitherproportionalorinverselyproportionaltofrequency. Mobility frequency responses of nondissipative flexible plants are exemplified in Z 1is falling.31(a).
The waveimpedance (mobi/ity) is theinputimpedance(mobility)ofthe structure extended to infinity so that no signal reflects at the far end and returns back to the input. The plot of lZI in logarithmic unitsis commonly nearly symmetrical about the wave impedance as shown in Fig.Chapter 7. 7. and the input impedance (mobility) of the structure equaling the wave impedance (mobility).Atthequarterwaveresonance. In electrical transmission lines.32 (a) Distributed structure. no resonances take place). Fig. where k and A4 are the stiffness and the mass per unit length of the structure.waves reflected at the ends of the structure.32(c). the impedance (mobility) of a distributed structureZ(s) possesses an infinite number of poles and zeros. In this case. are toalargeextentuncertain. Example I.This uncertainty becomes a critical factor in limiting the bandwidth of the available feedback. we can think of the structure as extending to infinity. the wave impedance is p = (Z/c)ln = . A motor with motor constantk is employed to rotate a spacecraft solar panel having moment of inertia J to keep the panel perpendicularto the direction to the sun. Matching between the signal source and the structure is provided when the signal source output mobility equals the wave mobility of the structure. Then. (e)its driving point mobility Correspondingly.Thewaveimpedanceofauniformlosslessdistributedstructure is the resistance p = 1 / f i . The desired motor output mobility can be created bycompoundfeedbackinthe driver. and if at either end no reflection occurs due to matching. 7. 7. This can be achieved by making the motor output mobility approximately equal to the solar panel wave mobility p. 7. as illustrated in Fig. as has been discussed in Chapter 4. We calculate the mobility using the voltagetovelocity electromechanical analogy. at a dashpot) with the impedance (mobility) equal to the wave impedance (mobility) of the structure. These poles and zeros can be viewed as produced by interference of the incident signal and the signal reflected back from the far end of the structure. Linear Links and System Simulation 231 number of elements as shown in Fig. the phase is27~$k(Zc)'~ where x is the length andZ . if mechanical.32(c). 27~frx(Zc>'~ = lc/2 wherefrom ( Z C ) ' ~ = 4&. Matchingat either end of the structure fullydamps it (sincetheresonancescanbeconsideredaresult of interference of the . Mafching at the farendmeansloadingthestructureatsomeresistor (or.inphysicalsystems. The solar panel torsional quarter wave resonance frequency is fr.andtherefore.32(a) and (b).The spacecraft attitude control can be improved by damping this resonance.and c are theinductanceandcapacitanceperunitlength. The frequencies of highfrequency resonances are very sensitive to small variations of the structure parameters. (b) its equivalent representation with lumped parameters.
Since the torque is thesamebefore andaftertheshaftflexibility. andthephaseangle of thefunctionisconstrained withinthe [go".orcurrent). The sensor and the actuator cannot always be placed exactly at the same location. 8 . The cx is the full capacitance of the line. (The control will not be collocated if there will be an additional rigid body with substantial moment of inertia at the point of the torque application. the actuator is a torque source and the sensor measures angular velocity dq~lldt (or a linear operator of the velocity. Otherwise. the actuator and plant . 90'1 interval.33 (a) Actuator is a torque source. In the mechanical arrangement shown in Fig.thecontrol remains collocated evenwhen the sensor cp2 is used.theplanttransferfunction P = cp/z is the mobilityof the mechanical system. (c) the case is considered in Section 7. 7. functions are reviewed in Appendix 2. for better control accuracy.4 Noncollocatedcontrol Contrary to the driving point lossless mobility. as shown in Fig. p' ~ ddt ddt $ 2 Fig. Linear Links System andSimulation ' inertia J . 9 1 or 1 p 2 providing .232 Chapter 7. the output mobility of the actuator is finite. Such control is called mentioned in Section 4 . 7 .33(b).4. .8. the control would still be collocated.As has been shown in Chapter 4. as was already (mobility or the inverse of mobility).) 7.8. (Properties of p. control is collocated when sensor is used and is noncollocated when sensor (92 is used IntherelatedblockdiagramshowninFig.which is analogous to the moment of Frequently. 7. p = 1/(4Jf) and the driver output mobility is 1 / ( 4 & & ) .7. For example. The driving point impedance (mobility) of a passive plant is psitiwe real (pr. (b) block diagram of collocated control.33(a).33(c). this phaseshift reduces the available feedback.).r. the fed back signal. If the plant were an ideal rigid body. the control is collocated when the first sensor is used and is not collocated. 6 . 3 . control is collocated with any of the sensors.Le. such as position or acceleration). or the actuatoran ideal force (or torque) source. and the phase shift of the transfer functions is not constrained.) This feature greatly simplifies the controller design and the stability provision. the sensor might be placed closer to the load. andthe feedback sensor is connected to the same portof the plant but reads the variablewhich is related to the actuator variableby the plant driving point impedance (mobility). transfer functions between different ports of a passive lossless structure may have several consecutive poles or zeros. 7. 3 Collocatedcontrol 1/(4cxh).. the actuator is apuresource of force(ortorque. Hence. The plant transfer function is in this case the plant driving point impedance or admittance collocated. Example 1.
or eddycurrents. p. Introducing structural damping can greatly improve the control system performance. the linear variable differential transformers (LVDT).34(a) measures the position of the plantwith respect to the base. weight. The dampers are. As has been shown in Section 4.. The dampers can be built using Coulomb friction.1. hydraulic energy dissipation.e.9. 7. with resolutionstill better than0. velocity) sensors.5”. For position and angle sensors.1 Position and angle sensors For position (or angle) variables’ measurements. generally. however. three categories of sensors are commonly employed: (a) position sensors. 7.3. and he must be able to give his recommendations for the required changes in the structure andto evaluate the posed solutions in real time during the meetings with the structural desigEr.. The LVDT shown in Fig. Withrespecttotheflexibility of theshaftconnectingthemotorandtheload.and (b) the control feedback loop bandwidth must be many times wider than the functional band of the system (as has been shown in Chapters 4 and 5). the potentiometers. relatively expensive. the control can be called collocated only as an approximation of the reality. structural designers tend to design structures with structural modes falling within the feedback bandwidth which prevents the control system from achieving high accuracy.1%. An electricalpotentiometer withthetapmovedmechanicallybytheplantcan provide high resolution and the position reading accuracy and linearity of 0. the resolvers. and (c) accelerometers. In this case. and to constrain the system cost. the optical encoders. It has three windings: the two symmetrical windings to which a signal from a generator is applied in opposite phases and the winding providing the output voltage.r.9. (b) rate (i. Trying to make the structure as inexpensive and as lightweight as possible. This is why the control svstem designer/dynamicist needs to be involved in the desigE process before the mechanical structure design is completed. the plant flexible modes restrict the available feedback. Linear Links System andSimulation 233 transfer function is not. using the dampers must be well justified by the available performance improvement.’’ The structural designer is mostly concerned with the structural soundness over the working frequency range and might underestimate the following factors: (a) the structural mode frequency is proportional to only the square root of the stiffness. and the star trackers are representative. The voltage is amplifiedand applied to a synchronous detector the output of which is a product of this voltage and the generator voltage. misunderstanding frequently arises between the designer of the mechanical structure and thecontrol loop designeraboutthemeaning of thestatement: “Higher frequency structural modesare much higher than the frequency range of operation. The dc component of the detector’s .Chapter 7. the laser interferometers.1Motionsensors 7.9 Sensor noise 7. when the second sensor is used.Less accurate potentiometers are universally used in small servos for radiocontrolled toys. the approximation being valid only over a limited frequency bandwidth.Usingthedampersallowsasubstantialreduction of theeffects of structural modes.6. and dimensions.
It has two stator windings and two rotor windings.To a pair of these windings an ac signal is applied in quadrature. there canbe only a tachometer is a dynamo mounted on the same shaft as the motor separate tachometerwinding onthemotor'srotororstator. The resolver shown in Fig.6.andcountsthefringes of the interference when the distance changes gradually.SpecializedICs are available which incorporateall necessary electronics. 7. 9 . There also exist rotaryversions of thedevice.themotorwindingsin brushless motors which are disconnected from the driver during certain rotation angle intervals. can be used as tachometer windings). the gyroscope. The (or. with simple patterns of alternating transparentand black lines. An electrical winding The rategyro is a gyroscope with a position servo loop. . the angle of rotation can be determined with high accuracy.2. the tachometer is not able to detect the angle of rotation when the rotation rate is very low since in this case the signalon the tachometer winding is below the noise level.234 Chapter 7.34 Position sensors: (a) LVDT.Lessaccurateandlessexpensiveinterferometers use modulated light beams. In contrast to the resolver. The emf on the tachometer winding is proportional to the motor angular velocity. From the signal induced on two the other windings. Quadrature incremental encoders have two 90"shifted readers. The incremental encoders must be accompanied by some electronics keeping track of the counts. 7. or incremental. Interpolation of the data from the readers additionally improves twice the angle reading accuracy. with complicated patterns on the optical disk which at any time give full information about the shaft angle. The interferometer can measure large distances withnanometeraccuracy. The encoders can be absolute. The optical encoder was already briefly described in Section 7.1" to 1" withprettygood linearity. 1 . 7 . (b) resolver The maximumstroke ofcommonLVDTs isfrom 0. and the resolution of precision LVDTs can be 1 microinch. The laser interferometer compares the phases of the incident laser light with the lightreflectedfromamirrorplaced on thetarget.34(b) is a rotating transformer.Linear Links System andSimulation output is proportional to the third winding displacement from the central position. I e Y l output Synchronous detector I I coswt Fig. 2 Rate sensors The most often used rate (velocity) sensors are the tachometer and. The star tracker is a small precision telescope equipped with an image recognition system. This improves the accuracy by a factorof 2 and also gives information about the direction of rotation.
withthefrequencydomaincharacterization.9. rates.theirnoisespectral density responses. piezoresistive. Gyros are worse in position determination than the star trackers at frequencies from 0 to 0.35. In an electromagnetic accelerometer.01 Hz. the magnetic proof mass is suspended on a spring. Linear Links System andSimulation 235 generates a torque preserving the relative position of the gyro to the base. it is common to use both the analog and digital outputs fromthe gyro.Chapter 7. An example of an accelerometer control loop is given in Section11 9 .ittakes considerable time to accumulate enough light to identify dim stars in a star tracker. current in the coil or a voltage producing electrostatic force) serves as the measure of the acceleration. from increases and the accelerometers become superior.g.Theprincipaldifferencebetweenthemistheirdynamicaccuracy characteristics. A set of three orthogonal accelerometers can be used to determine the vector of gravity force and. or. but it loses some higherfrequency information. motion sensors are interchangeable: suitable integrations and differentiations allow us to translate between positions.. The proof mass position relative to the base is kept constant by a servo feedback loop which applies electromagnetic force to the proof mass. this current is proportional to the base angularvelocity(i. and tunneleffect devices. . The frequencyresponses of typicalsensornoisespectraldensityconvertedto position data are shown in Fig. The current in the winding is the output signal of the gyro.3 Accelerometers The accelerometers use electromagnetic.7.Theratesignalisanalog. Motion of the baserelative to the proof mass produces electromotive force in a coil surrounding the proof mass which is mechanically connected to the base. kept constant by some force. gyros have drift(slow and gradual change in the reading caused by device imperfections) and therefore do not determine the position accurately after some time passes from the initial setup. AD conversion simplifies the data interface with the rest of the system.1.the rate ofthebaserotation).e. piezoelectric. This force creates acceleration nearly equal to the base acceleration so the proof mass remains still relative to the base.9.1. The noise power and the mean square error can be calculated by integration of the spectral density with linear frequency scale over the bandwidth of interest. andaccelerations. At even higher frequencies (say. but theiraccuracy decreases whenthepositionchangesrapidly..4 Noise responses Ideally. For these reasons. On the other hand. The signal producing this force (e.the inclination of a vehicle. 7. Position sensors typically give an accurate steady state value of the position.Forexample. therefore. 7. so that this sensor cannot react fast to the position changes of the spacecraft on which the star tracker is mounted. The value of the compensating force in accordance to Newton’s second law is the measure of the acceleration. via a servo loop. but get better at 15Hz and up) the gyro noise higher frequencies. whentheoutput data is postprocessed. There exist many different types of accelerometers using a suspended proof mass the position of which is measured by some sensor and.
Thus.2 Effect of feedback on the signaltonoise ratio The most common noise sources in control systems are those of the error amplifier (which is the first amplifier after the feedback summer). 7. the loading for the disconnected ports must be preserved while disconnecting the feedback path for appropriate comparison of the system with and without feedback. Therefore. of resistors.36. . when the output effect of the sensor noise is calculated. 7. (b) changing the ratio As was mentioned in Section 1. There is a caveat here:when the system performanceis compared with and without feedback.35 Frequency responseof sensor noise normalized to the sensor input 7. and of sensors.36. the transmission coefficient fiom the sensor to the system output should remain unchanged.1. Fig. 7. Particularly.236 Chapter 7. The difference is shown in Fig. not between the sensor output and the feedback path. the system shouldnot be changed in any other aspect.2. . Fig. A feedback loop which reduces both the signal source and the noise source effects does not change the ratio of the signal to the noise at the system output. The noise is commonly characterized by its spectral density. Disconnecting the feedback loop while (a) preserving the signaltonoise ratio at the system output.9.it is possible and often convenient to examine the signaltonoise ratio as if there were no feedback. the feedback loop should be disconnected between the system output and the sensor input. Linear Links and System Simulation  .
From here it follows that U = E + TE = ( T + l)E = FE as diagrammed in Fig. If. 7.TE.38 Analogy between (a) a feedback system and (b) a system with two parallel forward paths. feedbackandrepresentstheparallelconnection of two paths.l).3) are apparent. The feedback loop can be disconnected by setting 0 R2 = OQ.and(1.l).Feedbacktotwopoleconnectionanalogy Equations (1.37 shows an amplifier with the €eedbackpathfromtheennitterofthe outer stage to the emitter of the input stage.10. but because the resistance in the input contour increases which reduces the signal and increases thenoise.3) also describe the signals in the twopole connections depicted in Fig.38(b).Whenthefeedbackiseliminated by 1 = 0. The transfer functionsof the links in block diagram (b) recite Ohm’s law: voltage VI applied to the first twopole causes current I . 7.37 Amplifier with setting R since in this case the resistance in the input an emitter feedback path contour. Fig.3)remainvalidforthissystemwhichhasno IT+ It > 1. 7. When IT1 >> 1. and (1.decreases. the signaltonoise ratio remains the same with or without the feedback.the signaltonoise ratio improves Fig. Fig. The contour equation U1= U .and (1.2.U2 reflects the summer in the feedback loop. however. . The formulas (1. I 7.loaded onto loads equal to the loads that each port sees when the feedback path is closed. When introducing the channelT increases IUI and reduces the ratio E/U. Z ) . all the features of the feedback equations (l. both disconnected ports of the feedback path are . and this current applied to the second twopole produces voltage drop Uz. 7. This reduces the signaltonoise ratio not because of change in the feedback.1Feedbacktoparallelchannelanalogy The summer in Fig. ( l . l). 7.2).38(a) implements the equation E = U . then U 2: ET and U2/V = 1/B.10. (1. We will use this analogy in Chapter 12.Chapter 7. In this way. This analogy can be employed to analyze or simulate responses of system (a) when this system is unstable and system (b) is stable. 7.10 Mathematical analogies to the feedback system 7. (1.2).39(a). Linear Links System andSimulation 237 Example 1. 7.
the system is unstable with nearly all possible a.10.I U u.~cos(t))y =o (7. Y. + (a + 2. This is analogous to the closedloop transmission of a system with large feedback where the output is the input divided by the feedback path transfer function.as seen from the stability diagramin Fig. the output. + &) == U/&. Iotermdulation of thesignalharmonicsintheLTVlink 2~costproducescertain c m p w p f s a$ its output.Whenthe coefficient E is large. 7.19) which is representative of some LTV systems that might be encountered in practice.. the second twopole can be neglected when considering the current calculations: I = U/(l/zY. Linear Links and System Simulation I = u. IT1 >> 1.238 Chapter 7. We will use this analogy in Chapter 10.Thetimevariablecoefficient 2~cost changesthe solutions which are system behavior: some combinations of E and a lead to exponentiallyrising withtime. 7. andmight contain higher harmonics. 7.11 Linear timevariable systems Linear timevariable links (LTV) are described by linear equations whose coefficients explicitly depend on time.stabilitydiagram showninFig. Addition of these components to the signal passing through the LTI link a altersthephase of thesignalatthesummer’soutput. 7. is not necessarily sinusoidal.andothercombinationsintroducedampingintothe system. i. If this equation describes an LTI lossless resonator. Y. is that l&l>> ll/Yll. the impedance of the second twopole is much larger in magnitude than the impedance of the first twopole. A feedback system described by Mathieu’s equation is shown in Fig.WhenthesignalappliedtotheinputofanLTVlinkis sinusoidal.40depictstheareas of stability and instability in the plane of the equation parameters. uu. The analogycanbeemployedtousethepassivitycondition of anetwork of electrical twopoles for stability analysis of feedback systems. When several sinusoidal components are applied to the link input. The InceStrutt. LTV linksof digital compensators have already been analyzed in Section 5. In this sectionwe will consider Mathieu’s equation d 2y/dt E = 0.41(a).e. ’ .. = CAP U2=Z. contrary to the case of a linear timeinvariable link (LTI).40.7. the output contains intermodulation products. As a result. Thesolutionis on theboundarybetweenselfwith the angular frequency oscillationandexponentialdecay. 3 Z2=B Y1Z2= T Fig. the solution being a sinusoid &.7.39 Analogy between (a) a twopoles’ connection and (b) a feedback system This large feedback condition here.
In this case. nonlinear links can be seen as LTV links. This pumps the energy.This results in some uncertainty range in the link phase shift. . 7.41(b). When the compensators are LTV like in digital systems described in Chapter 5 or in the systems where the compensator parmeters are varied in search for a maximum of a certain performance index. and an equivalent mechanical system. the dynamics that limit the value of the feedbackand affect substantially the stability margins). the compensators are LTV. when controlling LTI plants.41 (a) Feedback system.the resonance frequency of the resonator. However. the rate of the compensator variations in time is chosen much lower than the rate of changes in the critical part of the signals and the system dynamics (Le. We will use this approximation in Chapter 12 for the analysis of process stability. The phase shift for the passing signal in periodical LTV links depends on the phase of the incident signal. on average over the length of the cycle. To increase the feedback about an LTV plant. in or out of the resonator. the voltage on the capacitor increasesand the stored energy increases (the energy is proportional to the square of the voltage). a controllercan be chosen to be LTV in such a manner that the loop transfer function is less dependent on time (i. it is generally appropriate to use LTI compensators. the available feedback is reduced. Such links are called quasistatic. The timevariable term periodically changes ..Therefore. but typically.41(c). the uncertainty of the loop transferfunctionreducesthepotentiallyavailablefeedback. the LTV link for the purpose of stability analysis can be considered LTI. In adaptive systems (see Chapter 9).The resonance frequency can be changed and the energy pumped into the pendulum resonator (c) by moving the center of mass of the swinging body up and down.e.. Therefore. while analyzing stability conditions. 7.40 InceStrutt diagram resonators described by Mathieu's equation An electrical circuit equivalent to the feedbacksystem is shown in the diagram in Fig.Chapter 7. in Fig. 7. Linear Links and System Simulation E 2 1 239 0 1 2 3 4 n " Fig. when the plant gain increases. For example. the worst possible case needs to be considered among all possible phases of theincidentsignal. when the resonator capacitance is reduced while the charge is preserved. and (b) and (c) Fig. the compensator gain decreases accordingly). 7. For small signal deviations from the current value.
02.42(a) for the case MI = 50. M = 50 kg. kt = 0.g. one plate sliding arbitrarily on the surface of another plate). What are the equivalents of Ohm's law for translational and rotational mechanical systems? For thermal systems? What are the equivalents of Kirchhoff 's laws for mechanical systems (consider two analogies) and for heat transfer systems? Draw the electrical analog circuit for the translational mechanical system Fig.72. and for the thermal system in Fig. 3. (c) anxy positioning system. J 2 = 3.42(c). 7.1.01. 7. (c) 10 (m/sec)/N. 8 Using equations (7. 7.42(a). 7.5).9. the rotational system in Fig. U = E 9 In Fig. or use SPICE.. 7. k. 10 The plant is a rigid body. k = 0.42(c) for the caseRT = 2. show that if ZL = 0. C = 100. (9) a fourwire junction of two electrical circuits. then I = IS. the loading curveis expressed as I = ( E . (d) Same as (c) for the case J t = IO. (b) 5 (m/sec)/N.. (e) TJ? for the systemin Fig. Problems Why is it convenient to preserve power while choosing the type of electromechanical analogy? How many independent variables are at (a) aball joint. 7.and when ZL= 00. (f) a threewire junction of two electrical circuits. A& 50. (a) V2/Ffor the systemin Fig. k (b) VdFfor the systemin Fig.2)(7. (b) a pin. (c) n d t for the systemin Fig. ki = 0.U)/Rs.42(a) for the caseMI = 100. 7. The viscous friction coefficient is 0.42(b) for the case J 1 = 20.01.03. (d) sliding planes (e.13. k 1 = 0. The actuator output mobility is (a) 1 (m/sec)/N. Derive the function using Lagrange equations or equations corresponding to an equivalent electrical circuit. and use MATLAB. Use MATLAB to plot the frequency response of the actuator together with the plant. 7. J 2 = 24.2. (e) a twowire junction of two electrical circuits. J 3 = 2. Fig.k 1 = 0.42 Examples of dynamic systems 7 Draw an equivalent electrical circuit for cooling a power IC with a heat sink.240 Chapter Linear 7.42(b) with torsional stiffnesses of the shafts k~and k. Links and System Simulation 7. J 3 = 12. A& = 5. What is the plant transfer function uncertainty 10 at Hz? 11 The plant transfer function is the ratio of the output velocity to the force applied to a . in Plot frequency responses of transfer functions: 1 = 2. Express I as a function O f RL.
Use MATLAB to plot the frequency response of the plant velocity to the input of the actuator. calculate the impedance of the simplified circuit. of theactuatoroutputmobility in in Fig. its input impedance is OQ. 20 kg e M e 50 kg. Find the resistorsRt.12. and then simply add this resistance to the obtained result. while calculating the input resistance.16 must be matched to the 50 R load.43. The actuator output mobilityis (a) l(m/sec)/N.000. initially. 17 The velocity sensor gain coefficient is 1 V/(m/sec). 20 kg < M e 50 kg. via a spring with stiffness coefficient k = 1. the force sensor gain coefficient is 3V/N (theoutpufs of both sensors are in volts).43 Examples of feedback amplifiers 15 Determine the output mobility of the motor (without the main feedback loop about the plant) of the feedback systems shown in Fig. 14 Calculate the input and output impedances of the circuits diagrammed the amplifier’s voltage gain coefficient is 10. The driver‘s output impedance is small for voltage drivers and large for current drivers. (b) unloaded? . 7. l/(sM). (c) lO(m/sec)lN. 7. The actuator output mobility is (a) 1(m/sec)/N. Make a conclusion about the effect of the actuator impedance on the plant uncertainty.Chapter Linear 7. and R3 such that signal losses in these resistors are not to be excessive and at the same time the attenuation of thecircuitry in thedirectionofthefeedbackpath is nottoolarge(makean engineering judgment).Links and System Simulation 24 1 rigid body. 7. for cases of the maximum and minimum mass of the plant. (c) 10 (m/sec)/N. What is the output mobility of the actuator? What are the outputs of the sensors when the actuator outputis (a) clamped. 16 Theoutputimpedanceofthefeedbackamplifiershown in Fig. 13 ApplyBlackman’sformulatothecalculation Fig. 7. Fig. disregard R3.44. (b) 5 (m/sec)/N. and its output impedance is very low. R2. 7. Find the output mobility (at the load) for cases (c) and (d). For circuit (d). The sensor outputs are combined to provide large compound feedback about the actuator. Use MATLAB to plot the frequency response of the actuator with the plant. What is the response uncertainty? 12 The actuator is driving a rigid body. (b) 5(m/sec)lN.
The return ratiois (a) 300/(s + 200).6 (radlsec)/(Nxm).44 Examples of mechanical feedback systems 18 The output mobility of a translational actuator is 40 (rn/sec)/N. the load mobility is 0. specifications of a brushless motor. 22 Draw a flowchart of a permanent magnet electrical motor that produces torqueof 10 Nm per 1 A of the current in the winding. RL= 0. 20 (a) A permanent magnet motor has k = 0. Calculate the back emf and the output mobility ofthemotorwhenthesource impedance (Le. RL= 0.5 V/sec“. 23 Draw a flowchart of cascade connection of a driver.2 N xm/A.. R. the free run angular velocity is R. and the mechanical losses in the motor are negligible. = 850 radlsec. and how they correlate with the flowchart in Fig.242 Chapter 7. a motor. the winding resistance being lOR. when no load is connected to the actuator).RS= 6 S2. Linear Links and System Simulation Motor 4 Ahaft Rigid plant 4 angle Optical encoder .24 N xm/A. RL= 1. What is the voltage of the source? The break torque? The torque? (b) Same fork = 0. The driver output impedance is RS= 15 R.” Make a good guess about what these parameters are.5R.PlottheBode diagram. & = 0.4 (radlsec)/(Nxm). The return ratio is 1OO/(s + 30) whenonlyaforcesensor is employed(Le.3). and an inertial load for the following data: (a) The driver has the voltage gain coefficient 5 and the output impedance 0. (d) Same for S2. . RS= 2.12 (rad/sec)/(Nm).. = 400 radlsec.IO). the load mobility is RL= 0.5 R. 7.. the output impedance of the driver) is 3S2 and the current is 2A.Shaft Rigid plant J I Motor Load (c) (dl Fig.0. k = 0. (c) Same fork = 0. 7. = 100 radlsec.19 where there is only one k.1 N x d A .17) for the output mobility of a motor.when only a velocity sensor is employed (Le. R. a gear.8 (rad/sec)/(N xm). = 200 radlsec. Find the loop transfer function whentheloadisa20kgmass. (b) 30/(s +. 21 Derive the expression (7.whentheoutput is clamped). (c) 84s +. RS= 25 R.5 N xm/A. it is written: 19 In the “kt = 72 oz xin/A.8 (radlsec)/(Nxm).
the load angle to the motor angle (or. sec. and the driver is a current source.04 Nm/(rad/sec). and the prototype with Bode step in given Chapter 4.the crossover is 6 Hz. rad. The gear amplifies the motor torque ten times. the load mobilityis 20/s (rad/sec)/(Nm). The gear amplifies the motor torque four times. is 1:4..05sin 168..5/s(rad/sec)/( N .) Use a current driver. is 1 :lo.3.02 sin 80... the torque variation model path 0. (c) (8 + 2)(8 + 40)/[(8 + 3)(8+ 5)sI. The load's moment of inertia is 0. Design the compensator and plot Bode and Nyquist diagrams with the function linmod. and the crossover frequency is m.Chapter 7.15 Nm. the load angular velocity to the motor angular velocity). and the driver is a voltage source. 7. the torque variation path is 0. (c) Do the same when k = 0.e. (e) (8 + 4)(s + 40)s/[(s + 2 ) ( 8 + 20)].e. the load mobilityis 2.The losses in the gear and bearings make B = 0. Le. the motor winding resistance is 4. the load angle to the motor angle (or. the cogging.04sin 160.06sin 89. the cogging pathis 0. The motor rotor's moment of inertia is 0. the cogging sensor is '12 bit 'encoder. or the zeros. or both? 26 Whataretheproperties ofthedrivingpointimpedanceofapassivelossless system?Canthederivativeofthemodulusoftheimpedanceonfrequencybe negative? What happens when the losses are small but not zero? 27 Indicatewhich of the following functions can be an impedance of a lossless twopole. 20 Will the signaltosensornoise ratio at the system output change when the feedback path is disconnected? How much? 29 (a) A capacitor and an inductor are connected in parallel. the torque variation path is 0.m) (b) The driver has the voltage gain coefficient 30 and the output impedance 5SZ. (The values are in model path is 0. the sensor is a 16 bit encoder.6 Nm. for initial part of the responses and for estimation of the accuracy of the velocity during the ramp command. the load angular velocity the motor angular velocity). 24 (a) Design a control system for a motor similar to that shown in Fig. Use SIMULINK. (Make the design using a PlD controller. The motor winding resistance is 30 SZ and the motor constant is 0. i.and ramp commands in position using different scales. and the motor torque variations on the accuracy the output position and velocity. i.the is sin 40. the sensor is a 10 bit encoder. Ohm. 25 What is in common between the driving point impedance and the transfer functions of a linear system: the poles. 15 Nm2. Le.4 Nm2.) Plot the output time history in response to step.2. Study the effects of Coulomb and viscous friction. Linear Links and System Simulation 243 The motor winding resistance is 2 SZ and the motor constant is 0. the plant transfer function is 55/s.5 Nm2. The load's moment of inertia is 0.1 12 Hz. kg.26 where k = 0. the crossover is 20 Hz. N.The motor rotor's moment of inertia is 0 .9.05Nm2. The losses in the gear and bearings make B = 0. (b) Do the same whenk = 0. What is the equivalent of the return ratio? .05sin 40. (d) (8 + 4 ) ( 8 + 40)s/[(8 + 2)(8+ 20)]. and plot the frequency response with MATLAB: (a) (8 + 2)(8+ 4)/[(8 + 3)(s2+ 5)sl. the gear ratio.08 Nm/(rad/sec).the cogging path is 0. (b) (8 + 2)(8 + 3)/[(8+ 4)(8 + 5)s].. the gear ratio.
2 (a) Three angles (three degrees of freedom) and three torques 7 Thediagram is shown in Fig. and its input impedance is nearly proportional . 7.38 and 7. thermal capacitancesof the case and heat sink can be neglected but theyneed to betakenintoaccountwhencalculatingthecollectortemperature during power bursts. 'heat sink radiation LJ " 1 ) / " . what is the analogy to the closedloop transfer function for the system Fig.45.39 for the case of 5 = 1. Thecollectortemperaturemustbebelowa certain specified temperature. 7.t and convection & Fig. When one is calculating the average collector temperature.39(a).000. The nonlinear resistance of radiation and convection cooling of the heat sink is depicted by a nonlinear power source that can be specified in SPICE by the required mathematical expression. using Ohm's law? 32 Redo the equations and block diagrams in Figs.39(b)? 31 What is the condition equivalent to large feedback. and the twoport output impedance is nearly proportional to the output impedance of the preceding link. (c) Same question about the parallel connection of a resistor and an inductor. in the connection of two two poles in parallel? How to explain the effect without using feedback theory. 7.to the input impedance of the following link. 30 In the system composed of two parallel pathsin Fig. 7.Z= R2/10. . Linear Links and System Simulation (b) Same question about the parallel connection of a capacitor and a resistor. 7. Answers to selected problems 1 Since power losses in good motors are small.T(=) = 10.244 Chapter 7. therefore this function can be a driving point impedance. 27 (a) The zeros and poles alternate.45 Electrical equivalent to heat sink 14 (a) The input impedance is: = R2.000. we can directly relate mechanical variables at one port to electrical variables at the opposite port. T(0)= 0.
and lowerbounding responses at the frequencies To satisfythespecification.theplanttransferfunction maps to an areaonthe Lplane. which happens to be Pshaped and the same at all frequencies in the example shown in Fig. and software packages for many of them are readily available.andextensionstocover MIMO cases.somewhatdifferentproblem statements.lineartimevariableplants. The shape is characteristic of the effects of parameter variations on the plant transfer function and is referred to as the planf femplafe. with brief developments of the basic ideas. QFT.in the plant. the design focuses on the variations of the closedloop.The idea is to design loop compensation and a prefilter so that the inputoutput transfer function remains between these bounding responses for allpossibleplantparametervariations. gain due to plant Parameter variations.(Disturbancerejectionrequirementscan be handled similarly.) Since the loop compensation and a prefilter can beimplemented with negligible uncertainty. the major contributor to the theory. This set is defined by upper. uses prefilters and loop compensation to provide the desired closedloop responses.) The compensator transfer hnction at each frequency is defined by shifting the plant template to' a proper location. The treatment hereis cursory. The methods discussed in thischapteruselineartimeinvariablecompensatorsandproducelinearcontrollaw which is optimal according to some performance index. The design beginsby determining an acceptable set of inputoutput transfer functions which satisfy the tracking performance requirements. (Some of his contributions to control theory have already been reflected in this book. the gain curves indicate whether the variations in MI 245 . The tolerances ai arethegains Oi. and provides sufficient stability margins.1 QFT The term Quantitative feedback fheory(QFT) has been coined by Isaac Horowitz.Chapter 8 INTRODUCTION TO ALTERNATIVE METHODS OF CONTROLLER DESIGN I This chapter surveys several important design methods and compares them with the classical Bode approach presented in the previous chapters. The QFT theorists aim to extend the Bode methods to handle performance issues more precisely. QFT relies on simplified relationships between the frequency and time domains. With the template in a particular location on the Nichols chart.1.and nonlinear problems.and lowerbounding frequency responses. Withthe allowableparametervariations. (The actuator is included.itisfirstnecessarytocalculatetheplanttransfer function for allpossibleparametervaluesateach of thefrequencies c u i . Most of these issues have already been addressed in the previous chapters of this book. 8. These alternative methods may be encountered in industry. For simplicity we consider the QFT design of a singleloop tracking system. the compensation takes the following form: at each frequency variation in the closedloop gain should not exceed ai dB for all possible plants defined by theuncertaintyranges of theplantparameters. and the basis for. spanned by the upper. considers sensor noise issues and actuator nonlinearities.) QFT is a frequencydomain design methodology which considers Bode methods to be a part of. 8. The QFT specification for the design of the loop of a certain set. andtheyaugmentthemwithadditionalformalizations.
the original gain variation 6 isdB. 1plane / 180° / phase Fig. There is a continuum of such shifted templates which satisfy the design requirement.asshown by thedashedlines in Fig. 8.2. the template is shifted until the difference betweentheminimumandmaximumgainisexactly a i . 8. and the edges or corners of the shifted templates with minimum closedloop gain form the minimum performance boundary B(UJ as shown in Fig.1 Plant templates on the Nichols chart B(oi ) forming the minimum performance boundary For each of the frequenciesq at which the system requirements are specified. the boundaries B(UJ mustbeplottedontheLplane.1. From the lines on the Nichols chart it is evident how the template must be shifted.2 Boundaries on the Lplane An additional highfrequency Lplane bound is included to guarantee system stability and robustness. If not. Suppose that the tolerancea i is= 1dB. 8. Intheexampleshownin Fig. 8. the next step is to search for a .246 Chapter 8. 8. Methods Alternative satisfy the QFT'design requirement. dB 15 10 20 15 10 5 0 5 0 5 100 20° 5 1 0 0 0 30° 40° 50° 60° 70° 80° 10 90° Fig.1. With the boundaries in place.
An elaborate set of rules exists for constructing the root loci from the openloop transfer function. At frequencies near the zerodB crossing and higher.the prefilter's uncertainty contributionis negligible.gain coefficient and to design additional loop compensation. the root locusanalysis is usually performed by computer. making the entire system better and cheaper to manufacture.) It can be shown that a solution to the QFT problem always exists.) The QFT design is focusedonsatisfyingtheperformancespecifications for the worstcaseplantparameters.2 Root locus and pole placement methods Another category of controller design methods focuses on the location in the complex plane of the roots of the closedloop transfer function. The design is performed by trial and error or by using specialized software. Generally. QFT methods have been developed to handle 'the design of MIMO systems with stable and unstable. Finally. although the resulting feedback bandwidth may be unacceptably large. The QFTdesigned system is not the best possible. Today. We prefertheBodeapproach for thefollowingreason:the cost differential between a substandard controller and the very best available is generally insignificant compared with the cost of the system.the compensator gainis shaped to follow the stability boundary. QFT design is far more complex than Bode design.Methods Alternative 247 rational compensator transfer function such that the loop gain at each of the frequencies C O i will be just over the minimum performance boundary. It makes little sense to lower the system performance just to reduce the feedback bandwidth. The Bode approach identifies the constraints on the bandwidth upfront. (Whatever the philosophical difference. Improving the control law might also relieve some of the requirements on other components of the device. the best design is taken as that which has the smallest feedback bandwidth while satisfying the'minimum performance and stability boundaries. and highfrequency noise. 8. timeinvariable and timevariable. * With the multiple templates to be calculated and plotted. . root locitochoosetheconstant The root4ocffs method usesplotsofthe multiplier of the loop . (Remember. but rather is just good enough to satisfy the closedloop response and disturbance rejection specifications. The QFT design philosophydeparts from the Bode approach in the following areas: The Bode approach is to maximize the performance (feedback) while satisfying the loop asymptote due to plant features and constraints on the highfrequency parameter variations.Chapter 8. and perhaps a few days of workby the control engineer (if he uses the Bode approach). The controllers differ only by several resistors and capacitors or a few lines of code. QFT pursues the inverse problem of providing the minimum acceptable performance while minimizing the feedback bandwidth.andneglectsoptimizingthenominalperformance. a prefilter is synthesized which corrects the inputoutput response to achieve the desired response. rather than minimizing the bandwidth and determining later whether it is still too high. linear and nonlinear plants. This may in turn affect decisions made about the development of the next generation of the system. This may or may not be an advantage. there is little doubt that engineers well trained in QFT can resolve these tradeoffs and design highperformance controllers.
Theguardpointgainstability margin is apparent. In addition. other system poles must be monitored for stability. The compensator design proceeds by trial and error.” meaning that the stepresponse of the closedloop system resembles the step response of a system with just this pole or these poles. a double pole at the origin and real poles at 30 and 100rad/sec.Finally.wheretherootlocuspassesveryclosetothe jwaxis but the system is quite robust. suppose that there is a variablegaincoefficient k intheloopwhichis gradually increased from 0 to 1. As the gain coefficient is increased.248 Chapter 8. The design goal is to move the dominant poles into areas on the splane with sufficient distance from the origin for thetransientresponsetobefast.not too close to the jwaxis) to prevent excessive overshooting. In fact. It would be instantly seen from the Bode diagram that the pole and the zero in thecompensator are inwrong places..Methods Alternative Example 1. the method is sometimes referred to as poleplacement. some for a feedback system poles cross over into the right halfplane. Consider a control system with plant P(s) = 100/s2. What are desirable pole locations? Usually the system is examined for a pole or pole pair which is “dominant.56) = 1ldB. The loci can be continued by increasing the gain coefficient past the nominal closedloop value Fig. A common choice is to place the closedloop poles in a Butterworth filter configuration. 8. Fig. (The possibility of a prefilter is not factored into the design.) Meanwhile. When the designer makes an a priori decision about the precise location of the system poles. andthephaseandgainguardpoint stability marginsare not balanced. andwith sufficient damping (Le.e.3 shows the root loci for our example. A major inadequacy of the rootlocus design method is that it does not allow the designer tojudge how close the system performance is to the best available. searching for compensation and a suitable loop gain which brings the closedloop poles into desirable locations on the splane.56. For the purpose of this analysis. the Bodeand Nyquist diagrams enable the designer to deal effectively with common nonlinearities. system performance in the nonlinear mode of operation is difficult to determine from the root loci. It is also not evident from the root locus whether the system is well designed. the poles move from their openloop positions to their closedloop positions.) Because of these . when k reaches 3. the poles of the system are just those of T = CAP. it is not.. i. (Aswe shall seein Chapters 9 1 1. actuator lOO/(s + 100) and lead compensator C(s) = 1O(s + 3)/(s + 30). Another problem is thecompletelack of visibilityintolowfrequencydisturbancerejection. although this label is often reserved for statevariable feedback control of MIMO systems as will be discussed below. A practical counterexample is an active RC notchfilter. With theloop open.The gain stability margin is therefore 20 log(3. Intheexample. The robustness of the system is still difficulttodeterminefromtherootloci. but the phase stability margin and even the guardpoint phase marginare not. It might be guessed that the distance of the roots from the joaxis would be a good indicator of system robustness. no convenient rules exist for designing good highorder compensators. butthisisnotalwaysthecase. 8.3 Root loci of 1.
3 Statespace methods and fullstate feedback From the classical control perspective. where n is the order of the combined actuator/plant transfer function. u is the input orcontfol vector. The controlinput matrix B wouldbe a column matrix of length n which distributes the scalar control input among the state derivatives. In statespace formulation the control states: u=Kx.13 and 4.e. i. and to conform to the convention adopted by MA'JLAB. Laplace transforms.. we can allow the control to affect the output directly by introducingthematrix D and . The following state spacesystem description is standard: i= Ax+ B ( u + r ) . It may be helpful to think about how a SISO system would fit into this format. The system matrix A would be n x n . but depends on the choice of states. the root locus method is not recommended for control system design. Note that the representation is not unique. a twoinput threeoutput system would have a controlinput matrix B that isn X 2 and an output x n. The system of equations can be transformed into a set of firstorder differential equations by introducing intermediate variables where necessary. Y'CX. Alternative Methods 249 deficiencies.1) is added to the u is a linear combination of the state vector. matrix C that is 3 The feedback loops are closed when the second component in (8. (8. root locus plots make very impressive presentations for highorder systems that have been already designed well using other methods. (8.5). dynamics of the actuatorand plant). r isthe reference. It is customary to try to choose states that correspond to some physical variable of the system.. Also. the linear control system is a block diagram of transfer functions. and on the nominimumphase lag in the link composed of several parallel minimumphase links (see Sections 3.Chapter 8.and C is the output matrix.. An advantage of the statespace notation is that it is easily generalized to multiinput multioutput systems by changing the matrix dimensions. X=(ABK)X+B~. The root locus method can be valuable for the analysis of the effects of certain parameter variations on stability.4) (8.e. For example. (8. B is the controlinput mafflx. Thesquarematrix A is referred to as the system matrix It describes the dynamics of the system without feedback (i. The output matrix C would be a row matrix of length n which reassembles the scalar output (which is a function of time) from the states. and y isthe output vector. 8.3) The closedloop system is then described by the equations where K is the gain matrix.5) To be more general.1) where x is a vector (column)of statevariables (or states). Of course the system can also be represented by one or several linear differential equations.
C.c.=[o 0 1 0 ] . is unity).d. The state vector consistsof a positionlikeand a velocitylike state x = [x1 ~ 2 1 Per ~ .4 Statespace block diagram of a feedback system (andthattheactuatortransferfunction represented as foilows: y=+ XI = x2.10) (8. The following command would then produce the openloop fiequency response plot: bode(a. and D matrices for our example into MATLAB. and D are as follows: . both scalars (functions of time). Suppose that we had manually entered the A. B.Theopenloopsystemcouldbe (8. Alternative Methods 1 "B* control matrix svstem matrix ~ A [ output matrix C II I Fig. and A.13) The nomenclature is important since systems are represented this way in MATLAB.12) Here u is the input and y is the output.Chapter 8. (8.l) . 8.=[~] C=[lO] D0 (8. B. our notation. C. n = 2.b.11) x2 = u .
For instance. The feedback bandwidth in these systems is typically limited by the sensor quantization noise.This mayseemlike a lot of overheadto calculate the frequency response of a doubleintegrator.5 hasbeenemployedforcontrolof position x (or the attitude angle) in many space systems. or by making the appropriate connections in the SIMULINK block diagram and rerunning linmod. A workaround is to expand the state vector to include some of the compensator dynamics. the acceleration is proportional to the actuator force (or torque. B. and instead focuses on matrix algebra. for attitude control).4) isoftenreferredtoas fullsfafe feedback. this is discussed in the next section. the actuator transfer function A is a constant. 8. Example 2. as is typically done to add integral control in the statespace version of the PID.and the plantP is seen as a double integrator. This draws the designer’s attention away from the physical elements of the control system. The missing states must be estimated using the available ones.4) restricts the compensator function to consist of a single unrealizable zero: C(s) = kl Ikzs. The plant is considered rigid. and D matJlices for further analysis: [a b c dl = linrnod(‘xnodelname’) After the gains kl and kz are chosen. the blockdiagramoriented SIMULINK hasafunction linrnod. The blockdiagram in Fig.) K. For thisreason (8. (We might already disagree with the practicality of such an approach since obtaining the desired closedloop response is not the only nor the main purpose of closedloop control in practical systems. An implicit assumption is that the states somehow available to be plugged into (8. especially those having rigidbody plants with small parameter uncertainty. (8. alongwith their limitations and imperfections.some implications of the Before we discuss the possible strategies for choosing x are statespace notation should benoted. Alternative Methods 25 1 The last argument is letting the MATLAB function know that we’re interested in the response of the output (all outputs in the general case) to the first input (the system in case hasonlyoneinputandoneoutput). A more insidious problem with the statespace approach is inherent to the representation of the systemby a setof linear matrix differential equations rather than a block diagram of transfer functions.4) and fed back to the input of the system.theorder of theactuatodplantcombinationexceedsthe numberofsensed outputs. . C. The statespace closedloop design problem is to choose the control matrix K to obtain the desired closedlooptransient response. Another feature of the statefeedback framework is that it does not allow compensators whose order exceeds the order of the actuator/plant. In our example above. In atypicalcontrol system. making fullstate feedback unrealistic. Fortunately. which createstheappropriate A. the closedloop frequency response can be obtained using MATLAB commands to manipulate the system matrices. The statevariable approach canbetovariousdegreesmixedwithconventional block diagram design methods. the matrices are usually created by other programs.Chapter 8.
from the noisy readings of the sensor. When the acceleration error is also 0. and the bandwidth of the acceleration loop is still wider. the output of C2is 0. and acceleration loops(a) and its SlSO equivalents (b) and (c) The position commandxcom. On the other hand. and La are map. .thethreeloopssubstantiallydiffer inbandwidth:the order (PD). but the higher order compensators are not easy to fit within this design. Alternative Methods Fig. velocity. a velocity source (a motor drivenby a driver with low output impedance) may improve the system accuracy. The torque source actuator (using a driverwith high output impedance) simplifies the analysis. especially when the plant is flexiblewith Coulomb friction.second.13). Comments on Example 2 . The transfer functions Lp. L. Still. the compensators’ order must be much higher than that of the PD compensators shown in the block diagram. first. 8. one loop at a time. are forwardedwithappropriategaincoefficientsto.and the velocity and acceleration commands obtained by differentiation. the signal at the actuator input 0. When the velocity error is also0. The errors are reduced by the three feedback loops.. and acceleration are formed by subtraction of the plant variable estimates from the signals arriving to the summing points. velocity.and acceleration summing points. The bandwidth of L. the compensators are typically lowand. second. The errors in position. The effects of saturation in the actuator limitthe useful bandwidthof the feedforward.respectively.5 Block diagram of a position control system with position. When the plant is flexible. The three feedback loops are coupled. since. It is seen that when the position error is 0. velocity. first. etc. is wider than the bandwidth of Lp and smaller than that of La. the filter distorts the output signal and. the design can be made by iterative adjustments. The filter cutofffrequencies must besufficientlylowtoextensivelyattenuate highfrequency sensor noise components.252 Chapter 8. the filter phase lag reduces the available feedback and the disturbance rejection. isThis control scheme canbe perceived as multivariable.. bandwidth of the velocity loop is wider than that of the position loop. A plant estimator (filter) generates the plant variable estimates X E . position. but not too low since. The differentiators in the feedforward paths are implemented in practice as lead links whose frequency responses approximate the response of the ideal differentiator over the required frequency band. and the phaselagresponses are relatedtothegainresponses by Bodeintegral (3. the output of the compensator Cl is 0.
Therefore.4 LQR and LQG The general plan of the so called modern control fheoryis to take the statespace description of the control system literally. The diagram in Fig.It is assumed that the desired state is x = 0. i. The complex diagram in Fig.Chapter 253 8.6 Singleloop equivalents (a) and (b) of the block diagram shown in Fig. i. 8.6(b) which follows the diagram in Fig. and then find the gain matrix K which is optimalforthisindex.e. the matrix R penalizes the control effort. Methods Alternative The controller can be augmented with inclusion of nonlinear links to reduce the windup and to improve the transient response for largelevel commands. This system is multivariable only fomally since it has only one sensor.1 (the loop transfer function about the plant is the same in these diagrams.6(b) includes only two independent linear links whose transfer functions are defined by the designer: the feedback compensator and the feedforward path. 8..*"C&G Fig.. 1L""""".5 can be equivalently transformed into the diagram in Fig. 8. performanceof the system shown in Fig. 2. but the initial condition is nonzero. Similarly. 8. so the matrix Q penalizes the state error in a meansquaie sense. Since the position.e.6(a) and further into the diagram in Fig.One suchapproachistominimizeaquadratic functional J of the state and control history for the system's step response: (8. set up some scalar performance index which quantifies the desirable features of the closedloop system. and acceleration have unique and simple interrelations. rate. is the same). .5 8. 8.5 cannot be superior to a conventional welldesigned system with a prefilter or a feedforward path. the'plant is rigid and the actuator is a force source. 8. and the inputoutput transfer function without the feedback.14) where the matrices Q and R are welghfing matrices. such a system can be equivalently and better described as a singleloopSISO system. with the sensor transfer function S = 0. 8.
since to achieve the specified performance. LQWLQGregulatortheory wasintendedtoresolvethetradeoffbetweenthe l u . Note that w and v the plant model is referred to are generallyvectors. or robustne. . theentirefuture of thestate can bepredictedwithcertainty. and that the actual plant differsfromtheplantmodel byan additional whitenoiseinput. methodology alone.Ananalyticallytfactableapproachisto assume that the measurements are corrupted by white noise. fullstate feedback is not practical. The onlyreasonablepossibilityistojudiciouslychoosethe weighting matrices Q and R.xE.the combined approachis referred to as an LQWLQG regulator. A common design strategy is to increase the control penalty matrix R until . The state estimate xE is to be propagated as and H isthe (8. and as LQR.1’7) where is the estimator gain matrix. (8. ( z H X ~ ) . the missing states must be continually estimated from the available measurements. In fact.ss to plant parametervariations. so thatthesystemdescription becomes X=Ax+Bu+Gw z = Hx+v. This is because the featuresof the control system which constrain the performance are not captured in the LQR framework. This estimatoris referred to asthe linearquadratic Gaussian. Suppose that the available measurements are linear combinations of the state variables. the remaining states can usually be reconstructed by repeated differentiation.Chapter 254 8. + ~ u ~+ .Thenoiseinthe measurements is referred to as the sensor noise and denoted w . If the LQR framework is to be used for practical problems. or LQG. To makethe estimationprocessnontrivial.16) measurement iE: =AX. where G isthe plantnoisedistributionmatrix matrix. Given the secondorder moments of the white processes w and v . Let themeasurementsbe z. The noise added to as the process noise.the largest expected transient does not result in saturation of the actuator. There has been no mentionof actuator saturation. The gain matrix K which minimizes J can be found by solving a matrix Riccati equation. As mentioned previously. and the plant model is perfect. The implications for system design are disastrous. If the measurements are perfect. The resulting controller is known as the linear quadratic regulator.Methods Alternative limits the control signals’ magnitude.15) (8. Whentheseestimates are usedinconjunctionwithanLQRcontroller. the methodology is referred to Although software is readily available to solve the matrix Riccati equation and thus determine the optimal gain matrix K . The inabilitytoaddressnonlinearities intheLQRframeworkiscrippling.and denoted v. andthenexaminetheresultingcontrolsystemusingclassicalfrequencydomain analysis. the actuator will be oversized to maintain linearity. This seems wise since actuator saturation can result inwindup or eveninstabilityforanLQRdesign.thestatespaceformulationhasto be augmented by introducing errors insensingandmodeling. run the LQR software to determine the “optimal” gain matrix K. the optimal estimator gain can be found which minimizes the mean square error in . it is not advisable to attempt to design a control system using the LQR. disturbance rejection. This generally entails several iterations.
the inputoutput formulations mean separating the system variables into the sets of local (internal for the blocks) and global (at the blocks' ports) variables.. who initiated the H .Chapter 8. As was exemplified in Chapter 7 with the twoports. norm is the limit on the magnitude of a vector in the Hilbert space. feedback control designmethod applies this norm to the closedloop .r.was as described in Chapter4.Mathematically. For the inputoutput formulations. and statespace models should came into picture only as internal models at the levelof computation and at the level of implementation of control systems. Zames. typically. Since this method by itself does not address the robustness issue. The computational aspectsof control system design have been already advanced to the degree whenthey cease to be critical for the design ofmost practical systems. often said that the processesof approximation in model building and obtaining statespace model do not commute [45].G method. and distribution of the available feedback over the functional bandwidth. The I f . and the quadratic norm is not appropriate for stability analysison the basisof the closed loop response. The addition of the loop transfer recovery (LTR) method to the LQG allows addressing the system robustness. instead of structuring the systems mathematically in sets of linear'matrices. the number of the global variables is much less then the number of the local variables.e. It solves in oneoperationthetwoproblemsthataresequentiallysolved withBode approach: maximization of the available feedback bandwidth withrelatedshapingoftheloop response over the frequency region of crossover frequency and higher. as aconglomerate of linear and nonlinear multiports interconnected via their ports.5 Et. G. I. inputoutput (black box) formulations is thepreferredframeworkforuncertain(practical)systemmodeling. method. andtheentiresystem. This response can be later modified with classical methods for better disturbance rejection at lower frequencies. The method is formulated such that it is directly applicable to multivariable control systems. and. building the system model and designing optimal compensators still presents a challenge. i. However. and linear matrix inequalities The statespace approach to control system design and the statespace performance indices are difficult to use during the conceptual design. The process of such design is however not simple.timedomainwiththe LQ. However. is an extension of the classical frequency domain design method. and allows adjusting the responses to provide the desired stability margins. LQG provides a loop response which is well shaped inthe crossover areaof the frequency band. linear black boxes can be described by the matrices of their transfer functionsandimpedances(mobilities). TheLTR method recalculates the frequency domain loop responses of thesystemdesignedinstatespace. 8. frequencydomain characterization arein many aspects more convenient than the timedomain ones.and are easier to accomplish with the inputoutput formulations. control system engineers should structure the systems as sets of physical blocks interconnected via ports. psynthesis. when the plant is known pretty well (say. The H. This norm is an extensionof the Chebyshev norm widely used in frequencydomain network synthesis.Methods Alternative 255 sensor noise and the disturbance rejection. it does not provide the best solutions tomost practical problems. In other words. with 1%accuracy) and the feedback bandwidth is limited by the sensor noise.
by using nonlinear controllers that will be studied in Chapters 913. It is required that with these links added. specified with weight functions. control and many other linear control and stability analysis problems can be formulated in terms of linear matrix inequalities (LMIs).The panalysis method introduces into the loop special links that imitate the plant uncertainty. design method is the method of linear control system design. method. in such diverse areas as combinatorial optimization. estimation. better. The H. The weight functions define at which frequencies disturbance rejection should be higher than that at other frequencies. method. the H. which is easy to do when the designed system is singleloop. A less conservative solution can been achieved with psynfhesis whichcombinesthe H. The weight functions should be calculated from the known disturbance spectral densities. Although it has long been recognizedthatLMIs are importantincontrol.Methods Alternative frequency responses from the disturbance sources to the system output. the nominal system should be still stable and perform well.it was onlywiththeadventofthe efficientalgorithms(based ontheinteriorpointmethods)thattheirpopularityhas increased inthe last few years. design andpanalysis in an iterative procedure.methodmayleadtoanoverlyconservativesystem. For the functional feedback bandwidth. and statistics. frequency responses of the disturbance rejection are first With H. the H. and by how much. design often results in Nyquiststable systems which are not absolutely stable and can burst into oscillation after the actuator becomes overloaded. or. The solutions to this stability problem are either making several iterations by relaxing disturbance rejection requirements and modifying the weight functions such that the resulting loop response be of absolutely stable type. Because of this. The LMI is the algebraic problem of finding a linear combination of a given set of symmetric matrices that is positive definite. The H . . Since it is not easy to properly shape the crossover area of the loop Bode diagram with H.256 Chapter 8. thenorm on F is nearly the same as the norm on 7 ' .It optimizes the system performance without paying special attention to the system global stability. Thenonlinearcontrollerdesignmethodsshouldbealso employed to further improve or optimize the system performance in the nonlinear state of operation when certain commands or disturbances overload the actuators. LMIs find applications outside of control.
Thefirsttypeusessensorreadings ofenvironmental parameters (temperature. time. and as explained in Chapter 2. The second method can be useful when the command profile is well known in advance. This identification provides most of the available benefits in the system performance.) and plant parameter dependencies on these environmental factors to correct the control law. the benefit of good plant identification.Whentheplantgaindrops. The third type uses the control loop response to specially generated pilot signals to correct the control law. that the plant is easier to identify in the frequency bands where the feedback in the main loop is not large. Correspondingly. feedback path.1.The improvedknowledge of theplantshould then be used for controller adaptation. Thefirstandthethirdtypes ofadaptationschemessubstantiallyimprovethe control when the plant changes at a much slower rate than the control processes.andforditheringsystems.thefeedbackandthe disturbancerejectionreduceandthesensitivity of theoutputtoplantparameter 257 . i. Increasing the feedback reduces both error components. and the pwJilter can be made closer to the ideal. Example 1. When plant parameter variations are large. The loop response with maximum available feedback bandwidth is the one when theplantgainislargest. and prefilter to reducethe output error. Adaptation schemes where the plant is identified first are called indirect.Theuncertaintycan be reduced by a p h t identification procedure which gives the plant transfer function estimation as P’. Assume the plant gain varies by 20 dB in a loop designed as shown in Fig. fb cfb. adaptation can significantly improve the system performance.Oneremedy is tousean adaptivecontrollawwhichchangesthetransferfunctionsofthecompensator.e. The feedback bandwidth is limited by sensor noise and/or plant resonances. adjustments of the compensator.the prefilter. 9. and the feedback path on the basis of accessible information about the plant. A brief description is provided for adaptive systems for flexible plants. On the basis of the information used for the adaptation.Chapter 9 I ADAPTIVE SYSTEMS Largeplantuncertaintyreducestheavailablefeedback..) Thesecondusestheplantresponsetothecommandor disturbances to correct the control law. the adaptive controllers can bedividedintothreetypes.Examplesof adaptive filters are described. the second component can be additionally reduced by using an appropriate prefilter. for disturbancerejectionandnoisereduction. If this is notthecase. (Obviously the dependencies must be known a priori.thesecondtypemayresult in asystemwithrapidlyvarying parameters whose stability analysis represents a formidable problem. pressure. (P’= P) is twofoZd: the feedback can be increased byspriate modifications to the compensator C.The rate and dynamics of the adaptation are defined by some adaptation law. 9. It is shown.1 Benefits of adaptation to the plant parameter variations Plantparameteruncertainlyimpairstheavailablefeedback. etc. The feedback system output error is contributed to by disturbance sources and by the plant parameter uncertainty.
the resulting loop gain uncertainty will be less than 0. if thesameremedyis contemplated . In thiscase. Adaptive variationsincrease. In the plant. lndirect adaptive algorithms do not use plant identification explicitly. Assume the loop in the previous example is designed as Nyquist stable to increase the feedback in the functional band. there exist many adaptive algorithms.258 ChapterSystems 9.compensation of the plant pole with a compensator zero . Fig.the plant identification needs to be Q times more accurate than in the previous example which might be difficult to accomplish. I. Therefore. not a real pole but a complex pole related to a plant flexible modewithhigh Q is uncertain.as the plant’s responses are smooth. The typical goal for such adaptive control is improvingtheresponsestothecommandsinthesystemwheretheplant parameters are varying slowly. and Dcoefficients of the controllers are adjusted correspondingly. Timedomain performance adjustment is most often used: the timeresponse parameters are measured and P.5 dB. a real pole position cango down by a factor of four from the initial position at 2f. To counteract this pole motion. the system can become unstable. This is an example of indirect adaptation. the plant gain needs to be monitored and the compensatorgainmodifiednot only tomaintainthedesiredlevel of disturbance rejection. after the plant identification. Example 3. To alleviatetheproblem.1 Loop gain responses with maximum and minimum plant gain Example 2. In the previous example. and not improving the higherfrequency disturbance rejection or the system tolerance to fast plant parameter variations. a zero can be introduced in the compensator transfer function and. The algorithms typically work well as long. Example 4.Fortuningthemost popular PID controllers. but merely tunethecompensatortoimprovetheclosedloopperformance. butalso just to preserve stability. The phase lag increases correspondingly and the system can burst into oscillation.plantgaincan bemonitoredandthe compensator gain coefficient continuously adjusted for the loop gain to remain equal to the loop gain of the case of the maximum plant gain. . The accuracy of the pole position identification need not be very high: if the zerotopole distance is within5% of the pole magnitude. 0. When the plant gain decreases. kept close to the pole position to cancelits effect.
and by the rate of the information processing. I Knowledge of how plant depends on environment Adaptation identification drivers Plant Fig. the adaptation is quasistatic.e. schedu/ing..which can be critical when the plant parameters vary rapidly . . 3. Therefore.andensuringthesystemstability requires increasing stability margins (and reducing the feedback) in the main control loop. 9.ChapterSystems 9. These systems. have several disagreeable features (aswe have seen in an example in Section 3.2. The plant is identified. 9.Such an adaptive cwtrol system is shown in Fig. Increasing the speedof the adaptationmakes the adaptationdynamic and leads to LTV systems with rapidlyvarying parameters. The rate of adaptation .12). the main loop can be designed asan LTI system. and the prefilter and the main loop links can be considered timeinvariant.2. In this case. 2. i. This is why the adaptation laws arecommonly designed as quasistatic. 9. Le. Commonly. Adaptive 259 9.Suchaprocessissometimescalled 7.3 Plant transfer function identification Fig. The pilot signal amplitude should be sufficiently small so as not to introduce substantial error in the system’s output. Using pibf signa/s generated and sent to the plant for identification purposes. generally.2 Prefilter and compensator regulation The rate of the plant identification is limitedby effects of noise and disturbances. during the system analysis. Using sensors of environmental variables and the known dependencies of the plant transfer function on these variables. Using the input and output signalsof the plant that are generated during normal operation of the control system as the resultof commands and disturbances. andthenthecompensatorand gain prefiltertransferfunctions are adjusted.An example of sucha system has been given in Section the environment varies much more slowly than the main control loop dynamics.4.is limited by the plant identification rate andalso by the system stability conditions. 9.3 illustrates three methods for the plant transfer function identification that are used with indirect adaptation: 1. Stability analysis of suchsystemspresentsaformidableproblem.. much slower than the processes in the main control loop.2 Static and dynamic adaptation Plantidentificationpresents noproblemsandcanbeperformedrapidlywhenthe dependence of the plant transfer function on the environmental variables is wellknown and the environmentalvariables are measuredaccurately. the main feedback loop dynamics can be assumed to be independent of the process of adaptation.
4 Flexible and n. Adaptive Systems Fig. Plant identification allows making a compensator whose transfer function has zeros to compensate for the unwanted plant poles. the frequencies where the feedback is large are relatively low which makes the plant identification process slow. from fb/4 up to 4fb. This bandwidth is importantforfeedbackbandwidthmaximization. the control ceases to be collocated and two poles in a row follow in the plant transfer function. In contrast. This limits the accuracy and speed of the adaptation. Athigher frequencies..4 (a) Plant with flexibility and (b) the uncompensated loop response . due to flexibility of the actuatortoplant shaft. however. For example. This increases the requirements on the sensors' sensitivity and accuracy.4.3 Adaptive feedback system with plant identification The plant is identified from its measured input and output signals. 9. flexible shaft Actuator o inertia 0 ~ Sensor (a) (b) Fig. the plant transfer function is that of a passivetwopoleimpedance.e.260 Chapter 9. Further reduction of the control errorcomponentsatthesefrequenciescanbeachievedmoreeconomically byusing nonlinear dynamic compensation. the problem of plant identification is wellconditioned over the frequency rangesof positive feedbackand small negative feedback. as explained in Chapters10 and 1 1. 9.p. and makes the identification complicated.Theplant identification over this frequency range provides tangible benefits and is relatively easy to implement since at these frequencies the pilot signal amplitudes can be larger and the measurements can be much faster. as shown in Fig. when thetorqueactuatorandtheangularvelocitysensorare collocated in a structural system with flexibility. Plant identification is illconditioned over bandwidth the of large feedback where the error frequency components are small and therefore the pilot signals must be very small. Also. The measurements are corrupted with the disturbance signals and sensor noise. plants Plant identification might especially improve the control when the plant is flexible and uncertain.andthephasevariationsdonotexceed 180". i. 9. 9.
thruster'spositionandthepropellantsloshinthe tank of a.7).p. as diagrammed in Fig. 9. 9. a filter with multiple narrow pass bands. The disturbances can also be compensated in a feedforward manner. plants.or some other adaptive algorithm.allowstheprovision of substantialfeedbackattheharmonics of a periodic of broadband feedback with the response shown disturbance as shown in Fig.12).p.5 Loop response adaptation for disturbance rejection Example 1. under the limitation of Bode integrals (3. (3. for example) can be removed from the main feedback loop by the addition of appropriate sensors for the plant identification.e. Adaptive Systems 26 1 For this application.. 9. instead Bokle by the dashed line. the feedback in both oftheresponsesbeinglimitedbythe integrals. The disturbance identification allows modifying the loop gain response to match the disturbance spectral density. the plant identification is less complicated and most beneficial over the frequency b&d where the feedback is either positive or small negative. 9..5 Disturbance and noise rejection The disturbance rejection can be improved by using adaptive loops within the compensator. Such a filter tunes bytrackingthedisturbancefrequency fa using a PLL. i. using frequency domain identification is also economical since the plant response needs to be identified only at higher frequencies. such as that of rotation and translation(duetothe. 9.5. within the bandwidth of positive feedback. disturbances /" Adaptation drivers f . spacecraft.7 Loop gain Bode diagrams . lag increases with frequency.6 Loop gain Bode diagram with adaptive equalizer systems for with different bandwidths Fig. N.p. and the best way to measure it is at higher frequencies. The n. l l Disturbance identification and signal processing Adaptive compensatorlactuator Fig.e.6. i. lag causedby parallel paths of the signal propagation. in the vicinity of the structural modes. Using an adaptive combf~/fer. We conclude that for the flexible plants and n.Chapter 9. 9. Fig. a frequencylock loop WL).
a sinusoidal pilot signal with the frequency in the crossover region and very small amplitude suffices to identify the plant . Or.Thisrequires increased rateof information transmissionby the pilot signals. 9. A spacecraft attitude control system using star tracker and gyro attitude sensors is shown in Fig. but when the camera turns in a direction where no bright stars exist. less accurate. and less reliable compared with acfive plant identification with specially generated pilot signals. The smallamplitude pilotsi.which are allfarfromoptimaltest signals for the plant identification. When several stars bright are inview. and the signaltonoise ratioof the tracker becomes excellent. the pilot signal can be generated in the system itself by selfoscillation. Adaptive Example 2 .Atanymoment.whenthefeedback bandwidth is limited by the gyro noise.  Startracker Kalman filter Gyro C. for the best performance. they are easily recognized when compared with the star map in the computer memory. The measurements are corruptedby the noise represented by star tracker depends on the the sources Ns.accurately. 9. there always is a tradeoff between the accuracy of the plant identification which increases with the level of the pilot signals. When only the constant multiplier of the gain coefficient of the plant is changing. and this change is slow. Therefore. andcommands. In the latter case. on the other hand.) Diagram (2). the shape of the loop response can be changed from (1) to (2) by an adaptation process. passiveidenlificalion relies on the feedback system’s responsestonoise.gnal can be superimposed on the command.(Thiskindofdiagramwasalsousedfor a spacecraft photocamera temperature control where the feedback bandwidth was limited by the quantizing noise of the temperature sensor.262 ChapterSystems 9. Diagram (1) in Fig. and the pilot signals must have larger amplitudes and/or broader spectrum. Consequently. A Kalman filler (a widely used algorithm minimizing the mean square error) performs this adaptation task continuously. Example 3.7 wasfoundoptimalfor a spacecraft attitude controlsystemwhichusesreactionwheelsinthetrackingregime. passive identificationof the plant is slower.thesignalfromthe particular stars that happen to be in its field of view. and the error the pilot signals themselves introduce at the system’s output. the bandwidth of the filter feedback path from the star tracker must be reduced. When pilot signals are used.whenthefeedback theoptimalfortheregimesofretargeting bandwidth needs to be wider and the gyro noise is of small importance.8.6 Pilot signals and dithering systems Without pilot signals.theadaptationmustbefaster. Highfrequency oscillation of small . Whentheplantchangesrapidly. + Fig. 9.8 Adaptive control system with a Kalman filter in the feedback path 9.distortions. is closer to of theattitudecontrol. NG. the tracker output is noisy and the attitude control in the should depend more on the gyro.
Adaptive 263 amplitude is called difhef. 9.. 9. In a conventional system.e. The attenuator varies the main loop gain so that the ditherlevelbecomesassmallasthereference. a provision of. the dither is selected by a bandpass filter. at this frequency the loop gain must be . This system has certain similarities with the automatic level control systemfor AM receivers described in Section 1.10 . 9. rectified. and compared with a reference. the loop gain at this frequency is 0 dB which allows an increase in the loop gainby these 8 to 10 dB. 9.9 Ditheringcontrolsystem The Nyquist and Bode diagrams for the system are shown in Fig. the error. In the dithering system. For this purpose. i. 3 times additional rejectionof the disturbances within the functional frequency range. dB dithering 0 conventional system frequency at which phase margin is 0 Fig. smoothedby a lowpass filter.Theditherloopcompensator C d implements the desired adaptation lawresultinginthedesiredtimeresponseofthe adaptation.10 Dithering and conventional control systems.10. is amplified and slowly and continuously regulates a variable attenuator in the forward path of the main feedback loop.8 to 10 dB to provide sufficient gain stability margin.ChapterSystems 9. Fig.9. amplified. (a) Nyquist diagrams and (b) Bode diagrams The dithering system oscillates at the frequency where the phase stability margin is zero. The difference between the measured dither and the reference. A dithering sysfernblock diagram is shown in Fig. The gain in the feedback loop needs to be adaptively adjusted in order for the amplitude of the oscillation to remain small. approximately.
and the second pair of balanced quadrature modulators returns the signal to the baseband. This system workswellwhentherequiredfrequencyresponsesarerelatively smooth.Theinputoutputfrequency responses can be varied by changing C(s) and using multifrequency carriers. 9. A transversal f i k r compensator is shown in Fig. also called linear filters. It consists of delay links z and variable gain links. Adaptive (In some systems. symmetrical regulators described in Chapter 7 can be used.Responses with sharperbends andresponseswithresonancemodeswould require using too many sections in the transversal filters.11. the dither signal reduces the loop gain by partially saturating the actuatornonlinearlink. Adaptivefilterscan be made with amplitude modulators. These adaptive filtersare particularly useful for rejection of periodic nonsinusoidal disturbances. Fig.) 9. .12. and reduced power available from the actuator to drive the plant. Two additional implementation examples are given below. Such a system can be classified as a selflearning system. larger error introduced by the dither. which are varied by some adaptation algorithm. the weights Wi. Example 1. The balanced quadrature modulators are used here to cancelmost of theparasiticintermodulationproducts.Thesesystems are simplersincetheyhaveneitheradither detector nor a variable attenuator. as for example in Fig. As the filters. with smaller range of adaptation.7 Adaptivefilters Adaptive controllers use linear links. but their performance is inferior.The first balanced pairof modulators (multipliers) with carriers in quadrature transfers the input signal from baseband to higher frequencies where the signal is filtered by filters C(s). 9.1 1 Transversal adaptive filter asa compensator The weights are adjusted by an adaptation algorithm (defining the dynamics of the adaptation loops) on the basis of the performance error produced by the performance estimator. With appropriate weights. 9.264 ChapterSystems 9. Example 2. any desired frequency response of the adaptive filter can be obtained.
Adaptive Systems 265 Fig.12 Adaptive filter with modulators . 9.Chapter 9.
output output output input output Fig.stability of linearizedsystems. lO.1 Characteristics of nonlinear nondynamic links: (a) soft and hard saturation. and (d) respectively. 10. and threeposition relay .conditionalstability.1 Nonlinearities of the actuator.I Chapter I O PROVISION OF GLOBAL STABILITY The actuator. These nonlinear links represent propertiesof typical actuators. The link has no memory.(b). 10. Fig. The Popov criterion is discussed and applied to control system analysis and design. Concepts are developedoflimitcycles..3 produces a logarithmic link.l(a). y(x). It is fully characterized by the inputoutput function. (c) dead zone and saturation. is about lo4. If the error is small. Nonlinear dynamics compensators (NDCs). The range of ii. feedback path. hard and smooth dead zone.2. 10. the input current ii. (b) soft and hard dead zone. and the range 266 . dead zone and saturation. = . and plant In a nondynamic link.global stability.i.2 Inverse operator Fig.are showninFig.3 Logarithmic amplifier Example 1. 10. When the feedback is large. the error e is small and input y + e approximately equals the signal fed back. (d) threeposition relay A nonlinear link y(x) can be placed in the feedback path to implement an inverse nonlinear operatorx@) as shown in Fig. Several examples of the inputoutput characteristics of nonlinear nondynamic links . the current value of the output variable depends only on the current value of the input variable andnot on its previous values. which ensure absolute stability without penalizing the available feedback. 10.(c). feedback path and plant nonlinearities are reviewed. 10. Putting an exponential link (using a semiconductor pn junction) in the feedback path of an opamp as shown in Fig.hard and smooth saturation. and absolute stability. are introduced.
40 dB even when the y is small and the feedback path coefficient is relatively small. Le. x and x' approach 0 when time increases indefinitely. the feedback bandwidth becomes 6.5. the feedback path gain coefficient increases of 2500.5(a). the feedback reaches 108 dB when iin is the largest.be smaller than. where x(($) is the signal time derivative.2 Types of selfoscillation The most frequently encountered oscillation x($) is periodic.4 (a) Robotic arm and those where the output nonlinearly (b) its forcetotorque ratio as depends on the current and the the function the of elbow angle previous valuesof the input variable. say. for the largest iin). 10. (a)(b).thesignalamplitudedecreases withtimeandthetrajectories approach the origin. For the error iin . variables output . When the oscillation is sinusoidal. Therefore. These are the sfafic can that nonlinearifies be characterized by the link's inputIn function.8 octaves smaller for the smallest To counteract this detrimental effect and to keep the functional feedback bandwidth wide enough when iin varies with time. Therefore. 10 dB/oct.S(b) depicts a stable secondorder system with some damping. In other for the trajectories. For elbow angle example. 68 dB. the trajectory is a circle.i to be small relative to iin .o depends on the output position.. 3 kinematic nonlinearifies. Starting with anyinitialcondition.10. words.Chapter 10. Provision of Global Stability 267 by a factor output voltageu is 4. They will be studied in Section 10.1. output with Q. the origin isthe sfafic affracfor . the forcetotorque ratio in a robotic arm driven by a rotational motor depends on the elbow angle as shown in10. when the inputsignalincreasesfromthemaximumtothe minimum. The unmarkettrajectoriesrelatethe informationabouttheshape of theoscillationbutnotaboutthefrequency of the oscillation. The phase plane (x. The stability conditions limit the feedback bandwidth when the loop gain is largest (Le. 1 1 .. liketheoscillation of differentamplitudesinasecondorder conservativesystemdepicted inFig. and (a) Dynamicnonlinearifies are Fig. lO.4 Fig. a nonlinear link with the characteristic approximately inverse to that in the feedback path can be installed in the forward path which will increase the loop gain and the feedback bandwidth for lowlevel signals. Twoexamples of plantnondynamicnonlinearitiesaretheheatradiationand turbulent liquid flow that were mentioned in Section 3. The trajectories on .. 10. assuming the loop Bode diagram slope is iin. and 13. 12.e. Fig.x').the plane reflect the time history of the process. Therefore. the relationbetweentheinputandthe b $2 rces. is shown in Fig. the feedback must not .7 and Chapters ~ 10.i.
for example. the limit cycle takes a rectangular shape as in Fig.Thistype of periodic oscillation is called limit cycle.6 Periodic oscillations with high harmonic content A nonlinear system might possess several limit cycles. 10. when the variable x needs to be constrainedby a e x e a. a nonlinear system with a limit cycle is shown. (b) dissipative second order system having a static attractor at the origin. and some initial conditions might lead to stability. (c) limit cycle for an oscillator of a sinusoidal signal. systems. 10. 10.6. The limit cycle and the basin of attraction of a feedback loop which is standalone unstable can be modified with linear and nonlinear compensators such as to facilitate stabilizing the system with other feedback loops (see an example in A1 3. the signal is triangular as in Fig. x($) is rich in high harmonics. with the ratex ’ changing by instant jumps. in the oscillations which are shown in Fig.5(d). This circular limit cycle describes sinusoidal oscillation. Because of this. 10. and thex timehistory is derived from the trajectories on the phase plane. lOS(c).268 Chapter 10. . Fig. the swifchlng lines x = a and x = a drawn on the phaseplane define the conditions for the relay to switch.6(a). as. each surrounded with its own basin of attraction in the phase space.althoughaperiodicoscillations do happeninsomeengineeringfeedback VCOs. for example. we will not consider itin thisbook. Provision Global Stability of Fig. for: (a) nondissipative secondorder system. systems suchas in improperly designed microwave Phase plane can be utilized for the design of loworder relay control.5 Trajectories on the phase plane. 10. In many systems. Aperiodic oscillation usually does not arise in control systems of moderate complexity near the border of global stability.In Fig. (d) limit cycle for an oscillator of a triangular signal Selfoscillation in physical nonconservative nonlinear systems is initially aperiodic butofienasymptoticallyapproachesperiodicity.For example.13). When. Initial conditions within the basin of attractionleadtothislimitcycle.Differentinitialconditionsleadtodifferentlimit cycles.
7(a).. Assume that it is also known that on the falling branch the device is stable if connected to a current source. The system is locally unstable when thecontourimpedancewhichis Z(u.+ bo m ' with the polynomial coefficients functions of u. Provision of Global Stability 269 10. coefficient a. bo.3.. i u nonlinear. 10. pass through zero at the bifurcation point .. or the output of an opamp with inphase current feedback) with an external dc bias voltage source u as shown in Fig.7(b) into the falling branch.on the basis of the system parameters linearized for small increments. therefore.1 Local linearization The First Lyapunov Methodfor stability analysis is applicable to nonlinear systems with differentiable characteristics. I Fig. the locally linearized impedance of the twopole Z(U. can be used to determine the system local stability.e. Assume it is known from experiments that outside the falling branch such a device is passive.e.. 10. and therefore the coeffkients a. i.. s) = ams + am. neon lamp. /oc~//Y.3 Stability analysis of nonlinear systems 10. 10. bnsn + b.7 Nonlinear twopole device (a) with Stype currentonvoltage dependence (b) According to the First Lyapunov Method. The differential dc resistance of the twopole is As the bias point moves along the characteristic in Fig. and b. twopole Y r( branch bifurcation points . 10. i... The differential dc resistanceduldi depends on u as shown in Fig.+ a. s) has a zero in the right halfplane of s. Z(s) has no poles in the right halfplane..is negative on the falling branchbetween the two bifurcation points. a.s""'+ ..7(b). are positive. Lyapunov proved that stability of the system equilibrium can be determined. the resistance and.sm1 +. Consider a nonlinear twopole (an electrical arc. The resistance .Chapter 10. Example 1.
Findingan appropriate Lyapunov functionis simple when the system is loworder and the stability margins are wide..eachvariableapproacheszero and Fig. are called vanishing. One such criterion is the Second Lyapunov Method.8 Lyapunov the is AGS. An AGS system remains stable following all possible initial conditions. typically. This necessitates devising convenient practical global stability criteria. as opposed to a conditiional/y stable system.270 Chapter 10. em 10. which is stable following some initial conditions butinwhichsomeotherinitialconditionstriggerinstability.remainwithinan envelope whose upper and lower boundaries asymptotically approach zero.Therefore. and the system is unstable.e. For a system where the deviations can be. Provision of Global Stability and both become negative on the falling branch. are determined by the highfrequency behaviorof the circuit (bystrayinductancesandcapacitances). 10. but after being overloaded.4 Absolutestability Many practical feedback systems consist of a linear link T(s)and a nonlinear link (actuator) that can be well approximated by a memoriless (i..The function V continuously decreases with time and approaches theorigin. finding the function for a practical nonlinear system with ahighperformancecontrolleris. when thetimeincreases.the contour impedance has a zero in the right halfplane of s. while the signal deviations from the solution are small.big.the NyquistandBodemethodsof stability analysis to determine whether a nonlinear system is stable locally. With a m positive and a.e. In particular. As theresult. negative. 2 Globalstability Signals that are initially finite and then. difficult. The FirstLyapunovMethod justifies using. the locallyapplied Nyquistand Bode stability conditions are necessary but not sufficient.e. i. It uses the socalled Lyapunov function which is a scalar function of thesystemcoordinates. However. k >1 are commonly used for the envelope boundaries. positive and nonzero) outside of the origin as illustrated inFig.isequal tozeroatthe origin. a Nyquiststable systemcommonly is stable when first switched on.3.Welldesignedcontrol systems must be AGS. the numeratorof 2: must have a positive real root..a m and b. andhasnegativetimederivative. is positive definite (i. 1 0 . 3 . On the other hand. 10. when thetwopoleisshorted(connectedtoavoltagesource). Asymptotic stability with such boundaries is called exponential stability. The systemiscalled asymptotically globally stable (AGS) if its responses are vanishing after any vanishing excitation.. andremain positive onthefalling branch. example function The Lyapunov function is often constructed as a sum of a quadratic form of the system variables and an integral of a nonlinear static function reflecting the system nonlinearity.8 for the case oftwo coordinatevariables. can become unstable. Unfortunately. Timeexponents be’. to directly test whether a system is AGS one would be required to try an infinite number of different vanishing signals which is not feasible. nondynamic) nonlinear . We will usethis result in Section 12.
. The system isAGS.9 This systemissaidtobe absolufely sfable (AS) if it is AGS with any characteristic v(e) constrained by 0 < v(e)/e c 1 (10. I. Since the flux has the same sign as the current. where q is somepositive coefficient and $(i)/i > 1 as shown in Fig. Theresistor andthetwopole Z(s)* drawand dissipate the energy stored in the inductor. The Popov criterion can be understood through the mathematical analogy between the feedback system and the connection of electrical twopoles (refer to Section 2.i. and the threelevel relay belong to the class of nonlinear characteristics defined by (10. 10.10).5. Let us connect this inductbr. The power dissipated ui = i($(i) .9. the energy qi$(t)/2 stored in the inductor is positive. Provision of Global Stability 27 1 link v(e). Consider also a nonlinear resistor with the dependence of voltage on current u = $(i) . 3(a). as shown in Fig.1 l(b). Hard and soft saturation.5 Popov criterion 10.ThePopovcriterioncan bereadilyappliedtothe systems definedby plots of their frequency responses. Since this energy cannot become negative. dead zone.Chapter 10. and is instrumental in developing controllers with improved performance.1) as illustrated in Fig.i) is (a) in the resistor . Lurie with theSecondLyapunovMethod. and a passive linear twopole with impedance Z(s) (that is positive real) in thecontourshown in Fig.10.e.1).10 Characteristicof the nonlinear linkin Fig. the energy cannot be overdrawn from the magnetic field. 10.1 Analogy to passive twopoles’ connection The cciterion due to V. M. 10. Popov is based on the Parseval’s theorem and is equivalent to the absolute stability criterion that has been previously obtained by A. Fig.10.10. Consider a nonlinear inductorwhich in response to currenti creates magnetic flux q$(i). the current i decays with time and approaches 0. Fig. 10.9 Feedback system with a nonlinear link Fig. i. 10. positive for all I. 10. .1 1 (a) Function @(r) and (b) an AGS circuit this resistor.
1 l(b) is.r. a feedback system consisting of a nonlinear link v and the linear link T(s)is AGS if T(s)is representable in the form (10.L U . 10.3) Therefore. The return ratioof the linear links of the loop is T(s)= [Z(s) . the system is AS i f there exists a positive q such that the expression Z(S) = (1 (10. d$(i) dt Using Laplace transforms U = L u and I = L i.1). Next. the diagram can be converted ’to that in (b). (10. according to the equations related to the upper branch. In other words.12 AGS equivalent feedback systems Since the link 1 canbeviewedasthefeedbackpathforthelink 1/Z(s). Z(s). the voltage u =$(i)i+q. 10.5) . The feedback system shownin Fig. 10.according to the Ohm’s equation for the the lower branch. Fig.272 Chapter 10. The diagram in (c) is equivalently redrawn with conventional clockwise signal transmission in (d). and at all frequencies Re [(l +jqa)T(@)]> 1. U = LCp(i). Notice that such v satisfy condition (10. This means that T(s)should have no poles in the right halfplane. 10. 10.12. Provision of Global Stability The voltage u in the circuit in Fig. is I = .12(d) is described by the same equations as the passivecircuit in Fig.I . reversing the direction of the signal propagation generates the diagram in (c) where the function v is the inverseof C p . The current.I + qs L$(i) = (1 + qs) L$(i) .1 1 and is therefore AGS. as) These equations describe the block diagram (a) in Fig. (10.3) with any p.4) + qs)T(s)+ 1 U o s i t i v e real ( p d .1341 + qs) .
By multiplying the loop transfer function by the Popov factor 1 + qs with large q. so that the Popov factor's angle is nearly +90°. for the phase shift of the Popov factor(1 +jqo) frequencies on the rolloff.l3(a). At this frequency.5) is satisfied at all fi.theinequalityis not satisfied.equencies. and since T is lessthan1. The phase shiftin this system varies from 0 to 150' as shown in Fig. loog and there exists a positive q such that (10. The realpart of thisfunctionispositiveatallfrequencies. since does not affect the left side of (10. Provision of Global Stability 273 OPE The Popov absolute stability criterionfollows: if the system is stable system isAS.If a positive a can be found such that the Nyquist diagram for (1 + qs)lJs) stays to the r i g w t h e vertical line 1. the system i s AS. 10. Consider a bandpass system with a lowfrequency rolloff slope of 10dB/oct and with the associated phase shift of 150' as shown in Fig. r\ 1 Example 3.3. Fig.5) is not satisfied. the phase shift of the expression in the brackets in (10. judgment and no can passed be whether on the %plane system isAS. one might plot the Nyquist diagram for (1 + qs)T(s). we obtain the function (1 + qs)T(s) with the angle from + 9 0 ' to 60'.13 Loop responsesfor (a) an AS system and(b) for a system for whichAS cannot be proved with the Popov criterion Example 2. 10. The is within the 0 to 9 0 ' limits.14 Nyquiststable system . q in Fig. Instead of the Nyquist diagram.14. and (10.14(b).This system therefore falls into the gap between the Nyquist and the Popov criteria.5) is within the 150" to 240' limits. Consider a system with a loop gain response which is flat at lower frequencies and has a highfrequency cutoff with 10 dB/oct slope. T must be real and less than 1 at some frequency as shown T isreal. Hence.& To check whether a system satisfies the Popov criterion. 10. Example 1. 1 l. In a Nyquiststable system. dB degr dB degr 0 0 180 Fig. equivalent Bode diagrams can be used.5) is satisfied. The Popovcriterion(which is sufficient but not necessary for AS) is more restrictive than the Nyquist criterion (which is necessary not but sufficient forAS).Chapter 10. Nyquiststable systems do not satisfy the Popov criterion. andtheabsolute stability condition (10. Therefore. 11.
The plot shown in Fig. The Popov criterion can be extended using more complex circuits than that in Fig. i.5) canbe rewritten as Re TGo). pefiodic oscillation with fundamentalfi cannot existin the system represented in Fig. 10. 10. .It passes through the point (1. The condition can be changed into an equivalent formwhich makes use of Bode diagrams.16. 10. This plot only guarantees the absence of certain types of oscillation.e.5. Therefore.274 Chapter 10. which is more convenient.8) is satisfied and the system is AS f a Popov line can be drawn entirely to the lei2 o f the Nyquist diagram on the modified dune. Adding qs to Z(s) does not change whether Z(s) is positive real since qjo is purely imaginary. the Nyquist diagram is vertically compressed at lower frequencies and expanded at higher frequencies.the system characterized by the plot shown in Fig.8) represents the P O ~ O line Fig.r.e. is not possible. For instance. (10.16 neither indicates that the system is stable nor that the system is unstable." (10. .12. at all frequencies larg[( 1 +jqo)F(jo)]l<n/2 .7) by plotting the Bode diagram for (1 +jqo)F(jo) and. Fig.. Therefore AGS can be established using the phasegain relations.g.7) is satisfied. oIm7). (10. " " " (10. W e r i o d i c selfoscillation can take place i f a Popov line can be It has been proven that drawn to the lefl of all the' points o f the modified Nyquist diagram that relate to the Fourier components of the oscillation. Provision of Global Stability 10.6) i. deciding whether (10.Condition(10.. When drawn on this plane.16 On the basis of this diagram.O) with slope l/q.a judgment onmodifiedNyquistplanecanbepassedthatalimitcyclewithfundamental f. but no judgment can be made about whether the system is AS.15 Popov line Fig.10. 10. The third form of the Popov criterion uses the modified Nyquist plane(Rel". (1 4qs)F(s) isp.2 Different formsof the Popov criterion Condition (10. inequality (10. For example. 10. 10.8) V shown in With the sign c replaced by = . From this follows the second form of the criterion: the system isAGS i f a real positive q exists such that (1 + qs)r(s)+ 1 + qs is p r .5) uses the real part of a function. e.qoIm T(jo)> 1 or oImT(jo)c ql[ReT(jw) + 11. circuitsincludingseveralvariousnonlineartwopoles and several .15 is AS.15.
6 Applications of Popov criterion Fig. Bandpass systemwith maximum feedback Considerabandpasssystem with the feedback maximized over a large relative frequency band.Chapter 10. 90".. arg(+jqo)T 1 varies the combinationof y = 0 and y ~ 0.5 only from n/2 to n/2.Satisfying this requirement dramatically reduces the maximum available feedback comparedwithusingthetypicalstabilitymargin ofy/6in . i. stability can be proved with the Popov criterion only when y12 1/12.18(a) and (b). Provision of Global Stability 275 is validwiththe passivetwopoles. 2 . the Popov sufficient stability criterion yll 80" and y 1 80°.with phase stability margin yn in the highfrequencycutoff region and yln in the lowfrequency rollup region. Making y 1 excessively large to satisfy the criterion imposes little impairment to the available feedback but requires an extension of the loop gain to lower frequencies.17 Application of Popov criterion to Bode optimal lowpass cutoff 1 0 . 6 . However. 10. exceed requires that the rollup and cutoff phase stability margins.7). as shown in Fig. in particular. Therefore.e. if l/q is chosen close to the mean square frequency of the passband. 10. the value of y1= 1/6 does well in practical design. It was proven.the bandpassoptimal . 10. It is seen that = satisfies condition (10. In the narrowband bandpass system.thatthecriterion function (1 +jwql)/(1 +jwqz) replacing thePopov factor (1 + 4s).
19. In the system shownin Fig.When the signal level is very large.The NDC can be built of linear and nondynamic nonlinear links. the compensator transfer function for small level signals is expressed as T'p/P.18 Bandpass system analysis with Popov criterion cutoff. For narrowband bandpass systems. Such systemsare not AS when the compensatorsare linear. the nonlinear link 1. Still. 10.v in the feedback path . The smaller the relative bandwidth. they can be made AS by using nonlinear dynamic compensafors (NDCS).v(e) represents a dead zone. Actuator Plant Fig.Forthe AS analysis. the return ratio in the NDCs local loop becomes G.because of thedeficientfilteringproperties of thePopovfactor (1 +j W .itcanbeassumedthe command is0. by 0 ) .19 Feedback system with nonlinear linkv in the actuator of the NDC and link 1 . regardless although this has not yet been proven theoretically. 10. Provision of Global Stability dB 0 Fig.7 Absolutely stable systems with nonlinear dynamic compensation 10.v. The rest of the links in theblockdiagram are linear. when v(e) is saturation. 10. 10. We denote by T'p the return ratio for the plant measured when the link v is replaced by 1 (and the link 1 . the smaller the feedback limited by the Popovcriterion. phase stability margins of n/6 bothathigherandlowerfiequencies of the relative bandwidth suffice for AGS of a system with a saturation link.7.276 %plane Chapter 10. As mentioned. Typically.v(e) in the local feedback of the NDC uses the same nonlinear functionv(e) as the nonlinear link of the actuator.Then. the Popov criterion is excessively restrictive.1 Nonlinear dynamic compensator Larger feedback is available in Nyquiststable systems. v(e) is a saturation link so that 1.
Provision of Global Stability 277 1 0 .9) (1 + G)(I + TE) . the diagram inFig. 10.20 shown in Fig. 10. (10.Chapter 10.11) states thatthe plant feedback is the product of the localfeedback in the NDCfor largelevel signals and the feedback in the equivalent system. with the same input signale and. using Mason's rule.10) From (10. 3 : Reduction'to equivalentsystem The diagram shown in Fig. The linear links within the dashed envelope form a composite linear link. the negative of the transfer function from the outputof the nonlinear link to its input is Tp G T E =.20 which contains only one nonlinear link v.19 depicts a system that has two identical nonlinear links v(e). 7 . 10. the plant feedback is 1 + Tp = I (10. lo). (10. Therefore. We denote the negative of its transfer functionas T E (equivalent return ratio). l+G is Given T E and Tp.2 1. the system can be modified equivalently into the one shown in Fig. For the sake of stability analysis. is further redrawn as Fig.21 Calculation of TE using two parallel paths and a loop tangent to both paths From this diagram. therefore. the areas of substantial positive feedbackin the G and TE loops should not overlap or else . the NDC linear link transfer function (10. the same output signal v. If T E satisfies the Popov criterion.11) Eq.20 Equivalently transformed system containing a single nonlinear link v To find the expression for TE.10. the system mustbe globally stable. 10. 10. r " " " " " " " " 1 r 1 """""""""""""""" I Fig.
22 where the actuator exhibits the same nonlinearity as the function v(e) in theNDC feedback path but for the ktimes larger signal. In the homing system diagrammed in Fig. correspondingly. and in the following Examples 13 we will use' systems with loworder links to illustrate such benefits of nonlinear dynamic compensation as making AS the system with rather steep Bode diagram. the dead zone in the feedback path of the NDC can be replaced approximately by a unity link. The lvDC forward path transfer function TpIP (a leadwithintegration compensator) and the plant return ratio €or smalllevel signals Tp are . but in a system with an NDC. and the system becomes a tracking system with the loop transfer function T P G+l (10. 10. Commonly. satisfies the Popov criterion.23. small.7.278 Chapter 10. in some cases. To achieve this.and thatTE.1 v(e) 111. ke actuator GPfT. or maximizing the stability margins while providing specified disturbance rejection response. and. .. crossoverfreauenq of TE(ja) must be either much smaller or much larger than the crossover fvequency of G(ja). Such a system has response without an overshoot if the guard point phase stability margin is 90" or so. improving the transient responses to large commands. For a largelevel signal.12) must be rather shallow over a rather wide frequency interval including the crossover frequency.. If positive feedback in each loop is substantial. 4.3 Design examples Withlowerordercompensation.thedisturbancerejection andthe benefits of NDC application decrease. 10. Examples with such responses will be demonstrated in the next section. Tp bust satis@ the Popov criterion.19 are also valid for the system shown in Fig. The equations derived for the block diagram in Fig. 10.. 10..22 Scaling down the nonlinear link in the NDC by a factor of k 10. there still remains some freedomin choosing the responses for G and TE.defined by (10. Still. they remain substantial. In these cases. We can see the advantage of a system with an NDC as follows: in a conventional system. and the plant P is a single integrator.12) and saturation in the actuator. b TdP e " rb k I I . there is no need to maximize to the full extent the disturbance rejection while keeping specified stability margins. v(e) is a saturation. the Bode diagram for (10.9). the only requirements are that Tp satisfies the weaker Nyquist criterion. Exploiting this extra design flexibility leads to better performanceas will be shown in the next section. I  Fig. Example 1.This freedom can be utilized for the provision of desired transient responses for large level signals. Global Provision Stability of the positive feedback in the plant will be excessive and the phase stability margin in the plant loop.
Chapter 10. Provision of Global Stability
Tp
279
"
 2( s + 0.5)
s(s+2) 2(s + 0.5) s2(s+2) '
P
Tp =
9
Fig. 10.23 Block diagram of a system with an NDC
The asymptotic Bode diagram for Tp shown in Fig. 10.24(a) is typical for simple lead compensation. The logarithmic Nyquist diagram forTp is plotted in Fig. 10.24(b) with nyqlog function from'Bodesteptoolbox described in Appendix A14:
np = [ 2 13; dp = [l 2 0 03; nyqlog(l,np,dp)
Nyquist diagram,x marksw = w b , + marksoctaves
dB 40
I
30 20
10 0
10
270 240 210 180 1
50 120
90
loop phase shift in degrees
Fig. 10.24 (a) Asymptotic Bode diagrams and (b) Lplane plots for Example 1(A)
In the linear mode of operation (for small signals), the guardpoint phase stability has overshoot. margins is 37" and the closedloop transient response 43% The general rules for selecting T E and G have been formulated in Section 10.7.2. Consider two choices of loworder TEand G.
(A) We ,willstart the designby a guessed response for TE:
TE=.
s(s
2
+ 2)
As shown in Fig. 10.24 his response merges with Tp at higher frequencies but has a shallower slope at lower frequencies and, correspondingly, lesser phase lag. system The with such a TEis AS (and even the processes in this system are stable; process stability is discussed in the next chapter). With such T E ,a from (10.10),
G=
s(s2
1
+ 2s + 2)
280
Chapter 10. Provision of Global Stability
The asymptotic Bode diagrams for these TEand G are shown in Fig. 10.24(a). The Lplane Nyquist diagrams are plotted in Fig. 10.24(b) with the following script (which includes the previously shownline for plottingTp):
np = [2 11; dp = [l 2 . 0 0 1 ; nyqlog(l,np,dp); hold on ne = 2; de = [1 2 01; nyqlog(l,ne,de); hold on ng = 1; dg = [l 2 2 01; nyqlog(0.5,ngtdg);zoom o f f gtext('TP'1; gtext('TE'1; gtext('G') %place labels with mouse
The transfer functionof the linkin the feedback path of the NDC is GP
"
s+2
_.
l;p
(s
+ 05)(s2 + 2s + 2) '
s3
The largeamplitude loop transfer function (10.12)
"
I;p
 2(s + 0.5)
+ 2s2 + 2s 2s3
_.
+5s2 + 6 s + 2
G+ sl 2(s+2 s) 3 + 2 s 2 + 2 s +s1 5+4s4+6s3+5s2+2s' The largeamplitudeclosedloopresponse(not shownhere) is dissimilartothe response of a Bessel filter; therefore, closedloop transient responses to large commands are not expectedto be acceptable.
(B)If, while retaining theNDC forward path link,we interchange the functionsfor G andTEmaking them TE =
s(s2
1
.+2s + 2)
, and
G =s(s
2
+ 2) '
the equations (10.9)  (10.11) are still satisfied since G and T E enter these equations symmetrically. The transfer function of the link in the feedback path of the NI>C GP "  2 I;. 2 s + 1 The stabilitymarginsforthis T E (i.e.,forGinFig.10.24) are smallerthanin version (A), but sufficient for the system to remain AS (although not processstable). The largesignal loop transfer function (10.12) is
" ' T P
2( s + 0.5) G +sl( s 2 + 2 s + 2 )
_.
*
This frequency response and the related closedloop response are plotted in Fig.10.25(a). The closedloopstepresponse is shown inFig.10.25(b);it hasno overshoot. These linear system responses indicates that the overshoot in the nonlinear regimes in response to large commands will be small or nonexistent.
Chapter 10. Provision of Global Stability
28 1
1o4
0
cn
W
10"
10"
10'
Frequency(radsec)
8
90
180 1OQ
1oo Frequency(radsec)
10" 10'
" 0
Time (sees).
10
20
(a)
(b)
Fig. 10.25 (a) Openloop and closedloop responses for large signals and (b) transient response to large step commands in Example 1
Indeed, the system transient responsesstep to commands obtained with SIMULINK and shown in Fig. 10.26 demonstrate NDC that improves the responses.
2
2
1.5 1 0.5 10
20
1.5 1 . A 0.5
0
1
10
20
0)
10
20
2
2
1.5 1 0.5 0
10
I
20
1.5 1 0.5 0
0.5 10
20
0.5
2
10
20
Fig. 10.26 Transient response to unity step command in Example 1, case (A): (a),(b),(c)  systems with threshold of saturation and dead zone set to, respectively, 0.1; 0.2; 0.5; (d) system without NDC; (e) windup in the system where threshold is set to 0.2, and the NDC feedback path disconnected; (f) actuator output for (a)
In the linear mode of operation (d), the overshoot is large, approximately 45%. In (a),(b),(c) the responses are shown for the systems with the unity step command and different saturation thresholds. The' larger the ratio of the command to the saturation threshold, the smaller the percentage of overshoot. Fig. 10.26(e) shows the windup responsewhich occurs when the feedback path in
282
Chapter 10.Global Stability Provision of
the NDC is disconnected and the actuator overloaded by a relatively large disturbance or command. The effects of the NDCs on the transient response for largelevel signals will also be considered in Chapter 13. Fig. 10.26(f) shows that the actuator output is constant over some time range and then rapidly drops to nearly zero. This indicates good utilization of the actuator output power capability. The system exhibits performance which is desirable formany homing systems: the absolutevalue of theovershootislimitedtoasmallvalue.Large(percentwise) overshoots occur in the responses to small disturbances, but not in the responses to large disturbances. The performance canbe further improved with higherorder compensation.
Example 2 . The systemblockdiagramisshowninFig. 10.27. The transfer function of the NI>C forward path and the plant smallsignal return ratio are
T p
P
rp
"
 2s2'+ 1.2s + 0.1
s3
+ 2s2
9
=
2s2 + 1.2s+ 0.1 s4 +2s3 ' 2(s + 0 . 5 ) ( ~ + 0.1) s3(s+2)
Fig. 10.27 Blockdiagram of a system with an NDC
1.e.,
r p=
As seen on the asymptotic Bode diagram in Fig. 10.28(a), ITpl is larger at lower fiequencies than that in Example 1 and therefore provides better disturbance rejection. The system is Type 3. The Nyquist diagram forTp is shown in Fig. 10.28(b), the system is Nyquiststable.
dB
50
1 8 dB/Oct
I , Tpin Example1
Nyquistdlagram, x marks w =w b , + marksoctams
Lv
270 240 210 180 150 120 90 loop phase shift in degrees
(a) (b) 2 Fig. 10.28 (a) Asymptotic Bode diagram and (b) Lplane plots for Example
Chapter 10. Provision of Global Stability
283
Let us consider three choices of TE and G, keeping them loworder.
(A) With the same
TE =s2
2
+ 2s
1.2s + 0.1
as in case A of Example 1, G is found from (10.10) as
G=
s2(s2
+ 2s + 2)
'
The asymptotic Bode diagrams for the TEand G are shown in Fig. 10.28(a). The Lplane plots are shown in Fig. 10.28(b). For largelevel signals, the system is Type 32 = 1 since the NDC local loop is Type 2. The NDC feedback path transfer function is GP Tp
"

(0.6s + 0.05)(s + 2)
(s2
+ 2s + 2)(s + O S ) @ + 0.1)
'
The system is globally stable and the stability margins in the equivalent system are large, butit does not qualify as a good homing system because the transient responses to largeamplitude signals (not shown here) are far from the best possible.
(B)With the functions T&) and G(s)interchanged, i.e.,
1.2s + 0.1
T E
= s2(s2 + 2s + 2) . .
%
and
G =
2
s(s + 2) '
the NDC feedback path becomes GP
s(s
2
s2 (s +.2)
Tp
+ 2) 2(s + O S ) ( $ + 0.1)

S
2( s + 0.5)(s
+ 0.1)

S
2s2 + 1.2s + 0.1 *
The type of the system for largelevel signals is 3  1 = 2. The transient responses for large magnitude step commands (not shown here) still deserve improvement.
(C) The numerator of the NDC feedback path transfer function is further adjusted by trial and error, in order to improve the transient response to a largeamplitude step command. WithNDC feedback path transfer function
GP Tp
"

1.6s + 0.16,
2s2 + 1.2s + 0.1 ' s3 + 2s2,+ 1.6s + 0.16
s3 + 2s2
s4
' 9
G=" GP T p T~ P
* P
 1.6s + 0.16
s3+ 2s2 '
G+1=
"
G +1

2s2 + 1.2s + 0.1
s(s3
+ 2s2 + 1.6s + 0.16)
'
Tp+l=.
+ 2s3 + 2s2 +,1.zs+ 0.1 s4 + 2s3
9
T E=
0 . 4+ ~ 1.04s ~ + 0.1
s4 + 2s3 + 1 . 6 + ~ 0.16s ~ '
284
Chapter 10. Global Provision Stability of
The large signal openloop and closedloop Bode diagrams are shown in Fig. 10.29(a). With such diagrams, the transient response shown in Fig. 10.29(b)has no overshoot, and the nonlinear system transient response shown in Fig. 10.30 becomes satisfactory. The Nyquist diagram for T ! is shown in Fig. 10.31, where the crossover frequency % = 0.65. The system is AS.
.
101
10 "
I o1
Frequency (radsec) .
"
0
El)
W
! i n
180
10" 1oo
1o1
Frequency(fadkc)
(a)
Fig. 10.29 (a) Bode diagrams and (b) transient response for
70
Tp/(G+1)[1+Tp/(G+I)
NNuist diagram, x marks w = w b , + marks octaves
2
1.5 1
60
50
0.5
10 20 Fig. 10.30 Transient response of the nonlinear system to large step commands in Example 2, case (C), obtained SIMULINK. with
*
*E . 30
8 20 8.
10 0 10
o /
0 7 ! 
240 210 150 180
1 2 0 90
phase loop
shift in degrees
Fig. 10.31 Nyquist diagram for
TE
Example 3. In Examples 1 and 2, the plant is Us. Let us next consider the system with a doubleintegrator plant. To retain the plant loop return ratio the same as before, let us multiply the NDC forward path transfer functions in Examples 1 and 2 bys, and to retain the same internal feedback in the NDC, let us multiply the feedback path by Us, The system with the l/s2 plant andthe modified NDC has the same G and TE, and is thereforeAS. However, since the NDC reduces the system's type for large signal amplitudesby the number of poles of G at dc, the output transient responses to largemagnitude step commands undershoot. Whengood largesignal transient responses must be assured in TE should be made steeper. Acceptable the nonlinear system, the Bode diagram for
Chapter 10. Provision of Global Stability
285
performance canbe obtained with
1 p="s2 *
"
T~ 2s3 + 1.2s2+ 0.1s P s3 +2s2 '
'
"
GP
T p

S
1 . 5+ ~1 ~. 5 + ~ 0.15s ~ + 0.002 '
Better transient responses can be achieved by making the dead zone in the NDC somewhat smaller than the actuator's threshold of saturation (although in this case the system analysiswith Popov criterion is no more valid, and the disturbance rejection for large signals will decrease,which might or might not be of importance). The presentedExamples 1,2, and3useexplicitalgebraiclowordertransfer functions. The followingthreeexamplesusegraphicalrepresentationforthe ,gain responses. The related phase responses can be calculated using either graphical methods given in Chapter4, or the programs from Appendices 4 or 14. After this, the responses for G andfor the compensatorTpIP can be approximated by rational functions.
fb = 8 Hz,
Example 4. Fig. 10.32 shows the diagrams for G + 1 as Bode optimal cutoff with 30" of phasestabilitymargin, and 10dBof amplitudestabilitymargin resulting in 40dB of feedback over the bandwidth [0,1]. T E is chosen with only lOdB 90". The over the functional bandwidth, and having the phase stability margin more than attained feedback ITp+ 1I of 50 dB exceeds the 40 dB available in the singleloop AS system. The system is Nyquiststable and AS.
dB
80
dB
60

8
f, log sc
.5
1
\ 2
20i
20
Flg. 10.32 diagrams Bode for the AS system of Example 4
Fig. 10.33 diagrams Bode for the AS system of Example 6
At lower frequencies the responses canbe reshaped, for example as shown by the dashed lines. The rational function approximation for the obtained responses can be found as described in Chapter 6.
Example 5. As indicated in Section 10.6, a bandpass transform of the lowpass Bode optimal cutoffwith a phase stabilitymargin smaller that 90" does not satisfythe Popov criterion. Application of an NDCresolves this problem. The bandpass transform of the Bode diagrams shown in Fig. 10.32 produces an AS bandpass system w i t h 50 dB of feedback. Example 6. The Bode plot for T E displayed in Fig. 10.33 is chosen as the Bode optimal cutoff with 40 dB of feedback and a 30" phase stability margin.The available feedback inthe AS system without the NDC 40dB. is
286
Chapter 10. Provision Global Stability of
A Nyquiststable local loop in the NDC is chosen to reduce the loop crossover frequency in order for the areaof positive feedback not to fall upon the areaof positive ' loop. The Nyquiststable plant feedback is80 dB in the operational feedback in the T frequency range,much larger than 40dB allowable in the AS system without an NDC. In Examples 4  6, no attention was paid to the closedloop transient responses. These responses, if desired, can be improved with the command feedforward technique or with additional nonlinear feedback loops. We can conclude that NDCs improve performanceof AS systems. Another approach to designing feedback systems with nonlinear dynamic compensation will be discussed in the next chapter.
10.8 Problems
(a) Make an amplifier equalizing the attenuation of a span of a coaxial cable (the cable attenuationin dB is proportional to the square root of the frequency) using a passivetwoportimitatingthecable;(b)makean N D converterfromaDIA converter; (c) make a voltagetofrequency converter from a frequencytovoltage converter. What is theinputoutputcharacteristicofafeedbacksystemwithahighgain forward path and a feedback path where (a) saturation (b) dead zoneis placed? Figure out the spectral density plots (Fourier transforms) for the signals in Fig. 10.6. Find the pointtopoint correspondence between the limit cycles in Fig. 10.5 and the timeresponsesfor x and x'.
~
Draw phase plane limit cycle curves for (a) a triangular symmetrical signal; (b) a triangular signal with different front and back slopes. What is the currenttovoltage characteristic of the output of an amplifier with inphase voltage feedback, if the amplifier gain coefficient is 20, the feedback voltage divider consists oftwo 10 WZ resistors, and the power supply voltage is 41OV? What external load resistance makes the circuit stable? Is the system with saturation AS if (a) T(s)= 1OO(s + 3)(s+ 6)/[(s + 1O)(s + 20O)sl; (b) T(s).= 100(~+ 2)(~+ 16) ~l O[) ( (S ~++ 20O)sl; (c) T(s)= 3 0 0 ( + ~ l)(s + 5)/[(~ + 1 0 0 ) (+ ~ 20O)sl; (d) T(s)= ~ O O S ( S + l ' ) (+ ~5 ) ( + ~ 5 0 0 0 ) / [ (+ ~100)(~ + 1 2 0 ) (+ ~ 130)]. (Draw asymptotic Bode diagrams or use computer generated Bode diagrams, and apply the Popov criterion.)
'
Are the systems from the previous problem AS when the saturation link is replaced by a dead zone link? By a threeposition relay?
A system is AS. Is it globally stable (yes, no, or no judgment can be passed) ifthe nonlinear link He) replaced is by nondynamic a link by the following equations:(a) v = e2/2; (b) e = $/2; (c) v = e3/2* (d) e = v = e  2/2; (f) e = v  e2/2;(9) e = 5v, (h) v = 5e; (I)e = v + 0.253; (j) e = (k) e = v/(v1) dver the range I vl e 1 , no solution outside this range.
Global Provision Chapter Stability 10. of
10 Draw the characteristic of the nonlinear link obtained by connecting unity link and the negative of a threeposition relay link.
287
in parallel a
11 Design an AS system with an NDC, with fb= 1 kHz, using the design prototype in Example1,Fig.10.24(a).MakeMATLAB or SPICEsimulations.Plottheoutput time responses to stepfunction input of small and large amplitudes. 12 Design an AS system with an NDC, with fb = 140 Hz, using the design prototype in Example2,Fig.10.24(b).MakeMATLABorSPICEsimulations.Plottheoutput time responses to stepfunction input of small and large amplitudes.
13 Using the block diagram shown in Fig. 10.22, scale down the saturation link for the NDC, when the plant saturation threshold is 20V, and the maximum output signal of the driver is V. 2
14 Whatisthemeaning of theword“asymptotic” in theterms:asymptoticBode diagram, AS, AGS. Can an asymptotic Bode diagram be used in determining AS?
15 Research project: DesignaNyquiststablesystemforaplantspecifiedbytheinstructor,withthe feedbackbandwidthspecifiedbytheinstructor,withsaturation in theactuator. Verify the design with SIMULINK. Apply a largeamplitude vanishing signal to the system’s input, and observe the following selfoscillation. Introduce a link with gain coefficient 100 in the loop after the compensator. Introduce local feedback about this link that makes the main loop response AStype. Introduce a dead zone into the local feedback path. 0 Transform equivalently this system to a singleloop system. Modify the linear links if necessary for the system to satisfy the Popov criterion. Study the responses to commands of differentshapes and amplitudes. Modify the linear linksif necessary to improve the responses. Comparethedisturbancerejectionofthedesignedsystemwiththedisturbance rejection ofanASsystemwithoutanNDCandwith (a) Bode step response or (b) PID response. 16 Researchproject:Dothedesignsimilartothatofthepreviousproblemfora system with a threeposition relay in the actuator and an appropriate nonlinear link in the compensator.
17 Research project: Do the same as in Problem 13 for the system with the characteristic ofthenonlinearlink in theactuatordescribedbytheequation e = v0.25@ and an appropriate nonlinear link in the compensator. 18 Research area: Design of nonlinear prefilters and feedforward paths complementing nonlinear feedback loops. 19 Research area: Design of NDCs with multiple nonlinear links.
20 All links in the system diagrammed in Fig. 10.34 are linear except the two identical links v. Which of the following conditions allows stability analysis with the Popov . criterion: (a) L 1 = L2; (b) L2 = L3; (c) r3 = 4;(d) L2 = 4;(e) L 1 = 2.74
Therefore. 10.6 6 The circuit is depicted in Fig. 20 The answer is (c) since in this case the input to both identical nonlinear links is the sameandtheoutputsare.thesame.35 Spectral density plots for oscillation shapes in Fig.forthepurposeof stability analysis. 10. uency frequency Fig. 10. Provision of Global Stability Fig. When the voltage exceeds VCC which is 10 V. The system is certainly stablewhen the load impedance is0 since this disconnects the feedback.7) or any ofthe other equivalent conditions. 10. .288 Chapter I O .therefore. The output impedance in the linear mode of operation can be calculated with Blackman’s formula asR = RoF() = (1 .36 (a) Amplifier with inphase voltage feedback and (b) its output currenttovoltage characteristic 7 (a) The system is AS.1.35.34 Feedback system Answers to selected problems 3 The sketches of the spectral density for the oscillations are shown in Fig. the amplifier is saturated. one of the nonlinear link can be removed.36(a). and the characteristic is shown in Fig. 10.36(b). and the output resistanceis Ro= 20 k d (weassumetheamplifieroutputimpedanceremains infinite in the linear mode of operation and while it is saturated). using MATLAB to verify (10. 10. 10k 3 ’ ’ I Fig. 10. as can be proved with q = 0. The negative impedance on the falling branch contains a positive real zero.10)20ks2 = 8 kd.
Isof and isoE responses are introduced. Since theharmonicsinteractinthenonlinearlinks. their stability is conditional when the only nonlinear links in the systems are the actuators’ saturation. Describing function (DF) stability analysis is simple and convenient. i. Simple approximate formulas are derived and used in4he design of nonlinear controllers. and hysteresis. So.1. Nonlinear interaction between local and common loops is reviewed. DFs are derived for most common nonlinear links: saturation.1 Harmonic balance analysis The condition for periodic selfoscillation (limit cycle) is unity transmission about the feedback loop. NDCs eliminate limit cycles in such systems.1 Harmonicbalance 11. A bangbang controller with a hysteresis link is described. This condition is called harmonic balance. AnNDC canbebuilt ofonenonlinearandseverallinearlinks.In this chapter. NDCs are neither expensive nor complex . For each of the harmonics. Although generally this procedure could be cumbersome. 11.cross section.an analog NDC can be made with an opamp and 5 to 10 passive elements. after passing about the loop. The effects of harmonics and intermodulation on the system stability and accuracy are discussed. NDCs are described with parallel nonlinear channels and with nonlinear local feedback paths.Conditionsare discussed for maximizing the phase advance for largelevel signals. the same signal must return to any initially chosen. It is known that SlSO and MlMO Nyquiststable systems provide larger feedback thus enhancing the disturbance rejection. harmonic balance analysis is simplified when the Fourier series can be 289 .. Stability margins for such systems are defined. each harmonic of the return signal is the same in amplitude and phaseas the harmonic of the signal at the beginning of the feedback loop. NDCs can be built in the same configurations as the NDCs designed for provision of global stability. A digitalNDC is trivial to program. However. dead zone. Nonlineardynamiccompensatorsarethelinksthatprovidephaseadvancefor largelevelsignals.thesuperpositionprinciple cannot be used here. NDCs also improve transient^ responses for large commandamplitudes. the transfer function’s equality to 1 should be verified in the presence of all other harmonics.For all these purposes. and the reader is warned about practical cases where such interaction can lead to a limit cycle.e. The chapter ends by describing the procedure for testing whether the system is AGS. The responses characterize the dependence of DF on the signal amplitude and frequency. Since periodic signals canbe presented in Fourier form. there is littleexcuseforloweringthesystem performance by using only linear controllers. We cannot consider a single harmonic in isolation fiom the others. A loop is analyzedwhichincludesanNDCandanadditional nonlinear link (actuator). resistors and capacitors.Chapter I I I DESCRIBING FUNCTIONS Stability analysis based on harmonic balance provides sufficient accuracy when applied to welldesigned control feedback loops having lowpass filter properties. threeposition relay.andreducetheeffectsofprocessinstability.
. the Fourier series for v((ut)is v(mt) = (4/. The feedback system diagrammed in Fig.]. the one sirnplifih the stability analvsis. The loop phase lag reaches at frequency 0. At the same time.aturation symmetry. the cross section at theinput to the nonlinear link is.1. Describing Functions justifiably truncated to a few terms.Because of this. Example 1. generally. 1 1. To establish whether this is possible. v(t) is sinusoidal. 10 . 11. This happens because v(t) is nearly doubly integrated by the loop linearlinks. e(t) anditsfirstandsecondderivatives are continuous. The selfoscillation initially grows exponentially until. Because of the s. 1 1 .1 contains a common type of plant withan integratorand two realpoles. the signale(t) at the input to the nonlinear link looks smooth. the signal e(t) is clipped in the saturation link andv(t) becomes nearly'trapezoidal.However. A selfoscillationtakesplace whenthe loopgaincoefficient exceeds 1 at this frequency.5 Hz. due to the saturation link in the loop. we generally expect the effects of the higher harmonics of e(t) on V to be small becauseof the following features of feedback control systems' typical nonlinearand linear links: e(t) have negligible efect on V. When the gain is barely sufficient for the selfoscillation to occur.l(b) for three particular values of k. the signal e(t) at the input to the nonlinear link is not exactly sinusoidal. Tlie knplitudes of the harmonics increase when the shape of v(t) approaches rectangular. This assertion is fair for common types (a) The nonlinear link characteristic v(e) is such that smallamplitude harmonics of of control . 11. I d ' T Fig. v(t)becomes nearly nshaped. e(t) does not differ sinusoid. Infact.Thecompensatorhasarealgain n: coeficient k (the compensator is not optimally designed).When k is large enough to make the loop gain Coefficient20 or more. With larger k.n)[sin a t + (1/3)sin 30t + (1/5) sin 50t + (1/7)sin 7 a t + ' I . v(t) is symmetrical and. contains only odd harmonics.l(c). the signal stabilizes with some specific amplitude and shape.In this case. which happens when k > 1 1. let us consider a typical feedback system example.1 (a) Block diagram of a feedback loop and.2 Harmonic balance accuracy During selfoscillation. The interference of its harmonics in the nonlinear element contributes to the fundamental V of v(t). 1l.the shapes of selfoscillation at (b) the output and (c) the input of ttie saturation link The shapes of the signal v(t) at the output of the saturation link are illustrated in Fig. Another explanation is that higher harmonics of v(t) are effectively filtered out by the much from the lowpass propertiesof the loop's linear links. As theresuit. therefore. as seen in Fig.290 Chapter 11.
the describing function can be different at different frequencies. the amplitude of the 34 dB third harmonic is 30 dB (i..in 1947 by L. Kochenburger in the USA (who introduced the term “describing function”)and several other scientists (in Germany and France). and in 1948. and they are attenuated 0 even more by the loop linear links as indicated in Fig. the 19dB third harmonic is 10dB below the fundamental. Therefore. however.p.2 Filtering higher looking at thefundamentalonly andneglecting harmonicsbylooplinearlinks the harmonics gives good a estimate for whether the system is stable. and for the valueof the stability margins.2) and 1 2% Imti=. Tustin in the U. 11. the slope of the gain response of the loop links is not steepand the phase condition of oscillation is met due to nonminimum phase lag or due to phase lag in nonlinear links.e. Goldfarbin Russia and byA. the phase condition of the oscillation arg T(ja) = n occurs at some frequency where the average steepness of the Bode diagram is 12 dB/oct. (b) r(j@gossesses proverties of a filterattenuating the return signal harmonics.2. 28 dB we conclude that in e($).K. With conditions (a) and (b) satisfied. Describingfunction Using harmonic stability analysis for nonlinear control systems was suggested almost simultaneously by several scientists. 1 1.describingthesignalattenuation andphaseshift. ja)= VIE (11.2. When. Describing Functions 29 1 system nonlinear links. Fig. 11. The real and imaginary partsof the describing function are defined by Fourier formulas as 1 2x Re ti = . This statement is called theconjecture of tl f. (1 1. Since in the IIshaped v(t) itself. The ratio of the fundamental’s complex amplitude V to the amplitude E of the sinusoidal signal applied to the link input known is as the describing function (DF): H(E.Ivcosatdat nE o ./ter.s. then harmonic analysis must involve not only the fundamental but also several harmonics of the signale@).. v(t) are even smaller. then. The filter conjecture is typically satisfied for welldesigned control systems with relatively smooth responseand small n.I v sinatdat nE o (11. the third harmonic is attenuated by 12 log23 = 20 dB dB relative to the fundamental by the linear linksof the loop. due to Bode phasegain relations. The higher harmonics in sc.Chapter 11. byR.3) . 30 times) lower than the fundamental.1) In general. If a periodic selfoscillation appears in such a system. and it is acomplexnumber.
11. equivalently. and the dead zone. the dc signal component needs to be taken into account.3 illustrates two ways to do stability analysis. The oscillation conditions (11.3 Using (a) DF and (b) inverse DF When H does not depend on frequency.4. further.3(a).5) (When the characteristic is asymmetric. When. The saturation threshold is e. However. when the inverse DF line in Fig.5) become satisfied whenthe DF traiectorv passes through the critical p a t in Fig.) Fig. approximately.4) T o o ) = lIH(E. Too) H(E. The difference is that H depends on E. 11. as shown in Appendix4. 11.is ed. and generally.The trajectories formedby the DF H(E) and inverse DF 1/H(E') at a specific frequencyand varying E are shown by dashed lines.3 Describing functions for symmetrical piece 1' mear characteristics 11.  .3. and of a threeposition relay are shown in Fig. the direct DF method is more convenient.. 11. analysis with inverse DF appears to be easier. or.of dead zone. the superposition principle and Bode integral relationships do not apply. 11.1 Exactexpressions Characteristics of hard saturation. . ju).jo)= 1 or (11.292 Chapter 11. the oscillation condition' in a feedback system is. of saturation with a dead zone. (11. DF analysis replaces the nonlinear link by an equivalent linear link with transmission functionH o o ) . the dc component of its output is zero. 11. Fig. when the DF is different at different frequencies which is typical for multiloop nonlinear systems and systems with nonlinear dynamic compensation.4.3(b) intersects the Nyquist plot. The integrals can be applied only in a modified form. Describing Functions When the characteristic of the nonlinear link is symmetrical. the harmonics at the input to the nonlinear links are negligible comparedto the fundamental. 11.
Le. When E e ed.4 are symmetrical. do not contain dc components. Any piecelinear characteristic can be obtainedby connecting a link k.. )Therefore.3) becomes0 . sincetheintegrand in (11. i. the output is sinot . (c) saturation with dead zone.. and (d) threeposition relay I The characteristics in Fig. in parallel .3)isanoddfunction. These DFs do not depend on frequency.e. the outputs are symmetrical. 11. Hence. Therefore.. arcsin(edE) (1 1. Describing Functions 293 Fig.Chapter 11. DF canbe found with only (11. 115 Inputandoutput evenfunctionandthefunction is symmetrical. signals of thedeadzonelink we can take the integral from arcsin(ed/E) to n/2 and multiplythe result by 4: ~~~ or arcsin E5 +A2E cos(arcsin i. k t us derive the DF for a deadzone link.ed/E up to the input output angle n .e.2). (In other words. The input and the output signals for the deadzone linkare shown in Fig. when the angle o t < arcsin(ed/E). %) %cos( arcsin %)] This expression for the dead zone DF is valid for the signal with the amplitude E > ed.arcsin(ed/E). (b) dead zone. 11.4 Characteristicsof (a) saturation. the output 0 is as long asE < ed. The output is not shifted in time relative to the input. Therefore. 11. the phase of the DF andthe 0 imaginary part of the DF are 0. Sinceshifted output not the is time in n 3d4 2n relative to the input. when the inputs to the links are sinusoidal. When e is positive.Becausetheintegrandis an Fig. the DF is 0. the links' DFs are real (the DFs have no phase shift).5.
.2) the describing hnction is H =or 4 a12 j sin cot dcot 7c 7c arcsin e..” “stop.. ed3. Therefore. the saturation DF equals 1 for E c e.2) is a linear function. and for E > e. is given by 7c E n E (11.7 Output signal of a threeposition relay for (a) E = 1.. . Therefore. here. 11.. subtracted from 1. 11. b. I  ko I I e edl ed3 L ” ” “ “ ” ” l Fig. .. the output Fig. c.294 Chapter 11.6 (a) Parallel connection of links with dead zones to implement piecelinear characteristic(b) For example. I E . . 11. shown in characteristic shown in Fig.. followed by linear links k l . as DF of the total link can be shown in Fig. Functjons Describing with several linkswith dead zones edl. are certain constants.7. k ~kf.6.05 and (b) E = 1.6) Fora fhreeposifion relay actuator (“forward.” “reverse”) withthe v(t) is twopolarity pulses. the saturation link DF is the DF for the link with the dead zone e. . from(11. Fig.. The obtained as the sum of the DFs of the parallel paths. 11.5 / The amplitude of the pulses is 1. 11. ed2.a.4(d). since the integral in (11.
9. then the pulsesare short. 1 1. 0 signaltothreshold ratio.7(b). the same: the system can oscillatewith the signal amplitude about twice . 11.4 dB. Example 1. dead zone.1 . The NyquistdiagramandtheinverseDFplotareshownin Fig. for theofDF of the inverseand the direct DF analyses threshold ratiois 2. 1 1. In this case.9(b).OOO(s + 500) s( s + 20)(s + isshowninFig. the harmonics in v(f) are therefore relatively small. 11. the loop gain for the fundamental is small. The oscillation is more likely to take place in practical systems to the maximum. The direct DF analysis is presentedin Fig.1 Hz.5. The loop gain of the linear links at this 5. 11.Chapter 11. Describing functions for saturation.8 Describing function characteristic for saturation. 6 dB. From the diagram in Fig.2 .. and the DF analysis is sufficiently accurate. the relay DF is maximum when E = 1. As seen in Fig. and threeposition relay are plotted in Fig. the second component under the square rolot can be neglected and ti (4/Z)ed/E. the pulses are rather wide as showninFig. the fundamental of v(t) is small. certainly. The conclusions are. dB 5 10 15 20 25 30 1 5 10 .5 15 . and oscillation cannot take when DF is close place.8.8.4 dB.11. Le. dead zone. The frequency at which the return ratio phase lag is 180" is 5.7) When j!?/edis large. and threeposition relay When E is only slightly larger than 1.03 Fig. the signaltofrequency is 5.05 . The stability analysis of a system with a threeposition relay and the loop transfer function T= 50.8. Describing Functions 295 (11.9(a). 11. 11.
27[E/(es. and a link 1/L as shown in Fig. and for E > e.6 dB.1)0.27(E/ed)l+ o. a saturation link. if L(s) approximates s over the frequency band of interest.27(i!?/ed)4. Functions Describing the threshold.8) . The threshold is eslL(ja)l. 2OOO 5  160' 140" 120" .1. H =z I. lSe. Hence.27(E/es)' . frequencydependent threshold .4(b) can be replaced by a parallel connection of a unity link and an inverting saturation link. a link inwhichthewidth of the dead zone is frequencydependent can be made. In a similar manner. 11.27/[1..9) A nonlinear link with characteristic including both dead zone and saturation is shown in Fig..9).0.7) which rapidly vanishes for large E. Then. For example.10 Nonlinear link with.10. 11.].{0.35dB correspondingly for E larger than. respectively. 0.6) for saturation DF is conveniently approximated by H = (4/n)(E/e.1dB.)' .4(c).27(E/eS)" (11. Calculations can be further simplified by omitting the second term in (1 1. (11.)4 . and an inverting saturation link with the threshold ed. 160' 140' 120' Fig.(ed/es)4]}(E/6?s)4 A saturation link with frequencydependent thresholdcan be made by cascading a linear link L. the DF for the dead zone link for E > e d is H z s 1 . for EE [ed. Fig. and the frequency of oscillation is approximately 5.)" or H 3 1. 11.1 Hz. It contributes less than2 dB. this is a rafe limifer..ed)]' .(E /ed)4] or HZ{1. 3 . 0.27(E/eS)" with the error smaller than 0.27[(E/e. e. 2e. 11. 11. It can be represented by parallel connection of a saturation link with the threshold e.0. H is as in (1 1.9 Nyquist diagram and the(a) inverse and (b) direct DF plots for a system with a threeposition relay 1 i 1 .27/[1 .296 Chapter 11.(4/n .(f?d/es)]}(E/e. e. The link with the dead zone characteristic of Fig. 2 Approximate formulas Expression (1 1.
3 arg H = . shown backlash shown in Fig.11) (1 1. Therefore.we have I E nE Jv. 11. branch v1 while the input e(t) is rising.de E + 1 E 7CE2 E j v2de= l E j (~.3. 0. 11. (11.lo).12(b) is typically in Fig.14rad . The characteristic for the caused by air gaps in gears.11 (a) Outputlinput characteristicof smooth saturation with hysteresis and (b) timehistoryof the input and the output The DF phase shift is arg H = arcsinFrom (11.The timehistory for the backlash link output when the input signal is sinusoidal is shown in Fig. 11.1 1. 0.3). 11. The output v(t) in Fig.11(a) shows the output/input characteristicof smooth saturationwith hysteresis. the area of the hysteresis depends on the signal amplitude. and branch v2 while e(8) is decreasing. 11. Describing Functions 297 11. ImH .12) Example 1.7n Example 2 . . V = 0. For the backlash.7E and S = 0.4 Hysthresis Fig.11.arcsin= O. Fig. H dsinat .The Schmitt trigger has a rectangular hysteresis characteristic. n EV (11. we have arg H = "arcsin S . The time delay of the output relative to the input indicates that the DF must have a negative imaginary component.11) and where S stands for the area within the hysteresis loop. By substituting H = VIE into the right side of (1 1.10) ImH =T 'v M O After replacing dsinot by E"de(t) and taking the integral of the output from E to E using vl(e). and backto E using vZ(e).Chapter 1 1. In Fig.l2(a).1 l(b) is foundby using. 1l.13.v~)de=M2E 7TE2 (11.
or else the overshoots in the feedback system become large or the system becomes unstable. 11. If the width of the backlash is 2.14 (a) Sawtooth signal generator and signal histories at (b) the Schmitttrigger output and (c) the integrator output . Hysteresis links are fiequently usedin oscillating feedback loops of which the sawtooth signal generator shown in Fig.[R1/(Rl + R2)]VCC.064 rad. h l . 1 1 14(c) are uh = t.14 is representative. The input thresholds of the Schmitt trigger shown by dashed lines in Fig. then the output signal amplitude is nearly the same as that of the input. and e = 20 sinot.where VCC is the power supply voltage.13 Timeresponse of Example 3.6'. output backlash link backlash the(b) and 11. 3. 11. Example 4. The backlash necessitates slowing the actions of the operator. Describing Functions Fig.Le. Because the phase lag reduces the available feedback.12 Characteristics of (a) SchmitttriggerFig..298 Chapter 1 1.Reducingthebacklash is important even for manually operated tools and equipment (like lathes) because of the manmachinefeedbackloopviatheoperator'stactileandvisualsensorsandthe operator's brain (compensator). and the phase lag is arcsin [2~20/(400n)] = 0. The feedback loop is composed of an inverting integrator with transfer function 1/(R&'s) and a noninverting Schmitt trigger. efforts aie always made to eliminateordecreasebacklash in gears andmachinery. 11. Fig. The integrator output is constrained within the dead beat (or dead ban$) c " .
15 (a) Bangbang temperature controller and(b) its output timeresponse When the plant in this feedback system contains extra highfrequency poles. to counteract additional plant inertia.14(b). switchingonand off withsufficientlyhighfrequencydoesnotpresenta problem. the phase lag in the Schmitt trigger will be less than 90" and the condition of oscillation will be not satisfied. the integrator output begins decreasing with constant rate VCCI(R3C) (V/sec) until the next switching occurs. The output signals' rates of rising and falling are the same in Fig. thus increasing the rate. but somewhat earlier in time or position. although sufficient for many applications.15(a). the control law is often augmented by additional rules. After the switching.the trigger output switches from VCC to VCC. and using PwlM is common. For example.Chapter 1 1. for example. However. The positive temperature rate depends on the power supplied bytheheater. 1/23 10  ~~ 0 t Fig. so that in one direction the resistance will be smaller and the integrator gain coefficient larger. It is seen in Fig.thehigheristhe control accuracy but also the higher is the oscillation frequency. An onoff (or bangbang) oscillatingtemperaturecontroller is shown in Fig. When the actuator is electrical. is also. 11. employing a Schrnitt trigger. i. The temperature of the single integrator plant oscillates within the dead band. 11. The period is therefore T F 4(udVCC)RsC = 4RlR3C/(Rl+ Rz). as when switching gas thrusters on and off. DF &sis cannot be used if the sawtooth signal at the input to the Schmitt trigger is replaced by its fundamental.andthenegativerate(thecoolingrate)dependsontheheat transferandradiationconditions. The accuracy of such an onoff controller. switching can be done not in the instant the output approaches the endof the dead band. . They can be made differentby adding in parallel to R3 an additional resistor R4 in series with a diode.e.11. seen that during the analysis of such a svstem the harmonics cannot be neglected. The oscillation frequency cannot be chosen to be excessively high since each switching in the physical actuators consumes some energy and/or wears out the contact mechanisms. bangbang control may be preferred. Example 5. is typically inferior to the welldesigned controllers using pulse width modulated (PWM) drivers. The actuator is fwoposifion a relay wifh hysferesis which can be implemented either as an electromechanical device or electronically. 11. when the switching cannot be fast and causes noticeable power losses (or propellant losses.14 that the linear link and the nonlinear link each lags the signal by 90"so that the signal comes back in phase after passing about the It loop.. Describing Functions 299 When the output voltage of the integrator arrives atuth.Thenarrowerthedeadbandis. since the thrust of such thrusters does not develop instantly).
Consequently. 1 1. 11. 11. 11. Fig. however. 11. An isoE Bode diagram of the loop gain of a feedback system with such an NDC is shown in Fig.300 Chapter 11.17 (a) NDC with parallel channels.17(c). 1 1. The Bode relations can only be used when constant E causes the DFof nonlinear links to be constant at all frequencies so that theDF can be equivalently replaced by a constantgain linear link. phase shift can be calculated from these diagrams . Fig. the The upper channel contributes relatively less to the output when E increases. the system is considered stable. We should keepin mind. DF is a function of E and cu (or j). 11. The DF is0f lines (or flines)are shown in Fig. The system is stable because the loop includes a link that provides a certain phase advance for largeamplitude signals.16. The AS analysis and design of the NDCs has been already introduced in Chapter 10. In this chapter we will employ DF methods. compensator isoE Bode diagrams are shown in Fig. (a) Fig. (b) isoE Bode diagrams From the isoE lines.I 6 (a) isof (solid) lines and isoE (dashed) lines. For the isolines shown in Fig. that these Bode diagrams do not.16 by the dashed lines. Example 1.5 Nonlinear links yielding phase advance for largeamplitude signals As mentioned before. log. iso€ Bode diagrams can be drawn as shown in Fig. Describing Functions 11. Eitherset of lines can be used for the stability analysis. the NDC lag decreases as E increases. and its iso(b) E Bode diagrams and(c) the loop isoE Bode diagrams. These methods allow designing certain NDCs which cannot be analyzed or designed with AS methodology. If no line passes over the critical point.uniquely define the phase shift. sc.17(b). arg DF increases with E.17(a) shows a simple proportionalintegral ( P I ) NDCusing parallel connection of a linear link with a nonlinear link. E increasing I c r. Theiso€ lines (or €lines) are shown by the solid lines. Because of the saturation.16(b) (MATLAB functions described in Appendix 14 can be used to plot the isoE diagrams for some typical NDCs). We will call such links nonlinear dynamic compensators (NDCs). 11.
For the purpose of stability analysis the loop can be crosssectioned at the input to either of the nonlinear links as shown in the figure.This can be seenby considering thatthecondition IT1= 1 is satisfied with some specific valuesof El = Elc and E2 = Ezc. and the valueElc appears at the inputof the NDC. Fig.18. does notdependontheposition of the cross section. these two isoE lines intersect when I T 1 = 1. some confusion can be seen with the gain stability margin: this stability margin seemingly depends on the crosssection chosen for examining the loop. 11. or to the input of the actuator with some amplitude E2. Functions Describing 301 This exakple shows that an effective NDC can be implemented rather simply. Nonetheless. Fig. Also. The proper cross section is that at the input to the NDC. although. and therefore. and the valueE ~ appears c at the input to the actuator. However. with some amplitude E l . the actuator. the value Elc returns to the crosssection. are not as frequency selective as the linear link in the loop with a single nonlinear link. Therefore. 11. and that usingisoE Bode diagrams is convenient for certain practical classes of NDCs. Before concentrating on the NDC design. as shown in Fig. the isof lines for El and E2 have a common point when IT1= 1. the loop DF is the same and the loop phase shift is the same. still lowpass filters. the same signals appear at the inputs of both nonlinear links. when the loop is broken at the input to the actuator and EZc is applied to the input of the loop. system with two nonlinear links therefore.f lines in a The nominal guardpoint phase stability margin.Chapter 11. if the loop is broken at the input to the N I X and Elc is applied to the input of the loop. when the feedback in the loop is large and the Bode diagramsof the linearlinks are reasonablysteep(theyarequitesteep when thesystemisNyquist . 11. the stability margin is just exactly satisfied. In both cases. 11. we first consider the NDC performance in the loop which also includes another nonlinear link. 11. the value EZCreturns to this crosssection. The loop return ratio DF can be measured by applying some testsignal either to the input to the NDC.19.19 Iso. 1 1. in Fig. uncertainty in the plant gain directly affects the loop return ratio: when plant gain reduces by a dB. The accuracy of DF analysis suffers when the loop incorporates two nonlinear links since in this case the linear links separating them. then the Nyquist diagram and the isof lines simply sink down by a dB. Hence. commonly. The NDC and the actuator are separated bysome linear link L.19. since in this case.18 Feedback loop with two nonlinear links separated by linear links Although different. Then.6 Two nonlinear links in the feedback loop A feedback system includingan NDC and a nonlinear actuator is shown in Fig.
To maximize the phase advance produced by the NDC.7 NDC with a single nonlinear nondynamic link For an NDC composed of several linear links and a single nonlinear nondynamic link with DF w.t& .od of this point. Therefore.13) where M and N are some functions of s. if the angle of M . Fig. even with the 390" phase error. corwspondingly. 1 1..case yhen a saturation link is used.240°. t " . 1 1. N = s. n/6 safety margin. T'his range Seems large:generally. This angle is. i.302 Chapter 1 1. 11. Hchangesfrom Milv to1andtherefore argH changes by arg(M/N). and is adequate for most practical applications.20 Vectors ( w + nn) and &e NDC to 4n/3. the NDC gives the phase advance no less than lo()"which is certainly betterthan no phase advanceat all. becomes H e&s~jvely sensitive to w. I as shown in Fig.21(c) &gernpfifjes. 11. using DF enables an easy design of a simple NDC providing phase advance of more than 120". are those equal in modulus to either MI or W . w + N) vanishes. arg N and arg M must be made as large as possible. the in and numerator wighborhq. bilinear& H(w) =w+N Wi" (11. the normalized NDC transfer DF depends on w as a ratio oftwo linear functions.20. Whenthesignalamplitudechanges. N) is n.13) is shown in Fig.21(a). eovegpgndingly. the error signal at the beginning of the upper branch decreases and the brgpck putpgt signal becomes negligible.suchuncertaintycan maketheentire difference between an unstable and a high performance system. ' A flowchart implementationof (11. T l & NW reducps the phase delay by 180"for largeamplitude signals.the vector w + M (or. Le. bounded by the continuity considerations. the phase uncertainty causedby the harmonics typically does not exceed "20".The most critical values of w. neither denominator larg MI nor larg NI should be allowed to exceed 2 d 3 . (or.resulting in a 180"phaseadvanceforlargeamplitude signals. w changesand H(w) changes. the transitionsin H are desked to be monotonic and smooth. with M = s. (Notice that when w j6 very large. 11.21(b) exemplifies the case where w is Q f&&giq pnplifier with the gain coefficientk and a dead zone. During the gradual change in the signal amplitude and in w.which is morethan ( w + N ) withminimumtmduli ewggh for all practical purposes. and M = Us. Describing functions stable). N = l/s over the ffequenGy rawe of interest. However. When w changesfrom 0 to 00.e. This requirement limits the phase shift of Fig. To avoid this oitvati9g with a . however.) Fig. For these values of w. 11. from this point of vkw..
DFforthe nonlinear element. The isof and isoE responsescanbefurthercalculatedusingtheset of isow responses.23. and a feedback path.1.Chapter 11.which still suffices for most applications.21(b).for three values of k x (DF of the dead zone link): 0.22. **** 11.k = 10. the output is VDB (6) G 6 6 0 4 5 1 R 6 6 0 1 * * * feedback path . 11. inverting G 4 4 0 3 0 1 e 4 4 0 1 R4 4 0 1MEG *** forward summer.ci. and the signal amplitude at the nonlinear element.21 NDC flowchart (a) and block diagrams (b). /SOw Bode diagrams can be used. * * * * ch9exl. Such a response can be calculated after replacing the nonlinear element with a constant linear element. * * * input integrator G 2 2 0 0 1 1 c 2 2 0 1 R2 2 0 1MEG *** feedback summer G 3 3 0 7 2 1 R 3 3 0 1 * * * kDF path: G5 =.21(b). These are the frequency responses measured while maintaining constant the value of the signal amplitudeat the input to the nonlinear element. or 10 G 5 5 0 0 3 1 0 R 5 5 0 1 *** forward path. 11. and 10. Describing Functions 303 Fig. a path in parallel with the forward path. 11.22 . theNDC includes a forward path with w.1. (c) To shift the phase over the full 'available range of 240°. 1.r for isow simulation of NDC Figs. the available phase shift change must not exceed 120" .e. 1.. and uses the schematic diagram in Fig. In addition to isof and isoE responses. when either M or N is zero. The isow Bode diagrams are shown in Fig. i. Example 1. The SPICEsimulationinput file is shown below. 11. Calculateand plot the isow Bode diagrams. When the NDC includes only a parallel path or only a feedback path. 11. In the NDC shown in Fig.
11. The first channel starts with a saturation link with unity threshold. . the feedback decreases with frequency. i. andtakesover when the fist channelbecomesoverloaded. Whenthesignalamplitudeislow. 11. 11.w response With the increaseof w. The second channel isoff when the input signal amplitude is less than 1.e. the signal amplitude after the summer is nearly constant at all frequencies where the loop gain is large. Correspondingly. 0 0 1 10 PROBE END .Inthiscase 2 ' 0 2 = CAP. with the increasein the signal at the inputto the NDC. Describing Functions *** VIN 1 0 AC 1 RIN 1 0 1MEG . The feedback system in Fig. and the second. 1122 SPICE model foriso. 1 1.24(b). as described in Chapter 6 ) and the PI controller shown in Fig.304 G 7 7 0 0 5 1 c 7 7 0 1 R 7 7 0 1MEG Chapter 11. and also. the phase lag decreases from dB 60 50 40 30 20 10 0 10.. Nonlinear nondynamic links are placed here in both channels. since the input block is an integrator. 11. IT to 0. The NDC can be equivalently implemented with a saturation link as shown in Fig. 20 30 40 50 Fig. the plot gradually changes from that of a double integrator to a constant gain response.8 NDC with parallel channels Simple examples of NDCs with parallel paths are theP I D controller with saturation in the Ichannel (or in the lowfrequency channel.withunitydeadzone. Fig.17.24(a) includes an actuator with saturation and an NDC with two parallel channels. .23 Isow Bode diagrams When the system is linear and the loop gain is large.thelooptransfer function is 2'01 = C1AP.AC DEC 20 .
26.24 (a) Feedback system with two nonlinear elementsin the NDC.25(a) are typically easier to implement. Fig. 11. Describing Functions 305 Fig. 5 I YE e s 1.11. The versions with saturationlinks showninFig. 1 1.24(b)andinFig.Chapter 1 1.5es 0 E.25. 1 . 11. Fig.or by plottingflines on theTplane as exemplified in Fig. log. 11.25 Three more NDC configurations The analysis and iterative design of the compensator can be performed by calculatingandplottingisoEBodeplots.11. 11.26 Intervals on Eaxis and on theisof line for a system with two parallel channels . sc. and (b) an equivalent diagram with a single nonlinear element in the NDC Three more equivalent block diagrams for the same NDC are shown in Fig.
The system has no limit cycles.and Hz increases from 0 to 0.28(a). The phase shift in the nonlinear dynamic link varies with the signal level.5. E < 1 andT = TO*.27 Nyquist diagram and isof lines for an experimental system 11. is comparatively short.8) for the intervalsof E shown in the left of the figure.] the expression(11 13) reduces to This piece of the isof line presents a segment of a straight line aimed at the endof the vector T02.26. Functions Describing o 2 H2(E) T = To1Hl(E) + T (11. lS I . while the output signal amplitude I of the composite link. 8 . On the fourth interval EE [e. 11. and does not deserve detailed analysis. short.27T02 (11.25 as can be verified with the H1 reduces from 1 to 0 plots in Fig.26. and its exact shape is not important for the stability analysis. 1: 1. e.27 shows the Lplane Nyquist diagram and isoflines measured in an experimental system designed withthemethodpresentedinthissection.9 NDC made with local feedback In many practical links the DFs modulus is the same as that for saturation.7) and(1 1. We can examine it piece by piece using simplified formulas (1 1. therefore. lSe. Fig. where EE [ 1 . 11.. and /E = 1. do not vary after the and. the isof line can be approximated by the sideof the parallelogram shown in Fig. 11. On the third interval where EE [1. This section is curvilinear.27(Tol + To2)IE T = 1.306 An isof line for Chapter 11.16) Example 1. This segment of the isof line is curvilinear.27T011E+ 1. The second interval.Wewillcallsuchlinks dynamic safurafion links. 11. A typical case is presented in Fig.]. In the vicinity of the stability margin boundary. Over this interval. On thefirst interval.. 1 1. . but the phas is nonzeroand changes withthesignallevel. .14) is displayed in Fig. f dB160 t Fig. the magnitude of the DF H saturation threshold is exceeded.7. On thefiJh interval E > lSe.
thephaseshiftchangesby arg(KB).e.29 Example of an NDC with dead zone in the feedback path The isoV Bode diagramsfor the NDC are shown in Fig. Since V and the dead zone link DF w are uniquely related and this relation does not depend on frequency. by the angle of the local loop phase shift. That is.the DF of thelinkchanges from " 3 to IC.30(a). When the signal is large. This value is sufficient for most applications. the link transfer functionI is C ( @ ) . This link is especially suitable for nonlinear dynamic compensation. The system has nolimit cycle. and K = 200. When the linear link loop gain is high. . 11. and the output signal is limited as if by saturation.e. and the isoV lines of the main loop on the Lplane might look as shown in Fig.. the closedloop transfer function is 1IB. 11. the full set of isoV Bode diagrams is the same as the full set ofisow diagrams. andthephaselag becomes that of K(@). and either one can be employed for the stability analysis.28(c) using similar links Kuo) and B.30(b). When the signal is small. IK(j0)l is large although decreasing with frequency.29.11. the phase lag increases with the signal level. and the dead zone link DF w changes from0 to1. the Bode diagram for B has constant slope8 dB/oct. 11.7.thedeadzoneDFapproaches1largefeedbackis introduced. 11. and B is real.That is. Still another version is shown in Fig.28 Dynamic saturation Fig. As indicated in Section 11. with'B =' 4 s (i.e.28(b) gives another implementationof dynamic saturation.. i. and the phase shift is 120'). 11. Fig. the describing function of theforwardpathdecreases. 11. Describing Functions 307 Nonlinear dynamic link Fig.phase lag reduces with the signal level.Chapter 11. For small signal amplitudes. Consider an NDC with dead zone in the local feedback pathas shown in Fig. As the signal amplitude gets much larger than thedeadzone. Example 1.. i. this value must be limited 120' to . the .thefeedbackbecomesnegligible. Here. The phase shift of the link will then be determined by the feedback path.
The proof mass and the soft springs the mass is suspended on are etched of silicon.31(a).31 (a) Silicon accelerometer block diagram. Fig. The voltage on the lower plate equals the voltage on the upper plate plus some bias. sc. It can be shown that with proper bias voltage. the upper plate voltage is proportional to the measured acceleration. 100k resonance mode susmnsion AM \ Fig. not shown in the picture. The accelerometer uses a tunnel effect sensor to measure the proof mass position. (b) compensator. 11.29 with the dead zone replaced by equivalent linear links. and (c) SPICE plant model .30 (a) Bode diagrams of the NDC shown in Fig.Theproof mass is gradually brought into the vicinity of the tunnel effect tip by an additional feedback loop using capacitive sensors. 11. 11. The position of the proof mass is regulated by electrostatic forcesbetween the proof mass and the upper and lower plates. The tunnel current flows between the proof mass and the sharp tunnel effect tip when thedistancebetween them issufficientlysmall.308 Chapter 11. Describing Functions I ' W H \ "I increasing t log. 11. and (b) isoV Nyquist diagrams for the main loop Example 2. A tunneleffect accelerometer is shown in Fig.
but when the gap is smaller. the tunnel current is exponentially larger. dB I Lplane phase shift 270 240 I degr 1: log. The mechanical plant might have some resonance modes with uncertain frequencies over 500 Hz. Global stability is provided by an NDC with a deadzone in the local feedback path. substantially increases the phase stability margin at frequencies below 200Hz. The deadzone element was chosen to be nonsymmetrical(a Zener diode) since the characteristic of the tunnel effect sensor is alsononsymmetrical. The Bode diagram and the Nyquist plot for signals of small amplitudes simulated in SPICE are shown in Fig. mass and suspension system to less than 3kHz.ForlowlevelsignalstheZenerdoesnotconduct. 11. This gain reduction reduces the slope of the Bode diagram. Two additional examples of applications NDCs of with local feedback incorporating a dead zone link are given in Appendix 13. only about 15 angstroms. Since in this case the derivative of the current to the gap width(thetunnelsensorgaincoefficient)increases.reducesthecompensatorgainatlower frequencies by approximately 30 dB. Without an NDC.e.theloopgainbecomes bigger than nominal. The normal value of the gap is approximately 6 angstroms. and improves the transient response of the closed loop which is important since the acquisition rangeof the tunnel effect sensor is very narrow.. Describing Functions 309 To achieve the desired accuracy.31(c). 11. The quality factorof the resonances is not higher than 20. scale 50I Fig. The lGS2 resistor is for the SPICE algorithm to converge.thesystemlocksrapidlyintothetracking modeandremainsstable whatever the initial conditions are. such a system would not be globally stable. 11. The SPICE model for the plant with such a resonance is shown in Fig. Two series RC circuits shunting the feedback path provide two leads giving sufficient phase stability margins over the range 200 to 3000 Hz. 26 dB. 1 1.thediodeopens andtheupper feedbackpath. The tunnel current is the exponent of the inverse of the tunnel sensor gap. which is an RC lowpass.32.32 Accelerometer (a) Bode diagram and (b) Nyquist diagram WhenthesignalexceedstheZenerthreshold.andthe compensator response is determined by the lower feedback path. In experiments. i.Chapter 1 1. Thefeedbackcrossover frequency fb is limited by the dynamics (structural resonances) of the proof. The compensator is shown in Fig. the feedback in the proof mass control loop must be madelarger than lOOdB atfrequencies up to 5 Hz.31(b). .
11. This and high sensitivity of H to the shape of the input signal (also the .a fullwave rectifier (i. The instantaneous amplitude of the output signal is determined by the upper channel output vl(t) shown in Fig. the integratorgain decreases with frequency without introducing the 90" phase lag of a conventional linear integrator.10 Negative hysteresis and the Clegg Integrator Let us consider two alternatives NDCs. 11. as seen in Fig.310 Chapter 11. two different linear links L1 and &.33 Negative hysteresis A generalization of the CIegg htegrafor showninFig. and to the noise.. 11. This. and a multiplier M.. 11. The negative hysteresiseffectcanbeachieved by switchingtheoutputatspecificlevels of the incident signal as shown in Fig. 11. the hope was expressed that this method would work well for smallamplitude signals. a highgain link with saturation realizing the sign operator.e. Describing Functions 11. 11.34consists of a splitter. is not possible. however. Therefore. A largeamplitudesecondharmonicispresentattheoutput of theCleggIntegrator. and it is very sensitive to the signal amplitudeand shape. more robust. when L1= kls and & = 1. absolute value link). When this idea was first introduced.12(a) but with reversed directions of the arrows on the branches of the characteristic.34(b).33(b).However. 0 Fig. Negative hysteresis links are rarely used since NDCs are simpler.thelinkdoesnotpasssignalswithsmall amplitudes. The signof the output signal is definedby the sign of the lower channel output. .33.34 (a) Clegg Integrator and (b) its signal histories In particular.34(c).(t) = Ivl(t)l sign vz(t) as seen in Fig. and able to provide much larger phase lead. thus allowing circumvention of Bode limitations and the causality principle.This harmonic's interference with the fundamental produces large phase lag for the fundamental. Fig.11. Such a link introduces phase lead up to 90" for signals of certainamplitudes. to Negative hysteresis is a link with characteristics as in Fig. 11. the composite link output signal is vou.
Hence. The value of the local feedback varies with the signal level.35 Local (actuator) and common loops with saturation When the main loop is disconnected at either of the cross sections (1) or (2). at either of the cross sections (1) or (2). a predistortion memoriless nonlinear link could be installed at the input to the actuator to make the total nonlinearity the saturation. the actuator's overload results in the reduction of the local feedback and in the introduction of a 120' phase lag in the main loop. 11. the system is stable (whentheactuatorloop. Consider an example of local feedback about the actuator in the block diagram in Fig. thus benefiting the main loop. c in one case. These linksmay introduce phase lag in the main loop that can result in a limit cycle. Another cominon problem is variations in the output impedance of the driver. For example. When it is not. This impedance dependson the signal level since local current or voltage feedback loops are often employed in the driver amplifier to make its output impedance correspondingly high or low. with parallel channels. the best actuator transfer characteristic is saturation. the Clegg Integrator and negative hysteresis are described here only to inform the reader that such ideas have already been explored and found not particularly useful. In addition to the common loop.11 Nonlinear interaction between the local and the common feedback loops Saturation links are located in the local and common loop actuators.an extra NDC with a 120' phase lead can be introduced in the loops. In bothcases.223~is s4B and B is real. . since it makes the loop gain constant up to the maximum of the output amplitude. or a local feedback about the actuator could be introduced. Fig. To prevent an oscillation.isproperlydesigned). Incontrolfeedbacksystems.theoptimalcharacteristicfortheactuator is hard saturation.Either of thecross sections canbe used for the stability analysis (the BodeNyquist criterion for successive loop closure hasbeendescribedinSection 3.4).Chapter 11. Describing Functions 31 1 disadvantage of negative hysteresis) prevent the circuit from being used in practice.theresultingloop includes a nonlinear d y n a ~ loop. in the other case. 11. Interference of the local loopwith the main loop is important to understand.when the feedback is limited by stability conditions.The nonlinear dynamic links can be analyzed as we did with NDCs.35. when L. 1 1. with local feedback. andthefeedbackintheseloopsisaffected by theactuatorordriver saturation. In fact.Localfeedback is employedwidelyinelectricaland electromechanical actuators to make their characteristics more linear and stable in time. a local loop is often employed in the ultimate stage of a linear amplifier tolinearize its characteristic.
when the driver isoverloadedandthegain inthelocalfeedbackloopvanishes. to limit the signal amplitude at the driver’s input. In other words. the diagonal terms are much larger than the others. Assumethateachloop is stable and robust intheabsenceofcoupling. The output current of the driver is proportional to the driver’s input voltage due to large local current . for example. solenoids. consider the effect of the coupling on the yloop.. let usconsiderthe main directiongaincoefficientstobe 1.kc. torque. or transducer + F. twooutput feedback system shown in Fig. Let us consider the twoinput. The effect will be easier to see if we redraw the system block diagram as shown in Fig.flux winding for electrical power generators. In this example.Then.38. 11. or z. This system can be. etc. to flux to or I. . andthe coupling coefficients. Describing Functions Example 1. Althoughtheplantgaincoefficientinthemaindirectionis substantially larger than the coupling coefficient. and can trigger a limit cycle. Le. the coupling is still not negligible. the transfer function of the actuator is a real number and the phase lag is 0. 11.36(b). and the actuator output force becomes an integral of this voltage. its output becomes a voltage.for magnetic windings of the actuators:voicecoils. asshown in Fig. overload .) However.thedriver’soutput impedance drops.12 NDC in multiloop systems AswasdiscussedinChapter 2.37. Since the output force (torque) is proportional to the coil current. Still.36(a).reaction wheels of spacecraftattitudecontrolsystems.output when the signal source impedance is high. 11. with some coupling between thex and y directions because of. and the plant is to a large extent decoupled. a resonant mode of the payload that is orthogonal neither to x nor to y.36 (a) Current feedback driver for inductive loads and (b) using extra‘saturation link at its input 11..introducesan extra integrator into the main loop. an extra satination link can be placed at the input of the driver. the coupling exists and might cause instability. To prevent this from happening. especially the coupling due to the plant structural resonances anddue to the plant nonlinearity.. without this feedback. 11. the driver amplifierwould be a voltage amplifier. or 1  9 back emf sensing resistor (a) Fig. say.+ 1+ Driver Currenttoforce. (Back emf does not affect the. as shown in Fig.312 Chapter 1 1.A typical example is using current drivers . an xy positioner.feedback. MIMO controlsystemsmostoftenhaveindividual actuators for each dimension of the system output. 11.
thefeedbackcan become positiveat any frequency below the crossover.Chapter 11. But if the feedback is positive. The with conventional loop gain response. for the Nyquist diagram.twooutputsystem Fig.39 shows the practically important case of the incident signal amplitude far exceeding the saturation levels. the composite link gain coefficient is close to kz.In the nonlinearmode of operation. This factor introduces additional uncertainty in the yloop.e. Harmonics lag could reach 12"in control systems higherthanthethirdalsocontributetothiseffect. the eliminates positive feedback in this loop and reduces the composite link gain.13. 11. the gain correspondingly increases.38 Equivalentblockdiagram The coupling resultsin a composite link with the transfer function 2 B X C X A X kc BxCxAx 1 + in parallel with the plant transfer coefficient. quitetheopposite.1 Harmonics DF analysis neglects the effects of harmonics on the output fundamental amplitude. whenthexactuatorisoverloaded.phasestabilitymarginsfortheisoflinesshould be larger by approximately 15" thanthephasestabilitymarginacceptedforthelinearstateof operation. Fig.The output v(t) is clipped most of the time. To ensure the necessary robustness. with the NIDC. Let us consider' the resulting error. or an NDC mustbe introduced in front of eachactuator .37 Twoinput..of v(t). stability margins can be reducedin the yloop and performance can be improved without sacrificing robustness.Whenthe NDC introduces some phase lead in the xloop. 11. . Describing Functions 313 Fig.even if the NDCs are notrequiredforthe individual loop operation (but they will certainly do no harm to the individual loops. 11.13 Harmonies and intermodulation 11.Therefore. either the stability margin must be increased with the resulting reduction in performance. This resultsin an extra phase lag for the fundamental of v(t). The third harmonic ine(t) delays the 0line crossing of e(t) and. When the feedback in the xloop is large. 11. i. theywillimprovetheindividualloopperformance!). Hence. therefore. this xactuator becomes overloaded.althoughtheircontributionsare smaller. which are indicated by the dashed lines.
as in feedback systems with resonance modes in the plant exemplified in Fig. be important while studying systems with resonant peaks andlor valleys on the loop Bode diagram and systems with large pure delay in which the loop phase lag can reach n without the loop Bode diagram being steep. DF analysis is still quite satisfactory at all frequencies except for those whose third harmonic is relatively large. 11. (a) (b) Fig.37. to the input of a nonlinear link results in the output signal v(t) containing not only frequencies rvlfi & nfi. Accordingly. At these frequencies. the effect of the third harmonic must be calculated since it can produce a change ofup to 30" or40"inthephaseshiftforthefundamental fo.3 14 Chapter 11. fi andfi.40. Describing Functions Fig.39 Effect of the third harmonic on the fundamental time delay Accounting for the harmonics in e(t) might. 1 1.someextraphasestability marginmustbeprovidedbyreducingthe steepness of the Bode diagram in the region close to& as shown in Fig. 1 1. 1 3 . When the Bode diagram for the loop gain is not monotonic. 2 Intermodulation Application of a signal having two Fourier components with different frequencies.thesystemrobustness computer simulation. the harmonics offi and but also intermodulation products with . mustbeverifiedwith 1 1 . 11.40 Openloop Bode diagram of a system with a flexible mode(a) reducing the gainof the fundamental and (b) increasing the gain at the third harmonic frequency Afterthe DF analysis and design.
For largeamplitude. the noiseis not noticeable. The IC’s designed for audio signal processing canserve the control system purposes as well.42(b).42(a). antenna pointing can beaffectedbywind disturbances. In control systems. Describing Functions 315 11. if vanishing disturbances in parameters of the oscillation (in the amplitude.e. as shown in Fig. 11. In this case. The process is stable. This function is performed by specially designed IC’s. t t is importanttohidethehighfrequencynoise of the magnetic head and the amplifiers.thesignalshowninFig. If the deviations from the periodic solution grow exponentially. Whileaudiorecording. for example) cause deviations from the solution which exponentially decay in time. loweramplitude components. (The intermodulation in the speakers can be reduced byusing separate speakers for higher and lower frequencies.. nonlinear distortions of the lowfrequency components will not affect the highfrequency components. higheramplitude components and higherfrequency. two different actuators for two separate bandsof distuibances can be employed. similar problems can occur. lowfrequency vibrations not to cause the actuator.by definition. 11. When the signal is clipped.41 (a) Clipping a multicomponent signal bya link with saturation (b) For example. The information would be preserved if the lowerfrequency and the higherfrequency components are first separated by a fork of lowpasshighpassfilters.Chapter 11. the amplifier bandwidth should be reduced to reduce the noise that otherwise would be clearly heard. after clipping. will not produce intermodulation. the process is not stable and the limit cycle does not take place. When the signal highfrequency components are small. For such selfoscillation to persist. i. This condition signifies a limit cycle and is illustrated in Fig. When highfrequency components of the signal are large.) Fig. to stop rejecting highfrequency disturbances. 11. Similar kinds of disturbances occurin vibration isolation actuators. the information contained in the highfrequencies components over the time of clipping is lost.and the highfrequency gain should be large for better signal reproduction.14 Verification of global stability The condition of harmonic balance at some frequency does not always lead to periodic selfoscillation at this frequency. 11. the latter having both lowerfrequency. components and smallamplitude highfrequency components is very typical for audio signals.41 withlargeamplitudelowfrequency For example.andthencombinedattheoutputafteramplification by separate amplifiers. the process of selfoscillation mustbe stable. .
pointing at instability of this process.thesignaldeflectsfromthesolutionexponentially (and very rapidly.43 Nyquist diagram and isof lines causing two limit cycles Forexample.qgetsinto the basin of attraction of some of the limit cycles.Processinstability willbefurtherstudiedinthenext chapter.fi.11.44 Nyquist diagram when the system oscillates with frequency (a) fa. An extra increaseof the signal level reduces the DF. This can be shown as follows. deviations of the signal amplitude from the equilibrium amplitude starts the process of exponential adjustment of the amplitude toward the equilibrium.e. 11.42 (a) Limit cycle and (b) unstable periodic solution Fig. and the signal starts risin Therefore. and thislimitcycle will proceed. the Nyquist diagram expands. i. From these three. but the analysis corresponding to oscillation with frequencyfi leads to the opposite conclusion. The simple test for global stability involves application of a stepfunction or a largeamplitude pulse to the command input. the limit cycles are associated with frequencies fi and f3. 11. Consider the limit cyclewith frequency A. the exponent corresponding to the fiequencyof oscillation has a negative real component.43. Such a test discovers most of the hidden . In this case. (b) A The analysis for the limit cycle with fiequency fi gives similar results. Then.. in asystemwithsaturationandtheNyquistdiagramshownin Fig.316 Chapter 11. if the oscillation at frequency A is created. Therefore. if the signal level getssmaller. The critical point 1 then occurs outside of the Nyquist diagram which is the mapping of the left halfplane of s. the critical point appears inside the Nyquist diagram which is the mapping of the right halfplane of s. and the signal amplitude reduces. by a jump). and the Nyquist diagram shrinks further. Describing Functions Fig. 11. the DF increases. thesaturation DF reducestheloopgainsuchthattheequivalentNyquistdiagram shrinks and passes through the point 1. practically. (a) (b) Fig. 11. andf3 cross the critical point. On the other hand.44(a). three isoflines related to the frequenciesfi. illustrated in Fig. and the solution at frequencyf2is unstable.
45. how to trigger selfoscillation? What is the relative amplitude of the 3th. of large amplitude and of various frequencies. 5th.45 Signals to excite limit cycles during a global stability check To estimate the stability margins.Functions Describing 11. how much smaller are the (a) 5th. 0 0 Fig.000. theoretically. such as those illustrated in Fig. To discover the suspected limit cycle. (f) T = 2500(~ + 100O)(s+ 1200)/[~(~ +30)(+ ~80)(~ + 180)l. but not always all of them. and 7th harmonics in the llshaped symmetrical periodic signal? During oscillation in a feedback control system where the loop phase lag is 180' and the slope of the Bode diagram is constant. using a set of bursts of oscillation isgood enough for testing stability of practical control systems.unstable. the initial conditions must belong to the basin of attraction of the limit cycle.15 Problems The loop transfer function is (a) 7= 200(s + 300)(s + 600)/[ s (s + 20)(s + 50)(s + 1OO)]. 11. is Why this failureof . (b) T = 100O(s + 400)(~ + ~OO)/[S(S +~O)(S +40)(~ + 80)]. even this test might not discover all limit'cycles. the initial conditions shouldbe chosen close to the conditions that will exist during the limit cycle. In other words. 1 1. to the harmonics) of the input signal. extra phase lag and/or extra gain is gradually added to the linear links of the loop until selfoscillation starts.000(~ + 100)(~ + 4 0 0 ) / [ ~+ (~ ~ O ) ( S+ 3 0 ) (+ ~SOO)] (d) T = 1 4 0 0 ( + ~ 600)(~ + 8 0 0 ) / [ ~+ (~ 15)(~ +20)(~ + 1200)l. Although. and use is as a counterexample to the (incorrect) statement that describing function analysis gives sufficient accuracy in all situations. 11. and these extra phase and gain valuesare considered to be the phase and gain stability margins. (9) T = 5. (c) T = 200. What is the frequency of oscillation (approximately).or conditionally stable?If it is conditionally stable. Such a test uses bursts of oscillation. Chapter 317 limit cycles. (b) 7th harmonics at the input to the saturation link relative to the fundamental? Invent a nonlinear link whose output's fundamental is very sensitive to the shape ( i a . is thesystemstable.Aretheoutputs of nonlinearlinksofcommoncontrolsystemsvery sensitive to the signal shape? DF analysis might failto accurately estimate the frequencyof oscillation in systems where stability margins are small over broad frequency bands. (e) T = 180(s + 272)(s + 550)/[s(s + 12)(s + 30)(s + 80)]. if there is a saturation link in the loop? If there is a deadzone link in theloop. and a counterexample system can be imagined where 'a limit cycle is triggered by only a special keysignal.000(~ + 5 0 0 ) (+ ~~ ~ O ) / [ S+ (S 3 0 ) (+ ~3 2 ) ( + ~3 0 0 0 ) l .
(c) 15 dB. with threshold 1 . 15 The area of the hysteresis loop is one half of the product of the sinusoidal input signal amplitude and the amplitude of the output signal fundamental. (b) dead zone 0. (f) T = 2 5 0 0 ( ~ + 1200)(~ + 1200)/[~(~ + 2 2 ) (+ ~ ~ O ) ( S+ 160)l.3.3. What is the phase lag of the linkDF? 16 Find the currentvoltage characteristic of the Schmitt trigger’s (a) transfer function (b) input. for the link with (a) dead zone 0.4. (9) T = 600000(~ + 6 5 0 ) (+ ~ ~ ~ O ) / [ S+ (S 1 6 ) (+ ~3 2 ) ( + ~ 300)]. (d) T = 140000(s + 600)(s + 8 0 0 ) / [ ~ +( 1 ~ 5 ) (+ ~ 2 0 ) (+ ~ 120)]. using the chart in Fig. 17 Calculate thedeadbeatzoneandtheperiod of oscillation in the system in .2 and saturation threshold 1. 8 Find the value of DF for a threelevel relay when the ratio of the signal amplitude to the threshold is (a)5 dB (b)15 dB (c) 25 dB (d) 5 times (e) 10 times. 11. 9 Do Problems 68 using approximate formulas for describing functions. (c) dead zone 0. 7 Find the valueof DF for a dead zone when the ratio of the signal amplitude to the threshold is (a) 5 dB (b).) 11 Thesignalamplitudeattheinputtoalinkwith 0. How many times does the DF change? 12 Write an approximate expression for DF which is valid for input sinusoidal signals with amplitudes more than the saturation threshold. What is the ratio of the signal amplitude to the threshold in the saturation link during selfoscillation? If the loop contains a deadzone link.(b) from2. How many times does the DF change? (Make an engineering judgment about what should be the accuracy of the answers.12345 deadzoneincreases (a) from 123 to 1230. using the chart in Fig. (b) T = 10000(~ + 4 0 0 ) (+ ~~ ~ O ) / [ S + (S lO)(s+ 3 5 ) ( + ~ 80)]. for: (a) T = 200(s + 300)(s+ 650)/[s(s+ 20)(s+ 45)(s+ go)]. 10 The signal amplitude at the input to a saturation link with the threshold 0.8. 11. (d) dead zone 0. 15 dB (c) 25 dB (d) 3 times (e) 5 times.3. what is the ratioof the signal amplitude at the input to the link to the dead zone? 14 Plot the transfer function on the Lplane with MATLAB and make inverse and direct DF stability analysis for a system with saturation.318 Chapter 1 1.12345 irweases (a) from 123 to 1230. Discuss the stability conditions. (c) T = 20000(~ + 1 OO)(S + 4 0 0 ) / [ ~+( ~ 1 O)(S + 3 0 ) (+ ~ SOO)].4 and saturation threshold 3. and the phaseis 180’ at some frequency. (b) from 314 to 628. 13 The linear loop links’ gain is (a) 25 dB. using the chart in Fig. 11.8 and saturation threshold 5.3 and saturation threshold 2. Describing Functions small importance for control system designers? 6 Find the value of DF for saturation when the ratio of the signal amplitude to the threshold is (a) 5 dB (b) 15 dB (c) 25 dB (d) 4 times (e) 8 times.72 to 7.8. (e) T = 18OO(s+ 272)(s+ 700)/[s(s+ 12)(s+ 35)(s+ 75)]. (b) 10 dB.8.
plot isow Bode diagrams for the NDC shown in Fig.29. To what values of the signal amplitudes do these DFs correspond? 20 In the nonlinear linkin Fig.thefeedbackpathbeing anintegrator. B = 204s + 20). 1117.23 and the related phase responses with MATLAB or SIMULINK.and the integrator elements are R 3 = 1 MIR and C = 10 nF. the heater power is 2 kW. whose gain response is flat and whose phase shift is zero over the band of interest.31 (a).I 5 if the preamplifier gain is 1. 11.5. When the saturation link becomes overloaded by a large amplitude signal.Plot isow and isoV Bode diagrams using MATLAB and the approximate expression for theof DF the dead zone link.5. (b) 10.(c) IOO? 21 Will the saturation in theforwardpathoftheNDC affect the NDCs performance? in Fig. (b) 0. 25 (a) With SIMULINK. the cooling losses are 200 W. 18 Calculate the oscillation period in the onoff control system shown in Fig. (c) 0. 23 Produce the isow Bode plot shown in Fig. The local feedback about this composite link makes the link gain response flat.05. When the link is overloaded by a largeamplitude signal. 11. with the threshold of the saturation link equal to 2.25(c)substantially 22 Plot with MATLAB theisoE Bode gain diagrams and phase diagrams for the link in Fig. Mark the plots with the values of €calculated with approximate formulas for DF.I4where VCC = 12V. 11.Chapter 11.isenloopedbylocalfeedback. What initial conditions must be used to verify by computer simulations that the system is globally stable? 29 Is the system with return ratio T = 20. and the dead zone (for each polarity) is I .31(b). 24 In the NDC shown in the block diagram in Fig. 11. 26 The composite link is a cascade connection of an integrator and a saturation link. what extra phase shift will be introducedin the main loop?Is this situation dangerous from the point of view of making the system not globally stable? 28 A Nyquiststable system is made globally stable by an NDC.11.I 7(b) when the DF of the saturation link at all frequencies is (a) 0.000(s + 500)(s + 550)(s + SOO)/[(s + IO)(s + 30)(s + 32)(s + 33)(s+ 3000)]stable in the linear mode of operation? Does the system have a limit cycle if there is a saturation link in the loop? At what frequencies is the phase stability margin equal to zero (read these frequencies from frequencyresponsesobtainedwithMATLAB)?What is theloopgainatthese frequencies? What is the frequency of the limit cycle? What is the value of the . Describing Functions 319 Fig. (b) Using SPICE. 11. 1 I . 11.1 . 1 I . plot isow Bode diagrams for the NDC shown in Fig. the dead beat zone is 1O°C. the Schmitt trigger resistors are RI = 1OOkS2 and R 2= 200 k S 2 . what extra phase shift will be introduced in the main loop? Is this situation dangerous from the point of view of making the system not globally stable? 27 A link.17(a) for the range of 1 e €e 100. Is the oscillation symmetrical? 19 Calculate the corner frequency (the frequency of the zero) in the plot in Fig. and the payload thermal capacitance is 2 Btu/deg.000. what is the phase shift at frequency 10mHz if the signaltothreshold ratio in the saturation link is (a) 1.
open loop G2 2 0 0 1 200 . zero at 300 L 2 2 8 1 G 3 3 0 0 2 1 . and compare the performance. at ii parallel RC twoDole with C = I and R equal to the inverse value of the p & e . the system is certainly conditionally stable. if the loop gain at this frequency is more 0 is 17 dB. 1 . When employing SPICE.46. the input file for the openloop frequency response calculation is Fig. The SPICE schematic diagram’is shown in Fig. or by simulation with SPICE.320 Chapter 11. with L = I and R equal to the zero value. zero at 600 L 3 3 9 1 R3 9 0 600 . Compare versions with different sequences of the linksin the local feedback paths. integrator . Each source should be loaded at a series. Compare versions with different sequences of the linksin the channels.output of the nonlinear link if the system oscillates? What (approximately) is the shape of the signal at the input to the saturation link? What kind of signal should be applied to the system to discover the limit cycle? project: Design a Nyquiststable system with an NDC with parallel 30 Research channels. use a chain of voltagecontrolled current sources. while the loop gain the loop. AS Answers to selected problems 1 (a)SincethephapeshiftofsaturationDFiszero. If a dead zone link is included in happens at f = 4. 1146 Schematic diagram for SPICE open loop simulation *chgocl.05 c 4 4 0 1 G5 . 1 1. to imdement a function Dole. pole at 50 R5 5 0 0.oscillationwilloccuratthe than dB.02 c5 5. 0 1 G 6 6 0 0 5 1 . RL twoDole to obtain a function zero.cir f o r determining oscillation condition *T = 200 (s+300) (s+600) / [s(s+20) (s+50) (s+lOO) I . This frequency whereargT= x. Describing Functions saturation DF if the system oscillates? What is the shapeof the signal at the . 31 Research project: Design a Nyquiststable system with an NDC with local feedback. pole at 100 R6 6 0 0. The loop response can be obtained by simulation with MATLAB.91 C 6 6 0 1 G 7 7 O O 6 1 . pole at 20 ~ 4 4 0 0 3 1 R4 4 0 0. gain. or. 5 0 0 4. 32 Research project: Design a Nyquiststable globallystable system with the approach and with the DF approach.2 Hz.
the system is conditionally stable. With a dead zone in the loop. To excite the selfoscillation. a single pulse or a burst of oscillation at frequency 4. a deadzone link and a summer should be added to the input file.5 dB . Check whether the loop gain VDB (7)is positive at this frequency.AC DEC 20 1 100 .2 Hz should be applied to the closed system . defloatingresistor c 7 7 0 1 VIN 1 0 AC 1 RIN 1 0 1MEG .Chapter 11.END 32 1 Use a cursor to find the frequency at which the phase shift VP (7) equals 180'.PROBE . and the loop should be closed. the limit cycle being at the frequency where the phase margin is zero.. (If it is desired to perform this simulation.input. source loading resistor . Describing Functions R77 0 1MEG .) 6 (a) 3 dB 7 (a) 9dB 8 (a) 4.
during experiments with. so they can be easily accounted for in a necessary and sufficient way by appropriate sizing of the stability boundaries.2 Absolute stabilityof the output process The system in Fig. any infinitesimal increment to the input process causes only an infinitesimal increment of the output process. 12. timevaried) system. process instability is not a critical design issue.Process insfabilify manifests itself in sudden bursts of oscillation or jumps in the output signal. This phenomenon is called jump resonance. by a jump). A system is said to be processsfable when the output processes are stable for all conceivable inputs. With stability margins of common values. The odd subharmonics may originate in control systems with saturation by a jump.The relationship between the amplitudeof the input sinusoidal signal and the amplitude of the fundamental of the erroris hysteretic. Boundaries on the Tplane specify the values and the thresholds of the jumps.the First Lyapunov Method.1 Process instability The output process is considered unstable ifaninfinitesimalincrementintheinput triggers a finiteor exponentially growing increment in the output. and troubleshooting feedback systems.As the input amplitude is gradually increased. process stability implies stabilityof the locally linearized (generally.the effectsof the output process instability need only to be bounded. testing. process instability and the associated nonlinear phenomena like jumpresonance and subharmonic oscillation may occur and can be very conhsing.I Chapter 12 PROCESS INSTABILITY In asymptotically processstable(APS) systems. a which point the error decreases (again.If the input amplitudeis then gradually reduced. In accordance with . However. and the majority of practical control systems are not APS . and the subharmonic originates smoothly. The APS requirement contradicts feedback maximization'. The second subharmonic only occurs in systems comprising nonlinear elements with asymmetric characteristics. The necessary and sufficientAPS conditions require the stability margins to be much larger than those commonly used. These phenomena contribute to the output error and therefore need to be limited. the process instability conditions and manifestationsneed to be well understood. In most cases. This is an important advantage of the frequency domain approach to nonlinear system design Theboundsforprocessinstabilityarefoundforsinusoidaltestsignals. For this reason. 12. subharmonics do not show up. the error and the output remain almost the same until ttie input amplitude is reduced to a certain value. The acceptable values of process instability can be reflected into certain boundaries on the Nyquist diagram. when a parameter of the system is varied continuously. the error amplitude increases smoothly until the error jumps up and then continues to increase very slowly.l(a) is said to be absolutely processsfable (APS) if it is processstablewith any characteristic of thenonlinearmemorilesslink v(e) whose differential gain coefficient is limited by 322 . 12.
i.e.. i. the monotonic characteristic illustrated in Fig.2 yields the equivalent electrical circuits.l). (12. output signal amplitude . The. shown in Fig.> 1. (12.it can be shown that when v(e) is saturation. the system in Fig.. If (12.2) To provethisstatement weneedtoshowthatthefeedbacksystemlinearized for deviations shownin Fig. 12. is resistive and positive.Hence. the nonlinear linkv(e) is replaced by a linear timevariable link with the gain coefficient dv g(t) = . Therefore.Chapter 12. (Notice that the analysis is valid even if the linear links in Fig. or l/g(t).1) and (b) someperiodicalinputsignalwiththe fundamental o.2).3(a) is stable.1Feedbacksystem (a) and its equivalent (b) Fig. 12.gradual change in the input signal amplitude renders a jump in the [PI. Therefore.1 is APS if it satisfies conditions (12.2 satisfies the condition (12.2) is not only sufficient but insomesensenecessaryfor APS. energydissipative. 12. 12. J Fig. Particularly. 12.10. ReT e 1 at some frequency o. de In accordance with (12. l). 12.that together bring forthan unstable output process. although timevariable.e. (c).3 are timevariable. the twopole in the lower part of the circuit in Fig.2 Characteristic of the nonlinear link with limited differential gain The system is APS (fat all freauencies Re T ( j o )> 1 .3(b).3) The analogy between the feedback system and the twopoles connection discussed in Section 3.1) and (12.twopole l/g(t) 1. 12. Instability Process 323 (12. 12.) LTV Condition (12.1). Here. Since F(s) = T(s) + 1 is positive real.2) is not satisfied.then there can be found (a) somefunction v(e) satisfying(12. (12. the coefficient g(t)E (O. and the input is a nshaped periodical signal with frequency o.1) dv occ1.3(c) is passive. de For example.2) gives a sufficient stability criterion for feedback systems. no selfoscillation can arise inthissystem.
324 Chapter 12. On the other hand. The required feedback Tplane system parameters should therefore lie somewhere between these two extremes. As the result.2) restricts the position of theNyquistdiagramasshownin Fig. Particularly.(c) described by the same equations The condition (12.andthenonlywhentheactuator is saturated and a rather. appropriate testsignals should beselectedamongthosetypicalforthepractical system inputs. . causes the output signal v(t) to change by a jump from the timeresponse (a) to the timeresponse (b) shown in Fig. excessive process instability is not acceptable. from U " to U " + 0. 12.to iII' .4 and requires the phase lag at large loop gains not to exceed 90". the figure 12=4 Restriction Of the in the following section) are employed Nyquist diagram by the APS of merit.While the input signal amplitude U is gradually increased. 12.3 Jumpresonance The amplitude and the shape of the output periodical signal generally are multivalued functions of the input periodical signal parameters.0. 12.an infinitesimal increase in the amplitude of the input.5. causes a jump down from the timeresponse (c) to the timeresponse (d) whentheinputamplitudeisreduced by an infinitesimalincrement. In practical systems not satiswng the A P S condition. and depend on the input signal prehistory. unusual command is applied to the input.3 Feedback system with timevariable real gain coefficient g(f) (a) and the schematic diagrams (b). 12. the amplitude of the fundamental of the output V increases by a jump. typically process instability doesnotcontribute muchtotheoutputerror. condition 12. For the systems where the inputs can beperiodical with largeamplitudes. To evaluate and certify the systems. the output might depend on whether the current value of the input signal was arrived at by gradually increasing or gradually decreasing the input signal amplitude.1(a) excitedby sinusoidal input u = U sin at.This limits the slope of the Bode diagram to 6 dB/oct and thereby reduces the available feedback. Jumpresonance can be observed in the nonlinear feedback system in Fig. gradual reducing the 'amplitude of the input starting at a value bigger than U" down to a certain value U. Process Instability Fig. Next.therelative jump values of the jumpresonance (described as. at a certainamplitude U = U" afterthesaturationlevel.
12.7 it is seen that the afterthejump amplitudeE.e.7. we assume that e($) = E sin cot. When the phase stability margin is rather small. .l(a) ( a scan be easily verified). 12. andalso.3.8. From the curves i Figs.Thesecurves allow one to specify the required stability margins from the values of the allowed jumps. This is exemplified in Fig.whileexperimenting.". Using the DF approach. Since the system is unstable.U between the points of bifurcafion marked dark. The proof that the solution on the falling branch is unstable a real with positive pole was given in Section 10. which implements the same relationships as the feedback system in Fig.6 Output signal amplitude for equivalent parallel channel system. 12.6issinglevalued. 12. .6. 12.tocalculatethestabilitymargins from theobserved jumps. vector addition of the outputs of the two parallel channels generates a collapse of U = 2E sin*/?) in the region where V IT1 A E.5 Shapes of the output signal v(t) before the jump up (a) and after the jump up (b).12.7 is. jumps . E jumps from beforethejump value Eb'l to the afterthejump valueE.between the two stable solutions In the system with saturation. 12. 'I) . and after the jump (d) down The conditions for the jumps can be found by initially analyzing the system in Fig.12. and when U is reduced to U '.' depends on the loop gainandphase. i. This plot comprises a falling branch over the interval (U'.Thecalculateddependenciesareplottedin. then E jumps down toEa'. say 15O. 12. threevalued. Instability Process 325 Fig. only the jumps down can be large. and (b) vectors' addition at the collapse Thedependence U(E) inFig.6 and 12.1.Chapter 12. WhenU passes U " while being gradually increased.Fig. 12. arg T is 165O.l(b).. the signal rises exponentially as shown by the arrows in Fig. ITI V U = 2Esin(yrc/2) Fig. before the jump down (c). (a) dependence of U on E.Theinverseoperator E(U) redrawn in Fig.
10.8 Lines of equal values of jumps downin a system with saturation The jumps occur in feedback systems when arg T is such that the curve U(E) in Fig.6 possesses a minimum. For feedback systemswith dynamic saturation (including those withNDCs). Such responseswere first observedduringthestudy of resonancetankswithnonlinear inductors having ferromagnetic cores. 12. On the plane (U. and therefore the jumpresonance is prominent. frequency frequency Fig.7 Threevalued dependence of the error amplitude on the input signal amplitude Fig. Process Instability dB r. 12. it can be shown 191 that the afterthejump amplitude ( 12. This takes place within certain fiequency intervals. 30 20 10 U' U" 0 30 60 Fig. 12. Because of the analogy described in Section 3.9.4) . 12. a parallel connectionof an inductor and a capacitor is describedby the same equations as a feedback systemwith the return ratio 1/(LCs2). 12.12.10 can be recorded while the input signal amplitude is kept constant.f ) in Fig. The phase stability margin is small (although it is nonzero due to the inductor losses). The jumps can be caused by varying either U or for both.326 Chapter 12. the limit case occurs of U' = U" and dU/dE = 0 with zerolength falling branch.9 Amplitudefrequency areas threevalued of amplitude U Fig.2. 12.1 0 Jumpresonance when the freqqency the of input signal is being changed Hysteretic frequency responses of the output signal amplitude similar to that shown in Fig. the area is traced where the dependence E(U) has three solutions. At the lowest and the highest frequencies.
5 dB is maximum. the subharmonics are not observedsince in this case the system behaves linearly. Instead. dB 15 10 5 0 5 10 Oo loo ZOO 30° 40° 50° 60° 70° 80° 90° Fig. M I = 1.Jumpresonance has been observed in the attitude control loop on the Mariner 10 spacecraft. henceE.85. Example 1. This implies a steep cutoff of the Bode loop gain diagram. It is difficult to measure the loop responses in RF and M W feedback amplifiers since these measurements require loading the port of the open feedback loop on a twopole withimpedanceimitatingtheimpedanceofthedisconnectedportas shown in Fig.Chapter 12. 12.86. 12. This is exemplified in Fig. We assume the saturation threshold 1.. phase stability margins at various frequencies can be found by observing the jumpresonance in the closedloop configuration and using the plot in Fig. the signalattheinputtothenonlinearlink mainly consists of thelowerfrequency components: E sin at + Esb sin(ot/n). On the isof line. Therefore. the subharmonics in lowpass systems with saturation could only become excited if the phase stability rnargin at. part of the time the nonlinear link is saturated by the signal.3.4. is Evidently.11 The jump value corresponds to the maximumof I MI on the isof line. with both E and Esb being so small that E + E& c 1.1 1.1 Odd subharmonics As will be shown below. Example 2 .4 Subharmonies 12. M I of 3.Le.8. 12.5. 12.the subharmonic frequency a/n is rather small. Instability Process 327 The denominator can be readily found by plotting the isof line on the Nichols chart. Let us examine whether odd subharmonic oscillation is possible with either small or big amplitudes of E and E&. . The gyro bearing nonlinearity lead to jumpresonance in the gyro loop. If E > 1 and Esb cc 1.' = 0. The problem was rectified by reducing the gain in the control loop. Nitrogen thrusters of the solar panels' attitude control excited largeamplitude periodic oscillation in the panels which have highQ structural mode. This in turn lead to the panel attitude control thrusters consuming an abnormally large amount of the propellant.
second subharmonic selfexcitation can be .4. shown in Fig. and the output of the saturation link v(t) accepts the shape of trapezoidal pulses. the input to the nonlinear link can be as shown in Fig. 12.13 control systems the phase stability margin is always bigger than x/6. 12.12 Thirdsubharmonicmechanism Fig. when the parameters of the input signal are manipulated slowly and continuously. On the other hand.2 Second subharmonic The second subharmonic can be observed only if the characteristicof the loop nonlinear link is asymmetric. 12. and the loop gain at this frequency is more than 1. the boundaries of the subharmonic's existence that correspond to different values of the signal amplitude are showninFig. the component with the frequency d n can be generated only due to the constant component of the Fourier series. 12. 12. The boundaries for [9]. If this angle exceeds the phase stability margin at the frequencyof the subharmonic. 12. the output increment is the product of sin(ot/n) and the Fourier expansion with the fundamental frequency o which characterizes the pulse element. if Esb >> 1 and E < E&.14 [9]. This signal is clipped at the levels shown by dashed lines. As seen from thefigure. Instability Process For small signal increments given to its input. These subharmonic pulses are shifted in phase by some angle due to the signal E sin ot.328 Chapter 12.and for higherorder subharmonics are displayed in Fig.Considering the subharmonic as the input increment. the nonlinear link does not introduce in the loop any phase shift for the subharmonic.thesecondsubharmonic is not generated 'in systems where the phase stability margins are greater than 3 0 " . Therefore. In this product. with nonzero steadystate amplitude E (this is called hard osci//afiron). Unlike odd subharmonic selfexcitation.12 that the phase shift is less than d2n. the wider are the areas in which the subharmonic exist. T. In practical the third. Since this component is real.12(b). For a system with onesided saturation. 12. a pulse element with the sampling frequency o. the odd subharmonics may only originate by a jump. dB 30 20 10 0 10 20 30 40 Fig.12(a). the link can be seen as an equivalent linear timevarying one.13 Lplaneboundariesforodd subharmonics ina system with saturation It is seen from Fig. a steadystate subharmonic oscillationmay be excited by creating appropriate initial conditions.12. and the greater the asymmetry is. and odd subharmonics are not observed.
6 Problems Prove that the systems in the block diagrams in Fig.~)(S + 0 . the phase stability margin is 3 0 ' and the slope of the Bode diagram below fb is 10 dB/oct. 12. what will be the change in the feedback at Hz? 0. In a system with saturation. = I . 12.2describes a processstable system with large feedback.5 Nonlinear dynamic compensation Nonlinear dynamic compensation can make the system processstable without sacrificing the available feedback. We may conclude that in singleloop systemswith saturation neither the odd nor the even subharmonics present real danger. 3 ) ] ? Prove the validity of the following equivalent forms for condition (12. what must be the slope for the system to be process stable? Approximately.03 Is a system with saturation APS if (a) T = IOO/(s + 0 .Chapter 12. since meeting mandatory the requirement eliminating for windup and substantial jumpresonance automatically excludes the subharmonics. 12.2) of APS: Re F ( p ) > 0 Re I/F(io) >0 (12.72)(s + 27.2)]? (f) T = 5000/[(~ + 0 . The conversion of the system with two nonlinear 10. Fig. Soft excitation means that when either U or w orbothare gradually changed along any trajectory entering the domain where the subharmonic can be observed. Is this system processstable?I f not.21)? (d) T = 500/[(~ + 0. with crossover frequency ti.(b) are described by the same equations. 0 1 ) ? (b) T = IOOO/[(s + 0.14 Lplane boundaries for second subharmonic existence in a system with single polarity saturation 12.7 links to an equivalent system with one nonlinear link that was described in Section is as well applied to designing NDCs to satisfy the process stability criteria.7.01)(~ + 2)]? (c) T = 123/(~ + 0. Example 3 in Section 10.1(a).02)(~ + 3)]? (e) T = 272/[(s + 2. the amplitude of the subharmonic increases steadily from zero.5) (12. 0 8 ) (+ ~ O. Process Instability 329 soft. without jumps.6) .
12.6 for a system with saturation. what must be the stability margin for the fifth subharmonic to be observable? . Instability Process T i )> 1/I T@)I (12. 7 OnaNicholschart.330 arg and cos Chapter 12. What is the phase stability margin (approximately) at this frequency? 9 In a system with saturation. (c) 4 dB. the outputinput amplitude dependence can be 5valued. What is the value of E after a jump down from the state of saturated output? 8 The loop gain is 20dB in a system with saturation at some frequency.6.8) 6 Show that in a system with dead zone and saturation. (12.5 times. Draw responses analogous to those shown in Fig.> T(j@ 1 I M(j w ) l = F(jw) sin arg T ( j w ) 5' Consider jumpresonance in a system with a dead zone. . 12. The jump down is (a) 2 times. by drawing a response analogous to that in Fig. (b) 1. (b) 8 dB.7) %.theminimumvalue of IM forasystemwithsaturation is (a) 6 dB.
This eliminates large transients caused by the initial switching on of the controller. Compositenonlinearcontrollersconsist of linearhighordercontrollersandthe means of transition between the linear controllers according some to participation rules. Windup is the large and/or long overshoot in nonlinear systems. The size of the regions where a single linear controller is operational and the complexity of the elementary control laws are discussed. Composite nonlinear controllers It was demonstrated in Chapters 10 and 11 thatsomenonlinearcontrollersperform much better than any linear controller. This can necessitate using more than two parallel branches and windows. For a small region in the variable space. The switching should not generate disruptive transients. the nonlinear optimal controller can be approximated well by a linear controller. the control bandwidth must be larger but the feedback can be much smaller. as multiwindow composite controllers. Many issues in multiwindow controller design have not yet been investigated. We already considered the design of NDCs using AS and DF approaches. generally. and so on. The design problem.p. nonlinear controllers perform better than linear controllers. Another typical application for the multiwindow controllersis timeoptimal control. it is important to guarantee that the combined transfer function remains m. NDC design methods are of profound interest. and the solutions are similar.3. It is relatively simple to implement and provides much better performance than linear controllers. is. In this chapter we consider the NDC design from yet another angle. another linear controller can be designed that would be optimal over thisregion. To perform each of these two tasks optimally. first. During the transition between the elementary controllers. precisely tracking it. nonlinear. When combining linear compensators via multiwindow nonlinear summer. The optimal controller in general. finding and locking into the target. and then.Chapter 13 I MULTIWINDOW CONTROLLERS Since. arises of integrating these locallyoptimal linear controllers into a 33 1 . Multiwindow control makes transitions between the elementary linear controllers onthebasisoftheamplitudeoftheerror. Stillanotherrelatedproblemistheproblem ofswitchingbetweentheexisting controllertoa new controller in anindustrialenvironmentwhenthemanufacturing process cannot be stopped. the control law cannot be the same: for acquisition. The widely used antiwindup controllers include nonlinear links directing the signal into different paths. Each of the linear controllers is a local linear approximation to the optimal nonlinear controller. it makes a big difference whether the controller to be puton is “cold or “hot. This problem is related to the acquisition and tracking problem. The transition between the two laws must be gradual in order not to deacquire the target. The chapter ends with a brief discussion of command feedforward in a system with multiwindow controllers. 13. therefore. It often occurs in the systems with a large integrating component in the compensator. depending on the signal level. and many questions remain as to how to make the best of such use controllers.” Hot controlters are those with the input connected to the signal. Several examples of nonlinear controller applications are presented. For an adjacent region. The acquisition and tracking problem is that of.
such controllers can be designed even by those ignorant of control theory. Hence. Fuzzy logic controllers break each smooth transition into several discrete steps. 13.i. With these rules and when the actions are scalars or collinear vectors. and the phase plane can be used for the. Over some transition interval of a variable or a condition. in particular.MultiWindow Controllers composite nonlinear controller and providing smooth transitions between these regional linear laws.. or actions) takes place.1. when I . fuzzy control design can be based on phaseplane partitions.m. What regionsize is optimal for composite controllers? There are two advantages tomaking the regions small. evenif some residue that needs to be cleaned out can be removed by either an acid or a base. anacidandabaseshouldnotbeusedasamixturewith gradually changing content. and on passivity theory expressed in state variables.332 Chapter 13. and not too shallow.Since them regions become very small. the precise shape of the participation functions does not matter much as long as it is monotonic. A smoother rule is illustrated in Fig. a monotonic shape of the participation functions does not yet guarantee the smoothness of the transition between different actions.e. and at the ends of the interval only one of the modes (or conditions.1(b). fuzzy logic control can. use loworder regional control laws. 13. also required that the of the adjacentcontrollaws mixwell.thatthecombinedaction adjacent controllers exceeds that of each individual controller. This increases the total number of regions with differentcontrollaws. Transitionsbetweenthecontrolmodescanbecharacterized by participation rules defining the set of participation functionsillustrated in Fig. The first is that the control lawsintheadjacentregionsmightbecome very'similar.1 Participation functions of control laws in composite controllers In general. 13.l(a). It is. As claimed by some fuzzy logic advocates. For example.1) where the scalark changes from 0 to 1 over the transition interval.whichenablessmooth transitions between them without taking special precautions. but these elements should not be combined in series or in parallel since they might produce resonances. The second advantage is that the linear controller can be of low order. Commonly.p. The transition rule can be linear and expressed as action = (1 . many variables need to be sensed and processed to define the boundaries of the regions. A lowpass link should not be carelessly mixed in parallel with a highpass link or else notches and n. However. This is not always the case. controller analysis and design. the transitions between the control laws are as illustrated inFig. 13. the controller action is the sum of the actions of the adjacent regional control modes. not too steep. interval I transition I " condition condition variable orvariable or " (a) (b) Fig. shift might result. For regulation of a reactive electrical current. a variable capacitor or a variable inductor can be used. (13. In fuzzy logic controllers.k) X action. Ik X actionz.
this approach is not difficult and leads to economical and nearly optimal controllers. In spite of the subclass simplicity. the output of a nonlinear operator can be approximated by applying the input signal to a bankoflinear operators. Nevertheless.notthephaseplanebut frequency domain methods should be used.k) log(action1) + k log(actionz). The windows divide the frequency spectrum (or. the number of decisionmaking algorithms and instruments for changing between the control modes becomes very large. Weiner. no other sensors and variables are employed to modify the control low. nonlinear dynamic links can be approximated by interconnectionsof linear dynamic links and nonlinear static links. This complicates both the controller design and the designed controller. we will consider singleinput nonlinear controller.2(a). MultiWindow Controllers 333 the number of the regions is large.do not requirehigh accuracy of design and implementation (i. In this case it is worth' considering logarithmic transition rule log(action)= (1 .e.. and the initial design is further optimized in practice by repeating the computer analysis. The signal components of the error can be divided into several sets bounded by twodimensional W ~ ~ O W shown S in Fig.andthereforeallowdevelopingsomerules ofthumb suitableforthe conceptual design and the design tradeoffs. and then combining various products of the linear operators' outputs.&herorder linear control laws can be made to remain nearly ptimal over a much broader region than loworder laws. 13. the'timeresponse behavior) and the amplitude range.3) ( actionz/nction$. this subclass allows for much richer varieties of the inputoutput relationships than those of the linear system. are robust). Within each window. The multiwindowcontrollersmakeasmallsubclass of suchlinks. This approach requires caution and application of certain rules discussed in Chapter 4 to provide good blending of the regional control laws at the boundaries between the regions. As shown by N. Le. rigorous strightforward methods for the synthesisof such controllers are yet to be developed. The partition between the regions should be also defined in the frequency domain. Forthedesign of higherorderregionalcontrollaws.. Unlike fuzzylogic controllers. equivalently. multiplication action = action1 X (13. This reduces the number of the regions and the numberofthe boundaries between them.2 .Althoughrelatively simple. Correspondingly. .Chapter 13. the number of boundaries between them becomes very large. the feedback error. In other words. a regional linear controller (compensator) is employed. the presentations in this chapterare not rigorousand rely largely on examples and simulations. As we concentrate on the applications. For controllers designed in frequency domain. summation (13.e.1) may cause nonminimum phase lag. 13. Multiwindow controllers perform significantly better than linear controllers. On the other hand.. i. Multiwindow control In the following.2) (13. The input of the controller is the output of the feedback summer.
The simplest multiwindow compensators are twowindow compensators. Such controllers provide good performance for a variety wide of practical problems. the exact shape of the participation rules is not critical.3 Diagonal windows The multiwindow nonlinear controllercan. This is the case.when the disturbance is a stochastic force with flat spectrum density. The criterion for minimum phase behavior is employed as the condition for smooth blending with the adjacent regions differing in freq. the components processed by thelinearoperators of thewindows.or postprocessing and for splitting and/or combining the signals. Due to this smooth blending. 13.3. Due to this correlation. and agreat number of useful nonlinear control schemes can be cast in this form. The controllers for these systems can be composed of linear operatorsof the windows and nondynamic (static) nonlinear functions for pre.2 The multiwindow control concept: Fig. in antiwindup schemes. and the threshold of the saturation need not be exact. The multiwindow compensator is nonlinear dynamic. the choice of the linear controller defined by the error amplitude and frequency content 13. and in Nyquiststable systems for provision of global and process stability.uency.time  frequencyw Fig. The twowindow controllers are widely employed.5). in particular. The regional linear controllers are optimized using Bode integrals as the performance bounds.334 Chapter 13. in acquisition and tracking systems. for example. and'when a displacement command profileis chosen in consideration of the limited force available from the actuator.thelowamplitudesignalsareprocessed withhigherlowfrequency gain. And conversely.The large amplitudecomponents are processed by theregionalcompensatorwithlowlowfrequencygain. the NDCs discussed in Chapters 10 and 1 1.andtheresultsaddeduporcombined by nonlinearfunctions.Thisarchitectureisreferredtoas multilrYindow.canbeviewedasmultiwindow controllers. whether they are made as a combination of parallel channels or as links withnonlinearlocalfeedback. the signal components that should go to a specific window can be selected eitherby the amplitude or by the frequency (the order of the selection is further discussed in Section 13. . The static nonlinearities used to implement the transition between the control modes can all be chosen to be of hard saturation type. applied to a rigid body and causing the body displacement with the spectrum density inversely proportional to the square of the frequency. MultiWindow Controllers frequency + . 13.be implemented as follows: the error is partitioned into components falling into different windows. A strong correlation exists in many systems between the error's amplitude and the error frequency contentso that the errors fall into a set of diagonal windows as shown in Fig.
13. 10Os/(s+10)2 I. but the outputs are both permanently connected to the actuator input as shown in Fig. large transients can result.5 Switching between controllers. n n Fig.linear compensators shown in Fig.35)1[s(s+2.” i. 13. Assume the inputs of several linear controllers are connected to the output of the feedback summer as shown in Fig.82(~+0.82)] Fig.Chapter 13. (b) at the output . 13. 13. A simple switching or some nonlinear law can be used to choose one of the outputs or a nonlinear function of all or some outputs to send further to the actuator. and switching between them will not create large transients. they process the error all the time. The following example illustrates the problem.4(b). Example 1.5(a).4 Switching (a) between “hot” controllers and (b) to “cold a controller On the other hand. and if the difference between the controllers is notexcessive. the transition between the regional laws can be made smooth.e.3 Switching between hot controllers and to a cold controller The sharpest participation rules are the instant switching between the controllers.” and its output signal is zero or some constant. when a nonlinear law allocates the inputs.4(a). The controllers are “hot. (a) at the input. the controller which has been off for a long time is “cold. When the actuator input is switched to this controller from even a similar but “hot” other controller. Since the controllers are hot. Even with the switching. 13. MultiWindow Controllers 335 13. m Plant I 2.theoutputsignals of thecontrollersaretosomeextentsimilar. Consider the system with a singleintegrator plant and two switchable ..
The compensators are permanently connected to the output of the feedback summer so that the compensators are “hot. MultiWindow Controllers h the The lower path compensator is a lead with an integrator.” and only after some time does the output signal drop to the steady state value. the compensator integrates the error. . 13. 0. 13.phenomenon this is called 8 windup. The overshoot for the largeamplitude input step is excessive and persistent .336 Chapter 13.thisdoesnothappenforlarge commands.4 Windup and antiwindup controllers The timeresponses of a system with saturation to small. The saturation thresholds are 100.and largeamplitude steps can be as shown in Fig. It is seen that in the system with hot compensators the transients caused by the switching are small Fig’ 13m6Transient responses while the transients caused by switching to the with switching to“hot“and “cold” compensators cold compensatorare large. since the return signal is reduced by the actuator saturation. andtheoutput I cold switches between 1 and 0.8(b). with some hysteresis to avoid frequent switching back and forth. The windup can be many times longer than the overshoot in the linear mode of operation (Le. 13.2. 3 Example 1. P 13.7 Transient response command smaller of amplitude).Thesaturationlimitsthe return signal and therefore prevents the error signal accu~ulated in the compensator integrator from being compensated. The asymptotic loop response is shown in Fig. In Fig..7. . shownin Fig. When the time comes at which this integrated error would be compensated by thereturnsignalinthelinear modeof operation. During the initial period after the step command is applied. 13(b). for the hysteresis link.The upper path provides much largerfeedbackbandwidthbutsmallerfeedbackatlowerfrequencies(these responses will be discussed in Section 13. it might take a long time for the feedback to compensate the integrated error.a similar system is shown but here the switching occurs from a hot to a cold compensator. and the lower path when 0 is toapplied 0 The output transient responses for both 0 systems are shown in Fig.1. Therefore. We will illustrate the windup with an example of a simple system with a singleintegrator plant. The error “hangs up. the overshoot for a step Fig. thethresholdsare0.6. linear mode (lower in curve) Windup typically is caused by a and windup (upper curve) combination two of factors: the error integration inthecompensatorandtheactuatorsaturation.8(a). 100. 13.” Switching between the compensators’ outputs occurs when the error magnitude achieves a certain threshold value.5). 13. The switch engages the upper path when 1 is applied to 1 +* the switch input. when the output is still low and the error is large.
Fig. 13. I V " " " " " " " " .m n+m error === 1 n command F n + m s3 + 2s2 s3+2s2+2s+1 actuator input = Ms= 2s2 + s command s3 +2s2 42St1 The timediagrams obtained with the MATLAB step command for this system saturation without are in shown . As seen from the asymptotic Bode diagram. Clippingthesignalpeakat the actuator by the actuator saturation windup. .Chapter 13. causes Fig.8 (a) System with saturation and (b) the loop asymptotic Bode diagram The compensator includes a parallel connection of an integrator and a unity gain path. the transfer function fkom the input to the error. It is seen that the signal at the actuator input has a large peak.9.25. .10 shows the output response the to stepfunction the command when saturation 0 2 4 6 8 10 Time (secs) level in the actuator is fl. 13. The increases the height and Fig. 13. and the transfer function to the input of the actuator are. n 2s+ 1 s3 +2s2 +2s+1 M=. The output response overshoot follows this peak with about 90" delaybecause of the plant integration. . respectively: 2s+l T== s3+2s2 m. the system must be stable. Compensator I (a) (b) Fig. The crossover frequency is1 rad/sec. The loop transfer function (return ratio). and a following link witha pole at 2 radsec. the closedloop transfer function.9 Timeresponses of a linear system the length of the overshoot. MultiWindow Controllers 337 . 13.
The value of the windup depends on the loop phase lag. the remedy suggested by this analysis is using separate channels for processing lowerfrequency and higherfrequency signal components.practically nonexistent.1 in front of the integrator in the compensator produces the response shown in Pig. the in integrator the of front0. simultaneously with the application of the stepcommand. the reset option is shown: transients to a step command. In Fig. 13. 1 3 . 13. by changing its output signal to zero the at risetime.11.19 in the command path. the capacitor is shorted for certain time(close to the risetime)to prevent the builtup charge. This makes the controller a threewindow one. dashed line by employing nonlinear dynamic compensation. two measures are shown which are widely used in analog compensators to reduce or prevent the windup. 13.e.MultiWindow Controllers Using the DF concept. the windup is.338 Chapter 13.1 in The windup can be reduced or eliminated compensator. 13.with large when the stability margin is 30" or added saturation with threshold smaller. Fig.12. In Fig.. 13.8. When Fig. corresponding to this low crossover frequency. or by resetting the integrator.10(b). A saturation link is sometimes placed as well in front of the Pterm as shown in Fig. 1 0 Timeresponsesforthe thephasestabilitymargin iS morethan '70°. the diodes (or Zener diodes) in parallel to the capacitor in the integrator limit the maximum voltage on the capacitor and.10 by the dashed line. but it is systeminFig. with a larger threshold than that of the saturation in front of the Iterm.6. In Fig. therefore. 13.1 l(a). . i. The explanation of the windup phenomenon can be also based on the idea of intermodulation: large amplitude lowfrequency component overload the actuator and prevent it from passing higherfrequency components.11 Analog integrators with (a) parallel diodes and(b) with reset switch Another methodof the windup eliminationis placing a rate limiter like that shown in Fig. Windup in a PID system is commonly reduced or eliminated by placing a saturation in front of or after the integrator. The resulting overshoot is long. the maximum charge in it that can be accumulated during the 13. placing an extra saturation link with saturation level 0. the qualitative explanation for the windup phenomenon goes as follows:Actuatorsaturationreducesthe describing function loop gain thus shifting the equivalent crossover frequency down. For example. solidline.
while the value of the overshoot reduces nicely.13 which exemplifies two types of architectures for multiwindow compensators. the time of the overshoot remains. However. 13.smallamplitudewindupisacceptable. 13. This order is different in the block diagrams in Fig.5Selection order The diagrams in Figs. The value of the threshold is not critical. 13. 13. 13.and Ppaths 13.theamplitude of the overshoot can be regulated by adjustments of the saturation threshold.Lp1 and highpass HPI) as in the left side of the block diagram inFig.13(a) shows compensators with parallel channels. .3 are somewhat ambiguous since they do not indicate whether the frequency selection or the amplitude selection is performed first. which helps to eliminate windup and to improve the transient response. When the saturation link follows the linear filter fork (a filter fork is a combination of lowpass and highpass filters for splitting or combining lowfrequency and highfrequency signals. the overshoot of 1% lasted 1 hour. saturation links are commonly placed in the lowfrequency channel since this channel's gain response dominates at lower frequencies. At large signal levels.Chapter 13.13 Multiwindow compensators with parallel channels Fig. Often a particular order is required. The overshoot can be reduced to say. in some cases.2 and 13. the saturation link reduces the lowfrequency gain. In such compensators.13(b). MultiWindow Controllers r""""""" 339 " " " " Fig.)Insomeapplications. and the phase lag of the compensator decreases thus reducing or eliminatingthe windup.Placingasaturationlinkwithanappropriate threshold in front of the Ipath commonlyreducesthelengthandtheheightofthe overshoot.13.12 Saturation links in front of the I. (In a versionof the temperature controller described further in this section. Fig.butnotinall applications. only 1%. Placing a deadzone element in front of the highfrequency channel with k c 1 reduces the phase lag at large signal amplitudes. in this case lowpass . but thehangoff can last a long time.
and the other half after. The polelzero frequency can be adjusted by a single knob. . and the lowfrequency components of this noise are substantial. if there is largeamplitude.10 and 12 with the absolute stability and process stability approaches. the integral term needs to be functional to reduce the static error but the integration is not effective if the peaks are being clipped by the saturation preceding the integration. Le.15 cafl be designed with the describing function approach as described in Chapter 11.a single pole.) Fig. 13. there is only one additional knob. 13. * Fig.(Forthehalfextraknobs are requiredcomparedwiththelinear integrator version. An example of the acquisitionltracking type is a pointing control system for a spacecraftmounted camera. When the error signal can be arbitrary.6 Acquisition and tracking Acquisitionandtrackingsystems. two PID controller. counting the saturation level. The best architecture of a PID controller with antiwindup depends on the sort of command the controller must respond to and also on the nature of the disturbances. two options exist: (a) halfofan integration can be placed before the saturation. the best performance (ina minimax sense) might be provided by a combination of the two strategies. placing the saturation first will prevent the accumulation of integrated error that causes windup. These implementations are indicated in Fig.15 Antiwindup NDC with local feedback 13. and one after. to cutoff the higher frequencies using . The NDC shown in Fig.14. are designed to operate in two modes: acquisition mode when the error is large. In this case itwould be better to lower the amplitude of the pulses by placing a lowpass filter (or an integrator) in front of the saturation link. short pulsetype random noise. When the only command is a step. like those used in homing missiles. the saturation link could be sandwiched between two lowpass filters.13.15 can also be designed as described 'in Chapters. However.340 Chapter 13. MultiWindow Controllers TheblockdiagraminFig. and tracking mode when the erroris small. 13.in which a rapid retargeting form a mosaic image of the maneuver is followed by a slow precise scanning pattern to object.one before the nonlinearity. or (b) using two firstorder links . cutting off lowfrequencies. Here. Another example clock is acquisition in the phaselocked loops of telecommunication systems and frequency synthesizers.. with a zero compensating the pole of the first link.13(b)shows an architecture of mixed orderof nonlinear windows and filter forks. But still.14 Possible /path implementations for antiwindup These NDCs and the used for the same purpose NDC with local nonlinear feedback shown in Fig. 13. 13.
25 . would produce largeamplitude transient responses. the acquisition response gradually sinks.16(a).5 (b) Fig.16(b).5. i. but the value of the feedback at lower frequencies should be made rather large to minimize jitter theand the tracking error.16(b): (1 3.thedifferenceinphasebetweenthetwotransfer functions at this frequency exceeds 180’. In the acquisition mode it is not necessary however that the loop gain be very large. As the transition from acquisition mode to track mode occurs. 13. in the tracking regime.1)of the acquisition and the tracking responses shown in Fig.) The nonminimum phase lag reduces the phase stability margin may and result in selfoscillation. as indicated in Fig. the tracking response rises. since the error is big and even with a small gain in the compensator. or by using nonlinear windows: the small errors are directed to the tracking compensator. (The conditions for the transfer function W l+W z to become n.the target can be deacquired. 13. As an example showing the importance of payingattentiontotheintermediate responses. or in a systemwith small stability margins and. even causing the target to be lost.4) and suppose that k smoothly varies from0 to 1. Whenthisfrequencyisstilllow. A special care must be taken to ensure that all intermediate frequency responses of the combined channel are acceptable. The transients generated while the system passes these values of k can be big and disruptive. and the frequency which at the moduliof the two components are equal. the system is in the acquisition regime. the actuator applies maximum available power to the plant. 13. the feedback bandwidth needs to be reduced to reduce the outputeffects of the sensor noise.Chapter 13. when each of the channels W l and W z is m. increases. intermediate tracking 10 cy log.e. were given in Chapters 4 and 5. in a modification of the system in Fig..m. 13.guaranteeingsmoothtransient responses during transition from acquisition to tracking is not trivial. In contrast. For example.5 where the compensators’ . scale . An improper intermediate responsemight result in an unstable system.p. The transition can generate large transients in the output and in the error signals. therefore.p. as shown in Fig.transfer function T has a zero in the right halfplane of s.. let the total loop response be the weighted sum (1 3. 13. the feedback bandwidth should be wide. and the large errors directed into the acquisition compensator. 13. MultiWindow Controllers 1 34 When the error signal is large.therefore the total . The transitionbetweentheresponsescan bemadebyswitching as shownin Fig.16 (a) Error timehistory and(b) acquisition/ tracking loop responses (a) While the determination of theoptimalfrequencyresponsesfortheacquisition modeandforthetrackingmode is straightforward. If the transients are excessively large. and the controller should respond rapidly.
With the regulation function +5s+64) eo>= l. 13. MultiWindow Controllers outputs are combined via nonlinear windows. although substantially better than a linear controller.7. To reliably avoid the nonminimum phase lag and excessive transients. the slopes of the Bode diagrams of regional responses should not different more than by 9 dB/oct in tivowindow controllers. the overshoot reached 500%. Hence.3) s(s+0.2.l(s+0.17) for the return ratio Tare shown in Fig. l6(b).25)(s+2)(s+3)(s2 s(s+05)(s+7)(s2 +lOs+lOO) and the intermediate return ratio T* ( 4= 250(s + 0. still does not allow implementation of the best possible responsesforacquisition and fortracking.005)(s2 +lOs+lOO) the responses of (6.342 Chapter 13.17.7. The appropriate responses can be also obtained by changing the response continuously. One method for this was described in Sections 6. Example 1.E 8 1 0 501 10l 0 I I I I11111 oo 1o1 Frequency (radhec) I I I I I1111 1o2 m 8 90 8 180 9 a 270 10” 1oo 10’ Frequency (radhec) 1o2 Fig. the twowindow controller.17 Responses for smooth transition from acquisition to tracking . 11. 50 E !! . 13.Thiscanbedonewith a threewindow controller using an intermediate frequency response like that shown by the dashed line in Fig.
A good approximation to a closedloop timeoptimal controller in a system with saturation requires using large loop gainoverthe wide bandwidth. and the openloop control entails considerable errors.18. After separation from the booster. 13. to about around the spacecraft. Timeoptimal controller is a relay controller. Whatever the chosen feedback bandwidth. Therefore. In order to obtain the best practical results. For example.The spacecraft booster is stabilized by spinning at 85 RPM. Formostcommonpracticalproblems of timeoptimalcontrol. the spacecraft 2 RPM. using an actuator with limited force or power.8Examples Example 2: Despin Control forSIC Booster Separation. (The yoyo is a weight at shown in Fig. the magnitude of the actuator action must be maximum all the time during the transition of the output variable between the lower force limit limits. Because of large uncertainty in the initial conditions after the separation.) The remaining spacecraft spin needs to be removed by firing thrusters. This requires using an NDC which can be implementedas a multiwindow controller as was shown in Chapter 10. the weight is also released and begins unwrapping the cable. and between the positive and the negative values. the feedback must be maximized under the limitationof keeping the system globally stable and without a windup. a Fig. and when left for some time uncontrolled. like in beam pointing of space optical telecommunication systems. Using a multiwindow controller can also help reduce the settling time.atwowindow controller suffices. off.Whenallthecablelength is unwrapped. 13. the cable is separated from the spacecraft. the controller design for the despin function must be made very robust. a proper balance must be kept between the achieved loop gain and the achieved feedback bandwidth. 1111 13. and in practical systems the loop gain cannot be infinite at all frequencies.7Timeoptimalcontrol I upper force limit Timeoptimal control changes the output variable position between the commanded levels in minimum time. at appropriate instants depending on the plant dynamics. Spinning of the spacecraft about the zaxis is unstable since the spacecraft is prolate and not absolutely symmetrical. the controller must be changed to provide better control accuracy in the cruise mode. the spacecraft will tumble. the timing for the switching cannot be exactly calculatedin advance. andat the same time..19(a] is despunby a yoyo. 1318 Timeoptimalcontrol force actuator in minimum time requires the force of theposition of arigid body profile shownin Fig. A stable closedloop largefeedback controlof an uncertain plant cannot employ a switch actuator since this actuator is equivalent to a saturation element with a preceding infinite gain linear link. and the yoyo takes away most of the rotation momentum. After the despin is complete. with various positions and spin rates and different types of coupling between the axes. it must performin a nearly timeoptimal fashion.Chapter 13. It switches the action on. When the spacecraft is the end of a cable wrapped several times released from the booster rocket. shifting a rigid body with. . When the plant is uncertain. MultiWindow Controllers 343 13. Morewindows may be necessary when the required settling error is very small. It has been proven that for the control to be timeoptimal. the despin should be fast.
344 X e Chapter 13. The initial simple P D controller wasemployedandprovidedsufficiently good despin and cruise control. the 2 basis of the absolute value of the I error each inchannel. The transient response for this controller is shown in Fig. It is seen that the controlis not timeoptimal. and zrotations. y. 13. with larger stability margins for the large error mode when the crossaxis coupling is the largest: (This twowindow controller was.however. initially at the rate of the booster. 13. r Y + (b) Fig..20 Timeresponse of zaxis despinning: control hws. The compensators are independent for thex. decoupling thruster logic Thrusters external forceiand torques ~ .20(b).20(a). MultiWindow Controllers PWM and b MI . DM separates the error vector into its Intheblockdiagram. and zrotations due to the spacecraft dynamics. 13. Due to large plant uncertainty.Le.complicated by coupling between the x. providinga large phase stability margin over the entire frequencyrange of possibleplantuncertainty and x. The multiplexer M does the opposite. the zaxis response was as shown in Fig.Thisrendersthecontrolofeachaxis independent to a certainextent. designedtoo late and has not been incorporated into the Mars Global Surveyor software.19Cb) uses pulse width modulated (PWM) thrusters.Theresultingcontrol (a) linearcontroller. y .thedemultiplexer components. This was done by passing the errors via withwindows saturationldead zone (a) (b) between the Fig.. A bettercontrolleris a twowindow nonlinear controller which 15 10 0 5 20 1510 0 5 20 changes control the law o n . the controller matrix is diagonal. however. 13. they are combined in pairs anddecoupled by thethrusterlogicmatrix.19 Spacecraft (a) local frame coordinates. 13.. The twowindow controller performs better and is at the sanie time more robust than the original linear controller. y. (b) twowindow controller law is nearly perfect for the despin function and as well for the cruise mode. including spinning of fuel and oxidizer. When the controllers' gains were chosen such as to despin the s/c without substantial overshoot.Theproblemis. and 2torques. Since each thruster produces x. andzcontrollers coupling. the despin was chosen to be proportional.) This example shows that even for complicated plants with multichannel coupled nonlinear feedback loops. (b) attitude control block diagram The controller shown in Fig. y. The despin time was reduced by 20%. a nonlineiir twowindow controller using only the error in individual channels for changing the control law 'provides' nearly timeoptimal ~~ 1 I I / .
its thermal controller electrical analogy (b).21 shows an electrical analogy toa thermal control system for a spacecraftmounted telescope (recall Section 3.5 and CLe= 0. The compensation for the highfrequency (HF) channel isa complex pole pair: CHF = 0. and the image in the focal plane to be clear.035 s+o.1 s2 +0. which are nonlinear.125s+O. and the torque is some fixedpositiveornegativevalue.Thetotalheater powercannotexceed 6 W. The plant is close to a pure double integrator. 13. Fig. and the controller configuration (c) The compensator is implemented in three parallel channels. MultiWindow Controllers 345 performance.)These controllers often do not include an Ichannel (lowfrequency disturbances are almost nonexistent). The thrusters are not throttled and not modulated. 13. substantially better than that of 1inear.21 Narrow Angle Camera (a). The plant is highly nonlinear because ofthe nonlinearlaw of heatradiationintofreespace. although there are flexible modes at high frequencies. and the pulsewidth timing resolution of 125 111s.5) ' The mediumfrequency (MF) and lowfrequency (LF) channels are firstorder: =s+0.l (1 3.5 ' (1 3.Chapter 13.6) . which is not discussed here. temperature of theprimaryandthesecondarymirrorswithin1. however there are also radiative losses G. . To avoid windup. Fig.6' (Another loop.The frequency response of the plant transfer function from the heater to the temperature differential is basically that ofan integrator.1). and only include a Pchanneland a highfrequencychannel. The primary and the secondary mirrors of the telescopemust be kept at approximately the same temperature in order for the mirror surfaces to match each other.Theheatersare pulsewidth modulated with the modulation periodof 6 sec. which is then considered to be the lowfrequency channel.Theheater H z isusedtokeepthe K of each other.orzero(similarlyto a 3positionrelay. drives the primary mirror heater HIwhich maintainstheabsolutetemperaturewithinthe 263O298OK range). Example 2: Cassini spacecraft attitude control with thrusters (without PWM).1 CMF 0. Example 3: Temperature controller for the mirrors of Cassini spacecraft's Narrow Angle Carmepa. The camera representsa small telescope.controllers.hey commonly use saturation in the Pchannel. .
The closedloop system transient response is notably insensitive to variations in this level. Industrial controllers. observed.) The antiwindupdeviceusedhereis a saturationelementprecedingthe LF path. unless an antiwindup device is provided. The heater power is maximum most of the time while the mirror is heated up. which often place the saturation link after the Ipath.23. and the loop frequency response is shown in Fig. the saturation threshold in the LF path was chosen to be 0. and the feedback bandwidth was Fig. The controller is nearly timeoptimal. The parallel connection of ’the M F and HE. Le. After a few (5 10) step response simulations were. 13. responses for thermal controller Fig. takes an excessiveamount of time to decay.This prevents the LF path from “integrating up” excessively when the actuator is saturated and the error is large.providing larger feedback at low frequencies. The poweron step response for the controller is shown in Fig. channels forms a Bode step on the Bode diagram near 3OmHz. The controller was implemented as digital. 13.22.22Parallelchannelcompensator ultimately limited by sampling effects. which makes a good level easy to determine. The separate and combined frequency responses of the compensator channels (ignoring the nonlinearity) are shown in Fig.24. MultiWindow Controllers and a saturation element precedes the LF compensatipn. frequently use integrator reset features to overcome this problem.346 Chapter 13. t . for transients in which the heater saturates. the heater saturates immediately. and the overshoot is insignificant. 13.. This would result in windup.8” I ( . 13.23 Loop frequency response for thermal controller The LF compensationsteepenstheresponseinthe 11OmHz range. excessive overshooting. Note that placement of the saturation element after the LF compensationresultsin a transientwith a winduperrorthat is small but. 13. (For the typical poweron transient.
6 In Example 1 in Section 13.9 is another example ofa twowindow controller. 13.9 Problems 1 In a system with saturation.Chapter 13. and make simulations with different saturation thresholds. 13. MultiWindow Controllers W 8 6 4 347 K O 0 K O 300  10 20 30 40 2000 2 0 1000 f.assume the plant gain is uncertain within 3 dB. What is the conclusion? 7 Study command feedforward with different frequency responses of the command feedforward linkin nonlinear modes of operation. The tunnel effect is an exponential function.andthesystemwouldbecome unstable if it were not for the gain reduction by the NDC. a double integrator plant. and increases the acquisition band of the tunneling condition. The antenna pointing controller described in Section 6. The actuatoris a saturation link with unity threshold. reduces the overshoot. The feedback bandwidth 200 is Hz (it is limited by sensor noise or the plant uncertainties). but also eliminates windup. PID and controller. Use SlMULlNK or SPICE.8. 250 I I I 0 5 10 r. 4 Make a simulation for the acquisition and tracking problem with switching between two responses. 3 Make SlMULlNK simulations of the system shown in Fig. 8 Study a command feedforward system for (a) largelevel commands and (b) large deviations of the plant response from the nominal. Plot the transient response for plant gain deviations up and down. a study the effect on the windup of the saturation links placed in front of the I and P paths. Design a good controller. and if the feedback loop were initializedwhen the distance in the tunnel sensor gap was much smaller than normal. 5 .5 is another example of a twowindow controller.usedeadzonefeedbackpathsabouttheintegraland proportional paths. 1 3 . Example 5. hours (a) (b) Fig.10.24 Step response for thermal controller Example 4: The microgravity accelerometerthat was described in Section 11. .This controller not only provides global stability with loop phase shift of n at frequencies where the loop gain is large. 2 In thepreviousproblem. sec . 5 The nominal plant is Po = l / [ ( s+ lO)(s + loo)]with up to +2 dB variations in gain. similarin shape (of the PItype) but one shifted relative to the other by an octave along the frequency axis.thentheloopgainwouldbemuchlarger.
prefilter. 10 Study a system with multiwindow compensators. and feedback path.348 Chapter 13. MultiWindow Controllers 9 Study multiwindow controllers with bounded internal variables in the plant and the actuator. command feedforward. .
Al. in cars. and why and how feedback system fail. and in TV sets.l Feedback block diagram For the purposes of analysisanddesignof a complicatedsystem. however. The term feedback was applied to closedloop system engineering . Understanding systems relies on understanding the blocks’ interaction and. Here.Appendices I Appendix 1 Feedback control.l are called block diagrams.everywhere where information in biological. 1 output input I I ~ Fig. In the modern world. is employed in spacecraft and missile control. 3 0 . particularly.2.2 Feedback control.by Harold Black of The Bell Laboratories in the 1920s. economical. biology. block‘s output is larger than the input (10 in this case) is gain coefficient. economics.5 ~ 2 5 .thesystemcanbe presented as an interconnection ofsmaller parts ofthesystemscalledsubsystems or blocks. The blockdiagrammethodbecamestandardincomputeraidedanalysisanddesign. and is widely used to explain and quantify the processes studied in biology. It is important to demonstrate not only how automatic control works. There is much more. A1. in response to the input value of 3. Feedback systems can and. and social sciences. denoting merely the obtaining of information on the results of one’s action. The fundamentals of feedback can be expressed in simple terms. We hope the following material will provide a better perception of how the systems of this world operate. Al. about the results comes back and influences the input. feedback. Al. The diagram describes this arithmetic: 3~ 1 0 ~ 3 0 . social and political systems . Pictures like Fig. to the quantitative meaning and methods of feedback. elementary treatment A1. the output of 5 is fed back to the input summer 349 .by which the. and social sciences. chemistry. elementary treatment A1. called the block’s In the block diagram in Fig. The word feedback is often employed in a much simplified sense. should be taught as a part of a science course in high school.2. in our opinion. but also how the system dynamics limit the speed and accuracy of control.1 A block diagram with a summer The factor. preceding and facilitating the teaching of physics. on understanding the feedback.1 Introduction The easiest way to comprehend complex systems is to break them into building blocks. the output value of 25 is produced. It describes regulation processes in engineering and .
the eye registers the rifle pointing direction gfld communicatesthis information to the brain as indicated in Fig. issues appropriate orders to the arms’ muscles to correct the aim. Ale2*2 Feedback control We start with examples. the flight computer calculates the direction to the planet and sen .3. Al.3 Block diagram describing aiming of a rifle Using the sights.3. In Fig.2 Block diagram of a feedback loop An inscription in the block (or close to it) often gives the name of a physical device represented by the block. Example 1. an eye looks at the target through the rifle sights.4 shows the process of the control system operation.andthebrain calculates the error and gives orders to the hands. The brain (a) calculates the error by subtracting the pointing direction fiom the command.Sights 4 Fig. While pointing a small spacecraft with a hardmounted telescope to take a good pictureof a planet. Al. command: Brain r ” ” ” ” ” ” ” I error I b “ ” “ “ “ a pointing direction Controller I Arms + Rifle pointing Eye 4. Al. A1.4 Block diagram describing driving a car Example 3.” The block diagram in Fig. Al. the “command” can be “drive west. r ” ” ” ” “ ” ” Brain driving Fig. The arithmetic is the following: output Fig. A1. While a rifle is being aimed. The pointing error is the difference between the direction to the target and the rifle pointing direction. Example 2. and (b) acting asalso a control/er. the rifle points down and to the left of the target.350 Appendix 1 forming afeedback loop.Theeyesestimatetheactualdirection ofthecar’smotion. While steering a fourwheeldrive car in a desert.
for example. but not The sensor is accurate. 6 General block diagram describing control of a plant Fig. The feedback control integrates the best features of both the actuator and the sensor. Al. The signal at the summer output is the difference between the command and the actual readings of the sensor.A1. Design a systemtomaintainthetemperatureof 1206’ C within an industrial furnace. Actuator 11111. A1. 1 1 1 .7 Block diagramof temperature control . The acfuafor drives the object of phf. as accurate as the sensor. but not powerful. a camera with a wider angle than the telescope. called the command: error  b Controller 11111. A1. The command here is: “1206’C. control. Al.” The actuator is now an electrical heater. It is widely employed in biological and engineering systems. command: calculated direction to planet the error Jets L . The controller’s gain coefficient is large. The pointing angle sensor. 1208’ Loaded furnace temperature  w Fig. It senses even a small error and aggressivel orders the actuator to compensate for the error.7. Spacecraft pointing angle measured spacecraft direction Angle sensor 4 Fig.6. the actuator is powerful. the error.6. no action is taken.5 Block diagram describing pointing a spacecraft We may now generalize the feedback co/?frolsysfem which is also called the c/osedloopsysfem to the form shown in Fig. the steering means can be the jets rotating the spacecraft. The plant is the furnace withthe payload. We use the general diagram of feedback control in Fig..e. A1. The sensor is an electrical thermometer. A If the error is 0. i. In a typical control system.Appendix 1 351 this data as a comrnand to the control system shown in Fig. Plant output l .here can be.5. The resulting block diagram is shown in Fig. Example 4. We now know enough to start designing control systems.
Now. This means that the command is not performed perfkctly.This link speaks two languages: its input understands degrees Fahrenheit.Al. . and the error is 0.352 Appendix 1 Example 5. At 100°F. An electronic thermometer. Design a block diagram of a biological system to produce a certain amount of a specific tissue. the tissue continues to be manufactured even when there is more than enough of it. how the tissue is manufactured and measured.15 atm. Example 6. the command is 2. the actuator is a pump. A1.05 atm. This may cause a serious health problem. Fig.01 V. Al. 3 Links Feedback systems are composed of links.as shown in Fig. The block diagram Fig. the thermometer output is V. Al.2 atm Pressure gaugen r4 pressure Fig. In other words. i. Al. (What is inside the blocks.e. this particular thermometer generates 0. Al.8 Block diagram of pressure control Assume that the pressure gauge reading is 2. A1. the output is 0.10 Thermometer link The electronic pressure gauge displayedin Fig.9 Block diagramof tissue manufacturing control When the feedback mechanism fails.8. the plant is the chamber. Design a system to maintain a pressure of 2.11 produces 1 V output for each atmosphere of pressure. 2 .2atm.01 V per each degree as indicated in Fig. is not considered here.lO.) Gene: make a Fig. its transmission coefficient is 1 V/atm..9 will do it.2 atmospheres in a chamber. 1 That is. for example. When the temperature is 1°F. and the sensor is a pressure gauge. produces electrical voltage proportional to the temperature. A 1 . 2. and its output speaks in volts.
12 shows a connection of two links making a composite link. this will not work. Al. the language must be common to the links.1 Dynamic links We assumed before that thermometers measure the temperature instantly. for example. Al. as shown in Fig.13 Links that cannot be connected At the link joint. Al.11 Pressure gauge link Fig. and not only because 13 is an unlucky number: the links speak different languages and do not understand each other.Appendix 1 353 Fig. for example. A mercury thermometer has to be kept in the mouth for several minutesforthereadingstoapproachthemouthtemperature.14 Equivalent composite link When several links are correctly connected in a chain. electrlcal CaYsec P power Heater Furnace 1111+ ' Thermometer in watts Fig.24 (cavsec)M/ in watts Loaded furnace temperature Fig. and produces0.12. electrical pow5 H e r heat. This is only approximately correct. .14. Al.24 heat. Al. in watts. but still not instantly. Thus. The heat raises the . the thermometer has memory. A1. A1. the resulting gain coefficient is the productof these links' gain coefficients. Al. but also on what the temperature was seconds and minutes ago. 15(a). connect an electrical thermometer to the output of the links of Fig. The thermometer readings depend not only on the instant temperature.13? No. We can.3.3 Why control cannot be pedect A1. calories each second per each watt. Fig. An electrical heater consumes electrical power from the 'input. CaVSec 0.temperature of the payload in the furnace chamber by an amount depending on the size of the payload. Electronic thermometers settle faster. as showninFig. Al.12 Composite link Can we connect two arbitrary links: a thermometer and a pressure gauge. as in Fig.
It depends not only on the frame I positionatthecurrentmoment. Let us push the frame by some distance.17 Control system's problem arises stopping of actuator the output time history action immediately after the error is reduced to zero. with larger controller gain. Devices or processes with memory are called ~ Y M M I I ~ C When . the Fig. The error decreases with time but does not completely vanish. reduces the error as exemplified in Fig. (c)temperature history after a delay Consider now a pendulum suspended from a frame. Another link with memory is the shower.17. the output can grow gradually. Certainly. A1. However. A more aggressive control. the change in the output is delayed. initially at rest.2 Controlaccuracylimitations After a step command (s) isissued. The pendulum has memory. curve (b).354 P 08O Appendix 1 frame position temperature P 1000 desired temperature 70° time the moment of placing thermometer in the mouth \ \ time the moment of moving the frame 500 \ time the moment of tuming the valve (a) (b) (c) Fig. a feedback system composed of dynamic links itself becomes dynamic.Theoutputdependsonwhat happened in the past. The output of the link is the pendulum position. Since the links in the feedback loop . . Al.it takes some time for the output of a dynamic control system to change. considering this distance the inputof the link. The output is delayed by the time it takes the water to flow through the pipe.A1.17. with bigger controller gain.15 Dynamic links time histories: (a) time history of thermometer readings.3.16 by dashed lines. (b) pendulum position history. and can be oscillatory. the input is changed instantly by a certain amount. A1. A1 .butalso onthe previous 1 frame position and when the position changed. as shown in Fig. the output is the shower water temperature. Al.15(c). The input is J the hot water valve position. as seen in Fig. I A1. The input and output time histories are shown FigaA1 *1 in Fig. The plot for the pendulum position alter the frame was pushed is showninFig. Al. curve (a).15(b).
of zero phase difference. the smaller must be the controller gain for the system to remain stable. A1. and the output oscillates as seen in the Fig. the oscillation opposite in sign. The actuator is the kid's muscles.17(c). Oscillation (c) is 90" delayed compared with oscillation(a). The kid jerks his body to produce extra tension in the rope. A 1.1 Selfoscillation The time history of an oscillation frequency = 2 osdsec. the information fiiom the sensor that the error is already zero comes back to the actuator with some delay. while a feedback control system is being designed. signals are in phase oscillationspersecondiscalled the oscillation frequency in Hertz (Hz).Oscillation(d)is delayed 180" which by is equivalent to changing the signof Fig. A single oscillation is a cyde. A1. A1.25 see. This curve is called a 1800 360° ~ i n u s ~ i dnumber The . A I . 18.4. A1. The larger the total delay of all dynamic links in the feedback loop. an error of the opposite sign will appear at the output.18(a). or having phase delay of 1800 remains in phase.e. time delay0. The sensors he uses are in his vestibular apparatus. with further increase of the controller gain. If the controller gain is even larger. The operation of a swing shown in Fig. curve (b). A1. He detects the proper timing for his movements by feeling zero tl .I 8 Time histories of motion the oscillation. Notice that if shifted by 360°. Rigorously speaking. and the actuator action proceeds for some time after the moment it should be terminated.2 Hz is shown in Fig.19 can be explained with the help of the block diagram in Fig.4 More about feedback A1. the oscillation amplitude increases and output will look more like Fig. and the oscillationgraduallydies.20. major attention should be paid to reduction of delays in the loop. + nn . and the less accurate and the more sluggish will be the control.Theoscillationamplitudecanbemaintainedconstantifthe frictioncaused lossesof energy are compensated by some mechanism injecting energy in the systemby some actuator. to sustain the oscillation.16 while a sheet of paper is being dragged in time the direction perpendicular to oscillation.Appendix 1 355 are dynamic. Al. the oscillation becomes periodical and with large amplitude like that in Fig. A simple way to explore this process experimentally is by trying to regulate the shower temperature while being very impatient. It can be drawn by a pen bound to the pendulum on Fig. Then. oscillation of the pendulum is not exactly sinusoidal. A1. i. Thus. The cycle is 360" long. phase delayof 90° 1 sec Oscillation (b) is in phase with oscillation (a).17. The process then repeats itself.
A 1 .22foragoodquality system with nearly equal gain at all frequencies from the lower frequencies ofHz 25 to the higher frequencies of 18. the phase shift in the loop is important as well . the return signal in the feedback system must be in phase with the swing motion and must be strong enough. speakers. with some special filling inside the enclosures) have wider and flatter gain responses. player I CD /’ ” ‘. Al. This method is used.21 CD player block diagram Fig.20 Block diagram for swing operation Fig.19 Swing Similar feedback systems are employed to generate radio and TV signals and in the dynamos generating electricity at power stations. with better magnets. + Muscles Brain signal to jerk + Swing Motion sensor motion Fig. Al. A I .356 Appendix 1 velocity in the rightmost and leftmost positions. 5 boombox frequency Fig. and for a portable boombox where the lower and higher frequencies are not well reproduced. for example. good system P CD + ) Amplifier disk 1111. in testing audio recording systems like that This systemcontainsaCDplayer.as we already know from the analysis of the swing. the system gain.A1.000 Hz. A I .A1.To sustain the selfoscillation. The gainfrequencyresponsesareexemplifiedinFig. thus making the sound different from the original. For feedback systems. Better and more expensive speakers (bigger. The input to output gain coefficient expressed in decibels (dB) is. however. 4 . with larger and firmer enclosures. Audiosystemsaretypicallycharacterized byonlythesoundamplituderesponses since our ears are to a large extent insensitive to the phase of the sound. 2 Loop frequency response Linksandentiresystemscanbetested with aset of sinusoidalinputswithdifferent frequencies.21.apoweramplifier. .and illustratedinFig.22 Frequency responses of a CD player The gain frequency response is not flat (as would be desired) because the speakers resonate at many frequencies with various amplitudes.
error x CAPS. Then.in aserieslinkconnection. During control system design.24 implements the following equations: output = error x CAP error = command fbs (A1. Using some algebrawe will find how small the error is.theequivalentgaincoefficientisthe product of the gain coefficientsof all the elementary links. A I . wherefrom .Appendix 1 357 A1. Actuator fed back signal (fbs) +Plant Sensor output 1 I Fig. A I . selfoscillation . and at those fkequencies where the signal is in a phase that! supports oscillation. the return signal is sufficiently small.2) (Al..3 Control system design using frequency responses Selfoscillation in conitrolsystemsispotentiallydisastrous.4. A controlsystemmust be $&&/e. the gain and the phase shift frequency responses about the feedback loop are first calculated with computers and then measured experimentally and displayedwith a signal analyzer as is shown in Al. Fig.2) we get error = command .. ) Actuator I . the feedback system on Fig.4 Somealgebra We already know qualitatively that when the controller gain is large.e. loop gain loop phase .3) fbs = error x CAPS command Controller II. As mentioned. Al.23.24 Feedback control system By substituting (A1‘3) into (Al. its phase is not such that supports oscillation. To prevent selfoscillation. 2 3 Measuring loop transmission frequency responses A1. and howmany times it is reduced by the feedback. the error is small.l) (Al. feedback control systemsare designed such that at fkequencies where the return signal is big.must not occur at any fkequency. i.Controller I =b Plant  Sianal Analvzer 4  Sensor 4 Fig.4.
floor vibrations caused by passing cars.e.. Al.i. By using position sensors and piezoelectric motors to move the desktop. and by people walking around.4) when using the disturbance as the command: disturbances at the output = D 1 D 1 + CAPS Without the feedback. and P = 1. if S = 2.0. The value of these unwanted components at the system’s output can be calculated with (A1.0.25 add some unwanted components to the plant’s output. very close to 0. we already have concluded that the feedback cannot be arbitrarily large.49988.4) is correspondingly0. Therefore. The expression 1 + CAPS is. numerically. the feedback. we find that 1 CAPS output = command S l+CAPS (Al. Appendix 1 This expression shows that the error is (1 + CAPS) times smaller than the command. much more than 1.5).25 Disturbances at the plant input For example. The feedback can be therefore used to reduce the effects of mechanical vibrations on some precision instruments and machinery. After substituting this expression for the error into (Al.4988. and 1 output = command x . Plant Sensor Fig. or 10.l).358 error = commund/(l + CAPS). And.488. A1. disturb precision optical systems mounted on a desk. A1. C A = 20. The larger the feedback. by the airconditioner motors.. or 100.4) When the product CAPS is large.4. the amplitude of the optics’ vibrations can be reduced many times. S (Al. then 1 in the denominator can be neglected.5) For example. the effect of disturbances at the system’s output would be DP.5 from (A1. .5 Disturbance rejection The actuator’s inaccuracy and the environmental disturbances D shown in Fig. w e e d b a c k reduces the output effectof disturbances ( I + CAPS) times. the smaller the error. the output from (A1.
many loops coexist to regulate various parameters: heartbeat frequency and strength.a plot of gain or phase dependence on frequency gain coeficient .description of physical system motion when forces are applied feedback . tactile and temperature sensors in the skin.a device like a motor.a device whose output equals the sum of its inputs variable . and the eyes. It would be difficult to count all feedback loops in a TV set .a measuring device sinusoid . muscle closedloop control. While operating our hands.a change in phase plant .a curve describing periodic oscillation summer .without phase delay link . A1.5 New words The following list was composed when the material of this Appendix was used by one of the authors to teach feedback control to his then elevenyearold daughter Helen.Appendix 1 359 A1. The list might be useful for those readers who will attempt a similar task of explaining feedback control to their children. actuator .a useful or paid for load to be moved or heated phase ship .a number of oscillation periods per second frequency response .return of signal or of information from the output to the input frequency in Hertz (Hz) . power amplifier. several sensors are employed.4.a number by which the signal is amplified in phase . amount of enzymes in the stomach. In complex systems.an object to be controlled positive feedback .feedback supporting oscillation quantitative expressed in numbers sensor . Sometimes. we use position and stress sensors in the muscles. and many others.a numerical description of a feature of a system (such as temperature or distance)   .6 Conclusion In this short introduction to feedback systems we considered singleloop control with a single actuator and a single sensor.feedback control . body temperature.representation of one variable dependence on another payload .using data from a sensor to correct actuator actions disturbance a source of error in the system dynamics . for example.there are hundreds of them.
theoutputsignalamplitudeis 2. when signal u&)= Umlsina t is applied toithe . and cp is the phhse shift between the output and the input signals.n phase . A2. o is the signalXrequency in radlsec.1 Frequencyresponses Linear systems have the property that when thev are driven by a:. Fig. A2. the m u t variable .1 Measurement of a link’s gain and phase shift Thegaincoefficient andthephaseshiftbothare.Thus. Signal generator Device under test +f Phase’ difference meter.’ or 4 O 0 . If a sinusoidal signalwith unity amplitude is the input. It needs to be measured to what extent the vibrations from the upper platform propagate through the truss structure to the second platform as theratio of the which is supposedtobequiet.input output signal of the link is uZ(t)= Udsin(at + cp).Thetransferfunctionisdefined measuredacceleration of a specifiedpoint on platform 2 totheforceappliedat a 1. For the purpose of the measuremeits.functions of the frequency.the same frequency. where Uml md U d are the signal amplitudes. the force is applied by a shaker. in fact. . Some relatively noisy (vibrating) equipment is placed on platform 1 (pumps.This motion generates reacting force F applied by the case of theshaker to theplatform. specific point to platform and the signals are sinusoidal. . The frequency axis is commonly logarithmic. link. tape recorders) while platform 2 is the place where some sensitive optics operates.with respect to the input by.is also sinusoidal with the of a link. and the phase difference between the signals can be seen a with twobeam oscilloscope or another phase difference meter. An accelerometer (a smallproofmassplacedon a piezoelement.n/3 (Le.sinusoidalsignal.l. motors. and the output sinusoidal signal is delayedi.2 depictsa truss structure (which can be a model of a stellar interferometer or something else)..1 shows how to measure the phase shift and the gain of an electrical link: a signal generator is connected to the link’s input. any variable of the system . thenthephaseshift is 7d3. The plots ofthesefunctionsarecalledgainandphaseshift frequency responses. and the gain is approximately 6 dB Fig. or a magnetic core of a coil suspended on a spring in the field of a permanentmagnet)producesanelectricalsignalproportionaltotheacceleration a. the‘gain coefficient is 2. by 60°). Fig. and. the voltmeters read the amplitudes of the input and the output. and 20 loglo( U d U m l ) is called the Example. Inside of the shaker there is an electromagnet that moves up and down a body called a proof mass.360 Appendix 2 Appendix 2 Frequency responses A2.and.generally. reaction wheels. The ratio U d U m l is called the gain coefficientof the gain of the link in dB. The measurements are performed in the frequency domain. A2.
and (c) the links' frequency responses The second link's gain coefficient is Mal = llo. amplifier signals of appropriate amplitude and frequency to the input driving the shaker. A2.3 (a) Force acting on a rigid body. This link's gain Coefficient is la/A = l/M.A2.(b) block diagram that relates the force to the body's position.Appendix 2 361 Insteadoftwovoltmetersand a signalgenerator. a signa/ ana/yzer is usedthat incorporates these three devicestogetherwith a displayand a computerthatsends of the power. Fig. The second link is an integrator:its input is the acceleration and its output is the velocity dB I . 20log(l/M) first link's gain 0 C log.and the link's output is the acceleration a = (1lM)Ssin at. and position is reflected in the threelink block diagram shown in Fig. Since the secondand the third links' gain coefficients are inversely proportional to .3(c). All three links can be integrated to form a single composite linkwith the gain coefficientl/(Maz) and the phase shift 1 8 0 ' . The gain in dl3 is plotted in Fig.3(a). scale Fig.gain of second or third link . velocity.3(b). and the calculation of the resulting acceleration. The input to the first link is the force. A2. A2."he application of force F = %in cot to a body with mass A4 as shown in Fig. and the phase shift is 0.A2.2 Measurements of a link's gain and phase with a signal analyzer Example 2 . The third link is also an integrator: its input is the velocity and its output is the position x = I v dt.and the phase is 90".
The transform is linear and exists for all practical stable functions At). can be seen magnitude and the phase of a complex transfer function.2 Complex transfer function The two scalar variables.flt) = t . A2. Example 1. means that the gain coefficientis 5.1.at3i2. A2.a?. as the Example 1. (infinitely narrow pulse whose area is l). Saying that a transfer function is 5exp(i. Here. the gain of these links decreases with frequency with a constant slope of 20 dB per decade.1 Laplace transforms F(s) = j f ( t ) e"dt 0 or F(s) = 4 m)I renders a unique correspondence between a function of time Jlt) and a function F(s) of Laplace complex variable s = CT +j a . and the phase shift is 30'. s+o . a ( s + b ) 2+ a 2 s+b (s + b)2 + a 2 The initialvalue and finalvalue theorems are the following: f(0) = lim sFO) S+ and t+ lim f ( t ) = lim sF(s). Example 2.362 Appendix 2 the frequency. the gain coeficient and the phase shift. for transform(s + a)2. Notice that it can be written as 1/[(ja)2Aq.e"(1 + at) sin at cos at e4'sin at ebtcosat +a)". The slope of the composite link gain is therefore 40 dBidec. Some of the transformsare shown in Table S(t) is the deltafunction A2. and 1( t )'is the stepfunction./6). In Section A2. for transform(s + a ) " . or.1'~example of a mass driven by a force. 12 dBioct. Inthe region of small times:A t ) =: 1 at. ~ l t = ) 812. equivalently. the total transfer function from force to position is 1/(Ma2). For the Laplace transform thefunction A t ) = e"'.3 Laplace transform and the splane The Laplace transform oo Table A2. (s 1 lis lis2 t + a)" + a)2 + a)" ai[@ + a)] a/[s2(s + a)] + a2/[s(s a/(s2+ a2) s/(s2 + a2) (s (s (s 11~3 812 e " ' te4 (812)e" Ie4 (at 1 + e4)ia 1 . or 5L30°.
4 Poles in the splane and related time exponents AZ. Example 2. multiplyina a Laplace transform by s is analogous to werentiating the corresponding timefinction. the Laplace transform ofthe output is + 1. A2.5s+ 7 (s k l)(s+ 2)s s3 + 3s2 + 2s 5(s _.AsseeninTableA2. Since the stepfunction is an integral of the deltafunction.4) . the h v l a c e transfer function o f a chain connection o f links is the product of the h p l a c e t r a n s f o n c t i o n s of& individual links.Appendix 2 363 When F(s) is rational. As long CT c 0 and the envelope of the signal is exponentially as the poleis in the left halfplane. presenting F(s) as a sum of partial fractions.4 Laplace transfer function The Laplace trmsfef fufwfion is the ratio output to that of its input. Conversely. i a splane / Fig.purelyimaginary poles make sinusoidal signals. summing the timefunctions that correspond to the fractions. 1 is Us.4)/(s + 2). is the hpluce transfer function o f an integrator. . I / . of the Laplace transform of a link's Example 1.4 shows various poles of F(s) in the splane and the related functions of time. I Since'theLaplacetransform is linear.thetransferfunction of several links connected in parallel equals to sum the of the Laplace transforms of the links. Since the stepfunction's Laplace transform is Us. ' .realpolesmakeexponentialsignals. The stepfunction l ( t ) is applied to the input of a link composed of two cascaded links with transfer functions 5 4 s + 1) and (s + 1.and then. evidently. Find the output timefunction. the related function At) can be found by. According 'to the convolution (Borel) theorem.1. complex poles make oscillatory timeresponses. A2. and the ratio of the stepfunction Laplace transform to the deltafunction Laplace transform according to Table 1 1 1 1 2 . Fig. first. narrowing with time.
To calculate the response to sinusoidal inputs.e.theoutputgrowsexponentiallyuntilitbecomes because of the power limitation in active devices.the FortheinputsignalRe(emsP')where transfer function is infinite. with the step command. When the input signal is an exponent Re(e'O' ) and so is a zero of the transfer function.1 that correspond to the partial fractions: 2e* . so largethat.e.1. s should be replaced by j a . i. the output signal is zero. When a pole of a transfer function is in the right halfplane of s. Laplace transforms and inverse Laplace transforms (i. the output signal becomes limited.verifyingtheseinequalitiesissimplerthanactually finding the roots.1. den) or with num = [5 71. impulse (num.5 X l(t) . as is illustrated in Section7. sp is a pole of thetransferfunction. The RouthHurwitz criterion indicates the presence of rightsided polynomial roots when certaininequalitieshold.. The output timefunction can be plotted usingMATLAB either with num = E5 71. To determine whether a system with a given rational transfer function is stable.364 Appendix 2 . den = [l 3 21. s t e p (num. .Such a systemis considered unstable since there always exist &function components in the input noise. and the system no longer can be viewedas linear.AB commands laplace and invlaplace.This results in a complex transfer function.thenumeratorandthedenominatoroftheLaplace transform of the output are used(astheresponsetothe&functionwhoseLaplace transform is 1). Inpracticalsystems. The timefunction of the output can be obtained bysumming the timefunctions from TableA2. the numerator and the denominator of the transfer functionare used. den = [l 3 2 01.. the output signal in responseto&functioninputis anexponentiallygrowingsignal. one might calculate the roots of the denominator polynomial and check whether there are roots with positive real parts.5 s s+2 can be found with the MATLAB function residue. The partial fraction expansion of the Laplace transform of the output 1.5e2' + 3. This can be done with MATLAB and many other popular software packages. den) Withthe impulse command.5 2  s+l ++ 3. The criterion is not described here since software packages calculate the roots in no time.2. the signal is infinitely amplified and becomes infinitely large at the output. finding fit) from F(s)) can be directly found with MATL.
e. It is evident that this vector is lowest in magnitude at lower frequencies where therefore the transfer function is largest. to nonminimum phase shift.. The pole is s = 5.thefrequency Example 1. i. i.InChapter 4.. At frequency o = 5 radhec.1 times larger than that of the input. the gain decreases with the constant slopeof 6 dB/oct.Le. Example 4. The vectorjcr. and the second one.+ 5 is shown in Fig. Whenthefrequencyincreases.This is a rational function with a double pole at the origin. Example 2.1/45".e.. 5 Poles and zeros of transfer hnctions The locations of the transfer function's poles and zeros manifest themselves in two important aspects: Theyshowwhichsignalsareamplifiedinfinitely(exponentialsignalswiththe exponents equal to the poles). The transfer function is lOO/(jcr.5 Gain frequency response (b) corresponding to the pole 5 on the splane (a) . ofpurelysinusoidaltestsignals. when the frequency doubles. A2.and which are not transmitted at all (exponential signals withtheexponentsequaltothezeros). Consider the transfer function L(s) = 50/(s gain frequency response is the plot of the function + 5).3. and the output is delayed by 45".. (a) (b) Fig.Appendix 2 365 Example 3. A2. Le.the vector becomes longer and eventually the gain coefficient decreases in proportion to the frequency. + 5). A2. If the input signal is Re[exp@~~~)J. and at this coo the Laplace transfer +jtp)]. the gain coefficient is halved. the function equals 14. I t is seen that s replacesjcr. In other function is ALq.. theformerissueis related'. the gain is 20dB andthephaseis 0. The 20 logll(jo)l= 20 log[SO/(j~ + 5)]. 0 Theyaffectthetransmission responses. That is. increases byan octave. decreaes by 6 dB.to system stability. then the output signalof the link will be ARe[exp(joo words. the output amplitude is 14. the signal is amplified A times and the phase is shifted (advanced)by (o radians. In the example shown in Fig. the Laplace transfer function is l/(h4s2). Therefore. At zero frequency. A 2 .5(a).
and IL(jo)lbecomes large.l) is. The frequency response is obtained by substitutingjo for s in the function.366 ' Appendix 2 When o changes from0 to . With a further increase of the frequencythedenominatorgetsbiggerandthemodulus of thefunctiondecreases. 0. A2. the higher is the peak K(jo)l decreases as the on the gainfiequency response. Example 3. /ao represents the damping coefficient for the pole. A2. the two vectors represent. square of the frequency.(0.. s +: a = s2 + 2c00s + 0:. is negative and 6 = 4 . Here.(0.(0. Since the vectors corresponding to the pole and to the zero are nearly the same. the phase changes from zero 4 to2 .6(a)showsthepolesonthe splane where c0is negative and small. consider a polezero pair with. Then. denominator becomes 200a0. )I[o .rt joo.Thetransferfunctionmultipliersforcomplexpoles(orzeros)are commonly written as [s. .At higher frequencies o >> oo. A2. And. 0.6(b) shows the gain frequency response for the function. . the . Fig. + ja. 2Olog(dq2) 0 Fig. +joo)J [s.theirpoles andzerosareeitherrealorcomeincomplexconjugatepairs. dominant poles and' zeros Next.and the gain decreases with constant slope 12 dB/oct. the first multiplier becomes a .20.s +  [s. The transfer function c L(s) = c s2 + 2<o. the multipliers in the denominator of the function.ja. there will be no substantial differencein these links' performance in any application. 11 hastwocomplexconjugatepoles. when o approaches a .7. . since the frequency response fully characterizes the performance of a linear link. Laplace rational transform functions for physical systems always have only real coefficients.Therefore.A2. as shown in Fig.6 Polezero cancellation. A2. Adding sucha pair to any transfer function will certainly change the order of the system but will have negligible effect on the frequency response.)J= s2 . ja. little distance between the pole and the zero.(0.6 (a) Complex poles and (b) the related frequency response The smaller the damping coefficient (and therefore lo. Fig. they compensate each other and insignificantly affect the frequency response. .
) Find the Laplace transforms of the signals: (a) &function. (d) 1 3sin(34f + 5). The closer the poles are to thejoaxis. What is the phasor of the sum: o f + 12) 12sin(o f + 12). Therefore. then the distance between the singularities must be very small for the replacement 0 to be adequate. (e) 4 L 60'. If thecluster is farfrom jmaxis s~lane the frequency range of interest on the joaxis.(c) 5sin(5t. (d) *cos( ot + 2). "7 P ~ l e .40L 60'. Such poles and zeros are called dominant. and only one or a few poles and zeros have prominent effect on the frequency response. . pye A2.8 Problems Can the system be linear if the input signal is 3sin(34 f + 5) and the output is: (a) 3sin(5f + 3 4 ) . the poles of the transfer function characterize the exponents of the solution toa system of certain linear differential equations. (c) 40L 229'. (e) 2.72f + 2. (b) s 5)(s + lo)] (Do it using Table A2. (e) 2.5).5). (f) 4 L 60' + 40L 30'.72L2.7 Timeresponses In the calculation of timeresponses.20L . (f) 3sin(340f . .(b) 3sin(34t+ 34).5). (e) 13(1 . (d) 44(s + Find the originals of the Laplace transforms: (a) l/s. (b) 2sin( o f + 2) (c) 3sin(of + 12).~ pair e~o interest. several poles and zeros mutuallycompensateorhavesmalleffect on the transfer function within the frequency range of Fig.1. (c) 2/(s + 3). What is the phasor for the function (express the phase in radians): (a) 22sin( o f+ 12). (b) 12sin(of + 12) 1 Osin( (c) 4 L 30'.72cos(o f + 2). In many systems. Much more about frequency responses' relation to the poles and zeros is explained in Chapter6.72). (a) 22sin( o f + 12). and using the MATLAB function invlaplace. (9) 58in(34f . (d) 1OOOL 30'. (d) 10fe2'. A2.72sin(2. (d) 4 L 30' + 40L 60'. the closedloop responses should not have poles too close to the joaxis in order that the output time history not differ substantially from the command.e43.72'. (c) 5f2. But if the cluster is very close to the imc frequencies of interest. (e) 2. (b) stepfunction with the value of the step equal to 8.2 Appendix 367 For example.40L 30'. (f)6sin2f. (b) 40L 59'. ' 0  What is the transfer function if the input signal phasor is 4L 49' and the output is: (a) 4 L 49'. the more oscillatory the solution is. the zero replacement is adequate even when the cluster is not very tight.a cluster of two poles and one zero canbereplaced by onepole.
(b) in parallel. (d) 2O(s + Q)(s+ 66)/[(~ + 6 3 ) (+ ~9 5 ) ( ~ + 2 4 0 ) (+ ~ 2700)]. (e) 2. 2O(s + 2 ) ( + ~ 26)/[(~ + 85)(~ + 250)]. (c) with the first and the second in parallel and the third in cascade. (e) 6O(s + 3)(s 16)/[(s+ 33)(s+ 75)(s + 200)(s+ 2000)l.2s + 20)/(s4+ 2 .72).) (0.72).72(s + 7)(s + lO)/[(s+ 70)(s + 13O)(s+ 1200)]. + 2 0 8 + 300). (c) ( 1 0 8 + 1Os +40)/(2s4 2s3 + 3 8 + 80). (f) (si! + 60s + 100)/(2s3 Find the link’s transfer function. (b) (si! + 30s + 4)/(s + 2s3 + 2 8 + 3). (d) (s? + 20s + 5)/(s + 5s3+ 8 + 3).) Appendix 2 (Use both the Table A2. (b) (8 + 3s + 24)/(2s3+ 1 2 8 + 60).368 (9) 4e4‘C0S5t. (d) (8 + 20s + 5)/(s + 5s3+ s? + 3). and are connected (a) in series (cascaded). (c) (8 + 4s + 42)/(2s3+ 1 2 8 + 80). 7 2 ~ +~ 7 8 + 2. (c) 1O(s+ 8)(s + 62)/[(s+ 50)(s + 65)(s+ 150)]. 12)/[(s+ 50)(s + 85)(s + llO)(s+ 1200)l.    10 Use the MATLAB command root to calculate the poles and z m s of the function: (a) (208 + 30s + 40)/(2s4 + s3 + 5 8 + 36).1 and the MATLAB function d@)/dt. 7 2 ~ +~ 7 8 + 2. (c) (10 8 + 1Os + 40)/(2s4+ 2s3 + 8 + 3). (si! + 30s + 4)/(s3+ 2 8 + 30). (f) (8 + 10s + 8)/(s4 + 12s3+ 1 2 8 + 130).1. (f) 1O(S + 2 ) ( ~22)/[(~ + 4 0 ) (+ ~ 6 5 ) (+ ~ 1SO)].728 + 27. respectively.2s + 20)/(s4+ 2 . 13 Three links with transfer functions. (d) (s4+ 5s3 + 8 + 20s + 5)/(s3 + 2 3 8 + 200s + 300).728 + 27. the output signal Laplace transform is (a) (8 + 2s + 4)/(2s3+ 4 8 + 30). (e) (8 + 50s + 40)/(2s3+ 2 2 8 + 120). (9) 2O(s + 2 ) ( + ~26)/[(~ +43)(~ + 85)(~ + 2 5 0 ) (+ ~ 2500)l. (b) 6O(s + 7)(s + 15)/[(s+ 53)(s+ 95)(s + 210)(s+ 2300)l. 9 Which of the following expressions can be transfer functions of stable systems: (a) (208 + 30s + 40)/J(s + 43)(s + 85)(s + 250)(s + 2500)l. (e) (2. (f) (8 + 1OS + 8)/(s4+ 12s3+ 1 2 8 + 33). (b) (8 + 30s + 4 ) / ( 2 ~ 2 8 + 30). (c) (10 8 + 1Os + 40)/(2s4+ 2s3 + 8 + 3). (d) with the first and the second in (I 08+ 1OS + 40)/(s4+ 2s3 + 3 0 8 + 1000) . (b) integral of 7 Find transfer function of the linear operators: (a) (c) double integral of fit) (Use TableA2. (b) (8 + 30s + 4)/(s + 2s3 + 2 2 + 36). 12 Write the frequency response function by replacing s by for: (a) (208 + 30s + 40)/(2s4 + s3 + 8 + 3). 11 Use the MATLABcommand poly to convert the following functions to ratios of polynomials: (a) 50(s + 6)(s +. 8 The input signal is the &function. (e) (2. (h) 4cos2t laplace. (d) (8 + 40s + 46)/(2s3+ 2 8 + 90).
Draw a conclusion. S 3 ) (+ ~ 16)/[(~ + 3 3 ) (+ ~ 7 5 ) (+ ~2 0 0 ) ( + ~ 2000)]. What is the value of the function at dc (Le. using functions: (a) (s+ l)/(s + 10). (b) ~ O ( + (c) 1 O ( s+ 2)(s + 22)/[(s + 40)(s+ 65)(s + 150)].1)]. (b) 1 O/(S + 10). (c) (s+ 2)/(s+ and (s+ 2)/[(s+ 7)(s + 14)j. (e) I 0 0 / ( 8 + 2s + 100). 15 Use MATLAB to plot the frequency response and the step timeresponse for the first. (b) (s+ 2)/(s + and (s+ 2)/[(s+ 9)(s + 1l)].Appendix . 5 + ~7. (c) (s+ 2 /(s+ and (s + 2)/[(s+ 7)(s+ 14)J. (b) 1 OO/(s + 1 O)2. / (10)~. (c) 100/(8 + 4 s + 100). 19 Plot and compare frequency responses of the functions with complex poles and zeros: (a) (s+ 2)/(s+ 10) and (s+ 2)(s + 5)/[(s + lO)(s + 5.58 ~ + 2. (b) (s+ 2)/(s + and (s+ 2)/[(s + 9)(s + 111. MATLAB toconvertthefunctiontoaratio of polynomials and plot the 16 Use frequency response for the function: (a) 50(s + 3)(s + 12)4(s+ 30)(s + 55)(s + 1OO)(s + lOOO)].4 + 25s + 20)/(s4+ 2s3 + 8 8 + 3). (d) (2. ~ (c) I O O O+ Describe the correlation between the slope of the gainfrequency response and the phase shift. hold on feature) of the .5).and secondorder functions: (a) lOl(s610). 14 Use MATLAB to plot the frequency response for the first.. when s = O)? What does this function degenerate into at very high frequencies? I 17 Compare the frequency responses of: (a) (s+ 2)/(s+ 10) and (s + 2)(s + 5)/[(s + 1O)(s + 5. (d) ( 2 8 + 22s + 20)/(s4+ 2 . (e) 2. and thirdorder functions: (a) lO/(s+ 10). Draw a conclusion. Describe the correlation beween the shapes of the frequency responses and those of the stepresponses. Find the transfer functions of the resulting composite links.1)]. second. (f) . 2 in the previous example. 18 Plot timeresponses to a step command for the functions Draw a conclusion. (d) 1 00/(8 + 4 s + 100). (d) 2O(s + 2 ) (+ ~26)/[(~ +43)(~ + 8 5 ) (+ ~2 5 0 ) ( + ~ 2500)l. 20 Plot the frequency responses (on the same plot.72(s + 7)(s+ 20)/[(s + lO)(s + 1OO)(s + lOOO)].2 5 ( ~ + 2 ) (+ ~44)/[(~ +55)(+ ~ 6 6 ) (+ ~7 7 ) (+ ~ 8800)l. (f) loo/(&? + s + 100).2 369 series and with the third in parallel to this composite link.
10 at high frequencies. I 22 Which of the following functions are positive real? (a) (s+ 2)(s+ 22)/[(s+ 40)(s + 65)(s + 150)l. rolling down inversely proportionally to the frequency at higher frequencies. (b) 10 athighfrequencies. flat at low frequencies and rolling down inversely proportionally to the square of the frequency at higher frequencies. rolling down inversely proportionally to the frequency at higher frequencies. (f) resonance peak response.   . (d) (s+ 2)(s + 22)4(s + 40)(s + 65)(s + 150)(s+ 250)]. flat at high frequencies and rolling up proportionally to the square of the frequency at higher frequencies. 100 at high frequencies. with different behavior at lower and higher frequencies. (e) 10 at medium frequencies.370 Appendix 2 (b) (s+ l)/(s+ 20). (b) (S + 2 ) ( ~22)4(~ +40)(~ +65)(+ ~ 150)].rollingupatlowfrequenciesproportionallytothe frequency. (9) resonance peak response. (c) (S + 2 ) ( + ~ 2 2 ) 4 (+ ~40)(~ 65)(+ ~ 1 SO)]. (d) (s+ 1 O)/(s + 1000). rolling up proportionally to the frequency at lower frequencies. rolling down at higher frequencies inversely proportionally to the frequency. Over what frequency range do the functions approximate the differentiator s ? 21 Make a transfer function to implement the following frequency response: (a) 10 at lower frequencies. (9) (s + 2 ) ( + ~ 22)/[(~ +40)(~ +65)(+ ~ 1 SO)]. (i) notch responses of three different kinds. (h) resonance peak response. (e) (s+ 2)(s+ 22)/[(s+ 40)(s + SS)]. rolling up proportionally to the frequency at lower frequencies. (d) 100 at low frequencies. (f) (s+ 2)(s + 22)(s + 1 50)/[(s+ 40)(s + 1 SO)]. (c) (s+ 1 ) / ( s+ 1 000). (c) 10 at low frequencies.
i. capacitances. There are no poles or zeros in the right halfplane s.r. This cannot be a property of a passive system. if the following three conditions are satisfied: A. and inductors connected in series and in parallel. Therefore.. In some cases. and the ratio of the sensor readings to the actuator action is the driving point impedance or admittance. Collocated controlis a feedback control of a passive plant where the actuator and the sensor are collocated. is A transfer function pole in the right halfplane means that the response to a finite input signal is an exponentially rising output signal..r.Poles and zeros on the jcoaxis are single. it is always possible to make a system composed of passive elements whose driving point impedance is the prescribed p.Re 6(iw) > 0 at all frequencies. but are not necessarily so. function. The function isp. This limits the range of the plant phase variations 180'. function can be realized (i. realization requires bridgetype circuits or transformers. All passive driving point impedances (a driving point impedance of the voltage to the current at the same port)are positive real(p. to Any p. The transfer impedances and admittancesof passive systems can be p. Le.). C. but are not necessarily so. For some p.e. Positive realness of a function 6(s) means that for all s in the right halfplane.Appendix 3 37 1 Appendix 3 Causal systems.A p. a system without sources of energy.r. Causal systems are those systems whose output value at anytl does not depend on the input signalat ? > 81. or a derivative or an integral of the impedance.e. The driving point impedancesof active systems can be p. the transfer function . . All stable systems are causal..r.r. the system is an arrangement of resistances. passive systems must have no transfer function poles in the right halfplane of s. Such an impedance hasnonnegativeresistanceforallsinusoidalsignalsandpositiveresistance for all risingexponentialsignals(whetheroscillatingornot).r. functioncanalso be realizedasadrivingpoint impedance of an active RC circuit. passive systems.r. however. is the ratio Re6(s) > 0.r. implemented) as a driving point impedance of a passive system. and positive real functions When s is equal to a pole. of B.r.impedancedoesnot generate but only dissipates power.r. The driving point admittancesof passive systemsare also p. functions.A p.
.Theintegralalongthelargearc equals dm where. the integral of the even part of 91s along thejaaxis equals 0 Fig.e. is m. then 9/s is analytical in the right halfplane of s and on the jaaxis. The integral along the arc equals IC&. A4. is the value of 9 at zero frequency.1). excepting the origin. of A AIl.p. The contour of integration may be viewed as composed of the jaaxis completed by a xradian arcof infinite radius R as shown in Fig. it accepts the form (A4.is the value of 9 at very high frequencies. as follows from (A4. So if the origin is avoided along the n:radian arc of infinitesimal radius as shown in Fig. 9(s) = A(s) function and B(s) is an odd function. the integral of 91s around the contour enclosing the right halfplane equals 0. On the jaaxis..2) A4." " " " 4 s A2 s2 B3 A4 s4 s3 . Therefore.2 Contour on the splane avoiding a pole at the origin . A.A. A4. where A(s) is an even % . Theintegralalongthesmallarc equals do where A.1 Integral of the real part The Laurent expansion at s + for an m. The expansion converges over the entire right halfplane of s.2 Integral of the imaginary part If 9 is m. along the positive semiaxis.1. is 00 + B(s).around the right halfplane of s equals 0. A4. the contour integral of 9 . the integral along the whole jaaxis equals twice the integralof the even part of the integrand.p. Therefore. A4. 0 2n: 4 (A4.p. as well.2.f. i..372 Appendix 4 Appendix 4 Derivation of Bode integrals A4. Therefore 0 0 Fig.1 Contour on the splane j(A%)dCO=..1) The function9 AIl.
€I(jo.and also define 0. " Re 6A.6) . function ReWorn) must be even and Im For those W reducing at higher frequencies to s. 2% "s ac 2 ' (A4. The sign of the radix at the sections must be chosen so that the whole contour of integration belongs to only one of the Riemann folds. W is purely imaginary. 1 1 0 (A43 A4.e. . Its real part is certainly zero: A m . Consider the function of s 1 s / j " w .Appendix 4 373 09 00 . On this contour.where A.4) &= d z isrealfor l o 1< 1 and i. The function JsW therefore alternates between being purely real or purely imaginary on adjoining sections over the joaxis.A. imaginary otherwise.e. The function SW is purely real.4) yields sw W = (1 + s2)/s. W is a setting to zero the integral of 8 reactance fu/WOn. 2 09 (A4. inparticular. positive or negative. +jB. i.A. the integral along the joaxis equals zeroas well. (A4.).and the integral along the large arc vanishes. Since the total contour integral is zero. )=(6Ac) s/j+o. on thejwaxis. JB(u) du = A. 00 sw " @A. and Bc are real.3 Generalrelation Let's define the frequency atwhich the phase shift is of interest as a .animpedancefunctionof a reactancetwopole. the Wurn).~ . 0=l 0=0 (A . either positive or negative. Another important relation between the real and imaginary components results from around the same contour. 0 then If.. On the j o axis. It has branch points at the joints of these sections. drn=2jRedo=O. This relation is known as the phase integral. odd. here.)darcsinrn = " B IJ_"d".3) where u = ln(o/oc). A. and (A4. the integrand decreases with s at least ass .
) d o . Since = 2o 2I do 2 ( A .A.o C ) . (A4.7) equals Bc = "[(A 7c . 2.8) where we denote: u 201. equals I slj Bds = " . we have oo j(AAc) 0 o2ac 2oc do=27cBc and finally B. . we see that theintegralalongthe joaxis equals 27cB. (A4. A4. As s approaches jo.) (A4. A4. . . . Therefore. and the secondmultiplier intheleftside of (A4. Next. equating the sum of all components of the contour integral to 0.0. the integral of the hnction taken around the contour shown in Fig. The integral along the arc of infinite radiusR equals 0 because oftheterm s2 inthedenominator of the integrand.A. thes u m of the integrals along these pieces is 0. . approaches Bc .j o .3 Contour on splane avoiding The integral along the small arc centered at jo..6) tendsto l/(s/j. .374 Appendix 4 This function is analytical in the right halfplane of s and on the jwaxis except at the points j o . equals imaginary zB. 6 . do v = 1o2 = 1ncoth 0.7) n oo2oc j% It 2oc d o =IUdV=UV(Vdu 02o. as well. The contour consists of several pieces. u = ln(o/o. Then the integral along the infinitesimal arc centered atja. I ui 2 IT 5 du 0 2 (A4.9) . Neglectingtheoddcomponent of the integrand whose integral is annihilated within these symmetrical boundaries.A c ) l n c o t h ~= ~1' a19 lncothdu. Fig.3 equals zero.
Hence.Appendix 4 375 The left side of the equation is real. B(o.10) . 2 w (A4. Since lncothU = jn+lncoth2 2’ U replacing In coth(u/2) by In cothlu/2l does not change the real components of (A4. lncothdu.9) and is therefore permitted.) = 1 =dA I ul rr j. We can therefore count the real components only.i. the function in square brackets becomes even andis annihilated becauseof symmetrical limits. Thus. the imaginary components on the right side are annihilated after summing. After the replacement.
Behavior of the magnitude outside the specified frequency interval is assumed to be an asymptotic extension of the response within the specified interval.m takes fiequency andmagnitudevectors as inputsandreturnsthe phase. tablemaker % calls the function creating the table load table % load data needed for toe and tail calculations: % table con hilimit lolimit numentries numintstep points = length(natfreq). The integrals of the tails are precomputed and stored in a lookup table createdby tablemaker. % magdb: row gain vector given in dB % freq: row frequency vector given in rad/sec % The Magnitude and Frequency vectors must be the same len % Before running this function prepare table tablemker.m. [row. The accuracy of the finite difference approximation of theintegralusedinthe routines is sufficient. a system that has 40 dB/decade roll off within the specified fiequency rangeis assumed rolling off at the same rate at higher fiequencies. toeslope = (magdb(2) . The programs' listing follows: . if col== 1. natfreq) % function [phase]= findshase2(magdb . integralu.magdb(numsteps)) / (flnfreq log(natfreq(numsteps))1.1 degrees. numsteps = points .magdb(1))/ (log(natfreq(2) 1ilnfreq) .p. As the number of data points increases. When the input is 100 logarithmically spaced points spread over 4 decades.m While these could be computed when needed. . The magnitude is prescribed over a finite frequency interval.376 Appendix 5 Appendix 5 Program for phase calculation The integral (A4.freq) % This routine uses the Bode Integral to generate phase dat % from a magnitude response of a m.Theerrorsintroduced by such handlingofthetailsmightbecomedominantwhenthe highfrequency andlowfrequency asymptotes are not well developed within the prescribed frequency interval. Functionfind_phuse2 function [phase]= findshase2(magdb.10) can findshase2. flnfreq = log(natfreq(p0ints) 1.withmore ofthemwherethe known function is rapidly changing. mag& = magdb'. accuracy improves. be calculated with MATLAB using numerical routines tablemaker and. planton the basis of. ifcol == 1. These three are among several routines for calculation of different Bode integrals written by Michael Kantner from Caltech. By placingdatapointsunevenly. tailslope = (magdb(points) . Thatis.p.1. The routines can be used to calculate the phase shift response of a physical m. end. The function findshase2. The functions have been tested on a variety of transfer functions. natfreq = natfreq'. end.col] = size(magdb).coll = size(natfreq1.m.its measured gain response. accuracy can be improved with minimal added computational cost.thelookuptableyieldsfastercode. function. with [row. %%% % The following variables are for the lookuptable (u domain) ilnfreq = log(natfreq(1)). typical errors are under 0.
m. looplnfreq = log(natfreq(i)).1)).l]table(ind'./dnatfreq.2) 'J = integralu(lel5..2))..loglO(hilimit). weights = (w2+ wc) .lolimit.numsteps) function [trapl=integral~u(u~.u2. wc = natfreq(i)*ones(wl).u2. (to tail lolimit = le9. .(wl + wc) .w2 = natfreq(2:points)t. for k = 2:length(vector). ind = (loglO(abs(u))+[loglO(lolimit)loglO(lolimit)])/con+[l 11. % number of steps for each integration vector =.l). numentries = 4001.numsteps) . should be odd numintstep = 100.*log(wl + wc) +.indl).natfreq(1:numsteps).table(k. called by function find~hase2.magdb(1:numsteps).1) = vector' .[looplnfreq]*[l 11 + [eps eps].indl). numintstep).nentries). [table(l.*log(abs(wcw2+eps)) . (wcw2). phase(i) = (deriv*weights + [tailslope toeslope]*tailtoe)/pi. deriv = dmagdb. % The next lines perform the integration .(wcwl). end Function tablemker. ~t~le~k. % number of table entries. table ( :. for I = l. u = [flnfreq ilnfreq]. ind = max([l l. end clear vector k con = logl0(table(7.Appendix 5 377 dmagdb = magdb(2:points) .points.This routine creates the lookup table for calculation the toe and le15 i s used for zero..rn % % % % % function tablemaker table_maker.*log(w2 + wc) . hilimit = 100. nnfreq = natfreq(1:numsteps).ind=min([numentriesnumentries. 650 is used for infinity reasonable limits: le9 to 100 This contains nearly all of the area 6 places).*log(abs(wcwl+eps)). numintstep). dnatfreq = natfreq(2:points) . wl = nnfreq'.2~l=table~kl.l). logspace(1og10(lolimit). save table con hilimit lolimit numentries numintstep table clear con hilimit lolimit numentries numintstep table Function integralu % function [trapl=integral~u(u~.1))loglO~table(6. tailtoe = abs(piA2/4*[1.~~+integral~u(table(kl.
numsteps+l). (ul==O) . the phase response corresponding boniqas . delta = vector(2:numsteps+l)vector(l:numsteps). to a gain response can be found using function . % It may not handle some special cases./(exp(vector)l))). %up = valvector (1:numsteps) *delta’ . ul=le15. For the function without highQ resonance peaks. %down = valvector(2:numsteps+l)*delta’.loglO(u2). u2=le15. % vector=linspace calculation(ul.Qm. disp(’Warning: NaN found in integration.5*(valvector(l:numsteps)+valvector(2:n~teps+l~~~*delta~. if trap==NaN.numsteps+l).u2. result set0‘) to end Different MATLAB programs for Bode integral calculations are listed in [6]. trap=(. however. end vector = logspace(logl0(ul). % This next calculation is the logufunction if valvector = log(abs((exp(vector)+l). end if (u2==0).m from Appendix 14. trap=O.378 Appendix 5 % This routeens integrates u in domain.
a more general description of the feedback system canbe used as a connection of a unilateral twoport w to a passive fourport B as displayed in A6. the return voltage U 3 appears at the port 3'. The return ratiois (A6.thedimensionality of thesignals at theinputandoutput of the amplifier and at the input and output terminals of the whole system does not influence thefollowinganalysis. andthesignalat its output by thecurrent 1 4 .However. Further.theamplifierinputand output impedances may be imitated by connecting passive twopoles in parallel or in series to the amplifier's input and output. Inprinciple. A6. A6. The activeelementwith transadmittance w = 14/E3 is therefore assumed to have high input and output impedances. we selectone of the possible versions and characterize the signal at the input to the active element by the voltage El. When the circuit diagram in Fig. When an external emf El is applied to the input of the amplifier disconneded from the Bcircuit.1 is appliedtotheanalysis of circuits withphysicalamplifiers. Fig. as shown in Fig.infinite or infinitesimal input and output impedances.Appendix 6 379 Appendix 6 Generic singleloop feedback system When a feedback system cannot be easily broken into a connection of links. Fig. these twopoles have be to integrated into the Bnetwork.1.tosimplifytheexposition.1) Fig. A 6 9 Disconnected feedback loop .1 Singleloop feedback system The twoport w is assumed to have zero reverse transmission and either.2. A6.
The output of the feedback system in Fig. i. the signal U3 is formed by superposition of the effects U. relative toE3. and as the ratio of the output voltage to the signal emf are. Hence.” u3=. whence Thus U3 = U.380 Appendix 6 When the return ratio T is. .. In particular.T(0)denotes the value of T measured while the input is shorted. the closedloop system transfer coefficients in voltage.3 Disconnecting feedback loop in the feedback path Generally. the twopole shunted by the emf E. 2’and 2 connected to the input and output of the broken loop provide appropriate loadingforthedisconnectedparts of theBcircuitinordertokeeptheirtransfer coefficients unchanged. Thus. A6. current. T = T(Z1). In the closedloop system. F = T + 1 is the return difference.The ratio Z’ need not be connected since it is equals 7’. and the ratio of the 1 when load impedance to the fourpole input impedance. the difference between the signals U3 and E.thevoltageoutputinputratio of a linearfourpolecanalwaysbe presented as the productof two ratios: the current outputinput ratio.while the inputis open. By virtue of (A6.being measured. and T().When T is measured this way.1 is a linear combination oftwo signal sources: the output of the amplifier and the signal source. respectively. the source impedance return ratio therefore depends on Zl. inaccordancewiththe & is connected to the port 1. the emf El of the source connected to thesystem’sinput 1 mustbereplaced by a shortcircuit.3 shows the crosssectioned feedback circuit. The latter ratio is found to be T is calculated. A6. T may be measured arbitrarily as the ratio of either voltages or currents.U3T. A6. The emf E6 applied to the input of the broken loop produces return signal Us.2) here.. u 3 0 F (A6.”and UJproduced respectively by the signal source and the output of the amplifier.2). the signal from the amplifier output is reduced I. Therefore. times.e. The external twopoles Fig. Le. Fig. . The superposition principle..
F(O). with the port terminals open.9) can be used for calculation of driving point impedance atany port n.4)into (A6. with the active element killed out).. If the terminalsare shorted. (A6. F(). or. impedance Z1to the system’s input terminals.anytwonodes of theBcircuitcanberegarded as input terminals of the feedback system.infinite impedance.8) into (A6. but not the current. Since. the input impedance in the system without feedback (with a crosssectioned feedback path. and return ratios T(0)and T(). i.3) (A6. the formula (A6. and km are the coefficients of direct signal propagation through the Bcircuit and are determined under the same set of loading conditions at the input terminals.9) The formula expresses 2 through the three easily calculated functions: 2 .7) 22 KOLI= KI . and F = F(&) arethereturn differencesmeasuredundertheconditions of connectingzeroimpedance. 2 0 (A6. and Z .Appendix 6 38 1 (146.3). kid.7). KOL.we get from which Blackman’s formula follows: (A6. andF(). and (A6.inprinciple.8) By substituting (A6. the voltage between them vanishes. or with w = 0.4) (A6. provided that F(0) is understood as F measured with the port n shorted.5) Here.e.KOL1. Then (A6. measured whilethefeedbackpath is disconnected. and KOLEare the open loop system transmission functions.6) (A6. respectively. Let 2 designate the input impedance. the feedback is called currenfmode (or series) with respect to the . For this reason.6). and kd.
the feedback is voltagemode (or parallel) if T(0)= 0 and T() = 0.9]. the feedbackis called compound. If neither of them equals 0. More detail along these lines is given in [2.382 Appendix 6 terminals n if with respect to these terminals T(0)= 0 and T() = 0. . Analogously. the input impedance is (A6 10) It depends exclusively on the Bcircuit and not onw. If the feedbackis infinite.
e.. +1 The formula expresses activemember mobility with feedback.5) results in Blackman’s formula as z =2 . and E = 0 when it is open. The return ratio T of the active member feedback loop is defined as the ratio of the return signaltothenegative of theinputsignal. (A7. Thus. i. The relative velocity V across the active member and the feedback return signal E.e. In this case. and d are the constants to be determined from boundary conditions.1) Here a... First.5) Comparing the obtained equations (A7.c)]F so that the active member mobilitywith feedback. Consider a mechanical system with an actuator accessed via a single mechanical port. E = 0.and T(oQ). This ratio iscertainlya function of the mobility of the structure ZL. To the active ZL.. induced in the active member. V = 0. at the end of the disconnected feedback loop can be each expressed as a linear function of the forceF and the signal E: V = aE + CE bF.c) (A7.ad/b)E.2) When the loop is closed. The case without feedback.becomes Zo=V/F= b (A7. the active member mobility without feedback.T(O). whenthe active member isclampedandwhenit is free to expand. when ZL = OQ.findtheexpressionsforthereturnratioforthe two differentloading conditions.gives V = bF. zero force is On the other hand.3) Second. For the purpose of analysis. i. Er= + dF (A7. The force F is member. the active member is connected.. (A7. = (c . the activemember is free to expand.the active member is rigidly constrained.. E. When ZL= 0.e. consider an active structural member including an actuator.astructureisconnected.e. T ( 0 )+ 1 T() (A7.1) gives V = [b + ad/( 1 . Notice that when the feedback loopis closed. the return ratio becomes T(O)=c+adlb.Thestructure’smobilityis measured between the active member and the structure.2.Appendix 7 383 Appendix 7 Effect of feedbackon mobility The following derivation of Blackman’s formula follows Blackman’s original proof. Hence. (447. = E. consider disconnecting the feedback loop at the input to the actuator and applying signal E to the actuator input. Zo.. case = cE and the return ratio is T(~)=c. i. however in mechanical terms [21]. T(ZL).e.4) i. in terms of three that do not depend on the structural system to which other functions. i.Zo.e. In this E. becomes Z = b + ad/(l . find the expressions for the mobility without and with feedback. . That is. T = Er/E.1) givesF = a/(bE) and E. c. i. E = E.2A7. b.2.
. for an electrical system. inductance of an inductor.Le.384 Appendix 8 Appendix 8 Dependence of a function on a s parameter The formulas derived in the two previous sections are bilinear. Generally. a Laplace transform transfer function of a physical linear system can be presentedas A/&.g. This function can be expressed as and can be used to analyze the effects of the plant uncertainty. where the main determinant A and the minor & are linear functions of the value of an element of the system (e..returnratioisabilinearfunction of thefeedbacksystemload impedance 2 . can be presented as a ratio of two linear functions. The proof can be found in [2]. gain coefficient of an amplifier). resistance of a resistor. . capacitance of a capacitor. Forexample. .
each one of the following equations z =z .the transmission from the output of the amplifier to each of the system output terminals is the same. the first condition from (A9. This are the conditions of balanced bfjdge.TheWheatstonebridge is balancedif Za/& = Zc/zd. I validates the other two. The desired output impedance of the amplifier (actuator)z d can be implemented by using a nested actuator feedback loop.Appendix 9 385 Appendix 9 Balanced bridge feedback As is seen from Blackman’s formula (A6. Fig. . Therefore.1) is satisfied. A9.A9. and connecting these terminals by any external impedance will not change the return ratio. Inthiscase. An examplecanbedrawn by inverting the direction of the amplifier in Fig A9.In such a system the feedback is not dependent on the external impedance. An example of a balanced bridge circuit is showninFig. the voltage across the bridge diagonal is zero.1 Balanced Zero transmission between system’s the output terminals and the input of the amplifier is bridge feedback of similaruse.1.9).1. and therefore the value of the feedback is not limitedby this impedance variations.
Let us determine the relations between the real and imaginary components of the describing function of a nonlinear link: Here. and the frequency response of the real part of the differentially linearized circuit A. for the continuously differentiable function A (the differentially linearized circuitmustbem.and amplitude of the output signal fundamental.lndU1. When.whichwillbe clearlyseenontheresults of themeasurements). The derivative is dN U dU 1 = exp(8 . we COS [ p.2) is rather complicated. For practical design applications.. Uz isthecomplex U. Assume that the conjunction of filter is applicable to the problem considered. by applying the Bode relation to the differentially linearized stable see that "  . I l _ ( i i .1 =adlnU.andnotcontainingsingularitiesonthe@axis.2)canbe determined with sufficient accuracy from the measurement results of A on a limited number of frequencies. The latter must be measured while simultaneously The applying to the link's input the main signal with amplitude U and frequency ao. the nonlinear link depends on frequency.n= J ~(u du l~"'lncoth 2~ d l .p. m.exp( A .386 Appendix 10 Appendix 10 Phasegain relation for describing functions For a describingfunction of a systemcomposedoflinearlinksand a nonlinear nondynamic link. however. This is the case of the isow curves. link. oo) can be calculated from the measurements of: the real component of the describing function H. Therefore. the signal amplitude at the input to the nonlinear link does not depend on frequency. dReH . Reits derivative. d lnU. Le.p. the relation B(A) becomes more complicated. ReHA) (A10. on the level of the inputsignal&e fld In&.2) the phase shift < p ( U1.H) . isow responses are preferable and should be used whenever possible.) de R In xe p U H (. U1 is theamplitude of thesinusoidalinputsignal.9) 1 . the relation between the gain and phase frequency responses B(A) is the same as in linear systems if the D l ? of the nonlinear link stays the same for all frequencies.2) Using (A10.theintegralin(A10. where the logarithmic transfer function for increments 8 = A +jB = lndU2 . dU. and dU1is real. . The relation (A10. and. magnitude A is easy to measure witha freequencyselective voltmeter if at this frequency no nonlinearproducthas a substantialamplitude. .Re H) COS( B .Sincethepower of theset of frequencies of these components is small in comparison with the power of all real frequencies.
Since the devices were relatively expensive and rather difficult to tune.and plant parameter variations? A: That sort of approach typically produces disastrous results in engineering practice. The Bell Laboratories scientists developed synthesis theory for highorder compensators to approximate the theoretical best. The old way of compensator implementation used the same technology as that for the actuation. it is stated that modern technology somehow relates to highorder compensators. an integrator device. and pneumatic gadgets whose outputs are proportional to the integralor to the derivative of the inputs. highorder compensators can be economically implemented in hardwareorsoftware. The conversion of most practical specifications is simple and transparent. the compensators remained loworder. The benefits of frequencydomain design far outweigh the trouble of converting the timedomain specifications. . they are not suitable to handle highorder controllers. these are condensed transcripts of conversations which took place with the authors’ colleagues and students. nonlinearities. however. There exist mechanical. all connected in parallel. On the other hand. The wayof themechanicalcontrollerimplementationimpactedthecontrol theory. andthecompensators are electrical. feedback amplifier technology was not limited to loworder compensators since small capacitors. Convenient methods have been developed for loworder controller design however. to whom we extend our sincere gratitude. is it advisable to optimize the performance while ignoring some details like noise. Nowadays. theinputs of theactuators are electrical. All. and a differentiator device. Teaching the control at the elementary level. resistors and chokes used in the compensators were cheap. still practicing some tools that had become not only obsolete but plain counterproductive and confusing try to make a precision controller design. Therefore. pneumatic. or else the time and resources allocated for design will be wasted on deadends. Is it worthwhile to convert them into frequencydomain specifications and then design the compensator with frequencydomain methods? A: Definitely. in mechanical. The sensor technology changed completely in 1980s and 1990s. when they had been used to Q: The specifications on the closedloop performance are often formulated in the time domain. What does the word “modern” have to do with the compensator’s order? Isn’t it used here just as an advertisement gimmick? A: It is more than the gimmick.andthecontrolcanbemadeclosetothebestpossible. hydraulic.1 Compensatorimplementation Q: In the Preface. Q: For initial design. These devices were combined to make the desired transfer function of the compensator. Even in the initial design. These theory and design methods only partially propagated into the socalled classical automatic control. In most cases. and also to address certain persistent misconceptions. the system should be addressed in its entirety. and hydraulic systems the outputs of the sensors are electrical. The mostcomplicatedamongthem were the PID controllers using a proportional device. lagged.Appendix 11 387 Appendix 11 Discussions The function of the “Discussions” sectionsis to anticipate questions and objections.
This procedure is discussed in Chapter 8. with small errors in the observed or calculated closedloop response possibly mapping into large errors in the calculation of the loop transmission.4 Elements (links) of the feedback system Q: Should the feedback path and the compensator be analog or digital? A: There are many factors to consider here. A . however. with B = 1 and without a prefilter.3 Tracking systems Q: Isn’t a common definition of a tracking system that the feedback path transmission is l? A: B = 1 is required for frequencies much lower than the crossover. cost. the parameters of an analog compensator can be made “programmable” by employing multiplying D/A converters. analog compensators are the first choice for fast feedback loops. they . that designs the system so as to satisfy the closedloop performance specifications in the presence of the worstcase parameter variations. Even rather large errors in the knowledge of C. A11. the closedloop transfer function CAP/(1 + CAPB) can be calculated. Therefore.) A11. and B can be identically 1 in a system with a prefilter. However.2 Feedback: positive and negative Q: Does the minus sign at the summer mean that the feedback is negative? A: No. and viceversa. The result is that the inverse problem is illconditioned. Q: From the transfer functions CAP and B. Although the methods involve a significant amount of calculation. (This may be unavoidable for homingtype systems. and the feedback link (and/or prefilter) is designed to provide the desired closedloop response with the nominal plant. Q: The compensator is optimized for disturbance rejection. the sign of the feedback also depends on the modulus and phase of the elements in the loop. To achieve much higher accuracy. and so forth. Is that all there is to it? A: It’s not quite that simple. When the feedback is large. For this reason. The summing device and the feedback path can be analog as long as the required relative accuracy is not better than about 0.388 Appendix 11 A11. the calculations or measurements of the closedloop response are commonly inadequate to determine the loop transfer function. and P do not cause much uncertainty in the closedloop transfer function. Many other factors will affect this decision. it should be pointed out that changing software is often neither trivial nor cheap. they should probably be digital. A digital compensator has the advantage that it can be reprogrammed in response to known changes in the plant. called quantitative feedback theory (QFT). If necessary. At any specified frequency. only the first calculation will be accurate. as it often is. But what happens to the closedloop response when the plant deviates from the nominal? A: A methodology has been developed.0001. A primary disadvantage of digital compensators is that they reduce the available feedback when the computational delay in the loop is significant. including power requirements. production quantity. the disturbance rejection and the compensator design need to be compromised to provide acceptable transient responses for the nominal plant.
Simple formulas for these calculations are given in Chapter 4.Problem 4 and the Answer to the Problems. Q: What if the plant is nonminimumphase? A: The nonminimumphase component of the loop phase lag must be compensated for by reducing the minimumphase component of the phase lag. and then determine the changes caused by the plant variations. Further. Is this correct? A: Not exactly. let’s consider the errors in the calculation. Q: What is wrong with design methods that use the closedloop response as the objective for the compensator design. more than 10dB.. the high frequency components of the command need to be sufficiently large. the appropriate value of feedback and the stability margins are not observed and preserved. equalizers have been routinely employed with 20th order transfer functions. and this information can be used to modify C such that the loop transmission function is as desired.e. Variations in the values of poles and zeros is not important as long as the frequency response doesn’t change much. modify the prefilter. in analog telecommunication systems. in the neighborhood of the crossover frequency fb.. and the feedback path are known. larger than the noise. The requirements for the control system compensators are by a long way easier. Right? A: Before deciding to do this. and the plant parameter variations may cause larger closedloop response variations than what is achievable with feedback maximization methods based on the openloop frequency response. The sensitivity of the frequency response to polynomial coefficients in typical cases is not high. Q: The command. and. if C is known. For most applications. However. or windup might occur. with the accuracy of implementation of the desired response of f0. A11. i. and the problem is illconditioned for all frequency components where the feedback is large. and the sensor is measuring the plant’s output. and the compensator order is in this aspect irrelevant. it is sufficient to first design the closedloop response with the nominal plant. Then. the command is not wellsuited to the goal of characterizing the plant.e. Q: The compensator transfer function must be high order to approximate well the required transcendental response. This can cause many problems: the system stability may be only conditional. AP can be determined exactly. rounding errors will make the system not robust. which is not usually the case. But. The sensor is not ideal (noisy).Appendix 11 389 are useful when parameter variations are very large. For example. the disturbance rejection may not be optimized.001 dB. This approach might be used for the frequency band where the feedback is positive. like the pole placement method? A: With such methods. This can be done by reducing the frequencyfb to increase the length of the Bode step. the prefilter. for this application. if necessary. i.5 Plant transfer function uncertainty Q: How accurate must the gain response approximation be for the accuracy of the related phase response to be 5 O ? A: About 112dB. . Q: Can the effects of the plant resonances on the potentially available feedback be calculated in advance? A: Sure. See Chapter 4. the poles and zeros of a highorder transfer function may be very sensitive to the polynomial coefficients. This data suffices to calculate CAP.
placingthepoles of aclosedloopsystemfar awayfromthe jwaxis doesnot guarantee global stability. the integral of negative feedback in the functional frequency band.we can always apply to its input such a signal that the system output will be as required. A practical counterexample is an active RC notch filter.reliable andwidelyused. Q:Is the definition of Nyquiststable the same as that sf conditionally stable? A: No. the statement is right. the sensitivity of the closed loop response to plant parameter variations mustbe limited. it doesn’t matter what is the actuator output impedance. process stability. Therefore. but the effects of the plant parameterdeviationsfromthe nominal are toalargeextentdependentonthe actuator output impedance. A Nyquiststable system with the usual actuator saturation nonlinearity may be conditionally stable but can be rendered globally stable by the additionof nonlinear dynamic compensation. or robustness. Is on the Nyquist diagram to reflect necessary Q: it convenient to define the boundaries and sufficient stability requirements? A: Yes. Q: Compound feedback produces some finite output impedance for the driver. and not by mathematically simpler circular boundary? A: The shape of the marginboundary is defined by globalandprocessstability requirementsand by theplantparametervariations.7 Actuator’s output impedance Q: Whatever the internal (output) impedance of the actuator.390 Appendix 11 A11.and phasemargins. process stability is not strictly achieved. The Lyapunov definition for conditional stability is that the nonlinear system stability depends on the initial conditions. All. For the nominal plant. but the stability margins in this feedback system are sufficient.by making voltage feedback and placing a resistor in series.) Also. (Often. Inthemajority of plants.6 The Nyquist stability criterion jcoaxis a Q: Is the distance of the poles of the closedloop transfer function from the better robustness measure then the stability margins on the Nyquist diagram? A: No. variations of’theplant gainand plant phase in the neighborhood of the crossover are not well correlated. A: The result willbe the same as long as the system remains linear. Q: Is much performancelost ifwe simplifythedesignmethodsandusecircular stability margin boundary? the stability margins from those appropriate to circular would A: Changing substantially reduce the integral of positive feedback in this frequency range and.Conversely. but the satisfaction of a certain norm on the errors in the nonlinear stateof operation is acceptable. Q: Why should the area surrounding the critical point be defined by gain. this can be achieved more simply. This method can be . therefore. This necessitates defining gainand phasemargins independently. But. Q: How are the values for the stability margins determined? A: Globalandprocessstability mustbeassuredinthepresenceoftheplantand actuator parameter variations. Its closedloop poles are very close to the jcuaxis. Right? A: Wrong.thesystemisgloballystable.
9 Bodeintegrals Q: How many Bode relations are there. and as a result.O) on the 2'plane. the actuator needs to be more powerful. They are usable on for IRF andmicrowavecircuitsdesign. A11.8 Integral of feedback Q: At what frequencies is the feedback positive? A: The feedback is positive within the circle of unit radius centered at (1. Q: But. However.10 The Bode phasegain relation Q: Why is the slope of the Bode diagram expressed here in dB/oct. of an expansion on certain nonlinear systems is given in Appendix these expansions are less important for practice. compared with the four fundamental Bode relations described in this book.than 5" error as needed for sound feedback system design. Q: Are the Bode integrals applicable to transcendental plant transfer functions? .Boderelationswereexpanded unstablesystems. total? A: There are about 30 different integral relations in Bode's book.andpositivefeedbackis confined to the frequency range where the noise and the disturbances are so small that even after being increased by the positive feedback. An example 10. there will be power losses on this series resistor. Fano expansions of the Bode integral of the real part.Appendix 11 39 1 used in the laboratory for testing various control schemes.Whentheloopgaindecreasesmonotonicallywithfrequency. Gain samples one octave apart typically suffice to calculate the. not barrels.themajor part of the area of positive feedback falls in the band 0. isn't the power dissipated anyway on the internal impedance of the actuator when the impedance is made finite by compound feedback? A: No. with a larger saturation level. Q: Have new relations or expansions of these relations been made since then? A: Yes.Also. And it is certainly convenient that the slope of a welldesigned Bode diagram is nearly 10 dB/oct. The integral of [9]).dB/dec? A: The same reason that supermarkets sell milk by gallons. Q: Does it make sense to consider a loop gain response having a constant slope of 10 dB/oct? How can sucha response be implemented in a physical system? A: It can be closely approximated by a rational functionof s.parallelpaths of the signal propagation. and the transfer function for disturbance isolation.6fb to 4fb.multivariablesystems. they still will be acceptable. not in. A11. However. but at others even worse thanit was before. which use not one but several terms in the Laurent expression. to Q: Well. The most important are the R.So.discretesystems. what is gained from the application of feedback? A: Negative feedback is used to cover the frequency range where the plant parameter variationsandtheeffects of disturbancesarecritical. Compound feedback causes no power losses and doesn't reduce the actuator efficiency. feedback atmuch higher frequenciesis negligibly small (see bibl. A11. the feedback makes the system better at some frequencies.phase response with less .
All. A Q: typicalconcern of theprojectmanager:thesystemdesigntradeoffsrequire knowledge of the available performance of its subsystems.392 Appendix 11 A: Yes.ranges of frequencies whereplantstructuralmodescanfall. and frequency. although it may not always be very precise.about the plant when designing the control law? A: Certainly.. up to four octaves over the estimated feedback bandwidth. What kind of data do we need for this estimation? What kind of mathematics is Q: involved? A: The following datais generally required: the sensor noise spectral density over three octaves in the vicinity of the planned feedback bandwidth. A11. the performince that can be achieved with the optimal compensator.11 What limits the feedback? Q: What are the physical factors which limit the feedback? A: The major factors are: the uncertainties in the plant transfer function. i. and sometimes more. but has bigger and faster feedback? A: TheBodeintegralapproachallowsthedetermination of thebesttheoretically available system performance.Incontrast. it is difficult to cope with three scalar variables: gain. and unmanageable manner.engineer striving to make the best of the design. and he reduced the number of the variables to only two scalars: gain and frequency. This increases the available feedback. the clipping of higher frequency components of the sensor noise in the actuator.e. Why is itso? A: Bode noticed that for an . threatening.and sensor. The available feedbackis infinite for hypothetical fullstate feedback. the teaching relies heavily on the root locus method. soon after me. Q: At a university where I have taught. How can I be sure that my competitor will not enter the market. and the level of noise in the system's output. How can the system tradeoffs be made without designing the feedback subsystems? A: The manager should request data about available subsystem performance. This sort of data is typically available. The available disturbance rejection can then be estimated using the Bode integrals.applyingtherootlocusmethodtoannthordersystem increases the number of the variables to n complex variables (roots) crawling all over the splane in a strange. Q: Does it pay . phase.to exploit asmuch as possible the available knowledge . plant. Afterward I started to work in aircraft industry and was surprised that here the engineers prefer Bode diagrams. the available (nondistorted)outputsignalamplitudefromtheactuator.12 Feedbackmaximization Q: I want to develop and market a product. and these subsystems often employ feedback. This approachallowshandlinghighordersystems(including MIMO andmultiloop systems).. with a superior product that uses the same actuator. Q: Is it necessary to approximatea physical plant transfer function (which is sometimes . and no feedback is available for a completely unpredictable plant. Using the Bode approach 'this can be calculated without actually designing the compensator.
Q: What ifthe plant is nonminimumphase? A: The nonminimumphase componentof the loop phase lag must be compensated for by reducing the minimumphase component of the phase lag. in order for the compensator to be accurate within 0.. i. Is this correct? of poles and zerosare not importantas long as A: Not exactly.5 dB. of the Q: How accurate must the gain response approximation be for the accuracy related phase response to be 5" ? A: About 1/2 dB. and the rational function approximation is introduced only in the final stage of the compensator design.. must be highorder for accurate approximation of Q: The compensator transfer function the required transcendental response. Q: What is therequiredaccuracy of theapproximation of thedesiredloopgain response? will approximate the A: The accuracy must be such that the related phase response desired phase response with an accuracy of about 5". Variations in the values the frequency response doesn't change much.of analog compensators? A: Equalizers for analog telecommunication systems have been routinely designed with up to 20th order transfer function. the sensitivity of the such function modulus to the polynomial coefficientsis typically less than 1. What is the difference with regard to digital controllers? ' . the polynomial coefficient of rounding is insignificant. by using asymptotic Bode diagrams.Thesmallremainingpossible feedbackincrease may notjustifyafurtherincrease inthecomplexity of the compensator. Simple formulas for these calculationsare given in Chapter4. Then. poles and zerosof a highorder transfer function may be very sensitive to the polynomial coefficients. accuracy need not be better than 1 to 5%. Q: How were these equalizers designed? A: By interpolation. Therefore. But. Therefore. 1dB of feedback is lost.e. with accuracy of implementation of the desired response of k 0. The typical compensator response has no sharp peaks and notches. This can be done by reducing the frequencyfb to increase the lengthof the Bode step.Then. over each of the 3 octaves of the cutoff. by adjustments. The calculated or measured plant from the desired loop response to obtain the desired compensator response. Q: Why 5"? A: In this case.001 dB. and the compensator order is in this aspect irrelevant. and by using the Simplex Method. by using Chebyshev polynomial series. and the effect Q: What isthe best available accuracy . by cutandtry procedures.inthe element domain.Appendix 1 1 393 transcendental) by a rational function and/or to perform plant "model reduction'' when designing a feedback system with the Bode method? and actuator transfer functionsare subtracted A: No. the rounding errors will make the system not robust. the average loop phase lag must stay 5" away from the stability margin boundary. 6% in the magnitude of the transfer function.theaverageslope of theBodediagramwillbelessthanthe maximum acceptable by 1/3 dB/oct (from the proportion: 90" for 6dBloct). by using the Second Remez Algorithm. Q: Only analog systems have been discussed.theselosses are marginallyacceptable. Typically.
the approximation error can be quite large. and a digitaltoanalog converter are timevariableand have an inherent delay that reduces the available feedback. For smooth plant responses.a digital filter. A11.14 Nonminimum phase functions Q: Is there a case where it is advantageous to apply compensators with nonminimum phase transfer functions? A: Probably not. More general and better results can be achieved with the frequency domain approach. Q: Why aren’t the ZieglerNichols conditions for tuning PZD controllers presented in this book? A: This method does not use a prefilter (or a command feedforward). and nonlinear dynamic compensation designallrequiretheuse ofthe Gplane. the technique is the same. stability margin definitions. A11. a compensator must introduce a phase delay in the loop to make the system stable. the expected improvementindisturbancerejection is 5 tolOdB when a PID controller is replaced by a higherordercontroller. as long as loop coupling is taken care ofby adjusting the responses and providing extra stability margins within certain frequency bands. which include an analogtodigital converter.394 Appendix 1 1 A: Sampled data systems. especially when it is used in conjunction with a nonlinear dynamic compensator. Multiloop systems can be designed one loop at a time. A11. when the plant has highfrequency structural modes. procedure Q: Isn’t it awkward to go back and forth from the Nyquist diagram on the Lplane to 8 the Bode diagram? Isn’t one type of diagram enough? A: It’s a case of the right tool for the right job: global and process stability characterization. butthetradeoffresolutionandthe . control falls short of the performance of a system with an appropriate highorder compensator. typically. having the about a proportionalintegralderivative (PZD) compensator Q: What transfer functionCis) = k. of the loopsis large? Q: Does this technique work well even when the number A: Yes.with the extra advantage of increasing the amplitude stability margin and reducing the noise effect at the actuator’s input. however. thePID compensator roughly approximates the response of PZD the optimal controller. a PID’s is the optimal controller? Q: But how much better off than A: It depends onthefrequencyresponse of theplanttransferfunctionandthe disturbance spectrum density. However.15 Feedback control design. Sometimes.For a plant with structural resonances. and is suitable only for specific plants. this phase delay can always be achieved with a minimumphase lowpass filtertype function . + ki/s + kds ? A: With appropriate gains. a5 to 30 dB improvement can be expected from the highorder controller.13 Feedback maximization in multiloop systems Q: What about loop response shaping for multiloop systems? A: Basically.
is seen when it is used not for the purpose of analysis only. robustness? A: Interaction of highorder harmonics can typically yield up' to 1520' of extra phase lag for the fundamental. say.Appendix 11 395 Q: A: Q: A: compensator design are simplified by using the Bode diagram due to a reduced number of variables.Q: DF analysis does not account for additional phase shifts created by higher So doestheDFdesignguarantee harmonicsinteractinginthenonlinearlink.16 Global stability and absolute stability in thedesignofhighperformance control systems? A: Devising appropriate Lyapunov functions is yet unmanageable for systems with a highorder linear part and several nonlinear elements. a threeinput. the accuracy of the analysis need not be high. Similarly.9). Q: At the 1st Congress of the International Federationof Automatic Control (IFAC) in 1960 in Moscow. This research impressed Americanprofessorsattendingtheconference andshiftedto a largeextentthe direction of research inAmericanacademia. which is rather easy to do since the typical NDC phase advance exceeds 100200". Russian rockets' control systems have been designed with frequency domain methods in very much the same way as American rockets (see A13. Q: IsitconvenienttouseLyapunovfunctions A11. The phase of the DF has an error of up to 2 0 ' compared with the phase calculated with exact analysis . researchers from Russian academia presented important theoretical results using timedomain and state variables. foolproof tool. will it take 3. and especially. . when nonlinear systems can be simulated well with computers? A: The DF advantage . longer than a singleloop system of comparable complexity for each channel? 3 times longer.Werethesemethodsthemethods Russian engineers employed to design control laws for their rockets and satellites? A: No. In many cases. Still. But multiloop system design is a natural extensionof this procedure. The NDC DF phase advance must be increased by this amount. Q: What is the purpose of using DF now. this phase is not of critical importance since NDCs can easily provide the required phase advance.but. It would be foolish to discarda Phillips screwdriverjust because it cannot drive all screws. conceptual design of control systems with several nonlinear links.9. Would it be right therefore to discard the DF approach altogether? A: No. or 27 times. but for the purpose of design. if you have to design. A11. although DF analysis is not a universal. Do the design phases discussed in the following Appendix 12 apply for the design of singleloop or multiloop systems? Singleloop. typically. threeoutput system. . it is fairly accurate when employed for analysis and synthesis of well designed control loops with steep monotonic lowpass responses.17 Describing function and nonlinear dynamic compensation Q: DF analysisfails to predict instability in certain systems.
When commands should be formulated in actuators’ actions (and sensors are not aligned with the actuators).18 Multiloopsystems when defining what is the Q: Why is it the number of saturation elements that matters multiloop system? A: Because the DF of a saturation link changes from 1 down to the inverseof the loop gaincoefficient.NDCs reduce theeffects of nonlinear coupling between the loops on the system stability. and using NDCs to provide global stability and good performance in the nonlinear mode of operation. since in this case the decoupling matrix doesn’t need to be precise. . changes in it caused by saturation could affect other loops. path? A: In the foryard path. i.e. verifying the m. How does coupling between the loops affect the system stability and robustness? Q: be analyzed by considering the change in one loop A: The effects of loop coupling can transfer function causedby the plant and the actuator transfer function variations in another loop. the coupling is more dangerous near the crossover frequencies where the feedback in the is loops positive. Q: Is it advantageous to use NDCs in MIMO systems? A . Q: Is the design of MIMO and multiloop systems difficult? A: It is not simple. Al1. character of parallel signal propagation paths. is coupling most dangerous? Q: At what ffequencies the wherethe loop’ gainislargeand A: In the nonlinearmode. However.in some cases.p.Inthiscase. this can be taken care of by using an NDC. In addition to providing the same advantages as for SISO systems. Compared with thiseffect. up to30 dB.. the decoupling matrix should be placed in the feedback path. Yes.in the forward path Q: Where is the better place to implement the decoupling or inthe feedback. considering the effects of loop coupling only in the narrow frequency ranges where the coupling is critical.396 Appendix 1 1 Q: HOWmuch improvement in disturbance rejection can be expected from using an NDC? A: It depends on the desired frequency shapingof the disturbance rejection. This effectneedsto be accountedfor during the stability analysis and the loop transfer functions synthesis. but several oftignored techniques greatly simplify the design: The Bode technique of making tradeoffs between the loops without designing the compensators. even the effects of the plant parameter variations are small. using appropriate output impedances for the actuators.19 MIMOsystems matrix . In the linear mode.atlowerfrequencies. without sacrificing the disturbance rejection.The robustness canbe provided for by correspondingly increasing the stability margins. A11.commands are formulatedinthesensors’readings. 100 or 1000 times.
in his definition the transfer function is the ratio of the input to the output. .which may be acceptable if we only limit the use of this term to a very narrow classof problems.20 The Bode’s book Q: In the bibliography. and expressions. to be the optimal. he used the word “corrector”. and not on analysis. Why? but mostly for synthesis. the authors of the Classical Feedback Control found no versions in the existing literature better or equally usable for the engineering design purposes than those employed in the Bode’s book. Bodedidn’t use the word “compensator”. of tools for these two purposes are. Bode formulas has been later modified andgeneralized for thepurposes of certain research goals and for the convenienceof teaching. we have to correct the response for it be toas prescribedby the theory. for feedback loop or for any other signal transmission channel. With the definition employed today of the transfer function as the ratio of theresponsetothesource. These definitions allowed integrating feedback theory with driving point functions analysis and synthesis. However. definitions. Bode himself named it as the book on analysis.Instead. function hasin fact maximum phase shift. and the sensitivity is the ratio of the relative changeof the cause to the that of the result. uses original Bode’s notations.Appendix 11 397 Al1. A: Bode developed several powerful approaches and he optimized the terminology to serve this wide of range applications.asatelecommunication engineer.p. function must be called minimum phase lag functions . we use the inverse definition of the transfer function which is. For example. we have measured the response (of the loop) as is. Nonetheless.p. with very few exceptions. Today. Q: What aboutthe terminology in the Bode’s book? valid for awide range of applications. This is why Classical Feedback Control. then. maybe. too narrow even for the control engineers. in most books it is inconsistently called minimum phase function. he defined certain functionsas ~ n i m u m phase as the functions having the smallest phase shift among all transfer functions with the same gain response. m. but are not as elegant and general. while in somecontemporaryworksthereturndifference is formallydefined as “sensitivity” . generally speaking. Following these definitions.. m. the book is characterized as a book on synthesis. however. The meaning of this word is: we know the theoretically best response. The choice A: The book describes the tools for analysis also.but still. quite different. To bring some sense to this. Bode used separate definitions for the sensitivityand for the return difference. intuitively better for the narrower area of applications.
test and tune the system performance (up to days). so it is important that this procedure be short. introduce nonlinear elements in the compensator andlor design a specialized nonlinear dynamic compensator. determine the frequency and time responses for the linear system (1 to 24 hours). an estimate of the average time is given for completion of each of the steps by an experienced designer. decide on the type of nominal closedloop gain frequency response that is close to the optimum(2 minutes to several hours). approximate the required compensator response using one of the methods discussed in Chapter 6 and obtain the compensator transfer fimction and specifications for the compensator hardwarehoftware (a day or two). findtheprefilter(orfeedback pathlink)toapproximate the requiredresponse (10 minutes to3 hours). . or a plant model. Design of the linear compensators and prefilters (up to 5 days) given the nominalplant response. decide what kind of nonlinear dynamic compensation (if any) will be employed. Phase 1 1 1 : Nonlinear compensator design and timeresponse simulations 0 0 0 Phase 11: 0 create anonlinearmodel for thesystemandfindtheresponsesforlargelevel signals (nonlinear mode) (up to 10 days). or. and reliable. subtract thenopinal plant gain responsefromthedesirednominalloopgainresponsetoobtaintherequired compensator gain response (10 to 30 minutes). B. makealinearsystemmodelandcalculatetheclosedloopfrequencyresponse (several hours). make all tradeoffs required by the frequency responses of the disturbance spectrums. andby the plant transfer function variation. draw the desired Nyquist diagram ontheLplanewhichmaximizesthefeedback (5 minutes). subtract thecalculatedclosedloopgainresponsefromthedesiredresponseto obtaintheprefiltergainresponse (10 to 30 minutes). find and drawthe desired Bode diagram (5 to 30 minutes).398 Appendix 12 Appendix 12 Design sequence In the following.reshapetheBodediagram (10 minutes to 22 hours). simple. Note that only Phase I is in the critical path for the global engineering system design (of which your feedback system is a subsystem). by other feedback loops in the system. alternatively. and define the stability bounds on the Lplane(5 minutes). Phase I might be repeated many times for different configurations of the global system before a final decision is made. Phase I: Calculation of the available performance (upto 24 hours) determine the feedback bandwidth6 (about 30 minutes).findthe desired responsefor the feedback path.determinethe available feedback and report the system performance to your manager. and for a notunusuallycomplex problem. 30 Subtotal timefor Phase III:up to 40 days.
A1 3. accuracy. Fig.. The plant gain fiequency response is shown in Fig.2 Temperature control system Fig. The signal amplitude of the thermocouple sensor TC is rather small. A13. A13. 1150" k 2"). The response of the PZD compensator can be adjusted by three manual regulators.1(a). two thermal capacitances. 1413. A13. the settling time needs to be minimized.Appendix 13 399 Appendix 13 Examples A13. temperature Heater power Fig.l(b).1(e).2. Thefeedbackbandwidthislimited by theeffects ofthesensornoiseandthe .Thesettlingtimedependsontheheater'savailableoutputpower.1 Industrial furnace temperature control Temperature control of a precision industrial furnace. A13. therefore. 3 hours) ata prescribed temperature (e. connected by the heat flux conductance as illustrated in Fig. the carrier being 60 Hz. The plant includes a sink representing heat dissipation into the environment.the feedback bandwidth. so that the error amplifier noise limits the system's. The fmace operation during the settling time is not paid for by the customer.1.g. A13. and (c) the plant frequency response The actuatoris a pulsewidth modulated electrical heater If. The regulator adjustments in relation to the different payloads are easy to make. H Fig. (b) its equivalent thermal schematic diagram. A13.1 (a) Industrial furnace with payload.3 Closedloop transient response The payload must be kept in the furnace for not less than a specified time (say. and the value of the feedback. is implemented as a singleloop SISO system depicted in Fig. The plantequivalentelectricalcircuitis showninFig. andtheplanttransfer function is the ratio of the payload temperature to the difference of the power of the heater and that dissipated to the environment. the furnace and the payload.
The control looks nearly The system transient response is shown timeoptimal because the settling time in the linear mode of operation is many times shorter than the fullpower preheating time. the controller simplified the interfacing and made the controller cheaperin mass production.themechanicalplant wasmodeled as an equivalentelectricalcircuit(the response for the electrical model was compared with that of the mechanical model and found to be exactly the same) and integrated with the controller's model in SPICE. A13.6.. Thus. windup was eliminated with a simple NDC that reduces the loop phase lag for largelevel signals. JM. the average phase stability margin must be about 60". I D Driver a " Compensator A LVDT electronics electrical to mechanical Fig. the feedback bandwidth must exceed 1 mz. rather big variations in the loop phase shift are expected over the frequency range with positive loop gain.sinceinthiscasethecompensatoradjustmentsforeachtype of payload need to be of higher precision than is practical.2 Scanning mirror of a mapping spectrometer ThemappingspectrometeroftheCassinispacecraftisaninsbxmentproducing an object image in the lightof specified wavelengths. but it cana be problem if the system is not well tuned. Then.the stiffness and the damping in the yoke's a r m s . Le. A13. and by the carrier frequencyof the pulsewidth modulation. with a staircasetype command as shown in Fig. anaLmT. The ayerage slope of the Bode diagram must therefore be about 8 dB/oct. A13. Later. Because of the loworder compensator employed.Themirrorangleshouldtrackthe command with a risetime less than2 $msec. A13. windup cannot be large. to scan the mirror in such a way that the light beam will remain longer in each of the focal plane photocells. for the minimum phase stability margin to be 40°. To improve the system robustness. however. With such stability margins.and the suspension flexure stiffness and damping.A13.4 Scanning mirror suspension actuator with scanning for mirror positioning (solenoid) sensor and (LVDT) Fig. The control goal is to increase the exposure time.5 Mechanical schematic diagram feedback with control The yoke is rotated by a solenoid actuator. Digital implementation of. A13. in Fig.400 Appendix 13 amplifier noise on the actuator's input.4.3. for the convenience of the analog controller design. The amplitude stability margin is 10dB. The analytical model of the system dynamics was derived and the system was simulated as a block diagram with unidirectional blocks. The plant's mechanical schematic diagram in Fig. The spectrometer's scanning mirror is attached toa yoke suspended on a flexure as shown in Fig. A13. The mirror angle is measured by a linear voltage differential transformer (LVDT). The average slope cannot bemadesteeper.5 shows the cores of the solenoid and the LVDT. . and the mirror characterized respectively by the momentsof inertia Js.
The plant and controller simulation has been performed in SPICE.Appendix 13 401 Fig. the feedback about the plant must exceed 60 dB at lower frequencies.Theplantphase. The feedback bandwidthis limited by the structural resonance of the solenoid core onitsarm. Thechosen stability boundaryon the Gplane is shown in Fig.and the solenoid core resonance on its a r m at 6kHz. the loop must be gainstabilized' at the frequency of the solenoid core resonance. the maximum plant gain in the 6kHz of ~2 dB. A13.7(a) and has a 45" phase margin. A13.7(b). neighborhoodisknownprettywell. The overshoot is about 30%.Thenominalresonancefrequencyis 6kHz withpossibledeviationsof ~ 2 0 %The .withtheaccuracy however. Thk stability margin of 45' is preserved over all frequencies where the loop gain is more ttian10 dB. the gain at the resonance must be below the amplitude stability margin of 10 dB. The feedbackpathtransfercoefficient (B) waschosentobe a constant. Theresonance frequencyof the LVDT core on its arm is higher than 10 kHz. which is marginally acceptable. is quite uncertain due to the effects of theLVDTresonance(at a higher frequency) and the stray inductances and capacitances in the coils. Therefore. A13. The Bode diagram for the designis shown in Fig. curve (a). is0.8.6 Closedloop transient response of the scanning mirror The size of a pixel. A13. The overshoot must not exceed 30%. Fig. this is excellent accuracy).The response of a system 'with such a B to a smallamplitudestepinput is shown in Fig. The plant parameter variations are such that for the beam to 'stay reliably within the pixel. anglewise.9 mad.. A13. damping of theresonanceisknownwiththeaccuracyof +20% (for damping. Here is seen the resonance of the mirror on its suspension at 8 Hz. i. The overshoot can be efficiently reduced (to about 10%)by either putting a lead filterin the . Therefore.7 (a) Nyquist and (b) Bode diagrams for scanning mirror control It is seen that variations in the frequency of the plant resonanceswill not cause the system to oscillate.e.
Therefore. The control loop compensator consists of a twopole lowpass filter having a corner frequency of 4 Hz and a highpass filter with a corner frequency of 0. or by an equivalent lowpass prefilter. The lowpass filter reduces the loop gain at the 10Hz carrier frequency to the acceptable level of 27 dB. Since the nutation frequencies are in the range of 0. the feedback loop is a bandpass one. it was decided to make the entire loop analog. Because of the rather large feedback bandwidth.8.11. and curve (b) to the threshold 5 times smaller than the maximum torque calculated for the linear system.8 Hz. Due to the 45" phase stability margin. application of an NDC is not required. . A13. However.9. this was not done since the actuator is not intended to operate in the linear mode. This was confirmed by computer simulation of the output response to the step command with different actuator saturation thresholds. The system block diagram and the simplified schematic diagram of the compensator are shown in Fig. The thrust is pulsewidth modulated (with the carrier frequency of 10 Hz) by a feedback loop using a gyro as a sensor.402 Appendix 13 feedback path. The filters are made analog as a costeffective solution to lowquantity production. A13.3 Rocket booster nutation control It was found that the nutation angle of a rocket booster shown in Fig. curve (a) relates to the linear system. small auxiliary thrusters were employed to produce stabilizing moments. The loop gain and phase shift frequency responses of the controller's original version are shown by curves (0)in Fig. A13. The highpass filter attenuates dc and lowfrequency components of the error to prevent them from overloading the thrusters. A13. In Fig. the output transient response is quite good. A13.1 Hz. the smaller is the real part of the nutation exponent. To fight this. linear system (without saturation).9 Analog controller and plant model for scanning mirror (simplified) It is economical to employ a small actuator whose maximum available torque is a small fraction of the maximum torque that would be seen in an idealized. large windup should not be expected. and required low power consumption. A13. Even with this small output power actuator.8 Closedloop transient response (a) without and (b)with saturation in the actuator Fig. It was shown that the larger the feedback is.2 to 0. 0 p+= output time LVDT electronics Fig. and the overshoot is reduced. small production quantities.10 increases exponentially during the main engine thrusting. A13.
The controller is globally stable. a substantial jumpresonance could be expected because of the phase stability margin of only 30' (the same as in the older system). The linearity of this repeater designed in Russia in the 1960s is comparable to the linearity of a concurrent American counterpart. the diodes are open and the NDC gain is 0 dB.Appendix 13 403 10  20 Fig. (b) schematic diagram Computer simulations showed that after 12 seconds of thrusting. to reduce the channel intermodulation to an acceptable level. are shown by the curves marked (m). A13. employing a notch filter to attenuate the carrier instead of the lowpass filter. scale (b) Fig.4 Telecommunication repeater with an NDC Linearity of repeaters for telecommunication systems with the frequency division multiplex should be very good.12(a). Both the feedback over the range of the nutation frequencies and the amplitude stability margin are increased.11 Bode diagrams for nutation controller The loop responses with the modified higherorder compensator. For large signals. despite the fact that the bandwidth of the . A1 3. The NDC schematic is shown in Fig. 10 Hz notch fllter (small slgnals) t log.12(b). A13. A13.10 Rocket booster nutation Fig. Although the process instability might not create a problem. A13. A13. the nutation amplitude in the modified system was 50 times smaller.12 Nonlinear dynamic compensator: (a) gain responses. it is safer to eliminate even the possibility of jumpresonance and windup by introducing an NDC whose frequency response for small signal levels is shown in Fig. however.
Al3. When the signal amplitude exceeds their threshold.further saturation in the ultimate stage does not cause periodical oscillation. J R = 1 Nm'.1 3 Bode openloop diagrams for a telecommunication repeater compensation amplifier feedback amplifier feedback Fig.15 Attitude control of a flexible plant . A13. A13. A1 3. thus changing the gain.14.1 4 Nonlinear dynamic in a telecommunication The system global stability was provided by a nonlinear dynamic local feedback loop in the penultimate stage. A13. The moment of inertia of the spacecraft is JSC = 100 Nm'. A13. . Many initial conditions were tried during experiments and none of them elicited oscillation. Fig.mass production in' large quantities and no stability problems wereever observed. upper curve. Fig.in the common loop as shown by the lower curve in Fig.15. so that according to the DF method. We consider here only the planar rotation about the center of rotation for the system. A1 3. This was achieved by the increasein the feedback obtained by using a Nyquiststable system with the Bode diagram shown in Fig. This Bode d i a g r q is less steep and the related phase lag is smaller. and of the reaction wheel.5 Attitude control of a flexible plant Consider the problem of pointing a spacecraft usinga reaction iheel that resonates ona flexible shaft at 30 H z inthesystemshowninFig. The simplified schematic of the qplifier with the NDC is shown in Fig.13.13. A13.404 Appendix 13 transistors was smaller and the linearity of the transistor in the ultimate stage was much worse than that of the special transistors developed in The Bell Laboratories. Thereactionwheel is rotated by the motorM. The diodes play the role of a deadzone link. the diodes start conducting ':and introduce negative feedback over the functional bandwidth. The repeater was put in.
16(a). and (b) for the nominal plant Alocalfeedbackloopaboutthedriver(not showninFig. A13. The loop at theresonancemustbegainstabilized. produces an area of uncertainty on the loop gain and phase responses.A13. A1 3. The output voltage is regulated by (a) changing the current in the field windingof the generator (main actuator) and (b) changing the grid bias of the bypass tube (vernier actuator). which further limits the feedback bandwidth as illustrated (not to scale) in Fig. themaximum voltage from the power amplifier that supplies the current to the field winding of the generator.as shown in Fig. The resulting Bode diagram is shown in Fig. vernier.17. and local loops The power supply voltage for a klystron transmitter is produced by a motorgenerator whoseoutput is rectifiedandpassedthroughalowpassfilter. A13. not to scale. The goal (the reference) for the main (slow)loop is to keep the voltage on the bypass tube to be on average 2 kV for the tube to be operational. resonating somewhere in the 13.16 Bode loop gain diagrams (a) for the flexible plant control with uncertain resonance frequency.2000 10 30 t V Fig.theavailablefeedbackbandwidthincreases by almostan octave. This reduces the peaktonotch swing from 54 dB to 28 dB compared with the case of the actuator being a pure torque source.A13. The gyro noise limits the feedback bandwidth to less than 1OHz. The 2OkV output voltage is the difference between the 22 kV at the filter’s outputand the 2 kV voltage drop on the bypass tube.Appendix 13 405 The amplitude stability margin was chosen to be 8 dB.6 Voltage regulator with main. The feedback bandwidth in the main loop is limited to 10Hz by the timeconstant of the generator.29 Hz region (nominally at 15Hz) with very low damping. The gainpeak depends on the value of the output mobility of the actuator. any other specified value within 0 the The range of the output voltage of the generator/rectifier/filter subsystem is from 0 to .42 kV.15)makesthe actuator output mobility equal to 0.03 (rad/sec)/Nm.and the noise of the voltage divider and the first stage of the amplifierin the compensator.Correspondingly. The goal (the reference) for the fast vernier loop is to finetune the voltage drop on the bypass tube for the regulator output voltage to be 20 kV (or to 40 kV range). dB 0 .1900 20 . The flexible appendage with the moment of inertia J A = 5 Nm2. .16(b). A13.
A13. However. so that its average output voltage needs to be maintained at the2 kV level (to be desaturated) by the slower main loop.18. The Bode diagrams and the Nyquist diagrams for the main loop return ratio TI and the vernier loop return ratio2’2 are presented inFig. Lplane T I T* 0 (b) Fig.17 Highvoltage power supply controller The vernier loop bandwidthwaschosentobe 60Hz toachievetherequired disturbance rejection.This regulator can only regulate within the rangeof 4 kV. The plant is highorder and nonlinear (because of the nonlinearity of the magnetization curve of the field winding). the saturation link S was not included in the system. and a substantial jumpresonance was observed.406 Appendix 13 Fig. initially. to .system was asymptotically stable following any initial conditions that take place in practice. The loop compensator’s orderis rather high. Analysis and experiments indicated that the reason for the jumpresonance was a nonlinear interaction between the vernier loop and the local loop T3. Experiments showed that the .18 (a) Bode and (b) Nyquist diagrams for a highvoltage power supply controller (4 This local loop is aboutthehighvoltagetransistordrivingthebypasstube. A13. A13.
a power transistor which is nearly an integrator.Theinputandoutputimpedancesare specified (commanded) by the transfer functions of the blocks 2. feedback can be used to make all these parameters of an active multiport as desired. Since then. Thethreeportssplitterandcombiner B. This loop contains an amplifier which is nearly an integrator.7 Telecommunicationrepeater' An electrical or a mechanical multiport can be characterized by transmission MIMO coefficients between the ports and the input impedances (mobilities) at the ports. A13. with 1 MHz feedback bandwidth.A13. F " " " L""4 Fig. and linear.5021.19. This increases the closedloop phase lagof this subsystem. 499 . when the transistor is overloaded by a largeamplitude input signal. A13. U . . The provision of the desired impedances and the inputoutput gain can be illustrated with the flowchart shown in Fig.19 Flowchart of a MIMO system The amplifieris a twoport described by linear relations between the two input and twooutputsignals: lain. stable in time. and laout.In other words.Appendix 13 407 stabilize its transfer function.therepeater's gainmustbecontinuouslyadjusted by automatic level control to match the attenuation of theopenwirespanbetweenthe vary within repeaters for all weather conditions.Uain. Thestabilitymargins inthetransistorloophavebeenincreasedand a signal amplitude limiter S was installed at the input to this loop. pp. which can cause the line attenuation to 40 dB at higher frequencies and within 20 dB at lower frequencies. and the feedback path which is nearly a differentiator. The nominal input and output impedancesof the repeater equal to2 = 150a. so that the phase stability margin in this loop is quite big and the standalone loop is globally stable. with . . . the inputimpedance.This is a MIMO systemhavingthreeinputsandthreeoutputs(not counting the slower gain response control loops). The repeater amplifier must have large feedback over the operational band to reduce the intermodulation between the channels. Over the operationalfrequencyband. An example of such a system is a repeater designed in the Bell Laboratories in the 1930s for a 12channel frequency division multiplex telecommunication system over open wires [2. The connect the signal source and the load to the amplifier and the feedback path ratios of the signals (variables) to be controlled are the amplifier transfer coefficient. thus increasing the phase lag in the vernier loop. However. the feedback in this loop reduces.andtheoutputimpedance. substantial jump resonance has not been observed.
with the summer S . . The repeater inputoutput gain depends on these two blocks and on the B donotaffecttheinputandoutput feedbackpath B. The stray elements bound the available signaltonoise ratio.20 Telecommunication repeater simplified schematic diagram The combiner and the splitter are made with transformers having bandlimiting stray reactive elements and multiple resonances. The loop gain response with a crisp Bode step and maximum available feedback over the operational bandwidthwas implemented with the accuracy better than 0.ame. the tradeoffs between these parameters were first part of an impedance that defines the resolved using the Bode integrals for the real available signaltonoise ratio. with the output current Zout summer Sz . and the output impedance.theseresponseswere approximated with highorder correctors and compensators that are implemented by incorporating extra electrical elements in the combiner and splitter. Thus.Then.gainregulationsin impedances. reflection attenuations.5 dB. The higher frequencies bypassed the feedback path B (whose transformers’ attenuate higher frequencies) via the highpass filter Bhf. They are more complicated than the idealized ones shown in Fig. A13.qualityof vacuum tubes available at the but the time. the bandwidth of FET feedback amplifiers can reach several GHz design principles remain essentially the s.20. A13. Today. on the splitter. The minimum  . and for the feedback. The frequency responses for the splitter and combiner in different directions of the signal propagationwerechosentobenotoverlycomplex. for the reflection attenuation. This diagram illustrates the benefits of structural design: over the functional frequency range. and the feedback. The feedback path was arranged in two parallel paths. The system is singleloop in the Bode sense since actuator with a nonlinear link: the saturation in the ultimate stage of the amplifier. The optimal solution to this tradeoff is different at different frequencies. the input voltage U n . During the design. The simplifiedschematicdiagramincludingthegaincontrolloopsisshownin it includes only one Fig. h Fig. with the order of the compensators over 25. A I 3.408 Appendix 13 the input current Zin and the outputs of the blocks Z being the command signals. The functional bandwidth over which the feedback sufficiently rejects the intermodulation was limited to 50 kHz by the .19: unlike in the block diagrams. the is controlled using the summer SI. and the output voltage Uout . splitting a signal in electrical circuits means splitting the incoming power into two (or many) loads. the input impedance depends exclusively on the combiner. and there is a tradeoff between how much power goes to one output at the expense of the second output.
that even a smallovershoot in a single regulatortransientresponsecauses a largeovershootattheoutput of the trunk as illustrated in Fig. A13. as shown in Fig. say. The slope of the Bode diagram and the available feedback over the operational band could be increased with Nyquiststable loop design. Three pilot signals with frequencies at the ends and in the center of the functional frequency band were being sent continuously. Bandpass filter I k I 4  reference Fig. Theanalysis ofsuchsystemsshows . A13. the system would not be globally stable.Appendix 13 409 phase property of the total feedback path Z3 + Bhf was preserved by appropriate shaping of the frequency responses B and Bhf. The automatic gain control d B keptthepilotlevelsattheoutputoftheamplifierconstant thus assuring that the total gain of the open wire line and frequency the repeater remained Odl3 at these frequencies and.23. When.21. thentomakesmalltheovershootofthetotal . A13. ThreeBodevariableequalizers inthefeedbackpathhavethefollowinggain regulation frequency responses: flat. there exists substantial interaction between them. Fig. approximately 0 dB over tbe entire frequency regulation frequency' band of operation. An analog computer in the form of a resistor matrix decoupled the automatic gain control loops. and convex. slanted. A simple NDC . A13.22.22 Cascaded signal level regulation in telecommunication system When the number of such regulators in a long telecommunication trunk is large. as depicted in Fig. and without such an NDC.andallthese automatic gain control system start acting to correct the signal level. However.a backparallel pair of diodes shunting the capacitive interstage load .21 Gain therefore.8 Distributedregulators The signal level along the telecommunication trunk canbecontrolled by sending a smallamplitudesinglefrequencypilotsignalandchangingtheattenuation of the feedback paths of the repeaters such that the levels of the pilot signal selected' by appropriate bandpass filters will be equal to the references.provided some phase advance for largelevel signals and significantly reduced the jump amplitude of the jump resonance. If allregulatorsareidentical. responses A13. the levels at the outputs of all repeaters starting withthenthbecomelargerthantheirreferences. the signal level on the nth repeater output suddenly becomes larger than the reference. A13. in those days (and until the 1960's) the methodsof providing global stability with NDCs were not yet well developed.
Q. thus reducing the coupling between the regulators. It can be seen that the plant contains many resonances due to structural dynamics and propellant slosh. although feasible. which. The digital computer compares the programmed attitude command history to the attitude feedbacksampledfromtheinertialplatform.24. .25 is a frequency map of the transferhnction from the gimbal actuators. Fig. presents substantial implementation difficulties. with a faster regulator following several slower regulators. It is part digital and part analog.24 Control block diagram for the Saturn V Fig. the attitude is sensed using an inertially stabilized platform.Theattitude error is passed through a I)/A converter to the analog flight control computer.l I I rComputations . Al3.9 Saturn V SIC flight control system For the flight control system for the first stage of the Saturn V launch vehicle. for example) with multiple spatially distributed regulators. t.I 1 DIA L . NDCs can be introduced in the regulator loops. the frequencies of which are uncertain. and the attitude rate is available from a rate gyro package. Note that the rate feedback and the compensators are analog because they must have a fast response. This computer consists of the highorder RC compensators Cl and C2. The analog feedback from the rate gyros goes directly to the flight control computer. A13. The controller block diagram is shown in Fig. or the regulators can be made differently. The torque is provided by gimballing four of the five F1 rocket engines. while the generation of the attitude errors in the digital computer can be rather slow. A13.23 Transient responses for (a) a single regulator and (b) a chain of regulators Alternatively. Chains or networks of regulation are employed for shape correction of flexible bodies (optical mirrors.""""" Launch vehicle digital computer I i I I !  4 c. A13.(4 I I ! C*(4 'I i Flight control computer L 1 + I 1 I """" Rate gyro  Inertial plafform 4  1 + Actuator and vehicle dynamics Fig. a 90" stability margin must be preserved in each regulator over the frequency band where the regulator loop gain is bigger than 60 dB.410 Appendix 13 response. A13.to the sensors.
A13. The control system has sufficient stability margins for all possible frequenciesof the flexible modes within the boundaries shownin Fig. However. via an opamp withRC compensation. twofeedbackloops are employed. A1325 Frequencymap of Saturn V dynamics The Nyquist diagram for the pitch channel is shown in Fig. A13. the spacecraft attitude changes linearly with time (if the solar windand the outgassing are neglected)andthemaximumdeviationfromthe nominal exceeds.The coupling between the two loops is small andbe can neglected. which is unacceptable.01 . A13.26.10 PLL computer clock with duty cycle adjustments The 20MHz computer clock of the Cassini spacecraft radar must have 50% duty cycle and must be synchronized with1OMHz a ultrastable quartz generator. ' 1 Making the error smaller would mean firing the thrusters more frequently and using more propellant. Thefilter's output voltage is the average value of the clock signal and is proportional to the duty cycle. thus changing the degree of asymmetry of the generated signal. The clock frequency is halved by a D flipflop and applied to one of the inputs of an exclusive OR gate which is used as a phase detector. This voltage is subtracted from a half of the VCC (obtained by a voltage divider).26Nyquistdiagram for SaturnV pitch control resonance modes of the structure and propellant appear as loops. The averaged phase detector output (phase error) is applied via an opamp with RC compensation to the VCO tuningdiode and corrects the clock frequency. To stabilize the frequencyandthedutycycle of theclock.4 modes I m a t o r dynamics1 . the output is forced by the feedback to havethe required 50% duty cycle. A13. A13. and the difference (duty cycle error) adds to thedc bias of the transistorin the VCO.To the other input of the gate is applied the reference 10MHz signal. The VCO is followed by a Schmitt trigger to make the clock shape rectangular.1 I nn n 10 frequency 100 Fig. Aerodynamics mode Btanks 1 41 Pitch (yaw) dynamics Bending. After being clamped by the Schmitt trigger. the required .11 Attitude control of solar panels The DS1 (Deep Space 1) spacecraft attitude is bangbang controlled by short bursts fired from theattitudecontrolthrusters.the cycle length of approximately 1000 sec.Appendix 13 Rigid Slosh. with .25. The dutycycle loop uses as a sensor a simple RC lowpass filter. A nearlysinusoidal 20MHz signal is generated by an LC VCO. The first loop is aPLL. The Fig. Between the firings.
which is not easy to achieve.27. Here. Thesamplingfrequencyis 0. alphacorn is the commanded angle. the settling time of thesystemisabout10seconds. and then returned to start the next scan.05 Hz.A simplified block diagram ip Fig.theclosedloop bandwidth = 0. and the settling time must be less than 0.412 Appendix 13 accuracy of pointing the solar panels outfitted with Fresnel lens light concentrators is 0. alphaenc is the encoder readings.5 Hz.27 Simplified block diagram for solar panels' attitude control The feedback controller has a common configuration of parallel connection of proportional control path and a lowpass link with saturation in front of it. Therefore. . pulsenun is the number of steps the motor must make during the sampling period. Therefore. I ~ I saturation pulsenum Driver. A13.12 Conceptual design of an antenna attitude control The Microwave Limb Sounder(MLS) af the Chemistry Spacecraft includes a TeraHerz radiometer. With this rise time.8 sec. in seven or so seconds after a step command this pathbecomesdisabledandthefeedbackcontroltakesover. The chosen stepper motors rotate the panels via a gear without backlash and make up to 40 steps per second. To prevent this from affecting the steady state accuracy. It begins with a highpass filter that does not pass dc. A bettersolution is to combine the advantagesof the openloop control and the closedloop control by using a of such a controller is shown feedbackfeedforward scheme. 2' sCan. The required settling accuracy after thereturnis lom5 (the ratioof 6 arcsec to 180').butforsmalltimesit supplies up to 12 pulses to the motor to' speed it up and to reduce the rise time and the settling time.However.theactionisfast. the panels' attitude toward the sun must be continuously corrected by motors.5'.07 Hz.whichistoolong. Whenthemotorsarecommandedopenloop. Motor.the possibility is not excluded that the motors might slip over one or several steps during extreme conditions of the thruster firing and excitation of structural modes in the large panels (the panels supply the power to the ion engine).Aftereach 20sec. theantennamustbeturnedaway by nearly 180' for calibration. Therefore. fbacknun is the number of pulses commandedby the feedback compensator. the system must be controlled closedloop using the data from the encoder placed on the solar panel shaft. and the rise time is about 5 sec. A13.TheradiometerantennamirrorscanstheEarthhorizon with 6 arcsec pointingaccuracy. fb = 0. f forwnum is thenumberofpulsescommandedbythefeedforwardpath. Encoder Fig. The upper path in the block diagram is the feedforward. A13.
a simple specialized modelof the system's dynamics has to be developed by the control system designer. In this case. The frequency of torsional resonance of the antenna on the shaft is 300Hz. the small settling time requirement necessitates a wide feedback bandwidth. The design of attitude controllers in the dual mirror system is simple since the off the optical pathof the switching mirror can be small and can be switched completely main mirror during the scan. about 60 Hz. but in this example there is no need for this. The singlemirror mechanismis shown in Fig.Appendix 13 413 The radiometer pointing direction can be shifted by 180' with an additional mirror. We will consider the worst case: the rocking mode motion aligned with the torsional modes. and a single servo loop for both scanning and switching.Thequestion regarding feasibilityof the singlemirror servo has to be resolved at the initial stages of the project when accurate mathematical models of the mechanisms' dynamics have not yet been developed. The coupling coefficients are not yet known at this stage of the project. a heavy 50 kg laser is installed.28(a). thebandwidthmightbelimited. A13. since the modes will be already substantially damped by the dissipative output mobility of the motor driven by a voltage source. . thus making the control collocated. It is placed close to the motor rotor. Therefore. The modes are coupled with the torsional motion of the rotor. and the scanning mechanism can include a gear. The entire mechanism is mounted on the spacecraft with a thrust structure. for the purposes of conceptual design and the control performance estimation. and of the laser modes. The employed models haveno provisions for modal damping. there will be two separate mechanisms: a scanning mirror and . There is a requirement that no structural mode can be lower in frequency than 50 Hz. The damping can be introduced in the model. The antenna mirror is connected to the motor rotor viaa shaft. The lowest frequency of the modes of the base is about 50 Hz.however. On the same instrument base. I Spacecraft I FIg. These projections use the moments of inertia of the rigid bodies about the rotation axis. A simplified mechanical schematic diagram for the rotor torsional motion shown in Fig. The angular encoder disk is used as the angle feedback sensor.a switching mirror. A13. by structuralresonances. A13. The spring coefficients. In the attitude controller for this mechanism.28(b) is obtained by projecting the rocking structural modes onto rotation about the motor rotation axis.28 (a) The TeraHerz antenna mechanism and (b) its simplified mechanical schematic diagram The major structural modesof the laser and the base are rocking modes. are chosen such that the torsionalmodesresultingfromtheseprojectionshavethesamefrequencies as the original rocking modes. An economic alternative is touse a single mirror directly driven by a motor.
A13. r. where mobility and to the mobility 1to1 electrical impedance representation k is the electromechanical motor constant.e. This simple model is valid at frequencies up to half of the frequency of the lowest structural resonance. .85. The initial rough estimation of the plant dynamics’ effects on the control loop based on the diagrams in Fig.29 (a) Electrical equivalent schematic diagram for the plant dynamics of the TeraHerz antenna attitude control. CM=: O. = 0. the resonance frequencies WL = 400 radlsec.COA = 2000 rad/sec. C.as long as the frequency rangeof interest is up to i. The antenna mirror moment of inertia JM = 0.0002 F. The instrument base.base moment of inertia is infinite. and the encoder constitutea rigid body with the moment of inertia about the motor rotation axis JB = 0. (b) modified diagram The schematic diagram modifiedby changing the orderof series connection of the twopoles about the contour is shown in Fig. For extending the control bandwidth beyond Hz. i.we can assume in the course of the system analysis that the .. Therefore. = 4 SZ.OlF.2 kgm’.g.29(b). The torsional stiffness coefficient of the mirror shaft k~ is replaced by L M = l / k ~The .5 F. A13.29(a). e. it is converted to ZS = kzr. A13. the stator.2 F.The resonances at 50 and 60H z of the base and the laser will not profoundly affect the loop because the base mobility is rather low relative to the antenna mobility. 15 the worstcase stability analysis can be made using the plant model in Fig.. up to 25 H z . 25 Hz. (the motor is driven by a voltage driver).e = 320 radlsec.01 F. The actuator output is shown as a source of velocity SZS with internal mobilityZSM.5 kgm’.01 kgm’. electrical contour impedanceof the driver’s output is the winding resistance r.28 is the following: The base moment of inertia is larger by more than by an order sf magnitude than the mirror’s moment of inertia. A13. The equivalent electrical schematic diagram is shown in Fig. with the of the mechanical to electrical parameters and variables. 1to1 ratio in the analogy with JM replaced by C M = 0. Fig. Here.414 Appendix I 3 The moment of inertia of the laser about the rotor rotation axis J L = 0. CL= 0.. CB= 0. A13.. the control bandwidth is below 15 H z .28(b). k = 0.e.
A13.346 1/zs motor toque rotor D’Alembert toque b + toque on mlrror shaft d 4 305e68+46. and its zeros 0.30 shows the block diagram model of the plant dynamics following the approach described in Fig.002s 0. Fig. However.035e6s4+31. mechanical variables are listed. I I 0.00001M1 25e6s 1955+1 4 ‘ negaHve of omega of base  delta omega onshaft w 1 07s Fig. On the SIMULINK block diagram.Appendix 13 415 This function is an impedance of a lossless twopole.6492+10e6 2 ‘ 53 0. fj400 and poles fj184. in this case the expressions for the impedances do not even need to be derived. (c) SIMULINK model .18(d) and (f).29 could be analyzed with SPICE. 7.30 (a) Ladder network plant model of the TeraHerz antenna attitude control. the plant model is worth building in SIMULINK. for the convenience of system design and integration. A13. The schematic diagram in Fig. A13.6S 61.kj700 alternate along the joaxis. (b) Block diagram model of the plant.
31 for the plant transfer function U(ZM)/ES is plotted with w = logspace(l.w) for two cases: for the nominal values of the plant parameters. and disconnecting the input to a link pointing up.B. disconnects the corresponding shunting branch.e. [A. a multiwindow controller. The two gain and phase responses in the picture are very close to each other.B. A13.C. and for & = 0. Based on this analysis. A13. Hence.C . 1.3). for an infinitely massive base.DI = linmod('mlsp1an') . The Bode diagram in Fig. . the structural resonances of the laser and of the base do not constrain the feedback bandwidth which can be therefore 30Hz. A13.416 Appendix 13 Notice that the diagrams (b) and (c) in Fig. a scanning mirror and a switching mirror.11) and loop shaping.l.C. i. command profiling (see Section 5.D. singlemirror attitudeandswitching control is not feasible. This device must be made with two separate mechanisms. bode (A. a decision can be made to use the economic option of a singlemirror mechanism. 10' 1o2 10 ' Frequency (radhec) 0 U B30 8 60 a 90 10' I I I I I Ill1 1o2 Frequency (rad/sec) I o3 Fig.W ) hold on % remove the connection to the input of block 22 bode(A.28 are very convenient for troubleshooting and investigating the effects of the series and parallel branches in (a): disconnecting the input of a link pointing down replaces the related series twopole by a short circuit.. With an appropriate prefilter. this feedback bandwidth ensures the required accuracy with substantial margins.31 Bode diagram for the plant of the TeraHerz antenna attitude control loop For a much larger primary mirror of the MLS GHz radiometer.D.B .
ZD is the driver amplifier output impedance. the PZT does not excite the structure. is circuit for the voice coil drive: (a) following the structure.The Fig.33. CM reflects the mass of and Rsuspreflect the VC the mirror.. The delay line includes two mirrors. the cart motor control loop will not be discussed here. flat mirror is moved by a piezoactuator (PZT) with a maximum stroke of 30pm. and k is the force/current . with the resonance at 5 Hz. A13.32(a). The pathlength must be controlled with better than 5 nm accuracy. PZT mirror Fig. A1332 Optical delay line The PZT consists of two piezo elements and applies a force between the smaller mirror and the countermass. spherical mirror is actuated by a voice coil (VC) capable of a maximum displacement of 1cm at low frequencies.33 Equivalent electrical VC mobility ZVc= (rvc+ ZD)/J'?where r. Since the forces applied to the supporting structure balance out. in turn. and the feedback bandwidth in this loop is not limited by the structural modes. The light bouncing between the mirrors along the variable pathlength experiences a controllable delay. (b) modified the coil resistance. and Lsusp (b) suspension. A reaction wheel assembly is employed for the interferometer attitude control. Here.5 kg against the flexible structure with mobility ZsbThe equivalent electrical schematic diagrams for the VC actuation of the mirror are shown in Fig. The VC 4USP I is.13 Pathlength control of an optical delay line System description The optical delay line of the stellar interferometer (SIM) is placed in the path of the light gathered from one of the widely separated optical elements called siderostats. The smaller. The voice coil moves the mirror having mass M = 0.Appendix 13 417 A13. A13. The VC desaturates the PZT. desaturated by placing it on a cart. The larger. A13. The delay is regulated to provide an interferencepattern on the focal plane where this light is combined with the light from another siderostat. The delay line is positioned on a truss structure as shown in Fig.
Over the frequency range of interest. on the average. E . = J k L .JM With the VC constant k = 0. 0 transient responses to largeamplitude vanishing disturbances that are neither excessively large in amplitude nor duration.as seen in Fig.. if the nonlinear phenomena responses are prolonged and violent and cannot be excluded by the design. so that the vibration amplitudes are. Hence. Numerical design constrains The PZT maximum displacement is 30 pm over the entire frequency range of interest. The VC maximum displacement amplitude is D m . the following characteristics are required or highly desirable: system robustness. The available feedback in the VC loop is affected by the mobility of the structure Zst. The feedback must be at least 60 dB at 16Hz. both the feedback and the maximum actuator output amplitude can decrease inversely proportional to the square of the frequency. . the gain and the slope of the loop Bode diagram can be substantially increased. the VC loop has to be designed as standaloneunstable. the structural modes and their uncertainties prevent the VC loop bandwidth from exceeding 100Hz . the gain in the VC loop at 16 Hz is only 26 dB. the frequency at which the VC maximum displacement . The principle design objective is to keep the mean square error below 5 nm.7. the loop gain slope must be close to 12 dBIoct. The sampling frequency for the PZT loop is 5 kHz. the sampling frequency can be chosen to be 1 kHz.= 3 A. inversely proportional to the square of the frequency.if the loop is designed as standalonestable.3 NIA and the current saturation threshold of the VC driver amplifier I . in this case. Higherlevel design objectives The control system performance index is the mean square error in the delay line pathlength. J is substantially smaller than the mobility of the contour it is in and does not much affect the velocity of the mirror (the voltage on CM).. D f c . in order that these nonlinear phenomena happen infrequently. In this case. For the VC loop.33. As can be estimated using an asymptotic Bodestep response. In addition to the principle design objective.00003 m. caused by the reaction wheels. good output responses to commands of different shapes and amplitudes. Therefore. A13. I (MD. is amplitude equals that of the PZT. = 0. Therefore. . or 0 a large disturbanceIcommand triggering threshold of nonlinear phenomena. less than the required 60 dB. The loop can be designed as a selfoscillatingdithering system as described in Section 9.418 Appendix 13 electromechanical coefficient of the voice coil. 1 I (2N = 40Hz * The feedback in the VC and PZT loops must be sufficiently large to reject the vibrational disturbances over the bandwidth up to 500 Hz. The disturbance forces’ spectral density responses are assumed to be flat over the 500Hz bandwidth. = (21~j’)~ kZm.
and nonlinear.e.. and uncertain plant must be reasonably close to the best achievable.on the PZT loop. To effectively reject vibrations at higher frequencies. The conceptual design employing Bode integrals and the Bode asymptotic diagrams should produce a solution in the vicinity of the globally optimal timeinvariable controller (the “optimality” here means best satisfaction of the higherlevel design objectives. system engineers). the feedback in the PZT loop must have wide bandwidth. not adaptive).the impedance can be . the only requirements for the VC and PZT loop gain shaping is the provision of stability and robustness. according to the opinion of the customers. If possible. Lower Level Design Objectives The lower level design objectives for the feedback system under consideration can be formulated as the following: 0 0 0 0 0 To effectively reject the vibrational disturbances of rather large amplitudes. We choose the controller to be multiloop.) At the frequency of 5 Hz. Therefore.e. is not necessary. but not overly complicated. therefore. i.the VC driver output impedance should be small to damp the VC suspension resonance.and the combined loop must be m. This is more easily accomplished using frequencydomain specifications. flexible. The system must be globally stable. the VC loop gain must exceed 50dB at 40Hz and the feedback must increase toward the lower frequencies. at higher frequencies. The design considerations are the following: The VC and PZT loops are nearly map... At frequencies over 50Hz. For this. The transient response to the commands of different amplitudes must be good.but it unnecessarily consumes a part of the area specified by a related Bode integral. this is not worth doing. PZT is the main actuator. The design begins with making some reasonable assumptions and translating the higherlevel objectives into a set of lowerlevel guidelines. making the loop cross at fcMSs loop which is not the main one does not hurt the system performance. Extra gain in the (Generally.p.Appendix 13 419 Design approach The controller for this nonlinear. the frequency of oscillation in this loop should be as high as possible in order for the amplitude of the limitcycle oscillation in this loop to be as small as possible and not to overload the PZT. If the VC loop is made unstable when standalone. At frequencies below fcross 1 40 H z the VC provides a larger stroke and is therefore the main actuator. it has been calculated that 600 H z bandwidth will suffice. these objectives and considerations should be formulated in a mutually decoupled (orthogonal) form to simplify the system tradeoffs and to speed up the design. rejection of the large amplitude disturbances depends atfcfcrosson the loop gain in the VC loop and at f > fcross. the parallel channels must be shaped appropriately. Outside of these ranges. but not time variable (i. however with smaller amplitudes. highorder. The latter consists of the design objectives and the design considerations.
At 40Hz.. From here.e. the PZT loop transfer function is employed. TVc= u/[s(s + 20)'] where a = 141(21~40)~ =2. The gain at 40 Hz must be 43 dB. since for this particular design task the advantages of using the better shaped diagrams exemplified in (b) do not justify higher complexity of the design. 141 times. industrial mainvernier systems do not include NDCs or only include simple NDCs for improving the responses to the commands. however. The PZT loop bandwidth is 600 Hz wide. and 12 dB/oct over the range 40 to 150 Hz. . When the feedback and the disturbance rejection that are achievable in such systems are not sufficient. Such systems are made globally stable by making all standalone loops globally stable. For this reason. An NDC provides global stability. with the dead zone equal to the threshold of the PZT actuator. Conventional design approach Most commonly. and 6 dB/oct below this frequency. The gain may gradually roll down at lower frequencies. The PZT loop phase margin at 40 Hz is approximately 40". At 40 Hz the gain in the PZT loop is therefore 12 dB/octx logz(150/40) + 20 = 43 dB. with the Bode diagram slope 10 dB/oct at frequencies 150 to 1200 Hz.24~10~. A13. The VC loop Bode diagram slope is 18 dB/oct down to 20rad/sec. To ensure the system's global stability and good transient responses to the commands. the phase difference between the loops is 270" .50]. an NDC will be included in the combined loop.34(a). The chosen design options The chosen design options for the system with an NDC are the following: The VC loop is standaloneunstable. It also necessitates a procedure to recover the system from the limit cycle which can be quite violent and damaging to the hardware. This ensures minimum phase character of the total loop with 50" safety margin. In this case the slope of the Bode diagram of the VC loop can be made steeper thus improving the disturbance rejection and the handling of largeamplitudedisturbances. causes the command generator to be complicated and the reaction to the commands sluggish.420 Appendix 13 made higher. i. The VC loop Bode diagram crosses the PZT loop Bode diagram at 40 Hz. thus making lZVclhigher in order for the loop transfer function at these frequencies to be less affected by the system structural modes. We choose the asymptotic Bode diagrams shown in Fig. This. this option is normally not used in industrial systems. the disturbance rejection can be improved at the price of making the mainvernier system only conditionally stable [25.140" = 130". with a Bode step response. As the linear block of the NDC.
dtot = conv(dvc.5e24 1. dtot) hold off The diagrams shown in Fig. bode(ntot.4. A1 3.dvc . dvc = [l 40 400 01. lag of sampling) is obtained with MATLAB function bos tep (from the Bode Step toolbox) and shifted in frequency with lp21p for the crossover frequency to be 21~600 rad/sec.w). This forms the standalone PZTloop transfer function.length(ntotl)length(ntot2)). dpzt = [l 2.5e17 1. . nvc.w) .34 Asymptotic Bode diagrams for standalone loop transfer functions.34(a) need to be approximated with rational function responses. ntot. ntotl = conv(npzt.dvc) hold on nyql og ( 3700 .5e9].p. (a) simpler and (b) higher performance Block diagram with rational transfer functions The asymptotic (transcendental) responses shown in Fig.w). and provides 69 dB disturbance rejection at 16 Hz.dp z t ) hold on nyqlog (3700.684e21 0 03.35 indicate that the system in the linear mode of operation is stable. nvc = [2. ntot2 = conv(dpzt. The Bode diagrams for the standalone VC and PZT loops and the summed loop Bode diagram are obtained with the following script: npzt = [2.2e281.24 x 109/[s(s+ 20)2].83e5 7e9 9.dpzt.14ell 1.5e21 9. A13.Appendix 13 42 1 I (a) \ \ I \ (b) \ Fig. bbode (npzt. np zt . The standalone VC loop transfer function has been already defined as 2.33e10 6. w = logspace(l.7e6 5.5). adz = zeros(l. robust. hold off zoom on The logarithmic plane Nyquist diagrams are plotted with nyqlog from the Bode Step toolbox: nyqlog(3700. ntot = ntotl + [adz ntot21. The frequencynormalized Bodesteptype loop response (including certain n.dtot.26e17 1.dvc) .nvc). hold on bode (nvc. A13.dpzt) .18e13 6.
A13.0003 implements nonlinear dynamic compensation (NDC). The dead zone equals the PZT saturation threshold. and the signal amplitudes have been increased 10 times relative to the real system. VC is represented by a velocity source with velocity saturation. and the combined loop. the dead zone. To simplify the analysis. This increases the numerical stability of simulation. A13. It does not . This scaling certainly does not affect the theory of operation and the responses' shapes. A1336 Simplified SIMULINK block diagram of pathlength control In this block diagram and in the simulations. the PZT. the thresholds of the saturation. (The combined response is the top one on the gain response.) The SIMULINK block diagram using these transfer functions is shown in Fig. Fig.36. + marks octaves 70 60 50 10' 1' 0 1oa g 10' :* 40 Frequency (raasec) 30 % 20 110 0 270 10' 1o2 10 10' 1' 0 Frequency (rad/sec) 1o6 20 270 240 210 180 150 120 loop phase shift in degrees 90 Fig. The pathlength is the sum of the displacements produced by the PZT and VC actuators.35 Bode and Nyquist diagrams for the rational transfer functions approximating the asymptotic diagrams for the VC. x marks w = wb. The feedback path via the dead zone dz0.422 Appendix 13 Nyquist diagram. using a detailed VC actuator model will not change the principle character of the results. and the intermediate one on the phase response and the Nyquist diagram.
The standalone VC loop is unstable. Without an NDC. Selfoscillationin the system without an NDC With frequency responses shown in Fig. the limit cycle of the standalone VC loop is in fact also the limit cycle of the system as a whole. Therefore. Largeamplitude vanishing signals with substantial frequency components of the limit cycle oscillation belong to the basin of attraction for the limit cycle. and the analysis can be performed with the describing function method. the system can be analyzed as having a single nonlinear link. 0 2 4 6 a (a) (b) Fig. when the vibrations’ amplitudes are large enough to saturate the PZT. the VC loop is left alone and the system bursts into the limit cycle selfoscillation at a low frequency with large signal amplitude. The oscillation approaches a limit cycle with the period longer than 5 sec. A13. Fig. The NDC effect on the global stability arid the transient responses to large amplitude commands will be discussed later. When the highamplitude highfrequency vibrations vanish and no longer saturate the PZT.Appendix 13 423 pass small amplitude signals and has no effect on the system performance in the linear state of operation.37 Oscillation triggered by a short pulse command in the system without an NDC (a) of the VC output pathlength component and (b) at the input to the saturation link Compared with the amplitude of this oscillation.37(b). In other words. 3 mm pulse command. The accuracy of such analysis is sufficient since the VC loop is a lowpass and the signal at the input to the saturation link becomes nearly sinusoidal when approaching the limit cycle as seen in Fig. the limit cycle oscillation does not change substantially since the amplitude of the output of the PZT is negligibly small compared with the displacement generated by the VC. A13. the saturation in the VC actuator. Limit cycle of the VC loop with the NDC Since the standalone VC loop is unstable and the NDC dead zone does not pass the . the system is stable in the linear mode of operation when both the PZT loop and the VC loop are closed.353. the PZT output is negligibly small. The system is conditionally stable. A13. A13.37 shows the pathlength oscillation triggered by a 30 msec.Thus. the system is not globally stable.
684e21 0 01.nvc) . A13. A13. Since the oscillation amplitude is small. dll = [ O nll + dl. the dead zone describing function approaches 1. and such a system. (a) (b) Fig.38 Equivalent block diagrams for describing function analysis. dvc = [l 40 400 01. the system becomes close to linear.14ell 1.5e21 9. The standalone linear part of the latter (the dashed box) is stable since this is a feedback system with parallel connection of VC and PZT loops. since for signals much larger than the dead zone. unstable. % nyqlog(l260. d2 = conv(dpzt. deq = dll + [ O 0 0 d21.dvc).18e13 6.33e10 6.39 is obtained with the script npzt = [2. It is important to emphasize. as has been shown before. The transfer function of the box is TPZT 1+Tvc + TPZT Notice that the signal about the loop via the saturation is applied in phase. w = logspace(0. for the conventional describing function methods to be applicable. the standalone VC loop is unstable even with the NDC. that the NDC radically changes the amplitude and the frequency of the limit cycle oscillation. the diagram (a) is transformed equivalently to the diagram (b) by replacing the dead zone with parallel connection of a unity link and a saturation link with the threshold equal to the dead zone. however.5e17 1. dl = conv(dvc.2e281. deq. The amplitude of the limit cycle oscillation is determined by the dead zone in the NDC. nl = conv(npzt.w ) . Therefore.7e6 5. nvc = [2. however.38(a). (a) with the unstable linear part and (b) with the stable linear part The Bode diagram for this transfer function shown in Fig.83e5 7e9 9.nl. A13.424 Appendix 13 signals of small amplitudes.dpzt) . and the system can be viewed as that shown in Fig. during the describing function stability analysis the saturation link in the VC actuator can be replaced by 1. dpzt = [l 2. This system’s linear part is. is stable with sufficient stability margins.5e24 1.deq) bbode (nl.26e17 1.4). oscillation can only take place with amplitudes not much higher than the dead zone.5e9].
Application of this signal makes the system stable.5 dB less than the gain in the linear state of operation.40. at 250 Hz.gain at this frequency is 3. the describing function of the PZT equals 0.Appendix 13 425 1 10 g o c 610 20 1oo 10 ' 1o2 1o3 Frequency(radkec) .67. the system is asymptotically globally stable.. or.00054. 8 ~ = 0. the gain coefficient is 1. The timehistory of the pathlength in this case is shown in Fig.27(E/e~)~ = 1/15 wherefrom E = 1 .27(E/e& . i. Le.e.. The loop . 1. is only 3.e. Therefore. using (113). approximately. (a) without VC that limits the slew rate. when the PZT actuator recovers from being saturated by disturbances or commands._) 1oo 10 ' 1o2 1o3 Frequency (radlsec) Fig. With this E. it is capable ofdeliveringanoutputsignalcomparablewiththeselfoscillation in the VC loop. and the oscillation in the VC asymptotically dies down.0.i. and (b) with saturation in the .5 dB. The oscillation amplitude E can be foundby equating the describing function to the inverse of the loop gain coefficient. A13. A13.. Since in the linear state of operation the stability and safety margins of order of 10 dB are provided.39 Bode diagram for the oscillation describing function analysis It is seen that the conditionof oscillation (zero loop phase shift while the loop gain is positive) occurs at approximately 1600 radhec.5.
5 0. nl = conv(nvc.83e5 7e9 9.04 0.684e21 nvc = [2. w = logspace(l.40 Pathlength timehistory after a pulse command in a system with the NDC and without a PZT actuator. so that the total system properties are those of a linear system. (a) without limiting the VC velocity and (b) with a M.02 0.w) is plotted in Fig.The Bode diagram of this transfer function found with the script npzt = [2.02 .41.) Thesystemthereforeincludesonlyonenonlinear link theVC saturation.04 ~ 0.5 l 0 0.dvc). .dpzt) .The transfer function of the equivalent linear link kom the VC saturation output to the VC saturation input is TVc/( 1 + T i ) .5e21 9. A13. is sometimes called exact linearization. 0. dvc = [l 40 400 01 .dpzt) .deq.2e281.4).18e13 0 01.426 Appendix 13 2. Application of the Popov criterion to this system shows that the systemis asymptotically globally stable.5e9]. (Augmenting system with an extra nonlinear subsystem.08 0.5 0 0 0.14ell 1.1 dsec limiter of the VC velocity Global stability of a system with the NlDC The deadzone branch is in parallel w i t h the PZT saturation branch.08 0. dpzt = [l 2.33e10 6. These two parallel a nonlinear branches equal a linear branch with the same transfer function.5e17 1. our in case by the dead zone. A13. dl = conv(dvc.5e24 1. deq = dl + [O d21.08 (a) (b) Fig. d2 = conv(npzt. bod'e (nl.26e17 1.06 0.7e6 5. 6.
and (b) with PZT. (a) Fig.42 Closedloop transient responses to 20 ms. 3 mm pulse command. A13.the NDC deadzonestopspassingthesignal. E 10 8 0 10 I I I I I I I111 I I I I I I l l 1 10 ' Frequency (radlsec) 1o2 1o3 10 ' Frequency (radlsec) 1o2 1o3 Fig.41 Bode diagram for transfer function TvJ(1 + TPZT) Transient responses to commands The output transient response to a 3 mm pulse command in a system.and the error dies down rapidly.Appendix 13 427 20 g .42. and PZT mirror displacement timeresponse. upper curves. A13. is shown in Fig. asymptotically globally stable. assuming no saturation in the VC actuator.thepathlength response coincides with the response of the equivalent linear system. lowergrves.43. (a) without PZT. After the PZT becomesdesaturated. without saturation in the VC actuator. A13. rapidly settling (b) The zoomedon parts of the responses are shown in Fig. A13. ending in a smallamplitude 250 Hz limit cycle oscillation. pathlength. .
rapidly settling Settling time to high accuracy is of that the linear system.A13. and (b) with PZT.44.5 0 5 0.04 0.035 0. (a) without PZT.1 I 0.45 shows the responses of the equivalent linear system of the real system (solid line). A13.02 0. Fig. A1 3.025 0 .08 0.04 Fig.06 0. A13.015 0.5 ' 0.428 x IO9 Appendix 13 x 10" 0. 0 3 0.04 0. 3 mm pulse command Theoutputtransientresponseto a 50 ms pulsecommandin a systemwith saturationinthe VC actuatorwiththreshold 0.08 0.12 0.08 0 0.01 0. The tail of the response is that of a linear system. (dotted line) and the response .02 0.asymptotically globally stable. 3 mm pulse command. The saturation limits the slew rate of the output. dashed lines.01 is showninFig. solid lines. A13. x IO" 3*5 5 9. ending in a smallamplitude 250 Hz limit cycle oscillation.02 0.12 (a) Fig. and PZT mirror displacement timeresponse. pathlength.43 Zoomedon piecesof the closedloop transient response to a20 ms. and in other aspects the responses are similar to those in Fig.1 0.42.44 Closedloop transient responses to 50 ms. with saturation in the VC actuator.
04 0.001 m and(b) larger amplitudes' commands . of the linear system ( llshaped with overshoots) and the mainvernier nonlinear system (with finit slew rate) These figures demonstrate the functionof an NDC: not only does it ensure global stability butalso it provides good transient response to large signals.41 is close toa single integrator response over the frequency range near the crossover.02 0. If needed. A13. but these are not needed because in absolute values the overshoot is small.Appendix 13 429 0 0.08 0. Such a loop response results in a singlepole closed loop transfer function which is known to produce a rather good closedloop transient response (although not as good as a higherorder Bessel filter response. In the largesignal (nonlinear) mode of operation.020.06 0. A13.6 0. Fig. A13.02 0.J. approximately.08 0. The 75% overshoot in the smallsignal (linear) mode of operation results from the stability margins that are justifiably chosen reasonably narrow (since the plant parameter uncertainty is small)..08 0 0.2 1 0.12 Fig.2 0 0 0. the smallsignal overshootcan be reduced by a prefilter or a command feedforward. from 1Hz to 15 Hz.45 Transient responses to step commands.04 0. A13.06 0.08 (a) (b) Fig.6 1.46 shaws the output response to the (a) small and (b) large step commands applied at the instant of 50msec.1 0. This loop response shown in Fig. the loop response is. 1.04 0.8 0.46 Transient responses to step commands of (a) 0. TJ( l+T. so there remains some room for improvement).8 x lo" 1.4 1.4 0.
This system satisfies the Popov criterion with conventional stability margins. After t & contr. Further reduction of the keshold 0.0001 5 .0001 to makes the srys@p unstableandcauses a small Fig. In most cases there was no reason to do this and it was not done. To such systems including more than one nonlinear link.12 Nceptable. we rely on computer simulations. tr).ol systep becomes operational (at least in simulations). the VC actuator. Within the .08 0.12). The iterative adjustments of the crossover iati+ and from the experiments converged rapidly.04 0.47 Transient responses to the pulse command in the system withPZf saturation mpljtude oscillation similar to that threshold 0. and for a device for taking mtwjsl g ~ p j e of s a comet or an asteroid in nearly zerogravity environment.1 0. it would be desirable to make the dead zone wider than the threshold of the F” saturation by 10 to 30%so as not to impair the PZT output stroke.06 0. the system is equivalent toa system with only one nonlinear element. the loop response at frequencies . A13.02 0. the Popov criterion is notdirectlyapplicable. However.2.5 0 0.430 Appendix 13 Robustn As long as the threshold in the PZT saturation is exactly equal to the dead zone in the NBC. Therefore.14 MIMO motor control having loop responses with Bode steps . for a rover to take samples of Mars rocks.42. and motion controls of g&ot mwp@d drills: for a robot designed to take samples of concrete in Chernobyl ~wjw station.47 remains 0.4. position. The same nominal loop response was used in most The ith loop response was therefore fully definedby the ade$fs between the different performance requirements sys&p .wgs reduced to finding the set of scalar numbers perf&na<ce. It yvis &o employed for several less precise attitude. the system must be also robust against yariations in the threshold of the mturation in the PZT. A13. and is tbrdore robust against variations in the linear links’ transfer functions and in the nonlineN characteristicof the VC actuator.andthefeedbackatlowerfrequencies was reduced. A1 3.3 An apprgyimation to the theoretical loop responsewith Bode step described in Section been employed in precision SISO motor control systems for retroreflector cgmiqgg motioncontrol and forgyroscopeattitudecontrol of an instrumentonthe ~ 1 ~ 4 spacecraft. q b lower than fbi can be reshaped following Bode relation (3. A13.0003 to 0. In t h w MIJ&3 control systems the feedback loops’ bandwidths are limitedby the s.00015 shown in Fig.Instead. but in some cases the required accuracy of the control loopwas able to be well satisfied with much lesserfeedback. the system remains stable and the 0respause to the pulse command shown in Fig. When the PZT saturation threshold is efranged from 0.
. . ". . ..~.*. .. .andthe Time (sec. . . . However. ..~. A I 3..) movements of theentiremechanicalsnake became agile and very impressive. each with the ability to bend in pitch and yaw directions. using the methods detailed in Chapter 7.... ... ..Appendix 13 43 1 limitationsimposed by theBodeintegrals. The last design step was the introduction of nonlinear dynamic compensation in some of the loops that improved the transient responses to large amplitude commands and disturbances.. . .~."..~.. the output mobilities of the actuators were made dissipative. ..".". force and feedback cschier@compuserve.".". . . ". . The damping introduced into the system reduced the effects of the waves' interference. La California. .. . Position feedback allowed achieving good transient response of a single link.".15 Mechanical snake control A gigantic animated mechanical snake employed in the movie Anaconda was constructed of chainconnected identical mechanical links.com>. . . . .forcefeedback was . .. . . ."".. . .~. Fig. A13.. .48 Transient responses This exmP1e Was contributed by of 5 links in chain a with pasition JAS Company. .~. .. . .". ...".". ..". . . . . .* . . . Crescenta...*. .~.~.thisloopgainreduction wastradedfor bigger phase stability margins and. the transient response of the chain of several links was drastically improved exemplified as in (j!. .. The oscillation was caused by the interference of mechanical Step Response waves propagating along the snake.I.added and.... therefore... better transient responses. ." (3 to 'I5 Fig. i . . . To correcttheproblem. . A13. transient responses of several such links in a chain were very oscillatory..48 forthe 5 linkchain.
asymptotic diagrams. BONYQAS (Asymptotic Bode diagram. for example.432 Appendix 14 Appendix 14 Bode Step toolbox The toolbox includesthe following MATLAB functions: 1.corn. and conceptual design 4 BONYQAS simplifies the system conceptual design. The loop shaping can be performed iteratively using this plot instead of the Bode diagrams. NDCP(Nyquistplotfordescribingfunctions of NDC with parallelpaths) 10. Plant modeling.frequencydenormalization. This plots' arrangement is convenient for loop shaping.similarto lp21p) 4. and logarithmic Nyquist PW 5. NYQLOG (Nyquist plot on logarithmic scale with frequency 'in octave marks) 2. 2 BOLAGNYQplotsfrequencyresponses of thegainandthephaselagstability margin. phase plot. B . The functions do the following: \ . TFSHIFI' (Frequencytransform. BOIN'IEGR (Breaking compensator into two parallel paths. Using the plot of the lag margin allows one to pay attention to only a small area of the plot where the gain response crosses the OdB level.Tdenormalizes an initially normalized frequency response similarly to the MATLAB function lp21p but does this by a polynomial transform in the sdomain instead of matrix manipulation in the tdomain (1p2 1p gives complex numbers as the answers for the gain and. BOCOMP (Calculation of the compensator function for a dc motor congol) 9. NDCB (Nyquistplotfordescribingfunctions of NDC withfeedbackpath) 12. BNDCB (Bode plot for describing functions ofNDCwithfeedback path) These MATLAB functions simplify designing linearhonlinear servo loops with Bode optimal loop responses and prefilters.3. BOSlXP (RationalfunctionapproximationtooptimalBodestepresponse) 6. and the 1ogaJrithmic Nyquist plot) 3. BOLAGNYQ (Bode diagram for the gain and the lag margin. and to read there the value of the guardpoint phase lag margin. BOCLOS (Prefilter design for closedloop low overshoot and fast settling) 7. Calculations 3 TFS)I[TT. produces inaccurate answers).) 8.2. BNDCP (Bode plot for describing functions ofNDCwith parallel paths) 11. oneof which is dominant atlow frequencies.Nyquist diagram on the logarithmic plane with octave marks. In particular. accomplish the following two tasks: it helps to . in problems like that described in Example 3 in Section 4. A. sometimes. Plotting routines 1 NYQLOG plots a. The mfunctions (and MATLAB scriptsfor the examples and the problems listedin the book) are available from the author's web page www luriecontrol . The marks allow reading the slope of the Bode diagram without plotting it. C. and plots the Nyquist diagram on the logarithmic plane with octave marks. It can be used.
one of which includes a variable or a nonlinear link (typically. E. 11 NDCB plots isoe describing functions on the logarithmic Nyquist plane for NDC an with a feedback path which includes a variable or a nonlinear link (typically. Completing theconceptualdesign of the feedback loop using asymptotic piecelinear Bode diagrams. Also. it generates the step response. F. The obtained parameters of the Bode step can be further used as the input file parameters for the BOSTEP function. and the frequency response of a 4thorder Bessel filter. 7 BOINTEGR converts the ' linear compensator transfer function into a sum of two transferfunctions. with the values of the describing function of the nonlinear element from 0 to 1. A series linkis also included that can imitate the rest of the feedback loopso that the loop Bodediagramscanbeplotted. Nonlinear dynamic compensator design 9 NDCP plots isoe describing functions on the logarithmic Nyquist plane for an NI)C withparallelpaths. the linear nonminimum phase lag coefficient. Calculating and plotting the phase shiftof the plant when only the plant gain has been estimated or measured (since the measurements of the plant gain are often less timeconsuming then the measurements of the phase shift).thelowpassone(ageneralization of an integrator of a PID controller) and the rest. Compensator design 8 BOCOMP calculates the compensator transfer function for the loop response with a Bode step of a servo witha dc permanent magnet motor. It allows for easy shaping of the responsesby changing the corner frequencies. It isusablefordesigning a prefilterrendering a goodclosedloop transient response. a saturation) link. A series linkis also included that can imitate the rest of the feedback loopso that . .Appendix 14 433 I. withthevaluesofthedescribingfunctionofthe nonlinear element from 0 to 1. the gain at the comer frequencies. A series link is also included that can imitate the rest of the feedback loop so that the loop Nyquist diagrams can be plotted. and the asymptotic slopes at low frequencies (the Type of the loop) and at high frequencies. and the feedback Type. The function generates the asymptotic Bode diagram and calculates and plots the phase and the Nyquist diagrams. a dead zone). asymptotic slope. 10 BNDCP plots isoe describing functions Bode diagrams for an NDC with parallel paths. I ) . obtainedwith BONYQAS) loop response with a Bode step for specified margins. a saturation).Feedback loop design 5 BOSTEIP generates a rational function approximation to the theoretical transcendental (or irrational. saturation can be introduced into the lowfrequency path to 'improve the transient response and to provide global stability. Then. 6 BOCLOSgenerates a normalizedclosedlooptransferfunctionwithoutandwith a prefilter. The input file includes the loop load transferfunctiongenerated by BOSTEP and the dc motor and the inertial parameters.one ofwhichincludes a variableor a nonlinear(typically. 1 1 . nonminimum phase.
with the values of the describing function of the nonlinear element from 0 to 1. 12 BNDCB plotsisoedescribingfunctionsBodediagramsforan NDC with a feedback path which includes a variable or a nonlinear link (typically.434 Appendix 14 the loop Nyquist diagrams can be plotted.1 Defaultldemo for nyqlog and bostep .18 and A14. nonlinear element from The functions’ HELP files include default/demo that generate plotswhen the name of the functionis typed in without arguments. Fig. A series linkis also included that can imitate the rest of the feedback loopso that the loop Bodediagramscanbeplotted. A14.A14. The plots are shown in Figs. a dead zone). withthevaluesofthedescribingfunctionofthe 0 to 1.912.
degr. degr Fig. A14.calculating and plotting a bandpass plant phase. radsec phase shift. + mark octaves fromwb 90 Fig.x marks wb.Appendix 14 435 % 0 ' 40 rf 20 '3 20 240 270 180 150 120 210 phase. Circles on the gain response mark the corner frequencies of the piecelinear approximation . A1 4.3 DefaulVdemo 1 for bonyqas.2 DefaulVdemo for bolagnyq Bode diagram Nyquist diagram frequency.
one peak. and the closedloop with the prefilter response is plotted. Circles on the gain response mark the corner frequencies of the piecelinear asymptotic Bode diagram for loop response with Bode step Dc motor servo design A dc motor control system (like that described in Example 2 in Section 4. the asymptotic slope. degr Fig. The MATLAB function can be used to calculate the closedloop response with a prefilter comprising two notches. Third.4. . A14. the frequencies of the beginning and the end of the Bode step. A14.Theresponsesatthispointarenormalized in frequency. First. and the attenuation in thefeedbackloop is sufficientatthefrequencies of structuralmodestoguarantee system stability with the chosen stability margins.5.theloopgain response has unity crossover frequency. with the function bcrstep a rational function approximation for the loop response is obtained.4 DefaulVdemo 2 for bonyqas.436 Bode diagram Appendix 14 Nyquist diagram 10" 10' 10 ' frequency.or with any subset of these. with bonyqas.1.e.. A14. like that shown in Fig.i.2.3) can be designed using the following sequence of the toolbox functions. the nonminimum phase lag. An example of the diagram is shown in Fig.2.3). asymptotic loop response. radsec phase shift. and a compensator function for the nominal singleintegrator plant with a small nonminimum phase lag is calculated (as was done in Example1 in Section 4. with the function boclos the closedloop frequency responseis plotted. and the system Type are determined during this conceptual feedback loop design. an asymptotic Bode diagram with a Bode step is chosen such that the feedback satisfies the disturbance rejection requirements. Second. The stability margins. A14. like that shown in Fig. and a lowpass . the prefilter is calculated.
without a prefilter (the one with 7 adB hump) and with a prefilter that consists of a notch.5 DefaulVdemo for boclos plotsseveralresponses:anopenloopgain response (the one with large gain at low frequencies).Appendix 14 437 The goal for the closedloop response with the prefilter a isresponse close to a denormalized Bessel filter response.as were those in Fig. radlsec 1. a peak. sec Fig. Fourth.7. . of the closed loop with the prefilter. 0. using the motor and load parameters. and a 3rdorder Bessel filter (the response with the widest bandwidth). 4thorder Besse filter response with nominal bandwidth 1 radlsec that has 15 dB attenuation at this frequency.5 3' . and of closedloop without a prefilter. the compensator and the plant transfer A14. The phase responses (the lower three) are of the Bessel filter.as exemplified by the plots in Fig. functions are calculated and plotted withbocomp. Fifth. A14. the compensator transfer function is split into a parallel connection of the lowfrequency path and the second path (for the medium and A14.5 0 0 5 10 15 20 time. The denormalized filter response can be obtained by shifting the nominal filter response along the logarithmic frequency axis. . The lower plot shows the closedloop transient response with the prefilter.6. closedloop responses. with the function bointegr. higher frequencies).
the gain and phase responses two parallel paths for the entire compensator and its Next. With the model. A14.7 DefaulVdemo for bointegr. 7. in the friction. the gain and phase responses for the compensator and the plant of a dc motor control system Fig. . A146 DefaulVdemo for bocornp. Themodelmustincludeflexiblemodes elements in the actuator (saturation in current and voltage). the controller is finetuned to perform well over the specifiedrange of theplantparametervariations. and in of the lowfrequency path of the the compensator (saturation is introduced in front compensator).438 Appendix 14 gain and phase of plant and compensator Fig.26. model. a SIIMULINK model of thesystemshouldbebuiltlikethatshownin of theloadandthenonlinear Fig.andthesystemperformanceand robustness are evaluated.
8.439 Appendix 14 NDC isoE describing functions The isoE responses for signal paths shown in Fig.5.c 1 for the saturation and He& c 1. I I 3 7 0 240 210 180 150 120 90 60 30 I 0 loop phase shift in degrees Fig. = 00. A14. 2. 1. 1.6. 2. 0. 6. A14. These values correspond to the signal levels He.30 40 I!.. 0.912 with the DF values 0.8 are exemplified in Figs.c.3. + c.1. x marksw = wl . A14.1 0 DefaulVdemo for bndcp .9 DefaulVdemo for ndcp 1 " 10" 10" 1oo 10' Frequency(rad/sec) i 10 ' U 8120 150 1 80 10" f 0. 0 . 3. 0 10 I I I I I 1\11 . A14.5. b .4.5. 3. 0. A14. 10" 1oo 10' 10 ' Frequency(radlsec) Fig.8 NDCs with (a) parallel paths and (b) witha feedback path Nyquistdiagram. 8 .  I+ Fig. 00 for the dead zone.+ marks octaws 70 60 50 % *E .2.
A14. A14.11 DefaulVdemo for ndcb .12 Defaultldemo for bndcb ." 10" 0 'T1 0) Q) 1oo 10' Frequency(rad/sec) 1o2 9 I o3 cn 90 2 a 180 IO" 1oo 1o1 1o2 1o3 Frequency (radhec) Fig.440 " 270 240 210 180 150 120 90 66 loop phase stift in degrees 3O 0 Fig.
2 H. J.Automatic Control Systems. Prentice Hall. N Y : 1961 (oneof the best books on using electromechanical analogies). Cambridge University Press. BoulderCo. When the reader becomes confused why the book doesn't followuponsomeapparentlyattractiveidea. NJ: 1993. CA: 1997. 1 and 2. Saletan. N Y : 1988 (the book containsmany.Bode. MA: 1986. Modern Control Engineering. Harper and Brothers. 44 1 . Upper Saddle River. Prentice Hall.lurieeontrol. 8 B. NY: 1963. Biernson. see www. 3rd ed. N Y : 1945 (orany of numerous later editions).J. why sometheoryisnot M e r generalized or.J. Thebookestablishesfrequencydomaintheory of feedbackmaximizationin causal singleloopand multiloop systems. Whileconcentratingonlinearsystems. E. H. Horowitz. C.KUO.thebookdevelopsdesignmethods applicable to practical systems with large parameter uncertainties and with nonlinear actuators. or even why the employed notations are not what the reader would consider appropriate. Schultz. The book cannot be recommended as a first book on feedback systems. Classical Dynamics: A Contemporary Approach. Menlo Park. Prentice Hall. he should be advised to suppress the urge to correct the classic and humbly do his homework of finding the reason why Bode did so. Bishop. Nonlinear Systems Analysis. W.N Y : 1980. 7 J.BIBLIOGRAPHY Textbooks 1 G. 2nd ed. 4 R. Pergamon Press. which is of immense importance for both theorists and practical engineers. M.Ogata. Vidyasagar. Jose.N Y : 1998. Dorf and R.QFT Publications. J. 5 P. Academic Press. practical examples of frequencydomain design). 11 C. Melsa. Principles of Feedback Control. C. L. Artech House. But afterward. Some of these issues are clarified in [6. Lurie. 13 M. 6 I.com. NJ: 1996.G. 3 P. simplified. LinearControl Systems. Modern Control Systems. Shock and Vibration in Linear Systems. 9 B. 2nd ed. N Y : 1997. Sidi.) 10 K. V. Spacecraft Dynamics and Control. 12 M. E. 5th ed.Rohrs. (The book describes the theory tind the methods for the analysis and design of certain linear and nonlinear feedback systems. conversely. D. John Wiley and Sons.&Crafton. Network Analysis andFeedback Amplijier Design. Cambridge University Press. and abstains from discussing those dedgn approaches that has been found nonpractical or inferior. Upper Saddle River. Dedham.9] and in the present book. Synthesis of Feedback Systems. N Y : 1993. NJ:'1997. the difficulty rises steeply since the book was written on the basis of lectures Bode gave for his colleagues at The Bell Laboratories where he later headed the mathematical and physical development. 1993. Feedback Maximization. McGrawHill. VanNostrand.Garnell. Englewood Cliffs. AddisonWesley. v. Guided WeaponControl Systems. Also: Qwantitative FeedbackDesign Theory (QFT). although the transparent and informative introductory chapter is worth reading for everybody interested in feedback.
L. 1998. J. 3. Lurie. 1992. vol. C. Lecture notes in control and information science. Bayard. no. Albuquerque. 20 G. vol. J. 1418 Oct. Nov.S. AIAA Guidance.128in [9]). Grogan. Navigation and Control Conference. R. Hench.28] (reference no. "J. of Intelligent Material Systems and Structures". 9. 21 G. H. Zavracky. 26 R. S . Johnson. Lurie. Proc. 1991. Integral Force Feedback for Structural Damping. Lurie. 19 G. San Diego.9] cover most topics specific to the present book.Neat. Many important papers are collected in[9. B. 104. B. First International Conf.'July 1996. PA. Bayard. Bernstein. Four and a Half Control Experiments and' What I learned From Them: A Personal Journey. San Diego. 17 J. Fanson. P. Active Member Vibration Control Experiment in a KC135 Reduced Gravity Environment. Bridge Feedback for Active Damping Augmentation. and P. v. Philadelphia. Berlin: Springer Verlag.7 11. AIAA paper 901243. 14 D.v. Blackwood. Some of these sources contain extensive additional bibliographies. of the Amerkan Control Conference. Chen and B. 2. J. Calvet. J. Montana. 25 J. 1990.. F. J.S . Necessary and sufficient conditions for LTI Representation . J. Lurie. R. Lurie. 22 B. Lawrence. Lurie. M. "J. T. Big Sky. 39. Hartley. System. Hawaii. September 1995. March 2000. Chen. Implementation of nonlinear control laws for the RICST optical delay line. FrequencyDomainProperties o f Scalarand Multivariable Feedback Systems. N M .Looze. Control and Dynamics". *F.1. no. American Control Conference. Albuquerque. of Guidance.S.G. J. ChengChih Chu.Y. Enright.P. Active Member Bridge Feedback Control for Damping Augmentation. Damping and Structural Control of the JPL Phase 0 Testbed Structure. 1994. Chen.. Transfer function design approach is compared with statevariable methods in C9. An AlgorithmforStateSpaceFrequencyDomainIdentification Without Windowing Distortions.MicroPrecisionInterferometer:PointingControl N Y . Nov.Hadaegh. Proc. 27 J. and B. Fust USIJapan Conference on Adaptive Structures. O'Brien. andR. June 1997.442 Bibliography Additional bibliography Classical control history is well reflected in the bibliography to [ 111. Lurie. 23 P. 43rd National Symposium of American Vacuum Society.W. June 1997. J. ElEE Aerospace Conference. The following is a short list of selected recent publications and patents (most of them written by the authors' colleagues at the Jet Propulsion Laboratory or by the authors) reflected in or related to the book material. 5. S . FreudenbergandD. The bibliographies in [6. of Adaptive Systems with Sinusoidal Regressors. Albany. R. 18 J. 1989. Nonlinear Multiwindow Controllers.O'BrienandG. 15. Electrostatic Actuationof a Microgravity Accelerometer. J. Microelectronical Mechanical Systems Topical Conference. Lurie.Dolgin.on Adaptive Structures. 24 J. B. 16 D. S . S. no. L. AC. and B. 1992. J. Grogan. B. Optical Delay Line Nanometer Level Pathlength Control Law Design For SpaceBased Interferometry. Maui.F. N M . 15 D. 1996. 4th Conference on Control Applications. 1987. IEEE Trans. B.
NM. no. 57 April 1999. Pullmn. April 1991.670.Tannenbaum.Y. June 12. GlobalStability of Bdanced Bridge Feedback ZCCON 89. "Proc. J. 38 B. June 2.PdohlinearCctntmlof the OpticalDelayLinePathlength. Lin.1998. Lurie. December 6. 43 B. O'Brien. Neat. Lurie. J.5 1. AIAA Conf. Lurie and F. 1994. Albuquerque. Active Suspensions for Vibration Isofation.Y. 1998.NavigaEtiodand Control. Alexandria. Sept. May 25. Asymptofiedly Globally Stable Multiwindow Controllers. 1990.AeroSense'99.J. 4. W. Radar and Communication Transmitter. Applications of Multiwindow Co%&din. B a l a n d Bridge Feedback Control System.v. and J. Nov. B. 3. Fanson. Motor Control w i t h Active Impedance of the Driver. Hench. 18th Power Modulator Synnptysim".Bibliography 28 Y. ADPA/NAA/ASME/SPIE Conference on Active M Structures. A h m e d .af Damping. G.Lurie. 1988. USA 27. 1997. F.Socha. no. Aug. S . W. 5. Space 2000. Lurie.M. Acquisition and Ti@kirig with Nigh&& Plant and Nonlinear Regulation of Bode Diagram. 46 J. J.003. System Trades with BodeStep Control Design.TheLegacyof GW@ s. llLAA Conf. Hadaegh. 1021 10. L. Y. Nonlinear Dynamic Compensation for 443 .SPIEc&%€ert%ce. 45 B. Atlanta. A. 44 B. HiltonHead. A. on Guidance.A. Control". Mettler. J. J. J. 1992. a letter to Bob's Mailbox. F. 47 S . Integral Relations for Disturbance Isolation. Lurie. Y. The DialaStrut C m & d l afar Sanrctut. 1997. DSP Motor Control Boosts Efficiency In @@@It3 Design. US patent 5.119. Hadaegh. J. NASA URC Tech. Lurie.371.Spacer& Systems. Sirlin. m o l System with Digital Sensor. €3at&m. pp. JanFeb 2000. 1998.Lurie. 43. "Intern. 30 B. Control and Dynamics. Three loopbalancedbridge f d b a c k pointing sicmtpo. m. no. Hadaegh. Albuquerque. Y.EEEi T r m s d o n Automatic Control. F L . New Orlean. 41. Navigation and G6ritfd. 1 . 48 A. 40 B. B. 32 B. of the American Control Conference". 34 B. Hadaegh. 42 B. 1998.M. Lurie. Fanson and R. 35 B. Y. Ahmed and F.1990. 29 B. Daegas. B. Dolgin. Integrated Modeling Me&iodd@gy Vatlid&n Using the MicroPrecisionInterferometerTestbed:Assessment d' CfmdLoop Perhmmce Prediction Capability. J. "Rc. MayJune 1997. American Control Conference. ~ ~ ~ June 1997. 20. Guidance. Lurie.H. Murray. J. Electronic Design. 1988. Lurie.270. Lurie. 32nd SDM Conf. B. Conf. Lurie. Multiloop Balanced Bridge Feedback in Applicationfieeisisn to Pointing. Jerus 31. 'Melody. Baltimore. Hadaegh.A. 1990. Hadaegh and E. vol. 39 B. A. J.. MarcW4 1996. Ahmed and F. J. An Improved HighVoltage DC Regulator for a.Orlando.US patent 5. I. Ghavimi. USA Patent 4. Laskin. J. J. J. vol. A. L d e . Lurie. J.J. Mitter andA.Schier. 33 B.onGuidance. J. lJBE International Conference on Control and Applications. Wa. Lurie.932. J. J. Tunable TID controller. J. Lurie. J. Febr.. J. andF. Control Conference. 36 B. Lurie. ~ . J. 199 1. 37 B. VA.TorqueSensorHaving P !!@eked Sensor Element Support Structure.
13 13 [23.113] 11. J. CA. Garnica. ‘511. ADPA/AI. LA. Neat.9. 56 A.65] Ch. Toulouse. 50 G. IEEE Transaction Control Technology.28]. Colorado. 1. Noise and Distortion Reduction in Amplifiers using Adaptive Cancellation.9.191760.9.MultitoneAdaptiveVibrationIsolationof Engineering Structures. T. O’Brien.9.1[9.3 [22]. 2 2. Feb.99.9. August 1997. C.9.1[9.7 [9. 4 4.110.102.39.113] are generic for those topics covered in this book which are not conventionaly included in contemporary textbooks.J.108.119.9 1 3 0 1 .1.F. Lurie.41] Ch. 7.9.106. 9.2E9.SPIEProc.24]. Talwar.3.and D.9.9.5 [47] Ch.5 [19.113] Ch.53] Ch. Vibration Attenuation Approach for Spaceborn v.Calvet.2 [54]. 6.106]. O’Neal.120.81.105. J.8 [9.105.1[9.107.9 [l. Spanos.444 Bibliography 49 G.1. or indicate earlier publications.1 [37]. 11.24. 1 1 11. J.86.122] Ch.21. 54 J. 3.4 [49. 8.9.9.9.4 [9.13 [9. 9 9. 7.30.9.vonFlotov.4.‘7. 17.9. W.2 1351. [9] and [9. 12 12. J. 1992.11. 9. May 1997.7 [26.9. Albuquerque.113.9. 6.2.J. Vibration Monitoring and Control.7. 10 10. 2. 4.39]. SMACS 2.M.9. 2.111.99] Ch.2 [28.3 [17].33]. W. Neat.51]. 7. July 1994.4.38. 12. 11. 10.9 [9. Nanometer level optical control of the JPL phase B testbed.NarrowbandControlExperimentsinActiveVibration Isolation. J.5 [9. April 1995.7. [6].2 [9. 3.9.51.9.9. Spanos and M.9].69.5 [9. W.126. The references listed below exempli@ and expand on selected topics.71].Rahman. 55 J.109] Ch.9.5 [29.44.2 [9. 36th AIAA/ASMEYASCE/AHS/ASC Structures.4 [9.15. 9. Won.9. 1992. 3.ActiveVibrationIsolationonExperimental FlexibleStructure.O’Brien. on Active Materials and Adaptive Structures. B.9.7 [14. 6.AA/ASME/SPIE Conf.9.Belvin.158. Sulla. 3. [41].9. 7. W. no. 226401. Keyston.9.9.42] Ch.Spanosand 2. 7.Nerheim.9. J. W.9.40.1 [3. Joint Langley Research Center/Jet Propulsion Laboratory CSI Experiment.9.3 [34]. References to chapters References [2]. Melody.3 [9.107] Ch.36. 51 G.9.103.95. Structural Dynamics and Material Conference.1. A. A. 52 2. SPIE Proc.12 [ 9. 15th Annual AAS Guidance and Control Conference.ControlTechnologyReadinessforSpacebornOpticalInterferometer Missions. 1 1. Optical Interferometers.9. 11.2 [33].R.113] 10.9. Rahman.105. 9.29. Microwave Journal.128. Neat. B.176. 53 J.99.Abramovichi. Lurie.N.571.9.2S.6. 6 6. Bayard. San Diego.55]. SO] 3.11.162. 7 7. Smart StructuresandIntelligentSystems.A.41] .Rahman.71].21. 3 3.113.3.111.103. Ch.6. Melody.6 [29321.5 [9.9. 2.].113] Ch. New Orleans. Spanos.7 [9.31.1998. 1993. 8 8.118.9. 4. France. 321.2 [9.and J. 8.38.
.
P(s). 76 2 ' . 324 V . 212 U .plant transfer function. 19 T . 205.return ratio. 216.4 M. L .weighting matrix. 320321 Uan) .N .207208 " mass of a rigid body. 381 open loop system ratio of the output voltage to the source 381 emf.return ratio in an equivalent system. Q . 194 W(w). 77 ZL .quality factorof a resonance.thermal resistance. 38 1382 T(=) . 381 open loop system transmission coeffkient in voltage.coefficients in bilinear function. 277 . 207208 W .amplitude of fundamental at the output of nonlinear link.return ratio when two specified nodes are disconnected. 2 12213 RT .381. R . 116. orimmitance.amplitude of nth harmonic ata system's output.absolute temperature.381 open loop system transmission coefficient in current.threshold valuesof U causing jumps in E.load resistance. 2 T .transfer Eunction (immitance) of a regulator.302 NA . 31.Po($). 324 V .transfer functionof nominal plant. 276 TE . 77. 216.383 K E KI KOL KOL I KOL E   7's  .regulator functions with zero or infinite values of the regulating element w .sensitivity.U" .impedance of a twopole that is not zero at infinite frequency.voltage.return 'ratio abouttheplant.admittance.216.admittance of a twopole that is not zero at infinite frequency.amplitude of sinusoidal signal at the system's input. 1 12 P. 195 Q = 1/(2c) .216 Zb . 254 closed loop transmission coefficient in current.Horowitz sensitivity.load impedance. 194 W(O). W() .velocity.noise at the actuator input.impedance (mobility) of a systemwithoutfeedback. 253 R L .weighting matrix. 38 1382 Tp .204205 A 4 = VF. R(s) prefilter transfer function. 17 SH .446 Notation estimator gain matrix. 16 V .sampling period.207208 V voltage. 77 I" .return ratio in a system with the two specified nodes connected. 1 Po . 208 S .inductance. 147 U . 137 Q . 77 2 .impedance. 212.regulation of a symmetricalregulator. 209 T(0).transferfunction. 194 Y . 253 33 R.
saturation threshold. 98 is dB..polynomial coefficients....223 k . 54 td . 109 & ..vector of sensornoise.frequency of a zero.82 fa.. 293 f . 194 w .226 p1.294.147 r reference.151152 e.coefficients in modified Popov criterion. . 131 n .polynomial coefficients. command.unity gain bandwidth.. 249 ua .frequency in Hz. kc . pz. 8 fb. 293 e. 271272 q1. . 148 fT . az. 98 fc. v(t) .coefficient of forward propagation in voltage. e(t) . q2. or transfer coefficient of a variable element.q2. 254 2 s     .highest frequency at which loop gainxisdB.. 132 k .amplifier transfer immitanceor coefficient in a feedback system.Popov’s criterion coefficient.delay time.gain coefficient. u(t) . 79 u. 381 koE .147 q . 298 Un signal sample at the input to trapezoid integrator.7 is dB. 109 fhfrequency of a pole.coefficient of forward signal propagation relative to the source emf.lowest frequency at which loop gainx1 highest frequency at which loop gain is x1 dB.dead zone.spring stiffness coefficient.time.coefficient of forward signal propagation.signal at system’s input.control vector. 54 tr rise time. 381 kor . 170171 fi.lowest frequency at which loop gainxisdB.signal at the output of a nonlinear link.362 t .. 320 u .deadbeatthreshold.120.signal at the input a tononlinear link.381 ko .width in octaves of a trapezium gain response segment.settling time. admittance.transformation coefficient.sampling frequency. . 147 w impedance. 82 w . . 254 Vn signal sample at the output of trapezoid integrator.coupling coefficient. 249 s = CT +j o . .slope coefficient of an asymptoticBode diagram. 97.275 41.Notation 447 source impedance.polynomial coefficients. 54 ts ..frequency at which loop gain 0 fc. . 207208 379.323 v .151152 bl. .vector of process noise.303 k .operational variable.313 n . 54 u = In ahc.. bz.323 e d .central frequencyof a segment of constant slopeof a Bode diagram.motor constant. 147 v. 379 w . 212 at. 132 f P fs.polynomial coefficients.
6364 y . 208.damping coefficient. 12 z b braketorque.74 0 . 254 zl delay operator. 147148 A .difference.g.waveresistance(mobility).     +  21# Z .denotes parallel connection of impedances. 3 12 z . measurements. 213 0..= 21&/(21+ &) . e.displacementalongyaxis. 64 vector of state variables.displacement. Notation  nominal value of variable element in Bode symmetrical regulator.nonlinear function. 78 A maindeterminant.448 VU.output vector. 7 o .state estimate.137 o .312 loweramplitudestabilitymargin indB.angular velocity. or yn . r d s e c . 271 p .frequency at which phase shift is calculated. 208 . 207 ..0(s).torque. 213 = 142Q). 65 x~ . 384 C2 . rad/sec. . 384 & minor. 000) . 254 y. radsec.angle of rotation. 79 # .phase stability margin.frequency.transfer function. 23 1 z . 249 x 1 upperamplitudestabilitymargin indB. 249 y . 195 x x x .free run angular velocity.
W.75.246. passim integral of admittance.392 step.269. 197199. 206207.385 Biquad. 2092 11 electrical to thermal. 179. 237. 17. 324 Active suspension (vibration isolation). 382384 Block diagram transformation.372 integral of real part. 205208 feedback to parallel channels. 97.107 diagram. passim Coil.: formula.98 Asymptotic Bode diagram. 152. 335 Collocated control. 1.357. 220.136. 7879 integral of imaginary part (phase). 101 Bias. highfrequency. 193.324 Analyzer. 133134 of describing function. 257265 Aliasing.392 integral of gain. 270 Automatic level control.386 Bridged Tcircuit. 219.386 Wheatstone. R.189 Cold controller switching.220. 188.371 Cauer. H. signal.27027 1 Accelerometer. 97100.349 feedforward. 1112. B.189 Bifurcation points. 40.4. 1 maidvernier arrangement. 213 Bridge: balanced. 188. 296 Asymptote.S . 153 Approximation: of constant slope Bode diagram. 139143 Brake (stall) torque.79.220. 257 Adaptive system. 10.372. directand indirect. 147.361 Antenna.25. 40.385 Bandwidth: functional.232. 7. 156157 Analogies: electrical to hydraulic. 269 Bifilar coil.325326 Bilinear function. 110 Bangbang control. 216. 17. 104. 5759. 7677 phasegain relation. 106. : twopole implementation.396 Cutoff. 13 11 33.228 Backlash.35.206207 Adaptation. 175 Burst of periodic signal.Absolute encoder.180 Chart: Nichols.152 Black. 4043 Bode. 297298 Balanced bridge feedback. 78.310 Closedloop transfer function. 322324 Absolute stability.228229 Actuator. 12. 268 Bessel filter. 194. 208209 electromechanical.393 integral of resistance. 234 Absolute process stability.7576. 14. 317 Causal system. 3940 motor. passim Asymptotic global stability. 183 Clegg Integrator. 299 Basin of attraction. 1161 19. 3536 Blackman.308309.360 Acquisition and tracking system. H. 11 1 of openloop feedback. 235. W. 1920 Back emf. 65. 373375. 149. passim piezoelectric. 260 449 . 150. 798 1.323 feedback to connection of twopoles.223.271. bifilar..224225 voice coil.327 RCimpedance. 11 1 of closedloop feedback. 77 integral of feedback. 3. 19. 237. .
212 back emf. passim for unstable plant. 270 Conjecture of filter. 182 switched capacitor. 294295 Design sequence.220. 184186 Compound feedback. 20. 137. 119. 175. 185 Index Delay.367 Dashpot.312. 6162 NyquistBode. 137139. passim Command feedforward. 9697.10.4 18 Dominant poles and zeros.193 Corner frequency. 190192. 341342 vector. 171172 Disturbances. 254 Exact linearization. 34 static (steady state). 3 11. 249 modern.394 "ID (tiltintegralderivative). 95. 1. 135136. 297 inverse.207 Dead band. 299 Dead zone.269. limit.398 Diagram: asymptotic Bode. 268 Damping coeficient.235 Driver. 4345.133. 249 Controller: composite multiwindow.228 Encoder. passim InceStrutt.232. 291292 approximate formulas 296 of dead zone. 55 Decoupling matrix. 239 loading. 367 Drift.381 . 2 13 Nyquist.253 multiwindow. 4445.218. dead beat band.8 Crosssectioned feedback circuit. passim Dynamic nonlinearity. 7.337338. 16. 110 common. 3 132 Commander. passim Bode. 143. 1. 21 8. 192. 33 1 input matrix.383 Conditional stability. 291 Control: bangbang. 25. 362 Current feedback.232234 timeoptimal. 2 balanced bridge. 173. 70 POPOV.325327 Fano R.173 Electromotive force(emf). 1 l*12. 292 phasegain relations for. passim Decade. lag. 294295 of three position relay. passim Criterion: Nyquist.44 Compensator. 262264.1131 14 Describing function. 115.54. 15 1. transport.382 current. 21 8.63 Crossover frequency. 387 of saturation. 6769 Differentiator. 186. 132 Coulomb friction. lead. Mathieu's.73 Estimator gain matrix.218. 9. 238 Error.234 Equation.450 Command. 2526 Cycle. 3738 compound.386 bandwidth (range) of closedloop feedback.85.382 Current regulator. 227 Coupling. 27 1275 Critical point. 174175. 332 PID (proportionalintegralderivative). 143. 1161 19. 299 collocated. passim Dither. 1 feedforward. 333347 noncollocated. 11 1 of openloop feedback. 293295 of hysteresis.312.25. 426 Falling branch.223. 174. 333347 fuzzy logic. 247 Dynamic range. 1. 33. 223. 260 fuzzylogic. 104. 11 1 functional. 154..146.392 Feedback. 135.
passim Floating capacitor.399 Fuzzy logic controller. 218.223.381382 voltage. 20921 1 Hysteresis.77.405 Hs method. 4 local. 297 negative. 176 state variable. 172173 Fourier: formulas. 3637 negative 3. 199. structures).6566.270.254 integral of.87 333. 8 Nyquist. 193194. 150.351.104. 7475. 213 Frequency: crossover. 1819 Hot controller switching.. parts.399 Homing system. IN. 217. 225226 Generator: sawtooth.97 parallel. M. 328 Harmonics. 339340 Kalman.360361. 335 Hydraulic systems. 18. flow chart. 186.96. 180.371 minimum phase. 371 transfer.298299 sweep. 34 Filter: active RC: multiple feedback. 101 Butterworth. 259 Gear. 28929 1 Heater. L. 5657 conjecture of.2 15 Flux. 67 Gyroscope. 255256 Hard oscillation.: theorem. 12. 5657 Chebyshev.76. 147153 Friction: Coulomb. 217 series. 25 1.97 path.245.394 multiloop. 209. 249 Gain scheduling. 229231. 180 . 208 Force: analogy to current.391.117118. 251. 3 136 Black’s. 1 rate. 198199 Global stability. 3536 command.230 canonical twopole.382 Feedforward. 3. 177 switched capacitor. 234. 184.394. 3 10 error. 291 Ground. 2 17.3 153 16 Goldfarb. 116119. 205207 analogy to voltage. 87. 264 Flexible (appendages.323. I. 7.S. 3 133 14 Harmonic balance. 96. 148 response. passim sampling. 332 Gain matrix. 379 Free run velocity. 220. 3738 maximization. 3133.385 lossless impedance. 176 SallenKey. 208 back electromotive.228 Foster. 5. modes. 363 Furnace.394395 positive real. 86 bilinear 149. 218. 5759.271 Follower. 262 linear phase.272. 291 filter fork. 152.441 sensitivity. 226227 Fullstate feedback.382 positive.76. 227 viscous. 351. plant. passim Horowitz.58. 1 fullstate. 115. 29 1 law.187 Guardpoint stability margin.Index 45 1 Fourport network. R.262. 57 transversal.254 Function: allpass.186 antialiasing.412 error.. 157 Bessel (Thompson).79. 7576 large.
234 Lead compensator. 3839 gain.158.238239 Links. 212213 resistance. 2 13 Loading diagram.172 switchedcapacitor. 207208 229230 load.7 tangent. 417 nested. A.228229 effect on feedback. 269270 Second method.390. 33 local. 7879 phase. 29 1 Ladder network. 6367.452 phase lag of. 188 Integrator Clegg. 23 1 Linear Quadratic Gaussian.394396 .75. 68 Inverse describing function. 233234 Lyapunov. 271 LVDT. 206207.417 Intermodulation. 23023 1 Lurie. passim opamp implementation. 7576 gain. 40. 33 phase shift. 270 Mainvernier system. 234 orbiting stellar. 21 1212 Kinematic nonlinearity. 256 Load: cell.311312 main and vernier. 135436.405406. 174175. 212 wave (characteristic). 7 inner. 38 outer. 1 time variable (LTV). 9..10 LMI. 213 Loading line.23 1 Incremental encoder. 297 Impedance. First method. 135136. 253255 Linear Quadratic Regulator. 324327 Junction of links. 363366 Laplace transform. 33 Integral of admittance.3940.175. 292 Is03 isoE. 362 Inner loop.267 Kochenburger.8687 Lag: compensator.174. 417 Margins stability. 3940.352353 Lplane. 3738. 7677 Instrumentation amplifier. 255 Lossless distributed structure. 206 RC. 134 link.182 Limit cycle. 212213.180 source. I.381 of lossless system.311312 crossed.154. 37. Index nonminimum phase. transmission. 184 Interferometer: laser. 113.8586. 362 Laser interferometer. 1. R.405406. 253254 Linear systems. 12. 199. 3 10311 discrete trapezoidal. 77 feedback. mechanical. 268 Line. 3 14315 Internal feedback. 78 resistance. M.. 17 1.74. A. 41 Loop transfer recovery. isow lines and responses.264 Loop: common. 146149 halfintegrator approximation. 213 Logarithmic amplifier.118. 219220 impedance. 171 Jumpresonance. 300306 Johnson (Nyquist) noise. 234 Initial value and finalvalue theorems.394 Laplace transfer f'unction.
174 unitygain frquoncy. H. 254 output. 3481347. 170174 instrumentation. 302 Mason's rule. 4142. . 7475. 249 MIMO system. 2112 Nyquist.394 Mobility.332 PkaYl08d.112 Nominal plant. 246 Minimum phase function.277 Mathieu's equation.10. with lscd feedback.133. 17.185 estimator gdn.113.17 Nonlinear dynamic compensator: based on absolute stability. 4769 frequency. 148 noise. 171 stability. 10. 206. 15. 193.417430 Pilot signal. 6465?374 Octave. 228. 185.25 1.Index 453 based on RF. p ~ s i r n Participation francdong.259260.364. 1 collocated. 254 of resistance. 170171 Optical encoder. 172174 pinout. 4445. 376378 Matrix: controlinput. 332 Participation rulm. 253 PID controller. 17 Nsnminimum phase shift (lag). 123124.387. passim effect on plant uncertainty. 9.234 Optimality.224225. 74. 351. 9496 Overshsot. 299 Opamp. passim Minimum performance boundary. 249 plant noise distribution.54. 115. 188 inverting. 99. 2 18 Multiloop feedback system. 254 gain. 171 noninverting. 394 Piezoactuator. 333347 Nichols chart.338. 254 system.337340.260 . 3 11312 Nonlinear product coefficient.213 series.59.341. 254 at system's output.341. 40. 1141 15 Noncollocated control.60. 4346.280. 359 Phase plane. 1121 13 process. 140141. 31. 87. 254 Norton's tborem. passim linear.337338. 206. 171 sensor.with parallel channels 305384. 186. 232234 Nonlinear distortions. 8586.327 Noise: at actuator input.16. 173 Onoff contrsl.79. 249 decoupling. criterion. 23 1 MATLAB @ commands and program listings: 79. 55 Offset voltage. 386309. 198192.70 diagram. 3637 Multiwindow controller. 214215 Motor. 405 Nonlinear inbraction betwean loa11 and common loops.333. 276286 Norm: H.223.262 Plant.. 20. 1314. 100104.255 quadratic. '238 Matching. 119. 394 safety. passim for unstable plant.213 permanent magnet. 150. 180181.25. 1171 18.246.232.111. 116118. 12. 249 measurement.243145. 6142. 344347 based an RF.
266. 105 Spectral density. 253255 voltage. 217 Rate limiter.266. 120121.218.2. passim Specifications. 8990. 56. 416 Signaltonoise ratio. 298 Sensitivity. 232233 template.303304.381382 Series. M.250 Source impedance. 156 Sampling frequency and period. 179 Rate sensor. 271 Popov stability criterion. 245247 Quality factor. 115. 322 Proof mass. Laurent. 101102 Prewarping.45 Reference. 1718 Horowitz.405 Ratestabilized.27 1275 Port. 147153 Saturation. 116119.5960. 178179.2 Servomechanism. 379 SIMULINK @. 6769. 299 Resistance: integral of.399 QETT. 16. 31 1312 with frequency depended threshold. 233 Return difference. 186 Rate gyro. 345 Siderostat. 213214 unstable.249. 213 thermal. 195196 current. 7677 Index load.120 Pole placement. 236 Singleloop generic system.2.380 Return signal (fed back signal).32032 1 . 299. 3334. 2526 Linear Quadratic (LQR). 137. 145 P'WM (pulsewidth modulation). 216 Prefilter. 54 Shaker. 296 Schmitt trigger. 208 Resolver. 123124. 1141 15. V.454 flexible.293 twoposition. 3 1.95.264 Regulator: Bode variable symmetrical. 154 Rules: Masons'. 2 12 Spacecraft. 254 nominal. servomotor. 309. 36. 372 Servo.302 SallenKey filter.5455. 150 Process instability.13 SPICE models or program listings. 4 142.9697.362 Safety margins. 18.19 Sensor noise. 234 RCimpedance chart.234. 115 noncollocated. 1.247249. rate feedback.227229. passim Relay: three position. 5 .111.293 in local and common loops. 253254 Linear Quadratic Gaussian (LQG).277 participation. 245 tolerances (uncertainty). 222. 19. 116. 329 for timedomain to frequency domain conversion. 119. 143145.. 360 PSPICE @.54 Root locus.2 Settling time. 144145. passim identification. 119. 110. 25726 1 noise distribution matrix. 247248 Popov.11. 183 Redundancy. 105. 176 Sample and hold.113 Series feedback. 392393 RTI (real time interrupt).1 Rise time. 153.380 Return ratio. 235.
325326 Thrusters. 213 Tracking. 85.405406. 230 with unstable plant.340342. 362 Tustin. 23 1 Transport delay (lag). 268 Static nonlinearity. 270 homing.228229.207. 54 Timeoptimal control. 97 Structural design. Type2 systems. 270.15 1 Twopole network: Cauer.211212 Subharmonics. passim multiloop.405. at links’ junction. 40.406 Wave impedance.416419 Voltage feedback.18 1 Foster.113114 Tunnel effect accelerometer. 149. rise. 45. 267 Static error. 347 Tustin A.196 System: absolutely stable. 180.405406. 27027 1 asymptotically globally stable. 180. 15. 270 global.388 Transfer function. 185. 343345 TID controller.390.382 Voltage regulator.58. 220 Twoposition relay.120 Vanishing signals. 6465.. 1841 86 Symmetrical regulator. 213 Vernier actuator. 343345 Timeresponse to step command.206. 6769. free running. 119. settling.2 systems.228229 Voice coil. capacitance.279. 2 12 Threevalued function. 14915 1 Transformer: balancedtounbalanced. 7274 with distributed parameters.1. 7274 Unstable plant. 207. brake (stall). 269 margins. 249 multiinput multioutput (MIIMO). 119120 .291 transform 149. 308309. 270 conditional. 2. 234 Thevenin’s theorem. 363 Transform: Laplace. 3637.417 Vibration suppression (isolation).18 1 TwoPo~~. 322 verification. 315317 Stall torque. 12. 1 tracking.Type 1.262 State estimate.99. 254 State variable filter. stiffness. passim Torque. 249 Static attractor. 3 16. 177 State variables. 67 of process.217.96. 327329 Switched capacitor circuits.268 Velocity. 3940. 1 singleloop. 56. 249 VCO (voltage controlled oscillator).59.73 Stiffness coefficient. 192. 231 Stability: absolute.2 13 Star tracker.417 matrix. mass. 45 Type 0 .315317 local. 108109.193 Time delay.46. 299 Type 0 . 225226 Transmission line. 189 flowchart. passim Nyquiststable. 234.3940. 27027 1 asymptotic. 282 linear timevariable(LW).394 guard point.208 Stray inductance. 6769. 212 statespace. 6367. 4346. 195.Index 455 Tachometer. passim singleinput singleoutput (SISO). 238239 mainvernier. 54. 158. 270 Variables.
219. 278279. ztransform bilinear): (Tustin..336338 psynthesis. Wind Wind antiwindup controllers. Weight 7980. .456 function. panalysis and 336 Windows: nonlinear. 191. 152 tables. 333334 Index Yoyo. 343 255 Zames.385 12 disturbance. 255 phenomenum. 253254 Whetstone bridge. 178179 in multiwindow controllers. G.255 Weighting matrix.