Actuators Basics and Applications

Actuators
H. Janocha (Ed.)
Actuators
Basics and Applications
With 303 Figures and 47 Tables
€1 Springer
Professor Dr.-Ing. habil. Hartmut Janocha
Universtat des Saarlandes
Fachbereich Elektrotechnik
Lehrstuhl fiir Prozessautomatisierung
66123 Saarbriicken, Germany
e-mail: janocha@lpa.uni-sb.de
Library of Congress Control Number: 2004109912
ISBN 978-3-642-08266-5 ISBN 978-3-662-05587-8 (eBook)

DOI 10.1007/978-3-662-05587-8
This work is subject to copyright. All rights are reserved, whether the whole or part of the material
is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication
of this publication or parts thereof is permitted only under the provisions of the German Copyright
Law of September 9, 1965, in its current version, and permission for use must always be obtained from
Springer-Verlag Berlin Heidelberg GmbH. Violations are liable for prosecution under the German Copyri-
ght Law.
springeronline.com
©Springer-Verlag Berlin Heidelberg 2004
Originally published by Springer-Verlag Berlin Heidelberg New York in 2004
Softcover reprint of the hardcover I st edition 2004
The use of general descriptive names, registered names, trademarks, etc. in this publication does not
imply, even in the absence of a specific statement, that such names are exempt from the relevant pro-
tective laws and regulations and therefore free for general use.
Typesetting: Data of authors, final conversion by LE-T£X Jelonek, Schmidt & Vockler GbR, Leipzig
Cover design: medio Technologies AG, Berlin
Production: LE-T£X Jelonek, Schmidt & Vockler GbR, Leipzig
Printed on acid-free paper SPIN 10545507 57/3141/YL 54321o
Preface
Actuators have wide-ranging applications in our everyday environment, from

CD players, cameras, communication devices, household appliances, radia-
tors, air conditioners, machine tools, cars, watercraft and aircraft to respira-
tors and artificial extremities. Actuators are an indispensable system compo-
nent whenever systems must be controlled and are equally essential for the
implementation of mechatronic systems and in microsystem technology. In
this regard, they are even "more necessary" than a sensor. Sensors only fulfill
real system functions in a closed control loop providing the processor with
all the required process information so that the processor, as an executive
component, is able to provide the actuator with the "correct" signals.
Seen from this point of view, it is surprising that during the past few
years, the focus has mainly been on sensor technology rather than the de-
velopment of actuators. One of the reasons might be the growing desire to
gain as much information as possible about processes with the aim of bet-
ter technical, economical or ecological control. This leads to a wide variety
of new sensor developments with respect to the type of measured quantity,
operating principle, properties of the device, and application field. The fact
that the price of computing power has decreased steadily led to additional
possibilities of gathering process information, such that the application of
microprocessors in very small systems seemed useful.
This development triggered new impulses for the field of actuator engi-
neering. With the establishment of microelectronics it became possible to
connect actuators directly to the processor. Additionally, the lower costs of
computing power allowed the introduction of more modern methods of con-
troller synthesis, improved control algorithms, and increased the importance
of actuator principles and materials. This development resulted in the inte-
gration of electronics into the purely mechanics-oriented field of measuring,
control, and automation technologies of the time. The final controlling ele-
ment became a microelectronic-compatible actuator, and the final controlling
equipment became an actuator system, in some cases with a standardized
interface, and sometimes equipped with sensor technology to improve the
imperfect transfer properties.
This book elucidates the ongoing development of actuators. It highlights
a broad range of potential application fields giving representative examples
of these applications. It is written for engineers of measuring, control, and
VI Preface
automation technologies, and especially for potential users of actuators and

actuator systems, as well as for students of the related sciences. For this
group of readers, it will serve as a reliable source of information and will
also provide valuable additional reading for lectures and courses. However,
it should not be mistaken for a reference book for those who "already know
it all." Readers of this book should be able to conceive actuator systems for
their own applications to be able to compare their own alternative solutions
carefully with other, existing potential solutions in terms of advantages and
disadvantages.
A good deal of the diverse material has been subdivided in the same way
as in the German book, "Aktoren," published in 1992 by the same editor.
Four other authors, all of them acknowledged experts in their field, agreed to
write four chapters of this book together with me. Some of them had already
participated in writing the German book. I wish to thank these authors for
their participation and their contribution to this book. Also, I wish to thank
the many members of staff, whose names cannot be listed here, but whose
support was essential for the publication of this book. Finally I would like to
thank the publisher Springer-Verlag for the appealing layout of the book.
This book is probably the first comprehensive work for experts about
actuators. Its authors, editor, and publisher will gratefully consider any sug-
gestions from readers, as they might contribute to the quality of this work.
Saarbriicken, May 2004 Hartmut Janocha

List of Contributors
Chapters 1 and 5: Prof. Dr.-Ing. habil. Hartmut Janocha

Lehrstuhl fiir Prozessautomatisierung
Universitiit des Saarlandes
66123 Saarbriicken, Germany
janocha@lpa.uni-saarland.de
Chapter 2: Prof. Dr.-Ing. Stefan Butzmann

Lehrstuhl fiir Elektronische Mess- und Schaltungstechnik
Ruhr-Universitiit Bochum
UniversitiitsstraBe 150
44 780 Bochum, Germany
stefan.butzmann@lems.ruhr-uni-bochum.de
Chapter 3: Prof. Dr.-Ing. Hans-Dieter Stolting

Institut fiir Antriebssysteme und Leistungselektronik
Universitiit Hannover
Welfengarten 1
30167 Hannover, Germany
stoelting@ial.uni-hannover.de
Chapter 4: Prof. Dr.-lng. Wolfgang Backe

Institut fiir fluidtechnische Antriebe und Steuerungen
Rheinisch-Westfiilische Technische Hochschule
SteinbachstraBe 53
52074 Aachen, Germany
wolfgang.backe@ifas.rwth-aachen.de
Dr.-lng. Andreas Klein
Adam Opel AG
Riisselsheim, Germany
Contents
1 Introduction
Hartmut Janocha 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
1.1 What is an Actuator? 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
1.2 Actuator as a System Component 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3
103 Actuators in the Fields of Mechatronics and Adaptronics 0 0 0 0 0 0 0 0 5
1.4 "Intelligent" and Self-Sensing Actuators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8
1.401 "Intelligent" Solid-state Actuator 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9
1.402 Self-sensing Solid-state Actuator 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10

Operator-based Modelling for Large-signal Operation 0 0 0 0 11
Reconstruction and Compensation Filter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 11
Example Application: Piezoelectric Positioning System 0 0 0 12
1.5 Design of Actuators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 13
References 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 7
2 Power Circuits
Stefan Butzmann 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 19
201 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 19
202 Power Switching Devices 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20
20201 Overview 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20
20202 Power Diode 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 22
20203 Bipolar Power Transistor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 24
202.4 Power-MOSFET 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 28
Static Behavior 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 28
Dynamic Characteristics 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 31
20205 IGBT 0 0 0 0 0 0 0 0 0 0 o 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 33
20206 Thyristor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 36
203 Drive Circuits for MOS-gated Devices 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 37
20301 General Remarks 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 37
20302 Low-side Drive Circuits 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 39
20303 High-side Drive Circuits 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 41
Drive Circuits for p-channel MOSFETs 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 41
Bootstrap Drivers for n-channel MOSFETs 0 0 0 0 0 0 0 0 0 0 0 0 0 42
Charge Pumps 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 44
X Contents
2.3.4Isolated Gate Drive Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Transformer-coupled Gate Drives . . . . . . . . . . . . . . . . . . . . . . 45
Gate Drives with Opto-couplers . . . . . . . . . . . . . . . . . . . . . . . 48
2.3.5 Speed Enhancement and Protection Circuits . . . . . . . . . . . . 49
Turn-off Speed Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . 49
Overvoltage Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Protection Against Gate-Source Overvoltages . . . . . . . . . . . 52
Overload Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
dV / dt Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Overtemperature Protection . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.4 Thermal Management....................................... 55
2.5 DC Switching Converters.................................... 60
2.5.1 Topologies.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.5.2 Electromagnetic Value Actuation with a Class-D Stage... 62
2.5.3 Stepper-Motor Drive Circuits . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.5.4 Zero-current Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.6 Current Development and Future Trends . . . . . . . . . . . . . . . . . . . . . . 74
2.6.1 Overview........................................... 74
2.6.2 Trench-FETs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.6.3 CoolMOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
2.6.4 "Intelligent" Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.6.5 Smart Power ICs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.6.6 Silicon Carbide. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3 Electromagnetic Actuators
Hans-Dieter Stolting............................................. 85
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.1.1 Definition of Electromagnetic Actuators . . . . . . . . . . . . . . . . 85
3.1.2 General View . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
3.1.3 Basic Construction Possibilities . . . . . . . . . . . . . . . . . . . . . . . 90
3.2 Self-clocked Motors with Mechanical Commutator . . . . . . . . . . . . . . 92
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
3.2.2 Electrically Excited DC Motors . . . . . . . . . . . . . . . . . . . . . . . 92
3.2.3 Permanent-magnet DC Motors........................ 94
3.2.4 Pulsating-current DC Motor . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.2.5 AC Commutator Motor .............................. 101
3.3 Self-clocked Motors with Electronic Commutator ............... 104
3.3.1 Brushless DC Motors ................................ 104
3.3.2 Servo Motors ....................................... 111
3.3.3 Switched Reluctance Motor ........................... 114
3.4 Externally Commutated Motors .............................. 115
3.4.1 Asynchronous Motors ................................ 115
Appliances and Design ............................... 115
Method of Operation and Performance Characteristic . . . . 117
Contents XI
Control Methods 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 122
Shaded-pole Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 124
3.402 Synchronous Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 125
Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 125
Permanent-magnet Rotor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 127
Hysteresis Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 130
Reluctance Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 131
3.403 Stepper Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 132
305 Limited Motion Drives 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138
30501 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138
30502 Electromagnets 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138
Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138
DC Magnets 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 139
AC Magnets 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 141
Swinging-armature Magnets, Vibrators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 142
30503 Voice-coil Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144
306 Closed and Open Loop Control Methods 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144
30601 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144
30602 Line-commutated Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 145
30603 DC Chopper Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 145
306.4 AC Chopper Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 146
30605 Converter for Polyphase Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 147
307 Applications 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 147
References 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 153
4 Fluidpower Actuators
Wolfgang Backe, Andreas Klein 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 155
401 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 155
402 Proportional Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 157
40201 Electromagnetic and Electromechanical Converters

as Pilot Stages 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 158
40202 Electrohydraulic Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 160
Servo Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 161
Proportional and Control Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 167
40203 Electropneumatic Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 171
Steady Pneumatic Directional Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 171
Pneumatic Proportional Pressure Control Valves 0 0 0 0 0 0 0 0 1 74
403 Switching Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 176
40301 Electrohydraulic High-speed Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 177
40302 Electropneumatic Switching Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 178
4.4 Fluidpower Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181
4.401 Hydraulic Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181
Rotary Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181
Translatory Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 183

XII Contents
Static and Dynamic Behavior . . . . . . . . . . . . . . . . . . . . . . . . . 184

Open Loop of a Hydraulic Drive ...................... 185
4.4.2 Pneumatic Motors ................................... 186
Rotary Motors ...................................... 186
Translatory Motors .................................. 186
Static and Dynamic Performance ...................... 186
4.5 Sensors ................................................... 192
4.5.1 Pressure Sensors .................................... 193
4.5.2 Flow Sensors ........................................ 195
4.5.3 Position and Rotary Sensors .......................... 196
4.6 Controller Types ........................................... 197
4.6.1 Control of Hydraulic Servo Systems .................... 198
4.6.2 Single-loop Control .................................. 199
4.6.3 Multi-loop Controls .................................. 200
4.6.4 Formulation of Control Design ........................ 202
Linear Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
Non-linear Additional Functions ....................... 205
4.6.5 Control of Pneumatic Servo Systems ................... 206
Position Control. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Pressure Control .................................... 211
4.6.6 Realization of Controls with Microcomputers ........... 212
4.7 The Fluidpower Drive as a Subsystem ........................ 214
4.8 Comparison Between Fluid power and Electric Drives ............ 215
Comparison of Characteristics ........................ 215
Advantages and Disadvantages of Fluidpower Actuators .. 216
4.9 Application Examples ....................................... 217
4.9.1 Automatic Anti-lock Braking System .................. 217
4.9.2 Servopneumatic Drives for Handling Systems ........... 219
4.9.3 Pneumatic Grippers ................................. 220
4.9.4 Simulation of Motion ................................ 222
4.9.5 Numerically-controlled Machine Tool .................. 224
4.9.6 Application and Operating Conditions
of Theater Hydraulics ................................ 227
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
5 Unconventional Actuators
Hartmut Janocha ............................................... 233
5.1 Introduction ............................................... 233
5.2 Electrochemical Actuator .................................... 234
5.2.1 Electrochemical Reactions ............................ 234
5.2.2 Technical Implementation ............................ 235
5.2.3 Application Examples ................................ 238
5.2.4 Development Trends ................................. 238
Contents XIII
5.2.5 Comparison to Competing Actuator Principles .......... 239

Expansion Elements ................................. 239
Metal Hydride Actuator .............................. 240
5.3 Actuators with Shape Memory Alloys ......................... 241
5.3.1 Shape Memory Effect ................................ 241
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
Heating of Memory Components ...................... 245
Dimensioning of NiTi Components .................... 246
Flap Operation ..................................... 250
Model Making ...................................... 250
Heavy Duty Actuator ................................ 251
5.3.5 Comparison to Competing Actuator Principles ......... 252
5.4 Piezoelectric Actuators ...................................... 254
5.4.1 Physical Effect ...................................... 254
Materials ........................................... 256
Design of Piezoelectric Transducers .................... 260
5.4.3 Electronic Power Amplifier ........................... 270
Position Controlled Actuator Operation ................ 273
Mirror Drive for Laser Gyro .......................... 274
Diesel Injector ...................................... 275
5.5 Magnetostrictive Actuators .................................. 277
5.5.1 Physical Effect ...................................... 277
Materials ........................................... 278
Design of Giant Magnetostrictive Transducers ........... 280
5.5.3 Electronic Power Amplifier ........................... 285
5.5.4 Comparison Between Piezoelectric
and Magnetostrictive Actuators . . . . . . . . . . . . . . . . . . . . . . . 286
Underwater Sonar System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Dynamic Vibration Absorber ......................... 289
Hybrid Actuator .................................... 290
5.6 Actuators with Electrorheological Fluids ...................... 292
5.6.1 Physical Effect ...................................... 292
Materials ........................................... 294
Design of Energy Transducers with ER Fluids ........... 298
XIV Contents
5.6.3Electronic Power Amplifier .......................... . 299

5.6.4Application Examples .......................... ..... 301
Positioning Drive .......................... .......... 301
Tactile Element . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
Shock Absorber .......................... ........... 303
5.6.5 Development Trends .......................... ....... 306
5.7 Actuators with Magnetorheological Fluids ..................... 306
5.7.1 Physical Effect .......................... ............ 306
5.7.2 Technical Implementation .......................... .. 307
Materials .......................... ................. 307
Design of Energy Transducers Based on MR Fluids ...... 310
5. 7.3 Electronic Power Amplifier .......................... . 313
5. 7.4 Comparison Between ER and MR Fluid Actuators ....... 315
5.7.5 Application Examples .......................... ...... 316
Brake .......................... .................... 316
Clutch .......................... ................... 317
Motor Mount .......................... ............. 319
5.8 Microactuators .......................... ................... 320
5.8.1 Principle .......................... ................. 320
5.8.2 Realization .......................... ............... 321
5.8.3 Application Examples .......................... ...... 322
Microdrives .......................... ............... 322
Positioning and Gripper Systems ...................... 325
Microoptics .......................... ............... 328
Microfl.uidics .......................... .............. 329
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
Index .......................... .......................... ..... 337

1 Introduction
Hartmut J anocha
1.1 What is an Actuator?
An actuator is a functional element which connects the information process-

ing part of an electronic control system with a technical or nontechnical, e.g.
biological, process. Actuators can be used to control the flow of energy, mass
or volume. The output quantity of an actuator is energy or power, often in
the form of a mechanical working potential "force times displacement" 1 . The
actuator control is always achieved using very low electrical power, ideally
without any power consumption, with currents and voltages which are, if
possible, microelectronically (e.g. TTL) compatible.
An actuator's functional structure can be described by introducing the el-
ementary functional components "energy controller" and "energy converter"
(see Fig. 1.1). The output variable of an energy controller is the energy pro-
vided by an auxiliary power supply which is controlled via the input variable
as it is typically done with transistors and valves (see Fig. 1.1a). An energy
converter's input and output variables are energies. In the case of current
transformers and torque converters these two energies are of the same kind,
whereas the input and output energies of electromagnetic and piezoelectric
transducers are different (see Fig. 1.1 b).
As an actuator is supposed to control flows of matter and energy, an
actuator must contain at least one energy controller. This is why an actuator
is usually a series connection of an energy controller and an energy converter.
The common understanding, however, leaves out one important property of
actuators, and that is their controllability with a low-power electrical signal.
Subsequently, the term actuator refers often only to the energy converter,
whereas the energy controller is called a power amplifier or power circuit.
These definitions are not standardized but are accepted and used by the
scientific community around the globe.
For further reference, see the German DIN standard 19226 Regelungs-
technik und Steuerungstechnik (closed and open loop control [1]). Figure 1.2
describes a control system according to this DIN standard with the official
translation of the technical terms. Within the actuator (Steller), the controller
1 Transfer elements with a nonmechanical energy /power output, such as power

semiconductors, can also be considered as actuators.
H. Janocha (ed.), Actuators

© Springer-Verlag Berlin Heidelberg 2004
2 Hartmut Janocha
E~nergy Energy
Energy
converter
Electronic control element: Current/voltage transformer Torque converter

Transistor stage (Transformer) (Gear)
c -1 s
Cp
!.
Fluid control element: Electromagnetic Piezoelectric
Valve transducer transducer
a b
Fig. 1.1. Elementary functional components of an actuator. a Energy controller,
b energy converter
·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·-·I
Fig. 1.2. Typical block diagram of a closed loop control system (terms as defined
in DIN 19226)
output variable Yc is turned into the manipulated variable y (StellgroJ3e)

which is used to drive the final controlling element (Stellglied). This final
controlling element will influence the flow of matter and/or energy. Subse-
quently, the actuator definitions mentioned above are much closer to the final
1 Introduction 3
controlling equipment (Stelleinrichtung) and final controlling element (Stell-

glied) in the DIN standard. It is worth noting that the term actuator used
in Fig. 1.2 conflicts with the actuator definition presented above which shall
serve as the basis for this book.
1.2 Actuator as a System Component
Many controlling tasks that are required in the natural and artificial en-
vironment can be described with an open loop control chain, as shown in
Fig. 1.3. The focus is placed on operations and processes that must be mod-
ified to achieve a certain goal. This is where actuators come into play. Their
input signals are microelectronically compatible and are produced by the elec-
tronic controls inside the information processing part of the control system.
The electronic controls are often distributed and can therefore be assigned
to the individual processes with respect to location and function. They are
usually program controlled and can be implemented by means of a personal
computer. The user may modify the process via a so-called man-machine
interface (MMI), composed, in the simplest case, of an alphanumeric keypad
and a computer monitor.
Energy Process, e.g.

matter mechanical processing
(information) chemical reaction
biological metabolism
Auxiliary
energy 0
~
Man-
Energy Power Micro-
machine
converter electronics computer
interface
Actuator
Fig. 1.3. Open loop control of processes
Automated processes are often controlled by means of a closed control

loop (see Fig. 1.4). One of its key functions consists in measuring the charac-
teristic process variables which are then preprocessed and fed into the con-
trol processor. The control processor compares the measured values with the
given set values and, depending on the difference between the two, determines
the control signal for the actuator or the corresponding power electronics by
means of control algorithms in accordance with a control strategy which has
4 Hartmut Janocha
Energy
matter Process
(information)
i"------s~~~~;-·a;·---------1 ' Sensor &

i__ ____ ~~~~!r.<:J~~~s.______ _rl !1------~~~~~~~~-i~-~-------'
t ·.~:
------------------------------------------11 t··-----------------------------------------
Auxiliary
energy L_~~~===~-J]
~------11/0
Man-
machine
Software, e.g. interface
process identification
controller synthesis
Fig. 1.4. Closed loop control of processes
been installed in the computer. Any parameters derived by the control pro-
cessor to describe the process, for instance in the form of a mathematical
model, are determined by the control processor during an identification cy-
cle. These parameters are the fundamentals of controller synthesis within the
computer. On a higher automated level, the controller adapts autonomously
to the process-related changes of the parameters, e.g. due to wear (adaptive
control, AC).
The symmetric system arrangement in Fig. 1.4 shows phenomenologically
the duality of sensor and actuator technology in the field of automation engi-
neering. It is interesting to note that an actuator alone features all the prop-
erties in terms of structure and function which comprise a complete control
system including sensors and a signal processing part. A good example is the
piezoelectric actuator whose displacement is detected by strain gages which
are mounted directly on the piezocrystal in order to eliminate - analogous
to the methods for compensating error signals from basis sensors 2 - tempo-
rary or technology-related imperfections of the actuator such as temperature
dependency, nonlinearity or hysteresis of the output-input characteristic (see
Sect. 1.4.1).
Treating an actuator as a system component automatically raises the
question regarding the type of its interfaces. The output or process interface
can vary just as greatly as the range of actuator applications and is deter-
2 In contrast to the sensors used for compensating the errors of the actuator, basis
sensors detect the process variables of interest.
1 Introduction 5
mined finally by the particular application. The actuator's input interface,

described above as microelectronically compatible, is much easier to describe.
Researchers have agreed on certain standards allowing them to connect an
actuator to any control processor with a standardized interface. As actuators
are often included in real-time system concepts, the control processor must
process the required user programs on time and practically simultaneously.
Ordinary personal computers (PCs) with a standard operating system usu-
ally cannot accomplish this task, in contrast to process computers that have
the necessary properties such as real-time capability and interrupt handling.
However, it is possible to upgrade a PC to a compact process computer with
commercially available hardware and software.
System theory describes and investigates all kinds of systems whereas
systems engineering deals specifically with technical systems. After all that
has been said so far, one should take a closer look at what part actuators
have to play in the field of systems engineering, that is, in mechatronics and
adaptronics.
1.3 Actuators in the Fields of Mechatronics

and Adaptronics
The term mechatronics was first coined by a senior engineer of a Japanese

company in 1969, as a combination of mechanism and electronics. Over the
years, the meaning of this term has been broadened and is now used in
technical jargon to describe the philosophy of engineering technology rather
than the technology itself. As a result of the broader meaning of the term
mechatronics a number of definitions can be found in the literature. The
most commonly used definition emphasizes synergy and states: Mechatronics
is the synergistic integration of mechanical engineering with electronics and
intelligent computer control in the design and manufacture of products and
processes. According to this definition, actuators are innately mechatronic
components. This has already been shown in Fig. 1.4, though in a different
context. Mechatronic systems can be implemented in many different ways and
cover the whole range from industrial robots and CNC machine tools to anti-
block and anti-slip systems in automobiles as well as CD players and photo
cameras. Here, actuators come in the form of solenoids and electro motors
(Chap. 3) valves and fluid-based motors (Chap. 4), as well as unconventional
actuators (Chap. 5), whereby dedicated power electronics (Chap. 2) have
a strong influence on the success of technical development.
In the 1980s there was a trend to miniaturize mechatronic systems. It
was at that time that in Germany the term Microsystem Technology (MST)
and in the USA the related term of Microelectromechanical Systems (MEMS)
were established. In this book both terms are defined as the development and
integration of sensors, actuators and other three-dimensional structures on
6 Hartmut Janocha
the scale of, and using the fabrication techniques of, integrated microelectron-
ics. Whereas the term of mechatronics emphasizes the aspects offunction and
design, this definition highlights the production technologies. The importance
of micro technology in the field of sensors is obvious ( "microsensors"), though
it is rather limited in the field of actuators, since the most common actuator,
namely those with mechanical output power, will perform only small dis-
placements and/or forces with a micromechanical structure (see Sect. 5.8.1).
Microsystem technology, however, becomes important when it comes to elim-
inating electronically the imperfections of actuators, such as the dependence
on temperature or nonlinear characteristics. Therefore, it is useful to minia-
turize sensors and circuits by means of micro technology and to fit them into
the transducer casing. The end-product is called an "intelligent" actuator
(see Sect. 1.4.1).
The term adaptronics initially introduced and defined in the German
language ( "Adaptronik") in 1991, is the generic term for disciplines that
are known on an international level under names such as smart materials,
smart structures, intelligent systems, etc. [2]. The term adaptronics des-
ignates a system and its development process wherein all functional ele-
ments of a conventional control loop are existent, whereby at least one el-
ement is applied in a multifunctional way. The conformity with a control
loop guarantees that the structure shows autonomic adaptive characteris-
tics and can thus adapt itself to different conditions. The difference from
the classical control circuit, where normally each single function is fulfilled
by a separately built component, is achieved through the use of multifunc-
tional elements. Materials represent the essential basis for all multifunc-
tional effects. The conception of multifunctional elements is therefore mainly
based on the made-to-measure production of functional materials, wherein
several functions are interlinked at a molecular level. Functional materials,
which are characterized by a high potential of functional and application
options, are amongst others: shape memory elements; bimetals; electrorhe-
ological, magnetorheological, thixotropic, and rheopex fluids; piezoelectric
elements; electrostrictors; magnetostrictors; chemochromic, electrochromic,
hydrochromic, photochromic, and thermochromic elements; and functional
gels [3].
Controlling actively the structure geometry is a typical task performed by
adaptronic systems. The response characteristic of structures and sensors is
detected on-line and is controlled by means of actuators. Piezoelectric stacks,
for instance, are used as active braces in truss structures, while piezoelectric
flexural transducers deform surface structures such as plates and shells (see
Fig. 1.5). In this particular application, the piezoelectric transducer can per-
form its actuator function and make use of its sensor properties at the same
time. These "self-sensing actuators" (see Sect. 1.4.2) allow the implementa-
tion of smart structures. Their operation now involves far fewer devices and
less installation effort (compare Figs. 1.5a and 1.5b).
1 Introduction 7
Controller Power supply
Power supply
Fig. 1.5. Controlling of surface structures. a With standard actuator-sensor con-

figurations (A: actuator , S: sensor) , b with linked self-sensing actuators (A/ S: adap-
tronic actuator-sensor module)
In a specific case, a number of self-sensing actuators is mounted on the

structure. These actuators exchange sensory information with each other and
with the host processor, for instance regarding the structure's eigenmode. The
processor provides a structure model which uses this information to generate
control signals for the actuator operation. These signals are fed directly into
the corresponding adaptronic modules allowing the user to control the sur-
face form for achieving the mechanical (light-weight construction, machine
tools, adaptive wing, active vibration damping) , optics or acoustics (adaptive
optic/acoustic, noise compensation) tasks at hand.
The fact that actuator and sensor properties are collocated is characteris-
tic of adaptronic actuator- sensor modules. This proves advantageous for the
design and the operation of the controller, as controllers with simpler stabil-
ity criteria can be implemented (e.g. PPF controller [4]). As a side effect of
8 Hartmut Janocha
the collocation, it is possible to test the piezoceramics and/or the structure

on-line for damages (auto diagnostics, health monitoring).
1.4 "Intelligent" and Self-Sensing Actuators
The concepts of "intelligent" and self-sensing actuators mentioned in Sects.

1.2 and 1.3 are exemplified below with solid-state actuators, such as piezo-
electric or magnetostrictive actuators (see Sects. 5.4 and 5.5, respectively).
The potential of both concepts is especially easy to recognize and to compare
when described in terms of system theory. We will start with the conventional
actuator.
The conventional actuator consists of the feedforward controller, power
electronics and solid-state transducer subsystems (see Fig. 1.6). By means of
the desired displacement Sct, the feedforward controller consisting of a linear
static transfer characteristic with a constant ks produces an electrical in-
put signal Xi for the power electronics. The power electronics generates the
energy-carrying output quantity X for the solid-state transducer from the
information-carrying electrical input signal Xi. The solid-state transducer
transforms the electrical energy quantity X into a displacement s against
a force F.
sd(t) r··--·-·r,;··-----~ X;(t) r··-----ril·------1 X(t) r···-··m··----1 s(t) ( ~
• '-------~~------__-___+-!--·+-L-___--~__===1-----------+-1--·+-j__-__..,__ ~I--------Ji-:-~-(t_)__,•...,J I~
Feedforward Power Solid-state
controller electronics transducer
Fig. 1.6. Conventional actuator (kv: transfer factor of the power amplifier)
However, even in quasi-static operation the actual displacement and the

desired displacement usually do not correspond. Internal imperfections such
as complex hysteretic nonlinearities (see the characteristic curves in Figs. 5.15
and 5.32) described by the operator IA 3 in Fig. 1.6 and external influences
such as load reactions via the surrounding mechanical structure are the main
reasons for the deviation between the desired and actual values. The for-
mer imperfection provokes ambiguities between the input and output of the
transducer; the latter one causes an additional deviation in the actual dis-
placement from the desired value due to the finite stiffness of the solid-state
transducer.
3 Operators are here used to describe mathematically the mapping between the
input and output time functions of dynamical systems.
1 Introduction 9
1.4.1 "Intelligent" Solid-state Actuator
According to general usage, solid-state actuators are called intelligent when

their transfer characteristic is determined by a functionally allocated and elec-
tronically integrated "intelligence", if necessary, with sensor support. Such
intelligent actuators can recognize deviations from the desired transfer char-
acteristic, which result from the hysteretic nonlinearities as well as from load
feedback, and correct them automatically. The position controlled actuator
in Fig. 1.7a is an example of such an actuator type. With this principle, the
compensation of internal imperfections and external disturbances is achieved
by a linear controller Gc, which receives information about the actuator
output from an external displacement sensor. With the reconstruction of
force Fr by means of the inverse filter JA-1, it is possible in this case to
give feedback about the actuator's current load situation to the superior
control system. An electrical circuit for the measurement of the electrical
quantity X is necessary for the implementation of this additional function.
Such an electrical measurement circuit can be an element of the power elec-
tronics [5].
The actuator concept in Fig. 1. 7b is sometimes used with piezoelectric
transducers. It clearly has a higher measure of integration. In this case, some
of the stack's ceramic disks are used as sensors in order to measure the force,
whereas the major part of the stack operates purely as an actuator. For
><;(t) r·········-..················1 x(t> r··---------------------------i

1----+---+---1 1---+---......~
L.r··--------------------------,
..........................J F(t)
·1
1----+o,__....J
Sm(t)--------------------~~---1-·-·-+·~-m..,...(t-)+-L.-...-..-1
.. ......................! y(t)
~--------·-------------. ------:
Feedback controller Measurement and Solid-state transducer
and reconstruction filter power electronics with separated sensor
a
..
><;(t) r••••••·······•··············-. X(t)..
1----~--~·---1 1-~-+·-~--~
Reconstruction and Measurement and Solid-state transducer

compensation filter power electronics with integrated sensor
b
Fig. 1. 7. Concept of intelligent solid-state actuators. a With separated sensor,
b with integrated sensor (kx, ky: transfer factors of the sensor to measure the
electrical driving quantity X and the dual electrical quantity y)
10 Hartmut Janocha
the accurate measurement of the force, the hysteretic transfer characteristic

of the integrated sensor must be compensated within the electronic signal
processing part by an inverse filter rs-
1 . In this case, the displacement Sr can
be reconstructed with the filter IA from the electrical quantity X and the
measured force F. Hysteretic nonlinearities and mechanical loading resulting
during actuator operation can be compensated by implementing the inverse
filter JA- 1 .
1.4.2 Self-sensing Solid-state Actuator
The self-sensing solid-state actuator shown in Fig. 1.8 has the highest mea-
sure of integration. However, its bidirectional function requires also the most
complex mathematical and electronic signal processing unit. Characteristic
of self-sensing actuators is the simultaneous utilization of both actuator and
sensor properties of the active material. In contrast to the intelligent con-
cepts of Fig. 1.7, they have power electronics which contain the electronic
circuits for measuring the given electrical quantity X and the dual electri-
cal quantity y carrying the sensory information. The central function of the
signal processing unit, which is responsible for the bidirectional function, is
in this case the linearization and decoupling of both sensor and actuator
operation.
Reconstruction and Measurement and Solid-state transducer

compensation filter power electronics with inherent sensor
Fig. 1.8. Concept of self-sensing solid-state actuators (the differences from Fig. 1. 7b
with respect to signal flow are indicated in bold)
In particular, the decoupling of both sensor and actuator operation for

force and displacement reconstruction (Fr and sr) according to Fig. 1.8 is the
main difference from the intelligent actuator concepts depicted in Fig. 1.7. In
the case of self-sensing actuators the output y of the sensory path is strongly
influenced by the driving quantity X of the solid-state transducer and must
be regarded as an external disturbance for the sensor operation. This is shown
in the right-hand block in Fig. 1.8. In intelligent actuators the output y of
the sensor path is not influenced by the driving quantity X of the solid-
state transducer, and a model-based decoupling of the sensor and actuator
operation is not necessary.
1 Introduction 11
Operator-based Modelling for Large-signal Operation
A solid-state actuator can be approximated mathematically by means of

a rate-independent electromechanical two-port system 4 which is coupled with
a dynamic model of the mechanical structure via the force F and displace-
ment s, which are mechanical quantities. In order to produce noteworthy
displacements during actuator operation, the solid-state transducer is driven
with electrical amplitudes X, which can cause more or less strong hysteretic
effects in the electrical or electromagnetical characteristic (y- X relationship)
and the actuator characteristic (s-X relationship). Moreover, during actu-
ator operation, the solid-state transducer is loaded with mechanical forces
F, which can cause hysteretic effects in both the sensor characteristic (y-F
relationship) and the mechanical characteristic (s-F relationship).
As a result, in electrical and mechanical large-signal operation there ex-
ists a rate-independent coupling with memory 5 between the electrical input
quantity X and the mechanical input quantity F which, in principle, re-
quires a mathematical description by means of vectorial hysteresis operators
[6]. This fact can be described by the operator equations
y(t) = Ts[X, F](t) , (1.1)

s(t) = JA[X, F](t) . (1.2)
Equation (1.1) is called the sensor equation and (1.2) the actuator equa-
tion of the solid-state transducer.
Reconstruction and Compensation Filter
The task of the reconstruction filter consists in reconstructing the solid-state

transducer's momentary displacements and load F by means of the electrical
measurement circuit outputs Xm and Ym of the power electronics. Therefore,
they must separate the measurement information which is contained in the
dual electrical signal y from that signal part which is caused by the electrical
signal X. Additionally, they have to eliminate the signal distortion which
results from the hysteretic nonlinearities. The compensation filter is required
to produce a control signal Xi by means of the desired displacement sd,
so that hysteretic nonlinearities occurring in the actuator transfer path as
well as influences on the transducer displacement by mechanical loading are
compensated.
The operator-based sensor model (1.1) and the operator-based actuator
model (1.2) are the starting point for the derivation of corresponding filter
equations. Then, the mechanical load can be calculated from the measured
4 This is a system with an electrical and a mechanical port. The transfer behavior
is single-valued or hysteretic.
5 This kind of coupling is independent of the rate of the input functions but de-
pends on the history of the input functions.
12 Hartmut Janocha
values Xm and Ym by inverting the operator describing the relation between

y and F. The reconstructed transducer displacement Sr is obtained by fitting
the reconstructed force Fr into the operator-based actuator equation (1.2).
The corresponding reconstruction filter equations are
Fr(t) = JS- 1 [ ~:, ~;] (t) , (1.3)
sr(t) = JA [ ~;, Fr] (t) . (1.4)
The filter equation
(1.5)
for the compensator follows from equation (1.2) by inverting the relationship
between s and X.
Example Application: Piezoelectric Positioning System
The self-sensing solid-state actuator concept illustrated in Fig. 1.8 was imple-
mented in a commercially available positioning system driven by a low-voltage
piezoelectric stack transducer (see Fig. 5.28a). Figure 1.9 shows the measure-
ment results obtained with the self-sensing actuator principle for electrical
large-signal operation [5]. It displays the characteristics of the three transfer
paths of the bidirectional actuator, illustrated in the form of s-sd, sr-s and
Fr-F trajectories. In case of a piezoelectric positioning system, the values
Xm and Ym in (1.3) and (1.4) correspond to the measured values of control
voltage and piezoelectric charge, respectively.
In Fig. 1.9a a huge deviation occurs between the load F and the recon-
structed load Fr because of the unconsidered hysteresis effect. The deviation
between the displacement s and the reconstructed displacement Sr on the
one hand, and between the desired displacement sd and the displacement s
on the other, are significant due to the unconsidered hysteresis effect. Us-
ing the operator-based filter, the influence of the hysteresis effect is taken
into account. As Fig. 1.9b displays, the deviation between F and Fr is now
comparatively small. One can hardly recognize the deviations between s and
Sr on the one hand and between Scl and s on the other. The deviation is
one-seventh as great as when using a linear reconstruction and compensation

model.
The self-sensing actuator concept requires the powerful mathematical ma-
chinery of complex hysteresis operators - first for reconstructing the mechan-
ical quantities by means of the measured values of electrical quantities and
second for compensating the hysteretic nonlinearities and the load depen-
dency. Whereas robust software tools exist for modelling, identifying and
1 Introduction 13
8.0 8.0
v
~m ~m
~ /.
4.8 4.8
ts A ~ L.6 ~
1.6 w
~
..<:Iii
s,
~
A~
-1.6 -1.6
I~
....
-4.8 -4.8 v
-8.0 -8.0 -50'----'----'--..__--'---'-----'
-5.3 -2.7 o.o 2.1 ~ ao -5.3 -2.7 o.o 2.1 f.IT1 ao -33 -17 o 17 N 50
a s- F-
8.0 8.0 50
N
/ "
~m ~m
I
4.8
ts
1.6 t
4.8
1.6
v t10
30
~~
/ s, 1/ II
-1.6
I'
/ -1.6
/ F.
':.1o ,d
-4.8
/ -4.8
/ -30 /
,I I
-8.0 -8.0 -50
-5.3 -2.7 o.o 2.1 ~ ao -5.3 -2.7 o.o 2.1 ~ ao -33 -17 o 17 N 50
b s- F-
Fig. 1.9. Function of the self-sensing actuator concept according to Fig. 1.8 for
electrical large-signal operation. a Linear reconstruction and compensation filter,
b operator-based reconstruction and compensation filter
compensating scalar complex hysteretic nonlinearities in practical applica-

tions, a considerable amount of research is necessary in the field of vecto-
rial hysteresis phenomena to obtain a similar status. Furthermore, attempts
are being made to map the computation intensive algorithms of reconstruc-
tion and compensation filters in FPGAs (field programmable gate arrays)
or ASICs (application-specific integrated circuits), in order that self-sensing
solid-state actuators will become available for highly dynamic applications.
1.5 Design of Actuators
As in most technical fields, actuators are increasingly designed with the help
of computers. The actuator and its surrounding mechanical structure are
simulated as a mathematical model by means of commercially available soft-
ware. Such models are fundamental for the simulation of the system response
characteristic in each specific case. In this way, it is possible to find out about
all the important properties of the system even before the actuator is built,
and the actuator's relevant parameters can be optimized to achieve the de-
sired values. This designing strategy is exemplified below with an auxiliary
14 Hartmut Janocha
mass damper which is able to withdraw kinetic energy from a host vibrating
system.
Such vibration absorbers are used for instance in the automotive and
aerospace industries where the vibration of car bodies or fuselages has to be
attenuated. Within the scope of a first rough model the mechanical structure
at the place of maximal vibration is described by the effective base mass m 1
which is excited by an unknown disturbing force F 1 causing undesirable vi-
brations (see Fig. 1.10). F 1 is thus a consequence of the interaction between
m 1 and the remainder of the mechanical structure which is excited by ex-
ternally or internally acting forces at other points. The task of the vibration
absorber is to displace the auxiliary mass m 2 in such a way as to generate
a secondary force Fz = m2 · a 2 that will compensate the primary force F 1
and thus counteract the excitation of mass m 1 [7].
Fig. 1.10. Vibration attenuation using a passive vibration absorber
When the force F 1 is narrow band, attenuation can be achieved with

a passive vibration absorber which has to be tuned to the disturbance fre-
quency through its parameters stiffness c, damping constant k and mass m 2 .
In contrast an attenuation of broadband disturbances requires the use of an
active vibration absorber whose mass is coupled to the vibrating main sys-
tem via an electrically controllable interface. From the formulation of the
dynamic balance of forces for the mass m 1 it follows that the acceleration a 1
of m 1 is a measure of the compensation effect of the active vibration absorber.
Thus the aim of this damper principle is to displace the mass m 2 through
appropriate feedback of the acceleration a 1 in such a way that the resulting
force F 2 will compensate the disturbing force F 1 and thus nullify the base
acceleration a 1 .
The starting point of the following specific example is a vibrating struc-
ture being stimulated to vibrate by imbalances within the rotating parts. Up
to now the vibration was dampened by a passive vibration absorber whose
resonance frequency is tuned to the fundamental frequency of the vibration
at 100Hz. The disturbing force F 1 affecting the passive vibration absorber
shows in addition to the 30 N value at 100Hz other noteworthy values of 20 N
and 10 N lying at 200Hz and 300Hz that cannot be compensated due to the
1 Introduction 15
narrow-band damping characteristic of the passive vibration absorber. Now

this task will be undertaken by an active piezoelectric vibration absorber.
The principal structure of the active vibration absorber corresponds ap-
proximately to the structure of the passive vibration absorber shown in
Fig. 1.10 whereby the passive elastic material between m 1 and m2 has been
replaced by a piezoelectric actuator and a displacement amplification system
to increase the achievable displacement of m 2 . The amplification system is
given in this example by an elastic suspension, similar to those illustrated in
Fig. 5.22c.
The mathematical model of the mechanical actuator system can be devel-
oped directly from the CAD design drawing by means of commercial FEM
software tools, e.g. ANSYS® [8]. This model is fundamental for the calcula-
tional modal analysis which serves to find the system's natural frequencies.
Figure 1.11 shows the FEM model of the active piezoabsorber and gives an
impression of the third vibration mode of the structure which is used in this
example for the vibration absorption.
Fig. 1.11. Third vibration mode of the absorber structure analyzed using ANSYS®
The active compensation of the disturbing force F 1 can now be achieved

by a suitable feedback of the measured base acceleration a 1 to the input of
the high-voltage source for the piezoactuator. The corresponding acceleration
feedback loop is represented in Fig. 1.12 with Gs as the sensor transfer func-
tion, Gc as the controller transfer function , Gv as the high-voltage source
transfer function, Gu as the guiding transfer function, and GF as the distur-
bance transfer function , respectively [7]. Based on this signal flow diagram,
which is always to be developed by the designer, the functionality of the
force compensation will be investigated on a computer with support of an
appropriate dynamic simulation and analysis software system, for example
MATLAB® [9].
Figure 1.13 illustrates several results of this simulation. The frequency
response in Fig. 1.13a shows the band rejection filter characteristic required
for the compensation of the force F 1 lying between about 70Hz and 329Hz. In
16 Hartmut Janocha
Gs ~------------------~
Fig. 1.12. Force compensation within a closed control loop (a1, a1rn: actual and
measured acceleration of the mass m1; F1: disturbing force; VR: control signal of
the high-voltage source; v: driving voltage of the piezoactuator; indices u, d: undis-
turbed and disturbed signal)
Fig. 1.13b the effect of the closed-loop force compensation is illustrated within
the time domain, over the time interval of 0-0.4 s. During the interval 0-0.1 s
the controller is idle, so that the vibration absorber operates passively. The
maximum amplitude of the acceleration a 1 emerging due to the excitation by
F 1 amounts in this operating state to about 5 m/ s 2 . The controller is switched
on at t = 0.1 s which excites the dynamics of the whole system.
This is indicated by a rapidly decaying high-frequency vibration corre-
sponding to the second peak in the amplitude response shown in Fig. 1.13a.
The high-frequency vibration is superimposed on a slower decaying low-
frequency vibration corresponding to the first peak in the amplitude response.
After the decay of all transient processes only the acceleration emerging due
-5.5 '---"---"---"---.J....__.J....___J
10' 102 10' 10' 10'1/s 10' -0.04 0.04 0.12 0.20 0.28 0.36 s 0.44
a w- b t-
Fig. 1.13. Active vibration absorber. a Amplitude and phase response, b compen-
sation effect within the time domain (GFc: disturbance frequency response of closed
loop)
1 Introduction 17
to the continued disturbance F 1 is still visible. The maximum amplitude of

the acceleration a 1 at steady state is approximately 0.25 m/ s 2 . Thus the force
affecting the base mass m 1 can be reduced by a factor of 20.
This analysis software naturally also allows the user, for instance, to test
and optimize the stability of the vibration absorber to avoid unpleasant sur-
prises after the prototype has been built.
References
1. DIN 19226, Regelungstechnik und Steuerungstechnik, Teil 4 (Begriffe fiir

Regelungs- und Steuerungssysteme), Beuth Verlag Berlin 1994
2. Janocha, H. (Ed.): Adaptronics and Smart Structures. Springer-Verlag Berlin
Heidelberg New York 1999
3. Neumann, D.: Adaptronics- a Concept for the Development of Adaptive and
Multifunctional Structures. In: Janocha, H. (Editor): Adaptronics and Smart
Structures, pp 5-12, Springer-Verlag, Berlin Heidelberg New York 1999
4. Schwinn, A.: Aktive Vibrationsdiimpfung mit verteilten Aktoren. PhD Thesis,
Saarland University, Germany, 2004 (83 references)
5. Kuhnen, K.; Janocha, H.: Self-Sensing Solid-State Actuators. Encyclopedia of
Life Support Systems. Accepted for publication.
6. Kuhnen, K.: Inverse Steuerung piezoelektrischer Aktoren mit Hysterese-,
Kriech- und Superpositionsoperatoren. PhD Thesis, Saarland University, Ger-
many, 2001 (60 references)
7. Kuhnen, K.; Janocha, H. (2002), Smart piezoelectric dynamic vibration ab-
sorber. Proc. 8th Mechatronics Forum Int. Conf. Mechatronics 2002, Twente
Enschede, Netherlands, pp 1048-1057
8. Product information ANSYS Inc., ANSYS 6.1 Full Set of Analysis Guides.
Canonsburg, PA, 2002, www.ansys.com
9. Product information The Math Works, Inc.: User's Guide, Natick, MA, 1988-
1998, www.mathworks.com
2 Power Circuits
Stefan Butzmann
2.1 Introduction
This chapter deals with power circuits and their use for actuator control.
Figure 2.1 shows the general structure of such a power circuit, where one or
several power devices are turned on and off by a driving stage. Protection
circuits are used to prevent the power devices from failure due to overload
conditions.
Single I
Integrated multiple
From driver ,.. To actuator
microcontro I~ power devices
Protection
Fig. 2.1. Structure of a power circuit
As most actuators are of inductive type - more than half of our electrical
energy is consumed only by electric motors in household and factory appli-
cations - the focus of this chapter is oriented on topologies, where the power
device is used as a switch. While linear drive circuits for inductive loads al-
ways result in a significant power dissipation in the power device, switching
topologies have the advantage, that the power dissipation in the drive circuit
can be minimized.
In the following, the most important power device types- the rectifier, the
bipolar transistor, the MOSFET and the IGBT- are discussed. As the plu-
rality of actuators is operated in low to medium power and voltage ranges,
where power-MOSFETs or IGBTs are generally the best choice for power
switches, different circuits for gate drive and protection of these devices are
described. An important aspect is thermal management, which is introduced
in the following section. Different topologies for DC and AC actuator control
are described next. The section "future trends" deals with coming technolo-
gies, gives an insight into intelligent switching devices and an outlook into
future developments.

20 Stefan Butzmann
2.2 Power Switching Devices

2.2.1 Overview
Advances in power circuits are strongly related to the continuous improve-

ment of power semiconductor devices. They are used to control the energy
flow from an electrical source to the load. The ideal power switch is charac-
terized by the following attributes:
- zero resistance during on-state,
- infinite resistance during off-state,
- zero transient time from on- to off-state and vice versa,
- zero driving power during static and transient operation,
- electrical isolation between control function and power flow.
Practical semiconductor switches are imperfect. They have a limited volt-
age and current handling capability, an on-resistance greater than zero and
finite switching times.
<(
"E
~
::J
()
Thyristor
1000 "F=---:----:--,....,..---..
Bipolar transistor
IGBT
100 1 = - - - - - - - , .
10 MOSFET
1
1 10 100 1000 10000 Voltage I V
Fig. 2.2. Application areas of power semiconductors
The first power devices broadly used in the 1950s were thyristors, which
today are able to switch voltages of more than lOkV and currents of more
than 10 kA. The lack of controlled turn-off-capability however makes thyris-
tors unsuitable for many applications where the turn-off of the power device
also needs to be controlled. Thyristor concepts providing this capability like
the gate-turn-off thyristor (GTO) or the MOB-controlled thyristor (MCT)
are generally only used in applications with power ratings in the megawatt
range. Even though thyristors are very cost-efficient devices, they are rarely
used for the drive of actuators as discussed in this book. For this reason, the
working principle of thyristors will only be briefly explained.
2 Power Circuits 21
A power device, which was used extensively in power applications up

to the 1980s, is the bipolar transistor. One of the main disadvantages of
bipolar power transistors is the relatively large base current needed to con-
trol the device and the long turn-off time limiting the maximum switch-
ing frequency to approximately 5-10kHz. In general, bipolar transistors are
only used in applications of medium frequencies and high voltages, where
they provide a cheaper solution than power-MOSFETs or IGBTs. However,
their more complex driving circuits in comparison to IGBTs has signifi-
cantly reduced their use in many applications. Today, typical remaining ap-
plications for bipolar transistors are electronic lamps or automotive ignition
circuits.
In contrast to bipolar transistors power-MOSFETs possess a high input
impedance and exhibit low power losses. Additionally, they allow for high
switching frequencies up into the MHz range and are still today the best
choice in low-voltage applications. Due to the increase of their on-state losses
towards higher operating voltages, their performance becomes unsatisfactory
in applications with supply voltages above 600 V.
The Insulated Gate Bipolar Transistor (IGBT) aims at the voltage range
above 600 V, which is no longer suitable for MOSFETs. It can be operated
with switching frequencies of up to 20kHz and currents up into the kA range.
IGBTs are frequently used in motor control as well as in advanced automotive
ignition, competing here with bipolar transistors. Their low-effort driving
circuit has led to an extensive use of IGBTs in many applications, even though
the device itself is more expensive than a bipolar transistor.
Table 2.1. Advantages and disadvantages of power switches
Advantages Disadvantages
BJT low on-resistance driving power
small area slow turn-off
inexpensive
MOSFET low drive effort high on-resistance at high
fast switching breakdown voltages
IGBT low on-resistance at higher medium switching speed
breakdown voltages low drive effort
Thyristor extremely high power no turn-off control via gate
capability
Power-MOSFETs and IGBTs are the most popular power switches used
for actuator control at the current time. As the drive of MOS-gated devices
is simpler than that of bipolar transistors because of the lack of base current,
complete gate drive circuits can meanwhile be integrated monolithically on
a single chip. The addition of protection functions and intelligence has pro-
22 Stefan Butzmann
vided a further step towards smart-power devices, which are discussed in the
last section of this chapter.
2.2.2 Power Diode
Most power applications require fast diodes as free-wheeling elements. The

V-I characteristic of the diode is shown in Fig. 2.3.
For a positive voltage from the anode to the cathode the diode is forward
biased and the current through the diode rises exponentially with the voltage
VF. In reverse bias (negative voltage from anode to cathode) only the leakage
current Is flows through the diode as long as the reverse voltage is smaller
than the breakthrough voltage V8 v. If the reverse voltage becomes larger
than the breakthrough voltage, the reverse current increases rapidly due to
the avalanche effect.
The breakdown voltage of the diode Vsv is usually specified for an am-
bient temperature of 25 °C. V8 v is a function of temperature with a positive
temperature coefficient. Also the leakage current Is increases exponentially
with rising temperature.
Of particular interest for power applications is the switching behavior of
power diodes, which is significantly different from that of small signal diodes.
Figure 2.4b shows the principal turn-on behavior of a power diode in the
configuration according to the circuit in Fig. 2.4a. When a constant current
is impressed into the diode, the device shows quasi-inductive behavior until
the pn-junction has been flooded with carriers (Fig. 2.4b). This yields a peak
voltage during turn-on, which is dependent on the current slope.
The turn-off behavior of the diode is shown in Fig. 2.5. Power pn-diodes
show a "memory" effect. If the voltage across a conducting diode is suddenly
reversed, the diode acts like a capacitor at first and a current back into the
voltage source builds up. Only after the injected minority carriers (holes in
then-region) have been swept out by the reverse current or by recombination,
the diode regains its blocking capability.
'o
Vsv
b
Fig. 2.3. Power diode. a Schematic symbol, b transfer characteristic
2 Power Circuits 23
I~
.t
tvo v~ Vp
....
a b t
Fig. 2.4. Thrn-on behavior of power diode. a Test circuit, b voltage output for
current step
The main parameter during turn-off is the peak reverse current JRR· For
a greater slope ldJ I dtl of the current during turn-off this reverse current in-
creases and causes current spikes, electromagnetic interference and additional
switching losses. It is therefore an important task in power applications to
limit dJ I dt during turn-off to keep the reverse current peak low.
The reverse recovery time trr describes the time between the point where
the turn-off current crosses zero and the point where the reverse current has
fallen back to 25% of the peak value. The reverse recovery behavior of the
diode is generally described by the softness factor S = tblta. If S « 1 a hard
snap-off is assumed.
In comparison to the described pn-diode a Schottky diode has a faster
switch-off behavior. Instead of an implanted p-region Schottky diodes are re-
alized by the junction between the barrier metal and the semiconductor. Since
both metal and silicon are n-type, the conduction occurs through majority
carriers only. Because of the lack of minority carriers injection, storage and
Fig. 2.5. Turn-off behavior of power diode

24 Stefan Butzmann
recombination, the storage time ts is virtually reduced to zero. This makes

the Schottky diode very suitable for fast switching applications.
The disadvantages of Schottky diodes are the relatively large leakage cur-
rent and the lower breakdown voltage in comparison to pn-diodes. Schottky
diodes are available with voltage ratings up to typically 150-200 V.
For fast-switching applications with breakdown voltages of more than
200V FRED diodes (fast recovery epitaxial diodes) are a good compromise.
These diodes contain gold or platinum atoms diffused into the epi-layer cre-
ating areas where the excess holes can recombine with electrons. This reduces
the minority carrier lifetime and improves the dynamic characteristics of the
diode. Alternatively, the minority carrier lifetime can also be reduced by ion
irradiation.
Table 2.2 summarizes the characteristics of the most important types of
power diodes. The selected examples of the rectifier diode 1N3768, of the
FRED diode 30EPH06 and of the Schottky diode 30CTQ100 (all Interna-
tional Rectifier) show that the forward voltage drop at a nominal current of
IN = 30 A is approximately 0.9 V for the Schottky diode, while the voltage
drop at the line-rectifier diode and at the FRED diode is more than a factor
of two higher. This results in lower static power dissipation of the Schottky
diode in comparison to the two other diode types.
Table 2.2. Comparison of the most important types of power diodes
Rectifier diode Fast recovery Schottky diode

diode (FRED)
Current range [A] 3000 500 100
Reverse voltage [V] 6000 2000 200
Switching losses medium small extremely small
Example 1N3768 30EPH06 30CTQ100

Rating lkV/35A 600V/30A 100V/30A
VF[V]@JN 1.8 2 0.86
2.2.3 Bipolar Power Transistor
Bipolar power transistors offer high breakdown voltages of up to 1500 V

and low saturation voltages VcE,sat· In contrast to MOSFETs, which exhibit
an extremely high gate-source impedance, power bipolar transistors require
a base current. The power control ratio of a bipolar transistor is described
by the current gain B = Ic/IB.
In order to achieve high voltage ratings as needed for power applications,
large base widths have to be realized, resulting in larger saturation voltages
and poor current gains B. Figure 2.6 shows the current gain of the bipolar
2 Power Circuits 25
--- r-- --....

!'-.....____
""' ~~·c
10 1
'\
19J=125°C
\
VcE=5V
Fig. 2.6. Current gain of bipolar power transistor BU208A as a function of collector
current [11]
transistor BU208A with a voltage rating of VcE,rnax = 1500 V and a maximum

collector current of Ic,max = 8 A. As shown in the diagram the current gain
is a function of the collector current and decreases below a value of ten for
the nominal current.
Bipolar transistors have no overcurrent capability, i.e. the collector cur-
rent Ic always has to stay below the specified value Ic,max· The maximum
collector--emitter voltage of a bipolar transistor is defined by VcEo and VcBo.
VcEo describes the collector-emitter breakthrough voltage with the base
open, VcBo the collector-base breakthrough voltage. A good approximation
for the relationship between VcEo and VcBo is given by:
VcBo
VcEo ;:::j VB. (2.1)
Figure 2.7a shows a bipolar power transistor used as a switch. Usually

the input base current JB will be chosen higher than necessary in order
to ensure saturation of the transistor. In consequence the collector-emitter
voltage will be reduced to VcE,sat, thus resulting in the power consumption
P =VcE,sat · Ic+VBE · IB of the device.
During the saturation mode with a forward biased base-collector diode
there will be an excess of charge in the base, which has to be extracted during
switch-off. If the base is open-circuited by switching off the base current JB,
the slow recombination of the excess carriers in the base will result in a long
26 Stefan Butzmann
15,-------~------~====~
---- 18 / A
_lc/A
······························-···
---------i
01-------!·······························~-----------
- 5 oL----~10---~2~0---~30
t! us
a
15
---- V IV 1
- Icc' A }_ ___________ ..
10
5 ,!
..........................; ..
:
01------f······························ !1--------l
i
:
I
'
-5 .................. ··········1···························--1···
:I !;
vc! -10 ----------!--- . ····---~----------
-150 10 20 30
t! us
b
Fig. 2. 7. Power bipolar transistor as switch. a Control via base current, b control
via base-emitter voltage
switch-off time toff· In order to reduce the switch-off time the base may be
returned to a negative potential via a low-valued resistor RB (Fig. 2.7b).
The slow recovery characteristic of the bipolar transistor is especially
critical when inductive loads are turned off. As the turn-off does not occur
equally distributed over the chip area, parts of the transistor continue to
carry the complete load current at the full operating voltage before the free-
wheeling diode is activated. This may result in "hot spots" and failure of the
transistor. This effect is known as "secondary breakdown" .
The turn-off time may also be reduced by prevention of saturation of the
device. A suitable anti-saturation circuit is the Baker clamp, which is shown
in Fig. 2.8. In this circuit, the diode D 1 prevents the collector potential from
falling below the base potential, thus avoiding saturation of the transistor.
Clamping the collector-emitter voltage in the described way, however, yields
an increase of VeE and thus in a higher power dissipation during the on-
state.
2 Power Circuits 27
Fig. 2.8. Baker clamp circuit
The largest disadvantage of bipolar power transistors in DC operation

mode is the low current gain. A load current of Ic = 10 A requires a base
current of IB = 1 A if a current gain of B = 10 is assumed. In order to
minimize the driving current of a power bipolar transistor, the Darlington
stage can be used, where two transistors are connected as shown in Fig. 2.9.
The current gain of the resulting Darlington transistor is the product of
the current gains of the two individual transistors. These advantages, how-
ever, are offset by the higher saturation voltage from the collector to the
emitter, which has been increased by VBE:
VcE,sat,Darl = VcE,sat,Tl + VBE,T2· (2.2)
Apart from this, special measures have to be taken to optimize the switch-
ing behavior. For this reason, power bipolar transistors have enormously lost
importance in favor of MOSFETs and IGBTs.
c·
B'
Fig. 2.9. Darlington stage

28 Stefan Butzmann
2.2.4 Power-MOSFET
Static Behavior
MOSFETs (Metal Oxide Semiconductor Field Effect Transistors) have been

in use for power applications since the early 1980s. The popularity of MOS-
FETs for power applications results from two main advantages over bipolar
transistors:
- MOSFET transistors are simpler to drive as their control electrode is
isolated from the current path. Thus, no gate current is required during
static operation.
- MOSFETs have a resistive nature , i.e. the voltage from drain to source
is proportional to the current flowing through the device. In contrast
to bipolar transistors where the saturation voltage VcE,sat has a negative
temperature coefficient, the on-resistance of MOSFETs exhibits a positive
temperature coefficient of ~ 1 %/K. This behavior is a great advantage
for parallel operation of multiple MOSFETs.
0
Fig. 2.10. Cross-section of power-MOSFET
For power applications the most popular MOSFETs are enhancement

types, which are in the off-state for gate- source voltage VGs of zero and in
the on-state for VGs > V'th (n-channel) or VGs < V'th (p-channel). Due to
the greater mobility of electrons over holes, n-channel types offer a lower
on-resistance per transistor area and are therefore more popular in power
switching applications.
The cross-section of a vertical power-MOSFET is shown in Fig. 2.10. The
actual MOS transistor is located at the surface near the gate and the source
contacts. This construction has proven to be suited for high voltages due to
the great distance of the gate and the drain contact. As a further benefit,
the power is dissipated over a larger volume than is the case with lateral
MOS transistors , which makes these transistors well suited for high-power
applications.
2 Power Circuits 29
150r-------~----r---.----.---,r---.----.
100
<(
~50 VGS
10----1~5----2~
0 ----~5~--~0----5~--~
-~q~ 2~5--~30
0--~
V05 / V
Fig. 2.11. Output characteristic of power-MOSFET
Figure 2.11 shows the output characteristic of a power-MOSFET. As long

as the gate- source voltage is zero, only a small leakage current flows, when
a positive drain- source voltage is applied. For a negative drain- source volt age
the intrinsic reverse diode conducts, so a reverse current will flow even for
Vcs = 0. W hen the gate-source voltage is increased, the transistor shows
a quasi-linear relationship between the drain current / 0 and the drain-source
voltage V08 as long as the MOSFET remains in the ohmic region.
The parasitic bipola r transistor and its base-collector capacitance made
early MOSFETs susceptible to turn-on for large gradients of the drain- source
voltage Vos- The source-metallization prohibits t his parasitic turn-on by
Fig. 2.12. Parasitic components in power-MOSFET

30 Stefan Butzmann
50V
·····························,----'~'--..., J:::::=5=0=0=V=::r 1OO%
Packaging
+ ......···..~--- - - - - - 1
M~=:='"t-----1"~ .8
c
0
:g
c
r:r.o
.0
............................. t - - - - - - 1 ·;::
c0
JFET
region
u
+
Expitaxial
layer
·····························t-----+
····-~-~~~~.r:~.~~........______.·············.t::===::i.o
Fig. 2.13. Relative contribution of different components to on-resistance of power-
MOSFETs with voltage ratings of 50 V and 500 V
shorting the base and the emitter, thus creating a pn-diode from the source
to the drain (see Fig. 2.12). This diode shows a long reverse recovery time
and its use is suitable only in low-frequency applications.
In switching applications the on-resistance of the power-MOSFET is one
of the most important parameters as it determines the power dissipation for
a given current. It is mainly contributed to by the following components:
- the resistance of the channel,
- the resistance of the parasitic junction-FET,
- the resistance of the drift region,
- the resistance of the substrate,
- the resistance of the bond wires and the metallization,
- the resistance of the source region.
Figure 2.13 shows the contribution of the different parasitic components
to the overall on-resistance for power-MOSFETs with voltage ratings of 50V
and of 500 V. At high voltages approximately 95% of the on-resistance RDs,on
result from the n- -epitaxial resistance, which is a function of the transistor-
breakdown voltage. For a chip-area of Achip = 1 cm 2 it is given by
Repi = 8 3. 10-9.
0 °
(UBR)
v
2.3
(2.3)
Therefore, the on-resistance of a MOSFET is dominated by the epi-

resistance particularly at high breakdown voltages. At breakdown voltages
of more than 600V the on-state voltage drop of usual power-MOSFETs is
higher than that of bipolar transistors of the same size. This makes the use of
IGBTs (Insulated Gate Bipolar Transistors) more attractive in high-voltage
applications.
2 Power Circuits 31
Dynamic Characteristics
The dynamic characteristics of a power-MOSFET are mainly determined by

the parasitic capacitances, which are the gate-source, the gate-drain and the
drain-source capacitance (see Fig. 2.14).
The gate-source capacitance Ccs consists of three components. These are
the overlap of the gate electrode with the p-channel region, the overlap with
the n- -source region and the capacitance resulting from the proximity of
the gate electrode to the source metallization. The gate-source capacitance
is widely constant over the operating range of the MOSFET. It has to be
charged to the threshold voltage before the drain current starts to build
up.
The Ccn capacitor is formed by the overlap capacitance of the gate elec-
trode with the n-drift drain region, the Cns capacitor by the junction capac-
itance of the body diode. Both capacitances are nonlinear and functions of
the drain-source voltage.
In datasheets, usually the values Ciss, Coss and Crss are found (see
Fig. 2.15) with the following relations to the depicted capacitances:
Ciss = Ccs + Ccn (2.4)

Crss = Ccn (2.5)
Coss = Cns +Gen. (2.6)
These capacitances have a strong influence on the switching behavior of

the MOSFET. Figure 2.16 shows the turn-on behavior of a power-MOSFET.
At first the input capacitance Ciss is charged and the voltage Vcs rises until
the threshold voltage of the MOSFET is reached. The MOSFET then draws
increasing load current yielding a decrease of the drain-source voltage Vns.
As the drain voltage is coupled with the gate voltage via Ccn, Vcs is kept
on a constant level while Vns is changing (Miller plateau). When V DS has
reached its final value Vcs rises to Vin·
A simple approach to compare the dynamic characteristics of power-
MOSFETs is the use of the total gate charge Qc, which represents the
Fig. 2.14. Parasitic capacitances in power-MOSFET

32 Stefan Butzmann
2400
C/pF
I I I
1800 f = 1 MHz
VGS =0v
1200
"--- C4ss
600 1\
,"- r--- Coss
'- Crss
0
0 10 20 30
Fig. 2.15. Ciss, Coss and Crss of MOSFET IRF640 [12]
current-time product required to turn the MOSFET fully on. The datasheet
of the IRF640 by ST, for example, gives a total gate charge of Qc = 55 nC
for a gate-source voltage of Vcs = 10 V. Additionally, the gate-source charge
(Qcs = lOnC) and the gate-drain charge (Qcn = 21nC) are specified. The
gate-source charge specifies the amount of charge necessary to bring the gate
voltage from zero up to the threshold voltage vth, the gate-drain charge de-
fines the amount of charge needed to overcome the Miller effect, while the
drain voltage falls.
A proper figure of merit to describe the static and dynamic behavior
of MOSFETs is the product of the MOSFET's on-resistance and the gate
charge.
Figure of Merit= Rns,on · Qc. (2.7)
60,-----~~----.=======~
-----------·---· . \ lf-Vas/V
y---V /V 08
40 .............................. , ......... ··········.
I
I
~ ~
~ ~
.............................. ~ ............. .li ....... ............................ .
I
I
I
I
; :
0 .................. :................~.- --~·"·;"·"-·"·"·~·"·"·"----
0.5 1.5
t/ us
Fig. 2.16. Turn-on behavior of MOSFET

2 Power Circuits 33
2.2.5 IGBT
Up to a drain- source voltage of Vos :::; 600 V MOSFETs are in most cases
the best choice for power switching devices. For higher voltages , however, the
bipolar transistor has a lower on-voltage and it is even less expensive. As
described before, it cannot be driven powerless in static operation due to the
limited current gain though.
A power device combining the high-voltage capability of the bipolar tran-
sistor with the input stage of the MOSFET is the Insulated Gate Bipolar
Transistor (IGBT). IGBTs are generally used in applications where the on-
resistance of MOSFETs becomes impractically high. The main difference be-
tween a MOSFET and an IGBT is the additional p-layer at the drain contact
which injects minority carriers into the n-zone and thus improves the con-
ductivity of the device. Figure 2.17a shows the non-punch-through (NPT)
structure of an IGBT, which consists of an n+pn - p+ -sandwich. With this
structure, the breakdown voltage of the device is determined by the doping
and the thickness of then- -layer, where the electric field can spread.
When the device blocks, the upper pn-transition is exposed to the max-
imum critical field strength. The area under the E-curve is proportional to
the collector- emitter voltage of the IGBT, while the slope is proportional
to the doping of the n- -layer. An alternative structure, the punch-through
(PT) IGBT employs an additional n+ -layer between then- and the p+ -layer
in order to increase the blocking voltage. This allows a thinner n - -layer and
reduces the on-state losses of the device at the expense of higher transient
losses.
This can be seen from Fig. 2.18, which shows the trade-off between the
on-state voltage at current densities of 100 A/ cm 2 and the turn-off-energy for
both punch-through and non-punch-through IGBTs with voltage ratings of
1200V.
i£1
c c
a NPT-IGBT b PT-IGBT
Fig. 2.17. Cross-section of IGBTs. a Non-punch-through, b punch-through

34 Stefan Butzmann
15 2 25 3
On-stage voltage IV
Fig. 2.18. Turn-off losses vs. on-state voltages at 100 A/cm 2 for 1.2 kV PT and
NPT-IGBTs [15]
Additionally, the turn-off losses of PT-IGBTs increase at a higher rate

than those of NPT-IGBTs, when the temperature increases. This behavior
can yield thermal instabilities for PT-IGBTs at high switching frequencies,
where higher switching losses result in increasing temperatures and vice versa.
In the worst case, the resulting thermal runaway can destroy the device.
NPT-IGBTs in comparison do not show this behavior and generally remain
thermally stable.
Figure 2.19 shows the equivalent circuit and the output characteristic of
a power-IGBT. The inherent pnp-transistor of the IGBT increases the on-
state voltage of the power device by the forward voltage of its base--emitter
junction.
The collector-emitter voltage of the IGBT is widely constant for changing
load currents. Thus, at high current levels the conduction losses are lower in
comparison to MOSFETs and at low current levels they are higher.
The saturation voltage VcE,sat of the IGBT is a function of the collector
current Ic, the gate-emitter voltage VeE and the junction temperature TJ.
As the IGBT is generally used as a switch, VcE,sat determines the static power
dissipation. It is contributed to by the on-resistance of the MOSFET stage
and the base-emitter voltage of the pnp-transistor. VcE,sat is temperature-
sensitive and can have both a positive or a negative temperature coefficient in
the area of operation depending on the doping of the device. In order to enable
parallel operation of several IGBTs, IGBTs are usually manufactured with
a slightly positive temperature coefficient of VcE,sat. The reverse blocking
voltage of IGBTs is determined by the breakdown voltage of the pn-junction
on the backside of the wafer and is typically in the order of 20 V.
2 Power Circuits 35
G~
E
- 200 oL----2~--~4----~6----~B~--~10
VcEIV ---+
Fig. 2.19. Output characteristic of an IGBT
While the turn-on behavior of IGBTs is similar to that of MOSFETs,

the turn-off behavior is determined by the tail current, which is due to the
storage of minority carriers in the epi-layer (see Fig. 2.20). As the IGBT's
MOSFET stage is turned off quickly, the turn-off behavior of the bipolar
stage is similar to that shown in Fig. 2.7a for the bipolar transistor as the
minority carriers in the pnp-base cannot be removed externally. The resulting
tail current continues until all minority carriers have recombined. This tail
current has a significant influence on the devices's power dissipation as it
increases the turn-off-time of the device.
50 ........ ;........... ;........... ;.......... .

:. : : '-----':----'
. .. ..
40 .......... : ...........;.......... , ........... ;........... ;..........., ..........., ......... .
~ ~ ~ ~ ~ ~ ~
30 ··········1···········~·····
. . ·····~···········~···········~···········~···········~··········
. . . . .
.: :. :. :. :. :. :.
20 ··········~···········~······ ····~···········~···········~···········'···········~··········
---~----~ ~ ~ Tail~current :
10 ··········!···········!\·.;.;··· .; ........... : ..........;...........;...........[......... .
:
:
:
:
'' .
:
:
:
:
:
:
:
:
:
0 ........ ) ......... ) ........... ~ ........... ~ ........... : : :
1 ~ \~ ~ ~
- 10 ··········:···········:··········~r-·---:-·--r---·:·--·-i-·--·-
-20 o~-0~.5~-~--'1.~5-~2-~2.~5-~3-~3.~5-~4.
tl ~s
Fig. 2.20. Turn-off behavior of IGBT

36 Stefan Butzmann
2.2.6 Thyristor
The thyristor combines high off-state voltages with high on-state currents
and has therefore been the preferred device in high-power applications for
a long time. Figure 2.21 shows a thyristor and its equivalent circuit.
The thyristor has a pnpn-structure with three terminals. During the off-
state both the npn- and the pup-transistor are nonconductive, and the leakage
current through the transistors and their current gain is so low that the device
remains off even for a positive voltage from the anode to the cathode. When
a proper signal is applied at the gate, the npn-transistor starts conducting and
biases the pup-transistor into conduction. The positive feedback structured
device latches and remains conducting until the anode-to-cathode-current is
reduced below the hold-current necessary to keep the device in the on-state.
The necessary gate current to fire the thyristor is dependent on the anode-
cathode voltage. If this voltage is raised above the breakdown voltage Vsn, the
thyristor will fire by itself. Furthermore, the gate current required to fire the
thyristor is also a function of the temperature. With increasing temperature
thyristors become more sensitive. Undesired firing of the thyristor can also
occur due to high voltage slopes dV/dt at the thyristor terminals, which result
in displacement currents through the internal junctions.
While the "vertical" triggering action in the thyristor occurs at high
speeds, the "horizontal" distribution of charge is only relatively slow. Par-
ticularly for large area elements, there is a danger of high current concen-
tration in the gate region during turn-on, which can cause local overheat-
ing and destroy the power device. For this reason, the current slope dJ /dt
during turn-on has to be limited. This is usually done by external snubber
circuits.
The turn-off of a thyristor is generally initiated by a negative voltage
from the anode to the cathode, as it occurs for example with a sinusoidal
supply voltage. Similar to power diodes, the current flows in the negative
direction during turn-off until recombination has been completed. Only after
,node
A
ga
G cathode
VGl lvs
c
a b c
Fig. 2.21. Thyristor. a Principle, b equivalent circuit, c symbol

2 Power Circuits 37
a minimum off-time may a positive anode-cathode voltage be applied again

without unintentional retriggering of the thyristor.
The lack of controlled turn-off-capability is a tremendous drawback for the
use of thyristors in actuator drive circuits. in particular, the development of
IGBTs, which achieve voltage ratings in the kV range, has provided enormous
competition for thyristors in the field of actuator drives.
Modified thyristors offering turn-off capability are GTOs and MCTs. As
these devices are generally used in high-power applications only, they will not
be discussed here though.
2.3 Drive Circuits for MOS-gated Devices
2.3.1 General Remarks
The ideal drive circuit for a power switch is able to charge or discharge the
input capacitance of a MOSFET extremely fast and has a low source resis-
tance. As MOSFETs themselves only sink or source current during switch-on
or -off, the drive circuit itself should need only low supply current, otherwise
the efficiency of the system would be degraded. Drive circuits can be classi-
fied into the following groups according to the position of the switch relative
to the load (see Fig. 2.22):
low-side drivers, where the load is connected to the supply voltage VB and
the switch connects the load to ground,
- high-side drivers, where the load is connected to ground and the power
switch connects the load to the supply voltage.
L D
D
fr~m ______ _j ~ PT
1
1
dovec
a b
Fig. 2.22. Transistor as low- and high-side switch. a Low-side switch, b high-side
switch
38 Stefan Butzmann
Isolated gate drives are used to provide high potential differences and
swings between the control circuit and the power switch. If the power switch
is used with an isolated gate drive, it can be applied both as a high-side and
as a low-side switch.
For each topology, different drive circuits are commonly used, which will
be described in the following. A first overview of the different circuits is given
by Table 2.3.
Table 2.3. Classification of drive circuits for MOS-gated devices
Type of drive Implementation Remarks

circuit
Low-side Complementary Vee-output swing
stage input signal required to cover Vee range
fast transition
inverting behavior
danger of cross-currents in the drive cir-
cuit during turn-on and turn-off
Totem-pole stage inverting behavior

no cross-currents
input signal with only small amplitude
required
Emitter-coupled noninverting behavior

automatic protection of gate against
overvoltages
no cross-currents
delay, as input signal has to change by
2VBE before switching action occurs
High-side p-channel driver only for p-channel MOSFETs, therefore

with level-shifting higher power dissipation in switch
stage simple circuit
Bootstrap driver for n-channel devices

external bootstrap capacitor required
high-side switch has to be turned off
regularly in order to recharge bootstrap
capacitor, therefore only for minimum
operating frequency
Charge pump for n-channel devices

additional supply voltage above battery
voltage created
higher effort
2 Power Circuits 39
Table 2.3. (continued)
Type of drive Implementation Remarks

circuit
Isolated Opto-coupled medium speed
drivers additional floating power supply needed
to drive power switch
Transformer- fast transition

coupled gate drives - no additional power supply on sec-
ondary side
expensive components
2.3.2 Low-side Drive Circuits
Figure 2.23a shows the complementary drive stage. This stage shows an in-
verting behavior and enables a fast turn-on and turn-off of the power tran-
sistor. This configuration of a driver output stage utilizes the Vee range to
its maximum as both driving transistors are operated with lowest drain-
source voltages. Depending on the implementation of the circuit high peak
currents through T 1 and T 2 can occur when both transistors are conducting
(see Fig. 2.23b), so careful design of the driving stage is necessary.
In integrated circuits additional logic or timing components are used to
prohibit the described cross-currents. The circuit in Fig. 2.24 shows the prin-
ciple of a complementary drive stage with delay logic.
If the input voltage Vin is constant, the output of the second inverter
INV 2 is equal to the input voltage. If Vin is high, then the output of the
NAND gate will become low resulting in T 1 conducting. If Vin is low, then
the output of the NOR gate will be high and T 2 conducts. Every time the
Vs
5
---- 'o.n' A
4 - JD,T2/A
vcJ
ov
0
b -10 20 40 60 80 100
a t/ ns
Fig. 2.23. Complementary drive stage. a Schematic, b simulation of cross-currents

through T1 and T2 during switching
40 Stefan Butzmann
Fig. 2.24. Complementary drive stage with delay logic to prevent cross-currents
input voltage changes, the output of INV2 and Vin will have different logic
levels for a delay time, which is determined by the RC network. During this
time, both transistors T 1 and T 2 will be off thus preventing a cross-current
from flowing through T 1 and T 2 .
A further approach of a low-side drive circuit is the push-pull stage with
coupled emitters (Fig. 2.25). The depicted circuit is a noninverting structure
with a low source resistance and is frequently used to boost current from IC
drivers.
At the beginning of a switching cycle the input voltage has to change by
2 VaE before a change of Vcs of the power transistor occurs. Consequently,
only one of the two driving transistors conducts at one time, which results
in the absence of the current shoot-through observed with the complemen-
tary stage. An advantage of this configuration is the ability to clamp the
gate-source voltage Vcs between Vee + VaE and GND- VaE thus providing
a protection of the power-MOSFET against too high gate voltages.
Vcc=15V
PT
Fig. 2.25. Push-pull stage with coupled emitters

2 Power Circuits 41
5VJ
ov
Fig. 2.26. Totem-pole stage
The totem-pole configuration in Fig. 2.26 with an inverting behavior also

prevents the current peaks, which are caused by the complementary-stage
architecture. However, for Vin =High a continuous current flows through R 1
and T2, which increases the power dissipation of the drive circuit. The mini-
mum gate voltage Vcs of the power-MOSFET is increased to the sum of the
saturation voltage of T 2 and the forward voltage drop of the diode. In order
to ensure a turn-off of the power-MOSFET when the driver supply voltage is
removed, the resistor R 2 can be added. This, however, yields a constant bias
current also for Vin = Low.
2.3.3 High-side Drive Circuits
Drive Circuits for p-channel MOSFETs
Applications with grounded loads require the switch to be placed between

the supply voltage and the load. The easiest way to implement a high-side
switch is the use of a p-channel MOSFET. The device is turned on by pulling
the gate voltage low.
Figure 2.27a shows the basic drive circuit for a p-channel power-MOSFET.
When the transistor T 1 is turned on the gate voltage is pulled low via the gate
resistor Rc. When T 1 is turned off, the gate-source capacitance is discharged
via R 1 and the power-MOSFET is turned off.
If the supply voltage VB is larger than the maximum allowed gate-
source voltage Vcs of the MOSFET, the circuit can be modified according
to Fig. 2.27b, where the maximum gate-source voltage is limited by a Zener
diode.
The main drawback of both circuits lies in the values of R 1 and R 2, which
should be high in order to minimize the bias current when the MOSFET is
42 Stefan Butzmann
a b
c
Fig. 2.27. Drive circuits for p-channel MOSFETs. a Principle, b with gate-source
voltage limiting diode, c with emitter-follower stage
turned on. This however yields large switching times. A simple solution is
the circuit according to Fig. 2.27c, where an additional transistor stage has
been added to provide a low-impedance driving output.
Bootstrap Drivers for n-channel MOSFETs
P-channel-MOSFET s are built on p-type epitaxial material and the major-

ity carriers are holes. Due to the lower mobility of holes in comparison to
electrons the on-resistance of p-channel devices is 2-3 times higher than that
of n-channel devices with the same chip size. This yields a higher power
dissipation, which is often not affordable.
For this reason, in most power applications n-channel devices are used. In
order to turn these devices fully on, their gate voltage has to be pulled above
2 Power Circuits 43
Fig. 2.28. High-side driver
the supply voltage VB. Figure 2.28 shows the principle of a high-side driver
with a bootstrap capacitor.
For Vin = High the power-MOSFET is turned off and the bootstrap ca-
pacitor CB is charged via the diode D1 and the load ZL. As soon as Vin
becomes Low the Transistor T 1 is switched off and the potential of the gate
of the power transistor PT is increased. As the source voltage is raised to
approximately VB the bootstrap voltage V* is raised to VB + Vee resulting
in a constant voltage Vcs of the MOSFET. Due to leakage currents in the
driving circuit the capacitor is slowly discharged, which yields an increase of
the on-resistance of the MOSFET and thus in a higher power dissipation.
The bootstrap capacitor has to be recharged by a return of the source volt-
age to zero, which is achieved by turning the power switch off. The described
------------------------------------------------------------'
Vee '
'
'
'
'
Pulse
generator
_n_
------------------------------------------------------------'
'
Fig. 2.29. Integrated level-shifting stage for high-side driver

44 Stefan Butzmann
behavior results in a minimum operating frequency, which is dependent on

the size of the bootstrap capacitor CB.
In integrated circuits, often a level shifting stage as shown in Fig. 2.29 is
employed. The turn-on and turn-off state of the power transistor are entered
by two short pulses generated by driving logic. An RS flip-flop, which is
coupled to the boosted supply voltage, translates these pulses into static
signals at its outputs, thus turning the power-MOSFET on and off. This
method of level shifting has the advantage that the actual driving stage is
decoupled from the level-shifting transistor yielding a faster turn-on of the
power switch.
Charge Pumps
In some applications, e.g. DC motors, the on-time of the high-side switch

may be undefined. As explained above, the previously described bootstrap
circuits suffer from the discharge of the bootstrap capacitor, which results
in a required minimum switching frequency. These concepts would fail in
applications where the on-time is greater than the size the bootstrap capacitor
permits. Charge pump circuits are a frequently used alternative in these cases
in order to create a static voltage above VB.
The inverter in the circuit in Fig. 2.30 is driven by a continuous pulse
train and pulls the capacitor electrode from ground to VB and back. When
the inverter output is low, the capacitor Cpump is charged via the diode D 1
to VB- VD, when it is high, the capacitor Ctank is charged from Cpump via D2
to 2VB- 2VD. If the turn-on time of the power-MOSFET is not critical, the
PT
Fig. 2.30. Charge pump circuit

2 Power Circuits 45
.---------.---------------~--o Vs
Fig. 2.31. Elementary version of charge-pump driver with increased turn-on time
driving circuit consisting of T 1 - T 3 can be saved by replacing the capacitor

Ctank directly by the power-MOSFET (Fig. 2.31).
Some integrated MOSFET drivers combine the bootstrap-technique with
the charge-pump technique in order to provide both fast switching action
at high frequencies and static turn-on conditions of the MOSFET. The en-
ergy for the gate switching action itself is taken from the external bootstrap
capacitor and the energy needed to maintain the gate charge constant is sup-
plied via the charge pump. With this architecture, the necessary charge-pump
capacitor can now be integrated as the power to be transferred is low.
2.3.4 Isolated Gate Drive Circuits
Transformer-coupled Gate Drives
In applications with large potential differences between the control circuit

and the power switch galvanic isolation may be desired. Apart from solutions
with opto-couplers, the use of transformers is the most common approach
for realization of isolated gate drive circuits. Figure 2.32 shows the basic
structure of a transformer-coupled gate drive.
PT
Fig. 2.32. Transformer-coupled gate drive

46 Stefan Butzmann
An important property of transformers is the fact that they can deliver

only AC signals. This results in a limitation of transformer drives to a duty
cycle of 50% or the necessity to realize large voltage swings for duty cycles
other than 50%, otherwise the flux in the core would constantly increase and
finally yield saturation of the core. A common approach to enable varying
duty cycles is the use of a Zener diode as shown in Fig. 2.33.
PT
Fig. 2.33. Transformer-coupled gate drive for varying duty cycles
0
-1 ~------L-------~------~------~------~
0 2 4 6 8 10
0 2 4 6 8 10
t/ ms
Fig. 2.34. Simulation results for circuit according to Fig. 2.33

2 Power Circuits 47
During turn-on of T 1 the voltage Vee is coupled to the gate-source elec-

trodes of the power-MOSFET via the transformer. When T 1 is turned off
again, the current is decreased via the Zener diode ZD. The voltage Vz of the
Zener diode has to be chosen to cause a total flux reset in the transformer
for the maximum occurring duty cycle. However, the (negative) coil voltage
of the secondary side of the transformer may not exceed the maximum spec-
ified negative Vcs value of the MOSFET, which is usually in the order of
20 V. Thus, with typical on-voltages of the power-MOSFET of 10-15 V, only
relatively small variations of the duty cycle can be realized.
The simulation in Fig. 2.34 shows an additional problem of the circuit.
When the current through the primary coil has been reduced to zero, the
current flow through the diode is cut off and the drain voltage starts to
oscillate.
A better approach allowing wider duty cycle variations and simultane-
ously preventing core saturation is the use of a decoupling capacitor on the
10.---------r-------~r-------~---------,
r- r-r-
>
......
0
~
>-10
-20L_________L_______ ~L-------~--------~
0 0.5 1.5 2
20.-------~---------r--------~-------,
15r- r- r- r- r- r- r- r- r- r- r- r-
::: 10
en
~ 5
0
0.5 1 1.5 2
b t/ms
Fig. 2.35. Transformer-coupled gate drive with decoupling capacitor on primary

and secondary winding for varying duty cycles. a Schematic, b simulation results
for secondary coil voltage and Vas of power-MOSFET PT
48 Stefan Butzmann
primary side of the transformer. In order to damp the resulting LC cir-

cuit formed by the decoupling capacitor and the transformer inductance
a damping resistor R 1 is often added. For duty cycle ratios, which are near
zero or 100%, extreme voltage levels may occur, where the turn-on voltage
has decreased so much that the power device will not safely turn-on. Fig-
ure 2.35 shows a circuit which uses a second capacitor and an additional
diode to shift the driving voltage levels back. When the output voltage
of the transformer is negative, the capacitor C 2 is charged to the trans-
former's output voltage minus the forward voltage of diode D. During the
positive half-wave the capacitor voltage is added to the transformer's output
(Fig. 2.35b).
The circuit shown in Fig. 2.36 uses an additional MOSFET to drive the
power switch. When the voltage across the transformer's secondary winding
is positive, the gate-source capacitance of PT 2 is charged via the intrinsic
backward diode of PT 1 . Even when the voltage across the secondary winding
falls to zero, the power transistor PT 2 will remain conducting. After the
output voltage of the transformer has become negative, PT 1 is turned on
and the gate-source capacitance of PT 2 is discharged. The transformer in
this circuit can have a smaller size as it is operated partly in saturation.
Fig. 2.36. Transformer-coupled gate drive with additional MOSFET
Gate Drives with Opto-couplers
In order to achieve galvanic isolation opto-couplers offer the following advan-

tages:
- high isolation voltages of up to 5 kV,
- signal transfer from DC to several MHz,
- simple interfacing to microcontrollers.
Figure 2.37 shows a common circuit used for opto-coupler gate drive iso-
lation, where the signal from the opto-coupler is amplified by two transistor
2 Power Circuits 49
]
f
,-------~-------,
PT
.. ________________________,.
Fig. 2.37. Gate drive circuit with opto-coupler
stages. A disadvantage of opto-couplers is the need of an additional power

supply for the driving circuit on the "secondary" side. Furthermore, the rela-
tively large propagation delay of typically several microseconds can be a fur-
ther disadvantage.
2.3.5 Speed Enhancement and Protection Circuits
Turn-off Speed Enhancement
Apart from dvjdt considerations, the turn-off process of the MOSFET should
take place as fast as possible in order to minimize the switching losses. Ideally,
Rc is tuned for the maximum allowed switching speed of the diode. For
different turn-on and turn-off speeds the circuits in Fig. 2.38 may be used.
In both circuits the turn-on speed is determined by the resistor Rm. The
turn-off speed is controlled by RmiiRG2 in Fig. 2.38a and by RG2 only in
Fig. 2.38b. The main advantage of the turn-off with the pnp-transistor is
given by the fact that the (sometimes significant) gate turn-off current is
held in the smallest possible loop between the gate and the source by the
pnp-transistor, thus reducing EMI problems and also reducing the power
dissipation in the driving stage.
Additional speed-up of the turn-off process can be achieved if a negative
supply voltage for the driving circuit is used. By this measure the charge can
be removed even faster from the gate capacitance of the power-MOSFET.
Overvoltage Protection
Overvoltages occur most commonly when the power switch is turned off while
the load inductance is carrying a current. As stated previously, the diode
starts to conduct only after it has been flooded with minority carriers. Until
this happens, the inductive load keeps a constant current and creates an
50 Stefan Butzmann
15V ____c IVin

-5V :=:r- t
Fig. 2.38. Speed-up circuits for faster turn-off. a Reduction of turn-off time by
use of diode, b use of pnp-transistor for additionally reduction of EMI and power
losses in driving stage
overvoltage at the drain of the MOSFET. Active clamping of MOS-controlled

transistors is a popular way to prevent overvoltages. The principle is shown
in Fig. 2.39.
An additional feedback path consisting of a Zener diode ZD and an ad-
ditional diode D is introduced into the circuit. When the drain voltage rises
above the sum of the breakthrough voltage of the Zener diode and the thresh-
old voltage of the power transistor, the gate voltage of the MOSFET is raised
and the transistor is turned on. The MOSFET in combination with the Zener
element thus acts like a power Zener diode. The diode D in the circuit has
the function to prevent current flow from the driver to the drain when the
power transistor is turned on.
Figure 2.40 shows a simulation of the described clamping circuit, which
was performed without a free-wheeling diode. For the diode ZD a Zener
voltage of Vz = 30V was chosen. When the power-MOSFET is turned off at
t = 5 ms, the current through the inductor increases the drain voltage of the
power-MOSFET. When the clamping voltage
2 Power Circuits 51
.----.---o Vs
ZD
PT
Fig. 2.39. Active-clamping circuit for overvoltage protection
40rr===~~====,-,-----~------,-----~
II- Drain voltage I
> 30 H- - · Gate voltage
~20
~ r-----------------------
~ 10
\. __________ _
0~---------------~
0 2 4 6 8 10
<( 10
_o
c
~ 5
:::J
u
c
-~
0
2 6 8 10
t I ms
Fig. 2.40. Simulation of active clamping circuit without free-wheeling diode
Vclamp = vth,MOSFET + Vz + VF (2.8)
is reached, the power-MOSFET turns on again. In the simulated case without

a free-wheeling diode, the MOSFET is kept in the on-state until the current
through the inductor has been reduced to zero. During this phase, the gate
voltage is held at the threshold level.
52 Stefan Butzmann
Protection Against Gate-Source Overvoltages
Possible causes for gate-source overvoltages are not only error conditions in
the driving circuit, but also the occurrence of a short circuit of the load when
the power-MOSFET is already in the on-state. In this case both the drain
voltage and the drain current rise very quickly. The large resulting voltage
slope dV/ dt in conjunction with the gate-drain capacitance can then yield
a further increase of the gate-source voltage. In order to protect the gate
against such overvoltages, usually two Zener diodes are connected back to
back between gate and source according to Fig. 2.41.
PT
Fig. 2.41. Protection against gate-source overvoltages
Overload Protection
Protecting the power device against short-circuit currents requires additional

circuitry in the driver output stage as the transistor should be turned off fast
after detection of the fault condition.
The most popular method of current protection is undoubtedly the use
of a protection circuit measuring the voltage over a shunt resistor, which is
connected between source and ground (see Fig. 2.42a).
This method is simple, but it requires a low resistance for large currents
in order to keep the shunt's power dissipation low. This again yields small
signal amplitudes, which are sensitive to interference and variations of the
ground level due to parasitic inductances. Measuring a current of 50 A with
a shunt resistor of Rs = 1 m!l gives a sense voltage of Vm = 50 m V and still
yields a power dissipation of P0 = 2.5 W in the shunt resistor.
Because of the small signal amplitudes over the shunt resistor a careful
layout is necessary; the use of differential sense amplifiers can be of further
advantage.
2 Power Circuits 53
Fig. 2.42. Current measurement with shunt resistor. a Shunt resistor in main
current path, b use of shunt resistor with sense-FET
Current transformers offer much lower susceptibility to interference and

even provide galvanic isolation between the power path and the control cir-
cuit. Their cost is much higher than that of shunt resistors, though.
In sense-MOSFETs a small source area is separated from the main current
path providing a second source contact. The current through this source
contact is proportional to the current in the main path reduced by the area
ratio of the two source contacts. An easy way of evaluating the sense current
is shown in Fig. 2.42b. In this configuration, Rs must be chosen to be small
as the measuring voltage over the shunt resistor reduces the sense current.
A more exact way of evaluating the sense current is shown in Fig. 2.43,
where the sense connection is virtually grounded via the operational ampli-
fier. The output of the op-amp is negative, though, thus the circuit requires
a negative supply voltage.
Fig. 2.43. Current measurement with sense-FET and virtual grounding

54 Stefan Butzmann
Figure 2.44 shows the principle of an integrated overcurrent limitation

circuit. The current source IRef creates a reference voltage VRef over RRef
and thus a constant potential at the base of T 3 . While the voltage over the
shunt resistor R 2 is lower than VRef, T 2 is operated in saturation creating
a voltage of VR2 + VBE,4 + VcE,sat,2 ~ 0.8 Vat the base of T5. In that case the
Darlington stage T 5,6 is not conducting and the gate of the power transistor is
maintained high via the current source h. If an overcurrent situation occurs,
then the base voltage of T 4 is raised above VRef and the current through T 2
and T 4 is cut off. This yields a turn-on of the Darlington stage and the gate
of the MOSFET is discharged until the current through the MOSFET has
been reduced to IRef · RRed Rz.
Fig. 2.44. Current limitation circuit
dV/dt Protection
When MOSFETs are exposed to high voltage slew rates during turn-off,
a current Ic = C · dVjdt will flow through the collector-base capacitance of
the parasitic bipolar transistor. In this case, it may happen that the product
of this current and of the resistance RB is large enough to spuriously turn the
transistor on. This uncontrolled action results in higher power dissipation and
undesired interferences. In the worst case, it can even destroy the MOSFET.
dV/dt-limiting has therefore become a central issue for drive circuits, also in
order to reduce the insulation stress in the load.
Figure 2.45 shows a dV/dt-protection circuit, where a capacitor is used
to limit the voltage slew rate during turn-off of the power transistor.
2 Power Circuits 55
Vcco-~-------------------.
PT
Fig. 2.45. dV/dt-limiting with feedback capacitor
Overtemperature Protection
The maximum allowed power dissipation in a power device is a function

of the ambient temperature, the thermal impedance of the setup and the
maximum allowed junction temperature, which is for typical power devices in
the order of 19j,max = 150-175 °C. If the junction temperature rises above the
allowed limit, the power dissipation in the device has to be reduced. Usually
the power-MOSFET is switched off above a certain junction temperature
19 0 ff and switched on again after the junction temperature has fallen below
a temperature 11on·
A simple way to measure the temperature is the use of the base-
emitter junction of a bipolar transistor, where the VsE voltage degrades with
:::::5 -2mV /K. More sophisticated approaches - restricted to monolithically
integrated smart-power devices- evaluate the current generated by a PTAT

current source (PTAT = proportional to absolute temperature), where the
current is a function of the absolute temperature [1].
2.4 Thermal Management
Temperature is certainly one of the most critical factors contributing to device

failure. The increase of power densities in modern mechatronic systems make
an effective thermal management more and more important.
Temperature-caused failures are either caused by mechanical stress or due
to temperature-dependent electrical failures. To the first group belong, e.g.
bond-wire fatigue- caused by badly matched thermal expansion coefficients
between the wire, the bond pad or the substrate - or die fracture in pack-
ages due to thermal expansion coefficient mismatching of the silicon and the
plastic. Failures of this kind are normally caused by thermal cycling of the
device.
56 Stefan Butzmann
RG,int
Fig. 2.46. Equivalent input circuit of MOSFET
Electrical failures are generated when the junction heats up until the
material reaches its melting temperature. Electromigration resulting from
high current densities can cause voids in the metallization which lead to
open circuit failure.
It is therefore extremely important that the maximum specified junction
temperature is not exceeded during device operation. Power dissipation in
power devices is contributed to by the following factors:
- Conduction losses, which are given by the product of the current during
the on-phase and the voltage drop over the device.
For power-MOSFETs the conduction losses are given by
Pn =Ran· f6 (2.9)
As power-MOSFETs do not possess a second breakdown, these devices

are operated safely as long as their junction temperature does not exceed
the specified value, which usually is in the range from 150 oc to 175 °C.
The limiting values of Vns and In are found in the datasheets of the power
devices in the safe operating area diagram (SOA diagram).
- Switching losses, which result from large voltages and large currents at
the same time during the transient phase between the on- and off-state.
Switching losses are proportional to the switching frequency. They are
generally negligible for low frequencies and may become dominant at high
frequencies. Switching losses highly depend on the drive circuit and the
reverse recovery characteristic of the other power elements in the circuit.
- Gate losses in MOS-driven devices are due to the power dissipation in the
internal gate resistance.
Assuming the equivalent circuit of a MOSFET or IGBT input stage ac-
cording to Fig. 2.46, the power loss at Ra,int can be described by
P.
a=
V?
in·
f · Ciss·R Ra,int
+R (2.10)
G,ext G,int
As the input capacitance varies with the drain-source voltage, the de-
scription via the gate charge provides a more suitable description:
2 Power Circuits 57
p,G = TT
Vin ·
f · QG ·
Rc,int
(2.11)
Rc,ext + Rc,int
Usually, the internal gate resistance is much smaller than the external one.
If a reasonably sized external gate resistance is chosen, it is safe to assume
that the greatest part of the gate drive losses is dissipated in the external
gate resistance. The internal gate resistance thus does not significantly
contribute to the heating of the device and is therefore negligible.
The maximum allowed power dissipation is determined by the ambient

temperature and by the thermal impedance Zth,ja of the power transistor:
'19j,max -'I?a = Zth,ja · P max (2.12)
with:
'19j,max maximum junction temperature,

'I? a ambient temperature,
Zth,ja thermal impedance from junction to ambient,
Pmax maximum allowed power dissipation.
The thermal impedance Zth,ja depends on the die size, the thermal pa-
rameters of the materials and on the geometry of the heatsink. It consists of
a real part Rth, which determines the static junction temperature for a given
constant power dissipation, and of a thermal capacitance Cth, which describes
the temperature transients.
In the manufacturers' datasheets, Rth,jc (the thermal resistance from
junction to case) is specified and is of the order of 1-2K/W for typical
power packages. The thermal impedance is generally estimated by the use
of finite-element programs. Figure 2.47 shows a power module construction
with a power-MOSFET soldered on a DCB (Direct Copper Bonding) sub-
strate. This setup, which is frequently used in modules designed for high
power dissipation, was simulated with a finite-element program in the exam-
ple shown. In order to calculate the static thermal resistance of the setup,
the bottom of the DCB substrate was set to a constant temperature '!?a. As
the setup is symmetrical only one quarter transistor was simulated. The tem-
perature difference between the hottest spot on the die and the base plate
with the constant temperature gives the thermal impedance in relation to
the dissipated power.
In a similar way the transient thermal behavior can be simulated. Fig-
ure 2.48 shows an electrical circuit used to model the thermal impedances.
The transient behavior of the transistor can be approximated by an RC cir-
cuit (Fig. 2.48) in order to describe the thermal behavior.
As stated earlier, the on-resistance of a power-MOSFET nearly doubles
during a temperature rise of the junction temperature from '19j = 25 oc to
'19j = 125 °C. Assuming a constant current flowing through the MOSFET,
58 Stefan Butzmann
Power
Al,O, module
Glue
Heat sink
(Aluminum)
; - - - - - - - - - -- - - - - - - , ANSYS 5 . Z
JlOV 18 1996
1 2: S7:45
PLOT DO .
NO DAL SOLOTIOII
TIP C• 7Z . l61
S!IX • 1. 716
. 190687
. 38l37t
. S7Z06
. ?62?47
. 953434
1.144
1 . 335
J . SZS
l . 716
Fig. 2 .47. Construction of DCB and finite-element simulation of temperature dis-

tribution [20]
the resulting power dissipation Po = Ros,on · I'5 yields an increase of the

junction temperature which in turn raises the MOSFET's on-resistance. This
again results in an increase of Po. This behavior is called the self-heating
effect (see Fig. 2.49) and can be simulated only by iterative simulations. The
results show higher junction temperatures than can be expected without
consideration of the self-heating effect.
A proper way to simulate the self-heating effect is the extension of the
transistor model by an additional resistor in series to the power-MOSFET,
which accounts for the rise of the on-resistance over temperature. The com-
plete electro-thermal model of a power-MOSFET is shown in Fig. 2.50. In the
model the power dissipation over the complete MOSFET is calculated and
the junction temperature {} J is calculated via the thermal model. The new
junction temperature is used to determine the on-resistance of the power
switch and the additional series resistor is adapted to the correct value
.D.Ros-
2 Power Circuits 59
Cth,1
P(t)
Cth,2
Cth,3
Fig. 2.48. Equivalent electrical circuit for modelling thermal impedances
Figure 2.51 shows a comparison of the simulations of the transient tem-

perature with and without consideration of the self-heating effect. The sim-
ulation was performed for the described DCB setup with a 200 V power-
MOSFET with a nominal on-resistance of Rns,on = 200 mO at a temperature
of {)a = 25 °C. It was assumed that the ·power-MOSFET conducts a drain
current of In = 10 A from t = 0 on. The simulation results show, that the
increase of the junction temperature is L\{}J ~ 30 °C, which corresponds to
a rise of the on-resistance L\Rns = 60 mO. Without consideration of the self-
heating effect, the junction temperature of the transistor is calculated to be
too low by L\{)J = 9 °C.
Power
dissipation
...,.. Heat
generation
~ /
Increase of
Roson
Fig. 2.49. Self-heating effect in MOSFETs

60 Stefan Butzmann
,..............................................................................................................................................................................................................................................
f1Ros = f(T) - TJ
D'
G~9 _i
i__ _______ ~-±--------------------------------------------------------------------------------------------------------------
Fig. 2.50. Electro-thermal model of a power-MOSFET with self-heating capability
60~======~=====~----,------,-----,
- - with self-heating
55 . - - · without self-heating
---- ------ ---------------
25.. . . . . . . . . . .- ........................ .
20 ~~------0~.~2----~0~.4----~0~.6~----~0.~8----~
tis
Fig. 2.51. Comparison of temperature simulations with and without consideration
of self-heating effect
2.5 DC Switching Converters
2.5.1 Topologies
DC chopper circuits are applied where an inductive load is to be supplied with

a variable current from a fixed DC supply voltage. A common application is
the drive of DC motors from a battery supply (see Sect. 3.2).
Figure 2.52 shows the principle of a DC chopper circuit with an inductive
load L.
The operation of the chopper circuit can be divided into two modes. When
the switch is closed, the current flows from the supply voltage to the load.
2 Power Circuits 61
Fig. 2.52. DC chopper circuit for one-quadrant operation (Buck-topology)
According to dJ Idt = VI L, the current through the inductance rises. When

the switch is opened, the current continues to flow through the inductance
and seeks its way through the free-wheeling diode D. It decreases slowly due
to the forward voltage drop of the diode and due to parasitic resistances.
The described converter topology is referred to as a single-quadrant chopper
with positive load voltage and positive load current (class-A chopper, buck
converter topology).
Another chopper circuit operates with positive supply voltage, but nega-
tive load current and is referred to as a class-B chopper (boost converter).
Figure 2.53 shows the circuit where the voltage source Vemr is for ex-
ample the back electromotive force of a DC motor. When the switch 8 is
closed, the current through the inductor rises. When 8 is opened again, the
inductor current flows through the diode D and returns energy to the supply
voltage VB.
Combining the two circuits allows operation with positive supply voltage
and bidirectional current (class-C converter; two-quadrant-operation) while
the class-D converter provides a unidirectional current and bipolar operation
(Fig. 2.54). The class E converter (Fig. 2.55) provides both motoring and
regenerating operation in both directions (four-quadrant-operation). Usually
the switches 8 1 and 8 4 are on simultaneously and the switches 8 2 and 83
are also on simultaneously, thus allowing the load to be connected to +VB, 0
and -VB.
!Vemf
Fig. 2.53. DC chopper circuit for one-quadrant operation (boost topology)

62 Stefan Butzmann
Fig. 2.54. Two-quadrant converters. a Class C, b class D
Fig. 2.55. Four-quadrant class-E converter
2.5.2 Electromagnetic Value Actuation with a Class-D Stage
Class-D and class-E converters are the most commonly used topologies for
driving inductive actuators. While class-E converters are mainly used in mo-
tor drives to enable forward and backward rotation (see Sect. 3.5.2, 4.2.1),
class-D converters are frequently applied for the drive of electromagnets,
where the direction of the produced magnetic field is of no further interest.
Examples are lifting magnets or valve actuators [21].
Figure 2.56 shows the structure of an electromagnetic valve actuator used
for camless engines. The actuator has two coils exposing an armature to
a magnetic field when a current flows through them. The resulting magnetic
force yields a movement of the armature and thus the closing or opening
action of the valve. Because of the relatively large valve lift of ~ 8-10 mm
two valve springs are employed in order to support the movement.
A typical driving circuit for each of the two coils is depicted in Fig. 2.57.
The power transistors T 7 and T 8 and the diodes D 2 and D 3 form the actual
switching converter stage. The transistors are driven by an integrated low-
side and high-side driver with the high-side driver being bootstrapped as
shown in Fig. 2.28.
The peak current of the circuit is determined by the output of the DAC
converter which itself is controlled by the microcontroller. The resistors R1
2 Power Circuits 63
Fig. 2.56. Electromagnetic actuator for variable valve train
and R 2 provide a hysteresis with a constant factor in relation to the DAC

output and thus determine the maximum and minimum current level. When
the lower transistor Ts is turned on, the current signal over the shunt resistor
Rs is amplified by a differential amplifier and compared with the current
levels. Depending on the output of the comparators, the RS flip-flop switches
the upper transistor on and off. When the upper transistor is turned on,
then the full battery voltage VB is applied at the actuator coil, when it is
turned off, the coil voltage is reduced to the negative forward voltage of D 2
in addition with the voltage drop over T 8 and the shunt resistor Rs. In this
case, the current through the coil is slowly reduced.
When both the lower and the upper transistors are turned off then the
current through the actuator coil continues to flow via the diodes D 2 and D 3
until it is reduced to zero. During this time, the coil voltage is inverted to
-VB and the current is rapidly reduced.
Careful consideration in actuator drive circuits should be given to layout.
As shown in Fig. 2.57, the ground was partitioned into an analog ground
AGND for the control circuit and a power ground PGND. The current
through the shunt resistor is measured with a differential amplifier and re-
ferred to AGND for the signal conditioning. As both the low- and high-side
driver have to supply their gate signals with respect to PGND the logic in-
put signals were converted to currents and transformed to PGND via the two
current mirrors T 2 ,3 and T 5 ,6 . These measures provide the desired decoupling
of the signal conditioning from the drive circuit.
For a given battery voltage VB only a limited gradient of the current
can be achieved, which is determined by the quotient VB/ L. The achievable
current gradients may often be too small for effective control of the actuator
64 Stefan Butzmann
0
> (")
z
(.!)
N
'<!"
0 a..
II
~
II
()
-5?
I 0::: I Cl)
0
z
~
Fig. 2.57. Complete driving circuit for one actuator coil

2 Power Circuits 65
Fig. 2.58. Modified power stage with dynamic high voltage boosting [22]
and higher current slopes are often desired. Figure 2.58 shows a modified
power stage, which provides an additional high voltage boost mode without
using an external supply.
The basic idea of the circuit is the feedback of the current via the diode
D 2 into a high-voltage storage capacitor CHv instead of the battery supply
during turn-off of the transistors T 1 and T 2 . When a high current gradient is
needed, T 3 instead of T 1 is turned on and CHv is discharged. As the upper
coil voltage is then pulled to Vs + Vc,Hv, the diode D 3 was inserted into the
circuit in order to prevent the intrinsic body diode of T 1 from conducting.
2.5.3 Stepper-Motor Drive Circuits
Stepper-motors are commonly used to provide a reliable controlled motion

based on digital information (also see Sect. 3.4.3). A desired motor position
can be reached with high precision by counting the number of steps applied
to the motor. Often, stepper-motors are driven "open-loop", which avoids
the need for expensive sensors and feedback-loops. In contrast, closed-loop
drives, where the motor position is controlled by position sensors, offer higher
reliability and a smoother motion due to the adjustment of the switching
sequence. In general, stepper-motors have two coils, which can be driven in
unipolar or in bipolar mode (Fig. 2.59).
66 Stefan Butzmann
E F D
Fig. 2.59. Stepper motor with permanent magnet and two coils
For unipolar drive the center taps of the two coils are connected to the
DC supply voltage, effectively resulting in four (half- )coils. A simplified drive
circuit for unipolar mode is shown in Fig. 2.60.
Here, the direction of the magnetic field produced is determined by the ac-
tivated half-coils. A rotation of the permanent magnet is achieved by switch-
ing the transistors as shown in Fig. 2.61. In the depicted case of a simple
drive circuit, the current through the motor coils is limited by the resistance
of the coil windings and the supply voltage.
While current limiting by the resistance of the coil winding is a convenient
option for low-torque motors, it usually results in low speed rates due to the
Vs
.-----------~-------------.----------~
Fig. 2.60. Drive circuit for unipolar mode

2 Power Circuits 67
L-R time constant of the system. An improvement could be achieved by the

use of a higher supply voltage and an external current limiting resistor. Figure
2.62 shows the simulation of the current through a motor coil with a nominal
current of IN = 1 A, an inductance of 10 mH and an internal coil-resistance
of 50. If the supply voltage is chosen as VB = 5 V, then no external current
limiting resistor is needed; however, the time needed for the current to build
up is determined by the ratio of the coil inductance and the coil resistance .
• t
~--;----+1,--~r---+---~--~~t
I
I
Rotor
position
Fig. 2.61. Transistor switching sequence and rotor position for unipolar drive
. . .
········:····.-·-·---_;...;--;-------:-----:----~
~ j , ,*' ~ ~ ~
....... -:·-I/ .......... .. . .................. .................. :· . . . .
0.8
:
: ;,
.. ·~.......... ·~ ~
:: :
:
::
········:··!:············· ·:··················:··················:··················:········
:..
:~
:
:
:
:
:
:
:
:
[i :
0.4
0.2
=• :
:
:
:
:
:
:
·_VB= ~V . .
--. VB= 25\/, Rext = 200
. _ .• VB =25 Vand current limitation circuit
-0·2 L___io_ _ _..!:::J2=====4====6====a=::::::!.J
t/ ms
Fig. 2.62. Simulation of current through motor coils

68 Stefan Butzmann
The current rises faster with a higher supply voltage and an external resistor,
but dissipates power in the external resistor.
A further approach is the use of an external current limitation circuit (see
Fig. 2.63), where the coils are exposed to almost the full supply voltage V8
when one of the transistors T2 or T3 is turned on. When the desired current
value is reached and the voltage drop over the shunt resistor equals VRef, the
pup-transistor T 1 limits the current through the coils. This circuit provides
further speed advantages (see Fig. 2.62). However, it also dissipates power in
the pup-transistor.
A common approach therefore is to limit the current in the motor coils
by chopping circuits. Figure 2.64 shows the realization of a chopping circuit
with constant off-time, where a user-adjustable mono-flop turns off the active
power transistor T 1 or T 2 for a constant time when the current through the
shunt resistor exceeds the maximum allowed value (VRef = I· Rs). When
the mono-flop resets, the power switch is turned on and the coil current rises
again. This principle has been applied for example in the SLA7024M (Allegro
Microsystems).
In contrast to the unipolar drive mode, the bipolar drive mode always
uses the full coils. The direction of the resulting magnetic field is determined
here by the direction of the current through the full coils. The bipolar mode
gives a higher torque, but requires two class-E power stages. In bipolar drive
mode, both windings are usually continuously energized, while the direction
of the coil currents determines the motor position (see Fig. 2.65) .
.---..----ovs
Fig. 2.63. Use of current limitation circuit

2 Power Circuits 69
,--------{) Vs
Fig. 2.64. Chopping circuit with constant off-time
Rotor
position
Fig. 2.65. Currents through motor coils in bipolar mode and the resulting motor
positions
Figure 2.66 shows a typical bipolar stepper-motor drive with the controller
1297 (ST) and the dual full-bridge driver 1298 (ST). In the circuit, a mi-
crocontroller generates the signals for forward or backward stepping, which
are translated by the 1297 into the necessary control signals for the power
stages.
70 Stefan Butzmann
+SV
r--t--OVB
osc
L297
INH1
L1
8
Step-clock
Half/full step
~c
Direction
Enable
Ds
03
c
INH2
L2
D
07
04
+SV
Fig. 2.66. Complete stepper-motor drive circuit

2 Power Circuits 71
The current through the power stages is supervised via two shunt resistors
and kept at a quasi-constant level for the active coil. For a particular direction
of current, each coil can be operated in 3 modes: fast current increase, fast
current decrease and freewheeling mode. For fast current increase in coil £ 1 ,
for example, output A of the 1297 is pulled high and output B is pulled to
ground. For fast current decrease, the IN HI-signal is pulled to ground and
all transistors in the power stage are switched off. In this case, the current
through the motor coil flows through the diodes D 2 and D3 until it has
decreased to zero. The freewheeling mode is realized by pulling signals A and
B to logic high, while IN Hl is also held at high level. In this case, the
current flows through transistor T 1 , the motor coil £ 1 and the diode D 2
and will decrease only slowly. In comparison to the other possible path for
freewheeling operation D3 - £ 1 - T 4 - Rs 1 , the power dissipation is lower as
the shunt resistor is not included in the current path.
Figure 2.67 shows the current limitation circuit used in the stepper-motor
controller 1297. In this circuit, the peak current through the motor coils is
monitored. Two flip-flops FF 1 and FF 2 are set at the rate of an internal
oscillator. When the voltage at the sense input, which is determined by the
current through the coil and the sense resistor, reaches the programmed peak
value, the flip-flops are reset and the following power stage is set into the
freewheeling mode until the next oscillator edge.
to power stage
Input Logic Output Logic
Fig. 2.67. Current limitation by peak-current detection

72 Stefan Butzrnann
Rotor
position
Fig. 2.68. Current waveforms and motor positions in half-step mode
The switching scheme depicted in Fig. 2.65 is the most popular for full-
step mode and provides the highest torque. A further drive method for
stepper-motors is the half-step mode, where one of the motor phases is with-
out current, when the motor is in the half-step position. This mode of opera-
tion provides an increase in the position resolution by a factor of 2. However,
it results in a great variation of torque between the full-step and the half-step
positions. Figure 2.68 shows the currents through the motor coils in half-step
mode and the corresponding motor positions.
2.5.4 Zero-current Converters
As discussed before, switching losses in the power transistor degrade the

efficiency of switching converters particularly at high frequencies. As these
switching losses are proportional to the product of the voltage over the switch
and the current flowing through the switch at the time of switching, they can
be drastically reduced if either the voltage or the current is zero when the
switch is opened or closed. Quasi-resonant converters provide a possible way
to achieve this goal. They have significant advantages over classical PWM
converters, which are low device stress, low electromagnetic interference and
high efficiency. This allows the application of high switching frequencies and
thus smaller-sized magnetic components.
A popular quasi-resonant converter is the zero-current converter, which
shall be discussed here as a typical representative of this family of converters.
Figure 2.69 shows a full-wave zero-current converter, where the switching
transistor T 1 is connected in series with the inductance Lr.
2 Power Circuits 73
Fig. 2.69. Zero-current converter of full-wave type
For an understanding of the circuit operation, the load current hL

through the inductance LL is assumed to be constant. The resulting wave-
forms of the voltage Vcr across the capacitor Cr and of the coil current
through Lr are shown in Fig. 2. 70.
The transistor T 1 is turned on when the current through Lr is zero. The
voltage across the capacitor Cr swings to approximately twice the supply
voltage Vs and - when hr is negative- back to almost zero. The basic idea of
the zero-current converter is to turn the transistor T 1 off after one oscillation
period, when the current through the transistor has returned to zero again.
However, the requirement for a precise timing of the turn-off of T 1 is relaxed
by the use of the anti-parallel diode, which takes over the current when the
transistor is turned off during the ring-back period of the coil current. The
Fig. 2. 70. Waveforms of capacitor voltage Vc and of current through inductor Lr

for full-wave zero-current converter
74 Stefan Butzmann
T1 01 L,
/LL
I
Vs! c, 02 LL
Fig. 2.71. Zero-current converter of half-wave type
amount of energy transmitted to the load is thus only determined by the

repetition rate of the turn-ons of the power switch. In order to increase the
output power, the frequency has to be increased. As the amount of energy
transmitted with each event is constant, a drawback of this type of control
is a high ripple, if only a low amount of energy is to be transmitted to the
load.
A modification of the full-wave converter is the half-wave zero-current
converter, which is shown in Fig. 2.71. In this circuit, the diode D 1 is con-
nected in series to the transistor T 1 . Thus, the diode prevents the current
from flowing back to VB and the capacitor is discharged only by the load
current hL·
2.6 Current Development and Future Trends
2.6.1 Overview
The goal of current development activities is the continuous improvement

of existing semiconductor switches. This includes the reduction of the on-
resistance and of the switching losses and for some applications the increase
of the maximum off-state voltage. This trend has led to an optimization of
the cell-designs of MOSFETs and IGBTs including the shrinking of the cell
areas due to self-adjusting production processes, but also fundamentally new
structures of power switches. Typical examples for the latter are trench ele-
ments and CoolMOS-FETs. Table 2.4 gives a comparison of the voltage rat-
ings and the on-resistances of trench-FETs, CoolMOS-FETs and the classical
DMOS-FETs. Both trench-FETs and CoolMOS-FETs have been optimized
for particular voltage ratings: CoolMOS-FETs are available with ratings from
500 V to 800 V, trench-FETs with ratings up to 100 V, thus targeting mainly
at the automotive area. The given examples in Table 2.4 show typical repre-
sentatives of each of the device families.
Simultaneously, a trend towards intelligent switches with integrated pro-
tection functions could be observed during the last few years. So-called
"smart-power" devices even provide the combination of power switches and
functional control on a single chip. Figure 2.72 shows the power ratings of
2 Power Circuits 75
T a ble 2.4. Overview of power-MOSFET technologies
DMOS Trench-FET CoolMOS

Voltage ratings 20V- 1000V 20V- 100V 500V- 800V
2.3
RoniVBR Ron~ VBR Ron~ VsR
Ron I
(m!1· mm 2 ) ::::: 200@ 75 V :::::90@ 75V ::::: 3000@ 600V
(typical values)
Example IRF1407 BUK7213-75B SPP03N60C3
(International (Philips) (Infineon)
Rectifier)
VsR IV 75 75 600
Ron I m!1 8 12 1400
Qg InC 160 41 13
Ron· Qg I pVs 1280 492 18000
discretes and smart-power devices versus their complexity. While monolithic

integration of t h e power switch and t he control logic offers the highest relia-
bility, the power ratings of smart-power devices are limited due to the higher
on-resistance of the integrated power switches. For this reason, there is a gap
between intelligent control, on one hand, and the use of h igh-power semi-
conductors, on the other, which can currently only be closed by the use of
discrete drive circuits.
Current I
10
MOS
10 4 Voltage IV
Complexity
Fig . 2. 72 . Power ratings vs. complexity of discretes and smart-power devices [24}
76 Stefan Butzmann
A higher versatility is given by the concept of an intelligent power con-

troller which provides driving intelligence and protective functions for discrete
power-FETs (see Table 2.5), thus closing the gap between high power and
intelligent drive.
Table 2.5. Classification of smart-power devices
Smart discretes Intelligent drivers Smart-power ICs

integration of intelligent driver IC power switch and
protective functions for discrete control functions on
on transistor die power-FETs one die
vertical current flow contains integrated targets at
integration of two protection and medium-current
or more switches on control functions applications
one die only with bus interface multiple isolated
common drain power outputs
also available in
multi-chip modules
Further integration yields even higher complexity, the so-called super-smart-

power devices, where microcontrollers and on-chip memories (EEPROM/
Flash-ROM) are integrated together with analog and power stages. Such
elements are intended for system integration in automotive and computer
applications, e.g. like printers.
2.6.2 Trench-FETs
Most power-MOSFET technologies available on the market use a horizontal

gate structure and a vertical current flow to the drain at the backside of
the wafer (see Fig. 2.10). Power-MOSFETs consist of a high number of the
described cells connected in parallel in order to reduce the on-resistance of
the complete device.
The most important MOSFET technology today is the DMOS-FET. The
cell-structure itself is different from manufacturer to manufacturer, but the
underlying principle of a lateral channel and a vertical current flow to the
drain contact at the backside of the die is the same. With such DMOS devices,
however, there is a fundamental limit to cell density due to the parasitic
junction-PET, which provides an additional resistance increasing with the
cell-density resulting from the pinch-off effect. This limits the optimization
of power-MOSFETs only by lateral shrinking.
A significant breakthrough was marked by the development of trench-
PETs, where the channel is vertically oriented and the gate structure is built
in a trench into the die volume, instead of being parallel to the surface of the
die (see Fig. 2.73).
2 Power Circuits 77
s G s
p p
Fig. 2. 73. Cross-section of a trench-FET
In a trench-FET the current flow in the channel is vertical from the source
to the drain contact on the backside of the wafer. As the junction-FET prob-
lem is now avoided, the cell density may be increased again, thus allowing
a further r eduction of the Ros ,on p er chip area (see Table 2.4). Driven mainly
by automotive and computer applications, today most trench-FETs are avail-
able in a voltage range from 20 V to approximately 100 V with specific on-
resistances of less than 200 mfl·mm 2 for a breakdown voltage of 100 V. In
comparison to standard power-MOSFETs with similar voltage ratings the
Ran · Qg-product is reduced by approximately a factor 2. An ongoing trend
is the increase of the number of dies per wafer , which gives additional cost
benefits or at least makes up for the more expensive manufacturing costs.
2.6.3 CoolMOS
While trench-FETs offer their m ain advantages for voltage ratings of up to

100 V , the development of medium-voltage transistors up to 800 V has been
marked by the introduction of CoolMOS.
CoolMOS transistors contain vertical implanted p-tubes as shown in
Fig. 2.74. These stripes introduce extra lateral pn-junctions, which are re-
verse biased, when the device is turned off. The result is a complete depletion
of the drift zone , while the off-state voltage is built up in the horizontal di-
rection. As a consequence, then- -drift area can be significantly reduced and
the reduction in the doping towards high off-state voltages as in conventional
MOSFETs is no longer necessary. The turn-on resistance of CoolMOS-FETs
increases only linearly with the breakdown voltage V BR, which is different
from the exponent of 2.3 as is the case for conventional MOSFETs. The in-
creased conductivity results in a n on-resistance a bout a factor of 5 lower than
that of conventional power-MOSFETs for a rating of 600V. Thus a higher
current capability p er chip area can be achieved and a lower gat e charge in
comparison to similarly rated conventional MOSFETs.
As CoolMOS-FETs with similar ratings as conventional power-MOSFETs
are much smaller in chip size, the parasitic capacitances are also reduced . This
78 Stefan Butzmann
Fig. 2.74. Cross-section of a CoolMOS transistor
yields switching losses about 50 % lower than those of conventional transistors

[15]. One disadvantage of CoolMOS-FETs, however, is the poor dynamic
behavior of the intrinsic body diode.
2.6.4 "Intelligent" Switches
Intelligent switches or "smart discretes" present an intermediate step from

discrete power devices towards more complex smart-power ICs. They com-
bine power-MOSFETs with protective functions like t emperature protection,
overvoltage or overload protection.
The integration of additional functionality into the power switch increases
the reliability of the device and allows the implementation of driving func-
tions into the switch, thus reducing the effort for the external drive circuitry.
Smart-power switches can be realized either by monolithical integration of
the control circuit and the power switch on one die or by multiple chip mount-
ing, where the control IC and the power switch are fabricated separately in
different optimized technologies and interconnected in the package. While the
realization of discrete MOSFETs uses approximately 6- 8 masks, t he mono-
Fig. 2.75. Architecture of a T empFET

2 Power Circuits 79
Statuso-- - - --------,
Supplyo-- - - - - - -----,
Input o---r=;::=::£:=::;===;-ttl~
Fig. 2. 76. Block structure of a TOPFET
lithically integrated power switch needs 12-15 masks, thus roughly doubling
the price per chip area.
Examples of a multiple-chip solution are Infineon's Temp-FETs, where the
power-MOSFET can be turned off by a thyristor, which in turn is controlled
by a temperature sensor.
More sophisticated devices, like the HITFET (Infineon) include additional
functions like overvoltage protection, overload protection, current and d V / dt
limitation and even a status feedback to the drive circuit via a digital flag.
Other manufacturers produce smart switches with similar functions by
monolithical integration of the control circuit and the power switch on a single
chip. Examples of these types of switches are ST's OmniFETs or Philips'
TOPFETs.
OL1
SL1
OLI
N.C. OL1
BATT BATT
Gt.D SH1
IH1 7 SHI
STATUS SH2
IH2 SH2
BATT BATT
N.C. OL2
SL2
L2 SL2
ot..2 OL2
Fig. 2.77. Full-bridge with "intelligent" switches (Courtesy of Royal Philips Elec-
tronics)
80 Stefan Butzmann
The latter device family combines the trench-FET technology with pro-
tective functions and is separated into low-side and high-side TOPFETs.
The low-side TOPFETs offer logic-level inputs, voltage clamping, dV/dt lim-
iting, overtemperature-, overload- and ESD-protection. An extended version
allows status reporting via an additional pin; however, it has to be supplied
by a separate 5 V source. The high-side version has an integrated charge
pump in order to provide the required additional supply voltage to turn the
n-channel MOSFET on.
Meanwhile, complete smart-power-MOSFET bridges are available in a sin-
gle package, where both high-side and low-side switches have been integrated
into one single package. Figure 2. 77 shows the lead frame and the output pin-
ning of an intelligent full bridge, where the two low-side switches are separate.
The two high-side switches in the center have a common drain connection
and share the charge pump and some of the diagnosis functions.
2.6.5 Smart Power ICs
While intelligent protection functions are implemented in "intelligent" swit-

ches, a further step towards smart-power ICs is the implementation of intelli-
gent control functions and power switches on one chip. Modern smart-power
technologies combine _6ipolar devices, QMOS logic and DMOS power tran-
sistors (BCD technology). This gives the designers the chance to select the
optimum technology for each circuit function: bipolar for low-offset ampli-
fiers and reference circuits, CMOS for high density circuit parts (digital and
analog) and DMOS for power functions. As a consequence, complex systems
can be built on a single chip.
The advantages of this approach are obvious:
reduced size of power control circuits,
reduced number of connections,
higher reliability.
All of the protection functions realized in intelligent switches like temper-
ature supervision or current limiting can also be integrated in a smart-power
chip. In motor applications, for example, simply protected transistors are
found as well as complete bridge drives with control and diagnosis functions.
While first generations of smart-power devices only featured low inte-
gration density, there is a trend towards the integration of increasing func-
tionality on the chip, including the use of microcontroller cores and DSP
blocks. This also yields significant demand of memory in today's smart-power
ICs.
As this development is driven in large part by the car industry, auto-
motive electronics is one of the main application areas of smart-power sys-
tems. With the increased use of the number of microcontrollers in modern
cars logic-level interfacing has become more and more important. The use
of intelligent power components becomes particularly advantageous, where
2 Power Circuits 81
transducers for disk drives, printers or motor control can directly be driven
from the microcontroller unit.
Besides automotive syst ems, other typical application fields of smart-
power ICs are power supplies, regulators and even audio amplifiers. In the
field of industrial electronics, integrated circuits in high-voltage technol-
ogy are used to realize off-line applications, e.g. power factor correction
circuits.
For the realization of smart-power devices, mainly three isolation tech-
niques are employed today. With self-isolation the power device is realized as
a vertical transistor. The power stage output is at the backside of the wafer
as is the case in conventional power devices. Examples for this technology
are the VIPower technology by ST or the Smart-SIPMOS technology (Infi-
neon). Thus, the voltage and the current ratings are also similar to those of
conventional discretes. This kind of isolation technique is usually applied for
high-voltage integrated circuits. The complexity of these devices is however
limited, as they share a common collector/ drain contact for multiple power
outputs. Yet, t he low on-resistance makes this kind of device an ideal choice
for high-side switches.
Figure 2.78 shows a sectional drawing of a smart-power device in a junc-
tion isolated smart-power technology. With this technology, quasi-vertical
DMOS transistors are implemented, where the drain current flows back to
the surface via a buried n+ -layer. As the transistors can be separated by p+-
areas, it is possible to realize electrically isolated devices on one chip, which
allows the implementation of low-side and high-side switches. BCD devices
may be more complex than vertical structures, however, their on-resistance
is higher.
The third technology used for smart-power systems is dielectric isolation
with completely isolated areas of silicon. The manufacturing process, however,
is relatively complex and the resulting ICs are quite expensive.
Where both complex control functions and the performance of vertical
transistor technologies are required, the separation of power control and the
power switches can be a good choice simultaneously offering a higher flex-
ibility in the design of the overall system architecture than is feasible with
n+ Buried layer
Fig. 2. 78. Realization of a vertical DMOS tra nsistor in a smart-power process

82 Stefan Butzmann
a complete integrated solution. The concept of "intelligent" power controllers

is heading into this direction.
Figure 2.79 shows the block diagram of an "intelligent" power controller,
which integrates the drive and control circuits for eight power transistors.
Each of the control channels is configurable by software as high- or low-
side drivers. The controller provides extensive protection functions like pro-
grammable current limiting, an SPI watchdog and diagnostic output to the
microcontroller via SPI. Thus, a wide range of MOSFETs for different ap-
plications and voltage ranges can be driven with one single IC controlled via
a serial interface , thus reducing the required number of output pins of the
microcontroller.
Control Logic &

Registers
Fig. 2. 79. Intelligent power controller block structure (Courtesy of Royal Philips
Electronics)
2.6.6 Silicon Carbide
In today's technologies , silicon is mainly used for production of power devices.

With increasing temperatures, however, the leakage currents grow unaccept-
ably large until the intended device operation is replaced by uncontrolled
device behavior.
For this reason , new materials are being explored with higher bandgaps
and increased thermal conductivities. Silicon carbide is such a material. Sil-
icon carbide has a bandgap, a thermal conductivity and a breakdown field
approximately 2-3 times higher than that of silicon (dependent on the chosen
polytype). The high operational junction temperatures , which are possible
with silicon carbide, permit the realization of power devices with extremely
high power densities. Due to the higher breakdown field and the greater
2 Power Circuits 83
bandgap higher switching frequencies can be realized with silicon carbide

devices thus allowing further shrinking of capacitors and inductors in the
circuit. The improved thermal conductivity allows a further increase of the
ambient temperature as cooling measures can be applied more efficiently.
Even though protoype SiC devices have shown excellent performance,
mass production has for a long time been prevented by the presence of defects
such as micropipes in the crystal structure. Only recently the first commer-
cial silicon carbide based Schottky power diodes have become available from
Infineon. The main features of these devices are the virtual absence of the
reverse recovery current and the virtually missing temperature influence on
the switching behavior in the specified temperature range. The forward volt-
age drop of the devices is significantly higher than that of comparable diodes
based on silicon, though. However, it can be expected that lower forward
drop devices will be available in the near future as the production processes
are continuously improved.
References
1. B. Murari, F. Bertotti, G.A. Vignola: Smart Power !Cs, Springer Verlag
2. Stengl, Tihanyi: Leistungs-MOS-FET-Praxis, Pflaum Verlag
3. Schroder: Elektrische Antriebe, Ed. 3: Leistungselektronische Eauelemente,
Springer Verlag
4. Schroder: Elektrische Antriebe, Ed. 4: Leistungselektronische Schaltungen,
Springer Verlag
5. M. Rashid: Power Electronics, Prentice Hall
6. B.J. Baliga: Modern Power Devices, John Wiley & Sons, 1987
7. Power Semiconductor Applications, Philips Semiconductors, 1994
8. Current Ratings of Power Semiconductors, Application Note AN-949, Inter-
national Rectifier
9. J.M. Peter: Characteristics of Power Semiconductors, Application Note
AN512, ST
10. Selecting and Designing in The Right Schottky, Application Note AN-968, In-
ternational Rectifier
11. Datasheet EU208, ST
12. Datasheet IRF640, ST
13. !GET Characteristics, Application Note AN-983, International Rectifier
14. C. Dorlemann: Geregelte Ansteuerung von Insulated Gate Bipolar Transis-
toren. Dissertation, Ruhr-Universitat Bochum, 2002
15. K. Sheng, F. Udrea et a!: PT and NPT !GETs up to 1.2kV: Which is op-
timum? Proceedings, 9th International Conference and Exhibition on Power
Electronics and Motion Control, EPE-PEMC 2000
16. B. Andreycak: Practical Considerations in High Performance MOSFET, !GET
and MGT Gate Drive Circuits, Application Note U-137, Unitrode
17. A.D. Pathak: MOSFET/!GET-Drivers- Theory and Applications, Application
Note AN0002, Ixys
18. Use Gate Charge to Design the Gate Drive Circuit for Power MOSFETs and
!GETs, Application Note AN-944, International Rectifier
84 Stefan Butzmann
19. HV Floating MOS-Gate Driver ICs, Application Note AN-978, International

Rectifier
20. S. Butzmann, J. Melbert: Elektrisch- Thermische Simulation von Leistungs-
bauelementen, GI/ITG/GMM-Diskussionssitzung ,Heterogene Systeme", 11-
menau, 1997
21. Pischinger et a!: Electromagnetically Operating Actuator, US Patent 4,455,543,
1984
22. S. Butzmann, J. Melbert: Device for Controlling a Regulator, Patent WO
00/09867, 2000
23. S. Butzmann: Sensorlose Regelung elektromagnetischer Aktuatoren fiir den
variablen Ventiltrieb. Dissertation, Ruhr-Universitat Bochum, 2000
24. J. Melbert: Power Semiconductors as a Key Component for System Integration
ISPSD '97, 1997
25. W. Pribyl: Integrated Smart Power Circuits Technology, Design and Applica-
tion Proceedings of the 22nd European Solid-State Circuits Conference, ESS-
CIRC96, Neuchtel, Switzerland, 17-19 September 1996
26. A. Lidow, D. Kinzer et a!: The Semiconductor Roadmap for Power Manage-
ment in the New Millennium. In: Special Issne on Power Electronics Technol-
ogy, Proceedings of the IEEE, Vol. 89 No. 6, June 2001
27. B.J. Baliga: The Future of Power Semiconductor Device Technology. In: Special
Issue on Power Electronics Technology, Proceedings of the IEEE, Vol. 89 No.
6, June 2001
28. B.J. Baliga: An Overview of Smart Power Technology. In: IEEE Transactions
on Electron Devices, Vol. 38 No. 7, July 1991
29. K. Satoh, M. Yamamoto: The Present State of the Art in High-Power Semicon-
ductor Devices. In: Special Issue on Power Electronics Technology, Proceedings
of the IEEE, Vol. 89 No. 6, June 2001
3 Electromagnetic Actuators
Hans-Dieter Stolting
3.1 Introduction
3.1.1 Definition of Electromagnetic Actuators
Electromagnetic actuators transform electrical and mechanical energy into

one another using the electromagnetic-mechanical principle. In the follow-
ing small electric machines, especially motors, are described with a nominal
power in the range about up to 1 kW, so-called fractional-horsepower motors
[8]. In the simplest case a mechanical switch serves as a controller. For open-
loop and closed-loop control of motors more and more electronic components
are used, mechanically switched or triggered by hardware or software. In this
chapter the combination of motor and electronic control is called an actuator
or drive.
As there are many special designs besides the usual function principles
and types of electromagnetic machines, the range of fractional-horsepower
motors is marked by a great variety of different types. This is because of
their unusually great application possibilities (Table 3.1.1):
Most of the motors are used as drives for consumer goods. In order to be
very cost-effective these motors are adapted constructively as extensively
as possible to the driven appliance and to the most economical production
technology.
Besides these, there are high-grade motors for capital goods with special,
often extreme features.
Therefore small electromagnetic actuators are special designs needing the

close cooperation of the manufacturer and the user. Doing so, one is often
faced with the problem that actuator development starts at the moment when
the development of the apparatus is almost finished. But the actuator has to
be available before the manufacturing of the apparatus starts.
The variety of motors and their applications impede standardization.
There are partial agreements on fixing dimensions, cable colors, etc. between
user groups (e.g. oil and gas burner) and motor manufacturers.

86 Hans-Dieter Stolting
Table 3.1. Characterising features of small cheap and high grade motors
Cheap motors High grade motors

Integration in the driven Optimal electrical and
apparatus; motor components constructional adaptation to the
take on apparatus functions. driven apparatus
Largely automatic production Small-sized production: cutting
(> 1 mio pieces p. a.): operation, fitting together by
punch-bend-joint technique, screws
using components with High grade components:
commercial (standard) electrical sheet steel or low-loss
dimensions (magnets, sheet, rare-earth or AlNiCo
bearings, capacitor etc.), magnets, rolling bearing
greater manufacture Special characteristics relating
tolerance band, rough to efficiency (low energy
dimension stages (outer and demand, low temperature rise,
inner diameter, length), small weight), speed (extremely
great airgap, low space high or low, great speed range,
factor, very easy winding, constant speed, non-cogging
staved-enamel wire, shaft or/ and synchronous operation)
steps as few as possible, dynamic (low mechanic or/ and
collar bearing, etc. electric time constant),
Power per weight and efficiency positioning, overload capability,
often insignificant: semifinished live time, robustness, freedom
sheet (often without annealing) from maintenance, low noise,
instead of electrical sheet steel, low vibration, electromagnetic
ferrite magnets. compatibility (EMC),
insensitiveness to environmental
conditions (temperature,
vibrations, accelerations,
pressure, pollution (water-,
dust-, gas-proof), electric or/and
magnetic fields, explosion
protection)
3.1.2 General View
For guidance principles of design possibilities of small electric machines are

given in Figs. 3.1, 3.2 and 3.3. With regard to their basic characteristics they
can be assigned to two groups: self-commutated and externally commutated
motors.
Self-commutated motors are characterized due to the fact that the voltage
applied to their windings is dependent on the rotor position. From that, their
special characteristics are:
The maximal speed depends on the dimensions of the drive and not on the
frequency of the power supply. Due to the very high speed these drives
3 Electromagnetic Actuators 87
Self commutated motors with mechanical commutator
Motor DC motors AC motors

type Magnet Separate or Compound Series Series
excitation shunt excitation excitation excitation excitation
fJ 8 8 8 ~
v v v
Basic
circuit
Speed-
torque
curve
N(min-1 )
~ ~·
2000 ...20000
T
< 6000
T
N~
<6000
T
k <6000
T
k_ T
3000...30000
P(W) 0.2... 1000 0.2 ... 1000 20... 1000 20... 1000 10... 1000
7) 0.4...0.8 0.3 ...0.7 0.3...0.7 0.3...0.7 0.3...0.7
T,/TN 4 ... 10 4 ...6 4...6 3... 5 2... 5
TmaxJTN < 10 <6 <6 <5 <5
Untapped
3 brush Untapped winding
Speed winding
chopper chopper controller
control phase angle
controller if required with rectification and DC link
control
@ Commutator motor @ Single-phase or three-phase motor

o..
V 0 t'D'_~~\._, Permanent-magnet, hysteresis or reluctance motor
~
f"VV"\ fVVV"\ Series or shunt (parallel, separate) winding 1.1 Permanent magnet
0 Squirrel-cage winding CA Starting capacitor C6 Running capacitator
TN Rated torque Ta Locked-rotor torque Tmax Maximum torque

P Output N Speed TJ Efficiency VE.A Excitation, armature voltage
p Number of pole pairs
Fig. 3.1. Direct mains-operated self-commutated motors
can be built with low volume and weight. Therefore they are the ideal
drives for portable apparatus and tools.
Speed control is easy, cheap and often lossless.
The motors are expensive because they need brushes and a commutator
for switching over the windings.
Motors with a mechanical commutator have a shorter life time, and are
more trouble-bone and noisier than externally commutated motors.
88 Hans-Dieter Sti:ilting
External commutated motors

Asvnchronous motors
Motor AC motors
Three-phase
type motor Three phase Two phase
capacitor Capacitor Resistance Split-pole
squirrel cage auxiliarv windinn auxiliarv windinc auxiliarv windina motor
Basic
circuit
kW
®
~
®
UtJc Uu
~ ®~ @z
Speed-
torque
curve
N~
T
N~ p2
T
wit!:P
N----------
aUJ(ihary
._..,.(nding
/ With
T
Ntz T
N(min- 1 ) < 3000 < 3000 < 3000 < 3000 < 3000
P(W) >50 > 500 50 ... 2000 < 500 < 200
T) 0.5...0.8 0.3... 0.7 0.5...0.75 0.3 ...0.6 0.05 ... 0.35
T.ITN 1...3 1. .. 2 C~2 .. .4, C8 :1 ... 2 0.2 ... 1 0.2 ... 1
Tmax!TN 1.5... 3 < 1.5 < 1.5 < 1.5 < 1.5
Phase angle
Speed- Frequency control Transformer
control converter transformer
pole changing

Svnchronous motors
Motor
Three-phase AC motors
type motor
Permanent Permanent
hybrid rotor maanet rotor Reluctance rotor Hysteresis rotor
maanet rotor
Basic
circuit
l4J 1Juc
E)z
·~9'
Uu Uu ~"q
9 ~0'~ I I
Nt2 Nc c
I , _ _..II \._ ..../ '-~
Speed-
torque
curve
N(min- 1) :;:; 3000

T
:;:; 3000
T
~ 3000
T
l2 TZ~3000
/ __/
~':
T
~ 3000
T
P(W) > 200 >500 :;:; 50 0.1 ... 500 0.01 ... 100
T) >0.5 :;:; 0.6 .;;0.6 < 0.05 ...0.5 < 0.05 ... 0.4
T.ITN <3 - - 0.5...2 0.2 ... 2
Tmax!TN < 1.5 < 1.5 < 1.5 < 1.5 < 1.5
Fig. 3.2. Direct mains-operated externally commutated motors

Self commutated motors

Motor
type Permanent-magnet AC commutator
DC motors Brushless DC motors motor
1-t-8----u-----® !1·::~
!3- * <
! 1...4-
:.......................::-:::::}'·---- 1; ~
~
rectifier electronic switching using
Hall elements or optical phase angle control
1-~ encoders with thyristor
1.
FD sensorless switching
1~
transistor driven in its r··------~
linear or active region : : 1...4-, •
without FD : :.------
~--------z--------------·
:
2. transistor driven in the speed control by tacho-
ON-OFF mode (Pulse- generator (T) phase angle control
Control Width-Modulation, with Triac
circuits PWM) r·------~
FD: free-wheeling diode speed control by
L_ ______ l:::~~;~::::::l ___ _j tache-generator
j:~ position control using a

pulse generator
PWM servo amplifier (P, encoder)
full, H-bridge circuit
speed control by tacho- j:~ ._ _________________ ,!
generator or electric
governor resolver (R) for speed
position control using a control position control
pulse generator (encoder) and switching

Motor
type
Asynchronous motor Stepper motor
1-~ 1-tSlEJ--®
3-* ..
frequency converter
phase angle control with acceleration and
Triac capacitor motor deceleration by a given
pulse serie
Control 1-~
circuits 3- 3- *
frequency converter
three-phase motor
speed control by tacho-
generator
Fig. 3.3. Electronically driven motors

- Motors with an electronic commutator are as robust and low-noise as ex-

ternally commutated motors. But they always need a rotor-position sensor
and expensive electronics, which make them comparatively expensive.
Externally commutated motors are characterized due to the fact that the
voltage is applied to their windings forcible switched by the power supply in
a definite order. From that, their special characteristics are:
- The maximal possible speed depends on the supply frequency. It is
3000 rpm at a frequency of 50 Hz.
- The speed control is very expensive, so that a variable-speed actuator is
relatively costly.
- The motors are cheap, robust and low-noise.
3.1.3 Basic Construction Possibilities
There are more different layouts of small electric motors than those of larger
motors, especially more rotor-stator configurations (Fig. 3.4):
- Internal-rotor motors predominate because of the ease of production, sim-
ple fitting, good stator-winding cooling and the low moment of inertia
(especially motors with slim rotors).
- External-rotor motors are produced for special applications, e.g. fans with
blades directly mounted at the rotor core (Figs. 3.30, 3.70), winders or
motors with good rotational accuracy (low pulsating torque, Fig. 3.21).
- Intermediate-rotor motors (bell-rotor motors) with moving rotor coil
(coreless rotor) have the lowest moment of inertia of all electric motors,
low mechanical and electrical time constant and no cogging torque. They
are relatively expensive and can be built only for low nominal power be-
cause the self-supporting rotor winding is not sufficiently solid at large
rotor diameter and high speed (Fig. 3.9).
- Intermediate-stator motors with self-supporting stator winding consist of
a permanent-magnet rotor inside the stator winding and an outside yoke
rotating with the rotor. Hence there are no eddy currents and the rotor
has good rotational accuracy (Figs. 3.22, 3.24 and 3.25).
- Disc-type or pancake motors and linear motors are built in the same
above-mentioned stator-rotor configuration (Figs. 3.10, 3.11, 3.21, 3.24
and 3.25).
- Regarding the die set or the magnetic field pattern motors can be classified
in single-axle or two-axle symmetrical types (Fig. 3.5), whereby the flux
lines are radially, diametrically or axially straightened. The first are often
easier to be built in and better to be cooled; the latter are often produced
at low cost.
- Most of the motors are of heteropolar type with alternating poles around
the air gap, but there are homopolar motors too, with alternating poles
along the axis. Heteropolar motors predominate because of their better
Stator
*
·-
Rotor
Fig. 3.4. Rotor-stator configurations
power /weight ratio. Homopolar machines with a large number of teeth

around the air gap rotate with a small angle per voltage pulse (Fig. 3.49)
or are suitable if low rotor revolutions are required.
- Windings distributed in slots are on the one hand expensive, and on the
other hand these windings generate a better, sinusoidal field pattern and
therefore have lower losses and parasitic torques.
Concentrated windings on salient poles are cheap to produce, but elec-
tromagnetically unfavorable in asynchronous and synchronous motors, as
they generate a nonsinusoidal magnetic field distribution.
Slotless or air-gap windings generate no cogging torques and prolong the
commutation because the inductance is lower due to the larger air gap
(Figs. 3.9, 3.10, 3.11, 3.20, 3.22, 3.24 and 3.25).
~
a ....__-_--_-_--_-_-_-_-__, b
~+I~::~
L.___ _ _ __J
Fig. 3.5. Intersection symmetry. a Two-axle, b single-axle symmetrical cross-

section
In the range of small motors there are sometimes special arrangements of

motors, for instance mirror-image mounted motors with different direction of
rotation for reversing or two smaller motors on one shaft for increasing the
output power.
As every small motor type can be built in nearly all of the above con-
struction variations, the designer of these motors has a lot of possibilities to
fulfill the demands of every special application.
3.2 Self-clocked Motors with Mechanical Commutator

3.2.1 Introduction
Typical features of self-clocked motors are:

- The torque-generating magnetic field is excited by permanent magnets or
coils (excitation or field winding) in the stator.
- The torque-generating current is fed to the rotor (armature) winding by
brushes and commutators.
The commutator works together with the brushes as a switch which
changes over the current in the armature coils, so that the best conditions
for torque generation are always given. That means the current direction
and field direction have to be perpendicular to each other. With increasing
speed the brush noise grows louder and the brush wear is greater because of
switching sparks. Small DC motors have neither commutating winding nor
compensating winding. The brushes are shifted out of the neutral axis against
the direction of rotor rotation to improve the commutation. The cause of this
is that the zero crossing of the magnetic field curve, lying at no-load in the
geometric neutral axis, is shifted against the sense of rotation by armature
reaction. Backward brush shift by about 1-1 ~ slot pitch is recommended if
the rotor always rotates in one direction. To protect the commutator from
damage by worn brushes they sometimes have an automatic disconnection
which lifts off the brushes before their brush-lead or cap drag along the com-
mutator. After changing the brushes the motor can be driven again. The
radio disturbance of the brushes always requires considerable expense of ra-
dio and television interference suppression devices. Because of the pulsating
rotor flux the rotor is made up of axially arranged punched and slotted sheet
steels. Therefore the eddy currents of the rotor caused by the alternating ro-
tor field are less. The pros and cons of the different DC commutator motors
are compared with each other in Table 3.2.
3.2.2 Electrically Excited DC Motors
In the lower power range DC motors with wound fields are also produced, but
only with a small number of pieces because of their higher costs compared
Table 3.2. Feature comparison of the of DC motors
Excitation Permanent- Separate or Series Compound

magnet shunt
Speed-torque more load- more load- load-dependent more load-
characteristic independent independent independent
Locked-rotor greater greater smaller greater
current
Speed range smaller greater greater greater
with the same
costs
Speed control similar similar expensive similar
Efficiency with greater lower lower lower
the same costs
Risk of exists not applicable not applicable not applicable
demagnetising
Interference more expensive more expensive similar similar
suppression
Costs lower mean mean greater
with permanent-magnet DC motors (Table 3.2). Therefore these motors are

built if special application demands are required, e.g. very low or very high
temperature or a wide speed range (tapped stator winding).
The types of electrically excited DC motors are given in Fig. 3.1. Shunt-
wound motors have a field winding connected in shunt with the armature
winding. Their speed-torque characteristic is linear decreasing with increas-
ing load. The field and the armature winding of separately excited motors
are supplied by different voltage sources. For instance the field winding is
connected with the AC system by a full-wave uncontrolled rectifier and the
armature winding in addition to such a rectifier by a PWM servo-amplifier
with a transistor driven in on off mode (a chopper like Figs. 3.60 and 3.61).
They have the same speed-torque characteristic as the shunt motors.
Series wound motors have field and armature winding connected in se-
ries and have a varying-speed characteristic which is more load-dependent
than the characteristic of shunt motors. Their locked-rotor current is smaller
in comparison with the locked-rotor torque because the torque depends ap-
proximately quadratically on the current. The current/torque dependence
of motor types with constant flux is linear. Series motors can be connected
to an AC voltage source too. Then they are called universal motors or AC
commutator motors (see Sect. 3.2.4).
Compound-wound motors have both a series and a shunt field, and they
therefore have a speed-torque characteristic intermediate between those of
the above mentioned motors. Because of their high costs they are only seldom
produced. Figure 3.6 shows the operating curves of a compound-wound motor
with its low-decreasing speed torque curve.
I N
A rpm
0
0 5 10 15
T: torque, TN: nominal torque, N: speed, P: output power, I: current, T): efficiency
Fig. 3.6. Performance curves of a compound-wound motor
Ordinarily, fractional-horsepower DC motors built for household or in-

dustrial applications have mostly two poles like universal motors (Fig. 3.37).
However, aircraft DC motors often have four or more poles. For easier and
therefore cheaper manufacturing the stator assembly is laminated too. The
armature winding is a drum winding, usually a lap winding. The brushes are
made of a mixture of copper and graphite. Solid brushes with more copper
have a longer life time, soft brushes with more graphite keep the noise down.
In every case brushes and commutator have to be selected very thoroughly
by testing. The same applies to permanent-magnet DC motors too.
3.2.3 Permanent-magnet DC Motors
For the most part fractional-horsepower DC motors are permanent-magnet

excited motors, as they are simpler and cheaper than motors with wound
fields. They have permanent magnets in the stator structure to set up the
required flux and a commutator winding in the rotor. The different kinds,
shapes and mounting manner of permanent magnets offer much more motor-
design possibilities than windings. Permanent-magnet motors can be adjusted
to the driven mechanism especially well. Therefore they are available in a wide
range of power and revolutions. Predominantly these motors are produced for
extra-low voltage(::; 42 V) and used for battery supply because of their good
efficiency [2]. Here are some examples: auxiliary drives for cars (windscreen
wipers, cooling fans, fuel pumps, sunroof and window lift motors, seat po-
sition motors, door locks, electric steering assistance etc.), toys (e.g. model
ee~e
a b c d
f g
Fig. 3. 7. Low-cost constructions of DC motors. a, b Diametral magnetization, c,
d radial magnetization, e, f, g fiat designs
railways, cars and airplanes) , garden and household appliances (grass shears,
tooth brushes, coffee grinder, tin openers, spits, door pass-keysystems 1 ; heat-
ing valves, ventilation and air conditioning technology, electro bikes, con-
sumer electronics (CD-player, magnetic-tape drives 1 ), machine tools (test
stands for small electrical motors 2 ) , cableless hand tools , auxiliary drives in
grinding and welding machines 1 2 as well as lathes), film and video technology
(cameras, camcorders, projectors) , medicine and laboratory appliances (mas-
sage apparatus, wheelchairs, X-ray units, dental apparatus, stirring appara-
tus, centrifuges, shaking and sieving machines) , vending machines 1 (vending
machines for goods, change machines, gaming and slot machines, jukeboxes),
printers, printing machines 1 , passport scanners 2 , golf cars, pneumatic-post
systems, automatic packaging and label machines .
Most small DC motors are of the internal-rotor type (cylindrical outside-
field type). Cost-effective motors (Fig. 3.7) are usually two-pole and have
a housing made of rolled or deep-drawn sheet. The core of the slotted ro-
tor is laminated. Usually the slot number is even for easy winding produc-
tion (loop winding) . There are especially narrow designs (Figs. 3.7e, 3.7f and
3. 7g), in some appliances also with an axial stator yoke simultaneously used
as a bearing bracket (Fig. 3.7e). Only cheap anisotropic ferrite magnets are
used. This material is temperature-sensitive. It has a comparatively low re-
manent flux density and BH product, but it is difficult to be demagnetize
because of its wide hysteresis loop. Therefore the air gap can be wider. That
reduces production costs. The magnets are thin, thin-walled hollow cylinder,
in dovetail shapes or plats. Since the flux density is low the axial length of
ferrite magnets is often longer than the armature length to get more flux.
The magnets are usually fixed with clamps or springs, seldom by adhesive.
1 often bell-type motors

2 often disc-type motors
Magnet
Pole shoe
a b c
Fig. 3.8. High-grade constructions of DC motors
Sometimes they are encapsulated in plastic. Flexible magnets are also used,
which adapt themselves very well to the housing (Fig. 3.70). Mostly the com-
mutator is cylindrical. Flat commutators are used if the motor length is to
be as short as possible without employing a more expensive disc motor (see
below). Double-pole motors usually carry lap windings, and multipole motors
often wave windings because only two brushes are necessary.
High-grade motors are usually fitted out with rare-earth magnets (SmCo,
but increasingly NdFeB) and with AlNiCo magnets (Fig. 3.8). The latter
material is temperature-stable and provides a high flux density, but a low
coercive force. Because of that it is very endangered by demagnetizing fields
generated by high armature currents. Therefore AlNiCo magnets have to be
long in the direction of magnetization (Fig. 3.8b). To limit the demagnetiza-
tion the currents have to be as low as possible. On the other hand pole shoes of
sheet steel are used to conduct the demagnetizing armature-axis flux, so that
it cannot affect the stator magnet (Figs. 3.8b, 3.8c). Besides them pole shoes
concentrate the exciter field to achieve a higher flux density in the air gap.
Because of that it is disadvantageous that the armature time constant and the
commutation get worse. The air gap should be as small as possible. All of this
makes motors with AlNiCo magnets expensive and has to be considered while
designing them. Today the importance of AlNiCo magnets is decreasing.
Other magnets can be small. SmCo has both high remanent flux density
and high coercive force. So the BH product is very much higher than that of
AlNiCo. It is hard to demagnetize, but it is very expensive. NdFeB magnets
are more cost-effective and have an even higher BH product than SmCo mag-
nets. However, high-grade NdFeB magnets are very sensitive to temperature
and corrosion. Cheaper NdFeB magnets with a lower magnetic remanence,
approximately like SmCo, can be used up to 150 °C and are corrosion-stable.
The development of permanent material is not yet completed. In the future
the BH product should increase further. This will influence the design of
motors with permanent magnets and intensify their importance.
In motors with slotted armature the permeability varies according to the
rotor position. This causes flux oscillations and t hrough that torque oscil-
lations. These so-called cogging torques make the true running of the rotor
worse. There are possibilities to reduce cogging: nonconstant air gap, odd
slot number , slot skewing, optimized width and shape of tooth tip, diagonal
magnetized or rhombic shaped magnets et c. These actions make motors more
expensive and do not suppress the cogging completely.
A winding arranged uniform on an unslotted armature core or self-
supporting winding (moving coil) in the air gap hardly creates torque os-
cillations. The latter are carried out as bell or disc shaped winding. The
inductivity of these motors is lower than that of motors with slotted rotors
because of their wider air gap. Therefore the commutation is better and the
electrical time constant is lower. The pot shaped rotor of bell-rotor motors
rotates around the internally placed double-pole magnet . Occasionally there
are four-pole motors. Figure 3.9 shows an exploded view of a bell-rotor mo-
tor with a cut-out housing part. The rotor winding is carried out as a skew
winding (Faulhaber winding) or as a rhombic winding (Maxon winding) . The
solid housing is used as the magnetic return path. These careless motors are
of the intermediate-rotor type (cylindrical-inside-field type). As the ironless
bell rotor has a very low moment of inertia, the mecha nical constant is low.
Therefore motors with such rotors have the best dyna mic properties of all
motors and are especially suited for control and servo drives as well as in ap-
pliances which require good true running. Commutator and brushes are made
of precious metal or are silver plated to reduce the contact resistance. AlNiCo
and now increasingly NdFeB magnets are used . These motors are produced
with an output power up to 100 W , occasionally with 250 W . Larger motors
are not able to be built because of the mechanical stability of the bell rotors .
Other coreless motors have flat rotors . Very small motors have four ferrite
magnetic poles in the stator, three flat self-supporting air-core coils and often
a flat commutator (Fig. 3.10). Motors with a power up to 200W have eight
or ten ferrite poles and a plastic disc with punched or printed conductors
on each site brazed together at the inner and the outer site. The conductors
form a wave winding, so that these motors need only two brushes. They are
made of graphite and axially arranged sliding on the inner part of the wind-
ing (Fig. 3.11a). Larger disc motors which are also known as pancake motors
have flat coils imbedded in plastic and eight or more poles of ferrite, AlNiCo
Rotor winding
Brush
Stator magnet
Fig. 3.9. Exploded view of a bell-rotor motor

or NdFeB magnets in the stator (Fig. 3.11b). Normally the commut ator is
cylindrical, but seldom are there fla t commutators. Because of their large
outer diamet er disc motors with an output power up to 500 W have a m e-
chanical time constant greater t han that of yc lindrical outside-field motors
with slotted rotors.
Magnet
Commutator
Fig. 3.10. Flat DC motor with an outer dia meter of 20 millimeters
a b Commutator
Fig. 3.11. Disc motors
Table 3 .3 g ivessome characteristic dat a of t he above described motors

and Fig. 3.12 shows the most important performance curves of a car auxiliary
drive which a re typ ical of all permanent-magnet mot ors.
The s peed r egulation of a p ermanent-magnet motor i s made byseries
resistance (additional losses, cheap) or by a n electronic cont roller (chopper :
low-loss, expensive). Motors of windscreen wipers somet imes have a t hird
brush, so t hat the armature winding is connected eit her with t he two norm al
brushes or with one of t hese brushes and t he t hird brush . In t he la tter case
Table 3.3. Typical catalogue data of permanent-magnet motors

Nominal Nominal Nominal Nominal Effi- Inertia Weight
voltage speed torque output ciency
power
v rpm Nm w % gcm 2 kg
Bell type 12 5000 0.02 6 72 16 0.13
motors 12 5000 0.05 20 78 47 0.27
12 12000 0.001 0.6 73 0.4 0.018
12 12000 0.0025 2 76 0.7 0.02
Disc type 24 3000 0.032 10 50 400 0.45
motors 24 3000 0.16 50 68 2100 1.3
24 3000 0.61 200 68 2900 3.9
24 3000 1.27 400 70 7900 6.5
24 3000 3.2 1000 82 10000 9.7
Servomotor 24 3000 0.064 20 55 120 0.58
24 3000 0.16 50 70 400 1.2
24 3000 0.32 100 76 950 2.1
24 3000 0.64 200 78 2900 3.6
24 3000 1.59 500 83 12000 8.8
Cost- 12 14000 0.0007 1 50
effective 24 14000 0.0034 5 55
motors 30 14000 0.0068 10 65
Car 4000 0.12 50 55 0.8
auxiliary 4000 0.24 100 60 1.8
drives 12 4000 1.19 500 65 4.0
or 4000 2.39 1000 65 4.0
24 12000 0.008 10 35 0.08
12000 0.04 50 55 0.25
12000 0.08 100 60 0.5
Pulsating- 220 7000 0.068 50 55 0.42
current 220 7000 0.136 100 57 0.7
motors
with AC
terminal
and recti-
fication
the commutation is worse (three brushes motor [3]). This method should be
used only for a short time.
3.2.4 Pulsating-current DC Motor
Pulsating-current motors are permanent-magnet DC motors which are op-

erated with rectifiers at the 230 V AC system without special smoothing
elements. The current ripple amounts to more than 10%. Table 3.4 explains
28 7000 .----r----~--~-----r----~----.----, 70
I _l:L p
A rpm
24 6000 r-~-r-+~~---r----+----+----1---~vv
20 5000 ~-~~~-~~~~~~~---~----~--~50
"'%
16 4000
12
4 10
0 0
5 TN 10 15 20 25 30_I_35
Ncm
T: torque, TN: nominal torque, N: speed, P: output power, /: current, fJ: efficiency
Fig. 3.12. Performance curves of a permanent-magnet DC motor
N I
rpm .; ....······· A
0.8
5000 I .; .;
..········
...··
..; . ...········
.; ..·· 0.7
IE-
4000
to=········
1\'. ·········· .... ~ .. .;
..·· ..···
...··· = Direct current 0.6
···v. *
.;
Diode
... ...·· 0.5
3000
~.\' ·, ····· ~- .,· ....······ ~ Uncontrolled
······· bridge
;;-..:, ....······· ~--....
.; 0.4
. .;
\
\
\ \. /~ ·········· .... ..
~Controlled
bridge(so·)
2000 .; ...···
':o.
... ... 0.3
\~ '\.····· ... ··········
... ·········· 0.2
.·\····· ········
.... ,
~
"""
1000 ~ ... ·..... N ··········
0.1
....··· \ ....
0
~ ~ ...... 0
2 4 6 8 10 12 14 16 18 20 T 24
Ncm
T: torque, N: speed, /: current (root-mean-square-value)
Fig. 3.13. Performance curves of pulsating-current DC motors with different elec-

tronic circuits
why these motors are not employed today to such a degree, as was expected
some years ago. They are used in easy household appliances, especially in
those which have a heating coil as series resistor (hair-drier, foil welding rna-
chine). Dependent on improvements in electronics and magnetic materials

it is to be expected that its field of application will extend considerably in
the future. Table 3.3 shows typical data of these motors. The easiest rec-
tification electronics need only one diode. Today two-pulse bridge connec-
tions are used even in simple appliances because they render possible better
motor utilization and they are cost-effective as semiconductor devices. Fig-
ure 3.13 shows speed-torque characteristics with various control methods
compared with DC supply. The curves for the uncontrolled and the con-
trolled bridge mark the region where the motor speed can be regulated by
a phase-angle control (Sect. 3.6.4). The latter curve applies to a timing angle
of a= 90°.
3.2.5 AC Commutator Motor
The AC commutator motor (universal motor) still has great importance in

the range of small-drive engineering. A8 it can be built with high 8peeds (up to
40,000rpm), its power/weight ratio is very favorable. Besides them the speed
can be varied in a wider range and more easily than every other motor type
compared to the costs. It is the most important drive of portable appliances
like household appliances, hand or dish tools: food slicer, mixer, coffee grinder,
washing machines, vacuum cleaner, sewing machines, garden pump, hedge
clippers, drilling and grinding machines, saws, milling and planning machines,
hot-air blower, spray-guns, high-voltage switches.
To easier appliances motors with pulsating current can be used. There-
fore in Table 3.4 typical characteristics of both commutator motors of the
Table 3.4. Comparison of the characteristics of DC and AC commutator motors

of the same size
Pulsating current DC AC commutator motor
motor
Torque identical
Production cost identical
Losses identical
Speed characteristic more load- independent very load-dependent
Speed control easier more expensive
Speed-regulation range at the smaller wider
same cost
Starting torque lower greater
Commutation better, longer brush life worse
Rectification at 230V-mains necessary not applicable
operation
Interference suppression more expensive easier
DC component in a fault case special protective not applicable
Demagnetizing danger actions if need not applicable
same size are compared. The wide speed range of AC commutator motors
is used in washing machines, where the speed difference between wash and
spin-dry amount up to 1:60. Its high starting torque is advantageous for some
appliances like stirring apparatus, drilling machines and switches. In contrast
to large DC series motors small universal motors are not able to reach un-
permissible high revolutions at unloading because of the relative high brush
friction.
AC commutator motors are always double-pole built, mostly with a double-
sided symmetrical section (Fig. 3.14a), seldom with a one-sided symmetrical
section (Fig. 3.14b) with the characteristics referred in Sect. 3.1.3. The two
coils of the field winding are situated before respectively behind the armature
winding. Because of that they act as interference suppression coils. Supplied
by a rectifier they work as a smoothing reactor improving the commutation
and because of that the life time of the motor. Table 3.5 puts together char-
acteristic data of universal motors. Figure 3.15 shows the performance curves
of a 500W-pump motor.
In former times tapped field windings were used for supplying universal
motors either with direct current or with alternating current (Fig. 3.16a).
The speed can be regulated
lossless by tapped field winding (Fig. 3 .16b),

- covered with losses by a series resistance or/ and a resistance parallel to
the armature winding (hardly used today, Figs. 3.16c, 3.16d),
lossless by variation of the terminal voltage.
Table 3.5. Characteristic data of AC commutator motors

Nominal speed Nominal Nominal output Efficiency Weight
torque power
rpm Nm w % kg
4000 0.12 50 40 1.5
4000 0.24 100 55 2.2
4000 0.48 200 65 4.5
6000 0.16 100 63 2.0
6000 0.32 200 66 3.0
6000 0.80 500 80 4.5
8000 0.12 100 65 2.0
8000 0.12 200 67 2.3
8000 0.60 500 80 3.0
10000 0.048 50 60 1.0
10000 0.095 100 67 1.5
10000 0.19 200 70 2.0
15000 0.255 400 75 1.1
15000 0.51 800 80 1.8
20000 0.334 700 75 1.5
20000 0.477 1000 80 2.4
Rotor or
armature winding
a b
Fig. 3.14. Electrically excited commutator motors
Tapped windings are used if a wide range of speed regulation is desired

(washing machines). Nowadays variation of the terminal voltage takes place
by phase-angle control (Sect. 3.6.4). In the future, universal motors may be
run by pulse-width modulation (PWM) both rectifying the motor current and
varying the motor voltage. Today this is expensive compared with phase-angle
control.
I p1 .Ji..
w
"
.!L
A rpm %
6 600 12000 k::::: ~
....... ...::::: /(
......... p1
5 500 10000 ~ 100
'> ~ P.-"
--
...........
-
4 400 800 80
......... .z;:::.':JI""
17
r--
~v
3 300 600 60
2 200 400
200 ~
v 40
0
100
0 v
0 0.1 0.2 0.3 0.4 0.5 TN 0.7 0.8 0.9 1.0 ..L..
0
20
Ncm
T: torque, TN: nominal torque, 17: efficiency, P1: input power, N: speed
Fig. 3.15. Performance curves of AC commutator motor

a b T
~N~
c T d T
Fig. 3.16. Speed regulation of universal motors. a Supplementary DC winding, b

field tapping, c series resistance, d stator parallel resistance
3.3 Self-clocked Motors with Electronic Commutator

3.3.1 Brushless DC Motors
Electronic motors have a polyphase winding in the stator and a permanent-

magnet or reluctance rotor. An electronic control switches the stator phases
in cyclic order dependent on the rotor position. Therefore these motors are
also designated as EC- (electronically commutated) motors or BLDC- (brush-
less direct current) motors [2, 4, 5]. They are robust and low-noise just as
asynchronous motors. They have a better dynamic performance than DC
commutator motors. They are applied not only in the field of capital goods
but above all permanently widening in the field of consumer goods, for in-
stance: tape and video recorders, hard-disc drives, camcorders, digital cam-
eras, plotters, printers, laser reflectors, optical storages, analysis and dosing
apparatus, bar-code scanners, robotics, winding and feed drives, welding ma-
chines, electro-cardiographs, dialysis apparatus, artificial-limb control, small
fans, car auxiliary drives. Meanwhile there are increasing applications in the
range of simpler apparatus. The constant development in the field of electron-
ics and of magnet technique reduces on the one hand the cost and improves
the properties on the other, so that new cost-effective designs, for instance
with stators according to the claw-pole principle, are possible (Fig. 3. 70).
More and more DC commutator motors are being replaced by BLDC motors.
In comparison to DC commutator motors the construction and the tasks
of stator and rotor are changed (Fig. 3.17) but without influence on the fun-
damental method of operation and performance. The stator is the armature;
the rotor excites the magnetic field. Often BLDC motors are drive alterna-
tives to synchronous (stepper- )motors. In contrast to them
they cannot pull out of synchronism because the stator phases are
switched just when the rotor position encoder indicates that the rotor
has reached the corresponding position,
their speed is load-dependent and not constant (shunt characteristic) ,

they are expensive because of the electronic commutator, and
their vibration ability depends on the resolution of the rotor position
encoder.
Fig. 3.17. Fundamental constructions of mechanically and electronically commu-

tated motors
For reducing torque oscillating conventional DC motors often have more

than three coils in the rotor, as shown in Fig. 3.17. Twelve up to 32 coils
are typical. Because of the cost at most four coils are used in BLDC motors
since the phase number of the control electronics has to be as great as the
phase number of the winding, essentially determining the cost. The electronic
commutation can happen in two ways. Either the phases are switched on
and switched out or their current direction is changed. In the first case the
current flows alternately in opposite directions in two coils per pole belonging
to one phase (unipolar connection). Here the control electronics is cheaper.
In the second case the alternating current flows only through one coil per
pole energizing alternately in the north and south pole (bipolar connection).
Now the motor efficiency is higher. Therefore most modern BLDC motors
are bipolar driven.
The driving of the winding phases is mostly done by Hall elements. They
are built in, so that they are excited by the rotor magnet itself or by a special
control magnet turning round with the rotor. It is also possible to evaluate
the induced voltage of the winding phase which is not used at this moment
(sensorless control). Further possibilities for determining the rotor position
offer pulse generators which supply the electronics with a certain number
of impulses per turn (magnetic or optical incremental encoder) or a definite
impulse code according to each rotor position (code disc). The latter method
detects the rotor position at standstill, too, so that the electronics are able to
start the motor automatically. In the other case a special starting connection
is required taking over the control, until a reset pulse of the rotating slit disc is
given to the counter which now begins to count. After every turn a reset pulse
is generated. The speed control is mostly made by a tacho-generator and -
with some restrictions relating to the accuracy- by evaluation of the induced
voltage of the just not used phase. Besides them resolvers are used which
106 Hans-Dieter St6lting
register the rotor angle in an inductive way. As with the above-mentioned

sensors, they are not only suited for winding switches but for positioning and
control purposes, too.
BLDC motors are able to reach more than 30,000 rpm, to cover a speed
range up to 1:3000 and to generate torques between 0.02Nm up to 100Nm.
Their life time amounts to several 10,000 hours. Typical data are given in
Table 3.6.
Table 3.6. Typical data of BLDC motors

Nominal Nominal Nominal Nominal Efficiency Inertia Weight
voltage speed torque output
power
v rpm Nm w % gcm 2 kg
24 5000 0.02 10 60 30 0.5
24 5000 0.057 30 65 70 0.9
42 3000 0.032 10 65 20 0.3
42 3000 0.095 30 65 200 1.0
42 3000 0.32 100 70 1400 2.5
230 3000 0.095 30 40 200 1.0
230 3000 0.32 100 45 1200 3.0
230 3000 3.18 1000 65 2000 5.0
230 5000 0.38 200 50 400 2.0
230 5000 1.15 600 60 600 3.0
The simplest BLDC motors have only one winding phase in the stator
which gets one pulse every turn generated by a Hall element (Fig. 3.18). The
useful torque is so low and it pulsates so intensively that it is used only for
simple appliances, for instance as a motherboard fan. The torque becomes
better if the winding gets two pulses per turn, so that the current flows
alternating in both directions (bipolar connection, Fig. 3.18c). This needs
four transistors. Two winding phases flowing through alternatively (unipolar
connection, Fig. 3.18b) are cheaper. In both of these cases the generated
torques do not overlap each other. That means that the resulting torque
disappears at the electrical angles of 0° and at 180°. Stopping at these points
the motor is not able to start self-supporting. Therefore an auxiliary torque
is superposed which is different from null at the above angles. Figure 3.19
illustrates the two possibilities to generate such a torque at examples of four
pole motors: the left side shows a motor with unsymmetrical poles exciting an
additional reluctance torque; the right side has no wound auxiliary poles and
unsymmetrical magnetized rotor. The connections (b) and (c) in Fig. 3.18
need only one Hall element. For one current direction the positive Hall pulse
is used, excited by the north pole of the rotor magnet, and for the other
direction the negative Hall pulse is used, excited by the south pole.
,]: W
H
R
T
Phase winding
Hall element
Resistance
Transistor
f g
unipolar bipolar
Fig. 3.18. Connections of BLDC motors. a One-pulse connection, b, c two-pulse

connection , d, e four-pulse connection, f three-pulse connection, g six-pulse con-
nection
0: not magnetized
Fig. 3.19. Two-phase motor with auxiliary torque

Four-pulse connections (Figs. 3.18d, 3.18c) need two Hall elements. As in

the two-pulse cases each winding generates a torque over an electric angle
of 180°. But now there are four torques which superpose each other. The
useful torque increases and the pulsating torque decreases. These motors were
used in former times when the cost of Hall elements and the accompanying
electronic circuit are very expensive. Figure 3.19 shows a four-pulse motor
with an air-gap winding describing the position of the four phases. The phase
numbers correspond to the numbers in Fig. 3.18. The phases are supplied in
the sequence 1, 3, 2, 4. Motor production is expensive. First the rotor is put
into a plastic cylinder. Then the four phases are wound axially round the
cylinder. Finally both parts are put together into the housing which consists
of a laminated core within an aluminium cylinder.
Three-phase motors are supplied with three or six pulses (Figs. 3.18f,
3.18g). The first one is only seldom used for simple actuators. Most important
is the latter type. It has a higher useful torque and an about 13% lower
pulsating torque than that of four-pulse motors. As electronic circuits and
Hall probes are cheaper than in the past, four-pulse motors are hardly used
today. Besides of the star connection in Fig. 3.18g the motor can be delta-
connected. The choice of the connection depends on the supply where it is
to be considered that in delta connections loss may arise generated by ring
currents if there are asymmetries in this circuit.
EC motors are produced with an internal rotor or with an external rotor
if this offers constructive advantages, if a good true-running is demanded or
if the magnetic material needs as much space as possible. Figure 3.21 shows
a motor with salient poles which may generate cogging torques. In slotless
round-frame-type motors a winding with axial coils or winding known as
Faulhaber or Maxon winding (like Fig. 3.8) can be fixed in the stator, divided
into three star-connected parts and directly driven via transistors. These
motors are produced in two types. The first one has a la minated stator core
in which the winding is pasted in like the motor in Fig. 3.20, but without
a plastic cylinder. This construction is expensive (Fig. 3.22 at the top). The
second one has a self-supporting winding. Outside of the winding a massive
magnetic return path rotates with the inside rotor (intermediate-stator type).
Fig. 3.20. Four-pulse motor with a ir-gap winding

Magnet pole
Coil
Fig. 3.21. Three-phase motor with external rotor (hard-disc actuator)
Because of that no eddy currents are generated. This construction is much

cheaper, but has a much higher moment of inertia. This is necessary if a drive
with good true running is required (Fig. 3.22 at the bottom).
In Fig. 3.23 there are given construction possibilities of BLDC disc motors
with ironless coils or self-supporting windings. Very small disc motors are
made with photolithographic windings. The stator has three phases with two
flat coils each. The rotor has eight poles. Figure 3.24 shows the so called
"penny-motor", a new micro motor. In this motor the return path rotates
together with the rotor, so that there are no eddy currents (see Fig. 5.64).
With wave-formed concentric conductors embedded in plastic the expen-
sive interconnection of coils is avoided, but on the other hand the manufacture
of this winding is more difficult and expensive (Fig. 3.25). Larger disc mo-
tors have a helically wound stator core often without slots. To reduce the
Stator winding
End shield Laminated core
Flux return path
Fig. 3.22. Exploded view of slotless BLDC motors

llO Hans-Dieter Stolting
EmB Winding
Magnet
c::::J
Fig. 3.23. Constructions of disc motors with radial and axial magnetic fields
Shaft
-- Rotor
Nonmagnetic stator
Fig. 3. 24. Disc motor ( "penny motor" )
Return path Stator wind ing Magnet
Fig. 3.25. Disc motor with waved-formed coil

axial force the stator is divided into two parts on both sides of the rotor
(intermediate-rotor type). The rotor has to be constructed to be distortion-
resistant to resist the stress caused by high pulse currents. This and the
essentially handmade manufacturing are the causes of the very high cost.
Because of the cost most small linear actuators are still produced as a ro-
tating motor with a gear, for instance with a gear rack, a toothed belt or
a thread bar with a nut. But today linear direct drives are appearing. The
disadvantages of gears (play, noise, losses) are avoided, but the size of the
drive becomes greater. The stator or the sliding element have to be as long
as the drive way. If the motor is not too small an air-lubricated bearing car-
ries the short slider with the field winding. It is necessary that the air tube
is fixed on the slider just as the electric connecting cable. This causes an un-
desirable torque which has to be taken in consideration if the motor should
be driven controlled. Very small linear BLDC motors have the field winding
in the stator, a permanent-magnet slider and a micro-ball bearing. All these
motors can be built with a two-axis motion (xy-motion), so that they are able
to follow any path. Besides them it is possible to realize three-dimensional
motions (xyz-motion) by combination of several motor systems. In addition,
a rotating element can be added [8]. The advantage of these motors com-
pared to voice-coil motors (Sect. 3.5.3) is that they can be driven in a larger
range.
3.3.2 Servo Motors
In the lower power range there are pilot or servo drives too. Their task is
to position mechanical elements at a given position within a given time and
with a given precision. These tasks are given in feed screws and feed drives
in machine tools 1 , in robot systems 1 , in conveyor belts 1 , smaller antennas,
smaller telescopes, butterfly valves, valves. Often several motors have to run
in synchronism. On servo drives heavy demands are made relating to the
dynamics, the positioning exactness, the peak torque (up to the fourfold
nominal torque), rotational trueness (especially at crawl speed), the speed
regulated range (up to 1:10,000), the efficiency, explosion protection, pro-
tection from environmental influences, etc. They are valid not only for the
motor but also for the other drive components, like power amplifiers, sensors,
control electronics and mechanical transfer elements. To reach an optimal
control dynamic the time constants all of the assemblies have to be as low
as possible. Often these motors are used in four-quadrant operation which
means the motors can be driven and broken in both directions. In the lower
power range electromagnetic actuators have prevailed against hydraulic and
pneumatic drives. The causes are given in Table 3.7 (see Sect. 4.8).
Some years ago only DC commutator motors were used as servo motors
because of their easy adjustability. Even though in the lower power range
1 Mostly motors with an output power greater than 1 kW
Table 3. 7. Comparison of hydraulic and electrical servo drives
Hydraulic drive Electromagnetic drives

higher energy density better control response
smaller mounting space higher efficiency
higher accelerating ability easier maintenance
lower cost easy adaption to different condi-
easy generation of linear motion tions
problem: leakage the same energy type for sensors,
control electronic and actuators
many servo actuators are still DC motors, today three-phase drives with elec-
tronic commutators increasingly prevail. As in DC motors with a constant
exciting field the torque depends linearly on the armature current, so the
regulation and control by the current is very easy. The advantage of a DC
commutator motor is that only one current value has to be measured and
controlled. Therefore the control electronics are simpler. Should the moment
of inertia be low, ironless bell-rotor motors are used up to an output power of
lOOW; at powers more than 20W ironless disc or (should the occasion arise:
slotless) bar-type rotors are used. Ironless rotors have especially low mechani-
cal and electrical time constants (see above). The rotors of three-phase motors
with a permanent magnet or with a squirrel-cage winding have a greater mo-
ment of inertia. In Table 3.8 the characteristic differences between DC motors
with and without brushes are listed. Figure 3.26 illustrates the performance
range of BLDC motors, enlarged relative to DC servo motors.
Table 3.8. Comparison of DC and three-phase servomotors
DC current drives Three-phase drives

- easy supply and control unit by one - control at least by two currents and
current capture of the rotor position
- easy servo-amplifiers -expensive converter, higher switch-
ing capacity
- noise and wear of brushes and com- -maintenance-free, robust, low-noise
mutators
- brush life up to 3000 h - brush life several 10000 h
- dynamic limited by commutation - higher protection grade
- heat dissipation over the shaft - ironless motors: essential lower mo-
ment of inertia
-at voltages higher than 400 V -lower weight, smaller size
matching transformator is necessary
- direct mains-operation
- more expensive
6 6 --,
T -Raffi:earth-ma9ne1s------ T
I
\ Rare-earth magnets
rN . . . . . ., TN
I
', Dynamic
I
I
I
4 : extreme 4 I
I
I range \
I
3 Ferrite magnets J
I
2 ~:;;:~· S3-25% '77T."77T."77T."77T.~\ Intermittent

~ . S3-40% / periodic
Vm77777777.S1 duty
Continuous
O+---~--+----<--~--duty
0.2 0.4 0.6 N 1 0.2 0.4 0.6 N
a NN b NN
Fig. 3.26. Performance range of servo motors. a BLDC motor, b DC motor; 25%,
40%: cyclic duration factor
In Table 3.9 AC motors (three phase induction motor) and BLDC motors
are compared. Both of these motors are dynamically equivalent. It is not
clear generally which of the two is more favorable. Is a higher speed range
wanted; is the AC motor the better one because of its larger field weakening?
If a higher positioning accuracy i8 required a BLDC motor is the better one.
In three-phase AC motors with squirrel-cage rotor the torque causing cur-
rent is generated inductively in the rotor. They need neither high-grade mag-
nets nor commutators or 8lip rings. Their speed i8 close to the synchronous
speed which depends on the frequency of the pole number and the electronic
control. Therefore the speed can be regulated by the frequency (see Sect.
3.4.1). The worse efficiency and power factor, in comparison with BLCD mo-
tor8, require a more powerful and more expensive supply unit. Therefore in
drive8 lower than 1 kW AC motors are 8eldom employed.
Figure 3.27 show8 the control of a servo commutator motor (m). The cir-
cuit exists of three in one another inserted control loops (cascade-circuit).
Table 3.9. Comparison of BLDC and AC motors
BLDC drive Three-phase drive

Robustness high very high
Overload capability high very high
Field weakening only restricted possible in a large range
Ohmic loss in the stator in the stator and in the ro-
tor
Reactive-power demand not applicable necessary
Size at the same power smaller higher
Motor cost high low
Control easier, cheaper large-scale, expensive
Total cost high very high
Three-phase system
L1 Lz L3
Fig. 3.27. Control circuit for servo drives
The desired position (setpoint position) is given by a system, the actual

value by an encoder (e). The close-loop position control determines, with
a setpoint/actual-value comparison, the desired speed N 8 • Comparing this
with the actual speed value Na measured by a tacho-generator (t) results
in the desired motor current is which is adjusted by the inner control loop.
A six-pulse bridge supplied by a three-phase system yields the direct current
ud dependent on the control signal Uc. The control circuit of an electronic
commutated motor is like this, but there at least two currents have be mea-
sured.
3.3.3 Switched Reluctance Motor
Besides the EC motors with permanent-magnet rotor, described up to now,

there are, however, very rarely, motors with reluctance rotors. Commonly sta-
tor and rotor are laminated because of the heavy flux pulsations. The motor
in Fig. 3.27 has three phase windings which are arranged on two opposite sit-
uated poles. The phases are star- or delta-connected. Dependent on the rotor
position the voltage is applied to them in a cyclic succession 1 . The switching
happens by three Hall elements which are displaced by 120°. The unwound
rotor follows the revolving field trying to put itself in such a position which
minimizes the magnetic resistance (reluctance). This is the case if two rotor
poles are in line with the just excited stator poles. As the two rotor positions,
drawn in Fig. 3.28, illustrate, the rotor rotates with half of the field speed
and opposed to it. Besides three-phase motors, four-phase motors with eight
stator poles and six rotor poles are suggested.
The motor is very robust and frequently used. That makes it interesting
and one hopes that it would prevail against the classic motor types. But there
are serious disadvantages. During the rotor motion the magnetic resistance
varies. Therefore the flux is fluctuating. With it the torque pulsates causing
great noise and vibrations. Many suggestions have been made to minimize
this, for instance by duplication of the number of poles per phase, special
1 If the phases are switched independent of the rotor position it concerns a syn-
chronous motor (see Sect. 3.4.2).
Hall
element
Fig. 3.28. Switched reluctance motor
shapes of the poles or by current control. All these methods make the motor
more expensive or reduce the utilization. A further problem is the high in-
ductivity of the winding which makes switching of the phase winding more
difficult. Therefore special measures have to be taken to accelerate the dis-
connection, for instance by current control. Otherwise currents are flowing in
several phases simultaneously. This generates braking torques. These are the
reasons why this motor type has not prevailed up to now and why it is not
suitable as a servo motor. Today it is used only in niches applications.
3.4 Externally Commutated Motors

3.4.1 Asynchronous Motors
Appliances and Design
Small asynchronous motors (AC motors, single-phase (induction) motors)

are mostly manufactured for operation with 230 V alternating voltage (AC
voltage), more seldom with 400 V three-phase voltage [1]. In the following
there are put together exemplary apparatus which use the typical charac-
teristics of these motors (cost-effective, robust, maintenance-free, low-noise,
low-vibration) and no speed regulation or at most a simple one is required:
household and garden apparatus (dishwashers, washing and ironing ma-
chines, clothes dryers, refrigerators, freezers, fan-assisted ovens, air dome 1 ,
garden chaff cutter, lawn mover, oil and gas burners, heating pumps and
other pumps, roller blinds, sun blinds, garage doors, fans 1 , stable fans), of-
fice machines (typewriters, copiers\ fax machines, printers, paper shredders,
apparatus fans 1 , coin counter machines), tool machines (circular saws, high
frequency tools, cement mixers, high pressure cleaners, compressors, grind-
ing machines), laboratory and medicine technique (stirring machines in ther-
mostats and cryostats 1 , analysis apparatus 1 , recording instruments 1 , air-bed
Laminated core
Stator winding
Rotor bar
Fig. 3.29. AC asynchronous motor, capacitor motor
Fan impeller
Fig. 3.30. Asynchronous motor with external rotor, radial blower
pumps 1 , incubators, pumps in electrocardiogram apparatus 1 ), picture and

sound technique (film projectors, cut dishes\ juke boxes, scanners).
1 often external-rotor motors

Stator and rotor are laminated, to minimize the eddy current losses gener-
ated by the pulsating magnet fields. Besides the most common internal-rotor
motors (Fig. 3.29) there are external-rotor motors. The latter are employed
among other things for fans because of the smaller axial length and the higher
fan efficiency. The fan blades are mounted directly onto the rotor housing
(Fig. 3.30). The rotor housing of winders can be used to roll up the mate-
rial directly. The stator winding being of two or three phases is distributed
mostly in slots. Only very small motors have concentrated coils on salient
poles (Fig. 3.31). The rotor has slots with one uninsulated bar each. All the
bars are connected at the end sides by rings (squirrel-cage winding). In a few
cases very small motors are equipped with massive iron cylinder or massive
bell rotors, possibly the latter as external rotor. The maximum torque of these
motors is the start torque. Ferraris motors with an aluminium or copper bell
rotor are suitable as positioning actuators, but today are hardly produced
because of the better suitable BLDC motors (see Sect. 3.3.2).
Fig. 3.31. Sections of four-pole double-phase asynchronous motors
Method of Operation and Performance Characteristic
The stator phase currents generate a magnetic field which rotates with the
synchronous speed
f
Ns = -
p
where
f line frequency
p number of pole pairs.
The revolving field of a two-pole motor rotates at f = 50Hz with N 8

3000rpm, the field of a four-pole motor with N 8 = 1500rpm, etc. When the
rotor turns round with a speed different from the synchronous speed, the field
induces voltages in the rotor bars and with it currents. The rotor currents
generate the torque together with the rotary field.
Increasing the speed the power /weight ratio can be improved; that means
the weight can be reduced at the same power. This is used for so-called high-
frequency tools. These are portable hand tools in factories (drilling and grind-
ing machines, etc.). Their motors are driven with frequencies from 200Hz to
400Hz, so that speeds are reached from 12,000rpm to 20,000rpm like uni-
versal motors (Sect. 3.2.5). Advantages related to universal motors are the
special characteristics described in the previous section. A disadvantage is
the worse possibility of speed regulation. In addition, an expensive frequency
converter is necessary.
Three-phase motors possess the best power/weight ratio of all AC motors,
but they are produced economically with an output power more than 50 W.
Their locked-rotor torque and their breakdown torque (maximum torque)
are about twice as great as their nominal torque. With a 400 V line generally
the winding is star-connected (Fig. 3.32b). Apart from the appearance of
parasitics the field rotates with a constant quantity (circle rotating field).
A typical speed-torque characteristic shows the curve (a) in Fig. 3.33. It is
a motor with a relatively high rotor resistance (high-resistance squirrel-cage
motor) to get a start torque as great as possible.
A three-phase induction motor can also be connected to an AC voltage
system, linking two terminals to the system directly and the third in series
with a capacitor. Using an existing, usually star-connected 400 V three-phase
motor it is to be delta-connected to a 230V system (Fig. 3.32c). As in this case
indifferent currents flow in the winding phases in contrast to a three-phase
motor, the torque pulsates during the revolution (elliptical rotating field). It
is pulsating with double the system frequency, so that this oscillating torque
superposes the useful torque which is constant at a certain speed. This os-
cillating torque causes noise which is typical for AC motors. The torque of
a single-phase motor with three windings is much lower than that of a three-
phase induction motor. Disconnecting one terminal of a three-phase motor
(Figs. 3.32e, 3.32f) the motor still produces a torque apart from the start-
ing direction like a one-phase motor. Such hand-started single-phase motors
(Fig. 3.32g) are not manufactured today, but often one-phase operation is
given.
To generate a rotating field two windings with out-of-phase currents are
enough. Mainly the phase displacement of the currents is done by capac-
itors, but also by increasing the ohmic resistance of one of the two phases
(Fig. 3.32h). By appropriate dimensioning of the winding and of the capacitor
a motor can be balanced which means the motor operates as a three-phase
AC induction motor. But this is possible only for one speed value. Should
.--------o t Right
• 0
t Left
t Right
• 0
k
______._____. t Left
VN
Fig. 3.32. Asynchronous motor, methods of connection

be that the nominal speed the motor gets a running capacitor C 8 which
is switched on constantly. However, in this case the locked-rotor torque is
very low because the capacitor is too small (Fig. 3.33e). If a high locked-
rotor torque is required the motor gets a starting capacitor CA which has
to be much higher than the running capacitor. Now the starting capac-
itor is badly suited to nominal operation. This causes inadmissible high
losses warming up the motor. Therefore the auxiliary phase with the start-
ing capacitor is to be switched off. Now the motor is single-phase driven
(Figs. 3.32i, 3.33d). Motors with a high locked-rotor torque and a high nomi-
nal torque get, as well as a starting capacitor, a running capacitor (Figs. 3.32j,
3.33£).
Mainly two-phase capacitor AC motors are produced because they need
a smaller capacitor than AC motors with three phase windings. Besides them
they have a better utilization than resistance-start motors. But the latter are
still more robust than capacitor-start motors.Therefore they are employed
Tb
0.8 1---+:::#::...._-+___::'-----t-""8.-----+------1
.................................. ··
................,.......... . . .
o~--~---L---L---~-~~
0.2 0.4 0.6 0.8 N
Ns
a Three-phase motor
b Single-phase motor, triple phase
c Motor with one phase
}
d Capacitor-start motor
e Running-capacitor motor Douple-phase
f Two-value capacitor motor AC asynchronous
g Resistance-start motor motor
Tb : breakdown torque of the three-phase motor

N5 : synchronous speed
Fig. 3.33. Speed~torque curves of different versions of asynchronous motors

N
1.0 rpm 4.0
I
A
0.8 3.2
0.6 2.4
0.4 1.6
0.2 0.8
0 0
200 400 600 800 T 1200
Ncm
T: torque, TN: nominal torque, N: speed, I: current, rr efficiency
Fig. 3.34. Performance curves of a running capacitor motor
where frequent switching is required (e.g. refrigerators, freezers). High ohmic

losses arise because of the high resistance of the auxiliary phase. Therefore
the auxiliary phase is switched on only for the start (Fig. 3.33g). Figure 3.34
shows the operation curves of a double-phase double-pole 175 W motor with
a 10 J.l.F running-capacitor motor. Table 3.10 gives the characteristic data of
single-phase induction motors.
Table 3.10. Typical data of single-phase induction motors

Number Nominal Nominal Nominal Effi- Running Weight
of poles speed continuous output ciency capaci-
torque power tor
rpm Nm w % J-iF kg
Capacity 2 2600 0.184 50 53 3 2.2
motors 2 2750 0.347 100 58 6 3.5
2 2790 1.71 500 67 12 8.0
4 1300 0.367 50 52 3 2.3
4 1370 0.697 100 55 4 4.5
4 1380 3.46 500 65 16 10.5
Shaded- 2 2600 0.037 10 22 1.0

pole 2 2650 0.18 50 32 3.5
motors 4 1350 0.071 10 20 1.2
4 1359 0.354 50 30 3.7
Control Methods
To change the rotation direction of single-phase induction motors either the
capacitor can be switched over from one phase to the other (Fig. 3.32k, 1)
or the current direction of one phase can be changed (Fig. 3.32m). The first
possibility is more cost-effective because only a one-pole switch is necessary.
But both phases have to be alike if the motor is to produce the same torques in
both directions. Often single-phase motors are made with two different phases
because such motors are better to be balanced, so that their performance is
like that of three-phase AC motors. Now the second switching possibility
is required. In contrast to three-phase AC motors the direction of rotation
cannot be changed while running in every case. This depends on the load-
torque type. Sometimes it is possible that the rotor rotates on in the old
direction. Therefore the motor should be stopped before the new direction
is switched on. Besides them the main phase should be switched of too.
Otherwise the motor rotates further.
The speed can be regulated by the following methods:
- The phase windings consist of two parts which can be connected in parallel
or in series with another or with the capacitor in different ways. For that
a complex switch is necessary and high losses are generated. For instance
this method is used in circulating pumps of heating systems. Because of
the constant synchronous speed the speed range is small.
- The terminal voltage can be varied by transformers or by phase-angle
control. Dependent on the square of the voltage the torque drops over-
proportionally with sinking voltage and the losses increase, especially if
phase-angle control is used because the voltage is no longer sinusoidal
and possesses large harmonics. Requiring an overdimensioned motor this
electronic solid-state speed regulator is cost-effective. Also in this case the
speed range is small because of the constant synchronous speed. To widen
the speed range high-resistance rotors, having a flatter speed-torque char-
acteristic, are employed (Fig. 3.35). Simultaneously the starting torques
becomes greater. Due to the intensive cooling and due to the convenient
speed-torque characteristic these methods are used for fans because the
motors are in the air flow, and for pumps with rotors turning in the
pumped liquid. This construction is cost-effective because a special seal
between pump and motor can be dropped. The stator is protected from
moisture by a very thin nonmagnetic tube in the air gap (canned or split-
cage motor).
- The change of synchronous speed can be achieved with two separate stator
windings each with another number of poles. The winding with the lower
number of poles can also be single-phase. The motor is started with the
low-speed winding before it is switched over to the single-phase winding.
Sometimes only one capacitor is enough for both windings.
- The most effective method to vary the synchronous speed can be done
by frequency variation with an electronic power converter. As the torque-
Speed N
regulation range
Fig. 3.35. AC motor, speed regulation by voltage variation
generating flux is about proportional to the voltage/frequency ratio, the

terminal voltage has to be varied to the same extent as the frequency.
In this way the torque and the utilization of the motor are constant
(constant-flux operation). Starting at a low frequency and at low volt-
age the motor already supplies its nominal torque. Reaching the nominal
voltage with rising speed, no more voltage improvement is possible. Be-
yond it the flux becomes lower. As a consequence of it the maximal torque
becomes lower too. As long as the nominal current is not exceeded, this
so-called field-weakening operation can be used. In this way, speeds are
possible more than twice as great as the nominal speed at 50 Hz opera-
I
1- Constant-flux NN
range
f<fN, lf;<lf;N
Field-weakening
--range-
f>fN, v; =v;N
IN
Fig. 3.36. AC motor, performance range by frequency variation

tion. The breakdown torque of larger motors is constant; the breakdown

torque of small motors becomes smaller with decreasing speed caused by
the influence of the stator resistance which is not negligible in the case of
small motors. This voltage drop can be compensated by a start voltage
greater than null. Only three-phase motors and no capacitor-start mo-
tors are employed because of their worse characteristics. On account of
the high electronic cost this method is still seldom used for high-grade
applications of lower-power drives.
Shaded-pole Motor
The shade-pole motor (split-phase motor) makes use of a further possibility

to achieve phase displacement. The auxiliary winding is short-circuited. Con-
nected with the AC line the main winding induces in this winding a voltage
like a transformer with a short-circuited secondary winding. The auxiliary
current excites a flux lagging to the main flux. These two fluxes generate an
elliptical field (very small ellipse). Therefore a very high oscillating torque
superposing the mean torque generates high losses. This is the reason why
all shaded-pole motors possess bad efficiency. Figure 3.37 shows two con-
structions of these motors. About a third of each pole is "split off'' and
surrounded by the short circuited winding (one to three turns). The coils of
the main winding are wound first and slid on the yoke (Fig. 3.37 left). Then
the yoke is moved into the stator packet. The rotor has the usual aluminium
squirrel-cage winding.
Because of the small phase of displacement of the currents in the main
and the auxiliary winding and because of the winding misalignment lower
than 7r /2 the start torque is very low. In addition, there is a saddle within
Main winding Short circuited winding Main winding
Squirrel-
cage
rotor
Single-axle Two-axle
Symmetrical cross-section
Fig. 3.37. Designs of shaded-pole motors

--
3 1500 r--
\ !]_
----
I N lf;j- %
A
2.5
rpm
: ---....... \
2 1000
) 100
: (J
1.5 / 80
500
I
( 60
)I 40
0.5
----- rr ......
\ 20
0 0
..........
...... ~---------
~ -~- -- -- /
0
20 T 60
Ncm
T: torque, TN: nominal torque, N: speed, J: current, r-,: efficiency
Fig. 3.38. Performance curves of a shaded-pole motor
the speed-torque curve at about a third of the synchronous speed (Fig. 3.38).
Sometimes this causes running-up problems. Therefore shaded-pole motors
are especially suited for small fans and pumps. Other applications are juice
presses, clothe driers, grills, simple butterfly control waves, massage appara-
tus, hot-air stoves and cabinet fans. Drives for reversing duties can be built
with two motors assembled homologously. Shaded-pole motors are low-cost
motors. Because of their low efficiency they mostly need intensive cooling. To-
day the importance of these motors is diminishing. They are being replaced
increasingly by permanent-magnet motors like single-phase synchronous mo-
tors (next section) or simple BLDC motors (Sect. 3.3.1).
3.4.2 Synchronous Motors
Introduction
In synchronous motors the rotating field is generated in the same way as in

asynchronous motors. Contrary to them whose rotors rotate slower than the
field, the speed of synchronous motors is equal to the field rotational speed.
Therefore the speed is strongly proportional to the supply frequency. Speed
variation is not possible with the help of voltage variation. Small motors are
used in open-loop control techniques. Because of their simple construction
and relatively high efficiency, if they have permanent-magnet rotors, they are
used in household appliances too: program controllers, clocks, time-delay re-
lays, writing measuring instruments, pulse recorders, studio camcorders, gyro
drives, slot machines, fans, pumps (washing machines, dishwashers, aquari-

ums), lemon squeezers, tin openers, recorder, answering apparatus etc.
Like AC motors the rotating field pulsates during the rotation. This re-
sults in huntings which entail disagreeable pitch pulsations if this is not com-
pensated by a high moment of inertia. Otherwise there are for instance loud
sound vibrations of audio apparatus. The rotor tries to follow the rotating
field. At no-load the pole axis of the rotor and field are in phase. The more
the load rises the more the angle between these axes (load angle) becomes
greater and greater, whereat the rotor is running after the field. If the load
becomes greater than the breakdown torque the rotor stands still. Added to
that the rotor swings at load variations (a special problem of stepper motors:
see Sect. 3.4.3). This property is not so marked with AC motors because
their squirrel-cage rotor has a damping effect. But the main problem of syn-
chronous motors is the start. The rotor does not reach the synchronous speed
immediately after it is switched on. For a certain running up there are several
possibilities which depend on the rotor construction demonstrated in the fol-
lowing sections. Small synchronous motors have no rotor coils and slip rings,
Table 3.11. Typical data of single-phase synchronous motors (230 V)

Number Nominal Nominal Nominal Efficiency Weight
of poles speed continu- output
ous power
torque
rpm Nm w % kg
Reluctance 2 3000 0.16 50 40 2.5

motors 2 3000 0.32 100 55 4.0
4 1500 0.32 50 40 3.0
4 1500 0.62 100 55 4.2
Hysteresis 2 3000 0.003 1 5 1.0

motors 4 1500 0.006 1 5 1.2
6 1000 0.012 1 5 1.5
12 500 0.024 1 5 2.0
Permanent- 8 750 0.013 1 30 0.2

magnet 8 750 0.064 5 35 0.6
motors 8 750 0.127 10 40 1.0
12 500 0.019 1 25 0.25
12 500 0.038 2 30 0.4
12 500 0.095 5 35 0.8
16 375 0.025 1 23 0.25
16 375 0.051 2 30 0.4
16 375 0.127 5 35 0.8
24 250 0.038 1 20 0.25
24 250 0.076 2 30 0.5
24 250 0.191 5 33 1.0
but magnet, hysteresis or reluctance rotors. Table 3.11 shows typical data of
the different types of 230 V motors.
Permanent-magnet Rotor
The stator of larger synchronous motors (over 100 W) is constructed like that
of AC motors, i.e. they have slots and are wound double- or three-phase. Con-
nected with the AC line the same elements for phase displacement are taken:
capacitors or resistances. Larger motors have varying rotor constructions.
Figure 3.39 shows two examples with a squirrel-cage winding for running-up.
Coming near to the synchronous speed the rotor pulls into synchronism.
Such motors are called synchronized induction motors or hybrid synchronous
motors (Merril motor). Here the squirrel-cage winding affects damping too.
If the load becomes greater than the nominal torque and finally exceeds
the synchronous breakdown torque, the rotor pulls out of synchronism and
runs further on, until it exceeds the resultant breakdown torque too. The
quasi-steady operation is illustrated in Fig. 3.40. Larger synchronous motors
without a squirrel-cage winding can be run up only by an electronic control
whose frequency arises gradually from near null up to the nominal frequency.
Smaller motors with a small number of poles are constructed like shaded-
pole asynchronous motors with a short circuited coil or without it. They
have a permanent-magnet solid or hollow cylinder as rotor. Therefore they
do not start right off. After switching on the AC voltage the rotor begins
to swing and then it pulls into synchronism within a few periods. Therefore
the moment of inertia has to be low and it is advantageous if the load is
not solid-coupled. Often the gear backlash is sufficient. To support the start
swinging the poles are asymmetrically constructed. Here the poles have steps;
in another case the air gap becomes slowly wider and wider from one pole
Fig. 3.39. Synchronized induction motors

Asyncronous torque
T
,..---.....::- - - Breakdown
torques
.I
N
Syncronous torque
Fig. 3.40. Synchronized induction motors, speed-torque curves
side to the other. The rotor takes different positions, when the current is
switched off and switched on. Figure 3.41 shows a typical design of a double-
pole one-phase motor. The motor in Fig. 3.41 is of great importance because
of its favorable cost and efficiency and has replaced shaded-pole asynchronous
motors as drives for small pumps (see above) and juice-squeezers. At start
the direction of rotation is unknown. In many cases this is not problematic.
To rotate only in one direction a backstop can be considered.
For control purposes often a low speed is desired. In this case the motor
gets a claw-pole stator, as shown in Fig. 3.42. A simple ring coil is mutually
enclosed by metal clips. All metal components are only punched, bent and
welded or clamped together. This is a very cost-effective manufacturing pro-
cess for getting a multipole motor. A disadvantage is that a large part of the
flux does not enter the rotor generating a torque, but as stray flux it goes
Fig. 3.41. Pump driven by a single-phase synchronous canned motor

Ring coil
Fig. 3.42. Claw-pole principle
over directly from pole to pole. Besides them eddy currents are caused by the
alternating field in the relatively thick sheet metal generating high losses.
When the alternating field becomes a rotating one, short-circuited rings
are built in like those of shaded-pole motors (Sect. 3.4.1). In Fig. 3.43 only
the stator of an external-rotor motor is presented in this account to give
Fig. 3.43. External-rotor motor, design with claw-pole and shaded-pole principle
System 1
System 2
Fig. 3.44. AC synchronous motor with two claw-pole systems
a better survey. The short-circuited rings influence the flux of the main and
the auxiliary claws to a different degree. Instead of the rings often cupper
plates are arranged at the front side of the motor. To support the start or
the running in a decided direction the claws are formed asymmetrically.
The rotating field of single-phase motors pulsates very considerably, so
that the mean torque is only low and a distinct noise is possibly generated. It
is more favorable if two claw-pole motors are built together. Their stators are
twisted with respect to each other by a half pole pitch and their rotor poles
are in line (Fig. 3.44). One of the stator windings is directly connected with
the AC line, the other in series with a capacitor or resistance (Fig. 3.32h or
3.32i). Should the speed direction be changed the capacitor is changed over
at the other phase (Fig. 3.32e). Likewise three claw-pole motors are built to-
gether with stators distorted like the phases of three-phase motors. They are
directly connected with the three-phase system. Also two smaller motors with
aligned stators can be built together to get a more powerful motor. The essen-
tial advantage of synchronous motors with permanent-magnet rotors is their
multiple greater torques compared with the other synchronous-motor types.
Speed regulation of large motors can only done by pole or frequency changing,
the latter for instance by pulse-width modulation (PWM; see Sect. 3.6.5).
Hysteresis Motor
The stator of hysteresis motors is like that of a magnet-rotor motor. The rotor
is a cylinder with a layer of hysteretic material. It is similar to magnetic ma-
terial, but nonmagnetic in normal conditions and possesses no poles. During
running-up the magnetization of the hysteretic layer is permanently reversed.
By that a torque is produced which accelerates the rotor up to synchronism.
In this state poles arise. Now the hysteresis motor works like a magnet mo-
tor, but with an about 20-30 times smaller torque. Switching off the motor
the magnetizing disappears. Its advantage is its self-supporting start and the
soft change-over from asynchronism to synchronism (Fig. 3.34). The eddy
currents in the rotor produce a torque additionally to the hysteresis torque.
Hysteresis motors are only seldom used today because of their low torque.
Reluctance Motor
The magnetic resistance (reluctivity) of reluctance motors varies along the

circumference corresponding to the pole number (anisotropic rotor). The ro-
tor tries to align itself with the rotating field, so that the reluctance of the
magnetic circuit becomes a minimum. That happens in synchronism. For
that the rotor of larger motors has a squirrel-cage winding like that of AC
induction motors. Figure 3.46 shows two construction possibilities.
The left rotor is an AC motor rotor in which four poles are milled. The
right rotor possesses non-magnetic spacers which force the flux on decided
paths. The slots, pole gaps and spacers are filled with aluminium. The pro-
duction of the left motor is cheaper, the right motor generates about double
the torque.
Very small multi-pole motors have a sheet-cylinder in which as many
windows as desired poles are punched. So an extremely low rotational speed
is given. The stator is of the claw-pole type like those of magnet motors.
Other rotors are formed like gear wheels with a tooth number like the
stator pole number. The stator has two or three phase windings. These
small motors generate a start torque by eddy currents. Contrary to hys-
teresis motors which pull in synchronism softly, this takes jerkily in re-
luctance motors. Pulling out of synchronism they continue running asyn-
chronously if the load torque is lower than the asynchronous breakdown
torque (Fig. 3.47). Today applications of reluctance motors are unimpor-
tant.
d ·······:r.·:r······
E ••••••••••••• •••• .....,.---Ideal
··········r························::::., motor
TH i Real
Ns-;r
TE: eddy-current torque, TH: hysteresis torque
Fig. 3.45. Hysteresis motor, speed-torque curve

Flux path
Fig. 3.46. Reluctance motors with squirrel-cage windings
---------Asynchronous
t
T
breakdown torque
-Synchronous
-Pull-in torque
I
I
I
I
I
N---
Fig. 3.47. Reluctance motor, speed-torque curve
3.4.3 Stepper Motors
For many positioning tasks step by step rotating motors are required instead
of motors with a continual motion [6]. Contrary to BLDC motors with rotor
position encoder, stepper motors are the cheaper option because they are
driven in open-loop control. There is no check-back signal whether the rotor
has really reached the given position. Therefore it must be sure that un-
der every admissible condition the motor always runs the desired number of
steps. So for a stepper motor it is characteristic that on every electrical pulse
always exactly one mechanical pulse follows. As BLDC motors, the current of
stepping motors is electronically commutated too, but with a frequency given
by the electronic control. Therefore stepping motors belong to the class of
synchronous motors and possess all the characteristics which are typical for
this motor type, like starting problems, pull out at overload and oscillations
under sudden load variations. These characteristics authoritatively influence
the design of a stepper drive because they evidently disturb the observance of
the above-mentioned condition "mechanical equal electronic pulse number"
or even prevent it. In the latter case BLDC motors have to be used.
Fig. 3.48. Fundamental constructions of stepper motors
Stepper motors are constructed in principle like BLDC motors. The stator
has a winding made of concentrated coils on distinct poles (Fig. 3.48) or has
a ring coil in a claw-pole system (Fig. 3.44). The rotor is a permanent-magnet
cylinder (permanent-magnet (PM) stepper motor), a gear wheel (variable
reluctance (VR) stepper motor) or a combination of both types (hybrid
(HY) stepper motor). In order to rotate in the desired direction, clockwise
or counter clockwise, the motor consists of two submotors (PM, HY) or of
three submotors (VR), in the following, called "systems". The systems can
be arranged in a plane side by side (Fig. 3.49) or axially one after another
Laminated stator core
Axially magnetized
permanent magnet
System 1
South pole ring gear
Fig. 3.49. Hybrid stepper motor

(Fig. 3.44). Either the system-stator::; are twisted towards each other or the
rotors by half a pole pitch. The other elements are in line.
Contrary to the VR motor which has to be controlled three-phase, PM and
HY motors need only a double-pha::;e control. The latter both have a higher
torque and the advantage that they produce a currentless torque if the sta-
tionary rotor is displaced by the load. This torque is similar to the cogging
torque of BLDC motors. There it is not desired, but here it is advantageous
because the rotor is fixed, when the stator winding::; are unexcited. This de-
tent torque should be not greater than 15% of the maximum steady torque,
when the windings are excited (holding torque). Otherwise large pulse o::;cil-
lations arise. The disadvantage of the PM motor contrary to the HY motor is
that not so many rotor poles can be magnetized along the circumference as
the latter has teeth. Table 3.12 gives characteristic data for the most essential
designs.
The ampere-turns of the systems change, alternating with every pulse.
If in Figs. 3.50a, 3.50b voltage is applied to the systems one after another,
the rotor moves on at 90° and makes four steps every turn. After a step
is finished it stands in the pole axis of the just-excited system. The torque
can be enlarged if two systems are excited simultaneously (Figs. 3.50c, 3.50d).
The number of steps at one turn is four too, but the rotor stops in the middle
of the pole axis of the two excited windings. The rotor makes a full step in
every case; it is working in "full-step mode". If alternately one, respectively,
two sy::;tems are excited in the sequence a, c, b, d, etc., the step angle is
halved and the number of steps doubled. In this case the motor is working in
"half-step mode". The number of steps is given by
2pm
Z=--
ks
where
f line frequency
2p number of poles of the aligning motor part
m number of systems
ks operation-mode factor (full-step operation: k 8 1·
half-step operation: ks = 0.5). '
The number of steps yields the step angle
360°
a=--.
z
Figures 3.50e and 3.50f show the alternating control of the ::;ystems of a VR
motor; Figs. 3.50g and 3.50h the alternating control of a HY motor.
Half-step operation is a possibility to diminish the step angle, but it has
the disadvantage that the torque is different step by step. To avoid this,
expensive current control is required. Therefore mostly full-step operation is
used like Figs. 3.50c, 3.50d, where the systems are reversed alternating. In
Table 3.12. Characteristic data of stepper motors

Number Step Holding Moment of Electrical Weight
of steps angle torque inertia time
(rotor) constant
deg Nm gcm 2 ms kg
Claw pole 24 15.0 0.05 3 0.7 0.09
motor 24 15.0 0.11 30 1.6 0.3
24 15.0 0.26 60 1.6 0.55
32 11.25 0.011 3 0.7 0.08
32 11.25 0.115 30 1.6 0.3
32 11.25 0.28 100 2.0 0.43
40 9.0 0.012 3 0.7 0.08
40 9.0 0.125 30 1.6 0.3
40 9.0 0.23 60 1.6 0.55
48 7.5 0.013 3 0.7 0.08
48 7.5 0.13 36 1.4 0.3
48 7.5 0.4 120 2.0 0.43
Hybrid 100 3.6 0.114 10 0.2

motor 100 3.6 0.26 18 0.21
200 1.8 0.18 52 0.34
200 1.8 0.48 120 0.53
200 1.8 1.0 330 1.0
200 1.8 5.5 2800 4.6
200 1.8 9.5 5500 7.25
200 1.8 15.0 8300 10.5
Hybrid 500 0.72 0.25 100 3.0 0.5

motor or or 0.85 400 8.0 1.3
with 5 1000 0.36 3.8 1800 10.0 3.5
systems 11.0 11500 20.0 12.5
PM motor 40 9.0 0.011 1 0.6 0.04

with disc 100 3.6 0.19 12 2.5 0.25
rotor 200 1.8 0.57 100 3.7 0.6
this case the motor is best utilized. With a still more expensive control the
ampere-turns of the systems can be changed in very small steps to move the
rotor in very small steps too. This mode is called "micro-step operation" .
This method has the additional advantage that the rotor reaches the next
latched position without swinging.
The step-number equation points at two further possibilities for dimin-
ishing the step angle:
An increase of the system number m follows a corresponding increase of
control-phase number raising the cost. Therefore mostly motors with two
systems are built. Seldom are there motors with up to five systems.
b d f h
Fig. 3.50. Control principles of stepper motors
- It is more cost-effective to increase the number of poles or teeth. In this

case more exact manufacturing is required. Otherwise the step-angle error
becomes too great. This error means the deviation of the desired position.
Because of the cost, stepper motors are often wired up unipolar, more
seldom bipolar. In order to follow the field during the performance of several
steps the control frequency is not allowed to exceed the so-called maximum
start-stop stepping rate. The corresponding curve is given dependent on the
load-moment of inertia in data sheets of stepper motors. To realize high
stepping rates, e.g. to minimize the positioning time the following methods
are given:
A resistance is connected (constant-voltage drive) in series with each

phase winding. Disadvantages are additional losses and a more power-
ful output amplifier.
- The windings are applied to a high voltage which is modulated so that
the mean current becomes the permissible value (constant current drive
or chopper drive).
- An additional voltage is superposed on the nominal voltage as long as the
nominal current is reached. Then the voltage becomes the nominal value
(bilevel drive).
The dying out of the rotor, after the motor is switched off, lengthens the
positioning time, so that dumping is indispensable (apart from the micro-step
drive). Today this happens generally by specificly placed pulses.
The most cost-effective stepper motor is the PM motor built as a claw-pole
type. It is constructed with two systems like the corresponding synchronous
motor (Fig. 3.44). Both stator systems are displaced to one another by about
half a pole pitch. Mostly each system has two ring coils where the currents
flow, alternating in one or other direction (unipolar connection). This connec-

tion needs a simple, i.e. cheaper, electronic control, but the motor is larger
because of the double number of windings. Occasionally each system has
only one coil, where the current direction changes alternately (bipolar con-
nection). In this case the electronics are more expensive, but the utilization
of the motor is better. The polarization of the ferrite-magnet rotor is con-
stant along the axis. Claw-pole stepper motors are produced with step angle
greater than or equal to 7.5 o. The great eddy-current losses and the stray
fluxes are the reasons for the low utilization of these motors. Drives with an
exceptionally low moment of inertia are constructed as PM disc motors, but
they are expensive especially because of the complicated rotor construction.
If small step angles are demanded without using gearing mostly HY mo-
tors come into question (Fig. 3.49). They have the advantage of the VR
motors, the small step angle, with the advantage of PM motors, the large
torque, but they are more expensive than claw-pole motors. The rotor has
two ring gears which are displaced to one another by about half of a tooth
pitch. Between the rings there is a rare-earth magnet which is magnetized
axially. The stator is laminated and generally has two systems in a plane.
Two opposite lying poles of a double pole motor belong to one system. The
poles are toothed too with the same tooth pitch as the rotor, but the teeth
of the both systems are displaced to each other by about half a tooth pitch.
Today bipolar connection is used mostly. In exceptional cases motors of this
type can have up to five systems with step angles down to 0.36°. But these
are very expensive actuators.
All of these motors can be designed as linear motors. Drives with such
motors are very expensive. Therefore mostly they are produced with a rotat-
ing motor whereby the revolving motion is formed into a linear motion by
gearing (feed screws, gear racks, toothed belts). The advantage of linear mo-
tors is that they directly actuate the working element and the disadvantages
of gearings (losses, play, noise) are avoided. But either the stator or slider has
to be as long as the working range and the advantage of gearing (enlarging
the driving force respectively minimizing the load-moment of inertia) is not
given.
PM-stepper motors with only one system like the one-winding-phase syn-
chronous motor (Sect. 3.4.2, Fig. 3.41) are used as clock drives (Fig. 3.66).
Figure 3.51 shows the principle of construction with unsymmetrical poles. If
the motor is switched off the pole axis may be situated in the position of
Fig. 3.51a. If the winding is supplied to voltage the current should generate
a flux pulse in the here drawn direction. On that the rotor rotates clockwise
and stops at the painted position in figure (b) after the current pulse dis-
appears. The next opposite directed pulse causes a further step of 180°. If
the rotor were to start in position (a) by a flux pulse like figure (b) it would
move about a small angle counter-clockwise and then fall back to the start
position. The next pulse is opposite to the first one and therefore the right
-
138 Hans-Dieter StOlting
-
<I>~
I I
a • +/ b -I
Fig. 3.51. Single-phase stepper motor
starting pulse is given. All clockwork motors are 180°-stepper motors rotat-
ing in only one direction. Because of that they are hardly used for any other
purpose than clock drives. After a new battery was put into a quartz clock,
it is possible that first a false pulse appears, so that the motor loses one step.
However, the next pulse is the right one. This trouble makes of the exactness
of the clock no worse.
3.5 Limited Motion Drives

3.5.1 Introduction
Many apparatus move only with a short unique or oscillating rotational or

linear motion. Then rotating motors with eccentric movements, rod drives
camshafts and tappets can come into question in individual cases. But today
they are used by way of exception because of the noise, the wear and the
cost. Actuators principally possessing the required motion are better suited.
They use the following electromagnetic principles of force generation:
- Forces at boundary layers of materials with different permeability (Max-
well force).
- Force on current-carrying conductors in a magnetic field (Lorentz force).
3.5.2 Electromagnets
Introduction
Electromagnets operate with Maxwell forces [10]. They have a fixed element
(magnetic core, body or yoke) with a coil (excitation winding) and a movable
element (armature). The magnetic core and the armature are made of ferro-
magnetic material for flux carrying. Switching on the current the armature is
attracted, switching off it falls back to its free position. There are character-
izing differences between direct and alternating current magnets influencing
essentially their applications. Many applications use these drives whereby
three magnet types are distinguished, solenoid and hammering actuators,
turning and swinging magnets:
Solenoid actuators: flaps, pneumatic and hydraulic valves, slide valves,

interlocks, brakes.
Hammering actuators: riveting, punching, stumping, chiselling machines.
Thrning magnets: throttle valves, control valves (hydraulic, pneumatic),
material support (for instance web of cloth, paper), turnout in transport
plants.
Swinging magnets, vibrators: electric razors, massage apparatus, piston
pumps, swing pumps, diaphragm pumps, small compressors, swing saws,
vibrating hauling plants, oscillating sieves, oscillating tables, swinging and
helical conveyors.
DC Magnets
The magnet body and the armature are made of solid iron and are very ro-
bust. Both components can be formed so that an appliance-matched torque-
per-way characteristic F(s) results. Figure 3.52 shows fundamental design
possibilities of solenoid and drag magnets serving for adjusting and position-
ing with their associated F(s)-curves. The construction (c) is a combination
of (a) and (b). Hammering magnets are designed similarly. But there the
electric energy is converted into kinetic energy generating mechanical pulses
for hammering, riveting, etc. The armature is reset at de-excitation by grav-
ity or springs. The magnetic forces are between lOmN and lOkN and the lift
amounts to some millimeters (small magnets) up to 20 centimeters (greater
magnets).
E A
a: magnetic core, b: armature, c: winding, d: nonmagnetic limit stop, A: start
position, E: end position
F: force, s: way
Fig. 3.52. Principles of DC magnets and their force-way curves
Besides translatory operating magnets, often rotational working magnets

are demanded. Their rotation can be made in different ways:
Moving magnets with axial air gaps are fundamentally constructed as
illustrated in Fig. 3.52. The lifting movement can be changed in a rotating
movement in several ways, for instance by an inclined ball race (Fig. 3.53),
so the shaft makes a small axial motion. The compressive load per unit
area of the balls causes greater wear. An advantage is the large angle of
rotation. Often an axial motion is undesired. In this case the armature
can be formed as a locked female screw with steep-lead-angle thread (the
shaft acts as a screw fixed for rotation). With this construction essential
large angles of rotation can be reached, but with very great wear. Another
type with a low rate of wear and without a linear motion is built of an
armature locked against longitudinal motion which is fixed with the shaft
and a core within the coil. Both of the components have helix shaped faces.
If the coil is currentless they form a large air gap. As soon as a magnetic
field is excited the armature tries to diminish the air gap by rotation.
Figure 3.54 shows this function with the aid of the unwound armature-
core principle. The production of these magnets is expensive because of
the need for milling the faces.
Solenoids with a radial air gap are rare because they generate only low
start and end forces and the angle of rotation is limited up to a maximal
65°. Figure 3.55 shows the general designs. The magnetic circuit consists
of the housing, one- or double-sided pole systems with the convenient
armature and the internal tube serving as magnetic return path. Pole-
systems and armature can be milled or pressed (porous metal). The reset
ensues by springs whose torque diminishes the working torque. For re-
Fig. 3.53. Rotating DC magnet with an inclined ball race
Start position End position

~------------~
~Core
A~aMe
Fig. 3.54. Rotating DC magnet with helix shaped faces of core and armature
Magnetic core
One-sided
pole system
Double sided
Coil Return path
Fig. 3.55. Rotating DC magnet with radial air gap
versing duty two solenoids are mounted, mirrored and excited according
to the motion direction. DC solenoids reach torques of about 1 Nm.
Generally the magnets are connected by rectifiers to an AC system. Excit-
ing the magnet coil is possible by a capacitor discharge. Should the magnet
react especially fast after it is switched on, the electric time constant can be
diminished for instance by a series resistance. But through that, additional
losses arise which require a more powerful supply. A further possibility to
diminish the pickup time of the armature is to switch it on at a higher volt-
age (over-excitation). But it must be dropped down to the nominal voltage
at once because the current increases to an unpermissible value. The voltage
change-over can be done by the magnet it selves time- or current-dependent
(bilevel operation).
AC Magnets
Because of the pulsating field in the magnet core and in the armature iron
losses arise, all flux-carrying parts of AC magnets have to be laminated. This
is done only for larger magnets, but not for small magnets because of the
cost. In the first case the switching time is shorter than that of DC magnets.
The magnetic force pulses with the double the system frequency entailing
a humming noise. To get a small retention force in the end position AC
magnets always have a short-circuited winding in which a current is generated
inductively. This results in a second displaced force so that the resulting force
is greater than null every time.
Magnetic
core
Winding
Short circuited winding

a b
t
F
\
Armature Start postion
c
F: force, s: way
Fig. 3.56. Principles of AC magnets and their force-way curves
There are lots of design possibilities to realize different force-per-lift char-

acteristics. Figure 3.56 shows three fundamental designs with the correspond-
ing F(s)-curves. Construction (c) yields the best utilization because of the
very low stray flux. AC magnets reach forces between 1 and 150 N with a lift
between a few millimeters and 10 centimeters. Larger magnet forces are gen-
erally associated with small lifts. Three-phase magnets are built with the
principle of Fig. 3.56c whereby the windings are fitted on three limbs. These
magnets produce forces between 50 and 1000 N with lifts of 20-60 mm. There
are few AC rotating magnets. Sometimes the rotary motion of a clapper-
armature magnet is used (Fig. 3.56b).
Swinging-armature Magnets, Vibrators
A to-and fro-motion of an armature is possible through when the magnet-

force generating current is switched on and off at the favorable moment or
its direction is changed. Drives with only one coil need a spring for falling-
back into the start position. If the armature swings beyond the position of
minimum energy a restoring force arises. The spring and the armature force
of inertia must be tuned so that the current switch-frequency is equal to the
natural frequency of the system. In drives with two coils the armature-swing
affects the alternating switch-on of both coils. By corresponding tuning the
resonance-effect can be obtained here too. This is the assumption for optimal
efficiency.
+
~''H
~- ......... ~~
Fig. 3.57. Swinging-armature magnets
Just as in the case of rotating motors, swinging-armature motors can

be subdivided in self-commutated and externally commutated motors with
similar properties as rotating motors (Sect. 3.1.2). In self-commutated drives
the armature switches themselves the one winding on and off respectively
to-and-fro between the two windings. The swing frequency is free eligible
because it depends on the motor design and the load. The switch arc is the
reason for switch wear and the limited life time. Also, these drives are loud.
Electronic commutation is quiet and robust, but more expensive because it
needs an armature position sensor and electronic control.
External commutated drives can be driven with alternating or direct cur-
rent. The first method is the cheapest one, the latter current must be switched
by more expensive electronics. These actuators are especially robust and
quiet. In every case the motor operates according to the synchronous-motor
principle, i.e. the switching rate is fixed. If the armature consists only of
iron the armature swings with double the system frequency (reluctance ar-
mature). If it is premagnetized it swings with the single frequency (magnet
armature).
Swinging-armature motors have magnet bodies and armatures which are
coupled together not mechanically, but magnetically. Only the armature
oscillates, whilst the magnet body stands still. All designs represented in
Figs. 3.52, 3.55 and 3.56 can be built as swinging drives. Figure 3.57 shows
widespread constructions. For instance they are used as drives in electric
razors.
Fig. 3.58. Vibrator

Vibrators are built similarly, but the magnet body and armature are con-
nected to one another by tuned springs, so that both elements are able to
swing (Fig. 3.58). If vibrators are fixed to conveying systems, so that the
swinging motion is directed upward-sloping (throw vibrator) goods can be
transported upwards. The same principle is used in winding-conveying appa-
ratus for the supply of components in automatic production lines.
3.5.3 Voice-coil Motors
Voice-coil motors (plunger electromagnet) are so called because they oper-

ate like electrodynamicalloudspeakers [8]. Here the force on current-carrying
conductors in a magnetic field is used. This force is proportional to the mag-
netic flux and the conductor current. Therefore the deflection of an elastic
fixed coil can be controlled by the current. By changing the current direction
the motion direction can be reversed.
Figure 3.59 shows two fundamental construction variants. In (a) a coil
is arranged coaxially to the magnet body. Instead of arranging within the
magnet the coil can lie outside of the magnet. Cylinder or box coils are used.
Part (b) shows a fiat building construction with a disc coil. As the design of
the coil's bearing and guide is difficulty to produce, the coil is often fixed and
the magnet movable arranged. But this construction often possesses a greater
force of inertia. In every case it is to be decided whether the magnet or the coil
should have the motion length. Both constructions have special advantages.
Fig. 3.59. Principle of construction of voice-coil motors
3.6 Closed and Open Loop Control Methods
3.6.1 Introduction
An overview of regulating and controlling methods of fractional horsepower

motors is given in Fig. 3.3 [9]. These methods are similar to those which are
known from larger drives. However, the type and the expense of the circuit
determine the cost. Therefore power semiconductors should be employed as
little as possible and no regenerating braking is used. Moreover, there are
circuits which are unsuitable in the higher power range. Fractional horsepower
motors are often driven in only one direction, so that the drive control must be
suitable only for one-quadrant operation. For reversible operation the current
direction has to be changed. In the range of low power mechanical switches
are often used. Should the rotor be stopped as fast as possible a mechanic
brake is chosen (e.g. spin drier). For instance in sunblind drives this is done
as follows. After the current is switched off a spring presses the rotor a bit
out of the stator against a disc brake. By switching on the current the rotor
is pulled into the stator and so the brake is lifted. Compared to electronic
brake circuits the advantage of a mechanic brake is that it works in the de-
energized state. If the motor is to work and brake in both rotational directions
a four-quadrant circuit is required.
3.6.2 Line-commutated Converter
Line-commutated converters are of use for DC commutator motors with

permanent-magnet stator. But they are also suitable for DC series motors
which are seldom used in the lower power range. Apart from the latter motors
the torque-generating current is practically constant. Therefore the torque is
proportional to the current. With the change of the load-torque the motor
current changes too if the speed is to be constant. The DC motor is driven
mostly at a half-controlled bridge. If it is supplied by AC voltage a recti-
fier is connected in series. For speed control the actual value is measured by
a tacho-generator and compared with the setpoint value. The analysis de-
livers the current setpoint. This value is compared with the instantaneous
current value. This yields the signal for the corresponding output voltage of
the converter (current-controlled speed limiting system).
3.6.3 DC Chopper Converter
At AC voltage supply DC chopper converter (DC chopper) own a line-

commutated rectifier which feeds a self-commutated converter. Figure 3.60
shows a simple converter for an electric hair-dryer with a voltage divider
(heating coil) for voltage matching. The transistor works as a switch. The
<>----"-..(_
220V-
0 ~
h fd
H: heating coil, r: rectifier, fd: free-wheeling diode, c: control,
pm: permanent magnet
Fig. 3.60. DC chopper for a hair-drier (single-quadrant operation)

Fig. 3.61. H-bridge (four-quadrant operation)
free-wheeling diode (FD) carries the current in the switch break. With the
H-bridge in Fig. 3.61 motors can be driven in four-quadrant mode. The cur-
rent course is marked by a solid line for one direction of rotation and the
additional course by a dotted line if the transistor T 1 is switched off. If the
other direction is desired the transistors T 2 and T 3 are switched on. The
speed control ensues in the same way, as mentioned in the case of a line-
commutated converter (see Sect. 2.5).
3.6.4 AC Chopper Converter
By AC chopper converters (phase-angle control) above all universal (AC com-

mutator) motors are driven, but seldom AC induction motors (see Sect. 3.4.1).
As a rectification of the alternating voltage is not necessary, this circuit is
simple and cheap. This is the reason why universal motors became the most
important drives of household appliances and handhold tools. Figure 3.62
Fig. 3.62. Phase-angle control of a universal motor

shows the fundamental circuit and the curves of the motor voltage UM and
of the motor current iM. Dependent on the resistance-position the capacitor
charges itself. If its voltage reaches the gate-trigger voltage of the Diac the
capacitor discharges over the Diac and the Triac, so that the Triac is trig-
gered and the motor voltage is switched on. By variation of the resistance the
firing angle a is varied. P is the phase angle between the voltage and the cur-
rent. Today an IC regulates the firing angle a. The speed-control of motors
is done with help of tacho-generators. Often they are very simple. A ferrite-
magnet rotating with the motor shaft induces voltage pulses in a coil. The
pulse frequency is raised with help of a claw-pole system (see Sect. 3.4.2).
It improves the analysis in an evaluation-logic for tuning the firing angle. It
must be protected against disturbing pulses generated by the commutation
which are especially distinct in phase-angle control.
3.6.5 Converter for Polyphase Motors
For small asynchronous and synchronous motors only frequency converters

with constant link voltage (PWM) come into question today and in the future
if the cost accounts for this. With the mark-to-space ratio the amplitude
can be changed as the frequency of the output voltage. By this frequency-
proportional voltage matching is possible, the assumption for constant-flux
operation (Sect. 3.4.1). If it is desired the motor can start with the maximum
torque. Is really the speed regulation of rotating-field motors expensive, the
more it is the speed control. Therefore alternative actuator solutions with
BLDC motors may be more advantageous. This is to be decided in every
case.
3. 7 Applications
The radial blower in Fig. 3.63 is driven by s permanent-magnet-excited 12 V

DC motor. To get an overall length as small as possible the motor is built
within the blower using an internal rotor. This construction is mechanically
more advantageous for a commutator motor and offers more space for the fer-
rite magnets used here. If BLDC motors are used for blowers or fans external
rotors are more effective. In this case the fan baffies are mounted directly on
the rotor housing.
Many new actuators have been developed for car technology. Figure 3.64
shows a compact actuator for power-assisted steering. The driver's hand-force
is measured by a torque sensor. The electronics control the motor current,
so that the on steering-gear acting torque, detected by a second sensor, is
a multiple of the hand-force torque. The ratio of both torques can be reg-
ulated depending on the driver's wishes and on the driving speed among
other things. There are advantages compared with the hitherto used hydraulic
Blade Blower wheel
Fig. 3.63. Radial blower
Torque sensors
Fig. 3.64. Drive for a power-assisted steering (Bosch)
power-assisted steering relative to driving behavior , built-in and working ex-

penses.
As increasing car-electrification leads to a thicker and thicker wiring har-
ness, new techniques for carrying energy and signal fluxes are being devel-
oped. Bus systems should replace the supply cables, for instance, to control
electronic motors. Among other motors the wind-screen wiper BLDC motor
in Fig. 3.65 should replace the currently used DC motor if the cost accounts
for this. In this way, the speed regulation possibilities are much more ex-
tensive, e.g. to regulate the wiper speed corresponding to the rain intensity.
This proves right for many other auxiliary actuators in cars, e.g. for the
Diaphragm
for pressure
compensation
Stator
core
Fig. 3.65. Drive for a wind-screen wiper (Bosch)
cooler-blower motor for the same advantages. In the future the starter motor
and the dynamo will be combined into one machine (starter-generator). This
will happen together with the introduction of a second vehicle-network of
42 V besides of the 12 V network. All the large current-users, e.g. the greater
motors, will be connected to the 42 V system. But today there are many
problems, especially a la rger and therefore much more expensive battery and
electronics.
In the clockwork of Fig. 3.66 similar single-phase synchronous machines
are built in, namely in a double manner as stepper motor and as generator.
These stepper motors are typical for quartz-controlled drives of analogous
clocks (see Sect. 3.4.3, Fig. 3.51). The same construction is chosen for the
Electronic
Generator
Fig. 3.66. Clockwork with stepper motor and generator (Seiko)

generator. It supplies a battery driven by an imbalance. It rotates up to

lOO,OOOrpm.
In Fig. 3.67 two hybrid stepper motors drive the slide of a coordinate
table with a spindle gearing in the x - and y-direction. Today such drives are
driven more and more by linear drives because they are more compact and
robust avoid the disadvantages of gearing (Sect. 3.4.3).
The light-gantry robot in Fig. 3.68 is suitable for positioning, mounting
and handling tasks in laboratories and of precision mechanics. Its head can
pick different tools (soldering apparatus, grabs, etc.). It is driven by three
five-phase stepper motors three-dimensionally with a resolution of 0.1 mm
and with a maximum speed of 0.7mjs. The programming of the head- and
tool-control is made menu-assisted on the screen of a personal computer.
Screw spindle
Fig. 3.67. Coordinate table (Phytron)
Fig. 3.68. Light-gantry robot (Berger)

Figure 3.69 shows a typical example of a voice-coil drive. It produces the

rotary motion of the read head in diskette units dependent on the coil exci-
tation. Most write-read heads are fitted out with such actuators. Mounting-
and measuring tables have two-dimensional voice-coil drives, e.g. machines
for laser-processing of printed boards.
The motor in Fig. 3. 70 is an example that very old inventions are still used
today. More than llO years ago the rotor of the first widely used three-phase
generator possessed a claw-pole system.
Fig. 3.69. Voice-coil drive of a read-head
Fig. 3.70. Apparatus fan . BLDC motor with claw-pole stator, external magnet
rotor and built-in electronics
152 Ha ns-Dieter Stolting
It was an ingenious idea but in large machines the disadvantages predom-

inate (see Sect. 3.4.2, Fig. 3.42). Employed in small motors t he advantage
is to get a polypole motor with only one coil. Claw-pole constructions are
to be found for instance in stepper motors , in bicycle dynamos , in motorca r
generators and in the BLDC motor of the fan in Fig. 3.70. Often the rotor
magnet is of elastic material. It is not so brittle and difficult to handle as
hard magnet material and adapts better to the rotor housing. Sometimes
the magnet m aterial is molded into the rotor housing. In the field of small
electrical machines there are often such constructions to be found which were
invented in former times for large machines, but because of their disadvan-
tages are no longer used and forgotten. Later they were rediscovered because
for small motors their disadvantages are not so serious compared to their
advantages 7[ ]. Other examples are the bell rotor and the skew winding used
in DC commutator motors and in BLDC motors too (Figs. 3.9, 3.2).
A h eart- lung machine can b e rep laced by two micro pumps as shown in
Fig. 3. 71. They are placed within a catheter in the left and the right cardiac
valve. They pump the blood during a heart operation. A flexible staff (spiral)
carries a ball on the top. At the bottom of the spiral there is the pump a nd
the motor within the pale plastic case. This is necessary caused b y a sep tic
Fig. 3. 71. Micro blood-pump and its placing in the left a nd right cardiac valve
demands because the blood flows along this case. By it the blood is warmed
up only about one degree. The three-phase two-pole synchronous motor has
an outer diameter of 6 mm and an output power of 3 W. It looks like the
motor in the top of Fig. 3.22. It is directly coupled with the pump rotating
at 35,000rpm. By this new method the opening of the breast can be made
very small so that the convalescence of the patient is faster and with fewer
problems.
References
1. Stepina, J. (1982): Die Einphasenmotoren, Springer-Verlag, Berlin/Heidelberg

2. Kenjo, T.; Nagamori, S. (1985): Permanent-magnet and brushless DC motors,
Clarendon Press, Oxford
3. Beisse, A.; Lebsanft, L. (1985): Betriebsverhalten permanenterregter Gleich-
strommotoren bei Verschiebung einer der Kohlebiirsten, Etz-Archiv H. 12, pp
389-394
4. T. J. E. Miller, T. J. E. (1989): Brushless permanent-magnet and reluctance
motor drives, Clarendon Press, Oxford
5. Nasar, S. A.; Boldea, 1.; Unnewher, L. E. (1993): Permanent magnet, reluc-
tance and self-synchronous motors, CRC Press, Inc., Boca Ration/ Ann Ar-
bor /London/Tokyo
6. Kenjo, T.; Sugawara, A. (1994): Stepping motors and their Microprocessor
Controls, 2nd edition, Clarendon Press, Oxford
7. StOlting, H.-D. (1998): Alles bewegt sich. Geschichte der Elektrotechnik, Bd.
16, Geschichte elektrischer Kleinmaschinen, VDE-Verlag, Berlin
8. StOlting, H.-D.; Kallenbach, E. (2002): Handbuch Elektrische Kleinantriebe,
2nd edition, Carl Hanser Verlag, Miinchen/Wien
9. Brosch, P. F. (2002): Moderne Stromrichterantriebe, 4th edition, Vogel
Buchverlag
10. Kallenbach, E.; Eick, R.; Quendt, P. (2003): Elektromagnete, 2nd edition, Teub-
ner, Stuttgart
4 Fluidpower Actuators
Wolfgang Backe and Andreas Klein
4.1 Introduction
In fluidpower actuators, the power of a liquid or gas flow is adjusted by

means of suitable devices. These devices are controlled by an electric input
signal, and transform the applied power into a specified mechanical variable
such as stroke (angle of rotation), speed (rotatory speed), or force (torque)
of a longitudinal or rotatory motion. The thermal energy of the pressure
medium is not utilized in any of these cases.
In hydraulics, relatively high pressures (up to 420 bar) and relatively low
flow velocities within the lines are used for energy transmission. However,
high flow velocities do occur, for example whenever there are great pressure
differentials at flow resistors or leakage gaps. These high flow velocities may
have undesirable consequences; e.g. throttle losses at leaks or impulse forces
that affect control elements.
Far lower pressures are usually used in pneumatics (6 to 10 bar). However,
much higher flow velocities within the lines are permissible.
In order to increase the accuracy with which fluidpower actuators adhere
to reference variables and compensate for any errors caused by disturbances,
the principle of feeding back the control variable (and, if necessary, further
auxiliary control variables) and constantly comparing it to nominal values is
employed.
The development of suitable controller concepts - often relying on digital
signal processing - has led to a significant increase in the speed and accuracy
of fluidpower actuators.
Fluidpower actuators are employed wherever large forces or torques have
to be brought to bear; e.g. according to a speed programme. By contrast,
pneumatic actuators are suitable for transporting low masses at high veloci-
ties, for example in the realm of transportation technology or in loading and
unloading production machinery.
Hydraulic energy can be controlled either through resistance (i.e. valve) or
through displacement control. One furthermore distinguishes between feeding
from a source with an impressed pressure or an impressed flow. Figure 4.1
shows the most important circuits of hydraulic drives.
Resistance control usually relies on an impressed pressure. The electrohy-
draulic valves used exhibit good dynamic performance, because low masses

156 Wolfgang Backe and Andreas Klein
Fig. 4.1. The most important types of electrohydraulic drives. a Resistance con-
trol with impressed pressure, b displacement control with impressed pressure, c
displacement control with impressed flow
4 Fluidpower Actuators 157
have to be actuated over short distances (10- 1 kg; 10- 1 to 10° mm). This
type of control has the disadvantage of high throttling losses at the valves'
control edges.
Displacement control is much more energy-efficient. Here, too, a valve con-
trol is required for adjusting the displacement volume; however, it is situated
in the signalling circuit, which has a much lower power level.
In the pump control (c), the displacement of the pump is adjusted while
the input speed remains constant. The characteristics of the impressed flow
determine the speed of the motor or the velocity of the attached cylinder.
The pressure varies with the load acting upon the motor or cylinder.
The dynamic performance of this type of drive depends on the dynamics
of the stroking unit and on the pressure build-up in the hydraulic capacity
of the liquid column between pump and motor.
In the displacement control with impressed pressure (secondary control),
a speed control ensures that the control motor is only stroked out as far as
necessary to overcome the load torque. In this case, the dynamic performance
is solely determined by the stroking unit for the displacement volume. As
the circuit design (b) requires a variable displacement unit, it can only be
implemented for rotatory movements.
Displacement units cannot be adjusted as quickly as valves, because
greater masses have to be moved over longer distances within them (10°
to 10 1 kg; 10 1 to 10 2 mm). The choice of an actuator's operating principle
needs to be based upon the specific requirements to be met. If dynamic per-
formance is paramount, valve control is the method of choice in spite of the
inherently high losses. If high powers are to be transmitted and good energy
efficiency is an important criterion, displacement control suggests itself [14].
4.2 Proportional Valves
Electrically controlled valves make it possible to combine a hydraulic power

system with an electronic control system into an accurate and highly dynamic
electrohydraulic drive system. They form actuators in which the power is
amplified from the low levels obtaining on the signal side into the high level
required for the power part. Generally speaking, valves are used for dosing
and controlling the flow direction of the pressure medium.
If controlling the flow direction of the pressure medium is all that is re-
quired, switching- and check valves are used. For reasons of energy efficiency,
it is important that little pressure be lost in flowing through such compo-
nents. This is an important selection criterion and a characteristic feature of
these valves.
Valves intended to dose the flow of the oil as well inevitably exhibit signif-
icant throttling losses. One distinguishes between two methods of influencing
the flow: steady valves are able to continuously adjust the flow, whereas un-
steady valves, mostly with fast-switching actuators, determine the flow by
varying the switching duration per time interval. In many applications, the
latter design would entail undesirably strong pressure pulsations, and hence
steady valves are primarily used.
4.2.1 Electromagnetic and Electromechanical Converters

as Pilot Stages
Electromagnetic or electromechanical converters are used to transform the

valve's electric input signal into a force and/ or a stroke to actuate the valve.
These converters adjust fiuidpower resistors in the form of control edges or
seat gaps. Other converter designs are hardly ever used, either because they
are incapable of generating sufficient forces or beset with other grave short-
comings. Figure 4.2 shows the electromechanic converters in common use
today with some of their characteristic values.
Proportional Moving coil Torque motor Linear motor

Magnet
Power
input (W)
Stroking 20 ... 1000 8 ... 80 2 ... 40 400 ... 2000
work
(Nmm)
Linearity 0.5 ... 6 1. .. 7 1. .. 2 0.5 ... 6
deviation
(%)
Frequency 10 ... 150 100 ... 200 100 ... 300 10 ... 200
limit (Hz)
Fig. 4.2. Common electromechanic converters
Apart from the torque motor, the proportional magnet, a derivative de-
velopment of the switching magnet, is becoming increasingly popular. Its de-
velopment made it possible to produce inexpensive proportional technology.
With special electrical controls, this type of magnet can deliver good dynamic
performance. However, it requires relatively high control power and can only
generate force in one direction, which means that it must always overcome
the resistance of a spring that generates the return force. Figure 4.3 depicts
the forces generated by proportional magnets of different size as a function
of their stroke.
The moving coil uses the electrodynamical principle. It can only generate
small forces, which is why it is primarily used in pilot stages only. Its advan-
Magnet size:
35 x 35 x 64 mm3 45 x 45 x 67 mm 3
80
N
.t. ~ ~
1\
60
..__
LL
Q)
~ -~ \
~
.,Q
40
1\ b .., - \
Q)
c
Ol
- f'l""
~
-~b \ ~ ~
\
"' \ \
''\ L~
::2 ~
""::::
20
"'
\
-
c
- ~ - _t-,? 1-- ,[\.., ~ - '-
I-
'--
0 2 4 6 mm B 0 2 4 6 mm B
Magnet stroke s Magnet stroke s
60 x 60 x 88 mm 3
160
120
bt-a-
-
-~ -- ...
\
a= 11im
b =0.6·/lim
\
LL
Q)
b c = 0.2·/lim
~
.,Q
Q) 80
~- --~ -- 1\
1\ \
c
Ol
"'
::2
\
--- -- "'
40
lhc '\
'-
......... ...___
1'- I--
0 2 4 6 mm 8
Magnet stroke s
Fig. 4.3. Force-stroke characteristic curves of various proportional magnets (ac-

cording to Binder)
tages consist in good dynamic performance, low control power requirements,

and the ability to exert both push and pull forces depending on the polarity
of the control current.
The torque-motor or linear motor conjoins the effects of permanent mag-
nets' and electromagnets' magnetic fields. Both generate great forces or,
respectively, torques in both directions and are highly dynamic. However,
their strokes or angles are low, and they are costlier to produce than propor-
tional magnets. Hence, these converters are mainly used in high-end valves.
Amongst more recent converter designs, piezoelectric actuators (cf. Sect. 5.4)
play an important role. In a stack design, they generate very large forces,
albeit only over small strokes. However, the use of piezoelectric actuators in
valve actuation has not become widely popular yet.
Figure 4.4 shows the work of stroking rate of the various principles of
valve actuation. Conventional converters generate inferior forces, but exhibit
working strokes of several millimeters, which is a necessary prerequisite for
directly generating a sufficient valve stroke.
• Linear motor
1000
Piezo
" ./·r I 1-
..
stack translator
#'Proportional
.9l Nmm
. "
##~ magnet
I!!
Ol ##
c:
. "
##
"
:l2 100
e
-
u;
0
~ . .... #
#
..;
.. ,:., ~ Moving coil
-
-
~ .~ :/
10
.,..,
I.,.
Piezo
. '\
Torque motor
I
disc trrslator
10 100 cm 3 1000
Volume of unit
Fig. 4.4. Work of stroking rate of different actuators for valves
4.2.2 Electrohydraulic Valves
Owing to the convergence of their technical characteristics, it has become

quite difficult in recent times to distinguish between proportional- and servo
valves. Servo valves developed as long as fifty years ago for the requirements
of the aerospace industry typically feature a highly dynamic pilot stage in
conjunction with a high-quality main stage design, the latter usually involving
the control edges being integrated into a steel sleeve.
For the less rigorous requirements of industrial hydraulics, robust pro-
portional valves were developed in the nineteen sixties. They frequently owe
their genesis to the combination of switching valves with proportional mag-
nets. Both in terms of their static and dynamic performance they are inferior
to the more sophisticated servo valves. In order to be able to cope with the
increasingly demanding control tasks in industrial applications, two modifi-
cations were subsequently introduced. On the one hand, servo valves were
rendered more robust by using pilot stages less sensitive to contamination
and by sometimes dispensing with the steel sleeve; upon the other hand,
the static and dynamic performance of proportional valves were improved
through the use of high-performance control electronics for the proportional
magnet and of sleeves. The differences between these designs thus having
been eroded, it makes sense to use the more encompassing term 'steady valve'
which describes their operating principle. However, the terms servo valve and
proportional valve have become so entrenched that we shall continue using
them in the following.
Depending on their purpose, one distinguishes between several types of
steady valves: steady directional valves (y rv i), steady flow valves (QL rv i),
and steady pressure valves (PL rv i). Amongst these, steady directional valves
are the most important in the realm of actuators.
Servo Valves
Design and Operation. The low input power (0.02 to 4 W) typical of servo
valves is amplified by means of one or more hydraulic pilot stages. This makes
it possible to overcome the flow forces acting at the hydraulic main stage,
and to achieve a high amplification of power (> 108 ).
In the main, two principles are used for converting the mechanical output
signal of the electromechanical converter into a hydraulic signal: throttling
the oil flow with variable mechanical resistors, or converting the kinetic energy
of an oil flow into static energy. Figure 4.5 schematically depicts the three
most commonly used designs.
The flapper-nozzle system and the jet pipe principle are preferentially
used as a first amplifier stage, because of the low moving masses involved.
However, these designs will inevitably cause an incessant flow loss due to their
principle of operation. Owing to its larger flow cross-section, the movable
jet pipe is considered less susceptible to contamination. In the area around
the middle position, a good linearity between the jet pipe's stroke and the
pressure differential existing between the control ports can be achieved.
The flapper-nozzle system is normally integrated into the resistor circuit
with two constant orifices shown in the illustration. Through the stroking of
the flapper, the two flow resistors arranged in opposite directions are actu-
ated. The pressure differential created, with which the next valve stage can be
actuated, also exhibits a linear stationary transfer characteristic over a broad
range.
The overall functioning of a servo valve shall be explained using the valve
depicted in Fig. 4.6 as a representative example of the many different designs.
The flapper-nozzle system is put out of balance by stroking the torque-motor
against a bending pipe and feedback spring. The pressure differential resulting
at the area of the main piston causes the piston to be accelerated against the
impulse forces of the flow and the feedback spring's force, both of which
increase with stroke. At the same time, as the piston strokes the bending
pipe and the feedback spring, the flapper is returned towards its original
position until torque balance with the torque motor is achieved, and thus
a reduction in the pressure differential at the piston. In a stationary state,
the pressure differential only acts against the flow forces.
Flapper nozzle
Stroking of spool by proof flapper by Stroking of jet pipe by

portional magnet torque-motor or moving torque motor
coil
JZfd =4 .. . 12mm JZfd = 0.25 . .. 0.5 mm JZf di = 0.12 .. . 0.2 mm
Ymax = ±(1. . . 4) mm Ymax =± (60 . . . 75) ~-tm Ymax = ±0.47mm
Po = ... 350bar Po = ... 350bar Po = ... 210bar
Qmax = 5 ... 2001/min Qmax = 0.3 . . . 2.51/min Qma x = 0.1. . . 2.51/min
Fig. 4.5. Design of mechanical-hydraulic converters
Armature
Bending pipe
Flapper
Nozzle Feedback spring
Filter
Fig. 4.6. Two-stage ele ctrohydraulic servo flow valve
The main stage is designed as a spool valve with four variable resistors.
The design of this two-stage valve is relatively complex, in particular due to
the low permissible dimensional tolerances.
Instead of the aforementioned mechanical position control with torque
balance, many servo valves feature an electric position feedback of the main
piston and an electronic controller. This operating principle, which is the
norm for servo valves with three or more stages, shall be treated of further
below in the context of a proportional valve.
Static and Dynamic Characteristics. In evaluating the static and dy-
namic characteristics of steady directional valves, one usually considers the
1.0
oiJ o.75
~
~ 0.5
Q)
E
::J 0.25
~ 0.5 ~
~ \<:'
0.25 ~0
e,'S
~
Hydraulic zeropoint ,~~
Fig. 4. 7. First quadrant of a steady directional valve's ideal performance map
flow-signal function, the pressure-signal function, the quiescent flow function,

the step response and the frequency response of the valve. These data, gath-
ered under defined conditions, allow one to derive characteristic values, with
the aid of which the valves can be compared in terms of their operating area
and their quality.
The parabolic flow characteristic describes the strongly non-linear perfor-
mance map of a valve with control edges exhibiting zero overlap (Fig. 4.7).
Equation (4.1) has already used the flow amplification, which indicates the
gradient of the flow signal function (Fig. 4.8) as a crucial valve characteristic.
The flow signal amplification is determined with connected load ports and
a valve pressure differential of Po = Pnom = 70 bar.
Po - pL(sgn i)
(4.1)
PNom
QL flow through the valve,

PL load pressure; difference between the port pressures
±(pA- PB),
Po system pressure,
Pnom nominal pressure (for servo valves: 70 bar; for pro-
portional valves: lObar),
i electrical input current applied to the valve,
VQi flow signal amplification for PL = Obar; Po= Pnom·
In order to obtain a high resolution in the valve's small signal range,

design elements intended to achieve progressive opening characteristics are
integrated into some valves. These valves are characterized by a broad control-
lable flow range(> 105 ). Their non-linear characteristics can be compensated
for by suitable measures on the signal side.
In regard to a controlled electrohydraulic drive's high load stiffness, the
pressure amplification Vpi is of great importance. It is defined as the gradi-
ent of the load pressure signal function and is measured at the valve with
blocked load ports (Fig. 4.8). In high-quality valves, a load pressure change
of pLfp0 =50% can be achieved with an input signal current of as little as one
per cent i/inom· The more precise the control edges of the spool are adjusted
to those of the spool, with as little radial clearance as possible, the more cer-
tainly the desired high pressure amplification can be achieved. Wear at the
valve's control edges manifests itself in a gentler slope of the pressure-signal
function.
As a further indicator of quality, the hysteresis value can be seen from
this function. Hysteresis induces an undesirable control deviation in the closed
superordinate control loop, and therefore ought to be minimized. A proven
countermeasure is to superimpose a high frequency alternating signal, the
so-called dither, onto the input signal proper. Typical values for the range
of the hysteresis signal are (0.01 ... 0.03) i/inom· With an electrical position
control, values smaller than 0.0001 i/inom are feasible.
The quiescent flow function is also determined with blocked load ports. It
yields information on leakage of the valve and thus, possibly, on the valve's
wear condition. In the performance map in Fig. 4.7, an operating point (OP)
has been randomly chosen, and the flow pressure amplification belonging to
that point is depicted as a tangent at this point. This characteristic value is
crucial for the damping behavior of a valve-controlled hydraulic drive. At the
hydraulic zero point (i=O, PL=O, QL=O), the following relationship obtains
between pressure- and flow amplification:
_Q_
Onom
0.5
50% i 100% -10% -5% 5% i 10%

inom Inlet and outlet inom
tan a= V0 ; edges~symetrically-0.5 tan (3= Vp;
underlapped
_1 Volume flow-signal Pressure-signal
amplification -1 amplification
Load pressure-
Volume flow-signal function signal function Pressure-signal function
Fig. 4.8. Static characteristic curves and characteristic values of servo valves
(4.2)
In assessing the dynamic characteristics of servo valves, the step responses

of the piston position and frequency response measurements are considered.
The step responses depicted in Fig. 4.9 show the role of step height and
system pressure. Whereas the valve behaves like a linear system and follows
an e-function when the input signal's steps are small, larger nominal value
amplitudes (in this case, i/inom > 0.25) cause the flow saturation of the
flapper-nozzle system to limit the piston's velocity. Moreover, with increasing
system pressure, the flow amplification of the pilot control stage is augmented,
which causes both the loop gain of the valve's internal control circuit to be
increased and the speed limit to be shifted towards higher values. In order
to sustain high and predictable valve dynamics when system pressures in the
power part are (temporarily) low, many valves offer the possibility to connect
a separate external pressure supply to the pilot stage.
In valves without an electronically position-controlled piston, determin-
ing the dynamic response characteristics is complicated, because commonly
available flow sensors are incapable of tracking the flow emanating from the
valve with the requisite speed. Suitable test rigs mainly involve a smoothly
running piston drive, with small moving masses and low leakage, actuated by
the servo valve to be analyzed. Its piston velocity is proportional to the flow
Q0 passing through the valve's main stage.
These devices are preferentially used to track frequency responses by ex-
citing the servo valve with sinusoidal signals and recording the amplitude
and phase of the response signal. Results are usually plotted in the form of
a Bode-diagram, as shown in Fig. 4.10 for a servo valve. Characteristics stated
include the -3dB-frequency (the frequency at which an amplitude drop of
approx. 30% occurs) and the -90°-frequency (the frequency at which a phase
lag of -90° occurs). However, one must bear in mind that these characteris-
tics are strongly dependent upon the system pressure and the amplitude of
the input signal i/inom· There are no norms for the parameters under which
frequency responses are to be measured, which means that the data given
in different manufacturers' catalogues cannot always be compared like with
like. The input signal's amplitude must be stated in any case.
The diagram in Fig. 4.11 presents an overview of the dynamic performance
characteristics of some commonly-used servo valves.
Mathematical Description of Response Characteristics. Equation
(4.1) has already described the static transmission characteristics of servo
valves. It is valid for valves with zero overlap and linear flow characteristics.
Giving an exact description of the valves' dynamic characteristics is difficult,
due to their complex structure. For designing a servo-controlled drive, it is
often sufficient to consider the valve as a proportional transmission element
with a second-order delay
1.25
System pressure p0 = 70 bar - - -
140 bar--
- . -1-·-=1.0
(_'7"
1.00 ---
~IJ 0.75
0.175
If I"'
r
Q) I
Jso
-"'
e I
I I
-- -l.
ii)
---
I
Q)
> 0.50
"
v
~
0.25
0.25 -- -- -- - --- -
0
0 10 20 30 40 ms 50
Timet
Fig. 4.9. Step responses of a two-stage servo valve
y
(4.3)
and to determine its characteristics empirically. For large valve input signals,
one must bear in mind that the dynamic transmission characteristics are
largely determined by the pilot control stage's flow limitation described fur-
ther above. The -3dB-frequency can then be approximated from the piston's
maximal velocity (Ymax):
f - 3 db Ymax
= 4 . 0 707 . . _i-
(4.4)
' YNom iN om
As a rule, the electric pilot stage's time constant can be disregarded,

because of the coils' low inductivity. This is particularly true if, as is com-
mon practice today, the control system uses a voltage-current-transformer.
Figure 4.12 depicts the simplified signal flow diagram of a servo valve.
1.6 -180°
0 1.0
II 0
~ II 0.8
.....
~
d' ·-
~
0.6
S-
<1=
c 0.4 -90° EC/)
0
~ Q)
= 0.5
C/)
~ -._i_ <ll
..c:
Q)
""0
/nominal a..
:ec. 0.2 -45°
E p0 = 70 bar
<( 140 bar
280 bar
0.1 oo
5 10 25 50 I 100! Hz 250
t3dB ~90'
Frequency f
Fig. 4.10. Bode-diagram of a two stage servo valve
Proportional and Control Valves
Proportional valves specifically designed for use within electrohydraulic con-

trol circuits are sometimes sold as control valves [38].
Design and Mode of Operation. The component characteristic of hy-
draulic proportional valves is the proportional magnet. As the current-force-
characteristics are independent of the stroke, if the stroke occurs against
a spring, a good proportionality between input current and piston position
can already be achieved in the open-loop control. Through this simple real-
ization of a steady actuation of valves with significant hydraulic power, the
performance of many hydraulic drives, including less sophisticated drives,
has been considerably improved. This trend is furthered by the circumstance
that proportional valves have the same hole patterns as switching directional
valves.
The most common type is the 4/3-proportional-directional valve. The
figures 4/3 indicate the number of connectors (4) and the number of possible
switching positions (3). In the valve shown in Fig. 4.13, the steady stroking
of the piston occurs in an electric position control.
The position signal determined by means of an inductive stroke measuring
system is compared to the valve's input signal. The controller converts this
difference into an actuating signal and actuates one side of the double-acting
proportional magnet (depending on the polarity) via an amplifier. The po-
sition control circuit largely compensates for any disturbance variables such
as flow forces or friction forces. This has a particularly beneficial effect on
the hysteresis and the response threshold. As is the case with servo valves,
Characteristic term Typical values

Nominal flow Qnom 0.6 ... 1000 !/min
Nominal current Inom 0.005 ... 4A
Flow amplification VQi < 105 1/min/A
Pressure amplification Vpi < 670bar/A (for Inom=0.03A)
Flow-pressure amplification VQp 0.5 ... 10- 4 1/min/bar (for hydraulic
zero bias)
Hysteresis 0.1. .. 5%
Temperature drift 0.02 ... 0.06 %/K
Leakage flow 0.2 ... 81/min
Natural angular frequency
W-3db see diagram, w = 27r f
w_ 90 o
Damping degree Dv 0.4 ... 0.7
400
Hz f. goo -.-1-·- = 100%
<:--" --
200 I nom
/
100 '-3db Supply pressure Ps = 210 bar
..... '
>-
u 60
-.....-- -
c: 40
Ql
:J
0"
~
LL
20
10
6 Load pressure: pnom =70 bar r-
4
6 8 10 20 40 60 100 200 400 1/min 1000
Nominal volumeflow Onom
Fig. 4.11. Dynamic characteristics of servo valves
Fig. 4.12. Signal flow diagram of a servo valve
proportional valves are often arranged in multiple stages whenever great hy-
draulic powers have to be controlled.
Figure 4.14 shows an electrically position-controlled multiple-stage con-
trol valve of Nominal Size 16, where the size indicates the diameter of the
valve's connectors. As a pilot control stage, a directly actuated control valve is
Control electronic:
- evaluation of position
sensor-signa l
- power stage for
proportional magnet
- position control
for valve spool
a A P B T
1.0
0.6
I I I I I
t 0.6
E 0.4 -
~0.2
I I I I I
Reve~al erro~ /
~ ..!
0 0
II'"
0-0.2
-0.4
v
-0.6
/
-0.6
-1 .0
-10 .a -6 -4 -2 0 2 4 6 v 10
vin-
b c
Fig. 4.13. Single-stage 4/3 proportional directional valve (Bosch). a D esign, b

symbol, c flow-signal function
used, which is also electrically position-controlled. It is equipped with a pro-

portional magnet acting against a spring.
Static and Dynamic Characteristics. Proportional valves are assessed
with the same test procedures and evaluation criteria as have already been
described for servo valves. The only difference to be noted is that , in deter-
mining the flow signal function according to relevant norms, a valve pressure
drop of Pv = 10 bar is set. The reason is that the opening strokes for propor-
tional valves are measured in millimeters, whereas they are only a few tenths
of a millimeter for servo valves.
Typical of many proportional valves is a dead zone in the flow signal func-
tion around the hydraulic zero point. This is because the valve has a positive
overlap, which ensures that, should a power failure occur, the load ports are
Fig. 4.1 4. Two-stage NS 16 control valve (Bosch)
400
Hz -.-i-= 100%
200 1nom
f.oo•
I
Supply pressure p 5 =140 bar
..... 100
>.
()
c: 60
~ ~- ----- -- \
Q)
::J 40 ~
CT
~ ~
LL
20 ' ,_
10
------ _L '-306
- -
6
4
Load pressure Pnom = 10 bar -
6 8 10 20 40 60 100 200 400 1/min 1000
Nominal volume flow Q

nom
Fig. 4 . 15. Dynamic characteristics achievable by proportional valves
locked in a fail-safe po~:>ition (in thi~:> case, the middle position) . A function
generator on the control card can compensate for this dead zone.
The inferior dynamic performance of proportional valves relative to that of
~:>ervo valve is abo due to their longer strokes. Figure 4. 15 shows an overview.
M a the matical D escription of the Tra n sition B e h avior . Unlike servo

valves, the valve's dynamic characteristics are not primarily determined by
the saturation of the pilot stage, but also by the proportional magnets' force
build-up, which is delayed relative to the input current . However, due to eddy
currents and other confounding effects, the relationships are so complex that,
just as with servo valves, it is best to describe them in terms of a PT 2 element
and the limit of the piston speed as a simple approximation.
4.2.3 Electropneumatic Valves
Steady valves in pneumatics can be sorted into two main categories:

directional or flow valves,
pressure control valves.
The first type in particular has assumed an important role in pneumatics
in recent years, because the use of modern control concepts has rendered it
possible to realize steady controls of electropneumatic drive systems.
Steady Pneumatic Directional Valves
The directional valves hitherto employed use a sliding piston with sharp con-
trol edges as the control element, which seals metallically through a tight
fit. Most valve designs use a cylindrical spool piston. Figure 4.16 shows the
operating principles of the most common valve designs.
One basic distinction is that between one- and two-stage valves. In two-
stage valves, the motion of an electromagnetic converter in a pilot stage is
transformed into a pneumatic pressure sig nal. The main advantages of these
systems consist in their compactness and low weight, because t he pneumatic
pilot stage has a greater power density than the electromagnetic converter.
Their key disadvantages consist in t he permanent energy loss due to the pilot
Nominal size 4. 5
Si n g le-stage - 90° - freq u ency 140Hz
wit h torque
Powe r input 2.7W
m o tor
Weight 1.5 kg
-
tA Nominal size 5
D o uble-stage - 90° - fre que n cy 100Hz
with moving Power input 1.5W
coil
W e ig h t 0 .14 kg
R~ fP
~
Single-stage Nominal size 6
wi t h -90° -frequ ency 7Hz
proportional Powe r in put 13W
m agn et W eigh t 0.6kg
RAP
Fig. 4 .16. Basic design of steady directional valves for pneumatics

air, as well as their greater sensitivity to contamination, and hence lower

operating safety, due to the narrow orifice cross-sections.
The rotary valve represents a different type of steady pneumatic valve.
Such valves are employed within the position control and hence exhibit ex-
cellent static and dynamic characteristics. Their main advantages are that
the valves consist of few components, are small, and consume little electrical
power. Figure 4.17 shows a sketch of such a valve.
~----------93 --------~
Spool Housing Motor Angular sensor
Nominal size 4 6 8
-90°-frequency 70Hz 70Hz 50Hz
Power input ca. 1. .. 5 W (max. 13 W)
Weight 0.14 0.14 kg 0.16 kg
Fig. 4.17. Pneumatic rotary valve (Kolvenbach KG)
The steady-acting miniaturized 2- or 3-way valve shown in Fig. 4.18 (di-

mensions ; : : : 10 x 18 x 49 mm 3 ) possesses a piezo bending element as an
electromechanical converter. With a supply pressure of 1 bar, it can steadily
actuate flows of up to 21/min (port 3 shut) or pressures of up to 1 bar in
three-way switching mode, by changing the control voltage from 0 to 40 V.
In choosing a suitable valve, not only the valve dynamics but also the
maximal flow through the valve matters. Methods of calculating the latter
will be presented in the following.
Unlike hydraulics, pneumatics involves a compressible pressure medium,
which means that flows on the high- and low-pressure sides differ. The
pressure-dependent flow characteristic can be depicted as an elliptic function.
Fig. 4.18. Miniaturized steady 2- or 3-way valve (Burkert)
The critical pressure ratio a, which is incorporated into the flow equation as
a parameter, varies between 0 and 0.528, with the higher value being achiev-
able with an ideal orifice. The a-value expresses the ratio of pressure upstream
and downstream of the valve at which a maximal flow occurs (reaching the
speed of sound). Figure 4.19 demonstrates various flow characteristics.
The equations for calculating the flow through pneumatic valves are given
in Table 4.1. Equation (4.8) can be used to determine the flow for the high-
pressure side of a control edge. Deviations of the overcritical flow of a nozzle
relative to that of an ideal orifice, caused by jet contractions, are considered
via the factor Wa (0 < Wa < 1). Equation (4.10) represents the simplified
formula for calculating the actual flow on a valve's high-pressure side.
Measuring point
for nominal flow
~-·--·----:-..::_-;----
1 1---· ----
1 ----.----
....... .......
~--b1=0 . 528 .....................~-
:---b2=0.2 . :,
1-·-b
I 3 =0 ""·
0
b1 6/7
Pressure ratio a = E2.a.h..
Plabs
Fig. 4.19. Pneumatic flow characteristics of orifices and nozzles

Table 4.1. Flow equations for pneumatic valves
1. Pressure relation a = E.2.a.ha.

Plabs
(4.5)
v1- 1
2. Pressure-dependent flow characteristic
w(a) = t~=:?~ ;b<a< (4.6)
w(a) = 1; 0 <a< b (4.7)
3. Volume flow at one control-edge
High pressure side Q1 = aoAvllimax v'2Rih w(a)
Av: valve opening
Vl = WmaxV2 R 273 K (4.8)
= 198.4 ~ (4.9)
Q1 = aoAvv1/ifj;w(a) (4.10)
Low pressure side Q2 =an A'(;v 1 /ifj; w(a) (4.11)
For pneumatic valves, the nominal flow is often given as a characteristic

value. It is measured with a supply pressure of 6 bar above atmospheric pres-
sure upstream of the valve and a pressure differential of 1 bar across the valve.
Pneumatic Proportional Pressure Control Valves
Pressure control valves are intended to maintain a preselected load pres-

sure, regardless of the input pressure or the air consumption. By selectively
throttling the flow, the electrically controlled pressure control valve converts
an electrical signal into a pneumatic pressure downstream of the valve. The
nominal value is usually impressed by changing the control signal's voltage
(0 to lOVolts) or amperage (0 to 20mA). However, there are valves available
which possess an integrated D /A-converter as a digital interface. Based on
their method of controlling pressure, electropneumatic pressure controllers
can be divided into two categories:
1. Systems in which the pressure to be controlled is mechanically measured

as a force, by means of a pressure-loaded area, and compared with a force
that can be adjusted through electrical signals.
2. Systems in which the pressure to be controlled is measured with pressure
sensors and in which the valve is electronically controlled in the pressure-
control circuit.
Systems using mechanical pressure control tend to be simpler and hence

cheaper than those with an integrated pressure sensor. However, owing to
mechanical hysteresis, their control accuracy is lower.
Figure 4.20 shows the design of a pressure control valve with a propor-
tional magnet, which serves to build up a reference force proportional to the
pressure. In general, as disturbance variables (such as friction or mechani-
cal hysteresis) are not compensated for, these valves exhibit a hysteresis of
10
bar 10 r----=:::::r---~-----,~--:--~--,
bar
8
~ (I) 8 k----+---+...::O..oc-
::l
::;
"'"'~
c.
6
"'~ 6
"5 4 c.
c. ~ 4 ~--~--~~~~~--~~~
"5 :;
0 2 o2
0
400 800 1200 mA 1600 0
Magnet current
Volume flow
Fig. 4.20. Proportional pressure control valve (Herion)
approximately 5 %. As the illustration shows, pressure drops with increasing

flow through the valve, because the valve works by comparing forces and
because t hrottling losses cannot be subsequently compensated for.
Figure 4.21 depicts a valve with an electronic pressure control circuit.
In this case, independently of the control variation of the pressure, control
electronics actuate a piezo bending converter, which actuates the valve spool
t hrough a pneumatic pilot stage. Owing to the electronic control, valves like
these exhibit hysteresis of less than 1 %. The flow characteristics curve ex-
hibits a constant pressure until the valve is fully opened, because the valve is
actuated through the direct electronic pressure control up to the point where
it is fully opened.
A further design of electronically controlled pressure control valves is
shown in Fig. 4.22. In this case, the actuator is actuated through being loaded
or unloaded with pressure by two fast-switching solenoid valves. The solenoid
valves are controlled by the valve's pressure-control loop, such that the out-
put pressure tracks the nominal signal. Owing to their integrated pressure
sensor, t hese pressure controllers exhibit a ow l hysteresis of approximately
0.5 to 1 %. Owing to the valves' pulse-modulated actuation, their response
times are in the range of 100 ms to approximately 1.5 s. These valves are pri-
marily used in applications where the flow changes slowly. Because of their
v '"'\
5
v
bar
~ 4
:0
U/
......_
~ 3
v~
a.
g_2
5
01 v
lL:
0 2 4 6 8v 10 0 200 400 1/min 800
Command voltage Volume flow
Fig. 4.21. Pressure control valve with pressure sensor and piezo bending converter
(SMC)
pneumatic amplifier stage, such pressure control valves are offered in nominal
sizes of up to 25 mm or more.
4.3 Switching Valves
Fast-switching valves open or close constant hydraulic resistors for vari-

able time intervals. They are usually controlled by means of a pulse-width-
modulated signal, in which the duration of the On-signal is proportional to
the actuation signal within a fixed cycle frequency [27, 28, 32].
Real-life valves cannot t rack this signal with t he requisite speed, due to
inertia and delays in the force build-up. However , avoiding large pressure
pulsations through high cycle frequencies and high resolutions necessitates
the use of valves that switch as quickly as possible.
The key reason for using switching valves is their reliability, due to their
simple design and low production costs.
Supply voltage
-+
Amplifier
Command value
Fast-switching
2-way valves
Feedback signal
>--+-----t-+----<~ 1-----+<>-----f--A. Pout
f----,,-----+-1> Exhaust
R
Actuator
10 .--.---.--.--.~~ 10
bar ba rr--
8
~ 8
:::J
-
~ 6 1---+---+--+-~'------11-------l
c.
"5
- ---
~ 4 1----+-~~-~---T~~~--1
0
2 4 6 8 v 10 0o 25 50 75 100 m3th 150

Command voltage Volume flow
Fig. 4.22. Electropneumatic pressure control valve with fast-switching valves (Hon-
eywell Lucifer)
In the layout of drives actuated by fast-switching valves, one often en-

counters a conflict of aims between maximal speed and achievable accuracy.
High speeds necessitate high flow coefficients, great accuracy requires low
reproducible flows, which in turn are increasingly difficult to achieve with
increasing flow coefficients.
4.3.1 Electrohydraulic High-speed Valves
Electrohydraulic high-speed valves currently commercially available have

mainly been designed for use in anti-lock braking systems. Table 4.2 lists
the performance data of various manufacturers' valves.
All valves listed are controlled by a switching solenoid as an electromag-
netic converter. Faster switching valves for this purpose with piezo control
are not yet available.
Figure 4.23 shows the cross-section of a high-speed valve. Owing to the
compensation measures integrated, it exhibits short and almost pressure-
independent opening and closing times. Apart from a static pressure balance,
the way in which the flow is directed largely compensates for flow forces.
178 Wolfgang Bach~ and Andreas Klein
Table 4.2. Overview of high-speed switching valves
SG/SO SSV88 MV2.02 MV3.01

Producer ContiTeves Rexroth Bosch Bosch
Design 2/2 2/ 2 3/ 2, 3/3 3/ 2, 3/ 3
Current loss opened/closed opened/closed
Max. pressure 180bar 200 bar 200bar 200 bar
Switch-on time 3 . . .5ms < 2ms 3 .. . 5ms 4 ... 7ms
Switch-off time 3... 5ms < 2.5ms 4 . . .7ms 6 .. . 9ms
Max. flow ca. 41/min 51/min ca. 61/min 8.51/ min
at p = 150bar 150 bar 150 bar 150 bar
¢orifice 0.84mm l.Omm
Application ABS ABS ABS, ASR ABS , ASR
Actuation el. Controller spec. IC el. Controller el. Controller
Pressure comp. no yes no no
Single-stage yes yes yes yes
Scaling element ball cone ball ball
·4·
~B
Fig. 4.23. High-speed switching valve (Continental Teves)
4.3.2 Electropneumatic Switching Valves
Switching valves only have a finite number of stable valve positions. That
means that the hydraulic characteristics pressure a nd flow can only be influ-
enced in regard to whether or not they exist at a particular load p ort. An
infinitely variable dosing of the flow, as with steady valves, is impossible. The
m ain application of switching valves is the shifting or r ele ase of flow c hannels .
As Fig . 4.24 shows, switching valves have a number of key features.
Characteristic valve Example

Usual design or size
performance #:1~+ I~T~Gt ~
mechanic, pneumatic
actuation electric
electric
l l l
number of pneumatic
2-3-4-5 5
ports
l l l
number of valve
2-3-4 3
positions
l l
electrically actuated
flow /nominal size NS 1.5 ... 25
5 /3-directional valve
Fig. 4.24. Classification and characterization of switching valves
Mechanically actuated switching directional valves are frequently used to

determine the positions of drives or work pieces in sequence control technol-
ogy, for example. Electrically actuated switching directional valves are used
for start/stop, as well as to change the moving direction of pneumatic drives.
Apart from the method of actuation, the number of ports and the valve
positions, the maximal flow that can be switched is an important criterion in
selecting valves. The flow through a valve is described by means of the stan-
dardized nominal flow. That is the nominal flow, calculated for conditions
of 0 °C and 1.013 bar atmospheric pressure. The nominal flow is the flow at
the suction state, measured with an absolute pressure of 7 bar upstream of
the valve and 6 bar downstream of the valve, and an ambient air temper-
ature of 20 oc (VDI 3290). The nominal flow is usually stated in standard
liters per minute or standard cubic meters per hour (Nl/min or Nm 3 /h). The
method for calculating the flow through the valve has already been presented
in Sect. 4.2.3.
In practice, a commonly used variable for characterizing the potential flow
through a valve is the nominal width of the valve. This refers to the inner
diameter of a tube, the area of which corresponds to the smallest cross-section
of the flow channel within the valve. Unlike the standardized nominal flow,
that variable gives only a rough indication of the valve's flow capacity, but
it is easily grasped and hence popular. Figure 4.25 depicts the relationship
between a valve's nominal width and its standardized nominal flow.
The valves' pneumatic ports usually take the form of screw threads, in
which screw connections are inserted. For economic reasons and because this
allows for a simpler and quicker assembly, valves are frequently mounted on
base plates today, such that hoses and piped do not have to be disconnected
whenever a valve is replaced. Figure 4.26 shows such a battery arrangement.
The base plates are standardized (ISO 5599), such that different manufac-
10000
;:: Nl/min
0
c;:: 1000 I--
roc:
.E
0 100
c:
-o
I/
Q)
!/)
~
10
"'c:
-o I
"'
U5
2 4 6 8 10 12 14 16 18 20 22 24
Nominal width
Fig. 4.25. Relationship between nominal width and standardized nominal flow
turers' valves with the same function and the same nominal width can be
mounted onto them.
Depending on their nominal width and control power, the valves' switch-
ing times range from approx. 8 to approx. 70 ms. However, as switching valves
are of simpler design and less sensitive than steady valves, a trend towards
developing very quickly switching valves for pneumatics has emerged in re-
cent times. These allow for quasi-steady controls of pressure or flow to be
realized.
The piezo pilot stage of the 3/2 directional valve depicted in Fig. 4.27 has
much lower switching times (:::; 2 ms) and very low switching energy require-
ments (0.014mWs). The pilot stage valve is supplied with a constant pressure
of 1.2 bar via a pressure controller and a fine-mesh filter. Upon switching, the
piezo converter causes a pressure to be build up above the membrane of the
control spool, switching the latter in such a w ay against the force of a spring
that P is connected to A. Valves of this kind are available for pressures of up
to 12bar and nominal flows of up to 1101/min.
Fig. 4.26. Battery arrangement of pneumatic valves

4 Fluidpower Act uators 181
Switch-off position Switch-on position

1: P, 2: A, 3: R
Fig. 4 .27. 3/2-directional valve with piezo pilot stage (Hoerbiger)
4.4 Fluidpower Motors
Fluidpower motors are used to transform hydraulic or pneumatic energy into

mechanical energy. Depending on whether a turning or linear movement is
created, one distinguishes between rotary or translatory motors.
4.4.1 Hydraulic Motors
Rotary Motors
The function of hydrostatic rotary motors is based on the displacement princi-

ple. Depending on the geometry of the displacement volume, one distinguishes
between piston motors, vane motors, and gear motors.
If, in a piston motor, the pistons are arranged axially with the drive shaft,
it is called an axial piston motor; if they are arranged radially, a radial piston
motor.
Contemporary motor designs are listed in Table 4.3. Each design has
specific properties rendering it suitable for specific applications.
Table 4 .3. Types of rotary motors

Schematic depiction Characteristics Design Displace- Rotary Working
m e nt speed pres-
• •
volume, range, sure,
cm 3 rpm bar
universally applicable ; constant 25 ... 800 750 . .. . .. 400

very good efficiency; motor , 8000
efficiency in wide range, variable
not much dependent on motor
pressure, speed and
Q torque; suitable for high
performance , typical
high speed motor
like swashplate motor, constant 25 ... 800 ... 8000 ... 400
suitable for low speed , motor,
high starting torque variable
motor
Swashplatemotor universally applicable, constant ... 100 ... 3000 ... 100
(rotating swashplate) very good efficiency, not motor
for higher speed because
of imbalance by rotating
swash plate
.I
Radialpiston motor universally app licable, constant 5 ... 7000 500 . . ... 350
(internal fluid exchange) very good efficiency, motor, 3000
s uitab le for hig h variable
performance motor
. I
Radialpiston motor universally applicable, constant 5 ... 7000 ... 2000 ... 200
(external fluid very good efficiency, motor ,
exchange) specially s uitable for variable
low speed and high motor
torque, typical low
speed motor
.I
Vane motor middle powerrange, constant 5 ... 2000 ... 3000 ... 200
quiet motor,
variable
motor
Rollingvane motor middle power range , constant 8 ... 1600 0.1. .. . .. 250
quiet, low degree of motor 2500
~
irregularity, suitable for
low speed, specially for
servo drives
Table 4.3. (Continued)
Schematic d e piction Characteristics Design Displace- Rotary Working

me nt s p eed pres-
volume, range , sure,
cm 3 rpm bar
Gear motor middle powe r range , constant 5 .. . 300 200 ... ... 280
•
simple d e sign , efficiency motor 3000
in wide ra nge n ot much
dependent o n pressure,
' s p eed and torque
e
Annulargear motor quiet , middle power constant 50 ... 900 10 ... . .. 250
range , suitable for low motor 1000
speed and high t orque
Translatory Motors
Translatory hydraulic motors are either single- or double-acting. Single-acting

cylinders are stroked out hydraulically, and exterior forces (weight force ,
spring force, opposing cylinder) take care of the return. In double-acting cylin-
ders, motion in both directions is created hydraulically. Figure 4.28 presents
an overview of the principal cylinder designs.
Cylinders with a single-sided piston rod offer the advantage of lower space
requirements for the same stroke, as compared to those with double-sided pis-
ton rods. However, even with the same flow, stroking speeds differ, depending
Single acting (plunger)
Double acting (differential piston)
Double acting (usual a= 1)
Fig. 4 .28. Types of translatory motors

on the direction of the stroke. This can only be avoided by taking specific
measures in terms of the control, or through the use of a symmetrical, double-
acting cylinder.
Static and Dynamic Behavior
In the design of hydraulic drives, rotation motors' displacement volume V2

and translatory motors' piston surface area A 2 needs to be determined. This
process takes into consideration the static and dynamic characteristics of the
motor. The equations required for calculating it are listed in Table 4.4 for
a cylinder with symmetrical surface areas and for a rotation motor.
Table 4.4. Design equations for hydraulic motors

Translatory motor
72,A2f x2jl'i! '

T T
PA' "o It 02 lp6 , "o
Force/torque M2th = ~ · (PA- PB)
Speed/rotation
n2 = 2.:w..
v2
speed
Acceleration ca-
pability
Nat ural angular
frequency WH = A 2 .J2·Eqj!
rn2·Vo
Hydr.-mech. effi-
ciency T]2hrn
Volumetric effi-
ciency T/2vol
Q2 = -172vol
1- 0
Q2th
Volumetric losses
Apart from the achievable force and the speed (or torque and rotatory
speed), the motor's acceleration matters, which is determined by both the
attached mass m (or J) and the pressure differential p A - PB applied to the
motor.
Theoretical values are reduced both by hydraulical-mechanical and vol-
umetrical losses. Hydraulical-mechanical losses comprise all friction forces
and torques. The term volumetric losses refers to the pressure- and speed-
dependent leakages, which flow from the high-pressure to the low-pressure
side within the motor (Index i) or which emerge from the motor externally
(Index e).
Owing to the compressibility of the pressure medium, every charged motor
is a spring-mass-system capable of oscillation, in which the oil column repre-
sents the spring. The corresponding natural angular frequency WH forms the
basis of a rough calculation of the motor's dynamic characteristics.
A further design criterion for rotary motors is the kinematic degree of
irregularity. The interaction of multiple displacement units such as pistons
or gears results in an irregular operating characteristic in terms of speed and
torque. As a rule, the larger the number of individual displacement units, the
lower the degree of irregularity.
Open Loop of a Hydraulic Drive
A hydraulic drive consists in a control element and a motor. The control

element can either take the form of a valve attached to a constant pres-
sure supply, or of a variable pump connected to the motor in a closed loop
(Fig. 4.1). The control element supplies a flow proportional to the input sig-
nal w, which causes pressure to be built up in the connector lines and drives
he motor. The load pressure PL then depends upon the load with which the
motor is impressed, the speed is determined by the flow of the control element
after all leakage flows have been taken into consideration.
If the dynamic response of the control element (valve or variable pump)
is at least thrice as fast as that of the drive (Ws > 3 WH), the control element
may be regarded as a proportional element in a linearized signal flow diagram
of the hydraulic drive (Fig. 4.29). The system's transmission characteristics
between the input signal w and the speed x (or '{J) may be regarded as
a system of second order in this simplified perspective. Attached downstream
is an integrator that yields the stroke x or the rotary angle zp.
Fig. 4.29. Hydraulic drive

The natural angular frequency WH and the damping rate DH of such

a drive do not solely depend on the motor, the oil volume and the coupled
mass, but also on the control element, the pressure-dependent internal leakage
flows (K1 Lip, K 2 Lip), or the flow pressure amplification VQP of the valve. The
relevant equations are listed in Table 4.5.
The static response characteristics obtaining between the input signal and
the speed are expressed in terms of the speed amplification C 0 .
The characteristics derived serve to assess the hydraulic drive's static and
dynamic response. This is particularly important if an open loop is connected
into a control loop by feeding back the stroke or the speed.
4.4.2 Pneumatic Motors
Rotary Motors
Compressed-air motors can be segmented into the types listed in Table 4.6.
The characteristic features of some compressed-air motors are depicted
qualitatively in Fig. 4.30.
One can see that vane motors are used for high speeds and low torques.
They are frequently employed to drive tools. Owing to their large displace-
ment volume and their medium speed range, gear motors have a high air
consumption, but are capable of delivering torques of up to 400 Nm. They
are primarily used in heavy industry and mining. Radial piston motors are
characterized by a low speed range with medium torques. Axial piston motors
are used for medium speeds and low torques.
Limited-angle rotary actuators constitute a particular type of rotatory ac-
tuator. These cylinders are mainly used for rotary movements covering angles
of up to 360°. They are particularly suitable as drives for valve spools, turning
equipment, shutter actuators, rudder actuation for ships, and so forth.
Figure 4.31 presents different types of limited-angle rotary actuators.
Thanslatory Motors
Compressed-air cylinders are termed pneumatic linear drives and may be

sorted into single-acting or double-acting cylinders. A further criterion for
categorization is the distinction between cylinders with rods and rodless ones.
In Table 4.7, the different types of linear drives, as well as the features of the
various cylinder types, are shown. Characteristics of pneumatic cylinders and
algorithms for the layout of cylinders and valves will be summarized in the
following chapter.
Static and Dynamic Performance
Just as the hydraulic drives shown in Fig. 4.29, pneumatic drives can be
connected to form an open loop. However, valves with four metering edges
Table 4.5. Characteristics of hydraulic drives
Translatory motor Rotary motor

Valve control Displacement control Valve control Displacement control
Natural angular 1+VH · VM·(K2Lix+A2) 1+ VH· VM ·(K2Licp+ V2/Z1r)
WH = WH =
TH-TM TwTM
frequency
1- · (_!__ + _.!._) DH- - 1- · (_!__ + _.!._)
TM
DH = -2·wH
Damping degree TH - 2·WH TH TM
Mechanic VM 6 VM = V2/27r
dx de;,
amplification
:Mechanical time "m2
TM = di: TM .!..2._
de;,
constant
Hydraulic 1 1 1
VH = Vqp+K2Lip VH = K1Lip+K2Lip VH = Vqp+K2Lip VH = K1Lip+K2Lip
amplification """
'Tj
Hydraulic time TH ------.fi! c TH = ----.!2H c 2"

Z·(Vqp+K2Lip) TH 2·(K1Lip+K2Lip) Z·(Vqp+K2Lip) TH = 2·(K1Lip+K2Lip) 0.:
constant '0
Vdead +A2 ·X2max CH Vdead + v2 /2 ~

Hydraulic capacity CH = EoiJ Eou "'
'~
()
Speed VH-VM·Vq; C _ VwVM·Vqi .,;
'"'"
Co= 1+VwVM·(K2Lix+A2) 0 - 1+VwVM·(K2Li,;+V2/27r)
amplification ~
0
'w"'
......
CfJ
-1
Table 4.6. Types of compressed-air motors
Type Features of motors

Piston motor piston with connecting rod, axial piston with swash plate,
radial piston with eccentric shaft
Membrane motor membrane with connecting rod, rolling membrane with ec-
centric shaft
Turning piston gear motor, helical gearmotor, screw motor, roots motor,
motor vane motor
t ~
I 1\
Input power
/, \
v:_
I"
\_
-·"]"' J- r-- -- -- --
t r----.. I I I
Output torque-
['.._
\
!----.,
-- -- -- -- -- -- -
'.
I
-"'"" X - X
t
v
1/-- I I I
Air consumption-
Gear motor
I
Radial piston motor
..-
-- -- -- - -
- X - X Axial piston motor
~~/ .A-~
X
~
Vane motor
Rotation speed n ~
Fig. 4.30. Qualitative representation of the characteristics for power, torque, and
air consumption of compressed-air motors
are almost exclusively used for that purpose. The equations for calculating
static and dynamic performance are summarized in Table 4.8.
The steady adjustment of the valve cross-section makes it possible to
continuously control the drive speed. Assuming a supply pressure of at least
6 bar, low load pressures at the outlet edges of the servo valve will induce an
approximately overcritical pressure ratio of flow. Thus, a further increase in
the supply pressure will have no effect on maximal speed. In this case, the
maximal speed solely depends on the drive geometry Az(Vdispl), the valve
Table 4. 7. Pneumatic translatory motors
Design Features
Single acting return movement by spring
cylinder installation length: 2 x stroke + 2 x guide length
II f!li\1\!111~
I
stroke: 20 ... 80 mm
Double acting movement in both directions by compressed air

cylinder installation length: 2 x stroke + 2 x guide length
Ill :
force dependent on direction due to different pis-
ton areas
I I
stroke: 10 ... 2000 mm
Cylinder with movement in both directions by compressed air

through running rod - installation length: 3x stroke+ 2x guide length
force independent on direction due to same piston
areas
stroke: 25 ... 300 mm
Membrane cylinder single acting

low friction
completely sealed
no lubrication necessary
installation length: 2x stroke + guide length
Rodless cylinder connection piston-slide

by rope
by band
direct
by magnetic force
installation length: lx stroke+ 2x guide length
force independent on direction due to same piston
area
Remarks:
rope cylinder: elasticity of rope problematic
band cylinder: sealing of band sensitive to wear
slitted cylinder: low stiffness of cylinder tube
magnetic cylinder: force-locking connection be-
tween piston and slide
,....
CD
0
Table 4.8. Characteristics of pneumatic drives

Cylinder drives Rotary piston motors ~
X ~
il'
::;
()q
Pv AvE AvA co
® T, 'J:. ~ ~
7;"
A, y
'J:.y p,J1P2,
~
y y Av=AvE=AvA "'
5
):L
Assumptions isothermal change of state, critical pressure ratio b=0.2 >

::;
):L
(!)
Constants ao, v = Wmax v2R273K, y = setvalue =valve stroke g;"'
Wmax = max value of flow function=0.484
~
ro
Maximum speed Xmax ~ ;:r;- nmax ~ aov rr;- s·
AzA aovy 29:iK vdispt. v 29:iK
dx ~<>DV rr;- dn ~~
Speed amplification Co= dy dy AZA V 29;iK Co,rot = dy dy vdispl. 29:iK
vrr;-
4Pc,absA~K Pc,abs Vlispl""
Natural angular frequency Wcyl = Wcyl,rot-
m(v~;l +Vdead,cyt) n 2 J(Vdispl +Vdead,rot)
Pc,al": absolute chamber pressure
V;;-I+Vdead,cyl ~ - 1 I KJPc,abs vJispl ( dn )

Damping degree Do= 4Pc.absA~Km dX Do,rot 4\ V,-l;.,....,J+V..:~.~.,...-~ ~~-~- dM jy=const.
a b c
Fig. 4 .31. Limited-angle rotary actuators. a Rotary piston, b translatory piston

with rack and pinion, c translatory piston with steep thread
aperture Av , the flow coefficient of the valve flow a o, the air temperature
T 1 upstream of the valve, and the critical flow velocity v1 at the outlet edge.
Figure 4.32 shows the stationary drive speed (rotatory speed) of cylinder
drive and rotary motor as a function of load pressure and reference variable
y. The performance curves depicted are based on an isothermic model [31] .
Comparing the illustration to the performance curves of the cylinder
drives and motors, one notices that even low load pressures lead to a drop in
the number of rotations per time. Provided that load pressures are low, drive
speeds in a pneumatic cylinder drive are much less sensitive to load pressures
than is the case in valve-controlled motor drives. In that working area, the
relevant damping ratio is much lower than for hydraulic drives.
The correlat ions reflected in the performance curves allow us to compare
t he stationary power output of pneumatic drives. Maximal power output is
roughly 25% of the maximal theoretical power for a rotary motor and 50 %
for a cylinder drive. The maximal theoretical power of pneumatic open loops
can be calculated as follows
(4. 12)
Just as in servo hydraulics, the dynamic performance of pneumatic open

loops can be described by means of a drive model of third order (PT2-I-
characteristic) . Again, characteristics occurring in that model are t he s pe ed
amplification (rotatory speed amplification) Co, t he natural frequency wo , and
the damping ratio D. The signal flow diagrams of pneumatic drives shown
in Fig. 4 .33 exhibit a structure comparable to t hat of the block diagram
for hydraulic drives shown in Fig. 4.29. The differences in structure between
1.4
_L
.L IYmaxl
~
1.0
7
·~IJ 0.5
"C
(])
(])
a. 0
/
/
/
~
~
(/)
"C
.l!l
ca
/
-0.5
(j)
~ L'
0::
-1.0
-1.4
-1.0 -0.5 0 0.5
t:.p
Related load pressure - -
Pv
a
1.0 r - - - - - - - r - - - - - , - - - - - - - r - - - - - ( j ) Mmax'Wmax
-1.0 '--------l...-----'----__...J'-------'
-1.0 -0.5 0 0.5

t:.p
Related load pressure - -
Pv
b
Fig. 4.32. Characteristic curves of pneumatic drives. a Cylinder, b rotary motor
rotary motor and cylinder drive result from the fact that the rotary motor is
more strongly damped through internal leakage, whereas the cylinder is more
strongly damped through speed-dependent friction.
4.5 Sensors
Properly using fluidpower drive systems requires suitable sensors. The most
important variables to be measures, either for monitoring or as control vari-
Av 2ao-vfVdispi"Pcabs"'
1t•J(\I.J;spi+Vdead) -
Pcabs •Vd1spl ·li dn I

21t·( \I.J;spl + '6aai dM v=const
2Dw
b
Fig. 4.33. Linearized models of the dynamic performance of pneumatic drives with
critical valve flows in the valve cross-sections. a Cylinder (middle position), b rotary
motor
ables, are pressure, flow, and stroke (or turning angle). Thanks to the evo-
lution of precision manufacturing technology and to the continuing increase
in the degree of integration of electronic components, highly accurate sensors
are available nowadays. Operating principles and features of the different
sensor types will be discussed in the following.
4.5.1 Pressure Sensors
Measuring techniques for registering pressure are divided into mechanical and
electrical procedures. In fluidpower drive systems, mainly dynamic pressure
fluctuations occur, and hence electrical pressure transducers are preferred.
In these sensors, the elastic deformation of an object is captured and trans-
formed into an electrical signal. This method exploits the material's physical
properties. In Table 4.9, the most important types of pressure transformers
are contrasted.
In measuring changes in pressure, often pressure transducers based on
strain gauges are employed. There is a multitude of different types, including
Table 4.9. Pressure sensors
·-
Function Mechanical Strain gear Inductive Piezoelectric Piezoresistive
principle
Schematic
-ED-
~ -e~
.
M easuring 0 . . . 4000 0 ... 3000 0 ... 1000 0. .. 7000 0 .. . 1000

range bar
Frequency 0 ... 5 0 ... 100000 0 ... 10000 10. .. 150000 0 ... 150000
range H z
Error% 1 0.25 0.5 1 0.25
Price range € 50 ... 175 150 ... 600 150 ... 400 350 ... 1000 25 ... 850
Remarks electr . output also with relativ. slow only for sensible to
possible amplifier dynamic overload
measurements
reasonably-priced ones with an integrated amplifier, for which even a volt-

meter suffices as a visual display unit. Piezoelectric transducers are suitable
for dynamic measurements, and they are available in very small sizes.
Piezoresistive sensors are the best solution for simultaneously analyzing
static and dynamic pressure signals. Vapor-deposited thick-film resistors on
silicon wafers constitute the measuring element. Highly-integrated sensors
contain the measuring element and the amplifier electronics in an integrated
component. These types are affordable, albeit somewhat sensitive to temper-
ature changes.
One crucial criterion in selecting a pressure transducer for fluidpower drive
systems is its frequency range. In Fig. 4.34, measured amplitude responses of
various pressure transducers are compared.
_;- -10
~
.$!
~ -20 1 : piezoresistive
"5 2: piezoelectric
0 db 3: strain gage with cycle
frequency 50 kHz
-30 4: strain gage with cycle
frequency 5 kHz
5: strain gage with integrated amplifier
100 1k 10k 20k 40kHz

Frequency f
Fig. 4.34. Amplitude responses of different pressure sensors

4.5.2 Flow Sensors
Even today, the exact measurement of flows is fraught with considerable

difficulties. Disturbance variables such as pressure pulsation, temperature,
viscosity of the pressure medium and the degree of contamination all influence
the measurement's result.
One distinguishes between two tasks , measuring volume and measuring
flow. Volume is measured for dosing procedures, for the open-loop synchro-
nization control of cylinders, and also for indirect stroke measurement.
Three approaches are used for measuring volume. In the displacement
principle, the total volume is split into multiple parts of defined size, which
are subsequently added up. In the flow principle, the effects of the flow are
captured via physical properties and then transformed into a signal propor-
tional to the volume flow. Simply measuring the velocity of the streaming
medium yields the volume flow at the point of measurement, provided that
the flow cross-section at that point is known. Table 4.10 presents a survey of
some sensor types.
Table 4.10. Volume flow sensors

Function G ear pump Turbine Floating Induction Ultra sound
principle flow meter body
Schematic A: transmitter-
~
Measuring 0 ... 1000 0.5 ... 20000 1. .. 600 0.5 ... 1200
range (1/min)
Dynamics low high high low high
Error(%) 0 .5 1.5 2 0 .5 2
Price range ... 3000 250 ... 1000 1000 ... 2000 1000 ... 2500 1500 ... 2500
(€)
R e marks sensitive to not accurate good linearity conductive without
contamina- at Q c:= 0 media feedbac k
tion effect
Gear-pump flow meters are equally suitable for measuring volume or flow .
They are highly accurate, but have the drawbacks of slightly higher pressure
losses and of being prone to seizing due to contamination. If low demands are
made on dynamics, the measuring turbine , easy to integrate into extant sys-
tems, is used. The medium flowing through the turbine drives a turbine wheel
that induces an electrical impulse with each blade. Counting the frequency
yields an output signal that is proportional to the volume flow .
High demands on dynamic performance with relatively lower accuracy are
fulfilled by floating-body flow meters. The flowing medium pushes a suitably-
shaped object against the resistance of a spring. The object's stroke is mea-
sured , and, if the object's geometry is suitably chosen, will allow for the
volume flow to be immediately inferred.
New measuring techniques low in feedback effects work with ultrasound
or magnetic induction. The ultrasound technique works in a similar way as
the Laser-Doppler principle, but is much less expensive. Ultrasound signals
are emitted into the medium and partially reflected by barrier layers caused
by particles or air bubbles. The phase change occurring according to the
Doppler effect is then evaluated.
If media contain water or are mixed with additives , the induction method
may be used. The flowing , electrically conducive medium is permeated with
a magnetic field , which induces a voltage dependent on the flow velocity. This
measuring technique is suitable for broad measuring ranges and has almost
no feedback effects on the flowing medium.
4.5.3 Position and Rotary Sensors
Position and rotary sensors have acquired a position of particular importance

throughout the entire realm of actuator technology. The exacting demands
on position controls and continuous path controls require the actual position
to be determined accurately. Many actuators come equipped with internal
sensors for measuring stroke and turning angle. The same principles apply to
measuring stroke and turning angle, and hence the following remarks shall
be restricted to stroke sensors. Table 4.11 presents an overview of common
metering principles and their features .
Potentiometric position measurement ranks amongst analogue measuring
techniques with resolutions that are infinite in theory. A sliding contact is
moved on a resistance coat and emits a position-dependent voltage. Sensors
based on this principle are easy to build and therefore cheap. Noise places
Table 4.11. Position sensors

Function Pote ntio- Diffe re ntial Inductive U ltraso und Optical
princ iple m e ter t r a nsforma tor
II
Sch em a tic
~~
M ea s uring
ra nge (mm)
... 1000 500 1000 1000
~ 1000
Recording 10 2 50 4
s pee d (m / s)
Error (%) 0.05 0.5 0.001 0 .05 0.001
Price r a nge 150 ... 350 500 750 600 250 . .. 600
(€)
R e m a rks sensitive t o s en sitive to eva lua tio n d e lay it me up re la ti ve o r
wear t e mpe r a ture costly to 1 m s a bsolute
encode r
a limit on the achievable resolution, and the resistance coat is sensitive to

wear and contamination.
Position sensors with differential coil are popular. An armature moves
through two coils arranged in line, and, with each change in position, shifts
these inductive elements of a bridge circuit. The differential circuit helps to
obtain a broad linear range of measurement, but accuracy is limited. Pressure-
resistant models are available.
Another way of constructing inductive position sensors involves a scan-
ning plate and a scale that is moved relatively to the plate. Both plates
feature meandering circuit conductors, charged with an oscillating signal on
the scanning plate. The signal induced on the scale is compared with the
input signal, and the phase shift provides clues concerning the position.
In ultrasound sensors, an ultrasound signal is sent through a wire. A mag-
netic ring around the wire indicates the position and causes reflections of the
signal. Position is determined by analyzing the running time. One disadvan-
tage of this system is the delay time inherent in the operating principle.
Incremental position sensors employ graded glass rules with embedded
grading lines. A reading head counts the impulses of the glass rule moving
past it. The method is very accurate and has a high resolution. The potential
danger of a loss of data caused by counting errors can be avoided through the
use of absolute encoders. In that case, a distance or angle is not divided into
equal gradations, but each position explicitly stated in form of a code. The
complex reading heads required for this type can exactly determine every
position.
4.6 Controller Types
Recent years have seen an increase in our knowledge of controlling fluidpower

that has led to a significant increase in the performance of such drive sys-
tems. These performance gains have been achieved through the realization
of controller structures more complex than the straightforward proportional
feedback used in the past. Owing to the characteristic properties of fluid-
power actuators, standard controller assemblies with PID-characteristics are
of little use. Freely-programmable digital controller assemblies open up new
avenues for fluidpower technology.
As fluidpower drives are so versatile in terms of their design (translatory
or rotatory motor), their control (primary or secondary control), and not
least the variable to be controlled (position, speed, pressure), no universally
applicable rules concerning controller structure and parameter setting can be
formulated. Rather, the available variety on the drive side requires the project
planner to carefully consider the task in hand. The following considerations
are intended to facilitate some fundamental design choices and parameter
settings, and to give some hints concerning aids to design [1, 15, 16].
4.6.1 Control of Hydraulic Servo Systems
Owing to the multitude of hydraulic open-loop controls, not all variants can
be discussed. However, models for all drives can be described as follows in
a way that may be sufficiently accurate for many tasks (Figs. 4.35 and 4.36).
One or two subsystems prone to oscillation, where many controller layouts

only take into consideration one such system.
.'f.oad
Valve characteristics: Cylinder characteristics: Load characteristics:

Vav : volume flow amplification A K : piston area m :load mass
Vap : volume flow-pressure Kup : leakage, FLoad : load force
amplification pressure dependent
Tv : time constant Kuv: leakage,
speed dependent
VA : volume of cylinder
chamber
dN :cylinder friction
Fig. 4.35. Signal flow diagram of a linear drive
Fig. 4.36. Signal flow diagram of the linear drive with related variables
An integrator that is freely connected according to the variable to be

controlled, and which may be positioned downstream of the subsystem
prone to oscillation.
In considering control technology, characteristics may be assigned that

describe the drive in question with sufficient accuracy for most purposes.
The natural angular frequency w 0 or natural frequency fo, respectively

(wo = 27rfo): the term natural angular frequency means the natural oscil-
lation of the significant subsystem prone to oscillation. Natural frequencies
of up to approximately 50 Hz are common.
The degree of damping D: it describes the damping characteristics of the
subsystem prone to oscillation for which wo was defined. Normally degrees
of damping range from 0.05 to 0.3, such that one is always dealing with
weakly-damped systems prone to oscillation.
Open loop amplification 0 0 : the drive's static open-loop amplification
upstream of an integrator is also termed the speed amplification, because
it reflects the cylinder's speed for a constant signal on the valve.
Valve time constant Tv: the valve can often be regarded as a PT 1 -element,
where Tv is the valve's time constant.
A further characteristic is the loop amplification VK. It reflects the static

gain of the open-loop control. For this purpose, the control loop needs to
be cut off immediately below the summation point, for otherwise any minor
loops used would not be properly taken into consideration.
4.6.2 Single-loop Control
In the following, the position control of an hydraulic cylinder drive with

valve control shall be treated of as an example. Single-loop controllers solely
rely upon the controlled variable as an input signal. The controller with
a strictly proportional transfer behavior (P -controller) is rarely used in hy-
draulics, because it further reduces the already poor damping. The PPT 1 -
controller has emerged as a good alternative. In that controller, a relatively
weak P-proportion is switched without delay, in order to maintain a high
response speed. The greater part is switched delayed via a PT 1 -element, in
order to counteract the control system's propensity to oscillate. As can be
seen from Fig. 4.37, the PPT 1-element may be interpreted as a P-element
with an additional positive speed feedback [1].
In hydraulic drives, this positive speed feedback leads to a slight decrease
in the natural frequency and a simultaneous increase in the degree of damp-
ing. If one regards the DT 1 -element as an ideal D-element (very low time
constant), the following formulae hold for the system's resulting damping
and natural angular frequency:
PPT-controller
Fig. 4.37. Interpretatio n of the PPT1-contro ller
Wo = woJ1- CoKl (4.13)

D
D*= -vfr-1=-===;;;c;::::o (4.14)
K:o=o=1
In order to avoid any lasting deviations that might be induced by friction

effects, control technology commonly incorporates an I-block into the con-
troller. This measure has no beneficial effect on hydraulic positioning drives,
- because the integrator positioned within the control system results in

a critical phase shift, which is only weakly compensate d for by the con-
troller's P-proportion , such that the I-block can only be weighted ex-
tremely weakly,
- because perpetual working cycles occur with low residual deviations from
the command variable,
- because the system's propensity to overshoot is aggravated.
4.6.3 Multi-loop Controls
Apart from the actual value to be matched to the command value, these
controls rely on further state variables of the open loop. In a single-loop
control, all state variables depend on one another, but only one state vari-
able constitutes the output variable (stroke, speed, pressure). From a control
perspective, each order of the system can be assigned a state variable. If
all significant state variables and the relationship s between them are known,
the system can be forced into every possible movement in theory. However,
Kx
Fig. 4.38. Three-loop state controller
the foregoing statement is not unconditionally true in practice, because of

the non-linearities existing in hydraulic open-loop systems and, mainly, be-
cause of the existing limitations to signalling and power (e.g. the valve spool
stroke). Rather, in practice, one attempts to imbue the open loop with a de-
sirable level of damping, in order to be able to commensurately increase the
control's P-proportion. The natural angular frequency can only be increased
to a trifling extent, because that would require a high and rapidly changing
energy input via the control element.
In many cases, multi-loop controllers have three loops. In those types, the
stroke, the speed, and the acceleration are fed back (Fig. 4.38).
Ak+Kiiv
.-----------~t=~----------~
DT1-element to suppress
the static pressure portion
VP: volume under pressure
Fig. 4.39. Load pressure feedback in the signal flow diagram

By means of these three state feedbacks, the natural angular frequency

and the degree of damping can be influenced.
The positive feedback of the acceleration (negative acceleration feedback)
has a damping effect. The negative feedback of the velocity reduces damping,
but increases the natural frequency. By means of a positive velocity feedback,
the damping can be further increased, as has already been shown above for
the PPT 1 -controller.
If the load pressure is to be the feedback signal instead of the acceleration
(Fig. 4.39), it must be connected through a DT 1 -element, because otherwise
a disturbing force would result in a lasting control deviation.
4.6.4 Formulation of Control Design
In large measure, the hydraulic actuator can be adequately described through

linear models. Accordingly, general considerations concerning the selection of
control algorithms and the corresponding parameters can proceed on the
basis of linear control theory. Non-linear simulations can improve the setting
of control parameters and allow for non-linear additional means to be tested
with the control [39].
Linear Method
On a fundamental level, linear descriptions of a fiuidpower drive in control

theory can either take the form of a frequency response or transfer function
(G(s), G(w) or G(z)) or proceed in state space. The state space description
has the advantage of comprising all internal process variables, whereas only
the input and output variables can be used with the transfer function. As
the aforementioned state controller has recourse to multiple state variables
(stroke, velocity, acceleration), state space description suggests itself.
The control theory literature suggests different methods of designing state
controllers. Pole assessment has emerged as a workable method in practice
[37]. In that method, the poles of the transfer function G(s) are assigned
corresponding desired poles. On the basis of these different poles the feedback
vector can be calculated.
The positions of the poles on the open loop are derived from the integrator
pole at the origin, from the valve pole on the negative-real axis, and from
the PT2-element's conjugated-complex pair of poles. As the PT 2-element is
particularly important, three varieties shall be shown here (Fig. 4.40):
- The aperodical boundary (D=l). The two poles coincide, such that this
element corresponds to a serial connection of two PTl-elements with the
same time constant.
- The fast step response, almost without overshoot (D=0.7). The poles
form a 45° angle with the imaginary axis; however, the step response
only exhibits minor overshoot.
w,x
lm
/ Re
0=1
Fig. 4.40. Influence of the degree of damping at a PT2-element
The undamped oscillation (D=O). The poles are situated on the imaginary
axis, the oscillation does not abate.
Using the pole assessment technique entails two inevitable problems, for
which there is no generally applicable solution, and which must always be
based on a sound understanding of the system in question.
In practice, the state controllers used do not feed back all state variables
existing in the open loop. The valve dynamics are neglected.
An ideal linear open loop can be imbued with every possible dynamic
performance by a state controller, but this is not feasible in practice (non-
linearities, limitations).
The following approach has proved useful for designing a state controller
(Fig. 4.41):
Considering the valve dynamics. The state controller is only used if the
valve's dynamic performance is much better than that of the open loop.
The valve will often be equipped with an integrated control. One valve
pole is left unchanged and used as a wanted pole.
Assessing the poles of the PT 2 -element prone to oscillation. In the pole-
zero-point plan, the PT 2 -element prone to oscillation with its character-
istics C 0 , w 0 , and D, manifests itself as a complex pair of poles, allowing
the variables w 0 and D to be read off. As the controlled drive with its
natural frequency cannot significantly exceed the uncontrolled loop's fre-
quency, the natural angular frequency of the uncontrolled drive is set as
the desired frequency. That way, the distance between the point of origin
and the desired pair of poles remains unchanged. As the step response of
a controlled drive is supposed to be fast but with little overshoot, a de-
sired damping of D=0.7 results. Thus, an angle of 45° between the desired
pair of poles is also determined.
Assessing the PT 1 - pole from the integrator of the uncontrolled open loop.
Potential positions for that pole are limited by the open loop's non-
linearities including the control element. In order to determine this pole,
~
e
Pole situation Step responses
t
lm
Open loop
Re l2:~
P-controller
+- F
XF
Re
~
XF
Complete
state controller e
Real
~
XF
three loop controller e
•
Optimized
controller
~ •
e
Fig. 4.41. Steps in pole assessment
first a robust P-controller is designed experimentally which shifts that

integrator pole to the left along the real axis. The distance of the result-
ing pole is doubled, and thus yields the final desired pole for the pole
assessment.
- Employing the feedback vector. The feedback vector can be calculated
from the position of these poles. The calculated vector consists in four
parameters feeding back the individual state variables. Since - as has al-
ready been mentioned- an improvement by the valve cannot be expected,
its parameter is not used, but only the parameters for the weighting of
speed, velocity and acceleration.
- Optimizing the controller through experimentation. As a rule, a controller
chosen in this manner will already exhibit a very good control performance
in response to a step in the command variable. In many cases, that per-

formance can be further improved through experimentation, because the
effects of non-linearities cannot be fully taken into consideration.
In Fig. 4.42, step responses of various controllers are shown.
u Three loop controller PPT1-controller
u.y
w0 =125 1/s
D =0.1
Tv =0.05 s
P-controller: VK = 0.1 1/s

PPTrcontroller: VK=0.15 1/s
Three loop controller: VK = 0.3 1/s
t
0.2 s 0.4 s
Fig. 4.42. Step responses of various controllers
Non-linear Additional Functions
The problems associated with an integrating part within the controller have
already been mentioned. In order to compensate for errors resulting from
a valve shifted out of position or from an external load, a switching integrator
is used. As Fig. 4.43 shows, it only intervenes as long as the actual value
remains within a range of tolerance.
The outer dead zone prevents an otherwise inevitable overshoot in reac-
tion to a command variable step, and the inner dead zone helps to avoid
continuous working cycles of the open loop caused by stick-slip friction.
Another important non-linear additional function is the liberalization of
characteristic curves. In order to achieve linear performance with elements
having non-linear characteristic curves or performance maps, these perfor-
mance maps are stored and the variables to be linearized used as indices for
the maps. This imbues those elements with those linear characteristics that
the design requires. However, such corrections can only be used if the perfor-
mance maps are time-invariant. A typical example would be the correction
of the performance map of a valve.
Controller
---
/
+
Signal is switched on, if
---- JL_ -
Integration to
-the compensation area compensate
is reached (external border) an offset
-the minimal deviation was not
reached (internal border)
Fig. 4.43. Switching integrator
4.6.5 Control of Pneumatic Servo Systems
Position Control
Owing to their particular performance characteristics and their good power-

to-weight ratio, position-controlled servopneumatic drives are particularly
suitable for use with handling systems. For many applications, running them
with asynchronous point-to-point (PTP) controls characterized by the step-
wise input of command variables will be sufficient. Continuous path controls
(CPC) are also used to satisfy more demanding requirements. The continu-
ous input of command variables involved in CPC-mode makes it possible to
synchronize multiple drive axes [29, 30, 31].
Asynchronous Positioning (PTP mode). Traditional control methods
cannot exploit the dynamic performance of pneumatic drives within the po-
sition control, because the drives' relatively low degree of damping means
that the closed loop control will be excited into oscillations even with low
loop amplifications. This is particularly true if working with stepwise com-
mand variable input, because the high accelerations excite the drives into
oscillation in a particularly forceful manner. The remedy is the three-loop
control concept already described for hydraulic drives (Sect. 4.6.3), which is
almost exclusively used, due to its positive effect on the natural angular fre-
quency and the degree of damping of pneumatic drives. Bases for the use of
that control structure are the dynamic models of pneumatic drives listed in
Sect. 4.4.2. Like hydraulic systems, they can be represented as delay elements
of second order with a downstream integrator (PT 2 -I-characteristic) if valve
dynamics and other non-linearities are neglected.
Over and above the advantages already described in the foregoing, the
three-loop control system's negative speed feedback reduces the dependency
on the working point of the speed amplification, which is pronounced in ser-

vopneumatic drives due to saturation effects of the flow function. The speed
amplification C 0 of the multi-loop controlled drive is calculated according to
the following formula:
Co
co=--~=- (4.15)
1 + CoKv ·
The comparison of the speed amplification of the open-loop system with
that of the controlled system, shown in Fig. 4.44a, underlines the reduction
in the dependency on the working point. As shown in Fig. 4.44b, the direct
dependency of the loop amplification on the speed amplification leads to
a marked liberalization of the loop amplification.
Linearizing the drive characteristics leads to an improved control perfor-
mance, such that a non-linear correction of the speed-signal-function can be
dispensed with in many cases [33, 34].
Path Control. Electrical robot- or CNC-controls are a sensible means of
controlling servopneumatic drives' paths. In order to simplify integrating the
drive control into a customary robot- or CNC-control, the three-loop PTP
control structure depicted in Fig. 4.45a is modified in a suitable manner. By
transforming the PTP-control's command variable equation
Open loop Speed controller

20+------.
__________c·o
a 5 v 10 y
P-controller P- and speed controller VK P- and speed controller

20 )-.. , ,
'•
1/s
10
P-controller
Kx 1 =1 V/cm
Kx2= 6 V/cm
00 5 v 10 y
b
Fig. 4.44. Linearizing the speed- and loop amplification through negative speed
feedback. a Speed amplification C 0 , bloop amplification Vk
y = Kx(w- x)- Kxx- Kx.x (4.16)
into the command variable equation of the CNC-control
y = Kx[Kv(w- x)- x]- Kx.x, K _ Kx (4.17)

v - Kx
one obtains the cascaded structure shown in Fig. 4.45b, which is also common
for electric drives.
Unlike the speed control of electric drives, pneumatic drives often feature
digital technology in the subordinate speed control, because that allows non-
liner control parts to be taken into consideration with ease. The result is an
overall control structure of the signal processing as depicted in Fig. 4.45c.
As the overall structure shown indicates, the product of the position de-
viation and the weighting factor Kv is utilized in calculating the speed com-
mand signal. For constant speeds, a velocity error occurs which is inversely
proportional to the loop amplification and proportional to the speed. In nu-
merical control technology, the factor Kv is often used to express the loop
amplification. However, the loop amplification of the cascaded CPC-control
of fluidpower drives is calculated according to
KxCo
Vk = Kv1 + KxCo (4.18)
As (4.18) shows, in the case of the control loops under consideration, it

would be inadmissible to equate the loop amplification Vk with the factor Kv.
The cascaded control has the advantage of allowing the speed- and
position-controllers to be set independently. As the external position con-
trol loop exhibits relatively low dynamic performance, longer sampling in-
tervals suffice for the generation of command values and for the position
control. Hence, only the speed controller requires a high sampling frequency
(Ts < 4ms).
Non-linear Means of Compensating Reversal Error. Within the range
of small command variables, stick friction and leakage lead to a reversal error
in the speed amplification C0 . If controller setting remain unchanged, the
control accuracy of position-controlled drives decreases in the area of the
reversal error. This property of servo drives can be compensated for with
non-linear measures. Examples given are the non-linear parameter setting
for a speed controller shown in Fig. 4.46 and the unsteady compensation of
the threshold shown in Fig. 4.47. These two measures are suitable both for
positioning and for path control.
If exacting demands are made on the positioning- and path accuracy of
servopneumatic drives, linearizing the loop amplification by means of a pro-
portional speed feedback will not suffice. Thus, the speed controller, which
in the cascaded position control can be adjusted without reference to the su-
perordinate position control loop is optimized with a view to compensating
:p
·-------------------------·'
b P: position controller, S: speed controller
Digital
Generating of path
signal
generation
Feedback
w
Digital P-controller
control X
-·················· _':-y_(':"_~~l.................. ,
Digital
x,x-controller
control
c
Fig. 4.45. Signal processing concept for the path control of servopneumatic drives.
a State controller, b divided in speed and position controller, c concept for the path
control
the speed amplification's non-linearity in the area of the reversal error. As

shown in Fig. 4.46, increasing the speed feedback at low speeds may help to
achieve a slow-running performance that is low in vibration. If an optimal
controller setting is determined for each command speed, the drive is able to
track a variable nominal speed profile.
0.6
r·
~Kx
Vs/cm Vs2/cm :x
~
K··
I
2
-"
•X
X
)(:
0.002
\
-"
(.)
0.4
"0
ro Q)
.c .2
"0
Q) c
.2
0.2 0.001 2
"0
Q) -......-....._ ~
co
Q)
~
c.
(/) ~
0
::t.
0
0 10 cm/s 100
b Speed x
1.00
Vs/cm
•X
0.75 ~
-"
·S: (.)
co
·><' .c
"0
"0 Q)
Q) 0.50 .2
Q)
c. "0
(/) Q)
Q)
c.
0.25 (/)
Timet
c 1: command of speed W, 2: actual speed x, 3: speed feedback Kx

Fig. 4.46. Non-linear speed control. a Controller structure, b control parameters,
c course of parameter adjustment over time
The unsteady compensation of the starting threshold depicted in Fig. 4.4 7

permits the positioning accuracy to be improved up to the limit of the mea-
suring system's resolution. The starting threshold compensation increases the
command variable, depending on the sign of the position deviation, by a set
increment, such that a dead zone in the speed-signal function can be com-
E
8
><
>< -19 ~-,---r:-;::::==+===:::::j-19 ~
Q)
:0 em em ro
-,
-~ -24.5 1+--____,----1+----+----'1 -24.5 ·~
ro
> '0
E
c c
0 -30 H--____,~---l+----+----'1:-3o
"""iii0 E
. 8
-ro -35.5 1+---+--it---+-------1:1-35.5 2
c.
:::>
t5
<( -41 ~==:±====i __L _ _jJ -41 ·~ Control structure:
I 0 1.75 3.5 5.25 s 7 [}_

Timet
1000 100
•>< cm/s 2
A cm/s
50
·><
\
§ 500 '0
)
Q)
Q)
~Q) ,........,., c.
en
0 0
] \\/ I
I
/ -ro:::>
t5
<(
-500
v ', - /
I
-50
-1000 -100
0 0.225 0.45 0.675 s 0.9
Timet
0.1
><
<1 mm
§ 0.05
~
·~
'0
c
0
' ~
,g -0.05
'iii
0
[}_ -0.1
0 1.75 3.5 5.25 s 7
Timet
Fig. 4.47. Unsteady compensation of starting threshold
pensated. An overcompensation of the starting threshold leads to persistent

working cycles.
Pressure Control
Pressure-controlled pneumatic components can be used in any application

where defined forces have to be exerted, for example when clamping or grip-
ping work pieces in handling technology. Designing a pressure control for
pneumatic components is much easier than designing a position control sys-
tem, because the former is not affected by friction as a disturbance variable.
In general, a pressure control is only used with low displacement speeds. The
models described in Sect. 4.6.5 are inapplicable in that case, because there
is no critical flow through the control edges. As long as inductive influences
are avoided, the rise in pressure during pressure control is almost entirely
dependent upon the valve opening:
T - ~ Vip = Vqy . (4.19)

P- 2Vqp' Vqp
Figure 4.48 shows the signal flow diagram and the step responses for the
pressure control. Whereas the dynamic performance of the pressure build-up
is correctly captured in principle, the high-frequency oscillation superimposed
on the measured signal does not appear in the linear model. On the one hand,
the frequency spectrum of the actual system's oscillation contains a portion
caused by the open loop's inductivity; upon the other hand, there also occur
pressure oscillations caused by pressure waves being reflected by discontinu-
ous changes in cross-sections. This portion of the pressure oscillation is both
time- and position-dependent, and hence cannot be captured in a simple
linear model derivation [44].
As the valve cannot compensate the aforementioned pressure oscillations
due to their high frequencies, the only solution is to use control concepts in-
corporating a delay, which can filter out these oscillations or avoid inducing.
In practice, the PPT 1-controller has proved itself, because one can balance the
requirements concerning accuracy and dynamic performance by fine- tuning
the P- and PT 1-portions to find a good compromise. For low TpTl, the PPT 1 -
controller acts like a pure P-controller; for high TPTl it exhibits PI-behavior.
Thus, the PPT 1-controller allows to achieve a compromise between the di-
rectly meshing P-controller with relatively low accuracy and the PI-controller,
which is frequently unsatisfactory in terms of its dynamic performance and
theoretically admits of no control deviation.
For a PPT 1-controller free of overshoot, a constant P-portion yields the
correlation between KPTl and TPTl shown in Fig. 4.49. With increasing KPTl
and TPTl, both the accuracy of the control and the duration of the adjustment
process increase. Thus, absolute values for both parameters can determine the
compromise between accuracy and control dynamics almost at will.
4.6.6 Realization of Controls with Microcomputers
Many of the tasks to be accomplished by a control algorithm can only be

inadequately fulfilled by analogue electronic circuits. Even the correction of
characteristics presents difficulties. Owing to the complexity of modern con-
trol tasks and the need for controllers to be flexibly adaptable to changing
conditions, the use of digital controller cards represents the contemporary
state of the art. Using controller cards based on microcomputers expands
the range of possibilities for controlling fluidpower drives, but some specific
properties have to be borne in mind.
bar -Simulation
o;
Kpt 0.3V/b~
.,
::J
!'-(\)
ti
o.-' .......
-E
8 3
o.-' 10,Ait.------
~
~
~~
I
i/
~ I
rn
rn I
~ Pactual
a.
m ~
:;:::>
c: 0
i --Simulation
~~~
~
~ - I
KPt 0.14 V/bar
i:5 I
I -Measurement
~----- KPt 0.14 V/bar
-3
0 50 100 ms 150
b Timet
Fig. 4.48. Pressure control loop. a Linear model for the rise in pressure, b com-
parison between model and measured data
The following points should be considered in the layout of a digital con-

troller.
The control process is not continuous, but proceeds in sampling steps.
All calculations involving time-invariant values are processed outside the
controller. The effects of the sampling time on control parameters can be
taken into consideration during the layout. Corrections of characteristics
may be stored as performance maps, correction equations are unnecessary.
Parameter adaptations can also be uncoupled from the pulse cycle, unless
the parameters to be adapted necessarily have to be calculated for each
pulse cycle.
Kpl=0.21; Kpr 1 =0.85; Tpn =200 ms

El E
0 0
r:f.c{ 0.5
o.IJ 0
KpL=0.21; Kpn=0.155; JPr1 =10 ms
~
::J
U)
U)
~
c..
0
50 100
0 50 100 ms
a Timet
10r-----~----,------r----~------r-----,-----~
~
1---
v
~bar
c:
0
~
!E
a.
E
ctl
~
gc: KPL =0.21 V/bar = const.
0
(.)
20 50 100 200 ms 500 1000 2000
b Time constant Tpr1
Fig. 4.49. Layout of the PT1-portion of the pressure control. a Step responses for
various controller settings, b controller parameters
4. 7 The Fluid power Drive as a Subsystem

Fluidpower drives are frequently employed as a drive unit within a plant.
Thus, a connection needs to be established between the fluidpower drive's
control and the superordinate process control, the overall control unit. There
are two main types of process control, computer-numeric control (CNC) and
memory-programmable control (MPC). In order to integrate the drive into
such a system, suitable hard- and software for the implementation is required.
Connecting a fluidpower drive to a CNC control proceeds by means of an
external controller unit, which receives as input variables the command vari-
ables of the CNC control and the position of the drive. Figure 4.50 illustrates
this.
CNC unit
Actual
value
Fig. 4.50. Connecting a hydraulic drive to a CNC and a MPC (Memory Pro-
grammable Control)
Memory-programmable controls (MPC) are characterized by their mod-

ular design. The control's central processing unit (CPU) and the controller
components communicate via a shared bus. Thus, a controller unit adapted
to the requirements of fluidpower technology can be used, which receives its
command variables from the CPU and, if necessary, may pass on its own
messages (error, position reached, etc.) to the CPU.
4.8 Comparison Between Fluidpower

and Electric Drives
The differences between fluid power and electromagnetic actuators are primar-
ily due to the force- or torque generation in the energy converters. In electric
motors, a torque is usually generated by utilizing the fact that a magnetic
field will exert a force perpendicular to any conductor carrying an electric
current. This method of force generation predestines the electromagnetic ac-
tuator to become a rotary machine.
In fluidpower actuators, a pressure exerts a perpendicular force on a dis-
placement surface, which makes it easy to construct linear drives.
Comparison of Characteristics
The ratio of weight or space to the power generated depends on the force
density within the energy transformer. Owing to the magnetic saturation of
the armature and stator materials, the force density in electric machines has
a theoretical limit of 15 or 16 bar (without air gap). It is much lower in real
machines, lying in the range of up to 4 bar.
om
DC-motor AC-motor Hydraulic motor
<ll
N
Ci5
.E
Cl
'(i)
5:
05
91
04
~~ • 01
<ll
u
Ill
0..
(/)
06 06 01
Mean values of different manufacturers at nominal conditions

(electric motors without ventilation)
Fig. 4.51. Weight and space of different servo motors
In fluid power machinery, the force density is limited by the strength values
of the mechanical parts (e.g. surface pressure, resistance to abrasion and
erosion). The force density of today's hydraulic converters is in the range of
320 t o 500 bar. The maximal rotatory speeds of hydraulic motors are lower
than those of electric motors, due to pressure commutation problems.
In a comparison of servo motors for machine tool feed drives, t he hyd ro-
motor with its superior force density offers advantages in terms of weight and
space, as shown in Fig. 4.51. For greater power outputs, the advantages of h y-
draulic displacement motors become more marked. The high energy density
of hydraulic actuat ors (up to 0.2kg/kW for axial piston machines) is only
possible because t he liquid pressure medium carries away the heat caused b y
power losses into t he tank. By contrast, t he power loss in electric machines
remains in the coils, a nd the only way to accelerate its dissipation is to use
an external cooling system.
Advantages and Disadvantages of Fluidpowe r Actuators
The key advantages of fluidpower actuators are summarized below.

The good open- and closed-loop control of the working variables pressure
and flow i s a precondition for many ways of solving control a nd drive tasks
where speed and accuracy matter.
The possibility of conducting fluidpower energy to stationary or mobile
consumers via hoses or pipes, constitutes an advantage of fluidpower tech-
nology over mechanical means of energy transmission, because the latter are
always bound by geometrically g iven energy paths (e.g. toothed belt drives).
Particularly in mobile applications such as vehicle or aircraft engineering,

the low energy weight and the low space requirements of hydraulic actuators
are extremely important.
The easy transformation of fluidpower into the mechanical power of
a translatory movement by means of piston and cylinder secures many appli-
cations in which large forces must be exerted for that type of actuator (e.g.
hydraulic presses, working hydraulics of mobile equipment).
In designing fluidpower actuators, particular attention must be paid to
some important problems.
The efficiencies of hydraulic displacement units are in the same range
as those of electromagnetic transformers. Energy-saving circuits (e.g. dis-
placement control) must be used whenever energy efficiency is especially im-
portant. In using fluidpower actuators, care must be taken to maintain the
pressure fluid (filtration, cooling, draining of ingressed water) and to avoid
leakages.
With ill-chosen designs, the great force density may lead to noise prob-
lems, unless primary and secondary noise reduction measures are taken.
4.9 Application Examples
4.9.1 Automatic Anti-lock Braking System
In the following, the functioning of a standard anti-lock braking system (ABS)

will be described. The purpose of an anti-lock braking system is to prevent all
the wheels from simultaneously locking during braking. This often reduces
braking distances and primarily serves to ensure that the vehicles remains
steerable during braking. The ABS is integrated into the connection between
the tandem master brake cylinder and the wheel brakes as shown in Fig. 4.52.
In the inactive state, each chamber of the master cylinder is connected to
two wheels by means of a currentless open switching valve (inlet valve). If
the ABS electronics sense that a wheel is about to lock, that inlet valve is
shut. If the inclination to lock persists, the currentless closed outlet valve is
opened. Thus, brake fluid from the wheel brake can flow into the low-pressure
accumulator and the braking pressure at the wheel is reduced. The pump
returns this bled-off volume back to the master cylinder. Once the wheel has
stabilized and almost reached the same velocity as the vehicle, the outlet
valve is closed and the inlet valve is opened (both without current). The
build-up of braking pressure at the wheel may proceed. The rise in pressure
is kept within moderate bounds by rapidly pulsing the inlet valve, in order
to achieve optimal contact between road surface and tire.
Realizing the control described in the foregoing requires fast-acting valves,
in order to keep short the time interval between identifying a wheel's propen-
sity to lock and reducing its braking pressure. Typical switching times are
Tandem master cylinder

with booster
Low pressure
accumulator
Fig. 4.52. Schematics of an ABS
less than 5 ms. The opening cross-section of the outlet valve must be di-
mensioned such that a drop in pressure can be achieved sufficiently quickly,
but also in such manner that the wheel has sufficient time to react properly,
thereby avoiding an unnecessarily steep drop in pressure. The inlet valve's
cross-section must be large enough to enable the driver to build up braking
pressure quickly, but also small enough to enable the ABS to modulate the
pressure build-up in a moderate way. Hence, the inlet valve requires switching
times similar to those of the outlet valves, but its orifices generally tend to
be somewhat smaller. These orifices open an area in the region of 1 mm 2 .
Over and above an ABS , modern systems include TCS (traction control
system) and ESP (electronic stability control) funct ions. From a fluidpower
engineering perspective, t he chief innovation is that additional valves allow
the device to build up braking pressure for each wheel independently of the
driver; in one case in order to prevent the driving wheel from spinning, in the
other, to prevent the vehicle from turning around t he vertical axis. For the
latter purpose, the ESP requires additional data concerning the driver's in-
t entions (via the steering angle) and the vehicle's real movement over ground
(via yawning torque and lateral acceleration sensors) .
Wheel brake
Plunger
unit
Electric
adjustable
friction brake
Controller L____j~~~~~~~~------'
Fig. 4.53. Anti-lock braking system with plunger cylinder
Recent anti-lock braking systems rely on the electro-hydraulic brake (hy-

draulic brake-by-wire). This renders it possible to use valves with variable
cross-sections, such that the aforementioned dilemmas, particularly that as-
sociated with the inlet valve (small cross-section to achieve good controllabil-
ity, large cross-sections so as not to prevent the driver from rapidly building
up braking pressure) can largely be resolved [17, 18, 43].
A further system, shown in Fig. 4.53, works in an infinitely variable man-
ner. A plunger unit, driven by electrically actuated friction clutches and an
electric motor, both allows it to disconnect the master cylinder from the wheel
brake and to vary the fluid volume (and hence the pressure) at the wheel brake
in a stepless manner. The clutch's reaction times are similar to those of the
solenoid valve type mentioned above. In comparing the energy consumption
ofthe electric motor, one must bear in mind that, owing to the operating prin-
ciple of the plunger-ABS, the energy consumption decreases with increasing
brake pressures, whereas the opposite is true for valve systems [35].
4.9.2 Servopneumatic Drives for Handling Systems
Pneumatics fulfils the most diverse tasks, mainly in the automation of pro-
duction and assembly processes. However, there are increasing demands in
that field for higher flexibility and accuracy not achievable with conventional
pneumatic systems. For that reason, servopneumatic linear drives were de-
veloped, which achieve an astoundingly good positioning behavior and good
dynamic performance. These characteristics are obtained through the use of
Stroke: up to 1500 mm
Pay load: up to 22 kg
Speed : max. 3.5 m /s
Repetition accuracy: ±0.01 mm

Weight (1000 mm): 9.7 kg
Fig. 4.54. Servopneumatic drive module (GAS)
fast steady valves and the three-loop state control of the drive by means of
microprocessors.
Owing to their lower space requirements, rodless cylinders are often used
as drive cylinders. The key advantage vis-a-vis electric drive systems consists
in their compactness and the superior dynamic performance. Figure 4.54 de-
picts a servopneumatic drive module, into which all required components such
as cylinder, valves, measuring system, brake, and guides have been integrated
[10].
However, more flexibility is also demanded of short-stroke drives, so
that these drives, too, are laid out as servopneumatic drives . These short-
stroke drives (up to approximately lOOmm stroke) are mainly used today
in mounting technology, as vertical axes. Fast drive performance and a com-
pact, lightweight construction are the main advantages. Figure 4.55 depicts
a servopneumatic short-stroke module into which all required elements have
been integrated. Additionally, the motion performance with different loads is
shown.
4.9.3 Pneumatic Grippers
Apart from axis drives in handling systems, the gripper is a central element
in handling and assembly technology. As the link between robot and handled
object, it has a decisive impact on the flexibility of the overall system. Over
80% of commercially available grippers are equipped with pneumatic drives.
In this application, the specific advantages of pneumatics, such as space-
and weight-saving design , can be brought to bear. Moreover, it is easy to
generate static gripping forces. Figure 4.56 shows various two-finger grippers
in contemporary use [44] .
Gripper (a) is actuated by means of a pneumatic piston. Rocker arms
transmit the piston's movement to both gripper fingers, which are guided
in parallel. Gripper (b) , too, uses a pressure-actuated piston. The piston's
4 F luidpower Actuators 221
100
{
\'\
m=Okg
'\\
"'
mm
"'
e
Q)
..>::
4i \
m=2.5 kg
'/
I
U5 50
~=1 \
\
kg \
\\ 1\
\
\
\
I \
/
0
100 200 300 400 500 600 ms 700

Timet
Fig. 4 .55. Servopneumatic s hort-stroke module (GAS)
a b c
Fig. 4 .56. Designs of pneumatic grippers. a Bosch, b Schunk, c IFAS

movement is transmitted directly onto one of the gripper finger and via a rack-
and-pinion system onto the other. The driving piston in doubled in order to
increase the gripping force. Finally, to cut down the effects of friction, gripper
(c) is equipped with roller membrane cylinders, which substitute the sliding
motion of conventional piston seals with a rolling motion. Again, this motion
is transmitted onto the gripper fingers by means of a lever mechanism.
In most cases, grippers are actuated by switching valves today. That
means that the fingers stoke is limited by end stops or by the work piece
to be gripped. The gripping force is determined by the pressure applied.
The use of servopneumatics for gripper drives permits a significant aug-
mentation of the advantages already inherent in conventional pneumatics,
because the application of position- and force-controlled drives can greatly
enhance the flexibility of the grippers.
Through position- and force-control of the gripper fingers, a broad range
of different work pieces can be gripped, and assembly and handling tasks
can often be accomplished without having to change grippers. Moreover, it
is possible to monitor the gripping procedure and to control it in an event-
dependent manner [4, 5].
4.9.4 Simulation of Motion
Simulators have become indispensable aids today, for example in the training
of airline pilots. They make it possible to practise even dangerous flight situ-
ations without risk. For these purposes, the cockpit replica has to be supplied
with environment data in such a way that the pilots' senses are deceived.
Flight simulation relies on computer systems in which the flight behavior
of the aircraft is stored, as well as a multitude of environment data, such as the
sight of various airports as it would appear to the pilot of a plane approaching
them. Apart from the visual simulation with computer-controlled graphics
and the actuation of the plane's instruments, maximal verisimilitude calls for
simulating the motion of the plane.
A movable platform with six degrees of freedom serves to simulate motion.
As Fig. 4.57 shows, the platform rests on six hydraulic cylinders arranged in
such a way that approximately one meter of motion in every direction is
possible from the middle position shown [26].
This application is characterized by very specific demands being made
on the hydraulic cylinders. For example, the cylinders have to have very
low friction, in order to preclude stick friction at low speeds (stick-slip).
Otherwise, experienced subjects could sense such a tearing-off, which does
not occur in reality.
One solution to this problem is to use hydrostatic guides, where mechan-
ical components do not touch because they are separated by an oil film.
Figure 4.58 shows a sketch of the effects of such a wedge-shaped guide. Be-
cause of the pressure differential between the two working chambers, an oil
Fig. 4.57. Motion simulator (source: Forschungsinstrument Fahrsimulator Daim-

ler-Benz AG , Report 11/90, VDI- Verlag GmbH, Dusseldorf 1990)
volume flows through the wedge-shaped gaps. When the piston is in an ec-
centric position, different flow conditions result, which in turn cause pressure
differentials. These pressure differentials serve to center the piston, and hence
contact between piston and cylinder wall cannot occur during normal use.
Oguide
a b
Fig. 4.58. Hydrostatic guides. a Wedge-shaped guide, b guide with external pres-
sure supply
External pressure supply
Fig. 4.59. Cylinder with equal areas
Another type of hydrostatic guide used relies upon an external pressure

source. The mode of operation is similar, but this design has the advantage
that the ability of the guide to bear loads does not depend on t he load
pressure differential, but on the pressure level of the oil source.
Accordingly, the use of cylinders with hydrostatic guides allows them to
fulfil an important requirement for the application within a flight simulator.
Another important requirement is that the cylinders exhibit as uniform as
possible a behavior in both mot ion directions. Cylinders of the type com-
monly used with only one rod do not meet that r equirement, because forces
and speeds differ with the direction of motion even for the same pressure
differential, because pressurized areas differ. The cylinder with equal areas,
which has two rods in most cases, does not suffer from these disadvantages.
However, for reasons of space, this design cannot be used in simulators, which
is why special versions are used, which require no more space than a differ-
ential cylinder, but which have equal pressurized areas for both directions of
motion. Figure 4.59 shows such a cylinder, which is als o equipped with t he
hydrostatic guides described further above.
4.9.5 Numerically-controlled Machine Tool
The economically viable use of grinding machines for finishing gears is deter-
mined by the achievable machining power. Figure 4.60 shows the layout of
a numerically-controlled profile grinding machine t hat makes it possible t o ex-
ploit the performance of novel CBN cutting materials (cubic-crystalline boric
Hydromotor (pitch axis a 2)
Hydrocylinder
(translatory axis x)
Fig. 4.60. Layout of a numerically-controlled gear profile grinding machine
nitrite) through high feed- and cutting speeds. The use of novel hydraulic cir-
cuit concepts and novel materials helps to create drives that combine good
dynamic performance with high efficiencies and extremely durable machine
components.
The machine's frame consists of resinated concrete, which exhibits good
thermal and mechanical properties. The CBN-coated grinding wheel is pow-
ered by a secondary-controlled hydraulic motor with a constant speed of
lO,OOOrpm. The entire drive unit is adjustable in height, and can be turned
for producing helical gearings. The gear to be produced oscillates back and
forth underneath the grinding wheel and is moved by an hydraulic linear
drive. The work piece slide consists of a high-performance composite fibre
material and is hydrostatically guided. This carbon-fibre-reinforced compos-
ite material exhibits virtually no material fatigue, does not expand at high
temperatures, and has a much higher stiffness-to-weight ratio than metallic
materials. The drive has low friction and a moved mass of only 62 kg.
After each stroke, the gear has to be moved on by the width of one tooth
gap. If it is a helical gearing, it also has to be turned while the grinding wheel
is working on it. These spacing- and screwing mot ions superimposed on the
gear feed are realized by means of a servo-valve-controlled hydraulic motor
of the bent-axis type.
Main Drive. A variable motor attached to a constant pressure net was
chosen for the grinding wheel drive. The hydraulic circuit diagram is shown in
Constant pressure net
Released
by
control
-n2command
Tank line
Fig. 4.61. Circuit diagram of a secondary-controlled hydraulic motor attached to

a constant pressure net
Fig. 4.61. The system's working pressure is impressed, i. e. independent of the

actual load. Hence, the stepwise changes in the load torque resulting from the
grinding wheel's operation can be compensated through a rapid adjustment
of the motor 's displacement without first having to alter the pressure in the
supply net. The good actuation dynamics thus achieved make it possible to
keep the speed constant within narrow bounds [11].
Inevitable losses due to the operating principle only occur at the control
element, thus giving the drive a good overall efficiency. A further reduction
in energy use is achieved by having the motor feed oil from the low-pressure
side to the high-pressure side when it is braking. The drive requires very little
space. Its weight is only 39 kg for a power output of 90 kW. Therefore, the
frame bearing it can be accordingly small and have thin walls.
Oscillating Drive of the Work Piece. The feed drive is fixed and thus
can be rigidly connected to a variable pump by means of short pipes. There-
fore , it is possible to actuate it through a hydrostatic gear, well known for
its high efficiency. To achieve the requisite dynamic performance, a radial
piston pump with a displacement volume of 63 cm3 was chosen as the vari-
able unit. Owing to its short actuation strokes and a high-pressure actuation
system with equal areas, it achieves a stroking time of only 20 ms from zero
to maximal stroke.
In addition, the gear's static pressure level was raised through a net of re-
sistors consisting in inlet and outlet orifices (Fig. 4.62). The hydraulic preload
of the actuator on both sides has a number of advantages. The system 's nat-
ural frequency is increased by a factor of J2. The load stiffness is doubled ,
because, starting with the static pressure level, pumping an oil flow makes
it possible to increase pressure on one side of the cylinder and reduce it on
the other. As both changes contribute to the generation of the required load
pressure, it can be adapted very rapidly.
mean pressure level: PN = PA + PB = po. Gz

2 Gz + GLE + GA
Fig. 4.62. Circuit diagram of the oscillation drive
With powering loads, the gear can also feed back energy back into the
electrical net, because the function of the hydromotor, pump, and E-motor are
solely determined by the direction of the power flow. This circumstance can
be exploited particularly advantageously in the present application, because
the oscillating movement to be generated requires a permanent acceleration
and braking [2, 3, 19].
4.9.6 Application and Operating Conditions

of Theater Hydraulics
Hydraulic drives are to be found in many different places in a modern theater.

Lifting devices of all sizes are used above and below stage level, in order
to make individual persons or entire stage sets appear or disappear within
a short space of time. Above and below the stage, stage movers, turntables
and mobile special constructions are to be discovered [12].
The use of actuators in the performances of large theaters is subject to
a number of specific restrictions. First to be mentioned are the stringent
safety requirements. Work usually takes place under great pressure of time,
and involves people moving around between set decorations and underneath
dangling loads weighing several tonnes in some cases.
The drives used in stage movers, backdrops and platforms therefore all
have redundant braking and control units, mechanical end stops, pipe break-
age protection systems, brake valves, and electronic monitoring devices.
Moreover, all hydraulic components may only be used with half the nom-
inal pressure stated by the manufacturer. In order to reduce fire hazards,
theater hydraulics are only used with non-flammable fluids.
The actuators predominant in stage technology are cylinders with long
strokes and slowly running motors for powering winches. The loads to be
borne by the drives are restricted to friction and weight forces. The con-
stantly changing coupled masses are often very large stage structures prone
to oscillation, and thus have to be moved very carefully. Therefore, drives
are required that exhibit a delicate start-and-stop performance despite great
differences in stroke, speed, and loads, and which also must not jerk to a halt
in the event of an emergency stop. The speed ranges required (up to 1 m/s
with passengers and 1.2m/s without humans) must be adjustable in small in-
crements and must deliver constant performance even with creeping motions.
Equally important is the ability to run several drives in parallel in various
constellations and to adapt the speed to the events taking place on stage.
A further desideratum in stage hydraulics are minimal noise emission lev-
els. So as not to interfere with the performance, the noise level on an empty
stage must not exceed 35 dB A. In this context, hydraulic drives have the ad-
vantage of allowing actuator and pressure supply to be physically separated,
and of not having fan noises occurring at the drive. Furthermore, if large-
capacity slow-running motors are used, gears can be dispensed with, which
eliminates another source of noise.
Just like any other application of technology, theater hydraulics must be
subjected to the test of economic viability. The great advantage of fluid-
power technology in this respect is that it is very easy to concentrate fluid-
power energy in accumulators. A theater is not a manufacturing plant, but
requires great power peaks for short periods of time during the performances.
Using electric drives would involve having the installed power match those
peak values, which would be very expensive. Theater hydraulics, however,
E
LO
..t
6m
Fig. 4.63. Hydraulic drive of a stage decoration (source: Mannesmann Rexroth)

use a central pump station equipped with large-capacity piston accumulators

with downstream nitrogen tanks.
Figure 4.63 shows an example of a special decoration that could only
have been realized with an hydraulic drive. The fingers of a giant hand,
symbolizing the invisible, omnipresent might of the Inquisition in Verdi's
opera Don Carlos, are driven by an hydraulic cylinder. This is an impressive
example of the way in which stage technology has become a part of the artists'
means of expression.
The layout of the hydraulic circuit can be seen in Fig. 4.64. The direction
of motion and the speed control rely on hand-actuated valves, such that the
motions can be made to match the rhythm of the performance. The sup-
ply pumps are situated beneath the stage and are connected to the actuator
by means of clutches and hoses. During the performance, the required pres-
Fig. 4.64. Hydraulic circuit diagram of a stage decoration's cylinder drive

sure oil is supplied by a charged accumulator hidden in the set decoration.

The accumulator's size is calculated such as to allow all movements to be per-
formed without having to be recharged. This enables the plant to be operated
noiselessly with the stage curtains open. As Fig. 4.64 shows, the cylinder is
hydraulically preloaded through a pressure relief valve, so that the motion
can be transmitted to the mechanical part smoothly and without jerks.
References
1. Anders, P. (1986): Auswirkungen der Mikroelektronic auf die Regelungskon-

zepte fiuidtechnischer Systeme und der Einsatz von Personal Computer als
Auslegungswerkzeug. Doctoral dissertation, RWTH Aachen.
2. Backe, W.; Berbuer, J. (1987): Neue Schaltungskonzepte fur hydrostatische
Getriebe. O+P Olhydraulik und Pneumatik, Vol. 26, No. 6.
3. Backe, W. (1982): Konstruktiv und Schaltungstechnische Massnahmen zur En-
ergieeinsparung. O+P Olhydraulik und Pneumatik, Vol. 26, No. 10, pp 695-
706.
4. Backe, W.; Eschmann, R. (1993). SSP: A Simulation Program for Pneumatics.
Proceedings of the 6th Bath International Fluid Power Workshop, Sep. 23 to
24 1993, Bath, UK.
5. Backe, W. (1986). The Application of Servo-Pneumatic Drives for flexible me-
chanical handling Technique. Robotics, 1986 N2.
6. Blackburn, J.F.; Reethof, G.; Shearer, J.L. (1960). Fluid Power Control. MIT
Press and John Wiley & Sons, Inc., New York and London.
7. Chochia, G. A.; Tilley, D. G.; Nguyen-Schaefer, H. (2001). Numerical and
Experimental Investigation on the Shockwave-related Cavitation Flow. Journal
of Systems and Control Engineering (Part 1), Vol. 215, No. 11.
8. Daachi, B.; Bellaleque, A. (2003). A Stable Neural Adaptive Force Controller
for a Hydraulic Motor. Journal of Systems and Control Engineering {Part 1),
Vol. 217, No. 14.
9. Franklin, G.F.; Powell, J.D.; Emami-Naeini, A.(1991). Feedback Control of Dy-
namic Systems. 2nd ed., Addison- Wesley Publishing Co., Reading, Massachus-
sets, Ch. 8.
10. Goedecke, W.-D. (1989). Servopneumatische Antreibssysteme. O+P Olhy-
draulik und Pneumatik, Vol. 33, No. 8.
11. Haas, H.-J. (1989). Sekundiirgeregelte hydrostatische Antriebe im Drehzahl- und
Drehwinkelregelkreis. Doctoral dissertation, RWTH Aachen.
12. Herz, E. (undated). Theatergerechte Hydrauliksysteme. Mannesmann Rexroth
RD 03141.
13. Hibi, A.; Ichikawa, T. (1975). Torque performance of hydraulic motor and its
mathematical model. 4th Int. Fluid Power Symposium, Sheffield, UK, April
1975.
14. lvantysyn, 1.; lvantysynova, M. (2000). Hydrostatic Pumps and Motors -
Principles, Design, Performance, Modelling, Analysis, Control and Testing.
Akademia Books International, First English ed.
15. Kett, R.; Murrenhoff, H. (1995). Circuit Design with Sophisticated Modern
Simulation Tool. 8th Bath International Fluid Power Workshop, Bath, UK.
16. Klein, A. (1993). Einsatz der Fuzzy-Logic zur Adaption der Positionsregelung
fluidtechnischer Zylinderantriebe. Doctoral dissertation. RWTH Aachen.
17. Klein, H.-C. (1989). KW-ABS Bremssysteme mit weiteren integriertenfunk-
tion. Automobil Industrie, Vol. 7, No. 2.
18. Leiber, H. et al. (1980). Antiblockiersysteme (ABS) fiir Personenkraftwagen.
Bosch Technische Berichte, Vol. 7, No. 2.
19. Lennevi, J.; Palmberg, J.-0.; Rydberg, K.-E. (1994). Microproocessor control
of hydrostat transmissions adapting to driving conditions. 11. Aachner Flu-
idtechnisches Kolloquium, Aachen, Germany. IHP /RWTH.
20. Lewis, H. J.; Stern, E. E. (1962). Design of Hydrostatic Control Systems.
McGraw-Hill Book Company, Inc., New York, Toronto, London.
21. McCloy, D.; Martin. H. (1980). Control of Fluid Power: Analysis and Design.
John Wiley, NY.
22. Merritt, H. E. (1967). Hydraulic Control Systems. John Wiley & Sons, New
York, London, Sydney.
23. Morse, A. C. (1963). Electrohydraulic Servomechanisms. McGraw-Hill, Inc.,
New York.
24. Moore, P.R.; Pu, J.; Harrison, R. (1992). Progression of Servo Pneumatics
Towards Advanced Applications. Proc. 5th Bath Int. Fluid power Workshop,
Sept 16 to 18 1992, Bath, UK.
25. Olems, L. (2000). Investigation of the Temperature Behaviour of the Piston
Cylinder Assembly in Axial Piston Pumps. International Journal of Fluid
Power, 1-2000, No. 1, pp 27-38.
26. Philipp, P.A. (1990). Forschungsinstrument Flugsimulator. Report Daimler-
Benz AG, 11, VDI-Verlag, Dusseldorf.
27. Pohl, J.; Sethson, M.; Krus, P.; Palmberg, J. 0. (2002). Modelling and Valida-
tion of a Fast Switching Valve intendend for Combustion Engine Valve Trains.
Journal of Systems and Control Engineering (Part 1), Vol. 216, No. 12.
28. Prest, P.; Vaughan, N. D. (1987). Drive circuits for pulse width modulated
valves. Proc. International Conference on Fluid Power, March 24-26 1987,
Tampere, Finland, pp 217-225.
29. Pu, J.; Weston, R.H.; Moore, P.P. (1992). Digital Motion Control and Profile
Planning for Pneumatic Servos. Transaction of the ASME Journal of Dynamic
Systems, Measurement and Control, December 1992, Vol. 11, pp 634-640.
30. Renn, J. L. (2002). Position Control of a Pneumatic Servo Cylinder using
Fuzzy-sliding Surface Controller. International Journal of Fluidpower, Vol. 3,
No. 3, pp 19-23.
31. Scholz, D. (1990). Auslegung servopneumatischer Antriebsysteme. Doctoral
dissertation, RWTH Aachen.
32. Sethson, K.M.; Vaughan, N.D. (1993). A model of magnetic characteristics
of fast switching on/off valves. Circuit, Component and System Design: Fifth
Bath International Fluid Power Workshop, Bath 23-24 Sept. 1993.
33. Surgenor, B.W.; Iordanou, H.N. (1994). Experience with Advanced Position
Control of Pneumatic Actuators. Proc. !FAG Int Workshop on Trends in Hy-
draulic and Pneumatic Components and Systems 1994.
34. Tang, J.; Walker, G. (1995). Variable Structure Control of a Pneumatic Ac-
tuator. ASME Jrnl Dynamic Systems Measurement and Control, Vol. 117, pp
88-92.
35. Tibken, M. (1990). Ein Neues ABS nach dem Plunger Prinzip. ATZ Automo-
biltechnische Zeitschrijt, Vol. 92, No. 1.
36. Thoma, J. (1969). Mathematical models and effective performance of hydro-

static machines and transmissions. Hydraulic and Pneumatic Power, Nov.
1969, pp 642-651.
37. Vaughan, N.D.; Plummer, A.R. (1990). Some issues in pole-placement control
of electro-hydraulic servosystems. 3rd International Fluid Power Workshop,
Bath, UK, September 1990.
38. Vaughan, N.D.; Gamble, J.B. (1996). The modelling and simulation of a pro-
portional solenoid valve. Journal of Dynamic Systems, Measurement and Con-
trol 1996, pp 12Q-125.
39. Viersma, T.J. (1980). Analysis, Synthesis and Design of Hydraulic Servosys-
tems and Pipelines. Elsevier Scientific Publishing Co., Amsterdam, Oxford,
New York.
40. Walzer, W. (1990). Elektronische Steuerung und Regelung von hydrostatischen
Mobilgetreiben. O+P Olhydraulik und Pneumatik, Vol. 34, No. 6.
41. Watton, J. (1987). The Dynamic Performance of an electrohydraulic
Servovalve-Motor System with Transmission Line Effect. Transaction of the
ASME Journal of Dynamic Systems Measurement and Control, Vol. 109, pp
14-18.
42. Wieczorek, U.; Ivantysynova, M. (2002). Computer-aided Optimization of
Bearing and Sealing Gaps in Hydrostatic Machines - The Simulation Tool
CASPAR. International Journal of Fluid Power, Vol. 3, No. 1, pp 7-20.
43. Wu, M.C.; Shih, M.C. (2001). Hydraulic Anti-Lock Braking Control using Hy-
brid Sliding-mode pulsewidth Modulation pressure Control Method. Journal
of Systems and Control Engineering (Part 1}, Vol. 215, No. 12, pp 177-187.
44. Zumbriigel, J. (1990). Flexible Servopneumatische Greifersysteme. Doctoral
dissertation, RWTH Aachen.
5 Unconventional Actuators
Hartmut Janocha
5.1 Introduction
Unconventional actuator is a rather pragmatic label for an energy con-

troller/ energy converter combinations whose functions are based on other
physical effects than described so far. According to this definition one would
have to consider a large number of actuator principles, including those which
are not really relevant in practice. A selection is therefore made under the
following two criteria: The first says that the actuators presented here are
based on special properties of new or improved materials, the second says
that they must have passed the laboratory stage of development and/or they
must be on the verge of a large-scale application in the industry.
Unconventional actuators have gained importance during the past few
years due to the growing number and variety of their applications. How-
Hydraulic
Electrostatic
0 0.2 0.4 0.6 0.8

Actuator efficiency
Fig. 5.1. Normalized actuator attributes: power-to-weight ratio versus efficiency

of actuators (by [4])

234 Hartmut J anocha
ever, even within the very same actuator family the most important pa-
rameters, such as deflection, force, operating frequency, etc. cover such wide
value ranges, that it does not seem possible at first to name family specific
properties which could simplify the user's pre-selection. In such cases, plots of
normalized attributes often pinpoint the fundamental differences between the
different actuator families. Figure 5.1 is such an example showing the power-
to-weight ratio being plotted against actuator efficiency (power out/power
in).
This report is based on a large actuator database developed by [4]. The
different actuator families overlap, but on the whole, they are widely sepa-
rated. Bi-metallic and shape-memory actuators are intrinsically inefficient
since they rely on heating and cooling. However, these materials possess
a high energy density (working capacity per volume) which makes their use
attractive in microactuators. Piezo and magnetostrictive actuators can de-
liver high power outputs even though the displacements they can perform
are very small. Both achieve high power outputs because they can be cycled
at very high frequencies. While reading the following sections, the reader
should keep these coherences in mind for his evaluation.
5.2 Electrochemical Actuator
5.2.1 Electrochemical Reactions
The principle of the electrochemical actuator (ECA) is based on evolving gas

during the application of a small voltage. In a closed system pressure builds
up and is then transformed via appropriate design features into mechanical
work. The pressure is reduced by means of a short-circuit or by pole rever-
sal. Depending on the requirements, distinct electrochemical reactions are
applicable. Some of these reactions are described below [2].
Gases which generate the working pressure in fuel cells are produced via
electrolysis, whereby the pressure build-up is quite slow. Depending on the
catalytic activity of the electrode surface and the actuator construction, one
to three minutes must be taken into account. The reverse reaction transforms
the gas back into water. It is initiated by a short-circuit and slower by a factor
of two to three. The maximal forces achieved by a test actuator lay between
1000 and 2000 N, and the displacement is 5 mm.
The reaction known as electrochemical oxygen pump transports oxygen
using a very low electrical potential. A pressure build-up is achieved by trans-
porting oxygen from the air via its liquid phase into the pressure chamber.
In the reverse reaction, the oxygen is exhausted back into the air. As a result
of this operating principle, the system needs to be open to the air on one
side. The electrochemical oxygen pump can be applied wherever a steady
movement in one single direction (stroke only) is required, as the reactions
5 Unconventional Actuators 235
are very slow, lasting up to several days. The electrochemical oxygen pump
can only produce small forces for long strokes (up to 30 mm and beyond).
The reaction of the silver electrode in an alkaline electrolyte is an example
of a solid-state reaction. Pressure builds up by the hydrogen produced on the
counter electrode while the silver oxidizes. The reactions can be summarized
in the following way:
Charging
Ag0+H 2 (5.1)
Discharging
In contrast to the fuel cell reaction,

the solid-state reaction does not require the separation of gases and the
division into two expansion spaces, and
the actuator can be reset without feeding energy by short-circuiting the
electrodes.
First actuators based on the silver-hydrogen system were built in the early
1990s [2]. Later, nickel-hydrogen systems which had been used in other fields
such as energy storage were applied. The reaction in these systems can be
described in the following way:
Charging
2Ni(OHh 2NiO(OH)+H 2 (5.2)
Discharging
A comparison of both cell reactions indicates that the latter

- is more favorable in terms of reversibility, as it does not produce water,
and
is more suitable for use in actuators since usually higher numbers of load
cycles can be achieved.
5.2.2 Technical Implementation
ECAs based on nickel-hydrogen technologies comprise an arrangement of elec-

trodes made of active carbon (cathode) where hydrogen is produced and of
nickel hydroxide (anode). The electrodes are separated by porous fleece ma-
terial, which has been saturated in an alkaline electrolyte. Its concentration
remains constant, as none of the water is converted. Such a basic cell has an
open circuit voltage of 1.32 V and is charged at 2 V DC in actuator operation.
In practice, several cells are connected in series, so that the supply voltage of
these ECAs can range from 2 V to 36 V. The maximum current consumption
can range from 100 rnA to 2 A.
The whole arrangement of electrodes is built into a bellows element made
of stainless steel which is hermetically sealed and which - depending on the
actuator's force- must cope with an internal pressure of up to 50 bar. During
the charging process, which begins at a certain minimal voltage, the bellow
236 Hartmut Janocha
Bellow Gas-proof welded
Electro-
a chemical cell (Sx)
b lime -
Fig. 5.2. Electrochemical actuator. a Design, b voltage, current and force-time

d iagram (by [3])
expands along its axis, whereby mechanical energy is stored. Some of this
energy is set free while the actuator i s retracted (discharged).
In the 1990s commercial actuators were produced with displacements of
up to 25mm and forces of up to 30kN. Actuation speeds ofO.l. . . 1mm/ s and
up to 100,000 cycles were reached. The temperature range for application was
between - 5 und +60°C. Due to its sealed design and the low current and
voltage it was applicable in explosive areas.
Figure 5.2 a shows the basic design of the E CA SK5 /300-AF from the com-
pany Friwo/Germany (operating voltage: 12 V; current : 300 rnA; max. stroke:
5mm; max. force: 300 N; dimensions (in mm) : ¢ 37 x 30). Figure 5.2b displays
the correlation of charging voltage, current and force, while driving t he ECA
up and down. The increase in force is approximately linear during charging.
During discharge its decline is also linear towards the starting position ( F start
= 30 N) . The application of higher currents changes t he progression of t he
voltage curves as shown in the figure . Figure 5.3 describes the correlat ion of
the achievable displacement and the required cha rge [3].
Since this principle features a correlation between the input c harge and
the hydrogen produced, the measured charge points out either the force or
F=30 N ~ I .
i 100 N/
I
I
I I
II I
/300 N
Y· .'
~/
0
0 5 10 15 20 mAh 30
a Charge-
c=5 N/mm
I
I
20 N/mm _.
~
I
I
I _.
_.
I
I
I _.
I
I 60 N/mm
_.
I/
_.
_.
0
w.:--
_. /
0 5 10 15 20 mAh 30
b Charge-
Fig. 5.3. Displacement-charge diagram. a Under constant load, b working against

a spring (by [3])
Table 5.1. Important properties of electrochemical actuators
low energy demand, directly slow (5 mm stroke > 20 s)
intermateable in microelectronic leakage problems (loss of gas
systems (1. .. 5 V, 1. .. 4 W) when operated during an
no holding energy required extremely long period of time)
(3-point-control mode) unknown long-term stability
adopting the safety position closed loop control (position
without feeding energy by sensing) is required
means of a short-circuit limited maximal stroke
noiseless operation
the displacement, as long as the other value is given and the temperature
remains constant. The manipulating speed of the ECAs is approximately
determined by the ratio of the electrode arrangement volume (as a measure
for the hydrogen produced per time) to the expansion vessel volume and is
therefore already predefined by the manufacturer. The user is able to influence
its durability, that is, the number of load cycles, by keeping the charging and
discharging current and the overall charge as low as possible, however, with
238 Hartmut Janocha
the negative consequence of reducing the displacement speed. Table 5.1 lists
the advantages and disadvantages of electrochemical actuators with solid-
state reaction.
5.2.3 Application Examples
A market segment for the electrochemical actuator with solid-state reaction,

as described in Sect. 5.2.2, is probably to be found between the well-priced
expansion element and the expensive motor valve. Imaginable areas of appli-
cation are therefore, among others,
- throttle control,
- regulation, proportioning of fluids and gas,
- positioning systems,
- room heating control.
Apart from charging and discharging, the ECA features a third operat-
ing mode, which is hold. Here, each value of position and force within its
operating range is "memorized" without further input. The unavoidable self-
discharge causes a reduction in force and distance, and from 1%/h at room
temperature it drops back soon to < 0.1 %/h. The possibility of discharge
across an active electronic load suggests the idea of using the energy of the
displaced ECA, e.g. during a power break down, as an emergency power
supply. Two tested applications are described below.
ECAs can be applied in level control systems and for low-frequency vi-
bration isolation of optic tables by applying them as electrically controllable
actuators and as gas operated compression springs. By including them in
a closed control circuit with a biaxial inclination sensor, it is possible to
achieve precise levelling of the table. ECAs produce the working medium,
the hydrogen, directly and therefore make the air supplies and compressors
known from pneumatic systems dispensable. In places that are difficult to
access, it is now even possible to produce forces of magnitudes as high as
they used to be reserved for hydraulic systems.
The other task consists in opening and closing a ventilation window with
electric control. So far, the problem has been solved by applying an electrome-
chanical drive, however, with a disturbing noise development and a ray of me-
chanically moving components prone to wear. These disadvantages have now
been eliminated by two push-pull operating ECAs that produce the pushing
and pulling forces required to actuate the ventilation window (single ECAs
produce pushing forces only).
5.2.4 Development Trends
Future R&D projects will aim at finding possibilities of expanding the work-
ing temperature range and raising the durability, as much as on miniaturising
the actuators. Another goal might be the integration of control electronics

into the ECA.
Additionally, R&D focuses on raising the actuation speed by one order
of magnitude. One constructive approach to achieve this is to replace the
(expensive) metal bellows by construction elements which feature a lower,
dynamic reducing dead volume.
Presently, ECAs are custom designed, as for instance the ECAs offered
by the Med-e-Cell company [5]. The manufacturers are focusing so much on
modular designs, that any combination of standardized electrode arrange-
ments and of expansion elements will be possible.
5.2.5 Comparison to Competing Actuator Principles
The ECA is comparable to thermal expansion elements and thermo-chemical

metal hydride actuators with respect to their actuating force, their displace-
ment and response time. They, too, can function as positioning element, for
example, in valves and fittings for heating, ventilation and general control
engineering. The advantage of the ECA is the possibility of direct influence
on its position and control behavior without need of a temperature change.
Therefore, rival actuators usually feature a higher demand of energy.
Expansion Elements
Expansion elements make use of the large volume-temperature dependence

of solid and liquid materials with a high thermal expansion coefficient. With
constructive measures, the volume expansion resulting from a rise of temper-
ature is converted into the movement of a piston. The piston's return motion
with dropping temperature is supported by a spring element [6].
The expansion material, e.g. wax, paraffin, or silicon oil is stored in a firm
vessel, that must resist pressures of up to 150 bar (15 MPa), depending on the
specific design. The design in Fig. 5.4 is based on a head-shaped elastomer
insert inside which the piston can travel. This design allows large strokes and
great loads at the same time.
Expansion elements possess very distinct stroke-temperature character-
istics, depending on the applied expansion material. Hydrocarbons feature
a higher volume increase within the area of melting temperatures, that is
a temperature range of 15 K, than in the following liquid stage. In compari-
son, liquid expansion materials have a wider linear control range, but smaller
specific stroke; see Fig. 5.4b.
Expansion elements are popular mass-produced articles in many areas,
such as heating and climate control. They mainly function as controlling
elements without auxiliary power and are available in product series. Their
most important parameters typically cover the ranges given in Table 5.2.
Due to a possibility of electric control, expansion elements become ac-
tuators in terms of the definition in Sect. 1.1. Primarily PTC resistors are
240 Hartmut Janocha
Working piston Stroke
Elastomer
element
Expansion
material
a Case
Max. temperature
t
s
E
E ~
oF±~~--------~-+--~~~
fJ -
b Operating range
Fig. 5.4. Expansion element. a Design with elastomer insert , b stroke-temperature

diagram for a liquid expansion material (by [6])
Table 5.2. Values of expansion elements (by [6])
Range of working temperature - 20 .. . +120 oc

Stroke within the operating range 5 ... 15 mm
Maximum stroke 6 ... 25 mm
Maximum force 250 ... 1500 N
Response time 8 ... 50 s
used to heat expansion elements. The temperature dependence of resistance

in PTC elements has a limit ing function and can in some cases eliminate the
need for a c ut-off switch.
Metal Hydride Actuator
The electrochemical actuator is related to the thermo-chemical metal hydride

actuator. Its function is based on the sorption of hydrogen in certain metals
and alloys. By means of an (electrical) heat source, the metal hydride is
reversibly pyrolyzed and hydrogen is released. This exothermic reaction is
reversed by cooling. Compared to electrochemical actuators, its operating
temperature range is much higher and could - depending on the applied

metal hydride- be as great as 1000 K. A prototype, designed in the 1990s by
the Japanese Aisin Cosmos company, featured a displacement of 15 mm and
produced forces of up to 4800 N for a thermal power of maximal 200 W.
5.3 Actuators with Shape Memory Alloys
5.3.1 Shape Memory Effect
The shape memory effect was discovered by chance in certain copper alloys
during the 1950s and was later found in nickel titanium (NiTi) alloys as
well. This unusual behavior is based on the reversible transformation from
the martensite into the austenite phase of a shape memory alloy (SMA). The
material adopts two different crystal structures depending on its temperature
(see Fig. 5.5) or external mechanical loading. SMA elongation exceeds the
elastic extensibility of metallic structure materials by far.
The temperature induced transformation is characterized by four temper-
atures: '!9Ms and '!9Mr during cooling and '19 As and '19 Af during heating. '!9Ms and
'!9Mr indicate the temperatures at which the transformation from the austen-
ite phase into martensite respectively starts and finishes (see Fig. 5.5); the
same applies to '19 As and '19 Af. The overall transformation describes a hystere-
sis on the order of 10 to 50 K in temperature. The temperature at which the
transformation occurs can be chosen in a range of -200 ... +200°C depending
on the composition and the microstructural constitution.
The martensite phase can also be induced by stressing the sample at
a constant temperature above '19 Af. The stress at which the transformation
occurs is almost constant until the material is fully transformed. When the
"C
Q)
§
~r:::
~
Q)
E
::::J
~ 0
5 ... 30 K
Temperature (b)
Fig. 5.5. Volume fraction transformed to martensite as a function of temperature

242 Hartmut Janocha
applied stress is released again, the reverse transformation (martensite to

austenite) will occur at a lower stress level as during loading. The material
is said to be superelastic or pseudoelastic, and reversible stress up to 8% can
be obtained.
SMAs have a high damping capacity in the martensite state as a conse-
quence of a strong amplitude dependent internal friction. During superelastic
cycling energy is dissipated as a result of the stress hysteresis between loading
and unloading. Important to notice is also that many characteristics, such as
the Young's modulus and the electrical resistivity, change drastically during
transformation. These changes can be used in specific applications and also
for feedback control of SMA actuators.
The unique property of superelasticity, combined with a good bio-compa-
tibility has lead to new, more versatile and less complex instruments and im-
plant devices in the medical field (surgical tools, stents, bone anchors, ... ).
The temperature induced transformation forms the basis for actuator appli-
cations, in a narrower scope [7], and is therefore explained below in more
detail considering the example of a shape memory spring.
One-way memory effect (Fig. 5.6a): The sample is deformed (A---. B)
and unloaded (B ---. C) at a temperature below 19Mr. The apparent plastic
deformation is restored during heating to a temperature above 19 Af ( C ---. D).
This effect is called the one-way memory effect, since only the hot shape
is memorized. The transition temperatures are determined by the alloying
processing parameters and can be varied by the SMA suppliers.
Two-way memory effect (Fig. 5.6b): A spontaneous shape change oc-
curs during cooling to a temperature below 19Mr (A---. B). This shape change
is recovered during subsequent heating to a temperature above 19Af (B ---. C).
In the two-way memory effect the material remembers both the hot shape and
the cold shape. This effect is only obtained after a specific thermomechanical
treatment by the SMA supplier, called training.
The following two cases are of interest for the application of actuators:
- When the shape recovery from the cold shape to the hot shape ( C ---. D in
Fig. 5.6a or B ---. C in Fig. 5.6b) is impeded, high recovery stresses of up
to 800 MPa are gradually generated during heating. This effect is used for
diverse clamping and fixation devices, such as SMA couplings and SMA
connectors.
- Starting from the points C and B in Figs. 5.6a and 5.6b a new operat-
ing point is determined in the F -c plane by an external force (actuator
load). Afterwards, the shape recovery during heating causes the spring to
move in the opposite direction of the force, which corresponds to work
production of up to 5 Ws/g.
Comparing both application possibilities, one comes to the conclusion
that the definition of actuators according to Sect. 1.1 is only absolutely cor-
rect in the second case, where the spring is deliberately heated to perform
A
{)<{)MI {)<{)MI {)<{)MI {)>{)AI
, D
Iirl
// /
h
ff / /
L 'A
B
//// //
A B c D
E
~ ///
a =;.-c
A,C
h
{)>{)AI {)<{)MI {)>{)AI
b
Irl A B c B
E
Fig. 5.6. Shape memory effect. a One-way effect, b two-way effect (strain, load,
and temperature are indicated by c , F, and TJ, respectively) [7]
a completely controlled action. This mode of operation therefore forms the

basis of the further SMA actuators to be discussed.
Interestingly, it is possible to achieve, for instance, in a wire shaped mem-
ory element, an almost linear dependency of the ohmic resistance on the
strain by means of a thermomechanical treatment, allowing the implemen-
tation of the self-sensing actuator principle according to Sect. 1.4.2, among
other things.
Finally, apart from the superelastic and thermic shape-memory effect,
there exists a magnetically controllable magnetic shape memory (MSM). It
occurs in ferromagnetic materials with martensite structure, e.g. Ni 2 MnGa,
which is characterized by a high magnetic-crystalline anisotropy with twin
structures. These twin structures grow in the magnetic field leading to macro-
scopic material deformation of up to about 10% for a field intensity of about
400 kA/m. A corresponding actuator consists, for instance, of a bar-shaped
MSM element, the magnetic circuit and a device in order to initiate the me-
chanical pre-stress along the bar axis. Upon applying a magnetic field per-
pendicularly to the bar axis, the twin structures orient themselves parallel to
the field direction causing the MSM element to lengthen along the bar axis.
244 Hartmut Janocha
As soon as the field has been switched off, the element, under mechanical
pre-stress, contracts to its original length.
When the MSM effect is applied, the influence of eddy currents and hys-
teresis must be taken into account. The literature describes several actua-
tor implementations with strokes of up to 5 mm and forces of up to 2 kN.
Their operating frequencies range from direct current into the kilohertz re-
gion achieving response times on the order of a millisecond [31]. Apart from
this direct MSM effect, an inverse MSM effect exists as well. The mechanical
actuator load influences the proportions of the twin structures and leads to
a change in the external magnetic field. MSM materials are single crystalline
metal alloys. Since they are expensive to produce, the future application of
the MSM effect depends strongly on the further price development of the
material.
Materials
Up to now, shape memory effects have been discovered in various materials,

which can generally be classified into noble-metal, copper, iron NiTi based
alloy systems and non-metallic shape memory materials. SMAs based on
nickel and titanium have provided so far the best combination of material
properties for most commercial applications. The shape memory effect in the
NiTi system was first discovered in 1962 at the Naval Ordinance Laboratory
(NOL), USA, and thus the system is named Nitinol® 1 , which is commonly
used when referring to NiTi-based alloys. NiTi is smelt, for instance, in an
electron beam or a vacuum induction furnace. The warm forming of casted
NiTi blocks is done by rolling or by extrusion molding. The same applies to
cold forming methods for manufacturing half-finished products.
Table 5.3 lists several important properties of NiTi and Cu based SMAs.
For most applications NiTi alloys are predominantly used these days. Their
one-way effect and their two-way effect are both stronger, the risk of overheat-
ing is lower, their long-term stability better, and the corrosion performance
is excellent. Besides, NiTi SMAs have a higher electrical resistivity making
electrical activation much simpler. The conversion temperatures of NiTi and
Cu based SMAs lay in the same range and are predetermined by the spe-
cific chemical composition of the alloy. Currently, alloys are available with
conversion temperatures reaching from -200 up to +200°C. With special
pre-treatments it is possible to achieve higher conversion temperatures for
alloys making use of the one-way effect.
1 The syllable NOL is also used in the name of the magnetostrictive material
Terfenol (compare Sect. 5.5.2) and indicates the same place of origin.
Table 5.3. Properties of shape memory alloy families in treated conditions

NiTi CuZnAl CuAlNi Units
Maximum one way 7 4 6 %
memory effect
Normal two way 3.2 0.8 %
memory effect
Normal working stress 100 ... 130 40 70 N/mm 2
Young's modulus 95 70 ... 100 80 ... 100 10 3 N/mm 2
Ultimate tensile 800 ... 1000 800 ... 900 1000 Njmm 2
strength, martensite
Elongation at fracture, 30 ... 50 15 8 ... 100 %
rnartensi te
Transformation temp. -100 ... +110 -200 ... +110 -150 ... +200 oc
range
Hysteresis 30 15 20 K
Maximum overheating 400 150 300 oc
temperature
Thermal conductivity 10 ... 18 120 75 WjmK
Thermal expansion 6.6 ... 10 17 17 10- 6 1/K
coefficient
Electrical resistivity 0.5 ... 1.1 0.07 ... 0.12 0.1. .. 0.14 10- 6 rlm
Density 645 790 715 103 kgjm 3
Normal number of 2': 100000 2': 10000 2': 5000
thermal cycles
Corrosion resistance excellent fair good
Biological excellent bad bad
compatibility
Heating of Memory Components
Shape memory elements can be heated by different means, namely by radia-

tion, conduction or convection and by direct inductive or resistive heating. For
a fast and homogenous response, resistive heating offers the most attractive
solution and is therefore also widely used. Because of the use of temperatures
below 100°C all radiation effects can be neglected. When the enveloping fluid
is not moving, thermal conduction is predominant. Convection is the main ef-
fect when the fluid is in motion (forced circulation). A favorable constellation
is a memory component located in a stream of a liquid or gaseous medium
whose change in temperature results in the deformation of the component.
In that case, the component combines the functions of a sensor and an actua-
tor. Resistance heating wires that are either braided or embedded in silicone,
as well as circumferential heat conductors and heating foils can be used as
contact heaters. For the initiation of the memory effect one may also use the
ohmic heating of the current-carrying component. With this way of warming,
it is possible to initiate the memory effect abruptly by means of a current
pulse.
The speed of deformation is determined by the speed of the heating up
and cooling down. Heating can occur within milliseconds but cooling can
only be improved by forced circulation of air or liquid or by a high surface to
volume ratio (strip rather than round wire). In this respect, SMAs with high
transformation temperatures would also improve the performance of this type
of actuator. Furthermore, reduction of the total mass of the actuator will also
246 Hartmut Janocha
Fig. 5. 7. Examples of memory alloy component implementations (left hand: cold

state, right hand: warm state; source: Krupp, Germany)
improve speed. As such, micro-devices can be an attractive application area

for SMA actuators (see Sect. 5.8.3).
Finally, it should also be taken into account that during cooling, the
martensite transformation occurs exothermically which means that extra heat
has to be removed.
Cooling-off behavior can be improved indirectly through constructive
measures by implementing the actuator's functions by means of agonist-
antagonist structures (compare Fig. 5.69), as occurs similarly in the muscles
of many living creatures.
Dimensioning of NiTi Components
NiTi SMAs are produced as custom-designed formed components or as or-

dinary standard components, such as taut wires, coil springs or torsion and
bending elements (see Fig. 5.7). The main advantage of a spring is that it
generates a large macroscopic displacement out of a relatively small micro-
scopic stress. But the stress distribution over the cross section of the spring
is not constant. This implies that a greater material volume is needed for
generating a given force. This has a negative effect on the efficiency and the
bandwidth of the spring based actuator because, for a given output, a larger
material volume has to be heated and cooled. When using wires as an active
element the required amount of work is generated with a minimal use of shape
memory material. Therefore tension loading has a much higher efficiency than
other load cases such as torsion or bending.
Further information onthe design of NiTi components for the production
of SMA actuators is given below.
Uniaxial Strained Taut Wire. It is very beneficial if uniaxial and tension

strained elements can be applied. Their dimensions are determined by the
limiting normal stress a= F/A. The memory effect is given by EM = D.l/l,
with the wire's elongation D.l (l: original length, A: cross-section). Table 5.4
lists the design data for uniaxial taut wires.
Table 5.4. Design data for the two-way effect with NiTi wires exposed to tensional
forces (heating temperature: 90°C < ilw < 150°C, cooling temperature: ilc < 45°C,
tensile stress during heating process: aw = 130 N/mm 2 , minimum reset stress during
cooling process: ac = 35 N/mm 2 , repeatable two-way effect: eM = 3.5%)
Wire diameter Tensional force Minimum reset Memory Useful work per
during heating force during shortening per lOOmm wire
process cooling process lOOmm wire length
length
mm N N mm Nmm
0.2 4 1 3.5 10
0.4 16 4 3.5 42
0.6 37 10 3.5 95
0.8 65 17 3.5 167
1.0 100 27 3.5 260
1.5 230 62 3.5 590
Table 5.5 lists data for the case that the memory wire is to be heated via
its ohmic resistance.
Table 5.5. Direct heating via the ohmic resistance (tensional force applied to NiTi
wires)
Wire diameter Heating voltage Maximum Heating time* Cooling time*
per lOOOmm continuous
wire length current
mm v rnA s s
0.12 16.6 250 3 4
0.2 11 470 7 7
0.4 5.5 920 15 50
* in 20° C calm air
Coil Spring. The dimensioning can be carried out according to DIN EN

13906. Since this area of application is very diverse, the memory specific di-
mensioning is complex. The basic dimensioning requires the following equa-
tions to determine for the spring force F and the spring stroke SF.
248 Hartmut Janocha
D »d (5.3)
(5.4)
d : wire diameter
T : limiting torsional shear stress on the inside of the coil turns
D : average diameter of the coil turns
n : number of turns
'I'M: memory effect of the spring (shearing).
Values of up to 100N/mm2 must be inserted for the limiting torsional

stress, depending on the case of operation. The memory effect 'I'M (shearing)
ranges from 1.2% to 3.5%.
Wires as Torsion Elements. The equations for a straight wire's maximum
torsional moment T and torsional angle r.p are:
(5.5)
(5.6)
: length of the wire

T : limiting torsional shear stress at the wire circumference
'I'M: memory effect of the wire (shearing).
Table 5.6. Design data for the two-way effect in NiTi wires exposed to torsion
(heating temperature: 90°C < {}w < 150°C, cooling temperature: {}c < 45°C, tor-
sional shear stress during heating process: T w = 130 N /mm 2 , minimum reset stress
during cooling process: r c = 35 N jmm 2 , repeatable two-way effect: /'M = 3.5%)
Wire diameter Torsion during Minimum reset Memory torsion Useful work per
heating process torsion during angle per 1 em 1 em wire length
cooling process wire length
mm Ncm Ncm rad ( 0 ) Ncmrad
0.6 0.6 0.1 1.2 (67) 0.6
0.9 1.9 0.5 0.8 (45) 1.1
1.2 4.4 1.2 0.6 (33) 1.8
1.5 8.6 2.3 0.5 (27) 3.0
Table 5.6 lists the design data for torsion wires. Table 5.5 is relevant for
the applications with an electrical resistance heater.
Bending Element. Beginning with a straight cantilever, the surface strain
when activated is:
s
eM = (±) 2R (5.7)
s : thickness of the strip

R : radius of curvature of the neutral axis.
One must consider that the memory effect is basically initiated in the
surface layers of the bending element. The maximum strain is determined
by the clamping point. The classical calculation methods for cantilevers are
valid here as well.
Table 5.7. Important properties of memory alloys used in actuators
almost stepwise deformation in stability of the memory effect

the interval of 10 ... 50 K depends highly on the alloy's
high energy density (working quality
capacity per volume) high price
diversity of deformations application requires good expert
(elongation, contraction, advice by the manufacturer
deflection, torsion) low energy efficiency
the effect can be limited to limited bandwidth due to
particular regions of the element heating and cooling restrictions
simplicity of mechanism (only a degradation and fatigue
phase transformation is used)
creation of clean, silent and
spark-free working conditions
Table 5. 7 lists the advantages and disadvantages of memory alloys that

are most important for use in actuators. In comparison to bimetals (see
Sect. 5.3.5), which due to their entirely different properties exhibit a de-
flection proportional to temperature, memory alloys exhibit strong hysteresis
and a considerably higher energy density.
Application areas of SMA actuators are

drive, control and release elements in automotive engineering,
safety and lock mechanisms in the appliance industry and space travel,
activation of safety devices in the fields of fire prevention and construction
of power stations,
flap adjustment devices in the fields of heating and climate control,
release devices as overheating prevention for clutches and bearings,
gripper for handling devices (e.g. industry robots),
250 Hartmut Janocha
- miniature drives for technical model building.

The market of shape memory alloy components is strongly growing and
includes all fields of modern industry. Applications in series now range from
space technology and electronics to medical technology and precision engi-
neering as well as toys.
Flap Operation
Figure 5.8 shows one application of the two-way effect. Earlier, bimetallic
strips (see Sect. 5.3.5) were used in ventilators to open and close the air flap
ribs automatically during switching on and switch off. As the development of
ventilators went on, bimetallic strips no longer fit into the compact cases due
to design constraints. The construction volume decreased considerably due to
the application of memory bending strips. The one-sided clamped NiTi strip
is heated up to the release temperature of 90 to 105°C with a PTC resistor.
The NiTi strip bends and opens the 6 fins with a force of 2.5 N. After the
fan has been switched off, the strip cools down to the ambient temperature,
straightens up and closes the fins. This is a series application [55].
Further development lead to a solution for bigger fan flaps, that is, exter-
nal flaps which require forces of up to 25 N. This solution is based on a tension
strained, 10 mm long memory wire with a diameter of 0.8 mm.
Plate fin,
Detail Z Memory bending element
~z open
( 1
}
'-.. I/
_L PTe-
resistor
0
\
\
Fan
Reset spring
Fig. 5.8. A ventilator's flap drive with a memory bending strip
Model Making
A more recent application of the two-way effect can be found in model mak-
ing. For instance, current heated memory torsion elements can be used as
miniature drives in order to lift and lower the current brushes of electric
Non-operating
state Wire Torsional Torsion Heating power
diameter moment in angle per Heating up Continuous
continuous e leme nt phase operation
operation length
mm Ncm 0 jcm w w
0 .6 0.6 65 1.2 0.9
1.0 3.0 40 1.5 1.2
1.25 5.7 30 1.8 1.4
Fig. 5.9. Miniature drive with memory torsion elements for model making
locomotives, to shift the track switches or to move gates remote-controlled.

Figure 5.9 shows a contact heated drive with the corresponding data.
The fact that memory drives can easily be miniaturized is an advantage
in the field of model making. A gate actuator for instance, consists of a
20 mm long and 0.6 mm wide NiTiCu wire. If thinner tension wires (00.12 x
60 mm) are applied, they are heated by their own ohmic resistance. Another
advantage are the individual movements which can be achieved with this type
of drive, if the heating is well dimensioned.
Heavy Duty Actuator
Figure 5.10 shows a shape memory actuated cylinder (SMAC) which can
replace traditional actuators. The so-called heavy duty SMAC uses a CuZnAl
coil as an active element. Special about this is that the copper alloy combines
unique properties such as shape memory effect, damping and superelasticity.
A self limiting electrical heating element (PTC) is built into the axle 4 and
heats up the shape memory coil 5. This coil works against a bias spring 2 via
a spring holder 3 and causes a linear motion. By nut 1 it is possible to adjust
the transformation temperature of the shape memory coil. Characteristics of
this heavy duty SMAC are: constant load up to 400 N, displacement of 10 mm
at 400N, any voltage is possible, weight 0.5kg [55]. Advantages include the
direct linear smooth motion, not needing any transformation of a rotational
movement into a linear one, and the complete noiseless operation.
2 3 4 5
Fig. 5.10. Heavy duty SMAC
252 Hartmut Janocha

SMA have past their introduction stage and are enjoying significant growth.
Design engineers have learned to work with nonlinear materials and appro-
priate models can be used to control or predict the SMA performance. In ad-
dition to the now very successful application of SMAs in the field of medicine,
microelectromechanical systems are expected to be applied soon as a large
scale. Moreover, due to market pressure there is a need to look for new or
improved materials: stable at higher temperatures, well-controlled hysteresis,
better fatigue life, higher damping capacity. Especially with regard to the ap-
plication in actuators, three major fields of research are to be distinguished:
- The search for stable and reliable SMA that can be applied at temper-
atures far above 100°C (relevant, e.g. for applications in the automobile
industry),
- the search for entirely new types of materials exhibiting similar functional
properties as classic SMAs,
- the development of new composites with woven, knitted or braided SMAs
embedded as actuator or sensor in the composite matrix.
5.3.5 Comparison to Competing Actuator Principles

In actuators with shape memory alloy and with the below described thermo-
bimetals, mechanical energy is produced by supplying heat. Due to their
similarities, their application areas overlap, but their physical effects and
therefore too the actuator relevant properties of both material groups are
completely distinct.
Thermo-bimetals are layered composites consisting of at least two com-
ponents that are inseparably connected and which feature different heat ex-
pansion coefficients. When heated, the so-called active component expands
more than the passive one causing the thermo-bimetal to bend. According to
DIN 1715 the specific thermic bend k is characteristic for the thermic sen-
sibility of thermo-bimetals and is determined by measuring the deflection A
of a thermo-bimetal strip that is set upon two supports with a temperature
difference of D.:{} (see Fig. 5.11a). Figure 5.11b shows a strip's deflection-
temperature diagram.
Two conditions had to be fulfilled in order to produce thermo-bimetals
as a commercial scale: the research and large-scale production of iron-nickel
alloys, as well as the
competence of cladding techniques, in order to connect these metals se-
curely and for a low price. The thermo-bimetal TB 1577 (DIN 1715) is the
most economic and technically favorable one and therefore the most applied
one as well. Its passive component consists of Invar (FeNi36), its active com-
ponent consists of an iron-nickel-manganese alloy (FeNi20Mn6). In the tem-
perature range -20 ... +200°C it stands out for its high thermal sensitiv-
ity [8].
8As
k = £2/::),.{)
a b
Fig. 5.11. Deflection-temperatu re diagram of thermo-bimetals. a Measurement
set-up, b diagram
<:(
CJ)
CJ)
c d
H t
H
e
s
~ 1
2
3
4
Fig. 5.12. Thermo-bimetal element designs. a One-sided clamped strips, b coils,
c U-shaped strips, d helix, e snap discs (according to [9])
When external forces work against the free deflection, the thermo-bimetal
develops a spring tension and can store and release energy. Figure 5.12 shows
several common applications. The advantage of disc shaped elements consists
in the relatively high spring force with little displacement. In addition, they
254 Hartmut Janocha
feature a higher working capacity and save space. Usually they are cambered
mechanically. A stack of discs placed back to back features especially large
strokes.
The heat that causes the deflection in the thermo-bimetal is, for instance,
supplied at its fixation point by means of thermic conductors or by irradia-
tion or convection out of the environment. The bimetal can also be heated
electrically. In this case, the heat is supplied by an electrical heating ele-
ment close to the bimetal or by direct current conduction. Direct current is
preferred for currents > 10 A, e.g. in circuit breakers in electrical circuits,
engines and machines. If thermo-bimetals with a specific electrical resistance
< 0.6 · 10- 6 Om are required, intermediallayers or coatings made of nickel or
copper are used.
Table 5.8 lists the advantages and disadvantages of thermo-bimetals re-
garding their application in actuators.
Table 5.8. Advantages and disadvantages of thermo-bimetals in the production of

actuators
easily available (as half-finished small forces

products according to DIN 1715) only one type of deformation
well-priced (bending)
configurable by the user low energy (working capacity
linear temperature-displacement per volume)
dependency
applications up to approx.
650°C
high stability of the deformation
effect (e.g. 20 million cycles)
5.4 Piezoelectric Actuators
5.4.1 Physical Effect
Certain crystals, such as quartz, feature a physical relationship between me-

chanical force and electric charge. When the crystal lattice ions are elastically
shifted relative to one another due to an external force, an electric polariza-
tion can be detected by means of metallic electrodes on the surface. This
so-called piezoelectric effect was first scientifically explained by the brothers
Jacques and Pierre Curie in 1880 and forms the basis for piezo sensors. The
effect is reversible and is then called reciprocal or inverse piezoelectric effect.
If, for instance, an electric voltage is applied to a disc shaped piezo crystal,
the thickness of the crystal changes due to the reciprocal piezoelectric effect.
It is this property that is made use of in actuators.
Describing analytically the piezo effect by linear state equations, the elec-
tric displacement density D and the mechanical strain S are combined with
the mechanical stress T and the electrical field strength E:
D = dT+sTE (5.8)
S = sET+dtE (5.9)
In this system of equations the piezoelectric charge constant d indicates

the intensity of the piezo effect; s T is the dielectric constant for constant
T and sE is the compliance coefficient for constant E. The mentioned pa-
rameters are tensors of the first to fourth order. A simplification is possible
by using the symmetry properties of tensors. Usually, the Cartesian coordi-
nate system in Fig. 5.13a is used, with axis 3 pointing in the direction of
polarization of the piezo substance (see below) [10].
All material dependent parameters can be described by matrices (2nd or-
der tensors ) 2 , whose elements are marked with double indices. Ins T, the first
index marks the orientation of D, the second the direction of E, and corre-
spondingly T and S in sE, and E und S in d. The examples in Figs. 5.13b
and 5.13c are based on the condition that the field strength works in the
direction of the polarization. The resulting elongation in the left hand fig-
ure points as well in direction 3 (longitudinal effect), in the right hand figure
however, it works in direction 1 (transversal effect). These two characteristics
of the piezoelectric effect are quantified by means of the piezo constants d33
and d31·
b c
Fig. 5.13. Definition of the axes in piezo materials. a The digits 4, 5 and 6 indicate
the shear on the axes 1, 2 and 3, b longitudinal (d33) effect, c transversal (d31) effect
2 dt in equation (5.9) is then the transpose of matrix din (5.8).

256 Hartmut Janocha
It is common to summarize all matrix elements in so-called coupling ma-

trices, which is exemplified in Fig. 5.14 for a lead-zirconium-titanate (PZT)
ceramic. From the coefficients in the coupling matrix it is possible to deter-
mine an important parameter of piezo materials, the coupling coefficient k.
For the coupling coefficient of the longitudinal effect k 33 applies for instance
(5.10)
Since k 2 corresponds to the ratio of stored mechanical energy to absorbed

electrical energy, achieving actuators with high elongation efficiency requires
substances with a large k.
sn 8]2 8]3 0 0 0 0 0 d31

812 8n 8]3 0 0 0 0 0 d31
8]3 813 833 0 0 0 0 0 d31
0 0 0 844 0 0 0 dl5 0
0 0 0 0 844 0 dl5 0 0
0 0 0 0 0 2(8n -812) 0 0 0
0 0 0 0 dl5 0 En 0 0
0 0 0 dl5 0 0 0 Ell 0
d31 d31 d33 0 0 0 0 0 E33
Fig. 5.14. Coupling matrix for PZT ceramics (symmetry group 6 mm)
In ferroelectric materials one must add to the linear piezo effect according
to (5.8) and (5.9) an elongation that depends on the square of the electric
field strength. This elongation share is negligibly small in the traditional
materials, but it can be increased systematically in order to reach the strength
of the linear piezo effect. This so-called electrostrictive effect is independent
of the polarity of the control voltage, and the corresponding diagram S(E)
shows a very small hysteresis. The effect is long-term stable (no creep, easily
reproducible), while on the other hand, the operational range of temperature
is limited to about 30 K, and the effect is not reversible. The electrostrictive
effect is presently of less significance for use in transducers.
Materials
Piezoelectric materials can be grouped into the class of natural crystals, such
as quartz or tourmaline, into one of polymers, such as polyvinyl fluoride
(PVDF) or that of polycrystalline ceramics.
For the production of actuators , sintered ceramics are mainly used , espe-
cially PZT compounds. After sintering, the domains of a ceramic body (i.e.,
the regions consisting of crystallites of uniform dipole orientation) will show
a statistically distributed orientation, i.e. , the macroscopic body is isotropic
and has no piezoelectric properties. Only when a strong electrical DC field
is applied, the dipole regions become almost completely arranged ( "polariza-
tion"). After switching off the polarization field, this arrangement remains to
a large extent, that is , the ceramic body features a remanent polarization Pr ,
combined with a permanent elongation Sr of the body (see Fig. 5.15).
Working range
Working range
SJ
curve
-1 .5 1.5 E
kV/mm
w
-1.5 -1 -0.5 0.5 1.5 E
kV/mm
-P,
s,
Fig. 5.15. Diagram P(E) and S(E) for a typical piezoceramic for T = 0. The
actuator's operation cycle starts at point E = 0, Sr
PZT ceramics are chemically inactive and can cope with high mechanic
loading, but are also brittle and therefore difficult to process. The permissible
compressive stress is considerably higher than the tensile stress. This is why
the elements need to be pre-stressed when extensive tensile stress is applied
(see Fig. 5.19). Table 5.9 lists some values of piezoceramics used in trans-
ducers with their comparing values for quartz. While quartz, for instance,
features k33 = 0.09, ceramics achieve values of up to k 33 = 0.7. However ,
these values are unfortunately more dependent on temperature than those of
single crystal. Other disadvantages are creep and the lower long-term stability
of the material's properties.
PZT ceramics belong to the group of ferroelectric materials which feature
a hysteretic behavior shown in the diagram P(E) in Fig. 5.15 3 . The satu-
ration polarization Ps, the applied remnant electric polarization Pr and the
coercive field strength Ec are characteristic points indicated on the curve. For
3 Due to the relation P = D- coE (P: electric polarization) and D = cE, the two
diagrams P(E) and D(E) differ merely by the term coE.
258 Hartmut Janocha
Table 5.9. Values of the two PZT ceramics PXE 52 and PXE 54 (multilayer
ceramics) compared to quartz
PXE 52 PXE 54 Quartz Units
Piezoelectric du 2.31 10-12 As/N

charge d14 -0.73 10-12 As/N
constant d31 -270 10-12 As/N
d33 580 > 450 10-12 As/N
Relative r::Tdr::o 3000 4.52
permittivity r::T3/r::o 3500 3000 4.68
Compliance sf1 16 12.8 10-12 m 2/N
20 9.6 10-12 m 2/N
coefficient 8~3
Young's c~3 ~ 110 107 10 3 N/mm 2
modulus c~3 60 ... 90 107 10 3 N/mm 2
Coupling ku 0.10
factor k31 0.39
k33 0.74 > 0.6
Compressive Tp > 600 > 600 ... 4000 N/mm 2
strength
Tensile Tt ~ 80 > 80 ... 150 N/mm 2
strength
Thermal >. 1.2 1.4 W/mK
conductivity
Specific heat Cw 420 783 Ws/kgK
capacity
Curie {)c 165 220 583 oc
temperature
Density p 7.8 7.9 2.65 10 3 kg/m 3
actuator operation the diagram S(E) of the polarized ceramic, the so-called
butterfly trajectory shown in Fig. 5.15 (right hand side), is crucial. The max-
imally achievable strain is limited by the saturation and the repolarization.
Precautions must be taken in order to avoid depolarization during actuator
operation due to electrical, thermal and mechanical overload. Piezoceramics,
for instance, gradually loose their piezoelectric properties even at operating
temperatures far below the Curie temperature 4 (depending on the material
120 ... 500°C, for multilayer ceramics (see below) 80 ... 220°C). When in cer-
tain applications the operating voltage is applied reversely, it may not exceed
20% of the rated voltage, or depolarization may occur.
Apart from the piezo effect, ferroelectricity and their mechano-thermal
solid-state properties, piezoceramics exhibit even more effects. Due to shifts
4 When the Curie temperature {)c is exceeded, ferroelectric and ferromagnetic

materials loose their special property and behave as any normal dielectric or
paramagnetic material. That is why operating temperatures of piezo transducers
should not exceed {)c/2.
in temperature, polarization and field strength, charges can occur in ferro-

electrics possibly resulting in charges on the surface and an electrical field
strength inside the material. This so-called pryoelectricity can be disturbing,
especially in low-frequency applications. Connecting an ohmic resistance in
parallel to the piezo transducer to reduce the pyroelectric charge raises the
lower cut-off frequency (see Fig. 5.19b).
Piezo ceramic elements are mainly available as plates or discs with
a quadratic, circular or ring-shaped profile and a length between 0.3 and
several millimeters, with or without metal electrodes. Most are designed to
make use of the longitudinal effect (see Fig. 5.16a), which due to the high
d33 value is the strongest effect. When making use of the transversal effect
the actuator stroke depends also on the dimensions of the material, whereby
the influence of the quotient s/l on stiffness and elongation is oppositional
(see Fig. 5.16b, left hand). In tube designs the transversal effect results in
a shortening of the length s, a broadening of the thickness d and a reducing
of the radius r if d 3 1 is negative (see Fig. 5.16b, right hand).
Since the 1980s, multilayer ceramics have grown more important. The
so-called green and several tens of micrometers thick ceramic foil is cut into
pieces and then coated with an electrode paste, similar to multilayer capac-
itors. The pieces are then placed on top of each other, pressed and sintered.
They form a kind of monolithic object that is used as a finished transducer
or as a basis for producing stacks (see Fig. 5.17). Multilayer ceramics reach
the maximum permissible field strength already at a driving voltage of about
100 V ("low voltage actuators"), and achieve therefore the same elongations
as ordinary piezo ceramics do for a driving voltage in the kilovolt range.
Apart from that , piezoelectric polymers are available as foils with a thick-
ness on the order of several tens of micrometers. Such polymers are known
since 1924; but a major milestone was marked with the discovery of the strong
a b
~s = yd31 V ~~ :::::: ~d3l v
E
Cp =
a·S
-E-- E a ·l
l Cp =~ ~r:::::: :ad31 V
833 . su · s
Fig. 5.16. Inverse piezo effect in polarized ceramics. Voltage V is applied in the
direction of polarization P. a Longitudinal effect (c~: stiffness of the piezo material
for constant field strength E), b transversal effect (Z: length of the ceramic tube, r:
tube radius, d: wall thickness)
260 Hartmut Janocha
Internal External
electrodes electrode
Fig. 5.17. Basic structure of a stack comprised of multilayer piezo ceramic
piezo effect in polyvinyl fluoride (PVDF) in 1969. Piezoelectric PVDF films

are produced by mechanically drawing the material and polarizing it in order
to form a useful transducer material. The drawing techniques include extru-
sion and stretching, and while processing the film the material is subjected
to a strong electrical polarization field. Typical for PVDF piezo constants are
d33 ~ -30 pC /N and d 31 > d32 > 0; the coefficient of coupling k 3 3 is about
0.2, and the Curie temperature is near ll0°C.
Recently, polymer foils made for example of polypropylene (PP ) have
become known with enclosed, lens-shaped vapor locks with dimensions in the
micrometer scale, forming a kind of foam structure. Upon applying a high
polarization voltage, electrical charges with opposite polarity are produced
on opposing bubble walls resulting in a piezoelectric behavior. While the d 33
values of PVDF foils are clearly below the values of piezoceramics, the values
can be much higher for PP foils.
For applications in the field of microactuators (see Sect. 5.8.3), very thin
piezoelectric films are preferably implemented with the help of sputter tech-
nologies. Frequently used materials include ZnO, ZnS and AlN. These are
placed on appropriate substrates, for instance, in the form of beams and mem-
branes, whereby it is also possible to produce multilayer designs. A strong
elotropy of the expansion rate leads to a distinct orientation of the polycrys-
talline layers, so that the piezoelectric values may reach approximately the
values of polarized ceramics under optimal precipitation.
Design of Piezoelectric Transducers
The user can either build a piezo transducer from piezoceramics t hat are
available on the market , or he may benefit from the broad range of avoid-
able standardized and cased transducers. Figure 5.18 gives an idea of the
transducer variety offered by a leading producer [19].
Stack Translator (Stacked Design). T he stack translator is t he "working
horse" among piezo actuators. Furthermore, it lends itself to explaining the
construction and properties of piezoelectric actuators.
Fig. 5.18. Examples of piezo transducers (source: Physik lnstrumente, Germany)
Structure. The active part of the transducer consists, for instance, of many
0.3 to 1 mm thin ceramic discs that are mounted with metal electrodes, e.g. made
of nickel or copper, for applying the operating voltage. The discs are stacked up
in pairs of opposing polarization and glued together. Highly insulating materi-
als seal the stack against external electrical influences. In other designs - the so-
called low-voltage actuators - the multilayer ceramics described above are used.
Figure 5.19 features the electric parallel connection and the mechanical se-
ries connection of the stack. Its displacement is the sum of the single element
elongations b.l. The applied field and the achieved elongation are in line with
the polarization, that is, the piezo constant d 33 is used (longitudinal effect). The
transducer can also handle tractive forces, if prestressed with a stotted cylinder
casing as shown in Fig. 5.1 9aor with an anti-fatigue bolt , as mostly done .
Static and dynamic behavior. The static diagram S(E) in Fig. 5.15
holds for no-load operation (T = 0 in (5.9)). The addend sET in (5.9) takes
into account the loaded piezo transducer's elastic deformation. Two cases are
distinguished:
- The load is constant, e.g. weight FG . In this case, the entire diagram is
shifted due to SE = A/c~l, T = FG/A, by
(5.11)
The spring constant c~ follows from (5 .9), if E = 0 (see Fig. 5.16). As
long as the maximum permissible load is not exceeded, the original no-
load expa nsion of the piezo substance holds (see Fig. 5.20a). The load is
dependent upon the displacement, e.g. spring force FF . From (5.9) follows
with FF = - cFb.l'
b.l' CFb.l' b.l
-= - --+- (5.12)
l cp l l
262 Hartmut Janocha
Insulator
1~1
\
/\. 1*1
1---- \ (
+
Piezo disc
Electrode i dq /dt c
J.A: I
! Rn cf l "' m~j~ ls
q1=d-F1
v s,
Bourdon '-:! s;= d-v
tube
a b
Fig . 5.19. P iezoelect ric stack t ranslator. a Structure, b ele ctromechanical equiv-
alent circuit and amplitude responses of the actuator and sensor transfer behavior
in small signal operation
instantaneously the effective elongation
6-l' 6-l ( Cp )
-l- = -l- Cp + Cp (5.13)
In this case, the origin of t he diagram does not move, but t he maximally
achievable elongation is reduced by t he actor
f cp j (c p+cF) (see Fig. 5.20b ).
In t he extreme case cp __, oo (fixed clamp support of t he transducer ), t he
transducer achieves its maximal force, t he so-called clamping force or
blocking force which also follows from (5.9), if S = 0.
Equations (5.8) and (5.9) show t hat an ideal pie zoelectric transducer input
can be considered as an electric capacitor with t he capacitance C and its
6./ 6./
.--.---~ - -- - ----- -- -- -
<i
X
~
_ E
X <]
_E
<]
v
<i
a b
F ig. 5 .20. Static displacement characteristic of a stack translator. a Constant lo ad,
b l oad that depends on the displacement
output as a mechanical spring with the stiffness cp. This is illustrated in

Fig. 5.19b 5 for the d 33 transducer, but the description holds in principle for
all piezo transducers. Since C is in reality always lossy and cp always has
a mass, the amplitude response lv/ Fl (sensory operation) has an electrically
determined lower cut-off frequency fc and a mechanical eigenfrequency f 0 .
When operated as an actuator, the electrical input is a voltage, that is, C
is constantly recharged, so that fc has no effect on the amplitude response
Is/vi, as shown in Fig. 5.19b.
In the upper frequency range the transducer operates like a slightly
damped second order system with the eigenfrequency
fo = .!..._
27r v~
m:; · (5.14)
If the transducer is fixed on one side, the effectively moved mass meff can
be approximated by m/3 (m: transducer mass). If loaded by a mass M, meff
is to be equated with M + m/3. For commercial stack translators, fo lays
within the kilohertz range. As a guide, standard transducers can be operated
up to 80% of their eigenfrequency.
The response time, which is the duration tr of time until the desired
displacement has been reached, amounts to
1
tr>:::;-. (5.15)
3fo
In dynamic operation, the material can experience accelerations as high as
several thousand times the acceleration of gravity.
Due to its capacitive behavior, piezoelectric transducers exhibit a flow of
electrical energy only when the material is experiencing a change in strain.
This energy flux corresponds with a shift in charge and subsequently with
the current
i = C dv (C = constant) 6 . (5.16)
dt
For a constant supply of current, the length of time required to build up
the voltage v in the piezoceramic is therefore
v
t-- I . c- (5.17)
This length of time is directly proportional to the transducer capacitance

and inversely proportional to the control current; this is crucial in order to
design the required control electronics (see Sect. 5.4.3).
5 There are four possibilities to represent the piezo effect using state equations on
which variables are selected as dependent and which as independent (see (5.8)
and (5.9)). Correspondingly, different electromechanical circuits can be gener-
ated, compare Fig. 5.19b with Fig. l.lb.
6 This condition is only approximately fulfilled, as the values listed for C in the
data sheets (usually for small signals) are dependent on the applied control volt-
age.
264 Hartmut Janocha
For a sinusoidal driving signal with the angular frequency w, (5.16) leads
to the current amplitude
(5.18)
which must be driven by an electronic power controller in order to charge
and discharge the ceramic capacitance. inversely proportional to the number
of ceramic layers, For constant transducer height the capacitance of a stack
transducer increases nearly quadratically with the number of layers. Conse-
quently, the capacitance values of multilayer translators lay in the microfarad
range, and the charging and discharging currents are considerably higher as
described by (5.18).
With (5.18) it is also possible to estimate the upper cut-off frequency for
the case that a transducer can provide the maximal output current i = imax·
In addition, one must consider that some electrical power is transformed
into heat due to hysteresis (see Fig. 5.15) during dynamic operation. Under
extreme conditions, this may lead to thermal depolarization of the ceramics.
Table 5.10 lists characteristic values of piezoelectric stack transducers and
other types.
Table 5.10. Characteristic values of piezoelectric transducers

Stack design Laminar Disk Stack with Units
design translator displacement
amplifier
Nominal 500 090 0 0.45 50.0 0200 0 0.100 ~m

displacement
Stiffness 18 00 02000 00 0 15 0.15. 0 00.3 00.1.4 Nf~m
Eigenfrequency 6oo.50 0 0 0 13 1.1. 0 02.5 00 .2.2 kHz
Compressive 0 0 030000 0 0.450 20.0 050 0 0.50 N
strength
Tensile strength 0 003500 0 0.100 . 0020 0. 050 N
Nominal driving 150. 001500 ... 1000 . 0.1000 ... 1000 v
voltage
Capacitance 0. 0 130 ... 145 16 ... 70 ... 70 nF
Coefficient of 0.1. 0. 0.8 . 0 .0.7 1.0 .. 04.8 ... 2.0 ~m/K
thermal expansion
Laminar Translator (Laminar Design). In contrast to the stack design,

the laminar design is based on the piezo constant d 3 1 and the transversal ef-
fect. The greater the quotient s/l of the piezoelectric element (see Fig. 5.16b ),
the bigger the effect. This leads to strip shaped elements with low stiffness.
Therefore, several layers of strips are piled up, similar to the stack design,
and form a so called laminate improving the mechanical stability. Since the
transversal effect is applied, the result are flat transducers which shorten pro-
portionally to the applied voltage, as d 31 is negative. Table 5.10 lists several
characteristic values while an example is visible in the foreground of Fig. 5.18.
Bending Elements. Bending elements feature the transversal effect as well.
They can consist, for instance, of a PZT ceramic mounted onto a piece of
spring metal (unimorph). If the length of the ceramic is altered while the
length of the metal core stays the same, the element bends in order to compen-
sate the different behavior, and is therefore phenomenologically quite similar
to the thermo-bimetal.
In the disk translator design, the elements consist of circular discs with
a diameter of a few centimeters, which allow a displacement of up to several
100 ~m, as shown in the middle of Fig. 5.18. An actuator design from the
company PBT [11] uses side-by-side unimorphs in which adjacent ceramics
are on opposite sides of a metal substrate. In its simplest form, the metal
forms a U-shape with two ceramic pieces; versions with 3 or 4 prongs are
available to give higher force or greater movement.
Similarly, it is possible to connect two thin ceramic strips one of which
shortens while the other expands (bimorph). One can distinguish between
two designs: In the series bimorph, the polarization of the two piezo layers
is inversely arranged, while it is codirectional in the parallel bimorph (see
Fig. 5.21). Compared to stack translators, bending elements feature a
greater deflection,
lower stiffness,
smaller blocking force and
lower eigenfrequency.
Shear Elements. Not long ago, Physik Instrumente [19] started offering
a line of actuators based on the strong d 15 -effect (shear effect), which has only
been used in piezo sensors till now. According to the definitions in Fig. 5.13a,
the quantities E and S work along the axes 1 and 5, i.e. upon applying
a voltage the piezo element experiences a shearing motion about its axis 2.
Making use of this effect, the end surfaces of block-shaped elements without
casing (cross sections of 3 x 3 to 16 x 16mm2 ) are shifted by up to 10 ~m
with respect to each other, while the shearing loads are limited to 300N.
Depending on the element size, the resonant frequencies lie in the range
97 ... 363Hz. By stacking two such elements, a x-y positioner can be created.
Adding a third piezo ceramic element based on the d33 -effect results in a
3-axis positioning system.
Piezoelectric Transducer with Displacement Amplification. In piezo-
electric transducers with displacement amplification the achieved deflection
+ p
v,
Fig. 5.21. Parallel bimorph transducer. V2 < V1 with bias voltage V1 and control
voltage v2 (according to [10])
266 Hartmut Janocha
Metal Piezo stack
~:e
Displacement
amplifier Displacement
Piezo stack amplifier
a b c
Fig. 5.22. Mechanical displacement amplification. a Implementation with elastic
hinges, b moonie transducer, c amplified piezo actuator
is increased by constructive means. The stiffness of such a design decreases

with the square of the displacement amplification ratio and is therefore much
smaller than in the stack design.
This kind of transducer used for displacements of up to 1 mm with forces
of several tens of Newtons is achieved, for instance, with elastic joints or
hinges. T hese elastic hinges transform small angular alterations into parallel
movements free of backlash. Figure 5.22 illustrates the principle, Fig. 5.18
shows in the middle a model.
The highly ela stic material region of the displacement amplifier in Fig. 5. 22a
is locally concentrated, while the designs in Figs. 5.22b and 5.22c make use
of the global elastic behavior of metallic materials. The so-called moonie
transducer in Fig. 5.22b consists of a piezoelectric disk sandwiched between
two metal end caps. The ceramic is poled in the t hickness direction and
uses the d 31 mode. In this way the small radial displacement of the disk is
transformed into a much longer axial displacement normal to the surface of
the caps. The moonie design is very simple and its manufacture can easily be
automated. It generates moderate forces and displacements, filling the gap
between bimorph and multilayer actuators [12]. Figure 5.22c shows a related
design in which the piezo stack and subsequently the d33 -mode are used.
The advantages of these APAs (amplified piezo actuator) are their relatively
Fluid Bellows
F z,S2
Fig. 5.23. Hydrostatic displacem ent a mplification

high displacements combined with its large forces and compact size along the
active axis [13].
Figure 5.23 shows an entirely different solution. A hydraulic force-dis-
placement transformer functions according to the two-piston hydraulic prin-
ciple. Leak-free operation is achieved in the presented design through the
use of two folding bellows of different effective diameters. This special con-
structive design keeps the enclosed oil volume small increasing the stiffness
of the whole design and minimizing the amount of error due to thermal fluid
expansion [14].
W ith the above introduced principle, it is usually possible to implement
an amplification factor of up to 10. Greater values are constructively possible
but quickly lead to a worsening of the dynamic behavior of the entire system.
Inchwo rm M o t o r. The term inchworm originates from the first version
of this linear motor which featured a movement similar to that of a worm.
Figure 5.24 describes the principle: A stator which holds three piezoceramics
elements, partly covers the length of a smooth shaft which is to be positioned
axially. The two elements at the ends are hollow cylinders and are placed
with a clearance fit on the shaft. When the elements are electrically driven,
they clamp the shaft (d 33 -mode). The piezo ceramic element in the middle
also has a cylindrical shape. It is also concentric to the shaft and expands in
the axial direction when voltage is supplied (d31 -mode).
~
IB
I I
I
i
II
II
§ (4)
I Fixed bearing
I"
.I
I
I
@ I (1)
® I"
~ ~~
Ii
(5}
I 8 I
i
"I ©
III §; (2)
i
(6)
I·I
I 8 I B
I
i I
i
(3)
II
i
@ (7)
i
Displacement 6 . . .200 mm
Resolution 2 ... 4 nm
Speed (vm ax/Vmin = 5 · 105 ) 0.5 ... 2 mm/s
Axial load 0.5 ... 1.5 kg
Fig. 5. 2 4. Inchworm motor. Motion and characteristic values (according to [15])

268 Hartmut Janocha
The movement is coordinated by an electronic controller as follows. First,

ring 1 clamps the shaft (1) , cylinder 2 expa nds (2) and moves the shaft to the
left. Afterwards, ring 3 clamps the shaft as well (3) , and then ring 1 reopens
(4). The shaft is now supported by ring 3 and cylinder 2 shortens again (5).
Now, ring 1 clamps the shaft (6) and ring 3 r eopens (7). Cylinder 2 expa nds
again, and the whole procedure starts a new.
The fit between shaft and piezo ring must be very precise. The tolerance
is temperature dependent and the surfaces subject to wear. Since the contact
is frictional, an exact positioning can only be achieved within a closed control
loop making use of a position sensor (see Sect. 5.4.4). Figure 5.24 lists typical
data of standard designs of this motor type.
Ultrasonic Motor. The ultrasonic motor 's drive uses m echanical oscilla-
tions in the ultrasonic range. The stator in Fig. 5.25 consists of two affixed
circular piezo ceramic disks. Each disk is composed of eight sectors of al-
ternating polarization. The disks are displaced rela tive to one another by
1/2 sector, that is >. j 4 , and are driven with two sinusoidal alternating volt-
ages that differ by 1/ 4 p eriod 7 . Four rotating ultrasonic waves r esult a nd are
transferred onto an elastic body which is affixed to the disks. The rotor rests
on the crest of the waves. It is a disk with suita ble frictional properties and
rotates in the opposite direction of wave propaga tion. The pret ension force
between the stator and the rotor is adjusted by means of a nut above a disk
spring.
Ultrasonic motors run relatively strong and slow, a re applicable for con-
tinuous operation only under certain condition, but can be operated in cer-
tain applications without a g ear. Without current supply they feature a high
holding torque, and in case of overload they adopt the function of a slip-
ping clutch. The motors operate quietly and without jerk. Since the power
transmission is implemented by frictional engagement , the positioning must
be carried out within a closed control loop. Known applica tions include focus
Nomin a l s peed 400min- 1

Nomina l torque 2. 5 Nc m
Driv ing vo ltage 12V (D C )
Efficie ncy 45%
Di a m e ter 40mm
Heig ht 12mm
We ight 70 g
Disc Nut Base Piezo

spring plate ceramic
Fig. 5.25. Ultrasonic motor. Examples of designs and charact eristic values
7 Note the similarity to the operating mode of asynchronous motors (Sect . 3.4.1).
Resonator
a (=Stator}
b
Fig. 5.26. Elliptec motor. a Basic structure, b examples of application (by [16])
setting in reflex and video cameras. Figure 5.25 shows the data of a prototype
from the Matsushita company.
In the past years, a number of further, mostly miniaturized , drives have
been introduced. The forces produced due to piezoelectricity are transferred
to the drive element by means of friction. Figure 5.26a shows the Elliptec
motor as an example. Ultrasonic oscillations are initiated in a multilayer stack
and cause the tip of a specially formed oscillator to rotate in an elliptical path
Table 5.11. Important properties of piezoelectric transducers
large forces achievable, high characteristic values of the
stiffness ceramic are dependent on
high electromechanical efficiency temperature and age
very short response time (range piezo effect can be lost by the
of 1-1s) influence of high temperature,
negligibly low power large electrical field strengths or
consumption in static operation mechanical shock
various couplings between the strong self heating of the
field and the strain axes possible ceramic in dynamic operation
large selection and availability of high-voltage power supply
different materials necessary for t he capacitive load
(up to several microfarad)
270 Hartmut Janocha
[16]. If for instance a spring presses this tip against a movable component,
movements of the translator or the rotor result as illustrated in Fig. 5.26b.
In a certain design of this motor type, the operating frequency of 85kHz
provokes the rotor to move forward, a frequency of 95kHz provokes it to
perform a backward motion. The voltage amplitude is 6 ... 8 V, the current
consumption is 50 ... 250mA (depending on the velocity). With velocities be-
tween 0 and 300mm/s, thrusts of 0.5 ... 1N are achieved. The displacement
increment is 10 ... 20 f.!m and the length of the motor is 25 mm. The auto-
motive, computer and toy industries as well as optics are further fields of
application among others.
Table 5.1llists advantages and disadvantages of piezoelectric transducers
for use in actuator applications.
5.4.3 Electronic Power Amplifier
If the short response time of piezoelectric transducers is to be taken advan-

tage of completely, the driving electronic power amplifiers must be capable of
supplying high pulse currents at high voltages (see (5.17) and (5.18)). Gen-
erally speaking, the two options are voltage control and charge control. Both
options imply advantages and disadvantages which are described below.
Amplifier with Voltage Control.
- The correlation between electrical voltage and displacement is hysteretic
(reducible by position feed back coupling).
- It is well suited for static displacements and for displacement signals with
a large DC component.
- The amplifier load (actuator capacitance) may vary by a factor of up to
100 without needing to adapt the amplifier.
- It offers a high security against overvoltage and low voltage in the piezo-
electric actuator, as the voltage amplitude is predetermined (controlled).
This applies as well when charge is generated in the piezo actuator, e.g.
as a result of thermal or mechanical loads.
One of the actuator's connectors can be electrically grounded.
Charge produced within the piezo actuator may lead to high current due
to the low-resistance amplifier output (limitation of the current is re-
quired). Here, quick control operations are possible with all their advan-
tages and disadvantages.
Amplifier with Charge Control.

- The correlation between the electrical current and the velocity, and respec-
tively between the charge and the deflection, is almost free of hysteresis.
- It is well suited when highly dynamic, almost non-hysteretic actuator
movements are desired. For static operation, position feed back coupling
is necessary due to drift.
It is not possible to conduct any significant load variations without chang-

ing the parameters in the amplifier, since the amplifier load (actuator ca-
pacitance) and the desired charge-time behavior must be accurately tuned
(if not, overvoltage and low voltage will occur in the actuator).
Accelerations or compressive and tensile forces in the piezoelectric ceramic
can be controlled by limiting the rate of current charge during charging
and discharging.
Electrical grounding of an actuator connection is not necessarily possible.
Due to the high-resistance amplifier output, the amplifier ignores mechan-
ical actuator vibrations. Therefore, the damping time may be longer.
Comparing both control methods, one comes to the conclusion that am-
plifiers with voltage output are more universally applicable, and that they
are less critical for the user. Thus, standard amplifiers are usually based on
this type of amplifier.
Analogue and Switching Amplifier. In addition, one must distinguish
between the use of a switching amplifier or an analogue amplifier [17]. Ta-
ble 5.12 compares all the properties of both amplifier types that are important
for solid-state transducers. It is worth noting that the switching amplifier can
recover the energy stored within the electric field of the transducer capaci-
tance by means of an inductance and a free-wheeling diode. During discharge,
the energy is set free and can be fed back into the transducer to improve the
actuator's degree of efficiency. Achieving the desired displacement-time be-
havior, however, requires a convenient dimensioning of the coil, as otherwise
the entire system could become too slow.
The idea of the hybrid amplifier is based on the minimization of losses
in an analogue amplifier. This can be achieved by supplying a variable op-
erating voltage by means of a switching amplifier. The operating voltage of
the analogue final stage is added to this variable voltage level and usually
amounts to 10% of the maximum output voltage. Subsequently, only 10% of
the losses of a comparable analogue amplifier occur [18]. However, one has
to add to the analogue amplifier a switching amplifier with a bit more than
the full rated power. The hybrid amplifier is therefore always bigger than
a switching amplifier, but it provides the actuator with the signal quality of
an analogue amplifier. From an energetic point of view, and due to its size,
it is only feasible for large power ratings.
Figure 5.27 shows a block diagram of a switching amplifier for the voltage
control of piezoelectric transducers. The control signal V with the desired
voltage-time characteristic within the amplitude range of 0 ... +10 V is con-
ditioned at the input stage of the power amplifier with a voltage offset and by
low-pass filtering. A voltage sensor measures the voltage across the piezoelec-
tric transducer, which is compared to the filtered input signal. An analogue
three-point controller calculates the controller output from the difference of
target and actual value. The command signal is transferred to a driver stage
via two opto-couplers which allow floating control of the power transistors.
272 Hartmut Janocha
Table 5.12. Comparison of different types of amplifiers for piezo actuators

Criterion Analogue Analogue Switching Hybrid
Class-A Class-C amplifier amplifier
amplifier amplifier
Losses in the very high even high while very low low
power transistors in idle position driving
Feed-back of not possible not possible possible possible (for the
stored field most part)
energy
Ripples of the extremely low very low high low
output signal
Relation typical 3.14 (1r) up to 100
pulse/ continuous
current a
Dynamics in extremely high very high low high
small-signal
operationb
Load on the very low very low high low
actuatorc
Electromagnetic very good very good poor (active poor (active
compatibility disturbance) disturbance)
(EMC)
Load ranged 100 100 ~5 100
(CA/CANom)
a important for the maximum slew rate of single rectangular pulses for a construction volume
(continuous current)
b below current limits
c load from parts of actuator current not resulting from input signal (e.g. current ripples,
discontinuous charging current
d range of variation of load capacitance around the nominal value without being forced to
change the actuator's control parameters
The two MOS field effect transistors T 1 and T 2 in the half bridge operate
in switched mode, that is, either T 1 connects and the transducer capacitance
Piez~
trans-
ducer
Fig. 5.27. Block diagram of a switching amplifier for voltage control in piezoelectric
transducers
starts to load or T 2 discharges allowing a low-loss operation compared to the

analogue amplifier (see Sect. 5.5.3). Free-wheeling diodes D 1 and D 2 protect
the transistors against high induction voltage when the current supply is
switched off and conduct the current until the energy in the inductance is
completely transferred into the piezo capacitance (charging, D2) or back into
the supply (discharging, D 1).
Position Controlled Actuator Operation
Due to hysteresis, the exact value of displacement is not clearly known for
the voltage control of piezoelectric transducers. This behavior does not affect
the positioning, as long as relative movements are carried out. In any other
case, absolute positioning is required. This task requires that the voltage
be controlled within a closed loop, and subsequently requires a sensor to
measure the actual position and a controller which controls the transducer
voltage according to the difference between the target and actual positions
(see Fig. 5.28).
The actual position is determined by measured values of displacement
or strain. In order to achieve a r esolution as little as ten nanometers, strain
gages are mostly used , whereas for finer resolution , inductive or capacitive
displacement sensors are used. Some manufacturers optionally offer piezo-
electric transducers with sensors integrated into the housing.
Advantages of closed loop control are
positioning free of hysteresis,

high absolute control precision,
no drift motion,
stable positioning in spite of alternating forces,
extremely high stiffness.
As an example, Fig. 5.29 shows a single-plane 6-axis positioning system.

Compared to multi-plane (stacked) multi-axis systems, this kind of structure
Auxiliary
power
6.1
a b v
Fig. 5.28. Piezoelectric transducers within a closed control loop. a Block diagram,
b static actuator behavior
274 Hartmut Janocha
3 x PZTfor z,
rot-x and rot-y
Hinge
Fig. 5.29. Single-plane 6-axis positioning system (source: Physik Instrumente, Ger-
many)
improves the accuracy, responsiveness and straightness. The travel is 100 J.l.m
in each of the x and y directions and 10 J.l.m in the z direction with a displace-
ment resolution of < 0.3 nm. In closed-loop operation, the deviation from
linearity is usually 0.03%, and a full-range repeatability of ±2 nm is possible.
The maximal load is 2 kg [19J.
The force exerted by the piezo drive pushes a multi-flexure parallelogram
via an integrated mechanical displacement amplifier (comp. Fig. 5.22). Four
capacitive sensors measure the position of the moving platform rather than
strain in the actuator (as common with lower-precision strain gage sensors).
Parallel metrology can "see" all controlled degrees of freedom simultaneously
and compensate for runout and crosstalk of orthogonal axes.
This kind of positioning system comprises the utmost achievable with
today's technology. Application examples include scanning microscopy, pre-
cision mask and wafer alignment, surface structure analysis and micromanip-
ulation.
Mirror Drive for Laser Gyro
In laser gyros based on the Sagnac effect, the optical path length and the
direction of propagation can change slightly by means of thermal influence
on the ring resonator. In order to correct these influences, mirror drives are
required (see Fig. 5.30a). One solution to the problem is the stack transducer ,
in which at least one of its ceramic disks is one-sidedly contacted with two
Piezo
transducer
Cross section A-B
~··
a Laser b
Fig. 5.30. Mirror drive for laser gyro. a Entire system (ring resonator), b principle
of the piezoelectric transducer (according to [20])
or four equally sized metal electrodes which are insulated from one another
(see Fig. 5.30b).
Depending on the two control voltages VI und v2 it is possible to achieve
all kinds of displacements, starting from a simple inclination - in which case
VI and v2 have the same value but different algebraic signs - to pure trans-
lational movement, in which case, VI and Vz have the same value and the
same polarity. In practice, the translator displacement amounts maximally
to 50 J.tm and the angle of inclination measures about 10 arc minutes. The
radiation intensity of the laser beam is the output quantity and is constantly
observed .
The very same idea can be applied to the tube shown in Fig. 5.16b. If the
external electrode of a tube is divided into four equal segments distributed
over the circumference, applying differential drive voltages ±v on opposite
electrodes will lead to bending of one end if the other end is clamped. Such
scanner tubes that flex towards the x and y directions are used in scanning-
probe microscopes.
Diesel Injector
In 2000, the scientific community awaited with interest the implementation

of piezoelectrically driven valves in the common-rail injection system of auto-
motive diesel engines (see Fig. 5.31). This was the first large-scale application
of piezoelectric stacks. Compared to ordinary electromagnetic valves , these
piezoelectric valves, or so-called injectors, enabled higher inj ection pressures
(1500 . .. 1800bar) , shorter response times and steeper switching flanks (in
the range of 100 J.tS). This way, up to seven injections per combustion stage
276 Hartmut Janocha
Piezo
Coupler
Control valve
Fig. 5.31. Third generation piezoelectric diesel injection valve (source: Bosch, Ger-
many)
are achieved improving the fuel combustion and decreasing the emission of
toxic substances and noise.
The multilayer piezoelectric stacks integrated in the diesel injector mea-
sure for instance 30 mm in height and consist of 350 layers achieving strokes
of 40 J..!m with a control voltage of 160V. A hydraulic coupler between piezo
stack and nozzle transmits the force from the actuator to the servo valve.
This accommodates manufacturing tolerances, temperature-induced changes
in length and any wear in the system. Similar concepts are being examined
for application in petrol engines. One could also imagine this type of piezo-
electric injection element for the application in paint and powder processing
improving effectively the fineness of dosing and atomization of the material.
Japan and the USA are leading world wide in the development of new piezo-
electric materials. In the field of multilayer technology, however, Europe has
reached the leading position in the past few years. For instance, the market
offers recently 40 mm high stacks consisting of 60 to 100 J..!m thick ceramic
layers from German manufacturers. In a few years time, it will probably be
possible to implement multilayer transducers with approx. 10 J..!m thick mul-
tilayer ceramics which would provide maximal control voltages of 20V. The
following trends seem to hold for the development of materials which are
relevant for actuators:
- Substitution of expensive precious metal electrodes (Pt, AgPd) in multi-
layer ceramics by lower-priced materials (e.g. Cu).
Stronger endeavors in substituting the lead share in PZT ceramics by

other materials due to legal restrictions.
Intensified research in order to produce relaxor based single crystals (very
high strain efficiency, small hysteresis characteristic) more economically.
Researching new application possibilities of the shear effect, which stands
out for its rather large piezo constant (d 33 > llOOpC/N).
Moreover, focus is set on the improvement of the mechanical structure,
for instance, in order to transmit the heat losses more efficiently away from
the piezoelectric material during dynamic operation, or to protect the mate-
rial from the disadvantageous effects of humidity during static operation of
the transducer. The focus on improving the entire system includes also the
development of small, powerful amplifiers with a large bandwidth and a high
degree of efficiency - especially in order to drive multilayer transducers with
their great capacitance values. Here, charge control may gain in importance
in comparison with voltage control.
5.5 Magnetostrictive Actuators

When a ferromagnetic crystal is magnetized its shape changes with increasing

field strength. This phenomenon is labelled magnetostrictive effect. The most
important part of magnetostriction is the Joule effect discovered in 1842.
It is based on the fact that the magnetic domains turn in the direction of
magnetization and shift their borders. Thus the ferromagnetic body changes
its shape, whereas its volume remains unchanged. The term magnetostrictive
effect usually refers to the Joule effect, as the magnetostrictive volume change
of common giant magnetostrictive materials can be neglected. In the following
the term magnetostrictive effect refers to the Joule effect in accordance with
the prevailing linguistic usage.
Although, strictly speaking, magnetostriction is to be taken as the mag-
netic counterpart of the quadratic electrostrictive effect (see Sect. 5.4.1), its
mathematical description in practice is given by a system of equations that
formally corresponds to the linear state equations for the direct and inverse
piezoelectric effects.
B = dT+ p7H (5.19)

S = sHT+dtH (5.20)
S and T symbolize the mechanical strain and stress, H and B stand for
magnetic field strength and induction, d is the magnetostrictive or piezo-
magnetic constant, J-L T is the permeability for constant mechanical stress T,
and sH refers to the compliance coefficient for constant field strength H. In
a formal mathematical sense, the equations (5.19), (5.20) are to be handled
278 Hartmut Janocha
in the same way as those of equation system (5.8), (5.9), only that it is pos-
sible to simplify equations (5.19), (5.20) even further. This is due to the fact
that the axis of the typical bar-shaped materials is always in line with the
magnetization, which means that only the axial strain and stress need to be
considered. Subsequently, d, 1-L and scan be treated as scalar values [21].
Materials
In alloys with components of iron, nickel or cobalt the magnetostrictive effect

can generate strains in the range of 10 to 30 1-Lm/m, whereas in giant magne-
tostrictive materials of rare-earth iron alloys the effect attains values of up to
2000 1-Lm/m. Giant magnetostrictive materials which are used in underwater
sonar systems have been developed in the USA since the early 1960s. The ma-
terial which was developed there later, Terfenol-D®, has a many times higher
energy density than piezoelectric materials. Terfenol-D is the name given to
the compound Tbo.JDYo.1Fe2 [25]. The first two syllables stand for terbium
and iron (lat. ferrum), the third reminds us of the place where the material
was developed: Naval Ordinance Laboratory. The letter D indicates that the
element dysprosium is used to minimize the anisotropic energy. Terfenol-D
is not easy to produce due to impurities in the basic materials and because
of the high reactivity of the rare-earth materials. The most applied processes
are therefore the Bridgman and the float-zone methods (free stand zone melt,
FSZM).
In the Bridgman method, the completely melted material is crystallized
starting with a seed crystal. The melt is then moved relative to the temper-
ature field of a heater with a constant velocity v < 0.1 mm/s so that it can
harden from one end. In order to produce greater rod diameters in spite of the
inner thermal stresses, the company Edge Technologies modified the Bridg-
man method. The maximum magnetostrictive strain that can be achieved
with such crystals is, however, smaller than that which can be achieved with
the float-zone method. In the float-zone process the melting zone is held in
suspension by the surface stress. With respect to price and quality, these
are so far the best methods to commercially produce giant magnetostrictive
materials with good properties. One advantage of the float-zone method is
that the melt can not be contaminated by the crucible material, as the two
are not in contact. One disadvantage is the limited rod diameter due to the
surface stress.
Positive strains S of 1000 ... 2000 1-Lm/m obtained with H-fields of 50 to
200kA/m are reported for bulk materials. Table 5.13 contains some charac-
teristic quantities and values of the Terfenol-D bulk material which presently
is the most frequently used material for magnetostrictive transducers.
Table 5.13. Characteristic values of the giant magnetostrictive material Terfenol-D

[22]
Magnetostrictive constant d33 1.5 10- 8 Vs/N

Relative permeability p,'fsJ P,o 9.3
/1>~3/ P,o 4.5
Young's modulus c~3 25 ... 30 10 3 N/mm 2
c~3 50 ... 55 103 N/mm 2
Coupling factor k33 ... 0.75
Resistivity Pel 0.6 w- 6 !1m
Compressive strength Tp 700 N/mm 2
Tensile strength Tt 28 N/mm 2
Thermal conductivity .\ 1100 WjmK
Specific heat capacity Cw 300 ... 400 Ws/kgK
Curie temperature {}c 380 oc
Density p 9.25 103 kg/m 3
Phenomenologically, giant magnetostrictive materials behave analogue to

ferroelectric materials. Their characteristics J(H) 8 and S(H), like the depen-
dencies P(E) and S(E) in piezoceramics, exhibit saturation and hysteresis
(see Fig. 5.32). It is interesting that the achievable strain also depends on
the pre-stress Tp of the material: the magnetostrictive strain increases with
increasing Tp and reaches its maximum at approx. Tp = 7 ... 17 MPa be-
fore decreasing. This behavior which can also be observed in piezoceramics
- though to a less marked degree - plays an important role for the optimal
design of magnetostrictive transducers.
It has to be considered that in real ferromagnetic materials other physical
effects occur which can affect the actuator application. For example, eddy
currents are induced in the magnetostrictive material due to time-variable
magnetic excitation. The active power produced within the material is trans-
formed into heat and thus leads to a thermal strain. In order to use the
material also for applications with higher frequencies the active material can
be laminated to reduce eddy current losses.
It is also known that the magnetization of a ferromagnetic material
changes under torsion (Matteucci effect). Independently of the direction of
torsion, magnetization first increases with increasing torsion and then de-
creases again. For practical application undesired torsional stresses must not
be introduced into the giant magnetostrictive material. This has to be guar-
anteed by suitable constructive precautions when mounting the transducer.
Giant magnetostrictive materials such as Terfenol-D are predominantly
produced in the shape of rods and feature a length of up to 300 mm, and
a maximum diameter of 70 mm. Plate elements with either circular or square
8 Due to the relations J = B- p, 0 H (J: magnetic polarization) and B = p,H the

characteristic J(H) and the usual diagram of B(H) differ merely by the term
p,oH.
280 Hartmut Janocha
s
10"3m/m
Tp= 2.8 MPa
100 150
_jj_
kA/m
-150 -100 -50 50 100 150

H
-1 kA/m
Fig. 5.32. Characteristics of J(H) and S(H) for the giant magnetostrictive mate-
rial Terfenol-D under different mechanical pre-stress
profile are also available. Designs with holes and laminations for higher fre-
quency applications are produced individually according to the customer's
needs [22, 23, 24].
Similar to piezoelectric materials, it is possible to produce very thin mag-
netostrictive layers with positive or negative magnetostriction by sputtering.
References state achievable strains even in the range of 500 ... 1000 11m/m
[25].
Design of Giant Magnetostrictive Transducers
Whereas the market offers a great variety of standardized piezoelectric trans-

ducers, the product range of magnetostrictive transducers available from
a small number of manufacturers is presently quite limited. In most cases,
it is therefore the user who is responsible for the transducer design, which
however allows solutions optimized to the application.
Figure 5.33 shows a commercialized transducer series. The Terfonol rod is
pre-magnetized by permanent magnets and pre-stressed with springs allowing
a symmetrical displacement. The values of stiffness and eigenfrequency have
been calculated or estimated from other available data.
Analogous to the piezoelectric transducer (Sect. 5.4) , equations (5.19) and
(5.20) lead to the conclusion that the electrical interface of the ideal mag-
netostrictive transducer can be considered as a coil with the inductance L
and its mechanical interface as a spring with the stiffness CM (see Fig. 5.34).
The considerations described in Sect. 5.4.2, for instance, concerning the static
characteristic and the amplitude responses apply basically to magnetostric-
tive transducers as well , as the piezomagnetic constant d33 (longitudinal ef-
fect) is used .
5 U nconventional Actuators 281
Nominal displacement ± 25 .. . ± 12 5 ~m
Stiffness 40 . .. 50 Nj ~m
E igenfrequency 1.5 .. . 15 kHz
Compressive strength ... 90 kN
Tensile strength . . . 3.5 kN
Blocking fo rce 2.2 kN
Nomi na l driv ing current 5 A
DC resistance 1 .. .5 rl
Induc t a nce 1.5. .. 9 mH
Fig. 5 .33. Exam ple of a commercialized t ransducer series wit h its characteristic
values (source: Etrema, USA )
Disk
i c
:;di
R L Fi F
..lA\~
--:::::~! if>l= d-FI
~= d.i l sl m~~? l s
~I .J\
\
111
1------< \ (
Coil Permanent
a magnet b
Fig . 5.34. Giant magnetostrictive t ransducer. a Structure, b electromechanical

equivalent circuit and amplitude responses of t he actuator and sensor transfer be-
havior in small signal operation
Gener a l Tra n sdu cer D esign. On t he basis of this example (Fig. 5.3 4a) ,
t he procedure for designing a giant magnetostrictive transducer is described
below.
D esig n procedure . F igure 5.35 shows a possible approach to design
a magnetostrictive actuator. Beginning with t he tensile force Ft,max , one de-
termines the mechanical pre-stress Tp,opt and chooses then the cross-sectional
area A for the rod-shaped material and the mechanical operating stress Tp
with the help of the S(H) characteristic. W it h the knowledge of the maximal
stroke value ~lm ax, it is possible to calculate t he required length of the rod
282 Hartrnut Janocha
Start
Characteristics
-·-·-·-·-·l
oj.
c.
,....------'------. ,.
i
Bl
A. Tp.opt =7... 17 MPa
0.8
0.6
I._l. A «A<A = Ftmax
min ; Tp.opt
.
\ ;~ints · - · - · - " ]
0.4 f"l TP= ~ \ . Maximum tension force·!
. \
I. i. 1
Flmax .
50 100 15~ I\ \., ! Transducer's maximum I

L.____ __!sAl!!!.J ....---------'-....----, .·I stroke A/max I
r·---;~:~netiz~=:--·l /1J/mln=
A/max
--s;;-
~j Longit~dinal
of a COli Aoo~ 1
section I
I.
•
nor .
Selection of permanent
magnets
\
smax--~ 11
. . max1mum current 1.
'III~,"'-off
I
.f . density Jrnax
\
Ma>dm"m ouneot ' - I
..,,..... I, I
Aoo~ 1 or Jmax I L~_up_pl~_v_ol~-ge_~_o__ j
I Hm.i8--j_ jmax<!; i
""'
Pd=Pd(Jmax•Aoo~l)
I n =n(Hmax• /max)
Zmax= 1Roo~l+j2-rrf0Loo~~l
i Vo>Zmax"/max
·-·--·---'
I
Stop
Fig. 5.35. Possible procedure for designing a rnagnetostrictive transducer
l and the maximal strain Smax. The characteristic magnetostriction curve

indicates then the maximal field strength H max.
There are different ways to find out about the values required for the de-
sign, as shown in Fig. 5.35. Based on the maximum permissible coil current
Imax and with knowledge of Hmax it is possible to determine the number of
coil turns n, and subsequently the coil's inductance Lcoil and its ohmic resis-
tance Rcoil· From the value of the actuator's maximum operating frequency
fc results the electrical impedance Zmax and the required supply voltage Vo
of the power amplifier. The other way round, the impedance Zmax and subse-
quently the cut-off frequency fc can be concluded from the given value of Vo.
Mechanical mounting. When conceiving transducers with giant mag-
netostrictive bulk materials, the transducers must be subject to an optimal
pre-stress in order to achieve up to 50% higher strains. A variable setting

for compressive pre-stress facilitates the transducer 's simple adaptation to
the mechanical boundary conditions of the application. The pre-stress of the
magnetostrictive rod can be performed for example by coil springs or plate
springs.
The temperature has to be considered as a further factor of influence
on the transducer function . When the magnetostrictive rod is subject to
a change in temperature of, for example, 100 K , the thermal strain of the
rod is in the same order of magnitude as the magnetostriction. By means of
design principles known from mechanical engineering, such thermally induced
longitudinal changes can be compensated.
Magnetic circuit. The magnetic circuit consists of the magnetostrictive
rod, the coil generating the magnetic field , the flux guide, and the perma-
nent magnets. The flux guide made of highly-permeable, electrically non-
conducting material (eddy currents!) reduces the magnetic stray flux and
thus increases the mean magnetic field strength in the magnetostrictive ma-
terial and provides for a field distribution as homogeneous as possible. Us-
ing permanent magnets a pre-magnetization can be produced allowing the
transducer to be driven with bipolar current signals, thereby reducing copper
losses.
As the relative permeability of Terfenol-D is small (J.Lr < 10), i.e. this
material only has a moderate magnetic conductance, the magnetic flux guide
has a great influence on the field distribution in the Terfenol rod. If a com-
pletely closed flux guide is for some reason not possible, simply the use of
highly permeable pole pieces at both coil ends can achieve approx. 90% of
the otherwise achieved mean field strength, and by increasing the outer ra-
dius rcoii the field inhomogeneity can be reduced to below 1%. Additionally,
a closed flux guide determines a reduction of copper losses in the excitation
coil by a factor of three [26].
A closer examination also shows that three possibilities for the config-
uration of the Terfenol rod, the coil and the permanent magnet as shown
in Fig. 5.36 are sensible. In the configuration of MTC, the requirement for
a sufficiently high static magnetic field in the Terfenol rod (maximum at ap-
T M C T C M M T C
Fig. 5.36. Sensible configurations of Terfenol rod (T), field coil (C) and permanent
magnet (M) (pole pieces not shown)
284 Hartmut Janocha
proximately 80 kA/m ) leads to an outer diameter of the Terfenol hollow rod

of at least 13 ... 20 mm; this minimum dimension restricts the application of
the MTC configuration so much, that it can only be used in particular cases.
Furthermore, it is shown that, compared to the TMC configuration, the TCM
configuration has a considerably lower field inhomogeneity, less copper losses
and a good coupling of the magnetic alternating field. It can be concluded
that the use of the TCM configuration should be given priority.
Worm Motor. Kiesewetter and Huang [27] designed the worm motor which
is a kind of magnetostrictive counterpiece to the piezoelectric inchworm motor
(compare Fig. 5.24). It produces an incremental motion without requiring
comparable clamp mechanisms. It is based on the idea that a Terfenol rod of
constant volume reduces its diameter when it is lengthened. This modification
is utilized to loosen the Terfenol rod from the clamping due to the tubular
fit around the Terfenol rod.
Figure 5.37 shows the worm motor's stages of movement. By exciting the
row of coils one coil after the other , the magnetic field moves.
In a current free state (1) the Terfenol rod is entirely supported by its
surrounding fit . If current is applied to the first coil section (2), the rod's end
expands at that spot while reducing its diameter. When the field moves (in
the figure this movement is symbolized by the shifted coil) because the coil
sections are sequentially driven from the left to the right (steps 3 to 6), the
(1)
(5)
(2) (6)
(3) (7)
(4) (8)
b
Fig. 5.37. Worm motor. a Course of motion, b increment with different driving
force (according to [27])
Table 5.14. Relevant properties of magnetostrictive transducers
large forces achievable, high characteristic values depend on
stiffness* temperature*
high electromechanical efficiency* brittle, difficult to machine*
very short response time (range of - electric power required even for
!J.s)* static displacement (can be
high power density reduced by pre-magnetization
(Terfenol/PZT ceramics > 10) with permanent magnet)
no stack design required sophisticated magnetic circuit
applicable over a wide range of - hysteretic characteristic*
temperature expensive and few sources
at present primarily d33-mode is
used
* properties similar to those of piezoelectric transducers
left end becomes clamped again. After reaching the right-hand end of the rod
(7) and switching off the current supply (8), the rod has travelled to the left
by the displacement increment lll. If this procedure is repeated often enough,
the rod can gradually travel as far as the tube and coil are long. A Terfenol
rod with a diameter of 20mm and with a non-magnetized length of 80mm
achieves an increment of 28 J..Lm while reducing the diameter by 14 J..lm. The
holding force is 800 N.
In the past few years, a whole range of versions based on the worm motor
principle have been publicized. In contrast to piezo motors, magnetostrictive
motors are not reported to have been applied at large scale in industry.
Table 5.14 lists some relevant advantages and disadvantages of magne-
tostrictive transducers.
After already having introduced the dual character of piezoelectric and mag-
netostrictive transducers (see Sect. 5.5.1), it is not surprising that in both
cases, analogue and switching amplifiers may serve as the electronic control
element. Since the first principle has already been handled in Sect. 5.4.3, we
directly continue with the second method (see Fig. 5.38).
At first, the control signal is scaled in and then low-pass filtered to deprive
the signal of undesired high-frequency signal fractions. A sensor measures the
current without a voltage drop and sends out a signal which is compared to
the filtered input signal. The analog controller forms a corrective signal from
the difference of the actual and target values which serves the four driver
stages of the power element as input signal both directly as well as phase-
shifted by 180°. The MOSFETs T 1 to T 4 form a full bridge and are driven and
controlled by the driver stages. The free-wheeling diodes D1 to D4 protect the
286 Hartmut Janocha
Fig. 5.38. Block diagram of an analog transducer for voltage drive of giant mag-
netostrictive transducers
transistors against high induction voltage peaks when the current is switched
off.
Analog to the piezoelectric actuator, it is possible with free-wheeling
diodes to regain the energy which is stored in the coil's magnetic field when
the field is decaying. This sort of circuit is only suitable for switching am-
plifiers in which the high voltage peaks that occur during the switching op-
eration can be used to charge a capacitor for energy storage. When using
an analog amplifier, the magnetic field may not decay any quicker than the
maximum rate of field increase, if the output signal is not to be strongly
distorted. That is why the voltage on the coil may not be any higher than
the supply voltage, which however impedes an easy recovery of energy.
5.5.4 Comparison Between Piezoelectric

and Magnetostrictive Actuators
The corresponding formal and analytical description of the magnetostric-

tive and the piezoelectric effect suggest a user-oriented comparison between
the most important material features. Transducers based on these materials
have maximum strains on the order of one part per thousand and huge forces
into the range of kilonewtons. The transducers' static characteristics are hys-
teretic, their response times lie in the range of micro- and milliseconds, and
they feature high electromechanical efficiencies. As both material types are
brittle, they cannot easily be machine processed.
Table 5.15 compares quantitatively the characteristic values of Terfenol-D
and the PZT ceramic PXE 52.
As for their application in actuators, the following differences might be
crucial criteria for the deciding which transducer technology is most appro-
priate:
Table 5.15. Magnetostriction compared to piezoelectricity
Property Magnetostriction Piezoelectricity Units

(TERFENOL-D) (PXE 52)
Maximum strain 00.1500 00.1500 10 6 m/m
Coupling factor 00.0.75 00.0.75
Young's modulus 25 ... 35 (cH) ~ 110 (c 0 ) 103 N/mm 2
50 ... 55 (cB) 60 ... 90 (cE) 103 N/mm 2
Sound velocity 1700 (vlf) 2800 (v~) mfs
2500 (v~) 3800 (v[?) m/s
Energy density 14.0 025 1 103 Ws/m 3
Curie temperature 380 165 000300 oc
Density 9.25 7.8 103 kg/m 3
Control quantity current voltage
Field H-Field D-Field
source free source field
At 380°C the Curie temperature of the magnetostrictive material Terfenol-

D is higher than that of piezoceramics (165 to 300°C), i.e. it can be oper-
ated at higher temperatures. As opposed to piezoceramics the transducer
properties of magnetostrictive materials recover without further handling
after the temperature falls below the Curie point 9 .
The energy density in giant magnetostrictive materials is considerably
higher than in piezo materials. Consequently, the construction of powerful
actuators requires less active material. This advantage, however, is partly
compensated by the space needed for the driving coil and the magnetic
flux guide.
In piezo transducers the electrical field is built up between metal elec-
trodes which have to be applied directly to the ceramic surface (source
field); the excitation field for the magnetostrictive transducer, on the other
hand, is a rotation field and can be generated by an excitation coil or
a permanent magnet at a distance from the active material.
Whereas the current control of magnetostrictive transducers avoids high
voltages, typically higher currents than in piezo transducers can occur.
(When the magnetic field is being switched off, high inductive voltages
are prevented with the aid of free-wheeling diodes.)
Compared to magnetostrictive materials the variety of commercially avail-
able piezoceramics is much wider. In addition, different effects (longitu-
dinal, transversal) can be utilized, whereas, at present, magnetostrictive
actuators make use of almost exclusively the longitudinal mode.
Piezo transducers can retain their static displacement almost without the
supply of electrical energy, whereas magnetostrictive transducers require
9 The Curie temperature {)c of PZT ceramics lies in the range of 120 ... 370°C.
Special piezoceramics with {)c = 500°C feature a smaller constant d and a smaller
degree of expansion.
288 Hartmut Janocha
magnetization with direct current or permanent magnets for a static dis-

placement.
Today mainly piezo transducers are used, which are ready-made and com-
mercially available in a great variety. For magnetostrictive actuators, how-
ever, there are only a few suppliers who have the magnetostrictive material
on offer in a few forms and dimensions. Moreover, magnetostrictive material
is, at present , considerably more expensive than piezoceramics.
Underwater Sonar System
The application of sonar systems under water for military operations was
one of the first applications of highly magnetostrictive materials. Their sound
velocity which is low compared to piezoelectric materials brings about a low
eigenfrequency favoring their use in low-frequency sonar systems. The lower
characteristic impedance allows a better acoustic matching by reducing the
reflection factor.
Especially in low-frequency operation, appropriately designed ultrasonic
magnetostrictive transducers outplay piezoelectric transducers in terms of
achievable acoustic pressure and their geometric dimensions. Butler and
Ciosek [28] implemented a sonar underwater transducer in the form of an
octagonal ring with a diameter of approx. 25 em (see Fig. 5.39) driven by 16
Terfenol rods (eight mechanically parallel rod pairs). It produces a maximum
acoustic power output of 360 W at an eigenfrequency of 775Hz.
Piston of Magnetostrictive
transducer
Fig. 5.39. Basic structure of an underwater sonar system (according to [28])

Dynamic Vibration Absorber
The reduction of structural vibrations by applying dynamic vibration ab-

sorbers (DVAs) on frame members is an effective way to reduce the cabin
sound pressure level in turboprop aircrafts. The vibration disturbance con-
tains pronounced tonal components corresponding to the first three blade
pass frequencies (BPFs) close to 100Hz, 200Hz and 300Hz. As disturbances
of this nature can be found in many transportation applications, the DVA
design described below, for which a patent is pending, has a large application
potential.
A magnetostrictive DVA was developed combining passive and active
functionality resulting in a so-called hybrid DVA. With a resonance frequency
corresponding to the first BPF, the hybrid DVA can passively absorb the
main vibration disturbance while actively operating as an effective force gen-
erator within the operating band 50 .. . 400Hz. A low-weight, compact design
was achieved in which the coils, the magnetostrictive rod and the backings
were effectively implemented as a seismic mass to generate dynamic forces.
Fixation beam
Backing
plate
Magneto-
strictive
rod
Coil
Elastic
a suspension
10 .------r----r---.---r--...---,r-----,
dB r-~:~.~-+--r-~-+-~~~
.:
t 426~~•.,1~~=i.~.....~-..~-----~----~----~·--
.....~·· · ··
~ 0 r---r.~~~~-·r---r---r---r---r-~
~ !
.~ -2 1----1+-1--1----1-
a.E -4 \! .... Active damping
<( _
6 \1 - - Passive damping
- Original system
-10
-a t---+---+--+- I
' ----'---'--.....1....--'--J._____jL____j
1 l
50 100 150 200 250 300 Hz 400
b Frequency-
Fig. 5.40. Magnetostrictive dynamic vibration absorber (DVA). a Cross-sectional

view, b amplitude response of mass acceleration
290 Hartmut Janocha
The resulting seismic mass comprises about 90% of the total mass of 325 g.
Displacement amplification is necessary to achieve the force requirements of
15 N, 9 N and 5 N at the three BPFs, respectively, while fulfilling restrictive
demands on size and weight. The design, shown in Figure 5.40a [29], makes
use of an elastic suspension to achieve a 90° transformation and amplifica-
tion a = 6 of the displacement generated by the 40 mm long active rod. The
elastic suspension also fulfils the function of preload spring, is longitudinally
stiff and laterally soft. As a result the overall DVA stiffness is dominated by
the contribution of the magnetostrictive rod.
A test and measurement setup was implemented to verify the performance
of the hybrid DVA. An electrodynamic shaker attached to a base mass gen-
erates the disturbance force. The solid line in Fig. 5.40b shows the constant
amplitude response of the mass acceleration relative to the disturbance force
over the frequency range of interest. The amplitude response represented by
the dashed curve indicates the capability of the DVA to dampen vibrations
passively in the region of its mechanical resonance. In the example a vibra-
tion reduction of 6 dB results. Operating actively in a closed control loop, the
DVA exhibits a broadband damping capability, as illustrated by the dash-dot
curve.
Hybrid Actuator
The hybrid actuator described here basically consists of magnetostrictive

and piezoelectric transducers oscillating at their electrical resonant frequency.
While the reactive energy is exchanged between the different transducer types
the mechanical work performed and the internal losses must be compensated
by energy supplied by an external source. Compared to applications with
separately driven solid-state actuators a higher efficiency can be achieved.
The linear motor shown in Fig. 5.41a provides an example of this hybrid
principle [30]. It operates in a similar way to the inchworm motor shown in
Fig. 5.24: Between clamps 1 and 3, which are activated by piezo stacks, lies the
cylindrical coil 2. These three elements grip a fixed magnetostrictive rod and
form the moving motor component. At the beginning of the motion, clamp
1 is closed and clamp 3 is open. The coil of the magnetostrictive transducer
is excited by a current, causing the magnetostrictive rod to expand. The
fixation is shifted from clamp 1 to clamp 3 and coil 2 is turned off. The
magnetostrictive rod thereby contracts to its original length. Finally, clamp 1
is activated and clamp 3 released- one cycle has been completed. In this time
the moving component has made one step corresponding to the displacement
of the magnetostrictive transducer.
In the corresponding switching amplifier in Fig. 5.41b the resonant circuit
formed by C 1 , C 3 , and £ 2 operates at its resonant frequency. At the beginning
of an oscillation period, the piezo transducer cl is charged to the operating
voltage vc1 = V0 and the auxiliary capacitor Cz to the voltage vcz· When
transistor T 2 is activated, the current i 2 flows through transducer modelled
Fixed magneto-
strictive rod
1 em
t------1
a
Differential
amplifier2
Magnetostrictive •......... _____ _

transducer 2 ! !
Vc3'
1 I!c 3
j Piezoelectric
! transducer 3
[__ _______ ......1
Differential
amplifier 1
b
Fig. 5.41. Hybrid actuator. a Linear motor based on the inchworm principle;
b switching amplifier (by [30])
by £2 and R2. Tz short circuits the capacitor Cz as soon as the differential

amplifier 1 recognizes the voltage Vcz = 0. The piezo transducer c3 is charged
until amplifier 2 detects a zero-current condition at resistor Rsh- Then, T2 is
blocked and T P opens, so that the charge is led back from C 3 to C 1 via the
parallel coil Lp, Rp· When the input voltage to amplifier 2 and therefore the
current are again zero, T P opens and the cycle is completed.
An experimental motor featured a working frequency of 650Hz and- with
an increment of 12 J...Lm- a velocity of 7.8mm/s and a feed force of 5 to 20N.
Due to the switching amplifier, this hybrid actuator has several advantages:
The actuator is capable of adapting automatically to a change in the

resonant frequency resulting from different loads.
The speed is nearly independent of the load and can be adjusted between
zero and the maximum value corresponding to the resonant frequency of
the oscillation circuit formed by the transducers.
292 Hartmut Janocha
- A time-delayed connection of the auxiliary coil Lp, Rp reduces the dis-

placement - the speed of the motor can be reduced to zero.
- Recovery of the reactive power results in a nine-fold increase of efficiency.
The properties of highly magnetostrictive materials are strongly linked with

the physical particularities (magnetic moment) of the rare earth metals ter-
bium and dysprosium. These two metals, and others of the same chemical
group of lanthanoids are imperative for the production of highly magne-
tostrictive materials. Other than the present production methods would ac-
tually not lead to a real decrease in the costs, because the total costs arise
mainly from basic material costs. Optimistic estimations state that, a greater
demand for highly magnetostrictive materials would sink costs to less than
1/5 of the current costs.
Another effect might open up new application areas. This effect can be
observed, because the Terfenol rod's axis coincides with what is called a
112 crystal axis. Dentric twin crystals can then feature stepwise changes
in the direction of a twin's magnetization, whereas the other twin's direc-
tion of magnetization features a continuous rotation. This behavior is re-
flected in the characteristic S(H) by a stepwise change of length. It can
be systematically used to build mechanical switches which transform most
efficiently magnetic into mechanical energy. That way, a worm motor, for in-
stance, would need a much smaller current in order to build up a transducer
field.
5.6 Actuators with Electrorheological Fluids
Already at the end of the 19th century, it was known that an electric
field could influence the viscosity of certain fluids. A field strength of
0.1. .. 1 kV /mm could increase the viscosity by up to 100%, if the fluid was
electrically conductive as well as exhibited polarized behavior. This behavior
was called the electroviscous effect.
It was in the 1940s of the past century in the USA that Willis M. Winslow
spoke for the first time of a much higher viscosity increase. His fluids were
suspensions of solid and non-metal hydrophilic particles with small amounts
of adsorbed water in non-conductive oils or solvents. Upon applying an elec-
tric field, these fluids "solidify" within a few milliseconds and take on the
behavior of a plastic body. In addition, the electric field can reproducibly
adjust the fluid viscosity to any degree from easy flow to a solid condition.
This effect is called electrorheological effect.
Even to this day, the reason for the electrorheological effect has not been
resolved. The most plausible explanation says that the suspended particles
are charged and surrounded by a diffuse cloud of counterions. When the
electric field is applied, the ion clouds are distorted and start to overlay each
other thereby hindering the particles free movement. Under the microscope,
one can observe that the polarized particles arrange aliphatically forming
a sort of bridge between the electrodes which breaks apart as soon as the
field is switched off [32].
Figure 5.42 shows the electrorheological effect as a set of characteristic
curves (shear stress vs. shear rate) for ideal electrorheological fluids (ERFs).
Without an electrical field (E = 0) the ERFs behave like Newtonian (viscous)
fluids, that is, the shear stress T depends linearly on the shear rate D. The
proportionality factor is called the dynamic viscosity TJ·
When a direct or alternating field is applied, the ERF solidifies and only
starts to flow when the yield stress Ty is exceeded. The flow behavior (viscous
and plastic contributions) of such fluids can be approximated by mathemat-
ical models. A commonly used, simple model is that of a so-called Bingham
body for which holds
T=Ty(E)+TJD. (5.21)
The yield stress Ty is raised with an increase of field strength leading to an
increase of the "apparent viscosity"
T
T)a = - (5.22)
D
by several degrees of magnitude 10 .

The increase of the yield stress along with the field strength while the
shear rate remains constant can be described by the relative shear stress
t
E
E=O
Fig. 5.42. Set of characteristic curves (shear stress vs. shear rate). Ideal electrorhe-
ological fluids
10 In comparison, the electroviscous effect is characterized by a turning of the char-

acteristic curve T(D, E = 0) around the origin depending on the field strength.
294 Hartmut Janocha
T(E) - T(E = 0)
Trel = T(E = O) (5.23)
with the numerator indicating the "control range" of the ERF. The ratio
T(E)/T(E = 0) is sometimes given the same label. Figure 5.42 shows that
Trel grows toward infinity with increasing D.
Materials
Real electrorheological suspensions consist of a nonpolar base fluid of small

conductivity and a low relative permittivity, in which polarizable solid par-
ticles with a comparably high relative permittivity are dispersed. Light oils,
such as petrolates and transformer oils, di-ether, naphthenic carbons, aro-
matic hydrocarbons, paraffins and hydrocarbon compounds are possible base
fluids.
The solid material share consists often of silicic acid anhydrides, such as
silica gel, aerosil, and diatomaceous earth. Other materials which are often
used are alumosilicates, metal oxides and other dispersing materials. These
particles measure between 1 and 100 11-m and their mass share in ERF usually
reaches from 30 to 60%.
Moreover, the presence of an activator in the form of a polar fluid which
is adsorbed by the solid particles is imperative. Usually water is used, but
since it evaporates at a higher temperature, the ERF's range of application
is fairly limited. That is why materials such as glycerine and amine are used
as well.
For practical use, ERFs are expected to be sufficiently resistant against
sedimentation and aging. That is the function of the fourth ERF component,
a surface-active agent, whose composition and percentage share is determined
empirically.
Table 5.16 lists some of the characteristic values of an electrorheological
fluid. The curves shown in Figs. 5.43 to 5.45 describe the same fluid. They
are acquired with a special rotation viscometer in which a control voltage
can be applied between the rotor and the external case in order to build
up the electric field strength E as well as in which the heating of the ERF
can be controlled. So far, no measuring method has been standardized to
characterize ERFs, and no measuring devices adapted to the ERF's properties
are at hand. Subsequently, the ERF data provided by the manufacturers are
acquired under differing measuring conditions making it difficult to compare
the different fluids with each other.
Figure 5.43a shows the relation of shear stress T and the field strength Eat
a constant shear rate D. The slope S of the linear part of the curve indicates
the intensity of the electrorheological effect. The threshold field strength E 0
is another parameter of interest which indicates the point in the curve where
the rise becomes linear und which can be found at the intersection of the
Table 5.16. Technical data of an electrorheological (ER) fluid (according to [33])
25 60 90 oc
Base viscosity 'T](E = 0)
for D = 100 s- 1 780 210 180 mPas
D = 1000s- 1 1330a 280 120 mPas
Field strength dependent viscosity
coefficient (50 Hz)
for D = 100 s- 1 1020 3960 3530 %/(kV/mm)
D = 1000s- 1 60 300 540 %/(kV/mm)
Threshold field strength E8 (50 Hz) 0.55 0.48 0.53 kV/mm
Field strength dependent shear stress 800 850 640 Pa/(kV/mm)
coefficientS (50Hz, E?:: Eo)
Field strength dependent electrical 8 35 206 (nS/m)/(kV /mm)
conductivity coefficienta (50 Hz,
E 2: Eo)
a extrapolated value
b mean value from several measurements under various shear loads
straight line with the slope S and the straight line T(E = 0). A good ERF
features a small Eo value and a large S value. Eo decreases with increasing
temperature, while S increases, which means that the ERF reacts quicker and
stronger to the electric field. But if the temperature is too high, the working
range of ERFs is limited due to the decreasing disruptive strength and the
increasing electrical conductivity.
Without applied field, the shear stress T depends approximately linear
on the shear rate D, see Fig 5.43b. The upper curve has been recorded by
means of a sinusoidal offset-free progression of field strength E with time
(f = 50 Hz), the result is a yield stress which grows with the increasing
field strength. Commercial ER fluids are driven at field strengths of up to
4 kV /mm. For shear rates D :=:::! 200 ... 1000 1/s shear stresses between 2000
and 4000 Pa can be achieved. If Fig. 5.43b is compared to Fig. 5.42, one
can see that the Bingham model is merely a rough approximation of the
flow behavior of real suspensions. The characteristic curve's nonlinear region
must be considered when control concepts are drafted for ERF actuators (see
Sect. 5.6.4).
The conductivity of real ERFs lies in the range of a few nS/m. Basically,
it depends on the electric field strength and the temperature, as well as on the
frequency of the applied field and the ERF's shearing load. Figure 5.44 shows
the conductivity li against E for different temperatures. The conductivity
grows in all curves along with the increasing field strength, and linearly in
the range of higher field strengths. The increase in conductivity gets stronger
as the temperature rises. As this results in a strong increase of the current,
the range of the field strength to control the viscosity is reduced when the
296 Hartmut Janocha
700 800
t 600
500 t 700
600
500
T 400
T 400
p;300 Pa 300
200 200
100
0 0.2 0.4 0.6 0.8 1 .0 1.2 1.4 0 50 100 150 200 250 300 350 400
E D
a kV/mm ____,.. b 1/s-
Fig. 5.43. Shear behavior of a real ERF. a Dependency T(E) with D = 116s- 1 ,
b dependency T(D) for two different electric field strengths, 1J = 25°C [33]
temperature grows. This disadvantage, however, is partly compensated by the

steeper slope of the static ERF characteristic when the temperature is higher.
Simple current measurements and the fact that the conductivity depends on
the temperature can be used to control the ERF's temperature.
As the electrorheological effect is independent of the electrical field po-
larity, it can be activated with either direct or alternating voltage. When
activated with an alternating voltage, the electrorheological effect is a func-
tion of the frequency. Figure 5.45 shows the frequency responses of ERF for
different temperatures. The effect is strongest, when a direct voltage or a low
frequency alternating voltage (40 ... 50Hz) is applied. At 25°0 and a driv-
ing frequency of 150Hz the shear stress has decreased to about half of its
maximal value. For increasing temperature, the cut-off frequency increases
as well. In other words, the response time of the ERF is shorter for a higher
temperature than it is for a lower temperature.
10 9/i
(Omf1
300
250
200
150
100
50
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

E
kV/mm
Fig. 5.44. ERF's conductivity depending on the field strength (D

according to [33])
T
Pa
700~-~
600
500
400
300
200
100
OL-~~~~~~~~~~~
50 100 150 200 250 300 350 400
f
Hz
Fig. 5.45. Amplitude response of the shear stress of an ERF (according to [33])
Figure 5.46 shows a real ERF's temperature dependence Trel ( '!9) for dif-
ferent field strengths. The optimal working range lies apparently between 70
and 120°C. Whereas it is possible now to shift the typically 50 K wide useful
temperature range of presently used fluids to lower temperatures, no way has
been found so far to broaden that range significantly.
Apart from the electrorheological effect, real ERFs feature a number of ad-
ditional effects which can influence the ideal characteristic shown in Fig. 5.42
in different ways. Thixotropy, electrophoresis, and particle coagulation are
some of them, as well as sedimentation with a subsequent agglomeration
and the formation of chains during a higher shear rate. The fact that the
characteristic values of ERFs depend especially on the temperature calls for
a careful conception of actuators.
In addition to ERFs with dispersed solid particles, homogenous ERFs
have been developed on the basis of oligomer carboxylic acids and their soap
T(E)
T(E=O)
90 E=1.32kV/mm
%
70
60
50
40
30
20
10
0 20 40 60 80 100 ·c 120 '!9

Fig. 5.46. Relative shear stress T(E)/T(E = 0) as a function of temperature
(according to [34])
298 Hartmut Janocha
in mineral oil, in which the flow resistance is controllable with an electric

field. In contrast to ER suspensions, the presence of an electric field does not
result in a yield stress. Electrophoresis, particle coagulation, sedimentation
and relative abrasion do not occur as there are no suspended particles. How-
ever, the effect that can be achieved with homogenous ERFs is smaller than
with ordinary ERFs and they respond slower to alternations in the control
field.
Design of Energy 'fransducers with ER Fluids
For actuators with ERFs one can distinguish between three operating princi-
ples which may occur separately or in combination: shear mode, flow mode,
and squeeze mode. In all three cases, the ERF is arranged between a usually
flat or cylindrical electrode assembly, to which the voltage V is applied (see
Fig. 5.47).
The shear mode features electrodes that are inversely polarized and move
relative and parallel to one another. The shear stress (see (5.21)) and con-
sequently the transferred power can be controlled by means of the electric
field. The shear modus is used in clutches and brakes as well as in vibration
absorbers.
As for the flow modus , the ERF flows between the fixed electrodes ar-
ranged in parallel. The controlling field influences the flow resistanc e and
consequently the pressure drop p 2 - p 1 (qv: volume flow). The flow mode
is typically implemented in electrically controllable valves performing tasks
in electrohydraulic actuators (see Sect. 5.6.4) . Such a valve, when used as
a controllable bypass, can also be utilized for the construction of vibration
absorbers.
The third principle is the squeeze mode. Here, the distance between the
parallel electrode plates varies thereby initiating a squeeze flow. The normal
stress to be transferred can be altered by means of the electric field. Actuators
working in the squeeze mode are especially useful for damping vibrations with
high dynamic forces and small amplitudes, for instance in machine tools.
Fixed electrode
a b c
Fig. 5.47. Operating principles of an ER energy transducer. a Shear mode, b flow

mode, c squeeze mode
C = CoE
co
G = Co"'
co
Fig. 5.48. Electrical equivalent circuit for ER transducers
Table 5.17. Important properties of actuators with electrorheological fluids
flow resistance electrically unpredictable long-term

controllable over several orders behaviour
of magnitude numerous effects which disturb
short reaction time (order of the electrorheological effect
milliseconds) properties are strongly
simple mechanical designs are dependent on temperature
possible expensive and low availability
numerous application high-voltage source necessary,
possibilities (clutches, valves, driving power up to several
dampers) hundred Watt
A simplified equivalent circuit for ER transducers consists of a capacitor

connected in parallel to an ohmic resistor (see Fig. 5.48). The capacitor C
is the product of the electrode plate's geometric factor C0 j Eo and the ERF's
permittivity c. The conductance G results from the product of the geomet-
ric factor and the electric conductivity r;, of the ERF. One must take into
account that the parameters c and ""' depend on the field strength and that,
consequently, the ER transducer's electrical behavior is strongly non-linear.
Table 5.17 lists important properties of actuators with electrorheological
fluids.
Actuators with ERFs represent an ohmic-capacitive as well as non-linear load

for the power amplifier. They feature, for instance, repercussions from the
mechanical output to the electrical input when the actuator's capacitance is
reduced, e.g. because the distance between the electrodes has been increased
or the effective electrode area decreased, while the stored charge remains
constant. The resulting increase in voltage counters the electrical discharge
- especially at high temperatures - due to the ERF conductivity which can
alter by a factor of 1000 in the temperature range 20 ... 80°C. Furthermore,
short circuits and arcing may occur due to air pockets, contamination and
300 Hartmut Janocha
sedimentation in the ERF. The driving amplifier must be able to cope with
all these cases without experiencing damage.
If the high dynamics of ER actuators is to be optimized, the amplifier's
output voltage must feature slew rates of several thousands of volts per mil-
lisecond. With a one-quadrant amplifier (unipolar output voltage and unipo-
lar output current) the actuator's capacitance is discharged only via the actu-
ator conductance which is temperature dependent. To discharge an actuator
quickly and in a controlled manner, independent of the ERF conductance,
a two-quadrant amplifier is needed, since it has a current sink. Alternating
control voltages with no DC component, as they are required to avoid elec-
trophoresis, can be achieved with four-quadrant amplifiers. Such amplifiers
operate as current sources and current sinks for positive and negative output
voltages. All amplifiers mentioned above can be implemented using analogue
or switching concepts.
Regardless of whether the analogue final output stage is achieved with an
A or C class approach, there is always the problem that the transistors must
securely block several kilovolts. Some semiconductor modules feature suffi-
ciently high blocking voltages, but they are designed for currents of several
hundred amperes, and are therefore not appropriate for ERF applications due
to their size. Since connecting transistors in series to achieve the high output
voltages in the range of several kilovolts might lead to further problems [35], it
is advisable to produce the required output voltages by means of a switching
power supply. Figure 5.49 shows an example of such a two-quadrant switching
amplifier.
24VDC
One-quadrant
switching
high voltage
amplifier
Controller
---------I
I
I
I I
I I
I I
I I
fi
[> ,..J I I
-1~ vA: G
I
I
1
I
Driver I
I
I, _ _ _ _ _ _ _ _ JI
Current sink ER actuator
Fig. 5.49. Two-quadrant switching amplifier

Basically, the power amplifier consists of a one-quadrant switching am-

plifier and an analogue current sink. Assuming that the high voltage is to
be generated directly from a direct current source, flyback or (single-ended)
forward converters are useful approaches for a high voltage supply. Pulse
width modulation (PWM) with fixed frequency is the classical approach to
controlling the switching power supply and can be achieved with voltage or
current mode control. In line with such design guidelines, an amplifier has
been implemented which is fed by a 24 V direct current power supply and
which transforms input voltages of 0 ... 6 V into 0 ... 6 kV. It features a per-
manent rated output power of 100 W and a rise time of < 1 ms for applying
the maximal output voltage across a load of 1 MDIII nF [18].
Finally, it should be mentioned that it is possible- similar to piezoelectric
transducers - to actually regain the energy which is stored in the field. But
due to practical problems with the high voltage and the small amount of
energy that would be regained - as the ERF conductance is quite high - it
is not advisable to do so.
This section presents some implementations of actuators that reflect quite

well the range of application of ERFs. However, we would like to underline
here, that most of these examples are prototypes. ERF applications are at
the verge of commercialization.
Positioning Drive
When ER fluids are used as a medium in a hydraulic circuit, it is possible to

implement proportional valves without moving parts making use of the flow
mode. Valves for electrorheological fluids mainly consist of coaxial cylindrical
electrodes or of a grouping of parallel plates (electrodes) (see Fig. 5.47b). By
controlling the electric field, the flow resistance and thus the pressure above
the valve are controllable. At a sufficiently high level of the control field
no volume flow occurs, i.e. the valve closes completely. Such a valve based
on ER fluids has a compact structure, can be controlled with little power,
and operates with negligible abrasion. Conventional proportional valves with
an electrical solenoid (proportional magnet) consist of a larger number of
precisely manufactured mechanical parts. Thus, compared to valves based
on ER fluids, conventional proportional valves are relatively expensive and
prone to abrasion. The high sampling frequencies that can be achieved are
an additional advantage of proportional valves based on ER fluids.
One possible application for valves with ER fluids is the positioning drive
shown in Fig. 5.50. The structure consists basically of a hydraulic cylinder,
four ER fluid valves that are connected to form a full bridge, a four-channel
high-voltage source and a controller. The position x of the piston is measured
302 Hartmut Janocha
v,
V2 ER fluid valve 2 ~ERfluid
valve 1
High
voltage
sources
V3 ER fluid valve 3 ER fluid
v4
~~4
Hydraulic
cylinder
Fig. 5 .50. Example of a positioning drive with ER fluid valves
with a displacement sensor and the force F exerted by the piston with a force
sensor. The pressure in the hydraulic network is produced by a pump that
operates at constant speed. Through a continuous control of the pressure
above the ER fluid valves, the piston pressure and thus the working direc-
tion and velocity as well as the holding force of the piston are continuously
controllable. In terms of control t heory, this is a non-linear, multi-variable
system which demands application of fitting methods of nonlinear control.
The above concept was the basis of an ER actuating cylinder which was
developed as a highly hydrodynamic power supply for a testing machine. It
made use of the ERF's hydraulic properties, since the elect rorheological (as
well as the magnetorheological) effect is as a matter of principle not capable of
generating forces directly. For maximum displacements of ±35 mm the actu-
ating cylinder produced forces of up to 300 N. Its working frequency reached
up to 100Hz a nd for smaller force and displacement amplitudes up to 400Hz.
Similar actuating cylinders were used in other applications in combination
with a hydraulic accumulator functioning as a spring-damper-element [36].
They produced forces of up to 5 kN (operating pressure 120 bar) and strokes
of up to ±85 mm; the working frequency reached up to several Hertz.
Tactile Element
With the aid of actuators based on ER fluids, the sense of touch provided
by surfaces can be adjusted. For this purpose actuators with ER fluids are
Tactile chamber ER fluid Membrane
Electrode
fT~f{!f-Tif-J~n7'"-;=;.~t;zjLER fluid
valve
Equalizer ERfluid Membrane

chamber
Fig. 5.51. Basic design of a tactile cell array based on ER fluid
arranged in an array. The smaller the actuators are, and the closer their ar-
rangement is, the higher is their local resolution. An example is shown in
Fig. 5.51. Every actuator consists of an upper tactile chamber and a lower
equalizing chamber. Both chambers are filled withER fluid and are connected
by a valve based on ER fluid. To the outside the chambers are closed with
a diaphragm or membrane. By pressing on the diaphragm of the tactile cham-
ber, e.g. with a finger, ER fluid from the tactile chamber is displaced through
the valve to the equalizing chamber. The counter pressure that results can be
changed with the electrical voltage applied to the valve. By arranging several
tactile actuators to form a matrix, the sense of touch that occurs when touch-
ing a certain surface can be modelled or reproduced. On the basis of these
tactile actuators, Daimler-Benz AG developed a pair of forceps for minimal-
invasive surgery [37]. The feel of the forceps' grip in the body is reproduced
by a tactile array based on ER fluids in the forceps' handle.
The squeeze mode offers another possibility for the implementation of
a tactile array. A particular arrangement is enclosed by a diaphragm on the
upside, and a grounded metal foil is fixed to the underside of the diaphragm
and serves as an electrode. The bottom of the tactile device consists of several
electrodes which are arranged to form an array and can be electrified by
high-voltage independently of each other. All high-voltage electrodes have one
grounded counter electrode in common, the metal foil on the underside of the
diaphragm. The ER fluid is located between the high-voltage electrodes and
the metal foil. By electrically controlling the single high-voltage electrodes
a surface tactile profile can be adjusted. The advantage of this arrangement
consists in the fact that there is no need for partition walls which could
influence the tactile characteristics between the single actuation regions .
Shock Absorber
Electrorheological vibration and shock absorbers operate according to the

shear or flow mode (see Fig. 5.52).
Electrically controllable shock absorbers for automobiles must meet the
following demands (the velocity of absorption is vo = 1 m/s [38]):
1. The maximum achievable damping force should be Fo max 2 3 kN.
304 Hartmut Janocha
Fixed
electrode
a b
Fig. 5.52. Principle of electrorheological vibration and shock absorbers. a Shear
mode, b flow mode
2. The mm1mum damping force should be Fn min :::; 200 N (Fn min =
Fn(E = 0)). Subsequently, Fnmax/ Fnmin = k ~ 15.
3. The response time should be tr :::; 5 ms, in order to enable continuous
control.
4. The effective volume of the ERF in between the electrodes, in order to
adapt to the conventional sizes of dampers, ought to be V :::; 0.51.
5. The power which is required to control the damper should be P:::; lOOW,
while voltages of up to 5kV are required.
6. The damper must function at least within the temperature range of
-40°C :::; {) :::; +120°C. This must be considered when the three pa-
rameters yield stress Ty, dynamic viscosity 77 and electric conductivity "'
of the ERF are specified.
Practical investigations have shown that most ERFs meet the second and
the third demands , but a high k can only be achieved when the distance
between the electrodes is optimal. This point is quite easy to understand, as
smaller distances mean that the viscous friction and subsequently Fn(E = 0)
increase more than the electrically added force component.
To fulfill the first and the fourth demands, the ERF must feature a min-
imum value Tymin for the yield stress. A high Ty value is often linked to
a higher electrical conductivity of the fluid, increasing the fluid 's volumetric
power loss according to the relationship
(5.24)
A graphic that shows the dependence of the dielectric strength En , yield
stress Ty and electrical power density P /V gives a good overview of an ERF 's
applicability. Figure 5.53 shows a state diagram in which the yield stress
Ty minresulting from demands ( 1) to (4) is drawn as a horizontal line and for
a typical ERF with rJ = 100mPas amounts to approx. 3000Pa. This is based
on a damper that operates in shear mode for which [38] holds
FDmaxVD7]
Tymin = (k - 1) (5.25)
Vk
As the viscosity decreases, the required value Ty min decreases as well and
vice versa. The shaded area in Fig. 5.53 takes this dependency into account
for 50 mPas :::; rJ :::; 250 mPas.
According to demands (4) and (5) , the power density may not exceed its
maximal value of ( P /V)max = 200 W /l. This cut-off value (vertical line in
Fig. 5.53) is to be compared with the value of the power density P/V which
occurs in the ERF and which can be determined by means of the values K and
ED. For variable volume V the yield stress Tymin depends on P , as indicated
in Fig. 5.53 for three different viscosities.
The values Ty min and P jV of available ERFs can be inserted into this
diagram where they can be compared. This has been done in Fig. 5.53 for
two ERFs which had been tested for use in shock absorbers. The viscosities
lie in the range of 50 ... 250mPas, depending on the temperature, whereby
the measurements were taken for alternating and direct current. The results
show that , in all cases, Ty min is too small by at least one decade. Whereas
2·104
25"C ox
so·c 4o•c·
o
so·c
t so·cx
102 L __ _ _____j,__ _ _ j_ _,L__---L._ _,
-
10 50 100 200 500 1000
PIV
Wfl
Fig. 5.53. State diagram of characteristic ERF values (x: direct voltage, o: alter-
nating voltage; according to [38))
306 Hartmut Janocha
the (P/V) values lie for the most part within the bounded region, at higher
temperatures the electrical conductivity increased and the limiting power
density was exceeded. Consequently, the ER fluids tested in this example are
not suited for use in automobile shock absorbers.
Research and development of recent years has lead to a successive improve-

ment of the rheological, as well as the electrical properties of electrorheological
fluids. The first ERFs, for instance, appeared to be rather abrasive, whereas
in modern ERFs that disadvantage has no more bearing. Nowadays, polymer
particles are used which are nonabrasive.
Sooner or later, most modern ERFs exhibit sedimentation and agglom-
eration of particles. Another disturbing effect is electrophoresis. As these
problems become more and more under control, ERFs with more stable and
reproducible rheological and electrical characteristic values will emerge. These
improved fluids represent an important condition for their successful usage
in many prospective areas of application, such as in the automobile sector.
The shear stress of future ERFs is going to be increased and their thresh-
old field strength Eo decreased. The characteristic values' dependence on tem-
pera.ture will be reduced and the operating temperature range will be broad-
ened. As the electrical conductivity decreases, the required driving power
can be lowered as well. This would reduce the complexity of the electronics
thereby increasing the attractiveness of ER actuator applications.
In research, many different ERFs are used, but so far there are hardly
any fluids used by industry. This is why the development focuses on robust
ERFs. The range of properties must be adapted to the specific requirements
of the application. That is, if there are different applications there must also
be different ERFs to suit them. The design of ER actuators must more than
ever be approached as with mechatronic systems. The ER transducer, power
amplifier or controller and the ER fluid must be seen as the components of
one system and must therefore be co-ordinated.
5. 7 Actuators with Magnetorheological Fluids

Magnetorheological (MR) fluids are suspensions consisting of magnetizable

particles in a low-permeability base fluid. Under the influence of a magnetic
field the suspensions change in their magnetic, electric, thermal, acoustic and
optical properties and in particular in their rheological behavior. The flow re-
sistance increases considerably with a growing magnetic flux density B. The
suspended particles form magnetic dipoles which align with the magnetic
field B and interact to form chains and agglomerates along the field lines.
These chains can be mechanically loaded and lead to the creation of a yield
stress and an increase in flow resistance. The first to describe this magne-
torheological effect was Jacob Rabinow at the end of the 1940s. The process
is fully variable and reversible, i.e. after turning off the magnetic field the
particles return to their original statistical distribution. The switching times
for the structural changes are within a range of a few milliseconds.
Since ER and MR fluids feature a similar rheological behavior, many
of the statements about ER fluids made in Sect. 5.6 correspond as well to
MR fluids; the electrical field strength E is merely to be substituted by the
magnetic flux density B. In the magnetic control field MR suspensions form
a yield stress Ty that is determined by the chain-shaped configuration of
the suspended particles. Without a control field, ideal MR fluids behave like
Newtonian fluids, i.e. when they are sheared, the shear stress T shows a linear
dependence on the shear rateD (compare Fig. 5.42). Under the influence of
a control field they have approximately the properties of a Bingham body,
i.e. the yield stress increases as the strength of the control field grows. While
flowing, a plateau of identical flow velocity evolves. The width of this plug
increases with the growing strength of the control field. When the plug fills
out the whole channel width, the volume flow ceases.
5. 7.2 Technical Implementation
Materials
For a long time the initially low reproducibility of the MR fluid properties, the
particles' strong sedimentation and the high abrasion 11 impeded a technical
development of magnetorheological applications. However, in the past few
years, the USA and Germany in particular, have developed commercial MR
suspensions which are characterized by a strong MR effect as well as by
a considerably improved sedimentation stability and a low abrasion.
MR fluids consist of the suspended magnetizable particles, a base fluid
and a stabilizer. The particles usually have a diameter ranging from 1 to
10 J..Lm and a density of approx. 7 gj cm3 . They often consist of car bony iron
alloys and can account for up to 75% of the suspension's weight. Mostly sili-
con and mineral oils are used as a low-permeability base fluid. The base fluid
is required to have a low viscosity and to be stable within a wide range of
temperature. The third component, the stabilizer, shall prevent the parti-
cles from sedimenting and coagulating. The density of the whole suspension
ranges from approx. 3 to 4 g/ cm3 .
At room temperature the base viscosity of common suspensions amounts
to several hundreds of mPa s. An increase in shear stress can be attained by
increasing the particles' share in volume or by using bigger particles while
11 The abrasive behavior of MR fluids is used to polish optical lenses: MagnetoRhe-

ological Finishing.
308 Hartmut Janocha
maintaining the particles' same share in volume. However, both methods

lead to a drastic increase in base viscosity and thus to a reduction of the
MR effect. Table 5.18 lists some of the characteristic values of two typical
magnetorheological fluids.
Table 5.18. Characteristic values of two MR fluids (19 = 25°C)
132LD (Lord) 252 (DEA) Units

Fluid base synthetic oil mineral oil
Operating -40 ... + 130 -25 ... + 120 oc
temperature
Density 3.1 3.8 gjcm 3
Solids concentration 81 75 %weight
Base viscosity D = lOs- 1 : 440 D=100s- 1 : 28 mPas
D=50s- 1 : 330 D=1000s- 1 : 647
Shear stress per flux unknown D=1000s- 1 : 28 mPa/mT
density (115 mT :S: B :S: 285 mT)
Particle size 1. .. 20 4 J.!.m
MR suspensions are superparamagnetic, i.e. ideally soft magnetic. They

have a saturation magnetization and a permeability that depends on the
magnetic field strength. Their magnetic behavior is free of hysteresis. The
transmittable shear stress T increases with growing magnetic flux density B
according to an exponential function. At medium levels of magnetic induction
a linear increase occurs (see Fig. 5.54a) [39]. When the magnetic flux density
approaches the saturation magnetization (up to 0.8 T depending on the type
of MR fluid), the increase of the shear stress declines more and more. The
shear stress which can be transmitted with an MR fluid is limited by the
saturation magnetization and can maximally reach up to 100 kPa.
The relationships in Fig. 5.54 have been measured in a modified rotation
viscometer with a coaxial double gap. Its rotor, that is, the shear surfaces
consist of non-magnetic material. The same MRF was examined in a mea-
surement set up where the rotor and the iron core were made of magnetic
steel. Figure 5.55 shows that the shear stresses were clearly higher than the
ones achieved with the non-magnetic rotor, because the particles stick much
better to the magnetic shear surface.
MRF DEA 252 exemplifies in Fig. 5.56 that, in contrast to ERF (comp.
Fig. 5.46) the shear stress decreases as the temperature grows. Without
a magnetic field, the shear stress increases much more with decreasing tem-
perature than with increasing magnetic field. Consequently, the MR effect
diminishes away as the temperature falls. In contrast to ERF, the MRF's
conductivity at room temperature as well as its dependency on the temper-
ature are so low that they are irrelevant in most applications.
8000
7000
v Pa
~a= 285 mT
t 50~~ v
./ t 6000
5000 r--
/ 1851 mT
I
T 4000 T 4000 /
3000 ./ 3000
115 mT
r-
/ 2000 l.-
2000
1000
___.. v 1000 o mT
1
0 00
0 50 100 150 mT 250 200 400 600 800 1/s 1200
a a- b o-
Fig. 5.54. Measured shear behavior of the MRF DEA 252 using shear surfaces
made of non-magnetic steel. a Relationship r(B) for D = 1000 s- 1 , b relationship
r(D) for different flux densities and 1J = 23°C (from [39])
25000 25000
a= 404 mT
Pa /
Pa 345mT
t15000
T /
/ t15000 .. II' ....... 27~mf
10000 /
T
10000
.r 206mT
/ ,/"'
I I
vv
5000 5000
64mT
0 f-- 0
0 100 200 300 mT 450 0 2000 4000 1/s 8000
a a- b o-
Fig. 5.55. Measured shear behavior of the MRF DEA 252 using shear surface~
made of magnetic steel. a Relationship r(B) forD= 1000 s- 1 , b relationship r(D)
for different flux densities and 1J = 28°C (from [39])
Apart from MR fluids based on micrometer particles there are also MR

fluids with particles that have a diameter of only 30 nm. The particles of
these nano-sized MR fluids consist of ferrite and account for approx. 60%
7000
Pa
&.
.... ...__, ...__, I I
a=185mT
t 5000
T-+-,
.......... ....__
~
T 4000
115 mT
3000
2000 '----t-+
..........
1000 ........... bmt- r--
0
-1 0 0 10 20 30 40 50 60 oc 80
v-
Fig. 5.56. Relationship r(iJ) of MRF DEA 252 for different flux densities and
D = 1000 s- 1 (from [39])
310 Hartmut Janocha
of the suspension's weight. The density of nano-sized MR fluids amounts to

approx. 2 g/ cm3 . Due to the very small size of its particles the nano-sized MR
fluid is extremely stable with respect to sedimentation, and its abrasion can
be neglected in practice. On the other hand, the maximally transmittable
shear stress of approx. 7 kPa is considerably lower than it would be with MR
suspensions based on carbon iron particles.
MRFs must not be confused with ferrofluids . Whereas in MRFs t he in-
teraction of particles is necessary to control the rheological properties, the
interaction of particles in ferrofluids is not wanted. So, in order to avoid that
interaction, the particle's size of a few nanometers and their share in volume
of approx. 10% may not be exceeded. Due to the particle's size and t he long
distances between them, their thermal energy is greater than the interaction
energy, preventing the fluid from forming chains. Ferrofluids are used as fluid
seals on rotating shafts as well as for substance transport and fixation , e.g.
for chemotherapy or hyperthermotherapy of tumors.
Design of Energy Transducers Based on MR Fluids
Energy transducers using MR fluids are based on three principles, which can
occur alone or in combination: shear mode, flow mode and squeeze mode.
In MR fluid energy transducers t he fluid is located in a gap in the magnetic
flux guide. The field lines of the magnetic control field run perpendicular to
the fluid's flow direction. The working modes of MR fluid energy transducers
are shown in Fig. 5.57 [39]. In the shear mode the plates move parallel to
one another, and the transferred force or the torque can be controlled by
the magnetic field. Possible applications of the shea r mode include clutches
and brakes as well as electrically controllable dampers. In the flow mode the
MR fluid flows through a gap in the magnetic flux guide. The control field
influences the flow resistance of the MR fluid and thus t he pressure drop
across such a valve. MR fluid valves do not need moving mechanical parts.
They are mainly used in shock absorbers and vibration dampers as well as in
a b c
Fig. 5.57. Working principle of MR energy transducers. a Shear mode, b flow

mode, c squeeze mode
T = __l:_ (1 + Rcoi!)
Rcoii Red
Fig. 5.58. Electrical equivalent circuit for MR transducers
hydraulic systems with the MR fluid as the hydraulic medium. In the squeeze
mode the distance between two parallel plates changes causing a squeeze flow.
In this mode it is possible to achieve relatively high forces, and it is especially
suitable for damping vibrations with low amplitudes up to one millimeter and
high dynamic forces (e.g. in machine tools).
The equivalent network in Fig. 5.58 describes the electrical behavior of
MR energy transducers with L symbolizing the current dependent inductance
of the magnetic circuit, including the MRF, whereas Rcoil stands for the coil's
ohmic resistance, and Red for the influence of eddy currents during alternating
field operation. The time constant T must be small to achieve a good dynamic
behavior of the transducer, in other words, when the transducer is designed,
L should be as small and Red as big as possible (e.g. by means of a laminated
flux guide).
General Transducer Design. A magnetorheological energy transducer
consists of the mechanical mounting and the magnetic circuit (MR fluid,
coil, flux guide and possibly permanent magnets). The first step to designing
a MR fluid energy transducer consists in the requirements of the mechan-
ical application, e.g. the minimum and maximum forces and the required
construction volume. Other requisites result from the type of application,
e.g. maximum electrical power requirement, response time, and from the sur-
rounding conditions, e.g. the range of operating temperature. The designing
process is briefly described below.
Magnetic layout. The MR energy transducer's essential drive parameter
is the flux density B which reacts in the magnetorheological fluid. Since the
maximum shear stress transmitted by MR fluid is limited by its saturation
magnetization, the fluid should be driven close to magnetic saturation.
It has to be considered that B depends on the respective MR fluid. If
different MR fluids are to be compared, the magnetic circuit has to be cal-
culated for each of them and B has to be determined in dependence of the
control current of the coil. In order to compare different types of MR flu-
ids regarding their specific applications, it is necessary in most of the cases
to perform a rheological and magnetic conception of the energy transducer
for every type of MR fluid, because the rheological properties of the MR
fluid determine the geometry of the MR fluid's working space in the energy
transducer, which has an essential influence on the magnetic circuit.
312 Hartmut Janocha
In order to achieve a compact design of the magnetic circuit, the flux guide
has to operate below its magnetic saturation. Then the magnetic resistance
of the circuit is almost exclusively determined by the working spaces of the
MR fluid, as MR fluids have a relative permeability of f.Lr < 10. This value
is of the same order of magnitude as that of Terfenol; thus, statements and
results of Sect. 5.5.2 can be adopted by analogy. It has to be considered that
the magnetic field lines in the working space of the MR fluid should run
perpendicularly to the shear direction.
With the aid of permanent magnets the operating point of the MR fluid
energy transducer can be adjusted and at the same time the electrical driving
power and the construction volume of the coil can be reduced considerably.
The cross-sectional surface area of the permanent magnets should be similar
to that of the MR fluid gap. The response time of a MR fluid energy trans-
ducer is considerably influenced by the inductance L of the magnetic circuit
(see Fig. 5.58). The impact of L can be reduced, for instance, by conceiving
the coil for high current densities. However, this leads to a strong increase in
the electrical driving power.
Rheological conceptualization. The rheological concept used in MR
actuators depends highly on each particular application calling for differen-
tiation on a case-by-case basis.
Clutches and brakes. Torque transmission with MR clutches can be im-
plemented with both flow mode and shear mode. The shear mode applies
the shear stress between the drive surface and the output surface, whereas
during the flow mode a driving force is applied to the fluid. This force can
be produced, for instance, by a rotating toothed disc putting low pressure on
the tooth's front side and excess pressure on the backside. Since the ratio of
transmissible torque and idle torque during shear mode is much better, this
is the exclusive way of power transmission in practice.
There are two basic structures to distinguish: The cylindrical design con-
sists of the MRF between two concentric rings for the drive and output.
The disc design consists of two parallel and circular discs with the MRF in
between. Compared to the cylindrical clutch the disc clutch has a smaller
construction volume and a lower weight which can be further reduced by
designing the clutch for lower maximum inductions in the MR fluid. Thus,
the electrical control power of the disc clutch decreases as well. However, it
has to be considered that at high revolution rates of the clutch and at low
magnetic fields the suspended particles move outwards through the MR fluid
gaps creating a depletion area of particles in the gaps.
If both shear surfaces which form the borders of the working space of the
MR fluid are made of highly permeable material then a stronger adhesion of
the particle structures to the shear surfaces in the magnetic field is guaran-
teed. Thus the attainable shear stresses are considerably higher than in the
case of non-magnetic shear surfaces. By a surface roughness which amounts
to two or three times the radius of the suspended particles, a further increase
of the forces transmitted by the MR fluid can be achieved at lower magnetic

flux densities.
Dampers for high vibration amplitudes. Dampers with MR fluids for high
vibration amplitudes are mainly built up making use of the flow mode, inte-
grating the valve with MR fluid in the piston, or in the case of double tube
dampers in the bypass. When designing valves with MR fluid, particularly
at high piston velocities, it has to be taken into account that the time spent
by the suspended particles in the magnetic field of the MR fluid valve must
be sufficiently long. If this time is considerably higher than the MR fluid's
response time to changes in the magnetic field, the damping force increases
considerably when the piston velocity increases. If, however, the time MR
fluid particles spend in the valve is lower than the response time due to the
small valve length or the high flow velocity, then the damping force is almost
independent of the damper's piston velocity. However, with the aid of the
valve length also a determined damper behavior can be adjusted depending
on the piston velocity. For example, at first, the damping force increases with
growing speed at low piston velocities, then at high velocities it changes into
an almost constant range [40].
Dampers for low vibration amplitudes. Dampers for low vibration ampli-
tudes are preferably built up in the squeeze mode in particular at high force
amplitudes. Due to low vibration amplitudes dampers in the squeeze mode
frequently display a viscoelastic behavior. The elastic vibration components
determine among other things a decrease of the energy dissipated by the
damper. A reduction of the elastic vibration behavior can be implemented
by an electrical control adapted to the damper movement. A possible appli-
cation field of squeeze flow dampers are machine tools in which the resonance
ratios that occur with high force amplitude can be reduced.
Table 5.19 lists advantages and disadvantages of actuators with magne-
torheological fluids.
Analogue and switching amplifiers (compare Sect. 5.4.3) are not always nec-
essarily the perfect solution, as the following example will illustrate. In the
case of a MR engine mounting (see Fig. 5.62) it turned out necessary to sup-
ply the load inductivity of 13 mH with a unipolar, rectangular current-time
characteristic. The current's peak value was to be 13 A, while the rise time
and fall time were to be <::::: 1 ms with a pulse width of 10 ms and a repetition
frequency of 25Hz. A feasibility study showed that, under these conditions,
an analogue amplifier suffers an average power dissipation of approx. 440 W.
Controlling the current amplitude of a switching amplifier with energy recov-
ery would require a switching frequency of > 10kHz; in connection with the
relatively high operating voltage of approx. 170 V this would lead to consider-
able losses in the switching transistors and higher eddy current losses within
314 Hartmut Janocha
Table 5.19. Important properties of actuators with magnetorheological fluids
flow resistance controllable over resistance of ageing not secured*
several orders of magnitude* careful conception of magnetic
short response time (range of circuit required (coil, flux guide)
ms)* material compatibility
simple mechanical constructions unclarified
possible* no data sheets available for
multiple applications (clutches, comparisons
valves, dampers)* no commercial measurement
control requires moderate technique available
currents and voltages
insensitive to soiling
magnetic resistance largely
independent of temperature
* properties similar to those of electrorheological fluids
the magnetic circuit, that is, the switching amplifier's important advantage,
the effect of its high degree of efficiency, would be considerably limited.
The circuit shown in Fig. 5.59a offers better results making use of electric
resonance and maintaining the current's set value with only a few switching
operations at a noticeably reduced supply voltage. current-time and voltage-
time diagram. The capacitor Co is charged with the voltage V, so that its
stored energy ~Co V 2 is equal to the energy ~ LI1 stored in the load inductiv-
ity L after the energy transmission is completed. The transistors T 1 and T 2
are connected in between the two points of time t = 0 and t = 1 rns. Energy
is transmitted from C 0 to the inductive load L increasing the current up to
its set value (see Fig. 5.59b). The transistor T 2 stays open from t = 1 rns
until t = 10rns, and T 1 is used to switch continuously between the two states
"recharge" from the comparatively low voltage supply and "hold" over T 2
.!t:. VA
A v
,/A 170
13
12 v ~v ------------····· t
10! ms
)i
VA \j -170
a b
Fig. 5.59. Power electronics for MR transducer. a Block diagram, b current-time
and voltage-time diagram
and D 1 . During that time, the step-up converter is inactive and by-passed
with D 3 . At the point of time t = 10 ms all transistors are opened. The load's
energy retransmits autonomously to the capacitor Co charging it ideally to
its original potential.
This electronics requires a power of approx. 20 W during engine mount op-
eration. Compared to an equivalent analogue amplifier it saves 95% of energy,
which is an important advantage for its application in automobiles. Finally,
it is worth noting that this switching concept is suitable for other induc-
tive loads with a small ohmic impedance component (e.g. magnetostrictive
actuators).
5.7.4 Comparison Between ER and MR Fluid Actuators
Magnetorheological fluids show rheological properties similar to those of ER

fluids. The MR effect is also reversible, and the MR fluids' response times
to changes in the magnetic field also lie within a few milliseconds. The main
differences between actuators based on ER fluids and those based on MR
fluids result from their different interactions with electric and magnetic fields
[37]:
Actuators with ER fluids are voltage-driven and they represent ohmic-
capacitive loads for high-power electronics; they require high voltage
(some kilovolts) at a low current flow (few milliamperes). On the other
hand, actuators with MR fluids are current-driven; voltages below 10 V
and currents below 2 A can be sufficient in order to control the field coils
under quasi-static conditions. The maximum driving power of comparable
ER and MR fluid energy transducers is similar at temperatures less than
50°C and is on the order of a few tens of watts.
- Already a small number of foreign bodies or included air bubbles can
cause a deterioration of the ER fluid's characteristics or result in electric
breakthrough in the ER fluid. In case of an electric breakthrough combus-
tion residues develop on the electrodes, which contaminate the ER fluid.
The breakdowns are very demanding as far as the short-circuit strength
of the control electronics is concerned. MR fluids are less sensitive to
contamination. However, both fluids are hygroscopic and thus should be
protected from moisture as that lends to particle coagulation.
The base viscosity at room temperature of common ER fluids with a value
below 100mPas is lower than that of most MR fluids. Hereby ER fluids
have considerably lower flow losses in hydraulic circuits. However, the
shear stresses transmitted with an MR fluid are higher by one order of
magnitude than those in ER fluids. With ER fluid shear stresses it is
possible to transmit up to approx. 10 kPa, whereas in the case of MR
fluids with micrometer sized particles shear stresses of 100 kPa have been
measured. That is why MR energy transducers require a lower volume of
active fluid.
316 Hartmut Janocha
- The electrical conductivity of ER fluids increases with rising temperature

according to an exponential function. Thus an upper temperature limit
in a number of applications (e.g. clutches, brakes, dampers operating in
shear mode) results not from the chemical stability of the ER fluid, but
from the high electrical driving power demand of the ER fluid energy
transducer. A comparable behavior does not exist in MR fluids. Common
MR fluids can be used at temperatures of 150°C and higher.
- With the aid of permanent magnets, the operating point of MR fluid
actuators (a particular flow resistance) can be achieved without electrical
power. Driving around this operating point can be performed with the
help of electrical coil magnets which strengthen or weaken the field of
the permanent magnet. A comparable possibility to adjust the operating
point without electrical power does not exist in actuators with ER fluids.
- If necessary, energy transducers with ER suspensions must be driven with
purely alternating electric fields, as otherwise the effect of electrophoresis
becomes noticeable: When driving an ER suspension with direct electric
fields over a long period of time, the particles migrate to the electrode
with a higher or lower electric potential, according to the surface charge.
Thus, an area with a very high concentration of particles and a depletion
area develop, which clearly decreases the ER effect.
- In ER fluids smooth particles can be suspended which have a considerably
lower density than the iron or ferrite particles used in MR fluids. In this
way a reduction of abrasion with a low tendency to sediment and a low
base viscosity can be achieved in ER suspensions. So called homogeneous
ER fluids in which an increase of the dynamic viscosity occurs in the
electric field show neither sedimentation nor abrasion. However, their ER
effect is lower.
Generally, both MR and ER fluids can be employed in actuators with
controllable fluids. The decision in favor of one of the fluids is determined by
the requirements of the single application. The boundary conditions of the
actuator application determine the type of fluid to be adopted.
5. 7.5 Application Examples

Brake
In 1995 Lord Corp. offered a disc brake as a first commercial product on the
basis of a magnetorheological fluid. Compared to eddy current brakes the
MR fluid brake produces a high torque already at low rotation rates, and
this torque can be controlled by the control current being over a large region
independent of the rotation rates [41]. The structure and the braking torque
M produced in dependence of the control current I are shown in Fig. 5.60.
The two gaps between the case and the brake disc are filled with MR fluid,
and the magnetic field generated by the coil runs perpendicularly to the shear
direction through the MR fluid.
7
Nm ~
t: I
T 3 /
2
7
/
~
,_
0 200 400 600 mA 1000
a b
Fig. 5.60. MR fluid brake. a Basic construction, b d ep endence of the braking
torque Ton the control current I in the range of revolution rates 200 min - 1 < n <
lOOOmin - 1
The brake has an outer diameter of d = 92 mm and requires a maximum

control power of approx. P = 10 W at a maximum control current of I = 1 A .
Mechanical power of max. Pw = 700 W is transformed into heat. By applying
little constructive changes this brake can be transformed into a disc clutch.
A possible application field of clutches with MR fluids in automobiles include
the power transfer to supplementary units of the engine, e.g. alternator or
ventilator. Thus it can be achieved that- independently of the engine's rev-
olutions - the alternator is always operated in the range of revolution rates
with the highest efficiency.
Clutch
The disadvantage of the cylindrical clutch design is that during the segrega-
tion of the MRF, the inner torque transmission surface is depleted of particles
due to the gravitation and the centrifugal force which affects the function the
clutch. The shear mode MR clutch with a V-shaped transmission gap, shown
in Fig. 5.61a circumvents this disadvantage [42] . Compared to the disc design,
the V-design , with equal torque, requires considerably less space. In the figure
can be seen that two magnetic circuits are arranged in parallel to produce
the required torque(> 10Nm). To prevent the MRF from leaking when not
operating (during operation, the MRF is maintained due to the centrifugal
force inside the clutch gaps), a special seal is used which consists of a circular
permanent magnet with an axial direction of magnetization and iron core.
Figure 5.61b shows some characteristics of this clutch.
Quickly switching safety clutches and clutches with controllable torque
are the potential areas of application for clutches with MRF as a torque
318 Hartmut Janocha
12 8=470mT
12 8=470mT
Nm Nm
10 393 r-
10 393
t: t: ----
287
287
219
219
146
~4 146
~ 74
~2 74
Np loa<;l operatioQ No load ooera i n
0 0
350 650 950 1250 1550 m1n 1 20 30 40 50 60 70 'C 80
b Rotational speed - Temperature -
Fig. 5.61. Slanted gap clutch. a Construction, b torque vs. speed at {) = 34°C
(left hand), torque vs. temperature for lOOOmin - 1 (right hand), each for different
flux densities
transmission medium. Compared to ordinary clutches, they have the advan-

tage of featuring a short response time, low level of abrasion and the fact
that the transmitting torque is exactly reproducible. Table 5.20 compares
the properties of ER and MR clutches.
Table 5.20. Comparison of clutches withER and MR fluids
Clutches based on ER Clutches based on MR

fluids fluids
Electrical control voltage-driven current-driven
Control signal V > lkV; I some rnA at V < 30V;
{) < 50°C; AC 0.5 A < I < 3 A; DC
Control power some lOW at{)< 50°C some lOW
Temperature exponential increase of little decrease of torque at
dependence control power at rising rising temperature, MR
temperature effect remains constant
Range of temperature -20°C < 'I'J < +80°C -20°C < 'I'J < +130°C
Maximum torque ~ lONm ~ lOONm
Reaction to very sensitive insensitive
contamination
Motor Mount
Automobile engine mounts must fulfil a wide range of tasks which require
contradicting properties: On the one hand, the (quasi- )static loads resulting
from the strong torque must be supported and since the engine is fairly lim-
ited in terms of motion within the engine compartment of the automobile stiff
mounts are required. On the other hand the strong oscillations of the engine
can only be decoupled with soft mounts. Additionally, one must consider the
vibrations provoked by the road surface which also require a certain spring
stiffness of the engine mounts. These contradicting goals can be reached by
means of load bearing elements which are able to adapt their characteristic
properties to the specific operating state of the engine (switchable system).
The best results can be achieved by directly influencing the transmission of
vibrations between the chassis and the engine.
Figure 5.62 shows an electrically controllable engine mount prototype
based on MR fluids which was built and tested within the scope of a funded
collaboration [56]. A conventional hydraulic mount is implemented as a base
mount to bear the static and quasi-static load. Since the MR actuator does
not have any direct spring properties it is connected in series with a spring
element. When the actuator is switched on, both elements are parallel to the
base mount; when it is switched off, only the stiffness of the base mount is
effective. The actuator is comprised of a shearing element which rotates in
the MR fluid activated by a lever. With this method it is easy to achieve the
required precise motion of the submerged element and an adjustable mechan-
MR actuator
* Passive motor mount (base mount)

Fig. 5.62. Principle of a semi-active engine mount based on MR fluids
320 Hartmut Janocha
ical translation (length of the lever between activation point and the axis of
rotation). The compartment filled with MR fluid is sealed with a conventional
radial seal. All parts of the magnetic circuit including the submerged element
are laminated to avoid eddy currents.
The MR mount's eigenfrequency is 150Hz, i.e. higher than the excitation
frequency of the engine. It is designed for dynamic forces of ±500 N and
displacements of ±2 mm. The maximum magnetic flux within the MR fluid
gap is 600 mT and is controlled via a special, highly efficient power amplifier
(see Sect. 5.7.3).
In the past few years the application of MR fluids has grown in comparison to
that of ER fluids. This might be the consequence of the better performance
of MRFs in many applications which have also lead to a whole range of
mass products with MR actuators. One is the electrically controllable shock
absorber in the automobile industry.
Many of the development trends of ERFs (compare Sect. 5.6.5) apply to
MRFs as well. Here, the situation is clearer, because the composition of MR
fluids and their properties are less complex than with ERFs. For instance,
the MRF's conductivity has no bearing due to its negligibly small value.
Redispersion and the resistance to sedimentation are important subjects
that remain to be worked on. The price for a higher resistance to sedimenta-
tion in MR fluids is usually a high base viscosity. A solution to that conflict
is one of the most important goals in development.
5.8 Microactuators
5.8.1 Principle
Microactuators are based on three-dimensional mechanical structures with

very small dimensions which are produced with the help of lithographic pro-
cedures and anisotropic etching techniques. For an actuator-like displacement
the most different principles of force generation are used, such as the bimetal
effect, piezo effect, shape memory effect and electrostatic forces. Characteris-
tic for microactuators in a more narrow sense is the fact that the mechanism
of force generation is integrated monolithically; in a broader sense, however,
also microstructures with a not monolithically integrated force generation are
numbered among microactuators.
With the downscaling of the actuator dimensions from the macro-range
to the micro-range of course also the actuator forces decrease. When down-
scaling a mechanical structure by a factor of m, electromagnetic forces -
depending on the technical requirements - are theoretically reduced by as
much as m 4 . Pneumatic, hydraulic or biological (muscle) forces and even
surface tensions behave in a more favorable way because they decrease only
by m to m 2 . Electrostatic forces are of particular importance in the micro
range: They decrease with m 2 , but as the breakdown electric field strength in
insulators increases with decreasing dimensions (Paschen effect), the electric
field strength may be increased by m- 0 · 5 which leads to a reduction of the
forces by only m.
5.8.2 Realization
For the manufacturing of extremely small mechanical components with mov-

able structural parts there are in essence the classic micromachining based
on single-crystal silicon, surface micromachining for the production of poly-
crystalline silicon structures, and the LIGA technology; furthermore there
are precision engineering techniques such as electrical discharge machining
(EDM) and micro injection molding.
The well-known silicon technology is the obvious manufacturing proce-
dure as silicon can be produced at highest purity in sufficient quantities
at low costs, and many procedures can be taken over from microelectron-
ics. Furthermore, single-crystal silicon has also good mechanical properties;
e.g. Hooke's law is valid at strains ten times higher than in other materials.
Apart from that silicon easily enables the monolithic integration of electronic
components, e.g. control circuits or monitoring sensors (compare Sect. 1.4.1 ),
whereby several hundreds of components per wafer and process can be manu-
factured simultaneously (batch processing). Existing surface micro machining
technologies on the basis of polycrystalline silicon in combination with a sil-
icon oxide sacrificial layer technology (SLT) have been qualified for up to 5
levels.
The LIGA technology, a combination of lithography, electroforming and
replication processes can be used for the cost effective fabrication of high
precision microstructures from a large number of polymers, ceramics and
metals or metallic alloys. The generation of the primary structures in thick
resist layers by deep X-ray lithography can be replaced in certain cases by
other microstructuring techniques, such as laser or particle beam based pro-
cesses. The polymeric primary structures are transformed into a metallic
secondary structure by electrochemical deposition processes, which may be
used as molding or embossing tools in various types of replication processes.
However, many "classical" materials for chemical technology can only be
microstructured by extending conventional precision engineering techniques
into the micro regime. In particular micro-milling, using diamond tools and
high speed spindles, as well as spark erosion based techniques have been
used successfully for the fabrication of microfluidic components. In many
cases technically and economically optimized solutions can only be realized
by a combination of different fabrication techniques. Very promising results
have been achieved, for example, by the use of high precision LIGA-fabricated
Cu electrodes in a micro die-sinking erosion process.
322 Hartmut Janocha
With microactuators the possibility of producing great actuating forces

or displacements is limited. They are mentioned here nevertheless, because
actuators with small measurements are advantageous in terms of velocity,
precision and the space they require to control miniaturized systems and for
handling small parts (see Table 5.21). Fields, such as micro-assembly, life
sciences, medical technology and communication devices can benefit from
these advantages.
Table 5.21. Important properties of microactuators
Advantages Dis ad vantages

batch production: compact, high investment costs for
precise, very dependable, production equipment
inexpensive only small displacements and
can be integrated into forces are possible
microsensors and
microelectronics
voltages and currents compatible
with microelectronics
Microdrives
High precision drives with millimeter size have become a key element for
minimal-invasive diagnostics and surgery as well as for miniaturized scanners
in datacom or positioning drives for example micro robots and hand held
consumer devices.
Electromagnetic Micromotors. The micromotor shown in Fig. 5.63 con-
sists of a diametrically magnetized rare-earth magnet rotor surrounded by
a coil system, which generates a rotating magnetic field.
A soft magnetic tube simultaneously acts as housing and return yoke
for the magnetic flux. In order to minimize the outer diameter of the mo-
tor a yoke material combining a high saturation magnetization and a very
low coercivity has been chosen. This motor design leads to comparatively
small heat losses and a lower power consumption. With its diameter of
1.9 mm and a length of 5.5 mm the motor generates a considerable torque
of up to 7.5~-LNm and rotational speeds of more than lOO,OOOrpm. All me-
chanical parts are fabricated by conventional precision engineering methods
[43, 44].
In addition to the longitudinal type a fiat electromagnetic motor with the
dimensions of a penny was developed (see Fig. 5.64). The rotor contains 8
Fig. 5.63. Construction and components of a 1.9 mm motor [43]
SGL
• diameter: 12.8 mm
• height : 1.4 mm (exci . shaft)
- torque : 80 ~Nm C!l 200 mA
- speed: up to 60,000 rpm
Fig. 5.64. Construction and realization of the penny motor [43]
permanent magnets and a soft magnetic return yoke which encloses the flat
coil array in the stator. The magnetic simulation shows how the ball bearing
can be pre-stressed by an additional soft magnetic ring in the stator in order
to avoid yaw motion of the rotor axle. Furthermore, a stray magnetic field
hardly occurs. Due to the very small radii extremely high accelerations and
rotation velocities up to 60,000 rpm in continuous operation are reached. The
motor torque scales almost directly to the volume and ranges up to 100 11Nm
[43, 46]. For series fabrication of the motor, precision engineering technologies
are combined with micro techniques which are advantageous for coil arrays,
electrical interconnection and encoders. Recently, a family of penny-shaped
micromotors has been developed [45].
324 Hartmut Janocha
Reduction Gears. In order to adapt the high speed of micro motors to

the industrial requirements of 100 to 5000 rpm and in order to increase the
output torque considerably, miniaturized reduction gears with similar sizes
and shapes are needed.
A gear realized for the 1.9 mm motor is a three-stage gear and contains
plastic parts fabricated of POM using the LIGA technique, see Fig. 5.65.
For mass production a through-going batch process has been developed
where hundreds of gear components are molded simultaneously in a wafer-
like magazine. Each stage is assembled separately. The gear may deliver an
output torque of up to 300 !l-Nm in short-time and 150 !l-Nm in continuous
operation [43, 44].
The penny motor (Fig. 5.64) with its flat shape is perfectly suited to
miniaturized rotary positioning systems, if a flat gear with high ratio and zero
backlash is integrated. Therefore, the harmonic drive with dynamic spline as
output proves to be an excellent solution but extremely difficult to minia-
turize. By realizing the wave generator by a simple planetary gear with two
flexible planets the overall diameter could be kept as small as 8 mm and a gear
ratio of 505 - in one stage - could be achieved. High aspect LIGA technique
turned out to be the pre-requisite to realize the prototype shown in Fig. 5.66,
which is illustrated impressively especially by the flex spline. resp. Gear size:
8x1 mm2. Gear ratio: 505 [1]. Its inner and outer toothing have moduli of
34 !l-ID and 35 !l-ID respectively. While the tooth width is as high as 1 mm the
ring thickness amounts to only 40 !l-ID. That means an aspect ratio of 25 is
achieved. Continuous output torques of 34mNm can be achieved by penny
motors with a gear.
The examples described show that miniaturization of reduction gears
down to millimeter size can be achieved by applying the LIGA technique. The
Fig. 5.65. Stage of the planetary gear. Gear modulus 55 J.lm. Gear stage ratio: 3.6
[43]
Fig. 5.66. Micro harmonic drive. Gear moduli: 34 ~-tm and 35 ~-tm resp. Gear size:
.08 x 1 mm 2 . Gear ratio: 505 (43]
development may be executed step-wise by using high aspect ratio lithogra-

phy and electroplating steps for prototypes and, eventually, injection molding
for series production.
Positioning and Gripper Systems
Miniaturized linear positioning systems have turned out to be a basic unit

for a wide range of applications. Examples are robots and grippers for the
production and handling of semiconductors and optical fibres, means for posi-
tioning, adjusting, focussing or zooming of lenses or fibre-optics in datacom,
telecom or consumer products and the manipulation of cells and tools in
medicine and biology.
Common modes for individual actuation are electrostatic, piezoelectric,
electromagnetic and electrodynamic. Pneumatic actuation is also an inter-
esting approach. When large displacements are needed , a thermal actuation
is favored.
Positioning Systems. A micro positioning device has been developed that
obtains a closed-loop control including an incremental optical encoder which
directly gauges the slider motion [43]. By applying modules like the micro-
motor (Fig. 5.63) and planetary gear (Fig. 5.65) and ultra-precise fabrication
of steel slider and housing by wire EDM , the performance was considerably
increased: 40 x 6 x 4 mm 3 overall size, 20 mm travel range with a velocity of
up to 400 mm/ s, a force of up to 0. 75 N and a positioning accuracy of ±5 11-m.
Using the electrostatic mode for comb-drive actuation much progress has
been made especially with respect to the displacement and linearity. Fig-
ure 5.67 shows the SEM photograph of a sensor-actuator system for applica-
tion in atomic force microscopy (AFM). Within this system an electrostatic
326 Hartmut Janocha
Fig. 5.67. SEM detail view of an electrostatic silicon actuator with integrated tip
for lateral AFM measurement [47]
actuator, a capacitive sensor and a silicon tip are included. The actuator is de-
signed to generate a large lateral displacement of about 20 J..Lm. While measur-
ing the sensor's tip is actively placed within this range and enables nanometer
resolution. Further applications of the low-force electrostatic principle are fo-
cusing on positioning/displacement of opto-mechanical components such as
switches, attenuators and filters or even microlenses.
Gripper Systems. In the last decade a few microgrippers have been built.
Mechanical gripping, gripping by vacuum, gripping with adhesive substances,
magnetism and electrostatics are the main principles which are applied. The
only tool which is applicable almost everywhere are mechanical grippers. For
those grippers various physical effects are used to generate the movement and
the gripping force such as electromagnetism, electro-statics, piezoelectric and
shape memory effect.
Fig. 5.68. Scheme of a microgripper made of silicon or glass [47]

Salim developed a gripper (Fig. 5.68) which is actuated by monomor-

phic piezoelectric ceramic. The gripper was made of silicon or photosensitive
glass. The necessary hinges are solid-state hinges. The piezoelectric actua-
tor is firmly connected to the gripper. When voltage is applied , the actuator
contracts (d31 -effect) and thereby moves the gripper. The displacement of
the piezoactuator is amplified by the solid-state hinges and the arms of the
gripper up to 100 times. The total system achieves a gripping force of 20mN
and a gripper-jaw displacement of 100 IJ.m.
Figure 5.69 shows a scheme of a SMA microgripper in open and closed
condition. The gripper is mounted on a substrate in a pre-strained condition.
Thus, a deformation is created in the beam structures, which can be con-
trolled by electrical heating. By selective heating of actuation unit I above
the phase transformation temperature, the folded beams recover their un-
deflected memory shape, which leads to a linear motion of the link. Conse-
quently, the circular beams are deformed and the gripping jaws are closed.
This condition can be reset by selective heating of the circular beams (actua-
tion unit II with inherent displacement amplification). The link between the
actuation units is designed to provide sufficient thermal isolation.
This microgripper has been fabricated by laser cutting of a cold-rolled
sheet of NiTi. For a prototype with 2 x 3.9 x 0.1 mm 3 size the gripping jaws
allow a maximum stroke of 180 11-m and a maximum gripping force of 17 mN.
A typical response time is 32 ms for 22 mW electrical driving power. The
cooling times are considerably longer, in the order of 300 ms. However, due
to the used antagonism the response times for opening and closing are only
determined by the heating performance of the corresponding actuation unit.
The cooling performance determines the maximum frequency of actuation
cycles.
Open condition Closed condition
Gripping
jaw
Actuation
} unit II
~Bond pads
(circuit 2)
Actuation
unit I
Fig. 5.69. Operation principle of a shape-memory alloy (SMA) microgripper [48]

328 Hartmut Janocha
Microoptics
Optical applications are a major sphere of actuator devices. For instance mi-
crolenses are positioned and focused , mirrors are actively shaped to a certain
curvature and micromachined gratings with electrically controllable trans-
missions have been investigated in the past years.
Microscanners. Microscanners have been reported frequently in connection
with a great variety of applications ranging from barcode scanners to scanning
image projection with laser. Arrays of micromirrors, developed up to several
millions of cells, are used for optical signal processing and image projection
or they act as phase grating, respectively.
The principal structure of microscanners is a tilting, movably suspended
mirror plate, attached to a substrate (see Fig. 5.70). On the one hand , the
rotational stiffness of the hinges and the mass moment of inertia define the
resonant frequency and often the maximum working frequency likewise. On
the other hand, the stiffness of the suspension determines the required driving
force for the application dependent maximum tilt of the mirror. Driving near
resonance increases the tilt by the Q factor up to several hundred.
Torsion beams are primarily used for elastic suspension in case of one-
dimensional scanners. Double axis scanning is achieved by special elastic
hinge structures or by orthogonal cascading of elastic suspension and drive.
The hinge material depends on the fabrication technology. Bulksilicon, poly-
silicon, various metals (Ni, Al) and polyimid are used.
Electromagnetic, piezoelectric, electrostatic, and thermal driving have
been reported. Most of the mirrors are electrostatically driven, whereby max-
imum working frequencies of 250kHz are possible. Voltages of up to several
Fig. 5. 70. SEM view of an electrostatically driven scanning mirror [47]

hundred volt are necessary depending on mirror size, maximum tilt and reso-
nance frequency. Specially arranged electrostatic comb drives, coaxial cascad-
ing of torsion-beam/mirror-systems and magnetic drives are new approaches
to reduce the operation voltage.
Microchopper. Detector arrays in spectrometer systems for the near-
infrared range must be operated in modulated light to achieve sufficiently high
signal-to-noise ratios. To realize a handheld infrared spectrometer, a minia-
turized light modulator has been developed [49].
The setup of the microchopper with lateral dimensions of 3.0 x 3.2 mm 2
is shown in Fig. 5.71. The LIGA structure of electroplated permalloy with
a height of 280 J..!m is supported by an aluminium oxide ceramic substrate. The
stop can oscillate at one end of two rectangular parallel springs with a width
of 10.8 J..!m. They are fixed to the substrate at the other end. The stop also
constitutes the anchor of the magnetic actuator. It moves between the two
poles of the chopper core with an air gap of a few micrometers. These two
parts are structured monolithically on the substrate. The resonant frequency
is approx. 1 kHz.
Coil of enameled
Coil core
Fixing block
Oscillating
stop
Fig. 5. 71. Setup of a microchopper [49]
The coil is inserted during assembly and fixed with catch springs. It con-
sists of a permalloy core with several hundred windings of enamelled copper
15 J..!m in diameter. When a current flows in the coil, the stop is attracted by
the core independently of the sense of current. For a periodic current, this
leads to an oscillation of the stop at twice the frequency of the current. This
is a major advantage for use in the microspectrometer system.
Microfl uidics
Microfluidic components and systems are currently gaining industrial signif-

icance in various technical fields, medical technology being among the most
prominent. However, the need for miniaturized fluid handling systems is be-
coming even more pronounced in chemistry and biotechnology.
330 Hartmut Janocha
Microvalves. There is a growing demand for miniaturized valves in a wide

range of applications: Valves in cold gas jet systems applied for space applica-
tions need to be lightweight and should deliver an ultra-short response time.
Valves for printing devices or for chemical micro reactors must be small and
cheap because a high number of valves has to be arranged in large arrays.
Ruby ball
--
Fig. 5.72. Principle of a micro ball valve [43]
A promising approach which solves the mentioned requirements is shown

in Fig. 5.72. In closed state, a ruby ball, which is commercially available down
to a diameter of 300 J.Lm, is pushed against the valve seat by the fluid pressure.
To open the valve, the ball is flung into the valve cavity by hitting the valve
seat with a piezo stack. When piezo actuation is finished, gas pressure of up
to 2000 hPa pushes the ball back in less than 1 ms. The valve seats are made
of silicon by advanced silicon etching in a wafer process. For the machining
of the housings a combination of micro die-sinking and wire EDM is used
which allows a series production in steel.
NiTi
microdevice
Valve
seat
Pins
Fig. 5.73. Cross-section of the microvalve. A, Band C denote valve ports [50]
The valve in Fig. 5.73 has been designed for operation in a normally open
condition where the pressure acts as a biasing force against the actuation force
generated by the SMA device. A polymethylmethacrylate (PMMA) substrate
of 10 x 20 mm 2 lateral size has been processed by mechanical micromachining
to establish a pressure chamber with inlet and outlet ports A and B, respec-
tively. The microvalve has been designed for a maximum pressure difference
of 1200 hPa. At this pressure limit, a stroke of 70 IJ.m, a gas flow of 1600 seem
and a work output of 35 IJ.m have been observed. The typical response time
for closing the valve is 0.5 s, the cooling time is 2 s.
For improvement of the work output, the SMA microdevice has been
stress-optimized. The basic idea of stress optimization is to design the shape
of active SMA parts in such a way that spatially homogeneous stress profiles
are obtained for a given load pattern. Thus, a maximum volume fraction of
SMA material is used for actuation, and fatigue failure due to local stress
maxima is minimized.
Development efforts on the microvalve involving reduction of the SMA
foil thickness, insertion of a metallic cooling ring above the microactuator
and a thermal insulator between the microactuator and the membrane have
reduced the above response times to 25 ms and 90 ms, respectively, at a heat-
ing power of 85 m W. Further potential to improve the valve dynamics can be
tapped by using SM alloys with higher transition temperatures.
Micropumps. Micropumps are attractive means for drug dosing in medicine,

reagent delivery on chemical applications or adhesive dispensing in the semi-
conductor industry. Furthermore, new application fields arise for dosing of
lubricants for machine tool bearings or with pump arrays for high-throughput
screening in chemical analysis.
The pump principle shown in Fig. 5. 74 is realized by silicon machining in
many places. considerably high due to the large lateral size of 10 to 20 mm.
Cost-efficient series production has been achieved by switching to polymers
like PC, PEEK or COC. The top and bottom parts, which are injection
joined, and membrane foils are connected hermetically by laser welding. Fi-
nally, the bimorph piezo plate is glued onto the outer foil. The self-filling
membrane pump is commercially available [51]. Maximal pressure and max-
imal flow rate are up to 2000hPa and 0.4ml/min for water, as well as up to
500hPa, down to -350hPa (vacuum) and up to 3.5ml/min for air.
The self-priming and bubble-tolerant operation mode was achieved by
enlarging the compression ratio - the stroke volume of the actuation divided
by the dead volume of the pump chamber- of the micropump. A compression
ratio of about 1:9 could be attained, which allows pumping of liquids and
gases as well as of liquid-gas mixtures [43, 52].
Figure 5.75 demonstrates mechanisms to transport highly viscous fluids
even with solid particles by the roll off of two gear wheels. The pump with
the oval wheels provides a larger pump room which increases flow rate and
332 Hartmut Janocha
Piezo
bimorph
Actuator
deflected
Outlet
valve
Fig. 5.74. Principle of the micro membrane pump [43]
Pump cavity
Fig. 5. 75. Principles for micro gear pumps with an external gear pair [43]
allows a higher size of solid particles or a more careful delivery of sensitive

fluids like blood.
A prototype of the circular pump has been realized by using LIGA wheels
with a modulus of 38 ~-tm and a brass housing which was conventionally milled.
Tests of the gear pump show a linear relationship between flow rate and rotary
speed of the wheels, while pressure hardly has an influence on the flow rate.
For a mixture of glycerine and water a maximum flow rate of 170 ~-tl/min
could be achieved at 2250 rpm up to a pressure of 1000 hPa.
Fluidic Microsystems. The dramatically increasing number of syntheses
and analyses in current biotechnological and pharmaceutical development
processes can only be performed by highly automated processing systems
using extremely small reagent quantities. Miniaturization is an essential pre-
requisite for the necessary enhancement of the p erformance of reaction and
analysis systems by integration and parallelization of many procedural steps.
Successful construction of complex biotechnological processing systems,
e.g. for mass screening or combinatorial synthesis methods, requires the use
Injector fluid __
Carrier fluid
+
Manifold
(glass, silicon)
Fig. 5. 76. Cross-view of a microliquid handling system [54]
of reaction chambers, mixing and heat exchanger systems, separation devices ,

pumps, valves and many types of sensors with extremely small dimensions
[53]. Interlinking of microfluidic components in a micro-channel system pro-
duces complex microliquid handling systems (see Fig. 5.76).
The layout of fluidic microsystems orientates itself by the advanced de-
signs of hybrid technology. This orientation is useful, as the building-up and
bonding techniques of hybrid technology are particularly suitable just for the
interfacing of components strongly differing in material and geometry. When
having a look at the large number of microfluidic components which have
been developed so far, the significance of the fluidic microsystem technol-
ogy as a technology which is able to interlink these components within the
microsystem efficiently, robustly and reliably becomes comprehensible.
For the past few years, there has been a trend to mix different micro t echnolo-
gies (e.g. Si and LIGA technology) opening up new ways of combining differ-
ent materials and possibilities of implementing complex 3D geometries. The
area of microfluidics tends to use increasingly synthetic components which
can be produced for more favorable prices than silicon components.
The usage of microfluidics in pharmacy and life sciences is a growing
market for microactuators. The big pharmaceutic companies focus on finding
lab-on-a-chip solutions for very different applications in the area of diagnosis.
Microsystems for analyzing the blood sugar of diabetics, for instance, have
most successfully been brought to the market, and microsystems are also
applied to dose insulin and other drugs.
In microoptics there is a growing trend to use microactuators which cover
a far range of possible applications reaching from the bar code scanner with
a movable mirror complex to mirror systems for display technology (Laser
TV, digital movie theater , etc.). The increased demand for optical communi-
cation technology will entail a greater demand for all kinds of optical switches,
and therefore at least some of the microactuator solutions will be used.
In most cases the control electronics is still much more voluminous than
the actuator element itself. Thus, a further development target consists in
334 Hartmut Janocha
integrating the electronics in the actuator housing when possible. In this

context first experiences have led to an unexpected result: The development
of a special ASIC often turns out to be the second choice. By a slight redesign
of the mechanical and fluidic construction, standard electronics can be applied
which reduces costs considerably.
Another impediment to application is due to the fact that users who
could employ microsystems with great advantages for small and medium-
sized series, rightly shy away from the high development costs. Application
fields like these then also remain closed to the producers of microsystems. For
this reason, industries and research institutes deal with the idea of realizing
small quantities to be manufactured as well as huge production lots in a unit-
assembly system.
References
1. Janocha, H. (Ed.): Adaptronics and Smart Structures. Springer-Verlag Berlin
Heidelberg New York 1999
2. Kempe, W.: Elektrochemischer Aktor. Technischer Bericht Nr. F2A-93-010,
Daimler-Benz AG Forschung und Technik
3. Briisewitz, M., Ciecierski, A.: Linear drives by electrochemical means- elec-
trochemical actuators. In: Mechatronics 10 (2000), pp 531-544
4. Zupan, M.; Ashby, M.F.; Fleck, N.A.: Actuator Classification and Selection-
The Development of a Database. In: Advanced Engineering Materials 2002, 4,
No. 12, pp 933-940
5. www.med-e-cell.com
6. www.btt-thermostate.de
7. Humbeeck, J. Van: Shape Memory Alloys: A Material and a Technology. In:
Advanced Engineering Materials, 2001, 3, No. 11, pp 837-850 (60 references)
8. www.vacuumschmelze.de
9. Schneider, F.E.: Thermobimetalle. Grafenau/Wiirtt.: expert-Verlag, 1982
10. Koch, J.: Piezoxide (PXE)- Eigenschaften und Anwendungen. Heidelberg: Dr.
Alfred Hiithig Verlag, 1988
11. www.pitgoadi.co.uk
12. Uchino, K.: Ferroelectric Devices. Marcel Dekker, New York Basel, 2000
13. www.cedrat.com
14. Wirkungsgradoptimierte Piezoantriebe fur hochdynamische Anwendungen in
der Flugzeughydraulik (PiezoServ): Joint project funded by the German Fed-
eral Ministry of Education and Research (BMBF), 16SV563, www.tib.uni-
hannover.de
15. www.exfo.com
16. www.elliptec.com
17. Janocha, H., Stiebel, Ch., Wiirtz, Th.: Power Amplifiers for Piezoelectric Ac-
tuators. In: Responsive Systems for Active Vibration Control, A. Preumont
(editor), Kluwer Academic Publishers, 2002, pp 379-391
18. www.dass.de
19. www.pi.ws
20. Rodloff, R.; Jungbluth, W.: Piezoceramic Servo-drive for Producing Transla-
tional Motion, Especially for Appliction to Ring Laser Mirrors. United States
Patent No 4 639 630, 1987
21. Wohlfahrt, E.P. (Hrsg.): Ferromagnetic Materials. Vol. 1. Amsterdam. North-
Holland Publishing Company, 1990
22. www.etrema-usa.com
23. www.materitek.com
24. www.txre.net
25. Quandt, E., Claeyssen, F.: Magnetostrictive Materials and Actuators (Re-
view). In: Proc. 7th Int. Conf. New Actuators (Bremen, 19-21 June 2000),
Bremen: ASCO-Druck 2000, pp lOQ--105 (61 references)
26. Schafer, J.: Design magnetostriktiver Aktoren. PhD Thesis, Saarland Univer-
sity, Germany, 1994 (98 references)
27. Kiesewetter, L.: The Application of Terfenol in Linear Motors. In: Proc. 2nd
Int. Conf. on Giant Magnetostrictive and Amorphous Alloys for Actuators and
Sensors. Marbella, Spain. October 12-14, 1988
28. Butler, J.L., Ciosek, S.J.: Rare earth iron octagonal transducer. J. Acoust. Soc.
Am. Vol. 67 (5), 1980
29. May, C., Kuhnen, K., Pagliarulo, P., Janocha, H.: Magnetostrictive Dynamic
Vibration Absorber (DVA) for Passive and Active Damping. Proc. 5th Euro-
pean Conf. on Noise Control, Naples, 2003, paper ID 159, pp 1-6
30. Clephas, B., Janocha, H.: Extended Performance of Hybrid Actuators. Applied
Mechanics and Engineering, 2000, vol. 5 No. 1, pp 157-168
31. Snorsa, 1.; Tellinen, 1.; Pagounis, E.; Aalitio, 1.; Ullakko, K.: Applications of
magnetic shape actuators. In: Proc. 8th Int. Conf. New Actuators (Bremen,
1Q--12 June 2002), pp 158-161 (14 references)
32. Block, H., Kelly, J.P.: Electro-rheology. J. Phys. D.: Appl. Phys. 21 (1988), pp
1661-1677 (Review, 97 references)
33. Oppermann, G.: Elektroviskose Fliissigkeiten (EVF). Information from Bayer
AG, Leverkusen, Germany
34. Ihrig, D., Koczar, P., Morgenstern-Biin, M.: Electrorheological fluids adapted
for viscous damped automotive devices and other applications in critical en-
vironments. In: Proc. 2nd Int. Conf. New Actuators (Bremen, 21-22 June
1990), VDI/VDE - Technologiezentrum Informationstechnik GmbH Bremen,
pp 231-235
35. Stiebel, Ch.: Leistungsverstiirker zur verlustarmen Ansteuerung kapazitiver
Lasten PhD Thesis, Saarland University, Germany, 2002 (61 references)
36. Adaptronische Transportsysteme mit elektrorheologischen Fliissigkeiten zur
Beforderung sensibler Giiter: Joint project funded by the German Federal Min-
istry of Education and Research (BMBF), 13N6987, www.tib.uni-hannover.de
37. BOlter, R., Janocha, H.: Aktoren mit elektrorheologischen und magnetorheol-
ogischen Fliissigkeiten. In: Automatisierungstechnische Praxis atp 39 (1997) 5,
pp 18-26
38. Hagele, K.H., et al: Continuously Adjustable Shock Absorbers for Rapid-Acting
Ride Control System {RCS). SAE-Paper Nr. 905125, XXIII Fisita Congress,
Turin, 7-11 May 1990, pp 37-46
39. Bolter, R.: Design von Aktoren mit magnetorheologischen Fliissigkeiten. PhD
Thesis, Saarland University, Germany, 1999 (74 references)
336 Hartmut Janocha
40. Bolter, R.; Janocha, H.: Performance of Long-Stroke and Low-Stroke MR Fluid
Dampers. In: Smart Structures and Materials. 1998: Passive Damping and
Isolation, Davis, L.P. (ed.) Proc. SPIE, Vol. 3327, pp 303-313
41. Carlson, J.D., Sproston, J.L.: Controllable Fluids in 2000 -Status of ER and
MR Fluid Technology. In: Proc. 7th Int. Conf. New Actuators (Bremen, 19-21
June 2000), Bremen: ASCO-Druck 2000, pp 126-130 (13 references)
42. Lampe, D.: Untersuchungen zum Einsatz von magnetorheologischen Fluiden
in Kupplungen. PhD Thesis, Technische UniversiUit Dresden, Germany, 2000
(39 references)
43. Michel, F., Ehrfeld, W.: Mechatronic Micro Devices. In: Proc. 1999 Int. Symp.
Micromechatronics and Human Science MHS'99 (Nagoya, Japan, 24-26 Nov.
1999)
44. Berg, U., et al: Series Production and Testing of a Micro Motor. In: Proc. 6th
Int. Conf. New Actuators (Bremen, 17-19 June 1998) Bremen: ASCO-Druck
1998, pp 552-555
45. www.mymotors.de
46. Nienhaus, M., et al: Design and realization of a penny-shaped micromotor. In:
Micromachining and Microfabrication (Paris, France, 30 March-1 April1999)
Proc. SPIE, Vol. 3680B-65
47. GeBner, T.: Recent Progress of Microactuators. In: Proc. 7th Int. Conf. New
Actuators (Bremen, 19-21 June 2000) Bremen: ASCO-Druck 2000, pp 62-70
(51 references)
48. Just, E., et al: SMA Microgripper with Integrated Antagonism. In: Proc.
Transducers '99, 10th Int. Conf. Solid-State Sensors and Actuators (Sendai
Japan, 7-10 June 1999) pp 1768-1771
49. Krippner, P.; Mohr, J.; Saile, V.: Electromagnetically Driven Microchopper
for Integration into Microspectrometers Based on the LIGA Technology. In:
SPIE Conf. Miniaturized Systems with Micro-Optics and MEMS (Santa Clara,
2Q-22 September 1999) SPIE Vol. 3878, pp 144-154
50. Kohl, M., Skrobanek, K.D., Miyazaki, S.: Development of stress-optimised
shape memory microvalves. Sensors and Actuators 72 (1999), pp 243-250
51. www.thinxxs.com
52. Woias, P.: A Silicon Micropump with a High Bubble Tolerance and Self-
Priming Capability. In: Micro Total Analysis Systems, J. Harrison and A. v. d.
Berg (eds.), Kluwer Academic Publishers, Dordrecht, Boston, London, 1998,
pp 383-386
53. Kamper, K.-P., et al: Microfluidic Components for Biological and Chemical
Microreactors. In: Proc. IEEE MEMS-97 Workshop (Nagoya, Japan, 26-30
January 1997), pp 338-343
54. Howitz, S.: Components and systems for microliquid handling. BioMethods,
Vol. 10 (1999), pp 31-73
55. www.amtbe.com
56. Aktives Lagerungssystem mit magnetorheologischer Fliissigkeit fur den Auto-
mobilsektor (ALAS): Joint project funded by the German Federal Ministry of
Education and Research (BMBF), 03N3105
Index
abrasion 298, 307, 310, 316 BLDC motor 104, 105, 132, 147, 148,
ABS 217 152
AC chopper converter 146 blocking force 262,281
AC commutator motor 93 Bode-diagram 165
AC induction motor 146 boost converter 61
AC magnet 141 bootstrap driver 42
AC motor 113, 115, 126, 216 brake 312, 316
achievable accuracy 177 breakdown field 82
active vibration absorber 16 breakdown voltage 24, 36
active-clamping circuit 51 breakthrough voltage 22, 25
adaptive control, AC 4 Bridgman method 278
adaptronics 5, 6 brushes 92,94
air-gap windings 91 Buck-topology 61
AlNiCo magnet 96
amplification system 15 canned motor 128
amplified piezo actuator 266 capacitor 118
analogue amplifier 271 capacitor AC motor 120
anti-lock braking system 177, 217 cascade-circuit 113
application areas 20 charge control 270, 277
asynchronous point-to-point controls charge pump 44, 80
206 cheap motors 86
axial piston motor 181 chopper 93, 98
clamping force 262
backdrop 227 class-A amplifier 272
backward brush shift 92 class-C amplifier 272
Baker clamp 26 classification of drive circuits 38
BCD technology 80 claw-pole construction 152
bell-rotor 97 claw-pole principle 104
bell-type motor 95 claw-pole stator 128
bending element 264 claw-pole type 131
bi-metallic actuator 234 close-loop position control 114
bimetal effect 320 closed loop control 4, 85, 237
bimetals 6 clutch 312, 317
Bingham body 293, 307 collocation 8
Bingham model 295 commutator 92
bipolar connection 106, 137 compensation filter 11, 13
bipolar power transistors 24 complementary drive stage 39
bipolar transistor 21 compound-wound motor 93
338 Index
compressible pressure medium 172 double-acting proportional magnet

computer-numeric control 214 167
concentrated windings 91 drive circuit 37
conduction losses 56 driving circuit 64
constant-flux operation 123 DSP blocks 80
contamination 172 duty cycle 46
control circuit 78 dV/ dt limitation 79
control processor 5 dV/dt protection 54
controlled bridge 101 dV/dt-limiting 55
controller 4 dynamic characteristics 31
controlling method 144 dynamic response 185
CoolMOS 77 dynamic vibration absorber 289
CoolMOS-FET 74
coreless motor 97 EC motor 104
cross-section 28, 33 electrically excited DC motor 92
Curie temperature 258, 260, 287 electro-thermal model 58, 60
current gain 25 electrochemical actuator 234, 236
current limitation 79 electrochemical oxygen pump 234
current limitation circuit 54 electrochemical reactions 234
current transformers 53 electrodynamical principle 158
cylindrical-inside-field type 97 electrohydraulic drive 156
electrohydraulic valves 160
electromagnet 138
damper 305, 313 electromagnetic actuator 63
damping rate 186 electromagnetic converter 158
Darlington stage 27 electromagnetic interference 72
DC chopper circuits 60 electromagnetic micromotor 322
DC chopper converter 145 electromechanical converter 158
DC commutator motor 112, 152 electronic commutation 105
DC magnet 139 electronic stability control 218
DC motor 147, 148, 216 electronically driven motors 89
DC switching converters 60 electropneumatic valves 171
DCB-substrate 57 electrorheological effect 292
dead zone 169 electrorheological fluid 6, 292, 293
decoupling capacitor 47 electrorheological vibration absorber
degree of irregularity 185 303
device stress 72 electrostatic forces 320
diametral magnetization 95 electrostrictive effect 256,277
dielectric isolation 81 electroviscous effect 292,293
differential transformator sensor 196 Elliptec motor 269
directional valves 171 EMI problems 49
disc motor 110, 137 energy controller 1,233
disc-type motor 90, 95 energy converter 1,233
disk translator 265 energy loss 171
displacement amplification 265 equivalent circuit 262,281,299,311
displacement control 155 ESP 218
dither 164 expansion element 239
DMOS-FET 74, 76 external commutated drive 143
double-acting cylinder 183, 189 external rotor 108
Index 339
external-rotor motor 90, 117 half-wave zero-current converter 74

externally commutated motor 86, 90 Hall element 105, 106, 114
hammering actuator 139
Faulhaber winding 97 handling systems 219
feedback vector 204 health monitoring 8
Ferraris motor 117 heteropolar motor 90
ferrite magnet 95 heteropolar type motor 90
ferrofluids 310 high grade motors 86
field weakening 113 high voltage boosting 65
field-weakening operation 123 high-frequency tool 118
finite-element program 57 high-grade motor 96
five-phase stepper motor 150 high-side drive 41
flapper-nozzle 161 high-side driver 37
flat rotor 97 HITFET 79
flight simulation 222 homogeneous ER fluids 297, 316
float-zone method 278 homopolar machine 91
floating body sensor 195 hot spots 26
flow amplification 163 HY motor 134
flow mode 298, 310 hybrid amplifier 271, 272
flow sensor 195 hybrid stepper motor 133, 150
flow valves 171 hybrid synchronous motor 127
flow-signal function 163 hydraulic drive 112
fluidic microsystem 332 hydraulic motor 216
fluidpower actuator 155 hysteresis 4, 164,174,241,252,273,
force compensation 16 308
four-pulse connection 107 hysteresis motor 126, 130, 131
four-quadrant circuit 145 hysteresis operators 12
four-quadrant mode 146
hysteretic effects 11
four-quadrant operation 61, 111
hysteretic nonlinearities 10
fractional- horsepower motor 85
FRED diodes 24
frequency converter 147 ideal power switch 20
frequency response 163 IGBT 21,33
frequency spectrum 212 inchworm motor 267
frequency variation 122 induction sensor 195
fuel cells 234 inductive sensor 196
full-step mode 134 injector 275
full-wave converter 74 intelligent actuator 8, 10
interference suppression 102
galvanic isolation 48 intermediate-rotor motor 90
gate charge 31, 56 intermediate-rotor type 97
gate losses 56 intermediate-stator motor 90
gate-source overvoltage 52 internal rotor 108
gear-pump 195 internal-rotor motor 90, 117
gripper system 326 inverse filter 9
inverse piezoelectric effect 254, 259
H-bridge 146 ironless rotor 112
half-controlled bridge 145 ISO 5599 179
half-step mode 134 isolated gate drive 38, 45
340 Index
jet pipe 161 microdrive 322

Joule effect 277 microelectromechanical systems
junction isolated smart-power (MEMS) 5
technology 81 microelectronics 6, 321
junction temperature 55, 57, 82 micromachining 321
junction-FET 76 micromirror 328
micromotor 109, 322
laminar translator 264 micropump 331
lap windings 96 microscanner 328
lead-zirconium-titanate (PZT) 256 microsensor 6
level shifting stage 44 microsystem technology (MST) 5, 333
lifting magnets 62 microvalve 330
LIGA technology 321 Miller plateau 31
limited-angle rotary actuator 186 moonie transducer 266
line-commutated converter 145, 146 MOSFET 21
linear BLDC motor 111 motion simulator 223
linear direct drive 111 motor mount 319
linear motor 90, 137, 159 moving coil 97, 158
load pressure 185 moving magnet 140
load stiffness 164 multi-loop controls 200
logic-level interfacing 80 multifunctional elements 6
longitudinal effect 255, 259, 261 multilayer ceramics 258, 276
loop amplification 208 multilayer stack 276
loop control chain 3 multiple chip mounting 78
low-pressure accumulator 217 multiple-stage control valve 168
low-side drive 39
low-side driver 37 nano-sized MR fluid 309
natural angular frequency 185, 186,
magnetic field curve 92 199
magnetic incremental encoder 105 N dFeB magnet 96
magnetic shape memory 243 Nitinol 244
magnetorheological effect 307 nominal flow 179
magnetorheological fluid 6, 306 nominal width 179
magnetostrictive actuator 8, 234, 277 non-punch-through 33
magnetostrictive constant 279
magnetostrictive effect 277 ohmic region 29
main stage 165 olenoid valves 175
man-machine interface (MMI) 3 OmniFET 79
Matteucci effect 279 on-resistance 30
Maxon winding 97 one-pulse connection 107
mechatronics 5 one-quadrant amplifier 300
membrane cylinder 189 one-way memory effect 242, 245
membrane motor 188 open loop amplification 199
memory-programmable control 214 open-loop control 85, 125
Merril motor 127 operator 8
metal hydride actuator 240 optical incremental encoder 105
microactuator 234, 260, 320 optical sensor 196
microchopper 329 opto-couplers 48, 271
microcontroller cores 80 output characteristic 29, 35
Index 341
overload protection 52, 79 power electronics 8-10

overvoltage protection 49, 79 power switching devices 20
power-MOSFET 28
pancake motor 90,97 power-MOSFET technologies 75
parallel bimorph 265 PPT 1-controller 212
parasitic capacitances 31 PPT -controller 200
parasitic components 29 pre-magnetization 283, 285
Paschen effect 321 pressure amplification 164
path control 207 pressure control 211
penny motor 324 pressure control valves 171
permanent-magnet DC motor 94 pressure drops 175
permanent-magnet motor 126 pressure oscillation 212
permanent-magnet stepper motor 133 pressure-signal function 163
phase displacement 118 process computer 5
phase-angle control 103, 122, 146 propagation delay 49
PI-controller 212 proportional magnet 158
PID-controller 197 proportional valves 157, 160
piezo actuator 234 protection circuits 49
piezo constant 261 protection functions 80, 82
piezo effect 320 pryoelectricity 259
piezo pilot stage 180 pulsating-current motor 99
piezoceramic 257 pulse-width modulation 103, 176, 301
piezoelectric actuator 4, 8, 15, 159, punch-through 33
254 push-pull stage 40
piezoelectric charge constant 258 PWM 103,147
piezoelectric effect 254 PWM servo-amplifier 93
piezoelectric elements 6 PZT 256
piezoelectric polymers 259
piezoelectric stack 275 quasi-resonant converters 72
piezoresistive sensor 194 quiescent flow function 163
pilot control 166
pilot stage 161 radial magnetization 95
piston motor 181, 188 radial piston motor 181
plunger electromagnet 144 radio disturbance 92
PM motor 134 rare-earth magnet 96
pneumatic gripper 220 real-time system 5
polarization 257 reciprocal piezoelectric effect 254
pole shoe 96 reconstruction filter 11, 13
porous metal 140 rectifier 93
portable appliances 101 reduction gear 324
positioning system 238, 265, 273, 325 relaxor 277
positive overlap 169 reluctance motor 126, 131
potentiometer sensor 196 resistance control 155
power amplifier 1, 270, 285, 299, 313 resistance-start motor 120
power circuit 19 reverse recovery current 83
power controller 76, 82 revolving field 118
power converter 122 rhombic winding 97
power diode 22 rodless cylinder 189, 220
power dissipation 56, 57 rotary motor 181
342 Index
rotary valve 172 smart discretes 78

rotatory speed 184 smart power IC 80
rotor position encoder 104 smart structures 6
round-frame-type motor 108 smart-power devices 74
running capacitor 120 SmCo 96
snap-off 23
safe operating area diagram 56 snubber circuit 36
saturation 25 solenoid 140
saturation voltage 34 solenoid actuator 139
Schottky diode 23 solid-state reaction 235
secondary breakdown 26 sonar system 288
secondary control 157 speed enhancement circuits 49
secondary-controlled hydraulic motor speed regulation 98
226 split-phase motor 124
sedimentation 297, 306, 307, 310, 316 squeeze mode 298, 310
self-clocked motor 92 squirrel-cage winding 117, 127, 132
self-commutated drive 143 stack transducer 274
self-commutated motors 86 stack translator 260, 263
self-heating effect 58, 59 stage mover 227
self-isolation 81 standardization 85
self-sensing 243 standardized nominal flow 179
self-sensing actuator 6, 8, 10, 12, 13 starting capacitor 120
sense-MOSFET 53 state controller 203
sensorless control 105 static response characteristics 186
series bimorph 265 status feedback 79
series wound motor 93 steady valve 161
servo drive 111 step response 163, 165
servo valves 161 stepper motor 104, 132, 149, 152
shade-pole motor 124 stick friction and leakage 208
shape memory alloys 241 super-smart-power devices 76
shape memory effect 241, 320, 326 superelasticity 242, 251
shape memory elements 6 surface micromachining 321
shape-memory actuator 234 swinging magnet 139
shear effect 265, 277 swinging-armature magnet 142
shear element 265 switched reluctance motor 115
shear mode 298, 310 switching amplifier 271, 272, 285, 300
shunt resistor 52 switching integrator 206
shunt-wound motor 93 switching losses 56
silicon carbide 82 synchronized induction motor 128
simulation 15, 58 synchronous motor 125
single-acting cylinder 183, 189
single-loop control 199 tacho-generator 105, 147
single-phase (induction) motor 115 tail current 35
single-phase motor 118 tandem master brake cylinder 217
single-phase stepper motor 138 tapped field winding 102
single-phase synchronous machine 149 TCS 218
six-pulse connection 107 temp-FET 79
skew winding 97 temperature coefficient 28
slotless windings 91 temperature distribution 58
Index 343
Terfenol-D 278 two-quadrant amplifier 300

theater hydraulics 227 two-quadrant converters 62
thermal conductivity 83 two-way effect 247, 248, 250
thermal impedance 57, 59 two-way memory effect 242, 245
thermal management 55
thermo-bimetal 252, 265 ultra sound sensor 195, 196
thermo-chemical actuator 239 ultrasonic motor 268
third brush 98 uncontrolled bridge 101
three-phase drive 113 unconventional actuator 233
three-pulse connection 107 unipolar connection 106, 137
throttling losses 157 universal motor 94, 101, 146
thyristor 20, 36
TOPFET 79 valve actuators 62
torque 184 vane motor 186
torque-motor 159 variable reluctance stepper motor 133
totem-pole 41 variable resistors 162
traction control system 218 vibration absorbers 14, 303
transformation temperature 241, 245 vibrator 139, 142
transformer-coupled gate drive 45 voice-coil drive 151
translatory motor 183 voice-coil motor 111, 144
transversal effect 255, 259, 264 voltage control 270,271,277
trench-FET 74, 76 voltage slew rate 54
turbine sensor 195 VR motor 134
turn-off 26
turn-off behavior 22 wave windings 96
turn-off losses 34 waved-formed coil 110
turn-on behavior 22,31,32,35 wheel brakes 217
turn-on resistance 77 windings 91
turning magnet 139 worm motor 284
turning piston motor 188
two-pulse bridge 101 zero overlap 165
two-pulse connection 107 zero-current converters 72

Actuators Basics and Applications

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Actuators Basics and Applications

Uploaded by

Copyright:

Available Formats

Actuators

With 303 Figures and 47 Tables

Library of Congress Control Number: 2004109912

ISBN 978-3-642-08266-5 ISBN 978-3-662-05587-8 (eBook)

Actuators have wide-ranging applications in our everyday environment, from

automation technologies, and especially for potential users of actuators and

Saarbriicken, May 2004 Hartmut Janocha

Chapters 1 and 5: Prof. Dr.-Ing. habil. Hartmut Janocha

Chapter 2: Prof. Dr.-Ing. Stefan Butzmann

Chapter 3: Prof. Dr.-Ing. Hans-Dieter Stolting

Chapter 4: Prof. Dr.-lng. Wolfgang Backe

1.1 What is an Actuator? 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1

1.2 Actuator as a System Component 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3

103 Actuators in the Fields of Mechatronics and Adaptronics 0 0 0 0 0 0 0 0 5

1.4 "Intelligent" and Self-Sensing Actuators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8

1.401 "Intelligent" Solid-state Actuator 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 9

1.402 Self-sensing Solid-state Actuator 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10

Example Application: Piezoelectric Positioning System 0 0 0 12

1.5 Design of Actuators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 13

202 Power Switching Devices 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20

20202 Power Diode 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 22

20203 Bipolar Power Transistor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 24

203 Drive Circuits for MOS-gated Devices 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 37

20301 General Remarks 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 37

20302 Low-side Drive Circuits 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 39

20303 High-side Drive Circuits 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 41

Drive Circuits for p-channel MOSFETs 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 41

Bootstrap Drivers for n-channel MOSFETs 0 0 0 0 0 0 0 0 0 0 0 0 0 42

2.3.4Isolated Gate Drive Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Control Methods 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 122

Shaded-pole Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 124

3.402 Synchronous Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 125

Permanent-magnet Rotor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 127

Hysteresis Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 130

Reluctance Motor 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 131

3.403 Stepper Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 132

305 Limited Motion Drives 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138

30501 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138

30502 Electromagnets 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 138

Swinging-armature Magnets, Vibrators 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 142

30503 Voice-coil Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144

306 Closed and Open Loop Control Methods 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144

30601 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 144

30602 Line-commutated Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 145

30603 DC Chopper Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 145

306.4 AC Chopper Converter 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 146

30605 Converter for Polyphase Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 147

307 Applications 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 147

401 Introduction 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 155

402 Proportional Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 157

40201 Electromagnetic and Electromechanical Converters

40202 Electrohydraulic Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 160

Servo Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 161

Proportional and Control Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 167

40203 Electropneumatic Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 171

Steady Pneumatic Directional Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 171

Pneumatic Proportional Pressure Control Valves 0 0 0 0 0 0 0 0 1 74

403 Switching Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 176

40301 Electrohydraulic High-speed Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 177

40302 Electropneumatic Switching Valves 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 178

4.4 Fluidpower Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181

4.401 Hydraulic Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181

Rotary Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 181

Translatory Motors 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 183