Professional Documents
Culture Documents
Actuators Basics and Applications
Actuators Basics and Applications
H. Janocha (Ed.)
Actuators
Basics and Applications
€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
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
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
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
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
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
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
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
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
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
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
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
Hartmut J anocha
E~nergy Energy
Energy
converter
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)
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.
Auxiliary
energy 0
~
Man-
Energy Power Micro-
machine
converter electronics computer
interface
Actuator
Energy
matter Process
(information)
Auxiliary
energy L_~~~===~-J]
~------11/0
Man-
machine
Software, e.g. interface
process identification
controller synthesis
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
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
Power supply
• '-------~~------__-___+-!--·+-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)
3 Operators are here used to describe mathematically the mapping between the
input and output time functions of dynamical systems.
1 Introduction 9
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-~-+·-~--~
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 .
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.
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)
Equation (1.1) is called the sensor equation and (1.2) the actuator equa-
tion of the solid-state transducer.
(1.5)
for the compensator follows from equation (1.2) by inverting the relationship
between s and X.
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
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
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. 1.11. Third vibration mode of the absorber structure analyzed using ANSYS®
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
References
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
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.
<(
"E
~
::J
()
Thyristor
1000 "F=---:----:--,....,..---..
Bipolar transistor
IGBT
100 1 = - - - - - - - , .
10 MOSFET
1
1 10 100 1000 10000 Voltage I V
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
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.
'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
""' ~~·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]
VcBo
VcEo ;:::j VB. (2.1)
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
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'
2.2.4 Power-MOSFET
Static Behavior
0
Fig. 2.10. Cross-section of power-MOSFET
100
<(
~50 VGS
10----1~5----2~
0 ----~5~--~0----5~--~
-~q~ 2~5--~30
0--~
V05 / V
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)
Dynamic Characteristics
1200
"--- C4ss
600 1\
,"- r--- Coss
'- Crss
0
0 10 20 30
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
; :
0.5 1.5
t/ us
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
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]
G~
E
- 200 oL----2~--~4----~6----~B~--~10
VcEIV ---+
Fig. 2.19. Output characteristic of an IGBT
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
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.
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. 2.24. 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
5VJ
ov
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.
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_
------------------------------------------------------------'
'
Charge Pumps
PT
.---------.---------------~--o Vs
Fig. 2.31. Elementary version of charge-pump driver with increased turn-on time
PT
PT
0
-1 ~------L-------~------~------~------~
0 2 4 6 8 10
0 2 4 6 8 10
t/ ms
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
]
f
,-------~-------,
PT
.. ________________________,.
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
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
ZD
PT
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
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
Overload Protection
Fig. 2.42. Current measurement with shunt resistor. a Shunt resistor in main
current path, b use of shunt resistor with sense-FET
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
Overtemperature Protection
RG,int
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)
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.
with:
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
Cth,1
P(t)
Cth,2
Cth,3
Power
dissipation
...,.. Heat
generation
~ /
Increase of
Roson
,..............................................................................................................................................................................................................................................
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
2.5.1 Topologies
!Vemf
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
0
> (")
z
(.!)
N
'<!"
0 a..
II
~
II
()
-5?
I 0::: I Cl)
0
z
~
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.
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
.-----------~-------------.----------~
• 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
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
,--------{) Vs
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
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
Rotor
position
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.
T1 01 L,
/LL
I
Vs! c, 02 LL
2.6.1 Overview
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
2.6.2 Trench-FETs
p p
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
Statuso-- - - --------,
Supplyo-- - - - - - -----,
Input o---r=;::=::£:=::;===;-ttl~
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.
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. 79. Intelligent power controller block structure (Courtesy of Royal Philips
Electronics)
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
Hans-Dieter Stolting
3.1 Introduction
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.
Table 3.1. Characterising features of small cheap and high grade motors
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
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
f"VV"\ fVVV"\ Series or shunt (parallel, separate) winding 1.1 Permanent magnet
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
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
Basic
circuit
l4J 1Juc
E)z
·~9'
Uu Uu ~"q
9 ~0'~ I I
Nt2 Nc c
I , _ _..II \._ ..../ '-~
Speed-
torque
curve
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
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
generator or electric
governor resolver (R) for speed
position control using a control position control
pulse generator (encoder) and switching
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
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
3 Electromagnetic Actuators 91
Stator
*
·-
Rotor
~
a ....__-_--_-_--_-_-_-_-__, b
~+I~::~
L.___ _ _ __J
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
3 Electromagnetic Actuators 93
I N
A rpm
0
0 5 10 15
T: torque, TN: nominal torque, N: speed, P: output power, I: current, T): efficiency
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.
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-
3 Electromagnetic Actuators 97
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
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
the commutation is worse (three brushes motor [3]). This method should be
used only for a short time.
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
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)
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-
3 Electromagnetic Actuators 101
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
Rotor or
armature winding
a b
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
a b T
~N~
c T d T
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.
3 Electromagnetic Actuators 107
,]: W
H
R
T
Phase winding
Hall element
Resistance
Transistor
f g
unipolar bipolar
0: not magnetized
Magnet pole
Coil
Stator winding
EmB Winding
Magnet
c::::J
Fig. 3.23. Constructions of disc motors with radial and axial magnetic fields
Shaft
-- Rotor
Nonmagnetic stator
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.
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
112 Hans-Dieter Sti:ilting
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.
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
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).
Three-phase system
L1 Lz L3
Hall
element
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.
Laminated core
Stator winding
Rotor bar
Fig. 3.29. AC asynchronous motor, capacitor motor
Fan impeller
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).
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.
118 Hans-Dieter Stolting
.--------o t Right
• 0
t Left
t Right
• 0
k
______._____. t Left
VN
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
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
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-
3 Electromagnetic Actuators 123
Speed N
regulation range
I
1- Constant-flux NN
range
f<fN, lf;<lf;N
Field-weakening
--range-
f>fN, v; =v;N
IN
Shaded-pole Motor
Squirrel-
cage
rotor
Single-axle Two-axle
Symmetrical cross-section
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
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).
Introduction
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
Asyncronous torque
T
,..---.....::- - - Breakdown
torques
.I
N
Syncronous torque
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
Ring coil
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
130 Hans-Dieter Sti:ilting
System 1
System 2
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-
3 Electromagnetic Actuators 131
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
d ·······:r.·:r······
E ••••••••••••• •••• .....,.---Ideal
··········r························::::., motor
TH i Real
Ns-;r
TE: eddy-current torque, TH: hysteresis torque
Flux path
---------Asynchronous
t
T
breakdown torque
-Synchronous
-Pull-in torque
I
I
I
I
I
N---
Fig. 3.47. Reluctance motor, speed-torque curve
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.
3 Electromagnetic Actuators 133
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
Axially magnetized
permanent magnet
System 1
(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
3 Electromagnetic Actuators 135
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.
136 Hans-Dieter Stolting
b d f h
Fig. 3.50. Control principles of stepper motors
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
3 Electromagnetic Actuators 137
-
<I>~
I I
a • +/ b -I
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.
3.5.2 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:
3 Electromagnetic Actuators 139
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
140 Hans-Dieter Stolting
~Core
A~aMe
Fig. 3.54. Rotating DC magnet with helix shaped faces of core and armature
3 Electromagnetic Actuators 141
Magnetic core
One-sided
pole system
Double sided
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.
142 Hans-Dieter Sti:ilting
Magnetic
core
Winding
\
Armature Start postion
c
F: force, s: way
Fig. 3.56. Principles of AC magnets and their force-way curves
~''H
~- ......... ~~
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.
3.6.1 Introduction
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.
<>----"-..(_
220V-
0 ~
h fd
H: heating coil, r: rectifier, fd: free-wheeling diode, c: control,
pm: permanent magnet
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).
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. 7 Applications
Torque sensors
Diaphragm
for pressure
compensation
Stator
core
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
Screw spindle
Fig. 3.70. Apparatus fan . BLDC motor with claw-pole stator, external magnet
rotor and built-in electronics
152 Ha ns-Dieter Stolting
Fig. 3. 71. Micro blood-pump and its placing in the left a nd right cardiac valve
3 Electromagnetic Actuators 153
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
4.1 Introduction
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].
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.
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)
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-
4 Fluidpower Actuators 159
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
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
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.
162 Wolfgang Backe and Andreas Klein
Flapper nozzle
Armature
Bending pipe
Flapper
Filter
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
4 Fluidpower Actuators 163
1.0
oiJ o.75
~
~ 0.5
Q)
E
::J 0.25
~ 0.5 ~
~ \<:'
0.25 ~0
e,'S
~
Hydraulic zeropoint ,~~
Po - pL(sgn i)
(4.1)
PNom
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
Fig. 4.8. Static characteristic curves and characteristic values of servo valves
4 Fluidpower Actuators 165
(4.2)
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
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
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
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
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
4 Fluidpower Actuators 169
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
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 -
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.
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.
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
~----------93 --------~
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
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
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)
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
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.
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
Fig. 4.22. Electropneumatic pressure control valve with fast-switching valves (Hon-
eywell Lucifer)
·4·
~B
Fig. 4.23. High-speed switching valve (Continental Teves)
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.
4 Fluidpower Actuators 179
~
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.
Rotary Motors
• •
volume, range, sure,
cm 3 rpm bar
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
4 Fluidpower Actuators 183
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
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.
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-
4 Fluidpower Actuators 185
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.
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
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
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
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
4 Fluidpower Actuators 189
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
Ill :
force dependent on direction due to different pis-
ton areas
I I
stroke: 10 ... 2000 mm
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
Pv AvE AvA co
® T, 'J:. ~ ~
7;"
A, y
'J:.y p,J1P2,
~
y y Av=AvE=AvA "'
5
):L
a b c
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)
~
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'-------'
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-
4 Fluidpower Actuators 193
Av 2ao-vfVdispi"Pcabs"'
1t•J(\I.J;spi+Vdead) -
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.
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
194 Wolfgang Backe and Andreas Klein
·-
Function Mechanical Strain gear Inductive Piezoelectric Piezoresistive
principle
Schematic
-ED-
~ -e~
.
_;- -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
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-
196 Wolfgang Backe and Andreas Klein
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.
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
4 Fluidpower Actuators 197
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).
.'f.oad
Fig. 4.36. Signal flow diagram of the linear drive with related variables
4 Fluidpower Actuators 199
PPT-controller
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,
4 Fluidpower Actuators 201
Kx
Ak+Kiiv
.-----------~t=~----------~
DT1-element to suppress
the static pressure portion
Linear Method
- 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.
4 Fluidpower Actuators 203
w,x
lm
/ Re
0=1
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,
204 Wolfgang Backe and Andreas Klein
~
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
u.y
w0 =125 1/s
D =0.1
Tv =0.05 s
t
0.2 s 0.4 s
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.
206 Wolfgang Backe and Andreas Klein
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)
Position Control
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
__________c·o
a 5 v 10 y
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)
: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
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
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:
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
Pressure Control
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:
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.
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
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
(.)
Fig. 4.49. Layout of the PT1-portion of the pressure control. a Step responses for
various controller settings, b controller parameters
CNC unit
Actual
value
Fig. 4.50. Connecting a hydraulic drive to a CNC and a MPC (Memory Pro-
grammable Control)
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.
216 Wolfgang Backe and Andreas Klein
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
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.
Low pressure
accumulator
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) .
4 Fluidpower Actuators 219
Wheel brake
Plunger
unit
Electric
adjustable
friction brake
Controller L____j~~~~~~~~------'
Fig. 4.53. Anti-lock braking system with plunger cylinder
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
220 Wolfgang Backe and Andreas Klein
Stroke: up to 1500 mm
Pay load: up to 22 kg
Speed : max. 3.5 m /s
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.
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
a b c
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].
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
4 Fluidpower Actuators 223
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
224 Wolfgang Backe and Andreas Klein
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
4 Fluidpower Actuators 225
Hydrocylinder
(translatory axis x)
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
226 Wolfgang Backe and Andreas Klein
Released
by
control
-n2command
Tank line
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].
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
References
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.
232 Wolfgang Backe and Andreas Klein
Hartmut Janocha
5.1 Introduction
Hydraulic
Electrostatic
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.
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
Electro-
a chemical cell (Sx)
b lime -
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
5 Unconventional Actuators 237
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-
Advantages Disadvantages
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.
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
5 Unconventional Actuators 239
Expansion Elements
Elastomer
element
Expansion
material
a Case
Max. temperature
t
s
E
E ~
oF±~~--------~-+--~~~
fJ -
b Operating range
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)
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]
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
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.
5 Unconventional Actuators 245
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
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).
(5.5)
(5.6)
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:
5 Unconventional Actuators 249
s
eM = (±) 2R (5.7)
Advantages Disadvantages
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
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
5 Unconventional Actuators 251
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
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
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.
Advantages Disadvantages
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)
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
(5.10)
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.
5 Unconventional Actuators 257
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
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
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
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.
5 Unconventional Actuators 261
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
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
5 Unconventional Actuators 263
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)
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.
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
~: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
Fluid Bellows
F z,S2
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. 5.25. Ultrasonic motor. Examples of designs and charact eristic values
7 Note the similarity to the operating mode of asynchronous motors (Sect . 3.4.1).
5 Unconventional Actuators 269
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
Advantages Disadvantages
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.
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
5 Unconventional Actuators 273
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
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.
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
5 Unconventional Actuators 275
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
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).
5 Unconventional Actuators 277
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
s
10"3m/m
Tp= 2.8 MPa
100 150
_jj_
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].
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
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 .
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
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
(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])
5 Unconventional Actuators 285
Advantages Disadvantages
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.
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
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-
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
Fixed magneto-
strictive rod
1 em
t------1
a
Differential
amplifier2
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])
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.
5 Unconventional Actuators 293
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
t
E
E=O
Fig. 5.42. Set of characteristic curves (shear stress vs. shear rate). Ideal electrorhe-
ological fluids
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
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]
10 9/i
(Omf1
300
250
200
150
100
50
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
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
C = CoE
co
G = Co"'
co
Advantages Disadvantages
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
Positioning Drive
v,
V2 ER fluid valve 2 ~ERfluid
valve 1
High
voltage
sources
V3 ER fluid valve 3 ER fluid
v4
~~4
Hydraulic
cylinder
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
5 Unconventional Actuators 303
Tactile chamber ER fluid Membrane
Electrode
fT~f{!f-Tif-J~n7'"-;=;.~t;zjLER fluid
valve
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
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
5 Unconventional Actuators 305
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.
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.
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
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])
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
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
T = __l:_ (1 + Rcoi!)
Rcoii Red
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
5 Unconventional Actuators 313
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
Advantages Disadvantages
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
5 Unconventional Actuators 315
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).
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
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
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
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
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
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
5 Unconventional Actuators 323
SGL
• diameter: 12.8 mm
• height : 1.4 mm (exci . shaft)
- torque : 80 ~Nm C!l 200 mA
- speed: up to 60,000 rpm
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
Fig. 5.65. Stage of the planetary gear. Gear modulus 55 J.lm. Gear stage ratio: 3.6
[43]
5 Unconventional Actuators 325
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]
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.
Gripping
jaw
Actuation
} unit II
~Bond pads
(circuit 2)
Actuation
unit I
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
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
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
Ruby ball
--
Fig. 5.72. Principle of a micro ball valve [43]
NiTi
microdevice
Valve
seat
Pins
Fig. 5.73. Cross-section of the microvalve. A, Band C denote valve ports [50]
5 Unconventional Actuators 331
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.
Actuator
deflected
Outlet
valve
Pump cavity
Fig. 5. 75. Principles for micro gear pumps with an external gear pair [43]
Injector fluid __
Carrier fluid
+
Manifold
(glass, silicon)
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
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
5 Unconventional Actuators 335
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