Sensors and Actuators A 219 (2014) 47–57

Contents lists available at ScienceDirect

Sensors and Actuators A: Physical

journal homepage: www.elsevier.com/locate/sna

Design of an impact drive actuator using a shape memory alloy wire

Shinya Hattori, Masayuki Hara ∗ , Hiroyuki Nabae, Donghyun Hwang, Toshiro Higuchi
Department of Precision Engineering, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan

a r t i c l e i n f o a b s t r a c t

Article history: This paper introduces an impact drive mechanism (IDM) that utilizes the rapid contraction property of
Received 2 May 2014 shape memory alloy (SMA) wire. In this study, possible structures for an SMA-wire-based IDM actuator
Received in revised form 11 July 2014 are investigated, and a prototype actuator comprising a main body, an inertia body, an SMA wire, and
Accepted 24 August 2014
a bias spring is developed. To verify the applicability of the prototype actuator as a positioning device,
Available online 3 September 2014
driving experiments were conducted under various conditions. The experimental results demonstrated
that the prototype actuator enables bidirectional step-like movement with several step sizes by changing
MSC:
the profile of applied voltage, such as amplitude, duty cycle, and frequency. In addition, we developed PI-
00-01
99-00
controller-based position control systems by using these three parameters as control input; the control
characteristics and potential applications of each method were discussed. These results implied that the
Keywords: SMA-wire-based IDM actuator has the potential to be used as a new linear actuator in sub-millimeter
Shape memory alloy order driving.
Impact drive mechanism © 2014 Elsevier B.V. All rights reserved.
Linear actuator

1. Introduction the effect of external energy such as electrical energy elements.

In recent years, technological advancements have acceler-
ated the trend of downsizing of electrical products. In particular,
mobile phones have undergone rapid downsizing, which nowa-
days support several functions such as high-resolution imaging
and high-speed wireless communication. For developing compact
products with multiple functions, a compact high-performance
actuator that can be driven in the limited space should be one
of the most essential components. Electromagnetic motors have
been widely used in actuators because of their high performance,
easy availability, and good controllability. However, the structure
consisting of magnets and coils prevents further size reduction.
Currently, the world’s smallest commercialized motor has 2 mm
in diameter and 5 mm in length [1]. Thus, the application of elec-
tromagnetic motors is becoming increasingly difficult as electrical
products are downsized. Hence, alternative actuators that are
smaller but offer high actuation performance will be necessary in
the near future.
Solid-state actuators have been also studied with the aim of real-
izing smaller and high-power actuators. Solid-state actuators use
active materials that generate micro displacement (or strain) under
∗ Corresponding author. Tel.: +81 3 5841 6465.
E-mail addresses: hattori@aml.t.u-tokyo.ac.jp (S. Hattori),
masayuki@aml.t.u-tokyo.ac.jp (M. Hara), nabae@aml.t.u-tokyo.ac.jp (H. Nabae),
donghyun@aml.t.u-tokyo.ac.jp (D. Hwang), higuchi@aml.t.u-tokyo.ac.jp
(T. Higuchi).
Materials commonly used in solid-state actuators are piezoelectric
elements, magnetostrictive materials, and shape memory alloys
(SMAs), etc. Among these materials, the piezoelectric actuator has
particularly attracted our interest owing to its large generative
force, precise sub-micron displacement, and fast responsiveness
[2]. Hence, various types of actuators based on these properties
have been proposed for application to various fields [3–9]. For
example, impact drive mechanism (IDM) actuators, which are well
known as precise positioning devices that utilize the impact (iner-
tia) and friction force produced when the piezoelectric element is
rapidly deformed [3], have been studied for application to scanning
tunneling microscopes [4] and cell micromanipulators [5]. Further,
Yoshida et al. [6] proposed smooth impact drive mechanism (SIDM)
actuator by improving upon the IDM principle. Currently, SIDM
actuators are implemented in digital cameras to facilitate auto
focus, zoom, and image stabilization. The present study also focuses
on the IDM because the application of the driving principle could
be very useful in implementing actuation in limited space.
The IDM mainly utilizes the fundamental physical phenomenon
that movement of an object accelerated on a friction surface by
external impact force stops due to the friction force between the
object and friction surface. In previous studies, various types of
actuators utilizing the impulsive force have been developed, in
which the impulsive force was mainly generated by electromag-
netic force [10], inertia force caused by rapid deformations of
the piezoelectric element or magnetostrictive material [3,11], air
pressures [12], and thermal expansions [13,14]. However, IDM
actuators that can offer better performance than piezo-based IDM

http://dx.doi.org/10.1016/j.sna.2014.08.013
0924-4247/© 2014 Elsevier B.V. All rights reserved.
48 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57
Main body Piezo Inertia body
(1) Start Base
(2) Rapid contraction Repeat
(3) Slow extension
(4) End
Displacement Voltage pattern
Fig. 1. Driving principle of impact drive mechanism [3]. As an example, the illus-
tration uses a piezo-based IDM actuator to explain the behavior in each step.
actuator have not been developed so far. Thus, this study examines
the possibility of using an SMA as the driving source in an IDM.
SMAs can be restored to a predetermined shape by Joule heat-
ing [15]. The possible strain and generative stress of an SMA are
obviously larger than those of a piezoelectric element and magne-
tostrictive material [16], although the responsiveness is not so high
(basically, the SMA is actuated less than 100 Hz [17]). Since SMAs
possess excellent energy density and tolerability for galling and cor-
rosion, SMA-based actuators can potentially be applied in various
environments. In this study, an IDM actuator using an SMA wire is
proposed, and a prototype device is developed to examine whether
the proposed actuator performs the expected movement [18]. This
paper mainly discusses the possibility of the SMA-wire-based IDM
actuator, investigating the basic performance and characteristics
of the prototype actuator. In addition, with a view to enabling the
practical application of the proposed actuator, we attempt to pro-
pose position control methods based on the characteristics of the
proposed actuator.
2. Impact drive mechanism
In this section, we introduce the basic principle of IDM to
roughly grasp the behavior of an IDM actuator with an SMA wire;
here, we take a piezo-based IDM actuator as an example. Fig. 1
shows the schematic diagram of a typical piezo-based IDM actuator.
The IDM actuator is mainly composed of three components: a main
body in contact with a base, a piezoelectric element to produce the
impulsive force, and an inertia body for causing the impulsive iner-
tia force to the main body. Fig. 1 also illustrates the driving principle
and the process sequence is as follows:
(1) In the initial state, the main body remains stationary condition
on the base.
(2) The inertia body experiences high acceleration toward the
main body when the piezoelectric element is activated so as
to rapidly contract, inducing an impulsive inertia force on the
main body. At this time, the main body slips slightly toward the
inertia body if the inertia force exceeds the static friction force
between the main body and the base.
(3) The piezoelectric element is activated to gradually extend until
the original length, keeping the inertia force below the static
friction force; the main body never moves in this period.
(4) Finally, the IDM actuator returns to initial state (1) with a small
drift toward the inertia body.
The process described in (1) to (4) produces a tiny step-like
displacement toward the inertia body. The step size depends on
the friction and the impact force, which is equal to the inertia
force generated by the deformation of the piezoelectric element. By
repeating this process, the IDM actuator achieves linear movement
consisting of continuous step-like displacements. Additionally, the
IDM actuator enables the movement in the opposite direction by
inverting the contraction and extension speeds of the piezoelectric
element.
As shown in Fig. 1, it should be noted that the coupled action
of rapid contraction and gradual extension (or rapid extension and
gradual contraction) is necessary for the IDM. With regard to the
application of an SMA wire to the IDM, rapid contraction can be
obtained by intensive heating of the SMA wire, and natural cool-
ing due to the air after the heating could realize gradual extension.
Thus, the piezoelectric element can be replaced with an SMA wire in
the structure shown in Fig. 1 and an SMA-wire-based IDM actuator
would be feasible. However, the rapid extension cannot be realized
by the natural air cooling because the cooling speed is not enough;
the use of other cooling sources such as a Peltier element might
allow faster cooling but the structure becomes more complicated
and further downsizing would be difficult. Thus, we expect that an
IDM actuator using an SMA wire can achieve only one-way move-
ment, i.e., movement toward the inertia body; the present paper
defines the direction of this movement as positive direction.
3. IDM actuator using an SMA wire
3.1. Design
In general, the application of bias force is necessary to repeat the
contraction and extension of SMA wire; the SMA wire would slack
after heating without the bias force. The bias force can be achieved
by applying an external force such as gravity, elastic restoring force,
or tension of another SMA wire that is activated antagonistically
[19]. Using these bias methods, some possible designs (structures)
of SMA-wire-based IDM actuator can be devised on the basis of the
structures of piezo-based IDM actuator. Fig. 2(a) shows the struc-
ture commonly used for the piezo-based IDM actuator, in which
a piezoelectric element is replaced with an SMA wire and a bias
spring. This structure would allow one-way movement toward the
inertia body because the SMA wire cannot extend rapidly as men-
tioned in the previous section. Hence, a symmetric structure of
Fig. 2(b) would be available to achieve bidirectional movement by
switching the activation of the two SMA wires although two SMA
wires, bias springs, and actuator drivers are necessary. The struc-
ture shown in Fig. 2(c) is stable in comparison with the structure
of Fig. 2(a). In this structure, it is not necessary to concern about
the instability due to the difference in weight between the main
and inertia bodies but the assembling and parameter tuning would
become more complicated. Fig. 2(d) shows a unique structure that
utilizes the gravity acting on the inertia body, in which the inertia
body is suspended by two SMA wires diagonally extended owing
to the weight of the inertia body. This structure could perform bidi-
rectional movement by controlling the activation of the two SMA
wires. Finally, Fig. 2(e) illustrates a structure based on the SIDM
 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57 49

Main body Spring Inertia body Groove

Screw holes
Base

(a) (b)
SMA wire Gravity 20 11

14.5
20 SMA wire
(c) (d)
Mover 20
(a) Side and top views (unit: mm)

Driving shaft Plastic screw Inertia body

(e) &
washer
Spring
Fig. 2. Possible structures enabling SMA-wire-based IDM actuator. Blue arrows indi- SMA wire
cate expected direction of movement. (For interpretation of the references to color
in this figure legend, the reader is referred to the web version of this article.)

Main body
[6,8] which consists of a movable object and a shaft that is sup-
ported by a spring and horizontally driven by the SMA wire. The
movable object moves horizontally toward the tip of the shaft by Eyelet
terminal Shaft
repeating rapid contraction and gradual extension of the SMA wire.
As shown in Fig. 2, several SMA-wire-based actuators would be fea-
sible. However, in the present study, we designed and developed an
SMA-wire-based IDM actuator based on the simplest structure (a)
since the main focus of present study is to examine the applicability Flexible electric wire Linear bushing
of an SMA wire as a driving source in the IDM.
Fig. 3 shows a prototype IDM actuator composed of a main body, (b) Cross-section view
an inertia body, a coil spring, and a SMA wire. The main (19.2 g) and
inertia (7.9 g) bodies were made of aluminum, and their surfaces Fig. 4. Schematics of side, top, and cross-section views of the prototype actuator.
were anodized for electrical insulation between the bodies and the
SMA wire. In general, SMAs exhibit thermal hysteresis between the
addition, a coil spring with 0.5 N/mm stiffness was applied, preload-
phases of heating and cooling, which is induced by the difference
ing the SMA wire with 1.5 N at room temperature.
in the phase transformations from austenite/martensite to marten-
Fig. 4(a) shows the side and top views of the prototype actua-
site/austenite. In the case of large thermal hysteresis, faster cooling
tor. The size was approximately 20 mm × 20 mm × 40 mm, where
is necessary to rapidly return the SMA wire to the initial length but
the longitudinal length varied depending on the tension of the SMA
it cannot be realized by the air cooling. Hence, the increase in the
wire. A few screw holes on the main body facilitated the adjustment
thermal hysteresis degrades the responsiveness of SMA wire in the
of the wire length. The SMA wire was attached in a U-shape; the
proposed actuator because the air cooling cannot be accelerated.
wire passed in a groove on the inertia body to simplify the assem-
In the present study, a NiTiCu alloy wire (NT-H7-TTR, Furukawa
bly and alignment. A cross-sectional view is illustrated in Fig. 4(b).
Techno Material), which had the smallest hysteresis among avail-
A linear bushing built in the main body and a shaft attached to
able SMA wires, was selected in order to allow the highest possible
the inertia body allowed stable and smooth linear motion of the
responsiveness. The NiTiCu alloy wire contains 42.6% of nickel and
inertia body, keeping the orientation of the inertia body flat. In
7.0% of copper. The phase transformation temperature in each state
addition, the electric wires for power feeding and both ends of the
are As = 74.3, Af = 86.7, Ms = 76.5, and Mf = 63.1 ◦ C under 200 MPa
SMA wire were crimped with eyelet terminals. These wires were
load. The diameter and length were 0.1 mm and 70 mm, respec-
bolted together with plastic screws and washers so as to avoid the
tively, and the resistance was approximately 12  at 20.0 ◦ C. In
electrical leakage.

3.2. Heating method

Rapid heating allows the rapid contraction of an SMA wire,

Fig. 3. Photo and schematic of a prototype actuator.
which is a necessary condition for realizing the IDM. The SMA wire
can be heated up electrically by applying voltage or current; the
applied voltage was mainly controlled in the present study. The
rapid heating can be achieved by applying a high voltage to the SMA
wire, however this involves the possibility of overheating, which
might result in some degradation of the actuation; in the worst
case, the SMA wire could be burn out. Thus, a very short heating
50 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57

Heating time
Voltage

Cooling time

0
Time
One cycle

Fig. 5. Input voltage pattern applied to the SMA wire.

time and sufficient cooling time are needed to avoid overheating.

Additionally, cyclic driving of the SMA wire is required to obtain
continuous step-like movement. Therefore, the input voltage
shown in Fig. 5 could be used for enabling the cycle of rapid heat-
ing and gradual cooling. In this paper, the ratio of heating time and
cooling time is defined as duty cycle. For instance, if 1% duty cycle is
set under 1 Hz frequency, it means heating for 0.01 s and cooling for
0.99 s. Hence, the rapid heating of the SMA wire can be controlled
by the amplitude, duty cycle, and frequency of the input voltage.
Estimating the maximum temperature that does not lead to
overheating of the SMA wire is helpful in designing the proposed
actuator and the input voltage. The amplitude of the input voltage
related to the maximum temperature could be estimated by the
heat transfer model of the SMA wire which consists of electric heat-
ing and natural convection (i.e., air cooling) [20,21]. However, there
are temperature dependent parameters to be considered in the heat
transfer equation, such as the specific heat [22] and the convection
coefficient [23], which are, in general, difficult to theoretically iden-
tify. For further precise simulation, we need to consider different
phase transformation equations between austenite and martensite
because the different phase transformations cause the hysteretic
behavior of SMA wire. Therefore, it is not easy to theoretically esti-
mate the maximum amplitude of the input voltage for safely driving
the proposed actuator. In the following experiments, the maximum
amplitude was empirically determined so as to avoid exceeding the
permissible temperature of the SMA wire.
4. Availability of the proposed IDM actuator
First, a simple driving experiment was conducted to reveal
whether the proposed actuator can achieve the movement toward
the inertia body. This section describes the details and feasibility of
the SMA-wire-based IDM actuator.
4.1. Experimental setup
Fig. 6(a) shows a schematic of the experimental system. Two
laser displacement sensors (sensor: Z4M-S40, amplifier: Z4M-W40,
OMRON Corp.) fixed on the base measured displacements of the
main and inertia bodies. Input voltage variously formed by a wave-
form generator (WF1946, NF Corp.) was applied to the SMA wire
after being amplified via a power amplifier (4015, NF Corp.). These
experimental data were taken and recorded with a digital oscil-
loscope (DSOX-2004A, Agilent Technologies) during the driving
experiment. The main body was placed on a V-shaped base made
of steel, as shown in Fig. 6(b), whose coefficient of static friction
was approximately 0.26.
4.2. Preliminary driving experiment
Fig. 7 plots the displacement of the main body (upper graph),
the relative displacement of the inertia body based on position
of the main body (middle graph), and the waveform of input
Fig. 6. Experimental system for verifying the availability of the SMA-wire-based
IDM actuator.
voltage (lower graph: 1 Hz frequency and 1.0% duty cycle). In the
upper graphs, the positive displacement indicates the movement
of the main body toward the inertia body (expected movement),
whereas the negative displacement represents the movement
in the opposite direction (unexpected movement). The upper
graphs also represent the movement of the whole actuator. In the
middle graphs, the positive value means that the inertia body was
attracted toward the main body. Fig. 7(a) shows that the prototype
actuator performed a step-like movement in the positive direction
when the amplitude of the input voltage was set at 16.0 V. This
result proved that the IDM could be realized by using the SMA wire
in the structure shown in Fig. 2(a). Focusing on the inertia body, its
behavior often became unstable immediately after the beginning
of voltage application but stabilized after a few cycles. This could
be attributed to inhomogeneous heating in the SMA wire in the
initial stage, which can be improved by preheating the SMA wire
before driving. On the other hand, the movement to the opposite
direction could be also obtained when the amplitude was set at
11.8 V, as shown in Fig. 7(b). This unexpected movement could be
produced by the difference between the inertia force of the main
body due to the rapid contraction of the SMA wire and the restoring
force of the coil spring; the restoring force would be larger than
the inertia force when the negative movement was achieved.
Because this unexpected movement was very crucial, the possible
mechanism was further analyzed considering the characteristics
of the prototype actuator. The details will be described in Section 6.
Finally, these results demonstrated that the prototype actua-
tor succeeded in bidirectional movement by simply changing the
amplitude of the input voltage, which is similar to the behavior of
the piezo-based IDM actuator, and the SMA wires can be potentially
used as a driving source in the IDM.
5. Characterization of the prototype IDM actuator
In Section 4, it was found that the proposed actuator has the
ability to achieve bidirectional movement by changing the input
voltage. This result implied that the direction of movement and
the step size could be controlled by modulating the parameters
of the input voltage shown in Fig. 5. This section examines the
locomotion performance of the prototype actuator by variously
changing the amplitude, duty cycle, and frequency of the input
voltage. All the characterizations were conducted by means of the
 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57 51

1.5 0.5
Main body Main body
1.0 0

0.5 -0.5

Displacement [mm]
Displacement [mm]
0 -1.0

-0.5 -1.5
2.0 2.0
Inertia body Inertia body
1.5 1.5
1.0 1.0
0.5 0.5
0 0
-0.5 -0.5
20 20

Voltage [V]
Voltage [V]

10 10

0 0
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
Time [s] Time [s]

(a) Positive movement (at 16.0 V) (b) Negative movement (at 11.8 V)

Fig. 7. Displacements of the prototype actuator when the amplitude of the input voltage was set at 16.0 V and 11.8 V under 1 Hz frequency and 1.0% duty cycle. The upper
and middle graphs plot the displacements of the main body and the relative displacements of the inertia body based on the position of the main body, respectively. In the
upper graphs, the positive displacement means the movement of the main body toward the inertia body, whereas in the middle graphs, it indicates the movement of the
inertia body toward the main body. The input voltage pattern is shown in the lower graphs.

same experimental system as that shown in Fig. 6. The maximum practical application, the amplitude from the maximum negative
amplitude was selected so as not to overheat or burn out the SMA step size to the maximum positive step size would be useful for
wire, which depended on the duty cycle and frequency of the input achieving the bidirectional movement, because this region shows
voltage. a linear increase. Moreover, it was found that the step size changed
according to the amplitude; the maximum step sizes in the pos-
5.1. Amplitude itive and negative directions shown in Fig. 8 were approximately
150 ␮m and 125 ␮m, respectively. These results proved that the
First, the amplitude of the input voltage was variously changed adjustment of the amplitude allowed the bidirectional movement
under 1 Hz frequency and 1.0% duty cycle. Fig. 8 plots the rela- with various step sizes.
tionship between the amplitude of the input voltage and the
step size. As shown in Fig. 8, for the prototype actuator, a zero- 5.2. Duty cycle
crossing point where the actuator did not move was found at
around 13.5 V. Based on this amplitude, the positive movement Second, we investigated how the changes in the duty cycle
was realized when the larger amplitude was applied, whereas the affected the locomotion. Fig. 9 shows the relationship between the
smaller amplitude caused the negative movement. Considering the
0.15
Duty cycle: 1.0%
0.15 0.10 Duty cycle: 0.5%
One step movement
0.10 0.05
Step size [mm]
Step size [mm]

0.05 0

0 -0.05

-0.05 -0.10
-0.10 -0.15
5 10 15 20 25
-0.15 Voltage [V]
6 8 10 12 14 16
Voltage [V] Fig. 9. Effect of the duty cycle of the input voltage on the step size. Blue round
markers and red square markers indicate the step sizes under 1.0% and 0.5% duty
Fig. 8. Effect of the amplitude of the input voltage on the step size. Blue round cycles, respectively. The amplitude of the input voltage was changed from 6.0 V to
markers indicate the step sizes when the amplitude was variously changed under 16.0 (at 1.0%) and 22.0 V (at 0.5%) under 1 Hz frequency. A double-headed arrow
1 Hz frequency and 1.0% duty cycle. Line arts drawn by red line illustrate the profile indicates an example showing that different movements could be realized at the
of one step movement. (For interpretation of the references to color in this figure same amplitude by changing the duty cycle. (For interpretation of the references to
legend, the reader is referred to the web version of this article.) color in this figure legend, the reader is referred to the web version of this article.)
52
0.15
0.10
Step size [mm]
0.05
0 −1.0
-0.05
-0.10
-0.15
0 0.2 0.4
Duty cycle [%]
Fig. 10. Relationship between the duty cycle and the step size when 16.0 V ampli-
tude was applied to the SMA wire under 1 Hz frequency.
amplitude of the input voltage and the step size at different duty
cycles; in the graph, the blue and red plots indicate the representa-
tive step sizes when the duty cycles were set at 1.0% and 0.5% under
1 Hz frequency. As shown in Fig. 9, 0.5% duty cycle also allowed
the bidirectional movement with various step sizes. A compari-
son of the two results shows that the trends in the linear region
were almost identical, but the plots at 0.5% duty cycle were shifted
toward the right side of the plots at 1.0% duty cycle. The interest-
ing finding was that for the same step size, the amounts of joule
heat given to the SMA wire were almost equal for the two condi-
tions. This means that the zero-crossing point can be controlled by
the duty cycle, maintaining the profile and the amount of energy.
The results also revealed that different movement could be realized
even if the same amplitude of the input voltage was applied to the
SMA wire. For example, at around 16.0 V (see the double-headed
arrow in the graph), the movement of the prototype actuator was
positive at 1.0% duty cycle; in contrast, at 0.5% duty cycle, negative
movement was achieved with different step size. Fig. 10 shows
the relationship between the duty cycle and the step size under
16.0 V amplitude and 1 Hz frequency. This result implied that under
a fixed amplitude, the direction of movement and the step size can
be controlled by modulating the duty cycle.
5.3. Frequency
Fig. 11 plots the velocity of the main body when various fre-
quencies were set under 16.0 V amplitude and 1.0% duty cycle;
the positive sign means the direction toward the inertia body. In
0.15
0.10
Direction change
0.05
Velocity [mm/s]
0 movement at around 9 s in Fig. 7, where the movement of the
-0.05
-0.10
-0.15
-0.20
-0.25
-0.30
0 2 4 6
Frequency [Hz]
Fig. 11. Relationship between the frequency and the velocity of the main body when
16.0 V amplitude was applied to the SMA wire under 1.0% duty cycle.
S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57
0.5
Heating start
0
Displacement [mm]
−0.5
20 Hz
6 Hz
−1.5
−2.0
0.6 0.8 1.0 −2.5
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Fig. 12. Displacements of the prototype actuator when the frequency of the input
voltage was set at 6 Hz and 20 Hz under 16 V amplitude and 1.0% duty cycle. Com-
paring the two conditions, it is found that smaller a step size could be realized in
20 Hz frequency although the velocity is almost the same.
the prototype actuator, the positive velocity could be achieved in
the lower frequency, and the negative velocity was maximized
at around 6 Hz. Beyond 6 Hz, the prototype actuator kept almost
the same velocity until about 20 Hz. Then, the velocity gradually
approached zero, and the prototype actuator did not move over
100 Hz because the impact force became smaller than the static
friction force owing to the shortening of the heating and cooling
times. However, the prototype actuator resumed the movement
in the negative direction again at around 200 Hz, which was near
the resonance point of the prototype actuator. Fig. 12 plots the dis-
placements of the prototype actuator at 6 Hz and 20 Hz frequencies.
As the graph shows, it should be noted that the step size changed
due to the frequency but the velocity was almost the same between
the two conditions; a smaller step size could be realized at higher
frequency.
6. Discussion
6.1. Generating mechanism of the negative movement
In Section 4, it was confirmed that the prototype actuator has the
ability to perform the bidirectional movement. This was an unex-
pected result because we assumed that the prototype actuator, in
which the SMA wire can achieve rapid contraction but not rapid
extension, moves only in the positive direction as discussed in Sec-
tion 2. Here, we consider a possible mechanism for generating the
negative movement based on the characteristics of the prototype
actuator.
Focusing on the one-step movement of the main body in
Fig. 7, a positive overshoot was found at every step in both the
cases, which we considered as a key factor for producing the
bidirectional movement. Fig. 13 shows the profile of the one-step
prototype actuator was sufficiently stabilized. It should be noted
that the one-step movement consisted of positive and negative
displacements, and the sign shifted from positive to negative
Decreasement
immediately after the contraction of the SMA wire stopped. The
first positive displacement was caused by impulsive inertia force
due to the rapid contraction of the SMA wire, whereas the second
negative displacement would be due to impulsive restoring force
generated by the coil spring. Even after the contraction of the SMA
8 10 12 14
wire was over, the inertia body still had some velocity toward the
main body, and it would continue to increase the potential energy
of the coil spring, saving the elastic restoring force. At this time,
since the SMA wire began to be extended, i.e., to lose tension by
 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57 53

Fig. 13. Profiles of one step movements in the positive and negative movements. The upper and middle graphs plot the displacements of the main body and the relative
displacement of the inertia body based on the position of the main body, respectively. The input voltage pattern is shown in the lower graph.

cooling due to the air, the elastic restoring force played a dominant
role in the movement of the main body. According to the results
shown in Fig. 13, it is assumed that the restoring force would be
smaller than the inertia force due to the contraction of the SMA
wire in the case of the positive movement, whereas the opposite
relationship could be found in the case of the negative movement.
Hence, the state transition shown in Fig. 14 would be caused when
the prototype actuator moved in the negative direction; the sum
of the displacements in states (2) and (3) finally gives the step size.
These results imply that controlling the balance between the
inertia force and the restoring force allows both positive and
negative movements. This characteristic could be a major advan-
tage compared to the piezo-based IDM actuator with the same
structure because the SMA-wire-based IDM actuator with a coil
spring enables the bidirectional movement with a simple input, as
shown in Fig. 5; the piezo-based IDM actuator requires sawtooth-
waveform input with high voltage (over 100 V) and a bias voltage.

Main body Spring Inertia body SMA wire during cooling

SMA wire during heating 6.2. Characteristics of actuation
(1)
The characteristics of the prototype actuator described in Sec-
tion 5 revealed that the direction of movement and step size of the
Impulsive force by SMA wire prototype actuator could be administrated by the parameters of the
input voltage—amplitude, duty cycle, and frequency. This section
(2) reviews and further discusses the characteristics of actuation due
to these three factors.
The amplitude and duty cycle are related to the amount of Joule
Impulsive force by spring heat given to the SMA wire. Thus, larger amplitude or higher duty
cycle enables faster heating, i.e., rapider contraction of the SMA
(3) wire, and it leads to greater inertia force on the main body. Hence,
movement in the positive direction, which occurs when the iner-
tia force is larger than the restoring force, can be obtained when
these parameters have higher values, as shown in Figs. 8 and 10.
However, the increase in the duty cycle leads to reduced cooling
(4) time, resulting in overheating of the SMA wire. Therefore, the step
size should be basically controlled by using the amplitude of the
Displacement input voltage. On the other hand, the zero-crossing point can be
Displacement Voltage pattern shifted by configuring the duty cycle. This characteristic could be
of main body
useful for designing an actuator driver that can supply lower volt-
Fig. 14. A schematic of generating mechanism of the negative movement in the age. Furthermore, the change of frequency enables bidirectional
prototype actuator. movement. However, the frequency range in which the movement
54 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57

with positive step sizes can be achieved is very narrow compared to

the other parameters. Hence, changing the direction of movement
on the basis of frequency would not be an efficient method.
The step size can be controlled by all the parameters as men-
tioned above. Focusing on the negative movement shown in
Figs. 11 and 12, it should be noted that finer movement can be real-
ized at higher frequencies while maintaining a high velocity. This
characteristic indicates that precise positioning could be achieved
in the negative direction without lowering the moving speed. Fur-
ther, we confirmed that the prototype actuator could move at
around the resonance frequency (approximately 200 Hz in the pro- Fig. 15. A block diagram of position control system based on a PI controller. The
totype actuator), although this paper does not show the result. This output of the PI controller (u) is the amplitude, duty cycle, or frequency of the input
fact implies that a smaller step size with high velocity could be real- voltage.
ized by designing the actuator with the resonance frequency; this
might be useful for further precise position control. Table 1
Parameters for position control.

7. Application as a positioning device Type I Type II Type III

P gain 0.30 0.01 0.40

As mentioned above, it was found that the movement (direc- I gain 10.00 0.05 3.00
tion and step size) of the prototype actuator can be controlled by Offset 1.35 V 0.7% 1.5 Hz
Amplitude Variable 1.6 V 1.6 V
properly modulating the parameters of the input voltage. Here,
Duty cycle 1.0% Variable 1.0%
we discuss position control methods for the proposed actuator Frequency 1 Hz 1 Hz Variable
using the amplitude, duty cycle, and frequency of the input volt-
*Offset in Type I and amplitude are the values at the waveform function.
age. Using the experimental system shown in Fig. 6 with a digital
signal processor (DSP: DS1104, dSPACE Inc.), the characteristics of
each control type were investigated to fundamentally show the to the output of the PI controller. The square wave based on the
possibility of using the proposed actuator as a positioning device. output value that the waveform function programmed in the DSP
generated was applied to the SMA wire after being amplified by
7.1. Position control system the power amplifier (gain: ×10). As a technique for achieving sta-
ble position control, a dead zone (under 10 ␮m in this study) in
Fig. 15 illustrates a schematic block diagram of position con- which the control system did not work was introduced to the PI
trol system. The main idea of this control system is to calculate controller; i.e., when the prototype actuator reached the dead zone,
the proper amplitude, duty cycle, or frequency of the input volt- the actuator was kept at the position until a new target was set. The
age based on the position error between a target position and a sampling time of the DSP was set at 0.05 ms.
measured position of the prototype actuator. The position of the
prototype actuator was measured by the laser displacement sensor, 7.2. Positioning characteristics
and a second-order Butterworth filter with 100 Hz cutoff frequency
was applied to the position error. A PI controller implemented in In the present study, the prototype actuator was controlled to
the DSP calculated the proper parameter to reduce the position move from the reference position (0 mm) to positions of ±0.5 mm
error close to zero; the reason why D control was not applied was away with amplitude-control type (Type I), duty-cycle-control type
that it might not be effective for the step-like movement that is (Type II), and frequency-control type (Type III). The parameters
considered as static movement. An offset amplitude, duty cycle, or used in each control type are listed in Table 1; note that the offset
frequency at which the step size became zero in the linear region in Type I and amplitudes are the values at the waveform function
(for instance, the offset amplitude was 13.5 V in Fig. 8) was added before being amplified. Figs. 16–18 show the representative results

Fig. 16. Step responses in the amplitude-control type (Type I).

 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57 55

Fig. 17. Step responses in the duty-cycle-control type (Type II).

Fig. 18. Step responses in the frequency-control type (Type III).

in each control type. All the control types succeeded in position-
ing the prototype actuator to the target position although in some
cases, the completion of positioning took a relatively long time
because of fluctuation in the displacement.
Focusing on positioning in the negative direction, it was found
that the smaller step size in the higher frequency (as shown
in Section 6.2) allowed less overshoot and stationary error in
the frequency-control type than the other control types. As for
the responsiveness, the frequency-control type showed a similar
response (i.e., time until the prototype actuator reached the target
position first) to the amplitude-control type; however, the response
in the duty-cycle-control type was relatively lower. Hence, the
frequency-control type could be the most effective method for pre-
cisely positioning the prototype actuator in the negative direction.
On the other hand, for positioning in the positive direction, the
three control types showed almost the same responses and step
sizes. Considering the overshoot and the convergence performance,
the amplitude-control type would be slightly better than the other
control types. From these results, a combination of the amplitude-
control type and frequency-control type could allow a fast and
precise positioning of the proposed actuator. In particular, we may
use the amplitude-control type so that the actuator moves beyond
the target position as fast as possible and with smaller over shoot,
and then switch the controller to the frequency-control type to
precisely approach to the target position in a few steps.
8. Conclusion
This paper proposed a linear actuator that utilizes the rapid con-
traction property of an SMA wire based on the principle of impact
drive mechanism (IDM). In particular, a centimeter-scale prototype
IDM actuator was developed using an SMA wire, and its applicabil-
ity and basic performance were examined. A preliminary driving
experiment verified that the prototype actuator can achieve similar
locomotion to the piezo-based IDM actuator and that the piezo-
electric element can be replaced with the SMA wire. The result
also demonstrated that the prototype actuator enabled the move-
ment in the negative direction (toward the main body) beyond our
expectation. Observation of the one step movement revealed that
each movement consisted of positive and negative displacements,
56 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57
which were produced by the inertia force due to the contraction
of the SMA wire and the restoring force of the coil spring, respec-
tively. Thus, the unexpected movement could be produced by the
balance between the first and second displacements. The analysis
implied that the prototype actuator performed the movement in
the negative direction when the restoring force was larger than the
inertia force acting on the main body.
Second, the actuation performance of the prototype actuator
was evaluated with a focus on the parameters of the input voltage:
amplitude, duty cycle, and frequency; the evaluation revealed
how the changes in these parameters affect the locomotion
performance of the prototype actuator. In addition, the position
control was attempted using these three factors. The findings are
summarized as follows:
• Changes in each parameter could produce bidirectional move-
ment with different step sizes under the conditions mentioned
in Section 4.
• Submillimeter step sizes could be realized in both the positive
and negative directions by changing the amplitude and the duty
cycle.
• The band of practically drivable voltage (linear region) could be
shifted by configuring the duty cycle.
• At higher frequencies, much smaller step sizes, i.e., more precise
positioning could be realized in the negative direction.
• The resonance frequency of the proposed actuator could be also
used to realize still smaller step size in the negative direction.
• Position control could be possible by a PI controller using the
amplitude, duty cycle, and frequency of the input voltage as the
output.
Future research will focus on further downsizing of the actuator,
which will be an important challenge in terms of practical applica-
tion [24]. Because the stiffness of the coil spring was not sufficient
to support the inertia body in the present centimeter-scale proto-
type actuator, a linear bush and shaft were embedded in the bodies
to make the actuation smooth and stable. In a smaller scale, the
inertia body will be lighter, and thus, these components could be
removed, which will enable further allow. Additionally, the present
prototype actuator uses electric wires for power feeding, which will
hinder smooth movement in the case of downsized actuator. Thus,
a novel method that will enable wireless power feeding should be
considered in a downsized actuator.
References
[1] Micromotor, Namiki Precision Jewel Co. Ltd., http://www.namiki.net/
(accessed 07.06.13).
[2] K. Uchino, Piezoelectric Actuators and Ultrasonic Motors, Electronic Materials:
Science & Technology, vol. 1, Springer, 1996.
[3] T. Higuchi, Y. Yamagata, K. Furutani, K. Kudoh, Precise positioning mechanism
utilizing rapid deformations of piezoelectric elements, in: Proceedings of IEEE
Micro Electro Mechanical Systems Workshop, 1990, pp. 222–226.
[4] H. Kawakatsu, T. Higuchi, A dual tunneling-unit scanning tunneling micro-
scope, J. Vac. Sci. Technol. A 8 (1) (1990) 319–323.
[5] K. Kudoh, T. Gotoh, K. Sato, Y. Yamagata, K. Furutani, T. Higuchi, Development
of piezo micromanipulator for cell micromanipulation, J. Mamm. Ova Res. 7 (1)
(1990) 7–12 (in Japanese).
[6] R. Yoshida, Y. Okamoto, T. Higuchi, A. Hamamatsu, Development of smooth
impact drive mechanism (SIDM). Proposal of driving mechanism and basic
performance, J. Jpn. Soc. Precis. Eng. 65 (1) (1999) 111–115 (in Japanese).
[7] M. Pozzi, T. King, Piezoelectric actuators in micropositioning, Eng. Sci. Educ. J.
10 (1) (2001) 31–36.
[8] R. Yoshida, Y. Okamoto, H. Okada, Development of smooth impact drive mech-
anism (2nd report): optimization of waveform of driving voltage, J. Jpn. Soc.
Precis. Eng. 68 (4) (2002) 536–541 (in Japanese).
[9] T. Morita, Miniature piezoelectric motors, Sens. Actuators A 103 (3) (2003)
291–300.
[10] T. Higuchi, Y. Hojjat, Application of electromagnetic impulsive force to precise
positioning (1st report): positioning mechanism and fundamental experi-
ments, J. Jpn. Soc. Precis. Eng. 55 (8) (1989) 1426–1431 (in Japanese).
[11] Zu Guang Zhang, T. Ueno, T. Higuchi, Magnetostrictive actuating device utilizing
impact forces coupled with friction forces, IEEE Int. Symp. Ind. Electron. (2010)
464–469.
[12] M.H. Kouklan, Y. Hojjat, T. Higuchi, Air impact drive for positioning by
pulse and continuous air pressure, J. Robot. Mechatron. 11 (4) (1999) 263–
268.
[13] Y. Yamagata, T. Higuchi, N. Nakamura, S. Hamamura, A micro mobile mecha-
nism using thermal expansion and its theoretical analysis. a comparison with
impact drive mechanism using piezoelectric elements, in: Proceedings of IEEE
Workshop on Micro Electro Mechanical Systems, 1994, pp. 142–147.
[14] O. Ohmichi, Y. Yamagata, T. Higuchi, Micro impact drive mechanisms using
optically excited thermal expansion, J. Microelectromech. Syst. 6 (3) (1997)
200–207.
[15] K. Otsuka, C.M. Wayman, Shape Memory Materials, Cambridge University Press,
1999.
[16] R. Kornbluh, R. Pelrine, J. Eckerle, J. Joseph, Electrostrictive polymer artificial
muscle actuators, in: Proceedings of IEEE International Conference on Robotics
and Automation, 1988, pp. 2147–2154.
[17] M. Zupan, M.F. Ashby, N.A. Fleck, Actuator classification and selection –
the development of a database, Adv. Eng. Mater. 4 (12) (2002) 933–
939.
[18] S. Hattori, D. Hwang, T. Higuchi, Linear actuator using rapid deformation of
shape memory alloy wire, in: Proceedings of 13th International Conference on
New Actuators, 2012, pp. 500–503.
[19] C. Mavroidis, Development of advanced actuators using shape memory alloys
and electrorheological fluids, Res. Nondestruct. Eval. 14 (1) (2002) 1–32.
[20] D. Madill, D. Wang, Modeling and L2-stability of a shape memory alloy position
control system, Trans. Control Syst. Technol. 6 (4) (1998) 473–481.
[21] M.H. Elahinia, H. Ashrafiuon, Nonlinear control of a shape memory alloy actu-
ated manipulator, J. Vib. Acoust. 124 (2002) 566–575.
[22] J.L. McNichols Jr., J.S. Cory, Thermodynamics of Nitinol, J. Appl. Phys. 61 (3)
(1987) 972–984.
[23] S.M. Dutta, F.H. Ghorbel, Differential hysteresis modeling of a shape memory
alloy wire actuator, IEEE/ASME Trans. Mechatron. 10 (2) (2005) 189–197.
[24] Y. Iwago, M. Hara, H. Nabe, T. Higuchi, Study on downsizing of impact drive
mechanism actuator using a shape memory alloy wire, in: Proceedings of the
14th International Conference on New Actuators (ACTUATOR2014), 2014, pp.
214–217.
Biographies
Shinya Hattori received the B.E. and M.E. degrees from
The University of Tokyo, Japan, in 2011 and 2013. He is
currently working at Komatsu Ltd., Japan. His research
interests include mechatronics and smart actuators using
shape memory alloys.
Masayuki Hara received the B.E., M.E., and Ph.D degrees
in engineering from Yokohama National University, Japan,
in 2003, 2005, and 2008, respectively. He was a Scien-
tist at the Laboratoire de Systèmes Robotiques (LSRO1),
Microengineering, School of Engineering, École Polytech-
nique Fédérale de Lausanne (EPFL), Switzerland, and was
involved in UT-EPFL Joint Lab Project from 2008 to 2010.
He is currently an Assistant Professor at the Department
of Precision Engineering, School of Engineering, The Uni-
versity of Tokyo, Japan. His research interests include
robotics, haptics, and cognitive neuroscience.
Hiroyuki Nabae, received the B.E. and M.E. degrees from
The University of Tokyo, Japan in 2010 and 2012, respec-
tively. He is currently a Ph.D candidate at the Department
of Precision Engineering, School of Engineering, The Uni-
versity of Tokyo, Japan. He is now mainly studying an
actuator using electromagnetic attractive force as a JSPS
research fellow. His research interests include mechatro-
nics, robotics, and actuators.
 S. Hattori et al. / Sensors and Actuators A 219 (2014) 47–57
Donghyun Hwang received the B.S. and M.S. degrees in
mechanical engineering from Ajou University, Korea, in
2008 and 2010, respectively, and the Ph.D. degree in pre-
cision engineering from The University of Tokyo, Japan,
in 2014. He was a Visiting Researcher at the Depart-
ment of Mechanical Engineering, University of Michigan,
USA, in 2010. He is currently a Post-Doctoral Research
Fellow at the Interaction and Robotics Research Center,
Korea Institute of Science and Technology (KIST), Korea.
His research interests include mechatronics, robotics,
smart actuator, and design and analysis of compliant
mechanism.
57
Toshiro Higuchi received the B.S., M.S., and Ph.D. degrees
in precision machinery engineering from The University
of Tokyo, Japan, in 1972, 1974, and 1977, respectively.
From 1977 to 1978, he was a Lecturer at the Institute of
Industrial Science (IIS), The University of Tokyo, Japan,
and an Associate Professor from 1978 to 1991. Since
1991, he has been a Professor in the Department of Pre-
cision Engineering, The University of Tokyo, Japan. He
was the Leader of the Higuchi Ultimate Mechatronics
Project, Kanagawa Academy of Science and Technology
from 1992 to 1997. His current research interests include
mechatronics, magnetic bearings, electrostatic actuators,
microelectromechanical systems (MEMS), robotics, and
manufacturing.