Proceedings of the IEEE
International Conference on Automation and Logistics
Qingdao, China September 2008
A New PCB-Based Low-Cost Accelerometer
for Human Motion Sensing
Dapeng Qiao, Grantham K.H. Pang
Mui Man Kit, David C.C. Lam
Department of Electrical and Electronic Engineering,
Industrial Automation Research Laboratory,
The University of Hong Kong, Pokfulam Road, Hong Kong
{ dpqiao, gpang }@eee.hku.hk
Department of Mechanical Engineering,
The Hong Kong University of Science and Technology,
Clear Water Bay, Kowloon, Hong Kong
[email protected]
Abstract - This article presents a new low-cost PCB-based
accelerometer. A metal film is adhered to PCB board, which
forms the two electrodes of the sensing capacitor of the
accelerometer. The sensor signal is proportional to the change
of capacitance, and it is obtained by a read-out circuit. This
signal is compared with the output signal of a commercially
available accelerometer, ADXL330, by Analog Devices. The
result shows that the new low-cost accelerometer can fulfill the
requirement of human motion sensing.
In recent years, the cost and size of accelerometers have
been reduced with advances in microelectromechanical
system (MEMS) technology. Other than their recent usage
in the video game industry, wide applications have been
found in Inertial Navigation System (INS), Hard Disk Drive
(HDD) head positioning control system, virtual reality
system and automobile control system. The MEMS silicon
accelerometers can be divided into two categories. In one
category, the acceleration is sensed and converted into
capacitance or resistance value, which is then measured and
output as voltage. Another category would translate the
sensed acceleration into a change of the resonant frequency
in mechanical resonators [3]. The frequency is measured
and output as voltage. Both categories of accelerometers
are based on the mechanical design structures of the
polysilicon device. The parameters such as bias and scale
factor stability changes with temperature,and the
performance of the accelerometer is affected.
When accelerometers are used in toys, remote controls
for video game or applications in human limbs, the
requirements of motion sensing are varied for different parts
of human limbs. A summary of the acceleration and
frequency range of human motions is shown in Fig. 1 [4].
Index Terms - accelerometer, PCB-based, human motion
sensing, read-out circuit.
I. INTRODUCTION
Accelerometers have found many interesting
applications in biomedical engineering, navigation and toys.
For example, the success of Wii, with sophisticated remote
controls for video game, has demonstrated the tremendous
potential for toys in the application of this technology.
Currently, the accelerometer used in the remote control of
Wii is ADXL330, a MEMS silicon accelerometer. Although
its price has already dropped to only a few US dollars in
volume purchase, it is still considered expensive when used
together with the custom-designed chips and interface
electronics associated with the remote control. Hence, a new
generation of low-cost accelerometers that can trigger more
diverse and wide spread applications is needed. To drive
down cost, accelerometer designed and fabricated using low
cost PCB processes and materials are explored..
Wii is the latest video game by Nintendo and it’s game
console is known for its wireless controller, called Wii
Remote [1]. The handheld device is among the few
innovative advanced gaming devices that detect acceleration
in three dimensions. Its distinguishing feature enables the
user to play interactive sport games such as tennis and
badminton against the computer, and brought new levels of
excitement to digital entertainment.
The heart of the Wii Remote is an accelerometer
(ADXL330) by Analog Devices [2], which is used to sense
the player’s hand position in three-dimensions (3D). Ideally,
an accelerometer is a device whose output is directly
proportional to the acceleration it is measuring. Very often,
this fixed ratio of output to input is valid only over for a
small range of operation, and for a small temperature range.
Fig. 1 Acceleration and frequency of different parts of human body
978-1-4244-2503-7/08/$20.00 © 2008 IEEE
56
The following is the characteristics of the different
types of devices:
- Head devices (e.g. head phone): average 3.5Hz
frequency with maximum frequency up to 8Hz;
and tilt less than 60 degree per second.
- Hand, arm, upper-body devices (e.g. tennis racket,
baseball bat): acceleration ranges from 0.5g to 9.0g
with frequency less than 12Hz.
- Hand, wrist, finger devices (e.g. pen, cell phone):
acceleration ranges from 0.04 to 1.0g with
frequency less than 8-12Hz.
- Foot-leg devices (e.g. walking distance measuring
device): acceleration ranges from 0.2g to 6.6g with
frequency less than 12Hz.
The range of acceleration for different human parts
suggests that one single design of the accelerometer may not
effectively meet all specifications. The designs of
accelerometers should be tailor-made for various
applications. One major field of motion sensing in consumer
products is the motion of human hands. Game controllers
and computer remote presenters sense the swinging and
waving motions of human hands. Measurements of human
hand acceleration were carried out by Verplaetse [4]. In the
experiment, hand accelerations of arm or hand swinging
motions were measured and were found to range from 0.49g
to 9.02g.
According to the experimental results, Verplaetse
concluded that most of the acceleration activities were
concentrated near the mean value of 2.2g. The frequency
range of the swinging of the human arm is from 0 to 12Hz.
Thus, to design an accelerometer for sensing human hand
motion, the range of operation must cover the acceleration
of 2.2g; and the frequency should be 12Hz or below.
The accelerometer used for human motion sensing in
game control should be inexpensive and PCB-based
accelerometer can be used as a substitute for costly silicon
accelerometer. Examples of successful micro-systems
demonstrated to be successfully transferred from a siliconbased technology to PCB based technology, include RF
switches, pressure sensors, micro-actuator and micro-fluidic
systems [5]-[8]. The material and processing technologies
was changed from silicon based substrates to organic
substrates, for example, epoxy, polyimide, and Liquid
Crystal Polymer (LCP).
Silicon accelerometers can be categorized into
piezoelectric, capacitive and thermal types. In a
piezoelectric accelerometer, a proof mass is suspended by
piezoelectric-active springs. Upon acceleration, the spring
changes its shape and changes the output electric signal. In
the capacitive type, the movement of the proof mass
changes the gap distance between the two plates of the
capacitor. In thermal accelerometers, temperature sensors
are used to measure the temperature of the gas surrounding
a heater. With acceleration, the gas moves and the change in
gas temperature is detected by the temperature and
correlated with acceleration. In this paper, the performance
of a capacitive PCB-based accelerometer is characterized.
II. PRINCIPLE OF ACCELEROMETER SENSOR
When an accelerometer is fixed to a moving object, the
model of the accelerometer can be simplified and shown as
a schematic diagram in Fig. 2.
Fig. 2 Schematic diagram of a typical accelerometer
The proof mass M experiences an acceleration force from
the object movement and the distance of displacement of the
seismic mass M is y when measured against the universal
reference. The free body diagram of the seismic mass is
shown in Fig. 3.
Fig. 3 Free body diagram of mass
The motion of the seismic mass can be described as,
(1)
My + Dz + Kz = 0
where,
z = yíx
(2)
and K is the spring constant and D is the damping constant.
After substitution, equation (1) becomes,
M ( z + x ) + Dz + Kz = 0
(3)
and
z+
D
K
z +
z = −
x
M
M
(4)
By applying Laplace transform, the second order transfer
function of the seismic mass is,
Z (s)
s2
s2
, (5)
H (s) =
=
= 2
X ( s) s 2 + D s + K s + 2ζωn + ωn 2
M
M
where the natural resonance frequency is,
K
ωn =
(6)
M
and the damping ratio of the system is,
57
ζ =
D
film. When accelerated, the proof mass moves and the gap
distance changes with acceleration, and changes the
capacitance of the sensor. The change in the capacitance
value would be converted to a voltage output by the
interface electronic.
Fig. 4 and Fig. 5 are the front view and the back view of
the prototype. A sensor and circuit that is part of interface
circuit are placed on the PCB board.
(7)
2 KM
The relationship between the output displacement of the
sensor and the input motion is shown in equation (5). The
sensor sensitivity is dependent on the seismic mass M, the
damping effect of the damper D and the stiffness of the
suspension structure K.
The damping of the accelerometer originates from the
squeeze film damping of the gas film underneath the proof
mass. The damping factor D of the gas film is dependent on
the viscosity of the medium, the gap height and the overlap
area between the proof mass and the base. The stiffness of
the suspension structure K is determined by the material
properties and the geometric configuration of the structures.
Under a constant acceleration a, equation (1) becomes,
Ma + Kz = 0
(8)
and,
zstatic
M
(9)
=−
a
K
Equation (9) represents the static sensitivity of the sensor
and the negative sign denotes the opposite moving direction
of the acceleration relative to the mass displacement. From
the equation, the sensitivity of the sensor is directly
proportional to the ratio of the weight of the seismic mass to
the stiffness of the suspension.
To reduce the damping in the accelerometer, one
approach is to reduce the gas pressure inside the package in
order to reduce the viscosity of air. This approach requires
costly advanced packaging with good sealing. An
alternative is to use porous plates to vent the gas underneath
the plate away during compression. In micro-machined
accelerometers, an array of holes is fabricated on the
movable thin-film electrodes to reduce the damping of the
accelerometer.
Fig. 4 The front view of the prototype
III. ACCELEROMETER SENSOR
Fig. 5 The back view of the prototype
The design specification of the PCB accelerometer
sensor is defined based on the human motion sensing
requirements. As most of the activities of the arm are at 2.2g,
this will be identified as the target measurement of the PCB
prototype. Characterization range of 5g with 2.2g at
approximately midway of the full range was used in this
study. The bandwidth of the sensor must cover the
frequency of arm motion, which is 12Hz.
The size of the accelerometer is the other design issue
to be considered. It should be minimized in order to reduce
the footprint occupied by the sensor and the cost of material.
The prototype discussed in this paper is a capacitance type
of accelerometer, which consists of a proof mass, a tap and a
piece of PCB board. A piece of metal film, that was etched
into a center plate with suspension attached to a frame, is
adhesively attached to the PCB board. The metal film and
the etched copper pattern on the PCB board facing the metal
plate forms the two capacitor electrodes separated by a gas
IV. EXPERIMENTS AND ANALYSIS
In the experimental setup, the sensor was connected to
the read-out circuit which produced an analog voltage signal
proportional to the capacitance of the sensor. The source
waveform for the sensor was a 5k Hz square wave which
had a maximum of 0.5V and minimum of -0.5V. This
square wave was produced by a waveform generator, and
DC power supply served the power source of the read-out
circuit. The analog voltage of read-out circuit was read by
oscilloscope.
A 3-axis silicon accelerometer from Analog Device
(ADXL 330) with analog voltage output was used as a
reference device. The ADXL330 has signal conditioned
voltage outputs in a single monolithic IC design. It measures
acceleration with a minimum full-scale range of f3g with a
sensitivity of 300mV/g. It can measure the static
acceleration of gravity in tilt-sensing applications, as well as
58
dynamic acceleration resulting from motion, shock, or
vibration [9]. The experimental setup was as follows:
Fig. 7 shows the output of these two accelerometers
under static acceleration. The waveform at the top is the
output signal of ADXL330, while the lower one is the
output of the PCB-based accelerometer.
From time 0 to 0.6 second, the two accelerometers were
hand held in vertical position. Then, the two devices were
placed in a horizontal up position until 2.5 second was
reached. Afterwards, the two devices were flipped to a
horizontal down position. The waveforms of these two
accelerometers are not straight since it was hand held. It can
also be seen that the waveforms are nearly the same. Both
have a response of ~0.3V for 1g of change, which is close to
the designed specification.
In a second evaluation, the two accelerometers were
shaken suddenly by hand, and their responses to abrupt
shaking are shown in Fig. 8.
PCB-based accelerometer
sensor
read-out circuit
Reference accelerometer
ADXL330
oscilloscope
Evaluation board
Fig.6 Block diagram of the experiment setup
ADXL330 has 3-axes acceleration output: X, Y, and Z.
In this experiment only X-axis was compared with the
output signal of the PCB-based accelerometer which has
only one axis. The bandwidth of ADXL330 was set to 50Hz.
An evaluation board (EVAL-ADXL330Z) provided by
Analog Devices was used to obtain an output voltage from
ADXL330.
To compare the output of these two accelerometers under
identical acceleration, they were affixed onto the same
substrate in the experiment.
Since the prototype was designed for human motion
sensing, thus these two accelerometers were vibrated by
hand to simulate human motion acceleration. The
ADXL330 would give an output voltage at X-axis of around
1.2V at zero g. The X-axis sensitivity varied from 270mV/g
to 330mV/g, and averaged 300mV/g. The PCB-based
accelerometer’s circuit was designed to be comparable with
ADXL330 and gives -0.2V at zero g.
2
Output signal of ADXL330
Voltage [v]
1.5
1
0.5
Output signal of PCB-based acceleromter
0
-0.5
0
0.5
1
1.5
2
2.5
Time [s]
3
3.5
4
4.5
5
Fig. 8 Abrupt shake responses
Three sudden shake motions are shown in the figure. The
first shake occurred at around 1.1 second and the two
accelerometers moved from left to right, and then suddenly
stopped. At about one second later, the two devices went
from right to left, and then suddenly stopped again. At
around 4.5 second, they were shaken again from left to right
with more force by the hand. These motions have been
clearly shown in the figure. Similar waveforms by both the
PCB-based accelerometer and ADXL330 have been
recorded.
The bandwidth of the read-out circuit is designed to be
16Hz, which covers the common frequency of human
motion. Fig. 9 shows the quick shake responses, and again,
the waveform of PCB-accelerometer is essentially identical
to that of ADXL330. The shake’s frequency is about 8Hz,
which is near the maximum of usual arm movement. From
the figure, it can be seen that the PCB accelerometer can
measure rapid arm shake. From the signal of ADXL330, the
peak from the base line is about 0.8V. From the
specification sheet provided, the sensitivity of ADXL330 is
0.3V/g, the maximum acceleration is about 2.7g.
Fig. 7 Static acceleration responses of PCB-based accelerometer compared
with the response of ADXL330
59
accelerometer.
Consequently, the expected ratio of
performance over price will be highly attractive. This will
enable PCB-based accelerometers to be adopted in toys,
biomedical and navigation applications that have low
measure range, low temperature requirements with no strict
limits on structure size.
2
Output signal of ADXL330
1.5
Voltage [v]
1
REFERENCES
0.5
[1] http://en.wikipedia.org/wiki/Wii_Remote.
[2] http://www.analog.com/en/prod/0,2877,ADXL330,00.html.
[3] M. Helsel et al., “A navigation grade micromachined silicon
accelerometerā, Proc. IEEE Position Location and Navigation Symp., Las
Vegas, pp. 51-58, 1994.
[4] C. Verplaetse, “Inertial proprioceptive devices: Self-motion-sensing
toys and tools”, IBM Systems Journal, Vol. 35, Issue 3-4 (1996), pp. 639650.
[5] R. Ramadoss, A. Sundaram, L.M. Feldner, “RF MEMS phase shifters
based on PCB MEMS technology”, Electronics Letters, Volume 41, Issue
11, 26 May 2005, Pages 654-656.
[6] J. N. Palasagaram, R. Ramadoss, “MEMS capacitive pressure sensor
array fabricated using printed circuit processing techniques”, IECON
Proceedings (Industrial Electronics Conference), Volume 2005 (2005), pp.
2357-2362.
[7] E. T. Enikov, K. Lazarov, “PCB-integrated metallic thermal microactuators”, Sensors and Actuators, A: Physical, Volume 105, Issue 1, 15
June 2003, pp. 76-82.
[8] A. Wego, S. Richter, and L. Pagel, “Fluidic microsystems based on
printed circuit board technology”, J. Micromech. Microeng., Vol. 11, Issue
5 (2001), pp. 528-531.
[9]http://www.analog.com/UploadedFiles/Data_Sheets/ADXL330.pdf.
0
-0.5
Output signal of PCB-based acceleromter
-1
0
0.5
1
1.5
2
2.5
Time [s]
3
3.5
4
4.5
5
Fig. 9 Quick shake responses
V. CONCLUSIONS
A low-cost PCB-based accelerometer has been designed
and fabricated. The performance of the PCB based
accelerometer is shown to be comparable to a commercial
accelerometer (ADXL330) in human motion sensing. With
a simple structure and no custom-design chips, we can
expect that this new accelerometer would cost less than one
USD, or around one tenth of the price of silicon
60