Dejan Duvnjak,
Dan Hebb, Alexandra Nadeau, School of Information Technology and
Engineering
Veronica
Wajda, Glen Torontow, Thomas March, Department of Mechanical
Engineering
University
of Ottawa
EXECUTIVE SUMMARY
This report describes a
myoelectric-controlled partial-hand prosthesis design. The
proposed design senses an electromyogram (EMG) signal from the
bicep when contracted. This EMG signal is amplified and
filtered and then sent to a microcontroller. The
microcontroller converts this signal to a pulse width modulated
signal to drive the servomotor; which opens the prosthetic
hand. Each stage is analyzed in detail and the report presents
simulation and experimental results of these stages.
INTRODUCT ION/OBJECTIVE
In today’s world there is great need
for prosthetics. Prosthetics is part of the field of
biomechatronics, which aims to integrate mechanical, electrical
and biological systems. They are typically used to replace
appendages lost to injury, missing from birth, or to replace a
defective body part. Prosthetics are important to improve
amputees’ lifestyles and in order to do so there have been many
studies aimed at allowing control over the prosthetic
appendage. In this report we will look at a myoelectric-controlled
prosthetic hand that opens when the bicep is flexed.
The proposed design can be split into
three parts:
1)
receiving/processing EMG signal from the bicep,
2) The
microcontroller interpreting the EMG signal,
3) The prosthetic
hand.
The report explains the stages
required for reading the EMG signal. These stages are filtering,
amplifying, rectifying and the microcontroller interpreting the
signal in order to the control the prosthetic hand.
ELECTRONIC STAGES/CIRCUITS
A.
The Electrodes
There are three electrodes
used to control the operation of the prosthetic hand. Two are
applied to the bicep and the other is connected to a neutral
part of the body to act as a ground. When the subject flexes
their bicep their brain sends a signal, also known as an action
potential through a series of nerves and excites the muscle
fibers to contract. This action potential has a fixed magnitude
and speed it can travel through the nerve fibers known as axons.
When this action potential travels through the synapse, from
axon to axon, there is a chemical reaction that creates a
changing electric field as it travels and this field can be
detected by a sensor attached to the subject’s skin. Since the
action potential is fixed in magnitude and propagation speed it
is the frequency of impulses that controls the magnitude of the
force your muscles apply. Therefore the electrodes must receive
the impulses but filter out noise from frequencies that are
outside of the possible rates of signaling of human nerve cells.
Even when muscles are inactive they still have a resting
potential which is what the ground electrode is going to detect
and allow us to remove from the signal.
Fig. 1.
Electrodes
B. The Instrumentation
Amplifier
The first stage of the system
is an instrumentation amplifier which receives inputs from the
two electrodes attached to the subject’s skin. The two
electrodes are connected to different parts of the bicep and
will receive impulses of 13-15ms in duration and of voltages
between 20-20000uv. The instrumentation amp has a very high
input impedance and doesn’t require impedance matching which
makes the design simple and efficient. The instrumentation
amplifier is essentially a difference amplifier which means that
it only amplifies the difference between the two electrodes
attached to the bicep which should cancel out noise which would
be equally affecting both inputs and therefore will not be
amplified. This implies that the placement of the electrodes on
the bicep must be far enough apart to have dissimilar signals in
order to get a coherent output from the instrumentation
amplifier. The output of the instrumentation amplifier will be a
signal consisting of the signal we are interested in between
50-500Hz and noise which is spread over the entire spectrum of
frequencies.
Fig.
2. Instrumentation Amplifier
C. The Low Pass Filter
The low pass filter is
connected to the output of the instrumentation amplifier and is
designed to remove frequencies that are above 500Hz. The low
pass filter removes the frequencies above 500Hz because that is
above the maximum signaling rate of the nerves in human muscles.
Therefore any energy in frequencies above 500Hz is noise and
will degrade the overall performance of the system if it is not
removed.
Fig.
3. Low Pass Filter
D. The High Pass filter
The high pass filter is
connected to the output of the low pass filter and filters out
frequencies that are below 50Hz. For the same reasons as with
the low pass filter we filter out any frequencies that are below
50Hz because we know that they are noise. Putting a low pass
filter is series with the high pass filter effectively creates a
band pass filter that only allows frequencies between 50-500Hz
to pass through without attenuation; which is the range of
frequencies human nerves can transmit signals.
Fig. 4. High Pass Filter
E. The Precision Rectifier
The precision rectifier is a
half wave rectifier that is configured so that the op-amp never
goes into saturation due to the diode D1 in parallel with the
resister R11. The output of the filter must be rectified so that
it can be an input to a comparator. The output of the precision
rectifier is a signal that is always positive and consists of a
series of impulses when the bicep is flexed and relatively close
to zero when the muscle is relaxed.
Fig.
5. Rectifier
G. The microcontroller
The microcontroller receives
the output of the rectifier and interprets the impulses and
converts them to a pulse width modulated signal to control the
servomotor. The microcontroller used was the Arduino
Duemilanove which is a cheap and relatively simple to use
microcontroller. The code implemented is provided below.
MECHANICAL
CONFIGURATION
The
prosthesis was designed using a 4-bar linkage to produce
clamping motion. The pressure of the clamp is derived from a
Hitech HS-311 servo motor. The servomotor produces up to a ±90°
rotation based on the programming of the microcontroller.
Inside
the servo motor is a simple potentiometer. The potentiometer, or
variable resistor, will allow the motor to rotate until the
intended position is reached. Once the position is reached the
motor will stop rotating. It will only be prompted to move again
if a new signal is provided by the microcontroller.
The hand, designed in
SolidWorks, was printed to full scale as a template to ease
manufacturing. The length of the follower link is based on the
open and closed position of the hand. A length too long and the
hand would not close; a length too short and the hand would not
open.
Figure 6. Hand as designed in Solidworks
TESTING
SIMULATION RESULTS
The input voltage is shown
in a 5 volts/div scale. Below is the result after each stage to
insure that each is performing as expected. The scale changes
from 5 volts/div to 2 volts/div after the instrumentation
amplifier and so we can clearly see that the input voltage has
been rectified and amplified for valid inputs to our
microcontroller.
A. The input voltage
B. The signal after the
instrumentation amplifier
D. The signal after the band pass filter
E. The signal after the
precision rectifier
EXPERIMENTAL RESULTS
The following
tests were done with a function generator for systematic
testing. The output being tested is the signal going into the
microcontroller.
From the above results, we see
that the band-pass filter is successfully filtering the correct
frequencies.
PROBLEMS
The
first attempt to build the circuit using 741 op-amps did not
work. There were very high levels of noise in the output that
were big enough to render our EMG signal undetectable. The other
sources of significant error were the components used to build
the circuit which receives the EMG signal and amplifies it. This
circuit was made out of standard 741 op-amps which produced a
significant DC offset. The resistors and capacitors were
accurate to +- 20% which also added to the DC offset. This
offset was compensated for by adjusting the supply voltage to
the op-amps which has eliminated much of the offset but lead to
unpredictable outputs. The circuit was rebuilt with a high
quality instrumentation amplifier and high quality op-amps which
reduced the DC offset to zero. The high quality components
produced an output with little or no noise and allowed for
accurate control the hand.
CONCLUSION
Overall
this project was a success and all of the components performed
as expected.
PICTURES
REFERENCES
[1] T. Pan, P. Fan, H. Chiang, R. Chang, and J.
Jiang, “A Myoelectric Controlled Partial-Hand Prosthesis
Project,” in proc. IEEE Transactions on Education, 2004.
[2] E. Widmaier, H. Raff, and K. T. Strang,
Human Physiology; The Mechanisms of Body Function, 9the. New York: McGraw-Hill, 2004.
[3] H. Takeda, N. Tsujiuchi, T. Koizumi, H. Kan,
M. Hirano, Y. Nakamura, “Development of Prosthetic Arm with
Pneumatic Prosthetic Hand and Tendon-Driven Wrist”, IEEE, 2009.
[4] Campbell and Reece, Biology, 6thed.
San Francisco: Benjamin Cummings, 2002.
