U.S. patent number 3,735,425 [Application Number 05/114,262] was granted by the patent office on 1973-05-29 for myoelectrically controlled prothesis.
This patent grant is currently assigned to The United States of America as represented by the Secretary of the Navy. Invention is credited to Charles H. Hoshall, Robert L. Konigsbert, Woodrow Seamone.
United States Patent |
3,735,425 |
Hoshall , et al. |
May 29, 1973 |
**Please see images for:
( Certificate of Correction ) ** |
MYOELECTRICALLY CONTROLLED PROTHESIS
Abstract
A control system that is suitable for use with prosthetic and
orthotic systems utilizing a single site, closed-loop servo system.
The closed-loop, position servo system comprises a sensor unit,
signal amplifier, control unit, and a power pack unit. The terminal
device opens in direct proportion to control signal amplitude as
the muscles contract. The control unit conserves battery power by
minimizing quiescent current when it is not necessary that the
drive motor be powered.
Inventors: |
Hoshall; Charles H.
(Burtonsville, MD), Seamone; Woodrow (Rockville, MD),
Konigsbert; Robert L. (Baltimore, MD) |
Assignee: |
The United States of America as
represented by the Secretary of the Navy (Washington,
DC)
|
Family
ID: |
22354230 |
Appl.
No.: |
05/114,262 |
Filed: |
February 10, 1971 |
Current U.S.
Class: |
623/25;
623/60 |
Current CPC
Class: |
A61B
5/30 (20210101); A61F 2/72 (20130101); A61B
5/389 (20210101) |
Current International
Class: |
A61B
5/0488 (20060101); A61B 5/0402 (20060101); A61F
2/72 (20060101); A61B 5/0428 (20060101); A61F
2/50 (20060101); A61f 001/00 (); A61f 001/06 () |
Field of
Search: |
;3/1-1.2,12-12.7
;128/418,2.6E |
References Cited
[Referenced By]
U.S. Patent Documents
Other References
"A Three-State Myo-Electric Control" by D. S. Dorcas et al., .
"Medical & Biological Engineering," Vol. 4, July 1966, pages
367-370 and e facing page 370. .
"Myo-Electric Control of Powered Prostheses" by A. H. Bottomley,
"The Journal of Bone & Joint Surgery," Vol. 47B, No. 3, 1965,
pp. 411-415. .
"Muscle Voltage Moves Artificial Hand" by G. W. Horn,
"Electronics," Vol. 36, October 11, 1963, pages 34-36. .
"The Control of External Power in Upper-Extremity Rehabilitation,"
National Academy of Sciences - National Research council, "Signal
Processing in a Practical Electro-Myographically Controlled
Prothesis" by A. H. Bottomley, pp. 271-273..
|
Primary Examiner: Gaudet; Richard A.
Assistant Examiner: Frinks; Ronald L.
Claims
We claim:
1. A control system for actuating a prosthesis from myoelectric
potentials, comprising:
a prosthesis;
electrode means for sensing myoelectric potentials developed by the
body, said electrode means being adapted to engage over a single
muscle area of the body;
means actuating said prosthesis in response to said sensed
myoelectric potential;
means for providing a position feedback signal to said actuating
means for identifying the degree of actuation imparted to said
prosthesis; and,
means for delaying in time said position feedback signal.
2. The control system as claimed in claim 1 wherein said electrode
sensing means comprise spaced first and second electrodes each
formed of a metal biologically compatible with the skin and capable
of making a low resistance contact therewith.
3. The control system of claim 2, wherein said means for minimizing
said interference noise signals comprises
a guard ring assembly encircling said sensing electrodes, the
assembly being connected to signal ground and adapted to contact
the skin surface,
a mounting plate adapted to be spaced from the skin surface and
forming the base of the guard ring assembly and having the
electrodes positioned on one side thereof, and
a preamplifier positioned on the mounting plate on the side thereof
opposite the electrodes, the mounting plate serving as a shield
between the preamplifier and the electrodes to prevent electrical
feedback and to minimize direct pickup from stray electric
fields.
4. The control system as claimed in claim 1 and further
comprising:
means for converting said sensed myoelectric potential into a DC
level signal; and
means for providing interfacing between said converting means and
said actuating means.
5. The control system as claimed in claim 4, wherein said
converting means is a detector and said interfacing means is a
buffer amplifier.
6. The control system as recited in claim 5 wherein said means for
providing said position feedback signal comprises a potentiometer
whose wiper arm is mechanically actuated in direct proportion to
the degree of actuation imparted to said prosthesis, the voltage
sensed by said wiper arm being received by said actuating
means.
7. The control system as claimed in claim 6 wherein said actuating
means comprises:
means for determining the error difference signal between said
position feedback signal and said DC level signal;
means for applying electrical energy to said prosthesis only when
said error difference signal is of a predetermined amplitude;
and
means for converting said electrical energy into mechanical energy
for actuating said prosthesis.
8. The control system set forth in claim 7 wherein said means for
determining said error difference signal is a summing amplifier
comprising:
a first input terminal for receiving said DC level signal;
a second input terminal for receiving said position feedback
signal;
means for providing a fixed threshold signal; and
means for obtaining the difference between said DC level signal and
said fixed threshold signal and producing a difference error signal
when the difference signal is of an amplitude greater or less than
said position feedback signal.
9. The control system as claimed in claim 8, wherein said means for
converting said electrical energy into mechanical energy for
actuating said prosthetic device is a DC torque motor.
10. The control system of claim 1 and further comprising;
means for minimizing interference noise signals;
means for amplifying said myoelectric potentials; and
means for attaching said electrode means to the body.
11. A control system for actuating a prosthesis from muscle
myoelectric potentials comprising:
a prosthesis,
means for sensing myoelectric potentials comprising
electrode sensing means adapted to engage over a muscle area of the
body,
means for amplifying said sensed myoelectric potentials,
means for converting amplified myoelectric potentials into a DC
level signal having an amplitude proportional to the intensity of
said generated myoelectric potentials,
means for identifying the exact physical position of said
prosthetic device and for producing therefrom a position feedback
signal,
means for actuating said prosthesis in response to said sensed
myoelectric potentials, said actuation means providing an
electrical signal sufficient to actuate said prosthesis and
including a summing amplifier for producing such electrical
actuation signal when the resultant threshold output amplitude of
said means for converting the amplified myoelectric potentials into
a DC level signal is greater than the amplitude of said position
feedback signal,
means for converting said electrical signal into mechanical energy;
and,
means for producing a control signal effective to actuate said
prosthetic device only when sensed myoelectric potentials are above
a predetermined threshold level, said degree of prosthetic
actuation being proportional to the amplitude of said sensed
myoelectric potentials, said means for producing said control
signal comprising a standard waveform generator having a
predetermined amplitude, said electrical actuation signal being
generated when said output amplitude of said summing amplifier is
greater than said predetermined amplitude of the standard
waveform.
12. The control system as claimed in claim 11, wherein said sensing
means comprises electrode means adapted to a single muscle
site.
13. The control system as claimed in claim 12, wherein said
electrode means comprises first and second electrodes.
14. The control system as claimed in claim 11, wherein said means
for converting said amplified myoelectric potential into a DC level
signal is a DC detector.
15. The control system as claimed in claim 11, wherein said means
for converting said electrical signal into mechanical energy
comprises a DC torque motor.
16. The control system as claimed in claim 15, and further
including gear means between said DC torque motor and said
prosthesis, actuation of said DC torque motor causing said
prosthesis to operate at a speed determined by the final ratio of
said gear means.
17. The control system as claimed in claim 16, wherein said means
for producing said position feedback signal comprises a
potentiometer having a wiper arm, said wiper arm being operably
connected to the output of said gear means.
18. The control system as claimed in claim 15, and further
including means for suppressing spurious responses generated by the
actuation and subsequent non-actuation of said DC torque motor.
19. The control system as claimed in claim 18, wherein said
spurious response suppression means comprises a plurality of
serially connected zener diodes all connected in parallel with said
DC torque motor.
20. The control system as claimed in claim 11, wherein said means
for generating said standard waveform is a triangular wave
generator.
21. The control system as claimed in claim 20, wherein said means
for generating said electrical actuation signal is a pulse width
modulator.
22. The control system as claimed in claim 11, and further
including means for delaying in time said position feedback signal.
Description
BACKGROUND OF THE INVENTION
This invention relates to a myoelectric control system for
prosthetic or terminal devices. More particularly, the subject
invention relates to a closed-loop myoelectric control system that
utilizes only one control site for proportionate mode control of an
externally powered prosthetic device.
When muscles contract they produce minute electrical potentials.
The potentials produced by the voluntarily controlled muscles of an
amputee are ideal control signal for prostheses. A prosthesis that
responds to muscle contraction in the same way that the replaced
part of the body responded could be used "naturally" with a minimum
of retraining.
Muscle emf's are known as myoelectric potentials. Electrical
conduction of these potentials through body tissue and fluids
results in potential differences that can be sensed on the skin.
Myoelectric potentials measured on the skin are much attenuated
relative to the amplitude of the signals at their point of origin
in the muscle. They are composite signals from many muscle fibers.
Surface electrodes in contact with (but not penetrating) the skin
are used for prosthesis control, despite the weak intensity of the
signal, because of the formidable problems encountered in the use
of percutaneous electrodes for any length of time.
The myoelectric signals acquired on the skin cannot be precisely
described because they are affected by many factors. Among these
are: (1) muscle type, function, and condition (including fatigue);
(2) characteristics of the tissue, bone and skin that lie between
the muscle and the electrodes; (3) electrode material, surface
texture, geometry, and spacing; and (4) the location of the
electrode relative to the muscle. However, some characteristics of
the myoelectric signal acquired on the skin are typical. These are:
(1) the signal is an AC voltage that is roughly proportional (in
amplitude) to the force developed by the muscles that generate it;
and (2) the power spectrum is such that the major portion of the
power lies between 30 and 500 Hz. Signal amplitudes on the order of
100 microvolts rms are typical for healthy muscles developing
modest tension. Paralyzed muscles often produce myoelectric
voltages, but their amplitude is usually much lower than for
healthy muscles. Some prostheses that are controlled by myoelectric
potentials are unsatisfactory because of difficulties encountered
in obtaining signals that are both sufficiently large in amplitude
and relatively free of noise and "crosstalk." Crosstalk results
when unwanted signals produce by antagonist (and other) muscles are
sensed along with the desired signals.
Since myoelectric potentials were first used to control prosthetic
devices, many types of systems and control mechanisms have been
developed. In most prior art systems the electrode structure
typically contains two stainless steel electrodes which make
contact with the surface of the skin over the muscle site or, more
commonly, sites; tensing of the muscle causes the generation of an
electrical signal -- the emg (electro-myographic) signal -- which
is AC in form with major frequency components in the 30 to 500 Hz
spectrum. The electrical signal sensed by the electrodes is fairly
low in level, generally in the 10 .mu.v rms to 1,000 .mu.v rms
range, depending on the degree of tension in the muscle and other
factors, and hence requires amplification if it is to be useful in
driving the synthetically powered device. However, such
amplification is usually beset with noise problems because of the
low level of the signal. The electric signal processing circuitry
must be designed to minimize these problems and provide a useful DC
signal, largely free of noise, for the externally powered
prosthetic device. Any particular signal level may be generated by
tensing the muscle to the desired degree. Typically, a small, high
speed DC motor is geared down to drive a mechanical linkage that,
in turn, operates the moving parts of the prosthesis. For a
self-contained drive mechanism to be satisfactory for a wide range
of applications it must fit within the space envelope of the
prosthetic device and must satisfy such diverse criteria as
moderately high opening and closing velocities and must be light,
silent in operation, and low in power consumption.
SUMMARY OF THE INVENTION
The major advantages of the control system of the subject invention
accrue from its utilization of only one myoelectric control site
and the associated simplicity of control. In these applications the
rest position, i.e., terminal device closed, corresponds to minimum
control signal voltage, i.e., the muscle is relaxed. The terminal
device opens in direct proportion to control signal amplitude as
the muscles contract. The technique of utilizing a prosthetic
control system in a proportional mode using only one control site
effectively eliminates difficulties with electrical crosstalk. Also
involved with the subject invention is a remote power pack concept
which obviates the requirement for packaging the motor and drive
system within the prosthesis. This concept also allows additional
flexibility in the selection of components and in the mechanical
design of the drive and control mechanisms and can also be used for
elbow and wrist rotation.
The myoelectric signal is acquired by surface electrodes that are
held in intimate contact with any muscle capable of producing
suitable myoelectric signals. The signals sensed by the electrodes
are amplified and detected. The output of the detector is applied
to a control unit that controls an electric drive motor which
operates the prosthetic device. With the muscles relaxed and
minimum myoelectric signal generated, the prosthesis is in its
closed position. However, when the muscles begin to contract, the
electrodes pick up an increasing signal intensity which causes the
control unit to drive the motor, consequently opening the
prosthesis. The prosthesis, in this case a hand, begins to open
until a feedback voltage proportionate to the hand opening position
is equal to the control signal. In this manner, the hand is servo
controlled for all positions between fully closed and fully
open.
An object of the present invention is to provide a myoelectrically
controlled prosthesis that utilizes a servo-mechanism control
system.
Another object is to provide a prosthesis that is controlled by a
closed loop servo system.
Still another object is to provide a prosthetic servo control
system which requires only a single site sensor assembly.
A further object of the invention is the provision of a servo
control system that can be utilized for both terminal devices and
above-elbow prosthetic devices.
Still yet another object is to provide a prosthetic control system
that is lightweight, sensitive, and reliable.
Yet another object of the present invention resides in the
provision of a prosthetic control system that operates in a
proportional mode.
A still further object is to provide a control system for use with
a prosthetic device that is operated by a small, self-contained
electric motor.
Another object of the present invention is to provide a prosthetic
control system that has power control circuits which conserve
battery power.
And a further object is to provide a unique prosthetic control
system that can be utilized with conventional prosthetic
devices.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING
FIG. 1 is a basic block diagram of the invention.
FIG. 2 is a perspective showing the electrode sensors and their
associated support apparatus.
FIG. 3 is a chart showing myoelectric signal waveforms generated by
neural stimulation of a muscle.
FIG. 4 is a detail perspective showing the physical arrangement of
the preamplifier.
FIG. 5 is a perspective showing the preamplifier employed, with the
shield cover in place.
FIG. 6 is a simplified diagram illustrating noise pickup resulting
from body capacitances.
FIG. 7 is a simplified diagram illustrating the effects of skin
impedances in the sensor region.
FIG. 8 is a circuit schematic of the preamplifier, detector and
buffer amplifier circuits.
FIG. 9 is a chart showing preamplifier and overall gain
characteristics.
FIG. 10 is a chart showing detector gain characteristics.
FIG. 11 is a schematic of the control unit.
FIG. 12a through 12g are charts showing the input and output
waveforms of the pulse width modulator.
FIG. 13 is a schematic of the power unit.
FIG. 14 is a cross-section of the power unit.
FIG. 15 shows an arrangement of the subject invention physically
located within an above-the-elbow prosthetic device.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
A block diagram of the control system is shown in FIG. 1. The
system is a closed-loop position servo with position feedback that
follows the myoelectric signal generated by the associated muscles.
The myoelectric signal is acquired by a pair of electrodes 1 and 2
that are held in intimate contact with the arm over the associated
muscle. The electrodes are spaced approximately 1 inch apart and
are placed close to the flexor or extensor muscles (not shown) that
control the fingers or anywhere else on the body where suitable
signals exist. The myoelectric signal sensed by electrodes 1 and 2
is applied at 3a and 3b to a preamplifier 4 wherein said
myoelectric signal is amplified and applied to a detector and
buffer amplifier unit 5. Output signal 6 of unit 5 is a DC control
signal that is approximately proportional in amplitude to the
amplitude of the sensed myoelectric signal. The output control
signal 6 of detector 5 is applied to a control unit 12. In order to
minimize electrical power consumption a pulse-width modulation
system is used to control the output torque of a D.C. torque motor
8 which actuates a prosthetic device 9. More specifically, a
pulse-width modulator 10 utilizes the output of a triangle wave
generator 11 and the output 13 of a summing amplifier 7 to provide
a pulse-width modulated signal which controls motor current. The
D.C. torque motor 8 is driven in one direction only. The output 13
of summing amplifier 7 corresponds to the difference between the
amplitude of control signal 6 and the position of the control cable
14 as represented by shaped position signal 15 of lead-lag circuit
16. Since output 13 of summing amplifier 7 reflects the above
mentioned difference it will hereinafter be referred to as error
signal 13. If the amplitude of error signal 13 is small, current in
motor 8 flows for a small part of the duration of the output of
triangular wave generator 11. Accordingly, the "on" time of the
motor is a function of the magnitude of the error signal 13. By
operating a power switching transistor 17 in a power switching mode
and causing motor 8 to actuate only when the output of lead-lag
circuit 16 is less than the control signal 6, relatively little
power is dissipated. When electrodes 1 and 2 are not sensing a
myoelectric signal, standby power consumption in the electronic
components is quite low, i.e., less than 300 milliwatts. No
mechanical switches or special power cutoff relays or circuits are
required to switch from "standby" to "operate" condition.
A potentiometer 18 provides a shaped position signal 15 via
lead-lag circuit 16 to summing amplifier 7. This feedback
arrangement provides high gain at low frequencies and less gain as
frequency is increased. Such signal processing makes the opening of
the terminal device very easy to control at all elbow flexion
positions with or without an object in the terminal device, i.e.,
the hand. This control technique facilitates a simple interface
between the amputee and his prosthesis. The amputee need generate
only one signal when he desires to open the hand, and when the
control muscle is relaxed the hand is automatically closed by
rubber bands and maintains a grasp force without any further effort
or attention on the part of the amputee. When the amputee wishes to
disengage the hand, he contracts the control muscle. When the
command signal provides the voltage needed for the hand opening,
the fingers open and the object is freed.
Referring now to FIG. 2, there is shown a sensor assembly 20 in
conjunction with an arm-band support apparatus 21. The sensor
apparatus consists of an aluminum (or any suitable metal) mounting
plate 22, a guard ring assembly 23, and two stainless steel
electrodes 1 and 2. In this particular embodiment the electrodes
are made of stainless steel but they can be made of some other
material, e.g., silver, gold, platinum, or silver-silver chloride.
(The sensor electrodes in FIG. 2 are shown positioned close
together for purposes of illustration only. Actually they are
spaced further apart.) As was previously mentioned, the sensor
electrodes 1 and 2 make contact with the skin over the muscle at
the control site. Neural stimulation of the associated muscle
causes myoelectric signals to be generated. As was previously
mentioned and is shown in FIG. 3, the myoelectric signals are AC
signals in which major frequency components lie in the 30 to 500 Hz
spectrum. Waveform a of FIG. 3 shows the myoelectric output when
the muscle is relaxed. When the muscle is in low tension, a
waveform such as that illustrated by waveform b of FIG. 3 is
generated. Waveform c of FIG. 3 shows a typical myoelectric signal
produced when the muscle is in high tension. The myoelectric signal
to be sensed is within the 10-1,000 .mu.v rms range of amplitude
values depending upon, among other factors, the tension of the
associated muscle.
Amplification is required if the myoelectric signals are to be used
to control an externally powered prosthesis. Amplifiers must be
carefully designed to minimize noise relative to the low amplitude
of the myoelectric signal. The subject invention was designed to
minimize these problems and to provide a useful demodulated signal
that is largely free of noise. The physical structure of
preamplifier 4 of FIG. 1 is shown in FIG. 4. Preamplifier 4 is
physically located on the reverse side of the sensor assembly 20 of
FIG. 2. Referring back to FIG. 2, the preamplifier circuit 4 is
mounted on the aluminum mounting plate 22 which forms the base for
the guard ring 23 and the surface electrodes 1 and 2. The aluminum
plate 22 serves as a shield between the preamplifier circuit 4 and
said surface electrodes 1 and 2 and thus prevents unwanted
electrical feedback. It also serves to minimize any direct pickup
from stray 60 Hz electric fields. FIG. 5 shows the entire sensor
assembly with a shield 24 covering the preamplifier circuit.
The analysis of the effects of 60 Hz noise pickup is a complex
3-dimensional problem. It involves distributed capacitances between
the noise source and the body surface area, leakage along the skin
(surface) and thru the internal tissues (sub-surface) of the body,
and distributed capacitances between the body surface area and the
external environment power ground to which the noise source is
returned. To simplify the problem and attempt to indicate how noise
pickup at the input of the preamplifier 4 can occur, a greatly
simplified model describing the noise problem is depicted in FIG.
6. In this figure, the distributed capacitances have been replaced
by lumped capacitances to major parts of the body. Noise currents
generated by noise source 30 (usually a 60 Hz source) may enter the
body parts by means of the stray lumped capacitances C.sub.N1
through C.sub.N5 shown at 31 through 35 respectively. There may
even be more than one noise source, but for simplicity only one
such source is shown at 30. Return of the surface and sub-surface
currents to a power ground 37 is through the stray lumped
capacitances C.sub.G1 through C.sub.G5 shown at 38 through 42
respectively. Because of the potential differences set up between
various parts of the body--due to the capacitance voltage dividers
being unlike--noise currents will flow on the skin surface and in
the sub-surface (internal tissues) from one body part node or
circuit node (i.e., from any of the nodes 43 through 47) to
another.
The sensor assembly 20 of FIG. 2 was chosen because of the
shielding of the two probes, against 60 Hz electric fields,
provided by the guard ring 23 and the aluminum plate 22. The guard
ring 23 when connected to signal ground 48 tends to establish an
equipotential region (at signal ground potential) on the surface of
the skin surrounding the two electrodes. As such, it tends to
syphon off surface 60 Hz noise currents to signal ground without
permitting them to develop a potential difference between the
electrodes. If this equipotential region were to extend into the
internal tissue region just below the skin surface, any internal
tissue currents flowing as a result of 60 Hz noise fields would
tend to be syphoned off to signal ground without permitting them to
develop a potential difference (which would be sensed by
electrodes) in the tissue underneath the electrodes. Because the
guard ring contacts only the skin surface, the degree to which the
equipotential region can extend itself into the internal tissue
region will depend on the skin surface to internal tissue
conductivity under the guard ring. Initially this conductivity can
be increased by moistening the skin in the contact area.
Fortunately, this conductivity should improve and maintain itself
with increased time of contact because of the likelihood of
perspiration and insensible water penetrating the skin underneath
the guard ring. Nevertheless, there will be some finite
conductivity between the surface skin and internal tissues and, as
a consequence, the internal tissue region underneath the guard ring
and probes will not be an equipotential region in general. Hence
some 60 Hz noise voltages will very likely develop in this region
and be sensed by the probes.
Since the problem is 3-dimensional, internal tissue currents may
flow from any direction underneath the probes. It is desirable,
therefore, to minimize these currents by establishing an
equipotential surface on the surface of the skin in all regions
near the electrode structure. For this purpose, the arm band 21
which encircles the arm, supports the electrode structure and holds
it in contact with the skin, should be metallic and tied to signal
ground.
FIG. 7 depicts the noise situation in the region of the electrode
structure. In FIG. 7, internal body capacitances between various
skin and tissue layers have been omitted for simplicity. Three
levels within and on the body are indicated: (1) skin surface; (2)
internal tissue level; (3) muscle bundle level. The body
resistances associated with each of these regions and between the
regions are also shown. The potential differences between the noise
sources developed at nodes 50, 51 and 52 cause currents to flow
through the various body resistances (i.e., through skin surface,
skin surface to internal tissue, and internal tissue resistances).
If any current is permitted to flow through internal tissue
resistance 53, an undesirable noise potential will be developed and
sensed by the probes unless an equal and opposite potential is
developed in the skin surface-to-internal tissue resistances 54 and
55--a very unlikely event in view of the highly variable nature of
skin and tissue resistances. This noise voltage could develop even
if the common node impedances 56 and 57 were infinite because
current may flow through resistance 58 (and develop a noie voltage
across the electrodes by virtue of current flow between nodes 50
and 51.
In the absence of a guard ring returned to signal ground, noise
currents would flow through resistances 59, 60, 58, 61, 62, and 63,
64, 53, 65, and 66 in FIG. 7. Resistances 59 through 62 represent
surface resistances along the skin between the respective nodes.
Resistors 53 and 63 through 66 represent internal tissue
resistances (i.e., subcutaneous tissues). Thus a noise potential
difference could easily be set up between nodes 1a and 2a (the
electrode contact nodes) and would be amplified by the preamplifier
undesirably. (Those resistors not identified represent inter-region
resistances).
If, however, a guard ring is installed (with connection to signal
ground) around the probes contacting the skin surface at nodes 67
and 68, an equipotential surface will tend to be set up at least on
the skin surface between nodes 67 and 68 between which are located
electrode nodes 1a and 2a. Skin surface currents normally flowing
through resistors 59 and 62 to resistors 58, 60 and 61 would tend
to be syphoned away from said resistors 58, 60 and 61 and would
flow into the guard ring (which is equal to signal ground), to node
52 (which is also equal to signal ground) and back to power ground
through stray lumped capacitance 69. These skin surface currents
would therefore not be allowed to flow through resistor 58 and
consequently no noise potential difference would be developed
across the electrodes. If the skin surface to internal tissue
resistances 70 and 71 were low, the equipotential surface would in
effect extend into the internal tissue region, thereby syphoning
currents that flow through the internal tissue resistances 63 and
66 to the guard ring and signal ground and away from internal
tissue resistances 53, 64, and 65. The degree to which this
syphoning action occurs will depend on how low resistors 70 and 71
can be maintained. With the guard ring and the "extended guard
ring" (i.e., the metallic arm bands) installed, adequate noise
rejection is possible even when no design precautions are taken to:
(1) balance the impedance in both input arms (resistances 72 and
73) of the preamplifier; and (2) insure that common node impedances
56 and 57 are high. Note that use of the guard ring does not reduce
the proportion of muscle bundle voltage (E.sub.1) reaching the
preamplifier inputs except for the increased loading caused by
resistors 60, 61, 64 and 65 which are, in effect, returned to
signal ground at nodes 67, 68, 74, and 75 (assuming resistances 70
and 71 are very small).
The effective source impedance of the muscle bundle voltage seen by
the electrodes can be as high as 1 megohm with poor electrode
contact and decreasing down to approximately 10K ohms with improved
contact. To minimize the attenuation of the portion of muscle
bundle voltage appearing at the electrodes, due to loading by the
differential input impedance seen looking into the preamplifier at
the electrodes nodes 1a and 2a, the latter should be made as high
as possible, preferably exceeding 200K ohms.
Referring now to FIG. 8, there is shown a detailed schematic
diagram of the preamplifier 4, and detector and buffer amplifier
unit 5.
Based on the expected emg signal level in the 10 to 1,000 .mu.v rms
range, the preamplifier differential gain should be in the 2,500 to
5,000 range so that its output may be detected at a useful level
where additional noise pickup at the output will be of little
consequence. The useful emg signal frequency spectrum has been
stated to be in the 30 to 500 Hz range with much of the energy
concentrated in the 50 to 150 Hz range. Some discrimination against
60 Hz noise can be built into the preamplifier by designing it to
have a band-pass characteristic with its center frequency in the
150 to 250 Hz range.
The preamplifier 4 is a bandpass amplifier [which uses a National
Semiconductor Type LM301A Monolithic DC Operational Amplifier
(DCOA)] having a nominal gain of about 4,350 at a center frequency
of about 150 Hz. The common mode impedance will be determined
essentially by the impedances external to the monolithic amplifier
which return to signal ground from each of the differential inputs;
the internal amplifier impedances leading from the differential
inputs to signal ground are usually much higher than these external
impedances and may be neglected by comparison (since they parallel
the external impedances).
With regard to equivalent input noise voltage of the preamplifier,
when working from high source impedances (of the order of 500K
ohms), this noise should preferably not exceed 5 .mu.v rms in the
30 to 500 Hz frequency spectrum. If held to this limit, the noise
will be insignificant with respect to normal emg signals. DC
currents will flow from the guard ring (which is at signal ground
potential) through the skin and body to the amplifier input circuit
by means of the probes. If the amplifier input DC bias currents are
I.sub.1 and I.sub.2 and the input offset voltage is E.sub.M, it can
be shown that the skin and body currents flowing into the probes to
the negative (inverted) and positive (non-inverted) preamplifier
inputs shown at 3a and 3b respectively are I.sub.a and I.sub.b,
where
I.sub.a = [R.sub.2 /(R.sub.1 + R.sub.2)] I.sub.2 + E.sub.M /R.sub.1
(1)
i.sub.b = [R.sub.2 /(R.sub.1 + R.sub.2)] I.sub.2, (2)
where I.sub.2 is the amplifier non-inverted input DC bias current
and R.sub.1 is the value the DC input resistor 76, and R.sub.2 is
the combined value of resistors 77a and 77b. The transfer function
G(s) for a differential input signal has the following general
form:
G(s) = -K (.tau..sub.1 s + 1)/(.tau..sub.2.sup.2 s.sup.2 +
2.zeta..tau..sub.2 s + 1) where .tau..sub.1 <
<.tau..sub.2
(3)
.tau..sub.1 is a function of the feedback network elements, while
.tau..sub.2 and .zeta. are functions of the input and feedback
network elements and the amplifier open circuit DC gain and its
first lag corner frequency.
Maximum differential gain in Equation (3) occurs approximately at
the frequency where .omega. = 1/.tau..sub.2 and this gain is:
.vertline.G(j.omega.).sub. max .vertline. = K.tau..sub.1
/2.zeta..tau..sub.2 with .omega. = 1/.tau..sub.2 (4)
Assuming an amplifier open circuit DC gain of 50,000 and a first
lag corner frequency of 20 Hz, the transfer function for the
preamplifier network elements is: ##SPC1##
The maximum gain magnitude is from Equation (4):
.vertline.G.sub.max .vertline. = 4,350, occurring at .omega. =
1/.tau..sub.2 = 1/1.062(10.sup..sup.-3) = 942 rad/sec or 150 Hz
If worst case tolerances for the amplifier open circuit DC gain
(10,000) and first lag corner frequency (10 Hz) are applied to
Equations (3) and (4), we would obtain: ##SPC2##
and .vertline.G.sub.max .vertline. = 2,730 occurring at f = 47.5
Hz.
For comparison purposes, the gain magnitude as a function of
frequency for Equation (5) (for DCOA open circuit gain of 50,000
and first lag corner frequency of 20 Hz) is plotted in FIG. 9. Also
plotted is the overall gain of the preamplifier 4 and detector and
buffer amplifier unit 5.
Because different subjects will have different emg potential
sensitivities, a gain control, potentiometer 78 is added to the
preamplifier output circuit to permit adjustment of the signal
level reaching the detector input.
The emg detector and buffer amplifier unit schematic is generally
shown at 5 in FIG. 8. The detector is an absolute value type with a
modification to make it frequency sensitive. Detection polarity is
determined by the direction in which the diodes 79 and 80 are
installed. To produce a positive DC rectified signal at the buffer
amplifier output 81, the detector output voltage at the terminal
82, fed to the inverted input 83 of the buffer amplifier DCOA 84
must be negative with respect to buffer amplifier non-inverted
input 85, since the buffer amplifier produces a sign inversion. The
negative potential is produced by installing the diodes as shown.
If the diodes are inverted, the detector and buffer amplifier
outputs will be inverted. The function of the buffer amplifier is
to provide isolation between the detector output and the load to be
driven.
By adding capacitors 86 and 87 the detector is made frequency
sensitive; its output will be essentially proportional to frequency
for frequencies considerably below and above 60 Hz. The gain of the
detector, i.e., the ratio of average value of the rectified DC
output (E.sub.o) appearing at the buffer amplifier differential
inputs to the RMS signal input (E.sub.1) appearing at the 10K gain
adjustment potentiometer wiper, is as follows:
E.sub.o /E.sub.1 = 4 .sqroot.2 (R.sub.F /R.sub.1) (f R.sub.F C)
(VDC/VRMS) (7)
subject to the restriction that:
f > > 1/4R.sub.F C (8)
where:
E.sub.o = Average value of rectified DC output voltage (i.e.,
differential voltage measured across buffer amplifier input
resistors)
E.sub.1 = RMS voltage at detector input
R.sub.F = Feedback resistance (resistor 92)
R.sub.1 = Input resistance connected to DCOA inverted input
(resistor 93)
C = Capacitance in detector output circuit (capacitors 86 or
87)
f = Frequency in Hz.
At frequencies much lower than f = 1/4R.sub.F C, the gain is
essentially:
E.sub.o /E.sub.1 = 2 .sqroot.2/.pi. (R.sub.F /R.sub.1) (VDC/VRMS)
for f < < 1/4R.sub.F C (9)
at DC, the gain is simply:
E.sub.o /E.sub.1 = R.sub.F /R.sub.1 (VDC/VDC) (10)
a plot of the detector gain as a function of frequency is given in
FIG. 10. Also shown for comparison purposes is a plot of the
theoretical gain given by Equation (7) which holds for frequencies
essentially above 30 Hz. Gain differences between the theoretical
and observed functions may be attributed in part to component
tolerances, particularly to the tolerance on capacitors 86 and 87,
and to the loading effect on the detector output caused by the
buffer amplifier input resistances 88 and 89.
As the frequency increases, E.sub.o would increase for a constant
E.sub.1 input, according to Equation (7) until either amplifier
saturation occurred or the open loop gain roll off (of the DCOA)
with frequency caused the output to fall off. To insure linear
operation and since the emg frequency spectrum of interest lies
between 30 and 500 Hz, an effective high frequency lag corner in
the gain characteristic was introduced by inserting resistors 90
and 91 in series with capacitors 86 and 87.
Referring now to FIG. 11, there is shown a schematic diagram of the
control unit 12, which receives from the buffer amplifier the
sensed emg signal 6 and passes this signal to summing amplifier 7
and then to pulse width modulator 10. The output of the pulse width
modulator is subsequently applied to the power unit (not shown)
which ultimately actuates the prosthetic device (the hand). In
order to minimize electrical power consumption within the power
units a pulse width modulation system is employed. The pulse width
modulator 10 utilizes the outputs 102 and 101, respectively, of the
triangle wave generator 11 and the output of servo amplifier 120 to
provide a pulse-width modulated signal 103 which regulates motor
current.
In order to more clearly describe the operation of the control
unit, any discussion of the lead-lag circuit 16 and the associated
feedback loops will be temporarily omitted. The output signal 13 of
summing amplifier 7 represents a DC signal that is proportional to
muscle tension. This DC signal is applied to input 104 of pulse
width modulator 10. Also, the output 102 of triangle wave generator
11 is received at input 104. The operation of the pulse width
modulator 10 can best be explained by additional reference to FIG.
12. The DC output 13 of servo amplifier 120 is shown by waveform a
of FIG. 12. In this condition, there is no sensed emg signal being
received from the buffer amplifier 6. The DC output signal as shown
in waveform a) is obtained from a potentiometer 113 and applied to
input 105 of operational amplifier 106. The output, E.DELTA., from
output terminal 99 of triangle wave generator 11, shown by waveform
b of FIG. 12, is applied to negative input 104 of operational
amplifier 106. Waveforms a and b are combined within amplifier 106
to produce the summed waveform c. As long as the summed waveform is
less than zero volts, the output of the pulse width modulator 10 is
as illustrated by waveform d which represents the control unit in
the "off" condition. Thus, the prosthesis does not respond since
there is no effective enabling signal. When an emg signal is
sensed, there is applied to input 119 of summing amplfier 120 a
positive DC signal. When combined with the threshold level signal
as produced by potentiometer 113, the signal as represented by
waveform e is thus applied to input 104 of amplifier 106. When
waveform e is combined with the E.DELTA. - waveform f), the
resultant signal as shown by waveform g is produced. Whenever the
upper peaks of waveform g exceed zero volts, the pulse width
modulator 10 shifts from the non-enabling E.sub.p1 voltage level to
the enabling E.sub.p2 level. In this manner does the prosthetic
device actuate only when the sensed emg voltage level exceeds a
predesignated (and variable) threshold. As the prosthesis is being
actuated, its physical position is indicated by the wiper arm of a
position potentiometer (not shown). The feedback position signal
107 is applied to the lead-lag circuit 16 at terminals 108 and 109.
The function of lead-lag circuit 16 is to prevent actuation of the
prosthesis by a short term, high gain signal, e.g., noise. Via
resistors 100, 110, 111 and 112 and capacitor 114, the prosthesis
will only respond to long term signal, thereby preventing the
prosthetic device from continually opening and closing upon every
sensed signal.
Referring now to FIG. 13, there is shown a schematic diagram of the
power unit. Emergent from pulse width modulator 10 is a series of
enabling pulses as shown by lower portion of waveform g) of FIG.
12. These pulses are applied to the motor 8 after being amplified
by power transistor 17. Transistors 121 and 122 serve as driver
transistors for power transistor 17. Upon energization, the
rotation of the armature of motor 8 causes like rotation in gear
reduction means 123. After the necessary gear reduction is
accomplished, the control cable 14 is connected to a pulley that is
attached to the last gear element (not shown). Attached to the end
of the control cable is the prosthetic device 9. In this manner,
rotation of the motor shaft causes activation of the prosthetic
device 9. Also connected to the gear means 123 is the wiper arm 124
of potentiometer 18. In this manner, position feedback signal 107
is provided for lead-lag circuit 16. A transient suppression
circuit is provided with the power unit for inhibiting undesired
interference with power transistor 17 and driver transistors 121
and 122 when motor 8 is switched on and off. The transient
suppression circuit consists of resistor 125, zener diodes 126 and
127, and diode 128.
The gear box reduction means is shown in the cross-sectional
drawing of FIG. 14. The output signal of power transistor 17 of
FIG. 13 is applied to brush assembly 130 causing rotation of the
rotor 131, and subsequently rotor pinion 132. Rotation of rotor
pinion 132 induces rotation in spur gear 133 and pinion gear 134.
Final gear reduction is accomplished by spur gear 135 to which is
attached at its upper end pulley 136 and at its lower end
potentiometer 18. Wrapped around pulley 136 is the control cable 14
which actuates the prosthetic device. The position feedback signal
is provided by potentiometer 18 and wiper arm 124 as heretofore
discussed.
Referring briefly to FIG. 15, there is shown one embodiment
illustrating the physical arrangement of the subject invention in
conjunction with an above-the-elbow amputee. The sensor unit 20 is
held in intimate contact with the biceps muscle of the amputee. The
output signal of the sensor unit is connected to the control unit
12 via signal cable 140. Control signals from the control unit 12
to the power unit 19 as well as position feedback signals from the
power unit 19 to the control unit 12 are transmitted via signal
cable 141. Upon receiving energization commands from the control
unit 12, the motor causes the control cable 14 to retract (as
described previously), consequently opening the fingers of the
prosthetic hand. Since the operation of the prosthetic hand (or
hook) is well-known in the art, only nominal attention will be
directed thereto. Retraction of cable 14 will cause forearm 142 to
raise if the elbow is unlocked. When the forearm is in the desired
position, the amputee pulls on locking cable 143 by means of
shoulder harness 144, thereby causing lever arm 145 to lock said
forearm 142 into position. Again, since this technique is well
known in the art, no detailed discussion will be accorded thereto.
When the forearm 142 is locked into position via lever arm 145,
further retraction of the control cable 14 will cause the fingers
of the prosthesis hand to open.
As was mentioned above, placement of the motor within the
prosthetic device (hand) is but one embodiment of physical
placement. Another embodiment that has been utilized is placement
of the power unit around the waist of the amputee when there is not
enough room for it in the prosthesis.
Other modifications, adaptations and embodiments of the present
invention are of course possible in light of the above teachings.
Therefore it should be understood that within the scope of the
appended claims the invention may be practiced otherwise than as
specifically described.
* * * * *