Docstoc

Lift Assist Arm Brace

Document Sample
Lift Assist Arm Brace Powered By Docstoc
					Lift Assist Arm Brace
              By


        Erin Gallagher
        Sudheer Potru
         Ronson Yong




ECE 445, SENIOR DESIGN PROJECT

         SPRING 2006




        TA: Hyesun Park



          May 1, 2006


         Project No. 44




               1
                                               ABSTRACT

The lift assist arm brace is a wearable device that provides mechanical assistance in raising simple
objects. The brace consists of electrodes that sense the contraction of the muscles, an amplifier circuit, a
noise filtration circuit, a variable pulse width modulator circuit, and an electric motor-driven brace
mechanism to bend the joint.

The signal sensed by the skin electrodes is filtered and amplified to control the motor. The
microcontroller measures the difference between bicep and triceps contractions and modulates the pulse
width to the motor accordingly, mimicking natural arm extension-contraction motions and eliminating
any inadvertent movement of the brace.

Our device works as specified, however there is a delay in switching from extension to contraction.
Improvements in the motor-driving circuit need to be made to resolve the switching delay.

We would also like to improve our device by making it out of more lightweight material and using a
lighter motor to increase the comfort and ease-of-use for the wearer.




                                                     2
                                                         TABLE OF CONTENTS


1.   INTRODUCTION ....................................................................................................................3
     1.1 Purpose ...............................................................................................................................4
     1.2 Specifications ......................................................................................................................4
     1.3 Subprojects ..................................................................................................................... 4-5
         1.3.1 Electrode Pair ............................................................................................................4
         1.3.2 Differential Amplifier............................................................................................ 4-5
         1.3.3 Noise Filter ................................................................................................................5
         1.3.4 Rectification ..............................................................................................................5
         1.3.5 Envelope Detection ...................................................................................................5
         1.3.6 PIC Microcontroller ..................................................................................................5
         1.3.7 Motor Driving Circuit ...............................................................................................5
         1.3.8 Motor/Mechanical Linkage .......................................................................................5

2.   DESIGN PROCEDURE AND DESIGN DETAILS ............................................................ 6-8
     2.1 Electromyelogram Circuit ..................................................................................................6
     2.2 Filtration and Amplification Circuit ...................................................................................6
     2.3 Rectification and Envelope Detection Circuit ....................................................................7
     2.4 Filtration and Amplification Circuit ...................................................................................7
     2.5 Filtration and Amplification Circuit ...................................................................................8
     2.6 Mechanical Linkage............................................................................................................8

3.   DESIGN VERIFICATION ................................................................................................. 9-13
     3.1 Testing and Verification .....................................................................................................9
         3.1.1 Filtration and Amplification ....................................................................................10
         3.1.2 EMG Voltage Output ..............................................................................................10
         3.1.3 Microcontroller Tests ..............................................................................................10
         3.1.4 Microcontroller Tests ..............................................................................................10
         3.1.5 Microcontroller Tests ..............................................................................................10
         3.1.6 Microcontroller Tests ..............................................................................................10
     3.2 Tolerance Analysis ...........................................................................................................11
     3.3 Conclusions.......................................................................................................................12


4.   COST ................................................................................................................................ 13-14
     4.1 Parts ..................................................................................................................................13
     4.2 Labor .................................................................................................................................13
     4.3 Total Cost………………………………………………………………………………..14

5.   CONCLUSIONS ....................................................................................................................15

     APPENDIX – TITLE .......................................................................................... ATTACHED

     REFERENCES .......................................................................................................................16




                                                                             3
                                          1. INTRODUCTION


The lift assist arm brace is a wearable device that provides mechanical assistance in raising simple
objects. The brace consists of electrodes that sense the contraction of the muscles, an amplifier circuit, a
noise filtration circuit, a variable pulse width modulator circuit, and an electric motor-driven brace
mechanism to bend the joint. The signal sensed by the skin electrodes is filtered and amplified to
control the motor. The microcontroller measures the difference between bicep and triceps contractions
and modulates the pulse width to the motor accordingly, mimicking natural arm extension-contraction
motions and eliminating any inadvertent movement of the brace.

1.1 Purpose

There are many health conditions, including stroke trauma, surgical operations, and aging, that limit a
person’s ability to lift everyday objects. It was our goal to use our engineering skills to design a
biomedical device that would mechanically assist the wearer in lifting light objects, thus restoring
normal functioning, preventing further injury, and improving the quality of life of the user.

We wanted our brace to be comfortable to wear and function smoothly so as to mimic natural motion of
the arm. The brace was designed to use the patient’s own muscle contractions to power the motor,
eliminating the need for external controls. The filtration and amplification circuits were optimized to
ensure a usable signal from each muscle contraction, and the microcontroller was programmed to allow
for smooth, controlled movement of the motor-brace assembly.

1.2 Specifications

Because the signal generated by the muscle contraction is only on the order of a few millivolts, the
amplification circuitry was designed to have a gain of 100 so that the signal would be large enough to
control the motor. The frequency of the muscle-generated signal (20-140Hz) required that the filtration
circuit only pass the frequencies in that range to eliminate noise and other interference in the signal. The
microcontroller was programmed so that only a significant difference in bicep and tricep contractions
would trigger the motor to move the brace. This difference was determined to be approximately .25V,
after amplification. To ensure the brace would be capable of lifting most light objects, a dc linear
actuator with 30-lb rated lifting power was used to power the brace. In order to avoid damage to
circuitry, the motor must not be supplied power greater than 36W.

1.3 Subprojects

The design of our project was broken down into separate modules, which each served a specific
function. (A block diagram showing the separate modules can be found in the appendix)

1.3.1 Electrode Pair

There are two pairs of electrodes attached closely together, one directly over the bicep and the other over
the triceps. The electrodes will pick up an analog voltage indicating the strength of muscle contraction.
A common-mode electrode will be placed over a bony area and will be common to both pairs of
electrodes.

                                                     4
1.3.2 Differential Amplifier

The electrodes will be inputted into the differential amplifier with gain = 1. This uses common mode
rejection principle to eliminate noise such as those from the heart.

1.3.3 Noise Filter

Noise filtration is required to remove ambient and AC electrical noise. A band pass filter will be
required to remove higher frequency noise components and low frequency motion artifacts. The EMG
signal will generally have frequency components between 20Hz and 450Hz. Using active noise
filtration circuitry, we can add additional gain to the signal because the voltage will be on the order of
millivolts.

1.3.4 Rectification

Since the raw EMG signal is biphasic, it will need to undergo rectification. The positive amplitudes
contain the main contraction strength information in the signal.

1.3.5 Envelope Detection

Another high pass filter is then required to smooth out the signal and serve as envelope detection.

1.3.6 PIC Microcontroller

A PIC microcontroller is used to input the voltage level of the filtered signals into an analog-to-digital
converter, compare internally, and, based on the difference between the strength of the bicep and tricep
contractions, output a PWM to drive the motor.

1.3.7 Motor Driver Circuit

A transistor, which receives the PWM output of the PIC determines if the motor is to be on or off. The
output of the transistor is then fed into a combination of relays, which switch the polarity of the motor
leads, depending on whether the brace should extend or contract.

1.3.8 Motor/Mechanical Linkage

The motor will power the lifting mechanism of the arm brace. The lifting mechanism will be a DC
linear actuator, which extends under a positive input voltage and contracts under a negative input
voltage. The actuator is mounted on the brace, which consists of two long support beams extending the
length of either arm that are attached by a moveable hinge and secured to the arm by four Velcro straps.




                                                     5
                         2. DESIGN PROCEDURE AND DESIGN DETAILS

2.1 Electromyelogram Circuit Design

The contraction of the “pull” and “push” muscles of the bicep and tricep produce an analog voltage
signal which can be detected and analyzed. In our design, each pull and push muscle from the bicep and
tricep is represented by an electrode, and these four electrodes (along with a common ground) were
attached to an AD622 instrumentation amplifier. However, we determined that the AD622 required a
dual voltage supply. Because we wanted to simplify the design and not have to use four power supplies
in order to adequately run it, we switched to an AD623 instrumentation amplifier chip, which has a
single-supply voltage. In addition, we had originally planned to set the differential amplification gain to
1, but in order to see the signal in any relevant capacity, we were forced to use a 100X gain on the
amplifier. Later, upon attempting to read the signal out on an oscilloscope, we discovered that a large
DC offset voltage was preventing us from obtaining a useful readout. Therefore, to eliminate this DC
offset, we built a simple RC high-pass filter with an 8.2 nF capacitor and a 1 MΩ resistance that
precedes the two inputs of each AD623; thus, we were able to overcome it.

2.2 Filtration and Amplification Circuit Design

The frequency of an EMG circuit generally ranges from about 20 to 450 Hz, so we originally designed a
simple first-order bandpass filter for these frequencies. We also assumed that a bandstop filter for
eliminating 60 Hz noise would be necessary, and were planning on calculating and designing a filter as
such, but it turned out that the differential amplification removed the 60-Hz noise and it was no longer
an issue. Design of the bandpass filter necessitated the following calculations.

       Choose C = 10nF.

       fc, lowpass = 1/ (2*pi*R2*C) = 450Hz
       R2 = 35k

       G = -R2/R1 = 1000
       R1 = 35

       fc, highpass = 1/ (2*pi*R3*C) = 20Hz
       R3 = 796k

       G = -R4/R3 = 1
       R4 = 796k

Upon testing of the signal, we discovered that a more appropriate frequency range would be from about
20 Hz to 150 Hz. Eventually, we decided upon a second-order Sallen-Key bandpass filter with cutoff
frequencies of about 20 Hz and 150 Hz, but simplified the resistor values to decrease the size of the
circuit and give us frequency cutoff values of about 18 Hz and 145 Hz (See Figure 3.2.1 for labeling of
parts). Choosing all capacitors to be either 1 uF or 2.2 uF (in the case where one had to be doubled),

fc,high = 1 / [sqrt (2 * pi) * (1 uF) * 22kΩ] = 18.13 Hz
fc,low = 1 / [(2 * pi) * (1 uF) * 1.1kΩ] = 144.69 Hz


                                                       6
We then built the filtration and amplification circuits with these components from the parts shop, and
after this revision, no updates were necessary.

2.3 Rectification and Envelope Detection Circuit Design

Because attempting to put negative voltage into a transistor accomplishes nothing, we realized that we
needed to rectify the signal before anything could actually be done with it. Rectification in our circuit is
performed by a simple 1N5819 Schottky diode, which serves to essentially cut off the bottom half of our
signal (i.e. the negative voltages).

Pulse-width modulation is dependent on being able to trace out the top of a signal, thus necessitating
envelope detection. The envelope detection circuit which accomplishes this task is a simple passive RC
high-pass filter with a cutoff frequency of about 3 Hz.

Choose C = 1 uF

Frequency ~ 3 Hz
R = 1 / (2 * pi * C * freq)
R = 53kΩ

After implementing the circuit with these components, we found that we were getting a signal of less
noise and lower amplitude, which was desired. No updates were necessary, and we used normal
components from the parts shop.

2.4 Microcontroller (PIC) Circuit Description

The PIC16F877A microcontroller is used to determine the difference between the bicep and tricep
muscle contractions. It also generates the control output to the motor circuit. Pin 1 is pulled high with a
47kΩ resistor to prevent resets. Pin 2 and 3 are used as analog inputs to an internal ADC. They convert
the analog 0-5V signal from the EMG circuit to a digital signal that can then be manipulated via the
PIC. Pin 4 and 5 are used to set up the Vref limits (0-5V) of the ADC. Pin 11, 12, and 13 are power,
ground, and clock signal respectively. Pin 17 is the PWM output generated from the difference between
the contraction strengths. Pin 19 and 20 generate a TTL signal that is used to determine contraction or
extension. When the contraction differential between muscles isn't great enough, both pin 19 and 20 are
low.

The code in the PIC implements the software flowchart found in Figure C.1. Specifically, the most
relevant section of the code is in Figure 2.1 below. It takes in the values of the bicep and tricep muscle
contractions and assigns them to bit numbers in ADC registers. After this, it compares the two values in
the ADC registers to determine which is more significant. If the bicep contraction is more significant
than a certain set threshold, it sets Pin_D0 to high, so the brace will contract. If the tricep contraction is
more significant than the same certain set threshold, it sets Pin_D1 to high, so the brace will expand. If
there is not a significant difference between the two contractions, then the duty cycle is set to zero and
the brace will not move in either direction.

               if(adc_reg1 > adc_reg2){
                     difference = adc_reg1 - adc_reg2;

                       if(difference > 10){
                             output_high(PIN_D0);
                             output_low(PIN_D1);
                             set_pwm1_duty(difference);

                                                      7
                       }
                       else{
                               output_low(PIN_D1);
                               output_low(PIN_D0);
                               set_pwm1_duty(0);
                       }
               }
               else{
                       difference = adc_reg2 - adc_reg1;

                       if(difference > 10){
                             output_low(PIN_D0);
                             output_high(PIN_D1);
                             set_pwm1_duty(difference);
                       }
                       else{
                             output_low(PIN_D1);
                             output_low(PIN_D0);
                             set_pwm1_duty(0);
                       }
               }
                               Figure 2.1 PIC Microcontroller Code

2.5 Switching Circuit Design

Essentially, the task of the switching circuit is to provide the linear actuator with an appropriate input, in
order for it to be able to decide whether or not to expand or contract.

Initially, we used two transistors to design this circuit, but we then realized that we would be using a
negative 12 volt input to one of the transistors, so this design would not work. So, instead we tried using
a 8L02-05-01 DC relay and a transistor, but we found that we were consistently blowing out parts.

Therefore, our final design is as follows. There is a IPS511 Intelligent MOSFET Switch that uses the
PWM control signal to switch a 12V source. The output is then fed into the DPST 8L02-05-01 DC
relays. One relay is controlled by pin 19 and the other is controlled by pin 20 of the PIC. For bicep
contraction, pin 19 is high which closes the two switches in the first relay when they would both
normally be open. This connects the red wire of the motor to ground and the black wire of the motor to
the output of the transistor. For tricep contraction, pin 20 is high; thus, on the other relay, the red wire
of the motor is connected to the transistor output and the black wire to ground. This allows for the
decision regarding whether or not the motor should be going up or down.

2.6 Mechanical Linkage Design

Lift Capabilities




                                                      8
110lbsf = 498.304N
550lbsf = 2446.52N

8” = .2032m
10” = .254m

Motion

.2032*498.304*cos(45) = Fweight * .254

Fweight = 276.792N
Weight = F/a = 28.2441kg = 62.2675 lb

Static

.2032*2446.52*cos(45) = Fweight, static * .254

Fweight, static = 1383.96N
Weightstatic = 141.22kg = 311.338 lb




                                                 9
                                      3. DESIGN VERIFICATION


3.1. Testing and Verification

Here are the tests and verification exercises that we performed on our device and circuit.

3.1.1 Filtration and Amplification Tests

Our testing of this circuit consisted almost entirely of obtaining a signal between 0 and 5 volts in order
to ensure that the signals could be used as inputs to the microcontroller, which takes up to five volts as
an input. We never had a problem with this, and as can be seen in Figures D.2 and D.3, we obtained an
adequate signal that could be used. We also tested the final amplifier in the circuit to ensure that it gave
us a linear increase in voltage as desired, and this is verified and presented in Figure D.8.

Last, we used the oscilloscope to essentially perform a manual frequency sweep of our filtration circuit
by putting in a simple 5V p-p square wave from a function generator, and it provided us with the data
found in Figure D.7.

3.1.2 EMG Voltage Output Tests

We wanted to see just how the EMG voltage would depend on the placement of electrodes and how
altering the weight of an object that was lifted during testing would affect the voltage as well. As can be
seen in Figures D.5 and D.6, neither had a significant impact on the voltage that we received, in the
sense that the average variation was less than 10% in both cases. In addition, it is noted in EMG
literature that voltage output can vary significantly from person to person, so we tried our circuit out on
all three of us, and received adequate readings like that in Figure D.3. all three times that we tested. We
concluded that our circuit accounted well for any spatial variations in electrode placement and lifting
weight, and that our brace would work under most circumstances.

3.1.3 Microcontroller Tests

We tested the PWM duty cycle versus bit number to ensure linearity, and it was verified in Figure D.10.
We also tested the analog voltage thresholds in the ADC to make sure that there was a linear increase in
voltage for an increase in bit number, and this is shown in Figure D.9. Last, we tested the output of the
PWM on an oscilloscope for physical verification and proof, and this is shown in Figure D.4.

3.1.4 Power Consumption

                                                     10
The power consumption of the front part of the circuit (minus the motor) was equal to 5 V * 29 mA =
0.145 W, which is not especially significant and will not result in any huge drains. The motor requires
12 volts at 3A max current, which equals 36W. This is something that needs to be resolved in future
work.

3.1.5 Temperature

Upon doing some research, we discovered that a temperature of 49 degrees Celsius (or 120 degrees
Fahrenheit) requires two minutes of continuous skin contact to cause a burn. We tested our device with
a thermometer and found that the highest temperature that the device reached was about 56 degrees
Celsius, and this will provide more than enough time for the wearer to remove the brace and avoid any
kind of burn.

3.1.6 Lift Capability Verification

To test this theoretical weight lifting capability, we fixed the brace to the wall and fully extended the
brace. We contracted the brace with 30 lbs (the arm brace’s rated maximum weight) attached to the end
of the arm brace. Attaching the actuator to 12V power source, the brace easily underwent a full
contraction and extension motion.

3.2. Tolerance Analysis

We tested our resistors and capacitors in the bandpass filter in our circuit to see if they could still satisfy
the necessary frequencies (namely, we wanted to ensure that a 10% change in either direction of our
acceptable signal would still fall over 10 Hz and under 100 Hz in frequency).

LPF:

R2 = R3 = 1 / (2*pi*C4*f)

fc = 1 / (2*pi*C4*R2)

C4, max = 1.1*1uF = 1.1uF
C4, min = 0.9*1uF = .9uF

R1, max = R2, max = 1.1*1.1k = 1.21k
R1, min = R2, min = 0.9*1.1k = 0.99k

119.575 Hz < fc < 178.625 Hz

The upper range of our frequency bandwidth is acceptable since much of the useable EMG signal is in
the between 10Hz and 100Hz.

HPF:

C1, max = C2, max = 1.1*1uF = 1.1uF
C1, min = C2, min = 0.9*1uF = .9uF

R6, max = 1.1 * 22k = 24.2k
R6, min = 0.9*22k = 19.8k

                                                      11
fc = 1 / (2*pi*C1*R6)

14.9866 Hz < fc < 22.387 Hz

The lower range of our frequency bandwidth is also acceptable as the lower bound is above 10Hz which
eliminates the motion artifacts.

3.2 Conclusions

We found that we completed most of the relevant testing successfully, and discovered that the tolerance
analysis that we performed allowed for a large variation in resistance and capacitance in our circuit. In
short, we feel that our brace is a safe, inexpensive device useful for anyone with muscle problems.




                                                   12
                                                  4. COST

The following is a list of all parts necessary to the design and the cost of each of them.

4.1 Parts

                             4.1.1 Parts and Cost for Electronic Circuitry

                                                   Cost
                                                                                Total Parts
     Description             Part Number            Per          Quantity
                                                                                   Cost
                                                   Unit
Disposable Electrodes             EL501            $0.70            20              $14
       (Biopac)
   Instrumentation
      Amplifier                AD623AN             $3.84            2               $7.68
  (Analog Devices)
 Pic microcontroller
     (Microchip)              PIC16F877A           $4.50            1               $4.50
Operational Amplifier
         Chip                  AD820AN             $1.67            6              $10.02
  (Analog Devices)
  1MOhm Resistor                20J1K0             $1.33            5               $6.65
  22kOhm Resistor               3W036              $0.17            2               $0.34
  11kOhm Resistor               3W336              $0.17            2               $0.34
 1.1kOhm Resistor               1W482              $0.11            4               $0.44
 100kOhm Resistor               3W424              $0.17            2               $0.34
    1uF Capacitor             75-5HKD10            $0.43            2               $0.86
   2.2uF Capacitor              DD050              $0.16            4               $0.64
   8.2nF Capacitor             SXM110A                                              $2.20
                                                  $0.55      4
    12V Batteries                                                                   $1.00
                                                  $0.50      2
   Schottky Diode             1N5819               $0.17            2               $0.34
     Transistor               2N3904               $0.10            1               $0.10
       Relay               KHAU17D11-12V           $4.55            2               $9.10
                                                                                   $58.55
                                                             SUBTOTAL


                       4.1.2 Parts and Cost for Motor and Mechanical Linkage
6" Stroke High Speed           ZYJ(s)07-8-12-6”

 65lbs force Linear                               $119.99           1             $119.99
       Actuator
     Steel beams                                   $3.09            2               $6.18
     2”, ¼ Bolts                                   $0.56            4               $2.24
      Lock nuts                                    $0.35            4               $1.40
   2 Velcro Straps                                 $4.99            2               $9.98



                                                     13
                                                                             $139.79
                                                         SUBTOTAL


Total Parts Cost = $58.55 + $139.79 = $198.34

4.2 Labor

Labor cost was determined using the formula below. Hours spent was estimated to be 130 hours per
person, including research, manufacturing, and debugging time. Ideal hourly salary based on annual
salary projection.

Labor Cost per person = (Ideal Hourly Salary x Hours Spent x 2.5)
                    = $50/hr x 130hrs x 2.5
                    = $16,250

Total Labor Cost      = 3 x $16,250
                      = $48,750

4.3 Total Cost

Total Design Cost     = Total Parts Cost + Total Labor Cost
                      = $48,948




                                                 14
                                           5. CONCLUSIONS

The project has had many accomplishments and some uncertainties that require future work. The entire
EMG circuit was quite good at extracting a signal from the muscles, filtering the signal for relevant
frequencies, and smoothing out the signal over a range of 0-5V. The PIC also functioned properly for
manipulating the two EMG input sources and outputting a proper control output. The programming can
also be optimized further to dynamically control EMG parameters such as the amplifier gains, the
frequency bandpass window, or the envelope detection frequency. This would make the brace more
robust and allow the brace to be used on different muscle groups which all have different characteristic
twitch frequencies and contraction strengths. The mechanical aspect of our project functioned quite
nicely also. The motor driving circuit, however, needs some more work. Another solution needs to be
found for the transistors/relays to accommodate the high power limits (~36W). Also, there is a thermal
performance issue to be resolved which may require looking at some heat sinks or alternate means of
cooling. A proper motor driving IC chip could solve all of these problems. The brace itself can further
be modified to include lightweight construction materials, adjustability, a different motor, and additional
safety features. Some lightweight materials could include aluminum which is light-weight and strong.
The brace can also include an adjustability feature to lengthen or shorten the brace for a custom fit.
Different motors, including servo motors, could be another solution to the weight and power issues.

Safety is the major ethical concern in this project. Some concerns include the brace pinching/cutting the
skin, components burning the skin or starting fires, and materials causing allergic reactions. Additional
safety features could include a quick release pin to disengage the motor or brace from the user. With all
our successes, this project is a good proof of concept for a lift assist brace but will still need additional
work to make it marketable.

We would like to thank Professor Carney and Ms. Hyesun Park for all of their help in accomplishing
everything that we did this semester. We could not have done it without your guidance and help, and we
appreciate all of your efforts.

Note: Appendix attached separately




                                                     15
                                         REFERENCES

Electromyelogram: The EMG. <http://moon.ouhsc.edu/dthompso/pk/emg/emg.htm> 25
February 2006.

Design of an Electrical Prosthetic Gripper using EMG and Linear Motion Approach.
<http://fcrar.ucf.edu/papers/tp2_ahabdima_fiu.pdf> 30 March 2006.

How to Build an EMG.
<http://www.univie.ac.at/cga/courses/BE513/EMG/> 28 February 2006.

DC Linear Actuator Data Sheet
<http://www.firgelliauto.com/Linear%20Actuator%20Catalog%202005.pdf> 17 March 2006




                                                16

				
pptfiles pptfiles
About