Power Assist Wheelchair by z0Ze2Va


									Power Assist Wheelchair

ECSE-4961: ECSE Design

 Written By,
    Eric L. Steed
    Joseph Lavigne
    Nate Diller
    Robert LaBella
    Tripp Hyde

 Advised By,
    Syed Murtuza

 December 9, 2002

I. Abstract
            This document details the design and construction of the ECSE
Design students’ project. The project was to design a lightweight, power-assist
wheelchair. The ECSE Design students’ role was to design the user input and system
control. We successfully implemented an open loop control system as well as a working
user input system.

II. Table of Contents
I.     Abstract                                    2
II.    Table of Contents                           3
III.   List of Figures                             4

       1.     Introduction                         5

       2.     User Input                           6

       3.     Slip Ring Research                   9

       4.     Slip Ring Manufacturing              10

       5.     Battery                              12

       6.     Motor                                15

       7.     Controller                           19

       8.     Controller (contingency plan)        21

       9.     Engineering Analysis                 23

       10.    Regulations and Environment          30

       11.    Summary and Conclusions              31

IV.    Appendix A
       1.    Project Costs and References          32
       2.    Purchased Parts Information           33
       3.    Simulation Calculations Spreadsheet   46
       4.    Matlab Simulation Code                48

V.     Appendix B
       1.    Resumes

III.      List of Figures
Title                                                               Page
Figure 1: System Block Diagram                                      5
Figure 2: FUTEK FR1010 force sensors                                7
Figure 3: Force sensor mounting                                     8
Figure 4: Amplifier circuit testing                                 8
Figure 5: Example of a commercial slip ring                         9
Figure 6: Constructed slipring and brush assembly                   11
Table 1: Comparison of different battery chemistries                13
Figure 7: Sample brushless DC motor circuit                         18
Figure 8: Micro II Scooter Controller                               21
Figure 9: Force sensor and differential op-amp circuit              23
Figure 10: Output of force sensors and amplifier                    24
Figure 11: Matlab model of power-assist wheelchair control system   24
Figure 12: Motor Dynamics Model                                     25
Figure 13: Motor Velocity (1 second on, 2 seconds off)              26
Figure 14: Motor Velocity (.5 second on, 1 second off)              27
Figure 15: Matlab model of the control system with wave shaping     27
Figure 16: Motor Velocity (1 second on, 2 seconds off, b=0.5)       28
Figure 17: Motor Velocity (1 second on, 2 seconds off, b=0.75)      28
Figure 18: Unmodified and Modified Input to the Controller          29

   1. Introduction

        Our team had been assigned the task of creating a power assist wheelchair to assist
users at a level above a manual wheelchair and below a powered wheelchair. This would
compete on with current market products such as the e-motion. Our long term goal for this
product is to work with the mechanical and material groups to create a functional and cost-
effective power assist wheelchair while concentrating on developing a working control system
and user interface for the wheelchair; for the single semester, our goal is to develop a control
system and interface with a power assist wheelchair integrated with the mechanical group’s
physical prototype design.
        Early in the semester we focused on developing only a controller, however we soon
realized the task was bigger than just that. After the mechanical group decided that a battery
powered motor was the most feasible, and we worked in conjunction with them to decide on a
drive system. Our group then produced a block diagram for our total design.

           User Input

           Slip-ring                     Battery

           Controller                    Amplifier                     Motor
                                                                       (output to wheel)

                               Figure 1: System Block Diagram
        The user input and transmission segments were not immediately thought about as the
line between the EE’s and mechanical group’s responsibilities had not yet been drawn.

2. User input
             The issue at hand was how to simply take the users push on the push-rim into a
         signal which we could use to drive the motors. We went through several requirements
         and brainstormed issues. We decided to use a force sensor in the end.
             One of the most important requirements was for the chair (user?) to have little or
         no learning curve. The most intricate part of that is the user input device. First, an
         afterthought to our group, we soon realized we had to create a system to get the input
         from the user and eventually to the “fifth wheel.” (give a brief description, including
         drawings, of mechanical design in the beginning) We found several options through
         brainstorming and then sat down to think through different possibilities and their
         benefits and shortcomings. We then eventually chose the one which was most feasible
         and effective.
           Accelerometer mounted on the wheel near the tire: This could be used to measure
            the user’s input as the derivative of acceleration (jerk); however, small bumps in
            the road could cause the mechanism to read falsely so that a bump in the road
            would cause it to brake. Also, wires or a wireless system would need to be
            implemented, as with many of these systems.
           Force sensor/strain gauge: The E-motion (no prior reference) uses a similar
            sensor mounted between the rim and the wheel. These sensors would be mounted
            between the push-rim and the wheel (include a sketch), basically be taking the
            place of some or all of the spokes. This method of measurement is very accurate.
            Attention to mounting and signal transmission is necessary.
           Linear potentiometer:    This would work just like any other potentiometer. A
            wiper attached to the rim moves along the base mounted on the wheel. The
            movement of the rim relative to the wheel changes the voltage across the
           Optical sensor:            This would have to create pulses by counting the
            spokes go by or some other IR method. Using the signal would be much like the
            accelerometer; however the drawback of hitting a rock and the user being jerked
            back even more is still a problem with this as well. The positive though is this
            could be mounted on the chair, instead of the wheel, thus eliminating the need of
            running any wires through the hub or having any radio transmitted signal.
           Magnet speedometer:        This would be similar to the speedometers that are
            used by bikes with a magnet mounted on the spoke and then counting every time it
            goes around relative to time. This shares the convenience of not having to run
            wires with the optical sensor. It also shares the same drawbacks of the optical
            sensor with the addition of being fairly inaccurate as the rotation of the wheel is
            relatively slow.
           Friction wheels mounted to chair: This would have two small wheels that would
            turn with the wheel and the push-rim. Spokes between the wheel and push-rim
            would have to be very flexible to allow larger deflection. The difference in phase
            would then be read as an acceleration/deceleration. This would have the feature of
            not having to run wires through the hub, however the difference in phase was
            thought by some to be nominal and thus have little or no effect.

              The main factors we used in choosing a possible design were: simplicity, cost,
        feasibility, and hands-on knowledge. We decided that the force sensor would be best as
        it would be the most accurate. After looking at several models, we chose a sensor from
        Futek.1 (Include brief description of the sensor …bridge circuit, resistance values,
        sensitivity etc.)

                          Figure 2: FUTEK FR1010 force sensors

This sensor was bought with intentions of mounting it in pairs on each wheel.
Due to limited time, efforts between the mechanical and electrical team could not
produce a working mount. The simplest idea for a mount is merely two L-brackets
mounted on both sides of the sensor, and then screwing that on the preexisting holes on
the wheel and slip-ring (include a sketch). This mounting process, however, would place
the push-ring approximately 1.25 inches from its current location; a better mounting idea
would be favorable (preferable?).
An amplifier circuit was necessary to amplify the signal from 10mV to a 0-5V range that
the controller could then use.
The force sensors were tested by connecting them to a block of wood with a notch cut out
of it. A screw below and above it could be adjusted to create the constant necessary force.

    http://www.futek.com model #FR1010 http://www.futek.com/product.asp?product=FP10000-00002-C

                             Figure 3: Force sensor mounting

At 5V input, the force sensor outputted up to 10mV or -10mV max based on the direction
of deflection. This small voltage was then amplified by a differential amplifier to get the
voltage into the 0-5V range needed by the controller. A differential amplifier was chosen
because the output of the force sensors not simply 10mV above/below ground (needs
editing), but rather the difference between a positive and negative signal. Also, high value
resistors were used for the op-amp circuit due to the relatively small resistance of the
force sensor bridge circuit.

                            Figure 4: Amplifier circuit testing

This circuit used 4.7 meg-ohm and 10 k-ohm resistors to create a gain of 470. Actual
maximums/minimums of the output of the circuit were near slightly higher that than
expected at approximately 4.8 V positive or negative. This confirmed the experimental

3. Slip Ring Research
       Being that we have sensors on our (needs editing)wheels we figured out that we
needed a way to transmit the signal that will be input by the user. This is a problem
because we have a part of the chair that spins relative to another. The problem therefore
becomes how to transmit a signal through a moving axle.

        We didn’t look into too many possibilities because there really aren’t very many
of them. The first that came to mind was radio transmission on the FM band. We could
mount an independently powered circuit on the wheels of the chair that would be
responsible for the receipt of the user input. From there the transmitter would have to
relay the information to the part of the chair that was stationary not in rorary motion? that
housed our controller. We also looked into the possibility of using Bluetooth technology,
which would allow us to transmit and receive signal, and we wouldn’t have to do any
circuit programming. Bluetooth is already at a level where both sides, the transmission
side and the receiver side are synchronized. The last possibility that we looked into was
using some manner of slip ring that would allow us to transmit signal through a set of
spinning rings and brushes.

        We finally decided on a slip ring for a number of reasons. Slip rings we thought
would be easy to implement, less expensive than a wireless option, and required less
mounting to the actual chair. We did more research and we looked into many options in
the realm of slip rings. Slip rings, as we found out are rather costly, in the area of $350-
$3000 each, so we did some more research. After a group meeting Professor Alben gave
us the idea of using a printed circuit board to construct a set of concentric copper rings
that would be able to spin freely and perhaps mount a set of brushes elsewhere on the
chair so that they could be moved, and tension adjusted. From there, Tripp (names
should be in an appendix ) took over the responsibility of actually manufacturing the
rings, after Robert saw and explained that purchasing them would be far too costly in
terms of time and money.

                       Figure 5: Example of a commercial slip ring

4. Slip Ring Manufacturing

        The original issue was transmitting the user input signal from the spinning wheels
to the stationary chair. We decided this would be accomplished using sliprings
manufactured by our team.

Options and Decision
        We had two options: concentric or cylindrical sliprings. We ended up choosing
concentric because the contacts and brushes would be more exposed and easier to work
with on the prototype. In a cylindrical slipring the contacts and brushes would be
enclosed inside at least one cylinder, making it impossible to get to in case something had
to be fixed. Also, the cylindrical sliprings would take up more space between the wheel
and mount because all the brushes and contact paths are aligned next to each other. In the
concentric slipring all the sliprings are in the same plane, just each one is bigger than the
one before. We also had to decide what kind of brushes to use. Originally we were
going to use steel brushes from an electric clutch, but after doing some testing, we
realized that the steel was scratching away the copper. Our other option was softer
carbon brushes that are self lubricating. We decided these would work better because
they would not scratch away the copper.

        The manufacturing (fabrication) of the sliprings was a great learning experience.
The first step was to create the rings themselves. This was done by etching a circuit
board. We first printed out the circuit (just four copper circles) on a piece of clear
acetate. We then stuck this to the light sensitive side of the circuit board and exposed it
to fluorescent light for the appropriate amount of time. The board was then washed in a
chemical to remove the rest of the light sensitive material and then washed in a second
chemical bath to remove the copper that was exposed by the previous wash. The final
step was to use yet another chemical to remove the rest of the photosensitive material and
expose the four copper rings. The final product was cut to shape with a Rotozip. The
next part of the process was the creation of the brushes. This was done by hollowing out
nylon bolts and putting the spring loaded brushes in side. The brushes were obtained
from the DeWalt service center in Latham for free. We then drilled out and tapped the
wheel spokes of the wheelchair in the appropriate places. This allowed for the brushes to
be adjusted to the right about of distance and pressure by simply threading them into the
tapped holes. The whole system fit together quite well, but we were never able to run a
signal all way through because the mechanical team never mounted the force sensors to
the rim of the wheel.

        We did not run into too many problems. The only minor problem you probably
can see in the picture at the end of the section. The wires on the slipring board were run
up through the board and soldered right to the contact paths. This probably would have
caused the brushes to jump and maybe lose contact momentarily. This might have cut off
the signal causing some control issues. Also, the sliprings were rather exposed. For the
prototype we did not bother to put any protection around the sliprings and outdoor
weather would have eventually ruined them.

        To next semester’s team we strongly suggest a whole new approach to the signal
transmission. Although the sliprings would work, they are rather bulky and probably a
bit lossy. Wireless transmission would be the best idea, but it would also require
additional research and some complicated circuitry. Also, the wireless transmission
circuitry would be much easier to protect from the weather because it simply be put in a
weather proof box. However, we chose to go with the sliprings because of possible
interference issues, but we still believe that the final product should have wireless

                   Figure 6: Constructed slipring and brush assembly

5. Battery
         One major part of the power-assist wheelchair will be the power supply. Most fully
powered wheelchairs use sealed lead acid batteries, which are the most common battery used
in cars. These batteries are normally 12V DC and have high capacities, but they tend to be
very heavy. We are attempting to use the most cost effective and lightweight power source in
our design.

        A few different battery chemistries were researched when looking for a power supply
for our power assist wheelchair design. A brief description of each type of battery that was
researched is given below.

Sealed Lead Acid – These are some of the most common batteries for providing power for
vehicles. Most motorcycles, cars, and powered wheelchairs use a sealed lead acid battery. As
the battery depletes, the voltage drops off very quickly until there is almost no voltage when it
is fully discharged. The battery can be recharged by a deep cycle recharge or by trickle

Nickel Cadmium (Ni-cad) – These batteries tend to be used in small applications, such as
cameras, CD players, etc. The voltage does not drop off much when the battery is depleted. It
will normally drop from 1.4V to 1.0V, for example. They can be recharged by a deep cycle
recharge or by trickle charging. These cells have a memory effect which may cause the battery
to not work properly without deep cycle recharging.

Nickel Metal Hydride (NiMH) – These batteries tend to be used in small applications, such as
cameras, CD players, etc. The voltage does not drop off much when the battery is depleted. It
will normally drop from 1.4V to 1.0V, for example. They can be recharged by a deep cycle
recharge or by trickle charging. These cells have a smaller memory effect which may cause
the battery to not work properly without deep cycle recharging every 30 cycles.

Lithium Ion (Li-ion) – These batteries tend to be used mostly in laptops, cell phones and many
other specialty applications requiring a power source with low power fluctuations. The voltage
does not drop off much when the battery is depleted. It will normally drop from 3.7V to 3.5V,
for example. They can be recharged by a deep cycle recharge or by trickle charging. These
cells have no memory effect.

        It would be necessary to make a battery pack with any of these battery chemistries.
Most Ni-cad and NiMH batteries are 1.2V. Li-ion batteries are usually 3.6V and lead acid
batteries are usually 12V. In order to obtain the necessary 24V to power the motor, we would
need to put these batteries in series to increase the voltage. With the Ni-cad batteries, we
would also need to put two battery packs in parallel to achieve the required 10Ah capacity
since there are no Ni-cad batteries with this capacity.
        Putting batteries in parallel is not recommended, however, “because keeping the
battery pack equally yoked during repeated charge and discharge conditions can be a
problem”2. This problem became evident during an experiment to test the motor and gearbox
combination. Two similar power supplies were connected in parallel to achieve the required
current to drive the motor and gearbox. One power supply outputted 7A while the other

    From http://www.powerstream.com/index.html , How to Design Battery Packs

outputted 3A. The outputs from the power supplies were reversed when the same test was
performed again. It was impossible to predict the current draw from either power supply
during the tests.
        These battery chemistries were compared by weight and cost, which are the two
main requirements for this project. Other items that were taken into consideration were
battery pack size, safety, lifetime of the battery, and ease of recharging. The comparisons
are shown in Table 1. All dimensions were for a battery pack at 24V and approx. 10Ah.

               Lead Acid             Ni-Cad                 NiMH                  Li-ion
Weight         15-30 lbs             12-14 lbs              8-11 lbs              1.5-5 lbs
Cost           $80-130               $140-240               $230-300              $???
                                     Memory effect does Must deep cycle
               Deep cycle or trickle                                        Deep cycle or trickle
Charging                             not allow trickle  every 30 cycles to
               charge                                                       charge
                                     charging           avoid memory effect
Size           6x8x4                 7.5x4x5.5              7.5x4x3.5             8.5x3x2.5

                                                   There's a chance of            Requires a
                                                   the battery                    protection circuit to
Safety         No major concerns No major concerns
                                                   overheating if it is           maintain current
                                                   charged too much               and voltage

               300+ deep cycles
Battery Life                         500+ deep cycles       700+ deep cycles      500+ deep cycles
               600+ half cycles
Table 1: Comparison of different battery chemistries3

         As shown in Table 1, the lithium ion battery pack is the lightest of the alternatives
and has the highest energy density. It is also the most expensive. Actual prices for the Li-
ion battery pack were unavailable. Using the IBM ThinkPad batteries as a reference, the
price of the Li-ion pack would probably cost between $800 and $1100. These batteries
also need a circuit to maintain a constant voltage and current. Another small disadvantage
is that the battery actually tends to “decay” over time if not stored in optimal conditions.
         The NiMH batteries have the next highest energy density, but are the next most
expensive. There is also the small safety concern with heat from the battery pack if the
batteries are overcharged. This is actually used in many situations, however, to determine
when the battery is fully charged which could be useful in the recharging process. The
memory effect is also not a major issue since it can be negated with one full discharging
of the battery every 30 cycles (probably once a month in the wheelchair application).
         The Ni-cad batteries have a similar energy density to the NiMH batteries, but
have the major disadvantage with the memory effect. This could be highly detrimental in
the wheelchair application if someone is expecting the wheelchair to assist him or her and
it actually ends up providing unexpected resistance to the user.

        The sealed lead acid batteries are the cheapest alternative, but they also have the
lowest energy density of all the alternatives. The weight and size may not be a huge
factor in the design of the prototype, however. They do have the advantages of trickle

 Costs and weights will vary depending on the manufacturer, but most costs were found at
http://www.powerstream.com/index.html and most sizes were found at

charging (which actually extends the life of the battery) and they are the most readily

Final Decision
        In the initial design of the prototype, we have decided to use a 24V, 10Ah sealed
lead acid battery that came with a scooter motor. This was decided based on cost,
availability, and that it was easily rechargeable since it came with a recharging circuit.


        If time permits, it may be beneficial to design and test a NiMH battery pack to be
used in place of the lead acid batteries to reduce the total weight of the wheelchair. The
sealed lead acid battery currently used weighs between 20 and 25 pounds and was rather
large. This may become even more of a factor if the individually powered wheel design is
used in the future. The decreased weight and size of the battery pack along with a
possible increase in performance may justify the price increase. It is also recommended
that next semester’s group look into the possibility of regenerative braking to recharge
the batteries during the wheelchair’s operation.

6. Motor
        The original issue was what to power the wheel chair with. We needed something
that would provide a very high torque (approximately 250 in lbs) at a very low velocity
(2-4 mph). If you view the attached sheet (need proper reference – figure number, table
…)created by the Mechanical team that details the torque calculations, you can see that if
the average push required by the user is 10 pounds, multiply this by 12 inches (radius of
the wheelchair wheel), and multiply by 2 (user pushes with 10 pounds on each wheel),
the result is about 250 in lbs (approximately 20 ft lbs). We looked at the average speed
of fully-powered wheelchairs (approximately 4-5 miles per hour) currently on the market
and the average speed a person walks (approximately 3 miles per hour) and tried to match
our calculations to fall between 3 and 4 miles per hour. After all was said and done, it
turned out to be a bit slower than this and that is discussed at the end of the “Initial
Options” section.

Initial Options
         The DMS team used some design methods to look at some very initial options for
powering this wheelchair from a mechanical spring to hydraulics. But research in the
current wheelchair market and simple efficiency led us to choose an electric motor. Of
course there are many types of electric motors, but the first thing we needed to decide
was whether to use AC or DC. At first the answer seemed simple. The batteries will
obviously supply DC current so the motor should run on DC. However, there were a few
flags that required us to at least look into AC: size, weight, and regeneration. I Wewill
address these issues below. After choosing the current type we then had to pick the
specifics of the motor. The specifics relied upon a handful of variables and some of these
were: drive wheel size, drive sprocket(s) diameter, drive inertia, distance to stop
(deceleration/electric braking on grade), and gear ratio. As one can see there are many
variables we had to consider.
         As mentioned above, we needed to choose AC or DC and also what voltage to run
the motor on. Regeneration (or regenerative braking) simply uses the motor as a
generator (a device that converts mechanical energy into electrical energy) to recharge
the batteries. When a DC motor is used as a generator it creates an AC current and in
order to charge the batteries we need it to be DC. This involves using a commutator on
the armature of the motor. The motor brushes touch the rotating commutator in such a
way that the polarity of the current moving in the armature is always connected to the
correct brushes. The net effect of this is that the generator output is always DC even
though the current inside the armature windings is always AC. The AC motor does not
use any brushes and does not need a commutator circuit. Therefore, the AC motor tends
to be smaller and lighter and, looking back on the project, it may have been wiser to
purchase a smaller motor because of the size constraints we had. However, we ended up
choosing a DC motor because of the way the current you supply to it is directly
proportional to the torque and speed of the motor. This is not necessarily true of the AC
motor because we would be supplying a DC current from the batteries and would have to
put it through a conversion circuit (using a sine wave generator). This creates some
additional complexity when trying to control the speed of the motor. We really wanted a
motor that had “Plug ‘N’ Play” capability. Simply apply the current and it would act
exactly how we wanted with no extra circuitry. Also, as it turns out, the mechanical team
did not design there drive system for both forward and reverse and, therefore, the

regeneration (and braking) would not work at all. This cancelled out the regeneration pro
for an AC motor. (This paragraph needs refinement)
        The battery pack we would be using is 24 volts and approximately 10 Ah. It is
simply two 12 volt SLA motorcycle batteries in series. Most low power DC applications
are 12, 24, or 36 volts and most of the motors we had access to were 12 or 24 volts.
Therefore the battery pack dictates that we will use a 24 volt electric motor. Also, the
higher voltage DC motors (90 V or 180 V) are probably more dangerous than is
necessary for our application, especially when dealing with a maximum of nearly 30
        The other major option was to purchase a brushless motor or a standard brushed
motor. With a brushless motor we would have needed a separate closed feedback loop
for the commutator to work properly because there are no brushes physically touching the
moving coil. Although this does seem to complicate things a bit, in the end it makes for a
better motor. In general, the brushless motors are smaller and lighter than standard
brushed motors. There is no mechanical component (brush) in the commutation and
therefore they operate more reliably at higher RPM’s and require less maintenance. Also,
the arcing between the brushes creates more noise and also emits ozone. One can use a
smaller DC motor with a higher-geared gearhead and save much space and weight (?).
Brushless motors have no physical drag from the brushes (increased efficiency), there is
no need for brush maintenance, and there is better efficiency over a wider range of
power. We eventually decided against all these reasons and chose to purchase a brush
motor because we did not want to deal with the external commutation circuitry. Also, we
had been in touch with Danaher Motion and they said that there were no commutation
controls available for our specific application and designing one would be difficult (not a
good idea to depend on one vendor).
        One of the other variables we also had to consider was how to gear the motor. All
leading powered wheelchairs carry heavy “transmissions” to do their gearing. We also
required some type of gearhead because the particular motor we ended up choosing
(more on this below) ran at 1750 RPM and 18 in lbs of torque. In order to get to the 250
in lbs of required torque we would need a 16:1 gearhead and this would put our final
drive around 100 RPM. A little slower than we originally required, but it is far more
important to sacrifice power for torque. If the wheelchair can hold itself on a 12 degree
slope, but goes 2 mph instead of 3 mph, we thought it was more important that it would
hold itself on the slope, even if it went slower. We ended up getting an 8:1 gearhead
from Nuegart that would work well with our application (see “Issues” for why we chose
only 8:1 and for the gearhead specifics).

Final Decision
        As mentioned above, we ended up choosing a ½ HP, 24 V, DC, permanent
magnet, brush motor, mainly because of the ease of control. The particular motor model
we chose was dependent on a handful of variables needed by Pacific Scientific. The most
important of these were the drive inertia as well as any requirements we had for electric
braking (which, as it turns out, the mechanical team did not design there drive system for
because it was too complicated). The inertia was worked out by the mechanical
engineers. The second major requirement, deceleration on a grade, was worked out by
our team on the ramp in the MDL using a tape measure and a stopwatch. Our team had
been collaborating with Blair Davis (an RPI graduate) of A&D Devices (part of Pacific
Scientific, owned by Danaher Motion) in Albany. We used Danaher Motion’s eMotion
software in collaboration with Blair and the above mentioned variables to pick the

specific model (see the end of this motor section for specifications). They also helped us
locate the 8:1 gearhead.

        The biggest problems we encountered were with the gearhead. At first we wanted
the 16:1 gearhead, but this would have had to be acquired from the manufacturer in
Germany. This would have taken about 5 weeks to receive and we did not have time for
that wait. However, there was an 8:1 gearhead in a warehouse out in Rochester and it
would not take very long to receive. We ended up going with this gearhead because the
mechanical team said it would not be a problem to do an extra 2:1 gearing with their two
drive sprockets. Also, the mechanical team did not seem to understand how large the
motor and gearhead together would be. We gave them the dimensions weeks before
either of the items arrived and they even made a cardboard mockup of the drivetrain.
They said it would not be a problem to fit the motor and gearhead. Although there was a
small wrench throw in the gears when we found out a ½” adapter plate was required for
the motor and gearhead to fit together, it still should have fit fine. However, when we all
arrived for the final presentation they said it was way too big and there was no way it
would fit. This was somewhat upsetting and we also had to deal with a handful of unfair
remarks as to how large the motor and gearhead actually were. Unfortunately they never
did the research into current power wheelchair drivetrains, as they would have found out
that our setup was actually much smaller and lighter than what is out there. (questionable

        To next semester’s team we strongly urge them to use a brushless motor.
Although it will require extra circuitry outside the motor, it will make for a much smaller
and lighter motor. For example, the equivalent Pacific Scientific DC brushless motor is
only 5 inches long. Also, we recommend that next semester’s team order all their parts as
soon as possible. Some of these drive components are hard to find and take a while to

Concluding Remarks
        All in all, we accomplished almost everything that we wanted in the way of the
motor. We did as much research as we had time to and, after speaking with
professionals, eventually picked out the proper motor for our application. We were able
to control the motor the way we wanted on the test bench and we think it would work
well with next semester’s team with the above recommendations taken into account. We
also learned a great deal about power engineering, invaluable information that really
would not have been learned in any other way. The one regret we have is that the motor
never ended up being mounted to the chair.

Figure 7: Sample brushless DC motor circuit (Q1-Q6 are the phase switches (for
reverse/regeneration) and the Hall effect sensor is, in effect, the commutator or phase sensor.)

7. Controller

        The controller was designed with three goals in mind: effective control of the
Brush DC motor, low cost, and adaptability to other motor and input options. The
resulting Pulse Width Modulator configuration can easily drive two outputs for an H-
Bridge amplifier, and can be adopted for use with either one –5 to 5 volt input range, or
two 0 to 5 volt ranges, as well as other peak voltages. It is also very cheap, so it can be
expanded to multiple channels, such as for a servo motor. The amplifier section uses
MOSFET transistors, which are high performance, and quite common on the market.
The ones chosen for this amplifier also accept a 5 volt trigger, so additional pre-amplifier
stages are unnecessary. Additionally, the entire controller was even smaller physically
than expected, so mounting on the wheelchair unobtrusively will be trivial.

Design Overview
        The controller section consists of the modulation circuitry, and the amplifier
circuitry. The modulation circuit accepts a signal within a voltage range, and outputs a
square wave pulse of proportional duty time. It does this using a timer driver to produce
a triangular periodic wave, which is combined with the steady voltage input signal using
a comparator. This yields the desired wave at a configurable constant frequency. The
amplifier section accepts the PWM signal, and drives the motor output with the same
signal at higher voltage and current.

        In order to provide the wheelchair with variable speed control, an appropriate
motor controller must be present. A motor controller uses a signal generator and an
amplifier to first create the drive signal, and then amplify it. Motor controllers in general
use DC input, and output a form of high frequency pulsing DC. The high frequency of
the output wave into an inductive load (motor) causes the output to appear as a constant
power output. This output is then modulated by the control signal on the low power side
of the amplifier to produce the actual desired output. In this way, either DC or AC output
can be easily achieved, allowing control of any motor type. This model is used to
minimize weight and cost, as the transistors involved are efficient and commonly
Analog input – The controller must accept one analog signal per direction
Stable and filtered output – It is important that transients in the input be kept to a
minimum in order to ensure smooth operation
Adjustable gains – The sensitivity of the system must be adjustable by the user
Low loss – The amplifier must have a high enough efficiency at all output levels to
ensure long battery life. Further, it cannot waste energy when the wheelchair is not in
Total variability – The motor output should vary smoothly and linearly across all output
High stall torque – The amplifier must be able to provide enough current for the motor to
operate at it’s rated stall torque
High reliabilitygraceful failure – All anticipated failure modes should either be
anticipated by the controller or clearly indicated when they occur

Regenerative braking – The controller should be capable of regenerative braking during
long deceleration periods.
Indicators – The indicators must clearly show the status of operation, including remaining
battery capacity, for both charging and operating modes
Adjustable governor – The controller should allow the user to set a maximum output

Experimental Data
        The controller section has not been tested with the 20A motor, but every aspect of
it has been bench tested, including load testing. However, no thermal testing has been
done, all tests were done at room temperature, and transient loads.
        The timer has been shown to operate over the 500Hz to 6KHz range, with no
degradation in output. The frequency is easily adjustable using a variable resistor, and
does not affect the amplitude of the wave. The comparator can drive one or two
MOSFET’s on one output, suitable for an H-Bridge configuration, with a change in
resistor value to increase output current. The MOSFET’s have no noticeable distortion or
other artifacts across the frequency range tested. Additionally, a low pass filter on the
input can reduce transients in the signal, and thus load on the motor, but no tuning has
been done to determine the amount of filtering that would be appropriate. The sensitivity
of the system should be adjusted using the sensor amplifier stage.

Design Verification
        The design appears to work satisfactorily and satisfy all of the design criteria.
Future options for the controller, such as filtering, bi-directional control, and servo motor
control can be accomplished without major design changes. The controller also appears
to have high enough performance that other closed loop control, such as a speed limiter,
could be implemented with relative ease.

8. Controller (contingency plan)

       When we began to approach the end of the semester, a degree of concern was
expressed about whether the electrical group would be able to deliver what we said we
would be able to deliver. Professor Alben as well as the mechanical engineers wanted a
way to be able to test their portion of the chair without having to wait for our controller.

        We saw that the mechanical group only wanted a way to propel the chair forward,
so we began to look into those possibilities. We thought at first that a simple single pole
single throw switch could work. We then looked into the possibility of using a high
capacity potentiometer to handle the current going to the motor. We felt that if we could
vary a resistance, that we could make the motor go as fast or as slow as we wanted to.
Finally we also looked into the possibility of a commercially available micro controller.
We thought that this could be on the expensive side, but we decided to look into it


                           Figure 8: Micro II Scooter Controller

        Above is a picture of the Micro II Scooter Micro controller. This micro controller
had many options that we wanted to implement, and the cost was relatively low as well.
At $135, we knew that this controller would work, and that we would be able to mount it
relatively easily, with the mounting holes provided. In addition it had more than enough
capacity to handle the power that we wanted to run through it and furthermore, it ran on
24 volts, which is what we have supplied for the motor anyway. The wiring diagram for
the Micro controller is located in Appendix A.

        The amplified sensor signal was used in place of the throttle potentiometer in an
attempt to test the force sensors with the micro controller. The controller was unable to
use this signal as an input and a ”throttle wig-wag” error was produced. This was due to
the fact that the potentiometer must have a 3.5V output at the wiper when the controller is
turned on. This is the neutral voltage for the controller and it will not produce an output at
this voltage. We then attempted to use the sensor in place of the high speed
potentiometer. The thought behind this was the throttle could be set to neutral at the start
and then engaged to full power when the chair was in motion. The force sensor – acting
as the high speed potentiometer – would then be used to control the maximum speed that
the wheelchair would move. When the user isn’t pushing, the sensor should output 0V
and the controller should have no output. When the sensor outputs a larger voltage, the
speed of the wheelchair should increase proportional to that voltage. This, however, was
not what happened in actuality and the force sensors were not effectively tested with the
        It is recommended that the next group contact the manufacturers of this controller
to determine if a sensor input instead of a potentiometer can be used to drive the

9. Engineering Analysis


Modeling the force sensors and differential amplifier circuit were done in PSPICE. The
force sensor was a full bridge circuit with resistances of 1200 ohms. Since the output of
these two legs were really not simply +10mV and -10mV, a differential amplifier4 circuit
must be used. The differential op-amp circuit was used to amplify the signal to a 0-5V
level so that signal loss in the sliprings would be reduced. Additionally, the controller
needed a voltage on that order.

                        Figure 9: Force sensor and differential op-amp circuit

The force sensor bridge (left) is an accurate model of the FUTEK force sensors
purchased. Since the input resistance was 1200ohms, then each leg is equal to that as the
bridge has 2 in series and in parallel with each other. Varying on of the resistors from
1190ohms to 1210ohms created an output voltage of 10mV to -10mV.
Resistors of large magnitudes were used for the differential op-amp as smaller ones
would disrupt the signal out of the relatively low-resistance force sensor. 4.7 meg-ohm
and 10 k-ohm resistors were used to create a gain of 470. The chart below shows how as
the force on the sensor changes the resistance, the output of the sensor varies from -10mV
to 10mV. Also, the output of the amplifier varies from 4.7V to -4.7V.

    differential amplifier circuit information found at http://www.ecircuitcenter.com/Circuits/opdif/opdif.htm

                     Figure 10: Output of force sensors and amplifier

Matlab Simulation

        It was important to model the system to determine the effectiveness of the system
prior to the project completion. This could give useful information to determine whether
wave shaping of the user input would be required and if the motor would drive the
wheelchair as desired. Figure 11 is the Simulink model used to test the wheelchair control
system in MatLab. The user input to the system is modeled as a pulse generator. This
would emulate a steady user input on a flat surface.

         Figure 11: Matlab model of the power-assist wheelchair control system

        The force sensor used for the input will have minimal delays associated with it, so
this part of the control system was included with the user input. The output from the
pulse generator was set to a magnitude of 24mV to incorporate this sensor. The first gain
is used to model the signal amplifier located on the wheel. The gain of this amplifier is
approximately 100. The second gain models the H-bridge controller. This bridge has been
modeled as a simple gain with a magnitude of 10.

The motor dynamics can be modeled as shown in figure 12.

                              Figure 12: Motor Dynamics Model

        The value of L is negligible compared to the rest of the values in the model. The
equation from this block diagram then becomes:
                                  (s)              KT
                                                                                   (1)
                                Vin ( s ) JRs  ( BR  K B K T )
        The back EMF and torque constants are equal. The motor resistance R and the
motor inertia Jm are given in the motor nameplate data. The inertia due to the load was
calculated from the equation below:
                                                mr 2
                                           JL  2
where m is the mass of the person and wheelchair, r is the radius of the drive wheel, and
n is the gear ratio. The final mass of the chair was estimated to be approximately 100
pounds. The mass of the wheelchair user needs to be a minimum of 200 pounds and a
maximum of 300 pounds. The values of JL at masses ranging from 140kg to 175kg are
shown in the Simulation Calculations spreadsheet in Appendix A. The mass used in the
simulation was 150 kg which would allow for a 230 lb user.

        The damping coefficient B needed to be found experimentally. Equation (1) can
                                                                         BR  K T
                          ( s)      K                     KT
be simplified further to                 where K               and a             . The value of
                         Vin (s) s  a                      JR              JR
K is known. Vin will be a step input that can be represented as A/s. The following
equations were used to determine the value of a from the experimental data:
                                                   K        A
                                        (s)           *
                                                 sa s
                                 ss  lim s 0 ( s *  ( s )) 
where A is the current draw of the motor at the steady state rpm (A is the magnitude of
Vin !).

        The current draw and the rpm of the motor were measured with the power supply
and stroboscope, respectively, at 12, 18, and 24V. Three trials were performed for each
voltage. The value of K, the rpm, and the average current value at each of these voltages

were substituted into the above equation to determine a. The results from this experiment
are shown in the Simulation Calculations spreadsheet in Appendix A. (see comments on
the model given on a separate sheet at the end, before appendices)

        The remaining portion of the model converts the angular velocity of the wheel
into the total distance that the wheelchair covers during the wheelchair’s operation. The
angular velocity needs to first be divided by the gear ratio to determine the angular
velocity of the drive wheel. The velocity then needs to be integrated and multiplied by the
circumference of the wheel to determine the displacement of this wheel. The distance this
wheel travels will also be the distance the whole wheelchair travels.

        Figure 13 shows the motor’s angular velocity for a user input of 1 second on, 2
seconds off over a 15 second time period. Figure 14 shows the velocity for an input
signal of .5 seconds on, 1 second off to show the effects at a higher frequency. With a
simple step input in the simulation, the motor velocity is approximately 1800 RPM. As
can be seen in the figures below, the motor velocity never reaches this maximum value of
1800. Both figures show large swings in velocity as the user grabs onto and lets go of the
wheel, which could lead to very jerky movement of the wheelchair. This is something
that needs to be avoided at all costs since this system should be invisible to the user while
they are using the wheelchair.

                 Figure 13: Motor Velocity (1 second on, 2 seconds off)

                 Figure 14: Motor Velocity (.5 second on, 1 second off)

        One possible solution to the problem would be to provide some wave shaping to
the amplified sensor signal. The new system would provide a smoother input to the
controller and produce a less jerky movement. Figure 15 shows the new model.

           Figure 15: Matlab model of the control system with wave shaping

        The value of b will depend on the circuitry involved. Figures 16 and 17 show the
effects of two separate values of b.

              Figure 16: Motor Velocity (1 second on, 2 seconds off, b=0.5)

             Figure 17: Motor Velocity (1 second on, 2 seconds off, b=0.75)

        Lower values of b will result in smaller differences between the peaks and valleys
in the motor output. This will, however, also increase the stopping time of the wheelchair.
Higher values of b result in a decrease in the maximum output of motor. In Figure 17, for
example, the motor RPM never exceeds 1100 RPM when it should reach almost 1500
RPM. Figure 18 provides some explanation of this phenomenon.

                 Figure 18: Unmodified and Modified Input to the Controller
        Smaller values of b will result in a faster system (from theory, the opposite is
true)in a more underdamped system. This will result in faster rise times of the system and
slightly faster fall times, but this difference will be minimal.

        The input from the user will definitely need to have some wave shaping involved
for the system to be invisible to the user. If this is not present, the wheelchair will not
have a smooth operation. The value of b will need to be experimented with more to find
the optimal value for the system (something more than a low pass filter is needed). It may
also be beneficial to perform another experiment to find the damping coefficient of the
motor. The values obtained in the experiment conducted this semester were very different
for the different voltages. The value at 24V was used in the simulation model, but this
may not have been a correct choice.

10. Regulations and Environment

FDA Regulations

The following is a list summarizing the federal regulations with regard to wheelchair
accessibility in public places. This information comes from the Americans with
Disabilities Act home page (http://www.usdoj.gov/crt/ada/adahom1.htm).

         Ground and Floor Surfaces.

         General: Ground and floor surfaces along accessible routes and in
         accessible rooms and spaces including floors, walks, ramps, stairs, and curb
         ramps, shall be stable, firm, slip-resistant.

         Changes in Level: Changes in level up to 1/4 in (6 mm) may be vertical
         and without edge treatment. Changes in level between 1/4 in and 1/2 in (6
         mm and 13 mm) shall be beveled with a slope no greater than 1:2.

         Carpet: If carpet or carpet tile is used on a ground or floor surface, then it
         shall be securely attached; have a firm cushion, pad, or backing, or no
         cushion or pad; and have a level loop, textured loop, level cut pile, or level
         cut/uncut pile texture. The maximum pile thickness shall be 1/2 in (13 mm).
         Exposed edges of carpet shall be fastened to floor surfaces and have trim
         along the entire length of the exposed edge.

         Curb Ramps

         Location: Curb ramps shall be provided wherever an accessible route
         crosses a curb.

         Slope: Transitions from ramps to walks, gutters, or streets shall be flush and
         free of abrupt changes. Maximum slopes of adjoining gutters, road surface
         immediately adjacent to the curb ramp, or accessible route shall not exceed

         Width: The minimum width of a curb ramp shall be 36 in (915 mm),
         exclusive of flared sides.


         General: Any part of an accessible route with a slope greater than 1:20
         shall be considered a ramp

         Slope and Rise: The least possible slope shall be used for any ramp. The
         maximum slope of a ramp in new construction shall be 1:12. The maximum
         rise for any run shall be 30 in (760 mm).

         a) Ramps: Curb ramps and interior or exterior ramps to be constructed on
         sites or in existing buildings or facilities where space limitations prohibit the
         use of a 1:12 slope or less may have slopes and rises as follows:

            (i) A slope between 1:10 and 1:12 is allowed for a maximum rise of 6
            (ii) A slope between 1:8 and 1:10 is allowed for a maximum rise of 3
            inches. A slope steeper than 1:8 is not allowed.

     It seems that these regulations are what drives the requirements for the design of a wheel
chair. Any motor used would need to be able to provide enough power to drive the chair and
at least a 250 lb passenger up the ramps as they are described in the “Slope and Rise” section
that is shown above.

Environmental Concerns

        The main environmental concern of the power assist wheelchair is the battery.
Batteries contain metals and acids that could be harmful to the environment and the
people in the nearby area if they are not disposed of properly. Fortunately, all of the
battery chemistries researched can be recycled. There are many places to recycle used
batteries. For example, I entered my hometown into this website:
http://www.rbrc.org/consumer/uslocate.html and 5 locations in Chicopee, MA showed
up. Public Works departments, small electronics stores, and cell phone stores are the most
common places to drop off used small batteries for recycling. Larger sealed lead acid
batteries can be recycled where they are purchased.
        The batteries are also not usually simply disposed of. They are usually
disassembled or broken open, the chemicals inside neutralized, and the metals recovered
and resold to battery manufacturers5. In some cases with Lithium ion batteries, the battery
goes through a process to render it non hazardous and placed in an environmental
protection facility.
        The electronic components may have a small affect on the environment. Some of
these components can contain some metals such as lead and mercury which can be
harmful to people. The chances of this occurring are very low, but if a large amount of
these materials are disposed of in one area, there may be environmental problems. Even
these materials can be recycled, though this process is far less extensive as the battery
recycling programs are. The components, metals, and even solder from printed circuit
boards can be retrieved from the boards and reused. The company that would
manufacture the wheelchair would probably need to send the components to be recycled,
however, since there are few areas where this is done.

11. Summary and Conclusions

       Each part of this system was tested individually on the bench. All parts worked
successfully as individual pieces, however, the integration of all of these parts was not
accomplished. If more time was available, we are confident that the whole system would
function properly.
Overall we have the following recommendations:
    A more robust and less expensive sensor
    Wireless transmitter for sensor signal
    Research further Ni-MH batteries and regenerative braking
    Use brushless DC motor with higher power density
    Use wave shaping to smooth user signal

    All information obtained from http://www.batteryrecycling.com/

A few comments about the motor and load model. In the equation on page 25 of the
          ss  lims 0 ( s * ( s)) 
 A is the amplitude of the input Vin , not the motor current, as is stated in the text. Also,
since  ss was found for motor with no load, K and a relate to motor J and B only. For
motor with load, calculate K and a using the equations:
                            BR  K T
         K  T and a                   where J and B are changed to include load inertia
              JR                JR
and load friction reflected to motor through the gear ratio. Now, you computed the load
inertia using the mass of chair and person. But the load friction can only be determined
by experiment, which, I think, the ME’s did. Overall a will be smaller and 1/ a , the time
constant, much larger than what is reported here. Also, K will be smaller.

IV. Appendix A

                                  Project Costs
Force Sensors

4 Sensors - $360.00


Circuit Board and Chemicals - $31.03
Brushes (Nylon Bolts/Wire) - $13.00

Motor and Gearhead

Motor - $345.00
Gearhead - $815.00


Microcontroller - $135.00
Op-amps - $20.00


User Input Sensor Information

                                Motor Information

1750 RPM, 1/2 hp, 24 Vdc
Low Voltage Rated PMDC


Speed (RPM)                                 1750
Voltage (DC)                                24
HP                                          1/2
NEMA Mounting Face                          56C
Enclosure                                   TENV
Continuous Current (Amps)                   19.5
Continuous Torque (lb-in)                   18.0
Peak Current (De Mag) (Amps.)               n/a
Torque Constant (lb-in/Amp)                 1.01
Resistance (Ohms)                           0.10
Inertia (lb-in^2)                           4.4
Inductance (millihenries)                   0.38
Length (inches)                             10.13
Weight (lbs)                                21
Replacement Brush Part Number
2 per motor
Tachometer Adaptable                        No

Important Features:

      NEMA 56C face with NEMA 56 removable base
      Class H insulation
      UL Recognized (UL 1004, File E61960)
      Designed for use with batteries
      Highly efficient
      For constant speed, motors are operated directly from a battery with no motor
       control interface.
      Low voltage motor controls are readily available. Pacific Scientific does not sell
       low voltage controls.

Torque Speed Curve at 24 Vdc:

Gearhead Information - WPLS 90, 8:1 gear ratio

Micro II Scooter Controller Connection Diagram

                                  Simulation Calculations

The table below was used for reference to determine what size person (in pounds) would
correspond with the mass values (in kg) in the left column.

Total     Wheelchair Person
Mass (kg) Weight (lb) Weight (lb)
      140        100            208
      145        100            219
      150        100            230
      155        100            241
      160        100            252
      165        100            263
      170        100            274
      175        100            285

  Below is the calculation used to determine the motor inertia from given motor nameplate data.

            Rated Motor Inertia
            4.4 lb*in^2
            2.2lb = 1kg
            .0254m = 1in
 Jmotor = 0.0012903 (kg*m^2)

 The equation and calculations below are to determine the load inertia at different kg values. The
                  total inertia at the motor shaft is given in the table on the right.
         Inertia Calculations

           Load Inertia
       r = 0.1016      (4in = .1016m)
       n = 16
Mass (kg)       J                     Total Inertia
      140    0.005645                    0.006935
      145    0.005847                    0.007137
      150    0.006048                    0.007339
      155    0.006250                    0.007540
      160    0.006452                    0.007742
      165    0.006653                    0.007944
      170    0.006855                    0.008145
      175    0.007056                    0.008347

The equation and calculations below are to determine the damping of the motor and to convert the
                                  torque constant to SI units.

Motor Ratings                          Conversion Factors
Kt = 1.01 lb-in/Amp                    1lb = 4.448N
       Kt = 0.114109                   1in = .0254m
            B = Kt*I/w
                Experiment Data                                      Calculations
  Volts       Amps     Ave. Amps             RPM       rpm/A      Kt     Kt/rpm/A B (N*m*rad/sec)
   12          1.8                           1000
   12          1.8    1.766666667            1000      566.04   0.1141   0.000202   0.001926043
   12          1.7                           1000
   18          1.9                           1500
   18          1.9    1.866666667            1500      803.57   0.1141   0.000142   0.00135671
   18          1.8                           1500
   24         1.95                           1850
   24         1.95    1.933333333            1850      956.90   0.1141   0.000119   0.001139322
   24          1.9                           1850

            Kt           B             R
            0.114109     0.001926      0.1

The table below was used to find the variable values for the Matlab simulation. The K2 values are
                              the Kt/JR and the a values are B/JR.

Mass (kg)       J     K2               a
      140    0.005645       164.5296          2.7972
      145    0.005847       159.8818          2.7182
      150    0.006048       155.4895          2.6435
      155    0.006250       151.3320          2.5728
      160    0.006452       147.3911          2.5058
      165    0.006653       143.6502          2.4422
      170    0.006855       140.0945          2.3818
      175    0.007056       136.7106          2.3243

MatLab code

%Give variables values

%Run Simulation
%To change the length of time of the simulation, open wheelv3, goto simulation,
%simulation parameters and change the stop time to the desired time (in seconds).
%To change the input signal, double click on the Pulse Generator and adjust
%the parameters as desired.

%Plot the rpm of the motor vs. time
xlabel('Time (sec)')
ylabel('Motor Velocity (RPM)')

%Plot the wheelchair displacement vs. time
%The value of d is divided by 60 since the motor velocity is measured in rpm.
%This will now be the distance traveled over the desired time period.
xlabel('Time (sec)')
ylabel('Wheelchair position (m)')

V. Appendix B
                                      Nate Diller
                                      16 BELLE AVE, TROY, NY 12180
                                           PHONE: 518 505 8302
                                          EMAIL: DILLEN@RPI.EDU

       To apply experience in live sound, engineering, and system design to event production
       and equipment installation.

       Rensselaer Polytechnic Institute                                           Troy, NY
           B.S. Computer and Systems Engineering, May 2003
               Concentration in Automated Control Systems

Work Experience:
      UPAC Sound, RPI Union                 1999 – Present                 Troy, NY
      Equipment Manager                     2002 – Present
          Purchase and installation of new equipment for two existing sound rigs
          Equipment troubleshooting and maintenance
      Rental Coordinator                    2001 – 2002
          Interact with event coordinators
          Allocate equipment for club rentals
      Team Leader                           2000 – Present
          Manage event production as well as setup and takedown

         Front House engineer for a wide variety of acts
         Stage setup and mike placement for a wide variety of instruments and applications
         Equipment troubleshooting both in the lab and during productions
         Monitor engineer for rock performances
         Extensive experience with computer hardware and software configuration
         Construction of mechanical and electrical systems

       Introduction to Engineering Design                                       Summer 2001
             Research, design, prototype, and develop a product
             Head designer – coordinate research and evaluation of various subsystems
       Signals and Systems                                                      Summer 2001
             Introduction to signal processing, both continuous and discreet, design and
                optimization of automated control systems
   Electric Circuits; Discrete Time Systems; Computer Architectures, Networks, and Operating
   Systems; Embedded Control; Dynamics

Activities and Interests:
    Beta Phi chapter of Alpha Sigma Phi fraternity, House Manager
            o Manage maintenance budget of $6000 per year
            o Coordinate capital improvement efforts
    Design of a sophisticated File System (extensive subsystem of an Operating System)

 Home Address                  School Address                 School Phone: (518) 276 – 2303
                                                              Home Phone: (860) 233 - 0700
 46 Sycamore Road              107 Sunset Terrace             Email: hydeh@rpi.edu
 West Hartford, CT 06117       Troy, NY 12180                 Website: http://www.rpi.edu/~hydeh       H. Holbrook
Hyde III

                           To obtain a B.S. dual degree in Electrical Engineering and Computer and Systems Engineering at
                           Rensselaer Polytechnic Institute and to pursue a career in computers/electronics/programming possibly
 Objective                 leading to management status in the automotive, aeronautical or interactive information technology
                           product design and/or services.
                           Summer of 2002
 Work                          Software engineer for the Commercial Power Systems group at United Technologies Fuel Cells

 Experience                Summer of 2001
                               New Service Technician for Connecticut Light and Power, part of Northeast Utilities
                           Summer of 2000
                               Laboratory Technician at Loctite North American Headquarters (Rocky Hill, CT).
                           Summer of 1999
                               Automotive Mechanic at Prospect Foreign Car Center, Inc. (West Hartford, CT). A twenty-year-old,
                                well-regarded automotive repair facility.
                           Summer of 1998
 Community                     Loaves and Fishes Soup Kitchen (Hartford, CT).
                           1999 School Year – Present
                               Food pantry services at local Watervliet Church (Troy, NY).
                           1999 – Present
 Honors and                     Advanced Placement Scholar.
 Achievements                    Honor Society at Kingswood-Oxford School (West Hartford, CT).
                                Captain of Varsity skiing and J.V. tennis teams at Kingswood-Oxford School.
                                Brother, past Secretary, past Ritual Chairman, and current House Manager of the New York
                                 Sigma Chapter of Phi Kappa Theta at Rensselaer Polytechnic Institute.
                                RPI Dean’s List student
                                Accepted into an RPI Undergraduate Research Program (URP) dealing with network and web
                                 administration second semester sophomore year
                                Accepted into another URP dealing with Ground Penetrating Radar (GPR) DSP data modeling
                                 first semester junior year, associated with CenSSIS
                           2000 – Present, Rensselaer Polytechnic Institute – Troy, NY
 Education                      Working on a dual major in Electrical and Computer Systems Engineering.
                                2.75 of 4.0 cumulative GPA
                                1995 – 1999, Kingswood-Oxford School – West Hartford, CT
                                Most Microsoft applications, Unix, Linux, Pro Engineer, SolidWorks 2000, Waterloo Maple, Most
                                 Adobe programs, HTML 4.0+, CSS, some Java/Javascript, C/C++, software design, most 3D
 Relevant                        fractal design programs, computer hardware maintenance, computer construction, Windows
 Computer                        networking, Buffalo interface, pSpice (circuit design), B2 Spice A/D 2000 (circuit design),
                                 Mathworks MATLAB, Minitab, Geophysical RADAN (DSP GPR software), Visual Basic, Visual
 Skills                          Basic for Applications, Rockwell RSView32, SQL, Oracle SQL Plus
                                Data Structures and Algorithms                 Laboratory Introduction to Embedded Control
 Relevant                       Engineering Graphics and Computer              Electric Circuits
                                 Automated Design (CAD)                   
 Classes                        Fields and Waves
                                                                                 Computer Components and Operations
                                                                                Signals and Systems
                                Introduction to Engineering Design             Computer Architecture, Networks, and Operating
                                Digital Electronics                             Systems
                                Computer Hardware Design                       Computer Graphics
                                Advanced knowledge of foreign and domestic mechanical/electronic automotive repair
 Other Skills                   Advancing knowledge of integrated circuits (components, construction)
 and Interests                   Advanced knowledge of the Motorola MC 68HC11 microprocessor (software and hardware
                                RPI Formula SAE race car team member
                                Novice hang glider, expert (backcountry, alpine) skier, mountaineer, hiker, mountain biker
                                Member (and website designer) of Hoen Racing Rally Team ( http://www.hoenracing.com )
                                Restored 1965 Jeep CJ-5, restored 1963 Sunbeam Alpine Series II

                 Member of Society of Automotive Engineers
                 Tool Manager of the Rensselaer Student Autoshop (RSAS)
                 Active member of RallyRPI

References:                Available upon request.


To top