What Happens When Your Test Stand Gets Out of Control
A More Sophisticated Robot (a *little* more sophisticated). This is a series of
web pages about Skinny. Motor Control, Sensors, and RF Control are discussed.
This is the basic frame
This is Skinny on his birthday. Skinny was built to explore navigation and PID
control using a robot. Skinny has ended up being four robots (or more), each
of which have different motor control implementations and different control
schemes. Look through these pages and the links to these pages to see how Skinny
grows. These pages were generated from about August '99 to present and were made
mostly because of a collaboration with Bill Ruehl. The two of us decided to make a
PID motor controller. After we got the controller half way working, we needed a robot to
put it on. I went the simpler route, and Bill went a much more sophisticated route. See some of Bill's robots at
Bill Ruehl's Web page
The Skinny robots use a plywood octagonal base (1/8") with aluminum angles for
reinforcement and for mounting heavy components. The wheels are turned on a lathe from
1/4" plexiglas and
the 'tires' are the O rings for a Eureka vacuum (series 1100). The hubs for the
wheels are brass and also turned on a lathe. The tail wheel is adapted from a
model airplane tail wheel (about $2 in a hobby shop) for a .60 wire axle. The
battery in this picture is a normal 7.2 volt NiCad RC battery for RC cars. Various
versions of Skinny use this battery or a lead acid Gel Cell. The cost for the
batteries are about the same - but a Gel Cell will get you 12 volts which is
a bit heftier for some motors. The motors keep changing on the different models and I will
mention them as we go along.
This is the frame with motor controller
The second picture is a little fuzzy (my last digital camera sucked with shots like this -
it did macros well, but in-between shots were rarely in good focus).
The motors are running (which is why the pattern is a bit more blurry than usual) and
you can just make out the Duty Cycle in the LCD.
The 16C73A is in the lower part of the picture in a 40 pin ZIF on a protoboard. The
connector by the wheel on the right side is for a ribbon cable to run to the second
deck of the 'bot when it becomes one. All the wiring going to the controller, etc will
be headed off there. The encoder sensors are mounted, but not
hooked up right now. The circuitry for them will go on the blank board in the
front. There are two possible motor sensors. The QRB1114QT-ND (3.5 mm focal length) which is the one
in the picture and the SY124 which is a 1 mm focal length thingy that Bill used in
his flexor sensors. I have a hard time with the little rascals and prefer the QRB sensor.
Skinny's new nerve system
Here is skinny with a redesigned motor controller. Notice the difference in the quality of
the photo? My new camera is much better (Sony Mavica - kewl!). Anyway, back to the 'bot: All the experimentation with the
previous design came up with this. You can set Skinny's PID, endpoints, speed, direction and most of his important
parameters through the PC (or through a serial connection with most MCUs).
The board is a standard Radio Shack protoboard
with a 16C73A (PIC chip), L293D (H bridge chip), 74HC04 (inverter), 74HCT14 (Schmitt trigger for
shaping the encoder waveform). Wouldn't it be nice if someone made all this in
Skinny at Tiffany's
Skinny trying on his second deck with a BasicX module on top. Just for show - lots to
do yet before we get this far. However, I did stick a Homebrew Stamp on Skinny - hooked
it to the motor controller umbilical and did a few turns. See the video below.
NEW! Real Movie of Skinny
Skinny is doing a four turn 20' run and returning in the above movie. His previous
video was off by six inches. He is a lot closer here. Improvements included a higher
encoder density (90 stripes or segments), locked anti-phase PWM, and discovering
that the left wheel was .025" smaller than the right wheel (made several inches
difference over a 20' course!).
Skinny Gets an Evil Sister Sarah #1 (SS1)
Bill asked me if I would like to make the mockup mobile so I made a second platform for
the mockup electronics. The following picture is of the encoder pickup that Bill made.
Bill's Flexor Encoder Design Using an SY124 (one of first prototypes after I had
bent it up a bit!)
Because the Pittman motors were a bit bigger than the Hsiang
motors I used for Skinny, I put them on the bottom. I also used a 12 volt Gel Cell for
power - got a couple of them from BG Micro for a good price.
The Pittman motors are faster than the Hsiang motors so I dropped the wheel size down
to 4" - makes the whole 'bot look smaller, but it really isn't. Sarah is shown here with
her umbilical running off to a PC somewhere. She has a little problem right now with her
encoder pattern so it may be a while before she is running as smoothly as Skinny. In any
case, here's a video of her taking a P (the course is a P shaped course). The video starts
out a bit late - Sarah actually starts at my feet in the video, so when she comes back, she
really lands on the same square that she started from. I also managed to cut the video off
before she gets fully turned around for her next run. In the video, note the dog bowls (empty) and chew
toys on the floor. These are natural obstacles left by our retrievers. One of whom is watching
the 'bot from behind me while I'm shooting this video. She doesn't quite know what to make of
the robots in her kitchen.
SS1 Takes a Turn (REAL Movie)