[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: your mail
On Wed, Mar 15, 2000 at 05:00:20PM -0500, Joyce Poon wrote:
>
> I'm leaving my room now. Should be back by 6pm.
Here is a copy of my lab entry from last night:
Thu Mar 16 EST 2000
- After discussion with Joyce, bought more pulleys, belts, and anything
that looked like it might work in place of the current pulley/belt
system. Cindy could not be reached. Interesting items purchased
are:
- A belt matching the cogs that I purchased yesterday. I could
only find one such belt (!), and it is too wide and would need to
be cut, but its teeth seem to match. Did not test.
- NB: small motors are great for melting holes into Lexan when
power drills are available.
- Many small metal pulleys of different shapes and sizes. Tried
some on one of the small test motors I bought. Odd. The test
motor is not powerful enough to move the robot. In fact, neither
of the two types of test motors I bought can move Zeta. This
suggests that our current motors are not so over-powered as they
may have seemed (to me). However, the belt gripped well-enough to
the brass pulley that it stalled the motor. This is good.
- Removed existing pulleys from actual motors. After some work, a
new brass pulley seems to be attached reasonably well to one of the
motors. It is a threaded pulley and was attached by holding it
to the motor shaft while the motor is turned on. It wound-in far
enough to hold, but Cindy will need to do something more permanent
later. Using this pulley with the small red belts provides pretty
decent acceleration. The belt will break in about 5s if the wheel
is stalled and the motor is running at max speed. Testing an
unloaded wheel, it ran for better than 5minutes before the belt
broke spontaneously. These results need to be improved, but now
at least the limiting factor is our belts and not the pulley.
Ran a comparative test with the other pulleys. As usual, the
heat-shrink was eaten away in seconds. The belts seem to have
very inconsistent strengths: the first one tested lasted for
quite a while even though it was being turned against the threaded
steel. The new brass pulleys definately have much better
acceleration. I believe that, on average, the belts last an
order of magnitude longer, but I don't have the time to test this
for quantitative results (esp. given belt inconsistencies). I will
attach the new pulleys onto the motors and hopefully it will work
well enough that we can finally show the TA a moving robot.
- During routine testing:
a) I managed to break both opto-encoders. They
are wired such that it is possible to connect the power supply to where
the comp line should be. I believe I did this on both of them and
thus destroyed them.
b) One of the power transistors fell out due to vibration caused by the
wheels turning. It will need to be resoldered.
c) The wires from the power supply board are annoying and poor. At the
least, they should have connectors added nearer to the power supply
board. Perhaps they should be replaced altogether.
d) We should drill some more holes for the opto-encoder wires so that
they can travel more naturally through the robot.
e) Currently the large motor batteries are located over the two low-
friction gliders. This is by far a non-optimal place. The main
robot wheels actually slip on the floor in my res room. I'm going
to see what Cindy thinks of moving the batteries so that they're over
the main wheels and provide some more traction.
f) I resoldered the power transistor that fell off. Another fell off.
It fell onto the Lexan. I was originally of the opinion that heat
sinks on the power transistors would not be necessary. After observing
one of them melt the Lexan, I have changed my mind. We should add
heat sinks.
g) The batteries really really need to be moved to the sides ontop of
the main wheels. If the main wheels keep on slipping, feedback from
the encoder will be plain wrong.
h) Front motor battery "F" needs to be recharged. I measured it at 4.5V
during PWM control.
i) Numbers on motors need to be re-etched. Should also etch in numbers
on Zeta's base corresponding to which impact sensor connects to that
location.
j) Front impact sensor towards left side not functioning consistently.
I think it is a problem with the circuit board and not with the
physical sensor.
I'm going to Supreme now to see if I can buy some new opto-encoders.
I'll ask to see the data sheets. I broke 'em, I'll fix 'em. If you
want to do only beacon (and IR sensors?) during this lab period, I'll
deal with the falling transistors, and quirky impact sensor, too.
--
Signature withheld by request of author.