yes. I would recommend thermal paste (heat sink compound). You can eeven get it at radio shack.
Allen,
I question what you are using for current monitoring. If you are at anything less than full power, most current meters will not be able to accurately measure the changes in current from a PWM output. If that is the case you are running more closer to stall (64 amps) then you think. Even at 34 amps (max power) you are asking the motor to dissipate over 360 watts, at about 41% efficiency you are dumping over 200 watts in heat inside the motor. That is a lot of heat. You also may have loose connections on the motors (FP motors are very difficult to terminate properly). Heat generated at the connection migrates inside the motor and raise the temp on the thermal cutout.
Hi-
Different member from the same team. We were using a clip on amp meter and were taking the readings on the wire leads coming off the victor. We agree that the thermal protection was probably put into place for a reason and are going to be taking steps to reduce heat and current draw. We are definitely going to be using a higher reduction gearbox (we don’t need the speed, and that should help reduce the stress on the motor) and are considering the heat sinks.
The problem of overheating the Fischer-Price motors was one we were aware of from the beginning. We had considered building custom heat sinks but decided that wasn’t feasible with the machining capabilities we had. Thanks for pointing us to the pre-fab ones. It’s sure going to help us out. We are still wanting to find a way to obtain non-thermally protected motors. Although we don’t think the thermal protection will trip after we’ve made the changes, it was decided that we’d rather not have that limitation pop up on us in the middle of a match. Thanks for all the help, this would have stopped our robot dead in it’s tracks without your help.
[RIGHT]-Will[/RIGHT]
Al - We’re using a Fluke 336 clamp on ammeter to measure our DC motor currents. Am I correct in believing that it should read the average DC current in the pulsed waveform out of the Victor? Is there something better we should be using?
Bruce
Bruce,
The 336, as with most clamp on ammeters and VOMs are only good to 400 HZ. The output of the speed controller is a 2 kHz rep rate so the pulses are even shorter than that. That allows for a lot of error to build up at less than full throttle. I am guessing that you might be reading 50-60% of what is actually taking place. It is a good tool to check between motors and will give accurate results when the controller is producing full throttle. It is a tool many teams don’t have in their tool box but should.
Heat sinks most likely will not solve your problem. Take a look at the gear box motor mounting. Did you block any vent holes? If so see if there is a way to reestablish the motor venting. The small can motors rely on the integral fan to move air and provide cooling. If the motors are venting then the load must be reduced.
Al - Thanks. I knew the output of the Victor was pulsed, but didn’t realize the rep rate was that high (Note to self: Read the nice manual.) I was also thinking the meter would read the average current under the pulses, regardless of width, a holdover from my old “analog thinking.”(<sigh> I miss the days of the old Simpson 260.) 
Check the IFI specs. The 883 was about 2k output feq. The output feq of the 884 is low about 120. This was done to give control below 10%. Using a UEI digital clamp on amp meter VS an old Amprobe analog clamp on I’m getting less than 10% variation in current readings.
I looked through the specs on the 884 on the IFI site and don’t see the output pulse rep rate. have you got a link to where it might be? I’d like to demo this to the students with a scope (if I can get near the robot this last week.) Thanks.
Bruce
I looked on the IFI website for both the 883 and 884. Last fall there was a post in the IFI forum looking at the differences between the 2. IFI detail the differences in the switching feq. Seams they have cleaned the posts up for the 2006 season and it’s not there. Would have to be re asked or give IFI a call to confirm. There where some threads where this topic was discussed on Chief Delphi. Again check the motor mounting blocking the vent slots. There was another post concerning this.
This stuff is legally “non-flamable”. But that doesn’t mean you can’t make a 5 foot fire ball in your kitchen when your wife isn’t home. Nevermind… (Don’t worry - you really need to try to make it burn like that - your robot won’t make it go whoosh. It’s OK… really it is.)
Has anyone ever thought of making aluminum blocks to mount to the motor, and storing them in a cooler of dry ice between matches? Actually… you may loose most of the cooling by the time the match gets started… just a thought…
I’d just like to re-iterate this. It is probably too late for you to do anything about it, but the vent holes in the front of the FP motor are critical to its cooling.
Matt
Try reading the current after the breaker, going into the victor.
This is not going to help. Actually, even starting the match with ice cold motors doesn’t help much either.
The problem is that the Cp of metal is just not that high – meaning you can’t store that much heat energy in them unless a phase change is involved (i.e. melting the metal).
In my opinion, even heat sinks on the outside of the motor can are of only marginal usefulness. Yes, heat sinks will make your motors able to runs somewhat higher current in steady state, but FIRST is not really steady state. Our uses come in bursts. Heat transfer takes time (especially conducting heat out of the windings of the armature, through the air or the shaft to the outer can to the thermal grease to the cooling fins). By that time, your windings are toast.
Air blowing across the armature – now THAT is cooling. Keep your current down and your armature speed up (1/4 of stall torque) and you’ll be fine.
…you mileage may vary…
Joe J.
I thought is was the other way around. (I have gotten this goofed up before though.) What may be confusing is that the input PWM is 120 Hz as is the internal loop reading the input.
The difference between the probes is pretty good. Still if you want to have accurate current data you must use another method. i.e. if you want to correlate current and breaker heating.
After a diligent search of the archives, I did find this…
It lists the 883 at 2kHz. However this sheet compares the 883 and 884 as having different deadband characteristics and the 884 being programmable as to the deadband.
http://web.archive.org/web/20020805205656/www.innovationfirst.com/FIRSTRobotics/Victor_884.htm
OK, I’m confused. 
We found this:
But I’m still not sure what that means. Is it talking about a code loop rate, victor input sample rate, or victor output pulse rep rate? I think somebody’s gonna have to hook up a 'scope.
I believe that it’s the Victor sample rate. It states that you cannot send it data any faster than it’s refreesh rate, which is 8ms. Thoug I believe that a pulse width is longer than that anyway.
Victor 883 - 2000 Hz
Victor 884 - 120 Hz
Direct from the designers.
-JV
Thanks,
Now I just have to figure a way to tuck that into my brain for next year when the question comes upt again.
OK, that’s pretty authoritative! Thanks, John.