This year for the first time we had recurring problems blowing victors.
The (6) FETs on one direction of the “H” would all burn out for some reason. Luckily no underlying damage was done, so it was only a matter of replacing 90 cent FETs, but we haven’t come up with a definitive reason for the failure.
The obvious suspects are metal chips, but the victors seemed to be clean when they are pulled apart.
Any of you have unusual victor issues this year, or have other interesting theories on blowouts?
Oddly enough nothing in particular. One on our wench blew as we were doing a systems check before it went into the crate on the ship date and pulling the wench cable out by hand. Luckily, only one blew during actual competition, so that one on a drill was probably stressed, but not unusually so. It wasn’t during a pushing match.
We didn’t have any problems with victors we’d rebuilt, so I suppose it could have been a bad batch. I found it interesting that all the FETs would be burned out, but that may have been because we weren’t noticing when just one FET went and only recognized it when power to a motor was gone completely after the remaining FETs were overloaded.
Were you powering the fans and were they running at the time of failure? What exactly was the failure? Did the FETs short or open? Did the circuit breakers click prior to the failure?
We had at least two Victors fail on us. We had a few extras and didn’t have time to mess around with figuring out what was wrong, but, yes, we did have problems with them…
My team had no problems with the Victors this year. We have run the robot even harder than we did at competition once we got it back. Today we ran it for almost 2 hours straight with breaks every once in a while, pushing objects around, showing off, etc. and we still havent had any problems…even after carrying me around. I do know that we check to make sure our fans are running before we start up the bot for an event but they have been fine. I still havent had a Victor die on me (883’s or 884’s) though i have melted the pwm cables going to them .
-Aaron
[edit] Also what type of drive system are you running? Are the Victors dealing with high current loads often? [/edit]
We had no problems with the victors (or fuses) at all this year. On the other hand, our driving was done with 2 drills, the bosch gearbox and a 2wd setup - According to what I’ve been told, that setup puts a fraction of the strain 4wd and 6wd bots on the motors. We ran the same 2 motors for almost our entire build season, the sacramento regionals (9 seeding rounds, and a few quarterfinals), and the practice day at atlanta. We saw the bot was moving slower than usual, and the driver said it was reacting slower. We had 4 extra drills in boxes, and they didnt take too long to swap.
We had spare victors ready to go, because we were told (and read on CD) about the magic smoke. Hopefully innovationfirst doesnt change them for the 2005 season … and we can use the spares if we need to
We had a lot of problems with them at Chesapeake I believe. Twice I think, we extensively went back over the code and wiring and then it was like “Maybe we should try a different speed controller.” and the new one worked. I dunno if we ever figured out what it was though. Too busy trying to get the robot ready.
our fan blow on us once. It was very mysterious. We were testing out a camera on the robot by driving around just the day before. when we came in the next day, the fan just wouldn’t work. Nothing serious, just replaced it, but its very puzzling.
Which FET’s did you use to refurbish your Victors, seems the guys repairing them on the Battlebots forums are finding the matching replacements costing MUCH more than .90 ea??
Were the cases damaged? i.e. blown off or burned. Simple electrical failure of some of the FETs without case damage, may indicate a static problem. Were the same FETs involved in all cases with the same motor? That might be an indication of a bad brush assy. or a bad drive part in one direction.
No case damage at all. All damage was limited to the FETs alone. The underlying substrate was unaffected.
It almost always involved different motors/speed controllers each time it happened (globe, drill, CIM). We never had a problem with a rebuilt Victor, and we didn’t replace any of the motors this year. By Nationals all Victor problems had gone away, but we’d replaced/rebuilt a half dozen by then.
Static is a possibility.
Any ideas on tests we can run in the off-season? I’d like to make it into a research project for the students, but I don’t want to risk expensive mistakes.
The type of failure you describe is unusual from my experience. Even the number of devices replaced is unusual. Is there a possibility that your controllers were mishandled before install? i.e. students walking across carpets and picking up the controllers. Were they ever used in another application, testing or demo? Ever installed near bare metal or wire? Ever connected directly to a battery and motor outside of a robot without circuit breaker protection? The controllers I have experienced with failure due to over current or foreign material show charred and blown apart cases, high current types of damage and burned circuit boards. On our team we regularly use 8 or more controllers per robot. Over the years with both a competition and proto robot running near year round we have only killed 2 or 3 controllers in the years since they were introduced. (Wildstang has been around since 1996 but there were Tekins then and they failed all the time.) At least one of those was due to drill filings falling into the case. Have you disabled the dead zone in software? Is there a possibility that the controllers were always “on” just a little? Do you calibrate the controllers? Can you show a picture of how they are mounted in your robot?
There’s no sure way of always knowing how students handled the controller’s while the engineers weren’t in the lab. While I was there they were being careful, but there were a lot of new students this year. The lab has no carpet, but clothing could still generate static. As a related measure we didn’t have trouble with the IR sensors being blown by static electricity as some of the other teams I was helping did, but that was a different sub-group with a much more limited number of students.
The electrical layouts were reworked many times as mechanical designs changed, so there was a lot of controller handling (but no powering up).
Some of the controllers were taken off last year’s robot when we started to run shy, so those were reused.
They were never used or installed without circuit breaker protection, or near bare wires.
We’ve never blown victors before, so initially we thought it was foreign object debris, but your description of the damage caused to the Victor case and the lack of other evidence seems to reduce the probability of that as a cause.
The dead zone is restricted by software with a smooth increase in power outside the dead zone compatable with the measured Victor deadzone. We verified via the dashboard the PWM output as neutral when it should be. The new joysticks didn’t usually center properly (weak springs). The blown Victors were a mix of factory calibrated and team calibrated. We did calibrate all later ones, because we weren’t measuring full power out of the Victors in bench tests with the factory calibration.
We had an issue early on with the drill/CIM victors receiving 127 according to the dashboard, but sometimes passing low power through to the motors anyway. That could be a related symptom. That problem also seemed to be gone by Nationals.
The wiring photos I have don’t show the Victor mounts clearly. I’ll have to wait for the robot’s return from Nationals.
I’ll check on the actual FET price and source (retail or wholesale) and get back to you.
Mark,
This is all very puzzling…Is there any chance that the controllers that failed were installed backwards or two were wired to one motor? I know I sound strange on this but I am grasping at straws. Everything seems in order except the unusually high rate of failure. In my regular job I repair and maintain electronic broadcast equipment. Finding a smoking gun (cause to the effect) is what I am trained to do.
Thanks Al, I really appreciate your help with ideas, maybe especially the “straws!” We too are grasping at straws. I’m only leery of straws when implementing potential solutions starts adversely affecting the operation of the robot. I had to fight to keep the Victor’s on “brake” this year, because we used coast exclusively the last few years and that change was a natural suspect. Any issues with “brake” that you know of?
From your experience if one FET goes would the others naturally follow from the increased workload?
No miswiring that I’m aware of, all replacements were dropped in and wired the same. Backwards would be more likely than two to one motor which wasn’t physically possible.
In the build season before we tightened up our construction practices and student (& engineer) training, metal filings could have gotten into the Victors, so that’s what we blamed initally. We haven’t uncovered any evidence to that effect and that shouldn’t have been the cause on later replacements, but you never know for sure.
It’s really a concern of course for next year when we won’t want a repeat of the issue. I’ll get together with the elec. students and other engineers to see if we can add to your straws, and get any stories of static discharge while handling any of the equipment.
Mark,
Foreign objects leave little parts of themselves in and on components. An inspection under a bright light will usually reveal an arc mark (burned deposit) on a component leg or on the board.
Brake mode should not have any adverse effects unless a controller was in brake for a full match while other motors were back driving it while running. To my knowledge, brake mode turns on the lower half of the “H” bridge thus putting a short on the motor through the two bottom sets of FETs. As you suggested earlier, the FETs died in one line, I took that as one set of positive FETs and one set of negative FETs, the two of which are in series through the motor.
If the controller is trying to push a lot of current, then there might be a domino failure. The first FET just pops and then the second and third have to share the current. When one of those fails, the last one is suddenly running all the current and blows it’s case apart as there is no place for the magic smoke to go. If the controller were wired backwards, it is possible but I think unlikely, that FETs could be turned on across the battery input. They might be able to fire up in reverse polarity under these conditions and the failure might be instantaneous vaporization of the internal wiring causing an open FET. I think that the power input is reverse polarity protected but the motor output cannot for obvious reasons. That would also explain why only two sets of FETs were failing and not all four.
We (one of our electrician mentors) reinspected the blown FETs and in our cases, the FETs shorted, which means it wasn’t a case of one going then the others sharing current, etc. Something, and the backdriving or motor brush theory is a real possibility, causes all three in any one quadrant to blow simultaneously. We didn’t notice any burn or flash marks, but that doesn’t mean that it didn’t happen.
.