Saturday night we had everything wired. We checked, double checked and triple checked all the wiring (by no less that three people). We booted up the robot and started trying to load code. Short story at this point is lots of trouble with cRio gave up (around 1:00 am) and went home but everything was recieving power including jags.
Come back to work today.
Swapped to new (translate known good that we have been using) cRio
Powered up machine (both classmate and robot) everything is fine
Imaged rio everything is fine
Loaded code (C++) everything is fine
Enabled driver station and poof went the jag
It got hot enough to completely melt the outer casing at the V+ connection. Take a look at the photos and if you can advise as to what might have happened I would be grateful so as not to repeat a mistake. No one was drilling anything and cRio is not mounted such that anything ‘should’ have fallen into the jag while swapping the rios. Will post pics as soon as they are approved…
Until you enabled the jag, you were never sending a PWM signal to it. I’m going to wager you got metal swarf in the jag, and the second you applied a signal it died.
To the best of my knowledge jags still do not have a conformal coating, so the smallest amount of metal inside them can be lethal. Victors are somewhat more reliable in that they have a conformal coating (someone correct me if the jags have changed).
So even though there was power (the lights were on on the jags but it was the no signal light) there was no control signal and that makes a difference?
Its really hard to predict what metal swarf will do, because it can short any two circuits together if they are close enough. Some swarf won’t do anything. Some will smoke as soon as power is applied. Sometimes, it will wait until you energize the power stage to convert itself to smoke.
You stated V+ but did you mean M+? That would make a little more sense as to why the Jag smoked when you enabled. Is there a possibility that the screws were loose and one of the terminals shifted into contact with the other one?
No Al, I did in fact mean the V+ connection. The M+ are clean. As soon as Delphi Admin approves photos will post them to this thread. As far as screws being lose when we removed the jag from the bot the connections were tight (the thought that crossed our minds was maybe too tight and perhaps a shaved thread?).
Can’t wait to see the pics then. It is also likely that the defect was present and on it’s way to catastrophic failure when you powered the robot. It was merely coincidental that you observed the result when you enabled. Was this in one of the new Jags? I haven’t opened one of the new ones, but there really isn’t much in the way of circuitry near the V+ terminal in the old ones. There is the 5 volt, three terminal regulator, that is towards the PWM connector and CAN Bus. Or you simply had what we call an ‘infant mortality’ failure of one of the FETs. This occurs when a device checks good, makes it through quality control testing but fails very early in it’s life cycle.
Finally got the photos approved (I don’t know why it takes so long, I mean it’s not like they are in the middle of build season or anything ) Ok, so take a look at the photos and see what you think. Just to be safe have taken all the remaining jags apart and blown them out with compressed air.
coach,
That appears to be very typical of foreign matter getting caught between the positive input terminal and other circuitry. It must have been spectacular. I would contact TI and see if they have any direction from here. If I was them, I would want to take a look a such a spectacular failure just to be sure that a particular part fail didn’t for some other reason.
The board that rsegrest posted is a slightly different version from the one shown on page 35 of the User Manual on the TI website. The damage is consistent with an excessively high current passing through the current sensing resistor, R35, causing it to vaporize itself and some of the adjacent copper track. Since there is no visible damage to the MOSFET’s, other than soot from the exploding resistor, it is not likely to be a shorted load that caused the fault current. It is likely that some conductive particle fell in and shorted some adjacent pins on one of the chips causing abnormal operation, say turning on both the upper and lower MOSFET’s. This would make the MOSFET’s try to short out the incomming 12 V supply. It is possible that the particle to have fallen in some time ago and lodged itself in a “benign” location then moved to where it could cause the Magic Smoke to be released.
Conformal coating is for environmental protection and is not normally considered a proper insulating material but at 12 V it is probably effective enough to make a difference.
We’ve actually recently had this problem, as long as multiple instances in the past. The first one I looked at had the MOSFETs shorted and the plastic casing around the MOSTFETs melted. Recently, one of the chips on a Jaguar shorted and burnt out, looking rather similar to your photos. From what I understand, the major cause cause for this is usually metal dust and debris. Metal dust can easily get into the electrical components, and since it is very small, it can get very deep inside some hardware. Since the dust is metal and therefore conductive, the result is a shorted circuit.
My teammates, mentors, and I discussed this issue earlier in the season, and now that we are mounting the electrical board, we are figuring out ways to prevent metal dust from getting into the system as well as protection from any debris or airborne objects. My idea was a curved “trunk” design that is hinged at the top, opening like a car trunk. Between the frame rods of the “shield” is a layer of aluminum wire mesh (cut from an old screen door) pulled tautly across the frame. On top of the wire mesh, which gives the structure its strength, is a thin layer of some sort of transparent plastic (like plasti-wrap, very thin lexan, etc.) to keep the dust out. I’m still in prototype phase of the design, but once it’s done, I’ll post some photos. This could be a good solution to problems like metal dust and damaged electronics.
In speaking with a number of teams while walking around the pit areas, it seems that they have less trouble with stuff falling in and shorting out the controllers if they are mounted on a vertical surface.
When you are first putting the robot together, you may want to cover up all the vents with tape.
Alternatively, use known bad controllers as “mechanical samples” to get the mounting holes marked and drilled (if you don’t pre-drill the panels). Once you are confident that all the drilling is done, replace the mechanical samples with the real ones. Of course, you will want to mark the mechanical samples prominently. I have done this at work with the IGBT’s at work when building up prototype inverters.
The new Jags come with this so leave it in place until you are ready to turn the robot on. Everyone should remember that controller fans blow down. So anything on the top or near the top or falling past the top will get sucked in and deposited within the controller. If you have to drill, cover the electronics. If you can’t cover then turn the robot on it’s side or top to insure the filings fall somewhere else. Vacuuming metallic dust is far superior to blowing the dust around with compressed air. The former takes the majority away, the later puts it somewhere/anywhere that you don’t want it to go.
Jaguars are notoriously known for unreliability. Our team has had many just give out during tests or competition. This year, we have switched to using all victors. they are much more reliable and durable. They also last much longer. Most of the Victors we are using, are over three years old, and even after extensive use, they still work like new. If you have the opportunity, I would recommend using only Victors. It will make your robot much more reliable. If you still use Jaguars, be very careful with them, you don’t want a blow out again.
Another tip is to make sure you keep Jaguars and any other components stored properly when not attached to a robot. That means putting them in their boxes, anti-static bags, etc. I organized all of our Victors last year and put them in a sealed plastic box (dubbed the Big Box-O-Victors). Haven’t had a problem with any of them since, after months of storage and nonuse.
) Ok, so take a look at the photos and see what you think. Just to be safe have taken all the remaining jags apart and blown them out with compressed air.