by JussiH » Sun Aug 18, 2013 1:24 pm
Hi Christof.
Sorry to see your beatiful creation shattered. Hope you will find time to cut some new parts and get it airborne soon again.
Looking at the photo of the ESC above, that is a blown FET. It can be quite powerful when they short out at high currents - They literally blow up and release a lot of gas in very short time. I have seen this happening due to overcurrent (because of wrongly set CL or shunt params), reversed voltage and feed wire short, but never that violent. 3 fully charged capacitors that is subjected to a short could also release some serious energy in a splitsecond, making matters worse. If you can find them, they could provide clues to what happened. If the ESC is shorted at high current then the ESC becomes so hot that components begin to desolder - and this can happen with a few seconds.
I am partial to thinking that it was the ESC that failed because of a input short and the burnt windings on the motor suggests that it was subjected to a permanent short across two phases. The current spike at >60A is another indicator that something went bad.
If the motor locked up, then the current limiter and bad detect functions should have been able to limit the current accordingly and disarmed the ESC instead of burning the motor by applying high current across 2 phases for long enough to burn the motor windings.
The fact that it flipped and inverted, could have been because of the current spike when the ESC shorted up. The voltage dropped quite dramatically - when that happens it usually ends up with a crash because it leaves the rest of the motors starved for power. Its the same thing that can happen if you forgot the time or didnt hear the battery alarm and the battery starves while correction is beeing applied.
There is nothing in the motor log to suggest anything but a well balanced and tuned vehicle and I dont see any signs of motor problems.
The fact that you mention problems with the HF heading, could indicate problems with magnetometers. It is a shortcoming in the logging system that we cant see the heading estimation, since that would help us know if there was a problem with heading beeing estimated incorrectly.
But looking at the mag reading in the log, the MagZ estimation jumps by more than 1 when you power up and there is a massive spike on Y and Z at the time of the voltage drop - this could indicate a power distribution problem resulting in a subsequent short that killed the ESC. I am not sure, since there is not much angular changes going on, but it seems to me that the Mag_Z is saturated and not showing any signs of change with horisontal angle changes.
I would go over the whole power distribution and look if there was signs of wear of insulation or burn damage that could indicate if a short was brought on by wire fatigue or other forms of wear damage. I dont see that kind of damage beeing consistant with a FW failure. But if it was, then its something we need to get to the bottom of soon.
Having said that, we have been seeing minor problems in ESC32 pertaining to startup lately, so its possible that something has crept in in one of the many recent updates of ESC32 FW that affects timing or switching.
We are looking at it, but I think its prudent that unless you specifically need "Servo mode" or one of the other recent additions, you are better off going with an earlier revision that is tried and tested for something as precious as that copter. Going with the latest always means you are running the risk of beeing the guy that finds out something is wrong with it. If you just need to run PWM with standard current limiting, you dont even need to update from the factory FW.