Sudden rolling causing crashes

Here is the tie down setup: … 3.jpg?dl=0

Possibly, however you posted the same log twice and it doesn’t contain RCOUT logging.

If your hypothesis is correct, this is what you should see in the RCOUT log, when it is in the air:
As #3 slows for whatever reason a roll to the left will begin as will a pitch up. To counter this the Pixhawk will request more power from #3 and less from #4, thus RCOUT for 3 goes up and RCOUT for 4 goes down. It sounds like you are suggesting that #3 then suddenly “comes back to life” and seeing a high PWM request, goes balls to the wall. This catches #4 by surprise as it’s powered down to compensate, thus a roll to the right. However #4 should come back on REAL fast to offset the roll, and the PWM RCOUT to #3 should fall as soon as the actual roll approaches zero. You’d have to take a real close look at the event timing to see all this. So maybe.

So…You’ve swapped out that motor, swapped out that ESC and checked EVERY solder joint for the power feed of that arm?

Oh, one caution about checking motor operation when tied down…At low throttle settings the craft is very sensitive to its levelness. That is, if it’s not quite level it might not spin some motors up as it tries to level itself. Once the throttle gets shoved up a bit that issue goes away.

Try this link:

I will check the solder once again, but it seemed fine when I checked it last time.

The motor and ESC were swapped for one on a different arm in the previous config and the issue still occurred.

The RCOUT don’t appear to change as I would have expected in the log.

Was my tied down test suitable or did you have something else in mind?

Thanks :slight_smile:

So after spending some more time looking at the graphs of RCOUT, Volt & Curr I see no correlation between the current drop (voltage spike) and motor output values. My thinking is the current drop is related to the dipping/slowing down of #3 but the RCOUT doesn’t change. The roll is at a minimum at around line 1150 but the current drop is around line 1320; strange?

#4 RCOUT does seem low and continues to decline as the test goes on, whereas #3 increases; is there something in this? The others seem to be similar to their spin direction partner.

The point of a tied down test is to provoke an unambiguous failure that won’t result in damage, like falling out of the sky will. But its limitation is that you won’t see the Pixhawk doing its control thing to counteract whatever is failing because it’s resting flat on a surface. And to test your theory you need to see actual roll/pitches compared to what and when the Pixhawk is asking for. Both types of tests have their usefulness, but one can be more expensive to do than the the other.

I’m not sure if a low #4 is an issue or not since the hex is on a flat surface. It’s possible the Pixhawk was sensing a slight off-level and winding down #4 in an attempt to compensate.

Perhaps your university project should instead be on the intricacies of system troubleshooting instead of “just” flying something?