After aborted Auto Tune arming copter flips on take off

After a lot of success with Auto Tuning I’ve been anxious to try it out with my heavy lift copter. There was a bit of a breeze so I wasn’t sure if it would have time to do the tuning before it drifted too far away but that turned out to be the case. On my last attempt of auto tuning the copter would drop during each cycle of the twitching. The distance it would fall increased each time so finally I flicked it out of Auto Tune and back to Stabilize. The I put it in Alt Hold to see how it would go and that was doing the same drop as during the Auto Tunes. I decided it wasn’t worth continuing as Auto Tune certainly isn’t going to work if Alt Hold is flakey. I should mention that when I started flying Alt Hold and Loiter were working perfectly.

All of that aside, the real issue happened after I landed it. The copter disarmed and it was in Stabilize mode. I had a look at the gimbal etc and then moved back to take off again. When I armed the copter the motors came up to full speed and the copter flipped on to it’s back. The crash detection must have kicked in as the motors stopped shortly after flipping.

I know that the throttle was not up as I arm with the same stick as the throttle. But even that shouldn’t allow the copter to spin up at full speed. And no, this was not the props just spinning up when armed.

I’d be most interested to know what caused the copter to give full throttle and flip when I armed it. Did it have something to do with the previous Auto Tuning? Or was it related to the Alt-Hold issues? Certainly, something was not right.

Flash log attached if anyone has time to have a look at the end of it to see what actually happened.

Using a new Pixhawk with Arducopter ver 3.1

There seems to be some kind of barometer problem. You can see that it’s fine at the beginning of the log but after about 6min it starts glitching and by the end of the log it’s glitching all the time. This causes the inertial navigation altitude estimate to climb and in fact it breaks the altitude fence.

I’ve actually never seen this kind of bad barometer behaviour. The closest I’ve seen is jumps caused by light hitting the barometer but even then it’s not normally this persistent.

The pixhawk is still within it’s case of course?

By the way, we’ve recently added a barometer vs inertial nav check into ArduCopter. I think with AC3.1.1 this would have have failed to arm and if you had telemetry enabled it would have at least printed a message on the HUD.

The pixhawk is still within it’s case of course?[/quote]

Yes. There’s also a tinted dome over the electronics. It was around 2.30pm so the sun was fairly high.

This is only the second flight I’ve done since swapping the APM out for the Pixhawk. It was flying great up to this point.

I can’t try again for a few days as the flip broke four of the eight props and I only have three spares. I’m using Xoar wood props and they break quite easily. I’ve ordered T-Motor CF props to replace them with as the Xoars are just too hard to source.

Thanks. I’ll make sure to load that before I do another flight.

The baro issue is being addressed with a software update. The issue has been resolved in ArduCopter-3.1.2-rc1 so I’ll mark this thread as solved.

Thanks to the Developers for all their hard work.