Crash during RTL (home at unexpected point)

Log Analysis -- deprecated Copter Logs -- deprecated
1

I just had a pretty bad crash with my quad. Its build with a Sky Hero Spyder and a Pixhawk (Arducopter 3.1) as a flight controller. I uploaded the last 30 seconds before the crash recorded by the GoPro.

[youtube]https://www.youtube.com/watch?v=xXa6YO1ShoI[/youtube]

The crash happened during a RTL triggered by a battery failsafe. It was only the second flight and the current sensor had not yet been calibrated, so the battery failsafe only relied on the battery voltage. In the attached logs the voltages seems to be at a ok level to be able to make it home. The quad really did not seem to just fall out of the sky, and actual roll and pitch seem to match desired roll and pitch.

Looking at the flight path on the map in mission planner I noticed that the home position seemed to be a about 70 meters away from the actual launch position. For reference I have also attached the previous flight where the home position is at the correct point. That is about where I launchd at the beginning of the second flight, too. The conclusion I have drawn is that the copter tried to return the the wrong home, wich would explain the strange flight path at the end.

I still am a bit confused though about the speed and the hight. To me it seems like the copter moved way to fast and a was loosing hight pretty fast.

The number of satellites and the hdop are at no point at levels which would explain the bad home position. So I am completely clueless what caused the hole mess.
Hopefully someone of you guys can help me figure this one out.

Thanks in advance
Gabriel

2

Have you checked the power module’s voltage readout compared to another volt meter? Maybe it’s lying…

Sent from my iPhone using Tapatalk

3

I looked at the log, and I can’t see exactly what happened. I would have more info if you had been using 3.2, because the EKF is much more intelligent.

However, I can tell you:

No problem with vibes. Very good PID tuning. No problem on GPS.

Your Mag1 which is the external is very very noisy. Practically unusable. This is not good. Strangely, the internal is much better.

On your earlier log, 42, it appears like at one point you were doing Alt-Hold tuning? There’s a section in the middle with really high throttle output noise, and corresponding Z-axis accel. Those periods with throttle output fluctuation show quite a lot of noise on the magnetometer, so I think this is the source of your problem.

The crash was simply the case that the copter didn’t know which direction it was facing, and tried to fly home, but headed in the wrong direction. The reason it didn’t know which way it was facing is likely due to these problems I highlighted with the mag.

It looks like you didn’t do the CompassMot procedure, this might have helped, though it’s hard to say for sure as the mag noise is really very bad.

I can’t comment on the home position, but I can tell you that the copter was not flying in the direction it wanted to.