RTL causes copter to behave erratically

I had a crash a few weeks ago with the 3D Robotics Iris. This was after receiving the unit back from 3DR after it was supposedly fixed from a previous crash.

dropbox.com/s/uvyk8ioy4aop4 … eCrash.mp4

After putting the unit back together, I set up a simple 3 waypoint mission. Something was a little off at the start. I would write the mission to the copter from mission planner and then read it back to confirm everything was ok. Well, in mission planner it would change things up from the original mission, everything from removing the LAND command, removing waypoints, etc. After talking with 3DR personnel, I decided to fly the unit anyways.

It was obvious from the start that something was off. The copter slowly was heading in the opposite direction of the first waypoint or any waypoints for that matter. So, once it was clear that it was not following the mission, I flipped the switch to RTL. At that point it changed directions, sped up and crashed into a tree in spectacular fashion.

Surprising the only damage was to the camera gimbal unit and it was easily repaired. I took it out the next day and flew it manually in Alt Hold mode with no issues what so ever. Looking over the log file I can’t see anything that would cause the copter to perform in this way. I am thinking that there is an issue in the pixhawk autopilot and that wasn’t found or fixed after the first crash. Please help. Thanks.

Copter – 3DR Iris
Autopilot – Pixhawk PX-4
GPS – U-Blox LEA-6H-0-002
Compass – above listed GPS has built-in compass
Radio – FS-TH9X PCM1024
Extras – Tarot Gimbal Kit, tall legs, GoPro Hero 3+
APM Copter firmware version – ArduCopter V3.1.5

You should upload the flash log from your Pixhawk SD card. Your tlog has some corruption or just very strange things like :

  1. the log shows that you were consuming 25 amps the whole time your bird was on the ground.

  2. All the time leading up to your takeoff your battery voltage was increasing hitting close to 14 volts right at the takeoff.

If your using a 3S battery then there is something wrong. The people that help out here could provide more direction if you copy and upload the flash log.


Attached is the file we finally extracted from the flight that day

The inertial navigation was definitely not behaving. GPS velocity doesn’t even remotely match NTUN.VelX/NTUN.VelY

This means one of 3 things:

  • GPS problems
  • Compass problems
  • Vibration problems

So the question is, what caused it?

  • I can rule out vibration problems off the bat by comparing IMU and IMU2 - they’re sampled at different rates, so when aliasing due to vibration is present they each behave differently.
  • Probably no GPS problems - your auto mission started with a decent sat count and HDOP

That leaves compass. I don’t think it was your calibration - your offsets are (-105, 63, 8), and the best offsets I could come up with were (-134, 124, -46). That difference produces a worst-case 8-degree heading error on your log, which wouldn’t cause the issue. So, could your compass be oriented wrong on the copter?

Yesterday I crashed my drone because I used RTL.

Are there known problem about this RTL issue or are this due to bad settings?

Is there someone who can analyze my bin or tlog file to trace the cause of this crash?

Thanks in advance!

Please don’t hijack others’ threads, make your own thread.