APM 2.6 quad crash when RTL invoked

This is a new problem for me. I’ve built several Quads and hex’s using APMs with very few issues and nothing like this. Testing a new build with a Steadidrone QU4D airframe. Flies perfect in loiter, alt hold, and manual. I invoked an RTL and it turned into a ball and fell out of the sky. Luckily the snow is deep and fluffy so it cushioned the fall. As it went down I switched back to stabilize and it actually leveled itself and started to fly out of the snow at which point the battery fell off and unplugged.

APM 2.6 (genuine 3DR)
3DR GPS/compass
3DR telem radio
3DR power module

In the log just after time stamp 1600, RTL is invoked and I can see the commanded pitch and roll strongly diverging from actual pitch and roll which says to me mechanical failure, but all motors work, ESCs good and the props were tight. HDOP is good, so I think this was a NAV PID or similar issue rather than loss of GPS position data. This is about as deep as my log analysis knowledge goes. Any help appreciated. Thanks!

Sorry to be annoying but do you have the dataflash logs? It’s much easier to see the errors in the dataflash logs rather than the tlogs.

It looks like the RTL was triggered by the CH7 switch.

I had a quick look over the parameters and nothing too serious but I noticed a couple of things:
The FS_THR_VALUE is 900 which means it may not catch a radio failsafe. Have you tested that your throttle failsafe is working ok according to the wiki (copter.ardupilot.com/wiki/throttle-failsafe/ )? On that page in the “Flight controller set-up” section you’ll find it says “Adjust the FSPwm field to be: above 910”.
The WPNAV_LOIT_SPEED is very low at only 200 (i.e. 2m/s). Not dangerous but very slow!

By the way, the “APM” that steadidrone sells are copies actually. You only mentioned that your frame was a steadidrone so I guess you got your APM directly from 3dr or another retailer…but just wanted to point that out.

Yes I understand that Steadidrone ships RCTimer branded boards with their RTF products. This particular bird had the RCTimer board ripped out and thrown in the trash bin since the telemetry port was bad. In the future, I only plan to use 3DR boards (RCTimer stuff has been nothing but problems). I have another QU4D that is getting a Pixhawk, still ironing out some bugs on that one. :smiley:

I did invoke RTL using CH7 switch. My radio failsafe does work, the throttle PWM value drops below 900 when the TX is switched off and is a crap radio so no programming options.

I also did slow down the WP_NAV loiter speed since I had a near fly away due to bad GPS data (I rushed taking off and didn’t get good GPS lock) so I slowed that down so that a GPS induced flyaway would be slower during this testing.

Attached is the log from the APM on board memory.

Thanks to all who take the time to offer support on these forums, without you there would be no Arducopter!

Well I have been experimenting and testing over the past week and have come to the conclusion that the compass was being interfered with, causing the loss of control and crash. I pulled my entire quad apart and put copper tape under the APM as well as under the compass/GPS module. The compassmot is now less than 2% (was 8%) and everything flies perfectly. Not sure if that counts as solved, but it seemed to do the trick.