3.5 rc6 - bad stab r/p value caused disarm in midair and subsequent crash?

I had a crash this morning caused by inadvertently setting too low a stab r/p (neglected to check ch6 tuning knob was making the correct changes before taking off).

However, though the loss of control was due to the bad gains, the actual crash was caused by the copter disarming in midair. There was an EKF Primary error followed by a CRASH message (subsys 12, ecode 1). The logs end abruptly in midair - props stopped and it fell about 3m, fortunately with minimal damage.

Is this expected behaviour - ie does the code automatically disarm, even in midair, in the event of a total loss of control?

If not, what is going on?

Thanks very much, log file attached. 2017-05-20 07-28-08.bin (4.2 MB)

I’m pretty sure this is expected behaviour - looking at your logs, just before the crash detection there is a large divergence in lean angles caused by the low PID gains.


There’s some conflicting info on this page though - @rmackay9 's video states the lean divergence must be >20 degrees to trigger the crash detection, while the documentation says 30 degress… I’m not sure which it is.

Will crash check occur in mid air ?
I doubt a brownout .

Thanks very much - looking at the logs, at the moment of the crash check there was divergence of ~40 degrees between des and actual pitch. The divergence no doubt caused by my stick movement trying to regain control. So I suppose that was what caused the machine to pass the crash check.

The crash_check page says the logic was changed in 3.3 from the version in 3.1 which the video describes, so that information is not up to date.

Good to know - I wonder if I had not made such large control movements trying to deal with an apparently unresponsive machine, could a more gentle landing have been achieved?