Dataflash log corrupting after aggressive control inputs from autopilot

ArduPlane VTOL Plane
1

Hi all,

We have observed three separate instances of dataflash logs being corrupted after aggressive control inputs commanded by the autopilot in a quadplane flight configuration. Fortunately, none of these aggressive control inputs were fatal to the drone. However, we find it suspicious that logs are only being corrupted after aggressive control inputs (we have plenty of logs from healthy flights). Furthermore, it is frustrating because log analysis of these unhealthy flights may uncover why the autopilot is commanding such aggressive control inputs.

A bit of information about our drone:

  • Skywalker X8 quadplane airframe (7kg TOW)
  • ArduPlane ver 4.0.9
  • Matek F765 flight computer
  • T-Motor MN5212 kV420 motors
  • T-Motor T-80A ESC
  • The VTOL PID control loops and input shaping parameters have been tuned and shown robust performance in strong winds.
  • The EKF has been tuned a little, with most of the default values working well.
  • The notch filter(s) have not been tuned but will be before the next flight test.
  • The aircraft has flown without issue for approximately 4 flight hours.

More information about the latest dataflash log corruption:

During the final descent in a fully automatic mission, at approximately 5m above the ground an aggressive forward pitching motion was observed. This was quickly damped out and the aircraft continued its descent. Then at approximately 1.5m above the ground, another aggressive forward pitching motion was observed, this turned into an oscillation that was seemingly growing and most probably would have developed into a full instability if the motors had not been disarmed when the aircraft touched down.
When the SD card was retrieved for log analysis, the following message popped up on Mission Planner

Log Browse will not function correctly without FMT messages in your log. These appear to be missing from your log.

As mentioned before, dataflash logs have been corrupted on two other occasions after overly aggressive control inputs were commanded in fully automatic modes of flight.

All thoughts/suggestions welcome.

Thanks,
Dorian

2

During the final descent in a fully automatic mission, at approximately 5m above the ground an aggressive forward pitching motion was observed. This was quickly damped out and the aircraft continued its descent. Then at approximately 1.5m above the ground, another aggressive forward pitching motion was observed,
this turned into an oscillation that was seemingly growing and most probably would have developed into a full instability if the motors had not been disarmed when the aircraft touched down.
When the SD card was retrieved for log analysis, the following message popped up on Mission Planner

  Log Browse will not function correctly without FMT messages in your log. These appear to be missing from your log.

As mentioned before, dataflash logs have been corrupted on two other occasions after overly aggressive control inputs were commanded in fully automatic modes of flight.

Could you provide a both a valid log from the vehicle and one of the
corrupt logs, please? Proximate in time, by preference.

One thingthat could couple these two things is a bodgy power supply; we’ve
seen compasses glitch bases on power supply, and it can certainly cause SD
card issues.

Dorian

Peter

3

Hi Peter,

Thanks for the quick response!

Here is a link with a good flight and a corrupted flight from the same day:
https://drive.google.com/drive/folders/1pIpYYRoLresCIzl7EB5E7CrLp-wrcYcl?usp=sharing

I’ll look into any potential power distribution issues now.

Best,
Dorian