Px4-autopilot: F450 with Pixhawk 3 Pro v1.7.0-rc3 crash

Created on 30 Nov 2017 · 9Comments · Source: PX4/PX4-Autopilot

Flight of DJI F450 with Pixhawk 3 Pro autopilot using v1.7.0-rc3 (66d4a1b3)

The pilot reported the vehicle took off, switched to Auto, reached the first waypoint and it started to climb above 300 meters, the pilot commanded RTL through QGC, the vehicle started to move towards home position but kept gaining altitude. The pilot sent land command through QGC but the vehicle didn't respond, it only responded to RC roll so the pilot engaged the kill switch.

Log

@dagar

bug

Source

r0gelion

All 9 comments

What history does this vehicle have? Does it have previous successful flights?

In mission mode when it was stuck in the climb the commanded throttle (blue line) went to the minimum (12%).

Combined with the relatively high PWM_MIN of 1230 us the minimum motor outputs were around 1300us.

Which ESCs does this vehicle have? Do they need to be calibrated? I would start with a short manual (stabilized) checkout flight.

dagar on 30 Nov 2017

These are the ESCs used on this vehicle. All calibrations performed.
p3p_esc

r0gelion on 30 Nov 2017

@dagar Is the green line a setpoint (z)position for the current mission? thanks :smiley:

xdwgood on 1 Dec 2017

@MaEtUgR @Stifael Could you please have a look?

LorenzMeier on 2 Dec 2017

👍1

Let's go over this again in detail tomorrow, but so far it looks like a physical setup issue. In the previous flight logs the nominal pwm output in a climb, hover, or forward flight is higher than the "flyaway" values in this log.

dagar on 8 Dec 2017

@Stifael Could you take another pass?

LorenzMeier on 8 Dec 2017

👍1

There is not much I can add to what @dagar already mentioned.
I looked at the entire pipeline:
for auto:
position setpoint -> velocity setpoint -> thrust setpoint -> thrust magnitude from vehicle_rates_septoint_s -> actuator controls -> actuator outputs
for manual:
stick inputs -> velocity setpoints -> ... (as above)

All the above messages seem to be correct in terms of flight intention and control outputs (they all go basically to the minimum).
@r0gelion do you know why the pwm minimum is set to 1230? Was it set that high on purpose? Is there anything that changed in between the crash and the previous flights such as weight? In the log files from previous flights the average pwm output is around 1500, but in the crash log the pwm output is in average at around 1325.

Stifael on 8 Dec 2017

@Stifael The PWM min is set to 1230 because it's the minimal value of the DJI ESCs and they don't support calibration.
If fact, that could be the issue: the ESCs are not the DJI ones. With this value, the lifting system produces too much thrust and can't descend.