My Kite3 has only done 2 flights but at the end of transition there is an alarming pitch down. According to the logs it looks like it’s when control is fully handed over to plane mode and a pitch up that has developed is suddenly corrected. The plane flies nicely in manual, FBWA and cruise. I believe there was at least one other post on this topic but I can’t locate them now.
Note flat desired pitch while actual pitch increases and the elevator is attempting to keep the nose down. Then after transition finishes there is a rapid pitch down correction. Interestingly I can’t find anything at the instant the pitch begins to reverse. The thrust power remains roughly the same.
I had a theory that the prop wash was causing a Venturi effect under the tail and the low pressure was pulling it down so I setup a workshop test. I did more off camera testing and actually printed a part to locate the cord exactly on the thrust line. Every time there is quite a strong pitch up. The previous version of this plane is virtually identical but I haven’t noticed this issue.
I could reprint the motor attach cup with the thrust line angled to push the nose down a bit. I’m wondering if I should reduce the transition delay (5s) after the speed is reached? Any other ideas welcome. The log file is in my recent post about ESC start delay nearly causing a crash.
EDIT Having checked the logs, this tendency was also in the previous model except it only got to about 8 degrees instead of 12 in this version.
I have recently experience something with some similarities, but while hand launching a fixed wing plane (not VTOL). It may be there is no relation, but lets take a look. In my case, throttle causes a significant pitch up. When thrown, the airspeed is low and speed_scaleris 2.1+. The pitch PID controller responds very quickly to the pitch up with a significant response from the integral. As the airspeed increases speed_scaler drops to ~0.4 and the integral is very, very slow to back off. I had one launch that wasn’t straight and as a fixed wing plane with no ailerons (rudder/dihedral) it was slow to recover wings level. The nose ended up level instead of up by the time the wings were leveled. Now the strong down contribution of the integral caused the nose to continue down and enough altitude was lost that, though it did start to climb again, a tree became a problem.
The potential parallel is pitch up with application of throttle followed by pitch down as airspeed increases.
The pitch integral coefficient and limit were at the default value. There are a number of components to a solution for my case:
Larger integral coefficient for faster recovery
Smaller integral limit (MUCH smaller than default)
Throttle → elevator feedforward
That last item isn’t an Ardupilot feature at this time, but it would address trim changes due to throttle.
If your only doing stabilized flights then it might be better to set the elevator trim for transition rather than cruise. Transition is the jumping off point where there is no I term, once your flying the pitch controller can keep on top the trim. But this would mean the trim was a little off at cruise speed which would be annoying if you wanted to fly in manual.
