VTOL: No Roll and no RTL in FBWA

During tilt-rotor quadplane maiden flight transition from QLoiter to FBWA went smoothly - but there was no aileron movement on roll input during the flight in FBWA nor in RTL mode. Elevon movement in FBWA was ok - plane gained altitude swiftly. Can anyone help to shed a light what could be the root cause here? Flight log can be found here: Dropbox - log_150_2021-11-14-15-39-30.bin - Simplify your life
Ailerons are servos 1 and 3 - while elevon is servo2.
In Manual mode on the ground ailerons seem to move correctly, responding to roll input.

After analyzing the logs - the root cause seems to be the fact that airspeed did not reach past FBW_MIN_ARSPD (17 m/s) during QLOITER → FBWA transition - so LEVEL_ROLL_LIMIT 5 degrees was in effect (causing plane only to react within 5 degrees on roll input). Another story is the fact that GPS speed was around +6-10m/s greater than airspeed - even when flying against the wind, which was around 4-6m/s (gps speed 17/ms vs airspeed 12m/s when flying against the wind for example). Im suspecting that airspeed sensor might be a bit damaged or blocked by dirt - ie showing lower values than it should.

However this behaviour seems to be a valid recipe for a VTOL plane flyaway scenario - as when FBW_MIN_ARSPD is not reached during transition (due faulty airspeed sensor for example), plane continues to fly out and LEVEL_ROLL_LIMIT will be in effect - so roll and also RTL won’t work - and plane is keeping its heading and continues to fly out even past radiolink coverage. And the only option to counter that would be to set Q_TRANS_FAIL - which would force plane to enter QLAND after Q_TRANS_FAIL timeout seconds (and potentially crashing plane on landing if terrain below is unknown).

Are there any ideas how to mitigate risks for this potential flyaway scenario?

Hi Andreas,

You have already found out the reasons for limited roll angle during transition
(LEVEL_ROLL_LIMIT 5) and for unsuccessful transition due to a faulty airspeed measurement.

I completely agree with you that in such a case constellation activated Q_TRANS_FAIL could do more harm than good. In my opinion, it also does not make sense to restrict VTOLS roll angle so much during transition.

But I am surprised that LEVEL_ROLL_LIMIT was active at all with Q_OPTIONS 0. With bit 0 of Q_OPTIONS not set the LEVEL TRANSITION (Keep wings within LEVEL_ROLL_LIMIT) should not be active.
Maybe it was active in your flight because of your older Arduplane Dev version, otherwise it would be a bug.

You always come out of the transition with a backshift into a Q-mode. If you are sure that the aircraft is fast enough, and only then, you can also switch very briefly to manual and then immediately back again. This will immediately - without delay - switch to fixed-wing flight. Be careful if you are too slow !

The possibility of faulty airspeed sensors also worries me, especially partially closed pitot tubes. For safety reasons, I have set the following functions on one switch each on the transmitter:
RCn_OPTION 106 Disable Airspeed Use
RCn_OPTION 91 Airspeed Ratio Calibration

Rolf

Thanks @Rolf for your input. Q_TRANS_FAIL transitioning immediately to QLAND is really risky behaviour - I wonder if it could be made to transition either to QRTL or at least to QHOVER/QLOITER in a first stage and QLAND triggered by critical voltage failsafe in second stage? That would give pilot some time to react and adjust - before going to autolanding approach. Also if Q_TRANS_FAIL is set to max 20secs - plane probably would not get too far and could possibly perform successful QRTL also. Currently faulty or misconfigured airspeed sensor would lead to possible flyaway - as without minimum airspeed reached RTL would not work - and there are very few options to actually recover the plane stuck in a transition phase.

@tridge has this potential quadplane flyaway case been spotted before?