Aircraft losing altitude in FBWB

I’m in the process of tuning the Plane autopilot for FBWB and Loiter.

I have a custom-built electric glider, twin engine, 20 ft span. The goal is to do autonomous endurance flights.

I went through the TECS tuning guide in FBWA and flew in Loiter for 1.5 h. Everything went fine. I then switched back to FBWB to do some straight legs at constant airspeed, which worked fine as well for a few legs, but then the aircraft suddenly started losing 40 m of altitude without any commanded input. When I then commanded max-climb rate, the aircraft did not respond. I switched to Manual and landed.

Does anyone have an idea what might be going on or where to look at?


params_pre_flight.param (18.1 KB)
params_post_flight.param (18.1 KB)
logs: 2023-04-27 11-47-56.bin - Google Drive

Some more info:

  • ArduPlane V4.3.4 (f73e812f)
  • 6 m Span
  • Electric twin motor
  • 4S Li-Ion
  • Pixhawk 6x
  • SDP3x Airspeed Sensor
  • SunnySky 3508 motor
  • ExpressLRS Diversity RC Receiver

The events are plotted below. Altitude starts dropping at the 2:50:30 mark.

As long as you have the Throttle above 1500 it maintains the height without problem. The moment you lowered the throttle it started to lose altitude.

Logically in FBWB mode it is supposed to maintain altitude but that is not what happens.

Maybe your ARSPD_FBW_MIN of 8m/s is insufficient to maintain level flight.

Another (important) aspect to consider; did you perform the TECS adjustment?

Thanks Lano.

we have flown at 8.0 m/s in FBWA level flight before. Stall speed is around 6 m/s.

We did go through the TECS tuning procedure.

Based on what I read here, there might be a glitch with certain receivers and THROTTLE_NUDGE = 1 (which is the case).

Looking at the logs, erratic oscillations does show in demanded acceleration as soon as throttle (demanded airspeed) is reduced.

I will try turning it off and try again, however I do like the ability to control airspeed with the throttle stick…

Looking at these records I find it difficult to draw conclusions. I can’t see level flight (maintained) at 8m/s and without touching the elevator.

There are many moments of throttle at zero and nose down but Throttle at zero and level flight (without touching the elevator) are practically not seen in the logs.

Personally I think that to get consistent data for TECS adjustment it is necessary to maintain the flight conditions sobre unos 8-10 segundos.

I had to push the nose down to stop it from climbing in FBWA. That’s because TRIM_PITCH_CD wasn’t tweaked yet. Here is 8 m/s level flight, the airframe is capable of that.

In any case, I’m not sure how is explains the oscillations in dspdem.