Fixed wing nosedives and crashes after takeoff

Hi everybody,

as a student team, we are in the process of developing a prandtlplane drone design. Sadly, it crashed during our test flight just after it had taken off. This is the second time it has crashed, in quite a similar way to the previous crash.
This time, we were able to retrieve the logs and capture the crash from multiple views. In the logs, the takeoff happens around 3.22, the crash happens around 3.26.

From our own analysis, we have some theories why this might happen:

  • The standard PID values for controlling the pitch angle don’t work for our design, as setpoint and estimated value for the pitch angle don’t really overlap.
  • By taking off in stabilized mode, which in hindsight wasn’t a logical decision, the pilot needs to respond too quickly when the plane takes off. Perhaps there is some kind of integrator windup which causes sufficient pitch downwards when the plane took off.
  • The pixhawk flight controller is vibrating too much, giving wrong sensor inputs, therefore making bad controller decisions.
  • The plane can simply not generate enough lift when the angle of attack changed after takeoff (i.e. the electronics seem fine, but there could be an aerodynamics problem).

Any other explanations or advice for future testing would be welcome.
Thanks in advance

I’d check your mechanical setup first.

Roll and roll sp are inverse so confirm that the deflection surfaces are moving the correct way in manual and then in stabilized.

Decrease the authority physically of the pitching deflection surfaces as a first test. Then set the maximum pitch in the parameter to a low level if your pilot is unable to keep it at low ascent angle (but preference is having the pilot do that without autopilot intervention which is generally the best path.

If you have lift AoA concerns you should do an Ansys analysis to get some rough CFD numbers for airspeeds, CG, etc as that should dictate where you are at with power systems and target test speeds.

Vibration is generally not that difficult to solve, mechanically you can isolate the autopilot but first look if you have prop imbalances or the autopilot is too close to the motor, etc. I’d also use the newest master PX4 and enable the dynamic notch filter as that will take care of things in the software side. Enable fft for logging (high rate logging.

Start there but you shouldn’t have too much difficulty getting this to fly, just need some basic covered first (stall speed, etc).