Vtol crash due to yaw error

At start of transition vehicle starts yawing right. Aborted the transition but the vehicle couldn’t stabilize itself because of big yaw error. Had a hard landing. max yaw rate is 20, max roll and pitch rates are 120. Couldn’t figure out why the vehicle couldn’t stabilize itself.

log link: https://review.px4.io/plot_app?log=67a73f20-7f04-41c1-a0a0-984f484b89e1

video link : https://youtu.be/EapXvsl9Omk

So, I don’t get any response from anyone in this forum. So having hard times fixing my problems with the PX4 firmware. I guess that is it. I will try ardupilot.

Had the same error today: https://logs.px4.io/plot_app?log=f12a1112-8cb4-4bad-b8e9-6a36bed561d4

I will create an issue on github for further discussions.

changing yaw related to wind or horizontal propeller Coriolis force effect. but autopilot couldn’t control yaw. yaw angular rate setpoint is zero but actuator_control_0 yaw show it changed for controlling yaw angle.

In general it doesn’t really make sense to limit the max yaw rate, yaw should rather be tuned by adapting MC_YAW_P (reducing it will lead to less yaw rate demanded). For VTOL I have that one usually at around 1.5.
Limiting the rate can make the control loop unstable (it will always be in saturation if limited to only 20° e.g.).

@dayjaby how does your vehicle perform in pure hover flights? Do you always have this large asymmetric load on the motors? If it’s a custom built vehicle, you may want to consider to increase the tilt on the motors to increase yaw authority.

Thanks for your suggestion!

The pilot went up with almost full throttle. We use the 4x geometry, even though the motors don’t form a square, but rather a long-stretched rect.

We had a successful flight today: https://logs.px4.io/plot_app?log=48e2e908-2540-4ce7-8d3a-dc83feb457bb

after changing VT_TILT_TRANS from 0.0 to 0.5. Apart from that we already had a measurement of 5m/s airspeed (instead of -5m/s as last time) before initiating the transition.

VT_TILT_TRANS only has an effect for tiltrotor, aren’t you using a standard VTOL?

For a long-streched rect you should adapt the mixer file PX4-Autopilot/quad_x.toml at 5868463d069eb652ff341427d3714133dde3c19d · PX4/PX4-Autopilot · GitHub

And not really related, but the fixed-wing doesn’t look that good - any idea where this huge roll angle error comes from?

Aldo very strong pitch oscillations, resulting in 6G (!) load on the vehicle .