But the fix suggested, tuning INAV algorithm to weight the accelerometers more, is no longer possible in copter 3.3.
I’ve added more foam by the baro, but it seems to have helped very little.
Are there parameters of the EKF that I can safely tune to try to address the issue? Or is vibration potentially a factor (bad vibrations making the EKF weight IMU less, more dependence on baro)?
I had a similar problem; a steady, shallow descent in horizontal flight, where I know that I am not exceeding the capability of the aircraft. Increasing EKF_ALT_NOISE from 1 to 3 seems to have helped this a lot. There is still room for improvement, but any time I can improve a physics problem with software is a win by me.