Garug, the IMU is very steadily fixed on specially printed details, I just fixed it on the another side to be sure the problem isn't in the IMU.
AndrewSupport, thx! Decreased pitch's power from 131 to 63 – vibrations gone. But number of poles is set as 12. If I set 22 (the real count of poles in the motor), vibration occurs again.
Is it acceptable to use wrong number of poles, in case it works?
What ever works, but you probably will have later on problems with estimate angles from motors etc if the pole count is wrong.
Vibrations are because something flex and vibrates. I suspect the IMU and payload installation (bot together). If you attach the IMU to the payload it is probably better. If you add an upper bar and attach the payload to it also, it is probably much better.
If you have 32 bit board, you can also do filtering.
crystalbit, I have the same problem. At first everything was fine, but at some point, broke the motor (like it is a visually and no faults, but a very negative effect on the sensor). I replaced the motor for the same. When angle the frame itself – is stable, when lean forward (away from you) resonance occurs at the pitch motor, similar to yours. I do not use filtering (currently not yet fully studied the subject). Adaptive settings for the pitch motor work well and resonance disappears, but it is not suitable for work, I want to set up the system perfectly without the use of adaptive methods (because before it worked).
I think that there can be two problems. First, the power and PID settings of the pitch-axis seem to be too hight. Second, the IMU has to be connected absoltely stable (no tape, no velcro fastener)