Yaw Unreliable

Frame: Tarot 650 hexacopter frame with two of the arms removed. (so, quadx configuration)
Motors: 570Kv
Props: Turnigy Cherry wood, 13x4.5

My issue is that there is a slow yaw drift clockwise, which sometimes increases especially after more throttle or other motions (sudden pitch or roll, as caused by autotune).

I’ve also noticed that even with quite a lot of yaw input, I can’t get much rotation counterclockwise. Noticed this when trying to counteract the issue and avoid crashes.

Attached are two logs. 86 is with a vibration damping mount, 87 is without. Is too much vibration the only possible reason for this issue, or could it be something other than that?

filedropper.com/86
filedropper.com/87

Created a new and improved vibration damping mount (IMU mismatch 0.13 on log autotest) but still experiencing unreliable yaw control and yaw drift. Seems not vibration related, in that case.

My internal pixhawk compass has very high compass offsets. I added an external compass to attempt to circumvent this issue. Is it possible the internal compass is still causing an issue, even though there is an external compass now available?

Updated to 3.3.1 and the issue is gone. This probably didn’t solve the issue directly, but unless it recurs I can’t be definitive about the cause