Clipping of IMU3 Z Accelerometer


A few days ago I pulled a log to check the vibrations on a newly build drone using a Cube Black. In general VIBE.VibeX, VIBE.VibeY and VIBE.VibeZ seems all good (well below 20). Clipping for X and Y also stays at 0 but for Z it seems to steadily increase, which as I understand is not good.

Looking at IMU3.AccZ I could see that it might be the culprit as it has much higher vibrations than IMU1.AccZ. Now according to the Cube specs and HW ID in Mission Planner both IMU1 and IMU3 are of type MPU9250. The Cube’s System Features also mentions that only 2 of the 3 IMU’s are vibration-isolated. So I guess this is why when I overlay IMU1.AccZ and IMU3.AccZ there is such a huge difference, IMU3 must then be the non-isolated one.

Below I have Z clipping in blue and the two AccZ plots in green (IMU3) and red (IMU1):

So I not sure what to make of this, is it a problem or not? If the steady increase of Z clipping can be contributed to IMU3 (which is not isolated) do I really have a vibration issue or is this kind of normal behaviour?

As a side question, does anybody know why would the Cube not add vibration isolation to all 3 IMU’s? Specially if they claim that no vibration isolation should be used, this means we would always have issues with IMU3…

I believe IMU3 is only meant as a backup and is physically on a different PCB to segregate it in case…

Have you tried finding out where the vibrations come from? Normally you have vibrations on X and Y if the propellers are unbalanced. Maybe try a FFT analysis?

That would make sense, would be a noisy backup but I guess better than nothing.

In general the vibrations seems fine so I did not really try to find a source, was thinking it could be a false warning as only the non-isolated IMU3 is complaining. Maybe anyway not a bad idea to do a FFT to see what is upsetting IMU3 so much.

Below are the vibes:

You do need to do hardware changes, until there is no clipping at all. After that use the notch filter and the dynamic notch filter to reduce the vibrations and only after that (re)tune the PID values.

So this is then not something I should just ignore and blame on IMU3 not being vibration isolated… Will then go through the steps to try and narrow the source down.