Calibration loss on changing orientation many times in BNO055

[joyaljose]

In BNO055 there is an issue with loss of magnetic calibration.

Aim : The thing we are trying to do in our application is, after a roll of 90 degrees, then if the device pitches upwards , yaw should not change along with pitch. So we reached at changing orientation to set this as initial orientation when roll is 90.

In our application, we are changing orientation when roll goes above 85 degrees and back to default orientation when it goes down below 85 degrees. This is needed so that we get stable yaw even on pitching after this roll. The orientation is changing as expected but we are losing the track of true north after some movements in this application. The calibration status of magnetometer also goes down.

We attempted to address this by writing the acceleration, gyroscope, and magnetic offsets that were set during calibration when the orientation returns to default. However, true north still deviates too much from its initial position after these movements.

We are looking for support, has anyone faced these issues?

[Ixck · Accepted solution]

YES! We encountered this issue about a year ago (as you may see from my post history), and up to that point, we'd never found anyone else that noticed the same issue.

The answer is simple, but unfortunate... The BNO055 firmware is terrible, and a workaround is either not possible or not worth it.

I'll try to recall the issue and trigger to the best of my memory.

The sensor would give quite stable and accurate readings until it was 'fully calibrated', and despite being able to load calibration offsets, they would be automatically re-written anyway in NDOF mode. The actual trigger was the sensor reaching "3/3 calibration" state. 0/3 to 2/3 calibration gave very good results, but 3/3... instantly broken. Once 3/3 calibration was reached, the heading would jitter and drift in either direction up to 10 degrees.

Our thought was maybe we could just keep loading offsets, we even tried saving the offsets to a separate flash chip, but the NDOF would constantly upload its own 'better' 3/3 calibration offsets as soon as they were available (which would happen when you're using the device).

Bosch Sensortec does not care about the issue, they do not care to assist you or me, nor the thousands of other people that have experienced this issue to some degree - as evidenced by people complaining about resetting calibration states.

I would strongly suggest going with another manufacturer. It's quite alarming that one of - if not THE - largest sensor manufacturers on planet Earth knowingly and deceptively continues to sell disfunctional sensors.

We foolishly switched to their newer architecture (BHI360) instead of going with a different brand and well... still struggling to get things working, constantly plagued with errors codes with no description.

[joyaljose]

Thank You for helping. We were having a hard time why this was happening.