I've got a conversation on-going with support but it's taking a long time to get info.
Are product has just shifted to MPU-9250 and we're finding between 10 and 20% of MPU-9250's have very poor noise levels on the magnetometer.
We even bought some of the official EMSENSR-MPU9250 breakout boards to prove a point - it's not our design.
Noise should be 0.85uT ish but we're finding up to 3.5 uT noise in some MPU-9250 samples. The noise also follows the IC, if we swap ic on our boards the noise profile follows.
Anyone else come across this issue ? Easiest way to test is log raw mag axis output and cal the standard deviation (e.g. in excel(). It should 2 or less.
Rob