i'm having the worst trouble with mags on the HK Pilot pixhawk clone. All mags are throwing wacky values near 65535 and then snapping back into normal readings. You can see the spikes in the graph below. These spikes cause flipping crashes when both the internal and external are spiking...I imagine the EKF math is garbage-in garbage-out at that point. It's less frequent on the externals but happens with both neo-6 and neo-7 while they are returning GPS data just fine. Flashing to 3.3 beta, 3.1.5, or 3.2.1 doesn't change anything and I can't find any posts about similar behavior so I don't suspect a software issue. When looking at live readings RAW_IMU_MAG xy or z looks normal, so does SCALED_IMU2. But the flash logs for the exact same moment show IMU1 or 2_MAGs as having the spikes. Has anybody heard of readings like this? Are these i2c bus errors?