gvm87 wrote:The screws are magnetical and thats the problem with gyroscope.
The problem is, the ferromagetic screws affect magnetic field sensor used by compass, not the gyroscope that may be used by compass application too but for quick movement detection rather than for determinig direction. Wrongly biased gyroscope may then cause compass rotating and giving "low compass accuracy". Unfortunately, the magnetic field is periodically disturbed:
![Screenshot_20170329-104230[1].png](./download/file.php?id=1745)
- Magnetic field sensor readout
- Screenshot_20170329-104230[1].png (172.8 KiB) Viewed 4526 times
I rotated the phone by 90 degrees in about 1/3 where the Y (green) readout drops and Z (blue) rises. There are high drops on Z axis together with smaller on the X axis and negligible rises on Y axis. It might be that the steel screw amplifies the magnetic field. I suppose it comes from power lines. When the processor requires more power, the current rises, the electromagnetic field gets stronger and the disturbance peaks as high as the earth magnetic field (which is really weak in fact). It's possible that removing the steel screw stops amplifying these disturbances. If it was just magnetized, the readout would be affected statically and could easily be overcame by calibration by 8-shape waving of the device.
Then, if removing the screw really helps (still not having proper torx tools to try), the steel screw should be replaced with one made of different material that is not ferromagnetic. Like titan or duraluminium for example.