Compass with ICM-20948 – high variance in readings

Sensors IMU & Motion
1

Hi,
I’m trying to build a compass using the ICM-20948, and while I can roughly determine north, my readings have a large variance ussualy about 10°.

I’ve performed calibration using Magneto, so I don’t think that’s the issue. However, the results are still unstable. What level of accuracy should I normally expect? Is this level of variance common?

Thanks!

2

From sensor noise alone, you can expect a standard deviation (not “variance”) of perhaps 2 degrees at most.

I just tested mine, sitting still on the desk, and collected several hundred measurements. Using the formula yaw = 57.3*atan2(magy, magx) I get a standard deviation in yaw of 1.5 degrees. If the compass is tilt compensated, the accelerometer noise will contribute as well.

For informed help, please provide the usual details regarding circuit, calibration, code and project environment.

1 Like