I am calibrating the AK8963 on the MPU9250. The first plot shows original mag readings and the last plot shows readings corrected for soft and hard bias. I am following the guidance of Kris Winer: https://github.com/kriswiner/MPU6050/wiki/Simple-and-Effective-Magnetometer-Calibration.
Do these plots look good enough or are they too flat? Is there a conventional metric to test for this. I was thinking of just plotting a perfect circle and calculating how well it fits.
each axis is in Microtesla