Verification and calibration of a commercial anisotropic magnetoresistive magnetometer by multivariate non-linear regression

Commercially available anisotropic magnetoresistive (AMR) magnetometers exhibit on the order of 1 nanotesla (nT) sensitivity in small size, weight, and power (SWaP) packages. However, AMR magnetometer accuracy is diminished by properties such as static offsets, gain uncertainty, off-axis coupling, and temperature effects. This work presents a measurement of the magnitude of these effects for a Honeywell HMC1053 magnetometer and evaluates a method for calibrating the observed effects by multivariate non-linear regression using a 24-parameter measurement equation. The presented calibration method has reduced the vector norm of the root mean square error from 4300 to 72 nT for the data acquired in this experiment. This calibration method has been developed for use on the AERO (Auroral Emissions Radio Observer) and VISTA (Vector Interferometry Space Technology using AERO) CubeSat missions, but the methods and results may be applicable to other resource-constrained magnetometers whose accuracies are limited by the offset, gain, off-axis, and thermal effects that are similar to the HMC1053 AMR magnetometer.