Fieldform — Magnetometer calibration

Input data

STEP 01

↥Drop a CSV file hereor click to browse · CSV / TXT

— OR —

CSV format
Use one header row with three axis columns. Supported labels include x,y,z, mx,my,mz, or raw_x,raw_y,raw_z (case-insensitive). You can also use mag_x,mag_y,mag_z.

Put one reading per row, separated by commas. Each axis value must be a number; do not include units in the cells. Keep all readings in the same unit (for example, counts or µT). A header is optional if the first three columns are X, Y, Z in that order.

How calibration works

GUIDE

Collect a full rotation

Rotate the sensor through every orientation. Aim for a spherical spread of samples.

Remove hard-iron bias

Subtract the estimated center of the distorted point cloud.

Correct soft-iron scale

Normalize the three axes to a common field radius.

Samples

—

points loaded

Hard-iron X

— counts

estimated offset

Hard-iron Y

— counts

estimated offset

Hard-iron Z

— counts

estimated offset

Magnetic field · axis projections

RAWCALIBRATED

X / Y PLANE

X / Z PLANE

Y / Z PLANE

Waiting for data · upload a CSV or load the sampleORTHOGRAPHIC PROJECTIONS

Calibration equation

STEP 02 · COPY TO FIRMWARE

Subtract the hard-iron bias, then apply the soft-iron matrix to each raw sample. The values below update when data is loaded.

Universal correction

V_cal = M_soft × (V_raw − B_hard)

Estimated hard-iron bias · counts

B_hard = [0, 0, 0]ᵀ

Soft-iron correction matrix

M_soft = [1, 0, 0; 0, 1, 0; 0, 0, 1]

C implementation

// Load data to generate calibration code

This quick-fit estimate uses per-axis min/max centers and diagonal scaling. It corrects offsets and axis scale, but does not estimate cross-axis rotation/shear. For best accuracy, collect samples covering a full 3D rotation; verify the calibrated plot is roughly circular in all projections.