Magnetometer calibration

Ellipsoid fitting using python numpy to calibrate magnetometers:

This tool performs magnetometer calibration using ellipsoid fitting to correct for hard iron and soft iron distortions. It can read data from CSV or TXT files, perform calibration, apply existing calibrations, and visualize results.

Web app

The same calibration runs in your browser — no install, nothing uploaded, and the same algorithm as calibrate.py:

• https://nliaudat.github.io/magnetometer_calibration/ (drop in a capture, read the quality report, copy the C code / JSON / ESPHome lambda, download the patched YAML);
• magcal.html — one self-contained file in this repository: double-click it, no server and no Python needed.

The port is validated against calibrate.py itself: tools/dump_golden.py records what the Python tool computes and web/tests checks the JavaScript against it — see docs/web-app.md. Locally, python tools/serve.py --open serves both the sources and the built site (double-clicking web/index.html cannot work: browsers block ES modules for file:// pages).

Usage :

1. output you raw data from sensor and name it mag_out.txt in format x,y,z (uT) or in a csv file

2. optionally set the MField Value according to Gravitation_Field.md

and use --field flag

3. run script Defaults to python calibrate.py with mag_out.txt

Example Usage: python calibrate.py -f magnetometer_data.csv --plot python calibrate.py -f mag_out.txt --json --save python calibrate.py -f data.csv --apply already_calibrated_matrix.json --plot

4. get "Hard Iron Bias" and "Soft Iron bias" matrixes (and c++ according code)

   // Taken from calibrate.py, run on the bundled mag_out_sample.txt with --field 515
   // (the 2021 script printed different soft iron values: it had the v1 index bug that
   // was fixed in commit 61f5ab1, see issue #1)
    float hard_iron_bias_x = 41.168866;
    float hard_iron_bias_y = -89.874657;
    float hard_iron_bias_z = 569.663929;
   
   
    double soft_iron_bias_xx = 2.711938;
    double soft_iron_bias_xy = 0.027825;
    double soft_iron_bias_xz = -0.113831;
   
   
    double soft_iron_bias_yx = 0.027825;
    double soft_iron_bias_yy = 2.750884;
    double soft_iron_bias_yz = 0.029724;
   
   
    double soft_iron_bias_zx = -0.113831;
    double soft_iron_bias_zy = 0.029724;
    double soft_iron_bias_zz = 3.357968;
    
    // get values x,y,z and subtract the hard iron offset
    float xm_off = id(hmc5883l_x).state - hard_iron_bias_x;
    float ym_off = id(hmc5883l_y).state - hard_iron_bias_y;
    float zm_off = id(hmc5883l_z).state - hard_iron_bias_z;
    
    // multiply by the inverse soft iron offset 
    float xm_cal = xm_off *  soft_iron_bias_xx + ym_off *  soft_iron_bias_yx  + zm_off *  soft_iron_bias_zx;
    float ym_cal = xm_off *  soft_iron_bias_xy + ym_off *  soft_iron_bias_yy + zm_off *  soft_iron_bias_zy;
    //not needed : float zm_cal = xm_off *  soft_iron_bias_xz + ym_off *  soft_iron_bias_yz  + zm_off *  soft_iron_bias_zz;
    
    //float heading = atan2(ym_cal, xm_cal);
    float heading = atan2(0 - xm_cal, ym_cal);
    
    //heading += id(magnetic_declination); 
    
    if(id(enable_magnetometer_serial_output) == true){
    ESP_LOGD("main", "%.1f,%.1f,%.1f", id(hmc5883l_x).state, id(hmc5883l_y).state, id(hmc5883l_z).state);
    }
    // Correct for when signs are reversed.
    if (heading < 0) {
    heading += 2*PI;
    }
    // Check for wrap due to addition of declination.
    if (heading > 2*PI) {
    //heading -= 2*PI;
    }
    float headingDegrees = heading * 180/M_PI; // Convert radians to degrees.
    return headingDegrees;
   

---

If you run esphome, a sample code is included to output raw data and get corrected values.

Working demo

If you want a sample or a working demo, look at https://github.com/nliaudat/weatherstation

Requirements :

calibrate.py requires numpy,scipy, pandas and optional matplotlib to plot the graph

Thanks :

John zhang12300 for issuing the bug

jremington for fixing the bug

domsl for improvements in PR#5

Sources :

    -  https://teslabs.com/articles/magnetometer-calibration/      
    -  https://www.best-microcontroller-projects.com/hmc5883l.html