Skip to content

Latest commit

 

History

History
219 lines (188 loc) · 10.8 KB

README.md

File metadata and controls

219 lines (188 loc) · 10.8 KB
Raw

MagAlpha-Calibration-Curve-Toolbox

Collection of tools and code examples designed to help users improving the accuracy of the MPS MagAlpha magnetic angle sensor by implementing a calibration curve.

Features

This project contains the following features:

  • Calibration curve generator
  • Angle Interpolation
  • Ma-Cal-Generator App

Calibration curve generator

C source code that generate a lookup table with the angle error.

The source code can be found in src/calibration-curve-generator

Angle Interpolation

C source code that compute the corrected angle using the previously generated lookup table.

The source code can be found in src/angle-interpolation

Ma-Cal-Generator App

C++ application that embed the two aforementioned source codes inside a executable test program.
The application take a CSV file containing the reference and measured angle as input parameter, generate a lookup table of the angle error and then use this lookup table to correct the measured angle. The application return a CSV file with the both the input parameters and the computed corrected angle. This output file will be placed on the output-files folder.

The source code can be found in src/ma-cal-generator

Installation

Ma-Cal-Generator App

The Ma-Cal-Generator application use Qt Open Source Framework. You can either directly use the compiled Windows executable or you can build it yourself.

Use the Pre-Built executable

The executable can be found in the bin.zip on the Releases page. The executable is provided along the required Qt DLLs.

Unzip the files and copy the bin folder into the main MagAlpha-Calibration-Curve-Toolbox folder. You should have something like this:

MagAlpha-Calibration-Curve-Toolbox
├───bin
├───docs
├───input-files
└───src
    ├───angle-interpolation
    ├───calibration-curve-generator
    └───ma-cal-generator

Build the project

Start by downloading and installing the Qt Open Source Framework.

Then open Qt Creator, go to File => Open File or Project... and select src/ma-cal-generator/ma-cal-generator.pro. The last step is to built the project. The executable will be located on the bin folder.

This project was built and tested with Desktop Qt 5.7.0 MinGW 32bit for Windows but it should works as well with the latest Qt version.

Calibration curve generator & Angle Interpolation

These source files doesn't require any installation, simply copy the .c and .h files in you project. They only require the math.h library to works and should therefore be portable to almost any microcontrollers, embedded systems and desktop environments that use C language.

How to use it

Ma-Cal-Generator App

Navigate to the MagAlpha-Calibration-Curve-Toolbox\bin folder and launch one of the following commands:

You can use the default input CSV file located in ..\input-files\calibration_data_input_example.csv.

ma-cal-generator.exe

Or you can specify the CSV input file you want to use.

ma-cal-generator.exe ..\input-files\calibration_data_input_example_add_75.csv

In both cases the output file will be located in MagAlpha-Calibration-Curve-Toolbox\output-files\calibration_curve.csv.

Input file format

The input file must use the following structure. You can use a much row as you want.

Reference Angle [degree] Measured Angle [degree]
258.047 0.703125
260.156 2.10938
262.266 3.51563
263.672 5.625
265.781 7.73438
... ...

Which gives the CSV format below.

Reference Angle [degree],Measured Angle [degree]
258.047,0.703125
260.156,2.10938
262.266,3.51563
263.672,5.625
265.781,7.73438
..., ...

Output file format

The generated output file will use the following structure.

Reference Angle Measured Angle Measured Angle with Zero Correction Angle Error Angle Error Fitting H1 H2 H3 H4 Ph1 Ph2 Ph3 Ph4 Number of points Corrected Angle Cst + Slope Angle Error after fit Cst + Slope Corrected Angle Cst + Slope Lin Search Angle Error after fit Cst + Slope Lin Search Corrected Angle Fitted Angle Error after Fit
258.047 0.703125 258.75 0.0596442 -0.461852 0.779883 3.42582 0.26283 1.64956 -1.56693 -1.56693 -1.56693 -1.56693 200 259.219 -0.469254 259.219 -0.469254 259.219 -0.469254
260.156 2.10938 260.156 -0.643115 -0.477579 0.779883 3.42582 0.26283 1.64956 -1.56693 -1.56693 -1.56693 -1.56693 200 260.655 -0.498378 260.655 -0.498378 260.655 -0.498378
262.266 3.51563 261.563 -1.34685 -0.496033 0.779883 3.42582 0.26283 1.64956 -1.56693 -1.56693 -1.56693 -1.56693 200 262.09 -0.527503 262.09 -0.527503 262.09 -0.527503
263.672 5.625 263.672 -0.643481 -0.530978 0.779883 3.42582 0.26283 1.64956 -1.56693 -1.56693 -1.56693 -1.56693 200 264.243 -0.571189 264.243 -0.571189 264.243 -0.571189
265.781 7.73438 265.781 -0.643115 -0.577405 0.779883 3.42582 0.26283 1.64956 -1.56693 -1.56693 -1.56693 -1.56693 200 266.396 -0.614876 266.396 -0.614876 266.396 -0.614876
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...

Which gives the CSV format below.

Reference Angle,Measured Angle,Measured Angle with Zero Correction,Angle Error,Angle Error Fitting,H1,H2,H3,H4,Ph1,Ph2,Ph3,Ph4,Number of points,Corrected Angle Cst + Slope,Angle Error after fit Cst + Slope,Corrected Angle Cst + Slope Lin Search,Angle Error after fit Cst + Slope Lin Search,Corrected Angle Fitted,Angle Error after Fit
258.047,0.703125,258.75,0.0596442,-0.461852,0.779883,3.42582,0.26283,1.64956,-1.56693,-1.56693,-1.56693,-1.56693,200,259.219,-0.469254,259.219,-0.469254,259.219,-0.469254
260.156,2.10938,260.156,-0.643115,-0.477579,0.779883,3.42582,0.26283,1.64956,-1.56693,-1.56693,-1.56693,-1.56693,200,260.655,-0.498378,260.655,-0.498378,260.655,-0.498378
262.266,3.51563,261.563,-1.34685,-0.496033,0.779883,3.42582,0.26283,1.64956,-1.56693,-1.56693,-1.56693,-1.56693,200,262.09,-0.527503,262.09,-0.527503,262.09,-0.527503
263.672,5.625,263.672,-0.643481,-0.530978,0.779883,3.42582,0.26283,1.64956,-1.56693,-1.56693,-1.56693,-1.56693,200,264.243,-0.571189,264.243,-0.571189,264.243,-0.571189
265.781,7.73438,265.781,-0.643115,-0.577405,0.779883,3.42582,0.26283,1.64956,-1.56693,-1.56693,-1.56693,-1.56693,200,266.396,-0.614876,266.396,-0.614876,266.396,-0.614876
...,...,...,...,...,...,...,...,...,...,...,...,...,...,...,...,...,...,...,...

Calibration curve generator

First you need to extract the harmonics components from the angle measurements.

#include "calibrationcurvegenerator.h"

//input parameters
const unsigned int sizeAngleArray = 200;         //change if needed
float referenceAngleInDegree[sizeAngleArray];    //fill with ref angle
float measuredAngleInDegree[sizeAngleArray];     //fill with sensor angle
//output parameters
float angleErrorArrayInDegree[sizeAngleArray];
float h1;
float h2;
float h3;
float h4;
float phi1;
float phi2;
float phi3;
float phi4;
extractAngleErrorHarmonics(  referenceAngleInDegree,
                             measuredAngleInDegree,
                             angleErrorArrayInDegree,
                             sizeAngleArray,
                             &h1,
                             &h2,
                             &h3,
                             &h4,
                             &phi1,
                             &phi2,
                             &phi3,
                             &phi4);

Define the lookup table size and angles to use.

//Generate the lookup table that will be use in the final application
const unsigned int lookupTableSize = 32;    //Define the lookup table size
float angleStep = 360.0/(float)lookupTableSize;
unsigned int i;
//output parameters
float lookupTableInputAngleArray[lookupTableSize];
for(i = 0;i<lookupTableSize;++i)
{
    lookupTableInputAngleArray[i]=(float)i*angleStep;
}

Compute the lookup table correction coefficients using the computed harmonics parameters. Their is several methods available to generate the calibration table.

Constants and slopes method

This method use two coefficients to represent the angle error:

  • angleErrorConstants represent the angle error offset for a given angle input
  • angleErrorSlopes represent the angle error curve slope for a given angle input

This method use more memory to store the lookup tables but is easier to use to interpolate the angle error of any given points.

//output parameters
float angleErrorConstants[sizeAngleArray];
float angleErrorSlopes[sizeAngleArray];
generateAngleErrorLookupTableUsingConstantsAndSlopes(lookupTableInputAngleArray,
    angleErrorConstants, angleErrorSlopes, lookupTableSize,
    &h1, &h2, &h3, &h4, &phi1, &phi2, &phi3, &phi4);

Fitted curve method

This method use a single coefficients to represent the angle error:

  • angleErrorInDegree represent the angle error value for a given angle input

This method use less memory to store the lookup tables but is slightly more complex to use to interpolate the angle error of any given points.

//output parameters
float angleErrorInDegree[lookupTableSize];
generateAngleErrorLookupTableUsingFittedCurve(lookupTableInputAngleArray,
    angleErrorInDegree, lookupTableSize,
    &h1, &h2, &h3, &h4, &phi1, &phi2, &phi3, &phi4);

Angle Interpolation

The function used to perform the interpolation depends of the method chosen to generate the lookup table.

Constants and slopes method

//input parameters
float measuredAngleInDegree;    //measured angle to correct
float zeroDegreeOffset = 0.0;   //Reference angle when the sensor return 0 degree (use 0.0 by default)
//output parameters
float interpolatedAngleInDegree;
float interpolatedAngleErrorInDegree;
interpolatedAngleInDegree=interpolateAngleFromConstantsAndSlopes(measuredAngleInDegree,
    angleErrorConstants, angleErrorSlopes, lookupTableSize,
    zeroDegreeOffset, &interpolatedAngleErrorInDegree);

Fitted curve method

//input parameters
float measuredAngleInDegree;    //measured angle to correct
float zeroDegreeOffset = 0.0;   //Reference angle when the sensor return 0 degree (use 0.0 by default)
//output parameters
float interpolatedAngleInDegree;
float interpolatedAngleErrorInDegree;
interpolatedAngleInDegree=interpolateAngleFromFittedCurve(measuredAngleInDegree,
    lookupTableInputAngleArray,angleErrorInDegree, lookupTableSize,
    zeroDegreeOffset, &interpolatedAngleErrorInDegree);