I ran a sketch to collect raw data while i rotated the imu but the values are too high and according to the blog, they need to be scaled down to be fed to the magneto program. My question is where can I find those gain factors for the sparkfun mpu9250?
I saw that but then I read that the Sparkfun and Kris Winer MPU-9250 code both have major errors and should not be used which is why i turned to that blog post and I am using the code from this post but in the blog they were using another sensor.
Well in this case, the files suggested to first run the mpu_9250_cal sketch to get the gyro offset and collect raw value samples while rotating the sensor. After that it said I have to input those values to magneto so it calculates and gives me corrections. It outputs combined scale factors, misalignments and compensation for soft iron. My issue is i don't know if I need to do the axis specific gain correction and if so I can't seem to find the gain scale to use from the datasheet.
For magneto, the "norm H" value is arbitrary, as none of the AHRS programs care about the absolute value of the magnetic field strength. They use only the direction of the magnetic field vector, and for input to the Mahony or Madgwick filters, the vector should be normalized (if it is not normalized elsewhere).
For Magneto, choose a value for the norm so that the diagonal correction matrix elements end up being about 1. The average magnitude of the input field vector is a fine choice.
So can I just use this or the data collected needs to be scaled down? I'm confused because the blog I read said the following
“Magneto expects to receive raw data in +- format (a value of zero indicating a null field in the current axis), but not necessarily normalized to +-1.0.
If your sensors have SPI or I2C outputs, they will usually directly produce the required format. For example, the MicroMag3 magnetometer directly produces counts from -3411 to +3411, and the the SCA3000 accelerometer directly produces counts from -1333 to 1333, and Magneto can process directly these values, without the need to normalize them to +- 1.0. I understand that a normalization may be desirable to avoid machine precision problems, but this has not been the case with these sensors.
If your sensors produce voltage levels that you have to convert to counts with an ADC, you have indeed to subtract a zero field value from the ADC output before using Magneto. You would then normally choose the maximum positive value as input to the ‘Norm of Magnetic or Gravitational field’.
It's for a small hovercraft I am trying to build and I want it to guide itself autonomously.
In the sketch where I have to add the corrected values I got
// vvvvvvvvvvvvvvvvvv VERY VERY IMPORTANT vvvvvvvvvvvvvvvvvvvvvvvvvvvvv
//These are the previously determined offsets and scale factors for accelerometer and magnetometer, using MPU9250_cal and Magneto 1.2
//
//The scale constants should *approximately* normalize the vector magnitude for accel and mag to 1.0
//The AHRS will NOT work well or at all if these are not correct
float A_cal[6] = {515.0, 279.0, 751.0, 5.96e-5, 6.26e-5, 6.06e-5}; // 0..2 offset xyz, 3..5 scale xyz
float M_cal[6] = {18.0, 28.3, -39.6, 0.01403, 0.01414, 0.01387}; // can make both 3x3 to handle off-diagonal corrections
float G_off[3] = {-45.0, 334.0, 141.2}; // newly calculated raw offsets
I replace the gyro offsets given by the sketch that calculates the raw values however, the raw data given to magneto included 6 columns: first 3 (xyz) for accelerometer and last 3 (xyz) for magnetometer. Do I need to separate them into 2 files and calculate them separately with magneto? I assume so since the offset values I have right now are way too high for the magnetometer.
alright I will do them both and just to be sure I understood it correctly. The combined biases calculated in magneto will go here as x, y and z and the diagonal values of the computed matrix A as scale factors diagonal_1,diagonal_2,diagonal_3 ?
To do the calibration right requires use of the full matrix. Since the MPU-9250 was discontinued several years ago, and I no longer have one to test, I never bothered to implement that correction.
It is done correctly for its much improved successor, the ICM_20948, so you can take a look at that code here:
Where you say Mag scale divide by 369.4 to normalize and Accel scale: divide by 16604.0 to normalize. Those scale values are taken from the mpu_9250_cal sketch where it calculates and displays rms for A and M at the end?
Those numbers were notes to myself. The code should normalize the vectors automatically.
In looking back, I see that I did add the full matrix correction to this MPU-9250 code. If you look at the function get_MPU_scaled(); you will see where the vectors are normalized.
Thank you so much I was going crazy trying to understand how all of it ties up together. I understand it now. I will rerun the calculation sketch and separate the accelerometer and the magnetometer readings into 2 files and use them separately in magneto to get my matrices.
