2 replies to this topic

[Jonathan Brandmeyer]

Are the magnetometer axes right?  I was measuring a field pointing up and to the south today instead of down and to the north (in N America) as expected.  Line up the figure from the datasheet on page 3 with the AHRS.  If the datsheet is displayed on a computer screen, then the AHRS should be held in front of you, with the bottom face facing you and the "front" arrow pointing to the right.  Then
chip +x points to AHRS -y
chip +y points to AHRS -x
chip +z points to AHRS -z

Now consider the code in ahrs.c, around line 346 (calibration code omitted):

mag_data.scaled.axis[0] = mag_data.raw.axis[1];
mag_data.scaled.axis[1] = mag_data.raw.axis[0];
mag_data.scaled.axis[2] = mag_data.raw.axis[2];

That can't be right: if two axes are swapped, and the chip has a right-handed coordinate system, then that would produce a left-handed coordinate system.  The way to flip it back to right-handed is to multiply all of them by -1, too.

Am I crazy?  If anyone else can confirm, then I'll go ahead and check in the fix.

[Jonathan Brandmeyer]

Or is this where the negative signs in the old calibration scale factors come from?  Which begs the question: Why is this being done in the calibration scale factors, instead of by the firmware itself?

[peabody124]

Correct, we rotate the axis of the mag to match the standard used by the INSGPS algorithm.  The scaling multiplication is done in firmware which includes the sign swap.