I am currently testing a FRDM-STBC-AGM01 with FRDM-K22F using NXP sensor fusion toolbox for windows.
I have taken the boards outside, calibrated the sensors and tested against a 360 degree baseboard.
Here are my error results below:
I understand that it is very difficult to get very accurate heading readings even with gyro compensation/calibration etc but it seems these errors are too high compared to other DMCs I have tested.
I have noticed results like these 'sine waves' before and it transpired that the compass was not calibrating properly.
Has anyone experienced something like this before with the NXP chip? if so how did you get the calibration to stick?
Thanks in advance, Adam.
Thanks for posting this question on our community website.It's really interesting to have such kind of characterization challenging our sensor fusion.
I would like to better understand how do you test our solution. What is the reference ? Do you use a motor to rotate the board with a defined step? How to you calibrate the sensor fusion ? (board it-self in the air and then it's mounted on your test bench?).
Theses questions are really important, because once you modify the external environement around the system (temperature, magnetic field or ferreomagnetic material ), it can make the calibration not accurate. Note that our FXOS8700CQ magnetometer in very sensitive due to its TMR technology. So any external disurption will be sensed and may bring heading error on the final result.
Hi Anthony, thanks for your reply.
So my reference is basically a flat wooden board with a 360 degree compass rose printout stuck to the surface.
I connected the laptop to the NXP FRDM and moved the laptop as far away from the wooden reference board as the cable would allow (~2M). I erased the calibration points before each test. To calibrate I moved the FRDM around in 3D space above the wooden board until it had taken sufficient number of magnetic buffer points in the sensor fusion toolbox for a 'calibrated' status on the main page. Then to take my measurements I kept the laptop and wooden board in the same place and manually lined up the NPX FRDM in 45 degree steps to the 360 degree printout, walking over to the laptop after each alignment to record the compass output.
I have achieved +/-0.5 degree accuracy using this method for a different DMC, admittedly at a more open location but the magnitude of error still seems uncharacteristic for a calibrated DMC even in the ~30x100m test area I used.
Hope this helps.
Thanks for adding details. It seems you see a 1st order distorsion. If we consider the magnetometer alone (no fusion with gyroscope nor accelerometer), it could be because of a miscalibration of the hard iron (offset). Can you confirm if you are using the Gyro Stabilized Algorithm? or the 2D Automotive Compass?