i have run obtained the six sets of raw data (based on the 6 directions of the LSM303D and then doing 36 equal divisions in a full rotation. (note: To keep this post from getting too long i will post the code for data averaging requested @kevin1961 separately after this one)
here is a scatter of the RAW (pre-calibrated) Mag data
OK. that looks like something i imagine that Magneto can really work with.
Then i look at the scatter for Accel and it’s a different story:
The values affected by gravity for each orientation are in the order of ~16000, so let’s zoom in and look at the spread of those middle ones. (here is the same data with max and min constrained to 4000):
this does NOT look to me like something that Magneto can deal with anyway, lets see…
Here are the raw data as two files: accel and mag each as a tab-delimited.txt file
170211MasterAccelRAW.txt (5.4 KB)
170211MasterMagRAWdata.txt (5.2 KB)
Now, running these data (one at a time, of course) thru magneto produces:
and for Accel:
how did it go?
putting the values back into pitch,Roll, Heading script and dissapointingly, the same problem persists!
As before i keep the heading constant so the red line (heading) should stay flat as i move first the roll(yellow) and then the pitch (blue). You can see it is a little better than it was for the roll but is USELESS on the pitch tilt compensation, as you can see!
I will add that the heading - if kept on the flat - is now VERY accurate. i.e on my compass rose a real heading 184deg returns the right value now. So that is some progress
BUT FOR TILT-COMPENSATION THE LSM303D = UNCALIBRATABLE …WHAT ON EARTH SHOULD I DO NOW???