by JussiH » Thu Nov 22, 2012 5:30 pm
Boris, I am running out of suggestions for things you can try. But here is a few more...
I am still pretty sure its something in your environment...static buildup in the insulation and/or expansion/contractions of the board as temperature changes is the best suggestions so far. Possible solutions is to try smaller temp range and other types of insulation/container - I know Bill uses only a tight fitting container made from styrofoam with the board mounted on plastic standoffs inside. I think that is possibly a better solution since the styrofoam wont absorb much moisture from the air. Sound isolation foam and mattress-foam may in some cases absorb a lot moisture from the air.
The board itself can also absorb moisture from the air in extremely humid environments. You can dry it out by placing it somewhere warm (up to 55C) and dry for a few days. The one I drowned and soaked in a pond for 24 hours was dried out on top of my wood-fired stove in a ceramic ovendish at 50-55C for 3 days, and it worked after replacing the pressure sensor.
You are letting it heat for 90 minutes, try 60 minutes or use a larger container/less insulation to slow the heating.
I am gonna research the board you sent back, but initial testing and drylogging of that board did not reveal anything wrong with it. I will see if I can generate temp compensation files for that board - if I succeed, we can exchange boards again. But it will take some time, since I am hanged up until middle of next week.
However, the file shown in the above file might still work for the temperature compensation, since the noise is pretty steady. The ones with just magnetometer noise could maybe also work for the same reason. You can see if a cal run with those is able to flatten out the sensor curves.
You can make temp calibration work with just 1 level file, so if you can manage to get a decent one, you should be able to generate a good result still - at least then you can progress with it...