I was once called upon to develop an accelerometer for a racing
application and used the ADXL202 with good success.  One problem I hit
was that the vehicle tilted slightly under acceleration.  However, it
was easily handled.  What I did was to wait for zero vehicle
acceleration and no significant vibration (meaning the vehicle was most
likely stopped) for an extended time (i.e. 10 sec), at which point I
"calibrated" the device for gravity (one sensor was in the vertical
axis).   I made the assumption that gravity would not change much over
the racetrack.  Once calibrated, it was simple trig to calculate what
the REAL acceleration was by sutracting out the tilt of the
accelerometer.  Since the vertical axis should have been a known value,
the difference was the tilt, at which point I recalculated the real
acceleration by subtracting the horizontal component from the
acceleration that was read from the device.  Simple.

I used another accelerometer for side-to-side acceleration measurement
and did the same thing in order to subtract vehicle lean from lateral
acceleration measurements.  I did it this way so that I didn't have to
use gravity in one device to calibrate another single-axis lateral
accelerometer.  It may have been OK, but I didn't trust it!  The other
advantage was I could call the same routine for both accelerometers.
 Made my job that much easier!

Does this make sense, or have I just clouded the issue??

P.S.
My contract called for all code rights to be turned over to the client,
so I can't give you example code.  SORRY!

Good luck!
Dennis

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads