Post Go back to editing

ADXL325 vibration bias

I am currently using the ADXL325 sensor to measure the acceleration normal to the ground of a moving vehicle (using Y-axis, facing downward) and have noticed a bias which occurs when the vehicle increases speed.

Attached is a plot of G vs seconds illustrating what I have described.  '02_BRF2.png' shows various sections when the vehicle is in motion and when it is at rest.  The signal has been corrected so the measurement reads 0G when the vehicle is at rest, but when it is in motion I would expect the signal to oscillate around 0G.  However, the smoothed signal, green, shows an obvious bias when the vehicle is in motion.

As per my discussion with Analog support I have tried various degrees of dampening, and will also try to reduce the bandwidth (currently at 50Hz) but wanted to know if anyone on here had some extra insight on the issue.