problem at very high g-values

Dear all,

we did some drop test of our wireless MPU-6000 measurement box and saw some strange results.

The acceleration scale is set to -+16 g and data is read at 1024 Hz.

When dropped onto the y-axis we expect to see high positive g-values (over 16 g) on the y-axis for almost 10 ms (estimated by high speed camera images). However on average we don't even get a positive sum of the g-values during the deceleration phase but instead get mostly negative values that look like folding in of very high positive g-values.

For some softer drops we get some values at the measurement range limit of either -16.0 g or 15.999 g, but often these values are again interrupted by what seems to be folding of much higher g-values.

My guess would be that the output values are correct in the range from -16 g to 16 g, for moderately higher or lower values the sensor output is still at -16 g or at 16 g, but for much higher g values in the range from 30 to 60 g the sensor returns any value between -16 g and 16 g.

Does anybody have an idea what our problem could be and how we could make the sensor to output -16 g or +16 when the sensor expriences lower or higher g-forces?

Best
Markus