We've got a few FXOS8700CQs that we're running through a temperature test chamber from 60°C down to -35°C in 10° steps. This data will be used to correct the for changes in the zero-G offset on each axis. The data that we're getting for the Y and Z axes is perfect; however, we're seeing a lot of noise on the X-axis, with amplitude outside of our acceptable range once we get down to approximately -20°C. Does anyone have any idea what could be causing this and/or how we can fix it?
Here's a graph to illustrate what I mean:
Top plot is the temperature, lower plot shows the axes output - note that the data labeled 'Y Axis' (blue) is actually the X axis data - we have to swap them around for the calculations we are doing. Z axis data is magenta.
We have 4 sensors on separate PCBs in the test chamber, all displaying the same behavior.