Accelerometer Self Test Function & Noise

Hi all

So I’ve been using these two ADXL362 accelerometers for a while now here’s the datasheet: http://www.analog.com/media/en/technical-documentation/data-sheets/ADXL362.pdf

I was just reading about this self test feature it has on page 35 & 41 but I don’t really understand exactly what it does. Regardless, I went about the process outlined on page 41 and fixed the accelerometer to a table using tape and tack then did the initial measurement. Then enabled self test and took another measurement. Subtracted the two values, but the readings are way off the max predicted in the table on page 4. The accelerometer is in the +/-2g range, which is the same as what all the specs are given too.

Here are the results from the two accelerometers:

Accelerometer axis: X1 Y1 Z1 X2 Y2 Z2
Normal values: 12 -2 -828 -25 -32 -808
Self test values: 1440 -1725 407 1395 -1775 414
Self test - normal: 1428 -1723 1235 1423 -1743 1222

In comparison to the output change section in the table on page 4, would this indicate my accelerometer is broken? Is this why I have ridiculous noise present, like 4 to 6LSB? Attached a picture of the X axis plot over roughly 7/8 hours. I’m thinking about applying a moving average to the data to see if that helps with the noise but I would ideally reduce it as much as possible initially.

z, X2.png