We're experimenting with using a pair of ADIS16228 to localize rattles on a surface, hopefully expanding out to eight. However, the signals guy on the team is nervous because the two we're looking at so far have pretty different offsets on the accelerometers. Resting the two on the bench, one reports acceleration of +0.92g, and the other +1.46g.On page 4 of the datasheet, the property "Offset Error" has a TYP of +/- 1g, and no stated MIN or MAX. I understand this to mean that the accelerometer is known to have an offset, and an offset up-or-down one gravity is nothing to be concerned about; in rare cases, there may exist even larger or smaller offsets (although not in my case). Am I correctly understanding this?
Registers 0x04, 0x06, and 0x08 all have scale adjustment, but that is related to the distance between adjacent steps of acceleration, and not with the overall offset. Is it the case that corrections for accelerometer offset cannot be done on-device, and should be done in post-processing?
I appreciate this is sort of a strange question; the ADIS16228 I think is supposed to be used more for its FFT capabilities than gathering raw acceleration data (given that AD has other devices purpose-built for gathering raw acceleration data). However, the signals guy is jumpy now about "can we rely on any of the data, if there's these weird offsets?"
To restate the questions:
Is presence of offset error in the accelerometers basically certain? (I think the answer is "yes")
Is there any value of offset error that indicates a problem with the device? (I think the answer is "no")
Is there a way to correct offset error on-device? (I think the answer is "no")