I'm in a debate with a peer who says that filtering accelerometer signals at the chip-level has nothing to do with mitigating the problem of drift
Often, additional software-level filtering is employed to smooth noise, depending on the application. But My question pertains to any filtering or processing that is happening on the chip, specifically as it pertains to drift (and not so much to smoothness of the output).
My assertion is that noise due to thermal, electrical, vibrational, and sampling jitter will invariably contribute to drift, and that this type of noise is filtered on the chip to help reduce the problem. By reading an Invensense product PDF, we know there's "signal conditioning" happening on the chip. But the documentation makes no real mention of what's going on in there. It seems to be proprietary, and reasonably so.
I don't need any proprietary information, but if anyone could speak as to whether noise filtering reduces drift, that would be tremendously helpful. It seems obvious to me, but perhaps I am mistaken.