apejovic, sorry for not responding sooner, I was hoping someone with more experience would chime in. Have you tried any more runs in the meantime? This might be better placed in the Calibrations forum, too.
It's interesting that the disturbances seem to be in similar times/temperatures, although not exactly. I would suspect something in the environment (A/C coming on, or something like that), and perhaps the timing is just a coincidence... but it does seem a bit strange.
One other thing you could graph would be voltages -- Acc sensor voltages are 8, 9, and 10, and for rate sensors it's 0, 1, and 2. This might tell us something about that noise at 400K.
I would not use the run 1 log because of that spike @ ~100K. The other 2 might be OK since the docs indicate that much noise would be acceptable, though if you could get cleaner logs that would be better. Again, I'm speaking from limited experience here, but I'm not sure you'll get a more definitive answer (I'm not sure anyone except Bill really understands the whole process
).
One other thing you could try is run your static logs files through the first 2 steps of the calculation process (cal --rate and cal --acc) and see what kind of MAE numbers you come up with (those 2 steps only require static logs). There's been some talk of trying to come up with a way to quantify the quality of the collected data -- there isn't really one yet but I think the MAE numbers from those 2 steps might be a good start.
-Max