So, I further investigated the attitude behaviour and how this might have an influence on the slow recovery of the accelerometers after a movement. I run a series of attitude test. Each test started by the ADIS being software reset and then configured : DEC_Rate = 0, Local Navigation Frame (ENU). Each time the ADIS was
- rolled -90° , then back +90°
- pitched -90°, then back +90°
- yaw'ed' -90°, then back +90°
For set 1 the system was placed flat on the table and a TARE command issued.
For set 2 the system was placed flat on the table and NO tare command issued.
For set 3 the system was placed on the table with -90° roll as its initial orientation and NO tare command issued.
I plotted the output in Matlab (and attached the .fig for zooming, etc). The rows correspond to the sets and the columns to 3 'tests'.
I can mostly understand set 2 and 3, but for set 1 only the yaw rotation shows what I would expect.
Roll and Pitch both seem to be multi-axial rotations which (at least in this magnitude) did not occur during the experiment.
Does the TARE command align all 3 axis with the current attitude? Or does it only act on the z axis?
The pitch test on set 2 and 3 shows multi axial rotations as I would expect since the system is not aligned with East North Up. However, in plot 2 roll is only positive. I might be understanding some Euler Angles wrong, but I would expect it to switch sign just like the yaw does?
Something is not quite right here and I'm wondering if this has anything to do with the slow recovery of the accelerometers after a movement?
