The ITG-3200

I have finally gotten some excellent calibration data out, and have determined one important fact. The ITG-3200 is a fine, fine sensor. Once the sensor is calibrated, with no noise filtering at all, it has an almost un-measurable drift over 17 minutes. It is good enough that you can just straight up integrate the body rates.

The ITG-3200 is available from Sparkfun, of course. Highly recommended for all your IMU needs.

The next task is to determine the relative orientation of the sensors to each other and to the frame of the Bridge board. This actually seems like it is quite doable. First, set up the turntable very carefully, using a circular bubble level to verify the spin axis is parallel to local gravity. I used about 10 sheets of paper to shim it.

Also, we are going to use the sides of the box on the turntable to make sure that the IMU box is always in the same orientation relative to the turntable.

Next, we do a calibration dance. The accelerometer measurement will be along local gravity, which we have made sure to make parallel to the IMU box frame. Any misalignment between local gravity and the accelerometer axes is attributable to misorientation of the sensor. When spinning, the vector body rotation will also be parallel to local gravity and the IMU frame, so we can identify its misorientation as well. Finally, I don't think that compass orientation matters that much, and I don't think that it is observable. Maybe by looking at the plane the measurement describes as the box is spun...