wwbrown:
How have you verified 0.5° accuracy? Have you calibrated the unit?
I never said 1/2 degree accuracy, I said 1/2 degree resolution. I know this because I can move the accelerometer around and I get 1/2 degree movements.
Also I have a granite surface plate with angle blocks. When zeroed on the granite all the angle blocks measure right.
The calibration is in the software. Look at the X,Y, & Z cal values. Those are set to 0 g. When you mount the accelerometer, first write a sketch that outputs only the value of analogRead in all 3 axis. If the object that the accelerometer is mounted to is level then the 3 numbers you get would be the calibration data.