11-20-2009 10:48 PM
I don't think there is a way to increase it. I don't really get how to measure the amount of detail that the accelerometer has but you might be able to find information in the "AccelerometerSensor" documentation.
11-21-2009 12:28 AM
I don't think the resoltion can be improved, although I'd love to be wrong. 18mg is my observation too.
Having said that, 18mg is quite sensitive. Are you sure you're looking at the correct axis of tilt?
11-21-2009 01:33 PM
Well, it isn't hard to just print all 3 and right now I don't remember if "1000" is 9.8m/s^2 or 980 is 1 g but close enough in either case. It isn't hard to turn resolution into angle, .018g seems to be a significant amount of tilt, empirically it seems to be more than a bubble level would need to show out-of-plane. But, for some simple artificial horizon and gross acceleration measurements its fine ( am I going down hill or turning left etc). If you had a way to vibrate it in some reasonable way you could force flucutations and average them and I guess even if there is enough sensor noise you can average sample if the sample rate is high enough. If the signal is clean, the ADC kills your hopes...