Accelerometer Experiment

Today I set out to find whether or not I could use an accelerometer as a tilt switch. The less pieces I have in my alchemy vessels the less likely hood of breakdown. Though tilt switches are hard to break, I would prefer to keep it as simple as possible. Due to the rather large amount of data and oddities associated with accelerometers I sent out to find whether they could guess orientation well. The answer is an astounding partially. The accelerometer that I am using ADXL 322 from Sparkfun could detect when it had almost entirely flipped upside down. Other orientation was not as consistent as I would have liked. However luckily for me I just need right side up and upside down orientation. The X data reads about 340 to 350 when sitting right side up. When flipped the data reads X as 640 to 650. While upside down there is about ten point drop in value of Y and Z. I restricted the data output on the processing end with a simple loop(I am using processing and arduino). The next step is to write a program the will detect the orientation rather than raw data. Once that is up I'll try to fool it.