Rustam's techno-feed

In this post I’ll try to investigate what signal is produced by WiTilt accelerometer/gyro controller when it is immovable, whether it is noisy or not.
I placed WiTilt (3-axis accelerometer and gyro controller) on a table and collected data from it. I got 350 samples of raw x,y,z,r (accelerations along 3 axes and speed of rotation) measurements from motionless controller. WiTilt was put on a surface in a way that z axis was directed upward. So, collected data from z accelerometer shall give acceleration of 1g and measurements from x and y axes shall give 0g value. Analyzing the data in Excel, I received the following results (all measurements are in raw ADC values).

The distribution of acceleration values from all axes is very similar to normal distribution. So, lets suppose that it is true.

x distribution

X acceleration is distributed according to Gauss distribution with Average=547.74 (0g), Average Deviation=2.0.

y distribution

Y acceleration is distributed according to Gauss distribution with Average=490.42 (0g), Average Deviation=2.0.

z distribution

Z acceleration is distributed according to Gauss distribution with Average=762.30 (1g), Average Deviation=2.06.

r distribution

Gyroscope rotation data is distributed according to Gauss (is it?) distribution with Average=466,73 (0 deg/sec), Average Deviation=6.01.

Converting raw ADC (Analog-Digital Converter) values to g’s.
During accelerometer calibration I found out that Z-axis accelerometer has the following values:
Z min = 259 ADC (-1g),
Z max = 762 ADC (+1g),
so 1g = (762-259)/2 = 503 ADC counts and 1 ADC = 1/503g
It means that Average Deviation of Z = 2.06 ADC = 0,004095g.

Lets see how accelerometers noise leads to visual vibrations (object is immovable and signal from accelerometers is not filtered and not smoothed):