Is there any data available about the matching of sensitivity between the two
axis? It is necessary for me to be able to calibrate an instrument utilizing
this sensor. If the matching is negligible then I have to find only the 0g
points of the axes, but if it is not, then I also have to look for the +-1g
points. The reason, why only matching is important for me, is that I do
calculations with the quotient of the measured values.
My other question relating calibration is the method for finding the accurate
zero of the sensor.
I have two ideas:
a. Find the +-1g points of that axis and do averaging b. Find the +-1g points
of the other axis and set that point to be the zero of the first axis Which
method is better? Are there any other methods better that theese?
What is the minimum deviation when I apply logic "1" to the self test pin, that
shows normal sensor functionality? It is necessary for me, because I have
problems with testing the sensor in the +1g position. Is it a reliable solution
to accept 150mg deviation when the sensor output reaches 1.2g?
A1. There is no specification for the matching of sensitivity of both axes. I
think it is safest therefore to assume that each axis can have sensitivity
variation of +/-60mV/g independent of the other axis. Even if we suppose that
the axes do tend to have similar sensitivity, this knowledge doesn't help much
if we don't know
2. Auto calibration is rather simple method, but a fixture is required. If the
PCB (with the accelerometer mounted on it) is mounted on a slowly rotating
fixture oriented such that the plane of the PCB is orthogonal to the surface of
the Earth the accelerometer will be exposed to ±1g in both axes as the PCB is
rotated 360°. The microcontroller should keep track of the minimum and maximum
output from both channels. Once this is known:
Zero g = (Minimum + Maximum) / 2
Sensitivity = (Maximum - Minimum) / 2
3. You should not rely on the ST pin to provide a precise aceleration stimulus.
It will give between 400mV and 1100mV of voltage change at 5V supply and this
number varies proportional to the cube of the supply voltage.