We want to calculate speed integrating acceleration on an elevator system. We
are studying to use ADXL345 and we are doing some trials. We measure one single
axis accelerometer, now board is placed horizontally so we are using Z axis.
As noise and offset are accumulated during integration, is there any “better
axis” to calculate this integration? (let say, using X axis measurement instead
of Z axis because may be this is less offset, lest drift or may be less noise
…) Lookin on page 4 of the datasheet it seems Z axis is “worse” specs..
You are correct. X/Y axis performance is better than Z axis.
You can make the board vertically to use X/Y axis.
Regarding integration, you have to calibrate the offset of the axis that you
are interested in or else the offset will dominate the total error.
Attached is a brief introduction about how to calibrate the offset of ADXL345.
Of course, if you use a high pass filter, the offset issue can be ignored.
Whether the HPF can be used or not depends on your interested frequency range.