Accelerometer Resolution/Accuracy

IMK

Hello,
·
I have a project that I am working that uses a tri-axial accelerometer to detect/measure low g forces in any direction.
·
Now I have been testing with the Analog adxl330 and the FreeScale 7260 and get similar noise results that I guess is the nature of them.
·
So as my application only requires a sample every 100ms (0.1 sec.) I was wondering how one best filter the noise for best resolution and accuracy. So is it best to use the R/C network and any recommended values please, or to sample the device as fast as possible on the A/D then compute the numeric mean.
·
Many thanks in advance IMK·
·

Tracy Allen

An RC noise filter with a time constant of less than 0.1 second would help with high frequency noise. That would be 10kΩ with a capacitor in the range of 0.1 to 1 µF. If that time constant is larger, it will affect the response to real accelerations above the 10 Hz cutoff. Digital filtering in the Stamp is a good option too, but there will not be time for more than a few readings in each 0.1 second interval, especially since you need to read 3 channels for the 3 axes. There are several possibilities for a digital filter, not only numeric mean. There could be an IIR lowpass filter, or a window or FIR filter, or a median filter. But they all will come up against the Stamp's limited sampling rate compared to the response to rapid accelerations.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Tracy Allen
www.emesystems.com