Customer is concerned about aliasing of vibrations using this device. How is this handled internally? Specifically, what's the sample rate of the ADC. Is it variable based on the filter settings or is the filter just used to decimate. One scenario would be a 200Hz output data rate, but he's got vibrations at 150Hz. Would the aliasing show up on the output data? Can you provide more information on the digital filter, perhaps the # of taps, and the response?

Thanks,

Tim

- 1 person found this helpful
Hi Tim,

I believe we've already been in touch on this topic. The accelerometer should be thought of as any sampled system would be. An output data rate has a corresponding bandwidth of roughly one-half the output data rate (so, the 200 Hz ODR would have a bandwidth of 100 Hz). Anything over the bandwidth will be attenuated by the digital filter; however, anything over 100 Hz is also in violation of the Nyquist theorem, which means aliasing will occur (although the attenuation will result in reduced amplitude). If the customer knows for sure a certain oscillation will be present, it would be advised that a suitable output data rate is selected to satisfy the Nyquist criterion ( the sampling rate, Fs or ODR for the accelerometer, should be at least twice expected maximum frequency content).

Hi Tim,

I believe we've already been in touch on this topic. The accelerometer should be thought of as any sampled system would be. An output data rate has a corresponding bandwidth of roughly one-half the output data rate (so, the 200 Hz ODR would have a bandwidth of 100 Hz). Anything over the bandwidth will be attenuated by the digital filter; however, anything over 100 Hz is also in violation of the Nyquist theorem, which means aliasing will occur (although the attenuation will result in reduced amplitude). If the customer knows for sure a certain oscillation will be present, it would be advised that a suitable output data rate is selected to satisfy the Nyquist criterion ( the sampling rate, Fs or ODR for the accelerometer, should be at least twice expected maximum frequency content).