Understanding Power Spectrum Output

[kinect-guy]

Hello all,

I have data from a text file which contains accelerometer data in three (x, y, z) axis. It was sampled at 100 Hz and I have data when the accelerometer was vibrating at 40, 60, 100, and 200 Hz.

I am trying to analyze it in an "FFT" style, where the data is graphed with the frequency in the x-axis and power in the y-axis. Currently this is being attempted by building a waveform with the data then computing the power spectrum with the spectral measurements vi. The output is what I would expect when I input the 40 Hz data, but becomes less accurate at 60 Hz, then plain wrong at 100 and 200 Hz.

I am hoping someone can clarify what is going on here, or recommend a better method for accomplishing this task. In the top right corner is an FFT vi which did not seem to work with my data. I have also tried the Amplitude and Phase Spectrum vi, which returned similar results as spectral measurements.

Thank you.

[dan_u]

I didn't look at your code (only have LV2012), but when you sample data @100 Hz you can only get a Power Spectrum up to 50 Hz (half the sampling frequency).

[royceP]

Hi kinect-guy,

Please take a look at this support article on NI's website about FFT Fundamentals. Look under the bold header "Maximum Resolvable Frequency" where you will find some helpful information. The information in this article agrees with what dan_u has said.

 http://zone.ni.com/reference/en-XX/help/372416B-01/svtconcepts/fft_funda/