Integrating a frequency domain signal

[m-s]

Hi

I am trying to extract the response of a frequency dependent signal :

1. First, I use the SignalFFT vi to get into the frequency domain.
2. Then I plan to integrate the FFT signal around a certain frequency and take a ratio of that wrt to the integral over the entire frequency spectrum to get a measure of how strong the signal is around that frequency.

I am choosing a range of frequencies above because the physical phenomenon producing the time domain signal produces a certain lifetime width. I need a numerical integration VI that can do step 2 above. The numerical integration VI that I found in Labview (using ver. 7.1) integrates the whole signal, and there is no way to define limits between which the numerical integration should be carried out.

Any ideas ?

Thanks.

[Tania_Lozoya]

Hi m-s,

The numerical integration.vi gets an array of doubles as a input, why don't you try to use the Array subset.vi to only integrate the part of the signal you are interested in?

Tania Lozoya

[m-s]

Thanks for the hint. Now how do I find out which array indices correspond to those limits ?

[Tania_Lozoya]

That a cool question!

This has to do with the sample rate, when you adquire the signal, you get set of samples, that hava a sample rate, and when you do an FFT on those samples, every sample corresponds to frecuency bin (kinda of frecuency samples). Example:

If you get 100 samples at 1kHz, you will get a 100 windows, this is 100 samples to represent 1kHz, however, only the half of the spectrum is valid, because of the Nyquist teorem, so this mean that from 0 to 49, you have a correct representation, from 50 to 99 you have a reflection of the spectrum. So in the first 50 "frequency samples" you have 0-500Hz

Tania Lozoya

[rpursley8]

The length of you acquisition determines your frequency step size.  1 second of data gives you 1 Hz resolution.  100 ms of data gives you 10 Hz resolution, etc...