surface graph of a fourier transform

[Cameroon]

Hi there -

I want to make a surface graph of my Fourier Transform. My z-axis should be the amplitude coefficients, My y-axis should be Time and my x-axis should be Frequency. However, when I use the FFT vi I only get a z matrix of one column. Yet the surface graph VI requires a 2D array for z. How can I account for this and make a surface graph with the information that I have?

Thanks

Cameron

[DavidJCrawford]

Cameron

Take a look at the shipped example 3D Surface Example - Fluctuating Sine Wave.vi.

Say you used this as your baseline and replaced the Create Waveform.vi with the FFT sample.

Does that help you at all?

David

[rpursley8]

Your original signal is in time.  Your FFT is in frequency.  To get a 3D plot of both you have to use some other type of transform.  Either you take an FFT of segments of your time domain waveform (STFT- short time Fourier transform) or maybe a Wavelet transform.  There are a myriad of transforms for this type of analysis.  I would suggest looking into a Wavelet transform.

[Cameroon]

Hi there-

I was thinking about the wavelet transform as well. However, doesn't it only give me a 1D array of coefficients, just like the fft does? And if that is the case, then my surface graph for the z coefficients will have a problem again because I need a 2d array. Or am I confusing something again?

Thanks for your help though

Cam

[rpursley8]

No, a wavelet transform essentially gives coefficients for frequency bands.  The higher the band the more coefficients,  The format of the output is usually a 1D array, but this array has all of the coefficients for all of the frequency bands (the highest band takes half the coefficients, the next a fourth, the next an eighth, etc.).  I guess to show it in a 2D array you will have to fill in each frequency band (except the highest) with interpolated or duplicated points to make each one the same length.

Your result should be several time-domain signals, each representing a certain frequency band (logarithmically spaced).

Message Edited by rpursley8 on 07-10-2006 11:30 AM