Generating a Power Spectral Density with Un-Even Spacing

[struclearner]

When Time Domain Data is converted to Frequency Domain Data in the form of Power Spectral Density Function, the amplitude data is obtained with even spacing of the Frequency. If the data has equal or very close amplitude values with in the frequency resolution, the PSD curve is almost flat.
How the Time Domain Data having amplitude values nearly equal can be converted to a PSD with Un-Even spacing reducing the number of data points accordingly.
What programs are available to obtain a PSD with Un-Even spacing.
Thanks for help.

 

[GregLocock]

You can't using normal maths. The line spacing of each frame of data is entirely controlled by the length of the sample in seconds.

The usual approach to reducing the number of lines is to bundle them into 1/3 octave (or whatever, constant bandwidth is sometimes used) bands.

The other approach, perhaps, in the case of a flattish spectrum, would be to curve fit to the psd amplitude.


Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[IRstuff]

Moreover, unless you are a wizard, you don't know, a priori, that you'll wind up with a flat spectrum. Nevertheless, I don't see why you wouldn't simply look at the PSD, and then decide whether you want to decimate the data, and as the data is a set of ordered pairs, you can do whatever you want to do with the data. However, unless you are trying to do something on a Casio calculator with 2 kB of memory, why would you even bother to decimate the data.

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[SomptingGuy]

If you only want values at specific values, why not just do it long-hand for the frequencies that interest you? No need to do a full transform.

Steve

 

[hacksaw]

You can use FFT with even spacing, but for uneven spacing you need to resort to the classical method using numerical integration of your autocorrelation.