Power Spectral Density vs. Amplitude Spectral Density

First of all the "density" term means that all amplitudes from the FFT process must be divided by the resolution band width. This is supposed to normalize measurements taken at different BW's so they all measure the same (this is really valid only for guassian noise, however).

Next, the averaged magnitude of the FFT (divided by the Resolution BW) will be the amplitude spectral density you asked about.

The power spectral density (PSD) is acheived by computing the averaged auto power spectrum, then dividing that by the resolution BW. The auto power spectrum, or APS, is computed by squaring the real and the imaginary terms of each FFT, adding them together, then averaging many of these (typically 100 averages for a pretty smooth PSD).

There are several off-the-shelf programs that will do all of this. One of the best is MatLab plus their DSP Tool box.

Here is a free presentation:

http://zone.ni.com/devzone/cda/tut/p/id/4186 by National Instruments. After doing that, GOOGLE "FFT tutorial" for more detail.

Generally the dat logged is analysed and is made available in PSD form . Is it possible to calculate the amplitude at a certain frequency (within the band ) from the available psd data??? for e.g i have my psd data filtered in 80 Hz and now i require to know the amplitude at 23 hz frequency from the psd data .....

The PSD data is just the magnitude data squared. If you take the square root of any PSD coefficient, you will have the voltage amplitude density at that frequency. Multiply it by the resolution band width to get the original voltage.

Thanx that was good for solving my query !!!!!!!!!!