Hello. I have a question about how to get spectral phase as exact as possible by using fft?

Hello.

I'm interested in spectral analysis on laser pulse. In order to get shortest pulse as possible, not only spectral amplitude is important, but also spectral phase is since the phase plays like how to contruct pulse by using given material appeared as the spectral amplitude.

Anyway, I tested fft on gaussian laser pulse and expected spectral amplitude is obtained while spectral phase is not. The analytic answer is that the phase profile are all zero over the spectrum. However, in this test, sawteech-like phase profile is obtained and applying unwrap command gives continuously decresing linear profile.

I uploaded similar questions before and I got answer that DFT is not able to give perfect phase profile. (I'm afraid that my interpretation of the answer is correct however it sounds like that.)

The following is line to my previous question.

http://www.mathworks.com/matlabcentral/answers/45296-why-does-spectral-phase-from-the-fft-of-gaussian-pulse-shows-sawteeth

Is it really true that getting phase by using fft is hard?

If so, is there any other way to overcome this difficulty?

I still believe that there should be something what I still miss..

[Ramya Ponna]

You need to use the unwrap command to extract the phase of the spectrum.In ur program u have used only angle command to extract the phase but u need to use unwrap(angle(fft of spectrum))

So,add this unwrap command in addition to angle you will get the expected spectral phase :)