0
$\begingroup$

I want to improve the frequency resolution of the FFT algorithm. Can CZT (Chirp z transform) be better to use after FFT. Because i read in some articles that CZT has better frequency resolution than FFT but it is computationally more expensive than FFT. So, if i use FFT first and then CZT will help me to improve my frequency resolution.

Please guide me. Thank you.

$\endgroup$
0
$\begingroup$

the FFT is an implementation of the DFT, which is a special case of the CZT. So, no, doing an FFT first and then another CZT will not work – it will take you back into something "time-domain-y".

You're probably just looking for other frequency estimators than the DFT, if you have knowledge about signal structure that allows you to take advantage of other CZTs.

So, I'd recommend asking a new question about your actual spectral estimation problem, describing why you think the CZT is an appropriate method here, and what you actually want to do!

$\endgroup$
6
  • $\begingroup$ I am currently working on weak pulsars signal, where i want to improve the frequency resolution of my signal, so i take the FFT of my signal, but the frequency resolution is not good so to improve the frequency resolution of my FFT i need a technique which can increase the frequency resolution. I tried padding and interpolation, but results is not promising. $\endgroup$
    – Tariq
    Dec 6 '21 at 7:52
  • $\begingroup$ @Tariq: what exactly are you trying to do? Determining the frequency of a signal with high precision or distinguishing two or more signal that are very close in frequency? $\endgroup$
    – Hilmar
    Dec 6 '21 at 12:27
  • $\begingroup$ yes, sounds like the FFT might simply not be the tool of choice. A different CZT instead of the FFT might work, but as far as I understand pulsars have relatively narrowband spectrum, you might want to use a significant amount of filtering to select bands, and then do a parametric estimation on these – after all, you're looking for exactly one tone. Maybe also talk to the radioastronomy buffs in chat.gnuradio.org/#/room/#radioastronomy:gnuradio.org $\endgroup$ Dec 6 '21 at 12:31
  • $\begingroup$ @Hilmar my goal is to find the period of a weak signal, for this i need higher frequency resolution. $\endgroup$
    – Tariq
    Dec 6 '21 at 19:46
  • $\begingroup$ @MarcusMüller yes, you are right. But FFT is widely used algorithm for pulsars, so i tried to use ZCT, but it is computational expensive more than FFT. $\endgroup$
    – Tariq
    Dec 6 '21 at 19:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.