Basic FFT megacore output question

Question

When scaled, the output of the FFT megacore will contain the frequency components of its input? I would like to use each of the frequency components individually, this is possible? I would expect it to be.

Thanks for any responses

altera_forum · Answer

the data will come out as a single stream. the order will depend on which mode the core is in

so to "see" individual frequency bins you'll need a counter and maybe a LUT (if you use bit-reversed addressing, for example)

altera_forum · Answer

so the frequency components are given serially?  I've been researching a lot but I still don't understand how the output is structured. Maybe I should look at it mathematically rather then conceptually.(uh oh)  I rather like what you said about using a counter to see the individual bins, but I'm afraid I don't see how this can be done. Can anyone suggest any paper or some kind of document which would provide some insight? THANKS

altera_forum · Answer

the input and output to the FFT core are basically in the same format:

you feed time domain samples sequentially into the streaming input

you receive frequency domain samples sequentially at the output

you should start with the FFT core in natural input order and natural output order mode. this uses more resources and has more latency, but its easier to work with. use the end of packet and output valid signals to reset a counter, increment the counter when the output valid signal is high. now you have a running number of the frequency bin. the bins will go in the order of a non-shifted FFT: 0 to N/2-1 then -N/2 to -1 (i think)

altera_forum · Answer

--- Quote Start ---

the bins will go in the order of a non-shifted FFT: 0 to N/2-1 then -N/2 to -1 (i think)

--- Quote End ---

samples 0 ~ N/2-1 represent frequency range 0 ~ Fs/2 (0 ~ +.5)

samples N/2 ~ N-1 represent frequency range -Fs/2 ~ 0 (-.5 ~ 0)

if you put last half of vector to precede first half (fftshift) then:

samples 0 ~ N/2-1 => frequency range -Fs/2 ~ 0

samples N/2 ~ N-1 => frequency range 0 ~ Fs/2

i.e. dc centred.

each sample can be thought of as brick walled frequency band(bin) with own centre. thus the exact bin centres fall half band from each end of sides.

thus frequencies occur at points separated equally as follows:

f=linspace(0,1-1/N,N) *Fs

and the resolution = Fs/N Hz representing bin width.

altera_forum · Answer

Ooh.  Thanks guys that's great, makes things much clearer

Forum Discussion

Basic FFT megacore output question

7 Replies

Recent Discussions

CPRI F-Tile Dynamic Reconfiguration

Quartus Programmer detects Agilex 7 AGFA014R24B device as AGFA014R24A

Agilex 5 PWRMGT IO pin connections

Shipping package information for 10M25DAF256C8G

Device migration about DD and DA