Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

FFT: How Twiddle Factors change with multiple input iterations

Status
Not open for further replies.

khaebler

Newbie level 3
Joined
Jan 22, 2018
Messages
3
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
36
I'm building an N=8192 input FFT with 8 channels in Matlab (later to be implemented on FPGA) and I'm having problems with how the Twiddle Factors should look like. The basic structure of the FFT I learned from a YouTube video, and the signal flow picture is attached here for N=8.

YouTube Tutorial.JPG

I implemented the above structure in Matlab and it works for N=8 data points, but as I increase N it no longer works. I suspect that it has something to do with the Twiddle Factors.

How do the Twiddle Factors change after each iteration of inputs? I assume they are exponentiated by a factor of 8 after each iteration. Or is there more to keep into account? I've made a signal flow diagram for my (not fully working) code in Matlab
Presentation1.jpg

Any advice is greatly appreciated
 

Hi,

There should be plenty of ready to use FFT code.
No need to code it on your own.

Klaus
 

Thanks Klaus,
But I'm actually doing this in Matlab as an exercise for how it will be implemented in FPGA. There are even built in functions in matlab to do it all for me, but I need to understand this first before making it in FPGA.

Would you happen to know how the Twiddle factors change after each iteration of inputs? Thanks
 

@ads-see, thanks for the suggestion, but the twiddle factors in that link are much different than the ones I'm looking for. For that matter, I have look extensively at a lot of documentation online about twiddle factors but have not found the answer I'm looking for.

All documentation, including the one you present, shows the twiddle factors where the number of inputs N is the same as the number of channels (that specific link has N=8). Finding the twiddle factors for that case I am quite aware of and have no problem with.

My question is how the twiddle factors change when N is much greater than the channels. In other words, after another iteration of 8 inputs. I did indeed post two pictures in my original question which are the same as what you provided me, so I am aware of those twiddle factors. What I don't know is how the exponent (or k value) changes after each set of 8 inputs.

Does that make sense? Looking at the difference between the signal flow diagram I made (the blue picture) and other pictures of butterfly diagrams will perhaps make it more clear that what I am trying to do is a modification of what most people are doing, and my twiddle factors change after each iteration of inputs unlike most schemes.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top