DislikedFirstly, many thanks to CB for the best thread I've ever read on these forums. At last, signs of intelligent life on FF
I've been down to my local university library and copied the Despain article on VFFT.
This isn't really my field, but my first thoughts were similar to CB's comment - that this paper is concerned with a quicker way of computing Fourier Transforms. FxTimer seems to hinting that there is more to it than this. I can't argue with him on this point because it's way over my head.
I've reduced the jpg's as much as possible...Ignored
You are right. I read Despain's paper during my degree and used some of the ideas in my PhD work.
It implements nothing more than a bog-standard Cooley-Tukey FFT with the old CORDIC trick, which avoids the use of floating-point multiplication. Before the advent of cheap CPU's this was a painful operation! CORDIC just uses cheap fixed-point operations.
There is absolutely no mystery here! I can categorically assure you that this is just a cheap and cheerful approximation to a standard FFT. Nothing more!