Technical discussions related to Analog Devices DSPs (including Blackfin, TigerSHARC, SHARC and ADSP-21xx DSPs).
|
Hello all, I see that the DSP library has two functions for forward FFT of real input, namely rfft_fr16 and rfftrad4_fr16. Those two take fract16 input and give complex_fract16 output. I do not see any functions for doing the inverse FFT, which should take complex_fract16 input and produce fract16 output. Is there any way of doing this? Thanks. John |
|
|
|
On Fri, 1 Aug 2003, john_baker_s wrote: > I see that the DSP library has two functions for forward FFT of real > input, namely rfft_fr16 and rfftrad4_fr16. Those two take fract16 > input and give complex_fract16 output. I do not see any functions for > doing the inverse FFT, which should take complex_fract16 input and > produce fract16 output. Is there any way of doing this? You can just take the complex conjugat of all your values and run that thru the fft (and there's a scale factor of 1/N I think as well). You can't assume the output of the inverse fft will be pure real, so you'll never find a routine that does that. Patience, persistence, truth, Dr. mike |