03-21-2017 01:49 AM
i have used fft ip core in vivado . i have generated a sine wave from dds compiler. the generated sine wave i have feed to fft core. so i got the result shown in the pic.
i have generated 10mhz sine wave
fft transform length i have choosen is 1024, radix-4 burst i/o,output ordering bit reversal..ect
so my dought is im getting the output but how to predict it frequency domail.
if i apply a signal with frequency F and amplitude A then fft output will be at F the amplitude will be A in frequency domain
plz help how to know the frequency of my fft output
03-21-2017 01:53 AM
03-21-2017 10:30 PM
thanks for reply sir
i got it ,but can u please say somewhat clearly
with following data
fft spike amplitude is 574412538467282136 (signed decimal)
at the spike my index value is 853
at point 23507.4nsec i got that spike
my input sine wave frequency 20mhz
how to find output freqency, for my input frequency, please help
03-22-2017 03:18 AM
03-24-2017 04:58 AM
sir im applying my input as sine wave to fft
sine wave frequency = 20Mhz; sampling frequency = 200Mhz;
with the same sine values in matlab i got a spike at 20mhz.
in xilinx i got spike at at index 745
form your formula N= 745; fz =200Mhz
so, 5*Fs/N - Fs/2N., 5*Fs/N + Fs/2N.
((5*200M /745) - (200M / (2*745)) =1208053
((5*200M /745) -+(200M / (2*745))= 1476510
this are the calculations
at spike it should be 20Mhz na.
12-28-2017 02:57 AM
I am working on FFT core 9.0 , facing the problem of input signal of FFT , ...I have read your comment you did this with dds compiler .... Can you help me how to generate a sine wave signal from dds compiler .......
thanks in advance,
mail id ---- email@example.com
02-11-2019 05:00 AM
Yes, staring at traces can be quite confusing. Manipulating the same traces within something like MATLAB or Octave can be simpler--and might really help you understand what's going on.
When I tested my own FFT implementation, I recorded the input to the FFT, Fourier transformed it using FFTW, and then compared it against the output. The comparison wasn't simple: all kinds of truncation errors accumulate, and the scaling through the system might surprise you. Internal overflow can really be annoying, and even difficult to track down. Further, you have to be careful to properly synchronize to the FFT coming out of your logic. My approach was to record the data going into the FFT, FFT it via FFTW, and then compare that to the data coming out of the FFT.
04-17-2020 12:11 AM
Sir I am facing a similar problem in calculating Tuser Index for FFT, I have generated SINE AND COSINE frequency of 4MHz using DDS , How many bits of TUSER Index will I consider, if:
Point size is N: 4096, and Sampling Frequency is fs: 100MHz ,
so that Frequency Resolution is 100M/4096 =24414.0625
The Expected index will be : I = fin/Frequency Resolution = 4M/ 24414=163 =1010 0011
According to pg109-XFFT document , TUSER width is calculated as: log with base 2 (min. point size) which in my case will be log with base 2(4096) =12
Now if I consider complete 12 bits than it gives a very high frequency, If I consider 6 bits (assuming 6 for cosine and 6 for sine ), than it overflows from expected index
Can You please help me where I am doing mistake in all this calculation.