--- Quote Start ---
I checked your inputs in matlab. All is ok if you scale the output by 2.
for example amplitude given by matlab fft for your input is 29569 which is pretty close to double your first example result.
Now you proved that the problem is indeed to do with fft IP scaling, that is what you started with anyway(-10 should be -1). You now either cautiously use your workaround of rescaling (exp=exp+9) and wait for other people on the form to respond with their experience or ask Altera for help.
--- Quote End ---
I will ask Altera, because waiting is not a solution.
The last idea I have refers to the input signal. Are x(n), the samples, really the real bins OR are they the magnitude? If it would be the magnitude, the real and imaginary bins would not be correct. But then another question would raise... Where is the phase information?!
By the way kaz, did you ever implement an Altera FFT? How does it work at your project?
Honored Contributor
