I'm afraid I'm not sure exactly what your question is. Are you having trouble doing an fft, doing a peak detection, or are you trying to do a peak detection on the results of an fft to find the primary spectral components. Peak detection generally works by taking the dirivitive of the function. The dirivitive of an fft is going to give some pretty strange results, so normal peak detecktion is probably not what you want to do here. You might just look for fft points above a certain threshhold.