ON 20010317@9:51:34 AM at page:
http://www.piclist.com/techref/microchip/fft/picspect.htm
OAB-CPS-I35a Oliver Antony Broad added 'Says

Regarding the comments about AN542 it sounds to me as if the apnote simply lacked a 'window', and was doing exactly what a 'by the book' FFT should do. In an exercise once I was required to simulate a wien bridge and investigate the output purity, the software was also very sensitive to whether an integer number of cycles were processed. I was suprised at this as it was the evaluation version of a professional CAD package.'