Weird apparition on a deconvolution

Started by Michel Rouzic February 23, 2006
I did a deconvolution between two signals, a sweep and an impulse
response (actually an impulse response with a very simplified response,
I lowpass filtered its magnitude bins in the frequency domain to make
it smooth) and as a result, i get what I expect, plus a weird peak
(actually two very close peaks) in the frequency domain. In the
magnitude, as everything goes smoothly between nearly 0.0 and 0.31, the
peaks go up to 105.2 and 19.8, and I have no idea why they are there.

Basically everything is normal, the magnitude of the result looks like
the impulse response's magnitude kinda upside down, but there as those
two peaks, and I have no idea where they're coming from.

Does it mean my deconvolution function might be broken? It's just a
fast convolution function that was proved to work right except that
instead of performing a complex multiplication it performs a complex