Date: 07/07/06 (Algorithms)    Keywords: no keywords

Hi

I'm trying to do a DFT using this (VB, sorry not very good at C):

Private Sub Transform()
For n = 0 To 500
Hann(n) = (ChIN(n) - 127) * (1 - Cos(n * 3.1416 / 500)) 'Hann Window
Next n

For k = 1 To 500
r = 0
i = 0
For j = 1 To 500
r = r + Hann(j) * Sin(k * j / 380)
i = i + Hann(j) * Cos(k * j / 380)
Next j
ChIN(k) = 1 / 100 * Sqr(r * r + i * i)
Next k
End Suba

It is working on a microcontroller measuring vibration, taking 500 samples (1byte resolution) and I don't really know how to get rid of the DC (big peak at 0Hz on the frequency spectrum). I though of dividing the signal in half (ChIN(n) -127) but with some signals work and with others it doesn't. Not very stable.... :(

Anyone with experience in DFTs or FFTs?

Source: http://community.livejournal.com/algorithms/81046.html