excel FFT amplitude normalisation

[peppiniello]

Hi to all

who can tell me the normalisation criterium Excel uses calculating the Fast Fourier Transform FFT(Sn), where Sn is a discrete signal?

What i do to obtain the FFT amplitude is the usual evaluation of the complex modulus of the FFT complex array


what i find is:

the original signal Sn has two main armonics, one of which (f1)has a peak of 100
but FFT(f1) is 60000 rather than 100!!

why?


moreover:

doin FFT on the same signal using first Excel , than Mathcad
i find :

Mathcad FFT amplitude = about 50 times Excel FFT amplitude

why again?!

 

[GregLocock]

Because different algorithm designers use different assumptions about the scaling. In my opinion the safest thing to do when using a new algorithm for the first time is to synthesise a signal with two tones of different, known amplitudes. Then FFT that signal. Then study the reported amplitudes. I know that both Excel and MathCad use non-intuitive scaling, related to the bandwidth and the number of lines, but can't remember the details of either. Of course, you could always read the manual, which specifies these things...

Cheers

Greg Locock