On Wednesday, January 30, 2013 8:41:42 PM UTC-8, Peter Mairhofer wrote:
> Hi,
>
> I create a signal x, let it through an FIR filter to create y. Then I
> divide the spectra and I should obtain the frequency response of the
> filter. However, what I get is very/noo noisy:
> ...
> Can someone explain this?
>
The frequency response of the digital exitation noise may be near zero at some frequencies, different frequencies for each independent noise vector. Either average many fft(x) and many fft(y) before dividing or try a signal that has non-zero response at all frequencies. An impulse, for example, even has equal non-zero response at all frequencies so there is no need to calculate fft(x).
Dale B. Dalrymple![]()
> Hi,
>
> I create a signal x, let it through an FIR filter to create y. Then I
> divide the spectra and I should obtain the frequency response of the
> filter. However, what I get is very/noo noisy:
> ...
> Can someone explain this?
>
The frequency response of the digital exitation noise may be near zero at some frequencies, different frequencies for each independent noise vector. Either average many fft(x) and many fft(y) before dividing or try a signal that has non-zero response at all frequencies. An impulse, for example, even has equal non-zero response at all frequencies so there is no need to calculate fft(x).
Dale B. Dalrymple