For a vector signal analyzer, an anti-alias filter is used to remove unwanted frequency before the sampling. A non-ideal filter has transition band due to roll-off. This transition band can also be used as a guardband. I've read from AN150-15 that when the analyzer computes the inverse FFT, the signals in the guardband are used to provide the most accurate time-domain results. Why is this so? Why not the passband?
:?:
Thanks!
:?:
Thanks!
If you do not include the guardband frequencies in the IFFT process then you will artificially filter the time domain waveform and this would not look the same as the original time waveform obtained from the hardware.
Regards -