In my opinion, since you never obtain an impulse in frequency domain for a pure sinusoid, because of implicit windowing that is taking place, i.e., rectangular windowing because you consider sinusoid over limited time. So applying window (hann, hamming etc) to time domain signal decreases the spectral leakage and improves FFT response.
what about analog design.. have you done any simulation of the DSM . here is my design.. but my output seems to be inverted.. anyone know if im doing anything wrong or can explain to me why my waveform is inverted.
my output bit steam when input at Vmax there should be more 1s at the top however my waveform stated otherwise.
my design is a 1st order SDM..
- - - Updated - - -
by the way are you able to show me how you design your LPF and downsampler?? and also how your output waveform look like?? in time /freq domain