I have a signal I want to apply some signal processing to, at a frequency fi. I have a couple of questions about the best way to approach this with respect to anti-aliasing filters. The signal of interest is pretty polluted with random broadband noise, so in terms of that fact alone, it requires filtering. There is strong noise at about 3fi that will impact the signal. The signal will be sampled with a single-ended 12 bit ADC at fs = 50fi (have lots of bandwidth available, it's low frequency stuff)
Provided the anti-aliasing constraints are observed correctly to cutoff at 2fi, is it better to just design a simple anti-alias filter of nominal attenuation, and do the other filtering in the processor, or is it better to design a higher attenuation active filter?
(perhaps this is a better question for the processor forum...but...it does involve filters) Now I guess the question is, will the FIR filter work as effectively with an anti-aliasing filter attenuation of say, 40dB at 2fi, than one with 80dB at 2fi? FYI - it may make a difference; the FIR filter will be a windowed type, probably a Parks-McClellan.
The signal will be further processed with an FFT for energy density at fi and it's harmonics.