CIC decimation filter increases aliasing and imaging in the signal bandwidth?

[boki]

Dear All,

I am trying [cicdecimate] toolbox, my signal is a 2nd order single-bit
sigma-delta modulator output, the sigma-delta modulator output
spectrum is correct, but the decimation filter output increase the
low-frequency noise, is this correct? I only use a CIC filter no other
low-pass filter connected.

CIC decimation filter increase aliasing and imaging in the signal
bandwidth?


BLOCK:

Voice band sinwave -> 2nd order SDM -> CIC decimation filter

SDM spectrum:
Good

CIC decimation filter spectrum:
Noise++ in signal band, quantization noise--


Thanks a lot.

Boki.

 

[Joel Kolstad]

the decimation filter output increase the
low-frequency noise, is this correct?

Yes. If you decimate by a factor of k, the spectrum gets sliced up into k
equally spaced pieces that are then all summed together. Hence, you get
more noise aliased to both the passband and stopband, but of course if
you're something like -40dB in the stopband this creates .087dB additional
ripple in the passband and is generally not noticeable. On the other hand,
each fold adds about 6dB noise to your stopband... ouch!

This is a basic property of decimation that any book by Mitra or
Oppenheim/Schafer will discuss (and go through the mathematical treatment
of) in length.

---Joel Kolstad

 

[boki]

Got it! Thank you very much.

Best regards,

Boki.