[Moved]: quantization noise modelling in frequency

[luca89]

I was wondering why the quantization error or noise is described in frequency as limited between Fs/2 and -Fs/2 . I mean I am aware of the meaning of the negative frequency, I just don't understand what limit the error to these limits. At first I though was the antialiasing filter but the quantization error is created after the antialiasing filter so there is no relation between them. Someone has any suggestion?

 

[erikl]

Perhaps this paper may be helpful. See p.65 (p.5 of the PDF): View attachment An_Overview_of_Sigma-Delta_Converters.pdf

 

[zorro]

You are telling about quantization noise or error, so the signal is in discrete time after sampling.
In discrete time, frequencies span the range -π/2<ω≦π/2 in radians/sample. That range correnponds to -Fs/2<f≦Fs/2 in Hz, being Fs the sampling frequency.
Assuming that noise samples are incorrelated, its power is uniformly distributed over that frequency range, giving the spectral density of your figure.

Z