[sdiy] BBD question

[Ingo Debus]

Am 03.09.2008 um 02:08 schrieb Nicholas Gregorich:

> Ingo Debus wrote:
>> Am 30.08.2008 um 23:36 schrieb Nicholas Gregorich:
>>> I guess this is the part I'm having a hard time with here, why is  
>>> there not aliasing between every single stage in a BBD?
>> Sampling creates copies of the baseband spectrum (you mentioned it  
>> yourself in anther posting), centered around multiples of the  
>> sampling frequency. If the anti-aliasing filter before the first  
>> stage does its job properly, there are no frequencies above half  
>> the sampling frequency present in the signal, so the sprectra do  
>> not overlap.
>> If this sampled signal is re-sampled (by the second and all  
>> following stages) with the same clock the copies of the baseband  
>> spectrum don't cause aliasing because they are folded back exactly  
>> into the baseband position (and into positions where already  
>> copies of the spectrum were).
>> Makes sense?
>> A different issue is, perhaps BBDs create other noises, not only  
>> the copies of the baseband spectrum. If these contain frequencies  
>> above half the sampling frequency then it would indeed make sense  
>> to place anti-aliasing filters between cascaded BBDs.
>
> I think I go wrapped up somewhere in between over complicating BBDs  
> into an A/D/A system and confusion over re-sampling with the same  
> clock.

The sampling theorem (which my explanations were based upon) applies  
to both BBDs and A/D/A systems. Aliasing is caused by sampling, not  
by A/D/A conversion.

Ingo