[castleofargh]

  Most NOS DACs don't have an AAF (strictly speaking its an AIF) at all or only a 1st order one. Whereas all oversampling S-D DACs do, in the digital domain at the least.
 
I have a hypothesis which I'm testing, that's that NOS DACs sound more analogue-like due to lower noise modulation than S-D types. When I added a sharp AIF to my NOS DAC, the noisefloor (subjectively) dropped considerably, resulting in an almost holographic soundstage. This sold me on steep AIFs but its an on-going experiment to answer the question 'how steep should they be?'.

ok noob question, I though that anti aliasing and anti imaging were the same thing. with one in an ADC and the other in the DAC but serving the same function. am I wrong about that part?

 

[RRod]

 
ok noob question, I though that anti aliasing and anti imaging were the same thing. with one in an ADC and the other in the DAC but serving the same function. am I wrong about that part?

 
The results of not using each one are slightly different, hence the distinction. Not using an AA filter gets you aliases, which will always be below ½ the Nyquist frequency. Not using an AI filter gets you images, which are always above ½ Nyquist. But yes, they're performing the same kind of mantra.
 
I would at least agree that not using an AA on the ADC causes more insidious problems than not using an AI on the DAC. Images will tend to be inaudible since they will be above 20kHz, but aliases can certainly be audible since they will map down into the audible spectrum. Of course, oversampling takes care of all this, so it still confuses me why people care so much.

 

[Sapientiam]

 
ok noob question, I though that anti aliasing and anti imaging were the same thing. with one in an ADC and the other in the DAC but serving the same function. am I wrong about that part?

 
What RRod said. Aliasing is more of a problem in practice than imaging - the aliases from not having the ADC filter fall within the audio band, hence are audible directly. They are enharmonic products hence sound terrible and can never be filtered out once created. OTOH images are inaudible directly but their presence does tend to squash dynamics - they raise the perceived noise floor through IMD in downstream electronics.

 

[RRod]

   
What RRod said. Aliasing is more of a problem in practice than imaging - the aliases from not having the ADC filter fall within the audio band, hence are audible directly. They are enharmonic products hence sound terrible and can never be filtered out once created. OTOH images are inaudible directly but their presence does tend to squash dynamics - they raise the perceived noise floor through IMD in downstream electronics.

 
By that ("raise the perceived noise floor") do you mean that the IMD products due to the hi-freq content of the images tends to be spread randomly across the spectrum? It will be interesting to see the results of your test.

 

[Sapientiam]

I think that is the case yes, based on my readings of various bits of literature relevant to audio amplifier distortion. For example the paper here - http://www.keith-snook.info/wireless-world-magazine/Wireless-World-1978/A%20new%20distortion%20measurement.pdf. See in particular the reference to the work of Brockbank and Wass (third column, first page).