[Audio Empire]

On Tue, 1 Mar 2011 14:03:29 -0800, Sebastian Kaliszewski wrote
(in article ):

Scott wrote:
On Feb 27, 5:14 am, "Arny Krueger" wrote:
"Scott" wrote in message



On Feb 26, 7:39=A0am, Andrew Haley
wrote:
Surely an ADC/DAC that
can be looped through 20 times with no audble
deterioration is flawless in the same way.
Hardly. All it takes is one brick wall filter that is
audibly degrading the original analog signal to shoot this theory
out of the water.
False.

No it is true, If the filter degrades the sound of a particular ADC it
does not matter how many times you loop the signal through the ADC/DAC
with zero degradation.

There is no zero degradation. There is inaudible degradation. Just since
sinc filter (ideal brickwall filter with 0 phase) is unrealisable (it
would require infinite delay) every iteration adds to sound degradation.
All phase nonlinearities will add, all not ideal brickwall effects will
add as well.

The system that system will degrade any
original analog signal.

And it will degrade it on each iteration.

The claim that the transparancy of the loop
proves transparancy of the system is fatally flawed for this reason.

It's not. Signal passes antialias and reconstruction filters on each
iteration, not once.

[...]
This is very poor logic.


There was no logic involved Arny. An assertion was made and I showed a
clear flaw in it.

You showed no flaw. Apparently you just misunderstand how DAC/ADC works.
Filter is an integral part of DAC as well as ADC component. When signal
runs through ADC/DAC chain many times it runs through antialiasing and
reconstruction filter the same number of times as well. All the signal
degradation accumulates as well.

It is a fact that if the filter of an ADC is colored
the ADC/DAC conversion of an analog signl will also be colored no
matter how many times you can run that signal through a loop without
degradation.

And in each iteration those colorations *will* accumulate.

That's the way I understand the process. Well described. Now perhaps if we
strung a series of A/D and D/A converters together and bypass their
antialiasing and reconstruction filters so that we are just re-digitising and
decoding a stair-step waveform over and over again, we might get the result
that Mr. Kruger predicts, but I'd have to think about it as I'm not 100% sure
that even that assumption is correct.

[Arny Krueger]

"Audio Empire" wrote in message


That's the way I understand the process. Well described.
Now perhaps if we strung a series of A/D and D/A
converters together and bypass their antialiasing and
reconstruction filters so that we are just re-digitising
and decoding a stair-step waveform over and over again,
we might get the result that Mr. Kruger predicts, but I'd
have to think about it as I'm not 100% sure that even
that assumption is correct.

Your assumption that we can't string a large number of ADC and DACs together
with their filters intact and not have a transparent chain has been shown to
be false many times. Figure 3-6 cheap ones and maybe 20 or 40 expensive
ones. For most applictions, one ADC and 1 DAC suffices. So, cheap ADCs and
DACs are 3X overkill.