The theory of a transformer "nicely smoothing out" the stair stepping is theory only, as long as transformers are not linear.
How do you think this filter will affect XXHE?
Nice question, and it really makes me think;
The analogue filter as John proposes is indeed something that should be there, ... in theory. Keep this latter in mind, because the theory for one reason, may imply destroyment for others. In fact, the theoritical reason for the analogue filter doesn't even apply in most - if not all cases, because this is about high frequencies (say above 300KHz) your amplifier may not be able to deal with. That is really all ! Furthermore, you can well say that *any* filtering at that stage destroys.
Mind the "at that stage", because it assumes that the necessary filtering to get rid of the harmonic distortion in the audio band has been applied already ...
With the latter in mind, let me twist the story :
What you see in that thread at Bottlehead, is a normal response from NOS lovers. However, down to the merits, and back to the quote above, this response of John makes *no sense* if not first the harmonic distortion has been removed. I know, any NOS lover doesn't know better, and accepts 30% (!!) HD as a fact, but why to add an analogue filter that "nicely smooths" etc. etc.; Harmonic distortion will still be 30% ... (at frequencies like 5KHz and up).
So you see, there is much more to it, and the NOS leage will never understand. They will never understand that it can also be done another way.
Now, it is by pure coincidence that XXHighEnd provides that other way. I mean, it is by pure coincidence that John's espression is valid afterall (the means "transformer" let alone) because Arc Prediction takes care of the real filtering, but in his own context it is not. I only want to say : his response can't be worth all that much in his context of 30% HD.
Please keep notice that I'm not trying to express negatively about John or anyone trying to make the best out of NOS, because it is just normal thinking in that area.
Having said this all, I can only try to explain what happened to myself at trying each and every I/V (current to voltage) stage that exists in this world, and each sounding THE MOST different from the other. This really ranges from 10 second long sounding cymbals (which apparently sound that long) to powerful bass you never heard before, while nothing comes together in one setting. I think I can make this clear by refferring to the Arc Upsampling, which is in te same leage of "being so much different", while in the mean time "so much different" is prone to be very wrong. With in fact all of the I/V stages this ends up to be wrong, because certain types of music won't "play" with it.
I ended up with my very own I/V stage (meaning, this is not an existing design), which bases on doing exactly nothing. As said, I have tried it all, but only "doing nothing" works out the very best. On that matter I don't think anyone can guess by far what that brings by itself. But think of another few jumps like 0.9y-4 is just one.
In the end it all depends. Aren't most of you the most happy already ? well, you all have other DACs, and even the OS ones workout for the better. So, you could just leave it to that ! but that doesn't mean it can't be way better again ...
I know my answers are vague a bit, but what I should tell - and which is hopefully a bit more clear - is that Arc Prediction anticipates on the DAC doing exactly nothing but D/A, and anything interfering with "nothing"
should workout less good. It needs 192KHz at least to do it (but 384 is better again), and it needs 24 bits to do it right (32 bits is better, and Arc Prediction works in that domain, but I think that is overdone).
As I said in the other topic (
Help me find a new DAC), I have an idea. I will work that out now in that topic.
Peter
