As I understand it, D-S implementations are much cheaper. If this meant that all D-S dacs are equally cheap, fairplay. But my point is that they're sold at a similar price, despite being significantly cheaper to produce.
I'll try not to be so black or white and all-encompassing, in my statements; some, like yourself, do seem to prefer D-S. (Although, I don't recognise any of your adjectives when it comes to describing the 'sound' of an R-2R dac.)
Each to their own, although musical preference probably has some influence.
Your first statement is technically true today - but it really does deserve clarification because, although accurate, it gives a false impression of the economies involved. It's not that, given a certain set of requirements, a D-S DAC chip costs $2, while a R2R chip costs $5. It's more like the fact that, if you want a ride to the airport, a ride in a car is a
LOT cheaper than a ride in a horse-drawn carriage - because there are so few horse carriages available nowadays that you will have to make expensive special arrangements to use one. If you want to design an audio DAC, using a "commercial off-the-shelf 24 bit audio DAC chip", I don't think you can buy an R2R one at
ANY price. Yggdrasil uses DAC chips that are
NOT intended for audio, which not only cost more, but require the designer to design all sorts of extra circuitry (to duplicate functionality that is already present in an "audio DAC chip").
So, in fact, it's
NOT that: "they save a few bucks by using the cheaper D-S chip".
It's more like: "if you don't want to use the standard parts everyone else uses, then you have to do it all from scratch".
(And "hand-made custom-designed" products always cost more - whether they're actually better or not.)
Imagine saying: "I really hate Microsoft Windows, but I also hate Apple software; I think I'll hire someone to design me a computer from scratch" and you're closer to the reality.
(And even the US military has gotten away from that extravagance wherever possible.)
16 bits is pretty much the dividing line between "it's a little more expensive to do it in R2R" and "it's
REALLY hard, and expensive, to do it in R2R, if it can be done as well at all".
The problem with
ALL adjectives is that they are not only imprecise, but they tend to mean different things to different people.
And the problem I personally have with every discussion I've heard lately about "D-S vs R2R DACs" is that they are all based on
someone's adjectives.
Oversampling DACs always alter the transient characteristics of the signal - at least slightly; it's an unavoidable "side effect" of the digital oversampling filter.
However, I am not aware of any specific technical difference in the performance of oversampling D-S and R2R DACs.
Yes,
SOME D-S DACs, with
SPECIFIC circuit designs, may produce a modulated noise floor, which
MAY be audible under some conditions.
And, yes, some R2R DACs have issues with glitching at the transitions, which can also be audible.
(You can't avoid oversampling with a D-S DAC, while an R2R DAC can use oversampling or not, but we seem to be assuming that isn't the issue.)
My point there is that I can show you the ringing of an oversampling filter in oscilloscope photos - and we can compare pix from different DACs.
While I have yet to see a single photo purporting to show how the output of a D-S DAC and an
OVERSAMPLING R2R DAC with a similar digital filter are different.
I want to actually see a picture showing me a D-S DAC that is exhibiting "mistracking" or whose noise floor is "obviously modulated"
(while a picture of that same signal, reproduced by an oversampling R2R DAC, obviously lacks that flaw).
Instead, what I actually see is a lot of technical discussion about highly technical flaws in the way in which D-S works.....
Followed by a bunch of subjective opinions about how specific R2R DACs sound better than other D-S DACs to certain people.
Usually followed by some vague assumption that the two must be related.
But what's missing is any specific correlation.
If the "modulation of the noise floor by D-S DACs" is really audible, then it should be easy to show that all those purportedly great sounding R2R DACs actually have a noise floor that looks and measures differently than the noise floor on all those purportedly bad sounding equivalent D-S DACs. However, if, when we look and take measurements, we find that the theoretical flaw is undetectable in practical terms, then perhaps a lot of people are barking up the wrong trees. Assuming that the assumptions are correct, the it shouldn't be that difficult to actually take the measurements, and show the photos.