[spruce music]

   
I'm all in favor of pursuing better sound to make it higher fidelity to the source recording.
 
BUT.....the choice of chip architecture is the area least in need of improvement.  Everything else in the reproduction chain, from the analog side to the headphones themselves and, most importantly, the recordings, are so measurably flawed in comparison to the purely digital portion (assuming low jitter) that it's just weird to obsess over nanoscopic differences there when everything else is much farther from being perfect.

I certainly agree with that.  I would place the biggest problems in order of increasing importance as microphones, speakers, and rooms.  Everything else is orders of magnitude better sorted out.  A special mention of the evils of terrible mastering of recordings that can undo everything else and is far too common.

 

[Sonic Defender]

 
I certainly agree with that.  I would place the biggest problems in order of increasing importance as microphones, speakers, and rooms.  Everything else is orders of magnitude better sorted out.  A special mention of the evils of terrible mastering of recordings that can undo everything else and is far too common.

It is nice to see somebody else mention the humble microphone as a limiting factor. When you consider that a microphone captures all through the air sounds in the studio I think that technology has way more impact on the final product than infinitesimally small differences in digital reconstruction.

 

[watchnerd]

 
It is nice to see somebody else mention the humble microphone as a limiting factor. When you consider that a microphone captures all through the air sounds in the studio I think that technology has way more impact on the final product than infinitesimally small differences in digital reconstruction.

 
Yes, 100%.
 
As much as possible I try to have phones and mics from the same maker (e.g. AKG, Sennheiser).
 
Oh, and it impacts imaging, too...the simple choices between omni, figure 8, cardioid, etc, all have a huge impact on imaging.

 

[disastermouse]

This blog helped me understand the basics. Though it's a bit biased in its conclusions I would agree with what he states at the end based on the differences I've heard.

http://www.mother-of-tone.com/conversion.htm
There's a lot of reading and learning ahead.

Delta-Sigma doesn't 'suck so much' on its own. It's just that after hearing a good R-2R implementation compared to a good Delta-Sigma implementation many people prefer the R-2R DAC for its superior natural reproduction of music. Purrin's thread is a good one to read on Head Fi (it's locked now). You also may want to go over to the DAC-19, Master 11, Master 7, Schiit Yggy, and Gungnir multibit threads. I should point out that I don't hear any less detail in R-2R than Delta-Sigma. I actually hear finer detail that gives me more cues to the timbre of real instruments. In comparison Delta-Sigma sounds unnatural while missing the low level detail.

Purrin's thread:
http://www.head-fi.org/t/693798/thoughts-on-a-bunch-of-dacs-and-why-delta-sigma-kinda-sucks-just-to-get-you-to-think-about-stuff

The basic reason all the major manufacturers are using Delta-Sigma instead of R-2R is cost of production, implementation, and size. Plain and simple. Here's a few good articles on the differences and a bit of history on the devepment of Delta-Sigma.

http://www.mojo-audio.com/blog/dsd-vs-pcm-myth-vs-truth/

http://funwithaudio.blogspot.ca/2012/01/today-in-electronics-everything-is-made.html

http://ankaudiokits.com/DAC-RR2-Part1.pdf

http://www.craigmandigital.com/education/pcm_vs_dsd.aspx

Delta-Sigma is essentially a rough approximation (a very good one) of the original signal, and the birthing point of DSD (or is it vice versa). In my opinion DSD and is a compromise and is basically a low resolution (1 bit) format that is simply super sampled out of necessity, not for audio fidelity. Hearing the difference first hand I now 'get it'. I'm picking on DSD because R-2R can't do DSD (and for a good reason) and recreates PCM in basically a 1-1 fashion without the rough approximation. Delta-Sigma = DSD.

In the end my conclusion is that once you've heard a good R-2R implementation it's hard to go back to Delta-Sigma. To me it was similar to going from a stock iPod Classic to the X5. The iPod was great, until I heard differently. I don't use my iPod anymore.

More on the differences between R-2R and Delta-Sigma (part 1 and part 2).

http://www.positive-feedback.com/Issue65/dac.htm
http://www.positive-feedback.com/Issue66/dsd.htm


As a side note here's a very good article on when higher sampling rates help or hurt audio fidelity. I learned a lot from this one.

http://www.trustmeimascientist.com/2013/02/04/the-science-of-sample-rates-when-higher-is-better-and-when-it-isnt/
Well, that's the problem. For an R-2R DAC to be 300-400 dollars it more than likely will sound like crap because of the implementation requirements.

As far as FOTM comments I feel it's more of a re-awakening than a FOTM, in no small part to the exposure Schiit has brought with the Yggy. I hope more manufacturers bring back R-2R, similar to how vinyl is making a mainstream comeback.

Is Yggdrasil R2R or just a better digital algorithm?

 

[watchnerd]

Is Yggdrasil R2R or just a better digital algorithm?

 
The Yggy is definitely an R2R DAC. The question is if what makes it special is the fact that it's R2R or the digital filter it uses?
 
There has been some comment from Schiit to the effect of (paraphrasing) "the megacomboburrito filter is the DAC."
 
How much of the magic is due to using high precision ladder resistor DACs vs the filter is something that only Schiit can answer (or at least can answer without a huge reverse engineering effort).
 
For me, personally, it's easier for me to conceptualize the majority of the magic happening in the filter.  Nothing wrong with that, BTW. Innovative algorithms are powerful things and the inventors should be rewarded.

 

[landroni]

 

Is Yggdrasil R2R or just a better digital algorithm?

 
The Yggy is definitely an R2R DAC. The question is if what makes it special is the fact that it's R2R or the digital filter it uses?
 

 
Delta-Sigma requires digital filtering stages (e.g. oversampling, noise-shaping...) that are unnecessary or optional in R2R. Same for the analogue output stage (e.g. the huge levels of out-of-band, high-frequency noise that are bound to fry equipment if left unattended).

 
From Mike Moffat's musings on Schiit's super-duper closed-form filter, I've never heard him opine that this filter was implementable with a Delta-Sigma chip. In my understanding most DS chips come with their own filter (based on Parks–McClellan algorithm, or by successive approximation) and some like ES9018 even impose their own analogue stage. Some manufacturers (like Bottlehead) work around the in-built filters by using FPGA. I've never read Moffat or Stoddard suggest or even hint that the filter could be implemented in their lower-end DS designs... If it were, I'd bet that we would have already heard something along these lines for Modi, Bifrost or Gungnir. Or maybe not.

 
More on Yggy's digital filter and what differentiates it from most Parks–McClellan implementations (by Mike Moffat):
http://www.head-fi.org/t/667711/new-schiit-ragnarok-and-yggdrasil/4950#post_11396780

 

[artur9]

I'm not usually one to enjoy the car analogy audiophiles like to use but then this this article by Ars Technica.
 
A quote
 

 Despite the obsession with numbers, cars remain devices of analog communication.

 
TL;DR
Measurements can't tell one everything.
 
So strange to see this argument made in a strictly mechanical field.  No quantum effects in the automative industry, unless @tonykaz says differently.

 

[watchnerd]

 
 
Delta-Sigma requires digital filtering stages (e.g. oversampling, noise-shaping...) that are unnecessary or optional in R2R. Same for the analogue output stage (e.g. the huge levels of out-of-band, high-frequency noise that are bound to fry equipment if left unattended).

 
From Mike Moffat's musings on Schiit's super-duper closed-form filter, I've never heard him opine that this filter was implementable with a Delta-Sigma chip. In my understanding most DS chips come with their own filter (based on Parks–McClellan algorithm, or by successive approximation) and some like ES9018 even impose their own analogue stage. Some manufacturers (like Bottlehead) work around the in-built filters by using FPGA. I've never read Moffat or Stoddard suggest or even hint that the filter could be implemented in their lower-end DS designs... If it were, I'd bet that we would have already heard something along these lines for Modi, Bifrost or Gungnir. Or maybe not.

 
More on Yggy's digital filter and what differentiates it from most Parks–McClellan implementations (by Mike Moffat):
http://www.head-fi.org/t/667711/new-schiit-ragnarok-and-yggdrasil/4950#post_11396780

 
Yes, Parks-MClellan is the default for DS DACs and I've heard Mike talk about how he doesn't like it.
 
My point wasn't that the Yggy filter couldn't be implemented in a DS DAC, but rather the opposite -- what makes the Yggy special sounding may all be in the filter, i.e. if a more generic R2R filter might sound more...well, generic.

 

[landroni]

 
 
Yes, Parks-MClellan is the default for DS DACs and I've heard Mike talk about how he doesn't like it.

 
My point wasn't that the Yggy filter couldn't be implemented in a DS DAC, but rather the opposite -- what makes the Yggy special sounding may all be in the filter, i.e. if a more generic R2R filter might sound more...well, generic.

Bifrost MB uses the same digital filter as Yggy (only slightly dialled down --- from memory, at 4x instead of 8x upsampling), but NO filter at 176.4/192 kHz. At those rates it's "pure" R2R --- no digital filtering is involved and the input is passed directly to the ladder of resistors. Might be fun, if not exactly informative, to test 88.2/96 material vs 176.4/192. The former would use the filter, the latter not.
 
My understanding/intuition on Moffat's filter is that it retrieves the exact form of the digitally captured waveform (i.e. exact points along the waveform), so that interpolated points aren't approximative/guesses, but perfect matches instead.

 

[watchnerd]

  so that interpolated points aren't approximative/guesses, but perfect matches instead.

 
I always find this point tautologically interesting:
 
If the recording as originally made with a sigma-delta ADCs, then the waveforms are an approximation, so then Yggy is making exact matches of an approximation?

 

[landroni]

 

  so that interpolated points aren't approximative/guesses, but perfect matches instead.

 
I always find this point tautologically interesting:

 
If the recording as originally made with a sigma-delta ADCs, then the waveforms are an approximation, so then Yggy is making exact matches of an approximation?

So it seems. Exact interpolation along the waveform as defined by the stored bits... It works with what it is being supplied. Or, as engineers quip: garbage in, garbage out. This still beats the hell out of "approximation of an approximation".
 
BTW, it seems Moffat is 10x unhappier with SD ADCs than with DS DACs.

 

[watchnerd]

 
So it seems. Exact interpolation along the waveform as defined by the stored bits... It works with what it is being supplied. Or, as engineers quip: garbage in, garbage out. This still beats the hell out of "approximation of an approximation".
 
BTW, it seems Moffat is 10x unhappier with SD ADCs than with DS DACs.

 
So here is where I find things get almost almost philosophical:
 
If the mastering engineers are using DS DACs (and most are) through their monitors / headphones while mixing, then isn't the use of a DS DAC closer to the original mixing environment and, thus, artistic intent?
 
In other words, which is more important, precision or accuracy?

 

[cjl]

   
I always find this point tautologically interesting:
 
If the recording as originally made with a sigma-delta ADCs, then the waveforms are an approximation, so then Yggy is making exact matches of an approximation?

There's also the fact that there will always be some error in the reproduction, so it doesn't really make sense to call either approach "exact". An approach involving successive approximation can be incredibly accurate, so really, both should be quantified by accuracy, not by some feelgood theoretical "exact" claim. From what I've seen, the Yggy is very good, with real world performance of about 21 bit. However, I'm very skeptical that it would be audibly distinguishable from a very good S-D DAC, and in fact, even a mediocre S-D DAC (so long as it measured good enough to be expected to be audibly transparent) should sound identical as well. I don't see a lot of point in speculating what makes the Yggdrasil sound so good until it is established that it does, in fact, sound audibly different from any other competent DAC.
 
That having been said, I am very impressed with the measurements they've managed to get out of that thing. Even if the difference isn't audible, I do love seeing a relatively price-is-no-object type approach taken just to see what can be done with it.

 

[Sonic Defender]

Looks like I may have a multi-bit gem on my hand. I was given a working Rotel RCD-955AX CD player which apparently has a nice Phillips mult-bit DAC in it. Does anybody in this thread no much if anything about this unit? I just read a Stereophile review of it (obviously older) where the author was quite convinced it was one of the best values in audio at that time feeling that it could keep up with the more expensive Theta's (although not quite as good). Hell I even have the remote and manual.
 
I used to run an electronic recycling depot and a lady brought this in along with a decent Denon receiver and a pair of very nice Celestion speakers (everything in mint and working condition). The former owner told me to keep them if I liked as she didn't have room anymore, but was happy they wouldn't be smashed.

 

[castleofargh]

for the very little we actually know of this filter, whatever claim of superior fidelity should come with measurements of such superior fidelity. and then some evidence of audibility in blind test.  else to me it's only marketing and I read it as such. they all have their own special sauce somewhere that is better in their own opinions. some are bound to be wrong.
if it's about sounding nicer, then it's subjective and everything goes. if the claim is on superior fidelity, then it should be easy enough to prove.
just like it should be easy enough to show if it's not all going down the drain once it has passed an amp and a transducer and a human ear. I have my personal doubts about this part.