[watchnerd]

 
Not much use IMO. You HAVE to have experience in product development, parts and design testing and actually produce a DAC that builds / exceeds existing model available at that price point to move the argument forward. It is all back to the theory of digital reproduction in the 80s which was flawed from the get go. Anyone can drown in fancy argue,nets in this subject. True product development and bench testing is how we get the sonic truth.
 
Going back to the Andy Grove article, and the FIR section, he is saying that the Audio Note approach is do the (required) filtering in the analogue domain, in the Audio Note case in the transformers.
The digital route is flawed (in his opinion).
 

 
 
1. On theory and fancy arguments:
 
The signal processing underpinnings of digital audio go back much much farther than the 1980s.  PCM for voice goes back to the 1930s and the Nyquist theorem to the 1960s.  The fancy arguments and product development depend on these mathematic foundations.  Without them, digital audio doesn't exist.
 
The 1970s and 1980s were simply when the economics of production (mainly in silicon) became cheap enough to consider doing it pervasively, first at the professional level and finally at the consumer level.  But digital audio existed prior to that.  Sony/Philips didn't invent digital audio, they invented the CD.  There is a big difference.
 
When it comes to following the theory, the challenge is that most folks, including many EEs working in consumer product development, don't have the background to dive into the heavy theoretical math involved.  And when you can get off the shelf DAC chips to work with, most EEs making consumer or professional products don't need to worry about it; that's the whole point of ASICs.
 
As for "theory of digital reproduction in the 80s which was flawed from the get go", we need to be much more specific than this broad statement because much has been learned and improved upon (especially in the implementations), but a lot of it is also the same, especially at the theoretical level.
 
2. On product development and testing:
 
Engineers who make DAC products and engineers who design DAC chips are usually two different sets of people.
 
The guys making DAC chips have expertise, often at the PhD level, in EE, mathematics, signal processing, etc.  Do you think they don't do product development and bench testing?  In fact many of them run so many tests that they end up writing IEEE papers about their results which are peer-reviewed.  That type of peer-reviewed research is an important aspect of how we get closer to the sonic truth.
 
The guys who make consumer DAC products also do testing, but they're also building on top of the testing that the chip developers do.
 
Product development and testing happens in both places.
 
3. Filters:
 
All filters, digital or analog, involve a set of compromises and trade-offs.  There is no filter that is completely penalty-free (this includes the passive crossovers used in speakers, BTW).  
 
Which filter is the "least bad" is an implementation choice and often has as much to do with the comfort level of the engineer with certain approaches vs others as much as anything.
 
Andy Grove may prefer to use analog filters, and there are plenty of reasons to have that preference, and it may result in good sounding products. But his explanation of FIR is still a pretty bad and inaccurate one.
 
4. Keeping Kosher
 
If one really feels that sigma-delta/delta-sigma approaches and digital filters do horrible things to audio (BTW, Schiit's R2R DACs use digital filters), you're left with some pretty hardcore approaches if you want to avoid being soiled:
 
a. Don't listen to any recent recordings because almost all of it is made using SD ADCs and digital filters.
 
b. Don't listen to any modern remasterings (high res or not) of older stuff originally made in analog  because most of that is now remastered using SD ADCs with digital filters, too.  Or went through an analog->DSD->PCM->DSD chain which involves all sorts of data transforms.
 
c. Track down CDs that were recorded/remastered roughly 10-15 years ago, or earlier.  Make sure they were recorded using R2R ADCs.  Listen to them on either vintage or modern R2R DACs.
 
d. Oh, but be careful about the late-1970s to mid-1980s stuff, too, because a lot of that is from the "Jurassic age of digital audio" (credit to Schiit for coining that term, which I love), which is also bad.
 
e. Go 100% analog as much as possible
 
5. What I Do
 
As for me, personally, when looking at high quality modern products (where the DAC chips themselves are good quality) I give a lot more weight to the quality of the analog portion of a DAC than the specific chip architecture inside. It's the last mile in the chain.
 
And it keeps me sane.

 

 

[jcx]

 ...the Shlitt Yggy is beating DS DACs at 5 times the price on a regular basis....

yet oddly Tyll and Katz didn't fall all over themselves praising the Schiit Yggy for any clear "multibit" advantage over the >5x priced Pro converter they compared that used Delta-Sigma converters
 
http://www.innerfidelity.com/content/big-sound-2015-participant-report-bob-katz#DCwdoB5iHq03VvBb.97
 
nor did they spontaneously comment on the "Holographic Imaging" of Moffat's  "Absolutely Proprietary", "Bit Perfect" 10k tap "megaburrito" filter

 

[watchnerd]

  yet oddly Tyll and Katz didn't fall all over themselves praising the Schiit Yggy for any clear "multibit" advantage over the >5x priced Pro converter they compared that used Delta-Sigma converters
 
http://www.innerfidelity.com/content/big-sound-2015-participant-report-bob-katz#DCwdoB5iHq03VvBb.97
 
nor did they spontaneously comment on the "Holographic Imaging" of Moffat's  "Absolutely Proprietary", "Bit Perfect" 10k tap "megaburrito" filter

 
Also noteworthy is this:
 
"both Tyll and I failed the blind discrimination test, scoring no better than 50-50."
 
Which comes back to the most important question: how much of the differences between R2R vs DS is actually audible under controlled testing?

 

[astrostar59]

   
 
Engineers who make DAC products and engineers who design DAC chips are usually two different sets of people.
 
 

Ooh I love that bit. I watch a video recently by one of the top engineers for the largest DAC chip manufacture in the USA. I will need to dig around for the video, so will come back here. It is on Youtube. He does a seminar at a big audio fest, talking about Delta-Sigma design and its development. Halfway through he says up to early 2000 they NEVER listened to the sound of the chips, only used measuring equipment.
He admitted they realised that measuring equipment was flawed. For more on that check out TotalDAC for their own experiences in that area (measuring equipment). Anyway, they had to employ 'golden eared' audiophiles, to further develop the chip designs. So in my book, digital engineers need to work hand in hand with audio manufacturers who actually listen and tune their product. Lets face it, no car manufacturer would design a car and sell it out the door without a test on the track with that car at it's limits, it would be suicide.
 
This is only part of the puzzle though. To me, regardless of what it says in the techno babble, I can say what I hear. And I seem to prefer any filtering done in the analogue domain. I also seem to prefer bit-perfect at Redbook, no upsampling. I have no interest in DSD or high res PCM, so I am not judging that format, as better or worse. We can have this argument for years, I use my ears, and listen to audio designers I respect in the industry, ones who already make superb amplifiers for example. Because a DAC is basically a digital board inside a pre-amplifier. If the power supply and pre-amplifier are not good, the type and implementation of the chip is wasted.
 
Going back to DS v R-2R, and a post above about price = sound quality, it does not always work. For example the 2k Schlitt Yggy is beating many 10K+ DACs that were up till recently very well regarded. 

 

[Yoga]

  Ooh I love that bit. I watch a video recently by one of the top engineers for the largest DAC chip manufacture in the USA. ... Halfway through he says up to early 2000 they NEVER listened to the sound of the chips, only used measuring equipment.

 
Manufacturing and selling DAC chips is not equivalent to manufacturing and selling audio equipment.

 

[watchnerd]

   
Manufacturing and selling DAC chips is not equivalent to manufacturing and selling audio equipment.

 
Indeed.
 
And, in fact, if you read Jason of Schiit's story about selecting a R2R DAC to use in their multibit products, it starts with them looking at spec sheets.  In that narrative, he then goes on to talk about the prototype board being built.  
 
He doesn't mention is if they grabbed a bunch of other R2R chips and made competing prototypes; one could infer that they didn't.  It reads like they knew what they were shopping for, found it, built it, and then focused on the other parts of the implementation to tune it.

 

[bigfatpaulie]

 
Going back to DS v R-2R, and a post above about price = sound quality, it does not always work. For example the 2k Schlitt Yggy is beating many 10K+ DACs that were up till recently very well regarded. 

 
I realize you keep flogging this point, but saying the Yggy "beats" many 10k DAC's is a comment I have a hard time agreeing with...
 
Beats?  Like in a physical fight?  Like emotionally beats it down?  Like 'sounds better'?  But you said:
 

  To me, regardless of what it says in the techno babble, I can say what I hear.

 
What YOU hear.  Yes, but that's very subjective.  Maybe you also mean, "others who have 'ears I trust' found the same."  Okay.  The problem is that there are just as many with golden ears who would disagree and feel just the opposite making it kind of a null and void.  You can find many who agree with you, reguarless of your opinion on the internet making it very easy to justify to yourself.
 
The spirit of my original comment was more about there being no real way to truly objectively contrast the two technologies.  What was your methodology, by the way? 
 
Saying you prefer it to other gear is 100% fine and good, but where I get my back up is the definitive "This is absolutely the better XYZ".  
 
Another analogy I like is the Rolex vs Timex one.  Who makes better watches?
 
Well, Rolex is better, they are prized status symbols that are built to last generations by artisans.
I agree.
But a Timex is more accurate, has more functions, and can be replaced hundreds of times for the price of a Rolex, so it's better.
I agree.
 
My point is, to me, I think this is all splitting hairs and really they aren't 'better' or 'worse' just different(ish).
 
P.S. Please don't take what I am saying to heart or be mad, it's just the opinion of some guy on the internet.

 

[watchnerd]

 
This is only part of the puzzle though. To me, regardless of what it says in the techno babble, I can say what I hear. And I seem to prefer any filtering done in the analogue domain. I also seem to prefer bit-perfect at Redbook, no upsampling. I have no interest in DSD or high res PCM, so I am not judging that format, as better or worse. We can have this argument for years...

 
Nothing wrong with liking what you like.  
 
But one has to be careful about implying that everything different is wrong without very strong evidence.  These debates can, and do, go on for decades when there is no obviously preferred solution.  The use of feedback in audio circuits (yes or no?), discrete vs opamps, tubes vs solid state, class A vs AB vs D amplification...all of these debates are very much alive.  R2R vs DS may just be another example.

 

[watchnerd]

   For example the 2k Schlitt Yggy is beating many 10K+ DACs that were up till recently very well regarded. 

 
BTW, the Yggy uses a digital filter, and a proprietary one at that.  So if it's getting good reviews relative other DACs, perhaps digital filters aren't completely and categorically horrible...

 

[yage]

   
I stand corrected.  I just had a look at that - very interesting.  I agree, that would probably be the closest you could get to a true A/B.  I would be extremely interested in hearing both Bifrosts side by side.  What I also find interesting is you read so often that R2R is "so expensive to make" and so R2R DAC's HAVE to be expensive yet here we have one for a mere $600: a refreshing change.

 
Yet this comparison is also flawed because if one or the other is definitively better, then it's only better in the context of the Bifrost - not across the board for all DACs. Maybe the Yggdrasil would sound even more amazing with ESS Sabre / TI / Wolfson pushing bits but we'll never know.
 
I like Schiit, but I find the whole R2R debate and the "sample-preserving" filter a good way to differentiate their products in the marketplace - but mostly that.

 

[astrostar59]

   
Yet this comparison is also flawed because if one or the other is definitively better, then it's only better in the context of the Bifrost - not across the board for all DACs. Maybe the Yggdrasil would sound even more amazing with ESS Sabre / TI / Wolfson pushing bits but we'll never know.
 
I like Schiit, but I find the whole R2R debate and the "sample-preserving" filter a good way to differentiate their products in the marketplace - but mostly that.

Yes I agree with all that, except the last statement. I can only speak from my own listening experiences, but the Audio Note R-2R DACs I have heard do sound quite different to DS DACs. However, as we have already discussed, this could be partly the tube regulated power supply and line stage. To me, and the thing that I disliked about digital music up to that point is the treble quality. Female vocals and saxophone, symbols, it just sounded edgy and false compared to my old turntable rig. Well designed tube based DACs can sound incredibly detailed so it is not the 'tubey' thing and rolled off treble going on. I know others who have experienced the same with 15K plus DS DACs and come over the R-2R at much lower entry price points as well. This is not to say all R-2R are better at the same price. It comes down to good design and parts quality as well. But in my mind, the way an R-2R chip turns the RedBook data into voltage is cleaner and simpler than the other methods. Simpler, but also more expensive to implement and is has resolution restrictions, which I ignore as I have no interest in DSD or higher res than Redbook.
 
So, maybe I have found what I want from music based around a digital system, but it is my opinion and my experiences. The best way forward for the DS DAC owners is try and get a demo of an Audio Note DAC, TotalDAC or maybe a Lampi Big 5 with the R-2R chip fitted. Then see if it sounds real to you. It is the only way to know for sure. Of course when you get to a certain level in the hobby there seems to be an issue with dealers and hearing gear side by side. Also Lampizator, TotalDAC, Schlitt Audio and others don't have dealers, they sell direct. This keeps the price down but you have to buy blind. Metrum Acoustics do however, though I have not heard the Pavane, only own the Octave MK2 (which is good for the money).
 
There is however a send it back in 12 days with those manufacturers I believe. The ideal dealer would be a broad selection of R-2R DACs alongside other DS DACs at that price point. Then this subject would be much clearer to all of us.... just hook it up to your system and then decide.

 

[pedalhead]

IMHO the argument boils down to the fact that most people (myself included) enjoy the way music sounds with some distortion added a la tubes, R2R, non-oversampling dacs and DSD. The distortion brings a level of realism to the music that perhaps reflects how we hear sounds in the real world. After playing with various types of dac, over and non-oversampling, software DSD upsampling etc this is my current theory at least.

 

[watchnerd]

 
Well designed tube based DACs can sound incredibly detailed so it is not the 'tubey' thing and rolled off treble going on.

 
It probably is the "tubey thing", actually, but not because of rolled off treble; the likely causes are other qualities that tubes exhibit.
 
The first is even-order (2nd, 4th, etc.) harmonic distortion that many people find euphonic and pleasant, myself included.  The distortion is also progressive, increasing in harmonic content (still even order) with volume, much like real life acoustic instruments.  All of this is easily measurable, easily audible, and most people can pick it out easily in an A/B test.  It's not subtle (unlike DAC chip differences).
 
The graph below is a perfect example of this. Two tube circuits were subjected to a 1khz sine wave test signal (red and blue spikes at 1 khz) and FFT analysis. The 6AU6A (red circuit and graph) is a pentode and exhibits a 2nd order harmonic (red spikes at 2 khz) only 48 dB below the test signal.  The 6SN7GTB (blue circuit and graph) is a triode and exhibits a 2nd order harmonic (blue spikes at 2 khz) a shallow 52 dB below the test signal. Both of these are very much above the noise floors of digital, and even above the noise floor of vinyl.  
 

 
 
Another distortion aspect that is makes tubes more euphonic is that when they do get overloaded, they exhibit soft clipping, unlike solid state can have sharp-edged clipping and square waves with huge amounts of high order distortion harmonics.    
 

 
 
Even-order distortion (in modest amounts) is so well known as being euphonic that you can buy (since at least the 1970s) recording and broadcast gear that adds it to the mix (vintage brochure below; they're much cheaper now):
 

 
 
The Aphex Exciter is the best example and has been in production for decades, both in hardware and now in software form.
 

 
Okay, so what does this mean for DACs and specifically the R2R vs DS debate?
 
1: Tubes Can Overwhelm DAC Chip Differences
 
Regardless of the DAC chip architecture, if tubes are used in any significant way in the signal path of the output stage the tubes will almost certainly dominate the perceived character to a greater degree than the DAC chip itself. This is due to the fact that tubes have irrefutable sonic effects both well within the main frequency range of human hearing and well above the noise threshold of any playback media.  It ain't subtle.
 
So if you like tube-based DACs (nothing wrong with that and I happen to like tubes in my signal path, too), that's a great thing to know about your preferences and what gives you enjoyment. But it's probably more due to the tubes than whether it uses DS or R2R DAC chips.  So if one likes tubes, I would keep an open mind about about chip architectures when shopping.
 
2. Why Put The Tubes in the DAC Itself?
 
Tubes operate on the analog chain.  But even if one likes the effect, why put the tubes in the DAC itself?  
 
Unless the DAC is also a preamp, I don't see a huge benefit to having them in the DAC itself, and lots of reasons not to do it there (power supply issues, noise, etc.), as opposed to external.
 
 
3. We're Stupid Consumers
 
If people like euphonic distortion (and there is a lot of evidence that they do), and euphonic distortion makes things sound more pleasant, live, and less fake or "digital", then we should stop wasting money on crazy audiophile voodoo.  We should go out and buy cheap pro-grade hardware or software exciters for a couple hundred bucks and start experimenting.  If we like it, we've probably saved ourselves thousands of dollars in niche audiophile gear that tries to accomplish the same thing that pros have been able to do for decades, and much more cheaply.  
 

 

 

[watchnerd]

IMHO the argument boils down to the fact that most people (myself included) enjoy the way music sounds with some distortion added a la tubes, R2R, non-oversampling dacs and DSD. The distortion brings a level of realism to the music that perhaps reflects how we hear sounds in the real world. After playing with various types of dac, over and non-oversampling, software DSD upsampling etc this is my current theory at least.

 
I concur.

 

[astrostar59]

I agree with most of that above. But we also have the inverse of the tube effect. The fact solid state amplification can and often does sound cold and artificial. I can't remember how many demo rooms I have heard at shows, with the same DAC and speakers even, and when the SS amp is swopped out for a SET, then we are talking, things start to sound more REAL.
 
I am unsure if the (slight) friendly distortion of tubes is the only key thing here. Transistor and FETs are amplify sound in a very different way. Also I seem to prefer tube rectified power supplies when I have heard the same pre-amplifier with both SS and tube PCB stage as an A/B.
 
The other thing I began to realise (in my view / taste) is having some tubes in the music seem somewhere helps bring a realism back to digital. Without is, i.e. SS straight through, I just didn't get it, it did not convince me it was real music, only impressive hifi. For example I bought a very good tubed pre-amplifier way back to drive my solid state mono blocks, it seemed to work well.
 
Back to tubes in a DAC and is it worth it? Well, a DAC of ANY description is a pre-amplifier of some sort or other i.e. it has to amplify the digital signal once it is brought over to voltage / analogue. At this point the signal is minute and has to be given gain to go anywhere. Some DACs have a true line stage that can drive a power amplifier directly like the TotalDAC and my modded Audio Note DAC. Even a DAC with a simple opp amp is going to affect the final sound hugely. Remember, when the signal is tiny and fragile (bit like in a phono stage with 50 DB+ gain) it needs to be protected and cherished, any damage at this stage will be amplified to an incredible degree further on. 
 
You know, there are many manufacturers who build tube amplifiers and believe the old faithful triode a more prefect amplifying device even against the most modern electronics.