Life after Yggdrasil: Watering the Ash

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Torq, Mar 1, 2017.

Tags:
  1. bengo

    bengo Friend

    Pyrate
    Joined:
    Apr 17, 2017
    Likes Received:
    1,371
    Trophy Points:
    93
    Location:
    Divisive Kingdom
    Home Page:
    Regarding Ares: I never understood the logic behind making a very low cost discrete R2R design balanced. Wouldn't it be better to spend that money on more accurate resistors (and other improvements)? Presumably it's a fashion / hype based decision, but there may be some solid engineering reason that I am missing.
     
  2. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    In theory, you can get an extra bit of effective resolution by running a pair of converters in differential mode.

    There isn't much headroom in finding more closely tolerances resistors than used in even the Ares, as they're specified at 0.01% already. 0.005% parts do exist, though not necessarily in packages that are suitable for inexpensive, automated, surface-mount assembly. Of course, understanding what those tolerances, and package types, actually mean in practical terms is, perhaps, just as big a concern as what the rated tolerance is. And these factors have a major influence on the specific price of a given resistor/specification.

    On reasons to do this ... well, there are a number ...
    • "Multi-bit" and/or "R-2R" are the current "hot-words" in the DAC world.
    • Audiophiles have a tendency to believe "discrete" implementations (of pretty much anything) are better.
    • Balanced-is-better is another common belief (as with all things ... "it depends").
    The most respected converters that hit these points are expensive. Progressively more so as more of those boxes get ticked. Someone that comes up with a low-cost unit that both sounds good and measures well is going to have a very hot product on their hands.

    Of course, those well-known and respected converters often employ additional technology and methodologies, often proprietary, to get around the very real challenges of coming up with an accurate, consistent, repeatable implementation of a discrete multi-bit converter. Most of that is necessary just to begin to approach the performance and consistency of an IC-based R-2R chip. And that is a big part of what drives up the cost of such units.

    From a cost-engineering perspective, as well as potentially an absolute-performance perspective, you either want to find a suitable/adaptable existing IC-based converter, or have one built to your specifications. Unfortunately the former are almost as rare as hens teeth (and still aren't cheap) and there are very few companies in the serious audiophile space that can afford to have custom ASICs fabbed for them (and even if they could, you'd not want to pay what they'd cost as they'd be so expensive as to be very low-volume ... further hurting pricing).

    --

    So, yes, there are serious challenges in doing a low-cost R-2R converter, discrete or otherwise. Perhaps so much so that, with the current state of the art the lower-price bounds to do it right might be above where some products are currently targeting (in my experience, so far, which certainly doesn't cover everything out there, about $1,500 is the lower limit for something, discrete at least, that I can say is properly competitive). But without trying, the state of the art won't move forward. The first to crack this is probably going to make out like a bandit.
     
    Last edited: Aug 20, 2017
  3. Luckbad

    Luckbad Traded in a unicorn for a Corolla

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    3,408
    Trophy Points:
    113
    Location:
    Holly Springs, NC
    Discrete R-2R seems to be the latest hotness, but man do I think that's a brazen approach to DAC design. More than half a dozen audio companies now believe they can outperform the decades of multi-million dollar research and development that went into the original (and current) multibit digital to analog converters on their first try, then price their products accordingly.

    Holo did a commendable job out the gate. Soekris had some early hiccups in the DIY scene and has started to do it right with carefully controlled manufacturing. Audio-GD, after well over a decade of designs based on PCM1704 chips, came out the gate and tripped on its face with discrete R-2R, sullying the brand in the eyes of many.

    @schiit has the grandfather of standalone DAC design (@baldr), and even they haven't gone so far as to make a discrete R-2R solution of their own, probably because Analog Devices can simply do it better. Will we ever see a discrete R-2R DAC from Schiit? Beats me. If anyone can do it well, Moffat is probably one of those people.
     
  4. Ringingears

    Ringingears Honorary BFF

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    3,659
    Trophy Points:
    113
    Location:
    Northern Californium Valley
    Excuse my ignorance. But did Wadia try a discrete R-2R DAC? I know they used a DSP algorithm as the filter. I have a (now non-functioning) 581i SACD player that I am fairly certain was based on PCM1704 chips. My experience with it was shit, not good Schiit.
     
  5. bengo

    bengo Friend

    Pyrate
    Joined:
    Apr 17, 2017
    Likes Received:
    1,371
    Trophy Points:
    93
    Location:
    Divisive Kingdom
    Home Page:
    I'm surprised we haven't seen any third-party designs based on Metrum Transient modules yet.
     
  6. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I now have the, very comprehensive, measurements performed on the Ares by @atomicbob. They have also been provided to @alvin1118. @atomicbob will post them to the dedicated measurements thread when he returns from some much deserved vacation.

    --

    A cursory look at the data shows that there are, indeed, differences in the noise and distortion components between the left and right channels.

    As to how directly audible those should be, that's another matter and more analysis/discussion will occur there. That said, much of my issue was, as stated, about fatigue over time rather than immediate A/B differences. Listening to either channel, independently, did not result in listening fatigue as quickly as when taken together (though one channel was better than the other in this regard nonetheless).

    It is, for me, most interesting at this point that there are measurable differences that do provide correlation with what I perceived during listening.
     
  7. soekris

    soekris MOT - Soekris Engineering

    Pyrate
    Joined:
    Mar 25, 2016
    Likes Received:
    954
    Trophy Points:
    93
    Location:
    Denmark
    I would like to claim that a good executed Sign Magnitude discrete R-2R DAC beat ANY current chip design, simply because there aren't any current Sign Magnitude R-2R DAC chips anymore.... Then there are other things that affect the sound, like the filters, power supply and buffers that might improve on a chip....

    So if t.ex. one of Torq's preferred DAC suppliers used the Soekris R-2R Sign Magnitude core instead of some industrial DAC chip, the resulting DAC would be even better and actually cheaper to manufacture....
     
  8. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    So ... something that exists is better than something that doesn't?

    Okay ... but that doesn't seem like much of a leap to me ... more in the realm of "the bleedin' obvious" ...

    I don't really have "preferred DAC suppliers". I have preferred DACs. As in individual products. And those span a number of manufacturers and, indeed, technical approaches to D-to-A conversion (and all the attendant periphery).
     
  9. soekris

    soekris MOT - Soekris Engineering

    Pyrate
    Joined:
    Mar 25, 2016
    Likes Received:
    954
    Trophy Points:
    93
    Location:
    Denmark
    Yeah, something that exists is better then something that don't, you have to design with what you can actually get, don't expect any chip company throwing money at something where there isn't a business case anymore.

    Most people agreed that the best Audio DAC chip ever made was the Burr-Brown (later TI) PCM63 or PCM1704. Those were manufactured in a bipolar process and required laser trimming during manufacturing, making them expensive to make, which ultimate resulting in their demise as the market for those parts got smaller and smaller and they became even more expensive to manufacture, while delta sigma parts made in CMOS without needing trimming got better and cheaper....

    Then the ultra precise resistors got smaller and cheaper and FPGAs got cheaper, making space for discrete R-2R DACs at lower cost, and with better Audio performance than any DAC chip ever made, the key is the Sign Magnitude architecture, just like the B-B parts. Today, discrete Sign Magnitude R-2R DACs are the right way of doing the best Audio DAC.

    Yes, Analog Devices still makes some very precise and expensive industrial DAC chips, as there are other kinds of customers for those that are willing to pay for them, but they're not really suited for Audio use....
     
  10. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    Like I said, that seems to be pretty obvious - so I was in agreement ... but people stating the obvious, as an argument, generally raises some kind of sarcastic comment.

    I'd agree with that in general. Personally I've never found a PCM1704-based unit that I liked (1702 worked better for me, or better still the PCM63). But the 1704 certainly has an enviable reputation.

    Some of the high-bandwidth, small-signal stuff I work on is so particular that we have to be concerned about things like variation in board-layer thickness and consistency of BGA connections (on a single device), which always makes me wonder about the degree of variation encountered with discrete resistor-ladder boards produced on normal commercial processes. While I have no doubt they're not nearly as sensitive to such things as the stuff I have to think about, and the effects might, in fact, be completely benign when it comes to the comparatively huge, coarse, stuff going on in an audio-targeted DAC, I do wonder how much variance there is between a "good" and a "bad" board (i.e. at the opposite ends of their respective tolerance windows). Though perhaps the comparatively high parts-count means that those variances within tolerance all roughly cancel out anyway. Either way it's just a point of curiosity/academic interest ... I don't care enough to get a bunch of boards together and measure them all.

    And yet one of the single best regarded, and measuring, DACs currently available makes just that sort of use out of them.

    I'd agree that they're not necessarily intended, or convenient, for audio-use, especially for the VAST majority of DAC vendors that just seem to slap fancy clocks, PSUs, bulky-casework and an XMOS or Amanero USB abortion on an otherwise data-sheet DAC-IC design and shove it out the door with comical pricing, the results being wrung out of them seem to indicate they're quite suitable.
     
  11. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    In the interests of, hopefully, driving more signal and less noise here, and de-biasing how impressions tend to go on DACs around here, I'm going to make a change to which DACs I'm willing to audition and what the requirements are in order for me to do so.

    Note that this does not apply to DACs I have a direct or personal interest in. Things like the RME ADI-2 Pro or the Metrum units with the DAC TWO modules I have sufficient interest in to acquire, review and write-up myself.

    For DACs not in that camp, and for which an SBAF tour is available ...
    • Enough interest in the unit to get a meaningful tour going.
    • A reasonable percentage of those on that tour actually make an sincere attempt to post proper impressions*.
    • I go later in the cycle OR hold my impressions until other's have posted their's**.
    And, again, this applies only to DACs that I don't have specific interest in myself.

    --

    *I have no issue with people not posting impressions on tour/loaner units. Not everyone can, not everyone wants to, indeed not everyone should. So this comment is only targeted at qualifying marginal-interest units with other's being willing to make enough effort of their own for me to bother with it also.
    **In the case where I hold impressions, I will share them with the likes of @Marvey, @Hands and @atomicbob) privately so as to avoid people thinking I'm basing my comments on the listening notes/impressions of others.
     
  12. landroni

    landroni Friend

    Pyrate
    Joined:
    Aug 12, 2016
    Likes Received:
    1,164
    Trophy Points:
    93
    ^ I guess this disqualifies the Mytek Manhattan II from the off.

    I bring it up for the simple reason that apparently these days everything gets compared to the Yggdrasil (duh), and the funny people at Stereophile (Herb Reichert this time) found it fitting to take another dig at Yggdrasil (emphasis mine):
    Given the recent controversies surrounding Stereophile I couldn't give a rat's ass about their flowery prose. But this review specifically seems to be an attempt to position every other lambda DAC as better than Yggdrasil, especially a DAC by a company that has historically very poor reputation around here. Plus an attempt to position MQA as a 'holy grail' for high-end audio, which would disqualify DACs not supporting it. Long story short, it would be useful and interesting to have something like this go through some trusty ears at SBAF to see if Stereophile is caught at it, again...
     
  13. bengo

    bengo Friend

    Pyrate
    Joined:
    Apr 17, 2017
    Likes Received:
    1,371
    Trophy Points:
    93
    Location:
    Divisive Kingdom
    Home Page:
    Femto clocks, decent power supplies, ES9038, MQA stuff... is that it? yawn. Wait, it's £5k? Must be amazing!

    Link to my short impressions below of £3.5k W4S DAC using the same chip. Full disclosure: I'm not sure if it was the SE version or not, but they all sound pretty similar to me.

    Not a lot of DAC talk on here...
     
  14. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    Well, that ... and the fact that of the Mytek units I've heard to date none have put in a performance that results in me wanting to repeat the experience - much less at the lengths required for proper integration into way I audition for "Life after Yggdrasil" posts.

    I'm sympathetic to the cause, but it's not my fight.

    And I don't read their drivel, or put any stock in any of the other "HiFi Press"; by which I mean any publication, site or poster where you can "follow the money" to get a fairly reliable read on how something is going to stack up, whether it'll be praised or panned, and how much nonsensical twatwaffle will get applied in it's name.

    That's a long winded way of saying, "If you take the paid HiFi press at their word, while I might not call you a moron, I'm probably thinking it."

    Easily dismissed by listening*.

    I've spent a fair amount of time comparing MQA to regular PCM, including using material with a known-common-master. MQA does sound different. And, in fact, superficially tends to be more immediately "impressive". But then so do the dynamically compressed shit-shows that pass for a lot of modern mastering. Listen longer, or coming from a point of familiarity with the material and a halfway decent setup (we're talking something in the hundreds-of-dollars, total, not TOTL exotica), and the shortcomings become readily apparent.

    I fully concur!

    As long as it's not my ears! ;)

    --

    *I am of the opinion that if you buy something without auditioning it, and your decisions are significantly swayed by paid media, well, we might have to revisit that "think vs. call" statement. So caveat emptor!
     
    Last edited: Aug 21, 2017
  15. Andre Y

    Andre Y Friend

    Pyrate
    Joined:
    Oct 4, 2015
    Likes Received:
    220
    Trophy Points:
    43
    Location:
    Southern California
    Can I just say how stupid the case of the Mytek Manhattan looks? There's a bunch of companies who use the same industrial designer or people who copy that look, and it's getting pretty bad.
     
  16. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    Always reminds me of the diamond-plate stuff that those of a mouth-breathing persuasion like to adorn their lifted trucks with.

    I honestly thought their latest little DAC/amp was going to be called the "Cletus", not the "Clef".
     
  17. msommers

    msommers High on Epipens

    Pyrate
    Joined:
    Nov 6, 2016
    Likes Received:
    2,750
    Trophy Points:
    113
    Location:
    Calgary, Alberta
    Home Page:
    It's unfortunate SBAF wasn't able to get a Venus (or Terminator) to compare to Yggdrasil. Since both Schiit and Denafrips seem to be touted as high-value, strong performing products, it seems almost a pointless exercise to compare lower priced models as it's highly unlikely either model would best the Yggdrasil, especially at ~$1,000+ less price difference. Was this due to asking for or receiving what was offered for the loaners?

    Denafrips seems pretty new on the scene and being well received and the "hype" seems to extend much farther than HF.
     
  18. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    The goal of my evaluations here is no longer to find an "Yggdrasil beater" - I'm just using Yggdrasil as a well-known, high-quality, reference.

    I don't think the majority here are seriously considering $4-6,000 DACs ... so things that are more likely to be of actual, realistic, interest take precedence, and in this case that's the Ares (~$650) and the Pontus (~$1,620).

    Otherwise I'd be focusing on listening to $40,000 Audionote transformers DACs ... ;)
     
  19. msommers

    msommers High on Epipens

    Pyrate
    Joined:
    Nov 6, 2016
    Likes Received:
    2,750
    Trophy Points:
    113
    Location:
    Calgary, Alberta
    Home Page:
    Audio Note Kits seem pretty intruiging actually given the quality parts used!
     
  20. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I've not looked into their kits.

    I have heard some of their finished products though.

    Didn't know that's what I was listening to at the time. I was shocked when I was told. Mostly because I didn't think it sounded as good as a unit that was about 1/30th it's price. Of course, the cheaper DAC couldn't be used as a kinetic energy weapon, where as the Audio Note unit would probably level a city block if it fell off a standard-height audio rack.

    I am heavily biased though ... I care about realized performance and value a lot more than I care about fancy parts and pretty cases (I'll admit aesthetics are a big thing for me ... otherwise I'd have bought a ZDS a LONG time ago "just because"). If it takes manufacturer "A" 5x the price to match the performance of a product from manufacturer "B", then it's easy math as to which product I'm going to pick.

    But, yes, they do use some very high quality parts. I'm just not sure they're getting results commensurate with their BoM as a result.
     

Share This Page