Schiit Yggdrasil Stereophile Review + Measurements

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by purr1n, Jan 20, 2017.

  1. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Whoa... "sub-optimal engineering"?

    sub-optimal.png

    "You might prefer the sound of truncation?" Kind of testy for whoever is behind that FB response. Did whoever write that line even bother listening to the "sound of truncation"?
     
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    19 kHz + 20 kHz IMD

    Sound of Truncation (Yggdrasil - into difficult 600 ohm load):

    [​IMG]

    "Good-performing product" (MSB Technology Analog into no-matter-what):

    [​IMG]
     
  4. The Alchemist

    The Alchemist MOT: Schiit - Here to help!

    Pyrate
    Joined:
    Sep 24, 2015
    Likes Received:
    1,189
    Trophy Points:
    113
    Location:
    The SchiitR
    Home Page:
    I find these comments ridiculous. It sounds like these "Haters" of Schiit Audio products are just jealous and have no idea what they are talking about. Quite honestly, it pisses me off to no end. Sure, I have never had the luxury of owning or even having a loaner unit to hear the Yggdrasil for myself, but damn, Schiit makes outstanding products and a great value and they know what the hell they are doing.

    And no, I am not kissing Schiit's ass, and no I do not have their higher-end DACs or Amps etc. but I do have a bifrost and a Valhalla 2 and they squeeze every bit of audio engineering into their products at an outstanding price and they sound incredible.

    These of course are my only personal opinions, but damn do these idiots even know what they are talking about? I am sorry for the ranting, but Schiit works very hard to squeeze every ounce of quality into their products and offer them at very good prices.

    Like I stated above, I think it comes down to jealousy and idiocy. I cannot even imagine what a Yggdrasil or a Ragnarok sound like when the bifrost and Valhalla 2 to me sound so good.

    Haters are going to be haters, and it really pisses me off that anyone would make such idiotic claims.

    Again, sorry for the rant, but this kind of ignorance really just pisses me off.
     
  5. Soliloqueen

    Soliloqueen Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    225
    Trophy Points:
    43
    spectral content like this is next-to-impossible in music so there's not really any point worrying about any of the artifacts
     
  6. msommers

    msommers High on Epipens

    Pyrate
    Joined:
    Nov 6, 2016
    Likes Received:
    2,750
    Trophy Points:
    113
    Location:
    Calgary, Alberta
    Home Page:
    The article seems much less ridiculous than on Facebook - that comment will definitely be something they wish they could take back.

    The end of the article did state this however:

     
  7. Psalmanazar

    Psalmanazar Most improved member; A+

    Pyrate Slaytanic Cliff Clavin
    Joined:
    Sep 27, 2015
    Likes Received:
    5,345
    Trophy Points:
    113
    The comparison to that Mytek ESS Sabre DAC is telling. Mytek are the Grado of digital gear.
     
  8. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    High resolution jitter spectrum (24-bit USB)

    Sound of Truncation (Yggdrasil)

    [​IMG]

    "High-performance DAC design" (Moon by Simaudio Neo 230HAD):

    [​IMG]
     
  9. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    To John Atkinson's credit, he did say "It's possible, of course, that the former will be rare with music, and that the latter will be obscured by the noise floors of recording". What I find strange is the disconnect between the seemingly dismissive response on Stereophile's Facebook page, and what John Atkinson wrote on Stereophile.com

    I would add that it's not only possible, but that there is 99% chance that of the truncated LSBs will not be an issue. We get 120db of dynamic range from 20 bits. How many playback chains, or even microphone preamps (after proper gain is set) get even close to 120db? The big elephant is the room is how much of your music collection is hires? The vast majority of recordings that I listen to, the best masters, or at least my favorite mixes/masters happen to be 16 bits.
     
  10. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Just read John Atkinson's summary of the Yggdrasil measurements. He concludes with the following:

    It's difficult to sum up the Schiit Yggdrasil's measured behavior. While the processor's analog circuitry is superbly well designed, its digital circuitry appears to have problems with high-level, high-frequency tones, and with the LSBs of 24-bit data. It's possible, of course, that the former will be rare with music, and that the latter will be obscured by the noise floors of recordings. But it does look as if the digital circuitry is not fully optimized. Hopefully, this could be addressed with a firmware upgrade.

    I've read enough of JA and his British-isms to know that he's pissing on the measurement results, e.g. "difficult". I assume that by "problems with high-level, high-frequency tones", he is referring to this IM measurement below, and the regular spurs.

    JA remarks: I then tested the Yggdrasil for intermodulation distortion with an equal mix of 19 and 20kHz tones (fig.10), and while the actual intermodulation products were between 90 and 100dB below the signal's peak level, the noise floor again looked ragged,as in fig.2. I suspect that the digital filter begins to overload with full-scale high-frequency tones.
    [​IMG]
    It's difficult to sum up the Schiit Yggdrasil's measured behavior. While the processor's analog circuitry is superbly well designed, its digital circuitry appears to have problems with high-level, high-frequency tones, and with the LSBs of 24-bit data. It's possible, of course, that the former will be rare with music, and that the latter will be obscured by the noise floors of recordings. But it does look as if the digital circuitry is not fully optimized. Hopefully, this could be addressed with a firmware upgrade.

    Sure, John Atkinson does frame the results in the proper context. However, I can't but feel that his summary statement for the Yggdrasil seems rather disingenuous as we look at the same measurement on the MSB Analog DAC.
    [​IMG]
    This measurement looks worse with distortion products more likely to be within the threshold of human hearing. However, John Atkinson's summary paragraph goes like this: It was a pleasure to measure such a good-performing product as MSB's Analog DAC. And I loved the top-mounted volume control, the display, and the ergonomics. It is rare to encounter an audio component so well thought out and so well engineered.

    Now, here's the same measurement for the NAD M51
    [​IMG]
    NAD's M51 Direct Digital DAC offers measured performance that is almost beyond reproach. Color me impressed.John Atkinson

    The Yggdrasil arguably measures better in this difficult IM measurement than the MSB Analog DAC and NAD M51, yet is labeled "difficult", while the Analog DAC and M51 are said to be "a pleasure to measure such a good performing product ... so well thought out and so well engineered" and "almost beyond reproach. Color me impressed." respectively.

    What gives? This is perplexing indeed.

     
    Last edited: Jan 20, 2017
  11. Smitty

    Smitty Too good for bad vodka - Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    562
    Trophy Points:
    93
    Location:
    Ca$hburn, NoVA
    Well, lesson learned - The Stereophile summaries for measurements are a waste of time.
     
  12. OJneg

    OJneg The Most Insufferable

    Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    3,923
    Trophy Points:
    113
    Location:
    Grand Rapids, MI
    Thing is JA is usually pretty decent with the comments. I think the reason we're surprised here is because there seems to be some inconsistency in standards or perhaps a conflict of interest or I daresay expectation bias for this particular product.
     
  13. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Worse. Wasting time is harmless. Interpreting the results like this seems to show favoritism or the opposite. I almost get the sense that Stereophile hates Schiit, with the measurement summary and testy "sound of truncation" Facebook retort.

    The Analog DAC review was from 2014. I've heard that DAC at the time (and was about to buy it), until I discovered that the performance of the Yggdrasil exceeded it in most respects.
     
  14. SSL

    SSL Friend

    Pyrate
    Joined:
    Nov 12, 2015
    Likes Received:
    1,239
    Trophy Points:
    93
    mfw rounding now means truncation.
     
  15. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Comparisons of my own IM measurements with JA's. I'm not sure how JA came to these conclusions:
    • its digital circuitry appears to have problems with high-level, high-frequency tones
    • digital circuitry is not fully optimized, Hopefully, this could be addressed with a firmware upgrade.
    • I suspect that the digital filter begins to overload with full-scale high-frequency tones. As music only very rarely contains such spectral content, perhaps the filter and DSP circuits have been optimized for low-level signals.
    I'll be the first one to say that these kinds of measurements are extremely difficult and in many instances impossible to replicate, to the point where I would be wary of rushing to judgement. You guys have heard me say every now and then: "Oh, it's a good measurement day! Few artifacts and spurs."

    Here is my measurement. I get the typical random noise from the modest QA400 analyzer, but I'm certainly not seeing the same pattern here as JA.
    [​IMG]
    [​IMG]
     
  16. Smitty

    Smitty Too good for bad vodka - Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    562
    Trophy Points:
    93
    Location:
    Ca$hburn, NoVA
    I'm going with conflict of interest, ad money pays the bills (mine included).
    I was considering the Analog DAC before Yggdrasil was announced, back when my BHSE was still just a down payment and a dream. Yggdrasil just made a lot more sense financially, I still was enthralled by MSB until I learned more, and realized their marketing stuff was BS. The really sad thing is that Stereophile measurements are a really important resource for the community, with a large number of tests with lots of gear, and with a consistent methodology. I almost hope it's just JA kissing some advertising partner's ass, and not letting personal bias/grudges get in the way.
     
  17. dmckean44

    dmckean44 In a Sherwood S6040CP relationship

    Pyrate
    Joined:
    Nov 17, 2016
    Likes Received:
    1,425
    Trophy Points:
    93
    Location:
    Peoria, IL
    JA has to know that there's no such thing as a 24-bit DAC and every audio DAC has an "equivalent number of bits" spec, normally somewhere around 18 or 19 bits. All audio DACs must round down 24-bit material.
     
  18. Merrick

    Merrick A lidless ear

    Pyrate
    Joined:
    Jan 6, 2016
    Likes Received:
    12,568
    Trophy Points:
    113
    Location:
    Portland, OR
    "Whoever wrote that line" is John Atkinson. The comment above the one you posted in that thread is: "We don't have a "social media guy" - I, John Atkinson, post to Facebook that which I think interesting and relevant. The datasheet for the DAC chip clearly states that it is a 20-bit part and my measurements in the review clearly reveal the truncation of 24-bit data."
     
  19. GanGreinke

    GanGreinke Friend

    Pyrate Contributor
    Joined:
    Mar 29, 2016
    Likes Received:
    512
    Trophy Points:
    93
    Location:
    New York, NY
    The Facebook headline and comments seem kind of dismissive of the reviewer, no? It's as if he is saying "Herb Reichert likes the DAC, even though it's outdated and doesn't pass my 'objective' measurements." You'd think that, as the editor, you'd want to highlight your reviewers, not your own personal takes.
     
  20. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I doubt this. I want to believe that he understands, but it's obvious he (or at least Stereophile) doesn't get it. You guys know how hard I've tried and what efforts I've taken to educate people on INL / DNL and how accuracy affects ENOB. But people still don't get it and oftentimes mistakenly think it's the same thing as the typical linearity plots.

    A technicality, possibly a matter of semantics, but Stereophile really should not have made this assumption. The AD5791 takes 20 bits for data in 24 bit register. The other 4 bits tell the chip what that data is for. Since the Yggdrasil runs a custom DSP, the DSP would be responsible for converting the 24 bits to 20 bits, assuming that hires material with 24-bit depth is what is being fed to it. As to how that conversion happens (there is 8x upsampling), I doubt Schiit would have just lopped off the 4 LSBs.

    Now what would be interesting...

    Maybe an analysis of hires music files to see how much of those 4 LSBs is actually noise instead of content. Might be a fun thing for me to do. Need to keep my programming and numerical analysis skills sharp.
     

Share This Page