SBAF DAC Talk II

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Maxx134, Jul 22, 2018.

  1. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Possibly useful data:
    Yggdrasil uses AD5791 converters and the data sheet can be found here:
    https://www.analog.com/media/en/technical-documentation/data-sheets/ad5791.pdf

    Temperature dependencies may be evaluated in figures 33 through 36:

    Figure 33. Full-Scale Error vs. Temperature
    Figure 34. Midscale Error vs. Temperature
    Figure 35. Zero-Scale Error vs. Temperature
    Figure 36. Gain Error vs. Temperature

    Take Gain Error vs. Temperature for an example. Assume device is at room temperature before power is applied and room temperature is 20 °C. We know the outside surface temperature will rise by approximately 13 ~ 15 °C over a period of 2 hours as measured by FLIR which is shown here:
    https://superbestaudiofriends.org/index.php?attachments/yggdrasil-2-hrs-warmup-jpg.716/

    Internal temperatures will be much higher than the external surface temperature. Using a conservative 10 °C rise we can estimate a change in DAC chip temperature from 20 to 45 °C. According to the chart, gain error will change from 0.25 PPM to 0.5 PPM if we assume a ±10V span implementation (the blue line.)

    For longer term differences, please refer to the inferred jitter measurements for measurements posted for:

    3 hours:
    https://superbestaudiofriends.org/index.php?attachments/20150430-yggdrasil-inferred-jitter-png.712/

    408 hours:
    https://superbestaudiofriends.org/index.php?attachments/20150517-yggdrasil-inferred-jitter-png.723/

    Note two significant changes. Reduced jitter spurs and lower noise floor for the 408 hour measurements.
    There are measurable differences both short term warmup and longer term thermal equilibration.
     
  2. Zhanming057

    Zhanming057 Friend

    Pyrate
    Joined:
    Dec 13, 2018
    Likes Received:
    1,033
    Trophy Points:
    93
    Location:
    NYC
    Warmup is a very real thing with both solid state and tube power amps, but the numbers I get are typically in the 15 minute to 1 hour mark and not days. More often it's a deliberate choice of running the system "hot" and then requiring time to get it there.

    I have a Nagra Classic INT that I run with the Mysphere 3.2 and the Raal SR1a. The Nagra is dead quiet despite pushing out 100 watts and is actually a pretty solid choice for a no holds barred "headphone" amp.
    This is helpful, but if a DAC truly needs to be powered on for 400 hours before reaching a performance optimum, that's really not the most user friendly of solutions.

    And the jitter performance difference seems to be quite small - there doesn't seem to be any reason why reaching the long term thermal equilibrium takes 400 hours, and I'm not sure about the significance of that particular measurement, unless it's in the context of, say, sample variation across DACs or differences measured throughout the day.

    At the end of the day, a single measurement isn't ever going to be as good as a solid, convincing explanation of what's behind the perceived changes. It seems that if there is some temp-related bit error correcting going on, the process should be reflected somewhere in the hardware design or firmware?
     
  3. AdvanTech

    AdvanTech Friend

    Pyrate BWC
    Joined:
    May 13, 2016
    Likes Received:
    1,667
    Trophy Points:
    93
    Location:
    NYC
    You're supposed to turn it on and forget about it. How is that not user-friendly?

    It's not just about jitter. You should hear the audible sound quality swings that the DACs go through as they warmup. It's about listening at the end of the day, no?
     
  4. Zhanming057

    Zhanming057 Friend

    Pyrate
    Joined:
    Dec 13, 2018
    Likes Received:
    1,033
    Trophy Points:
    93
    Location:
    NYC
    What if you want to bring it over to a friend's house or a meet? Or if you take long trips and would prefer to power down your apartment completely to prevent electric accidents? Or if you get a blackout? I had two day-long blackouts in the past year, one because of snow damage, the other because of a blown transformer in the neighborhood.

    Again, I am not doubting that there is a difference - but a single jitter and noise floor measure at 400 hours and a 2 hour IR camera measurement doesn't seem like a sufficient amount of explanation as to what's going on, or whether there is a real difference at all.
     
  5. ColtMrFire

    ColtMrFire Writes better fan fics than you

    Pyrate
    Joined:
    Jun 16, 2016
    Likes Received:
    6,226
    Trophy Points:
    113
    Location:
    Austin, TX
    Then go write a Yelp review.
     
  6. SoupRKnowva

    SoupRKnowva Official SBAF South Korean Ambassador

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,249
    Trophy Points:
    93
    Location:
    Austin, TX
    The amount of goal post moving and equivocation happening in this thread the last couple pages is staggering. One dude, @Zhanming057 , rolls in, and proceeds to ask a few, completely reasonable questions, in a non-troll-y manner and this is the response he gets? No wonder people think we're Schiit shills.

    You cant in back to back posts say hes answered the question loads of times, then say he doesnt have time to answer it, then say people dont care that he answers it. Just because we as a collective admire schiit products doesnt mean we cant be critical.

    Its fine for Mike to say he doesnt care to answer, or that he doesnt have time to answer. The problem that @Zhanming057 rightly points out is the pretty dismissive and handwavey manner in which he has answered it in the past. He wasnt condemning all schiit products or saying that Mike was an idiot, he was pointedly addressing one specific issue.

    we're better than this
     
    Last edited: Feb 12, 2019
  7. atomicbob

    atomicbob dScope Yoda

    Pyrate BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    18,652
    Trophy Points:
    113
    Location:
    On planet
    Some clarifications are in order. The measurement suites were performed at 3 hrs and 408 hrs power on times when time was available in the lab, not based on continuous monitoring of the DAC. The differences measured suggest that indeed changes are occurring. As pointed out those are only two points in time. More continuous measurements might be revealing but are rather involved and time consuming. I'm not inclined to invest that kind of time yet, especially when there is a high probability of results being met with dismissive arm waving.

    Our local meets see many using hot transport for DACs; a combination of UPS and 12V inverters, both with true sine outputs. Over the years I've come to appreciate a number of DACs benefit from many hours of power on time, not just Schiit MB.
     
  8. Zhanming057

    Zhanming057 Friend

    Pyrate
    Joined:
    Dec 13, 2018
    Likes Received:
    1,033
    Trophy Points:
    93
    Location:
    NYC
    Fair enough. That said I do see value in measurements rising up to skepticism. Keep in mind that if this was real, scientific research, the bar to establishing a difference would be much, much higher. As in multiple measurements on multiple different systems, with a diverse selection of measuring tools, and with adjustments for the natural dispersion related to minute component-level differences between individual samples of the Yggdrasil, a bit of statistics to work out error bounds, etc.

    I have no intentions of holding anyone here to that standard. It's a hobby after all. But I do feel that a balance could be struck somewhere, and that the criteria of evidence is somewhat higher than what I've seen in this thread and the other links.

    I actually know this guy in Hong Kong who jerry rigged his own UPS to go full take-it-onto-the-subway portable with a Mjolnir 2/ Hugo TT combo driving a pair of TH900's. I thought it was pretty funny and probably a good workout.
     
    Last edited: Feb 12, 2019
  9. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    While I can't fully agree with @ColtMrFire 's slightly cynical response, I can't agree with @SoupRKnowva either because I think @Zhanming057 is actually disrespecting sbaf in less offensive ways. It's pretty similar with the case of dudes requiring rigorous abx double blind testing without suggesting proper amount of funding and reward.

    There exist various degrees of scientific rigorousness (although this varies across different schools, areas, or industries). But I think it's foolish to parrot safe statements (quoted paragraph is a very typical example) that are already taken for granted -- more importantly EVERYBODY already knows. This community is on top of such consensus.

    Regarding multi-bit chip warm-up, I f'ing agree that it may be bothering. Personal bar regarding this issue must be highly subjective. If anybody hates it, he or she should find something else -- and this is partly why I decided going for convert-2 as opposed to yggy2. Truth is, the issue of dac warmup has been repeated every month every year. I emotionally agree with all kind of staggering toward repetitive rando questions that seemingly lack pre-(re)search.

    I don't think myself immune to silly questions. But I am OK that any matured members yell at me "hey go google first!". To me it's a totally acceptable part of sbaf culture.
     
    Last edited: Feb 12, 2019
  10. Zhanming057

    Zhanming057 Friend

    Pyrate
    Joined:
    Dec 13, 2018
    Likes Received:
    1,033
    Trophy Points:
    93
    Location:
    NYC
    You're literally quoting me at the cutoff where my next sentence is "I have no intentions of holding anyone here to that standard". Again, that's not the standard I'm looking for - but I'm not happy with the responses I have gotten so far.

    I did Google the question a couple times and while I am no DAC designer, I have written code related to digital audio, and I don't find the technical writeups impenetrable. I did not find anything convincing - to the contrary, nothing I found about INL suggests that it requires extensive warmup to adjust for. The SBAF quotes above are new to me and I admit that I could have spent more time running search terms, or used Google search on SBAF posts. That said, those responses, while being better than the rest of the discussion surrounding the issue, are similarly unconvincing.

    If this is a question that has been repeated here over and over again - why do multibit DAC's gain in performance upon extensive warmup far exceeding common warmup durations of audio equipment? With enough people asking, surely someone has at least speculation as to the mechanism behind the performance gains or better evidence of long term gains?

    You can't simultaneous say that 1. You're tired of talking about the issue and 2. that Mike doesn't need to address the issue without someone out there having actually addressed the issue. That's taking the easy way out of the question.
     
  11. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    That's "how to disregard sth without being banned 101". I'd call it operational politeness.

    If you didn't intend, DO NOT POST unnecessary bytes.

    That's mostly because we are talking about subjective experience/perception. That cannot be shared without similar audio experience. I figured out wtf people were talking about when I actually purchased Gungnir Multibit a1.

    If you want pseudo science talks, you'd better going somewhere else. Here in sbaf, we value every data point (both subjective and objective) with different weights, based on known history. Different people, different weighting, therefore different weighted average. No surprise.

    PS. DO NOT generalize. I don't think every multibit (or r2r) sounds automatically better. Just particular set of products are implemented well. Even within that manufacturer, I have different attitudes toward their current and previous products.
     
  12. Zhanming057

    Zhanming057 Friend

    Pyrate
    Joined:
    Dec 13, 2018
    Likes Received:
    1,033
    Trophy Points:
    93
    Location:
    NYC
    Okay, let me explain my argument again. Let's say that actual science research is a 100. I wanted a 10 and so far I've been getting 3's and 4's. So I said, hey, there is a 100 out there, and just because I'm asking for a 10 doesn't mean that I'm out to get Schiit and I think that we can arrived at a agreeable criterion, e.g. a well-founded explanation for the long-term warming effect.

    I'm pretty sure what I want is literally the definition of science. Being biased toward people you trust is literally not scientific, at least compared to evaluating evidence on its own merit and judging theories by how well they fit into real data and existing understanding about phenomena. I haven't voiced an opinion either way on this - I don't think that I've demo'd the Yggdrasil enough to have a solid conclusion on this matter. I do think that it's an excellent DAC, regardless of whether or not it performs better after hundreds of hours of heating.
     
  13. SoupRKnowva

    SoupRKnowva Official SBAF South Korean Ambassador

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,249
    Trophy Points:
    93
    Location:
    Austin, TX

    I'm not really sure where you're coming from here. He clearly did read about the topic extensively before posting, you could tell that just from reading his posts. He didn't walk in swinging his dick around asking for abx tests either or behave in a "bad rando" manner, he has been imminently reasonable in his questions and responses.

    Some of the responses from friends however cant have the same said, they have been been troll-y in their content. You for instance seemingly intentionally took his points out of context and then tried to blame him for it, you also built a strawman out of his actual argument. This behavior doesnt "respect" sbaf I wouldn't say.
     
  14. lm4der

    lm4der A very good sport - Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    1,461
    Trophy Points:
    113
    Location:
    Seattle, WA
    I think part of the problem is that there probably isn't going to be a satisfying answer to this question.
     
  15. yotacowboy

    yotacowboy McRibs Kind of Guy

    Pyrate Contributor
    Joined:
    Feb 23, 2016
    Likes Received:
    10,682
    Trophy Points:
    113
    Location:
    NOVA
    Home Page:
    Not trying to be a dick, but where on this forum does it specifically state that the purpose of any post here is to act as the continuation or extension of research with intent to publish to academic literature?
     
  16. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,831
    Trophy Points:
    113
    Location:
    East Coast
    @Zhanming057

    As far as I identified, three possibly separate issues are combined together, summarized as the following questions:

    1. Is notoriously-long warm up for yggy1/2 real? If so, isn't there any (objectively) convincing explanation?
    2. Are Mike's explanations enough or convincing (toward this issue)?
    3. Are Bob's measurements convincing?

    Regarding Q2, I don't disagree with you. And Mike obviously seems to have a firm philosophy toward multibit (as I recall, ultrabike refuted several times). I am not sure if it is valid to criticize Mike's professionalism. That's somewhat in the gray area imo. Conversations mostly done casually in several forums, rather than official places. I believe the best way is to directly contact Mike to ask specific questions, instead of "if I were".

    Regarding Q3, I am not fully convinced either. As Bob already said, it's more like something is happening, rather than proving the issue (in general I internalize nearly every digital-component measurement like this way -- occurs beyond human-hearing thresholds). Any measurement is not immune to reliability -- aka repeatability -- either. But still think voicing it explicitly is futile. Such limitation is already acknowledged by everybody. Also like already posted, "your result doesn't meet science but that's fine" kind of statements touched my nerve. You can freely put different weight (even 0 lol) on that specific single measurement in calculating "conceptual or emotional" weighted averages. But responding to the measurer with non-feasible standard does more harm than good. Even in actual research, at the non-funded stage, everything tends to go with extremely loose standards. Increasing rigorousness generally costs way more than self-voluntary things, let alone "not fun".

    Finally regarding Q1, I believe the former true, but doubt the existence of sufficiently good answers. As already said, I am not fully assured by currently-available explanations. Based on your scale, maybe 3 or 4, too. I've just experienced Modi Multibit (og) and Gungnir Multibit (og) at different time points (right out of the box, 3-6 hours later, 2-3 days later). What assured me was my own experience (and gumby1>>mimby1 in terms of warm-up). Extrapolating my experience with bit-hypothesis, yggy1/2 warm up time sounds possible although I've never heard cold yg. This is likely subject to all kind of cognitive biases, and I don't disagree. But isn't the nature of this hobby based on sharing anecdotal experiences, and putting personalized weights on different views? Back to the topic, if you want 10 but get 3, 7 is likely something you have to find own brutal ways to justify (nobody has the solid "7").
     
    Last edited: Feb 12, 2019
  17. skem

    skem Friend

    Pyrate
    Joined:
    Nov 1, 2017
    Likes Received:
    1,911
    Trophy Points:
    93
    Location:
    Charles River
    Desperately wanting to put an end to the above ...

    The week-long DAC orgy with Convert-2, Solaris, Lavry DA-11, Metrum Adagio, and PSA Direct Stream is winding down. I’m not sure how best to share my findings with SBAF, if I should at all. They largely agree with what Marv and others have already said. But If anybody has suggestions for a comparative study, I’m open.
     
    Last edited: Feb 12, 2019
  18. SoupRKnowva

    SoupRKnowva Official SBAF South Korean Ambassador

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,249
    Trophy Points:
    93
    Location:
    Austin, TX

    That kind of access and time? You gotta share at least something dude
     
  19. msommers

    msommers High on Epipens

    Pyrate
    Joined:
    Nov 6, 2016
    Likes Received:
    2,750
    Trophy Points:
    113
    Location:
    Calgary, Alberta
    Home Page:
    Yeah Skem...ya bestard

    Sharing is caring?
     
  20. skem

    skem Friend

    Pyrate
    Joined:
    Nov 1, 2017
    Likes Received:
    1,911
    Trophy Points:
    93
    Location:
    Charles River
    Note: no Yggdrasil or Gungnir in the mix. Woe.... What will we do? How will we know what to think?
     

Share This Page