1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Thoughts on a bunch of DACs (and why delta-sigma kinda sucks, just to get you to think about stuff)

Discussion in 'Dedicated Source Components' started by purrin, Dec 5, 2013.
386 387 388 389 390 391 392 393 394 395
397 398 399 400 401 402 403 404 405 406
  1. BassDigger
    Dude, soundchips, instead of DACs?!?
    You've elsewhere mentioned that you, pretty much, don't accept that cables or sources can make any difference.
    Aren't you just arguing that any/all 'High-Fidelity' equipment is unnecessary? Or would it be easier for you to say what 'HiF'i equipment is worthwhile? Do headphones make any difference? [​IMG] 
  2. Argo Duck
    Seriously? Let's dissect.

    Start with a personal opinion: "I doubt that...".
    Quantify the opinion: "That means over 99% of the people".

    Say what? How does your doubt generalize to a statement about others and their capacities or lack of them? Not just 'others' mind you. It's "the people".
    As in 'a whole population'? I and my colleagues didn't realise populations could be measured so easily! We could've spared ourselves a lot of trouble applying careful psychometric principles to population estimation by merely substituting our opinions.

    Next a breath-taking verbal sleight of hand: "since we are discussing the SQ as ... subjective, that percentage is very relevant". Hey you can make anything work by just conflating two independent ideas into the same statement. I guess what is meant here is something like 'because - based on a personal doubt - 99% of people would not subjectively notice a difference, this shows that anyone who subjectively does notice a difference is clearly in a very small minority (1%!) and thus their claim can be disregarded'.

    I kind of admire the spirit of this attempt to make the subjective objective in order that the subjective can then be dismissed, but I fear it fails as a credible way of tackling the problem and I won't be applying it within any of my clients' problem domains anytime soon.

    Next an appeal to authority based on claims of the pedigree of Tom's hardware. What, no "99%" this time? At any rate, no need for critical scrutiny. We can swallow the finding whole because "Tom's" says so!

    The cherry on this particular cake though is this: "I do however have every possible doubt about empty statements like...".

    Quite right. The post below is full of them...

  3. joeexp

    +1 Well put! Couldn't have said it better myself! 
  4. Feuergeist76
    I don't know why here and also in other threads there is the tendency to argue forwards some objectivity - in the real world most philosophers agree upon that there is not such kind of thing called objectivity.

    And especially in the high end hifi world I would say it's even more obvious that it's a pure subjective matter of preference.

    Who the hell likes a 2$ sound chip should be happy - he is saving a lot of money for fency cars, expensive holidays and/or sex, drugs and rock'n roll. But please - let the people spending 2000$ or 20000$ do what they want to do.

    And by the way - a Sabre high end audio chip called ESS9018 bla cost 10$ a peace. It's obvious that you can get a similar, a little bit less quality chip for 0.2$. And that's the story about.

    An AD Chip used in the Yggdrasil cost nearly 100$ a piece - this means that the material costs including labor and development costs of one piece will be around half the price of a unit. And sure the Schiit guys are earning money with their business. That's the story about.

    The transistors in the TotalDAC cost roughly one third of the selling price - just make up your own mind if the unit price is right or wrong.

    I can only tell you that in the industry it's common that around 25-50% of the selling price are production costs. If this is going more to the point that production costs are 1-10% I would felt ripped of. But this is only my point of view. The more unique something is, the more you have a USP. See what Apple is doing - 130$ unit costs and a selling price of 800$. And do you know what? I even like it.

    Back to topic - I personally found out what was mentioned hundreds of times in this thread already - I don't like the slightly robotic sound DS DACs are producing - some more, some less. I hope that the Yggdrasil will solve that - I'm still waiting for my unit.


  5. prot
    ugh. I think some of you are miss-directing your debating energy. We are talking about a test done by Tom's Hardware. If you think something was not right, you are welcome to point the error. And please do cause I tried and couldn't ... and since I have over $10K invested in audio, I am not particularly happy to hear that kind of stuff.

    On short about the few possible errors pointed above:

    There was no "amp issue" because because they compared complete DAC+amp packages: as provided by some MB, a complete o2/odac and a benchmark DAC with HP output. And yes a MB soundchip is basically a DAC ... and here are some measurements for the latest/best such chips. Pretty good if you ask me. And here is another similar test, posted right here on headfi.

    The 99% refers to the population at large. No, I do not have any studies/numbers, it's just an approx based on my exp and some extrapolation: e.g. go on the street and ask people, see if more than 1% know what a DAC even is... or another sample, there are many billions of people which live with less that $1 a day. Generally, hifi as a whole is a 1% game ... or even less. And not sure why would anyone want to debate that.
  6. Feuergeist76
    Hey Prot,

    I just only wanted to add this objectivity thing - it was not pointing at you,it should be more a contribution to the discussion, as I've read a lot argumentations where people would like to convince others that they are right. I think especially in HiFi and probably already in MidFi there is nothing but taste. If you and me would ask people what they think about spending 5figures in Audio stuff they would call us insane - ask my girlfriend ...

    And I totally agree to your last point - I would even say it's more 0.001% then 1%, but the absolute figure doesn't even matter.

    So please don't count my contribution to the discussion to a specific post or to specific postings - it's a more general point of view.

    By the way, for the tests for computer hardware I used to like tomshardware.com. It's quite some years ago that I've followed up their tests but ten years ago I really liked them.


  7. frenchbat
    If you think that comparing apples to oranges is a sound scientific method, I'd like to have a word with your science teacher. But to each his own. Not my concern. Still it would have been easy to redirect all streams to the DAC2's amp. How about the imperfect volume matching ? Impedance ?
    As to the anandtech test, a simple THD graph doesn't give a full picture of the situation, irrespectively of the final quality of said sounchip. For the record, I own a benchmark HDR and I've built 2 O2s, anyone who thinks they sound the same needs to seriously look at his audio chain.
  8. LingLing1337

    If you have $10k invested and cant hear the difference between gear, thats a personal problem. I cant imagine what would satisfy you in this argument other than coming to our houses and watching us pass an abx test. Why dont you stop worrying so much, sell your gear, and maybe take some cooking classes or somethint.
  9. prot

    your msg was not even on when I started writing my response, so you were by default not included :). Also I think yours was a very reasonable response. Same as this one

    I do not see any apples or oranges in Tom's test ... it's $2 dac+amp package vs. $200 dac+amp package vs. $2000 package. And, if anything, the extra amp variable would've skewed the results the other way (i.e. towards even bigger audible differences). Unmatched volume is also the best way to end up with "sounds different" results... didnt happen. In any case, they did match the volume and took care of pretty much all else too .. just read the whole article.

    However, your "for the record" sample sounds indeed quite fruity: HDR DAC vs. O2 amp.
    Anyway, we can discuss such details in the corresponding science thread (see link above). And you are more than welcome to find Tom's error(s) or do a better test that proves otherwise ... I'll be very thankful ... and relieved that I did not waste (most of) my audio money.
  10. LingLing1337

    So you will judge whether you made a good purchase based on someone else's tests rather than just enjoying the muaic...
    nicolo likes this.
  11. frenchbat
    Simple, a 1,5db difference is not trivial. Just try to EQ your bass 1,5db at 100Hz and tell me if there's no difference. No need for a blind test, it's gonna pop out. The fact they missed this means they've screwed up something down the line. Now when it comes to me making a test, I don't claim what these guys are claiming. The proper way to do would be to give the full results and the conditions for repeatability, which they don't.
    Just follow LingLing's advice, and stop worrying about that. Though I'd be you I'd thin the herd a bit. 10k isn't nothing [​IMG]
  12. KeithEmo
    Excellent - and very informative. However, for all its detail and everything it covers, bear in mind that this is still a "basic and general document", and so covers a lot of areas in relatively shallow detail (and a lot of it isn't especially relevant to audio applications).
    You also must be very careful not to take a lot of what it says out of context. This document covers all types of DACs, and all types of applications, so to it "audio" is a very special case, as are Delta-Sigma and Sigma-Delta DACs. For example, a lot of effort is spend discussing settling time, which is basically how fast the output voltage reaches the correct value after changing.... and the illustrations show a step response - which is similar to a square wave. This is an important issue on very high speed DACs (like video DACs), and on slow but very precise successive approximation DACs, but not really on audio DACs - since most non-S/A DAC chips have a settling time in the nanoseconds. This means that most DACs you're likely to see used in audio applications reach the correct output value to within a fraction of an LSB in a very small fraction of a single audio sample period, and so the "error" really only matters as it relates to THD. However, the ringing and time errors caused by the oversampling filter in a Delta-Sigma or Sigma-Delta DAC only get very brief mention (for example on 6.100).  Don't confuse the edge-ringing in the illustrations on settling time with the impulse ringing you see on a Delta-Sigma DAC because of the time response of the oversampling filter - they LOOK superficially similar, but they occur over very different time scales, and are for the most part unrelated. (Also note that AD differentiates between Delta-Sigma DACs and Sigma-Delta DACs, while most people generally interchange the terms in casual conversation.)
    What I'm getting at is that, for all its detail, and all the excellent design information it contains, this paper still barely mentions a lot of things that are very important for AUDIO DAC applications.... For example, it goes into serious depth about input data formats and timing, which is very important if you're designing a DAC, but not really useful otherwise, but barely mentions the types of errors caused by various oversampling filter topologies, and doesn't cover them in any significant detail. (In 6.100, they mention that "the settling time of the digital oversampling filter may make this type of DAC inappropriate for multiplexed applications", but they entirely fail to mention that some people might find the fact that the values may take several filter cycles to reach the correct value to be of possible significance in audio applications.)
    arnaud likes this.
  13. prot

    The realtek device had a 1.4db imbalance through the first few tests. And that made it identifiable. Not anymore after they corrected it. It was a clear mistake, but it just makes the whole thing even more credible & hard to swallow.

    And about thining those $10K, I am not 100% sure what you meant but I am open to suggestions ... e.g. for a start please tell me how to "thin" this

    as about Lingling's&co public displays of love, I am a shy guy and such things make me blush. Sorry guys, I'm still not the subject of this thread. But if you really cannot stop those effusions, just open a new thread and/or feel free to PM ... btw, I'm very partial to logic and beautiful audio gifts :wink:
  14. LingLing1337

    what info do you want? Which DACs they were?
  15. KeithEmo
    I haven't read the test you're talking about - but I do have some general comments on the comments about it..... based on my experience with "computer sound cards" and "motherboard sound chips / implementations".....
    If you look at the specifications on some (even relatively cheap) computer sound cards, they actually do seem to be quite respectable. However, in many cases, the actual performance you get seems to be not nearly as good as the numbers would lead you to expect. The last time I tried to make an actual recording (of a vinyl album) using a sound card that claimed "S/N of 96 dB", I was barely able to actually achieve a noise floor equal to the output of my phono preamp (since this was a sub-$100 phono preamp, which only claimed a S/N of around 70 dB, the sound card should have been a lot quieter and cleaner than it was). Honestly, the problems seemed to be related to the computer's having a poor ground (grounding things very carefully helped a little), and a very noisy power supply, and probably even the computer radiating enough noise into the room to cause some airborn interference with the phono preamp. This may suggest that "it wasn't the sound card's fault since the computer is a truly lousy environment for a sensitive audio component to be installed in", but the bottom line was that, with that sound card in the computer and connected to other equipment, I was totally unable to get even reasonably good performance. 
    Likewise, while the output specs on some sound cards (again even some reasonably cheap ones) suggest that many of them SHOULD sound quite decent, or even "audibly perfect", it's been my experience that none of them actually do manage to deliver good sound quality without bleed through from the screen image into the noise floor, and without unacceptable noise levels in general. Over the years I've had several HP and Dell desktop machines, several Dell, Gateway, and Asus laptops, and a few custom no-name units as well, some with stock MB sound cards, and some with mid-priced add-on cards, and none of them was able to deliver audio performance that wasn't "audibly limited" when compared to even the sound quality of a $59 external DAC. (Since a sound card relies on the computer for its ground and power, and internal sound cards also have to deal with proximity to a massive amount of radiated noise as well, I don't see this as at all surprising. Note that, since grounding seems to be a large part of the issue, plugging headphones directly into the computer is more likely to produce better results than connecting the output to a stereo system or outboard headphone amp because, with headphones, the computer is acting as a "floating" device - and so noise between it's ground and true ground is mostly irrelevant.
    I'm not suggesting that some computers may not have reasonably good sounding "sound systems" - at least under some conditions - however, for me at least, there are just too many variables, and too many opportunities for that not to be the case, for me to even consider using "the internal sound system in the computer" for "anything serious". (Since I have several computers, and lots of other equipment, I simply can't fuss a lot every time I hook up a different one in order to try and squeeze adequate performance out of a particular combination of components. I would really rather have a DAC that "I can trust to work with all my computers".)
    Currawong likes this.
386 387 388 389 390 391 392 393 394 395
397 398 399 400 401 402 403 404 405 406

Share This Page