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.
384 385 386 387 388 389 390 391 392 393
395 396 397 398 399 400 401 402 403 404
  1. Sapientiam
    I studied the two DSs until I found one example of what you were asking for, then stopped looking for any more. Its already been mentioned on this thread that subjectively the 1704 is a downgrade - what's interesting to me is that most of the measurements tend to point the other way. Incidentally in comparing performance of the two chips they're characterized at different sampling rates - the 1704 at 16X OS and the 63 at only 8X OS. So the 1704 is being asked to work harder.
  2. KeithEmo
    I've heard lots of different opinions about "what jitter sounds like".... My (informal) way of testing it was simply to compare the way a given DAC sounds by itself, and how it sounds when I feed the signal going to it through a "jitter remover" (one which I trust to, by the numbers, actually be reducing the jitter to very low numbers). What I tend to hear is that, with "higher jitter", wire brush cymbals tend to sound somewhat unnatural (more like a burst of steam than like metal on metal), and that sometimes sibilants sound odd (sort of emphasized but not necessarily louder). I hate going into details because these are clearly subjective, but I've asked other people their impressions of what was happening, and theirs seemed to agree with mine. I personally don't hear any significant difference with tracks that contain mostly vocals, or tracks with no strong high frequencies, and I find that only a few recordings seem to be "clear" enough to begin with that I notice the difference.
    Unfortunately, measuring jitter directly is very difficult, and very few companies are set up to do so. (The oft-mentioned J-test doesn't actually measure jitter directly; it delivers a "torture test signal" to the input of the device, then looks for increases in the jitter spectrum at the output. In short, while it provides a reasonable inference about the device's susceptibility to jitter, it neither delivers a jittered signal to test with, nor measures actual jitter at the output.) Note that the output noise spectra caused by jitter are its "symptom", so measuring them DOES give you an accurate picture of how much the sound quality of your output is being affected by jitter.... it just fails to provide detailed information about the error mechanism involved. (So "not seeing significant jitter spectra" really is sufficient proof that "jitter isn't a problem".)
    As for "how easy it is to design for low jitter" - that depends on your design and production process. You need to start out with parts that have very low jitter, and a design that doesn't introduce too much extra jitter, then you have to build it in such a way that you preserve those benefits. As a very wide generality, assuming you start with a perfect signal, commercial S/PDIF interface chips can maintain jitter at or under 100 pS (that's a quote - I haven't measured it). An asynchronous USB interface can deliver lower jitter since, because the clocking is done by the receiving device, the jitter of the incoming signal really doesn't matter.
    Unfortunately, designing for very low jitter tends to entail critical placement and spacing of circuit paths, which pretty well rules out any point-to-point wiring (you need to create a PCB design that works well and keep any interconnect wiring as short and as neat and consistent as possible). This makes it virtually impossible to "breadboard" a really low-jitter design. So, for a company designing a product, it tends to work out to "design the basic circuit, understanding that it won't work very well; once the basic design is good, design a board; send it out to be built; test it and see if it works like you thought; correct any errors; try again; repeat until you get it right" - which is a bit cumbersome for DIY.
    Currawong and Wildcatsare1 like this.
  3. KeithEmo
    Oversampling and upsampling are really two different words for the same process. Upsampling is a more general term for converting any digital audio content to a higher sample rate - regardless of where you do it. Oversampling is usually used to refer to doing that process inside a DAC, and usually in even multiples of the original sample frequency, but they really mean the same thing. (When it's done internally in a DAC, it's always an even multiple, because that's the easiest way to do it - and there's no benefit to using any other frequency instead.) In any case, since you cannot create new information that isn't present in the original audio data, you are using interpolation to "make up" new information. (Bear in mind that the purpose is NOT to create new information; the purpose is to simplify further processing by delivering the same information at a higher sample rate.)
    The math is slightly simpler to upsample to an even multiple, but I wouldn't really consider that an important factor these days.
    A clock is a clock - and it doesn't really matter how you get it... except... from a purely practical point of view the reason for using separate CLOCKS is that creating clock frequencies by dividing a clock by even numbers tends to give you a cleaner clock with less jitter than synthesizing a clock from non-even multiples. (There are chips that can "synthesize" any clock you ask them to, but the clocks they produce tend to not be as clean and jitter-free as clock signals generated by fixed-frequency dividers. Note that this is a purely practical consideration; all that really matters is how good the clock you're using is in the end.) Arguably, the best method would be to use a separate fixed-frequency clock for each clock you need, but the cost would be prohibitive.
    Clemmaster likes this.
  4. evillamer
    PCM1704(1.2mA) seem to have much less current output vs PCM63(2mA). Wonder if that has any effect to the bass slam or dynamics.
  5. Sapientiam
    Interesting point - if opamp I/V's being used then the opamp converting the '63s output has to work almost twice as hard, generating almost 6dB more PSU noise in the process.
    If passive I/V then the output impedance becomes important as it'll determine the PSRR - PCM1704 wins on that but the figure is still very low - 1kohm. If a 100R resistor's used for passive I/V then that only gives 20dB PSRR meaning PSU noise has to be low and stable.
  6. DreamKing
    Yes, I do get a sense that people that compared them like the 63 more. At least, that's all I've heard up to now. I may have to compare them at some point to get a better idea but so far I'm leaning more towards the 63 for a first step in R2R.
    The oversampling stuff is interesting, however.
  7. prot

    You need to concentrate and listen hard to hear diffs between almost any two DACs made in 2015 .. even motherboard soundchips sound very good nowadays and are very hard to abx .. and even if you sometimes do (or think you do) hear differences, the normal average-joe most probably wont.
  8. hbuus

    This may very well be true, but only up to a certain price point, I think.
    For example I recently had the opportunity to make a direct A/B-comparison between a NAD C165BEE and a NAD M51 (both are combined preamp/DAC.)
    The difference in sound signature was very apparent - even to the average joe, I'd suspect.
    So if you DON'T have $2,000 to spend probably not is what you're saying. :wink:
    I will never spend $2000 on a component. Ever. I just can't imagine such a thing. I know there are people with more money than I have or those that value such things more, but I can't see it for me personally. 
  10. LingLing1337
    I recently compared dacs costing $35, $175, $500, and $1500. There was a clear difference between all of them I think all of this "all dacs sound the same" talk has only been around a couple years and probably comes frm the younger generation who has less cash to spend and therefore wants to justify a lo-fi or mid-fi setup as endgame.
  11. hbuus

    Hehe, no :) I was merely referring to an example of two DACs I have listened to recently where there was a clear difference in sound, that's all.
    I totally respect your view on expensive audio btw.
    It's certainly not necessary to spend a zillion $$$ in order to get good sound nowadays, and if you don't feel the price of a piece of equipment is justified for you then indeed, why buy it.
    I think you hit the nail on the head when you say there are people who have more money than you and/or people who value such things more.
    For those people it may very well make sense to buy expensive audio.
    Look at it this way:
    Some people smoke 15-20 cigarettes a day and get enjoyment out of that.
    Some people like to travel to far-away destinations.
    Some people like to buy a new car every 3. year.
    etc. etc.
    We humans tend to spend our money where we get most enjoyment, and for some of us that includes expensive audio equipment :)
    LancerFIN and yuanathan like this.
  12. Sonic Defender Contributor
    Exactly brother, we all have different priorities. I am always shocked at the people who lecture me on my audio expenditures who feel it is more reasonable to have as fancy a car as they can, or to smoke, or travel etc.
    hbuus likes this.
  13. evillamer
    Found this PDF from analog. Really the most detailed and technical dac document you can find:

    dan.gheorghe likes this.
  14. evillamer
    There seem to be an unannounced AKM AK4497 chip  that has SNR 127db
  15. Wildcatsare1
    Hey Guys, the Monarchy 22B I picked up on EBay is shorting, the fuse blows immediately. The Seller claims it was working when he mailed it, if that is true any chance one or both of the transformers were damaged in shipping? Help!!!!
384 385 386 387 388 389 390 391 392 393
395 396 397 398 399 400 401 402 403 404

Share This Page