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.
377 378 379 380 381 382 383 384 385 386
388 389 390 391 392 393 394 395 396 397
  1. judmarc
    At a 44.1kHz sample rate, the reconstruction filter would have to be a well-nigh perfect brickwall filter, which isn't physically possible, in order to "sufficiently attenuate those images."  That's why "8x oversampling" (interpolation and filtering to raise the bitstream to 352.8/384KHz) became industry standard practice decades ago.
    That's why NOS DACs have fairly high measurable distortion when fed by RedBook.  You may like the sound (not you personally, but a given listener), but it's a result in part of that fairly high distortion.
  2. purrin
    I have not been able to audition any of the Emotiva DACs, but I trust those few who have heard them and compared them to other DACs. I consider Emotiva to be one of the good guys simply because you can see what's inside the box and know that they are not charging obscene amounts like $5000.
    The subtitle of this thread does not proclaim that all R2R is great. Careful readers will note that I dislike the vast majority of PCM1704 DACs, that I have criticized the Metrum ladder DACs for lack of subjective resolving ability, and that I have dismissed other R2R based units, including one which was based on the PCM63.
    Now it just so happens that my most favorite DACs before the Yggy ended up as the Theta, Spectral, and Sonic Frontiers (upgraded/modded) R2R models. PCM63, UltraAnalog, and PCM1702. What can I say? Is this causation or coincidence?
    There will be no hopping on the bandwagon for R2R or R2R/String DACs soon. The sort-of-currently-existing perpetually-soon-to-be-retired PCM1704 kinda sucks (but in a different way of suck from S-D chips), and the Analog Devices chips Schiit is using are a pain in the ass to implement for audio. The niche players like MSB and TotalDAC don't count because they make stuff for rich audiophiles, who are an endangered species in the USA because they are all too old and don't take care of their health (at least from what I saw at T.H.E. SHOW in Irvine.) Finally, I'm not seeing any signs that Analog Devices or other chip makers want to make expensive R2R audio DAC chips. Just no money in it.
    The wonderful thing is that those who like the R2R sound WITH resolution (not R2R syrupy warm thick sounding messes) are in luck. We now have at least some sort of choice at reasonable prices. I guess Schiit will be cleaning up for a few years, especially once multi-bit Gungnir and multi-bit beef-roast come out. (Evidently, Darko said the Gungnir boards are done.)
    azteca x likes this.
  3. blasjw
    I can send you a picture if you like.  [​IMG]
  4. purrin
    I don't think it's fair to have a volume production manufacturer scrounge for parts in the secondary market. DIY'ers can get away with that, but when one's income is tied to actually being able to make something in order to sell it, one does not rely on eBay for parts that were made in the 90s.
    BeyerMonster, azteca x and Ableza like this.
  5. Sapientiam
    Hi Jud - I very much enjoy reading your posts over on CA.
    You raise some important points - just how much is 'sufficiently' ? It depends how much frequency response flatness is needed in the filter and if any droop can be tolerated at 20kHz. The bandwidth available for the reconstruction filter is about 4kHz as the first image (of 20kHz) appears at 24.1kHz. I've played around with various LC filters in this role and its certainly possible to reach -93dB (the noise floor of dithered 16bit) and I'm sure the filter I simulated wasn't 'well nigh perfect' as I included losses. If the difficulty with the filter was the reason for adopting 8X OS as you're suggesting then it becomes considerably easier at 2X OS (the transition band expands from 4kHz to over 40kHz) so the rationale for going all the way to 8X is lost on me. Please do set it out here so I can learn something, or alternatively pick holes in the assumptions used.
    If 'sufficiently' is determined subjectively then -93dB stop band rejection turns out not to be necessary, I've settled on something around -60dB which is achievable with a quasi-elliptic filter (3 inductors, 7 caps). With such a filter, there is measurable distortion at 20kHz (perhaps even going above 3% of 24.1kHz) but subjectively there's no issue as the distortion over the majority of the audio band is considerably lower. Do you have any evidence to offer that the distortion is the reason people like the sound? I'd be interested to learn of it because my experience does not bear out that claim - installing the filter (which reduces the distortion) improved the SQ. If your claim were true I'd have expected a reduced enjoyment from putting the reconstruction filter in circuit.
  6. Sapientiam
    Depends what's meant here by 'fair' and exactly how big their volumes are. I don't think the volumes involved here are sufficient to rule out going to the secondary market, particularly at the prices involved. If the upfront investment were huge that would be a different matter - but at a price below $0.10 (as here) its only chump change for a going concern to buy 20k parts.
  7. AustinValentine
    Soon, definitely not. Schiit has a strong first-mover advantage here to capitalize on. Multi-bit beef-roast sounds pretty tasty. 
    But I'm not sure* that back in the early 2000s that anyone foresaw that a company producing chips for PC audio cards and budget DVD players would become such a ubiquitous presence in high-end audio a little over a decade later. Supply shifts left; ΔQ increases until supply and demand curves hit equilibrium. With enough market buzz to create demand, we'll see headlines like "OPPO Digital Chooses TI's new R2R DAC for Its Flagship Media Player" with incredible speed.
    *Note: I'm not sure because I wasn't in the hobby at the time. Seriously, did anyone see the ESS Sabre train coming down the pipe back then? Not a rhetorical question, I'd love to fill this particular gap in my knowledge.
  8. purrin
    I'm sure Keith made a business decision based on his assessment of risk (availability, expected units for lifetime of product, reliability, returns from breakage, etc.). But of course easier for you to say or judge since your business isn't on the line. Obviously, you are the expert running a successful business making decent audio stuff at affordable prices.
  9. Sapientiam
    Your assumptions are showing - presumably Keith made a decision based on not seeing any issues with S-D - why go out on a limb for no reason when you can get the results you want the easy way? And I've already made the investment in TDA1387s for my own business.
  10. purrin
    It's fine for you because you will sell 5 of them. Emotiva probably sells thousands.
  11. Sapientiam
    Assumptions again! I'm not selling any myself.
    In any case this is a deflection from the original point which is Keith's claim that D-S offers the best bang for the buck. I'm merely providing a counter example to that claim.
  12. purrin
    I merely saying because he runs a real business, he can't rely on hitting Alibabi or eBay 10,000 times in a year to buy 5, 10, 20 pieces of TDA1873 at once for $2.70 or $1.98 each (after shipment to USA). Also, many Chinese sources tend to be a waste of time when they say they have X units available or that can get something (trust me, I know this from experience.) Not to mention if he ended up buying defective units or units that ended up defective at a later time. ICs do last a long time, but c'mon, these are 20 year old parts, he would be totally ****ed. No sane person today would ever consider using the TDA1387 for a DAC which is meant to sell in the tens of thousands, or even more - which is Emotiva's business model, in case you haven't noticed.
    I'm sure Keith's claim was based on not putting the business at unnecessary risk. Maybe while Keith is on the Internet buying 10 pieces of TDA1873 at a time, he can also try to source Pacific Microsonics PMD100/200 HDCD digital filter chips.
    maverickronin and Ableza like this.
  13. Sapientiam
    You're setting up a straw man here. Not just one, several of them - but you're probably already aware of that. I'll only address the first by asking you a question - does Keith buy current production parts from Digikey several hundred times a year?
    If Keith considers it too risky to use a cheap, secondary market multibit part then that's fine by me. But his claim was about bang for the buck, not business risk.
    blasjw likes this.
  14. Poimandres
    Does anyone here have thoughts on the Oppo HA-1? I realize that it is a Sabre DAC however I am curious to read additional opinions on it.
  15. jcx
    Delta Sigma happens to have attractive properties for audio vs R-2R DAC
    differential linearity is important for audio - look at the GedLee Metric (highly weights "zero crossing" nonlinearity, tests well for correlation in listing tests)
    of course the differential linearity is "perfect" in single bit DS - but that has the other processing problems arising from the nonlinear saturation and audible patterns, "birdies" in early lower order implementations
    Multibit Delta Sigma does have to use clever tech to hide the linearity errors of the low (5-7?) bit count internal DAC - but it is proven and measured to work very well, gives differential linearity deep into the noise floor of practical electronics
    so Delta Sigma excel in low level linearity where we have evidence that our hearing cares most
    the trade is the high oversampling, complicated digital processing to achieve the noise spreading and filtering
    there can be issues around noise modulation - but at the levels in current "flagship" monolithic audio DAC chips the evidence for hearing these is scant
    in ESS presentation I believe the "bad competitor" DAC audio band noise floor rose ~ 10 dB from -117 dB to -106-7 dB as signal amplitude rose into the top -10 dB to 0 dB full scale of the converter - that should be Loud!

    using estimates of recording mic noise, home listening room noise floor, masking curves I simply don't see where that "bad" DAC's noise floor modulation is going to be audible with music played in the top 10 dB of the DAC, not even with 120 dB SPL peak system capability
    can you really hear noise modulation within 10 dB of our hearing threshold in quiet at the same time the music is blasting at >100 dB SPL?
    full bit depth R-2R DAC is pushing the tech very hard to keep the major bit carry error below desired audio DAC resolution requirements
    the differential nonlinearity of full bit depth DACs around zero crossing almost always are going to give correlated distortion
    so do you want correlated distortion that is audible or noise that may modulate at levels way below masking thresholds when the output is near full scale - and at a 10x cost differential too
    the yggy's AD5791 actually uses a segmented architecture with the top 6 bits being equal weighted instead of R-2R - its easy to see evidence of cycles of ripple in the INL and glitch vs code plots in the AD5791 datasheet
    I would say the full bit depth approach is "the hard way" today for "hi rez" digital audio
377 378 379 380 381 382 383 384 385 386
388 389 390 391 392 393 394 395 396 397

Share This Page