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.
393 394 395 396 397 398 399 400 401 402
404 405 406 407 408 409 410 411 412 413
  1. evillamer
    There has been quite alot of technology that tries to "restore" audio. Example will be BBE Sonic Max or Creative Xfi Crystalizer or Pioneer Sound Retriever to name a few.

    These audio processing software alter the transients quite abit. If done incorrectly or with certain music types, will likely exceed the volume headroom and distort.
  2. evillamer
    Seems like Absolute pitch is a god given/natural born talent/perk(fallout 4 anyone?).
    It seems that you can be trained to be relative pitch
  3. Sonic Defender Contributor
    I certainly subscribe to your way of thinking, but I'm not sure the new planar offerings from HiFi Man for instance would be a slam-dunk compliment as the 560 and now 1000 are very well extended with plenty of high frequency energy. Still, I get what you mean and agree. I wouldn't get a detailed implementation of a Sabre chip and use it to listen to poorly mastered rock with a Grado SR325. No sir. Now throw in an HE 500 or Oppo headphone ... different story.
  4. prot

    Oppo, older hfman, any audeze, senn 6x0 and prolly many others that I never tried .. and an even bigger selection of dark/smooth speakers. Or just use pretty much any tube amp or most of the small portable amps ... or (the horror), use an EQ :)
  5. jcx
    "perfect pitch" isn't necessarily a great thing to have - pitch standards have changed over the centuries, there are still a few concert pianists that like the last previous standard lower pitch - the whole orchestra has to tune to their piano
    relative pitch is what most everyone uses, gives the sensations of consonance/dissonance with chords, temperaments
    even the uncorrected Sana Clip+ is fine to listen by itself with its software 0.25% pitch error
    wahsmoh likes this.
  6. Sapientiam
    Its not really an explanation but the cause seems to be noise floor modulation - in other words signal-dependent changes in the background noise. Its not the first two of your suggestions, could be a combination of the last two. Quantisation needs to be correctly dithered to avoid modulation noise - that means a certain amplitude and probability density function noise source needs to be added in prior to word length reductions. From my (fairly limited) knowledge this precise dither source isn't used at the quantizer in S-D converters. If anyone has links to show me I'm wrong here and that it is, I'll be glad to follow them up.
  7. blasjw
    Here's a quote from Børge Strand-Bergesen which I think sums it up nicely:
    An intriguing side issue was brought up by Strand-Bergesen when he was explaining why he chose the AKM DAC chip to sit inside his chassis. One of the DACs under consideration was the current DAC flavour of the month, the ESS Sabre. Unlike many manufacturers out there, though, Strand-Bergesen rejected the Sabre out of hand, he believes that the design of this DAC is fundamentally flawed, “It’s riddled with maths trouble. The digital maths is disappointing. It will overflow internal registers, chopping up and generating a lot of output edges. It sounds distorted around the treble area. With a high power amp, it could kill the treble. I have a friend of mine who has access to the in-house data sheets of the Sabre. He actually has a work-around for it. Well, for the 9012 and 9018 chip models. Although on the 9022 and 9023, you cannot initiate the work-around because there are no pins to access the internal registers and bypass this problem.”
    evillamer likes this.
  8. Sapientiam
    He believes - that statement doesn't give anything much of evidence on which he's based his belief. For example, when does the overflow of the internal registers occur? Looks to me too much like FUD.
  9. blasjw

    I don't know, sounds pretty convincing to me.  As far as "when does the overflow of the internal registers occur?", why don't you ask him?  Maybe he can explain it in glorious detail.  [​IMG]
  10. wahsmoh
    If that guy is correct then it seems like a slam dunk conclusion and correlates to the ears of many who have heard a bright/neutral headphone with the average $250-$1000 Sabre. That isn't always a great pairing, but take the darker sound of the LCD-2.2 and you have a good combination for many. 
    Also moving back to the people who were talking about the DT880/HD800 with their treble peak. Taking the DT880 and pairing it with the R-2R Theta does not remove the treble peak or make it much more resolving. The Theta has a noticeably more circular soundstage and image versus the AKM Bifrost Uber and treble that is natural. I think the tonality of the Bifrost Uber is pretty spot on, but the overall image/soundstage/phase correction is influenced more by the burrito-filter DSP chip.
    Oh yeah and of course in the dynamics/transients/bass/slam department the Theta walks all over the Uber
  11. Sapientiam
    Maybe he could but since I have other reasons (unrelated to his claims) for not using Sabre DACs, I can't see how it would benefit either of us for me to ask. I put the question not because I'm curious to know the answer (I've looked at too many plots of THD+N for Sabre to know it can't be happening on normal sinewaves) rather to show how his statement could be made more persuasive.
  12. evillamer
    As you know in any type of digit based conversion, currency or floating point to integer, there's always chance for rounding errors.
    In the Sigma Delta world, when the PCM 16bit 44Khz data is converted into 5bit 1MHz Bitstream(for example), won't it introduce some form of rounding errors as well? Especially if we don't know if they use double precision or single precision or etc. Also during the process of conversion, won't any kind of electrical interferences and jitter affect the precision?
  13. Sapientiam
    Converting from 16bits down to 5bits certainly generates lots of rounding errors - called 'quantisation noise'. Its the job of the S-D modulator to make those errors inaudible - by forcing them up to higher frequencies (above 20kHz). Single or double precision applies only to floats, but here its fixed point.
    If electrical interference or jitter affects the result of a truncation, the design's broken. Digital is able to ignore noise completely below a certain level.
  14. evillamer
    The only thing about ESS is the lack of detailed datasheets(not including those NDA ones) like those you find on TI or AD or even AKM dacs. Especially things like Monotonicity, PSSR, Digital Filter Graphs etc.
  15. evillamer
    Nothing's wrong with this carpet or room, it's very clean and nice. [​IMG]
    Wildcatsare1 likes this.
393 394 395 396 397 398 399 400 401 402
404 405 406 407 408 409 410 411 412 413

Share This Page