Delta Sigma: An Inconvenient Truth

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by k4rstar, Jun 21, 2020.

  1. insidious meme

    insidious meme Ambivalent Kumquat

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,999
    Trophy Points:
    113
    Location:
    Sector 8023 of the Third Quadrant
    I'll just see myself out then.:p
     
  2. Dzerh

    Dzerh Friend

    Pyrate
    Joined:
    Apr 1, 2018
    Likes Received:
    425
    Trophy Points:
    63
    Location:
    San Ramon, CA, US
    In theory you can just unplug UPS from the grid during thunderstorms leaving the connected equipment on. Just need to mute UPS sound :)
     
  3. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,417
    Trophy Points:
    113
    What's inconvenient and true here, exactly? Not much other than the well-established notion that both DAC architectures have tradeoffs, and you pick your particular brand of vodka. Bricasti DACs (sigma delta) usually reside in one of my top 4 rooms at every audio show. They have a newish product which allows the user to dynamically switch between sigma delta and ladder conversion. The designer spent a fair amount of time allowing me to go back and forth between them last fall. This was pretty interesting (but not much more than that) to hear the difference between architectures under otherwise equal power delivery, dual mono isolation, etc design platforms. In talking with him, I got the impression he was simply trying to give users what they (think they) want because "all high end DACs are R2R!" (Bricasti products normally employ sigma delta chips). I think it's fair to generalize as objective truth that sigma delta converters offer the better price-performance ratio between the two on paper fairly universally. That's neither convenient nor inconvenient, and how it translates to user listening experience is of course not going to be universal.
     
  4. k4rstar

    k4rstar Britney fan club president

    Pyrate BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    6,942
    Trophy Points:
    113
    Location:
    Toronto, Canada
    i would be careful plugging in a $2,400 DAC into a UPS, surge protector, power conditioner, etc. in my opinion it really spoils the sound and paralyzes it. the Yggdrasil especially already has a choke-input, highly regulated power supply. additional AC filtering does more harm than good.
     
  5. Lyander

    Lyander Official SBAF Equitable Empathizer

    Pyrate Contributor
    Joined:
    Feb 25, 2017
    Likes Received:
    10,961
    Trophy Points:
    113
    Location:
    Philippines, The
    Edit @Cspirou: three including beemer, haha. Live performances I enjoy are more for different performers or rearrangements— studio recording of Comfortably Numb remains a desert island song but the Live in Gdansk closing solo is just transcendental.

    Anecdotal evidence: lent Bifrost MB A1 (granted, far from forum favourite) to a friend around Feb along with my main headphones, been without them since. Had micro iDSD back on loan from person I sold it to but when she asked for it back about a month ago I've only had laptop and phone to listen out of.

    I've been doing far less recreational listening, that is to say sitting back and genuinely immersing myself in music instead of just hearing things as part and parcel of other media, with each step "downwards". Phone and laptop are surprisingly passable to me now, but then again even AirPods can be so if they're all you listen to for extended stretches of time.

    In direct comparisons between the older Beefroast, the Chord Hugo v1, and the Micro iDSD (as a DAC), I always found myself preferring the tonality and sheer clarity of background of the iFi and sense of space and hyperclean resolution of the Chord, but something about the Schiit's presentation was, to borrow a snackage term, moreish. Greyer background and admittedly tubbier bass than either other two, but the sonic textures and nuance of spatial presentation (as distinct from just maximum shove-everything-far-away-from-"listening point"-headstage) was very much enticing. This is lower-tier shiz out of laptop USB and a basic surge protector/line filter or batteries where applicable, but regardless seems to validate the assumption that there are just tradeoffs between types and that one could easily substitute for the other depending on associated gear; lord knows I've got some sharpness to blunt.

    And hey, low-tier as it might be it was just about enough to discern differences between pubes on vinyl rips ;)
     
  6. beemerphile

    beemerphile Friend

    Pyrate Contributor
    Joined:
    Dec 19, 2015
    Likes Received:
    755
    Trophy Points:
    93
    Location:
    Danielsville, GA USA
    No, there are two of us.
     
  7. Jay

    Jay Facebook Friend

    Joined:
    Oct 2, 2015
    Likes Received:
    172
    Trophy Points:
    43
    Location:
    GA, USA
    FWIW, here's a copy of an email I had with Schiit a few years ago about the very subject of thunderstorms and UPSes:

    1) They shouldn’t but if you have unstable power it may be wise to power down the Gungnir

    2) A true sine wave should be fine, I would avoid anything else as they can have audible noise. The power sections should filter this but sometimes it is nearly unfilterable.
     
  8. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,778
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Well there's that. But also, there's stuff that can be easily confused either way. The Convert 2 could have fooled me, as well as many other AD-1955 based DACs such as the Lavry DA11 and Bricasti M2. If one associates the R2R sound to stuff like the UltraAnalog of yore, the AGD, or the Wadia PCM1704 stuff, then some of the Schiit and Holo R2 DACs could be a rude awakening for them.
     
    Last edited: Jun 22, 2020
  9. RobS

    RobS RobS? More like RobDiarrhea.

    Joined:
    Feb 21, 2019
    Likes Received:
    1,419
    Trophy Points:
    93
    Location:
    .
    Cross-posting from another thread, but k4 can you elaborate more on why discrete ladder DACs are so bad? I tried reading about the linearity issue but how it was described obfuscated more than made understandable. If you could shed more light on why they are bad I'd appreciate it. Do you have any recommendations on R2R DACs available for purchase now that aren't vintage? It's not that I'm against vintage gear, in theory I'd prefer them, but they might need servicing, no warranty, etc.

    I was looking at an old Arcam Delta Black Box that has that TDA1541 chip. It has separate power supplies for digital and analog, fully discrete class A output, and its an R2R. They sell pretty cheap like around $150, but seems it would be a huge step back (or multiple) from a Yggdrasil A2. I got laughed at by an engineer when I suggested I was looking at TDA1541 DACs but there are still dudes making new DACs with that chip.

    To get a meaningful, noticeable upgrade to another R2R from a Yggdrasil am I looking at paying $5k or more? My limit for DACs is around $2.5k, anything after that I'd rather go immediately into vinyl. FWIW I prioritize resolution, microdynamics, and detail.
     
  10. yotacowboy

    yotacowboy McRibs Kind of Guy

    Pyrate Contributor
    Joined:
    Feb 23, 2016
    Likes Received:
    10,696
    Trophy Points:
    113
    Location:
    NOVA
    Home Page:
    I'd propose a caveat: certain cheap brands of vodka don't have offensive after effects. But others give me a wicked f'ing hangover.
     
  11. Jay

    Jay Facebook Friend

    Joined:
    Oct 2, 2015
    Likes Received:
    172
    Trophy Points:
    43
    Location:
    GA, USA
    I hope no one minds that I take a stab at this.

    Let’s use a simple analogy to explain. Imagine you want a living room lamp with three levels of illumination. One approach is to select one with three different discrete lightbulbs. Let’s say they are incandescent. One 50 w, one 100 w, and one 150 w. Each will have slight differences in color warmth due to manufacturing variation in filaments and the white frost coat on the glass. Also, since they are discrete, each will maintain its own temperature. Lightbulb performance most definitely changes with temperature as they warm up. So going up and down your three lightbulbs will cause noticeable differences. Temperatures will change up and down based on use, and therefore so will performance.

    However, if you have a 3-way LED bulb with all three elements inside a single globe, not only will you have the same white coat on the glass, the internal temperatures will be more consistent.
     
  12. k4rstar

    k4rstar Britney fan club president

    Pyrate BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    6,942
    Trophy Points:
    113
    Location:
    Toronto, Canada
    i got a PM after posting this thread asking if i've tried the PS Audio DirectStream DAC. i'm happy to discuss gear in PMs - sometimes that's when the real opinions come out, but to give a proper answer i end up writing a lot and wish it was shared with more than just the recipient.

    before talking about the DirectStream DAC or any DAC that promises 1-bit DSD conversion inside of an FPGA, we need at least a rudimentary understanding of how D/S converters work. emphasis on rudimentary, laymans terms, etc. math was never my strong suite in school let alone quantization and i would be lying if i said i understood all the theory behind Delta-Sigma conversion. im about 34% of the way there.

    first let's establish when we're talking about digital audio and PCM we're always talking about approximations of the original signal. the original analog waveform captured by the ADC and encoded into PCM is forever lost, we only have the PCM encoding to work with at this point. i.e. what goes in is not 1:1 with what comes out. this is true unless you're one of those weirdo objectivists that believes ADCs and DACs can be perfectly transparent?

    the goal of the DAC chip, be it Wolfson, ESS, AKM, etc. is to convert an i2s digital data stream into an analogue audio signal which can be amplified. to do this samples are fed (the timing of which is controlled by a clock) into a delta-sigma modulator which analyzes the change (delta) in value between consecutive samples to generate something resembling an analog waveform.

    before this occurs, the signal is oversampled which is a means of using digital interpolation to increase the number of samples put through the modulator, for the same PCM encoding. in theory, this allows us to have more accurate conversion (really, more accurate 'guessing' of what the output bitstream should look like) because we have more samples to work with. this 'accuracy' is measured in SNR (noise) and ENOB

    after this occurs, despite our best efforts, we still have a bunch of noise left in the audio frequency band. at this point we need filters to push the noise out of the audio frequency band, and then chop off all information above a certain frequency. this has it's own effects on phase, pre-ringing, post-ringing, yadda yadda. you've already experienced the subjective effects of this if you've ever used a DAC with switchable filters.

    lastly, since we still have a really high sample rate we can't make use of in the analog domain, the digital filter is responsible for decimating or downsampling us back to a useable range. we now have a current signal, which we can convert to voltage in the analog domain through a variety of means and enjoy as music.

    all of this sounds awesome on paper, especially if you're a math geek. in practice though, this digital guessing game, upsampling, downsampling, filtering and molesting of the original PCM encoding is what is responsible for destroying the music. it's funny, there was some discussion in the turntable thread about how the workings of vinyl recording, pressing and playback are a wonder, but one rarely stops to consider what exactly is going on at the datastream level inside their DAC, and how that effects the playback of music.

    so how does this relate to the DirectStream DAC? basically Sony thought it would be cool to waste a bunch of R&D money to invent a method of super-giga-oversampling and patent it, that they could then use to create a monopoly over the recording and playback industry. of course, this failed spectacularly, and most savvy consumers know SACD/DSD recordings are a joke. it is only boomer audiophiles who fall for the DSD marketing promises. as an aside, Sony has always had a hard-on for oversampling, the earliest CD players designed and manufactured by Philips were not meant to have any oversampling, until Sony prodded them that 16-bit would sound a lot better than 14-bit in marketing literature and a 4x oversampling filter was added to the TDA1540 machines.

    DSD conversion is basically delta-sigma on steroids, for better or worse. the oversampling, noise-shaping and decimation is much more aggressive, all with the misguided notion that it will resemble a more accurate analog waveform. if you have ever listened to a DAC based on DSD conversion or even a DSD recording and though it sounded a little 'off', a little 'rounded' or unnaturally smooth, that's why. of course, a lot of people also like this sound and to them it's more high-end, so there's that.

    this image is good for marketing purposes only, but it shows what they were trying to do: https://www.cambridgeaudio.com/sites/default/files/paragraphs/image/infogram.png

    i have heard the DirectStream DAC more than once but never at home, so i can't recommend for it or against it. i will say there are several aspects of its design i really don't dig:

    1) input homogenization. i already talked about this with USB Gen 5 and Unison, but you can't make bad digital sources sound acceptable without putting a performance ceiling on the DAC's input section. for the DirectStream, this is done on all of the digital inputs using re-clocking and error correction schemes, as far as i can tell.

    2) transformer I/V conversion. remember how i was talking about that nice current signal and how we had a variety of options to transform it into a usable voltage one? some audiophool manufacturers thought it would be a great idea to use transformers for this application, because it has to be more transparent than an active device! wrong, so wrong. transistors are best suited to current mode amplification. using a passive I/V conversion via transformers or resistors may give you more detail in a 'hi-fi' sort of way but is definitely going to impact dynamics, rhythm and drive.

    3) firmware updates. apparently most owners of this DAC consider this a feature. really it's just the dude who designed it messing with the code for the digital filters to get better sound, which points to the volatile nature of this conversion method to begin with. don't get me wrong, it takes serious technical chops to even understand how this crap works let alone design a functioning product around it, but i am old-skool and don't want to update firmware on any audio gear. i have never experienced one of these firmware updates, i can't say if it's a difference of magnitude or a difference of kind.

    please forgive any errors made in the above technical explanations.
     
    Last edited: Jun 22, 2020
  13. k4rstar

    k4rstar Britney fan club president

    Pyrate BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    6,942
    Trophy Points:
    113
    Location:
    Toronto, Canada
    i was exaggerating when i said travesty, but let me try to explain.

    R2R DACs need a ladder of resistors to make conversion go. very tiny resistors in very tiny tolerances. if you are Philips in the 1980s and 1990s you have huge factories in Belgium and Taiwan pumping out these ICs for you with microscopic resistor networks that are trimmed by lasers, and QC processes to select the best chips for your top machines, while discarding the junk that doesn't pass muster.

    if you are a French dude making $20,000 DACs out of boutique audiophile naked film resistors and assembling the ladders by hand, you are just not going to match that level of precision. this is on the assembly front

    @Jay explained the temperature regulation issue well. resistors are temperature sensitive to perform within tolerances.

    no. spending $5k on a DAC is f'ing retarded. my advice if you are really curious, buy CDs, buy an old TDA1540, TDA1543 or TDA1541 CD player for cheap (there are like, a zillion models to choose from, i will write on the three i own in the CDP thread later) and f**k around with it. you can experiment by modding the I/V stage, bypassing oversampling, etc. this will give you a taste of what is possible in the path i have taken and decide if it's interesting to you or not
     
    Last edited: Jun 22, 2020
  14. bixby

    bixby Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    3,988
    Trophy Points:
    113
    Location:
    Northern Colorado
    And some live music can be about the sound. I have never heard air, space, or natural timbre reproduced via hifi like I do at a symphony. And then you have folks like Nils Frahm whose live show is so freakin' well done, no "stereo" is gonna come close!
     
  15. JoshMorr

    JoshMorr Friend

    Pyrate BWC
    Joined:
    Oct 4, 2015
    Likes Received:
    3,066
    Trophy Points:
    113
    Location:
    Portland, ME
    I see a real lack of actual descriptors in this thread. Inconvenient Truth was facts and science backing up what was a unpopular and unsexy topic. Just DS is bad and K4 likes R2R. Lots of theory and attributes of specific dacs, but I dont see any decent evidence of why we should lump R2R or DS dacs together on a subjective level. Ive heard great implementation of both and poor implementation of both. I'm looking for more than Stereophile vague terms like "musicality" "toe tapping" or to quote Rob "unsatisfactoriness". What specifically are we actually learning here?

    I know the strong opinions that people like k4 and Rob have been formed over years and several dozens of dacs (or 3 in Robs case), but why should we discount your previous statements and believe these ones? What have you learned this time that trumps all your previous writing and thoughts?

    Ps- im here for the debate, actually interesting conversation should come of this.
     
    Last edited: Jun 22, 2020
  16. Dzerh

    Dzerh Friend

    Pyrate
    Joined:
    Apr 1, 2018
    Likes Received:
    425
    Trophy Points:
    63
    Location:
    San Ramon, CA, US
  17. RedFuneral

    RedFuneral Facebook Friend

    Joined:
    Jun 8, 2016
    Likes Received:
    213
    Trophy Points:
    43
    Location:
    CT, USA
    I think you misunderstood my point(aka I was ranty and didn't word it well.) I think budget D/S as a category tend to be crap as they chase graph neutral sound(objective focus) & cut costs on the inputs. They don't separate enough from motherboard/smartphone audio & don't have well enough engineered inputs to cut off all the noise from the PC(which I feel is the most basic thing a DAC should accomplish.) Budger R2R makes more sense to me because its not competing with the motherboard or smartphone DACs, it's trying to accomplish something different. Neutral focused S/D I expect to be audibly superior to free or why even bother?

    I've yet to hear anything S/D under $400 that's worth keeping around. I've heard 4 S/D DACs in that category over the last decade: Schiit Bifrost 4399, Ressonessence Labs Herus, JDS Labs OL DAC, SMSL M300 MKII. All of them failed your 'listener curiosity' test, I found myself gravitating more toward entertainment videos & conference calls spending less time listening to music even compared to my phone/wireless headphones. I found exactly the opposite with budget R2R: Ultra-Fi DAC41, Metrum Flint, Soekris 1321. With those I was bouncing around all kinds of music to see how it sounded with the new equipment, these all did something(and had dead silent inputs!) which made me curious to hear them with as much music as I could find.

    As for premium S/D I had the same rush to hear as much music as possible through the Questyle 192D & Exogal Comet+ hence I liken them more to the R2R DACs I heard in listener curiosity. Worth noting my curiosity & long-term enjoyment of DACs is not connected. I would rather have one of the budget R2Rs over any of the S/D I've owned although I'm always hoping to land a DAC that has the best qualities of both implementations.
     
  18. Gazny

    Gazny MOT: ETA Audio

    Pyrate Contributor
    Joined:
    May 11, 2020
    Likes Received:
    2,209
    Trophy Points:
    93
    Location:
    open sky
    Some of this thought I believe is the driving for for the Burr Brown chips.
    https://www.ti.com/product/DSD1793
    [​IMG]
    • Top 6 bits: true multi-bit;
    • Lower bits: Delta-Sigma.
    Not sure what kind of nasties have to be solved when both are used. For my uneducated self it seems more complex than it usually is. I know a few forum members have the gear. Just wondering where this type of configuration excels.

    Any impressions of the middle ground?
    [​IMG]
     
  19. k4rstar

    k4rstar Britney fan club president

    Pyrate BWC
    Joined:
    Jun 11, 2016
    Likes Received:
    6,942
    Trophy Points:
    113
    Location:
    Toronto, Canada
    Josh i know you're breaking my balls here, but i'm not sure what kind of descriptors you're looking for. i gave up a long time ago trying to describe sound with typical audiophile lingo, and to do so is really difficult without sounding cliche or vague.

    IDK about Rob, but for me i was on record as saying D/S drools almost a year ago, and that was my final encounter with a D/S DAC https://www.superbestaudiofriends.o...teddypardo-teddydac-wolfson-masterpiece.8202/

    there are a lot of mediocre R2R DACs out there, stuff that the Convert2 will just go full Mike Tyson Punch Out! on, but i think even most of the mediocre ones do things the best D/S ones don't. in my time with D/S DACs, i was less likely to listen to music for extended periods of time or even sit through full album. i was less likely to try new musical genres or styles. i was more likely to stick to a smaller pool of songs i would play repeatedly trying to hear what i wanted but didn't understand why it was missing. as a last.fm user, i actually have some data to back these experiences up, coinciding with the ownership of certain DACs.

    again, i could put what that is in words but for some people it would sound cliche and vague no matter what; so i prefer to use ideas such as emotional connection, changes in listening habits and desire to explore new avenues of musical taste as a yardstick instead. the lengthy technical explanations only serve to theorize the cause of this phenomenon. :)

    i totally understand where you're coming from and feel the same. actually, my (mis)adventures with D/S DACs came out of frustration of trying to really like the A2 Yggdrasil and Gungnir, but failing. the Yggdrasil is actually the only DAC i bought and sold twice!

    my personal thoughts on the big boy Schiit DACs are that the approach has a unique sound no other R2R DAC quite has. depending on which way you look at it, you could argue the mega-combo-burrito filter manipulates phase information and creates a false sense of depth with stereo recordings. i thought i liked this, but then i heard other R2R DACs and decided it was weird on most recordings.

    further, Schiit is pretty limited on what they can do with the single-ended output stage. the AD5791BRUZ chips are used in the Yggdrasil in a differential configuration and are already handle I/V conversion internally. to make a significantly better output section for SE users would certainly mean increasing the cost of the DAC. i see now people are tending to emphasize in their gear impressions whether they used the SE or XLR outs of the DAC.

    the Bifrost 2 sounds a lot more normal through its SE output because Schiit started using next-gen op-amps in the output stage. they could do this with the Gungnir and Yggdrasil too, but then the ORFAS wouldn't buy them.
     
  20. YMO

    YMO Chief Fun Officer

    Pyrate Contributor
    Joined:
    Apr 1, 2018
    Likes Received:
    10,517
    Trophy Points:
    113
    Location:
    Palms Of The Coasts, FL
    The Inconvenient Truth is a chip is a chip and more of what makes a DAC a DAC is what they do around the chip. I get on with the WM8741 DACs from a few years ago (Audio-GD stuff, Cyrus CD6 I believe, and one more DAC I don't remember) to the AD chips on the Schitts. Honestly, almost all.of them do 24/192 at this point. Besides the chip, I'm curious on the Digital Inputs and power supply and blah blah blah.

    @k4rstar another Inconvenient Truth is the overwhelming of users aren't going to brother modding TDA DAC based CD players since more and more of us are on the streaming/digital store bandwagon at this point. Is there a TDA DAC SPDIF input unit out here? People still hunt down the Sonic Frontiers DAC for its sound signature, plus the DAC has digital input. The CD player per users here are also good but spinning CD is becoming less popular now. I love my music collection in hard drive plus Qobuz.
     

Share This Page