Rockna Wavedream DAC: Awesomesauce

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by purr1n, Nov 7, 2019.

  1. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Running Freya+ in the diff-amp "buffer" mode (I wish the Freya S implemented the diff-amp buffer as well as the Nexus). I prefer a single Aegir on this system vs dual-mono, so I needed a way to sum the balanced outputs of the Wavedream and Yggdrasil (as a known reference).

    I would seem that we have arrived at eerily similar conclusions of the Wavedream independently.

    The mellowless is hard to describe. I provided it more as a descriptor of tendencies for appropriate component matching, i.e. this probably isn't a good DAC to pair with laid-back sounding gear, e.g. Cavalli, rather than as a objectionable characteristic. And again, the big JBLs are on the mellower side. It might be interesting, perhaps more ideal, to pair the Wavedream with the big horn speaker.

    I'd say the mellowness is a function of transient behavior, deep soundstage, fluid presentation, and tonal balance (subjectively speaking). If I have heard anything like this before, I'd say it was on a custom x4 2A3 amp that Craig made using Tango amorphous interstage and output transformers.

    As far as many discrete R2R DACs sounding mellow, I would say that I have heard poor implementations of discrete R2R DAC sounding rather dull or flat as opposed to mellow. Not the same thing here.
     
    Last edited: Feb 24, 2020
  2. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    LOL. Surprise surprise. An $8k list DAC that actually sounds better, not worse, than $2k DACs. There so much garbage out there. And again, I really like what Rockna did here. Smart unique non-lazy actually engineered from scratch solutions. Not dumb dumb stuff like mega bank of 100 small caps (but hey let's skip the choke), or unnecessarily large power transformers more suitable for power amps just to add weight, or resurrecting "vintage" low-end POS R2R chips.
     
    Last edited: Nov 8, 2019
  3. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,287
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    OK, if mellow =/= dull or flat in this case, I get where you are coming from. I too am thinking more of the dull/flat characteristic in this case.
     
  4. monacelli

    monacelli Friend

    Pyrate
    Joined:
    Jul 23, 2017
    Likes Received:
    1,649
    Trophy Points:
    93
    Location:
    The Land of Enchantment
    If this is a silly question, feel free to say so. Does the Rockna sound DAVE-like at all? What are your thoughts on FPGA-based DAC implementations generally (understandable if it's not possible to generalize)? I seem to recall you not having a super favorable impression of Hugo 2, or really any Chord gear below DAVE. Wondering what these guys did to get so much fluidity out of an FPGA-based implementation, especially with a relatively modest number of filter taps (4820 per Rockna website).
     
  5. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I'm not sure if there is an "FPGA sound".

    It depends how it is utilized. The Wavedream FPGA is used specifically for UI logic, digital receiver, digital filter, and a few other things. The actual D to A component of the Wavedream is a discrete resister ladder. An FPGA inherently does absolutely nothing until it is programmed to do what we want it to do. Heck, we can use specialized ASICs, Atom chips, ARM running Linux, or even Motorola CPUs to do some of this number crunching or logic.

    As for number of taps, that's just bullshit with the way CHORD is marketing it. At some point, there are diminishing returns. I would surmise that the liquid sound of the Wavedream is more likely a matter of Rockna having conquered the problem of glitch with ladder architectures. Glitch is like the when you turn the vacuum cleaner or hair dryer on, and the lights dim and then flicker a little. Each time a bit is flipped on ladder DAC, there's a glitch. More bits changed at once, more glitch. This glitch can actually be mapped from one 16-bit (or 24-bit) word to another. But honestly, I don't know Rockna what did. The digital filter they used is very standard though - Parks-McClellan - but ultimately it's a matter of picking the filter parameters and executing the design.

    I mean it's obvious that the guys at Rockna had vision in mind, and through successive development has reached that vision or close to it. The CHORD stuff, with the 10000000 taps. It all fricking sounds different. The Qutest, DAVE, Mojo, Hugo - they all sound different. So I'm not sure what's up with the 10000000 taps spiel. It's just product differentiation marketing from a bunch of wankers taking advantage of dumb audiophiles thinking moar is always better.
     
    Last edited: Nov 8, 2019
  6. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Going forward, I will also be posting measurements in the subjective discussion subforum here instead of separately. I will provide a link in the measurements subforum to the measurements here. Unless I purely review or give impressions of gear based on measurements, I believe that both measurements and subjective impressions belong in the same place, with the understanding that simple measurements once they get good enough to a certain point, don't probably won't be able to tell us very much.

    See here: https://www.superbestaudiofriends.o...ically-incorrect-audio-blog.7261/#post-249242

    Due to lack of time, just did these very quickly. Output is SPDIF from the AverLAB. Linear filter. No dithering. 24-bit data.

    1kHz 0dbFS
    FFT 1k 0dbFS.png

    1kHz -30dbFS (normalized on plo y-axis at 0db)
    FFT 1k -30dbFS.png

    1kHz -60dbFS (normalized on plo y-axis at 0db)
    FFT 1k -60dbFS.png

    1kHz -90dbFS
    FFT 1k -90dbFS.png

    While SNIAD or THD+N for the full scale 0dbFS signal is not much better than 85db - the main culprit is the 3rd order harmonic distortion product - the THD+N performance actually improves as the signal gets smaller. For a -90db signal (last remaining one bit toggle if source material was 16-bits), we are still getting a lot of room, about 27-28db, before we hit any harmonics or AC mains noise.

    From a purely scientific point of review, we are unlikely to hear any of the measured nonlinearities. Distortion 85db below full scale would either not be audible at extremely loud listening levels (rock concerts) or at normal listening levels where SPL would not be high enough over ambient. And one is unlikely (it probably would not be possible) to listen at volumes where the -90dbFS tone is at 40dbSPL, because that would be the 0dFS max would have to be at 130dB. And even if we did, the distortion products of the -90dbFS /@ 40dbSPL would still be far below ambient at 13dbSPL, which is much lower than any ambient environment, certainly much lower than a quit home at night.

    The verdict: The measurements are not shit. While the nonlinearities are not superb beyond the scope of the most advanced instrumentation available today, they are beyond human hearing.
     
  7. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Here is something that should be interesting. I'm glad this DAC came with a dither on/off option. Without going too deep into it, dither is a way of smoothing out certain waveform patterns (digital quantization error) by injecting noise. In a way, it's an effective way to "cheat" on measurements.

    Some DAC manufacturers elect not to implement dither. I believe that Schiit's early R2R-2R DACs did not implement dither. I recall a conversation with Mike Moffat when we just asked me for my 2 cents, and just I replied: don't bother. For 24-bit signals, it doesn't matter. For 16-bit signals, why muck (add noise) to the least significant bit?

    Anyway, below are two measurements of the 90dbFS 1kHz signal. One is with dither. The other is without. Guess which is which?

    FFT 1k -90dbFS.png

    FFT 1k -90dbFS dither.png

    A good argument can be made that DAC manufacturers implement dither because the measurements look better.
     
  8. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Here are some THD+N plots per frequency, at 0, -30, and -60dbFS.

    We aren't seeing any particular frequencies where the distortion is worse than across the board. The sudden drop in distortion past 7kHz may be that the measurement is bandwidth limited (thus not including the higher order distortions). The distortion is similar across the different levels.

    0dbFS
    THDN 0dbFS.png

    -30dbFS
    THDN -30dbFS.png

    -60dbFS
    THDN -60dbFS.png
     
  9. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Here is THD+N for 1kHz from -120dbFS to 0dbFS

    SINAD (1kHz 0dbFS) is about 80db. What is interesting is that THD+N actually improves as we go down in level to about 90dbFS for a -4dbFS signal. Hahaha, maybe the effective SINAD is better if we listen edto quieter better mastered classic recordings?

    Just kidding. SINAD, as long as it's "good enough" (again, read my blog post), seems to be one of the most utterly useless measurement numbers when it comes to human subjective experience in regards to DACs. At least from the many level-matched blind listening tests which have been conducted by SBAF members.

    SINAD.png
     
  10. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,842
    Trophy Points:
    113
    Location:
    East Coast
    I'm having a little hard time in understanding odd harmonics -- not limited to this specific product.

    For amps, even > odd harmonics generally feel better (at least within my auditions/observations).
    And for dacs, I still strongly believe decent amount of harmonics sound far superior to near zero harmonics (concurrent chord products showing this character). But some products I really enjoyed (e.g., two ES9038 dacs) had odd > even harmonics -- not as strong as rockna though. I like Schiit multibit dacs (even > odd harmonics) over them but that's too much apple vs orange.

    For this uber dac, strong D3 and D5 played a rather beneficial role? Do you have any thoughts?

    On the dithering front, my guess is that the bottom one is dithered for two reasons: (1) noise level subtly raised; (2) seemingly higher (measured) dynamic range.

    Perception of dithering effects could be controversial. But I always prefer Schiit's recent revisions over early one. This applies to the whole products: Yggdrasil, Gungnir MB, Bifrost MB, and Modi MB. So it's safe to say that dithering DON'T shit. But I really don't know whether I prefer dithering if I can have an option to turn it on and off.
     
  11. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Lots of R2R DACs have odd harmonics, particularly at higher levels: Soekris, Denafrips, RDAC, etc. The general observation that even order is more pleasant than odd order harmonics does seem to get thrown out the window with R2R DACs.

    I also used the balanced outputs, which can suppress even harmonic distortion.
     
  12. Clemmaster

    Clemmaster Friend

    Pyrate Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    3,271
    Trophy Points:
    113
    Dithering is not just to make measurements look pretty. It has proven audible benefits - that for once happen to be backed by measurements ;). There's no reason not to implement/use it.
     
  13. Vtory

    Vtory Audiophile™

    Pyrate MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    10,842
    Trophy Points:
    113
    Location:
    East Coast
    Just curious. Could you let me know how the benefits were proven audible?

    In my understanding, improved dynamic ranges and bit resolution occur far beyond human threshold. If bits are extremely restricted, it's possible to have tangible gain -- I've seen a 8 bit image reduced to 1 bit image and how dithering really helped visibly. But are't 16 bit audio already too resolving?

    EDIT: and all above suppose the situation such as down-converted 16 bit playback from 24 bit files. For 16 bit finished files, and playing them with dithering.. my bet is at best no harm (occurring at too small levels). Can't think of any 'gain' in the latter.
     
    Last edited: Nov 21, 2019
  14. iFi audio

    iFi audio MOT iFi Audio

    Joined:
    Oct 3, 2017
    Likes Received:
    1,284
    Trophy Points:
    113
    Location:
    Many places.
    Home Page:
    Exactly. It's just a measure/tool to achieve an unspecified engineering goal you mentioned. No such thing as FPGA sound.
     
  15. ColtMrFire

    ColtMrFire Writes better fan fics than you

    Pyrate
    Joined:
    Jun 16, 2016
    Likes Received:
    6,229
    Trophy Points:
    113
    Location:
    Austin, TX
    So I'm at @famish99 place right now with a few glasses of rum in me listening to this DAC through my Starlett while the guys are out getting food and... I'm really quite taken with it.

    This thing is serious biznezz.

    More later.

    EDITED for boozy influence.
     
    Last edited: Jan 24, 2020
  16. Saleh84

    Saleh84 New

    Joined:
    Feb 1, 2020
    Likes Received:
    39
    Trophy Points:
    13
    Location:
    Dubai, UAE
    After reading about the rockna for weeks all over including a Russian forum using Google translate, I bit the bullet and placed an order for a signature version XLR to replace my Denafrips Terminator. I believe terminator sounds good for its price. But I'm craving more.

    From reading, this model witnessed hardware and software changes since its release. I was thinking its a dac that was released in 2014, and its 2020. Apparently i was wrong. Its even a more mature product in 2020, and hopefully will keep recieving updates to improve its performance.

    Recently a friend of mine bought an Aries Cerat kassandra ref2 dac. That 60kg beast is one of the best sounding dacs I have ever heard, but i haven't heard other dacs at this level. So looking forward to compare my rockna sig to this hulk of a dac when it arrives!

    Lets hope this month of wait pass quick))
     
    Last edited: Feb 1, 2020
  17. famish99

    famish99 Friend

    Pyrate
    Joined:
    Dec 26, 2017
    Likes Received:
    1,714
    Trophy Points:
    93
    Location:
    Austin, TX
    I withheld posting my thoughts on this largely because I wasn't sure if I was going to keep this piece of "audio jewelry" and I have a few judgmental and gossipy associates... but the noisepost above has inspired me to add some what I can only hope is meaningful impressions.

    tl;dr: Wow, but you may find out your other gear is shit.

    I have the Wavedream Edition Balanced (same version as @Thenewerguy009 ). Setup is JRiver/Roon > Rednet 3 (AES) > Wavedream/Gungnir Multibit A1/Bifrost 2 (BF2 courtesy @Joshvar ) > Af/J2 (J2 on loan)> Klipsch Forte 3 (using my modded 6xx and 800S once each). Most of my evaluation was spent with linear filter, dither off, local clock, but I did spend a good amount of time with dither on, and stream clock as well, and a small amount of time with the hybrid filter and NOS mode.

    Pinning how this DAC sounds was a bit of a head scratcher because on the surface it has the tonality of nicer DS DACs, like the Prism Atlas, but less warm yet not lean and with the multibit magic in the sustain/decay/microdynamics. As I started digging deeper, I struggled a bit with determining if some things I was hearing were a fault of the DAC or it was revealing faults in the rest of my setup. With the help of @Joshvar (big thanks bud!), we were able to find that the Af was significantly softening the attacks and causing a somewhat plastic timbre. However with the J2, the plastic timbre completely disappeared. I believe this would be my interpretation of what @purr1n was saying when he described the Wavedream as mellow. Especially side by side with the Schiit, the treble presentation on the Wavedream is far more relaxed than the aggressive presentation of the BF2. Having said all this, this is the hardest hitting MB DAC I've heard yet (Gungnir Multibit, Soekris 1321, Pavane).

    I largely agree with Marv's TV analogy regarding the dynamics, the BF2 really has an elevated dynamic floor compared to the Wavedream, which gives the BF2 this somewhat exaggerated plankton presentation but the details are all there on the Wavedream as well, usually presented closer to the black level (so to speak).

    Staging on the Wavedream is by far my favorite feature of the DAC, it has a row 25-30 presentation with exceptional width (going beyond the speakers), depth and natural placement. The BF2 by comparison tended to sound localized within the speakers, also with great placement, but the sounds off center tend to call more attention to themselves than they did on the Wavedream. The Wavedream never gets congested and you never lose the ability to pick out instruments like I often experienced on the Gungnir Multibit A1.

    Misc notes: Wavedream also sounds pretty good from cold, but I largely left it on out of laziness. Digital volume works well, I don't notice any dynamic loss with it connected to the J2 at lower volumes (but it could also be due to the Fortes excelling at low volumes as well).

    Would I say it's perfect? Not really, the remote leaves a lot to be desired in terms of switching inputs and since the display is OLED I preferred to leave it off, but it doesn't automatically turn back on when you try to change inputs/volume. Occasionally, I sometimes miss how warm and gooey the Gungnir Multibit A1 sounds with guitars, but I feel this DAC has the fewest compromises in the areas I care about most.
     
    Last edited: Feb 2, 2020
  18. roroh

    roroh New

    Joined:
    Feb 12, 2020
    Likes Received:
    6
    Trophy Points:
    3
    Location:
    Créteil
    [QUOTE = "Saleh84, post: 288492, membre: 8390"] Après avoir lu sur le rockna pendant des semaines, y compris un forum russe utilisant Google translate, j'ai mordu la balle et j'ai passé une commande pour une version signature XLR pour remplacer mon Denafrips Terminator . Je crois que le terminateur sonne bien pour son prix. Mais j'ai envie de plus.

    À la lecture, ce modèle a vu des changements matériels et logiciels depuis sa sortie. Je pensais que c'était un dac sorti en 2014 et son 2020. Apparemment j'avais tort. C'est un produit encore plus mature en 2020 et, espérons-le, continuera à recevoir des mises à jour pour améliorer ses performances.

    Récemment, un de mes amis a acheté un dass Ceris kassandra ref2. Cette bête de 60 kg est l'un des meilleurs DAC qui sonnent que j'aie jamais entendu, mais je n'ai pas entendu d'autres DAC à ce niveau. Alors j'ai hâte de comparer mon rockna sig à cette carcasse de dac quand il arrivera!

    Espérons que ce mois d'attente passe vite)) [/ DEVIS]
    salut

    You talk about the Rockna édition version or signature ??
    Thanks
     
  19. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,912
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    En anglais s'il vous plait!
     
  20. elmoe

    elmoe Friend

    Pyrate Banned
    Joined:
    Apr 23, 2019
    Likes Received:
    956
    Trophy Points:
    93
    Location:
    Phoenix, AZ
    Haha man I think he google translated the other rando's post to French, quoted it in French, but replied in English. Typical frenchman.

    Even the closing /QUOTE got translated into DEVIS (estimate)
     

Share This Page