How is Chord Dave?

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by mtoc, Jun 9, 2016.

  1. Chris F

    Chris F Boyz 4 Now Fanatic - Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    805
    Trophy Points:
    93
    Location:
    Ottawa, Canada
    To my ears my TT setup (in profile) is clearly better then Yggdrasil and you hear it right away. Tone, texture, sound stage, "realism" are all significantly better. The digital has an edge in speed/slam and of course noise floor but the gap is not as big as the advantages of the TT over Yggdrasil in the other areas. I think the masters have a lot to do with it as well as I do hear a lot of these qualities carried over in my LP transfers.

    In the very short time I got to hear DAVE at the SBAF meet I didn't hear anything that suggested it was head and shoulders above Yggdrasil. Better maybe. Slam dunk winner.... I wasn't feeling that. Without question though it was quality and definitely good enough that it deserves a look if you are in that market. I would love to get my hands on one for a week or so to play around with.
     
    Last edited: Nov 9, 2016
  2. Negura

    Negura Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    180
    Trophy Points:
    43
    Location:
    London, UK
  3. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,771
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    Looking at the Chord block diagrams, it appears that the FPGA is used for implementing their WTA filter (upsampling / noise shaping / digital filter, etc.). Can you explain how an FPGA would provide for more room for upgrades or "almost anything you want" as opposed to a Sharc or similar processor - at least for this purpose - the digital filter? Keep in mind that the sigma-delta "Pulse Array" DACs are implemented separately after the FPGA on the Chord DACs.
     
    Last edited: Nov 9, 2016
  4. Chris F

    Chris F Boyz 4 Now Fanatic - Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    805
    Trophy Points:
    93
    Location:
    Ottawa, Canada
    IMO the FPGA is a cop out from designing an ASIC which would be orders of magnitude faster. Rob Watts could have his 16bit perfect million tap interpolation filter in an ASIC no problem. With FPGA it will be another 4 years assuming the speed of the FPGA doubles each year. (probably not)

    Unfortunately designing an ASIC takes big $$$ and an ASIC design team which Chord (I assume) does not have so FPGA is the next best thing.
     
  5. Valolilol

    Valolilol Acquaintance

    Contributor
    Joined:
    Oct 29, 2015
    Likes Received:
    44
    Trophy Points:
    28
    @Marvey, I did not went through enough details in order to know what the FPGA was used for. You obviously knows better than me the internal architecture. If the FPGA is used only for that purpose, I do agree with you, you can go for a SHARC processor or whatever suits your needs.
    However, what I wanted to point out is that, the day you find something interesting to do, a new feature which is not embedded by your processor, then you are fucked up because you will have some hard time optimizing it whereas a FPGA let you cable gates directly and implement a rather good and optimized feature. This is what I meant.
    I want to point out something else too now that I know a bit more what the FPGA is used for in their system. With a FPGA, you have the choice of the architecture and this is IMO very important. I would understand somebody telling me that he rather goes for a dedicated architecture where he cables directly the output of the noise shaping block to the digital filter without any bus than a multi-pipelines or RISC architecture which are using buses to transfer data. All processors have pros and cons due to their architecture, a FPGA gives you more room regarding that point.
    @Chris F Your point is really good, I think by now, knowing what they are doing for quite sometime, CHORD should go for an ASIC. I see no point staying on a fixed FPGA architecture where you can have a dedicated chip for that. But maybe they have some explanation, maybe they update this block often and we do not know, I cannot tell. However it is clear to me that they should go for an ASIC if the architecture did not changed for the last 5-7 years.
     
  6. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,417
    Trophy Points:
    113
    I think what someone is paying for when they buy DAVE is a couple man years of R&D. From system design, the software, boards, to the FPGA revisions, etc. Hell if I'm not mistaken, even the USB implementation in that thing is custom instead of off the shelf. And personally, I applaud the effort. However, it's far from a given that it will all result in a sonically superior product. It's one fairly impressive achievement, though. But, yeah, I am with you. I heard it with the DSHA-3, and while it was definitely apparent that it was an excellent DAC, I couldn't say if it tops the likes of Yggdrasil without doing a critical A/B. And that makes me believe it's either not better, or only marginally so... But I could be proven wrong.

    Regarding FPGA vs ASIC, I don't know the architecture of the device, but it's not immediately intuitive as to why clock frequency should be a limitation for the design? I also don't know which FPGA they are using, but state of the art devices are going be capable of 100s of billions of MACs per second? Hard to believe he's hitting a ceiling there. ASIC makes no sense. He could work with a fab partner and not need ASIC expertise but just the NRE alone would be in the millions given that his volume is going to be infinitesimal. And again, not obvious what it buys.
     
  7. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    You're not mistaken.

    Even the, very primitive looking display, on DAVE is handled with hand-written gate-level logic (which is why it looks so primitive).

    I think Rob's approach with DAVE is interesting - eliminating as many extraneous components as possible and implementing equivalents (or near equivalents) on the FPGA. I don't think it is all upside though (nothing says implementing, say, USB on an FPGA has to be better than a discrete processor in ANY measure, let alone the important ones), which is always how it gets portrayed whenever he starts discussing it. And the way he talks about USB, and some of the more recent statements he makes there, make me shake my head and laugh.

    Also, I think Chord are missing big opportunities with their use of the FPGA ... and I think it's deliberate, but that's a different issue.
     
  8. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,417
    Trophy Points:
    113
    Out of curiosity, point me to the laughable statements. I remember reading a bit about the USB approach, and thought it had merit at the time. I haven't seen anything recently. But I just don't think it'd be terribly difficult to design a better "USB-specifically-for-audio" board than most of what's out there already. I wanted to believe they did a good job at it.
     
  9. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I'll have to do a bit of searching for some of them, but his most recent is this (the quote links to the post it was in) - and the implications that come with it:

    "My experience with USB sounding hard or edgy is limited to when the source is not sending bit perfect data."

    While USB audio doesn't have error-correction mechanisms, it is trivial to show that you're getting bit-perfect data all the way from the sample data on the disk to the bits arriving at the DAC. And, though bit-errors can certainly occur via USB, on sensible runs of to-spec cables, with properly implemented transceivers, they're not really common enough to be a real concern at the level that statement implies.
     
  10. Hekeli

    Hekeli Facebook Friend

    Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    149
    Trophy Points:
    43
    Location:
    Finland
    Nothing from that clause makes me think he is referring to USB level packet errors with "bit perfect".

    Btw it's been said that Chord ASIO drivers actually retry bad packets.
     
  11. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    What does it make you think then?

    Bit-perfect generally means the original sample data in the source file is delivered as the correct values, in the correct sequence. And as such, the bits received are either what is sent, or they aren't. The causes for that are either issues at the USB implementation/operation level, or they result from something in the source mangling the data (accidentally or deliberately).

    I would say that anyone doing anything with USB, audio or otherwise, that cannot write code to reliably output the correct bit-level data from a computer or embedded board is, simply, completely incompetent. Chipsets to handle this at the electrical and other layers are readily available.

    That's not to say there can't be other audible issues with USB (e.g. electrical noise that gets transmitted and affects the analog electronics in the DAC), but that precisely bugger all to do with bit-perfectness unless the problems are so entirely gross that they're corrupting the voltages on the data lines.

    Chord's ASIO drivers are, by definition, NOT "USB Audio". USB Audio has a very specific set of definitions - and they are not the same as "some random protocol that happens to be implemented on top of a USB connection".

    I have a prototype NAI that uses USB Mass Storage class mode to pull the entire track to be played across a USB connection - that's not "USB Audio" either, but it also has full ECC/retry/retransmit.
     
  12. JWahl

    JWahl Acquaintance

    Joined:
    Oct 2, 2015
    Likes Received:
    63
    Trophy Points:
    18
    Just to clarify my previous comment, and expound on Marv's question. First, I am not an engineer (yet) but am returning to school for E.E. in January so I enjoy getting into the technical details. But anyone who is more knowledgeable, feel free to correct me if I'm falsely assuming something.

    I think the biggest advantage of FPGA in a design is size of the end circuit and simplicity of physical hardware. Some functions, like an SPDIF receiver can be integrated in firmware IIRC. The design advantage here is more obvious with the Mojo, but less so with the Dave.

    A potentially good future use which I haven't seen yet (though may exist), would be placing an SSD storage device close to the FPGA and interfacing directly through the FPGA.

    https://www.xilinx.com/products/intellectual-property/1-f2v39l.html

    The above is link is more targeted toward data center servers though, because of the high parallelism of FPGAs reducing CPU load on storage access.

    In my idea, the concept is reversed to avoid the CPU's and OS affects on data transport. it's similar to a music server, except it could be accessed and controlled remotely through a PC, but no audio data would flow through PC or operating system. Instead directly clocking with high quality clocks, directly into FPGA with short signal paths and little chance for signal degradation. With all the money some of us spend on decrapifying devices, this has the potential to theoretically deliver the most ideal digital transport. It also eliminates the cost of hardware related to a full music server DAC Combo. Conceptually, this isn't all that different from using relays to switch inputs at the rear of a case, without routing the signal to physical switches on the front of the case. In this application, it wouldn't be a true server, because it would lack it's own internal operating platform, it would really be more of a remote SSD transport controlled through the PC. Yes, you could probably do the same with RAM buffering in to the FPGA, but you might need some kind of parity/error correction to reduce bit errors in transport and introduce a slight delay.

    Honestly, I was kind of hesitant to share the idea, since I'm not sure if it's been done yet or not. But I figure by the time I'm actually knowledgable enough to implement something like this, it will likely have already been done, if it hasn't already.
     
  13. Hekeli

    Hekeli Facebook Friend

    Contributor
    Joined:
    Sep 28, 2015
    Likes Received:
    149
    Trophy Points:
    43
    Location:
    Finland
    Obviously resampling or other software bit mangling. Why do you care what I think anyway and why do you go rambling here from a single clause that can mean many things? Perhaps go ask Rob join here so he can explain things in more detail. He even expressed some dissatisfaction with Head-Fi moderating, so who knows.

    I don't see how this is in any way relevant to the ramblings here. I was simply making a point that things like that have been thought of, and why wouldn't they be..
     
  14. Madaboutaudio

    Madaboutaudio Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    545
    Trophy Points:
    93
    Location:
    Singapore
    I still think the best approach to ultimate sound quality is the direct feeding of audio data into the dac via an integrated SD-Card I2S transport like how Resonessence labs does with their Invicta series:
    http://www.resonessencelabs.com/sd-card/

    This totally bypasses all the (additional jitter) AES/SPDIF/USB format conversions and any electrically noisy computers/external transports.
     
    Last edited: Nov 10, 2016
  15. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I don't think that's obvious at all, as the context of the discussion was USB, as an interface, being hard or edgy.

    If the software/player was re-sampling the data or otherwise mangling it, it would, by definition, still be an issue on any other digital interface and, therefore, wouldn't just be a USB issue.

    You brought it up, not me.

    All I did was point out that it's not "USB Audio". And those drivers are only useful on Windows PCs. They do nothing for any other source, computer or otherwise.
     
  16. Psalmanazar

    Psalmanazar Most improved member; A+

    Pyrate Slaytanic Cliff Clavin
    Joined:
    Sep 27, 2015
    Likes Received:
    5,345
    Trophy Points:
    113
    Well many Delta-Sigma DACs are better in some areas than the Yggdrasil, mainly that perceived 7khzish tape hiss region that the Yggdrasil will throw in your face. If one noise floor of a tape is higher than another in a multitrack recording, the Yggdrasil will make it too loud rather than just have the noise floor be higher or different on that specific instrumental track. The AK4399 Gungnir is supposedly like this too (I've never heard it but have been told so by multiple people) but the Gungnir Multibit doesn't despite being a lot warmer than Yggdrasil.
     
  17. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I disagree. I think he brings up a great point. An SD-Card interface may offer some benefits over other transports and conversion interfaces.

    I do agree that this discussion (SD-Card interfaces) merits it's own thread though.

    We know @Madaboutaudio, I don't think many of us know you though. Did you introduce yourself?
     
    Last edited: Nov 12, 2016
  18. zerodeefex

    zerodeefex SBAF's Imelda Marcos

    Staff Member Pyrate BWC
    Joined:
    Sep 25, 2015
    Likes Received:
    14,051
    Trophy Points:
    113
    We know he or she set his location up as JP but is joining us from China or through a Chinese proxy. We also know that he or she wasn't born in 1900 despite what his or her profile says.
     
  19. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,461
    Trophy Points:
    113
    Location:
    Winnipeg
    I heard the Dave the other day, under meet conditions, and not from my front-end but at least with some familiar music. Brief impressions:
    - I don't hate the aesthetic, the port is kinda neat, but ye gods it needs a better display. Or hey, use some glass and mirrors to make a pseudo hologram, that would be bonkers neato and actually work well for the given rudimentary graphics.
    - surprisingly heavy
    - no coax input? seriously?
    - nice (not amazing) with the Utopia; didn't give it much body but had a very smooth(ed over) tone and I didn't notice the paperiness up top like I did last time
    - seemed good with HE1Kv2 (very brief time; I did not push it either)
    - better with Ether-C-Flow, or maybe just the closed nature helped block out noise from everyone else
    - just no with the HE-6, zero oomph, this is an unfair test though
    - staging is flat-ish and down low in front
    - resolution is... fine? no anomalies that I could notice under meet conditions, but nothing jumped out at me either (Yggdrasil gave me the jump feeling my first time; Dave did not)
    - I really wish I could have fed it into other amps but I was too busy hosting and chatting up the cute girl at the meet (priorities!)
     
  20. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Meh. Too much. Ban hammer then.

    Carry on folks.
     

Share This Page