Focusrite RedNet (AOIP Nervosa)

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by 3X0, Oct 12, 2016.

  1. 3X0

    3X0 Friend

    Pyrate
    Joined:
    Jan 31, 2016
    Likes Received:
    563
    Trophy Points:
    93
    There's been much hullabaloo about the RedNet 3/D16 AOIP offerings recently and I figured I'd start a thread since I have one on the way (as well as there being considerable noise in the HF thread).

    I plan on updating this thread with some impressions on both setup on Windows 10 and OSX as well as subjective differences vs. USB.

    Recent (mini-)meet impressions have been very favorable in certain systems (RKML0007's Utopia system). I also believe some folks like Clemmaster and Luckbad have compared the RedNets to Lynx PCIE->AES offerings and ended up preferring the former.

    Besides escaping USB drudgery by going from Ethernet to coax/AES, it seems like the RedNets have some pretty neat features like JetPLL jitter reduction and well-designed power supplies (not sure if this applies to the 3, D16 or both).

    My main curiosities are:
    1. how these devices compare against pro sound cards with AES output in the typical computer system -- the only comparisons I can find seem to prefer the RedNets
    2. incremental difference from RedNet to RedNet + SPDIF reclocking (e.g. MC3+USB; not something I can speak to)
    3. Dante/network debugging (dropouts seem overblown, but can't be too sure)
    4. how much we hate how big/red/ugly these are
     
    Last edited: Oct 12, 2016
  2. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I won't say anything much about my findings in terms of sound quality using the RedNet 3 here (partly to avoid adding to any potential bias, mostly to save it for when I finally get around to posting my comparison of various AOIP/NAI devices).

    I will say that the RedNet 3, via AES, is one of the top three best interfaces (per above, you'll have to wait to see specifically which position it occupies) to my Yggdrasil. Though really I should say "top two", as one of the units in the top three really isn't practical or sane just to use as an AOIP/NAI.

    One thing I found consistently is that Yggdrasil performs its best for me via it's AES input. That has held true with every device I've hooked up to it that actually has that option. So far, whether it's been part of this test, or my other DAC evaluations, it's been AES > TOSLINK = COAX (no discernible difference between RCA or BNC) > USB.

    For those with the patience to get the thing setup, it's a pretty nice system. There are, however, much more approachable, if not necessarily less-expensive, ways to get fundamentally similar results. I didn't run into network issues or dropouts (my home network is a bit more robust than most would likely require, and isn't the typical mishmash of consumer-grade routers and switches) but there are lots of little "gotchas" that you'll no doubt have some fun working through (obviously several of us here can help there). If you like to fiddle with things, it's certainly an opportunity to do that!

    One big one "gotcha", which hopefully you're already aware of, is the studio-standards observed by its optical (ADAT rather than S/PDIF) and AES outputs. You can run a conventional cable for coax S/PDIF, but because the thing has support for so many channels the AES output is via a DB25 connector. So ideally you either ordered a suitable snake (which can be a pain in the arse, since they tend to come with multiple connections on the XLR end) or are planning to make up a cable yourself (which is what I, and I know @johnjen did).

    No comment, for now, on how it worked out for me with the MC3+USB re-clocker, or how that and/or the RedNet 3 itself, behave when given an external word clock, but I'll cover that in my eventual write-up (or at least write up something on dedicated word clocks to go along with it).

    It is very red and while not ugly on the level of a lot of pro-gear is still not something I'd want on display. Fortunately there's no particular issue running longer cables from the thing and hiding it, especially with an AES connection.

    Hope you enjoy it!
     
  3. Madaboutaudio

    Madaboutaudio Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    545
    Trophy Points:
    93
    Location:
    Singapore
    Does the rednet draw a lot of heat?
    Still thinking between this or the lynx/Rme
     
  4. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    No, not really. It runs cool.
     
  5. Madaboutaudio

    Madaboutaudio Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    545
    Trophy Points:
    93
    Location:
    Singapore
  6. 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
    Let's not go there, please. Next thing you know, we're talking about Ethernet cables and high end Juniper switches.

    I'm looking to hear from @Torq's sane structured approach and qualified (compared to other approaches) observations. One thing I can agree on is that the Yggdrasil's AES input sounds the best, BNC/Coax sound the same, and USB comes in last place.
     
    Last edited: Oct 12, 2016
  7. Azteca

    Azteca Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    1,581
    Trophy Points:
    93
    Location:
    Cincinnati, Ohio
    Unless you are running a bunch of intense media servers (like for editing 4K video from NAS), I can't imagine you needing anything like that. Listening to audio, even 24/192 FLAC, requires a comically small amount of bandwidth.
     
  8. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    No need for anything like that, it'll work on a 100 mbit network just fine (something I tested, though not how I actually use it). A faster network configuration is beneficial if you're going to run lots of channels, e.g. in a studio setting, but otherwise you don't need anything unusual or fancy.

    You will want to make sure your switch has proper QoS support if it's a 100 mbit device though, especially if it's shared with anything else. With a gigabit switch, even QoS isn't necessary, and perfectly good gigabit switches are readily available starting at $20 or so.

    As @Marvey says, I wouldn't go down the path of getting all tweaky when it comes to the network/cabling aspects of AOIP/NAI devices. And there are very solid reasons for this. Worrying about the clock over the network or timing is absolutely pointless as the network and protocols in use don't even guarantee in-order packet delivery. In other words, unlike USB or S/PDIF, blocks of sample data can (and often will) arrive out of sequence and it'll get put back together correctly at the receiver.

    Yes, really!

    You no longer have to worry about data arriving correctly either, since it's validated and can support re-transmit (within limits managed via latency settings). So unless your source is actually mangling data (i.e. your player/server is actually changing the bits sent vs. what's in the source file) the need or concern over fancy music servers and low noise PSUs on the computer end becomes entirely obsolete.

    So your nervosa focus becomes purely on what happens once the data hits the NAI.
     
  9. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    I should add/clarify ... worrying about what's going on on the computer end, given a bit-perfect player, with NAI devices is pointless.

    This is entirely different to using a computer source and a direct electrical connection between the computer and DAC (e.g. USB or an electrical S/PDIF or AES output via a sound-card or other PCIe device). With a direct connection, potentially every activity and device in, and occurring on, the computer affects load on the PSU and the electrical noise profile of the components in the machine ... and that can potentially cause issues with noise getting into the DAC which can have all kinds of interesting effects.

    Such things are not of concern with a NAI ... as it is completely isolated from the source computer with multiple levels of galvanic isolation, and completely independent timing mechanisms.
     
  10. Thad E Ginathom

    Thad E Ginathom Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    14,135
    Trophy Points:
    113
    Location:
    India
    Yeah. This is the way to go. Given a choice of the various wires that hang off a computer, ethernet is the one I back for the future of audio. In what time frame, I don't know, but my crystal ball shows all sorts of other stuff in the bin!

    Sadly, it also shows people inventing obsessive paranoid nervosa over whatever protocol/physical turn out to be the future of audio transport, so nobody need rush to sell their shares in the cable/etc companies.
     
  11. 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
    That would be funny if different firmware on that Juniper Internet traffic grade switch made the RedNet sound different.
     
  12. Kattefjaes

    Kattefjaes Mostly Harmless

    Pyrate
    Joined:
    Sep 5, 2016
    Likes Received:
    4,521
    Trophy Points:
    113
    Location:
    London, UK
    Having personally been involved with sample-accurate audio over IP, roundtripped between cities hundreds of miles apart (tested by nulling out), I'd say "ethernet" is underselling it a bit :)
     
    Last edited: Oct 12, 2016
  13. Torq

    Torq MOT: Headphone.com

    Pyrate
    Joined:
    Jan 15, 2016
    Likes Received:
    8,193
    Trophy Points:
    113
    Location:
    Seattle, WA
    You just know someone is going to go there ...

    And the "explanation" for it will undoubtedly be that the different firmwares cause different levels of hardware utilization in the switch, which will affect it's noise profile, and that'll get onto the network cable ... and into the NAI ... and ... so on ...

    Of course, even if that were a concern, unless the NAI is spectacularly poorly and incompetently engineered there's no way any of that noise is getting passed onto the digital outputs. Not that this is going to stop someone sticking an Ethernet Data Isolator (yep, they're real, and have a valid purpose ... but it's not THIS purpose) ahead of the NAI though and claiming major improvements. And then it'll all go down the rabbit hole, because that'll undoubtedly need to be improved with an LPS and so on ... and then there'll be a rash of audiophile-targeted isolators ...

    *Just idly pontificating about such things has me thinking I should finally get around to starting an audio-centric company to supply "solutions" to such issues, as for once I'd be "in at the start" as AOIP/NAI gets easier for normal consumers (or as normal as us audiophiles get, I guess). Generally I prefer not to take advantage of people, but at this point in life, if they're going to throw their money at such things anyway, I figure I might as well get some of it ...

    (*Even I'm not sure if I'm joking about this ... )
     
  14. Madaboutaudio

    Madaboutaudio Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    545
    Trophy Points:
    93
    Location:
    Singapore
    Maybe some(e.g. GUTB, RB2013) will go full blast nervosa and use this stupid cable: http://www.audioquest.com/ethernet/diamond :D

    Although switches may not matter that much(provided that the switch is dedicated to Dante Audio alone), basing on my experience with streaming high res files from my NAS, I find that Network interface cards do matter more. My preference is Intel based NIC. Realtek and especially Broadcom seems to have more packet loss or much higher cpu usage. I suspect it could be due to either poor performing hardware or bad drivers or combination of both.

    AOIP is still very bleeding edge technology. Only wish is for companies like Schiit to consider implementing Dante direct into Yggdrasil or at least let us know what they think of AOIP.
     
    Last edited: Oct 12, 2016
  15. perkele

    perkele Acquaintance

    Contributor
    Joined:
    Aug 28, 2016
    Likes Received:
    87
    Trophy Points:
    28
    Location:
    Northern Hell
    For those who will plant the RedNets to more congested networks: https://uk.focusrite.com/rednet-ethernet-switch-requirements . QoS section provides easy to follow guide for less experienced users (provided one uses Cisco SG series switches mentioned in text ).

    As many have said already - 1GB lan is overkill for 99% of the users. For some of us it's a true necessity. Luckily even the good switches do not cost much.
     
  16. jelt2359

    jelt2359 Friend

    Pyrate
    Joined:
    Nov 8, 2015
    Likes Received:
    157
    Trophy Points:
    43
    I have a different DAC. I use the Chord Dave, which is supposed to have as its best input USB; second is Toslink Optical, and last is AES/BNC.

    Comparing my microRendu with my Rednet 3 on this DAC (both fed by FMC optical converters), the two are very very close, which makes sense, since both are Ethernet in devices that avoid the USB connection from the computer. This AB test was exceptionally easy to do, since I used Roon as a player, grouped the Rednet3 and microRendu, and then just swapped between AES (rednet 3) and USB (microRendu) on my DAC with the click of a button.

    However, I have experienced significant differences when switching between different USB cables on the USB-out of the microRendu. That to me is a negative sign. I'm also currently only using the iFi switching power while waiting on my Reflektor-D, and apparently the microRendu is also very sensitive to power supply changes (to me, another negative sign- but this does mean that the microRendu could conceivably get better).

    Finally, a word on usability. My microRendu has far, far more connection drops when using the Roon->HQP combination. The RN3, using the Dante Asio protocol, is much more stable, and the connection from my server to the RN3 tends to drop far less.

    All in all, both of them are really good, but I'd probably lean RN3 for the ease of use issue, and the need to not fudge around with different USB cables and power supplies. But I'm at the point where- and I've given this advice to those who've PMed me on some comparisons- I think SQ is as good as it gets with either of these devices as a source. Go muck around with other things.

    YMMV, especially with a different DAC.
     
    Last edited: Oct 13, 2016
  17. Thad E Ginathom

    Thad E Ginathom Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    14,135
    Trophy Points:
    113
    Location:
    India
    I'm ashamed, as an ex-techie (hands-on, medium company, jack-of-all-trades system manager), that I'm never quite sure what word to use. I suppose "ethernet" is limiting it to a certain physical structure. Perhaps I should just say IP?

    This is the thing that troubles me with IP audio. All this specialist "audio" IP stuff is going to appear on the market, and it is not necessary. Specialist boxes for audio might have a purpose in the studio where dozens of channels are being dealt with, and, perhaps, all sorts of audio protocols are involved in the whole system. But in the home?

    Naah... you don't even need a switch: get a $5 hub. Except I'm not sure that hubs still exist since switch prices dropped to the point where hubs no longer any point.

    AOIP. whilst still under mucho development, is hardly edge, let alone bleeding edge, technology. This is bleeding-edge networking: 65Tbps over a single fibre. Getting a few (very few, in data terms. Very few compared to video in the home, let alone data transfer in the commercial world) bits of audio data from one side of your house to the other is a job for the bicycle, not the Ferrari. But audiophiles close their ears when people say
    (my added stress)
    My pet theory is that even an old 10Mb coax system would work, but, of course I don't have the parts to try it... and I'm not very good at numbers, so I stand to be corrected on this.

    One of my main reasons for supporting and wanting IP audio is that all the equipment is readily available, fully mature, and damned cheap. But I am well aware that I am on the wrong side of the audiophile glass wall on this one. Even now, some of you can see my lips moving, but you can't actually hear what I'm saying!

    \/
     
  18. Kattefjaes

    Kattefjaes Mostly Harmless

    Pyrate
    Joined:
    Sep 5, 2016
    Likes Received:
    4,521
    Trophy Points:
    113
    Location:
    London, UK
    "It depends". Some systems operates directly at layer 2 directly, so that would be yer actual ethernet or fiber etc.. That's not a million miles away from some of the crazy old stuff like AES over ATM, which is dying out for broadcast use now.

    The stuff I was alluding to, however, is IP based, and handles arbitrary media and metadata, cheerfully doing uncompressed 4k video too when you can get the networks in place. A few channels of audio from a MADI feed wouldn't be too hard.

    (..though getting a lot of modern switches to inter-operate reliably with links above 25Gb/s is proving a bit of a problem right now. A lot of them mostly work on paper.)
     
    Last edited: Oct 13, 2016
  19. Azteca

    Azteca Friend

    Pyrate
    Joined:
    Sep 26, 2015
    Likes Received:
    1,581
    Trophy Points:
    93
    Location:
    Cincinnati, Ohio
    We are getting several hundred channels of RedNet at work in the next two years. Going to be fantastic for routing all across the building. I assure you we won't be using audiophile networking hardware or cables.
     
  20. 3X0

    3X0 Friend

    Pyrate
    Joined:
    Jan 31, 2016
    Likes Received:
    563
    Trophy Points:
    93
    I believe Rb2013 and one or two other HF members found the MC3+USB to be a step-down as a master clock input, but a marginal improvement as a SPDIF reclocker from the RN3. I can't speak to their credentials, but I remember enough people conflating difference and improvement with the Mutecs to give Rb the benefit of the doubt on this one.

    I eagerly await corroboration in either direction (such as from Torq's upcoming evaluations). It may be premature for you to consider the RN3's master clock a con unless you've validated the supposed deficiency yourself.
     
    Last edited: Oct 13, 2016

Share This Page