[rb2013]

   
My only real response to this is... Yep.  
 
I won't go back to USB after having the D16 here the past month, even though the last thing I wanted to do was spend $1500 on a connection to the DAC.

I'm with you there - the $800 RD3 is sounding mighty good.  There is something different in the sound, an ease or musicality, like some kind of hidden distortion has been removed.  Very analog like - nondigital.

 

[occamsrazor]

I've been following this discussion for a while and it's been very interesting...If I may take a step back for a moment and look at the wider picture though. I'm just an amateur without the expertise or access to hardware that some of you have. It seems to me there's a few trends in audiophile computer audio recently, with people claiming benefits but going in different directions, but not so easy to incorporate them all....
 
1. Isolation/Reclocking of USB transmission - isolating the noise from the source/player hardware up to the DAC.
 
- On direct USB, such as Uptone Regen, Intona, etc
- Via USB Interfaces such as Gustard, Singxer F1/SU1, Mutec MC3+USB etc
 
2. Moving from USB to network/IP transmission:
- Via the network using player-specific protocols like HQPlayer NAA, Roon RAAT etc into small Linux boxes such as uRendu, SonicOrbiter, RoonReady devices.
- Via the network using OS-wide virtual soundcards such as in the Dante system and Focusrite Rednet, Ravenna, AVB, AES67.
 
3. Upsampling material to high DSD data rates
- HQPlayer, Audirvana, etc fed to DSD capable DACS at high rates like DSD256/512.
 
4. As a Mac user, many of these high-rate DSD DACs aren't accessible at those rates because of the need for DoP.
 
5. DSP/Room correction can bring significant benefits, but combining it as software with some of the above methods can bring a bunch of compatibility issues, and the available hardware such as MiniDSP restricts data rates, as well as difficulties combining with DSD.
 
So there's a lot of interesting methods being used, but it strikes me as difficult to combine them and some of them are incompatible with each other...
 
A. "Professional" DAC equipment seems to shy away from anything higher than 192khz, in many cases 96khz, and doesn't seem concerned much by DSD compatibility or high DSD rates
 
B. "Professional" network protocols such as Dante/Rednet of which there's been some very strong positive comments recently also limit the rates.
 
C. Methods that do allow high-rate upsampling and network transmission such as HQPlayer NAA require you to be locked into that particular player (or via Roon), and the endpoints such as those NAA-compatible like the uRendu bring you back to a USB connection to the DAC.
 
D. Connection methods that don't use USB such as AES, Toslink, SPDIF, with the exception of i2s that is still fairly rare externally, always seem to restrict the data rate.
 
Given the bandwidth of Gigabit ethernet, it would seem it is easily capable of higher rates than 24bit/192khz. These Dante devices can do dozens of channels simultaneously, but restrict a single channel's rate. In an ideal world we'd have some type of network endpoint that could receive very high rate PCM and DSD material across a LAN via ethernet, outputting to i2s then to DAC, I guess also with USB, AES, SPDIF options for compatibility, but that wasn't locked to a specific player such as HQPlayer and was system-wide.
Or that was built-in to the DAC so that the path from the network input to the DAC chip didn't involve an intermediate interface such as USB. Something like the Burl B2 Bomber DAC probably seems closest in the sense that you have a systemwide virtual Dante soundcard, going over the network to a network input inside a DAC. But it's limited to 192khz. Or the uRendu that allows high data rate over the network but then connects to the DAC via USB. It seems you can't have everything.... yet.
 
Like I say, I'm just an amateur with an interest in all this, and I have none of this hardware to use or try personally, and have limited budget. I just wonder what people's thoughts are on how best to try and combine some of these methods, or if that isn't possible then which to prioritise over others and which gives the most benefits.

 

[mourip]

To  occamsrazor:
 
Excellent synopsis of a long, exciting, and techical thread. I am sure that Rob will weigh in but I think that you have basically answered your own questions.
 
It seems to me that currently there is no cheap solution for either an advanced USB "chain" or a cutting edge ethernet solution. And in either case no way to pull DSD out of it.
 
Personally I would rather have perfected Redbook which has nearly unlimited acclaimed performances than to build for DSD. I had both a Vega DAC and also a Wyred4Sound DSD DAC for a while and never could really get behind either the DSD sound or it's unreasonable media cost. Possibly listening through Saber chips turned me off.
 
Right now I am using a Mutec USB DDC with AES to my Schitt Iggy and I find the sound to be wonderful. Having said this I too think that ethernet is the way to go and am looking at the current REDNet devices as an interim step up while waiting for a more dedicated(and cheaper) audiophile device to appear.
 
I would say that based upon Rob's, and others, pioneering findings that on a budget the F1 might be your best bet. You can always pile on USB tweaking devices to incrementally build up the SQ however as you do that you will be heading toward the magic $1000 where you find Mutec and REDNet waiting.
 
What ever you choose things are moving so fast that in a year it will all look different. Heck, in reading this thread things have changed in two weeks!

 

[occamsrazor]

  To  occamsrazor:
 
Excellent synopsis of a long, exciting, and techical thread. I am sure that Rob will weigh in but I think that you have basically answered your own questions.
 
It seems to me that currently there is no cheap solution for either an advanced USB "chain" or a cutting edge ethernet solution. And in either case no way to pull DSD out of it.
 
Personally I would rather have perfected Redbook which has nearly unlimited acclaimed performances than to build for DSD. I had both a Vega DAC and also a Wyred4Sound DSD DAC for a while and never could really get behind either the DSD sound or it's unreasonable media cost. Possibly listening through Saber chips turned me off.
 
Right now I am using a Mutec USB DDC with AES to my Schitt Iggy and I find the sound to be wonderful. Having said this I too think that ethernet is the way to go and am looking at the current REDNet devices as an interim step up while waiting for a more dedicated(and cheaper) audiophile device to appear.
 
I would say that based upon Rob's, and others, pioneering findings that on a budget the F1 might be your best bet. You can always pile on USB tweaking devices to incrementally build up the SQ however as you do that you will be heading toward the magic $1000 where you find Mutec and REDNet waiting.
 
What ever you choose things are moving so fast that in a year it will all look different. Heck, in reading this thread things have changed in two weeks!

 
Interesting thoughts, thank you... and agree with the perfecting Redbook idea to some extent. I think one of the things that prevented interest in DSD was the level of media availability. What I am seeing more, at least on these forums, isn't people playing actual DSD material so much as claiming advantages of playing Redbook upsampled to DSD. This is what appears to be behind the increasing interest in high DSD-rate DACs such as the T+A DAC 8 DSD, iFi Micro DSD etc.
 
What I'd love to see is a one-box solution like the Singxer SU-1 but with some type of Ethernet input... how great might that be?
 
Since I wrote that post have been doing some more reading and have noticed one other option for Dante. Bear in mind this is just from reading no experience whatsoever!. Rather than the Rednet dedicated hardware, I am wondering about a small PC with Dante input and i2s output. From what I've been reading the Dante DVS or Via software available on Mac/Windows could be used as a Dante "input" as well as "output". As there's no Linux version of Dante, it would need to run Windows. I wonder if there are any single-board small computers (like Raspberry Pi, Odroid, Beaglebone etc) that can A) Run Windows and B) Have i2s output.
 
https://www.audinate.com/products/software/dante-via
 
So you have a small PC running Windows and Dante Via as a receiver over ethernet connecting to the main Mac/PC that's also running Dante Via as sender. Then that small PC Dante "receiver" outputs the audio over i2s to a DAC with i2s input. I don't know much about i2s output hardware options but one might be something like this:
 
http://www.tnt-audio.com/sorgenti/pinkfaun_i2s_bridge_e.html
http://www.modelpromo.nl/PinkFaun_I2S_Bridge.htm
 
That of course doesn't get around the high-rate DSD issues, but might be an interesting and possibly cheaper way to add Dante connectivity to a DAC instead of something like the Rednet.

 

[somestranger26]

I think using a PC with ethernet input and i2s output would defeat one of the primary purposes of using audio over IP in the first place, which is to achieve galvanic isolation between the noisy computer source and the DAC.

 

[occamsrazor]

  I think using a PC with ethernet input and i2s output would defeat one of the primary purposes of using audio over IP in the first place, which is to achieve galvanic isolation between the noisy computer source and the DAC.

 
Yes that is probably true... was just trying to think of alternative Dante options.

 

[mtoc]

Any updates, OP?

 

[motberg]

   
Yes that is probably true... was just trying to think of alternative Dante options.

This would be a real stretch... but maybe an UP board (can run Win10)
http://www.up-board.org/kickstarter/up-in-partnership-with-pi-2-design/
 
And something like IanCanada's isolation HAT (may not be compatible with UP)
https://twitter.com/iancanadaTT
 
can be cobbled together...

 

[mtoc]

btw, does that bkue jeans cagmt cable sound good on rednet?

 

[mtoc]

pss, OP, you've mentioned UTP sounds better than STP on the lanrover stuff (I know it ain't lanrover but i can't remember its complex-name), how about utp vs stp on rednet 3?

 

[mhamel]

  pss, OP, you've mentioned UTP sounds better than STP on the lanrover stuff (I know it ain't lanrover but i can't remember its complex-name), how about utp vs stp on rednet 3?

 
Use UTP. STP is not necessary. If you do use it, the shield should only be connected at one end. but it really isn't necessary. You're passing so little traffic compared to what the cable is designed to handle already. 
 
Ethernet by nature is isolated since the data lines are transformer coupled. There is no need for additional isolation there or other boutique gizmos. While this isn't an inexpensive option, part of the beauty of it is in the simplicity. Look at the chain of gear that Bob was using and the hoops to jump through to get from point A to point B, none of which are required for this.
 
The BJC ethernet cable is just a very well-made, tested Ethernet cable. They test each one and make sure it is in spec and constructed correctly.
 
  -Mike

 

[jabbr]

Mike

Would like to hear your view on this quote taken from the page of Acousence about their GISO LAN isolator:
"...devices which serve as a "bridge" between the analog and digital world (AD or DA converters) have always been [using] {word added by jabbr} special components which are recommended (AES) or even obligatory (EBU) in the standardisation guidelines of the professional studio industry; so-called transformers - small hardware elements, which transfer a signal in a purely inductive manner without a physical connection to the conductor - prevent or at least decrease these disruptive influences."
Full text at: http://www.artistic-fidelity.de/index.php/en/giso-isolator

Though the Rednet itself isn't attached to the analogue device itself, it is the beginning of the chain that is.
Wouldn't the quote imply there are benefits to be had by adding a GISO in front of a Rednet?

Cheers

 

[mhamel]

Mike

Would like to hear your view on this quote taken from the page of Acousence about their GISO LAN isolator:
"...devices which serve as a "bridge" between the analog and digital world (AD or DA converters) have always been [using] {word added by jabbr} special components which are recommended (AES) or even obligatory (EBU) in the standardisation guidelines of the professional studio industry; so-called transformers - small hardware elements, which transfer a signal in a purely inductive manner without a physical connection to the conductor - prevent or at least decrease these disruptive influences."
Full text at: http://www.artistic-fidelity.de/index.php/en/giso-isolator

Though the Rednet itself isn't attached to the analogue device itself, it is the beginning of the chain that is.
Wouldn't the quote imply there are benefits to be had by adding a GISO in front of a Rednet?

Cheers

 
Honestly? I think this is marketing hype. Ethernet connections using USP cables are already transformer coupled on all connected pins by design, AES should also be transformer coupled by design.
 
So in connecting the PC to the Rednet box, then out through AES, you have isolation at multiple points without adding any special additional hardware.
 
PC -> Transformer -> RJ45 -> Ethernet Cable -> RJ45 -> Transformer -> RedNet Interface -> Transformer -> AES -> Transformer -> DAC.
 
Everything is encapsulated in packets on the Ethernet network, with built-in error checking and re-transmits. You can easily monitor the network connection to determine if there is any packet loss or errors in the data stream, in which case it is most likely a faulty cable, NIC card or other hardware issue causing the problem, not interference. If there is noise somewhere in the connection that injects data into the packet, the checksum will fail and it will be re-transmitted. With the relatively small amounts of data we're talking about, it isn't even remotely coming close to hitting performance barriers of the network - especially if you are connecting directly from the PC to the Rednet box.
 
I know I'll probable hear something along the lines of "It's audio, it's different" for saying what I'm about to say, but the reality is, the underlying network infrastructure and protocols do not care what the data is. They are designed to assure it gets from point A to point B without errors.
 
I work in IT, specializing in enterprise storage. We have storage racks in extremely busy data centers running 40 gigabit Ethernet connections across 4x 10Gbe using Cat 6a cables under the floor in data centers that have potential levels of electrical interference you would never even begin to see in a home environment. Even within the rack, you're talking about a rack switch for the management interfaces, two or more storage controllers, which are high end servers with 1.5TB of RAM and 32-64 CPU cores, then 8 disk shelves with 24x drives in each... over 1.5PB of raw storage in a rack. Multiple power supplies in each box, lots of cabling. Not an audiophile power conditioner, cable or other gadget in sight. Noisy, high speed/high volume fans all over the place. Yet these things can run for months at a time between maintenance windows or reboots, flat-out, pushing even the 40 gig network connection to it's limit, serving thousands of client machines... with zero packet loss or network errors.
 
I'm not saying there aren't potential ways to improve what gets from the PC to the DAC, not getting into clocking or anything like that. From a pure data integrity standpoint, talking about the data that the PC sends across the network - if all of your hardware and connections are good, if you're using a well built cable that meets or exceeds spec (like the BJC cables) the data you feed into the network is going to be *exactly* the data that arrives at the Rednet box, and no isolation device is going to change that.

 

[Muziqboy]

 
I'm not saying there aren't potential ways to improve what gets from the PC to the DAC, not getting into clocking or anything like that. From a pure data integrity standpoint, talking about the data that the PC sends across the network - if all of your hardware and connections are good, if you're using a well built cable that meets or exceeds spec (like the BJC cables) the data you feed into the network is going to be *exactly* the data that arrives at the Rednet box, and no isolation device is going to change that.

 
Very good point here Mike!
And out of curiosity's sake, I tried a bulk 3ft. Cat6 cable from FRY's and I was getting some clicks and drop-outs in the Music.
Hooked the BJC CAT6a back and no problems whatsoever. So I will order a short 3ft CAT6a cable from BJC since I only need that much length.
Does it matter what length I use between the PC and the RedNet 3? Is there a minimum length I should adhere to?
 
Thanks!

 

[mhamel]

   
Very good point here Mike!
And out of curiosity's sake, I tried a bulk 3ft. Cat6 cable from FRY's and I was getting some clicks and drop-outs in the Music.
Hooked the BJC CAT6a back and no problems whatsoever. So I will order a short 3ft CAT6a cable from BJC since I only need that much length.
Does it matter what length I use between the PC and the RedNet 3? Is there a minimum length I should adhere to?
 
Thanks!

 
No minimum length, 3 feet should be just fine.