Ethernet Cables, Audiostream test...

[barrows]

Michael at audio stream recently tested the Audioquest Ethernet cables. The results are going to be controversial. Let's for a moment agree that there is a difference in sound quality when streaming music files from a NAS to a server, or Renderer. Then, why/how could that be?

What I would like to see is some network analysis of the various cables. As Ethernet is an error corrected transmission protocol, it is going to be bit perfect. But, consider this: perhaps with a more precise cable there are less requests for re-sends, and if there are less less errors and subsequent re-sends, perhaps this higher performance results in less noise produced at the receiving end, and subsequently better sound quality. I am speculating wildly of course, and I do not know a lot about networking. Folks who do, what do you think? And can test equipment tell us how many errors and re-sends happen in Ethernet data transmissions?

[AudGuy]

On a Windows 7 platform we have access to tools that can determine some of these things.

Right click on the Task Bar select Task Manager select Performance Tab select Resource Monitor button.

In the Resource Monitor select the Network Tab then note the TCP Connections box.

 

Two columns to observe are the Packet Loss and Latency ... both will show network performance factors.

 

This is a start towards determining packet processing, but may not give send/resend numbers.

In large network systems more engaging software and hardware are used to determine these factors.

 

But, assuming that GIGA-net is used throughout, that is adapters and switches are all GIGA level ...

This should provide throughput far beyond the requirements for streaming music, even HiRes music.

This performance will be obtained using CAT5e or CAT6, I see no differences in my installation.

 

As Barrows points out, this is bit-perfect lossless error-corrected transmission ... what's to go wrong?

Sonic variation ... seems unlikely although a really robust test must be done to know for sure.

 

I suspect that if there is packet resends occurring then this is likely a result of factors other than cables.

Although a long expensive cable could be run point to pint and some observations made ...

 

One component to consider are consumer LAN switches, they are made to conform to a price point.

Geez, sonic difference in network streaming just makes me a little suspicious.

 

But I am open to the possibility that this may be so ...

[Audio_ELF]

I'm not going to debate if they sound different ... but has anyone considered a CAPS type computer (would need the Zuma motherboard and a larger case) and fitting a fibre adaptor rather than relying on Cat 5/6/7?

 

This would remove potential EMI problems which is the proposed reason that the Audioquest cables sound different (I believe Audioquest talk a lot about removing EMI on their website too).

 

Eloise

[Julf]

Folks who do, what do you think? And can test equipment tell us how many errors and re-sends happen in Ethernet data transmissions?

 

A proper analyzer can show you much more than that - it can show you noise margin and jitter (eye diagram), but they are expensive. To just look at re-transmits and dropped packets, just use wireshark (free) on a PC.

 

Anybody doing a cat 5 cable test should have done at least all those measurements.

[Cebolla]

I've no interest in debating the sonic difference too, but I hope we are all clear that network access in these situations is confined to file data, being used as an alternative to reading the same file data from a local storage device. The decoding of the file data to audio occours at a later stage, by player software if it's a computer, or firmware/hardware if it's a renderer/player/streamer device. Hence any sonic differences should be due to network data delays & dropouts and the player's handling of them.

[jcalleja]

With all respect I think the article points at noise through EMI beign captured by the cables and introduced in the DAC and rest of audio components as the key element of the differences

Not to anything related to the data transmitted

If I understood it correctly....

Regards

[Cebolla]

I haven't read the article, but would have thought that is so - certainly Audio_ELF's comment hinted at EMI noise. My last post was aimed at those who were considering responding to the OP's question (& presumably yours is too):

But, consider this: perhaps with a more precise cable there are less requests for re-sends, and if there are less less errors and subsequent re-sends, perhaps this higher performance results in less noise produced at the receiving end, and subsequently better sound quality. I am speculating wildly of course, and I do not know a lot about networking. Folks who do, what do you think?

We certainly don't want to start getting into another one of those debates equating network file data access dropouts with the loss of sound quality due to audio jitter, again!

[Julf]

We certainly don't want to start getting into another one of those debates equating network file data access dropouts with the loss of sound quality due to audio jitter, again!

 

Definitely not - but what I was pointing out is that if anyone has any doubts and speculations, it is easy to measure the data loss and transmission quality. If there is no correlation between those and perceived sound quality, we know we are talking about voodoo.

[Julf]

With all respect I think the article points at noise through EMI beign captured by the cables and introduced in the DAC and rest of audio components as the key element of the differences

 

And we do have to remember that ethernet cables are balanced/differential and galvanically isolated with transformers...

[Allan F]

Michael at audio stream recently tested the Audioquest Ethernet cables. The results are going to be controversial.

 

The problem I have with Michael's review is that he included a generic CAT5 cable for comparison purposes. Inclusion of an "industrial" quality CAT7 or even CAT6A cable for comparison, IMO, would have been far more informative.

[reverendo]

today I got the newest firmware for my Resolution Audio Cantata and it seems to be ready for prime time, at last.

this means that the ethernet that runs from my switch to the DAC might be important. I have 3 ethernet cables (CAT5e, CAT6 and CAT7) at home that I will compare to see if there are any sonic differences.

If there are I'll have a look at the ones from AQ.

let's see how this goes.

[Superdad]

Reverendo:

 

I urge you to put the BlueJeansCable Cat6a (mfrd. by Belden) into the mix. It is solid core, floating shield, bonded-pair, and rated to over 500Mhz. Very reasonable price and it easily won in my comparison test. Beware of cables claiming to be Cat7. The standard is not set and the actual construction varies all over the map (I got one on Amazon claiming to be Cat7 and it was made with copper-clad aluminum wire!). Below is an excerpt from something I wrote in another thread. Feel free to ask me any questions about some of the construction aspects I learned and what I heard.

 

Best,

ALEX C.

--------

 

Previously I never heard or even tried to hear Ethernet cable differences because I was always drawing my music from directly-atttched storage.

I should clarify the circumstance that has allowed me to hear the quality of the Ethernet cable (and BTW, it is the unique, Belden made bonded-pair 10GX Cat6a sold by BlueJeansCable that stands above the rest). Since the good sounding SDXC card slot of the 2010/11/12 Mac minis share the PCIe controller chip with the Ethernet port, and since I had just finished comparing--and hearing differences with every other interface with the range of drives I have (FW400, FW800, USB, SD card, internal SATA, RAM disk), I decided to try sharing a drive on the desktop mini (which I also use to control the headless music mini). It sounded okay, but there was some nagging, slightly edgy quality to it and some life and air went missing. The obvious step was to bypass my Cisco gigabit switch and run the 25ft cable just Mac-to-Mac. Wow, that really helped! This was just with my 8 year-old loose Cat5 cable whose locking clips were long ago broken off. Having read a couple of surprised-reviewer articles on the AudioQuest Ethernet cables, I figured I'd research a bit about Cat5/6/6a/7 construction on my own (I've been making DIY cables and commercial cables--with Hovland--since 1978, so wire research is fun for me). I ordered up several types (shielded outer, shielded pairs, floating shield, unshielded, bonded-pair, stranded and solid conductors), as many as possible from the same manufacturer to keep the variables down to just what I knew about the differences. All were 25-footers--and all from Amazon so I could return what I did not like.

 

Anyway, I don't want to make this post an Ethernet cable review. But I will say a few last things about it:

1) After I picked the cable I like best (Data Cables at Blue Jeans Cable), I was, for the first time, able to get Jumbo Frames (MTU 9000) to work between the Macs. The other cables would not support that.

2) Comparing standard MTU 1500 versus MTU 9000 gave a small but noticeable nod to MTU 9000.

3) This setup now equals or exceeds the SQ of the SD card (which allows me to remain booting from SD and gives me access to my 2TB hard drive music library--that's a big deal for me.)

[reverendo]

Reverendo:

 

I urge you to put the BlueJeansCable Cat6a (mfrd. by Belden) into the mix. It is solid core, floating shield, bonded-pair, and rated to over 500Mhz. Very reasonable price and it easily won in my comparison test. Beware of cables claiming to be Cat7. The standard is not set and the actual construction varies all over the map (I got one on Amazon claiming to be Cat7 and it was made with copper-clad aluminum wire!). Below is an excerpt from something I wrote in another thread. Feel free to ask me any questions about some of the construction aspects I learned and what I heard.

 

Best,

ALEX C.

--------

 

Previously I never heard or even tried to hear Ethernet cable differences because I was always drawing my music from directly-atttched storage.

I should clarify the circumstance that has allowed me to hear the quality of the Ethernet cable (and BTW, it is the unique, Belden made bonded-pair 10GX Cat6a sold by BlueJeansCable that stands above the rest). Since the good sounding SDXC card slot of the 2010/11/12 Mac minis share the PCIe controller chip with the Ethernet port, and since I had just finished comparing--and hearing differences with every other interface with the range of drives I have (FW400, FW800, USB, SD card, internal SATA, RAM disk), I decided to try sharing a drive on the desktop mini (which I also use to control the headless music mini). It sounded okay, but there was some nagging, slightly edgy quality to it and some life and air went missing. The obvious step was to bypass my Cisco gigabit switch and run the 25ft cable just Mac-to-Mac. Wow, that really helped! This was just with my 8 year-old loose Cat5 cable whose locking clips were long ago broken off. Having read a couple of surprised-reviewer articles on the AudioQuest Ethernet cables, I figured I'd research a bit about Cat5/6/6a/7 construction on my own (I've been making DIY cables and commercial cables--with Hovland--since 1978, so wire research is fun for me). I ordered up several types (shielded outer, shielded pairs, floating shield, unshielded, bonded-pair, stranded and solid conductors), as many as possible from the same manufacturer to keep the variables down to just what I knew about the differences. All were 25-footers--and all from Amazon so I could return what I did not like.

 

Anyway, I don't want to make this post an Ethernet cable review. But I will say a few last things about it:

1) After I picked the cable I like best (Data Cables at Blue Jeans Cable), I was, for the first time, able to get Jumbo Frames (MTU 9000) to work between the Macs. The other cables would not support that.

2) Comparing standard MTU 1500 versus MTU 9000 gave a small but noticeable nod to MTU 9000.

3) This setup now equals or exceeds the SQ of the SD card (which allows me to remain booting from SD and gives me access to my 2TB hard drive music library--that's a big deal for me.)

Hi Alex,

Thanks for your input. I'll definitely have a look at that.

Fact is that I listened to 3 different cables this afternoon and there are differences. The CAT5e came with the Netgear router, the CAT6 was bought at Micro Center and the CAT7 is from Rosewill that I got from Amazon. In my case my UPnP system looks like this:

Dell XPS 15 (in the office) > Netgear WN4500 (router) > Netgear WNCE4001 (wireless ethernet switch) > Resolution Audio Cantata (as Media Renderer)

This way I get to use ethernet similar to the way I used my modified SBT. I'm using UPnPlay, but am not completely happy. Bubble didn't detect the Cantata. I use MinimServer and Foobar to play the files. Any ideas on the app?

I haven't had time to sit down and find a cable that I would prefer yet, but differences were there. Fact is that it's already playing practically on the same level as the SBT + expensive USB cable.

Would love to hear your input and experience in regards to this. I'll try to do some serious listening and choose the one I prefer to make that a benchmark in comparison to the USB input with transport and other USB cables. The one from Belden might be a good place to start, but the ones from AQ seem interesting.

Best regards

[cjf]

So here's my take on this. IMO, any gains that one Ethernet cable may perceive to bring to the table over another would be from one of them using a better shielded jacket over the other to help isolate the electrical signal traveling thru it from other nearby wires in the general vicinity.

 

Depending on your setup, some may terminate this noisy ethernet cable stright into their DAC and others would terminate it into thier media PC. In either case, niether of these devices can afford to have their own noise compounded by all the other network equipment upstream of them which is now connected thanks to this cable. So basically you end up with a giant web of noisy systems all linked together by a bunch of Ethernet cables.

 

I wouldnt doubt that a crossover cable configuration from NAS to DAC/ Media Sever would yield better results because you would be minimizing the amount of crap that usually helps polute the connection ( multiple cheap switches, routers, wall warts..etc).

 

 

I've mentioned the use of Ethernet optoisolators in the past and while some may point and laugh at the sight of such a contraption it's hard to ignore the technology and how it breaks this dirty electrical link between all these components.

 

Why bother building a "green" media server with things like LION Batteries, separate battery powered USB cables and any other tweak you can think of if you are going to just turn around and link it up to an Ethernet connection directly. In doing so, you've just provided a direct path for all that noisy crap to counteract all the tweaks you just made.

 

So in summary, I feel that all these cables are doing is lessening the chance of furthur dirtying an already dirty connection by way of better shielding. The use of solid core verses stranded or gold verses silver for the conductors for this link in particular is just putting lipstick on a pig.

[Paul R]

HI Barrows - there is one thing in your post that stopped me cold - that would be where you called "Ethernet" a protocol. It isn't, it is a media.

 

What travels over that media is a whole bunch of different protocols, some at fundamentally different layers than other. For example, TCP is the Transmission Control Protocol that may govern the physical transmission of data on the cable. The transmission itself is a serial protocol. (i.e. subject to timing issues and jitter.)

 

TCP/IP refers to TCP using Internet Protocol, which is more than anything else, an addressing mode. TCP does put data in packets, but that data goes over the wire one bit at time, similar to USB media transmissions in many ways.

 

On top of (encapsulated in) TCP/IP or TCP/UDP is where our music rides. These encapsulated protocols, like SMB, NFS, RTP, telnet, etc. are usually thought to be somewhat immune to jitter or media concerns.

 

I would think TCP (IP or UDP) is where, if anywhere, cables could make an audible difference. At the application level, it ->should<- not be an issue, as by the time it gets to an application, it is buffered and the transmission timing is lost. That of course, may prove untrue. I certainly would not be totally shocked if it turned out to be a false assumption, though I don't have a clue what the reasons might be.

 

I do not doubt people are hearing differences, despite the fact I have not heard any difference from Ethernet cable. I just wanted to point out that Ethernet is not a protocol before other people jump on it and derail this entire interesting conversation. I will shut up now and go back to reading.

 

-Paul

 

 

 

Michael at audio stream recently tested the Audioquest Ethernet cables. The results are going to be controversial. Let's for a moment agree that there is a difference in sound quality when streaming music files from a NAS to a server, or Renderer. Then, why/how could that be?

What I would like to see is some network analysis of the various cables. As Ethernet is an error corrected transmission protocol, it is going to be bit perfect. But, consider this: perhaps with a more precise cable there are less requests for re-sends, and if there are less less errors and subsequent re-sends, perhaps this higher performance results in less noise produced at the receiving end, and subsequently better sound quality. I am speculating wildly of course, and I do not know a lot about networking. Folks who do, what do you think? And can test equipment tell us how many errors and re-sends happen in Ethernet data transmissions?

[coxhaus]

I am looking forward to the findings. I am using CAT5 and I am wondering if up grading would buy me anything. I feel like I am picking up like a ground noise when the Ethernet is connected. I am not a 100% sure but it seems like I have a blacker back ground when the Ethernet is not connected, physically unplugged. My computer runs from a 2500 watt power isolator and when I connect the Ethernet I lose some of the benefits of the power isolation. This is just a feeling not something I could pick blind yet.

[reverendo]

just pulled the trigger on the CAT 6a from Belden just to give me some perspective. should arrive Tuesday.

in the meantime I will give the cables here a more thorough listening session.

thanks for the idea, Alex

[Cebolla]

I am looking forward to the findings. I am using CAT5 and I am wondering if up grading would buy me anything. I feel like I am picking up like a ground noise when the Ethernet is connected. I am not a 100% sure but it seems like I have a blacker back ground when the Ethernet is not connected, physically unplugged. My computer runs from a 2500 watt power isolator and when I connect the Ethernet I lose some of the benefits of the power isolation. This is just a feeling not something I could pick blind yet.

Have you tried playing music with an ethernet loopback plug inserted (perhaps running some network loopback test software in the background, to get the network card to 'do something') and seeing if you are getting the same effect. It could be the network card itself, rather than the attached cable that might be causing the effects.

[barrows]

Paul: thanks for that, and you can probably tell, I am not a computer guy, and so use terminology incorrectly. What do you think about my idea that data errors, and the resulting re-sends may be causing sonic differences? My speculation is that re-sends result in more noise (spikes in the power supplies) which may be affecting the DAC itself. I certainly do not believe that data related "jitter" (as opposed to audio data stream jitter at the DAC) is an issue.

It seems clear that people are hearing differences with Ethernet cables, and testing by Blue Jeans appears to indicate that many brands of Ethernet cable do not meet spec for bandwidth. Higher bandwidth should result in less data errors, and hence less re-sends? Am I correct so far?

Thanks Alex for pointing to the BJ/Belden 6A cable, seems like a great choice. I would not be comfortable recommending anyone pay audiophile type prices for Ethernet cable (sorry Steve), as opposed to USB, I2S, and Analog cables.

[Paul R]

Hi Barrows, I think that your idea is plausible, though honestly, I am not sure exactly how the coupling would take place. 10BaseT and above network are usually connected to a switch, and there are not usually a whole lot of packet retransmits, but it is possible.

 

Another thing to think about, the 10, 100, and 1000 in front of the standards means the data is transmitted on the cable at 10Mhz, 100Mhz, or 1000Mhz. That's fast enough to be considered up there in the HF and low VHF bands I think. Could there be some kind of inductive coupling?

 

 

-Paul

[coxhaus]

Have you tried playing music with an ethernet loopback plug inserted (perhaps running some network loopback test software in the background, to get the network card to 'do something') and seeing if you are getting the same effect. It could be the network card itself, rather than the attached cable that might be causing the effects.

 

I have a loopback around here somewhere. It is tried and tested when I was working. I will look for it and start playing around with it.

[Superdad]

Hi guys:

 

Some random thoughts and questions to throw into the mix, all just based on the afternoons I spent first researching cable types (many items, not just Cat rating, see above) and then trying them out and adjusting settings.

 

1) Ethernet (hereafter just EN) cable only matters to SQ if you are pushing/pulling the music over the network to your DAC-connected computer. Only exception to that might be degradation caused by using an attached-shield cable (as opposed to floating shield or tied at one end only) wherein you lose galvanic isolation and allow in ground noise from EN switch even if you only use the EN for web-access/metadata lookup.

 

[bTW, all my comments regarding SQ of EN cable are with a 25-footer used to draw music files from a FW-connected shared drive on my desktop Mac mini (which I also use to control the headless mini in the music rack). Audirvana Plus, super-slimmed Mavericks, etc.]

 

 

2) Ethernet switches add to listening fatigue and somewhat mask the "character" of the EN cables. Not just the cheap ones. I tried running through three different switches. Two were older 10/100T consumer units; the third is the 10 port Cicso SG-300 gigabit managed unit I now use for my whole house network (fantastic switch for $197 on Amazon, BTW).

In fact the SQ difference between bypassing the switch--running the EN cable direct, Mac-to-Mac--versus through the switch was as large if not larger than the difference between any of the cables (except perhaps the one really awful cable--see below).

 

[so all my comments regarding SQ of EN cable are with direct connection--no switch! You can see my other post where I found a way to do that while restoring LAN and web access to my machines using the Apple Thunderbolt>Ethernet Adapter--but not for the music.)

 

 

3) Cat7 is B.S. No consistent standards exist, so companies can call something Cat7 even if it is garbage. An example is this one (of many) I got on Amazon: Amazon.com: KEYDEX 25ft CAT7 SSTP 600Mhz Gold-Plated Snagless Network Lan Ethernet Patch Cable - Blue: Computers & Accessories

Despite what the description claims, it is actually made with copper-clad aluminum wire. It sounded horrendous. Worse than my old, loose, totally unshielded Cat5.

 

 

4) Cat6 and Cat6a cable constructions are not all the same. Some Cat6a has the pairs individually foil shielded as well as an overall outer shield (foil or braid), some do not. Some cables use stranded wire, some use solid conductors. Conductor gauge varies from #23, #24, or #26awg. It sometimes takes some digging to find out all the details on a particular wire. Even Belden makes it tough to dig this stuff up.

 

5) I did not like the results with attached-shield cables. These are the ones you see with the metal around the RJ45 connector. They use a "drain wire" to tie the shield to the metal, and that provide continuity to the grounds of the two machines. I would not think that those cables which are also foil shielding the individual pairs are connecting those together to the drain wire, but who knows.

 

6) I thought the cables with solid conductors sounded best (this coming from a guy who prefers finely-stranded wire in almost every other application), but it is a bit hard to know since the number of permutations of construction types I tried was limited to six.

 

7) Both of the EN cables I tried that used a bonded-pair version of Belden's 10GX wire (which also has some another unique topology aspect you can read about here: IBDN System 10GX) sounded better than any of the other cables I tried, even the ones--of other brands--that had foil wrapped pairs.

I read many Belden documents and spoke to a tech at Blue Jeans Cable, and I am convinced that the particular Belden wire BlueJeans uses for its Cat6a is not available as a pre made patch cord from Belden. BJC uses a 10GX, bonded-pair, 24AWG, solid-core, with floating outer shield combination. (I think Beldon only sells that wire combination preterminated with the shield tied to a metal housing--something I don't think I'd like.) And the price BlueJeans charges for it is a bargain for the quality--$12.75 for a 10-footer--each one coming with an individual test printout. The only downside is that this stuff if really thick (7.1mm) and really stiff.

 

8) I think more the more efficient, lower overhead protocols will sound better. UDP via an NFS server/client share has potential, as does an iSCSI SAN set-up. But just upping the MTU of the standard Apple Bonjour AFP connection to 9000 (Jumbo frames) made a nice difference. Note that the only 25-foot EN cables that would allow me to do that were the Belden-wire versions.

I also experimented with lower rates--of both MTU and link speed (both 10-baseT and 100-baseT)--and those sounded worse.

 

That's all I can think of to say on the matter. It is all a bit nuts I know. The differences between the middle cables were subtle, but the differences between the best and the worst (even leaving out that awful CCA Keydex junker) were worthwhile. But you are unlikely to hear any of it while going through an Ethernet switch.

 

As for the Audioquest cables: I'm curious because they did make some good choices--such as solid conductors. It is shielded, but they tied the shield at both ends. If I tied it I'd probably warp some Kapton tape over the metal housing at one end. But it is pricey and I am VERY happy with the BJC. As someone else pointed out, Audiostream should have compared the AQ cables versus something better than cheap Cat5 of unknown construction.

 

Happy Holidays to all,

 

Alex C.

[Superdad]

Another thing to think about, the 10, 100, and 1000 in front of the standards means the data is transmitted on the cable at 10Mhz, 100Mhz, or 1000Mhz.

 

Hi Paul: Not MHz. It's 10, 100, 1000 megabits/second. Using multiple pairs. Quite different I think.

[reverendo]

Hi guys:

 

Some random thoughts and questions to throw into the mix, all just based on the afternoons I spent first researching cable types (many items, not just Cat rating, see above) and then trying them out and adjusting settings.

 

1) Ethernet (hereafter just EN) cable only matters to SQ if you are pushing/pulling the music over the network to your DAC-connected computer. Only exception to that might be degradation caused by using an attached-shield cable (as opposed to floating shield or tied at one end only) wherein you lose galvanic isolation and allow in ground noise from EN switch even if you only use the EN for web-access/metadata lookup.

 

[bTW, all my comments regarding SQ of EN cable are with a 25-footer used to draw music files from a FW-connected shared drive on my desktop Mac mini (which I also use to control the headless mini in the music rack). Audirvana Plus, super-slimmed Mavericks, etc.]

 

 

2) Ethernet switches add to listening fatigue and somewhat mask the "character" of the EN cables. Not just the cheap ones. I tried running through three different switches. Two were older 10/100T consumer units; the third is the 10 port Cicso SG-300 gigabit managed unit I now use for my whole house network (fantastic switch for $197 on Amazon, BTW).

In fact the SQ difference between bypassing the switch--running the EN cable direct, Mac-to-Mac--versus through the switch was as large if not larger than the difference between any of the cables (except perhaps the one really awful cable--see below).

 

[so all my comments regarding SQ of EN cable are with direct connection--no switch! You can see my other post where I found a way to do that while restoring LAN and web access to my machines using the Apple Thunderbolt>Ethernet Adapter--but not for the music.)

 

 

3) Cat7 is B.S. No consistent standards exist, so companies can call something Cat7 even if it is garbage. An example is this one (of many) I got on Amazon: Amazon.com: KEYDEX 25ft CAT7 SSTP 600Mhz Gold-Plated Snagless Network Lan Ethernet Patch Cable - Blue: Computers & Accessories

Despite what the description claims, it is actually made with copper-clad aluminum wire. It sounded horrendous. Worse than my old, loose, totally unshielded Cat5.

 

 

4) Cat6 and Cat6a cable constructions are not all the same. Some Cat6a has the pairs individually foil shielded as well as an overall outer shield (foil or braid), some do not. Some cables use stranded wire, some use solid conductors. Conductor gauge varies from #23, #24, or #26awg. It sometimes takes some digging to find out all the details on a particular wire. Even Belden makes it tough to dig this stuff up.

 

5) I did not like the results with attached-shield cables. These are the ones you see with the metal around the RJ45 connector. They use a "drain wire" to tie the shield to the metal, and that provide continuity to the grounds of the two machines. I would not think that those cables which are also foil shielding the individual pairs are connecting those together to the drain wire, but who knows.

 

6) I thought the cables with solid conductors sounded best (this coming from a guy who prefers finely-stranded wire in almost every other application), but it is a bit hard to know since the number of permutations of construction types I tried was limited to six.

 

7) Both of the EN cables I tried that used a bonded-pair version of Belden's 10GX wire (which also has some another unique topology aspect you can read about here: IBDN System 10GX) sounded better than any of the other cables I tried, even the ones--of other brands--that had foil wrapped pairs.

I read many Belden documents and spoke to a tech at Blue Jeans Cable, and I am convinced that the particular Belden wire BlueJeans uses for its Cat6a is not available as a pre made patch cord from Belden. BJC uses a 10GX, bonded-pair, 24AWG, solid-core, with floating outer shield combination. (I think Beldon only sells that wire combination preterminated with the shield tied to a metal housing--something I don't think I'd like.) And the price BlueJeans charges for it is a bargain for the quality--$12.75 for a 10-footer--each one coming with an individual test printout. The only downside is that this stuff if really thick (7.1mm) and really stiff.

 

8) I think more the more efficient, lower overhead protocols will sound better. UDP via an NFS server/client share has potential, as does an iSCSI SAN set-up. But just upping the MTU of the standard Apple Bonjour AFP connection to 9000 (Jumbo frames) made a nice difference. Note that the only 25-foot EN cables that would allow me to do that were the Belden-wire versions.

I also experimented with lower rates--of both MTU and link speed (both 10-baseT and 100-baseT)--and those sounded worse.

 

That's all I can think of to say on the matter. It is all a bit nuts I know. The differences between the middle cables were subtle, but the differences between the best and the worst (even leaving out that awful CCA Keydex junker) were worthwhile. But you are unlikely to hear any of it while going through an Ethernet switch.

 

As for the Audioquest cables: I'm curious because they did make some good choices--such as solid conductors. It is shielded, but they tied the shield at both ends. If I tied it I'd probably warp some Kapton tape over the metal housing at one end. But it is pricey and I am VERY happy with the BJC. As someone else pointed out, Audiostream should have compared the AQ cables versus something better than cheap Cat5 of unknown construction.

 

Happy Holidays to all,

 

Alex C.

Hi Alex,

Really appreciate the fact that you're taking the time to share your experiences with us. A few questions:

• it seems that you defend that ethernet switches will always interfere with SQ. Would you say the same in regards to the wireless switch (Netgear WNCE 4004) that I'm using. It actually enables me to place any computer far away from my system, giving me the practical and easy approach that my SBT gives me.
  
• because my system is basically wireless I'm thinking that only from the switch on the cabling will be important. does that make any sense from your perspective and experience. please note that my setup is different from most, since my DAC is also a renderer.
  

Thanks a lot in advance

André

[Superdad]

Hi Alex,

Really appreciate the fact that you're taking the time to share your experiences with us. A few questions:

• it seems that you defend that ethernet switches will always interfere with SQ. Would you say the same in regards to the wireless switch (Netgear WNCE 4004) that I'm using. It actually enables me to place any computer far away from my system, giving me the practical and easy approach that my SBT gives me.
  
• because my system is basically wireless I'm thinking that only from the switch on the cabling will be important. does that make any sense from your perspective and experience. please note that my setup is different from most, since my DAC is also a renderer.
  

Thanks a lot in advance

André

 

Hi André: I wish I knew how to answer your questions. I think it just takes experimenting. I am also not clear on how you are connecting your SBT and Cantata. Are you going wireless to the SBT? And then via USB to the Cantata? Or do you have the Cantata with an Ethernet cable to your Netgear switch?

 

Also, I forgot to mention that one of the biggest differences I heard when experimenting with all this Ethernet stuff was the reduction in edge and fatigue when I completely unplugged the old D-Link w-fi router that I had sitting on top of my Cisco switch (which is in my studio/office, on the floor next to my desk). I still have wi-fi in the other parts of the house, but unplugging that unit made a big difference. That might have been before I went with the direct Mac-to-Mac connection bypassing the Cisco switch, so I would have to put the D-Link back in to be able to guess if it was poisoning via air or cable to the switch.