Audiophile Ethernet Switch

Status
Not open for further replies.
I am fortunate enough to have bought a few redundant switches from work. These are Nortel/Avaya ERS series and are exceptionally good. But most people would not buy them. They're in the order of $2k each new. Actually owned by Extreme Networks now. They are nigh on indestructible. To my knowledge some of the American Armed forces only use these.
Kevin
 
Isn't it surprising that we agreed on the importance of a clearly state hypothesis that one wants to examine, but now in a miraculous way the verdict (i.e. the unknown hypothesis is untestable) is already casted in stone? ;)

@scottjoplin,

And interobject variability ;) - in Pavel's preamp listening test I preferred the poorer recording through the lower BW preamp and the better recording through the higher BW one. 100kHz v 1MHz in case anyone is wondering :D

The reason for Olson's experiment was that listeners did not prefer the extended bandwidth recording and reproduction when listening to such reproduction system.
So he designed an experiment with a purely acoustical instrumental orchestra along with acoustical filtering to compare the lowpass filtered to the extended version.
Therefore the quite high percentage of listeners preferring the lowpass filtered version came to my suprise.

In subsequently done experiments it became evident that the reason for the rejection of the extended bandwidth reproduction was due to high IMD of the equipment used. Unfortunately, while the publications about Olson's original experiment was comprehensive, I've not seen any publications about the other experiments, so that information remains a bit vague.
 
A quick google suggests there are some small PCs designed to run router SW that have SFP on. Would be an interesting test and at least prove the galvanic isolation case.

Of course thanks to toslink FUD audiophiles are scared of fibre ethernet :D

There's not much money to be made in fiber cables when everyone has to source their POF or glass from the same vendors :).
 
There's not much money to be made in fiber cables when everyone has to source their POF or glass from the same vendors :).
Indeed, but I have tried a TOSLINK cable and a coaxial cable from the same source into my preamp (AudioLab 8200CDQ) at the same time, played a quality CD (Jazz At The Pawnshop) for some friends (some AudioPhools and some not) and they all preferred the coaxial (of course I never told them which one I was using) and this was setup before anyone arrived. They actually were under the impression that I was using two different sources (I have 3 different CD/SACD front ends.)
In any case the TOSLINK connector is totally shirty - there is no decent clipping mechanism (like an LC connector commonly used on network fibre patch cables.) The LC connector is at least spring loaded so that the two interfaces have an airless connection.
 
Last edited:
Member
Joined 2011
Paid Member
That's correct Ferret, you get what you pay for, a cheap ethernet switch will have a plastic case, a marginal smps and doesn't meet any emc requirements and radiates rfi all over the place.

I agree, however then the issue is not at all about the Ethernet switch, it is about the power supply dumping noise onto the local power circuit. Which brings us full circle back to my sarcastic note about replacing your toaster, furnace, AC units, PCs, TVs, etc. etc. etc. with "audiophile-grade" devices. Might as well install those $300 gold-plated power sockets throughout your home as well.

Insanity. But you know this already.

A much better solution (if one is really concerned about it) would be to install a high quality power filter between the wall socket and all the components in the audio system. Of course, there are audiophiles who poopoo this solution because they claim it causes "veiling" (or any number of mystical issues). :D

Regarding RFI - well, we (and our treasured audio systems) are constantly bombarded by it. If we are really worried about it we shouldn't focus on the sources (which are impossible to control). I am sure there are audiophile Faraday cages available. :D I am willing to bet (without checking) that there are DIY ones around here.
 
Last edited:
Cogitech, dang your idea of a Faraday cage might not be too unobtanium!:cool:
But to be honest, the biggest problem on the power feed (at least where I am) is the horrible amount of phase shift. The electrical engineers (@ the varsity where I work) have done measurements in the past and have reported 30% and more depending on the time of day (or night,) and the season (lots more worse in winter.) I'm not certain how much that will affect our Hi-fi equipment
 
Last edited:
Member
Joined 2014
Paid Member
Indeed, but I have tried a TOSLINK cable and a coaxial cable from the same source into my preamp (AudioLab 8200CDQ) at the same time, played a quality CD (Jazz At The Pawnshop) for some friends (some AudioPhools and some not) and they all preferred the coaxial (of course I never told them which one I was using) and this was setup before anyone arrived. They actually were under the impression that I was using two different sources (I have 3 different CD/SACD front ends.)
which makes my point. because AT never took off for CD people think toslink proves fibre is bad.

In any case the TOSLINK connector is totally shirty - there is no decent clipping mechanism (like an LC connector commonly used on network fibre patch cables.) The LC connector is at least spring loaded so that the two interfaces have an airless connection.
low Mb/s over 10m and 10Gb/s over km are somewhat different requirements. toslink is fit for purpose. just...


Use batteries then, Bill likes batteries :)
I do. I wuss out on the power issues.
 
2. Running at 10/100MB/sec means that:

* You need only two pairs, not four, in your LAN cable. (Gigabit needs all four.) Removing the redundant pairs (4&5, 7&8 - keep 1&2, 3&6) makes for a surprising SQ difference. First, try a length of industrial-grade LAN cable (I use Excel CAT6 UTP as it's easy to get locally) and snip out the redundant pairs. If that works, try replacing the stock connectors with the likes of Telegartners. What matters is less whether the cable is CAT 6, CAT 8 or even CAT1024 and more how well the pairs are twisted and thus how well the all-important impedance spec is met.

Compare a length cut from an Excel cable with one from a typical piece of Amazonic crap to see what I mean - I'd wager that the Excel will typically outperform any cheapo CAT-whatever. There's a good explanation of this on Meicord's web site. See also "Is Your Cat 6 Cable a Dog?" on the Blue Jeans site. Measurements, even.

* You can get surprisingly good results from cheapo 10/100MB/sec switches. I use Zyxels but upgrade the on-board Vregs to an LT3045 device and drive them with Ciunas Audio LiFePO4 battery PSUs.

Hi Dave,

Thank you for the interesting recipe.
I just cut the 4&5 and 7&8 pairs from a CAT6 cable, and the sound is really different. With the low-speed cable, I found the sound is wilder, less-controlled, vibrant, and a bit harsh. With the original cable, the sound is smoother, well-defined, and a little dull. Currently I am not sure which one is more preferable to me, but the recipe really makes a different.
 
AX tech editor
Joined 2002
Paid Member
What is stated without proof can be dismissed without proof. Especially if there's overwhelming proof to the contrary.

People are so uncritical these days, uncritical even towards themselves. No iota of interest to understand stuff, no iota of interest in knowing facts.
It's a wonder society still functions, sort of.

Jan
 
Status
Not open for further replies.