AudioQuest Ethernet Cables

We are all waiting Gajdzin

And especially me as I ordered an NDX yesterday

It will be running on bog oridary CAT5 for a few weeks while final set up is fiddled with,  but I will be looking for better quite soon.  

Last evening I called on a buddy who has Linn DS & pestered him over Ethernet cables – I understand there should be sonic improvements over stnd CAT5,  but at this time I confused which way to go. 

It's good to read of others getting good results from the Cinnamon cables. I now have an 8m Pearl from the router to the switch (Cinnamon was too expensive) and two Cinnamons from the switch to the Naim bits. Before getting the Pearl I used a Supra. I didn't realise when I ordered it that it would be pale blue and really thick, so it had to go. It also had nasty looking plugs with a flimsy catch, whereas the AQ plugs seem much better quality and the catch is protected by a plastic bit. The combination of the Pearl and Cinnamon works really well.

Mike - I'd suggest that you get a switch really close to the system, then you can try a 0.75m Cinnamon to the NDX and it will only cost you £65 which, for the results it gives, is a total bargain.

Goat - I'm not very good at describing differences. At the time, more natural was my impression. I later discovered that the PRAT was being removed by the Chord Odyssey I was using. Now the NAC A5 has returned, the PRAT has returned too, and with the Powerline and AQ Ethernet leads, everything sounds really excellent.

Thanks HH,  I'm on the same wavelength believe me - I read, note, re-read & then go find out for myself. 

Re the switch close by: The phone co are moving the incoming point on Tuesday to be in the same area as the hifi. The router with 4 ports is the switch & will be within 0.75m of the NAS, but the NDX will need a 1.5m or 2m length. 

My questioning on what cable revolves around the theory of diminishing returns & my suspicion with "audiophile" labels & as such I have developed a built in snake oil & mirror smoke detection system. 

I am not looking for "bargains" as such; my priority is SQ 1st but maybe not stretching to the Chord STA @ 4 digits per metre. 

I have more or less settled on Cinnamon in my plans, thanks to the folks like yerself on the forum & last evening at buddy's with Linn DS.   He has gone thru a series of bog stnd stuff, then Supra & inbetween tried some rarefied stuff & finally Audioquest & rated 3 types as follows

Forest is good, better than Supra but less detail & muddled sound stage effects

Cinnamon gives much more detail than Forest and more of that airy feeling which Forest and Supra are missing.  It is the most natural sounding cable IHO

Vodka is awesome; music is highly polished but somehow too polite & missing something.  

His final choice was Cinnamon as a clear winner.  

I just realised, looking back at the beginning of this thread, that I'm simply repeating what I wrote back in February! Still, as least I'm being consistent.

Mike, I really would keep things like Ethernet switches/routers/you name it  well away from your audio equipment and interconnects. I would try and put a few metres between them, I have 6 metres.

The RFI does does tend to creep in robbing you of detail, and I found my FM tuner was most sensitive, but not exclusive to being affected.

If you are determined to use the boutique Ethernet cables, get a short one and use a RJ45 back to back connector to a standard Cat5e lead... With the Cat5e running to your switch. Also a ferrite choke around the lead closest to the audio equipment helps impede common mode noise.. Should you be experiencing it.

ethernet is great but it is quite electrically noisy.

Simon

What size diameter ferrite chock is best for ethernet cables?

Do you use the chokes anywhere else in your system?

thanks

John

SJB

Points taken Simon - THANKS

I am somewhat limited & as its planned the router position is fixed & the equipment & IC's will be between 1.5m & 2m away as the airwaves fly

But if there is any peep on the forum who is worried about RFI, its me.  My old job made me this way.  All my cables are shielded & my own made IC's are really minimal lengths - like 28cm

I always intended to use ferrite even tho the AQ cables are cat7 (reputedly) with the pairs individually shielded under an overall shield. 

Mike, you're nothing if not prepared. Personally i'd keep all the noisy stuff elsewhere - my NAS, Mac and Hi-Fi switch sit in the study with 30m or so of Cat 5e cable to the lounge around the outside of the house (with a LAN isolator and choke at the amp end of the Ethernet cable). Decent length patch cables are cheap and it may be worth trying a few combinations to make sure you are happy with the location of items before committing to your final installation.

Enjoy the NDX when you get it 

John, I use chokes designed for RG213 cable. They are slightly loose on the Ethernet cable, so I use a blob of blu-tak in the choke and clamp it round the Ethernet patch lead.. Appears to work well.

i also use chokes on my DC1 SPDIF lead between NDX and NDAC.

Mike I agree with James, you sound well prepared.

James your set up sound pretty good. There are potential advantages in having long  Ethernet leads like 30m, as there will be a greater voltage drop at the encoding frequencies used (100BASE-TX is 31.25 MHz), and so the energy from the send pairs will have reduced, interestingly I suspect Cat5e will perform 'better' here than Cat6a or Cat7. in addition any source initiated common mode noise, especially if high frequency will be attenuated.

Simon

That would be true in case of older switches and routers, but most modern equipment adjusts the voltage to cable length and if there's more line resistance the switch will simply increase the voltage on a given port, so I don't think the effect you describe would work...

Are you sure? Which RFC describes this? i was aware of some non standard proprietary methods but not a standard method - but I could be wrong..

802.3ab specifies the send differential voltages should be +-2V, +-1V and 0V so as to conform to 1000Base-TX Ethernet and +-1V and 0V for 100Base-TX Ethernet.

Simon

I'm not a networking specialist, but I was recently shopping for a little gigabit switch and both models I was checking out on the internet had a feature of adjusting the power consumption per port depending on the length of cable connected.

I realize ethernet has strict voltage specifications, but if there's a voltage drop on the port caused by long line's resistance, the switch could increase slightly its voltage output on that port to compensate, thus ending up with the regulatory +/-1V.

That's how I understand it, but I could be completely wrong.

Gajdzin, the low power ethernet standard 802.3az reduces power on links, but this is about reducing the power spent to maintain the link  when there is no data going across it. But from what I can see from the specs the link returns to normal when data is present. The gap between these bursts can be small - such as 100mS where the link drops to low power mode. Therefore there would appear to be another frequency as a possible source of EMI - Therefore 802.3az could conceivably make RFI worse, albeit you are saving in the home environment a small amount of power. In a data centre these savings could be immense however...

Simon

I agree, savings would be negligible in a home environment, but the standard you describe could be something else from what I mentioned. For example, see this switch: http://www.dlink.com/us/en/hom...gabit-desktop-switch. The Overview tab says: "It automatically powers down ports that have no link, allowing the switch to save substantial amounts of power by cutting power usage for unused ports or any ports connected to computers that have been shut down." Now that is probably an aspect of 802.3az implementation. But then it goes on: "It can also detect connected cable lengths, and adjusts power usage accordingly, helping you save energy without affecting networking performance." Now that sounds like something completely different And if I ask myself how can "adjusting power usage depending on cable length" be practically implemented, I can't think of any other adjustment than voltage...

Sometime this week or over next weekend I am listening to a few Ethernet cable on my buddies Linn DS (?)  

There will be four of us & we have a selection of 3m or 5m length cables from standard freebee Cat 5 stuff as the "control reference" & AudioQuest Pearl & Cinnamon & Supra C7 - all Cat7.  We hope to have Chord, but not sure if its the STA or the basic wire.

More later 

Mike, I would avoid Cat5 which is generally considered obsolete now and use Cat5e which is specced for 100BaseTX and 1000BaseTx connectivity.

Simon

Thats a given Simon, we are using what we think is Cat5 as a control reference (its a BT Home Hub freebee,  so you should know)

The real purpose of this test is to settle an argument between 2 of the people each claiming "mine is better than yours"  = AQ Cinnamon beats Supra & visa versa, with another guy bring AQ Pearl into the mix.

I'm tagging along as a very interested observer.  My quality minded head says go for the AQ as it oozes quality, whereas the Supra is just pug ugly & very utilitarian.   But my sensible head says the cables will not be seen in my set up & the Supra is reputed to be +/- as good as the Cinnamon & the cost is less than half - 3m = £99 vs £40.

But another question for you - if I chose AQ or Supra my set up will be Cat7 throughout – player/router - router/NAS.  But what are the Cat7 shields actually connected to?  the RJ45 has 8 connectors  - 4xpairs – so nothing for the shield connection except the RJ45 if its a metal body.  But looking at my BT router its obvious its not Cat7 ready as the sockets are all plastic.  So whats the downside using unconnected shield sections ??

Great question, I also want to know - I am having my house re-wired with CAT7 (not only for audio, but for 7 computers, a NAS, etc.) and I want to know - should I connect its shield at both ends of each run, or just one? And which one? And how do I check if my routers/switches have a metal contact for the shield (same concern as Mike-B's above)?

Why not email or phone Audioquest and ask them? Contact details are on their website.

Keith

Mainly because I want a factual answer & I suspect a cable mnft will add liberal doses of marketing hype.  

The fact that the shields are not interconnected is a simple electrical fact & the answer I suspect is also simple. 

I am open minded about differences between cables & am about to find out for myself. I suspect it may be possible to detect diffs between Cat5 & Cat7,  but am looking to be convinced there is a diff between one Cat7 &,another  ,   

Mike, indeed the shield(s) should connect to the grounding pad on the RJ45 connector, and of course the socket should ideally be able to ground these pads/shroud on the RJ45.

if the socket is fully plastic, the yes you are not going to get the full benefit of the shielding.. However I suspect there would still be significant benefit, as the shielding will still reduce any  EMI from the leakages from the twists and will be capacitively coupled to ground..(remember we are not talking DC) Of course you do need to watch out for earth loops with Ethernet shield earthing if both ends are grounded.

Thanks Simon,  you've confirmed what I had theorised.  

In the next few weeks I'll be able to provide some personal ear driven opinion around the cable side of things.  I've got an NDS/555 combo on order which will be connected to my Synology NAS via a gigE NetGear store and forward switch.

NDS to switch is about 8m, from switch to NAS about 1.5m.  I've ordered 8m and 1.5m AQ Cinnamon cables and also have a £10 8m cat 6 cable and similar 1.5m cable to compare against.

So hopefully I will provide some insight into cheap vs. expensive cables both from switch to streamer and switch to NAS.  Just don't ask me to explain WHY, if I hear a difference

Connector, or connectors - as in on both ends? And if one end, which one (in a scenario: NAS -> 1000baseT switch -> 30m of CAT7 -> 1000baseT switch -> Vodka -> ND5XS)?

I just bought those 30 m of CAT7, inexpensive stuff, 0,75 Euro per metre. The plugs were more expensive, at 10 Euro a piece - I bought ones that don't require a crimping tool (incidentally AudioQuest Vodka uses identical plugs). The instructions with the plugs say to connect the CAT7 cable shield on BOTH ends of the cable. But I worry about ground loops - wouldn't it be better on one end, and probably the one that's further away from the network player?

Shielded connectors can be a pain.  its probably safest to only earth one end, unless the shield grounds of your equipment are galvanically or capacitively isolated.

however a shield only connected at one end is more like a faraday cage, and so will be effective a low frequencies like 50 to 60Hz, but at higher RF then i would usually expect it to be grounded at both ends... But you could potentially get low frequency earth loop currents induced.

Therefore it's probably safer overall to use un shielded good quality twisted pair (UTP) for Ethernet!

i do use a shielded lead to my network streamer, but  both ends are floating. It does 'sound' better, but that just as easily could more accurate twists and less twist  leakage / induced non differential  current being created.

Simon