Ethernet Cables - Yes I'm going there... I was a skeptic and I was wrong


#1

I am running an UltraRendu powered by an Sbooster LPS. Software Audirvana + converting PCM to DSD128.

Ethernet path is wireless from Macbook to Modem/Router to FMC pair mod’d by SoTm. FMC’s are both powered by Jameco 5V linear wall warts.

I demo’d three pairs of Ethernet cables from cable company.

Wireworld Starlight Cat 8.

Audioquest Vodka Cat 7

Cardas Cat 7

All were compared to the stock cable I was using, Belden 6a.

I started out yesterday with the Wireworld without even listening to the stock. I cranked the SGCD to 35 resulting in about 88db sound levels. Normally I don’t crank the system that loud but it sounded really good. Didn’t know if it was the cable or maybe i was in the mood for loud.

Then I changed to the Vodka. Well… at 35 I walked away and within 5 seconds I was rushing back to the SGCD to turn it down 8 or 10 notches. There was a very brittle sound character that was not musical at all. Very bad sounding… i could not imaging listening to this system long term like this.

Switch back to Belden… the harshness is gone from the Vodka but much of the body I had perceived with the Wireworld wasn’t there. I wasn’t sure if the WW was better or if the AQ Vodka just screwed up my perception by being so bad sounding.

Switch to Cardas… not abrasive like the Vodka but again a little too much high frequency emphasis with lack of body.

Back and forth and back and forth… I would rank the Wireworld as a distant winner, this was a night and day difference that no one can miss. Smooth highs, nice body, a new full midrange that was previously lacking, and a live sound! 2nd is a draw between the $20 Belden and the $365 Cardas… these are different but I’m not sure which is better. Last is the Vodka which was a music killer for me… analytical way too agressive with the highs… sibilance off the charts.

I don’t know why the WW sounds so different. It could be the Cat8 but I’m also suspecting the non-twisted geometry and use of triple shielding. They say it reduces skew in the signal. Who knows, I just know this is the equivalent as a power cable change or greater to my ears in my system going with the Wireworld. The difference between the Cardas and Belden was maybe much more subtle… and the AQ was awful which I was not expecting.


#2

Interesting. I have two runs of AQ Vodka (NAS to switch and switch to Bridge II) and I don’t hear a harsh or brittle sound at all – and I am more sensitive than some people to aggressive treble. I wonder if the AQ was not broken it; this does matter, as I know from experience. I suppose the cables that you get on loan are supposedly ready to go, but maybe this one had never really been listened to, at least not in a long time.


#3

@magister thanks for the reply.

I think the deal with ethernet opens up a big rabbit hole with many fine details. It has been assumed that Ethernet is less prone to jitter, power supplies in the ethernet devices, and that the isolation transformers are immune to upstream noise. All of these notions are breaking down over time.

Take my system, we have a modem router, 2 FMC’s and an UltraRendu. The FMC’s were modifed by SoTm changing out the switching regulators to low noise linear and changing out capacitors to what they consider better sounding ones. This made a dramatic effect in my system even when the Belden 6a cable was installed. In addition, power supplies on both of those FMC’s make a difference. The cheap stock SMPS sounded kinda bad and mechanical compared to these Jameco Linear wall warts that I use. I highly recommend these for $10 a piece.

My point is that I have upgraded multiple ethernet components in their power sections which changed the noise profile of these devices. This may have had effects on the jitter of these devices as well. I am actually asking alot out of my system. I am upconverting to DSD on the fly in Audirvana, sending that wirelessly to a modem/router, going to an FMC which has to turn to optical, then another FMC going from optical back to ethernet, and then to the UltraRendu. BTW the UltraRendu does not have a huge buffer for these DSD applications… only about 6 seconds when the Ethernet cable is unplugged.

It is quite possible that the Vodka would fit better with the stock FMC’s with their different noise/jitter profile. In your case, your buffering capacity and power supply for bridge are different than my Sbooster external LPS for my Ultrarendu. John Swenson over at CA has been testing switches and ethernet devices and has discovered that different switches and different power supplies on those switches change the leakage currents and how much noise leaks through the ethernet line to the downstream device.


#4

Thanks for the followup. I completely accept that ethernet connections are not the simple things that people used to think, in terms of noise etc. In fact I have LPSs on my NAS and network switches.

What surprised me was how much worse the AQ sounded. If it had come in third by a small margin I would have put that down to system matching and thought no more about it. (I have no doubt that you are reporting accurately what you heard!) Clearly there’s something in your system that benefits from what the WW does and interacts negatively with the AQ, even though the latter can sound very good. Something about the cat8 construction of the WW cable?? Maybe someday we will understand how these things work. Or not.


#5

I was surprised too. Perhaps AQ doesn’t like my Synergistic Core speaker cables and interconnects… or that combination with the AC-5 power cords. Or maybe the Wireworld Silver Starlight USB doesn’t mesh with AQ? Maybe I’ve created the perfect storm against an AQ cable? Who knows.

The WW sounds like Synergy to me and the AQ is the complete opposite. I don’t believe they have to be extreme opposites in all cases. You could try the WW and find either a minor benefit or loss for all we know.


#6

Did the Cable Company confirm that the three burned in before your test? HDMI cables need burnin too.


#7

Some maybe remember a few years (maybe 5 or 6) back. We started this discussions on the forum. Some were skeptics and some were believers. Today i think most of us are believers. I have tried many many cables in the past and found out several differences in the sound with ethernet cabletypes. Even after 5 or 6 years I never found a cable that I realy liked over time. Al the cables had their own signature and brought something good and something less good to the sound, but they were never realy neutral to the sound. Today I think expensive ethernet cables are old news. Please don’t loose me here… the ethernet protocol is the only protocol that I know, that is not charged with a clock signal. The way to realy solve your trouble with the influence of noise to the signal is a galvanic isolation plug connected directly into your dac. The signal was already proven to be bit perfect, there’s no clock signal in the stream so there’s only noise comming from computers etc that has to be stopt and taken out of the signal. The result is perfect and I have never doubted the signal again.

Galvanic isolation directly in your dac is the way to solve ethernet noise. Galvanic isolation on other protocols like spdif (coax, AES-EBU), i2s, usb etc is great but with these protocols cabling is still important due to the clock signal, but again not for ethernet.


#8
the ethernet protocol is the only protocol that I know, that is not charged with a clock signal.
Maybe I am misinterpreting the comment here but ethernet data transmission does rely on clocking. And with clocks, come jitter and ethernet is not immune to jitter. You will find an oscillator on every network interface card as well as in any router/switch. Whether or not jitter degrades the signal to the point of affecting sound quality is another discussion, but it is present.

I do agree with your comments on galvanic isolation and feel that addressing it has had a positive influence on sound quality in my system. I replaced my standard off-the-shelf switch with an AQVOX AQ-Switch-SE and without doubt, noticed the improvement.


#9

Hi Jeff,

You are right, actualy there are clocks everywere in computers and networks. But they are not yet dealing with an audiofile but just with data. The timing for the audio file is for later concern. The data is indeed done by several clocks on the mainboard, the ethernet adapter etc. The data must first arive in the dac from there the dac must see if the data is ready to play or else it will be converted to playable data and from there the audioclock comes in sight.

This is different for other protocols for example spdif or I2s is dealing with the audiofile and the timing of the audiofile, from there it is transported to the dac. For these protocols you will still need good cabling besides galvanic isolation.

I’m certainly not an expert, but an experienced hobbyist. So I hope this helps!


#10

Some thoughts:

The real clock issue is which end is in charge of the data rate or is controlling how fast data is being consumed. USB and Ethernet (and technically I2S) are the bidirectional connections and both (all three) support the DAC being in charge of how fast the data is consumed and hence support the DAC having the master clock. USB supports either end being in charge but with most DACs having the DAC in charge is noticeably better - that’s a part of what Asynchronous USB Audio is about. Tho I2S is defined to support either end providing the master clock, but the master clock isn’t typically used that way for audiophile box to box connections.

I’d claim that, at least with the DS DACs, that these issues aren’t really relevant to the final sound of the DAC. The real issue appears to be dealing with noise, ground loops and potentially interference via power. Technically both USB and Ethernet can be providing power, but since most DACs don’t need this power and it’s not required in Ethernet that needn’t be a problem.

How various cables are grounded and if shielded, how the shield is grounded can affect both possible groundloops and how much noise is possibly transmitted from the source to the DAC, radiated to the rest of the system or received from the environment.

At least with the DS DAC, somewhat ironically people tend to forget that TOSLink wins big time on these theoretical issues, it only looses in terms of available bandwidth. Also FWIW the simpler protocols, S/PDIF and AES/EBU are easier to isolate than the bidirectional protocols (Ethernet and USB) They are also a little limited in terms of bandwidth compared to Ethernet and USB, but this isn’t all bad - the more bandwidth the more radiation for the rest of the audio system to deal with.

In the practical world all of this doesn’t really matter much - people should use whichever connection they find convenient (or believe in if they intend to have fun optimizing.) They all can be used to get great sound.


#11

Once again, thanks to Ted, I feel smarter for having read his post. At least until I forget what he said, which will be in approximately 20 minutes!


#12

I get regular e-mails from Blue Coast, which I assume others here do too, and the one I received today had a short piece on Ethernet cables. Bottom line, Cookie Marenco noticed a difference. You can read about it here


#13

I don’t know how a high-end ethernet cable improves the sound. I use “low-end” Audioquest ethernet cables, but have a fiber optical bridge between the router and Melco N1A server (having I guess more than 50 feet builder-grade ethernet cable between them in the wall), which definitely improves the situation, e.g., lower noise. Did anyone compare the “low-end” ethernet cable with fiber optical bridge vs high-end ethernet cable?


#14

Still a work in progress, but I have been listening with Ghent Audio Ethernet cables that utilize the “JSSG” shielding technique to address RF.

In combination with the “JSGT” SMPS negative shunt ground technique for high impedance AC leakage.

Along with insertion of Net Gear FS105 switches to deal with low impedance AC leakage.

All FMCs and isolation devices have been removed and not missed, I feel that even with the best power supply available, the FMCs had a negative influence.

This is after all of the previously mentioned techniques.

Previously, except for the ethernet connection to the WiFi bridge, all of my ethernet was via FMC.

With Blue Jeans unshielded Ethernet, the RFI emitted by the FMCs was easily noticeable as it was picked up by the cables.

Currently I am listening to the best sound that I have ever had.


#15

If it’s any help - my experience from playing in this area is that:

  • I got immense improvement from upgrading the wall wort on the modem/router combo

  • The SOtM DBL-CAT7 cable was the best sounding of the bunch in a test my buddy did on my system. He brought over a bunch of cables he was reviewing and that came out on top.

  • Incrementally, their ISO-CAT6 LAN Isolator had incremental improvements too but was not as big of a jump as just having their ethernet cables in itself.

  • The Supra Cat8 was the best value winner at about $50/cable.

Siao


#16

Hi Siao,

I’m curious about what wall wort you purchased for your modem and router? Also for what modem and router?

I’ve got an Audioquest Vodka and Supra Cat8 that I will be auditioning as soon as my DSJ arrives.

Shahram


#17

I got the SOtM SPS-500 for my TP-Link modem router CR1900. Hope that clarifies! I played around with the power cable to the SPS and got even more gain with it. Currently using a Vovox textura cable for that too.


#18

How close is the modem router to your stuff?


#19

They’re all pretty close…


#20

Even when my modem/router wireless was turned off it was polluting everything with a harsh character.

Replaced with a modified switch from SoTM with all linear internals. That itself was transformative. I reconfigured my network and moved the modem router two rooms down.