Valley of "audiophile" ethernet cables

(Sean) #388

I’m definitely not at all going to claim (i.e. guess) that the following is the root cause of your observations but ethernet’s little transformers block lots of 'the bad stuf’f but not all.

According to John Swenson, ‘high impedance leakage currents sail right through the ethernet transformers’.

I’ve posted links way up in this thread already but I’m guessing no ones read them lol, I’ll share again below. John’s said he’s needed to build custom test gear to measure this stuff…



Any reference to microRendu or ultraRendu can be substituted with any networked endpoint/renderer, not just those.

Fortunately he shares a relatively cheap trick to block the leakage currents from getting into the ethernet endpoint (therefore getting into the DAC).


Comparision is not so hard.

  1. We all agree ethernet cables transports DIGITAL data
  2. Audio is not easy to compare, because we cannot hold 2 streams side by side and search for differences
  3. Easy solution: change the audio to pictures and compare them
  4. did the picture with your new cable has more depth, more texture or look the colors like a curtain was removed?

Sounds absurd?
Why is it absurd to expect such result for pictures or maybe text, but for audio it’s not absurd?

(Sean) #390

But here, are you making the assumption that none of this stuff affects picture quality?

Is it possible that some (not all) of us here in this particular hobby of listening to music, care less about small differences in picture quality, than sound quality? So maybe picture quality is affected by some of this stuff but we don’t care? And therefore we don’t notice?

I ask these questions respectfully of course. I’m not an expert in imaging (or even audio) but I try to be careful with assumptions and generalisations.


Yes, if I load a picture from my NAS to my laptop, not one pixel is altered. Indepent of the transport way.
The whole internet would be a useless crap, if ethernet connection would not be able to transmit digital data unaltered.
Believe me, each error in my text here is from me and not from a bad cable.

(Sean) #392

But here you’re going into the bit perfect argument. I thought we established much higher up in this thread that none of this discussion relates to bits being altered…

At least that’s what I’d hope.

This entire discussion is (or should be) around assuming bit perfect playback…

Check out the links to John Swenson’s posts I posted (if you’re interested). It’s got nothing to do with bits being altered and data centres around the world collapsing.

It may be more to do with leakage currents and RF interference (both possibly linked)…

It comes up every month or two but can we please never discuss dropped or altered bits and the internet collapsing ever again :grin:

(Andrew J Shepherd) #393

Correct. Flipping digital bits would require a whole lot of noise and interference.

But if this were noise and interference instead propagating downstream and affecting the analog stages, then that would be measurable at the analog outputs.

So, somebody please produce analog noise floor measurements showing how these leakage currents and/or RF interference can translate into audio band noise. If true, that should be an easy task.



But you can compare them side by side i did so in my experiment by comparing the waveforms of different streams played out of my system. Any difference however small would be visible in the waveforms and I did zoom in incredibly far as to compare them, also to I was be able to play them all at the same time and solo anything track in the software this really helps to hear any difference how ever subtle. If you can do this @zoom25 then it might help see.of there is something. Now I did not test Roon in my system only upnp and usb playback so maybe RAAT is causing something it does transmit it’s data differently as its all in small chunks rather than one big one I believe . If I can get Roon to be at the same level as the rest of my system then I can retry but it the moment there is a slight volume differnee between local ,upnp and Roon so difficult to compare maybe this is part of the same problem. Can you also try upnp and see if you get same audible difference.

(Sean) #395

Hey Andrew, what DAC do you own?

What if your own DAC’s designer (if you asked them) agreed with these potential mechanisms, i.e. leakage currents and RFI potentially affecting SQ but they didn’t have the measurements to yet show this YET? Does that mean you’d lose respect for your DAC’s designer?

It might be worth everyone asking their own DAC designer what they think about John Swenson’s posts I linked. I’ve done that with a few (well known) DAC designers and they don’t disagree with John…

Then you can take up the arguments directly with your DAC designer if you disagree with them :slight_smile:

This is assuming you have a remote interest. Some people don’t care and that’s fine too! You don’t need to pay attention to any of this stuff to enjoy music. I just find it interesting.

(Anders Vinberg) #396

All networking sends small packets.
With uncoordinated timing.
Possibly out of order.
With possibility of packet loss, recovered by retry, with random timing.


I really wish the Naim forum was as open to this topic as it is here. Ive tried to participate there with my test and evidence but it mainly falls on deaf ears. No pun intended. It’s one way or another and never the twain shall meet.

(Anders Vinberg) #398

If I buy a $10,000 DAC and it improves “dramatically, like removing a veil” by the insertion of a $50 cable cleanup gadget? Yes, I would lose respect for that engineer.


So the way RAAT works is no different then and in theory make no difference. It’s definitely a busy protocol as they do say so themselves in the how-to set it up.

(Anders Vinberg) #400

There is a protocol stack.
Commonly numbered 1-7.
At the bottom you have Ethernet, and then IP, and the TCP, and so on, eventually you get application level protocols like RAAT.

IP is the norm.
But above that you can choose TCP, which adds a lot of reliability and order so you can relax, or you can choose UDP which does not and then the higher level protocol has take on that responsibility. Both approaches work, and in fact Roon has used both, switched from on to the other. So have I in different contexts. Advantages and disadvantages in both.

But the point is, eventually, at the top, you always have reliable and accurate delivery, and you never have reliable timing (so we can’t talk about jitter because jitter is deviation from correct timing and there is no claim for timing).


Hi Simon,

I don’t have Upnp with the BDP-1, otherwise I’d give it a try as well. The BDP-1 can be a DLNA renderer. I tried it once and sending it by Audirvana Plus. It gave me all kinds of problems with synchronization. It was unusable for both listening and testing. RAAT on the other hand is very responsive.


Ok would have been interesting to see if there was any difference in your setup. I’ll have to borrow the recorder again and see if I can compare the two. I did think I could hear a difference between native upnp and native Tidal comaped to Roon Ready with the Atom when I got it. I don’t notice this now or my ears/brain choose to ignore it. I also notice that my kit always sounds a little more focused after a few drops or two of booze so go figure. The brain is an odd thing.


True. But you are looking at only the digital domain here. The data is bit perfect in the digital domain. Music is converted to analog and the spurious electronic noise can affect the analog signal the DAC outputs. It’s that simple. So, as far as I am concerned, expensive Ethernet cables are a complete waste of money. The key is to make sure that as little electrical noise as possible makes it to the last Ethernet port in the chain. What happens before that is not really important.


RAAT is probably layer 5, the session layer.


For some it very well may be an RF issue where the cable is picking up noise or radiating it. If that’s culprit, then they should try to test for that and see if unplugging the cable changes the sound. For me it didn’t do anything. I tested it with the BDP-1 in the rack and 10+ feet away from the rack. Who knows, maybe in other’s case that noise is actually a problem. However, once you can eliminate that variable, and you still notice difference in sound, then you have to look elsewhere.

These Cat6 (especially the Belden bonded) cables are immune to 30 Mhz. On top of that, RF of that level is typically not a problem at homes. So how can cables still make a difference? Why does a linear power supply improve the sound? We know that they all get the bits across. No dropouts with any of them. I have a theory.

Remember signal integrity from way back? It might be that. We know there are no dropouts and bits are being turned into 0 and 1’s correctly. Has anyone looked at whether it’s easier or harder in some cases than others.

People try buying all kinds of cables with this shielding or that, fancy materials and this or that. Has anyone of them ever tried to control for length?

There might be something happening with the short cables in particular. I picked up a bunch more stuff from the data centre recently. All Fluke tested. I have multiple Cat6 UTP cables of varying length 1 feet, 2 feet, 3 feet, 1.25 m, 3 meters, 5 meters, 20+ meters.

They all work! Not one ever gives me a dropout. However, subjectively, it’s the short ones that sound different. As you start going long, differences quickly start going away.

I’ve read a few times on other network related forums (that have nothing to do with audio) and people troubleshooting for data centres that they generally recommend avoiding short lengths. Some insist that yes you can still make the short ones work, but the longer are better. I’ve read reflections as an issue.

If you have a switch, connect only the router to it along with the computer and network player to it. Put music on a USB drive and connect it to the computer so you can avoid a NAS and another cable. Use local playback and not from online source so the only cables involved in the path are the ones linking the computer and network player/DAC. I use a dummy router that’s not connected to net. So it’s there just to make things connect to each other. The port remains idle. If you want, you can even unplug it.

So for those two cables that are active, try varying the length. I know some people here won’t be able to vary the length of the cable from the computer to switch as the computer is in another room. In that case, you can try and change the length of the cable from the switch to the network player.

There are some other factors to mention as well which I’ll get to later. I have to leave soon.


Can you tell us which one you think is quieter or louder? If I had to guess Roon is the quieter one? Local is louder?

(Sean) #407

Sigh. No one is claiming the internet is collapsing with dropped bits and such dramatic night and day differences here.

These discussions with your own DAC’s designers are very worthwhile (I found). Especially the RF and leakage current mechanisms that John S discusses in my shared links.