Noise via Ethernet is a real thing, but the funny thing is that most audiophile cables are making the problem worse. The sad thing is that consumers are buying these things.
Twisted-pair Ethernet is a differential (balanced) connection so typical EMI/RFI issues are addressed through common mode rejection. That isn’t to say that Ethernet is immune to this type of noise, but in a domestic application you have to be really haphazard in cable routing to get real issues here.
Twisted-pair Ethernet is galvanically isolated when unshielded cables are used. This is part of the Ethernet spec which mandates transformer coupling (or some other similar air gap isolation) for all PHYs. This is in place to break potential ground loops (which can be deadly in real-world implementations) as well as to avoid current leakage and spikes over Ethernet links.
Twisted-pair Ethernet is an extremely reliable transport mechanism as long as the specifications are followed to the letter. This includes cable geometry, connector design, and termination. Ignore any one of these and the quality of transmission (and the noise rejection properties of the transport suffer).
Here’s how audiophile cables tend to screw up Ethernet:
-
“Creative” cable geometries which don’t follow the spec and ultimately result in noise issues and signal reflections that do impact data transfer performance.
-
Non-standard (answer to the question no one ever asked) connectors which look impressive but have issues with maintaining a reliable connection, impedance mismatches, and a higher likelihood of damaging the component they’re plugged into.
-
Use of shielding without understanding how Ehternet works. With a fully connected shield you’ve now linked the ground planes of the stereo with the rest of the network. If noise is a concern the this is just stupid. With the shield floating or only connected at one end you now have a very effective antenna connected to the ground plane.
-
Product designs which are make to look and feel impressive in order to justify ridiculous costs. This includes cables that are so heavy they literally rip the Ethernet jacks off circuit boards, connectors which tend to wallow out the jacks on the components, and cables which have no chance if having proper geometry unless they’re kept perfectly straight.
Fun fact… most audiphile Ethernet cables don’t pass the certification for their advertised category (i.e. some audiophile CAT7 cables are lucky to pass CAT5 requirements). Many audiophile Ethernet cables don’t certify remotely close to any of the Ethernet standards… yet for some reason they cost hundreds or thousands of dollars.
I’ve talked to a lot of audio cable manufacturers about Ethernet and the interesting thing is that none of them really understand how it works. They’re just trying to apply audio signal cable principles to a very different kind of connection. The good companies will admit that they don’t have the knowledge they need and hence their product offerings appear to be rather limited in scope (this list is very short).
Many report hearing sonic differences between different Ethernet cables and use this as a justification for purchase. If you’re happy with what you’re buying and feel that a cable provides a positive sonic change then there’s nothing wrong with that. It’s your money and system so if you’re happy then that’s great. Just don’t get caught up in the hype about noise reduction and magic dust. There’s a really good chance that the difference you’re hearing isn’t the result of noise reduction, just a change in the overall shape of the noise floor in your system.
Some notes here:
-
Ethernet is so ubiquitous and important that there are legions of very accomplished engineers constantly improving the specs and requirements. This work is in support of the real-world applications of Ethernet which are far more important than anything audio-related. The thought that the audio industry is going to make real improvements here when most companies lack the equipment to do proper measurements is beyond laughable.
-
This isn’t my opinion, it’s fact and based on actual understanding of IEEE 802.3 along with decades of experience in large scale data networking.
-
dCS has no official recommendations for network cables as long as they meet the Ethernet spec, don’t damage the component, and certify at their listed category. When customers report issues that are beyond basic configuration problems the first thing we request is the replacement of all audiophile Ethernet cables with something less exotic (and more correct). You’d be surprised how often this resolves the problem.
-
I have absolutely nothing against audiophile cables or the companies that make them. I do have an issue with products that are designed and marketed with an obvios ignorance toward their intended application.
We put a lot of effort into proper implementation of our Ethernet interfaces and can demonstrate excellent measured performance. We don’t violate standards so our Ethernet port shields are connected to the component’s ground plane. We go to great lengths to maintain a clean ground (which is especially critical in digital audio), but we can’t do anything when someone connects an antenna (poorly designed cable) to the Ethernet port.