You probably want to stick with shielded Ethernet, regardless of the Cat #
21 responses Add your response
I have very good experiences with the sotm cat7 Ethernet cables, they have a buildin filter: https://www.sotm-audio.com/sotmwp/english/portfolio-item/dcbl-cat7e/ Also using an isolation trafo is useful: https://www.sotm-audio.com/sotmwp/english/portfolio-item/iso-cat7/ and of course powering all components in your digital rig with linear power supplies is a must... |
Why? Many devices don’t have connectors that support the shield connection. Worse, shields connected at both end are prone to creating ground loops. They are designed for high speed communication of digital data, not for ensuring the lowest analog system noise. Shielding one end can work, but still back to the issue of whether your connection supports the shield.
This is another one of those audiophile group thinks (not offence @erik_squires ). People just assume shielding must be better, therefore it must sound better. Shielded at both ends cable was designed for noisy commercial environments, not for your typical home use. The shields are to protect from external noise, in a high noise environment. It was not targeted at home environments. The shield at both ends is a recipe for ground loops where they would not have existed with unshielded cable..
|
Only 1 end of an Ethernet connector needs to be shielded to work, and it’s nearly impossible to cause a ground loop via Ethernet cables. Ethernet signals are inherently isolated and balanced. There’s no galvanic path from them into the devices they connect to (barring PoE). I could see a streamer maker use a grounded plug which causes a loop, but that’s really bad design. |
@erik_squires , sorry that is wrong. Many shielded ethernet cables ARE shielded on both ends, and that shield connects to case ground on both ends. Audio circuits often have capacitive couping to case ground if not a direct connection from the DC ground to case. Surprisingly many linear supplies use a grounded plug, as do most things with metal cases and integrated AC supplies. POE, if compliant to the specification is also galvanically isolated. If you don't know how your shielded cable is built, you are better off in most cases not using a shielded cable because to your point, the data portion of the transmission is galvanically isolated.
|
Oh, so wrong. ALL shielded ethernet cables have shields at both ends. The cable shield is end to end, with the RJ 45 connectors on both sides having a shield connection. That doesn’t mean the equipment maker is forced to use a shielded socket or to ground it. No shield connection on the socket, no ground connection there.
Only if the equipment designer makes it a choice to do so. The shield on an Ethernet connection is not carried by the signal wires inside the jack, but on an outer foil like connector. It’s hard to see, but here:
As a streamer maker, you can either ignore that shield, or keep that shield separated from the rest of the system. You are not obligated to connect this shield at all, or to the equipment case. Don’t confuse the shield on an Ethernet cable with the case and safety ground connection that is made to an AC outlet. They are two different things. There is no safety requirement that a signal cable be case grounded, and in fact lifting or isolating signal grounds whenever possible is good practice. |
Signal grounds very commonly have capacitive connections to case grounds where they exist for noise reasons. In the average home environment, unless you are co-running your Ethernet in long run in parallel with a seriously noisy AC like in industrial/commercial, there really is no need for shielded. There simply are not the noise sources to make a difference. As you noted, it is already galvanically isolated. |
That is entirely up to the equipment manufacturer. I have a hard time believing a high end streamer maker would both ground for the sake of better EMI rejection AND leave themselves open to ground loops. I certainly wouldn't design a mixed digital/analog device like that. |
@erik_squires , I would have more faith in a lower cost, higher volume product from a lesser vendor, that has the proper equipment for testing, and probably more experience, that a boutique vendor who in many cases is winging it. I am not saying that is exclusively the case, but measurements by Stereophile/ASR have shown enough "high end" equipment to have questionable performance, even on the basics, that I have some confidence that is correct. I would expect the larger high end vendors to be designing "properly", at least I would hope that is the case. Fortunately, this is an easy one to test. Measure the resistance/capacitance from the Ethernet jack -shell to the case. |
With the twisted pair, you wound need to be running them pretty close. Packet bursts could demodulate to audible frequencies so there is a mechanism for the noise transfer even if the transmission frequency is high. I suspect this would be not much of an issue at 100M or 1Gig, more at 10M though since those high frequencies would be too high to even be demodulated as bursts at the power level. |
