Somehow I missed the remainder of Agaton's thread and since it's locked I'm opening it here. Maybe it belongs in Cable Talk though.
First things first: forget about what protocols are involved. Same goes for switches. It doesn't matter what the audio transport protocol is, be it TCP/IP, UDP or anything else. It's all to do with layer 1, not layer 3 where most people fixate about when scoffing about audiophile Ethernet gear.
Secondly: most audio streamers run at 100Mbps. Some more integrated models that need to do large transfers for firmware updates or video etc. may run at 1000Mbps but it's rarely seen on audio streamers. CAT6A is already good enough for 10Gbps @ 100m, let alone CAT7/8. So why fixate on cables that are optimized to cover an entirely different spectrum?
Test using non-audiophile cables have been done extensive overseas where you can order a long list of cables, delivered the same day and returned the next free of charge. The results, and I followed a number of them, are clear: 'cheap' CAT7/8 cables are among the worst you can do. If cheaping out, good old CAT5e does the best. And among them, often the best results are the ones with a flat ribbon-style construction, since it reduces crosstalk. Don't just discount some boring cable just because it looks (and is) cheap and only rated CAT5e.
When going up the ladder to the Furutechs et al I am not going to comment yet (although it is part of a collaboration project where measurements are ongoing), but they do have merit and will outperform the contenders. Which is best, there opinions are divided but the interwebs is full of comments. Sadly the pricing quickly gets astronomical.
Shielding, often a shield both sides can make things worse, depending on the system grounding. A common solution that is often effective is to cut the shield one side and properly earth the switch instead. Might be problematic with warranties of course.
Ferrites: a single clip-on ferrite tends to yield unpredictable response. Part of the problem is because data on the ferrites tend to be lacking, but generally good results have been had to rather apply an even amount throughout the cable length, not just in one or two places. The following picture shall illustrate.
Lastly though: results seem to favour a simpler, albeit less practical, solution. Only two bundles of the four (since only 100BASE-TX are used anyway), independent, using individual 100Ohm shielded twinax from aerospace materials. This cable below yields excellent results for example.
First things first: forget about what protocols are involved. Same goes for switches. It doesn't matter what the audio transport protocol is, be it TCP/IP, UDP or anything else. It's all to do with layer 1, not layer 3 where most people fixate about when scoffing about audiophile Ethernet gear.
Secondly: most audio streamers run at 100Mbps. Some more integrated models that need to do large transfers for firmware updates or video etc. may run at 1000Mbps but it's rarely seen on audio streamers. CAT6A is already good enough for 10Gbps @ 100m, let alone CAT7/8. So why fixate on cables that are optimized to cover an entirely different spectrum?
Test using non-audiophile cables have been done extensive overseas where you can order a long list of cables, delivered the same day and returned the next free of charge. The results, and I followed a number of them, are clear: 'cheap' CAT7/8 cables are among the worst you can do. If cheaping out, good old CAT5e does the best. And among them, often the best results are the ones with a flat ribbon-style construction, since it reduces crosstalk. Don't just discount some boring cable just because it looks (and is) cheap and only rated CAT5e.
When going up the ladder to the Furutechs et al I am not going to comment yet (although it is part of a collaboration project where measurements are ongoing), but they do have merit and will outperform the contenders. Which is best, there opinions are divided but the interwebs is full of comments. Sadly the pricing quickly gets astronomical.
Shielding, often a shield both sides can make things worse, depending on the system grounding. A common solution that is often effective is to cut the shield one side and properly earth the switch instead. Might be problematic with warranties of course.
Ferrites: a single clip-on ferrite tends to yield unpredictable response. Part of the problem is because data on the ferrites tend to be lacking, but generally good results have been had to rather apply an even amount throughout the cable length, not just in one or two places. The following picture shall illustrate.
Lastly though: results seem to favour a simpler, albeit less practical, solution. Only two bundles of the four (since only 100BASE-TX are used anyway), independent, using individual 100Ohm shielded twinax from aerospace materials. This cable below yields excellent results for example.
