Originally Posted by Ronnie
A summary of the above post is

a) the phenomena you describe above (including Skin Effect) have been well understood regarding data transmission many years ago

b) despite many trials and material combinations, nobody was ever able to come up with a material which would overcome the physical phenomena you describe

c) as a result, the Ethernet PROTOCOL was designed (and has been refined many times since to overcome these physical limitations

d) it is the Protocol that ensures that correct data is transmitted and received.....data that is "very similar" is eliminated, with 'blanks' left.....so that the receiver easily knows that the data is corrupt and missing data

Hi Ronnie,

Yes, I fully get what you're saying on Ethernet and error correction - don't worry. I read up on CRC error correction and it's pretty hairy-arsed for sure.

Not sure I made it clear that I had no intention of making specific assertions or recommendations about what cable to buy in which part of the system. I won't be buying any fancy cable between MD600 & 808v5, as the run is 20m and it would cost literally thousands.

My aim in the original post was just to throw in a bit more engineering-led thinking to sit alongside the obviously very deep well of IT-led knowledge that we have on this forum. The point was simply that, at the physical level (i.e. before error correction etc.), digital cables are affected to some extent by electrical, magnetic and mechanical influences that affect analogue cables and equipment.

BTW. Just to set the record straight, mercifully, the cable engineers did come up with adequate and cost-effective designs to account for skin effect in Ethernet cables - hence current ability to run Gigabit connections on cables costing a few quid. There�s the separate matter of the extent to which a problem is contained or reduced, so there are clearly more expensive designs that go further on this front, whatever one�s view of the value. It�s worth a quick search on Google to see what�s been done with successive generations of Ethernet cable design (purely in the IT world), not just around skin effect, but impressive tricks to minimise cross-talk as well to get data rates and bandwidth up.

Anyway, I have a question that�s bruising my brain a bit. I buy what you�re saying about the 100% infallibility of error correction over Ethernet, but this leaves me wondering about two things:
1. How does adding a GISO at the source end of an Ethernet run produce such a big improvement in sound for four contributors on this thread? Ralf told me that nearly all the gain is down to filtering out RF noise, so I�m wondering how a reduction in noise can help when TCP/IP is doing its thing. Not challenging here � just wondering where and why the gain is made. Could it be that the noise otherwise gets beyond the ID4x card, causing other damage to sound downline?
2. The ID41 card is widely reported to be a big step up from the ID41 and I�ve seen comments to the effect that this is largely down to better handling of noise. So again, given the reliability of Ethernet error correction, is the gain from noise reduction actually realised further downline?

Dave

808v5, MD600 & DSP7200.1
Dedicated audio mains supply