Originally Posted by mgkdragn
Kurt.. I've been wondering why some outfit can't come up with a RJ-45 to HDMI adapter.. thus allowing the use of the 1.4 Ethernet channel now.
Unfortunately, it wouldn't work. The Ethernet channel is a hideous, hideous kludge and it doesn't map to an RJ-45. Here's the deal.
There are three wires involved, which are the "plus," "minus," and "shield" for the Ethernet pair. These wires already existed in the old HDMI bundle, but had different functions. One, pin 17, was a ground for some low-speed data circuits and is now the Ethernet shield. Pin 14 was "reserved" for future use, and is now the "plus" side of the pair, while pin 19 was the "hot plug detect" pin and now has the "minus" side of the Ethernet pair.
A normal CAT5 type cable has two pairs in use, one to send and one to receive. But here there is only one pair, and the pair also has to carry the hot plug detect function, the grounds for other circuits, and the Audio Return Channel. So, to get all of this onto the one pair, what they've done is this:
(1) The Audio Return Channel will, in any application where Ethernet is active, run in common mode--that is, the signal will be sent simultaneously on pins 14 and 19, in the same phase. If this works right, the Ethernet circuit won't "see" the ARC signal because it's in common mode and will be rejected in the same way that common mode noise is.
(2) Since there is no "send" pair and "receive" pair, sending and receiving will be done simultaneously, both in differential mode, on the same pair. In order to not get confused, the circuit will not only be trying to keep out the common mode ARC signal, but also will subtract its own output signal from what it measures at the input.
This is a mess. The problem starts with intrapair skew. A tiny bit of skew makes it very hard to keep these signals from messing with each other. Return loss can make it worse--anything that reflects off of the destination and comes back to the source now, instead of just being a little edge lost on a bit transition, becomes noise interfering with the signal running the other way on the same cable. Will it work? I dunno. Probably, most of the time, but there are sure a lot of ways for it to fail.
It gets more complicated than this, but that's the gist of it. If I didn't want it to work, this is how I'd design it. My suspicion is that if you want RELIABLE Ethernet between devices, sticking with conventional and cheap RJ-45 patch cords is going to be the way to go.
Anyhow, though, to revert to your original query: as you can see, it couldn't be mapped to an RJ-45 by just using an adapter, because the send and receive are run on the same pair and are mixed with other things. If you did want to adapt to a conventional Ethernet cable, you'd need an active box to try to sift one signal from another.
KurtBlue Jeans Cable