[edit]I looked a little further into the question I had about Ethernet over coax, esp. RG-11 and RG-59. I didn't find an exact match since there is no single-ended coax PHY for GbE, but I found some video studio studies that were sending 1.5Gb digital data over various cable flavors, and they were quoting on average about 80m for RG59 and about 140m for RG11, I think it was. So based on that I still hold the view that no standard 802.3 flavor will work for the coaxial portion of cable plant, across the board. A slightly modified version may be OK in some cases since those distances are probably fine for some portions of cable plant (but I have no idea what percentage). But those proprietary ways lose (or reduce anyway) the cost-competitive egde of standard 802.3.
For Baseband ethernet, I also have some serious concerns over the amount of noise below 20MHz (which is quite large in some cable plant), and how well standard SE transceivers deal with this w/o interleaving and FEC. Being able to operate in a test deployment, and being able to operate longer term, at a low error rate, are 2 different things. You can have a pretty crummy error rate and high retries on a Gigabit link, and still do an impressive web browsing demo.
So I'm still a skeptic on this one until I get more positive, convincing, info. It's not that I don't think it can't be done, because it absolutely can if cost is no object.
[edit: Darn it! you edited your post on me<gg>. The part you added about the last drop being converted to TP may support my argument about issues with RG11, and differential TP would help immensely for all the external noise picked up along the run. BUT...they've still gotta trench for this] |
