> OK, I was pointed at http://www.linksys.com looks like they've got some
> good info. However, I just noticed something that has me wondering, looks
> like CAT5 only has 4 real wires made out of the 8 wires.
"Normal" CAT5 cable has four metallic pairs. 10-Base-T and 100-Base-TX
use only two of the pairs. (And not the two that you might expect.)
Other physical layers (including 1000-base-Tsomething) will use four pairs.
I generally recommend against trying to be cheap by putting in the minimal
required number of pairs.
> So... Is it possible to run at least 10BaseT over normal phoneline cable?
> It just happens that a totally unused phoneline was run a couple years ago
> at the same time I had a work phone put in.
I've successfully run 10-base-T on two spare pairs of prewire in my condo,
which probably doesn't even meet Category 2 specs.
The phone wiring for the whole place is a daisy chain, except for one branch
to the kitchen. The normal 10-base-T rule of thumb is that no two nodes on
the same network should be separated by more than four repeaters. (An
Ethernet hub is a multiport repeater.) Beyond that you're supposed to use
bridges or routers. (An Ethernet switch is really a bridge.)
This limit is required for several reasons, the most obvious of which is
the effect of propogation delay on collision detection.
However, I've gotten away with five hubs separation between some of my
nodes. I haven't seen any reliability problems. I suspect that I'm within
the timing requirements because my cable segments are much shorter than
the specified maximum.
What you *can't* get away with, though, is running 10-base-T over
flat "modular" cable. It really does have to be twisted pair.
Also, note that the limits for 100-base-TX are much more stringent. You
can't have more than two repeaters between nodes. Fortunately small
10/100 switches are getting cheap.
Eric
Received on Wed Oct 27 1999 - 16:44:14 BST