Thanks for the comments so far, generally in-line with my statements of
the obvious. Perhaps I should reiterate that the co-ax is already
installed and, because of its location, is **very** expensive to
replace.
So I am hoping for a solution that will provide best speed point to
point tcp/ip (not necessarily Ethernet but that seems the obvious
choice). 10Base2/5 achieves this using collision detection and
half-duplex. This may suggest a solution. If I can configure a
10/100BaseT interface to operate half-duplex and use a balun device to
merge rx & tx onto a single co-ax it might appear to the interface that
it effectively had a 10Base2 co-ax connection. Far fetched or possible?
Sounds pretty far-fetched, but I don't know that much about the signalling
scheme of 10base2.
However, I think John Woodgate's idea will work just fine: Terminate the
line in 75 Ohms instead of 50, and use stock 10Base2. You can literally
put a 75 Ohm resistor across the end of the cable. Don't use any balun or
transformer. As a terminology note, 100baseT2 has nothing to do with
10Base2. The former uses two twisted-pairs, and can span 200 Meters (hence
the 2) while the later uses a single piece of coax. The only similarity is
the 2, which designates 200 Meters in both cases.
I would require 100's of transceivers so custom hardware is OK.
It is not physically impossible to send data over a single cable in both
directions at the same time. It's just that you need to somehow subtract
the local TX contribution to get the RX part. And you have to make sure
the TX contribution doesn't get fouled up by the RX contribution. The
problem is that this all has to work with, in the case of 100BaseT, pretty
fast rise times. I think the rise time is one or two hundred picoseconds
or something. Oh, and you need to maintain 75-Ohm termination at both ends.
Still, you might be able to design a multiplexter that had a 100baseT to
coax conversion system. I'm just not sure it can be done with passive
components only.
I can almost envision how to do this with op-amps, but I don't know if it
would work in practice. If you want me to draw the circuit, let me know.
And now for the most far-fetched idea: Get wireless cards that can be
connected to external antennas. But don't use an antenna, just connect the
two cards to opposite ends of the cable. Please don't get mad at me if it
doesn't work, or if it ruins the cards. And do use a 50-Ohm to 75-Ohm
transformer if you can. Make sure it is rated for the frequency you are
using.
This actually might work, but if it doesn't, it will be because the signal
is too strong or too weak. If it is too weak, you can add an RF amp at the
feedpoint.
If it is too strong, you can put attenuators in line. It might be a good
idea to start with some attenuators in place and then gradually remove
them if you don't get a strong link.
Good luck!
Mac