geoff@suneast.uucp (Geoff Arnold) (02/19/86)
Does anyone have any knowledge/experience/opinions/etc. about a
product by Astra(?) of Mt.View(?) which claims to emulate Ethernet
over a twisted pair ring? I guess they offer a transceiver-like
interface. Sounds bizarrrrrre, but not unattractive.
--
#include <sys/disclaimer.h> /* co. lawyers: will this do? */
Geoff Arnold, Sun Microsystems Inc. (East Coast Division)
SnailMail: One Cranberry Hill, Lexington, MA 02173; 617-863-8870 x136
UUCP: [East Coast] linus!security!sunne!suneast!geoff
[West Coast] {hplabs,ihnp4,nsc,pyramid,decwrl}!sun!suneast!geoffphilip@axis.UUCP (Philip Peake) (02/26/86)
In article <135@suneast.uucp> geoff@suneast.UUCP writes: >Does anyone have any knowledge/experience/opinions/etc. about a >product by Astra(?) of Mt.View(?) which claims to emulate Ethernet >over a twisted pair ring? I guess they offer a transceiver-like >interface. Sounds bizarrrrrre, but not unattractive. I can't claim to have any experience of Ethernet on twisted pair, but, I do have some experience of "The Cambridge Ring", which uses twisted pair cables, and also of working for 5 years with a cable TV distribution system which used twisted pair cable. One of the problems of the current generation of LAN merchants is that they are too "digital oriented". A good example is the Cambridge Ring. This requires repeaters every hundred meters or so. It needn't. If some attention were paid to the basic principles of transmission of radio frequency signals over cables (which is really what we are talking about). Then *very* acceptable results can be achieved. With the TV distribution system long trunk routes of several miles were common, this is with 3 volts of drive ! On service routes with 10 volts of drive, and tap off points for every house in a street, it is easy to achieve half a mile between repeaters. Twisted pair cable is cheaper than co-ax, but, one must pay attention to the quality of cable used. More important still is the quality of installation of the cable. Any tap-offs must be done neatly, making the smallest possible interuption in the uniform twist of the cable. This achieves two things; it reduces the radiation from the cable - which is important if you don't want to upset your PTT, (at the same time reducing pickup of incident signals on your system), and reduces the reflections of the signal from the discontinuity in the cable impedance. Taking up another point I recently saw in this group, that of branches in Ethernet systems. I am fairly convinced that this is feasible with twisted pair networks. The trick is to used "back matched" transformers. Transformer technology is easier to apply to twisted pair (balanced) circuits than to co-ax (un-balanced). You can thus use a transformer with two secondary windings, one driving each branch. The problem is then to match correctly to signals coming from the "opposite direction". This should be possible. Obviously, this had no application in TV networks, so was not fully explored - however we did have transformers which sucessfully absorbed reflected signals in an internal resistor rather than re-reflecting it. I believe that telephone systems may actually use such a system to give two-way communication on a single cable. there is room for optimistic research here. So, basically, yes, it sounds like a good idea. BUT, you have to be carefull about installation and cable quality. ------------------------------------------------- Philip Peake. UUCP: {mcvax, vmucnam, ircam}!axis!philip