animal@ernie.NECAM.COM (Alan R. Silverman) (09/19/88)
ANC has moved to Tasman Drive, San Jose. I forgot the actual address and phone number. We have purchased one for testing purposes of running a single diskless node from one server. This highly restricts network growth, but is an excelent solution for small limited applications. Connecting two transceivers to a short length of cable can be expensive and a pain in the butt. A minimum length backbone cable is required. I don't know the length, but its around a hundred or so feet. An alternative that we've used succesfully in the past is a multi- port transciever box made by BICC, Excelan, and I believe Bridge/3Com. We have 2 BICC boxes currently in use ( and very successfully ). It's a small box with 8 transceivers built into it and a connection to a standard ethernet backbone. It is capable of standalone operations in which 8 ethernet devices can communicate as if on a full ethernet. It works great in the computer room where we have many hosts and often like a quick connection to test other devices, etc. The cost of this box is around $1000, which is compatible to the cost of putting together a small ethernet with cable, transceivers, and misc. parts. The best part, is no mess, no fuss. Couldn't be simpler! So much for my two cents. +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- ##### ^|- ##### ##### (0 o) Alan R. Silverman ##### ##### (+) NEC America, Inc. ##### ##### =(_)= (408) 922-3814 ##### ##### U aka animal@ernie ##### ##### ##### ##### Ack. Thpth. ##### ##### (trans. " So simple, even I can do it!!") ##### +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
rpw3@amdcad.AMD.COM (Rob Warnock) (09/19/88)
In article <157@ernie.NECAM.COM> animal@ernie.NECAM.COM (Alan R. Silverman) writes: +--------------- | Connecting two transceivers to a short length of cable can be expensive | and a pain in the butt. A minimum length backbone cable is required. | I don't know the length, but its around a hundred or so feet. +--------------- This is *NOT* true! There is no minumum length requirement at all! Even the 2.5 meter transceiver spacing rule is needed only on long cables with dozens of transceivers. You can safely put 2-10 (or so) transceivers on as small a piece of cable as will fit (say, 6 inches per transceiver). It's done all the time. (Just remember, though, you still need terminators at both ends.) Rob Warnock Systems Architecture Consultant UUCP: {amdcad,fortune,sun}!redwood!rpw3 ATTmail: !rpw3 DDD: (415)572-2607 USPS: 627 26th Ave, San Mateo, CA 94403
brian@ncrcan.Toronto.NCR.COM (Brian Onn) (09/23/88)
In article <22961@amdcad.AMD.COM> rpw3@amdcad.UUCP (Rob Warnock) writes: >In article <157@ernie.NECAM.COM> animal@ernie.NECAM.COM (Alan R. Silverman) >writes: >+--------------- >| Connecting two transceivers to a short length of cable can be expensive >| and a pain in the butt. A minimum length backbone cable is required. >| I don't know the length, but its around a hundred or so feet. >+--------------- > >This is *NOT* true! There is no minumum length requirement at all! Even the >2.5 meter transceiver spacing rule is needed only on long cables with dozens >of transceivers. You can safely put 2-10 (or so) transceivers on as small >a piece of cable as will fit (say, 6 inches per transceiver). It's done all >the time. (Just remember, though, you still need terminators at both ends.) I'm not disputing the idea of putting transceivers every 6 inches or so, but what kind of havoc does this wreak on the ethernet? I thought the 2.5 meter separation rule was there to prevent/minimize reflections on the medium? Whether or not you get reflections (at some transceiver separation) would depend (not entirely) upon the wavelength of the signal on the medium, right?. So what is the wavelength of a manchester encoded signal at ethernet speeds? Brian.
eshop@saturn.ucsc.edu (Jim Warner) (09/25/88)
In article <920@ncrcan.Toronto.NCR.COM> brian@ncrcan.Toronto.NCR.COM (The Super User) writes: > >I'm not disputing the idea of putting transceivers every 6 inches or so, >but what kind of havoc does this wreak on the ethernet? I thought the >2.5 meter separation rule was there to prevent/minimize reflections on the >medium? The purpose of the rule is to prevent an in-phase build up of the reflections from individual transceivers. You can't prevent the reflections themselves. The spacing rule guarantees that, when you get up near the 100 transceiver limit, that the reflections won't be in phase and create a big enough composite reflection to trash out a packet. If your network has a small number of transceivers (less than 10), even if their reflections are all in phase, their sum is too small to do any damage. jim warner U.C. Santa Cruz
henry@utzoo.uucp (Henry Spencer) (09/25/88)
In article <920@ncrcan.Toronto.NCR.COM> brian@ncrcan.Toronto.NCR.COM (The Super User) writes: >I'm not disputing the idea of putting transceivers every 6 inches or so, >but what kind of havoc does this wreak on the ethernet? I thought the >2.5 meter separation rule was there to prevent/minimize reflections on the >medium? Yes, but reflections matter only if the total end-to-end-and-back propagation delay is some significant fraction of the rise and fall times of the signals involved. If the cable is only a meter or so long, the delay will be only a few nanoseconds -- not significant at Ethernet speeds. Loosely speaking, reflections are the process by which the various parts of the system agree on how they will respond to the signal. When things are close together, agreement is reached more quickly than the signal can change, and essentially one has consensus at all times. (Much of what you get taught in an elementary-electronics course quietly assumes this.) When delays are long, though, consensus breaks down and the responses of distant parts trickle in slowly after a signal change. The trouble is that all this back-and-forth can be confusing to parts in intermediate places. So when delays are long, one has to take precautions to try to ensure that all parts respond the same way, and that any remaining nonuniformities cause minimum disturbance. But most of these precautions are irrelevant if the delays are slight. -- NASA is into artificial | Henry Spencer at U of Toronto Zoology stupidity. - Jerry Pournelle | uunet!attcan!utzoo!henry henry@zoo.toronto.edu