ta2@acci.com (10/02/89)
I need a good reference on the electrical rules/specs for Ethernet.
My Dad is a telecommunications consultant who is laying out the cable
plant for an office building. The non-technical lan manager person
for this building keeps laying out all kinds of rules, which Dad
being an old time RF jock doesn't see the point of. Specifically,
1) Why can't you use 75ohm cable for thin ethernet?
2) Is is legal to run a pigtail of two or three feet from the
base of the BNC T to the Ethernet card?
Please respond by E-mail, and thanks in advance!
--- Tom
Tom Allebrandi | Advanced Computer Consulting,Inc | Charlottesville, VA
804 977 4272 | VMS User's Network Working Group -- The DECUS UUCP people
---------- you should be able to find me somewhere in this mess -----------
Bix: ta2 | Internet: ta2@acci.com | Bitnet: ta2%esther@virginia.edu
DCS: ALLEBRANDI | DECUServe: ALLEBRANDI | Pageswapper: US142404
| UUCP: ta2@esther.uucp (...!uunet!virginia!esther!ta2)johng@trwind.UUCP (John Greene) (10/06/89)
In article <188.2526de30@acci.com> ta2@acci.com writes: > >1) Why can't you use 75ohm cable for thin ethernet? You would be creating an impedance mis-match with the Thin Ethernet transceiver. This will cause relections on the cable (transmission line) and distort the signal. The problem may not be noticed with a small number of nodes and short cable lengths but as the number of nodes increase and the length of cable increases, so will the trouble. > >2) Is is legal to run a pigtail of two or three feet from the > base of the BNC T to the Ethernet card? This is a definite no-no. Not only would you be radiating all over the place (an FCC taboo) but you would cause an even more severe mis-match than if you used a 75 ohm cable. The impedance of the pigtail will depend on the thickness of the conductors and the distance between them. The cards would probably not work with a pigtail an inch long let alone two or three feet. > -- John E. Greene "People are just like frankfurters....You have to decide if you're going to be a hot dog or just another wiener" DLR TRW Information Networks Division 23800 Hawthorne Blvd, Torrance CA 90505 ARPA: johng@trwind.ind.TRW.COM USENET: ..trwrb!trwind!johng
prc@erbe.se (Robert Claeson) (10/07/89)
In article <580@trwind.UUCP> johng@trwind.UUCP (John Greene) writes: >In article <188.2526de30@acci.com> ta2@acci.com writes: >>2) Is is legal to run a pigtail of two or three feet from the >> base of the BNC T to the Ethernet card? >This is a definite no-no. Not only would you be radiating all over the place >(an FCC taboo) but you would cause an even more severe mis-match than if you >used a 75 ohm cable. The impedance of the pigtail will depend on the thickness >of the conductors and the distance between them. The cards would probably not >work with a pigtail an inch long let alone two or three feet. I've actually seen this, and it worked. The site put a thin ethernet segment in the walls and had the base end of the T connector going out of the wall. When there were a need to connect a workstation, a Mac or whatever to the network, they just connected a few meter long cable to the T connector and to the thin ethernet connector on the Ethernet board in the machine. Judging from prior knowledge, this just should't work, yet still it does. Can anyone explain this? -- Robert Claeson E-mail: rclaeson@erbe.se ERBE DATA AB
roy@phri.UUCP (Roy Smith) (10/08/89)
In article <850@maxim.erbe.se> prc@erbe.se (Robert Claeson) writes: > When there were a need to connect a workstation, a Mac or whatever to the > network, they just connected a few meter long cable to the T connector and > to the thin ethernet connector on the Ethernet board in the machine. > > Judging from prior knowledge, this just should't work, yet still it does. > Can anyone explain this? To mis-quote Abraham Lincoln, "You can violate some of the rules some of the time. You can even violate some of rules all of the time. But you can't violate all of the rules all of the time." In any well-designed system, the configuration rules are conservative and are designed to guarantee that your system will always work if you follow all the rules. They don't guarantee the converse; that your system will be broken if you violate any of the rules. Ethernet is no exception. You can probably get away with cheating in one area or another, but eventually, if you cheat enough, it will catch up with you. By installing the "pigtails", you are introducing reflections into the system. If your network is small enough, you can probably get away with it, but eventually you will discover that you can't put the full number of stations on your net, or you can't run the full number of meters of cable, and still have things work right. Perhaps if you looked carefully, you might discover that even though your net appears to work fine, you really have an abnormally high number of corrupted packets or collisions but your software is silently correcting the problem. I'll give you a down-home example of what I am talking about. When we first installed our ethernet backbone, I misinterpreted the instructions about tranciver tap spacing. What you are supposed to do is make sure the taps are on 2.5m(?) spacings, i.e. right on the black stripes marked on the cable. I read it as not allowing taps to be *any closer than* 2.5m and made sure that the taps were between stripes, with at least 2 stripes between every two taps. I've since realized my mistake, and all new taps go on stripes, but we still have a dozen or so mis-placed taps. So far, everything stil works fine, but I'm constantly aware that if our net grows enough, some day I may have to pull out the ethernet trunk cable and replace it with one done correctly. -- Roy Smith, Public Health Research Institute 455 First Avenue, New York, NY 10016 {att,philabs,cmcl2,rutgers,hombre}!phri!roy -or- roy@alanine.phri.nyu.edu "The connector is the network"
rpw3@amdcad.AMD.COM (Rob Warnock) (10/08/89)
In article <580@trwind.UUCP> johng@trwind.UUCP (John Greene) writes: +--------------- | In article <188.2526de30@acci.com> ta2@acci.com writes: | >1) Why can't you use 75ohm cable for thin ethernet? | You would be creating an impedance mis-match with the Thin Ethernet | transceiver. This will cause relections on the cable (transmission line) | and distort the signal. The problem may not be noticed with a small number | of nodes and short cable lengths... +--------------- Actually, if you also use 75-ohm terminators at the cable ends [which is needed to match the 75-cable, no?] , you'll notice trouble even with *tiny* cables. You see, Ethernet uses a D.C. voltage-sensing method for collision detection. When transmitting, Ethernet transceivers inject a known current (*not* voltage) into a known 25-ohm impdedance [the parallel connection of "two" 50-ohm cables, namely, the cable to each side of the transceiver], which creates a voltage per E = I * R. If two stations transmit at once, the currents add, and thus so do the voltages, and the collision is detected by each station measuring that the local *voltage* is more than can be accounted for by the locally-injected *current*. Now if all cables were tiny and all transceiver's transmitters exactly the same, the best voltage threshold to use for detecting collisions would be exactly between the voltages of one transmitter and two concurrent transmitters, or 1.5 times once transmitter. But cables *can* be long, and the other guy's transceiver *can* be a little below par, so a better place to set the threshold is a little lower, say 1.25 or 1.2 times your own nominal output voltage (and that's in fact what commercial transceivers do). O.K. So you decide to use 75-ohm cable, 'cause it's cheap and you can get it at Radio Shack or whatever. And you terminate the ends with 75-ohm resistors to avoid reflections. And the first time any transceiver on the cable starts transmitting, it'll see a local voltage of 1.5 times what it expects (75 / 50 = 1.5), and will scream "Collision!" over and over. Oops! Stick to the 50-ohm cable. [And for long runs, I mean the *50* ohms of such cable as RG58/AU or RG58/CU, not *53* ohms of RG58/U.] And try to get all the cable for one Ethernet from the same spool, or at least the same lot number. [Note that you can use *any* kind of cable, or even bare wire, on a *really* short cable -- say, one foot -- as long as you have two 50-ohm (or one 25-ohm) resistive terminators. Useful to know for that late-night quick hack on a benchtop.] Rob Warnock Systems Architecture Consultant UUCP: {amdcad,sun}!redwood!rpw3 DDD: (415)572-2607 USPS: 627 26th Ave, San Mateo, CA 94403
johng@trwind.UUCP (John Greene) (10/09/89)
In article <850@maxim.erbe.se> prc@erbe.se (Robert Claeson) writes: >In article <580@trwind.UUCP> johng@trwind.UUCP (John Greene) writes: >>In article <188.2526de30@acci.com> ta2@acci.com writes: > >>>2) Is is legal to run a pigtail of two or three feet from the >>> base of the BNC T to the Ethernet card? > >>This is a definite no-no. Not only would you be radiating all over the place >>(an FCC taboo) but you would cause an even more severe mis-match than if you >>used a 75 ohm cable. The impedance of the pigtail will depend on the thickness >>of the conductors and the distance between them. The cards would probably not >>work with a pigtail an inch long let alone two or three feet. > >I've actually seen this, and it worked. The site put a thin ethernet segment >in the walls and had the base end of the T connector going out of the wall. >When there were a need to connect a workstation, a Mac or whatever to the >network, they just connected a few meter long cable to the T connector and >to the thin ethernet connector on the Ethernet board in the machine. > >Judging from prior knowledge, this just should't work, yet still it does. >Can anyone explain this? For one thing, this is not a "pigtail" as I took it to mean. When we refer to pigtails around here it is taking the coax and bringing it out to two separate leads. As to why the configuration that you descibe works...luck comes to mind. One reason that I can think of off the top of my head would be that the Ethernet cards that are being attached provide an excellent match to the cable. This will minimize the reflections going back to the "trunk" and won't interfere with the signal there. If the card does not provide a good match, it can mess up the network without being active. A good reason why this should not be done. > -- John E. Greene "People are just like frankfurters....You have to decide if you're going to be a hot dog or just another wiener" DLR TRW Information Networks Division 23800 Hawthorne Blvd, Torrance CA 90505 ARPA: johng@trwind.ind.TRW.COM USENET: ..trwrb!trwind!johng
tom@wcc.oz (Tom Evans) (10/10/89)
In article <580@trwind.UUCP>, johng@trwind.UUCP (John Greene) writes: > In article <188.2526de30@acci.com> ta2@acci.com writes: > > > >1) Why can't you use 75ohm cable for thin ethernet? > > You would be creating an impedance mis-match with the Thin Ethernet > transceiver. This will cause relections on the cable (transmission line) > and distort the signal. There's a far worse problem. The Ethernet transmitters drive a constant CURRENT into the cable of 90 +-4 mA. This develops a VOLTAGE across the cable that the receivers receive. If they don't see an average voltage of half of the peak 90e-3 * (50/2), they scream "COLLISION!!". The actual allowed range is -1.492V to -1.629V. A 75 ohm cable will develop 0.5 * (90e-3 * (75/2) ) = 1.69V average voltage, and will trigger the collision detect. Ergo elk. --------- Tom Evans tom@wcc.oz | Webster Computer Corp P/L | " "I Know" is just 1270 Ferntree Gully Rd | "I Believe" with Scoresby VIC 3179 Australia | delusions of Australia | grandeur." 61-3-764-1100 FAX ...764-1179 |
pat@hprnd.HP.COM (Pat Thaler) (10/12/89)
re a reference for 802.3 cabling rules: The best reference is the source, ISO 8802-3, ANSI/IEEE Std 802.3, available from IEEE 345 East 47th Street, New York, NY 10017. The media specs are in chapter 8 for thick and chapter 10 for thin. If you have an older copy of 802.3, it doesn't contain section 10 and you will need to have an old copy of the Supplements to 802.3 CSMA/CD or to buy a new copy. It is not friendly or tutorial. It tells the rules, it usually doesn't explain why. re why using a pigtail sometimes works: 802.3 MAUs (transceivers) present a high impedance to the bus to make it look as much like a continuous piece of 50 ohm transmission line as possible. Any shunt capacitance attached to the cable results in a reflection. The standard allows each MAU to cause no more reflection than a 6 pF capacitor. When you put on a pigtail you will be creating larger reflections than permitted. The reflections do two things. They reduce the power in the signal and they interfere with the signal as noise. Assume you have a number of taps done with pigtails. Depending on where you observe on the cable, where the transmitter is, where the taps are and what the data pattern is; the reflections may add or they may cancel each other. Also, some components in your system may be better than worst case allowing margin for the extra noise. So you may find things work fine or that most stations work fine, but A and B can't talk to each other. You can get some pretty flakey problems. It is not a good idea to do. Same thing applies to 75 ohm cable. Depending on the exact transmit level and collision detect of your MAUs, you may not see a problem. You may have some MAUs that don't work and some that do. You can get false collisions and you can get CRC/alignment errors. Pat Thaler
tom@wcc.oz (Tom Evans) (10/17/89)
In article <850@maxim.erbe.se>, prc@erbe.se (Robert Claeson) writes: > In article <580@trwind.UUCP> johng@trwind.UUCP (John Greene) writes: > >In article <188.2526de30@acci.com> ta2@acci.com writes: > > >>2) Is is legal to run a pigtail of two or three feet from the > >> base of the BNC T to the Ethernet card? > > >This is a definite no-no. Not only would you be radiating all over the place > > I've actually seen this, and it worked. OK. What we ALL want (:-) is to have the cheapernet in the wall where it can't be broken, tripped over, left unconnected. What we want is a BNC on the wall, one on the back of the machine and a simple cable connecting the two. Too bad the electrical specs won't allow it This continuous bus-topology stuff doesn't make intuitive sense to people who's main connection with electricity is their home telephone, power and audio wiring :-). Now if thinwire could be made to LOOK this simple. **************** Now you CAN have this. For those who will be dreadfully offended by something that looks like an advert, junk this article NOW! **************** AMP Inc (Harrisburg PA) have a wiring system called "LAN-LINE Thinnet Tap System". It LOOKs like a simple cable from the wall to your machine. What it ACTUALLY is is a twin coax, running from the wall-plug, up to the BNC on the end of the cable and back to the wall-plug again. When you plug it into the special tap-assembly on the wall-plate, it trips a switch and your thinwire network just got 12' (or 16.4' or 24' ...) longer. Yes, it just did an insertion trick. Yes, it does break the network, but it's make-before-break and 500ms max., which is a lot faster than I can do it with BNC Tee pieces. It's also nicely insulated, so if your network administrator has left your local segment ungrounded (see grounding discussion, this group), and someone in the lab has dropped his end of the cable into the middle of an open PC power-supply, you won't be the one to find out by getting 110V (or 240V here) across you. How come no-one's mentioned the SAFETY aspect of Ethernet grounding? This stuff can transport mains power better than it does data. Disclaimer - All I have is a brochure from the local distributor - no other connection at all. --------- Tom Evans tom@wcc.oz | Webster Computer Corp P/L | "The concept of my 1270 Ferntree Gully Rd | existence is an Scoresby VIC 3179 Australia | approximation" Australia | 61-3-764-1100 FAX ...764-1179 | D. Conway
johng@trwind.UUCP (John Greene) (10/17/89)
In article <410@wcc.oz> tom@wcc.oz (Tom Evans) writes: >In article <850@maxim.erbe.se>, prc@erbe.se (Robert Claeson) writes: >> In article <580@trwind.UUCP> johng@trwind.UUCP (John Greene) writes: >> >In article <188.2526de30@acci.com> ta2@acci.com writes: >> >> >>2) Is is legal to run a pigtail of two or three feet from the >> >> base of the BNC T to the Ethernet card? >> >> >This is a definite no-no. Not only would you be radiating all over the place >> >> I've actually seen this, and it worked. > >AMP Inc (Harrisburg PA) have a wiring system called "LAN-LINE Thinnet >Tap System". Unfortunately at $150+ *per connector* it is almost cheaper to install thick Ethernet. I have heard from someone who actually tried these connectors that they cannot handle many connects/dis-connects before something breaks or the contacts quit contacting. A neat idea, but.... -- John E. Greene "People are just like frankfurters....You have to decide if you're going to be a hot dog or just another wiener" DLR TRW Information Networks Division 23800 Hawthorne Blvd, Torrance CA 90505 ARPA: johng@trwind.ind.TRW.COM USENET: ..trwrb!trwind!johng
prc@erbe.se (Robert Claeson) (10/18/89)
In article <410@wcc.oz> tom@wcc.oz (Tom Evans) writes: >OK. What we ALL want (:-) is to have the cheapernet in the wall where >it can't be broken, tripped over, left unconnected. What we want is >a BNC on the wall, one on the back of the machine and a simple cable >connecting the two. Too bad the electrical specs won't allow it I've heard about someone (actually the SAS airline company's Danish headquarter in Copenhagen) who made it work using some scheme with coils and/or capacitors that the networking company who did it won't let me know. -- Robert Claeson E-mail: rclaeson@erbe.se ERBE DATA AB
leonard@bucket.UUCP (Leonard Erickson) (10/19/89)
tom@wcc.oz (Tom Evans) writes: >OK. What we ALL want (:-) is to have the cheapernet in the wall where >it can't be broken, tripped over, left unconnected. What we want is >a BNC on the wall, one on the back of the machine and a simple cable >connecting the two. Too bad the electrical specs won't allow it It can *be* that simple. We've got BNC connectors at the wallplate. We run a cable from the wall plate to the computer where it attaches to a tee and a terminator. The in wall wiring runs to a ptch panel in the local equipment room. So at that end of the cable we run a jumper cable from a DEMPR to the BNC on the patch panel. Everything in the equipment room is rack mounted. The DEMPRs connect to the DecNet backbone and we carry the backbone between buildings on fiber. It works quite well. We have several connections in most spaces and even if we do run out of connections in an area, we can just attach the new machine to an existing machine until we get around to adding a new drop (if needed) The best part is that the worst some user can do is knock his machine and and maybe one or two others that are daisy chained on it. Yeah, I know, it's a *star* configuration, but it *works*. We've got a few machines on some old wiring that predates the DecNet. They are a pain because unless you have a diagram handy you have to assume that you can't do anything we the cabling except disconnect the tee from the computer... I'll be glad when they're gone. -- Leonard Erickson ...!tektronix!reed!percival!bucket!leonard CIS: [70465,203] "I'm all in favor of keeping dangerous weapons out of the hands of fools. Let's start with typewriters." -- Solomon Short
jimm@haddock.ima.isc.com (Jim McGrath) (10/21/89)
In article <410@wcc.oz> tom@wcc.oz (Tom Evans) writes: > >OK. What we ALL want (:-) is to have the cheapernet in the wall where >it can't be broken, tripped over, left unconnected. What we want is >a BNC on the wall, one on the back of the machine and a simple cable >connecting the two. Too bad the electrical specs won't allow it > I recently ran across a product from AMP that provides this capability. It consists of a wall plate assembly and drop cable. The drop cable is really a double length coax, and thus increases your segment length by twice the length of the drop. There is an application note (DP 5727) that should be reviewed to get an understanding of potential problems, e.g. reflections caused by impedence mismatch since both segment and drop may have a mismatch of as much as 4 ohms (the standard is 50 ohms +- 2). Jim