jennifer@bsu-cs.bsu.edu (Jennifer Freeman) (09/18/90)
A company I am doing some work at needs to do some office rearanging as they are rapidly filling up the space they have. They are also about to get a multi-user Unix Box running on a 386 or a 486. One of the proposals is to locate the computer about 300 feet away from where the bulk (10-14) of the terminals will be. What type of wiring would be needed for this? Would shielded be good enough? Would some kind of line boosters be needed? We have some people with knowledge of micros, but not in wiring multi-user systems. Any suggestions things to look out for, tips etc would be much apriciated.
art@pilikia.pegasus.com (Art Neilson) (09/19/90)
In article <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: >A company I am doing some work at needs to do some office rearanging as they >are rapidly filling up the space they have. They are also about to get a >multi-user Unix Box running on a 386 or a 486. One of the proposals is to >locate the computer about 300 feet away from where the bulk (10-14) >of the terminals will be. >What type of wiring would be needed for this? Would shielded be >good enough? Would some kind of line boosters be needed? We have some people >with knowledge of micros, but not in wiring multi-user systems. Any suggestions >things to look out for, tips etc would be much apriciated. 50 feet is supposed to be the max for serial cable, however I've seen runs up to ~500 feet or so with no problems. Your milage may vary. 4-wire is usually good enuf, with a breakout box you can play with the signals and figure out which ones your box and tubes require. -- Arthur W. Neilson III | ARPA: art@pilikia.pegasus.com Bank of Hawaii Tech Support | UUCP: uunet!ucsd!nosc!pegasus!pilikia!art
lerman@stpstn.UUCP (Ken Lerman) (09/19/90)
In article <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes:
->A company I am doing some work at needs to do some office rearanging as they
->are rapidly filling up the space they have. They are also about to get a
->multi-user Unix Box running on a 386 or a 486. One of the proposals is to
->locate the computer about 300 feet away from where the bulk (10-14)
->of the terminals will be.
->What type of wiring would be needed for this? Would shielded be
->good enough? Would some kind of line boosters be needed? We have some people
->with knowledge of micros, but not in wiring multi-user systems. Any suggestions
->things to look out for, tips etc would be much apriciated.
Most modern terminals and drivers will run at 9600bps at up to 1000
feet. Depending on where you are and local codes, you may have to use
wire with special insulation which will add to your cost. If I were
you, I would look at running an ethernet the 300 feet to the center of
my office area and putting a 16 port terminal driver there. (I forget
what the things are called.) It will probably cost you about $200 per
line. But running a single ethernet instead of a bunch of wires saves
money.
Ken
jeff@marichal.austin.ibm.com (Jeff Johnson) (09/19/90)
In <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: >A company I am doing some work at needs to do some office rearanging as they >are rapidly filling up the space they have. They are also about to get a >multi-user Unix Box running on a 386 or a 486. One of the proposals is to >locate the computer about 300 feet away from where the bulk (10-14) >of the terminals will be. >What type of wiring would be needed for this? Would shielded be >good enough? Would some kind of line boosters be needed? We have some people >with knowledge of micros, but not in wiring multi-user systems. Any suggestions >things to look out for, tips etc would be much apriciated. The RS-232C standard (oxymoron?) limits the length of the cable to 50 feet. However, I have installed cables up to 250 feet with only minor noise problems. (We had an AM radio station transmitter 1/2 mile away and if the shielding wasn't grounded properly the wire acted like an antenna. It was very noticeable on our Merlin phone system cables that ran parallel to the serial cables.) We also ran a line about 500 feet using a passive modem from Black Box. We always had lots of trouble with this one at 9600 baud. It had a tendency of hanging the serial port or crashing the board. It worked better at 2400 or 4800 baud. I would suggest using 8-wire shielded twisted-pair and make sure you ground the shielding and pick complimentary wires for the twisted-pairs. Unless you have a very noisy environment or poor line drivers, it should work. You could also look into fiber-optics. -- Jeff Johnson 10926 Jollyville Rd #1420 Computer Consultant Austin, TX 78759 (512) 343 0675 Email -> uunet!cs.utexas.edu!ibmchs!auschs!marichal.austin.ibm.com!jeff Disclaimer: "My views and opinions do not reflect those of IBM"
usenet@carssdf.UUCP (John Watson) (09/20/90)
In article <11774@bsu-cs.bsu.edu>, jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: > locate the computer about 300 feet away from where the bulk (10-14) > of the terminals will be. > What type of wiring would be needed for this? Would shielded be > good enough? Would some kind of line boosters be needed? We have some people The computone cluster controller lets me (one of my customers) use one board in their 486 running Xenix 2.3.2, which connects to 4 cluster boxes. The boxes are several hundred feed away, check specks. Each box supports 16 terminals/printers. I have not tried modems on these boxes, net users have reported various modem/computone problems. The "trunk" cable is custom, you buy it from them. It has 3 pair shielded, using a protocal similar to RS422, balanced low impedance etc... If you try it with just long wires, the shielded wire tends to have high capacity/foot and may be a problem. Also long wires tend to pick up lightning charges like antenna, so have a pocked full of 7518[89] chips handy. The small guage "cheap" telephone wire tends to have a higher characteristic impedance and work better on the rs232 interface. Chose your pairs well, don't pair tx/rx for instance. Good luck, its a jungle. No matter what you do, someone will second guess you when that board breaks or lightning gets your terminal. I have read on the net that others have had good look with ARNET boards. Just trying to help, no ax to grind, not trying to make a buck. John Watson Middlesex NJ USA
bill@bilver.UUCP (Bill Vermillion) (09/20/90)
In article <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: >A company I am doing some work at needs to do some office rearanging as they >are rapidly filling up the space they have. They are also about to get a >multi-user Unix Box running on a 386 or a 486. One of the proposals is to >locate the computer about 300 feet away from where the bulk (10-14) >of the terminals will be. >What type of wiring would be needed for this? Would shielded be >good enough? Would some kind of line boosters be needed? We have some people >with knowledge of micros, but not in wiring multi-user systems. Any suggestions >things to look out for, tips etc would be much apriciated. You should probably look into one of the intelligent multiuser boards that supports RS422. There are several out there. Anvil's individual ports can be made selectively RS422 on a per/port basis, and I have seen at least one other manufacturer that uses a 422 transmission scheme to it's breakout boxes, where the first 8 ports can be up to 1000 feet from the computer, and each additional 8 port box can be 1000' from the previous box. The ports themselves are rs232. Using a 1.5Mbit transfer rate you are supposed to be able to support 32 terminals 38k. I do remember seeing the board, and it may be a Computone or Corollary - it has an ix in it's name. -- Bill Vermillion - UUCP: uunet!tarpit!bilver!bill : bill@bilver.UUCP
cpcahil@virtech.uucp (Conor P. Cahill) (09/20/90)
In article <1990Sep19.083601.21364@pilikia.pegasus.com> art@pilikia.pegasus.com (Art Neilson) writes: >50 feet is supposed to be the max for serial cable, however I've seen runs >up to ~500 feet or so with no problems. Your milage may vary. 4-wire is >usually good enuf, with a breakout box you can play with the signals and >figure out which ones your box and tubes require. The longer the run, the more help a shielded cable will be (over a non-shielded). We have runs of about 200' off of a megaport with no problems running at 38.4. They (Equinox) also have a special card that allows runs up to 1000'. -- Conor P. Cahill (703)430-9247 Virtual Technologies, Inc., uunet!virtech!cpcahil 46030 Manekin Plaza, Suite 160 Sterling, VA 22170
bruce@segue.segue.com (Bruce Adler) (09/21/90)
In article <jeff.653752463@marichal.austin.ibm.com> jeff@marichal.austin.ibm.com (Jeff Johnson) writes: >I would suggest using 8-wire shielded twisted-pair and make sure you >ground the shielding and pick complimentary wires for the twisted-pairs. >Unless you have a very noisy environment or poor line drivers, it should >work. Believe it or not, sheilding actually reduces the maximum usable length and speed. For long runs only use sheiled cable if you're operating in a noisy environment. Also, be careful about your grouding. Doing it wrong will introduce ground-loops and just make things worse. -- bruce@segue.com, ism.isc.com!segue!bruce, aero.org!segue!bruce
rjw@atti07.ATT.COM (Ralph J. Winslow x7774) (09/21/90)
-In <1990Sep19.083601.21364@pilikia.pegasus.com> art@pilikia.pegasus.com (Art Neilson) writes: ->In <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: -> ... multi-user Unix Box running on a 386 or a 486. One of the proposals is ->to locate the computer about 300 feet away from where the bulk (10-14) ->of the terminals will be. ->What type of wiring would be needed for this? Would shielded be ... - -50 feet is supposed to be the max for serial cable, however I've seen runs -up to ~500 feet or so with no problems. Your milage may vary. 4-wire is -usually good enuf, with a breakout box you can play with the signals and -figure out which ones your box and tubes require. That's true, but be careful to use twisted pairs if you use 4-wire. Actually, use twisted pair(s) whatever wire you use. --- -Arthur W. Neilson III | ARPA: art@pilikia.pegasus.com -Bank of Hawaii Tech Support | UUCP: uunet!ucsd!nosc!pegasus!pilikia!art -- Ralph Winslow ulysses!attibr!atti07!rjw
friedl@mtndew.Tustin.CA.US (Steve Friedl) (09/21/90)
[ How to get long-distance RS-232 connections ] Jeff Johnson writes: > I would suggest using 8-wire shielded twisted-pair and make sure you > ground the shielding and pick complimentary wires for the twisted-pairs. > Unless you have a very noisy environment or poor line drivers, it should > work. It was always my impression that twisted pairs were to be used for differential circuits (like RS-422?) where the two wires were electrically related or if one of them is a ground. While (say) CTS and RTS are "related" if one is looking at a datacomm protocol, electrically they have no relationship to each other, and twisting these wires together won't have any kind of meaningful benefit. Am I missing something? Steve -- Stephen J. Friedl, KA8CMY / I speak for me only / Tustin, CA / 3B2-kind-of-guy +1 714 544 6561 / friedl@mtndew.Tustin.CA.US / {uunet,attmail}!mtndew!friedl Jesse Helms for U.S. Supreme Court Justice
pavlov@canisius.UUCP (Greg Pavlov) (09/23/90)
> In article <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: >...multi-user Unix Box running on a 386 or a 486. One of the proposals is to >locate the computer about 300 feet away from where the bulk (10-14) >of the terminals will be. >What type of wiring would be needed for this? Would shielded be >good enough? Would some kind of line boosters be needed? We have some people >with knowledge of micros, but not in wiring multi-user systems. Any suggestions > The "official" limit is in the 50-75 ft range (I don't remember the exact no. and the context), but I you can go nuch higher. The longest successful runs I know of were in the 1300-ft range (most of it through an elevator shaft, no less). So you will probably be ok, tho I doubt that anyone will "guarantee" such an installation. But I would also look at a twisted pair ethernet/terminal ser- ver solution as an alternative. It may not cost that much more - you need to purchase port card(s) for the UNIX box and a substantial amount of wire and connectors, so you are faced with a pretty large expense in what you described The terminal server solution should yield quicker screen writes and may be more efficient cpu-wise, while the solution overall should be more amenable to future expansion. greg pavlov, fstrf. amherst, ny
jeffl@comix.UUCP (Jeff Liebermann) (09/24/90)
rjw@atti07.ATT.COM (Ralph J. Winslow x7774) writes: >-In <1990Sep19.083601.21364@pilikia.pegasus.com> art@pilikia.pegasus.com > (Art Neilson) writes: >->In <11774@bsu-cs.bsu.edu> jennifer@bsu-cs.bsu.edu (Jennifer Freeman) writes: >-> ... multi-user Unix Box running on a 386 or a 486. One of the proposals is >->to locate the computer about 300 feet away from where the bulk (10-14) >->of the terminals will be. >->What type of wiring would be needed for this? Would shielded be ... >That's true, but be careful to use twisted pairs if you use 4-wire. >Actually, use twisted pair(s) whatever wire you use. Wrong. The purpose of twisted pair wire is to reduce coupling between adjacent pairs and to reduce common mode pickup. Since RS-232c is a unipolar (unbalanced, common ground wire) scheme, neither of these are applicable. In fact, the twisted pair has a much higher capacitance than unshielded twisted pair. Adding a shield just increases the capacitance (to ground). The end effect is rise time deterioration. The higher the frequency (baud rate), the worse the effect until no signal makes it to the other end. I have had no difficulties with 500ft runs of 4 conductor #26awg telephone "station wire" at 19.2kb/s. Yet I have several 30ft runs that barely work. The difference is that in the 30ft runs, the AC power grounds were poor and about 10 volts (rms) of 60HZ crud appeared between the computer and the terminal grounds. Since the signal grounds between the computer and terminals are all wired together (pin 7), the signal appeared on top of the data. The solution for this kind of grounding problem (found between buildings, between floors, and creative electrical wiring) is to go balanced (differential) drivers at one or both ends. The idea is to eliminate any ground loops. There are "short haul modems" (which are sold by Inmac, Black Box, Telebyte, and others for about $70/ea) which convert unipolar RS-232 to balanced line and back. Some are advertised to go 6000 ft. They require 4 wires per terminal. Another solution is a cluster controller card. I use Corollary products. Their card will handle 4each of 8 port controllers to a max of 32 ports per card. Only 4 wires (balanced) go between the cpu and each cluster controller. I use these between floors and buildings. One nice feature is the ability to unplug a controller, watch the console complain, plug it back, and continue without interruption. -- # Jeff Liebermann Box 272 1540 Jackson Ave Ben Lomond CA 95005 # (408)336-2558 voice (408)429-0483 digital pager CIS:73557,2074 # PC REPAIR & RF DESIGN uunet!comix!jeffl ucscc.ucsc.edu!comix!jeffl # universe!milky_way!solar_system!earth!na!us!uunet!comix!jeffl
floydd@chinet.chi.il.us (Floyd Davidson) (09/25/90)
In article <1142@atti07.ATT.COM> rjw@atti07.ATT.COM (Ralph J. Winslow x7774) writes: >->What type of wiring would be needed for this? Would shielded be ... >- >-50 feet is supposed to be the max for serial cable, however I've seen runs >-up to ~500 feet or so with no problems. Your milage may vary. 4-wire is >-usually good enuf, with a breakout box you can play with the signals and >-figure out which ones your box and tubes require. > >That's true, but be careful to use twisted pairs if you use 4-wire. >Actually, use twisted pair(s) whatever wire you use. > Pray tell what does using twisted pair wire do on an unbalanced circuit? Nothing. Twisted pair will do wonders for common mode rejection on balance telephone lines, maybe on RS-422, but nothing for RS-232. Floyd
bruce@segue.segue.com (Bruce Adler) (09/26/90)
In article <1990Sep25.052558.11002@chinet.chi.il.us> floydd@chinet.chi.il.us (Floyd Davidson) writes: >Pray tell what does using twisted pair wire do on an unbalanced >circuit? Nothing. Twisted pair will do wonders for common mode >rejection on balance telephone lines, maybe on RS-422, but nothing >for RS-232. Not exactly nothing. Try this experiment at home, hook up about 200-300 feet of twisted pair to the RD and TD pins of your favorite 9600 baud terminal. Leave the far end disconnected (this is the trick). Now type something on the keyboard. The last time I tried this the unterminated cable run magically echo-ed every character as if it was really attached to a host port. Can you say "inductive coupling"? -- bruce@segue.com, ism.isc.com!segue!bruce, aero.org!segue!bruce
floydd@chinet.chi.il.us (Floyd Davidson) (09/28/90)
In article <3927@segue.segue.com> bruce@segue.segue.com (Bruce Adler) writes: >In article <1990Sep25.052558.11002@chinet.chi.il.us> floydd@chinet.chi.il.us (Floyd Davidson) writes: >>Pray tell what does using twisted pair wire do on an unbalanced >>circuit? Nothing. Twisted pair will do wonders for common mode >>rejection on balance telephone lines, maybe on RS-422, but nothing >>for RS-232. > >Not exactly nothing. Try this experiment at home, hook up about 200-300 >feet of twisted pair to the RD and TD pins of your favorite 9600 baud >terminal. Leave the far end disconnected (this is the trick). Now type >something on the keyboard. The last time I tried this the unterminated >cable run magically echo-ed every character as if it was really attached >to a host port. Can you say "inductive coupling"? Can you say "capacitive"? If the line was terminated and some current flowed or if it was so long (a bigger capacitor) that significant current flowed to charge it up, you might well get inductive coupling. But what you get with a couple hundred feet of unterminated cable is the same thing you would get by just putting a capacitor across the connector pins. (Look up transmission line theory for radio transmission and then realize that voice or even data is the same thing at a lower frequency.) Either way the last thing you want to use for RS-232 is twisted pair.
lerman@stpstn.UUCP (Ken Lerman) (09/28/90)
[...] ->Not exactly nothing. Try this experiment at home, hook up about 200-300 ->feet of twisted pair to the RD and TD pins of your favorite 9600 baud ->terminal. Leave the far end disconnected (this is the trick). Now type ->something on the keyboard. The last time I tried this the unterminated ->cable run magically echo-ed every character as if it was really attached ->to a host port. Can you say "inductive coupling"? ->-- ->bruce@segue.com, ism.isc.com!segue!bruce, aero.org!segue!bruce Or, more likely, capacitive coupling. Inductive coupling would require that there be a current path back from each wire. Of course, that current path could be capacitive and you could have a combination of inductive and capacitive coupling, but in this case, I would expect capacitive coupling to be the "culprit". Ken