SRA@wsmr-simtel20.army.mil (Gene N. Cartier) (11/22/90)
Last week I posted an inquiry to the net and asked about experiences in running 10Base-T in the same sheath as telco wiring. I was particularly worried about "punching holes" in data caused by ringing voltages. I didn't get as many responses as I had hoped, but attached are some of the more interesting ones. The general consensus seemed to be "Don't worry, but be careful." Listed below are some of the responses: >>Some respondents stated that there are products on >>the market that solve the problem. I tend to think that this >>is creative marketing >Cabletron has a 10base-T product they claim will not effect or be effected >by running in the same sheath as telco wiring. I'm on the verge of trying >this, so if you hear otherwise, I'd appreciate an update. > > Steven Cook <ls03cook%engrhub@hub.ucsb.edu> > UC Santa Barbara >I recently talked with our local BOC rep and Northern-Telecom >about their LANSTAR system. The selling point was the ability to run >2.56 Mbits on existing telco cable using the (typically unused) >two twisted pair of the three usually found ... my real advice: >test, test, test ... good luck! > > Lindsay Reed <LINDSAY%SLU.Bitnet@ricevm1.rice.edu> >>Other respondents say that they are running without any problems >>in the same type of environment. >In our office, we have the type of setup that you are >considering: six-conductor cable (three pairs) with voice on one >(Centrex service, AT&T STARPLUS phones), and 10BASET on one >(we use Ungermann-Bass NET/ONE boxes). It seems to work okay; there >is no noticeable crosstalk, and I've never seen any data dropouts >caused by phone ringing or anything else. The phone ring is at such >a low frequency compared to the data rate of the 10BASET that it >probably looks to them like just a DC offset, and they just filter >that out. In some places we run Appletalk (120 kbps) on the third >pair (using PhoneNet matching transformers) and that seems to work >okay too. > >I'll have to admit I'm a bit surprised; I'm one of those people >who didn't think Ethernet on UTP would ever work at all. I thought >for sure that there would be all kinds of crosstalk problems. I've >finally learned the lesson that my father tried to teach me: (never >underestimate analog EE weenies). > >One thing that you might to watch out for: it may be that voice and >data in the same cable is one thing, and two data pairs in the same >cable are quite another. I have seen an installation where RS232 >was run through Belden 24-conductor cable, and that was crosstalk >city, especially on long runs. A similar installation using >individually-shielded pairs worked quite well, however. > > David Cornutt <cornutt@freedom.msfc.nasa.gov> > New Technology Inc, Huntsville AL >I work at Northern Telecom Information Systems, and we run EVERYTHING >over twisted pairs. We run an eight-pair cable to each office, then >patch in whatever we need, including voice (Centrex), data lines (POTS), >Appletalk, Ethernet (802.3 from Synoptics), RS-232, and balanced 3270 >lines. All of it works fine, with no interference. Just make sure you >use cable with a high twist-per-foot count to avoid EMI problems. > > Heath Roberts <barefoot@catt.ncsu.edu> > NCSU Computing Center >Running 10BaseT and voice in the same sheath is no problem, PROVIDED THAT >THE PAIRS ARE TWISTED. In our new business building at Andrews University >they ran six pair to every office from a couple of closets, then four >25-pair cables to each closet from the mechanical service room. The >pairs were twisted in the 25-pair stuff, but not in the 6-pair. >10BaseT "must" run on "twisted pair"!! After we had purchased the >10BaseT equipment we discovered that it wouldn't run from the janitor's >closet to the offices. So we ended up running wire to each office any- >way, destroying any cost savings of 10BaseT. >Moral: Verify that the wiring TO THE OFFICES is TWISTED pair. If not, you >will be running new wiring even for 10BaseT. I know from sad experience >that not all 6-pair cable has each pair individually twisted. > > Bill Nickless <nickless@flash.ras.anl.gov> The bottom line is that I will run some tests to ensure that the CENTREX ringing doesn't effect the data and produce cross talk and maybe and run a prototype to ensure that we don't spend money needlessly. I'm still sort of queasy, but if its working in so many places maybe my fears are unfounded. Thanks to the people that responded. Gene Cartier SRA@WSMR-SIMTEL20.ARMY.MIL