Path: utzoo!attcan!uunet!know!sdd.hp.com!hplabs!pyramid!lstowell From: lstowell@pyrnova.pyramid.com (Lon Stowell) Newsgroups: comp.dcom.lans Subject: Re: Problems associated with Token-Ring and Twisted Pair Message-ID: <133122@pyramid.pyramid.com> Date: 5 Nov 90 21:45:06 GMT Sender: daemon@pyramid.pyramid.com Reply-To: lstowell@pyrnova.pyramid.com (Lon Stowell) Organization: Pyramid Technology Corp., Mountain View, CA Lines: 63 In article <28522@bellcore.bellcore.com> louie@cellar.bae.bellcore.com (Paul Louie) writes: >In the past few months, there were many auguments and counter auguments >against and for using Token-Ring and Unshielded Twisted Pair (UTP) wires. >The source of this debate I noticed range from systems integrators, LAN >supervisors, and trade magazines. > >I like to know the experiences of the netters out there. This would >resolve my curiosity and may help to steer away from a problem area. >So what is it? Is the design of TRN not suitable to UTP or maybe that >TRN has a very low tolerant and any bad connection or radiation can >set it crazy. It depends on whether you are running 4 or 16 Mbps Token Ring--and if 4, whether or not you intend to upgrade to 16. Token Ring runs quite nicely on UTP at 4 Mbit/sec. Since the FCC doesn't like radiation from the unshielded cable, a Media Access Filter is required (which may be imbedded in the T/R Adapter--a good choice..) to remove the higher harmonics of the Xmitted signal at each station's transmitter. This high-frequency roll-off, plus the lower transmission capability of UTP limits the physical span of T/R to ABOUT a third of that available with IBM Type 1 cabling. The problem is "jitter" with either implementation--UTP tends to hit the limit before Type 1... As long as you stay within the proper Lobe distance, keep the Main Ring Path within limits, UTP provides exactly the same error rate as Type 1. MOST installations have workstations well within UTP's range of the nearest MSAU.... You may combine Type 1 between MSAU's with UTP for Lobes for SOME increase in span in larger buildings compared to UTP (type 3) cable alone.... Several VAR's and aftermarket vendors push UTP well beyond IBM's cabling guidelines...as the IBM recommendations tend to be quite conservatively aimed at maximum reliability.... These aftermarket implementations seem to work quite well...symptoms of excessive length would be frequent Active Monitor changes and non-isolating error counts as jitter exceeds the ability of the stations to clock data reliably. 16 Mbit UTP is a different story. The IBM adapters APPEAR to work better than the T.I. based units (everyone but IBM...) although this is a hot area for manufacturer's and some fairly creative marketing claims from a few... There is nothing technically insoluble here, but tread carefully. If you have a ham radio background, look for Adapter cards with fully isolated ground and voltage planes for the Ring Interface circuitry with all signals buried between shielding layers. Look for good RF layout practices--even if at the expense of "pretty" component layout for ease of manufacturing. The TROLI modules from Pulse can help considerably, but poor RF layout can render even thes ineffective... Proteon has a unique solution for 16 UTP which (as I understand their implementation) re-clocks the ring data at each station. Although there are some possible losses in fault isolation, this re-clock approach has interesting possibilities for more reliable clocking--whatever the media....