Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!psuvax1!psuvm!UMAXC!CFINNEY From: cfinney@UMAXC.WEEG.UIOWA.EDU (Charles Finney) Newsgroups: bit.listserv.novell Subject: Re: ETHERNET PROBLEM Message-ID: <9002011730.AA15665@umaxc.weeg.uiowa.edu> Date: 1 Feb 90 17:30:22 GMT Sender: Novell LAN Interest Group Reply-To: Novell LAN Interest Group Lines: 15 Approved: NETNEWS@PSUVM Gateway X-To: NOVELL%SUVM@pucc.PRINCETON.EDU The specification do not say as regards the .5m stub. I haven't worked through the spec to determine if it is a requirement or not. We terminate our machines without a stub, but the question has come up here because of intermittent problems. As for lengths over 185m, some boards will work and some won't. We have a seriously mixed bag of adapters, with a Sniffer it became apparent that not all adapters packets would reach the other end of the cable intact. Shortened the cables to spec using multiport repeaters and the dropped packet problems disappeared. Another potential problem is maximum end to end delay is 950 ns. Collision detection depends on this timing, 185m is the limit at .65c for propagation rate. Unless you are willing to live with the increased error ratesand dropped packets, etc. it's not a great idea. Charles E. Finney University of Iowa