From owner-freebsd-scsi Sun Jun 8 11:29:16 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id LAA07744 for freebsd-scsi-outgoing; Sun, 8 Jun 1997 11:29:16 -0700 (PDT) Received: from GndRsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.241]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA07727 for ; Sun, 8 Jun 1997 11:29:11 -0700 (PDT) Received: (from rgrimes@localhost) by GndRsh.aac.dev.com (8.8.5/8.7.3) id LAA24046; Sun, 8 Jun 1997 11:28:39 -0700 (PDT) From: "Rodney W. Grimes" Message-Id: <199706081828.LAA24046@GndRsh.aac.dev.com> Subject: Re: 3940uw with long cables In-Reply-To: <199706071831.UAA00806@yedi.iaf.nl> from Wilko Bulte at "Jun 7, 97 08:31:18 pm" To: wilko@yedi.iaf.nl (Wilko Bulte) Date: Sun, 8 Jun 1997 11:28:39 -0700 (PDT) Cc: gibbs@plutotech.com, asami@CS.Berkeley.EDU, scsi@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL25 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-scsi@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > As Justin T. Gibbs wrote... > > > Perhaps. There are a few important things to keep in mind when > > tuning the bus length. > > > > 1) Having equal distances between all devices is sometimes more > > important than total bus length since the varying RC constants > > can cause strange signal skew conditions if the lengths aren't > > matched. > > In addition to this: don't use both the internal and the external > connector of a SCSI adapter (assuming the internal & external > connector are on the same bus). The differences in external versus > internal (flat)cable are also no-good. I haven't seen much problem with that as long as you use consitent impedence cables. Ie, as long as the whole bus is run with 90 Ohm cable it works, but you have to spend the bucks for high quality teflon or twisted pair internal cables to get any place close to 90 ohms. And of cource you have to buy double sheilded twisted pair 90 ohm external cables with gold contacts. One thing about doing what you state above is that this means you are pretty much stuck with what ever type of termination is on the SCSI controller. I don't know of any controllers that have FPT on them, and for long busses FPT is a _must_. > > 6) Honor the 6m and 3m bus length limits. 6m is the longest for 10MHz > > and 3m is the longest for 20MHz single ended SCSI. This assumes > > proper stub/gap matching, so real life buses may need to be shorter. > > Hm. The company I work for (==DEC) uses the rule that a single ended > F10 (10MHz) bus can be a max of 3 meters. Justin, can you double check these numbers, I'm seeming to recall that it is 6 meters for async upto 5Mhz, 3 meters for sync upto 10Mhz and 1 meter for sync upto 20Mhz, unless your running differential, which is a whole other set of numbers. -- Rod Grimes rgrimes@gndrsh.aac.dev.com Accurate Automation, Inc. Reliable computers for FreeBSD