Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!nosc!ucsd!ucsdhub!hp-sdd!ncr-sd!ncrlnk!ncrwic!encad!entec!jlohmeye
From: jlohmeye@entec.Wichita.NCR.COM (John Lohmeyer)
Newsgroups: comp.periphs
Subject: Re: Questions on SCSI device interfacing...
Summary: Build SCSI cables right.
Keywords: SCSI
Message-ID: <294@entec.Wichita.NCR.COM>
Date: 2 Sep 88 21:45:07 GMT
References: <433@icus.UUCP> <704@hodge.UUCP> <698@viscous> <14901@apple.App <8551@cup.portal.com>
Reply-To: j.lohmeyer@Wichita.NCR.COM (John Lohmeyer)
Followup-To: comp.periphs
Organization: NCR Corporation, Wichita, Kansas
Lines: 30

In article <8551@cup.portal.com> Thad Floryan writes:
> ... text deleted...
>One thing I've noticed (after adding over 1GB HD and a Fujitsu 190MB tape
>drive to my Amiga's SCSI bus) is that some device manufacturers take liberties
>with SCSI wires 20,22,24,28,30, and 34; some tie these to ground, some tie
>them to +5VDC, and some leave them (properly) unconnected.
                                    ^^^^^^^^^^
These pins should properly be tied to ground.  Unterminated signals can cause
noise problems.  Anyone who ties these lines to +5VDC not only violates the
SCSI standard but will cook some fuses (or failing that, some power supplies).

>Another caveat with multiple devices on a given bus is to be SURE that only
>one is supplying +5VDC on SCSI pin 26 for terminator power.  Some mfrs provide
>jumpers to enable/disable, some feed thru a diode, some feed the power out
>"raw", and some do nothing.

The best thing to do is have all initiators (host adapters -- there may be
more than one) provide TERMPWR through a diode and a fuse.  This is what
SCSI-2 requires.  Targets may also provide TERMPWR through a diode and a fuse.

Your comments on daisy-chaining are quite correct.  I like your diagram. Apple
and other vendors opted for the DB-25 connectors to save space. Obviously,
something had to go so they eliminated some ground signals. They may have shot
themselves in the foot.  These grounds affect their ability to transfer data
at high speeds.  At least they kept the 50-pin connectors on their external
boxes. If a future Apple processor needs really fast transfers, it will
probably have to use all 50-pins.  At least the peripheral boxes won't have
to be scrapped.


John Lohmeyer, X3T9.2 (SCSI) Chairman   j.lohmeyer@Wichita.NCR.COM