Ahh Kevin, you know that SCSI has to be properly terminated to work within
specs.  Some busses are more or less tolerant, and some are
auto-terminating, but the A3000 interface is not one of those AFAIK.  If the
controller is in the "middle" of the cable (typical if you have both
internal and external devices connected), you usually have to terminate both
ends of the cabling.

Bus noise the the bane of any signal travelling down a wire.  Proper
termination is intended to minimize signal reflections (i.e noise) caused by
the electrons hitting the end of the wire and seeing what is essentially a
dramatic impedance change.  The termination corrects for that impedence
change to make the wire infinitely long, electrically speaking.

  - Bruce Drake

"Kevin P. Neal" wrote:

> On Wed, Jan 10, 2001 at 07:49:24PM +0100, Ignatios Souvatzis wrote:
> > On Tue, Jan 09, 2001 at 09:40:46PM -0500, Kevin P. Neal wrote:
> > > BTW, has anybody ever noticed that at least the 3000 (at least mine,
> > > anyways) behaves terribly and randomly when termination isn't right
> > > on one of the SCSI busses in it? Quite irritating. I was getting pmap
> > > errors and random programs crashing before I removed the 2091 card.
> >
> > Thats probably rather the Z2 memory.
>
> Nope. I didn't tell loadbsd-2.14 to have the kernel use multiple memory
> segments, and the kernel said it was only using 16 megs (with less than
> that available for use, of course). I even tried running with less than
> 16 megs allowed (loadbsd option).
>
> I've still got some sort of problem with the 2091 gone, so I'm not in
> the clear yet. *sigh* At least things break less often now.
> --
> Kevin P. Neal                                http://www.pobox.com/~kpn/
>
> "It sounded pretty good, but it's hard to tell how it will work out
> in practice." -- Dennis Ritchie, ~1977, "Summary of a DEC 32-bit machine"