I can verify that, in that it's exactly what I've recently done. True, also,
that "HV" differential signalling, which was devised mainly to enhance cabling
options (by allowing for much longer cables) in small disk farms before
FiberChannel became a ubiquitous option, is in no way compatible with the
older 5-volt single-ended SCSI, whether wide or narrow.
There's a signal referred to as DIFFSENS that detects whether the interface is
LVD or SE, but it is only present on the more recent SCA types. My FUJITSU
SCA-2 drives, among others, detect this signal and automatically figure out
whether to use single-ended or LVD signalling. It also figures out whether
it's wide or narrow, with a little help from the controller, and is quite
capable of operating in either wide or narrow mode. Drives of the current
generation are capable of data widths up to 64 bits, which width none of my
drives/controllers can process, but the cable capacity is the rate-determining
step in this environment, and if your cables and controller can process SCSI
bus transactions at 20 MHz, then you can transfer the entire word width,
whatever it is in your case, at that rate. Normally, the bus width for older
devices is either 8 or 16 bits.
Where folks get into trouble is when they mix devices with different
synchronous speed capabilities. If they do so without setting the parameters
in the controller to match, then the device selection process can break down
and the entire bus then has to be slowed down to the "lowest common
denominator," so to speak, in order to allow business to be transacted
"normally."
Dick
----- Original Message -----
From: "Gary Hildebrand" <ghldbrd_at_ccp.com>
To: <classiccmp_at_classiccmp.org>
Sent: Tuesday, February 12, 2002 10:05 AM
Subject: Re: SCSI Bus Problem?
<snip>
> Then AFAIK, you can put an Ultra 160 LVD drive on an UW card, and it
> will revert to single ended UW operation. HV differential is a world
> unto itself.
>
> Gary Hildebrand
> ST. Joseph, MO
>
>
Received on Tue Feb 12 2002 - 10:55:11 GMT
This archive was generated by hypermail 2.3.0
: Fri Oct 10 2014 - 23:34:46 BST