Cable swap blows Ultra2 cards?

[Gingerich, Kevin]

* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* "Gingerich, Kevin" <k-gingerich at ti.com>
*
Gary, shorting LVD-SCSI drivers to ground should not cause any damage as
they are current limited. I wouldn't worry about connecting DIFFSENS first
but GND instead. It is most likely an electrical overstress on the silicon
of the host from a charged cable or ground potential difference between the
host's ground and that of the cable. The event is transitory and will not be
seen in a steady-state measurement of the ground potentials. /Kevin/

-----Original Message-----
From: Gary Watson [mailto:gary at trimm.co.uk]
Sent: Wednesday, October 07, 1998 2:27 AM
To: 't10 at symbios.com'
Subject: Cable swap blows Ultra2 cards?


* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* Gary Watson <gary at trimm.co.uk>
*

Hi all,
We are hearing an alarming number of reports that attaching a single ended 
disk subsystem to a powered-up PC with an LVD host adapter in it can result 
in catastrophic failure of the host adapter.  Obviously people ought to 
switch off everything before changing SCSI cables around, but technicians 
tend not to bother as the conventional wisdom is that you don't need to for 
SCSI.  Has anyone else seen this?  Could it be related to the fact that the 
host adapter is operating its multimode drivers in differential mode by 
default, and when a live cable is plugged on, DIFFSENS is not the first 
wire to mate, and effectively a bunch of active differential drivers get 
shorted to ground for a second or two?  I know this theory has a big hole 
in it as the host adapter ought to be tristated, but it's all I can think 
of unless there is some kind of odd construction being used in the 
multimode output pads which blows up if the voltage is out of the expected 
range for LVD operation when trying to run in LVD mode.  I thought about 
ground potential between the cabinets, but I've now measured this in two 
failed systems and found that the grounds were bonded perfectly.  So I am 
at a loss.  Is it possible to deduce the problem by post-mortem examination 
of the silicon?

Gary Watson			Trimm Technologies Ltd.
Engineering Manager		A Subsidiary of Robroy Industries, Inc.
gary at trimm.co.uk		Nottingham NG8 6AR  ENGLAND
http://www.trimm.co.uk		+44 115 951 9951


*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com