No revision 4 for X3T9.2/93 Rev 3 or X3T10/94 94

Paul D. Aloisi 603-429-8687 aloisi at uicc.com
Tue Apr 26 05:56:18 PDT 1994


X3T9/93 149 Rev 3 renamed to X3T10/94 38

There is no revision 4, the comments made in the SCSI working Group in
Newport Beach would cause system problems for companies other than the 
ones that made the comments. Their systems control the drives that are 
plugged into their systems, most systems do not have that luxury.

From:	CLUSTR::ALOISI       "Paul D. Aloisi 603-429-8687" 26-APR-1994 08:36:56.19
To:	MX%"gfrazier at ausvm6.vnet.ibm.com"
CC:	ALOISI
Subj:	Capacitance loading of Termpwr, and Hot Plugging


Giles,

The point I was trying to make that termpwr is bypassed at the input to
the terminator. This capacitor is on the termpwr line even if the terminator
is disabled. 

The capacitor on the output of the regulator will not be charged if the 
active terminator is disabled, only the capacitor on the input to the
regulator.

Even the passive terminators with the sip resistors removed, leaves the
termpwr bypassing capacitors on the termpwr line.

One vendor requires 47 uF bypassing on both the input and the output
of the regulator. A 47 uF capacitor charged through a cable, fuse,
and diode drops termpwr below 3.55 volts for 100 uS. Limiting the
bypassing capacitor to 10 uF maximum would reduce the termpwr droop 
time to 5.5 uS which the system can ride through with the capacitors
on the output of the regulator.

Some of our customers have had problems when they hot swap a drive
with the 47 uF capacitors. They could hot plug drives with our terminators
disabled, which require 4.7 uF bypassing capacitors, but the system
would not work when drives with 47 uF capacitors are hot swapped.

The terminator in a RAID box at the end of the cable or on the controller
would be allowed to have larger capacitors. 50 uF maximum was proposed.

(See X3T9.2 mailing #5 page 49 through 68) "RAID Issues for SPI"

There was a lot of opposition to the proposal in the March meeting. The
recommendations made by a couple of companies would only work with their
systems and our parts. This would cause a lot of problems with existing
devices. Currently there is no change to SPI proposed. It will be left
as a system problem.


Note: Some people think hot swapping would just be for RAID systems, but 
in the last company I worked for I use to do it regularly on a stand alone
workstation. I use to keep project information on separate disk drives,
and swap drives when working on the different projects. I had all seven
slots used on the bus, all the projects were kept on the same size disk
with the same ID. (It took about 20 minutes to reboot the system, and 5
minutes to shut it down. Loading a disk off of a tape takes about 30 minutes
plus another 30 to save the information that was on the disk.)


Paul Aloisi
Unitrode




More information about the T10 mailing list