Diff Sense Timing
Aloisi, Paul
aloisi at unitrode.com
Fri Nov 20 08:38:45 PST 1998
* From the T10 (formerly SCSI) Reflector (t10 at symbios.com), posted by:
* "Aloisi, Paul" <aloisi at unitrode.com>
*
The 100 milliseconds is a long time, it started out as a 50 Hz ground noise
filter. The jump from 20 milliseconds to 100 came at the end of the LVD SCSI
development. We originally never planned for the bus to switch dynamically
when the LVD Standard was defined. This came as an after thought for hot
plugging, which has major implications. We wanted the bus to go through a
reset cycle if it changed modes. All the speed negotiation should be done,
the best way would be a bus reset when the bus changed modes.
The expanders are required to do a bus reset when the bus changes mode on
one side to inform the system that it needs to renegotiate all the speeds.
It is the only single ended signal on the LVD bus with very low drive.
100 milliseconds is a long time for electronic components to drive into a
short to ground when the bus goes from LVD to SE, all 27 signal Positive
signal lines go into a short. Gene is right to have concerns.
The act of inserting a drive takes time, 10 -100 ms to make connection for
all the contacts. 300 ms seams like a reasonable time for the devices to
detect the bus change with the 100 ms maximum insertion - normally. 300 ms
all the devices come back to the bus with the correct mode and start with a
bus reset to reset all the speeds. Define the full sequence for the system,
we talk about a lot of things in the working groups, but very little of the
reason for the definitions are documented in the standard.
Thank you,
Paul
Paul D. Aloisi Phone: 603-429-8687
Unitrode MS 84 FAX: 603-429-8963
7 Continental Blvd Email: aloisi at unitrode.com
Merrimack, NH 03054 Web site: www.unitrode.com
<http://www.unitrode.com>
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com
More information about the T10
mailing list