September proposals/Timing diagram text

Elliott, Robert (Hou) Robert.Elliott at COMPAQ.com
Sun Aug 29 16:22:41 PDT 1999


* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Hou)" <Robert.Elliott at COMPAQ.com>
*
I've sent two SPI-3 proposals to John.  They should be on the T10 web site
Monday.

	99-225r0 	QAS Arbitration Fairness 
	(new)  		text changes to the Setup and Hold Timings section

The QAS proposal adds a QAS REQUEST counter to targets.  This could be
considered for SPI-3 or moved to SPI-4, in case anyone is still in a hurry
to finish SPI-3.  The other proposal restores some meaning that was lost
when the fast-10, 20, and 40 sections were merged.


One other topic for discussion in the SPI-3 editing session:

Background:  In figures 46 and 47 (the LVD measurement pictures), there are
some *s on the data signal that note: "use the crossing that yields the
shorter SETUP and HOLD time."  This was done so one picture could show data
going 0 to 1 and 1 to 0.

Problem:
The"may be detected" points in the CLOCK signal are somewhat vague.  During
the t(af) and t(ar) regions, "any signal structure may occur."  What if the
signal crosses 0V more than once?  Which 0V crossing is used for the setup
and hold timing measurements?  I assume it is the first one, but that needs
to be clearer. This affects SETUP time, ASSERTION period, and NEGATION
period.  Perhaps * could be placed by each "may be detected" and the the
description expanded to include ASSERTION and NEGATION period.
Alternatively, a note stating that "may be detected" means the first
transition might suffice.

The same problem exists at the "shall be detected" points.  During t(af) and
t(ar), according to the receiver mask, the signal may jump above the [100mV]
level.  The first transition does not necessary serve as the shall detect
point, however.  It's more like the first transition that is followed by a
t(m) region.  The * text could probably be overloaded here; the region
followed by t(m) yields the shortest HOLD time, ASSERTION period, and
NEGATION period.  The note mentioned previously would not work.

The same problem occurs for the data signals.  A single signal (e.g. 0 to 1)
may cross 0V more than once.  The *s are already in place and can be
interpreted as covering the issue.

Bill Ham suggested this wording:

The leading end of a timing duration defined by a nominally rising signal
shall be measured at the time where the signal crosses the 100 mV level and
then proceeds to stay above the 100 mV level for the required time. The
leading end of a timing duration defined by a nominally falling signal shall
be measured at the time where the signal crosses the -100 mV level and then
proceeds to stay below the -100 mV level for the required time.  After the
+-100 mV level is crossed and the required time expires any further
crossings of the +-100 mV level may be examined to determine if another
valid transition has been produced.

The trailing end of a timing duration defined by a nominally falling signal
shall be measured at the time where the signal first crosses the 0 mV level
after having stayed above 100 mV for the required time before reaching the 0
mV level. The trailing end of a timing duration defined by a nominally
rising signal shall be measured at the time where the signal first crosses
the 0 mV level after having stayed below -100 mV for the required time
before reaching the 0 mV level.   After the 0 mV level is crossed and the
required time expires, any further crossings of the 0 mV level may be
examined to determine if another valid transition has been produced.

 Note that after a transition has occurred in one direction that another
transition in the opposite direction is required before another transition
in the same direction can occur.



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






More information about the T10 mailing list