Draft Minutes of SPI-2 Working Group 1/6/97
Lohmeyer, John
JLOHMEYE at cossymwest.co.symbios.com
Tue Jan 14 15:49:00 PST 1997
* From the SCSI Reflector (scsi at symbios.com), posted by:
* "Lohmeyer, John" <JLOHMEYE at COSSYMWEST.CO.SYMBIOS.COM>
*
Minutes of SPI-2 Working Group
X3T10/97-110r0
Accredited Standards Committee*
X3, Information Technology
Doc. No.: X3T10/97-110r0
Date: January 14, 1997
Project: 1142-D
Ref. Doc.:
Reply to: John Lohmeyer
To: Membership of X3T10
From: Ralph Weber, Secretary X3T10
Larry Lamers, Vice-chair X3T10
John Lohmeyer, Chair X3T10
Subject: Minutes of SPI-2 Working Group
January 6, 1997 -- Dallas, TX
Agenda
1. Opening Remarks
2. Approval of Agenda
3. Attendance and Membership
4. Universal backplane [Wallace/Barnes]
5. LVD case 4 hot plugging [Barnes]
6. Changing driver modes when hot plugging (96-270) [Penokie]
7. Single-ended termination (96-245) [Wallace]
8. Proposed clarification to Fig. 24 [Ham]
9. Integration Issues [Lohmeyer/Lamers/Ham]
10. Bus Set Delay Reduction [Ham]
11. Meeting Schedule
12. Adjournment
Results of Meeting
1. Opening Remarks
John Lohmeyer, the X3T10 Chair, called the meeting to order at 9:04 a.m.,
Monday January 6, 1997. He thanked Jim McGrath and Mary Turco of Quantum
for
hosting and arranging the meeting.
As is customary, the people attending introduced themselves and a copy of
the
attendance list was circulated.
2. Approval of Agenda
The agenda was approved with the following additions:
9. Integration Issues [Lohmeyer/Lamers/Ham]
10. Bus Set Delay Reduction [Ham]
3. Attendance and Membership
Attendance at working group meetings does not count toward minimum
attendance
requirements for X3T10 membership. Working group meetings are open to
any
person or organization directly and materially affected by X3T10's scope
of
work. The following people attended the meeting:
Name S Organization Electronic Mail
Address
---------------------- -- -------------------------
-------------------------
Mr. Norm Harris P Adaptec, Inc.
nharris at eng.adaptec.com
Mr. Lawrence J. Lamers A Adaptec, Inc. ljlamers at aol.com
Ms. Lisa Huff V AMP, Inc. lisa.huff at amp.com
Mr. Ron Roberts A Apple Computer rkroberts at aol.com
Mr. Richard Wagner P Cable Design Technologies
rwagner at montrose-cdt.com
Mr. Edward Haske P CMD Technology haske at cmd.com
Mr. Siegfried Schmalz A Dallas Semiconductor schmalz at dalsemi.com
Mr. Bruce McLaren V Dallas Semiconductor
bruce.mclaren at dalsemi.com
Dr. William Ham A# Digital Equipment Corp.
ham at subsys.enet.dec.com
Mr. Roger Cummings P Distributed Processing
cummings_roger at dpt.com
Tech.
Mr. George Penokie P IBM Corp.
gop at rchvmp3.vnet.ibm.com
Mr. Dean Wallace P Linfinity Micro
75671.3443 at compuserve.com
Mr. Louis Grantham A Linfinity Micro
lgdatcom at ix.netcom.com
Mr. E. Jake Berzon P NEC Electronics, Inc.
jberzon at asic.mtv.nec.com
Mr. Skip Jones P QLogic Corp. sk_jones at qlc.com
Mr. Richard Uber V Quantum Corp. duber at tdh.qntm.com
Mr. Gene Milligan P Seagate Technology Gene_Milligan at notes.
seagate.com
Mr. Gerald Houlder A Seagate Technology Gerry_Houlder at notes.
seagate.com
Mr. John Lohmeyer P Symbios Logic Inc.
john.lohmeyer at symbios.com
Mr. Ralph O. Weber A Symbios Logic Inc. roweber at acm.org
Mr. Paul D. Aloisi P Unitrode Corporation aloisi at unitrode.com
Mr. Gregory Kapraun V Western Digital Corp. kapraun at wdroc.wdc.com
Mr. Doug Piper P Woven Electronics
doug.piper at internetmci.com
23 People Present
Status Key: P - Principal
A,A# - Alternate
O - Observer
L - Liaison
V - Visitor
4. Universal backplane [Wallace/Barnes]
John Lohmeyer asked what happens next (after the very through discussion
in
Palm Springs). Bill Ham expressed the belief that an informative annex
will
be written. John asked for volunteer(s) to write the annex. Other
possible
locations for the information were discussed. Dean Wallace volunteered
to
write the annex, provided Larry Barnes agrees to review it.
Dean asked how much of the "text book" material should be in the annex.
Bill
stated a preference for all of it. Others (perhaps the uninformed)
suggested
that the "how to build backplanes" information could be included by
reference.
This idea and several others were discussed. Dean received no clear
guidance
on the subject.
5. LVD case 4 hot plugging [Barnes]
John Lohmeyer announced that he had not received Larry Barnes' simulation
data
and thought that it had not been completed due to software problems.
John
noted that he will discuss the simulation with Larry when he returns to
the
office. John said Larry had been assigned to other projects and may not
have
time to finish the simulation. Larry Lamers asked if the data input for
the
simulation were available. John thought he could obtain the data.
Bill Ham, Dean Wallace, Larry Lamers and John discussed the usefulness of
the
simulation (and laboratory testing, for that matter). It seemed that
simulation and testing provided valuable starting points, but could be
overturned by "real world" experience. The details of the simulation
methodology also were discussed.
John agreed that the item will remain on the agenda for at least one more
meeting.
6. Changing driver modes when hot plugging (96-270) [Penokie]
George Penokie presented (96-270) a document proposing additions to
various
SCSI standards that provide for software notification of changes in
transceiver operating mode (particularly LVD to SE; Low-Voltage
Differential
to Single-Ended). John Lohmeyer raised concerns over the asymmetric
nature of
the required notifications. In particular, notification is required for
LVD
to SE but not for SE to LVD. In addition, George received several
editorial
corrections. After much discussion, George agreed to add symmetrical
requirements for notification of changes from SE to LVD.
Gene Milligan suggested that once a bus changes from LVD to SE, the only
way
to change back would be a power-cycle. The group questioned this concept
and
discussed the matter at length. Gene stated that the reason for his
statement
was that once a device enters single-ended mode, it drives DIFFSENS to
ground.
Several other people argued that multi-mode devices would not drive
DIFFSENS
to ground while in SE mode; they would continue to source current on
DIFFSENS
to attempt to re-enter LVD mode. Gene mentioned that some muli-mode
devices
might have a jumper to force single-ended mode. The group agreed that
such
devices should ground the DIFFSENS line when the jumper forces
single-ended
mode, because such devices effectively are only single-ended devices.
John Lohmeyer also discussed the timing issues around transition testing
on
the DIFFSENS line. John reviewed the discussion as recent editing
meetings.
He noted that the most recent editing meeting agreed that the 100
millisecond
transition delay should remain (unchanged) to guarantee filtering of
50/60 Hz
noise on the DIFFSENS line.
The group discussed issues concerning transitions to and from LVD
operation at
length. The discussion was complicated by the introduction of bus
expanders
to the proposed configurations to be covered. The discussion was lengthy
and
detailed. Virtually all present participated in the discussion at one
time or
another. Bus expanders appeared to present the greatest difficulty.
Eventually, the group agreed that a device (initiator or target) that
senses
change in DIFFSENS mode shall generate an internal reset. The group
further
agreed that a bus expander that senses a change in DIFFSENS mode shall
generate a hard reset on the other bus.
George agreed to generate a revision 1 of the proposal, including all the
resolutions reached during this meeting. George will bring the revised
proposal to the next working group meeting.
7. Single-ended termination (96-245r1) [Wallace]
Dean Wallace reported that several comments were received regarding the
first
revision of the proposal. He drew the figure and described the changes
proposed and the comments received from Seagate and others regarding the
proposed changes.
Bill Ham raised two additional concerns. Part of Bill's issue was that
he was
willing to make old terminators illegal in SPI-2, while Dean was
interested in
making minor changes that allow old terminators to comply with SPI-2.
Bill
pressed for resolution of four issues: maximum current, minimum current,
sinking current, and linearity. Dean noted that the proposal must be
revised
further to meet Bill's requirements. The idea of basing terminators on
80
ohms or 70 ohms (instead of 110 ohms) was also discussed.
Dean agreed to integrate the changes discussed by the working group into
a new
revision of the proposal.
8. Proposed clarification to Fig. 24 [Ham]
Bill Ham reviewed the proposed clarification for SPI-2 (rev 11) figure
24. He
noted that some might view the change as technical, but Bill expressed
the
belief that the change is not significant (about 30 pico-seconds in the
timing
budget). After a brief discussion, the group agreed to distribute the
proposal as a separately numbered X3T10 document for review and
consideration.
9. Integration Issues [Lamers]
1. Active negation - map of existence (see X3T10/95-295); map of
intensity
(use F20 envelope)
The group agreed that this issue was considered and resolved at the
last
working group meeting (see 96-257).
2. SE termination, max/min current & sinking of current (see X3T10/96-
222r1); suggest min current 20 milliamps at 0.2 v DC and a max of 25.4
milliamps at 0.2 v DC; don't require driver to sink this to avoid
re-qualify;
add a min at 0.5 v DC; how to deal with non-linear terminators; do we
need a
duty-cycle spec? suggest 20 asserted signals for wide, 12 on narrow,
37 on 32-
bit. Add exception for a contained bus (e.g., laptop) of less than
0.3
meters. Add a 12.5 pf max capacitance for terminator.
The group agreed that this issue was considered and resolved at the
last
working group meeting (see 96-257).
3. Leakage spec increased for LVD multi-mode drivers to 20 micro-amps
-
Resolve by specifying one for each driver type.
The group agreed that this issue was considered and resolved at the
last
working group meeting (see 96-257).
4. Latching & counting - require that they be an atomic action for hot
plugging - possibly an annex or implementation note.
The group agreed that this issue was considered and resolved at the last
working group meeting (see 96-257).
5. Ground - power, logic & ground drivers (25-pin power & ground
connected
(SFF-8040)), signal ground and ground ground for tables defining SE
signals.
The group agreed that this issue was considered and resolved at the last
working group meeting (see 96-257).
6. SE Receiver - hysterisis (.3), input levels (adopt F20), pin
leakage
(same), glitch filtering (enable of first detection of a transition
for 'x'
duration). Need an algorithm to determine 'x'.
The group discussed the desire to use the tighter Fast-20 requirements
(not
SPI) for the described values. Gene Milligan raised concerns about
historical
definitions of the requirements on the DIFFSENS line (for historical
High-
Voltage Differential implementations). Patent issues related to
glitch
filters were discussed; it is important that the standard not require
a
specific implementation to avoid various glitch filter patents that
may exist.
The group agreed to propose that Fast-10 SCSI use the same glitch
filtering
that SPI has, except that glitch filtering would be required for the
assertion
edge, as well as the negation edge. For Fast-20, glitch filtering
would be
required for the assertion edge, as well as the negation edge, and
that the
time during which signal reversals shall be ignored would be 7 ns,
instead of
10 ns. Fast-40 would have all the properties of Fast-20 with a glitch
filtering interval of 4 ns. Thus, x will be 10 for Fast-10 and 7 for
Fast-20
and 4 for Fast-40. The group agreed that filtering would apply only
to the
REQ and ACK receivers.
John asked that the silicon folks examine the above proposal (for REQ
and ACK
glitch filtering) and provide input on its feasibility at the March
SPI-2
meeting.
7. SE Drivers - slew rate (use F20 specs)
The group agreed to used the Fast-20 specs for slew rate.
8. 16 or 32 devices? 32-bit data path? Is it a single segment draft?
Should expanders be included? Suggest that SPI-2 is 16 loads per
segment max;
allow 32-bit data path and 32 SCSI IDs addressability when using VHDCI
connectors with primary and secondary cables
Larry raised the question of how SCSI is described in the foreword and
scope
of the SPI-2 standard. The group discussed various aspects of the
issue;
including cables, protocol chips, bus expanders and several other
hardware
components. Most of the discussion concerned implementation of 32 bit
wide
parallel SCSI busses.
The group agreed to leave the 32-bit data path and stick with a limit
of 16
devices per segment.
9. Case 4 hot plugging - requires SCA-2 connectors to allow
pre-charge, a
resistive contact may also work, needs research to determine max
disturbance.
It was noted that the glitch filter decisions (taken earlier in this
meeting)
tightens restrictions on case 4 hot plugging, at least with respect to
reflected glitches. Bill Ham expressed the belief that better glitch
filtering will not help resolve the case 4 problem. Larry Lamers
suggested
that this means that minimizing the disturbance produced by hot plugging
is
the only avenue to a solution. Lisa Huff indicated that disturbance
reduction
via resistive contacts was not fully designed and tested yet.
The group determined that additional research is required on this
subject.
10. Cable specifications - impedance (loaded & unloaded), skew, wire
gauge,
attenuation
Larry Lamers and Bill Ham asked if SPI-2 should use SPI or LVD/SPI-2
cable
requirements. Bill Ham proposed that cable vendors be asked, "If you did
use
the Fast-20 spec cable, what would that do to LVD?" and "Do the Fast-20
specs
defining the cable and the equivalent requirements in SPI-2/LVD describe
the
same cable?" and "If the two cables are different, what is required to
make
them the same?"
The group agreed to "go for the best cable and one cable for everything."
Bill noted that this may result in SPI-2 requiring a cable that is
different
|from all previously specified cables.
11. Add Q-cable pinouts
The group agreed to just do it.
12. Micro SCSI pinouts
Bill Ham and George Penokie discussed the usefulness of the micro SCSI
connector. Substantial resistance was encountered for including the
micro
SCSI connector. The group agreed to remove describing the micro SCSI
connector from the work list for SPI-2. It was noted that SFF will
document
the micro SCSI connector.
13. Reserved lines
Bill Ham proposed that SPI-2 clearly describe reserved lines as being
ungrounded, continuous from end-to-end, and not terminated. The group
agreed
with most of Bill's proposal, but final resolution of reserved line
handling
in the terminator was left open for additional discussion at a future
meeting.
14. Fair Arbitration
John Lohmeyer described the "Fair Arbitration" issue as a placeholder,
noting
that inclusion of fair arbitration is a work item for the final SPI-2
document. A previous working group has recommended that the IBM fairness
algorithm be included in SPI-2; work is still needed to convert the
proposal's
terminology to SPI-2 terminology.
10. Bus Set Delay Reduction (97-116) [Ham]
Bill Ham presented a proposal to change the Bus Set Delay from 1800 ns to
1600
ns. He described the reasoning behind his proposal. In a pathological
case,
the 1800 ns Bus Set Delay limits the maximum one way domain propagation
time
to 300 ns, under the worst condition. The existing, preferred, one way
domain
propagation delay is 400 ns.
John Lohmeyer described the SCSI timing model, from which the 1800 ns
time was
derived. He expressed the belief that Bill should consider modifying the
inputs to the SCSI timing model, the spreadsheet for the model, which
John
agreed to send to Bill. In the process of discussing the model, Bill
discovered that many other interesting parameters could be modified, if
inputs
to the model are adjusted. A key input to the model was the accuracy of
the
clock signal provided to a chip that is running the parallel SCSI
protocol.
The group digressed to a lengthy discussion of reducing arbitration
overhead.
Changing model inputs was only one topic discussed during the digression.
Other, more significant, changes to reduce overhead also were discussed.
Eventually, the group agreed that representatives of SCSI protocol chip
makers
will discuss the proposed reduction of Bus Set Delay with their design
engineers. Bill agreed to revise 97-116 based on the discussions at this
working group meeting.
11. Meeting Schedule
The next meeting of SPI-2 Editors is scheduled for Thursday February 6,
1997
in San Jose, CA with X3T11 hosted by Sun Microsystems. A subsequent
working
group meeting is planned for Monday March 10, 1997 in San Diego, CA.
The group discussed the need for every-month meetings. The need for
monthly
meetings appeared to be diminishing. Still, the group decided to
continue
monthly meetings for another bi-monthly cycle. Based on this decision,
John
will schedule a meeting in conjunction with the X3T11 meeting in April in
Palm
Springs, CA, hosted by Brocade at the Hyatt Suites (specifics of the
meeting
to be announced during the March X3T10 meeting week).
12. Adjournment
The meeting was adjourned at 5:52 p.m. on Monday January 6, 1997.
--
John Lohmeyer E-Mail: john.lohmeyer at symbios.com
Symbios Logic Inc. Voice: 719-533-7560
4420 ArrowsWest Dr. Fax: 719-533-7036
Colo Spgs, CO 80907-3444 SCSI BBS: 719-533-7950 300--14400 baud
*
* For SCSI Reflector information, send a message with
* 'info scsi' (no quotes) in the message body to majordomo at symbios.com
More information about the T10
mailing list