Minutes of SPI-2 Working Group -- April 18, 1997

John Lohmeyer John.Lohmeyer at symbios.com
Sun Apr 20 21:39:19 PDT 1997

* From the SCSI Reflector (scsi at symbios.com), posted by:
* John Lohmeyer <John.Lohmeyer at Symbios.com>
Minutes of SPI-2 Working Group                                 T10/97-160r0

Accredited Standards Committee*
National Committee for Information Technology Standards (NCITS)

                                             Doc. No.: T10/97-160r0
                                                 Date: April 20, 1997
                                              Project: 1142-D
                                            Ref. Doc.:
                                             Reply to: John Lohmeyer

To:         Membership of T10

From:       Ralph Weber, Secretary T10
            Larry Lamers, Vice-chair T10
            John Lohmeyer, Chair T10

Subject:    Minutes of SPI-2 Working Group
            April 18, 1997 -- San Jose, CA


1. Opening Remarks

2. Approval of Agenda

3. Attendance and Membership

4. Symmetric vs. Asymmetric Issue
   4.1 STA Recommendations to the SPI-2 Working Group [Mason]
   4.2 The case for Switching to Symmetric Drivers in SPI-2 [Bastiani]
   4.3 The case for Staying with Asymmetric Drivers in SPI-2 [Steele]
   4.4 Discussion

5. Universal backplane [Wallace/Barnes]

6. Changing driver modes when hot plugging (96-270r1) [Penokie] {May meeting}

7. Single-ended termination (96-245r2) [Wallace] {May meeting}

8. Proposed clarification to Fig. 24 (97-115) [Ham]

9. Integration Issues [Lamers]

10. Bus Set Delay Reduction (97-116) [Ham]

11. Hot-Plugging Data (97-144r0) [Ham] {May meeting}

12. REQ/ACK Glitch Filters [Ham]

13. Requirements for Fast 100 operation [Ham]

14. LVD backplane testing (Fast-40 and Fast-80) [Ham]

15. Meeting Schedule

16. Adjournment

                              Results of Meeting

1.    Opening Remarks

John Lohmeyer, the T10 Chair, called the meeting to order at 8:31 a.m., Friday
April 18, 1997.  He thanked Norm Harris of Adaptec 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 and changes:

      4.2 The case for Switching to Symmetric Drivers in SPI-2 [Bastiani]
      4.3 The case for Staying with Asymmetric Drivers in SPI-2 [Steele]
      4.4 Discussion
      7. Single-ended termination (96-245r2) [Wallace] {May meeting}

3.    Attendance and Membership

Attendance at working group meetings does not count toward minimum attendance
requirements for T10 membership.  Working group meetings are open to any
person or organization directly and materially affected by T10'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
Mr. Wally Bridgewater  V  Adaptec, Inc.             wally at eng.adaptec.com
Mr. Tak Asami          V  Adaptec, Inc.             asami at itc.adaptec.com
Dr. Robert Selinger    V  Adaptec, Inc.             selinger at adaptec.com
Mr. Tom Schneider      V  Adaptec, Inc.             schneid at itc.adaptec.com
Mr. Bill Gintz         V  Adaptec, Inc.             bgintz at corp.adaptec.com
Mr. Zack Mihalis       V  Adaptec, Inc.             zack_mihalis at corp.adaptec
Mr. Vincent Bastiani   V  Adaptec, Inc.             bastiani at corp.adaptec.com
Mr. Thomas W. Martin   V  Adaptec, Inc.
Mr. Joseph Tupy        V  Ancot Corp.               jozef at ancot.com
Mr. Charles Tashbook   P  Dallas Semiconductor      charles.tashbook at dalsemi.
Mr. Michael Smith      A  Dallas Semiconductor      mike.smith at dalsemi.com
Mr. Siegfried Schmalz  V  Dallas Semiconductor      schmalz at dalsemi.com
Mr. Robert C. Chang    V  Data Technology Corp.     dtc_eng at ix.netcom.com
Dr. William Ham        A# Digital Equipment Corp.   ham at subsys.enet.dec.com
Mr. Chris Nieves       A# Fujitsu                   cnieves at fcpa.fujitsu.com
Mr. Robert Liu         P  Fujitsu Computer          rliu at fcpa.fujitsu.com
Mr. Larry Ko           O  Initio Corp.              larryk at inito.com
Mr. Paresh Borker      V  Initio Corp.              pareshb at inito.com
Mr. Louis Grantham     A  Linfinity Micro           lgdatcom at ix.netcom.com
Mr. Alan Littlewood    O  LSI Logic                 alanl at lsil.com
Mr. Karl Nakamura      V  LSI Logic                 karln at lsil.com
Mr. Wayne E. Werner    O  Lucent Technologies       wewerner at lucent.com
Mr. Brian Davis        V  Mylex Corp.               briand at mylex.com
Mr. Doug Fields        V  Mylex Corp.               dfields at mylex.com
Mr. Skip Jones         P  QLogic Corp.              sk_jones at qlc.com
Mr. Ting Li Chan       A  QLogic Corp.              t_chan at qlc.com
Mr. James McGrath      P  Quantum Corp.             JMCGRATH at QNTM.COM
Mr. Henry Wong         V  Quantum Corp.             hwong at asic.qntm.com
Mr. Richard Uber       V  Quantum Corp.             duber at tdh.qntm.com
Mr. Dana Hall          V  Quantum Corp.             dhall at tdh.qntm.com
Mr. Michael T. LoBue   V  SCSI Trade Association    LoBue at scsita.org
Mr. Daniel (Dan) F.    O  Seagate Technology        daniel_smith at notes.seagate
Smith                                               .com
Mr. Mike Robinson      V  Seagate Technology        robinson at cdg.seagate.com
Mr. Michael R. Ham     V  Silicon Graphics Inc.     mham at engr.sgi.com
Mr. Vit Novak          A  Sun Microsystems, Inc.    vit.novak at sun.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. Frank Gasparik     V  Symbios Logic Inc.        frank.gasparik at symbios.com
Mr. Harry Mason        V  Symbios Logic Inc.        harry.mason at symbios.com
Mr. David Steele       V  Symbios Logic Inc.        david.steele at symbios.com
Mr. Jack Shiao         O  Tandem Computers          jack at loc3.tandem.com
Mr. Brett Philip       V  Temp-Flex Cable Inc.      brettp at ix.netcom.com
Mr. Kevin Gingerich    O  Texas Instruments, Inc.   k-gingerich at ti.com
Mr. Ted W. Pickerrell  V  Unitrode Corp.            pickt at uicc.com
Mr. Gregory Kapraun    V  Western Digital Corp.     kapraun at wdroc.wdc.com

47 People Present

Status Key:  P    -  Principal
             A,A# -  Alternate
             O    -  Observer
             L    -  Liaison
             V    -  Visitor

4.    Symmetric vs. Asymmetric Issue

4.1   STA Recommendations to the SPI-2 Working Group [Mason]

Harry Mason presented the recommendations from the SCSI Trade Association
resulting from their meeting of 14 March 1997.   The recommendations for
Ultra3 included a throughput of at least 80 Mega-transfers per second,
preservation of Ultra2 cable length and connectivity, and full backwards and
forwards compatibility.  The complete STA recommendations can be found in 97-

4.2   The case for Switching to Symmetric Drivers in SPI-2 [Bastiani]

Vince Bastiani presented data showing cable loss occurring on signals with
frequencies above 100 MHz.  He noted the dramatic signal degradation between
100 and 150 MHz.  He continued by showing signal traces for both symmetric and
asymmetric transmission on a no-loss and a loss cable.  He concluded that the
loss in the cable drives the differential signals under bias apart, resulting
in distortion of the differential signal crossings and loss of the zero
crossings in the sum of the differential signals.

Vince next presented data concerning the assertion and negation voltages under
three bias levels.  Vince concluded that asymmetric requires more power than
symmetric.  Bill Ham noted a potential 100 millivolt advantage for symmetric.

Wally Bridgewater presented additional data comparing symmetric and
asymmetric.  Wally showed how his asymmetric data reflected a case on the edge
of the signal balance tolerance specified in SPI-2.  Jim McGrath asked if the
data showed a problem in the tolerance in SPI-2.  When the answer was no, Jim
observed that the presentation might be interpreted to mean that the committee
will need to tighten the signal balance tolerance to reach 80 mega-transfers
per second.  Jim suggested that such a tightened tolerance could be an
expected result work on moving the 40 mega-transfers per second SPI-2 to an 80
mega-transfers per second future definition.

Bob Selinger concluded the presentation by presenting three roadmap choices
for the transition between Ultra2, Ultra3, and beyond.

The slides from Vince's and Bob's presentations can be found in 97-164.
Wally's additional data can be found in 97-165.

4.3   The case for Staying with Asymmetric Drivers in SPI-2 [Steele]

Dave Steele described the history and basis of the first symbol problem.  He
concluded that differences between asymmetric and symmetric offer no
advantages for either asymmetric or symmetric with respect to the first symbol
problem.  Dave then turned to the issues of the biased terminator.

Dave's next topic was driver current and actual signal measurements.  Dave
showed several signal traces of a FAST-40 LVD part running at 80 MHz,
including traces showing setup and hold times.  Dave presented calculations
showing that power usage is the same to achieve equal signal levels in both
symmetric and asymmetric cases.  Wally Bridewater noted that it would not
always be true.  Jim McGrath observed that the difference really is a matter
of how close the signal balance is to the ideal (or how much tolerance away
|from ideal the design or the specification allows).

The slides from Dave's presentation can be found in 97-159.

Frank Gasparik presented a transmission line model.  He described the process
by which the model and particularly the cable model was developed.  Frank then
showed how the model can produce the ratio of signal heights shown in one
sample of differential data from the March Adaptec presentation.  Frank's
results matched the observations made earlier in the day.  The slides from
Frank's presentation can be found in 97-166.

4.4   Discussion

Near the beginning of the discussion period, Bill Ham made the following

o     use bias termination, always
o     drive either symmetrical or asymmetrical for low speeds
o     use symmetrical drivers with adjustable offset receivers, applying the
      offset for Fast-80 during the data phase
o     use asymmetrical drivers for Fast-40 with 0 offset receiver
      use symmetrical drivers for Fast-80 with offset receiver
      (must negotiate for mode)

Kevin Gingerich described the transmission line problems of faster speeds and
concluded that shorter cable lengths would be needed to address the inter-
symbol interference (ISI) problem.  Bill Ham added that introducing expanders
would have the same effect.  Bill and Kevin also noted that point-to-point is
the easiest transmission line to work with.  Fast-40 was described as being at
the point of diminishing returns, with respect to transmission speeds unless
ISI is dealt with.

Three ideas for handling ISI were listed: 1) decrease the strength of
subsequent same value signals, 2) increase the strength of the first signal
after a string of opposite signals, and 3) increase the duration of the first
signal after a string of opposite signals.

Wally Bridewater described the differences between symmetrical and
asymmetrical drivers as follows:

1.    In asymmetric, the designer must pick a delta current, relative to the
   bias.  When multiplied by the AC impedance of the cable, the delta current
   produces different signal voltages (depending on the AC impedance of the
2.    Tolerance mismatch problems in asymmetric; the delta plus or minus
   tolerance of the termination voltage bias
3.              Assertion Voltage      Negation Voltage
   Asym.    140%                60%
   Sym.     100%              100%

Several people noted that Wally was comparing ideal symmetric drivers to
case asymmetric drivers.

Mike Robinson stated that Seagate is committed to products based on the
current LVD definition.  He continued, noting that the "train has left the
station" for Ultra2.  John Lohmeyer reported similar sentiments expressed by
George Penokie on behalf of IBM in a message to the SCSI Reflector.

The group reviewed the goals set for Ultra3 by the SCSI Trade Association.
This was followed by a lengthy meandering discussion of technical details of
all the information presented to this point.  At one point, Bill pressed for a
discussion of silicon design issues for the adjustable offset receiver that
was a key component of his earlier proposal.  But, no resolution was reached
because some designers wanted to develop new simulation data before making a

Many ideas for Ultra3 were discussed.  How to build terminators that switch
bias off for symmetric data signaling was discussed.  How to tell switching
terminators to switch was discussed.  Bill Ham proposed that both the leading
and trailing edges of the REQ and ACK signals be used for data clocking.

Bill Ham tried to gain the consensus of the group that no changes will be made
to the stabilized definition of LVD SCSI in SPI-2 revision 11.  Consensus was
not arrived at quickly.  After about thirty minutes of discussion, John called
for requests to recommend unstabilization of the LVD SCSI definition in SPI-2
revision 11 to the plenary.  There were no such requests.

Bill Ham led the group in an attempt to develop a list of ways to get parallel
SCSI to faster transfer rates (above Fast-40).  The resulting list was:

o     Switching termination (no bias at faster transfer rates)
o     Offset receivers
o     Pulse pre-compensation (voltage or time)
o     Use of both edges of REQ and ACK
o     Data encoding similar to 8b/10b (the 8b/10 code was rejected as being
      unlikely to work with parallel signals)
o     Expanders (or shorter segments)

Bill noted that timing budgets, tolerances, etc. will be critical in all
aspects of the Ultra3 definition process.  Bob Selinger and Skip Jones
questioned whether the last item on the list really met the goals brought to
the group by STA, and Bill, in effect, moved the last item to a different

Bob Selinger asked that use of symmetric signaling be added to the list as way
to make parallel SCSI work at faster transfer rates.  Bill disagreed and said
that in his opinion the differences between symmetric and asymmetric will be
swamped by the same problems at faster transfer rates.  Dave Steele expressed
the opinion that asymmetric will work equally as well as symmetric.  After a
little additional discussion, Bill was willing to agree that symmetric would
have slightly better margins at faster transfer rates.

Skip Jones and several others noted that several other issues deserve
consideration in the development of Ultra3, including: increased addressing,
lower overhead, arbitration fairness, and several other topics discussed at
previous working group meetings.

After a brief discussion of other topics, Bill continued his efforts to led
the group toward developing conclusions.  The discussion of asymmetric verses
symmetric resurfaced quickly and Kevin Gingerich was called on to show the
advantages of symmetric.  His conclusion was that symmetric gains about 0.5
mA.  Based on the specified driver currents for LVD, Bill Ham described the
advantage as being 10% to 12%.  Kevin and Wally discussed how the calculations
should be viewed, with respect to the experiments.  The specific percentage
range was debated both upward and downward.  After several minutes of
discussion, Bill was able to start the following list:

Technical Conclusions

o     For Fast-80 operation symmetrical drivers offer 5% to 22% improvement in
      margin (much of the value range resulting from differing opinions, not
o     Fast-40 as presently specified works
o     There are a number of options to consider for implementing faster
      transfer rates (see discussion above)
o     Lack of precise Fast-80 timing and tolerance specs prevented any
      conclusions on Fast-80 operation with present Fast-40 configurations

The group then turned to discussion the capacitance specs that might be
applied to Fast-80.  Bill asked that group members making components bring
their requirements on capacitance load and skew matching to the next meeting.

5.    Universal backplane [Wallace/Barnes]

John Lohmeyer presented a foil (97-167r0) that showed one set of numbers that
yield a `universal' backplane.  He said that Larry Barnes claimed there are
many other valid configurations.  Several questions were raised concerning the
interaction other signal traces with the SCSI signal traces.  Since Bill Ham
has data from a functioning universal backplane (see item 14), John proposed
that this agenda item be removed from future agendas.

6.    Changing driver modes when hot plugging (96-270r1) [Penokie] {May

Dan Smith, Bill Ham and John Lohmeyer discussed practical details of case 4
hot plugging.  The subject matter of the discussion was predominately
tutorial.  As noted on the agenda, the group agreed to discuss George
Penokie's proposal at the May meeting.

7.    Single-ended termination (96-245r2) [Wallace] {May meeting}

8.    Proposed clarification to Fig. 24 (97-115) [Ham]

Bill Ham strongly supported taking the 30 mV specs in 97-115 back to the
original 60 mV values in Figure 24 of SPI-2.  Everybody in the room agreed
with Bill.  Wally Bridgewater asked that Richard Moore be consulted one more
time regarding the 60 mV spec.  Bill also asked that the wording changes
proposed in the figure in 97-115 be considered to be non-technical.  These
changes add the words "may be detected" and "shall be detected" to several
places in Figure 24.

9.    Integration Issues [Lamers]

Action on this item was deferred to the May meeting.

10.   Bus Set Delay Reduction (97-116) [Ham]

Action on this item was deferred to the May meeting.

11.   Hot-Plugging Data (97-144r0) [Ham] {May meeting}

12.   REQ/ACK Glitch Filters [Ham]

Action on this item was deferred to the May meeting.

13.   Requirements for Fast 100 operation [Ham]

Action on this item was deferred to the May meeting.

14.   LVD backplane testing (Fast-40 and Fast-80) [Ham]

Bill Ham presented data from a multi-mode backplane developed in his
laboratory.  He showed a case where the system benefits from addition of an
expander between the long (between boxes) cable and the backplane.  Kevin
Gingerich led Bill to agree that the real problem is inter-symbol interference
(ISI).  Kevin suggested that shortening the cable length is a solution to the
ISI problem.  Bill suggested that introducing expanders would be a good way to
shorten the cable length while maintaining the end-to-end length.

15.   Meeting Schedule

The next general SPI-2 working group meeting is scheduled for Monday May 5,
1997 in Natick, MA.

16.   Adjournment

The meeting was adjourned at 5:25 p.m. on Friday April 18, 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--28800 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