May SSWG Minutes
R. S. Shergill
rss at berlioz.nsc.com
Tue May 31 12:20:32 PDT 1994
Accredited Standards Committee Operating under the procedures of
The American National Standards Institute.
X3 Secretariat, Computer and Business Equipment Manufacturers
Association (CBEMA)
1250 Eye Street NW, Suite 200, Washington, DC 20005-3922
Telephone: 202-737-8888 (Press 1 twice) FAX: 202-638-4922 or 202-
628-2829
X3, Information Processing Systems
Doc. No. X3T10/94-_____R0
Date: May 28,1994
Project:
Ref. Doc.:
Reply to: J. Lohmeyer
To: Membership of X3T10
From: Shergill/Lohmeyer
Subject: Minutes of X3T9.2 FAST-20 SCSI Ad-hoc May 5,1994
Agenda
1.Opening Remarks
2.Attendance and Membership, Introductions
3.Approval of Agenda
4.Document Distribution
5.Old Business
5.1 Testing Update (Ham)
5.2 Review of Outstanding Issues
a) System Level Issues
b) Chip Level Issues
5.3 Review of Project Proposal
5.4 Review of Current FAST-20 Draft
6.Meeting Schedule
7.Adjournment
Results of Meeting
1.Opening Remarks
John Lohmeyer convened the meeting at 9:30 AM. John thanked Jim
McGrath of Quantum for hosting the meeting.
As is customary, the people attending introduced themselves. A
copy of the attendance list was circulated for attendance and
corrections.
It was stated that the meeting had been authorized by X3T10 and
would be conducted under the X3 rules. Ad hoc meetings take no
final actions, but prepare recommendations for approval by the
X3T10 task group. The voting rules for the meeting are those of
the parent committee, X3T10. These rules are: one vote per
company; and any participating company member may vote.
The minutes of this meeting will be posted to the SCSI Reflector
and will be included in the next committee mailing.
2. Attendance and Membership, Introductions
Attendance at working group meetings does not count toward
minimum attendance requirements for X3T10 membership. Working
group meetings are open to any person or company to attend and to
express their opinion on the subjects being discussed.
The following people attended the meeting.
Name Organization
------------------------ ---------------------------
Mr. Wally Bridgewater Adaptec, Inc.
Mr. Mike Dreitlein Adaptec, Inc.
Mr. Norm Harris Adaptec, Inc.
Mr. Barry Litt Adaptec, Inc.
Mr. Craig Stuber Adaptec, Inc.
Mr. Mark Knecht Advanced Micro Devices
Mr. Mark Jander AT&T/ NCR Microelectronics
Mr. John Lohmeyer AT&T/ NCR Microelectronics
Mr. Dale Smith Cirrus Logic
Mr. Louis Grantham Dallas Semiconductor
Dr. William Ham Digital Equipment Corp.
Mr. Ron Roberts Maxtor Corp.
Mr. Robbie Shergill National Semiconductor
Mr. Ting Le Chan QLogic Corp.
Mr. Farbod Falakfarsa Quantum Corp.
Mr. James McGrath Quantum Corp.
Mr. Brian Davis Seagate
Mr. Stephen G. Finch Silicon Systems, Inc.
Mr. Vit Novak Sun Microsystems, Inc.
Mr. Pete Tobias Tandem Computers
Mr. Paul D. Aloisi Unitrode Integrated Circuits
Mr. John Osterlund Zadian
22 People Present
3. Approval of Agenda
The agenda was approved as proposed.
4. DOCUMENT DISTRIBUTION
The minutes of the March 31, 1994, ad-hoc meeting were distributed by R.
Shergill.
5. OLD BUSINESS
5.1 TESTING UPDATE
Bill Ham presented the results of further testing he had performed since
the last meeting.
Bill suggested Fast-360 as a name in order to use same units as serial
interfaces. John Lohmeyer suggested deferring this to the next plenary
meeting - this was done.
Bill has taken additional data on fully loaded single-ended bus.
At worst-case positions along the bus the assertion and negation levels
are badly degraded. 16 devices don't look feasible, 8 devices generally
work.
Wally Bridgewater asked if 3 meter cable with 16 devices would be
actually better because of wider separation between devices. Bill
thought that maybe possible.
Bill showed a 16-device scope picture where the waveform at the 4th
position from the driver barely gets to 2 volts and doesn't get below 1
volt. End-bus driving is worse than mid-bus driving. With 9 devices,
only the 4.9" stub case is marginal at 20MT/s.
Bill stressed that the differential bus has no problems - he has tested
30 MT/s operation at up to 21.5 meters.
Bill said that limiting the number of devices is best from backward
compatibility standpoint. The other options are - limit node capacitance
(will mean no switchable terminators), narrower thresholds, etc. Bill
added that he had tried doubling up the transceivers. This showed only
marginal improvement.
Bill concluded with the following suggestions -
- 1.5 meters
- 8 devices
- 4" max stubs
- 25pf
- cable impedance greater than 80 ohms
- plus all others - active negation, slew rate control, narrower
receiver threshold.
Norm Harris commented that all else is fine but Adaptec believes 3
meters is possible - especially if cable impedance is increased to above
100 ohms.
Vit Novak asked why the stub length can't be lower than 4"; because one
of the applications for FAST-20 is going to be backplane bus.
John Lohmeyer asked if the TI driver's extra strength - due to its
bipolar componentery - weakens this analysis? Bill Ham said the TI chip
has a pure CMOS driver.
Brian Davis commented that Seagate silicon's pin capacitance is about 12
pF. Paul Aloisi added that the termination chips' capacitance is about 3
pF now - there are actually devices out there that have 15 pF total
capacitance.
5.2 REVIEW OF OUTSTANDING ISSUES
A) SYSTEM LEVEL ISSUES
Norm Harris presented Adaptec's proposal:
- Backing up to 1.5 meters will hurt SCSI in the market with respect to
ATA-2. Therefore, we should stay with 3 meter cable length.
- Specify impedance to be 100-ohm +/- 10
- Specify maximum device capacitance to be 20pF
- Specify maximum devices to be 7; although will go along with the
agreement that exists currently on 8
- Specify 3:1 load spacing to stub length ratio.
Dale Smith asked if the termination can be strengthened. This will mean
stronger drivers - which will mean no backward compatibility. No one
wanted to do this.
Vit Novak pointed out that instead of simple cable impedance, what's
more important is the consistency of impedance all along the cable. 100
ohms is not good because it can't be achieved on backplanes easily. So
Vit suggested 90 +/-10. Adaptec argued for 95 +/-10 if 2:1 ratio is
desired. Vit agreed to this because he can get upto 85 on the backplane.
Bill Ham suggested 90 +/-6 for req/ack pair only and 90 +/-10 for all
other signals. This sounded like a good way to go if the cable folks can
do it. The group will expect the cable guys to provide input at
Harrisburg. Norm will put this info on the Reflector and also call a
couple of cable people.
On the issue of cable length, John Lohmeyer did not want to accept 3
meters until Bill Ham has tested it.
On capacitance, the group considered 20 pf as the target, broken down
this way:
12 for chip,
3 for chip term,
3 for trace
2 for connector.
Jim McGrath came into the meeting at about this time and objected to 20
pF limit. He hasn't seen a chip termination that specs worst-case
capacitance of less than 5-6 pF. 20pf limit maybe required for a fully
loaded bus, but what about the typical desktop application where you
only need 3-4 devices maximum? Paul Aloisi commented that his latest
termination chips have low capacitance - but McGrath raised the issue of
cost.
Robbie Shergill suggested that total device capacitance for the system
could be specified in addition to 25pF max limit per device. This way
you could configure a fully loaded system by using lower-capacitances
devices, or configure a lightly loaded system with 25pF devices. Adaptec
argued that this will cause two classes of devices. After much
discussion the group came back to one spec - 25pf - with an implementors
note or annex that explains that a fully loaded system should minimize
node capacitance. Vit Novak volunteered to write such a note.
Steve Finch raised the issue of how this capacitance is to be measured
and how applicable that capacitance number would be to the application.
Bill Ham suggested that ground offset should be specified. He suggested
50mv - no arguments. Bill Ham also pointed out that mixing 50 and 68-
wire cables (narrow/wide) would affect the signal delay skew and
therefore should not be allowed. This was also agreed to. In addition,
it was decided summarily that passive termination is not allowed with
Fast-20.
B) CHIP LEVEL ISSUES
Hysteresis:
Adaptec, NCR, SSI, and National did not want the levels changed from the
ones agreed upon in March meeting. It'll also cause problems for a
receiver to meet older spec. Ham will test to see if using 8 devices
will remove this problem anyway.
Slew rate:
Giles had commented on the reflector that 600mv/ns should be changed to
540mv/ns to remain compliant with SPI. Adaptec agreed to this.
Test circuit for ac-specs other than slew-rate:
NCR and Q-Logic said that 200pF max is what the load spec their chip's
AC parameters specs. No one had objections to using this as a test load.
Voh max:
Mark Knecht said that 5.25 v currently in SPI is too high for a 3.3v
technology. He wanted some number in the 3.7 - 4.0 volt range. his
concern is for DC levels higher than this - he understands that
reflected transient voltages will get up higher than this. Mark will do
a proposal. John Lohmeyer commented that it may not be too late to even
change SPI in this respect.
5.3 Review of Project Proposal
John Lohmeyer reminded everyone about the ballots on Fast-20 proposal -
ballots are due by May 13.
5.4 Review of Current FAST-20 Draft
No detailed review was done.
6. MEETING SCHEDULE
May 18, 1994 - in Harrisburg, PA
June 2, 1994 - in Milpitas, CA. (Will decide in Harrisburg if this
meeting is still required).
7. ADJOURNMENT
The meeting was adjourned at 1:00 PM.
More information about the T10
mailing list