Date: January 14, 1992 To: X3T9.2 Membership From: Lawrence J. Lamers, X3T9.2 Secretary Bill Spence, Chair SPI Working Group Subject: January 14, 1992 SPI Working Group Meeting Minutes Bill Spence called the meeting to order at 1:30 p.m. January 14, 1992. He thanked Pat Savage of Shell Corporation for hosting the meeting. The SPI working group is the umbrella for all contact, connector, cable, transceiver issues related to SCSI-3 Parallel Interface (SPI). Bill Spence chairs this working group which is chartered with developing a set of recommendations for the SPI standard that will get the physical plant to a reliable state. The final agenda was as follows: 1. Single-ended Cable Impedance Revisited (92-004) [Spence] 2. Driver MTBF with Aeronics/Trung Le Terminator. 3. Recommendations on Fast S-E SCSI (91-064R3, 91-185) [Steele] 4. New Terminator Concepts (92-028) [Jordan] 5. New Terminator Concept (92-017) [Samela] 6. Review of SPI group inputs for SPI document. 7. SPI Draft Standard. 8. Meeting Schedule. 9. Adjournment. The following people attended the meeting: Name Status Organization ------------------------------ ------ ------------------------------ Mr. Thomas Newman P Adaptec, Inc. Mr. Steven P. Ego O Aeronics Inc. Mr. Tony Castillo O Aeronics, Inc. Mr. Charles Brill P AMP, Inc. Mr. Bob Whiteman A AMP, Inc. Mr. Ken Schreder S AMP, Inc. Mr. Edward Hrvatin P Burndy Corp. Mr. John Geldman A Cirrus Logic Inc. Dr. William Ham A Digital Equipment Corp. Mr. Skip Jones A Emulex Corp. Mr. D. W. Spence A ENDL Associates Mr. Kurt Chan P Hewlett Packard Co. Mr. Dave McIntyre A Hewlett Packard Co. Mr. George Penokie P IBM Corp. Mr. Larry Grasso S IBM Corp. Mr. Thomas Forrer V IBM Corp. Mr. Kevin R. Pokorney A Intellistor, Inc. Mr. William D'Andrea S J.S.T. Corporation Mr. Lawrence J. Lamers P Maxtor Corp. Mr. Bob Masterson P Methode Electronics, Inc. Mr. Frank Samela O Methode Electronics, Inc. Mr. John Lohmeyer P NCR Corp. Mr. David Steele S NCR Corp. Mr. Stephen F. Heil A Panasonic Technologies, Inc Mr. Gerald Houlder A Seagate Technology Mr. Gene Milligan A Seagate Technology Mr. Vit Novak O Sun Microsystems, Inc. Mr. Ricardo Dominguez P Texas Instruments Mr. Dean Wallace O Texas Instruments 29 People Present Status Key: P Principal S Special Interest (frequent visitor) A Alternate V Visitor O Observer Results of Meeting 1. Single-ended Cable Impedance Revisited (92-004) [Spence] The first two people scheduled on the draft agenda not initially being present, Bill Spence led off by reconsidering the 95 +/- 15 ohm impedance of the last SPI working group. Seeking to regain that group's original objective of setting a tight control over cable impedance but arguing that 95 ohms is too high, he proposed a lower, tighter standard: 87 +/- 5 ohms. This figure was to apply to the average impedance of a cable. He presented extensive justification (see 92-004), including example cable specs meeting that standard. Unlike the December group, this group seemed to favor the lower impedance but had a different reason for a looser standard. Kurt Chan stated that HP specs a max impedance difference between core pairs and outide pairs. In his opinion, the standard should address this span. Bill reminded the group of the previous group's concern with maximizing noise margins in short (< 6 m) systems, but asserted even earlier groups' concerns with breaking out of the 6 m straightjacket, trading margin for distance. There is a question, of course, about reaching the 2 volt negation level when using lower impedance cables. How to spec the impedance and convey the right information? There was no shortage of suggestions for Kurt to edit into the standard. Use a range or a tolerance. Each has advantages and disadvantages. A set within a set for a spec. Total delta tighter for individual cable, wider range for several. Figure for lowest pair impedance, highest pair impedance, a statement to limit variation, a statement to bias to higher impedance unless going longer than 6 meters. Kurt promised to try to extract something practical out of it all. 2. Driver MTBF with Aeronics/Trung Le terminator. Steve Ego reported no new input re driver MTBF with Aeronics terminator. 3. Recommendations on Fast S-E SCSI (91-064R3, 91-185) [Steele] David briefly described the changes he had made to 91-064R2 after the December SPI meeting. It was agreed to take these up in the review of the SPI working document to follow. David reiterated his conviction that the limits on active negation driver strength established at the December meeting could be further tightened to the point of respecting the Boulay limits without unduly reducing the driver strength (see 91-185). There was support expressed for this viewpoint, and David agreed to provide input on active negation limits for the next revision of the SPI document. 4. New Terminator Concepts (92-028) [Jordan] Mark presented two very interesting developments. Both appear to be in late development phases. His presentation actually included a page marked "Proprietary Information," although he was aware that the meeting was a public forum which did not accept information as proprietary. He assured John Lohmeyer that the Proprietary status no longer applied and furnished a "clean" page for the published minutes. The first development is another non-linear terminator. Mark did not have a terminal V-I characteristic to show, but he described it as having very much the shape of the "Ideal" characteristic presented in 91-194; i.e., a fairly vertical (20 ma current source) region below 2.5 v, and a fairly flat (3.0 v clamped) region when sourcing current below 20 ma or when sinking current. As longas one device in a data exchange is at a bus end, his curves showed negation signals above 2.5 v with 70-ohm cable. The second development is an active negation emulator, designed to establish good noise margin even when a data exchange is between two devices interior to the bus. It is designed to feed a current pulse into any node which rises above a critical level, thus ensuring a clean rise to the negated state. The current feed is disabled whenever the signal is above about 2.5 volts. Mark asked for reaction and suggestions. One suggestion was that it could be of great benefit when applied at the terminator sockets of open-collector host adapters which are located interior to the bus, so that they would then behave as though located at a bus end. Mark estimated that his device might add 4 pf of capacitance. Further discussion: If current is coming from TERMPWR this could be a problem depending on number of signals acted on. Time delay of boost is the biggest hurdle. Could seriously affect the skew budget. Also the effect of multiple one-shots could stretch the signal duration beyond spec. 5. New Terminator Concept (92-017) [Samela] Frank also presented a new terminator design, aimed at providing good negation signal levels even with relatively low impedance cables. He even had an acronym: SLICK, for Signal Line Increased Current Kicker. According to his block diagram, it sources a customer-specified amount of excess current into an asserted line, with the current dropping back to 24 ma if the assertion lasts longer than 40 ms (may be customer-specified). The amount by which the sourced current exceeds 24 ma is chosen to provide a high level of negation signal in whatever line impedance the customer uses. Turn on and turn off is controlled to prevent spikes. Frank dealt with the question of the 48 ma limit on driver current by saying that his device "takes full advantage of the fact that controller drivers are capable of sinking short pulses of high current." It was pointed out that in synchronous data transfer, the -REQ and -ACK signals often maintain roughly 50% duty cycles--i.e., each "short pulse" is followed by an equally short off time and then followed by another pulse. 6. Review of SPI group inputs for SPI document. With Editor Kurt Chan present, the inputs proposed over the past meetings were reviewed. Many of them came from David Steele's 91-064R3, but there were a number of other sources. In many cases, it was agreed that only the "harder" material should go in the body of the document and the "softer" notes and comments should go into an Appendix, or Annex. Section 5.1: Changes from 91-064R3 and 91-179R1. Accepted with "soft" parts going to appendix. Section 5.2: Changes from 91-179R1. Gene Milligan objected to a "soft" bus length limit for S/E operations. But he was somewhat crunched between those who wanted the 6 m limit preserved, at least in soft form, and those who insisted that the hard 6m limit is intolerable. Section 5.2.3: Changes from 91-064R3. Three meter recommendation goes in in the soft form which David finally evolved. Section 6.1: Changes from 91-178. Accepted except for implementors note. Section 6.1.1: Changes from 91-064R2, 91-185, 91-214. Active negation limits: 3.0 v @ 24 ma, 3.24 v @ 7.12 ma. Rise and fall times set at 5 ns min. But John Geldman stated that process variation which would reliably respect 5 ns min could result in not being able to do fast SCSI. Corners are voltage, temperature, and process. Glitch suppression not addressed pending receipt of Florin Oprescu's promised input. Section 6.1.2: Accept David's wording for appendix. Hysteresis changed to 0.3. IIL changed to -20 to 0 ua. IIH changed to 0 to 20 ua. Section 6.4: Goes to appendix. Section 6.5.1: Goes to appendix. 7. SPI Draft Standard. It was tentatively agreed that Kurt would do an update on the SPI draft working document, based on the above inputs, as a basis for a possible final review and refinement process, with the objective of copies being available at the Austin meeting in February (Kurt's schedule permitting). 8. Meeting Schedule The SPI Working Group meeting for February is on Tuesday, February 18, 1992 at 9:00 am at the Doubletree Hotel in Austin, TX. The SPI Working Group meeting for March is on Tuesday, March 17, 1992 at 1:00 pm at Humphreys Half Moon Inn in San Diego, CA. 9. Adjournment. The meeting was adjourned about 5:45 pm.