Wednesday, May14, 9:00 AM 1) Chair, Howard Frazier, calls meeting to order, approximately 9:00 AM. 2) Introductions, around the room. 3) Selected recording secretary, Thomas Dineen (For this meeting only!). 4) P802.3z Status Report. -Presentation by Howard Frazier. 5) P802.3ab Status Report. -Presentation by George Eisler regarding status, plans, and interfaces. 6) Discussion of changes to agenda by Howard Johnson. 7) Motion to approve the agenda. -Motion: LaVigne -Second: Kolesar -Motion passed by acclamation. -Hard copy agenda distributed by Howard Frazier. 8) P802.3z Status Reports Continue. -Presentation by Howard Frazier. 9) Standards Development Time Line. - Presentation by Howard Frazier. - Resolve all TBDs and open issues by July Meeting. - Desire to go to Working Group Ballot in July. - Must come to consensus on open issues. - Focus on closing open issues. - We will vote to close open issues, which are not closed by sub task forces, on Friday Afternoon. 10) FTP Web Site. Presentation by Howard Frazier. - Do not distribute the password. - Electronic comment submission desired. - Please submit comments in the standard electronic forms. - Preferred method is the electronic form. - Link to ASCII form is provided for large volume of comments. - Do not submit free form comments. - Please submit comments early. - Q and A followed on comment submission. - Please do not send comments directly to the editor, use electronic procedure. 11) Objectives Overview. - Presentation by Howard Frazier. - Must meet 802.3z Working Group Objectives to go to Working Group Ballot - Call for review of objectives, to make sure draft meets objectives, so it can be forwarded to Working Group Ballot. 12) Distribution of Documents. - Documents distributed by Howard Frazier. - a) Copy of minutes. - b) Draft copy of P802.3z/D2.1. - c) Comments on draft P802.3z/D2.1. - Followed by discussion of comment document format. 13) P802.3z/D2.1 Draft Overview. - Presentation by Howard Johnson. - Includes updates from March meeting. - All comments must be on D2.1 and not on earlier versions. - Include page and line numbers. - Relationship to other standards discussed. - Covered outline of standard. - Reinstated small piece of clause 5 which is referenced by clause 30. - Naming conventions explained. - Meaning of change bars varies by clause. See first page of each clause for meaning of change bars in that clause. - Added overview of change bar meaning to next draft of document. 14) Call For Patents. - By Geoff Thompson - IEEE 802.3z call for patents. - License of patentable materials on reasonable and non discriminatory terms. Requested of all organizations which believe they have patentable materials. - Form letters provided as pages 2 and 3 of latest draft. - Howard Frazier, Continued discussion on call for patents. Response letters due by July 1. - Question and answers session followed. 15) Liaison Report. - Rick Taborek - Liaison to Fibre Channel. No new status. 16) Big Ticket Items. - Presentation by Howard Frazier. - Big Ticket Items as listed in agenda. - Listing and discussion, discussion by owners. - Discussion resulted in information used to fill in the agenda document blanks. {NOTE: THE SUMMARY OF BIG TICKET ITEMS PRESENTED HERE RELFECTS THE FINAL STATUS OF EACH ITEM AS OF FRIDAY AFTERNOON} ----- Item: 1 Title: HSSDC Connector Affected Clauses: 39 Owner: J. Thatcher, Christine Foster, Ed Cady, Lisa Huff Meeting Time: Meeting Place: Status: Closed Description: The short Copper HSSDC Connector needs material for D3. Plan: Produce specific test, figures, and references to be included in D3. Resolution: Done: Have materials ready for review by Thursday AM. ----- Item: 2 Title: PMD Parameter TBDs Affected Clauses: 38, 39 Owner: J Thatcher Meeting Time: Track 3 Meeting Place: Thursday AM. Status: Closed Description: Still a lot of TBDs. Plan: Recommend values for each and every TBD or "target value" in clause 38 and 39. Resolution: Presentations and motions Thursday AM. ----- Item: 3 Title: Jitter Affected Clauses: 38, 39 Owner: J Thatcher, Del Hanson Meeting Time: Meeting Place: Track 3 Status: Closed Description: Fibre Channel working group will probably not reach closure in time for our WG ballot. Plan: Revert to a more conventional specification for jitter. Use "Hybrid Proposal". Resolution: Review of proposal from S. Joiner. ----- Item: 4 Title: Management of Link Configuration. Affected Clauses: 30,35,37 Owner: Law, Grow, Taborek Meeting Time: Wed PM Meeting Place: Track 1, 2 Status: Closed Description: Need to trace Link Configuration status and control information from the state machines, up through the GMII management registers, and into clause 30 management attributes. Plan: Joint review to make sure that all control and status are covered and described consistently. Resolution: Done: Resolutions to be placed in the next draft. ----- Item: 5 Title: Thorough review of clause 37 Affected Clauses: 37 Owner: R. Taborek Meeting Time: Thur. AM Meeting Place: Track 2 Status: Closed Description: Clause 37 of D2.1 has been seriously revised since D2. This includes major changes to the state machines as well as text. Plan: Perform a thorough review of the entire clause. Line by line through the text and state machines. Resolution: Done: No outstanding comments. ----- Item: 6 Title: GMII Registers. Dependency on item 4. Affected Clauses: 30,35,36.40 Owner: B. Grow Meeting Time: Wed. PM Meeting Place: Track 1, 2 Status: Closed Description: The specification of GMII Registers and bits in D2.1 was a necessary modification of the March meeting discussion. Plan: The D2.1 version needs to be confirmed, and carefully reviewed for consistency with other clauses that either define other registers, or that use to determine the value of specific bits in the base register set described in clause 35. Resolution: Done: D2.1 confirmed by group. Move 35.2.5 to "Change to clause 22". ----- Item: 7 Title: GMII Electrical. Affected Clauses: 35,36,40 Owner: B. Grow, B. Quackenbush Meeting Time: Thur. AM Meeting Place: Track 2 Status: Closed Description: Reach closure on concept, and agree to a set of numbers. Plan: Recommend values for each parameter. Resolution: Done: Accepted Quackenbush proposal. ----- Item: 8 Title: SOP and EOP wording in 30.2.2.2.2. Affected Clauses: 30,35 Owner: D. Law Meeting Time: Meeting Place: Status: Closed Description: The repeater does not see the above delimiters signals. These signals are actually detected at the GMII level and as such we define SOP / EOP terms only in GMII. Plan: Produce new definition i.e. "beginning of a frame or end of a frame" in clause 35 and tie that definition to text in 3.2.2.2.2. Resolution: Define: Start of packet delimiters and end of packet delimiters, for the purposes of clause 30 Layer Management, in 35.2.3.6 as assertion and deassertion of RX_DV" ----- Item: 9 Title: Behavior of aMediaAvailable. Affected Clauses: 30, 35, 36, 38, 39, 40 Owner: D. Law Meeting Time: Friday PM Meeting Place: Joint Session Status: Closed Description: Need to validate the behavior text for "aMediaAvailable" attribute. Plan: Review text. Resolution: Presentation: David law Done: No changes required. ----- Item: 10 Title: Length of burst limit. Affected Clauses: 4 Owner: H. Frazier Meeting Time: Thur. PM Meeting Place: Track 1 Status: Closed Description: Value of burst limit still open for discussion. Plan: Review simulation data from M Kalkunte. Resolve via motion. Resolution: Motion: Increase burst limit to 64 KBits (65536 bits). Passed. ----- Item: 11 Title: IPG shrinkage. Affected Clauses: 4, 41 Owner: H. Frazier Meeting Time: Thur. PM Meeting Place: Track 1 Status: Closed Description: Need to add warning about IPG shrinkage. Plan: Review proposed text. Resolution: Done: Add note in next draft min = 64BT. ----- Item: 12 Title: Repeater management. Affected Clauses: 30, 41 Owner: Haddock, Law Meeting Time: Thur. PM Meeting Place: Track 1 Status: Closed Description: Substantial changes in D2.1 concerning repeater management. Differences between 10 Mb / s and 100 Mb / s. repeater management encountered. Plan: Review text of D2.1. Decide whether to make 1000 Mb / s repeater management behave like 10 Mb / s or 100 Mb / s. Resolution: Done: Changes reflected in the next draft. ----- Item: 13 Title: Issues related to 1000BASE-T Affected Clauses: 35, 40 Owner: Eisler, Frazier Meeting Time: Meeting Place: Status: Closed Description: Provide input to 1000BASE-T to assist with selection of line coding proposal. Plan: Discussion. Resolution: Joint meetings about GMII registers and electrical. ----- 17) Room Assignments. - Presentation by Howard Frazier on Track room assignments. - Presentation by Norm Harris, Meeting host ########################################################################## Minutes of TRACK I and II JOINT SESSION Wednesday Afternoon 14-May-1997 Wednesday - GMII Management Registers: May need a table similar to 37-9 for config_reg bits. Reg. 6 is an exact overlay from AN - remove subclause 37.2.4.1.6 Next Page is mandatory from Irvine meeting. Reg 8 is an exact overlay from AN - remove associated sub clause. - 8.14 should be defined as ACK (like 5.14) - Straw Pol: should 4.14 and 7.14 be defined as ack? yes - 11 no - 22 - TOGGLE is inconsistent with this in 7.11 but we need to fix this in clause 28 if we are going to fix this here. Mr. Law raised the issue of the division of Management Register definitions between clause 22 and clause 35. This creates complexity in clause 30 as well as to the reader of the standard. Mr. Frazier stated: "I think the best thing to do is to not change the name (e.g., Table 22-6 would still be MII management register set). This eliminates the need to search 802.3u and 802.3y for references to the register set. Only the bit definitions in the registers would be modified. 35.2.5 will include text stating that the GMII incorporates the MII management register set, with a pointer to 22.2.4." After discussion, with all comments in favor, the group agreed without opposition to move the content of 35.2.5, with appropriate editing, into a new chapter in the next draft as changes to clause 22. Table 22-6 will be modified per Mr. Frazier's recommendation to include both clause 28 and clause 37 definitions of the extended registers. Registers 0-1: same names, new bit definitions per clause 35 text. Registers 2-3: no changes. Registers 5-8: same names, reference both clause 28 and clause 37. Registers 9-10: no change from 802.3y. Register 15 will be identified as basic for GMII operation and defined in clause 22. This brought up a terminology discussion - what do we call the common bits between Auto-Negotiation & Link-Configuration. Option 1 - Use Auto-Negotiation in the name but use Auto-Negotiation or Link-Configuration in the description. Option 2 - Call both Auto-Negotiation, similar to "MAC", "PHY" or "Repeater". Differentiate between them using Clause 28 or 37. Straw Pol: Option 1 - 6 Option 2 - 15 The decision included renaming clause 37 "Auto-Negotiation function, type 1000BASE-X" and deleting or changing all references to Link Configuration. This was considered editorial and therefore we were going ahead with it based on the straw poll. ---- Scott Mason: Bit 0.5 (Negotiation Selection) seems to select more than AN type. It also seems to select the PHY. Should it be renamed? Should multiple PHYs exist in one device? Rich Taborek showed a table of possible combinations for 0.12 (Auto-Negotiation Enable) and 0.5 and after much discussion, the following results were decided upon (which I believe are unchanged from the original table.) 0.12 0.5 Definition ------------------------------ 0 0 Clause 28 - Manual 0 1 Clause 37 - Manual 1 0 Clause 28 - Auto 1 1 Clause 37 - Auto Scott suggested a change to the name of either Media or Phy select. Bob Grow was willing to accept Protocol select but neither of the above suggestions. Someone suggested that because there are 2 connectors, this should not be a problem. Bob Grow countered that 2 connectors does not preclude different PHYs in the same device. David Law suggested to make the use of write mode to bit 0.5 beyond the scope of this standard. He also asked what one should do with the disabled PHY if 2 existed in the same device. Bob Grow submitted that the bit switches both the protocol and the register set and that no changes would occur, except possibly to the name or to add clarification text. ---- In table 37-9, there is no definition of mr_lc_enable = 0. It was also suggested that there should be a definition of what Clause 37 should do when Clause 28 is enabled given the discussion above. Need to add Power Down and register 8 to this table. There is an inconsistency between Power Down (state) and Power On (event). ---- Repeater mode is a good thing but will likely be a pin on the device more than a bit in a register. There are already sufficient hooks between Clauses 36 and 41 with the repeater_mode variable. ---- Managed Objects: - aPhyType - 0.13,12,6,5 5.15:12 - aPhyTypeList - 15.15:12 Should we clear aPhyType during Auto-Negotiation? Straw poll was unanimously no. - aSymbolErrorDuring Carrier - once per real variable CarrierEvent A burst is a single carrier event to the phy. Is /V/ invalid or not? In 100Base, H was considered invalid - do we want to be different from 100Base? Across the GMII, a /V/ & invalid are indistinguishable and this counter may exist in the MAC. This pollutes bit error rate testing, as does 100Base in certain cases of bit level repeaters which generate H. To be clean, we could encode the GMII differently when receiving /V/ or invalid. Straw poll 17 - /V/ increments the counter - no change to GMII 4 - /V/ does not increment the counter - change how /V/ is signalled across the GMII. - aPhyAdminState - add Isolate to "Behaviour defined as" - aMauType - expand 1000Base-X to 1000Base-X, -LX, -SX & -CX - aMauTypeList - where do we get the LX, SX and CX information? Magic? Strike the second sentence and no one will know exactly where this info comes from so it will have to be generated design specifically. - aMediaAvailable - Left as an action item to be resolved Friday p.m. - aAutoNegAdminState - strike the last sentence - covered in aMauType - aAutoNegRemoteSignalling - use /C/ ordered_sets instead of LC code sequence. - aAutoNegAutoConfig - no parallel detection - no change - aAutoNegLocalTechnologyAbility - 1000Base-X or -T only (not LX, SX & CX) - Add reference to Clause 37 - Add ASYM_OUT, ASYM_IN and SYM for pause capability - aAutoNegAdvertisedTechnologyAbility - Add reference to Clause 37 - This was only a part of the base code word for Clause 28. - Because in Cluase 37 there is no split between technology and selector fields, change the text of this definition to say all the bits in the base page are written using this management object for Clause 37. - This means selector ability is not used for Clause 37. - aAutoNegReceivedTechnologyAbility - same comments as above. - aAutoNegLocalSelectorAbility - Not used in Clause 37. - aAutoNegAdvertisedSelectorAbility - Not used in Clause 37. - aAutoNegReceivedSelectorAbility - Not used in Clause 37. There are no attributes for Next-Page. These are now mandatory but since none are defined, none exist. When some pages become defined, also define the management attribute. ######################################################################### Thursday - D2.1 PCS Comment Resolution: Only those comments that had discussion will only be addressed. To completely follow these notes, the comments list should be read in conjunction with these notes. #4 There are 2 trade-offs, propagate the error as early as possible or enforce a /T/ on every packet and delay the error until after the /T/ or should it be until after the /T/R/ ? Allow the /T/ to be replaced by a /V/ in state END_OF_PACKET_EXT in the PCS Transmit ordered_set state diagram. Add a path from state RECEIVE to state EXTEND_ERR in the PCS Receive state diagram for the case of receiving a /V/. #12, #35 Global entries to the two states don't work because the conditions of the xmit variable are states not events. These global entries need to be changed to the xmit variable changing to these states rather than being these states. This will now be handled by a distributed top level state machine controlling how these different parts of the PCS Receive state diagram get entered. #23 Much confusion on the timing of the generation of rx_even. The problem is that given a PMA_UNITDATA.indicate with rx_code_group, rx_even is both assigned and evaluated. How should it be evaluated? As the value before it is assigned or after? After much discussion, the following was agreed to: The sync machine, upon detection of a PMA_UNITDATA.indicate, will generate a SYNC_UNITDATA.indicate, rx_code_group_latched and rx_even. This incurs no time delay but makes the interface easier to understand. PMA_UNITDATA.indicate |___|___|___|___|___|___|___|___|___|___| rx_code_group K D D D K D D D K D SYNC_UNITDATA.indicate |___|___|___|___|___|___|___|___|___|___| rx_code_group_latched K D D D K D D D K D rx_even E O E O E O E O E O The final names may be different than those listed above. #24 It is desireable to add hysteresis on rx_lc_invalid. Set rx_lc_invalid in state RXLC_INVALID. Transition to state RXLC_K if the next receive code_group is a K28.5, otherwise transition to state IN_CONFIG. Add the assignment of rx_lc_invalid to FALSE in state RXLC_K so that it is asserted for exactly one PMA_UNITDATA.indicate cycle. It was also noted that there is no transition from state IN_CONFIG directly to state RX_LC_INVALID. This is the case because invalids are only looked for when actually looping on /C/ or /I/. #25 Withdrawn #26 Include running disparity discussion - unanimous except for 1. #27 - #30 Make them real sentences. #31 The references to IPG and Receiver in the same paragraph is confusing. Move this to 2nd paragraph of subclause 36.2.4.1.4, drop the title and clean up the text to make it clearly informative. ######################################################################### LC Comment Resolution: #5, #12, #29, #68, #69 Scratch the desire_np variable. Because every PCS must be capable of NP exchange, you don't have to advertise it in the base page. This bit should only be set if you actually have next pages to exchange, not just to advertise the fact that you support it. This also allows you to remove the base_page variable. Clean up subclause 37.2.1.6.2 to remove the requirement for NP=1 in base page. #13 Scratch the reference to "upon power up". The default value should be something which would result in a normal operating mode. #20 There is a table of priority resolution for full & half duplex in Annex 28.B. With only two entries, no table is necessary for Clause 37. #25 (follow-on discussion), #31, #42, #47, #59, #67 Add various paths back to state LC_ENABLE when rx_config_reg=0 * ability_match=TRUE. #28, #50 - Superceded later by "Review of State Diagrams" discussion. Make rcv/C/ an event (.indicate) to fix the global entry condition into state LC_ENABLE. Also, to fix the global entry condition sync_status = FAIL, add the assignment of the variable xmit with the value IDLE. ######################################################################### Thursday Late Night session: #35 The direction is to define these registers once correctly then reference the definition later. Define tx_ and rx_config_regs in Clause 36 and mr_x_regs and how they get updated in Clause 37. #39, #52 This comment, along with several others, falls into the "Management Knows All" category. If there is management intervention in the form of forcing manual mode, then the management must make sure the values in the base pages result in a valid operational mode. This removes any requirement for default values on this register. #44 It was recommended that this be resolved by test description rather than any attempt to change the state diagram. #46, #65 Remove detection of break-link in state ABLITY_DETECT to eliminate oscillations if the local device transitions to this state before the link partner does. #51 The LC Expansion register is not the location for this information but it is useful information. It was agreed to that SYNC_STATUS will be the source of bit 1.2 (Link Status). #64 Add LSB and MSB labels to figures 37-2, 3 & 4 ######################################################################### Friday a.m. : #7, #15, #18, #32, #35 If the protocol is followed, then there should be no modifications necessary. 802.3y chose to move to 2 registers so the received base page wasn't overwritten with received next pages. When either a base page or a next page is received, the Page Received bit, 6.1, is set. As long as the the received page is read before the new next page register, 7, is written, there is no way that another next page will be received. 6.1 will continue to be cleared upon a read of register 6 when it is set. #45 The Toggle bit is defined in the register definition. The state diagram shows it as the inverse of the Toggle bit from the previously transmitted page, rather than coming from the register directly. It has been convention that it is not specified whether the state diagrams are inplemented in hardware or software and the generation of the Toggle bit is a perfect example of a case where it can be done either way and still be in compliance with the standard. #26, #27, #38 These were resolved once on Thursday, then changed to what was eventually agreed upon Friday morning. It was first decided that these variables (ability_match, acknowledge_match and idle_match) are cleared upon entry to every state then set by detecting the appropriate condition. The problem with this definition is that once set, the only way to clear is by changing state. It was decided that this would not provide the correct operation. The way these variable should be defined is that they should continue to clear upon entry to every state, continue to set upon detection of the appropriate condition but also clear when that condition has gone away. It was noted that this was how the variables were intended for use un Clause 28 Auto-Negotiation, also. Some clarification text should be added for these variables. #72 Remove all references to Link Monitor, in figure 36-2, replace Link Monitor with Sync & Auto-Negotiation. ######################################################################### Review of State Diagrams: Large discussion about what should happen when Auto-Negotiation is disabled in regards to the Arbitration state diagram. After the smoke cleared it was agreed that this condition forced entry to the state LC_ENABLE and you would remain there with the xmit variable assigned with the value DATA. This also inhibits LC_COMPLETE from setting which is a requirement for Auto-Negotiation disabled. If Auto-Negotiation is enabled, while in state LC_ENABLE, the xmit variable will be assigned the value CONFIGURATION and tx_config_reg will be assigned the value 0. This supercedes the previous discussion on comments #28 & #50. ######################################################################### PMA Comment Resolution: #4 Rejected - As long as the data meets setup and hold times, the max rise & fall time (of the data) doesn't matter. We don't want to be overly constrictive. ######################################################################### #30, #33, #34, #36, #48, #49, #53, #54, #60, #66 Timer/Counter Issue - Last Major Open Issue #30, #48, #49 Remove cr_change variable - use consistency match. #34, #36, #53, #54 Must transmit constant config_reg value for at least long enough to reach roundtrip on longest network. Whether this is accomplished with a counter or a timer is a style issue. They both satisfy the same result. In summary, it was agreed to stay with a timer. #33 Because of timer conventions from subclause 14.2.3.2, link_timer_done does not have to be explicitly defined. Reference this subclause just for completeness. #60 Roundtrip time is very small (5us/km) compared to the 100ms timer. Reasons for keeping this timer large (100 ms) should be based solely upon roundtrip delay, not to make this a large value "because it doesn't have to run fast". The arbitration state diagram is already designed to only run as fast as the slowest device and a timer should net be arbitrarily assigned a value to force this process to be slow if you have 2 fast devices trying to establish link. Technical Motion by Pat Thaler, 2nd by Ben Brown, to drop the link_timer_done value from 100 ms to 10 ms. The motion carried by acclamation. #66 After much discussion: Technical Motion by Steve Haddock, 2nd by Ben Brown, to not mandate any type of watchdog timers in the Arbitration state diagram. Techincal - requires >= 75% Yes:17 No:3 Abs:3 Motion Passed. ######################################################################### GMII MINUTES The GMII subtask force met Wednesday afternoon in joint session with the Management and Link Configuration subtask forces to discuss management issues. The group unanimously confirmed the approach to GMII registers as documented in 802.3z/D2.1. No significant changes are required in the clause 35 GMII base register set as a result of the management issues. The group also agreed without opposition to move the content of 35.2.5, with appropriate editing, into a new chapter in the next draft as changes to clause 22. Thursday morning, the GMII subtask force met to discuss the electrical portion of D2.1. The meeting included presentations and resolution of comments on the electrical section. Dr. Howard Johnson gave a brief presentation on GMII Overview and Relationship to PMA Interface. This presentation reinforced that the GMII and PMA interfaces are not the same. The GMII is defined to maintain compatibility such that pins can be used to multiplex either GMII or PMA operation for the data path signals. It is not necessary that the clocking or electrical specifications be the same to accomplish this, only compatible. Mr. Bill Quackenbush presented on GMII Electrical Specification Options. This presentation described alternative specification techniques and an analysis of their impact on implementation objectives. After presentation and discussion, Mr. Frazier moved and Mr. Dove seconded a Motion: Use pages 34 through 36 of "GMII Electrical Specification Options", Quackenbush, 5/1997, as basis for GMII electrical specification in the next draft of P802.3z. The motion passed Y=30, N=0, A=1. Comments on the electrical portions of D2.1 and their recommended responses, were discussed to and agreed to by the subtask force. On Friday morning the GMII subtask force met to discuss other comments and the proposed resolutions. The CRS changes in D2.1, 35.14 L48-51 were confirmed with typographical correction. A new subsection, 35.2.3.6, was approved. This subsection defines Start of Packet and End of Packet for clause 30 repeater management. The changes to the electrical specifications approved on Thursday were presented in detail and approved without objection. This includes modifications to Tables 35-11 through 35-14 and replacement of Figures 35-18 and 35-19 with two new figures, one illustrating setup and hold time, and the second illustrating clock period, time high and time low parameters. These figures are in addition to the two figures included from Mr. Quackenbush's presentation approved in the above motion. The subtask force also discussed, modified and confirmed comment resolutions to all comments included in the D2.1 comment database. The GMII subtask force then adjourned. ######################################################################### Minutes of Track I Thursday Afternoon 15-May-1997 MAC H. Frazier Reviewed comments from D. Wong on D2.1 Comment Concerning reference to receiveDataValid. Response proposed by editor accepted. Comment concerning latecollision threshold. Response proposed by editor accepted. Wording to be refined to properly describe latecollision threshold. latecollision threshold with be clearly identified in figure 4-7. Comment concerning "bursting on?" jump condition in control flow diagrams. Response proposed by editor accepted. Comment from M. Molle. Modify transmit frame control flow summary to omit increment of latecollision count, and insert test for late collision and > 100 Mb/s between increment attempts and test of too many attempts. Accepted, will be incorporated in next draft. Comments from S. Muller and M. Molle. The independent process control flow diagrams for BitTransmitter, BitReceiver, and SetExtending need an arc from the final action block back to the initial test condition. Accepted, will be incorporated in next draft. Reviewed note concerning IPG shrinkage. No objection to adding note. Agreed to change minimum value in note to 64 bit times, based on the calculation: Minimum IPG = Nominal IPG - repeater clock tolerance - preamble growth - guard band = 96 - 16 -8 -8 = 64 bit times Note as written applies to both full and half duplex operation. Dicussion of effect of IPG shrinkage on fairness based on nominal interframespacingpart1. Interpretation is that interframe spacing part 1 can be 0 to 63 bits. Mohan's presentation of analysis of burstLimit. Mohan Kalkunte presented the results of simulations he performed which analyzed the effect of increasing burstLimit to 64 kbits from the 12 kbits identified in D2.1. Motion: Increase burstLimit to 64 kbits. M: Kalkunte S: Molle Y:13 N:0 A:5 Motion passes Big Ticket item # 8. Add subclause 35.2.3.6 which will state that for the purposes of Clause 30 repeater management, the Start of Packet and End of Packet delimiters shall be defined as the assertion and deassertion of the GMII signal RX_DV, respectively. Clause 30 management Accepting comment from David Law concerning aDataRateMismatches, which also changes aRunts, aLateEvents, aCollisions. aCollisions Behavior Defined as: This counter increments for any carrier event signal on any port in which the collision event signal on this port is asserted. aLateEvents Behavior Defined as: {change second sentence to read} For a clause 27 and clause 41 repeater, this counter increments for each assertion of the collision event signal which occurs while activity duration is greater than the LateEventThreshold. Such a carrier event... {{note, leave alone the remainder of the text for this attribute}} aRunts Behaviour Defined as: {strike from (10 Mb/s operation) on line 18, thru (100 Mb/s operation) on line 20.} Same strike on line 21. aDataRateMismatches Behaviour Defined as: {strike from (10 Mb/s operation) on line 39, thru (100 Mb/s operation) on line 41.} Same strike on line 46-48. Accepted editorial comments from David Law Comment from David Law concerning 30.3.1.1.13, resolution is to remove reference to 7.2.4.6. Comment from David Law concerning 30.3.1.1.26, Comment withdrawn after discussion. Comment from David Law concerning 30.3.2.1.7, accept, editorial. Comment from David Law concerning 30.4.1.1.2, accept suggested remedy Comment from David Law concerning 30.4.3.1.2, accept suggested remedy Comment from David Law concerning 30.4.3.1.14. Comment regarding lines 39 to 51 accepted. Comment regarding lines 41 to 49 accepted, comment regarding line 53 withdrawn after discussion. Comment from David Law concerning 30.4.3.1.20 accept suggested remedy. COmment from David Law concerning 30.4.3.2.1 page 30.43 line 48. Accept suggested rememdy. Comment from David Law concerning 30.5.1.1.2 page 30.44 line 54. Accept in principle, but enumerations for 1000BASE-X, XHD, XFD, will be retained. Verbage in Behaviour defined as must be added to state that the 1000BASE-X, XHD, XFD should only be reported if the underlying PMD type is unknown. Comment from David Law 30.5.1.1.2 page 30.45 line 3 to 5. Accept suggested remedy. Comment from David Law 30.6.1.1.8. Accept suggested remedy. Clause 41 comments -Stephen Haddock. Comment from David Law concerning 41.5, page 41.18 line 15. Accept suggested remedy. Comment from David Law concerning 41.6.3 PICs. Accept suggested remedy Comment from David Law concerning 41.6.3. Accept suggested remedy Comment from David Law concerning 41.6.4.12 Accept suggested remedy. Comment from Don Wong concerning 41.2.2.1.6 repeater state diagram. Reject. Definitions for ALLXJIP, etc are described in 41.2.2.1.6. Comment from Don Wong concerning 41.2.2.1.4 page 41.10 line 6. Accept suggested remedy. Comment from Don Wong concerning 41.2.2.1.4 page 41.10 line 2. Accept suggested remedy. Comment from Don Wong concerning 41.2.2.1.4 page 41.10 line 24. Accept in principle. All timers will described in terms of the state machine that uses them and what conditions they time, without reference to specific states. Comment from Don Wong concerning 41.2.2.1.6 page 41.12 figure 41.2 repeater unit state diagram. Rejected. State diagram follows conventions in 1.2.1. Comment from Don Wong concerning 41.2.2.1.6 page 41.15 figure 41.5, transition from SSD PENDING WAIT to LINK WAIT. Accept in principle, this comment is resolved by resolution of Don Wong comment concerning 41.2.2.1.4 page 41.10 line 24. Change to state machine not required. Comment from Don Wong concerning 41.2.2.1.6 page 41.15 figure 41.5, second comment concerning transition to SSD PENDING WAIT to LINK WAIT. Rejected because change to state machine was not required to resolve previous comment on this state machine. ######################################################################### Minutes of the IEEE 802.3z Track III PMD Meeting 5/14/97 Ft. Lauderdale, Florida Jonathan Thatcher, Chair Karl Nakamura, Secretary Meeting Called to Order 5/14/97, 1PM Attendees: Karl Nakamura, LSI Logic, karln@lsil.com Sameer Muyyuru, TI, svuyymu@ti.com Michael Yam, Vitesse, yam@vitesse.com Christine Foster, Gore, cfoster@wlgore.com Vince Melendy, Methode, melendy@compuserve.com Doug Thomsom, OTI, dougt@ohmtech.com Shelto van Doorn, Siemens, schelto.vanDoorn@sci.sic Haluk Aytac, HP, haluk_aytac@hp.com Dave Smith, Honeywell, dsmith@micro.honeywell.com Arlen Martin, Lucent, stearm@micro.lucent.com Tremeont Miao, ADI, tremont.miao@analog.com Tom Debiec, Berg-Tek, tom_debiec@uscabla.aboctel.com Ed Cornejo, AMP, ecornejo@amp.com Colin Whitby-Strevens, SGS-Thompson, colinws@bristol.st.com Del Hanson, HP, del_hanson@hp.com David Cunningham, HP, dgc@hplb.hpl.hp.com Lisa Buckman, HP, lbuckman@hpl.hp.com Bob Musk, HP, bob_musk@hp.com Bryan Gregory, Molex, bgregory@molex.com Jay Neer, Molex, jneer@molex.com Ed Grivna, Cypress, elg@cypress.com John Fitzgerald, Bay Networks, jfitz@baynetworks.com Stan Swirhun, Vixel, sswrirhun@denver.vixel.com Bill Verheggen, Intel, william_c_verheggen@ccm.jf.intel.com Ed Cady, Berg, edcady@aol.com Ian Verigin, Packet Engines, iainv@packetengines.com Rick Lacerte, Cabletron Systems, lacerte@ctron.com Bill Kuypers, Tensolite, bkuypers@tensolite.com Norm Harris, Adaptec, nharris@eng.dapatec.com Todd Hudson, Siecor, todd_hudson@siecor.com John Bowerman, Honeywell, jbowerma@micro.honeywell.com Wen Tang, 3comm, Wen-Tsung_Tang@3mail.3com.com Steve Swanson, Corning, swansonse@corning.com Doug Harshbarger, Corning, harshbard@corning.com Lisa Huff, AMP, lisa.huff@amp.com Dan Brown, AMP, dan.brown@amp.com Graham E. Measor, GEC Douglas Sudjian, Microlinear, sudjiand@engmail.ulinear.com Jonathan Thatcher, IBM jonathan_thatcher@vnet.ibm.com Archana Roy, Synergy, archana.roy@synergysemi.com 1. Distribution of "PMD Goals for 5/97 Interrum Mtg." by Jonathan Thatcher. Copper PMD Objectives Add HSSDC Connector Close out discussion on electrical Tx/Rx values(mv-ppd) Set direction on grounding/shielding Clean up move from specifying cable to specifying jumper Close on comments about cable cross talk(NEXT spec) Close on 1394.2 common cable "opportunity" Figure out what to do with PICS Optical PMD Objectives Clear TBDs Close Jitter Specifications and Measurement Methodology Close Various Optical Parameters Dual Windows for 780 and 850 lasers RIM Modal Noise Penalty Connector Loss Rise/Fall Time Specifications(20-80%?) Modifications to Eye Measurements Modal Bandwidth of MMF Distances: SW and LW Other 9 or 10 micron SMF Clean up dispersion slope Measurement Techniques: RIN; Spectral Width Center Wavelength, Modal Noise Close on definition of Signal_Detect 2. Motion: That the PMA/PMD subgroup accept the direction of the parent committee to change Clause 39 to reflect Jumpers instead of raw cable, equalizers, and connectors. jumper= cable assembly(connector, raw cable . . .) Made by Ed Cady, Berg Second: Karl Nakamura, LSI Passed by Acclamation 3. Presentation: "Jitter Specification for 802.3z," by Steve Joiner of HP, presented by Del Hanson of HP Jitter is specified at 4 points 1.output of the serializer 2.output of the transmitter 3.input to the receiver 4.input to serializer Review of Jitter budget per Fibre Channel FC-PH rev 4.3 multimode link were presented. Fibre Channel Jitter Budget with Jitter Frequency Included Recommengations from the FC Methodology for Jitter Specifications Technical Report (Draft) are presented Consider 2 frequency bands for Jitter measurements -with jitter above cutoff frequency of clock recovery circuit -with jitter above frequency (fd) Specify maximum total jitter at interface Specify maximum deterministic jitter at interface Allow Random Jitter = max total jitter - Actual deterministic jitter Jitter Output Measurement of a Serialized Data Stream Diagram presented Required Jitter Tolerance Mask for Clock Recovery Circuit presented Proposed Fibre Channel Specification for Jitter Above fc>637 kHz Presented Proposed Conversion from Fibre Channel to 802.3z Assume same technology for physical interface Use same jitter (in ps) for FC and 802.3z Use same critical frequencies for jitter bandwidth -42.5 kHz (drift) -637 kHz (PLL loop bandwidth) Proposed Jitter Budget for 1000Base-SX Ser Opt Opt Des Out Tx Rx In TJ(UI) .24 .44 .47 .71 DJ(UI) .12 .22 .22 .45 RJ(UI) .12 .21 .25 .26 TJ(ps) 188 348 376 565 DJ(ps) 94 179 179 358 RJ(ps) 94 169 196 207 A Description "Jitter Measurements using Oscilliscopes, Revision 4/23/97" was included to describe the use of a typical digital oscilloscope to measure the random and deterministic jitter contributions for a transmitter and receiver." Discussion, Led by J. Thatcher There is concern that there's a lot more work to do on the specification, especially on the test methodology. FC Method Well Defined "Poor Rating" DONE FC Jitter Group "Poorly" Defined "Good Rating" NOT DONE Question: Can a Fibre Channel Jitter Group type solution be done in 3 weeks? No . . . So, we should adopt a traditional methodology for jitter specification. Motion: To use items 1-5 below to creat a jitter spec and test methodology for test points 1 and 2. 1. Measure total jitter at points 1 and 2. Tj normative 2. Dj and Rj informative at points 1 and 2. Specs and Test technique are informative 3. Use numbers from the Joiner Proposal 4. Test points 1 and 2 from the Joiner Proposal 5. Scope methods from Fibre Channel, adding the "golden PLL" test method from the Joiner Proposal. Motion by Colin W-S of SGS Second by Del Hanson, HP Y: 25 N: 1 Ab: 0 Motion Passes 4. Motion: Expand the previous motion to include points 3 and 4 from the Joiner Proposal Motion by Stan S, Vixel Seconded by Colin W-S, SGS Y: 23 N: 0 Ab: 2 Motion Passes 5. Motion: Use Jitter Tolerance mask for Clock Recovery Circuit from Joiner Proposal (with text) while exciting with both a +/- K28.5 data pattern and the Fibre Channel Jitter Working Group RPAT. Motion by Del Hanson, HP Second by Ed Grivna, Cypress Y: 29 N: 0 Ab 1 Motion Passes 6. Motion: To direct the chair of the PMA/PMD working group to request the MAC/PCS to provide a diagnostic function which includes the generation and detection of +/- K28.5 +/- K28.7 +/- D21.5 Pre Encode PRBS((2**7)-1) in order to support system level testing of PMD/PMA jitter specifications. Motion by Colin W-S, SGS Second by Karl Nakamura, LSI Y: 24 N: 0 Ab: 0 Motion Passes Meeting Adjourned 5/14/97, 5PM ######################################################################### IEEE 802.3z PMA/PMD Minutes for 5/15/97 Prepared by: Lisa Buckman 5/15/97 AM session Attendance: Name Company Lisa Buckman Hewlett-Packard Bob Musk Hewlett-Packard Dave Cunningham Hewlett-Packard Del Hanson Hewlett-Packard Grahame Measor GEC Plessey Takehiko Tokoro Hitachi Cable Ed Grivna Cypress Semiconductor Ed Cornejo AMP Dan Brown AMP Bryan Gregory Molex Arlen Martin Lucent Vince Melendy Methode Christine Foster Gore Karl Nakahura LSI Schelto VanDoorn Siemens Doug Harshbarger Corning Steve Swanson Corning Ken Taylor BOF Todd Hudson Siecor Paul Kolesar Lucent John Bowerman Honeywell David Smith Honeywell Norm Harris Adaptec Stan Swirhun Vixel Dick Schwarz Pulse Components Div. Keith Conroy Pulse Components Div. Petar Pepeljugoski IBM Jonathan Thatcher IBM Wen-Tsung Tang 3COM 9:10am: Lisa Buckman "volunteers" to be secretary 9:15am: Review of agenda Speaker: Del Hanson Topic: Joint Issues for both SX and TX (handout) Speaker: Steve Swanson Topic: Optical Cable Team - Assigned Work Items -Fiber and connector recommendations -Handout to be given out after lunch 9:30 am: Speaker: Del Hanson Topic: Review of Bandwidth Enhancement Issues (TIA group measurements) -Info. has come in since last Thurs. -Review of statements to give to whole committee -Position statements read by Del Procedural Motion: Howie Johnson Second: David Cunningham Motion: Move to appoint the following people to draft a statement on controlled launch BW for review tomorrow and [later] presentation at 802.3z: Jonathan Thatcher and Del Hanson. This motion passed with 26 for, 0 against and 3 abstain. 10 am: Distribute Clause 39 -will review after lunch -main points: HSSDC, "jumpers" 10:15 am: Speaker: David Cunningham Topic: RIN Calculations for link model (handout) Motion: David Cunningham Second: Petar Pepeljugoski Motion: Adopt RIN Calculation (Method A) that David Cunningham presented. This motion passed by acclamation. 10:25 am: Speaker: David Cunningham Topic: Modal Noise Calculations Suggested Penalties 62 MMF 50 MMF LWL .5 dB 1 dB SWL .15dB .3dB Motion: David Cunningham Second: Petar Pepeljugoski Motion: To accept suggested penalties above in table for mode-selective loss penalties. This motion passed by acclamation. Speaker: David Smith Topic: Data on connector insertion loss (handout) Motion: Del Hanson Second: Stan Swirhun Motion: To adopt 1.5 dB connector loss for total in-line connector loss. This motion passed with 22 for, 0 against, and 4 abstain. Speaker: Petar Pepeljugoski Topic: 780 nm window (presented table of parameters) Motion: Petar Pepeljugoski Second: David Cunningham Motion: To create 780-830 nm window of operation to allow tradeoff of wavelength and rise time only. (38.2 and 38.3 would need to be changed) Motion: Del Hanson Second: Petar Pepeljugoski Motion: Table motion until discussion of lambda^(-4) from fiber manufacturers (attenuation vs. lambda). This motion passed with 24 for, 0 against, and 5 abstain. Discussion of lambda^(-4): Corning says that lambda^(-4) is appropriate. Motion: Stan Swirhun Second: Del Hanson Motion: PMD accept lamda to the -4 as the power dependence in the attenuation line of the link model. Motion: Paul Kolesar Second: Ed Grivna Motion: Table above motion until 8:30 am tomorrow. This motion failed with 8 for, 9 against, and 14 abstains. The previous motion failed with 15 for, 9 against, and 5 abstain. Adjourn at 12:15pm for lunch 5/15/97 PM Session Attendance: Name Company Lisa Buckman HP David Cunningham HP Karl Nakahura LSI Del Hanson HP Mike Gardner Molex Jay Neer Molex Colin Whitby-Strevors SGS-Thompson Grahame Measor GEC Plessey Lisa Huff AMP Ed Cady BERG Christine Foster Gore Vince Melendy Methode Arlen Martin Lucent Petar Pepeljugoski IBM Carrie Munson Pulse Richard Schwarz Pulse Keith Conroy Pulse Bill Kuypers Tensolite Al Kelley Tensolite Sameer Vayyura TI Iain Vevigih Packet Engines Rick Lacerte Cabletron Takehiko Tokoro Hitachi Cable Jonathan Thatcher IBM 1:15pm: Review of Clause 39 changes (handout) redistributed CLS. 39 D2.11 Speaker: Karl Nakamura Motion: Karl Nakamura Second: Ed Grivna Motion: Change the 1000Base-Cx (Clause 39, Draft 2.1): TW style jumper cable specification from a 0.288 dB/m with 0.25 dB/connector specfication to a total link loss maximum of 8.8 dB, measured between point S and point R [detailed locations of changes were provided to assist the editor]. This motion passed with 13 for, 0 against, and 10 abstain. Editorial changes: include information on delay, move information to overview on length of cable (0-25m) and remove from Table 39-4 "operating range", add delay information to Table 39-5. 1:55pm: Technical Motion: Karl Nakamura Second: Colin Whitby-Strevens Motion: Change the 1000BASE-CX (clause 39, draft D2.1): transmitter minimum differential launch amplitude from 1100 mV to 700 mV, and the receiver minimum differential sensitivity from 400 mV to 300 mV [detailed locations of changes were provided to assist the editor]. This motion failed with 11 for, 6 against, and 10 abstain. 2:30pm: Speaker: Ed Cady Topic: MetaGig (handout) -will get common mode differential pair clarification -shows connectors, "header assembly", etc... (in handout) 2:45pm: Motion: Ed Grivna Second: Lisa Huff Motion: We [1000BASE-CX] will only entertain proposals for DC coupled shields in 1000BASE-CX implementations. This motion passed with 18 for, 0 against, and 9 abstains. Discussion of 1394.2 compatibility and pins in connector Jonathan will get on reflector info. regarding 1394.2 compatibility Karl and Ed will talk to 1394.2 group too. Editorial change: fix TX eye (done already). 3:15pm: Technical Motion: Christine Foster Second: Ed Grivna Motion: To add NEXT specification of 6% max at 85 ps Tr to table 39-5 jumper cable assembly specification. This motion passed with 15 for, 0 against, and 7 abstain. "Walk thru" of Clause 39: -Jeff's changes to overview - p. 39.2, 22.5-25 moved to overview delete editor's notes, p. 39.1 -Signal detect wording will be moved into CLS. 39 as per Jonathan. -Table 39-1, RJ, DJ will be worked out (see table) -Table 39-1, add RJ, DJ, TJ -Reviewed changes to p. 39.3. -Editorial change to Fig. 39.2..."load" instead of "circuit" -Request for editor to clear up intent of Fig. 39-4 vs. Fig. 39-3. -Editorial change of all test points to "1,2,3, and 4's" of Joiner proposal. -Table 39-4, delete 0-25m and put in overview. -add delay of jumper to table 39-5 Motion: Howard Second: Grahame Measor Motion: Add delay of jumper (253 bit times) to table 39-5. This motion passed with 14 for, 0 against, and 2 abstain. Editorial change: change title to "Jumper cable assembly", delete "1000Base-CX PMD" -Delete first note after table 39.t5 -p. 39.7, line 20, put into 39.1 -global change: all "cables" to "jumper cable assemblies" -p. 39.8, note stays until all TBDs filled in. Technical Motion: Ed Grivna Second: Ed Cady Motion: Accept contact style 1 assignment map for style 1 to style 2 connector as seen in d/2.1.1. This motion passed with 17 for, 0 against, and 0 abstain. -Discussion of +5VDC (optional) pin -Recommendation for editor, change "power" to "reserved" with footnote to say how may be used. Motion: Geoff Thompson Second: Ed Cady Motion: "Optional" is removed from Table 39-6 rows 2,3,7,8. Respective column 3 description to be changed to "reserved." Add a note (informative) about applications. The motion passed with 18 for, 0 against, and 5 abstain. -Recommendation: to tell plenary about above motion. -Informative beware: no spec. on ground wire. -Need Table 39-7 filled in. -Clarification of "PICS" given -Will have to be at July meeting on paper to vote on. Procedural Motion: Howard Second: Ed Cady Motion: Team of Cristine Foster, Ed Grivna, Lisa Huff, Jonathan Thatcher, [and] Ed Cady to provide 3z PMD S.W.G with complete PICs recommendation at July meeting. Put on PMD reflector one week before the meeting. This motion passed by acclamation. 4:40pm: Speaker: Colin Whitby-Strevors Topic: Jitter specifications for copper Technical Motion: Colin Whitby-Strevors Second: Ed Grivna Motion: To create a jitter specification and test methodology for test points 1, 2, 3, and 4 for 1000BASE-CX based on: 1) Measurement of total jitter at test points 1, 2, 3, and 4 2) Include DJ and RJ as informative 3) Numbers as presented 4) Test points from joiner proposal 5) Scope methods from FC adding golden PLL from Joiner proposal 6) with additional text: "A 1000BASECX compliant jumper cable, when driven with a worst case signal meeting the requirements of 39.3.1, shall deliver a signal that metts the requirements of 39.3.2. This motion passed with 15 for, 0 against, and 0 abstain. 5:15pm: review of objectives 5/15/97 Evening session Attendance: Name Company Lisa Buckman HP Gadi Lahat Madge David Cunningham HP Petar Pepeljugoski IBM Bob Musk HP Del Hanson HP Bryan Gregory Molex Paul Kolesar Lucent Stan Swirhan Vixel Ed Grivna Cypress Colin Whitby-Strevors SGS-Thomson Vince Melendy Methode Doug Harshbarger Corning Steve Swanson Corning Ken Taylor BOF Dave Smith Honeywell Todd Hudson Siecor Ed Cornejo AMP John Bowerman Honeywell Dan Brown AMP Schelto van Doorn Siemens Jonathan Thatcher IBM 5:30pm: -Review of 1000Base-CX work items -Need to find out if any measurement techniques required- will need for next meeting. Speaker: Del Hanson Topic: Review of decisions made on parameters -used RIN of -123 dB/Hz in model -changed .5 dB/km at 1300 nm -Discussion of parameters - ISO 11801 sets min required of links, ISO 11801 wants a 100 MHz link supported. Speaker: Petar Pepeljugoski Topic: Revisit motion to create 780-nm window Motion: Petar Pepeljugoski (from before) Motion: Petar Pepeljugoski Second: Del Hanson Motion: change 780 nm to 770 nm, and .34 ns to .32 ns in Del's table. Quesion called by Schelto. This motion passed with 11 for, 0 against, and 8 abstain. Motion to create new window (Petar's): This motion passed with 14 for, 3 against, and 5 abstain. (including changes in previous motion - 770 nm and .32 ns) Proposal - explicitly include jitter and rise/fall times 6:50pm Technical Motion: David Cunningham Second: Gadi Lahat Motion: To use the transmit eye mask for overshoot and undershoot only. Create a T-rise/T-fall specification @ 20-80% of max value 0.26 ns @ lamda > 830 and 0.21 ns with lamda less-than-or-equal-to 830 nm measured using a scope with BW of 0.75 BTF and correct for filter by multiplying by TBD. Assign Del and JT to fix eye based on new jitter specs and TBD for correction factor. -discussion of jitter and FC specs. -do we need to change eye mask to adopt above? -assign Del and JT to fixe eye based on new jitter specs This motion passed with 15 for, 1 against, and 1 abstain. 7:45pm: Motion: Del Hanson Second: Bryan Gregory Motion: Use SW Jitter specs for LW specs (from 5/14/97) This motion passed with 18 for, 0 against, and 1 abstain. Speaker: Steve Swanson Topic: Fiber parameters -shows table of parameters which could replace Tables 38.9 and 38.10 -discussion of parameters -agreed on attenuation, dispersion slope, and zero dispersion wavelength. ######################################################################### Minutes from IEEE 802.3z Track III meeting from 0845 to 1200, 5/16/97 Doug Harshbarger, Corning Incorporated Attendance: Name Company Email Doug Harshbarger Corning HarshbarDE@corning.com Ian Verigin Packet Engines iainv@packetengines.com Rick Lacertle Cabletron lacertel@ctron.com Tommy Leieng Asante TLeieng@Asante.com Grahame Measor Gec Plessey grahame_measor@gpsemi.com Takehiko Tokoro Hitachi Cable tokoro@rop2.hitachi_cable.co.jp David Smith Honeywell dsmith@micro.honeywell.com Petar Pepljugoski IBM Petarp@watson.ibm.com Stan Swirhan Vixel sswirhan@denver.vixel.com Mike Gardner Molex mgardner@molex.com Bryan Gregory Molex bgregory@molex.com Paul Kolesar Lucent pkolesar@lucent.com Schelto van Doorn Siemens schelto.vandoorn@sci.siemens.com Todd Hudson Siecor Todd.hudson@siecor.com Ed Cornejo Amp ecornejo@amp.com Dan Brown Amp dan.brown@amp.com Lisa Buckman HP lbuckman@hpl.hp.com Bob Musk HP bob_musk@hp.com David Cunningham HP ? Hon Wah Chin Cisco hwc@cisco.com Keith Conroy Pulse Components Division Vince Melendy Methode melendy@compuserve.com Ed Grivna Cypress edg@cypress.com John Bowerman Honeywell jbowerman@micro.honeywell.com Ken Taylor BOF foken@aol.com Jonathan Thatcher (JT) opened the meeting by displaying a table of CX specification issues that need test procedures. After explaining the content of the chart he left to make copies and Del Hanson (DH) took over the meeting. He displayed a slide that listed agenda items. The critical issue of fiber modal band- width specification was deferred until JT return. Other issues were dealt with as follows. Review of LX TBD's - After explanation of the issue by DH Ed Grivna (EG) moved to accept a single mode fiber connector loss value of 2 dB resulting in a 3 km link length. It was seconded by Shelto Van Doorn (SD) and approved by acclamation with 5 abstentions. - DH explained that the agreed to reduction in single mode fiber attenuation enabled the group to reduce receiver sensitivity from -20 dBm to -29 dBm in table 38-8 for a worst case link. SD moved accept a -19 dBm number and Bob Musk (BM) seconded. Stan Swirhan (StS) commented that the margin gain might be better spent elsewhere. Nevertheless the motion was approved by acclamation with 3 abstentions. -BM moved to introduce a statement defining receiver sensitivity as a value that inherently includes the extinction ratio penalty and is taken to occur at the eye center. The motion was seconded by SD. After some discussion it was accepted by acclamation with 5 abstentions. -DH described a change the to the RIN specification in table 38-7 from -122 to -120 in accordance with discussions of 5/15. BM moved to accept -120 dBM and Paul Kolesar (PK) seconded. The motion was approved by acclamation with 4 abstentions. -PK made a presentation (copy attached) on fiber modal bandwidth requirements. He mentioned that several individual countries use specifications taken from 11801. There are a variety of bandwidth cells specified in standards. His recommendation was to include both 160/500 Mhz.km and 200/500 Mhz.km for multimode fiber. This was based on parameters in ISO 11801. -Geoff Thompson (GT) asked if we need to specify link characteristics. - PK noted the characteristics are need for the link length model. -GT stated 160/500 is required by TIA/EIA 568A but is not permitted by ISO 11801. - Steve Swanson (SS) noted that most countries using 200/500 have merely passed through requirements from ISO 11801 in spite of that documents shortcomings. -PK, DT, Petar Pepljugoski (PP) continued a discussion of link lengths. - SS called attention to the large number of inconsistencies with ISO 11801. -PK stated that his proposal was a viable compromise - PP noted that PK's tables included the wrong operating range for source wavelength. At this point JT returned and instituted a time limit on further discussions of this topic. The guidelines required no more than 1/2 hour of discussion and each person will have one opportunity to speak. Several people then stated that they had presentation to give that should not be included in the time limits. - GT asked if the presentations support the requirement to be finished with the standard by the deadline. -JT stated that the presentations will cover 160 and 200 Mhz.km - Howard Frazier (HF) offered clarification that this standard must at least comply with ISO 11801, but additional items would not be excluded. - GT explained that not including 11801 requirements would result in a "no" ballot when we seek international approval. - SS explained that 160 Mhz.km is included in ISO 11801 and read a letter from the chairman of the 11801 organization (see next) Draft ISO/IEC 11801 Customer Premises Cabling and its migration to a distant future Dr.-Ing. Walter P. von Pattay, Siemens AG, München, Germany "... ISO/IEC 11801 specifies the minimum performance of links, and in its next update also of channels, independently from the actual length of a link or channel. In addition the minimum performance of installed components are specified: of components needed to meet the link performance for a maximum link length given by the standard. In many cases these components are also used for shorter links, which often results in a link performance which is above the minimum required to meet the standard. Such components might be even sufficient to implement short links of an higher link class. On the other hand components which do not meet the minimum requirements for the maximum link length may be used to im- plement shorter links, perfectly meeting the standard. A example would be the presently discussed Optical Fibre (OF) with a modal bandwidth of 160 MHz(km at 850 nm. This fibre is perfectly adequate to establish links with a bandwidth of 100 MHz as long as the link has a length of 1.6 km or less. To be frank: a better fibre could be an overkill for the many links in this distance window. Only when a higher link class is spec- ified in future which is met by the better fibre over the shorter distance there would be a significant return for the extra investment - should there be a price worth wile difference between the OF of 160 MHz(km and the next better one. Nevertheless, for longer links an OF is needed which has a higher modal bandwidth. As ISO/ IEC 11801 gives 2 km as maximum distance for that link an OF with 200 MHz(km is specified in the component clause. ..." - JT made a presentation recommending a reference to 200 but using 160 in the link distance model. - SS made a presentation noting that 3 out of 4 domestic fiber manufacturers prefer the 160 Mhz.km specification. He also displayed a graph that could be used in an informative annex that would specify link distances for higher bandwidth fibers. He also recommended specifying 200 Mhz.km effective bandwidth with a launch condition and measurement technique TBD. -PK noted that this would probability take too long to do. - David Smith (DS) stated that a restricted mode launch bandwidth measurement must be used as overfilled measurements are meaningless with laser sources. - David Cunningham (DC) made a presentation similar to SS's in which he proposed using a table in an informative annex to show link lengths beyond the specification. - DH noted that we can use bandwidths less than the specification and still reach 500 meter links lengths. - StS reinforced DS's point that the bandwidths we are concerned with are defined under restricted launch conditions and really do not relate to the overfilled bandwidths specified in ISO 11801. -DC again indicated we could use a table to define a range of bandwidths. - SS asked to have 160 Mhz.km placed in the table as the minimum value - StS recommended using the word "effective" to qualify the bandwidth value. - PK asked that this be included in an informative annex rather than the normative text. -DS stated that the existing standards are not applicable to our situation because they specify overfilled launch. Laser bandwidth is what must be defined and specified. -JT asked how that might affect the timing deadline the group is required to meet. - HF suggested we adopt a 200/200 specification. - DH suggested we do what DC presented. - Todd Hudson (TH) expressed concern that defining a new launch condition would provide a big problem for the installed base because it would all have to be tested for the new bandwidth value. - SD indicated that bandwidth would change with each different source. - SS agreed that a restricted launch is appropriate. He also indicated that the official US position on ISO 11801 is that it is wrong and must be changed. Finally, he indicated that changing to a 200 Mhz.km value would abandon the embedded base of 160 Mhz.km fiber. -StS stated that we can't fix the restricted launch issue now or in the next 4 months. Our best hope is to reduce risk somehow. - PP indicated that defining a launch condition could eliminate many lasers from use in GBE systems. - DC moved that we suspend the presentations and 1/2 hour of discussion. - StS seconded the motion and it passed by acclamation - DC moved to adopt Length and Bandwidth Data as an informative annex into the next draft of clause 38 for 1000 Base LX and SX (see below) -StS seconded the motion -Howard Johnson (HJ) noted that the standard simply needs a number and the number should be taken from ISO 11801. -HJ offered an unfriendly amendment to add (see below) 200 Mhz.km as the minimum modal bandwidth for both 62.5 um fiber and 50 um fiber. -PK seconded the motion. -SS explained that TIA/EIA 568A was the first standard and the most correct standard. ISO 11801 is inherently flawed. -DC explained that he viewed the amendment as unfriendly because the two issues should not be linked. -EG suggested specifying 200 and 160 - DH suggested that we could change item 13 which required that we comply with ISO 11801. -HF explained that this would require a change by 802.3 and 802. -StS stated that the current discussion does not relate to the motion and than he called the question. An technical amendment to this motion was proposed by Howie Johnson and seconded by Paul Kolesar: and specify the use of 200/200 fiber for SWL 62.5 applications as per ISO 11801. The vote for HJ unfriendly amendment failed : For 14 Against 17 Abstain 4 - PP asked if the link length software could be included with the standard. - HF explained that logistically this would be impossible - SD called the question on DC's motion. A technical motion was made by David Cunningham and seconded by Stan Swirhun: that we adopt link length versus effective modal bandwidth data and in model equations as informative information into the next draft of clause 38 for 1000BASE-LX and SX specifications. The vote for DC's motion passed : For 30 Against 3 Abstain 4 - HF indicated that we still did not have a solution. - SS indicated that our current solution does not comply with goal 11b which requires a 500 meter link length. - JT explained that we do comply. -DH explained that we must now pick a bandwidth number. - HJ asked what is the normative bandwidth and stated that we should specify it now. - PK supported HJ - DS stated that no solution that we have discussed complies with 11801. He further stated that we will be agreeing to write a fictional solution into our specification. We would then be required to fix it later. - StS stated that we need relief from the requirement to support ISO 11801. - TH stated that 200 Mhz.km specification would result in a premium product specification and would therefore increase the overall system cost. - DC stated that we do meet the 11801 requirement via the annex. - EG reiterated that we could include both. - JT clarified that the issue under discussion is that 160 Mhz.km does not meet objective 13 (ISO 11801). - GT stated that we are required to work from standards - John Bowerman (JB) stated that all three options should be included - DC reminded the group that an annex like the one here has precedent in 10Base F - DH stated that we also agreed to change the single mode fiber attenuation value from 1 to 0.5 dB/ km and that does not comply with 11801. At this point the group agreed to take a break while several motions were written. - Stan Swirhun moved and David Smith seconded a motion: To append an informative "effective modal bandwidth (EMB)" specification, where "EMB" is diffident as modal bandwidth measured under laser launch, to any accepted fiber specification and set the values to be 200/500 MHz*km for 62.5 micron and 500/500 for 50 micron. -HF asked if the amendment by StS duplicates the work by TIA FO 2.2 -DS & StS stated that this has a broader scope. -SS concurred - HJ concurred - DC stated that we an measure bandwidth with our transceivers - HF stated that we should not pass this motion because it would result in a delay to that standard which is not acceptable. - GT said he would prefer putting extremely conservative values in the standard rather than delaying it. The vote for StS's technical motion passed : For 25 Against 3 Abstain 4 A motion was made by Paul Kolesar and seconded by Geoff Thomson to: Adopt ISO 11801 bandwidth values for multimode specifications. Specifically: 200/200 MHz*km for 62.5/50 micron MMF with 850 SWL and 500/500 MHz*km 1300 LWL in the normative tables in clause 38. Provide reference to additional data as agreed in motion by David Cunningham. - DS asked if this motion means that an 11801 compliant but failing link will be permitted by the standard. - HF said that if the answer is yes than we have written a bad standard. - JB said that this implies shorter link lengths. The vote for PK's technical motion failed : For 6 Against 14 Abstain 7 -DH pointed out that supporting 11801 w/ .5dB will mean not supporting goal 11c. - JT presented a statement regarding RML to be read at the plenary meeting. - DS and StS objected due to its alarmist nature - HF clarified that the statement is accurate - DC objected based on technical grounds After some further discussion, a small group was tasked to rework the statement, the results of which were as follows: Various organizations have recently begun testing the performance of multimode fiber when driven by laser sources. Previous lab work indicated that the performance of multimode fiber when driven by a laser source might be better than the specified values. Preliminary results from some members of the test team indicate that the performance may be better or worse than expected. While it is too early to draw any conclusions from this preliminary data, there is some risk that we may have some difficulty specifying how to meet our 500m objective for multimode fiber by July, 1997. We will provide additional data about the final test results, and their implications, at our July meeting. HF presented this statement to the PMD sub task force, and asked if there were any objections to it. None were offered, and the statement as shown above was adopted as the position of the PMD sub task force. David Law was persuaded to be the recording secretary at 12:02 for the remaining 10 minutes of the meeting. Jonathan Reviewed the remaining issues. These were Signal Detect and the Del Hanson statement on single mode fiber. Del Suggested should just leave the 0.5dB/km value with support for 3km for now. Will deal with ISO/IEC11801 later. Jonathan We will move on to Signal Detect. Placed a copy of the new signal detect text. 38.1.1.3 PMD_SIGNAL.indicate This primitive is generated by the PMD to indicate the status of the signal being received from the MDI. This signal is optional. 38.1.1.3.1 Semantics of the service primitive PMD_SIGNAL.indicate(SIGNAL_DETECT) The SIGNAL_DETECT parameter can take on one of two values: OK or FAIL, indicating whether the PMD is detecting light at the receiver (OK) or not (FAIL). When SIGNAL_DETECT = FAIL, then rx_bit is undefined, but consequent actions based on PMD_UNITDATA.indicate, where necessary, interpret rx_bit as a logic ZERO. Note: SIGNAL_DETECT = OK does not guaranty that rx_bit is known good. It is possible for a poor quality link to provide sufficient light for a SIGNAL_DETECT = OK indication and still not meet the 10-12 BER objective. 38.1.1.3.2 When generated The PMD generates this primitive to indicate a change in the value of SIGNAL_DETECT. 38.1.1.3.3 Effect of receipt The effect of receipt of this primitive by the client is unspecified by the PMD sublayer. 38.1.2 PMD Signal Status function (optional) If implemented, the PMD Signal Status function shall report to the PMD service interface, using the message PMD_SIGNAL.indicate(SIGNAL_DETECT) which is signaled continuously. PMD_SIGNAL is intended to be a rough lights ON or OFF indicator. SIGNAL_DETECT is set to OK when the PMD circuitry senses the reception of light. Otherwise, SIGNAL_DETECT is set to FAIL. The receiving PMD shall indicate a FAIL condition when the link is unplugged or the transmitter to which it is attached is turned off. The optical power level for an off transmitter is defined in Table 0-1-. Individual PMD suppliers must guarantee that under worst case conditions there are no false positive OK indications. Though unspecified, this implies that there must be adequate margin between the SIGNAL_DETECT trip point and the inherent noise level of the PMD due to cross talk, power supply noise, etc. PMD suppliers must guarantee that under worst case conditions, an incoming signal at or above the minimum receive threshold (e.g. -17 dBm for 1000BASE-SX) will not indicate FAIL. Though unspecified, this implies that there must be adequate margin between the SIGNAL_DETECT trip point and the receiver sensitivity minimum. Table 0-1-To be stuffed into SX / LX tables Units Optical Power of off optical transmitter (max) dBm -30 Response time requirements are not specified. It is expected that SIGNAL_DETECT may be chatter at some optical input level. It is expected that the PMD service interface will be designed to handle this. If the Signal Status function is not implemented by the PMD, then the SIGNAL_DETECT indicator shall be set to OK.. Editor note (to be removed prior to final publication): We need to decide if there is a need to specify the optical/timing characteristics of the Signal Status. Del What is the difference between this and the text discussed on the reflector around the 1st April. Jonathan I believe there are no differences. Howie Is the intent to specify this loosely David Does not reflect the reflector discussion Jonathan There was not that much reflector discussion David So this will not protect against crosstalk Howie Pointed out that the text did state that the function should protect against crosstalk Del Pointed out that the whole function is optional anyway. Motion Move that adopt the signal detect text as written. Moved: Del Hanson Second: Howie Johnson There was no discussion. Motion was accepted by acclamation. Jonathan Review that additional text that was also included with the signal detect slide. This text was on Central Wavelength, Spectral Width, Optical power measurement, modal noise and RIN. Motion Move that adopt the additional text included on the signal detect slide as written. Moved: Schelto Second: Del Hanson There was no discussion. Motion was accepted by acclamation. The meeting adjourned at 12:17. ######################################################################### Friday PM Closing Session of IEEE P802.3z. 18) Chair, Howard Frazier called meeting to order. - 802.3z meeting only. Not a joint session with 802.3ab. 19) Review of resolution of big ticket items. - Resolution of big ticket items presented in item 16 above. - All big ticket items closed. 20) Repeater Sub Task Force Report. - By Steve Haddock. 21) Management Sub Task Force Report. - by David Law. 22) GMII Sub Task Force Report. - By Bob Grow. 23) PCS / PMA Sub Task Force Report. - By Rich Taborek. 24) Question: From Geoff Thompson: Regarding closure of TBDs? - Answer: From editors go over details of status. Remove editors notes which point to presentations. Result: No editors notes, No TBDs. Answer: From Howard Frazier: Clauses are complete and free of TBDs. 24) Report on interim meeting arrangements. - Hotel choices. a) Ramada Hotel at Gatwick Airport. or b) Strand Palace in London. - Comments and discussion. - Graham Measor - Colin Whitby-Strevens - Steve Haddock - David Law - Scott Mason - Pat Thaler - Straw pol results a) Ramada: 21 votes, b) Strand: 33 votes, Strand wins. - Frazier: How many plan to attend the meeting? Answer: 61 people raised hands. - Frazier: Plan for 90 to 100 people. 25) Length of London Meeting. - 3 or 4 day meetings? - Straw pol 4 day meeting wins, overwhelming. Based on these pols, the meeting will be held Monday September 8th through Thursday September 11th, 1997, at the Strand Palace in London. 26) October interim In Bay Area. - Frazier: Discussion of October interim held at bay area hotel. - Frazier: Decide details at July meeting. 27) Hotel Security. - Presentation by Derek Gisburne, director of guest relations. 28) PMD Status Of Big Ticket Items. - By Jonathan Thatcher. - Copper: Two open issues (TBDs) to be resolved at the July meeting. - Optical: Four open issues (TBDs) to be resolved at the July meeting. 29) Presentation And Statement. - By Jonathan Thatcher. - The following slide was presented: "Various organizations have recently begun testing the performance of multimode fiber when driven by laser sources. Previous lab work indicated that the performance of multimode fiber when driven by a laser source might be better than the specified values. Preliminary results from some members of the test team indicate that the performance may be better or worse than expected. While it is too early to draw any conclusions from this preliminary data, there is some risk that we may have some difficulty specifying how to meet our 500 m objective for multimode fiber by July, 1997. We will provide additional data about the final test results, and there implications, at our July meeting." Question: Geoff Thompson: About meeting 500 m objective? Thatcher: Confident in ability to meet 500 m goal at this time. Current specification method is at risk, not the 500 m goal. - Discussion, questions and answers. - Pat Thaler - Geoff Thompson - Shimon Muller - Scott Mason 30) New Business, Motions. Motion 1: Hanson / Kolesar "With respect to ISO 11801 Table 2.2 multimode optical fiber/cable parameters. Gigabit Ethernet supports the following link lengths (based on informative text in the document). Wave Length Minimum Effective Worst Case (nm) Modal Bandwidth Link Length (MHz*km) (m) 850 200 300 300 500 550 If the 850 nm min. effective modal bandwidth is 500 MHz*km for 50 mmf, the worst case minimum link length increases from 300 m to 550 m. We propose adding this case." -Discussion and questions. - Scott Mason - Shimon Muller - Howard Frazier - David Cunningham - Jonathan Thatcher - Steve Swanson Friendly Amendment: to motion 1 by Steve Swanson: - Consisted of the following table. - Response: No not friendly. Description Unit 50 um MMF 62.5 um MMF SMF @850 nm @1300 nm @850 nm @1300 nm @1300 nm Operating m 2to>500 2to>500 2to>250 2to>500 2to>3000 Distance Max. Atten. dB/km 3.5 1.5 3.5 1.5 1.0 Min. modal BW MHz*km 500 500 100 500 N/A Max. disp. ps/km*nm <= 0.11 <=0.11 <=0.093 slope Zero disp. um 1295-1320 nm 1320-1365 nm 1300-1324 nm wavelength Max. disp. ps/nm N/A N/A N/A N/A 18 Max. link db TBD TBD TBD TBD TBD atten. Conn. return db 20 20 20 20 26 loss Link Penal. db TBD TBD TBD TBD TBD - Amendment: Swanson - Second: Harshbarger - Questions and discussion: - S. Swanson - D Harshbarger - Bob Grow - Howard Johnson - Vote on amendment Yes: 6, No: 28, Abs: 24, Amendment fails. - Call the question, on motion, Yes: 29, No: 13, Procedural 50% passes. - Vote on motion: Yes: 27, No: 6, Abs: 27, Technical 75% - Hanson / Kolesar motion passes. Motion 2: Jonathan Thatcher: Diagnostics: -Motion: Thatcher -Second: Lahat "The PMD sub working group requests the upper layer sub working groups to add a diagnostic mode to generate and detect/test the following patterns: 1) +/- K28.5 2) +/- K28.7 3) K21.5 4) The FC jitter working group's "RPAT" For the purpose of doing link diagnostics and system level conformance." Amendment to motion 2: - Amendment: Johnson - Second: Grow "Add a diagnostic mode informative annex to describe the following patterns: 1) +/- K28.5 2) +/- K28.7 3) K21.5 4) The FC jitter working group's "CRPAT" For the purpose of doing link diagnostics and system level conformance." 2nd Amendment to motion 2: "Add optional diagnostic mode to generate and detect/test the following normative test patterns: 1) +/- k28.5 2)+/- k28.7 5 ones, five zeros contin = lowest freq 3)D21.5 4) The FC working groups "CRPAT" for the purpose of doing link diagnostics and system level conformance. 2nd Amendment: Colin Whitby-Strevens offered as friendly, not accepted as friendly, 2nd amendment withdrawn - Vote on Johnson/Grow amendment to motion 2: Yes: 36, No: 3, Abs: 6, Technical 75%, amendment passes. - Vote on motion, technical 75%, Amended version, Pass by acclamation. Motion 3: Jonathan Thatcher. - Motion: Thatcher - Second: Hanson "Move to adopt for clause 38 "Recommended Table"" (Presented below) Recommended Table: Description Unit 50 um MMF 62.5 um MMF SMF @850 nm @1300 nm @850 nm @1300 nm @1300 nm Operating m 2to>500 2to>500 2to>300 2to>500 2to>3000 Distance Max. Atten. dB/km 3.5 1.5 3.5 1.5 .5 Min. modal BW MHz*km 500 500 200 500 N/A Max. disp. ps/km*nm <= 0.11 <=0.11 <=0.093 slope Zero disp. um 1295-1320 nm 1320-1365 nm 1300-1324 nm wavelength Max. disp. ps/nm N/A N/A N/A N/A 18 Max. link db 1.9 0.81 1.13 0.81 1.62 atten. Conn. return db 20 20 20 20 26 loss Link Penal. db 1.05 4.60 5.42 4.11 3.34 (inc conn.) Technical 75% Vote on motion: Motion Passes by Acclamation Motion 4: Howard Johnson / Gadi Lahat Motion: Johnson Second: Lahat "Add to D2.1 the resolution of all comments as of this meeting, plus other editorial changes as necessary, to produce D3. Distribute D3 to 802.3 in satisfaction of the preview requirements for a WG ballot discussion at the July plenary." Technical 75% Vote on motion: Motion passes by acclamation. Motion 5: "Approve the minutes of the Irvine meeting." Technical 75% Vote on motion: Motion passes by acclamation. Motion 6: "Adjourn the meeting." Procedural 50% Vote on motion: Motion passes by acclamation. End of Minutes