Minutes of the 802.3z Gigabit Task Force Closing Plenary 11 March, 1998, Irvine, CA The meeting was called to order by Howard Frazier at 3:25 p.m. Nick Esser was conscripted as Recording Secretary for the session. SUB-TASK FORCE REPORTS: Howard Frazier, Clauses 3, 4, 5, and 31: Only one comment! No issues. David Law, Clause 30, Management: No significant issues. Howard Johnson, Overview Clauses: No significant issues. (See below.) Bob Grow, Clauses 6,22,35: All comments are closed. Steve Haddock, Clauses 41, 42: 4 Editorial comments, no Tech. Rich Taborek, clauses 36 and 37: "We're done!" Del Hanson, Clause 38, Optical PMD: Del offered perspectives on the Link length specifications -- I. The link analysis model used to define lengths is considered to be conservative. II. Optical transmitters, optical receivers and SerDes meeting specifications and the defined conformance tests will support the specified link lengths for MMF cable which meets its modal bandwidth specification, as listed in table 38-12. Link Length Cells were added were added for 1 additional Modal Bandwidth characterization for both 62 and 50 micron fibers, as follows: MMF Type Modal Bandwidth(MHz*km) Link Length (m) (SX/LX) (SX/LX) 62MMF 160/500 220/550 (existing) 62MMF 200/500 275/550 (new) 50MMF 400/400 500/550 (existing) 50MMF 500/500 550/550 (new) The LX link lengths above are achieved with an offset jumper cable. The 100BASE-LX SMF link length is increased to 5000m by the following: The SMF LX power budget has been increased from 5.5dB to 8.0dB to be compatible with the MMF launched power specifications of -11.5dBm with an external SMF offset-launch patch cord into MMF. In response to comment #57, with the additional power budget the link length has increased from 3000m to 5000m while retaining a 1dB unallocated margin. Andy Luque: What about the notes that were to be added? Dell Hanson (DH): The group decided to resolve all work around the table entries (as reported above.) Dell reported that all 56 comments from draft D4.1 have been resolved. The following comments relating to DMD-related jitter issues and meeting link specification PAR objectives are addressed by the response described below. Comment Commenter Type Notes 115 Thompson TR Recirculated 116 Lin TR Recirculated 117 Nowell TR Recirculated 122 Kolesar TR Recirculated 77 Cunningham TR D4.1 [84 Swanson TR D4.1] 85 Thatcher TR D4.1 86 Thatcher TR D4.1 92 Thatcher TR D4.1 114 Thompson T Recirculated The Optical PMD group took 7 actions to resolve outstanding TR comments related to DMD-jitter, interoperability, receiver bandwidth specification and supported lengths: I. Multimode fiber (MMF) differential mode delay (DMD) has been mitigated [not eliminated but made manageable] by defining a required transmitter coupled power ratio (CPR) range to limit launch conditions into MMF. II. The 1000BASE-SX optical power budget has been increased from 7.0dB to 7.5dB to support 400 MHz*km fiber/cable to 500m. This is achieved by increasing the minimum transmitter power from -10dBm to -9.5dBm. III. The receiver upper 3dB electrical cutoff frequency has been specified to be 1500 MHz to filter excess DMD-related jitter. [This requires action IV.] IV. The receiver stressed sensitivity and stressed contributed jitter conformance specifications have been imposed to ensure link interoperability. These stressed receiver tests impose specified intersymbol interference (ISI) and duty cycle distortion (DCD) - caused baud period shrinkage. V. Stressed receiver sensitivity and contributed jitter specifications have been defined. VI. The jitter budget has been revised to account for jitter contributions throughout the link [including DMD]. VII. The link analysis model has been updated to include DCD and jitter. This model has been confirmed experimentally and is the basis for defining worst case link lengths specified in this standard. The commenters have not yet been asked if they will sign off on this proposed resolution. There is also a TR comment that 1000BASE-LX RIN specification should be reduced from -120dB/Hz to -116dB/Hz, submitted by a non-voter. It was rejected and may be resubmitted by a voting member in the next cycle. Changes were proposed for the Jitter Budget, Table 38-10: 1. Add 96ps MMF Deterministic jitter 2. Keep (SerDes Tx) total jitter the same; Reallocate by reducing DJ & increasing RJ 3. Reduce (FO Rx Added) jitter due to increasing input test level of -12.5dBm 4. Reallocate (FO RX Added) jitter by reducing DJ & increasing RJ 5. Reduce Allowed SerDes Clock Recovery window The revised Table 38-10 for D4.2 is now: (Changed parameters in *xxx*.) Total Jitter Deterministic Random TJ(UI) TJ(ps) DJ(UI) DJ(ps) RJ(UI) RJ(ps) SerDes Tx, TP1 0.240 192 0.100 80 0.140 112 FO Tx Added 0.284 227 0.100 80 0.184 147 FO Tx Out, TP2 0.431 345 0.200 160 0.231 185 Fiber Added *0.170* *136* *0.050* *40* 0.120 96 Fiber Out, TP3 *0.510* *408* *0.250* *200* 0.260 208 FO Rx Added *0.332* *266* *0.212* *170* 0.120 96 FO Rx Out, TP4 *0.749* *599* *0.462* *370* 0.287 229 SerDes Rx Window *0.251* *201* Howard Frazier (HF): Concerning change II., was this made to support a specific comment? DH: Yes, Swanson's comment, #84. HF: Asked for confirmation that the commenters had not signed-off. DH: Yes, but there has been no time so far. HF: I need to know how many comments must go with the next recirculation. All Big Ticket items from D4.1 are closed: Description Clause Comment # Status Effect of DCD DJ on link length 38 85,86 closed Jitter budget and how to meet 38 87,88,114, closed 802.3z objectives 116,117,122 Receiver Bandwidth 38 119,120,121 closed Multimode fiber bandwidth cells 38 84 closed Increase LX distance on single-mode 38 57 closed There was a discussion of the TR comment on reducing 1000BASE-LX RIN, by Mike Dudek (MD), a non-voter. MD explained his comment, with the bottom line being that it would make it easier to make products. Comment: We specifically asked the manufacturers, and now it wouldn't be right to change. Comment: If we were to change this value, it would ripple through the spec. Wouldn't it involve changes to the testing methods? MD disagreed with this. Comment: Is there any support from voting members? MD: 2 have expressed support. Paul Kolesar (potentially supporting voting member): There may be a problem with more than 1 vendor. It deserves consideration. HF asked Geoff Thompson for advice on how to treat a nonvoting comment. Geoff said it should be treated as a valid comment, and be recirculated with D4.2. It is comment #49. Howard Johnson (HJ) brought up a comment by GT on Clause 7 (#112,) which will be recirculated. The response group believed it was outside the scope of the PAR. Pat Thaler: This will be included in Maintenance #5 Recirculation package, P802.3aa. The comment was that the clock spec. for 10Mb/s was inadvertently deleted by 802.3x. A response was drafted with no objection. HF: Does anyone know of any other TR comments outstanding? Response: There was a group of comments from Joe Quinn. Some were accepted, some rejected. They have all been worked on before; some comments were resubmitted by Joe. Comments 59 & 60 were accepted in principle. Comment #61 may be recirculated. No other outstanding TR's brought forth. HF: Is there any other business affecting this draft? MOTION #1: Moved: Jonathan Thatcher, Second: Rich Taborek *Create a new subsection called "36.A.5 Short continuous random test pattern." *Rename 36.A.4 "Long continuous Random test pattern." *36.A.5: This test pattern is identical to 36.A.4 except that: 1. The loop length is 29 2. Ther CRC is "2F E0 AA EF" 3. The IPG as seen on the link is: /T/R/I1,I2,I2,I2,I2 *Put the 10b - bit pattern on the reflector for reference by PMD manufacturers. (Technical) Discussion: Tom Dineen: What about the cost to implement? This is an 11th hour change. Jonathan: The existing pattern is too long for existing BERT testers. All that has to be done is to load this new pattern into the testers. It doesn't take very long. The motion was voted on: FOR: 46, AGAINST: 1, ABSTENTIONS: 13. PASSED. HJ proposed a revised schedule: 17 March: D4.2 to IEEE 1 April - 20 April: Sponsor ballot recirculation 30 April - 1 May: Interim Meeting 7 May: D5 to IEEE 15 May - 31 May: Recirculation (as necessary) 6 June - 19 June: SEC e-mail ballot (if necessary) 24 June - 25 June: Standards Board MOTION #2: Moved Jonathan Thatcher, Second: Bill Quackenbush Having adopted resolutions for all comments received on drafts D4 and D4.1 of P802.3z, the Gigabit Task Force requests that 802.3: 1. Affirm the work produced by 802.3z at this meeting. 2. Request a sponsor ballot recirculation of P802.3z/D4.2, with a copy to ISO JTC1/SC6. 3. Authorize 802.3z to respond to comments generated during the recirculation process, and to produce D5 based on those responses. 4. Request that 802 forward P802.3z to RevCom for consideration at the June Standards Board meeting, contingent upon successful completion of the recirculation process. (Technical) Friendly amendments for wordsmithing were offered and accepted, and clarifying questions were answered. The question was called with no objections. FOR: 74, AGAINST: 0, ABSTENTIONS: 0. PASSED. No corrections were offered to the minutes from the Seattle interim meeting. MOTION #3: Moved: Bob Grow, Second: Larry Ruben Accept the Seattle minutes as written. Approved by Acclamation. Information on the next interim meeting was given: Meeting to be in the Boston area, specifically Merrimac, NH, hosted by Ben Brown, Cabletron Systems. Dates: 28 April - 1 May Place: Merrimac Hotel and Convention Center Blocking: 130 rooms, may be too many. We need to take a count at the .3 closing plenary. Rate and cutoff date not yet known. Tentative scheduling: TSG - Tue. 28 April - Wed. a.m. 29 April .3ac - Wed. 29 April .3z - Thur. 30 April - Fri. 1 May .1 will co-locate, and will meet on Thur. 30 April, if at all. Meeting adjourned at 5:15 p.m.