MBI meeting 8 Mar 98 MBI 8Mar98 Irvine,CA John Abbott David McCallum Dan Brown Vince Melendy C Brown Peter Monarski Masum Choudhury Rodney Mui Phil Craine Hari Naidu David Cunningham Rick Neumann Chris Diminico Mark Nowell Thomas Dineen Petar Pepeljugoski Mike Dudek Dmitry Polonsky Joel Goergen Lutz Raddatz Del Hansen Wali Rousta Doug Harshbarger John Rutger Kenneth Jackson Dave Siljenberg Howard Johnson Steve Swanson Dick Kirk Jim Tatum Don Knasel Jonathan Thatcher Robert Lee Bruce Tolley Jeff Martin Luis Torres Arlen Martin Cynthia Uesato Tom Mathey Shelto Van Doorn Bob Mayer 1. Del Hansen to give us an update 2. Then discuss MBI investigation into the past events Introductions Del Hansen / HP Resolving comments and moving on with draft spec - 56 comments - 30 TR s - 9 technical - 17 editorial * jitter budget and not meeting 802.3 objectives * receiver bandwidth * additional MMF bandwidth * extended SMF link length from 3km to 5km * jitter budget and not meeting 802.3 objectives * new jitter table that was discussed in Seattle * reduce budget for receiver, allocate to other areas / fiber * * receiver bandwidth * additional MMF bandwidth * extended SMF link length from 3km to 5km Procedural question (Jonathan): concerned about global response .... Response should be real data to go into the document this week. Howie Johnson: - sponsor ballot comment resolution - problem : some comments are serious enough to do something about - min change to get it done - if legit problem, then must fix - 802.3 authorized MBI work plan - delay standard to collect more data - believes enough data to make final decision - ask PMD come to final resolution on all items and recommend one motion at 802.3 - next interim is in May and should have the standard done David C Agenda: - SerDes/jitter issues - Uesato / HP : SerDes 5min - Amcc and vitesse???? - Receiver conformance - Jonathan Thatcher 15 min - David Siljinberg ... 20 min - Del Hansen 10 min - Trans conformance - Mark Nowell 10 min - Jim Tatum 10 min - Rick Nuemann 10 min - Del Hansen 5 min - David Cunningham 15 min - Petar Pepulaguski 15 min - Fiber - John Abbott 10-15 min - Laptop announcement / 4675 security- to guard room Initial discussions - jitter DJ, etc Jonathan Thatcher SerDes Discussions: Uesato / HP - Jitter budget on receiver 40ps smaller window - Based on eye scan method ... or scope methode - RJ scope not adequate - Copies of pattern if people want to test with it - 10 to 11 parts sampled Receiver Conformance Discussions: Jonathan Thatcher / IBM: - 3.5db we have now and remove .5db random stuff - .36db penalty added to eye closure under assumption of 20% ... - Does not like 3.5db input to receiver .... Looking more at 2.5db - Believes 120ps at tp2 is max jitter we can accept - Believes we are at 200m using 160mhz*km fiber Dave Siljenberg: - Power supply 20ps - Data pattern : running disparity - Cross talk - Some parts fail 12.5db sensitivity - Spread of parts is worrysome - Variation in different parts from same supplier - Was done at normal ambient Del Hanson / HP: - 195ps opening out of receiver - Need conformance test for receiver - Time to do 100% test is too long - Doesn't know of the power supply was varied to min/max - Date k28.5 Break 10 min Mark Nowell / HP - Elipticity on near field after 10m - 100m more circular - Near fields change with patch cord - Question about repeatability - SX issue on mode conditioning Jim Tatum / Honeywell - Jitter table looks okay ???? Rick Neumann / Cielo - Fiber 2 orange - Should be some way to characterize good launch ....?? - Near field measurements: measured or parabola? - Dozens of eyes with various vixels ????? Del Hansen / HP - MBI in Boston - Fiber contributed jitter - Thinks we have to use numbers as they are - All vixel based transmitters - No data on CPR - Fall in range of CPRs we have today - Receivers: one of two lots reported before - BW 850mhz of receiver - 200ps of jitter in the fiber - (comment is serious problem) David Cunningham / HP - LX fiber - SX fiber - SX worst case modal bandwidth is 100mhz*km - Disagree we have time to specify conditioned launch - 7un spot: particular launch (reference figure 8) - 4ns per km was point 2sig distribution - normalized back to field population - Bandwidths in the field could show BW less then 160 - J Ritger: some of the fiber was not FDDI compatible and some failed mfg spec. - Then this stuff should not be included - was not FDDI grade - Failure in 62.5un fiber with an arbitrary launch shows about 7% error????? Petar Pepulaguski / IBM - Color codes indicate 'by transmitter' - DJ at transmitter is zero - Numbers show DJ in fiber is small (16ns)?????? - Keep same 160mhz*km, but reduce length to 200meters - Reduce fiber contributed jitter to 30ps Break - resume 2:35pm John Abbott / Corning - fiber David: Heard presentations and now are at the stage on how to finish it off. Del - Discuss link lengths first. Tom - Make a list of things we want to do and how much time. Chris - Resolution of the comments. This meeting was to come back and show what we had done to resolve TR's we had against the standard. Del - ???? Chris - ??? David - Tests, link lengths ???? Issues for Discussion - Link length - Receiver Conformance - Jitter Budget - Transmit test / MPD - Integration into Draft Issues for Discussion - Link length - What will happen should we change 550 meter objective. Howie - We have something that goes 550 meters, but how painful would reduced distance sx be? Jeff Martin / Bay Networks Bruce Tolley / 3Com Present graph of 'market potential VS distance' How would this graph change if we delay the standard further to get the preferred numbers? How many increments of time are we willing to take to get to a distance? Current install base vs new fiber How does curve change since we still support 500m How is cost involved? [8 companies signed up to make patch cord] Break until 4:15 - Receiver Conformance David: Companies have tried test and it is doable??? In stressing the receivers, there are additional penalties. 140ps high frequency jitter???? Fiber doesn't appear to add jitter, but rather close the eye ..... Distance effects ISI which effects test equipment. Chris: Summarize whether they believe they can support 200meters or 260meters more. Anyone object to receiver bandwidth test??? Mike: agrees if it stays high frequency test Del: ??? Jonathan: ??? Generally has support. Receiver stress testing: Comments: Difficult, but doable. This does a good job of sorting ... but painful. Time consuming test. 1. Allow ISI penalty 2. DJ Level 3. Voltage / temp corners 4. With TX turned on and asynchronous 5. Data pattern Howie: wants three people to propose solutions for the above (3,4,and 5) Del, Jonathan, and Mike Indication that some level of redesign may be needed on the receivers????? - Jitter Budget Allocated jitter to fiber and took some away from the SerDes. Kept the total the same for SerDes, juggle DJ and RJ / with no objection 1. change SerDes 2. squeeze the receiver window 3. squeeze the fiber - Transmit test / MPD MPD tests ... Bob: We should not add it as a spec today because there is a group working on it now. David: Can't add to document because it would be an experimental test method. Bob: ... Mark: Issue is time. For LX, conditional launch has been on the table for a while. The issues is how to get it on the table now. David: Spec for patch cord does not reference MPD ????? Many vendors have made a patch cord to these specs , all patch cords have been proven to work Bob: No question that patch cords work. Yes .. it is a solution. Generates a ..... If CPR is good enough, should be good enough for both. Mike: Is it possible to make something that passes the test, but fails in the field?? Dan: ??? David: Chris: deficiency on the SX spec???? Bob: Jonathan: thought short wave was in a hole, and there was a threat it may be out of the standard. Leaves you a different set of thoughts to solve this ???? All tests and verification takes time .... But don't think community will accept ....??? Mark: Different view of what conditioned launch will get you .... 260meters or more???? David: Conditional launch will help OFL and get us to 260 meters. On MPD, ????? Pat: If we could save the distance by going three months, might be okay. But anything else would be .... On the market place. Phil: Bruce: We thought we would have MPD test in Feb. Chris: Mark: Conceptually, right idea. We had a document that talked about MPD, but know one had done it. Questions some of the assumptions. Bob: Pat: Howie: Say we didn't do anything to the transmitter, is there anyone who thinks we could not get at least 200 meters? Bob: what is the bandwidth of the 1% of fibers that won't work? Howie: * Fiber: cutback bandwidth * Install base bandwidth * What about reducing 99% to 95%? Howard: * 500meter formal objective * time line objective - missed it * 99% Informally discussed only, not formal * Worth while to discuss how realistic and what are the problems Interpretation: A lot of fiber out there that doesn't meet the specs. [99% of the fibers that meet their over fill bandwidth] - Integration into Draft Howie: 1. Group of people to work out three items in the receiver test 2. Jonathan's proposal 3. 95% vs 99% 4. Jim Tatum proposal Break until 8pm 802.3z PMD 9 Mar 98 8:30 Start/Review Comments: 56 total 30 TRs [12 are carry-overs] 9 T 17 Editorial Key comment categories: Jitter Receiver Bandwidth MMF Bandwidth SMF length 3km to 5km Howie Johnson: Goal: Finish the standard Report on MBI: Recommendations from the MBI 1. Long morning session 2. Copy of all presentations .... By 5pm tonight 125 copies 3. Results of 8pm discussions * Link lengths * If we do anything at all, looks bad to customers * Receiver tests * Bandwidth test * Wide acceptance * Stress test * General agreement it was a good test * Discuss exact parameters * How much duty cycle distortion comes in to play * Transmitters create about 90ps distortion [ about -40meter] * Is it real? Group feels yes - no objections and cost 40meters link length * Jonathan looked at 220meters 62.5 SX and is going to propose something today * Del and Mike to report on temp, voltage and xtalk * Jitter budget changes * SerDes guys do not object to changes in their area * Transmitter conformance tests - MPD * At this moment, we can not define a test that works all the time * We can eliminate some bad launches - ie CPR test in the document * Does not eliminate DMD effect - or find ....??? * Fiber Specifications * Worst case TX, worst RX, worst length, worst DMD, etc, 99% of time it should work * Conclusion is with spec as is we will not meet it * Some TXs may cause 5% of fibers with worst DMD to fail. [95% pass] * Took some info from fiber guys to determine what percent of the time this could happen. Stats argument is are hard to make. But we can discriminate between 1% and 5% level. Lots of modeling and field measurements. * How often is something not going to work in the field. All worst case in the field is a small enough number that we should go ahead. * David Cunningham to quantify number * Straw pole: 24 [95%] to 2 [99%] should we continue to work on DMD test or should we take what we have and use probability of failure and move ahead with what we have. Customer would use different TX or fiber. * What happens with DMD studies? No formal action. We have asked TIA to look at the fiber. Can we make transmitters that are better? Most feel yes. * Chris would like to see us come out with a recommendation to get to 300 m. If we can not tell them what to do for 100% confidence, we need to tell them what to install. Pursue and calculate BW number. No further discussion Del: Introductions Jonathan Thatcher / IBM * Different kinds of DJ * Allocated 90ps for accordion affect / duty cycle distortion * 62.5 - 160mhz*km - 220meters / see proposal * Forgot receiver numbers in the changes, distance may go to 200meters ???? * DCD measurement???? how??? * [DCD seems to be a factor of receiver bandwidth] * Petar - ..... * Del - support DCD and TX, but not sure about receiver opening * David - primary fact of short pulse ... receiver has filter in it ... pulse is shortened in time, but the receive filters will broaden the pulse out again. Jonathan's numbers seem to be severe .. Del: 1. Data presented can be reviewed 2. Review editorial comments 3. Discuss Jonathan's comments Motion to go out and do work to evaluate DCD, etc on Jonathan's [Accept in principle the recommendations of the MBI group to accept RX test and new DCD imposed penalties and new jitter table with specific DCD values to be brought back by:] Work from 1-3 today during the plenary Jonathan 1st David 2nd Jonathan Thantcher David Cunningham Del Hansen Mike d Mark nowell Jim Tatum Shelto Petar David Siljinberg Lou torres Doug Harshbear Mohammid Jim: motion too strong, too much stuff, saying we are going to make change. Should say we are going to go work on it and come back with numbers. Chris: "evaluate recommendation" rather then accept in principle Tom: time for commitment .... Move along Del: group should come back with recommendations because .... Howie: two hours time ???? Del: Break until 11:30 until this group can figure out the numbers Vote : unanimous Break until 11:30 12 noon start Motion Accept the link analysis proposal from the ad hoc group. Table from Jonathan on Minimum average power penalty Jim Tatum 1st Jonathan Thatcher 2nd Based on unconditioned launch for SX. Chris feels 220meters and stress testing would accommodate 99% of fiber. Note: 50ps was moved to the receiver because the portion of the 90ps from the fiber - 40 belonged to the fiber - 50 actually belonged back to the receiver because of the filters and eye closure that was thought to be there. 62.5 / 260 meter required 185mhz*km 62.5 / 300 meter required 220mhz*km Howie called question Motion passed 27 yes, 2 no, 18 abstain Comment resolution #59 needs sign off - no objections #47 - use 123 Note - no objection to using Howie's comments in #123 [Accept new response] page 38.11 line 6 which states "RIN is referred to RIN12 in the referenced document" no objects #75 David Accepts No objections #118 Re-affirm what we have already done in this area. Defer for now based on editorial. #79 At line 5 change "conditioned launch (CL) produces" to "The CPR specification provides" Remove note c from table 38-3 Delete 38-6 paragraph lines 10-14 Accept ??? #48 Accept #119 Accept #88 Table 38-4, add (avg) to the units column Page 38-6 line 35 change to "receive power (max)" Page 38-6 line 37 change to "receive sensitivity (min)" Reject????? #87 Input optical power to replace input receiver Table 38-1 replace receive power with average receive power Break comments for Don's proposal, and then continue comment resolution at 7:30. Don Knasel / Corning 220meter lengths do not seem very useful Corning went off to see if there is something they could do to take another crack at this? 99% of the installed base of 62.5 multimode fiber has a modal bandwidth of greater then or equal to 185 mhz*km at 850nm. Paul Kolesar feels they are at 185mhz*km since 1995. Chris: are we proposing to revise a table????? Does not feel we can validate this information???? Doesn't feel comfortable ...... Optical fiber cable spec???? Motion: Change the BW used in the link model to 185mhz*km, which is a statistical representation of the installed base of fiber. This will increase the supported length to 260 meters for 62.5un SX. Don Knasel 1st Mike dudek 2nd Mark - friendly amendment add note to table 38-12 referencing this bandwidth to the 160mhz*km cell. Jonathan: ran model 260meters, closes at 190mhz*km, .24db ISI at 185mhz*km. For 185, would prefer the distance be 250 meters. Mike: thinks Jonathan's model is pessimistic and Petar's is optimistic. Jeff: Just a characteristic of 160mhz*km. Chris: selling fiber since 1990 and thinks this won't work. David: Thinks Petar's model is realistic to most fibers. He agrees his model is pessimistic. Agrees with Jonathan's numbers and wants it 260 meters. Howie: Speaks in support. Ask xcvr vendors to shave specs to meet this. Bruce: .... Bob: comes up with 255 meters ...... Howie calls the question Yes 34 No 3 Abstain 5 Resuming 7:30 7:30pm - meeting resumed #120 accepted #86 deferred #89 withdrawn #80 reject #49 yes 11 no 3 abstain 8 not agreed #60 accept #23 yes 15 no 0 abstain 13 #90 accepted #81 accepted #91 accepted #121 accepted #85 deferred #57 yes 21 no 0 absatin 5 #46 accepted #77 table #82 table #83 withdrawn #93 withdrawn #92 table #78 accept #122 table #76 accept #25 ??? #26 accept #61 rejected yes 24 no 0 abstain 2 ????? Motion: Present options to 802.3 (per Chris Diminico's slide) for consideration tomorrow with editorial changes as needed. Howard Johnson 1st Tom Dineen 2nd Amend option 3 to be 275meters at 200mhz*km Mike Dudek 1st Paul Kolesar 2nd Technical: yes 24 no 2 Abstain 6 Call the question Technical: yes 29 no 0 Abstain 2 Passes Motion: Add option 5 to include the following BW cells Fiber type Min Range Modal BW 62.5 2-220 160/500 62.5 2-275 200/500 50.0 2-500 400/600 50.0 2-550 500/500 Don Knasel 1st Paul Kolesar 2nd Technical yes 8 no 17 abstain 8 Technical yes 12 no 16 abstain 5 Fails Motion: Contingent upon referencing multiple band width cells for 62.5un multi mode fiber, propose the following bandwidth cells for 50un mullti mode fiber. Short wave 850nm 400/600 2-500 meters 500/500 2-550 meters 400/400 2-500 meters Jonathan Thatcher 1st Jim Tatum 2nd Call the question: yes 22 no 2 a ? Technical: yes 22 no 5 abstain 2 Passes Notes from 802.3 plenary Howard Frazier 1. Cabling BW spec inconsistent with link model cable BW. Modal BW assumed to be >=185mhz for 99% cables. 62.5 2-250 160mhz*km 2. Cabling BW spec consistent with link model cabling BW. 62.5 2-220 160mhz*km * Note : it is expected that 99% of the installed 160/500mhz*km 62.5um MMF meets this specification. 3. Two cabling BW specs consistent with link model cabling BW. * 62.5 220 160mhz*km * 62.5 275 200mhz*km 4. Cabling BW spec consistent with link model cabling BW. set the BW to 185mhz*km with 2-250meter distance. * 62.5 250 185mhz*km * Note : it is expected that 99% of the installed 160/500mhz*km 62.5um MMF meets this specification. Steve Swanson: 80% of install is 160mhz*km and 20% of install is 200mhz*km. Bob Grow: 160mhz*km is a US standard, 200mhz*km is an international standard. Steve Haddock: Note to three that 99% of fibers meet 185mhz*km. Geoff Thompson: 200mhz*km is not in TIA and not popular in the US Allen: different view on 20% being 200mhz*km David Cunningham: Not against note about 185mhz*km and favors option 3 Howard Frazier: Options 2 and 3 are most defensible and least risky. Tom Dineen: ? Howard Frazier: 1. Anyone can vote 2. Vote as many times as you want Straw Pole on options: 1. 3 2. 11 3. 128 4. 1 Given option 3 is successful, Similar straw pole for 50un fiber: 850nm short wave 50un fiber 1300nm 50un 400/400mhz*km 500meter 550meter 400/600mhz*km 500meter 550meter 500/500mhz*km 550meter 550meter Question: should we do this - support all three cells or just keep 400/400 (which is what was decided in Seattle). All three of them are offered by at least one manufacturer. Geoff: There is no 50un fiber spec in TIA. 50un is worldwide, but not US. TIA is considering adopting 50un fiber in the US. 1. one cell for 50um yes 16 2. more then one cell for 50un yes 77 Q:does a 400/600 require the use of a patch cord for 50un 1300nm fiber? A: Petar is unsure without further simulation. So - we are now able to propose more then one cell. Straw Pole: Add note in option 4 to option 3 62.5um fiber * Note : it is expected that 99% of the installed 160/500mhz*km 62.5um MMF meets this specification and actually goes 250meters. Yes 56 No 28 Wouldn't have passed as a motion, but expresses opinion. 802.3z pmd discussion ended. 2:10pm Tuesday Start Comment Resolution: 17 editorial comments in the document. Everyone should have looked over the editorial comments - yes Motion: Editorial comments accepted as proposed by Del Vince Melendy 1st Dean / amp 2nd Motion Passed Receiver conformance testing Four aspects of this Document on web site were the documents are: #52 38.6.11 propose new sub clause Can we add new sections? Howie: give them temporary section numbers at the end of the main section. Del: Anyone hear have any concerns regarding text? Mike: * vertical eye opening * signal used for testing "shall be the test pattern described in 36A.5 Receiver stress test pattern" * Table should have exact numbers * Heading in table should read "vertical eye closure penalty * Bottom line - 2.5ghz photo detector .... ????? Point is they may be incompatible .....???? Note inserted in first paragraph :Penalty shall be within the range. Bandwidth of test system vs filter [2.5ghz] - to close to 1875mhz filter {comment with-drawn } Page two of receiver testing Add optical attenuator Coaxial cable shown different Remove proposed 38.6.12 proposed new sub clause Upper cutoff frequency Ed: will edit and place in the document Test setup for receiver UCF measurement Do we need to define data pattern : should get the same results regardless??? But use the same one as defined in 36A.5 Del Accepts #54 38.11 line 6 add the sentence "For multi mode fiber measurements, the polarization rotator in ansi X3.230.1994 FC-PH should be omitted and the single mode fiber should be replaced with a multi mode fiber. Accept #55 reject #56 Reject - no need to define a separate annex. Handled as a note under a table of text. #117 accept - conformance test response will workng here and is easiest thing to do. Completed all comments except the bandwidth ones. Resume in solon B at 8pm Tomorrow 8:30am to 3pm Steve Swanson: wants to skip 8pm tonight. Howie: if we have a consensus, then not meet tonight. Del: Bandwidth cells issues Must be done at 5:45 tonight Bandwidth Cell issue. Howard : Motion Adopt option 3 for SX from our 802.3 meeting this am. David Cunning 1st Jim Tatun 2nd John Abbott: two things - 160/200 okay, the other is the lengths. 275meter BW may be optimistic and 220 meter BW may be pessimistic? David: This really doesn't say anything about the BW cell for 200mhz ????? Paul Kolesar: Will impact other tables. Bob Mayer: other numbers should relax more ????? Question called Technical vote yes 38 no 0 abstain 2 Passes Steve Swanson: Would like to see two cells in the table for 50un fiber: 400/600 and 500/500. 15% of 62.5 does not meet 160/500mhz*km. 15% of the 50un fiber base does not meet 400/600mhz*km. Source of data is ten years of production from 1988 to 1998. How much of the embedded base is corning ?????? Motion Adopt three cells for 50un MMF SW dist LW dist 400/400 500 550 400/600 500 550 500/500 550 550 Jonathan Thatcher 1st Vince Melendy 2nd Friendly amendment by Geoff thompson that the middle cell could be removed because it is already covered. Flatman 2nd Jonathan doesn't like it and wants discussion before he calls it friendly. 50um fiber - one country where 50un is used extensively with 400/400 worst case of base. Then see 500/500 as the top. Two levels of performance. John Abbott: 400/600 could run longer then 550meters and the index profile to make is much better then 400/400. Steve Swanson: Del Hansen: speaks in favor of two cells Chris Diminico: supports amendment and would go further to remove 500/500 ... Bob Mayer: agrees 400/600 cell. Discussions heard we could get rid of the patch cord at 400/600. Doesn't think this is true based on information from the field. Jonathan agrees amendment is friendly but we have to vote on it now because Howard Frazier said so. Steve Swanson: Distance possible with 400/600 cell could have distances with much longer lengths - up to twice the lengths. Allen: This approach is consistant that first line is consistant, the second line is consistant for tomorrow???? Steve Swanson: Differs because ..... Vote on call the question yes 37 no 7 abstain Question called On amendment yes 28 no 8 abstain 11 Amendment passes Discussion motion: Paul: 50un fiber install base 200/600 range in Japan. Technical yes 40 no 1 abstain 9 Passed One final question still open - note on the 62.5 Motion No "new" note for table 38-12 regarding extended lengths or OFL bandwidth Chris Diminico 1st Allen 2nd Paul Kolesar added friendly amendment to add "new". Okay by movers Bob Mayer: ???? Howie Johnson: would rule out of order any motion to put this back in ????? Bob Mayer: effective bandwidth out in the field has an effective BW of 185 or higher, and the numbers in the table shorten the lengths unnecessarily Geoff : John Abbott: Mike: supports note Paul Kolesar: back to discussion if fails Chris Diminico: ???? David Cunningham: 220 and 275 meters are solid based on data. Does not support note. Bob Mayer: Should have hard fast numbers that everyone should support .... But would vote no. Jeff: Certain portion of the market we can address with these numbers Del: Difficulty getting this through the overall group. Bill: If you include a note, then you open the door for many notes. Howard Frazier: agree with Bill Yes 33 no 13 abstain 8 Motion does not pass Resume at 8pm 8:05pm Start Summarize: Specification cases 62.5un mmf (sw/lw) 160/500 220/550 200/500 275/550 50un mmf (sw/lw) 400/400 500/550 500/500 550/550 Del's position: recommend we discuss note issue because he doesn't think there is anything we can do about it. Bob Mayer: wants to have a common motion if a motion will be made tomorrow. If we have here something that we all like ... we can present it tomorrow .... And really wants to resolve comments. Chris: wants to deal with the comments. Bob: Wants constructive input on his motion for tomorrow. Bruce: Wants constructive input on his motion for tomorrow. RX Stress Penalty Calculations Minimum Average Power Penalty Proposed table Added "unallocated margin" column Issue with 50un 400mhz*km 500meter distance shows -.48 unallocated margin. Dave gets 480meters to wash the margin. Proposal: use 65ps for all the calculations for ISI Penalty column. Motion The "DCD extended" link model used to evaluate performance will use a DCD=65ps at both TP3 and TP4. The link lengths and Bandwidths are the values shown on the attached table (as above ). Any unallocated margin is shown for each case [delete with friendly amendment: "rather then extending link lengths"]. The receiver stress test will be at 65ps DCD and the calculated stress ISI "is the worst case for SX and LX among four cases". Jim Tatum 1st David Cunningham 2nd Both need to agree on items above in "". Yes they do We aren't changing the spec, but changing the margins by changing ISI Penalty The stress test is really the same, not the requirement. Unallocated margins using 65ps rather then 80ps: Freq Unalloc margin 62.5un 200 .39 160 .61 400 0 (-.26) 50un 500 1.83 400 0 (.27) Note with uncertainty on unallocated margin and use the numbers as they came out. Yes 15 No 1 abstain 2 Passes Technical 11 March 98 Wednesday 8:30am Howard ask one resolution to state the status of the PMD group. RX Stress Penalty Calculations Minimum Average Power Penalty Fiber BW dist ISIp Stress ISI Stress RX sens cor unallocated SW 62.5 200 275 3.4 2.57 -12.53 .45 .39 62.5 160 220 3.39 2.56 -12.33 .45 .61 50 400 500 2.98 2.15 -13.37 .47 -.24 50 500 550 2.35 1.50 -13.85 .45 .27 LW 62.5 500 550 2.32 1.51 -13.85 .47 1.83 50 500 550 2.45 1.66 -13.56 .48 .04 50 400 550 3.38 2.55 -13.85 .47 .27 Paul: Either more power or less distance, but does not want a negative unallocated margin in the working table. Del: Would think we should drop this distance by what ever the numbers. Mike: Chris: this looks like an accounting tast. Feels all the distances are based on research. Jonathan: happy to adopt a -9.5db and raise the power budget by .5db. Motion Increase the 1000Base-SX minimum optical power from -10dbm to -9.5dbm. "and DCD = 80ps for link analysis. And DCD = 65ps for ISI Penalty" Jonathan Thatcher 1st Ed Grivina 2nd Jonathan: Accept friendly amendment defined in "" Ed: accept friendly amendment Del: Numbers give more stress for conformance and prefers 80ps Bob: Might be worth while to look into mods Jonathan made to the model and make sure the mods were made for real theory or ???? 1. Worth while as a team to look at the model and adjust numbers and 2. Still have random jitter in fiber that gets added and looking at those aspects of the model may give us the link lengths we have. Allen: Channel now confirmed to 540 meters ?????? Ed: clarify DCD allowances ???? Chris: fiber cell was never recommended anyway. Jeff: To reduce this distance may damage industry perception of us. Perception is gigabit not stable and thus not viable technology. Means all of us don't sell as many components into the gig space. John: ?? David: DCD gives shorter pulse - eye center comes down. Thinks Jonathan's table is accurate based on his independent calculations. Dan: Add .5db to link budget, what happens to the other cases? Simply increases unallocated link margin. Doesn't know of a conversation about this. Jonathan DCD 200ps/width of bit????? RJ on fiber - called modal noise. Not much we can do about that. Have modal penalties on the amplitude gain ..... in terms of distance, thinks we are fine .... Jitter ... eye saftey, the upper bound changed and not the lower. This could be hard on some TX suppliers. Del: Comment resolution phase. Need to make decision. Joel: okay with .5db change hp=live with, amp=pain, ceilo=pain, molex=yes, methode=yes Mark: does not agree with jonathan increasing the distance Technical: yes 18 no 3 abstain 11 Motions passed Honeywell voted no to motion - support of adding .5db to link budget, but would rather see the receiver sensitivity change and not the transmitter - feels less painful on that end. Reconciling Negative Budget for 50un 400mhz-km 500meter Straw poll A. For obtaining positive margins 1. Vote1=16 vote2=18 vote3=19 Reduce SX length for 50un 400mhz*km to 470meters B. Increase SX power budget .5db 2. Vote 1=6 Raise SX transmitter output min by .5db. 3. Vote 1=11 vote2=6 vote3=0 Increase RX sensitivity by .5db. 4. Vote 1=11 vote2=4 Raise SX transmitter power by .25db and increase SX receiver sensitivity by .25db 5. Vote 1=7 Delete 400/400 6. Vote 1=9 vote2=1 vote3=13 Delete -17.5db (informative) test + increase RX sensitivity by .5db Dave Silj: Thinks the last is the way to go. Motion Adopt option 1 ( a link length of 470meters for 400/400 fiber with 1000base-SX ). " With D4.1 power budget (7db) and D4.1 TXmin power -10db and DCD=80ps at TP4 for link analysis with modified model." Davis Cunningham 1st Chris Diminico 2nd His point is we need to get beyond the 500meters anyway to meet international cabling standards. Is also worried about xcvr companies if they can make this. Howie: are there any other numbers disputed?? Jonathan: friendly amendment to reverse last motion "" Chris and David agree Jonathan: would vote no because he hopes other motion to do the other. Del: supports motion and favors as a way to move forward. Jeff: making changes that will decrease demand for our products. Technical: yes 19 no 9 abstain 3 Motion fails [Strike words - Make the min RX sensitivity informative for both SX and LX and] Change the SX "RX Sensitivity" value to -17.5dbm and change min SX optical launch to -10dbm with an SX link budget of 7.5db. Use 80ps DCD at TP4 for Link analysis. Jonathan Thatcher 1st Jim Tatum 2nd This undoes a previous motion. Del: Friendly amendment to strike words - accepted by Jonathan and Jim [strike] David: add words "" accepted by Jim and Jonathan Jonathan: Effect of this motion is to add .5db to the receiver. Vince: Jonathan: This motion moves the .5db from the TX we just added it to back to the RX sensitivity. Jim: David: If this fails then we have a large consensus we can solve this. Take the pain in the TX or the RX Technical yes 8 no 13 abstain 8 Motion Fails There fore we have agreed to add .5db back into the transmitter to solve the 50um SX 400mhz 500meter distance issue. Del: 1. Jonathan needs to recalculate the new sheet 2. Comments to resolve * The rest of the group Pick ISI stress number and Stress Power: Test SX 62.5un 2.6 -12.0 50un 2.2 -13.0 Modify channel insertion loss table Wavelength at the wavelength indicated, not the worst case wavelength 850 1300 850 1300 SMF 220 275 550 500 550 550 5000 2.33 2.53 2.32 3.25 3.42 2.32 4.5 So, we have 9 comments unresolved at this point. Taken one of the 9 comments and put fields in #115 Howie moves to let Del modify the numbers in the comments response per all the work we have just done Jonathan 2nd Motion passed - unanimous Recommend that 802.3 adopt the resolution of all clause 38 comments per our meetings this week. Howie Johnson 1st Ed Grivina 2nd One issue by Mike that may cause problems Yes 21 no 0 abstain 0 All comments have been resolved. Table in 34-1 Change needs to be made to it. LX now meets 550meters and becomes normative. Adopt resolution to adopt this change into clause 34. Passed by all Letter to Scott / Molex - answer yes to his letter. We authorize Jonathan Thatcher to author a paragraph on test patterns for the standard. Howie Johnson 1st Del Hansen 2nd Motion passes PMD Adjourned