PMD meeting 11:30am 2Feb98 Monday Haluk Aytac Sharon Hakimi Vince Melendy Mark Sankey Harvey Blumsack Mike Hoickert Jack Merrow Steve Swanson Kirk Bovill Todd Hudson Larry Miller Jim Tatum Dan Brown Dave Hyer Bob Musk Pat Thaler David Cunningham Vic Kairis Mark Nowell Bruce Tolley Chris Diminico Paul Kolesar Satoshi Obaka Luis Torres Lois Fredirkson Bill Koypers Paul Pace Schelto VanDoorn Claude Godcharles Arlen Martin Bhavesh Patel Mike Witkowski Joel Goergen Jeff Martin John Ritger Ed Grivna Bob Mayer Dave Rolanl Del Hansen Chair Joel Goergen Secretary 1. Discuss comments 2. Take 1.5 hours to discuss data at hand 3. Take names of who wants to present something Propose we take time to organize this, then break for an hour lunch David: presentation order MDI update session Petar - 10 minutes Interpretation of jitter Mark - 2 minutes jitter update Chris- few minutes Lewis - 5 minutes fiber field tests Jonathan - till he is cut off Todd - slides on fiber bandwidth Ian - ?? About seventy minutes 1. break for hour and come back at 1pm 2. start now Agree to start promptly at 1pm Break Start 1pm Petar: Interpretation of jitter measurements Why is Linear jitter scaling wrong * methodology for deterministic jitter * why is isi penalty and deterministic jitter scaling with length wrong? Linear scaling is wrong based on graph * deterministic jitter affected by filter order * effect of signal shape on DJ measurements using BERT * conclusions and recommendations Linear scaling pessimistic jitter may become non-issue receiver design very important bandwidths above 800mhz may be okay perform all measurements on actual length David: I have done similar simulations to Petar and agree with the results Geoff: Understand linear is always worst case. Any number we pick will always be less then that. Petar: Then why should we measure the numbers at 300 plus meters when we can do it at actual lengths Jonathan: .... Mark: Jitter update Results: confirmed by Digital's measurements No correlation between Jitter Vs CPR As we increase the power levels, the jitter is reduced. Chris: Correlate Mark's data Same results, no correlation between CPR and jitter. Lew / HP Fiber filed tests 15minutes * Acknowledgments * Measurement goals * Functional block diagram * Optical pulse source * Optical source characteristics - short wavelength * Optical source characteristics - LX * Radialoverfilled launch * Rofl spacer mechanism * Rofl spacer disk fabrication and calibration * Rofl spacer disk accuracy * Overfilled launch * Transmitter unit front panel * Transmitter - inside * Transmitter unit - typical configuration Patch cords had sm rated connectors * receiver unit - inside * complete instrument carts * system impulse response * typical fiber measurements * over view of fibers tested * histogram of fiber bandwidth * Rofl and ofl bandwidths 625m5rofl Vs ocl bandwidth * rofl Vs OFL * rofl vs ofl * some interesting results - 62.5mm, LX Steve: Question, ?? * some interesting results - 50 mmf SX Jonathan : IBM system test How to do eye measurements across the boundary. 85 pages of data Petar will create pdf file for Jonathan and put it on the web Just about every case were fibers that were know to have problems We do know what length these fibers are. All the data can be calibrated. * worst combination - vxcel laser source with digital transceiver Bob: any testing done on transmitters Jonathan - IBM was up to specs, but Jonathan was not able to get the parts to confirm them In closing - hard to repeat in terms of deterministic jitter. Howie: Can we look at transmitters to look at the distribution Todd wants to deal with his issues in the comment resolution. Ian: Bristol university Discussion just about simulations and understand predictions we can make. * simulation methods * offset launch theory * modeling offset launch * analysis of bad MMF sample * objectives * simulation results:bandwidth gain * bandwidths after scaling using DMD criteria * average offset bandwidths and bandwidth gain * dependence of CPR on offset launch bandwidth * proposed specification for single mode offset launch with 50um and 62.5um MMF for gigabit Ethernet * shortwavelegth offset launch conditions preliminary work * bandwidths after scaling using DMD criterion * bandwidths after scaling using DMD criterion Paul: Offset with addition of angle Yes Paul: angle 8 degrees Yes Petar: Question of fiber loss with offset launch In terms of theoretical work we have got plots - can show those later on if he can find ?: just LX provided CPR is achieved by effective offset, then yes, if you have weird CPR, then no. ?A jumper cable might not be needed wouldn't want to go that far. It may be true, but not what w have modeled Jonathan: On the short wave cases, what is the profile of the launch CPR is just for LX, the SX is just the offset Assuming 5,7,15um Very preliminary results on 3um Petar: single mode type of launch Did that because if we started ... wouldn't know where to start Also, ............... So on purpose we went single mode If there is significant loss, likely to have mode conditioning loss into the picture. We have got plots to the number of nodes to the offset ..... Petar: Comment on percentage of modes to be excited Early 50% for offset in most cases. John / Lucent * present data on lucent fiber testing with 850nm and 1300nm xcvrs down a fiber Geoff: what size batch fiber may have come out of? Made back in late 80s, specs were much lower. Because of that and some other things, they tightened the specs. Geoff: Not clear ......... Howie: lengths of fibers 300 to 600 meter range Once you get over a 1000mhz, it doesn't matter, but what happens over 500mhz Petar: said something about a filter. Del: Wants to move on Plan on comment resolution 67 total comments 21 of these are MBI task related We as a group need to work through the comments that are MBI related in the following sequence .... As a group , work through these, reach a decision, modify- whatever, and then reach conclusion. Motion: Chris/Dec made motion to proceed with Del's resolution plan to comment 62 as basis to moving forward. Second: David C. Work Plan ... not just answer to comment 62. Jonathan: Does this close down the option of specing a better fiber. Del: no he does not see it that way This is the over all plan to deal with receiver characteristics Jonathan: fiber not receiver Del: point is we anticipate changing the jitter budget and performance changes are built in, we have just set a path to address comments. Jonathan: No new MMF parameters - Del: we don't have to build in new parameters to know where we are. .... We are not preventing any future work is the way he intended to write. ..... MBI related work item as described. Chris: here to resolve technical comments. If we just stand, we will never resolve the comments. The technical group has proposed to solve it in a certain way. Geoff: proposal is technical .... David:technical issue, real technical problems. Even if big group wants to move faster, we may have to convince group .... And schedule will slip anyway. Better to go with objectives we are all working towards. Not an option to just reduce lengths Mike: suggestion .... Del:purpose of schedule was to give guidance. ED: to leave out schedule would be derelict. David:not completely bought into time scale - best case Tom:How long will it really take. Motion before us: this would be the proposed response to 62 at initial pass at resolving these items. 29 Yes 1 No 0 Abstain motion passed comment 239 Motion JT: assured if there will be another choice to resolution response to comment. D.H.:Yes MBI will assemble before the 3pm Weekly was eventually done Proposed response accepted. approved by acclimation Passed comment 241: Discussion on CPR with offset launch and wants to know if CPR works for no offset launch. David: stub not worded out. Passed by acclimation Comment 221: Del: Should put on the table a proposed jitter budget. The idea is that this is what is in the document today. Del wants sub groups to put definitions to the table. This table, along with response, will apply to comment 221. Passes by acclimation Paul says he accepted. Comment 16: Tom Dinen Del: Has to do with patch cord description. Seems to be disagreement on brown color . Paul: would prefer we stick with colors already defined in international standard - Blue and Beige. Does not see a reason to pull up different color. Geoff: Title change to be 1000BASE-LX with rest of title the same. Change transmitter to equipment on the label. Then vote on color. Color of boot. Blue Beige acceptable for MMF end. Change "boot" to "color identifier." Ed: Motion all of above Second: Tom David: Shelto: Jonathan: Paul: call the question Del: wants to table until tomorrow Bob Musk would get group together to address this. Table passed: Motion suspended Comment 15: Howard Frazier Table this until discussion tomorrow. Table passed Comment 102: Any objection to proposed response? None Response accepted. Mark Nowell accepts direction / not comment Comment 212: Accepts direction Comment 219: Paul will sign off on SHALL. Del: we need to discuss this because we need a test method. Paul: If not mandatory, then why have a spec. Del: just there for guidance Jim is wondering what we are doing? Del: we can change direction 1. Re-assemble group at 8:30 tomorrow 2. Assemble three groups 3. Assemble 4th editing group to resolve non-MBI issues. Del: Needs someone to work with him on remaining comments Agreed to review his comments. Meeting ended until 8:30 am tomorrow. Special Session Start 4pm David: Where we ended in Ft Lauderdale meeting Three groups 1. transmitter issues 2. Fiber issues 3. Receiver issues See handout sheet Mark leads transmitter Don Leads fiber David Leads receiver 8:30am meeting start Motion to accept editorial comments as they are and have Ed start on making the changes. Editorial only. First: Chris Second John Rutger Motion Passed by acclimation Comment 80: Commenter: Joe Gwinn Note: (d) refers to all three items in 'signal detect value' column. Any problems with this resolution.? No Paste comment from 81 in here Passed by Acclimation Comment 226: Commenter: Paul Kolesar Proposed response is: Comment is rejected Paul: This is a table entry 38-11, but the reference wavelength is 1310 not 1300. Paul/Chris: 11801 spec uses nominal wavelength of 1310. Chris: as a cable, Paul is right, as a channel, maybe not. As a channel measurement, we agreed to measure at the nominal wavelength. Paul: Motion to take Paul's resolution as stands Ed: second Discussion: Tom: Change the number, but add note that channel characteristics are measured at 1310. No No objection to accepting Paul's comment as stands Motion Passed Note: 500Mhz 50um bandwidths Vs 400mhz to be discussed tomorrow. Comment 186: Commenter Tom Mathey Proposed group change: 'For example this allocation supports 4 connections with an average ......' Any objection to this? NO Motion passed by acclimation Comment 227: Any problem to having same resolution as 226, but to table 38-12? None Motion passed by acclimation Comment 187: Accepted by Acclimation Comment 63: Accept by acclimation. Comment 179: Ed wants to leave the current standard as is; reject comment Comment rejected by acclimation Comment 109: Any problems? NO Accepted by acclimation Note: John needs to hit accept on all editorial comments - has not been doing that on the data base. Comment 212: Paul: Will accept based on MBI investigations. Therefore: NO. Comment 16: Comment was tabled yesterday, and a response was give today. * Replace brown with blue * Title in text '1000base lx patch cord' * Equipment rather then transmitter * Minimum fiber length of 10cm each side of smf and mm for 62.5 and 50 multi-mode to table 38-13 Mark: by definition, still need a ferrel in there so there is a min length already to prevent the absurd case. Therefore, can strip the min length. Chris: Have cord already speced ......... David: have to have smf to mmf cable .... But don't have to define the length. Nuke the line on length Any objected to this proposed resolution: NO Accepted by acclimation. Comment 110: At this point, we are only able to show this works for lx mmf, not for sx Accepted by acclimation Comment 219: Petar: The bandwidth is not sufficient to achieve performance and it depends how you design the receiver. If higher then 1st order receiver, then you can get away with 800mhz. When you limit the bandwidth to 800mhz, you have ... but you reduce the amount of noise by 1db. Would propose we differ discussion until we here the report from the groups from last night. Del: Doesn't want to be designing receivers, but wants to give some guidelines. Do we provide guidelines or not. Jonathan: no ...... we don't want guidelines here. Intent to guarantee inter-operability - which is part of MBI Reject comment proposal and state a conformance test will be described for receiver. Trailing TR. Comment 100: David: TR stands until after the MBI concludes. Comment 99: Same as 100: David: TR stands until after the MBI concludes. 99 was withdrawn because it was duplicate to 100. Comment 214: Data supports doing this, but most want to leave this alone until after more MBI data. David will have to recalculate the CPR values and change the bandwidth to greater then 325mhz. Paul agrees to keep as trailing TR Comment 101: Leave as trailing tr Comment 59: Bob Musk .5db for the offset patch cord Comment Accepted Comment 216: Paul Kolesar Seems we are working on 237 at the same time. Note (a) change is in 237. Deleted note c. Any problem to this resolution resolving both comments Paul accepts with no trailing ... done Pat will have to agree later Comment 87: Should be resolved in the MBI investigations Comment 218: Should be resolve in the MBI investigations Comment 237: Done with this - see 216. Comment 221: MBI related issue. A proposed jitter table was given and left up to the MBI for further discussion. Comment 222: Accepted by Paul Comment 89: Dan: We need proper reference ccitt or what ever rather then sonet. Accepted by acclimation Comment 90: Add filter input and output in 38.6.5 Accepted by acclimation Comment 92: Any problems with response .. no Response Accepted by acclimation Comment 223: Paul: let this recommendation stand and will comment on it later on the next go around. Accepted by acclimation Comment 182 Any problem with this proposed resolution: NO Accepted by acclimation Comment 55: Pat: Make sure the chair of 802.3 gets a copy right to the test methods ... Accept by acclimation We have run through all the comments, and left open comment 15 which is bandwidth issue on mmf .... Pat: case for the mode conditioning cables is in sx section ......... something should not be there. Cabling should go into cable section and reference to it in the lx section. ??????? Pat's response to 237 is for the editor to move the reference on table 38.6 / 38.3.1 to the LX section 38.4.1 under table 38-7 ????? Ed: thinks he understands. Just trying to clarify things. What ever we put in the database has to match what he puts in the document. Pat accepts the resolution. Any other comments on these issues? We are through the resolution phase. Now wants to review the work phases from the three work groups. Come up with a time frame to do these tasks. Main 802.3z meeting is at 3pm. We will break from 12-1pm, then work from 1 to 3. Jonathan wants to have the editor do the comments for comment 55 - Apparently the notes are being put into the database. Jitter discussions: -missed some of it Jonathan: This table is the direction we are going. Motion Del Hansen's Jitter table PMD group adopts the direction implied in the Del Hansen table 38-10 proposal as target values to be refined and closed at the march plenary [for clause 38]. This reverses the Paul Kolesar motion in Montreal to not change jitter budgets. [we would need feedback from SERES vendors on rise/fall dependencies] First: Jonathan Second: Del Technical - 75% required David: techniques we can use have been developed in fiber channel ..... re-enforces the message that jitter budget is broken ...... short time. Jonathan: important to do for 1 reason - having a open ended study at this portion is impossible. Put up something on the table as a straw man - numbers not far from where we will end up. Del: HP motivated to start taking these measurements. Mike: lx and sx Jitter budget applies to lx and sx. Could have anther table for smf if we want. Jonathan: Is it possible that 38 and 39 jitter budgets can be different - is it reasonable to have different budgets? Concerned about TP4 .... 38: rail to rail 39: smaller shape Ability to hit same jitter targets is unrealistic. Current optics , with relief in the rise/fall, should not be hit. Do we know how to measure these points? We will be reviewing this. Jeff: Do we really think March is realistic? Pat - danger about not picking date........ Howie: group has choice as to when they meet - meet anytime we want. Sundays are okay ...... advise to schedule more time then we think we need. Any other discussion? Jeff - drive jitter budget issues .... Test then for rx and tx .....? Jonathan: implied by tighten budget, redoing the testing is required. 35 Yes 0 No 1 Abstain Motion passed by 75% margin Note - only comment 15 left. Completed work through the comment resolution. Clause 38: Item to close: Next plenary meeting March 9-13, Monday through Friday. Friday and Sunday with nothing on sat Sunday from 8:30 to whenever - need room for 30 people David - Update: Transmit section Receiver section Fiber section Break Start David Cunningham: * Del to publish fact that the PMD group will be adding new test and specifications in March Plenary. * Request testing using exact MBI test beds. * Howie J develop questions document for system vendors. * Work by phone until March. * SerDes jitter critical issue. * Make clear optical PMD signal eye pattern. * [some jitter might be from the +/- 10% power rather then at +/- 5%.] TX: * MPD measurements * Share results to ensure repeatability * Characterize sources * Theoretical work * Need to develop a range of sources to develop MPD template 1. By March, have a range of MPD sources 2. Also have more specific theoretical input to MPD template issue. * need optical spectrum of sources 3. missed this Fiber: * what is worst case at 850nm? * Provide samples to key transceiver companies * Corning to continue model for TX * Model for 50um * Further DMD information RX: * Complete test methods and specifications draft (next week) * Vendors to do measurements and report * Four corner measurement * Define high frequency cut-off test * Eye-opening recommendation * TP3-TP4 XXX?: Some designs may fall out and have to be redesigned. PMD Finished