"Review Of 1300 nm and 850 nm Optical PMD Specifications For IEEE 802.3z Gb/s Ethernet", 

Del Hanson, Hewlett Packard Company, Presented at 11/96 Plenary

Slide #1: Outline

* Review of certain 1300 nm Transceiver Specifications: SMF & MMF
* Review Of 770-860 nm Source Spectral Width
* Link Serial Jitter Specification Issues
* Summary Of Recommendations

Slide #2: Joint 1300 nm SMF & MMF Specifications Relative To 9/96 Report

* Transmitter Eye Mask
  + To maintain consistency between jitter specifications in the two wavelength proposals, a Fibre Channel (FC) mask will be used for both
* Eye Safety Limits
  + HP's implementation is unconditionally Class 1 CDRH & IEC 825-1 eye safe for both SMF and MMF
* Transmitter Power Range: -3 dBm to -13 dBm
  + Parameter allocation shows that both SMF and MMF transmitters can be supported in this range

Slide #3 & #4: Laser Eye Safety Limits Summaries

Slide #5: "Cross Section Of HP's "ELSA", Eye-safe Laser Sub-Assembly"

Slide #6: "Eye-Safety Characterization For IEC 825-1"

  + Shows that ELSA unconditionally meets eye safety limits

Slide #7: "Reconciling Proposed 1300 nm SMF and MMF Gb/s Ethernet Transmitter Power Level"
 
  + Starting with a nominal -9.3 dBm launched into SMF
  + Adding +/- 0.2 dB for power meter variability
  + Adding +/- 0.5 dB for setup repeatability
  + Adding +/- 0.5 dB for final test repeatability
  + Adding +0.8/-1.7 dB for laser assembly tracking
  + Adding +3.5 dB for maximum MMF coupling relative to SMF
  + Adding +/- 0.8 dB for aging over life
Results in matching the -3 dBm to -13 dBm specification range

Slide #8: "1300 nm SMF Source Spectral Width Relative To The 9/96 Report"

  + The SMF link length specification was extended from 2 km to 3 km at the 9/96 Interim meeting
  + To accommodate this increase in link length with the same spectral inter-symbol interference penalty, the spectral width is reduced from 6 nm to 4 nm.

Slide #9: "1300 nm MMF Link Length Relative to the 9/96 Interim Report"

  + The 7/96 Plenary motion stated that the link length be specified for over-filled launch (OFL) modal bandwidth, as with LED links, until the work of the TIA restricted mode launch (RML) committee completes its work.
  + The OFL modal bandwidth of both 50MMF and 62MMF is 500 Mhz*km at 1300 nm.
  + The maximum link length in my model is defined by the fiber exit response time being equal to the signal baud period.
  + My 1300 nm model for projecting exit response time has been checked in an industry test bed for 62MMF for the 155 Mb/s ATM Forum specification. Link analysis models were within 10% in projecting exit response times.

Slide #10: "62MMF Maximum Link Length For minimum wavelength of 1270 nm and rms spectral width of 4 nm"

  + An analysis plot shows a worst case link length of 870 m with 0.45 ns 10-90% response time and 930 m with 0.35 ns 10-90% response time. Increasing the modal bandwidth to 1100-1200 Mhz*km is required to support a 2 km 62MMF link length with these exit response time parameters.

Slide 11: "770-860 nm MMF Link Spectral Width Relative To the 9/96 Interim Report"

  + FC specifies 4 nm rms spectral width to accommodate CD LDs
  + The 10/23/96 GEA meeting suggested moving the spectral width to 0.85 nm rms to try to support 300 m with 62MMF
  + HP meets the 0.85 nm rms proposed spectral width with first generation VCSELs, but...
  + To accommodate next generation lower threshold VCSELs, the spectral width should only be reduced to 2 nm rms rather than 0.85 nm rms.

Slides #12 to #17: Maximum link length limits are shown for 62MMF and 50MMF with a range of LD response time & spectral parameters

  + 62MMF link lengths with modal bandwidth of 160 Mhz*km range from 230 m (4 nm rms spectral width and 0.45 ns 10-90% response time) to 300 m (0.85 nm rms spectral width and 0.35 ns 10-90% response time)
  + 50MMF link lengths with modal bandwidth of 500 Mhz*km range from 365 m (4 nm rms spectral width and 0.45 ns 10-90% response time) to 900 m (0.85 nm rms spectral width and 0.35 ns 10-90% response time)

Slide #18: "Remaining Optical Link Jitter Specification Tasks"

* We have agreed to:
  + Use the FC Transmitter Eye mask to trade off transmitter response time and jitter parameters
  + Adopt & incorporate the FC Jitter Study Group frequency dependent jitter methodology by 5/97

* Work Yet To Be Done:
  + Define SerDes transmitter serial output jitter
  + Define SerDes receiver window opening at BER
  + Target component response times & allocate jitter to achieve conformance with low cost implementation

Slide #19: "FC Transmitter Template With Tr = 0.35 ns and 0.30 UI pk-pk Jitter With & Without Filter"

Slide #20: "FC Transmitter Template With Tr = 0.45 ns and 0.24 UI pk-pk Jitter With & Without Filter"

  + Slides #19 and #20 show the trade-offs which exist between response times and jitter in meeting the FC transmitter eye template. This is part of the analysis needed to reconcile the over-all link response time and jitter limits.

Slide # 21: "Summary of Optical Link Specification Recommendations"

  + Use the FC transmitter eye template for all cases
  + Reduce the 1300 nm LD spectral width from 6 nm rms to 4 nm rms to accommodate the SMF link length extension from 2 km to 3 km, as decided at the 9/96 Interim meeting.
  + Reduce the spectral width of the 770-860 nm LD source at 2 nm rms to accommodate future lower threshold VCSELs
  + Use 0.45 ns 10-90% response time in projecting link lengths in the informative Annex
  + Define link serial jitter vs. response time allocations following the FC Jitter Study Group methodology