"IEEE 802.3 1.25 Gbd MMF Link Specification Development Issues", Del Hanson, HP, was presented at the IEEE 802.3 Interim Meeting in Wakefield, MA on 05/21-22/96. This presentation contained several performance analysis plots which are summarized here in text form. Slide titles are listed along with a text summary. Title: 1.063 Gbd Optical Power Specifications For 770-860 nm Lasers This summarizes the 1.063 Gbd Fibre Channel (FC) optical power budget. My comments related to reducing the FC receiver overdrive limit from 0 dBm to match the transmitter specification. The 0 dBm limit was introduced to anticipate a possible 4 dB upward shift in eye safety limits. Since the transmitter specification will need to change if the eye safety limits are relaxed, the receiver limit can be changed at the same time. Lowering the dynamic range results in 1-2 dB increase in sensitivity margin advantage for the -17 dBm receiver limit. Title: Reconciling MMF Time-Based Response Time & Pk-to-Pk Jitter With Templates Historically, MMF transmitter interface specifications have been based on worst case response times and time-based jitter. Gb/s interface specifications can be simplified by replacing this with a transmitter eye template and frequency dependent receiver jitter tolerance limits. Title: Time-based Jitter Specification Methodology & Limits A plot was shown to reconcile jitter response time limits to the signal Unit Interval (UI). Title: Fibre Channel (FC) 1.063 Gbd Serial Link Jitter Study Group Worst case time-based jitter is specified more tightly than necessary for multivendor interworking. A Jitter Study Group was formed to refine specifications based on templates and frequency-dependent jitter. This work should be leveraged for Gb/s Ethernet standard development. Title: Proposed FC Jitter Tolerance Test System and Eye Scan A proposed test system introduces data-dependent, random and sinusoidal frequency-dependent jitter to measure the receiver jitter tolerance 30% window opening at BER = 10^-12. Data shows the measured eye opening. Title: 62MMF Data Rate vs. Source Response Time & Modal Bandwidth. A plot shows the projected response time limits vs. source 10-90 % response time of 0.45 ns and 0.35 ns, and modal bandwidth of 160 and 200 Mhz*km. It shows the worst case link length ranging from 250 m to 340 m at 1.25 Gbd, depending on the parameter choice. Title: 50MMF Data Rate vs. Source Response Time & Modal Bandwidth. A plot shows the projected response time limits vs. source 10-90 % response time of 0.45 ns and 0.35 ns, and modal bandwidth of 400 and 500 Mhz*km. It shows the worst case link length ranging from 435 m to 590 m at 1.25 Gbd, depending on the parameter choice. Title: Impact of Spectral Width (Uw) on Maximum Data Rate for 62MMF. These plots for sources having FWHM Uw = 1, 7 and 9 nm show a worst case link length range from 275 to 235 m at 1.25 Gbd. This demonstrates the necessity of including the chromatic dispersion in the link specification. Title: Transmitter Template With Jitter & Without/With 0.75*Rate Filter Two plots are presented for 0.35 ns response time and related jitter and templates are defined by FC at 1.063 Gbd and SONET at 622 Mb/s. Analysis shown in plots show that 0.3 UI Pk-to-Pk jitter can be specified in the first case and 0.4 UI Pk-to-Pk jitter in the second case. It is critical to define the correct balance between jitter and response times at the transmitter and receiver interfaces. Title: Gb/s Ethenet Physical Link Specifications Conclusions. It is preferred that a transmitter template be specified rather than response times and Pk-to-Pk jitter since this eases measurement and allows trade-offs between response times and jitter. The FC Jitter Study Group should be leveraged in this development effort. Considerable effort is needed to reconcile and balance the specification limits with the lowest cost implementation. It is also necessary to include the chromatic dispersion in defining link length limits.