1. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 1 IEEE802.3aq Channel model ad hoc Task 2 and 4 Launch study Summary of progress 19 Jan 2005 Jonathan King

2. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 2 Launch study goals Specific inputs to task force on launch conditions and launch testing –in time for January meeting Activities OM1 link simulations and experiments OM2 model development prompted (in task1), and experiments OM3 link simulations and experiments –Simulation and experimental results for centre launch and range of OSL (offset single mode) launches, and a universal launch candidate (Vortex launch)

3. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 3 Findings: OM1 and OM3 With connectors - Center launch better for OM3, FDDI fiber slightly better with standard 62.5µm OSL (17µm – 23µm) No connectors - Center launch better for both OM3 and FDDI fiber Similar conclusions from simulation work (Joerg Kropp and S Botacchi) Supported by experimental data from many others (Joerg Kropp; F Sugihwo & J King; David Cunningham & Simon Meadowcroft; Yuri Vandyshev, Jim Mcvey, Hongyu Deng & Lew Aronson) Multiple launch options per fibre type improves coverage Results from Ewen_1_041215

4. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 4 Preliminary findings: OM2 OM2 model is under development in task 1 –50µm OSL (10µm - 16µm offset) proposed for primary launch, centre launch as secondary launch, based on reported experience with TIA 12-96 round robin fibres –to be confirmed with further experiments and simulations pending OM2 model Experiments comparing Vortex with CL and OSL on OM1 and OM3: –Varies less with connector offsets than CL, comparable to OSL –Compares favourably to OSL for OM1 –Worse than CL for OM3, but may be good enough – IPR / PIE Measurements on OM1 and OM3 Center, Offset and Vortex Launches (Yuri Vandyshev et al ) Simulations of Vortex launch show higher PIE-D value than per fibre type optimized launch, but has advantage of single patchcord interconnect at transmitter Jim Morris et al –Vortex launch would meet OM1 and OM2 primary launch EF definitions Preliminary findings: 'Universal' launch - Vortex launch

5. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 5 Experiments comparing Vortex with CL and OSL on OM1 and OM3: ('IPR / PIE Measurements on OM1 and OM3 Center, Offset and Vortex Launches', Yuri Vandyshev et al ) –Varies less with connection offset values than CL, comparable to OSL –Compares favourably to OSL for OM1 (PIE-D is 0.5-1dB lower than OSL) –Typically worse than CL for OM3 (PIE-D value up to 1.5dB higher than CL) but may be good enough Simulations of Vortex launch on OM1 and OM3 ('Vortex PIE calculations', Jim Morris et al, using 54YY fibre model and 850nm TIA OM3 delay set converted to 1300nm by P. Pepeljugoski) –Higher PIE-D value than per fibre type optimized launch, but has advantage of single patchcord interconnect at transmitter On OM3, with connector offsets, the 99% coverage PIE-D values were 4.8, 6.7 and 6.1 respectively for CL, OSL and Vortex lens with M=4. On OM1, the 99% coverage PIE-D values were 5.9, 5.8 and 6.1dB respectively for CL, OSL and Vortex lens with M=4. –Vortex launch would meet OM1 and OM2 primary launch EF definitions Preliminary findings: 'Universal' launch - Vortex launch

6. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 6 Summary Launch recommendations table prepared Primary and secondary launch recommendations for each fibre type –OM1: OSL, CL –OM2: OSL, CL –OM3: CL, OSL Encircled flux launch test definitions recommended for primary, secondary launches, and for 'universal' launches (implementation non-specific) Encircled flux launch test definition for a 'universal' launch Other points of note –Multiple launch options can increase coverage need to understand how/if to specify dual launches in standard - seeking end customer input –Confirmation that average mode power simulations give worst case 99% coverage PIE-D results (i.e. we don't need to explicitly model IPR variation due to dynamic effects, because the 'new' IPRs are already represented in the static IPR set)

7. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 7 Recommendations for changes to Launch section of Table 68-3-10GBASE-LRM transmit characteristics Fibre type OFL bandwidth Primary Launch TP2 encircled flux test criteria Alternative Launch TP2 encircled flux test criteria 'Universal launch' TP2 encircled flux test criteria 1OM1500/500< 30 % in 5 µm radius > 86 % in 23 µm radius note 1 > 30 % in 5 µm radius > 80 % in 10 µm radius note 3 < 30 % in 6 µm radius > 86 % in 18 µm radius note 4 2OM2400/400 500/500 < 30 % in 6 µm radius > 86 % in 18 µm radius note 2 > 30 % in 5 µm radius > 80 % in 10 µm radius note 3 < 30 % in 6 µm radius > 86 % in 18 µm radius note 4 3OM31500/500> 30 % in 5 µm radius > 80 % in 10 µm radius note 3 < 30 % in 6 µm radius > 86 % in 18 µm radius note 2 < 30 % in 6 µm radius > 86 % in 18 µm radius note 4 note 1: For example, 20 µm offset single-mode fiber offset-launch mode-conditioning patch cord, as defined in 38.11.4 note 2: For example, 13 µm offset single-mode fiber offset-launch mode-conditioning patch cord, as defined in 38.11.4 note 3: For example, single-mode centre launch note 4: For example, Vortex launch

8. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 8 Back up

9. 19 Jan 2005IEEE802.3aq Channel model ad hoc Task 2 & 4: Launch Study 9 Other valuable stuff - 1 Do dynamic effects introduce new impulse responses which need to be included in link simulations ? - No –Fibre models generate a set of impulse responses from static average modal power; Simulations indicate that the additional impulse responses due to polarization / dynamic effects are already represented in the static impulse response set –Average mode power simulations and individual mode power simulations (where input polarization was rotated) were compared: at 99% coverage the PIE-D results for the average mode power computation were the same or worse than individual mode power simulations. (Yu Sun 'PIE-D statistics comparison between averaged mode and individual mode computation method') Confirmation of the 7 micron value for a worst case connection offset –Extensive measurements of tolerances contributing to connection offsets were reported (Al Brunsting & Rick Pimpinella ' Lateral offsets for multimode fiber (MMF) connectors'