Improvements to Modal Noise Penalty Calculations Petar Pepeljugoski, Daniel Kuchta and Aleksandar Risteski IBM T.J. Watson Research Center Yorktown Heights, NY 1598
Outline Modal Noise (MN) penalty calculation changes Calculation of MN penalty for various launch classes Comparison of Overfilled launch (OFL) with Single Mode Fiber (SMF) Launch Comparison of OFL with Multimode Fiber (MMF) Launch Conclusions/Recommendations
Why do we need changes to the theory? Theory developed in the mid 8s to 9s Fundamental assumptions are not completely valid for today s links Original calculations assumed overfilled launch conditions not valid for center, offset or restricted launch Fiber transfer function assumed to be Gaussian Noise bandwidth and receiver filtering needs to be taken into account Modal noise bandwidth is at least equal to the signal bandwidth, may be much higher (or much lower)
Prior work: references for MN penalty calculations 1. R. J. S. Bates, D. M. Kuchta and K. Jackson Improved multimode fibre link BER calculations due to modal noise and non-self-pulsating lasers (Optical and Quantum Electronics, vol. 27, (1995) pp. 23-224) - Contains complete set of equations to evaluate the MN penalty Low frequency (Gaussian) and high frequency (Double Exponential) noise probability density functions - Basis for the work by the Gigabit Ethernet MN group and our simulations 2. T. Kanada Evaluation of Modal Noise in Multimode Fiber Optic Systems (IEEE JLT, vol. 2, 1984, pp. 11-18). - Modal Noise Bandwidth Calculated 3. K. Peterman Non-linear Distortions and Noise in Optical Communication Systems due to Fiber Connectors (IEEE JQE, vol.16, July 198 (pp. 761-77) - Take into account actual MPDs, connector effects, improve variance calculations 4. R. Dandliker How Modal Noise in Multimode Fibers Depends on Source Spectrum and Fiber Dispersion (IEEE JLT, vol. 3, 1985 (pp.7-12) - Speckle contrast at output of long fibers
Changes to Standard Approach Calculation of the low frequency modal noise std deviation in [1], eq. 4 modified not to use approximations Calculations of the high frequency modal noise std. dev. also affected Take into account actual fiber TF observed impact of launch conditions and fiber delay times on modal noise penalty Take into account the modal noise filtering in the receiver (if the bandwidth larger than the receiver bandwidth) First calculations without correction uncertainty in parameters, results will be worst case
Link Configuration Uses Worst Case MN Link agreed by the group 2 connectors with 7 µm offset separated by 1m fiber, one with 4 µm offset after 22m High Coherence Laser Diode with α=.98 and β=.47 (Fabry-Perot) Fiber bandwidth 5 MHz km First set of slides without correction for receiver bandwidth, assumes Gaussian fiber TF Second set takes into account actual fiber TF, and uses MMF launch Mode Partition Noise factor k is parameter
Comparison of MN Penalties for OFL Launch and SMF Offset Launch OFL conservative for smaller offsets, optimistic for larger offsets Penalties much smaller for 62.5 µm MMF, particularly for large offsets 4.5 4 3.5 k=.1 k=.25 k=.75 k=1. 2 1.8 1.6 k=.1 k=.25 k=.75 k=1. MN penalty [db] 3 2.5 2 1.5 OL Range OFL Launch Increasing values of k SM Launch MN penalty [db] 1.4 1.2 1.8.6 OFL Launch SM Launch Increasing values of k 1.4.5.2 5 1 15 2 25 Lateral Offset [µm] 5 µm MMF 62.5 µm MMF 5 1 15 2 25 3 Lateral Offset [µm] OL Range
Comparison of OFL Launch and MMF Launches for OM3 fiber Laser launch impacts the modal noise penalty Higher encircled flux = higher modal noise penalty But we need to evaluate the impact of the MMF launch on the equalizer performance Need to control encircled flux, steep rise in penalties for R(EF=86%)>18µm Similar results expected for OM1 and OM2 fiber 1.4 1.4 MN penalty [db] 1.2 1.8.6.4 k=.75 OM3 fiber Worst Case MN link OFL Launch MN penalty [db] 1.2 1.8.6.4 k=.75 OM3 fiber Worst case MN Link.2.2 2 4 6 8 1 12 14 16 18 2 Multimode Launch # 8 1 12 14 16 18 2 22 R(EF=86%) [µm]
Penalty vs. Total MSL Relationship between EF and MSL not linear Penalty curves follow expected trend, but at lower level Limiting encircled flux to 18 µm would guarantee low modal noise penalty Total Connector Loss [db] -.5-1 -1.5-2 -2.5-3 MN penalty [db] 2 1.8 1.6 1.4 1.2 1.8.6.4.2 BL - k=.1 GR - k=.25 RED - k=.75 CY - k=1 OFL Launch -3.5 8 1 12 14 16 18 2 22 R(EF=86%) [µm].5 1 1.5 2 2.5 3 3.5 Total MSL [db]
Conclusion Improved MN penalty calculations Takes into account launch conditions, fiber properties SMF launch (both offset and center launch) acceptable - fall in low MN penalty region MMF launch may be also acceptable if encircled flux is limited to 18 µm Need to check the impact of launch conditions on yield and residual ISI Compare center launch with offset launch Find the yield for MMF Measurements underway!