10GBASE-SR Link Fails Loss Budget on a 165 m OM4 Channel
A 10GBASE-SR link connects two access switches in the same building: 165 m of OM4, two LC patch panels, and one fusion splice per fiber at a mid-building splice closure, so each channel has four LC connectors and two splices. The design states a 2.6 dB channel budget for 10GBASE-SR. The contractor's OLTS test at 850/1300 nm reads 2.9 dB at 850 nm on pair 3 and 2.4 dB on pair 1. The OTDR trace on pair 3 shows a 0.6 dB event at the first panel, a 0.1 dB event at the splice, and a negative loss of -0.3 dB at the second panel. The contractor says the OTDR proves a bad panel and wants to replace both panels. The customer asks whether the link will support 10GBASE-SR and whether the OTDR evidence is reliable.
1. Calculate a component-based loss estimate for the channel and determine whether 2.9 dB passes the 2.6 dB budget. What should be checked before condemning the cable or panels?
Read the worked answer
Using common premises allowances: fiber 0.165 km × 3.5 dB/km = 0.58 dB; four LC connectors × 0.75 dB = 3.00 dB; two fusion splices × 0.3 dB = 0.60 dB; estimated total = 4.18 dB. That estimate is above the 2.6 dB channel budget, which shows the design is marginal even before measurement. The measured 2.9 dB is lower than the allowance-based estimate but still exceeds 2.6 dB, so pair 3 fails the stated budget by 0.3 dB. Pair 1 at 2.4 dB passes, but with only 0.2 dB margin. Before replacing panels, verify the test setup: inspect and clean all endfaces; confirm the OLTS reference method and that both reference cords are included or excluded consistently; check launch conditions and mandrel wrap; verify the meter is set to the correct wavelength and that the patch cords are not damaged. Re-test pair 3 after cleaning. If it remains above 2.6 dB, the channel is not compliant with the design, regardless of whether it happens to link. A passing link-light is not an acceptance test.
2. Evaluate the OTDR evidence, including the negative loss at the second panel. Is replacing both panels justified?
Read the worked answer
No. The OTDR trace is not reliable enough, as presented, to justify replacing both panels. A negative loss at the second panel is a classic OTDR artifact: a gainer can appear where backscatter changes, where a low-loss event is followed by a higher-backscatter fiber, or where the launch and receive conditions are wrong. It does not mean the panel has negative loss. The 0.6 dB event at the first panel is high for a mated LC pair, but it may include a dirty endface, a damaged adapter, a patch cord, or the panel's connector pair, and the OTDR's dead zone on a 165 m link can smear event locations. Require bidirectional OTDR testing with a launch fiber and a receive fiber, plus an OLTS measurement as the acceptance result. Inspect the first panel's endfaces with a scope, clean and re-test, and swap the patch cord before condemning the panel. Replace only the component that fails after cleaning and bidirectional testing. Replacing both panels based on a unidirectional trace with a negative event is over-repair and may not fix the 2.9 dB result.
3. If the link must remain OM4 and meet 2.6 dB, propose two design changes and state the trade-off for each.
Read the worked answer
Option 1: remove connectors. Eliminate one or both patch panels by splicing through or using splice-on LC connectors at the panel locations. Going from four connectors to two reduces the allowance by 1.5 dB (from 3.0 dB to 1.5 dB), bringing the component estimate near 2.68 dB and the measured pair 3 potentially below 2.6 dB. Trade-off: less operational flexibility, because a switch move requires a splice or a new patch cord rather than a simple panel-to-panel reconnection, and the splice closure must be accessible. Option 2: change the physical layer to OS2 single-mode. Use 10GBASE-LR optics with a 6.2 dB budget over OS2. Trade-off: new optics, new patch cords, and possibly new switch configuration, but the loss margin becomes comfortable and the same fiber supports 40GBASE-LR4 and 100GBASE-LR4 at 10 km. A third option, reducing the number of splices or using a lower-loss connector, may recover only a few tenths of a dB and should not be relied on. Whichever option is chosen, re-test with OLTS and document the revised loss budget.