Singlemode OSP Span With a Suspect Splice
A 3.6 km, 12-strand singlemode span between a main building and a gatehouse is ready for acceptance. The route has four fusion splices in two closures and one LC connector at each end. The design loss budget uses 0.35 dB/km at 1310 nm, 0.10 dB per splice, and 0.50 dB per connector. An OLTS at 1310 nm reads 2.9 dB end-to-end. An OTDR tested from the main building shows a splice at 1.4 km with a 0.85 dB event, while the other three splices read 0.08-0.12 dB. The OTDR also shows a reflective event at 3.6 km and a small ghost beyond it. The gatehouse end is not accessible today. The contractor wants to accept the span because "2.9 dB is under the 3.0 dB budget" and says the OTDR splice is a measurement artifact.
1. Calculate the design loss budget and compare it with the 2.9 dB OLTS result. Is the span acceptable as tested? Show the contribution of the 0.85 dB splice.
Read the worked answer
Fiber: 3.6 km x 0.35 dB/km = 1.26 dB. Splices: 4 x 0.10 = 0.40 dB. Connectors: 2 x 0.50 = 1.00 dB. Total = 2.66 dB. The OLTS result is 2.9 dB, which is 0.24 dB above the calculated budget, so the span does not meet the stated design budget as tested. The one high splice is 0.85 - 0.10 = 0.75 dB above its allowance; that alone is larger than the 0.24 dB excess, so other events or connectors may be better than allowance, or the readings have uncertainty. The contractor's "under 3.0 dB" claim uses a rounded number, not the calculated design budget; 3.0 dB is not the limit. Do not accept until the high-loss event is resolved and the final loss is within budget with margin.
2. The far end is not accessible. What can you conclude from the one-way OTDR, and what is the minimum additional action before ordering a re-splice?
Read the worked answer
A one-way OTDR splice loss can be biased by differences in backscatter and by launch conditions; a single 0.85 dB event is a flag, not proof. The event is non-reflective and located at a closure, which is consistent with a splice rather than a connector. Because the OLTS end-to-end loss is also over budget, there is corroborating evidence, but the far-end connector and the last span are not measured. Minimum action: clean and inspect both accessible connectors, use a proper launch fiber (and receive fiber if possible), repeat the OTDR from the main building, and obtain a bidirectional OTDR or an OLTS measurement from the far end when access is available. If the bidirectional average loss for that splice is still above the acceptance value, re-cleave and re-fuse it. Do not re-splice all four based on one one-way trace.
3. What acceptance records should be provided, and how should the ghost event be handled?
Read the worked answer
Records should include the link loss budget; OLTS results at the specified wavelength(s) in both directions; OTDR traces in both directions with launch/receive fibers; a splice loss table with locations; connector inspection and cleanliness records; equipment calibration certificates; and as-built fiber assignments. The ghost beyond 3.6 km is caused by a strong reflection, likely the far connector, re-entering the OTDR and appearing at a multiple of the distance; it is not a physical event. Use a receive fiber or longer launch to move the ghost away from the link, and verify with bidirectional traces. Acceptance should be based on corrected, bidirectional data, not the raw one-way trace with the ghost.