The Gainer After the Reflective Event at 5.6 km
You are characterizing an 8.2 km single-mode span at 1550 nm with a 100 ns pulse after setting the range and group index. The trace shows an event at 0 m of 0.45 dB with -52 dB reflectance, a splice at 2.40 km of 0.06 dB, a reflective event at 5.60 km of 0.55 dB with -38 dB reflectance, an upward step of 0.12 dB at 5.65 km (a gainer), and the far end at 8.20 km with a Fresnel reflection of -45 dB. The OTDR's total span loss reads 4.90 dB. A one-jumper light source/power meter measurement of the same span reads 4.20 dB. The span was built from 2.0 km of legacy single-mode cable and 6.2 km of new G.652.D, joined at the 5.65 km splice. Your acceptance report is due today and the customer wants to know whether the 5.60 km event is a failing connector.
1. Explain the gainer at 5.65 km. How do you report the splice loss there, and how do you report the 5.60 km event?
Read the worked answer
The 5.60 km event reflects -38 dB, which is strong enough that the tail of its reflection lifts the apparent backscatter level for some distance after it. The 5.65 km splice sits inside that disturbed region, so its loss is measured against an inflated baseline and appears as a gain. The change of fiber type at that same splice reinforces the effect: 2 km of legacy fiber and 6.2 km of G.652.D have different backscatter coefficients and mode-field diameters, so a splice between them shows a different apparent loss from each direction, and can genuinely look like a gainer from one end.
How to report it: take the OTDR trace from both ends and average the two directional splice losses: L = (L_A->B + L_B->A) / 2. That cancels the backscatter-coefficient difference. A one-directional gainer is never reported as a negative loss, and never as a real splice value either.
The 5.60 km event: -38 dB reflectance is worse than a clean UPC connector (typically -50 dB or better) and indicates an air gap - contamination, a damaged or scratched endface, or a UPC mated to an APC. Its 0.55 dB is also high for a mated pair. Report it as a suspect connector requiring inspection and cleaning, re-test, and if reflectance stays above spec, replacement.
2. Why does the OTDR total 4.90 dB differ from the power meter's 4.20 dB? Which number goes in the acceptance report?
Read the worked answer
The two instruments measure different things and neither is 'wrong'.
- The OTDR infers loss from backscatter levels. Its splice and connector losses are relative to the backscatter on either side, so reflectance tails, backscatter-coefficient differences and dead zones bias them - exactly the 5.60-5.65 km region here.
- The near-end connector is inside the OTDR's dead zone unless a launch fiber is used, so its 0.45 dB is an estimate, not a measurement.
- The far-end connector is partly obscured by the end reflection.
- The OTDR's own linearity is typically +/-0.05 dB/dB, i.e. about +/-0.25 dB on a 5 dB span, on top of the power meter's +/-0.2 to 0.5 dB.
Report the source/power meter result (4.20 dB) as the link insertion loss. That is what the cabling standard's field test specifies and what the channel budget is written against. Use the OTDR only for locating and characterizing events, with splice losses taken bidirectionally. Include the difference in the report as an explanation rather than as a discrepancy - the 0.6-0.7 dB gap is consistent with the reflective event and the fiber-type change.
3. Would a shorter pulse have improved this trace? Explain the trade-off.
Read the worked answer
A shorter pulse (e.g. 10-30 ns) narrows the dead zones, so the 5.60 km connector and the 5.65 km splice would be resolved more cleanly and less contaminated by each other's reflection tail. It would also sharpen the near-end event.
The cost is dynamic range: a 10 ns pulse launches roughly one-tenth the energy of a 100 ns pulse, so the noise floor rises by about 5 dB and the far end may become marginal. On an 8.2 km span with about 4-5 dB of loss and a -45 dB end reflection, that is affordable - but the practical approach is two acquisitions:
- 100 ns for the whole span - total loss, far-end continuity, overall event map.
- 10-30 ns over the 5.4-5.9 km region (zoom, and if the instrument allows it, a shorter pulse on the same span) - to resolve the connector and splice and to check whether the gainer is an artefact of the reflection tail or of the fiber mismatch.
Do not change pulse width mid-report: state which pulse each measurement used, because splice loss values from different pulse widths are not directly comparable.