A splice appears to gain 0.12 dB when tested from one end. From the other end it appears to lose 0.36 dB. Which result belongs on the splice record? Before opening the closure, establish that both traces measured the same joint, at the same wavelength, with enough resolution to separate it from neighboring events.
Table of Contents
ToggleWhy the direction changes the reading
An OTDR sends a pulse down one fiber and observes the light that returns. The slope between joints shows how the returned signal changes with distance. At a fusion splice, the instrument estimates event loss by comparing the backscatter levels on either side. Those levels depend on the fibers as well as the joint.
If the far-side fiber sends back more light than the near-side fiber, the trace can step upward across a passive joint. The event table may display negative loss, often called a gainer. Reverse the test direction and the same difference biases the reading the other way. The joint has not amplified the signal. A single reading cannot tell you how much of the step is splice loss.
Match the physical joint before doing the math
Start with the fiber identifier, the two terminal panels and the splice plan. A→B event 3 need not be B→A event 3: the reverse trace counts from the other end. Set both distance origins at the ends of the installed link, excluding OTDR launch and receive leads. Compare the positions and the sequence of neighboring joints.
Example: On a 4.00 km installed span, a joint 1.30 km from A should appear about 2.70 km from B. These are optical path distances, so use the trace settings and as-built records when comparing them with route-map distances. If two joints are close together, agreement in rounded distance alone is not enough.
Calculate the signed two-way estimate
For an isolated splice with usable backscatter on both sides, take the arithmetic mean of the two signed event losses in dB. Use readings taken at the same wavelength. Here the forward reading is −0.12 dB and the reverse is +0.36 dB, so the estimate is (−0.12 + 0.36) / 2 = 0.12 dB. Taking absolute values first would yield 0.24 dB and could trigger unnecessary rework.
The numbers are illustrative, not measurements from a product or a recommended acceptance limit. Compare the estimate with the splice criterion in the job specification. If the specification defines only end-to-end insertion loss, record the event separately rather than inventing a per-splice pass limit.
When the average is not usable
Look at the trace around the event before accepting the event table. A long pulse may combine two nearby joints; a strong reflection may leave the receiver recovering at the next splice. At the far end of a noisy trace, a fitted line may follow noise instead of the fiber slope. Averaging two unreliable numbers does not isolate the joint.
- Events merge: try a shorter pulse and check that the trace still reaches the joint with usable signal on both sides.
- One direction is obscured: inspect the launch arrangement and the preceding reflective event. Reacquire the affected direction before pairing traces.
- Only one end is accessible: report a one-way apparent event loss with its direction. Leave the two-way result unresolved.
Decide what to retest before opening the closure
If repeatable, correctly paired readings exceed the agreed event limit, retain the original traces and identify the tray and fiber before intervention. After an authorized repair, repeat both directions at the original wavelength and compare the same event. A favorable splice estimate alone cannot clear the whole link: an end-to-end insertion-loss test with a source and power meter answers a different question.
A useful handover record includes fiber ID, direction, wavelength, pulse width, lead lengths, event positions, signed readings, their mean and the applicable acceptance criterion. Keep both native trace files so the next technician can distinguish a new fault from the original backscatter step.
References and field support
For trace interpretation and the worked bidirectional method, see ITU-T G.650.3, Appendix I and the FOA guide to OTDR testing. For the physical repair location, see our splice closure installation guide.