Bidirectional OTDR Acceptance of an Apparent Negative-Loss Fusion Splice
A 18.2 km single-mode outside-plant span on G.652.D fiber is being accepted after fusion splicing six splices. The contractor specification is a maximum bidirectional splice loss of 0.15 dB; unidirectional OTDR values are not acceptance values. At 1550 nm the OTDR trace from the A end reports splice S3 as +0.31 dB, while the trace from the B end reports the same splice as -0.09 dB. At 1310 nm the A-to-B value is +0.18 dB and the B-to-A value is +0.02 dB. The splice was made with a calibrated fusion splicer and the estimated loss was 0.02 dB. No reflective event is visible at S3. The project manager wants S3 cut out and re-spliced because a splice cannot have negative loss. The span otherwise passes. You have the two bidirectional traces and the splicer log.
1. Calculate the bidirectional splice loss at 1550 nm for S3 and state whether it should be reworked under the project specification. Show the arithmetic and the reasoning.
Read the worked answer
The bidirectional average is the correct acceptance metric:
(0.31 dB + (-0.09 dB)) / 2 = 0.22 / 2 = 0.11 dB
At 1310 nm the average is (0.18 + 0.02) / 2 = 0.10 dB.
Both wavelengths are below the 0.15 dB bidirectional limit, so S3 should not be cut out on the basis of the +0.31 dB unidirectional reading. The +0.31 dB value is an OTDR measurement artifact caused by differences in backscatter between the fibers on each side of the splice; it is not the splice loss by itself. The correct action is to record both directional values, record the calculated average, and accept the splice if the average and all other acceptance criteria pass. Reworking a passing splice adds two new splices and risk without improving the link.
2. Explain why the B-to-A OTDR trace can report -0.09 dB at a passive fusion splice, and identify the field evidence that confirms this is an apparent gain rather than a real improvement.
Read the worked answer
An OTDR infers splice loss from the change in backscattered power across the event. If the fiber beyond the splice has a higher backscatter coefficient than the fiber before it, the trace can step upward even though the splice adds loss. The result is an apparent negative splice loss, often called a gainer. Mode-field diameter mismatch between the two fibers can also make the unidirectional result direction-dependent.
The confirming evidence here is: no reflective event at S3, a low splicer-estimated loss of 0.02 dB, a consistent 1310 nm result, and a 1550 nm bidirectional average of 0.11 dB. A true amplifying splice does not exist in passive fiber. The negative value should be averaged with the opposite direction, not treated as a real gain and not used to offset other losses.
3. What would you do if the reverse trace were unavailable or if the bidirectional average were 0.22 dB?
Read the worked answer
If the reverse trace is unavailable, do not use the unidirectional +0.31 dB as the final acceptance value. Obtain a reverse-direction measurement by testing from the far end, by using a temporary far-end launch and OTDR setup, or by an approved alternative such as an end-to-end power-meter loss check that isolates the splice. Document that the unidirectional result is provisional.
If the calculated bidirectional average were 0.22 dB, it would exceed the 0.15 dB limit. I would inspect the splice protection sleeve and tray routing first, verify the cleave and splice image, then cut out and re-splice if no bend or measurement problem explains the result. After re-splicing, retest both directions and record both values. The decision rule is the same: accept on the bidirectional average, not on a single direction or on the splicer estimate alone.