Ribbon Splice Loss and the Four-Hour SLA
A 24-fibre cable between two joint chambers is being spliced at night to restore a bank branch link. The technician has completed 18 splices when the fusion splicer, which has run about 2,100 arcs since the last electrode change, begins reporting estimated losses of 0.02-0.05 dB. The OTDR at 1550 nm, shot from the A end only, shows 0.45 dB at the fibre 19 splice. Re-splicing fibre 19 twice gives 0.31 dB and then 0.55 dB. On the third attempt the cleaver displays a cleave angle of 1.8 degrees. The splice protection sleeve was correctly positioned and centred. The cable is a 12-fibre flat ribbon and the two cables come from different manufacturers, with stated MFD of 9.2 and 8.6 micrometre at 1310 nm. Chamber humidity is high and the temperature is 8 degrees C. The customer SLA requires restoration within four hours; 3 hours 40 minutes have elapsed.
1. The one-way OTDR reading of 0.45 dB is being treated as the true splice loss. Explain why that number may mislead and state the correct measurement method and acceptance figure.
Read the worked answer
A single-direction OTDR splice reading is not a direct loss measurement; it is the difference between two backscatter levels, and it assumes both fibres scatter identically. Here they do not. A 9.2 against 8.6 micrometre MFD pair has an intrinsic mismatch loss of roughly 0.1-0.2 dB, and the backscatter capture differs on each side, so the reading becomes direction-dependent and can even appear negative (an apparent gain) from one end. Near-end dead zone is not the issue at fibre 19 mid-cable, but the mismatch is.
Correct method: measure the same splice from both ends and average the two values algebraically, so 0.45 dB from A and, say, -0.05 dB from B gives a true loss of about 0.20 dB. Record both directions in the splice sheet, never the single worst figure.
Acceptance figure: for a distribution link of this type, accept a bidirectional average of 0.15 dB or better; 0.1 dB is the target for trunk and backbone. A single-direction 0.45 dB with a bidirectional average of 0.20 dB is slightly over but far from the 0.45 dB that triggered the repeat work, so the panic re-splicing was aimed at a number that did not exist.
2. List the corrective actions in the order you would take them, and justify the order using the evidence in the setup.
Read the worked answer
Stop re-splicing and fix the process, because repeated attempts are making the result worse (0.31 then 0.55 dB) and consuming the SLA.
- Cleave first. A 1.8 degree cleave angle is far outside the normal limit of about 0.5 degrees and alone explains a poor arc and a weak splice. Re-strip with a fresh blade position, clean with lint-free wipes and anhydrous alcohol, and reject any cleave that does not show a flat end face under the viewer.
- Service the splicer. At about 2,100 arcs the electrodes are at or past their recommended life, which shows up as unstable arcs and unreliable loss estimates. Replace or rotate electrodes and run the built-in arc calibration.
- Compensate for the cold. At 8 degrees C with high humidity, arc power needs adjustment and condensation on the fibre and in the chamber is a real risk; warm the chamber, keep the fibre in its sealed box until immediately before use.
- Splice and verify. Treat the splicer's estimated loss as indicative only. Do the bidirectional OTDR average before accepting or rejecting.
This order matters: re-splicing with a bad cleave and worn electrodes can only produce another bad splice, which is exactly the 0.31 and 0.55 dB pattern already seen.
3. With 20 minutes of the SLA left, you must decide whether to chase the fibre 19 splice or restore service. State your decision, the acceptance logic, and what you record and hand over.
Read the worked answer
Decision: restore service now and plan a controlled follow-up, because the fibre 19 splice is not the cause of the outage and the residual excess is small compared with the link budget. A bidirectional average of about 0.20 dB against an acceptance limit of 0.15 dB is a 0.05 dB overshoot; on a link with several dB of margin that will not affect the customer's service, whereas missing the SLA has a contractual cost and delays the bank's branch opening.
Acceptance logic: judge the whole link, not one event. Complete all 24 splices, take bidirectional OTDR at 1310 and 1550 nm, and compare total loss with the design budget. If total loss sits comfortably inside budget, hand over. If it is within 1 dB of budget, still hand over but raise a planned-work ticket.
Record and hand over: the per-fibre splice loss table with both directions, OTDR traces saved as files with date, wavelength and pulse width, the splice count and electrode usage of the machine, chamber photographs, and the reason for the residual value on fibre 19 (MFD mismatch of 0.6 micrometre, which cannot be spliced away). The follow-up should use fibre from a matching-MFD drum or accept the intrinsic mismatch and re-baseline the acceptance limit for that splice. Notify the NOC in writing before leaving site.