融着接続機の損失推定値とOTDRの比較：なぜ0.00 dBでも接続不良になることがあるのか

A 融着機 loss estimate of 0.00 dB looks like a perfect joint. Then the OTDR reports 0.30 dB, 0.50 dB, or even a negative “gainer” at the same location. Which number should you trust?

First, identify what each instrument is actually reporting. The splicer calculates an estimate from alignment images and its internal model. An OTDR infers event loss from backscattered light. A light source and power meter measure the total end-to-end insertion loss. These values are related, but they are not interchangeable.

Why a Fusion Splicer Loss Estimate Can Read 0.00 dB

A fusion splicer normally observes the prepared fiber ends, aligns the fibers, applies the arc, and analyzes the resulting image. Depending on the model, the estimate may use visible core or cladding offsets, cleave condition, fiber geometry, arc data, and an internal loss model.

The splicer does not send a calibrated test signal through the completed link and measure the received power at the far end. A displayed 0.00 dB can therefore mean “the calculated result rounded to 0.00 on this instrument,” not “the splice has exactly zero physical loss.”

Treat the screen value as immediate process feedback. It can help catch gross alignment, cleave, contamination, or arc problems, but it is not a certification result.

読書	What it answers	Main limitation
Splicer estimate	Does the joint look acceptable to this splicer and algorithm?	Calculated locally; not a direct transmission measurement
One-way OTDR event loss	How does the event change backscatter from one direction?	May show false high loss or a gainer
双方向OTDR average	What is the best field estimate of this event’s loss?	Requires correct event pairing and signed averaging
OLTS/LSPM insertion loss	Does the complete link meet its end-to-end budget?	Does not locate one individual splice

ITU-T L.12 states that a splicer estimate or one-way OTDR can identify gross high-loss joints during construction, while actual field splice loss can be determined by averaging OTDR readings from both directions.

First Check: Is the “High Loss” Only a One-Way OTDR Result?

An OTDR measures returned Rayleigh backscatter rather than transmitted power. If the fibers on opposite sides of a splice have different mode field diameters or backscatter coefficients, the trace can step down too far in one direction and step up in the other.

The step-up is called a gainer. The splice is not creating optical power. It is a measurement artifact caused by different backscatter levels.

Use the signed readings from the same physical splice:

方向	OTDR event reading	Interpretation
A → B	+0.36 dB	Exaggerated loss may be present
B → A	−0.18 dB	Gainer in the reverse direction
Signed average	(+0.36 + −0.18) ÷ 2 = 0.09 dB	Best estimate from this trace pair

Do not remove the minus sign. Averaging the absolute values would incorrectly produce 0.27 dB.

This behavior is especially relevant when joining different single-mode fiber designs, such as G.652D and G.657A2. Corning’s OTDR assessment guidance documents cases where splicer predictions remained at or below 0.01 dB while one-way OTDR results showed exaggerated loss because of MFD-related backscatter differences.

For the fiber-specific background, see G.652DとG.657A2の光ファイバーを接続することはできますか？.

Three Reasons the Splicer and OTDR Can Disagree

1. The displayed 0.00 dB is a rounded estimate

The splicer has limited display resolution and uses a model rather than a direct through-power test. A very small predicted value may display as 0.00 dB. Different models may also produce different estimates from the same joint.

2. The OTDR result is directionally biased

Different mode field diameters or backscatter coefficients can make a good splice look like a high-loss event from one end and a gainer from the other. A one-way result is not the actual splice loss.

3. Real loss appeared outside the estimate

Genuine excess loss can result from contamination, a bad cleave, alignment error, the wrong splice program, poor arc conditions, a bubble, tapering, or damaged fiber.

Loss may also appear after the splice leaves the camera view—for example, when the protection sleeve is heated, the fiber is pinched in the tray, or a tight bend is introduced.

The OTDR may also be marking the wrong event. A nearby connector, mechanical splice, bend, splitter, or reflective fault can be mistaken for the fusion splice if route distances are not mapped correctly.

7つのステップからなるトラブルシューティングのワークフロー

1. Preserve the evidence before cutting the splice

Save the splicer record, fiber images, program name, estimated loss, cleave angles, date, operator, and splice ID. Save the original OTDR trace rather than only a screenshot.

Record the wavelength, range, pulse width, averaging time, event threshold, index setting, launch cable, receive cable, and test direction.

2. Confirm that both tools refer to the same physical event

Match the event to route drawings, cable markings, splice enclosure, tray, and fiber number. The same splice appears at a different distance when measured from the opposite end.

Check whether the event is reflective. A normal fusion splice is usually non-reflective. A clear reflection may indicate a connector, mechanical gap, contaminated interface, or incorrect event assignment. ITU-T L.12 states that a measurable spike from a confirmed fusion splice requires remake.

3. Repeat the OTDR test from both ends

Use the same wavelength and compatible range, pulse width, averaging, and analysis method. Use clean, suitable launch and receive cables.

Avoid changing several settings between directions. A long-pulse automatic trace and a short-pulse manually analyzed trace may not form a valid pair.

4. Average the signed event readings

Pair the same physical splice and calculate:

Bidirectional splice loss = (A→B reading + B→A reading) ÷ 2

Keep a negative gainer value negative. If one direction is high and the reverse result is negative by a similar amount, backscatter mismatch is probably influencing the result. If both directions are high and positive, a real defect is more likely.

5. Check the tray and protection before re-splicing

Inspect the joint and nearby fiber for:

• A sleeve that is bent, twisted, or not centered
• Fiber pinched under a cover or retaining tab
• A tight turn at the splice holder
• Buffer-tube pressure
• A sharp transition into the tray
• Loss that increases at the longer test wavelength

Correct routing only under the approved handling procedure, then repeat the measurements. Do not assume every post-splice loss is inside the fused glass.

6. Re-splice only when evidence indicates a real problem

If the bidirectional average fails the project limit, the confirmed splice is reflective, the image shows a true defect, or the protected joint has been damaged, cut out and remake it under the approved procedure.

For the remake:

1. Confirm both fiber types, coatings, and the documented splice program.
2. Strip and clean with the approved materials.
3. Make a fresh cleave and reject contaminated or visibly defective end faces.
4. Clean the V-grooves, clamps, and camera area as specified by the manual.
5. Check the cleaver blade position and condition.
6. 実行 arc calibration when required by the exact splicer manual, environment, or maintenance state.
7. Inspect electrode condition and follow the model-specific maintenance procedure.

AFL’s official troubleshooting guide lists poor cleaning, bad end faces, bubbles, V-groove or camera contamination, worn electrodes, uncalibrated arc power, incorrect programs, and unsuitable parameters among possible causes of high estimated loss. It also treats calibration procedures as model-specific.

Do not repeatedly apply an extra arc simply to reduce the displayed number. Additional arcs can worsen some splices.

For preparation, the verified LuLeey S09 cleaver page lists an average cleave angle below 0.5° and support for 250 µm, 900 µm, and 3.0 mm single-fiber constructions. The cleave length must still match the exact splicer and holder requirement.

7. Retest the event and the complete link

Repeat the bidirectional OTDR test after the repair or routing correction. Then use the project’s approved OLTS or light-source-and-power-meter method when the complete link must be certified.

Corning distinguishes these roles: OTDR traces locate and characterize events, but they do not replace Tier 1 end-to-end attenuation testing with an OLTS/LSPM. corning.com

Fast Decision Table

証拠	Most likely meaning	次のアクション
Splicer 0.00 dB; one-way result high; reverse is a gainer	Directional backscatter/MFD artifact may dominate	Average signed readings before remaking
Splicer 0.00 dB; both directions high	Real splice loss, bend, or wrong event is likely	Inspect mapping and tray; re-splice if the averaged result fails
Splicer 0.00 dB; reflective spike	Not a normal simple fusion-splice signature	Confirm event identity and inspect for a gap, connector, or defective joint
Splicer estimate high; bidirectional result passes	Estimation, image, or maintenance issue may exist	Inspect equipment; do not fail the link from the estimate alone
OTDR events pass; OLTS link loss fails	Cumulative loss, connector/reference issue, bend, or another event	Troubleshoot the complete link

What to Send for a Useful Technical Review

“The splicer says 0.00” is not enough evidence. Send:

• Exact splicer model, firmware, splice program, and maintenance state
• Fiber types and cable part numbers on both sides
• Original splicer images and records
• Original A→B and B→A OTDR trace files
• OTDR wavelength, pulse, range, averaging, thresholds, and launch/receive cables
• Event distance from each end
• Photos of the protected splice and tray routing
• End-to-end OLTS/LSPM result, if available

LuLeey can then help distinguish a preparation issue from an OTDR artifact or application mismatch. Use the LuLeey contact page and include the exact equipment model and trace files.

実践的な次のステップ

Do not cut out a splice because one instrument shows a surprising number. First map the event, test from both ends, and average the signed OTDR readings.

If the bidirectional result still fails—or the event is reflective, visibly defective, or mechanically stressed—repair the cause, make a fresh splice, and retest the complete link.

信頼できる参考文献

1. ITU-T L.12: Optical fibre joints
2. Corning AN3060: Guidance for OTDR Assessment of Fusion Spliced Single-Mode Fibers
3. AFL AN-3035: 90S+ Splicer Error Code and Troubleshooting Quick Guide