Acceptable Fusion Splice Loss: Why the Splicer Estimate Is Not the Acceptance Test

A field supervisor on a large commercial infrastructure project applied a strict rule: any single-mode fusion splice showing more than 0.02 dB estimated loss on the splicer had to be cut out and completed again. The work involved a 288-fiber backbone running nearly 2,000 feet between communications rooms. After several months, repeated re-splicing had become a production constraint. Many disputed splices showed clean alignment and no visible defect, yet the machine displayed 0.03 or 0.04 dB.

The dispute was not simply whether 0.02 or 0.05 dB was the better number. It was whether a machine estimate could become a rejection limit when the customer had not defined it in the project documents.

The number only has meaning when its role is clear:

The practical hierarchy is straightforward:

1. The approved project specification defines acceptance.
2. The loss budget defines allowable total attenuation.
3. OTDR testing characterizes individual events when required.
4. OLTS testing measures end-to-end insertion loss.
5. The splicer estimate supports process control.

The 0.05 dB value in the withdrawn ANSI/NECA/FOA 301-2016 document is a fusion-splice planning example, not a universal screen-based rejection rule. The same historical document contrasts it with a more conservative 0.3 dB allowance used in older TIA-based planning examples. FOA announced in 2025 that NECA/FOA 301 had been withdrawn, although it remains available as a historical technical reference. (ANSI/NECA/FOA 301-2016; FOA withdrawal notice)

Fluke also published a 2014 loss-budget example based on ANSI/TIA/EIA-568-C.3 that used 0.3 dB per splice. It illustrates the calculation method, but the cited cabling-standard edition is not current. TIA released ANSI/TIA-568.3-E, Optical Fiber Cabling Component Standard, in 2022. The project documents should therefore identify the standard edition and the actual limits being applied. (Fluke Networks loss-budget example; TIA announcement for ANSI/TIA-568.3-E)

Why a Screen-Based 0.02 dB Rejection Rule Can Fail

A low internal target can improve consistency. If estimates rise across several consecutive fibers, the team has an early warning to inspect cleaning, cleave quality, V-grooves, electrodes, arc calibration, splice mode and fiber compatibility.

The problem begins when that target is treated as a measured contractual limit. AFL describes splice-loss estimation as a prediction based on the fused-fiber image rather than a direct source-and-power-meter measurement. Different splicer designs can improve the estimate, but they do not convert it into an independent optical acceptance test. (AFL, Splice Loss Estimation with the Fujikura 41S Fusion Splicer)

Fujikura's 45S specifications make the same distinction. The displayed estimate is generated during the splice cycle, while its published average splice-loss figures are based on cut-back measurements after splicing. Fujikura also states that average splice loss changes with fiber characteristics and environmental conditions. (Fujikura 45S technical specification)

A screen-only rule can therefore create two opposite errors:

• A splice displaying 0.00 dB may still contain an optical, preparation or mechanical problem that the estimate does not represent accurately.
• A visually sound splice displaying 0.03 or 0.04 dB may remain below the approved OTDR event criterion and within the complete-link loss budget.

Repeated rework also consumes available fiber length and adds another stripping, cleaning, cleaving, fusing, sleeving and tray-handling cycle. On a high-fiber-count project, unnecessary handling can slow production and reduce the margin available for future restoration.

What the Splicer Estimate Should Trigger

The estimate is most useful when it changes the next field action rather than automatically deciding acceptance.

A sustained increase should trigger a process check

When the same operator, machine, fiber type and procedure normally produce estimates around 0.00–0.02 dB, a sustained shift toward 0.04–0.06 dB suggests that something has changed. The inspection should focus on:

• residue or contamination after stripping;
• cleave angle and end-face condition;
• fiber placement and V-groove cleanliness;
• splice-mode selection and fiber compatibility;
• arc calibration and electrode condition.

Fujikura notes that worn electrodes can eventually contribute to high splice loss, while the 45S analyzes cleave end faces and fiber brightness to stabilize the fusion process. (Fujikura splicer maintenance tips; Fujikura 45S technical specification)

A visible or mechanical defect should trigger immediate rework

A bubble, contamination, severe offset, deformation, incomplete fusion, unacceptable cleave warning or failed proof test provides direct workmanship evidence. Re-splicing is justified without waiting for a later optical test.

A clean borderline splice should trigger early verification

When the splice image is clean and symmetrical, the proof test passes and the project documents do not impose a screen-based limit, a value slightly above the internal target can be recorded for early OTDR verification.

Experienced installers often treat the estimate as asymmetrical evidence: a persistently high estimate may justify stopping and checking the process, but a low estimate does not prove that the event will pass OTDR testing. A 0.00 or 0.01 dB display can still be followed by a poor measured event, while a visually sound 0.03 or 0.04 dB estimate may test acceptably.

Field practitioners also report project-specific OTDR event limits such as 0.1, 0.2 or 0.3 dB on different classes of work. These values are examples of contract practice, not a universal hierarchy. They are meaningful only when the documents also state the test direction, averaging method, wavelengths and OTDR settings. (Practitioner discussion on acceptable fusion-splice loss)

If a customer requires 0.02 dB as measured by OTDR, the requirement should be confirmed in writing, including direction, averaging method and tester setup. Before production begins, the team should also verify that 0.02 dB was not intended to be 0.2 dB; a misplaced decimal point changes the requirement by a factor of ten.

How OTDR and OLTS Provide the Acceptance Evidence

OTDR and OLTS answer different questions. Neither should be replaced by the splicer display, and the two test results should not be treated as interchangeable.

OTDR characterizes the individual event

An OTDR sends pulses into the fiber and analyzes returned backscatter and reflections to locate splices, connectors, bends and other events. It reports apparent event loss and reflectance, but the result depends on the test setup and the fibers on each side of the splice. (Fluke Networks, OTDRs Are Essential for Testing and Troubleshooting Fiber Networks)

Single-mode splices can appear different from opposite directions because the two fibers may have different backscatter coefficients or mode-field characteristics. FOA gives an example of a splice reporting 0.25 dB from one direction and 0 dB from the other, producing a 0.13 dB bidirectional average. Bidirectional testing and averaging reduce the directional backscatter artifact when accurate event acceptance is required. (FOA OTDR FAQs; FOA 2025 technical note)

The project method should define:

• one-way or bidirectional testing;
• how directional results are averaged;
• test wavelengths;
• pulse width and averaging settings;
• event-detection thresholds;
• launch and receive fiber requirements.

A high result in one direction and a low result in the reverse direction should therefore be evaluated using the specified bidirectional method before the splice is rejected.

OLTS determines complete-link insertion loss

An OLTS uses a light source at one end and a power meter at the other to measure the total insertion loss of the link. Fluke identifies OLTS as the Tier 1 method for accurate end-to-end insertion-loss measurement, while OTDR provides Tier 2 event characterization. (Fluke Networks, OLTS & OTDR: A Complete Testing Strategy)

A link can pass its total OLTS budget while still containing an event that exceeds a separately specified OTDR limit. Conversely, a slightly elevated splicer estimate may be present even though neither the event criterion nor the complete-link budget is exceeded.

Fusion-Splice Rework Decision Matrix

A Practical Test Plan for the 288-Fiber Project

The project did not need to choose between re-splicing every estimate above 0.02 dB and waiting until all 288 fibers were complete. A staged test plan could limit both unnecessary rework and the risk of discovering a systematic problem too late.

1. Confirm the acceptance plan

Before production, the owner and contractor should agree on:

• whether the displayed-loss target is advisory or contractual;
• the maximum OTDR event loss;
• whether bidirectional averaging is required;
• the OLTS loss budget for each route;
• wavelengths, reference method and tester configuration;
• production-batch test frequency;
• physical defects requiring immediate rejection;
• required final records and calibration information.

The available field scenario does not provide these final decisions, so it cannot establish whether 0.02, 0.05 or another value ultimately applied.

2. Complete a representative trial batch

Use the intended production process and record the fiber ID, operator, splice program, estimated loss, image result, date and location. The trial batch should represent the actual cable, pigtail, machine, splice mode and working environment.

3. Correlate estimates with early OTDR results

Test the trial group before every tray or closure is fully dressed. If visually sound 0.03 or 0.04 dB estimates repeatedly produce acceptable bidirectional event results, automatic rejection at 0.02 dB may not be justified. If elevated estimates consistently correlate with excessive measured loss, the internal threshold is providing useful warning.

This correlation is local to the tested combination of fiber, equipment, operator and OTDR method. It should not be converted into a universal rule for other projects.

4. Move into controlled production batches

Correct systematic issues before mass production, then test by tray, ribbon group or another defined batch. Batch control limits the amount of completed work exposed to one unnoticed process problem and avoids reopening a fully dressed 288-fiber installation.

5. Complete final OLTS certification

After sleeves are heated, splices are secured, slack is routed and the panel or closure is complete, perform the specified end-to-end insertion-loss test. Retain the applied limit, reference method, wavelengths, measured loss, route length, OTDR event map where required, bidirectional averages and tester calibration information. Combined OLTS and OTDR records provide compliance evidence and a troubleshooting baseline. (Fluke Networks, OLTS & OTDR: A Complete Testing Strategy)

Protect the Accepted Splice

Passing the optical criteria does not remove the need for correct mechanical protection. Confirm fiber compatibility, match the protection sleeve to the coating and holder dimensions, center the joint in the sleeve, maintain tray bend radius and complete final testing with the sleeves and slack in their installed condition.

Glory Optical's guide to fiber pigtail types and splice selection provides related compatibility context, while its fusion-splicing procedure guide covers preparation and fusion workflow. Numerical acceptance statements on linked pages should remain consistent with the distinction between a splicer estimate, a planning allowance and a measured project limit.

Frequently Asked Questions