Fiber End-Face Inspection: IEC 61300-3-35 Zones and Pass/Fail Criteria

IZAK multi-channel instrument front panel with five fiber connector receptacles, the interfaces whose end faces must be inspected before mating

In short

  • Fiber end face inspection grades the connector by zone. IEC 61300-3-35 divides the end face into core, cladding, adhesive and contact zones, with the strictest limits at the core.
  • Single-mode is unforgiving. Under the widely used single-mode PC limits, no scratch and no defect is allowed in the 25 µm core zone. A 3 µm particle on the core can block about 19% of the mode, close to 1 dB.
  • Inspect, clean, re-inspect, then mate. Many rejects are loose debris that cleaning removes. A permanent pit or scratch in the core zone is a polish or handling problem.
  • Automate the verdict. A microscope that resolves 2 µm defects across a 250 µm field, analysed by software, gives the same answer every time and leaves a record per connector.

Why fiber end face inspection matters

A fiber connector works only if two glass end faces meet in physical contact over the core. Anything between them, a dust particle, a film of oil, a chip in the glass, changes how light crosses the joint. In single-mode fiber the light travels in a mode only about 9–10 µm across, so a particle far thinner than a human hair is enough to block a large share of it.

Contamination does three kinds of damage. It adds insertion loss by absorbing or scattering light. It lowers return loss because it creates an air gap and a reflecting surface at the joint. And it spreads and embeds: when the connectors are mated, a hard particle can be pressed into both end faces, turning a cleaning problem into two damaged connectors. In a published cleanliness study on single-mode connectors, contamination near the core raised insertion loss by up to 1.8 dB and changed return loss by 10 to 40 dB, while the same kind of debris on the outer ferrule had no significant optical effect.

That last point is why inspection standards grade the end face by zone instead of counting every speck equally. If you work with fiber assemblies, it also explains why coupling efficiency and end-face quality are usually tested together.

The four zones of IEC 61300-3-35

IEC 61300-3-35 is the international test method for visual and automated inspection of polished connector end faces with 125 µm cladding fiber in a ferrule. It defines concentric zones around the core centre and sets a limit for each zone, in sizes and counts:

A · Core zone0–25 µm (single-mode), 0–65 µm (multimode)B · Cladding zoneto 115 µm (110 µm in the 2022 edition)C · Adhesive zone115–135 µm, no limitD · Contact zone135–250 µm, large debris only
Inspection zones on a ferrule end face, drawn to scale by diameter. The green dot marks a 5 µm particle for comparison.

The zone sizes are diameters. Zone A covers the core and the light-carrying mode. Zone B is the cladding, where defects do not block light directly but can move to the core or keep the faces apart. Zone C is the ring of adhesive between glass and ferrule, which is not graded. Zone D is the contact area of the ferrule out to 250 µm, where only large debris matters because it can stop the two ferrules from touching.

The standard defines a defect as any non-linear feature: particles, debris, fluid residue, pits and chips. Its size is the diameter of the smallest circle that encloses it. A scratch is a linear feature, judged by its width. Loose debris counts as a defect, which is why an end face that fails before cleaning often passes after it.

Pass/fail criteria for fiber end face inspection

The table shows the limits most production specifications use for polished PC connectors, from the 2015 edition of the standard. Single-mode values apply to connectors with return loss of 45 dB or better; angled (APC) connectors and lower return-loss classes have their own tables.

ZoneSingle-mode PC: scratchesSingle-mode PC: defectsMultimode PC: scratchesMultimode PC: defects
A · CoreNoneNoneNo limit ≤ 3 µm, none > 3 µmUp to 4 of ≤ 5 µm, none > 5 µm
B · CladdingNo limit ≤ 3 µm, none > 3 µmNo limit < 2 µm, up to 5 of 2–5 µm, none > 5 µmNo limit ≤ 5 µm, none > 5 µmNo limit < 2 µm, up to 5 of 2–5 µm, none > 5 µm
C · AdhesiveNo limitNo limitNo limitNo limit
D · ContactNo limitNone ≥ 10 µmNo limitNone ≥ 10 µm

The third edition (2022) changed several things, so check which edition your customer or contract cites:

  • The outer edge of zone B moved from 115 µm to 110 µm, to allow for fixture tolerances.
  • Zones C and D are no longer graded. Instead, a cleanliness ring from 135 to 250 µm is recommended to hold no more than 10 particles of 5–10 µm and none larger than 10 µm.
  • A defect that is only partly inside the core is judged in the core only by the part that is inside.
  • Requirements were added for single-mode connectors with 35 dB return loss.
  • Visual inspection supplements, but does not replace, insertion-loss and return-loss measurement. A connector that meets its optical specification is not rejected on appearance alone.

The last point matters in production. Inspection is the fast screen that catches dirt and damage before mating; the optical test is the final judge of performance. A good process uses both.

Worked example: what a particle on the core costs

Take a single-mode connector at 1310 nm, where the mode field diameter (MFD) is about 9.2 µm, so the Gaussian mode radius is w = 4.6 µm. The fraction of the mode power inside a circle of radius a around the core centre is 1 − exp(−2a²/w²).

  • A 3 µm particle centred on the core (a = 1.5 µm) covers 1 − exp(−2 × 1.5² / 4.6²) = 19.2% of the power. If that light is absorbed or scattered, the extra loss is −10·log₁₀(1 − 0.192) = 0.92 dB.
  • The same particle 3 µm off centre still covers 8.9% of the power: 0.41 dB.
  • A 2 µm particle on the core: 9.0%, 0.41 dB.
  • A 5 µm particle on the core at 1550 nm (MFD 10.4 µm): 37%, 2.0 dB.
  • A 3 µm particle 20 µm from the centre, still inside zone A, covers practically none of the mode.

The last case is why the core zone is wider than the core. A particle 20 µm out costs nothing today, but it sits where mating pressure and the next cleaning wipe can push it onto the core, and it can hold the two faces apart. The numbers above are an upper bound for absorption and scattering loss only; an air gap held open by debris also lowers return loss, which can be the larger problem for analogue and high-speed links.

Lateral offset between the two cores adds its own loss on top of this. The ferrule inspection article has a calculator for core-to-ferrule offset; at 1310 nm a combined offset of 1 µm costs about 0.21 dB.

Calculator

End-face grader and core particle loss

Enter up to six features from your microscope image: type, size (defect diameter or scratch width) and distance of its centre from the core centre. The tool sorts them into zones, applies the 2015 PC limits for single-mode or multimode, and for single-mode estimates how much of the mode the core-zone defects can block.

Zones and limits: IEC 61300-3-35, 2015 edition, PC polish; zone by the distance of the feature centre. Particle loss: fraction of a Gaussian mode covered by an opaque disc, summed over core-zone defects, as an upper bound for absorption and scattering loss. Return-loss effects, APC tables and the 2022 edition changes are not modelled; the optical test decides.

Inspect, clean, re-inspect: the production routine

The order of operations matters more than the brand of wipe. A routine that works on the production floor and in the field:

  1. Inspect both sides before every mating. That includes the bulkhead or receptacle, which is the side most often forgotten and the one that contaminates every patch cord plugged into it.
  2. If it fails, clean. Start with a dry clean (lint-free click cleaner or wipe). If residue remains, use a wet-to-dry clean with a suitable solvent, finishing dry so no film is left.
  3. Re-inspect. Debris that moved is not debris that left. Often a fail turns into a pass after one or two cleaning cycles.
  4. If it still fails, look at the type of feature. A permanent pit, chip or scratch in zones A or B points to polishing, handling or mating damage, not dirt. Repolish or replace, and feed the finding back to the process step that caused it.
  5. Record the result. An image and a verdict per connector serial number make customer audits and failure analysis much shorter.

For high-power delivery fibers, such as those used in surgical and industrial lasers, contamination does more than add loss: it absorbs power, heats and can burn into the glass. These assemblies often use larger or bare fibers that fall outside the 125 µm ferrule scope of the standard, and manufacturers usually write their own, stricter acceptance criteria tied to transmission tests. The medical laser verification and validation guide shows how such tests fit into a device test plan.

Manual microscopes vs automated end-face analysis

The standard allows three methods: direct-view optical microscopy (A), video microscopy with an on-screen overlay (B) and automated analysis microscopy (C), where software acquires the image and compares it with programmable acceptance criteria. For the standard field of view the system must cover at least 250 µm and detect 2 µm defects and 3 µm wide scratches across the whole field.

QuestionManual (methods A, B)Automated (method C)
Who decides pass/fail?The operator, against a reticle or overlaySoftware, against stored criteria per fiber type
RepeatabilityDepends on the operator, lighting and fatigueSame image, same answer; variability comes from optics and focus
Record per connectorRarely keptImage, feature list and verdict saved automatically
ThroughputSlow for multi-fiber partsSuited to volume and multi-channel parts
Best forField work, low volume, troubleshootingProduction, incoming inspection, final QA

Two sizing checks help when you specify or build an automated system. Sampling: a 300 µm field imaged onto 1280 pixels gives about 0.23 µm per pixel, so a 2 µm defect spans about 8 to 9 pixels and is detected reliably; a 250 µm field on 640 pixels gives about 5 pixels per 2 µm defect, which is marginal once focus and noise are included. Optics: the diffraction-limited resolution 0.61·λ/NA is about 1.3 µm for an NA of 0.25 in visible light and about 0.8 µm at NA 0.4, so the objective, not only the camera, sets what the system can see. The 2022 edition also warns that results vary between inspection systems, even from the same supplier, so validate any system against reference end faces before you rely on its verdict.

Automated inspection also measures what a person cannot judge by eye, such as the offset between the core and the ferrule centre and the symmetry of the adhesive. For the background on modes, cores and cladding, see the beginner’s guide to fiber optics.

IZAK products for this application

IZAK builds automated inspection and test stations for fiber connectors and fiber assemblies, from a single ferrule check to multi-channel production lines.

Need a station built around your connectors and limits? See fiber optic design and custom electro-optical systems.

Setting up end-face inspection for production?Send the connector and ferrule types, single-mode or multimode, the volumes and the acceptance limits you test to. We will propose an inspection and transmission test configuration and a demonstration plan.

Frequently asked questions

QWhat are the zones in IEC 61300-3-35?
For single-mode connectors the core zone A is 0–25 µm in diameter, the cladding zone B extends to 115 µm (110 µm in the 2022 edition), the adhesive zone C is 115–135 µm and the contact zone D is 135–250 µm. For multimode fiber the core zone is 0–65 µm so it covers both 50 µm and 62.5 µm cores.
QIs any scratch allowed on the core of a single-mode connector?
Under the widely used single-mode PC limits (return loss 45 dB or better, 2015 edition), no scratch and no defect is allowed in the 25 µm core zone. In the cladding zone, scratches up to 3 µm wide are allowed in any number, and none wider. Check which edition and return-loss class your specification cites.
QCan a connector that fails inspection still be used?
First clean and re-inspect: many fails are loose debris. If a permanent feature remains, the 2022 edition states that visual inspection supplements but does not replace insertion-loss and return-loss tests, and a connector that meets its specified optical performance is not rejected on appearance alone. Many production specifications are stricter, so follow the customer’s limits.

Standards and publications

  1. IEC 61300-3-35:2015, Fibre optic interconnecting devices and passive components – Basic test and measurement procedures – Part 3-35: Examinations and measurements – Visual inspection of fibre optic connectors and fibre-stub transceivers.
  2. IEC 61300-3-35:2022 (edition 3.0), same title.
  3. T. Berdinskikh et al., “Development of Cleanliness Specification for Single-Mode Connectors”, APEX 2005.
>Tzachi Sabati Tzachi Sabati
CEO, IZAK Scientific
Physicist specializing in photonics and quantum technologies, with deep expertise in quantum sensors and advanced optical systems. Leads the Advanced Quantum Lab course at the Technion, bridging academic excellence with industry innovation. At IZAK Scientific, provides cutting-edge photonics-based solutions, developing customized inspection and sensing systems for R&D and production. Passionate about advancing quantum sensing applications and integrating novel technologies to meet industry needs.

 

Get new IZAK guides and tools by email

One short email a month: new technical articles, free calculators and product updates in laser measurement and electro-optics. No spam, unsubscribe anytime.

Get new IZAK guides and free calculators by email, one short email a month: Subscribe · Privacy Policy
Skip to content