A link is flapping. Receive power is 4 dB down on where it was commissioned. Someone swaps the transceiver, and it makes no difference. Someone swaps it again, for a different brand this time, and it still makes no difference. Two hours and three modules later, someone cleans the connector — and the link comes straight back.
This is the single most common fault in fibre networks, and it is also the cheapest to prevent. A dust cap is not a cleaning tool, a factory-fresh patch lead is not automatically clean, and the human eye cannot see a particle that will happily destroy a 10G link.
Why a speck of dust matters this much
The core of a single-mode fibre is about 9 microns across. A typical airborne dust particle is somewhere between 2 and 15 microns. Put one on the end face and it is not a small obstruction on a large surface — it is a boulder sitting on the whole road.
Multimode fibre has a 50 micron core and is correspondingly more forgiving, but not immune — and as the TIA Fiber Optics Tech Consortium notes, the tolerances tighten every time the grade moves up. The fibre-type distinction matters here for the same reasons set out in our guide to single-mode vs multimode fibre. Contamination causes three separate problems at once:
- Insertion loss. Light that hits a particle is absorbed or scattered instead of crossing the joint. A single contaminated mating pair can eat several decibels.
- Back reflection. The air gap created by a particle reflects light back towards the transmitter, which degrades the signal and, on some links, destabilises the laser.
- Permanent damage. This is the one people underestimate. Mating a contaminated connector grinds the particle between two ceramic ferrules under spring pressure. The particle transfers to the other end face, and often leaves a pit or scratch behind. You have now contaminated the clean side and damaged both — and no amount of cleaning afterwards removes a pit.
That last point is why the discipline is inspect-then-clean-then-inspect, not clean-then-plug. Every mating of a dirty connector makes the problem worse and spreads it.
The standard that defines “clean enough”
“Looks fine” is not a pass criterion. IEC 61300-3-35 — now in its third edition, published in 2022 — defines the visual inspection of fibre optic connectors and fibre-stub transceivers, and it exists precisely so that two engineers looking at the same end face reach the same verdict.
It works by dividing the end face into concentric zones and applying different tolerances to each:
| Zone | What it covers | Tolerance |
|---|---|---|
| A — Core | The light-carrying region | Effectively zero. Nothing of any size is acceptable on single-mode. |
| B — Cladding | The glass surrounding the core | Tight. A limited number of small defects only. |
| C — Adhesive | The epoxy ring | Loose. Cosmetic issues here rarely affect performance. |
| D — Contact | The outer ferrule surface that mates | Loose, but large debris here can hold the ferrules apart. |
The criteria vary by connector type — single-mode is stricter than multimode, and angled polished connectors have their own rules — and modern inspection scopes apply the grading automatically, returning a straight pass or fail. That automation is the point: it removes judgement from a task performed at 2am by someone who has been awake for nineteen hours.
A cleaning process that actually works
- Inspect first. Always, including on brand-new patch leads. Factory end faces pick up contamination from dust caps, packaging and handling, and “new” is not a synonym for “clean”.
- Clean dry first. A cassette cleaner or a one-click pen removes most loose contamination without introducing anything. Dry cleaning is the default because it cannot leave residue.
- Escalate to wet-dry only if dry fails. Oils, gels and fingerprints need solvent. Use a fibre-specific cleaning fluid on a lint-free wipe, then immediately dry-clean the same spot — solvent left to evaporate on its own leaves a residue ring that is worse than what you started with.
- Inspect again. Confirm the pass before the connector goes anywhere near a port.
- Cap it if it is not going straight in. A clean, capped connector stays clean. An uncapped one on a bench does not.
Two things that should never appear anywhere near an end face: compressed air, which drives contamination into the ferrule and can deposit propellant, and isopropyl alcohol from a general workshop bottle, which is rarely pure enough and leaves residue as it dries.
Do not forget the port
Half of all contamination lives on the side nobody inspects. The optical bulkhead inside a transceiver, the adapter in a patch panel, and the far side of any coupler are all mating surfaces, and they get dirty exactly like a patch lead does.
A transceiver that has been sitting in an unpopulated port without a dust plug will have collected debris on its internal ferrule. Plug a pristine patch lead into it and you have contaminated the patch lead. Cleaning sticks sized for the bore of SFP, QSFP and OSFP ports exist for exactly this, and an inspection scope with a long-reach tip lets you check the result rather than hoping.
MPO makes all of this harder
Parallel optics — the SR8 and DR4 modules behind 400G breakout — use MPO connectors carrying 12, 16 or 24 fibres in a single ferrule. Every one of those fibres has to pass inspection, and one contaminated fibre in a row of twelve takes down one lane of a multi-lane link. The symptom is not a dead port; it is a port that comes up and then runs with an error rate that nobody can account for.
MPO end faces also need a scope capable of scanning the whole array, not a single-fibre tip, and most MPO in data centres is APC-polished, which has its own inspection geometry. Add polarity to the list of things to get right before a parallel link will pass traffic at all — our guide to MPO polarity types A, B and C covers that side.
Where the budget goes
Contamination is not a category of its own in a link budget calculation. It shows up as insertion loss at a mating pair, and it competes for the same margin as fibre attenuation, splices and connectors — margin that was already tight on the day the link was designed. Our guide to optical link budget sets out the arithmetic, and it is worth doing before you conclude a transceiver is underpowered for the span.
This matters most on the links with least headroom: long-reach ER and ZR spans, CWDM and DWDM plans where a passive mux has already spent several decibels, and anything running multimode near its distance limit.
Catching it before someone notices
Every transceiver reports its own receive power. Record it at commissioning, trend it, and alarm on drift rather than waiting for a threshold breach. A connector that has picked up contamination shows as a step change in received power on one link while its neighbours are unchanged — which is about as clear a diagnostic signature as fibre gives you. Our guide to spotting a failing fibre link before it drops covers which values to trend and where to set thresholds.
It is also the fastest way to answer the question that started this article. If receive power fell and transmit power at the far end did not, the problem is in the path — not the module. Swapping transceivers will not fix it, and the two hours are better spent with an inspection scope.
A short checklist
- Inspect every end face before every mating, new leads included.
- Dry-clean first; escalate to wet-dry only when dry fails.
- Clean both sides — the port and the bulkhead, not just the lead.
- Never use compressed air or workshop-grade alcohol.
- Cap anything that is not going straight into a port.
- Record receive power at commissioning so you have a baseline to compare against.
- On parallel links, scan the whole MPO array, not one fibre.
When it really is the optic
Sometimes it is. Lasers age, and a module that has been running hot for years will eventually lose output — which is why every Carritech Optics module is tested before it ships rather than being taken on trust. Our article on the 22 checks a module passes before it reaches you sets out what that covers, and there is more on our transceiver testing page.
Carritech Optics supplies the full range of compatible optical transceivers from 155M to 1.6T, along with patch cords, DAC e AOC assemblies, all coded and tested for the platform they are going into and backed by a garantia vitalícia with UK and EU stock behind it.
Have OEM part numbers in hand? Our compatibility checker returns the tested Carritech equivalent in seconds. Or tell us what the link is doing and we will help you work out whether it is the fibre, the connector or the module — request a quote.
