On a high-power fibre, you do not just plug in the OTDR

In the specification you wrote “commissioning and maintenance with OTDR,” the same line you use for any other run. But does that backbone carry erbium-doped fibre amplifiers, or distributed Raman pumping? Above a certain power, connecting the instrument the usual way is not just an inaccurate measurement: it can irreversibly damage what you are measuring, and put whoever is doing it at risk.

A few hundred milliwatts

The reference is Recommendation ITU-T L.312 (01/2024), “Optical fibre cable maintenance support, monitoring and testing system for optical fibre cable networks carrying high total optical power,” Study Group 15, edition 2.0 approved on 13 January 2024, unique ID 11.1002/1000/15809 (edition 1.0, L.312/L.68, dates from 22 October 2007, Study Group 6). The Scope sets the boundary: it applies to “test equipment, optical switches for selecting fibre under test, test access modules […], testing optical fibre cords and optical connecting devices” that are part of the maintenance system, not the whole plant.

Clause 6 sets the threshold: “When high-power light is launched into an optical fibre cable maintenance system, the fibre-optic components in the system may be damaged in such a way that they no longer meet their specifications (e.g., optical loss).” And it quantifies it: “the term “high-power radiation” is used to refer to optical powers of several hundred milliwatts” — a few hundred milliwatts, not the watts you picture when you think of a dangerous laser: a low threshold, easy to cross without realising it on a run with amplification.

Two safety issues, not one

Clause 7 states it in the sentence that is the heart of the Recommendation: “There are two safety issues as regards optical fibre cable carrying high total optical power. One is human safety, and this problem relates to the exposure of eyes or skin to high-power radiation. The other is component safety. […] this may pose an overheating and fire-hazard in a worst case scenario.” Two distinct risks: the person handling the cable, and the component doing the measuring — which, with a larger-than-usual optical loss, can heat up to the point of becoming a fire hazard.

Clause 6 turns this into a requirement: the system “must be safe for network operators when handling optical cables, cords and fibre-optic components,” in accordance with ITU-T G.664, IEC 60825-1 and IEC 60825-2 — the same standards covered in our note on laser safety on fibre networks. Here the risk widens to the component under test as well, not just the source upstream.

A table that is already a specification

Table 7-1 lists four functions, the requirement and the method — and reads like a specification already written:

Connection: requirement “no fibre fuse or intense temperature increase”; method “use of fusion splices instead of optical connectors which require polishing and cleaning of connector endfaces.”
Termination: requirement “no tight optical fibre bends”; method “minimum bending radius R ≥ 30 mm for testing optical fibre cords in optical distribution frame (ODF),” “however fibres with improved bending capability will allow more severe operating conditions.”
Testing access for optical fibre line: same requirement as connection; method “use of fusion splices. Use optical branching components with high tolerance to high optical power exposure.”
Optical switch with butt-joint splice connection mechanism (e.g., fibre selector): requirement “no intense temperature increase or optical loss increase”; method attenuating the high-power light, or a “gap between fibres at butt-joint splice d < 10 μm.”

Clause 7.1 sums it up: to prevent “fibre fuse,” “it is recommended not to use connectors especially near the output of a high-power optical source. Instead, fusion splices should be employed.” And it recommends materials that do not easily induce a temperature rise for splice sleeves, “with a view to avoid overheating and fire hazards.”

See the service · Talk to an engineer

The order of operations, not a list

Clause 8 opens with auto-shutdown: “An auto-shutdown function is an effective way to prevent damage caused by accidents […]. Testing and maintenance should be undertaken in optical fibre cable maintenance systems after the auto-shutdown function has operated” — you work after the shutdown has triggered, not before. Four procedures follow, in sequence, not interchangeable:

“power monitoring using low power test lights should be performed in central offices”;
“reflection test and fibre end inspection should be carried out to check the endface near the output of a high-power optical source”;
“after confirming that there are no large loss points […], low-power optical loss testing or OTDR testing should be carried out to detect fault locations”;
“optical loss testing or OTDR testing using a high optical power should be performed” — high power is the last step, not the first.

Whoever plugs in the high-power OTDR as the first move has reversed a sequence that protects, one step at a time, what comes after.

Appendix I: the cord that never recovers

Appendix I “does not form an integral part of this Recommendation” — it says so explicitly — but it is the experimental proof of what the rest of the text prescribes to avoid. The context: DRA (distributed Raman amplification) and remotely pumped EDFAs, in WDM systems, launch high-power light into fibres and devices — the same scheme behind our note on DWDM and CWDM grids.

The experiment: sources at 1480 nm, maximum power 2 W; conventional single-mode tight-buffer cords, 1.1 mm and 1.7 mm in diameter (0.5 mm and 0.9 mm jackets); launched powers of 24, 27, 30, 31.7 and 33 dBm for 30 minutes; cords wound in 10 loops 10 mm in diameter; temperature measured with infrared thermography. The result, verbatim: “After this test at 33 dBm, the loss of the optical cords did not recover their initial loss. This is because the temperature increase caused catastrophic damage to the front few turns” — after the test at 33 dBm the loss never returned to its starting value, because the heat caused catastrophic damage to the front turns: melting of part of the 0.9 mm primary coating in the 1.7 mm cord; in the 1.1 mm cord, air bubbles that exploded between the fibre and the primary coating.

Like every ITU-T Recommendation: “compliance with this Recommendation is voluntary”: it gains force only when a client writes it by name into a specification, with the numbers from table 7-1 and the sequence from clause 8 — not the designation alone.

The two axes, applied to this maintenance

First axis: the specification distinguishes high-power runs from the others and states, for those, what changes — fusion splices near the source output, a minimum 30 mm bend radius for test cords in the ODF, branching components declared tolerant to high power, a tested auto-shutdown function, and the measurement sequence with low power before high. With dated, documented proof at acceptance: not “we measured,” but “we measured in this order, after verifying the automatic shutdown.”

Second axis: knowing which runs carry high optical power — where the amplifiers are, where the pumping is, which cords in which frame — is not something to look up on the day of the fault. Traces, measurements and as-built records become, with CSIDIA, the other company in the group, a single network map on which an AI does the diagnosis and the crew closes the fault: whoever arrives at night needs to know before opening the frame whether that fibre is dark or carrying two watts — the same principle behind the as-built register of runs. Within the client’s own perimeter: on-premise, on autonomous machines with no deep integration, or a dedicated cloud with a data centre in Italy, always with shared management.

Could you say, right now, which of your runs carry more than a few hundred milliwatts of optical power — and whether the written procedure knows it? Talk to an engineer: the site visit is at no cost, and out of it comes the list of your runs with, for each, whether it carries high optical power and from which equipment, whether the written maintenance procedure requires auto-shutdown and the low-before-high sequence, and who on the crew knows it before opening the frame — blank boxes included. It stays yours even if we do not go on together.