Dark fibre: you characterise it before service, or not at all
7 min read
You are about to sign a dark fibre contract, or to accept a newly laid run. In the specification, or in the acceptance report you have been asked to countersign, what did you actually ask to be measured — and against which numbers? If the clause says “tested to the standard” and stops there, you have bought a test whose tier you don’t know. On a run destined to become dark fibre, that tier decides whether, three years from now, you still have a figure to consult or only a signed PDF.
One recommendation, two tiers, not a single list
The reference is Recommendation ITU-T G.650.3 (08/2017), “Test methods for installed single-mode optical fibre cable links,” Study Group 15, approved on 13 August 2017, edition 3.0 (unique ID 11.1002/1000/13289; earlier editions 2007, 2008 and a 2011 amendment). The Scope states it from the first line, verbatim: “This Recommendation uses a tiered approach. The first level indicates measurements that are normally carried out to commission new optical fibre cable links. The second level indicates measurements that may be carried out to satisfy service level agreements (e.g., when a dark fibre contract is signed) or to verify attributes of older links that may be used at higher bit rates or over extended wavelength ranges.”
The difference between the two tiers isn’t a matter of technical nuance, it’s commercial: the first proves the run works and was installed to a proper standard of workmanship; the second proves it is fit to carry whatever you put on it in future, and to be sold as dark fibre without surprises for whoever buys it.
Tier 1: the new run works
Clause 6.1 states the purpose: “proving that the components used (fibres, connectors, etc.) meet their specifications and that the installation has been carried out to a good standard of workmanship.” Then comes the sentence that carries the whole specification: “The criteria for newly installed optical fibre links are defined by the operating company.” The Recommendation says which measurements to take; the numbers — per-splice threshold, per-connector limit, overall attenuation — are written by the client, combining attenuation coefficient, section lengths and splice count “according to the formulas of Appendix I of [ITU-T G.652].” A specification that points to G.650.3 without its own numbers has cited a designation, not specified anything.
Before any measurement, clause 6.1.1 puts connector end-face inspection first — “it is recommended that the end face of the connectors at each end of the link be inspected to ensure that they are clean and free from any damage” — and after cleaning, “the best practice is to re-inspect to ensure cleaning effectiveness”: you clean, then you verify the cleaning worked. Acceptance criteria for defects and scratches sit in IEC 61300-3-35, as we already covered in our note on connector face inspection.
For attenuation, clause 6.1.2 recommends the light source and power meter (LSPM) technique per IEC 61280-4-2, “the preferred technique for determining the total link attenuation,” typically at 1310 and 1550 nm, and at 1625 nm if the run will carry the L-band. On OTDR, the Recommendation warns: “The use of OTDR can shorten the measurement time, but also has risks. […] the OTDR can, for example, miss a broken fibre near the end of the link, unless tail cords are used. It can also allow crossed-over fibres to go undetected. […] When the results of such a check are marginal, follow-up with bidirectional measurement is recommended.” We have already written about reading this in the report and about choosing dead zones — see reading an OTDR report and OTDR dead zones.
Clause 6.1.3 sets the practice: OTDR “in both directions on every fibre using at least two wavelengths” — 1310 and 1550 nm for G.652-type fibre, 1550 and 1625 nm for G.655-type fibre — with “a launch lead that is long enough (typically 1 km to 2 km) […], a similar length tail lead should also be connected at the far end.” Without leads of that length, the connectors at both ends stay outside the measurement.
Tier 2: fibre fit to be sold as dark
Clause 6.2 introduces “fibre characterization” with the sentence that carries the whole article: “Often it is important to carry out full fibre characterization, particularly on a dark fibre contract, because once the fibres are brought into service it may not be possible to access the fibres again to assess their suitability for further upgrades in the future.” That is the argument for doing it before, not after: whoever accepts a run on tier 1 alone, then sells it as dark fibre, has bought a test that will not answer the question their own client asks a few years later.
Tier 2 includes all of tier 1, plus chromatic dispersion (CD) and polarization mode dispersion (PMD), and extends attenuation testing to further wavelengths — 1625 nm OTDR, or spectral attenuation across the O-, E-, S-, C- and L-bands for CWDM, measured “only […] in one direction” (clause 6.2.4). On PMD (clause 6.2.5): below 10 Gbit/s or on short runs it “may not be required”; where it is, “just one direction” suffices, and for a composite run the total PMD “is the square root of the sum of the squares” of the individual PMD values — added in quadrature, not simply summed. On CD (clause 6.2.6): not essential at low bit rate or short distance, but “useful for identifying the type of fibre installed,” with a precise conditional note: “If the cable section is less than 40 km long to comprise an optical link used for 2.5 Gbit/s transmissions, CD does not need to be tested. This measurement is conditional with respect to the planned bit rate, transmission distance and fibre type.”
Where it doesn’t apply, and why that’s a distinction worth stating
The Scope draws the boundary: the methods “are not intended for application to links that contain optical network elements, amplifiers, dispersion compensators, passive splitters or combiners.” On a PON run with a splitter, it doesn’t apply as it stands — it applies to the point-to-point run, not through the passive splitter, which we covered in our note on insertion loss on the ODN.
And, 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 tier and the numbers — the same logic we already met over the alarm threshold in optical monitoring, where L.391 leaves the same decision to the client.
See the service · Talk to an engineer
The two axes, applied to this test
First axis: the specification doesn’t just say “tested to G.650.3.” It states which tier, and with which numbers: wavelengths per direction, launch and tail leads of 1-2 km, per-splice and per-connector thresholds, an overall limit calculated with the Appendix I formulas of G.652, connector end-face inspection to IEC 61300-3-35 criteria before measurement — and, if the run may become dark fibre or carry 10 Gbit/s and above, CD and PMD measured before it enters service, not after, with dated, documented proof to show at acceptance.
Second axis: test traces don’t stay a PDF filed at the end of the job. Together with CSIDIA, the other company in the group, they become part of a single network map on which an AI does the diagnosis and the crew closes the fault: the commissioning OTDR trace is the reference against which tomorrow’s fault trace is read, and without it every later measurement is a figure with nothing to compare it to — the same principle behind the as-built register of runs. Within the client’s own perimeter: on-premise, on autonomous machines that need no deep integration into the existing network, or a dedicated cloud with a data centre in Italy, always with shared management.
Could you say, run by run, which tier of testing was actually carried out at acceptance — or whether, three years from now, you’ll have to ask for access to fibres already in service? Talk to an engineer: the site visit is at no cost, and out of it comes the list of your runs with, for each, the tier actually carried out, the wavelengths and directions measured, and whether CD and PMD were measured before entering service — blank boxes included. It stays yours even if we do not go on together.