OTDR instrument calibration: the certificate the specification forgets

A well-written specification asks for bidirectional OTDR testing, both wavelengths, launch and tail coils, a per-splice threshold — the difference between a tier 1 and a tier 2 test sits right there. The test passes, the report arrives with the traces attached, acceptance is signed off. Three years later, a fault on the run: the team pulls the acceptance trace to compare it against the new one and work out what changed. Before even reading the decibels, though, there is a question no one asked at acceptance: which instrument took that trace, and was it calibrated that day?

No one knows. The report has the signatures, has the native files; it does not have the calibration certificate for the instrument. A measurement is only a number if the instrument that produced it is traceable to a standard: without that certificate, the trace from three years ago is not a reference, it is just a file — the same question we ask when we write about what actually stays verifiable once fibre is installed. It serves neither the contractor nor the client, and least of all on the day of the fault, when it is supposed to serve as the benchmark.

A self-test is not a calibration

The two terms get confused often, and the difference is exactly there. A functional check is the self-test the instrument runs on itself: it confirms the electronics respond, and says nothing about measurement accuracy. Traceable calibration is something else: a comparison, done by an external laboratory, between the instrument and a higher-level reference standard, documented in a certificate with the indication error, the associated uncertainty, and the chain of comparisons that runs, step by step, back to a national standard. An OTDR that passes its self-test on every power-up can still have drifted in a way the self-test never sees.

The two scales that shift every event

Two calibrated quantities matter most for an OTDR. The distance scale depends on the group refractive index set on the instrument — as we discussed when writing about pulse width — and on how accurately it times the light’s round trip. The attenuation scale depends on how it converts the backscattered signal into decibels. An error on either one does not shift a single event: it shifts every event in the report, all at once, and no one notices in a single PDF, because the values still look plausible — just wrong.

Even the recommendation writers point elsewhere

We opened Recommendation ITU-T G.650.1, “Definitions and test methods for linear, deterministic attributes of single-mode fibre and cable”, to verify this ourselves: the edition in force is the ninth, approved on 13 January 2024 (Rec. ITU-T G.650.1 (01/2024), Study Group 15). Clause 6.4.2.1 — the alternative test method based on the backscattering technique, the same physical principle behind an OTDR — states: “Procedures for the calibration of backscattering equipment (for single-mode fibres) are provided in [IEC 61746-1]”. The same cross-reference reappears in Appendix II, on chromatic dispersion uniformity, in a near-identical sentence that drops the parenthesis about single-mode fibres.

The point is not a small one: even the text that defines how a fibre is measured does not specify how the instrument doing the measuring is calibrated — it points to a separate standard. Said with the same honesty we always use: G.650.1 is mainly about factory measurements, not the standard for testing installed network links (that is G.650.3), and conformance with ITU-T Recommendations remains voluntary, not a requirement.

IEC 61746: the standard exists, the text is paywalled

According to the public IEC catalogue, IEC 61746-1:2009, “Calibration of optical time-domain reflectometers (OTDR) – Part 1: OTDR for single mode fibres”, edition 1.0, published on 17 December 2009, 187 pages, Technical Committee 86, in force with a stability date of 2029. A Part 2, from 2010, exists for multimode fibre. The text is paywalled (around CHF 380): we have not read it, and we do not report its procedures. We know it exists, what it covers and who publishes it; we say nothing more.

Accreditation: what an accredited certificate changes

Accredia, Italy’s sole national accreditation body, accredits calibration laboratories against the UNI CEI EN ISO/IEC 17025 standard. An accredited laboratory calibrates by comparing the instrument against a higher-level reference standard, under a system verified by a third party: that is what makes the calibration traceable to a national or international standard, rather than merely declared as such. An accredited certificate states the uncertainty associated with every value; a certificate from a non-accredited laboratory can be entirely honest, but no one outside that laboratory vouches for the chain behind it.

Accredia publishes public databases of accredited bodies and laboratories, updated after every sectoral committee meeting. The search tool for calibration laboratories (LAT) filters by company name, accreditation number, region and, above all, by quantity and instrument: you can verify not just that a supplier is accredited in general, but that it is accredited specifically for optical fibre test equipment, and download the certificate with its validity dates.

The Italian regulatory picture: where it starts, and where it stops

Law No. 273 of 11 August 1991 established Italy’s national calibration system: primary metrological institutes and calibration centres, coordinated by the relevant ministries, with the task — under Article 1 — of ensuring “la riferibilità ai campioni nazionali dei risultati delle misurazioni” (our translation: the traceability of measurement results to national standards). It is the institutional foundation for the metrological traceability that Accredia’s accreditation also rests on. But it is a law that establishes a system, not a contractual obligation: as far as we have read, it does not require that fibre-network acceptance tests use calibrated instruments. That requirement, if you want it, has to be written into the specification — you will not find it anywhere else.

What to write, what to check

In the specification: that instruments used for acceptance measurements are calibrated with a traceable certificate, valid on the date of the measurement, with a stated uncertainty; that the certificate is attached to the test report; that the report states the serial number of the instrument used — the same number that belongs in the network’s as-built register, alongside the traces.

At acceptance, you are not checking a statement, you are checking a document: the certificate exists, it is tied to that serial number, and it is valid on the date the measurements were taken. Three checks, five minutes, before signing — not after.

See the service · Talk to an engineer

What we don’t know

We have not read the text of IEC 61746-1 or 61746-2, because they are paywalled: we know they exist, and what they cover according to the IEC catalogue, not their technical requirements. We have found no European or Italian rule that explicitly requires metrological traceability for contractual acceptance measurements on a fibre network: Law 273/1991 establishes the national calibration system, it does not bind private specifications. ITU-T G.650.1 is a recommendation with test methods, not a regulation. This is not legal or metrological advice: it is a reading of primary sources, with the limits set out above.

The two pillars, applied here

The first pillar is the lines to add to the specification listed above, and at acceptance the check falls on the certificate, not on someone’s word — with the evidence kept and dated in the file, alongside the traces.

The second pillar is what makes that evidence useful three years on. With CSIDIA, the group’s other company, the calibration certificate, the instrument’s serial number and the acceptance trace become part of a single map of the network on which an AI runs the diagnosis and the team closes the fault: comparing the fault trace against the acceptance trace only holds up if you know which instrument, calibrated when, took the first one. Within the client’s own perimeter — on-premises on self-contained machines with no deep integration, or a dedicated cloud with a data centre in Italy — always with shared management.

From the site survey, at no cost, comes the list of your tested runs — for each one, which instrument measured it, whether a calibration certificate tied to that serial number exists, and whether it was valid on the date of the measurement. Including the boxes that stay empty: it’s yours to keep even if we don’t go on to work together.