What a cal lab tracks that an inspection firm never has to

The instrument register is the spine of the system

A calibration laboratory's database is organised around instruments it does not own. Each record carries the customer, the make and model, the serial number, range and resolution, accessories that arrived with it, and the accumulated history of every visit. That history is the product as much as the certificate is: a customer deciding whether to keep a pressure transducer in critical service wants four years of as-found data, not the current pass mark.

Receiving is where data quality is won or lost. Intake captures condition on arrival, accessories, the customer's requested scope, the turnaround promised, and any damage noted before work begins. A duplicate record created at intake because a serial number was typed differently splits an instrument's history permanently, and no later report will show the drift.

As-found and as-left are distinct records, always. As-found is the condition the instrument arrived in and is what the customer needs for their own impact assessment when it fails. As-left is the condition it leaves in after adjustment. Systems that store one set of readings and overwrite it on adjustment destroy the only evidence the customer can act on.

Traceability: the chain has to be reconstructable, not asserted

ISO/IEC 17025 clause 6.5 requires metrological traceability through a documented, unbroken chain of calibrations to the SI, each contributing to measurement uncertainty. In software terms that means the lab's reference standards are first-class records, not attachments. Each holds its issuing laboratory and accreditation number, the certificate, the stated uncertainty, the due date, and its own upstream standard.

Every customer calibration must record which specific standard, by serial number, was used for it — not the model, the individual unit. That single link is what makes the chain reconstructable years later. It is also what makes an out-of-tolerance cascade a query rather than an archaeological project, which is the subject of the next section.

Standards carry due dates and intermediate checks of their own under clause 6.4.5, and a lab whose reference standard silently passes its due date has invalidated every calibration performed with it since. The due-date engine therefore has to treat reference standards with higher severity than customer instruments: a hard block on use, not a reminder.

Due dates, recall, and the scheduling the customer sees

Two due-date engines run in parallel and they must not be conflated. The internal one governs the lab's own reference standards and equipment, and it blocks work. The external one governs customer instruments and drives recall notices, quotes and inbound scheduling — a revenue engine as much as a compliance one, because a lab that reminds customers thirty days out retains work that otherwise goes to whoever calls first.

Intervals are not uniform. They come from the customer's own requirements, from the manufacturer's recommendation, from regulatory obligation, and from observed drift in the instrument's own history. A system that supports interval adjustment based on that history — extending stable instruments, shortening drifters — gives the lab a technical argument to have with its customers rather than a default it cannot defend.

Capacity is the other half of scheduling. Due dates cluster, customers ship in batches, and a lab that accepts everything arriving in one week misses turnaround commitments across the board. Scheduling that shows technician availability against promised dates is the same problem inspection firms solve in crew dispatch and scheduling, applied to a bench instead of a site.

Certificate production under clause 7.8

The certificate is the deliverable and the legal artefact. Clause 7.8.4 requires calibration certificates to carry the measurement results with their uncertainty, the conditions under which the measurements were made, and evidence of metrological traceability. Where results come from a customer-supplied item, its identification and condition are recorded. Where a statement of conformity appears, clause 7.8.6 requires the decision rule to be documented and communicated.

Because the certificate is generated from records rather than typed, the system must pull the standard used, its traceability, the technician's authorisation, the environmental conditions logged during the work, the as-found and as-left data, and the uncertainty budget for the relevant measurement points. Every one of those is already in the database. A lab that retypes them into a template introduces transcription errors into a document customers rely on for their own audits.

Certificates need revision control identical to inspection reports: numbered, immutable once issued, reissued as a new revision with reason recorded and the superseded version retained. Distribution needs to be recorded too, which is where a client portal replaces the emailed PDF and its unanswerable question about which version the customer holds.

Uncertainty budgets and decision rules as live data

Clause 7.6 requires the laboratory to evaluate measurement uncertainty for every calibration. In practice that is a budget per parameter and measurement point: the reference standard's uncertainty, resolution, repeatability, drift since the standard's last calibration, environmental contributions, and any correction applied — each with a distribution and a sensitivity coefficient, combined and expanded, conventionally at k=2 for approximately 95% coverage.

Held as spreadsheets, budgets rot. A reference standard is replaced with one of different uncertainty and forty budgets need updating; in reality three get updated. Held as data referencing the standard record, the budget recalculates when the standard's certificate changes, and every certificate issued afterwards carries the correct expanded uncertainty without anyone remembering to act.

Decision rules sit on top. Simple acceptance ignores uncertainty; guard banding subtracts it from the tolerance before judging; a risk-based rule targets a maximum probability of false accept. US laboratories working to ANSI/NCSL Z540.3-2006 face an explicit limit — §5.3(b) caps false-accept probability at 2%, with a test uncertainty ratio of at least 4:1 accepted as an alternative demonstration. The applicable rule differs by customer, so store it on the account and print it on the certificate.

Accreditation records are a second, parallel system

Accreditation adds a whole record set that has nothing to do with any individual calibration. The scope document defines which parameters, ranges and capabilities the lab may certify — work outside scope may be performed but cannot carry the accreditation mark, and software that lets a technician select an out-of-scope parameter without warning will eventually produce a certificate the assessor removes.

Around that sit proficiency testing and interlaboratory comparison participation under clause 7.7, personnel competence and authorisation under 6.2, internal audits, management reviews, complaints, nonconforming work and corrective actions. Assessors arrive on a cycle and ask for evidence across all of it. Labs that keep these in shared drives spend the fortnight before an assessment reconstructing them, which is the same pattern that drives audit preparation software demand on the inspection side.

Personnel authorisation deserves specific attention because it gates work. The matrix is parameter by equipment by person, with recorded evidence — training, witnessed performance, ongoing monitoring — and an expiry. Enforced at job assignment, it prevents the finding an assessor writes when a certificate is signed by someone whose authorisation for that parameter was never recorded.

Where an inspection firm's needs stop, and what Atlantis covers

An inspection firm running UT sets, MT yokes, PT kits and thickness gauges needs the register, the due dates, the certificate store, and a hard block preventing an out-of-calibration instrument from being used on a job or cited in a report. It does not need uncertainty budgets, CMC scopes, decision rules or proficiency testing — its calibrations are bought, not performed. That is the scope of equipment calibration tracking.

A lab that both calibrates for customers and inspects needs both models running side by side over one database, so an instrument in the internal fleet and an instrument in for customer calibration follow different workflows without being separate systems. Job costing across the bench and the field then works from one set of records — the subject of inspection job costing.

The Atlantis platform is Odoo-based, deployable cloud or on-premise, and configurable — which is what a calibration operation needs, because uncertainty budgets, certificate layouts, decision rules and scope structures are specific to each lab and cannot be delivered as fixed screens. Affordable. Accessible. Fully customizable. For a walkthrough against your own scope of accreditation, request a demo through contact, and for the compliance record set generally see compliance tracking.

Request a consultation