Quality Management Software for Calibration Laboratories (ISO/IEC 17025)

A UT thickness gauge goes out on a piping circuit inspection at a Gulf Coast refinery. The tech grabs it off the shelf, runs a two-point check against the site's step wedge, and gets to work. Three days later, back at the shop, someone notices the gauge's own calibration certificate expired the day before the job. Nobody caught it because it was tracked on a spreadsheet, in a tab nobody opened that week. Every reading on that job is now suspect, and if the lab is A2LA or ANAB accredited, dispatching an expired gauge is a nonconformance that has to be logged and root-caused. This is the scenario calibration lab QMS software exists to prevent — and why spreadsheets and calendar reminders stop being adequate once a lab is running under ISO/IEC 17025:2017 with real accreditation on the line.

For an NDT company that operates its own calibration lab — thickness gauges, hardness testers, MT yokes and ammeters, PT black lights and radiometers, RT densitometers and step wedges — this isn't an abstract IT decision. It's the difference between an audit that takes two hours and one that takes two days, and between field crews who can trust the case they grab off the shelf and crews who have to double-check everything because the tracking system is only as current as the last update.

ISO/IEC 17025:2017 Clauses 6-8: What Actually Drives Software Requirements

ISO/IEC 17025:2017 has eight clauses, but three of them are where a QMS platform earns or loses its keep. Clause 4 (impartiality and confidentiality) and Clause 5 (structural requirements) matter for governance; Clauses 6, 7, and 8 are different — each maps almost line for line onto a feature an assessor expects to see demonstrated on screen, not described in a policy manual.

• Clause 6 — Resources. 6.2 requires documented competence for every person who calibrates, reviews, or authorizes results. 6.4 requires equipment — including the lab's own reference standards — uniquely identified, on a defined calibration program, and protected from adjustment that would invalidate results. 6.5 is metrological traceability: every result links, through an unbroken chain of calibrations, back to the SI unit via a national metrology institute or an ILAC MRA-recognized lab.
• Clause 7 — Process. The longest clause and the one with the most software touchpoints: 7.1 review of requests, 7.2 method selection and verification, 7.4 handling of calibration items (receipt, unique ID, condition on arrival, storage), 7.5 technical records, 7.6 evaluation of measurement uncertainty, 7.7 ensuring validity of results (proficiency testing, interlaboratory comparisons, control charts), 7.8 reporting of results, and 7.10 nonconforming work.
• Clause 8 — Management system. Document control, control of records, corrective action, internal audit, and management review — the same system of record an assessor reviews for document revision history and CAPA closure, not a calibration tracker bolted onto the side of the business.

Every one of those sub-clauses generates a specific artifact — a competency record, a traceability chain, an uncertainty budget, a technical record, a CAPA entry — and assessors are trained to ask for the artifact, not the policy statement describing it.

Metrological Traceability to NIST: The Chain Your System Has to Preserve

Traceability is the most assessor-tested concept in the standard, and the one spreadsheets handle worst because it's relational, not tabular. Every certificate a lab issues has to show the reference standard used — a gauge block set, a multifunction calibrator — was itself calibrated by a source with a documented, unbroken chain back to the National Institute of Standards and Technology (NIST) or another recognized national metrology institute. That chain isn't a single link: a working standard traces to a reference standard, which traces to a calibration from an accredited external lab, whose own standards trace to NIST — each hop with its own certificate number, date, and uncertainty figure that has to be retrievable.

In a spreadsheet system, that chain lives as a scanned PDF filed under an asset number, cross-referenced (if you're lucky) in a column someone typed a cert number into by hand. When an assessor pulls a certificate issued eighteen months ago and asks for the traceability chain behind it, someone walks backward through folders manually. In a properly built QMS, that chain is a data relationship: the certificate links to the reference standard's asset record, which links to its own current and superseding certificates, which link to the accredited supplier's scope of accreditation — a lookup, not an investigation, and one of the fastest ways to keep a routine surveillance visit from becoming a suspended scope.

Calibration Intervals: Fixed Schedules vs. Reliability-Based Adjustment

Most labs start with fixed intervals — annual for a UT thickness gauge, six months for a black light radiometer — whatever the manufacturer or a generic industry table suggests. That's a legitimate starting point, but the standard (via Clause 6.4 and supporting guidance in ILAC-G24 / OIML D 10) expects a documented methodology for adjusting intervals based on actual performance, not habit: "automatic adjustment" (extend or shorten by a fixed ratio depending on whether the last calibration was in or out of tolerance), the "control chart" method (track drift and set the interval where projected drift approaches the tolerance limit), and "in-service time" methods for equipment with variable usage.

For equipment that sees hard field use — an MT yoke dropped on scaffolding, a UT gauge living in a truck through Houston summers — a fixed 12-month interval might be dangerously long if drift data shows out-of-tolerance trending by month eight. A rarely-used reference hardness tester in a climate-controlled lab might legitimately go to 24 months with data behind it, saving real turnaround time. The catch: reliability-based adjustment only works if someone tracks as-found/as-left data every cycle and runs the numbers — longitudinal analysis a spreadsheet resists (formulas break, a cell gets overwritten, the "master" file forks) and a database-backed QMS handles natively. Software that flags equipment off a static due-date is doing the bare minimum; software that shows a drift trend across cycles is doing what 17025 actually asks for.

Measurement Uncertainty Budgets: Why Spreadsheets Break Down at Scale

Clause 7.8.3.1(c) is unambiguous in a way the testing side of the standard isn't: a calibration certificate has to state measurement uncertainty whenever it's relevant to the result's validity or application — for calibration, essentially always. That figure isn't pulled from a spec sheet. Under GUM (Guide to the Expression of Uncertainty in Measurement) methodology, it's a budget: reference standard uncertainty, resolution of the unit under test, environmental drift, operator repeatability, and the reference standard's own drift since its last calibration, combined in quadrature and expanded by a coverage factor (typically k=2, ~95% confidence).

One uncertainty budget for one gauge type is a manageable spreadsheet. The problem is scale: a lab supporting UT, MT, PT, and RT equipment might run budgets across fifteen or twenty instrument types. Every time a reference standard is recalibrated and its stated uncertainty shifts, every downstream budget citing it should technically be revisited — a propagation that doesn't happen automatically in a spreadsheet, because nobody remembers which of thirty tabs reference that standard. That's the stale-data gap an assessor probes by asking when a budget was last reviewed against the standard's current calibration. A calibration-specific QMS that stores uncertainty components as structured data tied to the reference standard's live record can flag every affected budget the moment the parent certificate changes.

What A2LA, ANAB, and PJLA Assessors Actually Pull

US-based calibration labs accredit through one of three ILAC MRA-signatory bodies: A2LA (American Association for Laboratory Accreditation), ANAB (ANSI National Accreditation Board), or PJLA (Perry Johnson Laboratory Accreditation). What gets pulled during a surveillance or reassessment visit is consistent across all three:

• Scope of accreditation control — is every issued certificate within the lab's accredited scope (parameters, ranges, best measurement capability), with no quiet drift outside it on a one-off job.
• Proficiency testing and ILC records — Clause 7.7 requires ongoing validity monitoring via PT/interlaboratory comparisons: the provider's report, the lab's z-score or En-number evaluation, and corrective action for any out-of-range result.
• Technical record completeness — for a randomly selected certificate, can the lab produce raw as-found/as-left readings, logged environmental conditions, the technician's competency record, the reference standard's traceability chain, and the uncertainty budget, all cross-referenced.
• Nonconforming work handling — Clause 7.10 requires a documented process for identifying, containing, and dispositioning nonconforming calibrations, including client notification, with actual logged instances, not just the procedure.
• Corrective action closure — evidence that findings from the last audit cycle were root-caused and closed, not just opened.

A lab walking into a reassessment on spreadsheets and a filing cabinet spends the assessment finding things; a lab with a real QMS spends it showing things, which changes both duration and tone.

The Three-Day Gauge Problem: Due-Dates Meet Field Dispatch

Calibration due-date tracking sounds like a back-office concern until it collides with field scheduling, which is where it actually creates risk. An NDT company running UT surveys, MT/PT crack detection, and RT film work needs job dispatch and calibration status living in one system, not two that someone reconciles manually — the failure mode being the scenario at the top of this post: a gauge with three days left goes out on a five-day job and comes back non-compliant, or a radiometer due next week gets grabbed because it's the only one on the shelf.

The risk compounds under API 653, API 510, and ASME code work, where a client's QA specification may require proof of current, traceable calibration for every instrument used — not just at report time, but when the reading was taken. If calibration lapses mid-job, every reading after the lapse is technically unsupportable, which can mean re-inspection, a client complaint, or an integrity decision made on data that can't be defended. This is why due-date tracking belongs inside the same platform that handles scheduling: a scheduler assigning a tech should get an automatic warning if the equipment is inside a defined buffer window of its due date, not find out after the fact. Platforms like Atlantis NDT ERP are built around exactly this integration — calibration status, scheduling, and technician records in one place, so a gauge three days from due can't be assigned to a five-day job without someone consciously overriding a flag, itself a documented decision instead of a silent gap.

Personnel Competency Records: 17025 Is Not the Same File as SNT-TC-1A

This trips up a lot of NDT companies running their own cal labs, because both records answer "is this person qualified," and it's tempting to file them together. They're not the same thing, and an assessor treats them as two separate lines of evidence.

ASNT SNT-TC-1A is a recommended practice for qualifying and certifying NDT personnel — technicians performing UT, RT, MT, PT, ET, VT, or phased array inspections in the field — through documented training hours, experience hours, and a written/practical/general exam under an employer's written procedure. It answers whether a technician is qualified to perform and interpret a given NDT method. ISO/IEC 17025 Clause 6.2, by contrast, concerns the competence of laboratory personnel to perform calibration work specifically — operating calibration equipment, evaluating results, calculating uncertainty, authoring or authorizing a certificate — assessed through defined competence criteria, training records, and often witnessed calibrations, and it applies even to a technician whose SNT-TC-1A qualification has nothing to do with calibration bench work.

A technician can hold a current SNT-TC-1A Level II in UT and still not be competent to calibrate under 17025 if never trained and signed off on the lab's specific procedures and uncertainty methodology — and calibration competence under Clause 6.2 has no bearing on field inspection qualification. A QMS needs to keep these as genuinely separate records — separate criteria, evidence, and renewal triggers — rather than a single "certified" flag on an employee profile. Companies building out both sides often lean on structured NDT training and certification and outside ASNT Level III consulting to keep the two documented correctly and distinct from day one, since conflating them is a common finding on first-time accreditation.

Spreadsheets vs. Standalone Trackers vs. an ERP-Integrated QMS Module

In practice, a calibration lab chooses between three tiers of tooling, each with a real tradeoff profile:

• Spreadsheets and shared folders. Zero cost and infinitely flexible — and exactly why they're dangerous at scale: no enforced relationship between a certificate, its uncertainty budget, and its traceability chain, no automatic due-date flagging tied to dispatch, no audit trail on who changed what. Workable for a one- or two-gauge shop; a real liability past a few dozen assets.
• Standalone calibration-management software. Purpose-built trackers handle the calibration-specific work well — due-date automation, certificate generation, uncertainty budget templates, traceability linking — a real improvement over spreadsheets. The gap is integration: they typically sit outside the system handling job dispatch, so someone still has to manually check a separate system before sending a tech out, and the three-day gauge scenario slips through at the handoff.
• An ERP with a built-in calibration/QMS module. The advantage isn't necessarily deeper calibration features than a dedicated tool — it's that calibration status, technical records, and uncertainty data live in the same database as job scheduling and technician competency, so due-date enforcement happens automatically at dispatch instead of a manual cross-check. For an NDT company where the cal lab supports field operations, that integration usually matters more than deeper standalone features, because it closes the exact gap where nonconformances happen.

The real test of any of these three shows up during a reassessment audit's evidence pull. A typical A2LA or ANAB reassessment has the assessor select a handful of issued certificates at random and ask the lab to reconstruct everything behind them, live: the technical record, the technician's competency file at the time of the work, the reference standard's own current traceability chain, the uncertainty budget with its component values, and confirmation the reference standard was itself current on the job date. In a spreadsheet environment that reconstruction can take twenty minutes per certificate; multiplied across four or five sampled certificates, that's a half-day assessment becoming a two-day one. In an integrated system it's a handful of lookups, because the relationships were built into the data model from the start — and assessors read that speed as a proxy for whether the Clause 8 management system actually functions day to day. For labs also tracking asset condition alongside calibration, a digital twin platform extends that same connected-data logic to the physical equipment itself.

None of this replaces sound calibration technique or a well-written quality manual — software doesn't calibrate the gauge or write the uncertainty methodology. What it does is make sure the paper trail 17025 requires actually exists, stays connected, and can be produced on demand — the difference between a technically sound calibration program and one that survives contact with an assessor.

Atlantis NDT works with calibration labs and NDT companies building or tightening their ISO/IEC 17025 quality systems, from traceability architecture to the software that ties calibration status directly into field dispatch.

Atlantis NDT Products & Services

Atlantis NDT pairs field expertise with software: NDT inspection management software — Atlantis ERP, a digital twin platform for asset integrity, and NDT reporting software. Build your team with NDT training & certification (ASNT SNT-TC-1A) and ASNT certification pathways, or bring in ASNT Level III consulting. Affordable, accessible, fully customizable — book a free consultation.