What Is Calibration?

Calibration is the process of comparing an instrument’s readings against a known reference standard and adjusting it so its measurements stay within a stated tolerance. Every measuring device drifts over time - torque wrenches, gas detectors, scales, thermometers, pressure gauges - and calibration is the routine that catches the drift before it produces bad readings, failed work, or unsafe decisions. It sits alongside the inspection schedule as one of the recurring obligations attached to an asset.

What you will learn

• Why instruments drift and why the problem is invisible at the point of use
• How calibration works, from as-found readings to certificate
• Types of calibration and what “traceable” actually means
• Calibration vs verification - two words people use interchangeably but shouldn’t
• What certificates and records must contain
• How to set calibration intervals that follow the evidence
• What to do when an instrument is out of tolerance
• How to keep calibration on track in practice

Why instruments drift

Springs fatigue, sensors age, electronics heat and cool, and kit gets knocked about in vans and toolboxes. None of this announces itself: a torque wrench that reads low still clicks, a gas detector that has lost sensitivity still switches on, a scale that has crept off zero still shows a confident number. Drift is invisible at the point of use, which is exactly why calibration runs on a schedule rather than waiting for symptoms. By the time a drifting instrument produces a visibly wrong result, it has usually been producing quietly wrong results for a while.

How calibration works

A calibration compares the instrument against a reference standard whose own accuracy is traceable back up a chain to a national measurement standard. The typical sequence:

• As-found readings are taken first, recording how the instrument performed before anyone touched it.
• The instrument is adjusted (or assigned a correction factor) to bring it back within tolerance.
• As-left readings confirm the result.
• Everything is documented in a calibration certificate, with the next due date.

The as-found readings carry the real weight. If the instrument arrived out of tolerance, everything it measured since the last calibration is now in question - which can mean re-checking completed work, not just fixing the tool.

Types of calibration

Not every calibration is the same level of rigour, and the words on the certificate matter when an auditor reads it:

• Traceable calibration means the standard used can be traced, through an unbroken chain of comparisons, back to a national or international measurement standard (such as the SI units). This is the baseline expectation for most work.
• Accredited calibration (for example, to ISO/IEC 17025) adds an independent statement that the lab is technically competent and impartial, and usually reports the measurement uncertainty at each test point. It is the premium option and is often required in regulated industries.
• In-house calibration uses your own reference standards - themselves calibrated by an accredited lab - to check working instruments between external calibrations. It is cheaper and faster but only as trustworthy as the standard behind it.

Intervals can also be defined in different ways: by time (every 12 or 24 months), by usage (after a number of tests or operating hours), by event (after a repair, a shock, or a suspected knock), or by regulatory requirement. Most teams use time as the default and layer the others on top.

Calibration vs verification

These two words get used interchangeably, but they are not the same thing. Verification confirms that an instrument’s readings still fall inside acceptable limits - a pass or fail check where nothing is adjusted. Calibration compares the instrument against a traceable standard, records the actual error, and brings the instrument back into tolerance. A quick verification between scheduled calibrations acts as an early warning: if an instrument fails verification, you pull it and calibrate early rather than waiting for the due date. Neither replaces the other. Verification is cheap reassurance; calibration is the documented reset. The related idea of confirming a physical item is present and correct - rather than accurate - belongs to asset verification.

Certificates and records

A calibration certificate states the instrument’s identity (usually its serial number), the standard used and its traceability, the as-found and as-left results against tolerance, the date, who performed the work, and when the next calibration falls due. Customers, accreditation bodies, and auditors ask for these, and they ask per instrument: “show me the current certificate for this exact detector”. A folder of PDFs named scan0041.pdf does not survive that conversation, so the certificate needs to live with the asset’s record. The point of a certificate is not the piece of paper - it is being able to answer, at any moment, whether a given instrument was in tolerance on the day it took a particular measurement.

Setting calibration intervals

The manufacturer’s recommended interval is the starting point, not the answer. Tighten it for instruments that are heavily used, transported daily, or have a history of arriving out of tolerance; consider stretching it for stable instruments with a clean record. Criticality trumps convenience - kit used for safety decisions, like safety equipment and gas detection, deserves the conservative interval. Annual calibration is the most common default across industries, but the number should follow the evidence: an instrument that always passes can earn a longer interval, and one that keeps failing has already told you its interval is too long. A lapsed calibration has a habit of surfacing at the worst moment, turning into downtime when an instrument is pulled from service mid-job or fails an audit.

When an instrument is out of tolerance

An out-of-tolerance result is not just a maintenance problem - it is a data problem that reaches backward in time. Because the as-found reading captures the instrument’s state before adjustment, a fail tells you the instrument has probably been reading wrong for some unknown stretch. The disciplined response has three parts: fix the instrument, bound the affected period using the calibration history, and decide - and record - whether any of the work done in that window needs re-checking or a customer needs to be told. Teams that keep a clean per-instrument history can draw that boundary confidently; teams that don’t end up either re-doing far too much work or, worse, shipping suspect measurements they never flagged. This backward reach is the strongest argument for treating each measuring instrument, such as your test instruments, as its own tracked asset with a permanent record.

Calibration in practice

The common failure is not bad calibration work - it is the instrument nobody realised was due. The habits that prevent it: every measuring instrument gets its own asset record, the next due date is recorded as a field on that record rather than in someone’s head, the current certificate is attached alongside it, and out-of-service instruments are clearly marked so nobody grabs one from the shelf. In AMPthilly, teams add a calibration-due custom field to the asset record, attach each certificate as a document, and log the work through the service desk so the instrument’s full history stays on its record permanently.

Frequently asked questions

How often should equipment be calibrated? Start with the manufacturer’s recommended interval, then adjust based on evidence. An instrument that is used daily, travels in a van, or keeps arriving at calibration out of tolerance needs a shorter interval; one that sits in a controlled lab and always passes can often stretch its interval. Criticality matters too - anything used for safety decisions or customer-facing measurements should err on the side of more frequent checks. Annual is the most common default, but nothing about “annual” is magic; the interval should follow the instrument’s history.

What is a calibration certificate? The document issued after a calibration. It identifies the instrument (usually by serial number), names the reference standard used and its traceability, records the as-found and as-left readings against the stated tolerance, and gives the date, the person or lab responsible, and the next due date. Auditors and customers ask for certificates, so they need to be findable per instrument - not buried in a shared inbox.

What is the difference between calibration and inspection? An inspection checks condition and function - is the guard fitted, does it switch on, is the cable damaged. Calibration checks measurement accuracy - does the reading match a known standard within tolerance. A gas detector can pass a visual inspection while reading dangerously low, which is why measuring instruments carry both obligations, usually on different intervals.

What is the difference between calibration and verification? Verification confirms that an instrument’s readings still fall inside acceptable limits; it is a pass or fail check and nothing is changed. Calibration goes further - it compares the instrument against a traceable standard, records the error, and adjusts the instrument (or assigns a correction factor) to bring it back into tolerance. In practice teams often verify between calibrations as a cheap early warning, then calibrate on the scheduled interval.

What happens if an instrument fails calibration? The as-found reading tells you the instrument was out of tolerance before it was touched, so every measurement it took since the last good calibration is now suspect. You adjust or repair the instrument, but you also have to look backward - identify the work done in that window, decide whether it needs re-checking, and record the decision. This is exactly why a per-instrument calibration history matters: it lets you bound the affected period instead of guessing.

Related terms

• Inspection Schedule - the planned calendar of recurring checks that calibration dates slot into
• Downtime - what a lapsed calibration causes when an instrument is pulled from service
• Maintenance Backlog - the queue overdue calibrations quietly join
• CMMS - the software category built around scheduling recurring maintenance work
• Asset Uptime - the share of time an asset is actually available for use