Should You Calibrate Before or After Maintenance? The Right Answer for Each Scenario

The as-found / as-left principle

As-found calibration documents the instrument's condition before anyone touches it. As-left calibration documents its condition after any intervention. Together, they create a before-and-after record that supports quality investigations, return-to-service decisions, and customer or regulatory enquiries.

In a calibration laboratory setting, both are recorded on the same certificate: the as-found section shows what the lab found when they received the instrument; the as-left section shows what the instrument reads after any adjustments made within the calibration process itself. This dual record is best practice under ISO/IEC 17025 and is expected by quality auditors for instruments that have been found out-of-tolerance or have undergone repair.

The same logic applies to maintenance carried out in your own facility. Before your maintenance team touches an instrument, a calibration result (or at minimum a pre-maintenance measurement record) captures its state. After maintenance, a post-maintenance calibration verifies that the instrument has been returned to correct operation. Without both records, a gap exists in your instrument's history, and that gap can become a finding during an ISO 9001 or GMP audit.

Maintenance scenario decision table

The following table maps common maintenance scenarios to the calibration action required and the reason behind it. Use it as a reference when building or reviewing your calibration and maintenance procedures.

Why as-found data matters

If an instrument is found out-of-tolerance (OOT) during calibration or after a failure, the as-found data answers three questions that every quality investigation needs:

• The magnitude of the error. How far out was it, and by how much?
• The direction. Was it reading high or low, and on which ranges?
• The scope. Was the drift uniform across all ranges, or confined to specific measurement points?

This data is the input to your product impact assessment. The documented evaluation of whether measurements taken during the period the instrument was out-of-tolerance may have affected product quality, release decisions, or compliance records. An instrument with a +0.5% DC voltage error has a very different impact profile than one with a −5% error across all ranges. Without as-found data, you cannot answer the question: "Were my measurements likely affected?" You can only record that the instrument is now OOT, which leaves the investigation open.

Most SAC-SINGLAS accredited calibration labs record as-found data as standard. Confirm that your lab does this before submitting instruments, and if they do not, it is worth considering whether their certificates will fully support an OOT investigation when one is needed.

Routine preventive maintenance

Cleaning, lubrication, battery replacement, strap and case replacement, and firmware updates that do not touch measurement circuitry typically do not require recalibration in isolation. The practical test to apply in each case: "Could this maintenance have changed the instrument's measurement accuracy?"

If the answer is no (the maintenance only affected non-measurement components), recalibration is not required, provided the instrument remains within its calibration interval. The maintenance activity should be documented in the instrument record: what was done, who did it, and when. That record demonstrates to an auditor that the maintenance was controlled and that a judgement was made about whether recalibration was necessary.

Apply judgement at the component level. Replacing a worn probe tip on a standard multimeter (the probe tip is an accessory that simply completes the circuit. It is not part of the measurement transduction) does not require recalibration. Replacing the probe body on a precision contact thermometer (where the probe body houses the thermocouple junction that IS the temperature sensor), does require recalibration, because the new probe body is a new measurement element whose calibration characteristic may differ from the original.

When in doubt, consult the instrument's service manual. Most manufacturers specify which maintenance activities require post-maintenance verification and which do not.

Corrective maintenance (repair after failure)

Any instrument repaired after a measurement failure, display fault, or physical damage must be calibrated after repair before return to service. The repair is assumed to have altered the measurement chain until calibration proves otherwise. This is not overly conservative. Repair work on measurement circuitry, replacement of PCBs or sensor elements, and re-soldering of connections all introduce variables that only a calibration result can resolve.

Additionally: calibrate before the repair (as-found) to document what state the instrument was in. The as-found certificate created before repair may be needed for:

• Warranty claims to the manufacturer. Evidence of what failed, not just a description of the symptom
• OOT investigation for the period prior to failure. How long was the instrument in service while faulty?
• Customer dispute resolution. If the instrument was used to verify a product or process during the period of the fault

The sequence for corrective maintenance is therefore fixed: calibrate → repair → recalibrate → return to service. No shortcut in this sequence reduces risk; each of the three steps serves a distinct purpose.

Calibration during planned maintenance shutdowns

Singapore manufacturers frequently schedule instrument calibration during planned maintenance shutdowns. Chinese New Year, Hari Raya, or annual plant shutdowns. This approach efficiently combines calibration with any maintenance that was due in the same window. Best practice for shutdown calibration management:

1. Calibrate instruments before maintenance starts, as-found data captures the pre-shutdown condition across all instruments in the batch.
2. Perform the planned maintenance activities.
3. Recalibrate any instrument whose maintenance touched the measurement chain (sensors, references, measurement PCBs). This is the as-left confirmation that maintenance did not introduce error.
4. For instruments where maintenance only involved non-measurement items (cleaning, battery, housing), the pre-maintenance calibration may serve as the renewal calibration for that interval. Provided it was performed within the calibration schedule, the instrument was found in-tolerance (as-found), and the maintenance did not involve measurement components.

This approach minimises the total number of calibration events while maintaining full traceability for each instrument's state before and after the shutdown period.

Firmware and software updates

Modern digital instruments (spectrum analysers, power quality analysers, multifunction process calibrators, digital pressure gauges), receive periodic firmware updates from their manufacturers. These updates can range from user interface refinements to changes in the core measurement engine. The calibration question is straightforward once you know what the update changes.

When a firmware update affects a measurement algorithm, conversion formula, linearisation table, or compensation function (anything that changes how the instrument converts a raw signal into a displayed result), the instrument must be calibrated after the update. The update has altered the measurement chain, and the previous calibration certificate no longer describes how the instrument currently performs.

When a firmware update only affects the user interface, connectivity features, data logging, communication protocols, or non-measurement functions, calibration is not required. The measurement chain is unchanged.

The reliable source for this determination is the manufacturer's firmware release notes. Most instrument manufacturers explicitly state whether a firmware update affects measurement performance. Review the release notes before applying any update to a calibrated instrument, and document the outcome in the instrument record: firmware version applied, date, whether the update affects measurement function, and calibration action taken (or justification for no action).

Environmental changes

Instruments are calibrated in specified environmental conditions. Typically temperature and humidity ranges controlled to the calibration laboratory's specifications. A calibration interval is set on the assumption that the instrument will be used and stored in conditions reasonably similar to those during calibration. Moving an instrument to a significantly different environment can invalidate that assumption.

Moving an instrument from an air-conditioned metrology room or quality lab (23°C ±2°C, 45–65% RH) to a production floor (30–35°C, 70–80% RH) may cause the instrument to drift faster than the original calibration interval was set for. The calibration interval was not established for those conditions.

For instruments relocated to a significantly different environment:

• Calibrate before relocation. This establishes a baseline in the original environment.
• Allow four to six weeks stabilisation in the new environment before calibrating again. This allows thermally and hygroscopically responsive components to reach a new equilibrium.
• Calibrate in the new environment (or transport to the lab and calibrate). This confirms whether the environmental change has affected accuracy and by how much.
• Adjust the calibration interval accordingly if the drift rate in the new environment is higher than was expected based on historical performance.

The goal is not bureaucratic compliance. It is ensuring that the instrument's calibration interval remains appropriate for the conditions it is actually operating in, rather than the conditions it was originally characterised in.

Practical implementation for Singapore manufacturers

The simplest calibration-maintenance sequencing policy that satisfies ISO 9001 auditors can be written in four statements and referenced from the instrument control section of your QMS:

1. Corrective maintenance (repair after failure or damage): calibration before maintenance (as-found) and calibration after maintenance (as-left) before return to service. Both events are mandatory with no exceptions.
2. Scheduled preventive maintenance that does not involve measurement components: document what was done and retain the maintenance work order. No additional calibration is required if the instrument is within its calibration interval and the maintenance did not involve measurement-critical components.
3. Annual calibration: as-found data is always captured by the calibration laboratory; the as-left result is the final calibration result. Both appear on the certificate.
4. Environmental changes and firmware updates: flag in the instrument record; trigger a calibration review against the criteria above before deciding whether additional calibration is needed.

Write this policy into a procedure, reference it from the instrument control procedure, and train the maintenance team on the two scenarios that always require dual calibration: corrective maintenance and drop/damage events. Those are the situations where the temptation to skip the pre-maintenance calibration (to "save time") is highest, and where the missing as-found data causes the most problems later during an OOT investigation or audit.

Common mistakes we see in Singapore instrument records

Reviewing instrument histories for clients moving to Unitest from another calibration provider, a small number of recurring gaps account for most of the audit findings we see. The first is maintenance teams treating "the instrument still works" as equivalent to "the instrument is still accurate." A repaired unit that powers on, displays a stable number, and appears to function normally can still be measuring incorrectly; functional recovery and metrological recovery are different things, and only a calibration confirms the second. The second common gap is a maintenance work order that describes what was replaced ("sensor board replaced") without a corresponding note on whether that component is measurement-critical, which forces whoever reviews the record months later (often during an audit, under time pressure) to guess at whether recalibration should have been triggered. Writing that determination into the work order at the time of the repair, while the technician who did the work still remembers the detail, is far more defensible than reconstructing it later.

The third and most costly gap is skipping the pre-repair as-found calibration on the assumption that "we already know it's broken, so what's the point." This reasoning misses the actual purpose of as-found data: it is not there to confirm the instrument is faulty (that much is usually obvious), it is there to quantify how faulty, in which direction, and across which ranges, which is exactly the information a product impact assessment needs to determine whether anything measured with that instrument in the weeks before the failure needs to be reviewed. Skipping this step does not remove the need for the investigation; it just removes the data that would have made the investigation fast and conclusive instead of open-ended and defensive.

Deciding when to skip recalibration, and documenting that decision defensibly

Not every maintenance event needs recalibration, and over-calibrating (sending an instrument for a full accredited calibration after every battery change) is its own cost problem, not a safer default. The defensible middle ground is a written, component-level decision rule referenced above (the "could this have changed measurement accuracy?" test), applied consistently and recorded every time, including the times the answer is no. An auditor reviewing your instrument history is not looking for zero maintenance events without a matching calibration; they are looking for evidence that someone made a documented, reasoned judgement each time, rather than the decision being made silently or not at all.

In practice this means the maintenance work order itself should carry a short, standard field: "Measurement-critical component affected? Y/N, reviewed by [name]." For the "no" cases, that single line, cross-referenced to the instrument's current calibration certificate and due date, is sufficient evidence for most ISO 9001 and GMP auditors. For the "yes" cases, that same line becomes the trigger that routes the instrument to Unitest for as-left calibration before it goes back into service, closing the loop without relying on someone remembering to check weeks later when the audit season arrives.

Training the maintenance team to make the right call in the moment

Most gaps in as-found/as-left records are not caused by a bad procedure on paper; they are caused by a maintenance technician under time pressure, mid-shutdown, deciding in the moment that a quick fix does not warrant "the paperwork." Closing this gap is less about writing a stricter policy and more about making the correct action the fastest one. A short, laminated decision card at the calibration bench (the same "measurement-critical Y/N" test, with two or three worked examples specific to your instrument fleet) resolves most hesitation in seconds rather than requiring a technician to interrupt a shutdown to look up a procedure. Pairing this with a standing arrangement with your calibration provider (so a same-day or next-day as-found slot is genuinely available during planned shutdown windows, not a two-week queue) removes the practical excuse for skipping the pre-repair calibration that otherwise tends to erode compliance over time.