Key points
- Track every calibrated instrument in one register with an ID, tolerance, interval, due date and status, not in a folder of PDFs.
- Read every certificate before filing it: a certificate for the wrong serial number, or one without readings taken before adjustment, proves very little.
- An instrument found out of tolerance triggers a look back at everything it measured since its last good calibration.
- Set intervals from as-found history: lengthen them when readings stay well inside tolerance, shorten them when they drift out.
A service report and a calibration certificate are not the same record
Teams often keep both in the same folder and treat them the same way, and that causes trouble at audit time. A service report says what was done to a machine: filters changed, belts adjusted, a seal replaced. It is a record of work. A calibration certificate says how far an instrument's readings were from a reference standard, and how sure the lab is about that comparison. It is a record of measurement. A pressure gauge can be cleaned and serviced without anyone checking that it reads correctly, and it can be calibrated without anyone touching it.
Plenty of equipment needs both. A torque wrench gets serviced when its mechanism is cleaned or a part is replaced, and calibrated to confirm it still clicks at the set torque. A weighing scale gets serviced when a load cell is replaced, and that repair is exactly why it needs calibrating again before anyone uses it. So the rule is: any repair or adjustment that could affect a reading starts a new calibration. The register should make that link obvious.
Build the register first, file the certificates second
A folder of PDFs answers "do we have a certificate?" but not "what is due next month?" or "which instruments are we using right now that we should not be?". You need a register with one line per instrument, and the documents attached to those lines. These are the fields that do real work:
| Field | Why it matters |
|---|---|
| Your own ID, on a label on the instrument | Serial numbers can be hard to read, can repeat between brands, or can be missing. Your ID is what people on the floor will actually look for. |
| Manufacturer serial number | This is what the lab prints on the certificate. You need it to match the certificate to the right instrument. |
| Range and tolerance you need | Pass or fail depends on your process, not on the brochure. A gauge used from 0 to 10 bar to within 0.2 bar needs those limits written down. |
| Interval and basis | Record why the interval is what it is (manufacturer, standard, history). Otherwise nobody will dare change it later. |
| Last calibration date and due date | Count the due date from the calibration date on the certificate, not from the day the instrument came back. |
| Location and owner | An overdue instrument that nobody can find is the usual reason a recall falls apart. |
| Status | In use, due soon, overdue, at the lab, quarantined, limited use, retired. One value, always current. |
| As-found result of each calibration | Pass or fail before adjustment. This history is what you use later to set the interval. |
Check every certificate before you file it
Most problems with calibration records come from certificates that were accepted without anyone reading them. Five minutes with this list, at the moment the instrument comes back, prevents most of them:
- Serial number and model match the instrument in your hand. Labs handle hundreds of similar gauges, so mix-ups happen.
- Readings before adjustment (as found) are reported, not just the readings after adjustment (as left). A certificate that only shows the instrument passing after adjustment cannot tell you whether your past measurements were good.
- The test points cover the range you actually use. A gauge used at 6 to 8 bar, checked only at 1, 2 and 3 bar, says nothing about your working range.
- The measurement uncertainty is stated and is small compared with your tolerance. A common rule of thumb is that the uncertainty should be no more than about a quarter of the tolerance. If the uncertainty is close to your tolerance, a pass is not worth much.
- Traceability is shown: the reference standards used, their own certificate numbers, and ideally a lab accredited to ISO/IEC 17025 for that type of measurement. Accreditation applies to a scope, so check that the measurement you sent is inside it.
- You make the pass or fail call against your own tolerance. If the lab judged against the manufacturer's specification and yours is tighter, an instrument the lab passed may still fail for you.
Only after these checks do you update the register: new calibration date, new due date, as-found result, status back to in use, certificate attached. If any check fails, the instrument stays quarantined until it is resolved.
When an instrument comes back out of tolerance
This is where a calibration program either proves its worth or turns out to be paperwork. If the as-found reading was outside your tolerance, the lab will usually adjust the instrument and send it back passing. The instrument is fixed, but the problem is in the past: every measurement made with it since its last good calibration is now in doubt. Work through it in this order:
- Set the window. It runs from the date of the last calibration that passed as found up to today. You usually cannot tell when inside that window the drift started, so assume the whole window is suspect.
- Find what it measured. This is why each work order, inspection or checklist that records a reading should also record which instrument took it. Without that link this step becomes guesswork.
- Size the error against each use. A torque wrench reading 3 percent high, used on joints with a 10 percent tolerance, may not matter. The same wrench on a joint with a 2 percent tolerance does.
- Decide and act. Re-check, re-torque or re-test what matters, raise work orders for it, and record a reasoned "no action" for what does not.
- Shorten the interval for that instrument, and ask why it drifted: dropped, overloaded, used outside its range, or simply worn.
Set intervals from history, not habit
Twelve months is the default interval almost everywhere, mostly because it is easy to remember. Start there or with the manufacturer's advice, then let the as-found results move it. An instrument that comes back well within tolerance several times in a row is telling you the interval is cautious. One that comes back near the limit or outside it is telling you the interval is too long.
| Example: torque wrench, tolerance ±4% | As-found error | Decision |
|---|---|---|
| Year 1, 12 month interval | +1.0% | Keep 12 months |
| Year 2 | +1.5% | Keep 12 months, drift is steady |
| Year 3 | +3.6% | Near the limit: shorten to 6 months, check for wear |
| Next two checks at 6 months | +1.2%, +1.8% | Stable again after service: return to 12 months |
The figures above are an illustration, not data from a real wrench. The point is the method: change the interval for a recorded reason, and write the reason in the register. Never lengthen an interval just to cut calibration costs while the as-found history is short or mixed. Keep high-risk instruments on cautious intervals, the ones whose readings decide whether something is safe to release.
Between calibrations, simple intermediate checks catch drift early and shrink the window you would have to review. Examples: weighing a known check weight on a scale every week, or comparing two gauges on the same line each month. Record them as a short checklist against the instrument. A failed check sends the instrument to quarantine straight away.
Keeping it running week to week
- Review a 30 day look-ahead every week and send instruments out in batches, so you are not stripping one area of all its gauges at once.
- Label the instrument itself with its ID and due date. The person using it at 2 a.m. will not open the register first.
- Give quarantine a physical place, a red bin or a locked shelf, so a failed or overdue instrument cannot quietly go back into use.
- Use limited-use status honestly. A gauge that is fine at 0 to 5 bar but out of tolerance above that can stay in service with a label saying so, and that limit belongs in the register.
- Keep records for retired instruments as long as you keep the records of the work they measured, so a question about old work can still be answered.
Any tool that holds one line per instrument, attaches documents, calculates due dates and links readings to the instrument that took them will do the job, including a well-kept spreadsheet for a small fleet. In Solmex, calibrated tools and equipment sit in the register with their certificates and due dates, so calibrations can be scheduled like any other preventive task and every reading stays linked to the instrument that took it.