Calibration
Instrument calibration: how often, and how to set the right interval
There is no single correct interval for instrument calibration. Start from the maker's recommendation, then shorten it for heavy use, harsh conditions or tight tolerances. After every calibration, look at how far the instrument had drifted, and lengthen or shorten the next interval based on that evidence rather than habit.
How often should instrument calibration be done?
Most plants answer this with one word: yearly. It is easy to plan and easy to explain. It is also a guess. A pressure gauge on a hydraulic press that pulses all day does not drift at the same rate as a micrometer kept in a padded box in the inspection room.
The joint guidance from ILAC and OIML on recalibration intervals, ILAC-G24 / OIML D 10, says this plainly. There is no single interval that suits all instruments. Fixed intervals that are never reviewed are not recommended, unless the interval is set by a normative document such as a standard method. The guidance expects you to choose a starting interval with care, then adjust it from the calibration results you collect.
Your quality standard asks for the same thing in different words. ISO 9001 and IATF 16949 ask that measuring equipment is calibrated or verified at specified intervals, or before use, against standards traceable to national or international standards. They leave the length of the interval to you. What an auditor checks is whether you can explain why you chose it, and whether you act when an instrument comes back out of tolerance.
So the honest answer to "how often" is: often enough that the instrument is still inside its tolerance when you next check it, and no more often than that. The rest of this post shows how to get there without a statistician.
Set the first interval: a checklist to copy
Use this when a new instrument arrives, or when you clean up an old list where everything is yearly. The factors come from the ILAC-G24 / OIML D 10 guidance, put into plant language. Fill the last column for each instrument or group of similar instruments.
| Factor | Question to ask | Shorten the interval when | Your answer (fill in) |
|---|---|---|---|
| Maker's recommendation | What does the manual or datasheet say? | There is none, or your use is harder than the maker assumed | |
| Required accuracy | How close is the instrument's accuracy to the tolerance you measure against? | There is little margin between the two | |
| What depends on the reading | Does a product release, a safety trip or a customer characteristic depend on it? | A wrong reading would ship bad parts or hide a hazard | |
| Extent and severity of use | How many hours a day, how many cycles? | It runs every shift, or sees pulsation and shock | |
| Environment | Heat, dust, vibration, moisture? | It sits near a furnace, press or washer | |
| Measured quantity | Does the quantity itself wear the sensor? | Thermocouples at high temperature, gauges at the top of their range | |
| History | What did past calibrations and repairs show? | It was found out of tolerance or repaired before | |
| Intermediate checks | Is there a quick check between calibrations? | There is no check at all between calibrations | |
| Handling and transport | Is it carried around, dropped, sent out by courier? | It moves between stations or goes to an outside lab |
New instruments deserve a shorter first interval. The guidance suggests calibrating new equipment more often at first, so you learn how it drifts before you trust a longer gap.
Review the interval after every calibration
The simplest review method in the ILAC-G24 / OIML D 10 guidance is called the staircase. After each routine calibration, you look at the as-found result, the reading before any adjustment. If the error was well inside the tolerance, you keep or extend the next interval. If it was outside, you shorten it. The guidance leaves the exact cut-off to you, and its 2007 edition gave 80 percent of the maximum permissible error as an example. You can pick a stricter cut-off, such as half the tolerance, as the rule set below does.
This only works if the lab records the as-found reading. A certificate that shows only the reading after adjustment tells you the instrument is fine today. It tells you nothing about how far it drifted, so ask your lab for both.
Two other methods from the guidance are useful on the shop floor. One counts hours of use instead of months, which suits thermocouples in a furnace or gauges that wear with each cycle. The other relies on frequent quick checks, for example a daily check of a thermometer against a reference, with a full calibration only when the check fails. Control charts and other statistical methods also exist, but they need more data and effort than most plant teams have to start with.
Here is a rule set you can copy into your calibration procedure. The steps match common recurrences: quarterly, half-yearly and yearly.
- Found within half of the tolerance, twice in a row: extend one step, for example quarterly to half-yearly.
- Found between half and the full tolerance: keep the interval. Note the trend.
- Found out of tolerance: shorten one step, start an impact review of the readings taken since the last good calibration, and record the cause.
- Repaired, heavily adjusted, dropped or moved to harsher duty: treat it as new and go back to the shortest step.
- Never extend past the maker's or the method's limit, and never extend in big jumps. A long interval that fails means a long window of readings to review.
Worked example: a machining plant in Chakan
Worked example (illustrative)
Take an auto component machining plant in Chakan, near Pune. It has 142 instruments under calibration, all on a yearly interval because that is how the list was first set up. All numbers below are illustrative.
In the last yearly round, 9 of the 60 pressure gauges on the hydraulic presses and 3 of the 12 heat treatment thermocouples came back out of tolerance. Each one meant an impact review covering up to a year of production, with the quality team going through parts records to decide what to recheck. The micrometers and the utility temperature indicators all came back well inside tolerance.
The maintenance and quality heads group the instruments by type and duty and apply the rule set above:
| Group | Count | Last yearly round | New interval | Calibrations a year |
|---|---|---|---|---|
| Press pressure gauges | 60 | 9 out of tolerance | Quarterly, review after two rounds | 240 |
| Heat treatment thermocouples | 12 | 3 out of tolerance | Quarterly, plus a monthly check against a reference thermocouple | 48 |
| Inspection micrometers | 40 | All within a third of tolerance | Yearly, plus a weekly check on the setting standard | 40 |
| Utility temperature indicators | 30 | All well inside | Yearly | 30 |
| Total | 142 | 12 out of tolerance | 358, up from 142 |
What the example shows (illustrative). The calibration count goes up at first, which is the price of finding out how the gauges really behave. If the press gauges then pass two quarterly rounds within half their tolerance, they step up to half-yearly and the total falls to 238. The bigger gain is the size of each impact review: a gauge that fails now puts at most three months of readings in doubt, not twelve. The utility indicators stay yearly, and the plant can later ask whether any product decision depends on them at all.
Common mistakes
- One interval for everything. Yearly for all instruments is easy to schedule and hard to defend when a press gauge keeps failing.
- Extending because the lab is expensive. Cost matters, but an interval should move on calibration results, not on the budget. A failed gauge after a long gap can cost more in rechecks than the calibrations you saved.
- Reading only the after-adjustment value. Without the as-found reading you cannot tell drift from stability, so you cannot review the interval.
- No impact review when an instrument fails. An out-of-tolerance result is a question about every reading since the last good calibration. Record what you checked and what you decided.
- Counting from the wrong date. The interval runs from the date the calibration was done, not from the date the certificate arrived or was filed.
- Forgetting instruments fitted to machines. Gauges, thermocouples and load cells on equipment are easy to miss because they do not sit in the calibration cupboard. List them against the machine.
- Changing the interval without a record. If you shorten or extend an interval, write down the result that justified it. That note is what the auditor will ask for.
Sources
- ILAC-G24 / OIML D 10:2022, Guidelines for the determination of recalibration intervals of measuring equipment (opens in a new tab), OIML and ILAC
- ILAC guidance series (lists ILAC G24:2022) (opens in a new tab), International Laboratory Accreditation Cooperation
- National Accreditation Board for Testing and Calibration Laboratories (opens in a new tab), NABL India
Frequently asked questions
How often should instruments be calibrated?
It depends on the instrument, how hard it is used, its environment and how much accuracy you need. Start from the maker's recommendation, then adjust the interval from the as-found results of each calibration.
Is a yearly calibration interval acceptable to auditors?
It can be, if you can show why it suits that instrument and you act when one comes back out of tolerance. A yearly interval applied to everything without review is hard to defend.
What is the staircase method for calibration intervals?
After each calibration you look at the as-found error. If it is well inside the tolerance you keep or extend the next interval, and if it is outside you shorten it, one step at a time.
What should I do when an instrument is found out of tolerance?
Shorten its interval, review the readings taken with it since the last good calibration, decide what needs rechecking, record the cause, and fix or replace the instrument.
Does a NABL accredited lab set my calibration interval?
No. The lab calibrates the instrument and issues the certificate. Choosing and reviewing the interval is your responsibility, based on how you use the instrument.
Keep reading
Calibration management software
Schedules for machines and fitted instruments, with uploaded certificates and NABL details.
ReadCalibration traceability
The evidence chain auditors check first, from your gauge back to a national standard.
ReadNABL calibration records
What to keep from an Indian lab's calibration certificate, and what to do when a result fails.
Read