Equipment calibration process flowchart (as-found to return to service)
Equipment calibration process flowchart: recall the instrument, take the as-found reading, calibrate against a traceable standard, assess the impact of an out-of-tolerance result, then certify, label and return it.
How it works
Rename the lanes to your own roles
Replace Instrument user, Calibration coordinator, Calibration technician, External calibration laboratory and Quality manager with the roles you really have. On a small site the coordinator and the technician are usually the same person, so merge those lanes rather than drawing a handover that never happens. Keep the quality manager separate from the technician even then: the person who calibrated the instrument should not be the only person deciding what an out-of-tolerance result means for product. If everything goes to a third party, delete the technician lane and let the laboratory lane carry the calibration steps.
Define what is on the register, and what is not
Calibration applies to equipment whose readings you rely on to accept or reject something, not to every device on site. Decide which instruments are calibrated, which are checked but classed as indication only and labelled as such, and which are outside the system entirely. Give every calibrated item a unique identification number that stays with it for life, and record its location, its owner, its tolerance in use, its interval and its last certificate. Anything not on the register is, in practice, uncontrolled — and the first audit question is usually how the boundary was drawn.
Write the decision rule behind the pass
"As-found reading within tolerance?" is inert until someone states how the call is made. Take the tolerance from the job the instrument actually does — the drawing, the control plan, the test method — rather than the manufacturer's headline specification, which is usually tighter than anything you rely on. Then check the reference is comfortably better than the instrument under test; four times better is the working rule, and where you cannot manage it, say so and set a guard band that fails the borderline cases. Write that rule into the procedure, not into a technician's head.
Set the recall and extension authority
Decide how far ahead the due list runs — four to six weeks lets production plan around a recall — who chases an instrument that has not come back, and how long an item may run past its due date before use is suspended. Name who may authorise an extension, what has to be recorded when they do, and what evidence justifies it. Then settle the awkward case: treat a missing instrument as out of tolerance from its last successful calibration, so everything it measured since then goes through the same assessment as a failure. It is a strong incentive to keep the location field accurate.
Make the impact assessment executable
"Assess measurements since the last pass" is only possible if you can list what the instrument was used on. Decide now where that comes from — batch records, inspection records, test reports, the work order history — and how far back the search runs by default. Agree who performs the assessment, what evidence is kept, and the threshold at which it escalates to a decision about released product. Write the rule for the three branches of "Is released product affected?", including who signs off a no-impact conclusion, because that is the branch nobody documents.
Fix the certificate contents, then walk it through and publish
List what a certificate must carry before you accept one, in-house or from a laboratory: instrument identification, the reference standard and its own certificate, as-found and as-left values at each point, ambient conditions, uncertainty, date and signatory. Confirm the accreditation scope actually covers your measurement range. Then walk the finished chart through with a technician, an instrument user and the quality manager, correct any step people describe that is not drawn or is drawn but skipped, and publish that revision while keeping the earlier ones so anyone opening it later can see which version they are reading.
Frequently asked questions
What are the steps in an equipment calibration process?
The due date is reached on the calibration register, a recall notice goes to the instrument's user, and the instrument is either withdrawn from service or given a dated, authorised extension. It is booked in, its as-received condition recorded, and then either calibrated in-house against a traceable reference standard or sent to an accredited laboratory. The as-found reading is taken and compared with the tolerance before anything is adjusted. If it is within tolerance the as-left value is recorded and the certificate issued. If it is not, an out-of-tolerance report is raised, every measurement taken since the last successful calibration is assessed, and a decision is made about affected product. The instrument is then adjusted and re-verified, or retired if it will not hold calibration. Finally the certificate and traceability are filed, a label with the next due date applied, the instrument returned to service, the register updated and the interval reviewed against the drift history.
What is the difference between as-found and as-left calibration readings?
The as-found reading is what the instrument indicated when it arrived, before any adjustment, cleaning, battery change or repair. The as-left reading is what it indicates after the work is finished and it is ready to go back into service. They answer different questions. As-left tells you the instrument is fit for the period ahead; as-found is the only record of how it behaved during the period behind, and it is therefore the number the impact assessment depends on. A certificate that reports only as-left values will always look like a pass and tells you nothing about whether the measurements taken over the last year can be relied on. This is why the chart keeps "Compare as-found against the standard" as a separate step before "Adjust and re-verify the instrument", and why it is worth checking that any external laboratory reports both.
What do you do when an instrument is found out of tolerance?
Stop and treat it as a nonconformance before repairing it, because the repair destroys the evidence. Raise an out-of-tolerance report recording the as-found values and the size of the error, then work backwards: identify every product, batch, test or measurement the instrument was used on since its last successful calibration and judge whether the error was large enough to have accepted something that should have been rejected. The answer drives the next decision. Product still in your control is re-inspected or re-tested with a known-good instrument; product already delivered is escalated, and where the risk justifies it the matter leaves this process for a recall or a customer notification. A no-impact conclusion is legitimate but it has to be reasoned and recorded, not assumed. Only then does the instrument get adjusted, re-verified and returned to service, or condemned if it cannot be brought back.
How do you set and review calibration intervals?
The first interval is an estimate — the manufacturer's recommendation, adjusted for how hard the instrument is worked and how much rests on its readings — and the register exists to replace that estimate with evidence. The usual measure is the in-tolerance rate: the proportion of a group of like instruments that come back inside tolerance at calibration. Many programmes set a target of around 90% and shorten the interval for the whole group when it drops below, which is steadier than reacting to one device. Individual instruments still move on their own history: anything arriving out of tolerance should have its interval cut at once, because the interval has already been disproved. Lengthening an interval, though, also lengthens the period you would have to reassess if that instrument later fails, so cheaper calibration is bought with a wider exposure, and that trade is the one worth writing down alongside the change.
How is this different from a preventive maintenance process flowchart?
They share a trigger and almost nothing else. A preventive maintenance process, at /templates/preventive-maintenance-process, takes a machine out of production on an interval or a meter reading, plans the work order, kits the parts, isolates the asset, works a task list and tests it back into service; its output is a serviced machine, and the wider work order cycle including breakdowns is /templates/equipment-maintenance-process. Calibration also starts from a due date on a register, but its output is a certificate and a decision about measurement, not a repair. The consequence of failure is different too: a badly maintained machine breaks down in front of you, while a badly calibrated instrument keeps producing plausible numbers and the damage only appears when someone works backwards from a failed check. That reverse-traceability branch is the part of this chart that has no equivalent in a maintenance flow, and it is the reason to model calibration separately.