Calibration and traceability: building a laboratory record that survives an audit

What a calibration certificate must actually contain, how to read an accreditation scope, how to set your own intervals and defend them, and the intermediate checks that catch drift between calibrations.

Three sentences end an accreditation visit badly, and we have heard all three in Bangladeshi laboratories that were otherwise well run. The certificate does not state a measurement uncertainty. The calibration interval is whatever the supplier's service engineer suggested. And nobody checked the instrument between calibrations, so when it was found out of tolerance there was no way to say which of the last twelve months of results were affected. None of these is an equipment problem. All three are record problems, and they are cheaper to fix before the auditor arrives than after.

What a calibration certificate has to say

A calibration certificate is a measurement report. If it does not report measurements, it is a service visit record with a border printed on it. Check every certificate you receive against this list, and reject the ones that fail.

• Unambiguous identification of the item: manufacturer, model, serial number and your own asset identifier. 'Balance, Laboratory 3' identifies nothing.
• The reference standards used, with their own identification and their own traceability. This is the link in the chain, and it is the line most often missing.
• The measured results, with the measurement uncertainty and the coverage factor used. A result without an uncertainty cannot be compared with a tolerance, so it cannot support a pass or fail statement.
• As-found and as-left values. As-left tells you the instrument is right today. As-found tells you whether it was right yesterday, and that is the value you need when you have to assess the results already reported.
• The environmental conditions and the method under which the calibration was performed, because a temperature-sensitive measurement made in different conditions is a different measurement.
• The date, the person and the laboratory, plus the accreditation mark and reference number where the calibration is accredited.
• Any statement of conformity, and the decision rule behind it. If the certificate says 'pass' on a result sitting close to the tolerance limit, it has to say how uncertainty was handled in reaching that verdict.

A related trap at purchase: a manufacturer's factory test report is not the same as a calibration certificate, even when it carries impressive figures. It typically demonstrates that the instrument met the manufacturer's specification on the production line, without traceable reference standards named, without uncertainties, and without an accreditation scope behind it. Ask which one you are being offered before the equipment ships.

Traceability is a chain, and you are entitled to see it

Metrological traceability means your result can be related to a stated reference through an unbroken chain of calibrations, each with its own stated uncertainty, ending at a realisation of the SI unit. Written out, the chain runs from your working instrument, to the working standard used on it, to the reference standard held by the calibration laboratory, to a national metrology institute or another accredited laboratory, and so to the unit itself.

In Bangladesh the practical landmarks are the national metrology laboratory maintained under BSTI, and the Bangladesh Accreditation Board as the national accreditation body. The point that saves the most trouble: an accredited laboratory has a scope document listing exactly which quantities, ranges and uncertainties its accreditation covers. Read the scope, not the logo on the letterhead. A laboratory accredited for mass is not thereby accredited for temperature, and if the parameter you need is outside the scope then the certificate you receive is an ordinary commercial calibration, whatever badge is printed on it.

For parameters that no local laboratory can cover, and there are several in specialised instrumentation, the international mutual recognition arrangement between accreditation bodies is what makes a certificate issued abroad acceptable to a Bangladeshi auditor. Send the item, or arrange for a visiting calibration, and keep the accredited certificate in the file. It is a longer route, so plan it around the audit calendar rather than discovering it three weeks beforehand.

Setting the interval yourself, and defending it

The calibration interval is a risk decision that belongs to the laboratory, not a number the manufacturer hands down. Auditors rarely object to an unusual interval. They object to an interval nobody can explain. Base yours on five inputs: the manufacturer's recommendation as a starting point, the drift you have actually recorded at previous calibrations, how heavily the instrument is used, the consequence of reporting a wrong result from it, and the environment it lives in.

That last input matters more here than most imported quality manuals assume. Humidity through the monsoon months, dust in the dry season, mains disturbance, and instruments carried between sites in the back of a vehicle all accelerate drift. If your as-found data shows an instrument regularly arriving near the edge of tolerance, shorten the interval and write down why. If several cycles show almost no drift, you have earned the right to lengthen it, and the as-found history is your justification.

Instrument	What tends to drift	Sensible check between calibrations
Analytical balance	Sensitivity and corner load, aggravated by vibration, draughts and an unstable bench	Daily single-point check with a class weight at working load; monthly repeatability and corner load
pH meter	Electrode slope and offset; electrodes age quickly in warm storage	Two-point buffer calibration before each session, with the slope recorded and trended rather than merely accepted
Air displacement pipette	Piston and seal wear, worse with aggressive or volatile reagents	Gravimetric check on a calibrated balance at two volumes, monthly, with results plotted
Incubator, oven, refrigerator, freezer	Controller sensor drift, and a spatial gradient the fixed sensor never sees	Independent logger placed at the worst location found during mapping, reviewed weekly against minimum and maximum
Autoclave	Chamber temperature against gauge reading, drain and trap blockage	Independent data logger in a representative load, plus the indicator policy your discipline requires
Pressure gauge or flowmeter on a gas line	Mechanical drift and internal corrosion in humid air	Comparison against a calibrated reference gauge on a written interval
Fixed and portable gas detectors	Loss of sensor sensitivity, and poisoning that leaves no visible trace	Bump test to the written policy; span calibration at the set interval
Timers, thermometers and hygrometers	Slow drift that nobody notices because nobody looks	Annual comparison against a calibrated reference kept solely for that purpose

Intermediate checks: the part that actually protects results

Calibration tells you an instrument was correct on one day. Intermediate checks tell you it has stayed correct since, and they are what convert an out-of-tolerance finding from a catastrophe into an inconvenience. If your last calibration was eleven months ago and the instrument has just failed, the results at risk are eleven months of work. If you have a weekly check standard record, the results at risk are the ones since the last good check.

• Use a check standard: a stable item measured repeatedly under the same conditions. It need not be certified; it needs to be stable and to be treated as sacred.
• Plot the results on a control chart with warning and action limits derived from your own scatter, not copied from a textbook. A number in a logbook is data; a chart is information, and the trend appears on a chart weeks before it breaches a limit.
• Attach the check to an existing routine. A check that requires someone to remember it, find the standard and open a separate record will not survive a busy month. A two-minute check at the start of a run will.
• Cross-check against a second instrument where you have one, and against a second operator periodically, which catches technique drift as well as instrument drift.
• Record the failures, not only the passes. A file with no failures in three years tells an auditor that the checks are not being done honestly, and they will say so.

Building the record

1. Maintain an equipment register with a unique identifier for every item, physically labelled on the item itself and not only present in a spreadsheet. Include receipt date, location, status, and the person responsible.
2. Keep one calendar of calibration due dates, owned by one named person. Distributed responsibility means no responsibility, and the item that gets missed is always the one on the boundary between two departments.
3. Label each instrument with its calibration status: last date, due date, and any restriction on use. Anyone picking the instrument up should not have to open a file to know whether they may use it.
4. Write an out-of-tolerance procedure before you need it. It has to answer the difficult question: what do we do about the results we have already reported? That means defining how far back to look, how to assess the impact, who decides whether customers or clinicians are notified, and how the decision is recorded. This is the procedure auditors probe hardest and the one most laboratories have never written.
5. Define your decision rule for statements of conformity, so that pass and fail verdicts near a limit are made consistently and the treatment of measurement uncertainty is documented rather than improvised.
6. Quarantine and label failed equipment immediately, physically removing it from service. An instrument marked out of service but still sitting on the bench will be used.
7. Retain certificates for the period your quality system requires, indexed so that any certificate can be produced in under a minute. The speed with which you can produce a document changes how the rest of the audit goes.

Buy calibration with the equipment, not after it

Most calibration problems are created at purchase order stage and discovered two years later. Four lines in the bill of quantities prevent nearly all of them.

• An accredited calibration certificate at handover, with the accreditation scope reference stated, and the certificate emailed for review before shipment rather than produced at the end of installation, when your negotiating position has gone.
• The first scheduled recalibration inside the contract, with a named provider and a named date, so the first renewal is not an unbudgeted purchase.
• The reference items needed for your own intermediate checks: class weights, buffer solutions, a reference thermometer, a reference gauge. These are inexpensive against the instrument and impossible to obtain quickly once the audit is scheduled.
• The manufacturer's site preparation document supplied before civil works begin, which sets the environmental and supply conditions the calibration will assume.

One detail specific to importing into Bangladesh, and it catches people every year: the calibration interval runs from the date the calibration was performed, which is at the factory before dispatch. An instrument that spends weeks in transit and further weeks in customs clearance arrives with a meaningful part of its first interval already consumed. If the certificate is dated well before the equipment reaches your bench, plan the first recalibration from the certificate date, not from the installation date, and say so in the register so that the next person understands the dates they are looking at.

Back to all Insights