Intermediate checks: proving equipment stays in tolerance between calibrations

A calibration certificate tells you an instrument was in tolerance on the day it was calibrated. It says nothing about the six or twelve months until the next one. Drift, a knock, wear or a temperature excursion can move an instrument in between — and every measurement you made in that window rides on the assumption that it didn’t. Intermediate checks are how you keep that assumption honest.

What ISO/IEC 17025 asks

Clause 6.4.10 is short: “When intermediate checks are necessary to maintain confidence in the performance of the equipment, these checks shall be carried out according to a procedure.” Two words carry the weight. Necessary — you decide, from risk, how the instrument is used and how it tends to drift, whether checks are needed and how often. Procedure — they can’t be ad-hoc; the what, the how-often, the limits and the reaction all have to be defined in advance.

The check standard

An intermediate check re-measures something stable and watches for change. That “something” is a check standard — a gauge block, a reference mass, a fixed resistor, a stable artefact — that you own, keep under control, and measure the same way each time. You are not calibrating it. You are asking a simpler question: does my instrument still give the same answer it gave when it was known-good? Because you compare like with like over time, the check standard doesn’t even need a tight uncertainty of its own — its job is stability, not truth.

Reading the result: the Shewhart control chart

Plot each check’s deviation from the reference over time and you have a Shewhart control chart — the same tool used in statistical process control. Three lines matter:

• Centre line — the established mean, ideally near zero.
• Warning limits at ±2σ — cross these and pay attention.
• Action limits at ±3σ — cross these and act.

Here σ is the check’s own short-term variation, worked out from a baseline period of in-control checks before the limits mean anything.

“In control” is not “every point on the line.” Normal scatter is expected — the chart tells you when the scatter stops being normal.

More than one point: run rules

A single point beyond ±3σ is the classic alarm, but a process can drift out of control while every point still sits inside the limits. The Western Electric run rules catch that:

• one point beyond 3σ (an action signal);
• two of three consecutive points beyond 2σ on the same side;
• a run of several points (often seven or more) all on one side of the centre line, or steadily trending.

A slow, one-sided trend is exactly what instrument drift looks like — and it is precisely the pattern a fixed calibration interval will miss.

When a check goes out of control

An out-of-control check is a nonconformance, and the reaction should be written into the procedure before it ever happens:

• Quarantine the instrument so it isn’t used for more work.
• Investigate — check the check standard, the method, the environment and the operator before blaming the instrument.
• Assess the impact — every measurement since the last good check is now in question, so you may need to recall certificates or re-measure. This is “reverse traceability”, and it is the reason a check failure is taken seriously.
• Correct, then verify the instrument is back in control before it returns to service.

Handled this way, a bad check protects your customers instead of embarrassing you.

Where it fits

Intermediate checks fill the gap the other controls leave. Traceability proves your chain to the SI at a point in time; proficiency testing proves you agree with other labs; intermediate checks prove your equipment stayed put in between. They also feed better calibration intervals — if the chart shows an instrument drifting, you recalibrate on that evidence rather than waiting for the calendar.

More from the Cali blog: browse all calibration & metrology guides →