Torque wrench calibration under ISO 6789:2017
The 2017 revision split the standard into two parts and changed what a conformity certificate may claim. What Part 1 and Part 2 each cover, and why a Part 1 certificate is not an uncertainty statement.
The 2017 revision of ISO 6789 did something unusual: it separated what most people had been treating as one activity into two documents with different outputs, and made clear that only one of them produces a measurement uncertainty. A great many torque certificates in circulation are Part 1 declarations being read as though they were Part 2 calibrations.
Part 1 and Part 2
| ISO 6789-1 | ISO 6789-2 | |
|---|---|---|
| Purpose | Declaration of conformance | Calibration |
| Output | Pass or fail against the tool's class | Measured values with measurement uncertainty |
| Uncertainty stated | No | Yes |
| Use | Confirming a tool still meets its specification | Traceable measurement, and input to an uncertainty budget |
| Typical reason to choose it | Routine workshop control | Regulated work, or where the torque value feeds a calculation |
Neither is wrong. The error is buying Part 1 and then citing it as traceable calibration with uncertainty in a validation file, because Part 1 does not carry an uncertainty and was never intended to.
What the test actually involves
- The tool is exercised at maximum before measurement — several full-scale operations to settle the mechanism.
- Measurements are taken at 20 %, 60 % and 100 % of the tool's maximum torque, which is where the standard places them.
- A series of measurements is taken at each point, not a single reading, because scatter in a click mechanism is substantial.
- The tool is operated in its normal manner: a click wrench is taken to the click, and a reading wrench is read at the value.
- Direction matters: a tool used both clockwise and anticlockwise needs testing in both.
A click wrench should be stored wound down to its lowest setting, not at the value it was last used at. Leaving the spring compressed for months changes the tool, and the change appears as a shift at the next calibration that nobody can account for.
The length problem
Torque is force times distance, so where the hand is on the handle is part of the measurement. Tools with a marked hand position must be operated at that position during calibration and in use. For an adjustable-length wrench, or one used with an extension or a crow's foot adapter, the effective length changes and with it the torque delivered at the fastener — an adapter that extends the tool by 10 % of its length changes the delivered torque by roughly that proportion unless the setting is corrected for it.
This is why a torque certificate describes a tool in a configuration. Change the configuration and the certificate describes something you are no longer using.
What this is for
In regulated manufacturing a torque tool is usually in the file because a joint integrity, a closure or an assembly step depends on it: a container closure torque that holds a seal, a filter housing that must not be over-tightened, a cap on a drug product. The tolerance on that torque belongs to the process, not to the tool, and the tool has to be good enough to police it — the same test-uncertainty-ratio reasoning that applies to any other measurement.
That reasoning is also the reason to care which part of ISO 6789 you bought. You cannot form a ratio with an uncertainty you were never given.
Frequently asked questions
- What is the difference between ISO 6789-1 and ISO 6789-2?
- Part 1 is a declaration of conformance: the tool is tested against the requirements for its class and the result is a pass or a fail, with no measurement uncertainty stated. Part 2 is a calibration: it reports measured values together with the measurement uncertainty. If you need to demonstrate traceability with an uncertainty — for a regulated process, or as an input to an uncertainty budget — you need Part 2, and a Part 1 certificate will not do, however official it looks.
- At what points is a torque wrench calibrated?
- ISO 6789 places the measurements at 20 %, 60 % and 100 % of the tool's maximum torque, with a series of measurements at each rather than a single reading. The tool is exercised at maximum beforehand so the mechanism is settled. If you habitually use the tool at a value between those points, it is worth asking for an additional point there — the principle of calibrating where you actually work applies here as everywhere.
- Does an extension bar change the torque setting?
- If it changes the effective length of the tool, yes. Torque is force times distance, so an adapter that extends the distance from the handle to the fastener increases the torque delivered for a given setting, roughly in proportion to the change in length. A crow's foot used in line with the wrench does this; one used at right angles to it largely does not. Either way the certificate describes the tool in the configuration it was tested in.
- How should a click torque wrench be stored?
- Wound down to its lowest setting, not left at the value it was last used at. Storing the spring under compression for long periods changes the tool, and that change turns up as an unexplained shift at the next calibration. It is one of the few maintenance habits that visibly improves as-found results.
References
- [1]ISO 6789-1:2017 and ISO 6789-2:2017 — Assembly tools for screws and nuts: Hand torque tools
- [2]EURAMET cg-14 — Guidelines on the Calibration of Static Torque Measuring Devices
- [3]ISO/IEC 17025:2017 — General requirements for the competence of testing and calibration laboratories
- [4]JCGM 100:2008 — Evaluation of measurement data: Guide to the expression of uncertainty in measurement (GUM)
General technical guidance written against the cited sources. It is not regulatory or legal advice and does not replace the applicable standard, guideline or a qualified reviewer's judgement.
Have a question on this topic?
Ask ValiTrac AI and see the evidence and calculation behind the answer.
Ask ValiTrac AIFree calculator: Calibration interval
Paste an instrument's as-found history and see how fast it is actually drifting, how much of its tolerance that consumes, and how long the evidence supports trusting it. The strongest input to an interval is the instrument's own record — this reads it.
Open the calculatorRelated articles
How to set calibration intervals
There is no universal interval. Set it from risk, manufacturer data and — above all — the instrument's own as-found history.
What is a decision rule and why every certificate needs one
A decision rule states how measurement uncertainty is accounted for when declaring pass or fail against a tolerance. Simple acceptance, guard banding and non-binary statements are the common choices.
What is a decision rule and why every certificate needs one
A decision rule states how measurement uncertainty is accounted for when declaring pass or fail against a tolerance. Simple acceptance, guard banding and non-binary statements are the common choices.