Torque wrench calibration: ISO 6789 interval, tolerances and records

What ISO 6789 covers

ISO 6789 is the international standard for hand torque tools: the torque wrenches and torque screwdrivers used to tighten screws and nuts to a controlled torque. In 2017 the single 2003 edition was replaced by two parts. Both are first editions, published in February 2017.

ISO 6789-1:2017 is written for designers and manufacturers. It sets the tool types and classes, the maximum permissible deviation for each, the conformance tests a manufacturer runs and the declaration of conformance that comes with a new tool. It does not cover calibration certificates.

ISO 6789-2:2017 covers calibration. It sets the calibration method, the way to work out the measurement uncertainty of each calibration and the requirements for the torque measurement device the calibration relies on. Its Annex C gives minimum requirements for calibrating that device when no traceable certificate already states its uncertainty.

ISO is revising both parts. Until a new edition is published, the 2017 editions are the current ones. When a customer or a contract names ISO 6789, check which edition it means.

Torque tool types and classes

ISO 6789-1 sorts hand torque tools into two types. A Type I tool indicates the torque on a scale, a dial or a display. A Type II tool is set to a value in advance. When that value is reached it gives a physical signal that briefly reduces the torque, often with a click. Each type is divided into classes by design.

Permissible deviation by type and class

The maximum permissible deviation is the largest relative deviation a result may show, as a percentage of the target value on the tool (ISO 6789-1, 3.1.4). It depends on the type, the class and whether the maximum torque of the tool is above 10 N·m.

• Each result has to lie inside the limit on its own. A wrench with one reading outside it fails, even when the average is inside.
• Where the manufacturer claims a tighter limit, the tighter limit applies.
• A tool that works in both directions has to meet the limit in each direction the manufacturer specifies.
• Part 1 judges conformance without taking the uncertainty of the tool or of the torque measurement device into account (5.1.5).

• Free torque wrench calibration record (Excel) with these limits built in

What a calibration checks

ISO 6789-2 calibrates with the measurement method of ISO 6789-1, clause 6, and replaces that part's requirement for the torque measurement device with its own (4.2 and 4.3). The wrench is mounted on a calibration system and loaded through its handle. The reference value comes from the measurement device, which is either mechanical or an electronic torque transducer with a display.

Readings are taken at target values across the range of the tool. For Type I Classes A, B and D and Type II Classes A, D and G, the range runs from the lowest marked value to the maximum (ISO 6789-1, 5.1.3). For the other classes it is the range or value the manufacturer specifies. Calibration labs commonly test Type I tools and Type II Classes A, D and G at three target values: the lowest value of the range, 60% and 100% of the maximum. A fixed setting tool is tested at its set value, and a non-graduated adjustable one at its lowest specified or preset value.

The worked examples in Part 2 take five readings at a target value for indicating tools and graduated setting tools, and ten readings for fixed and non-graduated setting tools (5.2). Each reading is judged on its own.

How the torque goes in matters. Part 1 controls the rate of loading through the time taken to apply the last 20% of the torque, and the 2017 edition stresses avoiding parasitic forces such as side loads and bending. On some designs the reading also depends on where the load is applied on the handle. Part 1 has manufacturers test the effect of geometry changes such as flexible heads and extension bars, and tell users about it (5.1.8).

Part 2 turns these effects into uncertainty contributions. They include the repeatability of the tool, the variation from the point where the force is applied, and the geometry of the output drive and of its interface with the calibration system. Type I tools and Type II Classes A, D and G add reproducibility and the resolution of the scale or display. The measurement device adds its own uncertainty, and the expanded uncertainty uses a coverage factor of k = 2 for about 95% confidence.

The measurement device has limits of its own. At each target value its relative uncertainty interval may be at most a quarter of the interval expected for the tool. It needs a valid calibration certificate from a laboratory that meets ISO/IEC 17025, or from a laboratory that maintains the national measurement standard. Without a control procedure it is recalibrated at least every 24 months, and also after an overload larger than 20% of the top of its range, a repair or improper use (4.3).

• How to build a measurement uncertainty budget
• TUR and TAR calculator

What goes on the record and the certificate

Whether a wrench is checked in-house or calibrated by a lab, the record should let someone else see what was done and reach the same decision. Keep these items for each calibration.

• The ID, serial number, type, class and range of the wrench, and the directions it works in.
• The limit applied: the ISO 6789-1 value for its type and class, or the tighter figure from the manufacturer.
• The measurement device used, with its ID and its calibration due date.
• Each target value and each reading with its deviation and its pass or fail.
• The as-found readings before any adjustment, and the as-left readings if the wrench was adjusted or repaired.
• The date, who did the work, the next due date and, if you count cycles, the count at calibration.

• Free torque wrench calibration record (Excel)
• Calibration certificate requirements
• Calibration log template

The ISO 6789-2 calibration certificate

An ISO 6789-2 calibration certificate comes from a calibration laboratory, often one accredited to ISO/IEC 17025 for torque. Part 2 determines the calibration result at each calibration torque as a relative measurement uncertainty interval of the tool, worked out from an uncertainty budget (clause 7), and sets what the certificate contains (clause 8).

Part 2 expects that budget for each individual tool. The Type B parts may instead come from data on at least 10 tools of the same model, as long as the values are reviewed systematically (6.1). An accredited lab's certificate also carries the general elements that ISO/IEC 17025 clause 7.8 requires.

How often to calibrate a torque wrench

ISO 6789-2 clause 4.1 starts from your own procedure. If you control test equipment under a procedure, torque tools belong in it. You choose the interval from the maximum permissible error you need, how often the tool is used, the typical load, and the ambient and storage conditions. You then adapt the interval from the results of successive calibrations.

If you have no such procedure, the standard gives a default of 12 months or 5,000 cycles, whichever comes first. The interval starts with the first use of the tool. A shorter interval applies when you, your customer or legislation require one.

Whatever its due date, a wrench also needs calibration after a repair, after any improper handling that could affect it, such as a drop, and after an overload greater than the overload test of ISO 6789-1 (5.1.6). That test loads most tools to at least 125% of their maximum.

Cycles add up fast. A wrench that tightens 40 fasteners a shift, one shift a day and five days a week, reaches 5,000 cycles in 25 weeks. For that wrench the cycle count, not the calendar, sets the default due date.

• Calibration interval calculator

Justifying a shorter or longer interval

Once you have a procedure, the interval is yours to set, and the evidence behind it is what makes it defensible. These are the inputs that move it.

• Usage. Count or estimate the cycles per week from the jobs a wrench does. A heavily used wrench reaches 5,000 cycles long before a year is up.
• Overloads, drops and repairs. Each one calls for a calibration before further use, and a wrench with repeated incidents may need a shorter interval.
• As-found history. A wrench that keeps coming back in tolerance with readings close to target supports a longer interval. One that drifts toward the limit, or arrives out of tolerance, needs a shorter one.
• The joint. A wrench used on safety-critical joints usually warrants a shorter interval and more frequent checks.
• Storage and environment. ISO 6789-2 names ambient and storage conditions among the factors for the interval.

• How to set calibration intervals
• Drift rate and interval calculator

Checks between calibrations

A quick check on an in-house torque tester catches a damaged wrench early. Set the wrench to a value it is often used at, release it a few times on the tester and compare the readings with its limit. Record the result. A check like this is a verification, not a calibration, so it does not reset the due date.

Take a wrench out of use and check it when you see any of these signs.

• It was dropped, or pulled well past its maximum, for example on a seized fastener.
• The click is faint, late or missing, or the setting slips.
• The scale, ratchet, head or handle is damaged or loose.
• Interim checks show the readings moving toward the limit.

As-found, as-left and an out-of-tolerance wrench

Record the as-found readings before anyone adjusts the wrench. They show the condition the wrench was in while it was being used, which is what the work since the last calibration relied on. If the wrench is adjusted or repaired, take a full set of as-left readings at the same target values.

If any as-found reading is outside the limit, the wrench was out of tolerance for some unknown part of the time since its last good calibration. ISO 9001 clause 7.1.5.2 asks you to determine whether earlier results were adversely affected and to act as needed. For a torque wrench, work through these steps.

• Tag the wrench and take it out of use.
• Find the joints it tightened since its last good calibration, from job records, work orders or the tool crib log.
• Look at the direction and the size of the error. A wrench that under-tightens risks loose joints. One that over-tightens risks stretched bolts and stripped threads.
• Decide what to do with the affected work, such as re-torquing or inspection, and tell the customer when the contract requires it.
• Record the assessment and the actions, then review the interval and look for a cause such as an overload or a drop.

• As-found and as-left readings

Where Axiospec fits

Axiospec is calibration management software. It records each torque wrench with its calibration history, the as-found and as-left readings, the certificate and the next due date. A wrench logged as Fail goes into quarantine, and its record asks which earlier work was affected and what was done about it. Axiospec does not trace the joints a wrench tightened. You find those in your own records. Choosing the limit, the interval and the action stays with you.

• Gage management software