Calibration and measurement traceability for sheet metal

A first-article report lists a 6.00 mm hole as 6.01 mm, measured with a digital caliper. The callout is 6.00 +0.05/0. The instrument's own accuracy is about ±0.02 mm, so the reading carries an uncertainty that is 40% of the entire tolerance band. The calibration sticker on the caliper is fourteen months old. Three things are wrong, and only one of them is about the sticker.

Measurement traceability is the property that lets that 6.01 mm be compared with a national standard through a documented chain, with an uncertainty attached to every link. ISO/IEC 17025:2017 defines it, ISO 9001:2015 clause 7.1.5.2 requires it for measuring equipment, and an auditor will ask for the certificate rather than the sticker.

This page covers what the chain contains, which instrument suits which tolerance band, why temperature belongs in the chain on long sheet metal parts, and the four records a buyer should expect in a first-article pack. The standards and compliance hub covers the rest of the cluster.

What does measurement traceability actually prove?

It proves that on a stated date, in stated conditions, a stated instrument was compared with a stated reference, and the difference was recorded with an uncertainty. That is the whole content, and every one of those fields matters.

The chain behind the certificate runs from the shop instrument to a working standard, from the working standard to a reference standard, and from there to a national metrology institute and the SI definition of the metre. Each comparison has its own uncertainty, and the uncertainties combine — usually by root-sum-square — into the uncertainty on the shop floor reading. A certificate that reports a result with no uncertainty has broken the chain at the point where it matters.

The as-found column does more work than any other. A caliper found 0.01 mm out after a year is a stable instrument. One found 0.06 mm out tells you it was dropped, and it also tells you that every part accepted since the last calibration was accepted with a gauge that was reading wrong. ISO 9001 clause 8.7 then requires an impact assessment on the affected product, and that assessment is only possible if the gauge register links the instrument to the lots it measured.

Which instrument can judge which tolerance?

The working rule is the test uncertainty ratio: the instrument's uncertainty should be one quarter or less of the tolerance band. A 4:1 ratio leaves the decision clear at the limits. A 10:1 ratio is better and is sometimes demanded on safety-critical features. Below 4:1, the gauge starts deciding the result instead of the part.

Read the table against the drawing. A general sheet metal tolerance of ±0.5 mm is comfortably inside a caliper's capability. A hole position of ±0.1 mm, a pin diameter of ±0.01 mm or a flange height of ±0.05 mm is not, and the first-article report that uses a caliper for those features is producing numbers rather than evidence.

The 4:1 rule is also why the ±0.5 mm working tolerance on custom parts is not a limitation of ambition. It is the band that the standard inspection route can judge honestly, and tighter features are handled by agreement with a specific method: a fixture, a micrometer, or a third-party report.

Calibration accuracy is only half the input. The instrument has a stated accuracy, but the reading also carries the operator's technique: how squarely the jaws sit, how much pressure is applied to the thumb wheel, and whether the part is measured at the same point twice. That variation is what a gauge R&R study measures, and it is usually larger than the calibration error. On a 2 mm flange, two operators with the same calibrated caliper routinely disagree by 0.03–0.05 mm because one measures at the root of the bend and the other at the tip. The drawing can remove that ambiguity by naming the measurement point, which costs one dimension line and saves a rejection.

Temperature is part of the chain on long parts

Steel expands about 11.5 µm per metre per degree Celsius. A 500 mm bracket measured at 26 °C instead of the 20 °C reference temperature is 0.035 mm longer than its certificate says. For most sheet metal callouts that is irrelevant. For a ±0.05 mm feature it is most of the band.

At the other end of the range the numbers get serious. The 200 T brake has a 4,450 mm bed, and a bent rail from it measured in a shop at 28 °C is 0.41 mm longer than the same rail at 20 °C. That is the same order as the ±0.5 mm working tolerance on the part, and larger than any geometric tolerance a designer is likely to put on a rail. A drawing that calls a tight overall length on a 4 m formed part without a temperature note is asking for a measurement argument in July and a pass in January.

Three practical controls. Let long parts stabilise on the plate before measuring, record the shop temperature on the inspection report where the tolerance is under 0.2 mm, and agree the reference temperature in the drawing notes when a dimension is longer than about 1 m. For parts of a few hundred millimetres, the shop is close enough and the effort belongs elsewhere.

Aluminium doubles the problem. Its coefficient is about 23 µm per metre per °C, so a 2,000 mm extrusion moves 0.37 mm between 20 °C and 28 °C, and an aluminium panel clamped in a steel fixture is held while it wants to grow. That is one reason a large aluminium assembly is better checked on a granite plate with a temperature record than clamped in a steel holding fixture built for production.

What breaks the chain most often

Four failures account for most of the traceability gaps we see in incoming inspection reports.

Calibration outside the working range. A certificate for a 0–150 mm caliper may state the error at 25, 75 and 125 mm. A feature measured at 5 mm sits in a part of the scale that was not checked, and it is exactly where jaw wear shows first.

An in-house standard with no chain. Checking calipers against a shop-owned gauge block is only traceable if the block itself carries a certificate, is stored without burrs, and is used at a controlled temperature. A block bought at a tool shop and never calibrated ends the chain at the bench.

An interval that was never reviewed. Twelve months is a starting point, not a rule. A gauge used on the shop floor by six operators wears faster than one kept in the inspection room, and the interval should reflect that difference.

A gauge register without lot links. When a gauge is found out of tolerance, the register has to say which lots it measured. Without that link the impact assessment becomes a blanket recall, which is expensive, or a guess, which is worse.

What an auditor asks for, and what a buyer should ask for

An ISO 9001 audit of clause 7.1.5 looks for five things, and a buyer reviewing a supplier can use the same list.

1. A register of every measuring instrument, with an identifier, location, interval and last and next calibration dates.
2. Certificates for each instrument, retrievable by identifier, naming the reference standard and the uncertainty.
3. An interval rationale — why 12 months and not 24, and what evidence supports it.
4. Out-of-tolerance handling — the impact assessment, the lots affected and the disposition.
5. Traceability of the standards themselves, which is where a shop that calibrates in-house is most often thin.

The fifth item is the one that separates a real system from a folder. A shop that checks its calipers against its own gauge blocks has a chain only if the blocks are calibrated, the blocks are handled correctly, and the comparison is done at a controlled temperature. Gauge blocks are graded under ISO 3650, and a grade 2 set is adequate for shop comparison against a calibrated micrometer, provided it is wrung clean and kept free of burrs.

For a supplier audit, ask for the register and pick two instruments at random. If the certificate for either one is missing or does not name a reference standard, the rest of the folder is decoration. The supplier audit checklist covers the rest of the quality system.

What Ganesh runs, and how the chain is kept

Ganesh works inside ISO 9001:2015 and ISO 14001:2015 across five units in Mumbai (Vasai) and Nagpur, with first article inspection and PPAP available where the programme requires them. The measurement route follows the tolerance: laser-cut features are checked in a fixture built in the in-house tool room, formed features on a surface plate against the base datum, and coating on a DFT gauge with cross-hatch adhesion testing.

The fixture route is a measurement decision as much as a production one. A checking fixture built for a ⌀0.4 position zone tests the assembly condition in seconds, and its pins are themselves wear items that need a calibration interval. A fixture that has never been checked is a gauge with no certificate, and the parts it passes inherit that gap.

The stamped part inspection guide covers sampling and first-article content for press parts, and the bending tolerance guide covers the formed features that most often sit near a limit.

What to put on the drawing and in the purchase order

• The tolerance band, so the supplier can select an instrument that passes 4:1.
• The measurement method for any feature under ±0.1 mm: fixture, micrometer, or third-party report.
• The reference temperature for dimensions over about 1 m.
• The inspection record to ship: dimensions, instrument, uncertainty and the operator.
• The calibration evidence you want to see at first article, by instrument class.
• Whether the fixture is part of the tooling package and who owns its calibration.
• Any characteristic that needs a certified third-party report, named before the order.

The quotation request takes the drawing and the inspection requirements together. Naming the tight features at enquiry is what lets the route and the gauge be chosen before the first part is cut.

Where calibration stops

A calibration certificate says nothing about whether the gauge suits the feature. That is a measurement system question, covered by a gauge R&R study, and a gauge that passes calibration every year can still consume 30% of the tolerance band in reading variation. The gauge R&R guide sets out the study and the acceptance numbers.

Ganesh does not operate an accredited calibration laboratory, and we do not issue ISO/IEC 17025 certificates for customer instruments. Calibration of the shop's instruments runs through an accredited external laboratory, and where a drawing requires a certified material or dimensional report, that report comes from a third party. Say so at enquiry: it changes the lead time and the record pack, not the part.

Send the drawing with the tolerance bands and the characteristics that decide fit. The engineering team will return the method, the instrument and the inspection record for each one.