Stamping inspection: first article, in-process and final

Why one inspection stage is not enough

A stamped part is produced by a tool that wears, a press that deflects and a coil that varies. Inspection that happens only at the end of the run can tell you that the last parts are good, but it cannot tell you when the process moved or how many parts were affected.

The practical answer is three gates. The first article proves that the tool and the process can make the part to the drawing. In-process checks detect drift while there is still time to correct it. Final inspection confirms that the lot meets the agreed acceptance criteria before it ships.

Each gate answers a different question, and each one needs a different record. Treating them as one inspection event is how a supplier ends up with a full dimensional report on three parts and no evidence about the other twenty thousand.

The first article: prove the process

A first article inspection is a full dimensional and material verification of a part from the production tool, using the production process. It is not a check on a hand-made sample. If the tool is new, the first article is the moment when the drawing and the die meet, and it is the cheapest moment to find a mismatch.

Every characteristic on the drawing should be reported with a measured value, a nominal and the tolerance. That includes the features that are hard to reach, because those are often the ones that decide function. A report that omits a dimension because the gauge cannot reach it is a finding in itself.

The report should identify the drawing revision, the tool number, the material lot, the machine, the operator and the date. Add a ballooned drawing so that a reader can match each result to a feature. For a multi-cavity tool, report a part from each cavity.

• All drawing dimensions, including reference dimensions where the drawing asks for them
• Material grade, temper and thickness with the mill certificate
• Plating type and thickness on the functional areas
• Bend angles, flatness and burr height where specified
• Functional checks such as contact gap, force or crimp height
• Tool number, press, material lot and drawing revision

In-process: watch the feature that drifts

A progressive die does not drift evenly. Some features are set by a punch that wears slowly, others by a bend that moves with material temper, and others by a coining station that changes as the tool beds in. The in-process plan should focus on the sensitive ones.

For a typical stamped part, the first candidates are the bend angle, the overall length across several stations, the hole position relative to the datum and the burr height. For a terminal, add the contact gap and the crimp height. For a current-carrying part, add the width at the narrowest section and the flatness of the joint face.

The frequency should follow the process, not a habit. A new tool needs more frequent checks during the first production run. A stable tool can move to a longer interval, but the interval should be justified by the data, not by the wish to reduce inspection cost.

The check should be quick enough to run at the press. A calliper or a dedicated gauge at the machine is more useful than a CMM result that arrives the next day. Save the CMM for the first article, the periodic audit and the investigation.

A practical inspection plan

The table below is a starting structure for a stamped part. The exact features and frequencies should follow the drawing, the volume and the criticality of the part.

Final inspection and sampling

Final inspection is a lot decision. It should use a defined sampling plan, a defined acceptance number and a defined reaction to a reject. Inspecting a handful of parts and calling the lot good is not a sampling plan, even if the handful looks fine.

The final check normally covers the critical dimensions, the visual condition, the plating and the packaging. Packaging deserves a line of its own: a part that is dimensionally correct but arrives with bent legs or a damaged plating is a reject at the customer.

For a safety-related or high-volume part, add a capability check on the critical dimensions. A lot that is inside tolerance but centred close to a limit will produce rejects on the next run. The capability index tells you whether the process is centred, not just whether this lot passed.

Choosing the measurement method

The method should match the tolerance. A calliper is fine for a general dimension with a wide tolerance. A micrometer is better for thickness. A height gauge or a dedicated fixture is better for a bend height. A CMM or an optical system suits a position tolerance or a profile that cannot be reached with a hand tool.

The gauge itself needs to be capable. A gauge with a resolution close to the tolerance cannot separate a good part from a bad one. The usual check is a gauge repeatability and reproducibility study on the feature, not on a convenient test block.

For functional features, a attribute gauge is often better than a variable measurement. A go and no-go gauge on a contact gap tests the function directly, is fast and does not need a skilled operator. Use the variable measurement for the first article and the capability study, and the attribute gauge for the line.

Records that survive an audit

A record is only useful if it can be traced. The minimum set is the drawing revision, the tool number, the material lot, the production date, the gauge used and the result. Add the operator and the shift if the process depends on setup.

Photographs help for visual defects, but they should be dated and linked to the lot. A photograph of a burr with no scale and no lot number proves very little. A cross-section of a crimp with the lot and the crimp height is much stronger evidence.

Keep the first article, the setup sheets, the in-process charts and the final records together for the period the customer requires. If a non-conformance is found later, the records are what allow the supplier to bound the problem and recall only the affected parts.

When a result is out of limit

The plan should say what happens when a check fails. Stop the press, quarantine the parts produced since the last good check, identify the cause and record the correction. Then re-check the first parts after the correction before releasing the run.

The quarantine boundary is the important part. If the last good check was two hours ago, every part made in those two hours is suspect until proven otherwise. A supplier that adjusts the tool and continues without bounding the suspect period has not contained the problem.

Repeat failures point to a design or tool issue rather than an operator issue. If the same dimension moves again after adjustment, the die needs a change, or the drawing needs a tolerance that the process can hold. That conversation belongs in the periodic review, not in another shift of adjustment.