Error-free dimensional PPAP: a practical guide for suppliers.

Why the dimensional report matters

In most first sample rejections the part is not wrong: the report is. A characteristic is missing, the revision does not match, a geometric tolerance has been evaluated incorrectly or there is no way to show what it was measured with. The outcome is always the same: resubmission, a delayed start of production, repeated lab hours and the image of an unreliable supplier.

This guide summarises what is expected of a dimensional results report, the ten errors that cause the most rejections and a working method to avoid them.

1. What the “dimensional results” element requires

The three most common frameworks (AIAG PPAP, FAI to AS9102 and PPF to VDA 2) essentially ask for the same thing: to show, characteristic by characteristic, that the part made with the series production process meets the current product definition. The minimum content is:

• All characteristics in the product definition: 2D dimensions, tolerances in the 3D PMI, general notes, general tolerance, referenced specifications and special characteristics.
• Unambiguous numbering: every characteristic with a unique number on the ballooned drawing or model and in the results table.
• Nominal and limits in the drawing units, with the characteristic type.
• Design identification: drawing or model number, revision and, where applicable, engineering change.
• Measurement date and person responsible.
• Sample identification: part number, cavity, mould or line. In a PPAP, as a general rule, at least one part from each distinct process is measured; each cavity counts as one.
• Instrument or method for each characteristic, with its identification.
• Actual measured result, not a generic “OK”, for variable characteristics.
• OK/NOK verdict and, for any NOK, a reference to the approved deviation or the corrective action.

If any of these points is missing, the report is incomplete even if every measurement is correct.

2. The 10 errors that cause the most rejections

Error 1. Characteristics that exist only in the 3D PMI

More and more customers declare the 3D model as the master document. The PMI contains tolerances that are not on the 2D: a surface profile, the position of a clip, basic dimensions.

How to avoid it: go through every view and annotation in the model, not just the saved default views, and compare the count with the 2D. Every annotation without a 2D equivalent becomes a numbered characteristic.

Error 2. General notes and general tolerance left out

“Undimensioned radii R0.5”, “no flash > 0.1 mm”, “general tolerance ISO 2768-mK” or “DIN 16742 TG5”: these are all requirements. If they are not numbered, the customer assumes they have not been checked.

How to avoid it: number each note as a characteristic in its own right. ISO 2768 is intended for machined parts; for injection-moulded plastic the usual standard is DIN 16742. If the drawing of a plastic part cites ISO 2768, ask the customer and keep a record of the answer.

Error 3. Mixed units

Drawing in millimetres and measurement program in inches, or a decimal comma in some rows and a point in others. A 0.250 that is actually in inches is a 6.35 mm error.

How to avoid it: set the units and decimal separator in the header and configure the measurement software export to match the drawing. Check the angular characteristics.

Error 4. Wrong drawing revision

The part is measured against revision C when the order specifies D, or the 2D is at D and the 3D is still at C.

How to avoid it: before measuring, freeze the design package (drawing, model and specifications with their revision) and check it against the order and the latest engineering change. If the 2D and 3D do not match, ask before measuring.

Error 5. Cavities not identified

A four-cavity mould, four parts measured and a single report with no way of knowing which column is which cavity. If a NOK appears, nobody knows which insert to correct.

How to avoid it: one column or one report per cavity, with the number read from the cavity marking on the part, not inferred from the order in the box.

Error 6. Cpk calculated from too few parts

A Cpk of 2.1 from five parts does not prove capability: the confidence interval is huge. For initial samples, preliminary indices (Ppk) are usually requested.

How to avoid it: keep the part-by-part report separate from the capability study, follow the sample size in the customer’s manual and, if there are not enough parts, say so and agree a follow-up plan.

Error 7. GD&T modifiers evaluated incorrectly (MMC/LMC)

Hole Ø10 +0.2/0 with position Ø0.2 Ⓜ A B C. Measured diameter Ø10.15: it departs 0.15 mm from maximum material, so the available position tolerance is 0.2 + 0.15 = Ø0.35. Measured position Ø0.28: OK. If the report ignores the modifier, it rejects a good part; if it applies a bonus where there is no Ⓜ, it accepts bad parts.

How to avoid it: for every characteristic with a modifier, state the measured size, the bonus and the total available tolerance, and check that the measurement program applies the drawing standard (ASME Y14.5 or ISO GPS, which are not equivalent in every respect).

Error 8. Datums not declared

A profile of 0.5 relative to A|B|C measured with a best-fit alignment is not the same result. The customer cannot reproduce your measurement without knowing how you aligned the part.

How to avoid it: document the datum reference frame, the fixture or clamping, and whether the part was measured free or restrained. If you aligned it differently, declare it and justify it.

Error 9. Rounding

Upper limit 25.10; measured 25.104; report: “25.10 OK”. Limits are absolute: 25.104 is out.

How to avoid it: report one decimal place more than the drawing or at the instrument’s actual resolution, compare without rounding and, close to the limit, apply the decision rule agreed with the customer, taking uncertainty into account (ISO 14253).

Error 10. Instruments without traceability or MSA

“Measured with a calliper”, with no number, no calibration date and no repeatability and reproducibility study.

How to avoid it: every instrument with a unique code and calibration valid on the measurement date; an MSA study for special characteristics. Rule of thumb: instrument resolution at least one tenth of the tolerance.

3. The 7-step method

1. Freeze the design package: drawing, model and specifications with their revision, checked against the order. Inconsistencies are raised now.
2. Extract every characteristic from three sources: 2D, 3D PMI and notes. Unique number and a ballooned drawing or model.
3. Review the interpretation: datums, modifiers, free state, general tolerance and standard. Anything ambiguous is asked in writing.
4. Plan the measurement: method and instrument per characteristic, alignment, fixturing and MSA.
5. Measure by sample and cavity, respecting the agreed conditioning before measuring.
6. Evaluate and write up: unrounded values, bonuses applied, OK/NOK, and capability kept separate.
7. Independent review: a second person checks balloons = rows, revision, units, cavities, instruments and signatures.

4. Final checklist

To print it, use your browser’s print option: the page prints cleanly.

5. How automation helps

Many of the errors above are not metrology errors but transcription errors: copying hundreds of annotations into a spreadsheet by hand, numbering them, placing balloons and keeping everything consistent when a new revision arrives.

Tools that read the model’s semantic PMI generate the list with type, nominal, tolerances, modifiers and datums, plus the ballooned drawing or model. They bring completeness (no annotation is left out), consistency (balloon, row and annotation linked) and time to interpret and measure rather than copy.

Automation does not replace judgement: steps 3 and 7 still belong to people trained in GD&T. GALGA is designed for step 2. Try it for 7 days on your own parts or see the cases.