Engineering Answers / Prototyping & Validation
An FAI is not a dimensional report with a cover sheet. On silicone it is also a statement about when the part was measured, how it was held, and which compound and cure produced it — because without those, the numbers cannot be reproduced.
A usable first article inspection contains six things: a ballooned drawing with every dimension numbered, actual measurements against each balloon, CMM data for form and position, the compound lot and cure record, hardness and colour readings with the method stated, and a visual and functional check against agreed limits. On silicone the two most commonly missing items are the measurement timing after demolding and the method used on a soft part, and those two cause most of the disputes between supplier and customer.
A first article inspection exists to answer one question: is this tool, on this process, with this compound, able to make the part on the drawing? Everything in the report should serve that question.
1. A ballooned drawing. Every dimension, tolerance and note numbered, and the same numbers used in the results table. Without it, the report is a list of numbers with no map back to the drawing, and nobody can audit it later.
2. Actual measurements per balloon. Not a pass or fail column alone. The actual value, the tolerance, and the difference. A dimension at nominal and a dimension at the edge of tolerance are not the same information even when both pass, because one tells you the tool is centred and the other tells you it is about to drift.
3. CMM data for form and position. Flatness on a sealing face, concentricity, position of holes. These are the characteristics a calliper cannot see and the ones that decide whether the part assembles.
4. Compound and cure record. Compound grade, lot number, cure system, mould temperature, cure time, and post-cure time and temperature where applicable. This is the section that lets a second run be compared to the first.
5. Hardness and colour, with the method. Hardness to ISO 7619-1 at stated points and stated stack thickness; colour as ΔE against a reference, under a stated illuminant. A hardness number without a method is an opinion.
6. Visual and functional check. Against agreed limits: a boundary sample set, a flash limit by zone, a sealing or flow test where the part has one.
| Section | What it prevents | Commonly missing |
|---|---|---|
| Ballooned drawing | Untraceable results | Balloon numbers not matching the results table |
| Actuals per dimension | Hiding drift inside a pass | Only pass/fail reported |
| CMM form and position | Assembly failures found at the customer | Flatness on the sealing face |
| Compound and cure record | Irreproducible second run | Post-cure time and temperature |
| Hardness and colour method | Arguments about "it feels harder" | Stack thickness for hardness; illuminant for colour |
| Visual and functional | Subjective accept/reject | Flash limits stated by zone |
Silicone is the material where metrology most often becomes the argument, and the reason is simple: the part changes shape when you measure it.
A 1.0 mm wall at 40 Shore A will compress under a calliper jaw by more than the tolerance band. The inspector is not careless; the tool is wrong for the material. The methods that reproduce are non-contact optical measurement, a low-force gauge with a defined measuring force, or a cross-section cut through a potted sample for a truly critical wall.
Whichever method is used, it has to appear in the report. Two labs measuring the same part with different methods will disagree by more than the tolerance, and both will be right within their own method.
Timing matters as much as method. Silicone relaxes after demolding. A dimension taken ten minutes off the tool and the same dimension taken the next morning differ — typically by a few hundredths of a millimetre on a soft section, more on a thin wall under load. The report should state a minimum relaxation of 24 hours at controlled conditions, and both sides should measure at the same point in that window.
Almost every dimensional argument we have been pulled into was not a tool problem. It was one side measuring a freshly molded part with a calliper and the other measuring a relaxed part optically. Once both sides agreed on 24 hours and a non-contact method, the "out of tolerance" dimension landed in the middle of its band and stayed there for the life of the tool.
Two things get folded into FAI reports that should not be there, and both make the report harder to trust.
Biocompatibility and food-contact test reports. These are material and process validation, not dimensional first article. They have their own protocols, their own sample requirements and their own timelines. Putting a summary line in the FAI is fine; treating the FAI as the evidence is not.
Capability studies. Cpk needs a run of parts, not one. First article proves the tool can make a conforming part. Capability proves it can keep making them. They are consecutive steps, and the second one is the one that predicts your reject rate.
Ask for the FAI format at quotation stage, not after the tool is built. Three things in that format tell you whether the supplier measures seriously:
A supplier who cannot produce a sample report before the order is a supplier whose first report you will be arguing with after it.
At least 24 hours at controlled temperature and humidity, and the report should say so. Silicone continues to relax after demolding, and a part measured ten minutes off the tool reads differently from the same part measured the next day — typically by a few hundredths of a millimetre on a soft section, more on a thin wall. If the supplier measures hot and the customer measures cold, both are telling the truth and the numbers still will not match.
On a hard section, with care. On a soft wall or a sealing lip, no — the jaw pressure deforms the part and the reading depends on how hard the inspector squeezed. The methods that reproduce are non-contact optical measurement, a low-force gauge with a defined measuring force, or a cross-section through a potted sample. Whichever is chosen has to be written into the report, because "measured with callipers" is not a measurement method.
Yes, at least on the dimensions the re-machining affects. Refacing a shut-off land changes the cavity, and a tool that was signed off before the reface is not the same tool afterwards. In practice this means a partial FAI after any maintenance that removes material, and a full FAI after a new insert, a new cavity or a change of compound.