Key takeaways
- An assembly FAI proves a process as well as a set of dimensions.
- Every BOM line needs evidence, including the parts the assembler did not make.
- A first article built with hand selected components is a sample, not a first article.
- The pack should show the fixture, the torque values and the test results, not only the measurements.
An assembly FAI covers a process, not a part
A first article inspection on a single machined part is a dimensional exercise. The inspector measures the features on a drawing and records the results. An assembly is different. The dimensions still matter, but most assembly failures come from the process: a fixture that locates the wrong face, a torque driver that is out of adjustment, a component from a different supplier that sits 0.3 mm proud.
That means the pack has to show two things at once. It has to prove that the parts in the sample meet their drawings, and it has to prove that the process that put them together is defined, controlled and repeatable. A pack with perfect dimensions and no process evidence describes a lucky build.
Start the plan at the assembly drawing, not at the component drawings. The assembly drawing is where the interfaces live, and the interfaces are where the failures appear.
Treat the first article as a review document rather than a certificate. Its value is the open issues list, because that list is what the next build has to close.
Balloon the drawing, then follow the BOM
Balloon every dimension, note and symbol on the assembly drawing that affects fit, function, safety or the customer interface. That includes torque callouts, adhesive callouts, test pressures and the identification label. Then number the BOM lines and map each one to a component or to a sub assembly.
The mapping is the part buyers skip. A BOM line for a purchased latch is not covered by a dimensional result on the housing. It needs its own evidence: the supplier certificate, the incoming inspection record and, where the component is critical, a first article from that supplier on file.
Keep the balloon numbering consistent between the drawing, the inspection report and the BOM. When an auditor picks a random balloon and asks for the record, the trail should be short.
Component evidence: certificates and sub tier first articles
For each purchased component, decide what the evidence needs to prove. A standard fastener needs a material or property certificate and a dimensional sample check. A custom moulding needs a first article, a material data sheet and a certificate of analysis. A bought in contact or terminal may need a plating thickness record.
Copies of certificates are not the same as traceability. The certificate should carry a batch or heat number that matches the label on the packaging and the record in the build. If the assembler cannot connect the certificate to the parts in the sample, the evidence chain is broken.
Where a sub tier supplier changes a process, the change should reach the assembly FAI as a revision. A new mould at the sub tier, a new plating line or a new material batch can all move the assembly result without any change to the assembly drawing.
Ask the assembler to list the sub tier sources with the part numbers they supply. The list is often the first time anyone has written the chain down, and it usually reveals one line where the source changed without a drawing revision.
Assembly level measurements
Measure the features that only exist after assembly: the gap between two components, the flush condition of a cover, the position of a terminal block relative to the housing edge, the free length of a spring in its installed state, the force needed to latch a lid.
These are the measurements that a component first article cannot produce, and they are usually the ones the customer cares about. Write them as variables with a nominal and a tolerance, or as attributes with a clear pass or fail. Avoid a report that lists only the component dimensions that were easy to reach.
Record the measurement condition. A gap measured with the assembly resting on a bench and a gap measured with the cover screwed down are different numbers. If the drawing does not state the condition, the first article should, and the drawing should be corrected at the next revision.
A first article should include at least one unit measured at each extreme the process can produce, where the process allows it. Pulling a sample from the start, the middle and the end of a run shows the drift that a single unit hides.
Process evidence: fixtures, torque and test
A first article pack for an assembly should include a photograph or a drawing of the assembly fixture, the torque values and the tooling used, the test results with the method, and the process parameters that were running when the sample was built.
Torque records need the driver identification and the calibration date. A list of torque values with no tool reference proves that someone wrote numbers down. Test records need the fixture, the gauge and the acceptance limit, in the same form the production line will use.
Add the operator training record and the work instruction revision. An assembly is only as repeatable as the instruction the operator follows, and a first article built by the supervisor from memory is not evidence of a repeatable process.
Photographs of the fixture and the station are worth including. They show the reviewer what the operator actually sees, which is not always what the work instruction describes.
The PPAP elements that carry the weight
PPAP has eighteen elements in the standard submission, and not every programme needs all of them at the same depth. For an assembly, the elements that usually decide acceptance are the design records, the process flow, the control plan, the measurement system analysis and the dimensional results.
The process flow and the control plan should match each other line for line. If the flow diagram shows a torque station and the control plan does not, the pack has a gap that an auditor will find in minutes.
The control plan should name the characteristic, the specification, the sample size, the frequency, the method and the reaction plan. A row without a reaction plan tells the operator what to check and leaves out what to do when the check fails.
| Element | What it should contain | Usual weak point |
|---|---|---|
| Design records | Assembly drawing at the correct revision, with the BOM | Component drawings at mixed revisions |
| Process flow | Every step from goods in to dispatch, including test | Secondary operations left off the diagram |
| Control plan | Product and process characteristics, sample, method, reaction | Sample frequencies with no reaction plan |
| Measurement system analysis | Gauge or fixture study for each critical measurement | Study done on a different gauge from production |
| Dimensional results | All ballooned characteristics with the measured values | Only the characteristics that passed easily |
| Material and component evidence | Certificates traceable to the batch used | Certificates with no batch link |
| Initial process study | Capability for the critical characteristics | A study on a characteristic nobody controls |
The submission level agreed with the customer decides how many elements are supplied, not how many are prepared.
Submission levels and the usual rejections
Submission levels run from a warrant with limited supporting data to a full pack with samples. The level should follow the risk and the maturity of the supplier. A new assembler building a new design needs more evidence than an established line transferring a proven product.
Agree that level in writing before the run. A supplier that prepares a full pack when the customer wanted a warrant has paid for work nobody needed, and a supplier that prepares a warrant when the customer expected a full pack has to start again.
The rejections repeat. A first article built from hand picked components, a control plan that does not match the flow, a torque record without a calibration date, a certificate with no batch link, and a dimensional report that quietly omits the characteristic that was out of tolerance. Each one is avoidable with a checklist and a second pair of eyes before the pack is sent.
Keep the approved pack as the baseline. Every change to a component, a fixture or a process should produce a delta, not a new pack built from scratch, and the delta should be reviewed by the same person who approved the original.
Frequently asked questions
How is an assembly FAI different from a component FAI?
An assembly FAI covers the dimensions that exist only after assembly, plus the process: fixtures, torque, test and work instructions. A component FAI covers one drawing and one process. The assembly pack also has to include evidence for each BOM line, including purchased parts.
Do I need a first article for every component in the assembly?
Not always at the same depth. Critical and custom components usually need a first article or an equivalent approval. Standard fasteners may be covered by a certificate and a sample check. The decision should be recorded, not left to habit.
How many assemblies should be in a first article sample?
Enough to show that the process repeats, not just that one unit can be built. The number depends on the assembly and the risk, and it should be agreed with the customer before the run. Keep the units identified so they can be re-measured later.
What is the most common reason an assembly PPAP is rejected?
A mismatch between the process flow and the control plan, followed by certificates with no link to the parts used. Both are documentation problems rather than manufacturing problems, which makes them avoidable with a review before submission.
Sources and further reading
- IATF 16949:2016, Automotive quality management system standard
- AIAG PPAP manual, Production Part Approval Process
- AS9102, Aerospace First Article Inspection Requirement
Qumek Engineering Desk
Engineering team
The engineering team behind Qumek drawings, DFM notes, tolerances and inspection standards. Every guide is reviewed against the drawings and processes used on live programmes.
Building an assembly approval pack?
Send the assembly drawing and the BOM. The engineering desk will reply with the evidence list and the gaps a reviewer is likely to find.
Send an assembly drawing