Test Strategy: Combining AOI, X-ray, ICT and Functional Test

No single test finds every defect, and every test adds cost. The engineering question is not which method is best but which combination covers the failure modes of this product at an acceptable cost. Answering it requires knowing what each method can and cannot see, and where the boundaries between them lie.

What Each Method Covers

Automated optical inspection covers the visible surface: missing parts, wrong polarity, offsets, bridges and solder volume on exposed joints. X-ray covers what is hidden, including ball grid array joints, barrel fill and voids under large pads. In circuit test covers electrical continuity, component value and orientation on accessible nodes. Functional test covers the product as a system.

The four methods overlap in coverage but differ in fault isolation. Optical inspection tells you which joint looks wrong, in circuit test tells you which node is open, and functional test usually only tells you that the product does not work. Diagnostic value, not just detection, should drive the choice.

Where the Coverage Gaps Are

Gaps appear at the boundaries. A cold joint under a package is invisible to AOI and may pass a continuity measurement while still being unreliable. A marginal component value may pass in circuit test at room temperature and fail when warm. A functional test can pass while a latent defect remains.

Mapping the failure modes of the product against the methods is therefore the starting point. The mapping is a table with fault classes on one side and test methods on the other, and the empty cells are where escapes will come from. Our article on X-ray inspection of area arrays describes one of the largest cells in that table.

Sequencing the Tests

Order matters because early detection is cheaper. Optical inspection immediately after reflow catches assembly errors before the board is handled further, and in circuit test after that catches electrical faults before the board is assembled into a housing. Putting functional test first wastes time on boards that a cheaper test would have rejected.

Test content should also be deduplicated. Measuring a component value with in circuit test when the same measurement is made functionally adds cost without adding coverage, while measuring continuity on a net that functional test exercises directly may be redundant. The review should remove overlap rather than accumulate tests.

Coverage Numbers and Their Limits

Coverage is usually quoted as a percentage of nodes or of fault classes, and the number is only meaningful if the fault classes are defined. A statement of ninety per cent coverage says nothing about which ten per cent is missing, and the missing part is what matters.

The number also depends on the process. A stable process with a low defect rate needs less coverage than a new product on a new line, and the strategy should be reviewed as the process matures. Starting with full inspection and reducing it on evidence is a defensible path.

Cost of Test and Volume

Fixed costs dominate in circuit test because of the fixture, while functional test costs scale with the time each board occupies a station. At low volume, a flying probe or a manual functional test is often cheaper; at high volume, a fixture and an automated functional test pay back quickly.

Tooling cost also depends on how early the design is frozen. Building a fixture before the layout is stable produces a fixture that is obsolete and an expensive change. Our note on flying probe testing covers the strategy for the period when the design is still moving.

Diagnostic Value and Repair

When a board fails, the value of a test is the time it takes to find the cause. A test that reports a failing node saves an hour of debugging, while a functional test that reports only that the output is wrong may cost several. This is the strongest argument for maintaining in circuit test even where functional test covers the same nets electrically.

Repair also feeds back into the process. Recording the failure class and the root cause turns test data into process data, and the analysis method is the same as the one described in our article on first pass yield analysis.

Reviewing the Strategy

The strategy should be reviewed when the product changes, when the process changes and when the field failure rate changes. A fall in first pass yield indicates that a gap has opened, and a rise in escapes indicates that a test has stopped working rather than that the process has improved.

Documenting the strategy as a table with coverage, cost and diagnostic value makes the review concrete. Where a change is proposed, its effect on the table can be predicted and then measured, which is far more reliable than adding a test because a defect was found once.

Process Control and Verification

Running a first article through the same checks as the production panel confirms that the two agree, and that comparison is the cheapest form of process control available at prototype stage. Keeping a sample from the panel turns a dispute into a measurement, because the same coupon can be re-examined by both parties without rebuilding the batch.

Reviewing the design before the data is released is cheaper than correcting it after the panel is in the tank, because every step downstream inherits the decision made at the front end.

Checks Before Release

The checks that matter are the ones performed on the product rather than on a sample kept for the purpose, because a coupon that travels with the panel is the only evidence about that panel. Where a requirement can be measured, it should be measured at the point of manufacture and recorded against the board or the lot it applies to.

FAQ

Can functional test replace in circuit test? It can detect that a product fails, but it rarely identifies where. Where repair cost matters, in circuit test remains valuable for its diagnostic speed.

Is X-ray needed on every board? Only where hidden joints are present or where the product is safety critical. Sampling becomes viable once the process is stable and the void distribution is known.

How often should coverage be reviewed? At each product revision and whenever the field failure rate changes. A strategy that is never revisited tends to accumulate tests rather than improve coverage.