Why an incoming inspection plan pays off #
A component passes through several hands before it reaches your line: a franchised distributor, a broker, a reeler, a tester, maybe a second broker. Each hand is a chance for the lot to change character without changing its label. The fast way to lose money is not a headline shortage — it is a quietly wrong lot that ships accepted, gets built into a board, and fails at functional test or in the field.
An incoming inspection plan is the repeatable gate that catches those lots early. It is not a counterfeit witch-hunt on every reel, and it is not a full lab teardown on every shipment. It is a graded set of checks whose depth scales with the risk of the part, the source, and the consequence of a failure. Done well, it costs little on the safe lots and concentrates effort where it matters.
Start with a sampling level, not a full teardown #
The first decision is how many pieces to look at. Inspecting 100% of a high-volume passive shipment is wasteful; inspecting one piece of a flight-critical active is reckless. Most receiving teams anchor on a published sampling standard such as ANSI/ASQ Z1.4 (the U.S. adoption of the older MIL-STD-105 sampling plans), which maps lot size and an accepted quality limit (AQL) to a sample size and an accept/reject count.
The useful move is to hold a default level for trusted franchised stock and a tighter level for open-market or broker-sourced lots. When the source is unknown, when the part is on an active shortage list, or when the application is safety- or field-return-sensitive, pull a larger sample or move to 100% screening. The sampling level is itself a risk decision, not a formality.
External visual inspection: the first and cheapest gate #
Most problems that matter show up under a microscope before any test equipment is warmed up. The criteria live in references such as IPC-A-610 (acceptability of electronic assemblies) and IDEA-STD-1010 / AS6081 for component-level visual acceptance. What a buyer or inspector is actually looking for:
| Check | What it reveals |
|---|---|
| Top-marking and laser etch | Remarking, font mismatch, re-lasered date codes, inconsistent logos |
| Lead or ball condition | Scratches from reballing, bent leads, oxidation, residual flux |
| Package body and edges | Mold-flash, re-ground edges, non-factory tooling marks |
| Coplanarity | Leads or balls not flat — a reflow or placement risk |
| Label and barcode | Mismatch between reel label and device marking, duplicate lot numbers |
For a small-outline device such as the TI TPS54331DR buck regulator, the visual gate is fast: confirm the marking, lead form and coplanarity, and that the reel label matches the device. For a fine-pitch BGA such as the AMD Xilinx XC7Z020-2CLG484I, the bare eyes see little — the ball array hides the joints, which is exactly why the next step exists.
Dimensional and package checks #
Where the package matters to placement or mating, measure. Calipers or a vision system confirm body length, width, height and lead pitch against the datasheet outline drawing. A part that is mechanically correct but electrically wrong will pass here; a part that is the wrong package or a cloned outline will not. This step is cheap insurance for any lot where the marking looks plausible but the source is unfamiliar.
When to bring in X-ray #
X-ray inspection is the one tool that sees inside a sealed package without opening it. For BGAs and other area-array devices it is the standard way to check:
- solder-ball size, shape, coplanarity and missing or bridged balls;
- wire-bond placement and looping inside QFN or leaded packages;
- internal die placement, lead-frame alignment and unexpected voids or fillers;
- signs of a reballed or reworked device (uneven ball metallurgy, added substrate).
A Zynq-class BGA is a good example of where X-ray earns its place: a remarked or reballed part can look perfect on the surface and fail only at reflow or under thermal cycling. The decision rule is simple — if the joints are hidden and the lot carries any sourcing risk, X-ray before build, not after.
Electrical test: how far to go at incoming #
Electrical testing at incoming has two very different depths, and conflating them wastes money. A basic electrical test confirms the device powers up and meets a few key parametric limits — continuity, quiescent current, a switching or bias point. It catches dead, severely out-of-spec or wrong-die parts. A full functional test replays the device's real application conditions and is usually what the original manufacturer or a dedicated test house does, not the receiving bench.
For most incoming plans, basic electrical screening plus visual and X-ray is the right ceiling. Reserve full functional or curve-trace comparison for high-risk lots, high-value parts, or when a specific failure mode is suspected. The point is to catch the disqualifying defects, not to re-qualify the silicon.
Escalating to destructive physical analysis #
Some questions only a torn-open part can answer: is the die the right one, is the marking inside the package consistent with the top mark, are the bond wires and die-attach authentic? That is destructive physical analysis (DPA) — decapsulation to view the die, cross-sectioning to inspect bonds and metallization, and comparison against a known-good device or the manufacturer's published die photo.
DPA is slow and destroys the sample, so it is the last rung, used when visual, X-ray and electrical checks already raise a flag, or when the lot value and risk justify the cost. Treat a clean DPA as the final confirmation of a suspect lot, not a routine incoming step.
Tie it back to date code, MSL and traceability #
The inspection plan does not stand alone. Three surrounding facts change how hard you look:
- Date code and shelf life — an old code is not automatic rejection, but it triggers solderability verification (per J-STD-002) before reflow, and it changes how much you trust the lot's history.
- Moisture sensitivity — a moisture-sensitive package that exceeded its floor life needs baking before use; the inspection record should note MSL and opening time, not just pass/fail.
- Traceability — every inspected lot should carry its certificate of conformity, lot number and source documentation so a future failure can be traced back. Inspection without a paper trail is a snapshot you cannot act on later.
Conclusion #
A working incoming inspection plan is a graded ladder, not a single test. Sample by risk using a published plan, run external visual as the cheap first gate, add X-ray for hidden-joint parts like BGAs, keep electrical testing to basic screening at receiving, and escalate to destructive analysis only when something already looks wrong. The goal is not to prove every part perfect — it is to spend inspection effort where a bad lot would actually hurt, and to catch the wrong lot before it reaches the board.