PCB Electrical Test: Coverage and the Cost of Missing It

Every board that leaves a fabricator is tested for continuity and isolation, and the test is treated as a formality by most designers. It is not. The cost of a defect depends enormously on where it is found, and the electrical test is the last cheap opportunity to find one. Understanding what the test can and cannot detect also determines what the designer has to provide.

What the Test Detects

The electrical test verifies two things: that every net is continuous from end to end, and that no two nets are connected to each other. An open inside the board cannot be seen by inspection, and a short between inner layers cannot be repaired, so the test is the only mechanism that separates a good board from a bad one before assembly.

Leakage is the third property, and it is the most difficult. A resistive path between two nets that should be isolated may not register as a short at the test voltage but will degrade a high-impedance circuit in service. Testing for it requires a measurement rather than a continuity check, and the decision to do so depends on the application.

The Rule of Ten

The economics of defect detection are usually described by a rule of thumb in which the cost of correcting a defect grows by roughly an order of magnitude at each stage of the process. An open detected on the bare board may cost a repair or one scrapped panel. The same open discovered after components have been assembled costs the components, the assembly labour and the rework. Discovered after the product has been shipped, it costs the field service, the customer relationship and possibly the reputation.

That progression is the argument for testing early and testing thoroughly. It also explains why a fabricator will require test access even where the designer has not considered it, and why the cost of providing the access is almost always smaller than the cost of the failures it prevents.

Flying Probe and Bed of Nails

The two common methods have different economics. A flying probe moves a small number of probes across the board under program control, testing one point at a time. It requires no fixture, which makes it ideal for prototypes and small batches, and it can reach fine-pitch features that a fixture could not. The trade-off is speed: a dense board may take minutes per panel.

A bed of nails uses a fixture with a pin for every test point, so the whole board is tested at once. The fixture is expensive and takes time to build, which is why it is used for volume production, but the test itself takes seconds. The fixture also constrains the design, because the pins must reach every net from one side and must be positioned to avoid tall components. Where a design has no room for test points, the fixture cannot be built.

Test Coverage as a Design Requirement

Test coverage is the proportion of the nets that can be accessed, and it should be a design target rather than a result. Every net that must be tested needs a point that a probe can reach, with clearance around it for the probe or the pin. Where a net is accessible only through a via on the opposite side of the board, the fixture design may become impossible rather than merely inconvenient.

The layout should therefore reserve space for test points during placement rather than after routing, because the access features are as much a constraint as the components. Where the space genuinely does not exist, the designer should record which nets are not testable, so that the consequence is understood rather than discovered. The geometry around the test points follows the same rules as pads in general, which are described in this article on PCB pad design standards.

Agreeing the Test Specification

Because a fabricator is usually required to test every board, the test conditions should be agreed in the purchase specification rather than assumed. The relevant items are which nets are tested, the voltage and current used, the acceptance threshold for isolation resistance, whether leakage is measured, and how a failure is reported and handled.

Those items matter because the same board can pass one test specification and fail another. A continuity check at low voltage will not reveal a marginally resistive connection, and a test that includes leakage will find boards that a simple continuity test passes. Agreeing the specification removes the ambiguity, and it also makes the test data comparable between batches. The relationship between the test specification and the rest of the design data is described in this discussion of PCB design and fabrication.

Testing Beyond the Bare Board

Assembly introduces its own defects: solder joints that are open, components that are missing or reversed, and connections that are weak rather than absent. The bare board test does not detect any of them, which is why in-circuit and functional tests exist. The design implication is the same as for the bare board: the access has to be provided in the layout, and the test method has to be chosen before the design is released rather than after the first batch is built.

Where the product is manufactured in volume, the cost of the test access is almost always lower than the cost of the scrap it prevents. Where the product is a prototype, the access may not be needed, but the nets that cannot be tested should still be recorded so that a diagnosis problem in the next revision does not come as a surprise. The wider question of what a prototype must verify is discussed in this article on multilayer prototype requirements.

Containing the Repair Cost of Late Detection

The full repair cost of a defect is not the price of the rework itself. It includes the labour to locate the fault, the risk of damaging an adjacent joint while removing a component, the second test pass, the scheduling disruption and, if the board has already been assembled into a product, the cost of teardown and reassembly. That figure climbs sharply the further the defect travels from the process that created it, which is why bare board test exists at all. A short between two nets is trivial to find before components are placed and awkward afterwards.

Coverage decisions should therefore be made on consequence, not on convenience. Nets that carry power, nets that would be damaged by a wrong voltage, and nets whose failure would be hard to observe in final test deserve the most attention. Adding a test point to a dense board costs area and may force a layout compromise, but leaving a critical net unverified simply moves the cost to the customer. The rule of ten describes the escalation, and the only reliable way to interrupt it is to test earlier and more completely than feels strictly necessary.

FAQ

Which test method should be used? Flying probe for prototypes and small batches, because it needs no fixture and can reach fine-pitch features. Bed of nails for volume production, where the fixture cost is amortised over many boards and the test time matters. The choice follows from the volume and the density.

What limits test coverage? The accessibility of the nets. Every net that must be probed needs a reachable point with clearance around it, and on a dense board the space for those points has to be reserved during placement. Nets that cannot be accessed become untestable and cannot be diagnosed if they fail.

Why is early detection so much cheaper? Because the cost of a defect grows at each stage it survives. A fault found on the bare board costs a repair or a panel; the same fault found after assembly costs components and labour; found in the field, it costs service, transport and reputation. Each stage multiplies the previous one.