In-Circuit Test Versus Functional Test: Choosing the Right Gate

In-circuit test and functional test answer two different questions, and the argument about which one to use is usually resolved by deciding what the factory needs to know when a board fails. One test tells you which component is wrong, and the other tells you whether the product works, and a test strategy that provides only one of those answers always leaves a gap.

What Each Test Does

An in-circuit test measures individual components and nets while the board is powered down, using a bed of nails fixture or a flying probe to contact the nodes. It verifies that the right part is present, that it is the right value, that it is oriented correctly and that the nets are connected as the netlist expects. It is a structural test, confirming that the hardware was built as intended rather than that it functions.

A functional test powers the assembly and exercises it as a product, checking that the outputs respond correctly to the inputs across the operating range. It verifies the design as a whole, including the parts of the behaviour that no component measurement can reach. It is also the only test that can confirm the tuning of an analogue circuit or the timing of a digital interface.

Coverage and What Is Left Untested

In-circuit coverage is high for discrete parts and connectors, and it falls away for anything that cannot be isolated or measured in place. A component that is paralleled by another, a net that is loaded by an integrated circuit, or a part whose value drifts with temperature all reduce the practical test coverage that the fixture actually achieves.

Functional coverage depends entirely on how the test is written. A test that checks one operating point gives limited assurance, while one that sweeps the input range, monitors the supply current and checks the timing covers far more. Our short circuit inspection notes describe the faults that both tests are trying to catch. Neither test finds a fault that the design has made undetectable.

Fixture Cost and Lead Time

A bed of nails fixture is a significant investment and takes weeks to build, so it only makes sense where the volume justifies it. The fixture also has to be maintained, because worn probes and a warped board support both produce false failures that erode confidence in the test. Fixture cost is therefore a volume decision as much as an engineering one.

Functional test fixtures are usually cheaper and simpler, because they often need only a connector and a mechanical support rather than hundreds of probes. That makes functional test more attractive for low volume products, where the amortised cost of a nail fixture would dominate the unit price. Where the volume is uncertain, flying probe testing avoids committing to a fixture that may never be amortised. Our flying probe notes describe the alternative for prototype and low volume work.

Access, Nails and Test Points

In-circuit test depends on access, so the design has to provide test points that are reachable, on a sensible grid and not covered by another part. Where the layout provides no access on a critical net, the coverage figure quoted by the test engineer is optimistic rather than accurate.

Test points cost board area and they load the net they are attached to, so they are often removed to save space. The decision should be made with the test engineer rather than after the layout is frozen, because adding access later means a new revision. The same applies to test point size, which has to be large enough for the probe to land on reliably.

Diagnostics and Fault Isolation

The value of in-circuit test is diagnosis. When it fails, it reports a specific component or net, and the operator can go straight to the fault. That speed is worth a great deal on a high volume line where the cost of finding a fault is measured in labour minutes.

Functional test reports a symptom rather than a cause. A board that draws too much current or produces no output has to be investigated, and the investigation may take longer than the original assembly. That difference in diagnostics is often the deciding factor in the choice between the two. Diagnosis time is a real cost on a production line, and it is the cost that in-circuit test removes most directly.

Test Strategy for Low and High Volume

At low volume the usual answer is a flying probe for the structural checks and a functional test for the behaviour, with no dedicated fixture at all. That combination gives reasonable coverage without a large investment, at the cost of a slower test cycle.

At high volume the in-circuit fixture earns its cost, because the cycle time per board is short and the diagnosis is immediate. A functional test is usually added afterwards to cover the behaviour that the structural test cannot see, and the two together form the standard high volume strategy. The order matters, because a structural failure found early is cheaper to correct than one discovered after a functional test has already run.

Programming, Data and Maintenance

Both tests are software as much as hardware, and the software has to be maintained. A netlist change, a component change or a firmware change can invalidate a test, and a test that is not updated becomes a source of false passes rather than a control.

The data the tests produce is useful in its own right. A rising rate of failures at one component position points at the placement machine, and a rising rate of functional failures at one operating point points at a component supplier. Our test coupon notes describe the structural evidence that supports that analysis. That feedback loop only works if the failure data is recorded in a form that can be searched later.

Combining the Two Gates

The two tests are complementary rather than competing, and the usual arrangement is a structural test followed by a functional test, with a defined disposition for boards that fail either one. The structural test catches assembly errors cheaply, and the functional test catches the errors that assembly cannot explain.

Where only one gate is possible, the choice follows from the failure history of the product. A design that has suffered from missing and misplaced components needs the structural test, while a design with a history of marginal performance needs the functional one. Where the product is safety related, both gates are normally required regardless of volume.

Specifying Test Requirements

The test requirement belongs in the design documentation, because the access it needs has to be designed in. The drawing should name the test method, the access required and the coverage expectation, and it should say what evidence accompanies the delivered boards.

Where the customer specifies a test method, that requirement also determines the test points, the connector arrangement and the panelisation. Our design release checklist lists the items worth confirming before the artwork is released. Test access that is promised but never designed in is the most common reason a test strategy slips at the last minute.

FAQ

Can functional test replace in-circuit test? It can detect a fault but it cannot always localise one. For a product where diagnosis time matters, the structural test pays for itself even though it appears to duplicate work.

Is a flying probe an in-circuit test? It performs the same measurements using moving probes instead of a fixture. It is slower per board, which suits prototypes and low volume, and it requires no fixture investment.

How does gopcb support test strategy? We keep the test access the design needs, provide flying probe and fixture based structural test, run functional test to the customer programme where required, and record the results against the serial number for traceability.