ICT And Functional Test For Assembled Boards

Testing an assembled board is a question of what can be verified, at what cost, and at which point in the process. In circuit test checks the electrical continuity and isolation of the connections a board is supposed to have, and functional test checks that the assembled product does what it is supposed to do. The two answer different questions, and a production line needs both unless the design makes one of them unnecessary.

This article compares the two approaches, explains what determines the coverage of each, and describes the design decisions that decide whether testing is straightforward or expensive.

The deciding factor is usually access. A test method that needs a physical connection to every net requires the layout to provide it, and a layout that does not provide it removes the option before the test engineer is involved.

What In Circuit Test Checks

In circuit test, usually abbreviated ICT, places the board on a fixture with spring loaded probes, one per accessible net, and measures the resistance between them. Opens, shorts and wrong components are detected, and the measurement is fast because the fixture contacts every net at once. The technique depends on being able to isolate each component electrically, which requires the probes to drive the nodes around it.

Its coverage therefore depends on access and on the ability to isolate. A net that is not probed is not tested, and a component surrounded by low impedance paths may be difficult to isolate without damaging it or producing an ambiguous result. Capacitors in parallel, for example, cannot be distinguished from each other unless the surrounding network is broken.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/electronic-assumbler.webp" alt="Bed of nails test fixture under an assembled PCB” />

What Functional Test Checks

Functional test powers the board and exercises it, checking that the outputs respond correctly to the inputs. It verifies the design intent rather than the individual connections, and it can find problems that continuity testing cannot: a marginal timing path, a firmware issue, a component that is electrically present but out of specification.

The cost is in the fixtures, the software and the time. Each functional test needs a test program that models the expected behaviour, a fixture that provides the interfaces, and a pass or fail criterion that separates a real defect from normal variation. Where the product has several variants, the test program has to handle all of them.

Coverage And What It Costs

Test coverage is usually expressed as the percentage of the netlist that can be verified, and improving it has a cost that grows quickly. Going from ninety to ninety five percent may require additional probe points that need board area and a more complex fixture; going beyond that may require boundary scan or built in self test.

The decision should be made on the failure modes that matter. A board with a high value assembly and a low tolerance for field failure justifies a more expensive test strategy than a low cost product where a small number of escapes can be absorbed by the returns process.

Test Points And Layout

Every method that touches the board needs physical access, and the layout should be planned for it rather than adapted afterwards. Test points should be on a grid that the fixture can reach, clear of tall components, and away from parts that could be damaged by the probe force. A test point on the underside is often the only option, which requires the fixture to access both sides or the board to be tested in two passes.

Adding test points for every net is not always the right answer. Where the net is already accessible at a connector or at a component pad that the probe can reach, a dedicated point adds nothing. The in process inspection that happens during fabrication also reduces the number of defects that reach assembly, which changes the value of testing every net.

Boundary Scan And Built In Test

Where physical access is limited, boundary scan shifts the testing into the silicon. Devices that support it can drive and sense their own pins, which allows connections between them to be verified without probes. The coverage is limited to the nets between scan capable devices, but on a dense board that is often the majority of the connections that would otherwise be untestable.

Built in self test goes further and moves the test into the product’s own firmware, which allows the functional test to be performed by the device itself. The advantage is that no fixture is needed and the test can be repeated in the field; the disadvantage is that a fault in the test code can hide a fault in the product.

Fixtures, Probes And Mechanical Limits

A bed of nails fixture is a mechanical device and its limits are mechanical. Probe force, pitch and the flatness of the board all constrain how many probes can be used and how close together they can be. A board that warps will contact some probes and not others, which is why the panel and assembly processes affect test yield as well as the electrical result.

The fixture also wears. Probes lose their spring force, the alignment drifts and the contact resistance increases, so the maintenance schedule is part of the test strategy. A test that fails intermittently is more often a fixture problem than a product problem. Visual inspection complements the electrical tests, and the X-ray inspection used for hidden joints addresses the defects that no probe can reach.

Statistical Process Control

The measurements from in circuit test are data, and the data is more useful in aggregate than per board. Tracking the distribution of a component value, of a resistance or of a current draw across a production run reveals process drift before it becomes a failure, and it identifies batches that deserve closer inspection.

Establishing that requires the test system to record results rather than only to pass or fail them, and the effort is repaid the first time a supplier change shows up as a shift in a distribution. The wider incoming and in process quality controls provide the other half of the picture, covering the board before assembly.

Choosing The Strategy

The strategy follows from the product. A low cost, high volume board with simple assembly is often adequately tested by a functional test with a small number of probe points. A dense board with fine pitch devices and a high unit value usually needs in circuit test for structural coverage plus a functional test for behaviour.

The design’s contribution is to make both possible. Providing the test access, keeping the fixture requirements realistic, and allowing a defined test mode in the firmware cost very little at design time and are frequently the difference between a test that is cheap and one that is not.

FAQ

Can functional test replace in circuit test? Sometimes, where the design is simple and the firmware can exercise every interface. Where the board has many unpowered or inaccessible nets, structural testing catches defects that a functional test would pass, only to fail later.

How many test points should be provided? Enough to give the coverage the product needs, on a grid the fixture can reach, and placed during layout rather than added afterwards. In practice this is a decision made with the test engineer rather than alone.

Does a test point affect signal integrity? It can, because it adds a stub. On fast nets the test point should be short and placed close to the line, or the net should be verified by another method.