Test Point Design: Coverage, Placement and ICT Access

Test points are the interface between the board and the equipment that verifies it. They are cheap to add during layout and expensive to add afterwards, and their placement determines whether a bed of nails fixture can reach every net or whether the test engineer has to build a different strategy. Test point design is therefore part of the layout constraint set, not an afterthought once the routing is finished.

What Test Points Are For

A test point is not needed to build the bare board. Fabrication testing of the unpopulated board uses a fixture or a flying probe that contacts the copper directly, and at that stage every pad and via is accessible. Test points become necessary once the board carries components, because the probe can no longer reach the copper without touching a part.

Probing a surface mount component with a bed of nails risks cracking the part or the joint, and the measurement is unreliable because the probe may not make a clean contact. A flat test point gives the probe a defined surface, with a known position and a controlled force, which makes the measurement accurate and repeatable.

Test Point Types

The simplest test point is a dedicated copper pad, either a small rectangular pad or a circular via with a defined annular ring, which the fixture pin can contact. Where the design cannot spare the area, existing plated through holes can serve as test points if their position and surface are suitable, and component pads are sometimes used where the geometry allows.

A dedicated test point is preferred because it can be placed at a suitable pitch and kept clear of components. The pitch is set by the fixture: the probe pins have a minimum spacing, and if the test points are closer than that the fixture cannot be built.

Coverage is the first decision, and it determines how many of these points the layout will need.

How Much Coverage Is Required

The traditional target was that a high proportion of the nets should be accessible, often expressed as a percentage of the total. That target reflects the capability of the fixture and the cost of the test: fewer accessible nets means less fault coverage and more reliance on functional test. In practice the requirement depends on the product, the volume and the consequence of a defect reaching the customer.

Coverage should be decided early, because each net that needs a test point constrains the placement around it. Nets that carry sensitive high-speed signals may be excluded deliberately, because a test point adds capacitance and a stub, and the loss of signal integrity outweighs the test benefit. Power and ground nets usually need only limited access, while the nets that connect to connectors and power stages often need full access.

Placement Rules

Test points on one side of the board are easier to handle than test points on both sides, because the fixture then only needs one probe field. Where both sides must be probed, the fixture becomes more complex and the board handling more involved, so the design should keep the test points on the side with the fewest components if the geometry allows.

Spacing is the hard constraint. Probe pins have a body diameter larger than their tip, and adjacent pins need clearance for their bodies as well as their tips. Test points should also be kept away from tall components, which can prevent the fixture from seating, and away from the board edge, where the fixture frame supports the board.

Electrical Effects of a Test Point

A test point is a small area of copper attached to a net, and electrically it is a capacitance and, if it is connected by a length of trace, an additional stub. On a slow digital or analog net the effect is negligible. On a high-speed net it can be enough to degrade the channel, which is why high-speed nets are often excluded from probe coverage and verified by other means.

The parallel stub should be kept as short as possible where a test point is required on a fast net, and the point should be placed as close to the point being measured as the layout allows. Where a net carries a differential pair, probing one half of the pair introduces an imbalance and should be avoided.

Where test points cannot provide the coverage, other methods take over, each with its own limits.

Alternatives and Their Limits

Boundary scan testing uses the device itself to drive and sense its pins, and requires no physical access. It gives good structural coverage on nets that are connected to boundary scan capable devices, but it cannot test analog nodes or nets that are not part of the scan chain. Optical inspection verifies solder joints and component placement but does not measure electrical behaviour. X-ray inspection covers hidden joints under area array packages, which optical inspection cannot see.

None of these methods replaces in-circuit test completely. The practical strategy is usually a combination: in-circuit test for the nets that can be accessed, boundary scan for the dense digital interfaces, and optical or X-ray inspection for the joints that cannot be probed. Deciding the mix belongs in the quality plan for the board rather than in the test department alone.

Documenting Test Points

The test strategy has to be visible in the documentation. Test points should be identified on the assembly drawing, and their positions supplied in the placement data if the fixture is programmed automatically. Nets that are deliberately excluded from test coverage should be listed with the reason, so that a later reviewer does not treat the omission as an error.

Because test point placement is a layout constraint, it should be part of the placement review and of the general manufacturability review before the artwork is released. Adding test points after the routing is complete means moving components, and that is where the cost of the decision appears.

Coverage Targets and Probe Access

Network coverage is the share of nets that the fixture can actually contact, and it is the number that determines how much of the assembly can be verified without functional test. A common target for a mature product is 95 percent of nets, with the remainder excluded deliberately: power nets already verified by measurement, nets that carry a clock or a high-speed differential pair, and nets that exist only as a short internal connection. Tracking network coverage as a design metric, rather than discovering it when the fixture is quoted, keeps the decision in the layout phase where it is cheap.

Probe access is the physical side of the same problem. A probe needs a target of at least 0.9 mm on 2.54 mm centres for a conventional fixture, with a keep-out that stops the probe body from colliding with tall components, connectors or shields. Probe access also has to survive assembly: a test point under a component, or one covered by a label, provides no access at all. Where board area is tight, distribute the probes across the available surface rather than clustering them, and reserve a clear band along at least one edge for the fixture frame and its alignment pins.

FAQ

Do all nets need a test point? No. Coverage requirements depend on the product. High-speed nets are often excluded deliberately because the extra capacitance and stub degrade the channel more than the test benefit justifies.

Can a via be used as a test point? Yes, if the annular ring is large enough for reliable contact and the position suits the fixture. A dedicated pad is preferred where space allows.

Why not probe both sides of the board? A double-sided fixture is more complex and increases handling. Keeping test points on one side simplifies the fixture and reduces the risk of damage during test.