Test Point Design For In Circuit Test Fixtures

In circuit test works by pressing a bed of nails against the underside of the board and measuring each net. The nails are spring loaded probes, and each one has to land on a surface that is flat, clean, of a known size and clear of anything that would prevent contact. Those surfaces are test points, and they are designed rather than inherited from the pads that happen to be exposed.

This article explains what a fixture needs from the board, how test point size and spacing are chosen, which nets have to be accessible, and how the finish and the mask affect the result.

What The Fixture Needs

The probe has to make a reliable electrical contact with a defined force, and it has to do so without damaging the board or the components around it. That requires a flat surface of adequate area, a clear approach from the direction of the fixture, and enough support on the opposite side of the board so that the force does not flex the panel and shift the other probes. The board also has to be held in a known position, which is done with tooling holes rather than with the outline.

The mechanical requirements are as important as the electrical ones. A probe that lands on a component instead of a pad will damage the part and produce a false reading, and a probe that has to travel further than its spring allows will not make contact at all. The fixture design is therefore constrained by the height of the parts on the tested side and by the flatness of the board.

Size, Spacing And Grid

A test point needs a probe to land somewhere inside it with a margin for the positional tolerance of the fixture, which is typically plus or minus 0.05 to 0.1 millimetres. A pad of 1 millimetre diameter gives a comfortable target, and 0.8 millimetres is a practical minimum for a standard probe. Below that the probe has to be smaller, and a smaller probe has a shorter travel, a lower spring force and a shorter life.

Adjacent test points need a spacing that matches the probe and the fixture. A standard probe of 1.3 to 1.7 millimetres diameter needs a pitch of about 2.54 millimetres on a grid, and the fixture can be built at 1.27 millimetres with smaller probes at a greater cost. Test points on a regular grid are much easier to design a fixture for than points scattered at arbitrary positions, because a grid allows the fixture to be drilled on standard equipment.

Placement And Access

Test points are placed on the side of the board that the fixture will press, which is usually the underside, and they are placed so that the probe can reach them without passing over a tall component. A probe that has to reach past a connector needs a longer probe or a cut out in the fixture, and both add cost. Where a net is only accessible on the component side, the fixture has to test from both sides, which doubles the cost of the fixture and slows the test.

Clearance around each point matters as much as the point itself. A probe that lands next to a component may damage it, and a point that is under a part cannot be reached at all. A keepout of about 1 millimetre around each test point, free of components and of tall features, is a reasonable starting rule, and the layout rules for component clearance are described under manufacturable design guidelines.

Which Nets Need Access

The nets that must be accessible are the power rails, the grounds, the reset and enable lines, the programming and debug interfaces, and every net that has to be measured to confirm the function of the board. A common approach is to provide access to the power and ground planes and to the test points that the design already has, plus a defined set of signal nets chosen by the test engineer.

Ground access deserves particular attention, because the test equipment needs a low impedance return. Several ground points distributed across the board are better than one, since the return current then has a short path from wherever the measurement is made. The general principles of return path arrangement are described under ground routing and power trace planning.

Mask, Finish And Contamination

A test point is a mask opening over copper, and the copper has to be free of the coating that will be applied later. Where a conformal coating is specified, the test points are masked, and the masking has to be planned as part of the coating process rather than added afterwards. A coated test point gives a reading that drifts and eventually fails.

The finish also matters. A probe pressed onto a gold pad makes contact reliably and leaves little residue, while a probe pressed onto a tin surface or an oxidised copper surface builds up contamination and needs cleaning. Flux residue on a test point is a common cause of intermittent readings, and the cleaning requirement follows from the assembly process rather than from the test itself. The sequence as a whole is described under PCBA development process.

Verification

Verification is done with the first fixture. A board is placed in the fixture and the contact resistance of every probe is measured, which identifies the points that are too small, that are obstructed or that are contaminated. The measurement is repeated after the fixture has been used for a shift, because a probe that loads up with flux residue will drift.

The design check that prevents most of the problems is a review of the test point list against the fixture drawing before the board is released. It takes an hour and it catches the point that is inside a courtyard, the net that has no access and the pad that is smaller than the probe. Finding them at that stage costs nothing, while finding them after the panels are etched costs a layout revision and a new set of boards.

Flying Probe And Functional Test

A bed of nails is not the only way to reach a net. A flying probe moves a small number of heads over the board and touches each point in turn, which needs no fixture and no dedicated tooling, and it can test a design that is still in development. The penalty is time, since the heads move sequentially, so flying probe is used for prototypes and low volume while the fixture is used where the volume justifies it. Both need the same thing from the board: an accessible, reasonably sized surface on each net.

A functional test does not necessarily need probe access at all, because it exercises the board through its connectors. That is a different requirement, and a board that is designed for functional test only will often have no test points and will be impossible to diagnose when a unit fails. The practical compromise is to provide probe access for the power and ground structure and for the nets that a functional test cannot isolate, and to rely on the connectors for the rest.

FAQ

Can a component pad be used as a test point? It can be probed, but a solder joint is domed and its position is less certain, so the probe may slip. A dedicated flat test point is more reliable where the net must be tested on every board.

How many test points are needed? Enough to confirm the function and to isolate a fault, which in practice means the power rails, the grounds and the nets of any circuit that can be tested independently. The list is chosen by the test engineer with the designer.

Does a test point affect signal integrity? A short stub added to a high speed net does, and the stub should be kept short or moved to a point where the reflection is harmless. On a slow net the effect is negligible.