Flying Probe Test Point Design and Accessibility

Flying probe testing needs no fixture, which makes it attractive for prototypes and low volume, but it needs physical access that a bed of nails fixture would have provided through its own pins. Test points that are too small, too close together or covered by a component reduce coverage, and coverage lost at design stage cannot be recovered at test. This article covers the access rules and the layout practice that keeps them workable.

What a Flying Probe Needs

A flying probe places two or more moving probes on the board under machine control, so every net it measures needs a surface it can land on. The probe has a finite tip radius, a finite positional accuracy and a finite approach angle, and all three constrain the test point.

Because the probes move rather than being fixed, the access requirement is different from a fixture. There is no pin field to design, but there is a need for clearance around each point so that the probe can approach without touching a neighbouring component.

Test Point Size and Shape

A test point of 0.8 to 1.0 mm diameter is comfortable for most flying probes, and 0.6 mm is workable where the layout is tight. Below that the probe tends to slide off the pad, and the mark it leaves becomes a scrap criterion of its own.

Where a pad is used as a test point, it should not also be a solder joint, because probing a pad that carries paste disturbs the deposit. Square pads are measured accurately, while round pads tolerate slight positional error better, so the choice follows from the probe accuracy.

The probe leaves a witness mark, and the acceptance of that mark belongs in the drawing rather than in a discussion after the first boards are tested.

Spacing and Clearance

Adjacent test points need enough spacing for the two probes to land without touching each other, and a common rule is a gap of at least 1.5 times the probe tip diameter between neighbouring points. Components beside a test point need a keep-out of at least the probe diameter plus a margin.

Tall components are the practical obstacle, because a probe has to approach at an angle and a part standing 3 mm above the board shadows the area behind it. The keep-out around tall parts is larger than around a chip component of the same footprint.

Access from Both Sides

Double sided access doubles the available area and lets the design place test points on the underside where the top is crowded. The fixture or the machine holds the board on a frame rather than on a bed, so the underside remains reachable.

Where a board is tested from both sides in one programme, the handling has to turn it without damaging the components already placed. That constraint is considered at design stage, because a double sided test on a board with tall parts on both faces is difficult to automate.

Double sided access is a design decision rather than a test decision, since the test points have to exist on both faces before the programme can use them.

Coverage and What It Costs

Coverage is the proportion of nets that can be measured, and the figure that matters is the coverage of the nets whose failure would be dangerous rather than the average across the board. A power net without a test point is a larger risk than a signal net with one.

Where coverage cannot be complete, the gap is documented and the missing nets are covered by functional test or by inspection. Leaving the gap undocumented means that a later change can remove the last measurement without anybody noticing.

Layout Rules That Preserve Access

The rules that help most are a minimum test point size, a minimum gap between points, a keep-out around tall components and a requirement that every test point be visible from the machine’s camera. Those four rules, applied as design rules rather than as review comments, keep coverage high without a special effort.

They should be checked by the design rule checker rather than by eye, because a layout that satisfies them at the schematic stage can violate them after a late component change. Where the check is automatic, the coverage figure can be predicted before the board is built.

Probing Damage and Its Acceptance

A probe mark on a pad is normal, and the question is whether it is acceptable on the finished product. On a pad that will be soldered, a mark is usually harmless; on a gold contact or on an exposed trace, the mark can be a defect.

The acceptance is written on the drawing, with the pads that may be probed identified and the pads that must not be probed excluded. A blanket note that permits probing everywhere is easy to write and produces arguments at inspection.

Programme Design and Test Time

Test time on a flying probe is dominated by travel, so the order in which nets are measured matters as much as the number of probes. Grouping the points into a route rather than measuring in netlist order can halve the time on a dense board.

The programme also sets the probe force and the settle time, and both affect the mark left on the pad. Where the board is thin or the fixture is flexible, a higher force pushes the board down and the probe lands short, which reads as an open circuit.

Documentation

The drawing should state the minimum test point size, the spacing rule, the components that require a keep-out, the pads that may be probed and the pads that may not, and the acceptable witness mark. Those items are what the test house needs to build a programme without a discussion, and they are the same items that make coverage predictable. Continuity of the layout itself is covered in our fabrication notes.

Accessibility is the property the layout either provides or withholds, and it is best measured by the number of nets reachable from at least one side of the board. A crowded top side over an empty underside is usually a board whose test points were placed late in the layout work.

Test points arranged on a grid rather than scattered make the programme faster, because the machine travels between neighbouring points without long moves. The gain is measurable in test time on a dense board and it costs nothing at layout stage.

Where a net is accessible only on a via, the via has to be untented on that side, and that decision belongs to the fabrication drawing rather than to the test department. A tented via that was expected to serve as a test point is discovered when the programme is written.

Probe force is set in the programme and it interacts with the support the board receives. A board held only at its edges deflects under the probe and the contact is lost, which reads as an open circuit rather than as a mechanical problem.

Where one board is tested on two different machines, the test points have to satisfy the tighter of the two access requirements. Writing the rule for the best machine on the site produces a layout that cannot be tested on the second one.

FAQ

Can a component pad be used as a test point? Yes where it carries no paste, but a pad that is also printed with paste is a poor probe target because it disturbs the deposit.

How small can a test point be? Below about 0.6 mm the probe begins to slide off the pad, and the witness mark becomes the limiting factor rather than the electrical measurement.

Why is coverage measured on critical nets? Because an average across the board hides the case that matters, which is a power or safety net with no accessible point.