Test Point Design Guide

A test point is a piece of copper that exists only so that a machine or a person can touch the circuit. It takes no part in the function of the product, it costs area on every layer it passes through, and it is the difference between a board that can be tested in minutes and one that has to be debugged by hand. Test points are cheapest when they are placed during the layout, and most expensive when they are added afterwards.

What Test Points Are For

The first purpose is the in circuit test, where a fixture presses a probe onto each net and measures the components and the connections around it. The test needs access to every net that is not already exposed by a component pad.

The second purpose is debugging. A board that has to be probed with a scope or a meter needs somewhere to put the probe tip, and a via or a component pad is a poor substitute because the probe slides off and shorts to the neighbour.

The third purpose is programming and calibration, where a fixture or a connector has to reach a set of signals in a defined order. Those signals are usually brought to a header rather than to individual pads.

All three purposes are served by the same design decision: put accessible copper on the net, in a place where the fixture can reach it, and document it so that the fixture can be built.

Bed of Nails and Fixture Access

A bed of nails fixture carries a spring loaded probe for every test point, mounted in a plate that is drilled to match the board. The board is pressed down onto the probes, and the force has to be carried by the board without flexing it into the probes on the other side.

The probe has a travel of a few millimetres and a rated force per probe, so a board with two thousand test points needs a considerable total force. The fixture and the board support have to be designed together, and a board with no support in the middle will bow and miss contacts.

Access is decided by the side that the fixture probes. A single sided fixture only reaches one face, so test points on the other side have to be duplicated or the fixture has to be a double sided one, which costs more.

The probe also needs a clear vertical path. A tall component beside a test point will block the probe, and this is the mistake that turns a working layout into a fixture redesign.

Placement and Probe Clearance

Clearance is measured from the test point to any component that stands above the board, and to any neighbouring test point. A probe has a body of a few millimetres in diameter, so the spacing between test points is usually set by the probe rather than by the electrical rules.

The usual grid is 2.54 mm between test points on a single sided fixture, with a smaller pitch possible where the fixture uses a fine pitch probe. Component bodies, connectors and board edges all reduce the available area.

Test points on the underside are the most useful, because the top side carries the components. Where the layout allows, the underside of a via or a pad is a good test point, provided that the mask is opened over it.

The mask opening is often forgotten. A test point under solder mask is not a test point, and the error is found only when the fixture is built. The inspection methods used in production also depend on the same access.

Size and Shape

A test point pad is usually a square or a round pad of about one millimetre, and it is larger than a normal via because it has to accept a probe tip that is not perfectly aligned. A pad that is too small gives a contact resistance that varies with the pressure.

The pad should be solid rather than a ring, unless the net is a via that has to stay open for another reason. A ring gives a smaller contact area and the probe can sit in the hole, which is a common cause of a false failure.

Where several test points sit together, a square pad packs more densely than a round one and gives a larger contact area for the same pitch. The corners of a square pad should be rounded to avoid a sharp point that lifts the mask.

The solder finish on the pad has to be one that a probe can contact repeatably. A thick gold finish is the best for a fixture, while a rough tin finish oxidises and raises the contact resistance over the life of the fixture.

Electrical Effects

A test point is a stub of copper attached to a net, and a stub is a discontinuity. On a high speed net the stub adds capacitance and creates a reflection, and on a low level analog net it is an antenna for noise.

Where the net is sensitive, the test point should be as short as possible, placed at the end of the trace rather than in the middle, and connected with a short neck. A long thin trace to a test point is the worst case.

On a differential pair, both halves should have a test point or neither, so that the symmetry is preserved. A test point on one side only converts a differential signal into a common mode one, which is a frequent source of radiated emissions.

The fixture itself is also a load. A probe adds capacitance to the net while it is in contact, and the measurement has to be interpreted with that in mind. The flying probe method avoids the fixture but has the same electrical effect.

Design Review

The review starts with the net list and the test coverage requirement. Every net that has to be tested should have a test point or an accessible pad, and the coverage should be counted rather than assumed.

The second item is the mechanical access: the probe path, the board support and the fixture cost. A layout that needs a double sided fixture and a custom support is more expensive than one that does not.

The third item is the documentation. The test point list, with coordinates and net names, is what the fixture house needs, and it should come from the same data as the artwork. The test coupon is a separate feature and does not replace the test point list.

The last check is a trial fixture run on the first article, where the probes are checked for reliable contact. A test point that fails the trial is a layout change, and the release checklist should not be signed until the fixture passes.

FAQ

Can a via be used as a test point? Yes, if the mask is opened over it and the pad is large enough for the probe. A tented via cannot be probed, so the tenting rule has to exclude the test points.

How far apart should test points be? Usually 2.54 mm for a single sided fixture, set by the probe body rather than by the copper spacing. A fine pitch fixture allows less.

Do test points affect signal integrity? They add a stub and a capacitance, so on a high speed net the placement and the length of the connection both matter. The stub should be as short as the layout allows.