Shorting Plate Pinpoints Probe Deficiencies

Steve Brenne, RadiSys -- Test & Measurement World, 2/1/2000

The main reason for doing fixture maintenance, or probe replacement, is to maximise the productivity of your test process. As test-head probes begin to degrade, the rate of false test failures increases. These false failures either require time spent retesting failing boards, debugging the failures, or debugging the fixture or tests. In some cases, just to maintain throughput you can even end up adjusting the tolerance of a test to the point that the test’s effectiveness is reduced. In extreme cases, you might even disable a test if it’s not immediately clear that the cause of failure is poor contact due to failing probes. In these instances, the time you spend performing a controlled maintenance of your fixture can result in an overall increase in productivity and also result in higher quality tests.

For these reasons, we, at RadiSys, developed a technique to perform preventative maintenance on the fixtures for our HP 3070 ATE. Specifically, the technique enables us to identify and replace failing probes in the test head.

In a pristine controlled environment, a probe would last for one million cycles before failure. However, in today’s manufacturing use there are a number of factors that contribute to a probe’s degradation. Typically, this degradation shows up as an increase in the resistance between the probe and a board-under-test. The increase in resistance can result from:

Contamination, such as dust or flux in your manufacturing environment that affects the probe’s internal construction or its tip.

Wear that results from side forces during its travel, which can in turn deform the probe and reduce its working force. With less force, a probe will not be as effective in penetrating through oxide layers on printed circuit board test points.

Heavy Handed Maintenance
One method you can use to maintain ATE fixtures is to replace all probes after a certain number of test cycles. While this method is effective, there are also some trade-offs. Firstly, probes in a fixture do not degrade at the same rate. Changing all probes means that you may be ditching some probes that would have lasted much longer. Equally, there may also be a few probes that should have been replaced even before the maintenance cycle. Of course, you can simply wait until a probe breaks and then replace it.

Maintenance based on test cycles is only practical for high-volume manufacturing. In low-volume work, a fixture may not have enough use to justify setting up a maintenance cycle. In this case, factors such as contamination have more impact on probe life than cycle count. And, fixture maintenance may accompany a daily battle in getting boards to pass, which in the long run can result in lower test quality.

Probe Maintenance — On a Plate
At RadiSys, we developed our probe maintenance method for manufacturing in a low-volume, high-product-mix environment. The method is equally beneficial for high-volume, low-mix work. The method involves developing a unique test for each fixture that identifies individual probes that measure above a preset resistance threshold. Using the method, fixture maintenance becomes a controlled part of your process. When a probe fails, the method reports its location referenced to the test-head interface pin.

Our method involves using a shorting plate in place of a board-under-test (see Figure 1). A test run then performs simple resistance measurements between various probes. We select sets of three probes — designated common probes — in a fixture that are wired to each module of the HP 3070 ATE. We choose these common probes to be on different channels of each of the ATE’s test modules in order to avoid multiplexing conflicts. By using simultaneous equations, you can establish the individual resistance of each common probe (see “Modelling Your Test Fixture”).

Figure 1. Replacing the printed circuit board shown in this view by the shorting plate allows you to perform maintenance tests on individual probes.

The test run also measures the resistance between each of the remaining probes wired to a particular module and one of the common probes, depending upon multiplexing. You can now establish the approximate resistance of all probes by simply subtracting the common probe resistance. The test program delivers an error message for any probes with resistance measurements above a preset threshold. A key feature of the method is the test program that automates this preventative maintenance. Our software takes a few fixturing files as input and produces all the files we need for this test.

Plate Material Under Review
Another key to the success of our method is in selecting shorting plate material. Currently, we use solder-plated copper-clad boards. We expect these boards to wear out in due course. After repeated test cycles the probes will start to penetrate the copper cladding and contact the printed circuit board substrate, resulting in poor contact. We are exploring other shorting plate options.

In addition, we are using a similar technique to verify HP 3070 interface pins on the test head. The verification process will involve using the test-head diagnostic fixture. T&ME

Steve Brenne is a senior test engineer with RadiSys, having previously worked in test with Hewlett-Packard’s LaserJet Division.

Sidebar