Merging Flying-Probe and Fixture-Based In-Circuit Test Systems

For decades, manufacturers of loaded electronic circuit boards have relied on the capabilities of in-circuit test (ICT) systems to provide a fast and efficient means of finding component and manufacturing faults prior to functional test. ICT also has been used effectively to provide feedback information for process control.

The continued use of ICT, however, is becoming more difficult to justify because of growing limitations in making electrical contact to the board. This issue is affecting a dynamic change in ICT system design and spawning a new breed of testers that is a mix of bed-of-nails (BON) fixture-based ICT concepts with moving-probe ICT concepts.

BON vs. Flying-Probe ICT

The most prolific style of ICT today uses BON fixtures to contact each circuit net on the board, typically one nail per net. A relay matrix is used to switch the test electronics to the combination of nails needed to test each component or connection in the circuit.

Because of the speed of the matrix, BON testers can keep up with most high-volume production lines, depending on the size of the board. A unique fixture is required for each board type to be tested since the nails of the fixture must contact the test points in an exact pattern associated with that board type. However, the number of boards that are untestable or unfeasible to test with BON testers is growing rapidly because of test-point access issues and the costs associated with BON fixtures.

Several factors contribute to this trend, including:

• The cost of BON fixturing for low-volume board production typically is not economical.
• The cost of BON fixturing for boards requiring double-sided simultaneous access can be prohibitive.
• The cost of BON fixturing for boards with more than 3,000 nets can be prohibitive. In addition, large net count boards may exceed the tester’s pin-count limits.
• Revision changes to boards can render fixtures useless and cause fixture costs to escalate rapidly.
• The turnaround time to build fixtures can be very long.
• Nail-accuracy issues limit BON fixturing to boards with test-point targets greater than 20 mils in diameter.
• The cost of storing and maintaining BON fixtures can be high as the quantity increases.
• The lack of test comprehension for boards with few to no test-point targets minimizes or eliminates the effectiveness of BON fixturing.

Flying-prober (FP) testers are ICT systems generally classified as manufacturing defects analyzers (MDAs). They provide an alternative to the BON technique for contacting the board. FP testers use motion-system hardware to physically move test probes to the various test targets on the board for any given test, eliminating the need for a BON fixture.

There are two major benefits of FP testers over BON testers: no fixtures are required, and FP testers can hit much smaller contact targets than BON testers. However, the testers have some limitations:

• The speed of test can be from 30 to more than 100 times slower than BON testers, depending on the FP system.
• The number of motion probes for most systems has been limited to four and on one side only. This limits the speed of test to the motion rate of the mechanics of the system. It also restricts test coverage when more than four simultaneous contacts are required for guarding, functional test, or memory test/programming. 
  Several FP systems now allow opposite-side, manually positioned fixed probes to address these deficiencies. Also, some provide double-side access with up to 40 simultaneous motion probes.
• Popular vectorless test techniques for IC and connector open-pin tests require special probe heads and typically access the board from the same side as the stimulus. Some suppliers have addressed this problem with standard probes for all functions and double-side access.
• FPs have the capability to contact component pads for use as test points when no actual test pads exist. FPs currently hit test targets less than 10 mils in diameter. However, this can be accomplished only by probing from the side of the board where the component resides.

Consequently, the physical heights and positions of some components can severely limit probe access for most systems on the market. Some suppliers now offer systems with hardware and software that automatically provide object avoidance without compromising test comprehension.

Cost Considerations

Cost is the ultimate consideration when deciding which type of ICT technology to use or whether to use it at all. The desire is to spend as little as possible on test, while the need dictates otherwise. The reality is that faults do exist, and they must be detected and repaired—and preferably not repeated. Consequently, ICT can be cost-effective for most boards. The question is which type will be most cost-effective.

Table 1 lists some typical cost considerations for comparing FP and BON technologies. The toughest challenge is to get a handle on the existing cost of test to determine where the savings can be realized.

Table 1. FP and BON Cost Considerations

Merging BON With FPs

A trend has begun to combine BON techniques with FPs to enhance the speed and test comprehension of FPs. Many FPs on the market have the capability to manually set fixed-position probes for bottom-side contact while simultaneously using motion probes on the topside.

Up to 60% of all analog component tests will use Vcc or ground as one test connection. Consequently, using bottom-side fixed probes can have a big impact in test speed by freeing up 25% to 50% of the motion resources to effect more tests per motion.

Another benefit can be achieved by adding fixed probes, exceeding 40 in some cases. Unfortunately, the manual process of placing the fixed probes can be tedious and time-consuming, requiring an operator to use a topside camera for placement. Probe-placement accuracy also can be compromised in systems where using tooling holes for board alignment is not feasible.

One proposed solution would implement fixed probes as small BON fixtures on movable lift platforms. The mini-BON can be automatically placed under a board, rotated into place using a camera to find fiducials on the board, and then raised to make contact to the board. Not only can enhanced board tests be achieved with this technique, but also changeover from board to board can be quick. The drawback: This still is a BON fixture approach with all the negative baggage associated with it.

Bed-of-Moving-Nails

After each motion cycle, the system can implement a standard BON test process using a relay matrix to switch between hundreds of probe combinations to perform hundreds of tests in seconds. Each time the bed-of-moving-nails is moved to a new position combination, the BON test process is repeated. The technology to build such a system exists today in both hardware and software.

The effective implementation of a bed-of-moving-nails system requires two basic functions. The first is a hardware design that provides unrestricted probe motion. The second is sophisticated optimization operating in several dimensions simultaneously.

Unrestricted motion requires the probes to move independently from any other probe in the X, Y, and Z axes. In addition, the probes must have variable angle-placement capability to hit the target from any direction (360°) and any angle from vertical (typically 0° to 6°).

Given these capabilities, each probe can contact any point within its target area without moving the whole bed, which significantly increases test speed. Stacking probes in an array creates a bed-of-moving-nails.

To complete the function, the software optimization routine must take into account probe-to-target positioning and object avoidance (parts sticking up on the board). Then the software can calculate the optimum sequence of test execution and probe combinations, including simultaneous double-side probe access considerations.

Consider a bed-of-moving-nails with 60 probes on a side, each with a 2² × 2² target area and arranged in a 10×6 array. This configuration can completely cover a 20² × 12² board and produce the effect of a double-sided, 120-pin fixture. By moving the probes within the 2² × 2² target area, each motion and subsequent burst of tests have the same effect as if using a new 120-pin fixture with a different nail pattern.

This process can be repeated from four to 10 times a second. The only motion required of the bed itself, which can take significantly longer motion time, occurs when two to four probes need to be placed at or near the same target point or to avoid probe collisions with objects on the board. The close proximity of test-point targets causes the whole bed to be moved. For this probe-target scenario, the objective is to move the bed so the test-point targets lie between the probe target areas.

With a bed-of-moving-nails, the greatest motion distance required of the bed is the distance across a single probe target area. One approach to reduce or eliminate the need to reposition the bed uses multiple test points per net. By allowing the use of component pads as test points (which may be the only choice for many boards), multiple contact points generally will be available. Consequently, the optimization software can be programmed to determine the best combination of probes and test points to achieve the optimal BON effect with minimal motion of the bed.

Imagine a tester that doesn’t require any board-specific hardware, can contact very small test targets, doesn’t compromise test comprehension, whose test program can be written and modified quickly, and can keep up with most production lines. Then imagine the cost savings. I predict you won’t have to imagine much longer.

About the Author

Jack Ferguson is founder and president of ITA. For more than three decades, he has been involved with the design, development, and marketing of ATE systems. Mr. Ferguson received a B.S. in physics. ITA, 26242 Dimension Dr., Suite 120, Lake Forest, CA 92630, 949-583-1553, e-mail: [email protected].

Return to EE Home Page

Published by EE-Evaluation Engineering
All contents © 2000 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.

November 2000