For a decade now, we have been concerned with the cost of test and the cost of ownership. During this time, manufacturers of automatic test equipment (ATE) have successfully addressed these issues with impressive cost reductions.
These savings have been made possible both by technology advances in test techniques, such as boundary scan and vectorless test, and new ATE designed for specific test strategies such as the manufacturing process tester. As those testers have reached their limits in cost reduction for a specific test strategy, the next step has been to reduce the overall cost for a particular environment using multiple test strategies. This is the paradigm of the manufacturing test platform, providing state-of-the-art, uncompromised in-circuit and functional test with full compatibility.
Because ATE and inspection and test methodologies are interrelated, we must examine the choice of ATE and the overall test strategy as a complete entity, not as independent elements. But in today’s high rate-of-change production environment where time and money are everything, it has become an almost intractable problem to recognize and consider every variable in planning a rational test strategy.
In a low-mix, high-volume environment, a direct way to lower the cost of test is to reduce the test time per board as much as possible. In this situation, the test engineer focuses on reducing cost by fine-tuning the test strategy to improve throughput.
A high mix of board types introduces more challenges. The test strategy must adapt to different situations such as analog, digital, and mixed-signal tests; different layouts; and lack of access. Even if all the elements of the initial cost equation are the same as for a low-mix environment, the focus of cost reduction changes. Very quickly during the cost analysis of a high-mix environment, you see that the dominant expense is the bed-of-nails test fixtures.
In the simplest terms, we want to examine boards for manufacturing defects. And, we want the lowest-cost ATE that will provide the capability and productivity for our particular high-mix environment.
ATE Ownership
First, let’s look at the ATE. Figure 1 (see August 2000 issue of EE) shows the total cost figures for producing 50 board types per year for two types of test systems: an in-circuit tester (ICT) and a flying prober. The costs include initial acquisition, maintenance (4% of the acquisition cost per year), programming and debug (the same for the equivalent fault coverage on either system), and fixture fabrication and debug (which do not exist for a flying prober).
In-Circuit Tester
The most popular system for electrical process test is the ICT or manufacturing defects analyzer (MDA). In a high-mix environment, the manufacturing processes are more difficult to optimize for a small quantity of boards. As a result, an ICT is preferable to an MDA because in-circuit test provides better fault coverage by combining power-off and power-on testing.
An accompanying bed-of-nails fixture gives the tester access to every component pin on the unit under test. This tester-fixture combination offers excellent fault coverage and the fastest throughput.
There has been major progress in lowering the cost of ICTs. Basic ICTs cost $50,000 to $150,000 less than they did a decade ago.
Even with a cost reduction in the tester itself, many other factors are driving up the cost. For example, larger boards and high pin-count devices require more channels; more channels need more channel cards. Voilà, a low-cost system becomes a high-cost system.
Some cost reduction has been realized using boundary scan and vector-less test techniques, but it’s one step forward, two steps back. Even if you have just one board type with a high node count in your mix, either it will handicap the entire test cost for all board types or the board will not be tested in-circuit.
By far, fixturing generates the most expensive, recurring costs associated with ICTs, and you need one for each board type. Since the beginning of in-circuit test, the bed-of-nails fixture has been the albatross around the neck of the test engineer.
A fixture is mechanical, needs maintenance, and is not easily adaptable to change. Good progress has been made in fixture cost reduction over the years, but as boards become more dense and components become more complex, the fixture becomes increasingly difficult to build, and the cost goes up again.
In high-mix environments, fixturing is getting more clumsy and costly as board design and product variance proliferate and cycle times shrink. Even in a medium to low mix, the fixturing costs can exceed the tester acquisition costs in the first year.
The average capital cost of a basic ICT is about $125,000. A fixture kit starts around $3,000. Depending on size and complexity, a finished fixture could cost more than $15,000; however, for our comparison, we’ll use a conservative cost of $5,000 per fixture.
Flying Prober
A flying prober is an ICT that does not need a bed-of-nails fixture. Instead, four very sharp probes move over the board to perform electrical tests. Because the flying prober is a mechanical assembly and uses moving probes rather than switching relays, it is five to 10 times slower than a conventional ICT. In addition, the fault coverage is not as comprehensive as that of the ICT; digital testing must be done at the functional test stage.
On the plus side, a flying prober succeeds where a bed-of-nails fixture has reached its limits. Probe access to the board, no matter how dense, is close to 100%, and the precision of the moving probes is much better than any bed-of-nail probes.
Despite the significantly slower throughput, more flying probers are being integrated into the manufacturing line in high-mix, low-volume environments due to their fixtureless test capability, short ramp-up time, reliability, and repeatability.
Flying probers always have been used for prototype testing where fast cycle time is more important than throughput. Recent improvements such as the surface linear motor have made top-end probers precise and repetitive, eliminating false failures and mis-probes. Now that they are manufactured and supported by several companies, flying probers can be used in production as well as for prototype testing.
The average capital cost of a basic flying probe tester is about $225,000, nearly double that of the conventional ICT.
Factors Influencing Test Strategy
In Figure 1, the variable with the greatest impact on the cost of test is the test fixture or the lack of it. This does not reflect other factors that may be equally important to you. While cost of ownership is important, it should not be the sole basis for selecting ATE. Consider the inspection and test methodology requirements of your environment.
Time
What are the time requirements? The line beat rate rarely is the issue. In short-run situations, board handling at the beginning and the end of the production line is more apt to be the bottleneck.
Instead, we need fast programming time and quick program and system setup for different boards throughout the day. It would take too long to build and manipulate a fixture, and to stock all of them is an extra cost.
Is ramp-up time critical? The flying prober requires program debug of one to four days. A test fixture adds at least 13 days to the process: one day for data preparation, at least 10 days for fabrication, one to two days for debug, and one to four days for program debug.
Board Complexity
If the board types are simple, the best test strategy could be to run an exhaustive in-circuit test eventually complemented with functional test on the same tester. In that case, the cost of a test fixture is justified by the elimination of the functional test step.
If the board types are complex and must be tested with functional test in the manufacturing process, the fault coverage provided by a flying prober ensures that the yield at the functional test will be good. In this situation, the ICT and test fixtures can be efficiently replaced by the flying prober.
Node Count
Do you need to test a high node-count board? A fully loaded system with 5,000 test points will run $300,000 to $400,000 for an ICT and $225,000 for a flying prober. So, if volumes are low with high node-count boards and long test times are acceptable, the flying prober will be a smaller investment than an ICT.
Conclusion
No matter what else happens in the industry, it’s certain that the cost of test fixtures will keep pace with the evolution of technology. But don’t throw out the ICT yet. Where the expense of a single fixture can be spread among hundreds of thousands of boards, it still is your lowest-cost choice. Depending on volume, the break point for cost-effective in-circuit test is no more than 50 fixtures per year, but I have seen a manufacturing environment with 200.
In the end, everything we do is an economic choice, including test-strategy definition. One type of ATE cannot satisfy every manufacturer’s test requirements, but flying probers are coming into their own on the production line.
Flying Prober Tips . . .
If you’ve decided that a flying prober will be the lowest-cost ATE for you, carefully evaluate the systems on the market. There are at least five major suppliers from which to choose.
A well-designed flying prober will deliver the following:
- Rapid test development with minimum debug time.
- Precise probing accuracy and repeatability to consistently hit fine-pitch targets and test points that are difficult to contact or inaccessible with bed-of-nails fixtures.
- High surface-mount technology assembly process fault coverage using proven in-circuit test techniques.
- High mean time between failures.
- Global engineering and maintenance support.
In addition, you can expect the top-end systems to offer
vision inspection and power-on testing.
About the Author
Elie Bouskela is the flying prober product manager at Teradyne. Before joining the company in 1987, he was a design engineer of telecommunications VLSI components at Intel. Mr. Bouskela holds a degree in electronics from the University of Ville d’Avray and an advanced degree in electronics engineering from SUPELEC in France. Teradyne, Assembly Test Division, 2625 Shadelands Dr., Walnut Creek, CA 94598, 925-932-6900.
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.
August 2000