Test-fixture and probe manufacturers need the help of many technologies to bridge the gap between today's and tomorrow's design and test engineering needs.
Trends in PCB designs the push for quicker time to market and skyrocketing complexity, shrinking targets for bed-of-nails test, the loss of access to test nodes, and other technical challenges'show no signs of slowing. At the same time, reactions to those challenges depend on whom you talk to.
Designers tout the need for superior board performance, shrugging their shoulders at test engineers in a kind of What can we do? You ll have to cope. kind of gesture. Design for test never was a big seller in the best of times, and the accelerating pace of product advancements has intensified designers• resistance. In fact, many companies are modifying design rules and design-for-test guidelines because new designs cannot avoid violating the ones currently in place.
Test engineers, despite decades of attempts to work with designers to make boards more testable, often still have to address over-the-wall designer issues. That is, they receive the designs as faits accomplis and must devise the best possible test. Manufacturers of fixtures and test pins have to provide the tools to bridge that gap.
So what are the issues? And how do they differ among the groups of participants? Although many of the ideas here may seem familiar, they deserve a dusting off to address current and future needs.
Shrinking Test Targets
Perhaps the most obvious trend in PCB manufacturing in recent years has been the ever-smaller targets that bed-of-nails fixtures must hit to execute a test. Design engineers must improve and expand board features and increase speeds while reducing both size and cost. Comprehensively testing these new boards means that designs must include more test points. Shrinking real estate demands that those test points shrink as well.
Meanwhile, test engineers are chartered with ensuring that boards work when they leave the manufacturing process and that they continue to work throughout the product's useful life. This group demands the largest possible test targets to ensure that nails hit their targets accurately and provide reliable connections through which to execute tests.
The groups appear to be working at cross-purposes. Smaller test targets generally mean reducing fixture probe size. Smaller probes are less durable and require more preventive maintenance in production than their larger siblings. Fixture suppliers have had no choice but to develop new probe technologies to provide more robust probes despite tighter pitches.
Conventional probes are encased in tubes called receptacles that hold probes in place and create a tight vacuum seal. This architecture increases the effective diameter of the assembly, limiting how closely you can space probes together.
In the new so-called receptacle-less probes, the feature that holds and seals the probes is at the bottom of the probe, so the effective diameter becomes that of only the probe itself. Where an old-style probe might permit spacing of no less than 0.050″, comparable receptacle-less designs can achieve spacing down to 0.039″. Figure 1 shows both older and newer varieties.
Figure 1. Various Kinds of Receptacle and Receptacle-Less Fixture Probes
One note of caution: A single fixture should not include both probe styles. Their sealing techniques are so different that using both styles in the same fixture would allow air leaks. The air leaks, in turn, would produce significant levels of static electricity and might reduce probe pressure on densely probed components like ball-grid arrays (BGAs), potentially degrading test signals.
Meeting the challenge of tighter centers has encouraged new fixture manufacturing techniques as well. A spring probe's design limits its pointing accuracy. Too tight a fit between probe plunger and barrel would not provide sufficient clearance for the necessary sliding contact, and probes could stick. To address this issue, fixture manufacturers have adopted new drilling methods to create features in the top plate that funnel probe tips to their targets.
To avoid pin-to-pin shorts, fixture makers also have developed considerable expertise in drilling very small straight holes, controlling more drill parameters than in the past. Diamond-shaped tooling pins hold boards in place more securely than round ones to increase pointing accuracy and precision during test.
Board Damage During Test
Test must find PCB failures, not create them. Many current PCBs include small, surface-mount, low-power components that are more susceptible to damage than their predecessors. Available CAD tools allow designers to place hypothetical test points and run stress-deflection analyses on the resulting model.
Most models neglect to include fixture-related issues such as supports, push fingers, and backup blocks, considering only probe density. These other features remain the responsibility of the fixture supplier and are only as good as the tools that the supplier has developed.
Historically (and unfortunately), design groups have little interaction with test-fixture suppliers. As a result, too often the fixture damages boards under test.
So what is the alternative? Board designers must follow existing and accepted testability guidelines. Certainly the days of simply taking tester outputs and randomly sprinkling fingers and supports have passed. Designers and fixture suppliers must communicate with one another.
In many cases, today's complex boards demand treating the test fixture itself as a designed part of the product specifically to prevent damage during test. Both the designer and the supplier must consider placement of probes, push fingers, and supports as constraints on fixture design. For example, fixture push fingers must have sufficient clearance around component perimeters to prevent damage to sensitive components such as BGAs.
Managers expect today's test engineers to provide adequate tools to maintain a high-quality test product regardless of component packaging, point count, probe density, and probe forces during test. Fixture suppliers find themselves caught in the middle. Fixtures must support the board under test without damaging it, a real challenge as available real estate for push fingers declines dramatically.
Figure 2 shows some of the features competing for board and fixture real estate. Placing enough fingers and supports in the available space while assuring no contact with components becomes unnecessarily difficult. Too often initial product runs include broken, damaged, or degraded boards regardless of whether tests rely on vacuum, pneumatic, or mechanical fixtures.
Figure 2. Some of the Features Competing for Board and Fixture Real Estate
Test engineers must serve as a communications bridge between design and planning groups on one hand and fixture suppliers on the other. In addition to test data, test engineers should provide CAD information from designers and a bill of materials (BOM) from planning to assure the best possible product.
Maintaining this communications link can require more legwork at the outset but achieves the goal of the highest quality test fixture at the other end. Low-end fixture suppliers generally lack the knowledge of CAD, BOMs, and automated tools to modify fixture layouts to prevent board damage, relying instead on empirical methods. Strain-gauge measurements, for example, can offer information about board stresses but can provide only a snapshot of board stresses in high-risk areas.
Leading-edge fixture suppliers have developed software tools for automatically placing fingers, supports, and backup blocks properly, assuring that they provide enough support to prevent board damage. The software also runs a board stress analysis (BSA) to assure fixture-design integrity before fixture construction begins.
Figure 3 shows a board-flex map showing both the areas of maximum deflection and the components affected. Based on their experience and empirical data, fixture makers have proposed deflection limits. Without automated tools however, fixture manufacturers can verify fixture integrity only using conventional strain-gauge equipment after the fixture is built. In that case, the fixture maker often must resort to rework to lower measured strain to acceptable levels.
Figure 3. A Board-Flex Map Showing Both Areas of Maximum Deflection and the Components Affected
Lead-Free Solder
As of July 1, 2006, most products sold into China or the EU will require lead-free components and lead-free solder. Other countries will likely follow suit, and eventually the lead ban will likely spread to most other products as well.
Eliminating lead, particularly in solder, presents significant challenges. New solder compositions melt at about 217 C, more than 30• higher than the melting point of tin/lead solder and approaching tolerance limits for many components.
The new solder is more brittle, making good contact less certain and subjecting joints to possible breakage from mechanical forces such as board flex during test. It also doesn t wet as well or flow as easily as lead solder does, so a nail will more likely hit bare copper rather than solder. All companies incorporating bed-of-nails tests must overcome these obstacles.
In addition to addressing soldering issues, board designers must identify lead-free components. As a result, many companies will find meeting the July 1 deadline extremely difficult. It may be necessary to phase-in compliant products over a number of years as lead-free components and manufacturing techniques become more common. Complications caused by the need to track separate part numbers for leaded and lead-free components may encourage adoption of lead-free processes even for otherwise exempt products.
Test engineers will face increased failure rates. Likely failures other than solder-joint breakage include cracked components from the higher temperatures, tombstones, voids, solder bridges, and inadequate solder fill for through-hole components. Accurate probing will become more critical to achieve a reliable contact between test probes and targets. In addition, pooling of flux near test pads may create false test results.
Fixture suppliers must always provide products that deliver good contact and work closely with the company's test engineers to select appropriate test probes for the purpose. A hard, sharp probe, such as an 8-oz steel plunger, proves an effective choice for hitting lead-free solders.
Fixture houses have to design fixtures that subject lead-free boards to as little flexing as possible. Studies have shown that because lead-free solder is stronger and stiffer than its predecessors, when a lead-free PCB flexes, delaminating base copper from the substrate proves much more likely than with tin/lead solder. The need for higher spring-force probes exacerbates the problem.
Shrinking Test Targets and Boundary Scan Testing
Boundary scan techniques can minimize or eliminate test-net-access and tester resource allocation problems despite ever shrinking board geometry. Designers contend, however, that the limited availability of boundary scan components makes achieving comprehensive or at least largely boundary scan-based board testing much more difficult. Nevertheless, even boards that include only a few boundary scan components can eliminate at least some physical test points, alleviating some resource limitations.
Optimally, test engineers can wire together test access ports on multiple packages and test an entire board using a long data chain. Tests can include checking connections between devices as well as locating shorts between nets even without physical access to those nets.
Test programs can explicitly eliminate physical access to some nets, opening up board real estate for additional circuitry or easier routing. The technique often works well, but problems do arise when you apply it at in-circuit test.
When incorporating boundary scan into an in-circuit test, the long wires that connect fixture test points can seriously degrade signals that the board under test receives. Crosstalk between wires can corrupt boundary cell states, yielding false test results. The result incorrect fault diagnosis makes verifying test quality difficult.
Some test departments allow retesting failed tests, potentially allowing spurious noise to subside and thereby providing a test pass. Others consider such a result to be a hard failure, which boosts board-failure rates, so they have to find a way to reduce the noise.
Fixture suppliers have introduced technologies that only engage boundary scan points when desired. Bilevel or dual-stage fixtures fall into this category, although historically they were used to separate functional test probes from the usually much larger number of pins for in-circuit test.
Such a fixture contains test probes mounted on two levels. During functional testing, only probes necessary for that test contact the board. Then, when necessary, a second vacuum or mechanical stage brings the board into contact with the rest of the pins.
The technology has found new life for boundary scan operations. Dual-stage fixtures can completely eliminate crosstalk between wires by connecting only the test access port connections for boundary scan tests in the functional position.
Some fixture suppliers have attacked the noise problem by providing wireless fixtures with embedded ground planes. This approach requires careful attention to signal lines in the fixture/board layout process. Such solutions have been noticeably less successful in the marketplace for resolving crosstalk problems.
Cost: The Inevitable Leveler
Every company experiences intense pressure to lower costs. Although designers determine many costs even before a single board comes off the production line, test groups and the supply chain itself can exert considerable influence as well.
Designers must meet cost targets for each of a product's subsystems and each process step. Yet the lowest initial cost does not necessarily mean the lowest overall cost. Cost-effectiveness has no value if the resulting product proves too difficult to manufacture or test.
As an example, designers often strive to lower costs by minimizing the number of layers in bare-board substrates. Although board layout tools facilitate this approach, fewer board layers contribute to scarcity of board real estate and therefore reduced pin-target size.
To achieve product performance specifications, designers often must violate their own design-for-test rules. Recognizing these common board-design limitations, designers must involve test engineers as early as possible to allow them to compensate, or at least prepare, for the challenges that scarce real estate and small probe targets will bring.
Company managers fully recognize the cost implications of a comprehensive test strategy, to the point where some companies require sign-off by the chief financial officer before buying a fixture. This constraint makes test departments• jobs even more difficult because test engineers cannot react quickly to changes in board design, layout, or performance specifications that affect fixture requirements. Again, the least-expensive fixture does not guarantee the lowest overall test costs.
In one recent case, a test manager was directed to accept the lowest bid for a test fixture. Unfortunately, this bargain fixture:
• Missed targets.
• Never reliably made adequate contact with the smallest targets.
• Broke components.
• Took extra time to bring online in production.
• Never achieved the expected test coverage.
As a result, the fixture's overall cost widely exceeded that of more expensive fixtures. To complicate matters, the company planned to outsource production of that product to a contract manufacturer in Mexico where no local support would be available to address fixture problems.
In addition to supplying cost-efficient products, fixture makers must deliver them within a few days of receipt of the order and support materials, if possible. They must obtain advance notice of any unusual technology requirements or challenges beyond the norm.
Software tools have to eliminate board flex and interference during test. Drilling methods must allow probes to hit even small targets accurately and reliably. And, for most situations, it is important that the supplier offers global support.
The Future Is Like the Present
Trends toward denser, more complex boards with very limited real estate for test probes and other niceties will not go away. Designers, test engineers, and fixture manufacturers must face time-to-market pressures, increasing miniaturization, and similar challenges and still produce high-yield, reliable products.
For bed-of-nails test, all parties concerned should consider the following:
• To build the best, most reliable fixtures, test engineers should obtain CAD information from designers and a bill of materials from planning operations.
• A smart fixture design process will ensure that the fixture is right the first time, minimizing time and money costs and maximizing product quality over having to modify it later.
• A fixture to address today's test challenges also must accommodate brittle, high-melting-point, lead-free solder.
• To minimize pin-count and reduce costs, a fixture should take advantage of boundary scan logic whenever possible.
Ultimately, the fixture package must minimize overall costs. Higher manufacturing yields and higher throughput out the door will achieve that goal better than blindly considering only up-front costs.
About the Author
Gary St. Onge is the director of global kit operations and engineering at Everett Charles Technologies. He has 28 years of experience in the ATE industry and earned a B.S.M.E. and an M.S.M.E. from Union College in Schenectady. Mr. St. Onge also is a licensed Professional Engineer and holds eight U.S. patents that relate to test fixtures. Everett Charles Technologies, Test Services and Fixture Group, One Fairchild Square, Clifton Park, NY 12065, 518-877-3750, e-mail: [email protected]
FOR MORE INFORMATION
on test fixturing at Everett
Charles Technologies
www.rsleads.com/601ee-190