The traditional data-sheet metrics of handler throughput, index time, and other related specifications are important, but they do not provide enough information to choose a superior test-handling solution. The ultimate assessment of any piece of semiconductor capital equipment is the number of shippable units that a complete test cell can produce in a given period of time.
To guarantee success, equipment from different suppliers must be integrated, tested, and characterized long before first product delivery. This requirement goes well beyond what has been important historically and includes other key factors that must be managed concurrently.
New high-speed IC technologies demand higher test-cell performance—inclusive of the tester, handler, contactor, interface, and docking. More than ever, all the variables and their interdependencies must be managed early in the new product development cycle to ensure success.
There are three key areas to be addressed:
Traditional equipment performance indices such as throughput, mean cycles between assists (MCBA)/mean cycles between failures (MCBF), and index time need to be appropriate for the requirement. Solid performance in these areas, however, is not enough to guarantee success.
Superior mechanical and electrical performance of the docking and interface between the tester and the handler.
Global support giving you the capability to easily transfer equipment and technology between sites and regions.
Equipment and Factory Performance
The way each of us purchases a new car is probably much the same—we focus on performance and features. Is the color right? Do I like the interior? What is the gas mileage? Is the price right? Traditional methods of selecting capital equipment use much the same strategy, focusing on throughput, index time, and price.
While important, benchmark performance in these three metrics does not guarantee success. Certainly, they are essential, and failing to be at or near the industry benchmarks in any of these categories may be reason enough for discounting a particular supplier. But, we also should determine how a piece of equipment will perform over time.
Does uptime tell the story? What about the mean time to repair or assist? Are these sufficient? These metrics address the availability of a piece of equipment to test devices. But having a piece of equipment ready to test devices does not produce good shippable units.
For clarity, these metrics have historically been defined as:
Uptime—the percentage of time in a given period that a piece of equipment is available to test devices.
MCBA—the number of cycles between operator assists, or the jam rate.
MCBF—the number of cycles between failures of a component or mechanism.
Throughput—the peak or average number of units cycled per hour at a given set of test conditions.
Index time—the time that a tester is idle between testing devices.
Overall equipment effectiveness (OEE) addresses the issue of combined efficiency of the complete process. OEE is mathematically the product of the efficiencies of each process within a test cell. It is loosely defined as OEE = the actual number good units tested ¸ the theoretical number of good units that can be tested in a given period of time. The OEE value can never be greater than one.
All factory metrics must be high to achieve high OEE. It is possible to have most metrics near perfection and still have a low OEE if any key factor is low.
The corollary also is true. High OEE indicates a well-run factory.
Electrical and Mechanical Performance
Electrical performance ensures that the exact signal from the pin electronics card appears at the device. The interface—the mechanical docking and electrical contact between a tester and handler—also is critical for high-speed test.
Test frequencies approaching 1 GHz for single and multisite test applications demand that this area receives attention during the development of both the tester and the handler. Mixed-signal devices have been tested at these frequencies for many years, but the additional requirement to test up to 32 units in parallel places even greater focus on this area.
There are many issues that both tester and handler suppliers should address to ensure success:
Create solid docking platforms on both the tester and the handler that, when mated, ensure better than 0.002″ placement accuracy of the test head to the handler.
Plunge devices beyond the docking plane directly onto a high-frequency contactor placed on a loadboard with very few interconnects between the device and the pin electronics card.
Use a high-bandwidth, low-profile, reliable contactor to connect the device to the tester.
Manage the thermal performance of the device during test to ensure that the device temperature stays constant.
Tester and handler suppliers are recognizing the importance of test-cell integration and codeveloping docking and interface solutions early in the product development cycle, well ahead of product deliveries. A test and integration phase validates the design goals and performance objectives. This process guarantees superior electrical signal performance all the way to the DUT, plug-and-play integration, and fast production readiness of the test cell upon delivery.
Often, there are conflicting needs. But managed early in the process, all suppliers can make product changes that benefit each other and ultimately provide the best electromechanical interface possible.
Field Support
After delivery, field support is critical. Suppliers need to work as a team to certify that the test cell performs properly and the customer’s needs are satisfied. The suppliers should have trained local support, locally stocked spare parts, an infrastructure to manage day-to-day test-cell performance and escalations, and a lead supplier who serves as the single point of contact responsible for the performance of the cell.
Conclusion
With decreasing IC life cycles, time to market is more critical now than ever. Historically, the best price and a near benchmark throughput and index time were enough differentiation between suppliers to make a buy decision.
The demands of testing high-speed ICs require that the tester, the handler, and the interface operate together flawlessly, and the full test cell often is required to complete the product development of a new technology. Addressing the total performance of the test cell will make the difference in the ultimate success of any factory—delivering good devices on time to the customer.
About the Author
Kent Blumenshine, who has 18 years experience in IC test and manufacturing, is the product marketing manager at Delta Design. Previously, he was with Motorola’s Microprocessor Products Group. Delta Design, 5785 Kearny Villa Rd., San Diego, CA 92123-1111, (619) 505-1739.
Copyright 1999 Nelson Publishing Inc.
May 1999
|