Dr. James A. Forster on Sockets
From the user's standpoint, what are the most critical socket requirements? From a technical perspective, the key requirements are reliability and simplicity of operation. This means that the socket must function every time and never report a "false negative." This occurs when a good device fails burn-in due to a continuity problem between the device and the socket and the device is rejected when it is a socket problem, not a defective device. Ease of operation and a small footprint to maximize burn-in board density are also important-as are lower costs. Although socket purchasers continue to experience reductions in CSP socket prices as CSP volumes increase and the learning curve declines, these prices are still higher than sockets for conventional packages. From the manufacturer's standpoint, what socket attributes are the most difficult to design-in? The greatest challenge for any socket is to provide a reliable electrical connection. The contact must offer a variety of attributes:
A generic socket solution for all CSP packages is difficult since each CSP manufacturer has their own specific package design with different overall dimensions-even though the I/O array is identical to other devices which offer the same function. This requires that the socket design be capable of being easily modified without significant impact on delivery time or socket cost. What particular problems do RDRAM sockets present, if any, in terms of design and/or use? There are no specific problems associated with burn-in sockets for Rambus-based memory chips. The issues are essentially the same as those for burn-in sockets for all CSP packages. The unique 1.0 - 0.8 mm array pitch of the Rambus device allows some additional space for contact movement and increased wall thickness of the various molded components, when compared to the 0.75 mm and below pitch devices. The differences in the overall dimensions of the packages mean that a socket is customer specific and there is a cost or investment in tooling which must be made for each customer's device. Explain the difference is manufacturing technology and philosophy between sockets for test and sockets for burn-in use. The major differences between sockets for test and those for burn-in are:
What can users of sockets look for in the future? Burn-in sockets for CSP packages are in their infancy, since they have only been commercialized in the last three or four years. Users can expect continued improvements in the reliability of the contact, the manufacturing tolerances and the tools and techniques for manufacture and assembly of the components. These improvements will result in lower lead times and cost. As additional applications are converted to CSPs, and the total and specific application volumes increase, there will be a reduction in the price of burn-in sockets. TI estimates that prices for CSP burn-in sockets for high-volume, memory-type applications will approach $10 next year. How is the growth of portable communications ICs affecting sockets technology? The need for smaller package sizes required by applications in portable devices is driving the pitch for CSP devices down to 0.5 mm. The contacting options for this pitch are limited, and a number of companies have provided burn-in sockets. This is an area where there does not appear to be a clear burn-in socket leader, and some development activity remains. The costs of burn-in sockets at the 0.5 mm pitch are considered high, so much work remains in this area. As the number of CSP devices produced at a pitch of 0.5mm increases do you see further reductions below 0.5 mm? The move to any pitch less than 0.5 mm will present significant challenges-not only to the burn-in socket suppliers but also to the manufacturers of burn-in boards, the test and handling infrastructure and PWB board makers. HDI has demonstrated the commercial ability of producing 2 mil lines and spaces, but the problem will be the fan out of the signals at reasonable cost. These commercial difficulties suggest that it is unlikely that there will be widespread adoption of pitches below 0.5 mm anytime soon. Do you see consolidation among socket producers or more socket makers in the future? The current burn-in socket suppliers fall into two major categories. The first segment includes the larger global companies, such as Wells-CTI, Enplas, Texas Instruments and Yamaichi, each with socket sales in excess of $40 million per year. These companies provide a variety of solutions for volume applications to customers with locations around the world and have the resources to support the investments required for high-volume requirements. The second category, consisting of more than a dozen companies, includes those that offer excellent solutions and might be considered lower volume or custom application suppliers. Currently, it does not appear that there is any strong incentive for consolidation. The value statement of each company appears to be sufficiently different that combinations of any of the companies would not add "value" to the customer. Since the sale of Wells to CTI there has not been any significant activity in this field, and it appears that no company is on the acquisition trail. |
