A Critical Role in Back-end ATE Testing
BY ILA PAL
Failure to use appropriate test sockets can negate the accuracy of test results and significantly diminish the performance of million-dollar automatic test equipment (ATE). AC parametric and compliance testing of high-speed, high-pin-count integrated circuits (ICs) requires great care in the design, layout, and manufacturing of the load board and test socket. Because consistency is critical in ATE systems, socket selection plays a major role in the success of the production test phase. One choice is a commercially available socket, which the load board is modified to fit. Another option is a custom-designed socket modified to fit the load board. The cost and performance tradeoffs between these test sockets must be evaluated.
A typical test system includes ATE hardware and software to run different tests, load board, test socket, and handler. The test socket is mounted to the load board, which in turn interfaces with the ATE. A handler includes compartments for trays where devices under test (DUT) are stored. A vacuum head/plunger inside the handler will pick up the DUT and push it inside the test socket while the ATE performs necessary tests. Socket applications include production test, device characterization in the lab, in-system test, failure-analysis test, burn-in, and reliability test. This article concentrates on high-volume test before final production release where silicon is packaged into final form to be assembled onto circuit boards.
The test socket is the final link between ATE and DUT. A typical test socket consists of contacts for each pin on the DUT, housing to hold the contacts, and a precise alignment mechanism for the DUT. Ideally, the test socket should perform reliably for a minimum of 500,000 touchdowns. A variety of contacts are used in test sockets, including spring pins, particle interconnects, and shaped contacts loaded in elastomer. The contact geometry is important to achieve high bandwidth, low contact resistance, long cycle life, and high reliability. Socket contact requirements are driven by both electrical performance and mechanical reliability.
Typical electrical specification of a spring pin is shown in the following table (Figure 1).
Figure 1. Electrical specification of a spring pin.
Self-inductance and mutual inductance play major roles in high-speed test applications. Capacitance is driven by pitch. Fine pitch causes spring pins to be closer, which in turn increases capacitance and influences test performance. Minimizing parasitic inductance and capacitance reduces cross-talk effects. Contact goodness is defined by bandwidth, defined as the -1-dB insertion loss bandwidth. Insertion loss will be affected by characteristic impedance, discontinuities of spring pin geometry, and contact length. Housing materials, spring pin conductor loss, and DC contact resistance affect dielectric loss.
Mechanical reliability is defined by the total number of actuations withstood before failure. Contact performance at high temperature and humidity is another important factor defining the life of the test socket. Selecting the right contact material and designing appropriate geometry achieves both mechanical and electrical requirements for high-performance test applications.
Figure 2. Typical spring probe.
Double-ended spring probes comprise two plungers, barrel, and spring. The spring (gold-plated music wire) is sandwiched between two plungers (gold-plated hardened beryllium copper) inside the barrel (gold-plated phosphor bronze). The spring forces the plunger against the inner surface of the barrel. Therefore, the current flows on the outer surface of the pin and avoids the spring. This construction maintains a low inductance (<1 nH). The spring probe has contact resistance less than 0.04 W and high current rating of 4.0 A continuous. The operating temperature range is -55° to 150°C. An example spring probe is shown in Figure 2. An alternate to gold plating is palladium cobalt plating (PdCo) which is harder, has a lower coefficient of friction, and has comparable resistance to gold plating. PdCo plating is stable at high temperatures (150°C) and long duration (more than 30 days).
The socket design consists of two major components. One portion involves alignment and the even distribution of force on top of the IC pushing the solder balls into the spring pin. The other involves the proper housing design for holding the spring pin and allowing for appropriate compression to make perfect contact to the load board. Figure 3 shows a spring pin assembled inside the socket housing. The free state shows a spring pin assembly before mounting onto the load board. The assembled state shows a spring pin assembly compressed halfway when mounted on to the load board. The loaded state shows the fully compressed position of the spring pin when the IC is pushed down. Designing the top and bottom guide with appropriate hole sizes and depths defines successful and repeatable spring-pin performance.
Figure 3. Spring pin assembled inside socket housing.
Figure 4 shows a spring-pin socket mounted on a test board with an opened lid. The lid has an integrated compression plate. When the compression screw is turned, the compression plate moves down and applies an even pressure to the DUT. In the ATE process, the socket will not use the clamshell lid. The auto handler applies the downward pressure. There are different types of handlers available in the market. The socket surface should be customized to interface with these various handlers. For example, the vacuum head or plunger inside the handler picks up the DUT and places it into the socket. The socket surface should provide necessary clearance for the vacuum head/plunger and prevent over-compression of DUT.
Figure 4. Spring-pin socket mounted on test board.
A good quality spring pin should return to its original position and retain constant spring force, even after 500,000 device insertions and extractions. Mechanically, the spring pin should maintain constant force on each ball, regardless of the ball count. Since there are variations in ball heights, each spring pin will experience different deflection that, in turn, causes a different force. Variations in deflection can be controlled by socket housing design. The rigidity of load board is another important factor. Since the DUT pin counts easily surpass 1,000, overall compression force will be high, which the load board has to withstand without warping. Load-board deflection will cause uneven forces on each spring pin. A proper back plate support on the backside of load board is mandatory.
Spring Pin Endurance Test
Thirty-two sample spring pins were endurance-tested for a cycle life totaling 500 K employing a 4-wire Kelvin test setup. Under a cycle life testing program totaling 500,000 insertion and extraction cycles, the 32 sample probes have an average resistance of 27.38 mW throughout entire 500,000 life cycles. Figure 5 shows the contact resistance over 500-K life cycles.
Figure 5. Contact resistance over 500-K life cycles.
Burn-in testing determines contact resistance variability at high temperature. A socket was designed using 1-mm pitch, 676 spring pins, and mounted on a daisy- chained load board. A daisy-chained DUT chip was placed inside the socket. Adjacent pins were connected in the DUT chip. Alternate adjacent pins - pins that are not connected in the DUT chip - were connected in the test board. By testing the two end pins, the complete array is verified and contact resistance per pin is calculated. The two end pins were routed to four pads for Kelvin measurements using a 4-wire setting. The socket was placed inside an oven at 125°C. Total contact resistance was measured at regular time intervals. The results are shown in Figure 6. It can be seen from the graph that the contact resistance for the spring pin ranges between 25 to 58 mW over 200 hours.
Figure 6. Burn-in test results.
The test socket is the heart of an ATE system that provides the signal flow from the DUT to the tester. Spring probes serve as arteries that may withstand high temperatures and provide long cycle life when properly maintained. Replaceable probes allow users to change independently without changing the entire socket. Test systems should be designed based on the type of IC to be tested (memory, logic, digital, analog, RF, etc.), the type of test (DC parameters, AC parameters, RF, functional, in-circuit, etc.), the number of devices to be tested, and speed of the test. By careful analysis, the proper spring probe can be selected based on mechanical and electrical requirements. A properly designed socket using appropriate spring probes will connect the dots and complete the chain of any ATE system.
ILA PAL, director of R&D, may be contacted at Ironwood Electronics Inc., 990 Lone Oak Rd., Suite 120, Eagan, MN 55121; 800/404-0204; E-mail: firstname.lastname@example.org.