The Socket Response to Current Packaging and Test Trends
ADVANCES IN SPRING PROBE TECHNOLOGY DECREASE COST OF TEST SOCKET DESIGN
BY JAMIE ANDES AND ERIC BOGATIN
Denser, faster, cheaper are still the driving forces for semiconductor packages. Not only must the assembly infrastructure keep up, but the test infrastructure also must evolve to meet the changing needs of packages. This means pitches at 0.4 mm and tighter, application bandwidths greater than 5 GHz, quick turn times and reduced total cost of ownership.
Spring Contact Probes
Test sockets using spring probes have evolved over the past 10 years to keep up with changing package requirements. Meeting density and performance requirements requires shrinking probe dimensions: narrower pins to achieve tighter pitches and shorter pins to reach higher bandwidth applications. When spring probes are manufactured at a rate of more than 1 million /week, design for manufacture is just as important as design for performance. To optimize the probe design and evaluate their mechanical performance, an x-ray micro-imaging system has been implemented to observe the in-situ behavior of spring probes while they compress (Figure 1).
Figure 1. X-ray view of an exercised spring probe.
Even though an internal spring is used to create the force driving the pin between the device under test and the load board, no high-frequency current travels through the spring. Performance is related to the pin length, proximity of the return paths and socket material. Each generation of shorter pin and tighter pitch increases operating bandwidth.
The evolution of package requirements has driven the evolution of spring probe design through four generations — each capable of tighter pitch, higher performance and longer life.
The term "semiconductor probe" used to refer to a 10-mm-long, 1.27-mm-pitch, double-ended version of a standard bare board probe. Today, semiconductor probes are less than 2 mm long and can look more like a pill capsule than a spring probe. This evolution is a result of the constant push for smaller packaging. Major chip manufacturers are now showing roadmaps in which package pitches are comparable to chip pad pitches of 5 years ago (Figure 2).
Figure 2. Probe evolution over the past 7 years.
To keep up with this trend, spring probe design has not only shrunk, but the technology has improved dramatically. The concept of an electromechanical mechanism that can withstand a million cycles with a consistent resistance value of less than 40 mΩ is incredible.
Creating such a product involves taking the standard 4-piece design (plunger, barrel, spring, plunger) and building a plunger tip into the end of the barrel. By "floating" this probe in the socket, a smaller, more mechanically and electrically sound design became available (Figure 3). Manufacturing components and assembling parts for 0.5- and 0.4-mm-pitch devices is a feat mastered by less than a handful.
Figure 3. Cross-sectional view of a probe "float."
A growing concern that is beyond the control of socket and probe design engineers is the issue of sizable device tolerances. Even though package sizes and pitches continue to dwindle, tolerances are not declining at the same rate. This is of great concern to socket manufacturers, because pad sizes can drop to 0.2 mm in width, with positional and size tolerances of 0.1 mm. There is a definitive need for chip and package manufacturers to work with test handler manufacturers to improve these tolerances.
Higher Bandwidth Capabilities
The high-frequency performance of a socket is as much due to the design of the socket as the design of individual probes. The first order factor influencing performance is the impedance match of the signal-and-return path through the socket, with the typically 50-Ω impedance of the rest of the system. For the highest bandwidth applications, above 10 GHz, coaxial probes (which look like tiny sections of a coax cable) typically are used (Figure 4). For bandwidths less than 10 GHz, a pin field can give insertion losses less than -1 dB.
Figure 4. Coaxial pins with adjacent return pins in a metal block.
Three features influence the insertion loss of a pin field socket to first order: the ratio of the pin diameter to pitch between pins, the number and distribution of return path pins, and the dielectric constant of the socket material. The first step in optimizing the design of a socket is balancing these terms to achieve 50-Ω impedance through the pin field. The second step is to minimize the total path length of the pins.
There is a common misconception that pins with springs inside cannot be used in high-frequency applications. Though there is a DC connection between the springs and the plungers and barrels, no current actually flows through the springs — even at DC. The role of the spring is to push the plunger against the inside of the barrel's wall. This allows a total series resistance as low as 10 mΩ in some applications. All of the frequency components of the current above 1 GHz flow through the gold plating on the outside of the probes, so the under plate of nickel rarely plays a role.
Leveraging these techniques, cost-effective spring pin contacts can be routinely used in applications in excess of 25 GHz.
Decreased Lead Time
With an increased demand in quick-turn sockets, socket manufacturers must be able to not only provide innovative custom solutions, but also rapidly turn them. It is not unreasonable for customers to send in package drawings on the 1st of the month and expect a completed socket by the 21st. This is no easy task, because custom socket designs are required to push the limits of mechanical, electrical and thermal capabilities.
Design for manufacturability is a key in pulling off such a daunting feat. As with any great idea, if the design is not manufacturable it will not be successful. Common components are used whenever possible, but custom-machined parts are often necessary. Having experienced vendors who can machine 2,000 0.23-mm holes and hold them to the required 0.005-mm tolerances is critical. Socket manufacturers who also have complete control over their contract manufacturing have an edge in providing a repeatable, quality product.
Figure 5. A preliminary drawing of an online socket design.
Web innovations also have entered into the picture as a major contributor to decreasing lead times. Online socket designs are no longer a "thing of the future." Since customers generally need board footprints before a socket purchasing order has even made it through, web-based software has been created to provide socket drawings. Preliminary drawings allow customers to look at socket designs within minutes, rather than days, and get a jump on load board manufacturing (Figure 5).
Decreased Cost of Ownership
The most recent trend when defining the cost of test is to look not only at the original socket pricing, but to also look at the entire system and its lifetime costs. Limiting cost analysis to the up-front costs neglects costs associated with maintenance, downtime and scalability concerns. As is the case with many alternative sockets, the up-front costs may be attractive, but the setup and continual upkeep can be a nightmare.
For a poorly designed socket, maintenance, and eventually downtime, can add a large portion to costs associated with test. In high-volume production testing, contact replacement is necessary, but should not require a high-valued engineer's time or an entire toolbox. A field-friendly, replaceable solution is a valued commodity because of the limited tester downtime needed and the ability to optimize operator usage. A preventative maintenance process should be established with each test environment to decrease downtime caused by damage or major repairs.
Figure 6. SEM photos of cycled probe tips before (left) and after implementing improved material and plating processes (right).
The introduction of lead free to the market is having a significant impact on packaging, and cannot be overlooked on the test end. Lead-free solders, being much harder, wear probe tips at a greater rate. To assure contact to lead-free targets, harder materials, increased spring forces and altered tip geometries are being used. Furthermore, lead-free solders tend to create an increased amount of buildup on probe tips. To combat potential downtime problems, a number of plating options have been introduced to the market (Figure 6).
Follow the roadmap of any major chip manufacturer and a corresponding trend can be seen in spring probe technology. Moving this technology into test socket designs is achieving requirements associated with shrinking pitches, minimal lead times, high bandwidth capabilities and decreasing total ownership costs.
Editor's Note: This article is copyrighted by IEEE, and originally appeared in IEEE's proceedings of the 2004 International Electronics Manufacturing Technology Symposium, San Jose, CA.
JAMIE ANDES, semiconductor product manager, and ERIC BOGATIN, chief technology officer, may be contacted at Synergetix, 310 51st Street, Kansas City, KS 66106; (913) 342-0404; e-mail: firstname.lastname@example.org and email@example.com.