|Figure 1: Production socket with IC device in test condition. |
A typical IC's life starts with the concept/prototype phase, moves to the design validation phase, leaps into the application development phase, rides into production phase and ends with upgrade/replacement phase.
Let's zero in on the production phase of IC devices. In a final production test application, the requirement for accurate measurement of DC parameters — specifically resistance of the load line from IC to board to tester — is often critical. During the final test, IC devices are connected to the load board using interconnect medium. The spring probe is typical of the interconnect media being used over past few decades.
The weakest link in the load line is the semi-permanent connection of the IC device to the load board using spring probes. Spring probes have to be compressed to a certain height in order to make a reliable electrical connection. Since the IC devices are not perfectly planar, each of the spring probes may be compressed to different heights which causes variation in the electrical resistance for each signal path. This variation, in turn, causes load line value to be exceeded past the limit, sometimes causing IC devices to fail during production test. This is a false failure.
|Figure 2: Spring probe internal structure showing resistance network. |
It is customary to re-run the failed devices through the handler to confirm true failure. The metric used to analyze the results includes first pass yield, second pass yield, final pass yield. Typically, the difference between first pass yield and final pass yield ranges from 2 to 10 percent depending on the spring probe characteristics and the cleanliness of the pin tips. This means that recovering a small percentage of failed devices from the first pass represents a significant improvement.
Removing False Failures
A new feature that has been added to the spring probe now removes false failure phenomena in the final production test. This enhanced feature allows production workers to test and verify without multiple repeat testing.
To determine the contact resistance of spring probe, we need to understand the internal mechanics of the spring probe as well as the resistance network that allows the flow of current. Double-ended spring probes are primarily comprised of two plungers (bottom and top), barrel and spring. The spring — gold-plated music wire — is sandwiched between two plungers made of gold-plated hardened beryllium copper inside a barrel made of gold-plated phosphor bronze.
Compressing the Plunger
When assembled, the bottom plunger is compressed to the operating height. This will accommodate standard pad height variations on the target PCB. Similarly, on the top side, the device compresses the top plunger to its operating height to accommodate the device's co-planarity. In this compressed state, the tips of the plungers inside the barrel (both top and bottom) will engage with the barrel wall allowing current to flow through the cylindrical barrel. Because the device has wide co-planarity, not all of the spring probes are compressed to exactly the operating height. If the plungers are not compressed to the same operating height, the engagement to the barrel wall will be different, which in turn results in tens of milliohms of variation in contact resistance. In Figure 2, R1 represents the constriction resistance between the device lead and plunger tip. R2 represents the constriction resistance between top plunger and the inside wall of barrel. R4 represents the constriction resistance between the bottom plunger and the inside wall of the barrel. R5 represents the constriction resistance between the PCB pad and the plunger tip. R3 represents total bulk resistance of the top plunger, bottom plunger and barrel. Total resistance is R1+R2+R3+R4+R5.
|Figure 3: Stamped probe internal structure showing resistance network. |
In this current path, majority of resistance variation is attributed to R2 and R4. When the inside spring compresses a certain distance, a force is applied to the plunger engaging on the barrel inside wall. Variation in the compression distance causes variation in the force ultimately resulting in variation in the resistance. The biggest challenge is to eliminate contact resistance variation due to compression height variation. The design shown in Figure 2 eliminates one of the constriction resistances in the network and controls the other using both an external compression spring and an internal leaf spring. The current flows from the solid top plunger to the solid bottom plunger, eliminating the cylindrical barrel in the new stamped probe design.
The new design uses a pinch mechanism that slides in a controlled groove path between the two solid plungers. This pinch mechanism maintains continuity between moving components at all times and at different compression heights. R2 and R4 in a double-ended spring probe is replaced by only one resistance component R2 in the new stamped probe design. R2 values are further controlled by the pinching mechanism. The disadvantage to this option is that the pinching mechanism wears out after many repetitions of back-and-forth riding in the groove. The pinch tip geometry can be optimized for a certain number of cycles, which will suffice for semiconductor testing and the verification of IC devices.
To verify that the new stamped probe design works better than the existing spring probe design, we conducted an experiment comparing both designs side-by-side in a same experimental setup. The spring probe was referred as SS and the new stamped probe was referred as SBT. The test examined the relationship between contact resistances over the spring probes' life cycle count. An actual handler was used for this experiment.
Setting Up the Test
500 pins were assembled onto a test fixture that was mounted on the test board which was connected to a tester. A gold plated shorted device simulator was mounted on the handler head. The test setup was adjusted such that the head would move down 0.3mm which was the chosen travel for the spring probe. Initial contact resistance data was measured through the tester and the ATE (Automatic Test Equipment) was turned on. This moved the handler back and forth which in turn cycled the spring probes. A digital counter was inserted into the test setup to measure the cycle count. Contact resistance data was collected at different cycle intervals for SS pins. The test was repeated for SBT pins and data was collected and plotted.
It can be seen from the resulting graph that the average contact resistance of SBT pins was less than 15 mΩ consistently throughout 300,000 cycles. Standard deviation was also shown to provide an understanding of the data spread. Standard deviation for SBT pins was less than 3 mΩ throughout the life cycle.
Average contact resistance for SS pins started at 40 mΩ and shot up to 80 mΩ as the cycle count progressed. Standard deviation for SS pins fluctuated between 5 mΩ and 15 mΩ. These fluctuations resulted in false failure during final production test. Based on the graph, it can be concluded that the SBT pin design enables consistency in contact resistance throughout the life cycle, which in turn enables consistency in the final device test applications and thus no differences between the first pass and final pass yields.
|Figure 4: Resistance versus life cycle count for spring pin and stamped pin. |
The main objective of any production test is to be able to rely on test data and not waste time by repetitive tests, and to avoid false failures. The testing conducted here solved numerous production application needs. The test results compared the electrical and the mechanical characteristics of the two interconnection media. Stamped probes provided better contact resistance than the spring probes. Variation of contact resistance for stamped probes was less than 3 millΩ.
The improved geometry of stamped probes eliminated the cylindrical barrel and one of the constriction resistances from the network. The pinch mechanism between two solid plungers of the stamped probe maintained continuity between moving components at all times and at different compression heights. This feature will enable reliable test data and eliminate repetitive test steps. As time-to-market shrinks further, the main stream of the industry will embrace this technology — another step in the evolution of interconnection technology.
Contact: Ironwood Electronics, Inc., 11351 Rupp Dr., Suite 400, Burnsville, MN 55337 800-404-0204 or 952-229-8200 fax: 651-452-8400 E-mail: firstname.lastname@example.org Web: http://www.ironwoodelectronics.com