Test Contactor
A test contactor is the mechanical interposer that makes temporary, repeatable electrical contact between a packaged die's leads or balls and the ATE load board beneath it, so every unit on a tester can be probed, scored, and ejected in seconds without a single solder joint.
A contactor is not the same part as a probe card or a load board, and conflating the three is the most common mix-up outside test engineering. The probe card touches bare die on a wafer with sharp needles meant to pierce native oxide; the load board is the fixed printed-circuit interface that routes the tester's signals to the device position; the contactor sits between the load board and a packaged part, swapped out thousands of times a shift as new units cycle through, and it is the one component in that stack built to survive constant mechanical insertion rather than sit bolted in place.
Two contactor architectures dominate, and the choice tracks package type rather than preference. Pogo-pin contactors use an array of spring-loaded vertical probes, each a small plunger riding on an internal coil spring, well suited to fine-pitch BGA and CSP packages down near 0.3mm ball pitch where a clamshell's leaf-spring contacts cannot land reliably; clamshell or lidded sockets instead use a hinged lid that presses a QFN or QFP part's exposed leads down onto fixed contact beams, which tolerates the lead's own compliance better than a pogo pin would and is the standard choice for leaded packages. Johnstech, Multitest, Ismeca, Aries Electronics, and Yamaichi Electronics are the vendors test floors actually buy from, and a single high-pin-count fine-pitch socket can run from several hundred dollars to well over five thousand depending on pitch and pin count.
Contact resistance is the number the whole socket design is optimized around. A spec of under 50-100 milliohms per contact is typical, and engineers have to land in a narrow window: too little normal force and the contact goes intermittent, producing false fails that look like device defects but are actually socket defects; too much force and repeated insertions wear through the plating or deform the device's leads or ball coplanarity. Sockets are rated for a finite insertion-cycle life before resistance drifts out of spec, commonly 100,000 to 500,000-plus cycles depending on contact design, and that wear-out curve is tracked the same way a consumable tool is, not treated as a fixed fixture.
Kelvin, four-wire contact architecture solves a measurement problem the socket itself creates. Routing separate force and sense paths through independent contact elements cancels the contact resistance out of the measurement entirely, which matters for parametric tests reading threshold voltage or on-resistance in the low milliohm range, where the socket's own resistance would otherwise swamp the signal being measured — the same force/sense separation used in bench-level Kelvin probing, just built into the socket's pin map instead of a handheld probe.
Socket qualification runs on the same logic as any other consumable tool: measure the drift, retire before it costs yield. A new contactor design is characterized across its full rated insertion life, tracking contact resistance against cycle count rather than assuming a fixed pass/fail threshold holds forever, because the alternative — discovering wear-out failures mixed in with real device defects on a production floor — turns every failing unit into a debug exercise that has to first rule out the socket before it can even start looking at the silicon.
Read a final-test floor through the contactor, not the tester, and the economics line up differently: the ATE itself is a fixed capital cost, but the socket is the recurring line item that scales with volume, wears on a known curve, and sets the practical ceiling on how fast units can be pushed through test without the fixture itself becoming the yield problem.