Skip to main content
TOMPW Service logo
TOMPW
← Back to Knowledge Hub

ATE Time, Probe Cards and Load Boards: Where Test Cost Really Goes

Break down IC test cost into ATE time, tester class, probe card and load board NRE, and changeover time — and see which lever actually moves cost per good unit.

June 11, 2024Updated September 19, 20268 min readBy Henry
TestATEYieldASICCost

Ask what test costs and you will usually get a tester hourly rate. That number is real but it is the least interesting part of the equation.

What actually moves cost per good unit is how long each device sits on the tester, which class of tester it needs, and how much one-time hardware you have to buy before the first part ships.

Test Time Is the Dominant Variable

ATE is charged by the hour, so cost per device is essentially rate multiplied by test time, plus handling overhead. Cutting test time is usually a bigger lever than negotiating rate.

This is why DFT matters so much. Compressed scan and well-prioritized patterns routinely take a test program from minutes to seconds, and that difference is paid on every single unit.

  • Cost per device is roughly hourly rate times test time
  • Handler and changeover time is billed too, not just tester time
  • Multi-site testing divides the handler cost across more devices

Tester Class Matters More Than Hourly Rate

An advanced SoC tester costs far more per hour than a mature analog or mixed-signal platform. Paying that premium is only rational if your device genuinely needs the capability — pin count, scan depth, high-speed digital or precise analog instrumentation.

Putting a simple device on an over-specified tester is one of the most common silent cost leaks in a test plan. Matching tester class to device complexity should happen during test planning, not after the program is committed.

Probe Cards Are the Hidden NRE

A probe card is custom hardware: it has to contact your exact pad layout at your exact pitch. Fine-pitch and high-frequency work drives cost up quickly, and lead times are measured in weeks.

Cards also wear. Contact resistance drifts and tips get contaminated, so a production program needs a cleaning and replacement plan, not a single card. Budgeting one card and assuming it lasts forever is a common planning error.

  • Finer pad pitch and higher pin count raise card cost and lead time
  • RF and mmWave work needs specialized cards and calibration
  • Plan for wear, cleaning and spares from the start

Load Boards, Contactors and Changeovers

Final test has its own one-time hardware: a load board for the packaged part, plus contactors or sockets that wear faster than most teams expect. On high-volume programs, contactor replacement is a recurring line item.

Changeover time between lots or device variants is also billable. If you run many low-volume variants, changeover can dominate the test cost of each one.

Cutting Cost Without Cutting Coverage

The safe sequence is: improve DFT first, then prune redundant patterns, then increase multi-site parallelism, then renegotiate rate. Cutting patterns before improving DFT is what creates escapes.

The Test Cost & Yield Calculator shows how these levers interact: test time, tester rate, handler cost, volume and yield together, rather than one number in isolation.

Article FAQ

Quick answers tied to this topic

It varies widely by device and volume. Complex parts on advanced testers can spend a meaningful share of total manufacturing cost on test — industry discussions often put it in the range of roughly 15 to 30 percent for such devices — while simple parts in high volume sit far lower. Treat any single number as a starting hypothesis, not a budget.

Usually not immediately. Prototype volumes rarely justify the NRE; many teams start with FT on packaged samples and add a CP program once volume and yield data justify the card. If your pad ring is final and you expect volume, ordering the card early avoids a lead-time surprise later.

Related Tools

Plan this with a free calculator

Related Reading

Where this topic leads next

One deeper read on the same stage of the program, and one on the next decision you will hit after it.

References & Further Reading

Authoritative sources behind this article