Almost 30 years ago, while on a visit to Seattle, Wash., home of the suffering Boeing Corp., I saw a billboard inscribed, "Will the last person leaving Seattle please turn out the lights?" A severe recession, paired with a local economy totally dependent on aerospace, had gutted the Seattle economy. Two local businessmen, with a touch of irony in their humor," suggested in this billboard that it would be nice if the citizenry didn't leave any extra energy bills as they exited the city. Today, many of us live in places like Hsinchu, Taiwan, and Silicon Valley, Calif., where the local economy has been booming for years, but is now totally stagnant due to its dependence on semiconductors and high technology. Instead of waiting until the situation is so bad that we all have to leave, however, someone is already turning out the lights. And, the lights that are staying on are costing a lot more to keep lit. What does this have to do with IC testing? Changing Concerns Until now, the cost of test has been dominated by the cost of acquisition and unit throughput. On the second tier were issues of floor space requirements. Finally, and of least importance in generating a test cost model, were items like the cost of electricity. This cost includes not only the amount consumed, but also the initial costs of routing power and air conditioning during an ATE installation. Today, I would suggest, this is changing in a dramatic way. The cost of power can double in a short time if demand runs up tight against supply. And for some users, power has done worse than double. What does this mean? For high-power ATE, greater than 20-kilowatt systems, for example, energy costs can suddenly become a top-level concern.
The price of high-power ATE hides costs outside of direct consumption. Don't forget that a more power-hungry system emits more heat than a lower-power piece of equipment, which therefore adds to the power draw of the associated air-conditioning systems in the facility. Even if the system is liquid-cooled, it still needs a support system that conducts the heat away. The countermeasure against all this new-age energy expense is the use of lower-power equipment. Simply put, the savings can be dramatic. For instance, a memory tester of the mid-to-late '80s vintage consumes between 20-25 kilowatts. That tester can handle about eight memories in parallel. Today, a 32-site modern memory tester consumes about eight kilowatts. That's right! It's more than 12X more energy efficient. And don't forget, the modern tester is also twice as fast. There are similar gains available in logic and mixed-signal test. In addition, because these newer machines can test many more devices in parallel, they need fewer high-temperature handlers with their resulting high energy consumption and heat emissions. Combine the impact of lower power consumption and higher density on the floor space needed (and the associated power for environmental control of a smaller area), and there is a very compel-ling argument for newer, higher-efficiency equipment. Bottom-Line Costs What this all means to bottom-line costs is that the use of older ATE equipment makes sense only in the cost of acquisition. The real running cost is probably unsatisfactory, at best, and possibly very bad with today's high cost of energy. So, the next time the power on your test floor blacks out, and the testers shut down, leave the power-hungry ones off when the lights come back on. And if the recession continues too long, will the last one off the test floor please turn out the lights!
|
Copyright © 2001