April 2003 The International Reference for Chip-Scale Electronics, Flip-Chip Technology, Optoelectronic Interconnection and Wafer-Level Packaging Email the editor

Paul M. Sakamoto Contributing Editor-Test

By now you've realized that those of us who were bold enough (in this case, "bold" means ignorant) to have actually predicted some sort of business recovery by now have been proven radically wrong.

I was certainly one of the bold ones almost exactly one year ago. The next move that most of us make at this kind of juncture is to go forth and talk about how to turn crushing adversity into "opportunity."

This month, let's look at ways to take advantage of these prolonged "opportunities." The ones we will look at are the ones that can save immediate cash. After all, those are the things your boss might let you do. Furthermore, we will make sure that we don't need an initial cash investment, since I don't want you to be fired because of me!

In a previous column ("Turn Out the Lights," Aug.-Sept. 2001), we investigated the cost savings of discarding legacy equipment, even if it was $0 on the books, and replacing it with newer equipment to save electrical power.

Do you still enjoy betting millions of dollars upfront, or would you rather bet a few pennies per unit?

I've taken the investigation a step farther, and created the table. You will see that in Silicon Valley, electricity costs $0.15 to over $0.20 per kwh. A rough calculation (including air conditioning costs) quickly runs to hundreds of thousands of dollars in energy expense over a three-year period.

When you view the table, you'll see that the electrical consumption of the fully depreciated "free" tester on the left costs more than the likely new purchase price of the new low power tester on the right. It is assumed that the A/C (air conditioning) power consumption and operating costs are equal to the tester power consumption-a conservative estimate.

Equipment Depreciation

In the U.S., the average piece of equipment is depreciated in three years. Accordingly, if you are using a legacy 256-pin tester that consumes 34kw (this is the specification of a formerly great microprocessor tester), the electric bill amounts to $100,000/year.

If there were a low-cost replacement that cost under $300,000 and ran on 1kw, you could save money immediately by donating your old machine to a homeless shelter or university and replacing it with the new model on an operating lease.

I will leave it to your finance department to explain how this last part works.

Do such low-power machines exist? In many cases, yes! You should check to see, first, of course, whether they can handle your applications.

Another big opportunity is in the movement to designs that are fully testable in production using only DFT (design for test), BIST (built-in selftest) and other structural or in-socket testing.

Now, I'm aware that there are still some negative technical opinions on this subject, but I will just put those in the big bin of "opinions that will prove wrong."

If Moore, Noyce and Shockley displayed such negativity, we would still be using vacuum tubes.

Technical Challenges

It's just another set of incremental technical challenges to have a lot of fully structurally testable devices. The real argument, as always, is economics. There has been a view among many that relying on large scale ATE is OK because it has a long lifetime and becomes cheaper to use with each passing day.

In other words, if the equipment costs $2 million, and you test 200M units, what is the problem? The problem with that view is that you have to either put up $2 million or sign a long-term lease.

A better deal is to wrap the cost into each chip in the form of BIST and DFT, and pay incrementally.

To put it another way, would you rather bet a few pennies per unit as you make them, or do you still enjoy betting millions of dollars up front? In today's uncertain world, you may want to think this through for one or two seconds.

Now is the time that this kind of argument has great weight with your management. It is also a time when you might have enough time to get it done!

So, I hope that this has been helpful.

Remember; buy nothing until it is needed. Watch your power bill, it's going to go up when the first tank revs its engine in the desert. Save your company's cash and wait until the next (hopefully good) economic wave!