Current Issue The International Reference for Chip-Scale Electronics, Flip-Chip Technology, Optoelectronic Interconnection and Wafer-Level Packaging May - June 2002 Email the editor

Paul M. Sakamoto Contributing Editor

As I write this, I am on a very crowded 747 on my way home from a trip that included Japan and Taiwan.

Although the recession isn't over, this trip has convinced me that it will end, and that we are now facing the next semi-conductor production ramp. The big signs were crowded airplanes, restaurants and hotels, punctuated with many optimistic customer meetings.

Relief from the Doldrums

Overall, the pace of the trip was very hectic; a big relief from the past year of doldrums.

With my bold forecast of a business upturn in the second half of this year, we can now examine whether we are ready for the likely upturn in rush orders as customers strive to deal with inventories that were run to ground during the downturn.

If you followed the advice given in this space in the March-April issue, ("Things To Do During a Recession") than you are really ready with the equipment and support you built when you had the time.

If not, than we must think about a plan "B."

The first strategy is to look to the contract test and assembly companies. Although they are filling up again, they are not yet full. It is not too late, however, to put in a few contracts for flex capacity.

The very top-tier companies are filling up fast, so if you need their specific services, you need to move now.

Rather than name names, I would just comment that these are the subcontractors that provide the broadest range of services, highest overall quality and sophisticated test development as well as production test.

The next tier, which we can think of as high quality, but with a narrower range of services, is still open to giving out favorable contracts to get business.

In either case, get some relationships going now, because it will be too late in another quarter. Think about getting another 20% of flex capacity going this way.

The next piece of business is to think about changing your test flow to be more flexible.

For instance, can you replace a single socket test step that uses an SOC (system on a chip) test system to take advantage of two lesser testers?

This runs counter to normal cost and throughput engineering, but if you cannot invest in the capital or contracts to insure sufficient SOC capacity, this kind of fall back could insure your ability to ship during a crunch.

In some cases, it may not even cost more, when the broken out steps may be run on older, under-utilized, and fully depreciated equipment. By the way, your engineering staff will go out of their way to tell you that this is a really stupid idea, because it will really add to their work if this needs to be done.

Test Strategy

Another possibility is to use a scalable test strategy, such as BIST (built in self-test).

Many BIST and DFT (design-for-test) processes reduce the number of high-speed tester channels required per device.

Reduced channels may allow for either simpler, cheaper and more easily produced test equipment or for putting more devices in parallel on the same tester. Depending on the "yield limiter" of the device in question, this approach can increase capacity by allowing more rapid equipment acquisition and by making the use of that equipment more efficient.

By the way, if you are thinking of parallel test, don't forget to select and evaluate a parallel handler. Packaging changes rapidly obsolete handling equipment and properly evaluating these complex mech-anical contrivances takes some time.

So, overall, I hope my predictions are right. I also hope that you are able to build some flexibility into your test capacity so that we can all ride this forthcoming new wave of business.