The Historical Isolation of Assembly from Test Needs Changing

What's new in test handling? Let me start with the SEMICON West presentations given by Dennis Nelson of MCT and Tim Olson (formerly of FICO USA) on strip handling.

The solutions and equipment the pair discussed are not a panacea but are at least a breath of fresh air. And, after a quarter of a century of the same old thing in test, we're certainly overdue for a change.

Those of us who have been involved in getting this new concept brought forward to its current level are fully aware of the challenges ahead. The major challenge to acceptance and implementation is not hardware or software, it is, surprisingly, the historical isolation and segregation of test and assembly.

As Dennis said in his talk, test and assembly are separate entities under separate roofs and-in many cases-on different continents.

My concern is if this separation of processes and staffs continues, it may well prevent, or at least severely delay, any real improvement in testing philosophy or technology.

So the real challenge to the industry is to come to terms with the present industry structure and make the changes needed to implement test in assembly or assembly in test.

Some of the hot buttons are obvious in the questions being asked and in the challenges being made to the implementation of strip testing:

• How can you test in a lead frame or strip when all of the leads are shorted?
• Get rid of test and move it to assembly?
• Get rid of end-of-line assembly and move it to test?
• What about my job, why should I be out of work?
• What does assembly know about test?
• What does test know about assembly?
• How can you singulate parts after you've tested them and still guarantee that they're good?
• How can you sort good and bad parts?
• Who will be in charge, test or assembly?
• What value or interest is strip testing to me when all of the assembly is done in S.E.A. (South East Asia) and all of the test is done in the U.S.?
• What about defects at singulation? If you test before, you can't detect rejects.

  The one question that I hope to hear in the coming months:

• How can we address the issues, and are they really issues?

All processes in the manufacture and test of semiconductors involve some compromise. There are no perfect processes or designs, but we still manage to overcome the "show stoppers" and work with them (and around them), because the end goal is to get the parts out to the customer.

How does your company handle assembly and test? Are the reasons for isolation really in your company's best interest? Could you reduce costs and cycle times while improving yields-and profit margins-if the entire process were considered and not just viewed as assembly or test?

Before you make your mind up on strip testing, look at your reasons for questioning the process. If you want to reduce the cost of test and improve cycle times and yields, you should ask yourself (and others too), "How, when and where can we implement this change, and what changes do we need to make to our designs, processes and flows to make it possible?"

Remember, getting the correct answer is often the result of asking the right question.