Page 53 - Building A Succesful Board-Test Strategy

What Is a Test Strategy? 39

practical in the field because of fixture costs and the difficulty of maintaining a
complete fixture inventory for old or obsolete products.
One advantage of a common-machine strategy is the freedom to develop test
programs only once, rather than for each process step. Also, this solution allows
comparing failure data from each stage directly, permitting quality tracking from
"cradle to grave." Long-lived, low-volume products particularly benefit from
this approach, because field failures provide necessary information about the
manufacturing process. With low-volume products, a manufacturer appreciates
the traceability of each piece.
In high-volume operations, common data facilitate statistical process control.
That is, if analyzing a process indicates that a capacitive network is the most likely
circuit section to fail and field data confirm that fact, the test engineer can assume
a good understanding of the process. If that network does not actually fail, then
the process requires further study.
Of course, any test strategy requires gathering and analyzing test and failure
data. Computer-aided test equipment generates a lot of data. Unless those data
furnish a better understanding of the product or the process, they are useless.
Test people must remember that no test-strategy decision is cast in stone.
Strategies, even strategies that involve considerable investment, can be changed.
Also, no law says that a tester must always test to the limits of its abilities. A high-
end machine can still perform simpler tests if the occasion warrants it. This option
permits running equipment as close as possible to its maximum throughput capac-
ity, even when its most sophisticated capabilities are not needed for a particular
process, a particular advantage for a small company with a limited capital budget.
In addition, a manufacturer can change strategies without buying additional
equipment by reassigning what already resides within the factory. For example, a
product early in its life may require in-circuit testing. As the process matures, a
manufacturing defects analysis may suffice. The simpler test is faster, less expen-
sive to develop, and less expensive to conduct. If the in-circuit tester is not needed
elsewhere, it can also execute the new test, avoiding the necessity to acquire addi-
tional equipment.
Constructing a test strategy requires answering one other very important ques-
tion: How much will it cost? The next section begins to address this critical issue.

1.7 Test-Strategy Costs

Every test-strategy decision brings with it both technical and economic con-
sequences. If a particular test problem permits one and only one solution, either
because of technical or managerial constraints, then the cost of that solution
may be inconvenient, but it is unavoidable. Most manufacturing situations,
however, allow a range of options, with differing quality, scheduling, and cost
ramifications. Test managers must examine each alternative and select the one
that provides the best balance. One cardinal rule applies: You generally cannot
buy less than one of anything.