Page 34 - Building A Succesful Board-Test Strategy

20 BUILDING A SUCCESSFUL BOARD-TEST STRATEGY

Figure 1-5 100% test effectiveness is possible even with no test or 0% manufacturing
yield.

The effectiveness of any test strategy depends on the corresponding
effectiveness of the manufacturing process. Consider, for example, a perfect man-
ufacturing process that produces only good products. In that case, any test
strategy—even no test at all—will produce 100 percent good products, as in Figure
1-5. Assuming that the process remains in control, any expenditure on test would
prove,—at best—superfluous.
In contrast, a perfect test strategy would ship 100 percent good products,
even if every board and system off the manufacturing line contained defects.
Of course, the economics of finding and repairing those defects might present
a considerable challenge. Nevertheless, from the customer's standpoint, the
result would not change. The test effectiveness in both cases (a combination of
manufacturing and test processes) is 100 percent, although the strategies will likely
be very different.
Real manufacturing operations lie somewhere in between. They never achieve
the perfection of the first case, so some test strategy will be necessary. Nor do
they experience the horrors of zero yield, so a less-than-perfect test strategy will
probably suffice. The goal is always to strive for 100 percent test effectiveness,
regardless of results from the individual segments.
One drawback to eliminating redundancy between test steps is concern that
overelimination of test overlap will result in missed faults and, therefore, reduced
product quality. To some extent, this concern is justified—as always, depending
on the nature of the product under test. As with most engineering decisions,
the line between thorough fault coverage and test overlap represents a tradeoff.