Page 54 - Building A Succesful Board-Test Strategy

40 BUILDING A SUCCESSFUL BOARD-TEST STRATEGY

1.7.1 Cost Components

The total cost of any manufacturing and test strategy comprises a large
number of contributing factors. Most of these factors are interrelated, so that
saving money in one area often means spending it somewhere else. Design costs
represent the most obvious—although by no means the only—example of this
phenomenon. Designing a product for manufacturability and testability increases
costs at the design stage. The anticipated benefit is downstream costs that are lower
by at least enough to compensate.
Moreover, a product that facilitates manufacture and test permits easy
process changes to accommodate increasing or decreasing product demand, design
updates, or quality improvements. Conversely, designers concerned only with
keeping their own costs down condemn the product to delays in manufacturing,
quality problems, and higher-than-necessary overall costs.
Translating a theoretical circuit design into a practical system directly affects
purchasing and inventory costs. Consider, for example, a design that calls for a
number of resistor types, including 2-kO 2 percent resistors, as well as pull-up resis-
tors for some of the digital signal lines. For this application, pull-ups can range
from about 1 kO to 5kO. Nominal resistances can be off by 50 percent or more
and the circuit will still function adequately. The less-precise parts undoubtedly
cost less than the 2 percent versions. However, adding a part number to purchas-
ing and inventory tracking and allocating additional inventory space often cost
more than simply specifying the already stocked component for the less demand-
ing board position as well.
This simple tactic also has test implications. Just because a parts list calls for
a 2 percent pull-up resistor does not mean that the board test should necessarily
flag any part that fails to meet that specification. Test programmers must consider
the board function. Testing a pull-up resistor at 2 percent will produce more fail-
ures than testing it at, say, 25 percent. Boards will fall out of the production cycle
for diagnosis, repair, and retest, increasing test costs. But because the pull-up appli-
cation does not demand 2 percent precision, the higher costs will not improve
actual board quality at all.
A manufacturing engineer's job is to build a functioning product at the
lowest possible cost. Design engineers and test engineers must ensure that the
product works. Creating a test specification blindly from a purchasing or manu-
facturing document ignores differences in the responsibilities of each group. Devel-
oping test programs in this way is often less expensive than painstakingly modifying
automatically (or manually) generated programs to accommodate practical
differences from those specifications, but test costs will be higher. Again, the aim
is to reduce the total cost burden, regardless of the distribution among indi-
vidual departments.
Startup costs for any new-product introduction include the cost of idle
manufacturing capacity during retooling and rarnp-up, the cost of simulations and
physical prototypes, and the salaries of engineers, supervisors, and other non-
routine personnel who participate in this project phase. A product that is relatively