Page 38 - Building A Succesful Board-Test Strategy

24 BUILDING A SUCCESSFUL BOARD-TEST STRATEGY

percentage of that total. Therefore, changes will not affect the system's overall man-
ufacturing cost very much. An appropriate test strategy for this product tends to
be fairly inexpensive but must be reasonably exhaustive. A successful strategy must
ensure that the product will work, but many strategic choices can satisfy that cri-
terion. Test engineers should opt for the easiest implementation, rather than select-
ing the ultimate in state-of-the-art test equipment.
In contrast, a piece of magnetic resonance imaging (MRI) equipment,
another high-priced, low-volume product, requires edge-of-technology computer
engines, extremely fast data-transmission rates, digital-signal processing, lots of
memory, and other high-end capabilities. Testing such products is both more dif-
ficult and more expensive than testing elevators. Also, electronics make up the lion's
share of the product's technology, so that test represents a substantial portion of
overall manufacturing costs. Therefore, test-strategy decisions will directly affect
the company's bottom line.
As with the elevator, test strategies must stress quality, but this time test
equipment will likely be both state of the art and complex. Fortunately, low
production volumes permit long test cycles with little impact on schedules.
Sometimes the end use dictates the nature of the test strategy. Automotive
electronics are of the same level of complexity as, say, a PC. But an automobile—
which contributes heat, vibration, voltage spikes, and EMI—represents the most
hostile common environment in which electronics must function. Therefore, the
test strategy must stress ruggedness and durability. Also, the electronic content of
today's cars far exceeds the level of only a few years ago. So testing has become
more complex, more expensive, and represents an ever-larger percentage of the
overall manufacturing cost.
High-volume products, such as PCs, disk drives, and printers, also require
sophisticated test strategies. In these cases, however, fast test times are imperative to
prevent production bottlenecks. Also, because product generations change very
quickly, test-strategy flexibility and fast program development are prime concerns.
Corporate philosophy can also influence the design of a test strategy. Most
automobile companies, for example, will go to extraordinary lengths to avoid
increasing the cost of each board produced by even a fraction of a cent. They
prefer spending the money on test steps and test generation up front to avoid
higher material costs later. Advocating extra components or test nodes in that
environment will generally fall on deaf ears. Instead, test engineers must find ways
to invest during the development phase to reduce material costs during production
as much as possible.
Other companies, especially in high-mix situations, take the opposite
approach. They would prefer to increase material costs slightly to reduce the cost
of test development and setup.
What types of components will the boards contain? Will all components
come off the shelf, or will you use ASICs? Will all digital components require
similar voltages? Will there be ECL parts? Will microprocessors be relatively slow
and uncomplicated, or will they be fast, highly integrated, and state of the art? Will
the board include mixed-signal devices?