Page 18 - Building A Succesful Board-Test Strategy

What Is a Jest Strategy? 5

Suppose, for example, that the manufacturer of a digital device specifies a
"0" as less than 0.8 volts. Based on that specification, a parametric measurement
of a logic low at 0.81 volts would fail. Yet will the system actually not perceive a
voltage of 0.81 as a clean "0"? How about 0.815 volts? The answer, of course, is a
firm "It depends." The situation resembles the century-old conundrum: How many
raindrops does it take before a baseball field is wet enough to delay the game? In
that context, the question seems absurd. You can't count raindrops! Yet at some
point, someone must make a value judgment. Most baseball people accept the fact
that delaying the start of a game usually requires less rain than does stopping play
once the game has begun. Similarly, the question of how closely a circuit must
conform to published specifications may depend on surrounding circumstances.
The real question remains: Does the product work? As product complexity con-
tinues to skyrocket, the necessity to accept the compromises implicit in this
approach become glaringly apparent.
Compounding the challenge, issues of power consumption, heat dissipation,
and portable-product battery life have required drastically reducing operating volt-
ages for most digital systems. The 5V transistor-transistor logic (TTL) parts of the
past have yielded to devices operating at less then 3V, with more to come. As a
result, the gap between a logic " 1" and a logic "0" narrows every day. Devices must
perform more precisely; boards and systems cannot tolerate electromagnetic inter-
ference (EMI) and other noise that were commonplace only a few years ago. New
generations of test equipment must cope with these developments, and test strate-
gies must take them into account.

1.3.1 The First Step

Consider a (loaded) question: What is the single most important considera-
tion in developing a test strategy? The answer may seem obvious. Yet in board-test-
strategy seminars from New York City to Singapore, responses range from budgets
to design-for-testability to "Do we need to test?" to "Do we choose in-circuit or
functional test?" Before facing any of these issues, however, designing a successful
test strategy requires determining the nature of the product. That is, what are you
trying to test? What is the product? What does it look like? How does it work?
What design technologies does it contain? Who is designing it? Who is manufac-
turing it? Who is testing it? In many organizations, one obstacle to arriving at an
effective test strategy is that the people involved decide on test-strategy components
and tactics before answering these simple questions.
Test engineers do not design products. If nobody tells them what the product
is, how it is designed, and what it is supposed to do, their decisions may make no
sense. They might arrive at a correct strategy, but only by accident, and it would
rarely represent both the most successful and the most economical approach.
Test components or test strategies that work for one company or product line
may not be appropriate in another situation. If you do not know what you are
trying to test, you cannot systematically determine the best strategy. Even if you
find a strategy that works, thoroughly knowing the product will likely help you