Page 23 - Building A Succesful Board-Test Strategy

10 BUILDING A SUCCESSFUL BOARD-TEST STRATEGY

place machine, paste printer, or chip shooter. In these cases, test operations may
identify the problem, but minimizing or preventing its recurrence requires tracing
it back to its source, then performing equipment calibrations or other process-wide
changes.
The relative occurrence of each of these failure mechanisms depends on
manufacturing-process characteristics. Board-to-board process variation, for
example, tends to be most common when assembly is primarily manual. You can
minimize the occurrence of these essentially random failures by lightening
component specifications or automating more of the assembly process.
More-automated manufacturing operations generally maintain very high
consistency from board to board. Therefore, either almost all of a given board lot
will work or almost all of it will fail. Examining and correcting process parame-
ters may virtually eliminate future failures, a potent argument for feeding quality
information back into the process. Under these conditions, quality assurance may
not require sophisticated test procedures. For example, a few years ago, a pick-and-
place machine in an automated through-hole line was miscalibrated, so that all of
the device legs missed the holes. Obviously, the crimper failed to fasten the legs in
place, and when the board handler picked the board up, all of the components
from that machine slid off. Even a casual human visual inspection revealed
that the board was bad, and the pattern of failures identified the correct piece of
equipment as culprit.
It is important to recognize that even a process that remains strictly in control
still produces some bad boards. However diligently we chase process problems as
they occur, perfection remains a myth. Test professionals can rest assured that we
will not be eliminating our jobs or our fiefdoms within the foreseeable future.
You must determine your own failure levels and whether failures will likely
occur in design, purchased parts, or assemblies. Fairly low first-pass yields, for
example—perhaps less than 80 percent—often indicate assembly-process problems.
Very high board yields suggest few such problems. Those failures that do occur
likely relate to board design or to parts interactions. If board yields are very high
but the system regularly fails, possible causes include board-to-board interactions,
interactions of a board with the backplane, or the backplane itself. Understand-
ing likely failure mechanisms narrows test-strategy choices considerably.

1.4.1 Breaking Down the Walls
Test activities are no longer confined to the "test department" in a manufac-
turing organization. Design verification should occur even before prototyping. It
represents one of the imperatives of the simulation portion of the design process,
when changing and manipulating the logic is still relatively painless. In addition,
inspection, once considered a manufacturing rather than a test step, can now
reduce burdens on traditional test. In creating a test strategy, you must therefore
take into account the nature and extent of inspection activities.