Page 135 - Building A Succesful Board-Test Strategy

Inspection as Test 121

Although x-ray inspection can find many faults that elude other methods, it
is not a panacea either. It will not detect a wrong component unless it presents a
very different x-ray profile from the correct one. X-rays cannot detect faulty and
out-of-tolerance components, and will not notice cracked solder joints that cur-
rently appear intact. For x-ray inspection to be effective, you have to combine it
with some combination of process control, electrical test, and environmental-stress
screening.

3,3,5.1 A Catalog of Techniques

X-ray inspection covers a broad range of capabilities. With manual equip-
ment, a human inspector inserts a board into the system, then obtains the relevant
image and makes the pass/fail decision.
The image that the inspector sees may contain only the basic x-ray snapshot,
but it can also include metrology information, or even complete quantitative analy-
sis. Also, the software may enhance the image in some preset way or provide a
level of image processing to make the inspector's decision easier and more
consistent.
This approach offers economy, flexibility, and fast implementation. Cost will
vary depending on the amount of infrastructure and software support, but begins
at less than $50,000. Manual inspection examines boards one at a time, and usually
looks only at critical areas, rather than at the entire board. It works better primarily
for prototyping or during the ramp-up to full production, for random sampling
for process monitoring, and where the nature of the board makes a full inspection
unnecessary. Success depends on throughput requirements, and—depending on
the level of available software assistance—may vary like other manual techniques,
depending on the inspector's experience, the time of day, or the day of the
week.
Semiautomated techniques include an x-ray system and sophisticated image-
analysis software. This version inspects the board for device placement and solder-
joint integrity based on preset gray levels. More expensive than manual alternatives,
it requires longer lead times and a software model of the board for comparison.
However, it also provides much greater consistency, and generates far fewer false
calls.
Most elaborate are the automated systems (so-called automated x-ray inspec-
tion, or AXI), where the software makes pass/fail decisions based on established
heuristics. Long used for inspecting ball-grid arrays, it has become more popular
in the past few years in production because of the difficulty determining board
quality using more traditional test techniques. This alternative is faster than
manual and semiautomated techniques. It is also considerably more expensive and
requires longer startup times. In addition, depending on the throughput require-
ments of the production line, you may have to compromise between comprehen-
siveness and cycle time.
Programming x-ray systems can take two forms. Conventional programming
involves an image of the board—either a constructed image or one obtained from