Page 123 - Building A Succesful Board-Test Strategy

Inspecfion as Test 109

machines succeed better on the monotonous task of inspecting identical products,
Machines are faster, and inspection times do not vary from one board to the next,
It takes fewer workers to run automated equipment than to inspect boards manu-
ally, and automation requires less-skilled (and therefore less-expensive) workers,
Therefore, adopting AOI generally lowers labor costs. Also, automated systems can
resolve finer features than human beings can, although manufacturers have to trade
off resolution against throughput. The finer the required resolution, the longer it
takes to inspect a board.
AOI enjoys other advantages over manual inspection. By reducing the
number of false failures, it reduces costs and improves the productivity of rework
operations. Consistency of results allows feeding information back into the process
to prevent future failures, improving quality out of manufacturing and conse-
quently lowering test and repair burdens as the product matures.
When compared to conventional test, AOI can more easily reside in an auto-
mated production line with a minimum of human intervention. Because it can
detect many faults that otherwise rely on electrical tests, manufacturers can some-
times eliminate the process-test step altogether, reducing capital-investment costs
by avoiding a test-equipment purchase or freeing an existing tester for other
product lines. AOI provides data that test cannot on parts-placement accuracy,
which can affect future product quality. Also, positioning AOI after parts place-
ment and before reflow (and therefore before the possibility of electrical test) can
avoid extra reflow steps and thereby lower repair and rework costs.
Of course, as a wise soul once said, "No good deed ever goes unpunished."
AOI obviously requires a significantly larger capital investment than does manual
inspection. Equipment often costs hundreds of thousands of dollars, including
conveyors and other related infrastructure enhancements. Therefore, overall
startup costs are higher. The equipment requires "programming" in some form,
whereas human inspectors can generally work from existing documentation. For
the same reason, implementing engineering changes and product enhancements
takes longer and incurs more costs in the automated case.
Humans can make pass/fail decisions even when devices are partially hidden.
(A characteristic of humans that machines have thus far failed to match is our
ability to draw conclusions from incomplete information. The binary nature of
machine logic makes such "fuzzy" decision-making complicated at best.) Human
inspectors can also more easily allow for color, size, and other cosmetic variations
in the board's parts, as well as lighting variations and other less-than-optimal
conditions.
AOI looks for very specific features—part placement, part size, perhaps
board fiducials of a certain size and position, and patterns of light and dark, such
as bar codes. It can also look at label type fonts and sizes, although this level of
resolution slows the inspection step. Unfortunately, many boards include compo-
nents that exhibit large variations in package sizes and styles. Figure 3-12 shows
several examples of electrically identical components that appear different to an
AOI system. Automated inspection must also deal with changes in background
color and reflectivity and differences in silk-screen typefaces and sizes on allegedly
identical boards.