Page 70 - Building A Succesful Board-Test Strategy

56 BUILDING A SUCCESSFUL BOARD-TEST STRATEC7Y

reduced power consumption by a whopping 95 percent! Increased board func-
tionality led to a proliferation of new designs, wider applications, and much
higher production volumes. As designers took advantage of new technologies,
boards took on an increasingly digital character. Manufacturing engineers deter-
mined that if they could inject signals into digital logic through the edge-
connector fingers that connected the board to its system and measure output
signals at the edge fingers, they could verify digital behavior. This realization
marked the birth of functional test.
Unfortunately, functionally testing analog components remained a tedious
task for a technician with an array of instruments. Then, a group of engineers at
General Electric (GE) in Schenectady, New York, developed an operational-
amplifier-based measurement technique that allowed probing individual analog
board components through a so-called "bed-of-nails," verifying their existence and
specifications independent of surrounding circuitry. When GE declined to pursue
the approach, known as "guarding," the engineers left to form a company called
Systomation in the late 1960s. Systomation incorporated guarding into what
became the first true "in-circuit" testers.
In-circuit testing addresses three major drawbacks of the functional
approach. First, because the test examines one component at a time, a failing test
automatically identifies the faulty part, virtually eliminating time-consuming fault
diagnosis. Second, in-circuit testing presents a convenient analog solution. Many
boards at the time were almost entirely either analog or digital, so that either in-
circuit or functional testing provided sufficient fault coverage. Third, an in-circuit
tester can identify several failures in one pass, whereas a functional test can gen-
erally find only one fault at a time,
As digital board complexity increased, however, creating input patterns and
expected responses for adequate functional testing moved from the difficult to the
nearly impossible. Automated tools ranged from primitive to nonexistent, with cal-
culations of fault coverage being equally advanced.
Then, in the mid-1970s, an engineer named Roger Boatman working for Test-
line in Florida developed an in-circuit approach for digital circuits. He proposed
injecting signals into device inputs through the bed-of-nails, overwhelming any
signals originating elsewhere on the board. Measuring at device outputs again pro-
duced results that ignored surrounding circuitry.
Suddenly, test-program development became much simpler. In-circuit testers
could address anything except true functional and design failures, and most came
equipped with automatic program generators (APGs) that constructed high-
quality first-pass programs from a library of component tests. Results on many
boards were so good that some manufacturers eliminated functional test, assem-
bling systems directly from the in-circuit step. Because in-circuit testers were sig-
nificantly less expensive than functional varieties, this strategy reduced test costs
considerably.
As circuit logic shrank still further, however, designers incorporated more and
more once-independent functions onto a few standard large-scale integration (LSI)
and very-large-scale-integration (VLSI) devices. These parts were both complex