Page 103 - Building A Succesful Board-Test Strategy

Test Methods 89

For many electronics manufacturers, especially those of complex, low-
volume, and low-failure-rate products (such as test instruments and systems), func-
tional testers do not perform fault isolation at all. Bad boards proceed to a repair
bench where a technician, armed with an array of instruments and an under-
standing of the circuit, isolates the fault manually.
Because manual analysis takes place offline, it maximizes throughput across
the tester. Because finding faults on some boards can take hours (or days) even
with the tester's help, this throughput savings can be significant.
Manual techniques can be less expensive than tester-bound varieties. Most
test operations already own logic analyzers, ohmmeters, digital voltmeters (DVMs),
oscilloscopes, and other necessary tools, so the approach keeps capital expendi-
tures to a minimum. Also, this method does not need a formal test program. It
merely follows a written procedure developed in cooperation with designers and
test engineers. The repair technician's experience allows adjusting this procedure
"on the fly" to accommodate unexpected symptoms or analysis results. For the ear-
liest production runs, test engineers, designers, and technicians analyze test results
together, constructing the written procedure at the same time. This approach avoids
the "chicken-and-egg" problem of trying to anticipate test results before the
product exists.
On the downside, manual analysis is generally slow, although an experienced
technician may identify many faults more quickly than can an automatic-tester
operator armed only with tester-bound tools. The technique also demands
considerable technician expertise. Speed and accuracy vary considerably from one
technician to the next, and the process may suffer from the "Monday/Friday"
syndrome, whereby the same technician may be more or less efficient depending on
the day, shift, nearness to lunch or breaks, and other variables.
The semiautomatic fault-finding technique with which functional-test pro-
fessionals are most familiar is guided-fault isolation (GFI). The functional tester or
another computer analyzes data from the test program together with information
about good and bad circuits to walk a probe-wielding operator from a faulty
output to the first device input that agrees with the expected value. Performing GFI
at the tester for sequential circuits allows the tester to trigger input patterns peri-
odically, thereby ensuring that the circuit is in the proper state for probing. The
tester can learn GFI logic from a known-good board, or an automatic test-program
generator can create it.
Properly applied, GFI accurately locates faulty components. As with manual
techniques, it works best in low-volume, high-yield applications, as well as in pro-
totype and early-production stages where techniques requiring more complete
information about circuit behavior fare less well.
As with manual techniques, however, GFI is both slow and operator-
dependent. It generally occupies tester time, which reduces overall test capacity.
Long logic chains and the preponderance of surface-mount technology on today's
boards have increased the number of probing errors, which slows the procedure
even further. Most GFI software copes with misprobes by instructing the opera-
tor to begin again. Software that allows misprobe recovery by starting in the middle