Figure 4.7: An example of a circuit with sensitizable and Unsensitizable paths.
Figure 4.7 shows examples of a sensitizable path and an Unsensitizable path.
Considerfollowingpathinthecircuit: P2'. a-g2-g-g4-k. Topropagateatransition
in the path P2, d should be 1 and h should be O. Choosing c = 1 and b = O
will satisfy these constraints. Thus, P2 is a sensitizable (testable) path. However,
path pɜ: c-f-g3-h-g4-k is not sensitizable. Propagation of a transition in this path
happens if and only if g = O and e = O. To satisfy g = O, we should have a = 1
and d = 1. It contradicts with e = O. Thus, P3 is Unsensitizable.
To ensure that a path is sensitizable, we should generate two input vectors
for the circuit such that a transition propagates in the path. Creation such input
vectors might be very complex and take a long time. Thus, we determine a path
is testable or not in two steps: primary necessity check and using automatic test
pattern generation (ATPG) tools.
Primary necessity check is based on the partial path sensitization introduced
by Murakami et al. [56]. Using the topology and functionality of the circuit, they
introduce bf-pairs in the circuit. Each bf-pair consists of a b-line (back line) and
56