This preview shows page 1. Sign up to view the full content.
Design for Testability
47
(a)
(b)
2
/4/0
1/1/3
1/1/3
1/1/3
0.5/0.5/0.25
0.875/0.125/1
0.5/0.5/0.25
0.5/0.5/0.25
±
FIGURE 2.3
Comparison of SCOAP and probabilitybased testability measures: (a) SCOAP combinational measures,
and (b) probabilitybased measures.
measures and probabilitybased testability measures of a threeinput AND gate. The
threevalue tuple
v
1
/
v
2
/
v
3
of each signal line represents the signal’s 0controllability
(
v
1
), 1controllability (
v
2
), and observability (
v
3
).
Signals with poor probabilitybased testability measures tend to be difficult to
test with random or pseudorandom test patterns. The faults on these signal lines
are often referred to as
randompattern resistant
(RPresistant) [Savir 1984]. That
is, either the probability of these signals randomly receivinga0or1 from primary
inputs or the probability of observing these signals at primary outputs is low,
assuming that all primary inputs have the equal probability of being set to 0 or 1
and are independent from each other.
The existence of such RPresistant faults is the main reason why fault coverage
This is the end of the preview. Sign up
to
access the rest of the document.
 Spring '08
 elbarki

Click to edit the document details