Hardware Reference
In-Depth Information
I
2
= 0
I
1
= 0
Gate c
Vn
1
n
3
= effect
Gate a
Vth
c
1
Gate e
O =
effect
Rsh
0
Gate d
I
3
= 1
Gate b
Vth
d
n
2
1
n
4
I
4
= 0
Vn
1
5
5
#2
Vth
c
#6
0
0
R
1
C
R
2
C
Rsh
Fig. 2.11
Different excitations
(
Renovell et al.
1999
)
determined by the input vector. As an example, the vector #2 in Fig.
2.9
propagates
agates the effect through both gates 'c' and 'd' (with I
3
I
4
D
11). Different logic
gates have different logic thresholds. The value of the critical resistance is therefore
different through gate 'c' and through gate 'd'.
A very important point for propagation concerns the re-convergence of effects.
Considering vector #2, the effect of the defect is propagated through a single gate
(gate 'c') implying no re-convergence: the ADI is equal to [0; R
C
]. Considering now
gates (gates 'c' and 'd'). The two effects re-converge on gate 'e'. According to the
unpredictable resistance value, a defect-free or a defective effect may be propagated,
Between0andR
1
C
, the two defective effects cancel each other producing a
defect-free value on the primary output.
Between R
1
C
and R
3
C
, a defective and defect-free effects re-converge producing
a defective value on the primary output.
Between R
3
C
and infinity, the two defect-free effects obviously produce a defect-
free effect on the primary output.
The Analogue Detectability Interval corresponding to vector #3 is consequently
equal to [R
1
C
,R
3
C
]. This small example illustrates that the propagation of the ef-
fect has to be taken into account when defining the ADI.