Faults and Fault Detection
Last Edit July 22, 2001
Selecting a Chain
When all the possible links have been formed, there will be one or more
observable chains or sets of links. The longest chain defines the desired
Test Sequence.
Figure 9.5 Minimal Test Sequence for the Function Y = X3X2 + X1X0
Minterm
# |
Y |
X3 |
X2 |
X1 |
X0 |
5 |
0 |
0 |
1 |
0 |
1 |
29 |
1 |
1 |
1 |
0 |
1 |
9 |
0 |
1 |
0 |
0 |
1 |
27 |
1 |
1 |
0 |
1 |
1 |
10 |
0 |
1 |
0 |
1 |
0 |
30 |
1 |
1 |
1 |
1 |
0 |
6 |
0 |
0 |
1 |
1 |
0 |
23 |
1 |
0 |
1 |
1 |
1 |
5 |
0 |
0 |
1 |
0 |
1 |
Note that, for each vector, one input changes state and the observable
output changes state. Each input switches from 0-1 and from 1-0 during
the test sequence. If the internal nets X4 and X5 were added to the table,
they would also be observed to switch. This sequence provides 100% fault
coverage for the function.
There are cases when there is no longest chain. When a circuit is redundant
there may be two chains or sequences of equal length and only one needs
to be used for fault detection. Some faults will remain masked regardless
of the sequence selected.
Disjoint functions with terms that share no variable states will have
a disjoint sequence, existing as two or more chains. A stepped sequence
must be generated to connect the disjoint sequences, honoring the rule
that only one input may change per vector.
Advantages of the test sequence
The advantages of the Minimal Test Sequence are listed in Table 9-5.
Table 9-5 Advantages Of The Minimal Sequence
- Reduction of test hazards
- One observable output changes per vector allowing for straightforward
identification of the existence of an error.
- All variables are toggled from 0-1 or 1-0 and back
- Complete coverage of all detectable single faults
- Covers multiple faults
- Closed sequence which allows easy repetitive application
- Possible coverage of bridging faults (not yet researched)
|