Fault collapsing
In an ATPG process, once a fault has been detected it has to be dropped from the list of faults
There are some faults are either equivalent or some of them has dominance over others and identification of them has advantages in ATPG process in terms of no tests done and thus advantages on speed of operation
Equivalent faults
Faults f and g are functionally equivalent (or simply equivalent) if faulty outputs of them are identical for all test patterns
• Example: A/0 B/0 and C/0 are all equivalent faults for an AND gate so only {1,1} is sufficient to detect all the SA faults in A,B or C where A,B are input & C is output
• Equivalence relationship is symmetric, if fault f is equivalent to fault g, then fault g is equivalent to fault f
• Equivalence relationship is also transitive, if fault f is equivalent to fault g and fault g is equivalent to fault h, then fault f is equivalent to fault h
Equivalence Fault Collapsing
Faults f and g are functionally equivalent (or simply equivalent) if faulty outputs of them are identical for all test patterns
General equivalence rule:
- Input SA0 & output SA0 in AND gate
- Input SA0 & output SA1 in NAND gate
- Input SA1 & output SA1 in OR gate
- Input SA1 & output SA0 in NOR gate
- Input SA0/1 & output SA1/0 in NOT gate
For every such equivalence class of faults, one has to retain only one and rest has to be discarded
Dominant Fault Collapsing
If all tests for some fault fi also detects another fault fj , then fj said to be dominate fi
– for example, output stuck-at-1 dominates the input stuck-at-1 faults in an AND gate
For such dominance relation , we retain only fj and remove the fault fi, and patterns only for fj will be applied as in doing so also fi would also be detected
Note: When two faults are dominates each other they become a equivalent fault