Models in Memory Testing - Models in Hardware Testing - page 167

This set of FPs can be grouped to define a set of six well established and
characterized FFMs:

1. State fault ( SF x ): the logic value of the target cell flips in correspondence of
a given initialization value, even if no operation is performed. The state fault
should be understood in the static sense, i.e., the cell should flip in the short time
period after initialization and before accessing the cell. This fault is special in the
sense that no operation is needed to sensitize it and, therefore, it only depends on
the initial value stored in the cell. Two types of state faults exist: SF 0 D f FP 1 g ,

and SF 1 D f FP 2 g .
2. Transition fault ( TF x ): the target cell fails to undergo an up (0 ! 1)oradown
(1 ! 0) transition. Two types of transition faults exist: TF 0 D f FP 5 g ,and TF 1 D
f FP 4 g .
3. Read destructive fault ( RDF x ): a read operation performed on the target cell
changes the content of the cell and returns an incorrect value on the memory
output. Two types of read destructive faults exist: RDF 0

D f FP 9 g ,and RDF 1 D

f FP 10 g .
4. Write destructive fault ( WDF x ): a non-transition write operation performed on
the target cell, i.e., 0; w 0 ,or1; w 1 , causes the cell to flip. It is similar to the TF. In
both cases a write operation fails to work properly. Two types of write destructive
faults exist: WDF 0 D f FP 3 g ,and WDF 1 D f FP 6 g .
5. Incorrect read fault ( IRF d ): a read operation performed on the target cell returns
the incorrect logic value while keeping the correct cell content. Two types of
incorrect read faults exists: IRF 0 D f FP 7 g ,and IRF 1 D f FP 12 g .
6. Deceptive read destructive fault ( DRDF d ): a read operation performed on the
target cell returns the correct value while changing the content of the cell [ Adams
et al., 1996 ]. Two types of deceptive read destructive faults exists: DRDF 0 D

f FP 8 g ,and DRDF 1 D f FP 11 g .
The proposed set of FFMs is able to completely cover the set of FPs proposed in
Tab le 6.1 , and therefore any test able to cover these fault models is able to detect
any single-cell static faulty behaviors.
Additional fault models have been defined in the literature; nevertheless, us-
ing the proposed classification, they result in a combination of the six proposed
FFMs. For example the well known stuck-at fault model ( Van de Goor 1991 ), i.e.,
a cell is stuck at a given value for all performed operations, can be modeled as
follows:

SAF 0 D SF 1 [ TF 1 [ WDF 1 , denoting the stuck-at-0.
SAF 1 D SF 0 [ TF 0 [ WDF 0 , denoting the stuck-at-1.
In this case, SFA 0 is defined as a set of three FPs. Each fault primitive in this set is
able to sensitize this fault, i.e., any test that covers at least one of the fault primitives
in this set is able to cover the fault model.