The MMA Algorithm was conceived by S. D. Millman, E. J. McCluskey, and J. M. Acken as a way to diagnose bridging faults using only stuck-at fault information .
Byzantine Generalís Problem
The Byzantine Generalís Problem is one of interpreting node logical values given variable downstream logic thresholds. Please see "Beyond the Byzantine Generals: Unexpected Behavior and Bridging Fault Diagnosis" for how the Byzantine Generalís Problem relates to bridging faults .
Weighted Critical Area
A value generated by Carafe, representing the likelihood of a fault based on circuit layout and defect density information [4, 5].
A faulty circuitís response to a test set. Contains the faulty node(s), output(s) and test(s) used. This can be generated for various types of faults.
Composite Signature (for a node based fault)
The union of the two associated stuck-at signatures for a given node.
Composite Signature (for a bridging fault)
The union of the four stuck-at signatures associated with a bridging fault. (Each node in a bridge is going to have two associated stuck-at faults.)
Empty Composite Signature (for a bridging fault)
If there are no detectable stuck-at faults for the two bridged nodes, then there will be an empty composite signature.
When diagnosing a fault based on observed faulty behavior, a composite signature containing errors that are a superset of the errors contained in the observed faulty behavior, is a match.
When the nodes in the composite signature attempt to identify the nodes involved in the observed fault behavior, a correct match correctly identifies both of the nodes involved in the bridge.
When the nodes in the composite signature attempt to identify the nodes involved in the observed fault behavior, a partial match correctly identifies only one of the nodes involved in the bridge.
When the nodes in the composite signature attempt to identify the nodes involved in the observed fault behavior, a misleading match identifies neither of the nodes involved in a bridge.
The list of faults that best explain the response of the faulty circuit.
A correct diagnosis occurs when the list of faults contains a candidate fault that most closely corresponds to the actual failure mechanism.
An exact diagnosis is a diagnosis derived from only the correct match. (This will be a "list" of 1 fault.)
A partial diagnosis is a diagnosis derived from the correct match in addition to other matches. (The correct match is somewhere in the diagnosis.)
An incorrect diagnosis occurs when the list of faults does not contain the candidate fault that most closely corresponds to an actual failure mechanism. (The correct match is nowhere in the diagnosis.)
An incorrect diagnosis that is derived from only partial matches.
An incorrect diagnosis that is derived from only misleading matches.
An incorrect diagnosis in which the fault list is empty. This is the most desirable of all the incorrect diagnoses because it is clearly incorrect and can not mislead an engineer.
An improvement to the MMA Algorithm that flags vectors as incapable of detecting a particular bridging fault.
An improvement to the MMA Algorithm that flags vectors as dependably detecting a particular bridging fault.
Given a failed diagnosis, a failure recovery is attempted, and a best guess as to the possible defect is attempted .
If a bridging fault can create a feedback loop in the circuit, some test vectors may cause the circuit to oscillate. Such vectors are said to possibly detect the bridging fault. The inclusion of these vectors can lead to misleading and failed diagnoses.