Efficient Pattern Generation and Observation Point Insertion for Transition Fault Diagnosis

• Main Authors: Yi-DaWang, 王奕達
• Other Authors: Kuen-Jong Lee
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/4d5hj7

碩士 === 國立成功大學 === 電機工程學系 === 102 === Two major techniques are proposed to increase diagnosisability of circuits: 1. Diagnostic pattern generation (DPG) generates high quality diagnostic patterns to distinguish transition fault pairs and identifies equivalent fault pairs. 2. Observation point insertion inserts observation points into circuit to distinguish the fault pairs. The first technique mainly consists of two methods: 1. Fault Inactivation Method (FIM) generates diagnostic patterns to distinguish fault pairs by inactivating one fault and detecting the other. 2. Fault Propagation Method (FPM) generates diagnostic patterns to distinguish fault pairs by initializing both faults at the same time and creating different faulty responses of two faults on outputs. Experimental results show that the diagnosis resolutions in ISCAS89 (ITC99) benchmarks can reach 99.999999% (99.999995%). But few fault pairs cannot be distinguished or identified as equivalent fault pairs. We use the second technique to insert observation points to distinguish these fault pairs. Experimental results show that only 2 (109) observation points are needed for ISCAS89 (ITC99) benchmarks, and hence the diagnosis resolutions for all circuits are 100% with area overhead less than 1%. In addition, for the indistinguished fault pairs with large distance between two faults in a pair, we also adopt our second technique. Experimental results show that most of fault pairs can be distinguished by inserting few observation points.