An Implicit Grouping Algorithm for Multiple Faults Diagnosis of Full-Scan Design

• Main Authors: Yu-Chiun Lin, 林祐群
• Other Authors: Shi-Yu Huang
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/63064354003108133427

碩士 === 國立清華大學 === 電機工程學系 === 90 === It is a difficult task to locate the fault sites in an IC that fails manufacturing test. But knowing where the fault is helps us to find the manufacturing defect and improve the yield. A diagnosis tool can help us to shrink the suspect region and to ease the work of the failure analysis. Several types of diagnosis methodology have been proposed. For full-scan designs with only one fault, the accuracy has been improved significantly. However, the techniques for multiple fault diagnosis remain inadequate so far. In this thesis, we investigate a multiple-fault diagnosis scheme for combinational circuits or full-scan designs. In our approach, we first target the structurally independent faults based on a concept called “prime candidates”. Then, we perform a grouping algorithm to separate the rest of the suspect region into groups, each of which requires its own single-fault diagnosis process to find one fault. Experimental results on three real designs demonstrate that this approach is able to identify more faults than previous ones. On the average, we identify 3.84 faults in only 10 signal inspections with 5 faults in the chip.