Linear Regression-Based Searching for Minimum Supply Voltage in Semiconductor IC Testing

• Main Authors: LIANG, I-JUI, 梁益瑞
• Other Authors: YAN, JIN-TAI
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/54bf5h

碩士 === 中華大學 === 資訊工程學系 === 105 === Vdd(min) is defined as the lowest supply voltage at which a device can work on right logic states. For 65nm and smaller process of CMOS VLSI manufacturing, the IDDQ based outlier analysis can not be used to detect the defects, and Vdd(min) testing is required analysis for screening the defects. For Vdd(min) test, previous approaches to search the pass/fail point typically use a standard sequential or binary search which is used to search minimum voltage step by step, and searching becomes inefficient. Linear Regression analysis is a statistical methodology that utilizes the relation between two or more variables so that a response or outcome variable can be predicted. This thesis proposes a Vdd(min) determination approach based on Linear Regression which can predict a rough Vdd(min) value in the first step instead of unnecessary searches, and find an accurate Vdd(min) value in the second step based on standard sequential search.