Learning-Based Cell-Aware Defect Diagnosis of Customer Returns

Abstract

In this paper, we propose a new framework for cellaware defect diagnosis of customer returns based on supervised learning. The proposed method comprehensively deals with static and dynamic defects that may occur in real circuits. A Naive Bayes classifier is used to precisely identify defect candidates. Results obtained on benchmark circuits, and comparison with a commercial cell-aware diagnosis tool, demonstrate the efficiency of the proposed approach in terms of accuracy and resolution.

A comprehensive framework for cell-aware diagnosis of customer returns

“…In [28][29][30][31][32][33], we proposed several learning-guided solutions for CA diagnosis of mission mode failures in customer returns. All solutions are based on a Bayesian classification method for accurately identifying defect candidates in combinational standard cells of a defective IC.…”

“…In this paper, we first summarize the work presented in [28][29][30][31][32][33] and explain how these CA diagnosis solutions are used depending on the failing test scenario. Next, we extend the previous work by dealing with sequential cells and diagnosis of related defects in customer returns.…”

“…To this end, each cell-pattern for a sequential cell is considered as a tuple in which the first value represents the input clock signal, the second value is associated to the main input of the cell, and the third value is associated to a virtual input pin representing the previous value of the output pin of the cell. Owing to this cell-pattern representation, sequential cells can be handled in the same manner than combinational cells by the learning-based CA diagnosis flow utilized in [28][29][30][31][32][33] and summarized in Section 3. New experiments done on a silicon test chip and a customer return have proven the efficacy of our flow in terms of diagnosis accuracy and resolution for a given test scenario.…”

“…Section 2 summarizes our previous work. Section 3 presents the generic CA diagnosis flow used in [28][29][30][31][32][33] and summarizes the various contributions. Section 4 shows how defects in sequential cells can be dealt with the CA diagnosis flow.…”

A Learning-Based Cell-Aware Diagnosis Flow for Industrial Customer Returns

“…The proposed method is based on a Gaussian Naive Bayes (NB) trained model to predict good defect candidates. A generic description of this method was introduced in [19], with partial results obtained on benchmark circuits. In this paper, we propose a comprehensive description of our approach to show its superiority when compared to a commercial CA diagnosis tool.…”