Publications

Publications

< retour aux publications

Density estimation for analog/RF test problem solving

Auteur(s) : S. Mir, H. Stratigopoulos, A. Bounceur

Doc. Source: 28th IEEE VLSI Test Symposium

Publisher : IEEE

Pages : 41

Doi : 10.1109/VTS.2010.5469620

The reduction of analog/RF production test costs calls for the optimization of specification-based tests or their replacement by low-cost ones. A variety of techniques have been proposed in recent years, such as specification-based test ordering and compaction, alternate test to predict the specified performances from simpler measurements, and onchip circuitry to monitor IC performance while alleviating the complexity of test equipment. Since statistical evidence about test quality can only be gathered from large volume production test data, most innovative test techniques fail to gain acceptance for real products. This is because by the time this evidence could be obtained, most test costs would have already been engaged. At this stage, the change of the test procedure would hardly be acceptable and the modification of the design is no longer possible. An early evaluation of test quality at the design stage can be an essential step for promoting innovative test techniques from concepts to actual solutions. This evaluation can be based on a statistical model of the relationships between the specification-based tests and the proposed lower cost ones, that is, a model that represents the joint probability density function of the tests. This presentation will first introduce parametric and nonparametric density estimation techniques for building this model from a small sample of data typically obtained through Monte Carlo simulation. By sampling the statistical model, it is possible to generate a very large sample of synthetic devices. This large sample can effectively replace production data for addressing test problems before the actual production testing takes place. Density estimation will be next applied for solving a variety of analog/RF test problems, including: the selection of most suitable DFT/BIST techniques by evaluating parametric test metrics such as defect level and yield loss; the enhancement of the training phase of alternate test techniques that are b- - ased on machine-learning; the construction of defect (or outlier) filters suitable in particular for alternate test and for fault diagnosis; and the ordering of specification-based tests for the initial phase of production testing. These applications of density estimation will be illustrated using simulation and experimental data for RF devices.