Publications

PhD Thesis


< back to PhD thesis

« Built-In Self-Test solutions for high-performance and reliable analog, mixed-signal, and RF integrated circuits ».

Author: M. Barragan
President of jury: D. Dallet
thesis reviewer(s): M. Renovell, F.V. Fernandez,
thesis examinator(s): M.-M. Louerat, D. Morche,
HDR Université Grenoble Alpes
Speciality: Micro et Nano Electronique
Defense: July 09 2019
ISBN: 978-2-11-129254-3

Abstract

The integration capabilities offered by current nanoscale CMOS technologies enable the fabrication of complete and very complex mixed-signal systems. However, manufacturing processes are prone to imperfections that may degrade –sometimes catastrophically– the intended functionality of the fabricated circuits. Extensive production tests are then needed in order to separate these defective or unreliable parts from functionally correct devices. Unfortunately, the co-integration of blocks of very distinct nature (analog, mixed-signal, digital, RF, ...) as well as the limited access to internal nodes in an integrated system make the test of these devices a very challenging and costly task.

BIST techniques have been proposed as a way to overcome these issues. These techniques aim at including some of the ATE functionality into the Device Under Test, in such a way that each fabricated system becomes self-testable. Applying BIST to the digital part of a complex integrated system is a common and standardized practice. Many test alternatives broadly proven in practice are available, all of them based on defect test and fault models. On the other hand, AMS-RF BIST techniques are still lagging behind due to the strict requirements imposed by the analog circuitry. Since AMS-RF circuits are usually tested by measuring their functional specifications, this means that each measurement has to comply with strict accuracy constraints to match the performance of the circuits under test.

A promising solution to these issues is the combination of BIST strategies and machine learning-based tests. Machine learning test strategies replace costly analog, mixed-signal and RF performance measurements by a set of simpler measurements that can be performed on-chip by low-cost built-in test circuitry. The core idea is to build a mapping model from a set of simple measurements to the set of functional specifications. However, this test strategy is not free of shortcomings either.

My research has been focused on overcoming the limitations of current BIST and machine learning-based test for complex AMS-RF circuits, with the final goal of providing innovative state-of-the-art test solutions for these complex systems

pdf pdf