Prognostics of connection defects in electronics modules

Abstract

Electrical connection anomalies are widely known as problems in industrial applications. In machinery these can occur in a printed circuit board, for instance in a control unit, in the interfacing connectors, or even in the inter-connections within a sensor or an actuator. Different methods have been proposed in the last decades to detect these anomalies. Most of these methods are working on non-powered systems for instance by measuring reflections of a high speed pulses (reflectometry principle). However, the most challenging connection anomalies have intermittent nature, often in applications where vibration stresses are dominant (e.g. automotive sector). In these cases, detecting these anomalies in a live (powered) system is needed. Although some methods exist which allow such a detection, they are expensive, limiting the targeted applications (e.g. aeronautics). We developed low-cost methods to detect connections anomalies in live systems. These methods use measured signals, for instance across a sensor, not only to detect a fault in its connections but also to localize the position of such a fault. The used ‘diagnostic’ systems will trigger a warning at a very early stage of the connection anomaly. The proposed methods have been applied and validated on different industrial cases and proved to be able to localize within 30 cm accuracy and to detect connection degradation before corrupting the operation of the system.