Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Comparing prediction methods for LED failure measured with Transient Thermal Analysis

 
: Zippelius, A.; Hans, A.; Liu, E.; Schmid, M.; Perez-Velazquez, J.; Elger, G.

:

Institute of Electrical and Electronics Engineers -IEEE-:
21st International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems, EuroSimE 2020 : 5-8 July 2020, Cracow, Poland, Virtual Event
Piscataway, NJ: IEEE, 2020
ISBN: 978-1-7281-6049-8
ISBN: 978-1-7281-6050-4
S.110-117
International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE) <21, 2020, Online>
Englisch
Konferenzbeitrag
Fraunhofer IVI ()

Abstract
Accurately predicting the lifetime of an electrical component is an important part of the design process, both in the initial creation of the component and later when integrating it into larger devices. This is especially important in applications with high reliability standards, such as automotive LEDs. In this work, we compare methods to predict failures in solder joints of LEDs during accelerated stress testing (temperature shock tests) using transient thermal analysis (TTA). We compare a statistical approach and the use of artificial neural networks (ANN) with memory for two types of prediction: the state at a specific number of shock cycle and for a fixed number of cycles into the future. Our results show that the data-driven approach improves the prediction compared to classical methods in both scenarios.

: http://publica.fraunhofer.de/dokumente/N-633635.html