Infrared image processing strategies for die-level power devices inspection.
Infrared analysis is a non-invasive technique especially suitable for electronic devices monitoring, fault finding and thermal studies. Two post-processing strategies for IR thermography, boxcar averaging and Fourier coefficients-based time reconstruction (FoTiR), are analysed, implemented and asses...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2023 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/389285 |
| Acceso en línea: | https://hdl.handle.net/2117/389285 |
| Access Level: | acceso embargado |
| Palabra clave: | Thermography Infrared thermography reliability Fourier semiconductor IGBT RC-IGBT thermal image registration boxcar averaging Termografia Àrees temàtiques de la UPC::Enginyeria de la telecomunicació::Radiocomunicació i exploració electromagnètica::Teledetecció |
| Sumario: | Infrared analysis is a non-invasive technique especially suitable for electronic devices monitoring, fault finding and thermal studies. Two post-processing strategies for IR thermography, boxcar averaging and Fourier coefficients-based time reconstruction (FoTiR), are analysed, implemented and assessed on a power semiconductor device (RC-IGBT). Both approaches are capable of improving the time resolution currently available in IR thermography. The focus is set on enabling short-time inspection of power semiconductor devices under real operating conditions for reliability studies at both levels: die and converter. Moreover, an off-line solution is also provided to compensate the sample displacement during acquisitions by image registration techniques, i.e: image upscaling, image cross-correlation, image translation, and pixel interpolation. To this end, a specific Matlab-based post-processing program has been implemented for all solutions provided and used during the assessment phase. As a result, FoTiR outperforms boxcar averaging in terms of signal to noise ratio and data extrapolation capabilities. |
|---|