ON EFFECTIVE THERMAL CHARACTERIZATION OF ELECTRONIC PACKAGING

摘要

In this study, an effective methodology that integrates an infrared (IR) thermography measurement and a 3-D finite element (F.E.) model is proposed for thermal characterization of packages in a steady state under a natural convection environment based on JEDEC specification (EIA/JESD51). To perform the surface temperature measurement using the IR thermometer, the black paint coating is applied on the surface of packages so as to control the surface radiation. The emissivity associated with the black paint coating is assessed using a simple calibration experiment. In addition, the effect of the volume of the black paint coating on the surface temperature is carefully examined, and the optimum thickness of the coating is determined. To substantiate the proposed methodology, a typical 100-pin Thin Quad Flat package (TQFP) is considered as the test vehicle. The junction-to-ambient (J/A) thermal resistance as well as the chip junction temperature calibrated by the proposed methodology is considerably substantiated by using a thermal test die measurement. Additionally, the effect of both radiation and convection on the chip junction temperature is extensively studied through a simple modeling experiment. Data show that the effect of the convection is obviously more significant than that of the radiation; however, in a natural convection condition, both these effects are not negligible in the package-level thermal analysis. Furthermore, additional studies are performed to assess the appropriateness of the correlation models for heat transfer coefficients, suggested by Ellison, (1989), Aghazadeh and Mallik, (1990), Mulgaonker et al., (1994), Edwards et al. (1995), and Ridsdale et al. (1996). Since these models except the Ridsdale et al.'s only account for the convection effect, they are modified through the integration of the radiative heat transfer coefficient correlation model introduced in Ridsdale et al. (1996). These modified correlation models are extensively evaluated against the experiment (i.e., the thermal test die approach). The identified correlation model that provides the best approximation to the experiment can be employed in the F.E. model for subsequent parametric design of thermal performance of packages.