The influence of various common assumptions on the boundary-condition-independence of compact thermal models

It has been repeatedly shown that high accuracy can be obtained when comparing boundary-condition-independent compact thermal models (CTMs) to detailed model results. However, it should be realized that these results have been generated using certain assumptions (e.g., uniformly distributed boundary conditions), and the question remains about the validity of these assumptions when CTMs are used in "real-life" environments. The results show that the assumptions are justified, at least for the package studied, with the exception of boundary conditions associated with heat spreaders and heat sinks. For this type of boundary condition, we need to specify separate heat transfer coefficients for the two nodes at the surface. Finally, an important conclusion is that the "traditionally" generated CTMs perform very well in "real-life" environments.