Calibrating IC thermal models against experimental measurement is an exercise that engineers have endured since CFD (Computational Fluid Dynamics) became an essential part of the electronics design process. The benefits of calibration, or at least correlation, are generally accepted as reducing design time and improving reliability. Accurate predictive models allow more designs to be evaluated virtually leading to fewer prototypes built and tested.
Historically calibration processes involved the manual perturbation of thermal model inputs, such as material thermal conductivity, until the model results matched thermocouple or IR thermal camera measurements. A limitation with this approach is that it doesn’t capture the dynamic behavior of the IC component. Many of today’s electronics are designed into dynamic and often unpredictable environments where understanding the transient response of the IC is essential.
The challenge for designing electronics into dynamic applications consists of developing accurate predictive models in an acceptable amount of time. Developing an accurate predictive thermal model has been constrained by the finite amount of engineering time available, and the quality or appropriateness of the available measurement data. Using thermal measurement data that captures the transient response of the device offers the best opportunity for developing an accurate predictive model for electronics in dynamic applications. Using an automated calibration process that minimizes the difference between the analytical and measured results decreases engineering time.
This presentation discusses the challenges associated with thermal model calibration and introduces the transient thermal measurement and calibration process. The methodology is illustrated to calibrate an IGBT device.