Skip to Main Content
Transistor modeling is an essential process for circuits and systems design. It consists of two steps. The first is to propose a circuit topology. The second is to assign values to those components, by employing optimization techniques, to best fit experimental data. Most of microwave transistor models use a standard topology. Other topologies with different configurations have also been proposed in recent research works and in commercial software packages. To reach an optimum topology, tuning of the initial topology is required. This tuning process is usually based on a manual, trial-and-error procedure and is often specific to a given type of transistor. In this work, a new and efficient tool is introduced. It is able to automatically generate the most appropriate transistor topology as well as find its component values accurately. To illustrate the tool efficiency, experimental transistor data (S-parameters) was used. In this case the results show that the tool-generated topology fits the measured data better than the standard topology. The frequency range of the measurement data is 1 to 40 GHz. Compared to existing techniques, our approach is fully automated and requires minimal expertise from the external user. This point becomes critical when characterizing new components.