Hybrid small-signal model parameter extraction for GaN HEMT-on-Si Substrates based on the SPF method

This article proposes a parameter extraction method suitable for Si substrate based GaN HEMT small signal equivalent circuit models. The proposed method is based on a swarm intelligence optimization algorithm, which improves efficiency and accuracy by introducing a slope penalty factor (SPF) method for the objective function, rather than simply minimizing the error between simulation and measurement. By using PSO, WOA, and PNC-WOA for validation, we have demonstrated the advantages of the SPF method in extracting small signal parameters using swarm intelligence optimization algorithms. It is suitable for complex small-signal models that can describe the GaN HEMT-on-Si substrates, even if working up to 40 GHz.