Verilog-A Based ANN Large Signal Modeling of Gan Hemts

This paper presents an approach to implementing an Artificial Neural Network (ANN) model for simulating AlGaN/GaN High Electron Mobility Transistors (HEMTs) in the Advanced Design System (ADS) environment. The complexity of the ANN model increases with the number of layers, making analytical equation-based descriptions cumbersome. To overcome this challenge, it proposes a matrix-based framework that streamlines model development for varying layer configurations. The implementation employs Verilog-A coding, which is well-suited for circuit simulations, enabling seamless integration of the ANN model into ADS. This paper demonstrates the practicality of the approach by fitting DC-IV and RF large-signal measurements of GaN-on-Si HEMT.