New Neural Procedure for Extraction of Parameters of Microwave FET Noise Models

Abstract

A neural netwok based procedure for extraction of parameters of microwave FET noise models is proposed in this paper. The procedure is illustrated on the Pospieszalski’s noise model. A neural network is trained to predict the equivalent drain noise temperature for given values of the intrinsic circuit elements, intrinsic noise parameters, ambient temperature and frequency. In this way, the repeated extraction procedures in microwave circuit simulators are avoided. The presented approach is validated on a specific HEMT device noise modeling. The results are compared with the results obtained for the model parameters extracted by using the standard optimization based extraction procedures.

Keywords

• Artifcial neural networks
• equivalent noise temperatures
• HEMT
• MESFET
• noise model