| Type of Document |
Thesis |
| Author |
Martinez, Hector Abel
|
| Author's Email Address |
abelhect@hotmail.com |
| URN |
etd-07112005-103752 |
| Title |
Small Signal and Noise Temperature Modeling of Microwave MESFETs using Artificial Neural Networks |
| Degree |
Master of Science |
| Department |
Electrical and Computer Engineering, Department of |
| Advisory Committee |
| Advisor Name |
Title |
| Mark Weatherspoon |
Committee Chair |
|
| Keywords |
- Artificial Neural Networks
- Mesfets Modeling
|
| Date of Defense |
2005-06-29 |
| Availability |
unrestricted |
Abstract
This thesis presents a study of modeling of metal semiconductor field effect transistors (MESFETs) characteristics using artificial neural networks. A radial basis function artificial neural network (RBF-ANN) model is developed for scattering parameters and equivalent circuit parameters (ECPs), and a multilayer perceptron neural network (MLP-ANN) model is developed for incident available noise temperature parameters (TBs1B) of MESFETs. The training and testing data for these models is obtained from measured two-port scattering parameters, extracted ECPs, and measured TBs1B of a 4x50µm gate width, 0.25µm gate length gallium arsenide (GaAs) MESFET. A four-input, eight-output ANN is used to model the S-parameters of a microwave MESFET versus bias, temperature, and frequency; a three-input, eight-output ANN is used to model the ECPs of a microwave MESFET versus bias and temperature; and a two-input, one-output ANN is used to model the TBs1B of a microwave MESFET versus load reflection coefficients and load impedances. Comparisons of measured and modeled data are presented and the results show very good agreement. The average relative errors using the RBF-ANN models for the S-parameters and ECPs were 0.81% and 0.77% respectively. The results of the S-parameters and ECPs represent about 60% reduction in error when compared to Backpropagation ANN models of similar parameters of the same device. The incident available noise temperature model is a novel study and results are in very good agreement with the measured data with the best average relative errors of 0.0008% obtained for TBs1B versus load impedance Backpropagation ANN.
|
| Files |
|
