A scalable non-linear noise model for high-electron-mobility transistors

Noise is the unwanted random fluctuations in a signal, voltage or current. Over a frequency range from a few GHz to tens of GHz, the major noise contributors in the High-Electron-Mobility Transistor (HEMT) were thought to be partially correlated thermal drain-noise and gate-noise currents. However, the thermal origin of the drain-noise current cannot be explained with sub-micrometre HEMT theory. Furthermore, the measurements also reflect that both the noise figure and the optimum signal source admittance cannot be accurately defined using the conventional two noise sources. A simple but novel scalable non-linear noise model is presented in this thesis to predict the response and number of noise current sources in the HEMT in a frequency range from a few GHz to tens of GHz. The model uses probability theory, which is one of the best ways to predict the randomness associated with any quantity, for modelling noise in HEMTs, and describes how three noise sources are essential to completely define the noise performance of the HEMT over that frequency range. A comparison between the simulated and measured noise parameters is also carried out to test the validity of the model.