A 0.15 μm GaN HEMT Large Signal Statistical Modeling and its Validation Along with High Efficiency Power Amplifier Design in L and S-band Application

In this paper, a comprehensive nonlinear large-signal statistical model for 0.15μm GaN HEMT is proposed, which includes trapping effects, nonlinear electrothermal effects, and empirical equivalent circuits. The model integrates trapping effects using an effective gate bias (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{V}_{\mathbf{g s}_{\text {eff }}}$</tex>) based on the slotted gate structure, providing a realistic representation of device performance under large-signal conditions. 0.15μm GaN HEMTs are measured, ensuring robust data diversity. An in-house parameter extraction program is used to extract all model parameters systematically. To build the statistical model, the methods of Principal Component Analysis (PCA), Factor Analysis, and Multiple Regression are combined. This approach captures process variations and ensures reliability in the predictions. The Agilent-ADS is used to implement the statistical model, and S-parameters simulations are performed to validate the model. Measured and simulated data at 35dBm of input power for output power (Pout) of 44.8/44.5dBm respectively, large signal Gain of 10/9.5dB respectively, and Drain Efficiency of 65/63% respectively, show good agreement, confirming the model's accuracy. Using the model, broadband GaN HEMT power amplifiers are developed, to demonstrate its practical utility. The simulated DOE's results for lumped and distributed components over 5% tolerance, while simulated and measurement CW characteristics results for Drain Efficiency of 44-76/40-75%, Gain of 9-12/8.7-11.5dB, and Pout of 41-46/40.5-45.2dBm, are found to be in good agreement. The results highlight the model's capability for accurate design and design of experiment estimation.