Deep Learning-Driven Inverse Design of Doherty Power Amplifiers Using Pixelated Combiners and Dual-State Impedance Synthesis

2026-06-16

AI Summary

This work addresses the longstanding challenge in conventional Doherty power amplifiers, where the output combiner struggles to simultaneously achieve effective load modulation, impedance matching, and phase compensation, leading to complex design procedures. To overcome this, the study introduces a deep learning–driven inverse design framework that integrates a deep convolutional neural network, a pixelated electromagnetic structure representation, a genetic algorithm, and a dual-state impedance synthesis method to enable co-optimization of a three-port Doherty combiner tailored for both peak and 6 dB back-off operating conditions. Prototypes fabricated in GaN HEMT technology demonstrate, across the 2.6–2.8 GHz band, saturated output power exceeding 44.2 dBm, peak drain efficiency above 71.2%, and 64% efficiency at 6 dB back-off; after digital predistortion, adjacent channel leakage ratio improves to better than −51.3 dBc.

Abstract

The output combiner of a Doherty power amplifier (PA) integrates load modulation, impedance matching, and phase compensation within a single network, making its design and synthesis highly challenging. In this paper, we propose a three-port Doherty combiner design methodology that combines deep convolutional neural networks (CNNs), pixelated layout representations, and genetic algorithms (GA) with dual-state impedance synthesis to address both peak and back-off power conditions. As a proof of concept, two GaN HEMT Doherty PA prototypes incorporating three-port pixelated combiners are designed and fabricated. Both prototypes achieve a measured saturated output power exceeding 44.2 dBm with peak drain efficiency above 71.2% within 2.6-2.8 GHz. Furthermore, a drain efficiency as high as 64% is measured at the 6-dB back-off level. After applying digital predistortion, each prototype achieves an adjacent channel leakage ratio (ACLR) better than -51.3 dBc.

Problem

• Doherty power amplifier
• output combiner
• impedance matching
• load modulation
• phase compensation

Innovation

• Deep learning
• Doherty power amplifier
• Pixelated combiner
• Dual-state impedance synthesis
• Inverse design