Inverse Design of Compact and Wideband Inverted Doherty Power Amplifiers Using Deep Learning

2026-06-25

AI Summary

This work addresses the longstanding challenge of simultaneously achieving compact size, wide bandwidth, and high efficiency in conventional Doherty power amplifiers. For the first time, a deep learning–driven inverse design methodology is applied to a broadband inverted Doherty architecture, integrating convolutional neural networks with genetic algorithms to co-design a pixelated output combining network that concurrently realizes load modulation, impedance matching, power combining, and phase compensation. Fabricated in GaN HEMT technology and augmented with digital predistortion, the prototype demonstrates exceptional performance across 1.9–2.5 GHz: peak drain efficiency ranges from 51% to 63%, with 6 dB back-off efficiency maintained at 48%–54%; output power remains stable at 44 ± 0.3 dBm, and adjacent channel leakage ratio (ACLR) exceeds −53.2 dBc, significantly advancing broadband high-efficiency amplifier performance.

Abstract

This paper presents a deep learning-assisted methodology for the inverse synthesis of a compact, wideband inverted Doherty power amplifier (PA). Convolutional neural networks (CNNs) and genetic algorithms (GAs) are jointly employed to generate pixelated Doherty combiner networks that integrate load modulation, impedance matching, power combining, and phase compensation into a single structure. As a proof of concept, we design and fabricate a GaN HEMT Doherty PA with a pixelated output combiner. The prototype achieves a measured peak drain efficiency of 51%-63% and a 6-dB back-off efficiency of 48%-54% over 1.9-2.5 GHz. Within the same frequency range, the measured output power is 44+/-0.3 dBm. Furthermore, with digital predistortion (DPD) applied, the prototype circuit demonstrates an adjacent channel leakage ratio (ACLR) better than -53.2 dBc.

Problem

• Inverse Design
• Compact
• Wideband
• Doherty Power Amplifier
• Deep Learning

Innovation

• inverse design
• deep learning
• inverted Doherty PA
• pixelated combiner
• wideband efficiency