Design and Analysis of Complex Neutralization Gain-Boosting Technique With Low-Loss Power Combining for Efficient, Linear D-Band Power Amplifiers

Abstract

This article introduces a comprehensive design and optimization approach aimed at significantly improving the power gain of a given device to achieve the theoretical maximum stable power gain, denoted as $4U$ (with U representing Mason’s Unilateral power gain), across a wide bandwidth. To evaluate the wideband gain enhancement of the device, a device-level Gain-Bandwidth Product (GBW) metric is presented. The proposed technique leverages a high-order embedding network, specifically complex neutralization, applied to a differential power device pair. The detailed optimization process is presented alongside theoretical modeling. To address the limited output power at the D-band, a highly efficient power-combining network is co-designed with the output-matching network of the power amplifier (PA). To validate the proposed methodology, a D-band three-stage PA with two-way power combining was implemented using the GlobalFoundries 45-nm SOI process. The amplifier occupies a compact active area of $0.116~\text {mm}^{2}$ . Small-signal measurements demonstrate a peak power gain of 21.7- and a 3-dB bandwidth (BW) of 15 GHz, covering the frequency range from 117 to 132 GHz. The enhanced power gain enables the PA drivers to operate efficiently and linearly in class-AB biasing mode at 127.5 GHz, delivering a saturated output power ( $P_{\text {sat}}$ ) of 11.9 dBm, output power at 1 dB compression point ( $\text {OP}_{1\,\text {dB}}$ ) of 11.85 dBm, and a peak power-added efficiency (PAE) of 15%. This allows the PA to achieve an average output power of 7.1 (5.9) dBm under 64-QAM (128-QAM) modulation with a data rate of 27 (16.8) Gb/s. The PA shows an average modulation efficiency of 6.9% (5.15%) with an rms error vector magnitude ( $\text {EVM}_{\text {rms}}$ ) better than −24.8 (−25.7) dB.