With the rise in communication capacity, deep neural networks (DNN) for digital pre-distortion (DPD) to correct non-linearity in wideband power amplifiers (PAs) have become prominent. Yet, there is a void in open-source and measurement-setup-independent platforms for fast DPD exploration and objective DPD model comparison. This paper presents an open-source framework, OpenDPD, crafted in PyTorch, with an associated dataset for PA modeling and DPD learning. We introduce a Dense Gated Recurrent Unit (DGRU)-DPD, trained via a novel end-to-end learning architecture, outperforming previous DPD models on a digital PA (DPA) in the new digital transmitter (DTX) architecture with unconventional transfer characteristics compared to analog PAs. Measurements show our DGRU-DPD achieves an ACPR of -44.69/-44.47dBc and an EVM of -35.22dB for 200MHz OFDM signals. OpenDPD code, datasets and documentation are publicly available at https://github.com/lab-emi/OpenDPD

Digital predistortion (DPD) enhances signal quality in wideband radio frequency (RF) power amplifiers (PAs). As signal bandwidths expand in modern radio systems, DPD's energy consumption increasingly impacts overall system efficiency. Deep neural networks (DNNs) offer promising advancements in DPD, yet their high complexity hinders their practical deployment. This article introduces open-source mixed-precision (MP) neural networks that employ quantized low-precision fixed-point parameters for energy-efficient DPD. This approach reduces computational complexity and memory footprint, thereby lowering power consumption without compromising linearization efficacy. Applied to a 160-MHz-BW 1024-QAM OFDM signal from a digital RF PA, MP-DPD gives no performance loss against 32-bit floating-point precision DPDs, while achieving -43.75 (L)/-45.27 (R) dBc in the adjacent channel power ratio (ACPR) and -38.72 dB in error vector magnitude (EVM). A 16-bit fixed-point-precision MP-DPD enables a 2.8× reduction in estimated inference power. The DPD code in PyTorch is publicly available on GitHub.

This paper proposes a power amplifier (PA) correction technique to recover from load mismatch. It utilizes a main PA, two auxiliary PAs, and a coupler. By adjusting the input drive levels of the PAs it can recover the output power and to a great extent the efficiency of the main PA even when exposed to 2:1 VSWR mismatch conditions. When connected to 50O loading, only the main PA is active, for impedances below or above 50 O, besides the main amplifier, one of the auxiliary PAs is also activated. The power generated by the auxiliary PA adds in phase to the output power of the main PA, as such allowing the output power to be constant at the expense of a minor efficiency penalty.

This article presents an inverted Doherty power amplifier (IDPA) made load insensitive up to 2:1 voltage standing wave ratio (VSWR) across its fractional bandwidth with a very compact wideband impedance sensor embedded in its output power-combining network (OPCN). To correct for load variation, a low-loss tunable resonator (TR) is used to ensure ohmic loads to the main and peaking stages at the center frequency of operation. At off-center frequencies, TR is used to present an ohmic load for the main stage, while a digitally adjustable phase shifter is used to (re)align the main and peaking stage's current summation in the OPCN. For ohmic load deviations, the main and peaking stage supply voltages and input drives are adjusted to maintain the ideal Doherty's output power and efficiency profile related to nominal 50Ω loading across the bandwidth. To implement the control of the formerly mentioned technique, a wideband impedance sensor is proposed, which uses the orthogonality of incident and reflected waves and requires only four peak detectors. As proof of principle, a prototype 850-950-MHz IDPA featuring the proposed correction technique, the impedance sensor, and the control loop has been implemented as a printed circuit board (PCB) demonstrator. Measurement results show that the IDPA can maintain constant output power with a tolerance of only ± 0.2 dB while improving the drain efficiency and linearity across the entire fractional bandwidth (11% ) for a VSWR range of 2:1.

This article presents a low-loss correction technique for a self-healing load-insensitive power amplifier (PA) using a modified two-tap six-port network, wherein the varying (complex) load is first compensated for its unwanted susceptance part, followed by adjustment of the transistor output stage to the ohmic load variation, by modifying its supply voltage and drive level. This two-step approach avoids the high-Q conditions that occur in tunable matching network solutions, which aims to correct for both the real and imaginary load deviations, such as providing lower insertion loss and voltage stress. Next, to facilitate a fully automated load mismatch correction without the need for calibration, a modified two-tap six-port network for impedance detection and control loop approach is proposed. As proof of principle, a prototype 900-MHz class-AB PA featuring the proposed correction technique, as well as, the six-port reflectometer and the control loop, has been implemented as a PCB demonstrator. Measurement results show that the self-healing load-insensitive PA in the events of load mismatch significantly improves the performance and approaches the 50- $\Omega $ performance. On a 2:1 VSWR 360° mismatch trajectory and driven by a 64-QAM 3.86-MHz signal, the PA achieves a linear output power of 22 dBm with only ±0.1-dB variation and better than -45-dBc adjacent channel power ratio (ACPR).