RF Power Amplifier Research Articles

A Maximum Efficiency-86% Hybrid Power Modulator for 5G New Radio(NR) Applications

A hybrid power modulator is presented for the radio frequency (RF) power amplifiers of 5th generation mobile communication technology (5G) new radio (NR) applications. A hybrid power modulator utilizing a two-level switching converter and a broadband and high-efficiency linear amplifier is presented. A further improvement in the efficiency of the circuit is achieved by using an optimized supply voltage of the two-level switching converters of 4.5 V. In this way, the overall efficiency is improved by more than 5% compared to using a 5 V supply voltage. The linear amplifier consists of four stages. In order to improve bandwidth and circuit stability, a compensation circuit is added to the linear amplifier that eliminates the poles of the main amplifier by introducing additional zeros, indirectly pushing the pole distribution out of the bandwidth. Using this approach, the linear amplifier achieves a 3-dB bandwidth of 180 MHz and an efficiency of 51%. The hybrid power modulator achieves a maximum output power of 2.4 W and an efficiency of 86% when tracking a 100 MHz 5G-NR signal under a 4 Ω load in a 180 nm CMOS package.

Three-Level Hybrid Envelope Tracking Supply Modulator with High-Bandwidth Wide-Output-Swing

Envelope tracking is a dynamic supply modulation technique, which is mainly used to improve the efficiency of radio frequency power amplifier. For 5G NR mobile devices, this paper presents an envelope tracking supply modulator which is composed of a three-level switching converter and linear amplifier (LA) in parallel. The hysteresis control method is adopted in the supply modulator to improve the bandwidth of the switching converter and the loop response speed. At the same time, a rail-to-rail input-output class AB linear amplifier is designed to achieve a wide output swing and a super source follower is proposed to reduce the output impedance of LA. Designed with the 180 nm CMOS technology, the supply modulator operated at 5G NR 100 MHz envelope signal. The output power of the modulator reaches 1.5 W for a 6 Ω load resistor and the output swing is 4.3 V at 5.0 V supply, and the maximum efficiency reaches 85%.

Continual Learning Digital Predistortion of RF Power Amplifier for 6G AI-Empowered Wireless Communication

Artificial intelligence (AI) provides opportunities to enable high-efficiency wireless communication to dynamically adapt to the local environments and user demands. In this article, a continual learning digital predistortion (DPD) algorithm is proposed to linearize radio frequency (RF) power amplifier (PA) in 6G AI-empowered wireless communication. It is achieved by employing the ability of knowledge retention and knowledge transfer to merge the new PA operating state into the previous model, which makes the model more intelligent when dealing with multiple states. In addition, an augmented self-sensing model is proposed to accurately capture the common characteristics and key dynamic characteristics among different PA operating states to improve the knowledge transfer ability during the continual learning process. The experiments were carried out on a Doherty PA with dynamic configurations associated with power, bandwidth, and signal type. The experimental results have shown that the proposed technique can continuously learn when faced with unpredicted new PA operating states and so as to linearize all known states. The proposed DPD method performs with very low complexity in the long term, which is very suitable for the emerging AI-empowered dynamic scenario.

Digital Predistortion of RF Power Amplifiers With Decomposed Vector Rotation-Based Recurrent Neural Networks

In this article, we present a novel decomposed vector rotation (DVR)-based recurrent neural network behavioral model for digital predistortion (DPD) of radio frequency (RF) power amplifiers (PAs) in wideband scenarios. By representing memory terms of DVR with recurrent states and redesigning the piecewise modeling, we propose a novel recurrent DVR scheme. To ensure stable operation and enhanced modeling accuracy, we integrate the recurrent DVR into the gated learning mechanism of the modified Just Another NETwork (JANET). Experimental results confirm that the proposed DVR-JANET model provides much improved linearization performance with significantly reduced model complexity compared with the recent existing models.

A Uniform Neural Network Digital Predistortion Model of RF Power Amplifiers for Scalable Applications

In this article, a uniform neural network (NN) digital predistortion (DPD) model of radio frequency (RF) power amplifiers (PAs) is proposed for dynamic applications, which is suitable for RF PAs under various operating conditions without updating the coefficients. With the development of communication systems, it is difficult for the DPD to track the nonlinearity of the PA as the operating condition varies frequently. As one of the most promising achievements in recent years, the NN has shown excellent generalization ability, which is applicable to the DPD for scalable applications. In this situation, a uniform neural network model (UNNM), whose structure is a two-stage network, is proposed for scalable output power, scalable bandwidth, or simultaneous scalable power and bandwidth. The experiments are carried out on two sub-6 GHz broadband GaN Doherty PAs (DPAs). The experimental results show that the proposed model can achieve comparable performance without coefficient update in the scalable output power range of about 5 dB and the bandwidth range of 100 MHz, which outperforms the conventional fixed model with better than 3 dB power range and 40 MHz bandwidth range.

Analysis and fabrication of a small-scale radio-frequency balun for magnetic resonance imaging amplifier.

In this study, the authors report the design and fabrication of a small mixed-integrated balun for magnetic resonance imaging (MRI). The device was designed by using the positive anti-symmetric coupling method, which applies the lump surface-mount technology capacitors as well as mirror-symmetric coupling strips that were etched on the top and bottom layers of a printed circuit board. The capacitors reduced the length of the coupling strips and compensated for imbalances in the phase and gain due to errors in the fabrication process. The structure and equivalent even-odd circuit model of the device was modeled and examined using commercial software to optimize the design parameters. Following this, the device was fabricated and its performance was assessed through measurements using a network analyzer. The results showed that imbalances in the gain and phase were lower than 0.1 dB and 1°, respectively, and the insertion loss and the input voltage standing-wave ratio (VSWR) were lower than 0.4 dB and -25 dB, respectively. More importantly, the device was small, with dimensions of 50 × 60 × 1.5mm. This makes it suitable for MRI applications involving highly integrated miniaturized systems. The proposed device was integrated into a 3.0T radio-frequency power amplifier (RFPA) and reduced the dimensions of its power modules by 20% compared with the traditional balun. Finally, the RFPA module was used in an 3.0T MRI system for imaging experiments, and the results showed that the balun can help obtain high-quality scanning images.

A Manifold Regularization Approach for Low Sampling Rate Digital Predistortion With Band-Limited Feedback

Digital predistortion (DPD) is an effective linearization technique for RF power amplifiers (PAs), but conventional full sampling (FS) DPD systems use ADCs with three to five times signal bandwidth, and high-speed ADCs are expensive and power-hungry. In this article, we develop a novel band-limited DPD for reducing feedback sampling rate and acquisition bandwidth based on a general framework for semisupervised learning called manifold regularization (MR), which utilizes the geometry of unlabeled data to construct regularization terms for mitigating the overfitting problem. Considering the properties of DPD, we design a basis MR term and introduce it into the classical MR to obtain the extended MR (ExMR) method. To validate the proposed ExMR DPD method, experiments were conducted on two different RF PAs operating at 2.4 and 39 GHz, respectively. The test results demonstrate that the proposed ExMR DPD can linearize the RF PA with a 40-MHz acquisition bandwidth at 100-MHz input. The proposed method significantly reduces the cost and power consumption of the DPD system in comparison with the conventional FS DPD method, which provides a promising solution for broadband communication systems.

Distortion Analysis of RF Power Amplifier Using Probability Density of Input Signal and AM-AM Characteristics

When confirming the ACLR (adjacent channel leakage power ratio), which are representative indicators of distortion in the design of PA (power amplifier), it is well known how to calculate the AM-AM/PM characteristics of PA, input time series data of modulated signals, and analyze the output by Fourier analysis. In 5G (5th generation) mobile phones, not only QPSK (quadrature phase shift keying) modulation but also 16QAM (quadrature modulation), 64QAM, and 256QAM are becoming more multivalued as modulation signals. In addition, the modulation band may exceed 100MHz, and the amount of time series data increases, and the increase in calculation time becomes a problem. In order to shorten the calculation time, calculating the total amount of distortion generated by PA from the probability density of the modulation signal and the AM (amplitude modulation)-AM/PM (phase modulation) characteristics of PA is considered. For the AM-AM characteristics of PA, in this paper, IMD3 (inter modulation distortion 3) obtained from probability density and IMD3 by Fourier analysis, which are often used so long, are compared. As a result, it was confirmed that the result of probability density analysis is similar to that of Fourier analysis, when the nonlinearity is somewhat small. In addition, the agreement between the proposed method and the conventional method was confirmed with an error of about 2.0dB of ACLR using the modulation waves with a bandwidth of 5MHz, RB (resource block) being 25, and QPSK modulation.

A Low Running Complexity Model for Digital Predistortion of RF Power Amplifiers

In this letter, two basis function multiplexing-based behavioral modeling methods for digital predistortion (DPD) of RF power amplifiers (PAs) are proposed to reduce the running complexity of DPD. The proposed full basis-propagating selection (FBPS) model and reduced-complexity FBPS (RC-FBPS) model give two reasonable ways to multiplex even-order basis functions, extending the basis-propagating selection (BAPS) model which only uses basis function delay and odd-order basis functions. The experimental results confirm that both the proposed FBPS and RC-FBPS models can achieve a good tradeoff between running complexity and performance.

Segmented Spline Curve Neural Network for Low Latency Digital Predistortion of RF Power Amplifiers

At present, we are the dawn of a new era for wireless communication systems. The new dynamic approach to the operation of cellular networks requires an extension of the essential input–output hardware mechanisms, to include intelligence. Traditional neural network structures can be used to embed artificial intelligence into cellular network base stations; however, standard network structures are not an optimal solution for these use cases. In this article, we present a neural network structure, which is specifically designed to provide a more accurate and computationally efficient solution compared with the previous neural network solutions for predistortion of RF power amplifiers (PAs). The proposed network structure is directly compared with alternative neural network solutions, which have been successfully employed for digital predistortion (DPD). The operation of this network is validated for DPD with experimental measurements with wideband signals using the latest generation of commercially available RF hardware. The novel network structure proposed in this work is demonstrated, in practice, to have better performance for a normalized mean square error (NMSE), an adjacent channel leakage ratio (ACLR), and an error vector magnitude (EVM) compared with the most popular previously published neural network.

Neural-Network-Based Automated Synthesis of Transformer Matching Circuits for RF Amplifier Design

Rich experience and intuition play important roles in designing planar transformers (TFs) for contemporary radio frequency integrated circuits (RFICs). In general, RFIC designers have been heavily relying on multiple iterations of full electromagnetic (EM) simulations, which consumes much time and effort. Here, we propose an automated matching circuit synthesizer (AMCS) using neural networks (NNs). The proposed AMCS directly synthesizes a matching circuit combined with a TF throughout the entire design process, ranging from the desired performance to layout. In the AMCS, which is a “spec-to-layout” synthesizer, one NN returns physical parameters of matching circuits, and another NN estimates the electrical performance in two-port <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$S$ </tex-math></inline-formula> -parameters from the desired impedances. Before the NNs are trained, input feature design is conducted to avoid the one-to-many problem, which cannot be well characterized with an inverse NN. This significantly reduces the time and effort for iterative circuit and EM simulations. The AMCS generated the matching circuit layouts for simple single-stage amplifiers operating at different frequencies up to 70 GHz or in different bandwidths of up to 32.5%. The estimated <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$S$ </tex-math></inline-formula> -parameters of the amplifiers show good agreement with the EM simulation results.

Sub-25 ps timing measurements with 10 × 10 cm[formula omitted] PICOSEC Micromegas detectors

The PICOSEC Micromegas detector is a precise timing gaseous detector based on a Cherenkov radiator coupled to a semi-transparent photocathode and a Micromegas amplifying structure. First single-pad prototypes demonstrated a time resolution below σ= 25 ps, however, to make the concept appropriate to physics applications, several developments are required. The objective of this work was to achieve an equivalent time resolution for a 10 × 10 cm2 area PICOSEC Micromegas detector. The prototype was designed, produced and tested in the laboratory and successfully operated with a 80 GeV/c muon beam. Preliminary results for this device equipped with a CsI photocathode demonstrated a time resolution below σ= 25 ps for all measured pads. The time resolution was reduced to be below σ= 18 ps by decreasing the drift gap to 180 μm and using dedicated RF amplifier cards as new electronics. The excellent timing performance of the single-channel proof of concept was not only transferred to the 100-channel prototype, but even improved, making the PICOSEC Micromegas detector more suitable for large-area experiments in need of detectors with high time resolutions.

Design of radio frequency power amplifier for 2.45 GHz IoT application using 0.18 µm CMOS technology

PurposeThe purpose of this study is to show that due to the emergence of the Internet of Things (IoT) industry in recent years, the demand for the higher integration of wireless communication systems with a higher data rate of transmission capacity and lower power consumption has increased tremendously. The radio frequency power amplifier (PA) design is getting more challenging and crucial. A PA for a 2.45 GHz IoT application using 0.18 µm complementary metal oxide semiconductor (CMOS) technology is presented in this paper.Design/methodology/approachThe design consists of two stages, the driver and output stage, where both use a single-stage common source transistor configuration. In view of performance, the PA can deliver more than 20 dB gain from 2.4 GHz to 2.5 GHz.FindingsThe maximum output power achieved by PA is 13.28 dBm. As the PA design is targeted for Bluetooth low energy (BLE) transmitter use, a minimum of 10 dBm output power should be achieved by PA to transmit the signal in BLE standard. The PA exhibits a constant output third-order interception point of 18 dBm before PA becomes saturated after 10 dBm output power. The PA shows a peak power added efficiency of 17.82% at the 13.24 dBm output power.Originality/valueThe PA design exhibits good linearity up to 10 dBm out the PA design exhibits good linearity up to 10 dBm output power without sacrificing efficiency. At the operating frequency of 2.45 GHz, the PA exhibits a stability k-factor, the value of more than 1; thus, the PA design is considered unconditional stable. Besides, the PA shows the s-parameters performance of –7.91 dB for S11, –11.07 dB for S22 and 21.5 dB for S21.

3×3 MIMO 60-GHz OFDM-RoF System With Long-Distance Wireless Transmission Enabled by MIMO Volterra Filtering

A 3×3 60-GHz multiple-input multiple-output (MIMO) radio-over-fiber system with an MIMO Volterra filter was demonstrated. The main objective of the proposed scheme is to compensate for the nonlinear distortion caused by radio frequency (RF) power amplifiers (PAs) before the signals reaches the transmitter antennas—particularly when the PAs operate with a high input power to achieve a high output RF power for long-distance wireless transmission. When the required wireless transmission distance is only 7 m, the 3×3 MIMO system with careful control of the PA input power can achieve a data rate of 126 Gbps. However, the data rate decreases rapidly over a longer wireless distance because of the larger loss. By eliminating nonlinear distortion and interchannel interference simultaneously, MIMO Volterra filtering can significantly increase the achievable data rate of a long-distance system. When the required data rate is 90 Gbps (or 100 Gbps), the maximum wireless transmission distance is 118 m (or 70 m) with the aid of MIMO Volterra filtering, but it is only 74 m (or 40 m) without MIMO Volterra filtering.

Analytical Evaluation of Power-Amplifier-Based Charging Methodology and Energy Efficiency Optimization Framework for Aerial Base Stations

In this article, a radio-frequency (RF) power-amplifier (PA) system configuration for the on-board energy conversion is proposed for unmanned aerial vehicle base station’s (UAV-BS’s) battery charging. First, the PA system utilizes the internal RF source for energy conversion so that perpetual battery charging is possible without external power sources. Therefore, a continuous battery charging of the UAV-BS is possible while providing wireless network coverage by hovering over the target location. In addition, the proposed PA system offers a higher RF power control resolution than the existing PA works, which is desirable for precise power control of the UAV-BS. The objective of simultaneous RF power control and battery charging by the PA system is achieved by activating one Wilkinson power divider (WPD) with a specific power splitting ratio in the power divider bank. The simulations show that the dynamic activation of ten distinct unequal split WPDs of the proposed PA system improves the RF power control resolution by $157$ and $396{\,}\%$ compared to the existing PAs. Additionally, the simulations show that the PA’s power added efficiency fluctuation is reduced by $54$ and $82{\,}\%$ compared to the existing PAs. In addition, it will be shown that the proposed on-board battery charging increases the hovering time by $7.4{\,}\%$. Furthermore, an energy-efficiency optimization framework for UAV-BS’s is proposed considering real-world communication system’s hardware imperfections, and the proposed result suggests that UAV-BS’s flight time may decrease by $50{\,}\%$ depending upon the type of UAV-BS RF components.

Gain-enabled optical delay readout unit using CMOS-compatible avalanche photodetectors

A compact time delay unit is fundamental to integrated photonic circuits with applications in, for example, optical beam-forming networks, photonic equalization, and finite and infinite impulse response optical filtering. In this paper, we report a novel gain-enabled delay readout system using a tunable optical carrier, low-frequency RF signal and CMOS-compatible photodetectors, suitable for silicon photonic integration. The characterization method relies on direct phase measurement of an input RF signal and thereafter extraction of the delay profile. Both integrated silicon and germanium photodetectors coupled with low-bandwidth electronics are used to characterize a microring resonator-based, true-time delay unit under distinct ring–bus coupling formats. The detectors, used in both linear and avalanche mode, are shown to be successful as optical-to-electrical converters and RF amplifiers without introducing significant phase distortion. For a Si–Ge separate-absorption-charge-multiplication avalanche detector, an RF amplification of 10 dB is observed relative to a Ge PIN linear detector. An all-silicon defect-mediated avalanche photodetector is shown to have a 3 dB RF amplification compared to the same PIN detector. All ring delay measurement results are validated by full-wave simulation. Additionally, the impact of photodetector biasing and system linearity is analyzed.

Development of a wireless ultrasonic brain stimulation system for concurrent bilateral neuromodulation in freely moving rodents.

Bilateral brain stimulation is an important modality used to investigate brain circuits and treat neurological conditions. Recently, low-intensity pulsed ultrasound (LIPUS) received significant attention as a novel non-invasive neurostimulation technique with high spatial specificity. Despite the growing interest, the typical ultrasound brain stimulation study, especially for small animals, is limited to a single target of sonication. The constraint is associated with the complexity and the cost of the hardware system required to achieve multi-regional sonication. This work presented the development of a low-cost LIPUS system with a pair of single-element ultrasound transducers to address the above problem. The system was built with a multicore processor with an RF amplifier circuit. In addition, LIPUS device was incorporated with a wireless module (bluetooth low energy) and powered by a single 3.7 V battery. As a result, we achieved an ultrasound transmission with a central frequency of 380 kHz and a peak-to-peak pressure of 480 kPa from each ultrasound transducer. The developed system was further applied to anesthetized rats to investigate the difference between uni- and bilateral stimulation. A significant difference in cortical power density extracted from electroencephalogram signals was observed between uni- and bilateral LIPUS stimulation. The developed device provides an affordable solution to investigate the effects of LIPUS on functional interhemispheric connection.

Data Detection With CFO Uncertainty and Nonlinearity for mmWave MIMO-OFDM Systems

The nonlinear distortions attributed by radio frequency power amplifiers (PAs) are inevitable in millimeter-wave (mmWave) systems due to the high frequency and large bandwidth. This nonlinearity induces multiplicative distortion and intercarrier interference in mmWave multiple-input-multiple-output (MIMO) orthogonal frequency-division multiplexing systems, which in collusion with a frequency-selective channel and a carrier frequency offset (CFO) brings great challenges to signal detection. Furthermore, the nonlinearity destroys orthogonality of the training pilots and degrades the estimation accuracy of the channel gains and the CFO. Also, the target posterior distribution for data detection becomes analytically intractable due to the nonlinearity. In this article, we present an importance-sampling-based framework for estimating the CFO in the presence of PA impairment, which utilizes few pilots and initializes the channel estimator. We propose a particle-filter-based iterative algorithm for data detection, where the intractable posterior distribution is approximated by weighted random measures. Furthermore, a sequential-maximum-likelihood-based semiblind channel estimator in the presence of nonlinearity is proposed. Performances of the estimator and the detector are evaluated by means of normalized mean square error, mean square error, and bit error rate. To validate the efficacy of the channel and the CFO estimator, the Cramer–Rao bound (CRB) and the weighted Bayesian CRB are derived, respectively.

Residual RNN Models With Pruning for Digital Predistortion of RF Power Amplifiers

The paper presents several novel residual recurrent neural network (RNN) models for digital predistortion of radio-frequency power amplifiers. Then, unstructured and structured neural network (NN) pruning algorithms are proposed to reduce the per-sample computational complexity and the required memory, without compromising their modeling performance. The mathematical equations of the complexity and the memory are derived to analyze the hardware cost and design low cost models. An experimental platform is developed to measure the modeling performance in terms of adjacent channel leakage ratio (ACLR), normalized mean square error, error vector magnitude (EVM) and bit error rate. The platform uses an envelope tracking PA excited by 80 MHz IEEE 802.11ac signals. Compared with the state-of-the-art RNN model in (Sun <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">et al.</i> , 2019), the proposed models reduce the per-sample complexity and required memory by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$94\hbox{--}99$</tex-math></inline-formula> %, they also improve the ACLR and EVM by at least 3 dB and 1 dB at 0.45 dB and 3.15 dB gain compression, respectively. Furthermore, an observation path with time alignment is used to remove the transmit latency in the state-of-the-art approach. At 3.15 dB gain compression, the proposed models outperform the state-of-the-art feedforward and convolutional NN models by at least 1.3 dB in EVM, outperform the general memory polynomial model by at least 3.3 dB in EVM. The above evidences indicate that the proposed RNN models can be implemented with relatively low hardware cost and achieve the best EVM in high power efficiency scenario.

Comparative criteria for objective evaluation of the efficiency of RF power amplifiers

Objective comparison of radio frequency amplifiers, which are infinitely diverse in their parameters and purpose is a difficult engineering task. For its successful solution it is necessary to define objective criteria that allow to quantify qualitatively and quantitatively the scheme and mode of operation of the amplifier, without this assessment depending on the absolute values of current, voltage, power and parameters of active and passive elements in the electronic circuit.