Microwave passive component design is of particular interest to radio frequency (RF) scholars and engineers. Although a plethora of studies have been carried out over multiple decades, designing high-frequency structures that offer high performance still heavily relies on heuristic methods and even rules of thumb. Thus, the process is often inefficient, and outcomes are not guaranteed. This article proposes a novel cascaded convolutional neural network (CNN) model to speed up the design process of planar microwave passive components. Given target behavior specifications, our proposed neural network model can quickly and accurately suggest proper component structures for single or multiple frequency bands. The feasibility and reliability of our model are validated here by both electromagnetic (EM) simulation and a fully instrumented implementation. The experimental results demonstrate that the proposed model can design planar passive components, including two-port matching networks and three-port power dividers. Moreover, our model provides passive component topologies that are fundamentally different from canonical number-limited templates and, therefore, yields novel architectures for passive microwave components. It also facilitates rapid passive components design flow for targeted electrical behavior within a limited board area. The proposed cascaded CNN model and the associated methodologies in this article are generic and, thus, can be easily extended to the design of any symmetrical planar microwave passive components.

This paper presents a novel, tunable, frequency-division duplexing radio frequency (RF) front end that combines passive and active self-interference (SI) cancellation. An electrical-balance duplexer is used to passively cancel transmitter noise in the receive band, and an active canceller is employed to suppress SI in the transmit (Tx)-band. Subsystem specifications are developed, and a system-level analysis of noise and SI powers in this novel architecture is provided, thereby illustrating its operation. A proof-of-concept demonstrator, built from a software-defined radio and discrete RF components, has been characterized across a range of duplex configurations in the 700-950-MHz range and also at 1900 (Long-Term Evolution (LTE) band 3) and 2600 MHz (LTE band 7). The prototype achieves an impressive 6.0-7.4-dB noise figure in the presence of a +27-dBm LTE uplink Tx blocker for duplex separations of 47.5 MHz and above. The duplexer has also been tested against reference sensitivity test cases defined in the 3GPP LTE specification, demonstrating specification compliant sensitivity in LTE bands 28 (700 MHz), 3, and 7.

Microwave structure behavior prediction enables the estimation of circuit response over a frequency range, playing a crucial role in the design of radio frequency (RF) structures. Deep neural network (DNN) approaches have demonstrated their capability to simulate microwave structure behaviors. Nonetheless, the quality and utility of the model are constrained by the availability of data and computational capabilities. These inherent disadvantages hinder the extensive application of DNN in microwave structure behavior prediction. Transfer learning has recently been produced as a method offering improved accuracy and speed for predicting microwave circuit behavior. This paper proposes a novel transfer learning-based model to expedite the prediction process for a sequence of frequency samples. Through experimental validation, it is illustrated that the proposed methodology outperforms the conventional DNN techniques for microwave structure behavior prediction by effectively reducing the required data and shortening the training time. The proposed model also facilitates the fine-tuning of hyperparameters and reduces the simulator computing load.

Microwave structure behavior prediction is an important research topic in radio frequency (RF) design. In recent years, deep-learning-based techniques have been widely implemented to study microwaves, and they are envisaged to revolutionize this arduous and time-consuming work. However, empirical data collection and neural network training are two significant challenges of applying deep learning techniques to practical RF modeling and design problems. To this end, this letter investigates a transfer-learning-based approach to improve the accuracy and efficiency of predicting microwave structure behaviors. Through experimental comparisons, we validate that the proposed approach can reduce the amount of data required for training while shortening the neural network training time for the behavior prediction of microwave structures.

This article presents a design methodology for an asymmetrical Doherty Power Amplifier (DPA) which achieves a high average efficiency at back off across its operating bandwidth. It is shown that by combining continuous modes with post matching techniques, it is possible to achieve excellent efficiency performance whilst maintaining broadband operation. Analysis is provided on how the knee effect can reduce the effective potential efficiency of Class J and Continuous Inverse Class F modes. An optimum combination of $2\text{nd}$ and $3\text{rd}$ harmonic impedances is then proposed for the carrier PA which can minimise this knee effect impact on efficiency performance. Also, it is shown how the drain supply can be used to improve the bandwidth over which an intrinsic optimum load can be maintained. Based on this analysis, a simple iterative design procedure is then presented which can be directly implemented with standard RF design tools. This design procedure is then verified in the design and manufacture of a prototype DPA using the Wolfspeed CG2H40010F GaN HEMT. The realised PA operates between 2.1 and 3.2 GHz with a peak power output of between 43.9 and 44.5 dBm. The PA achieves a high average drain efficiency of 64.7 % at 8–9 dB of back off. The DPA has been tested with and without digital pre-distortion (DPD) by considering a 60 MHz LTE OFDM signal with 9 dB peak-to-average power ratio (PAPR). When DPD is enabled, the presented DPA achieves a drain efficiency of between 52.1–64.3 with an adjacent channel power ratio (ACPR) of between −42.2– −44.1 dB over the bandwidth of 2.1–3.2 GHz.

In this work, a digital predistortion (DPD) model for the linearization of RF power amplifiers (PAs) is presented. The model provides a linearized gain (DPD + PA) independent of the instantaneous transistor channel temperature within a predefined temperature window. Channel temperature variations due to varying ambient temperatures or changes in the signal probability density function (PDF) cause long-term memory effects, which results in dispersed (dynamic) AM/AM and AM/PM characteristics. The presented model is used to compensate for the memory effects due to self-heating and external temperature changes by estimating the transistor channel temperature through a linear single-pole Foster thermal network. The DPD model uses a first-order Taylor approximation to cancel out temperature-based nonlinearities. Gaussian pulses are used to extract the PA intrapulse gain at different temperatures without being affected by the signal PDF, thus allowing temperature- and signal-independent PA characterization. The model is validated from $20~^\circ \text{C}$ to $80~^\circ \text{C}$ and by considering a class-B 3.75-GHz 10-W gallium nitride (GaN)-on-SiC PA. The DPD performance is evaluated by considering the normalized root-mean-squared error (NRMSE), the output spectra, and adjacent channel power ratio (ACPR) with and without DPD for multiple signals bandwidths and peak-to-average power ratio (PAPR) and finally compared with other approaches.