Gallium Nitride (GaN) technology is transforming RF monolithic microwave integrated circuits (MMICs) for power amplifiers, switches, low noise amplifiers and more. Vendors are now producing GaN MMICs in volume and achieving outstanding performance. GaN's characteristics enable power amplifier MMICs with 3x to 5x the output power of GaAs alternatives, or much smaller die sizes from L-band through W-band. High power switches with low insertion loss up through 18GHz have been developed. Low noise amplifiers have been demonstrated with noise figures equivalent to gallium arsenide, but with much higher input power survivability. The market for GaN RF MMICs spans commercial and military applications including base station, CATV infrastructure, communications, radar and electronic warfare, among others. Gallium Nitride based transistor technology's characteristics of very high current density combined with high voltage operation have held promise to vastly improve many microwave circuit applications that presently utilize Gallium Arsenide (GaAs) devices. Today, GaN transistors are capable of operation up to 50V and higher while simultaneously demonstrating FT & Fmax characteristics more typical of lower voltage GaAs PHEMT devices. The potential benefits of GaN device characteristics combined with MMIC technology are many. Highly efficient switched modes of power amplifier operation should be possible at higher output power levels and frequency. High output impedance typical of transistors operated at three to five times the voltage of GaAs should facilitate lower loss matching networks due to the reduced transformation ratio. Alternately, transistor periphery and corresponding output power could be dramatically increased while maintaining impedance transformation ratios similar to that of existing GaAs PHEMT amplifiers. The higher output power density of GaN devices should lead to greatly reduced die size for GaN implementations of existing power amplifier functions. The improved heat flow realized by the high thermal conductivity Silicon Carbide (SiC) substrate material should allow for acceptable junction temperatures even with the much higher power dissipation. Very high power switches could be designed by using large control voltages and taking advantage of the high current capability (high Imax) of GaN. While the advantages of GaN are manifest, many of the features that make GaN transistors attractive can be shown to create significant issues that are typically not encountered with lower voltage technologies. In this talk, specific examples and scenarios are discussed highlighting the benefits and issues associated GaN MMIC technology.

The growing demand for high-performance microwave circuits in modern communication systems requires the use of accurate and efficient simulation tools, enabling short manufacturing cycles and low production costs, through a reliable prediction and correction of the circuit behaviour at the design stage. Simulation has added difficulties in the case of nonlinear circuits, which do not obey the superposition principle and can exhibit self-sustained oscillations. There is a possible coexistence of steady-state solutions for the same values of the circuit elements and parameters, which may be either stable or unstable. Only stable solutions are robust against small perturbations and, therefore, can be physically observed. When the goal is to obtain an accurate solution in terms of waveforms and spectral content, iterative numerical methods are generally preferred. The time integration method provides the entire evolution of the circuit solution from the initial values to the steady state. Fast time-domain methods, such as shooting and finite-difference algorithms, perform the time-domain analysis of the steady state only, avoiding the transient through the use of additional constraints. The modelling of linear distributed elements with loss and dispersion requires the computation of discrete impulse-response samples, as well as convolution operations. In contrast, the harmonic balance method uses frequency-domain representations for the linear elements, maintaining the time domain descriptions for the nonlinear devices. It is solved through an error minimization method, sensitive to the initial value. By default, only the spectral components at the frequencies of the input generators are initialized, missing the possible self-generated oscillations. The solution obtained can be unstable and, therefore, qualitatively different from the solution measured. This situation is often faced in circuits such as power amplifiers, so the use of a complementary stability analysis is essential. In this talk, the principles and characteristics of the main simulation methods will be discussed, considering the solution stability properties, as well as the potentially complex dynamics of the nonlinear circuit when varying a relevant parameter.

Doherty Power Amplifier (DPA) is an old and fashionable configuration, introduced by W.E. Doherty in 1936 at the Bell Telephone Laboratories (BTL) of Whippany, New Jersey, for the realization of vacuum tube high-efficiency amplifiers, that has been successfully resurrected in the last decade, in agreement with the last sentence of the original paper. "The new amplifier is believed to offer a most logical and practical solution to the problem of efficient operation of high power transmitters", W.E. Doherty said and it has been demonstrated that it was a true prediction. Actually, the proposed solution has become today a hot research topic, representing the most suitable approach for implementing high efficiency transmitters, as the many published state-of-the-art results demonstrate. Even if actual systems are extremely different with respect to the first broadcasting ones, in terms of active device technologies, power levels and modulation schemes, the DPA seems, in fact, to remain the best candidate in order to realize Power Amplifiers (PA) for modern and future generations of wireless systems. In this contribution, the evolution of the DPA from its original idea is presented, with the aim to stress its pros and cons. Starting from its original topology, the intrinsic limitations are reviewed together with new proposed solutions to overcome them, both in terms of architecture features and frequency limitations. Moreover, the theoretical aspects and the developed design methodologies are supported through experimental results, from MMIC or HIC realizations, highlighting different approaches and behaviors.

Circuit-based large-signal device models are extensively used to design integrated and hybrid microwave circuits. Furthermore they can profitably be applied under the form of a nonlinear embedding device model to help with the synthesis of multi-transistor power amplifiers (PA) such as Doherty and Chireix amplifiers. The development of a circuit-based microwave device model is usually an intensive process. However with the advent of NVNAs, circuit-based nonlinear microwave models can nowadays be extracted for a targeted range of operation from a few large-signal measurements. This talk will presents the example of SOS-MOSFET and GaN models with memory effects extracted from real-time active load pull (RTALP). The efficient phase sweeping of the RTALP drastically reduces the number of large-signal measurements needed for the model development and verification while maintaining the same intrinsic voltage coverage as in conventional passive or active load–pull systems. The bias dependence of the charges and IV characteristics are simultaneously extracted from these measurements using artificial neural networks (ANN). Memory effects such as the parasitic bipolar junction transistor (BJT) in the SOS-MOSFET are accounted for with the use a physical circuit topology. The impact of self-heating and the cyclostationary charging of traps in GaN HEMTs under large signal RF operation will also be reviewed. The verification of the model can be performed using waveforms, load-lines, output power & power efficiency load–pull from additional independent RTALP measurements. The compagnion ANN nonlinear embedding device model will also be presented and the application of nonlinear embedding device models to the design of Doherty and Chireix power amplifiers will be reviewed.

Millimeter-wave active imagers rapidly gain widespread acceptance not just for defense, but also for civilian applications such as automotive, security, or occupational safety. Resulting systems can have both very high range and angular resolution with small from factor antennas. Further, the advent of high performance Si/SiGe BiCMOS technologies enable an unprecedented integration density at millimeter wave frequencies, and, due to maximum frequencies of oscillation approaching 500 GHz even in commonly available foundry technologies. The authors recently demonstrated an integrated FMCW sensor with millimeter range resolution in F-band, with a chirp bandwidth of up to 56 GHz and on-chip antennas, using the IHP SG13 130 nm SiGe BiCMOS offering. Since these ICs are commonly used in array configurations, it is practical to supply a coherent signal from a lower frequency common chirp frequency source, sidestepping the challenges of making board-to-chip interconnects beyond 100 GHz. This makes the frequency multiplier a key on-chip component. Since harmonics create phantom responses in FMCW radars, harmonic suppression is a key performance parameter. The authors investigated several frequency multiplier approaches, mostly for multiplication by eight, using both analog four quadrant multipliers (Gilbert cells) and push-push topologies. The importance of maintaining properly balanced differential signals will be highlighted, using either active or passive on-chip baluns. The presentation will conclude with recent results for FMCW imagers using both on-chip and off-chip antennas. The presence of standard CMOS components also allows mixed-signal capabilities to be co-integrated on chip. While this has not been done yet for the F-band circuits, it will be briefly discussed on the background of a recently demonstrated phased array core chip for Ka band.

Average power power tracking (APT) and envelope tracking (ET) power amplifiers require high efficiency across a range of supply voltages, which is a change from old fixed voltage systems. The performance of these amplifiers is usually characterized by a so-called "waterfall curve". While circuit design choices can certainly be a factor in this roll-off of PAE with decreasing supply voltage, it is also necessary to evaluate device technologies to understand the trade-offs and opportunities they present for these applications. For technology development, it is advantageous to see if "simpler" measurements, such as RF-Knee correlate to these waterfall measurements since they are easier to collect. In this paper, the considerations for device performance for these applications will be presented and discussed.