Search

A state-space empirical nonlinear model for GaN-based field-effect transistors (FETs) is defined, along with the associated identification procedures based on a recently published double pulse measurement technique. Charge trapping phenomena are dealt with in terms of a nonlinear state equation, which describes the rate of change of the trap state as a function of its actual distance from the corresponding steady state. Model experimental validation is carried out, after on-wafer characterization of a 1-mm AlGaN-GaN on SiC FET, both under strong and mild nonlinear operation.

The paper describes the design of a V band single balanced mixer to be employed in the front end of an on-board receiver for space applications. The V band receiver is a demonstrator for the use of different monolithic processes and interconnection/assembly technologies in space applications at such high frequencies. The receiver front-end consists of a multi-stage LNA amplification followed by an image reject filter, the mixer and a frequency doubler for the local oscillator. All the chips are mounted over an LTTC substrate using bumps/hot vias technology. Chip to chip interconnection is provided by coplanar wave guide on LTTC. The mixer employs a rat race 180° hybrid to balance the local oscillator and a couple of Schottky diodes as mixing elements. The technology employed is a 0.15μm pHMET process that offers diodes with a cut off frequency higher that 300GHz. The choices made for the mixer type and topology starting from the system specifications are covered in the paper along with the actual description of the circuit design.

A novel predistortion approach well suited for MMIC design is discussed. The theoretical analysis shows third-order intermodulation product (IMP3) cancellation independent of the specific technology exploited for implementation and of the detailed behaviour of the power stage nonlinearity; further improvements can be obtained through optimization. The active stages of the predistorter are modeled through a nonlinear describing function, which can be experimentally characterized through standard LS S-parameters measurements. The predistortion scheme has been implemented within the Agilent ADS environment and applied to the linearity optimization of a power amplifier stage. In accordance with theoretical results, significant improvement in the linearity of the Pin-Pout is demonstrated together with IMP3 reduction. A sensitivity analysis on the optimization parameters is also presented, showing that this approach is robust and well suited for circuit design.

This paper describes a method for the detection of instabilities and their stabilisation under large-signal regime in multi-device paralleled power amplifiers. The symmetry of the amplifier is exploited for a simple and effective characterisation of odd- and even-order modes, and their reduction to equivalent single-transistor amplifiers. The nonlinear stability is then assessed by use of the conversion matrix. Examples of application are also given.

A new method is proposed for the empirical characterization of the nonlinear thermal resistance in GaN HEMTs. Low-Frequency dispersion due to self-heating is used as the sensing parameter of the channel temperature changes deriving from dissipated power variations. Since GaN HEMTs are also affected by LF dispersive charge-trapping phenomena, the thermal resistance description is embedded into a full electrothermal model, in order to extract the parameters by bestfitting the measured data. The method involves only multi-bias small-signal S-parameter and dc I/V measurements at different base plate temperatures. It is non-invasive, does neither require special-purpose device geometries/structures nor measurements in the conduction region of the gate junction. Preliminary validation of the method is based on comparisons with measured electrothermal data, with data from the literature on similar devices and on results provided by a large-signal RF device model embedding the thermal resistance description.

We present a new netlist-compatible interface for entering and implementing nonlinear dynamic models into circuit simulators. The interface is completely general and allows to implement nonlinear models with nonlinear memory starting from model equations either in Volterra Series expansion and/or from state-space equations. A detailed description of the interface is included in the paper as well as two application examples using two different nonlinear non quasi-static models.

The use of Digital Predistortion (DPD) for linearizing a millimeter-wave Power Amplifier (PA) is investigated. A PA operating at 38 GHz is designed using an accurate nonquasi-static transistor model taking into account both short- and long-term memory effects. A realistic test signal is then used for the identification of a Nonlinear Auto-Regressive Moving Average (NARMA) behavioral model of the PA. The NARMA-based digital predistorter is then derived and formulated in terms of Basic Predistortion Cells, especially suitable for efficient implementation in an FPGA. The performance of the predistortion solution is preliminarily assessed by means of computer simulations.

Electron device modelling requires accurate descriptions of parasitic passive structures connecting the intrinsic electron device to the external world. In conventional approaches, the parasitic phenomena are described by a network of lumped elements. As an alternative, a distributed description can be conveniently adopted. This choice has been proved very appropriate when dealing with device scaling and very high operating frequencies. In this paper, a distributed parasitic network is adopted in association with a nonlinear electron device model. In particular, it is shown how an equivalent intrinsic device and a suitably-defined distributed parasitic network can be accurately defined and modelled on the basis of standard measurements and easy electromagnetic simulations. Wide experimental validation based on GaAs and InP PHEMTs will be provided, showing accurate prediction capabilities both under small- and large signal conditions. The proposed model is shown to perform optimally even after periphery scaling.

A technology-independent,non-quasi-staticnon-linear model of electron devicescapableofaccuratepredictions at microwave and millimetrewavesisproposedin this paper. The model is based on thedefinitionofaquasi-staticassociated device, which is controlled by means ofequivalent voltages. In particular, in the paper it isshownhow to define and experimentally identify suitable voltage-controlled voltage sources, which modify the original domainof applied voltages and create a suitable control environment for the purely-quasi-static associated device. The advantage of this approach is that conventional purely quasi-static models can still be adopted even at very high frequencies, if suitable equivalent voltages are applied. Preliminary experimental validation of the approach is provided in the paper by means of a GaAs PHEMT.

This paper deals with the design of X-band T/R modules to be employed on a future generation Synthetic Aperture Radar (SAR) for space application. Since the final High Power Amplifier (HPA) stage of the transmitting chain directly affects the overall module performances (mainly efficiency and maximum output power), HPA topic has been deeply investigated. In particular, an HBT process has been extensively explored, because of its attractive features; however potential advantages can be really achieved only if thermal management is consistently addressed. In SAR applications the RF carrier phase control is of great concern and major causes of module output phase shifts have been identified: temperature variations of active devices, non-linear modulating phenomena (AM/PM conversion) and supply voltage dropping off. The designed HPA will successfully handle all of them without disregard efficiency. Expected improvements will be shown for the future generation T/R module employing the designed PA and future architecture solutions.

Large-signal modelling of electron devices for nonlinear MMIC design is a fundamental topic for the microwave community. Many different non-linear modelling approaches have been proposed in the last years, and quite often circuit designers suffer from the lack of reliable comparison criteria to identify which model (between those available) could be the most suitable for the desired application. Moreover, similar strategies are needed even from the research groups, whose activity is devoted to the model identification and extraction, in order to quantify the degree of accuracy achievable by the modelling approach adopted. In this paper an approach to verify large-signal model accuracy will be discussed, which is simply based on the comparison between de-embedded measurements and model predictions of Y-parameters versus the bias voltages at the intrinsic device ports.

A technology-independent, behavioural non-linear dynamic model, which has been previously proposed by the authors for the accurate experimental characterization of digital acquisition channels, is considered in this paper for the compensation of the overall non-idealities within these measurement subsystems. It is shown how the non-linear dynamic equations of the model, which inherently describe both the static and the dynamic non-linearities affecting the channel, can be numerically inverted, allowing for the reconstruction of the spectrum at the input of the system from the corrupted samples available at its output. Different recursive algorithms for the solution of the non-linear problem involved into the procedure proposed are described. The choice among these can be accomplished according to the applicationdependent best trade-off between numerical convergence robustness and computational cost. Experimental results are provided, which validate the technique proposed.

A silicon self-aligned-emitter bipolar process from STMicroelectronics for very high efficiency handsets power applications has been used to build two Dielectric Resonator Oscillators. Despite this technology addresses the mobile telephony frequency range at 1.8GHz, the oscillators generate a stable reference at 6GHz and 7.5GHz with good phase noise performance. A low frequency noise model has been identified and implemented in a Gummel Poon BJT nonlinear model. A design technique to optimize stability and phase noise performances has been used. The DROs exhibit phase noise of -116dBc/Hz and -107dBc/Hz at 10KHz offset from the carrier at 6GHz and 7.5GHz, respectively.

In the paper is shown that,when devices are strongly affected by low-frequency dispersion effects, simply deriving the i/v characteristics through direct integration of bias-dependent differential parameters leads to results which are only locally self-consistent and coherent with empirical data.Dynamic drain current deviations due to charge trapping and self-heating phenomena in FETs, introducing a more complex dependence on average applied voltages and power in the dynamic i/v characteristics,must be taken into account by means of accurate approaches, such as the recently proposed equivalent voltage model [2 ]. Instead,integration of the low-frequency differentialparameters in a suitably modified equivalent voltage domain is shown to be consistent with device physics. Experimental evidence is provided in the paper.

In this paper some considerations on different oscillator topologies are presented with the aim of analyzing their behavior in terms of phase noise performance. In particular, our analysis suggests that a 9db improvement in the phase noise level can be obtained when using a “Push-Push” oscillator architecture in comparison with a fundamental-frequency oscillator developed under the same conditions. Simulations of different oscillator topologies, which confirm the proposed theory, are shown in the paper.

A new modeling approach for electron devices up to millimeter-wave frequencies is proposed. It exploits a previously proposed distributed approach with the aim of deriving a more computationally efficient model still preserving good accuracy. In particular, a new electron device model based on a “single active device” and a “distributed description of the parasitic structure” is derived, which can easily be identified through conventional scattering parameter measurements and electromagnetic simulations. Preliminary validation results are provided in the paper.

The paper introduces a simple and efficient approach for the modelling of low-frequency dispersive phenomena in FETs. It is based on the definition of a virtual, non-dispersive associated device controlled by equivalent port voltages and it is suitable for modelling based on standard nonlinear dynamic approaches, such as lumped-element equivalent circuits. The proposed approach is justified on the basis of a physically-consistent, charge-controlled description of the device, but the results are general and provide a valuable tool for taking into account dispersive effects in FETs by means of an intuitive circuit solution, in the framework of any existing nonlinear dynamic model of the associated non-dispersive device. The new equivalent-voltage description, identified on the basis of conventional measurements carried out under static and small-signal dynamic operating conditions, allows for the accurate prediction of dispersive effects above the frequency cut-off, but the formulation is still compatible, without for al modification, for the modelling of the device behaviour under signal excitations having spectral components in the dispersive low-frequency range. Preliminary results are presented which conferm the validity of the proposed approach.

Stability analysis of structures composed of transistors “paralleled” by means of power dividers/combiners is a keypoint in the design of MMIC power amplifiers. However, techniques to cope with this problem are quite rare in the literature and standard, straightforward tools are usually not available in CAD programs. A new simplified approach, hich can be applied using conventional linear tools for microwave circuit analysis, is proposed in this paper.

A new approach, which takes into account both traps and thermal phenomena is proposed for the accurate modeling of deviations between static and dynamic (e.g., pulsed) drain current characteristics in III-V FETs. The model, which is based on the well known "backgating" concept, can easily be identified on the basis of conventional static drain current characteristics and small-signal, low-frequency S parameters. In the paper, the model formulation is developed along with some hints on its implementation in commercially available CAD tools for HB simulations. Experimental validation, consisting of comparisons between simulated and measured pulsed drain current characteristics, is also provided.

A new approach to the approximation of nonlinear characteristics for look-up table based electron device models is proposed. It relys on the strong analogy between signal sampling and characteristic measurements over a discrete grid of bias points. On this basis, nonlinear function approximation can be carried out by using approaches similar to well-known techniques for reconstruction of continuous-time signals from a given sequence of samples. With respect to conventional interpolation techniques, the method proposed guarantees the accurate reproduction both of nonlinear characteristics and as­sociated derivatives without non physical slopes or oscillating behaviour. Preliminary simulated and experimental results confirming the effectiveness of this approach are presented.