abstract: The drive towards device scaling and large output power in millimeter and sub-millimeter wave power amplifiers results in a highly non-linear, out-of-equilibrium charge transport regime. Particle-based Full Band Monte Carlo device simulators allow an accurate description of this carrier dynamics at the nanoscale. This work initially compares GaN high electron mobility transistors (HEMTs) based on the established Ga-face technology and the emerging N-face technology, through a modeling approach that allows a fair comparison, indicating that the N-face devices exhibit improved performance with respect to Ga-face ones due to the natural back-barrier confinement that mitigates short-channel-effects. An investigation is then carried out on the minimum aspect ratio (i.e. gate length to gate-to-channel-distance ratio) that limits short channel effects in ultra-scaled GaN and InP HEMTs, indicating that this value in GaN devices is 15 while in InP devices is 7.5. This difference is believed to be related to the different dielectric properties of the two materials, and the corresponding different electric field distributions. The dielectric effects of the passivation layer in millimeter-wave, high-power GaN HEMTs are also investigated, finding that the effective gate length is increased by fringing capacitances, enhanced by the dielectrics in regions adjacent to the gate for layers thicker than 5 nm, strongly affecting the frequency performance of deep sub-micron devices. Lastly, efficient Full Band Monte Carlo particle-based device simulations of the large-signal performance of mm-wave transistor power amplifiers with high-Q matching networks are reported for the first time. In particular, a CellularMonte Carlo (CMC) code is self-consistently coupled with a Harmonic Balance (HB) frequency domain circuit solver. Due to the iterative nature of the HB algorithm, this simulation approach is possible only due to the computational efficiency of the CMC, which uses pre-computed scattering tables. On the other hand, HB allows the direct simulation of the steady-state behavior of circuits with long transient time. This work provides an accurate and efficient tool for the device early-stage design, which allows a computerbased performance evaluation in lieu of the extremely time-consuming and expensive iterations of prototyping and experimental large-signal characterization. / Dissertation/Thesis / Ph.D. Electrical Engineering 2011

Kyoto University (京都大学) / 0048 / 新制・課程博士 / 博士(工学) / 甲第14049号 / 工博第2961号 / 新制||工||1439(附属図書館) / 26328 / UT51-2008-F441 / 京都大学大学院工学研究科電気工学専攻 / (主査)教授 和田 修己, 教授 引原 隆士, 准教授 久門 尚史, 教授 萩原 朋道 / 学位規則第4条第1項該当

Malgré les progrès faits dans les dernières décennies en CFD, les techniques RANS instationnaires pour les turbomachines multi-étages sont toujours très couteuses en temps de calcul, réduisant leur intérêt en conception industrielle. Grâce à une analyse de Fourier, les équations instationnaires de Navier-Stokes peuvent être considérées comme 2N+1 équations stationnaires couplées par un terme source. Cette approche calcule efficacement les écoulements instationnaires périodiques et montre de forts gains en terme de temps de calcul. Cependant, l’expression du terme source est algébrique, ce qui provoque des difficultés dans le calcul de l’inverse de la transformée de Fourier directe. Afin d’améliorer la robustesse et la précision de la méthode, une approche basée sur un échantillonnage temporel non-uniforme est adoptée. Pour réduire le domaine de calcul à un unique passage inter-aube des conditions aux limites de chrochronicité sont développées. Dans un premier temps, un rotor et un stator sont simules avec en une entrée du domaine de calcul une injection qui modélise le sillage de la roue directrice d’entrée. Ainsi, deux fréquences fondamentales sont vues par le rotor. L’influence du contenu fréquentiel dans le rotor (i.e. le nombre d’harmoniques de la fréquence de passage des roues et leurs combinaisons) est analysée. Les résultats sont valides contre ceux obtenus avec des simulations instationnaires classiques. Ensuite, la méthode est appliquée a deux compresseurs industriels, le compresseur transsonique ECL4 et le compresseur d’étude CREATE. / Despite the progress made in the last decades in CFD, the unsteady RANS techniques for multistage turbomachines are still very costly in computation time. Thanks to Fourier analysis, the unsteady Navier-Stokes equations can be seen as 2N + 1 equations coupled by a source term. This approach efficiently computes unsteady flows and showed significant savings in computation time. However, the source term of the method is algebraic, thus entailing some difficulties in the computation of the inverse of the direct Fourier transform. In order to enhance the robustness and the precision of the HBT, non-uniformly sampled time levels are chosen. To reduce the computational domain to a single blade passage, phaselag boundary conditions are developped. First, a rotor and a stator configuration with a wake injection at the inlet (accounting for an inlet guide vane) is modelled. Thus, two fundamental frequencies are seen by the rotor. The influence of the spectrum in the rotor is addressed. The results are validated against classical unsteady RANS simulations. Then the method is applied to two industrial : the transonic compressor ECL4 and the study compressor CREATE.

Universal Mobile Telecommunications Systems (UMTS) is the preferred third generation (3G) communication standard for mobile communications and will provide worldwide coverage, a convenient software technology and very high data rate. The high data rate, especially, requires the use of bandwidth-efficient modulation schemes such as Quadrature Phase Shift Keying (QPSK). But modulation schemes such as QPSK need, in turn, a very linear power from the output of the transmitter power amplifier in order to meet the spectral requirements. A linear power amplifier, traditionally, has very low energy efficiency. Poor energy efficiency directly affects operational costs and causes thermal heating issues in base station transmitters. Thus the power amplifier designer is forced to trade-off between linearity and efficiency. As a result of this trade-off a Class-AB power amplifier is most often used in QPSK based systems. Class-AB power amplifiers provide acceptable linearity at efficiency values around 45-50% typically. This compromise is not a satisfactory solution but is inevitable while using traditional power amplifier design techniques. This thesis details the use of a Class-F amplifier with carefully chosen bias points and harmonic traps to overcome this problem. Class-F amplifiers are usually considered as very high efficiency (80% or more power-added efficiency) amplifiers where the high efficiency is obtained through the use of harmonic traps (L-C filters or quarter-wavelength transmission lines), which provide suitable terminations (either open or short) for the harmonics generated. By doing this, a square wave drain voltage and a peaked half-sinusoidal drain current out-of-phase by 180 are produced. Since only a drain voltage or a drain current exists at any given time, the power dissipation is ideally zero resulting in 100% theoretical efficiency. These very high efficiency values are usually associated with poor linearity. However the linearity can be improved to meet the design standards but compromising on efficiency. Even after this is done, efficiencies are usually 10 to 15% greater than a traditional Class AB power amplifier with similar linearity performance. Thus efficiency can be improved without affecting linearity by the use of Class-F power amplifiers. In order to verify this theory, a Class-AB and a Class-F power amplifier are designed using Motorola's high voltage laterally diffused metal oxide semiconductor (LDMOS) transistor. The choice of bias points and the design of the harmonic traps are very critical for the Class-F performance and hence were designed after careful consideration. The designs were simulated on Agilent's Advanced Design System (ADS) and the simulated results were compared for three different power levels namely, the peak power, 3 dB below peak power and 6 dB below peak power. At all of these power levels it was noted that the Class-F and Class-AB power amplifiers have very similar linearity performance whereas the Class-F power amplifiers show about 10% improvement in efficiency in comparison to the Class-AB power amplifiers. / Master of Science

No description available.

No description available.

Trial and error and the use of highly time-consuming methods are often necessary for modeling, simulating and characterizing nonlinear dynamical systems. However, for the rather common special case when a nonlinear system has linear relations between many of its degrees of freedom there are particularly interesting opportunities for more efficient approaches. The aim of this thesis is to develop and validate new efficient methods for the theoretical and experimental study of mechanical systems that include significant zero-memory or hysteretic nonlinearities related to only small parts of the whole system. The basic idea is to take advantage of the fact that most of the system is linear and to use much of the linear theories behind forced response simulations. This is made possible by modeling the nonlinearities as external forces acting on the underlying linear system. The result is very fast simulation routines where the model is based on the residues and poles of the underlying linear system. These residues and poles can be obtained analytically, from finite element models or from experimental measurements, making these forced response routines very versatile. Using this approach, a complete nonlinear model contains both linear and nonlinear parts. Thus, it is also important to have robust and accurate methods for estimating both the linear and nonlinear system parameters from experimental data. The results of this work include robust and user-friendly routines based on sinusoidal and random noise excitation signals for characterization and description of nonlinearities from experimental measurements. These routines are used to create models of the studied systems. When combined with efficient simulation routines, complete tools are created which are both versatile and computationally inexpensive. The developed methods have been tested both by simulations and with experimental test rigs with promising results. This indicates that they are useful in practice and can provide a basis for future research and development of methods capable of handling more complex nonlinear systems.

Aero-engine assemblies are complex structures typically involving two or three nested rotors mounted within a flexible casing via squeeze-film damper (SFD) bearings. The deployment of SFDs into such structures is highly cost-effective but requires careful calculation since they can be highly nonlinear in their performance, particularly if they are unsupported (i.e. without a retainer spring). The direct study of whole-engine models with nonlinear bearings has been severely limited by the fact that current nonlinear computational techniques are not well-suited for complex large-order systems. The main contributions of this thesis are: • A procedure for unbalance response computation, suitable for generic whole-engine models with nonlinear bearings, which significantly extends the capability of current finite element packages. This comprises two novel nonlinear computational techniques: an implicit time domain integator referred to as the Impulsive Receptance Method (IRM) that enables rapid computation in the time domain; a whole-engine Receptance Harmonic Balance Method (RHBM) for rapid calculation of the periodic response in the frequency domain. Both methods use modal data calculated from a one-off analysis of the linear part of the engine at zero speed.• First-ever analyses on real twin-spool and three-spool engines. These studies illustrate the practical use of these solvers, provide an insight into the nonlinear dynamics of whole-engines and correlate with a limited amount of industrial experimental data. Both IRM and RHBM are directly formulated in terms of the relative response at the terminals of the nonlinear bearings. This makes them practically immune to the number of modes that need to be included, which runs into several hundreds for a typical engine. The two solvers are extensively tested on two/three-shaft engine models (with 5-6 SFDs) provided by a leading engine manufacturer using an SFD model that is used in industry. The tests show the IRM to be many times faster than an established robust conventional implicit integrator while achieving a similar level of accuracy. It is also shown to be more reliable than another popular implicit algorithm. The RHBM enables, for the first time, the frequency domain computation of the nonlinear response of whole-engine models. Its use is illustrated for both Single-Frequency Unbalance (SFU) excitation (unbalance confined to only one shaft) and Multi-Frequency Unbalance (MFU) excitation (unbalance located on two or more shafts, rotating at different speeds). Excellent correlation is demonstrated between RHBM and IRM.The parametric studies compare and contrast the frequency spectra for SFU and MFU cases. They also reveal the varying degree of lift at the unsupported SFDs. The sensitivity of the response to end-sealing and bearing housing alignment is also illustrated. It is demonstrated that the use of suitably preloaded vertically oriented “bump-springs” at the SFDs of heavy rotors produces a significant improvement in journal lift. It is also shown that the consideration of a slight amount of distributed damping in the structure significantly affects the predicted casing vibration levels, bringing them closer to measured levels, while having little effect on the SFD orbits.

MOR (Model Order Reduction) est devenu un domaine très répondu dans la recherche grâce à l'intérêt qu'il peut apporter dans la réduction des systèmes, ce qui permet d'économiser du temps, de la mémoire et le coût de CPU pour les outils de CAO. Ce domaine contient principalement deux branches: linéaires et non linéaires. MOR linéaire est un domaine mature avec des techniques numériques bien établie et bien connus dans la domaine de la recherche, par contre le domaine non linéaire reste vague, et jusqu'à présent il n'a pas montré des bons résultats dans la simulation des circuits électriques. La recherche est toujours en cours dans ce domaine, en raison de l’intérêt qu'il peut fournir aux simulateurs contemporains, surtout avec la croissance des puces électroniques en termes de taille et de complexité, et les exigences industrielles vers l'intégration des systèmes sur la même puce.Une contribution significative, pour résoudre le problème de Harmonic Balance (Equilibrage Harmonique) en utilisant la technique MOR, a été proposé en 2002 par E. Gad et M. Nakhla. La technique a montré une réduction substantielle de la dimension du système, tout en préservant, en sortie, la précision de l'analyse en régime permanent. Cette méthode de MOR utilise la technique de projection par l'intermédiaire de Krylov, et il préserve la passivité du système. Cependant, il souffre de quelques limitations importantes dans la construction de la matrice “pre-conditioner“ qui permettrait de réduire le système. La limitation principale est la nécessité d'une factorisation explicite comme une suite numérique de l'équation des dispositifs non linéaires . cette limitation rend la technique difficile à appliquer dans les conditions générales d'un simulateur. Cette thèse examinera les aspects non linéaires du modèle de réduction pour les équations de bilan harmoniques, et il étudiera les solutions pour surmonter les limitations mentionnées ci-dessus, en particulier en utilisant des approches de dérivateur numériques. / MOR recently became a well-known research field, due to the interest that it shows in reducing the system, which saves time, memory, and CPU cost for CAD tools. This field contains two branches, linear and nonlinear MOR, the linear MOR is a mature domain with well-established theory and numerical techniques. Meanwhile, nonlinear MOR domain is still stammering, and so far it didn’t show good and successful results in electrical circuit simulation. Some improvements however started to pop-up recently, and research is still going on this field because of the help that it can give to the contemporary simulators, especially with the growth of the electronic chips in terms of size and complexity due to industrial demands towards integrating systems on the same chip. A significant contribution in the MOR technique of HB solution has been proposed a decade ago by E. Gad and M. Nakhla. The technique has shown to provide a substantial system dimension reduction while preserving the precision of the output in steady state analysis. This MOR method uses the technique of projection via Krylov, and it preserves the passivity of the system. However, it suffers a number of important limitations in the construction of the pre-conditioner matrix which is ought to reduce the system. The main limitation is the necessity for explicit factorization as a power series of the equation of the nonlinear devices. This makes the technique difficult to apply in general purpose simulator conditions. This thesis will review the aspects of the nonlinear model order reduction technique for harmonic balance equations, and it will study solutions to overcome the above mentioned limitations, in particular using numerical differentiation approaches.

One of the major factors in RF circuit design is the ability to predict the performance of these circuits in the presence of uncertainty in the key design parameters. This is referred to as uncertainty quantification in the mathematical literature. Uncertainty about the key design parameters arises mainly from the difficulty of controlling the physical or geometrical features of the underlying design, especially at the nanometer level. With the constant trend to scale down the process feature size, uncertainty quantification becomes crucial in shortening the design time. This thesis presents a new approach to statistically characterize the variability of the Harmonic Balance analysis and its application to Intermodulation distortion analysis in the presence of uncertainty in the design parameters. The new approach is based on the concept of Polynomial Chaos (PC) and Stochastic Galerkin (SG) methods. However, unlike the traditional PC, the proposed approach adopts a new mathematical formulation that decouples the Polynomial Chaos problem into several problems whose sizes are equal to the size of the original Harmonic Balance problem. The proposed algorithm produces significant CPU savings with equivalent accuracy to traditional Monte Carlo and standard PC approaches.