abstract: Electric field imaging allows for a low cost, compact, non-invasive, non-ionizing alternative to other methods of imaging. It has many promising industrial applications including security, safely imaging power lines at construction sites, finding sources of electromagnetic interference, geo-prospecting, and medical imaging. The work presented in this dissertation concerns low frequency electric field imaging: the physics, hardware, and various methods of achieving it. Electric fields have historically been notoriously difficult to work with due to how intrinsically noisy the data is in electric field sensors. As a first contribution, an in-depth study demonstrates just how prevalent electric field noise is. In field tests, various cables were placed underneath power lines. Despite being shielded, the 60 Hz power line signal readily penetrated several types of cables. The challenges of high noise levels were largely addressed by connecting the output of an electric field sensor to a lock-in amplifier. Using the more accurate means of collecting electric field data, D-dot sensors were arrayed in a compact grid to resolve electric field images as a second contribution. This imager has successfully captured electric field images of live concealed wires and electromagnetic interference. An active method was developed as a third contribution. In this method, distortions created by objects when placed in a known electric field are read. This expands the domain of what can be imaged because the object does not need to be a time-varying electric field source. Images of dielectrics (e.g. bodies of water) and DC wires were captured using this new method. The final contribution uses a collection of one-dimensional electric field images, i.e. projections, to reconstruct a two-dimensional image. This was achieved using algorithms based in computed tomography such as filtered backprojection. An algebraic approach was also used to enforce sparsity regularization with the L1 norm, further improving the quality of some images. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2017

In this thesis, the applicability of Class-D amplifiers to integrated wide-band communication line driver applications is studied. While Class-D techniques can address some of the efficiency limitations of linear amplifier structures and have shown promising results in low frequency applications, the low frequency techniques and knowledge need further development in order to improve their practicality for wide band systems. New structures and techniques to extend the application of Class-D to wide-band communication systems, in particular the HomePlug AV wire- line communication standard, will be proposed. Additionally, the digital processing requirements of these wide-band systems drives rapid movement towards nanometer technology nodes and presents new challenges which will be addressed, and new opportunities which will be exploited, for wide-band integrated Class-D line drivers. There are three main contributions of this research. First, a model of Class-D efficiency degradation mechanisms is created, which allows the impact of high-level design choices such as supply voltage, process technology and operating frequency to be assessed. The outcome of this section is a strategy for pushing the high efficiency of Class-D to wide band communication applications, with switching frequencies up to many hundreds of Megahertz. A second part of this research considers the design of efficient, fast and high power Class-D output stages, as these are the major efficiency and bandwidth bottleneck in wide-band applications. A novel NMOS-only totem pole output stage with a fast, integrated drive structure will be proposed. In a third section, a complete wide-band Class-D line driver is designed in a 0.13μm digital CMOS process. The line driver is systematically designed using a rigorous development methodology and the aims are to maximise the achievable signal bandwidth while minimising power dissipation. Novel circuits and circuit structures are proposed as part of this section and the resulting fabricated Class-D line driver test chip shows an efficiency of 15% while driving a 30MHz wide signal with an MTPR of 22dB, at 33mW injected power.

O ruído de baixa frequência é um limitador de desempenho em circuitos analógicos, digitais e de radiofrequência, introduzindo ruído de fase em osciladores e reduzindo a estabilidade de células SRAM, por exemplo. Transistores de efeito de campo de metalóxido- semicondutor (MOSFETs) são conhecidos pelos elevados níveis de ruído 1= f e telegráfico, cuja potência pode ser ordens de magnitude maior do que a observada para ruído térmico para frequências de até dezenas de kHz. Além disso, com o avanço da tecnologia, a frequência de corner —isto é, a frequência na qual as contribuições dos ruídos térmico e shot superam a contribuição do ruído 1= f — aumenta, tornando os ruídos 1= f e telegráfico os mecanismos dominantes de ruído na tecnologia CMOS para frequências de até centenas de MHz. Mais ainda, o ruído de baixa frequência em transistores nanométricos pode variar significativamente de dispositivo para dispositivo, o que torna a variabilidade de ruído um aspecto importante para tecnologias MOS modernas. Para assegurar o projeto adequado de circuitos do ponto de vista de ruído, é necessário, portanto, identificar os mecanismos fundamentais responsáveis pelo ruído de baixa frequência em MOSFETs e desenvolver modelos capazes de considerar as dependências do ruído com geometria, polarização e temperatura. Neste trabalho é proposta uma técnica para análise de ruído de baixa frequência baseada na autocorrelação dos espectros de ruído em função de parâmetros como frequência, polarização e temperatura. A metodologia apresentada revela informações importantes sobre os mecanismos responsáveis pelo ruído 1= f que são difíceis de obter de outras formas. As análises de correlação realizadas em três tecnologias CMOS comerciais (140 nm, 65 nm e 45 nm) fornecem evidências contundentes de que o ruído de baixa frequência em transistores MOS tipo-n e tipo-p é composto por um somatório de sinais telegráficos termicamente ativados. / Low-frequency noise (LFN) is a performance limiter for analog, digital and RF circuits, introducing phase noise in oscillators and reducing the stability of SRAM cells, for example. Metal-oxide-semiconductor field-effect-transistors (MOSFETs) are known for their particularly high 1= f and random telegraph noise levels, whose power may be orders of magnitude larger than thermal noise for frequencies up to dozens of kHz. With the technology scaling, the corner frequency — i.e. the frequency at which the contributions of thermal and shot noises to noise power overshadow that of the 1= f noise — is increased, making 1= f and random telegraph signal (RTS) the dominant noise mechanism in CMOS technologies for frequencies up to several MHz. Additionally, the LFN levels from device-to-device can vary several orders of magnitude in deeply-scaled devices, making LFN variability a major concern in advanced MOS technologies. Therefore, to assure proper circuit design in this scenario, it is necessary to identify the fundamental mechanisms responsible for MOSFET LFN, in order to provide accurate LFN models that account not only for the average noise power, but also for its variability and dependences on geometry, bias and temperature. In this work, a new variability-based LFN analysis technique is introduced, employing the autocorrelation of multiple LFN spectra in terms of parameters such as frequency, bias and temperature. This technique reveals information about the mechanisms responsible for the 1= f noise that is difficult to obtain otherwise. The correlation analyses performed on three different commercial mixed-signal CMOS technologies (140-nm, 65-nm and 40-nm) provide strong evidence that the LFN of both n- and p-type MOS transistors is primarily composed of the superposition of thermally activated random telegraph signals (RTS).

Analisa-se no presente trabalho o problema das medidas de relaxação dielétrica a U.B.F. (Ultra Baixa Frequência). Propõe-se um modelo de ponte para tais medidas capaz de varrer o intervalo 10-3 Hz-10Hz, baseada na medida de fase conforme a discussão original de Van Turnhout e colaboradores. Analisa-se também os principais problemas experimentais com tal equipamento de U.B.F. e descreve-se a sua construção e funcionamento. Analisa-se também o modelo teórico de correlação entre medidas de U.B.F. e descarga de eletreto com correntes termo-estimulado. Em particular calculamos uma correção para a expressão integral dado por Van Turnhout e colaboradores que leva em conta a influência da energia de ativação na correlação entre as duas técnicas. Tal conversão é significante para valores da energia de ativação menor que 0,5eV e que ocorrem experimentalmente com frequência / In the present work we investigate the problem of U.L.F. (Ultra Low Frequency) dielectric relaxation. An experimental model is proposed for a bridge covering the range of 10-3Hz-10Hz, based on phase shift measurements originally proposed by Van Turhout and collaborators. We also analyze the main experimental problems with such U.L. F. measurements and describe its construction and performance. We have also investigated the theoretical correlation between U.L.F. dielectric relaxation and electret thermal stimulated currents. We calculate in particular, a correction for the integral expression given by Turnhout and collaborators that takes into account the value of the activation energy in the relation between the two techniques. The correction is important for values of the activation energy below 0,5eV, which occur frequently in dielectric relaxation processes

No presente trabalho descreve-se: a determinação da permitividade dielétrica complexa através de medidas de fase; a aplicação de técnicas de correlação e análise em séries de Fourier, para a determinação da fase, amplitude, qualidade e deformação do sinal atrasado; os circuitos eletrônicos do medidor de fase e do oscilador de U.B.F. programável; a programação desenvolvida em FORTRAN IV e MACRO-ASSEMBLER para o computador PDP 11/45, utilizado para operar o medidor de fase e o oscilador de U.B.F. / In the present work we describe: the determination of the complex dielectric permittivity by phase shift measurements; the application of correlation and Fourier series analyses to determine the phase shift, amplitude, quality and deformation of the shifted signal; the designed electronic hardware of the phase meter and programmb1e ultra-low frequency oscillator; the developed software written in FORTRAN IV and MACRO- ASSEMBLER for PDP 11/45, to operate the phase meter

Les travaux de thèse visent à optimiser la linéarité des HEMT GaN et intervenir dans la mise en oeuvre d’un banc load-pull multi-tons (MTLP). L’optimisation repose sur la simulation électrique d’un modèle du transistor et la quantification des effets des non-linéarités du modèle, les effets des parasites basses-fréquences soustraits au préalable. La mesure MTLP unifie potentiellement les caractérisations en linéarité traditionnelles. La réalisation de l’environnement de simulation permettant de simuler un modèle et la configuration MTLP accompagnent ces travaux. La modélisation considère un développement 8x0.25x75μm² et un point de repos (30V, 60mA). La caractéristique intrinsèque Id(Vgs, Vds) est identifiée de façon relativement rigoureuse et l’interprétation selon laquelle deux cadences d’émission (longue/rapide) et un seuil d’activation en Vds caractérisant les pièges longs est renforcée. Les modèles petit-signal et thermique sur Cgd sont directement extraits de paramètres [S] BF. L’émulation télécom est réalisé avec un signal à 8 raies pures irrégulièrement espacées satisfaisant aux conditions : PAPR=8.5 dB et loi gaussienne des composantes I,Q(t). Le choix des fréquences nous prémunit de l’interférence parmi IM3,5 (intermodulation d’ordre 3,5) et raies principales et l’acquisition (amplitude/phase) couvrent raies principales et d’IM3 de de relation fi + fj-fk et 2 fi-fj. Le critère de linéarité est celui du C/13 (taux d'intermodulation d'ordre3). Les signaux mono-/bi-porteuses sont parallèlement mis en oeuvre pour mesurer et comparer la linéarité de technologies GH25, GH50 et concurrente à 2 et 4 GHz. La nonlinéarité parasite est quantifiée sur la puissance disponible à l’entrée ( 1/2|a1|2 ) et l’analyse conjointe des caractéristiques temporelles d’enveloppe est proposée. / The PhD aims at optimizing the linearity of GaN HEMT in addition to the development and exploitation of a multi-tone load-pull bench (MTLP). The optimization is commonly based on electrical simulations implementing a model of the transistor and telecom signal and consists of identifying the influence of each non-linearity while annihilating low-frequency parasitics effects (models). The MTLP experiment has some capability to unify the classical linearity characterizations. The electrical simulation consistent with MTLP data is also suggested. The modelling focus on a 8x0.25x75μm² HEMT GaN sample and a (30V, 60mA) quiescent bias. The intrinsic Id(Vgs,Vds) characteristic is more accurately evaluated and more consistent look the interpretation that two types of emission time-scale (fast and long) plus a Vds trigger threshold (15V) characterizing the ‘long’ trap. The small-signal and the thermal model related to Cgd are straightforwardly extracted from low-frequency [S] parameters. A signal composed of unequally-spaced 8 tones has a PAPR=8.5 dB and Gaussian statistical rule of its I,Q(t) components (telecom properties). The frequency set implies no interferences among IM3,5 (3rd,5th order intermodulation ratio) and main tones. Power-wave acquisition cover the main tones and IM3 satisfying fi + fj-fk and 2 fi-fj. The C/I3 (3rd order carrier-to-intermodulation ratio) is the main linearity criterion. CW (Continuous Wave) and two-tone signals are also put into operation to survey and compare the linearity performances of GH25, GH50 and acompetitor. The parasitic non-linearity is quantified ( 1/2|a1|2 ) and the time-domain analysis give rise to interpretation about how trap effect may interact with linearity.

Low frequencies in small rooms are a problem that can affect the sound quality and therefore give an impaired listening experience if not treated. Even the slightest change of the subwoofer position can make a difference. Measurements have been made for two different subwoofer positions in a small control room, called the original and the alternative position. The original position represents the position that is normally used for the chosen control room. The alternative position was selected by measuring the position that had the flattest frequency response. The positions were tested with listening tests by 20 audio engineering students to see what differences existed between the two. The result shows that the original position was perceived to have more bass, which could be connected to the measurements. The subjects also showed that they could agree on the meaning of three of the given attributes used for the listening tests. If a flatter frequency response is wanted for the chosen control room, the alternative position is an option.

Two aluminum spherical mirrors with radii of 203.2 mm and radii of curvature also of 203.2 mm have been used to construct a tunable Fabry-Perót type resonator operational at frequencies as low as 500 MHz. The resonator has been incorporated into a pulsed nozzle, Fourier transform, Balle-Flygare spectrometer. The spectrometer is of use in recording low J transitions of large asymmetric molecules where the spectra are often greatly simplified compared to higher frequency regions. The resonators use is illustrated by recording the rotational spectra of bromobenzene and iodobenzene. In related experiments, using similar equipment, the pure rotational spectra of four isotopomers of SrS and all three naturally occurring isotopomers of the actinide-containing compound thorium monoxide have been recorded between 6 and 26 GHz. The data have been thoroughly analyzed to produce information pertaining to bond lengths and electronic structures.

Dans le domaine de l’électromagnétisme computationnel, les équations intégrales de frontière sont très largement utilisées pour résoudre certains des plus grands problèmes directs, grâce à leur grande efficacité. Cependant les équations intégrales du champ électrique et du champ combiné (EFIE et CFIE), deux des formulations les plus employées, souffrent d’instabilités à basse fréquence et à haute discrétisation, ce qui limite leur versatilité. Dans cette thèse différentes approches sont présentées pour obtenir des algorithmes applicables aussi bien à des problèmes de compatibilité électromagnétique qu’à des applications radar. Les solutions présentées incluent (i) l’extension des projecteurs dit quasi-Helmholtz (qH) aux modélisations d’ordre supérieur ; (ii) l’utilisation de ces projecteurs pour stabiliser l’équation intégrale du champ magnétique et former une CFIE extrêmement précise, augmentée par des techniques de type Calderón, qui ne souffre de problèmes ni à basse fréquence ni à haute discrétisation et qui n’est pas sujette aux résonances artificielles ; (iii) le développement d’une EFIE filaire, basée sur des B-splines linéaires et les projecteurs qH, stable aux deux extrémités du spectre. Ces travaux ont été suivis de l’ouverture d’un nouvel axe de recherche visant l’amélioration des techniques de résolution des problèmes inverses en électromagnétique, avec pour objectif principal l’augmentation des performances des interfaces cerveau machine (BCIs). Les premiers résultats obtenus incluent le développement de l’un des premiers systèmes libres de simulation de bout en bout de session de BCI ayant été publié après revue par les pairs. / In computational electromagnetics, boundary integral equations are the scheme of choice for solving extremely large forward electromagnetic problems due to their high efficiency. However, two of the most used of these formulations, the electric and combined field integral equations (EFIE and CFIE), suffer from stability issues at low frequency and dense discretization, limiting their applicability at both ends of the spectrum. This thesis focusses on remedying these issues to obtain full-wave solvers stable from low to high frequencies, capable of handling scenarios ranging from electromagnetic compatibility to radar applications. The solutions presented include (i) extending the quasi-Helmholtz (qH) projectors to higher order modeling thus combining stability with high order convergence rates; (ii) leveraging on the qH projectors to numerically stabilize the magnetic field integral equation and obtain a highly accurate and provably resonance-free Calderón-augmented CFIE immune to both of the aforementioned problems; and(iii) introducing a new low frequency and dense discretization stable wire EFIE based on projectors and linear B-splines. In addition, a research axis focused on enhancing Brain Computer Interface (BCIs) with high resolution electromagnetic modeling of the brain has been opened ; a particular attention is dedicated to the inverse problem of electromagnetics and the associated integral equation-based forward problem. The first results of this new line of investigations include the development of one of the first peer-reviewed, freely available framework for end-to-end simulation of BCI experiments.

付記する学位プログラム名: 霊長類学・ワイルドライフサイエンス・リーディング大学院 / 京都大学 / 0048 / 新制・課程博士 / 博士(理学) / 甲第22298号 / 理博第4612号 / 新制||理||1661(附属図書館) / 京都大学大学院理学研究科生物科学専攻 / (主査)教授 幸島 司郎, 教授 平田 聡, 教授 伊谷 原一 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM