O objetivo desta dissertação é propor o projeto de amplificadores de baixo rudo (LNAs) do tipo banda estreita e banda larga em tecnologia CMOS. O projeto de LNAs de banda estreita é representado através de um método de otimização conhecido como programação geométrica. Também, neste trabalho foi projetada uma topologia para LNAs de banda larga, aplicando a programação geométrica durante a fase inicial de projeto. Os layouts de ambos os circuitos foram desenhados e fabricados usando três processos CMOS diferentes. O aumento da utilização de circuitos digitais está reduzindo e substituindo a quantidade de circuitos analógicos implementados nos sistemas atuais. Nos transceptores de radiofrequência, a maior parte dos circuitos foi substituída por circuitos digitais equivalentes. A razão para esta substituição é devido a sua escalabilidade, variações PVT (Process, Voltage and Temperature) baixas, e menor tempo de projeto, resultado de um fluxo altamente automatizado. A redução do tempo de projeto representa um time-to-market menor e custos mais baixos. No entanto, o amplificador de baixo rudo é um dos blocos de radiofrequência que permanecem principalmente no domínio analógico, tornando a redução do tempo de projeto mediante a otimização do fluxo analógico como um bom foco de estudo. O LNA deve ser capaz de receber um sinal de baixa potência e alta frequência, e amplificá-lo adicionando o menor rudo possível, mantendo o casamento de impedâncias, baixo consumo de potência, e uma linearidade adequada a fim de evitar a distorção. Nesta dissertação, a maioria das especificações de desempenho citadas são formuladas rigorosamente e descritas como um programa geométrico. Além disso, vários scripts são escritos de forma a automatizar o fluxo de projeto. A programação geométrica é considerada como uma boa opção porque se o problema de otimização tem solução, o resultado é o ponto de otimização global, e pode ser atingido rapidamente (na ordem de segundos). Para um LNA fonte comum de banda estreita, o problema de projeto é completamente formulado como um programa geométrico, e alguns parâmetros normalmente desprezados, como as não idealidades dos indutores CMOS e a capacitância portadreno do transistor MOS são considerados no projeto. O problema de otimização é resolvido em minutos e testado em cinco processos CMOS diferentes, e para diferentes frequências de operação entre 1,5 GHz e 5 GHz. Os resultados são comparados e validados através de simulações, e dois layouts de LNAs para 2,45 GHz foram desenhados, fabricados e testados usando dois processos de 0,18 mm diferentes. Neste trabalho, também foi formulado um LNA de banda larga com cancelamento de rudo, e um bloco LNA-Misturador de banda larga é projetado incluindo a programação geométrica no cálculo da impedância de entrada e o cancelamento de rudo. Os layouts de dois protótipos diferentes do bloco LNA-Misturador de banda larga, operando na faixa de frequência entre 1 GHz e 5 GHz, foram desenhados e fabricados usando um processo de 0,18 mm. / This dissertation proposes the design of CMOS narrowband and wideband low noise amplifiers. The design problem of narrowband LNAs is represented as an optimization problem known as geometric programming. Furthermore, a topology for wideband LNAs is designed including the geometric programming in an early stage of the design. Both type of circuits were layouted and fabricated using three different CMOS processes. The tendency to increase the number of applications for digital-intensive circuitry, is reducing and replacing the amount of analog circuits implemented on systems nowadays. In radiofrequency transceivers, most of the circuits have been replaced by a digital-intensive counterpart. Digital circuitry is preferred over the analog one due to its scalability, low PVT (Process, Voltage and Temperature) variations, and shorter designing time result of a highly automated flow. The reduction of the designing time represents a faster time-to-market and lower costs. However, the low noise amplifier is one of the radiofrequency blocks that remain mainly in the analog domain, thus reducing its designing time by optimizing an analog design flow become a good focus of study. The LNA should be capable of receiving a low power and high frequency signal and amplify it adding the minimum noise possible, while maintaining good impedance matching, low power consumption and an adequate linearity in order to avoid distortion. In this dissertation, most of the performance parameters aforementioned are formulated rigorously and described as a geometric program. Moreover, various scripts are written in order to automate the design flow. The geometric programming is considered a good option because if the optimization problem is feasible, the result is the global optimum and can be obtained in seconds. For a common source narrowband LNA, the design problem is fully formulated as a geometric program and some parameters commonly neglected, as the CMOS inductors non-idealities and the gate-drain capacitance of MOS transistor are considered. The optimization problem is solved in minutes and tested on five different CMOS processes at different operating frequencies between 1.5 GHz and 5 GHz. The results are compared and validated through simulations, and two layouts for 2.45 GHz LNAs are drawn, fabricated and tested using two different 0.18 mm processes. In addition, a noise canceling wideband LNA is formulated, and a wideband LNA-Mixer cell is designed by including the geometric programming to estimate the input impedance matching and assure the noise cancelation. The layouts of two different prototypes of the wideband LNA-Mixer cells for the 1 GHz-5 GHz frequency band are drawn and fabricated using a 0.18 mm process.

Das impedanzwandelnde Hören ist eine wichtige Verfeinerung des Gehörsystems der Tetrapoden indem es einen Energieverlust während der Schallübertragung vermeidet. Anatomisch ist es durch eine Unterteilung des Foramen Metoticum in eine vordere Fenestra Pseudorotunda und ein hinteres Vagus Foramen charakterisiert. Dieses System trat mehrmals unabhängig in Amniota auf, und obwohl schon vorher vermutet wurde, dass es eine Homoplasie für Archosauria darstellt, wurde diese Hypothese bisher noch nicht geprüft. Demnach wurden 17 Hirnschädel Merkmale für 111 Taxa kodiert und auf einem informellen Supertrees optimiert. Die Analyse ergab, dass die Fenestra Pseudorotunda unbhängig acht Mal in Archosauria auftrat, mit fünf Umkehrungen. Während dieses Merkmal plastisch für Dinosauria ist, es trat nur einmal in Pseudosuchia auf. Eine Tree Shape-Analyse ergab, dass sechs Verschiebungen in den Diversifizierungsraten in Pseudosuchia und sieben in Dinosauria zu finden sind, von denen nur die der Ornithischia mit der Fenestra Pseudorotunda in Beziehung stehen. Viele Zustandsveränderungen erfolgen an der Basis der Dinosauria aber bei Pseudosuchia sind diese vor der Entstehung der Crocodyliformes und an der Basis der Notosuchia und Thalattosuchia konzentriert. Die Korrelationen zwischen Zustandsveränderungen und Verschiebungen sind höher in Dinosauria, während bei Pseudosuchia nur Mesoeucrocodylia eine ähnliche Anzahl aufweist. Daraus folgt, dass das impedanzwandelnde Hören nicht homolog in Archosauria ist, und dass es nicht als Schlüsselinnovation zur Erklärung der Vielfalt betrachtet werden kann, obwohl es eine Rolle in der Diversifizierung der Ornithischia spielte. Insgesamt ist die Anatomie des Hirnschädels der Dinosauria plastischer als die der Pseudosuchia. Die Positionierung des Pterygoid-Quadratum-Komplexes an die Seitenwand des Hirnschädels in Crocodyliformes stellt eine anatomische Beschränkung dar, welche nur von Notosuchia und Thalattosuchia durchbrochen wurde. / Impedance-matching hearing is considered an important refinement of the auditory system of tetrapods because it reduces energy loss during sound transmission. Anatomically, it is characterized by the sub-division of the metotic foramen into a posterior vagus foramen and an anterior fenestra pseudorotunda. Impedance-matching hearing has evolved independently in several tetrapod groups including archosaurs and although it has been suggested that it represents a homoplasy, this hypothesis has never been tested. Therefore, 17 braincase characters were coded for 111 taxa and mapped on an informal supertree. Optimization of the characters revealed that the fenestra pseudorotunda appeared eight times independently in Archosauria, with five reversals. While this character is plastic in dinosaurs, it appeared only once in pseudosuchians. A tree-shape analysis revealed that pseudosuchians had six shifts in diversification rates, while dinosaurs had seven. Of these, only ornithischian ones are correlated to the appearance of impedance-matching hearing. Many of the overall state changes occur at the origin of major dinosaurian clades, but for pseudosuchians they are concentrated prior to the origin of Crocodyliformes and at the origin of Notosuchia and Thalattosuchia. The overall number of correspondences between character state changes and shifts in diversification rates is higher for dinosaurs, whereas in Pseudosuchia only Mesoeucrocodylia has a similar amount. It is thus possible to conclude that impedance-matching hearing is not homologous for archosaurs, and that it cannot be considered a key innovation triggering diversification. However, it may have played some role in ornithischian diversification. In general, the braincase anatomy of dinosaurs is more plastic than that of pseudosuchians. The abutting of the pterygo-quadrate complex against the lateral braincase wall was a strong anatomical constraint for crocodyliforms broken only by notosuchians and thalatosuchians.

O objetivo desta dissertação é propor o projeto de amplificadores de baixo rudo (LNAs) do tipo banda estreita e banda larga em tecnologia CMOS. O projeto de LNAs de banda estreita é representado através de um método de otimização conhecido como programação geométrica. Também, neste trabalho foi projetada uma topologia para LNAs de banda larga, aplicando a programação geométrica durante a fase inicial de projeto. Os layouts de ambos os circuitos foram desenhados e fabricados usando três processos CMOS diferentes. O aumento da utilização de circuitos digitais está reduzindo e substituindo a quantidade de circuitos analógicos implementados nos sistemas atuais. Nos transceptores de radiofrequência, a maior parte dos circuitos foi substituída por circuitos digitais equivalentes. A razão para esta substituição é devido a sua escalabilidade, variações PVT (Process, Voltage and Temperature) baixas, e menor tempo de projeto, resultado de um fluxo altamente automatizado. A redução do tempo de projeto representa um time-to-market menor e custos mais baixos. No entanto, o amplificador de baixo rudo é um dos blocos de radiofrequência que permanecem principalmente no domínio analógico, tornando a redução do tempo de projeto mediante a otimização do fluxo analógico como um bom foco de estudo. O LNA deve ser capaz de receber um sinal de baixa potência e alta frequência, e amplificá-lo adicionando o menor rudo possível, mantendo o casamento de impedâncias, baixo consumo de potência, e uma linearidade adequada a fim de evitar a distorção. Nesta dissertação, a maioria das especificações de desempenho citadas são formuladas rigorosamente e descritas como um programa geométrico. Além disso, vários scripts são escritos de forma a automatizar o fluxo de projeto. A programação geométrica é considerada como uma boa opção porque se o problema de otimização tem solução, o resultado é o ponto de otimização global, e pode ser atingido rapidamente (na ordem de segundos). Para um LNA fonte comum de banda estreita, o problema de projeto é completamente formulado como um programa geométrico, e alguns parâmetros normalmente desprezados, como as não idealidades dos indutores CMOS e a capacitância portadreno do transistor MOS são considerados no projeto. O problema de otimização é resolvido em minutos e testado em cinco processos CMOS diferentes, e para diferentes frequências de operação entre 1,5 GHz e 5 GHz. Os resultados são comparados e validados através de simulações, e dois layouts de LNAs para 2,45 GHz foram desenhados, fabricados e testados usando dois processos de 0,18 mm diferentes. Neste trabalho, também foi formulado um LNA de banda larga com cancelamento de rudo, e um bloco LNA-Misturador de banda larga é projetado incluindo a programação geométrica no cálculo da impedância de entrada e o cancelamento de rudo. Os layouts de dois protótipos diferentes do bloco LNA-Misturador de banda larga, operando na faixa de frequência entre 1 GHz e 5 GHz, foram desenhados e fabricados usando um processo de 0,18 mm. / This dissertation proposes the design of CMOS narrowband and wideband low noise amplifiers. The design problem of narrowband LNAs is represented as an optimization problem known as geometric programming. Furthermore, a topology for wideband LNAs is designed including the geometric programming in an early stage of the design. Both type of circuits were layouted and fabricated using three different CMOS processes. The tendency to increase the number of applications for digital-intensive circuitry, is reducing and replacing the amount of analog circuits implemented on systems nowadays. In radiofrequency transceivers, most of the circuits have been replaced by a digital-intensive counterpart. Digital circuitry is preferred over the analog one due to its scalability, low PVT (Process, Voltage and Temperature) variations, and shorter designing time result of a highly automated flow. The reduction of the designing time represents a faster time-to-market and lower costs. However, the low noise amplifier is one of the radiofrequency blocks that remain mainly in the analog domain, thus reducing its designing time by optimizing an analog design flow become a good focus of study. The LNA should be capable of receiving a low power and high frequency signal and amplify it adding the minimum noise possible, while maintaining good impedance matching, low power consumption and an adequate linearity in order to avoid distortion. In this dissertation, most of the performance parameters aforementioned are formulated rigorously and described as a geometric program. Moreover, various scripts are written in order to automate the design flow. The geometric programming is considered a good option because if the optimization problem is feasible, the result is the global optimum and can be obtained in seconds. For a common source narrowband LNA, the design problem is fully formulated as a geometric program and some parameters commonly neglected, as the CMOS inductors non-idealities and the gate-drain capacitance of MOS transistor are considered. The optimization problem is solved in minutes and tested on five different CMOS processes at different operating frequencies between 1.5 GHz and 5 GHz. The results are compared and validated through simulations, and two layouts for 2.45 GHz LNAs are drawn, fabricated and tested using two different 0.18 mm processes. In addition, a noise canceling wideband LNA is formulated, and a wideband LNA-Mixer cell is designed by including the geometric programming to estimate the input impedance matching and assure the noise cancelation. The layouts of two different prototypes of the wideband LNA-Mixer cells for the 1 GHz-5 GHz frequency band are drawn and fabricated using a 0.18 mm process.

<p>As the interest in intentional electromagnetic interference (IEMI) increases, so does the need of a suitable antenna which endures those demanding conditions. The ultrawideband (UWB) technology provides an elegant way of generating high-voltage UWB pulses which can be used for IEMI. One UWB antenna, invented solely for the purpose of radiating pulses, is the impulse radiating antenna (IRA). In the course of this master thesis work, a suitable geometry of the IRA is suggested, and modelled, for the high-voltage application of 90 kV.</p>

<p>This thesis presents solutions and studies related to the design of wideband antennas for wireless handheld terminal applications. A method of electrically shortening the terminal chassis length to obtain resonance at high frequencies has been proposed and evaluated, thereby increasing the antennas impedance bandwidth. No significant effect on the lower frequency band in a dual-band antenna prototype has been observed, making the method suitable for multi-band applications. The chassis has further been utilized as a zero-thickness 0.9 - 2.7 GHz high efficiency antenna by inserting a notch in the chassis center, and a feasibility study for typical phones has been performed. Additionally, the effect of talk position on the chassis wave-mode has been investigated, where the standard equivalent circuit model for terminal antennas has been modified to include the presence of the users head. The model has been used to explain measured and simulated effects concerning frequency detuning, efficiency reduction and bandwidth enhancements when the terminal is placed in talk position.</p><p>The use of a hands-free earpiece cord is currently mandatory for FM radio reception as the cord is utilized as antenna. However, there is currently a market driven demand for removing the cord requirement since many modern phones are equipped with speakers and Bluetooth headsets. In this thesis, an active ferrite loop antenna is proposed as an internal replacement/complement with a performance of -23 dB (G/T degradation) compared to a full-size lossless dipole in urban environments. Also, a modification to the cord is suggested for DVB H reception.</p><p>Complex matching networks have been investigated to increase the bandwidth of dual band PIFA antennas, and a printed dual band dipole has been integrated with a modified Marchand balun for dual resonance at two separate frequency bands, thus covering the commercial cellular bands 824-960 and 1710-2170 MHz with a single antenna.</p>

As the interest in intentional electromagnetic interference (IEMI) increases, so does the need of a suitable antenna which endures those demanding conditions. The ultrawideband (UWB) technology provides an elegant way of generating high-voltage UWB pulses which can be used for IEMI. One UWB antenna, invented solely for the purpose of radiating pulses, is the impulse radiating antenna (IRA). In the course of this master thesis work, a suitable geometry of the IRA is suggested, and modelled, for the high-voltage application of 90 kV.

This thesis reports the analysis of the rectangular shaped defected ground structure (RS-DGS) and the application of the structure on some microwave devices. DGS is analyzed in terms of its superior properties, which enables the designers to easily realize many kind of microwave devices which are impossible to achieve with the standard applications. Within the scope of this thesis, the focus is on the rectangular shaped DGS and its characteristic properties. The basic slow wave and high impedance characteristics are utilized in the design of some microwave devices. The design is carried on at the two different frequency bands: X-band and Ka band, centering at 10 GHz and 35 GHz, respectively. Finally, using the high impedance property and the coupling between the defects, a wide band 1 : 4 beam forming network is designed and implemented at 10 GHz.

Submitted by Paulo Armando Panunzio null (330698) on 2016-03-15T14:50:59Z No. of bitstreams: 1 VERSÃOFINALTESE24022016V22.pdf: 3679161 bytes, checksum: 20211476e30e236ed54da33d4530b7e1 (MD5) / Approved for entry into archive by Ana Paula Grisoto (grisotoana@reitoria.unesp.br) on 2016-03-15T17:27:06Z (GMT) No. of bitstreams: 1 panunzio_pa_dr_guara.pdf: 3679161 bytes, checksum: 20211476e30e236ed54da33d4530b7e1 (MD5) / Made available in DSpace on 2016-03-15T17:27:06Z (GMT). No. of bitstreams: 1 panunzio_pa_dr_guara.pdf: 3679161 bytes, checksum: 20211476e30e236ed54da33d4530b7e1 (MD5) Previous issue date: 2016-02-16 / A proposta dessa Tese é o projeto e desenvolvimento de um sistema de fornecimento de energia elétrica conectada à rede sem a utilização de circuitos eletrônicos complexos. O sistema eletroeletrônico convencional utiliza um inversor de frequência que transfere a potência ativa do nível de tensão e corrente CC para o nível compatível com a rede elétrica CA em frequência, tensão e sincronismo de fases. Já o desenvolvimento do projeto teve como parte experimental a utilização de indutores adequados com valores variáveis de 200 mH a 500 mH entre os painéis fotovoltaicos e a rede convencional. A tensão e a corrente máxima de CC foi de 29,6 V e a máxima possível foi de 20 amperes. Utilizou-se do princípio do casamento de impedâncias entre o painel fotovoltaico e o sistema de fornecimento de energia CA. Assim o painel fotovoltaico fornece somente potência ativa para a rede não interferindo no sincronismo. Na onda de tensão e corrente CA há uma pequena alteração no nível CA em relação a simetria do eixo dos tempos, dentro dos limites previstos para a rede convencional de energia. Logo evidencia-se o fornecimento de potência ativa para a rede CA. Os resultados obtidos permitiram inserção na rede de cerca de 10% da potência ativa dos painéis fotovoltaicos, com a otimização dos valores dos indutores, ocorrendo a transferência de potência ativa dos painéis fotovoltaicos para a rede convencional de energia CA. / The purpose of this thesis is the design and development of a delivery system of electricity connected to the network without the use of complex electronic circuits. The electronics system uses a frequency inverter that transfers the active power level voltage and DC current to the level compatible with the mains AC frequency, voltage and phase synchronization. But the project was to develop experimental part the use of suitable inductors with variable values of 200 mH to 500 mH between the PV panels and the conventional network. The maximum current was 20 amperes. We used the principle of impedance matching between the photovoltaic panel and the power supply system CA. So the photovoltaic panel supplies only active power to the grid not interfering with the timing. In the wave of AC voltage and current for a small change in CA level against the symmetry axis of time. Logo is evident in the supply of active power to the grid CA. The results allowed inclusion in the network of about 10% of the active power of photovoltaic panels by optimizing the values of the inductors for the occurrence of power transfer of conventional photovoltaic panels for energy occurs CA.

L’objectif de ce travail est d’aborder la conception de réseaux d’adaptation d’impédance accordable (RAA) dans deux contextes bien distincts en radiofréquences : le RAA en faible signal et le RAA en fort signal.Concernant les aspects faible signal, des critères de performance de RAA ont été établis et étudiés. Une nouvelle expression générale de l’efficacité d’un RAA a été développée. Elle permet de prédire le rendement d’un RAA à partir des facteurs de qualité des composants dont on dispose dans une technologie donnée et du rapport de transformation à réaliser. Des abaques de couverture d’impédances en fonction des pertes d’insertion ont été calculés. Ils mettent en évidence les régions de couverture où le RAA apporte une amélioration à la performance du système, pour diverses topologies de RAA.Un démonstrateur sur PCB a été réalisé. Il est constitué de deux RAA, qui assurent l’adaptation simultanée d’un amplificateur de puissance sur une large plage d’impédances, comprises dans un cercle de l’abaque de Smith d’équation VSWR < 5 :1. La zone de couverture a été mesurée et présentée en fonction des pertes d’insertion, qui mettent en évidence les régions où les RAA contribuent à l’amélioration de la performance de l’amplificateur et celles où les pertes d’insertion du RAA n’arrivent pas à compenser le gain du à la réduction des coefficients de réflexion.Dans une seconde partie, la conception de RAA en fort signal a été traitée. L’objectif a été de présenter à la sortie d’un amplificateur de puissance les impédances qui optimisent son efficacité pour chaque puissance de travail. Un démonstrateur en technologie CMOS SOI 130 nm a été conçu et simulé. Il consiste en un amplificateur de puissance pour le standard WCDMA, fonctionnant à 900 MHz, et un RA accordable par des varactors MOS, capable de générer les impédances optimales correspondant à des puissances de sortie comprises entre 20 et 30 dBm. Les résultats ont mis en évidence le bénéfice apporté par l’insertion d’un RA accordable par rapport à un RA fixe. / The aim of this research was the design of tunable matching networks, in two different contexts: the TMN at low and high signals.Performance criteria of TMN were studied. A general expression has been developed that estimates the TMN efficiency in function of the quality factor of the components used. The impedance coverage of different RAA topologies was plotted in function of the insertion losses.For the small signal, we designed a prototype in PCB, composed by two TMN, which ensure the simultaneous matching of a power amplifier in a wide range of impedances, included in a circle of the Smith chart VSWR < 5: 1. The coverage area was measured and presented in function of the insertion losses, emphasizing the areas where the TMN contribute to improve the amplifier performance and those where insertion losses of the TMN are not able to compensate gain with the reduction of the reflection coefficients.Subsequently, we discussed the design of TMN for the large signal. The objective is to present at the output of a power amplifier, the impedances that optimize efficiency for each power of operation. A 130 nm SOI prototype was designed and simulated, consisting of a power amplifier for WCDMA standard, 900 MHz, and a MN tunable by MOS varactors able to produce the optimal impedances corresponding to an output power between 20 and 30 dBm. The results showed the benefit provided by inserting a tunable MN compared to a fixed one.

Made available in DSpace on 2014-12-17T14:55:10Z (GMT). No. of bitstreams: 1 ElderECO.pdf: 2183610 bytes, checksum: d02ab6a3abf8c0cd3e77eadf92d529a0 (MD5) Previous issue date: 2008-12-03 / Coordena??o de Aperfei?oamento de Pessoal de N?vel Superior / The microstrip antennas are in constant evidence in current researches due to several advantages that it presents. Fractal geometry coupled with good performance and convenience of the planar structures are an excellent combination for design and analysis of structures with ever smaller features and multi-resonant and broadband. This geometry has been applied in such patch microstrip antennas to reduce its size and highlight its multi-band behavior. Compared with the conventional microstrip antennas, the quasifractal patch antennas have lower frequencies of resonance, enabling the manufacture of more compact antennas. The aim of this work is the design of quasi-fractal patch antennas through the use of Koch and Minkowski fractal curves applied to radiating and nonradiating antenna s edges of conventional rectangular patch fed by microstrip inset-fed line, initially designed for the frequency of 2.45 GHz. The inset-fed technique is investigated for the impedance matching of fractal antennas, which are fed through lines of microstrip. The efficiency of this technique is investigated experimentally and compared with simulations carried out by commercial software Ansoft Designer used for precise analysis of the electromagnetic behavior of antennas by the method of moments and the neural model proposed. In this dissertation a study of literature on theory of microstrip antennas is done, the same study is performed on the fractal geometry, giving more emphasis to its various forms, techniques for generation of fractals and its applicability. This work also presents a study on artificial neural networks, showing the types/architecture of networks used and their characteristics as well as the training algorithms that were used for their implementation. The equations of settings of the parameters for networks used in this study were derived from the gradient method. It will also be carried out research with emphasis on miniaturization of the proposed new structures, showing how an antenna designed with contours fractals is capable of a miniaturized antenna conventional rectangular patch. The study also consists of a modeling through artificial neural networks of the various parameters of the electromagnetic near-fractal antennas. The presented results demonstrate the excellent capacity of modeling techniques for neural microstrip antennas and all algorithms used in this work in achieving the proposed models were implemented in commercial software simulation of Matlab 7. In order to validate the results, several prototypes of antennas were built, measured on a vector network analyzer and simulated in software for comparison / As antenas de microfita est?o em constante evid?ncia nas pesquisas atuais, isso devido ?s in?meras vantagens que apresentam. A geometria fractal aliada ao bom desempenho e comodidade das estruturas planares s?o uma excelente combina??o para projetos e an?lise de estruturas cada vez menores e com caracter?sticas multi-ressonantes e banda larga. Essa geometria tem sido aplicada em antenas tipo patch em microfita para reduzir o seu tamanho e evidenciar o seu comportamento multi-banda. Em compara??o com as antenas em microfita convencionais, as antenas patch quase-fractais apresentam freq??ncias de resson?ncia inferiores, possibilitando a fabrica??o de antenas ainda mais compactas. O objetivo desse trabalho consiste no projeto de antenas patches quase-fractal por meio da utiliza??o de curvas fractais de Koch e Minkowski aplicado ?s margens radiante e n?o-radiante de uma antena inset-fed patch retangular convencional alimentada por linha de microfita com reentr?ncias, inicialmente projetada para a freq??ncia de 2,45 GHz a ser observada. A t?cnica inset-fed ? investigada para o casamento de imped?ncias das antenas fractais, que s?o alimentadas atrav?s de linhas de microfita com reentr?ncias. A efici?ncia dessa t?cnica ? investigada experimentalmente e comparada com simula??es realizadas pelo software comercial Ansoft Designer, usado para a an?lise precisa do comportamento eletromagn?tico das antenas atrav?s do m?todo dos momentos e pelo modelo neural proposto. Nessa disserta??o um estudo bibliogr?fico em teoria de antenas de microfita ? realizado, o mesmo estudo ? realizado a respeito da geometria fractal, dando ?nfase a suas mais diversas formas, t?cnicas de gera??o desses fractais bem como sua aplicabilidade. Este trabalho ainda apresenta um estudo em redes neurais artificiais, evidenciando os tipos/arquitetura de redes utilizadas e suas caracter?sticas, bem como os algoritmos de treinamento que foram utilizados para sua implementa??o. As equa??es dos ajustes dos par?metros para as redes utilizadas nesse trabalho foram deduzidas a partir do m?todo do gradiente. Tamb?m ser? realizada uma investiga??o com ?nfase na miniaturiza??o dessas novas estruturas propostas, indicando o quanto uma antena projetada com contornos fractais ? capaz de miniaturizar uma antena patch retangular convencional. O estudo tamb?m consiste de uma modelagem por meio de redes neurais artificiais dos mais diversos par?metros eletromagn?ticos das antenas quase-fractais. Os resultados apresentados demonstram a excelente capacidade das t?cnicas neurais para modelagem de antenas de microfita, e todos os algoritmos utilizados nesse trabalho na obten??o dos modelos propostos foram implementados no software comercial de simula??o Matlab 7. Com a finalidade de validar os resultados obtidos, v?rios prot?tipos de antenas foram constru?dos, medidos em um analisador de rede vetorial e simulados em software para compara??o