Search

MgTiO3 is a material often used in different types of high-frequency capacitors, temperature compensating capacitors, and chip capacitors, so the enhancement of this material is still the focus of many research groups due to its remarkable dielectric properties. Outstanding features can only be achieved when the ceramics are highly dense. Densification of magnesium titanate by Spark Plasma Sintering (SPS) was the aim of this work. Magnesium titanate ceramics were prepared by applying mechanical activation as the first step. Powders prepared in this way were SPS sintered, at 1200 °C with a heating rate of 100 °C/min. After reaching the desired temperature, a uniaxial pressure of 50 MPa was applied. The dwell time at this condition was 5 min, followed by cooling to room temperature at 5°C/min. X-ray diffraction was performed in order to establish the phase composition of milled powders and obtained ceramics. Differences between samples milled in various times intervals, as well as sintered ceramics were examined by means of scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS). The presence of MgTiz0s phase was detected in XRD and was confirmed by EDS analysis for the non-milled ceramics. In the samples obtained from milled powders, no MgTi2O5 was detected in XRD patterns, but this phase was detected in EDS spectra in a lower amount. Dielectric measurements were performed at a wide range of frequencies, while the hardness of the SPS samples was measured at loads up to 10 N. The highest value of the hardness was obtained from powder milled for 15 min before SPS.

Perovskite ceramic material based on MgTiO3 is used in various types of electronic devices owing to its dielectric properties, high dielectric constant, and low losses. These features can be tailored by setting preparation conditions. Densification of magnesium titanate by Spark Plasma Sintering (SPS) was the aim of this work. First, the mechanical activation in the highenergy ball mill was applied on the powder of MgO-TiO2 mixed in mole ratio 1:1. Prepared powder mixtures, activated for different times, were SPS sintered, at 1200 o C with a heating rate of 100o C/min. After reaching the desired temperature, a uniaxial pressure of 50 MPa was applied, and dwelled at this condition for 5 min, followed by cooling to room temperature at 5o C/min. The starting powders, activated mixtures, and sintered ceramics bodies were investigated by X-ray diffraction, scanning electron microscopy (SEM), and energy dispersive spectroscopy (EDS). The presence of the MgTi2O5 phase was noticed for the nonmilled ceramics. In the samples obtained from milled powders, MgTi2O5 was detected in EDS spectra in a lower amount, below the threshold of the XRD method. Dielectric measurements were performed at a wide range of frequencies and temperatures. The highest value of the hardness was obtained from powder milled for 15 min before SPS. X Serbian Ceramic Society Conference - Advanced Ceramics and Application : new frontiers in multifunctional material science and processing : program and the book of abstracts; September 26-27, 2022; Belgrade

Our research group has developed a method for measurement of complex relative permittivity of various dielectric materials in the frequency range from around 1 kHz up to several GHz. Material samples have preferably a disk shape. The thicknesses of the samples can be in a wide range, from about 10 μm (thick films) up to several mm. We have designed and manufactured a set of coaxial chambers, which we use as test fixtures. We have also developed two numerical-simulation programs for the electromagnetic analysis of bodies with rotational symmetry. One program is suitable for the low-frequency analysis. It is based on an electrostatic approach. The other program is based on an electrodynamic approach and it is tailored for microwave frequencies. In measurements, we use impedance meters and network analyzers to obtain the input impedance of a chamber with a sample. Thereafter, we implement our software for the electromagnetic modeling to extract the relative permittivity of the measured sample. As examples of verification of our method, we present here results for the relative permittivities of two sets of samples whose sizes are on the extreme limits of the method. The first set comprises poly (vinylidene fluoride) and mechanically activated ZnO nanoparticle composite films, whose relative permittivities are around 1.8. The second set comprises large, high-density samples of spinel (aluminum magnesium oxide) ceramics, sintered under various conditions. The measured relative permittivities of these samples are around 7.5. In all cases, good agreement with other available data has been obtained.

The increase in the utilization of ceramic/polymer composites as components for electric devices is mainly based on their high chemical stability, mechanical strength, and flexibility. The polymers usually have poor dielectric performances. In order to increase the dielectric permittivity, ceramic fillers with high dielectric constant have been used as active fillers, and usually in large amounts, which have a detrimental effect on the mechanical properties of the composite. With the aim to enhance the dielectric properties of the composite without loss of polymer’s flexibility, 5 wt.% of BaTiO3/Fe2O3 core/shell powders, prepared by different synthesis conditions, was added into the PVDF matrix. The effect of the phase composition and morphology of the starting BaTiO3/Fe2O3 core/shell filler on the crystal structure and lattice dynamics was investigated. Based on the results of the thermal analysis, TG/DSC, various parameters of ceramic/polymer composites were determined. We were able to corroborate that differences in the phase composition and morphology of BaTiO3/Fe2O3 core/shell filler have an influence on the formation of various PVDF allomorph modifications, as well as the level of crystallinity. The dielectric performances of the pure PVDF and polymer/ceramic composites were measured. An enhancement in the dielectric permittivity and decrease in the loss tangent were revealed. IX Serbian Ceramic Society Conference - Advanced Ceramics and Application : new frontiers in multifunctional material science and processing : program and the book of abstracts; September 20-21, 2021; Belgrade

Corundum (alpha-alumina) is one of the most exploited materials in ceramic industry due to its good physico-chemical properties, mechanical properties, and, importantly, due to its low cost. Advantageous properties, such as hardness, dielectric and thermal properties, and refractoriness, open the door to a wide range of applications of corundum. Porous alumina ceramics is often used for filters and as a catalytic substrate, whereas dense alumina ceramics is frequently used in automotive and aerospace industry. Hence, there is a great interest in exploring this type of ceramics. The aim of this work is to explain the influence of preparation parameters (mechanical activation, addition of transition metal oxides, and sintering conditions) on the final electrical and mechanical properties of alumina. As a starting powder, a mix of a few alumina modifications (α-, κ-,γ - Al2O3) was used. It was doped with 1 wt % of Mn2O3, Cr2O3, and NiO, respectively. Three of the obtained mixtures were homogenized and three were ball milled for 60 min. Particle size analysis, SEM, and EDS were performed in order to demonstrate changes in the microstructure after milling. DTA/TG analyses were used to determine all characteristic temperatures. After sintering at 1400 oC in an air atmosphere, mechanical, electrical, and physico-chemical properties of the obtained ceramics were determined. Results showed that the mechanical treatment lead to a decrease in temperatures of phase transitions and sintering for approximately 100 oC. Incorporation of cations of transition metals into the crystal lattice of alumina was noticed. Activated and sintered samples showed a higher tensile strength and higher relative dielectric permittivity then the sample doped with Cr2O3. The main conclusion of this investigation is that the mechanical activation has the dominant influence on lowering the characteristic temperatures of all detected phase transitions and sintering processes. Using a mixture of several crystalline phases of alumina as a starting powder alleviates the incorporation of transition-metal cations into the alumina crystal lattice, leading to changes in microstructures of the prepared ceramics. Finally, changes in the microstructure and lattice disordering have the dominant influence on the final mechanical properties.

Magnesium aluminate and other alumina-based spinels are refractory ceramics with excellent properties, such as high hardness, high mechanical strength, and low dielectric constant, applicable in many modern industries. MgAl2O4 was produced by solid state reaction between MgO and alpha-Al2O3 powders. Mechanical activation represents a very efficient method for increasing the reactivity of powders, accelerating chemical reactions and decreasing the sintering temperatures. Mechanical activation of mixed powders was performed in a high-energy planetary ball mill in air for 60 minutes. Sintering was performed in air at temperatures ranging from 1200°C to 1600 °C with a 2 h dwell time. Initial powders and sintered samples were characterized using X-ray diffraction and scanning electron microscopy. Differential thermal analysis (DTA) and thermal gravimetry (TG) were used to determine the temperatures for synthesis reactions and phase transformations. Based on the DTA results, several different processes occured during heating, including evaporation of physisorbed water, decomposition of Mg(OH)2, and spinel formation. With mechanical activation, all characteristic temperatures shifted to lower values, and peaks were more pronounced. Raman spectra were used to characterize the degree of inversion as a function of sintering temperature for all of the sintered specimens.

We investigate the influence of a wire-based impedance-matching network for an axial-mode helical antenna on the radiation pattern. This network, proposed in our previous work, comprises a single wire attached to the helix. We show that the matching wire, mounted on a tubular dielectric support and attached to the helix close to the reflector, does not degrade the antenna radiation pattern. We also investigate limitations of the equivalent thin-wire model proposed in our previous work.

Mechanical activation is commonly used as a pre-sintering process in order to enhance the reactivity of materials, reduce the particle size, increase diffusion rates, accelerate the reaction, and lower the sintering temperature. The mechanical activation can affect the final electrical and mechanical characteristics. In this paper we consider the influence of the mechanical activation on the permittivity and the loss tangent. We outline methods for evaluation of these parameters, with emphasis on our coaxial-chamber technique for measurements in the VHF band.

The paper presents results of investigation of the influence of mechanical activation of shungite, a Russian natural mineral rich in silica and carbon, on its sintering behavior. The mechanical activation of the starting powder was performed in a high-energy ball mill in time intervals from 0 to 480 minutes. The phase composition of the starting mixtures and sintered samples was analyzed by the X-ray diffraction method. The scanning electron microscopy was performed in order to determine changes in the microstructure. Sintering was performed at various temperatures for 2 h, in an Ar and vacuum atmosphere. Dielectric properties of the sintered samples were measured in the frequency range from 1 to 500 MHz. The obtained results indicate that sintered shungite powder is a good candidate for applications as an absorber of electromagnetic waves in microwave engineering.

Sintered alumina powder represents suitable material for usage in various industry fields (e.g., as chip carriers in electronics, microwaves, jewelry production), due to convenient physical properties, such as sinterability, electrical and mechanical features. Those properties can be modified by addition of different oxides and/or mechanical treatment. Therefore, in this investigation the alumina was doped with 1 wt. % of Cr2O3, Mn2O3 and NiO, respectively, followed by 1 hour of mechanical activation at 400 rpm in planetary ball mill. Sintering of powder mixtures was tracked by sensitive dilatometer up to 1400 °C. The final density values varied from cca. 2 ̶ 3.2 g/cm3. Changes in microstructure were observed by means of SEM. The influence of additives along with mechanical activation is monitored trough changes in electrical permittivity and loss tangent. Compared to pure alumina, the additives lower the relative permittivity and increase dielectric losses. For a given mixture, the sintering increases the relative permittivity and decreases losses.

Recently, with the huge use of smart gadgets, developing of smart jewelry represents a very interesting segment in material science, as well as in electronic science. Alumina is widely used ceramic in many industrial fields as pigments, catalysis, microelectronics, etc., mostly because of its low cost and appropriate mechanical and electrical properties, high surface area and thermal stability. Also, modified alumina could be applied in production of smart jewelry. Thus, the main objectives of this investigation is to improve features of sintered alumina doped with Mn2O3 along with mechanical treatment, in order to obtain strong ceramic with low values of dielectric loss and low relative dielectric permittivity, as well as esthetic. Commercial alumina powder was doped with 1 wt % of manganese oxide and treated in planetary ball mill for an hour. Characteristic temperatures of both powders (non-activated and activated one) were investigated in detail by DTA and TG analyses. After sintering at 1200, 1300, and 1400 oC for 2 h, XRD patterns and SEM images were recorded. Furthermore, mechanical and electrical properties were examined for all sintered samples.

In this study, sintering of two systems was investigated in parallel, pure cordierite 2MgO:2Al2O3:5SiO2, and cordierite with addition of 5 mass % TeO2. Green bodies were prepared from powder mixtures mechanically activated for 0, 10 and 40minutes by uniaxial pressing at 20 MPa followed by cold isostatic pressing (CIP) at 1000 MPa. The pressure-less sintering of these specimens was performed at a temperature of 1350 oC for 1 h, with 10 oC/min heating rate. Densities over 96 % of theoretical values were obtained for samples activated for 40 minutes. XRD measurements of MAS-40 sintered sample pointed at a mixture that consisted of several phases, majority of cordierite (> 73 %), spinel MgAl2O4 (> 22 %), ZrSiO4 (2.5 %), and Al2O3 (1.8%), while XRD pattern of MAS-40-TeO2 contained majority of cordierite (> 72 %), spinel MgAl2O4 (> 23 %), ZrSiO4 (1.9 %), and Al2O3 (1.9 %). The relative permittivities of the samples were measured in an open test fixture. The results for the real part of the complex relative permittivity of the samples, measured at 200 MHz, are shown in Table I. The loss tangent for all samples was very small, below the resolution of the measurement setup (0.005). There is a strong correlation between the relative permittivity and the density, which agrees with our previous results.

We present an overview of electromagnetic numerical techniques and measurement apparatus that have been used for assessing electrical properties of ceramic samples. The complex permittivity of the ceramics is typically the first electromagnetic property of interest. In the measurement approach that we use, ceramic samples, in the shape of pills with metalized faces, are placed in either a proprietary coaxial chamber or in an open fixture. The scattering parameter (reflection coefficient) is measured using a network analyzer. For the deembedding of the complex permittivity of a sample one needs precise electromagnetic simulations, including detailed models of the chamber or the open fixture. For lower and intermediate frequencies we use a numerical technique for electrostatic analysis, based on the method of moments with Galerkin testing. We create a database that is used afterwards for evaluation of the complex permittivity practically in real time. For higher frequencies, a full electrodynamic model of the chamber is developed with WIPL-D software. For all our deembedding techniques, we have also developed a detailed model of losses in the SMA coaxial connectors, which is essential for a precise evaluation of the imaginary part of the complex permittivity for low-loss dielectrics. We present results of several measurements in order to illustrate the used techniques.

Mixtures of MgO-TiO2 powders were mechanically activated in planetary ball mill for time interval from 0 to 120 minutes. The influence of mechanical activation on phase composition and crystal structure was analyzed by XRD, while the effect of activation and sintering process on microstructure was investigated by scanning electron microscopy. Using a data obtained by XRD microstructure parameters, values of crystallite size (D), density of dislocation (ρD) and lattice strain (ehkl) were calculated. Dielectric measurements are performed in order to show difference in dielectric constant as a function of time of mechanical activation.

In this paper we present voltage-current-power characteristics of a plasma needle operating in the flow of helium at atmospheric pressure. In addition, we show some examples of how such a plasma affects plant tissues. In the characterization of the plasma needle, current and voltage waveforms were recorded by two derivative probes. These two probes are similar to the probes previously used by Puac et al for measuring transmitted power in low pressure CCP rf discharge. The instantaneous power was calculated from current and voltage waveforms and U-I characteristics of the discharge were determined. Regimes of operation with and without the grounding ring at the tip of the needle were considered. We have chosen two model systems to study the effect of the plasma needle on plant cells and tissues: sweet fern gametophyte (prothallus) and calli produced in vitro. Since the prothallus consists of a single layer of cells, the cytological effects could be easily examined. In addition, calli and prothallus are easy to manipulate and in vitro culture provides a possibility to work under constant and controlled conditions. null

The objective of this thesis is to systematically analyze and optimize nonuniformly-wound helical antennas, along with classical (uniform) helical antennas. The optimization of the nonuniform helical antennas has many degrees of freedom. Hence, the optimization space is large and the optimization task is challenging. It is shown that, in most practical cases, the optimal nonuniform helical antennas outperform the uniform helical antennas presented in the literature. It is also shown that the nonuniform helical antennas are the preferable choice when the losses are low or medium, whereas for high losses, the uniform helical antennas should be used. A large database is assembled from the optimization results, wherefrom a complete design procedure is developed for the nonuniform helical antennas. This procedure comprises all necessary equations and graphs for evaluating the optimal antenna parameters and estimating the antenna characteristics. The design procedure is verified experimentally, by measurements of a fabricated prototype. Quad (2 x 2) arrays of nonuniform helical antennas are also investigated. Their design procedure includes the optimization of single antennas along with their positions in the array. The solution of a real engineering problem is presented: a quad array that meets predefined specifications is designed and a prototype is fabricated and measured. Циљ ове тезе је систематична анализа и оптимизација неуниформно мотаних хеликоидалних антена, као и класичних (униформних) хеликоидалних антена. Оптимизација неуниформних хеликоидалних антена има много степени слободе. Стога је оптимизациони простор велики, а оптимизација изазован задатак. Показано је да, у највећем броју практичних случајева, оптималне неуниформне хеликоидалне антене надмашују по перформансама униформне хеликоидалне антене представљене у литератури. Осим тога, показано је да су неуниформне хеликоидалне антене бољи избор када су губици мали или средњи, док би у случају великих губитака требало користити униформне хеликоидалне антене. Резултати оптимизације чине велику базу података, на основу које је развијен поступак пројектовања неуниформних хеликоидалних антена. Овај поступак обухвата све потребне једначине и графике за одређивање оптималних параметара антена и процену карактеристика антена. Поступак дизајна је потврђен експериментално, мерењем карактеристика реализованог прототипа. Такође су испитивани низови од 2 x 2 неуниформне хеликоидалне антене. Процедура за њихов дизајн садржи оптимизацију засебних антена, као и оптимизацију положаја антена у низу. Представљено је и решење реалног инжењерског проблема: дизајниран је низ од 2 x 2 антене који испуњава унапред дефинисане спецификације, а прототип је направљен и измерен.

Представљена је квалитативна метода нумеричке електромагнетике за микроталасно формирање слике, која се ослања на решавање инверзног проблема расејања. У уводном делу дат је преглед литературе и укратко су дефинисане предности предложеног алгоритма у односу на већ постојеће методе. Након увода, дефинисани су основни постулати инверзних проблема и упоређени са добро познатом формулацијом директних електромагнетских проблема. Након тога, објашњене су потешкоће које настају при решавању инверзних проблема, односно показано је да су они у општем случају нелинеарни и некоректно постављени. Такође, детаљно је описана техника развоја по мултиполима као фундаментална алатка у аналитичкој електромагнетици, на којој се заснива приказана метода... An electromagnetic qualitative microwave imaging method, which relies on solving an inverse scattering problem, is presented. In the introductory part of this dissertation, the state-of-the-art is briefly summarized. Also, main advantages of the proposed method, compared to other known methods, are outlined. After the introduction, we define the basic idea of an inverse problem, compared to the well-known direct electromagnetic problem formulation. Then, we explain the main difficulties arising during an attempt to solve such an inverse problem, i.e., it is shown that these problems are generally non-linear and ill-posed. Also, the multipole expansion technique, as a fundamental tool in analytical electromagnetics, is described in detail...

Zbog velike zastupljenosti radiofrekvencijskih (RF) plazmi u industriji javila se potreba za definisanjem osnovnih fizičkih principa koji određuju proboj u RF električnim poljima. Uslovi, u vidu probojnog napona i pritiska, mogu se prikazati naponskom probojnom krivom, predstavljenom u V-p ravni. Oblik i položaj krive definisani su kako prirodom pozadinskog gasa tako i karakteristikama spoljašnjeg RF električnog izvora. Ova disertacija ima za cilj da nedvosmisleno definiše fizičke procese koji su odgovorni za proboj pri RF električnim poljima. Za to je korišćena Monte Karlo tehnika. Odabirom slučajnih brojeva, uz poznavanje preseka za određene procese, moguće je verno (čak se može tvrditi egzaktno) opisati kretanje roja naelektrisanih čestica između dve elektrode koje su pobuđene RF generatorom. Izgradnja modela je započeta posmatranjem kretanja elektrona u argonu pri frekvenciji od 13,56 MHz. Na primeru argona objašnjeni su osnovni mehanizmi proboja koji definišu oblik naponske probojne krive. Nadogradnja modela, kao verodostojne predstave RF proboja, uključuje efekte na površinama elektroda (refleksiju i emisiju sekundarnih elektrona), uticaj promene frekvencije i dvofrekvencijskih izvora, kao i teških čestica. Model je primenjen na različite gasove: helijum, kiseonik, sintetički vazduh i atmosfera Marsa. The wide range of radiofrequency (RF) plasma applications in industry rely on research of fundamental physical processes responsible for the RF breakdown. Conditions, such as the breakdown voltage and pressure, can be presented by breakdown voltage curves in the V-p plane. The shape and coordinates of the curve are defined by the nature of the background gas and characteristics of the applied RF electric field. The goal of this dissertation is to define physical processes responsible for the RF breakdown. For that purpose a Monte Carlo computational code has been used. If one knows cross sections for different processes by employing a random number generator, the transport of electrons between two electrodes can be determined. The development of the RF breakdown model begun by observing the motion of electrons in argon at the field frequency of 13.56 MHz. Argon was used to explain the basic mechanisms that define the shape of the breakdown voltage curve. Further upgrade of the model involved inclusion of the surface effects on electrodes (reflection and secondary electron emission), the influence of frequency and dual frequency RF power supplies, as well as the influence of heavy particles on the breakdown. The model was applied to different gases: helium, oxygen, and synthetic air, as well as the atmosphere of Mars.

U ovoj doktorskoj disertaciji razvijena je nova metoda koja omogućava određivanje nivoa električnog polja (u slobodnom prostoru) nepoznatog izvora korišćenjem skalarnih merenja u Faradejevom kavezu. Nova metoda omogućava upotrebu Faradejevog kaveza kao jeftine alternative standardizovanim mernim mestima za ispitivanje elektromagnetske kompatibilnosti (EMK), kao što su (polu)anehoične sobe, otvoreni merni prostor i reverberacione sobe (kavezi). Nova metoda, u odnosu na standardizovani merni postupak u reverberacionim sobama, omogućava uspešno merenje uređaja na niskim frekvencijama, koje su ispod najniže rezonantne frekvencije kaveza. Metoda se zasniva na aproksimaciji električnog polja zračenja ispitivanog uređaja nizom ekvivalentnih dipola. Parametri ekvivalentnih dipola su određeni primenom tehnike obrade retkih signala, odnosno primenom l1 regularizacije kojom se umanjuje neodređenost rešenja inverznih elektromagnetskih problema. Složenost prostiranja signala u sredinama kao što je Faradejev kavez povoljno utiče na tačnost rešenja. Predloženi metod je verifikovan skalarnim merenjima izvora u Faradejevom kavezu. Rezultati obrade su upoređeni sa onima dobijenim u anehoičnoj sobi i rezultatima procene parametara izvora korišćenjem analizatora spektra i numeričke obrade. Tom prilikom korišćen je namenski projektovan i napravljen izvor elektromagnetskog zračenja. U disertaciji je, osim toga, prikazano precizno modelovanje Faradejevog kaveza (uključujući razne nesavršenosti), a dat je i opis merenja odziva ekvivalentnih dipola u Faradejevom kavezu. Pri tome, bilo je neophodno odrediti mernu nesigurnost ispitivanja u Faradejevom kavezu i mernu nesigurnost modela Faradejevog kaveza. Takođe, precizno je definisana potrebna merna postavka pri merenjima izvora u Faradejevom kavezu. Na kraju disertacije predložene su smernice za budući rad. In this dissertation a new algorithm has been developed that enables determination of freespace field levels of radiating sources using amplitude measurements in Faraday cages. The new approach enables using Faraday cages as a low-cost alternative to the standard EMC test sites, such as (semi)anechoic chambers, open-area test sites, and reverberation chambers. In comparison to the standardized EMC tests in reverberation chambers, the new method enables measurements at low frequencies, even below the lowest resonant frequency of the cage. The algorithm computes the radiation parameters of a DUT in a free space, by means of equivalent dipoles and sparse processing. In order to extract the parameters of the equivalent dipoles, we use sparse processing and l1 regularization for solving ill-posedness of this inverse electromagnetic problem. By exploiting the multipath from the Faraday cage walls, floor, and ceiling, we improve the estimation accuracy. The algorithm has been verified experimentally by taking phaseless measurements of a known device in the Faraday cage. The results of the reconstruction have been compared with those obtained in an anechoic chamber and with directly assessed parameters of the source used a spectrum analyzer and a full-wave numerical model. The source of electromagnetic radiation was designed and developed specifically for this purpose. In this dissertation the model of the Faraday cage and numerical evaluation of the responses of the equivalent dipoles in the Faraday cage are described. Consequently, it has been necessary to determine the measurement uncertainty of testing in the Faraday cage and the measurement uncertainty of the model Faraday cage. Also, the measurements setup for measurements of the radiating sources in Faraday cages is defined. At the end of this dissertation guidelines for the future work are proposed.

Дисертација се бави унапређењем телекомуникационих система и у њој су предложена решења која се могу искористити у садашњим и будућим телекомуникационим системима. Након кратког увода у основне класе рада појачавача на микроталасним фреквенцијама, представљена су три појачавача снаге. Сва три појачавача су намењена модерним телекомуникационим системима. Ближе речено, сва три појачавача су конципирана и направљена да раде у више фреквенцијских опсега. У дизајнирању појачавача, акценат је био стављен на ефикасност појачавача. Стога су прва два појачавача дизајнирана и реализована да раде у класи АБ. Код првог, реконфигурабилног, појачавача једна од две могуће радне фреквенције бира се помоћу укључења/искључења PIN диоде у улазном колу за прилагођење. Други, подесиви (тјунабилни), појачавач је дизајниран и реализован тако да може континуално да мења фреквенцију рада, између две удаљене фреквенције, адекватном поларизацијом варикап диоде смештене у улазно коло за прилагођење. Трећи појачавач је дизајниран и реализован да ради у класи Ј. У вези с њим је најпре представљена теорија која се до сада није могла наћи у литератури. Та теорија представља допуну постојеће теорије везане за појачаваче у класи Ј. У оквиру новоразвијене теорије предложен је начин дизајнирања појачавача, где се на почетку одређују параметри појачавача за унапред задату ефикасност. Пратећи предложену теорију, дизајниран је и реализован појачавач у класи Ј. Резултати добијени симулацијом и мерени резултати су се одлично поклопили, што је верификовало ваљаност предложене теорије. На крају су дати резултати симулација нелинераности пројектованог појачавача у случају да му се на улаз доведу неки од сигнала који се користе у модерним телекомуникационим системима... This dissertation addresses the improvement of telecommunication systems. The proposed solutions can be implemented in both present and future telecommunication systems. After a brief introduction regarding the classes of amplifiers operating at microwave frequencies, three power amplifiers are presented. All three amplifiers are designed for modern telecommunication systems. More precisely, they are able to operate in several frequency bands. In the process of the amplifier design, the focus was set on the amplifier efficiency. Hence, the first two amplifiers are designed and fabricated to operate in class AB. In the first, reconfigurable amplifier, one of two possible operating frequencies can be chosen by switching on or off a PIN diode in the input matching network. The second, tunable, amplifier is designed and fabricated in such a way that the operating frequency can be continuously changed, between the lower and upper frequency boundaries. This is achieved by controlling the bias of the varicap diode, positioned in the input matching network. The third amplifier is designed and fabricated to operate in class J. Firstly, a novel theory describing the operation and design procedure of a class-J power amplifier is presented as an extension of the existing theory given in the literature. In the process, novel simple formulas are derived which govern the amplifier efficiency. An amplifier in class J is designed and fabricated following the proposed theory. Results obtained by numerical simulations and measurements on fabricated prototypes are in excellent agreement, thus verifying the proposed theory. Finally, simulation results are presented that characterize the amplifier performance when it is excited by some of the wideband signals used in modern telecommunication systems. The obtained results, which give insight in the linearity of the designed amplifier, are comparable with other results published in the literature, thus they additionally validate the applicability of the newly proposed theory...

Zahvaljujući svojim dobrim performansama i niskoj ceni po jedinici zapremine, gasni detektori su najčešće korišćeni detektori u eksperimentalnoj fizici visokih energija. Pored fizike visokih energija, ovi detektori nalaze primene u mnogim drugim oblastima, poput dozimetrije i zaštite od zračenja, medicine, fizike kosmičkog zračenja i geofizike. Postoje brojni modeli ovih uređaja. Bez obzira da li su analitički ili numerički, stohastički ili deterministrički, svi ovi modeli koriste transportne i/ili sudarne podatke za elektrone u gasovima. Nažalost, uobičajeno je da se unutar detektorske zajednice, ovi podaci kao i tehnike za njihov proračun primenjuju nekritički. Ova praksa, kao što je diskutovano u ovoj disertaciji, veoma često vodi ka pojednostavljenim modelima detektora i neadekvatnoj metodologiji za pristup i analizu. Imajući ove činjenice u vidu, osnovni cilj ove disertacije je da na osnovama transportne teorije rojeva elektrona, ukaže na čitav spektar važnih aspekata u proračunu i implemetanciji transportnih podataka u modelovanju kao i da na konkretnim modelima detektora pokaže kako podaci i njihova implementacija utiču na izračunate signale i performanse detektora. Osim toga, u ovom radu su detaljno analizirane netipične pojave u transportu elektrona indukovane eksplicitnim efektima nekonzervativnih sudara i njihove potencijalne implikacije u modelima. Specifičnosti transporta elektrona u ukrštenom električnom i magnetskom polju, ilustrovane su na primeru detektora tipa Time Projection Chamber (TPC) koji se koriste za trodimenzionalnu rekonstrukciju putanje čestica. Kod ovih detektora magnetsko polje ima ulogu u redukciji bočne difuzije od koje direktno zavisi prostorna rezolucija dok npr. linearnost rekonstrukcije zavisi od osetljivosti brzine drifta na temperaturu i nehomogenost magnetskog polja. U ovom kontekstu potencijalne optimizacije radnih uslova, a takođe imajući u vidu i neželjene varijacije parametara tokom rada, sistematski je razmotren uticaj električnog i magnetskog polja, temperature i pritiska gasa kao i udela nečistoća u gasnoj smeši, na transportne osobine elektrona u TPC detektoru. Posebna pažnja posvećena je detektoru sa pločastim elektrodama visoke otpornosti (Resistive Plate Chamber, RPC) koji se koriste za timing i triggering u brojnim eksperimentima fizike visokih energija kao i u drugim oblastima. Rešavanjem Boltzmannove jednačine i primenom Monte Karlo tehnike, koju smo sa jedne strane koristili da proverimo rezultate dobijene Boltzmannovom jednačinom, a sa druge strane za dobijanje prostorno razloženih transportnih parametara elektrona, identifikovani su iobjašnjeni transportni fenomeni elektrona poput negativne diferencijalne provodnosti i grejanja zahvatom elektrona koji su uočeni u gasnim smešama RPC detektora koji se primenjuju na eksperimentima u CERN-u. U ovoj disertaciji, razmotreni su teorijski principi klasičnog fluidnog modela i modela zasnovanog na difuzionoj jednačini i hidrodinamičkoj aproksimaciji na osnovu kojih je razvijen numerički 1.5-dimenizionalni fluidni model RPC detektora. Ovim modelom, ispitivan je razvoj elektronske lavine i strimera pod dejstvom efekata prostornog naelektrisanja i fotojonizacije u gasu. Pokazano je da nepravilna implementacija transportnih podataka, zanemarivanjem eksplicitnih efektata nekonzervativnih sudara, može dovesti do greške od nekoliko stotina procenata u proračunu indukovanog signala. Razmatrana je i osetljivost izračunatog signala detektora na promene setova preseka za rasejanje elektrona u individualnim gasovima gasnih smeša. Konačno, u ovoj disertaciji razvijen je i mikroskopski Monte Karlo model RPC detektora koji se zasniva na praćenju pojedinačnih trajektorija elektrona i njihovih sudara sa molekulima gasa. Pomoću ovog modela, ipitivana je stohastika elektronskog lavinskog procesa. Takođe, koristeći različite setove preseka za rasejanje elektrona u individualnim gasovima gasne smeše, izračunate su vremenska rezolucija i efikasnost RPC detektora koje se dobro slažu sa eksperimentalnim vrednostima. Owing to their good performance characteristics and low price per unit volume, gaseous particle detectors remained the most commonly used detectors in high energy physics experiments. In addition to high energy physics, these detectors have also found applications in other fields such as radiation protection and dosimetry, medicine, cosmic ray physics and geophysics. A number of methods to model particle detectors have been developed. Being analytical or numerical, stochastic or deterministic, detailed knowledge of electron swarm transport properties as well as reliable cross sections for electron scattering are required as an input in modeling. The highly applied nature of the field, has inevitably driven the modeling of these systems more towards empiricism, and unfortunately often away from its roots in the fundamental transport theory of electron swarms. Thus, the main goal of this work is to bridge the gap between the fundamental transport theory of electron swarms and applications in the field of particle detectors. This goal is achieved by considering many elements of the theory which are important for accurate calculation and correct implementation of electron transport data in modeling. In addition, we provide the examples of specific detector models in which the incorrect implementation of data affects the calculated signal and detector performance characteristics. In this work, we discuss atypical manifestation of electron transport phenomena induced by the explicit effects of non-conservative collisions and potential implications arising from their inclusion in the models. The peculiarities of electron transport in crossed electric and magnetic fields are illustrated using the example of Time Projection Chamber (TPC), a detector employed for three-dimensional reconstruction of particle trajectories. In this detector, the magnetic field suppresses the transverse diffusion, which directly affects the spatial resolution. On the other hand, the reconstruction linearity depends on the sensitivity of drift velocity on the gas temperature and non-uniformity of the magnetic field. With this as one of the motivating factors, and also having in mind the unwanted variation of detector parameters with time, we systematically study the influence of the electric and magnetic fields, gas pressure and temperature, as well as the impact of impurities in the gaseous mixture, on the electron transport properties in TPC. Special attention is given to the Resistive Plate Chambers (RPCs) which are used for timing and triggering purposes in many high energy physics experiments and elsewhere. The Boltzmann equation is used for the determination of electron swarm transport properties under conditions when transport is greatly affected by nonconservative collisions. A Monte Carlo simulation technique has been used with the aim of verifying the results based on the Boltzmann equation as well as for the evaluation of spatially resolved electron transport data. This segment of data is used to explain the existence of certain kinetic phenomena, including negative differential conductivity and attachment heating, which are important for the detector behavior. Within the framework of the classical fluid model and using the diffusion equation in association with the hydrodynamic approximation, we have developed a numerical 1.5-dimension fluid model of an RPC. This model is used to study the electron avalanche and streamer development under the influence of space charge effects and photoionization. We have shown that improper use of the data, especially the lack of correct representation of the explicit effects of non-conservative collisions, can lead to errors of a several hundred percents for the calculated signals. In addition, we discuss the sensitivity of the output detector signals with respect to the sets of cross sections for electron scattering. Finally, in this thesis we present our microscopic Monte Carlo model of RPC based on the tracking of individual electron trajectories and their collisions with the gas molecules. Using this model, we study the electron avalanche fluctuations and the related processes. The detection efficiency and timing resolution are calculated using different sets of cross sections for electron scattering. We have found that our results agree very well with the measured data.

Савремено друштво се у великој мери ослања на електронске сервисе, уређаје и системе. За успешно разумевање начина рада постојећих електронских уређаја и система, као и приликом пројектовања нових, неопходно је спровести одговарајућу електромагнетску (ЕМ) анализу. У већини случајева ЕМ анализу није могуће спровести аналитички, а до решења се тада мора доћи на неки други начин, најчешће нумеричком електромагнетиком. Метода коначних елемената (finite element method, FEM) представља једну од најпопуларнијих техника нумеричке електромагнетике. Са развојем технике и технологије пред FEM се стално постављају све изазовнији захтеви, као што су: ефикасно и тачно моделовање континуално нехомогених неизотропних средина, ефикасна просторна сегментација са високом тачношћу резултата, ефикасно моделовање отворених ЕМ проблема и имплементација прилагођена новим рачунарски системима. Како би се остварили савремени захтеви, основне формулације и имплементације FEM методе, познате од раније, морају се стално унапређивати. У овој дисертацији описано је нумеричко моделовање континуално нехомогених неизотропних средина FEM методом. Тензори параметара континуално нехомогених неизотропних средина добијени су применом теорије трансформационе електромагнетике, а описана је и теорија трансформационе електромагнетике у векторској и тензорској нотацији прилагођеној инжењерима. Како би се проширио дијапазон проблема који се могу решити FEM методом, описано је моделовање отворених ЕМ проблема, а посебно су продискутовани детаљи у вези са апсорпционим граничним условом (absorbing boundary condition, ABC). У циљу постизања резултата високе тачности предложено је ново просторно мапирање великих закривљених коначних елемената. Како би се боље искористили тренутно расположиви рачунарски ресурси, спроведена је и паралелизација делова рачунарског кода FEM методе... Modern society is heavily reliant on electronic services, devices and systems. When analyzing functionality of existing electronic devices and systems, as well as when designing a new one, it is necessary to perform adequate electromagnetic (EM) analysis. In most cases, the EM analysis cannot be performed analytically, i.e., in a closed form, and the solution must then be obtained in an alternative manner, usually employing computational electromagnetics. The finite element method (FEM) is one of the most popular computational electromagnetics techniques. With rapid development of technology, FEM is constantly confronted with increasingly challenging requirements such as: efficient and accurate modeling of continuously inhomogeneous and anisotropic media, efficient meshing for high accuracy results, efficient modeling of open EM problems and programming appropriate for the new computer architectures. In order to achieve these new requirements, basic FEM formulations and implementations, used in the past, must be constantly improved. In this dissertation, we propose, present and evaluate numerical modeling of continuous inhomogeneous and anisotropic media by the FEM. The parameters of continuously inhomogeneous and anisotropic media tensors in the presented models are obtained by transformation electromagnetics. In addition, the transformation electromagnetics is presented in a novel fashion, where vector and tensor notations are adapted for engineers. Moreover, to extend the range of problems that can be solved by the FEM technique, modeling of open EM problems is described with special emphasis on the absorbing boundary condition (ABC). In addition, novel spatial mapping of large curved finite elements for achieving high accuracy of results is also proposed and evaluated. Finally, for more efficient use of currently available computing resources, existing FEM computer code is partially parallelized.

Pojava brzih računskih mašina dovela je poslednjih decenija do pojave niza numeričkih metoda za proračun, odnosno modelovanje elektromagnetskog (EM) polja. Istraživanje novih numeričkih metoda je konstantno aktuelno jer postoji neprestana potreba za poboljšanjem efikasnosti, tačnosti i brzine računanja. Jedna od često i multidisciplinarno primenjivanih metoda je metoda konačnih elemenata (FEM). U skoro čitavoj raspoloživoj literaturi prikazana je tzv. slaba FEM formulacija. Ona se zasniva na funkcijama bazisa koje su izvan domena operatora koji opisuje analizirani problem, a koji je tipično diferencijalni operator drugog reda. Osim toga, najveći broj prikazanih metoda se zasniva na aproksimaciji niskog, odnosno prvog reda. Jaka FEM formulacija je malo zastupljena u literaturi i njoj je posvećen mali broj radova. Prvi radovi iz ove oblasti su pokazali da jaka formulacija može imati određene prednosti u odnosu na slabu. Ove prednosti se ogledaju u koncepcionoj jednostavnosti i jednostavnijem i prirodnijem definisanju graničnih uslova. Kako je razvoj jake FEM formulacije u začetku, prirodno je bilo da se ona prvo razvije za dvodimenzionalne statičke i kvazistatičke probleme. Tipičan primer ovakvih problema su oklopljeni vodovi. Osnovni cilj istraživanja ove doktorske disertacije je teorijsko i praktično zasnivanje jake FEM formulacije (Galerkinova varijanta FEM-a) za zatvorene dvodimenzionalne elektromagnetske probleme, tj. probleme kod kojih su granični uslovi zadati eksplicitno. Osim toga, pojedinačni ciljevi bili su kreiranje softvera na osnovu zasnovane numeričke metode, proučavanje karakteristika nove formulacije i primena nove formulacije na proračun vodova sa TEM talasom. Najpre je dat pregled u literaturi poznatih čvorno zasnovanih funkcija bazisa. Zatim su, polazeći od u literaturi poznatih jednodimenzionalnih funkcija bazisa formirane jake dvodimenzionalne funkcije bazisa. One zadovoljavaju granične uslove prve i druge vrste na granicama pravougaonih konačnih elemenata na koje je izdeljen računski domen. Funkcije bazisa su definisane tako da čine hijerarhijski skup, pri čemu je posvećena pažnja koncepcionoj jednostavnosti. Uočen je najniži red aproksimacije, odnosno minimalan red jakih funkcija bazisa. Proučene su funkcije bazisa za homogene, nehomogene, izotropne, anizotropne i biizotropne sredine. Zatim je jaka FEM formulacija primenjena u kvazistatičkoj analizi oklopljenih vodova, što je ilustrovano na većem broju jednostavnijih primera za koje u literaturi postoje ili analitička rešenja, ili rešenja visoke tačnosti. Dobijeni rezultati su upoređeni sa rezultatima dobijenim drugim numeričkim metodama, kao i sa rezultatima dobijenim slabom FEM formulacijom. Upoređivanjem rezultata dobijenih jakom FEM formulacijom sa rezultatima iz literature, uočeno je da postoji odlično slaganje. Pri tome se zadovoljavajuća tačnost postiže za nizak red funkcija bazisa, obično minimalan, a konvergencija rezultata je stabilna. Kako prikazane funkcije bazisa nisu čvorno zasnovane, povećanje stepena aproksimacije (reda funkcija bazisa) ne unosi dodatne čvorove. To znači da se dodatno ne komplikuje mreţa konačnih elemenata, što predstavlja veliku prednost nad čvorno zasnovanim funkcijama bazisa. Upoređivanjem rezultata dobijenih jakom formulacijom sa rezultatima dobijenim slabom formulacijom pokazalo se da jake funkcije bazisa, za istu tačnost rešenja, zahtevaju daleko manji broj nepoznatih. Dalje je izvršen pregled i kritička analiza formulacija FEM-a za otvorene probleme. Zatim je predložena nova, hibridna FEM-MoM metoda za otvorene statičke probleme zasnovana na specijalnim funkcijama bazisa na granici računskog domena i primeni Green-ovog identiteta. Specijalne funkcije bazisa su iz skupa jakih funkcija bazisa. Predložena metoda ne zahteva, barem ne eksplicitno, uvođenje novih nepoznatih. Na kraju su analizirani mogući dalji pravci istraživanja. The advent of powerful computing machines has in recent decades led to the emergence of numerous numerical methods for electromagnetic (EM) field computation and modeling. The research of new numerical methods perpetually constitutes a mainstream activity, since there is a continuous need for improvement of efficiency, accuracy and computation speed. One of frequently and multi-disciplinary used methods is the finite element method (FEM). In almost all available literature, the so called weak FEM formulation is described. It utilizes basis functions that are outside the domain of the operator describing the analyzed problem, which is typically a second-order differential operator. Moreover, the largest number of existing methods is based on the low order, i.e., first-order or second-order, approximations. Strong FEM formulation is underrepresented in the literature, and it is pursued in a small number of papers. The first papers in this field showed that the strong formulation may have certain advantages over the weak formulation, manifested in conceptual simplicity and easier and more natural definition of boundary conditions. Because the development of strong formulations is in its beginning, a natural choice was to start with a two-dimensional static and quasi-static problems. Typical examples of these problems are shielded transmission lines. The main aim of this PhD thesis is theoretical and practical establishment of the strong FEM formulation (Galerkin version of FEM) for closed two-dimensional electromagnetic problems, i.e., problems for which the boundary conditions are given explicitly. Additional objectives were to create the software based on the established numerical methods, study the features of the new formulation, and apply the new formulation in analyses of TEM-wave transmission lines. Firstly, an overview of the literature of the known node-based basis functions is given. Then, starting from the known literature on one-dimensional basis functions, two-dimensional basis functions are created. They satisfy the boundary conditions of the first and second kind on rectangular finite element borders into which the computational domain is segmented. The basis functions are defined as a hierarchical set, whereby attention is given to their conceptual simplicity. The lowest order, i.e., the minimal order for the strong formulation, approximation is observed. Basis functions for homogeneous, inhomogeneous, isotropic, anisotropic, and biisotropic media are studied. Afterwards, the strong FEM formulation is applied to the quasistatic analysis of shielded cables, which is illustrated by a number of simple examples for which analytical solutions or solutions of high accuracy exist in the literature. The obtained results are compared with numerical results obtained by other methods, as well as with the results obtained by the weak FEM formulation. Good agreement of the results obtained by the strong FEM formulation and the results from the literature is observed. At the same time, satisfactory accuracy is achieved using low-order basis functions, usually minimal, and good convergence is obtained. As the proposed basis functions are not node-based, increasing the degree of approximation (increasing the order of basis functions) does not introduce additional nodes. Hence, no additional complexness is introduced in the finite element mesh, which is a big advantage over the node-based basis functions. Comparing the results obtained by the strong FEM formulation with the results obtained by the weak FEM formulation, it is proved that strong basis functions require far fewer unknowns for the same accuracy. Subsequently, an overview and critical analysis are given of the FEM formulation for open problems. Then, a new hybrid FEM-MoM method for open static problems is proposed, based on special basis functions on the border of both computational domains and application of Green’s identity. These special basis functions belong to a set of strong basis functions. The proposed method does not require introduction of new unknowns, at least not explicitly. In the end, possible directions for future research are analyzed.

Potreba za bežicnim komunikacioniom linkovima velikih brzina prenosa podataka je podstaknuta ekspanzijom prenosivih uređaja i multimedijalnih servisa, uz pogodnost da priroda korišcenja dozvoljava a ponekad i zahteva ogranicen domet. Problem kapaciteta komunikacionih linkova i sve veceg broja korisnika se može rešiti prelaskom u opseg ucestanosti od 30 do 300 GHz, koji se naziva i milimetarski opseg. Visoka radna ucestanost pruža mogucnost korišcenja kanala velikog kapaciteta, kao i fizicki malih antenskih nizova za fokusiranje i prostornu lokalizaciju prijemnog i predajnog snopa. Milimetarski opseg nalazi primene i u ostalim oblastima, kao što su industrijske, medicinske i bezbednosne. U komercijalnim primenama od interesa je opseg ucestanosti oko 60 GHz, koji je dodeljen za nelicenciranu upotrebu širom sveta. Razvoj CMOS i BiCMOS tehnologija je omogucio da se sistemi u 60 GHz-om opsegu mogu integrisati u standardnim procesima. Pored viših radnih ucestanosti, skaliranje tehnologija uvodi i tehnološka ogranicenja koja degradiraju performanse ukoliko se njihov uticaj zanemari. Zanemareni efekti mogu doprineti vecim gubicima, koji povecavaju faktor šuma prijemnika i degradiraju efikasnost predajnika, ali i parazitnim preslušavanjima koja rezultuju neželjenim spektralnim komponentama. Stoga je potrebno razmotriti kvalitativne i kvantitativne pokazatelje uticaja tehnoloških ogranicenja na performanse i prilagoditi postupak projektovanja. Kriticni blokovi za domet primopredajnika su malošumni pojacavac na prijemnoj strani i pojacavac snage na predajnoj strani. U okviru teze predstavljen je postupak projektovanja malošumnog pojacavaca i pojacavaca snage za rad u 60 GHz-om opsegu i širokopojasnog delitelja ucestanosti. Uvedene su nove smernice projektovanja koje uzimaju u obzir tehnološka ogranicenja. Pokazano je da se pravilnim particionisanjem elektromagnetskog modela može postici dobro slaganje rezultata simulacije i merenja. Projektovana kola su fabrikovana u IHP Microelectronics korišcenjem 0.25 mm SiGe:C BiCMOS procesa (fT/fmax = 200 GHz). Parametri fabrikovanih kola su izmereni i verifikovani na stopicama cipa, upotrebom mikrotalasnih sondi... The need for high capacity wireless data links is driven by expansion of mobile devices and multimedia services, with the advantage that a typical use case allows, and sometimes demands, a limited range. Problems of limited communication link capacity and growing number of users can be solved by moving to frequency range of 30 - 300 GHz, also known as millimeter range. High operating frequency allows the use of high capacity channels, and physically small antenna arrays for beam steering and spatial localization. Millimeter region of spectrum is also suitable for industrial, scientific and security applications. Unlicensed 60 GHz band is available worldwide, and is attractive for commercial applications. Development of CMOS and BiCMOS technologies has enabled the integration of complete 60 GHz systems in standard processes. Technology scaling enables the use of higher operating frequencies, but imposes new design constraints which may degrade the performance if their effect is neglected. Neglected effects may contribute to higher losses, which increase the noise figure of receiver and degrade transmitter efficiency, and also to parasitic coupling which results in undesired spectral components. Therefore, qualitative and quantitative measure of technology constraints impact on performance degradation needs to be evaluated, and applied to circuit design process. Critical blocks for transceiver range are low noise amplifier on receiver, and power amplifier on transmitter side. Design procedures for 60 GHz low noise and power amplifiers, and wideband frequency divider are presented in this thesis. Guidelines for technology constraints aware design are used in the presented design flow. Good agreement of experimental and simulation results is achieved by proper electromagnetic model partitioning. Designed circuits have been fabricated in IHP Microelectronics 0.25 mm SiGe:C BiCMOS process (fT/fmax = 200 GHz). Test chip parameters have been measured and verified on-wafer by using microwave probes...

Analiza hoda je postala široko rasprostranjen klinički alat koji se koristi za objektivnu evaluaciju obrasca hoda, efekata hirurških intervencija, oporavka ili efekata terapije. Sve veći broj kliničara bira pogodne tretmane za lečenje pacijenata na osnovu informacija o kinematici i kinetici hoda. Procena i kvantifikacija parametara hoda je važan zahtev u oblasti ortopedije i rehabilitacije, ali takođe i u sportu, rekreaciji i posebno u razvoju tehnologija za ljude u procesu starenja. U cilju objektivne procene obrasca hoda, razvijen je bežični senzorski sistem čije su senzorske jedinice bežične, malih dimenzija i jednostavno se montiraju na segmente nogu subjekta čiji se hoda analizira. Senzorske jedinice podržavaju 3D inercijalne senzore (senzore ubrzanja i ugaonih brzina, tj. akcelerometre i žiroskope), kao i senzore sile. Osnovni cilj istraživanja je doprinos metodologiji za obradu podataka sa inercijalnih senzora i razvoj novih metoda obrade signala sa inercijalnih senzora u procesu određivanja kinematičkih veličina koje su uobičajene u analizi hoda (uglovi u zglobovima, brzina kretanja, dužina koraka). Ova metodologija je od posebne važnosti za objektivnu procenu nivoa motornog deficita, progresa bolesti i efikasnosti terapija, kao i efikasnosti primenjene motorne kontrole (prilikom funkcionalne električne stimulacije). U toku istraživanja razvijeno je nekoliko metoda za računanje uglova segmenata nogu ili zglobova, u zavisnosti od senzorske konfiguracije i složenosti algoritma. U disertaciji su odvojeno prikazani slučajevi u kojima je neophodno posmatrati kretanje u prostoru (3D analiza) i mnogo češći slučaj kad se kinematika može redukovati na sagitalnu ravan (2D analiza). Algoritmi uključuju i kalibraciju senzora, eliminaciju viii drifta, rekonstrukciju trajektorije i izračunavanje niza drugih relevantnih podataka koji karakterišu obrazac hoda. Dobijeni rezultati su poređeni sa postojećim sistemima za analizu hoda koji su validirani za kliničke primene. (sistemi sa kamerama, goniometri, enkoderi)... Gait analysis has become a widely used clinical tool which provides objective evaluation of the gait pattern, the effects of surgical interventions, recovery or therapy progress, and more and more clinicians are choosing therapy treatments based on gait kinematics and kinetics. Measuring gait parameters is an important requirement in the orthopedic and rehabilitation fields, but also in sports and fitness, and development of technologies for elderly population. In order to provide objective evaluation of the gait pattern, we have developed sensor system with light and small wireless sensor units, which can be easily mounted on body. These sensor units comprise 3-D inertial sensors (accelerometers and gyroscopes) and force sensing resistors, and our recommended setup includes one sensor unit per each segment of both legs. The main goal of this research is contribution to the methodology for processing of signals from inertial sensors (accelerometer pairs, or accelerometer and gyroscope sensor units). By using signal processing algorithms developed for this research, inertial sensors allow objective assessment of the quality of the gait pattern. This methodology is especially important for assessment of the motor deficit, progress of the disease and therapy effectiveness, and effectiveness of performed motor control (functional electrical stimulation). We have developed several methods for estimation of leg segment angles and joint angles, which differ in sensor configuration and algorithm complexity. Methods based only on accelerometers offer reliable angle estimations, which are limited to sagittal plane analysis, while the method using accelerometers and gyroscopes allows 3- D analysis. All this algorithms include sensor calibration, drift minimization, trajectory x reconstruction and calculation of numerous other parameters relevant to gait pattern analysis. The obtained results were compared with other commercial systems which are validated for clinical applications (camera systems, goniometers, encoders)...

Нумерички егзактно решавање интегралних једначина електромагнетског поља могуће је методом момената (МоМ). Када се примени на електрично велике проблеме (нпр. летелице), МоМ резултује великим системима линеарних једначина, чије решавање представља уско грло методе. Проблем је двојак: рачунарска меморија потребна за смештање матрице система је велика (сразмерна квадрату броју једначина), а рачунарско време решавања дуго (сразмерно трећем степену броја једначина, у случају решавања LU декомпозицијом). Један начин (делимичног) превазилажења овог проблема јесте паралелизација решавања система линеарних једначина. Корисницима обичних („персоналних“) рачунара на располагању су решења заснована на коришћењу вишејезгарних централних процесора или графичких процесора. На овај начин може се значајно скратити време решавања система линеарних једначина, али не и меморијско заузеће. У сваком случају, примена МоМ-а на персоналним рачунарима има јасне границе. Развијене су и бројне апроксимативне методе, између осталог засноване и на МоМ-у, за решавање електрично великих проблема. Свака од њих има своје предности и недостатке, али се ни за једну не може рећи да је универзална у погледу решавања произвољних проблема. Циљ нашег истраживања био је развијање нове апроксимативне методе, засноване на МоМ-у, за решавање интегралних једначина електромагнетског поља. Нова метода требало је да, на персоналном рачунару, омогући решавање проблема за један ред веће електричне величине од оних који се могу решити применом МоМ-а. Нова метода реализована је као итеративни поступак у којем се правац корекције одређује модификованом PO техником, а резултујући систем линеарних једначина је, за електрично велике проблеме, за више редова величина мањи од одговарајућег МоМ система. Нумерички експерименти су показали да је време потребно за добијање решења (задовољавајуће тачности) значајно краће од оног потребног за добијање МоМ решења, док је меморијско заузеће занемарљиво у односу на оно код примене МоМ-а. Електрична величина проблема који се могу анализирати персоналним рачунаром повећана је за најмање ред величине у односу на МоМ... Numerically exact solution of electromagnetic field integral equations is possible using Method of Moments (MoM). When applied to electrically large problems (e.g. airplanes), MoM results in large system of linear equations. Solving it is a bottleneck of the method, for two reasons: occupied computer memory is large (proportional to number of equations squared) and computer solution time is long (proportional to third power of number of equations, when using LU decomposition solver). A way to (partially) circumvent this problem is parallelization of solution process. Personal computer (PC) users have at their disposal software solutions based on multi-core CPUs and GPUs, which are able to significantly decrease solution time, but memory occupation stays intact. Anyway, application of MoM on PCs has clear limitations. Different approximate methods, many of them based on MoM, are developed for treating electrically large problems. Each of them has its own advantages and shortcomings, but neither is universal in respect of treating an arbitrary electromagnetic problem. The goal of our research is development of new approximate method, based on MoM, for solution of electromagnetic field integral equations. The new method should enable solution of the order of magnitude electrically larger problems than MoM. The new method is formulated as iterative procedure in which the direction of correction is determined using modified PO technique. For electrically large problems the resulting system of linear equations is much smaller than the corresponding MoM system. Numerical experiments show that (accurate enough) results can be obtained in much shorter time than with MoM, whereas memory occupancy can be neglected when compared to those of MoM. Electrical size of problems that can be analyzed using PC is increased by (at least) an order of magnitude...