Juan Rivas (with Prof. David Perreault)

New architectures for switched-mode dc/dc power conversion - The proposed architectures enable dramatic increases in switching frequency to be realized while preserving features critical in practice, including regulation of the output across a wide load range and high light-load frequency. This is achieved in part by how the energy conversion and regulation functions are partitioned. The structure and control approach of the new architectures are described, along with representative implementation methods. The design and experimental evaluation of prototype systems with cells operating at 100 MHz are also described. It is anticipated that the proposed approaches will allow substantial improvements in the size of switching power converters and, in some cases, to permit their integrated fabrication.

Assessing Cardiovascular Function using Electrical Circuit Analogs - Accounting for an estimated 2,500 deaths/day, cardiovascular disease (CVD) is the number one cause of death in the United States. With 34% of the American population having one or more types of CVD, monitoring the health of the heart and blood vessels is of prime importance, both on Earth and en route to Mars. In my presentation, I will introduce our group's recent work on circuit-based computational models of the cardiovascular system with particular emphasis on patient monitoring in intensive care units and on the monitoring of astronaut health during and after space travel. (Presentation PDF 2.4MB)

Photonics crystals for IR and near-IR applications:
thermophotovoltaics, thermal radiation sources, IR sensors and beyond - Photonic crystals are, in simple terms, meta-materials with periodic dielectric constant having periodicity that is on the order of the wavelength of light that is of interest. Although photonic crystals have been extensively studied for the last 20 years, most of the focus was on optical comunications and micro-photonics. In this talk we will try to shed some light on our research quest towards photonic crystals that can be used to generate, guide, filter and detect thermal radiation in near-IR and IR range. In terms of applications, we will focus on thermophotovoltaic power generation, thermal radiation sources and IR sensors and we will show the impact photonic crystals can have on performance of these devices. Furthermore, we will introduce the new class of partialy coherent, quasi-monochromatic resonant thermal radiation sources that are starting to change the common understanding of thermal radiation sources as isotropic incoherent radiation sources. (Presentation PDF 1.7MB)

Control of Stand-Alone Induction Generators - Most industrial motors are squirrel cage induction machines because of their simple and robust construction, low cost, minimal maintenance, and inherent overload protection. However, induction generators are much less widely used because the drive speed, electrical frequency, voltage, load, and equivalent terminal capacitance must be juggled to provide both the reactive excitation power to the machine and the varying real power to the load. The present investigation examines the use of high-speed induction generators to supply electrical power for ships and submarines. The project includes design and simulation of a 5 MW machine and verification using a 2kW demonstration model. This presentation will examine issues including initial excitation, a computer aided machine design strategy, methods to supply reactive power, and reactive power flow in a six-step inverter. (Presentation PDF 1.2MB)

Laura Zager (with Prof. George Verghese)

Graph similarity - Graph similarity measures have a broad array of applications, including comparing chemical structures, navigating complex networks like the Internet and WWW, and recently, analyzing different kinds of biological data. This talk will survey some interesting graph-theoretic applications and several different notions of similarity, then focus on a very intuitive class of algorithms that use the similarity of local neighborhoods to derive pairwise similarity scores between graph elements. We've developed an extension to these algorithms, and are exploring the application of this approach to the task of graph matching; some preliminary performance results will be presented. (Presentation PDF 0.6MB)

An ElectroQuasiStatic Induction Turbine-Generator - We present a microfabricated electroquasistatic (EQS) induction turbine-generator that has generated net electric power. A maximum power output of 192uW was achieved under driven excitation. We believe that this is the first report of electric power generation by an EQS induction machine of any scale found in the open literature. We also present self-excited operation in which the induction generator self-resonates and generates power without the use of any external drive electronics. The generator comprises five silicon layers, fusion bonded together at 700C. The stator is a platinum electrode structure formed on a thick (20um) recessed oxide island. The rotor is a thin film of lightly doped polysilicon also residing on an oxide island, 10um thick. We also present a generalized state-space model for an EQS induction machine that takes into account the machine and its external electronics and parasitics. This model correlates well with measured performance and was used to find the optimal drive conditions for all driven experiments. (Presentation PDF 1MB)

Ultrafast Tool Servos for Nano-surfaces - Fast Tool Servo (FTS) technology plays important roles in precisely manufacturing complicated free-form surfaces for use in modern optics, fusion experiment targets, metal molds for light enhancing films, and so on. The increasing complexity of surfaces requires more components in shorter spatial wavelengths, and thus drives simultaneously the need for high bandwidth, high acceleration and high accuracy of the FTS. However, the conventional piezoelectric stack-based micro positioning devices are not suitable for these requirements because the hysteresis-generated heat and the structural resonance limit their performance. Normal stress electromagnetic actuation mechanism promises to achieve higher acceleration at high frequencies combined with high bandwidth. I have designed, implemented, and tested such an electromagnetically driven fast tool servo with a stroke of 30mm, based on a novel ultrafast motor concept. Experimental results demonstrate that the ultrafast tool servo achieves 23kHz closed-loop bandwidth, as low as 1.7nm RMS error, 500G peak acceleration at 10kHz open-loop operation, and 2.1nm (0.04%) error in tracking a 3kHz sinusoid of 16mm p-v. This talk will describe the electromagnetic design, operation principle, magnetic material consideration, mechanical design, and experimental results. To drive and control this ultrafast tool servo, a 1kW linear power amplifier and a high-speed real-time computer with 1MHz sampling rate have been designed and implemented, which will also be presented in this talk. Project Webpage

42 Volts: The View from Today - A few years ago, the automobile industry agreed to adopt standards for a new voltage for the production and use of electrical power. The perception was near universal that 14 Volts was at the limits of its capability, and that 42 Volts would be adopted in a rush. The universal perception was wrong. Since then, much of the auto industry has encountered hard financial times. In a totally separate development, parts suppliers introduced innovations at 14 Volts, some of which a few years ago were thought to require 42 Volts. Today, there are 42-Volt cars and trucks for sale, but only at numbers far lower than necessary to begin to achieve economies of scale. But the factor which caused the industry to develop the 42 Volt standard, the growth of electricity use on motor vehicles, continues with no sign of letup. Further, the true technical obstacles to adoption of 42 Volts have been discovered and at least provisionally solved. The way forward to cost-effective solutions for advanced automobiles is clearer today than it was in the past. (Presentation PDF 0.2MB)

Abstract - The main focus of this work is to explore the dynamic behavior of an auction system for an electricity market. In order to cope with this complex problem, agent-based simulation has been previously used, where autonomous agents learn through the results of repeated auctions. In this paper, the replicator equation is introduced as a learning algorithm that can be applicable to agent-based simulation. A repeated electricity auction based on a Japanese power market is simulated with different strategy selection algorithms incorporating mechanisms for learning. The results are compared to show the applicability of the replicator equation.

Abstract - Electrochemical battery models have long been the domain of electrochemists. Electrical engineers, however, have primarily used some sort of equivlalent circuit model. However, with the recent interest in the propossed 42V electrical standard for automobiles, there has been increase interest in reliably modelling the performance of the automotive electrical system. In particular, there has been great interest among automotive electrical engineers to increase their understanding of the internal operation of the lead acid battery.
In this presentation, a simplified one volume per plate electrochemical model for a lead acid cell is developed. The simplifications have been made in such a way that an electrical engineer with a rudimentary understanding of chemistry should be able to understand the source and importance of each of the physical phenomena used to model the electrochemistry of the battery cell. The simplified model is then developed into a complete one dimensional model and then further developed into a 2 dimensional model. Time permitting, how this model can be adapted to the Lithium-Ion chemistry will also be presented.

A Markov Model-Based Analysis and Design Methodology for Fault-Tolerant Systems - In Safety-critical systems, such as nuclear power plants, aircraft, some automotive applications, and the space shuttle, it is imperative that the system function be performed, even in the presence of fault conditions. Design of such systems often involves trade-offs between multiple architectures. The comparative reliability of these architectures must be carefully assessed during the design phase. There are several different methodologies to analyze and compare the reliability of safety-critical systems, such as fault tree analysis, event tree analysis, or Failure Modes and Effects Analysis. With these methodologies, it is possible to compute and assign probabilities to the events of interest, in order to make a decision among different design solutions. However, problems arise when redundancy, reconfiguration schemes or common cause failures appear in the design. With traditional methodologies, it is difficult to include all these conditions. Although not perfect, Markov models provide the capability to handle all the aforementioned problems. In this presentation, an introduction to reliability analysis using Markov modeling is introduced. A Methodology for Fault-Tolerant system design optimization, based on Markov models is also presented. As a case study, a Power-Net architecture for application to new safety-critical automotive systems, e.g., steer-by-wire or brake-by-wire is introduced and analyzed.

Hilbert Space Approach to Modeling and Compensation of Reactive Power - In this talk we describe a Hilbert space approach that provides a unified exposition of various definitions of reactive (or inactive) power and of various compensation methods for systems with an arbitrary number of phases. The talk casts several compensation strategies (such as compensation without energy storage and compensation with linear shunt components) in a common framework, and utilizes the concept of orthogonal projections on suitable subspaces as the main analytical and computational tool.
We describe a physics-based methodology for decomposing the current (and consequently the apparent power) into mutually orthogonal components. The same set of tools can be used to quantify the tradeoff between line losses and compensator cost. It turns out that we can formulate and solve an optimization problem related to reactive power compensation with limited current bandwidth, where our solution generalizes some results from the literature. We deal with unequal phase resistances by introducing the notion of a weighted inner product. Our concept of reactive power can also be naturally extended to system transients, and it reduces to well known quantities in steady state.

Exploration, Processing, and Visualization of Physiological Signals from the Intensive Care Unit (Renjifo) - Physiological signals play an essential role in modern-day health care. Each day thousands of lives are improved by treatments based on the assessment of these signals. Even now, there is much we can still learn from physiological data to improve the quality of information given to doctors and nurses.
In this presentation we discuss some of our work in the areas of signal visualization and processing. We explore the application of phase space plotting and power spectrum analysis to cardiovascular data and introduce a novel method for data clustering based on the singular value decomposition. These techniques are applied to real data measured from a patient in the intensive care unit (ICU).

Cardiovascular Parameter Estimation using a Computational Model (Samar) - Modern intensive care units are equipped with patient monitoring devices, each continuously recording signals produced by the human body. Currently, these signals need to be interpreted by a clinician in order to assess the state of the patient, to formulate physiological hypotheses, and to determine treatment options. In some areas, such as hemodynamic monitoring, there is enough quantitative information available to formulate computational models capable of simulating normal and abnormal human physiology. Through tuning, such models could be used to track patient state automatically and to relate properties of the observable data streams directly to the properties of the underlying cardiovascular system.
The focus of our investigation is to use a cardiovascular model and to match its output to observable hemodynamic signals in order to estimate cardiovascular parameters. Tracking model parameters in time reveals disease progression, and hence can be very useful for patient monitoring purposes. As the observable signals are generally not rich enough to allow for the estimation of all the model parameters, we employ subset selection to identify those parameters that can be estimated robustly. Using simulated data at both intra- and inter-beat timescales, we were able to track clinically important cardiovascular parameters with reasonable accuracy.

Development of Three-Dimensional Passive Components for Power Electronics - As component and power densities have increased, printed circuit boards (PCBs) have taken on additional functionality including heatsinking and forming constituent parts of electrical components. PCBs are not well suited to these tasks. A novel fabrication method is proposed to develop an enhanced circuit board fabrication approach which overcomes this problem. This method uses a photoresistive scaffold and subsequent metal deposition to realize the proposed structures. These structures might be suitable as heatsinks, inductor windings, busbars and EMI shields among other applications. An application in heatsinking for power electronics has been explored in some detail and the results were not what was expected. (Presentation PDF 0.9MB)

Abstract - Semiconductor-device limitations to system miniaturization have receded, but exposed by their improvement numerous "ancillary" barriers which continue to preoccupy nearly every electronics industry. Prominent among these obstacles are package parasitics and heat, which have come to the fore as conventional circuits are applied in modern regimes of frequency and integration density. To an ever increasing extent, integration limits are symptoms of the fundamental frequency- and size-scaling limits of passive components. Power inductors and transformers, in particular, are challenging to miniaturize because of their poor performance when scaled down in size, and the difficulty of fabricating them with available planar processes.
A family of approximating networks for transmission lines, the focus of this work, enables miniaturization by internally circulating energy and exchanging delay fidelity for bulk energy storage. These multi-resonant components are substantially smaller than their lumped counterparts, in particular requiring less inductance, and enforce useful waveform symmetries that can be traded for higher power or higher efficiency. Lumped analogs of transmission lines, and delay-based means of processing energy in general, exploit rather than fight the parasitics which can restrict conventional designs to lower switching frequencies, and are compatible with RF power-conversion techniques.
Printed-circuit and wafer- or package-scale construction methods for multi-resonant structures will be presented, along with power-converter topologies that exploit the waveform symmetries they enforce. A new soft-switched RF power converter is introduced, in particular, that demonstrates reductions in peak device stress and passive-component size. Taken together, the construction techniques, networks, and converter topologies presented here extend the power levels and applications for which passive components can be manufactured in an integrated fashion, within a printed circuit board or at the die/package scale alongside semiconductor switches and converter controls.

An ElectroQuasiStatic Induction Turbine-Generator - We present a microfabricated electroquasistatic (EQS) induction turbine-generator that has generated net electric power. A maximum power output of 192uW was achieved under driven excitation. We believe that this is the first report of electric power generation by an EQS induction machine of any scale found in the open literature. We also present self-excited operation in which the induction generator self-resonates and generates power without the use of any external drive electronics. The generator comprises five silicon layers, fusion bonded together at 700C. The stator is a platinum electrode structure formed on a thick (20µm) recessed oxide island. The rotor is a thin film of lightly doped polysilicon also residing on an oxide island, 10um thick. We also present a generalized state-space model for an EQS induction machine that takes into account the machine and its external electronics and parasitics. This model correlates well with measured performance and was used to find the optimal drive conditions for all driven experiments.

Magnetic Machines and Power Electronics for Power MEMS Applications - Modern battery technologies have not kept pace with the rising demand for power by portable electronic devices. This has led to the need for alternative power sources such as MEMS-based electric generators that can produce 10­100 W of electrical power. One potential solution is a silicon micromachined gas turbine engine coupled to an electric generator. In order to produce high output power the electric generator must support high rotor spin (~1 Mrpm) and tip (500 m/s) speeds. In addition, it must be able to operate at high temperatures (~300°C) within the micromachined gas turbine engine.
This thesis presents the modeling, design, and characterization of microfabricated, surface wound, permanent-magnet generators and power electronics capable of generating Watt-level electric power for use in power MEMS applications such as the micro-gas turbine engine. The generators are three-phase, axial-flux, synchronous machines, comprising a rotor with an annular PM and ferromagnetic core and a stator with multi-turn surface windings on a soft magnetic substrate (stator core). Both passive and active power electronics have been built and tested. The passive power electronics consist of a three-phase transformer and diode bridge. The active power electronics consist of a switch-mode rectifier based on the boost semi-bridge topology which is used to convert the generators unregulated AC voltage to a regulated DC voltage of 12 V.
At a rotational speed of 225,000 rpm, one permanent-magnet generator demonstrated 9 W of mechanical-to-electrical power conversion and, coupled to the transformer and diode bridge, delivers 4.55 W of DC electrical power to a resistive load. This is the first ever demonstration of Watt level power generation and delivery by a microscale electric machine and is the highest output power ever delivered by a power MEMS device to date. At a rotational speed of 210,000 rpm, the switch-mode rectifier is able to deliver 3.7 W to a resistive load at a regulated output voltage. This Watt-scale electrical power generation demonstrates the viability of scaled PM machines and power electronics for practical applications.

Enhanced Next Generation Alternator - The power requirements in automobiles are increasing significantly due to the addition of new loads. This thesis makes several improvements in output power and efficiency of the alternator system while maintaining alternator size. The transition to the 42 V system introduces new concerns regarding the load dump overvoltage. This thesis studies the problem and develops circuit topologies that limit the load dump transient.
A recent development in alternator design has been the use of a switched-mode rectifier that enables load matching for improved power and efficiency. A design optimization of four alternator types is conducted with and without the switched-mode rectifier to obtain the least cost alternator that meets the challenging requirements of future automobiles.
Several enhancements are developed that enable improvements in output power and efficiency. The round wire field winding is replaced with copper foil, exploiting an increase in packing factor. This enables improvements in output power and efficiency. Several field circuit topologies are developed that accommodate the use of the foil winding and allow for low cost fast field de-excitation, which would be crucial to load dump protection. The first class of field circuits involves the replacement of the existing voltage regulator with a circuit employing a rotating transformer that provides contactless power transfer to the field winding without the need for brushes. Removal of the brushes improves efficiency and alternator life. The second class of field circuits utilizes a DC/DC converter on the rotor that creates a step up in current provided to the field winding.
The rotating transformer topology is designed, built and tested at standstill with a foil wound bobbin. A printed circuit board (PCB) transformer is used, which covers less space, has minimal cost, and is easily repeatable. A significant improvement in the field winding ampere turn excitation is achieved. The rotating DC/DC converter is designed, built, and embedded in a foil wound alternator with modified brushes. Considerable improvements in field ampere turn excitation and output power are achieved.

Synchronization in Stochastic Networks of Nonlinear Oscillators - Abstract: In this work, we study synchronization of complex random networks of nonlinear oscillators, with specifiable expected degree distribution. We review a sufficient condition for synchronization and a sufficient condition for desynchronization, expressed in terms of the eigenvalue distribution of the Laplacian of the graph and the coupling strength. We then provide a general way to approximate the Laplacian eigenvalue distribution for the case of large random graphs produced by a generalization, of the Erdos-Renyi model. Our approach is based on approximating the moments of the eigenvalue density function. The analysis is illustrated by using a complex network of nonlinear oscillators, with a power-law degree distribution.

An Electromechanical Valve Drive - Abstract: In traditional internal combustion engines, valve timing is fixed relative to crankshaft angle and piston position. If instead engine valve timing were flexibly and independently controlled, we could achieve significant improvements in fuel economy, engine performance, and emissions and obtain optimized engine performance at any load and speed conditions. Based on a previously proposed and conceptually proved electromechanical valve drive (EMVD), which incorporates a shear-force actuator driving the valve-spring system through a nonlinear mechanical transformer (NMT), a substantial amount of progress in modeling, control and NMT design has been made to reduce power consumption, peak torque and transition time. Experimental results also show small seating velocities. A second prototype with a much smaller motor is being set up and experiments will be conducted in the near future. (Presentation PDF 1.4MB)

Modeling and Monitoring of Cardiovascular Dynamics in the Intensive Care Unit - Lumped-parameter time-varying electrical circuit analogs for cardiovascular systems are frequently used in computational models for simulating and analyzing hemodynamics. These pulsatile models provide details of the beat-by-beat or intracycle dynamics. In other settings, however, such as when monitoring a hospital patient's hemodynamic state over time, it is more useful to track trends in the intercycle dynamics. In this presentation, we apply a cycle-averaging method to a simple pulsatile cardiovascular model to derive a cycle-averaged model for cardiovascular dynamics. (Presentation PDF 0.4MB)

Abstract - This research focuses on the enhancement of electroquasistatic and magnetoquasistatic nondestructive evaluation techniques. The terminals of the sensors involved are connected to conductors which are traditionally located on a single plane and have a spatially-periodic structure. The sensor operates as a two-port device with one conductor used to excite the sensor and a second conductor used to sense the response to test materials.

Fault detection and diagnostics for packaged rooftop air conditioners using nonintrusive power measurements - Over the last century, air conditioning has become increasingly common in both homes and businesses. A large segment of this market is occupied by split air conditioners, also known as rooftop air conditioning units (RTUs), as they are very effective for many small- to medium-size cooling loads and are relatively inexpensive when compared to other mechanical cooling options. As their popularity has soared, associated issues such as trends in their reliability and efficiency over the equipment lifespan have become more important and relevant to energy- and cost-sensitive owners.
This talk will examine a variety of techniques used to evaluate the performance of RTUs via field-based measurements. We will discuss the present state of research into developing fault detection and diagnostic procedures for this equipment, and we will focus on techniques currently being developed in LEES using the non-intrusive load monitor, or NILM. Some preliminary results indicating the efficacy of these new methods will also be presented. (Presentation PDF 1.8MB)

Space-charge: ebeams, insulators, batteries and ultra-capacitors - Space charges are an important component in the operation of electronic devices such as semiconductors, capacitors, and in most high voltage apparatus. Energy storage devices such as batteries, fuel cells and ultra capacitors rely heavily on space charge phenomena at their surfaces and in their electrolytes. Antistatic materials and filters rely on space charge effects in solid materials. Electron and ion beams used for processing, imaging and medical therapy inherently involve space charges and can cause significant charge accumulations in various beam targets. Charge accumulations in materials in the space environment pose hazard issues for space flight systems. And of course biological systems with cells, membranes and neurons are fundamentally space charge 'devices' as well.
Since they are difficult to directly detect and measure, space charges are often an inferred quantity from theory and terminal current measurements. However, new ultrasonic technologies have opened the door to direct charge imaging and in-turn such measurements reveal new details about the internal space charge conditions in materials and devices. This talk will discuss issues about space charges and their measurement and will give examples from recent research findings.

Low Voltage DC-DC Converters with No-Deadtime Capabilities - Applications of DC-DC converters require fast transient response, stringent EMI criteria, and high power density. Simply increasing switching frequency to improve the transient response is not always applicable to satisfy the design requirement because of the accompanying deterioration of the efficiency. This research presents the concept of no deadtime operation to keep the energy transmission smoothly from input to output. There are several benefits for a DC-DC converter to operate with no deadtime. For example, smaller filter inductance, faster output transient response, lower input ripple current, lower RMS of the current in transformer windings etc.
Based on these benefits, several new no-deadtime topologies will be introduced, all of which rely on the concept of introducing a switching cell with capacitors into a converter.

Natalija Jovanovic (with Prof. John Kassakian)
Photonic Crystals for Thermophotovoltaic Applications - This research investigates the use of two-dimensional (2D) photonic crystals (PhC) as selective emitters and means of achieving higher efficiencies in combustion-driven thermophotovoltaic (TPV) systems. The introduction of 2D photonic crystal selective emitters has potential of doubling the efficiency of current TPV systems. A review of TPV principles and history will be provided, along with and introduction to photonic crystals. The structure requirements and the development of a selective emitter fabrication process for high-efficiency TPV systems will be described. White light diffraction from 2D PhC sample (1MB) (Presentation PDF 1.3MB)

Resistance Compression Networks for Radio-Frequency Power Conversion - A limitation of many high-frequency resonant inverter topologies is their high sensitivity to loading conditions. This presentation will introduce a new class of matching networks that greatly reduces the load sensitivity of resonant inverters and radio frequency power amplifiers. These networks, which we term resistance compression networks, serve to substantially decrease the variation in effective resistance seen by a tuned rf inverter as loading conditions change. We explore the operation, performance characteristics, and design of these networks, and present experimental results demonstrating their performance. Their combination with rectifiers to form rf-to-dc converters having narrow-range resistive input characteristics is also treated. The application of resistance compression in resonant power conversion is demonstrated in a dc/dc power converter operating at 100 MHz.