Department of Electrical and Computer Engineering
差错编码-林舒Lecture1

6
• Hard-decision decont than soft-decision decoding. However, soft-decision decoding offers significant performance improvement over hard-decision decoding.
Lecture 1
Introduction to Error Control Coding
1.1 A Block Diagram for a Data Transmission or Storage System
1
2
Information Source
Source Encoder
u
Channel Encoder
11
1. Compute the conditional probability P (v|r) for every codeword v. ˆ that 2. Decode the received sequence r into the codeword v has the largest a posteriori probability P (ˆ v|r), i.e., P (ˆ v|r) > P(v|r) for v = v ˆ.
Error Control Coding for Information Transmission and Storage
Shu Lin Department of Electrical and Computer Engineering University of California, Davis Davis, CA 95616
Then the error probability of the decoding is P (E )
文献翻译中英文对照 (2)

Low Voltage Flyback DC-DC Converter ForPower Supply ApplicationsHangzhou Liu1, John Elmes2, Kejiu Zhang1, Thomas X. Wu1, Issa Batarseh1 Department of Electrical Engineering and Computer Science,University of Central Florida, Orlando, FL 32816, USAAdvanced Power Electronics Corporation, Orlando, FL 32826, USA Abstract —In this paper, we design a low voltage DC-DC converter with a flyback transformer. The converter will be used as a biased power supply to drive IGBTs. The flyback transformer using planar EI-core is designed and simulated using ANSYS PExprt software. Besides, anLT3574 IC chip from Linear Technology has been chosen for converter control. Finally, the converter modeling and simulation are presented and PCB layout is designed. Keywords:Flyback, anLT3574IC, PCBI.INTRODUCTIONThe goal of this project is to develop and build a prototype of a high-efficiency, high-temperature isolated DC-DC converter to be used as a biased power supply for driving a complementary IGBT pair. It is important that the converter can deliver the required power at an ambient temperature of up to 100℃; therefore it has to be efficient so that its components do not exceed their maximum temperature ratings. The final converter will be completely sealed and potted in a metal case. The input voltage range for this converter is from 9V to 36V. The output sides have two terminals, one is﹢16V and the other one is﹣6V. In order to get the desired performance, anLT3574 IC chip from Linear Technology is used. The key to this design is the flyback transformer. The transformer using planar EI-core is designed and simulated using ANSYS PExprt software. Finally, the PCB layout of the converter will be presented.II.KEY DESIGN OUTLINEFor this flyback topology, the output voltage can be determined by both the transformer turns ratio and the flyback loop resistor pairs. Therefore, at the initial design stage, we can choose a convenient turn’s ratio for the transformer, and modify it later on if necessary to make sure the output performance is desirable and the transformer will not saturate [1].The relationship between transformers turns ratio and duty cycle can be found asWhere n is the transformer turns ratio, D is the duty cycle, V O` is the sum of the output voltage plus the rectifier drop voltage, V IN is the input voltage of the transformer.The value of feedback resistor can be calculated asWhere R REF is the reference resistor, whose value is typically 6.04kΩ; αis a constant of 0.986;V BG is the internal band gap reference voltage, 1.23V; and V TC is normally 0.55V [1].With a specific IC chosen, the converter circuit can be designed based on a demo circuit and some parameters may need to be modified if necessary to optimize the performance. Furthermore, in LT Spice, a large number of simulations need to be done with different conditions such as load resistor values and input voltage levels. It is important to make sure that the output voltage can be regulated well with all these different conditions.The most critical part of the design is the flyback transformer. With high switching frequency, the AC resistance can only be estimated based on some traditional methods such as Dowell’s curve rule [2].In order to get more accurate values of AC resistance values; we propose to use finite element electromagnetic software ANSYS PExprt to do the design [3]. At the initial design stage, key parameters such as the worst-case input voltage, frequency, material, inductance values willbe decided. After that, these data will be imported to the software, from which an optimized solution will be generated.III.CONVERTER SIMULATION RESULTSWe choose LT3574 chip in this design. From the simulation results in Figure 1 and Table 1, it clearly shows that the output voltages which are﹢16V and -6V respectively can be regulated pretty well with the input voltage range from 9V to 36V. The voltage tolerance ranges are from ﹢15V to ﹢19V and -12V to - 5V, respectively. In addition, the current is also under control, which is around 100mA in this designFigure 1 . Output voltage and current simulation resultsTable 1 . LT Spice simulation resuitsIV.TRANSFORMER SIMULATION RESULTSWith the initial design parameters of the transformer, we use ANSYS PExprt to simulate and further optimize the transformer [4].Figure 2 shows the primary winding voltage. In order to make the transformer work correctly in all cases, it is important to make sure that it can work at the worst case, which is the minimum input voltage in the range. Figure 3 shows the current through the primary winding.Figure 2 . Voltage of the primary windingFigure 3 . Current of the primary windingSince it is a low power converter in this design, it is critical to minimize the power losses. We choose to use the planar type transformer structure. After doing the winding interleaving, the power loss can be reduced by approximately 25% and the temperature rise can be reduced byapproximately 15% [5].The structure can be found in Figure 4. The primary winding is marked in yellow, which has 6 turns in series. The first secondary winding is marked in red, which has 3 turns in parallel. The second secondary winding is marked in blue, which has 1 turn. It will be totally 6 layers in the multi-layer transformer structure [6].Figure 4 . Winding geometry by interleaving methodBased on the computer simulation, the 6-layer planar transformer winding structure can be drawn in Figures 5 -10. The primary side winding has 6 turns in series. In Figures 6 and 9, it clearly shows that the turns in different layers are connecting through via hole. In one of the secondary winding which is the +16V one, it has 3 turns in parallel as shown in Figures 5, 8 and 10. The one turn secondary winding (6V) is shown in Figure 7.Figure 5 . Top layer winding structure (secondary 1)Figure 6 . Inner Layer 1 winding structure (primary)Figure 7 . Inner Layer 2 winding structure (secondary 2)Figure 8 . Inner Layer 3 winding structure (secondary 1)Figure 9 . Inner Layer 4 winding structure (primary)Figure 10 . Bottom layer winding structure (secondary 1)The core loss of the transformer is approximately 47mW, comparing to the winding loss of 154mW, it i s about 30%, as shown in Figure 11 [7].Figure 11. Power loss of transformerThe E-I core transformer PCB in this design will be integrated into the converter’s PCB, rather than a separate board being added to the whole circuit [8], which will reduce the cost of the PCB fabrication since multi-layer PCB layout is expensive.V.CONVERTER CIRCUIT PCB LAYOUTIn this project, we make the transformer part layout as one component; it will be integrated into the whole circuit PCB layout. It has 6 layers totally. The isolation requirement is 1500V, so the layout takes a little more space than the one without any isolation rules. In Figure 12, we make the primary side components all in the right hand side of the board, the secondary sides all in the left hand side of the board, and the transformer in between them. The wire traces have been marked with different colors in order to show the specific layer that the traces are on The board area is about 1.4×07, It can always reduce the size of the board by adding more layers. However, the cost will be more expensive. It is important to balance these factors. The size of the PCB board meets the specs of the project.Figure 12. PCB layout of the flyback converterVI.CONCLUSIONIn this paper, a flyback DC - DC converter for low voltage power supply application has been designed. The modeling and simulation results are presented. Based on the design specifications, a suitable IC from Linear Technology is chosen. A large amount of circuit simulations with different conditions such as load resistor values and input voltage levels are presented to get the desirable output voltage and current performance. The transformer has been designed including electrical, mechanical and thermal properties. With all the specific components decided, the PCB layout of the converter has been designed as well.REFERENCE[1] Linear Technology Application Notes , Datasheet of Isolated Flyback Converter Without anOpto-Coupler, /docs /Datasheet/3574f.pdf.[2] P.L.Dowell, “Effect of eddy currents in transformer windings” Proceedings of the IEE, NO.8PP.1387-1394, Aug 1966.[3] S.Xiao, “Plana r Magnetics Design for Low- Voltage DC-DC Converters” MS, 2004.[4] ANSYS Application Notes, PEmag Getting Started: A Transformer Design Example,/download/ EDA/Maxwell9/planarGS0601.pdf.[5] K. Zhang; T. X.Wu; H.Hu; Z. Qian; F.Chen.; K.Rustom; N.Kutkut; J.Shen; I.Batarseh;"Analysis and design of distributed transformers for solar power conversion" 2011 IEEE Applied Power Electronics Conference and Exposition (APEC), v l., no., pp.1692-1697, 6-11 March 2011.[6] Zhang.; T.X.Wu.; N.Kutkut; J.Shen; D.Woodburn; L.Chow; W.Wu; H.Mustain; I.Batarseh; ,"Modeling and design optimization of planar power transformer for aerospace applic ation," Proceedings of the IEEE 2009 National, Aerospace & Electronics Conference (NAECON) , vol., no., pp.116-120, 21-23 July 2009.[7] Ferroxcube Application Notes, Design of Planar Power Transformer,低电压反激式DC-DC转换器的在电源中的应用Hangzhou Liu1, John Elmes2, Kejiu Zhang1, Thomas X. Wu1, Issa Batarseh1 Department of Electrical Engineering and Computer Science,University of Central Florida, Orlando, FL 32816, USAAdvanced Power Electronics Corporation, Orlando, FL 32826, USA摘要:在本文中,我们设计了一个低电压反激式DC-DC转换器。
POWER SYSTEM ANALYSIS

ECE 476 POWER SYSTEM ANALYSISLecture7 Development of Transmission Line ModelsProfessor Tom OverbyeDepartment of Electrical andComputer EngineeringAnnouncements●For next two lectures read Chapter 5.●HW 2 is 4.10 (positive sequence is the same here as perphase), 4.18, 4.19, 4.23. Use Table A.4 values to determine the Geometric Mean Radius of the wires (i.e., the ninthcolumn). Due September 15 in class.●“Energy Tour” opportunity on Oct 1 from 9am to 9pm. Visita coal power plant, a coal mine, a wind farm and a bio-dieselprocessing plant. Sponsored by Students for Environmental Concerns. Cost isn’t finalized, but should be between $10 and $20. Contact Rebecca Marcotte at for more information or to sign up./ 岩棉板岩棉管硅酸盐保温涂料硅酸铝 SDGE Transmission Grid (From CALISO 2009 Transmission Plan)Line Conductors●Typical transmission lines use multi-strandconductors●ACSR (aluminum conductor steel reinforced)conductors are most common. A typical Al. to St.ratio is about 4 to 1.Line Conductors, cont’d●Total conductor area is given in circular mils. Onecircular mil is the area of a circle with a diameter of0.001 = π⨯0.00052 square inches●Example:what is the the area of a solid, 1”diameter circular wire?Answer:1000 kcmil (kilo circular mils)●Because conductors are stranded, the equivalentradius must be provided by the manufacturer. Intables this value is known as the GMR and isusually expressed in feet.Line Resistance-8-8Line resistance per unit length is given byR = where is the resistivity AResistivity of Copper = 1.6810 Ω-mResistivity of Aluminum = 2.6510 Ω-mExample: What is the resistance in Ω / m ile of a ρρ⨯⨯-82 1" diameter solid aluminum wire (at dc)?2.6510 Ω-m 16090.0840.0127mm R mile mile π⨯Ω==⨯Line Resistance, cont’d●Because ac current tends to flow towards thesurface of a conductor, the resistance of a line at 60 Hz is slightly higher than at dc.●Resistivity and hence line resistance increase asconductor temperature increases (changes is about 8% between 25︒C and 50︒C)●Because ACSR conductors are stranded, actualresistance, inductance and capacitance needs to be determined from tables.Variation in Line Resistance ExampleReview of Electric Fieldse A 2To develop a model for line capacitance wefirst need to review some electric field concepts.Gauss's law:d =q (integrate over closed surface)where =electric flux density, coulombs/md =differential D aD a 2e area da, with normal to surface A =total closed surface area, mq =total charge in coulombs enclosedGauss’s Law ExampleSimilar to Ampere’s Circuital law, Gauss’s Law is most useful for cases with symmetry.Example: Calculate D about an infinitely long wire that has a charge density of q coulombs/meter.Since D comesradially out inte-grate over thecylinder boundingthe wire e A d 2q where radially directed unit vectorD Rh qhq π====⎰r r D aD a aElectric FieldsThe electric field, E, is related to the electric flux density, D, byD= εEwhereE = electric field (volts/m)ε= permittivity in farads/m (F/m)ε= εo εrεo= permittivity of free space (8.854⨯10-12F/m)εr= relative permittivity or the dielectric constant (≈1 for dry air, 2 to 6 for most dielectrics)Voltage DifferenceP P The voltage difference between any twopoints P and P is defined as an integralV In previous example the voltage difference between points P and P , located radial distance R and R f d βααββααβαβ-⎰E lR R rom the wire is (assuming = )V ln 22o o o R q qdR R R βααβαβεεπεπε=-=⎰Voltage Difference, cont’dR R WithV ln 22if q is positive then those points closer in havea higher voltage. Voltage is defined as the energy (in Joules) required to move a 1 coulomb charge against an ele o o R q qdR R R βααβαβπεπε=-=⎰ctric field (Joules/Coulomb). Voltage is infinite if we pick infinity as the reference pointMulti-Conductor Casei 1Now assume we have n parallel conductors,each with a charge density of q coulombs/m.The voltage difference between our two points, P and P , is now determined by superposition1V ln 2ni i i i R q R αβαβαβπε== where is the radial distance from point P to conductor i, and the distance from P to i.i i R R ααββ∑Multi-Conductor Case, cont’d ni i=11111111If we assume that q 0 then rewriting111V ln ln 22We then subtract ln 0111V ln ln 22As we more P to infinity, ln 0n n i i i i i i ni i n ni i i i i i i q q R R q R R q q R R R βααβααβαβαααπεπεπεπε=======+==+→∑∑∑∑∑∑Absolute Voltage Defined1Since the second term goes to zero as P goes to infinity, we can now define the voltage of a point w.r.t. a reference voltage at infinity:11V ln 2This equation holds for any point as long a ni i i q R αββπε==∑s it is not inside one of the wires!Three Conductor CaseAB C Assume we have threeinfinitely long conductors,A, B, & C, each with radius rand distance D from theother two conductors.Assume charge densities suchthat q a + q b + q c = 01111ln ln ln 2ln 2a a b c a a V q q q r D D q D V r πεπε⎡⎤=++⎢⎥⎣⎦=Line Capacitancej1111For a single line capacitance is defined as But for a multiple conductor case we need to use matrix relationships since the charge on conductor i may be a function of V i i in n q C V q C C q =⎡⎤⎢⎥=⎢⎥⎢⎥⎣⎦11n nn n V C C V ⎡⎤⎡⎤⎢⎥⎢⎥⎢⎥⎢⎥⎢⎥⎢⎥⎣⎦⎣⎦=q C VLine Capacitance, cont’dIn ECE 476 we will not be considering theses cases with mutual capacitance. To eliminate mutual capacitance we'll again assume we have a uniformly transposed line. For the previous three conductor exam a a a ple:q 2ince q = C ln a a V VS V C D V rπε=→==Bundled Conductor Capacitance 11c b 12Similar to what we did for determining lineinductance when there are n bundled conductors,we use the original capacitance equation just substituting an equivalent r Note fo adius r t()he R n nrd d b capacitance equation we use r rather than r' which was used for R in the inductance equationLine Capacitance, cont’d[]1m 13m1c b 12-12o For the case of uniformly transposed lines weuse the same GMR, D , as before.2ln whereD R () (note r NOT r')ε in air 8.85410 F/mn m c bab ac bc n C D R d d d rd d πεε=====⨯Line Capacitance ExampleCalculate the per phase capacitance and susceptance of a balanced 3 , 60 Hz, transmission line with horizontal phase spacing of 10m using three conductor bundling with a spacing between conductors in the bundle of 0.3m. Assume the line is uniformly transposed and the conductors have a a 1cm radius.Line Capacitance Example, cont’d 1313m1211c 118(0.010.30.3)0.0963 m D (101020)12.6 m 28.85410 1.14110F/m 12.6ln 0.096311X 260 1.14110F/m2.3310 -m (not /m)c b R C C πωπ---=⨯⨯==⨯⨯=⨯⨯==⨯==⨯⨯=⨯ΩΩACSR Table Data (Similar to Table A.4)Inductance and Capacitance GMR is equivalent to r’assume a D m of 1 ft.ACSR Data, cont’d7L 333X 2410ln 1609 /mile 12.0210ln ln 12.0210ln 2.0210ln m m m D f L f GMRf D GMR f f D GMR ππ----==⨯⨯Ω⎡⎤=⨯+⎢⎥⎣⎦=⨯+⨯Term from table assuminga one foot spacing Term independent of conductor withD in feetACSR Data, Cont.0C 6To use the phase to neutral capacitance from table21X -m where 2ln 1 1.77910ln -mile (table is in M -mile)1111.779ln 1.779ln M -mile m m m C D f C rD f rD f r fπεπ=Ω==⨯⨯ΩΩ=⨯⨯+⨯⨯ΩTerm from table assuming a one foot spacing Term independent of conductor with D in feetDove Example70.0313 feetOutside Diameter = 0.07725 feet (radius = 0.03863)Assuming a one foot spacing at 60 Hz12602101609ln Ω/mile 0.03130.420 Ω/mile, which matches the table For the capacitancea a C GMR X X X π-==⨯⨯⨯⨯=64111.77910ln 9.6510 Ω-mile f r=⨯⨯=⨯●Multi-circuit lines: Multiple lines often share a common transmission right-of-way. This DOES cause mutual inductance and capacitance, but is often ignored in system analysis.●Cables:There are about 3000 miles of underground ac cables in U.S. Cables are primarily used in urban areas. In a cable the conductors are tightly spaced, (< 1ft) with oil impregnated paper commonly used to provide insulationinductance is lowercapacitance is higher, limiting cable length●Ground wires: Transmission lines are usually protected from lightning strikes with a ground wire. This topmost wire (or wires) helps to attenuate the transient voltages/currents that arise during a lighting strike. The ground wire is typically grounded at each pole.●Corona discharge:Due to high electric fields around lines, the air molecules become ionized. This causes a crackling sound and may cause the line to glow!●Shunt conductance:Usually ignored. A small current may flow through contaminants on insulators.●DC Transmission:Because of the large fixed cost necessary to convert ac to dc and then back to ac, dc transmission is only practical for several specialized applicationslong distance overhead power transfer (> 400 miles)long cable power transfer such as underwaterproviding an asynchronous means of joining differentpower systems (such as the Eastern and Western grids).Tree Trimming: BeforeTree Trimming: AfterTransmission Line Models●Previous lectures have covered how to calculate thedistributed inductance, capacitance and resistance of transmission lines.●In this section we will use these distributedparameters to develop the transmission line models used in power system analysis.Transmission Line Equivalent Circuit Our current model of a transmission line is shown belowFor operation at frequency , let z = r + j L and y = g +j C (with g usually equal 0)ωωωUnits onz and y areper unitlength!Derivation of V, I Relationships We can then derive the following relationships:()()()dV I z dxdI V dV y dx V y dxdV x dI x z I yV dx dx==+≈==Setting up a Second Order Equation 2222()()We can rewrite these two, first order differential equations as a single second order equation ()()()0dV x dI x z I yV dx dxd V x dI x z zyV dx dxd V x zyV dx ====-=V, I Relationships, cont’d22Define the propagation constant aswherethe attenuation constantthe phase constantUse the Laplace Transform to solve. System has a characteristic equation()()()0yz j s s s γγαβαβγγγ==+==-=-+=Equation for Voltage1212121122121212The general equation for V is()Which can be rewritten as ()()()()()22Let K and K . Then()()()22cosh()sinh()x x x x x x x x x x V x k e k ee e e e V x k k k k k k k k e e e e V x K K K x K x γγγγγγγγγγγγ-----=++-=++-=+=-+-=+=+Real Hyperbolic FunctionsFor real x the cosh and sinh functions have the following form:cosh()sinh()sinh() cosh()d x d x x x dx dxγγγγγγ==Complex Hyperbolic Functions For x = α+ j βthe cosh and sinh functions have the following formcosh cosh cos sinh sin sinh sinh cos cosh sin xj x j αβαβαβαβ=+=+。
研究生留学美国石溪大学全新攻略.doc

研究生留学美国石溪大学全新攻略石溪大学是美国一所知名高校,其实力强劲,下面我就带大家一起看看石溪大学的优势专业以及申请条件。
美国石溪大学简介:石溪大学(Stony Brook University),又称纽约州立大学石溪分校(The State University of New York at Stony Brook,简称SBU),坐落于纽约市郊风景区长岛,是美国著名的研究型公立大学,在世界范围内享有很高的学术声誉。
石溪大学是北美顶尖大学联盟美国大学协会(AAU)的62个成员之一,享有公立常春藤的美誉。
石溪大学在202X US News世界大学综合排名中居第131名,美国大学本科综合排名中位居第96名。
卡内基基金会(Carnegie Foundation)将石溪分校列为全美最优秀的研究机构之一,约翰·霍普金斯大学(Johns Hopkins University)亦十分推崇石溪分校在学术研究方面的成果,根据该校的排名,石溪分校名列全美公立大学第二名,仅次于加州大学伯克利分校。
该校的许多学院入学竞争尤为激烈,如牙医学院录取率更低至3.5%。
石溪大学拥有许多一流学者,其教学品质和研究成果享誉国际,石溪的教授队伍中聚集着各领域的顶尖研究人员,著名核物理学家、诺贝尔物理奖获得者杨振宁在该校执教37年。
世界上第一张核磁共振图像诞生于此。
自1957年成立后,已经培养出多位著名人物,包括3位诺贝尔奖得主,1位图灵奖得主及菲尔茨奖、沃尔夫奖和阿贝尔奖得主约翰·米尔诺等。
美国纽约石溪大学优势专业石溪大学有很多专业在全美大学专业排名中名列前茅,如:生物化学、生物、电脑科学、应用数学与统计学、经济学、电子工程、工程学、英文、地质学、历史、数学、美国政治(政治心理学)、护理、音乐、物理和心理学等。
石溪大学独特的教学体系,使其毕业生因学识渊博和具备创造性的思考能力,而受到广泛好评。
研究生升学率为全国平均水准的两倍。
电子工程英文案例十三

Applied Program: Ph.D. program, Computer Science, Department of Electrical Engineering and Computer ScienceAs a child who has grown up in a family of scientific researchers, I have developed along-standing love for science since my early childhood and have long wished to make my own achievements in the field of science and technology. This aspiration was to some extent fulfilled in November 2003, a truly defining year for me in my personal development, when I was invited to make a presentation at XX as the author of the research paper XX. In my presentation, I proposed the novel idea that relatively high precise position estimation can be realized through the relatively low-cost commercial GPS received the unanimous appreciation from the experts attending the conference. This international conference experience, while allowing me to savor the initial sense of achievement, has all the more reinforced my determination to seek a research career in computer science.This achievement, among many others, would have been impossible without a quality education at the most prestigious university in China—XX University—where I have completed my undergraduate program and am on the verge of completing my Master’s program, both at the Department of Computer Science and Technology. During my undergraduate program, I relied on my strong background in mathematics and physics that I developed in middle school and on efficacious study strategies to score a well-above-average academic performance. I have particularly excelled in such specialized courses as Principles of Artificial Intelligence, Pattern Recognition, Hardware Design a nd Software Debugging. During my Master’s program, as in my undergraduate program, I have continued to pay close attention to the theoretical courses in computer sciences and have been able to speedily react to the most recent theoretical findings and new methodologies. In the elitist academic environment of XX University where competition is automatically exceedingly fierce, I know it is gravely inadequate simply to be a good learner. Only the fast good learner can be the fittest to emerge triumphant in co mpetition. It is this consistent “Fast-Good-Learning” principle that has brought me a spate of accolades including top-level scholarships and academic honors.As early as I was a junior student in my undergraduate program, I had contributed my efforts in the research experiments on speech recognition, digitalized remote controller of robots,intelligent battery charger and other projects. All those experiments initially proved to be major challenges to me as I discovered that my lack of relevant development experience put me in a grave disadvantage. To overcome my inadequacies, I resorted to intensive self-education and to auditing a lot of specialized courses until I could perform the experiments independently. In undertaking my graduation project for my Bac helor’s degree—Intelligent Air Data Sensors, I exercised my initiative of learning to a fullest extent. When I found that I lacked in-depth knowledge concerning the intelligent technology, I immediately set to consult a large amount of technical literature and within the shortest possible time mastered all the necessary knowledge regarding the fuzzy logic. This knowledge enabled me to solve all the major issues and to considerably enhance the precision of the sensors, resulting in the panel’s rating of my t hesis as “first-class” for its high quality and innovative research.In the course of my Master’s program, my academic interest and creativity have been brought into full play. Over the past years, while further modifying and improving the research on intelligent air data sensors, I have participated in the project XX. Working in a research team, my responsibility in this project was designing the electronic hardware circuit design of the robot and its control algorithm. In addition to fulfilling my own share of responsibilities, I communicated fully with other team members and the result of those effective communications was my proposal that distributed control architecture be adopted and individual processors be used to control the robot’s motion and comm unication. With efficient teamwork, our team,XX Mars, won XX. Subsequent to this project, I completed the research on the estimation system of unmanned helicopter based on GPS that I mentioned at the beginning of this Statement. The air data sensor and the GPS positioning system that I developedsingle-handedly have both been applied in the unmanned aerial vehicle and unmanned helicopter, playing a vital role in insuring the overall success of the two unmanned projects. An important award I have won is the third prize during the XX Cup 2002 Creative Electronics Design Competition of XX University. My computer software development skills have been strengthened by the part-time job as software engineer that I have been working on for the past five years at XX Co. Ltd. I have been responsible for developing the application programs under the embedded Linux operating system, including the text editor, file manager, desk-top manager and e-mail client terminal software.Currently, I am working on my Master’s diss ertation at XX. The paramount problem I am to work out is the time management and synchronic management in distributed system and the simplification method in terrain visualization. This project is very meaningful to the extent that the simulation system we develop will promote future control research of robotics and that all simulation modules can be re-used and replaced with minimal addition efforts.In retrospection of my past academic pursuit, I am very delighted to find that I have truly plunged myself into the formal research on computer science and have developed an unswerving interest in pursuing an academic career in this domain, specifically in computer application technology and robotics. In terms of robotics, I am interested in probing into such issues as the control of the autonomous robot, the application of artificial intelligence to robots, and the application of robots to practical purposes. I wish to study how to make robots act purposefully and successfully in a world in which almost everything is uncertain. Home to the most advanced computer technology in the world, the United States can offer me many fascinating concepts and technologies in computer science.A careful reflection on my background indicates that I am a good match for XX as its Department of Electrical Engineering and Computer Sciences offers curriculum and research programs that closely correspond to my interest. Among approximately 85 faculty members, roughly half are in the Computer Science Division, indicating an almost unparalleled faculty resource. This is further proved by the fact that the Department includes 22 members of the National Academy of Engineering and 9 Association for Computing Machinery (ACM) Fellows. Three faculty have also won the ACM Turing Award. Unde r XX ’s Ph.D. program, concentrations cover analysis of algorithms, artificial intelligence, complexity, theory of computation, computer architecture, operating systems, robotics, and computer vision. Therobotics study, in which I am most interested, has reached a leading international level at XX. Most importantly XX exalts extensive cooperation among faculty and students within ECS, in related departments on campus and with other research laboratories in companies. My own past experience shows me that cooperative research is vital to any scientific undertaking.It is my conviction that the excellence of the faculty and the breadth of educational opportunities in XX will expose me to the best education and research training in my chosen field. In your nurturing environment, my potential will be developed and my skills honed. I will endeavor to become an accomplished computer scientist worthy of the prestige of XX.。
电磁兼容技术——chapter 8

Introduction to Electromagnetic CompatibilitySecond EditionCLAYTON R.PAULDepartment of Electrical and Computer Engineering,School of Engineering, Mercer University,Macon,Georgia and Emeritus Professor of Electrical Engineering,University of Kentucky,Lexington,KentuckyA JOHN WILEY&SONS,INC.PUBLICATIONChapter8Radiated Emissions and SusceptibilityIn this chapter we will discuss the important mechanisms by which electromagnetic fields are generated in an electronic device and are propagated to a measurement antenna that is used to verify compliance to the governmental regulatory limits.Recall that for domestic radiated emissions the frequencies range of measurement is from30MHz to over1GHz.The FCC measurement distance is3m for Class B products and10m for Class A products.For CISPR22(EN55022)the measurement distance is10m for Class B products and10m for Class A products.Let’s recall the SAC.FIGURE2.7Illustration of the use of a semi anechoic chamber for the measurement of radiated emissionsThe lower frequency of30MHz is one wavelength at10m,whereas the frequency of1GHz is one wavelength at30cm.The product is therefore in the near field of the antenna for certain of the lower-frequency ranges of the regulatory limits and in the far field for the higher-frequency ranges.We will generate some simple models for first-order predictions(concept predictions)of the radiated emissions from wires and PCB lands in this chapter. For simplicity these models will assume that the measurement antenna is in the far field of the emission(the product),although this is not necessarily the caseover the entire frequency range of the regulatory limit.We will also investigate the ability of the product to be susceptible to radiated emissions from other electronic devices by deriving simple models that give the voltages and currents induced in parallel-conductor lines by an incident uniform plane wave.The incident wave is produced by a distant antenna such as a FM radio station.8.1SIMPLE EMISSION MODELS FOR WIRES AND PCB LANDSIn this section we will formulate some simple models that allow us to understand the factors that cause the radiated emissions from the currents on wires and PCB lands to exceed the regulatory limits.These will be derived for ideal situations such as an isolated pair of wires in free space distant from any other obstacles.The sole purpose of these models is to provide insight into the levels and types of currents with regard to their potential for creating radiated emissions.It is important to keep in mind that time-varying currents are the mechanismthat produce radiated electromagnetic fields.Hence currents on wires,PCB lands, or any other conductor in the system will radiate.The essential question is how well they radiate.Therefore our task in reducing radiated emissions is to produce“antennas”having poor emission properties.8.1.1Differential-Mode versus Common-Mode CurrentsConsider the pair of parallel wires or PCB lands of length L and separation s shown in Fig.8.1a.The two conductors are placed in the xz plane and are parallel to the z axis.Suppose that the currents at the same cross section are directed to the right and denoted as I^1and I^2.We are so familiar the equations below.FIGURE8.1Illustration of the relative effects of differential-mode currents I^D and common-mode currents I^C on radiated emissions for parallel conductors:(a) decomposition of the total currents into differential-mode and common-mode components;(b)radiated emissions of differential-mode currents;(c)radiated emissions of common-mode currents.Common-mode currents are inconsequential(not following logically as a consequence)in typical products,and,moreover,they often produce larger radiated emissions than do the differential-mode currents.In order to see why this occurs,let us consider the radiated electric fields in the plane of the wires and at a point midway along the line and a distance d from the line.The configuration for differential-mode currents is illustrated in Fig.8.1b. Observe that because the differential-mode currents are equal in magnitude but oppositely directed,the radiated electric fields will also be oppositely directed, and will tend to cancel.They will not exactly cancel,since the wires are not collocated,so the net electric field E^D will be the difference between these emission components,as indicated in Fig.8.1b.On the other hand,consider the emissions due to the common-mode currents shown in Fig.8.1c.Because the common-mode currents are directed in the same direction,their radiated electric field components will add,producing a net radiated electric field E^C.In the following sections we will show that for a l-m ribbon cable with wireseparation of50mils a differential-mode current at30MHz of20mA will produce a radiated emission just equal to the FCC Class B limit(40dBmV/m or 100mV/m from30to88MHz).On the other hand,a common-mode current of only8mA will produce the same emission level!This is a ratio of2500,or some 68dB.Thus seemingly inconsequential common-mode currents are capable of producing significant radiated emission levels.In this section we will derive simple emission models for a pair of parallel wires or PCB lands due to the currents on those conductors.This case of a pair of parallel wires or PCB lands represents an important and easily analyzed structure.It will therefore provide insight into the radiation mechanism of other structures.FIGURE8.2Calculation of the far fields of the wire currents.In order to determine this total radiated electric field of the two conductors, consider placing the two currents along the x axis and directing them in thez direction as shown in Fig.8.2.Each electric field of these linear antennas will be a maximum broadside to(in a direction perpendicular to)the antenna,that is,in the xy plane,θ=90°.Hence we will determine the maximum electric field if the xy plane.The start point for the latter discuss is the equations below.The term M^is a function of the antenna type such as Hertzian dipoles andhalf-wave dipoles.The subscription of the Eθmeans the direction of Eθ.8.1.2Differential-Mode Current Emission ModelIn order to simplify the resulting model,we make three important,simplying assumptions:(1)The conductor lengths L are sufficiently electrically short and the measurement point is sufficiently distant that the distance vectors from each point on the antenna to the measurement point are approximately parallel,(2)The current distribution(magnitude and phase)is constant along the line, and(3)The measurement point is in the far field of each antenna.We will also determine the radiated fields at a point that is perpendicular to the line conductors and in the plane containing them,as shown in Fig.8.3.For differential-mode currents,I^2=-I^1,it is a simple matter to show that a maximum will occur in the plane of the wires and on a line perpendicular to the wires(φ=0°,180°in Fig.8.2).In addition,the measurement point is at a distance d from the midpoint of the line.FIGURE8.3A simplified estimate of the maximum radiated emissions due to differential-mode currents with constant distribution.Also we substitute r=d andφ=0°(to give the fields in the plane of the wires). And finally,since we are considering differential-mode currents,we substitutein to(8.9).The result becomesWhere we replacesubstitutingand assuming that the wire spacing s is electrical small,so thatthe magnitude of(8.11)reduces toand is parallel to the wires.-----------------------Example8.1As an example,consider the case of a ribbon cable constructed of28-gauge wires separated a distance of50mils.Suppose the length of the wires is1m and that they are carrying a30MHz differential-mode current.The level ofdifferential-mode current that will give a radiated emission in the plane of the wires and broadside to the cable(worst case)that just equals the FCC Class B limit(40dBmV/m or100mV/m at30MHz)can be obtained by solving(8.12)to give-----------------------Generally,the formula for the maximum emission given in(8.12)is sufficient for estimation purposes.Please observe the(8.12).The maximum radiated electric fields vary with(1)The square of the frequency,(2)The loop area A=Ls,and(3)The current level I^D.Therefore,in order to reduce the radiated emissions at a specific frequency due to differential-mode currents,we have the following options:(1)Reduce the current level.(2)Reduce the loop area.-Mode Current Emission ModelCommon-Mode8.1.3CommonIt is quite easy to modify the preceding results to consider the case ofcommon-mode currents shown in Fig.8.7.FIGURE8.7A simplified estimate of the maximum radiated emissions due to common-mode currents with constant distribution.For common-mode current there isWe have-----------------------Example8.2As an example,consider the case of a ribbon cable constructed of28-gauge wires separated a distance of50mils that was considered earlier for differential-mode currents.Suppose the length of the wires is1m and that they are carrying a30 MHz common-mode current.The level of common-mode current that will give a radiated emission broadside to the cable(worst case)that just equals the FCC Class B limit(40dBmV/m or100mV/m at30MHz)can be obtained by solving (8.16a)to give-----------------------Generally,the formula for the maximum emission given in(8.16a)is sufficient for estimation purposes.The maximum radiated electric fields vary with(1)The frequency,(2)The line length L,and(3)The current level I^C.Therefore,in order to reduce the radiated emissions at a specific frequency due to common-mode currents we have the following options:(1)Reduce the current level.(2)Reduce the line length.8.1.4Current ProbesDifferential-mode currents are the desired or functional currents in the system and as such can be reliably calculated using transmission-line models or,for electrically short lines,lumped-circuit models.Common-mode currents,on the other hand,are undesired currents and are not necessary for functional performance of the system.They are therefore dependent on non-ideal factors such as proximity to nearby ground planes and other metallic objects as well as other asymmetries.Consequently they are difficult to calculate using ideal models.They can,however,be measured using current probes.Current probes make use of Ampere’s lawwhere C is the contour bounding the open surface S.Ampere’s law shows that a magnetic field can be induced around a contour by either conduction current or displacement current that penetrates the open surface S,as illustrated in Fig.8.9a.A time-changing electric field produces a displacement current.If notime-changing electric field penetrates this surface,the induced magnetic field is directly related to the conduction current passing through the loop.Current probes use this principle in order to measure current.A current probe is constructed from a core of ferrite material that is separated into two halves,which are joined by a hinge and closed with a clip.The ferrite core is used to concentrate the magnetic flux.The clip is opened,the core placed around the wire(s)whose current is to be measured,and the probe closed.The total current that passes through the loop produces a magnetic field that is concentrated in and circulates around the core.Several turns of wire are wound on the core,so that the time-changing magnetic field that circulates around the core induces,by Faraday’s law,an emf that is proportional to this magnetic field. The induced voltage of this loop of wire can therefore be measured and is proportional to the current passing through the probe.FIGURE8.9The current probe:(a)illustration of Ampere’s law;(b)use of the current probe to measure currents.A photograph of a typical current probe is shown in Fig.8.10a.FIGURE8.10(a)Photograph of a current probe and(b)its measured transfer impedanceIt is not necessary to carry out precise calculations of the resulting fields and induced emf in order to calibrate the probe.Simply pass a current of known magnitude and frequency through the probe and measure the resulting voltage produced at the terminals.The result is a calibration curve that relates the ratio of the voltage V^to the current I^asThe quantity Z^T has units of ohms and is referred to as the transfer impedance of the current probe.The probe manufacturer provides a calibration chart with the probe that shows the magnitude of the transfer impedance versus frequency.This calibration chart was obtained by passing a current of known amplitude and frequency through the probe and measuring the resulting voltage at the probe terminals.Usually this is given in dBΩ(relative to1Ω)asA typical such plot is shown in Fig.8.10b.There is an important assumption inherent in the transfer impedance calibration curve:the termination impedance of the probe.For example,in the calibration of the probe as illustrated in Fig.8.9b a voltage measurer such as a spectrum analyzer was used to measure the probe voltage in the course of determining the probe transfer impedance.Therefore the load impedance at the terminals of the probe is the input impedance to the measurement device,which is usually50Ω.Thus the calibration curve of the current probe is valid only when the probe is terminated in the same impedance as was used in the course of its calibration(usually50Ω).The probe measures the total or net common-mode current in the cable and the magnetic fluxes due to the differential-mode currents cancel out in the core.Thus the current probes will not measure differential-mode current unless it is placed around each individual wire.In fact,the current probe can be a useful EMC diagnostic tool throughout the design of a product.It is a simple matter to measure the net common-mode currents on all peripheral cables of a product or a prototype of the product in the development laboratory using a current probe and an inexpensive spectrum analyzer.8.1.5Experimental ResultsIn order to illustrate the relative magnitudes of differential-and common-mode current emissions,as well as to illustrate the prediction accuracy of the above models,we will show experimental results in this section.FIGURE8.12An experiment to assess the importance of common-mode currents on cables in the total radiated emissions of the cable:(a)schematic of the device tested;(b)photograph of the device.The first experiment is illustrated in Fig.8.12.A10MHz oscillator packaged in a standard14-pin dual inline package(DIP) drives a74LS04inverter gate.The output of this gate is attached to the input of another74LS04inverter gate via a1m,three-wire ribbon cable as shown in Fig.8.12a.The ribbon cable wires are28-gauge(Diameter=0.32mm)and havecenter-to-center separations of50mils.The middle wire carried the10MHz trapezoidal pulse train output of the driven gate to the gate at the other end,which serves as an active load.An outer wire carries the+5V power for the inverter active load,and the other outer wire serves as the return for both signals.The+5V power is derived from a9-V battery that powers a7805regulator asshown in Fig.8.12b.This provides a compact5-V source.There is no external connection to the commercial power system.This was intentional,so that radiation from the power cord of a power supply would not contaminate the measurements.The radiated emissions were measured in a semi anechoic chamber that is regularly used for developmental and compliance testing.The measured data to be shown were obtained over the frequency range of30–200MHz.The antenna and the ribbon cable were positioned parallel to the chamber floor, and both were1m above the floor.The separation between them was3m.A current probe having a probe transfer impedance of15dBΩwas used to measure the common-mode current on the cable for the prediction of the common-mode current radiated emissions.Equations used to provide the predicted electric field were derived from the (8.16a)and(8.20)The oscillator has a fundamental frequency of10MHz,so only harmonics of 10MHz will appear in the radiated emissions.A plot of the radiated emissions is shown in Fig.8.15.The predicted values are shown on the plot and are denoted by X.The predictions are within3dB of the measured data,except at50,80,and130 MHz.FIGURE8.15Measured and predicted emissions of the device of Fig.8.12.8.2S imple S usceptibility M odels for W ires and PCB L andsComplying with the regulatory limits on radiated(and conducted)emissions is an absolute necessity in order to be able to market a digital electronic product. However,as was pointed out previously,s imply being able to comply with regulatory emission limits does not represent a complete product design from the standpoint of EMC.If a product exhibits susceptibility to external disturbances such as radiated fields from radio transmitters and radars or is susceptible to lightning-or electrostatic-discharge(ESD)-induced transients,then unreliable performance will result and customer satisfaction will be impacted.The model that we will develop is a simplified version of the more exact transmission line model described,but it will be suitable for estimation purposes. We consider a parallel-wire transmission of length L that has a uniform plane wave incident on it as shown in Fig.8.21a.FIGURE8.21Modeling a two-conductor line to determine the terminal voltages induced by an incident electromagnetic field:(a)problem definition;(b)effects of the transverse electric field component and the normal magnetic field component;(c)a per-unit-length equivalent circuit.The wires are separated a distance s and have load resistances R S and R L.Weplace the two wires in the xy plane,with R S located at x=0and R L at x=L.The wires are parallel to the x axis.Our interest is in predicting the terminal voltages V^S and V^L given theeld E^i of a uniform magnitude of a sinusoidal,steady-state incident electric fi fieldplane wave,its polarization,and the direction of propagation of the wave.Two components of the incident wave contribute to the induced voltages.The component of the incident electric field that is transverse to the line axis,E^i t= E^i y(in the plane of the wires and perpendicular to them and directed upward), andThe component of the incident magnetic field that is normal to the plane of the wires, H^i n=H^i z(perpendicular to the plane of the wires and into the page),as shown in Fig.8.21b.The line will possess per-unit-length parameters of inductance l and capacitance c.For the parallel-wire line having wires of radius r w theseper-unit-length parameters were derived in Chapter4,and arewhereεr is the relative permittivity of the surrounding medium(assumed homogeneous and non-ferromagnetic).The essential modification required for the following model to apply to two parallel lands on a PCB are the use of the proper per-unit-length parameters of capacitance and inductance.A model of a∆x section of the line is shown in Fig.8.21(c),where theper-unit-length parameters are multiplied by the length of the section,∆x.The per-unit-length induced sources V^s and I^s are generated by the incident wave according to the following considerations.First consider the normal component of the incident magnetic field intensity vector H^i n:Faraday’s law(see Appendix B)shows that this will induce an emf (electric motive force)in the loop bounded by the wires asThis induced emf can be viewed as an induced voltage source whose polarity, according to Lenz’law,is such that it tends to produce a current and associated magnetic field that opposes any change in the incident magnetic field.Thus,forthe incident magnetic field intensity vector,normal to and into the page,the positive terminal of the source will be on the left.For a∆x section,theper-unit-length source will be given by dividing the result in(8.29)by∆x to giveThe per-unit-length induced current source I^s is directed in the-y(downward) direction,and is due to the component of the incident electric field intensity vector that is transverse to the line and directed in the+y direction.The incident fields at the position of the line may be produced by some distant antenna.The antenna producing these incident fields is assumed to be transmitting a radiated power P T,is located a distance d away,and has a gain G in the direction of the line.The incident electric field is[see(7.71)of Chapter7]The incident magnetic field,assuming a uniform plane wave,is obtained by dividing the electric field by the intrinsic impedance of free space,n0=120πΩ= 377Ω,to give-------------------------Example8.6For example,consider a half-wave dipole having a gain in the main beam of2.15dB(1.64absolute),transmitting1kW radiated power at100MHz.If the line is located a distance of3000m from the antenna,the maximum electric and magnetic fields in the vicinity of the line are------------------------If the line length is electrically short at the frequency of interest(L<=λ0/10),we may lump the distributed parameters by using one section of the form in Fig. 8.21c to represent the entire line and replacing∆x with L.We will make a final simplification that provides an extremely simple model that is valid for a wide variety of practical situations.In this simple model we ignore the per-unit-length parameters of inductance and capacitance.Neglecting the line inductance and capacitance is typically valid so long as the termination impedances are not extreme values such as short or open circuits.In addition,since the wire separation is much less than the wire length and is therefore also electrically short,the field vectors do not vary appreciably across the wire cross section,that is,with respect to y.Therefore(8.30)and(8.31)becomeThe simplified model is shown in Fig.8.23.From this model it is a simple matterto compute the induced terminal voltages,using superposition,asfields for a two-conductor line that is short,electrically.-------------------------Example8.7Consider,as a first example,the1-m ribbon cable shown in Fig.8.24a.The wires are28-gauge7x36(r w=7.5mils)and are separated by50mils. The termination impedances are R S=50Ωand R L=150Ω.FIGURE8.24An example illustrating the computation of induced voltages for a 10-V/m,100-MHz incident uniform plane wave with broadside incidence:(a) problem definition;(b)the equivalent circuit.The characteristic impedance of this cable isand we have ignored the wire dielectric insulation,εr=1.The line incident uniform plane wave has a frequency of100MHz and is traveling in the xy plane in the y direction.The line is1/3λ0at100MHz.This is probably marginal for the line to be considered electrically short.For illustration purposes we will assume that the line is electrically short and use the simplified model in Fig.8.23.The electric field intensity vector has a magnitude E i=10V/m and is polarized inthe x direction.The magnetic field intensity vector is therefore directed in the negative z direction(into the page)according to the properties of uniform plane waves,and is given H i=E i/n0=10/120π=26.5mA/m.Thus the component of the electric field transverse to the line is zero,and the component of the magnetic field that is normal to the plane of the wires is the total magnetic field vector.Therefore the induced sources are obtained from(8.35)asBecause there is no component of the electric field that is transverse to the line axis,the current source is absent.The equivalent circuit is shown in Fig.8.24b, from which we calculate(by voltage division)-------------------------8.2.1Experimental ResultsThere is something wrong with the experiment.The author used wrong equation to calculate V^s.8.2.1Shielded Cables and Surface Transfer ImpedanceCoaxial cables consist of a concentric shield enclosing an interior wire that is located on the axis of the shield.The intent of the shield is to completely enclose a circuit in order to prevent coupling to the terminations from incident fields outside the shield,as illustrated in Fig.8.30.If the shield could be constructed of a solid,perfectly conducting material,this would be the case.FIGURE8.30Illustration of incident field pickup for a shielded cable.We will assume that pigtails and other breaks in the shield are not present,so that the only penetration of an external field is through the shield.External fields penetrate non-ideal shields via diffusion of the current that is induced by the external field on the external surface of the shield.A typical way of calculating this interaction is to first calculate the current induced on the shield exterior by the external,incident field,assuming the shield is a perfect conductor and completely encloses the interior circuitry.Once the exterior shield current I^SH is computed in this fashion,the induced voltages in the terminations V^S and V^L are computed in the following manner. The shield current diffuses through the shield wall to give a voltage drop on the interior surface of the shield ofwhere the surface transfer impedance of the shield isand the propagation constant in the shield material isδis the skin depth,The shield inner radius is denoted by r sh and the shield thickness is by t sh.A plot of the surface transfer impedance is shown in Fig.8.31.This is normalized to the per-unit-length dc resistance of the shieldFIGURE8.31The surface transfer impedance of a cylinder as a function of the ratio of shield thickness to skin depth.and shows that the shield current on the exterior of the shield completely diffuses through the shield wall for wall thicknesses less than a skin depth,t sh<<δ,as we would expect.For wall thicknesses greater than a skin depth,the current on the exterior only partially diffuses through the shield wall,and the transfer impedance decreases with decreasing skin depth(increasing frequencies).This voltage drop on the interior surface of the shield acts as a voltage sourceZ^T I^SH∆x along the longitudinal interior surface of the shield.A per-unit-length equivalent circuit for the circuit enclosed by the shield is shown in Fig.8.32a, where r,l,g,and c are the per-unit-length resistance,inductance,conductance, and capacitance of the interior wire-shield circuit.FIGURE8.32The equivalent circuit of the interior of a coaxial cable for computing the pickup of external fields:(a)the per-unit-length equivalent circuit;(b)a simplified equivalent circuit for cables that are short,electrically.For an electrically short line we can approximate the solution by lumping the source and ignoring the per-unit-length parameters of the inner wire–shield circuit,as shown in Fig.8.32b,to giveProblems------。
带你了解美国ECE, EECS, EE, CS的专业设置情况(世毕盟留学)

带你了解美国ECE, EECS, EE, CS的专业设置情况(世毕盟留学)ECE: Electrical and Computer Engineering 电子与计算机工程EE: Electrical Engineering 电子工程CS: Computer Science 计算机科学EECS: Electrical Engineering and Computer Science,其实是把EE和CS放在了一起,但是一般提供EE和CS两个不同的项目有些少数学校比如MIT是把EE和CS结合在一起说的,所以他们提供的学位是Master of Engineering in ECE, Doctor of Philosophy (PhD) 和Doctor of Science. 但ECE偏硬件,所以大部分学校一般和EE放在一起说,CS偏软件,一般单独开设department of computer science 或者单独授予Master of science in computer science. 比如Princeton, Harvard, Yale, Chicago, Columbia, Stanford等等都是把EE 和CS分开来说的。
下面我们来看一下CS, EE, ECE和EECS主要的研究方向CS-我们以专排第一的CMU为例来看一下CMU是把CS单独放在了一个department 下(department of computer science);把ECE也单独放在了一个department 下(department of electrical and computer engineering). 分别提供MS和PhD两个学位。
CS主要的研究领域如下:Artificial Intelligence(人工智能);Computer Security(计算机安全);Graphics (图像);Programming Languages(程序设计语言);Systems(系统);Theory (理论)ECE主要的研究领域如下:在应用领域方面:Energy(能源);Healthcare & quality of life(医疗健康和生活质量);Mobile systems(移动系统);Smart infrastructure(智能设施)在系统和技术方面:Beyond CMOS,Compute/storage systems, Cyber-physical systems, Data/network science, secure systems在Theoretical & technological foundations方面:To innovate at the systems level and have impact in the real world, the discipline of electrical and computer engineering sits on strong technological and theoretical foundations.下面我们看一下Stanford 的EEEE主要研究方向如下:(主要研究嵌入式,电路,器件)Physical Technology & ScienceSubareas: Integrated Circuits and Power Electronics; Biomedical Devices, Sensors and Systems; Energy Harvesting and Conversion; Photonics, Nanoscience and Quantum Technology; Nanotechnology, Nanofabrication and NEMS/MEMS; Electronic DevicesInformation Systems & ScienceSubareas: Control & Optimization; Information Theory & Applications; Communications Systems; Societal Networks; Signal Processing & Multimedia; Biomedical Imaging; Data ScienceHardware/Software SystemsSubareas: Energy-efficient Hardware Systems; Software Defined Networking; Mobile Networking; Secure Distributed Systems; Data Science; Embedded Systems; Integrated Circuits & Power Electronics还有一些交叉的研究:Biomedical:Subareas: Biomedical Devices, Sensors and Systems; Photonics,Nanoscience and Quantum Technology; Nanotechnology & NEMS/MEMS; Biomedical Imaging; Information Theory & ApplicationsEnergy:Subareas: Control & Optimization; Energy-Efficient Hardware Systems; Integrated Circuits & Power Electronics; Energy Harvesting & Conversion下面来看一下Berkeley的EECSBerkeley 的department of EECS提供MS,PhD (in EE和CS), 申请EE的申请者需要申请EECS,申请CS的申请者需要申请CS.EECS的research areas 如下:Artificial Intelligence (AI) 人工智能Computer Architecture & Engineering (ARC) 计算机建筑与工程Biosystems& Computational Biology (BIO) 生物系统与计算机生物学Control, Intelligent Systems, and Robotics (CIR) 控制,智能系统和机器人Communications & Networking (COMNET) 通讯与网络Database Management Systems (DBMS) 数据库管理系统Design, Modeling and Analysis (DMA) 设计,模型和分析Education (EDUC) 教育Energy (ENE) 能源Graphics (GR)图形Human-Computer Interaction (HCI) 人机交互Integrated Circuits (INC) 集成电路Micro/Nano Electro Mechanical Systems (MEMS)微纳米机电系统Operating Systems & Networking (OSNT) 操作系统与网络工程Physical Electronics (PHY) 物理电子Programming Systems (PS) 编程系统Scientific Computing (SCI) 科学计算Security (SEC) 安全Signal Processing (SP)信号处理Theory (THY) 理论其中属于CS 的研究方向如下Artificial Intelligence (AI)Database Management Systems (DBMS)Education (EDUC)Graphics (GR)Human-Computer Interaction (HCI)Operating Systems & Networking (OSNT)Programming Systems (PS)Scientific Computing (SCI)Security (SEC)Theory (THY)属于EE:Communications & Networking (COMNET)Control, Intelligent Systems, and Robotics (CIR)Integrated Circuits (INC)Micro/Nano Electro Mechanical Systems (MEMS)Physical Electronics (PHY)Signal Processing (SP)交叉领域:Biosystems& Computational Biology (BIO)Computer Architecture & Engineering (ARC)Design, Modeling and Analysis (DMA)Energy (ENE)关于申请难度问题:好学校申请难度CS大于ECE,EE中低档学校(40/50左右开始往后),EE的申请难度大于CS, 因为CS的学校多,EE的学校少。
外文翻译原文

Stability of hybrid system limit cycles: application to the compass gait biped RobotIan A. Hiskens'Department of Electrical and Computer EngineeringUniversity of Illinois at Urbana-ChampaignUrbana IL 61801 USAAbstractLimit cycles are common in hybrid systems. However the non-smooth dynamics of such systems makes stability analysis difficult. This paper uses recent extensions of trajectory sensitivity analysis to obtain the characteristic multipliers of non-smooth limit cycles. The stability of a limit cycle is determined by its characteristic multipliers. The concepts are illustrated using a compass gait biped robot example.1 IntroductionHybrid system are characterized by interactions between continuous (smooth) dynamics and discrete events. Such systems are common across a diverse range of application areas. Examples include power systems [l], robotics [2, 3], manufacturing [4] and air-traffic control [5]. In fact, any system where saturation limits are routinely encountered can be thought of as a hybrid system. The limits introduce discrete events which (often) have a significant influence on overall behaviour.Many hybrid systems exhibit periodic behaviour. Discrete events, such as saturation limits, can act to trap the evolving system state within a constrained region of state space. Therefore even when the underlying continuous dynamics are unstable, discrete events may induce a stable limit set. Limit cycles (periodic behaviour) are often created in this way. Other systems, such as robot motion, are naturally periodic.Limit cycles can be stable (attracting), unstable (repelling) or non-stable (saddle). The stability of periodic behaviour is determined by characteristic (or Floquet) multipliers. A periodic solution corresponds to a fixed point of a Poincare map. Stability of the periodic solution is equivalent to stability of the fixed point. The characteristic multipliers are the eigenvalues of the Poincare map linearized about the fixed point. Section 4 reviews the connection between this linearized map and trajectory sensitivities.Poincare maps have been used to analyse the stability of limit cycles in various forms of hybrid systems. However calculation of the underlying trajectory sensitivities has relied upon particular system structures, see for example [7, 8], or numerical differencing, for example [6]. This paper uses a recent generalization of trajectory sensitivity analysis [9] to efficiently detemine the stability of limit cycles in hybrid systems.A hybrid system model is given in Section 2. Section 3 develops the associated variational equations. This is followed in Section 4 by a review of stability analysis of limit cycles. Conclusions and extensions are presented in Section 5.2 ModelDeterministic hybrid systems can be represented by a model that is adapted from a differential-algebraic (DAE) structure. Events are incorporated via impulsive action and switching of algebraic equations, giving the Impulsive Switched (DAIS) modelwheren x R ∈ are dynamic states and my R ∈ are algebraic states;(.)δ is the Dirac delta;(.)u is the unit-step function;,:n mnj f h RR +→;(0)(),:i n mng gR R ±+→; some elements of each(.)gwill usually be identicallyzero, but no elements of the composite g should be identically zero; the()i g± aredefined with the same form as g in (2), resulting in a recursive structure for g;,dey yare selected elements of y that trigger algebraic switching and state reset(impulsive) events respectively;dyandeymay share common elements.The impulse and unit-step terms of the DAIS model can be expressed in alternative forms:Each impulse term of the summation in (1) can be expressed in the state reset formwhere the notation x+denotes the value of x just after the reset event, whilstx-andy-refer to the values of x and y just prior to the event.The contribution of each()i g± in (2) can be expressed aswith (2) becomingThis form is often more intuitive than (2).It can be convenient to establish the partitionswherex -are the continuous dynamic states, for example generator angles, velocities andfluxes;z are discrete dynamic states, such as transformer tap positions and protection relay logic states;λ are parameters such as generator reactances, controller gains and switching times. The partitioning of the differential equations f ensures that away from events,x -evolves according to .(,)x y f x --=, whilst z and λ remain constant. Similarly,the partitioning of the reset equationsjhensures thatx -and λ remain constantat reset events, but the dynamic states z are reset to new values given by(,)jh y x z--+=-. The model can capture complex behaviour, from hysteresis and non-windup limits through to rule-based systems [l]. A more extensive presentation of this model is given in [9].Away from events, system dynamics evolve smoothly according to the familiardifferential-algebraic modelwhere g is composed of(0)gtogether with appropriate choices of()i g- or()i g+ ,depending on the signs of the corresponding elements of yd. At switching events (2),some component equations of g change. To satisfy the new g = 0 equation, algebraic variables y may undergo a step change. Reset events (3) force a discrete change in elements of x. Algebraic variables may also step at a reset event to ensure g= 0 is satisfied with the altered values of x. The flows of and y are defined respectively aswhere x(t) and y(t) satisfy (l),(2), along with initial conditions,3 'Ikajectory SensitivitiesSensitivity of the flowsxφandyφto initial conditionsxare obtained bylinearizing (8),(9) about the nominal trajectory,The time-varying partial derivative matrices given in (12),(13) are known as trajectory sensitiuities, and can be expressed in the alternative formsThe formxx ,xy provides clearer insights into the development of thevariational equations describing the evolution of the sensitivities. The alternative form 0(,)x t x φ, 0(,)yt x φ highlights the connection between the sensitivities and the associated flows. It is shown in Section 4 that these sensitivities underlie the linearization of the Poincare map, and so play a major role in determining the stability of periodic solutions.Away from events, where system dynamics evolve smoothly, trajectory sensitivities 0xx andxy are obtained by differentiating (6),(7) withrespect to 0x.This giveswhere/xf x f≡∂∂, and likewise for the other Jacobian matrices. Note that,,,xyxyf fg gare evaluated along the trajectory, and hence are time varyingmatrices. It is shown in 19, 101 that the numerical solution of this(potentially high order) DAE system can be obtained as a by-product of numerically integrating the original DAE system (6),(7). The extra computational cost is minimal. Initial conditions forxx are obtained from (10) aswhere I is the identity matrix. Initial conditions for 0zy follow directly from(17),Equations (16),(17) describe the evolution of the sensitivitiesxx andxybetween events. However at an event, the sensitivities are generally discontinuous. It is necessary to calculate jump conditions describing the step change inxx andxy . For clarity, consider a single switching/reset event, so the model (1),(2) reduces(effectively) to the formLet ((),()x y ττ) be the point where the trajectory encounters the triggering hypersurface s(x,y) = 0, i.e., the point where an event is initiated. This point is called the junction point and r is the junction time. It is assumed the encounter is transversal.Just prior to event triggering, at time τ-, we haveSimilarly,,y x++are defined for time τ+, just after the event has occurred. It isshown in [9] that the jump conditions for the sensitivitiesxx are given byThe assumption that the trajectory and triggering hypersurface meet transversally ensures a non-zero denominator for 0x τ The sensitivitiesxy . immediatelyafter the event are given byFollowing the event, i.e., for t τ+>, calculation of the sensitivities proceeds according to (16),(17) until the next event is encountered. The jump conditions provide the initial conditions for the post-event calculations.4 Limit Cycle AnalysisStability of limit cycles can be determined using Poincare maps [11, 12]. This section provides a brief review of these concepts, and establishes the connection with trajectory sensitivities.A Poincark map effectively samples the flow of a periodic system once every period. The concept is illustrated in Figure 1. If the limit cycle is stable, oscillations approach the limit cycle over time. The samples provided by the corresponding Poincare map approach a fixed point. A non-stable limit cycle results in divergent oscillations. For such a case the samples of the Poincare map diverge.To define a Poincare map, consider the limit cycle Γshown in Figure 1. Let ∑ be a hyperplane transversal to Γ at*x. The trajectory emanating from*xwill again encounter ∑ at*xafter T seconds, where T is the minimum period of the limit cycle. Due to the continuity of the flowxφwith respect to initial conditions, trajectories starting on ∑ in a neighbourhood of*x. will, in approximately T seconds, intersect ∑ in the vicinity of*x. Hencexφand ∑define a mappingwhere()kT x ττ≈ is the time taken for the trajectory to return to ∑. Complete details can hefound in [11,12]. Stability of the Paincare map (22) is determined by linearizing P at the fixed point*x, i.e.,From the definition of P(z) given by (22), it follows that DP(*x) is closely related to thetrajectory sensitivities***(,)(,)xxT T x x xφφ∂≡∂. In fact, it is shown in [11] thatwhereσ is a vector normal to ∑.The matrix*(,)xT x φis exactly the trajectory sensitivity matrix after one period of the limitcycle, i.e., starting from*xand returning to*x. This matrix is called the Monodromymatrix .It is shown in [11] that for an autonomous system, one eigenvalue of *(,)xT x φ isalways 1, and the corresponding eigenvector lies along **(,)f y x The remaining eigenvalues*(,)xT x φof coincide with the eigenvalues of DP(*x ), and are known as the characteristicmultipliers mi of the periodic solution. The characteristic multipliers are independent of the choice of cross-section ∑ . Therefore, for hybrid systems, it is often convenient to choose ∑ as a triggering hypersurface corresponding to a switching or reset event that occurs along the periodic solution.Because the characteristic multipliers mi are the eigenvalues of the linear map DP(x*), they determine the stability of the Poincarb map P(kx), and hence the stability of the periodic solution.Three cases are of importance: 1. Alli m lie within the unit circle, i.e., 1im<,i ∀.The map is stable, so the periodicsolution is stable. 2. Allim lie outside the unit circle. The periodic solution is unstable.3. Someim lie outside the unit circle. The periodic solution is non-stable.Interestingly, there exists a particular cross-section*∑, such thatwhere *ς∈∑.This cross-section*∑is the hyperplane spanned by the n - 1 eigenvectors of*(,)xT x φthat are not aligned with **(,)f y x . Therefore the vector *σthat is normal to*∑ is the left eigenvector of *(,)xT x φ corresponding to the eigenvalue 1. The hyperplane*∑is invariant under*(,)xT x φ, i.e., **(,)f y x maps vectors *ς∈∑back into*∑.5 ConclusionsHybrid systems frequently exhibit periodic behaviour. However the non-smooth nature of such systems complicates stability analysis. Those complications have been addressed in this paper throughapplication of a generalization of trajectory sensitivity analysis. Deterministic hybrid systems can be represented by a set ofdifferential-algebraic equations, modified to incorporate impulse (state reset) action and constraint switching. The associated variational equations establish jump conditions that describe the evolution of sensitivities through events. These equations provide insights into expansion/contraction effects at events. This is a focus of future research.Standard Poincar6 map results extend naturally to hybrid systems. The Monodromy matrix is obtained by evaluating trajectory sensitivities over one period of the (possibly non-smooth) cyclical behaviour. One eigenvalue of this matrix is always unity. The remaining eigenvalues are the characteristic multipliers of the periodic solution. Stability is ensured if all multipliers lieReferences[l] LA. Hiskens and M.A. Pai, “Hybrid systems view of power system modelling,” in Proceedings of the IEEE International Symposium on Circuits and Systems, Geneva, Switzerland, May 2000.[2] M.H. Raibert, Legged Robots That Balance, MIT Press, Cambridge, MA, 1986.[3] A. Goswami, B. Thuilot, and B. Espiau, “A study of the passive gait of a compass-like biped ro bot: symmetry and chaos,’’ International Journal of Robotics Research, vol. 17, no. 15, 1998.[4] S. Pettersson, “Analysis and design of hybrid systems,” Ph.D. Thesis, Department of Signals and Systems, Chalmers University of Technology, Goteborg, Sweden, 1999.[5] C. Tomlin, G. Pappas, and S. Sastry, “Conflict resolution for air traffic management:A study in multiagent hybrid systems,” IEEE Transactions on Automatic Control, vol. 43, no. 4, pp. 509-521, April 1998.[6] A. Goswami, B. Espiau, and A. Keramane, “Limit cycles in a passive compass gait biped and passivity-mimicking contr ol laws,” Journal of Au tonomous Robots, vol. 4, no. 3, 1997. 171 B.K.H. Wong, H.S.H. Chung, and S.T.S. Lee, ‘Computation of the cycle state-variable sensitivity matrix of PWM DC/DC converters and its applica tion,” IEEE Transactions on Circuit s and Systems I, vol. 47, no. 10, pp. 1542-1548, October 2000.[8] M. Rubensson, B. Lennartsson, and S. Petters son, “Convergence to limit cycles in hybrid systems - an example,” in Prepri nts of 8th International Federation of Automatic Control Symposium on Large Scale Systems: Theo y d Applications, Rio Patras, Greece, 1998, pp. 704-709.[9] I.A. Hiskens and M.A. Pai, “Trajectory sensitivity analysis of hyhrid systems,” IEEE Transactions on Circuits and Systems I, vol. 47, no. 2, pp. 204-220, February 2000.[10]D. Chaniotis, M.A. Pai, and LA. Hiskens, “Sen sitivity analysis of differential-algebraic systems using the GMRES method - Ap plication to power systems,” in Proceedings of the IEEE International Symposium on Circuits and Systems, Sydney, Australia, May 2001.[11]T.S Parker and L.O. Chua, Practical Numerical Algorithms for Chaotic Systems, Springer-Verlag, New York, NY, 1989.[12]R. Seydel, Practical Bifurcation and Stability Analysis, Springer-Verlag. New York, 2nd edition, 1994.。
- 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。
S-MAC cont’d
• Adaptive listening – To resolve sleep latency – Potential delay on every hop! • Basic idea – Let the nodes who overhear its neighbor’s transmission (ideally RTS or CTS) go to sleep and wake up for a short period at the endAC cont’d
• Additional collision avoidance – Record transmission time as NAV (Network Allocation Vector) – Check if NAV = 0 before transmission attempt – NAV = 0 indicates ongoing transmission is over
• Flat topology – Virtual cluster comprises nodes with common schedule – No coordination from cluster head
Accommodate both schedules
Schedule 1
Schedule 2
• General view on MAC protocol in WSN – Requirements – Problems • MAC protocols for WSN • Conclusion
General Purposes of MAC Protocol
• MAC protocol is to ensure that the channel can be accessed by multiple users, dealing with the situation of interference. • Has a direct bearing on how reliably and efficiently data can be transmitted
S-MAC cont’d
• Synchronization period – Nodes exchange schedules by sending SYNC packets to immediate neighbors at their scheduled listen time
S-MAC cont’d
S-MAC cont’d
• Cons: – Sleep latency – Active time must be long enough to handle expected highest load, inefficient when load is lower. • Essentially S-MAC trades energy with latency
overhearing
Tx
data
Rx
overhearing
Traffic Pattern
• Local broadcast – Schedule exchange/update between neighbors – Omni-directional transmission is desired • Nodes to sink report – Payload and signaling – In favor of directional transmission
S-MAC cont’d
– New schedule received by node A • Discard current schedule if node A has no other neighbor or • Adopt both schedules if node A has other neighbors Accommodate both schedules
S-MAC cont’d
• Coordinating sleeping – listen for a fixed amount of time • Following the existing schedule or • Establishing new schedule if no schedule exists and announcing new schedule by SYNC
Requirement of WSN
• Application-level performance is the goal as opposed to per-node fairness • Long battery life • Delay – Surveillance, Low traffic, Regular update – Emergency, Quick response, Bursty heavy load
T-MAC cont’d
Active Sleep Active Sleep Active Sleep
TA Frame 1
TA Frame 2
TA Frame 3
• An active period ends when no activation event has occurred for a time TA
S-MAC cont’d
• Message passing – Divide long messages into small fragments – Transmit fragments in a burst through reserved channel
S-MAC cont’d
• Features: – Loose synchronized due to large scale of intervals, no need for precise synchronization – Virtual cluster – Adaptive listen to reduce sleep delay – Message passing
T-MAC (Timeout-MAC)
• Minimize idle listening • Using timeout to be adaptive to traffic during wakeup period • Transmitting all messages in burst of variable length • RTS/CTS provides both collision avoidance and reliable transmission
T-MAC cont’d
• Definition of activation event: – The firing of a periodic frame timer (the beginning) – The reception of any data on the radio – The sensing of traffic (during collision) – The end-of-transmission of a node’s own data packet or acknowledgement – End of neighbor’s transmission (knowledge from prior overhearing)
Classification
• Basic idea – Control the active time of radio • Control the times of on-off switches • Scheduling: time-slotted system, wake-up by schedule, clock shift can be disastrous • LPL (Low Power Listening): preamble sampling, wake-up tone
Overhearing
• Receiving packets that are not destined to the node • Interception, waste of energy in receiving, error responding will cause potential collision
Requirement of WSN cont’d
• Adaptive to changing topology – Nodes die out – Mobile nodes • Applicable with limited computing capability
MAC Protocol for WSN
T-MAC cont’d
• Node successfully transmitting 3 packets
T-MAC cont’d
• Determining TA – RTS starts transmission, the TA should be at least long enough to hear the CTS – TA > RTS + CTS + Turn_around_Time
Frame
S-MAC cont’d
Wake-up sleep
Frame
– Total frame length is limited by latency requirement, buffer space, and active time – Active time depends on message rate; fragmentation is used while transmitting large messages
Energy Efficient MAC Protocol
for Wireless Sensor Network