ABSTRACT Title of Document: A SUNLIGHT TO MICROWAVE POWER TRANSMISSION MODULE PROTOTYPE FOR SPACE SOLAR POWER Paul Iven Jaffe, Doctor of Philosophy, 2013 Directed By: Professor Victor L. Granatstein, Department of Electrical and Computer Engineering The prospect of effectively limitless, continuous electricity from orbiting satellites for use on earth has captured many people?s interest for decades. The proposed approach typically entails collection of solar energy, its conversion to microwave energy, and the wireless transmission of the microwaves to the earth. This offers the benefit of providing baseload power while avoiding diurnal cycle and atmospheric losses associated with terrestrial solar power. Proponents have contended that the implementation of such systems would offer energy security, environmental, and broad technological advantages to those who would undertake their development, while critics have pointed out economic, political, and logistical barriers. Niche applications, such as provision of power to remote military bases, might better tolerate the higher energy costs associated with early operational systems. Among recent implementations commonly proposed for solar power satellites, highly modular concepts have received considerable attention. Each employs an array of modules for performing conversion of sunlight into microwaves for transmission to earth. This work details results achieved in the design, development, integration, and testing of photovoltaic arrays, power electronics, microwave conversion electronics, and antennas for 2.45 GHz microwave-based ?sandwich" module prototypes. Prototypes were fabricated and subjected to the challenging conditions inherent in the space environment, including solar concentration levels in which an array of modules might be required to operate. This testing of sandwich modules for solar power satellites in vacuum represents the first such effort. The effort culminated with two new sandwich module designs, ?tile? and ?step?, each having respectively area-specific masses of 21.9 kg/m2 and 36.5 kg/m2, and mass-specific power figures of 4.5 W/kg at minimum one sun and 5.8 W/kg at minimum 2.2 suns (AM0) simulated solar illumination. The total combined sunlight to microwave efficiency of the modules was shown to be on the order of 8% and 7% for vacuum operation in the 10-6 torr regime. These represent the highest reported combined sandwich module efficiencies under either ambient or vacuum conditions, nearly quadrupling the previous efficiency record. The novel ?step? concept was created to address thermal concerns and resulted in a patent publication. Results from module characterization are presented in context and compared with figures of merit, and practical thresholds are formulated and applied. The results and discussion presented provide an empirical basis for assessment of solar power satellite economic models, and point to several opportunities for improvements in area-specific mass, mass-specific power, and combined conversion efficiency of future prototypes. A SUNLIGHT TO MICROWAVE POWER TRANSMISSION MODULE PROTOTYPE FOR SPACE SOLAR POWER By Paul Iven Jaffe Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Doctor of Philosophy 2013 Advisory Committee: Professor Victor L. Granatstein, Chair/Advisor Professor Martin C. Peckerar Professor John Melngailis Professor Jeremy Munday Professor David Akin ii Dedication This work is dedicated to my wife, Rachael Schoenbaum, and our two young sons, Jules and Elliott. Without their unflagging support, it surely would not have been possible. iii Acknowledgements First and foremost, I thank my advisor, Professor Victor Granatstein. His sustained encouragement and confidence in my efforts buoyed me through many challenging times when I might otherwise have faltered. The debt I owe him I will strive to pay forward by assisting others in their own research and academic pursuits. The outstanding and engaged faculty and staff in the graduate office of the University of Maryland were key enablers of my work and studies, including Professor Martin Peckerar, Dr. Tracy Chung, Melanie Prange, and Maria Hoo. My journey and interest in this research area might never have started were it not for the inspiration and comprehensive technical overview supplied by John Mankins, former manager of Advanced Concepts Studies at NASA. His infectious enthusiasm, persistence, and tremendous depth and breadth of knowledge should serve as examples to researchers and program managers everywhere. The creation of the climate conducive for my study of this topic is traceable to the efforts of Dr. Gerry Borsuk, and the encouragement in the initial development of the proposal came from Dr. Jill Dahlburg and Dr. Rob Walters. I had the good fortune of receiving selfless and extensive comments on the original research proposal, as well as feedback during the course of the project and the thesis development process from luminaries in the field, notably Dr. James McSpadden, Professor Nobuyuki Kaya, and Dr. Frank Little. This endeavor could never have commenced without the opportunities afforded by my employer, the U. S. Naval Research Laboratory (NRL). My supervisor, Mark Johnson, and my organization?s management have been steadfast and accommodating iv in their support of my development and of the project itself, including Bob Towsley, Bill Raynor, Jerry Golba, John Schaub, Pete Wilhelm, and Dr. John Montgomery. Key administrative support was provided by Nancy Peaper and Arthur Espanta. The funding for this work came from an NRL Base Program 6.2 research effort. The NRL team that supported the development and testing effort is large and accomplished. Jason Hodkin supplied critical RF design prowess, Dr. Mike Nurnberger provided the antenna design, Clark Person delivered the highly efficient power electronics, Forest Harrington drafted the many iterations of the mechanical design, Bang Nguyen performed detailed thermal analyses, Susie LaCava managed the essential solar array development and carefully proofread this thesis, Dave Scheiman was invaluable in the creation and understanding of the solar simulation approach, Kathy Seymour assembled and reworked the module electronics, Mike Freeman guided our thermal feature and instrumentation implementation, and Trevor Specht and Ray Dixon aided the interfacing of all our test equipment to the facility?s electrical mains. Additional invaluable assistance came from Mike Brown, Kwok Cheung, Sheleen Spencer, Matthew Long, and Jim Pirozzoli. Phil Jenkins, Maria Gonzalez, Dr. John Pasour, Dr. Baruch Levush, George Flach, Dr. Kieran Carroll, Doug Sinclair, Sean Lynch, and Professor Kameswara Rao Bhamidipaty provided subject matter expertise, sage and vital inputs, and valuable advice at crucial points in the project. Numerous fabrication and test facility technicians played imperative parts in helping everything come together safely and effectively, including Mike Van Herpe, v Chris Calder, Paul Peffers, Tim Wilson, Bernie Lafrance, Paul Stencel, Mickey Dougherty, Frank Trimble, Mike Derosa, and others. Mark Kanawati and Dr. Dino Lorenzini of SpaceQuest, Ltd. were incredibly knowledgeable, patient, and flexible as our solar array needs evolved, and they delivered outstanding hardware that from any other source likely would not have met the technical requirements or would have exceeded the available resources. Many students performed indispensable roles during the project: Andrew Han was instrumental from its inception to completion, Grant Stewart was unparalleled in his ability to simply get things done, Karina Hemmendinger diligently kept the integrated testing on track, Ethan Hettwer developed and executed test scripts for the earliest RF environmental testing, Owen Nugent contributed rectenna work, Kate Aplin aided with background research and proposal review, Daniel Rhoades developed the original light field characterization methods, Sanjeet Das executed the pathfinding power beaming demonstration, Michael Long worked through endless solar panel geometries, and many others assisted as well, including but not limited to: Rashaad Patterson, Donald Thomas, Evan Siefring, Kenyetta Washington, Mike Taylor, Brad Segelhorst, Thanh Pham, Magda Moses, and Ben Adducci. I?ve gleaned tremendous knowledge and insights from other young researchers in the field, including Dr. Martin Leitgab, Ian McNally, Corey Bergsrud, and Bea Adkins. Dr. Seth Potter and Dr. Peter Schubert have also provided great professional inspiration. Colleagues who provided support and who were patient with me as I was consumed by this project and other responsibilities include Dr. Rich Fischer, vi Professor Steve Blank, Professor Reza Zekavat, Darel Preble, Steve Leete, and Bert Murray. I would be remiss if I did not acknowledge just a few of the many influential and inspirational teachers I?ve had over the years, including Professor Dan Jablonski, Professor Dan Dockery, Professor Chuan Sheng Liu, Professor Leonard Taylor, Professor Kazuo Nakajima, Judy Parsons, Ursula Alexander, Diane Albertini, and Barbara Case. While the outstanding staff at Joint Base Anacostia Bolling Child Development Center II provided first-rate childcare, the librarians of the community library provided a quiet place to think and write for countless hours, and free popcorn on some Fridays. Though I?m sure I?ve forgotten many others who made it possible, I will never forget my family, parents, and grandparents, who in large part made me who I am today by instilling and nurturing within me an insatiable curiosity and a love of learning. vii Table of Contents Dedication ..................................................................................................................... ii? Acknowledgements ...................................................................................................... iii? Table of Contents ........................................................................................................ vii? List of Tables ............................................................................................................... ix? List of Figures ............................................................................................................... x? Chapter 1: Introduction ................................................................................................. 1? Motivation ................................................................................................................. 1? Historical Perspective ............................................................................................... 3? Solar Power Satellite ?Reference System? ............................................................... 7? Modular Implementation Concepts ........................................................................... 9? The Sandwich Module ............................................................................................ 10? Goals ....................................................................................................................... 16? Chapter 2: Sandwich Module Prototype Development .............................................. 18? Thermal Analysis .................................................................................................... 18? Step Module Concept .............................................................................................. 23? Critical Tradeoffs .................................................................................................... 27? Photovoltaics ....................................................................................................... 27? DC-to-RF conversion .......................................................................................... 32? Antenna Elements ............................................................................................... 38? Module Architectures .......................................................................................... 41? Thermal Control Methods ................................................................................... 41? Design Iteration ....................................................................................................... 43? Module Fabrication ................................................................................................. 43? Chapter 3: Prototype Testing ...................................................................................... 45? Progressive Testing ................................................................................................. 45? Sun Simulation ........................................................................................................ 47? Xenon Light Source ............................................................................................ 47? Light Attenuating Screens................................................................................... 50? Light Field Characterization ............................................................................... 52? Space Environment Simulation ............................................................................... 58? Fused Silica Window .......................................................................................... 59? Thermal Vacuum Chamber ................................................................................. 62? Summary Comparison of Space, Simulated Space, and Ambient Environments ... 69? Supporting Equipment and Configuration for Module Testing .............................. 70? Chapter 4: Results and Discussion .............................................................................. 75? Overview ................................................................................................................. 75? Effects of Varying Illumination Conditions and Vacuum ...................................... 75? Ambient Testing with Varying Illumination ....................................................... 75? viii Vacuum Testing with Varying Illumination ....................................................... 77? Key Figures of Merit and Results ........................................................................... 81? Collect/Transmit Area-Specific Mass [kg/m2] .................................................... 81? Mass-Specific Transmitted Power [W/kg] .......................................................... 84? Combined Conversion Efficiency ....................................................................... 85? Additional Figures of Merit and Module Qualities of Interest ............................... 88? Chapter 5: Potential Economic Thresholds ................................................................. 90? Generalized Comparison of Energy Sources .......................................................... 90? Economic Analyses for Solar Power Satellites ....................................................... 91? Chapter 6: Conclusions ............................................................................................. 102? Summary ............................................................................................................... 102? Implications of Present Work ............................................................................... 103? Future Work .......................................................................................................... 103? Specialized Test Facility Development ............................................................ 103? Antenna Characterization .................................................................................. 105? Mass Reduction ................................................................................................. 106? Thermal Management Improvements ............................................................... 107? Thermal Instrumentation Improvements ........................................................... 110? Efficiency Enhancement ................................................................................... 111? Additional Module Functionality ...................................................................... 112? Alternative Sandwich Module Approaches ...................................................... 112? Appendix A: Microwave Power Transmission ......................................................... 115? Appendix B: SPS System Design ............................................................................. 123? References ................................................................................................................. 125? ix List of Tables Table 1: Projected RF signal and efficiency chain performance for tile module. ...... 35? Table 2: Screen designations with blockage and throughput percentages from manufacturer specifications. ....................................................................................... 50? Table 3: Screen designations and sun concentrations for initial tile module testing focus settings. .............................................................................................................. 52? Table 4: Comparison of Space and Simulated Space Environments for Sandwich Module Operation ....................................................................................................... 70? Table 5: Tile module mass breakdowns.. .................................................................... 83? Table 6. Efficiencies for the 2.45 GHz prototype tile module in vacuum at about one sun illumination (117W input over 0.087m2). ............................................................ 87? Table 7: Minimal inputs for a space solar power cost model ..................................... 92? Table 8: Four cases with varying assumptions to generate Levelized Cost of Energy values for solar power satellites .................................................................................. 97? Table 9: Comparison of simple solar power satellite Levelized Cost of Energy cases with conventional sources in $/MWh. Conventional energy data from the U. S. Energy Information Administration ............................................................................ 98? Table 10: Comparison of fully burdened cost of fuel kWh kerosene-based jet fuel equivalents to solar power satellite levelized cost of energy model results. ............ 100? Table 11. Comparison of selected means of amplification and DC to RF conversion. ................................................................................................................................... 120? Table 12: System designs examined by the URSI report ......................................... 124? x List of Figures Figure 1: Space Solar Power segments and efficiency figures adapted and updated from the DOE/NASA studies........................................................................................ 6? Figure 2: A recent 5.8 GHz derivative by the Solar High Study Group of the original 1978 DOE/NASA SPS Reference System concept.. .................................................... 8? Figure 3: Modular Symmetric Concentrator concept, circa 2007............................... 10? Figure 4: Depiction of the functional layers of the sandwich module. ....................... 11? Figure 5: Sandwich module layers showing subfunctions. ......................................... 12? Figure 6: The subset of sandwich module functions implemented in this research effort. ........................................................................................................................... 16? Figure 7: Solar power intercepted at one sun (AM0) by a 28 cm by 28 cm module and combined module efficiency with notional layer efficiency estimates. ...................... 19? Figure 8: Radiator area required to maintain temperature equilibrium for a 28 cm by 28 cm module at 23% efficiency. ............................................................................... 20? Figure 9: Temperature of a 28 cm by 28 cm module with both sides as black body radiators for various module efficiencies. ................................................................... 21? Figure 10: Thermal Desktop? simulation of a sandwich module under 3 suns of illumination . ............................................................................................................... 22? Figure 11: Photovoltaics and transmit antenna comprised of tile modules. ............... 24? Figure 12: Photovoltaics and transmission antenna comprised of step modules. ....... 24? Figure 13: Thermal Desktop? simulation of a step module under 3 suns of illumination. ................................................................................................................ 26? Figure 14: Layer stackup and I-V curve for the Spectrolab UTJ photovoltaic cells from the datasheet. ...................................................................................................... 28? Figure 15: Conversion module solar panel for a tile module. ..................................... 29? Figure 16: Conversion module solar panel for a step module. ................................... 29? Figure 17: Representative array string I-V and power data plot collected with direct sun-simulated illumination.......................................................................................... 30? Figure 18: I-V curves showing the effect of temperature on panel open circuit voltage with lamp output attenuated by screens to produce about one sun. ............................ 31? Figure 19. DC power and RF electronics baseplate for the tile module. .................... 33? Figure 20. Characterization of Power Added Efficiency performance of final stage amplifier. ..................................................................................................................... 34? Figure 21: Step module prototype with antenna mockup, electronics shown at left. . 36? Figure 22: A representative spectrum analyzer screen capture from monitoring of the RF bandwidth, harmonics, and center frequency. This capture is for the step module while powered by a solar array simulator. .................................................................. 37? Figure 23: Simulated surface currents of short backfire antenna . ............................. 39? Figure 24: Simulated gain pattern for a short backfire antenna at 2.45 GHz . ........... 40? Figure 25: Integrated tile module showing from top to bottom: solar array, conversion electronics with multilayer thermal blankets and black Kapton? tape, and antenna mockup. ....................................................................................................................... 42? Figure 26: Tile module integration and test flow overview. ....................................... 46? Figure 27: 4,000W xenon light source with power supply. ........................................ 48? xi Figure 28: Spectrum of 4,000W L.P. Associates xenon lamp. ................................... 49? Figure 29: The five light attenuating screens labeled with percent open area. ........... 51? Figure 30: BeamGage software screen capture. ......................................................... 53? Figure 31: A light field portion matching the solar array size that has been processed to show the number of suns of intensity incident on each cell area, adapted from . .. 55? Figure 32: Two lamp beam mapping showing equivalent number of suns of intensity over solar cell areas, adapted from . ........................................................................... 56? Figure 33: UV Grade Fused Silica Transmission Curve ........................................... 59? Figure 34: Solar Radiation Spectrum ......................................................................... 60? Figure 35: The fused silica window used with the vacuum chamber. ........................ 61? Figure 36: DC and RF electronics in thermal vacuum chamber with fused silica window sealing chamber opening. .............................................................................. 62? Figure 37: Thermal vacuum chamber prior to fused silica window installation. ....... 63? Figure 38: Module prototype in thermal vacuum chamber with fused silica window sealing chamber opening and vacuum chamber internals visible. .............................. 64? Figure 39: Liquid nitrogen shroud cooling apparatus. ................................................ 66? Figure 40: Depiction of step module in vacuum chamber configuration ................... 68? Figure 41: Step module installed in vacuum chamber for testing prior to installation of the fused silica window. ......................................................................................... 69? Figure 42: DC and RF electronics test support equipment. ........................................ 71? Figure 43: Subset of tile module temperature points collected during a representative vacuum test. ................................................................................................................ 72? Figure 44: Insertion loss measurement of the step module RF power measurement configuration. .............................................................................................................. 73? Figure 45: Test configuration for vacuum testing with illumination for the tile module ..................................................................................................................................... 74? Figure 46: Tile module RF conversion efficiency and solar array temperature during testing at ambient pressure under various illumination conditions. ............................ 76? Figure 47: Tile module RF conversion efficiency, solar array power, and RF output power vs. time with various illumination conditions at ambient pressure .................. 77? Figure 48: Tile module electronics efficiency, solar array output power, solar array voltage, and RF output power as a function of different operating conditions. Each cluster of three points represents the mean, minimum, and maximum. ..................... 78? Figure 49: Tile module temperatures as a function of different operating conditions 80? Figure 50: Mass contributions of major categories of tile sandwich module parts. ... 84? Figure 51: NRL RF Anechoic Chamber .................................................................. 106? Figure 52. Power collection efficiency as a function of ? with optimum power taper over the transmitting aperture . ................................................................................. 117? Figure 53. One-way sea level to zenith (straight up through the entire atmosphere) attenuations in clear sky conditions . ........................................................................ 118? Figure 54. Rectenna components . ............................................................................ 122? 1 Chapter 1: Introduction Motivation Global climate change and the consequent need for energy sources that avoid further contributions to climate degradation loom as significant societal concerns. It is widely realized that many sources of fossil fuels are either at risk of depletion or increasingly undesirable because of their contributions to greenhouse gases and growing scarcity. While many carbon-free or nearly carbon-free energy alternatives exist, they often suffer from significant problems such as intermittency, lack of scalability, locale dependence, or safety risks. One promising clean power source is the sun, which has an effectively unlimited energy supply. However, terrestrial collection of solar energy poses problems. The diurnal cycle, atmospheric attenuation, and weather effects all diminish access to solar power. Because of its intrinsic unpredictability, terrestrial solar power necessitates the implementation of some means of energy storage or use in conjunction with one or more predictable sources of power to achieve system viability. Collection of solar energy in space via satellite coupled with its conversion to microwaves for transmission to the ground largely overcomes these limitations, but it poses formidable engineering challenges and serious questions of economic plausibility. Though solar energy provided to the earth from space had been principally been considered in the past only for utility grid cases, the past decade has 2 seen greater interest in niche applications that could tolerate the higher expenses that would likely be associated with an initial capability. Such instances include power provision to remote locations with minimal infrastructure, military bases in forward areas, and areas devastated by natural disasters. A premium cost for power could be justifiably borne in these and other cases, particularly if it is still less expensive, more reliable, or more sustainable than existing alternatives. If such an implementation were pursued, it could serve to aid in the development and refinement of the needed technologies to improve the economic viability for the utility grid case. Solar power satellite (SPS) (also known as space solar power (SSP)) concepts have been examined in depth on several occasions in the past, and interest has been renewed in recent years in part because of improvements in a number of supporting technologies. These include: increased solar cell efficiency, increased solid state power amplifier efficiency, large space structures advances, and technology developments for robotic assembly in space. Recent solar power system studies have been significantly limited in their ability to accurately determine the costs and challenges of deploying an operational system by the small amount of actual hardware development that has been done to show the feasibility of key SPS technological elements. Recent space solar power system designs of widespread interest capitalize on highly modular architectures. However, assessments of their technical soundness are hampered by a dearth of substantive efforts to identify and resolve concerns about their component technologies, most notably those pertaining to the sandwich module. 3 The motivation for this research was the need for a critical examination of the challenges associated with sandwich module development. Historical Perspective The prospect of collecting a continuous, massive amount of solar energy in orbit and transferring it via wireless power transmission for use on earth has held the interest of a significant group of advocates and researchers for decades. Light from the sun is more intense in space because it is not attenuated by clouds or the atmosphere, and a satellite in geosynchronous orbit is illuminated essentially year- round, whereas terrestrial solar power systems must contend with seasonal sunlight variation and nighttime. Widely recognized as physically possible but economically prohibitive, interest in space solar power has resurged in recent years as a consequence of increased media attention resulting from government and non- governmental organization reports, as well as through efforts by private companies and national space agencies to develop or stimulate the development of practical space solar power systems. Thoughtful and reasoned criticisms [1][2] and counter- criticisms [3] have offered excellent summary insights into some of the major issues. Major funded studies were conducted by the U.S. Department of Energy (DOE) and the National Aeronautics and Space Administration (NASA) during the late 1970s [4], the National Research Council in 1981 [5], and again by NASA in 1995 and 1999 in the form of the ?Fresh Look? study [6] and Space Solar Power Exploratory Research and Technology (SERT) program [7], respectively. Many other funded and unfunded studies have been undertaken by the European Space Agency (ESA) [8], the Japanese Space Agency (JAXA) [9], The International Union 4 of Radio Science (URSI) [10], the U.S. National Security Space Office (NSSO) [11], the U.S. Naval Research Laboratory (NRL) [12], and numerous other national and international organizations [13]. In late 2011, the International Academy of Astronautics published ?Green Energy from Space Solar Power - The First International Assessment of Space Solar Power: Opportunities, Issues and Potential Pathways Forward? [14]. There have also been many books written about SSP, including the comprehensive text Solar Power Satellites [15] edited by Peter Glaser, the original patent holder of the SSP concept, and an excellent recently published introductory text by Flournoy [16]. As of August 2013, approximately half a dozen corporate entities are explicitly endeavoring to develop space solar power systems, including PowerSat [17], Mitsubishi Electric [18], Solaren Space [19], and Space Energy [20]. Several different approaches to SSP have been proposed. Fundamentally, each has a means of solar energy collection and a method for conveying the collected energy to the ground. For collection, photovoltaics and solar thermal have principally been considered. For transmission, microwave frequencies and lasers have been examined. Solar-pumped lasers and large spaceborne mirrors have been proposed as combined collection and transmission schemes. Each collection and transmission scheme offers distinct advantages and disadvantages. Photovoltaics (PV) are considered for their comparative reliability and simplicity, while solar thermal is advocated for its theoretical ability to achieve efficiencies transcending the Shockley?Queisser limit [21] that bounds photovoltaics. Laser transmission is cited for its ability to utilize smaller transmitter and receiver apertures, whereas microwave transmission is 5 favored for its greatly reduced susceptibility to attenuation from tropospheric effects. Microwave power transmission has been investigated extensively for space solar power applications and enjoys many decades of terrestrial demonstrations, as documented in detail by William Brown [22]. A discussion of microwave power transmission is included as Appendix A. Common technical concerns for all SSP implementations include total space segment mass, transmitted energy density on the ground, conversion efficiencies of the space and ground segments, power beam pointing and control, interaction between the power beam and the ionosphere and troposphere, and electromagnetic compatibility with other satellites and terrestrial services. These concerns and many others were examined exhaustively in the DOE/NASA studies of the 1970s, and in many cases works performed in association with these studies or their derivatives remain the exemplars for any future efforts. One such instance is the reference system functional breakdown and efficiency chain for the photovoltaic collection/microwave transmission scheme, an adaptation of which can be seen in Figure 1, notionally at 2.45 GHz. The efficiency figures are largely unchanged from the original assessment, with the notable exception of the PV efficiency. These are now on the order of 30% for commercially available space-rated photovoltaic cells, versus the 15% figure used in the DOE/NASA study. 6 Figure 1: Space Solar Power segments and efficiency figures adapted and updated from the DOE/NASA studies [23]. In this work, the focus was principally on the development and testing of a prototype that implements the first three segments from Figure 1: photovoltaics, DC- to-RF conversion, and the antenna. The other segments are addressed only to provide context and illustrate some of the considerations that would go into a complete system. Over a dozen major classes of solar power satellite (SPS) architectures have been proposed by different researchers. The most relevant to this work are those that employ photovoltaics for solar energy collection and microwave power transmission to deliver the energy to the earth, and in particular those that depend on the utilization of ?sandwich? modules. The attraction to this specific approach can be appreciated by 7 examining some selected forerunners and understanding their limitations and challenges. Solar Power Satellite ?Reference System? Any solar power satellite would have a large number of subsystems and design considerations. Though largely beyond the scope of this work, Appendix B: SPS System Design delves into general considerations such as space and ground aperture sizing and transmit frequency selection. For this discussion, focus is confined primarily to the DOE/NASA ?Reference System? and its variants. Figure 2 shows a recent derivative by the Solar High Study Group of the DOE/NASA ?Reference System? that arose from the eponymous studies, and which features the combination of photovoltaics and microwave power transmission. It consists of an enormous satellite in geosynchronous orbit (GEO) with separate solar collection and power transmission surfaces. These are connected by wire harnessing and a slip ring mechanism that must transfer thousands of amps of current. The satellite would send many GWs of electricity to the utility grid via a microwave downlink operating at either 2.45 or 5.8 GHz, which would be collected by large rectifying antenna (rectenna) receiving stations [24]. This approach fits within the ?perpendicular to orbital plane? class of SPS. For each perpendicular to orbital plane concept, the solar collection surface rotates on an axis perpendicular to the orbital plane in order to track the sun, and collected energy is routed to one or more transmission antennas that point at the earth to direct the energy beam. 8 Figure 2: A recent 5.8 GHz derivative by the Solar High Study Group [25] of the original 1978 DOE/NASA SPS Reference System concept. A 525 m diameter transmit antenna version was also proposed. The collection surface and the plane containing the antenna aperture must necessarily be pointed independently of each other, necessitating the aforementioned slip ring mechanism or some other means of energy redirection. Naturally, wiring to route the power around the satellite must be employed and this wiring undesirably contributes Transmit Antenna PV Panels 9 to the overall spacecraft mass. The slip ring mechanism poses intrinsic reliability concerns due to its representing a potential single point of failure, especially at high operating powers. These shortcomings led researchers to investigate alternatives, like those that use sandwich modules, which address these possibly crippling problems. Modular Implementation Concepts Implementation schemes that avoid the mass and reliability failings of the Reference System and other previously proposed designs are those based heavily on modular system elements, such as the Modular Symmetrical Concentrator (MSC), a derivative of the Integrated Symmetrical Concentrator concept [26], and the SPS- Arbitrarily Large Phased Array (ALPHA) concept [27]. These approaches utilize optical energy routing and a microwave transmit aperture constructed from essentially identical elements. This avoids the need for a large, conductive rotating joint and limits wiring mass compared to historical reference concepts since the transmitters that relay the energy are located in close proximity to the photovoltaic cells that collect it. The use of modular elements offers the possibility of improved economy through mass production. Employing solar concentration could reduce the required launch mass and as a result lower the system cost, but it increases the magnitude of the thermal challenges. A depiction of a proposed MSC satellite is shown in Figure 3. Though this image shows a monolithic structure, it might also be possible to use several satellites flying in formation to dispense with the connecting structures. 10 Figure 3: Modular Symmetric Concentrator concept [11], circa 2007. The SPS-ALPHA concept takes modularity a step further and calls for effectively every element of the satellite to be modular. This concept is described in detail by Mankins in [27]. The Sandwich Module The key element in most modular SSP architectures is the sandwich module, an idea which had first been seriously investigated in association with the original DOE/NASA studies [28]. The sandwich module as originally conceived performs functions separable into three layers: solar energy collection and conversion to direct current electricity, generation of a microwave signal of suitable frequency and amplitude for transmission, and transmission of the microwave energy. An array of modules is envisioned to be used in the formation of the immense spaceborne transmit antenna aperture that provides beam coupling sufficient to provide 11 meaningful energy transfer to the ground. Each sandwich module would act as an element or subarray in the large phased array antenna. A simple functional representation of a sandwich module appears in Figure 4. Figure 4: Depiction of the functional layers of the sandwich module. The lower of these functional layers can be further decomposed into subfunctions that are anticipated to be needed for a module that would be deployed in practice. In the DC to RF conversion layer, the subfunctions include DC power conversion, incorporation of frequency and phase information for beam control, RF amplification and phase shifting, and output filtering. In the antenna layer, in addition to the elements to transmit the power beam, there would likely be separate elements to receive the pilot signal to be used to control beam pointing retrodirectively. This pointing method has been demonstrated safely and effectively on many occasions and is described in [29]. Figure 5 shows a depiction of the subfunctions in context. Note 12 that the input sunlight could be any of a range of concentrations, depending on the system implementation. Figure 5: Sandwich module layers showing subfunctions. Chief among the design challenges of a practical sandwich module are the integration of the various required elements and effective thermal management under adverse conditions. Although these aspects have received some attention from 13 researchers in the past, there had not been any characterization of a sandwich module prototype?s performance in a realistic space environment scenario in conjunction with a comprehensive analysis of the limitations levied by heat transfer, materials, and specific power until this work. The first exhaustive examination of the sandwich module concept was by Owen Maynard in 1980 [28]. His NASA report ?Solid State SPS Microwave Generation and Transmission Study,? outlines many of the obstacles and sensitivities associated with the sandwich design. Maynard proposed using solid state field-effect transistor (FET) amplifiers as an alternative to or in conjunction with the vacuum electronics microwave sources that had been suggested in much of the DOE/NASA study documentation. He identified the maintenance of low junction temperatures of the solid state amplifiers used in a sandwich approach as a key point in assuring that acceptable operating lifetimes would result. Solid state amplifier efficiency plays a major role in determining the amount of heat that must be dissipated, as does the efficiency of the adjacent solar cell layer. Lower efficiencies produce more waste heat and thus raise the junction temperature. Maynard pointed out that an advantage of the solid state amplifiers over vacuum devices is that they do not require high voltages. The many thousands of volts needed for magnetrons and klystrons are difficult to manage in the space environment and necessitate the inclusion of high voltage power converters, introducing another source of conversion inefficiency. Among the issues and possible resolutions Maynard summarizes, charged particle radiation effects and topological considerations stand 14 out as some that specific part selection and module fabrication could address in a tangible fashion. Since Maynard?s work, there has been renewed interest in the sandwich module for space solar power and for the modular symmetrical concentrator concept. Japanese researchers in particular have performed analyses and even constructed prototypes of sandwich modules. The first comprehensive prototype was developed in 2000 by Hiroshi Matsumoto of Kyoto University [30]. The effort, dubbed SPRITZ for ?Solar Power Radio Integrated Transmitter,? culminated in hardware that included a solar illuminator, sandwich module prototype, and rectenna array for receiving the transmitted power. The solar cells used in the SPRITZ prototype were ?about 15%? efficient, resulting in considerable waste heat [31]. The reported RF system efficiency at >25W radiated power output was also 15%, excluding the solar array contribution but including feeder network and phase shifter losses [32]. Because the module was only operated in ambient conditions in which the convective cooling effect of air could assist in the heat dissipation, its probable performance in space could not be accurately characterized. More recently, Nobuyuki Kaya?s group at Kobe University has also produced prototype sandwich modules as part of an ongoing and comprehensive microwave power transmission and SPS technology development campaign. Though the sandwich module development of Kaya?s group focuses primarily on the antenna design and retrodirective control attributes required in an ultimate implementation, attention is also given to amplifier selection and operating conditions. Thermal concerns and amplifier and antenna configuration at 2.4 GHz are outlined in [33]. 15 Because of the complexities associated with addressing every possible functional aspect of a sandwich module, it was decided early on to focus on the most fundamental conversion elements. To this end, functions like implementing retrodirective beam control, phase shifting, output filtering, and actual antenna radiating were deferred for future possible prototypes so that resources could be focused on the photovoltaic conversion, DC-to-RF conversion, and to a lesser extent, antenna design and development. Figure 6 shows the subset of module functions which were implemented in this effort with the substitution of simulated sunlight for actual sunlight and a power measurement in lieu of antenna radiation. 16 Figure 6: The subset of sandwich module functions implemented in this research effort. Strikethroughs are functions that were not included, and red text indicates a variation from the model in Figure 5. Goals For this effort, it was proposed to experimentally investigate, analyze, and address thermal and integration problems inherent in the development of a sandwich module 17 prototype for photovoltaic collection, DC-to-RF conversion, and wireless power transmission for space solar power so that a prototype could be constructed and tested in a simulated space environment. Specifically, it sought to: ? Design, fabricate, and test the highest mass-specific power, highest total combined efficiency sandwich module to date versus previous efforts. ? Perform the first test of a sandwich module for space solar power under space- like conditions of vacuum and temperature, and to characterize its performance. ? Contribute to an empirical foundation for informed debates on the technical and economic viability of a prominent class of proposed space solar power systems. These goals were achieved, and the results are described in the remaining chapters. 18 Chapter 2: Sandwich Module Prototype Development The module prototype development began with an assessment of the principal design drivers. Because every layer of the sandwich module contributed to and was affected by thermal concerns, this was addressed first. Thermal Analysis A first order study of the thermal problem for the sandwich module showed some of the limitations imposed by the radiative heat transfer relation: 4ATP ??? (1) where P is the heat power transmitted, ? is the emissivity of the material, ? is the Stefan-Boltzmann constant, A is the radiating area, and T is the absolute temperature in kelvin. A view factor of one to a non-radiating body is implied. By assuming that a flat sandwich module could only use its top and bottom for radiating heat, since it would be adjacent to other modules needing to also dissipate heat at its edges, bounds were established by specifying the desired operating temperature, which in turn were set to allow usage of commercially available electronic components. In practice, solar cells and antenna surfaces can be decent radiators, so having these respective top and bottom surfaces modeled as efficient radiating surfaces was not unreasonable. 19 The sun?s yearly averaged power flux in space in earth orbit is approximately 1370 W/m2, though the actual flux presented to a conversion module may vary as a result of a concentrator implementation and orbital position. For the first order study, the flux was assumed to be constant. Multiplying notional efficiencies of the component layers gave the rough total module efficiency shown in Figure 7. In this case, a square module made of four rows of seven cells with each cell measuring 4 cm x 7 cm was assumed for simplicity. Figure 7: Solar power intercepted at one sun (AM0) by a 28 cm by 28 cm module and combined module efficiency with notional layer efficiency estimates. The required radiator area was calculated for different sun concentrations and emissivities for modules at different desired operating temperatures, as shown in 20 Figure 8. Because this was an idealized model, the temperature effects on the efficiency of the layers, most notably the photovoltaic and DC-to-RF conversion layers, were not accounted for. The efficiencies of these two layers will tend to decrease with rising temperatures, in turn resulting in more heat that must be dissipated, further raising the module temperature. Evident is that with the assumptions made, the available radiator area of a two- sided module will limit the solar illumination level to about two suns for an operating temperature of 100?C, and 6 suns for an operating temperature of 200?C. Figure 8: Radiator area required to maintain temperature equilibrium for a 28 cm by 28 cm module at 23% efficiency. 21 Inserting combined module efficiencies of varying levels of optimism allowed plotting the resulting temperatures as a function of sun concentration. This plot, seen in Figure 9, shows that to operate below 150?C with reasonable efficiency assumptions limits the sun concentration to about three suns. An increase in radiator area would necessitate a departure from the prototypical flat sandwich module. Figure 9: Temperature of a 28 cm by 28 cm module with both sides as black body radiators for various module efficiencies. Keeping the target module temperature below 150?C helps limit the efficiency degradation resulting from the temperature of the photovoltaics and the DC-to-RF converters. In reality, individual efficiency degradations would be dependent on the local temperature, which varies across different parts of the module. A thermal analysis using Thermal Desktop? performed for a sandwich module design with 22 realistic efficiency assumptions operating under three suns [34] shows the peak temperature indeed surpasses 150?C, as seen in Figure 10. Figure 10: Thermal Desktop? simulation of a sandwich module under 3 suns of illumination [35]. SOLAR ARRAY FACE TRANSMIT ANTENNA FACE PHOTOVOLTAICS FACE 23 The limits imposed by the thermal analyses presented here offered two possible paths to the creation of a sandwich module suitable for SSP: (1) a flat module at moderate sun concentration (below three suns), or (2) a departure from the traditional flat sandwich module to allow for an increase in radiator area for operation at higher sun concentration. Though the focus of this work was primarily on the former, a novel design embodying the latter was created, for which a patent application was published, U.S. 20130099599A1. It was observed that an array of modules that could still fulfill the implicit requirements of having a flat projected collection area to maximize solar energy collection and being able to serve as an antenna element could be formed using ?step? shaped modules. Step Module Concept In the step module design, upper and lower radiator surfaces are added to provide for additional heat rejection. The length of these radiators is arbitrary, but as the distance from the heat source increases, the benefit of the additional radiator surface will tend to diminish. The location of the electronics was moved from being in close proximity to the hot solar panel to a cooler place on one of the radiator panels. To visualize this departure from the traditional flat sandwich module, dubbed the ?tile? module to distinguish it from the newer concept, or ?step? module, consider the depiction in Figure 11, which shows a close-up view of the photovoltaics and transmission antenna of the Modular Symmetrical Concentrator implementation from Figure 3. 24 Figure 11: Photovoltaics and transmit antenna comprised of tile modules. Contrast this with the photovoltaics and transmission antenna portion comprised instead of step sandwich modules as seen in Figure 12. Figure 12: Photovoltaics and transmission antenna comprised of step modules. 25 Note that suitable transmit antenna apertures can be created using the step modules as elements to form structures in the shapes of cones, inclined ridged discs, and other shapes with varying radiator views of deep space for heat rejection [36]. An elongated spike formation at the tip of a cone could serve to increase radiator area further for additional heat rejection capability in the center of the array where the power density and waste heat would likely be greatest. In any case, from above or below the optical projection will very closely resemble the original flat disc, allowing it to be used essentially interchangeably with the rest of a given solar power satellite architecture that employs the sandwich module concept. Since the modular aspect is common to both the tile and step, techniques that envision self-organizing structures for assembly are still usable. A thermal simulation performed in Thermal Desktop? showed that the step approach effectively lowered the maximum temperature of the module when compared with the tile module under the same test conditions. Furthermore, the DC and RF conversion electronics operate about 20?C cooler than in the tile module. The results of the thermal simulation can be seen in Figure 13. 26 Figure 13: Thermal Desktop? simulation of a step module under 3 suns of illumination. 27 Both tile and step module concepts were pursued in order to evaluate the relative merits and disadvantages of each, and in particular to determine whether the thermal benefits of the step module were outweighed by the increase in mass required for the additional radiator area [37]. These results are presented in Chapter 4: Results and Discussion. Critical Tradeoffs Each layer of the module presented a plenitude of options for performing the required function. Prior to the design and fabrication of the prototypes, tradeoff studies were performed to assess the merits, disadvantages, and suitability of the various options for each layer for utilization in the modules. Photovoltaics For the module prototype, commercially available space-qualified photovoltaics (PV) were used. Though there exist many promising high efficiency technologies in laboratory settings, they were not available or practical for inclusion in the prototype because of prohibitive costs or lack of availability. PV cells commonly used for space are readily available from two companies: Emcore Corporation and Spectrolab Incorporated, a division of Boeing. Both offer triple junction cells with conversion efficiencies quoted near 30%. Spectrolab?s Ultra Triple Junction (UTJ) GaInP2/GaAs/Ge solar cells with a bare-cell efficiency of 28.3% were selected. The layer stackup and cell I-V curve (current and voltage measurements as a function of varying the load presented to the device?s output terminals) from the datasheet are shown in Figure 14. 28 Figure 14: Layer stackup and I-V curve for the Spectrolab UTJ photovoltaic cells from the datasheet [38]. SpaceQuest, Ltd. produced the solar arrays and employed fabrication methods to foster high temperature tolerance during cell-to-substrate integration. The finished solar panels for both the tile and step modules utilized two 14-cell strings in parallel to provide the desired voltage and current. A tile module solar panel is shown in Figure 15 and a step module solar panel is shown in Figure 16. 29 Figure 15: Conversion module solar panel for a tile module. Figure 16: Conversion module solar panel for a step module. 30 Since each 26.62 cm2 cell at its maximum power point outputs 2.35 volts and 434 milliamps at one sun AM0 illumination and 28?C per the data sheet [38], the total projected panel output was 28.0 W near room temperature accounting for a 2% loss anticipated due to cell coverglass installation. Because the initial projected operating temperature in vacuum was 100?C, the expected output power was 22.5W, accounting for the temperature effects on voltage (-6.5mV/?C per cell) and current density (1.2?A/cm?/?C) [38]. Prior to integration with the prototype, the completed solar arrays were characterized by collecting I-V curves. A representative I-V and power data plot is shown in Figure 17. Figure 17: Representative array string I-V and power data plot collected with direct sun-simulated illumination. 31 The I-V curve collection also afforded an opportunity to observe the temperature effect on the cell and string voltage and power output. Considering Figure 18, it can be seen that even over a comparatively small span of about 20?C the open circuit voltage drops by 1.5 volts. The output power drop is on the order of half a watt. This particular set of I-V curves was taken on one of the strings of a tile module panel while the beam from the xenon lamp used as a sun simulator was partially blocked by a set of screens, resulting in an illumination level equivalent to just over one sun. The basis for the estimated illumination level is described in the ?Light Field Characterization? subsection in Chapter 3: Prototype Testing. Figure 18: I-V curves showing the effect of temperature on panel open circuit voltage with lamp output attenuated by screens to produce about one sun. 32 Since the cells and assembly techniques employed were very similar to those used for actual satellites and space missions, no special modifications or accommodations were made to adapt the solar arrays for vacuum and illumination testing. DC-to-RF conversion The conversion electronics needed to simultaneously achieve cost, weight, efficiency, and output power requirements. Considerations and trade options are considered in greater detail in [39] and are outlined in Appendix A in Table 11. Because high power, high efficiency solid state devices had not been demonstrated at the likely greater than 30dB gain required to add the power to the input frequency to effectively utilize the available solar array power, the construction of a suitable amplifier chain was central to this layer of the module. Monolithic microwave integrated circuit (MMIC) options could be considered to implement this level of gain in the future. Power to feed the multistage architecture was generated by power circuitry painstakingly designed to minimize conversion losses, in large part by directly driving the final stage amplifier. The resulting prototype board demonstrated a DC to DC conversion efficiency on the order of 97%, and is shown in the context of the integrated DC and RF electronics baseplate shown in Figure 19. This high efficiency was due in large part to driving directly the final stage RF amplifier. 33 Figure 19. DC power and RF electronics baseplate for the tile module. For the prototype, all RF components were ?off-the-shelf? items operating at 2.45 GHz. They were selected to best match the expected output of the solar array at its peak power point under the projected operating conditions, with about 30 W for the one sun case. The frequency source was a voltage-controlled oscillator (VCO) with a 7.5 dBm output (the Mini-Circuits ZX95-2755+), which was filtered and attenuated prior to entering the GaAs driver stage (the Hittite HMC755LP4E). The driver stage provided about 31 dB of gain, and the signal was then fed into the final stage (the Cree CGH27015) where approximately 12 dB of additional gain was applied, resulting in an RF output power on the order of 15 W. When the final stage was 34 characterized on its own, greater than 55% Power Added Efficiency (PAE) in the region of interest was observed as seen in Figure 20. Figure 20. Characterization of Power Added Efficiency performance of final stage amplifier [35]. The projected maximum efficiency performance of the single chain used for the tile module is shown in Table 1. The measured efficiency matched the projection closely. 35 Table 1: Projected RF signal and efficiency chain performance for tile module. Temperature and vacuum testing of the complete electronics prior to module integration showed only limited variation in the output total electronics efficiency, on the order of 45?2%. The range of temperature testing for the electronics chain alone in vacuum was from -20?C to +95?C, with the final RF amplifier stage as the control point. The reduction from the results achieved with the final stage amplifier alone can be attributed to inefficiencies from the power electronics, driver stage amplifier, and cabling losses. For the step module, the RF chain was instantiated in triplicate, the approximate integer number of suns under which simulations suggested the step module should be able to operate. This scaled with the multiplicative current increase associated with increasing sun concentration. The layout can be seen in Figure 21. The outputs of the three parallel RF chains were power-combined for routing to the antenna. 36 Figure 21: Step module prototype with antenna mockup, electronics shown at left. Because the three legs were combined at the outputs of the final stage amplifiers, it was critically important to phase-match each leg to avoid heat dissipation in the power combiner. More importantly, a phase mismatch would also lower the total RF output power. A line stretcher preceding each driver input was used to tune the phases. The three line stretchers are visible in the upper left corner of Figure 21. The bandwidth and harmonics of the output were monitored using a spectrum analyzer. These proved effectively invariant over different illumination and temperature conditions, with the center frequency shifting over less than a 5 MHz range. A representative spectrum analyzer screen capture is shown in Figure 22. Antenna Mockup Line Stretchers 37 Figure 22: A representative spectrum analyzer screen capture from monitoring of the RF bandwidth, harmonics, and center frequency. This capture is for the step module while powered by a solar array simulator. Since the electronics used were not space-qualified nor in some cases even built to withstand military specified temperature ranges, there was some risk associated with exposing them to vacuum and temperatures that exceeded their datasheet ?Absolute Maximum? ratings. Though initial testing specifically avoided transgressing these limits, ultimately the limits on the voltage-controlled oscillator and driver stage amplifier were exceeded somewhat, with no apparent ill effects to their operation. Discussions with the manufacturers suggested that the ?Absolute Maximum? limits cited in the datasheets were established to ensure reliability over long operating periods and for the driver amplifier to specifically support one million hour mean time between failures. All electronics components were assessed for suitability of operation in vacuum, and aluminum electrolytic capacitors found on the final stage 38 amplifier boards were removed and replaced with tantalum capacitors to eliminate the possibility of a component rupture during vacuum operation. No parts were assessed or screened for radiation tolerance (neither total radiation dose nor single event effects) and no reliability or other requirements were placed on them beyond their ability to function for module testing. In an ultimate module for use in a spaceborne demonstration or operational space solar power system, electronics would also be implemented to effect the phase shifting needed for a retrodirective control scheme. Similarly, as previously shown in Figure 5, such a flight module would also require very narrow bandpass filtering on the output of the final stage to suppress amplified harmonics and thermal noise. Antenna Elements The antenna component needed to have a potential path in order to serve as an element in a large phased array that would comprise the spaceborne power transmitting aperture, or ?spacetenna.? This aperture would be quite large and would scale with the frequency selected for operation. Some examples of transmit antenna sizes for different frequency and power density parameters can be seen in Appendix B: SPS System Design. A wide variety of antenna types were considered for the microwave transmission face of the module. Ultimately, a short backfire antenna was chosen by virtue of its high directivity, high efficiency, ability to act as an effective thermal radiator, and because it was nearly an ideal physical size for the selected operating frequency and solar array dimensions. A CAD depiction showing the surface current distribution 39 appears in Figure 23 and the simulated gain pattern for a single antenna element is shown in Figure 24. Figure 23: Simulated surface currents of short backfire antenna [40]. 40 Figure 24: Simulated gain pattern for a short backfire antenna at 2.45 GHz [40]. The element pattern was designed to suppress the grating lobes that would otherwise result from having spacing between elements that exceeds a wavelength. In an SSP system, this antenna element would be one of hundreds of thousands in a filled array. Both the tile and step modules employ the short backfire antenna design described above. 41 Module Architectures Most previous sandwich module concepts have taken a hexagonal shape, ostensibly to maximize the usage of the volume of a cylindrical launch vehicle fairing. A preliminary analysis of current launch vehicle fairing volumes and throw weights suggested that a module of any reasonably achievable density will exceed the launch vehicle throw weight before exceeding the fairing volume. Because of this, module shape was selected based on other factors, such as photovoltaic cell coverage, though it is reasonable to anticipate that fairings for launches of SSP elements would be custom designed to maximize launch economy. Since currently available PV cells are essentially rectangular, a square or rectangular panel surface can be more efficiently filled than a hexagonal one. The module thickness is driven by the need to accommodate the DC-to-RF conversion electronics and the antennas, as well as by its ability to manage the transport and radiation of waste heat. Thermal Control Methods There are many effective and novel means for transferring heat from one area to another: diamond or graphene heat spreaders, pyrolitic graphite structures, microchannels, and two-phase heat pipes, among others. Regardless of the transport method, the waste heat needs to be radiated from the module. An appropriately anodized antenna surface closely approaches the emissivity of a black body. The area available for radiation of heat can be increased with judicious module design. In the tile module prototype for this effort, thermal grease between RF electronics modules and the substrate was employed, multilayer insulation (MLI) blankets were used to protect the power electronics from excess solar array heat, and black Kapton? tape 42 was used to maximize emissivity. Two thermal zones were created: one for the solar array which could tolerate and operate under higher temperatures, and a second lower temperature zone for the electronics to preserve their operating efficiency and to extend their reliability. The integrated tile module with some of the thermal features visible is shown in Figure 25. Figure 25: Integrated tile module showing from top to bottom: solar array, conversion electronics with multilayer thermal blankets and black Kapton? tape, and antenna mockup. For the step module, a layer of graphite sheeting was applied to the module substrate in an effort to facilitate heat transport. Black Kapton? to increase emissivity and thermal grease to enhance thermal conductivity between the electronics, baseplate, and substrate were again employed. Solar Array Thermal Blankets Antenna Mockup Black Kapton? 43 Design Iteration The original tile module design proceeded through a series of design iterations, beginning with a large hexagonal shape and migrating to a rectangular shape because of the considerations described in the ?Module Architectures? subsection. The size of the solar array was initially larger, but was reduced to more closely match the output levels of the anticipated RF amplifier components, as well as to fit within the confines of the available test chamber. Minor mechanical updates were made almost constantly as the antenna design progressed and went through a long series of iterations of its own. Once the solar array was ordered, as the costliest and longest lead element, it effectively constrained subsequent design changes. Each layer was deliberately kept separable, to the extent that the thermal control approach would allow, in order to permit replacement or revision of damaged or underperforming components. Thus, the module design was in itself maintained as modularly as possible. The step module was kept similar to the tile module in most regards to allow for more direct and meaningful comparisons to be made, though it required a different substrate to achieve the step shape and additional RF chains to handle the higher output power levels. Module Fabrication Except for the solar array and commercial electronics components, the modules were fabricated and integrated on-site at the Naval Research Laboratory using in- house machinists, technicians, and engineers. This allowed for rapid turnarounds in the event the design changed, and also enabled close oversight of the modules? build 44 progress. Multiple copies of critical assemblies were fabricated to support quick replacement of components in the event of hardware failures or other problems. 45 Chapter 3: Prototype Testing Since effective testing of the prototypes was fundamental to meeting the research goals, a treatment of the testing philosophy, facility, and approach is presented here. Because of the three layers and their various constitutive elements, testing was done progressively so that anomalies might be discovered prior to integrated module testing where problems would present larger setbacks with respect to time, and the ability to alter the architecture would be constrained. Progressive Testing As the various layers of each module type were completed, they were tested separately to at least a functional level prior to full module integration. This allowed individual characterization of each section of the modules in the event there was an unexpected interaction between the layers. In certain cases, this also provided an opportunity to use and learn lessons from test configurations that would be used for or would feed into final testing. For instance, an early RF electronics chain was tested in vacuum with the fused silica chamber window, verifying the approach to sealing the chamber, and allowing a chance to perform insertion loss measurements and attenuation compensation of the RF power measurement. Upon their completion, the integrated modules were characterized under ambient pressure and vacuum conditions. An overview of the process for the tile module leading up to this point is shown in Figure 26. ?Mass Properties? indicates a point at 46 which the subassembly or module was weighed and checked for dimensions so that figures of merit could be calculated. Figure 26: Tile module integration and test flow overview. Antenna mockups were used in place of the actual antennas during testing. Though it was initially considered to test with an output antenna radiating inside the vacuum chamber to a rectenna or power sensor to capture the transmitted microwave energy from the modules, routing the RF output via a coaxial cable through the chamber bulkhead was deemed sufficient to meet the research objectives. This avoided the need to install specialized, high-power, vacuum-rated RF absorbing material and eliminated the uncertainty that would have been introduced through the possibility of not being able to capture all of the emitted RF energy. 47 The prototype testing facility can be considered in three functional segments: (1) sun simulation, (2) space environment simulation, and (3) additional supporting equipment. These segments are addressed in turn, and the overall test configuration is discussed as well. Sun Simulation A critical subsystem of the test ensemble was that for the simulation of solar illumination. Comprised of two major sub-elements, one or two xenon lamps, and a series of attenuating screens, this subsystem produced the lighting conditions that a sandwich module might be expected to operate under while in space. Xenon Light Source A variety of sun simulation sources are available for various purposes. Ranging from simple arrays of common halogen lamps to xenon lamps with sophisticated filtering systems, a solution was needed that could provide intensity in excess of one sun (AM0) over a sizable area. A 4,000W xenon short arc lamp system, Model 6806 made by LP Associates, Inc. was ultimately selected for solar panel illumination. This lamp system, seen in Figure 27, has a parabolic reflector and is primarily used as a spotlight for sky and theatrical lighting. 48 Figure 27: 4,000W xenon light source with power supply. The lamp sports good beam collimation but uneven beam uniformity. Its focus could be adjusted via a motor drive that slightly changed the angle of the parabolic 49 reflector. It was set up initially to provide approximately a 56cm diameter beam, which was found experimentally to offer a reasonable balance of beam size, intensity, and uniformity. The resulting uniformity was adequate for testing but resulted in excess heat generation on the solar panel versus the expected illumination condition in space. Excess heat was also generated by overrepresentation of power content represented by the spike evident in the 850nm to 1050nm spectral range as seen in Figure 28; this spike in the spectrum is not present for sunlight in space. Figure 28: Spectrum of 4,000W L.P. Associates xenon lamp. 50 Light Attenuating Screens Combinations of five different light attenuating screens were used to control solar concentration. The screens were hung from a metal rack in the path of the xenon lamp?s beam. Table 2 shows the amount of light blocked and passed for each of the five screens individually. Table 2: Screen designations with blockage and throughput percentages from manufacturer specifications. Screen Allowed A 75.0% B 65.9% C 60.2% D 50.7% E 25.4% At a given focus setting, the xenon lamp produced a peak intensity that could then be approximately scaled by the various percentages shown. The screens themselves can be seen in Figure 29. 51 Figure 29: The five light attenuating screens labeled with percent open area. As the wire screens reduced the light intensity by an essentially constant amount over the entire beam, they did not materially address the matter of uneven beam power uniformity. Some investigation and thought was given to developing customized patterns, screens, lenses, or some manner of adaptive optics that could compensate for the beam unevenness, but each of these options appeared beyond the scope and wherewithal of the project. Instead, the light field was measured to quantify the power nonuniformity and determine its impact on module characterization. For the focus setting used for initial tile module testing, Table 3 shows the equivalent peak, minimum, and average solar concentration for each screen in terms of number of suns, where one sun is for AM0, 135 mW/cm2. These values changed as a function of the focus adjustments that occurred as testing progressed to slightly increase the illumination level on the module prototype. A 5% reduction in light power due to passage through the fused silica window is not reflected. This reduction factor was derived by comparing power measurements with and without the window. 52 Table 3: Screen designations and sun concentrations for initial tile module testing focus settings. Screen None A B C D E Peak 3.17 2.71 2.18 2.04 1.69 0.92 Average 2.23 1.90 1.56 1.44 1.21 0.66 Minimum 1.39 1.23 1.00 0.87 0.76 0.41 Light Field Characterization Determining the evenness of the field and the absolute intensity of the light field had particular challenges. Ultimately, the evenness of the field and the absolute intensity of the light source were measured and characterized using a Newport thermopile sensor optical power meter with a customized heat shield and an Ophir- Spiricon beam profiling system with a large Lambertian surface. The thermopile sensor?s output was compared against an Eppley global cavity radiometer by exposing both to similar conditions under an Oriel solar simulator. The reported power densities matched within 13.5% [41]. Digital image captures from the beam profiling system were taken of the unobstructed field in a plane in proximity to that in which the prototype module would be located, and again several times with the thermopile sensor in various points in the light field. The absolute intensity measurements from the thermopile sensor were logged and later referenced to the pixel locations in the unobstructed field image to create a mapping to show the relative sun intensities across the entire field. 53 The light field evenness and intensity characterization was done to ensure that the solar cell strings were exposed to a known minimum illumination condition, as the least-illuminated solar cell in a string current-limits the output of the entire string. Because the sun simulator had large intensity variations across the area of illumination, the efficiency calculations used the known minimum average illumination level. In order to ensure a minimum level of illumination of one sun (for AM0, 135 mW/cm2) over the entire solar array, some portions were exposed to greater levels. As a result, the thermal conditions were more oppressive in the testing than they would be in space, where the lighting conditions would be uniform. During a portion of the testing, the light field variations were exacerbated when the lamp focus was tightened to increase total intensity. A raw screen capture of the entire central light field of one lamp is shown in Figure 30. Figure 30: BeamGage software screen capture. 54 An example map of the beam uniformity was processed to show the equivalent numbers of suns averaged over each solar cell?s area (Figure 31). Though it appears the intensity of the field could be better centered, this was the portion of the beam that offered the best uniformity. The anomalies in the upper left and lower right around the perimeter were due to a slight mismatch between the framing of the panel area on the Lambertian surface and the actual capture area in the BeamGage software. 55 Figure 31: A light field portion matching the solar array size that has been processed to show the number of suns of intensity incident on each cell area, adapted from [42]. For step module testing, a second xenon lamp was added to increase the overall intensity. This helped even the uniformity of the field somewhat, but the inability to closely pack the lamp housings resulted in an incidence angle on the solar array face 56 slightly deviating from normal. This resulted in negligible cosine and refraction losses. A mapping showing the sun intensities per cell when two lamps were used is shown in Figure 32. Figure 32: Two lamp beam mapping showing equivalent number of suns of intensity over solar cell areas, adapted from [42]. 57 As with the single lamp light field map, the anomalies in the upper left and lower right around the perimeter were due to a slight mismatch between the framing of the panel area on the Lambertian surface and the actual imaging software capture area. While considerable effort was put into minimizing the unintended variation of the factors affecting the light field, not all factors could be absolutely controlled. Variables that were identified and addressed to the extent possible included: ? Lamp pointing ? Lamp distance ? Lamp warm-up duration ? Bulb age ? Focus setting ? Screen used ? Screen positioning ? Vacuum window presence ? Reflections from adjacent apparatus (pipe, chamber) ? Light incidence angle on solar array Further sources of error that could be introduced during the capturing and processing of the light field to derive the illumination intensity included: ? Imperfections of the Lambertian surface ? Pixel variability in the beam imaging camera 58 ? Inconsistency of optical power measurements due to sensor misalignment ? Misalignment of the measured light field and the light field exposed to the module ? Difference in the distance between the Lambertian surface and the solar array ? Inability to account for imperfections in the fused silica window Because of these factors, there are errors in the light field mappings that were difficult to bound with high accuracy, but the repeatability of the results from the beam mappings and module output power during testing instilled sufficient confidence in the findings. Space Environment Simulation As a major goal of the research, testing in a space-like environment was of paramount importance. While vacuum chambers are readily available in a range of research and manufacturing environments, it is not common to find chambers with windows larger than 30 cm in diameter, and not all chambers are equipped for the extreme temperature range found in space. As part of its spacecraft development facility, the Naval Research Laboratory?s Naval Center for Space Technology offers a range of small and very large vacuum chambers that can achieve pressures lower than 10-6 torr and temperature ranges at least as large as -150?C to +100?C. However, at the start of this effort, none routinely incorporated a vacuum window of suitable size. Because of this, a very large thermal vacuum and sun concentration test facility at NASA?s Glenn Research Center known as ?Tank 6? was considered. This facility offered a large chamber volume and sun concentrations up to eleven suns over small 59 areas. However, limited project resources and the anticipated need to frequently enter into and exit from the vacuum state, which typically scales in time needed and general arduousness as a function of the vacuum chamber?s size, required the creation of an alternative. This alternative needed to feature a sizable vacuum window that could transmit the solar spectrum effectively. Fused Silica Window Fused silica effectively passes the solar spectrum, as seen by comparing its transmission curve from Figure 33 and the solar radiation spectrum seen in Figure 34. Figure 33: UV Grade Fused Silica Transmission Curve [43] 60 Figure 34: Solar Radiation Spectrum [44] Though magnesium fluoride (MgF2), or to a lesser extent, calcium fluoride (CaF2) windows would have offered marginally better transmissivity, procuring a new window made of one of these materials of the required diameter and thickness was beyond the project?s budget. Fortunately, providence had it that a 68.6cm diameter / 4.5cm thick fused silica window that was designed to fit one of NRL?s much larger chambers, and which had lain forgotten and unused for many years, was usable with a smaller chamber in lieu of the vessel?s hinged door. This window, though it contained a few small blemishes and imperfections, proved effective and sufficient for the project?s purposes. The window is shown in Figure 35. The markings showing the locations of the flaws were removed prior to testing, and the window was cleaned with isopropyl alcohol. 61 Figure 35: The fused silica window used with the vacuum chamber. Ideally, an appropriate anti-reflective coating would have been identified and applied to the exterior face of the window to further enhance its transmissivity, but this was not done due to resource limitations. I-V curve collection testing with the window showed that it reduced the energy generated by the solar array strings on the panel from 4.1 to 5.8 percent versus without the window. This difference is principally due to reflection, as the window itself closely tracked room temperature, even when two lamps were used. During testing, the front door to the chamber was restrained open or removed so as to accommodate the fused silica window. The window was then placed on a custom- designed and fabricated pair of supports affixed to the front of the vacuum chamber mounting structure and then clamped into place, as shown in Figure 35. 62 Figure 36: DC and RF electronics in thermal vacuum chamber with fused silica window sealing chamber opening. As the air was pumped out of the chamber, the window sealed against the chamber O-ring gasket because of the pressure difference with the ambient surroundings. Thermal Vacuum Chamber The thermal vacuum chamber used was a liquid nitrogen-cooled vessel from Meyer Tools & Manufacturing (Part # 5928-01) capable of an internal shroud temperature range from -180?C to +100?C and of achieving a vacuum pressure of 1x10-8 torr. The vacuum is created by first using a mechanical roughing pump to approach the 10-4 63 torr regime, and then a cryogenic pump is employed to get to ?high vacuum?, the 10-6 to 10-7 torr range or lower. The standard configuration prior to fused silica window installation is shown in Figure 37. Figure 37: Thermal vacuum chamber prior to fused silica window installation. 64 The configuration that was used for testing in thermal vacuum conditions is shown in Figure 38. Here the fused silica window sealing the front of the chamber is nearly invisible, allowing essentially full illumination by the 4,000W xenon light source with only limited spectral filtering. The tile module prototype is suspended from the top of the chamber by insulated wire connect to a fiberglass support piece to minimize conductive heat transport during testing. Figure 38: Module prototype in thermal vacuum chamber with fused silica window sealing chamber opening and vacuum chamber internals visible. 65 During testing, once high vacuum had been reached, a solenoid-controlled valve was opened to allow liquid nitrogen to flow through the plumbing surrounding the chamber?s internal shroud. As the shroud temperature dropped, the illumination level was increased to prevent the module under test from dropping below about 5?C. Once ramped to the target temperature of -150?C to simulate the view of deep space, the shroud was maintained within ?5?C of this temperature with periodic flows of liquid nitrogen. The liquid nitrogen shroud cooling apparatus is shown in Figure 39. Note the container for receiving the nitrogen once it has circulated through the shroud to prevent excess nitrogen buildup in the test area. 66 Figure 39: Liquid nitrogen shroud cooling apparatus. 67 Though -150?C is about 123?K, which differs greatly from the background temperature of deep space of about 3?K, this shroud temperature is effective in measuring the module?s likely heat radiating ability because the difference of the temperatures of the shroud and module to the fourth power differs negligibly whether 123?K or 3?K is used for the shroud temperature. For example, using: )( 44 envTTAFP ?? ?? (2) Where F is introduced to represent a constant view factor between the radiating body and the environment (ranging from 0 to 1) and Tenv is the temperature of the environment. In the case of T = 300?K (approximately room temperature), for Tenv = 123?K or 3?K, the (T4 - Tenv4) factor changes by less than 3%. This difference decreases further with increasing T. Due to uncertainties about the vacuum chamber shroud?s capabilities, initial Thermal Desktop? simulations assumed a shroud temperature of -100?C (173?K); even this only changed the factor by about 11%. As the window end of the chamber was illuminated by the xenon lamp, it effectively simulated the thermal environment due to the sun. The opposite end of the chamber was not covered by a liquid-nitrogen-cooled shroud and thus maintained a temperature near the ambient room temperature. This heat input, though small compared to the heat input from the sun simulation, would not be present to the same degree in an actual space environment since the earth?s diameter only appears to occupy about 18? of the field of view from geosynchronous orbit. Even if the module were at the extreme window end of the chamber, the opposite end would occupy a 68 28? portion of the field of view. This only increase as the module is moved farther from the window end of the chamber. The step module is shown in a pictorial depiction of the configuration in Figure 40. Note that for the actual testing the radiating surfaces faced the sides of the chamber, rather than the top and bottom, in part because the bottom of the chamber had a fiberglass platform that was not cooled by liquid nitrogen, and thus was not a good target for simulating the space condition. Figure 40: Depiction of step module in vacuum chamber configuration[45] The step module installed in the chamber for testing is shown in Figure 41. 69 Figure 41: Step module installed in vacuum chamber for testing prior to installation of the fused silica window. Summary Comparison of Space, Simulated Space, and Ambient Environments Since it is not possible to recreate with exact fidelity the space environment on earth, an examination of the differences of the simulated space environment and the actual space environment is warranted. The comparison can provide insights into the effects of the environmental differences on module performance and allow for reasonable extrapolations. On the whole, the conclusion to be drawn is that the simulated space environment presents a more challenging heat rejection environment 70 to sandwich modules as compared to the actual space environment. Table 4 shows in summary form a comparison of some of the parameters of interest. Table 4: Comparison of Space and Simulated Space Environments for Sandwich Module Operation Parameter? Individual? Sandwich? Module?in?an? Array?in? Geosynchronous? Orbit? Individual? Sandwich? Module?in? Simulated? Space? Environment? Effect?of?Difference? Between?Space?and? Simulated?Space? Minimum?Temperature? of?Surroundings? ?270?C? ?150?C? Heat?radiation?is? less?effective?to? warmer? surroundings? View?of?Minimum*? Temperature?Area? Approximately? 3??steradians? Approximately? 2??steradians? Reduced?view?to? colder?surroundings? reduces?heat? radiation?ability? Pressure? 10?11?torr? 80% PAE,? in IEEE Journal of Solid-state Circuits, 42(10), 2007. [65] K. Yamanaka, Y. Tuyama, H. Ohtsuka, S. Chaki, M. Nakayama, Y. Hirano, ?Internally-matched GaN HEMT High Efficiency Power Amplifier for Space Solar Power Stations,? in Proceedings of Asia-Pacific Microwave Conference 2010, Paper No. WE3A-1. [66] R. S. Pengelly, S. M. Wood, J. W. Milligan, S. T. Sheppard, W. L. Pribble, ?A Review of GaN on SiC High Electron-Mobility Power Transistors and MMICs,? IEEE Transactions on Microwave Theory and Techniques, 60(6), pp. 1764, Jun 2012. [67] W. C. Brown, J. Triner, "Experimental Thin-film, Etched-circuit Rectenna," 1982 IEEE MTT-S International Microwave Symposium Digest, 1982, pp. 185. [68] J. McSpadden, L. Fan, K. Chang, "Design and Experiments of a High- Conversion-Efficiency 5.8-GHz Rectenna," IEEE Trans on Microwave Theory and Techniques, 46(12), pp. 2053-2060, Dec 1998. [69] Report of the URSI Inter-Commission Working Group on SPS, URSI Inter- Commission Working Group on SPS, pp. 13, 2007. [70] R. Cleveland, D. Sylvar, J. Ulcek, ?Evaluating Compliance with FCC Guidelines for Human Exposure to Radiofrequency Electromagnetic Fields?, Federal Communications Commission, OET BULLETIN 65, Edition 97-01, 1997. [71] IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3kHz to 300 GHz, IEEE C95.1-2005, 2005.