<%BANNER%>

Self-Powered Wireless Sensors: Remotely Powered Wireless Sensors Using Wireless Power Transfer


PAGE 1

SELF POWERED WIRELESS SENSORS: REMOTELY POWERED WIRELESS SENSOR S USING WIRELESS POWER TRANSFER By AHMAD HASSAN EL KOUCHE A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006 1

PAGE 2

Copyright 2006 by Ahmad Hassan El Kouche 2

PAGE 3

To my parents, brother, and two sisters. 3

PAGE 4

ACKNOWLEDGMENTS Foremost, I would like to express my grat itude to my research advisor Professor Abdelsalam (Sumi) Helal for all his support, in the many forms: his conf idence in my research work which encouraged me to work harder, hi s ambition and optimism that taught me to see obstacles in research as a start of a new solution and not a barrier, and his ability to approach me as a friend made my research tr uly enjoyable. I have never rea lly felt so comfortable working with a professor as much as I did with Dr. Hela l. His ability to switch hats from a research professor to a boss and then to a friend is abso lutely amazing. I have known Dr. Helal for almost four years now, in which I witnessed his first bo rn child Shadia. Without my advisors help and support, this research material would not have been possible. Al so, I would like thank Dr. Janise McNair and Dr. Shigang Chen for ag reeing to serve on my committee. I would like to thank my parents and thei r late night calls making sure I was eating healthy, when I was not, making sure I was gettin g enough fruit juice, when all I was actually drinking was black coffee, but most importantly I would like to thank them for providing me with strength at times when I was feeling tired. Also, I would like to thank my brother and two sisters for their endless love and patience. Also, I would like to thank all my friends in the Pervasive Computing Laboratory (Dr. Lim, Dr. Hisham, Dr. Bassam, Youssef, Jeff, Raja, Hen-I, and Steven). 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT...................................................................................................................................10 CHAPTER 1 INTRODUCTION................................................................................................................. .11 Problem Statement.............................................................................................................. ....11 Power Harvesting Sources......................................................................................................14 A History of Wireless Power Transfer...................................................................................17 2 DESIGN PLANNING AND CONSIDERATIONS...............................................................21 Friis Free Space Loss.......................................................................................................... ....21 Antenna Types........................................................................................................................26 Printed Patch Antenna.....................................................................................................29 Antenna Simulation Results............................................................................................34 Power Delivery.......................................................................................................................43 Attenuation.............................................................................................................................45 Feasibility...............................................................................................................................46 3 WIRELESS POWER TRANSFER SYSTEM DESIGN........................................................48 System Level Design..............................................................................................................48 System Design Constraints..............................................................................................49 Range Constraints............................................................................................................49 Cost Constraints...............................................................................................................50 Radio Frequency Generator....................................................................................................50 Radio Frequency Signal Source......................................................................................53 High Power Signal Amplifier..........................................................................................53 Antenna of Choice...........................................................................................................54 RF to DC Wireless Sensor Node............................................................................................55 Antenna Matching...........................................................................................................56 RF to DC Matching Network..........................................................................................56 Low Power BaseBand Processor............................................................................................58 Low Power Sensors................................................................................................................59 Low Power Wireless Transmitter...........................................................................................59 5

PAGE 6

APPENDIX A ELECTRONIC SCHEMATIC OF RF BASEBAD CONTROLLER....................................60 B REMOTE POWERED WIRELESS SENSOR C CODE.......................................................61 C RF BASEBAD C ONTROLLER C CODE.............................................................................64 D RECEIVER INTERFACE TO ATLAS C CODE..................................................................71 LIST OF REFERENCES...............................................................................................................74 BIOGRAPHICAL SKETCH.........................................................................................................75 6

PAGE 7

LIST OF TABLES Table page 2-1 Available ISM bands for wireless power transfer and applications..................................26 7

PAGE 8

LIST OF FIGURES Figure page 1-1 Moores law showing growth of transistor count for the Intel processor (dots) every 24 months. Figure taken from Wikipe dia.org...................................................................13 1-2 Power harvesting sources...................................................................................................17 1-3 Nicolas Tesla and the AC motor........................................................................................18 1-4 Tesla, master of lightning................................................................................................. .20 2-1 Basic system setup for wireless powe r transfer showing the antennas on both the transmitter and receiver, separa ted by radius distance R...................................................21 2-2 Received power (dBm) at the receiver vs. distance (meter) away from receiver, such that Pt =0dBm, Pr=0dBm, Gt=0dB, and Gr=0dB...............................................................23 2-3 Medium power amplifiers..................................................................................................24 2-4 Electromagnetic frequency spectrum and associated wavelengths. Borrowed from The RF and Microwave Handbook................................................................................26 2-5 Monopole whip antenna on top of a ground plane............................................................27 2-6 Dipole antenna............................................................................................................. ......28 2-7 Helical antennas........................................................................................................... ......29 2-8 Patch antenna.....................................................................................................................30 2-9 Effective length of patch after fringing effect Vs. ra tio of width to length k....................33 2-10 Effective length of patch after fringing effect Vs. ra tio of width to length k....................34 2-11 Patch antenna simulation using PCAAD........................................................................35 2-12 Square patch antenna using Ansoft De signer, with given dimensions, and output matched to standard 50 using stub stripline of length 12mm by 29mm.........................36 2-13 Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz......37 2-14 Antenna gain of patch antenna (dB) in terms of the swept frequency F(GHz).................38 2-15 Fabricated 2.45GHz antenna using homebrewed etching technique.................................39 2-16 Square patch antenna using Ansoft Designer matched to 50 at 2.45GHz......................39 8

PAGE 9

2-17 Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz......40 2-18 Antenna gain of patch antenna...........................................................................................41 2-19 Patch antenna for 2.45GHz ISM band, with 3.5dBi gain. Fabricated at PCBexpress.......42 2-20 Patch antenna array, matched at 2.45GHz, with gain equal to 6dBi.................................42 2-21 Improved wireless power received with additional 26.5dB gain.......................................43 2-22 High gain commercially available ante nnas suitable for remote power transfer...............44 2-23 Commercial chip antenna..................................................................................................4 4 3-1 Top level system design.................................................................................................... .48 3-2 Back-side view of RF generator........................................................................................51 3-3 Front-side view of RF generator........................................................................................52 3-4 Side view of RF generator, displaying tuning knobs.........................................................53 3-5 Signal generator local osci llator from MiniCircuits..........................................................53 3-6 RF amplifier ZRL-3500 from MiniCircuits.......................................................................54 3-7 Commercial patch antenna.................................................................................................55 3-8 RF to DC wireless sensor node system design..................................................................55 3-9 RF to DC matching network..............................................................................................57 3-10 RF to DC output voltage about 1.2 volts...........................................................................57 3-11 RF to DC output current about 1.2 mA..............................................................................58 9

PAGE 10

Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Master of Science SELF POWERED WIRELESS SENSORS: REMOTELY POWERED WIRELESS SENSOR S USING WIRELESS POWER TRANSFER By Ahmad Hassan El Kouche December 2006 Chair: Abdelsalam (Sumi) Helal Major: Computer Engineering Wireless sensor networks have recently starte d showing more interest towards issues of power consumption due to the high dependency on battery power. When the battery energy capacity is diminished to the point of inadequa te voltage level, the wi reless node would shut down, which might cause loss of information, re source limitation, intelligence, and perhaps cause network failure in some cases. Therefore, it is necessary to have a backup secondary power source that would deliver energy to the wireless node in case of a pow er scarcity. In the case of wireless nodes, backup power would have to be delivered, obviously, without using any wires, which means that, power would have to be delivered in the form of thermoelectric, solar power, mechanical, vibration of piezoelect ric material, or radio energy pr opagation. The latter is referred to as wireless power transfer (WPT), which is the main concentration and focus of alternative power source of this thesis, is discussed in deta il. The concept of WPT is as simple as generating a high power radio signal and beam ing it towards the wireless se nsor. The wireless sensor has the ability to detect this radio signal and turns it into usable DC voltage for storage. 10

PAGE 11

CHAPTER 1 INTRODUCTION Problem Statement As technology advances, we notice that elect ronic devices have the natural ability to eventually shrink in physical size in time. This can be obviously seen from the evolution of computers from mechanical computers, to transi stor IC computers, and now to ultra compact embedded sensor platforms, and wearable comput ers. The size and performance of integrated circuits ICs was predicted by Intels co-founder Gordon Moor e in 1965 that the number of transistors in a microchip would double every 18 months, which was then adopted as Moores law. Generally speaking, if the number of transi stors doubles every two years, then the power consumption would also double according to th e following equation for calculating rough estimate power consumption: Power Consumption = C V2 f (1-1) Where C is the transistor equivalent capacitan ce, V is the transistor voltage supply, and f is the transistor switching frequency. Therefore, th e amount of power consumed by a sensor node is proportional to the amount of power a single active transistor consumes s caled by the number of active transistors on the sensor node This tells us that the simple r the architecture deign is the less power is consumed. This of course has the draw back with the amount of intelligence and limitation a sensor node would have at that point, so there is a tr adeoff with sensor smartness and power consumption. This is an obvious argument si nce nothing is free, as system complexity grows, more transistors are used to get the job done, and thus more power consumption is used. The argument here is that the simpler you keep the design the more likely you would consume less power at the cost of se nsor intelligence lim itation. Another argument is how fast the transistor switches on and off, whic h is the effect of the variable f in the above equation, also 11

PAGE 12

known as the switching frequency or system clock. The switching frequency of the transistor is a tricky variable since one would basically think of reducing the frequency to consume less power, however, there is more to it than just operating at slower speeds. On the hardware level of things, using a high crystal clock requires more current to drive than a lower frequency crystal. For example, a microprocessor from Microchip PI C16F688 running on a 4 MHz crystal and voltage supply 2 volts would consume about 220uA vs. a 32 KHz crystal that would only consume 9uA running under the same conditions. However, the problem is that a 32 KHz crystal would take 128 times the time to complete the same job. With wireless sensor nodes the highest power consumption block is the radio frequency (RF) transmitter, which consumes constant power as long as there are information bits to be sent. Ther efore, if information bits are transmitted faster, then the RF stage would be on for a shorter peri od of time, and thus, system power consumption would be considerably lower. This method was successfully implemented and patented by Nordic VLSI company as the ShockBurstTM effect, such that informati on data is clocked in at a slower rate (32 KHz), but transmitted at 1 or 2 Mbps. Nordic ShockBurstTM radio architecture became popular in low power wireless devices such as keyboards, mice, computer presentation controllers, active RFIDs, heart m onitoring watches, and others. This clearly shows that a slower switching frequency is not always a lower power consumption approach, but that it really depends on the rest of system design and how it is affected by the switching frequency. Moores law states that the number of tr ansistors doubles every 18 months (SEMATECH research shows a 24 months roadmap), as seen in figure 1-1. This is actually because the transistor size (gate-length barrier) is reduced to half, and thus you are able to fit twice the number of transistors on the same silicon die area, which in effect reduces cost. 12

PAGE 13

Figure 1-1. Moores law showing gr owth of transistor count for th e Intel processor (dots) every 24 months. Figure taken from Wikipedia.org. This shows a simple observation that power c onsumption scales linearly with the amount of transistor doubling and the speed in which the transistor switche s states. Also, higher switching frequency requires higher bias voltages, which in return affects power on a quadratic scale. In order for a wireless sensor node to keep up with Moors law power consumption requirements, the battery industry would also have to double their battery capacity every two years, or put twice the amount of batteries on e ach sensor node. The la ter approach is highly unsuitable for wireless sensor network due to co st, size, and weight c onstraints. Also, some nodes have to be deployed in isolated environmen ts, which require long op erating life and years of maintenance-free operation. Thus, the need to reduce power consumption and the need for alternative power sources becomes a real concern for wireless sensor network.1,2,3,4 13

PAGE 14

To solve the ever increasing demand on pow er, wireless sensor nodes would require deployment of low power hardware design, soft ware selectable low power operation modes, short active duty cycles, and harv esting energy from different extern al sources. These constraints in wireless sensor network gave lead to the Zi gBee technology with its comparable low power consumption in full active mode of about 50 mA (Transmitter = 25 mA average, Receiver = 20 mA, and DSP = 5 mA) from Cirrone t when compared to a WiFi node that could consume up to a total of 700mA of active av erage current from DPAC and BlueTooth with an average of 150mA from BlueRadios. The advantage of ZigBee over Bluetooth includes lower power consumption, higher radius of operating range and larger number of supported nodes. Power Harvesting Sources Another solution for the high demand on power consumption is to harvest energy from the surrounding environment.5 Clean energy sources can be f ound around wireless sensor nodes and harvested into usable energy to recharge batteries or to charge large local capacitors, also known as super or ultra capacitors. Ultra capacitors have midrange capacity storage between rechargeable batteries and regular capacitors. However, unlike rechargeable batteries, ultra capacitors do not require a special way to be ch arged, as they would accept any kind of voltage level or waveform, that being AC or DC. This sets fewer constraints on the method of recharging super capacitors from different power source. Available sources for power harv esting include the use of solar panels seen in figure 1-2 A. Solar panels are found everywhere and are easy to harvest energy from, such that as the light intensity increases, the output produced voltage is a DC voltage easily charged on a super capacitor. Solar panels exhibit conversion effi ciencies between 10% and 20% in direct sun light[5]. For example, HARPs NE-Q5E2E photovoltaic module exhibits a 16.4% conversion efficiency under direct sun light Consequently, solar panel do not work very well indoors, in 14

PAGE 15

which they exhibit only 1% efficiency, and ar e useless for harvesting energy for a sensor network constantly operating in a dark environment. Piezoelectric material has gained some recen t popularity in power harvesting for wireless sensor network.6 As a matter of fact piezoelectric material goes back to 1880 when two brothers Pierre and Jacques Curie demonstrated that some crystals (tourmaline, quartz, topaz) generate electrical polarization from mechanical stress. This effect was then named "piezoelectricity" after the Greek piezein, which means to squeeze or press.7 Piezoelectric material really gained large popularity when it was used to maintain accurate timing in clocks, wrist watches, and crystal (XTAL) oscillators used in electronic systems today. However, the problem with using piezoelectric material for power harvesting is that you need a constant source of vibration or mechanical taps, stress, push, or pull to k eep generating power, which would be a perfect application to deploy a sensor network around vibr ating motors or engines in an industrial environment. Seen in figure 1-2 B is a flex ible piezoelectric sheet made by Measurement Specialties company that produc es very high open-circuit vo ltage close to a hundred volts when physically struck hard. However, when a low impedance load is used the output voltage becomes extremely low depending on the actually load value. Mechanical power can also be turned into el ectrical energy as seen in figure 1-2 C using windmills, and today using MEMS figure 1-2 D. A r ecent popular device in the market is the self powered flashlight uses a combination of mech anical shaking and elec tromagnetic induction of current. By shaking the flashlight, a magnet s lides between the north and south pole of the winded coil inducing current in the copper wire to be stored on a s uper capacitor, which is then used to power an ultra-bright LED when the on button switch is closed. The same exact device can be use to power a wireless sensor node, how ever running around to shake every single node 15

PAGE 16

becomes an impractical method, unless you are planning to use this system to work out, and I would guarantee you by the time you run to the s econd node, the message would already be lost. Another very popular device that uses mechanical energy to recharge a battery is the famous $100 laptop from MIT, which uses a hand crank to recharge the systems battery, low power processor, and low power software selectable us er modes. Future updates mention that the hand crank would be replaced with a power supply th at uses a foot crank instead, which reduces damage and stress caused to the laptops chassis. Thermal energy can also be transformed into electrical energy by using thermoelectric generators as seen in figure 1-2 E. Thermoelectric generators are composed of a thermocouple comprising a p-type and n-type se miconductor connected electrically in series and thermally in parallel. The thermoelectric generator produces an electrical curre nt proportional to the temperature gradient between the hot and cold junctions. An electric load is connected electrically in series with the thermoelectric gene rator creating an electric circuit, such that the temperature grading between the two electrode s can be harvested in to a super capacitor.8 Wireless power transfer is gaining signifi cant popularity and is heavily used in technologies like RFIDs seen in figure 1-2 F, wh ere batteries become e ssentially nonexistent. Such technology eliminates the bottle neck with power availability from battery capacity and the ever increasing demand on available energy cap acity. This technology essentially creates a transparent radio channel that either constantly f eeds power to the sensor node or charges a local capacitor over a constant duty cycle. Wireless power transfer requires two separate systems. The first system is the RF generator that genera tes the high power wireless energy and radiates it towards the wireless sensor node vicinity, and the second system, which is the wireless sensor 16

PAGE 17

node, receives the wireless power re ctifies it and turns it into usable DC power to be stored for future use. A B C D E F Figure 1-2. Power harvesting sources. A) Solar Panel. B) Piezoelectric sheets. C) Windmill D) MEMS E) Thermoelectric generators. F) RFID A History of Wireless Power Transfer Wireless power transfer was first explored by th e famous scientist Nicolas Tesla, seen in figure 1-3 A. Nicolas Tesla was the first to in vent the AC inductive motor in the late 1870s. Edison hired Tesla to work for him and proposed to pay him $50,000 if he was to succeed in creating AC power, mainly because Edison thought it was too difficult. Tesla took this as an easy job and came up with the AC power using his indu ctive AC motor, seen in figure 1-3 B. Edison was perhaps shocked when he found out that Tesla had succeeded and refused to pay him saying that the offer had been made in jest. Tesla imme diate resigned, and challenged one of the greatest scientists Thomas Edison with his AC power invention. Tesla took his AC power invention and considered it to be far more superior in nature than Edisons DC power station. 17

PAGE 18

A B Figure 1-3. Nicolas Tesla and the AC motor. A) Scientist Nicolas Tesla B) First AC inductive motor Tesla and Edison stood in competition to harn ess the power of the Niagara Falls. With Teslas Chicago exposition City of Light in 1890, where he proudly showed off his AC power station lighting hundreds of thousands of incandescent lamps in the city, and with the 27 million people who attended the fair, it was dramatically clear that the power of the future was the AC power. Teslas proposal to harness the power of the Niagara Falls was his childhood dream, and it was the only proposal to be accepted, and came into existence in 1893. Even Thomas Edison was finally convinced that AC power was a must and essentially powered all his labs with it. Another one of Teslas invention was the so called Tesla coil, where he would take his regular AC power and by using transformers, he was able to boost the voltage, while dropping current to maintain conservation of power, up to couple hundr ed of thousands of volts which created a high electrical field between the ne gative end terminal of the transformer and earth ground, which essentially caused sparks to jump off, or cha nnel through. He also noticed that the sparks where creating a high magnetic field that was illuminating incandescent lamps in the vicinity. In reality, this is the effect of electr ons releasing photons off the surf ace of the copper wires of the 18

PAGE 19

transformer and creating radio waves that would propagate around the transformer, such that the high power radio waves would light up the incandescent lamp in the vicinity. This discovery essentially consumed Tesla with excitement, and perhaps drove him crazy with the endless possibilities he could make of it. He immediat ely proposed his theory that power can be channeled through air to light up incandescent la mps and essentially an entire city, and was actually successful at de monstrating it, seen in figure 1-4 A. To prove his power transfer theory, he built the highest Tesla coil that stood as a tower in Pikes Peak looking over Colorado Springs as seen in figure 1-4 B. At th e point of completion and testing, Tesla turned on the giant Tesla coil and electricity arced out hundr eds of feet over the tower caus ing a complete burn out of all incandescent lights in the city and the Tesla tower had set itself on fire. T hus, causing a complete blackout of the entire city, and was then ch arged for the damages caused, which led to his bankruptcy. Tesla even proposed to transmit info rmation wirelessly around the globe, seen in figure 1-4 C, such that people w ould be able to share informati on such as messages, pictures, audio, and secure military communication, and was the first to propose the concept of the radar. He was also the first to create the first remote c ontrolled robot boat in 1898, as seen in figure 1-4 D. However people thought that all his proposal s where too dreamful and impractical which is understandable since he lived in a time period wh en people where astonished to even witness Edisons creation of the light bulb. However, it was not just one lamp he wanted to light up; it was millions of lamps and essentially the entire city. Tesla, with his over 700 patents, is now known as the scientist born out of his time.9 Indeed, here I am now, a 100 hears later demonstrating what Tesla had wanted to do. Tesla co il effect can be seen today in devices sold as taser-guns for self defense purposes. Also, ioni zers use the same concept Tesla proposed to ionize the air which gives us the feeling of fr esh air. However, Tesla wanted to use this 19

PAGE 20

electricity to create a channel to transfer the power to other locations without using expensive copper wires. If Nicolas Tesla wa s to say three last words before he died he would have said Wireless Power Transfer. A B C D Figure 1-4. Tesla, master of light ning. A) Tesla demonstrating wire less power transfer B) Tesla attempts to build a wireless power supply towe r. C) Tesla plan to transfer wirelessly power the world and transfer wireless information around the globe. D) First remote controlled robot built by Tesla. 20

PAGE 21

CHAPTER 2 DESIGN PLANNING AND CONSIDERATIONS Friis Free Space Loss In order to achieve wireless power transfer at high frequencies, we would need to closely study the derived equation of Harald T. Friis, know as the Friis free spac e loss equation, or just Friis equation.10 Friis explains through his derived equation what effects the radio signal exhibits while propagating through the air. However, befo re we look at the Frii s equation, we will set up the typical scenario of the sy stem from the transmitter (also know as the generator) to the receiver end (also known as the detector, or RF to DC sensor), as seen in figure 2-1. PT GT Transmitter (RF Generator) PR GR Receiver (RF to DC sensor) R Figure 2-1. Basic system setup for wireless po wer transfer showing the antennas on both the transmitter and receiver, separated by radius distance R. Shown below is the Friis equation describing RF signal strength received as a function of distance R, transmitted power Pt, and antenna gains Gt and Gr. P r 4 R 2G t G r P t (2-1) Such that, Pr is the power received, lambda is the wavelength of the radio signal, Pt is the power transmitted by the RF generator, Gt is the directivity gain of the antenna attached to the RF generator, and Gr is the directivity gain of the antenna attached to the RF to DC sensor. The receiver is also known as RF to DC sensor, wh ich is a description chosen by the author simply because the receiver could also be used as an RF radiation sensor for a particular band depending 21

PAGE 22

on the frequency of choice. By taking the 10Log( ) of the Friis Equation, we transform the power from units of mWatts to dBm, and thus we obtain the equation below. P r P t 20Log c f 20Log2 R G t G r (2-1) In equation (3) we have replaced lambda with its equivalent value (c / f), such that c is the speed of light (299,792,458 meters / second), and f is the frequency of the RF signal in Hz. Also, Pr and Pt are described in units of dBm, Gt and Gr both have units of dB, and R has units of meters. Next, we plot the effect of frequency f on the power received Pr in terms of the distance between transmitter and receiver, such that Pt =0dBm, Pr=0dBm, Gt=0dB, and Gr=0dB as seen in figure 2-2. We notice from Figure 22 that power attenuation is less for lower frequencies. For example, at one meter and 2.45 GHz signal, the power received is about -40dBm from transmitter. While at 434MHz, the power received is about -25dBm. This means that at higher frequencies the signal suffers from more atmos pheric absorption than lower frequencies. This effect is shown in figure 2-3. The question is now which frequency should be considered for wireless power transfer and at what cost? Th is question could be answered in many ways depending on the particular purpose of the a pplication. Also one should consider all the complications and effects the frequency has on th e system design. For example, the antenna size is characterized by the wavelength which is indirectly proportional to frequency. Therefore, as frequency decreases the antenna dimensions increase. This might be problem for self powered wireless sensors that have size constraints to meet, which would set a limit on the lowest choice of frequency. Therefore, due to size constraints, the author ha s chosen a higher frequency to enable wireless power transfer at 2.45GHz. Anot her advantage with operating at 2.45GHz is the high transmit power allowed at that frequency by the FCC. In addition, due to the popular 22

PAGE 23

2.4GHz ISM band, components are readily av ailable at relatively cheap prices. Figure 2-2. Received power (dBm) at the receiver vs. distance (meter) away from receiver, such that Pt =0dBm, Pr=0dBm, Gt=0dB, and Gr=0dB. The main focus of wireless power transf er is to increase the received power Pr as much as possible, such that one would be able to opera te at further distances. Looking at equation (2-1) we notice that we can play around with a number of variables that woul d allow us to achieve maximum power transfer, using the following methods: Increasing the transmitted power would be an obvious solution to in creasing the received power. The concept is simple here, the mo re power you transmit, the more power you receive. By choosing antennas with higher directivity gain on both the transm itter and the receiver would increase the delivery of wireless power to the receiver by means of concentrating the radio beam in one direction. 23

PAGE 24

Another obvious solution for in creasing received power would be to decrease the operating distance R, in other words, operate at a closer distance to the transmitter. By decreasing the signal frequenc y we would also be able to in crease the received power at the receiver. There are a number of complications for each of the above solutions. First, increasing the transmitted power means that the generator woul d have to use a higher power amplifier. High power amplifiers become more expensive and mo stly sold to defense contractors. Also, high power amplifiers consume significant amount of pow er and usually have low efficiency 40% to 60%, which means that the rest of the power is tr ansformed into heat. Therefore, heat-sinks and temperature sensors become a must in order not to destroy the expensive amplifier. The author was able to purchase a relatively cheap am plifier ZRL-3500 from minicircuits.com with maximum output power of +21dBm, which equiva lent to an output pow er of 125mW. The ZRL3500 is considered to be a relativ ely low power amplifier for the purpose wireless power transfer. However, ZRL-3500 is priced at $135 as compared to the ZVE-8G $1,095 with an output power of 1Watt, seen in figure 2-3. A B Figure 2-3. Medium power amplifiers. A) ZR L-3500 with output power 125mW B) ZVE-8G with output power 1Watt. Second, the antenna of choice is usually dependant on the type of the application such that the antenna size could be a constraint. If the ante nna size is not an issue, then one could design an antenna or choose to deploy antenna arrays with higher directivity gain at the cost of size and 24

PAGE 25

narrower radiation beam. Many othe r issues are involved with the choice of antennas, which is why there is a separate section that discu sses the choice of antenna in more details. Third, by brining the receiver closer to the transmitter, the signal received would be of higher power, which is a trivial solution. However, one might require the receiver to be at a minimum distance away from the transmitter for a pa rticular application to be feasible. On the other hand, some other applications, requiring wireless power transfer, might not have a constraint on a set distance between transmitter and receiver, such as recharging a device that is completely submerged in water. One would remove the device from under water to recharge it without disassembling it, which might cause wear and tear and eventually might start leaking water. Therefore, distance between the device and transmitter could be as close as possible. Finally, by decreasing the operating frequency one would be able to transmit power at further distances. However, there are many issues with choosing the right frequency for the right application, and in effect the fre quency of choice would affect all other variables. For this reason, one would have to carefully choose the operatin g frequency of the system. In addition, many factors could affect the choice of frequency such as the Federal Communications Commission (FCC) regulation, Industrial Scie ntific and Medical (ISM) ava ilable bands, antenna size, and available resources. FCC has a set of available license free bands called the ISM bands seen in table 2-1 located at different parts of the frequency spectrum seen in figure 2-4. ISM bands are regulated by the FCC with allowed effective radi ated power ERP, effect ive radiated isotropic power EIRP, active time period in transmit mode, interference, band width, and more. Table 2-1 shows the popular ISM bands av ailable with the allowed output power transmission. It is important to note the wavelength of the availa ble ISM band because antenna dimensions are dependent on it. 25

PAGE 26

Table 2-1. Available ISM bands for wirele ss power transfer and applications Frequency Band Allowed output power, application <125 KHz Near Field Inductive RFID 1.95, 3.25, 8.2 MHz Near Field Inductive theft tags 13.56 MHz Near Field Inductive RFID, RC Toys 27 MHz 0.1 Watt ERP 138MHz 0.05 Watt ERP, Duty Cycle <1% 402 405 Medical Implants, 25uW ERP 433.05 434.79 25mWatt ERP, Duty Cycle < 10%, RF controllers, and keyless entry 869.4 869.65 MHz 0.5Watt ERP, Duty Cycle <10%, RF controllers, and keyless entry 902 928 MHZ 4W EIRP 802.15.4 Zigbee, wireless modems, keyless entry, WPT 2400 2483.5 MHz 4W EIRP, 802.11b,g 802.15.4 Zigbee, WPT 5725 5875 MHz 25mWatt EIRP 802.11a 24.00 24.25 GHz 0.1Watt EIRP, police radars 61.0 61.5 GHz 0.1Watt EIRP 122 123 GHz 0.1Watt EIRP 244 246 GHz 0.1Watt EIRP Figure 2-4. Electromagnetic frequency spectrum and associated wavelengths. Borrowed from The RF and Microwave Handbook Antenna Types Antennas are made in many different shapes, sizes, lengths, power ratings, and for different purposes. However, antennas are mainly c hosen to fit the application rather than the other way around. Therefore, usually antennas are sp ecially designed or specially ordered to fit the purpose of the application. Th ere are, however, many commerci ally available antennas one can choose from, and due to the high availability of antennas, one could easily shop around for 26

PAGE 27

an antenna that would work for the application, or at least close enough. However, in a project such as wireless power transfer, we notice that the type of antenna is, indeed, an important factor of making power transfer feasible, which will be explained in later sections. Common popular antennas have relatively known attributes and aspects and are discussed below. Popular antennas include th e monopole whip antenna,11 which has omni-directional radiation pattern, directiv ity gain typically about 3dBi, length equal to wave length on top of a ground plane, and 10% band width, as seen in Figure 2-5. They are cheap to produce, and get the job done fairly well in communica tion systems. Monopole antennas might work well for wireless power transfer applications requi ring even power distribution in al l directions. However, the low directivity gain and omni-directi onal radiation pattern might also consider this type of antenna unattractive for an application re quiring high concentrated power that is focused in one direction towards an area of wireless sensor nodes. For a good performance monopole whip antennas, one could use a solid tinned copper wire anywhere from 14 to 18 AWG for good conductivity and low thermal loss. Ground Figure 2-5. Monopole whip antenn a on top of a ground plane The dipole antenna is also anot her popular type of antenna w ith very similar properties to that of a monopole. The difference between a dipo le and a monopole is that a dipole has twice the length of that of a monopole. Also, another main difference is that a dipole can be fed differentially. A dipole can be turned into a monopole by replacing one of the dipoles into a 27

PAGE 28

ground place, essentially what is seen in Figure 2-5. Dipoles are just as attractive as monopoles for wireless power transfer, and application dependent. A good point to make about both monopoles and dipoles is that they are greatly affected by ground planes, and the only ground plane in the vicinity should be constructed as seen in Figure 2-5. Mounting a monopole or dipole around objects and metals greatly ch anges the characterist ics of the antenna, such as impedance, radiation patter, and bandwidth. A dipole antenna is seen in Figure 2-6. Figure 2-6. Dipole antenna on Another similar antenna to the dipole and monopole is the helical antenna. A helical antenna operates in two complete different modes. The first mode is the normal mode, seen in Figure 2-7A, which is also know as the broads ide helical antenna. The naming broadside comes from the omni-directional radiation pattern of the farfield. In norm al mode, the main advantage is the short length of the antenna such that the windings of the antenna are shorter than the wavelength. In normal mode, the an tenna acts as short monopole w ith omni-directional radiation pattern and poor gain. However, the small factor of a helical antenna makes it attractive to mobile devices such as mobile phones. For wi reless power transfer, a normal mode helical antenna would not be good choice for RF generator due to its low gain. However, on the receiver side, the RF to DC sensor could be a good possible choice if the remote sensor is restraint to size. The second mode of operation is ca lled the axial mode, which is al so known as end-fire mode, as seen in Figure 2-7 B. The endfire mode nami ng comes from the high di rectivity gain of the antenna and concentration of radi ation at the end of the helical structure. The biggest advantage of the endfire mode is the high high gain >10dB and its simplicity. The disadvantage is the larger 28

PAGE 29

dimensions. This form of antenna is found in satellite communicat ion and long distance communication. As discussed in the previous section, higher direc tivity gain antennas can deliver more wireless power and increase operating dist ances. Therefore, axial mode antennas can be very useful and attractive for wi reless power transfer applicati ons especially on the generator side. However, an axial mode helical antenna mi ght not be very practical for operation on the wireless sensor node due to size constraints. Therefore one co uld choose two different antennas for operation the large axial mode on the gene rator, and the compact normal mode on the wireless node. A B Figure 2-7. Helical antennas. A) Normal mode he lical antenna B) Axial mode helical antenna The Yaggi antenna is a worthy antenna to menti on due to its high directivity gain >10dB. It was first discovered by a Japanese professor duri ng the world war two, and was named after him. Yaggi antennas are also very attractive for wire less power transfer for the same reason described in the axial mode helical antenna. There are so many different t ypes of antennas one can choose from, however, the author is mainly interested in printed patch antennas, which will be described in details in the next section. Printed Patch Antenna The patch antenna was chosen for this project due to the many advantages patch antennas have over others. Listed belo w are the basic advantages of using a patch antenna. Cheap fabrication costs over other antennas, su ch that it does not require any components other than etching a copper trace on a printed circuit board (PCB). 29

PAGE 30

Patch antennas have flat surface which give s them a low profile compared to other antennas like a monopole. Due to their flat surface, patch antennas are less susceptible to break or wear and tear. Patch antennas can be mounted on top of me tals without affecting antenna impedance. Readily available tools can simula te a patch antenna very easily. Patch antennas can be combined in arrays to increase directivity gain. Patch antennas have narrow bandwidth which acts as a natural band pass filter. The simplest form of a patch antenna is compos ed of three levels, seen in figure 2-8 A. The top layer is the copper patch of length L and widt h W. The resonant frequency is affected by the length L, such that L is approximately equal to before fringing. Therefore, the wavelength is initially equally to twice the length before fri nging. The middle layer is the dielectric material, which is the fiber glass material known as FR4 with relative di electric constant ( r) ranging from 4.2 to 4.7 and thickness h, as seen in figure 2.8 B. The third layer is the bottom ground layer. The simulation software used in this project to model the patch antenna is Ansoft Designeer, which uses dielectric material average for FR4 equal to 4.4 value. A B Feed Ground h L W L L L r FR4 h Figure 2-8. Patch antenna. A) Square patch antenna on a printed circuit board B) Cross section of board showing fringing effect. Patch antennas have the same characteristics as a transmission line, and are modeled the same way as a transmission line. Radiation occurs from the fringing fields, which extend the actually dimensions of the patch by L,12 as seen in figure 2-8 B. FR4 material was chosen for 30

PAGE 31

this project due to its low cost and availability. Other dielectric materials can be used for patch antennas that would actually enhanc e many features of the antenna.13,14 However, it would cost significantly more, which is inc onvenient for a large number of self powered wireless sensor nodes. Mathematical Description This section describes the mathematical m odel of a transmission line used as a patch antenna. A patch antenna is nothing but a tran smission line modeled to resonate at a high frequency. Since this patch antenna will be us ed for self powered wireless sensors, the dimensions will have to be fairly small to fit in as many situations as possible, which is done one way by increasing frequency. However, by incr easing frequency, the signal suffers more attenuation. Therefore, the aut hor chose to operate at the fr equency 2.45GHz, which is a popular ISM band used for WLAN, thus, com ponents are more readily availabl e than other frequencies. c f r (2-2) In equation (2-2), is the wave length in the dielectric material, f is the frequency (2.45GHz), c is the speed of li ght (299,792,458 meters / second), and r is the dielectric constant of the FR4 material (4.4). From equation (2-2 ), we get a rough estimate of the wavelength =5.834 cm, such that is equal to L which is equal to 2.917cm. However, we need a better estimate than that. wk ()k L (2-3) In equation (2-3) w is the width of the patch, L is the length of the patch before fringing, and k is a constant coefficient repr esenting the ratio of the width to length, such that k is equal to one for a square patch. In addition, the notation w( k) represents the width of the antenna as a function of k. 31

PAGE 32

uk () wk () h (2-4) In equation (2-4), u is the ratio of width w to the height h (1.5mm) of the dielectric material. For a square patc h, u(k) is equal to 19.455. re k () r 1 2 r 1 2 1 12 uk () 0.5 (2-5) In equation (2-5), re is the effective relative dielec tric constant. For a square patch, re is equal to 4.037. Notice that re is slightly lower than r because the fringing fields around the patch are not completely confined in the dielectric material. Lk ()0.412h re k()0.3 re k()0.258 uk ()0.264 uk ()0.813 (2-6) In equation (2-6), L is the added virtual length to the ac tual length of the microstrip patch antenna, due to fringing effect. For a square patch, L is equal to 0.7mm, such that the total length is now L + 2 L, or L + 1.4mm. This is a significan t change for a high frequency patch that is characterized by narrow ba ndwidth, and this is why we can not just estimate the length from equation (2-2). f o k () c 2L2 Lk () re k () (2-7) In equation (2-7), fo is the resonant frequency of the patc h antenna as a function of width to length k, and after the fringing effect. By plotting fo in terms of k, we can solve for k that would produce a resonant frequency equal to 2.45GHz, as seen in figure 2-9. 32

PAGE 33

Figure 2-9. Effective length of patch after fringing effect Vs ratio of width to length k. We notice from figure 2-9 that the resona nt frequency 2.45GHz occurs at a k value equal to 0.91, which is slight less that a square patch. L e k () c 2f re k () 2 Lk () (2-8) In equation (2-8), Le is the effective length after fringi ng, f is the resona nt frequency. By graphing Le as a function of k, we find the resonant le ngth L that corresponds to a frequency of 2.45GHz, as seen in Figure 2.9 below. 33

PAGE 34

Figure 2-10. Effective length of patch after fringing effect Vs ratio of width to length k. From figure 2-10 we notice that for a resonant frequency 2.45GHz, given k is 0.91 from Figure 2-9, we find that the effective length of the patch antenna is 2.917cm. From equation (2-3) we find that the wi dth of the antenna is about 2.654cm. Antenna Simulation Results There are many software that will do elec tromagnetic simulation based on method of momentum (MoM). The author had chosen two softwares (PCAAD and Ansoft Designer) for simulation to make sure that the two results were similar and consistent with the mathematical model. The first software PCAAD uses regular mathematical equations to calculate antenna characteristics as opposed to Ansoft Designer that simulates the electromagnetic field to find the characteristics of the patch antenna. Seen in figure 2-11 is the simulation results for a patch antenna using PCAAD. 34

PAGE 35

Figure 2-11. Patch antenna simulation using PCAAD. In figure 2-11 the author varied the antenna dimensions length and width to obtain a resonant frequency as close to 2.45GHz as possi ble. The length found was 2.81cm and the width being 2.9, which is close enough to the mathematical model described previously. We notice that the bandwidth is only 2.8%, e fficiency is about 50.2% and gain is 6.0dB. In addition, the magnitude of the input impedance of the feed lin e is about 104 which would need to be matched to standard 50 Using Ansoft Designer is not as straight forward as PC AAD, however, it allows for much more in depth analysis of the patch antenna Seen in figure 2-12 is the layout of the patch antenna in Ansoft Designer after optimizing dimensions 50 match. 35

PAGE 36

Figure 2-12. Square patch antenna using Ansoft Designer, with given dimensions, and output matched to standard 50 using stub stripline of length 12mm by 29mm. Seen in figure 2-13 is the re turn loss S11 (dB) in terms of the swept frequency F(GHz). Return loss basically shows at what frequency the 50 is happening. The lower the return loss in dB, the better the match is. We notice that the return loss at 2.45GHz ( pointer 1) is about 24.7dB, and the -10 dB bandwidth extends from about 2.4GHz to 2.48GHz, which is really good match for the 2.4GHz ISM band. One would try to match at the center of the 2,4GHz ISM band because after fabrication of the patch antenna, there will definite ly be process variations, such that the dielectric material would probably be di fferent than 4.4, or thic kness would be different, and so on. 36

PAGE 37

Figure 2-13. Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz. The gain of this antenna is shown in Fi gure 2-14, which has good distribution of gain centered at the operating frequency 2.45GHz. As disc ussed previously, the gain of the antenna is a really important feature of wi reless power transfer and self powered wireless sensor nodes. The gain obtained from the antenna at the cost of di rectivity is almost considered free power, since you are not paying for it with current consump tion as the case with the power amplifier. Therefore, one could cascade an array of antenna s to increase the gain of the system. However, the main disadvantage of antenna arrays is the larg er size, which doubles for every 3 dB of gain. 37

PAGE 38

Figure 2-14. Antenna gain of patch antenna (d B) in terms of the swept frequency F(GHz). The patch antenna was fabricated using a home-made etching technique using ferric chloride and blue press-n-peel sheets as seen in figure 2-15. The dimensions of the edges were not as accurate as simulated, and the antenna went through corrosion by time. Therefore, the performance of the antenna was not satisfactor y, especially for a sens itive application as remotely powered wireless sensors. The matchi ng greatly affects effici ency and gain, homebrewed antennas for this proj ect produce unacceptable results. 38

PAGE 39

Figure 2-15. Fabricated 2.45GHz antenna using homebrewed etching technique. Another antenna was simulated with a different matching technique, and was sent to a professional fabrication company PCBexpress for accurate fabricat ion results. The antenna was also simulated using Ansoft Designer and optimized to match at 2.45GHz, as seen in figure 2-16. Figure 2-16. Square patch antenna us ing Ansoft Designer matched to 50 at 2.45GHz. 39

PAGE 40

Seen in figure 2-17 is the return loss S11 (dB) in terms of the swept frequency F(GHz). We notice that the return loss at 2.45GHz (pointer 1) is -25.2dB, whic h is better than the previous version, and the -10 dB bandwidth ex tends from about 2.42GHz to 2.48GHz. Figure 2-17. Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz. The gain of this antenna is shown in Fi gure 2-18, and has a better gain distribution centered at 2.45GHz than the previous antenna. 40

PAGE 41

Figure 2-18. Antenna gain of patch antenna (dB) in terms of the swept frequency F(GHz). Gain is about 1.4dB or 3.5dBi. This patch antenna was sent to a professi onal PCB fabrication company PCBexpress as opposed to the previous antenna which fabricated in lab. The dimensions of the edges were precisely as designed and very close simulation results. In ad dition, the antenna has a green silkscreen which prevents the copper from corrosion. 41

PAGE 42

Figure 2-19. Patch antenna for 2.45GHz ISM band, with 3.5dBi gain. Fabricated at PCBexpress. This patch antenna can be used for both the remotely powered sensor node, and the RF signal generator. However, one might consider using a higher antenna ga in on the RF generator since size constraint would not be an issue. Therefore, for the 2.45GHz RF generator, an array antenna was developed similar to the previous antenna, but has twice the gain. The antenna is composed of two cascaded patches as seen in Figure 2-20. Figure 2-20. Patch antenna array, matched at 2.45GHz, with gain equal to 6dBi. The higher the gain the RF gene rator has, the more power is delivered to the wireless node. Therefore, one can purchase an already made an tenna array that has a larger bandwidth, and higher gain than a patch antenna fabricated on a FR4 board. 42

PAGE 43

Power Delivery Using the fabricated 3.5dBi 2.45GHz patch ante nna in figure 2-19 on the wireless sensor node, and the 6dBi 2.45GHz antenna at the RF gene rator, with transmitter output power equal to that of the amplifier in figure 2-3 A, we revisi t equation (3) and figure 22. In equation (3), Pt is set equal to +21dBm, which is the ma ximum output power of the ZRL-3500. Gt is set equal to 4dB (not dBi), which is the directivity gain of the patch an tenna array located on the RF generator seen in figure 2-20. Gr is set equal to 1.5dB, which is the directivity gain of the patch antenna located on self powered wireless sensor node. Therefore, we obtain the equation below. P r Rf ()20log c f 20log4 R 26.5 (2-9) In equation (2-9), Pt, Gt, and Gr add up to 26.5dB. Therefore, th e received power is 26.5dB higher than that of figure 2-2. Notice in figure 2-21 how all th e graphs have shifted up 26.5dB. Figure 2-21. Improved wireless power received with additional 26.5dB gain. 43

PAGE 44

Commercially Available Antennas for WPT Even though we have made a significant 26.5dB improvement, the theoretical received power at one meter is only -15dBm, which is not enough power for wireless power delivery. Therefore, we would need to increase the power delivered in any means possible. This means that a higher antenna gain at the transmitte r is required. HyperLinkTech sell HG2414P patch antenna array for WLAN use with high antenna gain equal to 14 dBi. This antenna would be used on the RF generator due to its large size. HyperLinkTech also offers Yaggi antennas and patch antenna arrays that have gains > 10dBi, and are readily av ailable for customer purchase, seen in Figure 2-21 A through E. A B C D E Figure 2-22. High gain commercia lly available antennas suitable for remote power transfer Unfortunately, for the remotely powered wireless sensors, it is difficult to increase antenna size due to size constraint. Howe ver, by using recent technology as chip antennas, one can cascade high gain chip antennas while maintain ing similar node dimensions. Therefore, the author had found the following chip antenna 2450AT45A100 suitable for the purpose of reducing the size of remotely powered wirele ss sensors, as seen in figure 2-22 A. A B Figure 2-23. Commercial chip antenna. A) Ch ip Antenna B) Dimensions of Chip antenna. Borrowed from JohansonTechnology website. 44

PAGE 45

The chip antenna is manufactured by JohansonTechnology with average gain equal to 1dBi, and maximum gain equal to 3dBi. Th e chip antenna is weakly matched to 50 at 2.45GHz, with a return loss equal to -14dB. Howeve r, the company also offers a matching circuit for better bandwidth(BW), gain, and re turn loss seen in figure 2-24 A, return loss figure 2-24 B. A B Figure 2-23. A) Matching network B) Re turn loss S11(-22.6dB), and BW (600MHz) Attenuation The main reason for the attenuation of the wire less signal in air is due to the atmospheric absorption due to the presence of water particle s in the air. As seen in figure 2-24, the atmospheric absorption is very high for 2.4GHz signal. This is the main disadvantage of selecting a 2.4GHz frequency for wireless power transfer. However, it is a great advantage for microwave ovens. As a matter of fact, it is th e only reason why microwave ovens operate at 2.4GHz because water absorption is very high at that frequency, which means that food containing water would cook very fast. However, one must not think that WPT is harmful for humans, for two simple reasons. First, the ouput power in its best is limited to 4 watts EIRP, which is well below that of a microwave oven 1000watts. Second, in order to cause any kind of severe harm to humans the wavelength would ha ve to be small enough to resonate with the length of human cells and DNAs such as that of X -rays. A quick note to men tion is that the radio signal produced by the microwave oven (2.4GHz) is not even a microwave scale wave. The 45

PAGE 46

wavelength of a 2.4GHz frequency is actually ab out 12.2 cm. However, it is called microwave because looking at table 2-1, you can see that it lies in the broad microwave spectrum. Multipath fading is also another reason for si gnal attenuation especially if the wireless power transfer is being done indoors. Multip ath fading could either by constructive or destructive. However, in most case it is usua lly destructive due to the signal bouncing off multiple walls back and forth. Multipath fading will reduce the amount of power transferred to the remotely powered wireless sensor and also cause communication errors in the channel. An open space would be a better environment for wire less power transfer. As a matter of fact, the best place for WPT is actually outer space where there is no water or oxygen absorption to attenuate the radio signal transmitted, but perhaps more interference and noise is the side effect. In addition, Friis equation mostly represents a theoretical received power in reality one should only consider about of the theoretical distan ce calculated, and this is being optimistic. Feasibility WPT would be feasible and useful with the availability of high power amplifiers starting with at least a one watt output power, and a high gain antenna in the order of 14dBi and higher. Also, areas with less humidity would be better environment for deploying remotely powered wireless sensors. This is the reason why Nicola s Tesla decided to build his Tesla coil tower high above sea level where humidity is much less. Th e authors best case scen ario is the available +21dBm power amplifier, 14dBi Antenna and 3dBi ch ip antenna. This gives a total system gain of about 34dB, which is about -7dBm theoretical received power according to equation (3). In the worse case scenario, the antenna would only have about 8dB gain, transmitter would transmit about 18dBm, and receiver chip antenna would be 0dB, such that the total system gain is now +26dBm instead, which is more of an expected gain rather than a worse case. Referring back to 46

PAGE 47

equation (3) a +26dBm system gain is equivale nt to about -15dBm received power. However, according to the author, in order to receive a d ecent amount of power to work with, one would need a minimum of 0dBm, and preferably gr eater than +10dBm. Assuming a high power amplifier such as a 1Watt or a +30dBm, w ith a high gain antenna about +22dBi on the transmitter and a 2 dBi chip antenna on the wire less sensor node, one can achieve about +50dBm of system gain from transmitter to receiver. This much power would transfer a theoretical +10dBm power to the wireless sensor node. However, a more practical received power would be to subtract 6dB, which is equivalent to divi ding the distance by 4, giving a +4dBm of received power, which is still a good amount of power to wo rk with. This would allow a wireless sensor node to receive wireless power and operate comfor tably about 2 meters away from transmitter. Two meters does not sound like much of a distance, however, its a good starting point for remotely powered wireless sensors. 47

PAGE 48

CHAPTER 3 WIRELESS POWER TRANSFER SYSTEM DESIGN System Level Design The system level of remotely powered wireless sensors is mainly divided into two parts. The first part is the RF generator and the sec ond is the self powered wireless sensor. The RF generator is responsible to generate the high power radio signal and radiate it towards the wireless sensor. Power for the RF generator is considered to be a bundant and constantly available. ON the other hand, the wireless se nsor maintains a low power state and energy conservation via energy management system and low power software selectable states. The wireless sensor has the responsibil ity to detect the high power signa l and rectify it into usable DC energy, stored locally for future use. Figure 31 shows the system level design of the system. Figure 3-1. Top level system design. In figure 3-1, the RF generator shown to th e left generates the hi gh power RF signal and propagates it in the direction of the wireless se nsor. Multiple antenna arrays can be used to propagate a concentrated RF signal in the general direction of sens ors in the vicinity of the RF generator. A splitter can be used to split the RF signal into multiple sources at the cost of less signal power. The middle part of figure 3-1 is the concentrated RF signal propagating in air, and on the right is the remotely powered wireless se nsor detecting and rectifying the RF signal. 48

PAGE 49

System Design Constraints The system constraints of the RF generator are far more flexible than that of the remotely powered wireless sensor. The RF ge nerator is considered to have infinite amount of available power, yet it would also have the capability to operate with a large battery if required. The current consumption of the RF generator is hi ghly dominated by the RF power amplifier. The ZRL-3500 consumes about 575m A at +12volts DC input. The remotely powered wireless sensor has far more constraints than th e RF generator. First of all the wireless sensor has to be optimized for lowest power consumption possible. The author recommends a total power consumption less than 1mWatt in active wireless transmission state, and less than 1uWatts in sleep mode. The remotely powered sensor has to be as small as possible in order to be adopted in any application and in any size limite d environment. Therefore, the wireless sensor node would use a chip antenna in stead of a patch to further reduce dimensions. Ultra capacitors can also be used on the wireless sensor nodes due to their smaller size and larger capacity. However, ultra capacitors have longer charging time due to their large capacity, which would be a problem for charging. Surface mount components are advisable for the wireless sensor node, and perhaps a final integrated AS IC would even further drop power consumption and size. Range Constraints Due to the main size constraint of the remote ly powered wireless sensor, the patch antenna would have to be replaced with the chip ante nna. This drops efficiency due to impedance mismatch between chip antenna and RF to DC se nsor. For this reason, th e total power received would be less, and thus the range would be redu ced. For a typical scenario, the RF generator is capable of outputting a typical +19dBm to + 21dBm of output power, transmitter antenna has a 49

PAGE 50

gain of about +8dBi to +12 dBi, and wireless sensor node has about -1dBi to +1dBi gain. Therefore, the total system gain ranges from +2 6dB to +32dB. According to equation (3), an average of +30dB system gain gives about -10d B of received power at one meter, and about +0dBm at 20cm away. As mentioned in previous sections, usable power would have to be at least +0dBm. Therefore, the rang e at the given constraints is limited to about 20cm. However, with a higher power amplifier, this could be easily extended. Cost Constraints The RF generator is perhaps the most expens ive unit in the design. However, there is only one RF generator in the system, so it is not a major issue in the system. The authors RF generator cost was about $400 in parts. However, this cost can easily increase with the replacement of the RF amplifier with a higher gain one. Amplifiers in the 1Watt range can easily start about $1,000 and up to couple thousand dolla rs for military RF amplifiers. The wireless sensor cost is much less than the RF generator, costing about $25 dollars with quantities greater than a hundred. The author did not address issues with reducing cost, but m onitored the increase in cost as the project develope d. In general, since wireless powe r transfer and remotely powered wireless sensors is not as popular yet, component s such as power amplifie rs are still expensive and range in thousands of dollars. However, in the future, the author believes as the concept picks up in the future, special designed high pow er amplifiers could be produced for cheaper prices. Radio Frequency Generator The RF generator is one of the two legs of the system. The RF generator design would either make or break the system. The system is mainly composed of a signal generator block, a high power amplifier, and a baseband processor. In addition, a large gr aphical LCD display is 50

PAGE 51

located on top of the RF generator for visual feed back. Figure 3-2 shows the back side of the RF generator. Figure 3-2. Back-side view of RF generator. Seen in figure 3-2 is the backside of the RF generator. Displayed are two fans mounted on top of a heat-sink. A temperature sensor monito rs the amplifiers temperature and turns on the two fans when the temperature goes over a preset limit. The fans are tied dow n with tie-straps to reduce vibration. Also see in figure 3-2 are 5 LEDs on the top left side. Four in which are yellow and correspond to up, down, left, and right, and one LED is green and displays a green color. These buttons are used for user input to navigate the menu on the LCD display. In figure 3-3, the front of the RF generator is displayed. 51

PAGE 52

Figure 3-3. Front-side view of RF generator. Seen in figure 3-3 is the frontside view of th e RF generator. First, there are two DC input power supplies with labels uC-DC +12-6V fo r the microcontroller power supply, and RF-DC +12V is used for for the RF power supplied to am plifier and signal generator. Second, there is a RS232 serial port, that would communicate locall y with the microcontro ller. Third, there is a microcontroller programmer uC Prog used to flash the microcontro ller program memory. Finally, there is the high power RF output port seen in the to p right corner with an SMA extended cable. Seen in figure 3-4 is the right side view of the RF generator. There are two turn knobs which are the RF tuner, and the LCD contrast tuner. The RF tuner wh en turned will adjust the the RF frequency. The frequency can be adjust ed from 2165MHz to 2650MHz, which in terms covers the required 2.45GHz range. The other knob ad justs the contrast of the display for better user view. 52

PAGE 53

Figure 3-4. Side view of RF generator, displaying tuning knobs. Radio Frequency Signal Source The main source of the RF gene rator is the local oscillator purchased from MiniCircuits with model number ZX95-2650 with tuning frequencies from 2165MHz to 2650MHz. The frequency is adjusted with the right knob seen in figure 3-4. The maximum output power of the oscillator is about +5dBm. Voltage supply ranges from +12 to +13 volts and current consumption is about 25mA. Seen in figure 3-5 is the oscillator ZX95-26 50 from MiniCircuits. Figure 3-5. Signal generator local oscillator from MiniCircuits. High Power Signal Amplifier The power amplifier was also purchased fr om MiniCircuits with model number ZRL-3500. The amplifier is powered from a +12 volts source and has an internal regulator at +5 volts. The 53

PAGE 54

maximum output power of the amp lifier is about +21dBm at best. This is considered to be a medium power amplifier. However, for wireless power transfer, this amplifier is considered to be a low power amplifier. Amplifiers with minimu m output power of 1watt should be considered for remote powered wireless sensors. The power amplifier is the most power hungry device in the system consuming about 575mA of current. Seen in figure 3-6 is the power amplifier ZRL3500 from MiniCircuits. Figure 3-6. RF amplifier ZRL-3500 from MiniCircuits. Antenna of Choice Antenna choice is a very important factor in WPT systems. Since the RF generator has less size constraints than the wireless nodes, the antenna size would not be a major concern. Therefore, the antenna size can be relatively large with high gain. Th e author recommends antennas with gain greater than 10dB, and prefer ably as high as possible. Antenna gain could make the wireless power transf er an easier task. However, th e higher the antenna gain, the narrower the radiation beam, or directivity, whic h means that the antenna would have to be directed towards the wireless sensor. The author chose the patch antenna array from HyperTech with model number HG2414P with directivity gain of +14dBi, seen in figure 3-7 A. In addition, the antenna has a 30 x 30 degrees vertical and hori zontal radiation beam se en in figure 3-7 B and 3-6 C respectively. 54

PAGE 55

A B C Figure 3-7. Commercial patch ante nna. A) Patch antenna HG2414P B) Vertical radiation pattern C) Horizontal radiation pattern RF to DC Wireless Sensor Node The RF to DC remotely powered wireless sensor is the other leg that makes this project walk. The RF to DC sensor is responsible fo r harvesting the radiated power from the RF generator and turning it into DC energy. The system design can be seen in figure 3-7. Figure 3-8. RF to DC wirele ss sensor node system design. Seen in figure 3-8 is the system level desi gn of the RF to DC remote powered wireless sensor. Starting from left to right, the first stage is the source of energy harvested. There are many sources of clean ambient energy that can be scavenged. We have discussed these sources in the introduction. Therefore, we will only disc uss the RF source in the green colored block. When the RF signal is received via matched ch ip antenna, the signal is fed into the signal rectifier. In this case, the si gnal rectifier is the voltage doub ling schottky diodes. The schottcky 55

PAGE 56

diode, also know as zero biased diodes (ZBD), have to have minimum forward voltage drop. When the signal is detected and turned into DC voltage, the voltage enters the DC to DC regulator, turning the received DC source into stable usable vo ltage such as 2.5volts or 2.0volts. The lower the required voltage is the less power is consumed, which is a concept we also discussed in the introduction. The regulated voltage is stored on a local capacitor, such as a supercapacitor. One should be careful with the capacitor capacity, too hi gh of capacitance would take too long to charge. The au thor recommends a one mFarad value capacitor for a fast charging response. The choice of capacitor is of course dependant on the application. The author uses a regular 1mFarad capacitor for storage. When the energy is successfully stored on a capacitor, the baseband section, seen in blue, can use this ener gy to monitor a sensor and send back a reading value. Antenna Matching The path antenna used is the 2450AT45A100 The chip antenna is manufactured by JohansonTechnology with average gain equal to 1dBi, and maxi mum gain equal to 3dBi. The chip antenna is weakly matched to 50 at 2.45GHz, with a return loss equal to -14dB. However, the company also offers a matching circuit for better bandwidth(BW), gain, and return loss seen in figure 2-24 A, return loss figure 2-24 B. By using the provided matching network, a better 50 match is possible with a wide bandwidth of 600MHz. A 1.5p Farad and a 3.9n Henry is required for matching. Impedance matching is im portant to achieve maximum power transfer with minimum power loss and reflection. RF to DC Matching Network It is important to match the RF to DC stage of the sensor. This will determine how efficient the wireless sensor is with converting RF ener gy to DC energy. Seen in figure 3-9 is the 56

PAGE 57

matching for the RF to DC sensor using smith char t technique. To match the input of the voltage doubler to the antenna a 13nHen ry inductor is required. Figure 3-9. RF to DC matching network. Seen in figure 3-9 is the voltage doubler RF to DC conversion with input matching to 50 The schottky diodes used are the ZBD HSMS-2850 from Agilent. The output voltage is as function of frequency is seen in figure 3-10, such that input power is 0dBm. Figure 3-10. RF to DC output voltage about 1.2 volts. 57

PAGE 58

Also the simulated current is shown in fi gure 3-11, such that the input power is also 0dBm. Figure 3-11. RF to DC ou tput current about 1.2 mA. The simulation shows that since the voltage a nd the current are of equal values, then the power is maximum, which means we have impedance matching at 2.45GHz. Low Power BaseBand Processor The baseband system is composed mainly of three blocks the PIC16F686 microcontroller from Microchip, the low power RF transmitter, and the piezoelectric sensor. When the microcontroller receives enough en ergy to power up, it will go di rectly into sleep mode to conserve the energy stored. The system is interru pt driven, meaning that when the piezoelectric sensor is actuated, the microcon troller wakes up from sleep and transmits its unique ID to the receiver base station at a frequency equal to 916 MHz. Most of the energy management is done by the microcontroller. The C code for the wireless sensor node is shown in appendix A. 58

PAGE 59

Low Power Sensors A unique sensor is used with the wireless se nsor node. The sensor is made from a piezo electric material that produces high voltage when stroke. This voltage has very low current, however, it is enough for CMOS technology to be trig gered. When the microcontroller is in sleep mode and the piezoelectric sensor is moved, it produces a voltage burst that interrupts the PIC from sleep mode. The importance of this kind of sensor is that it does not consume power to operate, it actually produces it. Low Power Wireless Transmitter The transmitter is a SAW based low power tr ansmitter. It operates at 916MHz frequency, which is different than the frequency suppl ying the power, so they dont interfere. The transmitter can operate on a low supply voltage as 1.2 volts and current of 1.2 mA which is much less than what a Zigbee node consume (50mA). Ther efore, with minimum duty cycles, the sensor can last forever, as long as there is a wireless power supply supplying power. 59

PAGE 60

APPENDIX A ELECTRONIC SCHEMATIC OF RF BASEBAD CONTROLLER This schematic was designed and implemented by the author. The schematic was designed you Protel DXP. 60

PAGE 61

APPENDIX B REMOTE POWERED WIRELESS SENSOR C CODE This is the C code implemented for the remo te powered wireless sensor node. The code was written using CCS compiler. #include <16F688.h> #device adc=8 #FUSES NOFCMEN //F ail-safe clock monitor disabl ed, switches to internal if external fails, page 33 for more details #FUSES IESO //Internal External Switch Over mode enabled //#FUSES NOIESO //Internal External Switch Over mode disabled, page 33 for more details //#FUSES BROWNOUT //Reset when brownout detected #FUSES NOBROWNOUT //No brownout reset #FUSES NOCPD //No EE protection #FUSES NOPROTECT //Code not protected from reading #FUSES MCLR //Master Clear pin enabled #FUSES NOPUT //No Power Up Timer, page 33 for more details #FUSES NOWDT //No Watch Dog Timer //#FUSES HS //High speed Osc (> 4mhz) #FUSES XT //High speed Osc (<= 4mhz) //#FUSES INTRC_IO //Internal RC Osc, no clkout //#FUSES INTRC //Internal RC Osc //#FUSES LP //Low power osc < 200 khz #use delay(clock=4000000) // 200K osc #use rs232(baud=38400,parity=N,xmit=PIN_C4,rcv =PIN_C5,bits=8) // Baud=2.5% Fosc %3.125 // assume Tx_bits= 32bits to be transmitted // time to transmit = Tx_bits 1/Baud = 32bits 1sec/3K bits = 11 msec // Power Consumed = 1.5v 1.5mA = 2.25mW for the transmitter // E = P* time = 2.25mW 11ms sec = 25uJ <<<<<<<<< Transmitter // will need 3 diodes in series, or half voltage divider 3.3v/2 = 1.6v // 1*Tao= 0.63*V=0.63*6.3v=4v // E = 0.5(C)(V)^2 = 0.5(C)(4)^2=25uJ => C=50uJ/16=3uF // E = 0.5(100uF)(4)^2=800uJ // Tao = R*C =10K*100uF = 1 sec to charge cap // P = 200uA*3.3v = 0.66mW (pic power consumed) // E = 0.66mW*11ms = 7uJ <<<<<<<<< PIC energy int i=0; 61

PAGE 62

//Baud Rates: // 110, 300, 1.2K, 2.4K, 4.8K, 9.6K, 19.2K 38.4K, 57.6K, 115.2K, 230.4K, 460.8K, 921.6K //#byte OSCCON = 0x8F //#byte OSCTUNE = 0x90 //#byte OSCTUNE = 0x90 //#byte TXSTA = 0x16 //#byte RCSTA = 0x17 #int_EXT EXT_isr() { //setup_oscillator(OSC_EXTERNA L); //OSC_31KHZ, OSC_125KHZ ,OSC_250KHZ, OSC_500KHZ //#use delay(clock=4000000) // 200K osc //#use rs232(baud=38400,pa rity=N,xmit=PIN_C4,rcv=PIN_C5,bits=8) // Baud=2.5% Fosc %3.125 delay_ms(10); //allow ascillator stability SETUP_UART(TRUE); printf("%cAA%C",0xFF,i); SETUP_UART(FALSE); output_low(pin_c4); i++; } void main() { char start_byte='A', device_id=' B', adc_data='C', crc_data='D', i; unsigned int int_counter=0; output_low(pin_c4); setup_adc_ports(NO_ANALOGS|VSS_VDD); setup_adc(ADC_OFF); setup_timer_0(RTCC_INTERNAL|RTCC_DIV_1); setup_timer_1(T1_DISABLED); //setup_timer_2(T2_DISAB LED,0,1); //for PIC16F684 setup_comparator(NC_NC); setup_vref(FALSE); 62

PAGE 63

ext_int_edge(L_TO_H); enable_interrupts(INT_EXT); //clear_interrupt(INT_EXT); enable_interrupts(GLOBAL); //#use delay(clock=4000000) // 200K osc //OSCCON = 0x21; while (1) { sleep(); } } 63

PAGE 64

APPENDIX C RF BASEBAD CONT ROLLER C CODE This is the C code implemented for the RF generator baseband processor. The code was written using CCS compiler. #include <16F777.h> #device *=16 #device adc=8 #fuses HS,NOWDT,NOBROWNOUT #use delay(clock=20000000) #include #include #include #use rs232(baud=19200,parity=N,xmit=PIN_C6,rcv=PIN_C7,bits=8,stream = pc) //#use rs232(DEBUGGER) #use fast_io(B) //#use fast_io(D) #include //LCD port definition -Done! #define gl_dat portb #define gl_a0 pin_d7 //output pin #define gl_cs pin_d6 //output pin #define gl_wr pin_d5 //output pin #define gl_rst pin_d4 //output pin //#define gl_rd pin_d4 // hard wired to VDD=+5V //Bump switches #define right_bump !input(pin_e0) //input bump switch #define left_bump !input (pin_e1) //input bump switch #define RF_ON output_high(pin_d1) #define RF_OFF output_low(pin_d1) //#define more_bump !input(pin_e2) //input bump switch #define pyro_an0 0 #define temp_an1 1 //IR port definitions #define GP2D02_VIN pin_d2 //green #define GP2D02_VOUT pin_d3 //yellow 64

PAGE 65

// servo constatnts //#define start_byte 0x80 //#define device_id 0x01 //LCD Constants used -Done! #define SYS_SET 0x40 #define SYS_SLEEP 0x53 #define SYS_CGRAM_ADDR 0x5C #define SYS_SCROLL 0x44 #define SYS_SCROLL_HDOT 0x5a #define SYS_SCROLL_RATE 0x5a #define SYS_CUR_FORM 0x5d #define SYS_CUR_ADDR 0x46 #define SYS_CUR_READ 0x47 #define SYS_CUR_DIR_RT 0x4c #define SYS_CUR_DIR_LT 0x4d #define SYS_CUR_DIR_UP 0x4e #define SYS_CUR_DIR_DN 0x4f #define SYS_OVER_LAY 0x5b #define SYS_MWRITE 0x42 #define SYS_MREAD 0x43 #define LCD_DISP_OFF 0x58 #define LCD_DISP_ON 0x59 #define LCD_INVERSE 0xff #define LCD_NORMAL 0x00 #define LCD_CR 0x20 //32 Chars/bytes per line #define LCD_W 0x100 //256 pixels wide #define LCD_H 0x80 //128 pixels high #define LCD_Lh 0x08 //Height of lin e (8x8 characters) #define LCD_L1 0x00 //Line 1 #define LCD_L2 0x20 //Line 2 #define LCD_L3 0x40 //Line 3 #define LCD_L4 0x60 //Line 4 #define LCD_L5 0x80 //Line 5 #define LCD_L6 0xA0 //Line 6 #define LCD_L7 0xC0 //Line 7 #define LCD_L8 0xE0 //Line 8 #define LCD_L9 0x100 //Line 9 #define LCD_L10 0x120 //Line 10 #define LCD_L11 0x140 //Line 11 #define LCD_L12 0x160 //Line 12 #define LCD_L13 0x180 //Line 13 #define LCD_L14 0x1A0 //Line 14 #define LCD_L15 0x1C0 //Line 15 #define LCD_L16 0x1E0 //Line 16 65

PAGE 66

#define LCD_GRH 0x1000 //Graphic home position // graphic LCD char line_string [33]; void gl_init(); void gl_grfclr(); void gl_grfhome(); void gl_strobe(); void gl_txtclr(); void gl_setaddr (int gl_cur, long gl_addr); void send_data (int gl_byte); void send_command (int gl_cmd); #separate void g_printf (char string[],long Line_num); void g_clear_line (long Line_num); void load_image(); void load_image() { int i,j; gl_txtclr(); send_command(SYS_CUR_ADDR); //CSRW command send_data(0x00); send_data(0x10); send_command(SYS_CUR_DIR_RT); //Cur movement right send_command(SYS_MWRITE); while(fgetc(pc)!=']') // after the ']' microc ontroller will lose 31 characters (becau se of software rs232) to initiate i and j again1: for(i=0;i<128;i++) //128/8=16 { for(j=0;j<32;j++) //256/8=32 { send_data(fgetc(pc)); } } goto again1; } 66

PAGE 67

int get_temp() { set_adc_channel(temp_an1); delay_ms(1); return (read_adc()<<1); } #include ////////////////// //////////MAIN()/////// /////////////////////////////////////// void main() { //LCD Global Variables used -Done! int gl_cmd; int gl_cur; int gl_byte; int gl_i; int gl_j; int gl_k; long gl_x; long gl_y; long gl_addr; long gl_w; int gl_read; int gl_old; long pix_x; long pix_y; long gl_addrlo; long gl_addrhi; int gl_nib; // servo control variables int servo_num; int right_arm_angle; int left_arm_angle; //general variables int i=0,j=0,k=0; //variables used for UltraSound long returned_distance; float distance; 67

PAGE 68

//variables used for ADC int pyro_adc; //16-bit integers float pyro_volts; int temp_cel; //should be long, but do ubt temperature would be that high float temp_volts; // CMUcam variables char cmu_char[10]; long cmu_param[8]; int ir_reading; int speed; //char bt_rx[33]; char bt_rx[3]; char rx_data[5]; int menu_option; //pwm variables: unsigned long pwm_freq; unsigned int t2p; unsigned int value; //some stuff setup_spi(FALSE); //setup_counters(RT CC_INTERNAL,RTCC_DIV_2); //setup_timer_0(RTCC_I NTERNAL| RTCC_DIV_8); //setup_timer_0(RTCC_INTER NAL| RTCC_DIV_64); //tim er_0 will overflow every 16.32 msec setup_timer_0(RTCC_INTERNAL|RTCC_DIV_128); //timer_0 will overflow every 32.7 msec, resolution = 128us (increments every 128us) //setup_timer_1(T1_DISABLED); setup_timer_1(T1_INTERNAL|T1_DIV_BY _1); //overflows every 65.5ms //setup_timer_2(T2_DIV_BY_16,255,16); //overflows every 65.5ms //setup_timer_2(T2_DISABLED,0,1); setup_comparator(NC_NC_NC_NC); setup_vref(FALSE); //set_timer0(0); //reset timer //set_timer1(0); //reset timer //set_timer2(0); //reset timer //counter0_overflow = 0; //counter1_overflow = 0; 68

PAGE 69

//counter2_overflow = 0; set_tris_b(0x00); //set_tris_d(0x00); //enable_interrupts(INT_TIMER0); //activate timer //enable_interrupts(INT_TIMER1); //activate timer //enable_interrupts(INT_TIMER2); //activate timer disable_interrupts(INT_TIMER0); //dectivate timer disable_interrupts(INT_TIMER1); //dectivate timer disable_interrupts(INT_TIMER2); //dectivate timer //enable_interrupts(GLOBAL); //activate global interrupt flag t2p = 255; setup_timer_2(T2_DIV_BY_16,t2p,1); //pwm_freq = 98KHz //setup_ccp1(CCP_PWM); // Configure CCP1 as a PWM setup_ccp2(CCP_PWM); //setup_ccp3(CCP_PWM); value = t2p>>1; //set_pwm1_duty(value); set_pwm2_duty(value); //LCD fprintf(pc,"\r\nInitia lizing Graphical LCD..."); output_low(gl_rst); delay_ms(100); output_high(gl_rst); //Make sure reset is high delay_ms(100); output_low(gl_cs); //Chip select active delay_ms(10); gl_init(); //Initialize display fprintf(pc,"Done!\r\n"); //////////////////////////////////// //////////////// //////////////// //////////////// 69

PAGE 70

strcpy(line_string, "Welcome to Wireless Power Trasfer!"); g_printf(line_string,LCD_L1); sprintf(line_string, "t2p=%u, value=%u",t2p,value); g_printf(line_string,LCD_L2); output_low(pin_c2); delay_ms(5000); //t2p=255; while (1) { //setup_timer_2(T2_DIV_BY_16,t2p,1); value = t2p>>1; //divide by 2 sprintf(line_string, "t2p=%u, value=%u",t2p,value); g_printf(line_string,LCD_L2); delay_ms(1000); t2p = t2p +1; } } 70

PAGE 71

APPENDIX D RECEIVER INTERFACE TO ATLAS C CODE This is the C code implemented for the receiv er interfaced to Atlas. The code was written using CCS compiler. #include <16F688.h> #device adc=8 #FUSES NOFCMEN //F ail-safe clock monitor disabl ed, switches to internal if external fails, page 33 for more details #FUSES IESO //Internal External Switch Over mode enabled //#FUSES NOIESO //Internal External Switch Over mode disabled, page 33 for more details //#FUSES BROWNOUT //Reset when brownout detected #FUSES NOBROWNOUT //No brownout reset #FUSES NOCPD //No EE protection #FUSES NOPROTECT //Code not protected from reading #FUSES MCLR //Master Clear pin enabled #FUSES NOPUT //No Power Up Timer, page 33 for more details #FUSES NOWDT //No Watch Dog Timer #FUSES HS //High speed Osc (> 4mhz) //#FUSES INTRC_IO //Internal RC Osc, no clkout //#FUSES INTRC //Internal RC Osc //#FUSES LP //Low power osc < 200 khz #use delay(clock=4000000) // 200K osc //#use rs232(baud=19200,parity=N,xmit=PIN_ C4,rcv=PIN_C5,bits=8,DISABLE_INTS) // Baud=1.5% Fosc #use rs232(baud=38400,parity=N,xmit=PIN_ C4,rcv=PIN_C5,bits=8) // Baud=1.5% Fosc // assume Tx_bits= 32bits to be transmitted // time to transmit = Tx_bits 1/Baud = 32bits 1sec/3K bits = 11 msec // Power Consumed = 1.5v 1.5mA = 2.25mW for the transmitter // E = P* time = 2.25mW 11ms sec = 25uJ <<<<<<<<< Transmitter // will need 3 diodes in series, or half voltage divider 3.3v/2 = 1.6v // 1*Tao= 0.63*V=0.63*6.3v=4v // E = 0.5(C)(V)^2 = 0.5(C)(4)^2=25uJ => C=50uJ/16=3uF // E = 0.5(100uF)(4)^2=800uJ // Tao = R*C =10K*100uF = 1 sec to charge cap // P = 200uA*3.3v = 0.66mW (pic power consumed) // E = 0.66mW*11ms = 7uJ <<<<<<<<< PIC energy int rx_char, valid_data,j; unsigned int rx_buffer; char byte1,byte2,byte3,byte4,byte5,byte6,byte7,byte8; 71

PAGE 72

//Baud Rates: // 110, 300, 1.2K, 2.4K, 4.8K, 9.6K, 19.2K 38.4K, 57.6K, 115.2K, 230.4K, 460.8K, 921.6K #int_RDA RDA_isr() //lots of work to be done here !!! { rx_char=getc(); printf("%c",rx_char); } void main() { //char start_byte='A', device_id='B', adc_data='C', crc_data='D', i; //char rx_string; char byte1,byte2,byte3, byte4,byte5,byte6,byte7,byte8; rx_buffer=0; //setup_vref(VREF_LOW|2); //setup_comparator(A1_VR); //setup_adc_ports(sAN0|sAN1|VSS_VDD); setup_adc_ports(NO_ANALOGS|VSS_VDD); setup_adc(ADC_OFF); setup_timer_0(RTCC_INTERNAL|RTCC_DIV_1); setup_timer_1(T1_DISABLED); //setup_timer_2(T2_DISAB LED,0,1); //for PIC16F684 setup_comparator(NC_NC); setup_vref(FALSE); //SET_TRIS_C(0x00); //SET_TRIS_A(TRISA|00000011); printf("hello world"); //enable_interrupts(INT_COMP); enable_interrupts(INT_RDA); enable_interrupts(GLOBAL); //OSCCON = 0x21; //SET_TRIS_C(0x00); 72

PAGE 73

while (1) { //wait for interrupt } } 73

PAGE 74

LIST OF REFERENCES 1. J. A. Paradiso and T. Starner, Energy Scav enging for Mobile and Wireless Electronics, IEEE Pervasive Computing Vol. 4, No. 1, 2005, pp. 18-27. 2. I. Stark, Thermal Energy Harvesting With Thermo Life, Proc. of BSN 2006 IEEE Computer Society Press, New York, 2006, pp. 19-22. 3. V. Raghunathan, A. Kansal, J. Hsu, J. Friedman, and M. B. Srivastava, Design Considerations for Solar Energy Harvesting Wireless Embedded Systems, IEEE IPSN 2005, pp. 457. 4. J. Kymisis, C. Kendall, J. Pa radiso, and N. Gershenfeld, Par asitic Power Harvesting in Shoes, Proc. IEEE International Conference on Wearable Computing 1998, pp. 132 139. 5. S. Roundy, P.K. Wright, and J. Rabaey, Energy Scavenging for Wireless Sensor Networks with Special Focus on Vibrations, Kluwer Academic Publishers, Boston, 2004. 6. S. Roundy, B. Otis, Y.H. Chee, J. Rabae y, and P. Wright, A 1.9GHz RF Transmit Beacon using Environmenta lly Scavenged Energy, Dig. IEEE Int. Symposium on Low Power Elec. and Devices Seoul, Korea, 2003. 7. S. Roundy, P.K. Wright, and J. Rabaey, A Study of Low Level Vibrations as a Power Source for Wireless Sensor Nodes, Computer Communications, vol. 26, no. 11, 2003, pp. 1131. 8. M. Loreto Mateu, Energy Harvesting from Passive Human Power, PHD Thesis, Universidad Politcnica de Catalua, January, 2004. 9. M.B. Schiffer, The Portable Radio in American Life Univ. of Arizona Press, Tucson, 1992. 10. S.A. Schelkunoff and H.T. Friis, Antenna Theory and Practice John Wiley & Sons, New York, 1952. 11. K. Finkenzeller, RFID Handbook 2nd ed., John Wiley & Sons, New York, 2003. 12. T. Milligan, Modern Antenna Design McGraw-Hill Book Company, New York, 1985. 13. J. Heikkinen, P. Salonen and M. Kivikos ki, "Planar Rectenna for 2.45 GHz Wireless Power Transfer," Proc. of IEEE Radio and Wireless Conference 2000, pp. 63-66. 14. K. R. Carver and J. W. Mink, Microstrip antenna technology, IEEE Trans Antennas 1981, pp. 2. 74

PAGE 75

BIOGRAPHICAL SKETCH I received my bachelors degree in electrical engineering from the University of Florida. My special undergraduate interest was radio frequency system level design. I did about two years of research in the ECE department and had two major publications in which I was the first author. I then joined a local company in Gainesville, Florida, in which I worked with high profile companies such as DoCoMo, Intel, and Advan ced Micro Devices (AMD) on a very specialized and detailed analysis, in which mobile handsets power usage is measured and benchmarked, and design refinements and changes are recommended. I was the chief RF engineer who was running and managing all work related to power be nchmarking. At a young age, I was managing two people, and delivering high quality results to customers. I flew to business meetings with Intel, DoCoMo, and AMD, and I met several of their VP s. This line of business generated significant revenues to the company. In the Fall of 2005, I joined the CISE department at the University of Florida to pursue the Master of Science degree in computer engine ering under the supervision of Dr. Abdelsalam (Sumi) Helal. I have known Dr. Helal for more than three years. I am honor to know him and work for him. It has always been a pleasure to work with him on vari ous projects and I am fortunate enough to have learned a lot from him. In the first semester of graduate school, I established myself as a research assistant fo r the pervasive mobile computing laboratory. I designed an RF system that allows sensor ne twork nodes to be power charged wirelessly and from a distance as my thesis, which is curre ntly prototyping the Pervasive and Mobile Computing Laboratory, and has been disclosed to the University of Florida, which has accepted the disclosure and filed for a patent applic ation. The docket number of this patent is PCT02/22/06. Currently, I am applying wireless pow er transfer to the ATLAS Sensor Platform, which is being commercialized by Pervasa, Inc, a University of Florida startup. 75


Permanent Link: http://ufdc.ufl.edu/UFE0017962/00001

Material Information

Title: Self-Powered Wireless Sensors: Remotely Powered Wireless Sensors Using Wireless Power Transfer
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0017962:00001

Permanent Link: http://ufdc.ufl.edu/UFE0017962/00001

Material Information

Title: Self-Powered Wireless Sensors: Remotely Powered Wireless Sensors Using Wireless Power Transfer
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0017962:00001


This item has the following downloads:


Full Text





SELF POWERED WIRELESS SENSORS:
REMOTELY POWERED WIRELESS SENSORS USINTG WIRELESS POWER TRANSFER




















By

AHMAD HASSAN EL KOUCHE


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2006

































Copyright 2006

by

Ahmad Hassan El Kouche




































To my parents, brother, and two sisters.









ACKNOWLEDGMENTS

Foremost, I would like to express my gratitude to my research advisor Professor

Abdelsalam (Sumi) Helal for all his support, in the many forms: his confidence in my research

work which encouraged me to work harder, his ambition and optimism that taught me to see

obstacles in research as a start of a new solution and not a barrier, and his ability to approach me

as a friend made my research truly enj oyable. I have never really felt so comfortable working

with a professor as much as I did with Dr. Helal. His ability to switch hats from a research

professor to a boss and then to a friend is absolutely amazing. I have known Dr. Helal for almost

four years now, in which I witnessed his first born child Shadia. Without my advisor's help and

support, this research material would not have been possible. Also, I would like thank Dr. Janise

McNair and Dr. Shigang Chen for agreeing to serve on my committee.

I would like to thank my parents and their late night calls making sure I was eating

healthy, when I was not, making sure I was getting enough fruit juice, when all I was actually

drinking was black coffee, but most importantly I would like to thank them for providing me

with strength at times when I was feeling tired. Also, I would like to thank my brother and two

sisters for their endless love and patience.

Also, I would like to thank all my friends in the Pervasive Computing Laboratory (Dr.

Lim, Dr. Hisham, Dr. Bassam, Youssef, Jeff, Raja, Hen-I, and Steven).












TABLE OF CONTENTS


page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ............_...... ._ ...............7....


LIST OF FIGURES .............. ...............8.....


AB S TRAC T ............._. .......... ..............._ 10...


CHAPTER


1 INTRODUCTION ................. ...............11.......... ......


Problem Statement ................. ...............11.................
Power Harvesting Sources ................. ...............14........... ....
A History of Wireless Power Transfer ................. ......... ...............17. ..


2 DESIGN PLANNING AND CONSIDERATIONS ................ ...............................21


Friis Free Space Loss ................. ...............21................
Antenna Types ................. ...............26.................
Printed Patch Antenna ................ ...............29........... ....
Antenna Simulation Results .............. ...............34....
Power Delivery .............. ...............43....
Attenuation .............. ...............45....
Feasibility .............. ...............46....


3 WIRELES S POWER TRANSFER SYSTEM DESIGN ......____ ...... .. ...............48


System Level Design ............. ...... ...............48...
System Design Constraints ............_...... ...............49...
Range Constraints............... ..............4
Cost Constraints............... ..............5
Radio Frequency Generator ............. ...... .__ ...............50..
Radio Frequency Signal Source .............. ...............53....
High Power Signal Amplifier ............. ...... .__ ...............53..
Antenna of Choice ............_...... ...............54...
RF to DC Wireless Sensor Node .............. ...............55....
Antenna M watching ................ ...............56...
RF to DC Matching Network .............. ...............56....
Low Power BaseBand Processor ............. ...... ._ ...............58...
Low Power Sensors ................. ...............59...
Low Power Wireless Transmitter ............. ...... ._ ...............59...











APPENDIX

A ELECTRONIC SCHEMATIC OF RF BASEBAD CONTROLLER .........._... ................60


B REMOTE POWERED WIRELES S SENSOR C CODE ......____ ........__ ................61


C RF BASEBAD CONTROLLER C CODE ....._......__. ..........._ ...........6


D RECEIVER INTERFACE TO ATLAS C CODE ................. ...............71........... ..

LIST OF REFERENCES ................. ...............74........... ....

BIOGRAPHICAL SKETCH ............. ..............75.....










LIST OF TABLES


Table


page


2-1 Available ISM bands for wireless power transfer and applications .............. .................26










LIST OF FIGURES


Figure page

1-1 Moore' s law showing growth of transistor count for the Intel processor (dots) every
24 months. Figure taken from Wikipedia.org ................. ...............13...............

1-2 Power harvesting sources............... ...............17

1-3 Nicolas Tesla and the AC motor. ........... ......__ ...............18

1-4 Tesla, master of lightning. ........... ..... ._ .....__....._ ........._......20

2-1 Basic system setup for wireless power transfer showing the antennas on both the
transmitter and receiver, separated by radius distance R. ................. .................2

2-2 Received power (dBm) at the receiver vs. distance (meter) away from receiver, such
that Pt =0dBm, Pr=0dBm, Gt=0dB, and Gr=0dB ......_.__._ ........... .. ...._._.........2

2-3 Medium power amplifiers. ........._.__...... .___ ...............24.....

2-4 Electromagnetic frequency spectrum and associated wavelengths. Borrowed from
"The RF and Microwave Handbook" ............. ...............26.....

2-5 Monopole whip antenna on top of a ground plane .............. ...............27....

2-6 Dipole antenna .............. ...............28....

2-7 Helical antennas. .............. ...............29....

2-8 Patch antenna. ............. ...............30.....

2-9 Effective length of patch after fringing effect Vs. ratio of width to length k. .................. .33

2-10 Effective length of patch after fringing effect Vs. ratio of width to length k. .................. .34

2-11 Patch antenna simulation using PCAADC. ............. ...............35.....

2-12 Square patch antenna using Ansoft Designer, with given dimensions, and output
matched to standard 5002 using stub stripline of length 12mm by 29mm. ........................36

2-13 Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz......37

2-14 Antenna gain of patch antenna (dB) in terms of the swept frequency F(GHz). ................3 8

2-15 Fabricated 2.45GHz antenna using homebrewed etching technique ................ ...............39

2-16 Square patch antenna using Ansoft Designer matched to 5002 at 2.45GHz. .....................39











2-17 Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz......40

2-18 Antenna gain of patch antenna ................. ...............41...............

2-19 Patch antenna for 2.45GHz ISM band, with 3.5dBi gain. Fabricated at PCBexpress_......42

2-20 Patch antenna array, matched at 2.45GHz, with gain equal to 6dBi. ............... ...............42

2-21 Improved wireless power received with additional 26.5dB gain ................. ................ .43

2-22 High gain commercially available antennas suitable for remote power transfer...............44

2-23 Commercial chip antenna. ............. ...............44.....

3-1 Top level system design ................. ...............48........... ...

3-2 Back-side view of RF generator. ............. ...............51.....

3-3 Front-side view of RF generator. .............. ...............52....

3-4 Side view of RF generator, displaying tuning knobs. .............. ...............53....

3-5 Signal generator local oscillator from Mini Circuits. ............. ...............53.....

3-6 RF amplifier ZRL-3 500 from MiniCircuits ................. ...............54........... ..

3-7 Commercial patch antenna ................. ...............55................

3-8 RF to DC wireless sensor node system design. ............. ...............55.....

3-9 RF to DC matching network. .............. ...............57....

3-10 RF to DC output voltage about 1.2 volts. ............. ...............57.....

3-11 RF to DC output current about 1.2 mA. .............. ...............58.___ .









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Master of Science

SELF POWERED WIRELESS SENSORS:
REMOTELY POWERED WIRELESS SENSORS USINTG WIRELESS POWER TRANSFER

By

Ahmad Hassan El Kouche

December 2006

Chair: Abdelsalam (Sumi) Helal
Major: Computer Engineering

Wireless sensor networks have recently started showing more interest towards issues of

power consumption due to the high dependency on battery power. When the battery energy

capacity is diminished to the point of inadequate voltage level, the wireless node would shut

down, which might cause loss of information, resource limitation, intelligence, and perhaps

cause network failure in some cases. Therefore, it is necessary to have a backup secondary power

source that would deliver energy to the wireless node in case of a power scarcity. In the case of

wireless nodes, backup power would have to be delivered, obviously, without using any wires,

which means that, power would have to be delivered in the form of thermoelectric, solar power,

mechanical, vibration of piezoelectric material, or radio energy propagation. The latter is referred

to as wireless power transfer (WPT), which is the main concentration and focus of alternative

power source of this thesis, is discussed in detail. The concept of WPT is as simple as generating

a high power radio signal and beaming it towards the wireless sensor. The wireless sensor has

the ability to detect this radio signal and turns it into usable DC voltage for storage.









CHAPTER 1
INTTRODUCTION

Problem Statement

As technology advances, we notice that electronic devices have the natural ability to

eventually shrink in physical size in time. This can be obviously seen from the evolution of

computers from mechanical computers, to transistor IC computers, and now to ultra compact

embedded sensor platforms, and wearable computers. The size and performance of integrated

circuits ICs was predicted by Intel's co-founder Gordon Moore in 1965 that the number of

transistors in a microchip would double every 18 months, which was then adopted as Moore's

law. Generally speaking, if the number of transistors doubles every two years, then the power

consumption would also double according to the following equation for calculating rough

estimate power consumption:

Power Consumption = V/2 C V2 41

Where C is the transistor equivalent capacitance, V is the transistor voltage supply, and f is

the transistor switching frequency. Therefore, the amount of power consumed by a sensor node is

proportional to the amount of power a single active transistor consumes scaled by the number of

active transistors on the sensor node. This tells us that the simpler the architecture deign is the

less power is consumed. This of course has the draw back with the amount of intelligence and

limitation a sensor node would have at that point, so there is a tradeoff with sensor smartness and

power consumption. This is an obvious argument since nothing is free, as system complexity

grows, more transistors are used to get the job done, and thus more power consumption is used.

The argument here is that the simpler you keep the design the more likely you would consume

less power at the cost of sensor intelligence limitation. Another argument is how fast the

transistor switches on and off, which is the effect of the variable f in the above equation, also









known as the switching frequency or system clock. The switching frequency of the transistor is a

tricky variable since one would basically think of reducing the frequency to consume less power,

however, there is more to it than just operating at slower speeds. On the hardware level of things,

using a high crystal clock requires more current to drive than a lower frequency crystal. For

example, a microprocessor from Microchip PIC16F688 running on a 4 MHz crystal and voltage

supply 2 volts would consume about 220uA vs. a 32 KHz crystal that would only consume 9uA

running under the same conditions. However, the problem is that a 32 KHz crystal would take

128 times the time to complete the same job. With wireless sensor nodes the highest power

consumption block is the radio frequency (RF) transmitter, which consumes constant power as

long as there are information bits to be sent. Therefore, if information bits are transmitted faster,

then the RF stage would be on for a shorter period of time, and thus, system power consumption

would be considerably lower. This method was successfully implemented and patented by

Nordic VLSI company as the ShockBurstTM effect, such that information data is clocked in at a

slower rate (32 KHz), but transmitted at 1 or 2 Mbps. Nordic ShockBurstTM radio architecture

became popular in low power wireless devices such as keyboards, mice, computer presentation

controllers, active RFIDs, heart monitoring watches, and others. This clearly shows that a slower

switching frequency is not always a lower power consumption approach, but that it really

depends on the rest of system design and how it is affected by the switching frequency.

Moore' s law states that the number of transistors doubles every 18 months (SEMATECH

research shows a 24 months roadmap), as seen in figure 1-1. This is actually because the

transistor size (gate-length barrier) is reduced to half, and thus you are able to fit twice the

number of transistors on the same silicon die area, which in effect reduces cost.












Moore's Law


Number of transistors doubling Bvery 111 months.

1,000,000.000-
Iranium 2







100,00.0001 Nme fnnltr -o~i en 2 4
'e --'e
10,000 -




197 1BD 99 200 00
Year

Figure0000 1-1. Mooe' law shwn rwho rniso on o h nelpoesr(os vr
24 moths.Figur takn frm Wikpedi0org

Thsshw asmpeobevainthtpoe cnumtonsaeslnery ih h aon





Fiue11 or' a hwn rwhof transistor doublin and the speed inwhchth tanisor switche stte. ls, ige





switching frequency requires higher bias voltages, which in return affects power on a quadratic


scale. In order for a wireless sensor node to keep up with Moors law power consumption


requirements, the battery industry would also have to double their battery capacity every two


years, or put twice the amount of batteries on each sensor node. The later approach is highly


unsuitable for wireless sensor network due to cost, size, and weight constraints. Also, some


nodes have to be deployed in isolated environments, which require long operating life and years


of maintenance-free operation. Thus, the need to reduce power consumption and the need for


alternative power sources becomes a real concern for wireless sensor network.1,23.4









To solve the ever increasing demand on power, wireless sensor nodes would require

deployment of low power hardware design, software selectable low power operation modes,

short active duty cycles, and harvesting energy from different external sources. These constraints

in wireless sensor network gave lead to the ZigBee technology with it's comparable low power

consumption in full active mode of about 50 mA (Transmitter = 25 mA average, Receiver = 20

mA, and DSP = 5 mA) from CirronetC when compared to a WiFi node that could consume up

to a total of 700mA of active average current from DPACC and Blue Tooth with an average of

150mA from BlueRadiosC. The advantage of ZigBee over Bluetooth includes lower power

consumption, higher radius of operating range, and larger number of supported nodes.

Power Harvesting Sources

Another solution for the high demand on power consumption is to harvest energy from the

surrounding environment.' Clean energy sources can be found around wireless sensor nodes and

harvested into usable energy to recharge batteries or to charge large local capacitors, also known

as super or ultra capacitors. Ultra capacitors have midrange capacity storage between

rechargeable batteries and regular capacitors. However, unlike rechargeable batteries, ultra

capacitors do not require a special way to be charged, as they would accept any kind of voltage

level or waveform, that being AC or DC. This sets fewer constraints on the method of recharging

super capacitors from different power source.

Available sources for power harvesting include the use of solar panels seen in figure 1-2 A.

Solar panels are found everywhere and are easy to harvest energy from, such that as the light

intensity increases, the output produced voltage is a DC voltage easily charged on a super

capacitor. Solar panels exhibit conversion efficiencies between 10% and 20% in direct sun

light[5]. For example, HARP's NE-Q5E2E photovoltaic module exhibits a 16.4% conversion

efficiency under direct sun light. Consequently, solar panel do not work very well indoors, in









which they exhibit only 1% efficiency, and are useless for harvesting energy for a sensor

network constantly operating in a dark environment.

Piezoelectric material has gained some recent popularity in power harvesting for wireless

sensor network.6 As a matter of fact piezoelectric material goes back to 1880 when two brothers

Pierre and Jacques Curie demonstrated that some crystals (tourmaline, quartz, topaz) generate

electrical polarization from mechanical stress. This effect was then named piezoelectricityy" after

the Greek piezein, which means to squeeze or press.7 Piezoelectric material really gained large

popularity when it was used to maintain accurate timing in clocks, wrist watches, and crystal

(XTAL) oscillators used in electronic systems today. However, the problem with using

piezoelectric material for power harvesting is that you need a constant source of vibration or

mechanical taps, stress, push, or pull to keep generating power, which would be a perfect

application to deploy a sensor network around vibrating motors or engines in an industrial

environment. Seen in figure 1-2 B is a flexible piezoelectric sheet made by Measurement

SpecialtiesC company that produces very high open-circuit voltage close to a hundred volts

when physically struck hard. However, when a low impedance load is used the output voltage

becomes extremely low depending on the actually load value.

Mechanical power can also be turned into electrical energy as seen in figure 1-2 C using

windmills, and today using MEMS figure 1-2 D. A recent popular device in the market is the self

powered flashlight uses a combination of mechanical shaking and electromagnetic induction of

current. By shaking the flashlight, a magnet slides between the north and south pole of the

winded coil inducing current in the copper wire to be stored on a super capacitor, which is then

used to power an ultra-bright LED when the on button switch is closed. The same exact device

can be use to power a wireless sensor node, however running around to shake every single node









becomes an impractical method, unless you are planning to use this system to work out, and I

would guarantee you by the time you run to the second node, the message would already be lost.

Another very popular device that uses mechanical energy to recharge a battery is the famous

$100 laptop from MIT, which uses a hand crank to recharge the system's battery, low power

processor, and low power software selectable user modes. Future updates mention that the hand

crank would be replaced with a power supply that uses a foot crank instead, which reduces

damage and stress caused to the laptop's chassis.

Thermal energy can also be transformed into electrical energy by using thermoelectric

generators as seen in Eigure 1-2 E. Thermoelectric generators are composed of a thermocouple

comprising a p-type and n-type semiconductor connected electrically in series and thermally in

parallel. The thermoelectric generator produces an electrical current proportional to the

temperature gradient between the hot and cold junctions. An electric load is connected

electrically in series with the thermoelectric generator creating an electric circuit, such that the

temperature grading between the two electrodes can be harvested into a super capacitor.8

Wireless power transfer is gaining significant popularity and is heavily used in

technologies like RFIDs seen in Eigure 1-2 F, where batteries become essentially nonexistent.

Such technology eliminates the bottle neck with power availability from battery capacity and the

ever increasing demand on available energy capacity. This technology essentially creates a

transparent radio channel that either constantly feeds power to the sensor node or charges a local

capacitor over a constant duty cycle. Wireless power transfer requires two separate systems. The

first system is the RF generator that generates the high power wireless energy and radiates it

towards the wireless sensor node vicinity, and the second system, which is the wireless sensor










node, receives the wireless power rectifies it and turns it into usable DC power to be stored for

future use.
























Figure 1-2. Power harvesting sources. A) Solar Panel. B) Piezoelectric sheets. C) Windmill
D) MEMS E) Thermoelectric generators. F) RFID

A History of Wireless Power Transfer

Wireless power transfer was first explored by the famous scientist Nicolas Tesla, seen in

Eigure 1-3 A. Nicolas Tesla was the first to invent the AC inductive motor in the late 1870s.

Edison hired Tesla to work for him and proposed to pay him $50,000 if he was to succeed in

creating AC power, mainly because Edison thought it was too difficult. Tesla took this as an easy

job and came up with the AC power using his inductive AC motor, seen in Eigure 1-3 B. Edison

was perhaps shocked when he found out that Tesla had succeeded and refused to pay him saying

that the offer had been made in j est. Tesla immediate resigned, and challenged one of the greatest

scientists Thomas Edison with his AC power invention. Tesla took his AC power invention and

considered it to be far more superior in nature than Edison's DC power station.























A B,


Figure 1-3. Nicolas Tesla and the AC motor. A) Scientist Nicolas Tesla B) First AC inductive
motor

Tesla and Edison stood in competition to harness the power of the Niagara Falls. With

Tesla' s Chicago exposition "City of Light" in 1890, where he proudly showed off his AC power

station lighting hundreds of thousands of incandescent lamps in the city, and with the 27 million

people who attended the fair, it was dramatically clear that the power of the future was the AC

power. Tesla' s proposal to harness the power of the Niagara Falls was his childhood dream, and

it was the only proposal to be accepted, and came into existence in 1893. Even Thomas Edison

was finally convinced that AC power was a must and essentially powered all his labs with it.

Another one of Tesla' s invention was the so called "Tesla coil," where he would take his regular

AC power and by using transformers, he was able to boost the voltage, while dropping current to

maintain conservation of power, up to couple hundred of thousands of volts which created a high

electrical field between the negative end terminal of the transformer and earth ground, which

essentially caused sparks to jump off, or channel through. He also noticed that the sparks where

creating a high magnetic field that was illuminating incandescent lamps in the vicinity. In reality,

this is the effect of electrons releasing photons off the surface of the copper wires of the









transformer and creating radio waves that would propagate around the transformer, such that the

high power radio waves would light up the incandescent lamp in the vicinity. This discovery

essentially consumed Tesla with excitement, and perhaps drove him crazy with the endless

possibilities he could make of it. He immediately proposed his theory that power can be

channeled through air to light up incandescent lamps and essentially an entire city, and was

actually successful at demonstrating it, seen in Eigure 1-4 A. To prove his power transfer theory,

he built the highest Tesla coil that stood as a tower in Pikes Peak looking over Colorado Springs

as seen in Eigure 1-4 B. At the point of completion and testing, Tesla turned on the giant Tesla

coil and electricity arced out hundreds of feet over the tower causing a complete burn out of all

incandescent lights in the city and the Tesla tower had set itself on fire. Thus, causing a complete

blackout of the entire city, and was then charged for the damages caused, which led to his

bankruptcy. Tesla even proposed to transmit information wirelessly around the globe, seen in

Eigure 1-4 C, such that people would be able to share information such as messages, pictures,

audio, and secure military communication, and was the first to propose the concept of the radar.

He was also the first to create the first remote controlled robot boat in 1898, as seen in Eigure 1-4

D. However people thought that all his proposals where too dreamful and impractical which is

understandable since he lived in a time period when people where astonished to even witness

Edison' s creation of the light bulb. However, it was not just one lamp he wanted to light up; it

was millions of lamps and essentially the entire city. Tesla, with his over 700 patents, is now

known as the scientist born out of his time.9 Indeed, here I am now, a 100 hears later

demonstrating what Tesla had wanted to do. Tesla coil effect can be seen today in devices sold as

taser-guns for self defense purposes. Also, ionizers use the same concept Tesla proposed to

ionize the air which gives us the feeling of fresh air. However, Tesla wanted to use this

















~_


electricity to create a channel to transfer the power to other locations without using expensive

copper wires. If Nicolas Tesla was to say three last words before he died he would have said

"Wireless Power Transfer."


Figure 1-4. Tesla, master of lightning. A) Tesla demonstrating wireless power transfer B) Tesla
attempts to build a wireless power supply tower. C) Tesla plan to transfer wirelessly
power the world and transfer wireless information around the globe. D) First remote
controlled robot built by Tesla.









CHAPTER 2
DESIGN PLANNING AND CONSIDERATIONS

Friis Free Space Loss

In order to achieve wireless power transfer at high frequencies, we would need to closely

study the derived equation of Harald T. Friis, know as the Friis free space loss equation, or just

Friis equation.10 Friis explains through his derived equation what effects the radio signal exhibits

while propagating through the air. However, before we look at the Friis equation, we will set up

the typical scenario of the system from the transmitter (also know as the generator) to the

receiver end (also known as the detector, or RF to DC sensor), as seen in figure 2-1.


SGT GR

Transmitter PT PR Receiver
(RF Generator) I 1 (RF to DC sensor)



Figure 2-1. Basic system setup for wireless power transfer showing the antennas on both the
transmitter and receiver, separated by radius distance R.

Shown below is the Friis equation describing RF signal strength received as a function of

distance R, transmitted power Pt, and antenna gains Gt and Gr.


r:= 4xR3 -rP (2-1)


Such that, Pr is the power received, lambda is the wavelength of the radio signal, Pt is the

power transmitted by the RF generator, Gt is the directivity gain of the antenna attached to the

RF generator, and Gr is the directivity gain of the antenna attached to the RF to DC sensor. The

receiver is also known as "RF to DC sensor", which is a description chosen by the author simply

because the receiver could also be used as an RF radiation sensor for a particular band depending









on the frequency of choice. By taking the 10Log( ) of the Friis Equation, we transform the power

from units of mWatts to dBm, and thus we obtain the equation below.


Pr:= Pt + 20Lo c 20Log(2x-R) + 4 + Gr


In equation (3) we have replaced lambda with its equivalent value (c / f), such that c is the

speed of light (299,792,458 meters / second), and f is the frequency of the RF signal in Hz. Also,

Pr and Pt are described in units of dBm, Gt and Gr both have units of dB, and R has units of

meters. Next, we plot the effect of frequency f on the power received Pr in terms of the distance

between transmitter and receiver, such that Pt =0dBm, Pr=0dBm, Gt=0dB, and Gr=0dB as seen in

figure 2-2. We notice from Figure 2-2 that power attenuation is less for lower frequencies. For

example, at one meter and 2.45 GHz signal, the power received is about -40dBm from

transmitter. While at 4341VHz, the power received is about -25dBm. This means that at higher

frequencies the signal suffers from more atmospheric absorption than lower frequencies. This

effect is shown in figure 2-3. The question is now which frequency should be considered for

wireless power transfer and at what cost? This question could be answered in many ways

depending on the particular purpose of the application. Also one should consider all the

complications and effects the frequency has on the system design. For example, the antenna size

is characterized by the wavelength which is indirectly proportional to frequency. Therefore, as

frequency decreases the antenna dimensions increase. This might be problem for self powered

wireless sensors that have size constraints to meet, which would set a limit on the lowest choice

of frequency. Therefore, due to size constraints, the author has chosen a higher frequency to

enable wireless power transfer at 2.45GHz. Another advantage with operating at 2.45GHz is the

high transmit power allowed at that frequency by the FCC. In addition, due to the popular










2.4GHz ISM band, components are readily available at relatively cheap prices.

Power Re~ceived (dBm) Vs5. Distance (m)


-2o

P R,2.45.1









1 010 101-0 1
-2R
Ditne ewe T n x mtr

F ,igre~S1 9 2-.Rcie oe dm tth eevrv.dsac mte)aa rmrciesc
th tatP=0 mP=dmG=0BanG=0 .

The,, mai fou fwrls owrtase st nces h eevd oe a uha
posbe sc ht n oldb betooeae tfrhe itnes oknga qato 21

we notic thtw a lyaon ih u bro aibe ht ol lo st civ
maxmu poe rnfeuigth olwngmtos

nrasn hetasmte owrwul eanovou ouio oinrain h rcie
poer The,,~ cocp ssml ee temr oe o rnsitemr oe o
receive

* Bycooig nena it ige drcivt gi o oh h tasmte adth ecie
wol nres h dlvr f iees oe o h eeve yman fcncnrtn
th rdo ea n nedretin










* Another obvious solution for increasing received power would be to decrease the operating
distance R, in other words, operate at a closer distance to the transmitter.

* By decreasing the signal frequency we would also be able to increase the received power at
the receiver.

There are a number of complications for each of the above solutions. First, increasing the

transmitted power means that the generator would have to use a higher power amplifier. High

power amplifiers become more expensive and mostly sold to defense contractors. Also, high

power amplifiers consume significant amount of power and usually have low efficiency 40% to

60%, which means that the rest of the power is transformed into heat. Therefore, heat-sinks and

temperature sensors become a must in order not to destroy the expensive amplifier. The author

was able to purchase a relatively cheap amplifier ZRL-3 500 from minicircuits.com with

maximum output power of +21dBm, which equivalent to an output power of 125mW. The ZRL-

3500 is considered to be a relatively low power amplifier for the purpose wireless power transfer.

However, ZRL-3 500 is priced at $13 5 as compared to the ZVE-8G $1,095 with an output power

of 1Watt, seen in figure 2-3.










A + IVE B

Figure 2-3. Medium power amplifiers. A) ZRL-3 500 with output power 125mW B) ZVE-8G
with output power 1Watt.

Second, the antenna of choice is usually dependant on the type of the application such that

the antenna size could be a constraint. If the antenna size is not an issue, then one could design

an antenna or choose to deploy antenna arrays with higher directivity gain at the cost of size and










narrower radiation beam. Many other issues are involved with the choice of antennas, which is

why there is a separate section that discusses the choice of antenna in more details.

Third, by brining the receiver closer to the transmitter, the signal received would be of

higher power, which is a trivial solution. However, one might require the receiver to be at a

minimum distance away from the transmitter for a particular application to be feasible. On the

other hand, some other applications, requiring wireless power transfer, might not have a

constraint on a set distance between transmitter and receiver, such as recharging a device that is

completely submerged in water. One would remove the device from under water to recharge it

without disassembling it, which might cause wear and tear and eventually might start leaking

water. Therefore, distance between the device and transmitter could be as close as possible.

Finally, by decreasing the operating frequency one would be able to transmit power at

further distances. However, there are many issues with choosing the right frequency for the right

application, and in effect the frequency of choice would affect all other variables. For this reason,

one would have to carefully choose the operating frequency of the system. In addition, many

factors could affect the choice of frequency such as the Federal Communications Commission

(FCC) regulation, Industrial Scientific and Medical (ISM) available bands, antenna size, and

available resources. FCC has a set of available license free bands called the ISM bands seen in

table 2-1 located at different parts of the frequency spectrum seen in figure 2-4. ISM bands are

regulated by the FCC with allowed effective radiated power ERP, effective radiated isotropic

power EIRP, active time period in transmit mode, interference, band width, and more. Table 2-1

shows the popular ISM bands available with the allowed output power transmission. It is

important to note the wavelength of the available ISM band because antenna dimensions are

dependent on it.















































1 10-1 10-2 10-3 310-5 10-6 410-7 10-8


<10-16


Wavelength (m)


Figure 2-4. Electromagnetic frequency spectrum and associated wavelengths. Borrowed from
"The RF and Microwave Handbook"

Antenna Types

Antennas are made in many different shapes, sizes, lengths, power ratings, and for

different purposes. However, antennas are mainly chosen to fit the application rather than the

other way around. Therefore, usually antennas are specially designed or specially ordered to fit

the purpose of the application. There are, however, many commercially available antennas one

can choose from, and due to the high availability of antennas, one could easily shop around for


mirwvs x
audio frqunce RF MF aio ilmt
VH teeiso su mlimtr ave
I I II" I I EE


Table 2-1. Available ISM bands for wireless power transfer and applications


Frequency Band


Allowed output power, application


<125 KHz
1.95, 3.25, 8.2 MHz
13.56 MHz
27 MHz
138MHz
402 -405
433.05 -434.79
869.4 869.65 MHz
902 928 MHZ
2400 2483.5 MHz
5725 5875 MHz
24.00 24.25 GHz
61.0 61.5 GHz
122 123 GHz
244 246 GHz


Near Field Inductive RFID
Near Field Inductive theft tags
Near Field Inductive RFID, RC Toys
0. 1 Watt ERP
0.05 Watt ERP, Duty Cycle <1%
Medical Implants, 25uW ERP
25mWatt ERP, Duty Cycle <10%, RF controllers, and keyless entry
0.5Watt ERP, Duty Cycle <10%, RF controllers, and keyless entry
4W EIRP 802. 15.4 Zigbee, wireless modems, keyless entry, WPT
4W EIRP, 802. 11b,g 802.15.4 Zigbee, WPT
25mWatt EIRP 802.11a
0. 1Watt EIRP, police radars
0. 1Watt EIRP
0. 1Watt EIRP
0. 1Watt EIRP


Frequency (Hz)


3-101


I


3-103


3-105


3-107


3-109


3 1011


3- 1014


3-10n16


>3-1024


107 106 105 104 103 102 10









an antenna that would work for the application, or at least close enough. However, in a proj ect

such as wireless power transfer, we notice that the type of antenna is, indeed, an important factor

of making power transfer feasible, which will be explained in later sections. Common popular

antennas have relatively known attributes and aspects and are discussed below.

Popular antennas include the monopole "whip" antenna,"l which has omni-directional

radiation pattern, directivity gain typically about 3dBi, length equal to '/ wave length on top of a

ground plane, and 10% band width, as seen in Figure 2-5. They are cheap to produce, and get the

job done fairly well in communication systems. Monopole antennas might work well for wireless

power transfer applications requiring even power distribution in all directions. However, the low

directivity gain and omni-directional radiation pattern might also consider this type of antenna

unattractive for an application requiring high concentrated power that is focused in one direction

towards an area of wireless sensor nodes. For a good performance monopole whip antennas, one

could use a solid tinned copper wire anywhere from 14 to 18 AWG for good conductivity and

low thermal loss.








Ground



Figure 2-5. Monopole whip antenna on top of a ground plane

The dipole antenna is also another popular type of antenna with very similar properties to

that of a monopole. The difference between a dipole and a monopole is that a dipole has twice

the length '/ h of that of a monopole. Also, another main difference is that a dipole can be fed

differentially. A dipole can be turned into a monopole by replacing one of the dipoles into a









ground place, essentially what is seen in Figure 2-5. Dipoles are just as attractive as monopoles

for wireless power transfer, and application dependent. A good point to make about both

monopoles and dipoles is that they are greatly affected by ground planes, and the only ground

plane in the vicinity should be constructed as seen in Figure 2-5. Mounting a monopole or dipole

around obj ects and metals greatly changes the characteristics of the antenna, such as impedance,

radiation patter, and bandwidth. A dipole antenna is seen in Figure 2-6.



11 I



Figure 2-6. Dipole antenna on

Another similar antenna to the dipole and monopole is the helical antenna. A helical

antenna operates in two complete different modes. The first mode is the normal mode, seen in

Figure 2-7A, which is also know as the broadside helical antenna. The naming broadside comes

from the omni-directional radiation pattern of the farfield. In normal mode, the main advantage is

the short length of the antenna such that the windings of the antenna are shorter than the

wavelength. In normal mode, the antenna acts as short monopole with omni-directional radiation

pattern and poor gain. However, the small factor of a helical antenna makes it attractive to

mobile devices such as mobile phones. For wireless power transfer, a normal mode helical

antenna would not be good choice for RF generator due to its low gain. However, on the receiver

side, the RF to DC sensor could be a good possible choice if the remote sensor is restraint to size.

The second mode of operation is called the axial mode, which is also known as end-fire mode, as

seen in Figure 2-7 B. The endfire mode naming comes from the high directivity gain of the

antenna and concentration of radiation at the end of the helical structure. The biggest advantage

of the endfire mode is the high high gain >10dB and its simplicity. The disadvantage is the larger










dimensions. This form of antenna is found in satellite communication and long distance

communication. As discussed in the previous section, higher directivity gain antennas can deliver

more wireless power and increase operating distances. Therefore, axial mode antennas can be

very useful and attractive for wireless power transfer applications especially on the generator

side. However, an axial mode helical antenna might not be very practical for operation on the

wireless sensor node due to size constraints. Therefore one could choose two different antennas

for operation the large axial mode on the generator, and the compact normal mode on the

wireless node.








AB

Figure 2-7. Helical antennas. A) Normal mode helical antenna B) Axial mode helical antenna

The Yaggi antenna is a worthy antenna to mention due to its high directivity gain >10dB. It

was first discovered by a Japanese professor during the world war two, and was named after him.

Yaggi antennas are also very attractive for wireless power transfer for the same reason described

in the axial mode helical antenna.

There are so many different types of antennas one can choose from, however, the author is

mainly interested in printed patch antennas, which will be described in details in the next section.

Printed Patch Antenna

The patch antenna was chosen for this proj ect due to the many advantages patch antennas

have over others. Listed below are the basic advantages of using a patch antenna.

*Cheap fabrication costs over other antennas, such that it does not require any components
other than etching a copper trace on a printed circuit board (PCB).









* Patch antennas have flat surface which gives them a low profie compared to other
antennas like a monopole.

* Due to their flat surface, patch antennas are less susceptible to break or wear and tear.

* Patch antennas can be mounted on top of metals without affecting antenna impedance.

* Readily available tools can simulate a patch antenna very easily.

* Patch antennas can be combined in arrays to increase directivity gain.

* Patch antennas have narrow bandwidth which acts as a natural band pass filter.

The simplest form of a patch antenna is composed of three levels, seen in Eigure 2-8 A. The

top layer is the copper patch of length L and width W. The resonant frequency is affected by the

length L, such that L is approximately equal to V/2 h before fringing. Therefore, the wavelength is

initially equally to twice the length before fringing. The middle layer is the dielectric material,

which is the fiber glass material known as FR4 with relative dielectric constant (Er) ranging from

4.2 to 4.7 and thickness h, as seen in figure 2.8 B. The third layer is the bottom ground layer. The

simulation software used in this proj ect to model the patch antenna is Ansoft DesigneerC, which

uses dielectric material average for FR4 equal to 4.4 value.







hA 4r

Ground A L L

Figure 2-8. Patch antenna. A) Square patch antenna on a printed circuit board B) Cross section of
board showing fringing effect.

Patch antennas have the same characteristics as a transmission line, and are modeled the

same way as a transmission line. Radiation occurs from the fringing Hields, which extend the

actually dimensions of the patch by AL,12 as seen in figure 2-8 B. FR4 material was chosen for









this proj ect due to its low cost and availability. Other dielectric materials can be used for patch

antennas that would actually enhance many features of the antenna.13,14 However, it would cost

significantly more, which is inconvenient for a large number of self powered wireless sensor

nodes.

Mathematical Description

This section describes the mathematical model of a transmission line used as a patch

antenna. A patch antenna is nothing but a transmission line modeled to resonate at a high

frequency. Since this patch antenna will be used for self powered wireless sensors, the

dimensions will have to be fairly small to fit in as many situations as possible, which is done one

way by increasing frequency. However, by increasing frequency, the signal suffers more

attenuation. Therefore, the author chose to operate at the frequency 2.45GHz, which is a popular

ISM band used for WLAN, thus, components are more readily available than other frequencies.



*Fr (2-2)

In equation (2-2), h is the wave length in the dielectric material, f is the frequency

(2.45GHz), c is the speed of light (299,792,458 meters / second), and er is the dielectric constant

of the FR4 material (4.4). From equation (2-2), we get a rough estimate of the wavelength

h=5.834 cm, such that V/2 h iS equal to L which is equal to 2.917cm. However, we need a better

estimate than that.

w(k) := k-L (2-3)

In equation (2-3) w is the width of the patch, L is the length of the patch before fringing,

and k is a constant coefficient representing the ratio of the width to length, such that k is equal to

one for a square patch. In addition, the notation w(k) represents the width of the antenna as a

function of k.









w(k)
u(k) :=
h (2-4)

In equation (2-4), u is the ratio of width w to the height h (1.5mm) of the dielectric

material. For a square patch, u(k) is equal to 19.455.

er +1 er- 112 \-0.5
2 2 u(k)j (2-5)

In equation (2-5), re, is the effective relative dielectric constant. For a square patch, re, is

equal to 4.037. Notice that re, is slightly lower than Er because the fringing fields around the

patch are not completely confined in the dielectric material.

Ere(k) + 0.3 u(k) + 0.264\
aLk): 042Ere(k) 0.258 u(k) + 0.813 (-6


In equation (2-6), AL is the added virtual length to the actual length of the microstrip patch

antenna, due to fringing effect. For a square patch, AL is equal to 0.7mm, such that the total

length is now L + 2AL, or L + 1.4mm. This is a significant change for a high frequency patch

that is characterized by narrow bandwidth, and this is why we can not just estimate the length

from equation (2-2).


fo(k) :=
2-(L + 2L00k))-( (2-7)

In equation (2-7), fo is the resonant frequency of the patch antenna as a function of width to

length k, and after the fringing effect. By plotting fo in terms of k, we can solve for k that would

produce a resonant frequency equal to 2.45GHz, as seen in figure 2-9.










R~esonanice Frequency Vs.. k--W/LT!


2.510'~ 2.5 -10


2.~110~2.442x109


2.44 -10'



2.42-10'

.412.4 -109
0.6 0.8 1 1.2 1.4 1.6 1.3
0.5 k 2
k= W;L (ratio)


Figure 2-9. Effective length of patch after fringing effect Vs. ratio of width to length k.

We notice from figure 2-9 that the resonant frequency 2.45GHz occurs at a k value equal

to 0.91, which is slight less that a square patch.


L (k) := 2 AL(k)
(2f Er~>(2-8)

In equation (2-8), Le is the effective length after fringing, f is the resonant frequency. By

graphing Le as a function of k, we find the resonant length L that corresponds to a frequency of

2.45GHz, as seen in Figure 2.9 below.










Eff;c~tive Length~_ Le~m) Vs. h--W /L.


2.4
2.3I
2.I







Figure 2-10. Effective length of patch after fringing effect Vs. ratio of width to length k.

From figure 2-10 we notice that for a resonant frequency 2.45GHz, given k is 0.91

from Figure 2-9, we Eind that the effective length of the patch antenna is 2.917cm.

From equation (2-3) we Eind that the width of the antenna is about 2.654cm.

Antenna Simulation Results

There are many software that will do electromagnetic simulation based on method of

momentum (MoM). The author had chosen two software (PCAADO and Ansoft DesignerC) for

simulation to make sure that the two results were similar and consistent with the mathematical

model. The first software PCAAD uses regular mathematical equations to calculate antenna

characteristics as opposed to Ansoft DesignerC that simulates the electromagnetic field to find

the characteristics of the patch antenna. Seen in figure 2-11 is the simulation results for a patch

antenna using PCAADC.





Figure 2-11. Patch antenna simulation using PCAADC.

In figure 2-11 the author varied the antenna dimensions length and width to obtain a

resonant frequency as close to 2.45GHz as possible. The length found was 2.81cm and the width

being 2.9, which is close enough to the mathematical model described previously. We notice that

the bandwidth is only 2.8%, efficiency is about 50.2% and gain is 6.0dB. In addition, the

magnitude of the input impedance of the feed line is about 10402, which would need to be

matched to standard 5002.


Using Ansoft DesignerC is not as straight forward as PCAAD, however, it allows for

much more in depth analysis of the patch antenna. Seen in figure 2-12 is the layout of the patch

antenna in Ansoft Designer after optimizing dimensions 5002 match.


Center frequency [GHz]. 1.5
Compute Frequency step [GHz) 1.01
Number of frequencies .. .I

Frequency Input Impedance [ohms] Bandwridth. .. .. 1.8
2.453 78.4 +j 68.6
Efficiency .. ... 15. 2
Directivity [dB) ..lO .


-A~ntenna Parameters ..
Patch length [cm) .. .. 1. Patch wridth [cm .. 1.9

1/ Substrate thickness (cm] 1. Dielectric constant 1.4
Dielectric loss tangent .. 1.0 F eed line wridth [cm) 1.3


d






































Figure 2-12. Square patch antenna using Ansoft Designer, with given dimensions, and output
matched to standard 5002 using stub stripline of length 12mm by 29mm.

Seen in figure 2-13 is the return loss S11 (dB) in terms of the swept frequency F(GHz).

Return loss basically shows at what frequency the 5002 is happening. The lower the return loss in

dB, the better the match is. We notice that the return loss at 2.45GHz (pointer 1) is about -

24.7dB, and the -10 dB bandwidth extends from about 2.4GHz to 2.48GHz, which is really good

match for the 2.4GHz ISM band. One would try to match at the center of the 2,4GHz ISM band

because after fabrication of the patch antenna, there will definitely be process variations, such

that the dielectric material would probably be different than 4.4, or thickness would be different,

and so on.











Ansoft Corporation
Return Loss


Y1~
dB(S(Port1,Port1
Setup 1 Sweep


-30.0





-40.00-
2.002.202.40 2.60 2.60 3.d
F [GHz]
X1= 245GHz IX2= 2 40GHz IX3= 2 50GHz
Y1= -24 69 IY2= -11 02 Y3= -6 66


Figure 2-13. Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz.

The gain of this antenna is shown in Figure 2-14, which has good distribution of gain

centered at the operating frequency 2.45GHz. As discussed previously, the gain of the antenna is


a really important feature of wireless power transfer and self powered wireless sensor nodes. The


gain obtained from the antenna at the cost of directivity is almost considered free power, since

you are not paying for it with current consumption as the case with the power amplifier.

Therefore, one could cascade an array of antennas to increase the gain of the system. However,


the main disadvantage of antenna arrays is the larger size, which doubles for every 3 dB of gain.










Ansoft C~orporation

Gain (dB)























2.202.302.40) 2.50) 2.E
F [GHz]

Figure 2-14. Antenna gain of patch antenna (dB) in terms of the swept frequency F(GHz).

The patch antenna was fabricated using a home-made etching technique using ferric

chloride and blue press-n-peel sheets as seen in figure 2-15. The dimensions of the edges were

not as accurate as simulated, and the antenna went through corrosion by time. Therefore, the

performance of the antenna was not satisfactory, especially for a sensitive application as

remotely powered wireless sensors. The matching greatly affects efficiency and gain, home-

brewed antennas for this proj ect produce unacceptable results.





























Figure 2-15. Fabricated 2.45GHz antenna using homebrewed etching technique.

Another antenna was simulated with a different matching technique, and was sent to a

professional fabrication company PCBexpressC for accurate fabrication results. The antenna was

also simulated using Ansoft Designer and optimized to match at 2.45GHz, as seen in figure 2-16.


























Figure 2-16. Square patch antenna using Ansoft Designer matched to 5002 at 2.45GHz.


















































X1=2.45G Hz IX2= 2.42G H z IX3= 2.48G H z
Y1= -25.22 Y2= -9.71 IYS= -10.33


Seen in figure 2-17 is the return loss S11 (dB) in terms of the swept frequency F(GHz). We

notice that the return loss at 2.45GHz (pointer 1) is -25.2dB, which is better than the previous

version, and the -10 dB bandwidth extends from about 2.42GHz to 2.48GHz.


Y1L

Setup 1 : Sweep


An soft Corporation

PlanarEM1


Figure 2-17. Return loss S11 (dB) in terms of the swept frequency F(GHz), matched at 2.45GHz.

The gain of this antenna is shown in Figure 2-18, and has a better gain distribution

centered at 2.45GHz than the previous antenna.










Y1lL
d~l(Gainlnput)
thelja=0lol, phi=0
Setup 1 : Sweep


An soft Corrporation

PlanarEM1


F [G H ]


Figure 2-18. Antenna gain of patch antenna (dB) in terms of the swept frequency F(GHz). Gain
is about 1.4dB or 3.5dBi.

This patch antenna was sent to a professional PCB fabrication company PCBexpressC as


opposed to the previous antenna which fabricated in lab. The dimensions of the edges were

precisely as designed and very close simulation results. In addition, the antenna has a green

silkscreen which prevents the copper from corrosion.





























Figure 2-19. Patch antenna for 2.45GHz ISM band, with 3.5dBi gain. Fabricated at PCBexpress.

This patch antenna can be used for both the remotely powered sensor node, and the RF

signal generator. However, one might consider using a higher antenna gain on the RF generator

since size constraint would not be an issue. Therefore, for the 2.45GHz RF generator, an array

antenna was developed similar to the previous antenna, but has twice the gain. The antenna is

composed of two cascaded patches as seen in Figure 2-20.













Figure 2-20. Patch antenna array, matched at 2.45GHz, with gain equal to 6dBi.

The higher the gain the RF generator has, the more power is delivered to the wireless node.

Therefore, one can purchase an already made antenna array that has a larger bandwidth, and

higher gain than a patch antenna fabricated on a FR4 board.










Power Delivery

Using the fabricated 3.5dBi 2.45GHz patch antenna in figure 2-19 on the wireless sensor

node, and the 6dBi 2.45GHz antenna at the RF generator, with transmitter output power equal to

that of the amplifier in figure 2-3 A, we revisit equation (3) and figure 2-2. In equation (3), Pt is

set equal to +21dBm, which is the maximum output power of the ZRL-3 500. Gt is set equal to

4dB (not dBi), which is the directivity gain of the patch antenna array located on the RF

generator seen in figure 2-20. Gr is set equal to 1.5dB, which is the directivity gain of the patch

antenna located on self powered wireless sensor node. Therefore, we obtain the equation below.



~i~"" =""p(Cf zols~*iw i(2-9)

In equation (2-9), Pt, Gt, and Gr add up to 26.5dB. Therefore, the received power is 26.5dB

higher than that of figure 2-2. Notice in figure 2-21 how all the graphs have shifted up 26.5dB.

Power Received (dBm) Vs. Distance (m)


-20








-115
-120
-125

I 010 1 1-0
E P~R9161dl R
Ditne ewe T n x mtr

Fiur 2-.Improvedl~ wieespwr eevdwthadtoa 2.d an









Commercially Available Antennas for WPT

Even though we have made a significant 26.5dB improvement, the theoretical received

power at one meter is only -15dBm, which is not enough power for wireless power delivery.

Therefore, we would need to increase the power delivered in any means possible. This means

that a higher antenna gain at the transmitter is required. HyperLinkTech sell HG2414P patch

antenna array for WLAN use with high antenna gain equal to 14 dBi. This antenna would be

used on the RF generator due to its large size. HyperLinkTech also offers Yaggi antennas and

patch antenna arrays that have gains > 10dBi, and are readily available for customer purchase,

seen in Figure 2-21 A through E.





~j~A BE~ I: C -- D E

Figure 2-22. High gain commercially available antennas suitable for remote power transfer

Unfortunately, for the remotely powered wireless sensors, it is difficult to increase antenna

size due to size constraint. However, by using recent technology as chip antennas, one can

cascade high gain chip antennas while maintaining similar node dimensions. Therefore, the

author had found the following chip antenna 2450AT45A100 suitable for the purpose of

reducing the size of remotely powered wireless sensors, as seen in figure 2-22 A.

LO10.5



A/ B




Figure 2-23. Commercial chip antenna. A) Chip Antenna B) Dimensions of Chip antenna.
Borrowed from JohansonTechnology website.










The chip antenna is manufactured by JohansonTechnologyC with average gain equal to

1dBi, and maximum gain equal to 3dBi. The chip antenna is weakly matched to 5002 at

2.45GHz, with a return loss equal to -14dB. However, the company also offers a matching circuit

for better bandwidth(BW), gain, and return loss seen in figure 2-24 A, return loss figure 2-24 B.







1.5pF -1~
3.9nH
A m l B

Figure 2-23. A) Matching network B) Retumn loss S11(-22.6dB), and BW (600MHz)

Attenuation

The main reason for the attenuation of the wireless signal in air is due to the atmospheric

absorption due to the presence of water particles in the air. As seen in figure 2-24, the

atmospheric absorption is very high for 2.4GHz signal. This is the main disadvantage of

selecting a 2.4GHz frequency for wireless power transfer. However, it is a great advantage for

microwave ovens. As a matter of fact, it is the only reason why microwave ovens operate at

2.4GHz because water absorption is very high at that frequency, which means that food

containing water would cook very fast. However, one must not think that WPT is harmful for

humans, for two simple reasons. First, the ouput power in its best is limited to 4 watts EIRP,

which is well below that of a microwave oven 1000watts. Second, in order to cause any kind of

severe harm to humans the wavelength would have to be small enough to resonate with the

length of human cells and DNAs such as that of X-rays. A quick note to mention is that the radio

signal produced by the microwave oven (2.4GHz) is not even a microwave scale wave. The









wavelength of a 2.4GHz frequency is actually about 12.2 cm. However, it is called microwave

because looking at table 2-1, you can see that it lies in the broad microwave spectrum.

Multipath fading is also another reason for signal attenuation especially if the wireless

power transfer is being done indoors. Multipath fading could either by constructive or

destructive. However, in most case it is usually destructive due to the signal bouncing off

multiple walls back and forth. Multipath fading will reduce the amount of power transferred to

the remotely powered wireless sensor and also cause communication errors in the channel. An

open space would be a better environment for wireless power transfer. As a matter of fact, the

best place for WPT is actually outer space where there is no water or oxygen absorption to

attenuate the radio signal transmitted, but perhaps more interference and noise is the side effect.

In addition, Friis equation mostly represents a theoretical received power, in reality one should

only consider about '/ of the theoretical distance calculated, and this is being optimistic.



Feasibility

WPT would be feasible and useful with the availability of high power amplifiers starting

with at least a one watt output power, and a high gain antenna in the order of 14dBi and higher.

Also, areas with less humidity would be better environment for deploying remotely powered

wireless sensors. This is the reason why Nicolas Tesla decided to build his Tesla coil tower high

above sea level where humidity is much less. The author's best case scenario is the available

+21dBm power amplifier, 14dBi Antenna and 3dBi chip antenna. This gives a total system gain

of about 34dB, which is about -7dBm theoretical received power according to equation (3). In

the worse case scenario, the antenna would only have about 8dB gain, transmitter would transmit

about 18dBm, and receiver chip antenna would be OdB, such that the total system gain is now

+26dBm instead, which is more of an expected gain rather than a worse case. Referring back to










equation (3) a +26dBm system gain is equivalent to about -15dBm received power. However,

according to the author, in order to receive a decent amount of power to work with, one would

need a minimum of OdBm, and preferably greater than +10dBm. Assuming a high power

amplifier such as a 1Watt or a +30dBm, with a high gain antenna about +22dBi on the

transmitter and a 2 dBi chip antenna on the wireless sensor node, one can achieve about +50dBm

of system gain from transmitter to receiver. This much power would transfer a theoretical

+10dBm power to the wireless sensor node. However, a more practical received power would be

to subtract 6dB, which is equivalent to dividing the distance by 4, giving a +4dBm of received

power, which is still a good amount of power to work with. This would allow a wireless sensor

node to receive wireless power and operate comfortably about 2 meters away from transmitter.

Two meters does not sound like much of a distance, however, it' s a good starting point for

remotely powered wireless sensors.









CHAPTER 3
WIRELESS POWER TRANSFER SYSTEM DESIGN

System Level Design

The system level of remotely powered wireless sensors is mainly divided into two parts.

The first part is the RF generator and the second is the self powered wireless sensor. The RF

generator is responsible to generate the high power radio signal and radiate it towards the

wireless sensor. Power for the RF generator is considered to be abundant and constantly

available. ON the other hand, the wireless sensor maintains a low power state and energy

conservation via energy management system and low power software selectable states. The

wireless sensor has the responsibility to detect the high power signal and rectify it into usable DC

energy, stored locally for future use. Figure 3-1 shows the system level design of the system.


RF


RF Wirels
Geneato sensor






Figure 3-1. Top level system design.

In figure 3-1, the RF generator shown to the left generates the high power RF signal and

propagates it in the direction of the wireless sensor. Multiple antenna arrays can be used to

propagate a concentrated RF signal in the general direction of sensors in the vicinity of the RF

generator. A splitter can be used to split the RF signal into multiple sources at the cost of less

signal power. The middle part of figure 3-1 is the concentrated RF signal propagating in air, and

on the right is the remotely powered wireless sensor detecting and rectifying the RF signal.










System Design Constraints

The system constraints of the RF generator are far more flexible than that of the remotely

powered wireless sensor. The RF generator is considered to have infinite amount of available

power, yet it would also have the capability to operate with a large battery if required. The

current consumption of the RF generator is highly dominated by the RF power amplifier. The

ZRL-3 500 consumes about 575mA at +12volts DC input.

The remotely powered wireless sensor has far more constraints than the RF generator. First

of all the wireless sensor has to be optimized for lowest power consumption possible. The author

recommends a total power consumption less than 1mWatt in active wireless transmission state,

and less than luWatts in sleep mode. The remotely powered sensor has to be as small as possible

in order to be adopted in any application and in any size limited environment. Therefore, the

wireless sensor node would use a chip antenna instead of a patch to further reduce dimensions.

Ultra capacitors can also be used on the wireless sensor nodes due to their smaller size and larger

capacity. However, ultra capacitors have longer charging time due to their large capacity, which

would be a problem for charging. Surface mount components are advisable for the wireless

sensor node, and perhaps a final integrated ASIC would even further drop power consumption

and size.




Range Constraints

Due to the main size constraint of the remotely powered wireless sensor, the patch antenna

would have to be replaced with the chip antenna. This drops efficiency due to impedance

mismatch between chip antenna and RF to DC sensor. For this reason, the total power received

would be less, and thus the range would be reduced. For a typical scenario, the RF generator is

capable of outputting a typical +19dBm to +21dBm of output power, transmitter antenna has a










gain of about +8dBi to +12 dBi, and wireless sensor node has about -1dBi to +1dBi gain.

Therefore, the total system gain ranges from +26dB to +32dB. According to equation (3), an

average of +30dB system gain gives about -10dB of received power at one meter, and about

+0dBm at 20cm away. As mentioned in previous sections, usable power would have to be at

least +0dBm. Therefore, the range at the given constraints is limited to about 20cm. However,

with a higher power amplifier, this could be easily extended.

Cost Constraints

The RF generator is perhaps the most expensive unit in the design. However, there is only

one RF generator in the system, so it is not a major issue in the system. The author' s RF

generator cost was about $400 in parts. However, this cost can easily increase with the

replacement of the RF amplifier with a higher gain one. Amplifiers in the 1Watt range can easily

start about $1,000 and up to couple thousand dollars for military RF amplifiers. The wireless

sensor cost is much less than the RF generator, costing about $25 dollars with quantities greater

than a hundred. The author did not address issues with reducing cost, but monitored the increase

in cost as the proj ect developed. In general, since wireless power transfer and remotely powered

wireless sensors is not as popular yet, components such as power amplifiers are still expensive

and range in thousands of dollars. However, in the future, the author believes as the concept

picks up in the future, special designed high power amplifiers could be produced for cheaper

prices.

Radio Frequency Generator

The RF generator is one of the two legs of the system. The RF generator design would

either make or break the system. The system is mainly composed of a signal generator block, a

high power amplifier, and a baseband processor. In addition, a large graphical LCD display is










located on top of the RF generator for visual feedback. Figure 3-2 shows the back side of the RF

generator.


Figure 3-2. Back-side view of RF generator.

Seen in Eigure 3-2 is the backside of the RF generator. Displayed are two fans mounted on

top of a heat-sink. A temperature sensor monitors the amplifier' s temperature and turns on the

two fans when the temperature goes over a preset limit. The fans are tied down with tie-straps to

reduce vibration. Also see in Eigure 3-2 are 5 LEDs on the top left side. Four in which are yellow

and correspond to up, down, left, and right, and one LED is green and displays a green color.

These buttons are used for user input to navigate the menu on the LCD display. In Eigure 3-3, the

front of the RF generator is displayed.































Figure 3-3. Front-side view of RF generator.

Seen in figure 3-3 is the frontside view of the RF generator. First, there are two DC input

power supplies with labels "uC-DC +12-6V" for the microcontroller power supply, and "RF-DC

+12V" is used for for the RF power supplied to amplifier and signal generator. Second, there is a

RS232 serial port, that would communicate locally with the microcontroller. Third, there is a

microcontroller programmer "uC Prog" used to flash the microcontroller program memory.

Finally, there is the high power RF output port seen in the top right corner with an SMA

extended cable.

Seen in figure 3-4 is the right side view of the RF generator. There are two turn knobs

which are the RF tuner, and the LCD contrast tuner. The RF tuner when turned will adjust the the

RF frequency. The frequency can be adjusted from 2165MHz to 2650MHz, which in terms

covers the required 2.45GHz range. The other knob adjusts the contrast of the display for better

user view.





























Figure 3-4. Side view of RF generator, displaying tuning knobs.

Radio Frequency Signal Source

The main source of the RF generator is the local oscillator purchased from MiniCircuitsC

with model number ZX95-2650 with tuning frequencies from 2165MHz to 2650MHz. The

frequency is adjusted with the right knob seen in figure 3-4. The maximum output power of the

oscillator is about +5dBm. Voltage supply ranges from +12 to +13 volts and current

consumption is about 25mA. Seen in figure 3-5 is the oscillator ZX95-2650 from MiniCircuits.








CrASiE STY~LE GB956

Figure 3-5. Signal generator local oscillator from MiniCircuits.

High Power Signal Amplifier

The power amplifier was also purchased from MiniCircuits with model number ZRL-3 500.

The amplifier is powered from a +12 volts source and has an internal regulator at +5 volts. The










maximum output power of the amplifier is about +21dBm at best. This is considered to be a

medium power amplifier. However, for wireless power transfer, this amplifier is considered to be

a low power amplifier. Amplifiers with minimum output power of watt should be considered

for remote powered wireless sensors. The power amplifier is the most power hungry device in

the system consuming about 575mA of current. Seen in figure 3-6 is the power amplifier ZRL-

3500 from MiniCircuits.










Figure 3-6. RF amplifier ZRL-3 500 from MiniCircuits.

Antenna of Choice

Antenna choice is a very important factor in WPT systems. Since the RF generator has less

size constraints than the wireless nodes, the antenna size would not be a maj or concern.

Therefore, the antenna size can be relatively large with high gain. The author recommends

antennas with gain greater than 10dB, and preferably as high as possible. Antenna gain could

make the wireless power transfer an easier task. However, the higher the antenna gain, the

narrower the radiation beam, or directivity, which means that the antenna would have to be

directed towards the wireless sensor. The author chose the patch antenna array from HyperTech

with model number HG2414P with directivity gain of +14dBi, seen in figure 3-7 A. In addition,

the antenna has a 30 x 30 degrees vertical and horizontal radiation beam seen in figure 3-7 B and

3-6 C respectively.



















A Vertical B Horizontal C

Figure 3-7. Commercial patch antenna. A) Patch antenna HG2414P B) Vertical radiation pattern
C) Horizontal radiation pattern



RF to DC Wireless Sensor Node

The RF to DC remotely powered wireless sensor is the other leg that makes this proj ect

walk. The RF to DC sensor is responsible for harvesting the radiated power from the RF

generator and turning it into DC energy. The system design can be seen in Eigure 3-7.


Thermoelectric
S RF Tx
Piezoelectrio -_ Signal Energy||
+ DC/DC uP
RadioH Ir.I Rectifier I I Storage||
Solar i II Sensor


Figure 3-8. RF to DC wireless sensor node system design.

Seen in Eigure 3-8 is the system level design of the RF to DC remote powered wireless

sensor. Starting from left to right, the first stage is the source of energy harvested. There are

many sources of clean ambient energy that can be scavenged. We have discussed these sources

in the introduction. Therefore, we will only discuss the RF source in the green colored block.

When the RF signal is received via matched chip antenna, the signal is fed into the signal

rectifier. In this case, the signal rectifier is the voltage doubling schottky diodes. The schottcky









diode, also know as zero biased diodes (ZBD), have to have minimum forward voltage drop.

When the signal is detected and turned into DC voltage, the voltage enters the DC to DC

regulator, turning the received DC source into stable usable voltage such as 2.5volts or 2.0volts.

The lower the required voltage is the less power is consumed, which is a concept we also

discussed in the introduction. The regulated voltage is stored on a local capacitor, such as a

supercapacitor. One should be careful with the capacitor capacity, too high of capacitance would

take too long to charge. The author recommends a one mFarad value capacitor for a fast charging

response. The choice of capacitor is of course dependant on the application. The author uses a

regular 1mFarad capacitor for storage. When the energy is successfully stored on a capacitor, the

baseband section, seen in blue, can use this energy to monitor a sensor and send back a reading

value .

Antenna Matching

The path antenna used is the 2450AT45A100. The chip antenna is manufactured by

JohansonTechnologyC with average gain equal to 1dBi, and maximum gain equal to 3dBi. The

chip antenna is weakly matched to 5002 at 2.45GHz, with a return loss equal to -14dB. However,

the company also offers a matching circuit for better bandwidth(BW), gain, and return loss seen

in figure 2-24 A, return loss figure 2-24 B. By using the provided matching network, a better

5002 match is possible with a wide bandwidth of 600MHz. A 1.5p Farad and a 3.9n Henry is

required for matching. Impedance matching is important to achieve maximum power transfer

with minimum power loss and reflection.

RF to DC Matching Network

It is important to match the RF to DC stage of the sensor. This will determine how efficient

the wireless sensor is with converting RF energy to DC energy. Seen in figure 3-9 is the









































Figure 3-9. RF to DC matching network.


Seen in figure 3-9 is the voltage doubler RF to DC conversion with input matching to 5002.


The schottky diodes used are the ZBD HSMS-2850 from AgilentC. The output voltage is as


function of frequency is seen in figure 3-10, such that input power is OdBm.

An soft Corporation Y1~
Vout mag(V(VPRIS:Vau
Circuit NWA1


----C---C---C---C---


matching for the RF to DC sensor using smith chart technique. To match the input of the voltage


doubler to the antenna a 13nHenry inductor is required.


c~o
I PNU;M=
RZ=1s'X)Chmn
lZ-Cim


Port1
PNUM=1
Pl=E8::hm
lZ='kism


F IGHz]


Figure 3-10. RF to DC output voltage about 1.2 volts.










Also the simulated current is shown in figure 3-11, such that the input power is also

OdBm.

An soft Corporation Y
lout mag~l(IPRB lout)}
Circulti WA




















F [G H ]


Figure 3-11. RF to DC output current about 1.2 mA.

The simulation shows that since the voltage and the current are of equal values, then the

power is maximum, which means we have impedance matching at 2.45GHz.

Low Power BaseBand Processor

The baseband system is composed mainly of three blocks the PIC 16F686 microcontroller

from Microchip, the low power RF transmitter, and the piezoelectric sensor. When the

microcontroller receives enough energy to power up, it will go directly into sleep mode to

conserve the energy stored. The system is interrupt driven, meaning that when the piezoelectric

sensor is actuated, the microcontroller wakes up from sleep and transmits its unique ID to the

receiver base station at a frequency equal to 916MHz. Most of the energy management is done

by the microcontroller. The C code for the wireless sensor node is shown in appendix A.










Low Power Sensors

A unique sensor is used with the wireless sensor node. The sensor is made from a piezo

electric material that produces high voltage when stroke. This voltage has very low current,

however, it is enough for CMOS technology to be triggered. When the microcontroller is in sleep

mode and the piezoelectric sensor is moved, it produces a voltage burst that interrupts the PIC

from sleep mode. The importance of this kind of sensor is that it does not consume power to

operate, it actually produces it.

Low Power Wireless Transmitter

The transmitter is a SAW based low power transmitter. It operates at 9161VHz frequency,

which is different than the frequency supplying the power, so they don't interfere. The

transmitter can operate on a low supply voltage as 1.2 volts and current of 1.2 mA which is much

less than what a Zigbee node consume (50mA). Therefore, with minimum duty cycles, the sensor

can last forever, as long as there is a wireless power supply supplying power.









APPENDIX A
ELECTRONIC SCHEMATIC OF RF BASEBAD CONTROLLER

This schematic was designed and implemented by the author. The schematic was designed

you Protel DXP.


~nlr~~
e~L~ ~F ~~



81~ a,




~"~ -n~


01B


~BIIII~ IIIIII III IIII


varar


9191.91


S ~ll i l~~l~a~la~ i l l il a~ll~


I 12









APPENDIX B
REMOTE POWERED WIRELESS SENSOR C CODE

This is the C code implemented for the remote powered wireless sensor node. The code

was written using CCS compiler.


#include <16F688.h>
#device adc=8

#FUSES NOFCMEN //Fail-safe clock monitor disabled, switches to internal if
external fails, page 33 for more details
#FUSES IESO //Internal External Switch Over mode enabled
//#FUSES NOIESO //Internal External Switch Over mode disabled, page 33 for
more details
//#FUSES BROWNOUT //Reset when brownout detected
#FUSES NOBROWNOUT //No brownout reset
#FUSES NOCPD //No EE protection
#FUSES NOPROTECT //Code not protected from reading
#FUSES MCLR //Master Clear pin enabled
#FUSES NOPUT //No Power Up Timer, page 33 for more details
#FUSES NOWDT //No Watch Dog Timer
//#FUSES HS //High speed Osc (> 4mhz)
#FUSES XT //High speed Osc (<= 4mhz)
//#FUSES INTRC IO //Internal RC Osc, no clkout
//#FUSES INTRC //Internal RC Osc
//#FUSES LP //Low power osc < 200 khz

#use delay(clock=4000000) // 200K osc
#use rs23 2(b aud=3 8400,parity=N,xmit=PINC4, rcy=PIN_C 5,bits=8) // B aud=2.5 % Fosc ,
%3.125

// assume Tx bits= 32bits to be transmitted
// time to transmit = Tx bits 1/Baud = 32bits 1sec/3K bits = 11 msec
// Power Consumed = 1.5v 1.5mA = 2.25mW for the transmitter
// E = P* time = 2.25mW 11ms sec = 25uJ <<<<<<<<< Transmitter
// will need 3 diodes in series, or half voltage divider 3.3v/2 = 1.6v
// 1*Tao= 0.63*V=0.63*6.3v=4v
// E = 0.5(C)(V)^2 = 0.5(C)(4)^'2=25uJ => C=50uJ/16=3uF
//E =0.5(1 00uF)(4)^'2=800uJ
// Tao = R*C =10K*100uF = 1 sec to charge cap
// P = 200uA*3.3v = 0.66mW (pic power consumed)
// E = 0.66mW*11ms = 7uJ <<<<<<<<< PIC energy

int i=0;









//Baud Rates:
//110, 300, 1.2K, 2.4K, 4.8K, 9.6K, 19.2K 38.4K, 57.6K, 115.2K, 230.4K, 460.8K, 921.6K


//#byte OSCCON = Ox8F
//#byte OSCTUNE = Ox90
//#byte OSCTUNE = Ox90

//#byte TXSTA = Oxl6
//#byte RCSTA = Ox17


#int EXT
EXT~isr()

//setup_oscillator(OSCEXTERNAL); //OSC_31KHZ, OSC_125KHZ ,OSC_250KHZ,
OSC 500KHZ
//#use delay(clock=4000000) // 200K osc
//#use rs23 2(b aud=3 8400, parity=N,xmit=PINC4, rcy=PINC5S,bits= 8) // B aud=2.5 % Fosc ,
%3.125
delay~ms(10); //allow ascillator stability

SETUPUART(TRUE);
printf("%cAA%C ",0xFF,i);
SETUPUART(FAL SE);
output low(pin_c4);
i++;


void main()


char start byte='A', device~id='B', adc_data='C', crc_data='D', i;
unsigned int int~counter=0;


output_10w(pin_c4);

setup_adc_ports(NOANALOGS|IVS SDD);
setup_adc(ADC_OFF);

setup timer_0(RTCCNTERNAL|RTCCDIV_1);
setup tim er 1(T 1_DI SABLED);
//setup timer_2(T2_DISABLED,0, 1); //for PIC 16F684

setup_comp arator(NC NC);
setup vref(FALSE);











ext~int~edge(L_TOH);
enabl einterrupts(INTEXT);
//cl ear~interrupt(INTEXT);
enable~interrupts(GLOBAL);

//#use delay(clock=4000000) // 200K osc
//OSCCON = Ox21;


while (1)

sleep();









APPENDIX C
RF BASEBAD CONTROLLER C CODE

This is the C code implemented for the RF generator baseband processor. The code was

written using CCS compiler.


#include <16F777.h>
#device *=16
#device adc=8
#fuses HS,NOWDT,NOBROWNOUT
#use delay(clock=20000000)
#include
#include
#include
#use rs232(baud= 19200,parity=N,xmit=PINC6,rcy=PINC7,bits= 8sra = pc)

//#use rs232(DEBUGGER)
#use fast io(B)
//#use fast~io(D)
#include

//LCD port definition -Done!
#define gl_dat portb
#define gl_a0 pin_d7 //output pin
#define gl_cs pin_d6 //output pin
#define gl wr pin_d5 //output pin
#define gl~rst pin_d4 //output pin
//#define gl~rd pin_d4 // hard wired to VDD=+5V

//Bump switches
#define right bump !input(pin_e0) //input bump switch
#define left bump !input(pin_el) //input bump switch
#define RF_ON output~high(pin_dl1)
#define RF_OFF output_10w(pin_dl1)
//#define more bump !input(pin_e2) //input bump switch

#define pyro_an0 0
#define temp_an1 1

//IR port definitions
#define GP2DO2_VIN pin_d2 //green
#define GP2DO2_VOUT pin_d3 //yellow










// servo constatnts
//#define start byte Ox80
//#define device id Ox01

//LCD Constants used -Done!
#define SYS SET Ox40
#define SYS SLEEP Ox53
#define SYS CGRAM ADDR Ox5C
#define SYS SCROLL Ox44
#define SYS SCROLL HDOT Ox5a
#define SYS SCROLL RATE Ox5a


#define SYS
#define SYS
#define SYS
#define SYS
#define SYS
#define SYS
#define SYS
#define SYS
#define SYS
#define SYS

#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD
#define LCD


FORM Ox5d
ADDR Ox46
READ Ox47
DIR RT Ox4c
DIR LT Ox4d
DIR UP Ox4e
DIR DN Ox4f


CUR
CUR
CUR
CUR
CUR
CUR
CUR


OVER LAY Ox5b
MWRITE Ox42
MREAD Ox43


DISP OFF
DISP ON
INVERSE
NORMAL


Ox58
Ox59
Oxff
Ox00


_CR
_W
_H
_Lh
L1
L2
L3
L4
L5
L6
L7
L8
L9
L10
L11
L12
L13
L14
L15
L16


Ox20 //32 Chars/bytes per line
Ox100 //256 pixels wide
Ox80 //128 pixels high
Ox08 //Height of line (8x8 characters)
Ox00 //Line 1
Ox20 //Line 2
Ox40 //Line 3
Ox60 //Line 4
Ox80 //Line 5
OxA0 //Line 6
OxCO //Line 7
OxEO //Line 8
Ox100 //Line 9
Ox120 //Line 10
Oxl40 //Line 11
Oxl60 //Line 12
Oxl80 //Line 13
OxlA0 //Line 14
OxlCO //Line 15
OxlEO //Line 16









#define LCD GRH


Ox1000 //Graphic home position


// graphic LCD
char line_string [33];
void gl~init();
void gl_grfclr();
void gl_grfhome();
void gl_strobe();
void gltxtclr();
void glsetaddr (into glcur, long gl_addr);
void send_data (int gl byte);
void send_command (int gl_cmd);
#separate
void g_printf (char string[],long Line~num);
void g_clear_1ine (long Line~num);

void loadrimage();


void loadrimage()


int 1,3;
gl txtclr();
send_command(SYS_CUR ADDR); //CSRW command
send_data(0x00);
send_data(0x 10);
send_command(SYS_CUR DIRRT); //Cur movement right
send_command(SYSMWRITE);


while(fgetc(pc)! =']')
// after the ']' microcontroller will lose 31 characters (because of software rs232) to initiate i
and j
again:
for(i=0;i<128;i++) //128/8=16


for(j=0;j<32;j++) //256/8=32
{
send_data(fgetc(pc));


goto again;












int get temp()

set~adc_channel (temp_anli);
delay~ms(1);
return (read_adc()<<1);



#include


////////////////////////////MAIN()//////////////////////////
void main() {

//LCD Global Variables used -Done!
int gl_cmd;
int gl_cur;
int gl byte;
int gl i;
int gl j;
int gllk;
long gl x;
long gl_y;
long gl_addr;
long glJw;
int gl~read;
int gl_old;
long pix x;
long pix_y;
long gl_addrlo;
long gl_addrhi;
int gl~nib;

// servo control variables
mnt servo num;
int right~arm_angle;
int leftarm_angle;

//general variables
int i=03=0,k=0;

//variables used for UltraSound
long returned_distance;
float distance;










//variables used for ADC
int pyro_adc; //16-bit integers
float pyro volts;
int temp_cel; //should be long, but doubt temperature would be that high
float temp volts;

// CMUcam variables
char cmu~char[10];
long cmu_param[8];

int ir~reading;
int speed;
//char bt~rx[33];
char bt rx[3];
char rx~data[5];
int menu_option;

//pwm variables:
unsigned long pwm_freq;
unsigned int t2p;
unsigned int value;


//some stuff
setup_spi(FALSE);
//setup_counters(RTCCINTERNAL,RTCCDIV_2);
//setup timer_0(RTCCNTERNAL| RTCCDIV_8);
//setup timer_0(RTCCNTERNAL| RTCCDIV_64);
msec
setup timer 0(RTCCNTERNAL|IRTCCDIV_128);
msec, resolution = 128us (increments every 128us)
//setup timer 1(T1_DISABLED);
setup timer_1(T1_NTERNAL|T1DIV BY_1); //
//setup timer_2(T2 DIVBY_16,255,1 6); //overfl


//timer_0 will overflow every 16.32

//timer_0 will overflow every 32.7


verflows every 65.5ms
ows every 65.5ms


//setup timer 2(T2_DISABLED,0, 1);
setup_comparator(NC NC NCNC);
setup vref(FALSE);


//set~timer0(0);
//set~timerl(0);
//set~timer2(0);

countere0 overflow
counterel overflow


//reset timer
//reset timer
//reset timer









countere2 overflow

set tris b(0x00);
//set tris_d(0x00);


//enable
nbl
//enable


_interrupts(INT_TIMERO);
_interrupts(INT_TIMER1);
_interrupts(INT_TIMER2);


//activate timer
//activate timer
//activate timer

//dectivate timer
//dectivate timer
//dectivate timer

//activate global interrupt flag

//pwmmfreq = 98KHz


disable~interrupts(INT_TIMERO);
disable~interrpts(INT_TIMER1);
disable~interrupts(INT_TIMER2);

//enabl einterrupts(GLOBAL);
t2p = 255;
setup timer_2(T2 DIV BY_16,t2p,1);


//setup_ccpl1(CCPPWM);
setup_ccp2(C CPPWM);


// Configure CCP1 as a PWM


//setup_ccp3 (CCPPWM);

value = t2p>>1;

//set pwml_duty(value);
set_pwm2_duty (value);


//LCD
fprintf(pc, "\r\nInitializing Graphical LCD...");
output low(gl~rst);
delay~ms(100);

output~high(gl~rst); //Make sure reset is high
delay~ms(100);

output_10w(gl_cs); //Chip select active
del aym s( 10);


gl~init();


//Initialize display


fprintf(pc, "Done!\r\n");









strcpy(line_string, "Welcome to Wireless Power Trasfer!");
g_printf(line_string,LCDL1);

sprintf(line_string, "t2p=%u, value=%u",t2p,value);
g_printf(line_stri ng,LCDL2);

output low(pin_c2);

delay~ms(5000);
//t2p=255;
while (1)


//setuptimer_2(T2 DIVBY_16,t2p, 1);
value = t2p>>1; //divide by 2
sprintf(line_string, "t2p=%u, value=%u",t2p,value);
g_pri ntf(l ine_stri ng,L CDL2);
del aym s( 1000);
t2p = t2p +1;









APPENDIX D
RECEIVER INTERFACE TO ATLAS C CODE

This is the C code implemented for the receiver interfaced to Atlas. The code was written

using CCS compiler.

#include <16F688.h>

#device adc=8

#FUSES NOFC1VEN //Fail-safe clock monitor disabled, switches to internal if
external fails, page 33 for more details
#FUSES IESO //Internal External Switch Over mode enabled
//#FUSES NOIESO //Internal External Switch Over mode disabled, page 33 for
more details
//#FUSES BROWNOUT //Reset when brownout detected
#FUSES NOBROWNOUT //No brownout reset
#FUSES NOCPD //No EE protection
#FUSES NOPROTECT //Code not protected from reading
#FUSES MCLR //Master Clear pin enabled
#FUSES NOPUT //No Power Up Timer, page 33 for more details
#FUSES NOWDT //No Watch Dog Timer
#FUSES HS //High speed Osc (> 4mhz)
//#FUSES INTRC IO //Internal RC Osc, no clkout
//#FUSES INTRC //Internal RC Osc
//#FUSES LP //Low power osc < 200 khz

#use delay(clock=4000000) // 200K osc
//#use rs232(baud=1 9200,parity=N,xmit=PIN_ C4,rcy=PIN_Cbis=,DISALNS //
Baud=1.5% Fosc
#use rs23 2(b aud=3 8400 ,parity=N,xmit=PINC4, rcy=PINC5,bits=8) // B aud= 1.5 % F osc

// assume Tx bits= 32bits to be transmitted
// time to transmit = Tx bits 1/Baud = 32bits 1sec/3K bits = 11 msec
// Power Consumed = 1.5v 1.5mA = 2.25mW for the transmitter
// E = P* time = 2.25mW 11ms sec = 25uJ <<<<<<<<< Transmitter
// will need 3 diodes in series, or half voltage divider 3.3v/2 = 1.6v
// 1*Tao= 0.63*V=0.63*6.3v=4v
// E = 0.5(C)(V)^2 = 0.5(C)(4)^2=25uJ => C=50uJ/16=3uF
//E =0.5(1 00uF)(4)^'2=800uJ
// Tao = R*C =10K*100uF = 1 sec to charge cap
// P = 200uA*3.3v = 0.66mW (pic power consumed)
// E = 0.66mW*11ms = 7uJ <<<<<<<<< PIC energy
int rx_char, valid data3j;
unsigned int rx~buffer;
char bytel,byte2,byte3,byte4,byte5,byte6,bt7ye8









//Baud Rates:
//110, 300, 1.2K, 2.4K, 4.8K, 9.6K, 19.2K 38.4K, 57.6K, 115.2K, 230.4K, 460.8K, 921.6K

#int RDA
RDAjisr() //lots of work to be done here !!!


rx_char-getc();
printf("%c",rx_char);






void main()


//char start byte='A', device~id='B', adc_data='C', crc_data='D', i;
//char rx_string;
char bytel,byte2,byte3,byte4,byte5,byte6,bt7ye8
rx buffer=0;
//setup vref(VREF _OW|2);
//setup_comparator(Al_ VR);

//setup_adc_ports(sANO IsAN1|VSSVDD);
setup_adc_ports(NOANALOGS|IVSSVDD);
setup_adc(ADC_OFF);

setup timer_0(RTCCNTERNAL|RTCCDIV_1);
setup tim er 1(T 1_DI SABLED);
//setup timer_2(T2_DISABLED,0, 1); //for PIC 16F684

setup_comp arator(NC NC);
setup vref(FALSE);
//SET_TRIS_C(0x00);
//SET_TRISA(TRISA| 00000011);

printf("hello world");

//enable~interrupts(INT_COMP);
enable~interrupts(INTRDA);
enable~interrupts(GLOBAL);
//OSCCON = Ox21;

//SET TRIS C(0x00);










while (1)

//wait for interrupt










LIST OF REFERENCES


1. J. A. Paradise and T. Starner, "Energy Scavenging for Mobile and Wireless Electronics,"
IEEE Pervasive Computing, Vol. 4, No. 1, 2005, pp. 18-27.

2. I. Stark, "Thermal Energy Harvesting With Thermo Life," Proc. ofBSN 2006, IEEE
Computer Society Press, New York, 2006, pp. 19-22.

3. V. Raghunathan, A. Kansal, J. Hsu, J. Friedman, and M. B. Srivastava, "Design
Considerations for Solar Energy Harvesting Wireless Embedded Systems," IEEE IPSN,
2005, pp. 457-462.

4. J. Kymisis, C. Kendall, J. Paradise, and N. Gershenfeld, "Parasitic Power Harvesting in
Shoes," Proc. IEEE International Conference on Wearable Computing, 1998, pp. 132-
139.

5. S. Roundy, P.K. Wright, and J. Rabaey, Energy Scavenging for Wireless Sensor
Networks nI ithr Special Focus on Vibrations, Kluwer Academic Publishers, Boston, 2004.

6. S. Roundy, B. Otis, Y.H. Chee, J. Rabaey, and P. Wright, "A 1.9GHz RF Transmit
Beacon using Environmentally Scavenged Energy," Dig. IEEE Int. Symposium on Low
Power Elec. and Devices, Seoul, Korea, 2003.

7. S. Roundy, P.K. Wright, and J. Rabaey, "A Study of Low Level Vibrations as a Power
Source for Wireless Sensor Nodes," Computer Communications, vol. 26, no. 11, 2003,
pp. 1131-1144.

8. M. Loreto Mateu, "Energy Harvesting from Passive Human Power," PHD Thesis,
Universidad Politecnica de Catalufia, January, 2004.

9. M.B. Schiffer, The Portable RadiRRR~~~~~RRRRR~~~~o in American Life, Univ. of Arizona Press, Tucson,
1992.

10. S.A. Schelkunoff and H.T. Friis, Antenna Theory and Practice, John Wiley & Sons, New
York, 1952.

11. K. Finkenzeller, RF7D H~andbook, 2nd ed., John Wiley & Sons, New York, 2003.

12. T. Milligan, M~odern Antenna Design, McGraw-Hill Book Company, New York, 1985.

13. J. Heikkinen, P. Salonen and M. Kivikoski, "Planar Rectenna for 2.45 GHz Wireless
Power Transfer," Proc. oflEEE Rad'io and' Wireless Conference, 2000, pp. 63-66.

14. K. R. Carver and J. W. Mink, "Microstrip antenna technology," IEEE Trans Antennas,
1981, pp. 2-24.









BIOGRAPHICAL SKETCH

I received my bachelor' s degree in electrical engineering from the University of Florida.

My special undergraduate interest was radio frequency system level design. I did about two years

of research in the ECE department and had two maj or publications in which I was the first

author. I then joined a local company in Gainesville, Florida, in which I worked with high profile

companies such as DoCOMo, Intel, and Advanced Micro Devices (AMD) on a very specialized

and detailed analysis, in which mobile handsets' power usage is measured and benchmarked, and

design refinements and changes are recommended. I was the chiefRF engineer who was running

and managing all work related to power benchmarking. At a young age, I was managing two

people, and delivering high quality results to customers. I flew to business meetings with Intel,

DoCOMo, and AMD, and I met several of their VPs. This line of business generated significant

revenues to the company.

In the Fall of 2005, I j oined the CISE department at the University of Florida to pursue the

Master of Science degree in computer engineering under the supervision of Dr. Abdelsalam

(Sumi) Helal. I have known Dr. Helal for more than three years. I am honor to know him and

work for him. It has always been a pleasure to work with him on various proj ects and I am

fortunate enough to have learned a lot from him. In the first semester of graduate school, I

established myself as a research assistant for the pervasive mobile computing laboratory. I

designed an RF system that allows sensor network nodes to be power charged wirelessly and

from a distance as my thesis, which is currently prototyping the Pervasive and Mobile

Computing Laboratory, and has been disclosed to the University of Florida, which has accepted

the disclosure and filed for a patent application. The docket number of this patent is

PCTO2/22/06. Currently, I am applying wireless power transfer to the ATLAS Sensor Platform,

which is being commercialized by Pervasa, Inc, a University of Florida startup.