<%BANNER%>

A Multiple-station experiment to examine the close electromagnetic environment of natural and triggered lightning

University of Florida Institutional Repository

PAGE 1

A MUL TIPLE-ST A TION EXPERIMENT T O EXAMINE THE CLOSE ELECTR OMA GNETIC ENVIR ONMENT OF N A TURAL AND TRIGGERED LIGHTNING By J ASON E. JERA ULD A THESIS PRESENTED T O THE GRADU A TE SCHOOL OF THE UNIVERSITY OF FLORID A IN P AR TIAL FULFILLMENT OF THE REQ UIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORID A 2003

PAGE 2

Cop yright 2003 by Jason E. Jerauld

PAGE 3

F or my parents, Ronald Jerauld and Janice Desrosiers.

PAGE 4

A CKNO WLEDGMENTS The w ork presented in this thesis w ould not be possible without the guidance of Drs. Vladimir Rak o v and Martin Uman, to whom I gi v e my sincerest thanks. Their scientic rigor and inte grity ha v e gi v en me no less than the best e xample of ho w to conduct myself as a scientist and a scholar I w ould also lik e to e xpress my appreciation to Dr Doug Jordan, Michael Stapleton, Robert Olsen III, Alonso Guarisma, Oli v er P ankie wicz, Thomas Rambo, Joe Richard, Clif ford Jordan, Julia Jordan, Andre w Owens, Matt Rile y and e v eryone else who dug a ditch or ran a ber related to this e xperiment. I o we a special debt of gratitude to Geor ge Schnetzer and K eith Rambo, who I belie v e taught me the most of what I kno w about good engineering and eld research. I w ould lik e to thank Dr Carlos Mata for w orking with me to pro vide the informati v e and aesthetically pleasing gures found in Chapter 4. Also, I w ould lik e to thank John Cramer of V aisala Corporation for pro viding the NLDN information presented in Chapter 4. I must also gi v e thanks to Kath y Thomson for all of the help and support she has gi v en me since my rst days in the lightning research b usiness. I am also grateful to the people of the linux and L A T E X 2 e communities, for without them the process of writing this thesis w ould ha v e been much more dif cult. In particular I gi v e my thanks to Ron Smith, who de v eloped the ufthesis document class. Finally I w ould lik e to thank my f amily and friends for their patience and support, which ne v er w ai v ered, no matter ho w much I complained about writing this thesis. i v

PAGE 5

This w ork w as supported in part by NSF Grant A TM-0003994, U.S. DO T (F AA) Grant 99-G-043, Sandia National Laboratories Contract K OMO42296, and by Florida Po wer and Light Corporation. v

PAGE 6

T ABLE OF CONTENTS page A CKNO WLEDGMENTS . . . . . . . . . . . . . . . . i v LIST OF T ABLES . . . . . . . . . . . . . . . . . viii LIST OF FIGURES . . . . . . . . . . . . . . . . . ix ABSTRA CT . . . . . . . . . . . . . . . . . . . xiii 1 INTR ODUCTION . . . . . . . . . . . . . . . . 1 2 LITERA TURE REVIEW . . . . . . . . . . . . . . 4 2.1 The Lightning Dischar ge Process . . . . . . . . . . 4 2.1.1 Natural Lightning . . . . . . . . . . . . 4 2.1.2 T riggered Lightning . . . . . . . . . . . . 9 2.1.2.1 Classical-triggered lightning . . . . . . . 9 2.1.2.2 Altitude-triggered lightning . . . . . . . . 13 2.2 Ov ervie w of the ICLR T F acility . . . . . . . . . . 16 2.3 Multiple-Station Field Measurements of Natural and Rock et-T riggered Lightning . . . . . . . . . . . . . . . . 19 3 INSTR UMENT A TION . . . . . . . . . . . . . . . 22 3.1 Ov ervie w of the 2001 and 2002 Multiple Station Experiments . . 22 3.2 Control System . . . . . . . . . . . . . . . 25 3.2.1 The PIC Controller . . . . . . . . . . . . 28 3.2.2 Softw are . . . . . . . . . . . . . . . 41 3.2.3 T riggering System . . . . . . . . . . . . 45 3.3 Measurement Implementation . . . . . . . . . . . 49 3.3.1 Electric Field and Electric Field T ime-Deri v ati v e Measurements 49 3.3.1.1 Analysis of a conducting at-plate antenna . . . 51 3.3.1.2 Flat-plate antenna implementation . . . . . . 64 3.3.1.3 Electric eld measurement implementation . . . 67 3.3.1.4 Electric eld time-deri v ati v e measurement implementation . . . . . . . . . . . 77 3.3.2 Magnetic Field and Magnetic Field T ime-Deri v ati v e Measurements . . . . . . . . . . . . . 80 3.3.2.1 Analysis of a loop antenna . . . . . . . . 80 3.3.2.2 Loop antenna implementation . . . . . . . 89 3.3.2.3 Magnetic eld measurement implementation . . . 95 vi

PAGE 7

3.3.2.4 Magnetic eld time-deri v ati v e measurement implementation . . . . . . . . . . . 104 3.3.3 Optical Measurements . . . . . . . . . . . 108 3.3.4 Current Measurements . . . . . . . . . . . 111 3.3.5 Measurement Bandwidth Summary . . . . . . . 118 3.4 Fiber -Optic Links . . . . . . . . . . . . . . 118 3.4.1 Opticomm MMV -120C Fiber -Optic Links . . . . . . 119 3.4.2 Nicolet Isobe 3000 Fiber -Optic Links . . . . . . . 120 3.4.3 Meret MDL288DC Fiber -Optic Links . . . . . . . 120 3.4.4 Fiber -Optic Cables . . . . . . . . . . . . 121 3.4.5 Fiber -Optic Link Calibration . . . . . . . . . 123 3.5 Digitizers . . . . . . . . . . . . . . . . 129 3.5.1 Y ok og a w a DL716 . . . . . . . . . . . . 129 3.5.2 LeCro y L T344 W a v erunner . . . . . . . . . . 131 3.5.3 LeCro y L T374 W a v erunner2 . . . . . . . . . 133 3.6 V ideo System . . . . . . . . . . . . . . . 133 3.7 T iming System . . . . . . . . . . . . . . . 135 4 PRESENT A TION OF D A T A . . . . . . . . . . . . . 136 4.1 Data Summary . . . . . . . . . . . . . . . 136 4.2 Example W a v eforms . . . . . . . . . . . . . 139 4.2.1 Natural Flash MSE-0202 . . . . . . . . . . 139 4.2.2 Natural Flash MSE-0203 . . . . . . . . . . 142 4.2.3 Natural Flash MSE-0205 . . . . . . . . . . 143 4.2.4 T riggered Flash FPL-0205 . . . . . . . . . . 144 5 RECOMMEND A TIONS FOR FUTURE RESEARCH . . . . . . 154 LIST OF REFERENCES . . . . . . . . . . . . . . . . 158 BIOGRAPHICAL SKETCH . . . . . . . . . . . . . . . 160 vii

PAGE 8

LIST OF T ABLES T able page 3–1 Measured parameters for the 2001 MSE. . . . . . . . . . 25 3–2 Measured parameters for the 2002 MSE. . . . . . . . . . 27 3–3 Coordinates of the MSE measurement locations. . . . . . . . 27 3–4 F ormat of a PIC controller command data pack et. . . . . . . . 38 3–5 F ormat of a PIC controller response data pack et . . . . . . . 38 3–6 Bit settings of PIC controller commands. . . . . . . . . . 40 3–7 Salient characteristics of the 2001 MSE electric eld measurements. . . 76 3–8 Salient characteristics of the 2002 MSE electric eld measurements. . . 76 3–9 Salient characteristics of the 2001 MSE magnetic eld measurements. . 104 3–10 Salient characteristics of the 2002 MSE magnetic eld measurements. . 104 3–11 Estimated bandwidths of the 2001 and 2002 MSE measurements. . . . 119 3–12 MSE ber -optic link summary . . . . . . . . . . . . . 119 3–13 O TDR measured optical lengths and corresponding time delays for armored ber -optic cables used during the 2002 MSE. . . . . . . . . 128 3–14 Calculated time delays for the ber -optic cables used in the 2001 MSE. . 130 4–1 Flashes recorded by the MSE netw ork in 2001. . . . . . . . . 137 4–2 Flashes recorded by the MSE netw ork in 2002. . . . . . . . . 138 4–3 NLDN-reported parameters for rst return strok es of natural ashes recorded by the MSE netw ork. . . . . . . . . . . . . . . . 140 viii

PAGE 9

LIST OF FIGURES Figure page 2–1 General locations of ground ash char ge sources observ ed in summer thunderstorms in Florida and Ne w Me xico and in winter thunderstorms in Japan, using simultaneous measurements of electric eld at se v eral ground stations. . . . . . . . . . . . . . . . . . . 5 2–2 Lightning cloud-to-ground dischar ges cate gorized by leader propag ation direction and polarity of char ge transferred to ground. . . . . . . 6 2–3 Fiber glass rock et with a spool of K e vlar -sheathed copper triggering wire mounted to the base. . . . . . . . . . . . . . . . 10 2–4 Sequence of e v ents observ ed in a typical classical rock et triggered lightning ash. . . . . . . . . . . . . . . . . . . . 11 2–5 Classically triggered ash S-0116, initiated on July 27, 2001 21:56:06 UT The v aporized wire channel (initial stage) is on the right with the indi vidual return strok e channels blo wn to the left by the wind. The launcher is located under ground. . . . . . . . . . . . . . . . . . 13 2–6 Sequence of e v ents observ ed in a typical altitude rock et triggered lightning ash. . . . . . . . . . . . . . . . . . . . 14 2–7 Unintentional altitude triggered ash FPL-0205, initiated on July 9, 2002 16:26:10 UT The launcher (the intended strik e point on the insulating to wer) is pictured in the fore ground. . . . . . . . . . . . . . 16 2–8 Sk etch of the ICLR T at Camp Blanding, Florida. . . . . . . . 18 3–1 Sk etch of the 2001 and 2002 MSE measurement locations at the ICLR T The arro ws roughly indicate the orientation of video cameras and optical sensors at those locations. . . . . . . . . . . . . . 23 3–2 Simplied diagram of the MSE control and data acquisition system. . . 26 3–3 The PIC controller A) Front vie w B) Side vie w . . . . . . . 29 3–4 Diagram of ho w a PIC controller is installed with a measurement. . . . 30 3–5 Diagram of the PIC controller communication topology used during the 2001 MSE. . . . . . . . . . . . . . . . . . 34 ix

PAGE 10

3–6 Diagram of the PIC controller communication topology used during the 2002 MSE. . . . . . . . . . . . . . . . . . 35 3–7 PIC RF unit enclosure mounted with a solar cell. . . . . . . . 38 3–8 Flo wchart representation of the nal 2002 MSE softw are control algorithm. 46 3–9 Schematic of the MSE trigger circuit. Enclosed in the dashed lines are the reference v oltage circuits used in the bistable circuits. . . . . . 50 3–10 Illustration of the electric eld boundary condition at a perfectly conducting surf ace. . . . . . . . . . . . . . . . . . . 51 3–11 Frequenc y-domain equi v alent circuit, using a Norton equi v alent current source, of a at-plate antenna sensor feeding a load (represented by Z L ). . 54 3–12 Aluminum at plate antenna used in the MSE. . . . . . . . . 64 3–13 Detailed mechanical dra wing of the aluminum at plate antenna used in the MSE. . . . . . . . . . . . . . . . . . . 66 3–14 Diagram of an installation of a MSE measurement utilizing a at-plate antenna. . . . . . . . . . . . . . . . . . . 67 3–15 Inte grator capacitor assembly used in 2001. A) Closed Pomona box. B) Box open to sho w interior . . . . . . . . . . . . . . 69 3–16 Measured (dashed line) and e xpected (solid line) test circuit responses for inte grating capacitor unit 01-01 ( 0 : 477 F ) . . . . . . . . . 70 3–17 Measured (dashed line) and e xpected (solid line) test circuit responses for inte grating capacitor unit 02-09 ( 0 : 209 F ) . . . . . . . . . 71 3–18 Schematic of the high-impedance amplier used in the 2001 MSE. . . 72 3–19 Measured frequenc y response of high impedance amplier 01-07. . . . 73 3–20 Diagram of a MSE electric eld measurement. . . . . . . . . 74 3–21 Diagram of a MSE dE/dt measurement. . . . . . . . . . . 77 3–22 Frequenc y-domain equi v alent circuit, using a The v enin equi v alent v oltage source, of a loop antenna sensor feeding a load (represented by Z L ). . . 83 3–23 Square loops of 50 W coaxial cable in 3 4 inch PVC pipe. A) Single loop at Station 4. B) Crossed loops at Station 4. . . . . . . . . . 90 3–24 Diagram (A) and equi v alent circuit (B) of a dif ferential-output coaxial loop antenna with both ends of the cable terminated in 50 W . . . . . . 92 x

PAGE 11

3–25 Diagram (A) and equi v alent circuit (B) of a single-ended output coaxial loop antenna with both ends of the cable terminated in 50 W . . . . 94 3–26 Schematic of an e xample acti v e inte grator . . . . . . . . . 96 3–27 Schematic of the acti v e inte grator used in the 2001 MSE. . . . . . 98 3–28 Measured (dashed line) and ideal (solid line) magnitude responses of one of the 2001 MSE acti v e inte grator units. . . . . . . . . . . 99 3–29 Schematic of the acti v e inte grator used in the 2002 MSE. . . . . . 100 3–30 Diagram of a 2001 MSE magnetic eld measurement. . . . . . . 101 3–31 Diagram of a 2002 MSE magnetic eld measurement. . . . . . . 102 3–32 Diagram of a MSE dB/dt measurement. . . . . . . . . . . 106 3–33 Schematic of the MSE optical sensor circuit. . . . . . . . . 109 3–34 Diagram of a MSE optical measurement. . . . . . . . . . 110 3–35 MSE optical measurement assembly . . . . . . . . . . . 112 3–36 Aluminum rock et launcher with Hof fman box mounted to the base. . . 113 3–37 Diagram of the MSE current measurements. . . . . . . . . . 116 3–38 T ime-domain equi v alent circuit of the MSE current measurements. . . 117 3–39 Example video frame from the MSE video system. Going clockwise from the upper left, the four quadrants represent the camera vie ws from IS1, IS2, IS4, and IS3, respecti v ely with the lightning being in vie w of the IS4 camera. . . . . . . . . . . . . . . . . . . 134 4–1 Frame of video sho wing the rst return strok e of natural positi v e ash MSE-0202. The channel is seen in the vie w from IS1, which f aces roughly south-west. . . . . . . . . . . . . . . . . . 141 4–2 Frame of video sho wing a channel from natural ne g ati v e ash MSE-0203. The channel is seen in the vie w from IS3, which f aces roughly south-east. The corrupted image seen in the vie w from IS4 is due to electromagnetic interference from the lightning. . . . . . . . . . . . . 143 4–3 Frame of video sho wing natural ne g ati v e ash MSE-0205. The channel is seen in the vie ws from IS3 and IS4, which f ace roughly south-east and east, respecti v ely . . . . . . . . . . . . . . . . 144 4–4 Recorded electric elds, magnetic elds, and optical signals for natural positi v e ash MSE-0202 on an 800 ms time scale. . . . . . . . 146 xi

PAGE 12

4–5 Recorded electric elds, magnetic elds, and optical signals for the rst leader/return strok e sequence of natural positi v e ash MSE-0202 on a 130 ms time scale. . . . . . . . . . . . . . . . . . . 147 4–6 Recorded electric elds, magnetic elds, and optical signals for the second leader/return strok e sequence of natural positi v e ash MSE-0202 on a 70 ms time scale. . . . . . . . . . . . . . . . . . . 148 4–7 Recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of natural positi v e ash MSE-0202 on a 10 s time scale. 149 4–8 Recorded electric elds, magnetic elds, and optical signals for natural ne g ati v e ash MSE-0203 on an 800 ms time scale. . . . . . . . 150 4–9 Recorded electric elds, magnetic elds, and optical signals for natural ne g ati v e ash MSE-0205 on an 800 ms time scale. . . . . . . . 151 4–10 Recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of natural ne g ati v e ash MSE-0205 on a 10 s time scale. 152 4–11 Recorded electric elds, magnetic elds, current, and optical signals for ne g ati v e altitude-triggered ash FPL-0205 on an 800 ms time scale. . . 153 xii

PAGE 13

Abstract of Thesis Presented to the Graduate School of the Uni v ersity of Florida in P artial Fulllment of the Requirements for the De gree of Master of Science A MUL TIPLE-ST A TION EXPERIMENT T O EXAMINE THE CLOSE ELECTR OMA GNETIC ENVIR ONMENT OF N A TURAL AND TRIGGERED LIGHTNING By Jason E. Jerauld May 2003 Chair: Martin A. Uman Major Department: Electrical and Computer Engineering This thesis presents a complete description of an automated e xperiment to measure the close (within a fe w hundred meters) electromagnetic en vironment of natural and rock et-triggered lightning. The e xperiment consists of a netw ork of wideband sensors spread about an area of approximately 0 : 5 km 2 at the International Center for Lightning Research and T esting (ICLR T), located at Camp Blanding, Florida and is a continuation of the w ork presented in ( Cra wford et al. 2001 ). This netw ork be g an operation during summer 2001 and measured quantities including the v ertical electric eld at eight locations, the horizontal magnetic eld at tw o locations, and the optical output of the bottom hundred meters or so of the lightning channel as observ ed from tw o locations. In 2002, the netw ork w as upgraded to include electric eld time-deri v ati v e measurements at four locations and magnetic eld time-deri v ati v e measurements at three locations. In addition, the induced current w as measured in an 14 m grounded conducting structure. The system is automatically turned on and of f by sensing the ambient electric eld amplitude; and is triggered by simultaneous signals from the tw o optical sensors vie wing the netw ork from opposite corners. In addition to a complete xiii

PAGE 14

description of the sensors and control system, representati v e data from both natural and triggered lightning, acquired in 2002, are presented. xi v

PAGE 15

CHAPTER 1 INTR ODUCTION Lightning is one of nature' s most f antastic visual and auditory displays. Lightning has played a k e y role in the mythology of man y cultures; se v eral of which had specic gods dedicated to the sk y weather and lightning. Man y of these gods were also considered gods of destruction, chaos, spite, or mischief. This association is not coincidental and can be easily concluded by an yone who has witnessed or has f allen victim to the a wesome destructi v e force of lightning. Although most people li ving in the twenty-rst century do not attrib ute lightning to supernatural forces, the y are no less susceptible to its deleterious ef fects. It is estimated that about 1000 people are killed from lightning each year around the w orld, with approximately 100 of those people li ving in the United States. Those who do survi v e, probably ten times those killed, are sometimes left with debilitating nerv e damage and chronic pain. According to the National Lightning Safety Institute (NLSI, http://www.lightningsafety.com ), lightning is responsible for a major portion (about 30%) of electrical po wer outages across the United States, costing tens of millions of dollars per year with total costs approaching $1 billion. In addition to damaging the electrical infrastructure, lightning is responsible for forest res, res to man-made structures, e xplosions of stored ammable substances (such as petroleum products), aircraft mishaps and upsets, and damage to electronic components. Most people ha v e lost or ha v e kno wn someone who has lost computer or telephone equipment as a result of lightning. In addition to being a destructi v e force, lightning is also lar gely responsible rene wing the Earth' s forests. F orest res started by lightning serv e to clear forests and pa v e the w ay for ne w gro wth. In addition, man y scientists theorize that se v eral billion 1

PAGE 16

2 years ago, lightning may ha v e pro vided the “spark” needed to turn the primordial ooze into the be ginnings of life on Earth; and may ha v e pro vided the chemical nutrients to sustain it. Ob viously there are man y practical reasons to study lightning. Monumental progress has been made in the eld since the rst systematic lightning e xperiments were conducted in the 1750s. Despite all that has been learned, there is much about the lightning dischar ge process that is poorly understood, and the study of lightning remains a v ery acti v e research area. Much that has been learned is the direct result of rock et-triggered lightning e xperiments, as discussed in Section 2.1.2 This thesis presents a description of an e xperiment intended to measure the close (withing a fe w hundred meters) electromagnetic en vironment of lightning, that w as elded during the summers of 2001 and 2002. This is kno wn as the Multiple-Station Experiment (abbre viated MSE in this thesis). The main purpose of this e xperiment w as to e xamine natural lightning, although data were also recorded for rock et-triggered lightning. While not the rst e xperiment of this type, it is the only one to combine time-domain measurements of electric and magnetic elds (and their time-deri v ati v es), optical signals, and induced currents from close natural lightning into a single e xperiment. Existing and future data obtained from this e xperiment should pro vide a wealth of ne w information about the ph ysics of natural lightning. The main goals for this e xperiment are Characterize rst return strok es in natural lightning. Determine whether subsequent strok es in natural lightning are identical to strok es initiated in rock et-triggered lightning. Chapter 2 gi v es a brief re vie w of the literature concerning lightning phenomena rele v ant to this thesis. Chapter 3 gi v es a complete description of the sensors, control softw are, ber -optics, digitizers, and video equipment; as well as ho w the y are combined to implement the e xperiment. When applicable, the distinction is made

PAGE 17

3 between congurations used in the 2001 and 2002 e xperiments, since the y dif fered signicantly in some cases. Chapter 4 gi v es a summary of all data recorded in 2001 and 2002 as well as a selection of w a v eforms recorded during the 2002 e xperiment. Finally Chapter 5 gi v es some conclusions and recommendations for future w ork. It should be noted that while the author w as primarily responsible for or g anizing and implementing the 2002 e xperiment, much of the 2001 e xperiment w as the w ork of Geor ge Schnetzer (a consultant on the project who is formerly of Sandia National Laboratories) and K eith Rambo (chief engineer at the Lightning Research Laboratory at the Uni v ersity of Florida). Their w ork in 2001 f acilitated the implementation of the 2002 e xperiment immensely

PAGE 18

CHAPTER 2 LITERA TURE REVIEW This chapter presents a brief re vie w of the literature concerning the lightning phenomena rele v ant to this thesis. Section 2.1 presents a brief introduction to the ph ysics of lightning, both natural and articially-triggered. Section 2.2 gi v es a brief o v ervie w of the ICLR T Finally Section 2.3 presents a re vie w of pre vious multiple-station lightning electromagnetic eld measurements. 2.1 The Lightning Dischar ge Pr ocess 2.1.1 Natural Lightning Lightning is an electrical dischar ge that is responsible for the rapid (within less than a second or so) transfer of char ge between the atmosphere and the Earth or between dif ferent parts of the atmosphere. Lar ge char ge centers are located in clouds termed cumulonimb us commonly referred to as thunderclouds. In Florida, these clouds usually e xhibit an “an vil” shape and are typically about 10 to 12 km in height with a lo wer visual boundary about 1 to 2 km abo v e ground ( Uman 1987 ). The char ge structure of a cumulonimb us can be crudely modeled as a v ertical tripole consisting, in temperate re gions, of a positi v e char ge center at a height of approximately 10 km a ne g ati v e char ge center at 5 km and another positi v e char ge center at 2 km. The tw o upper char ges are usually specied to be equal in magnitude and therefore form a dipole. The magnitude of the lo wer positi v e char ge is signicantly smaller than that of the dipole char ges. The general locations of ground ash char ge sources observ ed in summer thunderstorms in Florida and Ne w Me xico and in winter thunderstorms in Japan (using simultaneous measurements of electric eld at se v eral ground stations) are sho wn in Figure 2–1 4

PAGE 19

5 4 20 16 12 8 20 C 10 C 0 C 0 C Florida Storms Summer New Mexico WinterStorms JapanHeight, km Figure 2–1. General locations of ground ash char ge sources observ ed in summer thunderstorms in Florida and Ne w Me xico and in winter thunderstorms in Japan, using simultaneous measurements of electric eld at se v eral ground stations. Adapted from ( Krehbiel et al. 1983 ) and ( Rak o v and Uman 2003 ). Most lightning dischar ges occurs within a gi v en cloud. Although intra-cloud dischar ges are of particular concern to the a viation industry cloud-to-ground dischar ges are responsible for most lightning-related damage and injury Hence, the study of cloud-to-ground lightning dischar ges has man y practical applications. Cloud-to-ground dischar ges can be classied into four cate gories (Figure 2–2 ). This classication is based on the polarity of the char ge transferred to ground and the direction of propag ation of the initial leader propag ation. 1. Do wnw ard ne g ati v e 2. Upw ard ne g ati v e 3. Do wnw ard positi v e 4. Upw ard positi v e

PAGE 20

6 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + ++ + + + + + + + 1 2 3 4 Figure 2–2. Lightning cloud-to-ground dischar ges cate gorized by leader propag ation direction and polarity of char ge transferred to ground. Adapted from ( Ber ger and V ogelsanger 1966 ). Cate gories 1 and 2 ef fecti v ely lo wer ne g ati v e char ge to ground while Cate gories 3 and 4 ef fecti v ely lo wer positi v e char ge to ground. Cate gory 1 comprises about 90% of all cloud-to-ground ashes. Cate gory 3, while accounting for only about 10% of cloud-to-ground dischar ges, is of particular interest because of the lar ge peak currents in v olv ed. Cate gories 2 and 4 (relati v ely rare compared to cate gories 1 and 3) are most often observ ed on tall structures or mountain tops. A lar ge portion of lightning research has focused on do wnw ard-ne g ati v e ashes because of their o v erwhelming presence relati v e to the other three types of cloud-to-ground lightning. Ho we v er man y studies ha v e been performed using tall objects (such as to wers) that e xperience primarily upw ard lightning. A do wnw ard-ne g ati v e ash be gins with the initiation of a stepped leader from the ne g ati v e char ge center of the thundercloud. The stepped leader serv es to form a ne g ati v ely char ged plasma channel from the cloud to the ground. It is thought that

PAGE 21

7 this initiation is preceded by a preliminary breakdo wn process within the cloud. As the name suggests, the stepped leader propag ates in discrete b ursts or “steps” of e xtension. The stepped leader tra v els from cloud to ground at an a v erage speed of 2 10 5 m s 1 with an a v erage step length of tens of meters, and deposits ne g ati v e char ge along the channel. The length of each step and the time-interv al between steps is a function of height abo v e ground and it has been observ ed that both decrease as the leader approaches ground. The stepped leader phase typically lasts for some tens of milliseconds with each step lasting about 1 s and with the time between steps being tens of microseconds ( Uman 1987 ). As the leader approaches ground, an upw ard leader ha ving positi v e char ge, is initiated from the ground or other grounded objects (e.g., trees or other structures). Probably se v eral upw ard unconnected leaders are initiated from dif ferent locations. At some tens of meters abo v e ground, one of these upw ard leaders will connect with a branch of the do wnw ard stepped leader in a relati v ely poorly understood phase of the dischar ge kno wn as the attachment process. Once the tw o leaders ha v e connected, a lar ge sur ge of current, kno wn as the rst return strok e, tra v els back up the stepped leader channel neutralizing the char ge deposited by the leader ef fecti v ely lo wering ne g ati v e char ge to ground. The return strok e tra v els at about one third to one half the speed of light with speed decreasing with increasing height. The return strok e process can also be vie wed as a potential discontinuity tra v ersing the channel since the re gion ahead of the return strok e front (the ne g ati v ely char ged leader channel) is at a much higher ne g ati v e potential (near cloud potential which is ne g ati v e se v eral tens of me g a v olts or more) than the re gion behind the front, which is near Earth potential. Once the return strok e front has reached the cloud char ge, a subsequent do wnw ard leader may be initiated. T ypically this ne w leader kno wn as a dart leader follo ws the path of the pre vious channel and does not e xhibit stepping. In other w ords, a dart

PAGE 22

8 leader propag ates continuously not in discrete b ursts. Dart leaders tra v el at an a v erage speed of 10 7 m s 1 or about tw o orders of magnitude f aster than stepped leaders. Occasionally “dart-stepped” leaders are observ ed, which be gin as dart leaders b ut e xhibit stepping near ground. As the dart or dart-stepped leader approaches ground, another upw ard leader is initiated. Unlik e the rst return strok e, the attachment process for subsequent strok es typically occurs when the upw ard leader is only a fe w meters in height. The connection of the tw o leaders results in another return strok e. The characteristics of measured currents and elds from rst and subsequent return strok es are statistically dif ferent. This leader/return strok e process can occur man y times o v er the course of a ash, b ut 3 to 5 strok es is typical. Do wnw ard-positi v e ashes, which account for roughly 10% of the total cloud-to-ground dischar ges, are initiated by a leader process that is similar to that of ne g ati v e ashes. The do wnw ard leader initiated from the cloud deposits positi v e char ge along the channel and may or may not e xhibit stepping. Positi v e rst strok es can e xhibit a much higher peak current and char ge transfer than ne g ati v e rst strok es, and single-strok e ashes are much more common in positi v e than in ne g ati v e ashes. Upw ard ashes (Cate gories 2 and 4) are initiated in a completely dif ferent manner than do wnw ard ashes. In an upw ard ash, the rst leader is initiated from the ground-based object. A current, kno wn as the initial continuous current (ICC), o ws along the channel. T ypically this is is a steady current se v eral hundred amperes in magnitude lasting for se v eral hundred milliseconds and is not unlik e the initial stage current observ ed in classical rock et triggered lightning, which is discussed in Section 2.1.2.1 When the upw ard leader reaches the cloud base, there is a brief no-current interv al follo wed by subsequent dart leader/return strok e sequences, similar to subsequent strok es of do wnw ard ashes.

PAGE 23

9 2.1.2 T rigger ed Lightning The random nature of lightning mak es it a v ery dif cult phenomenon to study therefore a technique w as de v eloped in the 1960' s to articially initiate a ash “on demand. ” This most widely used and successful technique in v olv es the launching of a small rock et trailing a thin metallic wire. Other potential techniques in v olving, for e xample, lasers ha v e been attempted without apparent success. The “rock et and wire” technique can be di vided into tw o main cate gories, the classical and altitude triggering techniques.2.1.2.1 Classical-trigger ed lightning The classical rock et and wire technique is, by f ar the most-used technique to articially initiate a lightning ash. A small rock et, typically about one meter in length, is launched upw ard at an initial v elocity of about 200 m s 1 with a thin-g auge metallic wire trailing behind it. The spool of wire can either be tted to the rock et or the launcher itself, b ut in either case, one end of the wire is attached to the rock et and the other is attached to launcher The launcher is attached to the object to be struck or is grounded. The wire can be made of an y conductor although steel or K e vlar -reinforced copper ha v e been used most. Reinforcing the copper wire with K e vlar gi v es it enough strength to survi v e the launch without being brok en. The UF lightning research team currently uses spools of 700 m K e vlar -reinforced copper wire mounted to the base of a 1 : 15 m long ber glass rock et, as sho wn in Figure 2–3 Other congurations ha v e been used by the UF team in the past. In order to ha v e an y chance of a successful trigger typically three conditions ha v e to be satised. First, there must typically be a thundercloud o v erhead. Firing a rock et into “blue sk y” has v ery little chance of success, although being under the edge of the storm is occasionally acceptable. Second, in Florida the quasi-static eld at ground, as measured with an electric eld mill, must be belo w 5 kV m 1 with a v alue belo w 6 kV m 1 being desirable. Third, there must be lightning acti vity within se v eral

PAGE 24

10 Figure 2–3. Fiber glass rock et with a spool of K e vlar -sheathed copper triggering wire mounted to the base. kilometers, b ut only a moderate amount of acti vity is desirable since an o v erab undance of natural lightning in the area may disrupt the triggering process. All of these conditions are empirically deri v ed and based on years of e xperience b ut e xceptions can occur Furthermore, each metric indi vidually can be potentially misleading, hence the use of all three increases the lik elihood of success. This conserv ati v e approach is necessary due to the e xpensi v e nature of the rock et and wire spool conguration. Ev en when all of these conditions are satised, the success rate for the Uni v ersity of Florida group is only about 50%. The sequence of e v ents observ ed in a typical classical rock et triggered lightning ash is sho wn in Figure 2–4 The rock et and wire launch has the ef fect of quickly erecting a v ery tall grounded structure. Assuming the abo v e conditions are met, when the rock et reaches about 300 m abo v e ground, an upw ard leader will be initiated from the tip of the wire due to electric eld enhancement there. The polarity of this leader will be positi v e if there is a ne g ati v e char ge center o v erhead, and the leader will be

PAGE 25

11 1 4 3 5 6 ++ + + 2 3. Initial Continuous Current (ICC)4. NoCurrent Interval6. Upward Return Stroke 1. Ascending Rocket2. Upward Positive Leader5. Downward Negative Leader 200 m s1 1 2 s Hundreds of ms T ens of ms CopperW ire NaturalChannel W ire-trace Channel10 7 m s110 8 m s110 5 m s1300 m Figure 2–4. Sequence of e v ents observ ed in a typical classical rock et triggered lightning ash. Adapted from ( Rak o v et al. 1998 ). attracted to the cloud char ge. The mo v ement of char ge to the leader tip causes a quasi-static current to o w along the wire which increases in amplitude as the upw ard leader propag ates. When a current of a fe w hundred amperes is o wing through the wire, the wire e xplodes, briey interrupting the current. This interruption lasts about 10 s and can be manifested as either a se v ere drop in current amplitude or a complete cessation of the current. After the no-current interv al, the current is abruptly reestablished, often with a lar ge pulse, a signicant part of the process kno wn as the initial current v ariation (ICV). The details of the mechanism by which the current is cut of f and abruptly reestablished are poorly understood. The wire is replaced by a plasma channel and the reestablished quasi-static current, often superimposed

PAGE 26

12 with man y pulses, continues to o w for se v eral hundred milliseconds. This phase of the dischar ge is kno wn as the initial continuing current (ICC). The ICC in triggered lightning has been observ ed to be similar to that observ ed for natural upw ard ashes from tall structures. The upw ard leader initial current v ariation, and initial continuing current together constitute the initial stage (IS) of a triggered lightning ash. The initial stage is often colloquially referred to as the wir eb urn stage, since the triggering wire is destro yed in the process. By the time the ICC has ended (some hundreds of milliseconds from the initiation of the upw ard leader), the upw ard leader has long since entered the cloud. After the cessation of the ICC and a no-current interv al, a dart or dart-stepped leader ne g ati v ely char ged, may be initiated from the cloud and follo w the path of the pre vious upw ard leader and v aporized wire. As this do wnw ard leader approaches ground, a short upw ard leader is initiated from the launcher and probably from other grounded objects, as in natural lightning. The path to the launcher is at a higher temperature and hence less dense than the surrounding vir gin air and may ha v e higher conducti vity The do wnw ard leader attaches to the upw ard leader originating from the launcher and, as with natural lightning, a return strok e is initiated. Zero or o v er twenty additional strok es may follo w Ho we v er as with natural lightning, 3 to 5 are typical. Figure 2–5 sho ws a photograph (e xposed o v er se v eral seconds) of classically triggered ash with multiple return strok es. A considerable percentage of classically triggered ashes (40% or 16 out of 40 ashes during the 2001 and 2002 seasons at the ICLR T) contain the initial stage only and ha v e no return strok es. Usually the intent of a triggered lightning e xperiment is to study the return strok es and a wireb urn is considered an unsuccessful attempt. There is, ho we v er signicant scientic v alue in studying the initial stage. Therefore, more appropriate terms for a wireb urn are “classically triggered ash consisting of initial stage only” or “classically triggered ash with no return strok es. ”

PAGE 27

13 Figure 2–5. Classically triggered ash S-0116, initiated on July 27, 2001 21:56:06 UT The v aporized wire channel (initial stage) is on the right with the indi vidual return strok e channels blo wn to the left by the wind. The launcher is located under ground. It should be emphasized that all strok es in classically triggered lightning are initiated by dart or dart-stepped leaders and that the measured currents and elds are statistically similar to subsequent strok es in natural lightning. Therefore, the rst strok e, observ ed in natural lightning dischar ges, is not present in classical triggered lightning. It follo ws that, while e xtremely useful, the classical rock et and wire technique cannot be used to study rst strok es in natural lightning, which are clearly of interest.2.1.2.2 Altitude-trigger ed lightning Since the classical rock et and wire technique is incapable of initiating a rst return strok e, an alternati v e triggering technique kno wn as “altitude triggering” w as de v eloped. While f ar less ef cient and f ar more unpredictable than classical triggering, the altitude technique can initiate a a stepped leader while the classical technique cannot.

PAGE 28

14 + + + + + + + + + + 5 2 3. BiDirectional Leader 1. Ascending Rocket2. Upward Positive Leader 3 + + + 4 + + + + + 1 + + + 4. Upward Return Stroke5. Upward Positve Leader K e vlar Cable CopperW ire200 m s1200 m s110 5 m s110 5 m s13 s150 m 50 m 400 m CopperW ire10 5 m s110 5 m s112 km10 710 8 m s110 510 6 m s16 ms1 ms 10100s Figure 2–6. Sequence of e v ents observ ed in a typical altitude rock et triggered lightning ash. Adapted from ( Rak o v et al. 1998 ). The sequence of e v ents observ ed in a typical altitude rock et triggered lightning ash is sho wn in Figure 2–6 Essentially the same conguration is used for the altitude technique as for the classical technique. The only dif ference is that the wire used for an altitude trigger is not grounded. In one conguration, the rst 50 m of the spool consists of K e vlar reinforced copper wire and is kno wn as the intercepting wire. The ne xt 400 m consists of only K e vlar and the remainder of the spool is wire. When the rock et is launched, the wire is unspooled and the result is a length of ungrounded wire tra v eling upw ard. This oating wire is in the presence of a v ery strong electric eld, resulting in char ge separation within the wire. If the rock et and wire technique

PAGE 29

15 is used with ne g ati v e char ge centers o v erhead (as in Florida), a positi v e char ge will accumulate at the top of the wire and a ne g ati v e char ge will accumulate at the junction of the wire and the K e vlar section. As in the classical technique, an upw ard positi v e leader is initiated from the top of the wire. Se v eral milliseconds later a do wnw ard ne g ati v e leader is initiated from the wire/K e vlar junction. There is no plasma channel or other conducting path to ground due to the K e vlar so therefore the ne g ati v e leader must propag ate through vir gin air and hence e xhibits stepping. Ideally the do wnw ard ne g ati v e stepped leader will attach to an upw ard positi v e leader initiated from the intercepting wire, attached to the launcher Ho we v er since the stepped leader is not follo wing a pre-conditioned path, it is essentially free to w ander and may be attracted by upw ard leaders from ground or other grounded structures. Hence, the strik e point is highly unpredictable. As with the classical technique, the return strok e propag ates from ground to cloud along the path of the leader Moreo v er the return strok e tra v els three of four orders of magnitude f aster than the upw ard positi v e leader initiated from the top of the altitude wire and soon catches up with the leader tip and the current w a v e reects from it. Thus, return-strok e current w a v e is only allo wed to tra v el a kilometer or so, which is a fraction of the height of the cloud char ge. Therefore, the current and electromagnetic eld w a v eforms e xhibit a peculiar shape relati v e to those from a classically triggered ash for which the a v ailable channel length is of the order of se v eral kilometers or more. The current w a v eforms, measured at ground, rise sharply to peak and at rst decays normally b ut then decays rapidly gi ving the w a v eforms a “stunted” appearance. In addition, it is lik ely that current from the upw ard positi v e leader from the intercepting wire will not destro y it. The return strok e, ho we v er will destro y an y remaining wire and this results in an interruption in the current ( Rak o v et al. 1998 ). This phenomenon is manifested as a double-peak shape in the current w a v eform. After these processes, an ICC follo ws, similar to that in classical triggered lightning. Se v eral dart or dart-stepped leader/return strok e sequences

PAGE 30

16 Figure 2–7. Unintentional altitude triggered ash FPL-0205, initiated on July 9, 2002 16:26:10 UT The launcher (the intended strik e point on the insulating to wer) is pictured in the fore ground. may occur after the ICC. These leader/return strok e sequences are thought to be similar to those in classically triggered lightning and subsequent strok es in natural lightning. The success rate of altitude triggering is about 10% for the Uni v ersity of Florida group, compared to about 50% for the classical technique under the same thunderstorm conditions. Interestingly the triggering wire occasionally breaks during a classical-triggering attempt resulting in an unintentional altitude trigger In this case, the K e vlar cable is replaced with an air g ap, b ut the ef fect is similar A photograph of such an e v ent is sho wn in Figure 2–7 2.2 Ov er view of the ICLR T F acility The International Center for Lightning Research and T esting (ICLR T) occupies about 0 : 5 km 2 on the National Guard Army base at Camp Blanding, about 5 km east of Stark e, Florida. The f acility w as constructed in 1993 and initially operated by the Electric Po wer Research Institute (EPRI) and Po wer T echnologies Inc. (PTI). In 1994

PAGE 31

17 the Uni v ersity of Florida lightning research group took o v er operation of the site. Figure 2–8 sho ws a sk etch of the ICLR T as of early 2003. As currently congured, the ICLR T contains tw o permanent launching locations. One launcher is located atop an 11 m high w ooden to wer and can hold a battery of twelv e rock ets. The second launcher although currently non-operational due to maintenance needs, is located under ground with the top of the launching tubes ush with the ground and is capable of holding six rock ets. A 70 m 70 m metal mesh surrounds the hole and is attached to the launcher in order approximate an innitely conducting ground plane within the vicinity of the launcher When triggering from the to wer launcher operating personnel are housed in the Launch Control trailer approximately 80 m from the to wer The Launch Control trailer houses computers, video equipment, data acquisition equipment, and the rock et launching system in a centralized location. Pneumatic hoses run from the Launch Control trailer up to the launcher at the top of the to wer and air pressure is used to acti v ate pneumatic relays, located at the base of each rock et tube, which are used to re the rock ets. The use of pneumatics serv es to electrically isolate the rock et launcher from the launch control trailer Experiments using the under ground launcher are conducted from a trailer kno wn as SA TTLIF which stands for SAndia T r ansportable T rig g er ed LIghtning F acility SA TTLIF w as constructed by Sandia National Laboratories in 1990 and has been used in triggered lightning e xperiments at the K ennedy Space Center and F ort McClellan, Alabama, as well as at the ICLR T SA TTLIF has been placed at se v eral locations around the ICLR T site. It has been at its present location since 1999. Similar to the Launch Control trailer SA TTLIF serv es as a centralized location for personnel during rock et triggered lightning e xperiments and pneumatics are used to re the rock ets. In addition, se v eral portable launchers are a v ailable at the ICLR T In 2002, one of these launchers w as mounted to a truck with an e xtendable arm used for servicing

PAGE 32

18 UndergroundLauncher Tower Launcher MilitaryContainer BlastWall Control Launch SATTLIF Test Power Line Access Road Office Marsh IS1 IS2 IS3 IS4 RunwayICLRT FACILITY100 m TestHouseDuPont 3 phase lineN Figure 2–8. Sk etch of the ICLR T at Camp Blanding, Florida.

PAGE 33

19 po wer lines, commonly referred to as a b uck et truck. The launcher capable of holding six rock ets, w as mounted to the end of the e xtendable arm which could be raised to a height of o v er 10 m This mobile launcher can be relocated and recongured for ring in a fe w hours. Ev ery May through September se v eral major -funded e xperiments in v olving triggered lightning are conducted. Furthermore, natural lightning e xperiments are conducted year -round. These e xperiments ha v e in v olv ed direct strik es to a runw ay lighting system, a simulated house, tw o test po wer lines, and high e xplosi v es. In addition, se v eral multiple-station eld measuring e xperiments, for both natural and triggered lightning, ha v e been implemented at the ICLR T and are discussed in the ne xt section. 2.3 Multiple-Station Field Measur ements of Natural and Rock et-T rigger ed Lightning The term Multiple-Station Measur ements refers to a set of measurements from tw o or more sensors placed in tw o or more locations. T ypically the purpose of this type of e xperiment is to e xamine both the measured quantities indi vidually and as a function of location. Furthermore, multiple-station measurements can be used to mak e inferences about the sources of the quantities being measured. A tw o-station eld measurement system with each sensor separated by a certain distance is the simplest implementation of such a netw ork. Ev en a simple netw ork such as this can yield considerable information that a single sensor alone cannot. The rst multiple-station measurements of the electric elds on ground from relati v ely close lightning were performed by W orkman et al. (1942) and Re ynolds and Neill (1955) ( Rak o v and Uman 2003 ). A brief re vie w of this w ork is presented in ( Krehbiel 1981 ). The main goal of these e xperiments w as to obtain char ge solutions for intra-cloud dischar ges and cloud-to-ground return strok es. Numerous tw o and

PAGE 34

20 three-station electric eld measurement systems ha v e been implemented since then, with v arying time resolution and decay time constants. ( Krehbiel 1981 ) e xamined 10 cloud-to-ground, 21 intra-cloud, and three h ybrid ashes between 1976 and 1978 at the K ennedy Space Center The electric eld w as measured at a minimum of 9 and a maximum of 11 locations o v er an area 20 25 km in e xtent. Flat-plate antennas connected to char ge ampliers ha ving decay time constant of 10 seconds were used to sense the electric eld change. In addition, radar w as used for surv eillance of precipitation structure and de v elopment. The data were digitized real-time at 16 kHz and stored on magnetic tape. Since the o v erall electrostatic eld change w as observ ed, and not the ne details of the radiation elds, a lar ge bandwidth w as not required. The analysis of these data pro vided information on the location of the lightning char ge and char ge transfer as a function of time during the dischar ges. In 1997, a multiple-station eld measuring system w as elded at the ICLR T with the e xpressed purpose of measuring the distance dependence of electric and magnetic elds from triggered lightning. The electric and magnetic eld w as measured at se v en locations ranging from 5 to 500 m from the rock et launcher Fiber -optic links were used to transmit the analog data from the antennas to a central location where the y were digitized at or abo v e 10 MHz. A detailed description of the e xperiment is gi v en in ( Cra wford 1998 ). Data for v e triggered lightning strok es were recorded and some results of this analysis are gi v en in ( Cra wford et al. 2001 ). In 1998, another multiple-station e xperiment w as elded at the ICLR T Unlik e the 1997 e xperiment, the purpose w as to record the electric elds produced by rst strok es in natural cloud-to-ground lightning terminating within a kilometer or so of the site. T en electric-eld antennas were distrib uted about the ICLR T site occup ying an area of about 0 : 5 km 2 In addition, tw o pairs of orthogonal magnetic eld sensors were placed at opposite ends of the netw ork. Similar to the 1997 e xperiment, ber -optic

PAGE 35

21 links and digital storage oscilloscopes were used to transmit and record the data. This conguration is described in more detail in ( Cra wford et al. 2001 ). This netw ork operated through 1999 and data for o v er 50 return strok es were recorded within se v eral kilometers of the center of the netw ork. One strok e in particular is belie v ed to ha v e terminated within se v eral tens of meters from one antenna. Some results from this e xperiment are gi v en in ( Rak o v et al. 2003 ). This e xperiment is the direct predecessor to the e xperiment described in this thesis. The measurement locations and sensors remain essentially the same, although the sensor electronics and data acquisition equipment are signicantly dif ferent.

PAGE 36

CHAPTER 3 INSTR UMENT A TION This chapter discusses the instrumentation for the 2001 and 2002 Multiple Station Experiments. Section 3.1 presents an o v ervie w of the e xperiments. Section 3.2 discusses the MSE control system. Section 3.3 discusses the design and implementation of the MSE measurements, including the sensors and associated electronics. Sections 3.4 3.5 3.6 and 3.7 present descriptions of the ber -optic links, digitizers, video equipment, and timing system, respecti v ely used in the MSE. As mentioned in Chapter 1 the 2001 e xperiment w as instrumented by Mr Geor ge Schnetzer formerly of Sandia National Laboratories, and Mr K eith Rambo, of the Department of Electrical and Computer Engineering at the Uni v ersity of Florida. The 2002 e xperiment w as coordinated and instrumented by the author with signicant assistance from the abo v e-mentioned persons. When applicable, the distinction is made between the 2001 and 2002 congurations. 3.1 Ov er view of the 2001 and 2002 Multiple Station Experiments Figure 3–1 sho ws a sk etch of the MSE netw ork at the ICLR T Notable landmarks such as the Launch Control trailer and the of ce trailer are included as reference locations. In 2001, fourteen sensors were spread about ten locations, while in 2002 twenty sensors were spread about ele v en locations. In 2001, only electric elds, magnetic elds, and the optical output from the lightning channel were measured. During the 2002 season the netw ork w as augmented with electric and magnetic eld time-deri v ati v e measurements and current measurements. The measurement locations are referred to as Stations ; hence the term Multiple Station Experiment T en stations spread about the site are numbered 1 through 10 with each station housing one or more eld or eld-deri v ati v e measurements. It should be 22

PAGE 37

23 N E W S 100 m IS2Station 2 Station 6 Station 8 Station 4 Station 10 Station 1 Station 5 Station 9IS4Multiple Station Experiment at ICLRTIS1Access Road SATTLIF Office Trailer Runway Test Power LineDuPont Three Phase Line Access RoadBlast WallTower LauncherIS3Launch SW Optical NE Optical Control Underground Launcher Figure 3–1. Sk etch of the 2001 and 2002 MSE measurement locations at the ICLR T The arro ws roughly indicate the orientation of video cameras and optical sensors at those locations.

PAGE 38

24 noted that Stations 3 and 7 were disassembled prior to 2001 and will not be referred to in this thesis. The numbering scheme w as k ept the same so that there is continuity with pre vious multiple-station e xperiments at the ICLR T (as discussed in Section 2.3 ). Each measurement is designated by the type of measurement and its location. F or e xample, E-1 refers to the electric eld (E) measurement at Station 1 and dE-8 refers to the electric eld time deri v ati v e (dE/dt) measurement at Station 8. The magnetic eld (B) and magnetic eld time-deri v ati v e (dB/dt) measurement designations contain an e xtra parameter which refers to the orientation of the sensor F or e xample, B-4N refers to the magnetic eld measurement at Station 4 with the plane of the sensor (discussed in Section 3.3.2.1 ) oriented to true north. In addition, dB-1E refers to the magnetic eld time-deri v ati v e measurement located at Station 1 with the plane of the sensor oriented perpendicular to true north (i.e. east-west). The tw o optical measurements are designated NEO and SW O which stand for North-East Optical and South-W est Optical respecti v ely The tw o current measurements are designated I-High-T ower and I-Low-T ower where High and Low refer to the relati v e maximum amplitude measured (I-High-T o wer is capable of measuring currents about an order of magnitude higher than I-Lo w-T o wer). A computer program automatically controlled the acti v ation of the sensors as well as the arming of the netw ork and recording of calibration signals. The computer also monitored the battery v oltages of the indi vidual sensors and reported the status of the netw ork to personnel via an on-screen display and email. The program automatically turned on and of f the netw ork based on the output of an electric eld mill, b ut could be manually o v erridden when necessary The analog v oltage w a v eforms from all sensors were transmitted o v er ber -optic links and digitized on digital storage oscilloscopes. Furthermore, the digital storage oscilloscopes were triggered by the simultaneous output of tw o optical sensors vie wing the netw ork from opposite corners. F our -station video co v erage w as emplo yed to help in determining the location of the

PAGE 39

25 T able 3–1. Measured parameters for the 2001 MSE. Location Measured parameters Station 1 E Station 2 E Station 4 E B Station 5 E Station 6 E Station 8 E Station 9 E B Station 10 E North-East Optical Optical South-W est Optical Optical lightning and the geometry of the lightning channel. All computer data-recording and video-recording equipment were housed in the ICLR T Launch Control trailer A simplied diagram of the MSE control and data acquisition system is gi v en in Figure 3–2 T ables 3–1 and 3–2 list the parameters measured in the 2001 and 2002 MSE, respecti v ely As discussed in Section 3.3.2 multiple B and dB/dt measurements were present at some locations to sense dif ferent components of the horizontal eld. T able 3–3 lists the coordinates of the measurement locations. The coordinates were measured in 1999 with a dif ferential GPS unit and the author has not v eried the accurac y of the measurements. Furthermore, the coordinates of the South-W est Optical sensor were not measured. 3.2 Contr ol System Operation of numerous sensors spread about a lar ge ph ysical area poses se v eral logistical problems. These can only be o v ercome by a rob ust control system which automates as man y operations as possible. These logistical problems are outlined belo w Since the occurrence of thunderstorms can be unpredictable, manually acti v ation of the netw ork w ould require personnel to be on-site virtually all of the time, which

PAGE 40

26 Fiberoptic Receivers Trigger to all Digitizers Camera Video Control PC Control Software Activates / Deactivates Sensors Flatplate Sensor Digitizers Digitizers Sensor Loop Optical SensorsLAUNCH CONTROL TRAILERFiberoptic Cables Video Recorder Digital Arms / Disarms Control Software Electric Field Mill Figure 3–2. Simplied diagram of the MSE control and data acquisition system.

PAGE 41

27 T able 3–2. Measured parameters for the 2002 MSE. Location Measured parameters Station 1 dE/dt dB/dt Station 2 E Station 4 E B dE/dt dB/dt Station 5 E Station 6 E Station 8 dE/dt Station 9 E B dE/dt dB/dt Station 10 E North-East Optical Optical South-W est Optical Optical Launch T o wer Current T able 3–3. Coordinates of the MSE measurement locations. Location Coordinates Station 1 29.94390325 N 82.03451849 W Station 2 29.94404460 N 82.03497685 W Station 4 29.94329562 N 82.02931170 W Station 5 29.94323225 N 82.03248203 W Station 6 29.94160838 N 82.03561998 W Station 8 29.94145723 N 82.03053364 W Station 9 29.94026821 N 82.03396553 W Station 10 29.94068532 N 82.02925378 W North-East Optical 29.9440 N 82.02949 W Launch T o wer 29.94262236 N 82.03185467 W is not possible September through April, when rock et triggered lightning operations ha v e ceased at the ICLR T Since the electronics associated with each sensor (e.g. ber -optic transmitters and ampliers) are po wered by a battery the netw ork cannot be left acti v ated at all times; otherwise the batteries w ould drain within a fe w days. Furthermore, due the ph ysically lar ge size of the netw ork and the lar ge number of sensors, manual acti v ation of all of the sensors when a thunderstorm is present is infeasible. In addition, it is infeasible to monitor battery v oltages and the general health of the netw ork manually for e xtended periods of time. Therefore, in order to operate with minimum user attendance and maintenance, the MSE control system must fulll the follo wing requirements.

PAGE 42

28 The netw ork must be able to be acti v ated and deacti v ated automatically without an y user interaction, when appropriate thunderstorm conditions are present. Furthermore, the system must be able to automatically determine when these conditions are present based on predetermined criteria. Acti v ation includes turning on all of the sensors, calibrating all of the ber -optic links, and arming all of the digitizers. The control system must be able to gi v e instant feedback re g arding the status of the entire netw ork, including battery v oltages, calibration signals, and triggering status. Furthermore, all of this information must be a v ailable to both on-site and remote personnel. All data must be automatically recorded on non-v olatile media upon a system trigger since it is highly lik ely that no personnel will be on-site during the e v ent. The MSE control system consists of both hardw are and softw are components. The primary piece of control hardw are is a de vice kno wn as a PIC Controller A PIC controller w as placed in the eld with each of the sensors and the control softw are interacted with the PIC controllers in order to automatically acti v ate and deacti v ate the netw ork when necessary as well as monitored the status of the netw ork. 3.2.1 The PIC Contr oller The PIC controller is the foundation of the MSE control system. This de vice is used to remotely acti v ate each of the measurements, pro vide attenuation to each of the sensors, to check the status of each measurements battery and to calibrate the ber -optic link associated with each measurement. The PIC controller w as designed and de v eloped by Michael Stapleton, a project engineer and K eith Rambo. The de vice is kno wn as a PIC controller because it contains a PIC 16F873-207SP microprocessor The term PIC controller is a some what general term, since se v eral v arieties of PIC controllers ha v e been de v eloped for dif ferent applications at the ICLR T such as the automation of video recording. Unless otherwise noted, the term PIC controller refers to the model which is used to accompan y each sensor within the netw ork.

PAGE 43

29 Battery ConnectorLCD Display Power Connector Connectors Plastic Fiber HexadecimalSwitches Electronics for Measurement Connector BNC Connectors AB Figure 3–3. The PIC controller A) Front vie w B) Side vie w Figure 3–3 sho ws a picture of a PIC controller Each PIC controller has tw o female BNC connectors, a four -pin male microphone connector an Agilent HFBR-1523 ber -optic transmitter an Agilent HFBR-2523 ber -optic recei v er a DB-9 female serial connector and a tw o-wire po wer connector Each PIC controller is assigned a one-byte (8-bit) he xadecimal address, which is set by adjusting a pair of he xadecimal switches located ne xt to the DB-9 connector Each switch is capable of being set from 0x0 (0 decimal) to 0xF (15 decimal). Hence the range of addresses ranges from 0x00 (0 decimal) to 0xFF (255 decimal), with certain addresses reserv ed for special functions. Furthermore, each PIC controller is assigned a w ork group, which is programmed directly into the microprocessor itself and can only be changed by reprogramming the chip. Currently only tw o w ork groups

PAGE 44

30 BulkheadFeedthroughBNC Connector Cable Coaxial nnnnnn OUTIN Inline Terminator Data Fiber 12 V Battery Plastic Control Fiber Metal Enclosure PIC Controller Power to the Battery Supplies Supplies Power to PIC ControllerTransmitter the Fiberoptic FiberopticTransmitter PIC Controller Sensor Figure 3–4. Diagram of ho w a PIC controller is installed with a measurement. e xist, R TL (standing for Roc k et T rig g er ed Lightning ) and CAM (standing for CAMer a ), with the PIC controllers in the R TL w ork group being used to control the sensors and the ones in the CAM w ork group being used to control cameras. Currently no CAM w ork group PIC controllers are used in the MSE. The combination of a PIC controller' s assigned w ork group and he xadecimal address gi v es each PIC controller a unique identier Figure 3–4 sho ws a diagram of ho w a PIC controller is installed with a measurement. The output of the sensor is connected to the IN BNC connector while the OUT BNC connector is connected to the input of the ber -optic transmitter terminated in 50 W The po wer connector is connected to a 12 V battery while the po wer input of the ber -optic transmitter is connected to the microphone connector Pins 1 and 2 (ground) and 3 and 4 ( + 12 V) of the microphone connector are soldered together ef fecti v ely making it a 2-pin connector The female DB-9 connector is used to connect a tw o-line LCD display that can be used to monitor the status of the PIC controller in the eld.

PAGE 45

31 The PIC controller is essentially a combination of relays and attenuators controlled by a microprocessor Therefore the PIC controller can be described as a series of programmable switches and attenuators. Ho w the PIC controller beha v es is determined by the commands sent to it. The PIC controller communication scheme and command set is described in the follo wing sections. The PIC controller itself is po wered by a 12 V battery Ho we v er a relay inside is used to supply po wer to other electronics, which at a minimum includes a ber -optic transmitter via a cable and the microphone connector Hence, battery life can be conserv ed by simply disconnecting po wer from the electronics. Ho we v er the PIC controller must al w ays be po wered and dra ws a current of about a fe w tens of milliamperes at 12 V. When po wered by a 12 V, 24 Ah, battery the maximum battery life is se v eral weeks. The PIC controller is placed in series with the measurement between the sensor and the ber -optic transmitter via the IN and OUT BNC connectors and short lengths 50 W coaxial cable. The function of the PIC controller depends on what command is sent to it. First, the PIC controller can act as a 50 W in-line attenuator which reduces the output v oltage of the sensor increasing the full-scale range of the measurement. If no attenuation is set, the PIC controller has a g ain of 1 (0 dB) and does not af fect the measurement. The attenuators are resisti v e PI attenuators of v alues -3 dB, -6 dB, -10 dB, -14 dB, and -20 dB, which can be added in an y combination by sending the appropriate commands to the PIC controller The PIC controller adds the attenuators by switching the appropriate relays inside of the de vice. The attenuators are designed to be terminated in 50 W and will not pro vide the stated v oltage di vision if the output of the PIC controller is not terminated in 50 W The PIC controller has an input resistance of 50 W when the output of the PIC controller is terminated in 50 W re g ardless of the attenuation setting. If the PIC controller is terminated in a dif ferent resistance, the input resistance of the PIC controller will be that resistance only if no attenuation is used. While it is possible to use the attenuators when the PIC controller is not

PAGE 46

32 terminated in 50 W by calculating the actual v oltage ratio based on the attenuator circuit and a dif ferent load, this has ne v er been done and is not recommended due to the high possibility of confusion. The ability to remotely set the amount of attenuation used in a measurement eliminates the need for personnel to manually e xchange BNC in-line attenuators e v ery time an adjustment is required. In addition, the PIC controller can act as a w a v eform generator that calibrates the ber -optic link. When the appropriate command is sent to the PIC controller a relay inside the de vice disconnects the sensor from the ber -optic transmitter and injects a calibration w a v eform into the transmitter The calibration w a v eform is a 100 Hz square w a v e with a selectable peak-to-peak v oltage of 1 V or 0 : 1 V, when the PIC controller is terminated in 50 W When terminated in high impedance, the v oltage doubles to 2 V and 0 : 2 V respecti v ely The calibration w a v eform can be used in either mode as long as the operator is a w are of the ef fect of the dif ferent terminations. In addition, it is possible to also attenuate the calibration signal using the attenuation function, although this has ne v er been done. The calibration w a v eform is useful for estimating the g ain of the ber -optic link, assessing the amount of non-linear distortion present in the link, and determining whether the link is damaged. This eliminates the need for personnel to manually inject a calibration w a v eform into the link, a process that w ould be tedious and time-consuming in a lar ge e xperiment such as the MSE. The Agilent HFBR-1523/2523 ber -optic transcei v er pair (660 nm LED) is used to communicate with the PIC controller via 1 mm diameter optical plastic ber with snap-action connectors, when it is installed in the eld. The maximum length of ber that can be used, without the addition of an optical repeater is approximately 110 m at a maximum data rate of 40 kbps. The ber -optic transmitter con v erts serial TTL (0 to + 5 V) logic to pulses of light, while the recei v er does the re v erse con v ersion. No light corresponds to logical 0 while a pulse of light corresponds to logical 1. A standard PC can communicate with a PIC controller via its RS232 serial port and a ber -optic

PAGE 47

33 transcei v er pair by utilizing a RS-232 dri v er/recei v er such as a Maxim MAX232 chip, which con v erts RS-232 serial logic ( 12 V) to TTL serial logic. A PIC controller cannot send and recei v e data simultaneously and therefore the communication link is half-duple x. The method by which the control PC communicated with the PIC controllers dif fered between the 2001 and 2002 seasons. In 2001, the RS232 serial port of the PC w as interf aced with a MAX232 chip to translate the RS232 serial logic to TTL serial logic, as described abo v e. The output of the MAX232 chip w as connected directly to an Agilent HFBR-1523/2523 ber -optic transcei v er pair mounted on the same circuit board. One end of a plastic ber w as connected to the transmitter and the other w as connected to a repeater whose output branched out to three additional plastic bers. Each of these bers went into the eld, forming three loops which each e v entually returned to the Launch Control trailer At each measurement location, an incoming plastic ber w ould be connected to the ber -optic recei v er of the PIC controller and another plastic ber going to the ne xt measurement w ould be connected to the transmitter In this topology the PIC controllers also acted as ber -optic repeaters. Ho we v er in cases where the distance between successi v e measurements in the loop w as greater than about 100 m, additional repeaters were required. The plastic bers connected to the transmitter of the last PIC in each loop went back to the Launch Control trailer where the y were all connected to another repeater The output of this last repeater w as connected to the ber -optic recei v er of the MAX232 board which w as then connected to the control PC via the RS232 link. A diagram of this topology is sho wn in Figure 3–5 This method of communication posed se v eral problems, which are outlined belo w If an y PIC controller in a gi v en loop malfunctioned, e v ery measurement in the loop w as lost since each PIC controller acted as a repeater

PAGE 48

34 MAX232 PIC Repeater PIC PIC Repeater PIC PIC PIC PIC PIC PIC Repeater Control PC RS232Connection Loop 1 Loop 2 Loop 3 Plastic Fiber Figure 3–5. Diagram of the PIC controller communication topology used during the 2001 MSE. If an y plastic ber in a gi v en loop w as damaged, e v ery measurement in the loop w as lost since the series data path w as opened. Furthermore, these tw o issues were especially troublesome since nding the location of a malfunctioning PIC controller or a damaged ber w as a dif cult and tedious process. During the 2002 season these limitations were addressed by replacing the ber loops and repeaters with wireless 900 MHz RF links. These wireless RF units are kno wn as PIC RF units, since the y contain the same PIC 16F873-207SP microprocessor as the PIC controllers. The RF transcei v er itself is a MaxStream 9XStream-96 900 MHz transcei v er (902 928 MHz, unlicensed ISM band) capable of a maximum data rate of 9600 bps. Instead of running lengths of plastic ber to each PIC controller in the eld, PIC RF units were placed about ten meters a w ay from each measurement location. Each PIC RF unit w as equipped with an Agilent HFBR-1523/2523 ber -optic transcei v er pair and tw o short lengths of plastic ber

PAGE 49

35 PIC Control PC RS232Connection Transceiver Wireless PIC PIC PIC PIC PIC Plastic Fiber Antenna PIC RF PIC RF PIC RF PIC RF PIC RF PIC RF Figure 3–6. Diagram of the PIC controller communication topology used during the 2002 MSE. connect it to a PIC controller An RF transcei v er w as installed in the Launch Control trailer which w as interf aced to the control PC via the RS232 serial port. The RF transcei v er w as capable of directly encoding and transmitting RS232 serial logic, so no MAX232 chip w as required. The RF transcei v er transmitted the commands from the control PC and the PIC RF unit at each measurement location recei v ed and decoded the signal. The decoded data are then transmitted to the PIC controller o v er one of the lengths of plastic ber Similarly the PIC controller transmitted data back to the PIC RF unit o v er the other length of plastic ber The PIC RF unit then encoded and transmitted this data back to the Launch Control trailer where the transcei v er decoded the data and transmitted it to the control PC o v er the RS232 interf ace. Hence, the PIC controller communication scheme changed from a ring topology to a star -netw ork topology as sho wn in Figure 3–6 with the PIC RF units essentially acting as wireless repeaters.

PAGE 50

36 Multiple PIC controllers in an area could be interf aced with a single RF PIC unit by using an optical repeater A special repeater board w as designed for this purpose and can interf ace nine PIC controllers to a single PIC RF unit. The star -netw ork topology w as a signicant impro v ement o v er the ring topology used in 2001. No w if a single PIC controller or plastic ber f ails, only the measurement connected to that PIC controller f ails, instead of the entire loop, as before. This conguration made diagnosing system component problems much easier In addition, since the lengths of plastic ber were, at most, some tens of meters, the amount of plastic ber that could be damaged w as minimal. In this topology the PIC controllers themselv es no longer act as repeaters. Therefore, a simple hardw are modication w as required to remo v e this function. If this hardw are modication were not performed, a PIC controller w ould immediately repeat an y data sent to it back to the PIC RF unit that w ould then transmit it. If another PIC controller without the hardw are modication were present, it w ould recei v e this data and repeat it ag ain. Hence, this repeating could become an innite loop which making communication with the PIC controllers impossible since some PIC RF unit w ould al w ays trying to transmit, which essentially jams the transmitter at the control PC. The PIC RF units are electrically isolated from the lightning measurements to eliminate an y possibility of data contamination or equipment damage from lightning electromagnetic interference (EMI). The PIC RF units were placed in separate metal enclosures (modied ammunition containers) with the only ph ysical connection between the PIC RF unit and the measurements being the pairs of plastic bers. Each metal enclosure contained a PIC RF unit, a 12 V battery and optionally a repeater f an-out board for connecting multiple PIC controllers. The PIC RF unit itself w as mounted to the inside co v er of the box with a 15 cm antenna mounted to the outside of the co v er A 900 MHz tuning stub (shorted length of transmission line) w as placed between the antenna output and the ground of the wireless PIC controller At

PAGE 51

37 900 MHz, the tuning stub resonates and becomes a v ery high impedance b ut, at other frequencies, it looks lik e a short circuit to ground. Since the signicant EM radiation from close lightning occurs from DC to 10 MHz ( Uman 1987 ), an y induced current in the antenna from lightning should be directed to the ground of the PIC RF unit and not interfere with its operation. The PIC RF units will operate for a fe w days when po wered with a 7 Ah 12 V battery Fi v e-w att solar cells (450 mA short-circuit current) were installed at each PIC RF unit location in order to k eep the batteries char ged. The solar panels were mounted on top of a one-meter long 4x4 piece of w ood e xtending v ertically from the ground. The PIC RF unit enclosure w as hung from a pair of half-inch bolts scre wed into the w ood. The solar cell serv ed to shade the enclosure in this conguration. The solar cell itself w as encased in a metal “chick en wire” mesh with a hole-diameter of about v e centimeters to protect it from the induced ef fects of nearby lightning. A test cell w as placed in the eld in early June 2002 without this protection, and it is belie v ed that the electromagnetic signal from a close lightning strik e damaged the cell, rendering it useless. A tw o-wire po wer cable w as run from the solar cell into the PIC RF unit enclosure and connected to the 12 V battery This cable w as encased in metal shield braid with one end soldered to the chick en wire and the other end securely attached to the hole in the metal enclosure with a hose clamp. The plastic bers were run out of tw o additional holes in the enclosure. These holes were tted with PVC elbo w connectors with the openings pointed to w ard the ground so that no w ater could enter the enclosure. Figure 3–7 sho ws a PIC RF unit enclosure mounted with a solar cell. All command data pack ets sent to a PIC controller are formatted into a v e-byte (40-bit) array Each byte in the array can be represented as a character as dened by the American Standard Code for Information Interchange (ASCII). Hence, each data pack et can be represented as an ASCII character string, which is easy for humans to interpret. T able 3–4 describes the format of a PIC controller command data pack et.

PAGE 52

38 Metal BoxPIC RF Unit Containing Plastic Fiber Antenna Solar Cell Enclosed in "Chicken Wire" Mesh Figure 3–7. PIC RF unit enclosure mounted with a solar cell. T able 3–4. F ormat of a PIC controller command data pack et. Bytes 1-3 (bits 0-23) Byte 4 (bits 24-31) Byte 5 (bits 32-39) W ork group identication Address Command Communication with a PIC controller is implemented in a command/response format. What this means is that when a command is issued to a PIC controller the PIC controller performs the command, and then the PIC controller sends a response. The response the PIC controller sends is the last command data pack et that it recei v ed, which is stored in a b uf fer along with v e additional bytes corresponding to its battery v oltage and tw o bytes corresponding to its temperature. T able 3–5 describes the format of a PIC controller response data pack et. T able 3–5. F ormat of a PIC controller response data pack et Bytes 1-3 Byte 4 Byte 5 Bytes 6-10 Bytes 9-10 (bits 0-23) (bits 24-31) (bits 32-39) (bits 40-79) (bits 80-95) W ork group He xadecimal Last Battery T emperature identication address command v oltage

PAGE 53

39 Bytes 1-3 (bits 0-23) are the w ork group identication, such as R TL or CAM as discussed abo v e. These three bytes are interpreted as ASCII characters so that, for e xample, the byte array 0x52 0x54 0x4C translates into the character string R TL. A PIC controller will not respond to a command unless its w ork group identication matches the w ork group identication of the command. F or e xample, PIC controller 0x0E in the R TL w ork group will not ackno wledge a command designated for PIC 0x0E in the CAM w ork group. The address of the PIC controller is gi v en in byte four (bits 24-31). Therefore, each PIC controller can be communicated with indi vidually Address 0xFF is a global address, meaning that e v ery PIC will perform a command issued to address 0xFF re g ardless of its address setting. The w ork group constraint still applies, which means that 0xFF is only global within the scope of the w ork group. A PIC controller in the CAM w ork group will not respond to the 0xFF address of the R TL w ork group and vice v ersa. This is useful for sending a command to e v ery PIC controller simultaneously where sending the command to each PIC controller indi vidually w ould be time consuming. The dra wback of the global address is that none of the PIC controllers will send a response after performing their assigned command because there is no w ay the control PC issuing the command could understand multiple simultaneous responses. Hence, there is no conrmation that the command w as recei v ed. Byte v e (bits 32-39) designates the command issued to the PIC controller Each bit in character v e designates a certain function of the PIC, as sho wn in T able 3–6 The PIC controller can be told to perform multiple functions by setting specic bits in the command byte. F or e xample, the simplest command is the byte 00000000 (0x00), corresponding to ASCII character NULL which turns of f po wer to the electronics, turns of f all attenuators, and turns of f calibration signals. This is essentially a reset command. Moreo v er the byte 10100000 (0xA0), corresponding to ASCII character tells the PIC controller to pro vide po wer to the electronics and also

PAGE 54

40 T able 3–6. Bit settings of PIC controller commands. Bit Name Function 0 3 dB Pro vide -3 dB of attenuation 1 6 dB Pro vide -6 dB of attenuation 2 10 dB Pro vide -10 dB of attenuation 3 14 dB Pro vide -14 dB of attenuation 4 20 dB Pro vide -20 dB of attenuation 5 Cal ON T urn on calibration signal 6 Cal Choose between 1.0 (0) or V oltage 0 : 1 V (1) calibration signal 7 Po wer Pro vide po wer to electronics pro vide a 1 V calibration signal. This calibration signal could be attenuated by -13 dB, for e xample, by setting bits 0 and 2, which w ould change the command byte to 10100101 (0xA5) corresponding to ASCII character Some command combinations are essentially meaningless, such as 00010000 (0x10) corresponding to ASCII character Data link escape which tells the PIC controller to pro vide -20 dB of attenuation while not supplying po wer to the electronics. The byte 01111110 (0x7E), corresponding to ASCII character ~ performs a special function. While the literal interpretation of the bits is meaningless (no po wer to electronics, 0 : 1 V calibration signal, -50 dB of attenuation), the PIC controller interprets this command to be send a status response without performing an y function. Hence, the current status of the PIC controller including last command issued, battery v oltage, and temperature can be assessed without changing the current status. This command can also be used to check the status of an y wireless PIC controller The v e bytes corresponding to the battery v oltage are interpreted as ASCII characters. F or e xample, the v e-byte array 0x31, 0x32, 0x33, 0x34, 0x35 w ould be interpreted as ASCII characters 1 2 3 4 and 5 which indicates a battery v oltage of 12 : 345 V. Lik e wise, the tw o bytes corresponding to the ambient temperature are interpreted as ASCII characters. F or e xample, the tw o bytes 0x33 and 0x34 w ould be interpreted as ASCII characters 3 and 4 which indicates a temperature of 34 C. The

PAGE 55

41 PIC response is used to v erify that the correct PIC controller indeed recei v ed the proper command, as well as to assess the amount of battery life remaining and its temperature. 3.2.2 Softwar e The control softw are is the brain of the MSE and is responsible for the automation of the MSE. Using a sophisticated control algorithm, the softw are automatically acti v ates the sensors and arms the w a v eform digitizers when appropriate thunderstorm conditions are present. Furthermore, when the storm has passed, the softw are deacti v ates the sensors and disarms the digitizers. In addition, the softw are pro vides instant feedback to the user re g arding the status of the netw ork. Finally the softw are calibrates the ber -optic data links each time the netw ork is acti v ated and deacti v ated. The rst v ersion of the MSE control softw are, used during the 2001 season, is kno wn as LCA UT O (standing for Launc h Contr ol A UT Omation ) and w as written by Alonso Guarisma, at the time a graduate student in the Department of Electrical and Computer Engineering at the Uni v ersity of Florida LCA UT O w as written in C++ with the graphical user interf ace (GUI) written in Perl TK. LCA UT O ran on a He wlett P ackard P a vilion 6746C PC with a 733 MHz Intel Celeron processor 192 MB of memory and 30 GB of hard disk space running Red Hat Linux 6.2 with the 2.2 k ernel. The user interf ace allo wed personnel to manually query and send commands to indi vidual PIC controllers. F or e xample, the softw are could be used to command a specic PIC controller to supply a calibration signal so that the g ain of that ber -optic link could be check ed. In addition, a user could acti v ate or deacti v ate all of the MSE PIC controllers with the click of a single b utton. The most important feature of the LCA UT O softw are w as the ability to automatically acti v ate and deacti v ate the netw ork when appropriate thunderstorm conditions were present. The thunderstorm conditions were assessed by digitizing the output of an electric eld mill (a de vice to measure the ambient v ertical electric eld at ground with a bandwidth from DC to a fe w hundred hertz). An Adv antec data acquisition

PAGE 56

42 card, housed in a separate PC running W indo ws 95 digitized the output of eld mill at 100 Hz. Softw are written in GENIE (a scripting language by Adv antec that is similar to National Instruments LabV ie w) continuously monitored the digitized eld mill data and when the absolute magnitude of the v ertical quasi-static electric eld at ground e xceeded 2 kV m 1 a true ag w as written to a specic le on the LCA UT O PC. The LCA UT O softw are continuously monitored this le and when the true ag w as found, all of the MSE measurements were acti v ated. The list of MSE measurements w as k ept in a MySQL database, also written by Mr Guarisima. Each entry in the database w as composed of a measurement name, a PIC address, and an attenuation setting. The LCA UT O softw are interf aced with the database to obtain the proper settings for each of the MSE measurements. When the magnitude of the eld mill output fell belo w 2 kV m 1 for a period of 10 minutes, a f alse ag w as written to the le and the LCA UT O softw are deacti v ated all of the measurements by sending the reset command to all of the PIC controllers in the database. The LCA UT O softw are w as not capable of communicating with the Y ok og a w a DL716 w a v eform digitizer (discussed in Section 3.5.1 ); hence it w as the operator' s responsibility to mak e sure the Y ok og a w a w as properly set up and armed at all times. This digitizer w as capable of rearming itself after a trigger so this did not present a problem unless it w as accidentally disarmed. In addition, the LCA UT O softw are w as not capable of automatically acquiring calibration signals. Although the softw are could command the PIC controllers to send calibration w a v eforms, it could not tell the digitizer to acquire the w a v eforms. Hence, it w as the operators responsibility to mak e sure calibration signals were acquired, which w as especially important after a trigger since it is crucial to kno w the g ain of the ber -optic links (which can v ary as a function of temperature depending on the type of link) as close to the time of the trigger as possible. Ob viously this w ould not be possible if the trigger occurred when there were no personnel on site. In this case, the g ain of the ber -optic link w ould

PAGE 57

43 ha v e to be assumed to be some v alue, which is usually one. Depending on the type of ber -optic link used, this could introduce an ywhere from a fe w percent to o v er twenty percent error in the calibration, since the calibration is partially determined by the g ain of the ber -optic link. Also, the LCA UT O softw are w as capable of sending email at re gular interv als, such as e v ery 24 hours, that w ould gi v e the status of all of the measurement batteries in the netw ork. Therefore, a remote user could easily be informed of the status of the netw ork. During the summer of 2002, a ne w v ersion of the MSE control softw are w as de v eloped, that is kno wn as Mer cury This softw are w as written in National Instruments LabV ie w softw are by Robert Olsen III, presently a graduate student, and Alonso Guarisma. The Mer cury softw are ran on a tw o-processor 1 : 6 GHz PC with 1.5 GB of memory and 75 GB of hard disk space running Red Hat Linux 7.3 with the 2.4.18 k ernel. The ne w Mer cury softw are implemented all of the features of the LCA UT O softw are, such as a user interf ace where the status of measurements could be assessed indi vidually or in a group, along with some additional features. The Mercury softw are w as capable of interf acing with the digitizers o v er Ethernet and IEEE 488.2 (GPIB) interf aces, and therefore the digitizers could be automatically set up, armed, and disarmed along with the PIC controllers in the e xperiment. Furthermore, calibration signals could be automatically acquired, which minimized the time between a system trigger and the acquisition of calibration w a v eforms. Unlik e the 2001 season, the data acquisition card w as housed in the same PC as the control softw are. Additionally a National Instruments data acquisition card replaced the Adv antec card used in 2001. Dri v ers pro vided by the Comedi Pr oject were used to communicate with the card in LabV ie w for Linux.

PAGE 58

44 The Mer cury softw are could send email re g arding the status of the netw ork, as the LCA UT O softw are did. Ho we v er the Mer cury softw are also included a list of times when the netw ork w as acti v ated and deacti v ated in the email, although it did not include the times of system triggers. In addition, the settings for the MSE PIC controllers were no longer stored in a database, b ut in a set of ASCII te xt les. The conditions Mer cury used to decide whether to acti v ate and deacti v ate the netw ork were similar to that of the LCA UT O softw are, although the y were rened some what to minimize the amount of time the netw ork w as acti v ated unnecessarily The digitized output of the eld mill, sampled at 100 Hz, w as once ag ain used to determine whether appropriate thunderstorm conditions were present. These conditions e v olv ed o v er the course of the 2002 season and this e v olution is outlined belo w The netw ork w as armed if the magnitude of the output of the electric eld mill e xceeded 2 kV m 1 The output of the eld mill w as check ed e v ery 10 minutes and the magnitude of the output e v er fell belo w 2 kV m 1 the netw ork w as disarmed. Unfortunately a momentary glitch in the output of the eld mill or input of the data acquisition card w ould cause the system to arm unnecessarily The netw ork w as armed if the magnitude of the output of the electric eld mill e xceeded 2 kV m 1 continuously for tw o seconds. Ag ain, the output of the eld mill w as check ed e v ery 10 minutes and the magnitude of the output e v er fell belo w 2 kV m 1 the netw ork w as disarmed. This eliminated the problems due to glitches. Ho we v er the netw ork tended to remain armed for e xtended periods of time (sometimes hours) during the dissipating phase of thunderstorms, where the quasi-static electric eld at ground lingered at + 2 to 4 kV m 1 Although it w as desirable to acquire data from positi v e lightning, the chance of a positi v e lightning occurring w as belie v ed to be slim in such lo w elds. The netw ork w as armed if the output of the electric eld mill e xceeded + 4 kV m 1 or fell belo w 2 kV m 1 continuously for tw o seconds. After arming, the output of the eld mill w as continuously check ed and if it e v er went between 2 kV m 1 and + 4 kV m 1 a 10 minute countdo wn started. If, at an y point during the 10 minute countdo wn, the output fell belo w 2 kV m 1 or rose abo v e + 4 kV m 1 the 10 minute countdo wn w as reset. If, after 10 minutes, the output ne v er fell belo w 2 kV m 1 or rose abo v e + 4 kV m 1 the netw ork w as disarmed.

PAGE 59

45 The netw ork w as armed if the output of the electric eld mill e xceeded + 4 kV m 1 or fell belo w 2 kV m 1 Instead of requiring the eld mill output to e xceed the threshold continuously for tw o seconds, short-term a v eraging w as used to eliminate random glitches. After arming, the output of the eld mill w as continuously check ed and if it e v er went between 2 kV m 1 and + 4 kV m 1 a 10 minute countdo wn started. If, at an y point during the 10 minute countdo wn, the output fell belo w 2 kV m 1 or rose abo v e + 4 kV m 1 the 10 minute countdo wn w as reset. If, after 10 minutes, the output ne v er fell belo w 2 kV m 1 or rose abo v e + 4 kV m 1 the netw ork w as disarmed. This algorithm is sho wn in a o wchart in Figure 3–8 In addition, a calibration w a v eform w as acquired before each time the netw ork w as acti v ated and after each time the netw ork w as deacti v ated. 3.2.3 T riggering System The MSE w as intended to acquire data on close natural lightning, specically within the 0 : 5 km 2 ICLR T site. Therefore, a triggering system needed to be designed such that data were recorded when lightning occurred within the netw ork, b ut not when lightning occurred outside of the netw ork. This requirement is not tri vial and poses se v eral logistical problems. T ypically in triggered lightning e xperiments, the location of the channel is kno wn e xactly and the trigger threshold of the digitizers can be set based upon the distance to the channel of the sensor which triggers the digitizer and kno wn statistics of that particular ph ysical phenomenon. F or e xample, if the digitizers are to be triggered from an electric eld sensor 30 m from the lightning channel, an appropriate trigger threshold can be calculated based upon statistics of kno wn triggered-lightning electric elds at 30 m Furthermore, this technique can also be emplo yed if data on distant natural lightning are to be recorded. F or e xample, if data on natural lightning in a storm 10 km a w ay are to be recorded, the trigger threshold can be adjusted based on statistics on natural lightning at 10 km. Although no one ash will be e xactly 10 km a w ay from the sensor the a v erage lightning distance will be and this technique will w ork reasonably well, b ut not as well as for triggered lightning.

PAGE 60

46 START RecordCalibrationSignal RecordCalibrationSignal Digitizers Arm ActivateSensors DeactivateSensors Field Above Threshold ? Field Above Threshold ? Start 10MinuteCountdown Y N Y N N Passed? 10 Minutes Y Digitizers Disarm Figure 3–8. Flo wchart representation of the nal 2002 MSE softw are control algorithm.

PAGE 61

47 Unfortunately it is dif cult to emplo y this technique to e xamine close natural lightning since tw o v ariables e xist, distance and amplitude. The distance to the lightning channel from an y one sensor is unkno wn and cannot be assumed to be an y distance e xcept less than about 1 km. Furthermore, ( Ber ger et al. 1975 ) ha v e measured a range of peak currents from belo w 2 kA to abo v e 80 kA for ne g ati v e cloud-to-ground rst strok es and from belo w 4 : 6 kA to abo v e 250 kA for positi v e cloud-to-ground rst strok es. Therefore, with such a lar ge range of peak currents, a close lightning with relati v ely lo w peak current within the netw ork cannot be distinguished from a lightning outside of the netw ork with relati v ely high peak current based upon the amplitudes of the electric and magnetic elds alone since the eld amplitudes are a function of the peak current. In addition, the distance dependence of the electromagnetic elds from rst strok es in natural lightning is not kno wn. Since the magnitudes of the electric and magnetic elds from natural lightning are a function of both distance and current amplitude and w a v e-shape, a triggering technique w as required that can decouple the tw o dependencies. While a single electric or magnetic eld antenna is incapable of this, an acceptable solution can be obtained by utilizing tw o or more directional sensors. A directional sensor is one whose sensiti vity is a function of the azimuthal or ele v ation angle of the sensor relati v e to the source, such as a magnetic eld antenna made of a loop of coaxial cable. The magnitude of the output of the antenna is not only a function of the magnitude of the magnetic eld, b ut a function of the angle of the magnetic ux to the plane of the loop as well. Hence the output of a loop antenna is a function of the angle of the lightning channel to the plane of the loop. Although the output of a loop antenna is a function of direction, the direction information (angle to the lightning channel with respect to the plane of the loop) cannot be e xtracted from the output of a single antenna. In addition, there is a 180 ambiguity in the output of the antenna since the polarity of the magnetic eld produced by a ne g ati v e current is the same as

PAGE 62

48 that produced by a positi v e current 180 a w ay Adding a second antenna, oriented perpendicularly to the rst, will form a crossed-loop antenna pair which can pro vide directional information. Con v ersely a at-plate electric eld antenna laying ush with the ground is non-directional, ha ving rotational symmetry with respect to the normal of the plate, and does not pro vide an y direction information although it can be used to pro vide information needed to resolv e the 180 ambiguity in the crossed-loop antenna output. Detailed descriptions of the electric and magnetic eld sensors are gi v en in Sections 3.3.1 and 3.3.2 respecti v ely If tw o or more crossed-loop antennas are distrib uted about the netw ork, an algorithm could be de v eloped to determine whether a lightning is within the netw ork based on the output of these antennas. This type of triggering scheme w as emplo yed in the pre vious ICLR T multiple station e xperiment discussed in ( Cra wford et al. 2001 ) and ( Rak o v et al. 2003 ). Ho we v er this triggering scheme requires considerable resources since tw o antennas and tw o ber -optic links are required for each crossed loop pair A minimum of four loop antennas and four ber -optic links w ould be required to implement this triggering scheme. In addition, this triggering scheme could be fooled by horizontal channel sections or slanted channels. In vie w of the abo v e, in 2001 a triggering scheme w as emplo yed using tw o optical sensors. The principle behind this scheme and the crossed magnetic eld antennas is essentially the same; multiple directional measurements are used to dif ferentiate between lightning within and outside of the netw ork. The adv antage of using the optical sensors, ho we v er is that only a single sensor (and hence only a single ber -optic link) is required for each directional measurement. The optical sensors are directional because the light source must be in front of the lens of the detector for there to be an y output from the sensor Furthermore, the implementation of a simple optical detector is considerably less complicated than that of a crossed loop pair of magnetic eld antennas. A detailed description of the optical sensors is gi v en in Section 3.3.3

PAGE 63

49 One of the optical detectors w as placed at the north-east corner of the site, upon a twelv e foot high structure kno wn as the Blast W all f acing south-west. The second optical detector w as placed at the south-west corner of the site, upon a six foot high structure kno wn as the Military Container f acing north-east. The output of each optical systems ber -optic recei v er w as then connected into an AND triggering circuit, designed by Geor ge Schnetzer during the summer of 2001. Figure 3–9 sho ws a schematic of the triggering circuit. Each input is fed into a non-in v erting b uf fer circuit. The output of each b uf fer is fed into a bistable (Schmitt trigger) circuit. If the v oltage present at the input of the bistable circuit e xceeds the positi v e reference v oltage or f alls belo w the ne g ati v e reference v oltage, which are dened by zener diode/resistor circuits enclosed in the dashed box es, the output of the bistable circuit is dri v en to one of the supply rails. The reference v oltages are adjusted by v arying the potentiometer associated with each reference v oltage circuit. The output of each bistable circuit is then con v erted to TTL logic by a pair of op-amps and fed into a 7400 N AND g ate. The output of this N AND g ate pro vides the trigger to all of the digitizers in the e xperiment. Hence, the digitizers only trigger when the outputs of both optical sensors e xceed a certain threshold. The tw o sensors vie w the eld from opposite corners of the netw ork; therefore the lightning must be within the netw ork in order for both sensors to see the lightning and to cause the system to trigger Moreo v er as noted in Section 3.3.3 the ele v ation vie w of the sensors w as limited to k eep the system from triggering on a bright cloud dischar ge. 3.3 Measur ement Implementation In this section, the theory design, f abrication, and implementation of all MSE measurements is discussed. 3.3.1 Electric Field and Electric Field T ime-Deri v ati v e Measur ements In 2001, eight electric eld measurements were elded, located at Stations 1, 2, 4, 5, 6, 8, 9, and 10. In 2002, electric eld measurements at Stations 1 and 8 were

PAGE 64

50 + + + 10 k W 10 k W 5 pF 1 m F 10 k W 110 k W 20 k W 10 k W 110 k W 1 k W 1 k W 0.001 m F 0.001 m F LM318 LM318LM318 1N4148 1N4148 1 k W +LM318 1N4148 10 k W +LM318 1N4148 10 k W 10 k W + + + 10 k W 10 k W 5 pF 1 m F 10 k W 110 k W 20 k W 10 k W 110 k W 1 k W 1 k W 0.001 m F 0.001 m F LM318 LM318LM318 1N4148 1N4148 1 k W IN 2 +LM318 1N4148 10 k W +LM318 1N4148 10 k W 10 k W TRIGGER +5 V +12 V -12 V +12 V -12 V -12 V -12 V +12 V +12 V +12 V +12 V +12 V +12 V -12 V -12 V -12 V -12 V -12 V -12 V +12 V +12 V -12 V 5.1 k W 1 k W 5.5 V 0.1 m F 5.1 k W 1 k W 5.5 V 0.1 m F +12 V 5 k W 5 k W 5.1 k W 1 k W 5.5 V 0.1 m F +12 V 5 k W -12 V 5.1 k W 1 k W 5.5 V 0.1 m F 5 k W IN 1 OUT Figure 3–9. Schematic of the MSE trigger circuit. Enclosed in the dashed lines are the reference v oltage circuits used in the bistable circuits.

PAGE 65

51 sD nor mr s Figure 3–10. Illustration of the electric eld boundary condition at a perfectly conducting surf ace. con v erted to electric eld time-deri v ati v e (dE/dt) measurements and tw o additional dE/dt measurements were installed at Stations 4 and 9. No dE/dt measurements were made during the 2001 season. Conducting circular at-plate antennas were used to sense both the electric eld and its time-deri v ati v e. These types of antennas ha v e been used for man y natural and triggered lightning e xperiments, such as those described in ( Krider et al. 1977 ), ( Rak o v et al. 1998 ), and ( Uman et al. 2002 ). Furthermore, the y are generally considered standard equipment within the lightning research community There are man y types of electric eld and electric eld time-deri v ati v e sensors. Ho we v er when installed ush with the ground, the at-plate sensor has the adv antage that it theoretically introduces no eld enhancement. This is not the case, for e xample, with a whip antenna or an ele v ated plate antenna. Furthermore, the at-plate sensor can be implemented with entirely passi v e components and the equi v alent circuit analysis is typically straightforw ard (b ut not tri vial). 3.3.1.1 Analysis of a conducting at-plate antenna The The v enin or Norton equi v alent circuit for a at-plate antenna can be deri v ed by considering the boundary on a perfectly conducting ( s = ) surf ace, as sho wn in Figure 3–10 The boundary condition is ~ D ~ n = r s (3-1)

PAGE 66

52 The quantity ~ D is the electric displacement v ector at the surf ace and is e xpressed in units of C m 2 ~ n is the unit v ector normal to the surf ace, and r s is the surf ace char ge density and is also e xpressed in units of C m 2 Hence, ~ D ~ n is the component of the electric displacement which is normal to the surf ace. D nor m = r s (3-2) Note that since the surf ace is considered a perfect conductor the electric displacement inside of the conductor and the tangential component of the electric displacement along the surf ace of the conductor are both zero. If the medium abo v e the plate (air) is linear isotropic, homogeneous, and non-conducting, then D nor m = e E nor m (3-3) The quantity E nor m is the magnitude of the component of the electric eld which is normal to the surf ace of the plate (e xpressed in units of V m 1 ) and e is the permitti vity of the dielectric medium, which is essentially e 0 the permitti vity of free space (8 : 85 10 12 F m 1 ), for air Therefore, the e xpression for r s can be written as r s = e 0 E nor m (3-4) The total char ge on the surf ace of a conducting plate can be found by inte grating the surf ace char ge density o v er the area of the plate. Q pl a t e = Z S r s d A (3-5) The quantity d A is the dif ferential area on the surf ace of the plate. If the surf ace char ge density is uniform (which will be the case if the smallest w a v elength comprising E nor m is much greater than the plate diameter), then the total char ge can be found by multiplying the surf ace char ge density by the area of the plate. The boundary condition species that if the magnitude of the normal component of the electric eld is uniform

PAGE 67

53 o v er the surf ace of the plate, then surf ace char ge density along the plate must also be uniform. F or a circular plate, the electric eld along the surf ace of the plate can be considered non-uniform when the plate diameter is lar ger than about a sixteenth of a w a v elength. If the highest frequenc y component of E nor m is 30 MHz then the smallest w a v elength will necessarily be 10 m One sixteenth of a w a v elength is 0 : 625 m, which is lar ger than the diameter of the plates used in the MSE (discussed in Section 3.3.1.2 ). If the electric eld is uniform across a plate of area A pl a t e then the total char ge on the plate can be e xpressed as Q pl a t e = e 0 A pl a t e E nor m (3-6) The Norton equi v alent short circuit current, i ( t ) is necessarily the time-deri v ati v e of the char ge. i ( t ) = d d t Q pl a t e ( t ) = d d t e 0 A pl a t e E nor m ( t ) = e 0 A pl a t e d E nor m ( t ) d t (3-7) Hence, the at-plate antenna in the presence of a uniform time-v arying electric eld can be vie wed as a current source whose magnitude is proportional to the time deri v ati v e of the normal component of the electric eld. This Norton equi v alent current source is the basis of the equi v alent circuit. The The v enin equi v alent v oltage, which yields the same results, could also be used by performing a simple transformation. The equi v alent circuit analysis is most often performed in the frequenc y domain. Ho we v er a solution can be found directly in the time domain by solving a rst-order dif ferential equation. The relationship between the time domain and the frequenc y domain is gi v en by the F ourier transform. The F ourier transform, X ( w ) of a time-domain signal, x ( t ) is F f x ( t ) g = X ( w ) = Z t = x ( t ) e j w t d t (3-8) Dif ferentiation with respect to time in the time domain corresponds to multiplication by the comple x number j w in the frequenc y domain. Therefore, the e xpression for the

PAGE 68

54 Z S Z L I (w) Figure 3–11. Frequenc y-domain equi v alent circuit, using a Norton equi v alent current source, of a at-plate antenna sensor feeding a load (represented by Z L ). magnitude of the Norton equi v alent current source in the frequenc y domain becomes I ( w ) = e 0 A pl a t e ( j w E nor m ( w ) ) (3-9) The quantity E nor m ( w ) designates that the normal component of the electric eld is no w a function of angular frequenc y and not time. The Norton equi v alent circuit in the frequenc y domain is sho wn in Figure 3–11 The current source is placed in parallel with the source impedance, Z s and the load impedance, Z L The source impedance is the impedance of the antenna itself and the load impedance is the impedance of an y e xternal elements connected to the plate. In general, the source and load impedances can be resisti v e, capaciti v e, inducti v e, or a combination of the three. If the output of the antenna is tak en as the v oltage across the load impedance, then the e xpression for the output v oltage in the frequenc y domain is V ou t ( w ) = I ( w ) Z t o t al = I ( w ) ( Z s jj Z L ) = j we 0 A pl a t e E nor m ( w ) Z S Z L Z S + Z L (3-10) The output v oltage of the antenna in the frequenc y domain is the quantity j we 0 A pl a t e E nor m ( w ) scaled by the frequenc y-dependent quantity Z S Z L = ( Z S + Z L ) The frequenc y-independent g ain of the antenna is gi v en by the quantity e 0 A pl a t e If the quantity Z S Z L = ( Z S + Z L ) is equal to unity then the output is simply j w E nor m ( w ) scaled by the frequenc y-independent quantity e 0 A pl a t e If the quantity Z S Z L = ( Z S + Z L ) is equal to 1 = ( j w ) then the output is simply E nor m ( w ) scaled by the frequenc y-independent

PAGE 69

55 quantity e 0 A pl a t e Hence, the output of the antenna is characterized by the source and load impedance. T ypically the source impedance of a at-plate antenna is considered to be purely capaciti v e and an y resisti v e and inducti v e components are ignored. Therefore, the source impedance becomes Z s = 1 j w C an t (3-11) C an t is the capacitance of the at-plate antenna that can either be estimated theoretically or determined e xperimentally The antenna capacitance is a function of the plate area and the height of the plate abo v e a reference plate b uried in the ground. It may be possible to estimate the antenna capacitance using the relation C = e 0 A = d where d is the distance between the antenna plate and the grounded reference plate beneath it, b ut it is best to measure it. The load impedance is usually specically designed to pro vide the appropriate system response for the quantity which is to be measured. In other w ords, the load impedance will be dif ferent depending on whether one wishes to measure the electric eld or its time-deri v ati v e, as well as the bandwidth desired. The generalized load impedance will be considered to be a capacitor in parallel with a resistor; an y inducti v e components are ignored. The reason for this will become apparent when the antenna implementation is discussed. The capacitor C in t is typically kno wn as an inte grating capacitance and the resistor R is kno wn as the load resistance. Therefore, the load impedance becomes Z L = 1 j w C in t jj R = R j w C in t R + 1 j w C in t = R 1 + j w R C in t (3-12) Since the source and load impedances are both in parallel with the ideal current source, the y can be combined. The total shunt impedance, Z t o t al is Z t o t al = Z s jj Z L = 1 j w C an t jj R 1 + j w R C in t = R 1 + j w R ( C an t + C in t ) (3-13)

PAGE 70

56 Therefore, inserting the e xpression for Z t o t al into the Equation 3-10 yields V ou t ( w ) = j we 0 A pl a t e E nor m ( w ) R 1 + j w R ( C an t + C in t ) (3-14) The interpretation of the response of the antenna in the frequenc y domain depends on whether the antenna is to be vie wed as an electric eld or an electric eld time-deri v ati v e sensor From the dE/dt antenna perspecti v e, the response of the antenna, G d E d t ( w ) is gi v en by G d E d t ( w ) = V ou t ( w ) j w E nor m ( w ) = e 0 A pl a t e R 1 1 + j w R ( C an t + C in t ) (3-15) T ypically no inte grating capacitor is used with a dE/dt antenna ( C in t = 0 ) therefore the e xpression for G d E d t ( w ) reduces to G d E d t ( w ) = e 0 A pl a t e R 1 1 + j w R C an t (3-16) This is the response of a rst order lo w-pass lter with magnitude response gi v en by G d E d t ( w ) = e 0 A pl a t e R 0BB@ 1 h 1 + ( w R C an t ) 2 i 1 2 1CCA (3-17) The pass-band g ain is equal to e 0 A pl a t e R The -3 dB point (the frequenc y at which the output is approximately 0.707 times the output in the pass-band) of the response is w 0 = 1 R C an t (3-18) F or e xample, if R = 50 W and C an t = 30 pF, then w 0 = 6 : 67 10 8 s 1 The corresponding frequenc y is f 0 = w 0 2 p = 6 : 67 10 8 s 1 2 p 106 MHz (3-19) Therefore, this e xample at-plate dE/dt antenna has a -3 dB bandwidth of approximately 106 MHz. This at-plate design is a suitable dE/dt antenna for most

PAGE 71

57 wide-band applications. If both R and C an t are doubled, then the -3 dB bandwidth must necessarily decrease by a f actor of four yielding f 0 26 : 5 MHz. If the frequenc y range of interest lies f ar belo w the -3 dB point, then the e xpression for the response of the dE/dt antenna in the frequenc y domain becomes G d E d t ( w ) = e 0 A pl a t e R (3-20) The abo v e e xpression is only v alid for frequencies which satisfy the condition w w 0 = 1 R C an t (3-21) If the same condition is applied to Equation 3-14 (assuming C in t = 0), then the output v oltage of the dE/dt antenna in the frequenc y domain becomes: V ou t ( w ) = j we 0 A pl a t e RE nor m ( w ) (3-22) T o nd the output v oltage in the time domain, the in v erse F ourier transform is used. F 1 f X ( w ) g = x ( t ) = 1 2 p Z w = X ( w ) e j w t d w (3-23) Ho we v er the inte gral need not be computed in this case since the dif ferentiation property of the F ourier transform can be used. Therefore, the e xpression for the output v oltage in the time domain is v ou t ( t ) = e 0 A pl a t e R d E nor m ( t ) d t (3-24) Although this is a time-domain e xpression, the frequenc y constraint still applies since the time-domain e xpression is deri v ed from a frequenc y-domain e xpression with that constraint. In other w ords, the abo v e e xpression is only v alid if the dE/dt w a v eform has no signicant frequenc y content abo v e w 0

PAGE 72

58 If the at-plate antenna is vie wed from an electric eld antenna perspecti v e, the response of the antenna in the frequenc y domain, G E ( w ) is G E ( w ) = V ou t ( w ) E nor m ( w ) = e 0 A pl a t e R j w 1 + j w R ( C an t + C in t ) (3-25) T ypically an inte grating capacitor is used with an electric eld antenna. Depending on the measurement design, C in t may or may not dominate C an t ho we v er it best that it does since the antenna capacitance can potentially v ary with antenna mo v ement and surroundings. Combining the tw o capacitances into a single term yields C = C an t + C in t (3-26) G E ( w ) = e 0 A pl a t e R j w 1 + j w R C (3-27) The response of the electric eld antenna is that of a rst order high-pass lter with magnitude response gi v en by j G E ( w ) j = e 0 A pl a t e R 0BB@ w h 1 + ( j w R C ) 2 i 1 2 1CCA (3-28) The pass-band g ain is equal to e 0 A pl a t e = C The -3 dB point of the response is w 0 = 1 R C (3-29) F or e xample, if R = 500 k W and C = 0 : 1 F, then w 0 = 20 s 1 The corresponding frequenc y is f 0 = w 0 2 p = 20 s 1 2 p 3 : 2 Hz (3-30) Therefore, the output of this e xample electric eld antenna pro vides an adequate approximation of the electric eld w a v eform at frequencies do wn to about 3 Hz. It should be noted that w 0 is a lo w-frequenc y roll-of f when speaking in terms of an

PAGE 73

59 electric eld antenna and a high-frequenc y roll-of f when speaking in terms of a dE/dt antenna. The general time-domain e xpression for the output v oltage of an electric eld antenna is found by performing the in v erse F ourier transform on the frequenc y-domain e xpression. Rearranging Equation 3-14 (substituting C = C an t + C in t ) yields V ou t ( w ) = e 0 A pl a t e C j w 1 R C + j w E nor m ( w ) (3-31) This e xpression can be vie wed as the multiplication of tw o functions of w This can be e xpressed as V ou t ( w ) = X ( w ) Y ( w ) (3-32) The quantities X ( w ) and Y ( w ) can be dened as X ( w ) = j w 1 R C + j w (3-33) Y ( w ) = e 0 A pl a t e C E nor m ( w ) (3-34) Therefore, the time-domain e xpression for the antenna output v oltage, v ou t ( t ) can be found by the con v olution property of the F ourier transform. v ou t ( t ) = x ( t ) y ( t ) (3-35) The operator denotes linear con v olution and x ( t ) and y ( t ) are the in v erse F ourier transforms of X ( w ) and Y ( w ) respecti v ely Con v olution is performed by means of the con v olution inte gral v ou t ( t ) = Z l = x ( l ) y ( t l ) d l (3-36) Alternati v ely since linear con v olution is a commutati v e operation, v ou t ( t ) can be e xpressed as v ou t ( t ) = Z l = y ( l ) x ( t l ) d l (3-37)

PAGE 74

60 The e xpressions for x ( t ) and y ( t ) are found by using properties of the F ourier transform: First, x ( t ) is found by utilizing the dif ferentiation property of the F ourier transform and a well-kno wn F ourier transform pair x ( t ) = d d t F 1 ( 1 1 R C + j w )! = d d t e t R C u ( t ) (3-38) The term u ( t ) is the unit-step function and is dened as u ( t ) = 8>><>>: 1 t 0 0 t < 0 (3-39) The time-deri v ati v e can be computed by using the product rule of dif ferentiation and the f act that the deri v ati v e of the unit-step function is the Dirac delta function, d ( t ) which is dened as d ( t ) = 8>><>>: t = 0 0 t 6 = 0 (3-40) Therefore, the e xpression for x ( t ) is x ( t ) = 1 R C e t R C u ( t ) + d ( t ) (3-41) The quantity E nor m ( w ) is an arbitrary function of w therefore the e xpression for y ( t ) is y ( t ) = e 0 A pl a t e C E nor m ( t ) (3-42) The quantity E nor m ( t ) is simply the in v erse F ourier transform of E nor m ( w ) Substituting the e xpressions for x ( t ) and y ( t ) into Equation 3-35 yields v ou t ( t ) = 1 R C e t R C u ( t ) + d ( t ) e 0 A pl a t e C E nor m ( t ) (3-43)

PAGE 75

61 Recalling that the con v olution operator obe ys the distrib uti v e property and that an y function of t con v olv ed with d ( t ) is simply the function itself yields v ou t ( t ) = 1 R C e t R C u ( t ) e 0 A pl a t e C E nor m ( t ) + e 0 A pl a t e C E nor m ( t ) (3-44) The con v olution is computed via the con v olution inte gral, as sho wn in Equations 3-36 and 3-37 yielding v ou t ( t ) = e 0 A pl a t e C E nor m ( t ) e 0 A pl a t e R C 2 Z l = E nor m ( l ) e t l R C u ( t l ) d l (3-45) The ef fect of the unit-step function in the second term of Equation 3-45 can be incorporated into the limits of inte gration. Furthermore, the term e t R C can be f actored out of the inte gral. This yields v ou t ( t ) = e 0 A pl a t e C E nor m ( t ) e 0 A pl a t e R C 2 e t R C Z t l = E nor m ( l ) e l R C d l (3-46) The abo v e e xpression is v alid for an arbitrary electric-eld, E nor m ( t ) with no frequenc y constraints. The antenna output v oltage consists of tw o terms, the rst of which is the electric eld scaled by the quantity ( e 0 A ) = C This is the output of an ideal at-plate electric-eld antenna. The second term is the ef fect of the non-ideal lo w-frequenc y response of the antenna. As R approaches innity the second term approaches zero and the output is that of an ideal at-plate electric-eld antenna. This can also be seen by considering the frequenc y-domain e xpression for the antenna output v oltage and allo wing R to approach innity The same ar gument can be applied to allo wing the quantities R C and R C 2 to approach innity If C alone is allo wed to approach innity then in the limit the output v oltage will be zero. Ho we v er L 'Hopital' s Rule can be in v ok ed, and it can be seen that the second term will decrease to zero before the rst. Therefore, it can be said that for v ery lar ge v alues of C the antenna output v oltage will be approximately that of an ideal antenna, with v ery lo w g ain. Of

PAGE 76

62 course, “v ery lar ge” and “v ery lo w” are relati v e terms which are determined by the requirements of the e xperiment. The e xpression for the antenna output v oltage can be simplied some what if the electric-eld w a v eform is constrained to be of the follo wing form. E nor m ( t ) = s ( t ) u ( t ) (3-47) The quantity s ( t ) is an arbitrary function of time and u ( t ) is the unit-step response which w as dened in Equation 3-39 If this is substituted into Equation 3-46 the antenna output v oltage becomes v ou t ( t ) = e 0 A pl a t e C s ( t ) u ( t ) e 0 A pl a t e R C 2 e t R C Z t l = 0 s ( l ) e l R C d l (3-48) The antenna output v oltage can be specied e xactly if the electric eld w a v eform is specied e xactly The response of the antenna to a step-function is of particular interest since it can be used to crudely approximate the electric eld w a v eform from a return strok e. If s ( t ) is dened as the constant E 0 then the electric eld w a v eform is gi v en by E nor m ( t ) = E 0 u ( t ) (3-49) The quantity E 0 is amplitude of the step function. The corresponding antenna output v oltage is v ou t ( t ) = e 0 A pl a t e C E 0 e t R C + e 0 A pl a t e C E 0 [ u ( t ) 1 ] (3-50) The second term in the abo v e e xpression is zero for all times t 0, therefore the e xpression reduces to v ou t ( t ) = e 0 A pl a t e C E 0 e t R C (3-51)

PAGE 77

63 The term R C is kno wn as the decay time constant and is typically denoted by t The time constant is the in v erse of w 0 t = R C = 1 w 0 (3-52) When the electric eld is a step-function, the output of the antenna at time t = t is a f actor of 1 = e less than it w as at time t = 0. Therefore, the output of the antenna is only v alid for a short period of time relati v e to t This is e xactly what is e xpected since the response of the electric eld at-plate antenna is that of a high-pass lter in the frequenc y domain. F or times t t the output of the antenna, gi v en a step-function input of amplitude E 0 becomes v ou t ( t ) = e 0 A pl a t e C E 0 (3-53) If the e xample circuit parameters R = 500 k W and C = 0 : 1 F are ag ain considered, the decay time constant is t = ( 500 k W ) ( 0 : 1 F ) = 50 ms (3-54) Gi v en a step-function input, the output of this e xample electric eld antenna is only v alid for times much less than 50 ms Depending on the application, this time constant may or may not be acceptable. Increasing the time constant by increasing the capacitance necessarily decreases the output of the antenna by a f actor of 1 = C The resistance can be increased with no ef fect on the g ain, b ut there are practical limits to ho w high this resistance can be increased. T ypically both R and C are increased with C being adjusted to yield an acceptable g ain. It should be noted that if Equation 3-46 is solv ed for E nor m ( t ) as a function v ou t ( t ) it may be possible to “correct” the measured electric eld. In other w ords, it may be

PAGE 78

64 Antenna Housing Flat Plate Annular Gap Figure 3–12. Aluminum at plate antenna used in the MSE. possible to remo v e the ef fects of the non-ideal characteristics of the antenna from the measured data if the antenna can modeled in such a w ay that Equation 3-46 applies. 3.3.1.2 Flat-plate antenna implementation No w that the output of the at-plate antenna has been characterized in the time and frequenc y domains, the implementation of the MSE electric eld and electric eld time-deri v ati v e antennas can be discussed. Both the electric eld and the electric eld time-deri v ati v e measurements utilize identical aluminum circular at-plate antennas of area 0 : 16 m 2 and diameter 0 : 45 m as sho wn in Figure 3–12 The antenna consists of a hollo w rectangular aluminum housing with a circular portion of the top f ace isolated from the remainder of the top f ace of the structure by a 0 : 6 cm-wide annular air g ap surrounding it. The circular portion is k ept electrically isolated by this annular g ap and six n ylon standof fs which are used to mount it to the bottom (ground) of the housing. The circular portion is the sensor while the remainder of the housing is connected to a 3 m long ground rod by a short length of 12 A WG wire and a lug mounted on the side of the housing. A lug on the underside of the circular plate connects an approximately 20 cm length of 16 A WG wire to the

PAGE 79

65 center conductor of a female BNC connector mounted to the side of the housing. The inductance of a 20 cm length of 16 A WG wire is approximately 10 nH. The outer conductor of the BNC connector is connected to the grounded aluminum housing. Therefore, the v oltage measured across the BNC connector is the v oltage which appears between the circular plate and the grounded antenna housing. A detailed mechanical dra wing of the antenna is sho wn in Figure 3–13 The antenna housing w as placed on top of the ground. In order to simulate a at ground, pieces of wire mesh screen which e xtended from each side of the antenna were attached to the top of the the four sides of the metal antenna housing. One portion of the screen w as about one meter in length while the others were about a third of a meter in length. The electronics for each electric eld or electric eld time-deri v ati v e measurement w as stored in a metal Hof fman enclosure (also kno wn as a Hof fman box) which w as b uried under ground about one meter from the antenna. Each hole w as about a half a meter deep and the Hof fman box w as placed on a shelf angled do wnw ard so as to drain an y w ater acquired in the hole a w ay from the box. Furthermore, pieces of w ood were used to secure the box in the hole. The Hof fman box w as placed under ground so that it w ould be protected from the e xternal en vironment and the area surrounding the antenna w ould be as at as possible. In addition, a piece of reecti v e insulation w as placed o v er the hole to protect the electronics from the heat of the sun. The 1 m length of screen co v ered the hole and the insulation. A length of 50 W coaxial cable with male BNC connectors connected the antenna to a female BNC b ulkhead feed-through connector mounted to the side of the Hof fman box. This cable w as enclosed in metal shield braid which w as secured to the male BNC connectors on each end of the cable by metal hose clamps. Therefore, the shield braid and the Hof fman enclosure are electrically connected to the grounded antenna housing. This conguration serv ed to electromagnetically shield the coaxial cable and electronics, minimize the electric eld enhancement of the antenna, and protect the electronics

PAGE 80

66 Female BNCConnector 11 22 60 22 2 3/4 6 8.75 9 Ground Rod Attach Lug Connection Point Between BNC and Plate 2 3/16 Nylon Standoff Drip Edge TOP VIEW CROSS SECTION Drain Hole Figure 3–13. Detailed mechanical dra wing of the aluminum at plate antenna used in the MSE. Adapted from ( Cra wford 1998 ).

PAGE 81

67 Ground Rod Wire Screen Reflective Insulation Fiberoptic Cable Cable Coaxial Enclosure Metal Flat Plate Antenna Figure 3–14. Diagram of an installation of a MSE measurement utilizing a at-plate antenna. themselv es from the e xternal en vironment. A dra wing of this conguration is sho wn in Figure 3–14 This conguration is v ery similar to that described in ( Cra wford 1998 ). 3.3.1.3 Electric eld measur ement implementation As described in Section 3.3.1.1 essentially the only feature dif ferentiating an electric eld measurement and an electric eld time-deri v ati v e measurement implemented with identical at-plate antennas is the choice of the load impedance of the antenna. This choice is not tri vial, ho we v er and an improper choice of circuit parameters will almost certainly lead to a poor -quality measurement. Also, as stated pre viously the load impedance will be considered to be an inte grating capacitor C in t in parallel with a load resistance, R The inte grating capacitance af fects both the g ain and the decay time-constant of the antenna while the load resistance only af fects the time constant. The antenna capacitance must, in general, also be tak en into account b ut it can be ignored if the inte grating capacitance is much greater C in t >> C an t (3-55)

PAGE 82

68 In the frequenc y domain, this implies that the impedance of the antenna is much higher than that of the inte grating capacitor 1 j w C an t >> 1 j w C in t (3-56) In 2001, 0 : 5 F inte grating capacitors were used in the MSE electric eld measurements. The antenna capacitance w as measured to be about 80 pF (although this measurement w as v ery sensiti v e to the position of the antenna); hence the abo v e assumption holds and the antenna capacitance can be ignored. Eight capacitor units were b uilt by Geor ge Schnetzer in the summer of 2001. Each unit consisted of a 5 : 5 cm by 2 : 5 cm by 2 cm box with a male BNC connector on one end and a female BNC connector on the other Fi v e 0 : 1 F capacitors were placed in parallel inside each box to achie v e the desired capacitance. One of these capacitor units is pictured in Figure 3–15 The operating range of each inte grating capacitor w as tested by placing the capacitor in parallel with the 50 W input of an oscilloscope and feeding it with a 1 V sinusoidal v oltage source ha ving a 50 W source resistance. The e xpected magnitude response of this circuit is gi v en by V ca paci t or ( w ) = 1 q 1 + ( 25 w C ) 2 (3-57) The measured and e xpected test circuit responses for an inte grating capacitor unit used in 2001 are sho wn in Figure 3–16 The response of the capacitor ceases to be ideal at about 3 MHz, an undesirable lo w v alue. The cause w as belie v ed to be a series resonance of the capacitor and the inductance of the capacitor leads. The capacitors were placed in parallel in order to minimize the inductance of the leads b ut the inductance w as not decreased lo w enough. The resonant frequenc y of an L C circuit (either series or parallel) is dened as w R = 1 p L C (3-58)

PAGE 83

69 AB Connector Male BNC Connector Female BNC Capacitors in Parallel Pomona Box Figure 3–15. Inte grator capacitor assembly used in 2001. A) Closed Pomona box. B) Box open to sho w interior If C = 0 : 5 F and w R = 2 p 3 10 6 s 1 then L = 5 : 3 nH. Hence, only a v ery small lead inductance is required to achie v e a resonance at 3 MHz when 0 : 5 F capacitors are used. It follo ws that the magnitude response of the electric-eld measurements in 2001 w as limited to 3 MHz due to the non-ideal inte grating capacitors. F or 2002, in order to achie v e better frequenc y response, the capacitance w as reduced by a f actor of 2 : 5 to 0 : 2 F to increase the resonant frequenc y to abo v e 5 MHz If C = 0 : 2 F and L 5 nH then w R 2 p 5 10 6 s 1 In addition, military-grade ceramic capacitors were used and the number of capacitors placed in parallel w as increased, hence decreasing the series inductance. F or tw o of the ne w units, ten 0 : 022 F capacitors were placed in parallel while tw o other units consisted

PAGE 84

70 0.001 0.01 0.1 1 0 10 100 1000 10000 5 10 6 10 7 10 Circuit Response Frequency [Hz] Response of Circuit to Test Integrating Capacitor 01-01 0.001 0.01 0.1 1 0 10 100 1000 10000 5 10 6 10 7 10 Circuit Response Frequency [Hz] Response of Circuit to Test Integrating Capacitor 01-01 Figure 3–16. Measured (dashed line) and e xpected (solid line) test circuit responses for inte grating capacitor unit 01-01 ( 0 : 477 F ) of nine 0 : 022 F capacitors. One unit w as constructed from ele v en 0 : 018 F capacitors and another unit consisted of se v en 0 : 022 F capacitors along with tw o 0 : 017 F capacitors. All of the ne w capacitor units inte grated properly up to at least 5 MHz, although most e xhibited de viations from an ideal inte grator between 5 and 10 MHz. Some of the units e xperience a drop in g ain after 5 MHz while others e xperienced an increase. This situation w as considered acceptable since the measurements were to be band-limited to belo w 5 MHz by the b uilt in anti-aliasing lter of the digitizer The inte grating capacitors used in 2002 were tested in the same w ay as in 2001, with the e xpected response gi v en by Equation 3-57 The measured and e xpected test circuit responses for an inte grating capacitor unit b uilt in 2002 are sho wn in Figure 3–17 The load resistance, R is typically the input resistance of an amplier or a ber -optic transmitter The input capacitance of the ber -optic transmitter w as considered ne gligible and hence ignored. The input resistance of the Opticomm

PAGE 85

71 0.001 0.01 0.1 1 0 10 100 1000 10000 5 10 6 10 7 10 Circuit Response Frequency [Hz] Response of Circuit to Test Integrating Capacitor 02-09 0.001 0.01 0.1 1 0 10 100 1000 10000 5 10 6 10 7 10 Circuit Response Frequency [Hz] Response of Circuit to Test Integrating Capacitor 02-09 Figure 3–17. Measured (dashed line) and e xpected (solid line) test circuit responses for inte grating capacitor unit 02-09 ( 0 : 209 F ) MMV -120C ber -optic transmitter (described in Section 3.4.1 ) is 68 k W This yields decay time constants of 34 ms and 13 : 6 ms when 0 : 5 F and 0 : 2 F inte grating capacitors are used, respecti v ely Therefore, an amplier with high input resistance w as required to increase the time constant since increasing the inte grating capacitance further could result in further resonance problems. In 2001, K eith Rambo and Geor ge Schnetzer designed a unity g ain amplier with an input impedance of 5 : 1 M W A schematic of the amplier is sho wn in Figure 3–18 (the po wer supply circuitry is omitted for simplicity). An input resistance of 5 : 1 M W yields decay time constants of 2 : 55 s and 1 : 02 s when 0 : 5 F and 0 : 2 F inte grating capacitors are used, respecti v ely The decay time constant in either case w as suf cient to measure accurately the o v erall electric eld change of a natural cloud-to-ground lightning ash since ( Ber ger et al. 1975 ) documented the median ash duration, e xcluding single-strok e ashes, to be about 180 ms. In 2002, the g ain of the amplier w as increased to tw o in order to lo wer the

PAGE 86

72 + AD825 IN OUT 5.1 M W Figure 3–18. Schematic of the high-impedance amplier used in the 2001 MSE. full-scale range of the measurement. The output resistance of the AD825 op-amp is approximately 8 W although the feedback netw ork reduces this to a fraction of an Ohm. Therefore, the output of the amplier will be the same re g ardless of whether it is terminated in 50 W or high impedance. The amplier is po wered from a 12 V battery The amplier ground is at located 6 V abo v e the ne g ati v e battery terminal in order to bias the AD825 op-amp with 6 V. Therefore, if a single battery were to be used to po wer all of the electronics in the measurement, the output of the amplier w ould oat 6 V abo v e the common ground which is f ar abo v e the maximum input range of the ber -optic transmitter In order to alle viate this problem, both the input and output of the amplier w ould ha v e to be A C coupled (by a DC blocking capacitor), or the amplier w ould ha v e to be po wered by a separate battery than the remainder of the electronics. The second option w as chosen since it w as desirable to k eep the lo w-frequenc y roll-of f of the measurement to a minimum. The amplier w as connected to a separate 12 V battery through a relay The control circuit of the relay w as connected to the primary measurement battery (through the PIC controller) and the load circuit of the relay w as connected to the separate 12 V battery Therefore, when the PIC controller supplies po wer to the electronics in the measurement, it also connects the amplier to its battery via the relay Hence, the

PAGE 87

73 0.1 1 0 10 100 1000 10000 5 10 6 10 7 10 Gain Frequency [Hz] Measured Frequency Response of Hi-Z Amplifier 01-07 Figure 3–19. Measured frequenc y response of high impedance amplier 01-07. amplier is not po wered when the measurement is turned of f. The measured frequenc y response of an amplier is sho wn in Figure 3–19 A diagram of an MSE electric eld measurement is sho wn in Figure 3–20 The antenna is placed on the ground (surrounded by a metal screen as described in Section 3.3.1.2 ) and the electronics are enclosed in a metal Hof fman enclosure (indicated by the dashed line). In order to eliminate ground loops, all electronic components were isolated from each other and the metal box by pieces of plastic and Styrofoam. The inte grating capacitor C enclosed in a Pomona box, is directly connected to the female BNC connected mounted to the inside of the Hof fman box. The other end of the Pomona box is connected to the input of the amplier (ha ving g ain G am p and input resistance R ) via a short length of 50 W coaxial cable. Note that this cable is not terminated in its characteristic impedance of 50 W Ho we v er the length of the cable is short enough that reections in the cable should not be a problem. The output of the amplier is connected to the input of a PIC controller The output of the PIC controller is connected to an Opticomm MMV -120C ber -optic transmitter terminated in 50 W

PAGE 88

74 Integrating Capacitor in a Metal Box Fiberoptic Transmitter Amplifier Relay PIC Controller Supplies Power to PIC Controller Supplies Power to PIC Controller Amplifier 12 V Battery PIC12 V Battery Controller Transmitter Fiberoptic Metal Enclosure FiberopticCable C HiZ Amplifier Flat-plateAntenna G PI C G am p 50 W 50 W In-line T erminator Figure 3–20. Diagram of a MSE electric eld measurement. Since the PIC controller is terminated in 50 W it can be thought of as a 50 W in-line attenuator of v alue G PI C Using the e xpression for the ideal output v oltage of the electric eld at-plate antenna, the e xpression for the v oltage at the input of the ber -optic transmitter is v F O T ( t ) = e 0 A pl a t e C E nor m ( t ) G am p G PI C (3-59) The abo v e e xpression assumes that the output v oltage of the antenna can be approximated by the ideal output and the g ain of the amplier is at in the frequenc y range of interest. The output of the ber -optic recei v er is connected to a digitizer terminated in 50 W by a length of 50 W coaxial cable. The v oltage at the input of the digitizing oscilloscope is the v oltage present at the input of the ber -optic transmitter modied by the ber -optic link. If the ef fect of the link is assumed to be a frequenc y-independent g ain or attenuation (of v alue G l ink ), then the e xpression for the v oltage at the input of the oscilloscope is v sco pe ( t ) = e 0 A C G am p G PI C G l ink E nor m ( t ) (3-60)

PAGE 89

75 In 2001, A pl a t e = 0 : 16 m 2 C = 0 : 5 F, G am p = 1, and G PI C = 1. According to the manuf acturer specications, G l ink = 1, ho we v er this will be considered a nominal v alue since in practice G l ink is often not unity In addition, the v alue for C v aried between indi vidual electric eld measurements, hence C = 0 : 5 F is also considered a nominal v alue. Therefore, the nominal v oltage seen at the input of the scope for the 2001 MSE electric eld measurements is gi v en by v sco pe nom ( t ) = 2 : 83 10 6 E nor m ( t ) (3-61) v sco pe nom ( t ) is e xpressed in units of V when E nor m ( t ) is e xpressed in units of V m 1 If v sco pe nom ( t ) = 1 V and the abo v e e xpression is solv ed for E nor m ( t ) then E nor m = 3 : 53 10 5 V m 1 V 1 (3-62) If E nor m is e xpressed in units of kV m 1 then the abo v e e xpression becomes E nor m = 353 kV m 1 V 1 (3-63) Thus, nominally one v olt present at the input of the digitizer corresponds to an electric eld magnitude of about 350 kV m 1 In 2002, A pl a t e = 0 : 16 m 2 C = 0 : 2 F, G am p = 2, and G PI C = 1. If, as before, G l ink is assumed to ha v e a nominal v alue of unity the nominal v oltage seen at the input of the scope for the 2002 MSE electric eld measurements is v sco pe nom ( t ) = 1 : 42 10 5 E nor m ( t ) (3-64) Ag ain, solving the equation for E nor m e xpressed in units of kV m 1 yields E nor m = 70 : 6 kV m 1 V 1 (3-65) If the true capacitor v alue for each electric eld measurement is substituted into Equation 3-60 the e xpression for the electric eld as a function of the v oltage at the

PAGE 90

76 T able 3–7. Salient characteristics of the 2001 MSE electric eld measurements. Designation C ( F ) t ( s ) Units/V olt kV m 1 V 1 E-1 0.477 2.43 336.9 E-2 0.517 2.64 365.1 E-4 0.504 2.57 355.9 E-5 0.503 2.57 355.2 E-6 0.510 2.60 360.2 E-8 0.501 2.56 353.8 E-9 0.507 2.59 358.1 E-10 0.508 2.59 358.8 T able 3–8. Salient characteristics of the 2002 MSE electric eld measurements. Designation C ( F ) t ( s ) Units/V olt kV m 1 V 1 E-2 0.230 1.17 81.2 E-4 0.202 1.03 71.3 E-5 0.228 1.16 80.5 E-6 0.204 1.04 72.0 E-9 0.198 1.01 69.9 E-10 0.210 1.07 74.2 input of the digitizer becomes more accurate. Finally if the g ain of the ber -optic link is kno wn, the e xpression can be made e v en more accurate. In general, the capacitance v alues are measured once before being installed in the eld while the g ain of the ber optic link is measured with a square w a v e signal (as described in Section 3.4.5 ) at least once during each thunderstorm. T ables 3–7 and 3–8 summarize the salient characteristics of all of the indi vidual 2001 and 2002 MSE electric eld measurements, including the inte grating capacitance, the corresponding decay time constant, and the corresponding units per v olt (assuming G l ink = 1). Opticomm MMV -120C ber -optic links were used to transmit the analog w a v eforms to the Launch Control trailer where the y were digitized. The electric eld measurements were digitized continuously for 800 ms (200 ms pre-trigger) at 10 MHz on a Y ok og a w a DL716 digital storage oscilloscope. The electric eld w a v eforms were band-limited to 4 MHz ( -3 dB) by the anti-aliasing lter associated

PAGE 91

77 Fiberoptic PIC Controller Supplies Power to Fiberoptic Transmitter PIC Battery FiberopticTransmitter Cable PIC Controller Metal Enclosure Controller 12 V 50 W In-line T erminator 50 W G PI C Antenna Flat-plate Figure 3–21. Diagram of a MSE dE/dt measurement. with the DL716 digitizer Detailed descriptions of the Opticomm ber -optic links and the DL716 digitizer are gi v en in Sections 3.4.1 and 3.5.1 respecti v ely 3.3.1.4 Electric eld time-deri v ati v e measur ement implementation As stated pre viously no inte grating capacitor is used in a dE/dt measurement ( C in t = 0 ) and therefore the load impedance is simply the load resistance, R As with an electric-eld measurement, the at-plate antenna w as connected directly to the female BNC b ulkhead feed-through connector mounted to the side of the Hof fman enclosure. Inside the Hof fman box, a short length of 50 W coaxial cable connected the end of the b ulkhead feed-through BNC connector inside of the box to the input of a PIC controller The output of the PIC controller is connected to an Opticomm MMV -120C ber -optic transmitter terminated in 50 W Since the PIC controller is terminated in 50 W the PIC controller appears to be a 50 W in-line attenuator of v alue G PI C to the antenna. Therefore, the load resistance, R is 50 W In order to eliminate ground loops, all electronic components were isolated from each other and the metal box by pieces of plastic and Styrofoam. A diagram of the conguration is presented in Figure 3–21

PAGE 92

78 The capacitance of the 0 : 16 m 2 at-plate antennas w as measured to be about 80 pF. Therefore, using Equation 3-18 the -3 dB bandwidth of the response is w 0 = 1 R C an t = 1 ( 50 W ) ( 80 pF ) = 2 : 5 10 8 s 1 (3-66) This corresponds to a frequenc y of f 0 40 MHz. Since the dE/dt measurements were to be band-limited to 20 MHz (-3 dB) via anti-aliasing lters at the digitizers inputs, the response of the antenna can be considered uniform o v er the complete frequenc y range of interest. Therefore, Equation 3-24 can be used to approximate the v oltage output of the dE/dt antenna. The e xpression for the v oltage at the input of the ber -optic transmitter is v F O T ( t ) = e 0 A pl a t e RG PI C d E nor m ( t ) d t (3-67) The attenuation for the dE/dt measurements ( G PI C ) w as v aried o v er the course of the 2002 season. A v alue of 0.199 (-14 dB) w as originally used, ho we v er this resulted in an output v oltage that w as too lo w Therefore the attenuation w as changed to 0.316 (-10 dB) b ut this too resulted in a lo w output v oltage. Finally the attenuation setting w as changed to 0.501 (-6 dB) to obtain an acceptable output v oltage. Using this nal v alue for G PI C along with A pl a t e = 0 : 16 m 2 and R = 50 W yields the follo wing e xpression for v F O T ( t ) when dE/dt is e xpressed in units of V m 1 s 1 v F O T ( t ) = 3 : 55 10 11 d E nor m ( t ) d t (3-68) If dE/dt is e xpressed in units of kV m 1 s 1 the abo v e e xpression becomes v F O T ( t ) = ( 0 : 0355 ) d E nor m ( t ) d t (3-69) The v oltage present at the input of the digitizing oscilloscope is the v oltage present at the input of the ber -optic transmitter modied by the ber -optic link. If the ef fect of the link is assumed to be a frequenc y independent g ain or attenuation, G l ink

PAGE 93

79 then the e xpression for the v oltage at the input of the oscilloscope is v sco pe ( t ) = v F O T ( t ) G l ink = ( 0 : 0355 ) G l ink d E nor m ( t ) d t (3-70) If the nominal g ain of the ber -optic link is assumed to be one, the nominal v oltage at the input of the digitizing scope is the same as that present at the input of the ber -optic transmitter v sco pe nom ( t ) = ( 0 : 0355 ) d E nor m ( t ) d t (3-71) If v sco pe is assumed to be 1 V and the abo v e e xpression is solv ed for d E nor m = d t then d E nor m d t = 28 : 2 kV m 1 s 1 V 1 (3-72) Hence, nominally one v olt present at the input of the digitizing oscilloscope corresponds to a dE/dt magnitude of approximately 28 kV m 1 s 1 This v alue is only v alid for data sets where G PI C is equal to 0.501, ho we v er the result for an y attenuation v alue can be obtained by substituting the appropriate v alue into Equation 3-67 Opticomm MMV -120C ber -optic links were used to transmit the analog w a v eforms to the Launch Control trailer where the y were digitized. The dE/dt measurements were digitized on a LeCro y L T374 W a v erunner2 digital storage oscilloscope in se gmented memory mode. The internal memory w as di vided into four se gments with each se gment requiring a separate trigger signal; hence data for a total of four return strok es could be acquired. Unlik e the measurements digitized by the DL716, no data were acquired during inter -strok e interv als. During July of 2002 each se gment w as digitized at 50 MHz for 20 ms (19 ms pre-trigger). In August of 2002 the digitization rate and acquisition length for each se gment were changed to 200 MHz and 5 ms (4 ms pre-trigger), respecti v ely The dE/dt w a v eforms were band-limited to 20 MHz (-3 dB) by anti-aliasing lters at the digitizers inputs. Detailed

PAGE 94

80 descriptions of the Opticomm ber -optic links and the L T374 digitizer are gi v en in Sections 3.4.1 and 3.5.3 respecti v ely 3.3.2 Magnetic Field and Magnetic Field T ime-Deri v ati v e Measur ements In 2001, four magnetic eld measurements were elded. T w o measurements were located at Station 4, and the remaining tw o were located at Station 9. The tw o sensors at each station were arranged so that tw o orthogonal components of the horizontal magnetic eld could be sensed. This is v ery similar to the conguration described in ( Cra wford et al. 2001 ). In 2002, in order to conserv e equipment, only one magnetic eld measurement from each station w as used, and hence the total number of magnetic eld measurements w as reduced from four to tw o. Furthermore, this meant that only one orthogonal component of the magnetic eld could be sensed at each location. Although the same sensors were used in both 2001 and 2002, the electronics associated with the sensors dif fered for the tw o years. In addition, in 2002 four magnetic eld time-deri v ati v e (dB/dt) measurements were elded. T w o deri v ati v e measurements were located at Station 1, which were arranged so that tw o orthogonal components of the azimuth eld could be sensed. The remaining tw o measurements were located at Stations 4 and 9, with only one orthogonal component of the eld being sensed at each location. Square loops constructed from 50 W coaxial cable were used to sense both the magnetic eld and its time-deri v ati v e. 3.3.2.1 Analysis of a loop antenna The The v enin or Norton equi v alent circuit of a coaxial loop antenna can be deri v ed by rst considering F araday' s La w I C ~ E ~ d s = d F d t (3-73)

PAGE 95

81 The quantity ~ E is the electric-eld v ector ~ d s is a dif ferential length about an arbitrary closed path C and F is the magnetic ux through the surf ace dened by the closed path C and is e xpressed in units of Wb. F araday' s La w states that the line inte gral of the electric eld about an arbitrary closed path C is equal to the ne g ati v e of the time-deri v ati v e of the magnetic ux through the surf ace dened by C The magnetic ux is dened as F = Z S ~ B ~ d a (3-74) The quantity ~ B is the magnetic induction (or magnetic ux density) v ector and ~ d a is a dif ferential area on an arbitrary open surf ace S with ~ d a being normal to S ~ B is e xpressed in units of Wb m 2 or T. Therefore, F is equal to the surf ace inte gral of the magnetic induction o v er the open surf ace S If the open surf ace S is dened by the closed path C then Equation 3-73 can be re-written as I C ~ E ~ d s = d d t Z S ~ B ~ d a (3-75) Moreo v er if the medium is stationary then both C and S will be stationary and the e xpression can be further simplied further to I C ~ E ~ d s = Z S d ~ B d t ~ d a (3-76) Hence, if the medium is stationary then the line inte gral of the electric eld about the closed path C is equal to the ne g ati v e of the surf ace inte gral (tak en o v er the surf ace dened by C ) of the time-deri v ati v e of the magnetic induction. If the medium is linear homogeneous, and non-conducting (e.g. air), the magnetic induction, ~ B can be e xpressed as the magnetic eld intensity ~ H times the permeability of the medium, If the medium is air the v alue of is approximately 0 = 4 p 10 7 H m 1 which is the permeability of free space. Since ~ B and ~ H dif fer by only a constant in air the term “magnetic eld” is used to refer to either ~ H or ~ B

PAGE 96

82 If the magnetic eld is uniform o v er the entire surf ace and the entire surf ace lies in the same plane, then the e xpression further simplies to I C ~ E ~ d s = A l oo p d d t ~ B ~ n = A l oo p d B nor m d t (3-77) The quantity A l oo p is the total area of the surf ace and ~ n is the unit v ector normal to the surf ace. Hence, the line inte gral of the electric eld about the closed path C is equal to the ne g ati v e of the area of the surf ace bound by C times the time-deri v ati v e of the magnitude of the normal component of magnetic eld through the surf ace. The abo v e e xpression is only v alid if the magnetic eld is uniform o v er the entire surf ace of interest and the surf ace lies in a single plane. If ~ B is a component of an electromagnetic w a v e, then the e xpression is only v alid if the longest dimension of the surf ace is much smaller than a quarter of a w a v elength. The term on the left-hand side of Equation 3-77 is dened as the electromoti v e force, or EMF and is e xpressed in units of V. This can be interpreted by saying that if a perfectly conducting wire placed along the path C is brok en, the v oltage measured between the tw o open ends of the wire is equal the time-deri v ati v e of the normal component of magnetic eld through the surf ace bound by the wire times the area bounded by the wire. Hence, a loop of wire can be used to sense the component of the magnetic eld which is normal to the plane of the loop. The open circuit v oltage, v l oo p ( t ) of a brok en loop of wire in the presence of a time-v arying magnetic eld is gi v en by v l oo p ( t ) = I C ~ E ~ d s = A l oo p d B nor m ( t ) d t (3-78) A wire-loop antenna in the presence of a uniform time-v arying magnetic eld can be vie wed as a v oltage source whose magnitude is proportional to the time deri v ati v e of the component of the magnetic eld which is normal to the plane of the loop. This is the The v enin equi v alent v oltage of a loop antenna and the basis of the equi v alent

PAGE 97

83 Z S Z L V loop (w) Figure 3–22. Frequenc y-domain equi v alent circuit, using a The v enin equi v alent v oltage source, of a loop antenna sensor feeding a load (represented by Z L ). circuit. As with the electric-eld antenna, the circuit analysis for the magnetic eld antenna will be performed in the frequenc y domain. The frequenc y domain The v enin equi v alent (open circuit) v oltage source of a wire loop antenna is gi v en by V l oo p ( w ) = j w A l oo p B nor m ( w ) (3-79) The v oltage source is in series with source impedance, Z S and load impedance, Z L as sho wn in Figure 3–22 The source impedance is the impedance of the antenna itself and the load impedance is the impedance of an y e xternal elements connected to the loop. In general, the source and load impedances can be resisti v e, capaciti v e, inducti v e, or a combination of the three. If the output of the antenna is tak en as the v oltage across the load impedance, then the e xpression for the output v oltage in the frequenc y domain is V ou t ( w ) = Z L Z S + Z L V l oo p ( w ) = Z L Z S + Z L j w A l oo p B nor m ( w ) (3-80) The output v oltage of the antenna in the frequenc y domain is j w A l oo p B nor m ( w ) scaled by the frequenc y-dependent quantity Z L = ( Z S + Z L ) The frequenc y-independent g ain of the antenna is A l oo p

PAGE 98

84 T ypically the source impedance is considered a resistance, R l oo p in series with an inducti v e reactance, j w L l oo p Therefore, the source impedance is Z S = R l oo p + j w L l oo p (3-81) The load impedance is considered a resistance, R l oad and hence Z L = R l oad (3-82) Substituting the e xpressions for the source and load impedances (Equations 3-81 and 3-82 ) into the e xpression for the frequenc y-domain output v oltage of the loop antenna (Equation 3-80 ) yields V ou t ( w ) = A l oo p R l oad j w L l oo p + R l oo p + R l oad j w B nor m ( w ) (3-83) As with the at-plate antenna, the response of the loop antenna in the frequenc y domain depends on whether the antenna is to be vie wed as a magnetic eld sensor or a magnetic eld time-deri v ati v e sensor Unlik e the at-plate antenna, only the time-deri v ati v e perspecti v e will be considered for the loop antenna, since acti v e inte gration will be used to e xtend the loop antenna into a magnetic eld sensor From the dB/dt antenna perspecti v e, the magnitude response of the antenna, G d B d t ( w ) is gi v en by G d B d t ( w ) = V ou t ( w ) j w B nor m ( w ) = A l oo p R l oad j w L l oo p + R l oo p + R l oad (3-84) Lik e the dE/dt antenna, the magnitude response of the dB/dt antenna is that of a rst order lo w-pass lter The magnitude response is gi v en by G d B d t ( w ) = A l oo p R l oad R l oo p + R l oad 0BBBB@ 1 1 + w L l oo p R l oo p + R l oad 2 1 2 1CCCCA (3-85)

PAGE 99

85 The pass-band g ain is equal to ( A l oo p R l oad ) = ( R l oo p + R l oad ) The -3 dB point (the frequenc y at which the output is approximately 0.707 times the output in the pass-band) of the magnitude response is w 0 = R l oo p + R l oad L l oo p (3-86) If the frequenc y range of interest lies f ar belo w the -3 dB point, then the e xpression for the response of the loop antenna in the frequenc y domain becomes G d B d t ( w ) = A l oo p R l oad R l oo p + R l oad (3-87) The abo v e e xpression is only v alid for frequencies which satisfy the condition w w 0 = R l oo p + R l oad L l oo p (3-88) If the same condition is applied to Equation 3-83 then the output v oltage of the loop antenna in the frequenc y domain becomes V ou t ( w ) = A l oo p R l oad R l oo p + R l oad j w B nor m ( w ) (3-89) T o nd the output v oltage in the time domain, the in v erse F ourier transform (dened in Equation 3-23 ) is used. The inte gral need not be computed in this case since the dif ferentiation property of the F ourier transform can be used. Therefore, the e xpression for the output v oltage in the time domain is v ou t ( t ) = A l oo p R l oad R l oo p + R l oad d B nor m ( t ) d t (3-90) Although this is a time-domain e xpression, the frequenc y constraint still applies since the time-domain e xpression is deri v ed from a frequenc y-domain e xpression with that constraint. In other w ords, the abo v e e xpression is only v alid if the dB/dt w a v eform has no signicant frequenc y content abo v e w 0

PAGE 100

86 The general time-domain e xpression for v ou t ( t ) can be found by performing the in v erse F ourier transform on Equation 3-83 Equation 3-83 can be rearranged to yield V ou t ( w ) = 1 R l oo p + R l oad L l oo p + j w R l oad A l oo p L l oo p j w B nor m ( w ) (3-91) This e xpression can be vie wed as the multiplication of tw o functions of w This can be e xpressed as V ou t ( w ) = X ( w ) Y ( w ) (3-92) The quantities X ( w ) and Y ( w ) are dened as X ( w ) = 1 R l oo p + R l oad L l oo p + j w (3-93) Y ( w ) = R l oad A l oo p L l oo p j w B nor m ( w ) (3-94) Therefore, the time-domain e xpression for the antenna output v oltage, v ou t ( t ) can be found by the con v olution property of the F ourier transform. v ou t ( t ) = x ( t ) y ( t ) (3-95) The operator denotes linear con v olution and is e v aluated using the con v olution inte gral. The quantities x ( t ) and y ( t ) denote the in v erse F ourier transforms of X ( w ) and Y ( w ) respecti v ely The in v erse F ourier transform of X ( w ) can be found by using a F ourier transform table. x ( t ) = e R l oo p + R l oad L l oo p t u ( t ) (3-96) The quantity u ( t ) is the unit-step function, as dened by Equation 3-39 The in v erse F ourier transform of Y ( w ) is found by in v oking the dif ferentiation property of the F ourier transform. y ( t ) = R l oad A l oo p L l oo p d B nor m ( t ) d t (3-97)

PAGE 101

87 Substituting the e xpressions for x ( t ) and y ( t ) into 3-95 yields v ou t ( t ) = e R l oo p + R l oad L l oo p t u ( t ) R l oad A l oo p L l oo p d B nor m ( t ) d t (3-98) The e xpression for v ou t ( t ) is e v aluated using the con v olution inte gral as dened in Equation 3-36 v ou t ( t ) = R l oad A l oo p L l oo p Z l = d B nor m ( l ) d l e R l oo p + R l oad L l oo p ( t l ) u ( t l ) d l (3-99) The ef fect of the unit-step function can be incorporated into the limits of inte gration and the term e R l oo p + R l oad L l oo p t can be f actored out of the inte gral. This yields v ou t ( t ) = R l oad A l oo p L l oo p e R l oo p + R l oad L l oo p t Z t l = d B nor m ( l ) d l e R l oo p + R l oad L l oo p l d l (3-100) The inte gral is e v aluated via the inte gration by parts technique with the follo wing parameters. Z ud v = Z d B nor m ( l ) d l e R l oo p + R l oad L l oo p l d l = uv Z vd u (3-101) u = d B nor m ( l ) d l d u = d 2 B nor m ( l ) d l 2 d l (3-102) d v = e R l oo p + R l oad L l oo p l d l v = L l oo p R l oo p + R l oad e R l oo p + R l oad L l oo p l (3-103) Therefore, the inte gral becomes uv Z vd u = L l oo p R l oo p + R l oad e R l oo p + R l oad L l oo p l d B nor m ( l ) d l Z L l oo p R l oo p + R l oad e R l oo p + R l oad L l oo p l d 2 B nor m ( l ) d l 2 d l (3-104)

PAGE 102

88 T o e v aluate the inte gral in the second term, inte gration by parts is ag ain used and will result in another e xpression of a similar form containing a higher -order deri v ati v e and another inte gral. This result can be e xpressed as the follo wing innite sum. Z d B nor m ( l ) d l e R l oo p + R l oad L l oo p l d l = e R l oo p + R l oad L l oo p l X n = 1 ( 1 ) n + 1 L l oo p R l oo p + R l oad n d ( n ) B nor m ( l ) d l ( n ) # (3-105) The term d ( n ) B nor m ( l ) = d l ( n ) denotes the n t h deri v ati v e of B nor m ( l ) with respect to l Substituting the abo v e e xpression into Equation 3-100 and reintroducing the limits of inte gration yields v ou t ( t ) = R l oad A l oo p L l oo p e R l oo p + R l oad L l oo p t (3-106) e R l oo p + R l oad L l oo p l X n = 1 ( 1 ) n + 1 L l oo p R l oo p + R l oad n d ( n ) B nor m ( l ) d l ( n ) ## t l = If d ( n ) B nor m ( l ) = d l ( n ) < for all n then the term resulting from the lo wer limit of inte gration will be zero. Therefore, the e xpression for v ou t ( t ) becomes v ou t ( t ) = R l oad A l oo p L l oo p e R l oo p + R l oad L l oo p t e R l oo p + R l oad L l oo p t X n = 1 ( 1 ) n + 1 L l oo p R l oo p + R l oad n d ( n ) B nor m ( t ) d t ( n ) # (3-107) The multiplication of the tw o e xponential terms results in unity therefore v ou t ( t ) = R l oad A l oo p L l oo p X n = 1 ( 1 ) n + 1 L l oo p R l oo p + R l oad n d ( n ) B nor m ( t ) d t ( n ) # (3-108) Hence, the time-domain v oltage output of a loop antenna is the weighted sum of all of the time-deri v ati v es of B nor m ( t ) If the rst term in the summation is e xpanded,

PAGE 103

89 the e xpression becomes v ou t ( t ) = R l oad A l oo p L l oo p L l oo p R l oo p + R l oad d B nor m ( t ) d t R l oad A l oo p L l oo p X n = 2 ( 1 ) n + 1 L l oo p R l oo p + R l oad n d ( n ) B nor m ( t ) d t ( n ) # (3-109) Moreo v er Equation 3-109 can be further simplied to v ou t ( t ) = R l oad A l oo p R l oo p + R l oad d B nor m ( t ) d t R l oad A l oo p R l oo p + R l oad X n = 2 L l oo p R l oo p + R l oad n 1 d ( n ) B nor m ( t ) d t ( n ) # (3-110) The rst term in Equation 3-110 is identical to Equation 3-90 ; the time-domain output of the loop antenna obtained by taking the in v erse F ourier transform of Equation 3-89 assuming w w 0 The second term is the manifestation of the upper frequenc y response limit in the time domain. The second term will become zero if L l oo p = 0, which corresponds to w 0 = 3.3.2.2 Loop antenna implementation No w that the output of the loop antenna has been characterized in the time and frequenc y domains, the implementation of the MSE magnetic eld and magnetic eld time-deri v ati v e antennas can be discussed. Both the magnetic eld and the magnetic eld time-deri v ati v e measurements utilize square loops of 50 W coaxial cable as sho wn in Figure 3–23 The coaxial cable is placed in 3 4 inch PVC pipe to help k eep a rigid shape. The inner conductor of the cable is the actual wire comprising the loop antenna. The outer shield is necessary to k eep current from being induced on the inner conductor by an e xternal electric eld. The outer shield of the cable can be thought of another loop of wire placed at almost e xactly the same spatial location as the inner conductor Therefore, identical v oltages will be induced on the inner conductor and the shield of the coaxial cable. As discussed in the pre vious section, the v oltage output of a wire-loop antenna is

PAGE 104

90 AB Figure 3–23. Square loops of 50 W coaxial cable in 3 4 inch PVC pipe. A) Single loop at Station 4. B) Crossed loops at Station 4.

PAGE 105

91 measured between the tw o ends of the wire with the e xpression describing the output v oltage gi v en by Equation 3-110 In practice, since the antenna is constructed from coaxial cable, it is con v enient to use coaxial cable connectors (such as BNC or SMA connectors) and measure the v oltage dif ference across the connectors at the tw o ends of the cable. The output v oltage w ould then be the dif ference between the measured v oltages across the connectors at the tw o ends of the cable. v ou t ( t ) = ( v ic 1 ( t ) v oc 1 ( t ) ) ( v ic 2 ( t ) v oc 2 ( t ) ) (3-111) The subscripts ic and oc refer to the inner and outer conductor respecti v ely The subscripts 1 and 2 refer to the tw o ends of the coaxial cable. This conguration is the basis of a dif ferential output coaxial loop antenna. Ho we v er this conguration poses a problem since Equation 3-111 can be rearranged to yield v ou t ( t ) = ( v ic 1 ( t ) v ic 2 ( t ) ) ( v oc 1 ( t ) v oc 2 ( t ) ) (3-112) As mentioned pre viously the induced v oltage on the inner conductor and the outer shield are almost identical, therefore the output v oltage of the dif ferential coaxial antenna, as sho wn in Equation 3-112 w ould be close to zero. Ho we v er if the shield from the tw o ends of the cable is soldered together at the output of the antenna, then v oc 1 ( t ) will be equal to v oc 2 ( t ) and Equation 3-112 reduces to v ou t ( t ) = v ic 1 ( t ) v ic 2 ( t ) (3-113) This is simply the v oltage dif ference between the tw o ends of the inner conductor of the coaxial cable. Although soldering the shield together at the base of the antenna alle viates one problem, it introduces another Once the shield is soldered together the outer shield of the cable forms a closed loop. If an y current is induced on this loop by either an e xternal electric or magnetic eld, an unw anted magnetic eld will necessarily

PAGE 106

92 Input Terminal Input Terminal Differential Output Voltage Taken Across Inner Conductor Gap in Shield Shield Soldered Together Coaxial Cable Shield (Outer Conductor) Load Resistors Loop Resistance Differential Amplifier Input External A B 100 W Load R l oo p R l oad100 W 50 W V l oo p L l oo p R l oo p 50 W 50 W 50 W Figure 3–24. Diagram (A) and equi v alent circuit (B) of a dif ferential-output coaxial loop antenna with both ends of the cable terminated in 50 W be induced perpendicular to the loop. This induced magnetic eld will distort the e xternal magnetic eld to be sensed. Therefore, a small g ap is placed in the shield to inhibit an y shield current and hence pre v ent an y unw anted magnetic elds from the shield. T ypically this g ap is placed at the top of the loop, as pictured in Figure 3–24 In practice, each end of the coaxial cable is terminated in its characteristic impedance, which is 50 W for all MSE loop antennas. This termination usually tak es the form of the input resistance of the inputs of a dif ferential amplier that the the antenna is connected to. Since the output is tak en across both ends of the cable, the load resistance, R l oad is 100 W as sho wn in Figure 3–24

PAGE 107

93 The corresponding e xpressions for the ideal time-domain output v oltage and -3 dB bandwidth of a dif ferential output coaxial antenna with both ends terminated in 50 W are v ou t ( t ) = 100 A l oo p R l oo p + 100 d B nor m ( t ) d t (3-114) w 0 = R l oo p + 100 L l oo p (3-115) The v alue of L l oo p is determined by the geometry of the antenna and R l oo p the total resistance of the loop antenna including its inherent resistance and an y e xternally added resistance, is determined by the desired bandwidth of the antenna. If no e xternal resistance is added to the antenna, R l oo p is the resistance of the inner conductor of the cable, which is v ery close to zero. Both R l oo p and L l oo p af fect the g ain and bandwidth of the antenna, as sho wn by Equations 3-114 ( L l oo p is a function of A l oo p ), and 3-115 respecti v ely Decreasing R l oo p or increasing A l oo p (since increasing A l oo p increases L l oo p ) increases the g ain of the antenna. Ho we v er these same modications decrease the bandwidth of the antenna. Therefore, there is a g ain/bandwidth trade-of f associated with a coaxial loop antenna. A coaxial loop antenna can also ha v e a single-ended output, meaning the output v oltage is tak en from only one end of the cable. Lik e the dif ferential-output antenna, each end of the cable is terminated in its characteristic impedance, which is 50 W for all MSE loop antennas. Ho we v er the output v oltage is measured across only one end of the cable and the 50 W termination on that end of the cable tak es the form of the input resistance of an amplier The other end of the cable is terminated by soldering a 50 W resistor between the inner conductor and the outer shield. This soldering is usually located at the base of the antenna. This conguration is sho wn in Figure 3–25 The equi v alent circuit is the same as the dif ferential-output antenna, ho we v er the output is tak en across only one of the 50 W resistors, hence the g ain of the antenna is

PAGE 108

94 Voltage Taken Across SingleEnded Output Gap in Shield Inner Conductor Load Resistors Loop Resistance (Outer Conductor) Coaxial Cable Shield Shield Soldered Together Amplifier Input Single Ended to Shield External Resistor Soldered A B Terminal Input 50 W Resistor 50 W 50 W R l oo p 50 W V l oo p L l oo p R l oo p R l oad100 W 50 W Figure 3–25. Diagram (A) and equi v alent circuit (B) of a single-ended output coaxial loop antenna with both ends of the cable terminated in 50 W halv ed. Ho we v er R l oad is still considered to be 100 W since both resistors af fect the bandwidth. Therefore, the e xpression for the ideal time-domain output v oltage of a single ended coaxial loop antenna is v ou t ( t ) = 1 2 100 A l oo p R l oo p + 100 d B nor m ( t ) d t (3-116) The -3 dB bandwidth of the antenna is the same as for the dif ferential-output loop antenna, gi v en in Equation 3-115 It should be noted that a measurement utilizing a single-ended conguration is some what easier to implement than a dif ferential conguration since a dif ferential amplier is not required. Ho we v er the adv antage of the dif ferential conguration is that it has high common-mode noise rejection.

PAGE 109

95 3.3.2.3 Magnetic eld measur ement implementation The MSE magnetic eld antennas were square single-ended coaxial loop antennas of area 0 : 533 m 2 terminated in 50 W The antennas were arranged in orthogonal crossed-loop pairs as sho wn in Figure 3–23 No e xternal resistance w as added to the loop, hence R l oo p is approximately zero. R l oad as stated pre viously w as 100 W due to the 50 W terminations at each end of the cable. The output of the loop w as tak en from the base of the antenna and connected to a female BNC feed-through connector mounted to the side of a Hof fman box located on the ground approximately one meter a w ay In order to eliminate ground loops, all electronic components were isolated from each other and the metal box by pieces of plastic and Styrofoam. The end of the feed-through connector inside of the box w as connected, via a short length of 50 W coaxial cable, to the input of an acti v e inte grator with input resistance 50 W This 50 W input resistance is the termination of the output end of the antenna coaxial cable. As stated pre viously the 50 W termination of the other end of the cable consists of a resistor soldered directly into the cable at the base of the antenna. Therefore, assuming that the frequenc y range of interest lies within the re gion where the response of the antenna is at and substituting R l oo p = 0 into Equation 3-116 the time-domain output v oltage of this antenna is gi v en by v ou t ( t ) = 1 2 A l oo p d B nor m ( t ) d t (3-117) The inductance of this loop is approximately 4 H and the corresponding -3 dB bandwidth is w 0 = 2 : 5 10 7 s 1 or f 0 = 4 MHz. The acti v e inte grator is an essential part of a magnetic eld measurement for it inte grates the v oltage output of the loop antenna (which is proportional to dB/dt) and gi v es an output which is proportional to the magnetic eld. Ideally v in t ( t ) = Z t l = 0 v in ( t ) d l (3-118)

PAGE 110

96 + R 1 R 2 C 1 R 3 R out Input Output Figure 3–26. Schematic of an e xample acti v e inte grator In practice, acti v e inte grators ne v er ha v e an ideal response. Figure 3–26 sho ws an e xample acti v e inte grator An ideal acti v e inte grator w ould only ha v e a capacitor in the feedback loop of the op-amp, ho we v er the resistor is required to ha v e a DC feedback path. If this were not present, the acti v e inte grator w ould inte grate the small DC of fset v oltage present at the input of the op-amp until the output saturated. This feedback resistor has the ef fect of limiting the lo w-frequenc y response of the acti v e inte grator which can be vie wed in the time domain as introducing a decay time constant into the system. The reader can easily v erify using simple circuit analysis, that the decay time constant of the e xample acti v e inte grator is gi v en by t = R 2 C 1 (3-119) In addition, the output of an acti v e inte grator is not just the inte gral of the input, b ut the inte gral of the input multiplied by a constant. This constant is referred to as the inte gration constant, k in t and is e xpressed in units of s 1 F or the e xample inte grator

PAGE 111

97 k in t is gi v en by k in t = 1 R 1 C 1 (3-120) The upper -frequenc y limit of the acti v e-inte grator is typically determined by the op-amp. The upper and lo wer limits of the frequenc y response are the points in which the magnitude response of the inte grator de viates from k in t = w (the ideal magnitude response of an acti v e inte grator). Therefore, the output of the acti v e inte grator sho wn in Figure 3–26 assuming an ideal op-amp, is gi v en by v in t ( t ) = k in t Z t l = 0 v in ( t ) d l e t t = 1 R 1 C 1 Z t l = 0 v in ( t ) d l e t R 2 C 1 (3-121) Analogous to the passi v e inte gration used with a at-plate electric eld antenna, for times t t the time domain output of the acti v e inte grator reduces to v in t ( t ) = k in t Z t l = 0 v in ( t ) d l = 1 R 1 C 1 Z t l = 0 v in ( t ) d l (3-122) The 2001 and 2002 MSE utilized dif ferent models of acti v e inte grators, both of which were designed and b uilt by Geor ge Schnetzer The schematic of the 2001 acti v e inte grator is sho wn in Figure 3–27 This model is the same inte grator that w as used in the natural lightning magnetic eld measurements described in ( Cra wford et al. 2001 ). The inte gration constant v aried for each of the units b uilt, b ut is approximately equal to 1 : 96 10 4 s 1 : The ne g ati v e sign deri v es from the f act that the inte grator consists of only a single in v erting stage. The decay time constant, t of the 2001 acti v e inte grator is approximately 5 : 1 ms. The input and output resistance of the acti v e inte grator is 50 W The 2001 acti v e inte grator is po wered by a 12 V battery and a zener diode/resistor circuit is used to supply approximately 5 V to the LM318 op-amp. Ho we v er this forces the signal

PAGE 112

98 + Input Output LM318 51 W 51 W 5.1 k W 10 k W 510 k W 0.1 m F 0.01 m F 1.0 m F 1.0 m F 33 n F Figure 3–27. Schematic of the acti v e inte grator used in the 2001 MSE. ground to oat about 6 V abo v e the ne g ati v e battery terminal. This is a similar problem to that of the amplier described in Section 3.3.1.3 Therefore, tw o 1 : 0 F capacitors are used to A C couple the input and the output. These capacitors could ha v e been eliminated if the inte grator w as po wered by a separate battery than the rest of the electronics in the measurement, similar to what w as done with the amplier These acti v e inte grators were designed and constructed in 1997 and that option w as not practical at the time. Furthermore, in 2001, it w as e v entually deemed necessary to po wer the acti v e inte grator from a separate battery despite the f act that the input and output were already A C coupled. When all of the electronics, including the inte grator were po wered from the same battery a small v ariation of the supply v oltage of the inte grator w as observ ed when the ber -optic transcei v er pair located in the PIC controller w as acti v e. This v oltage v ariation manifested itself at the output of the inte grator Therefore, the 2001 acti v e inte grator needed to be po wered by a separate battery Lik e the amplier used in the electric eld measurements, a relay w as used to connect the po wer terminals of the acti v e inte grator to its battery when the PIC controller supplied po wer to the ber -optic transmitter The measured and ideal magnitude responses of one of the 2001 acti v e inte grator units are sho wn in Figure 3–28

PAGE 113

99 0.0001 0.001 0.01 0.1 1 10 0 10 100 1000 10000 5 10 6 10 6 5x10 Response Frequency [Hz] Frequency Response of a 2001 Active Integrator Unit 0.0001 0.001 0.01 0.1 1 10 0 10 100 1000 10000 5 10 6 10 6 5x10 Response Frequency [Hz] Frequency Response of a 2001 Active Integrator Unit Figure 3–28. Measured (dashed line) and ideal (solid line) magnitude responses of one of the 2001 MSE acti v e inte grator units. The lo wer and upper frequenc y limits were measured to be approximately 100 Hz and 2 MHz, respecti v ely In 2002, it w as desired to increase the decay time constant and upper frequenc y limit of the acti v e inte grator Therefore, another acti v e inte grator w as used. This is the same inte grator that w as used in the 2000 e xperimental conguration described in ( Rak o v et al. 2001 ) and the schematic is sho wn in Figure 3–29 As with the 2001 inte grator the input and output resistances are both 50 W Unlik e the 2001 inte grator the 2002 inte grator consists of tw o in v erting states. Hence, the inte gration constant is positi v e. The rst stage performs the actual inte gration while the second stage pro vides a g ain of ten. The tw o stages are A C coupled by a 22 F capacitor The v alue of k in t for each unit is positi v e and approximately equal to 2 : 5 10 5 s 1 This v alue tak es into account the g ain of ten introduced by the second stage.

PAGE 114

100 + Input Output LH0032CG 51 W 5.1 k W 2 k W 1 k W 1 M W 27 pF 0.02 m F 22 m F + 51 k W 51 W LH0032CG 1 k W 27 pF Figure 3–29. Schematic of the acti v e inte grator used in the 2002 MSE. The decay time constant for each unit w as measured to be approximately 15 ms. The lo wer frequenc y limit of the inte grator which w as measured to be about 10 Hz, is determined by both the rst stage and the A C coupling between the rst and second stages. The upper frequenc y limit w as measured to be approximately 5 MHz. This acti v e inte grator w as also po wered by a single 12 V battery b ut a DC-DC con v erter w as used to obtain 12 V needed to po wer the op-amp. This eliminated the problem with the v oltage uctuation observ ed at the output that plagued the 2001 inte grators. Therefore, the 2002 inte grators could be po wered by the same battery that po wered the rest of the electronics. In either case, the output of the acti v e inte grator w as connected to the input of a PIC controller by a short length of 50 W coaxial cable. The output of the PIC controller w as then connected, via another short length of cable, to the input of an Opticomm MMV -120C ber -optic transmitter F or the 2002 e xperiment, the output of the PIC controller w as terminated in 50 W while in 2001 it w as terminated directly into ber -optic transmitter which has an input resistance of 68 k W The 2001 inte grator could not be terminated in 50 W because the LM318 op-amp could not dri v e a 50 W load. The equi v alent circuit of the PIC controller is a 50 W in-line attenuator of v alue G PI C As described in Section 3.2.1 this equi v alent circuit only applies if the output

PAGE 115

101 PIC Controller Supplies Po wer to Fiber -optic T ransmitter Loop Antenna 12 V Battery Controller PIC PIC Controller Fiber -optic Cable Metal Enclosure T ransmitter OpticommFiber -optic 12 V Battery Inte grator Acti v e PIC Controller Supplies Po wer to Inte grator Relay k in t G PIC Inte grator Acti v e Figure 3–30. Diagram of a 2001 MSE magnetic eld measurement. of the PIC controller is terminated in 50 W Therefore, for the 2001 e xperiment, the programmable attenuators could not be used and hence G PI C = 1 Ho we v er this is not the case for the 2002 e xperiment, and the programmable attenuators in the PIC controller were used. Diagrams of the 2001 and 2002 MSE magnetic eld measurements are presented in Figures 3–30 and 3–31 respecti v ely If it is assumed that magnitude response of the antenna is at and the acti v e inte grator inte grates properly o v er the frequenc y range of interest of the magnetic eld to be sensed, then the e xpression for the v oltage present at the input of the ber -optic transmitter v F O T ( t ) is gi v en by v F O T ( t ) = 1 2 A l oo p k in t G PI C R F O T R ou t + R F O T B nor m ( t ) (3-123) The term R F O T = ( R ou t + R F O T ) is the v oltage di vision between the output resistance of the acti v e inte grator R ou t and the input resistance of the ber -optic transmitter or in-line terminator R F O T The output of the ber -optic recei v er w as connected to a digitizer terminated in 50 W The v oltage at the input of the digitizing oscilloscope is

PAGE 116

102 PIC Controller Supplies Po wer to Fiber -optic T ransmitter Loop Antenna 50 W 50 W In-line T erminator Inte grator Acti v e PIC Controller Fiber -optic Cable Metal Enclosure 12 V Battery Controller PIC T ransmitter OpticommFiber -optic k in t G PIC PIC Controller Supplies Po wer to Acti v e Inte grator Figure 3–31. Diagram of a 2002 MSE magnetic eld measurement. the v oltage present at the input of the ber -optic transmitter modied by the ber -optic link. If the ef fect of the link is assumed to be a frequenc y independent g ain or attenuation (of v alue G l ink ), then the e xpression for the v oltage at the input of the oscilloscope, v sco pe ( t ) is v sco pe ( t ) = 1 2 A l oo p k in t G PI C G l ink R F O T R ou t + R F O T B nor m ( t ) (3-124) F or the 2001 e xperiment, A l oo p = 0 : 533 m 2 G PI C = 1, k in t = 1 : 96 10 4 s 1 and G l ink will be assumed to ha v e a nominal v alue of one. The quantity R F O T is 68 k W the input resistance of the Opticomm MMV -120C ber -optic transmitter and R ou t is 50 W Therefore, the quantity R F O T = ( R ou t + R F O T ) is approximately unity The e xpression for the nominal v oltage seen at the input of the digitizer during the 2001 e xperiment, when B nor m ( t ) is e xpressed in units of Wb m 2 is gi v en by v sco pe nom ( t ) = ( 5220 ) B nor m ( t ) (3-125)

PAGE 117

103 If v sco pe nom = 1 V and Equation 3-125 is solv ed for B nor m ( t ) then B nor m = 1 : 92 10 4 Wb m 2 V 1 (3-126) If B nor m is e xpressed in units of Wb m 2 then Equation 3-126 becomes B nor m = 192 Wb m 2 V 1 (3-127) Thus, nominally one v olt present at the input of the digitizer corresponds to a magnetic eld amplitude of about 190 Wb m 2 The e xact same loop antennas were used for the 2002 e xperiment, hence ag ain A l oo p = 0 : 533 m 2 G PI C w as set to approximately 0.316 (-10 dB). Ag ain, G l ink is assumed to ha v e a nominal v alue of one. Also, as described abo v e, a dif ferent type of acti v e inte grator w as used, ha ving a nominal inte gration constant, k in t of approximately 2 : 5 10 5 s 1 Unlik e the 2001 conguration, the output of the acti v e inte grator w as terminated, through the PIC controller in 50 W and hence R F O T = 50 W Since R ou t = 50 W R F O T = ( R ou t + R F O T ) = 0 : 5. Therefore, the e xpression for the nominal v oltage seen at the input of the digitizer during the 2002 e xperiment, when B nor m ( t ) is e xpressed in units of Wb m 2 is gi v en by v sco pe nom ( t ) = ( 10500 ) B nor m ( t ) (3-128) If v sco pe nom = 1 V and Equation 3-128 is solv ed for B nor m ( t ) then B nor m = 9 : 5 10 5 Wb m 2 V 1 (3-129) If B nor m is e xpressed in units of Wb m 2 then Equation 3-129 becomes: B nor m = 95 Wb m 2 V 1 (3-130)

PAGE 118

104 T able 3–9. Salient characteristics of the 2001 MSE magnetic eld measurements. k in t t ( ms ) Units/V olt Designation s 1 (approximate) Wb m 2 V 1 B-4N 1 : 96 10 4 5 : 1 192.0 B-4E 1 : 96 10 4 5 : 1 192.0 B-9N 1 : 96 10 4 5 : 1 192.0 B-9E 1 : 98 10 4 5 : 1 190.0 T able 3–10. Salient characteristics of the 2002 MSE magnetic eld measurements. k in t t ( ms ) Units/V olt Designation s 1 (approximate) Wb m 2 V 1 B-4N 2 : 60 10 5 15 -92.3 B-9N 2 : 35 10 5 15 -102.4 Thus, nominally one v olt present at the input of the digitizer corresponds to a magnetic eld amplitude of about 95 Wb m 2 T ables 3–9 and 3–10 summarize the salient characteristics of all of the indi vidual 2001 and 2002 MSE magnetic eld measurements, including the inte gration constant, measured decay time constant, and the corresponding units per v olt (assuming G l ink = 1). Opticomm MMV -120C ber -optic links were used to transmit the analog w a v eforms to the Launch Control trailer where the y were digitized. The magnetic eld measurements were digitized continuously for 800 ms (200 ms pre-trigger) at 10 MHz on a Y ok og a w a DL716 digital storage oscilloscope. The w a v eforms were band-limited to 4 MHz (-3 dB) by the anti-aliasing lter associated with the DL716 digitizer Detailed descriptions of the Opticomm ber -optic links and the DL716 digitizer are gi v en in Sections 3.4.1 and 3.5.1 respecti v ely 3.3.2.4 Magnetic eld time-deri v ati v e measur ement implementation The MSE dB/dt measurements utilized square loops of 50 W coaxial cable with dif ferential outputs, as discussed in Section 3.3.2.2 T w o of the antennas consisted of a single loop while a third consisted of an orthogonal crossed loop pair yielding a total

PAGE 119

105 of four loops. The single-loop antennas ha v e ph ysical areas of approximately 0 : 120 m 2 and 0 : 108 m 2 These antennas were used to measure measure dB/dt in pre vious triggered lightning e xperiments, such as those described in ( Uman et al. 2002 ) and ( Schoene et al. 2003 ). The crossed loop antenna w as constructed during the summer of 2002 and each orthogonal loop has a ph ysical area of approximately 0 : 151 m 2 Unlik e the loops used for the magnetic eld measurements, a 470 W resistor w as soldered into the loop in order to increase the bandwidth of the sensor at the e xpense of g ain. Hence, R l oo p = 470 W Lik e the magnetic eld antennas, each end of the cable w as terminated in 50 W yielding R l oad = 100 W Therefore, when dB/dt is e xpressed in units of Wb m 2 s 1 (or in equi v alent units of Wb m 2 s 1 ), the time-domain output v oltage of this antenna is gi v en by v ou t ( t ) = ( 0 : 175 ) A l oo p d B nor m ( t ) d t (3-131) The inductance of each of the loops is approximately 1 : 2 H and the corresponding -3 dB bandwidth is w 0 = 4 : 75 10 8 s 1 or f 0 75 MHz. The dif ferential outputs of the loop were tak en from the base of the antenna and the end of each cable w as connected to a female BNC b ulkhead feed-through connector mounted to the side of a Hof fman box located on the ground approximately one meter a w ay from the antenna. The tw o lengths of cable were twisted together so that no v oltage could be induced on the cable by the magnetic eld. Inside of the box, one of the feed-through connectors w as connected directly to a 50 W in-line attenuator of v alue G a t t whose output w as connected to the input of a PIC controller by a short length of 50 W coaxial cable. The output of the PIC controller w as connected to the non-in v erting input of a Meret dif ferential input ber -optic transmitter with an input resistance of 50 W The other BNC feed-through connector w as connected directly to another 50 W in-line attenuator of v alue G a t t whose output w as connected to the in v erting input of the Meret ber -optic transmitter T ypically the PIC controller is

PAGE 120

106 Differential Output Loop Antenna Metal Enclosure PIC Controller 12 V Battery PIC Controller Fiberoptic Transmitter PIC Controller Supplies Power to Meret Fiberoptic Transmitter FiberopticCable G a t t G a t t G PI C1 50 W Attenuator Figure 3–32. Diagram of a MSE dB/dt measurement. used to pro vide programmable attenuation to a measurement; ho we v er it only has one input and is thus not compatible with sensors ha ving dif ferential outputs. While it w as possible to use tw o PIC controllers to pro vide attenuation, it pro v ed more cost ef fecti v e to use in-line attenuators. The disadv antage of this conguration is that an operator must ph ysically change the attenuators in order to manipulate the full-scale range of the measurement. The calibration function of the PIC controller ho we v er could still be used and this is wh y the output of the PIC controller is connected to the non-in v erting input of the Meret ber -optic transmitter A detailed description of the PIC controller is gi v en in Section 3.2.1 Figure 3–32 sho ws a diagram of a MSE dB/dt measurement. In order to eliminate ground loops, all electronic components were isolated from each other and the metal box by pieces of plastic and Styrofoam. If it is assumed that magnitude response of the antenna is at o v er the frequenc y range of interest, then the e xpression for the dif ferential v oltage present at the input of the ber -optic transmitter v F O T ( t ) is gi v en by v F O T ( t ) = ( 0 : 175 ) A l oo p G a t t d B nor m ( t ) d t (3-132)

PAGE 121

107 The output of the ber -optic recei v er is connected to the input of an A C coupled acti v e lo w-pass lter with an input resistance of 50 W by a length of 50 W coaxial cable. This lter is discussed in Section 3.4.3 The output of lter is connected to the input of a digitizing oscilloscope, terminated in 50 W by a short length of 50 W coaxial cable. The v oltage at the input of the digitizing oscilloscope is the v oltage present at the input of the ber -optic transmitter modied by the ber -optic link. If the ef fect of the link is assumed to be a frequenc y independent g ain or attenuation (of v alue G l ink ), then the e xpression for the v oltage at the input of the oscilloscope is v sco pe ( t ) = ( 0 : 175 ) A l oo p G a t t G l ink d B nor m ( t ) d t (3-133) During the 2002 season, G a t t w as set to approximately 0.500 ( -6 dB), hence the e xpression for nominal v oltage at the input to the digitizer v sco pe nom ( t ) is v sco pe nom = ( 0 : 0875 ) A l oo p G l ink d B nor m ( t ) d t (3-134) Unlik e the Opticomm MMV -120C and Nicolet Isobe 3000 ber -optic links, the Meret links, in this conguration, cannot be assumed to ha v e a nominal g ain of one. As discussed in Section 3.4.3 the g ain of the Meret links is strongly related to the quality of the ber termination with a poor termination resulting in high attenuation. Furthermore, the v alue of G l ink is dif ferent for each indi vidual dB/dt measurement and must be estimated e xperimentally Therefore, the e xpression for v sco pe ( t ) will be left in the form sho wn in Equation 3-134 The dB/dt measurements were digitized on a LeCro y L T344 W a v erunner digital storage oscilloscope in se gmented memory mode. The internal memory w as di vided into tw o se gments with each se gment requiring a separate trigger signal; hence data for a total of tw o return strok es could be acquired. Unlik e the measurements digitized by the DL716, no data were acquired during inter -strok e interv als. During July of 2002 each se gment w as digitized at 50 MHz for 10 ms (9 ms pre-trigger). In August of 2002

PAGE 122

108 the digitization rate and acquisition length per se gment were changed to 100 MHz and 5 ms (4 ms pre-trigger), respecti v ely The dB/dt w a v eforms were band-limited to 20 MHz (-3 dB) by A C coupled acti v e lo w-pass lters discussed in Section 3.4.3 A detailed description of the L T344 digitizer is gi v en in Section 3.5.2 Finally it is w orth noting that the ne g ati v e sign present in Equation 3-134 is some what arbitrary The polarity of the w a v eform recorded on the on the oscilloscope is related to the orientation of the loop antenna relati v e to the lightning channel. In addition, for a gi v en lightning location, re v ersing the connections to the dif ferential input ber -optic transmitter will also re v erse the w a v eform polarity The ne g ati v e sign w as k ept through the deri v ation for completeness. 3.3.3 Optical Measur ements T w o optical measurements were used in the MSE as a part of the system to trigger the digitizers to record lightning data. Since the optical measurements were originally intended to be only used as part of the digitizer triggering system, their outputs were not calibrated. There is ne v ertheless signicant scientic v alue in e v en the uncalibrated optical w a v eforms. One sensor w as placed at the north-east corner of the ICLR T site f acing south-west and the other w as placed on the south-west corner of the site f acing north-east. A detailed e xplanation of the MSE triggering system is gi v en in Section 3.2.3 The optical detectors themselv es are simple circuits consisting of a re v erse biased EG&G C30807E PIN photo-diode in series with a 1 k W resistor as sho wn in Figure 3–33 The output of the circuit is tak en across the resistor and is A C coupled by a 0 : 1 F capacitor The circuit is po wered by a 45 V battery yielding a measured quiescent current of a fe w A. The optical sensor circuit itself w as mounted ag ainst the inside of a metal Hof fman enclosure with a small hole drilled in it just lar ge enough for the lens of the PIN photo-diode. A c ylindrical piece of PVC of diameter 14 cm and length 6 : 5 cm w as

PAGE 123

109 1 k W 0.1 m F 45 V Output EG&G C30807E PIN PhotoDiode Figure 3–33. Schematic of the MSE optical sensor circuit. mounted on the outside of the Hof fman enclosure centered on the photo-diode. The inner w all of the PVC and the circular section of the Hof fman enclosure surrounding the photo-diode were painted black. A piece of Ple xiglas w as mounted to the open end of the PVC section and w as entirely co v ered by black electrical tape e xcept for an approximately 1 to 2 mm wide horizontal slit across the center The purpose of the slit w as to limit the amount of light recei v ed by the optical sensor and to limit the ele v ation angle from which light could be detected by the optical sensor Therefore, a v ery bright light source at a relati v ely lo w ele v ation angle w as required to produce a signicant v oltage at the output of the optical sensor The slit did not limit the azimuth angle at which light could hit the optical sensor although the length of the PVC pipe did. During the 2001 and 2002 seasons, the electrical tape comprising the slit w as replaced se v eral times due to deterioration from the rain and sunlight. In September of 2002, the electrical tape w as replaced with black paint, which is more resilient to the en vironment. The width of the slit w as ne v er measured precisely or f abricated uniform. Therefore it is impossible to calibrate the amplitude and w a v e shapes of the output of the optical sensors.

PAGE 124

110 Optical Sensor Metal Enclosure Mounted to PIC Controller 12 V Battery Amplifier PIC Controller Battery 12 V PIC Controller Supplies Power toAmplifier Relay PIC Controller Supplies Power to Fiberoptic Transmitter Transmitter Fiberoptic FiberopticCable HiZ Amplifier Metal Enclosure G am p10 G PI C1 Figure 3–34. Diagram of a MSE optical measurement. An y light shining on the lens of the PIN photo-diode will result in a current through it, with that current being superimposed on the quiescent current. The output of the sensor is tak en as the v oltage between the terminals of the resistor and is connected, through the capacitor to an amplier ha ving a g ain of ten and an input resistance of 1 M W This amplier is identical to that described in Section 3.3.1.3 e xcept for the g ain and input resistance. The output of the amplier is connected to the input of a PIC controller whose output is connected to the input of a ber -optic transmitter (not terminated in 50 W ). A diagram of an MSE optical measurement is sho wn in Figure 3–34 Due to limited resources, tw o dif ferent types of ber -optic links were used with the tw o dif ferent optical measurements. Ag ain, since the measurements were originally intended primarily to serv e as a trigger source, this w as not thought to be a problem. The sensor located on north-east corner of the netw ork w as coupled to a Nicolet Isobe 3000 ber -optic link while the sensor located on the south-west corner w as coupled to an Opticomm MMV -120C ber -optic link. The optical measurements were digitized continuously for 800 ms (200 ms pre-trigger) at 10 MHz on a Y ok og a w a DL716 digital

PAGE 125

111 storage oscilloscope. The optical w a v eforms are band-limited to 4 MHz (-3 dB) by the anti-aliasing lter associated with the DL716 digitizer Detailed descriptions of the ber -optic links and the DL716 digitizer are gi v en in Sections 3.4 and 3.5.1 respecti v ely A photograph of an optical measurement assembly is sho wn in Figure 3–35 3.3.4 Curr ent Measur ements In 2002, tw o current measurements were added to the MSE in order to record the current w a v eforms from upw ard unconnected leaders to the ICLR T launch to wer assembly (grounded by a v ertical run of shield braid) induced by nearby do wnw ard stepped leaders that e v entually initiate a rst return strok e some where nearby Upw ard positi v e leader currents can be inferred (based on studies of the initial stage in rock et-triggered lightning by ( W ang et al. 1999 )) to be on the order of 100 A although ( Mazur and Ruhnk e 2001 ) claim to ha v e measured currents around 5 kA. Therefore, it w as deemed necessary to measure currents both in the hundreds of amperes and the se v eral kA range. The current w as sensed at the base of the 2 m high aluminum rock et launcher located on top of the 11 m high w ooden launch to wer used to trigger lightning to a nearby test po wer line during the summer A horizontal conductor (connected to the launcher) w as suspended about 1 m abo v e the top of the launch tubes, yielding a total structure height of about 14 m. A metal Hof fman enclosure containing the current sensor and associated electronics w as mounted on the base of the launcher as sho wn in Figure 3–36 In this conguration, an y current o wing on the rock et launcher w ould also o w on the metal box. A T&M Research Products R-5600-8 1 : 233 m W current vie wing resistor (CVR) w as installed at the bottom of the box. A metal ring on the CVR which w as bolted to the box serv ed as the rst terminal of the resistor and the lug situated in the center of the CVR just outside (beneath) the box serv ed as the second.

PAGE 126

112 Figure 3–35. MSE optical measurement assembly

PAGE 127

113 RocketLauncher HoffmanBoxShieldBraid FiberopticCable Figure 3–36. Aluminum rock et launcher with Hof fman box mounted to the base. Therefore, assuming the launcher is perfectly conducting, there is a resistance of 1 : 233 m W between the launcher and the lug. If the lug is at a dif ferent potential from the launcher a current will o w through the CVR with the v oltage across the terminals of the CVR gi v en by Ohms La w Since this v oltage is directly proportional to the current, with slope determined by the resistance of the CVR, the current is sensed by measuring the v oltage. The lug w as connected to ground via a v ertical run of shield braid and w as connected to a ground rod located at the base of the to wer The measured grounding resistance is approximately 20 W The v oltage is tak en from a female BNC connector mounted on the end of the CVR inside of the box opposite the lug. The T&M CVR is specially designed to be non-inducti v e so that the v oltage measured is proportional to the current only and not the time-deri v ati v e of the current This is especially important in triggered lightning

PAGE 128

114 current measurements (for which this conguration is primarily used), where di/dt has been measured up to as high as 411 kA s 1 ( Leteinturier et al. 1991 ). The response of the T&M R-5600-8 CVR is limited by se v eral f actors. First, the CVR has an upper frequenc y limit of 12 MHz (-3 dB). Second, the 10-90% rise-time is limited to 30 ns. Third, the CVR has a maximum ener gy rating of 5600 J. If this rating is e xceeded, the CVR could be damaged or destro yed. It is possible that some of the impulsi v e components associated with upw ard leader currents will ha v e a higher bandwidth and f aster rise-time than the rating stated abo v e, therefore it is important that these limitations are kno wn. The ber -optic transmitters for the current measurements were located inside of the lar ge Hof fman enclosure mounted on the base of the launcher This box serv es both to shield the electronics from electromagnetic interference and to protect the electronics from the weather Since currents of se v eral hundred amperes (or possibly se v eral kA) are o wing on the outside of this box, it w as critical to eliminate an y ground loops inside of the box. Therefore, the electronics were placed in separate metal box es which are electrically isolated from the main box by pieces of w ood and Styrofoam. Furthermore, the electronics inside of the sub-box es were isolated from the sub-box by pieces of Styrofoam. T w o separate sub-box es were used, one for each current measurement. A diagram of the measurement conguration is sho wn in Figure 3–37 A single CVR senses the current for both measurements; ho we v er the full-scale range for each measurement is dif ferent. The full-scale range is changed by adjusting the attenuation setting of the PIC controller (discussed in Section 3.2.1 ) with a higher attenuation resulting in a higher full-scale range. The output of the CVR is connected in series with a pair of 50 W resistors placed in parallel. The end of each resistor which is not connected to the CVR is connected to the input of a PIC controller whose output is

PAGE 129

115 connected to the input of a ber -optic transmitter terminated in 50 W by an in-line terminator All cables are 50 W coaxial cables with male BNC connectors. The time-domain equi v alent circuit of this conguration is sho wn in Figure 3–38 The quantity i L ( t ) is the current o wing on the aluminum launcher and the Hof fman enclosure. The resistance of the CVR, R C V R is 1 : 233 m W for the T&M R-5600-8. The CVR is in parallel with tw o resisti v e branches, with each branch comprised of a 50 W resistor in series with a 50 W in-line attenuator in series with another 50 W resistor The coaxial cables ha v e no ef fect on the equi v alent circuit since the y are terminated in their characteristic impedance. The 50 W in-line attenuators are the equi v alent circuit of the PIC controller and their only ef fect is to attenuate the v oltage present at its input by a f actor of G PI C In this conguration, if either end of the 50 W attenuator is disconnected, the attenuator will look lik e 50 W as seen from the disconnected end. The input of the ber -optic transmitter is tak en at the junction of the output of the 50 W feed-through attenuator and the last 50 W resistor The v oltage across the CVR, v C V R ( t ) is determined by both the resistance of the CVR and the load. v C V R ( t ) = i L ( t ) ( R C V R jj 100 W jj 100 W ) (3-135) R C V R is the resistance of the CVR and the tw o 100 W v alues are the resistances of the tw o load branches (each comprised of tw o 50 W resistors in series). The 50 W attenuator does not change the resistance of the branch and can be ignored when calculating v C V R ( t ) Noting that since the CVR resistance is much smaller than the parallel combination of the load branches, the e xpression for v C V R ( t ) simplies to v C V R ( t ) R C V R i L ( t ) (3-136) The v oltage seen at the input of either ber -optic transmitter is determined by the attenuators and the v oltage di vision between the tw o 50 W resistors in a gi v en load

PAGE 130

116 T ransmitter Fiber -optic Isobe Nicolet PIC Controller Battery 12 V PIC Controller Battery 12 V PIC Controller 50W PIC Controller 50W V oltage Output of CVR Metal Box Mounted to Rock et Launcher Supplies Po wer to Fiber -Optic PIC Controller T ransmitter FiberOpticCable FiberOpticCable Ground Rod T ransmitter Fiber -optic Isobe Nicolet 50 W 50 W R-5600-8 CVR Shield Braid CVR Lug Isolated Metal Box Isolated Metal Box T&M G PIC G PIC Figure 3–37. Diagram of the MSE current measurements.

PAGE 131

117 i L (t) R CVR 50 W 50 W 50 W 50 W PIC Controller Equivalent Circuit 50 W Attenuator Valued G PIC Voltage Output of CVR Fiber-optic Transmitter Inputs Figure 3–38. T ime-domain equi v alent circuit of the MSE current measurements. branch. Therefore, the v oltage at the input of either ber -optic transmitter v F O T ( t ) is gi v en by v F O T ( t ) = 1 2 G PI C v C V R ( t ) = 1 2 G PI C R C V R i L ( t ) (3-137) F or R C V R = 1 : 233 m W v F O T ( t ) becomes v F O T ( t ) = ( 0 : 0006165 ) G PI C i L ( t ) The abo v e e xpression is only v alid o v er the range of frequencies in which the response of the CVR is constant. The v oltage present at the input of the digitizing oscilloscope, v sco pe ( t ) is the v oltage present at the input of the ber -optic link, v F O T ( t ) modied by the link. If the ef fect of the link is assumed to be a frequenc y independent g ain or attenuation, of v alue G l ink then the e xpression for the v oltage at the input of the oscilloscope is v sco pe ( t ) = G l ink v F O T ( t ) = ( 0 : 0006165 ) G PI C G l ink i L ( t ) (3-138)

PAGE 132

118 The abo v e e xpression is only v alid for the range of frequencies in which the resistance of the CVR and the g ain of the ber -optic link are constant. The quantity G PI C has not been e xplicitly dened because a wide v ariety of attenuation settings were used during the 2002 e xperiment. Nicolet Isobe 3000 ber -optic links were used to transmit the analog w a v eforms to the Launch Control trailer where the y were digitized. As discussed in Section 3.4.2 the g ain of an Nicolet Isobe link is selectable. T w o dif ferent setting were used during the 2002 season, hence the v alue of G l ink will not be e xplicitly dened here. The current measurements were digitized continuously for 800 ms (200 ms pre-trigger) at 10 MHz on a Y ok og a w a DL716 digital storage oscilloscope. The current w a v eforms were band-limited to 4 MHz (-3 dB) by the anti-aliasing lters associated with DL716 digitizer Detailed descriptions of the Isobe 3000 ber -optic links and the DL716 digitizer are gi v en in Sections 3.4.2 and 3.5.1 respecti v ely 3.3.5 Measur ement Band width Summary T able 3–11 presents a listing of the estimated bandwidths of all of the 2001 and 2002 MSE measurements (e xcluding the optical sensors) and the f actors that are belie v ed to limit the bandwidths. In all cases anti-aliasing lters were present at the inputs of each digitizer ho we v er as discussed in the preceding sections, other f actors may be responsible for bandwidths belo w the limits of the lter An e xample of this are the inte grating capacitors used in the 2001 electric eld measurements, as discussed in Section 3.3.1.3 3.4 Fiber -Optic Links Fiber -optic links were used to transmit the analog data from all sensors to the Launch Control trailer where the y were digitized. T w o dif ferent types of ber -optic links were used in the 2001 MSE while three dif ferent types were used in the 2002 e xperiment. This section presents a detailed description of all types of links.

PAGE 133

119 T able 3–11. Estimated bandwidths of the 2001 and 2002 MSE measurements. Lo w-frequenc y Lo w High-frequenc y High Measurement limit frequenc y limit frequenc y T ype Y ear (approximate) limiting f actor (approximate) limiting f actor E-eld 2001 0 : 1 Hz R C time 3 MHz Inte grating constant capacitor E-eld 2002 0 : 15 Hz R C time 4 MHz Anti-aliasing constant lter B-eld 2001 100 Hz Acti v e 2 MHz Acti v e inte grator inte grator B-eld 2002 10 Hz Acti v e 4 MHz Anti-aliasing inte grator lter dE/dt 2002 DC Not 20 MHz Anti-aliasing applicable lter dB/dt 2002 1 : 5 Hz Anti-aliasing 20 MHz Anti-aliasing lter lter T able 3–12. MSE ber -optic link summary Signal to Nominal Output noise -3 dB Input Output Input Range T ype ratio (approx) bandwidth Resistance Resistance Range (in 50 W ) Opticomm 59 dB DC 30 MHz 68 k W 50 W 1 V 1 V Nicolet 60 dB DC 15 MHz 1 M W 50 W Selectable 1 V Meret 40 dB DC 20 MHz 50 W 50 W 0 : 5 V 0 : 5 V In 2001, thirteen Opticomm MMV -120C ber -optic links were used along with one Nicolet Isobe 3000 ber -optic link. In 2002, four Meret MDL288DC ber -optic links were added along with tw o additional Isobe 3000 links. T able 3–12 gi v es a brief summary of the characteristics of each type of link. 3.4.1 Opticomm MMV -120C Fiber -Optic Links The Opticomm MMV -120C ber -optic links utilize frequenc y modulation (FM) with a carrier frequenc y of 70 MHz and operate at an optical w a v elength of 1310 nm The Opticomm links were intended by the manuf acturer to be used as video ber -optic links and therefore had an input and output resistance of 75 W (the standard resistance for video equipment). This w as not appropriate for the MSE, which is almost

PAGE 134

120 e xclusi v ely a 50 W system. Therefore, the input resistance w as modied to 68 k W by the manuf acturer A high input resistance has the adv antage of being able to be lo wered an y desired v alue by simply adding an in-line terminator The v alue of 68 k W w as not chosen arbitrarily; it w as the highest v alue the manuf acturer could achie v e without sacricing the performance of the link. In addition, the output resistance w as modied to 50 W from 75 W Finally the lo w frequenc y cutof f w as modied to DC from 5 Hz (-3 dB) by the manuf acturer The manuf acturer lists the signal-to-noise ratio to be about 67 dB, ho we v er this v alue is acquired using the short-haul RS-250C standard in which the signal is lo w-pass ltered with a cut-of f frequenc y of about 5 MHz Thus, 67 dB may not be an accurate representation of the true signal-to-noise ratio o v er the entire bandwidth. In practice the actual signal-to-noise ratio o v er the entire bandwidth is se v eral dB lo wer than the v alue obtained under the short-haul RS-250C standard. 3.4.2 Nicolet Isobe 3000 Fiber -Optic Links The Nicolet Isobe 3000 ber -optic links ha v e an input resistance of 1 M W and utilize a combination of amplitude modulation (AM) and pulse-width modulation (PWM). The input range of the transmitter is selectable from 0 : 1 V, 1 V, and 10 V. In addition, the g ain and of fset of the link can be manually adjusted at the recei v er end. The output range of the recei v er is x ed at 1 V re g ardless of the selected input range. F or e xample, if a 5 V signal is connected to the transmitter set to the 10 V input range, the corresponding v oltage at the output of the recei v er will be 0 : 5 V. Therefore, the Isobe ber -optic link ef fecti v ely attenuates the signal by -20 dB (0 : 1 V = V) when the transmitter is set to the 10 V range. Similarly when the transmitter is set to the 0 : 1 V range, the link introduces a g ain of 20 dB (10 V = V ). 3.4.3 Mer et MDL288DC Fiber -Optic Links The Meret MDL288DC ber -optic links utilize amplitude modulation (AM) and operate at an optical w a v elength of 820 nm. Unlik e the Opticomm and Isobe links,

PAGE 135

121 the Meret transmitters are not stand-alone units; the y must be mounted on a circuit board. The adv antage of this design, ho we v er is that an y electronics associated with a measurement such as a preamplier or an acti v e inte grator can be incorporated in the same circuit as the ber -optic transmitter In the early 1990s, Geor ge Schnetzer designed and b uilt se v eral dif ferent types of units incorporating the Meret transmitters. The three most common units included an acti v e inte grator a high-impedance preamplier or a dif ferential preamplier Although the Meret transmitter itself has a nominal frequenc y response from DC to 35 MHz, the other circuit components lo wer the upper limit. F or e xample, the units which incorporated an acti v e inte grator are limited to 10 MHz, while the units with a dif ferential preamplier are limited to 20 MHz. The 20 MHz dif ferential input units are the only type used in the MSE, and the other types will not be discussed further In practice, the Meret links suf fer from high noise and from DC of fset drift. This is corrected by utilizing A C coupled acti v e lo w-pass lters at the output of the recei v ers. The lo w frequenc y response is limited to about 1 : 5 Hz (-3 dB) by the A C coupling. In addition, the lters can pro vide g ain which is adjusted by a potentiometer These lters were designed to be paired with specic types of Meret units. F or e xample, a set 20 MHz (-3 dB) lo w-pass lters were designed to be used with the dif ferential input units. 3.4.4 Fiber -Optic Cables The Opticomm links utilize 62 : 5 = 125 m (core/cladding) graded inde x multimode ber -optic cable. In 2001, Optical Cable Corporation (OCC) AX series multimode K e vlar reinforced cables with ST connectors were used with the Opticomm links. During the 2001 season it w as disco v ered that the cables were being che wed and damaged by animals. Ho we v er the cable could be cut, re-terminated, and spliced with an ST b utt connector The optical loss of the splice, measured with an Agilent E6000C Optical T ime-Domain Reectometer (O TDR), w as typically around -1 dB.

PAGE 136

122 In 2002, OCC BX series w ater resistant armored cable replaced the AX series cable. Each armored cable consists of tw o armored outer jack ets surrounding either four (BX-04 series) or six (BX-06 series) sub-cables. Each sub-cable is K e vlar reinforced, similar to the AX series cable. The sub-cables, which are color coded, are twisted around a strength member running through the center of the cable. ST connectors were used as before. The armored cables are v ery dif cult to damage and problems due to animals were eliminated. In 2001, the Isobe links were used with 200 m multimode K e vlar reinforced duple x ber -optic cables with SMA connectors manuf actured by OFS Fitel Corporation. In 2002, six-ber armored cables, manuf actured by OFS Fitel were used. Ag ain, the indi vidual sub-cables are color coded and twisted around a strength member Each Isobe link requires tw o bers; hence each cable can supply three links. SMA connectors were used as before. These cables ha v e only a single layer of armor and were vulnerable to w ater intrusion through an y opening in the armor Therefore, care must be tak en to protect the ends of the cable from the elements. The Meret links ha v e been used in pre vious ICLR T e xperiments with 50 = 125 m although the y will operate using the 62 : 5 = 125 m OCC ber The Meret ber -optic transmitters use SMA0906 connectors, which w ould be dif cult to terminate in the eld. Con v ersely the ST and SMA connectors used with the Opticomm and Isobe links, respecti v ely were relati v ely easy to terminate in the eld. Therefore, the bers were terminated with ST connectors, as with the Opticomm links, and 3 m patch cords were used to con v ert the ST connectors to SMA0906 connectors. The insertion loss for an indi vidual patch cord ranged from about -0.5 to -1.5 dB, with tw o patch cords needed for each link. Furthermore, in pre vious e xperiments, the Meret links were used with a ber length of about 100 m ho we v er the y will operate at length of o v er 500 m using the 62 : 5 = 125 m ber An optical loss of o v er se v eral dB can be introduced by using such long ber lengths, b ut the links are still usable. The

PAGE 137

123 maximum signal-to-noise ratio w as measured e xperimentally to be about 40 dB by connecting a transmitter and recei v er together with tw o patch cords and a 1 m length of 62 : 5 = 125 m ber Since the signal le v el is decreases with increasing ber length, the signal-to-noise ratio for an y link in the eld will be less than the maximum v alue of 40 dB.3.4.5 Fiber -Optic Link Calibration The use of ber -optic links is a necessity in lightning e xperiments due to their electromagnetic immunity ho we v er the y introduce issues of their o wn. Introducing a ber -optic link into a measurement system introduces one more component which must be tak en into account when designing the system. The bandwidth, input resistance, output resistance, and delay of the link all ha v e an ef fect on the system. The input and output resistance are specied by the manuf acturer and are easily tak en into account. Furthermore, the bandwidth of the link is also specied, although it is a good idea to v erify this e xperimentally Finally the delay of the link (a function of both the link electronics and the ber -optic cable itself) must, in general, be determined e xperimentally In the frequenc y domain, a link is characterized by its frequenc y response, G l ink ( w ) In practice, G l ink ( w ) is not directly measured, ho we v er the magnitude, j G l ink ( w ) j and phase, \ G l ink ( w ) components are. The magnitude and phase components are related by G l ink ( w ) = j G l ink ( w ) j e j \ G l ink ( w ) (3-139) Ideally j G l ink ( w ) j = 1 and \ G l ink ( w ) = 0, which implies that the link has no ef fect on the measurement system. In other w ords, the v oltage present at the output of the ber -optic recei v er is e xactly the same as that present at the input of the ber -optic transmitter with no time delay While the rst condition is impossible to satisfy since no system has an innite bandwidth, the second is impossible since information cannot

PAGE 138

124 tra v el f aster than the speed of light. Ho we v er since the purpose of these links is to transmit data with minimal distortion and time delay their response should de viate little from the ideal o v er a specied frequenc y range. F or e xample, manuf acturers typically specify the -3 dB bandwidth of the link, such as DC to 30 MHz for the Opticomm MMV -120C links. Ov er a certain frequenc y range, which is smaller than the -3 dB bandwidth, j G l ink ( w ) j will be constant. This constant v alue will be denoted as G l ink Furthermore, G l ink is usually specied to be one o v er the stated frequenc y range, b ut it can v ary with temperature, ber length, and termination quality depending on what type of modulation scheme is used. F or e xample, in AM system such as Meret, G l ink is a strong function of ber and termination quality while FM systems such as Opticomm are more rob ust. In either case, it is necessary to estimate G l ink after each time data is acquired in order to assume the most accurate calibration possible. In addition, the phase response of a link, \ G l ink ( w ) should be a linear function of w Linear phase in the frequenc y domain corresponds to a constant time delay in the time domain. This is important since, if the phase response w as an arbitrary function of w then dif ferent frequenc y signals w ould arri v e at the recei v er at dif ferent times and introduce distortion in a signal composed of se v eral dif ferent frequencies. Assuming linear phase, the time delay of a gi v en link is a function of both the electronics and the type and length of ber used. The v alue of G l ink is estimated after each system trigger by introducing a calibration signal into the transmitter and measuring the signal at the output of the recei v er If the signal being fed into the transmitter is kno wn e xactly then G l ink is gi v en by G l ink = j v measur ed j j v sour ce j (3-140) The quantities j v measur ed j and j v sour ce j are the peak-to-peak magnitudes of the measured and source calibration v oltage w a v eforms, respecti v ely In order for this to be an accurate measure of the link, the entire frequenc y content of v sour ce must lie within

PAGE 139

125 the frequenc y range where j G l ink ( w ) j is constant. Furthermore, the phase response must be linear to ensure there is no phase distortion. As discussed in Section 3.2.1 the calibration w a v eform is a 100 Hz square w a v e with a peak-to-peak-v oltage of either 1 V or 0 : 1 V in 50 W The w a v eform generator circuit w as designed by Geor ge Schnetzer and is v ery immune to v ariations in time and temperature. Therefore, the v alue of v sour ce measured in the lab is an accurate measure of what v sour ce will be when the signal generator is used in the eld. In the frequenc y domain, the square w a v e consists of the fundamental frequenc y 100 Hz, along with an innite number of discrete harmonic frequencies, all of which are odd multiples of the fundamental. The magnitude of each harmonic is in v ersely proportional to the order of the harmonic, meaning, for e xample, that the magnitude of the third harmonic is one third that of the fundamental. Although the bandwidth of an ideal square w a v e is innite, it can be considered nite relati v e to the bandwidths of the ber -optic links used in the MSE. F or e xample, the magnitude response of the Opticomm MMV -120C link is at out to at least 20 MHz. A 100 Hz square w a v e can be assumed to ha v e zero frequenc y content at 20 MHz. Therefore, if j G l ink ( w ) j is constant up to 20 MHz and the link has a linear phase response, then the 100 Hz square w a v e calibration w a v eform can be used to estimate the g ain of the link. The same ar gument can be applied to the Nicolet Isobe 3000 and Meret ber -optic links, which ha v e -3 dB bandwidths of about 15 and 20 MHz, respecti v ely Ho we v er as discussed in Section 3.4.3 A C coupled lo w-pass lters are used at the output of the ber -optic recei v ers, which limit the lo w-frequenc y response of the Meret links to about 1 : 5 Hz. This does introduce some distortion into the recei v ed calibration signal. It is important to kno w the time delays associated with each ber -optic link so that all w a v eforms can be properly aligned when performing data analysis. Furthermore, it is desirable to kno w the time delays to within one sample (digitization) point. If each of the ber -optic links is of the same type and uses the same length of ber the

PAGE 140

126 relati v e delays between each of the w a v eforms will be identical. Ho we v er the MSE utilized three dif ferent types of links with man y dif ferent lengths of ber so accurate time delay measurements are critical. The total delay of a ber -optic link is determined by tw o f actors; electronic delays and ber delays. Electronic delays constitute an y delays that arise from the transmitter and recei v er electronics themselv es. Fiber delays are due to the propag ation of light in the ber -optic cable at a nite speed, which is slo wer than the speed of light. In general, the time delay of a link can be modeled as a constant delay (due to the electronics) plus a delay that is a linear function of ber length. Each indi vidual ber (2001) or armored cable (2002) w as cut and terminated in the eld, so the length of each ber or cable could not be measured directly Therefore, the optical length of each ber w as measured with an Agilent E6000C Optical T ime-Domain Reectometer (O TDR). The O TDR measures the length of the cable by sending a pulse of light do wn the ber and measuring the time it tak es for the pulse to be reected back to the light source. The distance resolution of the O TDR is in v ersely proportional to the pulse width of the light source (since the O TDR is actually measuring time); hence a narro w pulse will yield a more precise measure of ber length. In addition, when used with multimode ber the distance resolution of the O TDR will be limited by modal dispersion in the ber whereby the width of a pulse of light is widened due to multi-modal propag ation in the ber The optical length in meters of a ber l is gi v en by l = 1 2 t measur ed c N (3-141) The quantity c is the speed of light in a v acuum, equal to 3 10 8 m s 1 and t measur ed is the measured round-time of the light pulse emitted by the O TDR. The quantity N is supplied by the manuf acturer of the ber and is typically referred to as the group inde x, and is dened as the ratio of the speed of light in a v acuum to the

PAGE 141

127 speed of light in the ber In practice this v alue is often slightly dif ferent than the inde x of refraction of the glass which is used to mak e the ber The one-half scaling f actor is included because t measur ed is the measured round-trip time, not the one-w ay time. The group inde x is supplied by the manuf acturer and must be programmed into the O TDR. If the wrong v alue of N is used, the optical length measurement will be incorrect. The optical length measurement can be corrected if both the correct v alue of N and the incorrect v alue of N used during the measurement are kno wn. The ber delay t f iber can be found by rearranging equation 3-141 t f iber = 1 2 t measur ed = N l c (3-142) Hence, the time delay of the ber optic cable is proportional to both the l and N The OCC and OFS Fitel ber -optic cables (both armored and unarmored) ha v e v alues of N of 1.483 and 1.429, respecti v ely The corresponding ber delays are 4 : 943 ns m 1 and 4 : 763 ns m 1 respecti v ely The optical lengths, and hence time delays, of all bers were measured using the O TDR for both the 2001 and 2002 e xperiments. F or the 2002 e xperiment, the O TDR w as set to generate a 5 ns wide pulse at 1310 nm for the OCC armored ber and 850 nm for the OFS ber with the measurement being a v eraged o v er one minute. T able 3–13 gi v es the measured optical length and corresponding time delay for each ber The time delays were calculated by using Equation 3-142 Only the bers currently in service were tested; an y unused bers in an armored cable b undle were not. Each armored cable is uniquely identied by the location it w as run to, such as Station 1, for e xample. Each indi vidual ber is uniquely identied by the combination of the cable identier and the color of the ber in that cable, such as Station 1 Blue for e xample. The OFS bers are grouped into pairs (such as T ower Y ellow/Red for e xample) since the Nicolet Isobe links each require tw o bers. At the time of this writing, the delays

PAGE 142

128 T able 3–13. O TDR measured optical lengths and corresponding time delays for armored ber -optic cables used during the 2002 MSE. O TDR measured Corresponding Fiber designation Measurement length ( m ) time delay ( ns ) Station 1 Blue dE-1 499.7 2470 Station 1 Orange dB-1N 500.0 2472 Station 1 Green dB-1E 500.0 2472 Station 2 Bro wn E-2 129.2 639 Station 4 Blue dE-4 288.7 1427 Station 4 Green E-4 289.0 1429 Station 4 Gray B-4N 289.0 1429 Station 4 Orange dB-4N 289.3 1430 Station 5 Bro wn E-5 105.8 523 Station 6 Bro wn E-6 521.5 2578 Station 8 Blue dE-8 357.2 1766 Station 9 Blue dE-9 617.2 3051 Station 9 Bro wn E-9 617.5 3052 Station 9 Gray B-9N 616.2 3046 Station 9 Orange dB-9N 616.2 3046 Station 10 Bro wn E-10 527.8 2609 SW Optical Bro wn SW O 652.9 3227 NE Optical NEO 421.2 2006 Orange / Blue T o wer A I-High-T o wer 104.6 498 Orange / Blue T o wer B I-Lo w-T o wer 93.8 447 Y ello w / Red due to the electronics ha v e not been measured, and hence only the ber delays are sho wn. In 2001, the ber delays were measured by Geor ge Schnetzer Unlik e the 2002 season, the group inde x parameter of the O TDR w as set to some arbitrary and hence incorrect, v alue. Instead, an e xactly 500 m long piece of ber w as measured with the O TDR. The actual ber delay w as found by placing the same ber in a link and feeding a signal into the transmitter Both the input to the transmitter and the output of the recei v er were vie wed on an oscilloscope where the time delay w as measured. In order to subtract the delay associated with the transmitter and recei v er electronics, the

PAGE 143

129 same time-delay measurement w as performed using a v ery short length of ber The ber delay associated with this short length of ber w as ne gligible compared to that of the 500 m piece and hence only the delay of the electronics w as measured. It should be noted that calculated electronic delays were not a v ailable to the author at the time of this writing. Using this information, the time delay of an y length of ber could be calculated with the follo wing formula, assuming the same group inde x parameter w as used in the O TDR measurement. t f iber = l O T DR l O T DR 500 m t 500 m (3-143) The quantity l O T DR is O TDR measured length of the ber under test, l O T DR 500 m is the O TDR measured length of the 500 m piece of ber and t 500 m is the measured time delay of the 500 m piece of ber (with the measured electronic delay subtracted out). Equation 3-143 is only v alid if l O T DR and l O T DR 500 m are measured using the same arbitrary group inde x parameter of the O TDR. Each ber used during the 2001 season w as gi v en a numerical identier and the calculated ber delays are gi v en in T able 3–14 It should be noted that no measurement w as performed on the ber going to the north-east optical detector 3.5 Digitizers A single Y ok og a w a DL716 digital storage oscilloscope (DSO) w as used during the 2001 MSE. This w as augmented with a LeCro y L T344 W a v erunner and a LeCro y L T374 W a v erunner2 DSO in 2002. The full range of features of each DSO will not be discussed; only the features rele v ant to the MSE will be e xamined. A complete description of each digitizer can be found in the digitizers manual or from the manuf acturers website. 3.5.1 Y ok ogawa DL716 The Y ok og a w a DL716 is a 16 channel DSO with a maximum bandwidth of 4 MHz (-3 dB) and a maximum sampling rate of 10 MHz with 12-bit v ertical

PAGE 144

130 T able 3–14. Calculated time delays for the ber -optic cables used in the 2001 MSE. Fiber designation Measurement Calculated time delay ( ns ) 101 SW O 3221 102 E-1 2305 103 E-5 835 104 E-2 697 105 B-4E 1288 106 E-4 1264 107 B-4N 1269 108 E-10 2314 109 E-8 1455 110 E-9 2252 111 B-9N 2314 112 B-9E 2281 113 E-6 2562 resolution. The DL716 is capable of a maximum record length of 16 me g asamples per channel when all 16 channels are used simultaneously At the maximum sampling rate of 10 MHz the maximum record length is 1 : 6 s when all 16 channels are used. Therefore, the DL716 is ideal for recording a continuous full-ash record of lightning electric and magnetic elds. Since each sample is recorded with 12-bit resolution, tw o bytes (16 bits or one w ord) are required to store each sample (four bits are thro wn a w ay). Therefore, a 1 : 6 s record sampled on all 16 channels at 10 MHz requires 256 me g a w ords or 512 me g abytes. Each DL716 channel can be set from 5 mV per di vision to 20 V per di vision with a maximum peak-to-peak input v oltage range of 250 V. The input resistance of each channel is 1 M W shunted with 30 pF of input capacitance, either A C or DC coupled. In addition, each channel can be indi vidually congured with an internal lo w-pass lter of v alues 500 Hz, 5 kHz, 50 kHz, and 500 kHz (-3 dB). The DL716 is equipped with 9.2 GB of internal storage; hence man y acquired w a v eforms can be stored in the digitizer itself. The DL716 is equipped with an e xternal SCSI hard disk connector so more storage can be added. W a v eforms can be

PAGE 145

131 mo v ed to and from the DL716 o v er a 10Base-T Ethernet connection using the File T ransfer Protocol (FTP). The DL716 can be congured o v er an IEEE 488.2 (GPIB) b us and supports a rob ust command set. Almost e v ery setting can be manipulated o v er GPIB, therefore the DL716 can be congured and armed remotely The DL716 can be automatically congured for an e xperiment by a PC with a GPIB interf ace card and appropriate softw are such as National Instruments LabV ie w by either issuing a series of commands or loading a set of predened settings from the hard disk of the DSO. The DL716 can be triggered by an y of the 16 channels or by an e xternal TTL le v el trigger input. Furthermore, comple x triggering schemes can be generated by OR triggering an y combination of the 16 channels with the trigger le v el and slope of each channel capable of being set indi vidually One major disadv antage of the DL716 is that, for long records, it tak es se v eral minutes to write the data from memory to disk. During this interv al, the digitizer cannot trigger and hence no data can be recorded. 3.5.2 LeCr oy L T344 W a v erunner The LeCro y L T344 is a four channel DSO with a maximum bandwidth of 500 MHz at a maximum sampling rate of 500 MHz with 8-bit v ertical resolution. The L T344 is capable of a maximum record length of one me g abyte per channel when all four channels are in use. Since a sample is recorded with 8-bit resolution, one byte is required to store each sample; hence the total record length is one me g asample per channel. At the maximum sampling rate of 500 MHz the maximum record length is 2 ms. T ypically the L T344 is not used to acquire a single continuous record b ut rather is used in se gmented memory mode. In se gmented memory mode, the acquisition memory is di vided into multiple se gments and a separate trigger is required to record each se gment. F or e xample, if v e se gments are used, the acquisition memory is di vided into 200 kilosamples per se gment per channel. Se gmented memory mode is

PAGE 146

132 useful when acquisition memory is limited and multiple e v ents are to be recorded with the time length of each e v ent being v ery small relati v e to the length of time between e v ents. Se gmented memory mode is ideal for recording the electric and magnetic eld w a v eforms (or their time-deri v ati v es) from indi vidual return strok es. T ypically the return strok e w a v eforms occur on a tens of microseconds time scale while the time between return strok es is on the order of tens to hundreds of milliseconds ( Uman 1987 ). When pre-trigger is used, the pre-trigger setting applies to each se gment of the acquisition. F or e xample, if 10% pre-trigger is used, the rst 10% of each se gment record consists of data before that se gments trigger point. The input resistance of each L T344 channel can be set to 50 W or 1 M W either A C or DC coupled. Each channel can be set from 2 mV per di vision to 10 V per di vision with a maximum RMS input v oltage of 5 V and 280 V when 50 W and 1 M W input resistance are used, respecti v ely In addition, each channel can be indi vidually congured with an internal lo w pass lter of v alues 25 MHz and 200 MHz (-3 dB). The L T344 is equipped with a PCMCIA T ype III slot, which is used to add storage such as a hard dri v e or a compact ash card which come in a v ariety of sizes. A 128 MB compact ash card w as used in the L T344; hence a maximum of 32 w a v eforms could be stored if one me g abyte w as used per channel and all four channel were in use for each acquisition. W a v eforms can be mo v ed to and from the DSO o v er a 10Base-T Ethernet connection or the GPIB b us using LeCro y Scope Explor er softw are. Unlik e the DL716, the L T344 does not support FTP and uses a proprietary protocol for le transfers. The L T344 can be congured o v er GPIB or Ethernet and supports a rob ust command set. Similar to the DL716, the L T344 can be congured and armed remotely The L T344 can be triggered from an y of the four channels or from an e xternal trigger input, and comple x triggering schemes are a v ailable as with the DL716. Unlik e the DL716, the e xternal trigger input is not TTL, and the trigger le v el and slope can

PAGE 147

133 be adjusted. In addition, the input resistance and coupling of the trigger input can be adjusted.3.5.3 LeCr oy L T374 W a v erunner2 The LeCro y L T374 W a v erunner2 is the successor of the L T344 W a v erunner DSO. The L T374 is essentially identical to the L T344, e xcept that the maximum sampling rate is 2 GHz, the maximum acquisition memory is four me g abytes per channel and the input lters are selectable from 20 MHz and 200 MHz (-3 dB). The L T374 also includes se v eral ne w features and enhancements o v er the L T344 which are not used in the MSE and will not be discussed. 3.6 V ideo System The location of an y lightning within the netw ork is initially unkno wn. Therefore a video system is required which can be used to quickly determine the general location and geometry of the lightning channel. V ideo records are typically v ery useful for augmenting more rigorous location techniques. In 2001, a Cohu 1300 Series CCD camera w as installed in each of the four instrumentation stations (designated IS1 IS2 IS3 and IS4 ) located about the ICLR T site, as sho wn in Figure 3–1 in Section 3.1 The video signals were transmitted o v er 62 : 5 = 125 m ber by Opticomm MMV -110 ber -optic video links. All four bers were run to the Launch Control trailer where the outputs of the four recei v ers were fed into a quad-display security monitor The monitor con v erted the four indi vidual video frames into a single video frame di vided into four quadrants. Each of the four quadrants contained one of the four camera vie ws. This system pro vided a con v enient vie w of the entire ICLR T site at a glance, although a signicant amount of resolution w as lost. In addition, since four independent video signals were being combined into a single video signal, the four quadrants were not time-synchronized. Therefore, a single e v ent captured by all four cameras might not sho w up in all four quadrants of a single frame. F or e xample, quadrants 1 and 2 might display an e v ent in frame 1, while

PAGE 148

134 Figure 3–39. Example video frame from the MSE video system. Going clockwise from the upper left, the four quadrants represent the camera vie ws from IS1, IS2, IS4, and IS3, respecti v ely with the lightning being in vie w of the IS4 camera. quadrants 3 and 4 display the e v ent in frame 2. An e xample video frame is sho wn in 3–39 The single combined video signal w as a v ailable from an output of the monitor and this w as connected to the video input of a Son y SR2000 TIV O digital recording system. The amount of time a v ailable for recording is proportional to amount of hard disk space inside of the unit. The TIV O unit w as capable of recording about 18 hours of high-quality video with the f actory installed 80 GB hard disk. A second 80 GB hard disk w as added to increase the total recording time to about 36 hours. The TIV O can be programmed to record, although it can only be set to record at re gular interv als. F or e xample, the TIV O can be set to record a one-hour record at 12 pm e v ery day ho we v er there is no w ay to tell it to start recording when the netw ork is armed. Ob viously if there are personnel on site during a thunderstorm, the TIV O can manually be set to record.

PAGE 149

135 Therefore, due to the recording limitations, the TIV O w as set to constantly record, and it w as the operators responsibility to collect an y lightning video data before it w as o v erwritten. T o increase the total record length, a second TIV O (upgraded with a second 80 GB hard disk) w as added. Although the TIV O is a digital recorder there w as until recently no w ay to e xtract the digital video data directly from the hard disk. Therefore, a separate video digitizer which w ould digitize an analog video output from the TIV O, w as needed to e xtract the video from the 2001 season. The addition of this e xtra analog stage further decreased the resolution of the video. The T ytools softw are package w as used to e xtract the video from the recorder after the 2002 season. 3.7 T iming System A Datum bc627A T GPS timing card w as used to pro vide trigger times with microsecond accurac y The bc627A T card k ept accurate time by tracking and recei ving signals from se v eral GPS satellites in orbit. The output of the trigger circuit w as fed into an input of the card and the time w as latched into a set of internal re gisters when a system trigger occurred. The card w as specically designed to ha v e less than 100 ns delay between the rising edge of the trigger signal and latching of the time. Softw are written by the author interf aced with the card and logged all of the trigger times. This system w as used with both natural and rock et-triggered lightning in 2002, b ut only with rock et-triggered lightning in 2001.

PAGE 150

CHAPTER 4 PRESENT A TION OF D A T A In this chapter the e v ents recorded during the 2001 and 2002 e xperiments, for both natural and triggered lightning, are summarized. In addition, w a v eforms are presented for selected data acquired during the 2002 e xperiment. 4.1 Data Summary In 2001, data from tw o natural and six rock et-triggered ashes were recorded. In 2002, data were recorded from ele v en natural and se v en rock et-triggered ashes. T ables 4–1 and 4–2 present a summary of the recorded data for the 2001 and 2002 seasons, respecti v ely Each ash is gi v en a unique designation based on the year and the e xperiment ID. F or e xample, the ash designated MSE-0102 w ould be the second natural ash recorded by the MSE netw ork during the 2001 e xperiment. Similarly the ash designated MSE-0207 w ould be the se v enth natural ash recorded by the MSE netw ork during the 2002 e xperiment. The naming scheme for triggered ashes is slightly dif ferent. Although the data were recorded by the MSE netw ork, the ash name is based on the e xperiment for which the ash w as initiated (no triggered ashes were initiated for the sole purpose of being recorded by the MSE netw ork) and the order of the rock et red during the year F or e xample, the ash designated S-0105 refers to the ash initiated by the fth rock et red during the 2001 SA TTLIF e xperiment. Similarly FPL-0208 refers to the ash initiated by the eighth rock et red during the 2002 Florida Po wer and Light e xperiment. It should be noted that under this naming scheme for triggered ashes, the ash numbering is not necessarily continuous. F or e xample, if ten rock ets were red during the 2001 SA TTLIF e xperiment and shots 3, 4, and 9 resulted in lightning, then the consecuti v e ash designations w ould be S-0103 S-0104 and S-0109 In contrast, the numbering for natural ashes is 136

PAGE 151

137 T able 4–1. Flashes recorded by the MSE netw ork in 2001. Flash ID T ype Date T ime (UT) MSE-0101 Natural Ne g ati v e CG 7/23/2001 (19:21:48.305) MSE-0102 Natural Ne g ati v e CG 9/5/2001 (21:51:11.864) S-0105 Classical T rigger 7/20/2001 00:05:56.355 S-0106 Classical T rigger 7/20/2001 (00:12:16.142) S-0107 Classical T rigger 7/20/2001 00:17:32.523 S-0116 Classical T rigger 7/27/2001 (21:56:05.300) S-0118 Classical T rigger 7/31/2001 16:53:20.897 FPL-0107 Classical T rigger 7/27/2001 21:50:58.132 continuous since no rock ets are red in order to initiate the ashes. Flashes triggered for the Florida Po wer and Light (FPL) e xperiment were all initiated from the 11 m high launch to wer with the e xception of one ash initiated from a mobile launcher In addition, one unintentional altitude trigger (rock et red from the launch to wer) attached to Instrument Station 1. Flashes triggered for the SA TTLIF e xperiment were initiated from the under ground launcher Both f acilities are discussed in Section 2.2 The times gi v en in T ables 4–1 and 4–2 are those recorded by the GPS timing system discussed in Section 3.7 with the e xception of those in parentheses, that are the times reported by the U.S. National Lightning Detection Netw ork (NLDN). The NLDN is operated by V aisala Corporation. T able 4–3 lists se v eral NLDN-reported parameters, including calculated location and peak current, for the rst return strok e of each natural ash recorded by the MSE netw ork. In addition, the calculated distance to the UF launch to wer is gi v en. Only the information for the rst return strok e reported by the NLDN is gi v en, although multiple strok es were often recorded by both the MSE netw ork and the NLDN. The unique inter -strok e interv al timings were usually used to identify the ash in the NLDN records when GPS timing w as not a v ailable for the MSE data. Along with each NLDN-reported location, there is an associated “condence ellipse” or “error ellipse. ” This ellipse is dened as the elliptical area, centered at the NLDN-calculated most probable location of the strok e, for which there is a gi v en

PAGE 152

138 T able 4–2. Flashes recorded by the MSE netw ork in 2002. Flash ID T ype Date T ime (UT) MSE-0201 Natural Ne g ati v e CG 7/20/2002 (19:39:30.712) MSE-0202 Natural Positi v e CG 8/4/2002 (20:05:28.990) MSE-0203 Natural Ne g ati v e CG 8/30/2002 18:00:44.072 MSE-0204 Natural Ne g ati v e CG 8/30/2002 18:01:37.993 MSE-0205 Natural Ne g ati v e CG 8/30/2002 18:13:55.827 MSE-0206 Natural Ne g ati v e CG 9/15/2002 20:40:59.256 MSE-0207 Natural Ne g ati v e CG 9/15/2002 20:41:30.727 MSE-0208 Natural Ne g ati v e CG 9/15/2002 21:18:14.425 MSE-0209 Natural Ne g ati v e CG 11/10/2002 (14:29:05.611) MSE-0210 Natural Ne g ati v e CG 11/10/2002 (14:31:42.561) MSE-0211 Natural Ne g ati v e CG 12/24/2002 18:49:58.600 FPL-0205 Altitude T rigger 7/9/2002 16:26:10.767 FPL-0208 Classical T rigger 7/9/2002 16:35:05.073 FPL-0213 Classical T rigger 7/19/2002 21:58:05.813 FPL-0219 Classical T rigger 7/20/2002 20:26:32.545 FPL-0221 Classical T rigger 7/20/2002 20:51:42.721 FPL-0228 Classical T rigger 8/2/2002 00:20:15.862 FPL-0246 Classical T rigger 9/13/2002 19:18:14.084 probability that the actual strik e point w as within the ellipse. Each ellipse can be uniquely dened by the length of its semi-major axis, its eccentricity (the ratio of semi-major axis length to semi-minor axis length), and the angle of the semi-major axis with respect to true north. F or e xample, the 50% probability ellipse for the NLDN calculated location for the rst strok e of ash MSE-0204 has a semi-major axis length of 400 m and an eccentricity of one. Hence, the 50% condence ellipse for the rst strok e of ash MSE-0204 is a circle of radius 400 m. A circular condence ellipse ha ving a relati v ely lo w ( < 1 km ) semi-major axis length implies that the strok e w as detected by multiple NLDN sensors at v arious angles. A v ery eccentric ellipse, ha ving a lar ge semi-major axis, implies that the strok e w as only detected by tw o sensors with the major axis of the ellipse being oriented on a line between the tw o sensors. Ellipses for an y arbitrary condence le v el can be calculated by scaling the semi-major axis and semi-minor axis appropriately The 95% ellipses (95% probability of the strik e point being inside of the ellipse) are about twice the size of the 50% ellipses. V ideo records

PAGE 153

139 and other data, appropriately analyzed, can most lik ely be used (and will be in the future) to yield a more accurate location, b ut that is be yond the scope of this thesis. Furthermore, a c 2 v alue is calculated for each NLDN-calculated location. In general, c 2 is a measure of the quality of a solution, with a v alue of 1 corresponding to a good solution and a v alue much greater than 1 corresponding to a poor solution. In this case the solution is the set of calculated strik e point coordinates. Both the error ellipse and c 2 information are presented in T able 4–3 Muc h of the abo ve was tak en fr om the LP2000 user' s manual, copyright Global Atmospherics Inc. 4.2 Example W a v ef orms In this section, representati v e w a v eforms from four ashes (three natural and one triggered), recorded by the 2002 MSE, are presented. The graphics in all of the follo wing gures were generated by Dr Carlos Mata, a consultant on the project. In order to remo v e an y DC of fsets introduced by the ber -optic links, which can be signicant, the rst 100 samples of each w a v eform were a v eraged and the o v erall w a v eform w as v ertically shifted by this amount. 4.2.1 Natural Flash MSE-0202 Natural ash MSE-0202 w as a tw o-strok e positi v e cloud-to-ground ash that occurred at approximately 20:05:28.990 UT on August 4, 2002. Although the NLDN reports the strik e point to be about 4 km from the UF launch to wer video records indicate that at least one channel terminated within or v ery close to the netw ork. Both video and eld records indicate tw o separate termination points. Specically the magnetic eld (measured at Stations 4 and 9) w a v eforms from the second strok e ha v e opposite polarity relati v e to the magnetic elds from the rst strok e. This indicates that the tw o strok es were on opposite sides of the magnetic eld antenna at each location. In addition, the relati v e amplitudes of the leader electric eld changes corresponding to the rst and second strok es dif fer considerably at dif ferent stations. Figure 4–1

PAGE 154

140 T able 4–3. NLDN-reported parameters for rst return strok es of natural ashes recorded by the MSE netw ork (Courtesy of V aisala Corporation). NLDN reported Calculated 50% Ellipse Number of NLDN-reported peak current distance to UF semi-major Ellipse reporting Flash ID location ( kA ) to wer ( km ) axis ( km ) eccentricity c2sensors MSE-0101 29.942 N 82.043 W -38.8 1.1 0.4 1.0 1 17 MSE-0102 29.939 N 82.029 W -27.8 0.5 0.4 1.0 1 12 MSE-0201 29.980 N 82.040 W -10.0 4.3 5.5 6.9 1 3 MSE-0202 29.943 N 82.074 W +50.7 4.0 0.8 2.0 4 8 MSE-0203 29.994 N 82.035 W -18.1 0.4 0.5 1.2 1 8 MSE-0204 29.943 N 82.025 N -39.5 0.7 0.4 1.0 2 10 MSE-0205 29.942 N 82.039 W -14.7 0.7 0.7 1.7 1 6 MSE-0206 29.939 N 81.995 W -76.7 3.6 0.4 1.0 1 8 MSE-0207 29.949 N 82.019 W -38.5 1.5 0.4 1.0 2 10 MSE-0208 29.955 N 81.998 W -48.3 3.5 0.4 1.0 1 4 MSE-0209 29.931 N 82.043 W -135.1 1.7 0.4 1.0 1 10 MSE-0210 29.946 N 82.029 W -81.4 0.5 0.4 1.0 1 8 MSE-0211 29.941 N 82.033 W -17.7 0.2 0.4 1.0 1 6

PAGE 155

141 Figure 4–1. Frame of video sho wing the rst return strok e of natural positi v e ash MSE-0202. The channel is seen in the vie w from IS1, which f aces roughly south-west. sho ws a frame of video from each of the four video camera locations, with the vie w from IS1 (f acing roughly south-west) sho wing the rst return strok e of this e v ent. Figure 4–4 sho ws the recorded electric elds, magnetic elds, and optical signals for ash MSE-0202 on an 800 ms time scale. Non-functional measurements, such as E-10, are indicated on the gure. In addition, time 0 is set to be the trigger point of the Y ok og a w a DL716 digitizer (discussed in Section 3.5.1 ), and hence the time scale ranges from 200 to 600 ms. It should be noted that not much detail can be observ ed on an 800 ms time scale. Ne v ertheless, it does pro vide an o v erall picture of the ash. In particular the electric eld changes of the tw o return strok es, separated by approximately 53 ms, are seen. Figures 4–5 and 4–6 sho w the recorded electric elds, magnetic elds, and optical signals corresponding to the rst and second leader/return strok e sequences on 130 ms and 70 ms time scales, respecti v ely More detail can be observ ed on these time scales, b ut the microsecond-scale magnetic elds and optical signals are still unresolv ed. In order to vie w the magnetic elds and optical signals clearly a time scale that displays

PAGE 156

142 only a fe w tens of microseconds around the return strok e must be used. Since this section is only intended to pro vide some e xample w a v eforms, the optimal time scales for each w a v eform will not be sho wn. Figure 4–7 sho ws the recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of ash MSE-0202 on a 10 s time scale. 4.2.2 Natural Flash MSE-0203 Natural ash MSE-0203 w as a ne g ati v e cloud-to-ground ash that occurred at approximately 18:00:44.072 UT on August 30, 2002. The electric elds, magnetic elds, and optical signals were recorded for se v en leader/return strok e sequences. The video records indicate at least tw o termination points and the relati v e amplitudes of the electric eld changes for each strok e dif fer at dif ferent locations, as sho wn in Figure 4–8 which displays the electric elds, magnetic eld, and optical signals on an 800 ms time scale.. The Y ok og a w a DL716 digitizer triggered on the third leader/return strok e sequence. The electric and magnetic eld time-deri v ati v es were recorded for only this strok e and sho w man y leader -step pulses. It has not yet been determined which return strok es in the electric and magnetic eld records correspond to which termination point. As with ash MSE-0202, the details of the w a v eforms are hard to discern on this time scale, b ut these w a v eforms are only intended to be for surv e y purposes. Both of the optical w a v eforms sho w se v ere ne g ati v e DC of fsets which result in only the peaks of the w a v eforms being discernible for some of the return strok es, with the rest being completely unresolv ed. Figure 4–2 sho ws a frame of video from each of the four camera locations. A f aint channel can be seen in the vie w from IS3, which f aces roughly south-east. Ho we v er in other frames, a poorly resolv ed channel can be seen in another location. An e xtensi v e analysis of the video has not yet been performed.

PAGE 157

143 Figure 4–2. Frame of video sho wing a channel from natural ne g ati v e ash MSE-0203. The channel is seen in the vie w from IS3, which f aces roughly south-east. The corrupted image seen in the vie w from IS4 is due to electromagnetic interference from the lightning. 4.2.3 Natural Flash MSE-0205 Natural ash MSE-0205 w as a ne g ati v e cloud-to-ground ash that occurred at approximately 18:13:55.827 UT on August 30, 2002. The electric elds, magnetic elds, and optical signals were recorded for four leader/return strok e sequences while the electric and magnetic eld time-deri v ati v e signals were recorded only for the rst return strok e. The video records possibly indicate tw o termination points, although the lightning is poorly resolv ed due to the rain and reections on the windo ws in front of the cameras. Figure 4–3 sho ws a frame of video from each of the four camera locations. The channel can be seen in the vie ws from IS3 and IS4, which f ace roughly south-east and east, respecti v ely In the pre vious frame, a channel can be seen in the vie w from IS2, although it is not yet clear whether it is the same channel seen from IS3 and IS4. As with ash MSE-0203, an e xtensi v e analysis of the video has not yet been performed.

PAGE 158

144 Figure 4–3. Frame of video sho wing natural ne g ati v e ash MSE-0205. The channel is seen in the vie ws from IS3 and IS4, which f ace roughly south-east and east, respecti v ely Figure 4–9 sho ws the recorded electric elds, magnetic elds, and optical signals for ash MSE-0205 on an 800 ms time scale. One of the termination points w as apparently v ery close to Station 4 since the electric eld measured there saturated at approximately 55 kV m 1 As with ash MSE-0203, the optical w a v eforms sho w se v ere ne g ati v e DC of fsets which result in only the peaks of the w a v eforms being discernible. Figure 4–10 sho ws the recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of ash MSE-0205 on a 10 s time scale. 4.2.4 T rigger ed Flash FPL-0205 Flash FPL-0205 w as an unintentional ne g ati v e altitude-triggered ash (rock et launched from the 11 m launch to wer) that w as initiated at approximately 16:26:10.767 UT on July 9, 2002. This w as the rst ash recorded by the 2002 MSE netw ork. F our return strok es attached to IS1, which is adjacent to the Launch Control T railer A frame of video sho wing this e v ent is sho wn in Figure 3–39 in Section 3.6 and a still photograph (e xposed o v er se v eral seconds) is sho wn in Figure 2–7 in Section 2.1.2.2

PAGE 159

145 Figure 4–11 sho ws the recorded electric elds, magnetic elds, current w a v eforms, and optical signals for ash FPL-0205 on an 800 ms time scale. Only a fe w electric and magnetic eld time-deri v ati v e w a v eforms functioned properly and these are poorly resolv ed. This is because the problems with the system were still being w ork ed out and the optimal attenuation settings and digitization rates had not been found yet. These w a v eforms will not be presented here.

PAGE 160

146 Efield Station 10 ILowTower IHighTower 200 0 200 400 600 ms 35 0 35 kV/mEfield Station 5 200 0 200 400 600 ms 35 0 35 kV/mEfield Station 2 200 0 200 400 600 ms 0.05 0 0.05 SouthWest Optical 200 0 200 400 600 ms 40 0 40 uW/m2Bfield Station 9 loop NS 200 0 200 400 600 ms 35 0 35 kV/mEfield Station 6 200 0 200 400 600 ms 40 0 40 uW/m2Bfield Station 4 NS loop 200 0 200 400 600 ms 35 0 35 kV/mEfield Station 4 200 0 200 400 600 ms 0.05 0 0.05 NorthEast Optical 200 0 200 400 600 ms 35 0 35 kV/mEfield Station 9 N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–4. Recorded electric elds, magnetic elds, and optical signals for natural positi v e ash MSE-0202 on an 800 ms time scale.

PAGE 161

147 120 87.5 55 22.5 10 ms 0 10 20 30 kV/mEfield Station 5 120 87.5 55 22.5 10 ms 0 10 20 30 kV/mEfield Station 2 120 87.5 55 22.5 10 ms 0 0.02 0.04 SouthWest Optical 120 87.5 55 22.5 10 ms 40 30 20 10 0 10 uW/m2Bfield Station 9 loop NS 120 87.5 55 22.5 10 ms 0 10 20 30 kV/mEfield Station 6 120 87.5 55 22.5 10 ms 40 30 20 10 0 10 uW/m2Bfield Station 4 NS loop 120 87.5 55 22.5 10 ms 0 10 20 30 kV/mEfield Station 4 120 87.5 55 22.5 10 ms 0 0.02 0.04 NorthEast Optical 120 87.5 55 22.5 10 ms 0 10 20 30 kV/mEfield Station 9 Efield Station 10 IHighTower ILowTower Measurement Under Repair Measurement Malfunction Measurement Malfunction N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–5. Recorded electric elds, magnetic elds, and optical signals for the rst leader/return strok e sequence of natural positi v e ash MSE-0202 on a 130 ms time scale.

PAGE 162

148 5 20 35 50 65 ms 0 10 20 30 kV/mEfield Station 5 5 20 35 50 65 ms 0 10 20 30 kV/mEfield Station 2 5 20 35 50 65 ms 0 0.02 0.04 SouthWest Optical 5 20 35 50 65 ms 2 0 2 4 6 8 uW/m 2Bfield Station 9 loop NS 5 20 35 50 65 ms 0 10 20 30 kV/mEfield Station 6 5 20 35 50 65 ms 2 0 2 4 6 8 uW/m 2Bfield Station 4 NS loop 5 20 35 50 65 ms 0 10 20 30 kV/mEfield Station 4 5 20 35 50 65 ms 0 0.02 0.04 NorthEast Optical 5 20 35 50 65 ms 0 10 20 30 kV/mEfield Station 9 Efield Station 10 IHighTower ILowTower Measurement Under Repair Measurement Malfunction Measurement Malfunction N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–6. Recorded electric elds, magnetic elds, and optical signals for the second leader/return strok e sequence of natural positi v e ash MSE-0202 on a 70 ms time scale.

PAGE 163

149 0 5 10 s 5 0 5 kV/m/usdE/dt Station 4 0 5 10 s 10 0 10 W/m2/sdB/dt Station 1 NS loop 0 5 10 s 10 0 10 W/m2/sdB/dt Station 1 EW loop 0 5 10 s 5 0 5 kV/m/usdE/dt Station 1 0 5 10 s 5 0 5 kV/m/usdE/dt Station 9 0 5 10 s 5 0 5 kV/m/usdE/dt Station 8 0 5 10 s 10 0 10 W/m 2 /sdB/dt Station 4 NS loop dB/dt Station 9 NS loop N E W S 100 mLaunch Control SATTLIF Underground Launcher Station 9 Station 1 Station 6 Station 5 Station 4 Station 2 Station 10 Station 8SW Optical IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road NE Optical Blast Wall Military Container Figure 4–7. Recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of natural positi v e ash MSE-0202 on a 10 s time scale.

PAGE 164

150 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 5 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 2 200 0 200 400 600 ms 0 0.02 0.04 0.06 SouthWest Optical 200 0 200 400 600 ms 15 10 5 0 5 uW/m2Bfield Station 9 NS loop 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 6 200 0 200 400 600 ms 15 10 5 0 5 uW/m2Bfield Station 4 NS loop 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 4 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 9 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 10 200 0 200 400 600 ms 0 0.02 0.04 0.06 NorthEast Optical IHighTower ILowTower Measurement Malfunction Measurement MalfunctionMeasurement Saturated N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–8. Recorded electric elds, magnetic elds, and optical signals for natural ne g ati v e ash MSE-0203 on an 800 ms time scale.

PAGE 165

151 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 5 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 2 200 0 200 400 600 ms 0 0.02 0.04 0.06 SouthWest Optical 200 0 200 400 600 ms 10 0 10 20 uW/m2Bfield Station 9 NS loop 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 6 200 0 200 400 600 ms 10 0 10 20 uW/m2Bfield Station 4 NS loop 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 4 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 9 200 0 200 400 600 ms 60 40 20 0 kV/mEfield Station 10 200 0 200 400 600 ms 0 0.02 0.04 0.06 NorthEast Optical IHighTower ILowTower Measurement Malfunction Measurement MalfunctionMeasurement Saturated N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–9. Recorded electric elds, magnetic elds, and optical signals for natural ne g ati v e ash MSE-0205 on an 800 ms time scale.

PAGE 166

152 0 5 10 s 20 0 20 kV/m/usdE/dt Station 4 0 5 10 s 30 0 30 W/m2/sdB/dt Station 1 NS loop 0 5 10 s 30 0 30 W/m2/sdB/dt Station 1 EW loop 0 5 10 s 20 0 20 kV/m/usdE/dt Station 1 0 5 10 s 20 0 20 kV/m/usdE/dt Station 9 0 5 10 s 20 0 20 kV/m/usdE/dt Station 8 0 5 10 s 30 0 30 W/m 2 /sdB/dt Station 4 NS loop dB/dt Station 9 NS loop Sensor Malfunction N E W S 100 mLaunch Control SATTLIF Underground Launcher Station 9 Station 1 Station 6 Station 5 Station 4 Station 2 Station 10 Station 8SW Optical IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road NE Optical Blast Wall Military Container Figure 4–10. Recorded electric and magnetic eld time-deri v ati v e w a v eforms for the rst return strok e of natural ne g ati v e ash MSE-0205 on a 10 s time scale.

PAGE 167

153 Efield Station 5 Measurement Under Repair 200 0 200 400 600 ms 40 30 20 10 0 10 kV/mEfield Station 2 200 0 200 400 600 ms 0 0.02 0.04 0.06 SouthWest Optical 200 0 200 400 600 ms 10 5 0 uW/m2Bfield Station 9 NS loop 200 0 200 400 600 ms 40 30 20 10 0 10 kV/mEfield Station 6 200 0 200 400 600 ms 10 5 0 uW/m2Bfield Station 4 NS loop 200 0 200 400 600 ms 40 30 20 10 0 10 kV/mEfield Station 4 200 0 200 400 600 ms 0 0.02 0.04 0.06 NorthEast Optical 200 0 200 400 600 ms 40 30 20 10 0 10 kV/mEfield Station 10 200 0 200 400 600 ms 40 30 20 10 0 10 kV/mEfield Station 9 200 0 200 400 600 ms 1 0.5 0 0.5 1 kATower Current (Low) IHighTower Measurement Malfunction N E W SLaunch Control SATTLIF Underground Launcher 100 m Station 6 Station 9 Station 1 Station 5 Station 4 Station 2 Station 10 Station 8IS4 IS2 IS3 RunwayDuPont Three Phase Line Access RoadOffice Trailer SATTLIF Test Power Line Tower Launcher IS1 Access Road Blast Wall NE Optical Military Container SW Optical Figure 4–11. Recorded electric elds, magnetic elds, current, and optical signals for ne g ati v e altitude-triggered ash FPL-0205 on an 800 ms time scale.

PAGE 168

CHAPTER 5 RECOMMEND A TIONS FOR FUTURE RESEARCH The w ork presented in this thesis represents a rst step in obtaining a clearer picture of the ph ysics of natural lightning using the MSE netw ork at the ICLR T The data obtained thus f ar samples of which are presented in Chapter 4 and future data will require years of w ork to analyze fully both for the author as part of his Ph.D dissertation and for graduate students that follo w him. In addition, as with most research endea v ors, these ne w data, while pro viding some answers, will lik ely raise man y ne w questions. Se v eral changes to the MSE should be implemented, which w ould signicantly increase the quality of the recorded data. First, the bandwidths and digitization rates of the electric eld, magnetic eld, and current measurements should be increased. If the bandwidths of the measurements are increased, f aster rising signals can be recorded. This w ould require purchasing a ne w digitizer that has a higher bandwidth and digitization rate than the Y ok og a w a DL716 (described in Section 3.5.1 ) b ut also is capable of recording record lengths on the order of one second for at least 16 channels simultaneously In addition, as described in Section 3.3.1.3 the operating range of the electric eld inte grating capacitors is currently limited to about 5 MHz. Therefore, this limitation w ould ha v e to be addressed before increasing the bandwidth of digitizer Similarly the bandwidth of the magnetic eld antennas, currently limited to about 4 MHz w ould ha v e to be increased, possibly by using a smaller antenna. Calibrated outputs for the optical detectors (described in Section 3.3.3 ) w ould pro vide e xtremely useful information re g arding the relationship between the current, magnetic elds, and optical output of the lightning channel. 154

PAGE 169

155 No time-deri v ati v e measurements ha v e been obtained for natural subsequent strok es. Unlik e the electric and magnetic eld measurements, a separate trigger signal from the optical detectors is required to record each indi vidual return strok e from the deri v ati v e measurements. This limitation is due to the se gmented memory conguration of the digitizers, as discussed in Sections 3.5.2 and 3.5.3 The lack of time-deri v ati v e measurements from subsequent return strok es is presumably due to the relati v ely lo w (about a f actor of tw o lo wer) peak currents (and hence light output) of subsequent return strok es relati v e to those of rst strok es. The subsequent strok es' optical output w as apparently lo wer than the trigger threshold of the system. It should be noted that the MSE did trigger on se v eral classical rock et triggered e v ents, which contain return strok e peak currents similar to those in natural subsequent strok es. It is speculated that the reason the MSE triggered on these e v ents is because the launching locations were well within the boundary of the netw ork, and easily within the vie w of both optical detectors. In contrast, se v eral natural e v ents recorded are belie v ed to be near the boundary of the netw ork, and thus not in a v ery good vie w of at least one of the optical detectors. While the optical signal from a rst return strok e could be bright enough to o v ercome this problem, the signal from a subsequent strok e may not ha v e been. In addition, branching is common in natural lightning, and hence it is possible that the rst and subsequent strok es terminated at dif ferent points on ground with the location of the subsequent strok e being outside the boundary of the netw ork. Furthermore, e v en in the triggered-lightning case, not all strok es were recorded by the system, despite their f a v orable location. Therefore, it is recommended that the sensiti vity of the triggering system be increased in order to hopefully raise the probability of recording the electric and magnetic eld time-deri v ati v es from subsequent strok es in natural lightning. Ob viously the disadv antage of this modication is that the probability of triggering on an unw anted e v ent, such as a

PAGE 170

156 bright cloud dischar ge, is increased, thus disabling the system for a time period when a w anted cloud-to-ground ash could occur within the netw ork. It w ould be benecial to install Opticomm MMV -120C ber -optic links (discussed in Section 3.4.1 ) at all measurement locations and to eliminate the Nicolet Isobe and Meret links from the system. This w ould ensure that all parameters were measured as uniformly as possible. In addition, the Meret links, used in the dB/dt measurements, tend to be noisy and are A C coupled, both features being undesirable. The chief adv antage of the Meret links is that each unit has a b uilt-in dif ferential preamplier as discussed in Section 3.4.3 If these links were replaced with Opticomm links, separate dif ferential preampliers w ould ha v e to be b uilt. The video records, in general, pro vide useful information re g arding the approximate location and geometry of the lightning channel. Ho we v er as discussed in Section 3.6 the video system w as not automated in 2001 or 2002, and hence no video w as recorded for se v eral e v ents. Unfortunately without good video, e xact locations are impossible to determine. Therefore, a system to automate the video w as de v eloped in late 2002, and will be inte grated into the MSE during the summer of 2003. Furthermore, the addition of se v eral thunder sensors to the netw ork could potentially pro vide acoustic time-of-arri v al information for channel location and geometry reconstruction, although it may not be accurate enough. At the time of this writing, the MSE is still operational at the ICLR T f acility under the 2002 conguration, as described in this thesis. While continuing to record data from natural lightning, the lightning research group is de v eloping an e xperiment to completely characterize the initial stage (or wireb urn phase) of rock et-triggered lightning. This e xperiment in v olv es measuring the current, electromagnetic elds, and optical signals from the initial stage as well as using an e xtremely high speed (nanosecond resolution) camera to record the destruction of the triggering wire. Moreo v er the MSE will be in v olv ed in continuing the w ork of detecting high-ener gy

PAGE 171

157 radiation from both natural and triggered lightning, as described in ( Dwyer et al. 2003 ). Also, the MSE will be in v olv ed in an e xperiment to characterize lightning induced ef fects on a v ertical distrib ution line. Finally as noted in the rst paragraph of this section, the netw ork is e xpected to pro vide the data o v er the ne xt year or tw o that will be analyzed as part of the author' s Ph.D dissertation.

PAGE 172

LIST OF REFERENCES K. Ber ger R. B. Anderson, and H. Kroninger P arameters of lightning ashes. Electr a 80:23–37, 1975. K. Ber ger and E. V ogelsanger Photographische blitzuntersuchunger der jahre 1955-1965 auf dem Monte San Salv atore. Bull. Sc hweiz. Elektr otec h. V er 57: 599–620, 1966. D. E. Cra wford. Multiple-station measurements of triggered lightning electric and magnetic elds. Master' s thesis, Uni v ersity of Florida, Gainesville, Florida, 1998. D. E. Cra wford, V A. Rak o v M. A. Uman, G. H. Schnetzer K. J. Rambo, M. V Stapleton, and R. J. Fisher The close lightning electromagnetic en vironment: Dart-leader electric eld change v ersus distance. J ournal of Geophysical Resear c h 106:14,909–14,917, 2001. J. R. Dwyer M. A. Uman, H. K. Rassoul, M. Al-Dayeh, E. L. Cara w ay J. Jerauld, V A. Rak o v D. M. Jordan, K. J. Rambo, V Corbin, and B. Wright. Ener getic radiation produced during rock et-triggered lightning. Science Ma gazine 299: 694–697, 2003. P R. Krehbiel, M. Brook, R. L. Lhermitte, and C. L. Lennon. Lightning char ge structure in thunderstorms. In L. H. Ruhnk e and J. Latham, editors, Pr oceedings in Atmospheric Electricity pages 408–411, 1983. P aul R. Krehbiel. An analysis of the electric eld c hang e pr oduced by lightning PhD thesis, Uni v ersity of Manchester Institute of Science and T echnology Manchester United Kingdom, 1981. E. P Krider C. D. W eidman, and R. C. Noggle. The electric elds produced by lightning stepped leaders. J ournal of Geophysical Resear c h 82:951–960, 1977. C. Leteinturier J. H. Hamelin, and A. Eybert-Berard. Submicrosecond characteristics of lightning return-strok e currents. IEEE T r ansactions on Electr oma gnetic Compatibility 33:351–357, 1991. V Mazur and L. H. Ruhnk e. Ev aluation of the lightning protection system at the WSR-88D radar sites. T echnical report, National Se v ere Storms Laboratory Norman, Oklahoma, 2001. V A. Rak o v D. E. Cra wford, K. J. Rambo, G. H. Schnetzer M. A. Uman, and R. T ottappillil. M-component mode of char ge transfer to ground in lightning dischar ges. J ournal of Geophysical Resear c h 106:22,817–22,831, 2001. 158

PAGE 173

159 V A. Rak o v and M. A. Uman. Lightning: Physics and Ef fects Cambridge Uni v ersity Press, Cambridge, United Kingdom, 2003. V A. Rak o v M. A. Uman, D. E. Cra wford, J. Schoene, J. Jerauld, K. J. Rambo, G. H. Schnetzer B. A. DeCarlo, and M. Miki. Close lightning electromagnetic en vironment: T riggered-lightning e xperiments. In Pr oceedings of the 15th International Zuric h Symposium and T ec hnical Exhibition on Electr oma gnetic Compatibility pages 445–450, 2003. V A. Rak o v M. A. Uman, K. J. Rambo, M. I. Fernandez, R. J. Fisher G. H. Schnetzer R. Thottappillil, A. Eybert-Berard, J. P Berlandis, P Lalande, A. Bonamy P Laroche, and A. Bondiou-Cler gerie. Ne w insights into lightning processes g ained from triggered-lightning e xperiments in Florida and Alabama. J ournal of Geophysical Resear c h 103:14,117–14,130, 1998. J. Schoene, M. A. Uman, V A. Rak o v V K odali, K. J. Rambo, and G. H.Schnetzer Statistical characteristics of the electric and magnetic elds and their time deri v ati v es 15 m and 30 m from triggered lightning. J ournal of Geophysical Resear c h 108: 4192–4209, 2003. M. A. Uman. The Lightning Disc har g e Academic Press, San Die go, California, 1987. M. A. Uman, J. Schoene, V A. Rak o v K. J. Rambo, and G. H. Schnetzer Correlated time deri v ati v es of current, electric eld intensity and magnetic ux density for triggered lightning at 15 m. J ournal of Geophysical Resear c h 107, 2002. D. W ang, V A. Rak o v M. A. Uman, M. I. Fernandez, K. J. Rambo, G. H. Schnetzer and R. J. Fisher Characterization of the initial stage of ne g ati v e rock et-triggered lightning. J ournal of Geophysical Resear c h 104:4213–4222, 1999.

PAGE 174

BIOGRAPHICAL SKETCH Jason E. Jerauld w as born December 30, 1979 in Ne wton, Massachusetts and is the only child of Ronald Jerauld and Janice Desrosiers. In February of 1990, he mo v ed to Florida with his f amily In June of 1997, Mr Jerauld graduated from Lak e Re gion High School in Eagle Lak e, Florida and he entered the Uni v ersity of Florida in August of 1997. In the f all of 2000, as an under graduate, he became in v olv ed with the Lightning Research Laboratory In May of 2001, he recei v ed a Bachelor of Science in Electrical Engineering (with Honors) from the Uni v ersity of Florida. Mr Jerauld be g an his graduate studies in summer 2001 and w as a w arded the Robert E. Pittman graduate fello wship for 2 consecuti v e years. In addition, during the summmers of 2001 and 2002, he has participated in lightning e xperiments at the International Center for Lightning Research and T esting. He has co-authored tw o papers in re vie wed journals, six papers in conference proceedings, and v e technical reports. Moreo v er he has been seen on ABC W orld Ne ws T onight and Nightline. After the completion of his master' s w ork, he will continue to pursue a Ph.D under the guidance of Drs. Vladimir Rak o v and Martin Uman. 160


Permanent Link: http://ufdc.ufl.edu/UFE0000708/00001

Material Information

Title: A Multiple-station experiment to examine the close electromagnetic environment of natural and triggered lightning
Physical Description: Mixed Material
Creator: Jerauld, Jason E. ( Author, Primary )
Publication Date: 2003
Copyright Date: 2003

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: UFE0000708:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000708/00001

Material Information

Title: A Multiple-station experiment to examine the close electromagnetic environment of natural and triggered lightning
Physical Description: Mixed Material
Creator: Jerauld, Jason E. ( Author, Primary )
Publication Date: 2003
Copyright Date: 2003

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: UFE0000708:00001


This item has the following downloads:


Full Text











A MULTIPLE-STATION EXPERIMENT TO EXAMINE THE CLOSE
ELECTROMAGNETIC ENVIRONMENT OF NATURAL AND TRIGGERED
LIGHTNING
















By

JASON E. JERAULD


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


2003

































Copyright 2003

by

Jason E. Jerauld



































For my parents, Ronald Jerauld and Janice Desrosiers.














ACKNOWLEDGMENTS

The work presented in this thesis would not be possible without the guidance of

Drs. Vladimir Rakov and Martin Uman, to whom I give my sincerest thanks. Their

scientific rigor and integrity have given me no less than the best example of how to

conduct myself as a scientist and a scholar.

I would also like to express my appreciation to Dr. Doug Jordan, Michael

Stapleton, Robert Olsen III, Alonso Guarisma, Oliver Pankiewicz, Thomas Rambo, Joe

Richard, Clifford Jordan, Julia Jordan, Andrew Owens, Matt Riley, and everyone else

who dug a ditch or ran a fiber related to this experiment.

I owe a special debt of gratitude to George Schnetzer and Keith Rambo, who I

believe taught me the most of what I know about good engineering and field research.

I would like to thank Dr. Carlos Mata for working with me to provide the

informative and aesthetically pleasing figures found in Chapter 4. Also, I would like

to thank John Cramer of Vaisala Corporation for providing the NLDN information

presented in Chapter 4.

I must also give thanks to Kathy Thomson for all of the help and support she has

given me since my first days in the lightning research business.

I am also grateful to the people of the linux and ITEX 2F communities, for

without them the process of writing this thesis would have been much more difficult.

In particular, I give my thanks to Ron Smith, who developed the ufthesis document

class.

Finally, I would like to thank my family and friends for their patience and support,

which never waivered, no matter how much I complained about writing this thesis.









This work was supported in part by NSF Grant ATM-0003994, U.S. DOT (FAA)

Grant 99-G-043, Sandia National Laboratories Contract KOMO42296, and by Florida

Power and Light Corporation.















TABLE OF CONTENTS


ACKNOW LEDGMENTS ................................

LIST OF TABLES .................

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


ABSTRACT ...............

1 INTRODUCTION ........


. .. . x iii


2 LITERATURE REVIEW ........................... 4

2.1 The Lightning Discharge Process ........ . ...... 4
2.1.1 Natural Lightning .... .............. ... 4
2.1.2 Triggered Lightning ...................... 9
2.1.2.1 Classical-triggered lightning ... . ...... 9
2.1.2.2 Altitude-triggered lightning ............... 13
2.2 Overview of the ICLRT Facility ..... . . . ...... 16
2.3 Multiple-Station Field Measurements of Natural and Rocket-Triggered
L lightning . . . . . . . . 19

3 INSTRUMENTATION . . . . . . . 22

3.1 Overview of the 2001 and 2002 Multiple Station Experiments . 22
3.2 Control System . . . . . . . 25
3.2.1 The PIC Controller . . . . . . 28
3.2.2 Softw are . . . . . . . 41
3.2.3 Triggering System . . . . . . 45
3.3 Measurement Implementation . . . . . 49
3.3.1 Electric Field and Electric Field Time-Derivative Measurements 49
3.3.1.1 Analysis of a conducting flat-plate antenna . 51
3.3.1.2 Flat-plate antenna implementation . . . 64
3.3.1.3 Electric field measurement implementation . 67
3.3.1.4 Electric field time-derivative measurement
implementation . . . 77
3.3.2 Magnetic Field and Magnetic Field Time-Derivative
M easurem ents . . . . . . 80
3.3.2.1 Analysis of a loop antenna . . . . 80
3.3.2.2 Loop antenna implementation ... . ...... 89
3.3.2.3 Magnetic field measurement implementation . 95


page









3.3.2.4 Magnetic field time-derivative measurement
implementation . . . . . 104
3.3.3 Optical Measurements . . . . . 108
3.3.4 Current Measurements . . . . . 111
3.3.5 Measurement Bandwidth Summary . . . 118
3.4 Fiber-Optic Links . . . . . . 118
3.4.1 Opticomm MMV-120C Fiber-Optic Links . . . 119
3.4.2 Nicolet Isobe 3000 Fiber-Optic Links ...... . . 120
3.4.3 Meret MDL288DC Fiber-Optic Links . . 120
3.4.4 Fiber-Optic Cables . . . . . . 121
3.4.5 Fiber-Optic Link Calibration . . . . 123
3.5 D igitizers . . . . . . . . 129
3.5.1 Yokogawa DL716 . . . . . . 129
3.5.2 LeCroy LT344 Waverunner . . . . . 131
3.5.3 LeCroy LT374 Waverunner2 . . . . 133
3.6 Video System . . . . . . . 133
3.7 Tim ing System . . . . . . . 135

4 PRESENTATION OF DATA . . . . . . 136

4.1 Data Summary . . . . . . . 136
4.2 Example Waveforms . . . . . . 139
4.2.1 Natural Flash MSE-0202 . . . . . 139
4.2.2 Natural Flash MSE-0203 ................... ..142
4.2.3 Natural Flash MSE-0205 . . . . . 143
4.2.4 Triggered Flash FPL-0205 . . . . . 144

5 RECOMMENDATIONS FOR FUTURE RESEARCH . . 154

LIST OF REFERENCES . . . . . . . . 158

BIOGRAPHICAL SKETCH . . . . . . . 160















LIST OF TABLES
Table page

3-1 Measured parameters for the 2001 MSE. . . . . 25

3-2 Measured parameters for the 2002 MSE. . . . . 27

3-3 Coordinates of the MSE measurement locations...... . . 27

3-4 Format of a PIC controller command data packet. . . . 38

3-5 Format of a PIC controller response data packet..... . . 38

3-6 Bit settings of PIC controller commands. . . . . 40

3-7 Salient characteristics of the 2001 MSE electric field measurements 76

3-8 Salient characteristics of the 2002 MSE electric field measurements 76

3-9 Salient characteristics of the 2001 MSE magnetic field measurements. 104

3-10 Salient characteristics of the 2002 MSE magnetic field measurements. 104

3-11 Estimated bandwidths of the 2001 and 2002 MSE measurements. .. . 119

3-12 MSE fiber-optic link summary. . . . . . . 119

3-13 OTDR measured optical lengths and corresponding time delays for armored
fiber-optic cables used during the 2002 MSE....... . . 128

3-14 Calculated time delays for the fiber-optic cables used in the 2001 MSE. 130

4-1 Flashes recorded by the MSE network in 2001...... . . 137

4-2 Flashes recorded by the MSE network in 2002...... . . 138

4-3 NLDN-reported parameters for first return strokes of natural flashes recorded
by the M SE network . . . . . . . 140















LIST OF FIGURES
Figure page

2-1 General locations of ground flash charge sources observed in
summer thunderstorms in Florida and New Mexico and in winter thunderstorms
in Japan, using simultaneous measurements of electric field at several ground
stations . .............. .. ................. 5

2-2 Lightning cloud-to-ground discharges categorized by leader propagation
direction and polarity of charge transferred to ground..... . . 6

2-3 Fiberglass rocket with a spool of Kevlar-sheathed copper triggering wire
m mounted to the base . . . . . . . 10

2-4 Sequence of events observed in a typical classical rocket triggered lightning
flash.. . . .. . . . ...... ... 11

2-5 Classically triggered flash S-0116, initiated on July 27, 2001 21:56:06 UT.
The vaporized wire channel (initial stage) is on the right with the individual
return stroke channels blown to the left by the wind. The launcher is located
underground . . . . . . . . 13

2-6 Sequence of events observed in a typical altitude rocket triggered lightning
flash ....... ... ... .... ... ... ........... ..... 14

2-7 Unintentional altitude triggered flash FPL-0205, initiated on July 9, 2002
16:26:10 UT. The launcher (the intended strike point on the insulating tower)
is pictured in the foreground . . . . . . 16

2-8 Sketch of the ICLRT at Camp Blanding, Florida...... . 18

3-1 Sketch of the 2001 and 2002 MSE measurement locations at the ICLRT.
The arrows roughly indicate the orientation of video cameras and optical
sensors at those locations . . . . . . 23

3-2 Simplified diagram of the MSE control and data acquisition system. . 26

3-3 The PIC controller. A) Front view. B) Side view. . . . 29

3-4 Diagram of how a PIC controller is installed with a measurement. .. . 30

3-5 Diagram of the PIC controller communication topology used during the
2001 M SE . . . . . . . . 34









3-6 Diagram of the PIC controller communication topology used during the
2002 M SE . . . . . . . . 35

3-7 PIC RF unit enclosure mounted with a solar cell...... . . 38

3-8 Flowchart representation of the final 2002 MSE software control algorithm. 46

3-9 Schematic of the MSE trigger circuit. Enclosed in the dashed lines are
the reference voltage circuits used in the bistable circuits. . 50

3-10 Illustration of the electric field boundary condition at a perfectly conducting
su rface . . . . . . . . . 5 1

3-11 Frequency-domain equivalent circuit, using a Norton equivalent current
source, of a flat-plate antenna sensor feeding a load (represented by ZL). 54

3-12 Aluminum flat plate antenna used in the MSE. . . . . 64

3-13 Detailed mechanical drawing of the aluminum flat plate antenna used in
the M SE . . . . . . . . . 66

3-14 Diagram of an installation of a MSE measurement utilizing a flat-plate
antenna . . . . . . . . . 67

3-15 Integrator capacitor assembly used in 2001. A) Closed Pomona box. B)
Box open to show interior . . . . . . 69

3-16 Measured (dashed line) and expected (solid line) test circuit responses for
integrating capacitor unit 01-01 (0.477 F). . . . . 70

3-17 Measured (dashed line) and expected (solid line) test circuit responses for
integrating capacitor unit 02-09 (0.209 F). . . . . 71

3-18 Schematic of the high-impedance amplifier used in the 2001 MSE. .. 72

3-19 Measured frequency response of high impedance amplifier 01-07. .. . 73

3-20 Diagram of a MSE electric field measurement. . . . . 74

3-21 Diagram of a MSE dE/dt measurement . . . . . 77

3-22 Frequency-domain equivalent circuit, using a Thevenin equivalent voltage
source, of a loop antenna sensor feeding a load (represented by ZL) . 83

3-23 Square loops of 50 Q coaxial cable in 4 inch PVC pipe. A) Single loop
at Station 4. B) Crossed loops at Station 4. . . . . 90

3-24 Diagram (A) and equivalent circuit (B) of a differential-output coaxial loop
antenna with both ends of the cable terminated in 50 .. . 92









3-25 Diagram (A) and equivalent circuit (B) of a single-ended output coaxial
loop antenna with both ends of the cable terminated in 50 Q. . . 94

3-26 Schematic of an example active integrator. . . . . 96

3-27 Schematic of the active integrator used in the 2001 MSE. . . 98

3-28 Measured (dashed line) and ideal (solid line) magnitude responses of one
of the 2001 MSE active integrator units. . . . . . 99

3-29 Schematic of the active integrator used in the 2002 MSE. . . 100

3-30 Diagram of a 2001 MSE magnetic field measurement. . .. 101

3-31 Diagram of a 2002 MSE magnetic field measurement. . . 102

3-32 Diagram of a MSE dB/dt measurement. . . . . . 106

3-33 Schematic of the MSE optical sensor circuit....... . . 109

3-34 Diagram of a MSE optical measurement. . . . . 110

3-35 MSE optical measurement assembly. . . . . . 112

3-36 Aluminum rocket launcher with Hoffman box mounted to the base. .. 113

3-37 Diagram of the MSE current measurements. . . . . 116

3-38 Time-domain equivalent circuit of the MSE current measurements. .. 117

3-39 Example video frame from the MSE video system. Going clockwise from
the upper left, the four quadrants represent the camera views from IS1,
IS2, IS4, and IS3, respectively, with the lightning being in view of the IS4
cam era . . . . . . . . . 134

4-1 Frame of video showing the first return stroke of natural positive flash
MSE-0202. The channel is seen in the view from IS1, which faces roughly
south-w est . . . . . . . . 141

4-2 Frame of video showing a channel from natural negative flash MSE-0203.
The channel is seen in the view from IS3, which faces roughly south-east.
The corrupted image seen in the view from IS4 is due to electromagnetic
interference from the lightning . . . . . . 143

4-3 Frame of video showing natural negative flash MSE-0205. The channel
is seen in the views from IS3 and IS4, which face roughly south-east and
east, respectively . . . . . . . . 144

4-4 Recorded electric fields, magnetic fields, and optical signals for natural
positive flash MSE-0202 on an 800 ms time scale. . . . 146









4-5 Recorded electric fields, magnetic fields, and optical signals for the first
leader/return stroke sequence of natural positive flash MSE-0202 on a 130 ms
tim e scale . . . . . . . . . 147

4-6 Recorded electric fields, magnetic fields, and optical signals for the second
leader/return stroke sequence of natural positive flash MSE-0202 on a 70 ms
tim e scale . . . . . . . . . 148

4-7 Recorded electric and magnetic field time-derivative waveforms for the
first return stroke of natural positive flash MSE-0202 on a 10 |s time scale. 149

4-8 Recorded electric fields, magnetic fields, and optical signals for natural
negative flash MSE-0203 on an 800 ms time scale. . . . 150

4-9 Recorded electric fields, magnetic fields, and optical signals for natural
negative flash MSE-0205 on an 800 ms time scale. . . . 151

4-10 Recorded electric and magnetic field time-derivative waveforms for the
first return stroke of natural negative flash MSE-0205 on a 10 |s time scale. 152

4-11 Recorded electric fields, magnetic fields, current, and optical signals for
negative altitude-triggered flash FPL-0205 on an 800 ms time scale. .. 153















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

A MULTIPLE-STATION EXPERIMENT TO EXAMINE THE CLOSE
ELECTROMAGNETIC ENVIRONMENT OF NATURAL AND TRIGGERED
LIGHTNING

By

Jason E. Jerauld

May 2003

Chair: Martin A. Uman
Major Department: Electrical and Computer Engineering

This thesis presents a complete description of an automated experiment to measure

the close (within a few hundred meters) electromagnetic environment of natural and

rocket-triggered lightning. The experiment consists of a network of wideband sensors

spread about an area of approximately 0.5 km2 at the International Center for Lightning

Research and Testing (ICLRT), located at Camp Blanding, Florida and is a continuation

of the work presented in (Crawford et al. 2001). This network began operation during

summer 2001 and measured quantities including the vertical electric field at eight

locations, the horizontal magnetic field at two locations, and the optical output of

the bottom hundred meters or so of the lightning channel as observed from two

locations. In 2002, the network was upgraded to include electric field time-derivative

measurements at four locations and magnetic field time-derivative measurements at

three locations. In addition, the induced current was measured in an 14 m grounded

conducting structure. The system is automatically turned on and off by sensing the

ambient electric field amplitude; and is triggered by simultaneous signals from the two

optical sensors viewing the network from opposite covers. In addition to a complete









description of the sensors and control system, representative data from both natural and

triggered lightning, acquired in 2002, are presented.














CHAPTER 1
INTRODUCTION

Lightning is one of nature's most fantastic visual and auditory displays. Lightning

has played a key role in the mythology of many cultures; several of which had specific

gods dedicated to the sky, weather, and lightning. Many of these gods were also

considered gods of destruction, chaos, spite, or mischief. This association is not

coincidental and can be easily concluded by anyone who has witnessed or has fallen

victim to the awesome destructive force of lightning.

Although most people living in the twenty-first century do not attribute lightning

to supernatural forces, they are no less susceptible to its deleterious effects. It is

estimated that about 1000 people are killed from lightning each year around the

world, with approximately 100 of those people living in the United States. Those

who do survive, probably ten times those killed, are sometimes left with debilitating

nerve damage and chronic pain. According to the National Lightning Safety Institute

(NLSI, http://www. lightningsafety.com), lightning is responsible for a major

portion (about 30%) of electrical power outages across the United States, costing tens

of millions of dollars per year with total costs approaching $1 billion. In addition

to damaging the electrical infrastructure, lightning is responsible for forest fires,

fires to man-made structures, explosions of stored flammable substances (such as

petroleum products), aircraft mishaps and upsets, and damage to electronic components.

Most people have lost or have known someone who has lost computer or telephone

equipment as a result of lightning.

In addition to being a destructive force, lightning is also largely responsible

renewing the Earth's forests. Forest fires started by lightning serve to clear forests and

pave the way for new growth. In addition, many scientists theorize that several billion









years ago, lightning may have provided the "spark" needed to turn the primordial ooze

into the beginnings of life on Earth; and may have provided the chemical nutrients to

sustain it.

Obviously there are many practical reasons to study lightning. Monumental

progress has been made in the field since the first systematic lightning experiments

were conducted in the 1750s. Despite all that has been learned, there is much about

the lightning discharge process that is poorly understood, and the study of lightning

remains a very active research area. Much that has been learned is the direct result of

rocket-triggered lightning experiments, as discussed in Section 2.1.2.

This thesis presents a description of an experiment intended to measure the close

withing a few hundred meters) electromagnetic environment of lightning, that was

fielded during the summers of 2001 and 2002. This is known as the Multiple-Station

Experiment (abbreviated MSE in this thesis). The main purpose of this experiment

was to examine natural lightning, although data were also recorded for rocket-triggered

lightning. While not the first experiment of this type, it is the only one to combine

time-domain measurements of electric and magnetic fields (and their time-derivatives),

optical signals, and induced currents from close natural lightning into a single

experiment. Existing and future data obtained from this experiment should provide

a wealth of new information about the physics of natural lightning. The main goals for

this experiment are


* Characterize first return strokes in natural lightning.

* Determine whether subsequent strokes in natural lightning are identical to strokes
initiated in rocket-triggered lightning.

Chapter 2 gives a brief review of the literature concerning lightning phenomena

relevant to this thesis. Chapter 3 gives a complete description of the sensors, control

software, fiber-optics, digitizers, and video equipment; as well as how they are

combined to implement the experiment. When applicable, the distinction is made









between configurations used in the 2001 and 2002 experiments, since they differed

significantly in some cases. Chapter 4 gives a summary of all data recorded in 2001

and 2002 as well as a selection of waveforms recorded during the 2002 experiment.

Finally, Chapter 5 gives some conclusions and recommendations for future work.

It should be noted that while the author was primarily responsible for organizing

and implementing the 2002 experiment, much of the 2001 experiment was the work

of George Schnetzer (a consultant on the project who is formerly of Sandia National

Laboratories) and Keith Rambo (chief engineer at the Lightning Research Laboratory

at the University of Florida). Their work in 2001 facilitated the implementation of the

2002 experiment immensely.














CHAPTER 2
LITERATURE REVIEW

This chapter presents a brief review of the literature concerning the lightning

phenomena relevant to this thesis. Section 2.1 presents a brief introduction to the

physics of lightning, both natural and artificially-triggered. Section 2.2 gives a

brief overview of the ICLRT. Finally, Section 2.3 presents a review of previous

multiple-station lightning electromagnetic field measurements.

2.1 The Lightning Discharge Process

2.1.1 Natural Lightning

Lightning is an electrical discharge that is responsible for the rapid (within less

than a second or so) transfer of charge between the atmosphere and the Earth or

between different parts of the atmosphere. Large charge centers are located in clouds

termed cumulonimbus, commonly referred to as thunderclouds. In Florida, these clouds

usually exhibit an "anvil" shape and are typically about 10 to 12 km in height with a

lower visual boundary about 1 to 2 km above ground (Uman 1987).

The charge structure of a cumulonimbus can be crudely modeled as a vertical

triple consisting, in temperate regions, of a positive charge center at a height of

approximately 10 km, a negative charge center at 5 km, and another positive charge

center at 2 km. The two upper charges are usually specified to be equal in magnitude

and therefore form a dipole. The magnitude of the lower positive charge is significantly

smaller than that of the dipole charges. The general locations of ground flash charge

sources observed in summer thunderstorms in Florida and New Mexico and in winter

thunderstorms in Japan (using simultaneous measurements of electric field at several

ground stations) are shown in Figure 2-1.















Florida


-20 'C
-10 oC
() C_


Summer
Storms

New Mexico


Winter
Storms


Japan

\+i


0 oc


Figure 2-1. General locations of ground flash charge sources observed in summer
thunderstorms in Florida and New Mexico and in winter thunderstorms
in Japan, using simultaneous measurements of electric field at several
ground stations. Adapted from (Krehbiel et al. 1983) and (Rakov and
Uman 2003).


Most lightning discharges occurs within a given cloud. Although intra-cloud

discharges are of particular concern to the aviation industry, cloud-to-ground discharges

are responsible for most lightning-related damage and injury. Hence, the study of

cloud-to-ground lightning discharges has many practical applications.

Cloud-to-ground discharges can be classified into four categories (Figure 2-2).

This classification is based on the polarity of the charge transferred to ground and the

direction of propagation of the initial leader propagation.

1. Downward negative

2. Upward negative

3. Downward positive

4. Upward positive














S2










3 4




Figure 2-2. Lightning cloud-to-ground discharges categorized by leader propagation
direction and polarity of charge transferred to ground. Adapted from
(Berger and Vogelsanger 1966).

Categories 1 and 2 effectively lower negative charge to ground while Categories 3

and 4 effectively lower positive charge to ground. Category 1 comprises about 90%

of all cloud-to-ground flashes. Category 3, while accounting for only about 10% of

cloud-to-ground discharges, is of particular interest because of the large peak currents

involved. Categories 2 and 4 (relatively rare compared to categories 1 and 3) are

most often observed on tall structures or mountain tops. A large portion of lightning

research has focused on downward-negative flashes because of their overwhelming

presence relative to the other three types of cloud-to-ground lightning. However,

many studies have been performed using tall objects (such as towers) that experience

primarily upward lightning.

A downward-negative flash begins with the initiation of a stepped leader from

the negative charge center of the thundercloud. The stepped leader serves to form a

negatively charged plasma channel from the cloud to the ground. It is thought that









this initiation is preceded by a preliminary breakdown process within the cloud. As

the name suggests, the stepped leader propagates in discrete bursts or "steps" of

extension. The stepped leader travels from cloud to ground at an average speed of

2 x 105 m s 1 with an average step length of tens of meters, and deposits negative

charge along the channel. The length of each step and the time-interval between steps

is a function of height above ground and it has been observed that both decrease as the

leader approaches ground. The stepped leader phase typically lasts for some tens of

milliseconds with each step lasting about 1 |s and with the time between steps being

tens of microseconds (Uman 1987).

As the leader approaches ground, an upward leader, having positive charge, is

initiated from the ground or other grounded objects (e.g., trees or other structures).

Probably several upward unconnected leaders are initiated from different locations. At

some tens of meters above ground, one of these upward leaders will connect with a

branch of the downward stepped leader in a relatively poorly understood phase of the

discharge known as the attachment process.

Once the two leaders have connected, a large surge of current, known as the

first return stroke, travels back up the stepped leader channel neutralizing the charge

deposited by the leader, effectively lowering negative charge to ground. The return

stroke travels at about one third to one half the speed of light with speed decreasing

with increasing height. The return stroke process can also be viewed as a potential

discontinuity traversing the channel since the region ahead of the return stroke front

(the negatively charged leader channel) is at a much higher negative potential (near

cloud potential which is negative several tens of megavolts or more) than the region

behind the front, which is near Earth potential.

Once the return stroke front has reached the cloud charge, a subsequent downward

leader may be initiated. Typically this new leader, known as a dart leader, follows

the path of the previous channel and does not exhibit stepping. In other words, a dart









leader propagates continuously, not in discrete bursts. Dart leaders travel at an average

speed of 107 m s1, or about two orders of magnitude faster than stepped leaders.

Occasionally "dart-stepped" leaders are observed, which begin as dart leaders but

exhibit stepping near ground. As the dart or dart-stepped leader approaches ground,

another upward leader is initiated. Unlike the first return stroke, the attachment process

for subsequent strokes typically occurs when the upward leader is only a few meters

in height. The connection of the two leaders results in another return stroke. The

characteristics of measured currents and fields from first and subsequent return strokes

are statistically different. This leader/return stroke process can occur many times over

the course of a flash, but 3 to 5 strokes is typical.

Downward-positive flashes, which account for roughly 10% of the total

cloud-to-ground discharges, are initiated by a leader process that is similar to that

of negative flashes. The downward leader initiated from the cloud deposits positive

charge along the channel and may or may not exhibit stepping. Positive first strokes

can exhibit a much higher peak current and charge transfer than negative first strokes,

and single-stroke flashes are much more common in positive than in negative flashes.

Upward flashes (Categories 2 and 4) are initiated in a completely different

manner than downward flashes. In an upward flash, the first leader is initiated from

the ground-based object. A current, known as the initial continuous current (ICC),

flows along the channel. Typically this is is a steady current several hundred amperes

in magnitude lasting for several hundred milliseconds and is not unlike the initial

stage current observed in classical rocket triggered lightning, which is discussed in

Section 2.1.2.1. When the upward leader reaches the cloud base, there is a brief

no-current interval followed by subsequent dart leader/return stroke sequences, similar

to subsequent strokes of downward flashes.









2.1.2 Triggered Lightning

The random nature of lightning makes it a very difficult phenomenon to study,

therefore a technique was developed in the 1960's to artificially initiate a flash "on

demand." This most widely used and successful technique involves the launching of

a small rocket trailing a thin metallic wire. Other potential techniques involving, for

example, lasers have been attempted without apparent success. The "rocket and wire"

technique can be divided into two main categories, the classical and altitude triggering

techniques.

2.1.2.1 Classical-triggered lightning

The classical rocket and wire technique is, by far, the most-used technique to

artificially initiate a lightning flash. A small rocket, typically about one meter in

length, is launched upward at an initial velocity of about 200 m s-1 with a thin-gauge

metallic wire trailing behind it. The spool of wire can either be fitted to the rocket

or the launcher itself, but in either case, one end of the wire is attached to the rocket

and the other is attached to launcher. The launcher is attached to the object to be

struck or is grounded. The wire can be made of any conductor, although steel or

Kevlar-reinforced copper have been used most. Reinforcing the copper wire with

Kevlar gives it enough strength to survive the launch without being broken. The UF

lightning research team currently uses spools of 700 m Kevlar-reinforced copper wire

mounted to the base of a 1.15 m long fiberglass rocket, as shown in Figure 2-3. Other

configurations have been used by the UF team in the past.

In order to have any chance of a successful trigger, typically three conditions have

to be satisfied. First, there must typically be a thundercloud overhead. Firing a rocket

into "blue sky" has very little chance of success, although being under the edge of the

storm is occasionally acceptable. Second, in Florida the quasi-static field at ground, as

measured with an electric field mill, must be below -5 kV m1, with a value below

-6 kV m 1 being desirable. Third, there must be lightning activity within several






























Figure 2-3. Fiberglass rocket with a spool of Kevlar-sheathed copper triggering wire
mounted to the base.

kilometers, but only a moderate amount of activity is desirable since an overabundance

of natural lightning in the area may disrupt the triggering process. All of these

conditions are empirically derived and based on years of experience but exceptions

can occur. Furthermore, each metric individually can be potentially misleading, hence

the use of all three increases the likelihood of success. This conservative approach is

necessary due to the expensive nature of the rocket and wire spool configuration. Even

when all of these conditions are satisfied, the success rate for the University of Florida

group is only about 50%.

The sequence of events observed in a typical classical rocket triggered lightning

flash is shown in Figure 2-4. The rocket and wire launch has the effect of quickly

erecting a very tall grounded structure. Assuming the above conditions are met, when

the rocket reaches about 300 m above ground, an upward leader will be initiated from

the tip of the wire due to electric field enhancement there. The polarity of this leader

will be positive if there is a negative charge center overhead, and the leader will be







11




Natural 107m -1
105 m s1 Channel





O 200ms-1
A+ ; 108ms-1
SCCopper Wire-trace
I Wire Channel

Pa 300 m

1 2 3 4 5 6

1 2 s Hundreds of ms Tens of ms


1. Ascending Rocket
2. Upward Positive Leader
3. Initial Continuous Current (ICC)
4. No-Current Interval
5. Downward Negative Leader
6. Upward Return Stroke

Figure 2-4. Sequence of events observed in a typical classical rocket triggered
lightning flash. Adapted from (Rakov et al. 1998).


attracted to the cloud charge. The movement of charge to the leader tip causes a

quasi-static current to flow along the wire which increases in amplitude as the upward

leader propagates. When a current of a few hundred amperes is flowing through

the wire, the wire explodes, briefly interrupting the current. This interruption lasts

about 10 |s and can be manifested as either a severe drop in current amplitude or a

complete cessation of the current. After the no-current interval, the current is abruptly

reestablished, often with a large pulse, a significant part of the process known as the

initial current variation (ICV). The details of the mechanism by which the current

is cut off and abruptly reestablished are poorly understood. The wire is replaced

by a plasma channel and the reestablished quasi-static current, often superimposed









with many pulses, continues to flow for several hundred milliseconds. This phase of

the discharge is known as the initial continuing current (ICC). The ICC in triggered

lightning has been observed to be similar to that observed for natural upward flashes

from tall structures. The upward leader, initial current variation, and initial continuing

current together constitute the initial stage (IS) of a triggered lightning flash. The

initial stage is often colloquially referred to as the wireburn stage, since the triggering

wire is destroyed in the process.

By the time the ICC has ended (some hundreds of milliseconds from the initiation

of the upward leader), the upward leader has long since entered the cloud. After the

cessation of the ICC and a no-current interval, a dart or dart-stepped leader, negatively

charged, may be initiated from the cloud and follow the path of the previous upward

leader and vaporized wire. As this downward leader approaches ground, a short

upward leader is initiated from the launcher and probably from other grounded objects,

as in natural lightning. The path to the launcher is at a higher temperature and hence

less dense than the surrounding virgin air and may have higher conductivity. The

downward leader attaches to the upward leader originating from the launcher and,

as with natural lightning, a return stroke is initiated. Zero or over twenty additional

strokes may follow. However, as with natural lightning, 3 to 5 are typical. Figure 2-5

shows a photograph (exposed over several seconds) of classically triggered flash with

multiple return strokes.

A considerable percentage of classically triggered flashes (40% or 16 out of 40

flashes during the 2001 and 2002 seasons at the ICLRT) contain the initial stage only

and have no return strokes. Usually, the intent of a triggered lightning experiment is to

study the return strokes and a wireburn is considered an unsuccessful attempt. There

is, however, significant scientific value in studying the initial stage. Therefore, more

appropriate terms for a wireburn are "classically triggered flash consisting of initial

stage only" or "classically triggered flash with no return strokes."





























Figure 2-5. Classically triggered flash S-0116, initiated on July 27, 2001 21:56:06
UT. The vaporized wire channel (initial stage) is on the right with the
individual return stroke channels blown to the left by the wind. The
launcher is located underground.


It should be emphasized that all strokes in classically triggered lightning are

initiated by dart or dart-stepped leaders and that the measured currents and fields

are statistically similar to subsequent strokes in natural lightning. Therefore, the first

stroke, observed in natural lightning discharges, is not present in classical triggered

lightning. It follows that, while extremely useful, the classical rocket and wire

technique cannot be used to study first strokes in natural lightning, which are clearly of

interest.

2.1.2.2 Altitude-triggered lightning

Since the classical rocket and wire technique is incapable of initiating a first

return stroke, an alternative triggering technique known as "altitude triggering" was

developed. While far less efficient and far more unpredictable than classical triggering,

the altitude technique can initiate a a stepped leader while the classical technique

cannot.










S105 106ms-1


105 m s-


a 3 s 6 ms


1. Ascending Rocket
2. Upward Positive Leader
3. Bi-Directional Leader
4. Upward Return Stroke
5. Upward Positve Leader


1 ms 10- 100 Is
I I-- ^ -----


Figure 2-6. Sequence of events observed in a typical altitude rocket triggered
lightning flash. Adapted from (Rakov et al. 1998).


The sequence of events observed in a typical altitude rocket triggered lightning

flash is shown in Figure 2-6. Essentially, the same configuration is used for the

altitude technique as for the classical technique. The only difference is that the wire

used for an altitude trigger is not grounded. In one configuration, the first 50 m of the

spool consists of Kevlar reinforced copper wire and is known as the intercepting wire.

The next 400 m consists of only Kevlar and the remainder of the spool is wire. When

the rocket is launched, the wire is unspooled and the result is a length of ungrounded

wire traveling upward. This floating wire is in the presence of a very strong electric

field, resulting in charge separation within the wire. If the rocket and wire technique









is used with negative charge centers overhead (as in Florida), a positive charge will

accumulate at the top of the wire and a negative charge will accumulate at the junction

of the wire and the Kevlar section. As in the classical technique, an upward positive

leader is initiated from the top of the wire. Several milliseconds later, a downward

negative leader is initiated from the wire/Kevlar junction. There is no plasma channel

or other conducting path to ground due to the Kevlar, so therefore the negative leader

must propagate through virgin air, and hence exhibits stepping.

Ideally, the downward negative stepped leader will attach to an upward positive

leader initiated from the intercepting wire, attached to the launcher. However, since the

stepped leader is not following a pre-conditioned path, it is essentially free to wander

and may be attracted by upward leaders from ground or other grounded structures.

Hence, the strike point is highly unpredictable. As with the classical technique, the

return stroke propagates from ground to cloud along the path of the leader. Moreover,

the return stroke travels three of four orders of magnitude faster than the upward

positive leader initiated from the top of the altitude wire and soon catches up with the

leader tip and the current wave reflects from it. Thus, return-stroke current wave is

only allowed to travel a kilometer or so, which is a fraction of the height of the cloud

charge. Therefore, the current and electromagnetic field waveforms exhibit a peculiar

shape relative to those from a classically triggered flash for which the available channel

length is of the order of several kilometers or more. The current waveforms, measured

at ground, rise sharply to peak and at first decays normally, but then decays rapidly

giving the waveforms a "stunted" appearance. In addition, it is likely that current from

the upward positive leader from the intercepting wire will not destroy it. The return

stroke, however, will destroy any remaining wire and this results in an interruption

in the current (Rakov et al. 1998). This phenomenon is manifested as a double-peak

shape in the current waveform. After these processes, an ICC follows, similar to that in

classical triggered lightning. Several dart or dart-stepped leader/return stroke sequences




























Figure 2-7. Unintentional altitude triggered flash FPL-0205, initiated on July 9, 2002
16:26:10 UT. The launcher (the intended strike point on the insulating
tower) is pictured in the foreground.

may occur after the ICC. These leader/return stroke sequences are thought to be similar

to those in classically triggered lightning and subsequent strokes in natural lightning.

The success rate of altitude triggering is about 10% for the University of

Florida group, compared to about 50% for the classical technique under the same

thunderstorm conditions. Interestingly, the triggering wire occasionally breaks during

a classical-triggering attempt resulting in an unintentional altitude trigger. In this case,

the Kevlar cable is replaced with an air gap, but the effect is similar. A photograph of

such an event is shown in Figure 2-7.

2.2 Overview of the ICLRT Facility

The International Center for Lightning Research and Testing (ICLRT) occupies

about 0.5 km2 on the National Guard Army base at Camp Blanding, about 5 km east

of Starke, Florida. The facility was constructed in 1993 and initially operated by the

Electric Power Research Institute (EPRI) and Power Technologies Inc. (PTI). In 1994









the University of Florida lightning research group took over operation of the site.

Figure 2-8 shows a sketch of the ICLRT, as of early 2003.

As currently configured, the ICLRT contains two permanent launching locations.

One launcher is located atop an 11 m high wooden tower and can hold a battery

of twelve rockets. The second launcher, although currently non-operational due to

maintenance needs, is located underground with the top of the launching tubes flush

with the ground and is capable of holding six rockets. A 70 m x 70 m metal mesh

surrounds the hole and is attached to the launcher, in order approximate an infinitely

conducting ground plane within the vicinity of the launcher. When triggering from

the tower launcher, operating personnel are housed in the Launch Control trailer,

approximately 80 m from the tower. The Launch Control trailer houses computers,

video equipment, data acquisition equipment, and the rocket launching system in a

centralized location. Pneumatic hoses run from the Launch Control trailer up to the

launcher at the top of the tower and air pressure is used to activate pneumatic relays,

located at the base of each rocket tube, which are used to fire the rockets. The use of

pneumatics serves to electrically isolate the rocket launcher from the launch control

trailer.

Experiments using the underground launcher are conducted from a trailer known

as SATTLIF, which stands for SAndia Transportable Tri.g.'ere Lightning Facility.

SATTLIF was constructed by Sandia National Laboratories in 1990 and has been used

in triggered lightning experiments at the Kennedy Space Center and Fort McClellan,

Alabama, as well as at the ICLRT. SATTLIF has been placed at several locations

around the ICLRT site. It has been at its present location since 1999. Similar to the

Launch Control trailer, SATTLIF serves as a centralized location for personnel during

rocket triggered lightning experiments and pneumatics are used to fire the rockets.

In addition, several portable launchers are available at the ICLRT. In 2002, one

of these launchers was mounted to a truck with an extendable arm used for servicing














CD


CDC
lO 100m
2 ICLRT FACILITY N
o Access Road
a [ IS3 Blast
Office IS2 Wall
SSATTLIF Underground
SATTLIF L
Launcher

IS1 Marsh
Launch
I4 Control $
1 IS4
Test [] Tower
4 Test Power Line House Launcher
0O
I I
SRunway


Military
Z Container









power lines, commonly referred to as a bucket truck. The launcher, capable of holding

six rockets, was mounted to the end of the extendable arm which could be raised to a

height of over 10 m. This mobile launcher can be relocated and reconfigured for firing

in a few hours.

Every May through September, several major-funded experiments involving

triggered lightning are conducted. Furthermore, natural lightning experiments are

conducted year-round. These experiments have involved direct strikes to a runway

lighting system, a simulated house, two test power lines, and high explosives. In

addition, several multiple-station field measuring experiments, for both natural and

triggered lightning, have been implemented at the ICLRT and are discussed in the next

section.

2.3 Multiple-Station Field Measurements of Natural and Rocket-Triggered
Lightning

The term Multiple-Station Measurements refers to a set of measurements from

two or more sensors placed in two or more locations. Typically the purpose of this

type of experiment is to examine both the measured quantities individually and as a

function of location. Furthermore, multiple-station measurements can be used to make

inferences about the sources of the quantities being measured. A two-station field

measurement system with each sensor separated by a certain distance is the simplest

implementation of such a network. Even a simple network such as this can yield

considerable information that a single sensor alone cannot.

The first multiple-station measurements of the electric fields on ground from

relatively close lightning were performed by Workman et al. (1942) and Reynolds

and Neill (1955) (Rakov and Uman 2003). A brief review of this work is presented in

(Krehbiel 1981). The main goal of these experiments was to obtain charge solutions

for intra-cloud discharges and cloud-to-ground return strokes. Numerous two and









three-station electric field measurement systems have been implemented since then,

with varying time resolution and decay time constants.

(Krehbiel 1981) examined 10 cloud-to-ground, 21 intra-cloud, and three hybrid

flashes between 1976 and 1978 at the Kennedy Space Center. The electric field was

measured at a minimum of 9 and a maximum of 11 locations over an area 20 x 25 km

in extent. Flat-plate antennas connected to charge amplifiers having decay time

constant of 10 seconds were used to sense the electric field change. In addition,

radar was used for surveillance of precipitation structure and development. The data

were digitized real-time at 16 kHz and stored on magnetic tape. Since the overall

electrostatic field change was observed, and not the fine details of the radiation fields, a

large bandwidth was not required. The analysis of these data provided information on

the location of the lightning charge and charge transfer as a function of time during the

discharges.

In 1997, a multiple-station field measuring system was fielded at the ICLRT with

the expressed purpose of measuring the distance dependence of electric and magnetic

fields from triggered lightning. The electric and magnetic field was measured at seven

locations ranging from 5 to 500 m from the rocket launcher. Fiber-optic links were

used to transmit the analog data from the antennas to a central location where they

were digitized at or above 10 MHz. A detailed description of the experiment is given

in (Crawford 1998). Data for five triggered lightning strokes were recorded and some

results of this analysis are given in (Crawford et al. 2001).

In 1998, another multiple-station experiment was fielded at the ICLRT. Unlike

the 1997 experiment, the purpose was to record the electric fields produced by first

strokes in natural cloud-to-ground lightning terminating within a kilometer or so of the

site. Ten electric-field antennas were distributed about the ICLRT site occupying an

area of about 0.5 km2. In addition, two pairs of orthogonal magnetic field sensors were

placed at opposite ends of the network. Similar to the 1997 experiment, fiber-optic







21

links and digital storage oscilloscopes were used to transmit and record the data. This

configuration is described in more detail in (Crawford et al. 2001). This network

operated through 1999 and data for over 50 return strokes were recorded within several

kilometers of the center of the network. One stroke in particular is believed to have

terminated within several tens of meters from one antenna. Some results from this

experiment are given in (Rakov et al. 2003). This experiment is the direct predecessor

to the experiment described in this thesis. The measurement locations and sensors

remain essentially the same, although the sensor electronics and data acquisition

equipment are significantly different.














CHAPTER 3
INSTRUMENTATION

This chapter discusses the instrumentation for the 2001 and 2002 Multiple

Station Experiments. Section 3.1 presents an overview of the experiments. Section 3.2

discusses the MSE control system. Section 3.3 discusses the design and implementation

of the MSE measurements, including the sensors and associated electronics. Sections

3.4, 3.5, 3.6, and 3.7 present descriptions of the fiber-optic links, digitizers, video

equipment, and timing system, respectively, used in the MSE.

As mentioned in Chapter 1, the 2001 experiment was instrumented by Mr. George

Schnetzer, formerly of Sandia National Laboratories, and Mr. Keith Rambo, of the

Department of Electrical and Computer Engineering at the University of Florida. The

2002 experiment was coordinated and instrumented by the author, with significant

assistance from the above-mentioned persons. When applicable, the distinction is made

between the 2001 and 2002 configurations.

3.1 Overview of the 2001 and 2002 Multiple Station Experiments

Figure 3-1 shows a sketch of the MSE network at the ICLRT. Notable landmarks

such as the Launch Control trailer and the office trailer are included as reference

locations. In 2001, fourteen sensors were spread about ten locations, while in 2002

twenty sensors were spread about eleven locations. In 2001, only electric fields,

magnetic fields, and the optical output from the lightning channel were measured.

During the 2002 season the network was augmented with electric and magnetic field

time-derivative measurements and current measurements.

The measurement locations are referred to as Stations; hence the term Multiple

Station Experiment. Ten stations spread about the site are numbered 1 through 10 with

each station housing one or more field or field-derivative measurements. It should be






















Office Trailer Multiple Station Experiment at ICLRT


SATTLIFE


Station 1


IS4


CD
C '
5






o
C Q



rtC


CD





CD


0
O
(1






or




C
o .





o


Underground Launcher


Station 4


Station 5


Test Power Line


Runway




Station 8


Station 10


Station 9


1< >m
100 m


last Wall


Station 6



SW Optical









noted that Stations 3 and 7 were disassembled prior to 2001 and will not be referred

to in this thesis. The numbering scheme was kept the same so that there is continuity

with previous multiple-station experiments at the ICLRT (as discussed in Section 2.3).

Each measurement is designated by the type of measurement and its location. For

example, E-1 refers to the electric field (E) measurement at Station 1 and dE-8 refers to

the electric field time derivative (dE/dt) measurement at Station 8. The magnetic field

(B) and magnetic field time-derivative (dB/dt) measurement designations contain an

extra parameter which refers to the orientation of the sensor. For example, B-4N refers

to the magnetic field measurement at Station 4 with the plane of the sensor (discussed

in Section 3.3.2.1) oriented to true north. In addition, dB-1E refers to the magnetic

field time-derivative measurement located at Station 1 with the plane of the sensor

oriented perpendicular to true north (i.e. east-west). The two optical measurements

are designated NEO and SWO, which stand for North-East Optical and .Sin,-TT\t

Optical, respectively. The two current measurements are designated I-High-Tower and

I-Low-Tower, where High and Low refer to the relative maximum amplitude measured

(I-High-Tower is capable of measuring currents about an order of magnitude higher

than I-Low-Tower).

A computer program automatically controlled the activation of the sensors as well

as the arming of the network and recording of calibration signals. The computer also

monitored the battery voltages of the individual sensors and reported the status of the

network to personnel via an on-screen display and email. The program automatically

turned on and off the network based on the output of an electric field mill, but

could be manually overridden when necessary. The analog voltage waveforms from

all sensors were transmitted over fiber-optic links and digitized on digital storage

oscilloscopes. Furthermore, the digital storage oscilloscopes were triggered by the

simultaneous output of two optical sensors viewing the network from opposite corners.

Four-station video coverage was employed to help in determining the location of the









Table 3-1. Measured parameters for the 2001 MSE.

Location Measured parameters
Station 1 E
Station 2 E
Station 4 E B
Station 5 E
Station 6 E
Station 8 E
Station 9 E B
Station 10 E
North-East Optical Optical
South-West Optical Optical


lightning and the geometry of the lightning channel. All computer, data-recording

and video-recording equipment were housed in the ICLRT Launch Control trailer. A

simplified diagram of the MSE control and data acquisition system is given in Figure

3-2.

Tables 3-1 and 3-2 list the parameters measured in the 2001 and 2002 MSE,

respectively. As discussed in Section 3.3.2, multiple B and dB/dt measurements were

present at some locations to sense different components of the horizontal field.

Table 3-3 lists the coordinates of the measurement locations. The coordinates

were measured in 1999 with a differential GPS unit and the author has not verified the

accuracy of the measurements. Furthermore, the coordinates of the South-West Optical

sensor were not measured.

3.2 Control System

Operation of numerous sensors spread about a large physical area poses several

logistical problems. These can only be overcome by a robust control system which

automates as many operations as possible. These logistical problems are outlined

below.

Since the occurrence of thunderstorms can be unpredictable, manually activation

of the network would require personnel to be on-site virtually all of the time, which


























LAUNCH CONTROL TRAILER
a---------------------------------------
Trigger to ---_
all Digitizers
*_---I^^^ ^ ^


Di
Video


Optice





)les


gital
Recorder


I


da Sensors








Video
C'nm Prn


III









Table 3-2. Measured parameters for the 2002 MSE.

Location Measured parameters
Station 1 dE/dt dB/dt
Station 2 E
Station 4 E B dE/dt dB/dt
Station 5 E
Station 6 E
Station 8 dE/dt
Station 9 E B dE/dt dB/dt
Station 10 E
North-East Optical Optical
South-West Optical Optical
Launch Tower Current

Table 3-3. Coordinates of the MSE measurement locations.
Location Coordinates
Station 1 29.94390325 N 82.03451849 W
Station 2 29.94404460 N 82.03497685 W
Station 4 29.94329562 N 82.02931170 W
Station 5 29.94323225 N 82.03248203 W
Station 6 29.94160838 N 82.03561998 W
Station 8 29.94145723 N 82.03053364 W
Station 9 29.94026821 N 82.03396553 W
Station 10 29.94068532 N 82.02925378 W
North-East Optical 29.9440 N 82.02949 W
Launch Tower 29.94262236 N 82.03185467 W


is not possible September through April, when rocket triggered lightning operations

have ceased at the ICLRT. Since the electronics associated with each sensor (e.g.

fiber-optic transmitters and amplifiers) are powered by a battery, the network cannot

be left activated at all times; otherwise the batteries would drain within a few days.

Furthermore, due the physically large size of the network and the large number of

sensors, manual activation of all of the sensors when a thunderstorm is present is

infeasible. In addition, it is infeasible to monitor battery voltages and the general

health of the network manually for extended periods of time.

Therefore, in order to operate with minimum user attendance and maintenance, the

MSE control system must fulfill the following requirements.









* The network must be able to be activated and deactivated automatically, without
any user interaction, when appropriate thunderstorm conditions are present.
Furthermore, the system must be able to automatically determine when these
conditions are present based on predetermined criteria. Activation includes
turning on all of the sensors, calibrating all of the fiber-optic links, and arming
all of the digitizers.

* The control system must be able to give instant feedback regarding the status of
the entire network, including battery voltages, calibration signals, and triggering
status. Furthermore, all of this information must be available to both on-site and
remote personnel.

* All data must be automatically recorded on non-volatile media upon a system
trigger, since it is highly likely that no personnel will be on-site during the event.

The MSE control system consists of both hardware and software components. The

primary piece of control hardware is a device known as a PIC Controller. A PIC

controller was placed in the field with each of the sensors and the control software

interacted with the PIC controllers in order to automatically activate and deactivate the

network when necessary, as well as monitored the status of the network.

3.2.1 The PIC Controller

The PIC controller is the foundation of the MSE control system. This device

is used to remotely activate each of the measurements, provide attenuation to each

of the sensors, to check the status of each measurements battery, and to calibrate the

fiber-optic link associated with each measurement. The PIC controller was designed

and developed by Michael Stapleton, a project engineer, and Keith Rambo.

The device is known as a PIC controller because it contains a PIC 16F873-207SP

microprocessor. The term PIC controller is a somewhat general term, since several

varieties of PIC controllers have been developed for different applications at the

ICLRT, such as the automation of video recording. Unless otherwise noted, the term

PIC controller refers to the model which is used to accompany each sensor within the

network.














Battery Connector




Power Connector
for Measurement
BNC Connectors Electronics






Plastic Fiber
Connectors LCD Display
Connector


Hexadecimal
Switches

Figure 3-3. The PIC controller. A) Front view. B) Side view.


Figure 3-3 shows a picture of a PIC controller. Each PIC controller has

two female BNC connectors, a four-pin male microphone connector, an Agilent

HFBR-1523 fiber-optic transmitter, an Agilent HFBR-2523 fiber-optic receiver, a DB-9

female serial connector, and a two-wire power connector.

Each PIC controller is assigned a one-byte (8-bit) hexadecimal address, which is

set by adjusting a pair of hexadecimal switches located next to the DB-9 connector.

Each switch is capable of being set from OxO (0 decimal) to OxF (15 decimal). Hence

the range of addresses ranges from Ox00 (0 decimal) to OxFF (255 decimal), with

certain addresses reserved for special functions. Furthermore, each PIC controller is

assigned a work group, which is programmed directly into the microprocessor itself

and can only be changed by reprogramming the chip. Currently, only two work groups










In-line Terminator

SFiber-optic Data Fiber
STransmitter
PIC Controller Battery Supplies
S Power to the
Coaxial Supplies Power to PIC Controller
Cable the Fiber-optic
STransmitter ,is cro
Sour+
SPIC Controller 12 V Battery
Sensor C IN



Bulkhead
Feed-through
BNC Connector

Metal Enclosure
Plastic Control Fiber

Figure 3-4. Diagram of how a PIC controller is installed with a measurement.


exist, RTL (standing for Rocket Tri-.1eYIred Lightning) and CAM (standing for CAMera),

with the PIC controllers in the RTL work group being used to control the sensors and

the ones in the CAM work group being used to control cameras. Currently, no CAM

work group PIC controllers are used in the MSE. The combination of a PIC controller's

assigned work group and hexadecimal address gives each PIC controller a unique

identifier.

Figure 3-4 shows a diagram of how a PIC controller is installed with a

measurement. The output of the sensor is connected to the IN BNC connector,

while the OUT BNC connector is connected to the input of the fiber-optic transmitter,

terminated in 50 Q. The power connector is connected to a 12 V battery, while the

power input of the fiber-optic transmitter is connected to the microphone connector.

Pins 1 and 2 (ground) and 3 and 4 (+12 V) of the microphone connector are soldered

together, effectively making it a 2-pin connector. The female DB-9 connector is used

to connect a two-line LCD display that can be used to monitor the status of the PIC

controller in the field.









The PIC controller is essentially a combination of relays and attenuators controlled

by a microprocessor. Therefore the PIC controller can be described as a series of

programmable switches and attenuators. How the PIC controller behaves is determined

by the commands sent to it. The PIC controller communication scheme and command

set is described in the following sections. The PIC controller itself is powered by a

12 V battery. However a relay inside is used to supply power to other electronics,

which at a minimum includes a fiber-optic transmitter, via a cable and the microphone

connector. Hence, battery life can be conserved by simply disconnecting power from

the electronics. However, the PIC controller must always be powered and draws a

current of about a few tens of milliamperes at 12 V. When powered by a 12 V, 24 Ah,

battery, the maximum battery life is several weeks.

The PIC controller is placed in series with the measurement between the sensor

and the fiber-optic transmitter, via the IN and OUT BNC connectors and short lengths

50 Q coaxial cable. The function of the PIC controller depends on what command is

sent to it. First, the PIC controller can act as a 50 Q in-line attenuator, which reduces

the output voltage of the sensor, increasing the full-scale range of the measurement.

If no attenuation is set, the PIC controller has a gain of 1 (0 dB) and does not affect

the measurement. The attenuators are resistive PI attenuators of values -3 dB, -6 dB,

-10 dB, -14 dB, and -20 dB, which can be added in any combination by sending the

appropriate commands to the PIC controller. The PIC controller adds the attenuators by

switching the appropriate relays inside of the device. The attenuators are designed to

be terminated in 50 Q and will not provide the stated voltage division if the output of

the PIC controller is not terminated in 50 Q. The PIC controller has an input resistance

of 50 Q when the output of the PIC controller is terminated in 50 Q, regardless of

the attenuation setting. If the PIC controller is terminated in a different resistance,

the input resistance of the PIC controller will be that resistance only if no attenuation

is used. While it is possible to use the attenuators when the PIC controller is not









terminated in 50 Q by calculating the actual voltage ratio based on the attenuator

circuit and a different load, this has never been done and is not recommended due to

the high possibility of confusion. The ability to remotely set the amount of attenuation

used in a measurement eliminates the need for personnel to manually exchange BNC

in-line attenuators every time an adjustment is required.

In addition, the PIC controller can act as a waveform generator that calibrates the

fiber-optic link. When the appropriate command is sent to the PIC controller, a relay

inside the device disconnects the sensor from the fiber-optic transmitter and injects a

calibration waveform into the transmitter. The calibration waveform is a 100 Hz square

wave with a selectable peak-to-peak voltage of 1 V or 0.1 V, when the PIC controller

is terminated in 50 Q. When terminated in high impedance, the voltage doubles to 2 V

and 0.2 V, respectively. The calibration waveform can be used in either mode as long

as the operator is aware of the effect of the different terminations. In addition, it is

possible to also attenuate the calibration signal using the attenuation function, although

this has never been done. The calibration waveform is useful for estimating the gain

of the fiber-optic link, assessing the amount of non-linear distortion present in the link,

and determining whether the link is damaged. This eliminates the need for personnel

to manually inject a calibration waveform into the link, a process that would be tedious

and time-consuming in a large experiment such as the MSE.

The Agilent HFBR-1523/2523 fiber-optic transceiver pair (660 nm LED) is used

to communicate with the PIC controller, via 1 mm diameter optical plastic fiber with

snap-action connectors, when it is installed in the field. The maximum length of fiber

that can be used, without the addition of an optical repeater, is approximately 110 m at

a maximum data rate of 40 kbps. The fiber-optic transmitter converts serial TTL (0 to

+5 V) logic to pulses of light, while the receiver does the reverse conversion. No light

corresponds to logical 0 while a pulse of light corresponds to logical 1. A standard

PC can communicate with a PIC controller via its RS232 serial port and a fiber-optic









transceiver pair by utilizing a RS-232 driver/receiver, such as a Maxim MAX232 chip,

which converts RS-232 serial logic (12 V) to TTL serial logic. A PIC controller

cannot send and receive data simultaneously, and therefore the communication link is

half-duplex.

The method by which the control PC communicated with the PIC controllers

differed between the 2001 and 2002 seasons. In 2001, the RS232 serial port of the PC

was interfaced with a MAX232 chip to translate the RS232 serial logic to TTL serial

logic, as described above. The output of the MAX232 chip was connected directly to

an Agilent HFBR-1523/2523 fiber-optic transceiver pair mounted on the same circuit

board. One end of a plastic fiber was connected to the transmitter and the other was

connected to a repeater, whose output branched out to three additional plastic fibers.

Each of these fibers went into the field, forming three loops which each eventually

returned to the Launch Control trailer. At each measurement location, an incoming

plastic fiber would be connected to the fiber-optic receiver of the PIC controller,

and another plastic fiber going to the next measurement would be connected to the

transmitter. In this topology, the PIC controllers also acted as fiber-optic repeaters.

However, in cases where the distance between successive measurements in the loop

was greater than about 100 m, additional repeaters were required. The plastic fibers

connected to the transmitter of the last PIC in each loop went back to the Launch

Control trailer where they were all connected to another repeater. The output of this

last repeater was connected to the fiber-optic receiver of the MAX232 board which was

then connected to the control PC via the RS232 link. A diagram of this topology is

shown in Figure 3-5. This method of communication posed several problems, which

are outlined below.


S If any PIC controller in a given loop malfunctioned, every measurement in the
loop was lost since each PIC controller acted as a repeater.










Loop 2


PIC


Repeater
Loop 1 Loo

PICNIC Loop 3
// IC Repeater



MAX232 oMA /2

Connection



Plastic Fiber



Control PC

Figure 3-5. Diagram of the PIC controller communication topology used during the
2001 MSE.


S If any plastic fiber in a given loop was damaged, every measurement in the loop
was lost since the series data path was opened.

Furthermore, these two issues were especially troublesome since finding the location of

a malfunctioning PIC controller or a damaged fiber was a difficult and tedious process.

During the 2002 season these limitations were addressed by replacing the

fiber loops and repeaters with wireless 900 MHz RF links. These wireless RF

units are known as PIC RF units, since they contain the same PIC 16F873-207SP

microprocessor as the PIC controllers. The RF transceiver itself is a MaxStream

9XStream-96 900 MHz transceiver (902 928 MHz, unlicensed ISM band) capable

of a maximum data rate of 9600 bps. Instead of running lengths of plastic fiber to

each PIC controller in the field, PIC RF units were placed about ten meters away

from each measurement location. Each PIC RF unit was equipped with an Agilent

HFBR-1523/2523 fiber-optic transceiver pair and two short lengths of plastic fiber















1:a]les

P II
PIC
,C RE

PIC R


P PIC
RF rPIC
PIe C Antenna
Wireless
Transceiver
RS232
Connection
Plastic Fiber







Control PC

Figure 3-6. Diagram of the PIC controller communication topology used during the
2002 MSE.


connect it to a PIC controller. An RF transceiver was installed in the Launch Control

trailer, which was interfaced to the control PC via the RS232 serial port. The RF

transceiver was capable of directly encoding and transmitting RS232 serial logic, so no

MAX232 chip was required. The RF transceiver transmitted the commands from the

control PC and the PIC RF unit at each measurement location received and decoded

the signal. The decoded data are then transmitted to the PIC controller over one of

the lengths of plastic fiber. Similarly, the PIC controller transmitted data back to the

PIC RF unit over the other length of plastic fiber. The PIC RF unit then encoded and

transmitted this data back to the Launch Control trailer where the transceiver decoded

the data and transmitted it to the control PC over the RS232 interface. Hence, the

PIC controller communication scheme changed from a ring topology to a star-network

topology, as shown in Figure 3-6, with the PIC RF units essentially acting as wireless

repeaters.









Multiple PIC controllers in an area could be interfaced with a single RF PIC unit

by using an optical repeater. A special repeater board was designed for this purpose

and can interface nine PIC controllers to a single PIC RF unit. The star-network

topology was a significant improvement over the ring topology used in 2001. Now, if

a single PIC controller or plastic fiber fails, only the measurement connected to that

PIC controller fails, instead of the entire loop, as before. This configuration made

diagnosing system component problems much easier. In addition, since the lengths of

plastic fiber were, at most, some tens of meters, the amount of plastic fiber that could

be damaged was minimal. In this topology the PIC controllers themselves no longer

act as repeaters. Therefore, a simple hardware modification was required to remove

this function. If this hardware modification were not performed, a PIC controller would

immediately repeat any data sent to it back to the PIC RF unit that would then transmit

it. If another PIC controller without the hardware modification were present, it would

receive this data and repeat it again. Hence, this repeating could become an infinite

loop which making communication with the PIC controllers impossible since some PIC

RF unit would always trying to transmit, which essentially jams the transmitter at the

control PC.

The PIC RF units are electrically isolated from the lightning measurements to

eliminate any possibility of data contamination or equipment damage from lightning

electromagnetic interference (EMI). The PIC RF units were placed in separate metal

enclosures (modified ammunition containers) with the only physical connection

between the PIC RF unit and the measurements being the pairs of plastic fibers. Each

metal enclosure contained a PIC RF unit, a 12 V battery, and optionally a repeater

fan-out board for connecting multiple PIC controllers. The PIC RF unit itself was

mounted to the inside cover of the box with a 15 cm antenna mounted to the outside

of the cover. A 900 MHz tuning stub (shorted length of transmission line) was

placed between the antenna output and the ground of the wireless PIC controller. At









900 MHz, the tuning stub resonates and becomes a very high impedance but, at other

frequencies, it looks like a short circuit to ground. Since the significant EM radiation

from close lightning occurs from DC to 10 MHz (Uman 1987), any induced current in

the antenna from lightning should be directed to the ground of the PIC RF unit and not

interfere with its operation.

The PIC RF units will operate for a few days when powered with a 7 Ah, 12 V

battery. Five-watt solar cells (450 mA short-circuit current) were installed at each PIC

RF unit location in order to keep the batteries charged. The solar panels were mounted

on top of a one-meter long 4x4 piece of wood extending vertically from the ground.

The PIC RF unit enclosure was hung from a pair of half-inch bolts screwed into the

wood. The solar cell served to shade the enclosure in this configuration. The solar cell

itself was encased in a metal "chicken wire" mesh with a hole-diameter of about five

centimeters to protect it from the induced effects of nearby lightning. A test cell was

placed in the field in early June 2002 without this protection, and it is believed that

the electromagnetic signal from a close lightning strike damaged the cell, rendering

it useless. A two-wire power cable was run from the solar cell into the PIC RF unit

enclosure and connected to the 12 V battery. This cable was encased in metal shield

braid with one end soldered to the chicken wire and the other end securely attached

to the hole in the metal enclosure with a hose clamp. The plastic fibers were run out

of two additional holes in the enclosure. These holes were fitted with PVC elbow

connectors with the openings pointed toward the ground so that no water could enter

the enclosure. Figure 3-7 shows a PIC RF unit enclosure mounted with a solar cell.

All command data packets sent to a PIC controller are formatted into a five-byte

(40-bit) array. Each byte in the array can be represented as a character, as defined by

the American Standard Code for Information Interchange (ASCII). Hence, each data

packet can be represented as an ASCII character string, which is easy for humans to

interpret. Table 3-4 describes the format of a PIC controller command data packet.









Solar Cell Enclosed in "Chicken Wire" Mesh
I


Antenna

Metal Box
Containing -
PIC RF Unit


V
Plastic Fiber

Figure 3-7. PIC RF unit enclosure mounted with a solar cell.


Table 3-4. Format of a PIC controller command data packet.

Bytes 1-3 (bits 0-23) Byte 4 (bits 24-31) Byte 5 (bits 32-39)
Work group identification Address Command


Communication with a PIC controller is implemented in a command/response

format. What this means is that when a command is issued to a PIC controller, the

PIC controller performs the command, and then the PIC controller sends a response.

The response the PIC controller sends is the last command data packet that it received,

which is stored in a buffer, along with five additional bytes corresponding to its battery

voltage and two bytes corresponding to its temperature. Table 3-5 describes the format

of a PIC controller response data packet.


Table 3-5. Format of a PIC controller response data packet

Bytes 1-3 Byte 4 Byte 5 Bytes 6-10
(bits 0-23) (bits 24-31) (bits 32-39) (bits 40-79)
Work group Hexadecimal Last Battery
identification address command voltage


Bytes 9-10
(bits 80-95)
Temperature









Bytes 1-3 (bits 0-23) are the work group identification, such as RTL or CAM as

discussed above. These three bytes are interpreted as ASCII characters so that, for

example, the byte array 0x52 0x54 Ox4C translates into the character string RTL. A

PIC controller will not respond to a command unless its work group identification

matches the work group identification of the command. For example, PIC controller

OxOE in the RTL work group will not acknowledge a command designated for PIC

OxOE in the CAM work group. The address of the PIC controller is given in byte four

(bits 24-31). Therefore, each PIC controller can be communicated with individually.

Address OxFF is a global address, meaning that every PIC will perform a command

issued to address OxFF, regardless of its address setting. The work group constraint

still applies, which means that OxFF is only global within the scope of the work group.

A PIC controller in the CAM work group will not respond to the OxFF address of

the RTL work group and vice versa. This is useful for sending a command to every

PIC controller simultaneously, where sending the command to each PIC controller

individually would be time consuming. The drawback of the global address is that

none of the PIC controllers will send a response after performing their assigned

command because there is no way the control PC issuing the command could

understand multiple simultaneous responses. Hence, there is no confirmation that

the command was received.

Byte five (bits 32-39) designates the command issued to the PIC controller. Each

bit in character five designates a certain function of the PIC, as shown in Table 3-6.

The PIC controller can be told to perform multiple functions by setting specific

bits in the command byte. For example, the simplest command is the byte 00000000

(0x00), corresponding to ASCII character NULL, which turns off power to the

electronics, turns off all attenuators, and turns off calibration signals. This is essentially

a reset command. Moreover, the byte 10100000 (OxAO), corresponding to ASCII

character d tells the PIC controller to provide power to the electronics and also









Table 3-6. Bit settings of PIC controller commands.

Bit Name Function
0 3 dB Provide -3 dB of attenuation
1 6 dB Provide -6 dB of attenuation
2 10 dB Provide -10 dB of attenuation
3 14 dB Provide -14 dB of attenuation
4 20 dB Provide -20 dB of attenuation
5 Cal ON Turn on calibration signal
6 Cal Choose between 1.0 (0) or
Voltage 0.1 V (1) calibration signal
7 Power Provide power to electronics


provide a 1 V calibration signal. This calibration signal could be attenuated by -13

dB, for example, by setting bits 0 and 2, which would change the command byte to

10100101 (0xA5) corresponding to ASCII character N. Some command combinations

are essentially meaningless, such as 00010000 (0x10) corresponding to ASCII character

Data link escape, which tells the PIC controller to provide -20 dB of attenuation while

not supplying power to the electronics. The byte 01111110 (0x7E), corresponding

to ASCII character -, performs a special function. While the literal interpretation of

the bits is meaningless (no power to electronics, 0.1 V calibration signal, -50 dB of

attenuation), the PIC controller interprets this command to be send a status response

without performing any function. Hence, the current status of the PIC controller,

including last command issued, battery voltage, and temperature can be assessed

without changing the current status. This command can also be used to check the

status of any wireless PIC controller.

The five bytes corresponding to the battery voltage are interpreted as ASCII

characters. For example, the five-byte array 0x31, 0x32, 0x33, 0x34, 0x35 would be

interpreted as ASCII characters 1, 2, 3, 4, and 5 which indicates a battery voltage

of 12.345 V. Likewise, the two bytes corresponding to the ambient temperature are

interpreted as ASCII characters. For example, the two bytes 0x33 and 0x34 would be

interpreted as ASCII characters 3 and 4, which indicates a temperature of 34 C. The









PIC response is used to verify that the correct PIC controller indeed received the proper

command, as well as to assess the amount of battery life remaining and its temperature.

3.2.2 Software

The control software is the brain of the MSE and is responsible for the automation

of the MSE. Using a sophisticated control algorithm, the software automatically

activates the sensors and arms the waveform digitizers when appropriate thunderstorm

conditions are present. Furthermore, when the storm has passed, the software

deactivates the sensors and disarms the digitizers. In addition, the software provides

instant feedback to the user regarding the status of the network. Finally, the software

calibrates the fiber-optic data links each time the network is activated and deactivated.

The first version of the MSE control software, used during the 2001 season,

is known as LCAUTO (standing for Launch Control AUTOmation) and was written

by Alonso Guarisma, at the time a graduate student in the Department of Electrical

and Computer Engineering at the University of Florida LCAUTO was written in

C++ with the graphical user interface (GUI) written in Perl TK. LCAUTO ran on a

Hewlett Packard Pavilion 6746C PC with a 733 MHz Intel Celeron processor, 192 MB

of memory, and 30 GB of hard disk space running Red Hat Linux 6.2 with the 2.2

kernel. The user interface allowed personnel to manually query and send commands

to individual PIC controllers. For example, the software could be used to command a

specific PIC controller to supply a calibration signal so that the gain of that fiber-optic

link could be checked. In addition, a user could activate or deactivate all of the MSE

PIC controllers with the click of a single button.

The most important feature of the LCAUTO software was the ability to automatically

activate and deactivate the network when appropriate thunderstorm conditions were

present. The thunderstorm conditions were assessed by digitizing the output of an

electric field mill (a device to measure the ambient vertical electric field at ground

with a bandwidth from DC to a few hundred hertz). An Advantec data acquisition









card, housed in a separate PC running Windows 95 digitized the output of field mill at

100 Hz. Software written in GENIE (a scripting language by Advantec that is similar

to National Instruments LabView) continuously monitored the digitized field mill data

and when the absolute magnitude of the vertical quasi-static electric field at ground

exceeded 2 kV m1, a true flag was written to a specific file on the LCAUTO PC. The

LCAUTO software continuously monitored this file and when the true flag was found,

all of the MSE measurements were activated. The list of MSE measurements was

kept in a MySQL database, also written by Mr. Guarisima. Each entry in the database

was composed of a measurement name, a PIC address, and an attenuation setting.

The LCAUTO software interfaced with the database to obtain the proper settings for

each of the MSE measurements. When the magnitude of the field mill output fell

below 2 kV m 1 for a period of 10 minutes, a false flag was written to the file and the

LCAUTO software deactivated all of the measurements by sending the reset command

to all of the PIC controllers in the database.

The LCAUTO software was not capable of communicating with the Yokogawa

DL716 waveform digitizer (discussed in Section 3.5.1); hence it was the operator's

responsibility to make sure the Yokogawa was properly set up and armed at all times.

This digitizer was capable of rearming itself after a trigger, so this did not present

a problem unless it was accidentally disarmed. In addition, the LCAUTO software

was not capable of automatically acquiring calibration signals. Although the software

could command the PIC controllers to send calibration waveforms, it could not tell

the digitizer to acquire the waveforms. Hence, it was the operators responsibility to

make sure calibration signals were acquired, which was especially important after a

trigger since it is crucial to know the gain of the fiber-optic links (which can vary as

a function of temperature depending on the type of link) as close to the time of the

trigger as possible. Obviously, this would not be possible if the trigger occurred when

there were no personnel on site. In this case, the gain of the fiber-optic link would









have to be assumed to be some value, which is usually one. Depending on the type of

fiber-optic link used, this could introduce anywhere from a few percent to over twenty

percent error in the calibration, since the calibration is partially determined by the gain

of the fiber-optic link.

Also, the LCAUTO software was capable of sending email at regular intervals,

such as every 24 hours, that would give the status of all of the measurement batteries

in the network. Therefore, a remote user could easily be informed of the status of the

network.

During the summer of 2002, a new version of the MSE control software

was developed, that is known as Mercury. This software was written in National

Instruments LabView software by Robert Olsen III, presently a graduate student, and

Alonso Guarisma. The Mercury software ran on a two-processor 1.6 GHz PC with

1.5 GB of memory and 75 GB of hard disk space running Red Hat Linux 7.3 with the

2.4.18 kernel.

The new Mercury software implemented all of the features of the LCAUTO

software, such as a user interface where the status of measurements could be assessed

individually or in a group, along with some additional features. The Mercury software

was capable of interfacing with the digitizers over Ethernet and IEEE 488.2 (GPIB)

interfaces, and therefore the digitizers could be automatically set up, armed, and

disarmed along with the PIC controllers in the experiment. Furthermore, calibration

signals could be automatically acquired, which minimized the time between a system

trigger and the acquisition of calibration waveforms. Unlike the 2001 season, the data

acquisition card was housed in the same PC as the control software. Additionally, a

National Instruments data acquisition card replaced the Advantec card used in 2001.

Drivers provided by the Comedi Project were used to communicate with the card in

LabView for Linux.









The Mercury software could send email regarding the status of the network, as

the LCAUTO software did. However, the Mercury software also included a list of

times when the network was activated and deactivated in the email, although it did

not include the times of system triggers. In addition, the settings for the MSE PIC

controllers were no longer stored in a database, but in a set of ASCII text files.

The conditions Mercury used to decide whether to activate and deactivate the

network were similar to that of the LCAUTO software, although they were refined

somewhat to minimize the amount of time the network was activated unnecessarily.

The digitized output of the field mill, sampled at 100 Hz, was once again used to

determine whether appropriate thunderstorm conditions were present. These conditions

evolved over the course of the 2002 season and this evolution is outlined below.


* The network was armed if the magnitude of the output of the electric field mill
exceeded 2 kV m1. The output of the field mill was checked every 10 minutes
and the magnitude of the output ever fell below 2 kV m-1, the network was
disarmed. Unfortunately, a momentary glitch in the output of the field mill or
input of the data acquisition card would cause the system to arm unnecessarily.

* The network was armed if the magnitude of the output of the electric field mill
exceeded 2 kV m 1 continuously for two seconds. Again, the output of the field
mill was checked every 10 minutes and the magnitude of the output ever fell
below 2 kV m-1, the network was disarmed. This eliminated the problems due
to glitches. However, the network tended to remain armed for extended periods
of time (sometimes hours) during the dissipating phase of thunderstorms, where
the quasi-static electric field at ground lingered at +2 to 4 kV m1. Although it
was desirable to acquire data from positive lightning, the chance of a positive
lightning occurring was believed to be slim in such low fields.

* The network was armed if the output of the electric field mill exceeded
+4 kV m-1 or fell below -2 kV m 1 continuously for two seconds. After
arming, the output of the field mill was continuously checked and if it ever went
between -2 kV m1 and +4 kV m-, a 10 minute countdown started. If, at any
point during the 10 minute countdown, the output fell below -2 kV m or rose
above +4 kV m1, the 10 minute countdown was reset. If, after 10 minutes, the
output never fell below -2 kV m 1 or rose above +4 kV m-1, the network was
disarmed.









* The network was armed if the output of the electric field mill exceeded
+4 kV m 1 or fell below -2 kV m-1. Instead of requiring the field mill output
to exceed the threshold continuously for two seconds, short-term averaging
was used to eliminate random glitches. After arming, the output of the field
mill was continuously checked and if it ever went between -2 kV m 1 and
+4 kV m1, a 10 minute countdown started. If, at any point during the 10
minute countdown, the output fell below -2 kV m1 or rose above +4 kV m1,
the 10 minute countdown was reset. If, after 10 minutes, the output never fell
below -2 kV m1 or rose above +4 kV m1, the network was disarmed. This
algorithm is shown in a flowchart in Figure 3-8.

In addition, a calibration waveform was acquired before each time the network

was activated and after each time the network was deactivated.

3.2.3 Triggering System

The MSE was intended to acquire data on close natural lightning, specifically,

within the 0.5 km2 ICLRT site. Therefore, a triggering system needed to be designed

such that data were recorded when lightning occurred within the network, but not when

lightning occurred outside of the network. This requirement is not trivial and poses

several logistical problems.

Typically, in triggered lightning experiments, the location of the channel is known

exactly, and the trigger threshold of the digitizers can be set based upon the distance

to the channel of the sensor which triggers the digitizer and known statistics of that

particular physical phenomenon. For example, if the digitizers are to be triggered

from an electric field sensor 30 m from the lightning channel, an appropriate trigger

threshold can be calculated based upon statistics of known triggered-lightning electric

fields at 30 m. Furthermore, this technique can also be employed if data on distant

natural lightning are to be recorded. For example, if data on natural lightning in a

storm 10 km away are to be recorded, the trigger threshold can be adjusted based on

statistics on natural lightning at 10 km. Although no one flash will be exactly 10 km

away from the sensor, the average lightning distance will be and this technique will

work reasonably well, but not as well as for triggered lightning.























































Figure 3-8.


Flowchart representation of the final 2002 MSE software control
algorithm.









Unfortunately, it is difficult to employ this technique to examine close natural

lightning since two variables exist, distance and amplitude. The distance to the

lightning channel from any one sensor is unknown and cannot be assumed to be any

distance except less than about 1 km. Furthermore, (Berger et al. 1975) have measured

a range of peak currents from below 2 kA to above 80 kA for negative cloud-to-ground

first strokes and from below 4.6 kA to above 250 kA for positive cloud-to-ground first

strokes. Therefore, with such a large range of peak currents, a close lightning with

relatively low peak current within the network cannot be distinguished from a lightning

outside of the network with relatively high peak current based upon the amplitudes of

the electric and magnetic fields alone since the field amplitudes are a function of the

peak current. In addition, the distance dependence of the electromagnetic fields from

first strokes in natural lightning is not known.

Since the magnitudes of the electric and magnetic fields from natural lightning

are a function of both distance and current amplitude and wave-shape, a triggering

technique was required that can decouple the two dependencies. While a single

electric or magnetic field antenna is incapable of this, an acceptable solution can be

obtained by utilizing two or more directional sensors. A directional sensor is one

whose sensitivity is a function of the azimuthal or elevation angle of the sensor relative

to the source, such as a magnetic field antenna made of a loop of coaxial cable. The

magnitude of the output of the antenna is not only a function of the magnitude of

the magnetic field, but a function of the angle of the magnetic flux to the plane of

the loop as well. Hence the output of a loop antenna is a function of the angle of

the lightning channel to the plane of the loop. Although the output of a loop antenna

is a function of direction, the direction information (angle to the lightning channel

with respect to the plane of the loop) cannot be extracted from the output of a single

antenna. In addition, there is a 180 ambiguity in the output of the antenna since

the polarity of the magnetic field produced by a negative current is the same as









that produced by a positive current 180o away. Adding a second antenna, oriented

perpendicularly to the first, will form a crossed-loop antenna pair which can provide

directional information. Conversely, a flat-plate electric field antenna laying flush with

the ground is non-directional, having rotational symmetry with respect to the normal

of the plate, and does not provide any direction information although it can be used to

provide information needed to resolve the 180 ambiguity in the crossed-loop antenna

output. Detailed descriptions of the electric and magnetic field sensors are given

in Sections 3.3.1 and 3.3.2, respectively. If two or more crossed-loop antennas are

distributed about the network, an algorithm could be developed to determine whether

a lightning is within the network based on the output of these antennas. This type of

triggering scheme was employed in the previous ICLRT multiple station experiment

discussed in (Crawford et al. 2001) and (Rakov et al. 2003). However, this triggering

scheme requires considerable resources since two antennas and two fiber-optic links

are required for each crossed loop pair. A minimum of four loop antennas and four

fiber-optic links would be required to implement this triggering scheme. In addition,

this triggering scheme could be fooled by horizontal channel sections or slanted

channels.

In view of the above, in 2001 a triggering scheme was employed using two

optical sensors. The principle behind this scheme and the crossed magnetic field

antennas is essentially the same; multiple directional measurements are used to

differentiate between lightning within and outside of the network. The advantage of

using the optical sensors, however, is that only a single sensor (and hence only a single

fiber-optic link) is required for each directional measurement. The optical sensors are

directional because the light source must be in front of the lens of the detector for there

to be any output from the sensor. Furthermore, the implementation of a simple optical

detector is considerably less complicated than that of a crossed loop pair of magnetic

field antennas. A detailed description of the optical sensors is given in Section 3.3.3.









One of the optical detectors was placed at the north-east corner of the site, upon

a twelve foot high structure known as the Blast Wall, facing south-west. The second

optical detector was placed at the south-west corner of the site, upon a six foot high

structure known as the Military Container, facing north-east.

The output of each optical systems fiber-optic receiver was then connected into

an AND triggering circuit, designed by George Schnetzer during the summer of 2001.

Figure 3-9 shows a schematic of the triggering circuit.

Each input is fed into a non-inverting buffer circuit. The output of each buffer

is fed into a bistable (Schmitt trigger) circuit. If the voltage present at the input of

the bistable circuit exceeds the positive reference voltage or falls below the negative

reference voltage, which are defined by zener diode/resistor circuits enclosed in the

dashed boxes, the output of the bistable circuit is driven to one of the supply rails.

The reference voltages are adjusted by varying the potentiometer associated with each

reference voltage circuit. The output of each bistable circuit is then converted to TTL

logic by a pair of op-amps and fed into a 7400 NAND gate. The output of this NAND

gate provides the trigger to all of the digitizers in the experiment. Hence, the digitizers

only trigger when the outputs of both optical sensors exceed a certain threshold. The

two sensors view the field from opposite corners of the network; therefore the lightning

must be within the network in order for both sensors to see the lightning and to cause

the system to trigger. Moreover, as noted in Section 3.3.3, the elevation view of the

sensors was limited to keep the system from triggering on a bright cloud discharge.

3.3 Measurement Implementation

In this section, the theory, design, fabrication, and implementation of all MSE

measurements is discussed.

3.3.1 Electric Field and Electric Field Time-Derivative Measurements

In 2001, eight electric field measurements were fielded, located at Stations 1, 2,

4, 5, 6, 8, 9, and 10. In 2002, electric field measurements at Stations 1 and 8 were





























TRIGGER
OUT
10 kQ


Figure 3-9. Schematic of the MSE trigger circuit. Enclosed in the dashed lines are
the reference voltage circuits used in the bistable circuits.









Dnorm = Ps







G =oo

Figure 3-10. Illustration of the electric field boundary condition at a perfectly
conducting surface.


converted to electric field time-derivative (dE/dt) measurements and two additional

dE/dt measurements were installed at Stations 4 and 9. No dE/dt measurements were

made during the 2001 season.

Conducting circular flat-plate antennas were used to sense both the electric field

and its time-derivative. These types of antennas have been used for many natural and

triggered lightning experiments, such as those described in (Krider et al. 1977), (Rakov

et al. 1998), and (Uman et al. 2002). Furthermore, they are generally considered

standard equipment within the lightning research community.

There are many types of electric field and electric field time-derivative sensors.

However, when installed flush with the ground, the flat-plate sensor has the advantage

that it theoretically introduces no field enhancement. This is not the case, for example,

with a whip antenna or an elevated plate antenna. Furthermore, the flat-plate sensor can

be implemented with entirely passive components and the equivalent circuit analysis is

typically straightforward (but not trivial).

3.3.1.1 Analysis of a conducting flat-plate antenna

The Thevenin or Norton equivalent circuit for a flat-plate antenna can be derived

by considering the boundary on a perfectly conducting (t = c ) surface, as shown in

Figure 3-10.

The boundary condition is


D-. = ps


(3-1)









The quantity D is the electric displacement vector at the surface and is expressed

in units of C m 2, n is the unit vector normal to the surface, and ps is the surface

charge density and is also expressed in units of C m2. Hence, D-. is the component

of the electric displacement which is normal to the surface.


Dnorm = Ps (3-2)


Note that since the surface is considered a perfect conductor, the electric

displacement inside of the conductor and the tangential component of the electric

displacement along the surface of the conductor are both zero. If the medium above

the plate (air) is linear, isotropic, homogeneous, and non-conducting, then


Dnorm = E,,norm (3-3)


The quantity Enom is the magnitude of the component of the electric field

which is normal to the surface of the plate (expressed in units of V m 1) and e is the

permittivity of the dielectric medium, which is essentially Eo, the permittivity of free

space (8.85 x 10 12 F m 1), for air. Therefore, the expression for ps can be written as


Ps = EoEnorm (3-4)

The total charge on the surface of a conducting plate can be found by integrating

the surface charge density over the area of the plate.


Qplate= PsdA (3-5)

The quantity dA is the differential area on the surface of the plate. If the surface

charge density is uniform (which will be the case if the smallest wavelength comprising

Enorm is much greater than the plate diameter), then the total charge can be found by

multiplying the surface charge density by the area of the plate. The boundary condition

specifies that if the magnitude of the normal component of the electric field is uniform









over the surface of the plate, then surface charge density along the plate must also be

uniform. For a circular plate, the electric field along the surface of the plate can be

considered non-uniform when the plate diameter is larger than about a sixteenth of a

wavelength. If the highest frequency component of Enorm is 30 MHz then the smallest

wavelength will necessarily be 10 m. One sixteenth of a wavelength is 0.625 m, which

is larger than the diameter of the plates used in the MSE (discussed in Section 3.3.1.2).

If the electric field is uniform across a plate of area Aplate, then the total charge on

the plate can be expressed as


Qplate = coAplateEnorm (3-6)

The Norton equivalent short circuit current, i(t), is necessarily the time-derivative

of the charge.

d d dE, (t)
i(t) = d Qplate() = d (EoAplateEnorm(t)) = oAplate dEnt (3-7)

Hence, the flat-plate antenna in the presence of a uniform time-varying electric

field can be viewed as a current source whose magnitude is proportional to the time

derivative of the normal component of the electric field. This Norton equivalent current

source is the basis of the equivalent circuit. The Thevenin equivalent voltage, which

yields the same results, could also be used by performing a simple transformation. The

equivalent circuit analysis is most often performed in the frequency domain. However,

a solution can be found directly in the time domain by solving a first-order differential

equation. The relationship between the time domain and the frequency domain is given

by the Fourier transform. The Fourier transform, X(o), of a time-domain signal, x(t),

is

F {x(t)} = X(o) = x(t)e-jtdt (3-8)

Differentiation with respect to time in the time domain corresponds to multiplication

by the complex number jo in the frequency domain. Therefore, the expression for the











I(co) Z, TZL



Figure 3-11. Frequency-domain equivalent circuit, using a Norton equivalent current
source, of a flat-plate antenna sensor feeding a load (represented by ZL).

magnitude of the Norton equivalent current source in the frequency domain becomes

I(CO) = CoAplate (jAEnorm(CO)) (3-9)

The quantity Enorm (o) designates that the normal component of the electric field is
now a function of angular frequency and not time.
The Norton equivalent circuit in the frequency domain is shown in Figure 3-11.
The current source is placed in parallel with the source impedance, Zs, and the load
impedance, ZL. The source impedance is the impedance of the antenna itself and the
load impedance is the impedance of any external elements connected to the plate. In
general, the source and load impedances can be resistive, capacitive, inductive, or a
combination of the three.
If the output of the antenna is taken as the voltage across the load impedance, then
the expression for the output voltage in the frequency domain is

( ZSZL
Vout(mCO) = I(CO)Ztotal = I() (Zs ZL) = joAplateEnorm(CO) ZS ZL (3-10)

The output voltage of the antenna in the frequency domain is the quantity
jCmoAplateEnorm(CO), scaled by the frequency-dependent quantity ZsZL/ (Zs +ZL). The
frequency-independent gain of the antenna is given by the quantity FoAplate. If the
quantity ZsZL/ (Zs +ZL) is equal to unity, then the output is simply jaCEnorm (O), scaled
by the frequency-independent quantity oAplate. If the quantity ZsZL/ (Zs + ZL) is equal
to 1/(jco), then the output is simply Enorm(cn) scaled by the frequency-independent









quantity EoAplate. Hence, the output of the antenna is characterized by the source and

load impedance.

Typically, the source impedance of a flat-plate antenna is considered to be purely

capacitive and any resistive and inductive components are ignored. Therefore, the

source impedance becomes

Z, (3-11)
jOICant
Cant is the capacitance of the flat-plate antenna that can either be estimated

theoretically or determined experimentally. The antenna capacitance is a function of

the plate area and the height of the plate above a reference plate buried in the ground.

It may be possible to estimate the antenna capacitance using the relation C = oA/d,

where d is the distance between the antenna plate and the grounded reference plate

beneath it, but it is best to measure it. The load impedance is usually specifically

designed to provide the appropriate system response for the quantity which is to be

measured. In other words, the load impedance will be different depending on whether

one wishes to measure the electric field or its time-derivative, as well as the bandwidth

desired.

The generalized load impedance will be considered to be a capacitor in parallel

with a resistor; any inductive components are ignored. The reason for this will become

apparent when the antenna implementation is discussed. The capacitor, Cit, is typically

known as an integrating capacitance and the resistor, R, is known as the load resistance.

Therefore, the load impedance becomes

1 Rc R
ZL= -1 R j- (3-12)
jmCint R- 1 + jRCint

Since the source and load impedances are both in parallel with the ideal current

source, they can be combined. The total shunt impedance, Ztotal, is

(Zt = ZZ ( R R3-13)
Ztotal = Z IZL jCat I + jcoRCint 1+ joR (Cat +Cint)3)









Therefore, inserting the expression for Ztota into the Equation 3-10 yields


Vot (o) = jAoFoAplateEnorm(mC) (1 + R ( +c) (3-14)
j1 + jR (Cant + C int)

The interpretation of the response of the antenna in the frequency domain

depends on whether the antenna is to be viewed as an electric field or an electric

field time-derivative sensor. From the dE/dt antenna perspective, the response of the

antenna, Gd (co), is given by
dt

Gd (coo) LO oAplate (3-15)
dt jcoEnorm(c) 1 + jR (Cant + Cint) (3-15)

Typically no integrating capacitor is used with a dE/dt antenna (Cint = 0),

therefore the expression for GdEC (c) reduces to
dt

Gd (CO) = oAplateR (1(RC,,) (3-16)
dt I + jCRCant

This is the response of a first order low-pass filter with magnitude response given

by

Gd (O) =o0ApiateR (3-17)
dt r ^-] 2
II + (RCant\)2
The pass-band gain is equal to coApiateR. The -3 dB point (the frequency at which

the output is approximately 0.707 times the output in the pass-band) of the response is

1
C0o0 = (3-18)
RCant

For example, if R = 50 Q and Cant = 30 pF, then oo = 6.67 x 108 s-1. The

corresponding frequency is

noo 6.67 x 108 s 1
fo -- 106 MHz (3-19)
27 27

Therefore, this example flat-plate dE/dt antenna has a -3 dB bandwidth of

approximately 106 MHz. This flat-plate design is a suitable dE/dt antenna for most









wide-band applications. If both R and Cant are doubled, then the -3 dB bandwidth must

necessarily decrease by a factor of four, yielding fo a 26.5 MHz.

If the frequency range of interest lies far below the -3 dB point, then the

expression for the response of the dE/dt antenna in the frequency domain becomes

GdE (o) = oApiateR (3-20)
dt

The above expression is only valid for frequencies which satisfy the condition


CO <(c o =RCa (3-21)

If the same condition is applied to Equation 3-14 (assuming Cit = 0), then the

output voltage of the dE/dt antenna in the frequency domain becomes:


Vout (O) = jCWOoAplateREnorm (CO) (3-22)

To find the output voltage in the time domain, the inverse Fourier transform is

used.

F {X(m)} = x(t) =X(co)ejXtdco (3-23)
27c
However, the integral need not be computed in this case since the differentiation

property of the Fourier transform can be used. Therefore, the expression for the output

voltage in the time domain is


Vot (t) = oAplateR dEOm(t) (3-24)
dt

Although this is a time-domain expression, the frequency constraint still applies

since the time-domain expression is derived from a frequency-domain expression

with that constraint. In other words, the above expression is only valid if the dE/dt

waveform has no significant frequency content above coo.








If the flat-plate antenna is viewed from an electric field antenna perspective, the

response of the antenna in the frequency domain, GE (c), is

GE(Co) = Vot (CO) EoAplateR 1 + (3-25)
Eorm ,,, (C) 1 + j(R (Cant + CinGt)

Typically an integrating capacitor is used with an electric field antenna. Depending

on the measurement design, Cit may or may not dominate Cant, however it best that

it does since the antenna capacitance can potentially vary with antenna movement and

surroundings. Combining the two capacitances into a single term yields

C = Cant + Cint (3-26)



GE (co) = oApiateR 1 oRC (3-27)

The response of the electric field antenna is that of a first order high-pass filter

with magnitude response given by


IGE ()) = oApiateR + RC)(3-28)
1 + ( j(RC)2 2

The pass-band gain is equal to (EoApiate) /C The -3 dB point of the response is

1
00 = RC (3-29)

For example, if R = 500 kQ and C = 0.1 rF, then oo = 20 s1. The corresponding

frequency is
coo 20 s1
fo 3.2 Hz (3-30)
2x 2x
Therefore, the output of this example electric field antenna provides an adequate

approximation of the electric field waveform at frequencies down to about 3 Hz. It

should be noted that coo is a low-frequency roll-off when speaking in terms of an









electric field antenna and a high-frequency roll-off when speaking in terms of a dE/dt

antenna.

The general time-domain expression for the output voltage of an electric field

antenna is found by performing the inverse Fourier transform on the frequency-domain

expression. Rearranging Equation 3-14 (substituting C = Cant +Cit ) yields


Vout(m) = ate --jo Enorm(0) (3-31)

This expression can be viewed as the multiplication of two functions of co. This

can be expressed as

V(o)() = X(o)Y(co) (3-32)

The quantities X(m) and Y(o) can be defined as


X(o) = (3-33)
C + CO



Y(o)= ApateE (C) (3-34)

Therefore, the time-domain expression for the antenna output voltage, Vo,t(t), can

be found by the convolution property of the Fourier transform.


Vout (t) = x(t) y(t) (3-35)

The operator denotes linear convolution and x(t) and y(t) are the inverse Fourier

transforms of X(co) and Y(co), respectively. Convolution is performed by means of the

convolution integral

Vout(t) = x(l)y(t )dl (3-36)

Alternatively, since linear convolution is a commutative operation, Vout(t) can be

expressed as

Vot(t) = y(l)x(t- l)dl (3-37)
J/ o--0








The expressions for x(t) and y(t) are found by using properties of the Fourier
transform: First, x(t) is found by utilizing the differentiation property of the Fourier
transform and a well-known Fourier transform pair.

x(t) = F = e-Ru(t) (3-38)
-v C R- +jc) d7

The term u(t) is the unit-step function and is defined as

1 t>0
u(t) = (3-39)
0 t<0

The time-derivative can be computed by using the product rule of differentiation
and the fact that the derivative of the unit-step function is the Dirac delta function, 8(t),
which is defined as
CO t=0
s(t) = (3-40)
0 t O

Therefore, the expression for x(t) is

x(t) =- e Cu(t) + 6(t) (3-41)

The quantity Eor,,,(o) is an arbitrary function of co, therefore the expression for

y(t) is

y(t) = OAplateEort) (3-42)

The quantity Enorm (t) is simply the inverse Fourier transform of Enorm (o).
Substituting the expressions for x(t) and y(t) into Equation 3-35 yields

vt(t) = (-Re RFu(t) +8(t)) *OApateEnorm(t) (3-43)
V,,=1i+~) KL I ~la








Recalling that the convolution operator obeys the distributive property and that any

function of t convolved with 8(t) is simply the function itself yields

I t 1 -oAplate 8oAplate
Vout(t) = -Re R-cu(t) CAp Eor, (t) + OApaEnorm(t) (3-44)

The convolution is computed via the convolution integral, as shown in Equations

3-36 and 3-37, yielding

OAplate oAplate ,O O
Vout(t) = Apate orm(t) Aplate Enorm(l)e uR (t- l)dl (3-45)

The effect of the unit-step function in the second term of Equation 3-45 can be

incorporated into the limits of integration. Furthermore, the term e RC can be factored

out of the integral. This yields

Vou(t AplateEnorm) Aplatee t Enorm(l)edl (3-46)
Vot (t)C2 e RC- n Rmomt

The above expression is valid for an arbitrary electric-field, Enorm(t), with no

frequency constraints. The antenna output voltage consists of two terms, the first of

which is the electric field scaled by the quantity (eoA) /C. This is the output of an

ideal flat-plate electric-field antenna. The second term is the effect of the non-ideal

low-frequency response of the antenna. As R approaches infinity, the second term

approaches zero and the output is that of an ideal flat-plate electric-field antenna. This

can also be seen by considering the frequency-domain expression for the antenna

output voltage and allowing R to approach infinity. The same argument can be applied

to allowing the quantities RC and RC2 to approach infinity. If C alone is allowed to

approach infinity then in the limit the output voltage will be zero. However, L'Hopital's

Rule can be invoked, and it can be seen that the second term will decrease to zero

before the first. Therefore, it can be said that for very large values of C, the antenna

output voltage will be approximately that of an ideal antenna, with very low gain. Of









course, "very large" and "very low" are relative terms which are determined by the

requirements of the experiment.

The expression for the antenna output voltage can be simplified somewhat if the

electric-field waveform is constrained to be of the following form.

Eorm(t) = s(t)u(t) (3-47)

The quantity s(t) is an arbitrary function of time and u(t) is the unit-step response

which was defined in Equation 3-39. If this is substituted into Equation 3-46, the

antenna output voltage becomes


vot (t e) (t)(t)- Apatee-R t s(l)eRdl (3-48)
C RC2 1J=o

The antenna output voltage can be specified exactly if the electric field waveform

is specified exactly. The response of the antenna to a step-function is of particular

interest since it can be used to crudely approximate the electric field waveform from a

return stroke. If s(t) is defined as the constant Eo, then the electric field waveform is

given by

Eorm (t) = Eou(t) (3-49)

The quantity Eo is amplitude of the step function. The corresponding antenna

output voltage is

LoAplate Ft OAplate
Vout(t) = oApatEoe R + Eo [u(t) -1] (3-50)
C C

The second term in the above expression is zero for all times t > 0, therefore the

expression reduces to
/ Plate
Vout(t) = AplatEoe RC (3-51)
C









The term RC is known as the decay time constant and is typically denoted by T.

The time constant is the inverse of coo.

1
S= RC =- (3-52)
(CO
oo

When the electric field is a step-function, the output of the antenna at time t = T

is a factor of 1/e less than it was at time t = 0. Therefore, the output of the antenna

is only valid for a short period of time relative to T. This is exactly what is expected

since the response of the electric field flat-plate antenna is that of a high-pass filter in

the frequency domain.

For times t << the output of the antenna, given a step-function input of amplitude

Eo, becomes
0oAplate
Vout (t) = Apa Eo (3-53)
C

If the example circuit parameters R= 500 kQ and C = 0.1 rF are again considered,

the decay time constant is


T = (500 kQ) (0.1 F) = 50 ms (3-54)


Given a step-function input, the output of this example electric field antenna is

only valid for times much less than 50 ms. Depending on the application, this time

constant may or may not be acceptable.

Increasing the time constant by increasing the capacitance necessarily decreases

the output of the antenna by a factor of 1/C. The resistance can be increased with

no effect on the gain, but there are practical limits to how high this resistance can be

increased. Typically both R and C are increased with C being adjusted to yield an

acceptable gain.

It should be noted that if Equation 3-46 is solved for En orm(t) as a function out(t)

it may be possible to "correct" the measured electric field. In other words, it may be


























Figure 3-12. Aluminum flat plate antenna used in the MSE.


possible to remove the effects of the non-ideal characteristics of the antenna from the

measured data if the antenna can modeled in such a way that Equation 3-46 applies.

3.3.1.2 Flat-plate antenna implementation

Now that the output of the flat-plate antenna has been characterized in the time

and frequency domains, the implementation of the MSE electric field and electric field

time-derivative antennas can be discussed.

Both the electric field and the electric field time-derivative measurements utilize

identical aluminum circular flat-plate antennas of area 0.16 m2 and diameter 0.45 m as

shown in Figure 3-12.

The antenna consists of a hollow rectangular aluminum housing with a circular

portion of the top face isolated from the remainder of the top face of the structure by

a 0.6 cm-wide annular air gap surrounding it. The circular portion is kept electrically

isolated by this annular gap and six nylon standoffs which are used to mount it to

the bottom (ground) of the housing. The circular portion is the sensor while the

remainder of the housing is connected to a 3 m long ground rod by a short length of

12 AWG wire and a lug mounted on the side of the housing. A lug on the underside

of the circular plate connects an approximately 20 cm length of 16 AWG wire to the









center conductor of a female BNC connector mounted to the side of the housing. The

inductance of a 20 cm length of 16 AWG wire is approximately 10 nH. The outer

conductor of the BNC connector is connected to the grounded aluminum housing.

Therefore, the voltage measured across the BNC connector is the voltage which

appears between the circular plate and the grounded antenna housing. A detailed

mechanical drawing of the antenna is shown in Figure 3-13.

The antenna housing was placed on top of the ground. In order to simulate a flat

ground, pieces of wire mesh screen which extended from each side of the antenna were

attached to the top of the the four sides of the metal antenna housing. One portion

of the screen was about one meter in length while the others were about a third of a

meter in length. The electronics for each electric field or electric field time-derivative

measurement was stored in a metal Hoffman enclosure (also known as a Hoffman box)

which was buried underground about one meter from the antenna. Each hole was about

a half a meter deep and the Hoffman box was placed on a shelf angled downward so

as to drain any water acquired in the hole away from the box. Furthermore, pieces

of wood were used to secure the box in the hole. The Hoffman box was placed

underground so that it would be protected from the external environment and the area

surrounding the antenna would be as flat as possible. In addition, a piece of reflective

insulation was placed over the hole to protect the electronics from the heat of the

sun. The 1 m length of screen covered the hole and the insulation. A length of 50 Q

coaxial cable with male BNC connectors connected the antenna to a female BNC

bulkhead feed-through connector mounted to the side of the Hoffman box. This cable

was enclosed in metal shield braid which was secured to the male BNC connectors

on each end of the cable by metal hose clamps. Therefore, the shield braid and the

Hoffman enclosure are electrically connected to the grounded antenna housing. This

configuration served to electromagnetically shield the coaxial cable and electronics,

minimize the electric field enhancement of the antenna, and protect the electronics











































3/4 r


2 Ground Rod Attach Lug


2 3/16 DE Nylon
W DnpEdge J Stand off


Connection Point
Between BNC and Plate


v-J


CROSS SECTION

Figure 3-13. Detailed mechanical drawing of the aluminum flat plate antenna used in
the MSE. Adapted from (Crawford 1998).


\ Q /






/ 8.75 "









@ 9" @





I\ ^-/
I---60
6"










-----_---------- -----------------
- - - -


TOP VIEW
22 "


Female BNC
Connector









Wire Screen
Reflective Insulation
SFiber-optic Cable
Flat Plate Antenna Fiber-optic Cable




Cable


Ground Rod

Figure 3-14. Diagram of an installation of a MSE measurement utilizing a flat-plate
antenna.

themselves from the external environment. A drawing of this configuration is shown in

Figure 3-14. This configuration is very similar to that described in (Crawford 1998).

3.3.1.3 Electric field measurement implementation

As described in Section 3.3.1.1, essentially the only feature differentiating

an electric field measurement and an electric field time-derivative measurement

implemented with identical flat-plate antennas is the choice of the load impedance

of the antenna. This choice is not trivial, however, and an improper choice of circuit

parameters will almost certainly lead to a poor-quality measurement. Also, as stated

previously, the load impedance will be considered to be an integrating capacitor, Cint, in

parallel with a load resistance, R. The integrating capacitance affects both the gain and

the decay time-constant of the antenna while the load resistance only affects the time

constant. The antenna capacitance must, in general, also be taken into account but it

can be ignored if the integrating capacitance is much greater.


Cint >> Cant


(3-55)









In the frequency domain, this implies that the impedance of the antenna is much

higher than that of the integrating capacitor.

1 1
>> (3-56)
jOCant JOCint

In 2001, 0.5 rF integrating capacitors were used in the MSE electric field

measurements. The antenna capacitance was measured to be about 80 pF (although

this measurement was very sensitive to the position of the antenna); hence the above

assumption holds and the antenna capacitance can be ignored. Eight capacitor units

were built by George Schnetzer in the summer of 2001. Each unit consisted of a

5.5 cm by 2.5 cm by 2 cm box with a male BNC connector on one end and a female

BNC connector on the other. Five 0.1 rF capacitors were placed in parallel inside each

box to achieve the desired capacitance.

One of these capacitor units is pictured in Figure 3-15. The operating range of

each integrating capacitor was tested by placing the capacitor in parallel with the 50 Q

input of an oscilloscope and feeding it with a 1 V sinusoidal voltage source having a

50 Q source resistance. The expected magnitude response of this circuit is given by

1
Vcapacitor (C) = (3-57)
/1 + (25cC)2

The measured and expected test circuit responses for an integrating capacitor unit

used in 2001 are shown in Figure 3-16.

The response of the capacitor ceases to be ideal at about 3 MHz, an undesirable

low value. The cause was believed to be a series resonance of the capacitor and the

inductance of the capacitor leads. The capacitors were placed in parallel in order to

minimize the inductance of the leads but the inductance was not decreased low enough.

The resonant frequency of an LC circuit (either series or parallel) is defined as

1


(3-58)


cR =
= 7c









Pomona Box










Male BNC Female BNC
Connector Connector











Capacitors in Parallel

Figure 3-15. Integrator capacitor assembly used in 2001. A) Closed Pomona box. B)
Box open to show interior.

If C = 0.5 iF and oR = 27 (3 x 106) s 1, then L = 5.3 nH. Hence, only a

very small lead inductance is required to achieve a resonance at 3 MHz when 0.5 rF

capacitors are used. It follows that the magnitude response of the electric-field

measurements in 2001 was limited to 3 MHz due to the non-ideal integrating

capacitors.

For 2002, in order to achieve better frequency response, the capacitance was

reduced by a factor of 2.5 to 0.2 rF to increase the resonant frequency to above

5 MHz. If C = 0.2 kF and L 5 nH, then coR 2 2n (5 x 106) S-1. In addition,

military-grade ceramic capacitors were used and the number of capacitors placed in

parallel was increased, hence decreasing the series inductance. For two of the new

units, ten 0.022 rF capacitors were placed in parallel while two other units consisted










Response of Circuit to Test Integrating Capacitor 01-01
1 ; ;(------------------------ =

\\\




0.1 -

CL



0.01





0.001
0 10 100 1000 10000 5 6 7
10 10 10
Frequency [Hz] 10 10 10

Figure 3-16. Measured (dashed line) and expected (solid line) test circuit responses
for integrating capacitor unit 01-01 (0.477 RF).


of nine 0.022 RF capacitors. One unit was constructed from eleven 0.018 RF capacitors

and another unit consisted of seven 0.022 RF capacitors along with two 0.017 RF

capacitors.

All of the new capacitor units integrated properly up to at least 5 MHz, although

most exhibited deviations from an ideal integrator between 5 and 10 MHz. Some of the

units experience a drop in gain after 5 MHz while others experienced an increase. This

situation was considered acceptable since the measurements were to be band-limited

to below 5 MHz by the built in anti-aliasing filter of the digitizer The integrating

capacitors used in 2002 were tested in the same way as in 2001, with the expected

response given by Equation 3-57. The measured and expected test circuit responses for

an integrating capacitor unit built in 2002 are shown in Figure 3-17.

The load resistance, R, is typically the input resistance of an amplifier or a

fiber-optic transmitter. The input capacitance of the fiber-optic transmitter was

considered negligible and hence ignored. The input resistance of the Opticomm










Response of Circuit to Test Integrating Capacitor 02-09
1; ------------------------.





0.1

05










0 10 100 1000 10000 5 6 7
10 10 10
Frequency [Hz] 10 10 10

Figure 3-17. Measured (dashed line) and expected (solid line) test circuit responses
for integrating capacitor unit 02-09 (0.209 rF).


MMV-120C fiber-optic transmitter (described in Section 3.4.1) is 68 kQ. This yields

decay time constants of 34 ms and 13.6 ms when 0.5 rF and 0.2 rF integrating

capacitors are used, respectively. Therefore, an amplifier with high input resistance

was required to increase the time constant since increasing the integrating capacitance

further could result in further resonance problems. In 2001, Keith Rambo and George

Schnetzer designed a unity gain amplifier with an input impedance of 5.1 MQ. A

schematic of the amplifier is shown in Figure 3-18 (the power supply circuitry is

omitted for simplicity).

An input resistance of 5.1 MQ yields decay time constants of 2.55 s and 1.02 s

when 0.5 rF and 0.2 rF integrating capacitors are used, respectively. The decay

time constant in either case was sufficient to measure accurately the overall electric

field change of a natural cloud-to-ground lightning flash since (Berger et al. 1975)

documented the median flash duration, excluding single-stroke flashes, to be about

180 ms. In 2002, the gain of the amplifier was increased to two in order to lower the













AD825 0 O UT
IN o -

5.1 MQ




Figure 3-18. Schematic of the high-impedance amplifier used in the 2001 MSE.

full-scale range of the measurement. The output resistance of the AD825 op-amp is

approximately 8 Q, although the feedback network reduces this to a fraction of an

Ohm. Therefore, the output of the amplifier will be the same regardless of whether it is

terminated in 50 Q or high impedance.

The amplifier is powered from a 12 V battery. The amplifier ground is at located

6 V above the negative battery terminal in order to bias the AD825 op-amp with 6 V.

Therefore, if a single battery were to be used to power all of the electronics in the

measurement, the output of the amplifier would float 6 V above the common ground

which is far above the maximum input range of the fiber-optic transmitter. In order to

alleviate this problem, both the input and output of the amplifier would have to be AC

coupled (by a DC blocking capacitor), or the amplifier would have to be powered by a

separate battery than the remainder of the electronics. The second option was chosen

since it was desirable to keep the low-frequency roll-off of the measurement to a

minimum. The amplifier was connected to a separate 12 V battery through a relay. The

control circuit of the relay was connected to the primary measurement battery (through

the PIC controller) and the load circuit of the relay was connected to the separate 12 V

battery. Therefore, when the PIC controller supplies power to the electronics in the

measurement, it also connects the amplifier to its battery via the relay. Hence, the










Measured Frequency Response of Hi-Z Amplifier 01-07
1 -1- ----- ^--------- --------------- W-------^-------^------------ -------- --------
0 1 i i -















0.1
0 10 100 1000 10000 5 6 7
Frequency [Hz] 10 10 10

Figure 3-19. Measured frequency response of high impedance amplifier 01-07.


amplifier is not powered when the measurement is turned off. The measured frequency

response of an amplifier is shown in Figure 3-19.

A diagram of an MSE electric field measurement is shown in Figure 3-20. The

antenna is placed on the ground (surrounded by a metal screen as described in Section

3.3.1.2) and the electronics are enclosed in a metal Hoffman enclosure (indicated

by the dashed line). In order to eliminate ground loops, all electronic components

were isolated from each other and the metal box by pieces of plastic and Styrofoam.

The integrating capacitor, C, enclosed in a Pomona box, is directly connected to the

female BNC connected mounted to the inside of the Hoffman box. The other end

of the Pomona box is connected to the input of the amplifier (having gain Gamp and

input resistance R) via a short length of 50 Q coaxial cable. Note that this cable is not

terminated in its characteristic impedance of 50 Q. However, the length of the cable is

short enough that reflections in the cable should not be a problem. The output of the

amplifier is connected to the input of a PIC controller. The output of the PIC controller

is connected to an Opticomm MMV-120C fiber-optic transmitter terminated in 50 Q.










PIC Controller PIC Controller
Supplies Power to Supplies Power to
Flat-plate Amplifier Relay Fiber-optic Transmitter
Antenna
S Hi-Z Amplifier PIC Controller
GP \c _Fiber-optic
S amp GP0 Transmitter Fiber-optic
S\ -- Cable

Amplifier PIC
Controller
Battery
12 V Battery
12 V 50 Q In-line Terminator
Integrating Capacitor in a Metal Box
Metal Enclosure

Figure 3-20. Diagram of a MSE electric field measurement.


Since the PIC controller is terminated in 50 Q, it can be thought of as a 50 Q in-line

attenuator of value Gpic.

Using the expression for the ideal output voltage of the electric field flat-plate

antenna, the expression for the voltage at the input of the fiber-optic transmitter is


VFOT (t) = Aate Enorm (t)GampGPIC (3-59)


The above expression assumes that the output voltage of the antenna can be

approximated by the ideal output and the gain of the amplifier is flat in the frequency

range of interest.

The output of the fiber-optic receiver is connected to a digitizer terminated in

50 Q by a length of 50 Q coaxial cable. The voltage at the input of the digitizing

oscilloscope is the voltage present at the input of the fiber-optic transmitter modified by

the fiber-optic link. If the effect of the link is assumed to be a frequency-independent

gain or attenuation (of value Glink), then the expression for the voltage at the input of

the oscilloscope is


Vscope(t) = -- GampGPicGlink Enorm(t) (3-60)
( oA









In 2001, Aplate = 0.16 m2, C = 0.5 gF, Gamp = 1, and Gplc = 1. According to the

manufacturer specifications, Glink = 1, however this will be considered a nominal value

since in practice Glink is often not unity. In addition, the value for C varied between

individual electric field measurements, hence C = 0.5 rF is also considered a nominal

value. Therefore, the nominal voltage seen at the input of the scope for the 2001 MSE

electric field measurements is given by


scope-nom(t) = (2.83 x 106) Enorm(t) (3-61)


Vscope-nom(t) is expressed in units of V when Enorm(t) is expressed in units of
V m1. If ,scope-nom(t) = 1 V and the above expression is solved for Enorm(t), then


Enorm = 3.53 x 105 Vm1 V1 (3-62)

If Enorm is expressed in units of kV m1, then the above expression becomes


Enorm = 353 kV m1 V1 (3-63)

Thus, nominally, one volt present at the input of the digitizer corresponds to an

electric field magnitude of about 350 kV m1.

In 2002, Aplate = 0.16 m2, C = 0.2 kF, Gamp = 2, and Gpc = 1. If, as before, Glink

is assumed to have a nominal value of unity, the nominal voltage seen at the input of

the scope for the 2002 MSE electric field measurements is


Vscopenom(t)= (1.42x 10 5)Enorm(t) (3-64)

Again, solving the equation for Enorm, expressed in units of kV m1, yields


Enorm = 70.6 kV m1 V 1 (3-65)

If the true capacitor value for each electric field measurement is substituted into

Equation 3-60, the expression for the electric field as a function of the voltage at the









Table 3-7. Salient characteristics of the 2001 MSE electric field measurements.

Designation C (rF) T (s) Units/Volt (kV m- V-1
E-1 0.477 2.43 336.9
E-2 0.517 2.64 365.1
E-4 0.504 2.57 355.9
E-5 0.503 2.57 355.2
E-6 0.510 2.60 360.2
E-8 0.501 2.56 353.8
E-9 0.507 2.59 358.1
E-10 0.508 2.59 358.8

Table 3-8. Salient characteristics of the 2002 MSE electric field measurements.

Designation C (rF) T (s) Units/Volt (kV m- V-
E-2 0.230 1.17 81.2
E-4 0.202 1.03 71.3
E-5 0.228 1.16 80.5
E-6 0.204 1.04 72.0
E-9 0.198 1.01 69.9
E-10 0.210 1.07 74.2


input of the digitizer becomes more accurate. Finally, if the gain of the fiber-optic link

is known, the expression can be made even more accurate. In general, the capacitance

values are measured once before being installed in the field while the gain of the fiber

optic link is measured with a square wave signal (as described in Section 3.4.5) at least

once during each thunderstorm.

Tables 3-7 and 3-8 summarize the salient characteristics of all of the individual

2001 and 2002 MSE electric field measurements, including the integrating capacitance,

the corresponding decay time constant, and the corresponding units per volt (assuming

Glink = 1).

Opticomm MMV-120C fiber-optic links were used to transmit the analog

waveforms to the Launch Control trailer where they were digitized. The electric

field measurements were digitized continuously for 800 ms (200 ms pre-trigger)

at 10 MHz on a Yokogawa DL716 digital storage oscilloscope. The electric field

waveforms were band-limited to 4 MHz ( -3 dB) by the anti-aliasing filter associated










Flat-plate
Antenna


S.50. ._______ Fiber-optic
Q Cable

PIC
Controller
Battery
12 V 50 Q In-line Terminator


Metal Enclosure

Figure 3-21. Diagram of a MSE dE/dt measurement.


with the DL716 digitizer. Detailed descriptions of the Opticomm fiber-optic links and

the DL716 digitizer are given in Sections 3.4.1 and 3.5.1, respectively.

3.3.1.4 Electric field time-derivative measurement implementation

As stated previously, no integrating capacitor is used in a dE/dt measurement

(Cit = 0), and therefore the load impedance is simply the load resistance, R.

As with an electric-field measurement, the flat-plate antenna was connected

directly to the female BNC bulkhead feed-through connector mounted to the side

of the Hoffman enclosure. Inside the Hoffman box, a short length of 50 Q coaxial

cable connected the end of the bulkhead feed-through BNC connector inside of the

box to the input of a PIC controller. The output of the PIC controller is connected

to an Opticomm MMV-120C fiber-optic transmitter terminated in 50 Q. Since the

PIC controller is terminated in 50 Q, the PIC controller appears to be a 50 Q in-line

attenuator of value Gpic to the antenna. Therefore, the load resistance, R, is 50 Q.

In order to eliminate ground loops, all electronic components were isolated from

each other and the metal box by pieces of plastic and Styrofoam. A diagram of the

configuration is presented in Figure 3-21.









The capacitance of the 0.16 m2 flat-plate antennas was measured to be about

80 pF. Therefore, using Equation 3-18, the -3 dB bandwidth of the response is
1 1
C = 2.5 x 108 s1 (3-66)
RC,,nt (50 Q) (80 pF)

This corresponds to a frequency of fo a 40 MHz. Since the dE/dt measurements

were to be band-limited to 20 MHz (-3 dB) via anti-aliasing filters at the digitizers

inputs, the response of the antenna can be considered uniform over the complete

frequency range of interest. Therefore, Equation 3-24 can be used to approximate the

voltage output of the dE/dt antenna. The expression for the voltage at the input of the

fiber-optic transmitter is

dE,/A n nor ) (3-67
VFOT (t) = FoAplateRGPIc dEnrmt (3-67)

The attenuation for the dE/dt measurements (Gpic) was varied over the course of

the 2002 season. A value of 0.199 (-14 dB) was originally used, however this resulted

in an output voltage that was too low. Therefore the attenuation was changed to 0.316

(-10 dB) but this too resulted in a low output voltage. Finally, the attenuation setting

was changed to 0.501 (-6 dB) to obtain an acceptable output voltage.

Using this final value for Gpic along with Aplate = 0.16 m2 and R = 50 Q yields

the following expression for vFOT (t) when dE/dt is expressed in units of V m1 s1


VFOT(t) =(3.55 x 1011) dErm (3-68)

If dE/dt is expressed in units of kV m 1 s 1, the above expression becomes


VFOT(t) = (0.0355) dEO(t) (3-69)
dt

The voltage present at the input of the digitizing oscilloscope is the voltage

present at the input of the fiber-optic transmitter modified by the fiber-optic link. If the

effect of the link is assumed to be a frequency independent gain or attenuation, Glink,









then the expression for the voltage at the input of the oscilloscope is


scope(t) = FOT (t)Glnk = (0.0355)Glink dEorm (3-70)
Vcope .3dt

If the nominal gain of the fiber-optic link is assumed to be one, the nominal

voltage at the input of the digitizing scope is the same as that present at the input of

the fiber-optic transmitter

dE .orm t )
Vscopenom(t) = (0.0355) dEnm(t) (3-71)

If Vscope is assumed to be 1 V and the above expression is solved for dEnorm/dt,

then
dEnorm, Io IV-
dE = 28.2 kVm s-1 V1 (3-72)
dt

Hence, nominally, one volt present at the input of the digitizing oscilloscope

corresponds to a dE/dt magnitude of approximately 28 kV m1 s-1. This value is only

valid for data sets where Gpic is equal to 0.501, however the result for any attenuation

value can be obtained by substituting the appropriate value into Equation 3-67.

Opticomm MMV-120C fiber-optic links were used to transmit the analog

waveforms to the Launch Control trailer where they were digitized. The dE/dt

measurements were digitized on a LeCroy LT374 Waverunner2 digital storage

oscilloscope in segmented memory mode. The internal memory was divided into

four segments with each segment requiring a separate trigger signal; hence data for

a total of four return strokes could be acquired. Unlike the measurements digitized

by the DL716, no data were acquired during inter-stroke intervals. During July of

2002 each segment was digitized at 50 MHz for 20 ms (19 ms pre-trigger). In August

of 2002 the digitization rate and acquisition length for each segment were changed

to 200 MHz and 5 ms (4 ms pre-trigger), respectively. The dE/dt waveforms were

band-limited to 20 MHz (-3 dB) by anti-aliasing filters at the digitizers inputs. Detailed









descriptions of the Opticomm fiber-optic links and the LT374 digitizer are given in

Sections 3.4.1 and 3.5.3, respectively.

3.3.2 Magnetic Field and Magnetic Field Time-Derivative Measurements

In 2001, four magnetic field measurements were fielded. Two measurements were

located at Station 4, and the remaining two were located at Station 9. The two sensors

at each station were arranged so that two orthogonal components of the horizontal

magnetic field could be sensed. This is very similar to the configuration described in

(Crawford et al. 2001).

In 2002, in order to conserve equipment, only one magnetic field measurement

from each station was used, and hence the total number of magnetic field measurements

was reduced from four to two. Furthermore, this meant that only one orthogonal

component of the magnetic field could be sensed at each location. Although the same

sensors were used in both 2001 and 2002, the electronics associated with the sensors

differed for the two years.

In addition, in 2002 four magnetic field time-derivative (dB/dt) measurements

were fielded. Two derivative measurements were located at Station 1, which were

arranged so that two orthogonal components of the azimuth field could be sensed.

The remaining two measurements were located at Stations 4 and 9, with only one

orthogonal component of the field being sensed at each location. Square loops

constructed from 50 Q coaxial cable were used to sense both the magnetic field and its

time-derivative.

3.3.2.1 Analysis of a loop antenna

The Thevenin or Norton equivalent circuit of a coaxial loop antenna can be

derived by first considering Faraday's Law


E s = d (3-73)
c dt









The quantity E is the electric-field vector, ds is a differential length about an

arbitrary closed path C, and ( is the magnetic flux through the surface defined by

the closed path C and is expressed in units of Wb. Faraday's Law states that the line

integral of the electric field about an arbitrary closed path C is equal to the negative

of the time-derivative of the magnetic flux through the surface defined by C. The

magnetic flux is defined as

S= jA.da (3-74)

The quantity A is the magnetic induction (or magnetic flux density) vector and

da is a differential area on an arbitrary open surface S, with da being normal to S. B

is expressed in units of Wb m 2 or T. Therefore, ( is equal to the surface integral of

the magnetic induction over the open surface S. If the open surface S is defined by the

closed path C, then Equation 3-73 can be re-written as


E .ds = Bda (3-75)
c dt s

Moreover, if the medium is stationary, then both C and S will be stationary, and

the expression can be further simplified further to

dA-
ds=- -- da (3-76)
Jc Jsdt

Hence, if the medium is stationary, then the line integral of the electric field

about the closed path C is equal to the negative of the surface integral (taken over the

surface defined by C) of the time-derivative of the magnetic induction. If the medium

is linear, homogeneous, and non-conducting (e.g. air), the magnetic induction, B, can

be expressed as the magnetic field intensity, H, times the permeability of the medium,

pu. If the medium is air, the value of u is approximately Po = 47n x 10 7 H m1 which is

the permeability of free space. Since B and H differ by only a constant in air, the term

"magnetic field" is used to refer to either H or B.









If the magnetic field is uniform over the entire surface and the entire surface lies

in the same plane, then the expression further simplifies to

1i^i d (" i r -^dBnorm
I E i s = -Aloop d ( = -AodBnorm (3-77)
Sloop dt dt

The quantity Aloop is the total area of the surface and n is the unit vector normal

to the surface. Hence, the line integral of the electric field about the closed path C is

equal to the negative of the area of the surface bound by C times the time-derivative

of the magnitude of the normal component of magnetic field through the surface.

The above expression is only valid if the magnetic field is uniform over the entire

surface of interest and the surface lies in a single plane. If B is a component of an

electromagnetic wave, then the expression is only valid if the longest dimension of the

surface is much smaller than a quarter of a wavelength.

The term on the left-hand side of Equation 3-77 is defined as the electromotive

force, or EMF, and is expressed in units of V. This can be interpreted by saying that

if a perfectly conducting wire placed along the path C is broken, the voltage measured

between the two open ends of the wire is equal the time-derivative of the normal

component of magnetic field through the surface bound by the wire times the area

bounded by the wire. Hence, a loop of wire can be used to sense the component of

the magnetic field which is normal to the plane of the loop. The open circuit voltage,

vioop(t), of a broken loop of wire in the presence of a time-varying magnetic field is
given by

vods = Aloop dBr(t) (3-78)
cVloopW = dt
A wire-loop antenna in the presence of a uniform time-varying magnetic field can

be viewed as a voltage source whose magnitude is proportional to the time derivative

of the component of the magnetic field which is normal to the plane of the loop. This

is the Thevenin equivalent voltage of a loop antenna and the basis of the equivalent













V100O( a)) ZL




Figure 3-22. Frequency-domain equivalent circuit, using a Thevenin equivalent
voltage source, of a loop antenna sensor feeding a load (represented
by ZL).

circuit. As with the electric-field antenna, the circuit analysis for the magnetic field

antenna will be performed in the frequency domain.

The frequency domain Thevenin equivalent (open circuit) voltage source of a wire

loop antenna is given by


Vioop (C) = -jClAoopBnorm (CO) (3-79)

The voltage source is in series with source impedance, Zs, and load impedance,

ZL, as shown in Figure 3-22. The source impedance is the impedance of the antenna

itself and the load impedance is the impedance of any external elements connected

to the loop. In general, the source and load impedances can be resistive, capacitive,

inductive, or a combination of the three. If the output of the antenna is taken as the

voltage across the load impedance, then the expression for the output voltage in the

frequency domain is

ZL ZL
Vout (CO) =-Z + Z o(C)= Z + loopBnorm (CO) (3-80)
zS+ZL Zs+ ZL

The output voltage of the antenna in the frequency domain is -jCOAioopBnorm(CO),

scaled by the frequency-dependent quantity ZL/ (Zs + ZL). The frequency-independent

gain of the antenna is A0oop.









Typically the source impedance is considered a resistance, Rloop, in series with an

inductive reactance, j0Lloop. Therefore, the source impedance is

ZS = Rloop + jLloop (3-81)

The load impedance is considered a resistance, Rload, and hence

ZL = Rload (3-82)

Substituting the expressions for the source and load impedances (Equations 3-81

and 3-82) into the expression for the frequency-domain output voltage of the loop

antenna (Equation 3-80) yields


Vot, (C) = ( A-oJ-Rloadload joBnorm (CO) (3-83)
jcoLloop +Rloop +Rload /

As with the flat-plate antenna, the response of the loop antenna in the frequency

domain depends on whether the antenna is to be viewed as a magnetic field sensor

or a magnetic field time-derivative sensor. Unlike the flat-plate antenna, only the

time-derivative perspective will be considered for the loop antenna, since active

integration will be used to extend the loop antenna into a magnetic field sensor. From

the dB/dt antenna perspective, the magnitude response of the antenna, GdB (C), is given
dt
by
Gd (o) Vout (co) _AloopRload (3-84)
GjdB (C) = (3-84)
dt jBBnorm (C) jLloop + Rloop + Rload

Like the dE/dt antenna, the magnitude response of the dB/dt antenna is that of a

first order low-pass filter. The magnitude response is given by



GdB(o)= AlolpRoad--1 (3-85)
t Roop +Rload + Lp 2
SAlpRldloop load
Gt ( )RI +Rod [ i1O'~l~ ] ~ (-









The pass-band gain is equal to (AloopRload)/(Rloop +Rload). The -3 dB point

(the frequency at which the output is approximately 0.707 times the output in the

pass-band) of the magnitude response is

SRloop + Rload (3-86)
Lloop

If the frequency range of interest lies far below the -3 dB point, then the

expression for the response of the loop antenna in the frequency domain becomes


GdB () = AloopRload (3-87)
dt Rloop +Rload

The above expression is only valid for frequencies which satisfy the condition


S ( Rloop +Rload) (3-88)
Lloop
If the same condition is applied to Equation 3-83, then the output voltage of the

loop antenna in the frequency domain becomes


Vt (o) = oopRoad Bom () (3-89)
Rloop +Rload

To find the output voltage in the time domain, the inverse Fourier transform

(defined in Equation 3-23) is used. The integral need not be computed in this case

since the differentiation property of the Fourier transform can be used. Therefore, the

expression for the output voltage in the time domain is

VOW (t) AloopRload dBnorm(t) (3-90)
Vot (t) = (3-90)
Rloop +Rload dt

Although this is a time-domain expression, the frequency constraint still applies

since the time-domain expression is derived from a frequency-domain expression

with that constraint. In other words, the above expression is only valid if the dB/dt

waveform has no significant frequency content above coo.








The general time-domain expression for Vot (t) can be found by performing the

inverse Fourier transform on Equation 3-83. Equation 3-83 can be rearranged to yield


1 () = RloadAloop joB or(o)) (3-91)
Rloop+Rload Loop +
\ Lloop i o

This expression can be viewed as the multiplication of two functions of co. This

can be expressed as

Vout(o) = X(o)Y(o) (3-92)

The quantities X(m) and Y(o) are defined as

X() = 1 (3-93)
Rloop+Rload1 + j



Y(co) = RloadAloop jBr ) (3-94)
Lloop
Therefore, the time-domain expression for the antenna output voltage, Vo,t(t), can

be found by the convolution property of the Fourier transform.


Vot (t) = x(t) y(t) (3-95)

The operator denotes linear convolution and is evaluated using the convolution

integral. The quantities x(t) and y(t) denote the inverse Fourier transforms of X(o)

and Y(o), respectively. The inverse Fourier transform of X(o) can be found by using a

Fourier transform table.
(Rloop +Rload t
x(t) =e Lloop / u(t) (3-96)

The quantity u(t) is the unit-step function, as defined by Equation 3-39. The

inverse Fourier transform of Y(co) is found by invoking the differentiation property of

the Fourier transform.
t RloadAloop dBnorm (t) (-
y(t) p (397)
Lloop dt