i5, a.11
HIGI{ SPECTRAL RESOLUTION STUDIES
OF THE
ATOMIC OXYGEN, ),630 nm, DAYGLOI,I
A Thesís for the degree of Doctor of Philosophy
submítted by
TERRY DOUGLAS COCKS, B.Sc. (Hons)
THE MAI/üSON INSTITUTE FOR ANTARCTIC RESEARCH
UNIVERSTTY OF ADELATDE
MARCII, L977. ABSTRACT
Using two Fabry-Perot Interfe.rometers (F.P.I") in scrics, the line emission of atomic oxygen at À630nrn tras ireen isolaled from rhe large background of scaEtered sunlight, with a spectral resolutio4 of 2.1 x 10s, permitting estimates Eo be made of the tenperature and wind velocity characteristic of the neutral thermosphere (¡,200 - 250km) during daytime. The data also yr'-e1<1 information on the emission intensity and the Ring effecÈ.
The research project was devel-opmental in nature and this report is biased towards a description of the equiprnenL and techniques used.
The design and construction of a 1oçø resolution, mechanical-ly scanne A daÈa analysis scheme is described that requires only empirical informaEion to be used. The parameters relating to temperature, intensiLy and wind velocity are estimated by a least squares fitting routine performect in Ëhe Fourier transfonn domain. Ttre re-Liability of the analysis routine and the experimental techníque are established by numerically simulatíng the observational data. Observations r^rere made at Mt. Torrens (34oS, l39oE). Results are presented for the perio The thermospheric temperature \¡ras found Èo vary from abouÈ 800oK duríng the morning t\dilight to a maximum of 1200oK near 1400 hours LMT. r1- ïhe zonal vrinds are predominantly westwarcl ("r,75nt s-l¡ drrring the day wíth a maximum a few trours after Sunrise and reversíng to an eastward direction near 1900 hours LMT. The meridional winds are equatorward duríng the morning twilight with the daytime velocities being little different fTom zeTo. The emission intensity was found to vary consisLently from abou¡ 0.35 kR at a solar zenith angle of 95o to about 2.5 kR at a solar zeníth angle of l0o, The intensity variation had a broad maximum at about 1200 hours Ll{T- This is the first experiment to derive neutral i¡ind velocitj-es fronl observations of the À63Onm emission line duríng the day :rnd is the first ground based experiment to yield data of sufficÍ-ent accuracy ancl reliability to permit a study of the variaEion of the daytime Ehermospheric te-mperature. The successful application of techniques to measure the spectral characteristics of the atomic o)çygen dayglow rro\^7 means that Èhe thermosphere can be directly monitored by a ground based observatory over the full diurnal cycle. lii This thesís cont¿ríns no material which has been accepted for the award of any other degree or dipJ-oma in any UniversiEy, and, to the best of the authorrs knowledge and belief, iÈ contains no material previously published or written by anot-her pe-rson, except when due reference ís made in Ehe texÈ. (T. D. COCKS) iv ACI(I{OI,ILEDGE}MNTS The author is indebted for the support and co-operation received fr<¡m the personnel of the Mawson lrrstitute throughout this project. The author thankfully acknovrledges the encour:agement and guídance provided by his supervisor, Dr. F, Jacka, Director of the Mawson Institute. It was he who inscigated Ëhe use of Fabry-Perot inEerfero- meters i-n nighrglow observations at this laboratory, and proposed Èhe dayglow project. I'fr. D. Creighton made many valuable contributions to the construction of the spectrometer, particularly the electronics. The author wishes to thank Drs. P. tr^Iilksch and A. Bower for providing a high resolution F"P.I. that performed so relíably throughout this work. The rnany discussions held with Dr. P. Wilksch were of immense help. The mechanical constructíon work vras undertalcen with the assístance of Mr. F. Fone and l"Ir. F. Koltai. Their willingness to continually modífy pieces of equipment at a momentrs notíce was much appreciated. The author wishes to express his deep appreciatíon to his wife, Kathy. Her constant supporÈ and encouragement contributed much to the success of f-his projecÈ. For a part of the period spent by the author on this project, he wás supported by a Commonr¿ealth Postgraduate Scholarsl-rip. He ís indebted to his parents and parents-in-law for the willing assistance they provided. The auÈhor is grateful to Mrs. tr^lyaÈt for the typing of this thesis. v CONTENTS AsSTRACT l-l_ STATEMENT l_v ACKNOI,JLEDGEMENTS v I INTRODUCTION I 1.1 Optical Radiation Sensing of Thermospheric Ternperature an 1.2 Observations of the À630nm [Of] Airglorv 3 1.3 Dayglow Observations 5 1.3.1. Introduction 5 L.3.2. Expected Values of Ternperature, InJind Velocity and Emission Intensi-ty 5 1.3.3. The Background of Scattered Sunlight 6 I.3.4. The Ring Effect 7 1.3.5. Doppler Shifts of the Solar Spectrum B 1.3.6. Absorptj-on by Atmospheríc O2 B I.4 Previous À630nm Dayglow Observations 9 1.5 The Mawson Institute Dayglow ExperimenÈ 1l 1.6 Summary l4 2 TIIE DUAL ETALON FASRY-PEROT INTERFEROMETER THEORY 15 2.L Introduction 15 2.2 The Spectrorneter Transmission Profile, Recorded Spectrum and Transmitted Flux I6 2.3 Spectrometer Selection 20 2.1+ Fabry-Perot Theory : Single Etalon 22 2.4.1. General Princíples 22 2.4.2. Effect of Plate Defects 24 vi 2.4.3. Effect of FiniÈe Field of View 27 2.4 .4. Srrnrnary 30 2.5 Fabry-Perot Theory : Dual EÈalon 31 2.5.L. IntroducÈíon and General Polyetalon Principles 31 2.5,2. Etalon Coupling 3B 2.5.3. Number of Etalons Required 39 2.5.4. Instrumental Profile 4l 2.5.4.I. StatemenË of the Problem 4L 2.5.4.2. Effects of Plate Defects 42 2.5.4.3. Ef fect of a Finite Field of View 44 2.5.5. Srlmmary 45 2.6 Choice of Operating Parameters 47 3. TI]E IIIGIT RESOLUTION FABRY-PEROT INTERFEROI'Í!]TER 52 3.1 Introduction 52 ala Optical Flats/Plates and Reflective Coati-ngs 53 3.3 Parallelism Control 53 3.4 Separation Control 54 3.5 General Structure 56 4 THE LOI^J RESOLUTION FABRY-PEROT INTERFEROMETER 57 4.1 Design and Construction of the Low Resol-uti-on l¡abry-Pero t Interf erometer 57 4.1.1. Intro<1uctÍon 57 4.I .2. Design Concepts 5B 4. r.3. Optical Flats/Plates 60 4.L.4. Mechanical Details 60 4.L.s. Piezoelectric Ceramic MounÈs 62 4.I.6. Temperature Compensation 64 4 .L .7. Plate llountings 65 4.2 Desígn and ConsÈruction of the Etalon Enclosure 66 4.2.L. Origínal Design Concepts 66 vii 4.2.2. General Enclosure Description 66 4.2.3. Temperature Control. 67 4.2.4. Inner Chamber 6B 4.2.5. Oufer Chamber 69 4.3 Scanning and Parallelísm Control 70 4.3.I. Introduction 70 4.3.2. Fringe Viewing System 7L 4.3.3, Electronic Controls 7I 4.4 Operation an.d Performance 73 4.4.1. Scanning the InterferomeÈer 73 4.4.2. Instrument Profile MeasuremenEs 74 4.4.3. Settíng the Order 75 4.4.4. Piezoelectric Ceramic Characteristics 75 4.4.5. Finesse Measurements 76 4.4.6. Parallelism and lfean SeparaEion Srabi-lity 76 5 THE DUAL ETALON FASRY-PEROT SPECTRO}TETER DESIGN, CONSTRUCTION AND OPERATION 7B 5.1 Introductíon 7B 5.3 The Optical System 79 5.3.1. The OptÍcal Configuration 79 5.3.2. Mechanical Details of the Couplíng System B1 5.3.3. Alignrnent Procedure 82 5.3.4. The Interference Filter B3 5.4 The Períscope Bl+ 5.5 Photon Detect,ion 86 5.5.1. The PhoEomultiplíer B6 5.5.2. Digital DetecEion B6 5.5.3. Analogue Detection BE 5.5.4. The Combined DeÈecÈion SysÈern B9 viií 5.5.5. I'fonitoring the Sígnal Levels B9 5..6 Data Accumulatíon and Associated ElectrrrnÍc. Controls 90 5.6.1. Introduction 90 5.6.2. The Scan Generator 90 5.6.3. The }lultichannel Analyser 9T 5.6.4. The Data Acquísition lìoutine 92 5.6.5. The System Configuration 94 5.6.6, Data Handling 95 5.7 Operating Procedures 96 5.7.L. Tuning the Etalons 96 5.7.2. Achj-eving Scan Synchronism 97 5.7.3. InÈensity CalibraLions 99 5.7.4. Instrument Profíle Measurements 99 5.7.5 The Use of Polarízer f-or Background Di-scrimínation l0l TABLE I 103 6. DATA ACCUI'TULATION AND ANAI.YSIS 106 6 .1 RedefÍnition of the Instrument Profile 106 6 .2 Data Accumulation to7 6.2.I. Digítal and Analogue Detectíon IO7 6.2.2. Gaussia¡r Line Profile i11 6 .3 The Dayglow Spectra TL2 6.3.1. Introduction Lr2 6.3.2. Isolation of the Emission Feature 113 6 ..4 Data Analysis : Theory tt7 6.4.I. Statement of the Problem LT7 6.4.2. Analysis Schemes llB 6.4.3. Convolution and Applicatíon of the Discrete Fourier Transform 119 6.4.4. Description of the AnalysÍs Scheme L22 lx 6.4.5. The Least- Squares Fitting Routj-ne L2.4 6.4.6. Analysis of T\rilight Data T2B 6 5 Intensity Calíbrations I29 6 6 Data Analysis : Implementation 131 6.6.I. The Instrument Profile 131 6.6.2. The Data Analysis Programme t34 6.6.3. CalculaÈions of the Residuals r36 6.6.4. Mean Separation Drifts : Analysís r36 6.6.5. Wind Velocity Determination 138 7 . NI.]MBRICAL SIMULATION OF THE OBSERVATIO}trAL DÀTA 140 7.L Introduction 140 7.2 SÈatistical Errors and the Power Ratio 141 7.3 Variations in Instrument Profile Shape r42 7.4 Simulation of the Data r43 7.4.1. IntroductÍon 143 7.4.2. The Instrument Profile 144 7.4.2.1. The Tnterference FilEer L44 7.4.2.2. The High and Low ResoluÈion F.P.I.'rs t44 7.4.2.3. The Dual Etalon E.P.I. L44 7.4.3. fire Slcy and Solar Spectra t45 7.4.4. The Recorded Spectra t46 7.5 The Analysis Scheme r47 7 .5.I. Adjustment of the Scaling Factor L47 7.5.2. The Ring Component 148 7.6 Spectral DisÈortio¡rs 150 7.6.1. Scattered Light from the Periscope 150 7 .6.2. Atmospheric Oz Absor:ptíon Lines 151 7.7 Summary L52 B. OBSERVATIONAL DATA 153 x B .1 Introduction 153 8 .2 Neutral Thernospheric'Iemperatures 156 8 .3 Neutral l¡Iind Velocities L6L B .4 Emission Intensítíes L63 B .5 The Ring Effect l-66 9, CONCLUDING REMARKS ]*69 APPENDIX I PIEZOELECTRIC CERAI'IIC COEFFICIENTS !7L r.1 Introductíon L7I r.2 Creep L72 r.3 Hysteresís and Lí-neariÈy L73 T.4 Variation of Coefficients 174 r.5 Effect of Non-Continuous Cyclíng 175 r.6 Summary t76 APPENDIX II PULSE COUNTING LOSSES t7B APPENDIX III - OBSERVATIONS OF THE O2 ATMOSPHERIC ABSORPTION 181 LINBS APPENDIX IV - POLARIZATION OF SCATTERED SUNLIGIIT LB2 IV.1. IntroducÈion tB2 IV.2 Analysís of Lj-nearly Polar:ized Light r84 IV.3 The Use of Polarizers ín the Dayglow Experiment 1 84, IV.4 MÍnimisation of Spectral Dístortions 185 APPENDIX V CORRECTIONS FOR BACKGROUND INTENSITY VARIATIONS 186 APPENDIX VI EXAMPLE OF COMPUTER DATÀ ANALYSIS PRINT OUT 189 BIBLIOGRAPITY 190 1 CHAPTER I INTRODUCTION 1 1 Optical Radiation Sensing of Therrnospheric Temperature and tr{ind VelocitY Until recently, observations of the optical radiation emanat.ing from Ehe upper aÈmosphere have cont-ributed little to the present state of our unclerstanding of the thermal and dynamical behaviour of rhe thermosphere. The bulk of such informatj-on has been provided by satellite d.rag, orbital inclination and mass spectrometer c1ár.a, lviCh incoherent scatt.er radar making contributions in recent years. In the past, the main emphasis for observing radiati.ons from the upper atmosphere was the idenËification of Èhe radi.ation sources and the explanation of their excitation processes. Using low resolutioD. devices such as photometers, this study is norv at an a.clvanced stage. With recenL irnprovemerits in the sensitívj-ty and rel-iabílity of high resolution photoelectric specËrometers, successful observations have been made of Ehe emíssíon line of aËomic oxygen at a r¡/avelength of À630.03nm, yielcling estimates of temperatures and wind velocities characterisÈ.ic of the neutral thermosphere. Some of these observations have been conducted over extended periods. (Bower 1974, !trilksch 1975, Hernandez and Roble L976) " The OI emission lines, resulting from Èhe forbidden transitíon rD -> ís the rnore intense and has been extensively studied over about Èhe Iast twenty years, mostly wiEh photometers. The rD sÈat" of atomic oxygen is excited by the followíng mechanisms. (a) dissociative recombination, or* + e + o(rD) + O 2 + \^/here ttre 02 ions are proclucecl l-ry l-Tre charge exchange reaction, o* + oz -,oz-F + o (b) photo clissociation of molecular oxygeû by solar radiation ín the Schumann-Runge continuum (Àf35 - Àl75nm), 02 + hv -> o(rD) + o (.) collísions with photoelect.rons, either local or conjugate (d) excitation by thermal electrons. This is expected to contribute only a small amount under normal conclít j-ons. (e) collisions with high energy particles, giving ríse to such phenomena as aurora ancl SAR arcs ' The excited state has a long lifetirne (110 secs.) and is cluenched mainly by collisions with molecular nítrogen. Consequently the emission only becomes significant above about 200krn where such collisions become infrequent. Theoretical considerations ancl rocket fligtrt data locate the emissíon layer peak at about 250 lcm near trvilight and recent studies by Roble, Noxon and Evans (L976) irrdicate the peak i-s at about 200km duríng the day. At these alLitudes ttre collision mean free path is large and so the temperature and wincl velocity gradients are smal1. (J¿rcchía I9-ll). l'he lifetíme of the atomic oxygen excited state is sufficiently long to enable the excited atoms Eo attain thermal equilibrium r¿ith the neutral atrnosphere before radiating. Consequently a measuremenL of the tenperature of the emitting atoms yields a resulÈ Lh¿rE is represefitative o,E a broad regÍ-on of the thermospl'rere-. It has been demonsÈratecl thal the temperature derived frorn Ehe emission line is basi-cal1y the local temperature at the peak of emission (Roble e'b aL. L968, Hernandez eL aL. L975) and this value closely follows the exospheric temperature . Bullc moËion of the emitLing atoms results irr a doppler shifted emission line. Although this shift is small, it is resolvable with high resolution spectrometers. l{easurements of thís shift provide estímates of the wind velocity at the emissíon height. This value ís 3. agai-r expected to l¡e representaf-ive of a broad reg-ion of the therrnosphere. ObservaEj-ons of the OI emission at À630nm ¿lre of importance to the study of the upper aËmosphere because Ehey yielcl direct nteasurements of the temperature and rvind velocity. That is, the results are not obtaj-ned by having to assume some atmospireric moclel in order to interpret the claca in ternts of temperature and ç'rind velocity. Mány aspe-cts of the OI e-mission and of thermospheric behavíour mentioned above arrd in t-he next fer,r sections are díscussed in more det.ail (with appropriate references) in Chapter B. I.2 Observations of the À630nm [or] Airglow' Ground base-cl observations of the À630nm airglotv to determíne teÌr,peratLrre (e.g. tsiondj- and Fe-Lbelman 1968; Hays, Nagy and Roble L969; Cogger, NeJ-son, lliondi, Élake and Sipler L970; Armstrong and Bell 1969) an (Blaniont and Lut--on 1972; Blamont, Luton and Nisbet 1974) has produced a l-arge anìount of data on the temperature of the sunlit thermosphere, there have been no comparable ground based studies using the À630nm airglow. If one is ínvolved in the study of the response of Èhe atnosphere to the driving forces providcd by the electromagnetic and corpuscular radiation of the sun, serious limítations are írnposed if information in unavailable for that period of the diurnal cycle when the al-mosphere is experi-encing direct excítation by the sunts eJ-ecEromagnetic radiation. In the past t.his was the case with ground based observations of the thermosphere using À630nm emissions. Now it is arguable that if such information is available frorn satellite measurements, then it is not íinportant that the effort be made to develop ground basecl 4. observatíonal techniques. Vlhile the pros and cons of saEellite ancl ground. based observations can be pursued, a decision t-o obtai.n ground based observaEioo of the sunlit thermosphere can be jr-rsEifÍ.ed si.mply on the basis of a good scientific practice. Namely, ttre results of one e-xperimental technique should be cross checked with another independent technique. Ground basecl observations of the airglor¡/ offer such an alternative. In fact, satellite, rocket and ground based measurements should be consiclered complementary to each other, as each provides some information not obtainable by the others. High resolution studies of the 0I emission line at À630 nm duríng the day using ground based techn-iques offer the iollowing possibilities: (a) as previously mentioned, the OI emission pro.rides a direct measurement of thermospheric temperature and wind velocity; (b) whereas sarellit.es carì provide an extende.cl geographical coverage, the sampling time at any one location j-s limited to the orbi¿al per-iod of the satellj-te. The ground based sampling time is only limited by the sensit.ivity of the instrument and so is more likeJy to be useful for studying tv/iligtrt phenomena and the sudden onset of rnagnetic storms; (") as of this time, no níghttime information has been obtained from satellites because the OI emission is too we-ak durÍ-ng Èhis period to be cletectecl with the current instruments. The successful development of ground based observations during the day would provide a complete diurnal coverage. (d) as of thj-s time, no direct observations of the wind velocity have been made from sate11ítes. Such measurements are possíble from the grouncl. 5 1.3 Dayglow Observations 1.3.1. Introduction. Many features of the claygloÍ{ spectrum have been detected and identifíed ín recent years by experimenters usÍng rocket, satellíte, balloon and glound based techniques. A discusisÍon of these results can be found ín the reviews by l^Iallace and McElroy (1966), Noxon (f968) and Llewellyn and Evans (1971). Ground based observations of ttre dayglor¡r are made difficult by the overwhelming background of scattered sunlight prese-nt .ln the day sky. tr^lith sufficient resolution and spectral isolation, the radíation of atomic oxygen is, in principle, detectable witl-r present photoelectric spectrometers. However, there are fe¡,¡ experiments that have detecÈed the À630nm line and none of these have yie-lded rvind velociÈy results and none have had the reliabiliEy or sensitivíty to produce tenperature measurements that can be meaning.Eully compared rvith the resr-rl Es of atmospheric models or satellrlte data. Before several of the clayglow experiments are discussed, iL is instructive Èo consider the proble-ins associated with ground base I.3.2. Bxpected Values of Temperature, tr'Iirrd Velocity and Enrission Intensity. From the results of satellite data, íncoherent scatter radar, atmospheríc models and nÍghtglow observations, the temperature of the thermosphere duríng the 1300oK with a daily variation of about 200oK. The resulting width of the doppler broadened emission line Ís such that spectTometers with resolvances in excess of about 1.5 x lOs are required to measure the temperaËure. Increased ion drag on the neutral components of the thermosphere 6 result in dayEirne wind velociLies being smal-Ier than Èhose observed at night. The daytime velocities are- expected to be aboul 50ms-1 maxírnum. The more relíabJ-e of inte-nsiby measurerûents Índicate intensities in the range I to 10 ¡¡¡;c, although values higher than Lhese have been observed (Noxon L961+). Recent theoretical studies have indicatecl that a'more likely range is betrveen 1 and 5 kR. 1.3.3. The Background of Scattered SunlÍght. O.bservations of the day sky in wavelength regions near 630nm have ínclicated that the intensity of the scatteretl sunlighE is exper:ted to be about 5 x lO4kn .toJt. (Noxon and Goody 19623 l,Iille-r ancl Fastie 1972). Although the OI emission is one of the brightest features of the visible dayglow spectrum it contributes only a small fraction of fhe Eotal radiatíon received from a cl-ear day sky. Some degree of discrirninaÈion is achieved by limitíng the specEral band wiclth of the. specÈroìneter. For a band width cornparable to the wídth of the emission line, the emission line contribuEes between 17" ar.d 2% of. ttre total signal from a 5 x 104tn nm t background. Clearly Ehen" the maj-n difficulry ín "ky observing the OI emíssion line during the day from the ground is the isolation of a weak feature from an oven+helming baclcground and beíng able to achieve this such that Ehe statistical errors of the têmperature and velociEy estímaEes are small enough to permit meanÍngful comparísons with Ëhe results predicted by various atmospheric model-s. This process is further complicated by the spectral structure of the background. The majority of rhe sky light is Raylei-gh scattered 'tÏhe Rayleigh, denotecl R, is defÍned as a unit of radiance or äPparent surface brightness where 0 101 -l 2 -t 1R photons sec nì SÏ 4t¡ 7" sunlight. Consequenrly fhe spectrcl fe¿rtures of the solar spcctrum (the Fr:nunhofer abscrption 1ínes) a.re t:eprocluced in the sky liglit. Figure 1.1. illustråtes thís spectral strucl-ure iir a synthetic spectrum computed for purposes of modelling the dayglow experiment. The scale of the diagram is such tirat the emission line and ihe scattereci background correspond to inteusi.ties of 4kR ancl 5 x lQakR n*-l t."pectívely. Suppose one observes the day sky with a spectrometer of high resoluti-on and good spectral isolation (i.e. r¡/ith litÈle sígnal- origirrating from outside the principal bandpass of lhe spectrometer) then the OI emission line will appear near the bottom of ¿ srnall Fraurrhofer 1ine, This absorption line is due to atomic oxygen ín the solar photosphere an<1 is about 1.1 x lO-2nrn wide and aborlt 47" deep referred to rhe loca-l continuum. (Delbouille, Neven and Roland 1973). The e,mission line Ehus appears in a region of corrsiclerable spect-ral Stlucture which must be retnoved in order to examine the spectral details (the r'/idth ancl wavelength) of the emissiorr 1ine. L.3.4. The Ring Ef fect. The Ring effecL (Grainger and Ring L962) a.rises from Ehe presence of a continuous cotnponent in the sky lighf. This component has been identified as being associafed with the lower atmosphere and is unpolarized. (Noxon and Goody 1965). At present, Ehe source of this component is uncerEain but it has been observed with values up to a few per cent of the sc.attered sunlight intensity. Consequently, the sky and solar spectra at À630nm díffer not only by the presence of an emissíon line but also by the presence of the Ring conponent. In a dayglorv observation, the Ring effect manifesÈs itself as fol1ows. Suppose ttre spectral structure of the scattered sunlight j-s to be removed by the subtractíon of a suitably normalised spectrum of direct sunlÍght ancl that. the slcy light has a Ring component p.."",tt which has a magni-tude a, expressed as a fraction of the sky light FRAUNHOFER ABSORPTION SPECTRUM l.o OI FeI ScII TiI +- Øz tr, 2 Atm ô tí o.5 U' = Atm É Fel \ Fel zo Atm O, FcI OI Do¡t3i+rr I o.o 629.5 630.,O 630"5 WAVELENGTH (nm) Figure 1.1 The couplex spectral. structur:e of the skylight ín the region of 1,630 nm is iLLu-strated by a s)'nÎ:hetíc spectrum conrpuled for use ln the numeríca1 rlodellí.ng of Ëhe dayglow experiment. The intensiEy scale repïesents a 501000 kR/nn sky background and a 4 kR oxygen emissíon 1íne at nm, these beíng typical. of values observed at 600 to the zenii-h.^630"03 rr.o continuum intensity at ).. (Figure 1,2), Af ter t-he solar spectrum has been normalLzed such that íts ve.lue at À" ís equal to the value of the sky spectrum at À., the Fr¿unhofer lj-ne in Ehe slcy spectrun appears less deep than in the sol-ar spectrum. Upon subtractíon, a posítive feature resrrlts. The presence of this feature wJ-ll conpllcaEe the anal.ysís of any emission line that is isolated by the subtraction process. It can be seen that Èhe deeper Ëhe Fraunhofer 1ine, the more pronounced is the subtract.ion result. In subsequent discussions of the Ring effec.t, the ccntinuous Ring spectrurn will be referred to as the Ríng component and the spectral shape resulting from normalization and subt,raction of the solar spectrum v¡ill be referred to as the Ring spectrum. 1.3.5. Doppler Shifts of rhe Solar Spectrum. The velocity of an observer on the earth relative to the sun varies continuously during the day and the velocity of the sunts approach is v = -rürcos (0) sin (h) cos (6) ( 1 . 1) r+heie r and o are Lhe radíus and angular veloeíty of the earth, 0 is the laÈitude of the observer and h and ô are t-he hour angle and cleclÍnatíon of the sun. Thus the solar spectrum is continuously doppler shifted with the maximum rate of wavelength shift occurring at local noon. The implications of this doppler shift are dÍscussed in section 1.5 . 1.3.6. Absorptíon by Atmospheríc O2. In the region of À630nm, there exist several absorpt.ion lines due Ëo atmospheric molecular oxygen and r^/ater vapour. (nabcock and llerzberg 1948; Moore, Mínnaert and HouEgast L966). The presence of these lines means that Ín general, the sky and solar spectra are not identical at these wavelengths (exc1-uding Êhe Ring effect). -The absorption lines of Oz are more interrse than those of HzO near À630nm (N} SOLAR SPECI-RUM SKY SPECTRUM (B} I,O 1 À" t-- !'- za zrn Lrl frj zF t-z RI¡$ COMPONENT o.o o.o WAVËLENGTH 5CALED SPECTRA (CI F. iõ z_ lri 2 .--lDl o.o Fig. L.2 The Ring effect This effect ar"ises from an unpolanised, continuous component pnesent in the sþlight. The sky spectnum in the ::egion of a Fnaunhofen line is ill_ustnated in (B) and has a fnaction of i-ts continuum va-lue added as a gney spectru.m. Upon nonmalizing the solan spectrunr(A), at a wavel-ength, À", the sky F::aunhofer'l_ine is less deep.than the soLan liner(C), and rleveal-s a positive fea"Lure upcn sub-. tnaction, (¡). This effect has been obsenvecl right across the visible spectnum and infonma'Eion on the behaviour of the Ring effect is obtained as a bypnoduct of the analys.is of À630 nm dayglow data. 9 ancl so are potentially more deLrimental to any daygl-ow observaLion. If the bandpass of the spectrometer is not srrfficiently fr:ee of sidebarrd transmission, rleakaget o:E inforrnetion from the region of the absorptíon lines can result in spectral distorr.ions upon subtraction of a solar spe-ctrum. This problem ís discussed in more detail in sectíon 7 .6.2" I.4 Previous À630nm Dayglorv Observations The first successful grouncl based observation of ttre ).630nrn daygloiv was performed by Noxon and Goociy (1962) using a spectral scanning polarimeter. More extensive observations were later reported by Noxon (L964). The üechnique used relied on the change of t-he clegree of polarization as tire spectrometer.was scannecl across the emission line. This clrange in ytolarizat-ion was due to the facE that rhe daylight was partially linear polarized and the emission line rvas unpolarized' No observation of Ehe dj-rect sunlight !ùas requi.red, The spectral banclr,ridth of the spectrometer lvas 0.lnm and consequently only emissiorr intensr'-ty datawere obtained. Reported intensities varied fron 5 to 50 lcR, the most conmron being in the range 5 to I0 kR. Large varÍations rvith periods of days or even hours r,^/ere reported with no noticeable magnetic activíty. Successful temperaLur:e determinations r'rere claimed by JarreL ancl Iloey (1963, 1964). These results were obtained by photographing the sky through an F.P.L and interference filter. Subse-quenL observaCions and criEicísms by Bens, Cogger and Shephercl (1965), Cogger and Shepherd (1965), Flenderson anrl Slater (1966) and Noxon (1968) pointed oul t-he inacceptability of these results. In a somewhat naive approach to a complicated probl-ern, Jarret ancl Hoey failed to make Ehe one ot¡servation that is so important in dayglow observations; namely an observation of the direct sunlight to see if the feature iclentifíed as the ernission line is still prese-nt. The importa+ce of the Jarret arrd Hoey ejxperinrenÈ 10" r¡/as thaË it macie experimenters niore a\.raÌte of the possibiliÈies for misinterpr:etation of dayglo¡¡ data. The first high resolution experimeuL to isol,ate the À630nm emission line was reporLed by Bens, Cogger and Shepherd (1955). Using two F.P.I.rs ancl an iLrterference fj-lter in series, a spectrum of Èhc day sky was recordecl and comparerl wíth a spectrum of ttte direcE sunlight. Upon subtracEion Ehe emission line was isol.¿rted" Using a resolution of 86000, they reported intensities of 6 to 5C kR and the exanple of the emission line profile presentecl inclicated a temPerature of 17001 7500K. The evolution of this e: The authors labelled their results preliminary. Using a Pepsios instrument (three F.P.I.ts in series), Barmore (I972) ísolated the emission line by numerically comparing trigh resolution spectra of the clay sky r¿ith a spectTruo of The data analysis ascounted for the doppler shifÈ of solar spectrurn ancl the Rirrg effecc" ÉIov¡ever, the signal to uoÍse ratio of the resulËs r¡¡as such that e-ven after averaging many observations, the temperature v/as only deternined to \^/Íthin +450oK for a solar zenith ang1.e of 55o. Barmore l:eported íntensities of 6 to I kR for solar zenith angles of 50o to 600. The intelsity decreased srnoothly to about I kR at a solar zenith angle of 95o. The observed magnitude of the Ríng effect was in the range of 0.2% to 2,0''Á. These results rrere later publíshed (Barmore 1975). Thís experiment yiel-ded no rvind velocity measurements because the observatíolr.s v/ere rnade in the zenith. Those experiments discussed above represented Ëhe 'fstate of the artrr at the time this work began. Clearly large improvements lüeïe required, both in the relj-ability of instrumentation, and in Eerms 11. of the statisÈical erro::s associated with the tempeïaLu-.ûe measut:erÍ'.errts. 1,5 The lfarvson Institute Dayglow Experiment. The dayglow experinent reportecl here was inítiated at a tíme when a 150mm aperLrrrr: F.P.I, Tr'/as beíng developed at the Þfawson Institute. Thís instrument ï¡/as used for observations of the À630nm nightglow to obtaín estimates of thermospheric temperature and r'¡j-nd velocity. (Bower I974, l{illcsch 1975). The projecÈ undertaken by Èhe author rras to extend the capabil-itíes of this instrument to permit measurements of temperature anrl wind velocit-y cluring the d:ry, tht¡s giving this laboraxory the capability of mouitoring the thermosphere over the full diurnal cycle. The Mawson experimen-t ríes based ott the work of Bens, Cogger and Shepherd (1.965), despiËe some scepticisrn about the potential oE this mettrod" (e.g. Noxon 1968, "It is likely that the resul.ts obLained by Bens et aL. are aboul- as good as can be hoped for from this meEhod".) It r¡as felt rhac the rationale of the Bens et aL. experiment was sound and that the dayglow line coul-d be reliably observed rvith two Ir.P.I.ts in series if the initial- experimental t-echnique was improved and extended. Basically the experir¡ental technÍ-que is as follcv¡s, Ttre background of scattered sunlÍght, and in particular its spectral structure, is removed by recording a high resolution spectrum of direct sunlighc over the same spectral interval as the recorded sky sPectrum. The solar spectrun is normalised so thaL at some hravelength À", the solar spectrum is equal to the sky spectrum. I'lhen the solar spectrua is subtracEed, the emíssion line profile remains. The wavelengEh À" is chosen as beíng representative of the local continuum near Èhe Fraunhofer lines of ínterest (e.g. see Fígure 6.5.). Bens e'b aL" achieved the solar spectrum normalisaËion by attenuating the direct sunlight until the solar signal equalled the slcy signai aL Àc. This procedure is considererL unsatisfactory f.or three reasons. 12. Firstly, the intensity of the slcy backgrouncl is l-Llce-l.y t-o chetnge cluring the course of the observatiorr, ,n.ki.lg the nornLalisal-ic¡n .incoïrect. Seconclly, the maximum emission line si-gnal will only contribute about 17" oL Ehe total signal. and to successfull-y anal1¡se the dnta, the line must be isolated liith any re-mnant Fraunhofer structure being less than about 5 x IO-2% of the total signarl at Ehe emission wavelength. It j.s unlikely t-hat this precision could be ¿rchieved using Ëhe technique of Bens et aL. Part of the intensity variations reported by these authors was probably caused by iu.complete subtraction. Thirdly, by the nature of the subtractíon process, the statistical fl-uctuat¡'-ons on the solar spectrum contribute Eo the fluctuatiolls on the subtraction resulE. By attenuaEing the sunlight, this effect is errhanced. A tnore leliable rnethod of remciving the backgroun normalise the solar spectrum during subsequent data analys:is. Horvever, there are t,{o proce-sses that can result in incompleEe srrbtractíon ancl spectral distorEion. Fírstly, srrppose the solar ancl sky observatíons are separaLed by some frlnite l-ime interr¡a.t. During this time the absor:pLion lines are doppler shifted by a change in Ëhe relatíve velocity of Ehe sun and earth. If the two spectra are norr subtrarcted, spectral structure sirnilar to the dif fere,ntial of the spectrun is Íntroduced. The recorded spectrum resulting from a scan through the OI Fr:aunhofer line with a dual F.P.I. exhibits approximacely a 3"/. signal variation (Figure 6.5). From equation (1.1.), the maxímum rate of wavel-ength shift Ís about 3 x 10-anm per l0 ninutes cluring the summer at a latitude of 35oS. If the sky and solar spectra were observed 10 minutes apart, a resitlual of amplj-tucle abouÈ 5 x IO-2% of che total signal would resul-t. Thi-s r¿ould be detrimental to the successful anall'sis of the emission line profile. For a sample interval of I Co 2 mínutes, the level'of distorEion is acceptable. 13. Seconclly, l.ong ír-rte.rvals betr¡een obse-rvat:ions can also resuLc in Èhe iniroduction cf sinilar distol:tions clue to drifts j-n the wavelerrgth calibratj-on of the spectrometer, smalI changes in the banrlpass characteristÍcs *fl" spectrometer clrif wavel-ength "f and Ls in che of maxintun LransutitL¿,Lnr:e of the ínterference fílter. tiith the paranreters chosen in thi.s experi-ment, a change in the mean sepâraLj-on of the À/:00 hi.gh resolution F.P.I. of at L630nm in l0 minures would result in a dístortic¡n with an arnplitude of 5 x l0-2i(. Ttre published resul-Es of Bens et aL. (i965) are reproduced in Figure 1.3. The large ¡:scil-l¿rLions in the wings of the emission line pr:ofile could be the result of the above vravelength shif t effects. ('Ihe tirne bet-r,reen observations r,ras not published. ) In the author t s experience, such structure rnrould make the subtraction rcsulÈ unacceptable as far as any analysis is concerned. Bartnore (I972) overcane the problem of the doppler st'rif tíng of the Fraunhofer lines by scanning over an interval of 0.l8nm sr;l that the recorded spectrum inclurJed the iron (Fe) Fraunhofer line at À630.15lnm. the clata analysis scheme adjusted che posi.tion of the solar spectrum orì the wavelength sc¿rle until the Ire i.íne in both spec.tra \rere coj-ncident. Horvever, to avoid dísÉortions, this has to be achieved vrith hígh precision. In the Mawson experiment, several scans of ttre stcy spectrun are accumulated in a signal- averager follor.recL by the accumulation of several- scans of ttre solar spectrum. This sequence has a period of 1 to ?. ninutes and 1s repeated cont'inuously, the results of each sequence being acirted to the accumulated results of the previous sequences. This proce.dure permits extended periods of The solar spectrum is obtained by observing a cliffusLng sc::een illurninatcd by the sun and is about l0 Lines as iniense irs the sky lighÈ. conse<1uently, it is advantageous to make more scans of the sky than of BO 1.0 Sky speclrum 60 E .98 _40o ÞC i; 20 96 (o) 6300 0 6300 2 6300.4 \ 80 Solor speclrum 1.0 60 Ec € .98 ',i 40 6 ! cú c, tn :UI 20 .s6 P E (b) zo 6300 0 6300 2 6 300 4 \(A¡ 5 @ .01 OlôÉ\ , ¡ I t , I ro o I , I { II t ¡ o É , ¡ 5 I I ô o , I o E I c o o I o 9 o ott Ø (c) 0 6300t O 6J0O.e 6300t 65004 6500t ¡rÅl o 5 o Fígure 1.3 The recorded spectra from the Bens, Cogger and Shepherd (1965) dayglow experíment. SpecËra (a) ancl (b) represent scans through the OI and Sc II absorpÈion lines in the skylíghf and the dírect sunlight respectively. Spectrum (c) is a subtraction of (b) from (a). The cmíssion line ís obviously present and the dashed oK curve represents the result expected for a 1700 emission line. 14" the sun per sequence. The Lwo Ir.P.I.ts are mec,hanically scanned using piezoel-ectríc ceramics and so it is possible Ëo complete each scan in about 6 seconds as opposecl to a minute or ntore with pressure scanned F.P.I.rs as used by Bgns et a'1. (f965) ancl Barnore (1972). Being able to scan the spectrum quickly reduces the effecËs of backgrouncl intensiry variations (important at. twilight) and ít allows a greaLer number of scans Èo be accumulated per observational period l¡hj-ch results in the random fluctuations of sliy brightness being more effective-Ly averaged to zero, The results are expected to be r-rltimately lirnited by the sEatistical fluctuatíons on the data and so it ís important that as much time as possible is spent accumulating useful daÈa. To maximise chå efficiency of this experiment, the v¡avelengEh interval scanned is limited Eo 3.6 x 10-3nm. This is just sufficient to scan through the OI FraunhoÊer 1ine. As illustrated in Figure 1.3, Bens et aL. scanned over an interval of 6 x lO-3nm and so íncluded the ScII absorption line aE (Note, publíshed wavelengLh scale is íncorrect (Figure 1.3).) ^630.068nrn. 1.6 Summary The rvork reporEed here was basically undertaken to demonstrate conclusively that a spectroneter consísting of trvo F.P.Lf s ín serÍes coulcl be used to reliably monitor the lhermosphere during the day. The isolation of the weak emission líne frorn a 1arge, spectraj-ly comple:<, time varying background requires híghly stable instruntents and sound experimental techníque. The precision required in this experínent is not conmonly required in other spectroscopic studies of the upper aËnosphere. Consequently, the contenËs herein are biased torsards a detailed description of instrumentaE.ion, experimental technique and data analysis. 15 CIIAPTER 2 TTIE DUAL ETAIO}I T'ABRY-PEROT I NTER}.E ROIVIETE R TIIEORY 2 .1 Inl-rodtrct ion The ttreoreLícal descriptiorr of a polyeËa1on F.P.I. presented in this chapter is developed from the basic concepts of specErometry via the single etalon theory. Specific ernphasis is placed on the dual- etalon F.P.I. The interactions of various Fabry-Perot spectrolneter parameters are complex and a conpleEe analytical description of ttre F.P. spectrometer is not presented. Rather, the theory presented íllustrares the principal gtrídeline-s by which an F.P. spectrometer ls designed. Comparison of varíous spectrometers (secEion 2.3) led to the selecËíon of a clual etalon F.P.I. for the study of Ëhe À630nnr daygl-orl. However, the theoretical description of a single etalon F.P.I. ts inclrrded here in sone deEail Eo illustrate the interaction of various F,P.I. pal:ameËers and their effect on the performance of the spectrometel" The desígn considerations of a sÍngle etalon F.P.I. are applicable in modifÍed form, to poly-etalon F.P.I.fs. In the particular dual et,alon configuration chosen most of the specÈromeËer propertÍes are controlled by only one F.P.I. The Èheoretical clescripËion of a single etalon I-.P.I. has been presenred ín deuail- by Chabbal (1953, 1958). Jacquinot (1954,1.960>, ïli]-l (1960), llallik (1966), Sroner (1966> and llernandez (1966) have also contril¡uted to l-he theory. The theoreLical description presented here is based on the \,rork of the above authors. t6. Recently l^Iilksch (I975) presented a resËatement of the Eheory using the order of interfeïerrce as the indepenclent variab.Le as opposed to the moïe conmìon varj-¿rb-Le, r,lavenurnber. Although the use of wave- number has some advanËages, iË is felt that waveJ-engËh is more comrnonly used in the dcsign aspects of 11 P. spec.tromeËers. Consequently tlre Èheory presented here uses wavelength as the inclependent variable. chabbal (1958), Mack er a1. (1963), McNutt (1965), Stoner (1966), Roesler and Mack (1967>, Roesler (1968, 1974) and Daehler and Roesler (196S) haye presented various aspects of polyeEalon theory and clesign. The conÈents of section 2.5 is based ori these rrorks. . 2.2 The Spectromeler Transmission Profile, Re-corded Spectrum and Transmittecl Flux The instrument trarrsmission profile (someti-nes referred to as the instrument. profile) is defined as the spectroineter transmittance as a funcËíon of vral¡elength, Fígure 2.1 illustr¿rtes the instrument profile of several common spectrometers. The specLral content of an emíssioll. or absorption source is examined by varying sorne parameter such that the instrr:nent function is rsrveptt across the wavelengl-hs of ínterest. The radiaEion transmiÈtecl at each poínt of Ehe tsweept or scan (ËhaE is, at each presented to a detector, usually a photomulLiplier, which Ào' ) ís clevelops a signal proportíonal to Èhe flux. This s:'-gnal ís recorded as a function of the wavelen8th Ào. In thís cliscussion, it ís assumed that Èhe spectrometer profíle does not change its shape during the scan and Èhat Èhe source has a uniforrn surface radiance and fills the field of, víew of the spectrometer. The spectral radiance of the source B(À) i-s d.ef ined as the t I (") Ào tr (b) a) o É d +l ----) *91 ¡J 'r'{ aÉ É þCÚ ¡,0 H .r (") Ào l^IavelengÈh À Figure 2.l The instrument transmissíon profiles of several comlnon spectro- meters r¿here ôÀ- is the bandwidth at the !t- points. ((a) a single etalon FPI, (b)to dual et¿lon IPI ancl (c) ¿rtdiffrac-tíon grating spectro- meter working at a resolution defined by the slit width only). ß (À) c) (.) Ê (ü .il +J l{ oÊ É ql t{ H À6 À þ¡r ¡, ¿\z- )' -l I.gqr"¿¿ The flux at the detector, originating within a small spectral' d),, is proportional Eo the area under the curve, I(À-Ào) S(tr), -inTõaãLin rhat interval whãre I(À-Ào) is the instrumenL Profile and B(À) is the specÈral radíance of the source. L7. radiance per uni-t wavelength interv¿rl and has urriEs of photons *-2s-lsr-l (urrit wa',¡elengtl-r.)-1. trIith Ehe spectrometer tuned to Ào, the flux transmitted in a spectral element À to À f dÀ ís, dOl = sQI(À-Ào)B(À)dÀ Q.L) o and the total flux at the detector is obtained by ínEegrating ecluation (2.1) over al1 values of À where the spectrometer has non zero values of transmittance (Fi.gure 2.2) su-ch that, Ào- +A"À o(r r(À-À )B(À)dÀ (2.2) o )=Sf¿ o tro- -ÀrÀ where S ís the aperture area of the spectrorneter and fì is the sol-id angle of the fielcl of view. The recorded function Y(À^),'o ís defined as 0(À_)-'--q' (2'3) Y(À-) = u Sf) Equation (2.2) clefine.s Èhe reco::ded fuuc,Èion as the cross correlaÈion of the instrumenÈ profile ancl rhe source profÍle, denoted, in general wavelength terms, âs¡ Y(À) = B(À) * I(l) (2.4) The spectral conÈenr of B(À) Ís reflected in the sËructure of Y(À). It is convenient to define a normalised transmission profil-e I (À-À sueh Èhat, o o ), I(I-Ào) = .II (À-À ) I (0) I (2.s) o o o The flux at the Sf,ltrYo(À (2.6) 0(Ào) = o) where Yo(À) = B(À) ft ro(À) (2.7) The flux at the detector can only be calculated wíth a knowledge of Yo(À) which Ís rlependent on the shapes of B(À) and I(À). 18. However, lríth several assumptions, reasonable flux estímaÈes can be ma,]e. Fírstly, assume the spectrometer has only one transrnission peak of high contrast that contributes to the rletected flux. This conditíon can be arranged by using a pre-monochromator or interference filter if the specÈromeLer has naturally occurring side bands as does the EPJ. The filter: must Ehen be cÕnsidered as part of fhe spectrómeter. If the source consÍsEs of a narro1ll emission line, the condition ís fulfilled. naturally. Secondly, assume the spectrometerts requivalentt width is equal to the widÈh of the transntj-ssion peak aË ha1.f heíght, ôÀr. (Note: henceforth the width of any function at half height \^7i11 be just referred to as the widÈh.) Fo:: Ehe more conmon spectrometerst this is reasorrable. If the source can be repïesented as a simple line emissien of rvidth 6ÀU or a simple continuum, estimates of flux are easi-ly made when (a) ôÀT. << 6À8 (b) ôÀr >> ôÀB (c) ôÀ, 'v ôÀu Under rhe condiEion (a), the shape of Yo(À) approximates B(À) with an area ôÀr, thus, 0a(À) tu sQrrB(À)ôÀr (?-.8) If the maximum value of B(À) ís denoted B*, Èhen the maxímum flux ís, SQI_B 6À_ (2.e) 0AM "r, IM I This expression ís also valid Íf the source ís a contirruum. under condition (b), the shape of Yo(À) approximaÈes the mirror image of Io(À), % (À) tu sarrrð(À)BmôÀB (2.10) and Õbor t sQ'rrB*ôÀu (2. rr) Under c.onditíon (c) , the flux depends specifícally on the relative widÈhs and shapes of I (À) and B(À), such Ëhat o Q r sfJt-'r-B 6À- Q.Iz> cm I.6m 15 The factorr' is introduced because the cross-correlation operation broadens the resulting profile but preserves area. Hence the peak transmiEtance must decrease; tU ( l. I.t the widths are nearly equal (2.9) (2.1L) and and Ehe shapes are sj¡rilar, TB tu 0.7. Equatíons ' (2.I2) permít flux estimates for sources of interest. to the r,¡ork reporEed here, namely a line emissíon source of Gaussian shape and a quasi-conËÍ¡ruous source, the solar specÈTum. The light gathering power (or luminosity) of Èhe specErorneter is defined as the flux transmítted per unit radiance anrl for thís discussion is given as 0 r il (2. 13) -- uorôÀ" Hence ô^ L=S0r T for ôÀ, ôÀu (2. L4) I ôÀB L = sfJ'r, for ôÀr >> ô^B Q.ls) L = Sfl.rr.r' for ôÀa n, ôÀU Q.I6) The quantÍty, SQ, ís known as the 6Èendue of the spectrometert U = Sf,t Q.l7) If a spectrometer can resolve two rnonochromatic emissíons separated by a wavelength interval 6À, then the resolving power or resolvance of Èhe spectrometer ís defined as T R = ii (2.18). 2.0, \nrhere À is the average waveleugth" The .resolvíng pro.oertÍes of a spectïometer are complelely deLermíned by the representatiot-t j-È gives to a norlochromaric line. Equations (2"8) , (2.14), (2.9), (2.15), (2"10) and (2"16) define the conditiorrs uncler which a spectroneter is usecl . If 6ÀI << ôÀU, the recorded flux accurately reflects the source shape, but the light gathering poqrer ís lor¿. If ôÀI ,t ôÀU, the líght gaEheri-ng power is naximised but no ínfor:matíon is gaine exhibits a broad maximum ¿rbout ôÀ, "t' ôÀU. The L-R product is a constant for a particular spectlolneEer configuration arrd is often used as a figure of merit rvhen conpaling the performances of variorls spectrometers. 2.3 Specttometer Selection The choice of speccrometer depends on the details of the radiatiorr to þe stu f ol-lowing: (a) ease of obLaining the desired resolving porver. (b) the rejection efficiency Í.ot conEaminating sources. (c) ease of scanrri.ng the spectrum and the precision wíËh which this scan is exectrted, í,e. precision of wavelength calibration. (d) the spectïometerrs light gnthering poþrer at a given resolving poûIer. Although (a) - (c) are ímportant in most applicatíons, spectro- meEers are basically compared r,¡iÈh respect to Èheir L-R producÈ' Jacquinot (f954) compared the light gathering polt/ers of spectrometers 2I. havíng a prism or a blazed díffraction gratírrg as their rlÍspersive element wíth a FP specËrometer and found tlr.e latter to be srrperíor. Thetr'P specÈrorneter, aE a given resolutíon, has a larger light gathering potrer be.c¿ruse it can accept lighÈ from a much greaÈer solid angle Èhan the grating spectrometer. The lumír-rosíLy of a Mj-chelson interferometer is comparable Ëo that of an FP (Hunten et a1., 1967), whereas t-he wid.e angle Micheison (I^IAl'fI) greatly sulpasses the F.P. spectrometer (tt1ttiard and Shepherd, L966>. At this poinE it .is convenient to consider Ehe- spectrometer properties required to observe the atomic oxygen dayglow at À630 nrn. 'ihe lcinetic temperatures characterisric r¡f the neutral thermosphere are deterruined by Ehe estimat,ion of the width at half-height of a (2.2o) 6À=&c where l. is the emissíon wavelength, v ís the líne-of-síghÈ velocity and c is the velociLy of light. Wind velocitíes in Ehe day thermo- sphere are expecfed to be approximately 50 ns-1, giving a rvavelength shíft of I x 10-+ rrm. Thus the spectrometer mtrst be capable of a wavelength calibration accuraey better Èhan 1 part ín 107. As mentioned in chapter l, the emi.ssion line appeals ín Ëhe .) ¿.L.'') presence of a 1-arge background with a complex:¡ncl tj-me depcndent structure. Thus Ehe spectrolneter must be capab.Le of iso.l-¿:-tirr.g a narror^r spectral regíon \^rith little f Iux arising :Erom neighbouri.ng wavelengths. The spectroneter also requj-res a large light gat-herirrg pov/er because the emission feature tuíl-l be cloririnatecl by the photon noj-se of the large background unless suff:íclent photons are detecced. The Michelson spectrometer is besE suited for measuTemenEs of emission lines of known analyticaj- shape and r¡hen the background structure is easily determined. This latter condition is not satisfied during the F.P.I.ts operating in series would be required and since they couJ-d be coupled without J-oss of 6ten<1ue, the decrease ín f-ransmittance due fo their coupli-ng is smal1 enough such that thr: light gathering advantage over a graÈíng spectrometer is still maintainerl. Thus a polyetalon T"P. spectrometer \^ras chosen for observaÈions of r-he À630um tOf¡ emission during the day. 2.4 Fabry-PeroÈ Theory Single EEalon 2"4. L General Prirrciples The ideal Fabry-Perot etalon consists of a pair of flat and parallel transparent plates whích have thelr inner surfaces coated with a semi-transparent mirror of reflectance, B, absorptance A' and 23. transmittance T. The convenEional conf igur:ai-ion of an F.P.I. l-s shov¡n in Figure 2.3. Light from an e: The wavelength Eransmission characterísLíc of the ideal etalon is described by the Àiry function, -rA (2.2r) A(À) = lI?}.a,s)- À where 'r^ Ís the tran-cmission coeff ícienL, .{, is the plaEe separation, A ¡t is the refractive index of Ehe spacíng medium and Ç = cos0 where 0 i-s the angle of ínciclence betrveen the ¡rlates. For a gÍven wavelength À, the transmission is maxi.rnised when the order of interference m, ís íntegral. 2vl"t' o, = ï* (2.22) For a given 9" and Ç, the eEalon has a mu1-tiple beurdpass transmíssion profile. The spacing between transmissiorr peaks is knom as the free spectral range, a,\ \! ... (2.23) ^l - m The free spectral range is dependent on wavelength but for large orders, AÀ varies little over a ferv spectra-l- ranges. extended source 9. etalon ï focusing lens { field stop (aperture) Figure 2.3 The conventional confíguration of arr TPI. Radiation front an extencled sotrrce undergoes multiple re-flections in the etalon and is l-hen transmitted Eo the detector through a fíe1d stop. 24. If the Aíry frrnction has a wídth of ôÀA, Èhr:n the raLio of free spectral range üo wj-dth is termed the reflective fínesse, NR' For val-ues of R greater than about 0.5, t- r, (2.24) "R-ôÀ^-(1-R)--^¿--I(5i: The transmission coeffj-cient depends on the rnirror propertíes such that _ : _-y__ .A = Tt-Ðu =- rr-,,4-.1(r-R) e.z5) l't -/ I where T = I - R - A, hence the importance of keeping A srnall if R is large. The area under one order of the Airy funcÈion is knor,rr as the ordinal area AO, where 12 ^^À =- clft¡,nr Q '26) Figure 2.4 illustraLes the form of ttre Airy functÍ.on, 6Ào and AÀ. Since for an íntegral value order mo, the wavelength À, has a 2vLE-, À transmission profíle can maxjrnum transrnittance where = *o the be made to execuËe a wavelength scan by the variati.on of { (spatial- scanning) , p (pressure scannín,g if the rnedium is a gas) or .[, (separation scanning) . 2.4.2 Effect of Plate Defects Variatj-ons in the opÈícal separation over the area of the etal-on modify the form of the transmission functíon. Variatíons may be caused by surfrrce defects, lack of paral-lelism or non- uniformities in the reflecÈi-ve coaËings. If the FPI is operated neaï norm¿rl íncidence, l-he etalon can be considered to consist of a number of elementary etalons (Ctrabbat, 1953) of different separation. The fracÈíonal- otut f, of an elementary I AÀ TA o o É (Ú {J .rl+J É (/) É (Ú t¡ r *-ùÀ.0, EJ u/2 A^ 0 Ir 't,Iavelength f Figure 2.4 The Aíry function. 25 etalon l^rith a separation be{:\^reen JÙ * x and 1, * x -F dx is equal to D(x)dx, where D(x) is the uníE ârea defect frrnction, _1,i" (2.27) D(x) Sdx ancl L is the rnean etalon separation. For an elemenEary etalon of separation L, the transmitEance of wavelength À' is gíven by equation (2.2I) ancl j-s íllustrated in Figure 2.5. For an elementary et-alou o.f separation g' * x, (x > 0, say), the transmission maximun of order mo shífts to a larger wavelength. Thís is illustrated in }igure 2.5. At a separation of .Q, * x, the transmittance of Àr is equal to the transmittance of Àt .. À* at a separation of 9.. If the elernenÈary etalon of separation 1, -l x is illuminated with monochromatic radíation of wavelength Àt, iEs conÈríbution to Èhe Èransmittance of thís wavelength by the entire etal-on ls, dE¡, = A(À'-\)D(x)dx Q"zB) Definíng a urril- area defect function D(À--), as lds D(À = (2.2e) x) ----s dÀx equation (2.28) can be written as dE¡r = A(À'-trx)D(Àx)dÀx (2.30) where and dx are related via equaEíon (2"?-Z). The total dÀx transmittance of the r,ravelength Àr, is obtaíned by integrating equation (2.30). Expressed ín general terms, Èhe transmission function of an etalon rvith plaËe defecEs is E (À)= A(r-À )D(À ) dÀ (2. 31) X x x This is a convolution operaÈion and the function E(À), knor¡rr as Èhe etalon function, is denoted as (a) 4-mo <--.-mo'l separation f, T. -Àx c) CJ É (d .u ]J .rl Þ Ø À À À É (J þ H (b) separat ion .{,*x T"r t Àt Wavelength Figure 2.5 (a) fne ilstrument functiorl over orders m9 and m9-1 for a separatíon [. (b) At a separatíon .0J-:< (*>0), the orclers mo and m0-1 transmít at larger \^7ave1ength. The transmittance of Àr in (b) ís equal to Ëhe transmittance of l' -).x in (a) ' 26. E(À) = A(À)tt¡¡1¡¡ (2.32) If the width of the ôÀD = ¡¿ ut, (2.33) The defect finesse and the etalon fínesse are- defined as, ND= AÀ = Àr (2.34) ¡it 2u E ô-"D and NE= (2. 3s) q,^À respectively where ôÀU is the widch of the etalon .Function E(À). The two rnost common defecÈs are a spherical error or a misalignment of the plates. If the perfectly flat pJ-ates are out of parallel by an amourrt À/n, the defecE f inesse is *o = Zdi, Q-36) If a spherícal defecÈ of sagitEa X/n, is present, Èhe defect finesse ls, ,n No = i Q.37) If there are tr^Io or more definable defects present, the rescrltÍng clefect function is the convolution of Èhe individrrai- defect functions (lernandez, 1966) The properties of the convolution operation dictate that Lhe area of E(À) be the same as A(À), narnely A¡ per free spectra,l range. Since the convolution entails a broadening of A(À), the peak trans- mittance of E(À) must be decreased to preserve area. The peak transnittance of E(À) is (2. tE ='f'f .D "A 3B) where the value of TO is determined by the relative widths and shapes 27 " 'rO of A(À) and D().) . Chabbal (1958) investigated Èhe behaviour of N.o NR t ís smal-I, rD tu 0.7 as a functíon of Nl and found that r D I if N; Lf *; tu 1 and for *; greaÈer LharÌ one, 'rO rapi NR > ND, and Ehe etalon ls tuned for rnaximum transmittance of a mono- chronatic 1ine, some areas of the eÈalon have a transmittance of less than 0.5. This is equi'"'alent to dec.reasing the effectíve area of the etalon'and the tertn tautomaskingt (Mack et al ., 1963) has been used to describe the effect. The defect resolvance is definecl as o À Q.3s) "D= 6Ào 2.4.3 Effect of Fínite Field of Víer¡ An on axis aperture of finite radius r, samples the frj.nge pattern in the focal plane- of the focussing lens (Figure 2.3). Across the aperture, the cosine of the angle of incidence between the plates varies from l to 1- tSE. If the angle of incidence is always small, then 2 (2.4o) ô6 ^, %(ï+) where f is the f <.¡cal length of the lens. aE order m at normal If the vravelength Ào is t.ransmitted o incidence, then for an íncídence angle cosine of E, the wavelength transmitted at order mo is À +À where, o I À- tt-'1 (2.41) \ #o Thus À- is always negative. ç Each element- of the aperture varyíng from E to E + dE subtends a solid angle citrt, at the lens where dt¡ = -2ndE ,.. (2.42) 2B The unit area aperture functiorr is clefinecl as, F(q) I clul (2.42> ç¿ dE where f) is the total solid angle subtended at the lens such that, f¿ = 2nôt e.ß) The aperture fturctíon ís of a rectangular shape (Iigure 2.6). An aperture functíon, in Lerms of çvavelength, is defined as I dô F(À (2.44) ) f) dÀ E Basically equation (2.44) sfates that the fraction of the solítl angl-e do ä, that transmits wavelengths À, to ÀE -F dIE less than the wave- length transmitted at normal incidence í-s equal to F(ÀE)dÀ8. The aperture functions F(E) and lr(Àr) are related via equation (2.22.) and if rhe widrh of Fc6) is ôË, rhen rhe widrh of F(Àr) ís IÌ r^F --TôE o Q.45) Using concepLs siml-lar to those of sectíon 2.4.2', it. can be seen that if the etalon ís íll.uninated with a monochromati.c source of wavelengEh Àt, the conEril¡uEion to the transmittance of an apert-trre element of 6 to { -l- clf, ís, dr,, = E(À'-tr-)F(q)dqq' Q.46) ^' or dr¡r = E(À'-^E)F(ÀE)dÀ6 (2.47) Integrating equaËion (2.47) over all À for the aperËure and using general noEatÍon, Èhe instrument profile I(À), ís expressed as a convol-ut i.on" That is ¡ôl' F (2.48) r(À) = f n (À-Àr) F (16) dÀE Jo or I (À) = E (À) ,kF (À) (2.4e) F (E) r r-6 1 F(l F ) ôÀ, 0 E Figure 2.6 The aperture function both in terns of the incidence angle cosine, Ç, and the wavelength. 29" The aperture finesee No, is given as À o Nr AT (2. so) T_\ 2u¿(<58-1) The aper:ture finesse is only meaningful- of the pJ-ate separation Ís specified. The instrurnent f inesse N, ís def inecl as, A^ |f= t^, (2. sL) where ôÀ, is 'Lhe width of rhe instrument profile. Again the convclution broadens the profile and the peak Eransmittance decreases, such that the peak transmittance of I(À) is tr tFtE (2,s2> where'r,, is clependenÈ on the relative widths and shapes of F(À) and E(À). The ínstrument Èransmission profile is now given as I(À) = A(À)* D(^)'bl'(À) (2. s3) and tr = tAtDtF (2.54) The aperture resolvance RUr is related to the solíd angle sub- tended by the apertur:e; from equatíon (2.43) ancl (2.45) h : 2rr¿ Q'ss) By the nature of the F.P.L, the eff ective resol.vance r¿ill always be less than eiÈher RA, RF or RO, thus, 2tr R< -R- (2.56) This then places a fundamenral restrainÈ on Èhe size of the aperture Íf ttre F.P.I. is to be operated at a given resolvance. 30. 2.4.4 _9"ry. The precedíng sections have shol,;rn thaE Ehe síngle etalon F.P"l. has a multíple bandpass transmission profile. The wiclth, shape and peak transmittance of any one passband are clependent, in a complex rnanner, on the relative widths and shapes of the Aí-ry, defect and aperture functions. Chabbal (1953) made an extensive study of these Ínter- relations. which permitËed Ehe formulaÈion of several design criÈeria. Chabbal C1953) has shown that there is a broad maxímtun in the L-R product rvhen NA t ND and N, t NE" The ef fectí.ve resolvance of the spectrometer Ís then, t 0.5RD .r, 0.7\ R ^, 0.5IA Q,57) It is useful for desígn purposes to be able to estÍmate the resulEing finesse r,vhen two or more functions are convolved and although the follorving expressions relate specífically to Gaussían funccions, they provide useful estimaEes for present purposes. ôÀ! r ôÀi + orf + 6111 (2.s8) or N-2 r, *;' * rrro2 -l- llo'z Q.59) The flux t.ransmittecl to a detector when scannÍ-ng a source B(À), 1s, from equations (2.4) and (2.53)' o(À) = slì{B(À) *' tA(À)*t(r)*u,^r, ... (2.60) If ts(À) is a símple line source, the L-R product is again maximised when ôÀ, t ôÀf. However, ttris conditíon results in signifícant broadening of the recorded profÍle, impl-ying Ëhe need for deconvolution if the r¡idth ôÀBr is to be determined. The defect finesse is,fíxed by the qualíty of the plates ancl can represent a serious limítation to Èhe performance of the F.P.I' From equation (2.6), Ëhe flux and hence the light galhering power is proprotional to'r., which is proportional to TATD. Nor,r'rO can be 31. increased by decreastng Ehe absorptance of che reflective- coatings or decreasing NO. However it ís of ten desirable fo utaint.ain a high NU and so care must, be Èaken to ensure that i" not much gre.ater than $ffNo one, otherwise the transmittance of the F.P.I. is seriously affected. The fÍnesse N, can be interpreLed as the nutrber of spectral elements that can be resolved per free spectral range and because ìI < ND, Ëhe importance of NO is agaín illustraÈed. The resolvance and the fínesse are relat.ed by R = mN (2.61) Now for rnost F.P.I. ts, a f inesse of 30 is reasoital¡le and thus high resolution implies a high order and consequently ¿t small spectral range. If the specËrum to be invesËigated is complex anrl extensive, equation (2.61) implies a serious limitation on the applíca-ti.ons of a single etalon F.P.I. ThaÈ is, the ePpeer:ance of flux fr:om sídebands that cannot be easily suppressed by simple ntearrs such as an interference filter. The defeet fùnction is usually difficult tr¡ express anal)'Lically and if the shape of I(À) is required in some st-rbl;equerì.t analysÍs scheme, iE is best to determine the fonn of I()r) e:cperirnentally alEhough some authors (ilernandez, L966) have expressed I(À) analytj-cally for various sÍrple defecL frrnctions. 2.5 Fabry-Perot Ttreory : llual- Etalon 2.5.I Introductíon and General PolyeÈalon Priacì.p1.es The interaction of varíous F"P.I. parameters in a polyetalon sys'Eertr are even more complex than for a single eÈalon an a theoretical descríption is dífficult. Some aspects of polyetalorr tlreory have been presente franrer,¡orlc for the exact, forrnul-¿rticn of i:he thecry Ís pr:esented ln secÈion 2.5.4. I{ower¡e:lr the essenti-al properties of a polyetalon spectrorrreter aïe sEill r^evealed by much sinrpler cortsiderations, dretwíng heavily on the theory of a sÍ-ir.gle F.P'I. As prevíour;ly shor,m the single F.P.I. transmii:s rad-'Lation in the form of a ch¡rnnel spectrum with br¿rusrniss-Lon maxima separated by a qtrantity knoran as Lhe f ree specEral range, ÀÀ (equation 2.23). If the F.P.I. ís sc¡rmted over an order a-nd there exists radiation at wavelengths or¡tsicle the free specËral range of interest, fhe resultant re.cordecl spectlîum may be ambiguous ín its inEerpretalion. Essentially one requires the recorclecl spectrum to conEairl j.nformation derived from r,¡ithin only one baldpass. Thus the sidebands i¡ust be suppressecl an means of one or more F.P.I. in series rvith the original. Broad spectral isolaEion is provicled by an interference filter which plays an Ímportant role i.n Ehe design of a polyr:talon F.P. spectrom-eter. To isolate a banclpass at wavelengl-h ÀO, all the Ir.P. etal-ons are tuned for maximun transinj.ssÍon of this wavelerr6;Eh, buÈ the separation of each etalor: is chosen such thal thein respective sl'-cle'- balds occur at different wavelengEhs. Since the resulÈing transmission profile is approximaLely Ehe product of the jndividual profiles, there ís a large recluction in peak transmitl-ance aE wa'¡elengths r¡/heTe the siclebands are not irr wavelength coíncíderrce. Uowever, for any combínation of separaÈions, exact or near coincidences occur at some wave-lengths" Because of the fínite width of the bandpass' near coincidences calL result in large transniftarices. Consequently it is important EhaË at these vlavelengths, thê ínterJ:erence filter provides strffícient suppression or that they occur in a region free of spectral structure in the- source. ts Suppose thaü thre coml¡irration of tr,ro F.P.I. is to be. such El'rat 33. there exists no exact co-Lr,cidence ín a spectral j.nLerval from Ào. ^À", This reqtrires that ti-re free spectral ïanges of the ll'P.T.Îs den-oted AÀr an AÀ2 = AÀ"/xz Q'63) In general, the desired resoltrËion specifies AÀ1 and thus ki is set by the desired ÀÀ". As illustratecl in Figure 2.7 (for integral ratios) ¡ x2 cátfl vary from I to k1 - 1. The plate separatlon of the trvo E P. etalons are relaËed &r t<2 (2.64) I'z - kr Chabbal (1958) classes the combination as follows, lst type x2 tu 1 Also denoted as a high-low or monochromaÈic coubination. 2n as being arì extension of the Lst type) ' The Èr¿rnsmíssion profile has finite co1¡trast and so compleEe nullíficaÈíon of siclebands cloes not occur, t:he situafion being the appearalce of ghost-s or parasitic bands as illustrated ín tr'igtrre 2'B' The size ancl position of these parasitíc bands is of fundanental iuportanee to the desígn of a polyetalon IlP. spect]iomeEer. Reflections beLween etal-ons can be reducecl to insignÍf ic.ance by etther tÍlting one etalon relative Eo the other or by ínËroducing a small amount of absorptance between etalons. In the following dis- cussiofi, iE vtill be assumed thaL the ínter-etalon reflections have been elírnínated. For a polyetnlon spectroineter consisting of perfectl.y parallel etalons and operating with an infínÍtely stnall fíeld of view, the transmíssion profile is the multiplication of the Aíry frrnctions of the Âtrr (") t=5 ).¡ (b) =lcr -1 2 (c) x =kr-2 2 (d) x =kr-3 Àq 2 (e) x=1 k1AÀ1 Tigure 2.7 Principle of a polyetalon FPI. The transmj.ssíon sidebands of etalon (a) are suppressed r,¡ith a second etalon such t-haL Èhe first side- band occurs at l¡t5AÀr. The (a)-(b) combination Ís classe-d as the 2nd typei (a)-(c) or (a)-(d) as the 3r:d type and (a)-(e) as the lst type. (a) (1, ) (c) (d) I I (e) I i (f) (e) ll À¡ l'j-gure 2.B Princi.ple of a polyetalon FPI" Si-mílar to Figure 2"7 blut --Tfe païasitic sidebands due to the fínite wj-dth of Ehe ínstrunent profiles are illustrated. A combirraLion of tTre lst type, (a)-(b)' results ín (c); 2nd Eype, (a)-(d), results ín (e); 3rd rype, (a)-(f), results in (g). (after Roesler, L974). 3l¡. individual, F.P. ' s (section 2.5. 4.I) " I (À) =. Ar (I)42 CÀ) A*(^) (2.65) where the ith etalon has a reflecÈance R, and a reflecti-ve finesse, Nn.. This discussion is primarity .onJlned with F.p.r.'s operating l- at high resolvance wtrich by equation (2.61) implíes large orders of interference for a finite finesse. Thus the assumption that for a particular etalon, AÀ is constant over a few orders does not introduce any serious errors and simplífies the probtrem. The v¡idth of the polyetalon F.P.I. profíle ís smaller than the wídth of the highest resolution member. This is illustrated in Figure 2.9 where 6À is the resultant v¡idth, U^O, is the widtTr of the hígh resoluÈíon F.P.I. and 6ÀO, ís the widtll of the lor¡ resolution F.P.I. If Rr = 32-, then ôr.A'z - -&t Q'66) - 9"2 Tf.A1 Thus for a combination of the lst type, equatíon (2.66) ís equal to k¡ and usually large (5 or more). For a combinatÍon of the second type, i.t is approximarelY one. Figure 2.9 ill-ust.rates the important fact Èhat for a combination of the first type, the resolut:Lon of Èhe spectrometer ís very nearly that of the highest resolution rnember. The adclition of a thírd' even lower resolution eE¿rlon, has very little effect. Figure 2.10 ill-ustraÈes the increase ín resoluEion for a conbinati-on of the second type as strccessive eLalons are added. The existence of parisitic bands causes tleakager of ínformation from wavelengths outside the spectral range of interest. In the observatÍon of a c.ontinuous or a simple absorption spectrum, one is concerned with Èhe total flux arisi.ng from the parasitic bands, thís beíng proportional to the area under the Ltansmission profile outside 1.0 ôÀ ôÀ A1 0.5 0 9\r - r'' ôÀo, 9.2 ôÀ Tieure 2.9 The combínation of two íde¿rl etalons of r'ridth À1 ôÀO, results ín ari instrumenÈ profile of wídËh 6À' Tor a combinatíon of the 2nd tyPe, ôÀA2/ôÀAt-l and for the 1st type, (a) (b) has ôÀA2 /ô¡,Al >>1 . (" . g. , in Figu re 2 "B , combination - ôÀA-/ôIA=5ífbothetalonshavethesamefinesse). 1.0 ôÀ mA. 0.5 0 246810 Number of etalons Tieure 2.1-0 The resultanË width ô)., is illustrated for: a combination of the 2nd type as a function of the number of etalons user-l . 35 a speclfied region. To descríbe ttre perfoTmanoe of a spectrometer ín thís respect, Chabbal (1953) defj.nes a qrraliiy cal-led the t filteraget e, T #¿ôÀ o r (À) dÀ (2.67) À o-%ôÀ +æ r (À) dÀ -oo where ôÀ is the widÈh of bandpass centred on Ào, and I(À) is the spectrometerrs transmissíon profíle, including the interference filter. Ior'a s-Lngle order of an ideal- F.P., e = 0.5. The definítíon of e ís somewl-rat arbitrary but ít does provirle a means of cómparing spectrometer petformance. In the observatj-on of a complex line em-isslon spectrum or a cornplex absorption spectrum, one is concerned with the relative amplÍtudes of the parasitíc bands. As can be seen in'l-igure 2.8, kr - 1 parasltic bands arise from Ehe higher resolutior¡ F.P, and xz- I from the lower i:esolution F.P.I. and the largest parasític bancls e-xist at posÍÈíon.s corresponding to the transmission utaxima of Lhe high - n",flz- p varies from resolutíon F.P.I. They are located at À^o Kr r¿here L to k1 - 1. 'fhe relatiwe height of the pth ghost ís, Ar (Ào+p^À zlkt)Az (Ào+p^À z/lct) (2 68) h= Ai (Ào)Az (Ào) " o#) S inc.e Ar (Ào) = Ar (Ào * , then A2 (¡o+pAÀ 1) h- T lyz -l (2.6e) [1 +. "i"' ff I 36. The relati.¡e helgtrt Ís thus independent of fhr: type of conbinatl-on under consíderatíon, The stroirgest oT pr:íncípa1- ghost occurs for p = l, and so equation (2.69) can be slmplified Ëo, h = [1 + 0./,[rå2"1n2 *rr-t (2.7o) or 2_ h=\(#) kr>4 (2.70a) . ^tR, Fig. 2.11 illustrates the value of **, that must be aehieved if the princípal ghost is to be below L7" or 2% Lor a given val-ue of k1. Figure 2,12 iJ-J.:rtstrates equation (2.69) for various values of **r. If pth ghost greater than can be for a given k1 and N-llz' , the is tolerated, it can be furÈher suppressert by íncreasíng NRz. However, íf thís valu" of *O, j-s íncommensurate wíth NOr, ttrere wÍ1.1 be a decrease in transmittance whích could be greater than would be obtaj.ned by the addition of a third rrP,I. The peak transmittance of a series of n eÈalons 1s Tn'n = TA,"....TA Q.7I) For normal. values of tO (tuO.7), a rapid decrease c'f transmiÈÈance is experienced wíth increasíng n. For a polyetalon F.P. of Ëhe lst type containj-ng n etalorrs, the free spectral ranges are rel-ated as (2..72) AÀn -k n-r,AÀ n-r (n) (2 3) kt k r-taÀt .7 ^À An lncreasíng subscript denotes decreaslng sepzrratÍon. For a combinatíon of tlie 2nd type, the relations are Al -1 AÀ (2 .7 4) n (kÎ- -1) 100 h-0. 01 BO h=0.02 60 *u, 40 20 5I01520 kr Figure 2,11 For a conbination of tv¡o ideal etalons, the value of *o, required to give a relative transmittance of I7" or 27" for t he prÍncipal ghost or sidebanci j-s illustrated as a function of k1. 0.5 0.1 0. 05 h 0. 01 =2O 0.005 N =80 N =40 R2 R2 0. 001 o 0.1 0.2 0.3 A.4 0.5 plkt Fíg11¡e 2,L2 The relative transmittance h, of the pth sideband as a function of k1 is given for a combination of t-r+o ideal e-talons each haví.ng the finesses indícated. :\7 . (n) AÀ = tT"lAÀ, (kÎ-I 1) (2.7 s) n ^À N the serÍes of n etalons is, ,An overall finesse T' for M(') (2 .7 6) Nr (n) ôI where of") iu the width of the resurtant profil-e. For a combinaÈion of the Ist type, ôÀ(n)tu ôÀ1 an<1 for a combina¡ion of the second type, O¡(tl" relared to ôÀ1 via fígure 2.10. If the factor I( is ¿efined as the gain in overall finesse occasioned by the addition of another etalon, then K tu kr for a combination of the first type and K .\, 1.4kr for a combination of the second type where k1 is ttte factor by which the free spectral range is íncr:eased by the aclditÍon of that etalon. Chabball (1957) sruclled the filtrage factor of ideal polyel-alons and concluded ttrat combinations of the second type l¡lere stlperíor to that of the fír:st type. Figure 2.13 illustrates the filtrage of a of titr, dual etalon F.p. The filtrage is essentíally a function 5-tnR, , l-ittle. dependence on No_ althotrgh the maximum occurs at Kl-'t\r a ---3- tor a combination of Èhe first type. ctra.bbal used "ra, *u, integrai:íon limirs of ! k1 A)'1 to calculate Èhe total transmitted flux of equatíon (2.67). Figure 2.14 j-LLustrates the filtrage factor for combínetions of fhe first type usíng Ët47o and three etalons' A combination of Èwo etalons Ís superior to Èhat of three eËalons only if the f íltrage facLor i-s -Less than 0.5' ThefactÈhatapolyeüal.onsystemcanhaveafiltragefactor greater Èhan Ehat of a síngle bandpass single eËalon ís related Eo Èhe der:r:ease of transmission ín Ehe vrings of the central band caused by the transmittance multipl icatí-on process' 1.0 NO=30 ¡{ o {J CJ (ú rH o nd ò0 2 type (Ú 0.5 ${ .u St -l'rl 1 tyPe t¡{ o.25 0 o.25 0.5 0.75 K t*, Figure 2.13 The filtrage factoT, e, for a dual e¡alon F?' either ín a lst qr 2nd type conrbinaEion. t{ o Þ (.) d three. etalons tH o) ò0 d 1-{ 0.5 Ð r-{ 'r{ two etalons Ftt *'-- o.25 0 0.1 0.2 0.3 0.5 o.75 K **, Figure 2.14 The filtrage, e, for a Ëwo or three etalon combínatíon of the l-st type. 38. So long as the facÈor by whích the free specllal range is Íncreased by the addition of each etalon is less than the fÍnesse by an apprecíabl.e factor, then the filtrage factor is always close to 0. 5. 2.5.2 Etalon Coupling It is importan'E, Lhat each successive F.P. etalon is incl-uded wiÈhout a loss of 6tendue, SQ. Equation (2.57) ind,icat-es that the resolvance - solj-d angle product Ís a constant under the condíÈions of optimum light gatheríng poü¡er. The érendue of the original F.P.I. ís to be maintainecl and so each successive eÈa1on must have an 6tendue at least as large as the original. This is arranged by constructing the polyetalon with etalons of successj-vely lower resolvance. From equation (2.56) and the- requirement of maíntaining átendue, st Sr (2.77) R. R1 l- rh where S. and R, are the area and resolvance of the i etalon. l- l- rtrt a combination of the second type'Ri tu Rt' thus the eEalorrs have- comparable area and operate with Èhe same solid angle. Figure (2.15) illustrates the coupling norrnally usecl for a dual etalon combination of the 2nd type. R, For a combination of the lst type, r and by equaËion (2'77) =ar(l ' ' the lower re-solvance etalons can operaEe at larger solíd angles and smaller areas. A smaller area has a significant cost advantage and a superior defect fínesse can usual-ly be obtained. Thus combinations of the lst Eype (and also the 3rd type) are usually couple magnificatíon provicte Figure 2.15 Optical coupling of tvTo FP etalons in a combinatíon of the 2nd type. E2 (a) E1 E2 (b) E¡ Iigure 2.16 Optical coupling of t\^ro FP etalons in a combination of the lst type. (a)'positíve-negativet coupling. (b) ¡posiLi,ve-positivet coupling. In (b), the inter-etalon clístance can be reduced by the use of- a field lens at the common focal point between the etalonS. 39. but v-Lgnetting carr be inportant, partÍcurLarLy al l.o'n¡ ¡:esoluEion. Coupling (b) is usu¿r-lly pre.l.erred anci. vígnetcirrg .i-s rninirnised by usÍ.ng a fíeld lens aE ttre iul-ermedía-ry field stop to image the low resolrrtion F.P"I., 82, on Èo the high resoluÈion Iì.P.I', Et, if iÈ is desirable to minirrrise the interetalon distance. fn the coupl--ing of ttre type in Figrrre 2.!5, vígnetting ís mínimísed if the beam dive-rgence is srnall (usually the case for etalons of high resolvance) and by keeping the etal-ons as close to eac.h other as possible. If the áEeudue of each etalon is Ehe sarne, then Er= &!=Di Q'78) Rz 9.2 DZ =:1ik1 where D is the eEalon diamete.r. For a coml¡ination as in Fígure 2"L6b, f l- = Dl- ÊL (2.7e) Í.2 D2 ßr wliere f is the lens'focal length ß:L is the angular spread of Ëhe "ttd beain Lhrough [he ith eta-ton. 2.5.3 Number of Eüalons Rer:luired The number of etal-ons reqlríred depends on the type of obser:vat,ion to be na,de but some índication of the number required ís obtained from a corrsider--¿ri-Lon of the maxinum allor¿able relaÊíve transmÍttance of the principal gtros! and the suppressíon required by the inEerference fílter at the wavelengÊh of banclpass coincidence. Sr-rppose that the addif-ion of each successive etalon of reflective finesse ll*, extends the overall fínesse by a factor, I(. For a polyetalon combínation of n etalons, lhe overall finesse ís, Nr = N*(r)o-l (2.80) 40" Thus the spectral range free of ghosts greÐ.Ler l:han a- cerLaj-n Ïreight ís AÀ(t) t ôÀLN- = u- ôÀ,(K)n-l I-iere Ehe resul-tant profile widttr has been assumed ec1ual t.o l;.he widEh of the highest resol-ut-ion F.P,I. The addition of successive etalons can cease whe-n the overlap wavelerrgth is rerjuced to a prec'leterniined valrre by the ínterference fil ter. Suppose this wavelength occurs aL x times the f -Llter wiclth, thus n is limited to the case rvhere , (2,82) 6À¡N,,(K)t-1i( *6Àr, or --t r I (K)" ' t "\r\-Iì. , (2.83) r¿here R and Rr'are the resol-ving porvers r:f the polyetalon coirririnatíon and the inEerference f iluer respectÍuely and ôtrrOis the f il.ter r'¡í.dth. The value x for a given filter Lransmittance clepends on the filter shape. Irigure 2.17 illustrates ecluat.ion (2. 83) for various v¿l Lues of reflective fínesse, assumíng the princj-pIe ghost Eo be less than l%. This sche-me should only be taken as a rough gr-ricle s;ince perfect etalons have been assumed, The principle ghost heighE will- be irrcreasecl by plate defect.s and Ehe use of a finiÈe fiel one assulnes a reflect-ive finesse that r'¡ould be compatible with Ëhe expectecl rlefect f inesse and then reduces the allowable ghost transmittance by a factor of 3 ro 5. Thís sche-me is al.so sornewhaË arbitrary in Èhe fact that lar.ge changes of some gtrost transmittances can be accomplishecl by relatively small changes in the eEalon separation rati-os. N*=50 N*=40 double eEalon triple etalon 1000 xR N =20 Rtp R 100 10 I 2 3 n (number of etalons) Fígure 2.17 The nurrber of ideal etal-ons of equal fj-nesse, for various finesses, requÍ-reci if the relatíve transníttance of the principle sideband Ls l% and the polyetalon and interference filter have resolvances of R anrl RrU respectively. 4'.1. 2,5.4 Tns Èrurnental Pr.rf il c 2,.5 .4 " r. StatemenË of L:he Frc¡bl-r:rn Chabbal (1958) proposerl that the instrument profíle of a dua.l etal-c¡n F.P.I. cor-rp1ed withour magnífi.cation (.2rrd type) ls, . r (tr) = i tnr (r),kD] (À)l tAz (À)'rDl(À)l 1 '* ¡(À) (2 '84) rh vrhere À. ancl D. are the Airy ancl clefect funcÈioirs of the i et¿rlon l_ tL and F i-s the aperLure fuuction definecJ by the soli colrùnon to both eta-tons. Tor coupling wJ.Lh magnif Ícation (1st type) , Chabbal pr:oposed r(tr) =. {nr(r)t(D1(À)'kFr(À)}{42(À),kDz(À),kFz(À)} ... (2.85) r¡here F1 and Fz are the aperture functions clefinecl by the etal.ons j-ndiviclual solid angle. To the lcnowle .¡¿rlidity. Ttrese two ecluaËions fail in tvro funclamental consideratíons" Firstly, it is r.¡e1.1 knc¡wn (Maclc eÈ a1 ., L963, Roesler 1968, Barmore Lg72) that tTre snrface defects of orre etalon inÈer¿rct, wÍÈh tl-re defe-cts 'of the oLhe:r such that the tr.ansmittance can be ch:rnged by varying the relative axial or:.Le.ntatíon of the etalons. Thís ís not predÌ.cted by equations (2.84) or (2.85) and stents from the facÈ that the defecÈ function (section 2,4.2) ís incler:endenl- of the spal-:ià1 dísLrj-brrLíon of the defects, Seconrlly, for ¿r combinatíor-r with magnif icaEíon where both etalons are tunecl to have rnarcirnum transmiËtance of Ào at normal incÍ.dence, Ehe two etaLons become detuned aclîoss the common f:ield stop. Again tlris is not conEainecl ín eqr-ration (2.85). Roesler (I974) suggestecl a solution to this probl,em o.Ê detiining (see 2.5-4.3) " IË thus seems cles.irable to reformulaLe Ehe Eheorel-ícal descriptiou of a d.ual F.P,I. (Chis can be extendecl to a pol-yetalon 42. F. P. I. ) and to investi_gate un'der rshat con i.ncidence 2.5.4.2. Ef f ects of Pl-ate Defects The variation in separation over ttre etalon surface is clescribed by a function L(r,o), which is the sep:rre.tion in excess of the mean separirtíon at the posi.Eion (r,0). IÈ is instructive to initially consider a single etalon, at normal íncidence. If an element of area, r dr clQ, sítuaËed at (rr0) is illumj-natec{ by nonochrornatic radíat.ion oJ:- wavelength À, then íts contril¡ution to the transmitËance is, (see secLior. 2.4.2) dEÀ = A(¡,-\) t*r-!{ . (2. 86) where = L(r,q) (2.87) ^, * Thus drÞ E (À) A(À-ÀL) {_dr (2. BB) =il S rþr For an ideal etalon, L(r,Ô) = 0, À, = 0, thus E(À) -- A(À) as expectecl. Norv ecluation (2. 84) and (2. 85) are equival-ent. hÍgh resoluEj_on FpI (r ) low resolution FpI (2) I I 9.2 apert,ure (a) .cr (2) (1) rotated by n radiaLrs t- ï - - ¿lperture 9'z .Cr angu T discontlnuity (b) Fígure 2.18 The optical- combínatíon of two FP ef:al-ons as in (a) ís equival-ent to the coml¡ination (b) where etalon (2) is rotated by n radians. 43. Tor a dual etalon of any type cperating at nc¡rm¿ll íncídence, the contríbutíon to tÏre tr¿rnsmittance o, À rto* an e-l.ement of area at rrQ is d0. dE = Ar(À-À. )A2(À-À. r Ë-4' (2.B9) rr l Lz-' S where Lr and Lz describe Ehe eEs.lon defecbs and Àr, and Àr, are found frorn equation (2.87). IntegraËing over Ëhe total cortrnon surface, the transmission prof ile i-s E(À) = a, (À-Àr,, ) A2 ( À-À12 )r-uå-gg (2.eo) 0 t Chabbal woulcl presenÈ this as' E'(À) = {Ar(À¡'t¡r(À)}{Az(À)*Dz(À)} (2.9L) Equation (2.90) car: be simplified under the followíng conclitions. (í) If borh etalons are perfecE, Li(r,0) = Lz(r,Q) = 0, thus E(À) = Al (À)42 (À) (2.92) as expectecl. (íi) If either eE¿rlon ís perf ect, ttrat is L. (r,$) = 0, then J E(À) : iA. (À)v.1. (tr)l Aj (À) (2.e3) where i = I i.f j = 2 and vice versa. llhe axíal dependence from equation (2.90) is re-rnoved if (a) either or both etalons are perfecË (b) eitþer or both etalons have a symneÈ.rícal defect such as a spherical <1ef e.cË. Thus equation (2.90) is reduced to 2t¡ E(À) = Ar (À-ÀLr)nz (À-À" dr (2,e4) S r)r T oEher simplifícations occur if Ll(r,0) = cL2(r,0) where c can l¡e posiËíve or negative. The integrals above have noÈ been developed frrrther at t-his time 44. but numerical irrtegraE:ion wor-ild be sEraight fr-¡r'r¡ar:â fot tile tnore conmon. eìef ects (spLrerical , mic::osLructure or par:illelisrn n:lsarJ-ignrltenL), llowever, i-t is easily shown by simple example that the fo:n¡ulalion of Chabbal emd equation (2"90) are not cornpatible- . Roesler (1968) consiclerecl a polyetalon of Ltre second type, e¿rch etalon havi¡g the same spherícal defect. Front the results presented, it appears as though equa-tion (2.84) underestj-rnate-s the peak traus'- mittance by abou: 5%. Thís difference. is only of srnall conÉrequerlce r,rhen the general design of a polyetâlon spectrometer is bej.ng considered. Comparison of prof iles gerrerated by eciuation (2.85) r:sing ernpirically determinecl etalon funcÈíons and measurements of ttre instL:ument profile (chapter 7) incl,ic.ates that íf equ,ation (2. 85) was used in the design aspecLs of a polyetalon spectrometer, litf1e error would re-sulC' The presence of plate defecLs in a polyetalon systen fles consequeïrces simj-Iar f-a Ëhose in a single etalon F.P.I. ' fianìe1y ír recluction ín rcsoluì:i-o¡r and a seveïe -Loss of light gal"]neti-r'g powet It Ehe refl-ectance of Ehe coatings is Èc¡o high. Tl:re dc'.fect func[ion incre¿rses the ghost Lransmittances, (espec-ially ttrose nearest the central maxinum of ttre lov¡ r:esoh-rtion F.P.I.) but has littl-e effect ori the f i1trage factor (Chabba.l , !957) f-o-r a give-n reflecti.vs: fíïresse. , 2"5.4"3. EfiecE of a Einite .Fie1c1 of Víer^I As carr be seen from equatíon 2.41 or 2.45, the rate cll- change Of r^laveleng¡h transmitted across Ehe aperture sEop depencls only on the raEe of change of the angle of :'-ncidence cosinr: which depencls on the f ocal J engËh of the f ocussir.tg lens (2'9s) E == 1 - 4G/Ð2 ... for small angles of incídence. In a clual ecalon Ì-.P.I. of the f irst Èype, coupled wíttr map,nífication and ¿1 cotnmon fielcl stop, Èhe rate of change of f, í.s 45. dífferent for each etalon. AU a radius. vj' the high and low resolution F.P.I" ts Èr¿nsrnit wavelengths Àr = Ào (vu," (*r¡i)) tr > I ('2's6) and (2-s7) ' r,- = Ào 0 -"+ ) ".. respecl-ively where t 1 is the ratio of focal lengths of the couplíng lens and f1 is the focal length of the low resolrrtiol F.P.I. lens and both F,P.I.ts are tuned Èo Ào at normal incidence. This detuning j-s illustrated ín Figure 2.19. AE a radius ï, the relative transmittance of À1 is , 4R ,2m2 ,a . .',-t Az(Àz-Àr) = (t + 1fty sin'z [ (Àz-Àr)J for À2 - À1 small and Ào t À1 and where m2 is the order at Ào in the low resolution eÈal-on. This function is shol,rn in Figure 2.2O for a value of E¡ = 3, R = 0.94, ft= 300 mn ancl mz = 1100 (símilar to values used in thís expe-rirnent). It can be seen that a transmitLance of unity is not maintajne 2.5.5 -S"t*arJ It ís simple e-nough to couple several F.P. eLalons r'¡ithout loss of étendue l¡ut the light gathering po\¡¡eï achieved depencls on the ntrmber of eÈa1ons and their respective reflective and defecE fínesse' I^Ihi1e it ís in general true th¿rt a greater fl-exibilíty ín strppressing parasític bands is obtaine{ by usíng three etalons instead of two, the introductj-on of a ttrirrl etalon may be accompanied by a 1ow high resolut resolutíon FPI FPI o o Ë d ]J ¡J 'rl oÈ1 É GJ tr H À'z Àr À¡ À Fígure 2.19 The high and 1ow resolutíon FPIIs are tuned to À¡ at normal inciclence. At some off axis posiLíon ín the aperture, Lhey are tuned to Àr and Àz respectívely, resulting in a transmission loss. 1.0 0.9 0.8 m=1100 Az (Àz-Àr ) 0.7 m=1500 0.6 0.5 1.0 1.5 2.0 (r*) Figure 2.20 The relative tr:ansmittance across the aperture for IPI parameters símílar to those used in thís experiment. (see text p.45) 46. signif -LcernI reduction in EransmittarÌce" IÈ is n;hus ot- sonle importance not to oversper:ify [he requi-rerl Í:iit-rage factor or relative sideband transmj-Etaïces. These specÍ-f ica'cions are only sel: by consideration of Ehe type of observation to be macle. . In a dual etalon F.P.I., the filtrage factor and sidebancl transmittânces clepend pri.marj.-Ly on the char:acteristics of the low resolut'j¡on F.P. for a combination of the f irsL type. I'Ihile it is j-mportant to maj.ltain a high contrast (high reflective finesse) in the l.ow resolution F,P., sone gaitr in transrnittarrce ís obtainecl if tl-re reflective finesse of Ëhe high resolution I¡.P, is decreasecl when significanÈ coating absorptance is presenL. This ntak-es little 10). If the free spectral range is extencled by too gTeat a fact-or (say, kr > 15) and yel small sídeband transrnittances ¿,rre stlll requirecl , a greater reflective f Ínesse is requi::ed, irrelspectj.ve of ttre combinational Eype. This fines;se may be incompatible r,;¡ith the defecÈ finesse, r:estrlti-ng in a severe clecrease ín transm-Lttarrce. T-n general-, one follorqs the clesign criteria of er síng1e etalon F.P.I. as faÏ afi opLinisation of the L*R product is concerned. The presence c¡f plare clefects and a finÍte field of 'vieiv in-cre¿ses the relative sicleband transrniÈtances but has little effecL on the filtrage factor. The determínaÈion o:i the best spacíng ratio is soneÈímes clifficulr. McNutr (1965) and Stoner (1966) have developcd sone theory in relat-Lon Eo a triple et¿rlon system, holvever, the clua1 eEalon is much sirnpler and can almost be considered intuitj-vely wtren consideratíolr is given to the proposal of dec,reasíng the separat:i-on of Èhe low resolution F.P. t.c.¡ (:ompensate f or mistuning across Èhe aperture' The complex j-nter*etalon interactíon of plat.e defects makes it advisable to determine the instrumentts transmissíon' profile 47. empiricålIy. slmilarly, íntensity calibratiorrs strould be made empírically. 2.6 Choice of Operating Parameters polyetalon . The nature of the observations to be made r.¡ifh a spectr:orneter has significant irrfluence on Ëhe c"hoice of paraneÈers" Tn general, the more relaxed the desígn criteria, the higher the spectronteter ligtrÈ gathering pol^Ier because of the ability to use fewe:: etalons and to use coaÈíngs with reflectanc,es compatibl.e wíth the expected defect fínesses. Observations of the dayglow require the spectrometer to scan through the solar Fraunhofer absorption line at À630.031lnm. The emission f eature j-s ísolated by the subtraction of a sui¡ably nornìalisecl direct sunlight spectruin from the sky spectrurn. The data analysis scheme requires Ehe 1ocal corrtinuum to be sampled on eíther sj- The spectrometer bandwi-dth ís set by consirltr-,:ations of Èhe fluxes at the cletector origirr.at:ing fron the emission line and the quasi- cont-inuous backg::ound of scatterecl sunl:'-ght" Equation (2.9) gíves tire f l-ux froin a continuum whích has a spectral racliane.e B*. Superírnposed on this íni-ense coniinuum is a weak emissíon line wiÈir a maxímum spectral radiance of a Bor. In the case of the dayglol'/ aE À630nm, t-he facËor, a, is expected tD be about 0.01. Thus the noise on the recorded sigiral is rlue almosË entirely to the background. For an observation peliocl oi t seconds, the signal Ëo noíse ratio j-s, (assuruíng Poisson statistics) 48, l'4, 0 t' L (2.ee) s/n = t- Q"'' where 0a ancl Õ" are the fluxes due Eo the contínuurn and the enissiort líne respectively. usíng equaLions (2.9) and (2.12) for a given t' Usíng the graphs presented by Chabbal (1953) for rB as a function of ôÀU , a broad maximum in Èhe signal to noise ral-:io is foun'd to exist T'À---I at ôÀU t 6Àt. This Ís the sane as the condition for opËí'nu'o L-R product. The maximum toler:able sideband ËransmitËances or ttre filtrage we.re determined from two consideratíons. Firstly, for a single banclpass of wiclth comparable to Èhe line r,¡idth, the emission line fl-ux at the detector ís expectecl to be al¡oul I% of tire J-lux frorn the back- ground continuum. If Ëhe flux origì-nilting from wavelengths ourside thís single band (here a single banci includes all Eransmittances wíthin t%^À1) j-s b Èi:nes thaÈ inside the band' the.n ttre signal !o noise rati.o is' 1.0*2 o t- C (2.100) s/n = t4 t-2 (0c+b0c) The integratÍon time required to achíeve a given signa-l- to rtoise ratio is, r cx (1-Fb)2 (2'rol) Under assumption that Õa is constant as b varíes, Èhere is no great disadvanÈage if b is of the orcler of 0.2 to 0"3. Thus Èhe filtrage requiremenÈ can be quiEe relaxed. If b htas to be significantly l-ess than 0.2, j-t woul-d probably necessitate the use of more etalons and as j-n a consequence Õa vrould decrease and Èhe increase the requlre Second.Iy, the nraxintum Eolerable side.b¡lrd transmittances have to be ccnsidered frorn tlr,r: viewitcj-nE of tleakager from spectral regions occupied by afrnospheric absorpLion lines (ehaiiter -1- and 7). As a first estimate, it rvas considered aclecluaLe if the prÍ,ncipal s-lrlebarrd had a transmittance of J-7" and that the first overlap be suppressed to about 0.5% by Èhe- interference filEer. This allor,red for the Íncrease in sideband transmittance rvherr the defecL ¿-r,ð, aperEure functions were considered. .Ihis increase coull be partially compensated for loy the jurlicious manipulatirrg of the etalon separatíon rat-Los once Che broad clesign eriteria h¿rd beeu set. Àt the time when the basic parárneters r,/ere beirrg debermi-ned, the diameEer of the plaLes of the highest resolution etalon lracl been set at 150 nrm, brrE fiuesse measurements \4te-re not ¿rv¿riIable. Consequently, the parameËer:lì l)/ere chosen on the ¿rssur¡pLion thaf a defect finesse oJ. 50 could be achieved, Thís turnecl out to l¡e ao over opi:imistic estinate. The assumed defect- f j-ne',sse would be wltched by a reflective finesse of 50" It r,/as also assumed ttrat al-l subsequent. etalons j-n ì:he specEromeËer would have similar: firresses. F-Lnally, the :í-nEerference fil ter characteristícs are requi.¡:ed Ín orcler to decÍcle how mauy etalons are to be used. InLerference filters r¡rere cotnnercially available wittr rvidËhs of aboul- 0"3nm aE À630nm and peak trarrsmitÈances of about 50%. TtiEh a trvo period fil-ter, the 0.5% transmíttance points (relative to the maximum transrnittance) occrtr at about twice the filter widËh from the w¿lvele-ngLh of maxÍmum transmittance. Vlith a reflecEíve finesse of 50 and a maximum tolerable principal glrost of I%, Figure 2-.L2 impLies a value of K = 10. The reflecLíve resolvance vras assumr:d to be aboul: 1" 7 times the overall resol-vance 50, (set b¡r the requÍrenent ôÀ, t 6À8). Thus equatj.r-rir (2"81j) inilj-catecl that trvo etal-ons rvould be adequ:rÈe. It was decidetl thaE the etal.on combinat:1on should be of the first type, Èirus I( = kt = 10" From equation (2.78),it can be seen thaL a 50 rul diarneter etalon wa.s required, the separ;rtion ratio being about. 10:1. The use of a com'birration of Ehe firsË t-ype had the followíng advantages. (i) it is much more ecouomÍcal to use 50 run opticirl:Elats" (ii-) a Lar,¿er defect f inesse coulcl be expected. (íii) since the spectrometer profile j-s fundamenta.l-ly deEermined by the high r:esoluÈion F.P.I. for a combination of the first Èype, sufficíenÈ performance should be obtained from the low resolution F"P.I. without resorting Eo elaborate servo controls for separaEion arrcl parallelí-sm. The opt-ical coupling v/as to be o.E Ehe r:ype illusLr¿ited j.n Figure 2.I6b, ttrat is, a'positiverpositive' system. As prevlously mentioned, Ehe sysrern étendue is deterrnÍrred by the higher resolution F.P.I. The choice of solicl angle is dictated by ecluation (2"55), where if the L-R product is opEímised, R tu 0.7RF, Thus the sol-Ld angle of the high:r:esolution F.P.L is gi-reir as n, ={o"zR Q" LOz) Horueve.¡:, it was in this choice of aperture Ehat thr: 5y5¡sn <1eparËed from optimumization. The defect fj.nesse of the hÍgh resoluÈion F.P.I. turned out Ëo be much less than 50 buÈ it ¡nr¿rs sti.Ll important to retain a free specEral Ta1-r-ge of about 0"04 nrn' Thus if the aperture finesse was set fo:: optimum performance, the resulta;nL ínstruruenL vri-dÈh would have been in excess of the i-i.ne v¡idth. IË was initía1ly thought ËhaÈ Ít rvoul-cl be beneficial if light gatheríng por,irer was sacrificed for resoluríon" The slightly higher: resr-rlÈant 51. fínesse of the low resol.ution I',P,I. rvould aid in the suppr:essiorr of the parasitÍc bands. Thus a solid angle of abcrut h¿¡lf the oipimum rvas chosen. Later consíclerations showed that an ac.tual improvement in resr:lÈs woulrl be obtained by usíng Ëhe optimum solid angle. The observat.ional results .reporÈed in chapEer B, were ¡na The separaL.ion of the etalon plates ís coarsely set by the desired free speciral range AÀr, and the separaÈion ratj-o trc1. The aciual values r¿ere chosen as follows. Because the higtr resolution F.P.L rlras to operate at night as we-ll , the order ri/as seÈ to avoid OII contarnination (Ilernandez, L97l+) and to facilitate easy wavelength calibration (chapter 5). The separation of the low resolution F.P.I. r,ras set small.er than dicEated by the value of k¡ above.. The choice r¡ras governecl by the rnents (chapter 5). The separafion \,ùas decreased .f:rom the l0:1 r:atío. This decrease in sepat:ation of Ehe lor¡ resolutiorr F.P.I. íncreased the Èransmittance of the firsÈ few sidebands but thi-s r¿as considered tolerable. The value of k1 thus chosen r¡Ias al¡oug, 13.3. Th-Ls value is also more- in line with the proposal macle by Roesler (1974). (see secËion 2.5.4.3). The achíeved degree of parasític band suppressí-on is discussed in detail in chapter 7. 52. CHAPTER 3 TTIE ITIGH RESOI,UTION FAßRY_PEROT IN]IERFEROMETER 3.1 Introcluctíon The high resoluLion F.P,I. was cleveloped ¿rL Ëhe ì[awson Inst-itute to make high resolution specLral studies of the- nightglorv, in particular the emissions from aËomic oxygen at À630nrn and À558nm, The instrument has been descríbed in detail in Ehe theses of Bower (1974) and llilksch (1975), and Ehe description presented in this chapter has been included to present a mole compleÈe-.description of the dual F.P"I. developed for dayglow observaEions. 1'tre modifícaÈ-ì-ons made by the author to the existing equipmenE are clescribed -ln Chapter 5. Details of the phoLon detection scheme are also presented in ChapEer 5. The high resolution F.P.I. is a large aperËure scanning .llabry-Perot spectromeLer incor,ooratj-ng seïvo-mechanical con-trol of paral1elj-sm ancl plaEe separation. The F.P.I. is scanned by the variation of the plaLe separat-ion using piezoelectric ceramic Èransducers which are also involved in Êhe parallelísm contïol schene" The ins+-rument is desígned to wo-r-lc over a large range of rvavelengths and a large xarrge of plate separations. Photon detection Ínvolves an BMI 9558 photomultipi-ier with an S-20 phoEocathode. The effectíve quantum efficiency is enhancecl by the use of an Hirschfeld cone and the clark current can be minimísed by cooling and magnetic defocussing. The high resoluËion F.P.I. Ís a versatile instrument and was ideally suited for its role in the dayglo¡v observations. 53. 3.2 OptÍ.caj- F1¿rfs/Pl-ates and P."eflecLi.ve Coat:Lngs The opËÍc.al fl.ats consist of twcr trS0rnm C-Liarnete..r, 25rnm Ehick fused silj-ca plates witl-r 45o facc¡.l:s orr L.he lowe-r plzLte as strov¡u in Fi.gure 3"1. The piat.es r,re.r:e- hand pol-i-str.e<1 by J, Cole of CoIe PrecisÍon Optics, Adel.aide, ancl the reElecLive coatirrgs l./ere applied by J. I'Iard of !ü" R. E", Salisbury. The reflecEive coaLings were of a s,Llver-diele-ctric composi-tÍ-on ill-usErated in Figr-rre 3.2. This coating was chosen in preference to a multi.-layer dielect::ic coaf:iirg becarrse it- h¿s less vari¿l.tj-on of optícal properties vrith wavelength. The broacl band nature oi the coatiugs ruas lequired because Ehe instrument was to be used at several wavelengths across the visible specÈrum. The high reflecEance of Lhe coatj-ngs in Ëhe near infra-re Table T contains a list of Ehe o¡:tÌ.cal prooerties of the co;rtíngs" Coat:r"-ngs of the- same design r\¡ere appl-:ied to the l.o'rr resolution F.P.I. T?re unc-oated opÈj-cal Í1ats were est:imated to have a rlefect fíne-sse of aborrt 30 ancl so coatings with a refiectr'-r¡e fj-nesse of 50 r,¡ere applíecl . llorrrever, the final defect finesse \¡ra,s rleasure-d aÈ 16.5" This m.isrnatch o-Ê defect and refl-ecEive fj-nesses resulted in a loweríng of ttre etalon ¡-ransmíssion. The Ltansmission is estimated at abouL 23%" It is belie-vec1 that the decrease i-n the defect fi.nesse I'ras clue to non-uníform- iEj-es in Èhe coalings causing varíation of phase change on reflect:lon. 3.3 Parallelism Control- Satisfactory perfornìa11ce of an F,P.I. requires that the p1:rEe.s be maíntaíne c D + Fígure 3.1 Principle of the servo-mechanícal parallelísm scheme illustrated for one of the two orthogonal control axes. CeO À ß À500 nm 2 4 MgF i @ À356 nm 2 4 Ag 30 flm Iusecl Silica Figure 3.2 The component layers of the reflective coatings applied to the FP etalon. 54 A coll.imated beam of broad band Ligirt: is:i-ntc'rnally rc+f1.ecte-d by the facets of the Lo'r;,¡er plate as i.n I'igur:e. 3, i. \^iTren Èhe pattrs AB=CD, the channe]- spectr¿r resultr'-ng f::clm e'ach ;r¿55 throrrgh the etalon gap mal:ch, pr:oducj-ng a maxi:num intensily at Èhe detector¡ a FTN diode. Any departure from lhe con í.ntensiEy at the cletectot. The sense of the correction to be macle Lo restore para.l1el:í-sm is cleterrnj.ned by the applica.ti-on of a 4kLI.z, 3tln amplitude rwobbl-es to the top plate u.sing the piezoelecEric ceranj-c tr:ansducers. The ampli-fied ctetector signal is fed Lo a phzrse serrsiËj-ve cleEector whose out-puL controls rhe parailelism via a negaL.Lve. feedbaclc loop. Parallel,is¡¡ is sensr:d and e.djusted alon-g tr¡o orthogonal axes' X and Y in such a rvay Ëhat the mean spacing rerira-ins constant. Light sc¿rttered from the paralle.lism conErol scherae arid cle-tected by Èhe photomultiplier is r:rj-nimised by restricting the wavelengths; used to Ehe near ínfra-red. iL-he PTN dj.ocle st-111 has a good resporrse in this regíon whereas the photornultiplie.r has a quantliln eff iciency of less tlran 0 ,I% at ÀB50nm; 3 .4 Separa h-i,on Cont-rol- Long tern separat-it-rn srabilj.ty a-nd lirrearity c¡f the scari are. rìecessaïy in an F,P.I", especially f.or nreasurenìents of the small. doppler shif ts neede-d to esti.mate wi-nd velc¡ciEie,s" 'fhe non-lineariEies cf the píezoelectric cel-am-ics re,luired the separat-ion to be sensecl al- each point of the scan and Ehe approp::iate correction app.lÌed. Because of the Large plate sep.arar-ions trsed (S co 10mm), ctranges Ín refractive index of Èhe air due t-o prcìssure changes are sígnifj-canL. The pressure varíations must be measurecl and corrections nacle to Ehe separa*"ion such Èhat the optical r;eparation is Lcept const¿1nt" The core of t-he separation contro-L is a capa-,r::Ltive dísplacement transducer. IÈ consists of two sets of inLerlear¡ed sEeel discs, which 55" aïe ínteï-coïtn.ected to fo'rrrr fcur par-'al-Iel p-Lace etir-spaced capacj.Eors ín a bríclge formation as íllustr:atecl :l-i¡ Fig''rre 3.3. One set j.s fixed i:o the lower plaEe suppoïL and the c¡ther moveable seü is connecte-d to the ilppe:r plate support via an invar rod. The air gaps al:e maintaíned ¿r.t- a spacing of about 501rm by snall leaf spr:íngs ancl each capac-i-tor has a capacitance of abouE 200 p-l-. A lOkllz signal ís applíeci ro th¿ brictge and the capa-citors aj:e;:djusted to balance the briclge by zr Ehe rocl posi-tion, Any clisplacemenL of the pl-ates causes an unllalance in the- briclge, the signal generated across the bridge beíng proporEíonal- to the clisplacetnenÈ. Pressure variations are sensed by an aneroirl cell and a displacement transducer (Uer,rlett Packard 24DCDT-100). The signal from the transdticer is amplified with a gain thaC is set- according Eo the mean plerEe separâ E j.on. The separatíon control system (Figure 3.3) also provicles fclr marrual corrtrol of the plate separatÍon and for scarrning the I'.P.I. The- nranual offset, baromeLer and capacitor. Erauscluce.r signals are suntnecl ¿Lt Ehe equal-izer: and conpar:ed to the scan inpuE sígrral" Corre(:tions are rnacle t-o the pJ.aÈe separâtíon vj-a Ehe píezoelecEríc ceramíc tran¡;c.l.ucers such that r-he summe The clisplacement trans freque-nt waveleng rh calibrations . Cffset 3 5Ì0k lOkHz Bandpess Frlter 5l0k 4 l¡iverter Drrve Rectifier L/ 12 K Circuit Equiralent 600v Extemal Supplies Scan Ínput Equalizer Variable lnrer Rcd A.ttenr-rator 2 Silica Piug D <-hp 2aDüDT Filler Ring Pressure Anercid Transducer ,.1 -Steel -MicaWasher Cetl 5 $\--S'eel DC lnpra I l/,r ¿A 4 D'sC Capacitance Dìsplacement Transcjucer Figure 3.3 Block 3 .5 Gener:al- S crucËure The F.P.L .Í-s enclosed in a Large' fr:arne rvith insulated i'¡ooclen panels. The air w:lçhin L,he enclosure ís lemperature controlled' The components of Ètre II"P.I. are contained in a long r:ylinder i^rhich is supported on steel springs. I'igure 3.4 shows the I'.P.I. as used for níght-Eime observations. A complete list of all- component par'ameters is conr--ained in Table 5.1. Pí2¡'lscapÈ prøf ilte i' & imcrging lcns fiøld stop spring support compound <3r lens collimoting lens cell €.> øto lon etolon chqmbør 2m focusing løns æ shutter photomultipliør enhoncømant conø I chamber photomultipliør Fígure 3.4 The spectrometer structure an CIIAPTER 4 TIIB LOI.I RE SOLUTTO}I FABRY-PEROT TNTERFEROI.{ETER 4.I Design and Construction of the Lorv Resolu Eion Fabry-PeroË InLerf erometer 4.I.l. Introductíon The 10w resolut.ion F,P.I. consists of a 50mm diameter eta10n vrirh irigh f inesse optical f l-af-s. Coarse parallelÍsm and spacing adjustments are made by means of dj-fferential screr¡/s. Fíne adjustment and scanning j-s accomplíshed by the use of piezoelecl-ric ceramic tubes. The operatirrg orcler is 1 136 at- L630mn and the Eemperature compensation Ís opti-mised for thís spacing. 'Ihe eËalon is operatecl ín a Èemperature stable environment ancl is i,solated from vibration by a damped-piston suppoït system, The etalon ís set parallel by vierving frínges of equal j-nclínation (FP fringes) using a He-Ne laser for i1lumínal:íon. A fellors resea"rch strrclent, R" Base-dor.r, used a 50mm F"P.I" of sinilar de.s:Lg1 Èo nJeasure OH rotation¿rl tempeïafures and conseq.uently the ínjËi¿rl part of our respective, projects ruas spent collaborating on the clesígn, constructiol and Èesiing of the e.Èalon. Once the design vras. shown to tscaled produce a sEal¡le ancl reliable eÈalon, it v¡as upt for use as a 75rnm aperEure ínstrument by M. Chamberlaín to measure racliations from atmospheric H^ and N]. The same design has been used at the SouËh Þ AustralÍan IrrsÈituËe of Technology to construcE a 50mm F.P.I.; iníÈially for use in undergraduate teaching, buL even'tua1-ly to be used as a research instrumenL" Perfo::mance figures for Lhe F.P.I. are given in Sectí'ons 4.4.5 atd 4,4 .6 . 58. 4.1..2. Design Corrcepts " It was Ínit-iatly proposecl Ltr;lt ¿n etalon be- cletsignecl ancl sonstr-'ucted to operate ¿t a fíxed separation. llowever, the ciesigtr wers to be slrch Ehat ihe mean separation conlcl be changeci Írom a ferv mici:ons to abouE 2nrn. Coarse parallelism adjus.;tnerì.ts r.rere tÐ be .'¡ade mech¿n.icall-y ancl fine adjustmc-nts by piezoelecEric cerarnics. It ivas also pr-oposerl ÈtraL no servo systems were Eo be usecl for elthei: paralleJ,ism or separa-tion c.ontrol. This pJ-aced a high stability recluiternent into the cle.sign specif ications ancl was clif ficult to achieve. -tn th-e eai:ly stages of the projecÈ it \,,/as not realised horv crj-tie-al r-he long ternr stabil-ity o-E Lhe e.talon would be to the success of the daygl-ow observations' LaLer modifi-cations to Ehe associated electronics pernij-l-ted the etalon Lo be used in a scannj-ng mocle. The ela|on r{as Èo be operated r¿ith Ehe p,l-are.s horizontal , this being the mosÈ stable configuration as far as gravit-y is concetned. Sagging of the plates r*oulcl be negligible j-f a high ttriclcness to dianteler r:atio r,¡as used (a r:aE.io of 0.25 was chosetr). The et.:rlon rvas al-r:o to have: a high degree of immuni-t-y 1tetn t-he e-ffects o.Ê tempê-TaEure variat:i-ons. Ruggeclness t,tas also another design crii-.er,La. Misa-1-1grrment of f:[re plat.es by accidental knoclcíng could irr'¡olve lengthy reartljttst-menfs. I'trís meant that the plates hacl to l¡e firmly clampecl in their rnounts but: in such a vray âs noÈ to defortn them. Many publicatÍons harre reported corrsEt:uctional dei:ails of F.P.I'rs and t-hose. referenced here represent only a parr; of Ehe total. Sc;rnning the band pass of the inslrument acïoss the wavelengths to be measured can be irnplenented in a vari-ety of ways, Pressttre scanuitrg (Jacquirrot ancl Dufour 1948) and spaEial scanníng (Shepherd e.t a.L. 1965) are sLÍll rrsed by some wor'lcers buf: many of: the more recenf- i.nsEruments use mechanícal scanníng, that ís, changirrg t-he physicnl separation beLrn¡een the optical f1ats. This can be achievecl by pur:e]y ntechanical me-ans 59. (Greenler: 1957, Berrre-¡r 1971), inagrìeIosl:r.:Íctively (Slater et a'I. 1965) and e-lectrornagneiícally (tsruce ancl I:Iil 1 196I). Ilc-r.,.reverr the use- of piezoelecLric material-s ís nor,r very comnorr, p;ri:ticularly Ètre l-ead z:Lrconate - leac1 LÍt¿lnai:e (PZl'> ce.r¿Lrìri.cs (Rarosay L962, Greig L963, Jaclcson ancl Pj.lce 1968, llernandez 1970, Smeethe irnd James 197I, Clarlce et aL" 1975) . Pi.ezoelectric. cer¿mÍcs oEfet advantages such as a rvicle range of sc.:m pe.riods ancl separatj-on changes. Constructíon of F.P.I.ts with piezoelectric scanning is mechanically simpler than tr'.P.I.ts involving pressure scanning rvhich has an added clisadvanEage of re.-qui-ring large pr-.essure changes for et¿lons operaEj.ng at low orclers. By usÌ.ng stacks of dj-sks (Ramsay and Ì{ugr:idge 1962, SmeeEh,e ancl James I97L, J¿lckson and Pike 1968) or Ëhin rrall-ed tubes (Ilernanclez 1970), scans in excess of one order can be obEaine T.n applir:a-l-j"ons where Èhe scurce int.ensity may be r:apidly ehangiug, a fast scan across the specErum is essential , During twilight measuïÉìmeni-s of aEmcspher:i-c radiations, sc¿l.ns r¡-'rth periods of a few seconcls are required to nrinj-mise spectral clÍsEorLions. This scan perr'-od is easily achi-eved rvil-h p:LezoelecLric caramics bub not- with pressure scanning. IE was the versatilj.Ey of pi-ezoelectríc ceramic drives LhaE 1ed to their use ín the h-Lgh resolution F.P"I. (Borver L974, ililksch L975) and ín i.he present 1or^¡ re-¡;ol.Ltion F.P"I. Detaj-led discussion on the performance of the PZT cei:amics is containerl in Appendi.x I. Methcds o:E niechanícal parallelj-sm adjustne.nt such as compressing pads of deforrnabl-e rnateriat (Shepherd 1960, Berney Lq7l, Hinclle et a'L. 1967, Slater el; a'1. i965) \^/ere to be avoided because of the possíbility of creep ín the m¿rteríal- under compression. A double threaded or differential scre'"y system rrras usecl in the high resol.uÈiorr F.P"I, (Borver L974, Wi1lcsch 1975) and was foun '\,/ere hancl po1-ished by J. Cole, Cole Precisicn OptÍ-cs, Arlel.aide. The plartes are 12.mn thick ancl trave a small ledge ¿irc¡rrncl- bhe-Lr-' circum-Ee::encè- The plates are suppoi:Eed on this leclge as descril¡ecl in section 4"L"7. Each plate has a weclge angle of several minutc.s of arc Lo suppress Ehe effects of reflections Êrom the llncoated surf-ac'.es. The ref|ecEive c,oatings rvere applÍed by J" I'lard, I'leapons Resear:ch Establ.ishmenL, Salisbury. The eoatirrgs r'¡ere dc-signed to hc.v'e a high re.îlectance ancl k¡w absorpEance oveï a rvide range of wa'¡eiength-s (À530nm to À1000nm) and are simí14ï Lo those zrpplied to the high resolution F.P,I. 'Ihe coating layers are illusl--raterl ín l¡igure 3'2, and Eheir characteristics are listed in Table I. These coatings ai:e very robust and have shown rro sigrrs of deleriorat:Lon a-fter several years. The uncoa[ecl p1aËes r,rere assessed by vierving f::ínges of equal j-nclination irr .refl.ection usÍng a l-le-Ne laser. The pirir of plai: cluri-ng operâLion of tire- F.P.I. 4.L"4, Ilechani-cal Details Support structures for Fabry PeroÈ etalons have bee-n consl:ructed fronr a variely o:E rnalerÍals such as rnilcl steel (iUirdle et eL. 1967) ancl aluminiurn (Srneethe and Janies L97L, llernandez 11)70) but the most conìmon nnterial usecl is the lor.¡'the::nal expansion ¡:oefficient alloy, j-nvar, (Slater eí; aL. 1965, Berney 197'1, Clarke eb aL" L975>. However, this nateríal was not reaclí1-y availabl-e ín large cluantit:Les and the 50nm e-talon suppoïts Inlere conslructed of a bronze select-.ecl for íts hígh clirnensíonal sÈabílÍLy. A:FEer m¿rchining, t-he su1:porÈ:s wÐre stress relieved ancl' chernj-cally blac'l The mechanícal- Basically Ehe upper anc{ 1o'¡er p.Late sLlpports each consist of tlvo concenËric annuli ¡r¡ittr a cross*sectior-ra.-l- si.::e of L4 x l1rmn, I'he t'lp inner ¿rnnulus is nachined to contain the plate ancl moves relaËj-ve l-o the outer annuk-Ls and the other plate by rne-ans of Ehe ad.justrnent à"rur" describe*cl below. Each of the inner annrrl-i have Èltree. tLonguest protruding ac I20o spacing srud it is by these tongues that they are aÈtached Eo the outer annul.i. (Figures 4" 1 and 1r"6) The l.orver inne.r annulus is clampecl t-o the lower ouber annulus usíng a brass w¿rsher t-o conErol the coarse separation of the plates. Ori.ginally it wzrs pr:oposed [h¿lE t¡oth th-e upper anci lower anr-rttll- have acljustment scre.b/s, but ttre operation of the etalon r¡¡as such that the lorver scre.\¡rs $/ere superfluous. It was also felr that there míght be a srability gain by removing them -- -less moving parts. The nachíning of the plate supports in the lower annu.l-tts is Ídentical Eo thal- of the upper excepE for ttrr:ir physical l-ocation. The plaEe support and clarnping systern are descr:jbed in secEion 4"1.7. 'Ihe tongues of t-he inner anoul-i fil 'i-nto slightly oversize recesses cut in the outei: annuli. It is ¿Lt Etre locaLion oil Ehese recesses that Ehe actjustment screr/¡s are sj-tuated. The adjrrstment. scl:er^/s have two threads cuË on cl:'-.Eferent size clj-ameters. 12 Ehreacls per inch (tpi) Ehread is cut on the larger ^ dienreter and ¿t 7ro Epi on [he- smaller. 'Iension is applied to the scre\üs by means of a stiff, steel spri-ng whictr has a lir"tle less Lhan one turr. . A rubber washer was initially È-riecl buL r,ras found Lo be rrnsuitable because ils ela-slic propertj-es varied greaLl.y over ttre range of acljustnent recluired ancl it \,/as snspected that the rubber worrl-d deÈeriorate sevc:z:ely wiÉh age. The steel spr-ing r^ras limíEed to less tlran one Èurn to minilníse the vertical si-ze of l-he etalon and Ehus increase íts lateral stabílity. The spri-ng has a r'lrotlcing range of aborrÈ lmm. differential screw uPp er support clamping bezeL ríngs annuli spring ¡,¡asher steel sprang. - upper FP plate t torrgue +-píezoelecËric Ëransducer spacing lower tr'P plate ¡vasher lower support annuli <- base plate Ficure 4.1 Mechanícal details of the low resolution FPI. 62" The nuts conEainj-ng Lhe 76 Lpi thieacl are cemenLecl in positj-on to prevent Èhern r:otirEing. If boËh 1-ire scre-vr and \-lne 72 tpi nut are rotaEed togeËher, large sepelrriEj-on changes per revolutic¡n occtlr" Smaller adjustments are made by turnirrg i.tre scr:el¡ only" One revolution of the screws cha.nges the plate separation by 60 orclers at À630nm. Ttre hearl of the screr/ has holes drille 4 .L.5. Piezoe-1ectrÍc Cerami c Mounts. To date, the mosE cornmon form of pi-ezoelectrj-c displacement: l-ransducer for use in arr F.P,I. consisted of a stacic of disks, interleaved with copper r'¡ashe'rs, bonde-d togeEher r.ríth conclucting resin. Stacks of this type, using truo PZT-SH discs (Vernitron Corp., U.S.A.), were irrit:tal1y usecl ín the lorv resolution F.P.I. It lvas suspecred that sorne of the initiel inscability problerns in Ëhe l'.P.I" \tere due to the layer:s of resin beLwe-en [he components of l:he stack-" AlEhough 63, thís ivas noÈ conclltsi-r'ely pro\,'en, i,t.nras at- thís tí.me-: thaÈ bhe r'¡otlc of Hernanciez (1970) bec¿;me tnuroo' llemandez preferre-d Ehe use oi thin wal,1-ed piezoeJ-ect'r:i.c c.eraro:ic tubes and found that the PZT*A ceramir: had betLer l:lnearíEy ch¿rrac'.teristics than PZT-5H and that tubes hacl better linearity lhan discs. Tests in this TahoraLory (Appencli-x I) confirmed hj-s conclusíons anrl the 1ow re-solution F,P.I. is operatecl r,rirh three piezoelectric ttansclucers, each consisting of one r:eranj.c. tube. The loca-tir:n of tlrese ceramic Eubes ís illustrated in I'igur:es 4"1 an:d 4.6. IrÍgure 4.2- il-lust-rates the constructional details of the Lransducers. AJ-thorrgh t-he Lube assembly lacks the r:igidicy of Lhe- cementecl sLack, Èhe long te.rm drifts thaE still appear in the I'.P"-t. cannot be iclentifíed as l¡eing caused by the ceramic nounts As illustratecl in Tigrrce 4.2, a brass tccttoit reelr ís attac.Jre-d to the lower brass end cap by a bolt, thus locating a Ehin silíca Lube vrasher r¡hích is seatecl on a r-iclge the satne diarneter as Èhe " The vertical di.mensions of the silica washers an<1 the brass r:ncl caps depend on Ehe p-l-ate separaií.on chosen and Lhe t:eqirLrements of rhe thermal compensat-i-on schene dj-scrrssed in secÈion 4"1.6, The cer¿rrnic tube whj-cir h¿rs its ends ground Êl-at and pa:ral-Lel to rrr:Lthin orre minute of arc., has ,ro*a the rcoËtou reel'. "l-"rttor,."" The uppe:: silica ruasher is also 1at-era1ly l-ocaÈed by Elre tcoitoit .'ceeit. Ther top brass cap ís kept i.n position by the pressure suppl-iecl try a srna1l spring, thus clamping Ehe whole assembly tcgellrer. The spring teusíorr j.s reclucr.-d r¡hen the eLalon is assemblecl ancl in r'-ts place of operation. The inner wall c¡f the tube is helcl at ground potenblal, eleclrical contact being rnzrcle hy two strí¡rs of berylium-copper shim solclerecl to Lhe rcotton reelt, The ouEer wall is at a posíLive potential, electrical corrtact. being made via a spring coil of gold plated bronze- rnrire located af- about the centre of ti're tube.' The a.ppli-cation oÊ a racli¿l electric :fielcl catlses a.r a*iäl contr¿rctiori of the c:eramic butre. s l-eel sprÍ-ng lcm 1¡rass bolÈ * end cap washer /- insulatlon -sl-lica piezoelectric +600 v ccramic tube I cotton reelt earthÍng contact silica rvasher: <--- end cap j-ng Tigttre 4 "2 The rnechairical detaíls of tire pÍ-ezoelectríc transducer us ceraml'-c tubes. (¡4. 4 . L.6. Tempera tur:e Corn¡rens;r ti-on. Although F.P. eEalons usual-Ly operate ín an ern(.uir:onment l¡hich tras sorne degre-e of tcmperatur-e scabl'-h'-ty, it is desírable to clesign the e laton support s tructure in such a !'/ay as to nrinimise the e.[f ects of tcrnperature variatíons. firis temper¿ÌLure compensation is actr:leved by ensuring Ehat Eemperature iircluced expansions Lhat tend Eo decr:ease Ehe pl¿te sparcing ;ire collnLer-acte 50mm etalon design include tenperatllre compensation, even if it were operated in a highly temperature stable environmenE. The method used in the clesign ¿Ìssumes thaE any temperaÈure r:hange is exirerienced by all cor.rponents of Ehe etalon simul-taneously. Once Ehe location of Ehe plate-s in ttre inner annul.i and fhe plate spacings have- been selected, two simu-Ltalleolrs equations, describing the expansions increasirtg and decr*easing r-he plate spacÍ-ng, have Èo be solve PZT Eubes havi-ng been predeLer:míned. C¡rlc.r-ilations indicate. th¿ri this compensation schene shoulri give soacing change.s of less t-h¿n À/i00 at À630nm per oC change in temperatuïe when the lilcely errors in coefficients and machin.ing are consiclerecl . Iiorvever, the conclitÍons of the assumption are never met in practice, as there is alrvays differential heating of the etalon sllppoits. For example, the top annuli are not well thernrally couplecl to the lower annuli and the efficiency of radiative coupling wj-th Ehe surrounclíngs varj-es throughour Èhe etalon. Consecluently, this scheme is of rnarg-Lnal use when the etalon is subjected to large, rapid changes in te.mperaL.ure but is expected to rvork well for small, slolv vari¿rtions. This implies 65" Ehat as much therm;rl capac.',r-Ly as possibl-e sh.otLl':l l-le inco.rpor¿rted ín tl-re dtalon support structure ancl i-ts re¿rr environment ancl that the t,emperature should be controll-ed to rnuch better: L1'Lan 0.loc. The matte-r of temperatllTe índuced insrabilitie,s and temperature control are discussed in sect.Lons l¡"2.3 and 4..4.6 respectívely. 4 " L7 . Plate ìfourrEings. The positions of the plates in the supporting annuli are illustrated in Tigure 4.1, rvhich also il-lustrates the ntounting sc.heme. A thre,e point suppor:t system rvas chosen to minirnise the bucklir.rg of the plates. The ledge thal supports the plaLes h¿rs beerr machinerd such that there are three areas v¿hich rernain higher tha-n the adjacent areas. These raised areas are at l20o spacing, 0.5run hi-gh, 0.5nm r,'ide and abour lrnm in length. If there is any flexir-rg in the inner annr-rli it ís expectecl t-o originate ín the re-gion of the ttonguest, so 'Lhe areas of plate supporL are rem<¡ved by 600 frorn ttre ttonguest. lfhe- plates are clampecl by applying pressllre on the p-l-ate le-clge directly above the support areas. The pressul:e j-s lc¡ca,l-ísed ¿Ìi these positions by nieans o.E ttrree snall tabs on the imrer circurirference of a 0.6mn thiclc br¿rss \,/asher. Ttre tabs on Lhe i-nrrer circr:rnference protrude over the plate leclge ¿rnd are pressed dorvn by a tc¡:inl¡-.Ledt spring washer under a beze,L ring. The t-abs are lcepE in registration by a lug on hhe or.rter cit:cumferelrc-e of l-tre rvasher" This lug Ís locate.d :Ln a recess in the ínne.r annul-us. hrith this tiounting sche.ne, it i-s still possible to buclcle the plate.s if the bezel ríng pressure is too great, ;rlLtrough sufficient pressure to firmly clarnp the plates can be applied withour obvious buckling. The brrchlíng of Lhe plate..s rLocier high pressure could be minimlsed if the support ledge on the plate was located near the- uncoaLecl sr-rrf ace. 66. 4.2 De-sign and Const-rttctíon of t.he Etal.on Enclosure 4.2. L Original Design Concept:s" l-t was originally proposecl that the lors resol-ut.ion etalon be cont.rined ín Ehe top of a tube mountecl direcrly above the hígh resolution F.P.L such thaE the optícal axes of the two F.P"I"!,s were colinear aud had a contmon f ield stop. For nightt.i-rne observatj-ons, the entire assemlrly'of the low resolution F.P.L woulcl ha've to be removed. A unít of lhis clesign r,/as constructed but laboratory testing indicated it woul,l not be suitable for use at the fi-e1d station. The t-ernperature control and thermal insulation were insufficíent and the mechanical shocks received by the etalon while moving the unit caused se-vere loss of parallelísm in the eËalon. IE was Eherefore decided that the lor.v resolution F.P.L rr¡oulcl be located permanently to one side of the high resolution F.P"I. and that optíca1 coupling be achj.eved by tr¿o mirrors and two lenses. This arrangement is described ín detail ín Chapter 5. In t.he redesign of bhe eEalon enclosure, careful al-tention çvas paj-d to ttre quesl-ions of temperatr-rre control ancl vibraÈion isolaLion. /+.2,2. Gener:al Errclosure Descriptíon. Figure lr.3 is a schematic diagram of the low resolution etalon ancl its encl-osure. Basically the enclosure consists of a wooclen box, with l8mm th-Lck rvalls, paclced with polystyrene foam and contains trüo steel ch.ambers wL'-ttr wa1l thicknesses of 6mm. The outer chamber serves as a mount for a m-irror, interference filter and a parabolic reflecl-or as we-l1 as being an integral part of the temperalLrre conrrol- scheme. The etalon Ís conE.ained in the inner chamber. Both chambers are supported by a woclden partition locaÈed near the centre of the box. Both the upper and lower box lids are easíly removed, permí.Eting access t-o var:ious parts of the F.P.I. The lower lid supporLs the frínge insulating foam window outer charnber filter reflecEor rvindow t-nneT chamber etalon focusing lens optical axís -+ ml. rror fringe viewing system ,t Figure 4,3 Sche-ma[ic diagram of the low resolutj-on ]'PI enclosure. The reflector and frínge viewing systern are wíthdrawn from the optical beam duríng observatíons' 67. 'box viewing sysLem clel.cribecl .i-n secLjon 1+"3"2" DeE¿lj-ls of tl-re support scherne- are üiscrrssed i.n secl-i-on 5"1' 4.2.3.'Iemper:a-ture- Cor-rtrol. Temperature stability ín the regiorr. of [he etalon is critical to i-E.s long terrn per-forlnance and ís achievecl by the use of he¿ters rr¡ouncl on the f-rvo chambe-rs, thick layers of irrs;u-l-atj-on and by the- largä [hermal capacÍ.ty of the inner chamber. The heaters consíst of niclìrome wire, errclosecl in plastic, wound on the outer surf aces of the chambers. All surf¿rces ¿Ìre heated l¡ith constant po\^/er per urrit area. The heater wire is helcl in position with alu¡rinium adhesive t¿lpe. This Eape is al-so useful for the in.Era-recl r:adiative ciecottpling of varrious coiliponents in the enclosure' Both the inner ancl ouE.er chamber ile¿rters rlissipate 30 waEts (maximum) and are con.tro-L-l,ed independently by using l-hernlistors mountecl in the chamber: ¡valls arrcl electrottic circuitry developed foi: the high resolution F.P.1. (Bower 1974). The resisEarlce of the sensing t-frermisto¡-is comparerl to a stanrlard. resistance and a c1i,ÊEer:ence signal -Ls developecl lvtii-ch controls Ette heater power. Tirne differenfia'tion of the sigLral perrnit-s i¡creasecl po\,úer if the fempelatu::e::us fal1--Lng rap:1<11y. The mains supply (21+Ov, 5Ol1,z) is swr'-tched by a triac [rlgger a1-- zello crossj-ng to nrin:Lrnise radiation irrierference. Over a cert¿lín ïange the círcuiU provicles pïoportion¿ri control of the he¿rter poe/er' Orrtsíde this range it ac-ts merely as an on-off sr.¡itch. The inner s1.rrface of the outer chamber is plated r¡ith caclmium and the j.lsulatÍng fgam surfaces are also covered ç¡ith alunrj-niunt tape to recJuce racliative coupling. The etal-on and the inner surf aces of the inner chamber are- blackenecl to plomoÚe radiative coupling, thus hasteníng the acirievement of thermal equiliþrium in this regiorr' The t-eirLperatuïe of Lhe etalon is monítored by a therrnistor rnounLecl on the outer ;rnnulus, The temperature is SeE at 30oC, th-is 68, beirrg between 50 and 10oC above the ¿¡mbierrt tempoÏattlre of the 'Lield- station. l-'he orrter chamber i-s mairr.tained a'L a telì1pç)rature about 0'5oC bel-ow that of ttre inner chantber. ithe temperature of the etal-on is rnalntairre.cl to rvittrin 0.003oC for periocls of up to abouL fotrr hours typically, and rvj.thin 0.02oC for l-nctelaÐ.lte:. perrocls. If the ambieni temperature f l-uctuates 1e-ss than 2oC, then ttre fi.gures cluote-rl above can be improved. llhe enclosure ha's a J-ong time constarrË ¿ncl after large tenperature changes (several "C) expelienced cluring po!.Ier faj-lttre ancl mairrtenance, Lherntal equilibríurn is not attained for B to L2 hours. Since the low resolution F"P.f . is locatecl just bc'.low the level of tl-re false ceiling (chapter 5), it is in a region that exPeriences large anibient ternperature incre¿rses rlur:ing sufilmel: days. Tc' minimise this increase, 75mn of polystyrene Eoatn is centented on top of the false ceiling an<1 dr-rring very hot- clarys, cool air frcm the []-oor: is blc¡r'rn j-nto the region of Ehe enclorìllre. 4"2.4. Inner Chamber. The consir--uctional. deEails of the inner c?rantber are il.lustrated ín Figure 4.4. The chaniber weights l4kgm, mos;t of the ntass being ín the steel base. The brass base.plate of the et-?"1on is suppor-terl on three c|oth pacls ancl ís not bolted clirectly to Ehe ste.el base beca'use of the possib:Llity of buclcling r'-ntroducerl by Ltre differing ther:mal expansion coefficients of steel and brass. A steel anrlulr-ls is }¡olted Eo the \^/ooden partition and Ehe inner chamber is supporte<| on it by three s[eel springs at 12-00 spacing. These springs form part of the vibration dampíng scheme desi:rj-bed belorv. DirecEly above the etalonts parallelism adjustment scre\^Is, small holes in the chamber allow a tool to be inserted into the sc-rews for parallelism ancl spacing adjustments without seriously upseEting the t-hermal stabilíty of t1-re etalon. vibration oil filled dash pot r¿ocden parËition bolt boilr etalon cloth pad wrre j-s Figure 4,4 The ínner chamber of Lhe Iow resulutÍon FPI enclosi-lre. (The FPI noË represenËeó in fulI The vibration damping system provides ruech¡.ni.c¿rl. isolati.on of Ehe etalon. Details of tlìis sysËem are shown in l'j-gure- /r.4. A piston in a dash pot containing a rnedium viscosíty oi-l pr:orr-Ldes damp-i ng to ahy verÈical vibrational motion, The resonant frequerrcy of the sysEem is ¿rbout 4llz and all vibrations arísíng from ¿rn impulse ar:e damped after only a few cycles. The sides o[ the piston are slightly rounded Lo permit small LatetaL motíons t'¡ílhout the piston seizing. The dampí-ng sysEem Ís located in and below a threaded cylinder. These cylinders act as levelling scre\ús during atrto-collimation of [he optical system lvhen the etalon plates are being set perpendicular to the optical axis, 4.2.5. Outer Chamber. The outer chamber is in t\^ro sections. The lovrer sectíon is bolted Èo the wcoden partitíon ancl r-he upper is locatecl by a steel ring buL ís removable to allow access to the inner chautber. The heat-er wire on the two secÌ-j-ons operaËe in paraltel . r\.s illustrated in ltigure 4.3, the lower secLion supports a fr:ont surfaced, 45o mirror and a single component lens, The nirror can be rotated ancl nlovecl verLically si-nce it is motl1l Ee-d on a cyl-incler r¡hích is a bearing fit inside another. The lens posi-t-Lorr j-s noË acl.l usrable. The upper seciiou of the chamber supports an inr-erfet:ence filter an Ëhat it can be pushe.d in and out- of the optical beam by a rod j-nserted Ehrough the box wa1-1; again v¡iËh no disturbance to the F.P.I. 70 4.3 Scanning ¿rnd Paral-le1:i.srn Cttrrf-rol 4.3.f. Intrad-uction. As the low resolution t?"P.I. lras not to b(: s:err¡o-'controllecl for paral-lelism, a scheme for the visual- assessment and conLrol of paralJ-elism h/as requíred. The degree of par.allelt'-sm can be assessed by Ewo rnet-hocls; a. viewing interierence Er:-Lnges aE infinicy (Fabr:y-Perot fringes) ancl cletecli-ng cÏranges in the fringe dianr-eters as one tscanst across the etalon. The correction required depends on whe,ther the fr:inges contract or e-xpancl" Ttle se-nsítivity of thís metho<í depencls on the apparant angular di¿rme-ter of the fringes and so is more- applicerble to eralons working at low orders if one rvishes to avoicl the use of trigh mergniEication telescopi-c optics. b. viewing fringes in the plane of the eLalon plates (Fizeau frí-uges) and assessin-g the eve-nness of intensi-ty acr:oss the etalon vrhen it is tuned to trarìslnr'-t rhe spectral line being used. 'Ih-is inethod is very sjrlrìsiEive and acljustnents can be made quickl.y. Irlovreve;,', the eEaJ-crit ntus-u be illuminaEed urríf ormly. lloth meth It l¡as desi-rable that the i-rrterfere,nce fiiter be. situated. i-n a region that was tenìperature controlled, easily accessible and be placed in a beam of litcle convergence. The obvíous place was just-- above Èhe low resolution etalon. However, this .rneant that sirLce regular parallelism adjustrnents mighc be requ'ired, j-t was inconvertient to illumin¿rte Ehe etalon wíth a source above bhe etalon but or.rtsícle the enclosure. A simple solution was t-.o use a He-Ne laser and illuminate the etalon via 71" a lÍght-pipe ancl pare-rbolic ref lector si[uated beLu'eelr the fi]-ter arlcl etalon. (!'igure lr.5) " Si.nce the :r:ef lector provides ttneven illumin¿tion across the e.talon, method (a) r'"ras chosen. 1his method recluires some i.nterpretative slcills to be clevel-oped by tire operator and sorùe appreciation of the topography of Che plates. IJith pra-ctice, charrges in paralle-lisrn o'f- X/l00 can be clel-ectecl; this being arlequate since the p-Lafes have a clefect finesse of about 30. 4 .3 .2. Fri.nge View-i.ng Sys Èem Since iË is clesirable tha¡ at leasL one fringe is always in the fj-eld oF view, Ehe viewing system must accept light from angles of incj-clence over a range corlespoftding to abouE one order. Using a He-.Ne laser: (À633nm) with an ol:der of inter:fererlce near 1140, Èhe viewing system ulust accepE angJ-es of about 40. The beam from the perrabolic reflector is dj.f Errsed by roughened perspex to ensure illuminating of ttre .Eull field of víew. 'I'he perspex ís not au isotropic scatterer and the beam stil1 has a goo<1 forr¡ard inl-ensity, assuring a bright image is rr¿lintained. ilhe opEics of the fr.lnge vie-..,ring system (Figure lr.5) consists of a lOQmm focal leLrgth lens rvtrich ruas chosen because it pr:ovidecl t.tre systern wiEh clímensic-,ns that were ptrysíca1ly convenierrt. Al1 Ehe opt,Lcal conponenls are mountecl in a tube that can be retr¡rc.ted from the optj-c¿rl path afte-r all adjustmenf-s are compl-eEed. 4"3.3. Electronic Controls Even i:E the F.P.I. \4ras to operaie \,rith a fixed spacing, iL is advantageous to trave a scanning capability in order to measlll-e the finesse ancl to determine the order. Provisions fi.rr scanning were originall-y incorporaLed ín the ele-ctronic control círcuitry ancl ín fact the F.P.I. no\^7 operates in a scanning mode. o"Dl:ical fi1¡re f r:oirL 1¿1sÐr --> parabol:Lc reflector dif fusing screetl *.---+ etalon f ocrrsi-ng lerrs mrrlor (2) mí-rror (1) Fígure 4.5 Schematic clíagram of l-he frínge .víer'ring system. exit, prrpil mirror (3) 12. The para1.lel.:Lsm is assesse-d ¿lnd cc.lntrol1e11 alon¡¡ tr,nlo ortliogonill axes, X and Y, as illrrstraEed in Itigure 1i.6. '.the piezoelectric ceratnic transilucers are labell,ed Y, X1 and X2, Iroi: adjustnìentsj along i:he ' X axis, rLl unit of volLage is applie Èo X1 apd X2. Thus the mean spacing is kept constant du.ring parallelism . adjus tmen ts Scanning is implemented by the applicatíon of time varying voltage Eo the PZT ce-ramics, the r:ate of increase of voltage at a.ny one ceramic is inversely pr:oportional to iEs piezoelectric c.oefficient. 'Itre control circuitry (Figure 4.7) provicles X-Y axis parrallelism acljgstments, seanning with inclividual gain sett:Lngs, coarse (n'I order) tuning r+ithout individual gain control and fine tuning with gain conl-r:ol. The input scan clrive signal can be.rmplified wiLh variable gain or r¿ir-h a fixecl gain. The 600V supp,Lies, developed for trse in the hÍ.gh resolutíon Í'"P-I. (Bower Ig74) are esserrti.ally h,Lgh gain operaticnal amplífíers prorricling an output propor:tional- to tire input currenL-.. The aniplifier has sumrn-Lng irlpu[s to permit the combinaLj-on of several inpr,rt signals. There is one suclr srrpply for eaclt 'piezoel-ectric cerarnic transducer. Ttre scan clrive signal is developecl exte-rnally and entets via a uniLy ga-Ln input brrffer. Tl-ris signal is then sunlned ¡¿ith Che 11 .c. tune signal.. Lolv gain ís appliecl bo the scan clrive if the F"P.I. r'-s scanning ín synchronisrn rvíth Lhe high lesolution l¡"P.I. The ga.in is vzrr-Led by G1 to ac-hj.eve exact wa-¡ele-ngth synchronism as explainecl in section 5.-l .2. The scan clrive is aniplified with near unity ga-Ln if the etalon is to be scanned over about an or Gxt, Gx2 and Gy, se.t the ratio of the vo|tages appliecl , vía the 600V supplies, to the ceramics Xr, X2 ancl Y respectívely. The parallelj-srn and tune signals are developed from resistor: P1 piezoelectríc transd,ur:er DS dífferential screw I IDS Pr (xz) PT (xr ) Y axís DS (Y) IPT X a:xis Figure lr.6 Diagrarnatlc plan Virer,/ of ttre low resolut:Lon FPf- showing the piezoelectric tralrsducer and dif f erent'íal scre\¡l l-ocations and the X and Y parallelísm cont-rol axes. +3v BK 2K c) Y IK (,J) 2K ^l -3v fnput buffer 9K 3t( scan drive fnput 9K 5K G G 2oK v 1K Gt Y à Figure 4.7 llhe low resolu tion I'PI control ci-rcuit::y' Orttputs Y, Xl and X2 are inPuts to the 60Ov suPPlies of the PZT transducers. Control f unctions; (a) change mean separation (d) ireaa with plates Parallel, (b) X axis, (c) Y axis and separation withouL maj-nt eríning parallelisrn. 73. di.zider netrvorks connect.ecl to stabl.e volLage cupplies. The i:esi-stors are of the metal foil Cype and higir stabil-ii:y, mu-Lti-turn potenti-ometers are used. The performance of the circuit is such ttraL it cloe.s not' contribute sj-gni-ficantly to the long Lernt separation and parallelism varial-ions of the I .P " I . 4.4 Operatiorr and Performance 4.4. L" Sc¿tnning the Interf eroineter . The interferomefer must be set to scan with the eËalon p1-ates parallel, hlthough Che pla¡es may be set paretllel at Ehe beg'it-lning of a scan, they will not, in genera-1-, be par:rlle-t at some other point j.rr the scan because of the variation of piezoelecEric coefficienLs between the ceramics irr the etal-on" Compensatj-on must be. m¿rcle for Lhis var-Lation. The etalon is set scanoing parallel by the followíng methocl" At drive is the beginning of a scan (at lorv scan voltage) " the scan stoppecl and Ehe p1-ates are set par:a.Llel using the X and Y parallclism controls. A positige gc¡ing scan voltage is appl-Lecl and the scan ís again stopped neaï ü¿lxirtllm scan vo-Ltage. At th.Ls po,lnt the pJ-ates are again seì: paral-lel by acljusting the voltage applied to each t;:a.nsducer using the tgaínr pot-entiometers Gx¡, Gx2- and Gy (Iígure 1+.1). This 1'rroceclu::e is repearted several- Cimes. Smal1 irLodj,f-Lcatirtns to lhe control setti¡gs are macle rshile ttre ínter.EeroLneter is contínuously scanning. Extensive tes¡s on the scanning peïformance of the P"P.I', nacle by R. Basedow, indLcate f-trat beEter performance- is achíeverl if the interferometer is scanned continuously rvíth a Eriangular sc¿Ìrì. dríve tfly-'backt havi.ng slope ratj-os between 1:1 and 15:1. The rapÍ-cl experienced r^¡hen lhe interferoraeÈer is sc¿rnnerl'¡iEh a sar¡7 tooth drive is detrirnenta-l to its long term perf ormance. This ís a rlore :important consícleration j-f ltre j-nt.erferometer is scanned over an appreciable 74. fract:lon of ¿rn orcler (I/3 or more). Durin¡¡ normal dayglow observa,tj-ons, the lors resol-ul-ion F"P.L is scannecl ove.r about 0"06 of an order and so the 'fly*backt problern is of no si.gnifícanr:e. 4 "4 "2.. Instrument Prof ile Me¡lsurelnents . Measurements of the insLrur,rental profile of the lol¡ resolution l-.P.I. are macle to âssess its stabrí-lity, rùe¿Ìsure finesse, cleternine the or:der anrl Eo obEain a digitized record for comp'.tter analysis. Thêse measurernerlts are made using an expe-rimental conEi.guratíon as ill-ustrafed in Figure lr. B. The deÈector, a PIN diode, Í-s situated at the focus of the objective lens. The size of the diode is suc.tr Ehat the profile is not aperture- bro¿rdened, and its shape is thus determined by'the convolution c,f the Airy and clefect functions. The low current amplifi,er (HP-4254) has a narroe/ banclr¡idth necessitating large period sc¿Ìns ('u2C0 secs) if distortion ís to be avoicled. If a digital record -Ls recluire.ci, l-he si-gnal is digit--î,ze-d by a vo1-tage contr:ol.lecl oscillal-or ancl acc,:umu.Late-d in a multichannel artalyser. The nemory conte-nt of the ilnalyser ís i:hen rùriLten olr tape by a digítal cassetEe tape recorder for subsequent analysis" The scan ge.nerator, developed in this -Labol'atory, provicles a trí-angular scan signal of variable ampLitude and period. The si-gnal is íncremented in steps; the number of sEeps per cycle being variable but usually set to match the number of channels:in the analyser" The gerier¿ÌLor also provides clock pulses Lo advance the anal-yser rrlerìory address at the same rate as the scan voltage is increnrentecl; the zero channel address being synchronised rvith the beginrring of the scan" Tlie scan drive can be sÈopped at any point in the cycle and then restarLed a.t the same point. This facility is used rvhen arìjustments are beíng nrade as describecl in section 4.4.I. ilhe versatility of the scar] gerrerator is rveil suitecl to the type *Ë laser or sPectral sour:ce Ç> dispersíng lens . diffusing screen etalon Ëo piezoelectric transducer v scan dri-ve waveform PIN diode scan rror generator X_Y plotÈer low current amp. rnt¿1ti- channel ^lD analyser Fígure lr.B sche-natic of the experímental arrangenent used to measure the ínstrumefit profile of the low resolution IPI. ?q of measurement-s that vrere requirerJ. to be rn¿rde on the F.P.I" All neasurements clescribed in sections 4.4"5 and 4./+,6 we:ce rtiade w-Lth this sys ten. 4. 4.3. lelti¡s_lbe_order. It ís important in Ehe dayglow experiment that the orcler be set 1-o 'F2 orders cf the desirecl value (section 5"7. f). Coarse setF-íng of the order (1f00) is achieve :lmaged on a screen by a long focal length l.ens (f,'r,28Onun) . The or:dcr is given by (),o¡2 M (4. t) .r2 D2- "i+1 - i where D. is the diameter of the ith fringe t- Finer adjustments recluire the interferometer to be sc¿rnnecl over a range of about lli orders aU À630nn and the illumination of the eta-lon rvith several dif f erent spectral- sources. The scans are recot;de X-Y plotter and the order is determined by the relat-ive posi-tions within the scan of the various spectral line transtnissiorr peaks, Usíng the Na doublet at À5$9nm, the order can be set to t10 anci by using Hg (À546nin) and the tle--Ne laser (À633nrn) as we1l, the order: can be set unl.que1y. By using all the spectral lines mentioneC, the errors caused by phase changes on refl-ection varying viith t,ravelength (ir- sma1-l eff e.cL with ttre present coatings) , píezoelectric ceramj.c non-lirreariL:tes and rneasureûtent. limitatiorrs do not detract from the uniclueness of the order deEermí.nation. 4.4.4. Piezoelectr-Lc Ceramic Characteristic.s If an F.P.I. Ís sc¿lnned by using piezoelectric ceranícs rvithou'r the benefit of t-he parall-elism and separation beí-ng servo control-leti, the characterístics of the ceramics musE be well unde::stoocl . Extensi.¡e tests were made on the piezoelectric ceramics.used ,'ln the lorv resolution F.P.I. The r:esulr-s of these [ests Lre. ,lescribecl ín Apperrdix I. 76" l¡ /+.5 . I¡ines;se }{e¿tsttrettlents " The finesse is rne¿rstirecl rnain-Ly âs an assessnen¿ of the qualj-ty of the optic¿r1 flats. IJsing the experiment¿rl at:rangenìerrt- descrjbe-d in t+ I orders aL À633nm. section "4.2, the F,P.I. is scanned ove-r about " 3 Ìt'igure 4.9 shows trlro transmissj.ol peaks of the [Ie-Ne laser:, the eEa-l-on lraving been scannecl wj-th PZT-| cerauiic tubes. The prof iles are recorded duri-ng the íncreasing voltage half of the- sca¡ cycle. The piezoele-cËric ceramicts hysteresí-s câLlses the n -l-1 orcler pealc to be broader than the m.o peak (Appendix I) . The laser ís chosen as f-he s1>ectral sorlrce because of its narrow line wictrh and its; r¡avelength proxirnit-y to À630nm. The finesse resulting from each peali is measured graptiically ancl Table 4.1 presents a series of measurements Lrsj-ng PZ'l-4 ancl PZ'I-SU tubes. The larger hyster:esis of PZT-'51I causes a greafer dj.f ference betweerr the m tl and the m order finess;es. O o The finesse being ineasure'l .r-epresjents the- wi<1th of Ehe- profile re.sulting from the convolrrtion of Èhe Airy and defect f uncEions' The spread of the values in T¿rl¡le /r,1 is an indication of t-he repeatability of alig¡i-ng the plates and the set[-'Lng of the í,ntlividual piezoelectric transducer gains. It is proposed that rlhe ave-rage valtte of ¿rll Ìulne PZT-4 results be taken as the besL estimate of finesse-, thus N = 25 ! 0"6. This implies a def ect f inesse of al¡out 29 and. a clef ec r. s-Lze of ¿rbout À/60 at À630mrt (assurning a spherícal defect). I¡.1¡,6 ancl SeparaELon S'tabil-iiy. " Para,Lle-Li-sm l{ean The best m,ethocl of assessing the nean sp¿lcing stabí1ity of an ll .P.I. is to períorlical-ly scan â spect-ral line and ¿ccuraLely deLermine iLs pealc posi¡ion. The norrnal operation of the low resolution F'P'I' is such that this is not feasible" The F.P"i. is only used Èo scan a êper-:tral line when the orcler is bei-ng set ancl this is usuall-y afEer Èhe eLalon has bee-n mechanically acljrLsLed and has; experiencecl a large 1.0 oz. U') an E tn z_ tr 0-5 cl i,t-u J É: oz 0"0 ffio*'! rb ORDER (wíih the iow resoluËion FPI measured aÈ À633 nrl. Fieure 4.9 The instrument Profíle l{F very large) of The profile was recorrJed in the increasing voltage half cf the scan di:ive cycle as índicated by the arrows. TABLB ¿r.1. LOI^I RESOLUTION FINES SE }IEASI]REMI]NTS Finesse, NU, at orders of interference rn ancl rno*l using PZT-4 and PZT-5H piezoelectric ceramic tubes' _ PZT - tt PZl 5H m m +1 m + 1 o o o 23 26 2.4 26 27 24 27 z2 25 23 26 22 27 24 26 2l 26 25 27 26 average 26.3x0.8 z4.3tI 2.6.3xO.5 22xO"B 77. Èemperature varLaE.ion. Since the F"P.I. requires sever¿rl cla¡rs to settle down, Èhe stabiliÈy during these periods is not a goocl indicat.ion of :i-ts typical performance. On one occasion, the F.P.I. \^/as lefL scanning over l1', order:s for two days an Durj-ng normal operation, the mean separation stability is assessed by noting the tËuner setting variations when the l'.P.I. is tuned rvith the white light souïce (section 5.7 .l). It is Èuned every 90 minutes and typical drifts al:e about 0.6rrm per lìour or X/1000 pe.r hour ¿rt À63Onm. This is about the accuracy to which the F.P.I. can be tuned. This result is sirnj-lar lo thaE obLained by R. Basedow by periodl-cal-Ly scanníng a spectral line source cluring normal F.P.L operation. Parallelism stabílity is more difficulc to assess alttrough sone indícation is gí-ren by noting the varíation of the X and 1 axis poËentiometeï set'tingä. During observations, ttre parallelísm is checkecl but not necessarily changedevery 4 hours. During the cour:se of one day, there are typically no signif icanE acljustments requj-r'ed- The F.P.I. can be, left for several months and adjustmenf-s less Lhan À/50 on each a: CI]APTER 5 TFIE DUAL ETALOIT FABPJ*PEROT SPECTRO}TETER DES]GN COI'ÍSTRUCTIO}.I À}ID OPERATION 5.1 Introduction The previous two chapl-ers have cliscussed the two F"P.I.rs to be used in the dayglow experiment-. In this chapter Ëhe design, r:onsËrucÈion and operation of the clual etalon F.P. spectrometer is presented. The dayglow experiment reported here rvas under:taken to develop reliable techniclues of measuring ther:mospheric temperature and rvind velocity during the clay, so that the fut-l dj-urnal cycle would be open to experimental i.nvestigation. ConsequenËly, the two I.P.I.ts have been couplecl together irr such a rray that tlìe system has a capability for continuoLrs 2/+ hour operation. A-l-though many systern parameters were ctrosen rvittr specífic r"f".rerr.. to the À630nr,r emission of atomic oxygen, [tre clual eEalon Ir.P. specrometer is easily adaptable io <¡ther racliations such as Lhe À558nrn emission of atomíc o: 5.2 Mechanical Details ' The fielcl station has a row of hatctres down the centre of íts roof . Observations of the sky are nade through the'.se hatches ancl the síze of Èhe low re-solution F.P.I.ts enclosure and the desirability of haví-ng it permanently located to orie síde of the high resolution F.P.I. meant that the dayglow observations would have to be made through the hatch adjacenL Eo the one used for níghtglow observaf-ions" The desígn of the hatches was such ttrat this re Th.js is achíeved by supporting the l-orv resolutiorr F.P.I. enclosure on a fíel<1 staL1on.roof -periscope \ false ceiling coup,lÍ.ng low resol-ution FPI sy sLeflr enclosure fringeT viewing optics supporE frame hígh ::esol-ul,lon FPT énclosttre its dayglow .'llgft.tl-.t- Schematic Ciagrarn oÍ tire- clual- etalon IPI in conÍiguralion. 79. frane ntacle of 50x6ntr angle i.rorì as :i-lltls tr¿r Lecl ,i,rr Ïip;rrr:r: 5 . I . One s:id.e of t¡is frarne ís br¡lted Lo thei fra'me of thr: high resol-ution T.P,I. enclosr¡re. The low resoluti.on [r"P,J-. enclcrsure is supporied on a crad1-e th:rt permits thc enc-l-osure to be rotaLed about three- o::thogorlal a)iÈs centred on F-he focal poi-nt of the iot^¡ res;cl-utít>n F.I'.I" objecti'¡e- lens. This facility is usecl clu;:ing the al-Lgnment proced'-rrc'- (seclion 5.3.3). The encl-osure penetrates into ttre h.atch ale'-a such t-hai- the top is only 2cm below the false ceilirrg on rvhich the periscope is mc¡unted. Part of the false ceil-ing is removable to allow easy access to Ehe periscope ' , An electronics rack, mounte 5.3 The Optical SYstern 5.3.1. The 0p tical Con f-i.gurat:í-on. As discussecl iir section 2"6, an ûtal-on sepnrrat.i-on raËío Of rrbout l0 to l was chose.il ancl frorn ecluai.ion (2.'i9), the- foca-l. lerrglh of tha lorv resolution it.P.I. object-ivr-ì lens is recluirecl to be 3o0iruL:i' (:ihe focal lelgth of the high resolulion 1,'.P"1^ objeclirre lens is 970trm). lhe two etalons are conibirràC jrt a !positi-ve-Irositiver systelrt as Ehe i-nter-etak¡n clist:arìce r,¡as not a prj-me conside.rafion. In facL, the structural de,t-ai-l-s of the Êield st.aEj.on (section 5.2) l'rere suctr l'hat the i-n[er*et¿rlon distance had t-o be ex[endecl' The cptical sysLem of the drral þ'.P.I. is illustrated in lrj-gttre 5.2, 'in'Iable ancl deL¿r.ils of the 1n,livic1ual c.omponents are listed 5'L' The cornl:onents 1, 4, and 5 are síngle conponent, p1-a.no-convex lenses' Componerrt 3, the f Íelcl lens, is bi--corrvex. 'lhe dual- F.P.I. operaEes over a very narrol'r Spectral range so there Ís no rree- 1 2 l t- b s4 3 Op tícal Compon.ents 1 300rmn focal length, plano*coilvex' f6 2 7 fr:onÈ surfac:ed Ag mirrors' 3 63mm foc.e1- lengt-h, bi-convex .r.6 4 5 137mn foca.l- length, plano-convex, 6 stop r-hat defirres f j-elil of vier'¡ ()ô 970mrn focai 1.eugt-h, triPl-et' f 6 I 3OOinm focal ,1-e.ngttr, fz Lrig,rrre 5"2 The opLical. sYstem of the c1ua1 . Ï¡'or níght:g1.ovr observaliorrs, e Eal.on ''ì'Pl. mirror 7 is temoved and those. coi"rponenÌ:s labellect 10 a'e 1nsÈa-l'lecl" B high resolution TPÏ 9 photorriul Lípller 80. ruittr ne-ar axís bearus;, conse(ll1ent"1-y s;i-ngle compoilertr 1¿.:irses vere: considered sai:isfecbor-v. The hi-gh r:esoluLicin t'.P=1.., cles,ignerl pr:Lrnarily for nigtrL tinLe observatiorìs of rhe À630nrir anC À558nrn aírglow, has an achromatic t-riplet object-íve trens, B. '.lhe extensj-ori of the inter_etalorr clj-stance Ís achieved by the couplí-ng lenses , 4 arrd 5. A double -Lens system rüas chosen to provicle gre-ater flexibiliEy in design and acljtrsrment (sections 5.3.2 and 5.3'3). The inte-i:ference fil.ter acts as Lhe erit::ace pupil of the syct-em with a useable dianeter of aborrt 46mm, The fielcl l.e,ns, 3, images the interference fjlt.e-r onto Ëhe high reso-l-rrtion etalon. This mr'-ni-nrj-ses f ignett.ing ancl provi.cles a useful al:ignnrent cliagnclstic (section 5.3 '3). The two mirr:ors , 2. aod 7 , were originrrlly fronL surfacecl aluminium. A í:hirct aluminium rnirror r¡as userl in the periscope (r;ection 5,4). tleasureme-nts oI the relative tlansLnissj.orr r:f the siirgle ancl dual F.P.Lts i.ndicatecl that the cltr¡i.l F"P.I" had a Lransnissiorr much l.olver than e; The rnechanical secLions o.E ttre F"P",t.rs hacl been cle:s:igned to accouoilaLe 6mrn thick, frorrt surf aced i¡rir:roi:s. FronL surf acecl sj.lver mirrors worrld soon deteriLrrate ¿rnd bactc surfaced rirírrors r.lou:l-cl have required rnajcr: mech¿rnical ¿rlterations" Consequeritly a nirror r^ras designed a¡rd consl-rucl:ed as foll-ows. Silver rvas evapor¿rted cnto the b¿rclc of lmm thiclt r;heeL o'[ glass. tJ1. Thj s was Lhen coct¿rc tecl i:o :r 6urnr LÌrì-cic, ei .l -i.p L:.i-c¿1,1. nt:il-'.i:or: b1.ank, 'L¡.1;rll'. ttre edges r¡/ere tilen ground an<1 sealerl . 'i'ire pravicies the ¡íl-ce-ss¿ìt:), nech¿rnicai .rigicli t-;v f or ti-tt: tnr-r:r:or , Tliir+ Eron L surf a ce \üa.s coate \./ere coated with a \/+ anti-reflection liryer, as l'Ieíe the t-l-rre-e çv-indo'.vs above Ehe etalon, The ::emaioing -Losses ensure tire suppr:r'-ssion of rnult-iple reflectioLts betrseerr the et¿ilons. The- Lranemitt¿rnce of all opticerl compoRents, rvinclows iucluclecl , j-s est-imatecl ¿rt about 0 '6. This es;Eimate exclr-rcles scattering by dust- 'llhere are F-wo fielcl stops in the systcm (3 arrd 6). Tlhe sLop nurnber 6 is the smaliest, and so clerfines Lhe fielci of vieru of the spec troäle-t-er . T[re system is conr¡e.rLecl for: nlghtglow otoservaLj-ons by relrrovi-ng nirror 7 ¿,n.d. instal-tÍng t-he pe'r:i-s;cope anci c,omponrlnLs lal¡e1led 10 irr Fígune 5"2" 5 .3 .2. J'{ecLr¿rrric¡,r L Dc'-iai I s of the Coupl.ing System - In order 1:o âccomoclal--e the -Lor,r resolution Ir,?.Í., the topinost strricÈure of the Trigh resol-ution Ir.P.T-. tracl to be- consíder:ably g0o modified. The cptical ¿1.: et-.¿rJon being sr-'bjecte cap FPI fibreglass low resoluti-on cylinder enclcsure wall teflon bearings concer-Lillered p1aËe rubber tube coup'! ilo lenses /\ 7 5 6 .f 4 ( support for hígh res. high resoluËion FPI FPT. encl-osure wa11 Fi re 5.3 Mecl:anicaldetailsofthecouplingSystemusedinthedualeÈa1onFPI. oL.o1 attached to'Ehe cyJ-Ínc'lcr wh-Í-cl-i corrf¿.rius the h:i.g,h res;olution et-lllon (l-igure 3"¿t) " lllhis ri-i::ro'r-' is¡ ulounl:,:t1 t-o parnr"i-t: 1:cLaEjr:na.J- ancl long,L-- tticij-nal adjus ttnenLs. The ouEer fj-breglass cyl:Lnde-r:, i,-ir,i-ch j-o. Lhe nightglovr coirfiguraLion support.s the pe:riscopr¿ and rotal-eíi on teflcn bea::íngs, is fixed in posiËion for dayglow oþserv¿rtíons. A cap provides aa efEective light seal . 'Ihe coupL.i¡g system ,í-s connected t-o this cylincler by a concerËínered rubber Èlrbe. The couplirrg system conÈainíng -Lenses 4 aûd 5, i.s bolted f-o the frame of the higtr resolution F.P.I" enclosrrre. Each lens :i-s cotrLainect -Lrr er cylincler rvhich í.s at-t¿rchecl to an end plate. These e-rid plaies conÈain the field stops, located j-n the foc¿rl plancs of 1e-nses 1¿rnd B' The trvo cylinders are sleeve l[ean separaEion drj-Íts o.[ the iligh r*eso]-u.Lír¡n F.P.I" arÉl moniL.ored by removing both lens cy1 Lrrrlers and íllumina1::i.ng the fj-eld stop vrít.h clíf fuse ligirt frorn a Iìg-198 lantp as íllusErated in FJgur:e 5,1+" T,he size of tl-re cliffusing screen ensures that Ehe entii:e fielcl of view of the ttigh re:soluLíon F.P.I. is ill-umínaLe-d. Tests i¡ 5.3,3" Älignrnent Procedtrre. The aiigrment of lhe- optical system j-s I I diffusing I screerrs I i i I i I ¡ i I I I .urrJt. 1:.de I I Figure 5.4 Arra;rgenent ior calibration of the high resolution FPi wiËh the Hg-i98 source. Mechanical Cetails as in lti.gure 5.3. 83. be renoved to permit :il.luru-in¿rl:1on f ron belirw Llte irígh lesol-rr.l-iç:11 etalon, The alignnrent woulC be str;righl-:[orrva-rcl aiLd ¿,-:cr-',i'¿Ll:o if ¿t .J-aser beam rvas shone frorn below a.Long the optical. a: The optícal component-s associated with eactr F.P.[. ¿rnd the coupling syste,rn are focr-rssed and aligned using autocollimation-. l.Ior¿ever." the optical axis of the hígh resolution,tt'P.I., as projected through the cou,pling systern has to be- colinearr'¡ítl-rthe opEical etxís of che lor'r resolution F.P.l-" T.his is actri,evecl by pi'votting the lorv resolution F.P.I. axís about the focal point of lens 1. Thís alignment is car:riecL out lvith the low resolution eta-l.on removecl an 5.3.4,'Ihe IrrterferenceÌ¡ilter. Itlhe fjlter selected for the ctayglol.r obr;ervaticrrrs l.ras rnanrrt-acttrred by Spec.tro-Fi1m, Inc", U.S.A. It is a t.¡/o period filter, blockcd to l1tm and hrls a ful1 aperture- pe-alc lransru.Lttanc€'- ol 0,47. PleasuremenËs of Ehe fj-l.ter pro[ile have showrr Ehe r,¡idi:h to be 0.3:!0.01nm and che peak tr:ansm-Lttance occurs within l-0.025nm c¡f the avera1e over lhe full 50mr,r aperture, Àlt?rough narror^rer sÍngle per:iod filters are availab,Le-, the E.ransmittance in Lhe w:Lngs of the ban-clpass ir.; highe-r ancl undesirable for this experíment. B4 .Pei:j,si:ope,." 5 "L The The- proposei[ ciayg io',s o b s r*Lv¿rtio na-L s shr:me r:eqr:Li-L:er,i th'.: s-,ilntpI j ng of the sky anci sol¡r:: speotra at tiine ir.r1,t:rv¿r-Ì-s over: ç¿hi,cÏr the clopple:: ¡;hif t of the ¡;o1ar si)ectrrun ancl j-ru;trunLenta.l dr-Lf rs would have a iregligible efíect, Ttris is achievc:cl by the use of a periscope illus Era.led ,Ln F:i-gure 5 .5 . Às jilustraLecl , r-he spectr:omeLer would be sarnp-Li-rrg the solar spectr_uil by observing the d:iffusing ecree.n rvhich is illumirrated v¡íth dire-cl sun1.í-ght. The zenithal. ,Jrientation of the scre-en is f jxe The per:isq¡.per i.s construcl;ed of IjRBP tubr'-ng rvhictr is rirade lighl: tighl- by rhe applicaticn of a layer oÊ a1uminj.ur¿ ¿idhesive Ì-ape orr. ttre outer surfaces, Äl-1 inner su:rf¿rces are ccvered ruith black vel,l¡et c1olh to mirrimisc stray re-8,1-ecti-ons . llhe t s<¡1ar tube I cc¡ntai-ns a ft'ont surfaced Al mirror and the diff r-rsíng screen. Tlhis [ube slides :r1-ong t tverti,cal- a cr:osÍ; Lube- t tha E is inser i:r¡cl Llirc-rugh the tubef , In the posit.Lon illus Lr¿r t-ec1 , Lhe ' sky rarcliatj-on is sealecl oJ. f . Ttre c.on Lact betryeerri l-he w;,Llls of the t so.l-ar t--ubet ¿rncl the t c:ioss tu[¡e I províries a,n ef f ec l;ir¡e se¿:.I f or the sol¡*r.' radi¿'r t-Lon r,¡hen the t r;olar r-ube | -ts t t drivr.-n up the cross tube . On a coniman 5,6.4), a smerll noLo-r clL:ive.s the tsolar tLlbet up the tcross t-ubef rrntil- the. m.Li..L:or is a.t l1-re ¡,osj-F--lou. m¿rrke<1 b,v thc dottecl -Line in lr.igrrrr: 5.5. The pr:cÍ-tion oF t-he'solar tubc.-f is sensed by inic-r,osr¡itches at boLh ends of .¡l-i:s travel. Ilpon colEact, [he rnoEor cu]:renb i-s cut off ancl cert¡rj.n c',onrl,Lli-ons are set in ttre controi electronics" Af te-r the current -i-s cut oJ-f , the r'-riertj-a of Ehe systr:m tensior:rs a spri.ng :in the lnotor chai-n drive " The gear i-n¡3 ratj-o of tl-re motor :is tiigh enough lh¿lt the spr:Lng then keeps the solar [ube firmly irr position. cårp I ù mt-rl:or t t sol-ar tube diffusing screen S.R.B.P ;jv'binz I vert ícal tube I mounting / block for mirror mo bor CTOSS tubel alignment screü7 eflon bear:ing false ceihlng lov¡ resolution TPI enclosure Figure 5.5 The períscope used ín dayglow observ¿rtions' 85. Às a compromise l,e-tween tj:re van lìli:ijn ínt.erisì-t1¡ en-l:ancement anrl ttre lhazí.nessr oi. t-.he clay sky at hígher zeni.Lh ang,Ler;, it ur:rs Cecicle The slcy is observed via a mirror on top of Ehe per:iscope. This mirror, fixed in angle, is e,nclosed j.n a cap rvh-Lch can be rotatecl Ín azimuth, independent- of the azj-m.rrth¿rl orienLatj-ou of the tsolar fuber. Iìor zeniLh observations or observations of the r'rhi.Êe lighf source (sectj-on 5"7 "L.), the cerp conEaínj-ng the mirror is rernovecl . To prevent direct sunJ-ight being scattered inEo the periscope when observing the sky., a shield of bla,ckened cardboard, rnor-rrrted on Èhe câp, is use-d Eo casE a shador,/ over the aperture; The periscope is located by a short aluminiuin cylinder attachecl [o Lhe false ceiling. A teflon pad in tlie outer r,¡al1 of the perisccpe provides the be-a,ring surface, for rotatíon. A syste-m of j-nterl-e¿rvecl cylinclers provides the coupling to the low resolu{:ion T,P.I. encl.osure" Tlris coupli.ng is li-ght tighr but ttrere Ís no physi.cal cotLl-.).cL betv¡een [he periscope and the enclosure. Th-Ls preveilts pe]:iscope generated ví.br¿¡ t- Lorrs; a.'Ff r:ct í-r-ig the. e Ealon. 'Itre ruotor drj-ves the rsolar: tubet fr:on one- pos1ticn lo Lhe otirer in ¿rbotrt 2 se-concls; an.c1 is ¿ictivaLecl about 45 iimes per hor-rr. Orre unsaE-Lsiactory parL of the design \^/as 1-Lre laclç of bafflcs ín the tsol.ar t.ubet. \littroul the velvet clr¡Eh" ref lections Í::onr the Lube w¿rl-l.s retsu"Ll-ed in severe spectrai- distorLions. I^lith the velvet cloLh., Elrese dj-stortions âre still presenE but they aie very sma-t-L and easily ta.ketr inÈo accouitl-. (sectiori 7 .6.I) . The c1-Lffus;.Lnr3 sc-::eeo iLl-uninatect rvj-t-h dj-recÈ sunJ-i,ghi, ¿llso receíves radj-aLíorr from tLre sk;r but me¿rsurerìents j-ndicate ttraE the contribution is ¿¡boul: 37" o'f L-,hr: Ì:oÈa1 signal.. ThÍs amoLlnt has a neglígible- effect orr Ehc derivecl ternper:atrrres ancl wind vel,ociry. It would lor,rer the est-Lmated line emj,ssion intensÍ.Ly by about 37". 86" 5 ,5 Ìrl-toton Ðeleci-íon -5 "5 .1 . The Ptro Eomr-r.i-ti 1-i-er. The elements of the ctetecf ioll sysl-em clescrilbe.d .it Lhe next: five sections we-re assembled for use in t:he original worlc of the higtr resol-ution !"P.I. (Bcwer 1974,I^lj-Llksch 1975), bttt scme modífications and extensions to the exístirrg equípment vrere rcquired for the daryglow experiment. The ptrotornul-típlíer is an EI.II 95588 tube rvith an S-20 (tri-allca1í) photocathocle. T.ris is operated aL a currenC gain of ab¡:ut l:c 106 using a dynocle divirler ch¿rin supplying equal inter-clynode '¡oltages and a l50V zener dioCe betr^¡een the cathode and Ehe f irst d)'nocle " 'Ihe Iast two clynocles are decoupled and Ëhe last dynocle i-s decoupled lo grouncl because of ttre use- oE pulse corrntiug during periods of lorv sorlrce radiance, The phctonultipi:Ler is equipped rvith a ltirschfelcl corle enha.ncement rlevice which i-ncreases the effective cluanturn effj-c:iency at À6-?Onrn fron about 67" to L17". 'Ihe photomultiplier chambei j.s aEtached to the sgspe'-ndecl cylinde:: r,¡hjch contains Lhs <-rptics ar'.d r:ta]-on of Ehe [rL6itr resok',tion F.P,1. The photomultiplI'-er is cooled l;y punping a rnetl'ranol-v/ai-er mixt,,rre througll a cooling .iacket sLtrrcLlncling the tut¡e. Typ:1cal photocathocle tempeïatures are -l.6oC, resu-1-ting in ¿r res;i-tltral- dark count of about 30 pulses per second. 5.5 "2. Ðj gita1. Detection. tJnrle;: conclítions rvhere Ehe si-gna-l from t-he phoCcmrrltiplier Ís comparatble to iLs clarik current, digit.al detection o:r ptrlse cüur:iting is known to be super.í-or to analogrre detection' A1-Ehougir cligital- detectj.on has no ¿rdvanLage over ana-Logue clet-ecLion ín the clayglorv observations because of ttre large signal t-o darlc current (>1000: l), it røas originally planned to use cLigital cletecEiort for da-ytinte, tvri-l.ighÈ ancl nj-ghÈ time observatj.cns . 87" 'Il-re pu1-se ratc- from Ehe ¡;hotomultip.l.ier v¡hen cbe lìpr-'.(:trLrmeter is oltservÍng the rii-ffus;ing screerì illumin.:itecl b1r Ll-re- ':j.irect sunlighË ca'n lîeach 5 x 105 ccrrrrEs per second.. This signal:Ls ro be accumu-Late-d ín a nultichannel analyser (sec'.Lion 5.6"3) that requ-Lrcs ¿rn avei.ege of Bps to add one çiou.nt to i!:s memory. Thjs dead time límiLs the s,i".rnal puJ.se r¿tte into the anal.yser if clead time- losses are to be íns.i.glri,fí,cant. z\ scal-Lrig circrrit r.ras ccostructed to achieve pulse cor.i.nt rates comt)atib-Le wiEh tlte tie¿rrl tr'-me of tfie analyser. The prearrrpli.fi.er: and discrj-minators are as used by Borver (l97tr¡, however, the outprrt urorlos't¿rt¡le r,¡iclth was reiluced to 60ns and adçlitional circuitry was add.ed to Crive the 3 metre-s of 50Q cabie Eo the scaler. The preampl.ifier h¿rs 50dB of gain and a b¿nclwidth oÍ. 2-5ùHz" The upPer ¿rnd lower ciiscriminators are high spee,J dif fe-ren1:-i.al comparatol:s. The lower level is set to ¡naxj-nr-i.se Ehe signal to noise ratj.o undcr low s;ignal conclitions. Ttr-í-s circuiE, i1lc'slralecl in f:igure 5.6, c¿Ln r"sqllre pulse pa.i-rs se-parate-d by 0.14Us; [[rj,s being li.rnitecl by Lhe rnonos tabl.e pulse wiclth. The scaler:, Figur:e 5 , 7, cons Ertlc l.ed in TTL elec tronj-cs , scales by 1to 16 rvlth ¿r sv¡j-tcheil divicle by'2 option. If the RlilfOTE st,¡j-tch is c.l,osed, t-he. sc--alel' cLivicles by 16 (ot 32) whenever ¿ signal i.s appl.-ied t<¡ the REl4OllIl ínpul, otherrvise it d-i-vide-s by l:he selected value' On the sc¿-l.e of 32 rertge, t-he scaler c:an r:esolve pu1.se pa-irs separaEecl by 0,10ps at- å m€ran i:ate of 400lstiz. The scaler is thus faster l-han the pu1-se- counting c,Lrcuit. The outptrt. ¡rulse r,/idth of the scal-er is l,21ts, rhis beíng reqrrired by the analyser. AL the count rates expelience-cl arrcl the ctwell times used, [he scaling cloes not câl-lse signi.f:icanL overflcrw .Ercnt one memory channel to the next. IË w¿rs p:loposed that the scaling r-anp¡e be chosen sr.rch that for a gíven signerl ccuilL:r:âte, the dead time l-o¡;s(ls -Ln the analy'ser rnrcluld be much srnaller than lhose iir the pulse countj-ng circuit âs rliscussed Í.n Appendi)< IL llornever, experiments seemed to indic¿rte that the from phoLomult í p1íe.r + 27A pulse nmpli-f -Íer 4'7 0 íb. 2N36t+Q comp¿tratoTs _n_ 680 0ns (7 1o) output t:rí9" 2N3 646 Si ?-20 m,onosLable (7 4r2r> upper level level + discrimj.nator sett ings Figure 5"6 PhoEomuli:Í-piier pulse det-ection circuít' 16 B R]JJLIOTE fl_ -t .2ps monostable (7 4L2L) output ns 222 2 inptrt -Jl_60 7 493 c?"e, ri counter tl !'ulse P¡ + trans;f o MR 5. Pulse scal-Í-ng circuiÈ. 2 L i'igure 7 up/clown I¡Ihen the *2 switch ís closed, ttre 't" counl-êr (741e3) circuj-t scales by a factor tw:Lce that se.Lected lry the rof-ary swilch" BB. count l.osses aïe not fully descri-bet-l by ttte statist,Lcal niodel developed and 1,he systenì dead [imes used" Conse-qu.er.l:i-]7 anirlogr.te clet.eci-íon is used for <íayg-l.orv obserr¡alíons and digital cletection j-s used for twilight :rnd ní-ghrt.i.me observatj-ons, 5 .5 .3 . Analogue Det-ectj-on, The photomu-ttiplier anode currenË is ainplifi.ed by the use of, arì operational aurplifier (Fairchild InsErunentacíon AD0-2-4) ¿ls ¡r l:r¿ìns-resistar-rce sEage with tlee gain ancl bandv¡idth being switch selectable. The low current amplifíer prorricles a 0 to I volt oLttpul proportional to the anode current for the gain sele-ctecl. The gain ís changed by changirrg Ehe portion of outprlt voltage applíed to t-he feeilback re-sistor as illrrstrated in Figure 5.8. This ainplifier was developed for: use in airglow photorneters arrd is; suíted to ttre dayglorv experime,rrt be.cause of its high stability. (Schaeff er 1970). The aurplifier: rvas always operated ç,¡iEh the 1-argest available bandwidth giving the R-C response times, ËRC, list.ed ín Table 5.2, for l:he ty1>ica1 gains used r"'ith tire type oJl trbse-r-vatj-ons i-Î-stecl . ]1.¡r trse in the dayglow experiment:, a rel ay 'Ls in,clrrde-d in the. c.írcu-í-t such thaL rvhen a sigrral is applied to the coil input, tl:e l-orvest garín is selected-, otherr.rise the gilin is as selecteci by cltc switctr. 'Jlhr,: output of the low cr-rrrent zrroplifier is converted to a pu.Lse traln. by a voltage c.ontrolled o¡;cillaior (v.c.o.) with a clynamíc rarrge cf ?-k.lP.z to 2Oktlz. Ttris pulse train js Ehen accumulared in ihe inr-r-l-L:i.c.hannel analyser. The amplifiel has ¿r small oIfset volLage thaE is gai.rr. clependent and so t-he count rz'tte for zero sl.gnal (rhj-s:i-nclucles arnp"Lifíer offset, photomultipler darlc current and v.c.o. of:Eset) is routine-Ly rneasur:ed fc¡r ttre gaj.n settings usecl durí-ng the course of an ohservat-íon. 2.4K 5"6K L6K y 6K1 2' rel.ay srvítch 3 " 3I"l t60K 60K t0M input: ADO-24 out¡rut + Figure 5.8 The lor,v current amplifier. tr'lhen posit,Lon 2 of tl-re relay switch is closed, the. lowesÈ current- gain" 333 nA/v, is selecLed. I^Ihen position 1 is closecl , the gain is as set by the robary switch. TABLE 5.2. LO[.I CURP.ENT À}IIT.I}-T,ER GAI.\S .\ND RC T].}II RESPONSfi Type of 0bservatior-r lllypical Gaí-n (nr\/v) tnc (secs) - solar spec trLlm 333 1.6 x I0 .-q r'¡híte light spercÈrLur 100 5.2x10- _t, sky spectrum 33 1.6x10- _It llg-198 source 1.0 5.2 x 10 89, 5.5,4, Ttre Combi-necl DetecLi-on .S stem" .To facililaÈe a qui-clc c.harnge of cietecLir:n syst:ens, irotl'r detectol: circuíLs are rlounLerf sicle by sirle on, the piroi;cnlu-itiplicr: irousing in aluminÍum calls r^¡hich p::ovide ncise inrmunity' 'fhe outpur- of the phoEomultiplier tube is srv:Ltched Ec-r either circuit by a relay which ir-r its unactivatecl sfate-, selects Ltte pulse countÍ-ng círcuit. A multiple pole srvitch rnounted on one of f-he- aluininium cans connects or clísconllects tlte relay coil Eo the por¡7er suppl.íes, provícles liower to the appropriate circuit and se.lecLs u'hich orrtput is taken to a connector on the eieclrcnic-s pzrnel of the high resolution F.P.I. The gaj-rr swiLch for l-he low cur:rent amplifíer is located on Ehis elecEronics pane1. Although the change of cletection systen entaj-ls ope¡|pg E1-re higli resolution l',P.I. e¡rclosure, t-his can be t.lone qui.ckly enough to prevent any thenn¿tl effec[s, As shoq¡n in Figure 5.9, the orr[put from the cletect-or system then goes t-hrough eittrer Ëhe scaler or the r¡olt-¿rge contr:olled oscillato.r ancl into l-he mtLl-tichanne-l. analyser via a rateme ter (se,cti.on 5.5 "5). 5. jj.-5. I'ionitori.rrg the SÍgnal Levels ' rtant, as far as tt'.ning f-he tr'"1?"-L. ts and assessr-nent of L,lie instrllmenl:ts perfor:ntance i.s concelrned, that a vj-sual tl-isplay of the deEecLecl count rate is ava-i-lable. The ratemeter r-tse-d (Iìower 1914) d.e-rre-Lops al analogue,s-Lg-na-1-r proportiotrai to Lhe i¡rput count ratet whÍc-lr is clispla.yed on ¿.ì. nìeter. The ftrll- scale l:¿lnp5e of Ehe- rneler is