advanced optical design and control of multi-colored ssl

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Oct 06, 2021

Advanced optical design and control of multi-colored SSL system for stage lightingapplication.

Chakrabarti, Maumita

Publication date:2016

Document VersionPublisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):Chakrabarti, M. (2016). Advanced optical design and control of multi-colored SSL system for stage lightingapplication. Technical University of Denmark.

https://orbit.dtu.dk/en/publications/b29f7c18-def0-41cb-80b5-c2f28cd7ead4

PhD Thesis Doctor of Philosophy

Advanced optical design and control of multi-colored SSL system for stage lighting application Maumita Chakrabarti

Technical University of Denmark Kongens Lyngby, Denmark 2016

Advancedopticaldesignandcontrolofmulti‐

coloredSSLsystemsforstagelightingapplications

February29

Preface: The present thesis represents theoutcomeofaPhDprojectincollaborationwiththeDanish company Brother, Brother & Sons Aps,funded by The Danish National AdvancedTechnology Foundation in 2012. The PhDworkwas carried out at the Technical University ofDenmark, DTU‐Fotonik‐department of PhotonicEngineering, from 1st January 2013 to 31stDecember 2015. The experimental work hastakenplaceintheQualitylabatRisøcampus.ThePhD project was supervised by Senior ScientistCarstenDam‐Hansen and co‐supervised by Prof.PaulMichaelPetersen.

PhDThesis

Advanced optical design and control of multi‐colored SSL systems for stage lighting applications

Page i of 173


Page ii of 173

All rights reserved Printed in Denmark by DTU Photonics, 2016 Diode Lasers and LED Systems group Department of Photonics Engineering Technical University of Denmark Frederiksborg 399 4000, Roskilde Denmark www.fotonik.dtu.dk

Dedicatedto

Myparents,ArindamAdhikariandDTULEDgroup


Page i of 173


Page ii of 173

Acknowledgements

The successful completion of this thesis has been possible only because of theprofessional and personal support from colleagues, friends, and family. Therefore Iwouldliketoexpressmyacknowledgementsforthiskindhelp.

Firstofall, Iwould like toexpressmysinceregratitude tomysupervisorsSeniorScientist Carsten Dam‐Hansen and Prof. Paul Michael Petersen for their pricelesssupervisionthroughouttheyearsandforgivingmethewonderfulopportunitytoworkwithLEDrelatedtechnologywhichwasacompletelynewfieldtomethreeyearsago.Youbothbelieveinmeandthatconfidenceprovidesmethefuelanddrivetochallengemyselftogothroughamostinterestingproject.Inparticular,IwouldliketothankyouCarstenDam‐Hansen,foralwaysbeingtherewithyourinsightfulguidance,mentoringanddiscussions.Youalwaysencouragemeandprovidethemoralsupportevenwhenthingsweredifficult.IwillalwaysbegratefultoyouforintroducingmetotheworldofLEDlightingandrelatedtechnology.IamalsogratefultoProf.PaulMichaelPetersenforyourvaluabletimeforcounselingonmyPhDworkandimportantsuggestionsrelatedto thesame.Youhavealwayshadpatience to listenmeevenwhenyouwerebusy.Thank you both for being my supervisors and mentors, which I have enjoyedthroughoutmyPhDtime.WhatIhavelearnedfrombothofyouwillbemycapitalformyfuturecareer.

IwouldliketothankPeterBehrensdorffPoulsenforhiskindhelpandmoralsupport.Being the project leader of the PhD project, you have always tried to provide therelevantinformationwhichwashelpful.Ihaveenjoyedyourhumorandourfriendship,whichmakemytimeeasytopass.Yourmoralsupporthasgivenmeconfidence.

ThePhDworkcouldnotbepossible to finish if Ididnotgetkindsupport fromaDanish lightingcompanyBrother,Brother&SonsAps. Inparticular, Iwould liketothank Thomas Brockmann, Peter Plesner, Christian Poulsen, and Esben Mallan forcollaboratingwithmefortheprojectandalltheirkindsupport.

I would like to give a warm gratitude to Prof. Emeritus Steen G. Hanson forintroducingthespeckleworldtome.Earlier,IhadalimitedknowledgeontheABCDmatrixandonspeckle.Youhelpedmetoenhancemydepthinthesamefield.Ihavealsoenjoyedyourgoodhumor.ItwaslovelyworkingwithyouandMichaelLindeJakobsen.Ienjoyedverymuchwhenwesharedoursuccesswhileeatingcake. Iwould liketothankyoubothforgivingmetheopportunitytoworkingwithyouboth.Inthisregard,IwouldliketothankDr.HaroldYurafromUniversityofCaliforniaforbeingmyexaminerduringthecourseevaluationprocess.IwashonoredwhenIcametoknowthatyouareoneofthepupilsofNobelLaureateProf.RichardFeynman.

IamtrulyfortunatetogetkindhelpfromProf.EmeritusP.S.RamanujamandMariaLouisa Rosenberg Welling for their technical and editorial checking of my thesis,respectively.Becauseofyourreviewingcomments,mywritingwascorrectedandIfeel


Page iii of 173

moreconfident.Iwouldgratefultobothofyouandthankyouboth.ThroughouttheyearsMariahasnotonlycheckedmythesis,shehasalsocheckedallmyarticles.Iwouldliketoexpressmysincerethankstoyouforyoureditorialsupport,kindhelp,andforyour friendship. I would also like to thank Linda Christel for your kind help andfriendship.YouhavehelpedmelottofulfilthePhDformalities.

Ihavecollaboratedwithdifferentgroupsanddifferentpeoplewithregardstoprojectwork packages throughout three years. I would like to thank all of you for yourcollaboration and kind help andmoral support. In particular, Iwould like to thankHenrik Chresten Pedersen, Thøger Kari Jensen, Jørgen Stubager, Dennis Dan Corell,JesperWolff, Jørgen Jepsen,AndersThorseth, StelaCanulescu, and JohannesLindén.Althoughyouaremycolleagues,Ihavealsoenjoyedyourfriendship.Inparticular,IwillmissAndersThorsethandDennisDanCorellforyourkindhelpasafriend,andourconversationsanddebatesaboutdiversetopics.Youbothbecomemyclosefriends.IhaveappreciatedtheusingofMatlabtoolbox,whichyoubothmade.

I would like to express my sincere gratitude to Anders Kragh Hansen for yourconstantmoralsupportandkindhelpasafriend.ManytimesIgottechnicalhelpinMatlabfromyouwhichIappreciated.WheneverIhadanyissue,Isharedwithyouandyouhavealwaystriedtosolveitasatruefriendwithoutanycomplains.Iamgratefultohaveyouasmyclosefriend.SoIwouldliketothankyouandwillmissyou.

Iwouldalsoliketothankmyfriends,colleagues,officemates, includingourteam‐members, and senior members. In particular, I would like to thank Rebecca BoltEttlinger,NaghmehalsadatMirbagheri,ThierrySilvioClaudeSoreze,MekbibWubishetAmdemeskel,SaraLenaJosefinEngberg,AndreaCarloCazzaniga,DominikMarti,JakobMunkgaard Andersen, Aikaterini Argyraki, Ahmed Fadil, Sune Thorsteinsson, SørenStentoftHansen,PeterJensen,ChristianPedersen,OleBjarlinJensen,HaiyanOu,JørgenSchou,andLars‐UlrikAaenAndersenforyourmoralsupport,friendship,andvaluableadvicerelatedtomyPhDstudy.Iwillmissthelunchtimetalkswithallofyou,whichIhaveenjoyedlot.IalsowouldliketothankmyflatmateEdgarNuñoMartinez,andbusmateKasperGrannAnderssonwhobelieveinmeandprovidemoralsupportandkindhelp.

Iwould like to thank JanKühn from the company JJKühnA/S andThomasFichPedersen from Force technology for helping me to fabricate and characterize themicrolensarraywhichisakeycomponentinmyPhDwork.IwouldliketothanktothecompaniesJJKühnA/S,InMold,andPolerteknikfortheirhelpstomakeTIRlens,mirrorcoating,andtherawmirror,respectively.InthiscontextIwouldliketothankHenrikChrestenPedersenagainforcollaboratingwiththeprojectfromInnovationConsortiumLICQOP,grant#2416669.IwouldliketothankHagenSchwitzerfromGermanyforhiskindhelptosimulateaRGBcolormixingsystembytheUnifiedOpticalDesignsoftware(WyrowskiVirtualLabFusion)fromLightTransUG.IwouldalsoliketothankVladimirLelekofromGermanyforhiskindsupportandforsendingasampleofdiffusermadebyAdvancedMicroopticSystemsGambhfortheopticalcharacterization.


Page iv of 173

IwouldliketothanktheSwedishEnergyAgency–LightingEducation(Sweden)andLightResearchcenter(LRC,USA)forgivingmetheopportunitytograsptheknowledgeofthestate‐of‐the‐artintheLEDlightingtechnologyfield.Inthisregard,IwouldliketothankmysupervisorCarstenDam‐HansenonceagaintosendmeinUSAforthecourse.

IwouldliketothankOttoMønstedsFondfortheirfinancialsupportduringmyvisitatthe SPIE conference andTheDanishNationalAdvancedTechnologyFoundation fortheirfundingformyPhDproject.IwouldliketothanktheITdepartmentatDTU,theAdmindepartmentatDTUFotonik,andofcoursethePhDSchoolatDTUFotonikfortheirhelpandsupport.IwouldliketothankeachandeverypersonwhodirectlyandindirectlyhelpedmetocompletemyPhDformalitiessmoothly.

I would like to thank the examiners Dr. Christophe Martinsons from CentreScientifique et Technique du Bâtiment (CSTB), France; Prof. Kjeld Pedersen fromAalborgUniversity,Denmark;andProf.KarstenRottwittfromTechnicalUniversityofDenmark,DenmarkforevaluatingthePhDwork.

Lastbutnottheleast;IwouldliketothankmyparentsandmyclosefriendArindamAdhikarifortheircontinuousmoralsupportandpatiencewithmeduringthreeyearsofPhDstudies.


Page v of 173


Page vi of 173

Summary

The thesis deals with a novel LED color mixing light engine which is designed,developed,andsubsequentlydemonstratedbymakingaprototypeofthesame,whichisexperimentallyinvestigated.Further,thedesignoptimizationsolvestheproblemsofachieving collimated high luminous flux in a color mixing system and provides asolutionwhich is capable of replacingboth the Fresnel lens spotlightHalogen lamp(2kW) and the commercially available LED luminaire (~160W), which haveapplicationsinstagelighting,theaterlighting,TVstudiolighting,etc.SincetheopticaldesigncomprisesLEDs,thelightoutputfromthelightengineisenergyandopticallyefficient as well as environmentally friendly. The light output stability during theoperationaltimeisinvestigatedbyusingtheMonteCarlosimulationandacolorsensoris implemented alongwith the pre‐calibrated lookup table to a feedback system inorder to provide controlled color and intensity variations within certain limits. Byimplementing the controlmechanism, system‐to‐system calibration is possible. ThecontrolmechanismcanbegeneralizedtobeusedinanyotherSSLsystem.Insteadofusingacolorsensor,thevariationinwavelengthforalasercolormixingsystemcanbedetectedbyanewspecklebasedwavemeter,whichiseasyforuseandcost‐effectivecompared to the commercially available expensive spectrometers. The thesis alsoreportsonthefabricationofatoolforreplicatingthemicrolensstructureusedforbeamhomogeneity,whichovercomestheexpensiveandtediousmanualpolishingordirectdiamond turning. The tool fabrication provides an easy and inexpensivemold andhenceacosteffectiveinjectionmoldingreplicationprocess.


Page vii of 173

Resumé

Afhandlingenomhandleroptiskdesign,optimeringogeksperimentelundersøgelseafetnyemultifarvedeLEDoptisksystemer.Deterresultatetafethøjteknologifondsprojektudført i samarbejde med Brother, Brother & Sons omkring lyskilder og lamper tilprofessionelbelysningafstudierogteatre.ResultateteretnytLEDoptisksystemkaldetV8 light engine som muliggør en fokuserbar lysstråle med variable korreleretfarvetemperaturfraen50mmgate,kombineretmedenmegethøjlysstrømpåmereend20.000lumen.Dettemuliggørerstatningaftraditionelletungstenhalogenlyskilderpå2kWogerenmarkantforbedringaftilsvarendeLEDbaseredelamperpåmarkedet.Dererudvikletogafprøvetennyogprisbilligmetodetilmasseproduktionafmikrolinsearray som er den for farveblandingen vigtigste komponent i systemet. Dette er etafgørenderesultatforkommercialiseringenafV8lightenginedamikrollinsearrayeterden potentielt dyreste komponent i det optiske system. Afhandlingen fokusererderudoverpådestyringsogkontrolmæssigeproblemerafsådannemultifarvedeLEDoglaserbelysningssystemersomkanvarieresilysetsspektralesammensætning.Dereropbygget enmodel for et farvekontrol feedback system baseret på en farvesensor.MonteCarlosimuleringerharvistnødvendighedenafogresultaterafimplementeringafetsådantsystemtilV8lightengine.Dererudvikletogindsendtenpatentansøgningpåetnytspecklespektrometer,dervilkunnebenyttestilfarvekontrolaflaserbaseredelyskilder. Afhandlingens resultater er publiceret i ti reviewed journal papers ogkonferenceproceedings.


Page viii of 173

Listofpublications

Articlesininternationalpeer‐reviewedjournals:

1. MaumitaChakrabarti,MichaelLindeJakobsen,andSteenG.Hanson,“Speckle‐basedspectrometer”,Opt.Lett.Vol.40,No.14,2015.*

2. M.Chakrabarti,A.Thorseth,D.C.Corell,andC.Dam‐Hansen,“Awhite‐cyan‐redLED lighting system for low correlated color temperature”, Lighting Res.Technol.2015;0:1–14;*DOI:10.1177/1477153515608416

3. M. Chakrabarti, C.Dam‐Hansen, J.Stubager, T.F.Pedersen, and H.C.Pedersen,“Replicationofopticalmicrolensarraysusingphotoresistcoatedmolds”,Opt.Express24(9),9528–9540,2016.*

4. Chakrabarti,M.,Thorseth,A.,Jepsen,J.,andDam‐Hansen,C.,“ColorcontrolfortuneablewhiteLED lightingsystem”,manuscriptsubmitted inLightingRes.Technol.on29thFebruary’2016.*

5. Chakrabarti, M., Pedersen, H.C., Petersen, P.M., Poulsen, C., BehrensdorffPoulsen,P.,andDam‐Hansen,C.“Focusable,colortunablewhiteandefficientLED light engine for stage lighting”, manuscript submitted in OpticalEngineeringon29thofFebruary2016.*

Articlesininternationalpeer‐reviewedconferences:

1. Chakrabarti, M., Thorseth, A., Jepsen, J., and Dam‐Hansen, C. “Monte CarloAnalysisofmulticolorLEDlightengine”Proceedingsof28thCIESession2015,OP60,526.*

2. MaumitaChakrabarti,AndersThorseth,JørgenJepsen,DennisD.Corell,andCarsten Dam‐Hansen, “A color management system for multi‐colored LEDlighting”,SPIEproceedings,Vol.9571‐10,2015.*

3. Steen G. Hanson, Michael Linde Jakobsen, and Maumita Chakrabarti,and,“Specklebasedwavemeter”Proc.SPIE9660,SPECKLE2015:VIInternational


Page ix of 173

Conference on Speckle Metrology, 96600U (August 24, 2015);doi:10.1117/12.2195629*

4. SteenG.Hanson,MichaelLindeJakobsen,MaumitaChakrabarti,andH.T.Yura,“Findingsmalldisplacementsofrecordedspecklepatterns:revisited”Proc.SPIE9660, SPECKLE 2015: VI International Conference on Speckle Metrology,96601J(August24,2015);doi:10.1117/12.2195631

5. Chakrabarti, M.; Thorseth, A.; Corell Dan, D. and Dam‐Hansen, C., “Opticaldesignformulti–coloredLEDlightingsystemsformuseumlightingapplication”3rdEuropean‐AsianworkshoponLight‐EmittingDiodes,D2‐02,2014.

Patentapplication:

1. MaumitaChakrabarti,MichaelLindeJakobsen,HenrikCPedersen,andSteenG.Hanson, “Speckle‐basedspectrometer”PA201500158,submitted inMarch2015

*After receiving the permission from the respective publishers and also from co‐authors,theauthorhasusedidenticalphrasinginthethesisasinthearticlesinsomeinstances.


Page x of 173

Abbreviations

BSDF:Bidirectionalscatterdistributionfunction

BRDF:Bidirectionalreflectancedistributionfunction

CCT:Correlatedcolortemperature

CIE:CommissionInternationaledel'Eclairage

CMFs:Colormatchingfunctions

CNC:Computernumericalcontrol

CRI:Colorrenderingindex

CQS:Colorqualityscale

DMX:Digitalmultiplex

ESTA:EntertainmentServicesandTechnologyAssociation

FWHM:Fullwidthhalfmaximum

IP:Internetprotocol

IS:Integratingsphere

IVL:Inversesquarelaw

JND:Justnoticeabledifference

LED:Lightemittingdiode

LID:Lightintensitydistribution

MA:Microlensarray

nm:Nanometer

PAC:Photochemicalactivepolymer

PCB:Printedcircuitboard

PMMA:PolymethylMethacrylate

PVD:Physicalvapordeposition

RGB:Red,greenandblue

SPD:Spectralpowerdistribution

TIR:Totalinternalreflection

TLCI12:TelevisionLightingConsistencyIndex


Page xi of 173


Page xii of 173

TableofContents

Preface……………………………………………………………………………………i

Acknowledgement…………………………………………………………………..ii

Summary……………………………………………………………………………….vi

Resume´…………………………………………………………………………………vii

Listofpublications…………………………………………………………………viii

Abbreviations……………………………………………………………………………x

Chapter1:Introductionandbackground…………………………………..1

1.1 Motivation………………………………………………………………………………1

1.2 Aim of the project……………………………………………………………………2

1.3 Outline of the chapters……………………………………………………………3

1.4 LED lighting technology…………………………………………………………..5

1.5 Measuring techniques for LED lighting……………………………………8

1.5.1 Photometry……………………………………………………………………………….8

1.5.2 Colorimetry .................................................................................. 13

1.6 Summary……………………………………………………………………………….21

Chapter2:Traditionallightingsystem…………………………………….23

2.1 Introduction………………………………………………………………………….23

2.1.1 Drawbacks of traditional lighting ................................................ 24

2.1.2 LED theatre lighting ..................................................................... 25

2.2 Characterization of 2 kW Halogen‐Fresnel luminaire…………….26

2.3 Characterization of LED luminaire…………………………………………28


Page xiii of 173

2.4 Summary………………………………………………………………………………29

Chapter3:OpticaldesignandmodellingofV8lightengine……..31

3.1 Introduction………………………………………………………………………….31

3.2 Description of V8 light engine……………………………………………….31

3.3 V8 light engine components and design optimization……………34

3.3.1 PCB shape and size ...................................................................... 34

3.3.2 LEDs ............................................................................................. 35

3.3.3 Reflector ...................................................................................... 36

3.3.4 Lens .............................................................................................. 38

3.3.5 Gate ............................................................................................. 43

3.3.6 Microlens arrays (MA) ................................................................. 43

3.3.7 Fresnel lens .................................................................................. 53

3.3.8 Electronic Driver .......................................................................... 53

3.3.9 Heat sink ...................................................................................... 54

3.4 Optical modelling in Zemax……………………………………………………54

3.5 Optical component parameters for V8 light engine design……57

3.6 Summary……………………………………………………………………………….59

Chapter4:Modellingofcolormixing……………………………………….61

4.1 Introduction…………………………………………………………………………..61

4.1.1 Color mixing technique ............................................................... 61

4.2 Color mixing portfolio for V8 light engine………………………………62

4.2.1 Spectral modelling ....................................................................... 63

4.2.2 Color choice and color gamut ..................................................... 66

4.2.3 Color quality in V8 light engine ................................................... 67

4.3 Discussion on the utility of the technique……………………………..69


Page xiv of 173

4.4 Summary……………………………………………………………………………….70

Chapter5:ColorconsistencyofV8lightengine……………………….71

5.1 Introduction………………………………………………………………………….71

5.2 Monte Carlo simulation…………………………………………………………71

5.2.1 LED model and pre‐calibrated lookup table ................................ 72

5.2.2 Uncertainty analysis by the Monte‐Carlo simulation .................. 76

5.3 Color and luminous flux control mechanism………………….………83

5.4 Discussion on color control……………………………………………………85

5.4.1 Why this control technique is useful? ......................................... 93

5.5 Color variation in laser‐color mixing system…………………………..94

5.5.1 Speckle based wavemeter ........................................................... 94

5.6 Summary……………………………………………………………………………..103

Chapter6:ExperimentalinvestigationofV8lightengine……….105

6.1 Introduction…………………………………………………………………………105

6.2 Forward luminous flux measurement for LED………………………105

6.3 Reflectance from the parabolic reflector……………………………..108

6.4 TIR lens loss…………………………………………………………………………108

6.4.1 TIR lens tolerance ...................................................................... 111

6.5 Color sensor and related measurement………………………………111

6.5.1 Calibration of the color sensor .................................................. 112

6.5.2 Color sensor dynamic gain settings ........................................... 116

6.6 Microlens array characterization…………………………………………117

6.6.1 Surface topology ........................................................................ 119

6.6.2 Working performance of MA .................................................... 121


Page xv of 173

6.7 Measurement of the prototype…………………………………………..123

6.7.1 Forward flux and LID measurement from the prototype .......... 123

6.7.2 Beam homogeneity ................................................................... 127

6.7.3 CRI & CQS score ......................................................................... 129

6.7.4 Warmup time ............................................................................ 130

6.7.5 Temperature sensitivity ............................................................ 131

6.7.6 Flickering test ............................................................................ 131

6.7.7 Comparison with the traditional lighting .................................. 132

6.8 Summary……………………………………………………………………………133

Chapter7:Conclusionsandoutlook…………………………………….135

7.1 Conclusions……………………………………………………………………….135

7.2 Outlook……………………………………………………………………………….138

7.2.1 Beam spot .................................................................................. 138

7.2.2 Light quality and visual perception ........................................... 138

7.2.3 V8 light engine components ...................................................... 139

7.2.4 Scope for the customisation ...................................................... 139

7.3 Application…………………………………………………………………………139

7.3.1 Business potential ..................................................................... 140

Chapter8:Bibliography………………………………………………………..143

Appendix A: ……………………………………………………………………………………………153

Appendix B: ………………………………………………………………………………………….153

Appendix C: …………………………………………………………………………………………..154

Appendix D: …………………………………………………………………………………………..154

Chapter 1 1 Introduction and background

Introductionandbackground

‘For,usuallyandfitly,thepresenceofanintroductionisheldtoimplythatthereissomethingofconsequenceandimportancetobeintroduced.’–ArthurMachen

Chapter1 :Introductionandbackground

1.1Motivation

Thetraditionallightlikehighintensityincandescent/tungstenhalogenlights[1]havebeenusedparticularlyforstudioorstagelighting.Theelectricalpowerrequirementsforsuchlightsaresubstantiallyhigh(~2kW)toilluminatethestagesufficientlywithmorethan10,000lmluminousflux.Hencethoseluminairesproduceenormousheatand infrared radiation during the operating time and thus are not environmentalfriendly.Also; such lights requiremulti‐colormechanical/optical filterdeviceswhicharesubtractivemethodofcolorfilteringtoprovideeffectivechangesinprojectedlightcolorultimatelymaketheluminaireevenlessefficient.Inaddition,therangeofcolorsinsuch light sources is limited. Looking into those mentioned problems, in presentscenario the traditional stage lights are replacedby light emittingdiodesdue to theimprovementinefficiencyandperformancereliability[2],[3].Thearraysofmulti‐colorLEDscanalsobecontrolledselectivelywithcertainintensityprovidingawiderangeoftunablecolorsandlightintensities[4]–[6].However,thereareseveralLEDluminairesavailableinthemarketwhichuseswhiteLEDwithfilterasasubtractivecolormixing.Theefficientwayofcolormixingtechniqueisadditive.SomeLEDluminairesusecolorbeams from individual color LEDs through lenses, and color mixing mechanism ispossiblebyoverlappingthedifferentcolorbeamsintoeachother.ThemixingofthreecolorLEDs(Red‐Green‐Blue),fourcolorLEDs(Red‐Green‐Blue‐AmberorRed‐Green‐Blue‐White)areafewexamplesofsimilarkindsofluminaires[7].AlthoughfourcolorLEDsareprovidingabettercolorrenderingindex(CRI)thanthethreecolorluminaire,theybothhaveshortcomingsofthecolorshadows[8].Thereareotherpossibilitiestogetunlimited colormixingwith thepackageofmulti‐colorLEDdie, integratedwithsinglelens.Howevertherewillbeatradeoffontheflexibilityofgettinghighluminousfluxfromonesinglepackage.Thus,ingeneralitisrequiredtohaveamulti‐coloredLEDluminairewithahighlumenandalsoahighCRI.Inaddition,itisalsonecessarytohaveverycompactilluminationdevicesforhandling,whichisdifficulttoachievewhenmorelight sources are integrated into the same illumination device. Due to etendue


limitations[9], [10], the task of combining light from an unlimited amount of lightsourcesintoanarrowlightbeamisdifficult.Onesolutiontothisproblemcouldbethecasewhereanopticalsystemcollectsthelightfromtheopticalgateandisadaptedtoimagetheopticalgateatatargetsurface.Recently,thelightingarchitectureonsuchasystemaredescribedbyDennisJørgensen[11]andGadegaardet.al[7].However,thisdesign cannot sufficiently produce light ofmore than 10, 000 lm. Thus the presentmarketclearlysetsthedemandofanewkindofluminairewhichwouldprovidethesolutiontowardsthestatedproblem.Thestagelightingprerequisitesfocusablewhitelight.Duetocollimationpropertyoflaserlight,itcanbeanotherpotentialcandidateasan illuminator.There isastate‐of‐the‐artavailable forwhite lightcoming fromcolormixingbylaser[12].However,thelaserhasamuchnarrowerspectrallinewidththanlightemittedfromLEDsorphosphorswhichmaynotcreatebettercomfortinvisualperceptionthanabroaderspectrum.Furthermore,duetothedemandofhighluminousflux,thecolormixingfromseveralcolorlaserscannotbeaninexpensivesolutionforthestagelighting.Toaddressthoseissues,anadvancedopticaldesignoftheadditivecolormixingtechniquebycombiningseveralcoloredLEDbeamswiththehelpofmicrolensarray (MA), positioned at the output gate of the luminaire, is proposed in thisthesis[13]–[15].

Toobtainflexibilityinopticaldesignitisalwaysrecommendedtosimulatetheopticalmodel beforehand. The optical simulation allows the design optimization and themodification easily and provides the cost effective solution. Thus initially, the colormixing technique and the proposed advance optical design are investigated bygeometricalaswellasspectralsimulationconsideringasimpletricolorLEDlightingsystem[16].TheopticalsimulationofthattricolorsystemhasbeentakenandextendedasatooloftheopticalsimulationforfivecolorLEDlightingsystem,calledasV8lightengine.

Thewhite lightproduced fromthecolormixingwouldsuffer fromcolorvariationovertime[17],[18].Thereforethereisarequirementofintelligentcontrollooptoresistthecolorvariation.Thecontrolsystemcouldworkeffectivelyifknownuncertaintiesoftheinfluencingparameterscouldbeinvestigatedandusedasaninputintoafeedbackloop[19], [20]. This leads to explore the spectrometric information from the lightoutput.Toexamine the colormixingby laserand the corresponding colorvariationovertimepersuadestodiscovertheideabehindspecklewavemeter,whichisabletodetect thewavelengthchange inthescene[21], [22].However, inordertoknowthecolor variation form the LED based light output, the use of the specklewavemetercannot be seen as an optimal solution. Implementing the color sensor inside thefeedbackloopisthesmartwayofcontrollingthecolorvariationwhichalsogivesthecosteffectivesolutionthanusingaspectrometer.Thus,thisthesisalsodocumentstheproposedworkonthisfield[23]–[25].

1.2Aimoftheproject

The scientific idea behind the PhD project “Advanced optical design and control ofmulticoloredSSLsystemsforstagelightingapplications”istodevelop,investigateand


analyzeanopticalsystemthatcollimatesandcombinesthelightbeamsfromanumberof coloredhighpowerLEDarrays into a single focusable beam. Through intelligentcontroloftheindividualcoloredLEDstheresultinglightwillbeableto:

1. coveralargecolorgamutarea

2. coverarangeofwhitelightcorrespondingtodifferentcolortemperaturesandcharacterizedbyhighcolorrenderingpropertiesandhighbeamhomogeneity.

Thiswillprovideafundamentallynewfocusable lightenginethatfullyutilizesthepotentialsofLEDcomponentsinenergyefficiencyandhighlightquality.Thistypeoflightsourcehas,however,notyetbeenrealizeddespitetherapidachievementsinLEDtechnology.TherearetwomajorproblemswithLEDlightsourcesusingadditivecolormixing today[26], [27]. They suffer from either poor color rendering or generatecoloredshadowsbehindtheilluminatedobjects.Inordertoovercometheseproblemswithoutsacrificingefficiency,newinnovativeideasontheopticalsystemsarerequired.Furthermore, a better understanding of color rendering, control, uniformity andstabilizationisneeded.Iftheseproblemsareovercome;theresultswouldbeofmajorinterestintheLEDandlightingR&Dcommunityaswellasinthelightingindustry.ThesolutionalsowillbecapableofreplacingtheHalogen‐Fresnelspotlamp(2kW)aswellasthecommerciallyavailableLEDluminaire(~160W)whichhasapplicationsinstagelighting,theaterlighting,TVstudiolighting,etc.

ThePhDprojectispartoftheR&Dproject“LIGHTENGINEV8–Agreenrevolutionfor colored light”, which has been granted by The Danish National AdvancedTechnologyFoundation in2012.DTUFotonikisrunningtheproject incollaborationwith the Danish company Brother, Brother & Sons Aps, who would take the LEDtechnologyintothefutureofprofessional lightingintheentertainmentindustry.Thepossibleapplicationareaswouldbefarmorewidespreadandarchitecturalaswellasdisplaylightingisgoingtobeinvestigatedfurtherforcosteffectiveness.Thescientificcontributionsmade in relation to the study period of this project are patented andpublishedinpeer‐reviewedscientificarticles.

1.3Outlineofthechapters

ThestudyandtherelatedworkofthePhDprojecthasbeendocumentedinthisthesisdividedinsevenchapters.Theshortdescriptionaboutthechaptersandtheirrelationinconnectiontoscientificarticlesareoutlinedbelow:

Chapter1:describesthemotivationofthestudyonthetopicofthePhDprojectandtheachievablegoalsbehindtheideaofresearchingonadvancedopticaldesignformulti‐coloredLEDsforstagelightingapplication.Italsoelaboratesthestate‐of‐the‐artonLEDlightingtechnologyandcorrespondingmeasuringtechniquesofLEDluminaires.

0:providesthestate‐of‐the‐artonthetraditionalstagelighting.Theshortcomingsandadvantagesof traditionalstage lightingcomparetoLEDstage lightingare listed.Theexperimental results from tungsten‐halogen luminaire and LED luminaire (market


available) show the capability of present stage lighting system. The chapter alsocommentsontheforecastontheupcomingstagelighting.

Chapter3:introduceswithV8lightenginecoveringthearchitecturaldesignaspectofthe same. It also defines the required optical components used to demonstrate theprototype of the V8 light engine. The geometrical simulation of the optical designoptimizes the individual optical componentswithin the light enginewith respect tobeamhomogeneity,opticallossandluminouslightoutput.Thereisapeer‐reviewedscientific articleon thisnewadvancedopticaldesignof amulti‐coloredLEDsystemwhich provides the tunable, focusable and collimated while light with dimmablecapability.ThechapterdesignatesthedesignaspectoftheMA,whichhasbeenusedasasecondaryopticsinsidetheV8lightingdesign.Italsoelaboratesonthenewfabricationtool used to produce the MA and the characterization technique to investigate theworkingperformanceofthesame.

Chapter4:presentsthestate‐of‐the‐artonadditivecolormixingtechniquesandtheadvantage of using the same concept on subtractive filtering method. The spectralsimulationofcolormixing technique isverifiedby takingasimpleexampleof threecoloredLEDsystem.TheV8systemwhichismorecomplicatedhavingfivecolorLEDs,however,itessentiallyusesthesamebuildingblockswhilesimulatingthedesign.Thepeer‐reviewed article is considered to publish the corresponding scientificcontributions. The three colored optical system is demonstrated subsequently bymakingaprototypewhichisusedinthedisplaycasesoftheexhibitionatRosenborgCastleinCopenhagen,Denmark.

Chapter5:elaboratesonthecolorconsistencyaspectoftheLEDdeviceandtherelatedresearch on spectral color control of the resulting beamwhich optimizes the colorgamut, color rendering and optical efficiency of the light engine. Initially, theuncertainties in the multi‐colored LED device are analyzed by the Monte Carlosimulation and the new algorithm to reduce the uncertainties inside the system isproposedbyimplementingacolorsensorandafeedbackcontrolloop,respectively.Thespectralmodelcanbevalidatedbyproposedexperiments.Thischapteralsodescribesthepublishedpeer‐reviewedscientificarticlesrelatedtotheabovementionedstudy.For laser based colormixed luminaire, the spectrometric information is required todetect the color variation. This chapter therefore discloses the new idea behind aspecklebasedwavemeterandtherelatedpeerreviewedarticlewhichcanbeusedtoanalyzethewavelengthchangeofthelasersbeingusedduringtheoperation.

Chapter6:establishestheexperimentalfacilityandthecharacterizationoftheV8lightengine.The experimental resultshavebeenvalidatedby theoptical simulation.TheresultsalsoindicatethemodificationoftheshapeandsizeoftheopticalcomponentsusedintheV8designtoachievethetargetluminousfluxfromthelightengine.Thereisan extensive investigation of the optical loss encounters from the different opticalcomponentsinsidethesystemwhichguidesthetotallosscalculationfromtheopticalsystem.FinallythereisacomparisonoflightoutputbetweentheHalogen‐FresnelspotluminaireandtheV8lightengine.


Chapter7:concludesthe3yearsstudyaswellastheworkforthePhDproject.Italsoindicatesthefuturerecommendedworkplanasoutlookonthesamefieldwhichdidnotfinishwithintheshortperiodofthreeyears’time.Thechapteralsogivesamarketsurveyonthepresentscenarioofstagelightingandtheirprosandconsinconnectiontomarketdemands.There isa thoroughdiscussionon theadvantageofusingV8asastagelightingsystemoveravailablelightingsystems,likeArri,Desistietc.Thechapteralso directs the application areas where the V8 light engine could be used as anilluminatingdevice.

Chapter8: offers the listingofpublicationdetailswhichhave followedas referenceduringthestudyperiod.

ThenextsectionexplainsaboutthebasicphysicsbehindtheLEDtechnologyanditsadvantagesoverthetraditionallighting.

1.4LEDlightingtechnology

TheilluminationfromLEDisefficient,extremelylonglasting,andenormouslyversatile.Asthefuturelightingtechnology,LEDsoffermaximumflexibilitytolightingdesigners,architect,electricians,andprivateconsumers.Thetechnologyprovidescostsavingsinpowerconsumptionandmaintenance.Lightmanagement systems(LMS)withLEDsprovidetheperfectcontrolforplanningvaluableLEDlightingsolutions.ThefollowingaretheadvantagesofLEDtechnologyovertraditionallightingtechnology:

Lowpowerconsumption

Highefficiencylevel

Longlifetime

Continuousdimmingcapability

Smallestpossibledimensionsandcompactmodule

Highresistancetoswitchingcycles

Immediatelightwhenswitchingon

Widerangeinoperatingtemperature

Highimpactandvibrationresistance

Reliabilityondifferentenvironmentalconditions

AbsenceofUVorIRradiation

Highcolorsaturationlevelwithoutfiltering

NomercuryisusedinthemanufacturingofLEDs.Thusitismercuryfree

As a result the LED technology is exploited in many applications in differentindustriessuchasindoorlighting(home,office,hospitals,retails,etc.),outdoorlighting(sports,stage,theater,studio,street,etc.),backlighting,automobilelighting(headlight,


indicator, sensor, etc.). For those applications, the efficiency of the LED lighting isattractive.Inthiscontexttheluminousefficacyisameasureofhowwellalightsourceproducesvisiblelightforagivenelectricpower.Thus,theluminousefficacyistheratioof luminous flux toelectricalpower (lm/W).The followingFigure1‐1provides theguidelinesonefficiencyofanyrangeofwhitelightfromdifferentkindoflightsources.Thefractionofelectromagneticpowerwhichisusefulforlightingcanbeameasureforluminousefficacyofradiation.Theluminousefficacyofradiationcanbeobtainedbydividingtheluminousfluxbytheradiantflux.Since,thelightwithwavelengthsoutsidethevisiblespectrumcontributestotheradiantfluxwhiletheluminousfluxofsuchlightiszero,thosewavelengthsreducesluminousefficacy.Therefore,wavelengthswhichareneartothepeakoftheeye'sresponsecontributemorestronglythanthoseneartotheedges. For the caseofmonochromatic light at awavelengthof555nm (green), thePhotopic(explained in1.5.1) luminousefficacyofradiationhasamaximumpossiblevalue of 683 lm/W corresponds to an efficiency of 100%, whereas the Scotopic(explainedin1.5.1)luminousefficacyofradiationreachesamaximumof1700lm/Wfornarrowbandlightofwavelength507nm.Sothewhitelightwhicharethemixtureofseveral wavelengths, never can achieve the luminous efficacy of 683 lm /W. TheprogressofwhiteLEDsismuchmoreprominentnowadaysandwhiteLEDs'luminousefficacyhasincreasedfrom5lm/Wtotoday'scommercialwhiteLEDof150lm/Wwhichistillnowthehighestluminousefficacyofallwhitelightsources.However,asexplainedabove,thetheoretical limit forwhiteLEDs isabout260‐300lm/W.Sotheluminous flux efficacy of LEDs continues to reach higher values than other lightsources[28]–[31]inlaboratory.Theavailablevalueof265lm/Wisalreadyachievedinthe laboratory[29]. However, the shortcoming of this specific design is that theluminousfluxisabout14.4lm.Bymodifyingthedesign,theresearchersconfirmedtwootherwhiteLEDswithvaluesof203lmand183lm/Wat350mA,andwithvaluesof1913lmand135lm/Wat1A,respectively.Figure1‐1showsthetheoreticalborderofavailablecommercialwhiteLEDatthepresentscenario.TheimprovementofefficacyinluminouslightoutputbyLEDleadstoannualenergysavingsofUSD1.8billionintheUSmarketaccordingtothereportin2013fromUSdepartmentofenergy[32].FortheUSnation,thepotentialpayoffforthesolidstatelightingisenormousanditisestimatedthat converting entirely into LED lightingwould save the countryUSD39billion inannualenergycost,reducingtheenergyconsumptionfromthelightingbymorethanhalfandavoid280MTofcarbonemissioneachyear.Thus,potentialdemandsofusingLEDasaprimaryilluminatingsourceisincreasingdaybyday.


Figure 1‐1: The present scenario of optical luminous efficacy for white light from the different kind light sources (Source Ref.[33])

ThebasicprincipleofanLEDconsistsofmultiplelayersofsemiconductingmaterial.Lightisproducedintheactivelayerwhilethediodeisbeingusedwithdirectcurrent.The produced light is decoupled directly or emits by reflections. In contrast tocontinuousspectrumemittingfromincandescentreflectorlamps,anLEDemitslightofadistinctwavelength.Thelight'swavelengthorcolordependsonthecompositionofthecrystalcompoundsofthesemiconductormaterialused.InordertoproduceLEDswithahighdegreeofbrightnessinallcolorsfrombluetored,twomaterialsystemssuch as phosphide or arsenide are mainly used. These could create variouscombinations, each of which releases varying amounts of energy according to thematerial's bandgap.When charge carriers (holes and electrons) are recombined byapplyingspecificvoltageinanLEDprintcircuitboard,photonsareemittedintermsoflightaccordingtospecificdiscreteenergylevels.ThisspecifiestheparticularlightcolorforLEDs.Forexample,bluelightisproducedifahighlevelofenergyisreleasedandalowerlevelofenergyisresponsibleforredlightemission.EachLEDlightcolorislimitedto a very narrow range of wavelength called the dominant wavelength of thecorrespondingcolor.SinceLEDproducesonlymonochromatic(singlecolor)light,theonlyspectrumthatcannotbeproduceddirectlyfromtheLEDprintcircuitboardisthewhite light spectrum,which represents amixture of all light colors. There are twomethodsforproducingwhiteLEDlight:photoluminescenceandadditivecolormixing.ThusthewhitecolorisobtainedeitherfromluminescenceconversionofbluecolorLEDorbymixingred,greenandblueLEDs(RGB). Thestate‐of‐the‐artforadditivecolormixingisdiscussedindetailinChapter4.


There are few factors which influence the reliability and operating life of theindividual LED and also on the entire LEDmodule.When light is produced due toincreaseofheatintermsofincreasingjunctiontemperature,boththelifecycleandtheLED'sluminousfluxaffects.Thatiswhyitisessentialtomanagetheheatbythebestpossibleinstallationmethodorbysuitableheatsinks.ThecoolerLEDhasthelongerlifecycle, and has a more efficient and brighter light output. The LED operatingtemperaturealsocaninfluencetheworkingperformanceofLEDinsameway.EveryLEDcanbeoperatedwithinaspecificcurrentrange.Thelowerthecurrentiswithinthisrange,thelessenergyisreleasedandthelowertheheatproductionwouldbe;whichhas a direct effect on the operating life. The following section would provide themeasurementtechniqueoflightoutputfromLEDs.

1.5MeasuringtechniquesforLEDlighting

Thetermcalledlightingonlyexistsduetothevisualperceptionofthehumaneye.Theradiantenergythatisbasicallyanelectromagneticwave,capableofexcitingtheretinaand the constructionof avisual sensation inside the retina is called light.Thereareseveraldistinctregionsofthiselectromagneticspectrumnamely;ultraviolet,visibleandinfrared,etc.,whicharedesignatedby the rangeofwavelengths.Thevisibleportionextendsfromabout380nmto780nm.Inthis,wehavedealtwithvisiblelightemittedfrom LEDs in the V8 system. To characterize the light, we need to know the basicknowledge behind the photometry and colorimetry as well as the measurementtechniquewhichhavediscussedinthissection.

1.5.1Photometry

Photometry is the science concerned with measuring visible light in terms of itsperceivedbrightnessbythehumanvisualresponse.CIEdefinesthespectralresponseoftheaveragehumaneyebymeasuringtheadaptedlightbyeyesandcompilingthedataintothephotopiccurve.Thecurveindicatesthatpeoplerespondstronglyforthecolorgreenandarelesssensitivetothespectralextremeregionsredandviolet.Theeyehas a different response characteristic in the dark‐adapted state and as such it hasdifficultyindefiningcolor.Thisleadstoasecondsetofmeasurementsandthecompileddataiscalledascotopiccurve.Figure1‐2showstheresponsecurveforphotopicandscotopic, respectively.Both thephotopic curves (CIE1931 [34]andCIE1978)havepeaksat555nm,whereasthepeaksareshiftedto507nmforscotopiccurve.InthisthesisallcalculationsarebasedonthephotopicresponsecurvebyCIE1931andCIE1978,respectively.


Figure 1‐2: The CIE 1931 V(λ), the CIE 1951 V(λ) and the CIE 1978 V(λ) functions are shown (Source Ref.[35])

Todefinethevariousphotometricquantitiesitisrequiredtoknowtheterminologyoflightingterms,symbolsandunits(Table1‐1).

Table 1‐1: Photometric quantities with the units (Source Ref.[36])

Quantity Symbol Metricunit Englishunit

Luminousintensity I candela(cd) candela(cd)Luminousflux Φ lumen(lm) lumen(lm)Illuminance E lux(lm/m2) fc(lm/ft2)

Luminousexitance M lm/m2 lm/ft2Luminance L cd/m2 cd/ft2

Luminous intensity (I): is assumed as the fundamental quantity, defined by theunitcalledcandela(cd).Theunitofluminousintensitywasestablishedin1860byusingacertaintypeofcandleofspecificweightburningataspecificrate.Thefirstincandescentlamphadaluminousintensityequalto16candles,andthecorrespondingvoltagewas110volts(AC).Itisalsodefinedasthequantityofluminousfluxemitteduniformlyintoasolidangle.Thus,1cdisequivalenttoonelumenpersteradian.

Luminousflux(Φ):isanotherquantity,definedbytheunitcalledlumen(lm).Itistherateatwhichluminousenergyisincidentona1‐m2surfacefromadistance1mfromauniformpointsourceof1cdintensity,asshowninFigure1‐3.Itisameasureofthetotalopticaloutputofavisiblelightsource.


Figure 1‐3: 4 π lumens emits from 1 cd uniform point source (Source Ref.[36])

Illuminance(E):isthe‘ ’densityincidentonasurface.InEnglishunits,onelumenoffluxfallingononesquarefootiscalledafootcandle.Themetricequivalent,onelumenper squaremeter, is termed a lux (10.76 lux = 1 footcandle).When light incidenceoccursonanobject, ithasreflection, transmissionandabsorptiondependingonthemediumofincidence.Italsofollowscertainrules.Thefollowingarethecommonrulesconsideringthemeasureoftheilluminance.

Inverse square law (ISL): for point source is when the illuminance on a surface isinversely proportional to the square of the distance from the source. The largestdimensionofthelightsourceismuchsmallerthanthedistancewheretheilluminanceismeasured.Inthatcase,theilluminanceshouldmeasurethedistanceoffivetimesthesourcesize,calledthefivetimesrule.However,forthestrongdirectionalsourcesuchasalasercannotfollowthefivetimesrule.Inthatcasetheminimumdistancecanbetentimestothesourcesize.

Lambert’scosinelaw:isthattheilluminanceonanysurfacedependsonthecosineofthelight’sangleofincidence.Thefollowingequation(1)describestheformula:

(1),

whereincidentangle‘ ’ismeasuredwithrespecttothesurfacenormal.

LuminousExitance(M):is the density of ‘ ’ leaving the surface in all direction known as exitance. The unit is the same as illuminance which is lm / m2. The luminous exitance measurement is most appropriate in emitters with a flat surface.

Luminance(L):isalsoreferredtoasthephotometricbrightness,itisthedensityof‘ ’leavingthesurfaceinaparticulardirection.Thetechniquetakesintoaccounttheareaofthesurfacemeasured,andtheanglesubtendedbyanobserverlookingatit.Thus,thefollowingequation(2)definestheformula:

∝

where ‘A’ is the projected area and ‘α’ is the subtended angle. For perfectly diffuse(Lambertian)surfaceboth‘M’and‘L’isrelatedtotheequation(3).

(3)


1.5.1.1Luminousfluxmeasurement:

Tomeasureaccuratelythetotalradiantflux,the‘ ’andthespectralpowerdistribution,integratingsphere(IS)photometeriscommonlyused.Bythismeasurement,thelamplumen,luminousefficacy,CRI,andcolorappearanceparameters(e.g.CIEx,y,andCCT)can be determined[37]–[39]. An IS is a hollow sphere with its interior uniformreflectancecoatingbybariumsulfatetocreatenearlyperfectdiffusereflectivityasthelambertiansurface.Thereflectancedependsonthethicknessofthecoating.Thecoatingis thermally stable at approximately 100° C, and generally the commonly availablereflectanceisfrom80%to95%.Figure1‐4describesthebasicprincipleoftheIStheory.The light reflectsmultiple timesbeforeentering thedetector indirectlyas thedirectlightisblockedbythebaffle.Themeasuringtotalof‘ ’isfoundbyusinganilluminancemeter, where the measurement by the photometer is compared to the similarmeasurementobtainedfromacalibratedlampofknownoutput,commonlyreferredtoasthe“comparisonmethod”.Theresultsofthiscomparisondeterminethelightoutputofthetestlamp.ThefollowingEquation(4)‐(6)[36]showstheformulaetomeasurethefluxfromtheIS:

Figure 1‐4: Total luminous flux measurement using an integrating sphere

1 ⋯ .1

(4)

4 1

(5)

1

(6),

where‘ ’isthereflectanceoftheinteriorsurface,‘ ’istheluminousfluxofthelamp,‘ ’istheradiusofthesphere,‘ ’istheilluminanceontheinterior,‘ ’istheconstant


determined using a standard reference lamp. A spectrometer at the detector portenables fluxandcolormeasurements.Themeasuring total ‘ ’usinga spectrometermeasures the radiant power as a function of wavelength. The luminous fluxinterpretations for the sensitivity of the eye by weighing the radiant flux at eachwavelengthwith the human eye response function. Equation (7) is the formula formeasuringthetotal‘ ’fromthespectrometer:

683

(7),

where ‘ ’ is the wavelength, ‘ ’ is the SPD, ‘ ’ is a function representing thewavelength‐dependentsensitivityoftheeye.Theluminousefficacyofalightsourceismeasured by the total ‘ ’ emitted by the lamp divided by the total lamp power(electrical)input.Itisexpressedinlumensperelectricalwatt.Inthisthesis,theISof1mand 2m diameters respectively equipped with an array spectroradiometer (CAS‐140CT)areusedaccordingtotheCIELEDteststandard[40]byusingthequalitylab(Q–lab)facilityoftheLightQualityMeasurementLaboratory,atDTUFotonik.Thelightwascoupledthroughthesideport(15cmdiameter)inaforwardfluxconfigurationandaself‐absorptioncorrectionwasapplied.

1.5.1.2Intensitydistributionmeasurement:

The luminous intensity distribution of the luminaires can be measured by agoniophotometer or a goniospectrometer. The goniophotometer and coordinatesystemsarecategorized intothreetypeswhichareTypeA,TypeBandTypeC.TheangulardistributionofthespectralandluminousintensityoftheluminaireismeasuredinthisthesisbyaTypeCnearfieldgoniophoto/spectrometer(RiGO‐801)[41],whichisequippedwithanarrayspectrometer(specbos1211),acolorcamera(LMK98‐4color),and a photometer(PH‐ST‐C8‐TH‐L). Figure 1‐5 shows the goniometric coordinatessystemfortheTypeC.ThisequipmentisalsosituatedintheQualitylab(Q–lab)facilityoftheLightQualityMeasurementLaboratory,DTUFotonik.Themeasurementdistanceis1.4m.Thus, thegoniometricmeasurementcanbeusedas far fieldmeasurementwherethemeasurementdistance1.4miswithinthefivetimesofsourcesize[40],[42].ThecameracapturestheimagesofluminancedistributionsoftheluminaireatdifferentanglesusingVλfilters.BychangingthefilterstoX1,X2,andZ,themeasurementscananalyzethevaluesofX,Y,andZ(tristimulusvalues)ofthelightatdifferentangles.TheCCTandDuvcanbecalculatedafterwardsatdifferentangles fromthose tristimulusvalues.TheintegratedvaluesofmeasurementswithVλfilterwouldgivethetotal‘ ’ofthe luminaireasVλcanbetakenas ‘ ’of the light.ThespectrometricmeasurementwouldalsoprovidetheinformationofCCTdistributionofthelightatdifferentangles,ifthe luminaire isof small size compared to themeasurementdistance (followed fivetimes rule), where the goniospectrometer measurement acts as a far fieldmeasurement. However, for the larger size of luminaire the spectrometricmeasurement would introduce an error. The photometer measures the ‘E’ of lightdistributionforalldirectionalandintegratedvalueofalldistributionsprovidethetotal


luminous flux ‘ ’ of the luminaire. All measurements are calibrated with a totalluminousfluxstandardlampfromaNationalMetrologicalInstitute(NMI).

Figure 1‐5: Goniometer Coordinate System (Ref. [41])

1.5.2Colorimetry

Theassessmentandquantificationofcolorisreferredtoascolorimetryorthe“scienceof color” [35]. To characterize the light qualityof the light output, it is necessary tomeasurethecolorparameters likeCCT,Duv,CRI,etc.Thefollowingwoulddefinetheparameterswhichareusedinthisthesis.

1.5.2.1Colormatchingfunctions(x(λ),y(λ),andz(λ)):TheCIEhasstandardizedthemeasurement of color by means of color‐matching functions and the chromaticitydiagram(CIE,1931)[34]asthesensationofthecolorisasubjectivequantity.Thethreecolor‐matchingfunctionsareobtainedfromaseriesofadjustingtherelativeintensitiesofthered,green,andbluelight, inwhichthesubjectsetstheintensitiesofthethreeprimary lightsrequiredtomatchaseriesofmonochromatic lightsacross thevisiblespectrum.TheCIE1931color‐matchingfunctions (λ), (λ),and (λ)areshowninFigure1‐6.Thegreencolor‐matchingfunction, (λ),ischosentobeidenticaltotheeyesensitivityfunction,V(λ),i.e.:

(8)


Figure 1‐6: CIE (1931) 2° color matching functions (CMFs)

Thethreecolor‐matchingfunctionsportraythefactthathumancolorvisionpossessestrichromacy,andthecolorofanylightsourcecanbedescribedbyjustthreevariables,whicharedimensionlessquantities.Therehavebeenseveraldifferentversionsofthecolor‐matching functions and of the chromaticity diagram as any of them are notunique.

1.5.2.2Tristimulusvalues(X,Y,Z):givethestimulation(i.e.power)ofeachofthethreeprimaryred,green,andbluelightsneededtomatchthecolorofP(λ)whichispowerspectral density or power spectrum. The relation between P(λ) and the tristimulusvaluesareasfollows:

x λ P λ dλ

(9)

y λ P λ dλ

(10)

z λ P λ dλ

(11)

According to the above definition, tristimulus values should have the unit ofwatt.However, the tristimulus values are usually considered as dimensionless quantitieswheretherewillbeapre‐factorofwatt‐1beforeintegral.

1.5.2.3Chromaticitycoordinates (x,y,andz): are the stimulationof eachprimarylightdividedbytheentirestimulationwhichareasfollows:

(12)


(13)

1

(14)

Asseen inequation(14),zchromaticitycanbefoundfromxandy, thoughit isnotrequiredtouse.Thus,thechromaticitydiagramismadebythexandycoordinatesinFigure1‐7.Inthisdiagram,thereddish,greenishandbluishhuesarefoundforlargevaluesofx,yandz.Figure1‐7alsoillustratesthelocationofthePlanckianlocus,whichis theblack line inthediagram.Allcolorscanbecharacterizedwithrespecttotheirlocation in the chromaticity diagram. The figure also shows the equal‐energy pointfoundinthecenterofthechromaticitydiagramwhichhas(x,y,z)=(1/3,1/3,1/3).Theopticalspectrumcorrespondingtotheequal‐energypointhasaconstantopticalenergyperwavelengthintervaldλacrossthevisiblespectrum.Theoutcomeofsuchspectrumresultsinequaltristimulusvalues,i.e.X=Y=Z.Thecolordifferencesofcloselyspacedpointsinthechromaticitydiagrammusthaveaminimumgeometricaldistancetoyieldaperceptibledifferenceincolor.In1943MacAdam[43]showedthatthecolorswithinacertainsmall region in thechromaticitydiagramappear identical tohumansubjectsandthoseregionshavetheshapeofellipses.SuchellipsesarecalledastheMacAdamellipses.Italsoindicatesthatthegeometricdistancebetweentwopointsinthe(x,y)chromaticitydiagramdoesnotscalelinearlywiththecolordifference.

Figure 1‐7: CIE 1931 (x,y) chromaticity diagram where monochromatic colors are found at the perimeter and white light is situated in the center of the diagram

1.5.2.4Uniformchromaticitycoordinates(u,v,u ,́v´): If thecolordifference inthechromaticitydiagramisproportionaltothegeometricdifference,itisconvenientanddesirable. In 1960, the CIE introduced the (u, v) and in 1976 the (u′, v′) uniformchromaticitycoordinatestoformtheuniformchromaticitydiagram[44][45].Fromthetristimulusvalues,theuniformchromaticitycoordinatesarecalculatedaccordingtothe


belowequations(15)‐(16).TheCIE1960andCIE1976(u′,v′)uniformchromaticitydiagramsareshown inFigure1‐8(a‐b).The(u,v)and(u′,v′)uniformchromaticitycoordinatescanalsobecalculated fromthe (x,y) chromaticity coordinatesandviceversaaccordingtotheequations(17)‐(20).

Figure 1‐8: (a) CIE 1960 (u, v) and (b) CIE 1976 (u´, v´) uniform chromaticity diagrams calculated using the CIE 1931 2° standard observer respectively

415 3

,615 3

(15)

´

415 3

, ´915 3

(16)

´

42 12 3

(17)

62 12 3

, ´9

2 12 3

(18)

9 ´6 ´ 16 ´ 12

,2 ´

3 ´ 8 ´ 6

(19)

32 8 4

,2 ´

2 8 4

(20)

Thecolorchangesmuchmorerapidlyinonedirection,e.g.thex‐direction,comparedtotheotherdirection,e.g.they‐directionin(x,y)chromaticitydiagram(Figure1‐7)asthecolor differences between two points in this diagram are spatially non‐uniform.Althoughthedeficiencyofthe(x,y)chromaticitydiagramisnoteliminatedfullyinthe(u, v) and (u′, v′) uniform chromaticity diagrams, the color difference between twolocationsintheuniformchromaticitydiagramis(approximately)directlyproportionaltothegeometricaldistancebetweenthesepoints.

1.5.2.5Alternative to theMacAdamEllipses: The CIE recommends that the (u',v')chromaticitydiagramdenotesthemostuniformcolorspaceforlightsources.InFigure1‐9, the five‐step MacAdam ellipses (solid black line) are plotted in the CIE (u',v')


diagram,ontopofwhichcircleswithradiusof0,0055(dashedredline)areplotted.Asshowninthefigure,theseellipsesandthecirclesarecloselyoverlapping,whichindicatethatinthewhiteregionnearthePlanckianlocusinthe(u',v')diagram,circlescanbeusedasanalternativetotheMacAdamellipses.Thus,itwasalsofoundthatthefive‐stepMacAdamellipsesareapproximately equivalent to the circleswith radius0.0055 in(u',v').Theu'v'circleisspecifiedwithacenterpoint(uc ,́vc´)andradius‘r’onthe(u',v')diagram,andexpressedbytheequation(21)[46].ForconsistencywithintheMacAdamellipsesinthe(u ,́v´)chromaticitydiagram,theterm”n‐stepu'v'circle”isdefinedbyCIE[46]asacircleinthe(u',v')diagramwitharadiusofntimes0.0011.Foracenterpoint(uc ,́vc´),then‐stepu',v'circleiscalculatedbytheequation(22)where‘n’correspondstoan‐stepMacAdamellipse.

Figure 1‐9: Five‐step MacAdam ellipses in IEC 60081 (IEC, 1997) and circles (radius 0. 0055) in the CIE 1976 (u',v') chromaticity diagram (Source Ref.[46] )

´ ´ ´ ´ (21) ´ ´ ´ ´ 0.0011. (22)

A just noticeable difference (JND) is recommended by CIE at 50% probability isconsideredtobe(0.0011×1.18=)0.0013in(u',v')coordinates.Inthewhiteregionofthe(u',v')diagram,chromaticitydifferencesbetweenthetwopoints(u´1,v´1)and(u´2,v´2)on the (u ,́ v´)diagramwhich is expressedby thedistance in (u',v') coordinatescalculatedbyΔu´v ́isshownintheequation(23).

∆ ´ ´ ´ ´ ´ ´ (23)

1.5.2.6Chromaticitydistance:istheclosestdistancefromthePlanckianlocusonthe(u',2/3v')diagram,with‘+’signforaboveand‘‐‘signforthebelowthePlanckianlocusanddenotedbyDuv[47].TheFigure1‐10indicatestheDuvsignbyarrow.


Figure 1‐10: magnified CIE 1976 u´‐ v´ diagram (Source Ref. [48])

1.5.2.7Colortemperatureandcorrelatedcolortemperature:Thecolortemperature(CT)ofawhitelightsourceisthetemperatureofaPlanckianblack‐bodyradiatorwhichhasthesamechromaticitylocationasthewhitelightsourceconsideredandtheunitofCTiskelvin.Withincreasingtemperatures,itglowsinthered,orange,yellowishwhiteandfinallyabluishwhiteandaccordingtothatcolorvariationitisdistinguishedbyfewregionscalledwarmwhite,neutralwhiteandcoolwhiterespectively.IfthecolorofawhitelightsourcedoesnotfallonthePlanckianlocus,thecorrelatedcolortemperature(CCT)isusedwhichisdefinedasthetemperatureofaPlanckianblack‐bodyradiator,whosecolorisclosesttothecolorofthewhitelightsource.TheCCTisalsogiveninunitsofkelvin.

1.5.2.8Colorgamut:Thecolorgamutrepresentstheentirerangeofcolorsareathatcan be created from a set of primary sources. Color gamut is a polygon positionedwithintheperimeterofthechromaticitydiagram.Dependingonthenumberofprimarycolor,theshapeofthegamutchangeasforthecaseofthreeprimarycolors,thecolorgamutisatriangle.Theproducedwhitelightfortheadditivecolormixing,allmixingcolorscreatedbythevertexpoints(primarycolors)ofagamut, is locatedinsidethesame gamut (dash line in Figure 1‐11). Thewide color gamut is desirable to haveflexibilityofcreatingwhitelightinanadditivecolormixingprocedure.


Figure 1‐11: The area surrounded by dash line is the color gamut for four primary colors

1.5.2.9Colorrenderingindex(CRI):Theabilitytorenderthetruecolorsofaphysicalobject(e.g.,fruits,plants,humanskin,flowers,wood,furniture,clothes,etc.)whicharebeingilluminatedbythewhitelightsourceismeasuredintermsofthecolor‐renderingindex(CRI)[49].TheevaluationofCRIismadebycomparingthecolorrenderingabilityoftestlightsourcewiththecolorrenderingabilityofreferencesource.ThereferencesourcerecommendedbyCIE[49]followscertainruleswhichare:

(i) Planckian black‐body radiatorwouldbe the reference sourcewith thesamecolor temperatureas the test sourcewouldbe considered, if thechromaticitypointofthetestsourceislocatedonthePlanckianlocus

(ii) ThereferencesourcewouldbeaPlanckianblack‐bodyradiatorwiththesamecorrelatedcolortemperatureasthetestsourceifthechromaticitypointofthetestsourceislocatedoffthePlanckianlocus

(iii) For other instances, one of the standardizedCIE illuminantswouldbeusedasareferencesource

(iv) Itisalsoconsideredthatthetestsourceandthereferencesourceshouldhavethesamechromaticitycoordinatesandluminousfluxoutput

Conventionally,theCRIofthePlanckianblack‐bodyreferencesourceisanticipatedtohaveCRI=100andthetestsourceneverhasCRI>100.ThesensitivitydependencyofCRIindicatesthattheselectionofreferencesourceisalwayscriticalwhilecalculatingCRI of test sources. As the test‐color samples are defined in terms of their spectralreflectivity,thetest‐colorsamplesplayanimportantroleforcalculatingtheCRIofatestsource inaddition to the reference sourceand the test source.Thus, in intensionofhaving international standardization, a set of fourteen test color samples has beenrecommendedforthepurposeofdeterminingtheCRIofthetestsource.AccordingtotheCIEstandard,generalCRI,whichisanaverage,iscalculatedasfollows[35]:

∑ , (24)


where iscalledasspecialCRIsforasetofeighttest‐colorsamples.ThespecialCRIsarecalculatedaccordingtothefollowingformula[35]:

100 4.6∆ ∗, (25)

where ∆ ∗represents the quantitative color change while a test‐color sample isilluminatedwith, firstly, thereferencesource,andlaterwiththetestsource[50].ThegeneralCRI(Ra)iscalculatedfromtheeighttestcolorsamples(i=1–8).Inadditiontotheeighttest‐colorsamples,sixsupplementaltest‐colorsamples(numbers9–14)areusedtofurtherassessthecolorrenderingabilitiesoftestsourceswhileCRI9toCRI14arereferred to as the special color rendering indices 9–14. After investigating thereflectivitycurves, thecolorsof the test‐colorsamples9–14haveparticularlystrongcolorswith relativelynarrowpeaksand thus these supplemental test‐color sampleshavethefollowingcolorseffect:9–strongred;10–strongyellow;11–stronggreen;12–strongpurplishblue;13–complexionofwhiteperson;14–leafoftree. Althoughpeople show several shortcomings of judging the color rendering capabilities of thelightsourcebyCIE’sCRImethod, this isstill internationallyagreeduponbypopularstandard.ThusthelightqualityischaracterizedinthisthesisbycalculatinggeneralCRI(Ra) and other special CRI (R9 – R14) considering the CRI for the Planckian locusilluminationsourcesastheV8lightengineproduceswhitelight,whichcanbetunedfrom3000Kto6000KwithlowDuv(<5*10‐3)values.

1.5.2.10ColorQualityScale(CQS):TheNationalInstituteofStandardsandTechnology(NIST) has developed another color standard (CIE TC1‐69) [51], [52] with closecontactsinthelightingindustryandtheCIEtoaddresstheproblemsoftheCIECRIforsolid‐state light sources and to meet the new needs in the lighting industry andconsumersforcommunicatingcolorqualityoflightingproducts.CQSevaluatesseveralaspects of the quality of the color of objects illuminated by a light source such aschromaticdiscriminationandobserverpreferences.TheCQSincorporates importantmodificationsinCRItoovercomeitsshortcomingsandfocusesonabroaderdefinitionof colorquality.TheCQSuses a set of 15Munsell sampleshaving themuchhigherChroma, span the entire hue circle in approximately even spacing, and arecommerciallyavailable.TheCRIcanresultinnegativevaluesforsomelampswhilsttheCQSimposesa0‐100scale.

1.5.2.11Televisionlightingconsistencyindex(TLCI‐2012):AnothercolorstandardisrelevantinthiscontextwhichisTLCI–2012[53].Thisprogramisintendedtobeusedfortheassessmentoflightsourcesforuseintelevisionlighting.Themetricdefinesa‘standard’camera,andthenusesittoanalyzeasetoftestcolorswhenilluminatedbyastandardsourceandbythesourceundertest.Thecolorimetricdifferencesbetweenthetwoexposureconditions,whenviewedonareferencedisplay,arecalculatedaccordingto theprinciplesespousedbyW.N.SprosonandE.W.TaylorofBBCR&D, in: ‘Acolortelevision illuminant consistency index’ by BBCR&DReport 1971/45. The primaryresult fromthismethodprovidesasinglenumber,rangingfrom0‐100,representingthe TLCI of the test light source. In general the higher number the better, with aperfectlyrenderinglightsourcehavingaTLCIof100aswithCRIandCQS.Usually,anylightsourcewithaTLCIof85orgreaterwouldbeusablewithavideocamerawithlittle


ornoadjustmentstothecamera.Asthescalegoesdown,generallythesourceswithaTLCIbetween50and85wouldstillbeusablebutwouldneedcorrectioninthevideochainsetuptogetacceptableresults.Finally,asourcewithaTLCIscalelessthan50maynot be usable, evenwith substantial correction, principallywhen used on sensitivecolorslikeskintones.

1.5.2.12IESTM‐30‐15:Becauseofthecomplexityofcharacterizingthecolorrenditionpropertyofalightsource;presently,thereisnoonemetricorscalethatcanaccuratelyenumerateallaspectsofcolorrenditionand/orclassifythemostdesirablelightsourceforeveryapplication.Thus,recentlyIESTM‐30‐15[54]–[56]elaboratesonamethodforevaluatingalightsourcecolorrenditionthatquantifiesthefidelity(closenesstoareference) and gamut (increase or decrease in Chroma) of a light source byincorporatinganobjectiveandstatisticalapproach.However,usingFidelityIndex(Rf)and Gamut Index (Rg), user could utilize the two‐dimensional characterization todeterminewhatismostappropriateforthespecificapplicationinconsideration.ThemethodalsocreatesacolorvectorgraphicthatspecifiestheaveragehueandChromashifts,whichalsointerpretsthevaluesofR1andR9.

1.6Summary

Thischapterdescribesthebackgroundoftheprojectasamotivationoftheproject.ThisisfollowedbytheprojectgoaltoachieveanadvancedopticallightingdesignforcolorcontrolinLEDstagelightingwhichwouldfulfilthepresentmarketdemandofachievinganenergyefficient,tunablewhite,homogeneousandpowerful luminouslightoutputwithoutsacrificingthelightcolorquality.Thischapteralsoprovidesthesubstancesofthefollowingsevenchapters.Fromsection1.4onwardsthechapterelaboratesonthestate‐of‐the‐art of the LED technology and the measuring technique of the LEDluminairewhichisadoptedinthisthesis.Thefollowingchaptercoversthestate‐of‐the‐artintraditionalstagelightingandtheprosandconsoverLEDstagelighting.

Chapter 2 23 Traditional lighting systems

Traditionallightingsystems

‘DeepajyothiparabrahmaDeepasarvatamopahahaDeepenasaadhyatesaramSandhyaadeeponamostute

Iprostratetothedawn/dusklamp;whoselightistheKnowledgePrinciple(theSupremeLord),whichremovesthedarknessofignoranceandbywhichallcanbeachievedinlife’

Chapter2 :Traditionallightingsystem

2.1Introduction

AsourmainpurposetoreplacethetraditionallightingbyLEDlightingforimprovementinenergysavingsandothercriticalaspects,accordingtotheapplicationrequirements,itisimportanttoknowindetailabouttheprosandconsofbothofthelightingsystems.Hence,thischaptermainlydealswiththetraditionallightingsystemscharacterizationwhichisespeciallyusedonthestageofatheatreorinastudio.However,beforegoingthere,weneedtoknowwhatstagelightingisandthehistorybehindit.

Therearetwotypesoflightingequipmentrequiredtoilluminatethestage;namely,general illumination and specific illumination. The general illumination provides adiffuse, shadow‐less, wash of light over the entire stage space. Candle, oil, gas, andelectriclampsareexamplesofgeneralillumination[57]–[60].Candleswereintroducedastheatrelightinginthelate15thcenturyinItalyandEngland.Inthelate17thcentury,aSwisschemistdevelopedthemodernoillampwhichsoonreplacedthecandlesastheprimarylightsources.Theworld’sfirstgasstagelightingsystemwasinstalledattheChestnutStreetTheatreinPhiladelphiain1816.Fromthenon,gaslightingwaswidelyadaptedandusedinexperimentalpurposesinmostcountriesoftheWesternWorld.Inthe 1880s, the incandescent mantle (burner) was introduced producing a muchbrighterandsaferlight.Hereafter,ThomasEdisonmadethefirstpracticalincandescentelectric lamp in 1879. By the end of the 19th century, most modern theatres hadreplaced gas lights by the much safer electric lights. In 1903, the Kliegl Brothers


installedanelectric lightingsystemwith96resistancedimmersat theMetropolitanOperaHouseinNewYorkCity.

Specificilluminationprovidesasharp,highlycontrolledshaftoflight,introducedbythelimelightinthemiddleofthe19thcentury.Thoseshaftswereusedtohighlightasmall area of the stage, a principle actor, or to create the illusion of sunlight ormoonlight.Theconceptofusinglightasspotlight,floodlightandfootlightwerestartedby using lens boxes or other alternatives. Thomas Drummondwas the inventor ofcalcium light,alsoknownas lime lightorDrummond light.Lateron, thenewerandbrightercarbonarclampbegantoslowlyreplacethelimelightinthemoderntheatre.Althoughthesaferincandescentspotlightstartedtograduallyreplacethecarbonarclamp,thearclampcontinuedtobeusedasafollowspotuntiltheendof20thcentury.The era of electric spotlight started in 1904. Kliegl brothers started to introduceinnovative design of planoconvex spotlights, Fresnel lens spotlight and ellipsoidalreflectorspotlightwhicharestillbeingusedtothisdateinstagelightingsystems.

Generally,afloodlightproducesalarge,almostuncontrollablewashoflightwhereasaspotlightproducesasmallhighlycontrolledpooloflight.Floodlightsaremostlyusedtolightthebackdrop(suchasscenery)whilespotlightsareusedtoshinelightontheactor.Table2‐1willexplainthecategoryandpropertiesoftheinstruments[60].

Table 2‐1 Instrument details, using in the stage lighting of theatre

Instrument Category Beamangle Spill Edgequality

Ellipsoidal Reflectorspotlight Fixed Verylittle HardFresnel Lensspotlight Variable Quiteabit SoftPARCan Spotlight Fixed Some SoftScoop Reflectorfloodlight Fixed Alot Soft

Borderlight Floodlight Fixed Alot SoftFollowspot Spotlight Variable Almostnone Hard

2.1.1Drawbacksoftraditionallighting

Recently the movement of LED light fixtures into the theatre lighting is graduallyincreasingduetothedisadvantagesofthestandardlighting[2],[3],[6].Thestandardlightingfixturesmostlyuseahalogenlightratedfrom300to1000W,andareenergyinefficient.Althoughtheselightsproduceasolidcoloraswellasagoodbrightnessatadistance,theyhavemanydrawbacks.Thereisalargeamountofenergywastedwithhalogenlightsasaresultoftheheatproduced.Thisdemandstoputanextraventilationsystemintheatres.Thus,thetheatreshavetobedesignedtoadequatelyventtheheataway from the lights. Beside this, the use of halogen lights requires a large energyinfrastructuretogetalargeamountofelectricity.Thedesignandinfrastructureforthelightingsystemtoaccommodateanextrasupplyofelectricityandventilationincreasestheinitialcostofthetheatres.Anotherdrawbackofusingthehalogenlightsistheshortlifespanofthelightbulbs,rated2000–4000hoursofuse[61].Thelifespanofbulbsisrelativelyshortintermsofhoursusedbyatheatre.Generally,alightusedfor3hoursaday,5daysinaweekwillhavealifespanbetween130and200days.ThereplacementcostforbulbsforincandescentlightsrangefromUSD15‐USD25/bulb,whichinturnneedtobereplacedabouttwiceayearforeachlight.Duetonothavingthecolortuning


facility, anotherdrawback to standard theatre lighting is theuseof gels togenerateothercolors.Theheatproducedfromtheincandescentlightfixtureslimitsthelifespanofthegelasitismadefromplasticmaterials.Thetransitionfromfloodpositionoflightto spot, there is almost~37%energy loss in the traditional lightingwhich is alsoashortcoming.

2.1.2LEDtheatrelighting

Withadvancedtechnologytheuseoflightsintheatresisgrowingdaybyday.ThereareseveraladvantagesofusingLEDlightingovertraditionallighting.EnergyconsumptionislessinLEDlightingcomparedtoincandescentlamp.Insteadofhaving300–1000Wfor incandescent lighting, LED lighting consumes only 100 W, the lower energyconsumptionfurtherreducesthecostoftheatreinfrastructure.LEDlightingproduceslessheatwhichhelpstoreduceenergyconsumptionoftheHVACsystem.AnothermoresignificantbenefitofLEDlightingsystemovertraditionallightingisthelongerlifespan.TheLEDmodulesusedintheatresareratedfrom50000hoursat70%operationwhichis12.5to25timeslongerthantraditionalbulbs.Thus,theLEDmoduleswouldlastfornine years. The maintenance cost and associated problems of changing bulbs aresignificantly reduced in LED lighting systems. They also eliminate the problems oftrapping dust and other particles in lights. To obtainmulticolor lighting, gel is notmandatoryforLEDlightingsystemsandLEDlightfixturescontainseveralcoloroptionswhicharepreprogrammedintothelightfixture.ThecolorchangingandmovingLEDlightsareverypopulartoobtainvariouscolorsinthatscenario.

TherearefewdrawbacksofusingLEDlightingsystemsintheatres.Basically,LEDlightingismoreeffectiveinashortdistanceasitislessbrightcomparedtotraditionallightingsystems.Thustheuseoftraditionallightfixturesinlongrangelightingisstillsuperior.Thewhite light fromonecompanyisnotthesameasthewhite light fromanothercompany.Thedifferenceincolorandqualitywillkeepthecustomercommittedtothesamecompanyinthefuture,shouldalightfixtureneedreplacing.Besidethis,thepriceofanLEDlightfixtureismoreexpensivethanthetraditionallightingsystems.Onthetopofthat, theLEDlightingtechnologyisboomingandthenewdevelopmentinLEDtheatrelightingiscreatingbetterlightseveryyear.Thelightfixturewouldlikelybereplaced insteadof the just thebulbas the technology continues to improve,whichmakestheearlierinvestmentinlightingsystemwastage.

Theaforementionedissuewouldberesolvedbythesolutionwhichisdescribedinthis thesis. The thesis demonstrates the light enginewhichwould provide the costeffectiveenergyefficientlightoutputconceptualizedbythelatestLEDtechnology.Theintelligentcolorcontrolinsidethelightengineregulatesthelightoutputwithincertainlimitsothattheusercannotdistinguishthelightvariationamongseveralsimilarlightengineswithinascene.Anotheradvantageofthedescribedopticaldesignisthattheenergylossfromfloodsettingtospotsettingisimprovedbyafactorof6.

ThereareseveraltypesofprojectorssuchasPROFILESPOTS(fittedwithknifeedgesoragobo),FOLLOWSPOTS,AMBIANCESPOTS,etc.,usedintheatresoronstage.Tounderstand the working functionality of traditional lighting and available LEDluminaire, the light output is characterized from the spotlight Halogen‐Fresnelluminaireassinglelensluminaire(Figure2‐1(a))andthecommerciallyavailableLED


luminaire (Figure 2‐1 (b)). The following sections describe the characterization indetail.

Figure 2‐1: (a) Halogen‐Fresnel spotlight (b) LED luminaire

2.2Characterizationof2kWHalogen‐Fresnelluminaire

Compact,highlyefficientFresnellensspotlightusingtungsten–quartz–halogenfilamentluminairehavingmaximumpowerof2kWfor120Vischaracterizedbymeasuringitslightoutput.Duringtheoperatingtime,theluminairewaspoweredby1.8kWintheQ–lab at the Technical University of Denmark. Both photometric and radiometricmeasurementswererecordedbya2mISandphoto‐goniometerandthemeasurementtechniques are already explained in Chapter 1 under section 1.5.1.1 and 1.5.1.2,respectively. The photometric integrating sphere measurement done by 2 πmeasurements isshown inFigure2‐2(a).Figure2‐2(b)and(c)showthe luminairealignmentduringthenear‐fieldgoniometricmeasurement.Thesizeofthepooloflightisdeterminedbythethrowandthepositionofthefocusingknob:SpottoFlood.TheexperimentalresultsareshowninTable2‐2.Themaximummeasurementdeviationbetweentwomeasurementsis0.4%.Fromthisresultitisclearthatthemeasurementresultshaveconsistentvaluesbetweenthetwomeasurements.Fromthetableitisseenthatthereis~37%lossinluminousfluxwhilethefloodlightingpositionchangestothespotlightingposition.

Figure 2‐2: (a) 2 π photometric measurements by 2 m integrating sphere; (b) Near field goniometer alignment of the luminaire; (c) Near field goniometric measurement


Table 2‐2 : Compared measured value of flux by an integrating sphere and goniometer, respectively

Halogen‐Fresnelluminaireat1.8kWelectricalpower

ΦMeasuredbyIntegratingsphere[lm]

ΦMeasuredbygoniometer[lm]

ΔΦ%

Flood 18745 18803 0.3Spot 11777 11717 0.3

Hand held Luxmeter (PRC‐ RdioLux 111)were used tomeasure the illuminancevalue of the luminaire at a field distance of 5.28m after setting the spot and floodpositions of the luminaire. For a different angle of illuminance, the intensity of theluminairewasalsocalculatedbyusingthefollowingequations:(26)‐(28)where‘r’isshowninFigure2‐3.Theluminancecanbemeasuredbytheequation(29).If =0,thentheintensitywascalculatedaccordingtoequation(30),where‘D’isshowninFigure2‐3.TheexperimentedresultsaredocumentedinTable2‐3.

(26)

, (27)

(28)

(29)

(30)

Figure 2‐3 : Diagram of LID for different angle of θ

Table 2‐3: Comparison of the experimental illuminance and intensity value for flood and spot position


Halogen‐[email protected]

E =0[lux]@3m(calculated)

E =0[lux]@5.28m(measured)



I =0[cd] Fieldangle[deg.]

FWHM[deg.]@50%

Floodbyluxmeter

3763 1212 941 418 33902 69 50.6

Floodbygoniometer

28345

Spotbyluxmeter

26364 8510 6591 2929 237273 24 10.8

Spotbygoniometer

243075

Inthetable,theilluminance(E =0)atdifferentdistancesforfloodandspotpositionofthe luminaire aswell as the intensity of the light by the luxmeter is reported. Themeasurementvalueofintensitybygoniometerisverifiedandthedeviationbetweentwomeasurementsismaximum2%.TheexperimentedLIDforfloodandspotpositionfromtheluxmeterandgoniometermeasurements,respectively,hasbeengraphicallyplottedinFigure2‐4.ThemeasuredCRIwas99.9andaverageCCTvaluewas3055K.

Figure 2‐4: (a) & (c) represent LID at different angles for flood and spot position respectively, measured by a luxmeter; (b) & (d) depict the same LID measured by goniometer, where the c‐planes

are depending on luminaire mounting direction

2.3CharacterizationofLEDluminaire

ThelightoutputfromtheLEDluminairehasamaximumofΦ=4000lm.Therefore,thepresent LED luminaire is also not fulfilling the current demand of stage lighting.


However,thetotalΦandLIDaremeasuredbyusinga2mISandgoniometer.TheΦforfloodandspotpositionsat3000Kare4052lmand3412lm,respectively.Figure2‐5(a)showsthemeasuredΦandcorrespondingmeasuredCCTofactualCCTtuningfrom2800Kto5800Katfloodposition.ThemeasuredCCThasalinearrelationwithactualCCT,R2=0.99.Forthefloodposition,themaximumΦis4975lmat4000KandtheminimumΦis3465lmat2800K.TheCRIatallCCTrangesisbetween97and90,seeninFigure2‐5(b).

Figure 2‐5 (a) Measured CCT and measured Φ for tuned CCT (2800 K – 5800 K); (b) Measured CRI for tuned CCT (2800 K – 5800 K)

ThegoniometricmeasurementsareshowninFigure2‐6forfloodandspotpositions.Theenergylossfromthefloodlightingpositiontothespotlightingpositionis18%.

Figure 2‐6:(a) LID at flood (b) spot positions respectively for the LED luminaire where the c‐planes are depending on the luminaire mounting direction

2.4Summary

This chapterdescribes theprosandconsof traditional stage lightingandcomparedthose properties with an LED luminaire. There are two examples of lightcharacterizationofthetwoluminaires,oneisthetraditionalHalogen‐FresnelluminaireandtheotheroneistheLEDluminaire.Boththeluminaireshowsthehighenergylossduringthetransitionfromthefloodlightingpositiontothespotlightingposition.TheexperimentalresultsprovethatthepresentLEDluminairecannotfulfiltheimmediatemarket requirements of LED stage lightingwith a high luminous flux (>10,000 lm)


withoutcompromisingthelightquality.ThefollowingchapterswillintroducewiththeV8lightenginebyelaboratingontheopticaldesign(Chapter3),colormixingtechnique(Chapter4)andcontrolsystem(Chapter5)oftheluminaire.TheexperimentalresultsfromtheV8lightengine(Chapter6)willshowthebettermentofstagelightingovertraditionallighting.

Chapter 3 31 Optical design and modelling

OpticaldesignandmodellingofV8lightengine

‘Musicisthearithmeticofsoundsasopticsisthegeometryofthelight.’–ClaudeDebussy

Chapter3 :OpticaldesignandmodellingofV8light

engine

3.1Introduction

Asmentionedinthesection1.2inChapter1,theaimsoftheV8opticaldesignare;(1)toreplacethetraditionalstagelightingbyimprovingtheopticalefficiency,(2)toobtaincompact focusable, homogeneous and high luminous flux, (3) to provide dimmableintensityandtunablewhitelightoutputfrom3000Kto6000K,(4)tohavehighlightqualityintermsofCRI(>85),(5)toimprovetheenergylossfromfloodsettingtospotsettingand lastnot the least, (6) toget intelligentspectralcontrol toprovideoutputstability.ThischapterdealswiththeopticaldesignandmodellingoftheV8lightengineto simulate optical efficiency of the entire light engine, which would address theaforementionedgoals.Inprinciple,themodellingwillleadtoanoverallestimateoftheluminous light output from the outer end of the optics. The V8 light engine is theintegrationofmorethanoneopticalcomponent.Thesimulationwillbetakingcareofoptimizationofcriticalopticalcomponent in termsofobtainingmaximumluminouslightoutput,propercolormixingandoptimumcolorhomogeneityoftheentireopticalsystem.ThenextparagraphwilldescribetheV8lightenginefollowedbythedetailsofoptimizationofthecomponentusedintheV8opticalsystem.

3.2DescriptionofV8lightengine

TheobjectiveoftheV8lightengineistobeapowerful,focusablelightengineinstagelightingapplicationwhichcouldbeusedintheatres,auditoriums,orsimilarotherstageshows.Itisdesignedtoproducetheluminouslightoutput(Φ fromthelightenginetoexceed20,000lmwithanextremelyhighcolorhomogeneityintheaimedregion,whichisgoingtobepublishedinascientificjournal[14].Theschematicdiagramofthelight


engineisseeninFigure3‐1(a)whichconsistsof:(1)Printedcircuitboard(PCB),(2)LEDs,(3)Totalinternalreflection(TIR)lenses,(4)Parabolicreflector,(5)Opticalgate,(6)Microlensarray(MA),and(7)Fresnellens.TheparabolicreflectorandTIRlensesareusedasprimaryopticsandtheMA,andFresnellensareusedassecondaryoptics.The optical gate is surroundedbydifferent coloredLEDs andTIR lenseswhich aresittingonthePCB,makinganLEDclusterofsevenringsinsidetheV8lightengine(asseen in Figure 3‐1 (b)). The quasi‐collimated beam from the individual LEDstransmittedthroughthetotalinternalreflectionlens(TIR)isreflectedbytheparabolicreflectorandisdirectedtowardstheopticalgate,whichissituatedatthefocalplaneofthereflector.Theopticalgateisactingasanapertureandlimitssomeofthereflectedbeam to pass through the gate. The hexagonal patterned double‐sided convexMA,calledaKohler integrator, isusedatthegatepositiontocombinea largenumberofquasi‐collimatedbeamsfromtheindividualcoloredLEDsarrivingattheMAatdifferentanglesintoahomogenouscolormixedbeam.Toobtainbettercolormixingandcolorhomogenizationfromthelightengine,theMAiscustomized,whichisalsogoingtobepublishedinascientificjournal[13]anddiscussedin3.3.6.Thelightenginemixesthecolor from fourcoloredLEDs (Green,Red,Blue, andRoyalBlue)withseveralwhiteLEDstoproduce intensitydimming(minimum5%of total luminous flux)andcolortuning(3000Kto6000Kwitha100Kresolution)whitelight.Finally,theFresnellensisusedtocollectalllightoutputfromthegateandtoprovidecollimatedlightoutputfromthelightengine.ThetransitionoftheFresnellenstowardsthegatecanchangethefocusablespotpositiontofloodsetting.

TomaximizethelightoutputsuchasΦandtoachieveagoodlightqualitywithahighCRI(>85)andalowDuv(<10‐3),theusedcomponentsinsidethelightengineneedtobeoptimized.AspectralmodelforoptimizingthecolormixingofLEDbeamisdescribedinChapter4.TheoptimizationonthegeometricalopticaldesigntoaccomplishahighΦisdonebyusingtheopticalsoftwareZemaxwhichiselaboratedonthenextsection.Thecomponent description and functionality used inside the V8 light engine is alsoparticularized on the next section. The V8 light engine also incorporates the colorfeedbacksystemwhichwillhelptogetagoodconsistencyofluminouslightoutputaswellascoloroveroperationaltimewhichisdescribedinChapter5.


(a)

(b)

Figure 3‐1: (a) Schematic diagram of the V8 light engine design with the position of optical components (not in scale); (b) Seating arrangements of four colored and the white LEDs on the PCB forming seven rings where beams coming from the LEDs are collimated through the individual TIR

lenses; (Ref. Figure 1 of Chakrabarti et. al [14])


3.3V8lightenginecomponentsanddesignoptimization

TheoptimizationparametersforthecomponentsusedintheV8lightenginearelistedinFigure3‐2,elaboratesonthemethodsandoutcomesfromtheoptimizationprocess.

Figure 3‐2: Description of the component optimization order, parameters, outcome, and the process tool

3.3.1PCBshapeandsize

ThePCBusedinV8ismadebytwolayersusingmetalcorePCB(MCPCB)[62]–[64].Thefirst layer is made in aluminum. Then there is an isolation layer made bydielectricum[65]onwhichacopperlayerisplacedandthenaphotoresistandetchingprocess is applied. After that, soldermask or solder resist is applied for protectionagainstoxidationandtopreventsolderbridgesfromformingbetweencloselyspacedsolderpadsandfinally,itissilkprinted.TherearesixsegmentstowherethecolorLEDsis distributed in a pattern to optimize the colormixing (Figure 3‐3). The LEDs areconnected to eachother by serial electrical connection to avoid thermal runway byparallelconnection.EachLEDstringhasamaximumvoltageof60V/stringduetoLEDdriverlimitationsandhighvoltagesafetyregulations.


Figure 3‐3: The PCB used in V8 distributing LEDs within six segments (made by Brothers & Brothers Sons)

The size and shape of the PCB was designed using SolidWorks in order toaccommodatemaximumnumberofLEDswithinagivenspace.ThefollowingFigure3‐4(a)indicatestherelationfoundintermsofPCBsizeandtheLEDnumberstobeaccommodatedwhereyistheLEDnumberandxisthePCBsizeinmm.Figure3‐4(b)showstheprobableLEDpositiononthePCB.

Figure 3‐4: (a) The relation between PCB size and number of LEDs to accommodate within the PCB (b) The pictorial indication of LED position on the PCB

3.3.2LEDs

Keepinginmindofachieving~20,000lmofluminousflux,210numbersofLEDsaredistributed innumbersof18, 24, 30, 12, and126Royal‐blue,Blue,Green,Red, andWhiteLEDs,respectively.ThedifferentcoloredLED’sarepositionedinsidetheclustermakingapattern(asseeninFigure3‐5)bywhichthemixtureofbeamsfromdifferentcoloredLEDsprovideagoodoverlapwitheachother.


Figure 3‐5: Projected light on the wall from the V8 light engine when all (a) Royal‐blue, (b) Blue, (c) Green, (d) Red, (e) White, and (f) All LEDs are on at 100% DMX value

Generally, a bared LED has a large viewing angle (~110°‐130°), which makes itdifficulttotailorthelightforprocessing.Hence,aTIRlensisusedontopofeachLEDtocollimatetheLEDbeamcomingfromtheindividualcoloredandwhiteLEDs,whichislater focused into the optical gate by the parabolic reflector as seen in Figure 3‐1.Chapter4(Section4.2.2)andChapter6(Section6.2)describemoreabouttheLEDsandtheirpropertiesrelatedtotheV8lightengine.

3.3.3Reflector

The reflector as primary optics inside theV8 light engine plays a significant role tomaximize the light output. The light extracted from the TIR lens is reflected andafterwards directed towards the gate by the reflector. To direct almost all reflectedbeamstowardsthegate,theimportantparameterstobeconsideredare:(a)reflectortype,(b)shapeandsize,and(c)material.Therearedifferenttypesofreflectors.Amongthemparabolicreflectorandfreeformreflectorarepopular.Aparabolicreflectorhastheshapeofaparabolawith the lightsource typically locatedat thereflector's focalpoint.Therefore,thelightraysparalleltooneanotherinasymmetricalpatternandcanfocustheavailablelightintoatightbeam.Intheparabolicreflector,movingthelightsourcetowardthereflectorwillspreadthelight,whereasmovingthesourceof lightawayfromthereflectorwillcausethelightraystoconverge.Theparabolicreflectorisutmostefficientinapplicationsthatdon'trequiresharpcut‐offs;however,allowallofthereflectorsizetobeutilized.Thus,alargerparabolicreflectorprojectsmorelightforthoseapplicationswhicharenotrestrictedbyavailable installationspace. Freeformreflector [66], [67] allows the reflector's surface to be calculated point by point toilluminatecertainareasofinterest.Thedesignoffreeformreflectorsurfacetechniquehasmadeitpossibletoachieveafavorablebeampatternsolelyproducedbytheshapeofthereflector.Theinvestigationsonthefreeformreflectorandtheparabolicreflectorbysimulationarecarriedouttomaximizethelightreflectionfromthereflector.Thereflector size and shape are optimized by simulation in order to obtainmore lightoutputfromthelightengine.Thatmeansthecomparisonwasmadebyevaluatingtheoptical efficiency of the simulated light engine by using both types of reflectors(parabolic and freeform). Optical efficiency of a system is generally defined bymeasuringhowmuchlumensofthetotallumenssentintothatsystemcomeoutofit.As


anexample,ifopticalefficiencyofacommoncollimatorlensforanLEDhastaken,itmeansthatfirstthetotallumensoftheLEDaremeasuredseparatelyinanintegratingsphere.Asasecondstep,allthelumensoutofthissystemaremeasuredinthesameintegratingspherewherethelenstobemeasuredisplacedontopoftheLED.Finally,theopticalefficiencyistheratioofthesetwolumenvalues,multipliedwith100,togetouttheresultin%.IntheV8lightdesign,itisobservedinsimulationthattheparabolicreflectorprovideshigheropticalefficiencythanareflectordesignedbythefreeformoptics. This is because the parabolic reflector in overall delivers higher opticalefficiencies for theTIR lenses that arenear to theoptical gate than the free formedreflector(Figure3‐6).InFigure3‐6,theTIRlensespositionedatfirstandninthringaresituatedat6.45mmand110mmawayfromtheedgeoftheopticalgate,respectivelywhiletheotherringnumbersaresituatedinbetweenofthefirstandtheninthring.However,theopticalefficiencydecreasesgraduallyfortheparabolicreflectorwhenthepositionoftheTIRlensisfarfromthegate.IrrespectiveofpositionoftheTIRlenswithrespecttothegate,theefficiencyisconstantuntilthesevenringpositionforthefreeformedreflector.AsseeninFigure3‐6,attheeightringposition,theefficiencyishigherforthefreeformedreflectorcomparedtotheparabolicreflector.Thisisbecausetheshapeof the free formedreflector isoptimized inordertoreceivemaximumopticalefficiencyonlyattheeighthringpositionwhereastheshapeoftheparabolicreflectorisoptimized to receive over allmaximum optical efficiency from the systemwhich isirrespectiveoftheparticularposition.

Figure 3‐6: Simulated optical efficiency vs LED position in the ring (Ref. Figure 6 of Chakrabarti et. al [14])

Todesignaparabolicreflectorthebasicequationofparabolaneedstobeconsideredisequation(31):

(31),

whereyisthesymmetryaxisoftheparabola,directrixisonxaxisandAandCaretheconstants:

∴ 2

(32)


∴

1 44

14

(33),

wheref isthefocallengthofthereflector.Fromequation(33),weseethatthefocallengthisindependentofxandthusallraysparalleltotheyaxiswillpassthroughthefocusafterreflectingfromthemirror.Thatalsomeanstheoff‐axisrays,whicharenotparallel to the axis, will not pass through the focus after reflection. Although thedivergenceangleofrayscomingfromLEDsthroughTIRopticsisless(~±5°–±8°),theyarenotconsideredaspurelycollimatedlightwhentheyfallintothesurfaceofparabolicreflector.Somostofthelightreflectedfromtheparabolicreflectorcannotmeetatthefocalpointoftheparabolicreflectorandthuseventuallyatthefocalplanetherewillbeaspreadofmeetingbeams.Thisisexploitedtodecideonthegatesizefromwherethefinallightoutputistobeextracted.Tomaximizethelightextractionfromthegate,thegateisplacedatthefocallengthofthereflector.Theparabolicreflectedsurfaceismadebytwolayers.Initially,itisspincoatedbyhydrogensilsesquioxane(HSQ)andthenaphysicalvapordeposition(PVD)ofhardcoatingofaluminumisappliedbysputteringmode[68].Initially,theprocessofcoatingwasnotgoodandbecauseofthatthelightengine had 30% of reflectance loss from the non – uniformity in the reflector.Thereafter,thereflectorwasmodifiedwithacontrolledcoatingofaluminumandtheprocesshelpedtogetagoodsurfaceuniformitywithreflectanceof~89%(Figure3‐7),whicheventuallyimprovedanaverageof24%inluminousfluxoutput.AlthoughtheflatnessofthereflectorisnotacriticalparameterfortheV8lightengineapplication,thereflectedlightvariationfromdifferentpartofthereflectoriswithin~2%.

Figure 3‐7: The uniform reflectance of the parabolic reflector at visible wavelengths

3.3.4Lens

The viewing angle of an LED is normally~110° – 130°which for beam shaping isextremelydifficulttoworkwith.Thus,theTIRlensisusedintheV8lightenginedesign,so thatwhen the LED beams incident on the surfaces of the TIR lens, after havingseveralreflections;thetransmittedbeamcanbecollimated.However,beforeusingtheTIRlens,thecollimationfromanormalcollimatedlensisalsoexaminedandcompared


withtheopticalefficiencyoftheTIRlens(Figure3‐8).FromthefigureitisseenthattheTIRlenshasalmostthesame~78%opticalefficiency(higherthancollimatedlens)attheexitanglefrom20°to70°whichisusefulfortheV8lightengineapplication.

Figure 3‐8: Optical efficiency comparison between a TIR lens and a normal collimating lens

TheTIRlensisdesignedbyfree‐formopticsusingMathematicawherethesurfacesofthelensisoptimizedthroughapiece‐by‐pieceoptimizationprocessconsideringapointsourcerayanglesranging from0° to90°withastepof1° ineachtime. Initially theprototypeoftheV8lightengineusedaTIR lensofdimension11.7mm(±0.2mm),which laterwasmodifiedwith a bigger size of dimension of 12.7mm (± 0.2mm)(Figure3‐9)toimprovetheoverallopticalefficiencyoftheV8lightengine.

Figure 3‐9: The TIR lens showing a point source at 0° and then tilted by 30°, 60° and 90°, respectively


For the two different sizes of TIR lenses, Figure 3‐10 ((a)‐(b)) shows theexperimentalopticalefficiencyforwhiteandblueLEDsaccordingtotheirpositionatPCBwithrespecttothegate.Theexperimentalmethodisexplainedinthe6.4sectioninChapter6.Asseeninbothfigures,theopticalefficiencyishigherforbothLEDs(whiteand blue) when the system is assembled without a MA. That means that the MAobstructsthelightfrombeingtransmittedthroughthegate,whichwebelieveisduetothecutoffanglefromtheMA.TheTIRlenseshaveadivergenceofmaximum±8°withrespecttotheparallelbeam.Duetothisdivergence,theincomingbeamtowardsthegateafterreflectionfromtheparabolicmirrorbecomesmoreobliqueiftheLEDsaresituatedfarfromthegate.TheviewingangleoftheblueLEDishigherthanforthewhiteLED.Thus, thedivergence isalsohigher for theblueLEDs than for thewhite. ThatexplainsthattheopticalefficiencyoftheblueLEDsisalwayshigherthanforthewhiteLEDs,whilethesystemisnotusingaMA.Thesizeofthelensalsoinfluencesthelightextractionfromthelenswhichconsequentlycontributestothefinallightoutputfromthesystem.TheFigure3‐10(c)showsthesimulatedcomparisonoflighttransmissionbetweentwodifferentsizesoftheTIRlenseswhilealightcollectingdetectorissetataconstantdistance.ThefigureshowsthatthelighttransmissionfromabiggerTIRlensisalmostconstantfortheLEDprintcircuitboardsizeupto2mmwhereastheLEDprintcircuit board size influences the light transmission for a smaller TIR lens. Thus theoptimallenssizeof12.7mm(±0.2mm)isusedinthemodifieddesign.Theincreasedlenssizeaidedtoimprovethelightoutputfromthelensbyaverageof~56%comparedtotheprototype.WehaveoptimizedtheTIRlenssizewithrespecttotheilluminationfromthewhiteLEDs.Hence,thereisamaximumof~50%opticalefficiencyhikeforthewhiteLEDsbyusingthelargersizedTIRlens.ThebiggerTIRlensalsohelpstheblueLEDstoincreasetheiropticalefficiencyby~65%.However,thelightextractionfromtheblueLEDsstillneedstobeoptimizedforahigherefficiencywhichisplannedforfutureresearchproject.AlthoughtheMAandtheTIRlenssizeareoptimizedtoenhancethelightoutputfromthesystem,thepositionoftheLEDswithrespecttothegatestillinfluencestheLEDefficiency.Thus,theLEDefficiencydecreaseswhilegoingawayfromthegatewhichisindicatedintheFigure3‐10(a)(b).However,byusingwhiteLED,itisnoticedthatthetransmissionfromTIRlenswith12.7mmsizeis86%.Therefore,188lmlightwasobtainedoutoftheTIRlensifthedrivingcurrentofthewhiteLEDis1A.


(c)

Figure 3‐10: (a) and (b) Optical efficiency vs LED position to the gate for small and big TIR lenses, respectively (Ref. Figure 7 of Chakrabarti et. al [14]) (c) Optical transmission from the TIR lens vary

according to LED print circuit board size

Further light transmission from different parts of the TIR lens is investigatedrigorouslybysimulationusingZemax.AsexpectedthedispersionofraysatthegateduetomarginalrayswithregardtotheTIRlensislesserthanthecentralorprinciplerays(Figure3‐11).DuetothetransitionoftheTIRlensfromonesurfacetoanother,therayscannottransitproperlyleadingtounrealisticdataresultswhicharedisregarded.


Figure 3‐11: Dispersion of ray angles at the gate due to displacement of the ray from a center position of the TIR lens due to LED print circuit board size

IntheV8 lightenginedesign,eachLEDisencapsulatedbythe individualTIR lenswhichisalsoplacedonthePCBinsertingits3legsintotheholesinthePCB,andisfixedbyapplyingglueasseeninFigure3‐12(a).ThematerialusedforTIRwasPolymethylMethacrylate(PMMA)andtheinjectionmoldtechniquewasadoptedtomaketheTIRlens[69]–[71].

(a) (b)

Figure 3‐12: (a) The TIR lens is placed by applying glue (b) 210 TIR lenses were placed in the first prototype of V8


3.3.5Gate

Theoptimizationofthesizeofthegateanditspositionwithrespecttothereflectorinordertomaximizethelightextractionfromtheoutputendofthegate(Figure3‐1)isdonebyusingsimulationsinZemax.Figure3‐13showsthesimulatedcumulativeΦinlmattheoutputendofthegatewhilevaryingthegatesize.However,thepositionofthegatewaskeptconstantwhichisatthefocalplaneofthereflector.ThefigurealsoshowsthetotalΦat individualringpositionfortheaforementionedcases.Althougha55mgatesize ishavingthehighestcumulativeΦvalue, thereflectorsizeneeds tobe0.8timeslargerthantheusedreflectorsizeforthegatesizeof50mmwhichisthesecondhighestΦcontributor.Therefore,theoptimizedgatesizeischosenas50mmfortheparabolicreflector.Althoughattheeighthringposition,thefree‐formreflectorprovidesthe highest totalΦ, it is seen that the used free‐form reflector is having the lowestcumulativeΦvaluewithagatesizeof50mm.

Figure 3‐13: Considering the varying gate size, cumulative as well as individual luminous flux for each ring number

3.3.6Microlensarrays(MA)

TheV8 light enginedesignuses a double‐sidedhexagonal patterned convexMA, toperformcolormixing.ThepurposeoftheMA,calledaKohlerintegrator[72],[73],istocombinealargenumberofquasi‐collimatedbeamsfromtheindividualcoloredLEDsenteringtheMAatdifferentanglesintoahomogenouscolormixedbeam.

3.3.6.1TheKöhlerIntegrator

TheKöhlerintegratororrelatedilluminationsystemconsistsoftwolensarraysandacondenser lens formingside‐by‐sidemultipleKöhler illuminationsystemsasseen in


Figure 3‐14.InitiallyafirstlensarrayshownasLA1dividestheincidentlightandcreatesmultipleimagesofthelightsourcesintheapertureplane.ThefirstlensarrayLA1alsoassists as an arrayof fielddiaphragmsdescribing the illuminatedarea in theobjectplane. The following conditions can apply for the Köhler integrator as explained inVoelkelet.al[70]:

A.TheLA1actingasacollectorlensimagesthelightsourcetotheplaneoftheapertureplane.

B.TheapertureplaneispositionedinthefrontfocalplaneoftheLA2andcondenserlens.Bythis,eachpointintheapertureplaneisimagedtoinfinityatobjectplane.

C.Thefieldplaneisimagedtothetargetplanebyproperadjustmentofthedistancefromthecondenserlenstotheobjectplane.

Thisadjustmentensuresauniformilluminationofboththeobjectandtheimageplane.Intheobjectplane,realimagesofthesub‐aperturesoftheLA1superimposetoLA2.Assumingthatthelightirradianceisapproximatelyuniformandsymmetricovereachsub‐aperture of LA1, the overlapping of all images provides a uniform intensitydistribution in the object plane. Generally, 10x10microlens arrays are sufficient toachievegoodflat‐topuniformity.Therefore,intheV8lightenginedesign,severalLEDsact as light sources incident light on the first surface of the double sidedMA afterreflecting from parabolic reflector. After that the real images of microlenslet arrayproduced by the first surface of MA superimpose to the second surface of MA toconstructinguniformillumination.Fresnellensisusedasobjectivelens.

Figure 3‐14: Köhler Integrator for uniform illumination (Ref..[74])

AstheMAisacriticalcomponentforV8lightenginedesign,theextensiveresearchwasdone for the lensdesign andon the fabricationprocess. Finally a cost effectiveprototype of the MA was made in‐house which is reported through scientificpublication[13].TheworkwaspartiallyfundedbytheDanishCouncilforTechnologyandInnovationunder the InnovationConsortiumLICQOP,grant#2416669andTheDanishNationalAdvancedTechnologyFoundationundertheV8 lightengineproject(037‐2011‐3).


Thissectionalongwiththefollowingsubsections,willdescribeanovelprocess,fortoolfabricationutilizedforinjectionmoldingofthemicrolensstructuresbasedonthepreviouslypublishedarticlebyChakrabarti et. al[13].Theprocess exploits standardCNCmilling to make a first rough mold in steel. After that, a surface treatment isintroduced by spray coating with photoresist to obtain an optical surface qualityrequiredforMAmoldandhenceinexpensiveinjectionmoldingreplicationprocess.Thefabricationprocess isdemonstratedby theproductionofø50mmdouble‐sidedMAdesignedforcolormixingfortheV8lightengine.TheMAisdesignedandoptimizedthelensletparametersforlargeangularacceptanceangleandthemoldingtoolisproducedfromthese.TheMAistheninjectionmoldedinPMMAfromboththeuncoatedroughmold and the photoresist coatedmold. The injectionmoldedMA are characterizedthrough surface measurement by 3D microscopy. The working performance isobservedbyangulardependenttransmissionandinvestigationoflightscattering.

3.3.6.2DesignofMA

Figure3‐15illustratesthebasicfunctionalityoftheMAbylookingattheraypathsinanoptical simulation, using Radiant Zemax® software, for three collimated beamsincident on a single lenslet demonstrated by three different colored beams and themicrolensarray issimulatedherewith five lenslets.Figure3‐15(a)showsthecolormixing mechanism for collimated beams at incident angles of 0°, Θ1° and ‐ Θ1°,respectively.Thethreecolorbeamsareallfocusedonthesecondsurfaceofthesamelenslet. The displacements in position of the focal points in the focal plane aredependent on the incidence angles. Themaximum incident angleΘ is governed bynumericalaperture(NA)of the lenslet(equation(34))and isexpressedbyequation(35). From the second surface the three color beams exit the lenslet in the samedirectionwiththesameexternaldivergencehalf‐angle,asΘ.

sin Θ

2 ∗ 2

≅2 2

≪ (34),

∴ Θ asin ≅ asin

2

(35),

HereDisthewidth,andfisthefocallengthofthelensletandnistherefractiveindexofthelensmaterial,asindicatedinFigure3‐15(a).Hereristheradiusofcurvatureandt is theseparationdistancebetweentwoopposite lensletasseen inFigure3‐15(a),respectively.tneedstobeequaltof,sothatacollimatedbeamisfocusedonthesecondsurface of the lenslet. It is seen that a large divergence angle Θ requires a largenumerical aperture, e.g. large n and D compared to f. In the normal operationalsituation,theincidentcollimatedbeamsarelargerindiameterandhenceilluminatealarge number of lenslets, and the beams are split into a corresponding number of‘emitters’ fromeach lenslet.Hence, thedifferent coloredbeams incident atdifferentanglesontheMA,arespatiallysplitintoamultitudeofemitterswithequaldivergenceanddirection,andresultsinauniformcolormixingoftheincomingbeams.


(a)

(b)

Figure 3‐15: Basic color mixing functionality of a double‐sided, convex microlens array, with incident beams at angles (a) lower than and (b) higher that the acceptance angle. (Ref. Figure 1 of Chakrabarti

et. al [13])

If the incidence angle is larger than the angle Θ given in Eq. (35), the incidentcollimatedbeamisfocusedandcoupledintotheneighborlenslet.ThisisillustratedinFigure3‐15(b)bythecollimatedbeamcomingfrombelow(redrays).Itisseenthatiftheincidentangleexceedsthecriticalangleattheinterfaceofthesecondlenssurface,the beam is entirely internally reflected, where a part of the beam returns in thebackwarddirection. In thatscenario, ideally therewillbeno transmittance fromthesecond surfaceof theMA.Thus, this angle canbe called the cutoff angle,Θ=Θc , orequivalently;theacceptanceangleoftheMAaswithinthatangletheMAallowsbeamstobetransmittedfromthesecondsurfaceofthelenslet[72].Figure3‐15(b)illustratesthecasewhereacollimatedbeamis incidentatanangle larger thantheacceptanceangle,Θ2>Θc(bluerays).Inthiscasetheraysaretransmittedintheforwarddirectionfromtheneighboringlenslet,butatlargeangleslargerthanthedivergenceangle,Θ=Θc.Therefore those rays do not contribute to the effective color mixing within thedivergenceangle,Θc.


According to the above, the transmission of theMAwould therefore be high forangleslowerthanΘcandclosetozeroatthisangle,andthenincreaseagainatlargerangles.Figure3‐15onlyillustratestheprinciple,soinordertogetabetterideaabouttheefficiencyofthesystem,asimulationismadeofthetransmissionthroughtheMAusingacollimatedlargebeamandvaryingtheincidenceanglefrom0to50°.Inordertoinvestigatetheoutputefficiencyasafunctionofexitangle,thesimulationisdoneforadetector that measures the light transmitted, within an exit cone of varying angle.Figure3‐16showsthetransmissionefficiencyoftheMAforthevaryingincidentanglesandforvaryingexitconehalfangles.Itcanbeseenthatforincidenceanglesfrom0‐30°,morethan90%ofthelightistransmittedwhentheexitconehalfangleislargerthan40°.ThiscorrespondstothecaseinFigure3‐15(a),wherethebeamsareincidentatangleslowerthantheacceptanceangleandwheretheexitconecoverstheacceptanceangle.For incidenceanglesaround40°almostnolight istransmittedwithintheexitconeof 40°. This corresponds to the casewhere the incidence angle is close to theacceptanceangleasillustratedinFigure3‐15(b)bytheredrays.ThetoppartofthegraphinFigure3‐16correspondstoalargeexitconeangleofaround80°e.g.collectingall the transmitted light. Here efficiency higher than 40% is observed for incidenceanglesaround45°,correspondingtothecaseinFigure3‐15(b)(bluerays),however,thistransmittedlightdoesnotcontributetothemixedlightwithinthedivergenceangle.

Figure 3‐16: Simulated optical efficiency in % of double‐sided microlens array as a function of incident angle while exit angle varies (Ref. Figure 2 of Chakrabarti et. al [13])

To achieve a high optical efficiency of the light output from the light engine, theacceptanceangleoftheMAneedstobeoptimized.ThereforetheMAfortheV8lightenginehasbeendesignedtohavealargeacceptanceangle.ItisseenfromEquation(35)thatalargedivergenceangleΘrequiresalargenumericalapertureNA,e.g.largenandD compared to f. However, the overall transmittance decreases linearly with theincreaseofnofthematerialduetoFresnelloss,i.e.reflectionsatsurfaceairinterfaces.Thus there would be a tradeoff between transmittance and acceptance angle. For


injectionmoldingthechoicesofmaterialsmostwidelyusedarepolycarbonate(PC)orPMMA.AlthoughthehighernofPCyieldsalargeracceptanceanglethanusingPMMA,the transmittance is lower.A simulationof the transmittanceof theMA for the twodifferentmaterialshavebeenperformedusinganincidentconeoflightintegratingoverall angleswithin the coneangleanda large exit coneangle. Figure3‐17 shows thissimulated transmittance as a function of the incident cone angle for PC andPMMArespectively.ItshowsthatthetransmittanceofthePMMAMAishigherthantheoneusingPCforall incidenceconeangles.ThetransmissionthroughPMMAmicrolensis~3%higherthanforPCatalmostperpendicularincidence.PMMAalsohas~5%largerefficiency at high incident cone angle of 50°. In our application the transmittance isprioritizeandchosePMMAasthematerialforthemicrolensfabrication.

Figure 3‐17: Material refractive index dependence transmission for double‐sided microlens array as a function of incidence cone (Ref. Figure 3 of Chakrabarti et. al [13])

InordertoenhancetheopticalefficiencyintheV8lightengine,thehighNAvalueisalsorequired.Thelensletsaresphericalwitharadiusofcurvature,r,andthediameterD of the single lenslet varies with the sag [75] which is also responsible to give aspherical shape of the lenslet. A thorough study was done on different lensletparameters (Figure 3‐18) targeting the enhancement of the acceptance angle andfinally,thedesignwasoptimizedbysimulation.TheFigure3‐18(a)indicatesthatlowf#yieldsahighacceptanceangleasforlowf#thelenswidthishigh.Thehighsagprovideshighacceptanceangleraisingthelensletwidth(inFigure3‐18(b)).However,whenthesagalmostappearsclosetor,theacceptanceanglebecomeslow.


(a) (b)

Figure 3‐18: The variation of acceptance angle with respect to (a) f# and (b) sag of the lenslet respectively

The followingTable 3‐1describes thedesigned lensparameters used for theMAfabrication after optimizing the parameters using simulation by Radiant ® Zemaxsoftware.

Table 3‐1: Lens parameters for fabricated Microlens (Ref. Table 1 of Chakrabarti et. al [13])

MaterialMicrolensarraysize[mm]

Totalthickness(doublesided)

[mm]

Lensletcurvature[mm]

Sag[mm]

Lensletwidth[mm]

Focallength(mm)

Acceptanceangle[°]

PMMA 65 2.1 0.85 0.55 1.45 1.4 82


3.3.6.3Fabricationprocess

Inthissectionthefabricationofthedouble‐sidedMAisdescribed,fromtheComputerNumericalControl(CNC)fabricationofmold,thecoatingprocedurewithphotoresistandtheinjectionmolding.Computernumericalcontrol(CNC)machineperformsboththefunctionsofdrillingandturningmachines.CNCmillsarecategorizedaccordingtotheir number of axis and are conventionally programmed using a set of computergenerated codes that represent specific functions. Therefore, the CNCmachinewasloadedbythecomputergenerateddrawingfile,madebySolidWorkssoftware,whichholds the design optimized lens parameters information (Table 3‐1) about theMA.After that the CNCmachine inscribes the loadedMA structures on a steel block asshowninFigure3‐19.Asimilarkindofblockneedstobemadeandheldontopofthebottomblock.Thealignmentofthetwoblocksrequiresprecisionsothatthepositionsof themicrolens structures in both blocks canoverlap eachother.As shown in thefigure,thereisaninletforinjectionmoldingattheedgeofthemicrolensstructure.TheaccuracyoftheCNCmachineis+/‐50µm.

Figure 3‐19: Inscribed microlens array structure on a steel block (Ref. Figure 4 of Chakrabarti et. al [13])

The surface roughness of the steel mold can be reduced by photoresist coating.Photoresist is defined as a viscous polymer resin (solution) which containsphotochemical active polymer (PAC). There are two types of Photoresist, namely,negative and positive photoresist, respectively. Spin or spray coating is the most


commonwayofapplyingPhotoresistonwaferoranysubstrate. Spincoating isnotpossible for this application and in order to overcome this problem MicroSprayphotoresistisused.Microspray[76]isapositiveacting,aerosolnovolakresistandsuitsourapplication.It iscosteffective,easytouseandsprayeliminatestheirregularitiesandpeak–valleyvariationsonthesurfaceofthemoldinaccordancewiththeprojectrequirements.Forthebestuse,theMicrospraywasplacedatroomtemperature(~20°)foranhourpriortouse.Beforeapplying,theMicrosprayneedstobeshakenstronglyfor10timesandfurtherneedstowaitfor5minutestoremoveanyairbubbles.Beforecoating, the steel mold needs to be cleaned by a metal cleaner. After applying thecoating,themoldrequirestobesoftandhardbaked.Thethermalcrosslinkingachievedthrough thehardbakingmakes thephotoresist robustduring the injectionmolding.TheblockdiagraminFigure3‐20showstheentirecoatingprocess.

Figure 3‐20: Photoresist process sequence (Ref. figure 5 of Chakrabarti et. al [13])

ThecoatingthicknesswasinspectedthroughDektakmeasurementsonaseparatesample,tobebetween10‐14 m,wherethestandarddeviationofthethicknessis~2.6%.ThepolymerPMMAisthebestcandidatetouseasthematerialfortheMAfortheapplicationasexplainedabove.Italsoholdstheadvantagesofthematerialpropertiesespecially the versatile polymeric material is well suited for imaging / non-imaging application. The mold was installed in a standard injection molding machine, in which the second steel block was aligned by dowel pins with 50 micron accuracy and placed on the top of the first block, and PMMA was injected through the center hole of the top block.Whenthesecondsteelblockisalignedpreciselyandplacedonthetopofthefirstblock,PMMAwasinjectedthroughthecenterholeofthetopblock.ThePMMAwasthenchannelizedthroughtheinlet(Figure3‐19).Afterwardsaheavypressureof2500barwasappliedtothe polymer melt. The process cycle of the microlens array is ~20 sec. Meltingtemperature of PMMAwas 210 °C and themold temperaturewas 70 °C. Injectionmoldedmicrolensarrayswereproducedinthefirstiterationusingtheuncoatedroughmold. After photoresist coating of both steelmolds (top and bottom one) a second


iterationofinjectionmoldingofmicrolensarrayswasperformed.About100microlensarrayswereproducedwithoutanysignofwearoffthephotoresist,andwithoutanymeasurabledifferencesbetweentheitem1and100.

The injection molding was performed by the Danish company JJ Kühn.Afterfabrication of the MA, the surface quality is characterized looking at pre and aftercoatingprocessresult.Thelensparametersofthefabricatedlensletsweremeasuredby analyzing the array under 3D confocal and interferometric microscopy. Theanalytical results for surface quality and the quantitative measurement on lensparametersarediscussedinChapter6undersection6.6.

3.3.6.3.1Fabricationprocessutility

Toachievethehighopticalsurfacequality,directfabricationoflensarraysbydiamondturning inpolymermaterials havebeendemonstrated [77]–[82]. The advantageofdirectdiamondturningprocessisthatitisaonestepprocess.However,duetoprocessdifficultyandexpensivetooling, the fabricationprocesscostcouldbe intheorderofUSD 100 – 1000 /mm2. Presently, electrostatic‐induced lithography [83] and inkjetprinting [84] have been explored for direct fabrication of polymer‐based shapecontrolledlenses.TheOpticalefficiencycanconsiderablyreduceduetothelargedeadspace between lenslets. This has been observed in fabricating themicrolens arrayswhich considered initially in this project. Through 3D optical printing process, themicrolensarraysareunsuitablefortheapplication.

Sincesuchdirectwritingprocessesarenotsuitable,awayof improving thesurfacequality of a CNC machined mold is explored and reprted in this thesis. For anyreplicationtechnique, thecriticalparameter is thesurfacequalityof themoldwhichconsiderably affects thequality of the finalmicrostructures and the applicability foropticalcomponents.TheCNCmachiningiseasyandinexpensive;however,thesurfacequalityisnotgoodenoughforopticalcomponents.InmanycasestheCNCmachinedmoldispolishedmanually,whichisbasicallytedious,expensiveandnotapplicableforlarge microlens arrays and also mass production. The procedure of spray‐coatedhydrogensilsesquioxanecoatingstoreducesurfaceroughnesswaspresentedrecently[85].Thisprocessneedsheatingthemoldpartupto400°Cinaninertgasatmosphere.Suchahigh temperaturemaybenotuseful for thesteelmoldsassomeof thesteelimpuritiesmaysegregateandcauseembrittlementofthesteel.

Thereforethethesisdescribesanovelprocess,fortoolfabricationutilizedforinjectionmolding of microlens structures. To make a first rough mold in steel, the processexploits standard CNCmilling. After that a surface treatment by spray coatingwithphotoresist is introduced toobtain anoptical surfacequality required formicrolensarrays.Bythisway,theexpensivetediousmanualpolishingorshortcomingofdirectdiamond turning process can overcome. The process also provides an easy andinexpensivemoldandhencedirects for cheap injectionmolding replicationprocess.Theaddedadvantageisthatthecoatingprocessonlyrequiresaheattreatmentofthemoldpartupto175°C.


3.3.7Fresnellens

Anoff‐the‐shelfpositivefocallengthFresnellensisusedatgrooves‐inpositionintheV8design.Thegrooves arewith constantwidth.TheFresnel lens ismadeby colorlessBorosilicateglasswith∅250mmandfocallengthof100mmwhichisabletocapturealllightoutputemergingfromtheoutputendofthegatewithamaximumobliqueangleof±40°andprovidetherequiredcollimationtoproducespotlightfortheV8lightengine(Figure3‐21).ThetransmissionlossfromtheFresnellessis~8%.FutureFresnellenswillbemadefromPMMAAcrylicwhichisdiscussedinChapter7.

Figure 3‐21: Fresnel lens made by Borosilicate glass

3.3.8ElectronicDriver

Presently,thedigitalmultiplexsignal(DMX)512‐A[86]universeisusedintheV8lightengine to control the electrical power input to LED channels using a DMXDecoder/Driver. DMX512‐A is the current standard and is maintained by ESTA(EntertainmentServicesandTechnologyAssociation).TheDMX512signalisasetof512separateintensitylevels(channels)thatareconstantlybeingupdated.OnesingleV8lightenginerequires4channels.However,itisaonewaycommunicationandasalightingdeskitcanhandleseveralsimilarV8sbyusingadistinctaddressfrom1‐512,eachlevelhaseither8bit(28=256)or16bit(216=65536)stepsdividedoverarangeof0 to 100percent. As an example ifwe take 12 similar V8s, then the total channelsrequiredwouldbe:12*4=48channels,wherethesecondV8lightenginewillstartfromthefifthchannelandsoon.Presently,newDMXRDMisavailableinthemarketwhichhasextendedtheDMXasitsupportslimitedbi‐directionalcommunication.

In theprototypeof theV8 lightengine, thereare206LEDswhichneed~666WelectricalpowerstodrivetheLEDsunderafullload.TheLEDdriverutilizingtheboostfromLinearTechnologycreatesDCcurrentwithtriangularcurrentrippleof10%withafrequencyof500KHz.,whichisthereasonthereisnoflickeringobservedandthatiswhy it isbetter thanswitchingON/OFF inpulsewidthmodulation (PWM).Step‐Up(Boost) regulator LED drivers generally produce the high voltages crucial to drivemultipleLEDsinseries,confirmingcurrentandthereforebrightnessmatchingbetweentheLEDs.TheadvantageofboostregulatorLEDdriversfromLinearTechnologyisthattheyofferthehighestefficiency,lowestnoise,andthesmallestfootprints.OtherfeaturesincludeintegratedSchottkydiodes,whichprovideaccurateLEDcurrentmatchingandmultipleoutputcapability.Tofacilitatethedimmingeffect,theV8lightengineusesapersonalizedfileofvirtualdimming,whichissquaredimmingcalculatedoutoflineardimmingasacompensator.Asquarelawdimmingappliesamultiplederivedfromthe


square root of the control level (with full output equal to 1.00) to increase voltageresponseatlowcontrollevelstocompensatefortheinfraredloss.

3.3.9Heatsink

Aheatsinkisadevicethatisattachedtoalightingdevicetokeepitfromoverheatingbyabsorbingitsheatanddissipatingintotheair.Generallytoavoidthefailurefromthecomponents,anydevicetemperatureshouldnotruninexcessof50‐55°Cwhileundera full load. Especially for luminaires made by an LED module is sensitive to thetemperature. Thus,water cooling is used in the prototype of theV8 light engine totransportheat fromLEDmodule to theouter surfaceof theV8 light engine so thatexcess heat can be dissipated and also maintain the water temperature at 10 °C.Althoughwatercoolingisveryeffective,itisinconvenientforhandling.Thus,infutureaproposalofusingaheatpipeaswellasanelectricfanisdiscussedinChapter7.Fortheaterapplication,noisefromfancanbeanissue.

3.4OpticalmodellinginZemax

The purpose of doing optical simulations of the optical design is to provide designflexibilityandoptimizationoflightthroughput.Italsohelpstofinalizethepositionandsizeoftheopticalcomponents.OpticalsimulationoftheV8lightenginedesignhasbeendoneusingZemaxopticalsoftware.AstheV8lightengineisacomplicatedsystemtosimulate,initiallyasanexample;thegeometricalsimulationofasimpleopticalsystembased on the color mixing of three LEDs is considered, which is documented byChakrabarti et. al in scientific article[16]. The article describes the investigation andoptimizationprocedurefortheangulardistributionoftheluminousintensityaswellasthe color distribution, considering a 3D geometrical ray–tracing model of the LEDopticalsystemusingtheopticalmodellingsoftwareZemax.Theresultoftheangulardistribution isanalyzed for luminous intensityandhomogeneityofchromaticityandCCT. The three LEDs are modelled by using ray source files provided by themanufacturerswith 500,000 rays in each, and themeasured SPD for the LEDs areincluded as spectral weights for the rays. The reflectance of the Ba2SO4 coating ismodelledbyusingtheLambertianscatteringmodel.ThethreecoloredLEDsystemalsousedadiffuserplate(5mmx5mmx1mm).Thediffuserhelpstocreatehomogeneouscolormixingfromthetri‐colorLEDsystemasexplainedbyChakrabartiet.al[16].Thebidirectional scatter distribution function (BSDF) of the diffuser has beenexperimentallymeasuredandthescattermodeldataisincludedinthemodel.AppendixB shows the technique of the BSDF calculation. A virtual polar detector is used toanalyzethespatiallightdistribution.Thisisdonebysummingupthenumberofrayshittingtheangularspaceddetectorelements.Fromthisanalysis,theluminousintensitydistributionandthetristimulusvaluesofthelightforthedifferentangulardirectionscanbefound.ByadjustingtheradiometricpowerofeachLEDrayemitter,themodeltothespecificoptimizedsettingsisabletotunablethroughthefoundsettingsofspectralmapping.Thechromaticitycoordinatesarecalculatedfromtheangularspaceddataforthetristimulusvaluesofthepolardetector,andtheyareusedtocalculatetheCCTofthelightindirectionscorrespondingtotheΘandφangles,wheretheZ‐axisisconsideredastheopticalaxis,correspondingtoΘ=0°,wheretheoriginisatthecenteroftheopticalmodel(Figure3‐22(a)).TheXaxisisgivenbyφ=0°andΘ=‐90°,respectively.


TosimulatetheV8lightenginedesign,asimilarmethodisadopted.TorepresenttheparabolicreflectorandtheTIRlenses,correspondingCADfilesareloadedas .stlfiles.ThematerialoftheTIRlensisdefinedasthePMMAwhilethereflectorisdefinedasamirrorwherereflectionlossesareconsidered.Theotheropticalcomponents(e.g.,PCB,gate,MA) are defined as Zemax objects. Initially, the design consisted of 7 rings onwhich210LEDswereplacedas seen inFigure3‐22.The ray files for theLEDsareloadedintotheZemaxfilefromtherespectivesourcefileswhichisavailablefromthecorrespondingsupplier’swebsites.EachcoloredLEDwiththeirrespectiveringpositionwassimulatedseparatelythroughtheTIRlenstoinvestigatetheradiationlossaswellasthetotallightoutputfromanindividualscenarioofcolortuningbetween3000Kand6000K.However,toeasethesimulationprocessfortheentireV8lightengine,insteadof simulating 210 LEDs, onewhite LEDwith an individual TIR lens at each ring isconsidered.Thismethodsavescomputationaltimeaswellasreducesthecomplexity.Therayfilewith500,000raysforwhiteLEDareloadedintotheZemaxfilefromtherespectivesourcefile.ThetotallightoutputwasfinallyestimatedbyconsideringtheothercoloredLEDswiththeirrespectivepositionintheringandtheircorrespondinglightoutput.Thetotal lightoutputfromthesimulationwasinvestigatedbyplacingavirtualdetectorasascreenata5mdistancefromtheopticalgatewheretheexitconewas±40°.Figure3‐23(a)representsthesimulatedbeamspotfromtheV8lightenginewhilethelightisprojectedata5mdistance.Thefigureindicatesthatthelightoutputhasahomogeneousintensityinthemiddleexceptforoneredspotandthenitgraduallyfades out towards the perimeter. As mentioned earlier, the simulation is done byconsidering few white LEDs inside the optical design; the light output cannot beuniformlydistributedand shows thehighest intensitywherebeams from thewhiteLEDsinfluencethemost.Thebeamspothashexagonalshape,whichisbecauseofthehexagonalshapeoftheMA.TheissueisdiscussedinChapter7undersection7.2.1andtheshapeisexperimentallyvalidated(Figure6‐25).

Figure 3‐22: (a) Top view of TIR lens position on the PCB in simulation software (b) Color mixing light coming out from the gate (Ref. figure 3 of Chakrabarti et. al [14]


When the total light outputwas estimated by the simulated system, the spectralinformationsuchastheCCTofthelightwasunabletobeinvestigated.Thus,thereisnoscopeoftuningtheCCTrangingfrom3000Kto6000Kbythesimulation.However,4350KwhitelightwassimulatedbymixingoffewcolorLEDsinsidethelightengineandthecolorhomogeneityisobservedbycheckingtheviewingangle(Θ=±50°).ThetwoplanesXZandYZare consideredwhereXZ represents=0°andYZ represents=90°(Figure3‐23(b)).ThevariationinCCTismaximum~4%withinΘ=±50°.

Figure 3‐23: (a) Simulated light output from the V8 light engine when the virtual detector is placed at 5 m from the gate (b) Simulated white light at 4350 K from the V8 light engine

Thebeamhomogeneityof thewhite lightoutput isalsocheckedbysimulatinganoptical design using Unified Optical Design software (Wyrowski VirtualLab Fusion)fromLightTransUG,wherethecollimatedbeamfromRed‐Green‐Blue(RGB)LEDprintcircuitboards(comingfromdifferentangles)aredirectedtowardstheMA[87],[88].Onevirtualdetectorisplacedat5mfromtheMAsothattheprojectlightcreatesthebeamspot.Figure3‐24representsthebeamspotwherethebeamsizealongtheXandtheYdirectionsaresame(~±2.5m).Thespotlookshomogeneousinthemiddleandshows thechromaticaberrationat theperipherydue to thedispersionof light.Thisissueisdiscussedin7.2.1inChapter7.ThebeamspotshapeishexagonalduetotheshapeoftheMA.


Figure 3‐24:, White beam spot created by color mixing of RGB LEDs using the MA projected at 5 m distance (Ref. [87], [88])

Theparametersofthereflector,theTIRlensandtheMAhavebeenoptimizedwithrespecttoobtainingahighlightthroughputfromtheopticalgate.Thesection3.3hasalreadydiscussedthebenefitsoftheoptimization.

3.5OpticalcomponentparametersforV8lightenginedesign

ThedimensionsoftheindividualopticalcomponentsalongwiththeLEDnumbersusedintheopticaldesignoftheprototypeoftheV8lightenginearelistedinTable3‐2.Toobtainhigherlightoutputfromtheopticalsystem,Table1alsoindicatestheoptimizeddimensionsofthecorrespondingopticalcomponents.


Table 3-2: The dimensions of the optical components in the V8 lighting design (Ref. [14])

TheluminousefficacyofthewhiteLEDsateachringpositionissimulatedaccordingtothemodifieddesign1(Table3‐2).Figure3‐25directsthecumulativeluminousfluxateachringafteraddingthetotallightoutputfromtheLEDsateachringposition.Theindividual luminous fluxdecreasesgraduallyas theringnumber increases(far fromgate).However, the cumulative luminous flux increases initiallywith increasing ringnumber.This isbecause the ringsize increasesgraduallygiving roomtoholdmoreLEDsthanthepreviousone.Afterringnumber7,thecumulativeluminousfluxagaindecreasesas thehighTIR lens losscannotcompensate theeffectof increase inLEDnumber.Thefigure indicates thatthetotal luminousflux fromtheV8 lightengine is~25900lmbeforeFresnellens.Theaveragedeviationinopticalefficiencyofthelightenginebetweenthesimulationandexperimentalresultsis~9.7%,whichisdiscussedinsection6.7.1inChapter6.

Figure 3‐25: Cumulative luminous flux and individual luminous flux at each ring respectively (Ref. Figure 7 of Chakrabarti et. al [14])

Optical

s ystem Numbers Parabolicreflector[mm] PCB

size[mm]

Lenssize[mm]

Gatesize[mm]

Fresnellenssize[mm]

MAwidth[mm]LED Ring Diameter Curvature Focal

lengthOriginal

Prototype 210 7 240 ‐380 170 206 11.7±0.2 50 250 1.2

Modified

design1 210 7 240 ‐330 165 206 12.7±0.2 50 180 1.45

Modified

design2 288 9 340 ‐400 180 270 12.7±0.2 50 180 1.45


3.6Summary

This chapter introduces the V8 light engine by defining the components and theirrespective functionalities to produce a desiredwhite light output from the V8 lightengine,whichhasastageapplication.ThegeometricaldesignoftheV8lightengineisopticallysimulatedusingZemaxandusedopticalcomponentsareoptimizedinordertoyield>20,000lmlightoutputfromthelightengine,whichisoneofthecriticalgoalsoftheproject.TheMAusedinsidetheV8lightenginedesignasabeamhomogenizerisdesignedforthisapplicationtomaximizetheopticalefficiency.Itisfabricatedinhouseand characterized for the lenslet parameters, surface quality, light scattering, andacceptance angle. The fabrication process of the MA overcomes the expensive andtediousmanualpolishingordirectdiamondturning.Theprocessprovidesaneasyandinexpensivemoldandhenceinexpensiveinjectionmoldingreplicationprocessbyfar.Thefinaldimensionsoftheusedcomponentsaredocumentedinatable.ThesimulatedlightoutputindicatesthattheV8lightengineiscapableofproducingaluminousfluxof>20,000 lm.The followingchapterwillbeabout thecolormixing techniqueandtheapplication in theV8 light engine design required in order to get an effective colormixingwithgoodlightquality.

Chapter 4 61 Modelling of color mixing

Modellingofcolormixing

‘Everysooftenchangeyourpalette.Introducenewcolorsanddiscardothers.Youwillgainknowledgeofcolormixingandyourworkwillhaveaddedvariety.’–KennethDenton

Chapter4 :Modellingofcolormixing

4.1Introduction

ThischapterwillprovidethebasicinsightofthecolormixingtechniqueadoptedinthemodellingofcolormixingfortheV8lightengine.SincetheV8lightenginecollimatesandcombinesthelightbeamsfromanumberofcoloredhighpowerLEDarraysintoasinglefocusablebeam,thesystemneedstomixtheindividualcoloredLEDsincludingwhiteLEDinadeptwaysothattheresultantofthecolormixingwouldcoveralargecolorgamutareaandatthesametimecoverarangeofwhitelightcorrespondingtodifferent color temperatureswhich shouldbe characterizedbyhigh color renderingproperties.Other critical requirements for theV8 light engine are; (1) the resultantwhite light needs to be free from color shadows, (2) focusable spot light output.However,initiallythecolormixingoflaserlighthasbeeninvestigated,asspotlightcaneasilybemadefromlaserlightduetothecollimatednature[12][89].ThelaserhasamuchnarrowerspectrallinewidththanthebroadspectrumoflightemittedfromLEDs.Thusduring colormixing time, LED spectrumshave capacity to overlap eachotherwhile laser spectrums rather stay discrete after mixing. Therefore, the white lightproducedbythelasercouldprovidediscomfort inavisualperceptionofcolorwhileseeingthecolorobjectswhichdoesnotattractthepeoplepreference.Itcouldaffectthecolorrenderingoftheoveralllightoutput.Secondly,LEDsareacosteffectivesolutioncomparedtolaser.Foranexample,agreenlaserof1Wprovides~600lm,whereasabluelaserof1Wprovides~20lm,andaredlaserof1Wprovides219lmluminousflux[90].Thus, toobtaina20,000 lm lightoutput fromtheV8 lightengine,weneedmorethan20lasers[91]whicharecostlycomparedtothepriceofanLED.Hence,thecolormixingfromtheLEDilluminationswereusedfordesigningtheV8lightengine.

4.1.1Colormixingtechnique

Whitelightcanbegeneratedbypropermixingwithred,greenandblueLEDs[8],[92],[93].ButnowadayspeopleprefertouseanadditivecolormixingmethodwheremorethanthreeLEDsareusedtogeneratethedesiredwhitelight[94].ThewhitelightcanalsobeproducedbythedifferentcoloredLEDswithwhiteLEDsitself.Theadvantages


ofusingthisadditivecolormixingmethodaretoachieveashighacolorrenderingaspossible,andas lowachromaticdistancetothePlanckian locus. Inthiscontext, it isassumedthatthepowerdistributionoftheLEDsismuchnarrowerthanthevariationsinanyofthethreecolormatchingfunctions(x(λ),y(λ),andz(λ)).Thechromaticitycoordinatesof themulti‐colored light (e.g. threecomponentsof light)are ina linearcombination of the individual chromaticity coordinatesweighted by the Li (i=1,2,3)factorswherePi(i=1,2,3) is thecorrespondingopticalpoweremittedbythesourcesand(x(λi),y(λi),andz(λi))arethecolormatchingfunctionscorrespondingtothepeakwavelengthintheequations(36)‐(38)[95].

̅ ̅ (36) ̅ ̅ (37) ̅ ̅ (38)

Using the abbreviations, the chromaticity coordinates of the mixed light can becalculatedfromthetristimulusvaluestoyield(x,y)coordinatesinequations(39)‐(40).

(39)

(40)

Figure4‐1(a)showstheexampleofthreecolormixingLEDsproducingresultantSPDof white light (2200 K) on the Plankian locus and corresponding CIE 1931 (x, y)chromaticitydiagramatFigure4‐1(b).

Figure 4‐1: (a) The resultant spectral power distribution (dashed line), and spectral power distributions of the individual LEDs (cyan, white and deep red) normalized to the peak of the resultant SPD (b) CIE 1931 chromaticity diagram for the tri‐color LED optical system where solid blue line of triangle shows the full gamut of the LED system, white (2152 K)‐cyan (502 nm) – deep red (666 nm)

(Ref. Chakrabarti et. al [16])

4.2ColormixingportfolioforV8lightengine

TheV8lightengineusesthemethodofadditivecolormixingoffourcoloredLEDsandawarmwhiteLEDwiththeirspecificSPDs,whereresultant‘SPD’representsthelinearsummationofSPD(λ)perunitwavelengthemittedfortheentirerangeofwavelengths


from360nmto830nm.ItisbelievedthattheaddingofpoweroflightemittedperunitwavelengthforeachLEDwillproduceamoreaccurateresultantSPDthanaddingonlySPD (λP) at λP of each LED, where λP is the peakwavelength of each colored LED.Moreover,sincethewhiteLEDisusedasthemixingcolorcomponent, forthewhiteLED, therewillbeanuncertaintyof finding theprecisepeakwavelengthwithin thebroadspectrum.Thus,equations(36)‐(38)cannotbeusedtosimulatetheresultanttristimulus functions of the produced white light output. To find out the resultanttristimulus functionof theproducedwhite light, the following equationsneed tobederived.LetusassumethatS1(λ),S2(λ),…S5(λ)[96]arethespectralpowerdistributionsper unit wavelength for the five LED components. The resultant spectral powerdistribution‘S(λ)’perunitwavelengthcanthenbecalculatedbyequation(41).

S λ S λ S λ S λ S λ S λ (41)

Thetristimulusvaluesfromequation(41)aregivenbytheequations(9)‐(11),wherex λ , y λ , z λ arethecolormatchingfunctionsatcorrespondingwavelength.Fromtristimulsvalue,CIE1931(x,y)chromaticitycoordinatescanbecalculatedbyusing(12)‐(14).Finally,fromCIE1976(u’,v’)[97]andCIE1960(u,v)uniformchromaticitycoordinatescanbederivedfromtheequations(17)‐(18).

4.2.1Spectralmodelling

Torealizeandanalyzetheabovementionedmethod,thespectralmixingofasimpletri‐colorLEDopticalsystemhasbeenusedasatoolofLEDspectralmodellingwhichisdocumentedinthearticleofchakrabartiet.al[16].Insteadofatri‐colorsystem,theV8lightengineisdesignedforfivedifferentcolorLEDs.However,thebasicmethodologywasimplementedinasamemanner.Thebasicmethodologyforthetri‐coloropticalsystemiselaboratedinthefollowingparagraph.

Spectralmodellingofthelightoutputhasbeenperformedbyextensivemappingofthepossible resultingSPDof theLEDoptical system.Thegoalof thiswas toobtainwhitelightwithalowDuvandagoodcolorrenderingatlowCCTwithcommerciallyavailable LEDs. For color rendering, the thesis considers mainly the general colorrenderingindex,Ra,andthewidelyusedsupplementaryspecificCRIforredobjects,R9,aswellasothermetrics,R10andCQS.SimilarlytoCorelletal[98],itisassumedthatthiscouldbeobtainedinthetri‐colorsystembyusingacyanLEDtofillthegapbetweentheblueandthegreenspectralregionsintheSPDandbyusingaredLEDtoadjusttheredspectralcontent.Thisassumptionwasverifiedbythesimulationmethodshowninthissection.Thereisatradeoffbetweenthesimplicityofthetri‐colorsystemandbroaderparameterrangeofcolorqualitiesofwhitelightfromthetri‐colorLEDsystemoverasystemwithmore than three colored LEDs. Thismappingmethod is based on thespectralcharacterizationoftheindividualLEDswhichismodifiedfromreference[99]and describedmore thoroughly in reference [100]. The results are visualized as anoverview of the colorimetric possibilities and make the identification of individualsettings for the LEDs with suitable color characteristics easily accessible. Thissimulationisnotinitselfoptimizedtowardsasingleparameter;however,itiscarriedoutinanexhaustivenumericalmappingofallthepossiblecolorcharacteristicsthatthesystemcanproduce.Thisisdonebysimulatingaconstantcurrentsetting(700mA)forthewarmwhiteLEDandthencalculatingthecolorimetricresultsfromdifferentscaling


combinations[Equations(42)and(43)]oftheredandcyanLEDsusingadditivecolormixing.

, , (42)

, , (43)

whereΦredandΦcyanareradiantflux,αredandαcyanarescalefactors,λisthewavelengthand , spectralradiantfluxat300mA, , spectralradiantfluxat90mAfromLEDs for equationEquations (42) and (43), respectively. The resultant SPD isobtainedbyaddingcontributionsfromthethreeLEDs(Equation(44).

, , (44)

whereΦ(λ)istheresultantspectralradiantfluxofthethreecombinedLEDsandΦwhite(λ)thespectralradiantfluxforthewhiteLEDat700mA.Theresultingmapsofthegeneral CRI, and specific CRIR9, CCT, andDuv can be seen in Figure 4‐2.Using thismethod,themappingofthecharacteristicstoatwo–dimensionalproblemreduces,thatcan be calculated relatively fast for the whole parameter space. Finding optimalsolutions for color characteristics can then be done by searching for the desiredcharacteristicswithin theavailablegamutareaof theLEDoptical system.ThemainfocuswasonthesolutionswherethechromaticityisclosesttothePlanckianlocus,i.e.thelowestpossibleDuvwithhighcolorrendering.Figure4‐2showstheresultofthecalculationwherethehorizontalandverticalaxesofthefoursubplotsareαredandαcyanforredandcyanrespectively,whereasthewhiteLEDiskeptataconstantscalingvalue.TheoptimizationoftheLEDsettingsforthedesiredCCTsisdiscussedinthefollowingparagraph.


Figure 4‐2: Spectral mapping of the possible color characteristics of the optical system. Mapping of the general CRI Ra(a), special CRI R9 (b), CCT (c) and Duv (d) for varying content of red and cyan LED light in the LED optical system. Experimental validation results are represented by ‘s’ while ‘x’ shows the locations of simulations with corresponding CCT and Duv values from the exhaustive numerical

simulation. The solid thick black line represents parameter sets which have zero Duv. From Figure 4‐3 (c), the black line corresponds to a narrow CCT range from 2094 K to 2106 K (Ref. Figure 4 Chakrabarti

et. al [16])

ThespectralmodellingproducesthemappingofthecolorimetricparametersCRIRa,CRI R9, CCT andDuv for relevant settings of the αcyan (20% to 100%), αred (40% to100%),andconstantwarmwhiteLEDsasshowninFigure4‐2.Theoptimalsettingsofallabovementionedcolorimetricparametersarefound.CRIRaandR9aremappedwiththe spectral contentof adeep redLEDversus the spectral contentof the cyanLEDaccordingtoFigure4‐2(a)and(b),respectively.ThespecificCRIforgoldenobjectsasyellowobjects,R10, isequally important.However,aclosecorrelation in thespectralmappingbetweentheR10andRawasfoundandhenceonlygraphsforRaandR9areshown.TheCQSshowsasimilarcorrelationwithRa.Theinterestingobservationisthatthe optimum CRI regions,where both Ra > 95 and R9 > 90 appear as lineswith anegative slope of αred andwhile αcyan is limited up to 40% for Ra and 80% for R9,respectively.ThemaximumachievableRaandR9byspectralmappingsare98.6and99.9,respectively.

Asseen inFigure4‐2(c)and4d,bothCCTandDuv levels increasewithanegativegradientacrossthemap.Figure4‐2(c)presentsCCTmappingwheretheintendedCCTrangesfrom2000Kto2400Kcouldbeachievedbylinearlyincreasingscalingfactorsof


both cyan and deep red. In order to reach 2000 K, αred should be 80% and thecorrespondingαcyan 20%.At 2400K, both αcyan andαred is at 100%.Hence, for thatparticular CCT settingRa andR9 values cannot be greater than90. Thepresence ofminimumDuvandoptimumCRIareonlyfoundatonespecificCCT,wheretheoptimumlinesintersectwitheachother.Forthisparticulartri‐colorsystem,theintersectionhasaCCTvalueof2090KandCRIisat98.6,respectively.However,thiscancompromiseCRIandDuv in order to achieve a larger tuning range of CCTs. In this case, if the CCTincreases, the cyan level increases andDuvbecomeshigher (Figure4‐2(d)) and thataffects the quality of thewhite light output. Sowith the present tricolor LED colormixingsystemtherewillbealimittoachievingthedesiredvalues(lowDuvandhighCRI) fortheabovementionedCCTrange.This limitation isoverruled intheV8 lightenginewherethesystemisdesignedforfivedifferentcolorscoveringalargespectralrange.However,recentfindings[101],[102]showthathumanpreferenceforlowCCTfavorsanegativeDuvandthattheoptimummightnotbeatzeroDuv.

4.2.2Colorchoiceandcolorgamut

ThetechnicaldatasheetvaluesfortheusedLEDsintheV8lightengineisdocumentedin Table 4‐1. The corresponding experimentallymeasured SPDs at 350mA drivingcurrent for the five LEDs are represented by Figure 4‐4(a). The correspondingmeasuredexperimentalvaluesaredocumentedinChapter6.AsseeninFigure4‐4(b);theproducedwidecolorgamutbytheabovementionedfiveLEDsareindicatedin1976(u ,́v´)uniformchromaticityco‐ordinates.ThewidegamutproducedbythefiveLEDsiscapableofprovidingthetunablewhiteCCTrangefrom3000Kto6000K.


Table 4‐1: LED portfolio according to technical data sheet used in the V8 light engine [103]–[105]

LEDtype LEDcolors

Partnumber Datasheetin2014Current[mA]@85°C

Luminousflux[lm]

Peakwavelengthλp[nm]

LuxeonZ Royalblue LXZ1‐PR01‐0600 700 823mW 447.5LuxeonZ Blue LXZ1‐PB‐01‐0048 500 40 470CREEXPE Green XPEGRN‐L1‐0000‐00E03 350 105 528CREEXPE Red XPERED‐L1‐R250‐00801 350 54 625CREEXPE2 White XPEBWT‐H1‐0000‐00BE7CT‐ND 350 97 3000K,82CRI

(a)

(b)

Figure 4‐4 (a) SPDs for five used LEDS at 350mA driving current (b) Color gamut produced by five color LEDSs in CIE 1976 (u’, v’) uniform chromaticity diagram

4.2.3ColorqualityinV8lightengine

Byvarying theproportionalpoweroutputgeneratedby fivecolorLEDs,all thebestpossibleSPDsforwhitelightrangingfrom3000Kto6000KfortheV8lightenginearesimulatedusingtheoptimizationcriterialistedinTable4‐2.Thelightqualitiesofthose


simulatedwhitelightarealsovalidatedbyexperimentalfindings.Figure4‐5showstheexperimentallygeneratedthreeCCTsandTable4‐3representsthecorrespondingcolorqualities.ItindicatesthecolorqualityoflightbyevaluatinginCRI[49]andTLCI12[53]scalewhicharedefinedin1.5.2.9and1.5.2.11sectionsinChapter1.Accordingtotheapplication,themetricscoreforCRIandTLCI12aregoodforthreeCCTsettings.ToevaluatethescoreforTLCI12,theTLCI2012softwareisusedwhichispublishedbytheEBUtechnicalcommittee[53].Theobtainedluminousfluxwaslow(<10,000lm)forthreeCCTsettingsbecauseofthelossesencounteredinsidethesystemoftheprototypewhich is modified afterwards by optimizing optical components, discussed in thesectionof3.3inChapter3andChapter6,section6.7.1.

Table 4‐2: Optimization criteria for spectral simulation

Luminousflux CCT DUV CRIHighestachievableluminousfluxattargetCCT TargetCCT+/‐20K <5*10‐3 >85

Figure 4‐5: Experimental results of white light at three CCTs

Table 4‐3: Light Output from V8 Light Engine at Three CCT Settings (Ref. Table 3 of Chakrabarti et. al [14])

TargetCCT

Proportionofusedcolorcomponents ProducedlightqualityRoyalBlue

Blue Green Red White CCT CRI(Ra)

CRI(R9)

DUV(10‐4)

TLCI12

Luminousflux[lm]

3000 1 0.5 17.3 100 100 3018 89.2 41.9 ‐6.6 75 79014500 14.2 45.9 77.5 79.3 100 4514 91.1 89.1 0.88 91 92955800 22.6 90.3 100 3.5 100 5814 88.5 76.2 1.5 87 9315

TheproducedCCTsareplottedat thePlanckian locus inCIE1960 (u, v)uniformchromaticitydiagraminFigure4‐6.ThereareeightquadrangleBINsseeninFigure4‐6byredcolorwhichincludewarm(2700K,3000K,3500K),normal(4000K,4500K),andcold(5000K,5700K,6000K)rangeofCCTasaquadrangleforSolidStateLighting(SSL)products[106], [47]. These eight BINs are approved by the ENERGY STAR and astandardisdocumentedasANSIC78.377(AmericanNationalStandardInstitute).Thisstandardspecifiestherangeofchromaticitiesrecommendedforgenerallightingwhilethechromaticityoflightisrepresentedbychromaticitycoordinates(x,yandu’,v’).Thestandard also defines the 7‐steps from Planckain locus. According to the plot, theexperimentalsolutionfor4500KislocatedonthePlanckianlocusandthesolutionsfor3000Kand5800KarehavinganegativeandpositiveDuvvaluerespectively.Sinceall


Duvvaluesare<±10‐3, they indicatewhite lights.However, the lightwhich ishavingnegativeDuvvaluehaspinkishtintwhereastheDuvwithpositivevaluehasgreenishtint.

Figure 4‐6: Three experimentally found CCT points on the CIE 1960 (u, v) uniform chromaticity diagram (Ref. Figure 5 of Chakrabarti et. al [14])

4.3Discussionontheutilityofthetechnique

Theabilitytocreateagreatvarietyofcolorsisanimportantqualityfordisplays.Thecolor gamut represents the entire range of colors that canbe created froma set ofprimary sources. Color gamut is a polygon positioned within the perimeter of thechromaticity diagram. All colors created by additive mixtures of the vertex points(primarycolors)ofagamutarenecessarilylocatedinsidethegamut.Inourcase,itisdesirablethatthecolorgamutprovidedbythefivelightsourcesisaslargeaspossibletocreatedisplayswhichareabletoshowbrilliantsaturatedcolors.Thecolorconsistencywithin360nmto800nmcanobtainintheresultantSPDofproducedwhitelightbyusing five color componentswhich can be seen in Figure 4‐4(a). The color gamutcreatedbythefivecolorsLEDsinFigure4‐4(b)ischosentoprovidealargecolorgamutareaincludinghighlysaturatedcolorsandcoversarangeofwhitelightcorrespondingtodifferentcolortemperaturesandcharacterizedbyhighcolorrenderingproperties.TheblueandtheredLEDsareusedwithotherthreeLEDs(royalblue,green,andwhiteLEDs)tocovertheentireblackbodyradiatedcolortemperaturefrom3000Kto6000K.AwarmwhiteLEDisaddedinthegamuttoprovidebaseCRI(~82)andthenCRIcouldimprovebyaddingmorered.Thoughmoreredcannotbeusedasitwilldecreasetheefficiencyofthesystem,therewillbestillhavingflexibilitytooptimizetheCRI,aswellastheefficiencybyotherremainingcolors(namely,royalblue,blue,andgreen)andthatistheuniqueadvantageofthespectralmixingproceduresreportedinthisthesis.Table4‐3indicatesthehighcolorqualitiesoftheproducedwhitelightatthreedifferentCCTs.GoodspecificCRIRa(>85)ateachCCTwillhelptorendercoloreffectively.Goodspecific


CRIR9 (>40) implies important red content andhelps to light rawmeator interiorlighting decorations in display lighting. The score of TLCI 12 > 75 signifies that theproducedwhitelightcanbeusedinastudiowithouthavinganycorrectionsdonebycamera.Theluminousfluxinawarmcolortemperatureregimewouldnotbeashighasina cold color temperature regimebecause inwarmcolor temperature regime, theproportionofusedredLEDswouldbehigherthanotherLEDs.However,theluminousfluxwillbeimprovedgraduallyasitmovestowardsacoldercolortemperatureusingmoreproportionsofgreen,blue,androyalblueLEDsbycompromisingwithhavingahigherDuv value.However, CRIwill improvemore than in awarmCCT regime. TooptimizetheCCTpointagainwithhighlightqualities,thesimulatedresultantSPDusetheoptimizationcriteriadescribedinTable4‐2.

4.4Summary

This chapter provides the methodology for the color mixing technique adopted indesigningtheV8lightengine.Thespectralmodellingisbasedontheoptimizationofcolor qualities e.g., high CRI indices (Ra>85, R9 >40) and low Duv (< ±10‐3) at thecorrespondingdesiredCCT.ToprovideanexamplefortheexperimentalvalidationofthesimulatedSPDs,themeasuredlightqualityandluminousfluxforthreetargetCCTsettings(3000K,4500Kand5800K)withtheircorrespondingSPDsfromtheV8lightengine are documented. The created wide color gamut in 1976 (u ,́v´) uniformchromaticity diagram by five differentmulti‐colored LEDs is projected to show thecapabilityoftheV8lightenginetotunetheCCTrangingfrom3000Kto6000K.ThefollowingchapterwillexplainthecolorcontrolmechanismoftheV8lightengine.

Chapter 5 71 Color consistency of the V8

ColorconsistencyoftheV8lightengine

‘Consistencyisthelastrefugeoftheunimaginative.’–OscarWilde

Chapter5 :ColorconsistencyofV8lightengine

5.1Introduction

UpontillnowthesubjecthasbeenontheadvantagesoftheV8lightengineasastagelighting luminaire over the traditional lighting system, the light engine could havedeficienciesintermsofthelightingsystem’sstabilityinluminousflux,andcolor(e.g,CCT,Duv,CRI,etc.)duringtheoperationaltimeifthesystemdoesnotencompassanycontrolsystem.SincetheV8lightenginecomprisesmulti‐coloredLEDs,therearealsooften noticeable luminous flux and color variations in light output for similar V8systems.Thelongtermandshorttermoperationalstabilityofthelightoutputdependsonmanyinfluencingfactorsi.e.LEDbinning(wavelengthandflux)andaging,ambienttemperature, LED junction temperature, driver electronic such as digital control ofcurrent,systemruntime,etc.[99].Thus, the luminous fluxandcolorvariations fromlamptolampandstabilityoverbothshortandlongtimeoperationareverycriticalandshouldbecontrolledtoremainwithincertainlimits.Toprovidecontrolledlightoutputwithout compromising the efficiency and light quality, several methods wereimplementedbasedontemperaturefeedback,opticalfeedbacksuchasfluxandcolorfeedback, or a combination of temperature and optical feedback[107]–[112]. Toprovideoptical feedback, inmanycases,simplecolorsensorsarepreferablyusedtoexaminetheinstabilityandtoprovideinformationofcolorvaluestoafeedbacksystem.Thesecouldbe:CCT,Duv,andluminousflux[17],[18],[113].Therefore,thefeedbackcontrol system could work effectively if known uncertainties of the influencingparameterscouldbeinvestigatedandusedasinputintothefeedbackloop[20],[114].

5.2MonteCarlosimulation

TheuncertaintiesoftheabovementionedinfluencingparametersareexperimentallystudiedandevaluatedbyuseoftheMonteCarlosimulationasexplainedinthearticlesby Chakrabarti et. al [23]–[25]. Since theMonte Carlomethod avoids the analyticalpropagationofuncertaintybycombinationofsimulatedandexperimentalfindingsofthe uncertainties of individual input parameters of colored LEDs, there is norequirement for deriving the partial derivatives of the complicated equations.Furthermore,theuncertaintycanbeexaminedforanycolorquantitiesinaneasyway.ThusmanyapplicationscouldusethisMonteCarloSimulationtoestimateandanalyze


measurementuncertaintybasedonthepropagationofdistributions[115]–[117].TheMonte Carlo method can provide and analyze the uncertainties of chromaticitycoordinatesoverthetunablerange,andcanrelatethemtothecolorimetricvaluese.g.CCTandDuv,andCRIsforthemulticoloredV8LEDengine.TheMonteCarloanalysispresentedhereisbasedonanexperimental investigationofthebasicsensitivitiestoexternalchangeswhichwascarriedoutontheV8LEDlightengine.Itwasalsoaimedtoanalyzetheachievablestabilitybyimplementingacolorsensor.Therefore,theanalysisincorporatesmanyofthefactorsinfluencingthecoloruncertaintyforsimilarsystem,i.e.LED binning and aging, ambient temperature, driver electronic as digital currentcontrol,andcolorsensorspectralresponse.

Amethod thatmakes use of aMonte Carlo simulation approach to estimate theresultingpropagationofuncertaintiesofanycolorquantityofthemulti‐coloredV8lightenginewillbedescribedinthissection.Thefirststepofthemethodistobuildamodelof the LED system. The second step is to obtain experimentally or otherwise theuncertaintyof thecontrollableparametersaswellastheuncertaintyof the inherentpropertiesofeachLED(peakwavelengthbin,wavelengthshiftwithtemperature,fluxbin, expectedaging,etc.).The thirdstep is toapply theMonteCarloMethodon thismodel,calculatingasetofresultingspectralpowerdistributions(SPDs)fromthemodelwithrandomvaluestakenfromtheprobabilitydistributionofeach inputparameterandderivingthecolorimetricresults.ByimplementingafeedbackandcontrolloopsinthemodeloftheV8lightengine,theuncertaintyinthecolorimetricvaluesisdecreased.Theassociateduncertaintiesrelatedtocontrolsensorsandelectroniccircuitswillhavetobeincludedwhenthecontrolfeedbackloopisapplied.

5.2.1LEDmodelandpre‐calibratedlookuptable

ThemodeloftheLEDsystemissimulatingthespectralradiantfluxforeachcoloredLEDused in theV8 lightengineas input tomix thespectral contentofaLEDcolormixingmodelwhichisreportedinthearticlesbyChakrabartiet.al[23]–[25].Itisthemodel using four colored LEDs to supplement and change the SPD from thewarmwhiteLEDat3000K(Table4‐1).TheoutputfromindividualLEDsisaddedtoproducetheresultantSPD.TheresultantSPDprovideswhitelightoutputfromthelightenginewithoptionoflightintensitydimming,achievingaminimumlowerdimmingto13%,andvaryingtheCCTrangingfrom3000Kto6000K.Thereisapre‐calibratedlookuptable which is associatedwith the white tunablemulticolored V8 light engine. ThelookuptableconsistsoftargetCCTs(3000Kto6000Kin1%stepsizeofpreviousone)andcorrespondingtargetluminousflux,ΦV,(1000lmto8000lmin3%stepsizeofitsprecursor)atoptimizedDuv(<±2x10‐3)(seeninFigure5‐1).TheblackdotsinthefigurerepresentthepossibleguidedpointsinthelookuptablefortheCCTrangingfrom3000Kto6000Kwithdimmablecapabilityfrom1000lmto9000lmcorrespondingtoeachCCTsetting.Thegreendots implythe foundsolutionsfromtheV8lightenginecorrespondingtoblackdots.Asseeninthefigure,thereweresomedifficultiestofindsolutionsfortheregionabove6500lmfor~3000K.Asimilaroccurrenceisobservedintheregionfrom~4200Kto4300K.ThiscouldappearduetothelimitationoftheLEDsusedinthecolormixture.


Figure 5‐1: Pre–calibrated lookup table consists of target CCTs (3000 K to 6000 K in 1% step size of the previous CCT and corresponding target luminous flux 1000 lm to 9000 lm in 3% step size of the

previous ΦV) at optimized Duv ( < ±2 x 10‐3) (Ref. Figure 1 of Chakrabarti et. al [24])

The lookup table isgenerated fromaspectral calibrationof theLEDcolormixingmodel.AftersettingthetargetofCCTandΦVasreference,thelookuptableisusedtofindtheoptimumadjustmentsfortheindividualLEDswhichareoptimizedtolowestDuv, and close to target CCTs. The target color settings for the light output are alsooptimizedbyahighgeneralcolorrenderingindex(CRI)Ra>85andspecificCRIR9>40forrenderingthestrongredobjects(seeninFigure5‐2).InFigure5‐2(a)and(c),foreachCCTsetting,thefoundDuvandCRI(Ra)fordifferentΦVaredistributedinvertically.Asexpected,theCRI(Ra)inFigure5‐2(c)islowinthewarmwhiteregioncomparedtothecoolwhiteregion.However,Duvare low(<±2x10‐3) throughouttheoperatingrangeinFigure5‐2(a).SimilarlyinFigure5‐2(b)and(d),foroneΦVsetting,thefoundDuvandCRI(Ra)forCCTrangingfrom3000Kto6000Karedistributedvertically.NosuchpatternssimilartoFigure5‐2(a)and(c)arefoundforDuvandCRI(Ra).


Figure 5‐2: (a and b) Optimized Duv, (c and d) and CRI (Ra) vs target CCTs and ΦV respectively (Ref. Figure 2 of Chakrabarti et. al [24])

ThetargetpointinthelookuptableoffersinformationaboutthecorrespondingDMXvaluesfortheindividualcolorchannels(seeninFigure5‐3).InFigure5‐3(a)‐(e),foreachCCTsetting, thecontribution fromdifferentLEDs fordifferentΦVareverticallydistributed.Figure5‐3(f)representsthesameresultwhileplottingeverycolorLEDinthesamegraph.Asexpected,toconstructthewarmwhite,thecontributionfromtheredLEDismorethanthegreenandtheblueLEDsanditdecreasesgraduallytowardsthecoolwhite.Similarly,thegreenandtheblueLEDcontributesmoreinthecoolwhiteregion than the red LED. The contribution from thewhite LED is almost the samethroughouttheentiretheCCTrangeasitisusedasabaseLED.


Figure 5‐3: Corresponding driving current information for individual LEDs for target CCTs for different ΦV (Ref. Figure 3 of Chakrabarti et. al [24])

Figure5‐4showsthebestfoundsolutionsforfourCCTs(3000K,4000K,5000K,and5800K)at8000lumenandtheseSPDsareconsideredastargetSPDsasblackline.InFigure5‐4,thepurplelinesarereferenceSPDs,whichforCCT<5000KarerepresentedbyPlanckianradiationspectrumandSPDsforCCT>5000KarerepresentedbyD–illuminant.TherestofthecoloredspectrumsinFigure5‐4representthecontributionfromindividualcoloredLEDstoproducethetargetSPD.TheSPDsfromindividualLEDsused in the model are calculated from spectral measurements (Chapter 6), withvariation in the various controllableparameters (current, junction temperature etc.)whicharealreadyexplainedinChapter4andinthearticlesofChakrabartiet.al[23].


Figure 5‐4: Resulting target SPDs (black line) of five color mixing LED system at four different target CCTs (3000 K, 4000 K, 5000 K and 5800 K in (a) – (d), respectively) and the corresponding reference

SPDs (purple line). Individual SPDs for the colored LEDs are also shown by different colors (Ref. Figure 1 of Chakrabarti et. al [23])

5.2.2UncertaintyanalysisbytheMonte‐Carlosimulation

The Monte Carlo simulation is developed by incorporating many of the inputparametersi.e.LEDbinning,ambienttemperatureanddriverelectronicssuchasdigitalcontroloncurrent,whichinfluencethecoloruncertaintywithinthelightengine.Theabovementionedinputcomponentsaredistinguishedbythreecategories(system,loopandhistory)basedontheirinherentnatureofinfluence.AlthoughLEDaging,abruptfailure of LEDs, and contamination inside the systemmay cause variations in lightoutput,thisthesisdidnotconsiderthoseinfluencesintothesimulationandfutureworkwilladdressthoseissues.ThefollowingTable5‐1describestheindividualcomponentsindetailwhichisalsoreportedinthearticleofChakrabartiet.al[23].


Table 5‐1: Components responsible for uncertainties in light output from light engine (Ref. Table 1 of Chakrabarti et. al [23])

ComponentDescription/unit

Typeofparameter Sensitivityfrom(Cause) Sensitivityto(effect)

Relation

LEDbinFlux(Φ)inlm,

wavelength(λ)innm

SystemandStatic Binwidth Φ,λ linear

DrivercurrentinmALoopandDynamic Digitalcurrent

controllerΦ,λ linear

Digitalcurrentcontrollerinnumber

LoopandDynamic Current,drivingcontroller

Φ,λ Droopcurvewithsteps

LEDjunctiontemperaturein°C

LoopandDynamic Current,runtime,andenvironment

Φ,λ linear

Temperaturein°CHistoryandDynamic

Ambienttemperature Φ,λ linear

Colorsensorspectralresponsein

number

SystemandStatic Tristimulusvalues Φ,λ linear

Thesimulationhas runcontinuously to createa realistic scenario for theV8 lightengine in case the engine would be exposed to different varieties of LED bin,environment,etc.TheflowchartinFigure5‐5describesthemethodoftheMonteCarloanalysisofthemulti‐coloredV8lightengine.Thechromaticitycoordinates(u ,́v´)ofthesimulatedSPDswillbeplottedintheCIE1976chromaticitydiagramandΔu´v ́willbecompared in the n–step u´v ́ circle (where n=1.18 and just noticeable chromaticitydifferenceΔu´v´=0.0013)asdefinedinCIETN001:2014[46].Theuncertaintiesoftheu′,v′chromaticity,CCT,Duv,andΦVhavebeencalculatedasoutputparametersfromthesimulationusingtheprobabilitydistributionoftheinputparametersforeachcoloredLED. The outputs from the simulation are analyzed subsequently by calculating thecoverageintervalandtheirsensitivitytotheinputparametersatdifferentCCTsettingsofthedynamiclightengine.


Figure 5‐5: Block diagram of the Monte Carlo simulation with N iteration for the multi‐colored LED system (Ref. Figure 2 of Chakrabarti et. al [23])

The experiments on the individual LED provide the variation in light output andpeak/dominatedwavelengthduetochangesintheinputparametersforeachcoloredLED.ThegraphsinFigure5‐6illustratestheradiantfluxandpeakwavelengthvariationwithrespecttochangesinthedrivingcurrent(top)andcasetemperature(bottom)forindividualLEDsataconstantcasetemperature(top)anddrivingcurrent(bottom).Theradiantfluxandpeakwavelengthvariationsarecomparedwithvaluesobtainedata300mAdrivingcurrentinFigure5‐6(a)and(b).Figure5‐6(a),showstheradiantfluxincrease almost linearly to the increased driving current.However, the slope of thecurveismuchhigherforaredLEDthanothercoloredLEDs.InFigure5‐6(b),thepeakwavelengthshiftstowardshighervalueswiththeincreaseofthedrivingcurrentfortheblueandredLEDs.However,theeffectisoppositeforagreenLED.ForFigure5‐6(c)and (d), the radiant flux and peak wavelength variations are compared to valuesobtainedatcasetemperatureof20°C.Figure5‐6(c)shows,theradiantfluxdecreaseforeachcoloredLEDalmostlinearlybyincreasingthecasetemperature.Inthiscase,theslopeofthecurvefortheroyal‐blueLEDishigherthanothercoloredLEDs.Figure5‐6(d)representsthesensitivityofthepeakwavelengthwithrespecttochangesinthecasetemperatureforeachindividualcoloredLED.Inthisgraph,peakwavelengthshiftstowardshigherwavelengthforeachcoloredLEDsbyincreasingcasetemperature.Asexpected,thesensitivityishigherfortheredLEDcomparedtoothercoloredLEDs.


Figure 5‐6: Sensitivity of radiant flux (a) and peak wavelength (b) for the individual colored LED by varying driving current to 300mA when the case temperature is at 20 °C. Sensitivity of radiant flux (c) and peak wavelength (d) for the individual colored LED by varying the case temperature to 20 °C when

the driving current is at 700mA (Ref. Figure 3 of Chakrabarti et. al [23])

Asweunderstandfromtheaboveexperimentalresults that theLEDperformancevarieswithoperatingcondition,thereisachallengetogetastablelightoutputfromtheLEDlightingsystemwithoutconsideringanycontrolledoperatingconditions.Thus,itiscritical to investigate and consider the uncertainties in the operating conditions ofindividualLEDs.Duetotheuncertaintiesintheoperationconditionontheindividualdriving current of LEDs, the light engine could not produce the light output withdesirabletargetvalues.TheMonteCarlosimulationsimulatesandappliesthevariationon operating conditions of the individual LED internally forN = 2000 iterations byconsidering theuncertainties (mentioned in Figure 5‐5) as input parameters. Theseresultsleadtosimulate2000resultantSPDsoftheLEDcolormixingsystem,actingasa2000virtualV8 lightengines.Figure5‐7showsanexampleof thevariationofSPDsfrom the individual LEDs due to the above mentioned uncertainties as inputparameters. By using the estimated uncertainties in output parameters due to theuncertaintiesininputparameters,theMonteCarlosimulationcalculatescorrespondingcolorquantitiesasoutput(namelyu ,́v ,́CCT,Duv,ΦV).


Figure 5‐7: Illustration of variation of SPDs with the variation of uncertainties as input parameters (Ref. Figure 4 of Chakrabarti et. al [23])

Asanexampleto illustratetheoutputoftheMonteCarlosimulation,5000Kwith8000 lmas targetCCTandΦV is considered (Figure5‐4 (c) and the correspondingvaluesareshowninthelook‐uptable).After,theMonteCarlosimulationisappliedonthetargetSPD.Fromtheresultofthe2000SPDs,thechromaticitycoordinates(u’,v’)arecalculatedandplottedintotheCIE1976diagraminFigure5‐8(b)asgreendotsalongwiththemeanpoint(blue‘+’)ofthosevariations.InFigure5‐8(a)and(c)thecorrespondingprobabilitydistributionofvariationsinv ́andu ́areplottedandindicatethelowerandupperboundof95%coverageintervalareabyredarrows.InFigure5‐8(b),theisothermalline(redline)representsΔDuvandthePlanckianlocus(blackline)represents ΔCCT,which is analyzed in Figure 5‐9. By taking themean point as thecentral point (blue ‘+’), a circle with a radius of 0.0013 (u´v ́ circle in CIETN001:2014[46])isdrawninFigure5‐8(b).Thefoundoutcomesaremostlyoutsidefrom the just noticeable chromaticity difference circle (blue). Even the target pointrepresentedbythebrown‘+’isalsooutsidethejustnoticeablechromaticitydifferencecircleindicatinganoffsetfromthelookupvalues.TheCCTandDuvarecalculatedfrom(u ,́v´)forthe2000iterations.TheprobabilitydistributionofthevariationsinCCTandDuvareshowninFigure5‐9(a)and(b),respectively.Consideringa95%coveragearea,ΔDuv=0.0095withthemeanof‐0.0046andΔCCT=339KwiththemeanofCCT5068Kare found.Similarly theMonteCarlosimulationsarerepeated for the targetsettingsfrom3000Kto6000KwithtargetΦVvaluesat8000lmtoobtainthesimilaroutputparametersasmentionedabove.Eachtimethesimulationrunsfor2000iterations.


Figure 5‐8: (a) and (c): represent probability distributions for v´ and u´ where the red arrows are indicating lower and upper bound of u´, for a 95% coverage area respectively. (b): All possible

outcomes for the 2000 iterations of the Monte Carlo simulation are plotted into u´,v´ diagram (CIE‐1976 chromaticity coordinates) at 5000 K (2:1 axes) (Ref. Figure 5 of Chakrabarti et. al [23])

Figure 5‐9: Probability distribution functions of CCT (a) and Duv (b) at 5000 K CCT settings by calculating the coverage interval of the corresponding colorimetric quantities where ΔCCT=339K and ΔDuv=0.0095

for 95% coverage interval (Ref. Figure 6 of Chakrabarti et. al [23])


The Monte Carlo simulation is analyzed to find the correlation and sensitivitybetweenthe inputparameters foreachcoloredLEDtooutputparametersof theV8lightengine.Thesensitivityestimationsareusedtofindthecriticalpartsinthesystemfor correction and system stability of the light engine at different CCT settings. TheresultsareanalyzedandplottedinFigure5‐10forthesensitivitytooutputparametersCCTandDuv.Ingeneralwithinthewarmwhiteregion,thecontributionfromtheredLEDishigherthanforthecoolwhiteregion.Similarly,inordertoobtainanSPDinthecoolwhiteregion,theblueLEDcontributesmorecomparedtotheredLED(asseeninFigure5‐3andFigure5‐4).Asanexample,inFigure5‐10thesensitivityinCCTandDuv,withrespecttousedindividualcurrentforeachcoloredLEDareplottedforthetargetCCTsrangingfrom2800Kto6000K.ThesensitivityofCCT(SCCT)inK/mAandSDuvmA‐1forDuvaremeasuredaccordingtofollowingequations.

∆∆

(45),

WhereΔCCTisthechangeinCCT(K)forchangeincurrentΔI(mA)oftheindividualcoloredLED.Therefore,theunitofSCCTisK/mA.

∆∆

(46),

WhereΔDuvisthechangeinDuvforchangeincurrentΔI(mA)oftheindividualcoloredLED.Therefore,theunitofSDuvismA‐1.

Asexpected,CCTandDuvareverysensitivetotheredLEDinthewarmwhiteregioninFigure5‐10anditisalmostzerotowards6000K.Similarlytheeffectwillbereversedinthe case of theblue/royal blue LED.The twodifferent coloredblue LEDs aremoresensitivewiththechangeinCCTandDuvat5000KcomparedtotheothercoloredLEDs.The green and the white LED contributions are always required in both abovementionedregions.TheCCTissensitiveinthewhiteLEDmorethanDuv.ThegreenLEDinfluencesbothCCTandDuv.TheblueandtheroyalblueLEDshavealmostthesameinfluenceonCCTandDuv.


Figure 5‐10: Sensitivity of output parameters namely in CCT (top) and Duv (bottom) at CCT settings ranging from 2800 K to 6000 K due to the individual five colored LEDs (Ref. figure 7 of Chakrabarti et.

al [23])

5.3Colorandluminousfluxcontrolmechanism

AsseeninFigure5‐8(b),thesizeofthedistributionofthepossiblefoundoutcomesforthe 2000 iterations has amuch larger spread than the just noticeable chromaticitydifference(JND)circle(asdefinedinCIETN001:2014[46]).Thevariationscorrespondston=8stepcircles(accordingtoequation(47))whichisfarfromthejustnoticeablechromaticity difference variations of n = 1.18. Hence the average human eye couldeasilynoticethedifferenceoflightoutputforthesametargetCCTsetting.Alargeoffsetbetweenthemeanvalueandthetargetvaluesisobserved.Therefore,thevariationofchromaticity and luminous fluxneed to control.A color control systemneeds tobe


implementedandthedevelopedMonteCarlosimulationwillneedtoevaluatethecolorfeedbacksystem.

∆ ´ ´0.0011

(47)

Therefore, thecolorcontrolmechanismpresentedhere isdivided intothreepartswhichare:1)controlalgorithm,2)hardware,and3)statisticalanalysisofuncertaintycontributors/components,calledas“uncertaintyparameters”(listedinTable5‐1).Thecolorcontrolmechanismincorporatesthepre‐calibratedlookuptableandacalibratedcolor sensor thatmonitors the chromaticity of theuncorrected resultingwhite lightoutputprovidingafeedbacktothecontrolsystemtoregulatethecolorandthelightoutputvariationswithinacertainlimit.Inordertocorrectthelightoutput,thecontrolunitoffersanewsetofdrivingcurrentstotheindividualcolorchannelbychoosinganew lookup point. The stability of the system is characterized by evaluating theresultingchromaticities in theuniformchromaticitydiagram(CIE1976[46])withatargettolerancewithinacircleradiusof0.0026(n=2).Aschematicdiagramofthecolorcontrol system is shown in Figure 5‐11. The following paragraph elaborate on theprocessagainindetailwhichisallreportedinthearticlebyChakrabartiet.al[25].

Figure 5‐11Schematic diagram of the color control system of the multi‐colored tunable (3000 K to 6000 K) V8 light engine, Tamb = ambient temperature and Tjun = junction temperature of the LEDs (Ref.

Figure 1 of Chakrabarti et. al [25])

First,theusersetsthetargetCCT(3000K–6000K)andΦV(1000lm–8000lm)fortheV8lightengineasreference.Thecontrolalgorithmfindsamatchinthelookuptableclosesttothetargetvalues.Intheidealcase,thelightingsystemdeliversthetargetbyusingcorrespondingDMXvaluesofindividualcolorchannelsfromthatlookuppoint.


Theaforementioneduncertaintyparameters (Table5‐1) inside thesystem influencethelightoutputwhichwillbeanalyzedbythestatisticalanalysisusingtheMonteCarlosimulationprocesswithiterationN=2000(explainedinthesection5.2).Thereafter,theMonte Carlo simulation process investigates the colorimetric and photometricvariationsofthespectrallightoutput(u ,́v ,́andcorrespondingCCT,Duv,andΦV)basedon the uncertainties such as current response (lm/mA, nm/mA), DMX response(mA/DMX),LEDfluxandwavelengthbin(lm,nm),junctiontemperatureofLED(Tjun,°C),ambienttemperature(Tamb,°C),andcolorsensorresponse(W/nm)asshowninFigure5‐11(extremerightredblock).TocontroltheCCTandΦVvariationsinreal‐time,the color sensorhasbeen implemented tomonitor andget feedback from the lightoutput (as seen inFigure5‐11).The color sensorusedhas approximate tristimulusresponsivity( , and ̅)andneedstobecalibratedclosetotheoperatingchromaticityrange (explained in the section 6.5.1 in Chapter 6) in order to minimize spectralmismatcherrors.Thecolorsensorwascalibratedbyusingahandheldspectrometer(GS‐1150)[118].Afterhaving themeasured tristimulusvaluesof the resultingwhitelight[119],thecontrolunitcalculatesthecorrespondingerrorcorrectionforCCTandΦVwhichdirects touseanewlookuppoint fromthe lookuptable.Thenew lookuppointprovides thecorrecteddriving informationforthe individualcolorchannelstodeliveracorrectedlightoutputwhich isclosertothetargetCCTandΦVvalues.ThelookuptablehasalreadytheoptimizedDuvvalues(<±5x10‐3)foreachlookuppoint.Thus,thenewlookuppointshouldhaveDuvvalue<±5x10‐3.However,ifthecorrectedlightoutputcannotprovidethecontrolledlightoutputwithinacertainlimit[46],theDuvneedstobecontrolledusingtheindividualcolorsensitivitytoDuv(Figure5‐10(b)).IftheobtainedDuvshowsthehighervaluethanthedesiredDuvvalue,thegreenchannelneedstogodownalongwithaddingmorefromredand/orbluechannelsdependingontheCCTregion.FornegativeDuvcorrection,thegreenchannelneedstogoupwhiletheredand/orblueneedtodrive less thantheearliersettings.Finally, thechanges indriving currents at the respective color channels leads to choose new lookup pointwhich is evaluatedbyplotting in theCIE1976 (u ,́v´)diagram.The controlledpointinsidetheCIE1976(u ,́v´)arenotwithinJNDcircle.However,95%ofcontrolledpointsarewithinn=2circle.theyareTheproposedcontrolmechanismcannotcontrolcolorrenderingindices(CRI(Ra),CRI(R9)[120].However,thecontrolledlightoutputhasgoodlightqualityintermsofCRIRa(>85)andR9(>40)observed.Thecontrolmechanismcanbegeneralized tobeused inanyotherSSLsystem(suchasadditive laser colormixingsystem)where insteadofusingthecolorsensor,aspectrometermayuseforsensingthespectralinformation.

5.4Discussiononcolorcontrol

Thesimulatedlightoutputfortuningrange4000Kto6000Kshowsanaverageof8.5%variationonΦVwithina95%coverageintervaloverinitialsettingof8000lmduetouncertaintiesoftheinputparametersiftheTambchangesfrom20°Cto45°C.Generally,redLEDsarehighlysensitivetowardsachangeincurrentandtemperaturecomparedtootherLEDs(Figure5‐6).Thus,duetolargercontentofredLEDsthanotherLEDsinthewarmandneutralwhiteregion,thehighercolorvariationisobservedwithinthatregion.ThehighestuncertaintyontheCCTat4000Kisfound1020K.AsanexampletoillustratethecolorvariationandthecorrespondingcolorcontrolintheV8lightengine,


thevaluesof5500Kand8000lmastargetCCTandΦV,respectively,areusedandthecorresponding DMX values for individual color channels from the lookup point aretransferredtothelightingsystemtosimulatetetargetSPD.Afterwards,theMonteCarlomethodisappliedonthetargetSPDtoimposethevariationsduetoinputparametersfor2000iterations.Forexample,basedon2000iterationsFigure5‐12representstherandomdistributionofheatsink temperaturesbetween25 °Cand50 °C.Due to thevariationoftheheatsinktemperature,thejunctiontemperatureoftheLEDalsovaries.Figure5‐13illustratesthechangeinthejunctiontemperatureforindividualLEDsduetochangesintheheatsinktemperature.

Figure 5‐12: Random distribution of heatsink temperature for 2000 iterations

(a) (b)


(c)

(d)

(e) (f)

(g) (h)

Figure 5‐13: Relation between junction temperature of individual LED and heatsink temperature

Figure5‐13(a)showsthejunctiontemperaturevariationwithrespecttothechangeintheheatsinktemperaturewhileallLEDsaresetatfullelectricalload.Thatmeansthewhite;thegreen,theblueandtheroyalblueLEDsareoperatingata1000mAcurrentwhiletheredLEDisoperatingwitha700mAcurrent. Inthisscenario, the junctiontemperatureofthegreenLEDisalwayshigherthantheotherLEDsanditreaches~110°Cwhile the correspondingheatsink temperature is at60 °C.Figure5‐13 (b) to (h)


representsthejunctiontemperaturevariationsoftheindividualLEDsat3000K,3500K,4000K,4500K,5000K,5500K,and6000K,respectively.AsseenfromFigure5‐3,the contribution for each LED is not the same for all regions of the white lightgeneration(warm,neutral,andcool).Therefore,atthewarmwhiteregion,(Figure5‐13(b)to(d))thejunctiontemperatureoftheredLEDishighcomparedtotheneutralandcoolwhite regions (Figure 5‐13 (e) to (h)). Similarly the green LED takes over thepositionof the redLED forhaving ahigh junction temperature if theCCT from thewarmwhitetransitstowardsthecoolwhiteregion.However,thechangeinthejunctiontemperatureinfluencestheLEDstochangetheluminousfluxand/orwavelengthshiftmayoccur(Figure5‐6).Duetotheinfluencefromtheuncertaintycontributors,thelightoutputfromtheV8lightenginecannotprovidethedesiredtargetlight.ThegraphinFigure5‐14 isacomparisonbetweenthe threeresultantSPDsobtained fromapre‐calibrated lookup point (as target), uncontrolled light output (due to uncertaintyparameters), and a new pre‐calibrated lookup point (for controlled light output) at5500K.Asshowninthegraph,duetowavelengthshiftinthesystemfromthetarget,theSPDsforcontrolledlightadjustcorrespondinglyandshiftinthewavelengthspacecomparedtothetargetlight.However,thecontrolledlightoutputdeliversthedesiredlightoutputwhichisclosertothetargetCCTandthetargetΦV.

Figure 5‐14: Compared SPDs between target, before and after implementing the control system

Fromtheresulting2000SPDs, thechromaticitycoordinates(u ,́v´)arecalculated.TheprobabilitydistributionsofvariationsinCCTandDuvfor2000systemsarederivedand are shown in Figure 5‐15(a)‐(b). The coverage interval of 95 % from thedistribution implies thatΔCCT=738Kwithmeanof5709KandΔDuv=0.0252withmeanof0.0028,arerepresentedbythegreencolorinthesamefigure.InFigure5‐15(c)and(d)thecorrespondingprobabilitydistributionofvariationsinCRI(Ra)andCRI(R9)derivingfromtristimuluscolorsensoroutput[119] , indicatinga95%coverageintervalbyalsogreencolorhaveplotted.Withinthe95%coverageintervalΔCRI(Ra)=15.4withmeanof91andΔCRI(R9)=95.7withmeanof43,respectivelyhavefound.Similarly, the sameoutputparameters fromMonteCarlo simulations for targetCCT


settingsfrom4000Kto6000KwithatargetΦVof8000lmhaveevaluated,whichislistedinTable5‐2.

Figure 5‐15: (a) ‐ (d) represent the probability distributions within a 95 % coverage interval (green) of CCT, Duv, CRI (Ra), and CRI (R9), respectively, for 2000 iterations while the red color is indicating the

same for outside of the 95 % coverage interval (Ref. Figure 3 of Chakrabarti et. al [25])

Table 5‐2 CCT and ΦV variations within 95% coverage interval at target settings of target CCT and ΦV respectively (Ref. Table 1 of Chakrabarti et. al [25])

TargetCCT[K]atΦV=8000lm

ΔCCT[K](uncontrolled)

ΔCCT[K](controlled)

ΔΦV[lm](uncontrolled)

ΔΦV[lm](controlled)

4000 1020 125 737 1694500 932 155 704 1675000 898 211 684 2305500 738 282 614 1856000 836 381 653 187

FromTable5‐2 it canbenoticed that theuncontrolledΔCCTandΔΦVvaluesarehigherat4000KcomparedtootherCCTsettings.ThisisbecausetheuncertaintyontheoutputfromtheredLEDishighestatthissettingandthereforeitscontributiontotheuncertaintyishigh.


Figure 5‐16 (a)‐(b) represents the variation of CRI(Ra) and CRI(R9)for the 2000iterationbybluedotsbefore implementing the control systemandgreendots afterimplementationthesame,respectively.Here,thepurple‘+’representsthetargetvaluesfortheCRI(Ra)andCRI(R9).Asthecolorcontrolsystemcontrolsthecolorvariationupto a certain level, the green dots are closer to the target value than the blue. ThestandarddeviationofCRI(Ra)andCRI(R9)variationsarealsoplottedintheFigure5‐16(a)‐(b),wherethebluelines(65%coverageinterval)ofthelightoutputinthesystemarelargerthanthegreenlines(65%coverageintervalafterimplementingthecontrolsystem).SimilargraphscanbeplottedforΦV,CCT,andDuv.

Figure 5‐16: The variation of mean and standard deviation of (a) CRI (Ra) (b) CRI (R9) for 2000 iteration for before and after implantation of the control mechanism, respectively (Ref. Figure 4 of Chakrabarti

et. al [25])

Figure5‐17showsthedistributionofthefourparametersforthe2000iterationsafterimplementingthecontrolsystem.Thegreenlinesshowthedistributionwithina95%coverageinterval.Figure5‐17(a)and(b)showsthatthevariationofCCTandDuvhas


beenreduced toΔCCT=133KandΔDuv=0.0011within the95%coverage interval.Similarly,Figure5‐17(c)and(d)showΔRa=0.18andΔR9=4.9within95%coverageinterval.

Figure 5‐17: (a) ‐ (d) represent the probability distributions within 95% coverage interval (green) of controlled CCT, Duv, CRI Ra, and R9, respectively, for 2000 iterations while the red

color is indicating the same for outside of the 95% coverage interval (Ref. Figure 3 of Chakrabarti et. al [25])

Figure5‐18indicatestheworkingperformanceofthecontrolsystembyshowingthecolor variations of the 2000 iterations in the u ,́ v ́ diagram (CIE‐1976 chromaticitycoordinates)beforeandafterhavingthecolorcontrolsysteminsidetheV8lightengine,respectively. InFigure5‐18, the isothermal line (blackdash lines)demonstrates thevariation in Duv (ΔDuv) and the Daylight locus (solid black lines) characterizes thevariationinCCT(ΔCCT).Figure5‐18alsoshowsaJNDcircle(redline)andn=2circle(dashedredline)witharadiusof0.0013and0.0026,respectively,[121](u ́v ́circledefinedinCIETN001:2014)bytakingthetargetpoint(purpledot)asacenterpointofthecircle.ItisclearthatthetargetpointindicateanoffsetfromtheDaylightlocus.Thisis because the lookup point values are not optimized to Duv at zero, rather; it isoptimizedto lowDuv(<±10‐3)asrecentfindings[101], [102]showthatthehumanpreferenceforCCTandanoptimumDuvmaynotbeatzeroDuv.FromFigure5‐18itisobserved that the found solutions before implementing the control system (blue ‘.’)


almostallareoutsidefromtheJNDcircle(red)andnotevenclosetothetargetpoint.Similarly, the green ‘o’ implies the solutions for those 2000 iterations afterimplementingthecontrolsysteminFigure5‐18.Itisclearfromthefigurethatthecolorcontrol systemcouldnarrowdownthevariationalong theDaylight locus (ΔCCT)aswellasalongtheisothermalline(ΔDuv).Althoughtheallcontrolledpointsarenotyetwithin the JND circle[46] (50% of human observer cannot preserve the colorvariations),95%pointsarewithinn=2circle.Thepresentlookuptableismadeintwodimensions, by registering the variations in CCT and ΦV, where each lookup pointrepresentsoptimizedDuvvalue(<±10‐3).ThefutureworkwillincreasethedimensionofthelookuptablewherethevariationsinDuvwillalsobestoredinthirddimension.ThethreedimensionallookuptableswillresolvetheissueanditwillbeexpectedthatthecontrolledpointswillbeinsidetheJNDcircle.

Figure 5‐18: The example of uncontrolled and controlled solutions from the lighting system at 5500 K in u´‐ v´ diagram (CIE‐1976 chromaticity coordinates) (Ref. Figure 4 of Chakrabarti et. al [25])

Figure5‐19outlinesthecontrollingrangeoftheV8lightengineatΦV=8000lm.TheV8lightengineistunablefrom3000Kto6000Kinstepsof100K.TheredsolidsquaredatapointsrepresentsthecontrolledCCTvaluefoundatthecorrespondingtargetCCTandtherelationbetweenthemislinearwithR2=0.965.ThecontrolledluminousfluxfoundatthecorrespondingtargetCCTissignifiedbythebluesolidcircledatapoints,whilethedottedblacklineshowsthetarget:ΦV=8000lm.AlthoughtheΦVvariationduringthetuningCCTrangesfrom3000Kto6000K,itiswithin~1%,thefoundCCT


valuesarehigherthanthetargetvalueattherangebetween3000Kand3500K.ThisisduetothefactthatwithinthatregionsomelookuppointsinsidethelookuptablehavehaddifficultyinfindingtheinitialSPDsforthecorrespondingCCTsettingsexplainedearlierinsection5.2.1.

Figure 5‐19: Controlled light output from the lighting system while the CCT tuned from 3000 K to 6000 K with 100 K resolution at 8000 lm (Ref. Figure 5 of Chakrabarti et. al [25])

5.4.1Whythiscontroltechniqueisuseful?

Although themulti‐colored LED lighting systems are popular nowadays, they oftensufferfrompoorcolorstabilityandreproducibilityfromsystemtosystemduetocolorandintensityvariationovertime.Toaddresstheseproblemswithoutcompromisingtheefficiencyandlightquality,amethodologyofcontrolmechanismbyintroducingacolorfeedbackinsidethesystem,coveringawidewhitegamuttoachievecontrolledchromaticity and stability during operation time is proposed. The V8 light engineproduces white tunable light output, ranging in CCT from 3000 K to 6000 K withintensitydimmingcapabilityofminimum13%loweron8000lmateachCCTsetting.ThecolorcontrolsystemimplementedintotheV8isacombinationofapre‐calibratedlookuptableutilizingasetofpreviouslyoptimizedsettingsforeachtargetCCTandacalibrated color sensor for chromaticity control. The pre‐calibrated lookup table ismade by an algorithmwhich stores the resultant SPDs as a lookup point from thespectralmixturesoffivecoloredLEDs,optimizedtolowDuvandahighgeneralcolorrenderingindex(CRI)ofRa>85andaspecificCRIR9>40forrenderingthestrongredobjects.ThecolorcontrolsystemisstatisticallyanalyzedusingaMonteCarlosimulationprocesswhichsimulatestheSPDofthemixedlightandhencevariationsinCCT,DuvandΦV as output basedon theuncertainties inside the input parameters (LED flux andwavelength bin, junction temperature, driving current, ambient temperature, driver


electronics, and color sensor response)of theV8 system.Theuncertainties in inputparametersareeitherfoundexperimentallyortakenfromdatasheets.Therearenotmany people who use Monte Carlo for investigating the uncertainties of inputparameterstoevaluateuncertaintiesofoutputparametersforadynamicsystem.TocontroltheseCCTandΦVvariations,thecolorsensorhasbeenimplementedtoprovidetherealtimefeedbackfromthelightoutput.ThecalibratedcolorsensormonitorsthewhitelightoutputfromthelightingsystemandevaluatesontheerrorcorrectioninCCTandΦVsothatanewlookuppointcanbechosentoachievethedesiredlightoutputstabilityfromthesystem.Thelongtermstabilityandaccuracyofthesystemhasbeeninspected with a target tolerance within a circle radius of 0.0013 in the uniformchromaticity diagram (CIE 1976) which is recommended in CIE TN001:2014. Thecontrol systempresentedhere does control theDuvwithin two step just noticeabledifferencecircle,andthiswillbeimplementedinfuturework.Thecontrolmechanismcanbegeneralized tobeused inanyotherSSLsystem(suchasadditive laser colormixingsystem)where insteadofusingthecolorsensor,aspectrometermayuseforsensing the spectral information. Three LED color mixing systems are underdevelopmentandwillbeusedtoexperimentallydemonstratebothsystemstabilityandsystemtosystemreproducibility.TheMonteCarloresultswillbecomparedwiththeexperimentalresultsandfurthersimulationworkontheuncertaintycontributionsduetoabruptfailureofLEDs,LEDaging,andcontaminationoftheopticalsystemwillbeconsidered. In this way it will open up new possibilities for analysis of the coloruncertaintyofmulticoloredLEDsystemswhichwouldeasilybeexpandabletolargerand more complex systems, without dealing with analytical propagation ofuncertainties.

5.5Colorvariationinlaser‐colormixingsystem

Sincelaserlightisapotentialcandidateformakingthespotlight,asmentionedintheChapter 4, the research area for color mixing by laser light has been investigatedinitially.Thereforethecolorvariationfromsuchasystemisalsoconsideredinthepartof theresearchandthusthe ideaofaspecklewavemeter isdiscussed.Normallythelightoutputfromthelaserissensitivetotemperature[122].Generally,therewillbeatemperaturedependence[123]onthewavelengthshiftforalongoperationaltime.Thischange in wavelength can be recognized by the speckle wavemeter which isinexpensivecomparedtothecommerciallyavailablespectrometers.Therefore,insteadofusingthecolorsensor,thespecklewavemetercanprovidethewavelengthchangeinused laserof the laser colormixingsystemwhichcanbe converted to the (u ,́v´) toprovide the error correction in the feedback system of the control mechanism(explainedabove).ThefollowingparagraphswillelaborateonthefunctionalityofthespecklewavemeterbasedonthepreviouslypublishedarticlebyChakrabartiet.al[21]andHansonet.al[22].

5.5.1Specklebasedwavemeter

Manymaterialshavesurfaces thatare significantly roughon thescaleof theopticalwavelengthandtheirlateralscaleusuallyexceedsthediameteroftheilluminatingspot.Whenanearlymonochromaticlightisreflectedfromsuchasurface,theobjectsviewed


incoherentlightacquireagranularappearanceoftheintensitywherethestructureofthis granularity is chaotic andunorderedwithan irregularpattern that canonlybedescribedbythemethodsofprobabilitytheoryandstatistics.Thispatternisnamedaspecklepattern[124]–[126].Numerousapplicationshavebeenpresentedwherethedisplacement of this speckle pattern arising from a dynamic scattering surface isinvestigated.Bymeasuringspecklecross‐correlationbeforeandafterdeformationoftheobjectusingaphotodiodearray,Yamaguchi[127]–[129]directlyfoundstraindueto small object deformations. Later, Yamaguchi et al. [130] also showed roughnessmeasurementbycalculatingspecklecross‐correlationbeforeandafterachangeoftheincidentangleofalaserbeam.Anabundanceofapplicationsinthisfieldisavailable.

Here,thecross‐covarianceisusedbetweentwospecklepatternsobtainedwithonlyadifferenceinthewavelengthtomeasurethechangeinthewavenumberofalaserbeam,and thereby propose and demonstrate a novel attractive, inexpensive specklewavemeterwith a reasonable resolution. Brandon et. al [131] proposed a low lossspectrometer by using speckle pattern images generated by interference duringpropagationinmultimodefibers.Thestandardwayofmeasuringwavelengthsisbyagratingspectrometer[132],Fabry‐Perot(FP)interferometers[133],wavemeters,orbyheterodyne methods. In the grating‐based spectrometer, light with differentwavenumbers ( ∆ ∆ )aredeflected indirectionsdeterminedby thewavelengthoflight.Thissetupisdelicateifahighresolutionisneeded,andfurthermorethetemporalresolutionislow.FPspectrometersaresensitivetoalignandthetemporalresolutionisusuallylimited.Thismethodisbasedonallowingthelighttopassbetweentwohighlyreflectivemirrorswithavariabledistance.Whenthemirrorspacingmatchan even number of standing waves, the resonant wavelength will pass, and thewavelength can be found within certain limitations, but unfortunately with aredundancygivenbytheso‐calledfreespectralrange(FSR)oftheFPinterferometer.Comparedwith other expensive spectrometers, the proposed specklewavemeter issimpleandproviding thepossibilityof ahigh temporal resolutionanda reasonablygoodspectralresolution,however,onlyprovidingchangesinwavelength.Inthisway,theproposedprecisesystemiscomparablewiththeFPinterferometerwithoutsomeofits limitations. The following section will provide the basic theory for the system,supportedbytwosimpleexperiments.

Figure 5‐20 shows the basic setup, here for free space propagation. A coherentincidentbeam illuminates thesurfaceatanobliqueangleand thespecklepattern isobserved with a CMOS array, being it linear or 2D, placed at a distance L. As thewavenumber is changed, the speckle pattern will shift laterally, as shown in thefollowing.Forthesakeoftheoreticalsimplicity,theincomingandoutgoingbeamsaresupposedtobeinthesameplane,asisthenormaltothescatteringsurface.


Figure 5‐20: Experimental setup for probing wavelength changes by observing the speckle displacement (Ref. figure 1 of Chakrabarti et. al [21])

Thecross‐covariance , ; ,C k k P p betweenthespecklepattern‐beforeandafterashiftinwavenumber‐isgivenbytheequation(48).

, ∆ ; , ∆2

⟨∆2;

∆2

∆2;

∆2

⟩

⟨∆2;

∆2

⟩ ⟨∆2;

∆2

⟩

(48)

wherethevariableshiftoftheintensityIisgivenby p andthecenteroftheCMOSarrayispositionedatP .Angularbracketsindicate“ensembleaverages”.Assumingthefield toobeycircularsymmetricstatistics– i.e.assumingthespeckle field tobe fullydeveloped– thecross‐covarianceof the intensitiescanbereducedtocalculating thesimplerfieldcorrelation:

, ∆ ; , ∆

2 |⟨ ∆ /2; ∆ /2 ∗ ∆ /2;

∆ /2 ⟩|

(49)

Thefieldscatteredoffthesurface(r‐plane)andobservedintheCMOSplane(p‐plane)canbewrittenas:

;

2

∗

∗2

2 .

(50)

Herethescatteringphases r willbeassumeddelta–correlatedinaccordancewith

the surface giving rise to a fully developed field, i.e. the ensemble average being

21 2 1 24 / k r r r r .Theensembleaverageisdenotedbyangularbrackets


andtheincidentdeterministicfieldis inU r .Thelinearphaseoftheincidentbeamand

theoutgoingbeamis 1 2sin[ ] and sin[ ]kx kx , respectively,havingassumedthatthe

scatteringandincidentwavevectorslieinthex‐plane.Inordertotrackthefieldfromtheroughsurfacetothedetectorarray,wehereusethematrixformalism[134],[135],wheretheGreen’sfunctionisgivenbythecomplexnumbersA,BandDdefiningthe

optical system as 2 . , including apertures

(Appendix C). In this formalism all apertures are assumed Gaussian apodized. Twoopticalsystemsaretreatedhere,namelyfreespacepropagationA=1,andD=1andaFouriertransformsystemwiththescatteringsurfaceinthefrontfocalplaneofalenswithfocallengthfandtheCMOSarrayinthebackfocalplane,withoutanyaperturesA=0,andD=0.Ananalyticalexpressionforthecross‐covariancecannowbeobtainedbyinsertingtheexpressionofequation(50)intheexpressionforthecross‐covarianceinequation (49) taking into account the delta – correlation of the scattered field. Acommon factor in front of the cross‐covariance has been omitted as it carries littleinformationonthespecklesandtheirdynamicbehavior.Further,theresultisgivenforthecenterofthedetectorarray,i.e.P=0.Foropticalsystemswithoutapertures,theon‐axiscovariancebecomesthen:

0, ∆ ; , ∆k ∝ 4

∆ ∆ cos16 cos ∆

(51)

This expression is valid for a collimated incident beam of width 0 (1/e2 intensity

radius). The scattering wavevector in the plane of the object is given by where 1 2sin sin . The peak position 0x

p for the cross‐covariance reveals thechangeinwavenumbergivenby

∆ ∆ (52)

Incaseofafreespacesystem,asdepictedinFigure5‐20,theB‐elementequalstheopticallengthL,andAisthemagnification,hereinunity.IncaseofaFouriertransformsystem,anintermediatelensislocatedbetweenthesurfaceandtheCMOSarrayplacedinthebackfocalplaneofthelens.Here,theB‐elementequalsitsfocallengthfandA=0.Incasea limitingaperture is included foraFourier transformsystem,averyminorchangeinthespeckledisplacementgivenbyequation(52)willoccur.

It is noticed that a slight uneven displacement is observed, as a function of theposition in the observation plane, which would be revealed by writing the fullexpressioninequation(51)where P 0 .Infact,thespecklepatternexpandsfromthepointwherethelightreflectedoffthesurfacewouldhittheplaneoftheCMOSarray,had the surface not been specularly reflecting. The effect of an uneven speckledisplacementwillthusdecreaseastheprojectionofthescatteringwavevectorintheobject plane increases,which fortuitously corresponds to the ratio between speckledisplacementandchangeinwavevectorbeinglargest,i.e.theconfigurationgivingthe


highestsensitivity.Besides,afinitecurvatureoftheincidentbeamwillnotinfluencethescalingderivedinequation(52),neitherwillitchangethespecklesize.But,needlesstosay,ahighdegreeofmechanicalstability isrequired,as is thecase foramajorityofinterferometers. A slight change in angle of the incident beam between the twoexposureswillgiverisetoanerrorinthedeterminationoftheshiftinwavevectoraswillanin‐planedisplacementoftheobjectincaseofafreespacepropagationsetup.IntheFouriertransformcase,onlyachangeinanglewillgiverisetoashiftinthespecklepattern.Alineardisplacementofthescatteringstructurewillinthiscaseonlygiverisetospeckledecorrelation.

Forfreespacepropagation,equation(51)willgivethespecklesizebythewidthofthecross‐covariance in xp for 0.k The speckle size becomes

1 0 0 22 cos[ ] / ( cos[ ).L K In the same way, the decorrelation due to a change in

wavenumberisgivenby 2 21 04 cos[ ] / ,decork L ingoodagreementwiththeknown

expressionforthelengthofthespeckles.Oneexperimentwascarriedoutinordertoverifytheconcept.Acollimatedbeamfromaverticalcavityemittinglaser(VCSEL,typeTO510)emittingat awavelengthof760nmwasusedgivinganalmost collimatedbeamatthetarget,herebeingasandblastedaluminumplate.ThespecklepatternswererecordedwithaCMOSarray(1024x1280pixels,pixeldistanceof5.2m).ThespeckleimageswererecordedsuccessivelyafterchangingthedesiredparameterfortheVCSEL,afterwhichthecross‐covariancebetweenthetworecordingswascalculatedandthepeak position found with a Gaussian fitting procedure. A representative specklerecording is shown in Figure 5‐21. The displacement incurred during a change ofwavevector is horizontal in Figure 5‐21, i.e. along theminor axes of the ellipsoidalspeckles.Theelongationofthespecklesisduetothegrazingangleoftheincidentbeam,herebeing 0

1 85 . Theangleofobservationwas 2 =30,andthediameterof theilluminatingbeam–beforeincidence–wasapproximately1mm.ThedistancefromtheobjecttotheCMOSarraywas20cm.

Figure 5‐21: Speckle pattern from a slanted object resulting in a needle‐like speckle pattern (Ref. figure 2 of Chakrabarti et. al [21])

AseriesofrecordingsweremadewithaconstantcurrentappliedtotheVCSEL.Atthesametime,theVCSELtemperaturewaschangedwithaPeltierelementplacednexttotheVCSEL.AcalibratedthermistorwasplacednexttotheVCSELandthetemperaturewaschangedandsubsequentspeckleimageswereacquiredforvarioustemperatures


afterastabletemperaturewasobtained.Atwo‐dimensionalcorrelationwasperformedwithsubsequentimagesandthepeakpositionofthecorrelationfunctionwasfoundwithsub‐pixelaccuracybyGaussianfittingof7pixelsand42pixelsinthehorizontalandverticaldirection,respectively,aboutthepeakpixelvalue.Thespecklewidthwas3.5pixels,FWHMandtheheight,perpendiculartothedirectionofmovement,was39pixels.Ithastobenotedthattheentire2Dcovariancefunctionisusedforestimatingthespeckle shift. Mathematica software is used to compute the various Correlationfunctionsof speckles.Thepeak location isdetectedbycentroidandparabolic fittingafterdoingFouriertransformoftwoimagesasshowninFigure5‐22.Atthatmomentitisnotdonebyrealtime.However,theprocessistakinglessthan1minute.

Figure 5‐22: Speckle pattern of (a) first image (b) second image (c) cross correlation


Figure 5‐23: Change in peak position of the cross‐correlation function measured in pixels as a function of the change in VCSEL temperature for constant injection current (Ref. figure 3 of Chakrabarti et. al

[21])

Figure5‐23showstheresultwherethespecklepatternisshiftedasafunctionofthetemperature change in injection current.The first readingwas fora temperatureof27.78°C,andthe lastreadingwasmadeat45.16°C. Inthissetup,achange inpeakpositionofonepixelcorrespondstoachangeinwavelengthof11.44picometer,where1picometerwavelengthchangecorrespondstoafrequencychangeof520MHzatthiswavelength.Therecordingwasmadebycorrelatingallspecklepatternswiththefirstoneacquiredatthelowesttemperature,andsubsequentlyfindingthepeakpositionsfor each of following. The measured change in wavelength as a function of thetemperaturebecomes46.1pm/ °K,where the specification for theVCSEL reports atypical valueof55pm/ °K.The injection current, atwhich thiswas stated,wasnotstated by the manufacturer. Figure 5‐24 shows the result where the speckledisplacementisdisplacedasafunctionofthechangeininjectioncurrent.Inthissetup,achangeinpeakpositioncorrespondstoachangeinwavelengthof25.3picometer.Themeasured change in wavelength as a function of the change in injection currentbecomes0.706nm/mA,wherethespecificationfortheVCSELreportsavalueof0.3nm/mA. In this experiment, the temperature of the VCSELwas not kept constant,whichcouldexplainthedifference.


Figure 5‐24: Change in peak position of the cross‐correlation function measured in pixels as a function of the change in VCSEL injection current (Ref. Figure 3 of Hanson et. al [22])

The change in the emittedwavelength as a function of the natural heating of theVCSELwasfoundataconstantinjectioncurrent.Hence,aseriesofmeasurementswereobtainedbyacquiringadatapointforevery30seconds.Thefirstspecklerecordingwasusedasareferencewithwhichthefollowing9recordingswerecorrelated.ThetemporalchangeinpixeldisplacementisshowninFigure5‐25.ThespecificationfortheVCSELcallsforathermalchangeinwavelengthof60picometer/K.Usingthisvalue,theheatingoftheVCSELforthegiveninjectioncurrentis10m°K/sec.

Figure 5‐25: Temporal change of the peak pixel position as a function of time, here giving a wavelength change due to natural heating of the VCSEL (Ref. Figure 4 of Hanson et. al [22])


Thesensitivityofthesystemheavilyreliesontheaccuracy,withwhichthespeckledisplacementcanbeprobed.AccordingtoZhou[136],anaccuracyof0.005pixelcanbeobtained by developing an iterative, spatial‐gradient based algorithm, by using onlyfirst‐order spatial derivatives of the images before and after deformation. Later, amethodbasedonnewalgorithmsfor2Dtranslationimageregistrationtowithinasmallfractionofapixelthatusenonlinearoptimizationandmatrix‐multiplydiscreteFouriertransformswereputforward[137].Here,accuracyforafastmethodfor2Dcorrelationgaveanaccuracyofcloseto0.01pixel.IftheresultsformFigure5‐23areusedandnoerror isassumed in the temperaturemeasurementand furthermore,a totally lineardependencyofthewavelengthversustemperaturerelationareassumed,theaccuracyin that experiment would become 0.03 pixels. This again would correspond to ameasurementaccuracyof180MHz.

5.5.1.1Utilityofthespecklewavemeter

Therefore a novel, yet simple, speckle‐basedwavemeter aimed atmeasuringminorchangesinwavelengthhavebeenpresentedhere.Experimentsthatsupportthevalidityofthemethodhavebeengiven,althoughnoattempthasbeengiveninordertoobtaintheultimateaccuracyachievable.Anaccuracyofatleast100MHzcanbeobtainedbyusing a CMOS arraywith a higher number of pixels and having a setupwhere theincidence and observation angles aswell as the spot size on the target are furtherincreased.Finally, theuseofa1Dcorrelation functionrowbyrowwithsubsequentaveragingwillfurtherincreasethesensitivity.

Thestandardwayofmeasuringwavelengthchangeswouldbewitha commercialscanningFabry‐Perotinterferometer,withwhicharesolutionofdownto7.5MHz.butwiththedrawbackofhavingafreespectralrangeof1.5GHzincaseofafinesseof200.Adelicatealignmentisneeded,andaseriesofconcavemirrorshavetobeacquiredinordertocoverthevisibleandnear‐infraredregime.ThescantimeforFPinterferometerisgivenbythemechanicalresponseofthemechanicalsystemandthecontroller.Hereascan time of 10msec. is representative, as it will be for the time interval betweenrecordingswithaCMOSarray.Inthisway,thepresentsystemisinferiortoaFP‐systemwithrespecttoresolution,butitsrealizationissimpler.

Although the speckle wavemeter cannot detect the wavelength change in the LEDbasedsystem,itcaneasilydetectthetemperaturedependentwavelengthchangewithin the laser‐based colormixing systemwith an accuracy of a picometer during theoperationtime.Knowingtheaveragewavelengthoftheincidentradiation,theshiftinwavelengthistacitlygivenbythephysicalparametersofthesetup,i.e.theangle(s)ofincidenceandthedistances.Theabsolutewavelengthoftheincidentradiationcanbecoarsely determined by the speckle size, but in real applications, the incidentwavelengthhastobeknown.Ifoneallowstwomeasurementsofthespecklepattern,betweenwhichsomephysicalparameterhasbeenchanged(e.g.distance,tiltofobject),theabsolutewavelengthmightbederivedwithahigheraccuracy.


5.6Summary

ThevariationsincolorpropertiesandlightintensityareduetotheuncertaintiesinLEDbinningandaging,operationalandenvironmentalconditions,etc.ThischapterreportsthedevelopmentofaMonteCarlosimulationprocesstoanalyzecolorvariationsofthemulti‐coloredLEDlightingsystem.Thisisneededfore.g.studiolightingwherevariationincolorfromlamptolampandstabilityoverbothshortandlongtimearecriticalandshouldbecontrolledtocertainlimits.Thedevelopmentofthesimulationisdoneforawhite tunable V8 light engine, using the experimentally measured sensitivities ofindividualcoloredLEDswithrespecttouncertaintiesininputparameterswhicharefedinto the simulation process. The Monte Carlo simulation provides the resultantuncertaintyofthechromaticityforthetunableCCTrangeoftheV8lightengine.Itisanalyzed in the Uniform Chromaticity Scale Diagram comparing the variations inchromaticitydifferenceswiththe“n‐stepu′v′circles”forvisuallynoticeablevariations(varyingfromn=3‐8circlescorrespondingtothen‐stepMacAdamellipses forthetargetCCTsettings3000Kto6000K)andwillalsobe importanttocomparetoTVcamerasensitivities.Thisinformationisusedinthecolorcontrolsysteminrealtimetonarrowdowntheuncertaintybyusingapre‐calibratedlookuptableandacalibratedcolor sensor for feedback. In future it is aiming to incorporate the uncertainty ofindividualLEDlifetimeandfailuremode.Thereisalsoaplantovalidatethecontrolsystemexperimentallyonthreesimilarkindsofsystems.TheMonteCarlomethodisused to evaluate theuncertaintiesof theoutputparameters for theV8 light engine,whicharefinallyusedasinputforthecolorcontrolsystemsothatdynamicallyoutputparameterscanbecontrolledwithinacceptablevariatione.g.withinajustnoticeablechromaticitydifferencecircleasdefinedinCIETN001:2014.Theresultsshowthatthestabilizedsystemdecreasesintheaverageuncertaintywithin95%coverageintervalonluminousfluxfrom8.5%to2.3%andsimilarlyreducestheuncertaintyontheCCTat4000Kfrom1020Kto125K.LowerreductionsintheCCTuncertaintiesareobservedathigherCCTs.Thepresentcontrolmechanismcontrolthelightwithinacertainlimitsothatthecontrolpoint intheCIE1976(u ,́v´)diagramwouldbewithinn=2circle,takingthetargetpointasthecenterofthatcircle.

Theeffectivenessoftheproposedcontrolmechanismisthatthecontrolmechanismcanuse inanySSLsystem.Therefore, insteadofusing thecolorsensor,anovelandinexpensive specklewavemeter isproposed touse in the laser colormixing systemwhichcanindicatethewavelengthvariationinthelasersystemasafeedbacktothecontrol unit. Thewavelength change canbe converted to (u ,́v´) for further analysispurpose.Itisexperimentallyverifiedthattheminimumvariationinwavelengthoftheorderofthepico‐metercanbedetectedbythespecklewavemeter.

The following chapter will elaborate on the experimental investigation and thecorrespondingresultsrelatedtotheV8lightengine.

Chapter 6 105 Experimental investigation of V8

ExperimentalinvestigationofV8lightengine

‘Atthebeginningofallexperimentalworkstandsthechoiceoftheappropriatetechniqueofinvestigation.’–WalterRudolfHess

Chapter6 :ExperimentalinvestigationofV8light

engine

6.1Introduction

The previous three chapters have elaborated on the V8 light engine and it’scharacteristicsasamulti‐coloredtunablewhitelightsource[138]–[143]whichcanbeappliedintostagelighting.ThischapterstudiestheexperimentalvalidityoftheV8lightenginedesignbyevaluatingthelightoutputfromtheprototypeofthedesign.Therewillbeacomparisonbetweentheluminaireswhichareavailableonthemarketforstagelighting application. The chapter also registers the experimental results and dataanalysisforcharacterizingthedifferentcomponentsusedfortheV8lightenginedesign.

6.2ForwardluminousfluxmeasurementforLED

This section will register the experimental results for the light output from theindividual5differenttypesofLEDprintcircuitboards,usingthe1mISaccordingtotheCIE standard [40] mentioned in the section of 1.5.1.1 in Chapter 1. The light wascoupledthroughthesideport(15cmdiameter)inaforwardfluxconfiguration(asseeninFigure6‐1)andaself‐absorptioncorrectionwasapplied.Theexperimentalresultsare compared to the respective datasheet values [103]–[105] and there is a goodagreementwith a ~7% variation. Figure 6‐2 represents the value of luminous fluxwhenthedrivingcurrentisrampingupwith50mAstep.Thefigureindicatesthattheincreased driving current almost linearly increases the luminous flux from theindividualLEDexceptforthegreenandwhiteLEDs.TheexperimentalresultsarelaterusedforevaluatingtheTIRlenslossdiscussedinthesection6.4.


Figure 6‐1: Experimental setup showing the IS port


Figure 6‐2: The relation between the driving current and the luminous flux from the individual LED at case temperature 25° C

Table6‐1representstheluminousfluxwhilethedrivingcurrentissetat350mA,500mA, 700mA and 1000mA, respectively, with a case temperature at 25 °C for theindividual LED. The temperature is maintained at 25 °C by a Peltier temperaturecontroller.ThesimilarexperimentsetupisshowninFigure6‐3.TochecktheLEDbinconsistency,anotherprintfromeachLEDtypehasbeenmeasured.ItwasobservedthattheredLEDhasmaximumvariation(~12%)betweentwoLEDprintcircuitboards.


Table 6‐1: Experimentally measured luminous flux @ 25° C case temperature

LEDtype LEDcolors

Luminousflux[lm]@350mA




LuxeonZ Royalblue 13 18 24 33LuxeonZ Blue 30 40 52 66CREEXPE Green 122 153 186 220CREEXPE Red 70 97 125 125(@700mA)CREEXPE2 White 102 136 174 218

Figure 6‐3: Experimental setup for forward flux measurement from LED at varying current when case temperature is at 25 °C

6.3Reflectancefromtheparabolicreflector

Theparabolicreflectoroftheprototypehasbeentestedforitsreflectancepropertyandtheuniformityofthesurface.Thereisarelativeluminousfluxmeasurementdonebyusing1mIS.Theusedcollimatedwhitesourcefortheexperimenthadknownluminousflux values which are measured through the forward flux measurement using IS(explained in 1.5.1.1 in Chapter 1) and then the same source was measured afterreflectingfromtheparabolicreflector.Theresultsforthosetwoinstancesarecomparedtofindoutthereflectancewhichisalreadyshowninthesection3.3.2inChapter3.Thesurface uniformity of the reflector has been analyzed by comparing the reflectancevaluesfromthedifferentpartsofthereflector.

6.4TIRlensloss

TIRlensisacriticalcomponentoftheV8lightengine.NormallytheLEDhasalargeviewingangle(~110°‐130°)andisdifficulttocollimate.However,TIRlensesareusedinthedesignoftheV8lightenginetogetcollimationoftheLEDbeamaftertransmittingfromtheTIRlens.TheopticallossfromtheTIRlenshasbeencharacterizedseparatelyfortwosizesoflenses.TheexperimentalsetupisseeninFigure6‐4.


Figure 6‐4: Experimental setup for the lens efficiency measurement inside the V8 light engine

Accordingtotheexperimentalsetup,theLEDbeamistransmittingthroughtheTIRlens, gets reflection by the parabolic reflector which is situated at the focal planedistance (reflector) from the lens position. The reflected beam passes through themicrolensarray(MA),placedatthegateandthenentersintotheIS.ThetransitionrailisusedtosetthelensatatimetothecenterofthegatefollowingtheringpositioninsidetheV8 lightengine(Figure3‐1(b)).Thus,thenearestpositionisconsidered32mmfromthecenterofthegate.ThethreeaxesmicrometertransitionalstageisusedsothattheLEDprintcircuitboardcanbeatthemiddleoftheTIRlensandtheTIRlenscansitontheprint.TheblueandthewhiteLEDsareusedfortheexperimentastheblueLEDprintcircuitboardissmallerthanthewhiteLEDprintcircuitboard.ThedrivingcurrentoftheLEDswas1000mA.Toevaluatethelensloss,luminousfluxvaluesoftheLEDsarecomparedandmeasuredappliedwithandwithouttheTIRlens.TheexperimentalresultsandrelatedanalysisarealreadydiscussedinChapter3,section3.3.3.TheopticalefficiencyinFigure3‐10showsthelossesduetothepositionofthelensfromthegate.However,theabovementionedlossismixedwiththereflectionlossfromthereflectoraswellastheraycone.Therayconeiscreatedbecausethereflectorissituated~170mmawayfromthegatehaving∅50mm.Therefore,tofindthelossbasedonlyontheTIRlensaswellastheraycone,anotherexperimentisarranged(Figure6‐5).


Figure 6‐5: Experimental setup for TIR lens loss measurement

TheFigure6‐6showstheopticalefficiencyobtainedfromtheTIRlens(biglenswithsize12.7±0.2mm)whilethelensholderinFigure6‐5isplacedataparticulardistanceaway from the IS (2cm, 14 cm, 19 cm, 29 cm, 39 cm and 58 cm) to produce thecorrespondingrayconeconsideringfivedifferentLEDs.With100°raycone,almostalltransmittedlightfromtheTIRlensenterintotheIS.Thereforetheopticalefficiencyis~80%forthefivedifferentLEDs.ThatincludestheFresnellossfromtheTIRlensandtheraylossduetothegeometryoftheTIRlens(asseeninFigure3‐8).Asweknow,theopticalefficiencystartsdecreasingastherayconedecreases.Fromarayconeof<20°,theopticalefficiencyforthegreen,thered,andthewhiteLEDsarelessthantheroyalblueandtheblueLEDs.ThisisbecausetheLEDprintcircuitboardissmallerfortheblueandtheroyalblueLEDsthanotherthreeLEDs.AccordingtotheV8lightenginedesign,araycone~15°isexpectedandshowsgoodagreementwithFigure3‐10(b)when theTIR lens is situatedat the first ring. Similarly, theTIR losshasbeenalsocharacterizedforthesmalllenssize(11.7±0.2mm).


Figure 6‐6: Loss characterization from the TIR lens

6.4.1TIRlenstolerance

TherewasathoroughinvestigationonthetoleranceoftheTIRlens.Byusingthethreeaxesmicrometertransitionalstage(seeninFigure6‐6),theoptimizedpositionoftheTIRlensforthreeaxeswithrespecttotheblueLEDprintcircuitboardisfoundwhichcorresponds to the maximum luminous flux transmitted from the TIR lens. Thetolerance of one side for the X, Y, and Z axes are 0.51mm, 0.4mm, and 0.36mm,respectively, (Figure 6‐7) while the transmission loss is 10% of FullWidth at HalfMaximum(FWHM).Similarly, thewhiteLEDprintcircuitboardshowsthetolerancevaluesfortheX,Y,andZaxes0.64mm,0.6mm,and0.64mm,respectively,whilethetransmissionlossis10%ofFWHM.

Figure 6‐7: TIR lens tolerance for the optical efficiency loss

6.5Colorsensorandrelatedmeasurement

ThecolorsensorusedforthecontrolmechanismisadevelopmentkitDKMTCS‐INT‐AB4ofMAZeT(Figure6‐8),andincludestheOEMMTCS‐INT‐AB4sensorboard,fittedwith optical interface, a standard I²C‐to‐USB converter, as well as software. Thesoftwareenablestocontroltheconverter,calibratethesensorandallowsdataloggingoptions[144],[145].TheOEMsensorboardMTCS‐INT‐AB4isasmallPCBforgeneral


colormeasurementandcontrolapplications, includesaTrueColorsensor(MTCSiCFXYZcolorsensor)basedontheinterferencesensortechnology,havingfunctionalityoncolorteaching,relativecolormeasurement,andabsolutecolormeasurement.ThecolormeasurementbasedonDIN5033colorimetrywithinternationalstandardobserverCIE1931 spectral function, a special analog‐to‐digital converterMCDC04EQwith a highdynamicrange(1:1,000,000),anEPROMforsensordata,anintegratedtemperaturesensor,2LDOmicropowerregulators(LT1761ES5‐3,LinearTechnology)tomanagetheanalog,anddigitalsupplyvoltage.I²Cisrequiredfortheexternalcommunication,configurationofthesensor,readoutofthesensordata,aswellaswritingandreadingofthememory.TheOEMsensorboardusesastandardI²C‐to‐USBconverter for initialtestingreasonsviaUSBandPCsoftware.TheconverterIOW24‐DGisconnectedtothesensor board via cables and controls uncritical functions (not synchronized) of thesensorboardviaPC.

Figure 6‐8: Jencolor (DK MICS‐INT‐AB4) true color sensor from MaZet

6.5.1Calibrationofthecolorsensor

IntheV8application,thecolorsensorisconnectedthroughUSBcableandmonitorsthegeneratedwhitelightfromtheV8lightengineandprovidesthecorrespondingdatatothecontrolmechanismoftheV8lightengineforprocessinginternally(asseeninFigure5‐11).However,thereadingfromthecolorsensoriscalibratedwiththehandheldGS‐1150 spectrometer[118]. Initially, thedifferentmonochromatic aswell as thewhitelight(coveringthewarm,neutral,andcoolwhite lightregions)aremeasuredbythehandheldGS‐1150spectrometer(intotal32setsofmeasurements)andcorresponding(x,y)coordinatesofmeasuredlightareplottedintheCIE1931diagraminFigure6‐9.Simultaneously,thesamelightismeasuredbycolorsensoralso.Figure6‐10showstheexperimentalsetupforthecolorsensorcalibration.Inthefigure,thelightboxisabletoprovide the desired light (monochromatic/white) which is seen by a 2° observerwindow(accordingtoCIE1931)ofthespectrometeraswellasthecolorsensor.TheilluminatingplanewasthesameforthespectrometeraswellasthecolorsensorandthelightboxisprovidingtheuniformlightthroughouttheilluminatingplanesothatthespectrometerandthecolorsensorcanseethesameLIDcomingfromthelightbox.ThemeasureddatabythespectrometeraswellasbythecolorsensorareloggedintothePCthroughtheUSBcablesforthefurtherprocess.


Figure 6‐9: (x,y) co‐ordinate in CIE 1931 diagram for inspected light

Figure 6‐10: Experimental setup for the calibration of color sensor


Tofindoutthecalibrationmatrixforthecolorsensorthefollowingequations(53)‐(57)need tobe abided. Let us assume thatXs,Ys, andZs are the tristimulus valuesmeasured by the spectrometer for measured light which can be related to thetristimulusvalues(XJ,YJ,andZJ)measuredbythecolorsensoraftermultiplyingbythecalibrationmatrixA (assumed linear relation) as seen in equation (53), where thecoefficientofthecalibrationmatrixisdefinedinequation(54).

∗

(53)

,

(54)

Nowtodothecalibration,wehavetofindoutthecoefficientofthecalibrationmatrix.Three calibration points are chosen from the 32 sets of light measurements andcorrespondingtristimulusvaluesarenotedfromthespectrometeraswellasfromthecolorsensor.Inequation(55)SandJrepresentthethreesetsoftristimulusvaluesfromthespectrometeraswellfromthecolorsensor,respectively,whereasAisthesameforeachcase.Therefore,Acanbeevaluatedbytheequation(56).FinallythetestvectorTcanbedeterminedbyusingequation(57),whileTrepresentsthetristimulusvaluesforanylightmeasuredbythecolorsensor.

∗

,

(55)

∴ ∗ (56)


∴ ∗ ∗ ∗

(57)

Since the controlmechanism evaluates the color characteristics according to CIE1976,aftercalibratingthecolorsenorreading,themeasuredlightsarevalidatedandplottedintheCIE1976diagram(Figure6‐11).IntheFigure6‐11(a)‐(c),‘◊’representsthe threecalibratingpoints tocreate thecalibrationmatrixwhilecharacterizinganywhitelight.Byusingthatcalibrationmatrix,thecolorsensorreadingsarecalibrated,(u ,́v´)coordinatesareevaluatedandplottedintheCIE1976diagramalongwiththereadingforthesamemeasuredlightsfromthespectrometer.AsseenintheFigure6‐11(b)‐(c), with the three calibrating points, the measured white light from thespectrometer as well as the color sensor are in good agreement with each other(averagevariation~2.7x10‐6)andtheyarewithinthejustnoticeabledifference(JND)circle[46].However,forthecharacterizationofmonochromaticlight,thesecalibrationpointsarenotgood(Figure6‐11(a))asthespectrometricandcolorsensorarenotingoodagreement.Therefore,itisrequiredtofindanotherthreecalibrationpointsforthemonochromatic light to avoid the error due to a spectralmismatch. There is ~6%measurementdiscrepancybetweenthespectrometerandthecolorsensor(asseeninFigure6‐11(d))byusingthethreecalibrationpointsindicatedinFigure6‐11(a)‐(c).AppendixDwilldescribe themethod if32measurementpointshavebeen taken toevaluatethesamecalibrationmatrix.

Figure 6‐11: (u´,v´) co‐ordinate in the CIE 1976 diagram after measuring the spectrometer and the color sensor (a) monochromatic light (b) mixed light (c), and white light; (d) characterizing the points

according to JND definition [46]


6.5.2Colorsensordynamicgainsettings

Thecolorsensorgainsettingsareacriticalaspectforthemeasurementoftristimulusvaluesaseachsettingcorrespondstoaparticularcalibrationmatrix.Thereforethegainsettingsofthecolorsensorhaveinfluenceonthemeasurementduringtheoperationaltime. There are two ways of changing the gain settings of the color sensor, whichultimately change the reading of ADC count (X, Y, and Z). Table 6‐2 represents theavailablereferencecurrentandtheintegrationtimesettingsforthecolorsensorwhichareresponsibleforthegainsettings.Foracertainreferencecurrent,ADCcount(X,YandZ)increasesiftheintegrationtimeincreases.However,theADCcount(X,YandZ)decreases if the reference current increases at integration time. Themaximum andminimumADCcountforthesettingsis65536and0.TheoperatingrangeoftheADCcountbetween65000and30aregood,ascounts<30and>65000havetheprobabilityofdetectingtheerrorinthemeasurement.Figure6‐12representstheADCcounts(X,Y,andZ)fordifferentgainsettings.

Table 6‐2: Gain settings for the color sensor

Referencecurrent[nA] Integrationtime[ms]20 6480 128320 2561280 5125120 1024

In the figure, theyaxis represents theADCcountwhile thex axis represents thelogarithmicscaleofthereferencecurrent.ThelinewithasolidsquarerepresentstheXADCcount,thedash‐dottedlinewithasolidcirclerepresentstheYADCcountandthedashedlinewithasolidtrianglerepresentstheZADCcount.ThedifferentcolorinthelegendrepresentsthedifferentsetofX,YandZADCcountsfordifferent integrationtime. For an example, theblack color represents theADCX,Y andZ counts for theintegrationtime=64ms.

Figure 6‐12: For different sets of integration time, the ADC X,Y and Z counts variation on reference current for the color sensor

Asseeninthefigure, irrespectiveofanysettings,theZADCcountisalwayslowerthantheXandYADCcount.SincetheADCcountdependsonthelightintensityincident


onthecolorsensor,itiscriticaltosetthegain‐settingsattheoptimumconditionsothattheADCcountcannotreachsaturationorcannotcount<30.Tofacilitatethedynamicgainsettingsaccordingtothelightintensityincidentonthecolorsensor,softwarecodesarewritteninMATLABandincorporatedintothecontrolunitofV8lightengine.Thethorough investigation has been done on the ADC count depending on the lightintensity. Therefore there will be a default settings (reference current: 80 nA andintegration time: 128ms) forwhen themeasurementwill start. The correspondingcalibrationmatrixwill be taken during the calculation of tristimulus values of light(equation(57)).IfthelightintensityishighenoughtoreachthesaturationintheADCcount,thecontrolunitchangestoahigherreferencecurrentorlowertheintegrationtimeaccordingtothesituationandtheactionwillbeviceversa forthe lowlighting.However,foreachcase,thecalibrationmatrixwillbedifferent.Thestandarddeviationforthemaximumerroroccursduetodifferentsettingsis~0.0004.

6.6Microlensarraycharacterization

Thesurfacequality lookingatpreandaftercoatingprocessresultsof the fabricatedlenslets aremeasuredandanalyzedby3D confocal and interferometricmicroscopy(SensofarPlu‐neox).Confocalmicroscopy(Figure6‐13(a)) isused inthisapplicationbecause it offers several advantages over conventional optical microscopy, whichincludes shallow depth of field, elimination of out‐of‐focus glare, and the ability tocollectserialopticalsectionsfromthickspecimens.Thereforeconfocalmicroscopyisanoptical imaging technique for increasing optical resolution and contrast of amicro‐graph by introducing a spatial pinhole placed at the confocal plane of the lens toeliminate out‐of‐focus light. By this way, within a thick object, it enables thereconstruction of 3D structures from the acquired images by accumulating sets ofimagesatdifferentdepths.

Thewhitelightinterferometricmicroscopesystem(Figure6‐13(b))exploitsamethodknownas two‐beam interferometry tomeasuresurfaceprofiles.Twobeamsof lightinterferewitheachotherastheterm"two‐beaminterference"suggests.Thelightfromthelightsourcedividesintwowhenithitsahalfmirror.Onepartreflectsoffthemirrorand the other transmits through it. The reflected light then travels to the referencemirror,whilethelightthattransmitsthroughtravelstothematerialunderobservationon the stage. However, both beams of light bounces off the respective mentionedobjects.Thebeamsinterferewitheachothertoformanimageinthecamera,wherethey produce striped patterns known as interference fringes. This microscopicmeasurement is used to get bright contrast between the interference fringeswhichhelpstomeasureaccuratelytheheightoftheopticalcomponent.


Figure 6‐13: Basic principle of (a) Confocal microscopy (Ref.:[146]); (b) Interferometric microscopy (Ref.:[147])

The analytical results for surface quality and the quantitativemeasurement on lensparametersareshowninthe6.6.1section.Theopticalperformancewithregardtotheapplicationwasinvestigatedbymeasuringtheangulardependenceofthetransmissionand investigating the light scattering effect. Figure 6‐14 illustrates the setup forsimulatingthetransmittedlightasafunctionofincidentconeanddifferentexitcones.Italsodescribes the transmissionmeasurement,where a collimatedwhite light beamwithabeamdiameterof10mmfromafibercoupledsource(OceanOpticsHL2000)wasused.Theincidenceanglewasvariedovertheinterval±45°.Thetransmittedlightwasmeasuredas a functionof incidence angleusing a6 inch integrating sphere ascollectionoptics.TheMAwasmountedatthe1inchdiameterportopeningandtheexitconehalfangleisnearly90°,whichcorrespondstothetoppartofthesimulationresultsshowninFigure3‐16.TheresultsfromthemeasurementareshowninFigure6‐17.Thetransmittedlightwasmeasuredusingaspectrometer(QE65000)fibercoupledtothesphere. Further, the light scattering from the surface of the MA is investigated bylookingatthereflectedlaserlightofasinglelensletanddiscussedinthe6.6.2section.


Figure 6‐14: Sketch illustrating the simulation and experimental set‐up for measuring angular transmission and acceptance angle of the MA where dashed line represents the optical axis (Ref.

Figure 6 of Chakrabarti et. al [13])

6.6.1Surfacetopology

Thepropertiesoftheinjectionmoldedsamplesfromtheuncoatedandthecoatedmoldtooliscompared.TheconfocalmicroscopymeasurementwasdoneusingtheobjectiveofNikon50xwithNA0.95.The3Dimageshavebeentakenfor5consecutivelensletsbythemethodofstitching3x3imageswiththeconditionof25%overlap.Fromeachofthetwomicrolensarrays5individuallenseshavebeenanalyzed.Thediameterofthelensarrayincludingtheouterringis85mmasshowninFigure6‐15(a).Thepositionofthecentrallenswas39mmfromtheedgeoftheperimeteroftheouterringandthefiveconsecutivelensletsweretakenforthemeasurementsfromthatregion.ThediameterofthehexagonallensesismeasuredatanedgetoedgedistancewhichismeasuredbyFouriertransformoftheimage(20xmicroscopeimages)toidentifyaunitcellorlenslet.Thus,thewholeimageisusedforthemeasurement(seeninFigure6‐15(b)).Thetotalheightofthelensismeasuredfromthestitchedimage(1x850x)asseeninFigure6‐15(c)wherethecenterandtheedgeofthesinglelensletisvisible.Theverticaldistancefromtheedgetothecenterofthelensletgivestheheight.Theradiusofcurvatureismeasuredforeachlensbyfittingaspheretothe3Dsurface.Theaberrationfromthissphereisalsoinvestigatedasa3Dimage.Theroughnessofthesurfaceisalsomeasuredaftersuitablewavinessfiltration.Table6‐3summarizesthemeasuredresultsforthelensparameters.Thestandarddeviationbetweenallmeasurementsfortheradiusofcurvature is 35µm. The table shows the repeatability and reproducibility of thefabricationprocesswithin±50µmaccuracy.Thevaluescorresponding touncoatedandcoatedmoldareequalwithintheuncertaintiesandshowsthatthecoatingprocessdoesnotaffectthelensletparameters.


Figure 6‐15: (a) Photo of the injection molded sample; (b) Magnified view of the selected portion of the five lenslets for measuring the lens width; (c) Magnified single lenslet for height measurement

(Ref. Figure 7 of Chakrabarti et. al [13])

Table 6‐3 : Measure Single Lens Parameters by 3D Microscopy (Ref. Table 2 of Chakrabarti et. al [13])

Microlensarraytype Radiusofcurvatureforlenslet[mm] Sag[mm] lensletwidth[mm]Uncoated 0.863±0.040 0.503±0.001 1.457±0.005Coated 0.869±0.035 0.492±0.001 1.46±0.005

Thesurfacequalityofthelenslethasbeeninvestigatedthroughsurfacetopographymeasurements.Figure6‐16comparestheimagesofthetopsurfacesoflensletfromtheuncoated(left)andcoated(right)mold,afterfittingwithasphereofradius846.1µm.ThesurfaceroughnessiscalculatedbymeasuringRa,Rq,andRz[148]whichareshowninFigure6‐16.Thereissomecastdefectoccurrenceattheedgeoftheperimeterofthearray.However,thosedefectsdidnot influencethelightpath. It isobservedthatthephotoresistcoatingofthemoldimprovesthesurfaceroughnessoftheinjectionmoldedmicrolensarrays,inreducingtheRa,Rq,andRzbyfactorsof2,2.5,and2.6.Itisseenfromthecross‐sections(Figure6‐16(c)and(d))thatthemaximumvariationisreducedfrom1.2µmto0.2µmforthesespecificscans.


Figure 6‐16: Topography image with height contours for each 2 µm in the z‐direction for microlens lenslet from (a) (shown 600 x 600 µm

2 area) uncoated and (b) (shown 400 x 400 µm

2 area) coated

mold and corresponding surface variation along cross section in (c) and (d) (Ref. Figure 8 of Chakrabarti et. al [13])

Table 6-4: Roughness Measurement According to ISO 4287:1997 [148]

MAtype Ra*a[nm] Rq*b[nm] Rz*c[nm]Uncoated 75±29 110±56 441±194coated 38±1 42±2 173±26

*aRa:arithmeticaverageofabsolutevalueforsurfaceroughness;*bRq:rootmeansquare(rms)averageoftheroughnessprofileordinates; *cRz:themeanroughnessdepthwhich is the arithmetic mean value of the single roughness depths of consecutivesamplinglengths

6.6.2WorkingperformanceofMA

TheworkingperformanceoftheinjectionmoldedMAfromthecoatedmoldhasbeeninvestigatedthroughmeasurementoftheangulartransmission,Figure6‐17showsthemeasured transmission as a function of the incidence angle and the correspondingsimulationisshownaswell.ThisisusedtodeterminetheacceptanceangleoftheMA.Figure6‐17showsthecorrespondingresultsforanotherMAwithD=1.2mmusedinthefirstprototypeoftheLEDlightengine.Theacceptanceangleforthisis37°.FortheoptimizeddesignoftheMAproducedinthiswork,thenumericalaperture(NA)hasbeenincreasedbyincreasingthewidthDtoavalueof1.45mm.ThemeasurementforthisinFigure6‐17,showsamaximumtransmissionof~92%whichisalmostconstantwithin±40°,henceanacceptanceangleof40°.TherefractiveindexofPMMAis1.489.Thus, the Fresnel reflection loss per interface for 2.1 mm MA is ~3.9 % andtransmission from the array was ~ 92%. Again, the calculated acceptance angleaccordingtoequation(35)is~41°.Hence,thereisagoodcorrespondencebetweensimulationandmeasurementsontheproducedMA.Duetomachineaccuracyinlensparameters;thereisadiscrepancyof~2°inacceptanceanglebetweenthesimulation


andtheexperiment.Theincreasingmeasuredandsimulatedefficiencyatanglelargerthan±40°correspondstothecouplingoflightthroughtheneighborlenslets,illustratedinFigure3‐15(b).

Figure 6‐17: Comparison of MA by both simulation (Sim.) and experiment (Exp.) (Ref. Figure 9 of Chakrabarti et. al [13])

Figure6‐18showstheresultsofthevisualandlightscatteringinvestigationofthesurfacequality.The investigation is on a single lenslet,where the images to the leftshowsphotosof the topsurface takenbyopticalmicroscopeandthe imagesontherightarephotosof thereflected lightwhena lenslet is illuminatedbyaHe‐Ne laserbeam.TheimagesinFigure6‐18(a)arefortheinjectionmoldedMAfromtheuncoatedmold. Irregularitiesareseenonthemicroscope imageandthereflectedHe‐Ne laserbeam isstronglydistortedandscattered.Thissurfacequality isnotgoodenough toconsiderforthemicrolensapplication.TheMAproducedfromthephotoresistcoatedmold,hasamuchimprovedsurfacequalityasshowninFigure6‐18(b),therearemuchfewerirregularitiesinthemicroscopeimageandthetransmittedHe‐Nelaserbeamisnearlyundistortedandlessscatteredlighthasbeenobserved.Thissurfacequality isapplicablefortheLEDlightengineapplicationanda5%increaseoftheefficiencyofthiswasobservedwhenusingtheMAproducedfromthecoatedmoldcomparedtotheonefromtheuncoatedmold.

Figure 6‐18: Optical microscopic images of a single lenslet (left side) and photo of transmitted He‐Ne laser beam from a single lenslet (right side) in injection molded MA from the (a) uncoated mold and

(b) photoresist coated mold (Ref. Figure 10 of Chakrabarti et. al [13])


6.7Measurementoftheprototype

Figure6‐19showstheprototypeoftheV8lightengineinoperation.ThissectionwillcharacterizethebeampropertiesofthelightenginesuchasLID,totalluminousflux,CRIscores,beamhomogeneity,lightoutputstabilityintermsofoperationaltimeaswellastemperature,etc.fortheprototype.

Figure 6‐19: White light output is coming out from the prototype of V8 light engine

6.7.1ForwardfluxandLIDmeasurementfromtheprototype

Initially the luminous flux from each of the colored LEDs are measured from theprototypebyforwardfluxmeasurementusing1mISaccordingtotheCIEstandard[40].Figure6‐20representsthetotalluminousfluxobtainedfromthecorrespondingDMXvaluesfortheindividualcolorLEDsinsidetheprototype.Asseeninthefigure,exceptfortheredLEDs;allotherLEDsachievingsaturationintheluminousfluxafterreachingaDMXvalueof90%.Thisisbecause,ata90%DMXlevel,thecurrentlevelforallcoloredLEDs(excepttheredLED)reachsaturationpointat~984mA(Figure6‐21).Therefore, during the operational time of the V8 light engine, it is safer to use themaximum90%ofDMXforallLEDs.ItisalsonoticedinFigure6‐21thattherelationbetweenthecurrentandDMXarenotsamefor theredLEDasothercoloredLEDs.Therefore, the uncertainty of the luminous flux for the variation of DMX values isconsideredintheMonteCarlosimulationinsectionof5.2inChapter5.


Figure 6‐20: Increased luminous flux with increasing DMX for all; (a) the royal blue, (b) the blue, (c) the green, (d) the red, (e) the white LED with reflector loss at 30%


Figure 6‐21: Change in current with a changing DMX for the individual colored LEDs

Table 6‐5 compares the measurement of luminous flux at 100% DMX from theindividual colored LEDs transmitted from the TIR lens using 2 m IS with themeasurementofluminousfluxat100%DMXfromtheindividualcoloredLEDsfromtheprototypeusing1mIS.Totallossinsidetheprototypeis20316lm.However,thelosswasdecreasedto18153lmwhenthereflectorwasmodified.Theopticalefficiencyof the prototype was ~42%. The position of the LEDs, with respect to the gate,influencesthelossintheprototype,whichisdiscussedinsection3.3inChapter3.TheprintsizeoftheroyalblueandtheblueLEDaresmall.ThereforethelossfromtheTIRlensislessforthoseLEDscomparedtoothers.

Table 6‐5: Total luminous flux for the individual color LEDs from the prototype

LEDs LEDnumbers

Luminousflux(lm)@individualcolorLEDs+TIRlensat100%

DMX

Luminousflux(lm)@individualcolorLEDsat

100%DMXforprototype

Loss(%)forindividualcolor

LEDs

Royalblue 18 466 238 51Blue 24 1396 688 49Red 12 1224 352 29Green 30 5597 1675 30White 126 22960 8375 36Total 210 31643 11327 39

Theprototypewasnot able to produce the required luminous flux (~20,000 lm)fromtheV8lightengine.Therefore,theoptimizationoftheV8lightenginedesignwasneededand theoptimization stepsareelaboratedon inChapter3, section3.3. TheluminousfluxvaluescanbeincreasedafteroptimizationofMAandTIRlenswhichareshowninTable6‐6.TheincreaseinnumberofLEDsfrom210to288alsoincreasesthefinallightoutputby56.5%comparedtotheprototype.However,theoverallsizeand


weightwouldincreaseaccordinglyby25%and27%,respectively.Inadditiontothisthe mechanical design difficulty increases since making such a large reflector iscomplicatedandcostly.

Table 6‐6: Improvement in luminous flux after optimization in simulation

Luminousflux(lm)(Prototype)

Luminousflux(lm)(afterMA

Optimization)

Luminousflux(lm)(afterTIRlenssizeOptimization)

Luminousflux(lm)

(288LEDs)

11327 16624 22284 25990Improvement 5297 5660 3706

Duetotimerestrictionsaswellasthebudgetarylimitation,thesecondprototypewasnotmadeafterthedesignoptimization.Figure6‐22showstheexperimentalvalidationwithrespecttothesimulationwheretheopticalefficiencyofthewhiteLEDtransmittedthrough the TIR lens is characterized inside the optical system at different ringpositions.Theaveragedeviation inopticalefficiencyof the lightenginebetween thesimulationandexperimentalresultsis~9.7%,whereasthemeasurementuncertaintyfortheexperimentis~6%.

Figure 6‐22: The comparison between simulation and experimental results considering the optical efficiency of white LED through the TIR lens at each ring position

Theangulardistributionof the luminous intensityof theprototypewasmeasuredusinggoniospectrometers;ahorizontalc–typemanuallycontrolledgoniospectrometer,which is equippedwith an optical probewithdiffuser (EOP‐146) for the irradiance


measurement.ThedetectorwasfixedinspaceandtheLEDsystemwasrotatingwithaspecific angle. The measurement distances was 8.65 m and the diameter of theluminous area, i.e. the hemisphere of the optical system, was D = 125 mm. Themeasurementdistanceswas68xDandthus,themeasurementwasdoneinthefarfield[40].Forthe100%DMXsetting,themeasurementsweredoneforthefloodandspotposition of theV8 light engine. The experimental results are plotted in polar graph(Figure6‐23).

(a)

(b)

Figure 6‐23: The LID for (a) spot; (b) flood position of the V8 light engine

6.7.2Beamhomogeneity

Figure6‐24(a)representstheexperimentalresultsforlightintensityatfloodpositionfromtheprototypemeasuredbyPRCphotometer,whiletheprojectedlightdistancewas 5 m and the manually controlled gonio‐spectrometer respectively. Both themeasurementshowstheFWHMofthebeamsizeis~60°andtheuniformintensityis


foundwithin ±20° (Figure 6‐24 (a)). There is 25%variation in intensitywithin theentire beam size (FWHM) in the PRC measurement. The data point for themeasurementbygonio‐spectrometerwaslessandthustheintensityvariationwasnotobserved.ThehandheldspectrometerisusedtoinvestigatethebeamhomogeneityinCCTandCRI (Figure6‐24 (b)), andalso inu ,́ v ́parameters (Figure6‐24 (c)). ThemeasurementshowstheuniformityinCCT,CRI,u ,́andv ́arewithin±25°.Duetothemeasurementuncertaintyoflowlightlevel,thebothendofthegraphsinFigure6‐24(b)andFigure6‐24(c)showthelargevariations.

Figure 6‐24: (a) Distribution of light intensity with respect to beam size (b) Distribution of CRI and CCT with respect to beam size (c) CIE 1976 (u´, v´) with respect to beam angle (Ref. figure 8 of Chakrabarti

et. al [14])

ThespotpositionfortheV8lightengineisfoundbyplacingthelightengineatthefocalplaneoftheFresnellens.Toobtainfloodposition,theFresnellensispositionedascloseaspossibletotheMA.Thepictureistakenbyprojectingthelightonawhitepieceof cardboard for the spot and for the floodposition to check the visual appearance(Figure 6‐25). It looksuniform to thenaked eye.However, the shapeof the spot ishexagonalduetothehexagonalMA,whichisalreadyseenbythesimulation(Figure3‐23).Theoutlooksection7.2.1inChapter7willdiscussthisissuefurther.


Figure 6‐25: (a) the spot; (b) the flood from the V8 light engine

Insteadofusing theMA in theV8 lightengine, thecolormixing fromthediffuser(Gaussian laser beam shaper LBH(DG)‐GT‐DV84° at 633 nm from a s) was alsoexperimentedonbyusingthesamesetupasmentionedinFigure6‐14.Thediffuserisin formof the2D–refractivemicro‐opticarray for transformingthebeamintensitydistribution into the Gaussian beam. The beam divergence of the diffuserwas 84°.Although,theoutputfromtheV8lightengineafterusingthediffuserdidnotshowthehexagonalspotatthescreen,luminousfluxwaslesscomparedtotheMA.Figure6‐26representsthetransmissioncomparisonbetweentheMAandthediffuser.Incaseofthediffuser,thetransmissiondecreasesgraduallyfromthepeaktransmissionvaluewhichisnotdesirableforthebeamhomogeneity.

Figure 6‐26: Light transmission comparison between the MA and the beam shaper (diffuser)

6.7.3CRI&CQSscore

Asmentionedearlier,theCRImatricesscoresarehighfortheV8lightengine;Figure6‐27representsthemetricscoresforfourCCTpositions.ItisseeninthefigurethatCRI(Ra)aswellasCQSare>85foreachCCTvaluesandthelowestCRI(R9)is41at3000K.CRI(R10)isalso~80foreachCCT.Accordingtotheapplication,themeasuredmetric


scoresateachCCThavegoodvalues.ThegoodvalueofCRI(R9)rendersthestrongredcontentswhilstCRI(R10)isgoodforrenderingyellowobjects[120],[149]–[154].

Figure 6‐27: Metric score for four tuned CCT positions (3000 K, 4500 K, 4600 K and 4800 K )

6.7.4Warmuptime

The temperaturedependencyof theprototypewas investigatedduring thewarmuptime (Figure6‐28).Theprototypewas stable for25minutes after switchingon thedrivercontrolunit.Fromthefigureitisobservedthattheluminousfluxwasdecreasedby~1.5%forincreasingofthetemperatureby9%.

Figure 6‐28: The temperature dependency of the luminous flux during warmup time


6.7.5Temperaturesensitivity

The color characteristics stability (e.g CCT, Duv, CRI(Ra), and CRI(R9)) with varyingtemperature from61°Cto35°CwasexaminedwhenallLEDs insidetheprototypewere set at 100% DMX (Figure 6‐29). Temperature sensitivities for CCT, Duv arecalculatedfromthismeasurementwhich isusedasuncertainties intheMonteCarlosimulationinsection5.2inChapter5.

Figure 6‐29: Variation of color characteristics with varying temperature

6.7.6Flickeringtest

There was no temporal periodic intensity variation observed during the intensitydimmingofthecolorchannelintheV8lightengine.Theexperimentalanalysishasbeendone by feeding the measured signal from the detector to an oscilloscope(Scopein@Box).


6.7.7Comparisonwiththetraditionallighting

The measured light output from the Halogen‐Fresnel traditional lamp (2 kW) iscomparedwith the lightoutput from theprototypeof theV8 lightengine (0.6kW).Table 6‐7 shows the comparison between the two lighting systems. Although theHalogen‐Fresnellampisproducing~34%moreluminousfluxcomparedtotheV8lightengine,itconsumes70%moreelectricalenergy.Duetohavingawiderbeam,theV8lightenginehaslessintensitythantheother.However,afterimplementingthedesignmodification,thefinalproductoftheV8lightenginewithalowelectricalconsumption(~0.9kW)willprovide~20,000lmwhichiscomparabletothelightoutputfromtheHalogen‐Fresnel.Moreover,thetransitionfromfloodtospotinthecaseoftheHalogen–Fresnel lamp lose ~37% optical efficiency, whereas the loss in optical efficiency is~10%fortheV8lightengine.Figure6‐30showsthecomparisonofnormalizedlightintensitydistribution(LID)betweentheHalogen‐FresnellampandtheV8lightengine.AsseeninFigure6‐30(a),thedistributionsoftheHalogen‐FresnellampandtheV8lightenginearesimilar forthe floodposition.However, forthespotposition, theV8 lightengineprovidesabroaderlightoutput(~±3°)thantheHalogen‐Fresnellamp.

Table 6-7: Measured total ΦV, luminous efficacy, intensity and beam angle for the 2kW Halogen-Fresnel spotlight and the V8 prototype light engine; and corresponding values

for the optimized simulated V8 light engine (Ref. Table 4 of Chakrabarti et. al [14])

Lightingsystem Fresnellens

adjustment

ΦV[lm] Luminousefficacy[lm/W]

Intensity[cd]

FWHM@50%

[degree]

Fieldangle[degree]

Halogen‐Fresnelspotlight

Flood 18745 10 33902 50.6 69

V8prototype Flood 12350 19 15680 55 70.8V8simulated Flood 22284 33 28292 ‐ ‐

Halogen‐Fresnelspotlight

Spot 11777 7 237273 10.8 24

V8prototype Spot 11100 17 115494 16.2 29.2V8simulated Spot 20060 30 208721 ‐ ‐

Figure 6‐30: LID comparison between the Halogen‐Fresnel lamp and the V8 light engine for (a) the flood; (b) the spot position, respectively


6.8Summary

The experimental investigation results in the light output from the prototype areanalyzed and discussed thoroughlywithin this chapter. The optical efficiency of theprototypewas~39%.Theopticalefficiencywasenhancedaftertheoptimizationoftheoptical components of the V8 light engine. However, the final simulated opticalefficiencyfromtheV8lightengineis~56%,whereastheaveragedeviationbetweenthesimulation and themeasured value is~9.7%. The results indicate that the beam isuniforminintensityaswellasincolor.TheCRI(Ra)isfound>85andCRI(R9)>40forwarm,neutral,andcoolwhite.ThecomparisonbetweentheV8lightengineandtheHalogen‐Fresnel lampshowsthattheV8lightengineloses10%opticalefficiencybychangingthepositionfromspot to flood,whereastheHalogen‐Fresnel loses37%ofoptical efficiency for the same. The LIDs for the flood position are similar for theHalogen‐Fresnel lamp and the V8 light engine. However, the V8 light engine has abroaderspectrumthantheHalogen‐Fresnellamp.

Chapter 7 135 Conclusion and outlook

Conclusionandoutlook

‘Ithinkandthinkformonthsandyears, Ninety‐nine times, theconclusion is false. Thehundredth time I am right.’–AlbertEinstein

Chapter7 :Conclusionsandoutlook

7.1Conclusions

Thegoalofthethesisistodescribeanovelopticaldesignandcontrolofmulti‐coloredLEDlightingsystemforstagelightingapplicationswhichwascarriedoutwithinthreeyears (2013 – 2015) at the premises of DTU Fotonik, department of PhotonicsEngineering incollaborationwiththeDanishcompanyBrother,Brother&SonsAps,fundedbyTheDanishNationalAdvancedTechnologyFoundationin2012.

Theobjectivesoftheprojectweretodesign,develop,andexperimentallyexamineanopticalsystemthatproducescollimatedtunablewhite light fromthecombinationofmulti‐coloredLEDsclusterintoasingleenergyefficientfocusablebeam.TheintelligentcolorcontroloftheindividualLEDsinsidetheopticalsystemwouldbeabletoprovideagoodqualityoflightoutputfromtheopticalsystemsuchas:highluminousflux,highcolor rendering, and light uniformity in intensity as well as in correlated colortemperature(CCT).Thecolorvariationaswellasthelightintensityvariationfromthelightingsystemneedstobewithinajustnoticeabledifference(JND)circlesothattheprobabilityof50%ofthehumanvisualperceptiondoesnotnoticethecolorvariationsofthelightoutputfromthesystemduringtheoperationaltime.Therefore,thelightingsystemshouldbecapableofreplacingtheFresnellensspotlightHalogenlamp(2kW)aswellasthecommerciallyavailableLEDluminaire(~160W)whichhasapplicationsinstagelighting,theaterlighting,TVstudiolighting,etc.

Beingconsciousoftheaimoftheproject,Chapter1describesthestate‐of‐the‐artofthepresentLED lighting technology aswell as themeasuring techniqueof theLEDlightingwhichisusedduringtheprojectwork.0elaboratesonthestate‐of‐the‐artofthetraditionalstagelightingandprovidestheexperimentalresultsonthelightoutputfromtheHalogen‐FresnellampandthecommerciallyavailableLEDluminaire.

FromChapter 3 onward the thesis reports on the simulated design of an opticalsystem,called theV8 lightengine,whichcanaddresstherequirementof theprojectgoal. The simulated design is subsequently demonstrated by making a prototype.Keepinginmindthegoalofachieving~20,000 lmof luminousflux,oneLEDclusterhaving210five‐coloredLEDsareusedinsidetheopticaldesignoftheV8lightengine.


The quasi‐collimated beam (maximum divergence ~±8°) from the individual LEDstransmittedthroughthetotal internalreflectionlens(TIR)isreflectedbyaparabolicreflectorand isdirected towards theopticalwindow(gate),which is situatedat thefocalplaneof the reflector.Onehexagonalpatterneddouble‐sidedconvexmicrolensarray(MA),calledaKohlerintegrator,isusedatthegatepositiontocombinealargenumberofquasi‐collimatedbeamsfromtheindividualcoloredLEDsarrivingattheMAatdifferentanglesintoahomogenouscolormixedbeam.Toprovideacosteffectivesolution in making an MA, a new process for tool fabrication utilized for injectionmoldingofthemicrolensstructuresisproposed.TheprocessexploitsthestandardCNCmillingprocesstomakeafirstroughmoldinsteel.Asurfacetreatmentisintroducedafterwards by spray coating with photoresist to obtain the optical surface qualityrequired for the MA. The process overcomes the expensive and tedious manualpolishingordirectdiamondturningandprovidesaneasyandinexpensivemoldandhence an extraordinarily inexpensive injection molding replication process. Theexperimentalinvestigationofthedirect3DprintedMAindicatesmorelosscomparedtothe MA made through the proposed procedure. The fabrication process isdemonstratedbyproducingadouble‐sidedMAwithø50mmfortheprototypeoftheV8LEDlightengine.Theworkingperformanceandthesurfacequalityofthelensarrayarecharacterizedexperimentally.

Toachievethegoalofhighluminousflux(~20,000lm),thesimulatedopticaldesignof the V8 light engine is optimized through the experimental investigation of theprototype which is discussed in Chapter 6. The modified simulated optical designindicatesthepossibilityofreceivingtheluminousflux>20,000lmfromtheV8lightengine. The light transmission from theMA depends on the refractive index of thematerialaswellasthelensletparameteroftheMA. TheoptimizationonthelensletparametersallowsahighanglebeamtotransmitthroughtheMA,andenhancestheopticalefficiencyoftheV8lightengine.Therefore,thecriticaldesignoptimizationoftheMAhelpstoincreasetheacceptanceangleoftheMAfrom±30°to±40°,increasingtheopticalefficiencyfurther.Anextensiveopticallosscharacterizationwascarriedoutfortheoptical componentsused inside theV8 lightengine.Usinga larger total internalreflection (TIR) lens and optimizing the parameter of the reflector, furtherenhancementoftheopticalefficiencyisfound.Thefinalopticalefficiencyofthesystemis~56%whichdelivers thegoalofhaving luminous flux>20,000 lm from the lightengine.Themodifieddesignisnotexperimentallyvalidatedduetoinsufficientfinancialsupport.However,thereisanagreementbetweentheprevioussimulationsandrelatedexperimentalresultswithamaximumaveragevariationof~9.7%.

To offer tunable white light and high color rendering, the additive color mixingtechniqueofmulti‐coloredLEDsisusedwhichisexplainedinChapter4.Initially,thecolormixingtechniqueisinvestigatedbysimulatinganexampleofasimpletri‐colorLEDsystem.AwhitelightcomplementedbycyananddeepredLEDsareusedinsidethesimulationtoachievehighcolorrenderingatalowcorrelatedcolortemperature.Aspectralmappingofthepossiblecombinationsoftri‐colorLEDsisusedto locatetheoptimal solutions within the color gamut, emphasizing chromaticity and the colorrenderingindices.Thismappingisfurtherexploitedasatooltosimulateacomplicatedsystem like theV8 light enginewhere the colormixing isdoneby five coloredLEDclusters.Ageometricalopticalmodelofthetri‐colorLEDsystemisusedtodesignandoptimize the light intensity distribution as a function of angle. The resultant design


produces a diffused homogeneous white light with a tuned low correlated colortemperature(CCT).Theexperimentalcharacterizationof the tri‐coloropticalsystemindicates that it is capableofreplacing the incandescent lighting inmuseum lightingapplication.Usingthetri‐colorsystemasatool, theV8 lightenginedesignuses fourcoloredLEDstosupplementandchangethespectralpowerdistribution(SPD)fromawarmwhiteLEDat3000K.TheoutputfromtheindividualLEDsisaddedtoproducetheresultantSPD,whichprovidesdimmablewhitelightoutputfromtheV8lightengineataCCTrangingfrom3000Kto6000Kwithahighcolorrendering.Themeasuredlightoutputfromthefirstprototypeshowsgoodcolorrendering(CRI(Ra)>85,CRI(R9)>40)andlightuniformityinintensityaswellasinCCT.Todeliveracollimatedwhitelight,alaser color mixing technique could be a potential candidate. However, it would beexpensivetousemorethan20lasers inordertoreachtheaimofa luminousflux>20,000 lm. Moreover, the color rendering of the resulting mixed light is not goodcomparedtocolormixingbyLED.This isduetothenarrowbandwidthofthelaser.Therefore,unlikecolormixingbyLED,thecolormixingbylaserproducesaresultantSPDwhichcannothaveabroadspectrum.Therefore,theproducedwhitelightwillbeuncomfortableforviewinganobjectduetothenarrowbandwidthinlaseremission.

Anotherstringentrequirementforstagelightingistohavelightingstabilityduringtheoperationaltime.Chapter5 isaboutthecolorconsistencyoftheV8lightengine.SincethelightengineconsistsofseveraldifferentcoloredLEDs,someinstabilityinlightintensitymayoccuraswellasincolor.LEDbinning(wavelengthandflux)andaging,ambient temperature, LED junction temperature, driver electronic such as digitalcontrol of current, system run time, etc. are the influencing factors that provideuncertaintyinthedesiredlightoutput.ThechapterdescribesthemethodofanalysisforthoseinfluencingfactorsonthelightenginebytheMonteCarlosimulation.Theknownvaluesoftheinfluencingfactorscanhelptocontroltheuncertaintyinthedesiredlightoutput. Inthechapter,asimulatedcolorcontrolsystemmechanismisproposedandexplained,whichincorporatesapre‐calibratedlookuptableaswellasacalibratedcolorsensortoprovideafeedbacktothecontrolsystem.ThestabilityofthecolorvariationfromthefinallightoutputischaracterizedbytheJNDcircleduringtheoperationaltime.Thesamecontrolmechanismcanbeappliedtothelasercolormixingsystemwhereinstead of the color sensor; a proposed novel speckle wavemeter can provide theinformation on the change in wavelengths (accuracy 100 MHz) for the used laserduring the operational time, which can further be converted into the desiredparameterstoprovidethefeedbacktothecontrolsystem.Experimentalresearchwascarriedouton theVCSEL laser tovalidate theperformanceof theproposedspecklewavemeterexaminingthechangeinwavelengthduetotemperaturevariationduringtheoperationaltimewhichisdescribedinChapter6.

Finally, the lightoutput fromtheprototype iscomparedwith theHalogen‐Fresnellamp(2kW)andshowsthattheV8lightengine(0.9kW)is55%moreenergyefficientthan traditionalstage lighting.Thecomparisonalso indicates that theopticalenergylossduetochangingfromfloodtospotpositionis10%fortheV8lightenginewhilethetraditionallightingloses37%.

Thefollowingsectionwilldiscusstheideaswhichcanbeimplementedinfutureforthebettermentoftheluminaire,whichisproposedinthisthesis.


7.2Outlook

Due to time restrictions aswell as the budgetary limitations, some ideas related tofurther extension of the project couldn’t be materialized. However, this sectiondiscusses the ideaswhichcouldbe implemented in future for theaspectsof furtherimprovement of the light output from the V8 light engine as well as for betterunderstandingoftheV8lightengineasacommercialproduct.

7.2.1Beamspot

DuetotheuseofahexagonalpatternedMAintheV8lightenginedesign,theprojectedlightoutputfromtheV8lightengineasahexagonalshape.Inaddition,ithasapositivedistortion called pincushion due to the geometrical aberration of the lenses. Thisdistortioncanchangewiththechangeofworkingdistancefromthelensanddistortionsare different for different wavelength of light. Since the V8 light engine comprisesseveralwavelengths,thechromaticaberrationispredominantattheperipheryofthelight spot if the spot is projected on a screen. The pincushion distortion can beaddressedbyalenscorrectionmodel[155],[156].Tomodifytheshapeofthespotfromthe hexagonal pattern to the circular one, several ideas can be adopted. Using thecircularaperturecansolve theproblemwitha tradeoffofenergy loss. TherandompatternofthelensletoftheMAcanalsobeconsideredtosolvetheproblem.However,todosofurtherresearchofthefabricationdifficultyoftheMAneedstobedone.Toaddresstheissueforthechromaticaberrationattheedgeofthebeamspot,thefilmwithafly‐eyepatterncanbeaddedonthetopsurfaceoftheFresnellenssothatthebeamwillbeevenlydistributed.However,thepatternandthefilmmaterialneedtobeoptimizedaccordingtotheapplication.

7.2.2Lightqualityandvisualperception

ToenhancetheopticalefficiencyoftheV8lightengine,aphosphorconvertedbluelaser(actinglikewhitelaser)withthecorrectopticsforthebeamshapingcanbeusedinsidetheopticaldesign.Thepresentthesisreportsonthetunablewhitelightfrom3000Kto6000K.However,thereisapossibilityofobtainingonlysaturatedcolorwithdimmingabilityastheV8lightenginecoversawidecolorgamutintheCIE1930chromaticitydiagram.

The spectral control which is reported in the thesis minimizes the variations inluminousfluxandCCTtosomelimit.However, itcannotcontrolthevariationinDuv.ThereforeitisobservedthatthecontrolledoutputisspreadalongtheIso‐thermallineinFigure5‐18.Futureworkwilladdressthis issueandproposetheerrorcorrectionmethodinDuvconsideringthetristimulusvaluesfromthecolorsensor.

Three LED color mixing systems are under development and will be used toexperimentally demonstrate both system stability and system‐to‐systemreproducibility.MonteCarloresultswillbecomparedtotheexperimentalresultsandfurthersimulationworkontheuncertaintycontributionsduetoabruptfailureofLEDs,LEDaging,andcontaminationoftheopticalsystemwillbeconsidered.

ThesameCCTandDuvvaluescanbeobtainedfromtwodifferentwhitelightswhilethelightsareproducedbytwodifferentSPDs.WhentwolightsfromtwodifferentV8


lightengines,whicharedifferentinSPDs,butwiththesameCCTandDuvvalues,areprojectedonanobjectsimultaneouslysidebyside,thehumaneyecandistinguishthedifferencebetweentwolightstosomeextentaccordingtotheobjectreflectanceandthevisualperceptionofthehumaneye.Theseeffectscanbeexaminedbyusertest,wheretwoexposuresshownsidebysidesimultaneouslyfromthetwoabovementionedlightsources to the object and viewers will be asked to express their opinion on theexperiencedlightonthescreen.Thisexperimentwillleadtoaresultofthetoleranceofthehumaneyetowardsthedifferentlightexposure.

7.2.3V8lightenginecomponents

Presently,theV8lightengineisusingwatercoolingasapassivecooling,whichisnotconvenient to use inside a compact product. Therefore, there is a research on toreplacingthewatercoolingbyactivecoolingusingaheatpipe,atemperaturesensor,togetherwith a fan. This replacementwill further enhance the stability of the lightoutputbycontrollingtheheatgenerationinsidetheV8lightengine.

Thepresentparabolicreflectorhasa10%reflectionloss.Thereisastate‐of‐the‐arton the surface quality enhancement of the reflector and the commercial product isavailable[157]toachievea5%reflectionlossfromtheprecisionparabolicreflector.Therefore,infuturetherewillbeapotentialitytouseareflectorwithabetterquality,whichwillleadtoenhancementofthelightoutputfromtheV8lightengine.Thesurfaceofthereflectorneedstoprotectfromcorrosion,whichwillincreasethedurabilityandin this context, the utility of applying HSQ is quite significant[158], [159]. Furtherresearchcanleadtoimprovementofthesurfacequalityofthereflector.

TheV8lightengineusedaFresnellensismadebyglass.InfutureitcouldbereplacedbyanacrylicFresnellens,whichhasalighterweightalongwithadesiredpatterntoeliminatethechromaticaberrationfromthelens.

7.2.4Scopeforthecustomisation

There is a room for customizing the V8 light engine design according to therequirements.TherewasresearchmadeontherelationbetweenthesizeandthelightoutputfromtheV8lightengine.Itwasobservedinthesimulation,thatthesmallestV8lightengine(~10cm)hasthecapabilityofilluminating5000lm,whichcanbeusedinmanyapplicationssuchasphoto‐voltaticlightingetc..

7.3Application

By meeting the entertainment industry's critical demands for color gamut, colorrenderingofwhitelightandintensity,theV8lightenginefinallypermitsarevolutionaryand attractive replacement of the energy consuming halogen based light sources inprofessional lighting. As an integrated light engine combining advanced optics,technology, and software, it is promising to integrate the V8 light engine as a lightsourceinawiderangeofproductsinaprofessionallighting;exploringthefullpotentialoftheLEDtechnology.Optimizedandsizereducedtechnologywillalsoberelevantfore.g.retrofitarchitecturallightingandlightingproductsfortheconsumermarket[160].


7.3.1Businesspotential

TheV8lightenginehasobviouscommercialpotentials:(a)asaretrofitlightenginetoreplaceenergyconsuming lightsources in functioningSpotlights(ETC,Altman,ADB,RobertJuliat,Spotlight,DeSisti,Selecon,Strand,etc.).(b)asanintegratedlightsourcefornew fixtures,automated lightsandarchitectural lights. (c)asa lightsource forarevolutionaryLEDspotlight,designedbyBB&S.

TheincreasingpriceofoilandtheglobalneedtoreduceC02emissionleadstoshifttowardsenergysavinglightsourcesingeneral.MostinstitutionsinDenmarkareforcedtomakeenergysavinginvestmentswithin4‐5years.V8lightengineisanticipatedtomeetthisgoalata5hoursdailyuse.Ifsavingsoncooling,maintenance,manhours,etcare involved, itwouldbeeven less.Thistendency isprojectedtoberelevant forthemajority of leading institutions in the global entertainment industry, facing thechallenge of energy reduction. As a retrofit, the V8 light engine offers a new LEDsolutiontoluminariesstillinfunction,increasingthelifetimeofthepreviousluminaryinvestments. Apart from meeting the high standards of conventional halogen lightsources,theV8lightenginealsodeliverscompletelynewfunctionsandfeatures,basede.g.ontheadditionalcolormixingsystem.TheV8lighttakestheLEDtechnologyfrom"justasgood"to"different".

TheV8lightenginemeetsthecustomers(suchas:theater,TVstudio,sports,stadium,etc.)basicneedsfor:

1. Anenergyreducinglightsource.

2. Alightsourcethatmeetsthecriticallevelforlightintensityandcolormix.

3. Acosteffectivesolutionconcerningpreviousinvestments(retrofit).

TheV8lightenginecreatescustomervalueby:

1. Reducingenergyusebyupto90%(selectedcolors)

2. Reducingtheenergyconsumptionrelatedtocooling.

3. Providinglightsourceswithalongerlifetime.

4. EliminatingIRandUVradiation

5. Offering amore cost efficient extension of the lifetime of functioning lightfixtures.

6. Reducing other costs related to lighting (lower frequency in changing lightsources,nooverheatedfixtures,reducingmanhours,filters,e.t.c.)

TheV8lightengineisuniquetochallengingLEDproductsby

1. Proposinganoptimaladditionalcolormixof5ormoredifferentcoloredLEDunits,overcomingthetechnologicallimits/barriersofRGBcoloringmixing

2. OfferinghighqualityLEDlightsourcestoawiderangeofprofessionallightingproductsinauniquelightengine


3. Providing an integrated spotlight light source where the optical systemcombines the light source, advanced driver electronic, thermal cooling andcolorcalibration,colorcontrol

Chapter 8 143 References

References

“True leaders do not makechoices with reference to theopinion of the majority. Theymake choices based on theopinionofthetruthandthetruthcan come from either themajority or the minority!” ‐IsraelmoreAyivor

Chapter8 :Bibliography

[1] ADBlighting,“‘EUROPE’RANGEPCandFresnelSpotlights,”2000.[Online].Available:http://www.adblighting.com/files/DS5113‐EEuropePC&Fresneldatasheet.pdf.

[2] J.K.KimandE.F.Schubert,“Transcendingthereplacementparadigmofsolid‐statelighting.,”Opt.Express,vol.16,no.26,pp.21835–21842,2008.

[3] A.DeAlmeida,B.Santos,B.Paolo,andM.Quicheron,“Solidstatelightingreview‐PotentialandchallengesinEurope,”Renew.Sustain.EnergyRev.,vol.34,pp.30–48,2014.

[4] J.T.BaileyandP.L.S.Husar,“Multi‐colorfocusableLEDlight,”US005,752,766A,1998.

[5] K.J.Leblanc,L.D.Hagan,andB.Hammond,“LEDfloodlightsystem,”US2007/0019415A1,2007.

[6] B.Chemel,M.K.Blackwell,C.Piepgras,andJ.S.Warwick,“MethodandapparatusforprovidingLED‐basedspotlightilluminationinstagelightingapplications,”US2010/0204841A1,2010.

[7] J.GADEGAARD,T.K.JENSEN,D.T.JØRGENSEN,P.K.KRISTENSEN,T.SØNDERGAARD,T.G.PEDERSEN,andK.PEDERSEN,“High‐outputLED‐basedlightengineforprofilelightingfixtureswithhighcoloruniformityusingfreeformreflectors,”Appl.Opt.,vol.55,no.6,pp.1356–1365,2016.

[8] C.‐C.Sun,I.Moreno,Y.‐C.Lo,B.‐C.Chiu,andW.‐T.Chien,“Collimatinglampwithwellcolormixingofred/green/blueLEDs.,”Opt.Express,vol.20,no.1,pp.A75–84,2012.

[9] H.Murat,D.Cuypers,andH.DeSmet,“DesignofnewcollectionsystemsformultiLEDlightengines,”Proc.SPIE,vol.6196,no.2006,pp.619604–619604–11,2006.

[10] A.Wilm,“RequirementsonLEDsinetenduelimitedlightengines,”Proc.SPIE,


PhotonicsMultimed.II,vol.7001,p.70010F–70010F–10,2008.

[11] D.Jørgensen,“Colormixingilluminationdevice,”US20140185285A1,2012.

[12] A.Neumann,J.J.Wierer,W.Davis,Y.Ohno,S.R.J.Brueck,andJ.Y.Tsao,“Four‐colorlaserwhiteilluminantdemonstratinghighcolor‐renderingquality.,”Opt.Express,vol.19Suppl4,no.May,pp.A982–A990,2011.

[13] M.Chakrabarti,C.Dam‐Hansen,J.Stubager,F.T.Pedersen,andC.H.Pedersen,“Replicationofopticalmicrolensarrayusingphotoresistcoatedmolds,”Opt.Express,vol.24,no.9,pp.9528–9540,2016.

[14] M.Chakrabarti,H.C.Pedersen,P.M.Petersen,C.Poulsen,P.P.Behrensdorff,andC.Dam‐Hansen,“Focusable,colortunablewhiteandefficientLEDlightengineforstagelighting,”Opt.Eng.,vol.Manuscript,2016.

[15] H.C.Pedersen,T.Brockmann,andC.Dam‐Hansen,“Lightengineforanilluminationdevice,”EP2843301,2015.

[16] M.Chakrabarti,A.Thorseth,D.D.Corell,andC.Dam‐Hansen,“Awhite–cyan‐redLEDsystemforlowcorrelatedcolourtemperaturelighting,”Light.ResTechnol,vol.0,pp.1–14,2015.

[17] M.Aldrich,“Dynamicsolidstatelighting.”MassachusettsInstituteofTechnology,2010.

[18] G.Eppeldauer,“SpectralresponsebasedcalibrationmethodofTristimuluscolorimeters,”J.Res.Natl.Inst.Stand.Technol.,vol.103,no.6,p.615,Nov.1998.

[19] J.L.Gardner,“ComparisonofCalibrationMethodsforTristimulusColorimeters,”J.Res.Natl.Inst.Stand.Technol.,vol.112,no.3,pp.129–138,2007.

[20] H.Li,X.Mao,Y.Han,Y.Luo,andV.Author,“Wavelengthdependenceofcolorimetricpropertiesoflightingsourcesbasedonmulti‐colorLEDs,”Opt.Express,vol.21,no.3,pp.3775–3783,2013.

[21] M.Chakrabarti,M.L.Jakobsen,andS.G.Hanson,“Speckle‐basedspectrometer.,”Opt.Lett.,vol.40,no.14,pp.3264–3267,2015.

[22] “Specklebasedwavemeter,”inSpeckle2015:VIInternationalConferenceonSpeckleMeterology,2015,p.96600U–1.

[23] M.Chakrabarti,A.Thorseth,J.Jepsen,andC.Dam‐Hansen,“MontecarloanalysisofmulticolourLEDlightengine,”inCIE,2015,p.OP60526–10.

[24] M.Chakrabarti,A.Thorseth,J.Jepsen,D.D.Corell,andC.Dam‐hansen,“Acolormanagementsystemformulti‐coloredLEDlighting,”inSPIE9571,LED‐basedIlluminationSystems,2015.

[25] M.Chakrabarti,A.Thorseth,J.Jepsen,D.D.Corell,andC.Dam‐Hansen,“ColorcontrolfortunablewhiteLEDlightingsystem,”Light.Res.Technol.,vol.Underrevi,2016.

[26] J.M.BornhorstandT.DeSoto,“Additivecolormixingsystemwithvariablehueandsaturationlightsources,”5,031,078,1989.

[27] S.GotohandJ.Yao,“Colourmixingmethodforvariablecolourlightingandvariablecolorluminaireforusewiththemethod,”US005384519A,1995.


[28] T.Fujii,Y.Gao,R.Sharma,E.L.Hu,S.P.DenBaars,andS.Nakamura,“IncreaseintheextractionefficiencyofGaN‐basedlight‐emittingdiodesviasurfaceroughening,”Appl.Phys.Lett.,vol.84,no.6,p.855,2004.

[29] S.Keeping,“LEDEfficacyImprovementShowsNoSignsofSlowing,”2011.

[30] S.Pimputkar,J.S.Speck,S.P.DenBaars,andS.Nakamura,“ProspectsforLEDlighting,”Nat.Photonics,vol.3,no.4,pp.180–182,2009.

[31] Y.Narukawa,M.Ichikawa,D.Sanga,M.Sano,andT.Mukai,“Whitelightemittingdiodeswithsuper‐highluminousefficacy,”J.Phys.D.Appl.Phys.,vol.43,no.35,p.354002,2010.

[32] EnergyEfficiency&RenewableEnergy,“ChallengesinLEDresearchanddevelopment,”U.SDepartmentofEnergy,2013.[Online].Available:http://energy.gov/eere/ssl/text‐alternative‐version‐challenges‐led‐research‐and‐development.

[33] OSRAM,“LEDknowledge:professionalknowhow.”[Online].Available:http://www.osram.com/osram_com/news‐and‐knowledge/led‐home/professional‐knowledge/index.jsp.

[34] CIE,“CIE1931.”CambridgeUniversityPress,Cambridge,1932.

[35] E.F.Schubert,Light‐EmittingDiodes,2ndEditio.CambridgeUniversityPress,2006.

[36] LightResearchCenter,“SwedishEnergyAgency‐AdvancedLightingSeminar.”RensselaerPolytechnicInstitute,pp.1–817,2013.

[37] C.DeCusatis,HandbookofAppliedPhotometry.AIPPress,1997.

[38] M.Rea,LightingHandbook:ReferenceandApplication,9thed.IilluminationEngineeringSocietyofNorthAmerica,2000.

[39] JosephB.Murdoch,Illuminationengineering:fromEdison’slamptothelaser.Technology&Engineering,1994.

[40] CIE,CIES025/E:2015:TestMethodforLEDLamps,LEDLuminairesandLEDModules.2015.

[41] Technoteam,“GoniophotometerRiGO801Generalspecification.”2015.

[42] IESNALM‐79‐08,“IESApprovedMethodfortheElectricalandPhotometricMeasurementsofSolid‐StateLightingProducts.”ANSI,2008.

[43] D.MacAdam.,“Visualsensitivitiestocolordifferencesindaylight,”J.Opt.Soc.Amer.,vol.32,no.5,pp.247–74,1942.

[44] D.B.JuddandG.T.Yonemura,“CIE1960UCSDiagramandtheMullerTheoryofColorVision,”vol.74,pp.23–30,1969.

[45] CIE,“CIES014‐5/E:2009CIE1976L*u*v*ColourSpaceandu’,v'UniformChromaticityScaleDiagram,”inCommissionInternationaledeL’Eclairage,2009.

[46] CIETN001:2014,“ChromaticityDifferenceSpecificationforLightSources,”2014.

[47] AmericanNationalStandard,“ANSI_ANSLGC78.377‐2008,”2008.


[48] Y.Ohno,“CalculationofCCTandDuvandPracticalConversionFormulae,”inCORM2011Conference,2011.

[49] CIE,“MethodofmeasuringandSpecifyingColourRenderingPropertiesofLightSources,”1995.[Online].Available:http://www.cie.co.at/publ/abst/13‐3‐95.html.

[50] G.WyszeckiandW.Stiles,ColorScience:ConceptsandMethods,QuantitativeDataandFormulae.WileyNewYork,1982.

[51] CIE,“TC1‐69:ColourRenditionbyWhiteLightSources.”2009.

[52] W.DavisandY.Ohno,“Colorqualityscale,”Opt.Eng.,vol.49,no.3,p.033602,2010.

[53] MikeWood,“TelevisionLightingConsistencyIndex2012,”2012.

[54] A.David,P.T.Fini,K.W.Houser,Y.Ohno,M.P.Royer,K.a.G.Smet,M.Wei,andL.Whitehead,“DevelopmentoftheIESmethodforevaluatingthecolorrenditionoflightsources,”Opt.Express,vol.23,no.12,p.15888,2015.

[55] IESTM‐30‐15,“IESMethodforEvaluatingLightSourceColorRendition,”2015.

[56] “Correspondence:InsupportoftheIESmethodofevaluatinglightsourcecolourrendition,”Light.Res.Technol.,vol.47,no.8,pp.1029–1034,Nov.2015.

[57] LouisHartmann,Theatrelighting;:Amanualofthestageswitchboard.NewYork:DBS,1970.

[58] BobbiOwen,LightingDesignersonBroadway:1915‐1990.NewYork:GreenwoodPress,1991.

[59] YaronAbulafia,TheArtofLightonStage:LightinginContemporaryTheatre.LondonandNewYork:Routledge,1977.

[60] J.MichaelGillette,TheatricalDesignandProduction:AnIntroductiontoSceneDesignandConstruction,Lighting,Sound,CostumeandMakeup,4thed.California:MayfieldPublishingCompany,2000.

[61] I.NavigantConsulting,“Life‐CycleAssessmentofEnergyandEnvironmentalImpactsofLEDLightingProducts,”2012.

[62] SanFranciscoCircuit,“MetalCorePCB/MCPCB,”2016.[Online].Available:https://www.sfcircuits.com/pcb‐production‐capabilities/metal‐core‐pcb.

[63] Multi‐cb,“PrintedCircuitBoardsfromMultiCB,”2016.[Online].Available:http://www.multi‐circuit‐boards.eu/en/products/printed‐circuit‐boards/special‐production/metal‐core‐pcb.html.

[64] EpecTec.,“FLEX&RIGID‐FLEXCIRCUITS.”[Online].Available:http://www.epectec.com/flex/.

[65] ELFA,“Dielectricum,”2016.[Online].Available:http://doctord.dyndns.org/courses/bei/ee360/Topics/Capacitors.pdf.

[66] F.R.Fournier,W.J.Cassarly,andJ.P.Rolland,“Fastfreeformreflectorgenerationusingsource‐targetmaps.,”Opt.Express,vol.18,no.5,pp.5295–5304,2010.

[67] R.Zhu,Q.Hong,H.Zhang,andS.‐T.Wu,“Freeformreflectorsforarchitectural


lighting,”Opt.Express,vol.23,no.25,p.31828,2015.

[68] W.D.Sproul,“Physicalvapordepositiontoolcoatings,”Surf.CoatingsTechnol.,vol.81,no.1,pp.1–7,1996.

[69] P.Benítez,V.Díaz,andJ.Alonso,“DesignandManufactureofAsphericLensesforNovelHighEfficientPhotovoltaicConcentratorModules,”2001.

[70] I.Wallhead,T.M.Jiménez,J.V.G.Ortiz,I.G.Toledo,andC.G.Toledo,“DesignofanefficientFresnel‐typelensutilizingdoubletotalinternalreflectionforsolarenergycollection,”Opt.Express,vol.20,no.S6,p.A1005,2012.

[71] C.‐C.A..ChenandF.‐C.Lee,“FLOWFRONTANALYSISOFTIRLENSOFLEDSWITHINJECTIONMOLDING,”2008.

[72] J.C.Miñano,M.Hernández,P.Benítez,J.Blen,O.Dross,R.Mohedano,andA.Santamaría,“Free‐formintegratorarrayoptics,”inProc.ofSPIE,2005,vol.5942,p.59420C–59420C–12.

[73] R.VölkelandK.J.Weible,“LaserBeamHomogenizing:LimitationsandConstraints,”inProceedingsofSPIE,2008,vol.7102,p.71020J–71020J–12.

[74] G.Naumann,H.;Schröder,BauelementederOptik :TAschenbuchfürKonstrukteure.CarlHanserVerlag,1983.

[75] D.MeisterandO.World,“MethodsforEstimatingLensThickness,”Opt.World,vol.26,no.201,pp.1–5,1997.

[76] Micro.Chem,“MicroSprayTM,”2016.[Online].Available:http://www.microchem.com/Prod‐Microspray.htm.

[77] B.McCallandT.S.Tkaczyk,“Fabricationofplasticmicrolensarrayforarraymicroscopybythree‐dimensionaldiamondmicromilling,”Opt.Eng.,vol.49,no.10,p.103401,2010.

[78] A.Y.YiandL.Li,“Designandfabricationofamicrolensarraybyuseofaslowtoolservo,”Opt.Lett.,vol.30,no.13,pp.1707–1709,2005.

[79] S.Scheiding,A.Y.Yi,A.Gebhardt,R.Loose,L.Li,S.Risse,R.Eberhardt,andA.Tünnermann,“Diamondmillingorturningforthefabricationofmicrolensarrays:comparingdifferentdiamondmachiningtechnologies,”Proc.SPIE,vol.7927,no.0,pp.79270N–1–11,2011.

[80] H.M.Leung,G.Zhou,H.Yu,F.S.Chau,andaS.Kumar,“Diamondturningandsoftlithographyprocessesforliquidtunablelenses,”J.MicromechanicsMicroengineering,vol.20,no.2,p.025021,2010.

[81] J.Yong,F.Chen,Q.Yang,G.Du,H.Bian,D.Zhang,J.Si,F.Yun,andX.Hou,“Rapidfabricationoflarge‐areaconcavemicrolensarraysonPDMSbyafemtosecondlaser,”ACSAppl.Mater.Interfaces,vol.5,no.19,pp.9382–9385,2013.

[82] S.Scheiding,A.Y.Yi,A.Gebhardt,L.Li,S.Risse,R.Eberhardt,andAndreasT¨unnermann,R.Eberhardt,andA.Tünnermann,“Freeformmanufacturingofamicroopticallensarrayonasteepcurvedsubstratebyuseofavoicecoilfasttoolservo,”Opt.Express,vol.19,no.24,p.23938,2011.

[83] G.Liu,W.Yu,H.Li,J.Gao,D.Flynn,R.W.Kay,S.Cargill,C.Tonry,M.K.Patel,C.Bailey,andM.P.Y.Desmulliez,“Microstructureformationinathickpolymerbyelectrostatic‐inducedlithography,”J.MicromechanicsMicroengineering,vol.23,


no.3,p.035018,2013.

[84] W.‐C.Chen,T.‐J.Wu,W.‐J.Wu,andG.‐D.J.Su,“Fabricationofinkjet‐printedSU‐8photoresistmicrolensesusinghydrophilicconfinement,”J.MicromechanicsMicroengineering,vol.23,no.6,p.065008,2013.

[85] B.JiriCecha,H.Pranovb,G.Kofodb,M.Matschukb,S.Murthyb,andR.Taboryskia,“Surfaceroughnessreductionusingspray‐coatedhydrogensilsesquioxanereflow,”Appl.Surf.Sci.,vol.280,pp.424–430,2013.

[86] Lutron,“DMX‐512Fundamentals,”2010.[Online].Available:file:///O:/FTNK/Public/RESOURCE‐EDI/Projects/LEDDTU/Maumita/Thesis/Reference/basicknowledge/DMXwebinar_7‐29‐2010.pdf.

[87] WyrowskiPhotonicsUG,“Unifiedopticaldesignsoftware‘WyrowskiVirtualLabFusion.’”LightTransInternationalUG,2016.

[88] WyrowskiPhotonicsUG,“WyrowskiVirtualLabFusion,”2016.[Online].Available:http://www.wyrowski‐photonics.com/virtuallab/news.html.

[89] J.M.P.J.Verstegen,D.Radielovic,andL.E.Vrenken,“Anewgenerationdeluxefluorescentlamp‐combininganefficacygreaterthan80lm/Wwithacolorrenderingindexofabout85,”J.Illum.Eng.Soc.,vol.4,no.2,p.90,1975.

[90] KarlGuttagonTechnology,“DiodeGreenLasers.”2011.

[91] DHgate,“Datasheet.”2015.

[92] Y.‐C.Lo,I.Moreno,B.‐C.Chiu,W.‐T.Chien,J.‐Y.Cai,Y.‐Y.Chang,andC.‐C.Sun,“ColormixingcollimatinglampbasedonRGBLEDs,”inProceedingsofSPIE‐TheInternationalSocietyforOpticalEngineering,2012,vol.8486,p.84860T.

[93] S.Muthu,F.J.P.Schuurmans,andM.D.Pashley,“Red,green,andblueLEDsforwhitelightillumination,”IEEEJ.Sel.Top.QuantumElectron.,vol.8,no.2,pp.333–338,2002.

[94] CREE,“LEDColorMixing:BasicsandBackground,”2016.

[95] E.F.Schubert,“Colormixingandcolorrendering,”inLight‐EmittingDiodes,2nded.,CambridgeUniversityPress,2006,pp.313–327.

[96] D.B.JuddandG.T.Yonemura,“CIE1960UCSDiagramandtheMullerTheoryofColorVision,”J.Res.NotionalBur.Stand.‐A.Phys.Chem.,vol.74A,no.1,pp.23–30,1969.

[97] ISO11664‐5:2009,“Colorimetry—Part5:CIE1976L*u*v*Colourspaceandu’,v'uniformchromaticityscalediagram,”2009.

[98] D.Corell,H.Ou,C.Dam‐Hansen,P.‐M.Petersen,andD.Friis,“LightEmittingDiodesasanalternativeambientilluminationsourceinphotolithographyenvironment.,”Opt.Express,vol.17,no.20,pp.17293–17302,2009.

[99] S.Chhajed,Y.Xi,Y.L.Li,T.Gessmann,andE.F.Schubert,“Influenceofjunctiontemperatureonchromaticityandcolor‐renderingpropertiesoftrichromaticwhite‐lightsourcesbasedonlight‐emittingdiodes,”J.Appl.Phys.,vol.97,pp.1–8,2005.

[100] A.Thorseth,“Optimizationoflightqualityfromcolormixinglight‐emitting


diodesystemsforgenerallighting,”inProceedingsofSPIE,2012,vol.8278,p.82781O–82781O–11.

[101] M.S.ReaandJ.P.Freyssinier,“Whitelighting:Aprovisionalmodelforpredictingperceivedtintin‘white’illumination,”ColorRes.Appl.,vol.39,no.V,pp.466–479,2014.

[102] Y.OhnoandM.Fein,“VISIONEXPERIMENTONACCEPTABLEANDPREFERREDWHITELIGHTCHROMATICITYFORLIGHTING,”inCIE2014LightingQualityandEnergyEfficiencyIssue:CIEx039:2014,pp.192–199.

[103] PhilipsLumileds,“ColorandWhiteLEDportfolio‐Technicaldatasheet.”2014.

[104] CREE,“Cree®XLamp®XP‐ELEDsproductfamilydatasheet.”pp.1–38,2014.

[105] CREE,“CreeXLampXTFamilyLEDs®®Binning&labeling.”pp.1–27,2014.

[106] Y.‐F.Hsieh,O.‐Y.Mang,T.‐W.Huang,andC.‐C.Lee,“DeterminationofoptimalconvertingpointofcolortemperatureconversioncompliedwithANSIC78.377forindoorsolid‐statelightinganddisplayapplications,”Opt.Express,vol.20,no.18,pp.20059–20070,2012.

[107] ThomasNimzandFredrikHailerandKevinJensen,“SensorsandFeedbackControlofmulti‐colourLEDsystems,”LEDprofessionalReview(LpR),pp.2–5,2012.

[108] S.RobinsonandI.Ashdown,“PolychromaticOpticalFeedbackControl,Stability,andDimming,”inProc.ofSPIE,2006,vol.6337,pp.633714–633714–10.

[109] M.M.SistoandJ.Gauvin,“AccuratechromaticcontrolandcolorrenderingoptimizationinLEDlightingsystemsusingjunctiontemperaturefeedback,”inProc.SPIE,2014,vol.9190,p.919002.

[110] K.ManandI.Ashdown,“AccuratecolorimetricfeedbackforRGBLEDclusters,”inProc.ofSPIE,2006,vol.6337,pp.633702–1.

[111] G.HeandJ.Tang,“SpectraloptimizationofcolortemperaturetunablewhiteLEDswithexcellentcolorrenderingandluminousefficacy,”Opt.Lett.,vol.39,no.19,pp.5570–5573,2014.

[112] S.BusoandG.Spiazzi,“Whitelightsolidstatelampwithluminanceandcolortemperaturecontrol,”inCOBEP2011‐11thBrazilianPowerElectronicsConference,2011,pp.837–843.

[113] MAZeTGmbH,“TrueColormanagementforlightingsystems,”2007.

[114] J.L.Gardner,“ComparisonofCalibrationMethodsfor,”vol.112,no.3,pp.129–138,2007.

[115] M.CoxandP.Harris,“EVALUATIONOFMEASUREMENTUNCERTAINTYBASEDONTHEPROPAGATIONOFDISTRIBUTIONS,”vol.46,no.9,pp.9–14,2003.

[116] Z.Liu,S.Liu,K.Wang,andX.Luo,“OpticalAnalysisofColorDistributioninWhiteLEDsWithVariousPackagingMethods,”IEEEPhotonicsTechnol.Lett.,vol.20,no.24,pp.2027–2029,Dec.2008.

[117] R.QiandD.P.Bajorins,“UncertaintyanalysisforchromaticitycoordinatesandluminousfluxmeasurementsofLEDs,”vol.9190,no.Ccd,p.919008,Sep.2014.


[118] GammaScientific,“GS‐1150HandheldSpectrophotometers.”2015.

[119] MAZeTGmbH,“AccuracyoptimizationofTrueColorSensorsolutionsforLEDcolorcontrolofmultipleLEDlightsources,”2012.

[120] CIE,“Methodofmeasuringandspecifyingcolourrenderingpropertiesoflightsources,”CIE13.3.p.16,1995.

[121] CIE,“ChromaticityDifferenceSpecificationforLightSources,”2014.

[122] A.G.Ryabtsev,E.V.Lutsenko,G.I.Ryabtsev,G.P.Yablonskii,A.S.Smal,B.Schineller,andM.Heuken,“StudyoflaserthresholdtemperaturesensitivityinopticallypumpedGaNepilayers,”Phys.STATUSSOLIDIC,no.1,pp.479–482,2002.

[123] M.Fukuda,T.Mishima,N.Nakayama,andT.Masuda,“Temperatureandcurrentcoefficientsoflasingwavelengthintunablediodelaserspectroscopy,”Appl.Phys.BLasersOpt.,vol.100,no.2,pp.377–382,2010.

[124] L.I.Goldfischer,“Autocorrelationfunctionandpowerspectraldensityoflaser‐producedspecklepatterns,”JOSA,vol.55,no.3,pp.247–252,1965.

[125] J.W.Goodman,“Somefundamentalpropertiesofspeckle,”JOSA,vol.66,no.11,pp.1145–1150,1976.

[126] J.C.Dainty,“Laserspeckleandrelatedphenomena,”inBerlinandNewYork,Springer‐Verlag(TopicsinAppliedPhysics.Volume9),1975.298p,1975,vol.9.

[127] I.Yamaguchi,“SpeckleDisplacementandDecorrelationintheDiffractionandImageFieldsforSmallObjectDeformation,”Opt.ActaInt.J.Opt.,vol.28,pp.1359–1376,1981.

[128] YamaguchiI.,“Alaser‐specklestraingauge,”J.PhysE:Sci.Instrum,vol.14,1981.

[129] I.Yamaguchi,K.Kobayashi,andL.Yaroslavsky,“Measurementofsurfaceroughnessbyspecklecorrelation,”Opt.Eng.,vol.43,no.November2004,p.2753,2004.

[130] I.Yamaguchi,“DigitalHolographicInterferometryandSpeckleCorrelation,”AIPConf.Proc.,vol.1236,pp.46–51,2010.

[131] BrandonReddingandHuiCao,“Usingamultimodefiberasahigh‐resolution,low‐lossspectrometer,”Opt.Lett.,vol.37,no.16,pp.3384–3386,2012.

[132] J.E.Webb,“CONCAVEGRATINGSPECTROMETER,”US3,229,568A,1966.

[133] J.M.Vaughan,TheFabryPerotinterferometer,0thed.Taylor&FrancisGroup,1989.

[134] H.T.YuraandS.G.Hanson,“Opticalbeamwavepropagationthroughcomplexopticalsystems,”JOSAA,vol.4,no.10,pp.1931–1941,1987.

[135] H.T.YURAandS.G.HANSON,“OpticalBeamWave‐PropagationthroughComplexOptical‐Systems,”J.Opt.Soc.Am.A‐OpticsImageSci.Vis.,vol.4,no.10,pp.1931–1948,1987.

[136] P.ZhouandK.E.Goodson,“Subpixeldisplacementanddeformationgradientmeasurementusingdigitalimage/specklecorrelation(DISC),”Opt.Eng.,vol.40,no.8,p.1613,2001.


[137] M.Guizar‐Sicairos,S.T.Thurman,andJ.R.Fienup,“Efficientsubpixelimageregistrationalgorithms.,”Opt.Lett.,vol.33,no.2,pp.156–158,2008.

[138] J.H.Oh,S.J.Yang,Y.‐G.Sung,andY.R.Do,“Excellentcolorrenderingindexesofmulti‐packagewhiteLEDs,”Opt.Express,vol.20,no.18,p.20276,2012.

[139] LeiJiang,GuXin,andM.Liu,“SpectralassemblageusingLEDstoobtainspecifiedcharacteristics,”Appl.Opt.,vol.53,no.35,pp.8151–8156,2014.

[140] M.Zhang,Y.Chen,andG.He,“Colortemperaturetunablewhite‐lightLEDclusterwithextrahighcolorrenderingindex.,”ScientificWorldJournal.,vol.2014,no.9,p.897960,Jan.2014.

[141] X.Zhu,Q.Zhu,H.Wu,andC.Chen,“OpticaldesignofLED‐basedautomotiveheadlamps,”Opt.LaserTechnol.,vol.45,no.1,pp.262–266,2013.

[142] X.Zhu,J.Ni,andQ.Chen,“AnopticaldesignandsimulationofLEDlow‐beamheadlamps,”J.Phys.Conf.Ser.,vol.276,p.012201,2011.

[143] E.ChenandF.Yu,“DesignofLED‐basedreflector‐arraymoduleforspecificilluminancedistribution,”Opt.Commun.,vol.289,pp.19–27,2013.

[144] MAZeTGmbH,“DATASHEETDKMTCS‐INT‐AB4,”2015.

[145] C.Mercinaries,“IOW24DGConverterI2CtoUSB.”.

[146] Nikoncorporation,“Confocalmicroscope,”Thesourceformicroscopyeducation,2015.[Online].Available:http://www.microscopyu.com/articles/confocal/confocalintrobasics.html.[Accessed:09‐Jun‐2016].

[147] NikonCorporation,“Whitelightinterferometricmicroscopicsystem,”2015.[Online].Available:http://www.nikon.com/about/technology/life/instruments/wl_interferometric/index.htm.[Accessed:09‐Jun‐2016].

[148] ISO4287:1997,“GeometricalProductSpecifications(GPS)‐‐Surfacetexture:Profilemethod‐‐Terms,definitionsandsurfacetextureparameters,”1997.[Online].Available:http://www.iso.org/iso/catalogue_detail.htm?csnumber=10132.[Accessed:20‐Sep‐2015].

[149] E.Bretschneider,“Efficacylimitsforsolid‐statewhitelightsources,”PhotonicsSpectra,vol.41,no.3,pp.72–81,2007.

[150] N.J.MillerandM.Royer,“ColorSpacesandPlanckianLoci:UnderstandingallthoseCrazyColorMetrics.”pp.1–49,2012.

[151] X.GuoandK.W.W.Houser,“Areviewofcolourrenderingindicesandtheirapplicationtocommerciallightsources,”Light.Res.Technol.,vol.36,no.3,pp.183–199,2004.

[152] W.DavisandY.Ohno,“Towardanimprovedcolorrenderingmetric,”Proc.SPIE,vol.5941,no.2,p.59411G–59411G–8,2005.

[153] W.DavisandY.Ohno,“Evaluationofcolordierenceformulaeforcolorrenderingmetrics,”NIST,2006.[Online].Available:http://physics.nist.gov/Divisions/Div844/facilities/photo/.


[154] A.Žukauskas,R.Vaicekauskas,F.Ivanauskas,R.Gaska,andM.S.Shur,“Optimizationofwhitepolychromaticsemiconductorlamps,”Appl.Phys.Lett.,vol.80,no.2,pp.234–236,2002.

[155] D.Clausanda.W.Fitzgibbon,“Arationalfunctionlensdistortionmodelforgeneralcameras,”2005IEEEComput.Soc.Conf.Comput.Vis.PatternRecognit.,vol.1,pp.0–6,2005.

[156] W.Hugemann,“Correctinglensdistortionsindigitalphotographs,”IngenieurbüroMorawski+HugemannLeverkusen,…,pp.1–12,2010.

[157] EdmundOptics,“PrecisionParabolicMirrors.”1992.

[158] J.Creusa,H.Idrissia,H.Mazille,F.Sanchetteb,andP.Jacquotb,“CorrosionbehaviourofAllTicoatingelaboratedbycathodicarcPVDprocessontomildsteelsubstrate,”ThinSolidFilms,vol.346,pp.150–154,1999.

[159] B.K.Hwang,K.Fujihara,U.I.Chung,K.H.Kim,S.I.Lee,andM.Y.Lee,“ApplicationofHSQ(HydrogenSilsesquioxane)BasedSOGtoPre‐MetalDielectricPlanarizationinSTC(StackedCapacitor)DRAM,”inSymposiumonVLSITechnologyDigestofTechnicalPapers,1996,no.1996,pp.112–113.

[160] DTUFotonikandBB&SAps.,“Advancedtechnologyprojects.”2011.

[161] B.Rose,S.G..Hanson,andH.T.Yura,“ComplexABCDmatrices:Ageneraltoolforanalyzingarbitraryopticalsystems,”inOpticalsensorsandmicrosystems:Newconcepts,materialsandtechnologies,1998.

[162] S.G.Hanson,M.L.Jakobsen,andH.T.Yura,“Complex‐ValuedABCDMatricesandSpeckleMetrology,”inLinearCanonicalTransformsTheoryandApplications,vol.27,no.1,Springer,2010,pp.397–426.

[163] K.J.Beck,J.V.;Arnold,Parameterestimationinengineeringandscience.WileyNewYork,1977.

[164] A.Tarantola,InverseproblemtheoryandMethodsforModelParameterEstimation.Siam,2005.

Appendix 153 Light output geometry

AppendixA

Thefollowingfiguredefinesthelightoutputgeometryinstagelightingapplication.

Figure 8‐1 Describing the light output geometry in terms of FWHM at different distance

AppendixB

Agoniospectro‐photometer,called“REFLET”isusedformeasuringtheback‐scatteringandforwardscatteringlightfromthediffusingsample.TheBSDF(bidirectionalscatterdistributionfunction)measurementofthediffuser(section3.4inChapter3)wasdoneinENTPE,France.Afterthat,themeasureddatawasanalyzedandthescattermodelwasfedintotheZemaxsoftwareforfurtherprocess.Figure8‐2representstheZemaxBRDF(bidirectionalreflectancedistributionfunction)formatwherethelightincidentonthediffusedsurfacecreatesalightconeuponreflection.Similarly,therewillbealightconeaftertransmissionfromthediffuser.Thedetectorinsidethegoniometer,measuresreflectedandtransmittedlightcoveringthefullhemisphere(2πsr)byrotation.Thus,therearetwoanglestobeconsidered;scatteredradialandscatterazimuthduringtherotation.

Figure 8‐2: Diagram according to Zemax BRDF format which is adapted for the BRDF measurement

Appendix 154 Light output geometry

AppendixC

LightbeampropagatingthroughanycomplexopticalsystemcanberepresentedbyanABCD ray‐transfermatrix[161], [162].Mathematical approximations are required ineachstepifthefieldfromoneelementtothenextneedstobetraced. ThecomplexABCDmatrixmethodsolvestheproblemdifferently,becausetheapproximationsareachieved on the optical system itself. Thus, the result acquired is effective for anarbitraryABCDopticalsystem.However,theconsiderationislimitedandonlytrueforthecasewhereallilluminatingbeamsareGaussianshaped(onlyforTEM00lasermode).This is because,Green’s Functions in the Fresnel regime for free space propagationalongwithkernel forquadraticphase changepresentedby a lens areof a complexGaussianform.

AppendixD

Thecalibrationmatrixcanbefoundaccuratelyusingall32measurementpointsinsidetheSmatrixandalsoutilizingthecorrespondingentireJmatrix.Inthatscenarioleast‐squareestimateneedstoconsider[163],[164].

Therefore,

(58),

WhereSandJrepresentarethe32setsoftristimulusvaluesfromthespectrometeraswellfromthecolorsensor,respectively,whereasAisthecalibrationmatrix.

IfmatrixFofmeasurementerrorscanintroduce,thencanbere‐writtenas:

(59)

advanced optical design and control of multi-colored ssl

Documents