Optical configuration and analysis of the AMBER/VLTI instrument

15 Pages·2007·2.58 MB·English

A&A464,13–27(2007) Astronomy
DOI:10.1051/0004-6361:20065432 &
(cid:1)c ESO2007 Astrophysics
Special feature
AMBER:Instrumentdescriptionandfirstastrophysicalresults
Optical configuration and analysis of the AMBER/VLTI instrument
S.Robbe-Dubois1,S.Lagarde2,R.G.Petrov1,F.Lisi3,U.Beckmann4,P.Antonelli2,Y.Bresson2,
G.Martinot-Lagarde2,7,A.Roussel2,P.Salinari3,M.Vannier1,6,13,A.Chelli5,M.Dugué2,G.Duvert5,S.Gennari3,
L.Glück5,P.Kern5,E.LeCoarer5,F.Malbet5,F.Millour1,5,K.Perraut5,P.Puget5,F.Rantakyrö6,E.Tatulli5,
G.Weigelt4,G.Zins5,M.Accardo3,B.Acke5,14,K.Agabi1,E.Altariba5,B.Arezki5,E.Aristidi1,C.Baffa3,
J.Behrend4,T.Blöcker4,S.Bonhomme2,S.Busoni3,F.Cassaing8,J.-M.Clausse2,J.Colin2,C.Connot4,L.Delage10,
A.Delboulbé5,A.DomicianodeSouza1,2,T.Driebe4,P.Feautrier5,D.Ferruzzi3,T.Forveille5,E.Fossat1,R.Foy9,
D.Fraix-Burnet5,A.Gallardo5,E.Giani3,C.Gil5,15,A.Glentzlin2,M.Heiden4,M.Heininger4,O.HernandezUtrera5,
K.-H.Hofmann4,D.Kamm2,M.Kiekebusch6,S.Kraus4,D.LeContel2,J.-M.LeContel2,T.Lesourd7,B.Lopez2,
M.Lopez7,Y.Magnard5,A.Marconi3,G.Mars2,P.Mathias2,P.Mège5,J.-L.Monin5,D.Mouillet5,16,D.Mourard2,
E.Nussbaum4,K.Ohnaka4,J.Pacheco2,C.Perrier5,Y.Rabbia2,S.Rebattu2,F.Reynaud10,A.Richichi11,A.Robini1,
M.Sacchettini5,D.Schertl4,M.Schöller6,W.Solscheid4,A.Spang2,P.Stee2,P.Stefanini3,M.Tallon9,
I.Tallon-Bosc9,D.Tasso2,L.Testi3,F.Vakili1,O.vonderLühe12,J.-C.Valtier2,andN.Ventura5
(Affiliationscanbefoundafterthereferences)
Received13April2006/Accepted28August2006
ABSTRACT
Aims.ThispaperdescribesthedesigngoalsandengineeringeffortsthatledtotherealizationofAMBER(AstronomicalMultiBEamcombineR)
andtotheachievementofitspresentperformance.
Methods.Onthebasisofthegeneralinstrumentalconcept,AMBERwasdecomposedintomoduleswhosefunctionsanddetailedcharacteristics
aregiven.Emphasisisputonthespatialfilteringsystem,akeyelementoftheinstrument.Weestablishedabudgetfortransmissionandcontrast
degradationthroughthedifferentmodules,andmadethedetailedopticaldesign.Thelatterconfirmedtheoverallperformanceoftheinstrument
anddefinedtheexactimplementationoftheAMBERoptics.
Results.TheperformancewasassessedwithlaboratorymeasurementsandcommissioningsattheVLTI,intermsofspectralcoverageandreso-
lution,instrumentalcontrasthigherthan0.80,minimummagnitudeof11inK,absolutevisibilityaccuracyof1%,anddifferentialphasestability
of10−3radoveroneminute.
Keywords.instrumentation:highangularresolution–instrumentation:interferometers–methods:analytical–methods:numerical–
methods:laboratory–techniques:highangularresolution
1. Introduction – Minimum magnitude: K = 11, H = 11 (goals: K = 13,
H =12.5,J =11.5).
AMBER (Petrov et al. 2007) is the near infrared focal – Absolutevisibilityaccuracy:3σ =0.01(goal:σ =10−4).
V V
beam combiner of the Very Large Telescope Interferometric – Differential phase stability: 10−3 rad (goal: 10−4 rad) over
mode (VLTI). A consortium of French, German, and
oneminuteintegration.Thisallowsustocomputephaseclo-
Italian institutes is in charge of the specification, design,
surethatisnecessaryinthesearchofbrowndwarfsandextra
construction, integration, and commissioning of AMBER
solarplanets(Segransanetal.2000).
(http://amber.obs.ujf-grenoble.fr). This instrument is
designed to combine three beams coming from any combina- Thesespecificationshavebeenthestartingpointofaglobalsys-
tionofUnit8-mTelescopes(UT)orAuxiliary1.8-mTelescopes tem analysis (Malbet et al. 2003) initiated by a group of inter-
(AT).Thespectralcoverageisfrom1to2.4µm withapriority ferometristsfromseveralFrenchinstitutesandcompletedbythe
totheK band. InterferometricGRoup(IGR)ofAMBER.Thisworkledtothe
AMBERisageneraluserinstrumentwithaverywiderange currentdefinition of the AMBER instrumentwhose broad out-
of astrophysical applications, as is shown in Richichi et al. linesarerecalledintheparagraphsbelow.
(2000) and Petrov et al. (2007). To achieve the ambitious pro- To reach a sufficient sensitivity, ESO provided a 60 ac-
grams, Petrov et al. (2001) presented the associated specifica- tuator curvature sensing system MACAO (Multi-Application
tionsandgoals,suchas: CurvatureAdaptiveOptics) (Arsenaultetal. 2003)specifiedto
deliver at least a 50%Strehl ratio @ 2.2 µm for on-axisbright
– Spectralcoverage:J,H,andKbands,from1.0µmto2.3µm sources (V = 8) under median seeing conditions (0.65(cid:2)(cid:2)) and
(goal:upto2.4µm). a 25% Strehl ratio @ 2.2 µm for faint sources (V = 15.5) un-
– Spectralresolutions:minimumspectralresolution30 < R < der the same seeing conditions. The required high accuracy of
50,mediumspectralresolution500<R<1000,andhighest the absolute visibility measurements implies the use of spatial
spectralresolution10000<R<15000. filters (Mège et al. 2000; Tatulli et al. 2004) with single mode
Article published by EDP Sciences and available at http://www.aanda.orgor http://dx.doi.org/10.1051/0004-6361:20065432
14 S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument
fibersbasedontheexperienceofothersmaller,successfulinter-
ferometerssuchasIOTA/FLUOR(CoudéduForesto1997).The
atmosphericnoiseisreducedtophotometricfluctuations,which
canbemonitored,andtoOpticalPathDifference(OPD)fluctu-
ationsbetweenthedifferentpupils,whichcanbefrozenbyvery
shortexposuresoradaptivelycorrectedbyafringetracker.
Thesimultaneousobservationofdifferentspectralchannels
isensuredbydispersedfringes.Thisverysignificantlyincreases
the number and the quality of the measurements and subse-
quently the constraints imposed on the atmospherical models.
The modularity of the concept was a strong argument in fa- Fig.1.SchemeoftheAMBERconfiguration:(1)multiaxialbeamcom-
vor of the multi-axial scheme, as carried out on the Grand biner;(2)cylindricaloptics;(3)anamorphosedfocalimagewithfringes;
Interferomètreà 2 Télescopes (GI2T) of the Plateau de Calern (4) “long slit spectrograph”; (5) dispersed fringes on 2D detector;
(Mourardetal.2000).Inaddition,itwasdemonstratedthatthe (6)spatialfilterwithsinglemodeopticalfibers;(7)photometricbeams.
instrument must correct the atmospheric transversal dispersion
in J and H (Tallon-Bosc 1999). The need of an image cold
stopwasassessedbyMalbet(1999)toreducethethermalback-
the beam size and position, flux optimization, OPD equal-
groundcomingfromtheblackbodyemissionofthefiberheads,
ization,polarizationcontrol,andcombinationofthespectral
whichcanbegreaterthanthedetectorRON,especiallyforlong
bands.
timeexposuresinthe K-Band.Apupilmaskalsoactsasacold
– ANamorphosisSystem(ANS)tocompressthebeamsinone
stop.Toperformthedatareduction,theABCDalgorithm(Chelli
directionwithoutperturbingthepupillocation.
2000;Millouretal.2004),asusedwithco-axialconfigurations
– CooledSPectroGraph(SPG).Thiselementincludes:disper-
with a temporal coding, was chosen. The associated complete
sionwithdifferentresolutions,thermalnoisereduction,pupil
datacalibrationprocedurewasthenfullydefined(Hofman1999;
configuration,photometriccalibration,andspectralfiltering.
Chelli2000;Tatullietal.2007).
– Cooled DETector (DET), which detects the dispersed
On the basis of the general instrumental concept resulting
fringes.
from the global system analysis, we defined the main mod-
Theauxiliarymodulesare:
ules and necessaryaccessories (suchas the alignmentunits) of
– System to correct the atmospheric transversal dispersion
AMBER. We established a budget for throughputand contrast
degradation through the different modules, made the detailed (ADC)in JandH.
– CalibrationandAlignmentUnit(CAU)necessarytoperform
optical design,and performeda completeopticalanalysis. The
the contrast calibration (Millour et al. 2004; Tatulli et al.
latterconfirmedtheexpectedoverallperformanceoftheinstru-
2007).
mentintermsofsignal-to-noiseratio(SNR).Thisentireprocess
– Remote Artificial Sources (RAS) allowing for the align-
isdescribedinthepresentpaper.Theproceduredescribedhere
to allocate the specifications of the different modules of an in- ments,thespectralcalibration,andfeedingtheCAUforthe
contrastcalibration.
terferometer could be used, after some changes, for the design
– MatrixCalibrationSystem(MCS)scheduledtocalibratethe
ofotherinterferometricalinstruments,suchas,forexample,the
contrasttoachievespecificscientificgoals.
VLTIsecondgenerationinstruments.
– ABYPass(BYP)oftheSPFtoalignthewarminstrumentin
thevisible(forcontrollingthepupil,theimagelocation,and
2. OverviewoftheAMBERimplementation the beamseparationandheight),andtoinjectlightdirectly
toSPG.
The concept of AMBER is illustrated by Fig. 1. Each input
beam is fed into a single mode fiber that reducesall chromatic We will describe each module in detail in the following para-
wavefront perturbations to photometric and global OPD fluc- graphs, from the entrance of AMBER to the detector, starting
tuations (6). At the output of the fibers, the beams are colli- withthemainmodulesandcontinuingwiththeauxiliaryones.
mated,maintainedparallel,andthenfocusedinacommonAiry
disk (1). The latter contains Young fringes with spacings spe-
cifictoeachbaseline,allowingustoseparatetheinterferograms 3. Thespatialfilters
in the Fourier space. This Airy disk goes through the spectro-
ThethreeVLTIbeamsattheentranceofAMBERhaveadiam-
graphslit(3)afterbeinganamorphosedbycylindricaloptics(2).
eter of 18 mm and equal separations of 240 mm (see Fig. 2).
Thespectrograph(4)formsdispersedfringesonthedetector(5),
The three AMBER beams are separated by 70 mm at the fiber
whereeachcolumnallowsustoanalysetheinterferogramsina
differentspectralchannel.A fractionof eachbeam is collected entrance.Theseparationsattheentranceareachievedbyadjust-
ing theVLTI beaminjectionoptics, allowingusto compensate
before the beam combinationto monitorthe photometryvaria-
fortheopticalpathdifferencewithadditionalpathlengths.The
tions(7).
chosenconfigurationensuresperfectsymmetrybetweenthein-
Figure2showstheglobalimplementationofAMBER, and
terferometricpathsandallowsfortheuseofsmallsizeoptics.
Fig. 3showsa pictureoftheinstrumenttakenattheendofthe
integrationatParanal(March2004).Thecoreoftheinstrument
iscomposedofthefollowingmodules,fillingspecificfunctions: 3.1.Characteristicsofthespatialfiltering
– SPatial Filters (SPF) to spatially filter the wavefront per- Single mode fibers cannot be efficient over a too large wave-
turbationandreach high-visibilityprecisionmeasurements. length domain.The full J, H, K rangefrom1 to 2.4µm needs
The functionsof this elementare also: spectralband selec- atleasttwodifferentfibers.Themostefficientwayistouseone
tion(J, H,andK),interferometricarmselection,controlof spatial filter by spectral band, avoiding dividing the H-band in
S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument 15
Fig.2.GeneralimplementationofAMBER.ThetopschemeshowsthelightpathfromtheVLTItothedetector.Detailedconfigurationbelow.
TheVLTIbeamsarriveinthelowerleftcorner.OPM-CAU:CalibrationandAlignment Unit.OPM-BCD:beaminvertingdevice(Petrovetal.
2003).OPM-POL:polarizationselectingdevice.OPM-SFK(SFH,SFJ):spatialfiltersfortheK-,H-,andJ-bands.OPM-ADC:correctorforthe
atmosphericdifferentialrefractionin H and J.OPM-ANS:cylindricalafocalsystemforimageanamorphosis. OPM-OSI:periscopetoco-align
thewarmandthecoldoptics.OPM-BYP:movablebypassdirectlysendingtheVLTIbeamstowardstechnicaltoolsortowardsthespectrograph
tocheckVLTIalignmentandacquirecomplexfields.SPG-INW:inputwheelwithimagecoldstopanddiaphgraminsidethespectrograph(SPG).
SPG-PMW:pupilmaskwheel.SPG-IPS:beamsplitterallowingtheseparationbetweeninterferometricandphotometricbeams.SPG-DIU:light
dispersion(gratingsorprism).SPG-CHA:SPGcamera.DET-IDD:chip.SPG-CSYandDET-CSY:cryostatsoftheSPGandoftheHawaiidetector
(DET).Duringfinaloperation,thetwocryostatsareconnectedbyacoldtunnelandsharethesamevacuum.
two,whichwouldresultinthelossofapartoftheH-band.The Thespecificationsonspatialfilteringaredrivenbythequal-
spatial filtering modules SFJ (H, K) (Fig. 4) receive the light ity of the optical fibers. The fiber length of about 1.30 m en-
fromdichroics.Parabolicmirrorsinjectthelightinsilicatebire- suresagoodtransmissionwhilemaintaininga10−3 attenuation
fringent single-mode fibers. At the exit of the fibers the same of the high order propagationmodes (Malbet et al. 2003). The
opticalsystemisrepeated. polarization-maintainingisachievedbyfiberswithanelliptical
core causing the so-called “form birefringence”, or by strong
birefringence caused by two stress members applied on oppo-
3.1.1. Theopticalfibers site sidesofthecore(bowtie-andpanda-type).Thefiberscre-
ate an intensity modulationin some of the images recordedby
LAOGprovidedequalized J-fibers, H-fibers,and K-fiberswith
AMBER.Figure5showsthismodulationforopticalfibersused
a maximal accuracy of ±20 µm. The characteristics of the previouslytothoseofTable1.Thiseffect,fainternow,ispresent
presentlyusedfibersaregiveninTable1.
16 S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument
Fig.5. Left: illustration of photometric images (extreme) and in-
terferometric image (center) produced with the artificial source
(H-band, ∆λ = 150 nm). The zero OPD is not exact.
Note the intensity modulation generated by the fibers. Right:
cross-sections of polarization maintaining fibres: elliptical core,
bow tie, and panda fibers (http://www.highwave-tech.com/;
http://www.fibercore.com/).
Fig.3.PictureoftheAMBERinstrumentattheendoftheintegration
atParanalinMarch2004(byA.Delboulbé,LAOG).
Fig.4. Pictures of the K-spatial filter entrance. Left (picture by
Y.Bresson,OCA):thethreebeamsmeetthedichroicsandtheparabolic
off-axismirrorsbeforetheinjectionthroughtheSibirefringentsingle-
modefibers;Right(picturebyA.Delboulbé,LAOG):theopticalcon-
figurationisrepeated attheexitofthefibers.Diaphragms control the
beamsizeattheexitofthefibersandshuttersselecttheinterferometric
arms.
Fig.6.Aboveisthesurfaceprofileasmeasuredwiththemicro-sensor
of SAVIMEX on a test element: maximum roughness of 60 nm PTV
Table1.ManufacturerandcharacteristicsofthefibersintheK-,H-,and andrmsroughnessofRa6.5nmandRq8.2nm;Belowisthepicture
J-bands.NA:numericalaperture,λ :cut-offwavelength,(cid:3):diameters, ofthemirrorallowingthemanufacturingofthethreeinjectionopticsin
Conc.:coreconcentricity. c theSFJasobserved through themicroscope (fieldof 400×300µm).
ThePTVroughnessisabout1/4thto1/8thoffringe,i.e.,80to40nm.
Thermsroughnessisabout6nm.Thesurfaceopticalqualitymeasured
K-band Highwave NA=0.16;λc=1900nm withthecollimatinglunetteisbelow30nm,i.e.,λ/20PTV@633nm
Silica Core(cid:3)=9.7µm;Conc.<5µm
onan18-mmdiameterdisk.
Ellipticalcore Coating(cid:3)=245µm
Absolutelength=1.30m±0.01m
Lengthdifferenceafterpolishing:≤11µm
H-band Fujikura NA=0.15;λ =1150nm 3.1.2. Fiberinjection
c
Silica Mode(cid:3)=5.5µm@1300nm;Conc.0.2µm
Pandacore Coating(cid:3)=245µm The parabolic mirrors, designed by the French company
Absolutelength=1.30m±0.01m SAVIMEX (Grasse), are metallic off-axisdiamond-turnedmir-
Lengthdifferenceafterpolishing:≤12µm rors.TheywerecontrolledbySAVIMEXusingaprocedurede-
J-band Fibercore NA=0.14;λ =944nm veloped in collaboration with OCA using a microscope and a
c
Silica Mode(cid:3)=6.3µm@1060nm;Conc.0.28µm micro-sensorfortheroughness,andacollimatinglunetteforthe
Bowtiecore Coating(cid:3)=245µm surfacecontrol(Fig.6).ThefocallengthFisrelatedtothefiber
Absolutelength=1.30m±0.01m numericalapertureNAandthebeamdiameterD.Thebestcou-
Lengthdifferenceafterpolishing:≤20µm plingefficiencywasgivenbyZemaxforNA.F/D=0.46,taking
thetelescopeobstructionintoaccount.
3.1.3. OPDcontrol
in the science as well as in the calibrator source and can then
be reduced below the specifications in the calibration process. ThissystemwasdesignedbyOCA,andmanufacturedandtested
However,it dependson the fiber temperatureand on the injec- at the Technical Division of the Institut National des Sciences
tion conditions (conditioned by the Strehl ratio). As far as the de l’Univers (DT/INSU, Paris). The static OPD in K is con-
highest accuracygoals are concerned(especially that of reach- trolled with a 4 µm accuracy within a few mm range. While
ingveryhighaccuracydifferentialphasemeasurements),a fast thestaticOPDisadjustedforagivenwavelengthineachband,
correctcalibrationofthiseffectisnecessary. thereexistsanOPDdriftbetweenwavelengths.AMBERneeds
S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument 17
to correct this differential OPD in J- and K-bands respective 3.1.6. Polarizationcontrol
to the H-band (ESO fringe sensor unit functioning)during the
Thefollowingwasdonetominimizethepolarizations’effects:
observations. This chromatic OPD due to the atmospheric re-
fraction is introducedduring the telescope pointing. It is given – Use the same number of reflections/transmissions between
by: Bsinz(n(λ1) − n(λ2)), where n(λ) is the refractive index the2interferometerbeams.
of the atmosphere, B is the baseline, and z the zenithal an- – Requireidenticalcoatingsfortheopticswiththesamefunc-
gle. Considering the extreme case for which B = 200 m and tions between 2 beams: same substrate, same structure for
z=60◦,thenecessaryOPDrangeisabout30µminKand70µm thedifferentlayers,simultaneousmanufacturings.
in J.Suchadjustmentsareachievedthroughdriftsoftheentire – Usepolarization-maintainingfibers.
AMBER K, H, or J spatial filter entrance parts. They are per- – Selectonepolarizationdirectionatthefiberentrance.
formedeveryfewminutes. – Controltheincidentangleonreflectingopticswithanaccu-
racybetterthanafractionofadegree.
Nevertheless, such translations cannot compensate for any
chromaticOPDgappresentinsideeachspectralband(δλequal Prior to the spatial filters, the polarizersselect onepolarization
to 33 nm in K, 32 nm in H, and 24 nm in J). This chromatic directiontogetridof thecross-talkinside thefibersandofthe
OPDgapisintroducedwhenadifferenceintheglassthickness phasedifferencebetweenbeams(variabledifferencesduringthe
or in the fiber length between two interferometric arms exists. telescope pointing). The selected direction is that which is not
Limitingthisrelativethicknessto0.5mm,thecontrastfactoris affected by the multiple reflections inside the instrument (per-
thus ensured of being less than 0.99. The relative fiber lengths pendiculardirectiontothebeampropagationplane).
inallthespatialfilterswerecontrolledbyLAOG(Grenoble).It Eachpolarizerisassociatedwithoneblade.Theorientation
was shownthatatthe minimalresolutionof AMBER, thecon- ofthetwoelementsismechanicallycontrolledtoensurethedi-
trast degradation factor due to differential dispersion is better rectionofthelightbeamrelativetotheopticalaxis.Theairblade
than0.99insideeachspectralband(Robbe-Duboisetal.2003). located between the two prism constituents of each polarizer
mustbeparallelfortheopticalsystem torespectthechromatic
dynamicOPDspecifications(seeSect.3.1.4).
3.1.4. RapidOPDvariations The relative polarization control also has an impact on the
opticalcoatingquality,in particularfor the dichroicselements.
Instabilities of the VLTI beams can generate rapid achromatic Alltheopticswiththesamefunctionsaresimultaneouslycoated
and chromatic OPD changes. The first type of problem results bythemanufacturer,inparticularthedichroicsandtheinjection
inadeviationoftheentirefringepatternsideways.Thisiscom- parabola located between the polarizersand the fiber entrance,
pensatedbytheVLTIitself.Thechromaticchangesdegradethe to reach a flux difference of a few % after each reflection or
interferencepattern,curvingthefringeshapeatthetimescaleof transmission and a minimal phase difference generated by the
these instabilities. It leads to the presence of a blurred pattern, differentlayerthicknesses.
especiallyatthesidesofeachspectralband(eveniftheOPDis At the spatial filter exits, an error on the dichroics layer
wellstabilizedatthecentralwavelengthofeachspectralband). thicknesscouldgenerateacontrastdegradation.Thenumberof
Themainsourceofsuchinstabilitiescomesfromthedifferential elements being small (1 reflection/arm in K, 1 reflection/arm
positioningshiftsoftheVLTI beams,combinedwith thetravel and 1 transmission/arm in H, and 2 transmissions/arm in J),
of light towards wedged dispersive glasses. Consequently, the the overall effect is almost null. From Puech & Gitton (2005):
requirementisto ensurethatdynamicchromaticOPD rmsval- a typical 2% of error on the layer thickness generates less
ues are lower than the uncertainty that comes from the funda- than1%contrastlossinK.Theneutralaxesatthefiberentrance
mental noise levels. The goal is to reach the (mainly photon) are controlled with a ±3◦ accuracy to compensate for the po-
noisecorrespondingtoa1-minmeasurementwitha5-magstar larizationdirectionrotationsgeneratedbytheresidualmanufac-
(observing with 2 UTs and the AO of the VLTI). In terms of turing differences between dichroics and by the incident angle
chromaticOPD, betweenthe centralwavelengthandthe wave- differences. This ensures the maximum coherent energy inside
lengthatthesideoftheconsideredbandwidth(mostdemanding thefibers.Atthefiberexit,thedifferentialpolarizationbetween
case: R = 35), this translates into OPD(noise) = 5.8 e−11 m beamsbeforecombiningiscontrolledtowithinafewdegrees.
in K. Vannier et al. (2002, 2004) performed a complete study
that led to the instrument requirement analysis concerning the
3.1.7. Acousticperturbations
wedge angle of the prismatic optics, and the surface quality
and the relative thickness of the dispersive elements located Acousticperturbations(duetostep-by-stepmotorsforinstance)
prior to the fibers. This study included elements such as opti- can modify the behavior of optical fibers that are sensitive to
calfibers, polarizers,anddichroics,butalso thoseof theauxil- pressurevariations.Itcanbeshownthatatypicaltalkproduces
iarymodulesdescribedbelowinthispaper,suchastheNeutral a 60 dB acoustic intensity, which implies a phase instability
Densities(NDN),theMatrixCalibrationSystem(MCS),andthe of10−7rad,farbelowthespecification(Perraut,internalreport).
AtmosphericDispersionCorrector(ADC). ToavoiddisturbingotherVLTIequipmentintheinterferometric
laboratory,each instrumentdoes notgenerateacoustic noise in
excessof40dBat2minallthedirections.
3.1.5. Dichroics
4. Fromthespatialfilterstothedetector
The dichroics satisfy the photometric requirements: ≥0.95 in
reflection in the highest spectral band and ≥0.90 in transmis- Thegeneralparametersofthemodulesfromthespatialfiltersto
sioninthelowestspectralband.Anabsorptionispresentatthe the detector are given: beam configuration, pupil diameter and
endofthe H-band(transmissionfrom86%at1800nmto80% separation, spectral resolution, and signal sampling. The mod-
at1850nm),butitdoesnotaffecttheglobalthroughput. ulesarethendescribedindetails.
18 S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument
Fig.7. AMBER pupil configuration in J, H, and K with no
anamorphosis.
Fig.8.Illustration:simulatedopticaltransferfunction(OTF)inwhich
thecoherentandincoherentenergypeaksarenotcompletelyseparated.
The pupil separation ischosen such that thefringe peak center isnot
4.1.Generalparametersofthesystemmodules
affectedbytheincoherentpeak.
Beam configuration: AMBER is a “dispersed fringes” instru-
ment operating in the image plane. The three-pupil configura-
tion (Fig.7)isa nonredundantline set-up.Themaximalbase- spectrograph(SPG)throughaperiscopeandafocalizingoptical
lineislargerthantheminimalonebyafactorthree.Thispupil componentadjustingbothaxesandimagepositionsattheinter-
configuration produces three systems of fringes corresponding face between the warm optics and SPG. The anamorphoser is
to the following baseline: B , 2B , B = 3B (the indices m a Chretien hypergonarmade of an afocal system of two cylin-
m m M m
andMmeanminimumandmaximum,respectively). drical mirrorsinserted in a parallelbeam section. The anamor-
Pupil diameters and spectral resolution: the pupil size de- phosis factor, 15.6, is the ratio between the focal length of the
pends on the spectral resolution and on the number of grooves twomirrors.Theanamorphosisdirectionisperpendiculartothe
per mm of the grating. The central wavelength (λ = 2.2 µm baseline.Suchanafocalsystemcontainsaneutralpoint,located
0
in the K-band)involvesa limitation of the numberof lines per nearbyafterthefocalplaneofthesmallest(2nd)cylindricalmir-
millimeter for the spectrograph grating (about 500 lines/mm). ror.Byputtingtheconjugateofthespectrographcoldstopatthis
To optimize the recordedflux, the spectral channelis a bit un- neutral point, we avoid having the ANS introduce a difference
dersampled (λ /D on one detector pixel). To obtain a spec- between the longitudinal and transverse pupil positions, mini-
0
tral resolution of 10000, the pupil diameter in the K-band is mizingtheaberrationsandimprovingthecoldstopbaffling.The
D=40mm.Thefringesamplingisthesameforallthespectral difficultylayin themanufacturingofthe 1stopticsoftheANS
bands.Thisimpliessmallerpupildiametersintheotherspectral shared by the 3 interferometric beams: a conic 220 × 50 mm
bands(30mmforthe H-bandand23mmforthe J-band).The rectanglewitha2-mcurvatureradiusinitslengthdirection.The
instrument pupil is set by a cold stop inside the spectrograph. PTVopticalqualityof633nm/5wastestedonaspecificoptical
Thispupilplaneiscombinedwiththeneutralpointofthecylin- benchatOCA.
drical optics, roughlysuperimposedto the pupilmasks located
after the collimating parabola at the exit of the spatial filters.
4.3.Thespectrograph(SPG)
Theprecisionofthisconjugationhasanegligibleimpactonthe
performanceofthissinglemodeinstrumentwithafieldofview ThecoldspectrographSPGincludesthefollowingfunctions:
limitedtoanAirydisk.
– Filteringofthermalradiationattheinputimageplane.
Pupilseparation:intheopticaltransferfunction(OTF),the
– Formationofaparallelbeam.
fringes with the lowest frequency produce a coherent energy
– Accuratespatialfilteringofthepupils.
peak close to the incoherentenergypeak(Fig. 8). To avoidthe
center of this fringe peak to be affected by the central single – Separation of interferometric beams from photometric
pupil peak, the minimal baseline B is: B > D(1+ λ /λ ). beams.
m m M m
For a full band observation in the K-band (λ = 2.0 µm and – Spectralanalysisatthreeresolvingpowervalues.
m
λ = 2.4 µm), B must be greater than 1.2D. For this reason, – Formation of images on the detector plane. The list of the
M m
SPGisgiveninTable2,withtheproducttreedefinitionsand
thedistancesbetweenthethreepupilsare1.3D,2.6D,and3.9D.
functions.
Signal sampling: anamorphosis factor and camera focal
length:consideringdata reductionrequirements,itisnecessary Theopticsandtheopticalbencharecontainedinavacuumtight
to sample the fringes produced by the combination of the fur- cryostat that allows the cooling of all functionsto the working
thest beamsby aboutfourpixelson the detector. Each spectral temperatureofabout77Kbymeansofliquidnitrogenatatmo-
element(λ0/D) is analyzed by one pixel. The magnificationof spheric pressure. The spectrographcryostat does not lodge the
the beams between the spatial direction and the spectral direc- detectorthatstaysinasecondcryostat;thetwocryostatsareme-
tion must be different. This anamorphosisfactor a is given by: chanicallycoupled,theysharethesamevacuumandworkatthe
a = 4BM/D = 15.6. The size p of the detector pixel, equal same temperature, but have two independent cooling systems.
to 18.5 µm, is linked to the camera focal length fc by the re- The coupling is achieved by a flexible bellow and an interface
lation 4 p = a fc (λ0/BM). The fc parameter is then deduced structure that allows for a certain degree of angular and linear
from: fc = p/(λ0/D)≈350mm. adjustmenttoalignthedetectoritselftothespectrographoptics.
TheopticaldesignofSPG(Fig.9)followsthegeneralpatternof
the grating spectrograph,with the necessary modificationsdic-
4.2.Anamorphosersystem(ANS)
tatedbytheopticalandmechanicalaccuracyrequestedbyinter-
At the exit of the spatial filters, the beams enter the ferometry.AmoredetaileddescriptionofSPGisinLisi(2003).
cylindricalopticsanamorphoser(ANS)beforeenteringthecold
S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument 19
Table2.ListofAMBERspectrographmoduleswithacronymdefinition For the purpose of spectral analysis, a rotating device
andfunctions. (SPG-DIU) allows forthe choiceamongthree dispersingcom-
ponents: two gratings (497 g/mm and 75 g/mm for respective
Opticalelements Functions resolutionsof10000and1500),andaprism.Thesupportofthe
SPG-INW Inputslit. gratingsandtheprismismotorizedtoallowustoselectthespec-
InputWheel Imagecoldstop.
tralrangeandresolution.Theangularaccuracyofthissupportis
SPG-CSY Spectrographopticscooling. about3 pixels of the detector (≈30(cid:2)(cid:2)), which impliesa spectral
CoolingSystem Vacuum,temperature,andnitrogen
calibrationprocedure.
levelcontrol.
Thermalfluxreduction. The final optical function before the detector is the camera
SPG-ISD Coldstopforthespatial SPG-CHA, designed around three mirrors. It is composed of
ImagingandStopping filtermode(thermalfluxreduction). twoasphericalmirrorswithatotalfocallengthof350mm(see
Device Technicaloperations(widediaphragm). Sect. 4.1.4)andanapertureofaboutF/2.A planemirrorsteers
Calibration(dark). the beams coming from the camera unit to send it to the DET
Beamcollimation. cryostat.AspectralfilterisinsertedintheJ pupilmasktoelim-
SPG-PMW Thermalfluxreduction.
inatethebackgroundcomingfromtheK-band,whileobserving
PupilsMasksWheel Beamsizedefinition.
withthesecondspectralorderofJ.
SPG-IPS Thermalfluxreduction.
InterferometricPhotometric Beamsplitting.
Splitter Beamsdeviation.
4.3.2. Opto-mechanicsandperformance
PhotometricimagesdeflectiononDET.
SPG-DIU Spectraldispersion. Therequirementsfortheopticaldesignreflecttheneedtomount
DispersionUnit Spectralwavelengthselection.
alltheopticsinsideacryostat,wheretheaccessibilityforalign-
Spectralresolutionselection.
mentis reducedandthedisplacementsofcomponentsafterthe
Spectralmodulation.
cooling are very large. The tolerance analysis shows a fringe
SPG-CHA Beamcombination.
contrast degradation factor of about 95% under the following
Camera Beamfocalization.
Sampling. constraints: a total positioning error of the optical elements
of 0.1 mm, an angular positioning error (tilt) of 2.3 mrad, and
a quality of optical surfaces better than 5 fringes in focusing
and 1 fringe in irregularity.This performancedependsonly on
AMB-SPG-CSY AMB-SPG-CHA
theuseofthedetectorpositionalongtheopticalaxisasacom-
pensator, with a total displacement of less than 2.3 mm with
respect to the nominal position. The surface quality (micro-
roughness) of the metallic mirrors has an impact on the effi-
ciency. The machining of the aspherical and plane mirrors al-
AMB-SPG-DIU AMB-SPG-PMW lowed us to produce the respective roughnesses of less than
10 nm and 5 nm rms, to ensure that the loss of light be less
AMB-SPG-IPS
than 1.5% on each mirror. One feature of the opto-mechanical
AMB-SPG-INW design is keeping the optics aligned at room temperature and
atliquidnitrogentemperaturewithoutadjustments.Testsofthe
Fig.9. Optical design of the SPG. For the acronym definition see SPGopticseitheratroomoratoperativetemperatureshowedno
Table2. significant differences of performance between the two sets of
measurements,confirmingthedesignconcept.
4.3.1. SPGopticalconfiguration 4.3.3. Vacuumandcryogenicsystem
The OPM-OSI module (Fig. 2), located at the side of the
Thewholesystemislodgedinsideavacuum-tightcasemadeout
SPG cryostat,injects the beam into the spectrographthrougha
of weldedsteelplateswithsuitable reinforcingribs.Theliquid
CaF window.Thebeamisspatiallyfilteredbytheslit,mounted
2 nitrogen vessel is a box-like structure (worked out of a single
onawheelthatalsocarriesseveralcomponentsusefultothecal-
aluminum block completed by a welded cover), whose bottom
ibrationandalignmentoperations.
plate is the cold optical bench. The external case supports the
Acollimatingmirrorproducestheparallelbeam.Thecolli- coldbenchbymeansofanhexapod(composedofsixfinesteel
matingmirroris a symmetricparaboloidalmirror(with respect beams),dimensionedtoallowtheSPGsystemtobeplacedona
tothefocalpoint)ofthefocalizingmirror;thisconfigurationhas side withoutundergoingpermanentdeformation.Allthe optics
thepurposeofminimizingtheopticalaberrations.Ithasafocal areenclosedintheradiationshield,intightthermalcontactwith
lengthof700mmandanapertureofF/3. thecoldbench.Themodelofthethermalbehaviorshowsthatall
The wheel SPG-PMW carries the pupil masks, located on itspointsareatmosttwodegreesoverthecoldbenchtempera-
the plane of the pupilimage, which help to minimize the stray ture; this is confirmedby measurements.The SPG opticalsys-
thermalbackground.TheopticalcomponentSPG-IPSisabeam temincludesatotalofthreemovingwheels,theaperturewheel,
splitter that steers part of the flux coming from each telescope thepupilmaskwheel,andthegratingwheel.Tosimplifytheen-
from the parallelbeam for photometriccalibration, leaving the gineering, these three functions are implemented by cryogenic
largestfractionofthefluxfortheinterferences.Thebeamsthat motors (Berger Lahr 5-phase), modified according to the ESO
carry the photometric information run along the same optical experience.Positioningoftheassociatedwheelsisperformedby
pathastheinterferometricbeams,buttheyaresuitablydirected a worm-wheelgearthat is substantially irreversibleand acts as
toformseparateimagesonthedetector. astoptotheforceexertedbythespiralspringusedtoovercome
20 S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument
cable for power and RS232
J
H
fiber optic
cable (double)
K
elec-
detector dewar tronics
housing detector LCU
(SUN UltraSparc)
fiber optic cable P1 P2 In P3
(single)
Fig.11. Image with the CAU light: dark current, photometric
sensor signals beams(P*),andinterferometricbeam(In).
Fig.10.PhysicaloverviewoftheAMBERdetectorhardware. 4.4.3. Detectorcharacteristics
Detectortype:RockwellHAWAII-1FPA(focalplanearray,one
thebacklash.Thetemperatureofthethreemotorsisconstantly quadrantinuse)
monitoredbythecontrolsystemusinglocalPt100sensors.The Detectorsize:512×512pixels
nitrogen vessel can store up to 19 dm3 of liquid N , while the Pixelsize:18.5µm×18.5µm
2
massofaluminumtobecooledisabout24kg.Thedesigntotal Operatingtemperature:77K
thermalloadisoftheorderof25W.Aturbo-molecularpumpes- Quantumefficiency:>50%for1−2.4µmwavelength
tablishestheoperativevacuumandinnormaloperation,asmall Other properties of the detector chip were measured and are
quantityofactivecharcoalkeepstheinternalpressureatthelevel listedbelow:
of2×10−5mbarforseveralmonths.Theliquidnitrogensupply -detectornumber:#159;
lastsforabout30h. -fullwellcapacity:63670e-;
-1%nonlinearity:26289e-;
-conversionfactor:4.70µV/e-;
4.4.Thedetector(DET) -readoutnoise(CDS@500kHz):11.6e-;
-numberofbadpixels:1489;
4.4.1. Hardwareoverview
-clustersofbadpixels(≥4):≈10.
The detectoris locatedin a dewar that is cooled downto 77 K
withliquidnitrogen.Thedetectorelectronicshousingisdirectly
4.4.4. Imageonthedetector
attached to the dewar to avoid electronic interference resulting
fromlongsignalpaths(Fig.10).Itisconnectedtothesensorby Figure 11 shows one image recorded on the detector with the
twoshortcables.Inadditiontothis,itcomplieswithchallenging CAUlight.Fromlefttorightarevisualizedthedarkcurrent,two
constraints concerning heat dissipation, interference, and elec- photometricbeams, the interferometricbeam,andthe 3rd pho-
tromagneticcompatibility,forexample.Thepowersupplyisnot tometricbeam.Thefluxinthe J-bandwasnotoptimized.
installedclosetothedetectorelectronicshousing.Adistanceof
amaximumof15misallowedbetweentheelectronicshousing
andthepowersupply.Theconnectionismadebyasinglepower 5. Auxiliarymodules
cable. Thiscable also containsthe galvanicallyisolated RS232
5.1.Remoteartificialsources(RAS)andcalibration
seriallineforcontrollingtheelectronics.Thepowersupplyrack
andalignmentunit(CAU)
isinstalledintheinstrumentcontrolcabinetinthestorageroom.
Thedigitalimagedata istransmittedto thedetectorLCU viaa Artificial sources are provided in the module RAS for align-
fiberopticscable. mentsinthevisible,fluxandOPDcontrol,contrast,andspectral
calibrations. These sources are: one laser diode and one halo-
gen lamp allowing alignments and calibration of the matrix of
4.4.2. Functionaloverview
the “pixel to visibility” linear relation (P2VM). The halogen
Thedetectorelectronicsconsistsofthefollowingmodules: source feeds two differentsingle-modefibers, one dedicatedto
theK-band,theothertotheJ-andH-bands,totransportthelight
1. An infrared detector (HAWAII-1 focal plane array from up tothe CAU, whichcan simulatethe VLTIin theintegration
Rockwell). and test phase (Fig. 12). The use of the same fiber in J and H
2. A sequencergeneratesclock patternsnecessary for reading is a compromisesolutionallowingustosave space andmoney
and sampling the sensor. It can be configured by data sent withoutlosingtoomuch(afewtensofa%)injectedlightinboth
throughthegalvanicallyisolatedserialline(RS232).Italso bands.TheexitsofthetwofiberscomingfromtheRASprovide
generatesa headercontaininginformationaboutimagefor- almostpoint-likesourcesin J,H,andK.
mat,readoutmode,etc. The retained optical configuration of the CAU chosen on
3. Aclockdriverbooststhedigitalsignalsfromthesequencer. the basis of generatingan achromaticpath lengthuses a wave-
4. A video amplifier supplies all necessary bias voltages and frontdivisionconfiguration.ThespectralbeamsexitingtheRAS
preparestheanalogsignalfromtheIRsensorforsampling. fibersarecollimatedandrecombinedinaglobalwavefront.The
5. AnanalogtodigitalconverterADCsamplestheanalogsig- latter is magnified and divided in three parts via a set of plane
nal. On the ADC board there is also digital logic for aver- mirrors to be injected inside AMBER via a movable 45◦ mir-
aging several samples (subpixel sampling). A fiber optical ror. The equivalent K-magnitudeof the CAU was estimated to
transmitter on the same board feeds the image data into a be −1.4 for the medium beam of AMBER and 0 for the ex-
fiberopticscableconnectedtothedetectorLCU. treme beams (different because of the Gaussian distribution of
6. Apowersupply. theglobalwavefront).Theinstrumentalcontrastgeneratedwith
S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument 21
Fig.13.RelativeaxialrotationofoneADCsystemrelativetotheother
oneatz=0◦andatz=60◦.
Fiber head 1-µm diameter
z=61°
H
J
z=50°
Fig.12.CalibrationandAlignmentUnit(CAU). z=0° H
H J
J
the CAU light is not 100%. Given that the cores of the fibers
have a finite dimension, the CAU does not provide perfectun- Fig.14.CombinedopticalandpositioningerrorsoftheADC.Thespots
representtheimpactsofthebeamfor3wavelengthsinsideeachband,
resolvedsources.Nevertheless,thiscontrast(from0.75to 0.87
J(blackspots)andH(whitespots),for3valuesofthezenithalanglez,
dependingonthespectralresolutionandbaseline)istakeninto
andtakingintoaccountthefollowingerrors:prismanglesmanufactur-
accountin the P2VM procedureand does not affect the instru-
ing, error on the prism gluing, sensitivity of the axial rotation of the
mentalcontrastofAMBERilluminatedbytheVLTI. twogluedprismsrespectivetothefirstone,tip/tiltofthisrotationaxis,
The CAU light is also used to perform a calibration of tip/tiltofthe3-prismsassembly,sensitivityofthe360◦-rotationofthe
the medium spectral resolution. A blade can be inserted at the two ADC prism system, and inclination induced during this rotation.
spectrograph entrance generating Perot-Fabry-like interference Theworseexpectedcouplingdegradationfactoris0.85inJandH.
fringeswitha30%contrastandaperiodicityofabout0.05µm.
In low spectral resolution, we use the spectra as they are de-
finedbythethreeJ,H,andKdichroicstransmissioncurvesand ofAMBERrequireadifferentconceptionthantheRisleyprism
calibratedwithspectroscopicreferencestarsandlamps.Wesuc- generallyused.TheoriginalsystemofAMBERiscomposedof
cessivelyobservedthespectrumofeachindividualspatialfilter twosetsof3prismsrotatingwithrespecttoeachother(Fig.13)
byclosingtheshuttersofthetwoothers,whichyieldsasubpixel and inserted in each beam prior to the J and H spatial filters.
calibrationwith an accuracyof about0.01µm, enoughfor this EachsystemiscomposedofafirstprismaticbladeinBK7and
lowresolutionmode.Thehighestspectralcalibrationmodecan ofadoubletofbladesinSF14andF2gluedtogether.Duringan
onlybecalibratedusingspectroscopiccalibratorsand/ortelluric observation with the UTs of objects located from 0◦ up to 60◦
lineslikeanyotherhighresolutioninfraredspectrograph. from zenith, the correctionof the transversaldispersion is per-
formedbyanaxialrotationofthetwo systems.Duringthisro-
tation,themaximumimageshiftduetocombinedopticalmanu-
5.2.Calibrationsystem facturingandpositioningerrorsislessthan1µmontheentrance
To calibratethe P2VM itis necessaryto introducea controlled fiber heads so as not to affect the coupling efficiency by more
phase delay (between60◦ and 120◦) between the interferomet- than15%in J and H (Fig.14).Thiscouplinglossistakeninto
ricarms.Inthepresentstateoftheinstrument,thepiezoelectrics accountinthethroughputbudget.
usedforthechromaticOPDcontrolareaccurateenough(afew
nanometers)toperformthisphasedelay.Nevertheless,arequire-
5.3.1. Bypass(BYP)
ment of 10−4 rad on the repeatability of the phase value was
initiallydefinedtoreachsomedifferentialinterferometricgoals. A specific system (BYP) with a commutable mirror was de-
To achieve this performance, a specific set-up (MCS) was de- signedtobypassthespatialfilterssothatpreliminarymechanical
signed and could be used in a near future if necessary. It con- adjustmentsand opticalalignmentsin the visible couldbe per-
sists of couples of slighty inclined glass blades placed in each formedwithtechnicaltools(CCD,lunette)andinjectedtowards
beampath,thesecondonehavingatiltoppositetothefirstone the final detector(DET)via the spectrograph.Itis also usedto
(leadingtoaVglassshapeineachbeampath).Thissystemal- ensurethattheCAUfibersfortheJH-bandandtheK-bandare
lowsforverygoodtoleranceontheexactabsolutethicknessof superimposed.ItappearsthattheBYPcanbeusedonskyforac-
theblades,astheirthicknessvariationswillbecompensatedfor quisitionofcomplexsourcesortechnicalcontrolsduringVLTI
througha slightinclinationcontrolduringtheopticsmounting. troubleshootings.Thesamplingis23mas/skywithUTsandthe
Thistypeofconfigurationenablesarepeatablephasedelayeven non-vignettedfieldabout1as.
ifthesystemundergoessomeinclinationduringitspositioning.
5.3.2. Neutraldensities(NDN)
5.3.Atmosphericdispersioncorrector(ADC)
TwosetsofNDNcanbeinsertedintheVLTIbeamstoavoidthe
TheroleoftheADCistocorrectthedifferentialtransversaldis- detectorsaturation.Onesetischosenasafunctionofthetarget
persion of the atmospherein J and H. The specific constraints brightness.Twofluxattenuationsarepossible:10and102.
22 S.Robbe-Duboisetal.:OpticalconfigurationandanalysisoftheAMBER/VLTIinstrument
6. Opticalstudyandpresentperformance
TheperformanceofAMBERisgivenbythesignal-to-noisera-
tio(SNR)ofthevisibilityderivedfromAMBERinterferograms.
Malbet et al. (2003) showed that the instrument contrast V
inst
must be 80% (90%) and the optical throughput tA Ceff larger
than2%(5%)toreachthemagnitudespecification(goal)defined
for a fringe detection at SNR = 5. The parameter t includes
A
bothopticaltransmissionandthefibercouplingdegradationfac-
tor due to misalignments. Ceff is the fiber coupling efficiency
taken to be about 81%, considering the telescope obstruction.
TheotherconsideredparametersintheSNRcalculationare:
– E : flux of a zero-magnitude star at the considered
0
wavelength;
– m:expectedmagnitudeoftheobservedobject;
– S: telescope surface area (consideringa telescope diameter
of8mfortheUTsand1.8mfortheATs);
– t :VLTIopticalthroughput(20%in J,26%in H,and30%
V
inK);
– SR: Strehl Ratio. The SR in K for an on-axis reference
source is equal to 50% (when the science and reference
sourcesare1arcminaway,theStrehlinKisdividedbytwo);
– ∆λ:spectralbandwidth;
– τ: elementary exposure time (τ = 10 ms for the high ac-
curacy mode, 50 ms for the high sensitivity mode, and up
to100sforthelongexposuremode);
– η:detectorquantumefficiency(0.6);
– n:numberofpixelsforonevisibilitymeasurement(about16
for3telescopes).
The computationof the magnitudeassumes the use of a single
polarization.TheintegrationtimeτdependsonAMBERobserv-
ingmodes(Malbetetal.2003).
In this section, we present the allocations for the contrast
and throughoutof AMBER with the associated errorsbudgets.
The results of the optical study and the expected instrumental
stability are compared with measurements in laboratory. Then
Fig.15. Contributions to the throughput and instrumental contrast
theresultsofthetwocommissioningsatVLTIarediscussed. (Bsinz=51.2m).
6.1.AllocationsforthecontrastandthroughputofAMBER
numericalaperture.Theerrorbudgetonfibercouplingrequire-
The diagram of Fig. 15 summarizes all the identified ments was then comparedto VLTI performanceto validatethe
contributions to the throughput and instrumental contrast hypotheses. It was the input for the definition of the degrees
(B sinz = 51.2m).Moredetailsontheallocationsaregivenin of freedom necessary on each optical element located at the
Table3.Thethroughputresultsarebasedonconservativevalues entrance of the fibers and to achieve their detailed tolerance
from manufacturer information on coatings. The next sections analysis.
givetheerrorbudgetsonthefibercouplingandthecontrast.
6.1.2. Contrasterrorbudget
6.1.1. Errorbudgetonthefibercoupling
Results on contrast-error budget in the K-band (priority of
This paragraph gives a list of the errors that can contribute to AMBER) with a 35 spectral resolution (most demandingcase)
the degradation of the fiber coupling Ceff. To derive this error areshowninTable5.Thiserrorbudgetconcernsdifferentialer-
budget,analyticalsimulationswere performed(Escarrat2000). rors between interferometric beams. A degradation factor ρ is
TheseresultswereconfirmedwiththeraytracingtoolZemaxin suchthatthemeasuredinstrumentalcontrastV isequaltothe
inst
whichtheopticalinjectionoflightinsideafibercanbeestimated idealcontrastvalueV timesthisfactor(V = ρ V).Itwasas-
i inst i
(Wagner&Tomlinson2004). sumedinafirstapproximationthattheerrorswereindependent.
Table4showstheerrorbudgetonthefibercouplinginterms Someoftherequirementsconcerningstaticerrorswereanalyti-
of fiber axis inclination respective to the incident optical axis callyderived.Othersrequiringdynamicalanalysisneededsome
(tilt),lateralshiftofthefiberhead,anddefocuscomparedwith simulation. The latter considered a Gaussian wavefront going
thelocalizationoftheinjectionopticalfocalpoint.Couplingef- througha simple 2-telescopeinterferometerusingthe AMBER
ficiencieswithoutmisalignmentswerealsoestimatedaftermod- spectral bandwidths and resolutions. Four pixels were consid-
ification of the ratio F/D, where F is the injection optical fo- eredtosamplethefringes.Theconsideredpupildimensionsaf-
cal length and D the pupil diameter. This ratio depends on the ter anamorphosisin the horizontaldirectionwere 40×2.6mm