• CRIRES Optical Model for Wavelength Calibration V1.0


  •   
  • FileName: crires_model_f.pdf [preview-online]
    • Abstract: the location of the illumination in the detector focal plane at sub-pixel accuracy for a ... instant accuracy at the sub-percent level (1 pix over 1000 pixel arrays) ...

Download the ebook

ST-ECF
Instrument Science Report WIRE – 2005 – 002
CRIRES Optical Model for
Wavelength Calibration V1.0
Overview
P. Bristow, F. Kerber
M. Fiorentino & M. R. Rosa
February 2005
Abstract
We present a streamlined model of the CRIRES optical path that enables calculation of
the location of the illumination in the detector focal plane at sub-pixel accuracy for a
given wavelength and instrumental configuration. The instrumental configuration is
described in terms of the tips and tilts of optical surfaces, their optical properties and
environmental conditions. The physical model will be used together with a minimisation
algorithm that is capable of using multiple realisations of the model to find the
instrumental configuration which results in the optimal match between simulated
wavelength data and observed calibration exposures. This is a component of the
Wavelength for Infra-Red Echelles project. We are now ready to apply the model to
laboratory testing tasks for CRIRES. During this phase the parameterisation will be
refined and the optimisation technique introduced.
1. Introduction
In June 2004 the Space Telescope European Co-ordinating Facility (ECF) and ESO’s
Instrumentation Division and Data Management Division agreed on the project
Wavelengths for IR Echelles in which the expertise of the ECF’s Instrument Physical
Modelling Group (IPMG) would be used to contribute to the 2D wavelength calibration
of CRIRES. The mutually agreed work plan calls for the development of a model
describing the optical elements of the spectrograph based on engineering information.
This model will then be used to support the laboratory testing and qualification of the
instrument. In order to deliver a high fidelity model of the spectrograph, input data of
compatible quality is required.
CRIRES is a cryogenic echelle spectrograph covering the wavelength range from 950-
5300nm at high spectral resolution (Rmax~100,000). Pre-dispersion is achieved by means
of a massive, reflective ZnSe prism with a wedge angle of 15° and a length of about
150mm used in double path (Delabre 2001). An echelle grating with a blaze angle of
63.5° and a groove density of 63.5lines/mm provides dispersion in the main spectrograph,
whilst the detector plane is equipped with four 1024x1024 (512x1024 illuminated) 27µm
pixel InSb detectors.
Traditionally the wavelength calibration of spectrographs relies upon an empirical
approach. An exposure of a source, usually an emission lamp, with clear, laboratory-
calibrated features is obtained. The location of features on this wavecal exposure are then
matched to the catalogued wavelengths of the source, and a low order polynomial is fitted
to the data points to provide an empirical relation between positions on the detector and
wavelengths. A meaningful polynomial fit will require a sufficient density of useful lines
distributed over the wavelength range of interest. Since such an empirical polynomial fit
has zero predictive value outside the range defined by data points, a lack of calibration
lines at the limits of the wavelength ranges and detector boundaries is particularly critical.
We replace this empirical way of wavelength calibration by using our physical
understanding of the instrument. We know from the design process that even
sophisticated spectrographs can be accurately modelled (Ballester & Rosa, 1997 and
Ballester & Rosa, 2004). However, with the exception of the model driven bootstrap for
the still canonical wavelength calibration of the VLT/UVES instrument, that approach
has previously been perceived as to complex and to difficult to become the backbone of
the wavelength calibration of a pipeline.
In order to demonstrate the superior performance of an entirely model based approach ,
we have developed a package that replaces the empirical wavelength calibration parts of
the HST/STIS pipeline in its most difficult modes, the UV 2D echelle formats (Rosa et al.
2005). A fast ray trace instrument model is at the heart of this technique, allowing for
iterative evaluation for many wavelengths and orders. Parameters are optimized using the
Monte-Carlo type “Adaptive Simulated Annealing” technique (Fiorentino et al. 2004)
The design of the CRIRES optical model presented here is based on the generic kernel of
the HST/STIS package, with a major modification to appropriately trace rays through the
double pass pre-disperser. In order to avoid computationally expensive, but for the
present purpose unnecessary level of detail, those surfaces which do not affect the
relative geometry on the detector (eg. plain folding or pick off mirrors) are neglected.
However, the ray tracing at all relevant surfaces is performed by the appropriate 3D
matrix transformation. In order to bring the models full potential of accuracy to bearing,
we have acquired highly accurate data for the wavelength and temperature dependencies
of the refractive index of the CRIRES pre-disperser prism (Kerber & Bristow 2005).
The model in its version 1.0 as documented here can be used interactively as a powerful
predictive and investigative tool. During integration of CRIRES in the laboratory one
obvious use is to quickly simulate the distribution of light on the detectors for any given
source and instrumental configuration. In addition one can easily gain an insight into the
effects of varying a specific parameter. In combination with the Adaptive Simulated
Annealing optimizer the very same model kernel will then be able to support a model
based wavelength calibration in operational mode (cf. Rosa et al 2005).
The present document describes the parameterisation adopted and the input data required
for the CRIRES model V1.0.
2. Representation of CRIRES Optical Components
The optical design of CRIRES as presented in some detail in the CRIRES Optical Design
Report (Delabre 2001) was our main source of information regarding the layout and
specifications of the components. Here we give examples of how key optical components
are represented in the model and how we chose which components to include.
2.1. Choice of contributing surfaces
The purpose of the ray trace kernel is to simulate the relative geometry of the dispersed
light on the detectors. We are therefore interested in those surfaces that ultimately affect the
position at which a given ray intersects the plane of the detector array. In principle this
would include every surface in the instrument. However, in the interest of efficiency we
restrict the model to simply those surfaces that have wavelength dependent properties (the
pre-disperser prism and diffraction grating) and the entrance and intermediate slits and the
detector array itself.
We thus treat the instrument in the geometric optics, on-optical axis limit. Consequently
our model will not be able to predict from first principles any aberrations. We know from
the HST/STIS implementation (Rosa etal 2005) that this very basic approach provides
instant accuracy at the sub-percent level (1 pix over 1000 pixel arrays). Aberrations and
alignment issues, as will become measureable against real data from the instrument, will
then be dealt with case by case to reach the sub 0.1- percent accuracy required in closed-
loop operation in a pipeline. Small misalignments of the components that we ignore will
manifest themselves in the alignments of the components that we do consider. This is easy
to see in the case of plane mirrors that simply bend the light path. A slight rotation of an
intermediate plane mirror can simply be represented as a rotation of the frame of reference
for the subsequent surface. Similarly, in the case of surfaces with power, the effective
distances along the optical path between the surfaces present in the model are altered.
Summary for the current instance (V1.0) of the CRIRES model :
• None of the surfaces before the spectrographs entrance slit are modelled.
• Pre-disperser Section: - only the entrance slit, the prism itself and the intermediate order
selection slit are explicitly included. Any small rotation of the folding mirror will
manifest itself as a tip or tilt of the prism. The effect of the off-axis parabola is
indirectly included in the computation of the starting vector for a given position on the
entrance slit (see below).
• Main Spectrograph – Only the Echelle grating itself is included in this model. The
collimator/camera (a three mirror anastigmat) is not included, though it ultimately
affects the scale length of the projection onto the detector array (see below). Two
folding mirrors are also ignored here.
2.2.Component details
Entrance Slit
The entrance slit is a rectangular aperture, which can, in principle, be inclined at any
angle to the optical axis. In the model, light can originate from any position on this
rectangular aperture. In order to obtain an image of the entrance slit on the detector array
we simply run the simulation for many different, equally spaced, initial ray locations
covering the entrance slit. Alternatively we can simulate a given illumination pattern on
the slit by choosing random starting points from a probability function describing the
desired illumination pattern.
For a given initial ray position we calculate the unit vector in the direction of its
intersection with the optical axis at a distance equal to the focal length of the off-axis
parabola (listed here with the parameters of the entrance slit) preceding the prism and
take this as the starting vector. Diffraction effects are not considered.
Parameters Nominal Values
Slit Width [mm] 0.062
Slit Height [mm] 7.65
Position Angle [deg] 0.0
Rotation about x axis [deg] 0.0
Rotation about y axis [deg] 0.0
Focal length of off-axis parabola [mm] 1500.0
Table 1 : Parameters used to describe the entrance slit.
Prism Pre-Disperser
Pre-dispersion is performed by a ZnSe prism. The ability to rotate this prism also allows
the selection of the wavelength range by determining which part of the refracted spectrum
falls upon the intermediate slit. Essentially, the prism is tilted such that the central
wavelength of the desired Echelle order falls at the centre of the intermediate slit. An
additional use of this model is to find the exact prism tilt required for each grating order.
This is achieved by iteratively running the model for the required central
wavelength/order up to calculation of the intersection with the intermediate slit, each time
adjusting the angle (using Newton-Raphson convergence) until the ray passes through the
centre of the slit.
The prism pre-disperser is modelled as three surfaces. For each surface the ray is first
rotated into the frame of reference of the surface and then either refracted or reflected
depending upon the surface. Details of the computation of the refraction are given in
appendix A.
The refractive index of ZnSe, nZnSe, is itself dependent upon wavelength and temperature.
We have obtained data for nZnSe covering the appropriate range of temperatures (15-
310K) and wavelengths (540-5600nm) from the CHARMS (Cryogenic High-Accuracy
Refraction Measuring System) facility at NASA’s Goddard Space Flight Center (this data
was kindly provided to us prior to publication by Leviton and Frey, see Kerber & Bristow
2005). For any given wavelength and temperature we use a fit to the grid of
measurements covering these parameters. First we fit refractive index against wavelength
at each temperature using the Sellmeier formulation that describes well the variation of
refractive index with wavelength (see, eg. Ghosh 1998):
3
si "2
n 2 ( ") #1 = $
i=1 ( " # "i )
2 2
where n is the refractive index, λ is the wavelength and si and λi are constants determined
by a least squares fit. Then we perform a linear interpolation between temperatures
! bracketing the operational temperature. The data provided by CHARMS includes
measurements at 5K intervals, for CRIRES we are interested in the values at 65K, 70K,
75K and 80K.
Figure 1 displays the CHARMS nZnSe data in the planned operational temperature and
wavelength range of CRIRES. As expected for prisms, the higher gradient (dnznse/dλ) at
shorter wavelengths provides the increase of dispersion with decrease in wavelength.
Figure 1: nZnSe as a function of wavelength and temperature in the parameter space of CRIRES
operations
Parameters Nominal Values
Refractive index Wavelength and temperature
dependent, see text
Position Angle of prism [deg] 0.0
Rotation of entrance/exit surface about x axis [deg] 0.0
Rotation of entrance/exit surface about y axis [deg] Mode dependent, see text
Prism angle [deg] 15.0º
Table 2 : Parameters used to describe the pre-disperser prism.
Intermediate Slit
The intermediate slit selects the wavelength range that will be transmitted to the Echelle
grating. It is a rectangular transparent aperture. In the model rays that do not pass through
the aperture are not traced any further. Diffraction effects are not considered.
Parameters Nominal Values
Slit width [mm] 0.7
Slit height [mm] 40.0
Position Angle of slit [deg] 0.0
Rotation about x axis [deg] 0.0
Rotation about y axis [deg] 0.0
Table 3 : Parameters used to describe the intermediate slit.
Echelle Grating
The diffraction grating is modelled as in Ballester & Rosa (1997):
m#
v" =
x + vx
$
Parameters Nominal Values
-1
! Grating Constant [mm ] 31.6
Position Angle of grating [deg] 0.0
Rotation about x axis [deg] -64.11
Rotation about y axis [deg] 0.0
Table 4: Parameters used to describe the Echelle grating.
Detector Array
The detector array consists of four 512x1024 pixel CCDs (actually each is 1024x1024
pixels, however only one half of each detector is illuminated) aligned to form a combined
array of 512x4096.
Parameters Nominal Values
Focal length of collimator [mm] 1500.0
Position Angle of detector array [deg] 90.0
Rotation of detector array about x axis [deg] 0.0
Rotation of detector array about y axis [deg] 0.0
x offset of 1st detector from optical axis [mm] -53.622
y offset of 1st detector from optical axis [mm] 0.0
Position angle of 1st detector relative to detector array [deg] 0.0
x offset of 2nd detector from optical axis [mm] -17.874
y offset of 2nd detector from optical axis [mm] 0.0
Position angle of 2nd detector relative to detector array [deg] 0.0
x offset of 3rd detector from optical axis [mm] 17.874
y offset of 3rd detector from optical axis [mm] 0.0
Position angle of 3rd detector relative to detector array [deg] 0.0
x offset of 4th detector from optical axis [mm] 53.622
y offset of 4th detector from optical axis [mm] 0.0
Position angle of 4th detector relative to detector array [deg] 0.0
Table 5 : Parameters used to describe the detector array.
3. Preliminary Results
As yet there are no calibration data available, therefore we are unable to begin the
optimisation process in order to determine the detailed component configuration. However
we present here predictions made regarding the prism spectrum at the intermediate slit
which were requested during the assembly of the instrument, an example of the simulated
image of the slit on the detector and further examples that illustrate the potential for
supporting the CRIRES laboratory-testing phase.
3.1.Intermediate Slit
Here we use the model to compute the wavelengths passing the intermediate slit, the width
of grating orders at the intermediate slit and the size of the slit image on the detector. We
follow the Code V description and set the prism angle as 39.412º for order 35. However we
find that this leads to an intermediate slit location 7.1mm from the optical axis, different
from that in the Code V listing. In generating the results below we chose to fix this as the
intermediate slit position for all orders. For this reason these results should be considered to
be purely illustrative.
M λcen λmin λmax αprism λslit_bot λslit_top Worder xmin xmax Dprism
Å Å Å º Å Å Mm mm mm Å/pix
11 51468 49230 53919 37.58 49474 53461 0.8231 30.39 29.56 153.7
12 47179 45291 49230 37.61 45216 49141 0.7023 30.33 29.63 151.3
13 43549 41937 45291 37.63 41632 45467 0.6123 30.29 29.68 147.9
14 40439 39044 41937 37.66 38593 42284 0.5484 30.27 29.72 142.3
15 37743 36525 39044 37.68 35994 39492 0.5040 30.24 29.73 134.9
16 35384 34312 36525 37.69 33717 37051 0.4648 30.23 29.76 128.5
17 33302 32351 34312 37.71 31755 34850 0.4434 30.21 29.77 119.3
18 31452 30602 32351 37.73 30050 32855 0.4363 30.21 29.78 108.2
19 29797 29033 30602 37.74 28530 31063 0.4336 30.22 29.78 97.70
20 28307 27617 29033 37.76 27144 29470 0.4262 30.21 29.78 89.71
21 26959 26332 27617 37.78 25905 28013 0.4263 30.20 29.78 81.33
22 25734 25162 26332 37.79 24772 26695 0.4258 30.21 29.78 74.19
23 24615 24091 25162 37.81 23780 25450 0.4488 30.22 29.77 64.41
24 23589 23108 24091 37.83 22858 24320 0.4710 30.23 29.76 56.36
25 22645 22201 23108 37.84 21923 23368 0.4390 30.21 29.77 55.73
26 21774 21364 22201 37.86 21177 22372 0.4910 30.25 29.76 46.06
27 20968 20587 21364 37.88 20402 21534 0.4804 30.23 29.75 43.65
28 20219 19864 20587 37.90 19686 20752 0.4740 30.24 29.77 41.14
29 19522 19191 19864 37.92 19095 19949 0.5521 30.27 29.71 32.93
30 18871 18562 19191 37.94 18444 19298 0.5160 30.25 29.73 32.92
31 18262 17973 18562 37.95 17843 18682 0.4915 30.24 29.75 32.36
32 17692 17419 17973 37.98 17380 18004 0.6202 30.30 29.68 24.073
33 17156 16899 17419 38.00 16844 17468 0.5834 30.28 29.70 24.064
34 16651 16410 16899 38.02 16339 16963 0.5497 30.27 29.72 24.057
35 16175 15947 16410 38.04 15874 16477 0.5369 30.27 29.73 23.243
36 15726 15510 15947 38.06 15488 15964 0.6417 30.31 29.67 18.383
37 15301 15097 15510 38.09 15063 15539 0.6076 30.29 29.69 18.377
38 14898 14705 15097 38.11 14695 15102 0.6745 30.34 29.67 15.696
39 14516 14332 14705 38.14 14322 14710 0.6714 30.33 29.66 14.970
40 14153 13979 14332 38.16 13962 14345 0.6475 30.32 29.67 14.754
41 13808 13642 13979 38.19 13652 13964 0.7541 30.37 29.61 12.060
42 13479 13321 13642 38.21 13323 13636 0.7189 30.35 29.63 12.055
43 13166 13014 13321 38.24 13010 13322 0.6861 30.33 29.65 12.050
44 12867 12722 13014 38.27 12742 12991 0.8238 30.41 29.58 9.5854
45 12581 12442 12722 38.30 12458 12703 0.7966 30.39 29.59 9.4768
46 12307 12175 12442 38.33 12184 12430 0.7627 30.37 29.61 9.4727
47 12045 11918 12175 38.36 11926 12164 0.7554 30.38 29.63 9.1613
48 11794 11673 11918 38.39 11683 11905 0.7754 30.38 29.60 8.5570
49 11554 11437 11673 38.41 11443 11664 0.7444 30.36 29.62 8.5533
50 11322 11210 11437 38.45 11248 11397 1.0579 30.55 29.49 5.7801
51 11100 10993 11210 38.49 11030 11171 1.0799 30.53 29.45 5.4423
52 10887 10783 10993 38.53 10811 10963 0.9613 30.47 29.51 5.8807
53 10682 10582 10783 38.56 10605 10758 0.9259 30.45 29.53 5.8775
54 10484 10388 10582 38.60 10415 10553 0.9862 30.52 29.53 5.3156
55 10293 10200 10388 38.64 10229 10358 1.0177 30.50 29.48 4.9657
56 10109 10020 10200 38.67 10045 10174 0.9822 30.48 29.50 4.9628
Table 6 : Simulation results for central wavelengths of orders 11 to 56 for the intermediate slit.
Key:
m : Required spectral order for the echelle grating
λcen : Central wavelength for this order
λmin : Minimum wavelength of this order
λmax : Maximum wavelength of this order
αprism : Tilt of the pre-disperser prism which puts λcen at the centre of the intermediate slit
λslit_bot : Wavelength passing the bottom edge of the intermediate slit for αprism
λslit_top : Wavelength passing the top edge of the intermediate slit for αprism
Worder : Width of the order at the intermediate slit
xmin : Location of λmin on the plane of the intermediate slit
xmax : Location of λmax on the plane of the intermediate slit
Dprism : Dispersion of the prism for λcen
m λcen x’cen W’order S sbottom stop
Å mm mm mm mm mm
11 51468 27.80 723.5 547.2 -187.6 359.6
12 47179 27.80 623.9 601.9 -201.8 400.1
13 43550 27.80 556.2 653.4 -213.8 439.6
14 40439 27.80 504.6 691.3 -221.6 469.7
15 37743 27.80 463.1 709.2 -225.1 484.1
16 35384 27.80 428.7 729.9 -228.7 501.1
17 33303 27.80 399.5 712.4 -225.7 486.7
18 31453 27.80 374.3 666.4 -216.5 449.8
19 29797 27.80 352.3 620.5 -206.3 414.3
20 28307 27.80 332.8 591.7 -199.3 392.5
21 26960 27.80 315.5 554.1 -189.5 364.7
22 25734 27.80 300.0 523.0 -180.7 342.3
23 24615 27.80 285.9 464.8 -163.2 301.6
24 23590 27.80 273.1 418.2 -148.1 270.1
25 22646 27.80 261.6 432.5 -152.9 279.7
26 21775 27.80 250.8 364.8 -129.5 235.3
27 20969 27.80 241.1 358.3 -127.2 231.1
28 20220 27.80 232.1 349.4 -123.9 225.5
29 19522 27.80 223.5 285.5 -99.8 185.7
30 18872 27.80 215.8 295.9 -103.8 192.1
31 18263 27.80 208.6 300.9 -105.7 195.2
32 17692 27.80 201.5 227.7 -76.5 151.3
33 17156 27.80 195.3 235.1 -79.5 155.6
34 16651 27.80 189.4 242.5 -82.5 159.9
35 16176 27.80 183.8 241.1 -81.9 159.2
36 15726 27.80 178.3 194.5 -62.5 132.1
37 15301 27.80 173.4 200.1 -64.8 135.2
38 14899 27.80 168.6 174.8 -53.9 120.9
39 14517 27.80 164.1 171.0 -52.3 118.7
40 14154 27.80 159.9 172.9 -53.1 119.8
41 13809 27.80 155.7 144.1 -40.5 103.7
42 13480 27.80 151.9 147.7 -42.0 105.7
43 13166 27.80 148.4 151.2 -43.6 107.6
44 12867 27.80 144.6 122.5 -30.7 91.8
45 12581 27.80 141.4 123.9 -31.3 92.6
46 12308 27.80 138.3 126.6 -32.6 94.1
47 12046 27.80 135.3 125.1 -31.9 93.3
48 11795 27.80 132.3 119.2 -29.2 90.0
49 11554 27.80 129.6 121.7 -30.3 91.4
50 11323 27.80 126.3 83.1 -12.5 70.6
51 11101 27.80 123.7 79.8 -11.0 68.9
52 10887 27.80 121.5 88.2 -14.9 73.3
53 10682 27.80 119.2 89.9 -15.6 74.2
54 10484 27.80 116.8 82.6 -12.2 70.4
55 10294 27.80 114.5 78.5 -10.3 68.2
56 10110 27.80 112.5 80.0 -11.0 69.0
Table 7: Simulation results for central wavelengths of orders 11 to 56 for the detector array.
Key:
m : Required spectral order for the echelle grating
λcen : Central wavelength for this order
x’cen : Location of λcen in the plane of the detector array
W’order : Width of the order at the plane of the detector array
S : Width of intermediate slit image at the plane of the detector array
sbottom : Location of λslit_bot at the plane of the detector array
stop : Location of λslit_top at the plane of the detector array
3.2.Wavelength Calibration
The table below gives a very simple example of output useful for wavelength calibration.
Using order 35 and the nominal configuration values of section 3 we derive the x
(dispersion direction) pixel position for light of a given wavelength (passing through the
centre of the entrance slit). The table simply lists the results for 1Å intervals (about every
13 pixels), but we could equally well calculate this to higher precision, making
interpolation trivial.
Wavelength x Wavelength x Wavelength x Wavelength x
(Å) (pixel) (Å) (pixel) (Å) (pixel) (Å) (pixel)
16117 12 16138 270 16157 517 16177 779
16119 24 16139 283 16158 530 16178 792
16120 37 16140 295 16159 543 16179 805
16121 50 16140 308 16160 556 16180 819
16121 63 16142 321 16161 569 16181 832
16123 76 16143 334 16162 582 16182 845
16124 89 16144 347 16163 595 16183 858
16125 102 16144 360 16163 608 16184 871
16126 114 16146 373 16165 621 16185 885
16127 127 16147 386 16166 634 16186 898
16128 140 16148 399 16167 648 16186 911
16129 153 16149 412 16167 661 16188 924
16130 166 16150 425 16169 674 16189 937
16131 179 16151 438 16170 687 16190 951
16132 192 16152 451 16171 700 16191 964
16133 205 16153 465 16172 713 16192 977
16134 218 16154 478 16173 726 16193 991
16135 231 16155 491 16174 740 16194 1004
16136 244 16156 504 16175 753 16195 1017
16137 257 16176 766
Table 8 : Example, low resolution, wavelength/pixel calibration list
3.3.Comparison of Emission Lines from ThAr, Ne and Kr Calibration
Lamps
As a further example we use the model to provide a visual impression of the density of
lines that would be seen in a CRIRES exposure for three possible calibration lamps. In this
example we show only the wavelength range corresponding to grating order 44 (~1272nm -
~1301nm), as displaying the entire wavelength range of CRIRES, even at this modest
resolution would be impractical here. However, the range that we choose to show is by no
means untypical. Figures 2 a, b and c show the simulated detector array illumination for,
Krypton, Neon and Thorium Argon calibration lamps respectively. The dark regions
represent, to scale, the gaps between the chips in the detector array. Note that the lack of
lines in the bottom chip (shortest wavelengths) in figure 2c simply reflects the fact that our
input line list for Thorium Argon does not extend to this region. On the other hand, the
Neon and Krypton spectra do not have any emission lines in the wavelength range covered
by the bottom chip. The sources for the input linelists were, Hinkle et al. (2001), Sansonetti
& Blackwell (2004) and Saloman & Sansonetti (2004) for Thorium Argon, Krypton and
Neon respectively.
3.4.Entrance Slit Image
Figure 3 shows a section of a simulated CRIRES detector array illumination, actually taken
from the ThAr simulated data of figure 2c. The slit is uniformly illuminated and the ThAr
emission lines are modelled as delta functions at the wavelengths listed in Hinkel et al
(2001). The dispersion direction is vertical in this figure. Note that some curvature is
apparent in the slit images.
4. Input File specifications
4.1.Calibration Lamp Line Lists
Given a list of wavelengths and intensities as an ascii file named wave.dat, the model will
generate a corresponding list of locations of the lines in the detector plane and a FITS file
showing the entrance slit image. The line list should be two columns, tab separated as
follows.
1st column 2nd column
Quantity Wavelength Relative Intensity
Units Å NA
Data type Real Real
Table 9 : Specification for calibration line lest input file format
4.2.Component Configuration File
The optimisation process will generate the configuration files in ascii format. These are a
simple list of parameters in a predefined order. We give an example in Appendix C,
however, since the final list of parameters is not yet finalised, the list in appendix C is
purely for illustrative purposes and should not be used as a template. The parser does not
evaluate the string to the left of the “=” sign, these are simply to help the user. Instead the
parser assumes the given order and reads the value starting one space after the “=”.
a
b
c
Figure 2: Simulated detector array for calibration lamps in order 44. a) Kr b) Ne c) ThAr. Note that for ThAr no input data is available
below 1270nm whilst Ne and Kr simply do not have features in this part of their spectrum.
Figure 3: Close up of an area of the second chip in the simulated data of figure 2c (ThAr). The
dispersion direction is horizontal on the page, the spatial direction is vertical. Note the pronounced
curvature seen in the two-dimensional detector array.
4.3.Order vs Prism Angle File
The code for the instrument model has a mode which generates this file. It does so by
iteratively solving for the prism angle which places the centre of a given order at the centre
of the intermediate slit (see section 3.3 above).
1st column 2nd column
Parameter Order Prism Angle
Units NA Degrees
Data Type Integer Float
Table 10 : Specifications for input file containing default prism angles.
4.4.ZnSe Refractive Indices
The data are provided by CHARMS/GSFC (see 2.3 above). We fit Sellmeier coefficients to
their tabulated data for each temperature The model reads these coefficients from an ascii
file named znse.dat.
1st column 2nd column 3rd column 4th column 5th column 6th column 7th column
Parameter Temperature Sellmeier Sellmeier Sellmeier Sellmeier Sellmeier Sellmeier
coefficient coefficient coefficient wavelength wavelength wavelength
s1* s2* s3* λ1* λ2* λ3*
Units ˚K NA NA NA µm µm µm
Data type Float Float Float Float Float Float Float
Table 11 : Specification for Sellmeier coefficient input file.
3
* si "2
Sellmeier equation for refractive index: n 2 ( ") #1 = $
i=1 ( " # "i )
2 2
!
References
Ballester, P. & Rosa, M.R., 1997, A&A Supp, 563, 126.
Ballester, P., Rosa, M.R., 2004, ADASS XIII< ASP Conf. Vol 314, p 418.
Delabre, B., 2001, VLT Instrumentation Plan: CRIRES Optical Design Report,
VLT-TRE-ESO-14500-2096. ESO.
Downes, R., Hartig, G., and Plait, P., 1997, STIS Instrument Science Report ISR 97-06.
Fiorentino et al., ADASS 2004, ASP. Conf. Ser. in press
Ghosh, G., 1998 Handbook of Optical Constants of Solids, Elsevier Academic Press.
ISBN: 0-12-281855-5.
Hinkle, K.H., Joyce, R.R., Hedden, A., Wallace, L., 2001, PASP, 113, 548
Kerber, F. & Bristow, P., 2005, WIRE-2005-003, in preparation.
Leviton, D.B., Frey, B.J., 2004, private communication.
Rosa, M. R., Bristow, P., Fiorentino, M. & Kerber, F., 2005, WIRE-2005-001, in
preparation.
Saloman, E.B., Sansonetti, C.J., 2004, private communication.
Sansonetti, C. J. & Blackwell, M., 2004, private communication.
Appendix A: Refraction at prism surfaces
The prism pre-disperser is modelled as three surfaces. For each surface the ray is first
rotated into the frame of reference of the surface and then either refracted or reflected
depending upon the surface.
At the refractive surfaces we compute the angle of incidence, i, as:
2 2
vx + vy
i = arctan
vz
where vx, vy and vz are the components of the unit vector describing the optical path of the
ray passing through the instrument (see Rosa et al 2005 and Ballester & Rosa 1997 for a
! more detailed discussion of the reference frames that we use) and also:
#v &
" = arctan% x (
%v (
$ y'
We then apply Snell’s law to compute the angle of refraction:
sin i" = n sini
!
and require that the refracted ray is in the same plane as the incident ray and the normal
to the surface (i.e. the angle θ does not change):
! v z = sin i"
v x = sin # cos i"
v y = cos# cos i"
! Appendix B: Representation of detector array
The detector array consists of four 512x1024 pixel CCDs (actually each is 1024x1024
pixels, however only one half of each detector is illuminated) aligned to form a combined
array of 512x4096.
After rotation into the plane of the detector array, the projection onto the detector array is
computed as follows:
vx
x = f col
vz
vy
y = f col
v + v z2
2
x
where fcol is the focal length of the collimator. The x,y position on a chip (if the ray falls
on a chip) is then given by:
!
$ (x " x i )cos # i " (y " y i )sin # i '
offset offset
x pix = 0.5N x dim + &
& ) " 0.5
)
% spix (
$ (y " y i )cos # i " (x " x i )sin # i '
offset offset
y pix = 0.5N y dim + &
& ) " 0.5
)
% spix (
where Ny dim and Nx dim are the x and y array dimensions for each chip (512 and 1024)
respectively; xioffset and yioffset are the x and y offsets of the ith chip from the centre of the
! detector array respectively and θi the rotation of the ith chip with respect to the detector
array. In this way, if the bottom left pixel is (0,0) and x-xioffset=0 and y-yioffset=0, then the
ray will arrive at the precise centre of the chip. Note however that such conventions are
not critical when the model is used in the optimisation process in conjunction with real
data, the values off xioffset and yioffset will automatically converge to the correct values.
Appendix C: Configuration file
Example CRIRES configuration file:
%Any number of comments preceded by a percent symbol
mues = 0.0 % mu angle for entrance slit (deg)
nues = 0.0 % nu angle for entrance slit (deg)
taues = 0.0 % tau angle for entrance slit (deg)
es_s = 7.65 % entrance slit height (es_s x 2) (mm)
es_w = 0.062 % entrance slit width (mm)
%
fcp = -1500.0 % focal length of collimator (mm)
%
mup1 = 39.412 % mu angle for Prism entrance/exit surface
(deg)
nup1 = 0.0 % nu angle for Prism entrance/exit surface (deg)
taup1 = 0.0 % tau angle for Prism entrance/exit surface
(deg)
%
mup2 = -15.0 ;% mu angle for Prism reflection surface (deg)
nup2 = 0.0 ;% nu angle for Prism reflection surface (deg)
taup2 = 0.0 ;% tau angle for Prism reflection surface (deg)
%
%Intermediate slit
muis= 0.0 % mu angle for intermediate slit (deg)
nuis= 0.0 % nu angle for intermediate slit (deg)
tauis= 0.0 % tau angle for intermediate slit (deg)
fis= 750.0 % Focal length of prism collimator (mm)
slitx= 0.0 % intermediate slit x offset (mm)
slity= -7.8 % intermediate slit y offset (mm)
slitw= 40.0 % intermediate slit width (mm)
slith= 0.7 % intermediate slit height (mm)
%
mug = -64.11 % mu angle for grating (deg)
nug = 0.0 % nu angle for grating (deg)
taug = 0.0 % tau


Use: 0.1326