Chapter 2. Planning the Observing Run

Sections: Diffraction Gratings | Slit Width and Spectral Resolution | Spatial Resolution | Filters Available | Slit Lengths and Dekkers | Efficiencies | Detector Characteristics
Previous: The RGO Spectrograph | Next: Description of the Spectrograph | CONTENTS

This chapter contains all the system parameters needed to plan and apply for an observing run using the RGO Spectrograph.

2.1 Diffraction Gratings

Table 2.1 lists the current set of gratings. The name of each grating indicates the number of grooves per mm, and the optical band of the blaze for first order. Column 2 gives the old name used in early log books. Dispersions listed are in Å/mm, and Å/pixel for the EEV CCD (13.5 µm/pix). For more information on these gratings, see the Gratings Manual.

Table 2.1: Available Gratings

Name  No.a Order  Blaze  Dispersion 
(Å/mm) 		 Dispersion 
(Å/pix) 
82cm  25cm 
(Cam) 		 25cm 
(Col)b		 82cm  25cm 
(Cam) 		 25cm 
(Col) 
1200B  6  I  4300  10  26  33  0.14  0.35  0.45 
1200V  1  I  5000  10  25 
33 
 0.14  0.34  0.45 
1200R  2  I  7500  9  20  33  0.12  0.27  0.45 
II  3750  5  11  17  0.07  0.15  0.23 
1200I    I  10000  9  15  33  0.12  0.20  0.45 
II  5000  3  8  17  0.04  0.11  0.23 
III  3333  2      0.03     
600U  5  I  3500  20  58  64  0.27  0.78  0.86 
600V  8  I  5000  20  57  66  0.27  0.77  0.89 
600R  9  I  7500  19  53  66  0.26  0.72  0.89 
316Rc   I  7500  38  108  125  0.51  1.46  1.69 
300B    I  4500  40  114  132  0.54  1.54  1.78 
270R  3  I  7600  45  130  142  0.61  1.76  1.92 
a Original number of the grating, useful for interpreting old logbooks
b Note that blaze to collimator is the default configuration; see below.
c The 316R grating is frequently in use at the UK Schmidt for the 6dF galaxy survey. Contact Paul Cass (cpc@aaocbn.aao.gov.au) for unscheduled use of this grating. 

Blaze

Gratings can be oriented with the blaze direction towards the collimator or towards the camera. For the 82 cm camera there is relatively little difference in the nominal dispersion between working blaze-to-camera and blaze-to-collimator, except for the higher orders of the 1200 line gratings. For the 25 cm camera, significantly lower dispersions occur when working blaze-to-collimator. For most projects it is advised that the grating be operated at blaze-to-collimator on the 25 cm camera because of the significantly wider entrance slit that can be used for the same projected slit width (see Section 2.2). Note, however, that some vignetting will occur at small grating angles (Section 2.6). For the larger format CCDs, the edges of the spectrum are distorted at blaze-to-collimator - this effect is not seen blaze-to-camera.

Wavelength Selection

The central wavelength is selected by changing the grating angle, and by the use of order sorting filters. A utility program called RGOANG is available on the AAO web site at http://www.aao.gov.au/cgi-bin/rgoang_inputs.pl, on the VAX, and on the SUN workstations under the AAT utilities (see Chapter 4). This routine supplies the required grating angle as well as the spectral dispersion and slit projection factor (Section 2.2). The wavelength specified refers to the centre of the field. RGOANG uses the general grating equation: where m is the order of diffraction, a is the number of grooves per mm, theta is  the grating angle (in degrees) from the readout and C1 and C2 are offset angles (C1 ~ 65° for the 82 cm camera and 5° for the 25 cm camera while C2 ~ 50° for both the 25 cm and 82 cm cameras).

Wavelength Dispersion

The wavelength dispersion for all gratings is given in Table 2.1. The dispersion is given in Å/mm at the blaze wavelength for the Littrow mounting. The blaze wavelength is reduced by about 8% by the 25 cm camera and about 1% by the 82 cm camera. The dispersions are calculated for the 25 cm camera with the grating mounted blaze-to-camera and blaze-to-collimator. The equivalent dispersions in Å/pixel are shown for the 13.5 µm pixels of the EEV CCD. (The MITLL CCDs have 15 µm pixels, and the Tektronix has 24 µm pixels.) Note that for low-resolution observations of faint objects with the EEV it may be beneficial to bin x 2 in the wavelength direction (i.e. 27 µm pixels). The variation of plate dispersion with wavelength may be calculated from the general dispersion equation:
d(lambda)/dx = a cos(theta - C1) / mf  (Eq. 2.2)
where x is the distance along the detector, f is the focal length of the camera (25 or 82 cm) and the remainder of the symbols are as for Equation 2.1.

Wavelength Coverage

The wavelength coverage can be estimated by multiplying the dispersion for the grating by the physical size of the detector. The EEV CCD is 55.3 mm wide in the wavelength direction, but for blaze-to-collimator only 45 mm is useful. Likewise, the MITLL CCDs are 61.4 mm wide, but only 45 mm is useful blaze-to-collimator. The Tektronix CCD is 24.6 mm wide.

For example, any 1200 line/mm grating in first order, blaze to collimator, on the 25 cm camera with the  EEV CCD will have ~1400 Å coverage; 600 line gratings with the same setup will have ~3000 Å coverage; and the 300B grating will cover ~5700 Å.

Grating Angle Limitations

Gratings can not be put at angles less than -10°, or greater than 110°, and vignetting becomes severe near these limits (Section 2.6). The setups affected by these limitations are high order observations with the 25 cm camera. Second order observations with blaze to collimator can only be made blueward of 5000 Å and with blaze to camera blueward of 6500 Å.

2.2 Slit Width and Spectral Resolution

The spectral resolution obtained is dependent on wavelength dispersion, pixel size of the detector, instrument resolution, projected slit width, accuracy of focus and even size of seeing disk in some cases. While spectral resolution can be predicted, it should be checked during setup if critical. Spectral resolution is estimated from the gaussian FWHM of comparison arc spectral lines which are free of blending and saturation.

Instrumental Resolution

The minimum resolution (with a 10 µm slit) of the RGO is ~33 µm, corresponding to 2.5 EEV pixels, 2.2 MITLL pixels and 1.4 Tektronix pixels. In fact the resolution can be better, especially in the UV, and redward of 6000 Å the resolution element increases due to charge spreading, and is typically 40 µm at 8500 Å.

Slit width and Slit Projection Factor

The spectrograph slit width is variable between 10 µm and 2.5 mm. As 1 arcsec on the sky is equivalent to a slit width of 150 µm, the range of widths is 0.07 to 16.7 arcsec. In the RGO spectrograph, the projected slit is normally less than the actual slit width. The projection factor  (i.e. the ratio between the actual slit width and the width projected at the detector) is given by:
g(theta) = G cos(theta - C1) / cos(theta - 50) (Eq. 2.3)
where C1 ~ 65° G = 1.46 for the 82 cm camera, and C1 ~ 5° G = 4.8 for the 25 cm camera. These functions are plotted in Figure 2.1.
 
 

Figure 2.1 Projection factor vs grating angle for the 25cm and 82cm cameras.

Projection Factor

Sampling

For most applications, a minimum resolution of 2 pixels is required for proper sampling of the spectrum. Undersampling results in residuals of sky emission lines and atmospheric absorption lines, mismatch in summing spectra, and aliasing of the Fourier transform. Oversampling is recommended for velocity measurements and will increase accuracy of sky subtraction and removal of atmospheric bands.

Calculation of Spectral Resolution

The spectral resolution for a setup is approximated by:

R2 = RI2 + [(S x Fg) / (P x g(theta))]2 (Eq. 2.4)

where R is the resolution in pixels, RI is the instrumental resolution, S is the slit size in µm, P is the pixel size in µm, g(theta) is the slit projection factor (Equation 2.3), and Fg is the geometric factor. The geometric factor corrects for the fact that the slit is not a circular aperture, and has been determined experimentally to be 0.66 ± 0.01. Note that this is only an approximation, and the actual resolution should be checked on arcs during setup. Slit sizes < 1 arcsec should generally be avoided - assume an average seeing of 1.5" for Siding Spring.

2.3 Spatial Resolution

The spatial scale at the detector is 32.1 arcsec/mm for the 25 cm camera, and 9.75 arcsec/mm for the 82 cm camera. For the EEV CCD, this corresponds to 0.43 arcsec/pixel and 0.13 arcsec/pixel respectively.

2.4 Filters Available

A range of neutral density and colour filters is available both above and below the slit, and are listed in Table 2.2. Laboratory efficiencies for the colour filters are plotted in Appendix C.1 and nominal density values for the neutral density filters are given in Appendix C.2.

Table 2.2 Available Filters

Position Star 1  Star 2  Arc 1  Arc 2  Below Slit 
1  Clear  Clear  Clear  Clear  GG495 
2  0.34G  CuSO4  0.5Q  CuSO4  Clear 
3  0.7Q  BG24  1.3Q  GG495  CuSO4 
4  1.0Q  BG38  2.5Q  OG570   
5  1.3Q  UG11  3.0Q  0.9Q   
6  1.6Q  GG385  4.0Q  1.5Q   
7  1.8Q  GG495       
8  2.1Q  OG530       
9  2.6Q  OG570       
10  3.0Q  RG630       

Notes

The CuSO4 filter in the star wheel is a 5 mm thick liquid filter. The other CuSO4 filters are crystal filters. The UG and BG filters are 2 mm thick and the GG, OG and RG filters are 3 mm thick.

Order Sorting Filters

The use of order sorting filters is essential for higher order work, and for first order work redward of 8000 Å. Extremely blue objects may have significant second order light from 6500 Å. The CuSO4 filter is also useful for first order work around 3300 Å, where red scattered light can be significant. Order sorting is also necessary for calibration arcs. Though second order arc lines can be useful for 82 cm camera observations, especially at higher orders where line distribution is sparse, they cannot be used for 25 cm camera observations as the wavelength-dispersion relationship is not linear.

While the below-slit filters are preferable as they filter both the starlight and calibration lamps, they are not often used. There are only three positions available, one of which is clear. The use of a below-slit filter changes the focus setting significantly, so frequent changes of below-slit filter are not recommended. In addition, the below-slit filters vignette the ends of the slit (Section 2.5).

The above-slit star filters are used for starlight and for chimney calibration lamps, and the internal calibration lamps have a separate wheel of filters with a more limited selection. They do not affect the focus.

Neutral Density Filters

These filters are not often used for objects since the current CCDs have a large dynamic range and are robust to saturation. They are useful to extend exposure times of bright objects to increase timing accuracy, and are required for red observations of the Cu Ar arc. If ND filters are used for the photometric standard, note that the dependence of transmission on wavelength has not been measured redward of 8000 Å.

2.5 Slit Length and Dekkers

The maximum slit length is 38 mm, or 4.2 arcmin. In fact, the ends of the slit are severely vignetted, so the usable slit length is 3.4 arcmin. Further vignetting is obtained if a filter is used below  the slit, particularly the CuSO4 filter, which reduces the usable slit length to 2.6 arcmin. Figure 2.2 shows the actual slit length obtained with and without a below-slit filter. The slit length can be also limited by the choice of a range of CCD windows or by use of a dekker.

Because of the vignetting, observing two objects at once is only efficient if the objects are less than about 1 arcmin apart. In addition, since the field of the slit-view mode of the acquisition system is only 30 arcsec, acquisition of widely spaced objects is a problem. Thus separate observations are typically more efficient for such objects.

Dekkers are available for the selection of limited portions of the slit. Details of the present dekkers are shown in Appendix D. For dekkers with multiple holes or slots, the separation is specified between inside edges. These diagrams are slightly misleading in that they suggest that the width (i.e. spectral dimension) is much narrower than the length (spatial direction). Actually the width of all dekkers is larger than the widest slit width available and is often considerably greater than the length.

Dekkers still in common use are 50 (clear); 56 (narrow) to provide a synthetic star for instrument setup; and 73 (double) for spectropolarimetry. However all listed dekkers are still available.

Figure 2.2 Vignetting function along the RGO slit for observations made with and without a below-slit filter

Vignetting of the slit

2.6 Efficiencies

A S/N calculator is now available for most configurations. Sufficient information is presented below to enable you to calculate count rates more accurately, and for other setups. S/N per pixel is calculated using:

S/N = O / SQRT(O + S + DN2) (Eq. 2.5)

where O is the number of counts per pixel for the object, S the number of counts per pixel for the sky and DN2 is the square of the readout noise times the number of spatial pixels extracted times the number of exposures to be summed.

Atmospheric Transmission

The fraction of the flux density transmitted by the atmosphere as a function of wavelength is shown in Table 2.3.

Table 2.3 Atmospheric Transmission


Wavelength
(Å) 		 Fraction 
transmitted 
3500  0.54 
4000  0.66 
4500  0.79 
5000  0.83 
5500  0.85 
6000  0.86 
6500  0.90 

 

Optical Elements

Table 2.4 presents estimates for typical efficiencies of various components of the telescope, RGO spectrograph and associated systems.

Table 2.4 Efficiency of the RGO


    25 cm  82 cm 
(i)  Primary mirror  0.85  0.85 
(ii)  Secondary mirror  0.85  0.85 
(iii)  Diagonal flat  0.87  0.87 
(iv)  Collimator mirror  0.87  0.87
(v)  Two lenses in camera (4 glass surfaces)  0.9854 - 
(vi)  Mirrors in camera  0.902 0.90 
(vii)  Flat mirror  -  0.87 
(viii)  Field flattening lens (2 glass surfaces)  0.9852 - 
(ix)  Window (to seal dry gas
ahead of the detector 		 -  0.9872
  PRODUCT  0.40  0.42 

 

Gratings

The efficiencies of the gratings are given in the Gratings Manual.

Vignetting

The 25 cm camera (only) experiences vignetting for extreme grating angles. Table 2.6 shows the calculated vignetting.

Table 2.5 Vignetting


grating angle  transmission %  slit projectiona
-10  35  8.2 
0  64  7.2 
+10  74  6.2 
+20  82  5.2 
+30  > 92  4.6 

Slit Throughput

The amount of light transmitted by a slit is dependent upon both the slit width and the size of the atmospheric seeing disk. In Table 2.6 the dependence of throughput on slit width is shown for various seeing disk sizes. In these calculations we assume that the star is centred on the slit and that the seeing is Gaussian in shape with a FWHM (in arcsec) given by the label.
 
 

Table 2.6 Slit Throughput


Slit Width
arcsec 		 1 arcsec
seeing 		 2 arcsec
seeing 		 4 arcsec
seeing 
0.1  0.08  0.04  0.2 
0.2  0.15  0.08  0.04 
0.5  0.38  0.20  0.11 
1.0  0.66  0.38  0.20 
2.0  0.90  0.66  0.38 
5.0  0.95  0.92  0.75 

 

Detectors

Efficiencies for the current detectors are presented below.

2.7 Detector Characteristics

The CCD characteristics are described in detail in the relevant web pages for the EEV, MITLL2A, MITLL3 and Tektronix. A summary is presented here. Figure 2.3 shows the efficiency curves of the current detectors.

Figure 2.3. Efficiency curves of the current CCDs.

CCD efficiencies

EEV

The EEV has 2k × 4k 13.5 µm pixels. Usually the longside is aligned with the dispersion direction. However, readout speed is much faster if aligned in the spatial direction (and suitably windowed), so this option may be useful for time-resolved work. You have to give warning well in advance of your run if you want to orientate the CCD in this way.

The EEV has a blue-sensitive coating, and is the preferred CCD for observations blueward of 6000Å. It suffers from strong fringing in the red, which appears to subtract well using flatfields, with only a small residual. A report from the commissioning run of the RGO with the EEV CCD is available here (coming!). Table 2.7 gives the characteristics of the EEV.

Table 2.7 Characteristics of the EEV CCD

SPEED  INT
(us) 		 GAIN
(e-/adu) 		 READNOISE
(e-) 		 ALPHA
(x1e-6)		 SAT
(Ke-) 		 READ
RATE
(us/pix) 		 Read Time (s)
Full 2x2 5x5
NONASTRO  1+1 6.8 8.2 1.16 200 7.5 69 30 11
FAST 2.5+2.5 2.7 4.4 0.59 168 11.5 105 43 54
NORMAL 5+5 1.3 3.4 0.35 81 16.5 150 54 17
SLOW 20+20 0.32 2.7 0.11 21 46.5 416 123 31

MITLL2A

The MITLL2A has 2k × 4k 15 µm pixels. It is Lumogen-coated to enhance blue sensitivity. It has the broadest range of sensitivity though it is less sensitive than the EEV in the blue, and the MITLL3 in the red. The MITLL2A suffers from some pixel smearing, leading to somewhat lower resolutions than might be expected. Table 2.8 gives the characteristics of the MITLL2A.

Table 2.8 Characteristics of the MITLL2A CCD

SPEED  INT
(us) 		 GAIN
(e-/adu) 		 READNOISE
(e-) 		 ALPHA
(x1e-6)		 SAT
(Ke-) 		 READ
RATE
(us/pix) 		 READ TIME
(s) 
1x1 
 2x2  5x5 
NONASTRO  1+1  4.8  5.26 -1.48 125  6.5  60  24  9 
FAST  2.5+2.5  1.8  3.2 -0.16 115  11.5  105 42  15 
NORMAL  4.5+4 .5 0.9 2.4  0.02 59 19  170  59  19 
SLOW  12+12  0.37  1.8  0.04  24  34  300  93  24 
XTRASLOW  48+48  0.091 1.4  0.01  6  106  940  260  56 

MITLL3

The MITLL3 has 2k × 4k 15 µm pixels. It is not possible to bin this CCD. It has the highest red sensitivity. It provides higher resolution than the MITLL2A, but has a higher cosmic ray rate and no blue sensitivity. At the time of writing, the MITLL3 is undergoing repairs and is not available. Check the webpage for news of its current status.

Tektronix #2

The Tektronix #2 CCD has 1024 × 1024 24 µm pixels. It has significantly lower efficiency than the EEV, and in general would not be the CCD of choice. Intending users of the Tek should discuss this question with the instrument scientist.

This detector can be used at 170 K or 200 K. Use at 200 K results in a 10% to 20% increase in QE at expense of a much increased dark current (0.1 e-/pix/s compared with 0.55 e-/pix/2000 s at 170 K).

Table 2.9 Characteristics of the Tektronix

Readout
Speed 		 Readout
Noise
(e- rms) 		 Gain
(e-/ADU) 		 Linearity
(alpha)
(×106		 Saturation
Level
(ADU) 		 Readout
Time
(s) 
XTRASLOW  2.3  0.34  negl  65000  394 
SLOW  3.6  1.36  negl  65000  120 
NORMAL  4.8  2.74  -0.03  65000  75 
FAST  7.2  5.5  -0.07  65000  52 
NONASTRO  11  11  -0.14  35000  33 

Observers can design their own CCD window if desired. Popular windows currently available are TEK1K_RGO, 375 columns which include the full slit; TEK1K_BX1, 150 columns covering the central half of the slit; TEK1K_BX2, the same as the previous window with spatial binning × 2.

CCD Windows

The RGO slit covers only 20 - 30% of the CCD, so you should use a narrow CCD window. A range of CCD windows suitable for RGO observing are supplied in the default area. Table 2.10 gives the most popular unbinned windows for each CCD. The spectral direction is usually in the Y direction (for pixel sizes in Angstroms for your chosen setup use RGOANG http://www.aao.gov.au/cgi-bin/rgoang_inputs.pl) and the spatial direction along the slit is usually in the X direction (spatial scales given in Section 2.3).  Typical readout times are 1 - 2 minutes in SLOW mode. More information on CCD windows is given in Appendix A.

Table 2.10 Popular CCD windows for the RGO


Window X Size (Pixels) Y Size (Pixels) Overscan (Pixels)
EEV2_RGO 600 4096 1
EEV2_RGO_300 300 4096 20
EEV2_RGO_150 150 4096 20
MITLL_RGO 463 4096 30
MITLL_RGO_200 200 4096 30
MITLL_RGO_100 100 4096 30
TEK1K_RGO 349 1024 26
TEK1K_RGO_200 200 1024 26
TEK1K_RGO_100 100 1024 12


Sections: Diffraction Gratings | Slit Width and Spectral Resolution | Spatial Resolution | Filters Available | Slit Lengths and Dekkers | Efficiencies | Detector Characteristics
Previous: The RGO Spectrograph | Next: Description of the Spectrograph | CONTENTS
 


Ray Stathakis
Last update 21/3/2002