Skip to content. | Skip to navigation

Personal tools
You are here: Home Users Guillermo Blanc LDSS-3 LDSS-3 User Manual

LDSS-3 User Manual

This is a revised version of the LDSS-3 user manual made in December 2016 by K. Boutsia, G. Blanc, and Y. Beletsky. This is a live document meant to be kept up to date.

1 - Instrument Description


The ‘Low Dispersion Survey Spectrograph' (LDSS-3) is a high efficiency spectrograph and optical imager installed in the Nasmyth West port of the Magellan Clay telescope. Its high throughput and excellent sky subtraction lead to spectacular performance at very faint limits while its relatively large field of view and multi-object capability provide a large multiplex advantage for statistical projects. LDSS-3 is the upgraded version of the LDSS-2 spectrograph that was in operation on Magellan from 2001-2004. The upgrade includes a new collimator, camera, CCD device, filters, and grisms. LDSS-3 was designed to be very red sensitive - allowing efficient multi-object spectroscopy out to 1 micron. LDSS-3 excels at obtaining spectra of large numbers of faint objects.

The instrument operates as follows. The telescope is focused onto a multi-aperture mask held in an 8-position wheel. The light then passes through various apertures cut in the mask and enters the collimator, which converts the input f/11 beam into parallel light, before passing it through either a filter and/or a grism. The light is then focused by the camera onto an external detector with a final focal ratio of f/2.5. Removable Hartmann masks are provided in the filter and grism wheels to aid in focusing the instrument. Three grisms are provided and mounted in the grism wheel at any one time. These cover a range of spectral resolution of several hundred to several thousand. Up to 7 filters can be mounted in the filter wheel at once. By using clear positions in the aperture and grism wheels, LDSS-3 can be used to give direct images in the chosen filter pass band over a wide field of view. It thus doubles as a wide-field imager.

LDSS3 Photo
Figure 1: LDSS-3 installed on the Nasmyth West port of Clay (Magellan-II)

2 - Basic Parameters

Focal Ratio: f/11 

Wavelength Coverage: 3600-10000 Å 

Field of View: 8.3 arcmin diameter field, trimmed to 6.4 arcmin in spatial direction (i.e. x-direction on detector). 

Plate Scale at Focal Plane: 2.88 arcsec/mm

Pixel Scale at Detector: 0.189 arcsec/pixel (15 μm pixels at detector plate scale of 12.6 arcsec/mm)

Aperture Wheel: 8 positions (7 masks + 1 clear)

Filter Wheel: 7 positions (SDSS griz + OG590 + 1 clear + user filter on request)

Grism Wheel: 3 grism positions + 1 clear + 2 Hartmann masks

Slits: A series of standard long-slit masks exist ranging from 0.75” to 2.15”

Calibration: External lamps are used for calibration including He,Ar,Ne for wavelength calibration and Quartz lamps for dome flats.

CCD Detector: S3-171, 2048 x 4096 (15 μm pixels).

Minimum Exposure Time: The GUI allows 1 sec exposures but to avoid shutter response effects a minimum of 7sec is recommended.

Maximum Exposure Time: The maximum exposure time that can be used with LDSS is 30min (if longer the files cannot be saved, thus the exposure is lost). Because of cosmic rays accumulation it is advisable to limit exposure times to 15-20min. 

Operating Modes:

  • Imaging Mode: SDSS griz filters, a broad VR filter and a series of bandpass filters available on request. An updated list of the filters can be found in Table 1.
  • Long-slit and Multi-object Spectroscopy: low-dispersion grism spectroscopy: R=860 – 1900 (0.75” slit)
  • Nod & Shuffle Mode: this is a technique that allows excellent sky subtraction by chopping the spectra between two ends of the slits and nodding the telescope in the opposing direction.

3 - Available Filters


Up to 7 filters can be mounted in the filter wheel at once. Four SDSS broad-band filters (griz) and a Harris B-band filter are supplied for direct imaging. An OG590 blocking filter is also available for spectroscopic observations. In addition, a number of filters have been purchased by observers including a broad VR filter and a series of bandpass filters. These user filters are kept at the Observatory and are available for general use. To assure that these filters are available for your run, please specify them in the Instrument Setup Request Form at least 3 weeks prior to your run. Observers may also supply their own filters. However, the use of such filters must be coordinated with Guillermo Blanc (LDSS-3 PI: gblancm<at>, ldss3<at> at least 2 months prior to the observing run. The standard filters have diameters of 100mm and thickness less than about 10 mm. A full list of the available filters can be found in the following table:


Filter Wavelength Range
Data (ASCII)
Plot (PDF)
data plot
SDSS g'   data plot
SDSS r'   data plot
SDSS i'   data plot
SDSS z'   data plot
OG-590    data plot
GG-495    data plot
KG-650    data plot
  data plot
KG-850   data plot
KG-915   data plot
KG-950   data plot
W-630-950    data plot
BPF-380-590   data plot
BPF-480-780   data plot
B-old    data plot
VR   data plot
I   data plot

Table 1: Available LDSS-3 Filters 


4 - Imaging


By using clear positions in the apertureand grism wheels, LDSS3 can be used to give direct images in the chosen filter passband over a wide field of view.


4.1 - Imaging Sensitivity


The most recent zero-point measurements (i.e. the magnitude of a source that produces 1 e-/s at airmass=1.0) for the four SDSS broad-band filters are given in the next table:

Filter ZP
1 e-/s @ airmass=1.0
SDSS g' 27.64
SDSS r' 27.80
SDSS i' 27.82
SDSS z' 27.86

Table 2: Broad-band imaging zero-points (measured in August 2017 - after new M3 installation)


4.2 - Ghosts (under revision)

Imaging observations of bright targets using LDSS3 will clearly show ghosts. The location of these ghosts relative to their parent objects shows that they arise from light which bounces of the detector, passes back through the camera (and hence gets re-collimated to a parallel beam), reflects off the filter and returns once again through the camera to be imaged on the opposite side of the field. The intensity of the ghosts is principally a function of the reflectivity of the CCD surface and the steepness of the filter cut-on/off (over the wavelength range where the filter drops from transmitting to reflecting it acts as a ~50% mirror). 

We have measured the ghost intensity and behaviour using direct images of masks, which provide a recognisable pattern of bright slits, and find the following: 

The ghost  amplitude is less than 1% at almost all field locations for all four SDSS filters (griz). The ghosts are strongest for parent objects near the center of the field and decline approximately linearly with radius. This is likely due principally to the wavelength shift of the filter bandpass induced by the steepening incidence angle of the reflected light from far field locations, which reduces the effectiveness of the filter as a reflective surface.

The ghosts produced by the above process are also not perfectly in focus, because the collimated beam is not perfect (the collimator leaves in a residual spherical aberration and the position of the pupil is wavelength dependent). This residual aberration results in a slight smearing of the ghost image relative to the parent image that is wavelength and field location dependent. The FWHM of the ghost for a parent point source can be described as a linear function of field radius increasing towards large radii from the apparent center of reflection. 


5 - Spectroscopy 

There are three modes to use LDSS-3 for spectroscopy (described in detail below).   
  • long-slit mode
  • multi-slit mode 
  • nod & shuffle mode 
All these modes can be used with different combinations of grisms and slit-widths that translate in different spectral resolution and wavelength coverage. 

5.1 - Grisms 


LDSS-3 currently has three grisms available for general use. The new VPH grisms were designed by Mike Gladders and assembled by Ivan Baldry, Karl Glazebrook and the JHU Instrument Design Group. A feature of these grisms is that the wavelength coverage and blaze vary substantially over the detector. All relevant properties for available grisms can be found in the table below. 


Grism Lines 
(0.75" slit)
wavelength (A)
Nominal wavelength
range (A)
Linear dispersion at
nom.wav. range (A/pixel)
Peak total system
efficiency (%)*
VPH-All 400 860 7100 4250 - 10000 1.890 32
VPH-Blue 1090 1900 5000 3800 - 6200  0.682 32
VPH-Red** 660 1810 8000 6000 - 10000 1.175 37

* measured July 2016, 

** For the VPH-Red filter, the OG590 filter should be used to eliminate second order contamination that is substantial above 7000A for blue sources.

Table 3: LDSS-3 grism properties 


The throughput and wavelength coverage for the three available grisms can be found in the figures below (measured in July 2016):


Throughput LDSS

Figure 2:
Total system throughput with the three LDSS-3 grisms. The curve for the VHP-Red grism includes the transmission of the OG590 filter. 

WARNING!! Second order contamination: For the VPH-Red grism the OG590 filter should always be used to eliminate second order contamination, which can be substantial redward of about 7000A for objects with significant blue flux. The plots below show the difference in the VPH-Red spectrum of the spectrophotometric standard star EG274 (a white dwarf) with and without the OG590 filter. Not using the blocking filter can translate in up to an 80% level contamination at the red end of the spectrum for blue sources such as this one. Therefore it is highly recommended that the filter is always used when observing with the VPH-Red grism.

EG274 OG590  EG274 OG590 ratio
Figure 3: Left: VPH-Red spectrum of the white dwarf EG274 with the OG590 filter (red) and without it (white). Right: Ratio between the two. 

5.2 - Standard Long-slits


A series on standard long-slit masks of different widths are available for LDSS-3. The appropriate masks should be requested on the Instrument Set-up form, which should be submitted three weeks before an observing run. The long-slits are centered near the center of the field and at offset positions 2' on either side of the center. As mentioned above the wavelength coverage and blaze of the VPH grisms vary substantially over the detector, therefore the offset slits can beused to obtain different grism efficiency curves (skewed towards the blue or the red). The direction of the bandpass shift is noted in each mask (+2’ red slit, - 2’ blue slit). The two following figures show images of the three types of masks and their relative system throughput.


Blue Center Red 1.25" Slits

Figure 4: Images of the three 1.25" x 4' long slits masks: Blue (offset +2'), Center, and Red (offset -2').


Relative ThroughPut LDSS

Figure 5: Relative system throughput for the VPH-All grism using the Blue (offset +2'), Center, and Red (offset -2') long-slits. Curves are normalized to the peak system efficiency with the Center slit.


The following table lists all the available long-slit masks for LDSS-3. These include 4' long slits of different widths (as the ones shown in Figure 4) as well as masks with several 16" long slits of various widths (see Figure 6). All these masks are available in the Blue, Center, and Red configurations (i.e. cut with a +-2' offset). Users can also create specific long slits using the software for designing multi-aperture masks (see below). 

Mask Name Description
0.75"x4' center Width=0.75", Length=4', Position=Near the center of the field
0.75"x4' red Width=0.75", Length=4', Position=2' (red coverage)
0.75"x4' blue Width=0.75", Length=4', Position=2' (blue coverage)
1.0"x4' center Width=1", length=4', Position=Near the center of the field 
1.0"x4' red Width=1", length=4', Position=2' (red coverage) 
1.0"x4' blue Width=1", length=4', Position=2' (blue coverage) 
1.25"x4' center Width=1.25", length=4', Position=Near the center of the field
1.25"x4' red Width=1.25", length=4', Position=2' (red coverage) 
1.25"x4' blue Width=1.25", length=4', Position=2' (blue coverage)
1.5"x4' center  Width=1.5", length=4', Position=Near the center of the field 
1.5"x4' red Width=1.5", length=4', Position=2' (red coverage) 
1.5"x4' blue Width=1.5", length=4', Position=2' (blue coverage)
varwidthx16" center Variable Widths=2.15", 1.4", 0.85", 0.5", 0.65", 1.10", 1.75", length=16", Position=center of the field 
varwidthx16" blue Variable Widths=2.15", 1.4", 0.85", 0.5", 0.65", 1.10", 1.75", length=16", Position=2' (blue coverage)  
varwidthx16" red Variable Widths=2.15", 1.4", 0.85", 0.5", 0.65", 1.10", 1.75", length=16", Position=2' (red coverage) 
varwidthx16" blue + red Variable Widths=1.10", 0.85",0.65",0.5", length=16", Position=2' either side (allows red and blue coverage)
Table 4: Name and description of available LDSS-3 standard long-slit masks.
Variable Slit Width Masks

 Figure 6: Variable width masks containing several 16" long slits of various widths. These mask are also available in the Blue, Center, and Red configurations.

5.3 - Multi-aperture Masks


LDSS-3 multi-aperture masks consist of a number of slits cut with a laser machine. Observers must prepare the files necessary to generate their aperture masks and submit mask files at least 6 weeks prior to their observing run. The masks are then fabricated at LCO. Instructions on how to prepare these multi-aperture mask files are available here. The old LDSS-3 mask design instructions page (now deprecated and not kept up to date) can be found here and might provide useful information.

WARNING!! If the masks are not submitted at least 6 weeks before the first night of the observing run their timely fabrication cannot be guaranteed.  

Observers should bring to Chile printouts of the postscript files created by the mask generating software and finder charts of each field with the alignment stars labeled. Necessary masks for the run must be specified on the Instrument Setup Request Form at least 3 weeks in advance of your run. The day crew will mount them in holders and load them in the aperture wheel before you arrive at the telescope.

Users must submit their LDSS3 mask files using the LCO on-line LDSS3 Slit Mask Manager. For LDSS3 masks it is necessary to submit the final .nc file with up to 7 sub-masks included. Please do not use any form of compression on the uploaded files. The status of the masks can be monitored using the mask LDSS3 Slit Mask Database. Observers will be charged a fee of $35 US for each aperture mask cut.


5.4 - Nod & Shuffle


Nod & shuffle spectroscopy is a technique that uses very short slits (typically a few arcseconds) and chopping the spectra between opposite ends of the slits, while at the same time shuffling the charge back and forth on the CCD and nodding the telescope in the opposite direction. The advantage of nod & shuffle spectroscopy is that it allows excellent sky subtraction for long exposures since the object and sky spectrum are sampled by the same pixels on the detector in a quasi-simultaneous manner. However, the technique is considerably more complicated than standard multi-slit observations and should only be attempted by experienced observers. For more on the technique see Glazebrook & Bland-Hawthorn 2001, PASP, 113, 197. 

Nod & shuffle spectroscopy is possible with LDSS-3 in two modes: 

  • 'Microshuffle' mode: nods of a few arcseconds and shuffles of up to about 20 pixels. This mode has been successfully tested on LDSS-3 and appears to work very well. 
  • 'Macroshuffle' mode: shuffles of a significant fraction of the CCD width. While this mode can in principle be very powerful, early tests suggest it is not an optimal mode for LDSS-3. Therefore, this mode is not recommended and it is not supported either. 

Nodding with LDSS-3 is done by moving the guide box on the guide CCD and not by moving the guider probe. This effectively limits nods to some fraction of the width of the guide probe field of view or approximately 30". 

Note: This mode should be requested by the observer in the Instrument Setup mode well in advance (3 weeks) since a dedicated instrument setup is required that should be performed during daytime. 


6 - CCD detector, Field of View, and Shutter:


The LDSS-3C CCD is a 2kx4k chip provided by Fermilab and it is the same type of detector used in the DECAM project. The final science grade chip used in the instrument is #S3-171. 

The device uses a two-amplifier read mode. Each amplifier produces its own fits file, identified by ccdxxxxc1.fits and ccdxxxxc2.fits. The 'lstitch' iraf script, in the ldss iraf package, reconstructs the two frames into a single image. The stitched image is named ccdxxxx.fits. CCD cosmetics are acceptable, with some bad columns on the first amplifier (C1). In December 2016 new bad pixels masks have been created for both amplifiers (you can download them here C1 and C2). 


6.1 - CCD parameters:  

  • Dark current: 25 e-/pix/hr (@173K)
  • Full well (e-): ~205,000 e- (10% non-linear) 
  • Non-linearity (1%): > 175,000 e-
  • Saturation: 65,536 ADU
  • Quantum Efficiency: Peak QE=90% @ 900 nm (QE>50% at 400 and 1000 nm, see plot below)
  • QE stability: 0.6% / K @ 1000 nm 
  • Readout Modes: Slow, Fast, Turbo 
  • Gain (updated in July 2016):
Speed amplifier 1 (C1)
amplifier 2 (C2)
Slow         0.15          0.13
Fast*         1.67         1.45
Turbo         3.04        2.63
* Updated in June 2017 

  • Readout Noise (updated in July 2016): 
Speed amplifier 1 (C1)
amplifier 2 (C2)
Slow         4.4         5.3
Fast         7.0         7.2
Turbo*         ~10        ~10 
*Not measured lately - not to be used for science 
  • Readout Time (updated in July 2016): 
Speed Bin 1x1 Bin 2x2
Slow  166 sec    50 sec
Fast   30 sec    13 sec
Turbo   25 sec    11 sec 


The CCD has three readout speeds. The "fast" speed readout is appropriate for most spectroscopy and certainly for imaging. The "slow" speed should achieve the lowest noise at the cost of a longer read time, but experience has shown that there is no real positive benefit using this mode. Thus it is recommended to use "fast" speed also for faint sources. The "turbo" readout is intended for setup frames and tests, thus it should not be used for science or calibrations. There is a low level readout noise pattern (amplitude <1.5 e-) in all readout modes. The pattern phase is time variable and the amplitude is significantly smaller than the readout noise, so averaging a few frames should remove it completely. 

WARNING!! In order to avoid damage to the detector, please respect the following guidelines: 

  1. Never observe when the sun is above the horizon (i.e. no sky flats with the sun above the horizon).
  2. Avoid morning flats if possible. At any rate, choose short exposure time (~1sec) for the first flats and/or when changing filters to determine safe exposure levels.
  3. Never point at the moon.
  4. Switch off Vsub in the LDSS-3 GUI whenever the camera is not in use (i.e. in the morning after the run is over). 

The rotation of the CCD is fixed and will not be adjusted for observers. This maintains consistency between the data reduction software and the orientation of the spectra. This also ensures that the mask alignment software will work. Users should expect that the spectra will not align perfectly with the cardinal directions of the CCD. This is true regardless of the rotation of the CCD because of distortions in the instrument. 


6.2 - Linearity:

 The linearity of the CCD has not been analyzed in detail. However, preliminary tests suggest it is linear to better than 1% up to 40,000 ADUs. Figure 7 below shows number of counts vs. exposure time.


LDSS-3 Linearity

Figure 7: Flat-field median count level as a function of exposure time showing better than 1% linearity up to 40,000 ADU.


6.3 - Quantum Efficiency:

As mentioned above the QE of the detector is >50% at 1000 nm. This is a considerable improvement with respect to the previous LDSS-3 detector (replaced in 2014) as shown in Figure 8 below. 

Figure 8: S3-171 Detector Quantum Efficiency as a function of wavelength, compared to previous LDSS-3 detector.

6.4 - Cross-talk  (under revision) :

Cross-talk between amplifiers has not been measured for the S3-171 detector. The DECAM instrument which uses the same detectors and similar readout electronics shows crosstalk at the 0.05% - 0.1% level, linear up to 60k ADU (260 ke-) where it becomes non-linear.  No 1-amplifier readout mode is currently offered, nor do we envision offering such mode in the future. The 2-amplifier cross-talk is sufficiently small and the 2-amplifier readout is the only supported CCD readout mode for this instrument. 


6.5 - Field of View: 

Since the new CCD is only half the size of the old CCD, the 8.3' diameter FoV of LDSS-3 is not completely covered, but rather trimmed to 6.4' in the horizontal direction. About 10% of the light at the edge of the old FoV is lost. See the footprint plot on Figure 9 below. 


Figure 9: LDSS-3 Field of view (black) and region covered by the 2k x 4k S3-171 detector (red).


6.6 - Shutter:


LDSS-3 uses an iris-type shutter that sits in the collimated beam. Since the shutter is in the collimated beam, there are no significant variations in exposure time across the CCD. In December 2016 data was obtained to perform tests on LDSS-3 shutter timing. In Figure 10 we plot the ratio of the actual exposure time to the requested exposure time. This shows that for exposure times equal to or longer than 7 sec, the shutter timing is good to 1% or better. 


Figure 10: Ratio of actual to requested exposure time as a function of requested exposure time. Shutter timing is good to 1% for >7 second exposures.

7 - Calibration Unit: 


The calibration lamps are run from the DCU/Clay GUI (Figure 11), that can be found among the applications, if it is not already running on the observer's desktop. 


Figure 11: DCU/Clay GUI which controls the Calibration Unit.

The row of buttons represents individual arc or quartz lamps mounted near the flat field screen. Below the row of lamps there are the controls for the flat field screen. Generally observers should only be concerned with the "deploy" and "retract' commands from the drop-down menu "FF-Screen". These command move the screen in and out of the light path. When you are ready to take some calibration exposures, select "deploy" item from the "FF-Screen" menu to move the screen into the light path. If this is done during night-time observations, make sure to warn the telescope operator that the screen is being deployed in order to stop guiding. While the screen is moving, the word "moving" appears in the GUI box. After the state on the GUI box becomes "deployed", it is possible to turn on the necessary arc lamps by clicking on the lamp buttons. Once the lamp buttons turn read, which means they are on, you may take exposures. 

For the spectroscopic configurations of LDSS-3, the helium (He), argon (Ar) and neon (Ne) lamps are useful. The integration times needed for arcs will depend on the grism you are using and the width of your slits. A line atlas of comparison spectra for LDSS-3 can be found here.     

Quartz lamps for dome flats are also controlled from the DCU/Clay GUI. There are two lamps available, Quartz low (Ql) and Quartz high (Qh). For lower intensities, better suited to broadband imaging, users are advised to use the independent quartz lamp power supply installed in the dome beneath the NASW platform. Setting of 1-3 volts should be sufficient for most filters and the power supply should never be set higher than 5 volts. For using this lamp, first-time users should coordinate with the Instrument Specialist. In the table below we provide some reference numbers of imaging flat counts obtained in the g'r'i'z' filters and suggested lamp intensities (measured in July 2016):


Filter Counts (Fast bin1x1) Exposure time (sec)  Voltage (V)
  g'         12000           30      2.0
  r'         14000           10      1.6
  i'         15000           10      1.2 
  z'         20000           10      1.0

 Note: For regularly updated values, please check the "Calib" section 


8 - Observing with LDSS-3: 

8.1 - Preparing for the observations 

8.1.1 - Observing catalogs

More details on how to create observing catalogs can be found here

8.1.2 - Producing an LDSS-3 mask  

More details on how to create LDSS-3 masks can be found here
Masks should be submitted at least 6 weeks prior to the observing run following instructions found here

8.2 - At the telescope 

8.2.1 - LDSS-3 Control Software (GUI) 


The CCD camera and motion of the various wheels (i.e. filter, grism, apertures) are controlled by the observer using the CCD Camera GUI. The CCD Camera GUI will usually be setup by the day crew and already be running when the observer arrives at the telescope. If not, or if a restart is needed, the observer can easily start the GUI themselves, launching the Configuration window by typing ldss3 in a terminal, on the control computer. This command is in the observer's path and no parameters are necessary. This GUI allows the observer to define the system setup without the need of editing any of the setup-files. In most cases, the only thing the observer need to change is the "Observer" parameter (see Figure 12). Click "OK". This will launch the CCD Camera GUI (Figure 13). 



Figure 12: LDSS3 configuration GUI


Figure 13: The CCD camera GUI


Note that the data files will by default be written into the user's home directory. The data path can be changed in the LDSS-3 GUI using "DataPath" which is found in the "Options" pull-down menu. The user should make sure the data path has been properly set before taking any exposures. 

ExpTime: is the exposure time in seconds for the next observation. Once the exposure time has been set, it cannot be changed (the ExpTime button turns grey and it is not selectable). The only way to stop an exposure is by clicking the "Stop" button before readout (see details below). 

Loops: allows obtaining a series of exposures with identical exposure times. This can be a convenient way to take bias frames, flat fields or make synoptic observations. 

ExpType: specifies the type of exposure (e.g. object, bias, flat, etc). This information is written to the image header and can be useful for data reduction purposes. 

Subrasters: As of May 2017 the subraster mode presents readout problems and should not be used until further notice!

The software also provides for a very convenient way to set up subrasters using the menu to the right of "ExpType". When "Full" is selected, the entire CCD is read out. If "Subraster" is selected, a second window is launched (see Figure 14) that allows the appropriate pixel values to be entered. Note: the pixel values for the subraster must be the values associated with the Quick Look Tool and not from a stitched image. 



Figure 14: The Subraster GUI


As many as 8 subarsters can be selected; the user fills in a table of X0 and Y0 coordinates and the size of each sub-array. The cursor on the Quick Look display can be used to select the center of the subraster: type "a" to transfer the position and set the subraster size. Taking a picture with the subraster feature will save time, since the CCD is clocked without reading the data between the readout of subrasters (at whatever binning). However, as the number of subrasters increases, this advantage will diminish. Data for subrasters can be stored separately by choosing SaveMode=minimal; or embedded in a full frame with zeros filling the non-subrastered areas, by choosing SaveMode=full, in the Subraster dialog box. The SaveMode=full is the default. Be careful if you are reading the subraster values from a file, that you input the correct file name. If the file name is misspelled, the code will accept it but will not really apply the subraster. 

Aperture, Filter and Grism are used to move the wheels in LDSS-3. To move one of the wheels, press the left button of the mouse on the box with the current position name. This will pull up a menu with the various wheel positions. When a selection is made, the box will remain yellow until the move is complete. 

The Edit buttons are password-protected and can only be used by the day crew during initial set-up. 

Focus is used to change the focus value of the instrument. When the option "Auto" is selected (green) the GUI automatically adjusts the focus with temperature, when the difference to the current focus is more than 4 units. 

Start begins a science exposure. The exposure can be paused (e.f. for clouds) and resumed by toggling with the pause button, but observers should remember that cosmic rays are accumulating while the exposure is suspended. The progress of the exposure can be seen in the graph-bar that fills to the right; as the CCDs read out the bar empties to the right. Test picture will typically be taken rebinned to save readout time. 

The Snap button automatically sets the 4x4 mode, takes and exposure and, unlike the Start command, does not update the frame number at the end of the exposure. 

Stop: At any time the user can stop the exposure and a readout will immediately begin. An image with the actual exposure time (the time the exposure has actually lasted and not what has been inserted in the ExpTime field) will be saved. A wipe of the chips follows, but since the user will not generally know when it is finished, it is strongly recommended that a "Snap" of minimal exposure time be taken to insure a clean chip, before any science exposure is begun. 

Abort: The 'Abort' button stops the exposure and does not readout the CCD, thus no image is saved and the frame number is not updated. 

File#: sets the file number. If the same file number is entered twice then the older image will be overwritten 

Speed: sets the readout speed. The CCD camera has three readout speeds. The 'fast' speed readout is the default, with a readout time of 30sec. The 'slow' speed achieves the lowest noise at the cost of 166sec readout time, but it presents a noise pattern higher that the other two modes. The 'turbo' readout is intended only for setup frame and tests with a 25sec readout time. 

All modes present a diagonal pattern on the bias. The rough amplitude of the pattern is: 

  • Slow: +/- 8 ADU (+/- 1.2 e-) 
  • Fast:+/- 0.8 ADU (+/- 1.3 e-)
  • Turbo: +/- 0.5 ADU (+/- 1.5 e-)

All patterns have a time dependent phase, so they will average out when averaging several exposures. In terms of data quality, there should be no difference in terms of the impact of readout patterns between the three readout modes. Nevertheless we recommend using 'fast' mode even for faint sources, because the 'slow' mode does not give a strong advantage and the readout time is considerably longer. 

Disk: shows how much disk space is available in the current directory. 


8.2.2 - Afternoon and twilight calibrations - Mask verification 

Once your masks have been mounted in the aperture wheel, you are ready to take some calibration and preparation images. The first thing to do is to take images of your slit masks in the afternoon. Typically a 10sec integration will be sufficient - there is no need to turn on any lights, since there should be plenty of light leaking in from the dome. These images should be used to verify that the masks have been mounted properly in the holders. When displayed with iraf, the slits should run perpendicular to the columns, unless you have used tilted slits, and along the columns if they were created for nod & shuffle mode. Compare the image with the postscript output from the mask generating code. 

Observers are also encourages to become familiar with the mask aligning software (see cookbooks) during their first afternoon at the telescope. Several iraf scripts have been written to assist with the alignment of multi-aperture masks. To run these, start up iraf on the control computer (by typing goiraf in a terminal) and type "ldss3". This will load all appropriated scripts, dedicated to LDSS-3, that are described in detail in the spectroscopy cookbooks. - Biases 

Bias frames should be obtained in the afternoon. To take a sequence of bias images, in the LDSS-3 GUI, set "Loops" to the number of exposures you wish to take (5 to 10 frames should be enough), set "ExpType" to "Bias" and Binning X and Binnning Y values to whichever values you plan to use at night. At this point you may click the "Start" button and execute the loop sequence. Note: It is necessary to take biases for all readout/gain setting you plan to use during the night. - Flats and Arcs 

Ideally, you would want to take some dome and arc exposures during the afternoon. However, there is severe light leakage in the Clay telescope dome, so it is not possible to properly illuminate the flat field screen during the day to take filter flats. Thus, all dome flat field images should be taken at night or significantly into twilight. For imaging, twilight flats are recommended. 

For spectroscopy, it is best to take the arc and dome exposures you need for each mask while you are observing the field (i.e. immediately before or after an observation). However, it is useful for observers to take some test dome and arc exposures in the afternoon to determine appropriate exposure times. Below we provide some reference numbers for exposure time and suggested quartz lamps in order to obtain a flux level of ~30000 ADU with various longslit widths for a standard setup of: speed:fast / gain:low / binning:1x1

Grism(lamp) / Slits  0.75 - center 1.0 - center 1.25 - center 1.5 - center
 VPH - All (Qh)    12 sec    10 sec     7 sec     6 sec
 VPH - Red (Qh)   12 sec     9 sec     6 sec     5 sec
 VPH - Blue (Qh+Ql)   90 sec   70 sec   60 sec   50 sec

 Note: For regularly updated values, please check the "Calib" section - Focusing the spectrograph 

LDSS-3 can be focused using an internal stage, which moves the location of the camera along the optical axis. Filters sit in the collimated beam and the system is almost achromatic. The focus offset of each filter has been measured and it is very small. Typically, blue filters focus at slightly smaller focus encoder values. Under normal operations the focus compensation with filter changes is automatic and the user should not need to change the focus manually when changing filters. This will be true as long as you are using one of the standard filters. 

Observers are encouraged to verify that the focus does in fact change when the filter is changed the first afternoon of their observing run. Note that some user filters do not have filter focus values measured and thus observers may need to change the focus manually for these filters. The focus of LDSS-3 is temperature dependent. As mentioned above, temperature adjustments are built into the automated system (see "focus" section in 8.2.1). Currently, the Instrument Specialists measure a focus value during setup, for a given temperature, and then the system offsets this focus value to the appropriate number for night time operating temperatures. The focus value is changed using the "Options -- StdFocus" menu to enter the focus of the Sloan-R filter and the temperature when it was measured. The GUI presets the temperature with the current instrument temperature (not the last used focus temperature), thus if the focus measurement was obtained in a different moment the appropriate temperature value should be edited as well. The focus system moves to the nearest commanded encoder position with a coarseness of 5 units (i.e. 510, 515, 520, etc.). Note that a one-pixel image separation (i.e. 0.189 arctic) in a Hartmann mask test of LDSS-3 corresponds to 182.4 encoder units, thus a coarseness of 5 units is more than adequate. 


 8.2.3 - Beginning of the night 

When you arrive at the telescope after dinner, the night assistant will have already opened the done and the telescope mirror covers. Next, the telescope should be focused. This process can take a little time (10 to 15 minutes) and generally does not require the astronomer to do anything. 

If this is the first night that LDSS-3 has been on the telescope, the observer should verify the orientation of the CCD on the instrument has not changed. Typically for the West Nasmyth port of the Clay telescope, a rotator offset mode and angle of (27.5 EQU) will align the standard long slits, and slit view mode, along the N-S directions. Targets in the South will have N up and E to the right of the array. Targets in the North will have N down and E to the left of the array. 

To align the slits along the parallactic angle, use (-62.5 HRZ). If you are using multisite masks, the latest version of the mask generation software now contains a program named "obscat" that will generate the appropriate catalog file entry with input of the .SMF file. The fiducial values for the rotation are known to shift by several degrees. The latest values are usually documented in a figure that should be posted in the control room. Multi-slit users should note any offset to the angle of the first mask and align and update their observing catalog appropriately. 

8.3 - Using LDSS-3 


8.3.1 - Longslit spectroscopy Cookbook 

8.3.2 - Multi-slit spectroscopy Cookbook 

8.3.3 - Imaging Cookbook

8.3.4 - Nod & shuffle Cookbook 


8.4 Reducing LDSS-3 data  

LDSS-3 data can be reduced in a fairly straightforward fashion using the reduction package COSMOS 























Document Actions