The Dark Energy Spectrometer (despec): a multi-Fiber Spectroscopic Upgrade of the Dark Energy Camera and Survey for the Blanco Telescope



Download 4.5 Mb.
Page9/10
Date conversion08.07.2018
Size4.5 Mb.
1   2   3   4   5   6   7   8   9   10

4.C DESpec Unit Spectrographs and Optical Fibers

DESpec spectrographs will separate the light into component wavelengths and focus it onto CCDs. The key elements driving the optical design of the spectrographs are the wavelength range, the required spectral resolution, and the diameter of the optical fibers carrying the light from the focal surface. To accommodate 4000 fibers, each of 10 spectrographs must accept ~400 fibers. There are several options for the physical location of the spectrographs: they could be mounted off-telescope or, if sufficiently small and light-weight, they could be mounted near the top of the telescope and arrayed around its upper ring to minimize fiber length and thereby increase the throughput of the instrument. The off-telescope option for the location of the spectrographs has the feature that the fiber positioner provides a new versatile facility for CTIO. The part of DESpec up to the DESpec spectrographs can be used as-is (perhaps with the upgrade of adding an ADC) to feed other spectrographs that the observatory or the user community may wish to install, enabling a generic wide-field spectroscopic capability in the southern hemisphere. Off-telescope spectrographs can be maintained and upgraded without disrupting telescope operations.

As we are still exploring the science and survey requirements at this time, we are considering two spectrograph designs. The first design is a single-arm spectrograph with a wavelength range 550<<950 nm. The second is a two-arm (dichroic) spectrograph in which the blue side has wavelength range 480<<780 nm and the red side covers 750<<1050 nm.

4.C.1 Single-Arm (VIRUS) Spectrographs



The DESpec unit single-arm spectrograph optical design is based on VIRUS, the instrument being built for the HETDEX survey (Hill et al. 2008) on the 11-meter Hobby-Eberly Telescope in Texas. HETDEX requires 200 of these spectrographs, so they must be inexpensive and simple enough to produce in bulk. The VIRUS design (Hill et al. 2010, Marshall et al. 2010) is a high-throughput, single-arm spectrograph. Figure 4.8 shows the optical design for VIRUS, which can be easily optimized for the DESpec wavelength regime and spectral resolution.

The VIRUS instruments are composed of the main body, housing the collimator and the disperser of the spectrograph, and a vacuum vessel that houses a Schmidt-type camera. A volume phase holographic (VPH) grating acts as the dispersing element. The collimator unit consists of a spherical collimator mirror, folding flat mirror, and the VPH grating. The Schmidt camera consists of an aspheric corrector lens that serves as the Dewar window, a spherical camera primary mirror, an aspheric field flattener lens, and a CCD detector that is held in the center of the camera on a spider assembly. DESpec, like VIRUS, is a set of simple, fixed spectrographs with no moving parts, making it inexpensive, quick to assemble, and easy to mount on or near the telescope.

Figure 4.8: Optical layout of one of the VIRUS unit spectrographs. Light enters on the left and is reflected by the collimator mirror and fold mirror before it passes through the VPH grating. The CCD is just to the left of the field-flattener lens. The overall length of the VIRUS is 0.75m.

With the single-arm spectrograph design we use 100 m diameter (1.75”) fibers with a 2.6 pixel/fiber resolution. That requires an f/1.3 camera (HETDEX VIRUS has f/1.33). We use 20 DECam red-sensitive, fully-depleted CCDs, two for each spectrograph. Each spectrograph has 400 fibers, 200 per CCD, with spectrum along the 4k direction. Table 4.1 lists some of the parameters for the single-arm spectrograph.



Parameter

Single-Arm

Fiber Diameter

100 m (1.75”)

Wavelength Range (nm)

550<<950

CCD

DECam 2kx4k

Resolution (nm/res. el.)

0.263 nm

# pixels/fiber

2.6

Camera f/#

f/1.3

Camera Type

Reflective (VIRUS)


Table 4.1. The one-arm spectrograph concept.

4.C.2 Two-Arm Spectrographs

The two-arm spectrograph design enables an increase in the wavelength range and improves the spectral resolution because we spread the light out over twice as many pixels (CCDs).


Parameter

Blue Side

Red Side

Fiber Diameter

100 m (1.75”)

Wavelength Range (nm)

480<<780

750<<1050

CCD

E2V

DECam 2kx4k

Resolution (nm/res. el.)

0.228 nm

0.228 nm

# pixels/fiber

3

3

Camera f/#

f/1.5

f/1.5

Camera Type

refractive


Table 4.2. An example of a two-arm spectrograph with the same resolution (in nm) on the red side as on the blue side. The break at about 760 nm is to separate the two spectra at the location of a strong sky absorption feature.

Parameters for a two-arm spectrograph covering the wavelength range 480 <  < 1050 nm are shown in Table 4.2. The spectral resolution on both sides is 0.228 nm, sufficient to easily separate the redshifted 3237A O II doublet. An example of a spectrograph of approximately this design was proposed for WFMOS. It is a high-throughput, 2-arm spectrograph with all-refractive optics and VPH gratings (Smee et al. 2006). Another example of this type of design is the conceptual design proposed for the GMACS wide-field, multi-object optical spectrograph for the Giant Magellan Telescope. GMACS will use a fully refractive spectrograph (Marshall, et al. 2011) divided into a blue and red channel by a dichroic. VPH gratings disperse the light. Five (six) lenses focus the red-side (blue-side) of the light onto CCDs, which are at the end of each optical train. The DESpec version would be scaled down to an appropriate physical size. The benefit of a design such as GMACS is the maximal throughput of the system, in part because the beam is not occulted by a CCD in the center of the camera, higher efficiency from the two VPH gratings, and the increased wavelength.

We have adopted the two-armed spectrographs as our reference design and retain the one-armed spectrographs as a possible de-scope option. The larger spectral range (480 – 1050) afforded by the two-armed option allows, for example, both O II 372.7 and H alpha 656.3 to appear within the spectral range for all redshifts between 0.3 and 0.6, which will enhance the certainty of redshift measurement for weak-lined faint objects. The factor-of-two increase in pixels for the two-armed option translates into roughly the same increase in science reach, yet the cost of the two-armed option is much less than twice that of the one-armed option.

4.C.3 Optical Fibers

The light is carried from the fiber positioner to the spectrographs in optical fibers. The diameter of the fibers depends on the expected source flux distribution and the sky background and should be chosen to maximize the S/N ratio of a sky-dominated object spectrum.

The median delivered point-spread-function (PSF) of the Mosaic-II prime focus camera on the Blanco has been 0.9”, and with upgrades related to the installation of DECam it may be a bit better. The plate scale with the DECam corrector is 0.27” per 15-micron pixel. We have initial results from a study that optimized the fiber diameter for a flux-limited survey in the presence of a dominant sky background. The calculations start with a magnitude limit and compute the total number of galaxies per sq. degree at this limit. Next a fiber diameter is chosen, and the rate at which spectra are collected to a pre-determined S/N ratio as a function of galaxy magnitude and radius is calculated. The distribution of galaxy radii and magnitudes comes from the COSMOS simulation (Jouvel, et al. 2009), as described in Sec. 3. Because large galaxies tend to have low surface brightness, they are the most difficult spectroscopic targets. The rates are calculated using a weighted average over a distribution of CTIO-measured seeing. A contribution of 0.6'' from the optics PSF is included. For magnitude limits in the range 22 to 24, the optimal fiber size is in the range 1.8 to 2.0 arcsec. However, the rate at which redshifts are collected depends only weakly on fiber size: the range in fiber sizes that have success rates within 10% of the peak rate correspond to diameters ranging from 1.5 to 2.4 arcsec. The intersection of these ranges that is near-optimal for all limiting magnitudes is 1.7 to 2.1 arcsec. Because the optimization has determined that the fiber diameter is a soft minimum, we expect that the final diameter can be selected based largely on other considerations, such as spectrograph resolution or fiber positioning accuracy.

The input f-ratio, f/3, is ideal for high-efficiency in capturing the light and for high throughput and low focal ratio degradation (Murphy, et al. 2008). The length of the fibers is 10 to 30 meters and depends on where we mount the spectrographs. Figure 4.9 shows fiber throughput versus fiber length for six different wavelengths from 500 to 1100 nm in a typical fiber.

4.C.4 R&D



R&D in this area will lead to a design for the DESpec unit spectrograph and its optics. The choice among designs will come from the science and survey requirements. The R&D will optimize the fiber diameter (for instance, sky-background in some wavelengths suggests smaller fibers improve the signal-to-noise), fiber length, light throughput, and signal-to-noise ratio of the observations. We would also determine how many fibers could be read out on each CCD. An increase here could decrease the number of spectrographs by 30%, and therefore, the total cost. In addition, there is substantial R&D being performed around the world on concepts (e.g. Content & Shanks 2008; Content et al. 2010) for multi-object spectroscopic instruments on wide-field telescopes. We are watching these efforts carefully in case a sensible technical alternative for the DESpec arises.



Figure 4.9: Fiber throughput versus fiber length at six different wavelengths for Polymicro broadband fibers. This assumes optimal f-number, which is f/3 to f/4. Note that for shorter wavelengths the fibers have a lower throughput, which gets progressively worse for <500 nm.

4.D CCDs and Readout

4.D.1 CCDs



The light will be dispersed by the spectrographs onto CCDs. DESpec will use 2k x 4k backside-illuminated, red-sensitive CCDs designed by LBNL, for either the one-arm spectrograph or for the red side of the two-arm spectrograph. These CCDs have high quantum efficiency (QE) at near infrared wavelengths. They are 250 microns thick and attain good (~5 micron) dispersion characteristics from a 40V substrate bias. The 4-side buttable CCD package is suitable, so existing spare, tested, packaged, science-grade DECam CCDs (Estrada et al. 2010) can be used on



Figure 4.10: The absolute QE of three typical CCDs produced for the Dark Energy Camera.

DESpec, providing a significant cost saving, although we include their costs for the purpose of making cost estimates at this time. Fig. 4.10 shows the quantum efficiency of 3 DECam CCDs. The blue side of the two-arm spectrographs could also use DECam 2kx4k CCDs. E2V 2kx4k devices would also work and have QE that is a little better around 500 nm.

The DECam CCD readout electronics can read out the 2kx4k CCD in ~17 (45) seconds with ~7 (3) electrons/pixel noise. They could be modified relatively easily for the E2V CCD if necessary. Even 3e- read noise is somewhat larger than desirable for faint-object spectroscopy as these resolutions (e.g. Saunders et al. 2012 Proc. SPIE 8446-196), but ongoing R&D at Fermilab on low-read-noise techniques (e.g. Estrada et al. 2012 Proc. SPIE 8453-50) is already achieving acceptable readout noise of 1-2e- at the required speeds.

4.E Interchangeability with DECam

The Dark Energy Camera has been designed for efficient installation and removal from the Prime Focus Cage. The time required to change between DECam and DESpec may be limited by the warm-up time for the DECam imager. Instruments at the f/8 focus can be used for time when the prime focus is not available. Figure 4.11 shows the camera installation fixture positioned in front of the Prime Focus Cage, mounted on the Telescope Simulator at Fermilab (Diehl, et al. 2010). To change from DECam to DESpec, one tilts the Blanco over to the northwest platform and uses the camera installation fixture to remove DECam. DECam is then stowed off of the telescope with its Dewar window, the camera’s final optical element (C5), in place. DESpec, which will have been stowed either off-telescope or on the telescope structure, is connected to a similar installation fixture for inserting into the cage. DESpec would be approximately the same weight as DECam; differences in weight would be corrected for by adjustment of the prime focus cage counterweights.



The DECam Project is supplying CTIO with a platform that allows easy installation and removal of the filter-changer and shutter. The platform is with the telescope at the North position. DESpec’s ADC and (the same) shutter will be installed and removed from the same platform using a similar procedure.



Figure 4.11: The camera installation fixture (in the foreground at top left, schematic at top right, in close-up at bottom) at Fermilab. Bottom image shows the installation fixture being used to mount the camera in the Prime Focus Cage (black, at right). The cage itself is attached by fins to the white and yellow rings of the telescope simulator; the inner white ring has the same dimension as the ring at the top end of the Blanco. The simulator was used to test DECam in all configurations it will encounter on the telescope as well as the mounting and dismounting procedures.
1   2   3   4   5   6   7   8   9   10


The database is protected by copyright ©dentisty.org 2016
send message

    Main page