WIYN Instrumentation:
Bench Spectrograph Upgrade

These project will commission and document all of the new subsystems, and provide software for scientists to explore and optimize different spectrograph configurations.

A working proto-type of this configuration tool is available. This GUI can be used to optimize setups and calculate exposures for all gratings and fiber cables (calibrated with measured system throughput values where available; see Simulator). Modification of the camera-collimator angle (for conventional gratings) can optimize blaze-wavelength and anamorphic factors.

A more complete description can be found in our Sept 2003 report to the WIYN SAC and Board, which serves as our Concept Design, in the links above, and in the project schedule and status.

< Collimator.

As of Apr 2006, the project is currently working toward a preliminary technical design review of the collimator optical design and concept-design review for the opto-mechanical design. An optical design (Feb 2006) and initial tolerancing (Apr 2006) has been completed. Opto-mechanical design and layout is in progress, and in the early stages.

A suite of two initial VPH gratings are part of the Upgrade project.

• Commissioning is complete on one (740 l/mm) VPH grating, and is in regular use. See this link for more details.
• A second (3300 l/mm) VPH grating has been designed, and fabricated by CSL (delivery in Feb 2006). Preliminary testing on-spectrograph and on-sky (April 2006) are under analysis. Future development includes coating for high incidence angles (50-70 degrees), design and fabrication of the grating cell, and final commissioning. Further information about this grating is available below at these links: [1] [2].

CCD and Controller.

A new CCD and controller is planned for the upgraded system. The rear-element of the camera will be replaced assuming the device is significantly flatter than the original CCD used on the spectrograph.

The new CCD system will have smaller pixels (better sampling of small-fiber high anamorphic-demagnification configurations important for the highest-resolution work), lower read-noise (critical for sky-limited work at high dispersion which is read-noise limited with current systems), and faster read-out time. CCDs are currently being culled from test-lots from wafers used in development of OTA devices for ODI. The nominal upgraded controller systems is MONSOON. Fall-back options will be explored if foundry runs yield no suitable devices, and if MONSOON is delayed or cannot yield low-noise environment. So far, results on CCDs look promising.

Optical Design

Description.The new collimator consists of an off-axis parabolic segment with an all-spherical, all-glass, transmissive corrector. The collimator is "faster" than the existing collimator by roughly 20%, i.e. with a focal length decreased from 1021 mm to 800 mm. The primary aim of the shorter focal length is to capture more of the fiber-output beam into the 150mm collimated beam for which the gratings and camera were designed, while preserving the spectral resolution achievable in the highest-resolution settings with the echelle grating where anamorphic factors are large.

The design is heavily constrained by the project-level requirement to keep the existing camera. The initial design consisted of a 3-element corrector with tilted elements (akin to the Wynne triplet, but using tilted, full spherical segments). A preliminary tolerance analysis showed the Bench implementation was likely unbuildable. The current off-axis design has 4 corrector elements and yields improved image performance than the existing on-axis design. Other considerations included:

spectrograph geometry

beam profile

throughput budget

delivered spectral resolution:

geometric demagnification

anamorphic demagnification

optical aberration

pixelization

S/N

The details of these considerations are found in two reports to the WIYN SAC and Board:

• Sep 2003 Concept Design
• Mar 2004 Trade Study: S/N vs line-characterization precision

Spectrograph Geometry: Detailed notes on optical element size and location, as well as obstructions. This is used for determining both the required layout of the upgrade Bench, as well as the throughput budget.

The existing surface-relief grating suite for the WIYN Bench Spectrograph delivers a wide range of coverage in wavelength and resolution, as shown here in the left-hand figure. A completed Bench upgrade may include a set of VPH gratings which replace or augment the current capabilities of the existing gratings in this plane. Some examples of possible VPH grating suites are shown in the right-hand figure.

Two gratings resulting from the initial VPH effort, as contracted to Centre Spatial de Liege (CSL), will be part of the initial Bench Spectrograph upgrade: 740 l/mm and 3550 l/mm gratings. These are shown as red curves in the above, left figure. At this time, testing is underway on the 740 l/mm grating. The development is mature enough to offer the grating in Shared Risk mode for 2005B. We have taken delivery of a test-version of a small high-line-density grating (3300 l/mm). This was made on float glass and is not science grade. The high-density (3550 l/mm) science-grade vph grating is still under manufacture as of April 2005.

Grating Pages:

• v740a:     740   l/mm grating -- science grade
• v3300a:   3300 l/mm grating (CSL/WP3200) -- test grade
• v3550a:   3550 l/mm, large high-density grating (CSL/WP4200) -- science grade
  • Design
  • Optimization:   3550 l/mm, 510.6 nm @ 65^o
  • Expected Performance 

• Opto-mechanical   issues

Notes-

See v740a   page for performance and optomechanical details. The effective d for this grating apparent comes from "RCWA interpolation" (Pierre-Alexandre Blanche, CSL, private communication, 2004.02.09)
See v3300   page for performance and optomechanical details.
Diamant float glass vendor: Saint-Gobain Glass.
Green numbers are measured values as delivered.

A new CCD detector will be put in the existing dewar, and upgraded with new control electronics. The current CCD is warped, has large (24 micron) pixels, moderate read-noise (4-10 e- rms delivered), slow read-time (2+ minutes) and no enhanced red or blue response. While the chip is cosmetically very clean, significant improvements can be made with newer technology, including:

• a flat detector (improved focus over field)
• smaller pixels (better sampling)
• higher QE in red and blue
• lower read-noise (RN; 2e- rms goal)
• faster-read time at low noise

These design goals can be compared to as-delievered, on-telescope performance of two state-of-the-art systems using Leach-2 controllers:

• The ESI on Keck, using a MIT-Lincoln Labs 2048 x 4096 device with 15 &mu m pixels, slow read-out time of 120 sec and RN of 2.7 e-, with QE peaking at or above 80% between 500-650 nm, and above 60% between 400-800 nm. (See this link.)
• The RSS on SALT, using 3 E2V 2048 x 4096 device with 15 &mu m pixels, and slow read-out time of 42 sec and RN between 2.3 and 2.9 e- (amp dependent), with QE peaking above 75% between 350-650 nm, and above 50% down to 900 nm. Full well is measured at 150-180 k e-.

CCD Subsystem: Design Requirements

version 1 - internal draft
version 2 - working document reviewed by WIYN SAC
version 3 - final project document

One ramification of this upgrade is that with a flat detector, the last camera element will be changed. The opportunity will be taken to re-coat the camera element to improve red and blue response and at the same time decrease scattering.

Lot 1 foundry run

As of September 2004, a successful foundry run at Lincoln Labs for OTA CCDs has also produced 10 very promising, standard devices for the Bench. These are 2600x4000 12 micron pixel devices. Three to four of them have a single hot or blocked column. Five of these are now at Mike Lesser's lab for cold testing.

WARM images taken with eight of these are shown below. Note the left side of w10 and w11 are confused, and may be the same half-device - the bad column is exactly in the same place on both. These are all warm, relatively high illumination images - some things get better cold and others get worse.

w01 w03 w04 w05
w07 w09 w10 w11

Lot 2 foundry run

This URL shows images of each of the 12 CCDs in Lot 2. Each is labeled as wN.gif, where N is the wafer number. Wafers 9-12 are made with the high resistivity material that allows for better red response. Also see George's notes from the report on CCDs for the last telecon (pre 19 Dec 2005).

Surface flatness for one device

This detector (SN 5652) is flat over the great majority of its imaging area to +/- 5 microns, and is typical of the devices from Lot 2, and the device (5644) ultimately adopted for the Bench system.

Device: STA1 (SN 5644), 12 micron 2600x4000 pixels, low-resistivity silicon

Controller: MONSOON

Parameter summary     [Detailed information provided in links]

• pixel size: 12 &mu m, unbinned
• binning modes: 1x1, 2x1, 2x2, 4x2, 4x3 (spatial x spectral) enabled
• gain modes:

  High

  0.21 e-/DN

  Medium

  0.44 e-/DN

  Low

  0.87 e-/DN

• read-out modes: single-amp; two-amp option in future
• read-out times: seconds, single-amp (two-amp is one-half)

  Gain/Binning:	1x1	2x1	2x2	4x2	4x3
  High	227	131	67	43	29
  Medium	140	87	45	31	22
  Low	97	65	34	25	18

• detector-noise: e- (rms)

  		Detector
  Gain/Binning:	Overscan	1x1	2x1	2x2	4x2	4x3
  High	3.4	3.7	4.0	4.1		4.4
  Medium	3.8			4.3		4.4
  Low	5.5			5.9

• QE:^* >80% 460-770 nm, >50% 350-870 nm at 160 deg (K)
• dark current: 0.0008 e-/pix/sec @ 152 deg (K) typical
  
  • 50% < 7.1 x 10^-4 e-/pix/s
  • 80% < 6.3 x 10^-4 e-/pix/s
  • 90% < 1.3 x 10^-3 e-/pix/s
  • 95% < 1.5 x 10^-3 e-/pix/s
  • 99% < 2.1 x 10^-3 e-/pix/s
  • mean of top 1% (100836 pixels) = 0.012 e-/pix/s
  • max: 1.5 e-/pix/s
• overscan: 50 pixels
• linearity: 0.2% (10,000 e-/pix), 1% (63,000-70,000 e-/pix)
• CTE:^* 99.99998% (vertical), 99.99996% (horizontal) at 160 deg (K)
• cosmetics: No bad columns
• data-taking interface: GUI
• data format: FITS
• data header: standard IRAF keywords

^*Note: These values should be verified at final operating temperature.

Detector noise and bias structure

o Overscan vs detector noise
o Charge-injection and bias structure

QE

The STA1 device provides a 30% throughput performance gain relative to T2KA due to higher QE alone, flat across wavelength. The improvement relative to the original detector, T2KC, is roughly half this value. The following figure (Bershady et al. 2008, SPIE, in prep) illustrates and compares laboratory and telescope measurements.

Associated data files:
o on-telescope measurements
o lab measurements

Dark Current

o As a function of detector temperature.

Data-taking interface

Data headers

The following set of header key-words were populated as part of the requirements document. The only issue outstanding concerns the defintion of a detector-noise parameter as distinct from read-noise (RDNOISE), which is populated with the rms of the overscan (not relevant to data processing or performannce).

• TCS keywords
  
• Bench keywords
  
• CCD keywords
  
• Hydra keywords