Big BOSS Preconceptual Design Concepts Michael Sholl UCB

Big. BOSS Pre-conceptual Design Concepts Michael Sholl (UCB) DOE Review of Big. BOSS, Dec. 6 -7, 2011

Overview • The team • Visible & NIR ELGs, LRGs & QSOs, the low-hanging fruit of dark energy research • The survey • Spectrograph trades and baseline • Telescopes suitable for big. BOSS • Widefield corrector • Fiber positioners • Systems engineering: throughput • Risk reduction and targeted R&D — — Corrector Optics Spectrograph Actuators Thermal 2

Who’s working on this? • • • • • Mark Ackerman (UNM) Eric Anderssen (LBNL) Marco Azzaro (IAA) Charles Baltay (Yale) Chris Bebek (LBNL) Santiago Becceril (IAA) Robert Besuner (UCB) Arjun Dey (NOAO) Peter Doel (UCL) Jerry Edelstein (UCB) Will Goble (NOAO) Bruce Grossan (UCB) Henry Heetderks (UCB) Patrick Jelinsky (UCB) Dick Joyce (NOAO) Robin Lafever (LBNL) Michael Lampton (UCB) • • • • Michael Levi (LBNL) Ming Liang (NOAO) Nick Mostek (LBNL) Paul Perry (LBNL) Claire Poppett (SSL) Francisco Prada (IAA) Eric Prieto (LAM) Michael Raffanti (UCB) Natalie Roe (LBNL) David Sawyer (NOAO) Christoph Schenk (LBNL) David Schlegel (LBNL) Joe Silber (LBNL) Chao Zhai (USTC) Zengxiang Zhou (LBNL) • …and others 3

Atmospheric transmission good to 1 micron, poor beyond http: //www. gemini. edu/sciops/Obs. Process/obs. Constraints/oc. Trans. Spectra. html J band H band K band 1. 6 mm H 2 O at Gemini North M. Sholl, P 3, 6 December 2011 4

Atmospheric emission manageable to about 1 micron http: //www. gemini. edu/sciops/Obs. Process/obs. Constraints/oc. Sky. Background. html Sky brightness at visible wavelengths: not too horrible (linear scale) Sky brightness at NIR wavelengths: horrible (note log scale) M. Sholl, P 3, 6 December 2011 5

Low-hanging fruit for Big. BOSS • Low absorption and emission over 360 -980 nm band • Lyman-α from Quasi-Stellar Object (QSO) — 121. 6 nm rest frame (1 nm=10Å) — At z ≥ 2, Ly-α redshifted to (2. 2+1)× 121. 6 nm ≥ 389 nm (Big. BOSS!) • [OII] doublet from emission line galaxies (ELG) — 372. 7 nm and 372. 9 nm rest frame — At z ≤ 1. 6, [OII] redshifted to (1. 6+1) × 372. 9 nm ≤ 970 nm • 4000Å break of luminous red galaxy (LRG) — 4000Å rest frame — At z ≤ 1, LRG redshifted to (1+1)*400 nm ≤ 800 nm (Big. BOSS) 6 M. Sholl, P 3, 6 December 2011

Telescope and Survey • Main goals: — 20, 000 galaxy spectra — 14, 000 square degree survey — [OII] at redshift range of z=0. 5 to 1. 6 • S/N=7 at a line flux of 9 × 10 -17 ergs s-1 cm-2 • 1000 second exposure • Considered in exposure time calculator — Detector readnoise, dark current, quantum efficiency — Effective aperture — Reflective losses — Transmission losses (if prime focus corrector) — Spectrometer throughput — See N. Mostek talk, B 4. 5 • A 3. 8 m telescope and a 5, 000 fiber multi-object spectrograph (MOS) meets these requirements 7 M. Sholl, P 3, 6 December 2011

A MOS survey is driven by greed. • How many fibers can you fit on your spectrometer(s)? • How many fiber positioners can you afford? • How many fiber positioners can you accommodate? • Our goal in defining the Big. BOSS architecture is to maximize science throughput by evolutionary, not revolutionary means. • 5000 fiber positioners baselined after examining existing spectrometer and fiber positioner technology suitable for adaptation to Big. BOSS. (LAMOST has 4000 positioners) • These choices lead to the following questions: — How small/large should the positioners be? — What FOV is optimal? M. Sholl, P 3, 6 December 2011 8

Optimizing FOV & fiber patrol radius Parameter Value Survey 14, 000 deg^2 Galaxy Density 2, 800 galaxies/deg^2 Target Density 2, 300 galaxies/deg^2 ELG Exposure 1 time unit LRG exposure 2 time units QSO exposure 5 time units Fibers 5000 First Visit Observe this target Success = • Assume galaxy distribution in fiber patrol radius follows Poisson distribution • • Vary FOV and Patrol Radius Compute fiber efficiency and survey time Second Visit Observe this target Success = nth Visit Observe this target Success = 9

5000 Fiber Optimal FOV 2 14, 000 deg 2 survey, observed density of 2, 300 galaxies/deg 2 FOV (°) 2. 0 2. 2 2. 5 3. 0 3. 5 4. 0 M. Sholl, P 3, 6 December 2011 Galaxies Fiber Spacing Visits/FOV per patrol Efficiency (mm) radius 0. 63 3 8. 0 2. 15 0. 60 4 8. 8 2. 60 0. 73 4 10. 0 3. 36 0. 81 5 12. 0 4. 84 0. 82 7 14. 0 6. 59 0. 85 9 16. 0 8. 61 Survey Time (QSO) (Galaxy) 1. 35 2. 25 1. 49 1. 86 1. 15 1. 44 1. 00 1. 03 1. 01 10

Spectrometer Considerations • 5000 fibers • Three bands (dichroic splits) — 360 nm < λ < 660 nm: R > 1500 — 620 nm < λ < 840 nm: R > 3000 — 800 nm < λ < 980 nm: R > 4000 • Science-specific constraints on the PSF — Maximize survey speed by matching fiber to target size • Goals of pre-conceptual design — Mass production, cost-effective high-throughput design 11 M. Sholl, P 3, 6 December 2011

Spectrometer Considerations • Benefits of pre-conceptual baseline design — Eliminate oil, grease, coupling fluids from design (air-spaced) — Use standard, robust and widely available materials — Eliminate motorized alignment motors (set once, not touched again) — Minimize optics size — Minimize obscurations — Room-temperature cameras — Off-the shelf sensors 12 M. Sholl, P 3, 6 December 2011

Spectrometer Landscape • • “Before attempting to create something new, it is vital to have a good appreciation of everything that already exists in the field” 20 existing/planned spectrometer examined — Refractive versus reflective — Collimator f/number — Camera f/number — Lens materials (by designer, evolution over time) 13 M. Sholl, P 3, 6 December 2011

Survey of Collimators used on ground-based spectrometers 14 M. Sholl, P 3, 6 December 2011

Generalizations (Exceptions likely exist) • Other than designs by Epps, collimators tend to be reflective • Faster than f/4, all existing designs are reflective • Schmidt collimators work well below f/4, but are larger than refractive collimators. — Stray light issues from light reflecting off fiber slit block — Fibers must be routed through collimator pupil (exception: VIRUS) • Pre-conceptual baseline design (Prieto talk B 5. 5) has f/4 refractive collimators and f/2 refractive cameras 15 M. Sholl, P 3, 6 December 2011

Pre-Conceptual Design (current baseline, September of 2011) UV NIR VIS 16

Resolving power based on 50% encircled energy dλ 17 M. Sholl, P 3, 6 December 2011

Next step, investigate ~4 m telescopes are compatible with Big. BOSS f/4 spectrometer? • Prime focus configuration selected for Big. BOSS • Rationale: — Cassegrain telescopes allow high magnification with a short package — Wide-field surveys don’t require high magnification • M 2 does not magnify significantly, it acts as an aspheric plate (and a very expensive one) — A fully-baffled wide-field Cassegrain telescope requires 50% or greater central obscuration 18 M. Sholl, P 3, 6 December 2011

Design study: Can Big. BOSS-like f/4. 5 prime focus correctors be placed on 4 m class telescopes? • • Faster speed M 1 = More difficult Larger M 1 = More difficult (due to C 1 diameter) 19 M. Sholl, P 3, 6 December 2011

f/4. 5 3º Prime Focus Correctors ~4 m telescopes, increasing M 1 f/# WIYN M 1: f/1. 8 Ø 3. 5 m William Herschel M 1: f/2. 5 Ø 4. 2 m AAT M 1: f/3. 2 Ø 3. 9 m DCT M 1: f/1. 9 Ø 4. 2 m Mayall/CTIO M 1: f/2. 8 Ø 3. 8 m Galileo TNG M 1: f/2. 2 Ø 3. 6 m AEOS M 1: f/3. 0 Ø 3. 7 m Calar Altar M 1: f/3. 5 Ø 3. 5 m NTT-ESO M 1: f/2. 2 Ø 3. 5 m ESO M 1: f/3. 0 Ø 3. 6 m CFHT M 1: f/3. 8 Ø 3. 8 m 20

Corrector constraints • • FOV: 3° Chief ray normal design — f/4. 5 f/4 (spectrometer input) • <0. 5º (mean) chief ray deviation • FRD — Jerry Edelstein to talk, 2. 3, 6 December 2011 • LLF 6 (used on existing Mayall corrector)is no longer available. Use fused silica, N-BK 7 and LLF 1 for corrector. — Maximum size of LLF 1: ~0. 9 m diameter • • Maximum corrector lens element size: 1. 25 m (cost-prohibitive beyond this diameter) Focal plate may be curved, but ROC should be greater than 2. 5 m (as large as possible, to avoid constraints on actuators) M. Sholl, P 3, 6 December 2011 21

Corrector Challenges • f/2. 8 f/4. 5 magnification — Required for fiber injection margin, and spectrometer input — Spectrometer designed for f/4 input (accommodate 500 fibers) — Magnification and chroma correction requires large, thick elements — Chromatic aberration correction necessary • 0. 36 to 0. 98µm PSF must enter 120µm fiber core • ADC uses LLF 1 for chromatic correction and control of atmospheric dispersion • All other corrector elements can be fused silica or N-BK 7 — Vignetting-free design for maximum throughput • Better performance can be achieved by allowing pupil to lie between M 1 and the prime focus. This leads to 5 -10% vignetting. (See Ackerman OMA designs) M. Sholl, P 3, 6 December 2011 22

Corrector Challenges • Comparable projects and other options — LSST and DECAM are filtered imagers, and do not magnify to f/4. 5 • Not suitable for MOS as currently configured (chromatic correction only over narrow filter bands) — Cassegrain and Gregorian options (use M 2 for magnification) do not work well • focal plane mid-way between M 1 & M 2 • Geometric performance is poor — Options that use ~flat M 2 for aberration correction only are unduly expensive 23 M. Sholl, P 3, 6 December 2011

Pre-conceptual Corrector Cutaway Focal Plate Fiber Actuators C 3 & C 4 (fused silica) ADC 1 & ADC 2 (LLF 1, N-BK 7) C 1 & C 2 (fused silica) rre a r. B o t ec teel) r r (S Co l nts u o lls M c s Ce to i r me Len aces l o t s 8 rf re Ela var 3 Inte Bar In ure ctor x e Fle orr C 24 M. Sholl, P 3, 6 December 2011

Atmospheric Dispersion Correction Geometric raytrace shows eects of atmospheric dispersion on telescope point spread function. a) A heavily chromatically aberrated view of the sky 60 from zenith. Overall scale is 1 m square, PSF exaggerated by factor of 106. This chromatic aberration is removed by rotating the ADC prisms 85 as shown in b). The dispersion being compensated here is 3 arcsec. 25

How do you correct atmospheric dispersion? Wynne (1984), Liang (2004 & 2009) Pair of doublet, zerodeviation wedge prisms LLF 1 High dispersion N-BK 7 Low Dispersion s 2 x DC e riv D A 26 M. Sholl, P 3, 6 December 2011

Atmospheric Dispersion Compensator ADC 1 Dispersion ADC 2 Dispersion + Atmospheric Dispersion ADC 1 Dispersion = Net ADC Dispersion ADC 2 Dispersion + Image Dispersion Atmospheric Dispersion = Net ADC Dispersion Image Dispersion 27 M. Sholl, P 3, 6 December 2011

Note on ADC • Near the horizon, differential dispersion across the FOV (M. Azzaro, M. Lampton) will require motion of the some fibers during an observation. This motion, called “dead reckoning” must work without the fiber view camera. 28 M. Sholl, P 3, 6 December 2011

d 95 Statistics • 1530 kg glass • LLF 1, N-BK 7, Silica • Six groups — Two aspheres — Constrained aspheric slope departure (<30µm/mm) — ADC (spherical external prism surfaces) — 15 mm radial on lenses for mounts • 2. 7 m ROC convex focal surface, 0. 95 m diameter 29

Pre-conceptual baseline corrector (d 95) geometric blur performance 30

Fiber Positioners (actuators) • 5000 fiber positioners — 3˚FOV — 12 mm actuator pitch • Fiber positioning technologies — Plug Plate (a-la BOSS) • Too slow to reposition • Strongly curved focal surface (>2. 5 m ROC) — Echidna, tilted spine (Can work at 12 mm pitch, but tilt is high) — LAMOST θ-θ (25. 6 mm pitch) • Reference design includes 12 mm USTC actuators 31

Can Echidna work for Big. BOSS? • • 12/sqrt(3)=6. 92 mm patrol radius 160 mm spine length 2. 5˚ tilt relative to chief ray! Throughput loss! • System at telescope focus is f/4. 5 (12. 5º) • Collimator designed for f/4 input beam (14°) • • • System margin for fiber misalignment lies between f/4. 5 and f/4. At 1. 5º field angle, an f/4. 5 telescope beam requires f/3 collimator to work with Echidna! Tilt of Echidna spines (2. 5˚) leads to focal ratio degradation (FRD) and reduces system throughput significantly AND throughput is a function of fiber position within patrol disk! Actuator tilt throughput loss can be minimized through use of a low-tip actuator topology (θ-θ or R-θ) Requires multiple iterations, which are not suitable for “dead reckoning” 32

Cobra Actuators • Cobra (θ-θ) — Innovative design — Cobra has a pitch of 8 mm • Could be increased to 12 mm? — Have not been deployed on a spherical focal surface — Uses Squiggle Motors, somewhat complex electronics • Reliability for 5000 actuator deployment not established — Need 6 -8 iterations with fiber view camera • Not useful for dead reckoning — Expensive unit cost (Newscale) • Prototypes: $50 k for five prototypes — Will investigate during R&D period 33 M. Sholl, P 3, 6 December 2011

Low-tilt actuators • • LAMOST 25. 6 mm pitch, theta-theta (USTC) SIDE 30 mm pitch, theta-theta (IAA) LBNL: 20 mm pitch, r-theta Efforts underway to miniaturize actuators for 12 mm pitch • USTC, IAA and LBNL all have promising 12 mm pitch designs, with the low-tip inherent to theta-theta actuators (<0. 5 degrees tip) • Please see C. Zhai talk 5. 4, 7 December 2011 34 M. Sholl, P 3, 6 December 2011

Big. BOSS Systems Engineering • Main task of Big. BOSS Systems Engineering is management of a throughput budget — Atmospherics — Obscurations — M 1 reflectivity — Corrector throughput • Material absorption • Coating reflection — Fiber injection efficiency • Chief ray normal orientation at fiber • Galaxy image size at fiber • Errors in lateral position of fibers — Fiber throughput — Spectrometer M. Sholl, P 3, 6 December 2011 35

Optical System Throughput * * * WFC=Wide Field Corrector M. Sholl, P 3, 6 December 2011 36

Fiber Injection Efficiency • Fiber injection efficiency — Galaxy image size relative to fiber diameter • • Galaxy size (Sersic inverse exponential galaxy, 0. 3 arcsec EE 50 radius) Atmospheric seeing (Moffat 1 arcsec FWHM, β=3. 5) Geometric blur (residual phase, as-built, thermal & dynamic) Angular Misalignment of chief ray with respect to fiber normal — Lateral misalignment of galaxy image relative to fiber • Galaxy image convolved with circular fiber to estimate throughput • Sources of misalignment between fiber and galaxy image are key to throughput budget — — • Astrometry errors Focal plane thermal shifts Dynamics (rotational mode of corrector on spider support vanes) Fiber positioning errors (mechanical and fiber view camera) Throughputs may be estimated by RSSing uncorrelated misalignment components or multiplying resulting throughputs 37 M. Sholl, P 3, 6 December 2011

Fiber Misalignment Throughput 0. 30 arcsec, EE 50 radius Sersic galaxy 1 arcsec FWHM seeing Moffat Beta parameter of 3. 5 M. Sholl, P 3, 6 December 2011 38

Lateral Galaxy Position Error Breakdown • Largest uncertainties are astrometry — Currently 200 mas • Tracking/pointing errors currently at 100 mas — Big. BOSS focal plane trackers should improve • Fiber mechanical lateral error: 125 mas • Fiber defocus error: ± 60µm 39

Lateral Galaxy Position Error Breakdown • Largest uncertainties are astrometry — Currently 200 mas — Will improve over time • Tracking/pointing errors currently at 100 mas — Big. BOSS focal plane trackers should improve • Fiber mechanical lateral error: 125 mas • Fiber defocus error: ± 60µm • Small numbers individually, but costly to throughput when included in an overall system budget 40

Pre-conceptual Baseline, Risks & R&D Plan • Mayall 4 m telescope • d 95 widefield corrector with ADC • September 2011 LAM f/4 -f/2, 500 fiber transmissive spectrometer • USTC 12 mm pitch θ-θ actuators • Convex IAA focal plate 41 M. Sholl, P 3, 6 December 2011

Pre-conceptual baseline risks • Mayall 4 m telescope — Risks • Will Kitt Peak be shut down? • Is 4 m telescope and mount compatible with Big. BOSS — Strength and deflection — Cooling services — Structural dynamics — Mitigation • Ongoing investigation as to suitability of other 4 m telescopes — CFHT — Blanco 4 m with heavily modified DES corrector • Semi-weekly coordination meetings with Kitt Peak — FEM of existing telescope and Big. BOSS implementation — Accelerometer measurement of 4 m dynamics underway (accelerometers installed on Mayall prime focus cage and gathering data as we speak) — Investigation of routing paths for BB services underway 42 M. Sholl, P 3, 6 December 2011

Pre-conceptual baseline risks • d 95 widefield corrector with ADC — Risks • • Large corrector Glass availability? Corrector magnification Glass procurement — Mitigation • Ongoing pre-conceptual structural modeling, tolerancing, mount design, exploration of alternate configurations • Glasses selected from current production capability of Schott and Corning • Examination of alternate optical configurations ongoing and TBD during R&D phase over next two years • Re-evaluation of possibility of faster, perhaps Schmidt-collimator-based spectrometer during R&D phase 43 M. Sholl, P 3, 6 December 2011

Corrector Optics R&D • • Work with LBNL, Kitt Peak (M. Liang) and independent contractor (M. Ackerman) to develop and simplify corrector Example: M. Ackerman One Mirror Anastigmat concepts 44 M. Sholl, P 3, 6 December 2011

One Mirror Anastigmat, five group, f/4. 3 • 1260 kg glass 45

One mirror anastigmat, five-group f/3. 53 (requires 8 mm actuator) • 1127 kg glass 46

What about DES corrector? • Current design works over 2. 2º FOV • Vignettes significantly for larger fields (even 2. 5º, see next slides) • Can DES corrector be modified to work with ADC? — Yes, with minor changes, to 2. 2º FOV — Yes, with major structural modifications, to 2. 5º FOV 47

DECAM Baseline Corrector, 2. 2˚ FOV M. Sholl, P 3, 6 December 2011 Starting Point: 2602 -v 203 -070122 -distrib 48

Vignetting of DECAM at 2. 5˚ Starting Point: 2602 -v 203 -070122 -distrib M. Sholl, P 3, 6 December 2011 49

Unacceptable vignetting of DES Corrector over 2. 5° FOV M. Sholl, P 3, 6 December 2011 50

Can horrendous vignetting at 2. 5º be fixed? • Main vignetting on: — C 4, C 5 • Can C 4&C 5 these be enlarged to reduce vignetting? (yes) — C 4 diameter Ø 604 mm Ø 640 mm — C 5 diameter Ø 542 Ø 568 mm 51

Vignetting of DECAM with redesigned back end (C 4&C 5) M. Sholl, P 3, 6 December 2011 52

Can an ADC be added to baseline DECAM? • Wavelength band: — 0. 37 -1. 0µm • ADC must be inserted in filter gap (1 cm gap between ADC prism elements (including sag) and filter gap • Existing C 4 must be replaced to extend FOV to 2. 5º • Aft end lenses that replace C 5 should be behind DES cryostat mount flange • Preliminary design study results: YES — Maximum chief ray deviation: 0. 2° — Blur at Zenith: 15. 1µm RMS, 0. 626” FWHM • Although the design could be made to work at 2. 2° and 2. 5°, the survey is diminished significantly (when compared to a 3° corrector) due to aforementioned galaxy Poisson statistics 53 M. Sholl, P 3, 6 December 2011

Big. BOSS Corrector R&D • The ADC, and performance at low elevations is a key challenge of the system. Further studies will be performed to understand the differential dispersion and higher order effects during the R&D phase — TBD by M. Liang, M. Sholl & M. Ackerman • Work with Kitt Peak (M. Liang), LBNL and independent contractor (M. Ackerman) to develop and simplify corrector — Example: M. Ackerman One Mirror Anastigmat concepts — Recent studies show that even simpler and lighter (<800 kg glass) configurations exist, albeit at f/3. 5 — This leads to the question of whether a 5000 fiber spectrometer system can be made to work at roughly f/3 54 M. Sholl, P 3, 6 December 2011

Recall survey of collimators presented earlier 55 M. Sholl, P 3, 6 December 2011

GYES Spectrograph at CFHT: S. Mignot et al Proc SPIE v. 7735 2010 500 fibers, 152 mm slit height, f/3. 5 collim, f/1. 2 cameras 4 k spatial pixels, 8 k spectral pixels, 2. 8 pixels/resel. 56 M. Sholl, P 3, 6 December 2011

VIRUS 57

J. Gunn “A Large Multifiber Faint-Object Spectrograph for SUBARU” Schmidt collimator f/2. 5; Schmidt cameras f/1 http: //subarutelescope. org/Science/Subaru. UM 2010/files/Gunn_sumire. UM. pdf 58 M. Sholl, P 3, 6 December 2011

Spectrometer R&D • R&D explorations — Investigate Schmidt collimators at f/3, and Schmidt and transmissive f/1 to f/1. 5 cameras • • Cryogenics Dichroic splitting Geometric performance Obscuration (in case of Schmidt camera) — Investigate manufacturability of reference spectrometer • Materials • Aspheres • Alignment — Can a 2 -arm spectrometer work for Big. BOSS? • VPH grating efficiency and scattering • Spectrograph pupil size — Work TBD by J. Edelstein, P. Jelinsky, E. Prieto and M. Sholl 59 M. Sholl, P 3, 6 December 2011

Actuator and Focal Plate R&D • Work to define structure of focal plate and interface to actuators, trackers, corrector barrel and the fiber system. — IAA/LBNL/USTC • Further R&D work on USTC 12 mm actuator — Packaging — Power dissipation — Precision and accuracy — Repositioning time • • Further R&D work on LBNL R-θ and IAA θ-θ actuators Work to be done by USTC, IAA and LBNL 60 M. Sholl, P 3, 6 December 2011

Corrector Thermal System R&D • Heat is generated by the actuators, within the telescope line of sight. — How can heat be dissipated from actuators without adversely effecting seeing? — Novel concepts for removing heat without a risky liquid cooling system at the prime focus are proposed for Big. BOSS — Convective heat transfer analysis in infancy (see Sholl Cage & Barrel & Optics, talk 2. 2, 7 December 2011) — Much work to be done in next two years • Conduction/Natural Convection/Forced Convection analysis — CFD — Experiment • Work to be done by LBNL, IAA and USTC 61 M. Sholl, P 3, 6 December 2011

Conclusions • • A reasonable pre-conceptual baseline design for the Big. BOSS mission was presented Goal is maximum system throughput During 2 -year R&D phase, all aspects of the pre-conceptual baseline will be re-evaluated to ensure maximum science/dollar R&D areas include — Site — Corrector — Spectrometer — Fiber system — Actuators — Thermal • Big. BOSS is of key importance in the study of dark energy. The team understands the risks associated with the pre-conceptual system, and has in place a viable risk-reduction plan to be implemented in the R&D phase M. Sholl, P 3, 6 December 2011 62