1. Overview – LSST Camera, sensors, science raft subsystem P. O’Connor BNL Jan. 25, 2012

2. LSST PROJECT

3. LSST project Imaging survey of the entire sky with a large ground-based telescope NSF/DOE joint project: Key innovations: – Wide field of view – Wide spectral coverage – Fast readout – Camera located in optical beam NSF (lead)DOE Provides: TelescopeCamera Site Data management Science goals: solar system inventoryDark energy Milky way mapDark matter Optical transients

5. Preliminary Design Review Tucson, Arizona August 29 th – September 2 nd, 20115 The LSST Optical System - Modified Mersenne- Schmidt/Paul-Baker Design Primary and Tertiary Mirrors Secondary Mirror Camera Lenses Rest of camera

6. 6 STAR LSST size compared to STARdetector 8.4m

7. LSST CAMERA AND FOCAL PLANE

8. Camera – cross section 1.6m

9. Focal plane Science rafts (3 x 3 CCDs) Corner rafts Guide sensors Wavefront sensors

10. 10 TOWER CCDs + front end electronics 180K operation An autonomous, fully-testable and serviceable 144 Mpixel camera thermal straps FEE boards housing (cold mass) cooling planes RAFT 9 CCDs coplanarity 13.5  m 12.5 cm 4K x 4K CCD 10  m pixels = 0’’.2 extended red response 16 outputs 5  m flatness 4 cm CRYOSTAT FOCAL PLANE WITH 21 SCIENCE RAFTS 21 “science rafts” make up the 3.1Gpix focal plane

11. 11 LSST’s camera will surpass existing survey instruments in four areas: The largest focal plane –LSST: 3.1Gpix(189 CCDs) –PanSTARRS GPC1: 1.4Gpix (60 CCDs) –HyperSuprimeCam: 940Mpix(112 CCDs) –DECam: 500Mpix (60 CCDs) –CFHT MegaCam: 340Mpix (36 CCDs ) The fastest focal ratio –LSST: f/1.23 –SuprimeCam:f/1.87 –DECam:f/2.7 –PanSTARRS: f/4 –CFHT MegaCam:f/4.2 The fastest readout time –LSST: 2s –PanSTARRS GPC1: 6s –DECam: 17s –CFHT MegaCam: 40s –Suprime-Cam: 18s The highest data rate –LSST: 1.0TB/hr –PanSTARRS GPC1: 0.22 –HSC 0.03 –DECam: 0.004 –CFHT MegaCam: 0.003 DECam HSC MegaCam GPC1 LSST

12. REQUIREMENTS

13. 13 LSST system performance goals High etendue: –primary mirror effective aperture 6.7m, field of view diameter 3.5 degrees High throughput: –> 80% (temporal), > 90% (spatial), > 80% spectral efficiency (visible) Short exposures: – two back-to-back 15s exposures per “visit” Image quality: –the system contribution to delivered image size should not exceed 15% Sensitivity: –achieve 5  image depth of r=24.7 in 30s (about 10 -4 photons/cm2/s) Photometric repeatability: – 0.5%, accuracy 1% Astrometric accuracy: –.05” (240nrad) Survey lifetime: –10 years (3x10 6 visits over 20,000 square degrees of sky, ~40PB image data)

14. LSST’s angular resolution is about 0.5 arcsec …15 more lines 60,000 pixels

15. 15 Unique science goals drive sensor design Large field of view implies physically large focal plane (64cm  ) Modular mosaic focal plane construction 21 rafts, 9 (4K) 2 CCDs/raft 189 CCDs total 3.1Gpix Fast f/1.2 beam, shallow depth of focus Tight alignment and flatness tolerance Flatness: 5  m Alignment (z axis): 10  m Plate scale 20”/mmSmall pixels, close butting Pixel: 10  m Chip-chip gap: 250  m Fast readout (2s) with low noise (5 e - ) Highly parallel readout electronics 16 amplifiers/(4K) 2 CCD Broadband, high spectral sensitivity Thick silicon sensor, back illuminated, AR coat 100  m thickness for IR sensitivity Thin conductive window Seeing-limited image quality Internal electric field to minimize diffusion High resistivity, biased silicon (> 3 k  -cm, -50V)

16. LSST CD-1 Review SLAC, Menlo Park, CA November 1 - 3, 201116 LSST science demands a camera sensor that pushes the state of the art in imaging technology Broadband wavelength coverage with a single detector type –Sensitivity from 320 – 1050 nm Flat imaging surface in f/1.23 beam –Shallow depth of focus (13.5  m p-v within a raft) Fewest possible wasted photons –90% packing fraction mosaic focal plane –80% of time spent exposing PSF size and shape control –<10% degradation due to sensors Read noise < darkest sky background shot noise –7 e- rms Manufacturability –Largest number of CCDs in any astronomical camera –Reproducible characteristics, no per-device tuning

17. 17 Requirements Quantum Efficiency Image Quality –Image size and ellipticity Charge diffusion Charge transfer inefficiency Beam divergence in silicon Non-flatness and z-height variation –Full well capacity –Residual image –Dark current and cosmetic defects Electrical Requirements –Read noise and readout time –Linearity –Crosstalk –Power dissipation Fill factor and pixel size

18. LSST CD-1 Review SLAC, Menlo Park, CA November 1 - 3, 201118 Summary of the SRFT requirements to which sensors contribute ParameterMinMaxunit QE u41% QE g78% QE r88% QE i81% QE z75% QE y414% Pixel size10 mm Diffusion5.5  m rms Read time2sec Read noise/pixel/exposure7e- rms Video crosstalk.05% Gain stability.05% Nonplanarity13.5  m p-v in raft Inactive area8% Complete set of SRFT requirements captured in controlled document LCA-57 sensors sensors + electronics sensors + RSA

19. 19 Noise LSST signal-to-noise ratio should be limited by sky background: The dominant noise contribution at SNR=5 is the sky in all bands except the u-band.

20. 20 (a)Estimated read noise for a 4K x 4K sensor constrained to 2s readout time. Noise estimates based on CCD output transistor properties with three values of sense node capacitance. (b) Proposed layout of a 16-fold segmented, 4K x 4K CCD for fast, low-noise readout. Segmentation vs. Read Noise

21. 21 Additional electrical requirements Linearity 3% for signals from 100e - to full well limit Driven by requirements on photometric accuracy Crosstalk 10 -4 from all electrical effects (sensor + readout electronics) Power dissipation –Sensors dissipate power in their source-follower amplifiers and while clocks are being driven. –Target power dissipation of same order as IR head load from L3 lens + warm cryostat walls. –Approx. 0.4W per 4Kx4K CCD

22. 22 Technology selection Sensor technologies considered: –Monolithic and hybrid CMOS –CCD CMOS technologies promising due to: –low power –integrated electronics, flexible addressing modes –No CTI –“electronic shutter” Monolithic CMOS difficult to reach high sensitivity and NIR response Hybrid p-i-n/CMOS limited use in astronomy to date: –Noise higher than CCD at same frame rate –Significant number of isolated dark pixels (bump bond failures) –Dark current density relatively high –Interpixel capacitance induces undesired correlations –Problems with oversaturated illumination: Residual image at the level of 8e-/pix/s at 120K, persists for hours after 100X full well illumination Permanent threshold shift of ROIC input transistor Although hybrid and monolithic CMOS would simplify some engineering aspects of LSST’s large-area focal plane, concerns about performance, technological maturity, and high cost led us to choose CCDs as the baseline technology for the science, wavefront sensing, and guiding arrays.

23. 23 Mechanical design Non-imaging periphery of a FDCCD has to include space for amplifiers, electrical busses, and field-terminating structures (guard rings). –Guard ring area > 2.5X chip thickness to keep lateral fields in substrate negligibly small. 4K x 4K, 10  m pixel CCD –Require 4-side buttable package on 42.5mm pitch. Imaging surface flatness, package tip,tilt, and piston controlled to maintain focus: –Require imaging surface of all chips 13.000 .005  m above baseplate. 0.8W heat removed through package frame and mounting points to the raft: –Require thermal impedance ≤ 5°C/W.

24. SENSORS: IMPLEMENTATION

25. 25 These design choices lead to the LSST sensor reference design 100  m-thick, high resistivity silicon CCDs, fully-depleted with transparent conductive window – for broadband QE and small PSF 4K x 4K format with 16-fold parallel output –for fast, low noise readout 10  m pixels –for optimum sampling at LSST plate scale Buttable, >92% fill factor packaging –for minimum inactive area Flatness and alignment tolerance to bring image surface to 13.000 ±0.0065mm from baseplate –for use in fast f/1.2 beam Mounting and alignment feature geometry specified –ensure mechanical compatibility between vendors Vendors allowed to develop implementation details (window formation, amplifier design, package, electrical interface, etc.)

26. 4K x 4K science sensor Functional characteristics 10 um pixels 16 segments each 2k x 512 16 readouts per CCD 100um thick silicon, back- illuminated Fully depleted Operated at -100C Key requirements: Broadband spectral response Fast parallel readout 500 kpix/s  2s Flat imaging surface 5um peak-valley Small PSF 5.5um rms Low read noise 5 e- rms

27. 27 wirebonds frontside bond pads bump bonds multilayer ceramic frame alignment pins backthinned CCD (a) (b) Buttable Package Construction

28. 28 RSA fill factors Gap between outermost imaging pixels on adjacent chips 0.75 – 1.8mm (15 – 37 arcsec)

29. RAFT TOWER MODULE

30. 12K x 12K science raft tower

31. Components of the science Raft Tower Module (RTM) 3x3 array of CCDs on flat SiC baseplate -100C Clock and bias buffering Analog signal processing Differential line drivers/receivers -100C Video digitizing Data serializer Monitoring and diagnostics Slow controls -40C Raft-Sensor Assembly (RSA) Raft Control Crate (RCC) Front End Cage (FEC) power, control, cooling pixel data

32. LSST 2009 Camera Meeting – Raft Tower Modules32 Deliverables List 21 Science Raft Tower Modules (RTMs), each of which consists of: –Nine 4Kx4K science CCDs –Raft baseplate –Front-End Electronics (FEE) Three FEE boards Type A Three FEE boards Type B FEE board supports FEE enclosure –Raft thermal management components Thermal planes to cryoplate Raft thermal straps Raft temp sensors Raft heaters –Raft mechanical hold-down components –Raft conductance barrier –Shipping box –Performance report Spare raft tower modules –Quantity TBD Documentation –Operating manual –System safety information (thermal/electrical limits) –Installation/repair/rework procedures

33. CONSTRUCTION PLAN

34. LSST 2009 Camera Meeting – Raft Tower Modules34 Integration flow in production CCD raft Electrical/ optical test Metrology Assemble and align (warm) Data base raft CCD Integrate electronics test (warm) Cold acceptance test; ship to SLAC Data base (225 units) (25) (180) FEB ASICs (1200) purchased component purchased component LPNHE, Paris U. Penn Harvard (30) RSA BEE RTM

35. LSST CD-1 Review SLAC, Menlo Park, CA November 1 - 3, 201135 We have put in place a phased prototype development program with external vendors Phase 1 Technology development Phase 2 Full spec prototypes 2a – pkg 2b – operable Phase 3 Extended mfg. demo First articles CD1CD3aCD2 3 vendors 1 vendor Production CD0 39 mo A more detailed schedule will be shown in the subsystem section of the talk 2 vendors Add 3 rd vendor? Teledyne e2v ITL/STA e2v ITL/STA 2 vendors

36. RESULTS

37. LSST CD-1 Review SLAC, Menlo Park, CA November 1 - 3, 201137 CCD prototypes from 2 vendors e2v CCD250 (operable) STA3800 (mechanical sample) (cold probe, Aug. 24)

38. LSST 2009 Camera Meeting – Raft Tower Modules38 CCD250 measurements at BNL 37-wire interface, side A CCD250 handling jig shorting plug connector saver 6”CF w/triple 150-pin feedthrough

39. LSST 2009 Camera Meeting – Raft Tower Modules39 Two CCD250 buttable packages on raft (e2v)

40. 40 Package mechanical samples on silicon carbide raft prototype

41. LSST 2009 Camera Meeting – Raft Tower Modules41 Initial assembly of mechanical samples to raft 13.5  m window Expect improvement with flatter sensors torque and spring control