Notes from SNAP Collaboration Meeting Feb 9, 2002 Michael Richmond 9:00 Perlmutter: Welcome 9:02 Levi: The goal of the meeting speakers should submit electronic version of their talks visitors: ?? FNAL Blucher U Chicago Adam Riess Ray Kutina STScI note: this is first collaboration meeting, may get fiesty review is coming up need to come together on the goal of the experiment not everything we hear today is ready for Prime Time Lehman reviews pushing to get Lehman here for a review no way we can pass such a review yet will take place June 25-27, 2002 will find problems considers planning and science responded to comments from previous review? are there any new issues which have arisen? do we all agree on the baseline SNAP? see agenda working lunch Saturday party tonight at Levi's house will hear about recent HESSI launch informal reception here Sunday Monday working lunch for working group coordinators Tuesday AM: working group meetings RoadMap for Particle Physics SNAP is on the list! next decision is 2004-5 (into construction) SNAP schedule not on agenda for SageNap (yet) review 2002 CD0 2002/2003 spacecraft 2005-2008 telescope 2005-2008 launch vehicle 2006-2009 NRC review Mar 2002 DOE Lehman review Jun 2002 NASA/SEU release roadmap Sep 2002 CNES review Oct 2002 What is the baseline? must decide this weekend focal plane, orbit, etc. technology readiness need to prove that it can be done Level 5 by CDR (Sep 2003) breadboard test in lab Level 6 by PDR (Sep 2004) prototype in relevant environment What is cost of mission? around $350 million has been estimated by "others" (includes launch) SNAP needs to make up its own estimate Action items from previous review error budget calibration plan what uncertainties in measurements are required cost/benefit of weak lensing computing plan Questions are risk areas known? is mission feasible? are instrument concepts optimal/sufficient? ex: does SNAP need spectrograph? what are science requirements? where can one cut corners? Q: why is SNAP _not_ on agenda for SageNap? A: don't know focus is on neutrinos and underground experiments 9:20 Bebek Fixed Filter Focal Planes for visible and IR detectors why one big focal plane? multiple focal planes too hard to keep in focus simul so put all instruments in the same plane guiding easier big filter wheel unwieldy fixed filters drag field across focal plane, it goes through various filters as it moves requirements S/N > 15 at 2 mag below peak zodiacal light is limiting background so, calc S/N for various sitations visible: CCD QE = 60% dark 0.02 e-/sec/pix read 4 e- include well depth effects, assume 0.4 um Airy disk could be important for nearby SNe z = 0.2 need 4 exp of T = 200 s readnoise, dark not important z = 0.1 and large Texp can saturate central pixels use 4 or 9 pixels in aperture (?) looks good for up to and a bit beyond z = 1.0 at least to 2.5 mag below peak not quite to 3.8 mag below peak check effect of cosmic rays near-IR: HgCdTe QE = 40% dark 0.02 e-/sec/pix read 4 e- (not achieved yet) goes 1100 - 2200 Angstroms (but not final) want S/N = 15 at 2.0 mag below peak need lots of exposures to get desired S/N! check saturation -- not an issue want to avoid adding up more than 4 pixels (in aperture?) large background of galactic light is important need lots of exposures how can we beat it down to match CCD values? if T = 200 s and 4 exposures, can't meet requirements at z < 1.0 how to solve the extra exposure time needed for IR? dedicate more focal plane area (x2) use time dilation: need fewer exposures for high z, can check once per 8 days instead of each 4 days buys another factor of 2 orbit of SNAP is important if get factor 2x2 = 4 extra for IR, then IR still lags spec at discovery level 3.8 mag below peak matches 2.5 mag below peak spec to z=1.4 conclude: z < 1.3: can meet requirement but discovery level 3.8 mag below peak tough for z > 1.0 time dilation effects are important Q: SN scientists and analysts must work hard to define exactly the requirements on observations satellite constraints: focal plane layout assume SNAP field is 1 x 10 degrees position of solar cells and radiator change with orbit 4 90-degree rotations over course of a year causes focal plane to rotate 90 degrees scan over entire strip every 4 days CCD and IR detectors must have integer size relationship if scan in simple way and get exactly same exposure time in each type of detector how many different filters? factors K correction photo z Type Ia/II discrimination Alex Kim says 6 filters in visible, 3 filters IR will work maybe add one more filter in IR occasionally B-band cutoff CCD z=1 V-band cutoff z=1.23 [yeah, yeah] shows cartoon of tiling focal plane with 6/3 filters gives same exposure time in each filter vis, and IR one plan: focal length 21.66 meters donut field of view R inner = 0.3438 deg R outer = 0.7449 deg sky area = 1.37 sq deg approx 0.7 sq deg has detectors all stars fall in all filters observation fields should be organized as linear strips which are one focal plane wide rather than as squares what is yearly SN yield? assumes 2-day orbit want discover and get spectrum in 75% of a year (elliptical orbit causes data dump for several hours a day, no observations) how much time given to spectroscopy is very important! all results below are optimistic -- no calibration time, etc. doesn't take into account a 3-day orbit we'll hear about at this meeting 280 SNe per year per sq. deg. z < 1.7 spectroscopy drives mission time takes 2 hours z=1.0 takes about 5 hours z=1.5 total IR detector 0.34 sq deg total CCD 0.28 sq. deg (more cracks between chips) about 30 SNe per bin z=0.03 per year only to 20 per bin if long spectral exposures up to 50 per bin if short spectral exposures most of them are high-z if pick 50% high-z, 50% low-z, get maybe 40 per bin per year -- so helps a little bit (so ignore some high-z SNe) note that you do discover more SNe, but don't get spectra for them all now, some people worry that can't build so many IR detectors idea: split focal plane visible/IR many ways to do it shows a candidate (some discussion -- good for SNe, not good for other purposes....) 50% of stars fall in CCD+IR, 50% CCD only yields slightly fewer SNe followed up with spectra another candidate plan for split stars may miss some filters again, fewer SNe with followup comparison of candidates symmetric focal plane does best at z > 0.7 candidates with more CCDs do better at z < 0.7 what hasn't been discussed? data buffers size, telemetry observation time cost of near-IR read noise dithering method how many sub-pixel dithers are required? worry about sub-pixel variation in sensitivity comment: 2x2 dither is enough orbit period effects on sicence no telemetry during observations tradeoff photometry/spectroscopy 10:10 end Q: tradeoffs of fixed vs. movable filters? A: want more focal plane go to IR spinning filter wheel makes sense if one type of detector harder when two types of detector also, giant spinning filter wheel strikes some as dangerous Q: can't you do the calculations to show yield with spinning vs. fixed filters? A: haven't done it yet still have to move field of view Q: each pixel has its own filter passband with fixed filters A: yes Q: look at it the other way: how many SNe do you need, and at which redshift, to answer geometry questions? A: (someone else will address later today) Q: what about narrowband filters? could do better job of distinguishing Type Ia/II ramp filters have passband change with position on filter so move SN across filter to get low-res spectra A: that works only if you use them in triggered = targeted mode not ordinary scanning mode A: spectrograph is low-res already, ramp filters not a big win Q: what about objective prism? A: trouble with bright sky unless SN is nearby and bright 10:21: Lampton SNAP telescope satellite comes in many pieces spacecraft, telescope, instrument, etc. must have very clear interfaces between teams on each piece reviewers will look at interface documents payload currently 2.0 m aperture 1.37 sq. deg field optics kept near 290K transverse rear axis side Gigacam location passive detector cooling combines CCDs and IR detectors spectrometer(s) share focal plane few moving parts shutter is included (aha!) optical config annular field prolate ellipsoid primary hyperbolic convex secondary flat oval 45-deg feeds transverse flat focal plane science requirements need S/N = 50 at peak brightness zodiacal light is major noise source geometric blur must be kept below diffraction limit Airy disk at 1 micron = 0.12 arcsec FWHM wavelength 0.4 - 1.7 microns optical design three-mirror astigmat (TMA) is best choice no refracting elements flat field industry can build the design several possible mirror materials Corning ULE glass -- expensive Schott Zerodur - lower cost, used on ground Astrium/Boostec SiC-100: unproven in space carbon-fiber plus resin for structure conventional grind/polish mirror at first finish with ion-beam if desired no show stoppers image quality characterize via Strehl ratio actual vs. perfect peak light concerns: alignment errors in 1-g vs. 0-g gravity release launch-induced distortions aging and creed how many adjustments on-orbit are needed? primary mirror dominates budget: most expensive to figure non-optical figures attitude control system stability transparency and optical depth in Si in practice, usually specify desired Stehl ratio at 633 nm but SNAP looks at longer wavelengths, too must be aware of errors vs. wavelength could cut corners in visible but be OK in IR could save big $$ use geometrical ray trace to check performance for a given design can put most rays within Airy disk, no problem (as long as optics meet design....) diffraction losses as function of spider shapes 4-vane spider has spikes at 10^-3.5 level 8-vane spider has spikes at 10^-3 level spikes still high at larger radii but much stiffer design main R&D issues how good a Strehl do we need? can we afford? need input from science/simulation teams need to understand tolerances figure errors misalignment errors need to identify schedule risks ex: military customer takes precedence perform trade studies stray light issues main mirror is long-lead time issue need to create requirements document what have we done so far? reject off-axis designs reject 5,6 mirror designs reject eccentric pupil designs developed annular field chose warm optics over cold one focal plane or two? currently prefer one focal plane low-CTE metering structure vs. constant-T structure? prefer low-CTE currently in progress build prototype concurrrently with real spacecraft? ULE vs. Zerodur mirror material mass, $$, schedule issues full-aperture vs. sub-aperture test requirements OTA is the critical path item stray light news keep stray light below zodiacal light natural zodiacal = 1 photon/pixel/sec/micron total stray = 0.02 photon/pixel/sec/micron mostly scattered Earthlight ASAP software in place ASAP training achieved preliminary telescope ASAP models being built ASAP illumination environment models being built studying baffle arrangments risks identified primary mirror main worry easy thermal environment few disturbances on orbit OTA is main lead-time concern error budget must be communicated clearly to vendors need plan for contamination control starting plan for stray light control potential vendors have been identified all say "we can do that" have preliminary study on ULE mirror faceplate 7mm thick ribs 3.2 mm thick OTA team: 6 people now but HESSI launch frees up some new people to join goal: write biddable requirements document biggest task is partitioning work into packages 10:57 end Q: how much is ULE blank? A: $5-9 million Q: how well is Strehl ratio related to encircled energy? A: is complicated comment: Strehl = 80% is perfectly OK most critical in blue visible, where SNe are brightest Q: has anyone looked a optical errors across field of view, including diffraction from spiders, etc. A: irrelevant for photometry [ahem. not] comment: problem is scattering from bright objects in the field comment: 4-fold symmetry in spiders better than 3-fold, because field rotates by 90 degrees is complicated to figure out effects properly Q: are you setting limits on amounts of potassium, U, Th in glasses? we have found Schott glass with 10% potassium watch out for radioactivity! A: new to me 11:07 AM end (discussion break starts, return at 11:30 AM) 11:32 Bebek: GigaCAM requirements current design shutter light shield focal plane OTA mount thermal links cable plant local electronics data collection network calibration hardware [aha!] instrument control environmental monitoring all sits behind primary main R&D issues IR detector performance filters on or above the sensors? focal plane layout shutter must develop this -- is new challenge readout electronics interface control documents thermal mechanical electrical shutter rotating bowtie arrangement hole is 6 inches in diameter open/close in 25 msec needs LOTS of design and testing GSFC recommends big program of test/qualification needs about 1 million cycles in mission Q: has anyone considered a plunger-type shutter? A: no sliding parts are bad, rotating parts are good shield thermal stray light blocks particles esp. solar protons current design: aluminum brick about 1-inch thick easily modified sensor integration mechanical thermal focal plane runs at 140 K temperature stability of CCDs isn't big issue temperature stability of IR detectors _may_ be, don't know for sure filters guiders build at LBL, or purchase off the shelf spectrograph light goes through holes in focal plane spectrographs mounted behind focal plane on cold plate unknowns cable/thermal links how many spectrographs? operating temperature cable plant CCDs: 100 40-wire strips IR: 20-40 37-wire cables guiders: dunno spectrographs: dunno all carry low-level analog signals as far as 0.5 meters radiator thermal load looks managable a plate of area 2 sq. meter is sufficient use laminate of hundred(s) of Al sheets need to R&D performance (Michael misses some stuff) electronics analog processing CCDs IR detectors guiders spectrographs will we do attitude control here? data collection instrument control exposure, readout, erase, etc. calibration equipment instrument monitoring power systems for the camera, not the entire satellite calibration hardware don't know what hardware, if any, is needed for sensor calibration [aha!] end 11:51 Q: what tests on temp sensitivity for CCDs? A: dark current very low QE at edges is function of temperature Q: have you considered phonons? A: yes, is part of standard analysis see presentation on Monday don't need to hold CCD temps const to 0.1 deg K Q: near-IR detectors have stronger sensitivity? A: see later presentation reference pixels on IR arrays help Q: how realistic is radiation damage testing of CCDs? space environment is different than ground A: see report on tests Monday have considered protons alpha particles not tested as much Q: where to fold in issues of testing and calibration? A: this is a big system with many pieces is very complicated will require testing facility need discussion of on-board calibration Bobek: only calibration which makes sense in space is through-the-aperture calibration lamps don't make sense comment: lots of calibration comes for free [but not enough, I think] Q: what about annealing? A: not relevant Q: what is long-term damage expectation? A: do know how CTE is likely to change over long-term due to protons Q: what about high-z nuclei events? A: no idea, but they are very rare comment: HST has small change to sensitivity CTE is the thing that changes comment: have done tests on CCDs in proton beam see presentation on Monday 12:02 Tarle: Near-IR camera near-IR sensor development is way behind CCDs 1 year ago, design was a SINGLE array now, covers half of focal plane many details will be covered in talk on Monday this talk: science drivers for IR detectors science objectives for near-IR comparison of near and distant SNe distant SNe have spectra shifted into near-IR so can't use CCDs grey dust to detect grey-ish dust, use largest possible range of wavelengths near-IR adds lots of range to visible SN trigger want to be able to identify SNe early, helps to have galaxy redshifts before SN appears large range of photometric measurements allows one to estimate galaxy redshift can get photo-z from z=0 to z=3 is important in distinguishing Type Ia from Type II SNe Type II generally fainter, will occur in low-z galaxies auxiliary science comes for free, for the most part galaxy evolution and clustering photo-z to 25-50 million galaxies morphological information census of "dropout" galaxies for R, I, z, J uses position of restframe 4000 Angstrom break goes out to z=10 with near-IR QSO luminosity function to z=10 with near-IR, get candidates to z=10 vs. z=6 with visible only study GRBs to high redshift can study afterglows and orphan afterglows not sure how many we'll find probe structure at re-ionization recombination occurs between z=1000 and z=6 find galaxies, QSOs at z > 6 is first step discover coolest nearby objects low-mass T,L stars, brown dwarfs best studied in near-IR lensing studies at z > 1 SNAP high spatial resolution important weak lensing measurements need thousands of sources must be resolved to see alignment discover solar system objects in Kuiper Belt and beyond which are cool, of course identification of targets for NGST esp. in near-IR science requirements FOV = 0.34 sq. deg 0.17 arcsec/pix 2048x2048 H2-RG HgCdTe covers 1.0 - 1.7 micron (this chip does not yet exist) detector temp = 130 K readnoise 5 e- doesn't go exactly as sqrt(N) needs R&D dark 0.05 e-/sec/pix QE > 50% (really? or is this the hope?) is very early in development of these detectors is currently 65% at best, but falls off to short lambda so 50% is a reasonable target Q: will reviewers buy this? A: see presentation later on Monday relative photometry accurate to 1% need to know intra-pixel sensitivity current devices have large variations "we are told" that future devices will be better but don't know for sure studies of devices are behind schedule won't know answer for a while IF intra-pixel sensitivity is large, and IF varies from pixel to pixel, THEN big trouble filters 3 special filters to cover 1.0 - 1.7 microns main R&D issues intra-pixel variations in IR devices more important for SNAP than NGST, others sampling strategy to reduce read noise optimization of planned SNe observations in IR see Gary's talk readout electronics and data volume establish facilities for receiving and verifying near-IR devices other groups claim that one needs several test facilities to get 36 devices in flight NGST plans to have 3 test facilities and that's probably not enough have 215 million pixels, not much more than SNAP we can copy their plans, of course optimize calibration techniques and strategies technical challenges dealing with intra-pixel sensitivity dithering pattern? map entire detector? establish read-out strategy optimal calibration strategy to maintain 1% accuracy develop plan for testing lots and lots of near-IR devices build near-IR team for doing all these tasks currently, our team is too small we are not credible! 12:30 end Q: have you tried to quantify amount of work needed? A: is similar to effort needed for NGST Q: may need to revisit requirements A: yes Q: do we believe that ground-based pre-launch calibration would hold over time in space? A: good point. We don't know yet, must build test facilities to find out Q: the devices you need to test do not yet exist A: very true we will obtain one readout card from Rockwell April 2002 at same time, they will deliver H1RG device to Don Viger [is 1024x1024, I think] [H2RG means 2048x2048, I think] we will have to purchase 3 or so H1RG devices plug into our readout system and test _maybe_ could start testing by Nov 2003 will be tough to meet schedule of Phase I Q: what about 2048x2048 vs. 1024x1024 devices? A: we can't control what Rockwell builds Q: what are prospects of having HgCdTe devices work in visible, too? A: requires high-risk thinning is very expensive Q: what about using different mixtures of materials in sensors? A: too late to consider that Q: CCD group has use 5-micron dot to map out gate response of some CCDs A: rah! ( more discussion on devices Monday afternoon ) 12:40 break for lunch