The Eddington Photometric Camera Working Group Eddington System Studies WG meeting ESA - HQ November 20 th, 2002 revised on Nov. 28th CAB W G

Eddington SWG progress meetingESA-HQ 20th November Contents Scientific Requirements analysis Instrument configuration and operation Example

Eddington SWG progress meetingESA-HQ 20th November Scientific Requirements analysis

Eddington SWG progress meetingESA-HQ 20th November Latest version  “Eddington High level Science Requirements” Claude Catala and the EST July 2002 General comments:  There are some key requirements, which affect technical definition and are missing: ­Maximum allowable defocusing (to avoid crowding) for AS and PF ­Number of stars to be monitored for AS (could be derived from “Typical star densities for the Eddington mission” (Claude Catala; September 2002) ­Color discrimination still TBC (?) for AS and PF  Photometric requirements should be clarified for the complete magnitude range and translated to directly measurable engineering parameters Scientific Requirements analysis: general

Eddington SWG progress meetingESA-HQ 20th November Requirement: ­AS: (goal 3) ­PF:17-11 General considerations:  Bright stars produce saturation in the CCD  Weak stars can not reach the required photometric accuracy INTA studies based on:  CCD E2V 42-C0: 3072x2048 pixels; full well capacity of e-  Astrium preliminary design  Defocusing: star box size 16x16 pixels (170 microns on the focal plane) Scientific Requirements analysis: magnitude range

Eddington SWG progress meetingESA-HQ 20th November Integration time for saturation: Bright stars saturation problem Star Magnitude Integration time for saturation (sec) 50,17 60,50 71,25 83,25 97, , For Ti=100 µs ; Tr=1 µs = 1MHz 1x1 => 3072(300 µs µs) = 7,366 sec. 2x1 (or 2x2)=> 1536(400 µs µs) = 3,837 sec. 4x1 => 768(600 µs µs) = 2,072 sec. For Ti=50 µs ; Tr=500 ns = 2MHz 1x1 => 3072(150 µs µs) = 3,683 sec. 2x1 (or 2x2) => 1536(200 µs µs) = 1,918 sec. 4x1 => 768(300 µs µs) = 1,036 sec. EddiSim Data CCD 42-C0 Data Sheet Readout time for different binnings

Eddington SWG progress meetingESA-HQ 20th November AS PF Readout time

Eddington SWG progress meetingESA-HQ 20th November PF:  No saturation problem within the required magnitude range  Potential saturation problems induced by bright stars present in the FOV (magnitude below V=11) Bright stars saturation problem

Eddington SWG progress meetingESA-HQ 20th November AS:  Scientific requirements are not accomplished. Different options can be considered: 1.Larger defocusing 2.On chip binning: reduction of integration time 3.Smaller effective FOV: reduction of integration time 4.Use of two readout ports simultaneously 5.Use of integration times shorter than readout time 6.Different operations on one telescope, optimized for bright stars 7.Combination of some of the above options Bright stars saturation range

Eddington SWG progress meetingESA-HQ 20th November Larger defocusing:  Maximum defocusing is constrained by the expected crowding Bright stars saturation problem: options Defocusing (microns) PSF box V5V6V x160,250,501, x170,250,751, x180,500,752, x200,501,002, x220,501,253, x240,751,503, x250,751,754, x260,752,004, x271,002,255,50 Integration time for saturation (sec) EddiSim Data PROS: Longer integrations could be used without saturation  CONS: Crowding Photometry accuracy  due to larger background contribution  needs to be analysed

Eddington SWG progress meetingESA-HQ 20th November

Eddington SWG progress meetingESA-HQ 20th November Binning alternatives: 4x1 Bright stars saturation problem: options PROS: Readout time is reduced to 2 sec (1MHz) or 1sec (2MHz) Data volume is reduced by 4  CONS: If working with CCD readout speed of 2MHz the electronic chain has to work at 2 MHz Very poor PSF spatial sampling Photometric accuracy  due to higher background contribution 2x2 PROS: Data volume is reduced by 4 PSF better sampled with 2x2 binning It would allow to use a readout of 2MHz for the CCD and 1MHz for the electronic chain  CONS: Readout time is reduced to only 3.8 sec (1MHz) or 2 sec (2MHz)

Eddington SWG progress meetingESA-HQ 20th November Binning alternatives: Bright stars saturation problem: options EddiSim Data Input image CCD output 1x1 binning CCD output 2x2 binning CCD output 4x1 binning

Eddington SWG progress meetingESA-HQ 20th November Smaller FOV (windowed readout): Bright stars saturation problem: options PROS: Readout time is reduced Data volume is reduced (smaller processing requirementS)  CONS: FOV is reduced by a factor 4 (number of stars is reduced) Strong constraints on the readout port; loss of redundancy 1650 pixels 625 pixels 2048 pixels Claude Catala Proposal Image area of 10,89 Mpixels instead of 37,8

Eddington SWG progress meetingESA-HQ 20th November Use of two readout ports simultaneously: Bright stars saturation problem: options PROS: Readout time is reduced  CONS: Duplicated readout chain Loss of redundancy 5. Use of integration times shorter than readout time: PROS: Integration time could be adapted to the required value  CONS: Gaps between integrations required to read the image and “clean” the CCD; effective observation time is reduced Photometry accuracy for weak stars could not be acceptable due to the loss of effective integration time 6. Optimization of one telescope for bright stars: PROS: Defocusing could be adapted for bright stars With a filter the integration time for saturation could be longer  CONS: If a filter is installed, redundancy between telescopes is lost Photometry accuracy for weak stars will be worse

Eddington SWG progress meetingESA-HQ 20th November Combination of some/all of the above options: There are lot of possibilities It is recommended that the operational solution:  Does not reduce redundancy  Maintains the same HW configuration for the four telescopes; differences should be only in the operation Example:  4 identical telescopes  3 of them with operations optimized for weak stars: Bright stars saturation problem: options DefocusingBinningIntegration time 16x162x22sec, continuously  1 of them optimized for bright stars (but also observing weak stars): DefocusingBinningIntegration time 18x182x20,5 sec integration + 3,5 sec integration efficiency: 66%

Eddington SWG progress meetingESA-HQ 20th November Requirement: ­AS: Noise level in amplitude Fourier space  1.5ppm in 30d for m v = 11 in frequency range mHz ­PF: noise level in the light curve  1e-5 in 39 hrs (average 3 transits) = 6.3e-5 in 1hr for late-type dwarfs Both requirements should be translated into measurable instrument parameters and should be expressed for the whole magnitude range. Scientific Requirements analysis: photometric requirements

Eddington SWG progress meetingESA-HQ 20th November We have assumed the following definition: SNR -1 telescope = Noise/Signal =  / signal For one single measurement with the telescope the accuracy of this single measurement is given by: S measured   SNR -1 instrument = SNR -1 telescope /  Number of telescopes = SNR -1 telescope / 2 Scientific Requirements analysis: photometric requirements

Eddington SWG progress meetingESA-HQ 20th November How is SNR -1 instrument calculated?:  Directly considering only photon noise: SNR -1 telescope = Noise/Signal = 1 /  counts per telescope SNR -1 instrument = 1 / 2  counts per telescope  Using EddiSim: Scientific Requirements analysis: photometric requirements EddiSim (1 telescope) S (for a given star magnitude and type) Noise (distributions of photon, readout, background, etc.) N times = N samples of S * (N around 400) Calculation of:  *  S *  SNR -1 telescope =  * /  S *  SNR -1 instrument = SNR -1 telescope /2

Eddington SWG progress meetingESA-HQ 20th November Results considering only photon noise: Scientific Requirements analysis: photometric requirements Star magnitude 1 telescope photons/s 1 telescope counts/s 4 telescopes counts/s 101,0E+66,8E+52,7E+6 114,0E+52,7E+51,1E+6 142,5E+41,7E+46,8E+4 151,0E+46,8E+32,7E+4 164,0E+32,7E+31,1E+4

Eddington SWG progress meetingESA-HQ 20th November V 5

Eddington SWG progress meetingESA-HQ 20th November Results using EddiSim: Scientific Requirements analysis: photometric requirements V11V16V11V16 54,29E-44,30E-34,22E-44,41E ,03E-43,04E-32,82E-42,72E-3 301,75E-41,76E-31,56E-41,49E ,59E-59,62E-48,18E-58,40E ,91E-53,93E-43,53E-53,63E ,03E-53,04E-42,57E-52,51E ,59E-69,62E-58,59E-69,24E-5 SNR -1 instrument Direct calculation SNR -1 instrument Using EddiSim Acumulated integration time (sec) EddiSim data have been obtained: considering only one integration and not taking into account the saturation for a PSF box of 16x16 pixels

Eddington SWG progress meetingESA-HQ 20th November

Eddington SWG progress meetingESA-HQ 20th November Photometric accuracy could be improved by:  Increasing the telescope aperture (worsening of the saturation problem)  Implementing more telescopes (not realistic)  Increasing the accumulated integration time (longer sampling time, still compatible with the detection of transits) Scientific Requirements analysis: photometric requirements

Eddington SWG progress meetingESA-HQ 20th November Requirements:  PF : ­Time sampling:  600 sec (bottomline)  30 sec (goal) ­Number of stars to monitor> late-type dwarfs with PF1 S/N > all types with lower S/N  Assumed by INTA studies Scientific Requirements analysis: number of stars and sampling time Sampling time (sec)Total Number of stars

Eddington SWG progress meetingESA-HQ 20th November Requirement:  AS : ­Time sampling:  30 sec (baseline)  5 sec (for some targets) ­Number of stars to monitornot included; estimation could be done with “Typical star densities for the Eddington mission” – Claude Catala, Sep.02  Assumed by INTA studies Scientific Requirements analysis: number of stars and sampling time Sampling time (sec)Total Number of stars 5 

Eddington SWG progress meetingESA-HQ 20th November Instrument configuration and operation

Eddington SWG progress meetingESA-HQ 20th November General comments  In order to start the instrument definition and preliminary sizing it is necessary to establish an instrument configuration and operation baseline for both science modes: AS and PF  The parameters that should be set are the following: Instrument configuration and operation: Telescope operation Identical or different operation Defocusing Integration time Number of stacking areas Image area size Binning CCD readout frequency + readout port (1 or 2) Sampling time Number of stars to be monitored Telescope Configuration Identical or different (filter for example)) Number of CCDs per telescope CCDs type and characteristics

Eddington SWG progress meetingESA-HQ 20th November How do these parameters affect the instrument definition and sizing? Some examples:  Defocusing/binning: determine the number of pixels in which the information is contained  number of pixels to be processed required processing capability  Image area and binning: affects directly the required onboard memory  Number of stars: gives the number of photometric points to be processed required processing capability  Sampling time: it constraints the time in which the processing has to be done required processing capability Instrument configuration and operation:

Eddington SWG progress meetingESA-HQ 20th November In addition, the scientific proocessing algorithm has to be defined to dimension the instrument. Instrument configuration and operation:

Eddington SWG progress meetingESA-HQ 20th November Example of instrument dimensioning

Eddington SWG progress meetingESA-HQ 20th November Instrument configuration and operation baseline: Example Telescope Configuration Identical telescopes 6 CCDS per telescope CCDs 42-C0 type (3072 x 2048 pixels in the image area) Telescope operation Identical operation Defocusing 16 x 16 pixels (170 microns) Integration time = 2 sec Number of stacking areas = 2 Image area size = 3072 x 2048 pixels Binning = 2x2 CCD readout frequency: 2MHz 1 readout port readout time = 1.9 sec Sampling times: AS: s PF: s Number of stars: AS: 120 (6s) (30s) PF: (30s) (600s)

Eddington SWG progress meetingESA-HQ 20th November Example: instrument data flow configuration CCD ADC Binning 2x2 1 read-out port Integration time 2 sec 16 bits ADC per binned pixel (availability TBC option suggested by MSSL with 2 x12 bits ADCs) 3 Bytes per binned pixel 4 Bytes per binned pixel Spacewire bus (100Mbits/s) 1.5 MBytes/s Adder 1 Adder 2 DPU Intermediate buffer  9MBytes Pre-processor STACK 1 6s/30s 4.5MBytes STACK 2 30s/600s 4.5MBytes Image area Storage area 6.3 Mpix 3.15 Mpix 2 MHz Mpix/2sec 1 MHz Output Register Readout time 1.9 sec  2sec Readout Amplifier 1 MHz 4.5 Mbytes per CCD image 3 Mbytes per CCD image 6 MBytes per CCD image

Eddington SWG progress meetingESA-HQ 20th November Example: DPU configuration Based on the design developed by CRISA for PACS on Herschel Constituted by: + CCD I/F interface module, based on SMCS332 Spacewire links at 100 Mbps + scientific processing unit, based on 1 TSC21020E processor at 20 MHz + extended memory boards + instrument control unit, with an independent processor + OBDH I/F module based on the 1553B bus at 100 kbps + monitoring, synchronization and power supply modules  This DPU is already being built and is fully compatible with the Herschel bus

Eddington SWG progress meetingESA-HQ 20th November At the beginning of each observing period (once per month), a reference image (binning 1x1) is obtained by combining different integrations during around 1 hour. The reference image is downloaded to ground using the highest available TM (10 minutes per CCD at 300 kbps without compression). The reference image is processed on ground, obtaining the reference photometric value for each star of interest. A table containing the identification of the stars to be monitored, as well as several bits indicating the kind of processing to be performed, is uplinked to the spacecraft. The table will include also the photometric mask to be used for each star: Example: scientific processing strategy

Eddington SWG progress meetingESA-HQ 20th November The photometric mask contains 1 bit per position (64 bits for 8x8 PSF box). Depending on the bit information, the corresponding binned pixel will be added or rejected. The masks will allow to minimize the impact of overlapping stars, CCD edges, defect pixels or columns,... They will be obtained on ground from the reference images, in order to optimize the results. Example: scientific processing strategy

Eddington SWG progress meetingESA-HQ 20th November The DPU will add only the pixels marked with 1 in each PSF box. The value so obtained will be subtracted from the reference value, computed on ground from the reference image with the same algorithm: the reference background is computed on the same pixels than the star itself!. This difference will be sent to ground with 4 bytes per value. The values will be mostly zero or very small numbers, allowing for a high degree of compression. Example: scientific processing strategy

Eddington SWG progress meetingESA-HQ 20th November In addition, a TBD number of complete windows (8x8 pixels) will be sent to ground to monitor the evolution of the background and the health of the CCDs. The real photometric value will be reconstructed on ground. Cross-correlation of the 4 photometric series on ground will allow to discard the effect of cosmic rays. Computations with the EddiCam simulator show that for V < 16 the images stacked up to 600 s effective integration time (300 frames) remain photon noise limited. Example: scientific processing strategy

Eddington SWG progress meetingESA-HQ 20th November The preliminary estimated TM requirements are the following (assuming all values sent to Earth with 3 bytes coding): AS: kbps for stars ( kbps for 100 background windows) every 30 s: ( x120) stars/telescope + (6x100) 8x8 windows (33.000x3)x4 telescopes + (600x64x3) = bytes PF: 81.9 kbps for stars (+ 5.4 kbps for 100 background windows) every 600 s: ( (20x20.000)) stars/teles + (21x100) bkg windows ( x3)x4 telescopes + (2.100x64x3) = bytes  Well within the Herschel TM capabilities (  100 kbps sustained rate), assuming some moderate data compression! Example: scientific processing strategy

Eddington SWG progress meetingESA-HQ 20th November Processing analysis tool support:  DEIMOS Space S.L. is supporting INTA with the processing requirements dimensioning  Emulations of Eddington image processing are being done using TSIM Professional host simulation tools Processors under study:  ERC32 (TSC695E) with 32 Mbytes RAM, at 20 MHz AS:  22 % CPU load PF:  26 % CPU load  A single DPU can handle the 6 CCDs of each telescope  TSC at MHz under evaluation, but similar results expected Example: system simulations

Eddington SWG progress meetingESA-HQ 20th November Conclusions Not all the present scientific requirements can be accomplished simultaneously with the present Eddington mission concept  Major problems with saturation vs crowding vs large dynamical range But feasible instrument configurations would allow to comply with most of the requirements The fine tuning of the present designs requires the agreement on which scientific drivers should be optimized: a task for the EST

Eddington SWG progress meetingESA-HQ 20th November OMC first light 5ºx5º image (limit magnitude  13) Each star spreads over around 50”, similar to a 16x16 pixels PSF on Eddington