SPECTROMETER DETECTORS From science requirements to data storage

1
SPECTROMETER DETECTORSFrom science requirements
to data storage
2
SUMMARY

1. Basic requirements for science
2. NIR Baseline
3. NIR Dark Current
4. NIR and SN Cosmics, Readout, memory size and
   reduction
5. NIR and Galaxies Readout, memory size and
   reduction
6. CCD Science, Baseline, requirement and memory
   size
7. Summary
8. Addressed questions

3
SCIENCES GOALS

• Results shown during science talks concerning the
  SN and WL for the spectrometer have been obtained
  doing some assumptions on the detectors.

CCD NIR
array size (Mpxl²) array size (Mpxl²) 3,5x3,5 2x2
pxl size (µm) pxl size (µm) 10,5 18
number of detectors number of detectors 2x½ 2x½
T (K) T (K) 140 110
RN (e-) RN (e-) 2 5-7
DC (e/pxl/sec) DC (e/pxl/sec) 0,0003 0,002
depend of the integration time. see later
4
Spectro NIR Baseline
NIR
array size (Mpxl²) array size (Mpxl²) 2x2
pxl size (µm) pxl size (µm) 18
number of detectors number of detectors 2x½
T (K) T (K) 110
RN (e-) RN (e-) 5-7
DC (e/pxl/sec) DC (e/pxl/sec) 0,002
l
Minima requirements used for the science
simulations
spatial
Readout modes and memory size have to be thinked
for each science
depend of the integration time. see later
5
Spectro NIR / Dark Current

• From dark current Imager requirement (0.02
  e/pxl/sec.) to spectro requirement (0.002
  e/pxl/sec.)
• Achievable down to 130K
• T requirement 110K

• Remarks/questions
• have to be tested on more devices
• does the CCD work at this temperature
• if not it means that two temperature are
  necessary on the spectro focal plane
• 2 structures
• insulation
• not the same thermal straping for CCD and NIR
• heater?

R.Smith M.Bonati,2006-01-07
6
Spectro NIR and SuperNovae

• Because of the readout noise limitation, science
  plots have shown clearly that you need long
  integration time and our baseline for the spectro
  have to be 3000sec.

• how to manage such a long time with cosmics rays?
• how to reach readout noises at the level of 7e-
  per exposure?

7
Spectro NIR and SN / Cosmics

• With18µm pixels calculations show a
  rate1,310-3/s/pxl
• ? 1 cosmic/800s/pxl (TBC)
• The strategy to reject cosmics is to use an
  up-the-ramp readout

You have to know the previous slope good enough
to reject the cosmic
? choose the appropriate readout cadence (10-50
TBD)
Some calculations have already shown that the
loss of S/N using up-the-ramp and cosmic
rejection is affected of only -1.6 for 3000sec
exposure (TBC)
(S/N no cosmicS/N cosmic )/ S/N no cosmic
8
Spectro NIR and SN / Readout Noise
Actual measurements (our baseline) show that
being read noise limited larger integration time
are necessary but restriction on readout noise
have to be under control With 3000 secondes of
exposure time Fowler 100-500 (TBC) are necessary
to limit the readout noise
9
Spectro NIR and SN / Readout strategy

• Need of 3000 secondes exposures
• Need of a Up-the-Ramp (cadence 10-50 sec.)
• Need of Fowler 100-500

signal
This ideal readout mode HAVE TO BE TESTED WITH
REAL DATA
3000sec
Reset
N Fowler Samples stored
N Fowler Samples stored
Slow Up-The-Ramp stored maximum clocking but
not stored
10
Spectro NIR and SN / memory size

• hypothesys 8h/day divided in 3000sec. exposure
  read with Fowler 300
• 2 X ½ NIR 2x2 (Mpxl²) x 2 bytes 8 Mbytes
• 8 Mbytes/frame x 3h/day / 3000 sec/expos x
  600frames
• 17 Gbyt/day
• to be compared to the 63Gbyt/day NIR imager
  without data reduction

11
Spectro NIR and SN / memory size reduction

• To conclude on that point
• check the hypothesys of that 17GByt/day
• possibilities to reduce by a factor of 4 (at
  least) this volume

Use lossy compression reduction by a factor up
to 5!!!
12
Spectro NIR and Galaxies

• science inputs
• readout noise7 e DC0.002 e/pxl/sec
• Integration time is drived by the imager
• 4 x 300sec exposure with 3.5 pixels dithering

300sec. integration time R.Smith curves show
that a Fowler 30 reach the 7 e goal
14 electrons noise in 1200s
13
Spectro NIR and WL / data storage

• with 300sec. integration time
• 220exp/day x 60frame/exp x 8Mbyt/frame
  106Gbyts/day
• with 1200sec. integration time
• 55exp/day x 200frame/exp x 8Mbyt/frame
  88Gbyts/day

• Other possible reduction of volume
• depending of the dynamic range of the galaxies
  one can code on 8bits instead of 16
• check the dynamic range (e/ADCU) versus the
  number of g of galaxies (TBD)
• use lossy compression factor 5 achievable?

14
Spectro CCD Baseline

• what is new for the spectro CCD?
• use of the same than the imager
• LBNL 3.5x3.5 Mpxl² 10.5µm each same thickness

• Due to the thickness we have the same cosmic
  limitation on the integration time than for the
  imager 300sec whatever the science!

• In terms of readout noise requirement we plan to
  have 2e- and this is achievable changing the
  readout frequency from 100kpxl/sec (imager) to
  50kpxl/sec.

• In terms of Dark Current a requirement of
• 0.0003 e-/pxl/sec (1 e/pxl/hour) is needed
• ? questions is it feasable at 140K? does it
  work in space?

15
Spectro CCD requirement summary and impact on
data storage
CCD
array size (Mpxl²) array size (Mpxl²) 3,5x3,5
pxl size (µm) pxl size (µm) 10,5
number of detectors number of detectors 2x½
T (K) T (K) 140
F (kHz) F (kHz) 50
RN (e-) RN (e-) 2
DC (e/pxl/sec) DC (e/pxl/sec) 0,0003

• DATA STORAGE
• 3.5 x 3.5 (Mpxl²) x 2 (Byt) 24.5 MByt/expos.
• SN
• 24.5 MByt /exposure x 36 exposures (3h/day)
• 0.9 GByt /day without compression
• 0.45 of the CCD imager
• ?Galaxies
• 24.5 MByt/exposure x 220 exposures (100day)
• 5.4 GByt /day without compression
• 2.4 of the CCD imager

16
Spectro CCD summary
imager spectro
CCD array size (Mpxl²) array size (Mpxl²) 3,5x3,5 3,5x3,5
CCD pxl size (µm) pxl size (µm) 10,5 10,5
CCD number of detectors number of detectors 36 2x½
CCD T (K) T (K) 140 140
CCD F (kHz) F (kHz) 100 50
CCD RN (e-) RN (e-) 6 2
CCD DC (e/pxl/sec) DC (e/pxl/sec) 0,03 0,0003
CCD Readout time (sec) Readout time (sec) 30 30
CCD SN integrated time 300 300
CCD SN exposures per day 220 (100 day) 36 (3h/day)
CCD SN memory size (Gbyte) 194 0,9
CCD WL integrated time 300 300
CCD WL exposures per day 220 (100 day) 220 (100 day)
CCD WL memory size (Gbyte) 194 5,4
without compression
17
Spectro NIR summary
imageur spectro spectro
NIR array size (Mpxl²) array size (Mpxl²) 2x2 2x2 2x2
NIR pxl size (µm) pxl size (µm) 18 18 18
NIR number of detectors number of detectors 36 2x½ 2x½
NIR T (K) T (K) 140 110 110
NIR RN (e-) RN (e-) 9 7 7
NIR DC (e/pxl/sec) DC (e/pxl/sec) 0,02 0,002 0,002
NIR Readout time (sec) Readout time (sec) 1,7 1,7 1,7
NIR SN integrated time 300 3000 3000
NIR SN exposures per day 220 (100 day) 5 (3h/day) 5 (3h/day)
NIR SN Read out mode fowler 16 coads on boards Fowler300Up-the-Ramp50sec Fowler300Up-the-Ramp50sec
NIR SN memory size (Gbyte) 63 17 17
NIR WL integrated time 300 300 1200
NIR WL exposure per day 220 (100 day) 220 (100 day) 60 (100 day)
NIR WL Read out mode FowlerN coads Fowler30 coads Fowler100Up-th-Ramp50
NIR WL memory size (Gbyte) 63 1,8 88
without compression
18
Addressed questions

• stable assumptions inputs from science (TBC)
• CCD DC _at_ 140K TBC
• CCD DC OK in space?
• cosmic flux and inpact on NIR TBC
• Up-the-ramp readout cadence for cosmic rejetion
  have TBC
• Fowler N (100-500) with 3000 sec. TBC with other
  devices
• FowlerNUp-the-ramppermanent clocking have to be
  really tested
• with many devices
• _at_ different T (110-140K)
• NIR DC verify with different devices
• CCD works down to which T (is 110K OK?)
• dynamic range have to be confirmed
• for each science
• for darks
• for stars calibration
• dynamic range versus g flux have to be confirmed
  for galaxies distributions
• data storage size and reduction have to be
  studied
• for galaxies should we move from 300sec up to
  1200sec ?
• does the CCD readout have effect on NIR
  integration/readout performance?

hard tests. Could be done both in US and FR?
19
Spectro CCD and NIR summary
CCD NIR NIR
array size (Mpxl²) array size (Mpxl²) 3,5x3,5 2x2 2x2
pxl size (µm) pxl size (µm) 10,5 18 18
number of detectors number of detectors 2x½ 2x½ 2x½
T (K) T (K) 140 110 110
F (kHz) F (kHz) 50 TBD  TBD 
RN (e-) RN (e-) 2 7 7
DC (e/pxl/sec) DC (e/pxl/sec) 0,0003 0,002 0,002
Readout time (sec) Readout time (sec) 30 1,7 1,7
SN integrated time 300 3000 3000
SN exposures per day 36 (3h/day) 5 (3h/day) 5 (3h/day)
SN Read out mode   Fowler300Up-the-Ramp50sec Fowler300Up-the-Ramp50sec
SN memory size (Gbyte) 0,9 17 17
WL integrated time 300 300 1200
WL exposure per day 220 (100 day) 220 (100 day) 60 (100 day)
WL Read out mode   Fowler30 coads Fowler100Up-th-Ramp50
WL memory size (Gbyte) 5,4 1,8 88