Title: Star Counts
1Star Counts
- M.Lampton
- Sept 2002
- Updated Sept 2003
2Motivation
- Can SNAP guide itself satisfactorily?
- Are there enough guide stars?
- bright enough for low photon shot noise
- numerous enough so that a reasonable size guider
field is 99.99 certain to get a star
3Two things cause pointing errors...
Dynamics
Disturbances
Coarse star trackers
Wheels
STATE VECTOR s/c attitude
Coarse sun sensors
Controller
Jets
Coarse/fine gyros
Torquers
focal plane guider
Environment orbit, Sun, Earth, stars....
Cassegrain guider
commands data
sensor noise
What is the disturbance torque spectrum? What are
the various sensor noise spectra? What is the
closed-loop response?
4Previous WorkSecroun et al Experimental
Astron. v.122 2001
- Calculated the expected sensor position errors vs
magnitude and integration time - centroiding 2x2, 3x3, 4x4 pixel groups
- Calculated the Poisson statistics for nominal
mean star densities (13ltVlt16) at NEP, SEP - average star densities from C.W.Allen 3rd Edition
- observed star densities from GSC1 Vlt14.5,
extrapolated to V16 - Result gt95 probability of finding a Vlt16 star
provided that the guider field is 200x200 arcsec
or larger - Concerns
- V is not silicon band R is more nearly correct
- need actual counts, not extrapolations or
averages - 95 is NOT good enough
- Aldering region is not NEP but is closer to NGP,
fewer stars!
5Big Picture
6SDSS Early Data Release 462 sqdeghttp//archive.
stsci.edu/sdss/edr_main.html
7Aldering Region in GSC 2.2RA244.0,
dec55.0DeltaRA8.75deg, DeltaDec1.5degon
sky 5.0 deg x 1.5 deg 7.5sqdegGSC 2.2, DPOSS
IIR F bandIIIaFRG610 0.65umhttp//www-gsss.
stsci.edu/support/data_access.htm15512
objectsall non-stars, Kodak objects etc
rejected
8Integral star counts at mid-galactic-latitudesAld
ering Region at (l,b)(85,44)
9References to star counts
BS Bahcall Soneira, Ap.J.Supp. v.55,
67-99 1984 Allen C.W.Allen
"Astrophysical Quantities" Third edition 1973
p.243 Basel Bahcall et al, Ap.J. v.299
p.616-632, 1985 SDSS Newberg Richards
Richmond Fan, "Catalog of four color
photometry... 2002 see also... Chen et
al, ApJ v.553, pp.184-197, 2001 EDD
http//star-www.st-and.ac.uk "EDDINGTON
Cumulative Star Counts" MS O.Yu.Malkov
O.M.Smirnov, "Testing the Galaxy Model with
GSC" ADASS III ASP Conf. v.61 1994. GEMINI
http//www.shef.ac.uk/cgi-bin-cgiwrap/phys/compst
ars.ps Doug Simms Aug 1995 "Longitudinally
Averaged Cumulative Star Counts" GSC2.2
http//www-gsss.stsci.edu/support/data_access.htm
10Analysis of region using box0.05degreesThis is
180x180Slightly smaller than Secrouns
200x200
11100000 random guider locations in Aldering
regionSquare guider box size 0.05, 0.10, 0.15
degHistograms of brightest star within guide box
12What does this mean?at 30 frames/sec...
13...or at 3 frames/sec...
14...or at 10 frames/sec and grasp360nm...
15Typical CCD QE curves
Front illuminated CCDs typical QE 30 typical
BW 400nm typical QEBW 100 to 200nm
16Kodak KAF-3200MEfront illuminated, 2184 x
1472ITO gates not polysiliconLensletsQE BW
300 nm
17Conclusions
- If we insist on full video rate 30fps
- 4 guider chips 1K x 1K is NOT sufficient
- 16 guider chips 1K x 1K is OK
- If we can make do with 10fps
- 4 guider chips 1K x 1K is marginal
- 4 guider chips 1K x 1K with higher QE is OK
- Sample rate requirements depend on disturbance
spectrum and behavior of optimized Kalman filter - ACS dynamic model is needed!
- SDSS map with u-g-r-i-z would allow better SNR
calc - Need to validate the Secroun centroid SNR estimate
18Future Work
- GSC 2.2 contains some duplicate objects
- overestimates log(N) curve
- does not affect Monte Carlo calc
- GSC has poor accuracy -- roughly 0.4 mag RMS
- bias could invalidate our predictions
- We will probably have two guiders focal plane
and cass focus - require a guide star in FP guider and in CF
guider - would convert 99 into 98 success rate
- no impact if we are 100 covered
- We have non-Aldering fields! Weak Lensing, cal
stars.... - Dont we want to be able to guide anywhere on
the sky? even NGP? - guiding affects PSF -- WL work demands tight
guiding - Use todays SDSS on NGP region try mowing some
stripes - Enlarge SDSS to Aldering region
19Guider CCDs located within GigaCam
Guider CCDs located within rear metering
structure, on optical axis
20Guider thermal structure creep?
- Cass guider corrects for s/c pointing, primary
and secondary motions, but not small motions of
folding flat, tertiary, or GigaCam - Assume 1 degC peak-peak over 3 day orbit, coffin
and GigaCam - dT/dt 1E-5 degC/sec, or 0.01 degC over a 1000
second exposure - Coffin material is CFRP cyanate ester
CTE1ppm/degC - assume dryout is complete after first month on
orbit - GigaCam foundation plate is molybdenum CTE5.4
ppm/degC - Creep within GigaCam baseplate
ltltFOCAL PLANE GUIDER - Given by Texp dT/dt CTEmoly Distance 2nm
- utterly negligible! about 0.2 millipixel or 0.02
milliarcsecond - Creep of coffin center with respect to Gigacam
surface ltltCASS GUIDER - Given by Texp dT/dt CTEcfrp Distance 10nm
- negligible about one millipixel or 0.1
milliarcsecond - Compare best gyro errors 40 mas/10seconds 40000
times worse
21Additional Suggestions, 20 Sept 2002
- Medium format CCDs might be agile able to
quickly dump 99 of a field, and read a selected
1 region slowly with excellent SNR. Rockwell
HiVISI addressable CMOS chip? - Medium format scientific LBL CCDs could have
excellent QEBW products! We should use them, if
staff permits. Of course we need a blind storage
area to eliminate the need for a shutter. Could
run at low pixel rate since only a few rows would
have to be read out repeatedly dump the other
rows 2K x 5rows x 10fps 100kHz. We would
also need a full frame search mode to perform
initial localization, probably with a much higher
pixel rate. - Although we clearly benefit from having a large
available chip area, any one given field will
need only one CCD running -- dont need 16 full
field CCDs running in parallel. We can switch to
a different CCD and a different row group when we
move to each new field of stars.
22Additional Suggestions, continued
- Bad columns could seriously spoil the linearity
with which a star centroid is recovered, hence
radiation damage might cripple a fraction of the
guider area. Best to have plenty of extra sky
field available on board for tracking so that we
can always pick a good guide star located in a
functional CCD column. - Guider (x,y) centroids control two axes, but how
about the third (roll) axis? Dont we need a
really good roll guider as well? Would a Ball
Aerospace CT-602 serve? Do we need diametrically
opposite guide stars in our focal plane? - Algorithm for the centroid must be robust against
CR hits perhaps confine centroid calc range to
2x2 or 3x3 pixels and perform sanity trend check
of each result.