Background information for users of STIS
description
Transcript of Background information for users of STIS
Background information for users of STIS
Charles R. Proffitt
STIS Presentation 2
Outline of Topics
More on Bright and Faint limits with STIS Calibration Lamps Wavelength Calibration Target Acquisitions CCD Operations and Characteristics MAMA Characteristics Time Resolved Observations Observing Overheads Summary of Data Products Spatial Undersampling of STIS Data A few selected data artifacts
STIS Presentation 3
Bright Object Limits
STIS IHB gives tables of worst-case limiting magnitude as a function of grating and source spectrum.
Normalization can vary enormously, depending on grating, aperture, source SED, reddening, etc.
Cool stars especially tricky NUV flux very sensitive to all stellar parameters esp. metallicity FUV flux often dominated by chromospheric emission
Not included in Kurucz or other photospheric models Strongly affected by stellar activity
Epsilon Eri: Kurucz model vs. observed spectrum
STIS Presentation 4
STIS Spectroscopic BOP Limits
Limit for CENWAVE with highest countrate Assumes slitless 1st order; 0.2X0.2 for echelles
STIS Presentation 5
Bright and Faint Limits - example
Example: bright and faint limits for an A0 star Faint limit defined as S/N=10 in one hour For CCD bright limit will saturate CCD in 0.1 s
@gain=4 For MAMA bright limit determined by local or global
BOP limits
STIS Presentation 6
Approximate Bright and Faint Limiting Mag for A0V star at a single wavelength using typical clear apertures
(don’t take exact numbers too seriously)
Grating Wavelength Mag to give S/N=10 in 1 h
Bright limit mag
delta
G750L 7000 20.8 1.1 19.7
G750M 19.0 -1.1 20.1
G430L 5500 20.8 1.5 19.3
G430M 18.4 -1.3 19.7
G230LB 3000 18.3 -1.6 19.9
G230MB 15.4 -4.4 19.8
G230L 2600 18.4 10.4 8.0
G230M 14.4 6.5 7.9
G140L 1350 16.7 8.1 8.6
G140M 13.4 5.8 7.6
E230M 2700 13.2 6.6 6.6
E230H 11.6 5.1 6.5
E140M 1400 10.7 4.5 6.2
E140H 1350 9.8 4.0 5.8
NUV-PRISM 2300 20.6 11.9 8.7
Calibration Lamps
STIS Presentation 8
STIS Calibration Lamps Cal Insert Platform
Flatfielding lamps Tungsten (4 lamps) Krypton (130 - 170 nm) Deuterium (165 - 310 nm)
Echelle wavelength cal PtCr/Ne (LINE)
Cal insert mechanism (CIM) blocks external light & acts as additional external shutter
Hole in the Mirror (HITM) PtCr/Ne (HITM1/2)
1st order wavecals Locate aperture during
target ACQ
STIS Presentation 9
Flat fielding Lamps
For small scale pixel-to-pixel flat fielding Krypton for FUV Deuterium for NUV Tungsten for CCD
Also used for IR fringe flats for G750L & G750M at > 7500 Å
STIS Presentation 10
LINE and HITM lamps spectra
Low dispersion STIS spectra of LINE and HITM1 lamps
Wavelength Calibration
STIS Presentation 12
Wavelength calibration
Causes of Wavelength mis-alignments MSM positioning does not repeat exactly.
Projection of target/aperture shifts by a few pixels Thermal flexure of STIS bench can also shift
projection on detector by a couple of pixels Any drift/mis-centering of target in aperture will
cause corresponding shift in wavelength scale
STIS Presentation 13
Wavelength calibration
Wavecal observations must be adjacent to science No intervening MSM motions because of non-repeatability
Recommend repeating wavecals every 40 minutes Slit-to-slit alignment & repeatability is good
No need to use same slit for science and wavecal For best alignment do ACQ/PEAK in small aperture
G430L Wavecal with 52X0.2 Aperture- Aperture bars allow offsets in cross dispersion direction to also be determined.
E140M Echelle Wavecal
STIS Presentation 14
Wavelength calibration
AUTO-Wavecals meet needed requirements May not always schedule at most efficient time
AUTO-WAVECALS may be turned off for visit GO-WAVECALS may then be specified by observer No automatic enforcement of timing requirements for GO-
WAVECALs
Target Acquisitions
STIS Presentation 16
Need for STIS Onboard Acquistions
With GSC1 typical rms pointing errors were ~ 1” GSC2 is more accurate - 0.1”-0.3” accuracy expected
Many STIS apertures smaller than this
Pointing errors along dispersion direction, translate directly to wavelength errors
Basic STIS ACQ procedure desiged to centroid to ~1/5 CCD pixel or about 0.01”
STIS Presentation 17
Target ACQ exposures
ACQ procedure does the following:1) Images target using 5”x5” subarray1 2) For point source ACQ, use flux weighted
centroid around brightest 3x3 checkbox (extend source algorithm also available).
3) Move spacecraft to put target at reference location on CCD
4) Re-image target1 & centroid again5) Image reference aperture using HITM1
lamp & locate aperture6) Move spacecraft to put target at center
of 0.2X0.2 reference aperture
First image
Second image
Lamp image of 0.2X0.2 aperture
1Each external ACQ image is actually made from 2 subarray images dithered by 3 pixels in x and y. They are shifted into alignment and then combined by taking the minimum value at
each pixel to eliminate cosmic ray hits and hot pixels.
STIS Presentation 18
Selecting Target ACQ parameters
For point-source ACQ exposure S/N > 40:1 suggested More is better. If 40:1 SN needs < 0.1 s minimum exposure time,
see if 0.1 s exposure is unsaturated before switching to less sensitive setting.
But don’t let ACQ saturate. Allowing central pixel to overfill and bleed along columns may affect centroid in y direction.
If there are multiple close stars, be sure which one will be brightest in chosen ACQ filter.
Point source ACQ accurate to 1/5 pixel or 0.01” Diffuse source ACQ algorithms also available
• Larger checkboxes (up to 101x101 pixels or ~ 5”x5”)• Choice of flux Weighted or geometric centering
STIS Presentation 19
ACQ/Peak exposures
Peakups recommended for apertures ≤ 0.1” in size
Do after ACQ Always done using CCD Peakups measure flux through
small aperture and move spacecraft to maximize flux
Need to peakup in both directions for small & short apertures (0.1X0.09)
Special procedures for 0.1X0.03 peakups
Peakups can use images or dispersed light
Accuracy ~ 5% of slit width
STIS Presentation 20
Fixing Orientation on Sky
STIS long slit can be oriented to put extended or multiple targets in aperture
Orient in APT should be (degrees east of N) + 45
Usually 180 degree alternative is just as good
STIS CCD Operations
STIS Presentation 22
CCD Operations
STIS CCD Format
AMP D
Bias and dark correction Daily dark and bias observations and more intensive pre-and post
anneal observations used to create weekly superbias and superdark images used for OTFR pipeline reduction.
Super-bias image subtracted from science image. Serial and parallel overscan regions used to provide 2D correction
to bias levels of image. Superdark is subtracted from science image.
For side 2 data, superdark scaled for CCD housing temperature
CCD includes physical and virtual overscan regions.
Four amps, but most science uses AMP D.
STIS Presentation 23
CCD Operations
Science data also divided by pixel-to-pixel flat field images based on data collected in yearly campaigns.
Some models also have low order flat field images to correct for vignetting.
Monthly anneals warm CCD from ~-85 to ~ +5 C Heals ~80% of transient hot pixels; Increasing numbers of permanent ones accumulate.
STIS Presentation 24
CCD Dark Current & Hot Pixels Initial dark current low: median value ~0.0015 e-/s
Extrapolation predicts 0.009 cnts/pixel/s for Cycle 17 Increased over time due to radiation damage On side-2 no closed loop T control
CCD temperature & dark current varies with T Use housing temperature to scale dark current before dark subtraction
Inexact scaling is an additional source of noise
Monthly anneal (warm from -85 C to + 5C) to heal hot pixels
STIS Presentation 25
CCD Read Noise AMP D has always had lowest read-noise and is
used for science At Gain=1, read-noise initially ~ 4 e-
Increased to 4.5 e- after SMOV3a After switch to STIS side-2, additional 15-18 kHz electronic
noise increased read noise to ~ 5.5 e- (herring bone pattern)• careful Fourier filtering can sometimes remove this
Gain=4 showed pick-up noise even on side-1 ~7.3 e- on side 1; ~ 7.7 e- on side 2
From STIS ISR 2001-05By Tom Brown
STIS Presentation 26
CCD Options
Gain: 1, 2, 4, or 8 Only Gains values of 1 and 4 supported for GO observations. Gain=1 has lower read-noise, but amps saturate at ~33,000 e-. Gain=4 has higher read-noise, but allows full well of CCD to be
used (144,000 e- at center, ~ 120,000 e- at edges) In saturated GAIN=4 images, electrons bleed to other pixels
(perpendicular to dispersion direction), but are not lost. Total response remains linear, allowing very high S/N with special processing techniques.
Binnng at readout by 1, 2, or 4 in either axis or both Binning data during read-out reduces read-noise and file size Increases impact of bad pixels and cosmic rays. With older, noisier detector, usually not worthwhile
STIS Presentation 27
CCD Options - cont
CCD Sub-arrays Can save only part of image array on read-out
Reduces file size and number of buffer dumps required Decreases readout time, allowing increased cadence.
For GOs only support reducing AXIS2 size (perpindicular to dispersion direction) Discards virtual overscan in parallel direction, but retains physical overscan in serial
direction to aid in bias removal Lack of virtual overscan does make bias subraction more difficult
Reducing AXIS1 is an available-but-unsupported mode. Reducing both discards all overscan regions, greatly increasing difficulty of accurate bias
removal. Different clocking patterns used by any CCD sub-arrays may introduce artifacts,
and invalidate assumptions of empirical CTI corrections algorithms. Cosmic Ray rejection is normally done by taking multiple images.
CRREJECT=2 is default
AXIS1
AXIS2
STIS Presentation 28
CCD Charge Transfer Inefficiency
During parallel transfers some electrons get trapped
Trapped e- be released later during read, causing extended “tail”.
Number of free traps depends on flux level that has moved through that pixel.
CTI gets worse with increasing radiation damage
No sufficient pixel based physical model, so need empirical corrections.
Loss increases with # of transfers (1-CTI)n Putting target near readout amp reduces losses E1 positions defined near row 900 Typical exposures of faint targets with the STIS
CCD in cycle 17 might experience 20-30% CTI losses when target at center of the detector, but only 5-8% if at E1 near row 900.
STIS MAMAs
STIS Presentation 30
FUV MAMA Dark Current
FUV MAMA initially had very low dark current (7x10-6 counts/lo-res-pixel/s), but occasionally showed enhanced glow.
Initially glow present only rarely became more frequent over time
Lower edge & lower right hand corner remains mostly dark (near original 7 x 10-6 counts/lo-res-pixel/s).
Physical basis of FUV dark current glow is unclear
STIS Presentation 31
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
131 darksApr 1997Aug 1998178809 s
Mean 0.21Glow 0.37D.C. 0.161x 10-5 c/p/s
(hi-res-pixel)2048 x 2048
126 darksAug 1998Nov 1999173880 s
Mean 0.45Glow 1.08D.C. 0.165x 10-5 c/p/s
141 darksDec 1999May 2001194580 s
Mean 0.56Glow 1.24D.C. 0.161x 10-5 c/p/s
125 darksMay 2003Aug 2004172500 s
Mean 0.65Glow 1.65D.C. 0.154x 10-5 c/p/s
Dark Corner
Glow region
STIS Presentation 32
FUV MAMA Dark Current FUV MAMA dark current increases
~ linearly with time since HV turn-on
Increases faster at higher T Rate of increase has gone up over
the years Hot pixels also increasing
STIS Presentation 33
FUV MAMA Dark Current
Mitigation strategies Use only first orbit of each SAA free period for observations that
need low dark current. (only 1 orbit per day). Keep FUV HVPS off when detector not in use (ops change). Cool detector (NUV MAMA off). Place target on darker part of detector.
New D1 aperture position defined near bottom edge of detector
The count rate summed in each column over a seven pixel high region of the mean dark image covering the period between May 2003 and August 2004. The dotted line gives the results for a region near the standard 1st order spectral location, and the solid line gives the results at the new D1 position
located near the bottom edge of the detector.
STIS Presentation 34
NUV MAMA Dark Current
NUV MAMA dark current dominated by a different physical mechanism than the FUV MAMA
Meta-stable states with lifetimes of days to weeks are populated by high-energy particle impacts, leading to a phosphorescent window glow.
Long term trend depends on low-earth orbit radiation environment
STIS Presentation 35
NUV MAMA Dark Current Effect of temperature changes on dark current is complex Short term changes lead to a large increase in the dexcitation rate,
leading to a large, but temporary, increase in the dark current, including daily cycling as MAMA warms up.
Over the long term, a smaller equilibrium number of populated states partially balances the higher excitation rate caused by higher average T.
If detector cold for long time, large but temporary increase until a new equilibrium is reached.
STIS Presentation 36
MAMA Pipeline Dark Images
Low dark rates require averaging hundreds of images to make useful dark image.
NUV darks semi-empirically scaled for time and temperature changes and subtracted in pipeline.
Secular changes are seen in shape of NUV dark current over time.
Unpredictable nature of FUV glow makes subtracting it in OTFR pipeline impractical
Only base dark current and hot pixels subtracted by pipeline - users need to to custom extraction of glow.
In background limited observations, FUV hot pixels should just be masked out because poor statistics makes subtraction difficult.
STIS Presentation 37
MAMA Flat Fields
On-orbit lamp images used to provide MAMA pixel-to-pixel flats collected during occasional campaigns. MAMA flats very stable once data binned to lo-res
(1024x1024) Can use same pixel-to-pixel flats for essentially all
data. Flat fielding of unbinned 2048x2048 hi-res images not
repeatable - significant structure remains hi res mostly useful for filtering out hot pixels.
Low order MAMA flatfields provided for selected modes (mostly FUV modes).
STIS Presentation 38
MAMA Observation Modes
ACCUM mode Keeps track of how many events fall on each pixel. For medium and high dispersion modes, the pixel
locations are corrected for spacecraft doppler motion as image is accumulated.
STIS data buffer can hold 1 hires (2048x2048) image or up to 7 lowres MAMA + CCD full frame images or 1 hires image + 3 lowres or CCD full frame images
Hires format default for MAMA science, lores for wavecals
STIS Presentation 39
MAMA Observation Modes
TIME-TAG mode Records x-y location and time of each event with 125 micro-
second resolution. Corrections for spacecraft Doppler motion done on ground, not on
spacecraft STIS buffer divided into two sections for time-tag
Each half of buffer can hold 2 x 106 events. One half of buffer can be dumped while other half is recording. User must predict rate and specify buffer time so that buffer is
dumped before one half fills, otherwise gaps will appear in sequence. If global rate < 20,000 counts / s, continuous observations can be
sustained for extended periods (up to 30 buffer dumps). For some projects needing time resolved data, a series of
ACCUM observations may be better than time-tag mode. For CCD observations, the use of subarrays may increase
cadence.
STIS Presentation 40
Time resolved STIS Observing
Detector Mode Minimum Sample time (texp)
Time between samples (dt)
Max time for uninterrupted time series
MAMAs Time-tag 125 s 0 6e7/R s for R < 20,000 cnts/s
4e6/R s for R > 20,000 cnts/s (R=global count rate)
MAMAs Hi-res ACCUMs
0.1 s 30s if texp > 3 m
2 m if texp < 3 mNo limit
MAMAs Lo-res ACCUMs
0.1 s 30s if texp > 3 m
1 m if texp < 3 m
No Limit
7
CCD Full Frame ACCUMs
0.1 s 45 s No limit for texp > 3 m
(texp + dt) x 7
CCD 1060x32 subarray ACCUMs
0.1 s 20 s No limit for texp > 3 m
(texp + dt) x 256
STIS Presentation 41
Other MAMA Constraints
STIS MAMAs cannot be used in any SAA impacted orbit Optical isolators scintillate from cosmic rays and can
cause random bit flips in MAMA electronics STIS low & high voltage turned off during deepest SAA
passages; not practical to turn on MAMA for only part of individual orbits.
Allows use during only one ~ 5 - 6 orbit block per day Observers required to separate CCD and MAMA
science observations into separate visits when practical
STIS Presentation 42
Summary of Overheads
STIS Presentation 43
STIS Data Products
Selected STIS data file types:
opppvvnnd_tag.fits - table of time tag events opppvvnnd_raw.fits - 2d image of unproccesed data opppvvnnd_flt.fits - flat fielded image opppvvnnd_crj.fits - cosmic ray rejected image (CCD) opppvvnnd_x1d.fits - fits table with 1D extracted spectra opppvvnnd_sx1.fits - 1D spectra from summed images opppvvnnd_x2d.fits - 2D spectral image (rectified and flux
calibrated) opppvvnnd_sx2.fits - 2D spectral image (rectified and flux
calibrated) from summed images
STIS Presentation 44
1D spectral extraction
In 1D spectral extraction, an extraction box is centered on spectrum, and summed over cross dispersion direction at each pixel in dispersion direction.
Extracted spectrum is then background subtracted and flux calibrated
Corrections for aperture throughputs, time-dependant sensitivity changes and CTI losses (CCD only) are applied.
Geometry for extraction of 1st order STIS spectra
STIS Presentation 45
1D spectral extraction - cont
For echelle modes, a separate 1D extraction is done for each spectra order
Background subtraction is done using a special algorithm that models the scattered light (see STIS ISR 2002-001 by Valenti et al
STIS Presentation 46
2D spectral extraction
Image rectified so that wavelength and spatial scales are linear and aligned with x and y coordinates.
X2d image is flux calibrated (science images only) Corrections for aperture width and time-dependant
sensitivity changes are applied (no CTI correction).
STIS Presentation 47
Spatial Undersampling of STIS
Critical sampling of PSF requires about 2 pixels per PSF FWHM
STIS CCD spatial scale of ~0.051”/pixel undersampled by ~2x @ 5000 Å.
STIS MAMA spatial scale ~0.0245”/pixel undersampled by ~2x @ 2500 Å.
Undersampling can produce artifacts when extracting spectra at small spatial scales (affected by tilt of spectrum on detector)
Dithering along long slit to sub-sample spatial scale recommended if spatial structure is significant.
STIS Presentation 48
Selected Data Artifacts
CCD window reflections.Brightest ring about 1% of flux
Some MAMA modes also show imaging ghosts.
Airy rings produce spectroscopic fringes
IR fringing due to multiple reflections in CCD. Need contemporaneous tungsten fringe flats to correct properly.