TIPS/JIM December 15th 2011
£  News:
£  INS – Roeland van der Marel
£  JWST – Massimo Stiavelli
£  Flexible Workplaces – Peg Stanley
£ Pixel-based CTE Correction of ACS/WFC: CTE Time and
Temperature Dependence- Leonardo Ubeda
£ Correction of the post-SM4 ACS/WFC signal dependent
bias shift– David Golimowski
£ Confluence – Ann Gonella
Next TIPS/JIM January 19th
INS News TIPS/JIM Meeting December 15, 2011 •
•
•
Please email Margaret Meixner if you are available to help with the INS Contribution (Drinks Station) to the STScI ERC Holiday Party, Tuesday, December 20 at 3:00 p.m. If you are attending the AAS meeting in January 2012, please advise Jason Kalirai (JWST) or the HST Mission Office if you are available to help out and spend some time at the STScI booth. ITSD has rolled out a new Leave Request calendaring system. See Margaret Meixner’s email from earlier this week. ITSD will present and discuss the tool at the next TIPS meeting in January. JWST Status
Massimo Stiavelli, December 15th,
2011
Space Telescope Science Institute
Agenda


Mission Progress
STScI Progress
Mission Progress



NGAS delivered the first release of the flight software.
Completed electronics simulator for ISIM
Installed Helium shroud floor at Johnsonʼs Chamber A
Schedule Milestones 
Completed four HQ tracked Project milestones
Milestone
Due Completed
Prototype Exposure Time
Calculators
June 2011
May 9, 2011*
WaveFront Software
Subsystem Integration
November 2011
September 27, 2011**
Proposal & Planning
Subsystem Design Review 3
November 2011
October 31- November 1
Data Management
Subsystem Design Review 1
December 2011
December 5-6
* Objective was to release in advance of the Frontier Science
Opportunities Workshop, June 6-8.
** Early completion resulted from mitigation actions to address a late
start to WAS, MCS & WEx integration. Progress at STScI

Data Management System



SODRM 


SDR review passed in early december
Built 0 completed: integration of up-the-ramp pipeline module and workflow
manager.
Ongoing work involving most of WIT staff.
Karl Gordon gave a presentation to the SWG soliciting their input.
MIRI Acceptance Review


STScI staff participated to the MIRI acceptance review kickoff meeting.
Actual delivery scheduled for April (Project Milestone)
TIPS/JIM December 15th 2011
£ Pixel-based CTE Correction of ACS/WFC: CTE Time and
Temperature Dependence- Leonardo Ubeda
£ Correction of the post-SM4 ACS/WFC signal dependent
bias shift– David Golimowski
£ Confluence – Ann Gonella
Next TIPS/JIM January 19th
ACS/WFC CTE evolution
Leonardo UBEDA
Jay ANDERSON
Space Telescope Science Institute
method
introduction
results
website
New CALACS
1) automatic removal of bias stripes
2) bias shift correction ? David’s talk
3) pixel-based CTE correction
Current data products: _CRJ _FLT _DRZ
New data products:
_CRC _FLC _DRC
Space Telescope Science Institute
Anderson & Bedin (2010, PASP, 122, 1035)
method
introduction
results
website
New CALACS
1) automatic removal of bias stripes
2) bias shift correction ? David’s talk
3) pixel-based CTE correction
Anderson & Bedin (2010, PASP, 122, 1035)
Current data products: _CRJ _FLT _DRZ
New data products:
_CRC _FLC _DRC
ACS Team tests:
The new CALACS:
Linda
Integration onto CALACS
Matt D./Warren
The pixel-CTE correction code:
Jay/Luigi
New reference files:
Amber/Pey-Lian/Sara/Matt M.
Verification of new CALACS for point sources:
Marco/Jay
Verification of new CALACS for extended sources: Ray/Norman/Marco
Verification of new CALACS using simulated data: Roberto/Jay
ACS/WFC CTE evolution:
Leonardo/Jay
Space Telescope Science Institute
introduction
method
results
website
What
Study the evolution of the correction for charge
transfer efficiency
Why
Verify that code is working to 10% accuracy
Verify correct integration onto CALACS
How
Study of hot pixels CTE trails
With
Use of ACS/WFC dark frames
Space Telescope Science Institute
introduction
method
results
website
Method: dark current data set
~5400 _RAW dark frames
2002 − 2011
_RAW
240 Gb raw data
CALACS
~ 5 days
_FLT
_FLC
Space Telescope Science Institute
2.0 Tb processed data
introduction
method
results
website
Method: hot pixel CTE trails
B
C
D
WFC2
WFC1
A
2003
Space Telescope Science Institute
2011
introduction
method
results
website
Method: identification of hot pixels
pixel should have dark current > 0.08 electrons/second
pixel should be present in at least 80% of _FLT dark frames in each anneal cycle
pixel must be 1500 px away from amplifiers
pixel should be isolated within 4 pixel radius
Space Telescope Science Institute
introduction
method
results
website
Method: identification of hot pixels
Space Telescope Science Institute
-77°C
-81°C
% of pixels
103 pixels
pixel should have dark current > 0.08 electrons/second
pixel should be present in at least 80% of _FLT dark frames in each anneal cycle
pixel must be 1500 px away from amplifiers
pixel should be isolated within 4 pixel radius
method
introduction
results
website
Method: pixel distribution
We study the evolution of number of
electrons in pixels P1, P2, and P3 as
a function of time and hot pixel flux
January 2004
Space Telescope Science Institute
October 2009
August 2011
method
introduction
results
website
Method: pixel distribution
We study the evolution of number of
electrons in pixels P1, P2, and P3 as
a function of time and hot pixel flux
case: hot pixel with 1000±50 electrons
January 2004
Space Telescope Science Institute
October 2009
August 2011
introduction
method
results
website
Results
case: hot pixel with 1000±50 electrons
GAIN = 1
GAIN = 2
_FLT CTE correction turned OFF
-77°C -81°C
Space Telescope Science Institute
introduction
method
Results
case: hot pixel with 1000±50 electrons
_FLT CTE correction turned OFF
_FLC CTE correction turned ON
Space Telescope Science Institute
results
website
method
introduction
Results
results
website
P1
P2
P3
P1
P2
P3
P1
P2
P3
100±10 e−
1000±50 e−
10000±1000 e−
Space Telescope Science Institute
introduction
method
Results: we are online
HP = 1000±50 electrons
Space Telescope Science Institute
results
website
www.stsci.edu/hst/acs/
TIPS/JIM December 15th 2011
£ Correction of the post-SM4 ACS/WFC signal dependent
bias shift– David Golimowski
£ Confluence – Ann Gonella
Next TIPS/JIM January 19th
Correction of the Post-SM4 ACS/WFC
Signal-Dependent Bias Shift
David Golimowski
Anatoly Suchkov
Jay Anderson
Norman Grogin
12/15/11 TIPS/JIM
David Golimowski
1
Post-SM4 electronic artifacts
•  Degraded CTE (source: radiation damage)
•  Aperture photometry correction (Chiaberge ISR ACS 09-01)
•  Pixel-based CTE correction (Anderson & Bedin 2010)
•  New CALACS module
•  Horizontal bias stripes (source: CEB-R)
•  1/f noise on Voffset between CDS and ADC
•  Corrections by Grogin et al. (ISR ACS 11-05) and Anderson
•  New CALACS module
•  Static 2-D bias gradient (source: CEB-R)
•  5−10 DN for dual-slope integrator (DSI)
•  Removed by CALACS (BIASCORR)
•  Amplifier cross-talk (source: CCD and CEB-R)
•  5x10−5 at 50,000 e− (Suchkov et al. ISR ACS 10-02)
•  CALACS implementation TBD
•  Signal-dependent bias shift (source: CCD pre-amp and CEB-R DSI)
•  0.02−0.3% among quadrants
•  CALACS implementation TBD
12/15/11 TIPS/JIM
David Golimowski
2
Signal-dependent bias shift
Saturn in quad B far from amp
A
Saturn in quad B close to amp
B
A
Readout
Direction
C
Readout
Direction
D
12/15/11 TIPS/JIM
B
C
David Golimowski
D
3
Bias shift affects 1/f noise removal
Before “destriping”
12/15/11 TIPS/JIM
After “destriping”
David Golimowski
4
Signal Dependent Bias Shift
•  Two electronic components cause bias shift:
1. The high-pass AC filter in the CCD preamp
2. The resistor mismatch in the CEB-R DSI
•  The AC filter causes the DC offset of the preamp to
vary with pixel signal. If many bright/dark pixels are
sampled consecutively, the DC offset rises/falls.
•  The DC offset feeds the DSI, which integrates down
for the reset (bias) level and integrates up for the signal
level. The mismatched resistors cause mismatched
gains in the up and down stages, thereby amplifying
the signal-dependent DC offset (or bias shift).
•  Bias shift can be removed using a parametric algorithm
developed by Markus Loose (ML), but parameters must
be tuned for each quadrant.
12/15/11 TIPS/JIM
David Golimowski
5
Bias shift correction algorithm
•  1-D along each image row, but wraps around to next row after gap of 146 virtual pixels
(parallel shift time / pixel sample time: 3212 µs / 22 µs pix–1 = 146 pix)
•  Four tunable parameters (per quadrant):
τAC
AC filter time constant (45–50 ms ?)
RDC
Ratio of DC offset shift vs pixel signal (0.3 ?)
RDSI
DC sensitivity of DSI
Pgap
Value of virtual pixels in gap
•  For pixel signal Pn and bias Bn : Pn,cor = (Pn – RDC RDSI Bn) – 10/22 RDC (Bn – Bn-1)
where:
Bn = FAC Bn-1 + (1–FAC) Pn ,
FAC = exp(–1 / τAC fsamp) , and
fsamp = (22 µs)–1
•  ML tuned parameters from high-signal, RAW flat fields obtained during SMOV
è poor leverage on τAC
è susceptible to amplifier cross-talk
è poor “correction” of bias gradient
•  Cycle 18 CAL program obtained images of Saturn strategically placed near and far
from each amp to observe wrap-around effect and constrain τAC
12/15/11 TIPS/JIM
David Golimowski
6
CAL prop 12395 image strategy
Stack of 8 images of Saturn placed
near and far from all 4 amplifiers
optical ghosts
12/15/11 TIPS/JIM
David Golimowski
bias shift
7
Raw frame (post-BLEVCORR)
Saturn in quad B far from amp
12/15/11 TIPS/JIM
Saturn in quad B close to amp
David Golimowski
8
Initial (post-SM4) correction
Saturn in quad B far from amp
12/15/11 TIPS/JIM
Saturn in quad B close to amp
David Golimowski
9
Bias gradient problem
Bias shift correction algorithm does not apply to bias gradient
è
è
è
12/15/11 TIPS/JIM
1D algorithm cannot correct 2-D gradient
Time constant of gradient (7–15 ms) is much smaller than τAC
DC and DSI sensitivities also differ from those of bias shift
David Golimowski
10
Proposed CALACS revision
•  Remove bias gradient from image before bias-shift correction
•  Set Pgap = 0; re-optimize τAC, RDC, and RDSI, as needed
•  Perform bias-shift correction first to simplify bias stripe removal:
Step 1: Subtract bias gradients from science image using contemporary superbias
Step 2: Perform bias shift correction, then remove bias stripes
Step 3: Restore bias gradients to corrected image
Step 4: Proceed with usual CALACS bias/dark subtraction, CTE correction, etc.
12/15/11 TIPS/JIM
David Golimowski
11
Optimizing τAC, RDC, and RDSI
•  ML fixed values of τAC and RDC based on presumed characteristics of preamp
components for each CCD quadrant.
•  ML then adjusted RDSI to yield lowest residuals in prescans of flat field image
•  However, images of Saturn placed near and far from amplifiers provide better
leverage on τAC with minimal impact from cross-talk
•  Leaving RDC fixed at ML’s value, iteratively adjust τAC and RDSI to minimize
bias-shift residuals in prescan for both positions of Saturn in each quadrant
•  Flattening prescan is the best alternative to having a “truth image” of Saturn
(although we cannot separate the relatively small influence of serial CTI)
•  This proxy analysis leads to moderate revisions of τAC and RDSI
12/15/11 TIPS/JIM
David Golimowski
12
Initial (post-SM4) parameters
Undersubtracted bias shift
12/15/11 TIPS/JIM
David Golimowski
13
Revised Parameters
12/15/11 TIPS/JIM
David Golimowski
14
Revised Parameters
Param
Description
Quad A
Quad B
Quad C
Quad D
τAC
AC filter time constant (ms)
38.25
36.00
38.25
42.50
RDC
Ratio of DC shift to signal
0.3
0.3
0.3
0.3
RDSI
DC sensitivity of DSI (× 10−3)
2.40
10.23
11.94
3.36
Pgap
Signal during parallel shift (DN)
0
0
0
0
fsamp = 45.4545 kHz,
τpar = 3.212 ms
Nota bene:
These revised parameters, while systematically derived, are one of many
“degenerate” sets of parameters that yield indistinguishable corrections
based on quantitative and visual inspections of residual bias shift.
12/15/11 TIPS/JIM
David Golimowski
15
Final results – Quad B
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
16
Final results – Quad B
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
17
Final results – Quad C
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
18
Final results – Quad C
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
19
Final results – Quad A
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
20
Final results – Quad A
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
21
Final results – Quad D
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
22
Final results – Quad D
BLEV and gradient subtracted
12/15/11 TIPS/JIM
Bias-shift corrected and de-striped
David Golimowski
23
Summary
•  We have applied the parametric bias-shift correction algorithm derived by
Markus Loose (ML) to high-signal images of Saturn strategically placed in
each ACS/WFC quadrant to constrain the parameters.
•  ML’s 1-D bias-shift correction algorithm cannot be used to remove the 2-D bias
gradient, so the gradient must be removed before bias-shift correction is applied.
•  Iterative processing of Saturn images yields “optimized” parameters that minimize
the bias shift residuals in the prescans and permit effective removal of bias stripes.
•  The bias shift is 3–5 times stronger in quads B and C than in quads A and D.
•  3-stage bias-shift correction will be implemented in CALACS as a new first step,
i.e., before de-striping, superbias/superdark subtraction, CTE correction, etc.
(Step 1) subtract bias gradient (via superbias or polynomial surface fit) è
(Step 2) apply bias shift (& cross talk) corrections è
(Step 3) restore bias gradient and proceed with rest of CALACS
12/15/11 TIPS/JIM
David Golimowski
24