INS Division Status Summary for TIPS, 3/15/07
•
ACS Technical Reviews are done. Bob William’s committee’s reviews are done!
Special thanks to
o Marco Sirianni, Linda Smith, Nolan Walborn, Charles Proffitt, Diane
Karakla, Linda Dressel, Ron Gilliland, John Biretta, Marco Chiaberge,
Stefano Casertano, Dave Golimowski, Massimo Robberto, Massimo
Stiavelli, Roeland van der Marel, Andy Fruchter, Bill Sparks, Tom Brown,
Ed Smith, and Andre Martel.
o TTRB members: Howard Bond, Bill Sparks, Dave Soderblom
NEW: Linda Smith, Kailash Sahu, Stefano Casertano
o ACS Bus “C” test in ~2 weeks. Will inform SM4 repair decision.
•
Plan to decide on final organizational structure of division early next week.
o Still need to talk to ~2 dozen people + affected team leads
o Will hold an all-hands meeting to explain.
o Open positions will be posted around same time. Open for a week;
interviews for a week; select leads by 1st or 2nd week in April.
•
Rodger and Doris are still working on getting backlogged Mac orders approved by
HST Project.
•
Getting feedback on new contract; negotiations in progress. WBS #s will change on
May 1. Stay tuned.
•
New HST Ground System Manager is Moonie Ahmed. Meeting today with Rodger
and Director’s Office.
•
TAC meeting next week. Biretta presenting on WFPC2. He, Gilliland & Koekemoer
will provide on-site support.
•
STUC meeting April 12 & 13. Planning presentations.
•
AAS Abstracts due next Wed., April 21. Still haven’t heard from all teams.
TIPS-JIM Meeting
15 March 2007, 10am, Auditorium
1.Two-Gyro Performance: Scheduling and Acquisitions
Merle Reinhart
2. Quantum efficiency of a near-infrared array detector
Peter McCullough
3. JWST calibration error budget
Jerry Kriss
Two-Gyro Performance: Scheduling and
Acquisitions
Merle Reinhart
Scheduling
•
•
Two-Gyro Mode scheduling rates are currently at 72 orbits per week (74 orbits per week
excluding the 5 weeks of ACS hardware issues in 2006 when we scrambled to schedule
other SIs).
Our prediction prior to Two-Gyro Mode entry was 71-73 orbits per week.
Scheduling
•
Our TGM average failure rate (Fine-Lock Backup not considered a failure) is 5.01%.
–
–
–
3.10% in 2005
5.66% in 2006
8.18% in 2007 through the end of the 07.057 SMS
•
•
•
~5.00% for 07.036 to end of the 07.057 SMS
However, we noticed a significant upturn in the failure rate in December, 2006 that continued
through mid-January, 2007.
The Institute and SEs did a lot of analysis and have found two impacting effects
–
–
Coma Galaxies being attempted as guide stars
KDDL-83 missing S-Curves during the acquisition (no further discussion in this presentation as PCS is still
investigating).
Plot from
Mike Wenz
GSC Issues
•
Historically, “bad stars” are typically caused by:
– Double Stars
– Dim Stars, Galaxies, Variable Stars, Blends, Plate Defects, Asteroids, etc
– Failure rate due to catalog issues is ~2.42%
GSC Issues
Galaxies
•
Two issues contributed to the 2006 increase
1. Identified a bug in the GSC2 catalog construction in late Summer as a
result of some preliminary Coma Galaxy Cluster observations.
•
•
•
•
Galaxy information from GSC1 was not consistently propagated into GSC2.
Corrected catalog was delivered to Operations on Sept 12, 2006 (day 255).
However, only the Coma Observations were explicitly reprocessed to use the
updated catalog.
~0.77% of the 2006 failure rate was this bug
2. Dramatic increase in Observations in the Coma Galaxy Cluster in
December.
•
•
•
This is primarily two large programs using ACS/WFC.
Since the galaxy to star ratio is very high in this area, there is a higher chance
of trying to use a mis-identified galaxy as a guide star.
To help mitigate the second problem, we have begun manually prechecking the guidestars used by the Coma Cluster programs.
–
6 galaxies have been found and avoided to date
GSC Issues
Other GSACQ Issues
Mispointings
• Pointing error after successful ReAcq following a failed GSAcq/ReAcq (1/2
Gyro Mode specific)
– Dec 7 - One of the ACS Instrument Scientists, Tom Brown, was not able to identify
the ACS/SBC field in an observation taken the previous day. This was immediately
recognized as a Bright Object Health & Safety concern.
– He also mentioned that a different ACS/WFC observation taken a couple of months
earlier was mispointed by ~80” after a similar GSAcq/ReAcq failure sequence of
events.
– Using the ACS/WFC observation, a quick analysis indicated that the mispointing
was almost exactly the size of the decenterline maneuver.
– Turned over to OTA SE Lead for further analysis. Problem was identified to be a
timing problem between two commands in the failure path of the GSAcq PLCP
group.
– Dec 8 - Based upon the understanding of the issue, computed where the Dec 6
ACS/SBC observation was pointed. Instrument Scientist confirmed the pointing
within a few arcseconds of the prediction.
– Dec 20 - FRR for PLCP changes. Implemented for the 06.359 SMS.
• 2 weeks from identification to FRR
• 2 1/2 weeks from identification to flying
Other GSACQ Issues
Mispointings
•
Pointing error after Single-GS ReAcqs with SAMs in the previous orbit.
– This problem was identified by the OTA SEs in some follow-on analysis of the
previous problem’s test results. However, the cause of this problem is different.
– Problem only occurs with Target-ReAcq style Acqs/ReAcqs where both PASS and
the onboard ReAcq process are effectively accounting for a portion of the SAMs in
the previous orbit.
– Currently utilizing a workaround of doing a full GSAcq every orbit. This is difficult
to impossible to implement for some science observations.
– Solution is to have PASS use the gs centerline acquisition position of the primary
gsacq for all the target-reacq style reacqs (this is what the onboard process expects).
– Discovered that this problem can also affect pair acquisitions in certain limited
situations.
– PASS fix was put into place for the 07.071 SMS.
TIPS-JIM Meeting
15 March 2007, 10am, Auditorium
1.Two-Gyro Performance: Scheduling and Acquisitions
Merle Reinhart
2. Quantum efficiency of a near-infrared array detector
Peter McCullough
3. JWST calibration error budget
Jerry Kriss
Quantum Efficiency Measurements of a
NIRSpec infrared sensor in the ODL*
Peter McCullough, Project P. I.
Mike Regan, ODL Lead
Kevin Lindsay
Eddie Bergeron
Thanks to Bernie Rauscher and Don Figer, et al.
See JWST-STScI-001053 for details.
Presented to TIPS at STScI
by McCullough
Mar 15, 2007
*ODL = Operations Detector Lab at STScI and JHU
Monochromator
Optical Coupling
Lamp
Lockin and cable
for PbSe diode
Blackbody
Image of Blackbody
8
•Blackbody
•Window
•Filter
•Pinhole
•Detector
•Dewar
Simple  Predictable
R
Contours are in
counts (ADUs)
Insert Integrating Sphere to make flat fields
Flat Field
4.5 um
(Subsampled to 512x512)
White: QE=1.05
Black: QE<0.85
Bad pixels: << 1% by
number
Corners have low QE
Reference pixels visible as
black edges
Flat Field
3.9 um
(Subsampled to 512x512)
White: QE=1.05
Black: QE<0.85
Bad pixels: << 1% by
number
Corners have low QE
Reference pixels visible as
black edges
(Same caption as
previous slide, just
different wavelength).
Gain requires accurate nonlinearity correction
0.8 um photons give different
gain than 1.8 um photons.
Interpreted as due to a quantum
yield. Analysis in preparation.
Gain versus
wavelength
and binning
2
Bin factor
1
2
3
4
5
1
0
0.8
Wavelength (um)
2.7
Apparent Gain depends on
binning, due to interpixel
capacitance.
0
z
0
z
1-4z
z
0
z
0
(circuit diagram from Finger et al. 2005)
Summary
JWST-STScI-001053
• Rockwell (Teledyne) 2kx2k sensor H2RG-S010 tested in
Operations Detector Lab at STScI.
• Median QE = 95% within 5% (1 sigma) at tested wavelengths
(3.38, 3.9, 4.5 microns).
• Interpixel capacitance (IPC) is significant and blurs images
due to the 2.5% crosstalk to each of the 4 nearest-neighbor
pixels. Neglecting this crosstalk would cause the QE to be
overestimated by 2*4*2.5% = 20%. This is apparently typical
for epoxy-filled H2RG devices.
• New analysis (binning method) invented to account for IPC in
measurement of system gain; alternative to autocorrelation
method previously used. Intuitive and more resistant to EMI.
• Analysis of measurements from 0.5 to 6.0 um in preparation.
TIPS-JIM Meeting
15 March 2007, 10am, Auditorium
1.Two-Gyro Performance: Scheduling and Acquisitions
Merle Reinhart
2. Quantum efficiency of a near-infrared array detector
Peter McCullough
3. JWST calibration error budget
Jerry Kriss
JWST Calibration Error Budget
Jerry Kriss
JWST Flux & Wavelength
Calibration Requirements

SR-20: JWST shall be capable of achieving data
calibration into physical units with absolute accuracies
shown in the following table.
Table 8-2. Required Calibration Accuracies
Absolute
Calibration
Accuracy
Flux (%):
Imagery
Flux (%):
Coronagraphic
Imagery
Flux (%):
Spectroscopy
Wavelength (%
resolution element):
Spectroscopy
NIRCam
5
5
N/A
N/A
NIRSpec
N/A
N/A
10
12.5
MIRI
5
15
15
10
FGS-TFI
5
10
NA
10
There are no requirements on absolute flux calibration for the FGS-Guider.
15 March 2007
2/14
JWST Astrometric Calibration
Requirements
3.7.1.5.4 Field Distortion Uncertainty
MR-120 After calibration, the uncertainty in the
Observatory field distortion correction within any SI
and the guider shall not exceed 0.005 arcsec RMS.
(There are similar ISIM requirements on knowledge
of relative placements of instruments in the focal
plane.)
These requirements enable the precise target
acquisition required for multi-object
spectroscopy with NIRSpec and coronagraphy
15 with
March 2007NIRCam, MIRI, and FGS-TFI.
3/14
Required Error Budgets

NIRCam



NIRSpec







All tied to OTE/PSF Stability
Imaging: Flux
Coronagraphy: Flux
Both tied to Pointing
Low-resolution Spectroscopy: Flux and Wavelength Error Budget
Medium-resolution Spectroscopy (IFU): Flux and Wavelength
FGS-TFI



All tied to OTE/PSF Stability.
MSA Spectroscopy: Flux and Wavelength
Fixed Slit Spectroscopy: Flux and Wavelength MSA & slits tied to Pointing Error
Budget.
IFU Spectroscopy: Flux and Wavelength
MIRI


Imaging: Flux
Both tied to OTE/PSF Stability
Coronagraphy: Flux
Imaging: Flux and Wavelength
Coronagraphy: Flux and Wavelength
Flux for both tied to
OTE/PSF Stability
Astrometric Calibration
15 March 2007
4/14
Generic Error Budget
Error in Final
Result
Statistical
Error
Absolute Flux
Error
Routine
Calibration
Errors
Pointing
Errors
15 March 2007
Background
Subtraction
Errors
Errors due to
Detector
Effects
Function of
Source
Brightness
Stability
Function
of Source
Brightness
Data
Processing
Errors
OTE/PSF
Stability
5/14
Elements of the Error Budgets (1)

Routine radiometric calibration

Ultimately limited by quality of other calibration
observations.


For NIRCam, since the current known standards are too bright, a
second tier of calibration standards will be required, which
introduces more errors.
Can tie into other error budgets. E.g., pointing errors affect
corrections for slit transmission.
15 March 2007
6/14
raw f ile from 1 SCA;
1 reset of d etector;
M MULTIACCUM s
Flag s set as FITS
head er keyw ord s
CalNIRCam A
Reference f iles
used b y p ip eline
MASKCORR
Ap p ly Mask
MASKFILE
REFCORR
REFCORRTYPE
Ref Pixel Correct
NOISCALC
Calculate Noise
NOISFILE
BDCORR
Sub Bias/Dark
BIASDARKFILE
NLINCORR
Non-lin Correct
NLINFILE
CRID
CR Id entify
SLOPEFIT
Slop e Fit
FLATCORR
Flat Correct
FLATFILE
PHOTCALC
Calculate Phot
PHOTTAB
Calib rated
Im ag e set
Basic JWST Image Processing
Radiometric Flux Calibration
Count rate in spectral bin i
The flux in spectral bin i is
1
1
ri
f! = I! # #
#
s! LFi "!i
Inverse Sensitivity
Width of spectral bin i
Slit transmission
Low-order flat correction
Non-linearity
Raw counts
Corrected counts in pixel j
rj =
15 March 2007
nj
texp
Exposure time
Bias Level
Ref. Pix. Corr.
G( N j ! B j R) 1 1
=
texp
FF j PF j
Fringe Flat
Pixel Flat
8/14
Radiometric Error Propagation
Standard propagation of uncorrelated errors then gives the
fractional error in the absolute flux as
2
2
2
. 2f / &. I2/ . s2/ . LF
) 1 #
. -2/ , . PF
. FF
'
=$ 2 + 2 +
+ 2 + **
+
!
2
2
2
2 '
f/
s/ LF
-/ + PF
FF ( n pix !"
$% I /
15 March 2007
9/14
Elements of the Error Budgets (2)

Background subtraction

More important for fainter sources. Fractional errors scale as
&, B )
$** . ''
$%+ f . (


2
2
2
, - PF
- PF ) 1 #
**
' !
+
2
2 '
FF ( nB !
+ PF
"
Dominated by systematic errors for faint sources
Several potential sources:





15 March 2007
“Sky” (includes scattered light from outside the instrument)
Internal scattered light
Overlapping spectra
Overlapping orders
Light leaking through closed shutters (similar to optical ghosts)
10/14
Elements of the Error Budgets (3)

Detector Effects

Some can be characterized and corrected in routine processing:



Others that are more unpredictable:



Non-linearity
Intrapixel sensitivity
Persistence
Electronic ghosts & EMI pattern noise
Can depend on source brightness:


15 March 2007
External signals can swamp source photons
Bright sources cause their own problems in non-linearity and persistence
11/14
Elements of the Error Budgets (4)

Stability

May affect all aspects of routine calibration:




Flat fields
Flux calibration
Biggest external tie in is to the OTE and PSF
Data Processing Errors


These are expected to be small, but we must manage them.
Examples: Approximations, round-off/truncation errors
15 March 2007
12/14
Example: Error Tree for NIRSpec Flux
Calibration
15 March 2007
13/14
Conclusions


JWST calibration requirements, although “lax” by
HST standards, will be difficult to meet.
The challenges:

For flux calibration, key concerns are:





Current standards are too bright for NIRCam.
PSF stability will have impacts at the few percent level.
Detector effects need to be characterized and quantified better.
Scattered light and backgrounds will be limitations for spectra,
long-wavelength MIRI observations, and coronagraphy.
For wavelength calibration, knowledge of target placement
is crucial. Relying on target acquisition is not enough.
15 March 2007
14/14