NIRISS NRM bad pixel tolerance analysis
David Lafrenière
2012 February 21
Simulation description
•
•
Master PSF
•
•
•
•
OTE phase maps RevV
NRM mask G7S6SC
Polychromatic, using F430M filter profile
Created with oversampling of 11x11 compared to NIRISS
pixels
•
Soummer, Pueyo, Sivaramakrishnan, Vanderbei OpEx 2007, same as WebbPSF algo)
Pointing
•
•
•
Telescope pointing error for acquisition and dithers: 15 mas
RMS
Telescope pointing jitter while guiding: 5 mas RMS
Simulated at 0.01 pixel precision using oversampled PSFs
Simulation description
•
Detector
•
•
•
•
•
•
Pixel flat field error: 0.1%
Non-uniform intra-pixel sensitivity
•
•
1 at pixel center, decreasing to a mean of 0.8 at pixel corner
(with RMS of 0.05)
Gaussian shape
21 e- read noise per CDS
Mean dark current of 0.012 e-/sec
Inter-pixel capacitive coupling included
Bad pixels included (see later for details)
Simulation description
•
Observation
•
•
•
•
•
•
L’=7.5 mag star
256x256 subarray (tframe=0.66 s)
Read mode, TFIRAPID, Nframe=1, Ngroup=14
•
Peak pixel kept at <70000 e- in last read
9 dithers on a 3x3 grid with 4” step
•
Assumed that the central 7x7 pixel box at each dither
position was free of bad pixels
121 integrations at each dither position
3 hour clock-time total on target
•
•
plus a similar sequence on a calibrator
12 min exposure on each of target and calibrator
Bad pixels
•
•
•
Varied fraction of bad pixels from 0% to 5%
Randomly distributed, assumed that 5/6th of bad pixels are
grouped in a “cross pattern” while 1/6th are individual pixels
Bad pixels assumed completely unusable
128x128
subarray
Bad pixel fraction 1%
Bad pixel fraction 5%
Bad pixels overlaid on PSF
•
A single exposure, on a log display scale
Bad pixel fraction 1%
Bad pixel fraction 5%
Analysis
1. Mask bad pixels. Using integer shifts only, center all PSFs such
that their brightest pixel is aligned
2. Average all images together to obtain a “clean” PSF image
3. For each image, shift the “clean” PSF image by fraction of pixel
to align it precisely with the image, then substitute bad pixels for
the corresponding values from the shifted “clean” image.
4. Apply a fractional shift to all images to precisely register them to
a common center.
5. Average all images to obtain a new, and better, “clean” PSF.
6. Repeat 3 & 4 using the new “clean” PSF image to substitute
bad pixels.
7. Apply the usual procedure to extract the CP and SqV from each
image, then average them.
Results – detection limits
•
Limiting contrast (in difference of magnitude) as a function of
angular separation
0.06”
0.08”
0.10”
0.20”
0.30”
0.40”
0%
7.75
8.26
8.54
8.89
8.93
8.98
0.5%
7.66
8.17
8.45
8.76
8.73
8.79
1%
7.66
8.16
8.43
8.72
8.67
8.75
2%
7.42
7.92
8.19
8.49
8.42
8.49
3%
7.12
7.63
7.92
8.24
8.18
8.24
4%
6.93
7.45
7.74
8.08
8.02
8.08
5%
6.66
7.20
7.50
7.86
7.81
7.85
Results – detection limits
•
Magnitude loss in contrast compared with 0% bad pixels
0.06”
0.08”
0.10”
0.20”
0.30”
0.40”
0%
0.00
0.00
0.00
0.00
0.00
0.00
0.5%
-0.09
-0.09
-0.09
-0.13
-0.20
-0.19
1%
-0.09
-0.10
-0.11
-0.17
-0.26
-0.23
2%
-0.33
-0.34
-0.35
-0.40
-0.51
-0.49
3%
-0.63
-0.63
-0.62
-0.65
-0.75
-0.74
4%
-0.82
-0.81
-0.80
-0.81
-0.91
-0.90
5%
-1.09
-1.06
-1.04
-1.03
-1.12
-1.13
Results – detection limits
•
Factor of loss in contrast compared with 0% bad pixels
0.06”
0.08”
0.10”
0.20”
0.30”
0.40”
0%
1.00
1.00
1.00
1.00
1.00
1.00
0.5%
1.09
1.09
1.09
1.13
1.20
1.19
1%
1.09
1.10
1.11
1.17
1.27
1.24
2%
1.36
1.37
1.38
1.45
1.60
1.57
3%
1.79
1.79
1.77
1.82
2.00
1.98
4%
2.13
2.11
2.09
2.11
2.31
2.29
5%
2.73
2.65
2.61
2.58
2.81
2.83
Conclusion
•
•
Relatively good tolerance to bad pixels using dithers
To limit the loss in contrast at any separation to less
than 0.5 mag compared with 0% bad pixels, the bad
pixel fraction should be less than 3%