Instrument Science Report WFPC2 2001-002
Noiseless Preflashing of the
WFPC2 CCDs
A.B. Schultz, I. Heyer, J. Biretta
February 12, 2001
ABSTRACT
We present results from analysis of WFPC2 images obtained prior to and following preflashing the CCDs (WFPC2 CAL program ID: 8450). A crowded field in the globular cluster Omega Cen (HD116790) was observed, and the WFPC2 calibration lamp was used as
the preflash source for the CCDs. A modest enhancement of ~3% in the detected counts
was measured for point sources far from the horizontal shift register (at Y=800) for the
preflash exposure. For the preflashed exposure, the second image after a preflash showed a
decrease in the detected counts. The CCDs essentially returned to a previous sensitivity
level.
Introduction
This report summarizes the current assessment of preflashing the WFPC2 CCDs before
obtaining a science image. The technique is to obtain an exposure with the calibration
lamp (30 sec. INTFLAT lamp exposure) and reading it out prior to the start of the science
exposure. Reading out the preflash should in principle leave traps filled while contributing
no extra noise to the science image, hence the term “noiseless” preflash. This technique of
preflashing the CCDs was suggested by Biretta and Mutchler (1998) and implemented for
calibration program 8450. However, due to failed guide star REACQs (re-acquisitions), all
of the observations except for the short exposures (16 sec.) were lost. Some star images
were trailed, while other images were lost due to closing of the shutter. HOPR 587 was
Instrument Science Report WFPC2 2001-002
filed against program 8450 and a repeat was approved. Repeat observations, visit 51, were
obtained on February 11, 2000, and analysis of these images is presented here.
Schultz, Heyer, and Biretta (TIR WFPC2-99-02) reported preliminary results for the
usable images (16 sec. exposures) from the first visit for 8450. Photometry of faint star
images showed on average a 3.0 +/- 0.9% enhancement in the stellar counts (at Y=800) in
the preflash exposures. This is consistent with the noiseless preflash giving only a partial
reduction in charge transfer efficiency (CTE). This indicates that the majority of the CTE
effect must be due to traps which release their charge on time scales of less than two minutes. Analysis of the corresponding images from visit 51 yielded similar results to the first
visit data.
In this report, we present analysis of the new 16 sec., 80 sec., and 400 sec. data
obtained on February 11, 2000. The 80 sec. and 400 sec. exposures were obtained with the
target positioned in WF2. The telescope pointing remained constant for these observations. Two star guiding was achieved. There was a bright star in WF3 which made the 80
sec. and 400 sec. WF3 exposures unusable for this project. The WF2 and WF4 data were
used for the long exposure analysis.
Charge Transfer Efficiency (CTE)
The CCDs are clocked vertically to shift charge packets through adjacent potential
wells by changing the bias of each well by clocking. The charge packets are clocked vertically to the horizontal shift register. The output of the horizontal shift register is connected
to a pre-amplifier which converts the charge into a voltage signal.
WFPC2 Charge-Coupled Devices (CCDs) have a CTE problem (ISR-97-05, ISR-9708, TIR-98-01, ISR-98-02). CTE is a factor that measures the loss of electrons during
transfer from one pixel to the next pixel. The measured CTE for the WFPC2 CCDs detectors is under certain conditions ~0.9995 reading down a column. A target will appear to be
fainter when observed at the top of the CCD (Y=800) compared to photometry obtained
when the target is positioned at the bottom of the CCD (Y=1). This is due to a loss of
charge during the transfer from one potential well to the next. This is a serious problem for
photometry of faint sources (< 500 counts).
CCD Traps
Bulk traps in CCDs are electrically active regions due to defects in the crystalline silicon. They are related to impurities and imperfections in the silicon material. As charge
packets are transferred through the device, charge is lost to all empty traps in the material.
Bulk traps are active at all operating temperatures.
High energy incident radiation can create silicon vacancies and ionization damage by
the passage of particles through the material. High energy protons can displace silicon
2
Instrument Science Report WFPC2 2001-002
atoms from their lattice positions by coulombic or nuclear collision. Displacement of silicon atoms can create trapping sites in the CCD’s signal channel which degrades CTE.
These silicon vacancies typically migrate to and get trapped near impurity atoms (Janesick
et al. 1990).
In principle, poor CTE at low light levels (i.e., loss of charge due to traps) can be prevented by filling the traps before reading out the CCD. The CCD can be preflashed by a
light source, such as the calibration lamp for WFPC2, to provide a pedestal of charge to
fill the traps. This induced pedestal is sometimes called a “fat zero”. Once traps are filled,
the lifetime to remain filled is of fundamental importance.
Data
The globular cluster Omega Cen (HD116790) was observed on February 11, 2000 (program ID: 8450, visit 51). The globular cluster was observed before and after a preflash (30
sec. INTFLAT lamp exposure). Each observational sequence started with two darks followed by two target observations, preflash, and two target observations. The dark
observations should have removed any residual charge from the CCD arrays remaining
from previous exposures. The INTFLAT observations were read out (i.e. noiseless) prior
to the start of the science observations. Two sets of data were obtained. For the first set, the
target was positioned in WF4, WF3, and WF2. For the second set (for the longer exposures), the target was positioned only in WF2 with other regions of the cluster being
imaged in WF3 and WF4.
List of Observations
Table 1 presents the first set of new exposures (16 sec.) where the target was positioned in WF4, WF3, and WF2. Table 2 presents the second set of new exposures (16 sec.,
80 sec., 400 sec.) where the target was positioned in WF2. For completeness of the two
data sets, the 16 sec. data with the target positioned in WF2 are repeated in both tables.
Due to the critical time dependence of the preflash, the UT start times (TIME-OBS) of the
observations are listed in Tables 1 and 2.
Calibration & Data Reductions
The data were calibrated using the On-The-Fly Calibration (OTFC) option within StarView upon retrieval from the HST Archive. No additional calibration steps were
performed. Pairs of images were not combined to remove cosmic rays. Individual images
were evaluated to avoid confusion between the effects of preflashing the CCDs and to
determine any affect upon the photometry of residual charge due to reading out images of
the dense star field.
3
Instrument Science Report WFPC2 2001-002
Table 1. WFPC2 CAL Program 8450 Visit 51.
Orbit
LINENUM
Obs.
Aperture
TIME-OBS
Comments
1
51.010
u5ka5101r
-
14:55:14
dark, exp=1800 sec.
51.010
u5ka5102r
-
15:27:14
dark, exp=1800 sec.
51.011
u5ka5103r
WF4
16:02:14
Omega Cen, 16 sec.
51.011
u5ka5104r
WF4
16:04:14
Omega Cen, 16 sec.
51.012
u5ka5105r
WFALL
16:07:14
INTFLAT, exp=30 sec.
51.013
u5ka5106r
WF4
16:09:14
Omega Cen, 16 sec.
51.013
u5ka5107r
WF4
16:11:14
Omega Cen, 16 sec.
51.020
u5ka5108r
-
16:17:14
dark, exp=1800 sec.
51.020
u5ka5109r
-
16:49:14
dark, exp=1800 sec.
51.021
u5ka510ar
WF3
17:24:14
Omega Cen, 16 sec.
51.021
u5ka510br
WF3
17:26:14
Omega Cen, 16 sec.
51.022
u5ka510cr
WFALL
17:29:14
INTFLAT, exp=30 sec.
51.023
u5ka510dr
WF3
17:31:14
Omega Cen, 16 sec.
51.023
u5ka510er
WF3
17:33:14
Omega Cen, 16 sec.
51.030
u5ka510fr
-
17:39:14
dark, exp=1800 sec.
51.030
u5ka510gr
-
18:11:14
dark, exp=1800 sec.
51.031
u5ka510hr
WF2
18:46:14
Omega Cen, 16 sec.
51.031
u5ka510ir
WF2
18:48:14
Omega Cen, 16 sec.
51.032
u5ka510jr
WFALL
18:51:14
INTFLAT, exp=30 sec.
51.033
u5ka510kr
WF2
18:53:14
Omega Cen, 16 sec.
51.033
u5ka510lr
WF2
18:55:14
Omega Cen, 16 sec.
2
3
4
Instrument Science Report WFPC2 2001-002
Table 2. WFPC2 CAL Program 8450 Visit 51, target positioned in WF2.
Orbit
LINENUM
Obs.
TIME-OBS
Comments
3
51.030
u5ka510fr
17:39:14
dark, exp=1800 sec.
51.030
u5ka510gr
18:11:14
dark, exp=1800 sec.
51.031
u5ka510hr
18:46:14
Omega Cen, 16 sec.
51.031
u5ka510ir
18:48:14
Omega Cen, 16 sec.
51.032
u5ka510jr
18:51:14
INTFLAT, exp=30 sec.
51.033
u5ka510kr
18:53:14
Omega Cen, 16 sec.
51.033
u5ka510lr
18:55:14
Omega Cen, 16 sec.
51.040
u5ka510mr
19:01:14
dark, exp=1800 sec.
51.040
u5ka510nr
19:33:14
dark, exp=1800 sec.
51.041
u5ka510or
20:15:14
Omega Cen, 80 sec.
51.041
u5ka510pr
20:18:14
Omega Cen, 80 sec.
51.042
u5ka510qr
20:22:14
INTFLAT, exp=30 sec.
51.043
u5ka510rm
20:24:14
Omega Cen, 80 sec
51.043
u5ka510sr
20:27:14
Omega Cen, 80 sec
51.050
u5ka510tr
20:34:14
dark, exp=1800 sec.
51.050
u5ka510ur
21:06:14
dark,exp=1800 sec.
51.051
u5ka510vr
21:52:14
Omega Cen, 400 sec.
51.051
u5ka510wr
22:01:14
Omega Cen, 400 sec.
51.052
u5ka510xr
22:14:14
INTFLAT,exp=30 sec.
51.053
u5ka510yr
22:16:14
Omega Cen, 400 sec.
51.053
u5ka510zr
22:25:14
Omega Cen, 400 sec.
4
5
5
Instrument Science Report WFPC2 2001-002
Pairs of images were combined using the STSDAS task gcombine to create finder
charts. This allowed distinguishing faint stars from cosmic ray hits in the individual
frames and to identify those star images that were seriously affected by cosmic ray hits.
For the first set of exposures, identical stars were manually selected from the respective
WF2, WF3, and WF4 frames. For the second set of exposures, a list of stars was selected
from the longer 400 sec. exposures, and this list was used as input for the photometry of
the respective 80 sec. exposures. Star selection was determined by the presence of contamination from saturated star images, close or binary stars, and cosmic rays mis-identified as
stars.
Photometry was performed using the DAOPHOT task phot with the list of star positions manually identified as input. An aperture radius of 3.5 pixels was used with the sky
annulus defined to be between 4 and 6 pixels. The centering algorithm was set to “centroid” with parameters cbox=3 and maxshif=1. All photometry values were written to
tables for subsequent cross-correlation of identical stars in the respective wide field
frames.
Data Characteristics
There were no reported guiding anomalies during the 8450 visit 51 observations. Two
guide star guiding was achieved. As the exposure times were increased from 16 sec. to 80
sec., and finally to 400 sec., stars that were relatively bright in the shorter exposures
became saturated with extended halos. The background levels increased for the longer
exposure observations due to a multitude of faint stars not visible in the 16 sec. exposures,
wings of the PSF, cosmic ray hits, and light scatter. Faint stars near brighter ones became
lost at longer exposure times due to overlap of the brighter star’s PSF. Some close pairs of
stars at the shorter exposures merged into a common image in the long 400 sec. exposure
data. There was a bright star in WF3 which made the 80 sec. and 400 sec. exposures unusable for this project. The WF2 and WF4 data were used for this analysis.
Table 3. Mean background counts for the 8450 visit 51 non-preflashed data.
CCD
16 sec.
(counts)
80 sec.
(counts)
400 sec.
(counts)
PC1
-0.004 +/- 0.042
0.064 +/- 0.046
0.586 +/- 0.062
WF2
0.088 +/- 0.028
0.741 +/- 0.295
3.383 +/- 0.113
WF3
0.062 +/- 0.031
0.608 +/- 0.107
2.821 +/- 0.022
WF4
0.035 +/- 0.042
0.577 +/- 0.118
3.138 +/- 0.041
Table 3 presents the measured background for each CCD chip from the non-preflashed
data, and Table 4 presents the measured background for the first image following the preflash. The counts in several small 21x21 pixel apertures randomly selected in each CCD
6
Instrument Science Report WFPC2 2001-002
were sampled and the mean counts determined. Regions selected were relatively free of
cosmic ray hits, faint stars, bright stars, and diffraction spikes. For the WF3 background
statistics, the region in the image not affected by the bright star was sampled. A better way
to determine the background for the short exposure data is to scale the background for the
400 sec. exposures to the 16 sec. and 80 sec. exposure times.
Table 4. Mean background counts for the 8450 visit 51 preflashed data.
CCD
16 sec.
(counts)
80 sec.
(counts)
400 sec.
(counts)
PC1
-0.010 +/- 0.039
0.112 +/- 0.061
0.989 +/- 0.127
WF2
0.168 +/- 0.031
0.757 +/- 0.053
4.333 +/- 0.602
WF3
0.160 +/- 0.040
0.755 +/- 0.041
3.783 +/- 0.305
WF4
0.178 +/- 0.020
0.668 +/- 0.072
4.017 +/- 0.522
The amount of charge on the CCD chip, and thus available to fill traps, depends upon
the lamp-on time. The flat field exposures were 30 seconds in the F555W filter, yielding
approximately 2,000 counts on average to the wide field detectors (gain=7 e-/DN). Table 5
presents the mean counts in a region [100:800,100:800] for each chip. The vignetted
regions for each chip near the edges with adjacent chips were not used to determine the
mean counts. The lamp-on exposure time was set to produce a maximum effect on the
wide field chips. The PC1 aperture is not important for this report.
Table 5. F555W Filter Internal Flat Field Statistics (exp=30 sec.).
APERTURE
NPIX
MEAN
STDDEV
MIN
MAX
PC1
491401
409.
54.3
2.329
1648.
WF2
491401
2320.
280.8
4.597
3778.
WF3
491401
2235.
285.5
4.287
3785.
WF4
491401
1544.
255.8
1.697
3791.
Analysis
The measured and corrected counts (FLUX) for each star were ratioed with the counts
for the same star after preflashing the CCD chips. The preflash ratio is defined as the ratio
between preflash and non-preflash measurements (preflash/non-preflash) of the same star
at the same location on the same chip. Y-preflash is defined as the percentage of ratio
increase (slope) over 800 pixels in the Y direction.
7
Instrument Science Report WFPC2 2001-002
Arithmetic operations on the data tables were performed using the ttools package task
tcalc. The trends in the data, over 800 pixels, was determined by fitting a linear fit to the
ratios using the task polyfit within the utilities package. No corrections for CTE were
applied. Photometry of stars of all intensities were used for the analysis. A few points outside of the range 0.9 < # < 1.1 (presumably due to cosmic rays, etc.) were discarded and
not used for the fit.
Effect of Preflash on Y-CTE
8450_v51_wf2_ratio_09_11.da
102
1.2
WF2
1.1
Number
Flux Ratio (preflash/non-preflash)
In this section, we present analysis of the 16 sec., 80 sec. and 400 sec. exposure data.
Figures 1-7 show comparisons of the “preflash ratio” (preflash/non-preflash) for the wide
field chips. A summary of the different Y-preflash values is presented in Table 6.
1
101
.9
Y-preflash = 2.2 +/- 1.5%
.8
0
200
400
600
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
Figure 1: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510kr/u5ka510ir)
vs. vertical position on the WF2 chip for 16 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
Table 6. Y-preflash. The increase in counts in % over 800 pixels for the 8450 visit 51 data.
CCD
16 sec.
80 sec.
400 sec.
average
WF2
2.2 +/- 1.5%
2.1 +/- 1.1%
2.3 +/- 0.8%
2.2 +/- 0.1%
WF3
2.5 +/- 1.6%
-
-
-
WF4
3.0 +/- 1.3%
3.6 +/- 1.3%
3.7 +/- 1.0%
3.4 +/- 0.3%
average
2.5+/- 12.8%
2.8 +/- 26.3%
3.0 +/- 23.3%
2.8 +/- 21.4%
8
8450_v51_wf3_ratio_09_11.da
102
1.2
WF3
1.1
Number
Flux Ratio (preflash/non-preflash)
Instrument Science Report WFPC2 2001-002
1
101
.9
Y-preflash = 2.5 +/- 1.6%
.8
0
200
400
600
0
800
500
1000
1500
2000
2500
3000
3500
Counts
y-position
8450_v51_wf4_ratio_09_11.da
102
1.2
WF4
1.1
Number
Flux Ratio (preflash/non-preflash)
Figure 2: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510dr/u5ka510ar)
vs. vertical position on the WF3 chip for 16 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
1
101
.9
Y-preflash = 3.0 +/- 1.3%
.8
0
200
400
600
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
Figure 3: Left, the ratio of the preflash/non-preflash flux in DA (u5ka5106r/u5ka5104r)
vs. vertical position on the WF4 chip for 16 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
9
Instrument Science Report WFPC2 2001-002
u0or_0rm_09_11_2.da
102
1.2
WF2
1.1
Number
Flux Ratio (preflash/non-preflash)
Figures 4-5 show comparisons of the “preflash ratio” (preflash/non-preflash) for the 80
sec. exposure data.
1
101
.9
Y-preflash = 2.1 +/- 1.1%
.8
0
200
400
600
800
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
u5ka510or_0rm_09_11_4.da
102
1.2
WF4
1.1
Number
Flux Ratio (preflash/non-preflash)
Figure 4: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510rm/u5ka510or)
vs. vertical position on the WF2 chip for 80 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
1
101
.9
Y-preflash = 3.6 +/- 1.3%
.8
0
200
400
600
800
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
Figure 5: Left, the ratio of the preflash/non-preflash flux in DA (u5ka510rm/u5ka510or)
vs. vertical position on the WF4 chip for 80 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash
10
Instrument Science Report WFPC2 2001-002
u0v_0ry_09_11_2_test.da
102
1.2
WF2
1.1
Number
Flux Ratio (preflash/non-preflash)
Figures 6-7 show comparisons of the “preflash ratio” (preflash/non-preflash) for the
400 sec. exposure data.
1
101
.9
Y-preflash = 2.3 +/- 0.8%
.8
0
200
400
600
800
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
u0vr_0yr_09_11_4.da
102
1.2
WF4
1.1
Number
Flux Ratio (preflash/non-preflash)
Figure 6: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510yr/u5ka510vr)
vs. vertical position on the WF2 chip for 400 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
1
101
.9
Y-preflash = 3.7 +/- 1.0%
.8
0
200
400
600
800
0
500
1000
1500
2000
2500
3000
3500
Counts
y-position
Figure 7: Left, the ratio of the preflash/non-preflash flux in DA (u5ka510yr/u5ka510vr)
vs. vertical position on the WF4 chip for 400 sec. exposures. The increase in % over 800
pixels (Y-preflash) is presented. Right, histogram of the measured stellar flux for stars
used to determine the Y-preflash.
11
Instrument Science Report WFPC2 2001-002
Care must be exercised in the interpretation of Figures 1-7 as the data constraint 0.9 <
# < 1.1 may artificially flatten the distributions. The effect of preflash on Y-CTE for stars
with measured counts of 50 to 200 and 200 to 500 counts is presented in Appendix I. The
mean value of the Y-preflash increase is approximately 2.8 +/- 0.6% for the wide field
chips. Due to the limited number of stars and the number of observations, the chip-to-chip
differences between the measured Y-preflash values on the different chips are not significant. These plots indicate that the noiseless preflash increased the stellar counts at high
CCD Y numbers by about 3%, while at low CCD Y numbers there is no effect. This
behavior is consistent with a reduction in the CTE photometry ramp.
Residual charge
The comparisons of the measured stellar flux ratio between the first and second images
after a dark (case 1) showed a mean increase of ~1-2% in the flux ratio, while the same
comparison between the first and second images following preflashing with the calibration
lamp (case 2) showed a mean decrease of ~1-2% in the flux ratio. In the first case, the
increase in the stellar flux ratio indicated that reading out the image of the star cluster
would leave some traps filled. For the second case, reading out the image a second time
following preflashing with the calibration lamp returned the flux ratio to that achieved by
reading out the image of the star cluster. It appears that reading out the image of the star
cluster reduces the CTE ramp, but it is not as well as preflashing with the calibration lamp.
This indicates that preflashing with the calibration lamp to fill traps only works for the first
image following readout of the flat field. These results raise the question as to whether it is
the readout time or the time constant for the traps (or both) that is important when preflashing the CCDs.
CTE measurement
In this section, we present analysis of the data for which the telescope pointing was
changed to position the target in WF4->WF3->WF2. Identical stars were manually
selected from each of the three WF frames. Stars at the top of WF4 (Y=800) were found at
the bottom of WF2 (Y=1). Hence, comparing the counts for a star on WF2 and WF4 provides a simple measure of CTE.
Figures 8-9 show the CTE measurements between apertures WF2 and WF4 before and
after the preflash. Only stars with counts in the 3.5 pixel aperture in the range 100 to 500
DN were included in this analysis. The detected counts were corrected for X-CTE using
formula (1b) from Whitmore, Heyer, and Casertano (1999), so as to isolate Y-CTE effects.
12
Instrument Science Report WFPC2 2001-002
100 < DN < 500
non-preflash
1.5
Number
Flux Ratio (WF2/WF4)
u5ka5103r_0hr_cte_100_500.da
102
2
1
101
.5
Y-CTE = 25.6 +/- 1.7%
0
-500
0
∆Y
0
-100
500
0
100
200
300
400
500
Counts
Figure 8: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
non-preflash data for WF2 vs. WF4 (u5ka510hr/u5ka5103r) with the Y-CTE loss in %
over 800 pixels. Right, histogram of the measured stellar flux for stars used to determine
the Y-CTE.
u5ka5106r_0k_cte_100_500.da
102
2
preflash
1.5
Number
Flux Ratio (WF2/WF4)
100 < DN < 500
1
101
.5
Y-CTE = 24.1 +/- 1.6%
0
-500
0
∆Y
0
-100
500
0
100
200
300
400
500
Counts
Figure 9: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
preflash data for WF2 vs. WF4 (u5ka510kr/u5ka5106r) with the Y-CTE loss in % over
800 pixels. Right, histogram of the measured stellar flux for stars used to determine the YCTE.
13
Instrument Science Report WFPC2 2001-002
The formula applied to the data are:
X-CTE = 2.5 * [1 + 0.341(0.00720 - 0.0020log10CTSobs)*(MJD - 49471)]
CTScor = [1 + X-CTE/100 * X/800] * CTSobs
The comparison between the non- and preflash Y-CTE loss measurements (Figures 8
and 9) suggests that the preflash reduced the Y-CTE loss by 1.5 +/- 2.2% over 800 pixels.
The Y-CTE for stars with measured counts of 50 to 200 and 200 to 500 counts is presented
in Appendix II. The basic assumption is that electrons fill the traps and remain in the traps
on a time scale long enough so that charge traps are not available to trap electrons in the
targets as they are read down a column. It appears that the majority of traps have time
scales shorter than about 2 minutes which is the time difference between the start of the
preflash exposure and the start time of the following observation.
CTE web tool
A web based CTE tool has been created by Mike Wiggs and John Biretta as an aid to
WFPC2 observers. This tool allows users to estimate the effects of CTE losses on point
sources. The correction formulae were taken from Whitmore, et al., PASP, 1999, 111,
1559. The URL for the tool is:
http://www.stsci.edu/instruments/wfpc2/Wfpc2_cte/wfpc2_cte_calc.html
The web tool yields consistent results for the same star observed at high and low Y-position. For example, Table 7 presents the measured flux for the same star observed in WF2
and WF4. The data (gain=7) were obtained in back-to-back orbits, for the WF4 data
mjd=51585.67308686 and for the WF2 data the mjd=51585.78697575. The equation for
the line (obs. cts. range 100 < # < 500) in Figure 8 is:
–4
ratio equ = – 3.010182 ×10
× ∆Y + 1.009857
Table 7. WFPC2 web tool CTE correction comparisons (* counts corrected for X-CTE
but not Y-CTE).
Star
CCD
xpos
ypos
obs cts
CTS*cor
Corrected
(ratio equ.)
Corrected
(web tool)
#1
WF2
755.34
752.0
191.9307
205.065
253.06
258.61
WF4
88.0
91.23
240.2348
242.057
251.09
251.70
WF2
691.874
81.608
262.9186
278.278
287.85
285.53
WF4
160.801
759.091
215.4408
218.501
270.58
292.80
#2
14
Instrument Science Report WFPC2 2001-002
Conclusions
The WFPC2 CCDs exhibit a less than ideal CTE. The CTE loss for faint stars on faint
backgrounds has increased since 1995 from ~3% to as high as ~40% for a single faint star
at the top of the chip (Y=800) (Whitmore 1998, Whitmore et al. 1999). As reported earlier
from the analysis of the 8450 visit 01 data, the noiseless preflash yields a modest decrease
in the CTE effects (Schultz, Heyer, Biretta, 1999). The noiseless preflash increased stellar
counts at Y=800 by an average of about 2.8 +/- 0.6%, while the measured Y-CTE loss was
8 times this amount. This result is independent of the length of the science exposure. It is
evident from these results, as well as those reported earlier, that the noiseless preflash electrons are not held in traps long enough to significantly reduce the effects of CTE loss on
aperture photometry of faint stellar targets.
More recently, Dolphin (2000) compared WFPC2 observations with ground based
observations of Omega Centauri and NGC 2419, using a baseline through March 2000,
roughly a year longer than available in Whitmore et al. (1999). In general he finds good
agreement with the Whitmore et al. (1999) results. In addition, the improved functional
form, longer baseline, and more extensive data set used by Dolphin result in less scatter in
the residuals. In particular, he finds similar corrections to within a few hundredths of a
magnitude in all cases except for recent (1998 and later) data with low counts. In these
cases, the Dolphin corrections are larger than the Whitmore et al. (1999) corrections.
Whitmore (2000; private communications) is currently analyzing an August 2000 dataset
to see which of the two corrections provides better results.
It is recommended that if possible, a faint target should be imaged close to the pyramid
apex at pixel location (150,150) to reduce the effects of CTE loss. Placing targets closer to
the pyramid apex than this position one risks the target landing near the vignetted regions
and affecting the resulting photometry. For the wide field CCDs, aperture=WALL is recommended. The aperture reference point for WALL is at pixel (133,149) on the WF3 chip.
For PC1 imaging, it is recommended that a POS TARG be used to move the target from
the aperture reference point (420.0,424.5) to the desired position (150,150) using (POS
TARG -12.292,-12.491).
Acknowledgements
We wish to thank Vicky Balzano and Wayne Baggett (Commanding) for discussions
about the WFPC2 timing sequences when commanding an internal flat field and the following science exposures.
References
Biretta, J. and Mutchler, M. 1998, Charge Trapping and CTE Residual Images in the
WFPC2 CCDs, WFPC2-ISR-97-05.
15
Instrument Science Report WFPC2 2001-002
Casertano, S. and Mutchler, M. 1998, The long vs. short anomaly in WFPC2 images,
WFPC2-ISR-98-02.
Clampin, M. 1992, WFPC-II CCDs, WFPC2-ISR-1992-06
Dolphin, A.E. 2000, “The Charge Transfer Efficiency and Calibration of WFPC2”,
PASP, Oct. 2000, in press.
Janesick, J. Soli, G., Elliott, T. and Collins, S. 1990, “Predicting the Effects of Proton
Damage on Charge-Coupled Devices using the Radiation Transfer Technique”, paper No.
103, IEEE 27th International Nuclear & Space Radiation Effects Conference, Reno, NV,
July 16-20, 1990.
Schultz, A.B., Heyer, I., and Biretta, J. 1999, Preliminary Results of the Noiseless Preflash Test (prop. 8450), WFPC2-TIR-1999-02.
Whitmore, B and Heyer, I. 1995, A Demonstration Analysis Script for Performing
Aperture Photometry, WFPC2 ISR-95-04.
Whitmore, B. and Heyer, I. 1998, New Results on Charge Transfer Efficiency and
Constraints on Flat-Field Accuracy, WFPC2-ISR-97-08.
Whitmore, B. 1998, Time Dependence of the Charge Transfer Efficiency on the
WFPC2, WFPC2-TIR-98-01.
Whitmore, B., Heyer, I., and Casertano, S. 1999, “Charge Transfer Efficiency of the
Wide Field and Planetary Camera 2”, PASP, 111, 1559-1576.
16
Instrument Science Report WFPC2 2001-002
Appendix I: Effect of Preflash on Y-CTE
8450_v51_wf2_50_200.da
102
1.5
WF2
50 < DN < 200
1.25
Number
Flux Ratio (preflash/non-preflash)
Photometry of faint star images is more affected by CTE than brighter targets. The
analysis presented in the main body of this report used stars of all intensities. In this section, we present the analysis of the 16 sec. exposure data for stars with measured counts in
the range 50-200 and 200-500 counts. Photometry was performed with an aperture radius
of 3.5 pixels and a sky annulus defined to be between 4 and 6 pixels. No corrections for
CTE were applied. The effect of preflash on Y-CTE is defined to be the increase in the
ratio of counts in % over 800 pixels.
1
101
.75
Y-preflash = 6.3 +/- 5.8%
.5
0
200
400
600
0
-100
800
0
100
y-position
200
300
400
500
Counts
Figure 10: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510kr/u5ka510ir)
vs. vertical position on the WF2 chip for 16 sec. exposures. Right, histogram of the measured stellar flux for stars used to determine the Y-preflash.
Table 8. Y-preflash for stars with measured counts (flux) in the range 50-200 and 200-500
counts for the 8450 (visit 51) 16 sec. exposure data. The Y-preflash is the increase in the
ratio of counts in % over 800 pixels after preflashing.
Counts
WF2
WF3
WF4
50-200
6.3 +/- 5.8%
4.7 +/- 5.8%
1.4 +/- 4.1%
200-500
-1.1 +/- 2.8%
6.1 +/- 2.9%
6.8 +/- 2.4%
17
Instrument Science Report WFPC2 2001-002
WF3
50 < DN < 200
1.25
Number
Flux Ratio (preflash/non-preflash)
8450_v51_wf3_50_200.da
102
1.5
1
101
.75
Y-preflash = 4.7 +/- 5.8
.5
0
200
400
600
0
-100
800
0
100
200
300
400
500
Counts
y-position
Figure 11: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510dr/u5ka510ar)
vs. vertical position on the WF3 chip for the 16 sec. exposure. Right, histogram of the
measured stellar flux for stars used to determine the Y-preflash.
WF4
50 < DN < 200
1.25
Number
Flux Ratio (preflash/non-preflash)
8450_v51_wf4_50_200.da
102
1.5
1
101
.75
Y-preflash = 1.4 +/- 4.1%
.5
0
200
400
600
800
y-position
0
-100
0
100
200
300
400
500
Counts
Figure 12: Left, the ratio of the preflash/non-preflash flux in DA (u5ka5106r/u5ka5104r)
vs. vertical position on the WF4 chip for 16 sec. exposures. Right, histogram of the measured stellar flux for stars used to determine the Y-preflash.
18
8450_v51_wf2_200_500.da
102
1.5
WF2
200 < DN < 500
1.25
Number
Flux Ratio (preflash/non-preflash)
Instrument Science Report WFPC2 2001-002
1
101
.75
Y-preflash = -1.1 +/- 2.8%
.5
0
200
400
600
0
-100
800
0
100
200
300
400
500
Counts
y-position
Figure 13: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510kr/u5ka510ir)
vs. vertical position on the WF2 chip for 16 sec. exposures. Right, histogram of the measured stellar flux for stars used to determine the Y-preflash.
WF3
200 < DN < 500
1.25
Number
Flux Ratio (preflash/non-preflash)
8450_v51_wf3_200_500.da
102
1.5
1
101
.75
Y-preflash = 6.1 +/- 2.9%
.5
0
200
400
600
800
y-position
0
-100
0
100
200
300
400
500
Counts
Figure 14: Left, the ratio of the preflash/non-preflash flux in DN (u5ka510dr/u5ka510ar)
vs. vertical position on the WF3 chip for the 16 sec. exposure. Right, histogram of the
measured stellar flux for stars used to determine the Y-preflash.
19
Instrument Science Report WFPC2 2001-002
WF4
200 < DN < 500
1.25
Number
Flux Ratio (preflash/non-preflash)
8450_v51_wf4_200_500.da
102
1.5
1
101
.75
Y-preflash = 6.8 +/- 2.4%
.5
0
200
400
600
0
-100
800
0
100
200
300
400
500
Counts
y-position
Figure 15: Left, the ratio of the preflash/non-preflash flux in DA (u5ka5106r/u5ka5104r)
vs. vertical position on the WF4 chip for 16 sec. exposures. Right, histogram of the measured stellar flux for stars used to determine the Y-preflash.
Figures 10-15 show comparisons of the “preflash ratio” (preflash/non-preflash flux) for
the wide field chips. A summary of the different Y-preflash values is presented in Table 8.
The ratio is sensitive to the flux of the individual stars, which depends upon the exposure
length, the background and the sky subtraction, and the number distribution of the stars in
the sample. The Y-preflash variations are not significant for these two samples of stars, 50
< DN < 200 and 200 < DN < 500.
The Y-preflash values in Table 8 could be consistent with an average enhancement of
4.0 +/- 72%. There are no reasons to ignore the low Y-preflash values other than that they
are low. For these samples, there may be a preference to select stars near the center of the
chips over those from the top and bottom of the chips. Ignoring the two low values in
Table 8, the average enhancement in the stellar counts (Y=800) becomes 6.0 +/- 13.4%.
20
Instrument Science Report WFPC2 2001-002
Appendix II: CTE measurement
In this section, we present the Y-CTE analysis of the 16 sec. exposure data for stars
positioned in WF4 and WF2 with measured counts in the range of 50 to 200 and 200 to
500 counts. Identical stars were manually selected from each of the two WF frames. Stars
at the top of WF4 (Y=800) were found at the bottom of WF2 (Y=1) and vise versa. The
detected counts were corrected for X-CTE using formula (1b) from Whitmore, Heyer, and
Casertano (1999), so as to isolate Y-CTE effects. The measurement of the Y-CTE in the
body of this report used stars with intensities of 100 to 500 counts.
50 < DN < 200
non-preflash
1.5
Number
Flux Ratio (WF2/WF4)
u5ka5103r_0hr_cte_50_200.da
102
2
1
101
.5
Y-CTE = 32.9 +/- 3.1%
0
-500
0
0
-100
500
∆Y
0
100
200
300
400
500
Counts
Figure 16: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
non-preflash data for WF2 vs. WF4 (u5ka510hr/u5ka5103r) with the Y-CTE loss in %
over 800 pixels. Right, histogram of the measured stellar flux for stars used to determine
the Y-CTE.
Table 9. Y-CTE for two samples of stars from the 8450 (visit 51) 16 sec. exposure data,
stars with counts (flux) of 20-200 and 200-500 counts.
Counts
non-preflash
preflash
50-200
32.9 +/- 3.1%
32.1 +/- 2.9%
200-500
24.6 +/- 1.7%
20.0 +/- 3.6
21
Instrument Science Report WFPC2 2001-002
200 < DN < 500
non-preflash
1.5
Number
Flux Ratio (WF2/WF4)
u5ka5103r_0hr_cte_200_500.da
102
2
1
101
.5
Y-CTE = 24.6 +/- 1.7%
0
-500
0
∆Y
0
-100
500
0
100
200
300
400
500
Counts
Figure 17: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
non-preflash data for WF2 vs. WF4 (u5ka510hr/u5ka5103r) with the Y-CTE loss in %
over 800 pixels. Right, histogram of the measured stellar flux for stars used to determine
the Y-CTE.
50 < DN < 200
preflash
1.5
Number
Flux Ratio (WF2/WF4)
u5ka5106r_0k_cte_50_200.da
102
2
1
101
.5
Y-CTE = 32.1 +/- 2.9%
0
-500
0
∆Y
0
-100
500
0
100
200
300
400
500
Counts
Figure 18: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
preflash data for WF2 vs. WF4 (u5ka510kr/u5ka5106r) with the Y-CTE loss in % over
800 pixels. Right, histogram of the measured stellar flux for stars used to determine the YCTE.
22
Instrument Science Report WFPC2 2001-002
u5ka5106r_0k_cte_200_500.da
102
2
preflash
1.5
Number
Flux Ratio (WF2/WF4)
200 < DN < 500
1
101
.5
Y-CTE = 20.0 +/- 3.6%
0
-500
0
∆Y
0
-100
500
0
100
200
300
400
500
Counts
Figure 19: Left, the detected flux ratio vs. ∆Y for the F555W filter. The figure shows the
preflash data for WF2 vs. WF4 (u5ka510kr/u5ka5106r) with the Y-CTE loss in % over
800 pixels. Right, histogram of the measured stellar flux for stars used to determine the YCTE.
Figures 16-19 show the Y-CTE measurements between apertures WF2 and WF4
before and after the preflash for two samples of stars, those with counts (flux) of 50-200
and 200-500 counts. A summary of the different Y-CTE values is presented in Table 9.
The comparison between the non- and preflash Y-CTE loss measurements for stars with
counts within the range 50 to 200 counts indicates no improvement in CTE due to preflashing the CCD, while for stars with counts within the range of 200 to 500 counts a 4%
reduction in CTE loss is indicated though the uncertainties are large.
23