Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/wfc3/documents/ISRs/WFC3-2008-30.pdf Дата изменения: Unknown Дата индексирования: Tue Feb 5 15:27:40 2013 Кодировка: Поисковые слова: http astrokuban.info astrokuban

Instrument Science Report WFC3 2008-30

WFC3 TV3 Testing: IR Channel Dark Current
B. Hilbert September 25, 2008

ABS

TRACT

Using data taken during WFC3's Thermal Vacuum 3 (TV3) testing campaign, we have characterized the dark current behavior in the IR Channel, which contained IR-4 (FPA165). The Contract End Item (CEI) Specifications call for dark current to be below 0.4 e-/sec/pixel, with a goal of 0.1 e-/sec/pixel. We measure dark current values well below the spec and goal values. The longest exposure time ramps show dark current values in the 0.045 ­ 0.060 e-/sec/pixel range. Initial dark current calibration files will be created from these TV3 data.

Introduction
WFC3 underwent TV3 testing in the spring of 2008 with the goal of collecting data necessary to create an initial set of calibrations for use after launch. As part of this calibration effort, we collected dark current files using each sample sequence in the IR channel. As this was the first and only ground test data collected with the IR flight detector (IR4), these dark current data were used to characterize the dark current behavior of the IR channel, as well as to create mean dark current ramps. These mean ramps will be used by CALWF3 for data reduction purposes during Servicing Mission Orbital Verification (SMOV).

Operated by the A ssociation of Univ ersities for Research in A stron omy, Inc., for the National A eron autics and Space Adm inistration


WFC3 Instrument Science Report 2008-30

Data
Dark current data were collected using three Science Mission Specifications (SMSs). Each SMS collected full-frame ramps using one of the three flavors of sample sequences. IR01S03 was used to collect RAPID and SPARS ramps, while IR01S04 and IR01S05 used STEP and MIF sample sequences, respectively. Several complete sets of full frame dark current files were collected using each of the two instrument electronics sides (referred to in this study by the main electronics board number, MEB1 and MEB2). Table 1 lists the details of the dark current data collected during TV3 testing. Four iterations of the IR01S03 SMS were completed during TV3. Three of these were completed with the MEB1 electronics, and one with the MEB2 electronics. Similarly, IR01S04 was run 5 times, with four iterations on MEB1 and one on MEB2. IR01S05 was run twice during TV3, once on each MEB. We keep data taken on each side (MEB) of WFC3 separate, so that we can check for any side dependence. All data were obtained at the nominal detector operating temperature of -128.7C. The temperature of the IR detector was constant throughout the dark current data collection to within 0.2oC. Median dark current values are reported below on a quadrant-by-quadrant basis. For this, we follow the previously defined convention for quadrant numbering. Quadrant 1 is defined as the upper left quadrant of the detector. Quadrant numbers increase in a counter clockwise direction. Figure 1 shows a read from a dark current ramp, with the 4 quadrants labeled.

Figure 1: Dark curren t image from the final read o f a 2800 second ramp. The stretch is histogram equaliza tion, with pixel valu es ranging from -0.5 to 2.4 e-/sec/pixel. The circular feature a t the bo ttom of quadrant 2 has been dubbed th e "death star" and is a co llection of poorly performing pixels. These p ixels are ignored in all future calcu lations.

2


WFC3 Instrument Science Report 2008-30

Sample Sequence IR01S03 RAPID SPARS SPARS SPARS SPARS SPARS S S S S S T T T T T E E E E E P P P P P

10 25 50 100 200

No. of Ramps MEB1 / MEB2 9/3 9/3 9/3 9/3 9/3 9/3 12 12 12 12 12 / / / / / 3 3 3 3 3

IR01S04

25 50 100 200 400

IR01S05

MIF600 MIF900 MIF1200 MIF1500 Table 1: TV3 observations useful for char

3/3 3/3 3/3 3/3 acterization of dark current.

Analysis
Prior to dark current analysis, all ramps were run through the WFC3 IDL data reduction pipeline (Hilbert 2004), in order to remove the bias signal and pixel-to-pixel variations in the zero level. We also multiplied by 2.26 e-/ADU, which is the gain value calculated from TV3 data (ISR forthcoming). The first analysis performed on the dark current ramps examines the behavior of the signal up the ramp. For a given ramp, we calculated the median signal in each quadrant of each read, and plotted these medians versus time. These plots revealed a consistent, unexpected trend. The first read in all of the dark current ramps showed an elevated signal. Measured signal decreased between the first and second reads, and then increased at the expected rate after the second read. Figures 1 and 2 show plots of representative signals for the SPARS sample sequences. The elevated signal in the first read of each is apparent. This implies some sort of reset anomaly with the FPA. For the purposes of this study, where dark current rates are reported as a single number for each sample sequence, we ignored this first read in subsequent calculations. As a result, while dark current

3


WFC3 Instrument Science Report 2008-30 measurements reported here are a single number for each sample sequence, the actual dark current behavior is more complex. The dark current correction that will be applied by the CALWF3 pipeline will involve subtracting a sample sequence-specific dark current ramp from the data ramps, rather than simply scaling a single dark current number to the data. This should remove any reset anomaly effects from the data.

Figure 2: Median m easured signal up th e ramp for on e ram p each o f SPARS10, SPARS25, SPARS50, SPARS100, and S PARS200, with ea ch ramp vertically offset from that b elo w b y 6.75 e-. Each ramp sho ws an apparent reset anomaly in the first read, even after th e zeroth read has been subtracted.

Figure 3: Median signal up the ramp for 9 S PARS25 ramps. All ramps sho w the eleva ted signal in th e first read.

4


WFC3 Instrument Science Report 2008-30 Median Dark Current Rates Dark current rates were calculated on a pixel-by-pixel basis in each ramp. For a given pixel, we fit a line to the signal up the ramp (ignoring the first read), and collected the slope as that pixel's dark current rate. We then created a histogram of the dark current rates in each quadrant. The peak position of a Gaussian fit to the histogram was recorded as that quadrant's "median" dark current rate. This will be slightly smaller than the true median of the dark current, due to the long tail of distribution created by the small fraction of hot pixels. A typical histogram and Gaussian fit for a SPARS200 ramp are shown in Figure 3. This process was repeated for all ramps of a given sample sequence, and the median of these values was recorded in Tables 2 and 3. The uncertainties were derived using the effective noise measurements reported in Hilbert (2008). These noise values, which are dominated by the contribution from the 21 e- CDS readnoise, were calculated by combining the readnoise value with the shot noise associated with the dark current signal. The two noise sources were combined following equation 2 in Hilbert (2008), which takes into account the correlation between measured dark current signals in the 16 reads of a ramp. These noise values were then propagated through the line-fitting process, in order to arrive at the uncertainty associated with the best-fit slope (and therefore dark current rate) of a ramp. Results are printed in Tables 2 and 3.

Figure 4: Dark curren t histogram and Gaussian fit for quadrant 4 of a SPARS200 ramp. To improve the fit, we have ignored th e high dark current tail and only fit a Gau ssian do wn to 80% o f the maximum histogram value.

5


WFC3 Instrument Science Report 2008-30 MEB1 Median Dark Current and Uncertainty (e-/sec/pi Quadrant 1 Quadrant 2 Quadrant 3 RAPID -0.008 / 0.257 -0.015 / 0.257 -0.016 / 0.255 SPARS10 0.041 / 0.078 0.041 / 0.080 0.037 / 0.079 SPARS25 0.051 / 0.032 0.045 / 0.032 0.046 / 0.031 SPARS50 0.052 / 0.017 0.047 / 0.017 0.050 / 0.017 SPARS100 0.051 / 0.009 0.047 / 0.009 0.049 / 0.009 SPARS200 0.050 / 0.005 0.046 / 0.005 0.048 / 0.005 STEP25 0.051 / 0.037 0.044 / 0.037 0.042 / 0.037 STEP50 0.047 / 0.020 0.042 / 0.020 0.040 / 0.020 STEP100 0.054 / 0.012 0.047 / 0.012 0.047 / 0.012 STEP200 0.057 / 0.007 0.051 / 0.007 0.050 / 0.007 STEP400 0.058 / 0.005 0.051 / 0.004 0.051 / 0.004 MIF600 0.050 / 0.014 0.045 / 0.014 0.045 / 0.014 MIF900 0.049 / 0.009 0.044 / 0.009 0.044 / 0.009 MIF1200 0.049 / 0.007 0.044 / 0.007 0.044 / 0.007 MIF1500 0.049 / 0.006 0.044 / 0.006 0.044 / 0.006 Table 2: Median IR dark current values for each sample sequence wi WFC3. These data are plotted in Figures 5-7. MEB2 Median Dark Current and Uncertainty (e-/sec/pi Quadrant 1 Quadrant 2 Quadrant 3 RAPID 0.005 / 0.253 0.004 / 0.252 -0.007 / 0.252 SPARS10 0.053 / 0.078 0.049 / 0.078 0.047 / 0.078 SPARS25 0.059 / 0.032 0.054 / 0.032 0.056 / 0.032 SPARS50 0.062 / 0.017 0.057 / 0.017 0.060 / 0.017 SPARS100 0.060 / 0.009 0.056 / 0.009 0.058 / 0.009 SPARS200 0.058 / 0.005 0.054 / 0.005 0.057 / 0.005 STEP25 0.053 / 0.037 0.049 / 0.036 0.047 / 0.036 STEP50 0.055 / 0.020 0.050 / 0.020 0.052 / 0.020 STEP100 0.059 / 0.012 0.053 / 0.012 0.055 / 0.012 STEP200 0.062 / 0.007 0.056 / 0.007 0.058 / 0.007 STEP400 0.063 / 0.005 0.057 / 0.004 0.058 / 0.004 MIF600 0.057 / 0.014 0.053 / 0.014 0.058 / 0.014 MIF900 0.056 / 0.010 0.052 / 0.009 0.057 / 0.009 MIF1200 0.052 / 0.007 0.049 / 0.007 0.055 / 0.007 MIF1500 0.053 / 0.006 0.049 / 0.006 0.053 / 0.006 Table 3: Median IR dark current values for each sample sequence wi WFC3. xel) Quadrant 4 -0.024 / 0.266 0.033 / 0.082 0.042 / 0.033 0.046 / 0.017 0.046 / 0.009 0.045 / 0.005 0.036 / 0.038 0.037 / 0.021 0.043 / 0.012 0.047 / 0.007 0.048 / 0.005 0.043 / 0.014 0.042 / 0.010 0.042 / 0.007 0.041 / 0.006 th MEB1 of

xel) Quadrant 4 -0.006 / 0.252 0.042 / 0.078 0.051 / 0.032 0.056 / 0.017 0.054 / 0.009 0.053 / 0.005 0.040 / 0.036 0.047 / 0.020 0.051 / 0.012 0.054 / 0.007 0.054 / 0.004 0.053 / 0.014 0.051 / 0.009 0.050 / 0.007 0.049 / 0.006 th MEB2 of

6


WFC3 Instrument Science Report 2008-30

Along with median dark current rates, we used the dark current histograms to collect statistics on anomalous pixels in the IR detector. This included pixels with dark current values above the CEI Spec value of 0.4 e-/sec/pixel, dead pixels, and hot pixels (defined as having a dark current above 1.0 e-/sec/pixel). Table 4 shows the results of these calculations for STEP400 and SPARS200, the two longest sample sequences. These are the highest signal to noise dark current data, and give the most accurate count of the variously misbehaving pixels. Roughly 0.7% of the light-sensitive pixels in the IR detector exhibit a dark current above the CEI spec value of 0.4 e-/sec/pixel. This number decreases to roughly 0.4% if the criterion is changed to be 1.0 e-/sec/pixel. Only ~2,000 pixels across the detector appear to be dark (i.e. dark current between 0 and 0.01 e-/sec/pixel). Roughly 9 times that number of pixels displays negative dark current across the 2,800 second ramps. All 4 flavors of bad pixels are distributed randomly and evenly across the area of the detector.

Percentage of Non-nominal Pixels MEB1/MEB2 Above CEI Hot Pixels Dark Pixels SPARS200 Quad 1 Quad 2 Quad 3 Quad 4 STEP400 Quad 1 Quad 2 Quad 3 Quad 4 0. 0. 0. 0. 9 7 5 5 / / / / 1. 0. 0. 0. 0 7 5 6 0. 0. 0. 0. 6 4 3 3 / / / / 0. 0. 0. 0. 6 4 3 3 0. 0. 0. 0. 02 02 01 02 / / / / 0. 0. 0. 0. 03 01 01 02

Negative Pixels 0.4 0.05 0.05 0.2 / / / / 0. 0. 0. 0. 4 04 04 2

1. 0. 0. 0.

0 8 6 6

/ / / /

1. 0. 0. 0.

1 8 6 6

0. 0. 0. 0.

7 5 3 4

/ / / /

0. 0. 0. 0.

7 5 4 4

0. 0. 0. 0.

02 01 01 01

/ / / /

0. 0. 0. 0.

02 01 01 01

0.4 0.04 0.04 0.2

/ / / /

0. 0. 0. 0.

4 04 04 2

Table 4: Fraction of light-sensitive pixels with non-nominal behaviors. The first column represents pixels with dark current values greater than the CEI Spec value of 0.4 e/sec/pixel. The hot pixel column is for those pixels with dark current greater than 1.0 e/sec/pixel. Dark pixels are those with dark current less than 0.01 e-/sec/pixel and greater than zero, and negative pixels are those with dark current values less than 0 e-/sec/pixel.

7


WFC3 Instrument Science Report 2008-30

Figures 5, 6, and 7 show the median dark current for MEB1 for the RAPID/SPARS, STEP, and MIF sample sequences, respectively. The scatter in the dark current values decreases as the exposure times of the ramps increase. Comparison of Figures 5, 6, and 7 with one another, as well as with similar plots for other MEB1 iterations of the SMSs, as well as with MEB2 data confirm the uncertainties presented in Tables 2 and 3. One result to note is that the MIF data consistently show a lower median dark current than the SPARS and STEP sample sequences. In most cases, these values are within the listed uncertainties.

Figure 5: Median dark current in each quadrant, from the RAPID and SPARS sample sequences. Ea ch blue p lotting symbol represents a difference sample sequence. Da ta were taken in order of in creasing exposure tim e. X's (not always visible due to large scatter) show th e RAPID dark current. Squares are SPARS10, stars are SPARS25, p lus signs are S PARS50, dia monds are SPARS100, and triangles are SPARS200. Each po int represen ts the med ian dark current for a sing le ramp. No te the decreasing scatter as the exposure times in crease. The red stars sho w the temperature of the detector during the test.

8


WFC3 Instrument Science Report 2008-30

Figure sq u a re ra m p s, has the

6: Median dark current for th e STEP sample sequen ces. Th e star, plus, diamond, triangle, and represent the mea sured dark curren t in the STEP25, STEP50, STEP100, STEP200, and STEP400 respectively. As with Figure 4, we see tha t quadrant 4 has the lo west dark current, and quadrant 1 highest.

Figure 7: Same as Figure 5, bu t for the MI F sample sequen ces. Stars, pluses, diamonds, and triangles represent th e m edian dark curren t for MIF600, MIF900, MI F1200, and MI F1600 ramps.

9


WFC3 Instrument Science Report 2008-30

Conclusions
The IR Channel in WFC3 meets (with considerable margin) both the CEI Spec 4.8.4 and goal for dark current. Only 0.7% of the active pixels in the IR channel fail to meet the CEI Spec of 0.4 e-/sec/pixel. The median dark current measured in TV3 is between 0.045 and 0.060 e-/sec/pixel. Only 0.005% of light-sensitive pixels are nonfunctional (cf. CEI 4.8.11.3) due to their dark current being greater than 100 times the mean dark current, i.e. > 6 e-/sec/pixel. These dark current levels have been measured ignoring the initial read of each dark current ramp, which exhibits an elevated signal that we attribute to a reset anomaly of the FPA. The effects of this anomaly will be removed from WFC3-IR data by CALWF3 in the data reduction process, where dark current ramps are subtracted on a read-by-read basis from data ramps.

References
Hilb ert, B., WFC3 TV3 Testing: IR Channel Read No ise. WFC3 ISR 2008-25. http://www .stsci.edu/hst/wfc3 /documen ts/ISRs/WFC3-2008-25.pdf July 2008. Hilb ert, B. Basic ID L Data Reduction Algorithm for WFC3 IR and U VIS Channel. WFC3 ISR 2004-10. http://www .stsci.edu/hst/wfc3 /documen ts/ISRs/WFC3-2004-10.pdf 10 June 2004. Petro, L. and T. Wh eeler. New IR D etector Sa mple Times. http://www .stsci.edu/hst/wfc3 /documen ts/ISRs/WFC3-2006-06.pdf 2 Oct 2006.

10