HRI block diagram (Ref. 3, used with permission); FPA, focal plane array; TLM, telemetry; LVPS, low-voltage power supply; LVDS, low-voltage differential signaling interface.
MRI/ITS block diagram (Ref. 3, used with permission); acronyms are defined in Fig. 1 caption.
CCD architecture (Ref. 3, used with permission).
Generalized timing patterns (Ref. 3, used with permission). (a) Timing pattern for a single exposure when the light blocker is used between each exposure. If exposures are commanded, then this pattern is repeated times. (b) Timing pattern when the light blocker is not used between each exposure and there is no additional commanded integration time beyond the minimum. In the second case the number of exposures, , can be commanded as 1. Both figures show when the time stamp saved in the image header is collected with the arrow and label “VTC.” In both cases the time stamp is collected at the end of the exposure period.
IR focal plane array architecture (Ref. 3, used with permission). (a) Architecture of the full IR FPA; (b) spectrometer mapping onto the FPA. Only Quads II (A) and IV (B) are used; Quads I and III are not used.
IR focal plane readout timing. Timing for two different readout modes, (a) interleaved and (b) alternating, are shown.
A full-frame, HRI vis image taken shortly before impact, displayed with the FITS convention. This orientation reproduces a true sky image (ecliptic north is to the right). The first and last bytes are those read from the FITS file and are not connected with the order of detector readout. Quadrants A–D noted throughout this paper are labeled in the image.
A full-frame, MRI image taken at nearly the same time as the HRI vis image in Fig. 7. Displayed with the FITS convention, a true sky image is reproduced (ecliptic north is to the right). The first and last bytes are those read from the FITS file and are not connected with the order of detector readout. Quadrants A–D noted throughout this paper are labeled in the image.
A full-frame, HRI-IR image taken shortly before impact, displayed with the FITS convention. For this FITS display, the wavelength increases as the fastest-varying axis increases to the right. The slowest-varying axis is the spatial direction along the slit. The first and last bytes are those read from the FITS file and are not connected with the order of detector readout. IR quadrants A and B noted throughout this paper are labeled in the image.
HRI CCD encounter LUTs for converting DNs to DNs.
MRI CCD encounter LUTs for converting DNs to DNs.
ITS CCD encounter LUTs for converting DNs to DNs.
HRI IR FPA encounter LUTs for converting DNs to DNs.
Central strip of the MRI alignment image of star cluster NGC 3532, taken after launch, heavily stretched to show dimmest star images, noise, and bias differences in the four quadrants. Note the triangular shape of the heavily overexposed images due to stressing of the optics by the mounts and six offset diffraction spikes due to the nonradial secondary mount spider vanes.
PSF FWHM in MRI quick alignment images for SNR larger than 100 (1296 stars) as a function of distance in pixels from the center of the detector.
The PSF FWHM of the MRI after bakeout for 707 stars with peak SNR greater than 100 as a function of distance in pixels from the center of the detector.
The PSF FWHM of the ITS from the quick alignment images of 1638 stars with peak SNR greater than 100 as a function of distance in pixels from the center of the detector.
The PSF FWHM of the ITS star images after bakeout for 710 stars with peak SNR greater than 100 as a function of distance in pixels from the center of the detector.
The measured PSFs through the nine filters of the HRI, as determined by a short exposure of a bright star (to minimize pointing drift during the exposure). The second column displays the logarithmic stretch of a single image of the PSF intensity to show in detail the extendedlow-intensity part of the PSF. The third column shows a single image of the measured PSF intensity, scaled to linearly display minimum brightness as black and maximum as white. The fourth column displays a subpixel sampled drizzled PSF derived by combining ten nearly randomly positioned images of the stars 16 Cyg A and B taken on 14 June 2005 using the STSDAS drizzle function (see text). It appears that the PSF for the Clear-6 filter is contaminated by significant drift during at least one of the exposures, presumably the very first one, which was the reference image.
An example of the deconvolution step in the calibration pipeline applied to a HRI image through the clear filter in wheel position 6. Left panel shows the original image; right panel shows the deconvolved image. Small-scale features that are blurred in the left panel can be distinguished in the right panel. Also note that noise is amplified by deconvolution. The horizontal dark line at the center of the deconvolved image is a deconvolution artifact caused by a very slight dark line (1%–2% level) in the original image due to flat-field residual at the edge between the upper and lower pairs of CCD quadrants.
HRI vis response linearity. Results for the various quadrants are offset by 1000 DN for visibility.
Residuals to a linear fit of the HRI vis response.
Photon transfer data for HRI vis.
Dark current as a function of CCD temperature for quadrant C of the HRI instrument. The model is the solid line and the data are the squares.
exposure, dark subtracted, Gaussian stretched image showing frame transfer smear.
Version of Fig. 25 corrected for frame transfer smear using parallel overclocked pixel value.
(Color) Absolute spectral irradiance of selected photometric standard stars.
Camera response during inflight star calibrations ratioed to those obtained in May 2005 (HRI, top; MRI, bottom). When error bars cannot be seen, they are smaller than the data symbol.
Ratios of actual inflight stellar response rates to those predicted using the prelaunch camera models.
Ratios of actual inflight stellar response rates to those predicted using the adjusted camera models.
Modeled camera response rates to an extended gray source of radiance of : (a) HRI vis, (b) MRI, and (c) ITS.
Dropoff in ITS signal rate just prior to impact.
Bad pixels for HRI, MRI, and ITS in mode 1. The figure is not to scale for illustration purposes. See the text for the definition of marks .
HRI (top), MRI (middle), and ITS (bottom) flat field for mode 1 and filter 1 (no filter for ITS). Each flat field is a frame covering all four quadrants. Data span the range from 0.8 (black) to 1.2 (white) in this display with the mean response normalized to 1.0. The upper left quadrant corresponds to quadrant A in the text, the upper right quadrant to B, the lower left quadrant to C, and the lower right quadrant to D.
Profile of HRI (top), MRI (middle), and ITS (bottom) flat field corresponding to Fig. 34 along row 250.
Profile of a stimulator image before (left) and after (right) flat-field correction for HRI (top), MRI (middle), and ITS (bottom). The flat-field correction reduces the pixel-to-pixel variations from to rms.
Comparison of the profile of a coma image with the new flat field including the stimulator pixel-to-pixel relative-response variations and the TV4 flat field.
An example of a recurring low-level noise sometimes seen in all three CCD imagers. A segment from one quadrant of a HRI CCD image is shown below, and two DN profiles across the image are plotted above. The duration of the noise corresponds to about two times the readout time for one row. Profile (a) is a row with the noise, and profile (b) is a row without the noise. Profile (b) is offset by DN for clarity.
CCD room-temperature dark current pattern. The center of a room-temperature HRI image are shown.
Histograms of HRI CCD STIM images showing uneven ADC bit weighting: (a) full histogram, (b) expanded view.
Saturated image of Canopus before (top) cross-talk correction and after (bottom) cross-talk correction. The electronic ghosts nearly disappear.
The stray light pattern for HRI. Canopus is shown scaled as a reference. There are five images surrounding the boresight image. Each image has an exposure time of . The Canopus image saturates with an exposure time of about . The secondary mounting structure (the spider) is the major contributor to the observed stray light.
The same HRI data as in Fig. 42 converted to polar coordinates with 1° increments.
One of the bright columns from the HRI polar plot in Fig. 43 (at 94°) normalized with respect to the total Canopus signal showing the worst-case HRI scattered light response to a point source outside the FOV.
The stray light for MRI. The moon is shown scaled as reference. There are five images surrounding the boresight image. Each image has an exposure time of . The moon saturates with about exposure time. The MRI telescope is similar to the HRI, so similar stray light patterns can be seen.
The same MRI data as in Figure 45 converted to polar coordinates with one degree increments.
One of the bright columns from the MRI polar plot in Fig. 46 (at 190°) normalized with respect to the total lunar signal.
HRI vis in-field scattered light shows the comparison between the clear filter and the IR filter. Total stray light is expressed as a fraction of the total signal within the masked area. Each image is .
MRI vis in-field stray light. Each image is .
MRI in-field scattered light response down column 650. Signal level is in arbitrary units with all filters normalized to the same signal amplitude.
CCD saturation and charge conservation. The plot shows an idealized cross section of signal of an unsaturated star (blue line without symbols) and a saturated star (green line with triangles) in the case where the physical limitation of the CCD full well is above the range of the ADC. The signal level is shown relative to the maximum value digitized by the ADC. The resulting signal level in the CCD is also shown (red line with squares). Any signal above the dashed line is digitized as 16 383 counts. The figure in the upper right of the plot is a HRI image of Canopus saturated about , showing the charge bleeding up and down columns.
The number of cosmic ray events per (top) and mean values of DN (bottom) vs the number of pixels affected by an event at a peak of solar activity (crosses) and for out-of peak activity (stars) on dark MRI images with .
Calibrated MRI image 6 002 420 . From left to right: initial image, pixels recognized as CRs by DI̱CRREJ, CRFIND, IMGCLEAN, and RMCR (at and ).
Central part of calibrated MRI image 9 000 907 . From left to right: initial image, pixels recognized as CRs by DI̱CRREJ, CRFIND, IMGCLEAN, and RMCR (at and ).
Calibrated HRI image with a large star (at maximum radiance ). From left to right: initial image, pixels recognized as CRs by DI̱CRREJ, CRFIND, IMGCLEAN, and RMCR [at and ].
Detailed layout of the SIM (Ref. 3, used with permission).
Relative boresight alignment of the HRI vis and IR spectrometer as measured prelaunch (boresight offsets in figure are not drawn to scale; only the central subarea of HRI and the central 10% of IR spectrometer slit are shown).
Relative locations of the HRI-IR slit with respect to the MRI and HRI fields of view measured in flight (boresight offsets in the figure are not drawn to scale; only the central subarea of HRI, the central subarea of MRI, and the central 25% of IR spectrometer slit are shown).
IR FOV during scan of 47 Tuc. The FOV is 100 scans wide and 512 physical pixels tall. To have matching spatial scale in both directions, the image has been stretched by a factor of 2 in the scan (horizontal) direction.
An extracted subimage of the MRI FOV containing the FOV of the IR spectrometer during its scan of 47 Tuc.
Difference in across-slit IR and MRI star positions as a function of along-slit position.
The inflight variation of the centers of Gaussian fits to the image of the calibration star Beta Hyi as a function of IR spectrometer pixel (column) number. The total variation from one edge of the IR detector to the other is almost exactly one physical pixel.
The inflight variation of the centers of Gaussian fits to the image of the calibration star Beta Hyi as a function of IR spectrometer wavelength. The total center variation in row numbers from minimum to maximum observed wavelength is almost exactly one physical pixel, and the variation is very close to linear with respect to wavelength.
The IR spectrometer spatial resolution (FWHM of the Gaussian fit) as a function of pixel (column) number for the 10 May 2005 observation of calibration star Beta Hyi. Note that the spatial resolution is approximately constant at a value of FWHM for pixels 6 through about 700 and then rises approximately linearly to nearly for pixel numbers around 1000 due to optics diffraction.
The IR spectrometer spatial resolution (FWHM of the Gaussian fit in physical pixels) as a function of wavelength for the 10 May 2005 observation of calibration star Beta Hyi. Note that the spatial resolution is dominated by the defocused PSF of the HRI telescope for wavelengths shorter than about and by the diffraction limit of the telescope for wavelengths longer than about . One physical pixel is , so is about .
Along-slit traces of a star image in the IR spectrometer for 3.5, 4.0, and (top to bottom). Also plotted are Gaussian fits; each data set is normalized by the maximum value of the Gaussian fit and then respectively displaced by 0.5, 0.25, or 0.0 intensity units upwards. The data points are joined by lines, and the Gaussian fit values are dotted.
The number of IR spectrometer DN measured in the left quadrant A using a resistant mean vs integration time for each imaging mode is shown. Imaging modes are identified by color, and the month of the event is defined by linestyle.
The IR spectrometer response rate is calculated using the data shown in Fig. 67 for quadrant A. Plot A shows the subframe imaging modes with a higher response rate. Plot B was created by applying the correction ratios of 0.95 and 0.92 for BINSF1 and BINSF2, respectively. As in Fig. 67, imaging modes are identified by color, and the month of the event is defined by linestyle. The nonlinearity in the IR response becomes more apparent in these curves.
The IR spectrometer nonlinearity equations are derived after applying the subframe scaling ratios, and all data are normalized to at 5000 DN. Plot A shows the data from quadrant A, and plot B shows the data from quadrant B. Data from each month are represented by a different color.
IR spectrometer photon transfer data in the unbinned (top) and binned (bottom) modes.
IR spectrometer background level as a function of prism temperature for mode 4 ground-based data at zero integration delay. (a) Background as a function of the prism temperature. Symbols indicate the measured values of the total background signal, while the line indicates the computed values for the best fitting contribution from the instrument glow. (b) Residual background as a function of the FPA temperature (Ref. 3, used with permission). The symbols indicate the residual signal after the computed contribution from the instrument glow has been removed. The line shows the best fit for the dark current contribution.
IR spectrometer background rate as a function of the prism temperature for the in-flight data. Symbols indicate the measured values of the total signal, while the line indicates the computed values for the best fitting modeled contribution from the instrument glow.
Fractional residuals in the background level of inflight IR spectrometer data after the computed FPA dark current and instrument-glow contributions are removed. Residuals for different modes are shown in different colors.
Background level in a TV2 sequence of 32 IR spectrometer images. The level in the first frame of the sequence is much higher than those in the following images, though subsequent frames continue to drop with time throughout the sequence.
Normalized IR spectrometer background level as a function of the intersequence gap for the first image in a sequence in mode 1. (a) The full range of observations showing a constant overshoot level (dashed line) for longer gaps. (b) Blowup of the first , showing the linear rise with intersequence gap time. The diamonds are measurements obtained for mode change from 2 to 1, and the squares are for no mode change (1-1). Linear fits to the rising portion in each of the two cases, indicated by the dashed lines, have the same slope but are offset in time.
Falloff in the IR spectrometer background level after the first frame in mode 1. The dotted lines and diamonds show the falloff of the background when the first-frame overshoot reached the plateau, while the solid lines and squares depict the sequences in which the first frame did not reach the plateau level.
Example comet spectrum showing the effects of improper IR spectrometer background removal. The heavy line shows the result for the “optimum” background level found as described in the text. The upper thin line shows the result for a 15% decrease in the background level, and the lower thin line shows the result for a 5% increase. Around the effect is , while at the shortest wavelengths, it overwhelms the signal.
Wavelength vs unbinned column number for the IR spectrometer for a bench temperature of .
Absolute spectral irradiance of selected photometric standard stars.
IR spectrometer total response during inflight star calibrations: (a) binned mode, behind antisaturation filter except for minicals; (b) binned mode with expanded scale; (c) unbinned mode, all outside antisaturation filter; (d) unbinned with expanded scale.
IR spectrometer inflight calibration curves; binned behind antisaturation filter on the top; unbinned outside antisaturation filter on the bottom.
Best-fit averaged IR spectrometer calibration curves; behind antisaturation filter on the top; outside antisaturation filter on the bottom.
IR spectrometer antisaturation filter spectral transmission curve: black, inflight measurement; red, prelaunch measurement scaled to fit the inflight measurement.
Residual ratio between the component-level IR spectrometer calibration curve (excluding beamsplitter ripples) and inflight measured curve.
Best-fit IR spectrometer calibration curves and their ratio to the inflight measurements.
Bad-pixel map for IR spectrometer unbinned full-frame modes 4 and 6 based on data acquired during the June science calibration.
Flow chart defining the algorithm for deriving the IR spectrometer flat field from images of a spatially uniform source radiance.
Flat-field IR spectrometer images obtained during TV4 with the integrating sphere through the fused-silica window; brightnesses are normalized to the mean value. From top to bottom on the left are modes 5, 3, 2, and 1 flats, and on the right are modes 6 and 4. Wavelength varies from from left to right in each image. The residual vertical structures on the right side of each frame result from regions of minimal input flux to the instrument due to atmospheric and fused-silica window absorptions.
Flat-field IR spectrometer images obtained during TV4 with the W lamp through the ZnSe window; brightnesses are normalized to the mean value. From top to bottom on the left are modes 5, 3, 2, and 1 flats, and on the right are modes 6 and 4. Wavelength varies from from left to right in each image. Residual structures, some likely due to the illumination pattern of the W lamp reflecting off a rough gold surface, are apparent in both the dispersion (horizontal) and spatial (vertical) directions.
Application of IR flats to a TV4 HRI-IR Ar-lamp image. Signal from several selected rows is plotted. Squares, original data. Diamonds, data after flat-field division. Little to no improvement in the scatter of the data results from application of the TV4 flat.
Flat-field IR spectrometer image obtained from the lunar scan in mode 4; brightnesses are normalized to the mean value. Compared to the TV4 W-lamp/ZnSe window flat, there is almost no structure found in the long-wavelength (rightmost) regions. The horizontal green/red lines are at the boundaries of the central region of the array that is inside the antisaturation filter; some low-frequency structure is apparent at these boundaries. Low SNR at the shortest wavelengths and the longest wavelengths behind the antisaturation filter limits the usefulness of the flat at wavelengths ( 475) and ( 950) behind the antisaturation filter.
Inflight IR spectrometer dark flat field produced using the thermal emission of the optical bench as the illumination source; brightnesses are normalized to the mean value. There is significant spatial structure in the derived flat compared to the TV4 and lunar measurements.
IR spectrometer linearity flat derived from a series of dark measurements in a linearity test. The low-frequency structure imposed by the instrument self-illumination pattern, seen in the dark flat image, has been removed by normalizing the response in each pixel to the median signal of its neighboring pixels in a area. The result leaves the high-frequency variations produced by the differences in gain from pixel to pixel. Relative-response values shown here range from 0.8 (black) to 1.2 (white).
Histogram of IR spectrometer dark image showing modest uneven ADC bit weighting.
IR spectrometer response profiles across a knife edge from the ground-based in-field scattered light test. Signal is detected away from the knife edge at a level of 3%–4%. Column numbers represent increasing wavelengths from across the detector. No change in scattered light vs wavelength is seen.
Gaussian-stretched, dark-subtracted IR spectrometer images of the Moon (composite RGB: 1.15, 1.90, ) showing scattered light around the lunar limb and terminator from two different observations. The arrows indicate the locations of the profiles plotted in Fig. 97.
Horizontal (top) and vertical (bottom) IR spectrometer profiles of scattered light off the lunar limb (see arrows in Fig. 96). Scattered light levels of 1% are detected from the limb in the horizontal direction and of 2% at from the limb in the vertical direction. The scattered light profiles at all three wavelengths (2.5, 3.0, and ) behave similarly.
Gaussian-stretched, dark-subtracted, composite IR spectrometer image of three bands near from a scan in the direction of the star Canopus. As discussed in the text, a ghost image is clearly detected towards the end of the scan (bottom of image).
Vertical profiles across the IR spectrometer Canopus scans. Top: Scan in the direction showing a ghost image at 13 frames after the star image. Bottom: scan in the direction showing a ghost image at 36 frames after the star image. As described in the text, when corrected for differences in scan rates, the position and magnitude of the ghost images are very similar.
The number of clusters detected as CR events per vs the number of pixels affected for out-of-peak solar activity in a typical calibrated dark IR spectrometer frame ) with .
A flowchart describing the image processing pipeline used to calibrate Deep Impact images. Note that some modules are not applied to all instruments.
HRIvis data compression LUT loaded at launch.
MRI data compression LUT loaded at launch.
IR spectrometer data compression LUT loaded at launch.
IR spectrometer data compression LUT loaded on 26 May 2005.
Deep Impact instrument suite summary.
Deep Impact filter characteristics (Ref. 3, used with permission).
Visible (CCD) imaging modes (Ref. 3, used with permission).
IR spectrometer operating modes (Ref. 3, used with permission).
Prelaunch calibrations of the DI instruments.
Summary of inflight tests and calibrations using the DI instruments.
Focal-length solutions for each HRI filter position.
Image frame misalignments between the MRI and HRI.
Best-fit gain factor and read noise for each quadrant of each vis camera.
Bias (DN) for modes 7 and 8, where there are no serial overclocked pixels, and values of the parameters and (eV) used for the dark current model (see equation in text).
Radiometric conversion constants and effective wavelengths for each filter in each vis camera.
Comparison of encoding step size to signal shot noise for VIS data compressed using various LUTs.
Cross-talk ratios for HRI, MRI, and ITS. The ratio is the fraction of the primary signal that is produced in another quadrant.
IR spectrometer focal plane and optical bench temperatures from the cruise calibration events.
IR spectrometer linearization equations calculated for both quadrants with and without the July calibration data. DN value; DN value. Differences in the polynomials are less than 0.15% below 11 000 DN.
Width of spectral lines observed in the IR spectrometer.
Number of IR spectrometer bad pixels identified using unbinned, full-frame dark data from ground calibrations (TV2 and TV4) and inflight calibrations in 2005.
IR spectrometer bad-pixel maps used in the pipeline.
Comparison of encoding step size to signal shot noise for IR spectrometer data compressed using various LUTs
Noise parameters determined in ground tests of all instruments.
Article metrics loading...
Full text loading...