INTRODUCTION WARNING: This manual is written in the present tense. At the release of this edition (February 1995) direct imaging and the cross-dispersed grisms were available, but the long slit JHK grisms and imaging polarimetry were not available. Please contact Peter McGregor on (06) 249 0228 or peter@mso.anu.edu.au for information on the current availability of functions discussed in this manual. This manual describes the operation of CASPIR, the Cryogenic Array Spectrometer/Imager on the ANU 2.3 m telescope at Siding Spring Observatory. CASPIR uses a Santa Barbara Research Center (SBRC) CRC463 256 x 256 InSb detector array to provide direct imaging and spectroscopic capabilities in the 1-5 micrometer wavelength range. Two direct imaging focal plane scales of 0.5 arcsec/pixel and 0.25 arcsec/pixel are available, as well as long slit J, H, and K grisms giving two pixel resolving powers of ~500 through a 1 arcsec x 128 arcsec slit, and IJ, JH, and HK cross- dispersed grisms giving two pixel resolving powers of ~1100 through a 1 arcsec x 15 arcsec slit. Coronograph and imaging polarimetry functions are also available. The AAO IRIS Users Manual contains valuable information about observing technique with infrared array cameras which is equally applicable to CASPIR. I recommend that CASPIR users also be familiar with the IRIS manual. Nomenclature for the array readout parameters is common between IRIS and CASPIR, so many of the concepts discussed in the IRIS manual also apply to CASPIR. However, the IRIS and CASPIR readout methods are numbered differently. The CASPIR array is quite robust. It is not damaged by exposure to strong infrared illumination, and has little remnance. However, four warnings are in order: o THE ARRAY IS STATIC SENSITIVE SO UNDER NO CIRCUMSTANCES SHOULD A USER DISCONNECT THE CASPIR DEWAR FROM THE ACE2 DRIVE ELECTRONICS o THE HYBRID DETECTOR ARRAY CAN DELAMINATE AND THE CRYSTAL OPTICS MAY CRACK IF TEMPERATURE CYCLED FASTER THAN 20 DEGREES K/hour. THE DEWAR HAS BEEN DESIGNED TO PASSIVELY TEMPERATURE CYCLE AT THE MAXIMUM SAFE RATE. UNDER NO CIRCUMSTANCES SHOULD THE TEMPERATURE CONTROLLER BE USED TO HEAT THE ARRAY OR THE CAMERA MORE QUICKLY! o THE ARRAY MUST BE STORED UNDER VACUUM AT ALL TIMES TO AVOID SURFACE CONTAMINATION. DO NOT BACK-FILL THE DEWAR WITH ANY GAS o HIGH PRESSURE HELIUM HOSES ARE ATTACHED TO THE DEWAR. DO NOT ROTATE THE INSTRUMENT ROTATOR OUTSIDE THE RANGE -270 TO +110 DEGREES The Latex source files for this manual can be found in the directory ~peter/latex/manuals/caspir on the MSO Sun network. The file manual.ps in that directory contains a postscript version of the manual. This file can also be obtained via anonymous ftp by typing: ftp merlin.anu.edu.au username: anonymous password: type your internet id cd pub/peter get manual.ps bye This manual is also available on the MSSSO WWW home page (http://meteor.anu.edu.au/home.html) A CASPIR users email group exists for occassional dissemination of information about the instrument. Contact Peter McGregor (peter@mso.anu.edu.au) for inclusion in this list. A Detailed Look at the Hardware System Overview CASPIR operates within the environment common to all infrared instrumentation on the 2.3 m telescope. All mechanical functions are controlled from MOPRA through a DECNET link to the LSI-11/23 located in the Cassegrain Instrumentation Rack in the Nasmyth Lab, and from this to the instrument control subrack mounted on the Instrument Mounting Box (IMB). The detector array clocks, biases, and signal processing are performed by the SBRC Array Control Electronics (ACE2) also mounted on the IMB, close to the CASPIR dewar. The ACE2 is controlled directly by MOPRA through a 9600 baud RS-232 connection Data from the four detector output channels are digitized in the ACE2 using four 16-bit, 500 kHz Burr- Brown ADCS, and serialized using four transputer Link Adaptors. The serial data are transmitted from the Cassegrain focus to the Nasmyth Lab where four T800 transputers receive the data and process it as necessary. When the requested integration sequence completes, the data are transferred to MOPRA through a transputer link to Q-bus interface and are displayed on the workstation screen and stored on disk. The detector array temperature is controlled by a commercial controller located in the Nasmyth Lab. MOPRA communicates with this controller through an RS-232 line. The CASPIR Dewar CASPIR is a cryogenic reimaging camera with a 50 mm long, 10.4 mm diameter collimated beam section. The camera body is cooled to ~60 K by the first stage of a closed cycle helium refrigerator, and the detector array cooled to ~32 K by the second stage of the cooler. The dewar incorporates a novel design which uses five 16-position annular wheels mounted coaxially around the cooler to produce a compact vacuum system. The wheels are driven by motors located on the dewar base plate. The CASPIR dewar mounts on port A of the IMB. The rotatable dichroic mirror in the IMB directs the f/18 telescope beam to the dewar. The dewar window is a Sapphire/CaF2 doublet which acts as field lens to image the telescope exit pupil (the secondary mirror) onto an internal cold stop. A cold gold-coated mirror then directs the beam down, parallel to the dewar axis. The telescope focus is located immediately below this mirror at the Aperture Wheel. The Aperture Wheel contains baffles for the 0.5 arcsec/pixel and 0.25 arcsec/pixel focal plane scales, a range of slits for the grisms, coronograph masks, and the field mask used for imaging polarimetry (Appendix, Table 12). The diverging beam then passes to a fixed MgO/CaF2 doublet, collimator lens which produces the collimated beam section. Immediately below this, in the collimated beam, is the Upper Filter Wheel which contains the filters listed in Table 13 of the Appendix. Next is the Utility Wheel which is located at the pupil plane. This wheel contains the direct imaging cold stop, the MgO/CaF2 doublet slow camera lens for the 0.25 arcsec/pixel focal plane scale, the six grisms, and the Wollaston prism polarimeter analysers (Appendix, Table 14). Below the Utility Wheel is the Lower Filter Wheel which contains the filters listed in Table 15 of the Appendix. Note that some of the broadband filters require blocking filters located in the Lower Filter Wheel. Both filter wheels contain clear positions. The detector array should not be exposed to optical light while cold, as the software prevents both clear position being selected at the same time. Note also that the Lower Filter Wheel is located in an f/10.4 converging beam when the slow camera is used, so some refocusing between filters may be required. The final wheel is the Lens Wheel which contains the MgO/BaF2 fast camera lenses for the 0.5 arcsec/ pixel focal plane scale. These are rotated out of the beam when the slow camera is used. The detector array is located at the lower end of the dewar at the camera focus. The Array and Its Read-out Methods The detector is an SBRC CRC463 256x256 InSb array which is sensitive from ~0.9 micrometers to ~5.5 micrometers. The array has four output channels corresponding to four interlaced columns (12341234...) The array is controlled by the SBRC ACE2 drive electronics which is mounted close to the CASPIR dewar. Communications with the ACE2 is through an RS-232 connection to MOPRA. The CRC463 is a hybrid device in which the InSb detector material is bump-bonded to a silicon multiplexer through indium bumps. The multiplexer is a switched FET read-out device, which operates differently to a CCD. Each pixel (or unit cell) contains four FETs; the two row select FETs and the reset FET are switches which can be thought of as closed when activated. The fourth FET acts as a source-follower amplifier which continuously samples the voltage on the detector node without affecting its value. Vgg provides a load for the unit cell source-follower FET. This load FET is located in the column biasing circuitry. The output FET acts as a second source-follower amplifier with its external 10 KOhm load resistor in the ACE2 electronics rack. The two column select FETs also act as switches. Pixels in the array are sequentially addressed by pulsing column and row shift registers which activate the column and row select FET switches. Once a pixel is selected, the voltage of the detector node can be non- destructively read via the source-follower amplifier signal train, and the detector node voltage can then be optionally reset to Vdduc by activating the reset FET switch by pulsing the Phirst clock line. Note that the stored charge is not transferred across the array like a CCD. Instead each pixel is sequentially reset and read. This results in a time delay in the integration window across the array of one frame readout time between the first and last pixels. The array readout scheme permits the use of a variety of readout methods which are now described. Readout Method 1: Fast Sampling In the fastest readout method, speed is considered more important than accuracy so we take only one sample per pixel and reference this voltage to electrical ground (Vss). The unit cell is reset, integrated, and sampled once at the end of the integration ramp. This method should be used for imaging in the 3-4 micrometer band and at M where the high thermal background flux significantly fills the detector wells in of order the frame readout time. In this situation, the dominant noise source is photon shot noise from the background flux, so readout noise is not an issue. The minimum readout time for this method is 0.2 sec. Readout Method 2: Relative Sampling Under less extreme background conditions, significant improvement in stability can be made by referencing the signal level to the reset level, instead of electrical ground. This is done in readout method 2. Note that the detector node voltage jumps when the reset FET is switched off, and this pedestal level is not removed in readout method 2. The pedestal level is different for each pixel in the array so this imprints a pedestal structure on the image, which is difficult to remove completely. The uncertainty in the value of this pedestal is known as kTC noise. For direct imaging through broad band filters with high sky levels, the dominant noise source is still the photon shot noise of the background flux so readout method 2 gives acceptable performance. The minimum readout time for this method is 0.3 sec. Readout Method 3: Double-Correlated Sampling The pedestal structure can be removed using a readout method which references the detector node voltage at the end of the integration to the voltage at the beginning of the integration. In principle, this readout method is susceptible to electrical 1/f noise since it differences two samples separated in time by the duration of the integration. In practice, this is unlikely to be an important noise source. This is the preferred readout method for broad band imaging because it does not imprint the pedestal pattern on the data. The minimum readout time for this method is 0.4 sec. Readout Method 4: Triple Correlated Sampling The potential 1/f noise problem with readout method 3 can be overcome, at the expense of two more reads, by referring both the start and end reads to their respective reset levels. Naively, this should be the most accurate readout method to adopt. However, we must now delve into the more obscure operating characteristics of the CRC463 array to see why other effects dominate. Readout Method 5: Fowler Sampling The FET switches in the CRC463 multiplexer are not perfect switches as assumed above, but instead have finite gate capacitances that act as sinks of charge that would otherwise remain on the detector node capacitance. The reset pedestal (i.e., the amount the detector node voltage jumps by when the reset is taken off) is due to a redistribution of charge from the detector node capacitance to the reset FET gate capacitance that occurs when the the reset FET gate voltage (i.e., the reset clock voltage) moves positive to switch the reset FET off. This amounts to ~100 mV of lost detector reverse bias (or well depth). Similarly, a further ~400 mV of detector bias is lost when the row and column select FETs are switched off to deselect the pixel. This constitutes a movement of ~500 mV of charge off the detector node compared to the normal operating detector reverse bias that remains of only ~200 mV. The true pedestal is therefore significantly larger than the reset pedestal seen if each pixel is reset and read on one pass through the array. Fowler & Gatley (1990, ApJ, 353, L33) show that the read noise can be reduced by performing multiple nondestructive passes through the array at the beginning and end of the integration ramp. By resetting each pixel on one pass through the array, and sampling the detector node voltage on subsequent passes through the array, the true pedestal is removed from the data. Each time a pixel is selected, charge is redistributed from the row and column select FETs back onto the detector node capacitance. Fowler claims that the read noise is predominantly due to the kTC noise associated with this charge redistribution. By performing multiple non-destructive passes through the array at the beginning and end of the integration, the read noise is reduced by the square root of the number of passes. Readout Method 5 implements Fowler sampling in this way. The number of reads at each end of the integration is set by the FNDR parameter (CASPIR/FNDR=...). For applications where low read noise is required, at the expense of increased frame readout time, method 5 is the preferred readout method. This is likely to be the case when using the grisms. Well Depth Considerations The detector node capacitance for the CRC463 array is ~0.06 pF. The well depth is then proportional to the actual reverse bias voltage across the detector; q = CV. Consequently, a detector reverse bias of ~160 mV is required for a well depth of 60,000 e. Our array has an odd-even column effect which causes the applied detector reverse bias on even columns to be lower than that on odd columns. This means that even columns have well depths ~80% smaller than odd columns. Saturation becomes severe at signal levels >= 6500 ADU, corresponding to a well depth of 58500 e-. Linearity Correction When the bias-subtracted linearity data are plotted as signal rates it is apparent that the CASPIR array has a quadratic non-linearity that must be allowed for during data reduction. The best description of this non-linearity is currently given by the equation: Linear Counts = Raw Counts + 6.4 * 10^-6 * (Raw Counts)^2 It is possible that different quadratic coefficients apply to different pixels Further investigation of this effect is required. Setting Array Voltages It should not be necessary for users to alter array voltages. However, the following description of the procedure is included for reference, and to describe some of the cosmetic features that indicate that adjustment is necessary. Voltage adjustment is best done using the Idle Display and readout method 3. Ensure that the array has stabilized at its operating temperature of 32 K. All voltages are negative, but absolute values are discussed here, i.e., increasing a voltage means make it a bigger negative number. Look at something bright (the dome) with 0.5 arcsec pixels and the 5 arcsec x 15 arcsec grism slit so that the illuminated pixels saturate. Look for a dark vertical stripe down the full length of the array at the location of the slit. If this is seen, increase V3 to remove it. Now look for a brightening at the left hand side of the top array row in the Idle Display and a darkening of the middle of the bottom array row. If this is seen, decrease V3. V3 must be made small enough to remove the edge row effects, but large (i.e., negative) enough so as not to have vertical darkening around saturated objects. Now reduce Vgg(on) until the vertical darkening reappears, and increase it again so the vertical darkening has just disappeared. The default array voltages are read from the file IR_CASPIR:ARRAY_PAR.DAT. This file must be edited if new array voltages are to be set on startup. Transputer Preprocessor CASPIR is designed to operate in the high background conditions encountered at long wavelengths. It uses four 16-bit 500 kHz analog-to-digital converters to digitize the data, with the rest of the data train being capable of sustained data rates of at least 2 microsec/pixel. The requirement that individual frames be coadded at this data rate to build up the image means that the data cannot be input directly into MOPRA. Instead a transputer-based preprocessor is used. A transputer is a fast microprocessor chip with considerable in-built parallelism and which uses four fast serial "links" for I/O. Each link can be connected to another transputer, or to an external device through a "Link Adaptor", which is essentially a bidirectional serial to parallel converter. This architecture has made transputers popular in parallel computing applications. The CASPIR system uses four transputer link adaptors on the ADC cards to serialize the data at the Cassegrain focus. Four serial lines then bring the signals to the Nasmyth Lab where the transputer preprocessor is located. This consists of four T800 transputer boards which each have 1 Mbyte of memory and each are responsible for processing the data from one serial line. The current frame is DMAD into transputer memory at the same time that the previous frame is being coadded to an accumulation array where the summed image is stored. After the requested number of coadd cycles, the accumulated image is divided by the number of coadd cycles, optionally has a bias frame subtracted and is divided by flatfield frame, and is copied to MOPRA via a transputer link to Q-bus interface with the four separate data channels correctly interlaced to form the final image. The result of each sequence of coadd cycles is displayed on MOPRA's workstation screen and is stored on MOPRA's disk. The individual frames from each cycle are not normally saved (this only occurs in the occultation observing mode). Instrument Mounting Box All infrared dewars mount on a box attached to the Cassegrain focus of the 2.3 m telescope that is known as the Instrument Mounting Box (IMB). Two dewars can mount on the IMB at the same time and a dichroic mirror in the IMB can be positioned to direct infrared light from the telescope to either dewar. The IMB output ports are identified by the labels A-D. From the Cassegrain Access Platform, with the control electronics racks on your left, the port positions facing you is 'A', the one to your right (not used) is 'B', the back position is 'C', and the control electronics mount on face 'D'. A dichroic refexer directs the infrared beam to the selected IMB port while an acquisition system views the same field in optical light. Auxiliary modules can be mounted above the dichroic refexer in the IMB. A calibration lamp module is available, and a polarimetry module is under construction. A manual dust cover is located in the mounting flange at the top of the IMB. The open and close positions are clearly marked on the flange, and it should be used. Acquisition System The IMB acquisition system contains either the ICCD acquisition TV, SBIG autoguider, or tip-tilt sensor mounted on an X-Y stage at the bottom of the IMB. These systems view the telescope field in optical light through the dichroic reflexer, and allow offset guiding or selection of bright tip-tilt reference stars. The travel of the X-Y stage is ~(+/-31.2mm) in each direction in the f/18 (5 arcsec/mm) telescope beam. A TV filter/focusser unit mounted on the X-Y stage directly in front of the TV camera converts the f/18 telescope beam to f/5.76, provides a pupil mask that acts as an optical sky baffle, allows for focussing of the optical image independent of the infrared image, and contains B, V, R, I, and RG630 filters as well as a clear position. Calibration Lamp Module The calibration lamp module can be used to obtain wavelength calibration arc spectra. The module contains Xenon and Argon lamps as well as an incandescent lamp which may be useful for flatfielding. A rotary mirror is used to select one of the three lamps and a flip mirror is placed in the telescope beam to direct the lamp light to the detector. THE LAMP LIGHT IRRADIATES THE TV CAMERA. TURN DOWN THE TV CAMERA BEFORE ACTIVATING THE CALIBRATION LAMPS. Polarimetry Module The polarimetry module contains an achromatic half-wave plate borrowed from the polarimetry module used at the AAT with IRIS, and a 50mm diameter wire grid for position angle and efficiency calibration. The warm half-wave plate is rotated in the infrared beam to rotate the intrinsic plane of polarization of an astronomical source with respect to a cold analyser mounted within CASPIR. The analyser is either a Wollaston prism mounted in the Utility wheel, or a wire grid mounted in the Upper Filter wheel. The Wollaston prism separates the parallel and perpendicular polarization components equally on the array, so is a dual beam system. The wire passes only one polarization, but it does allow the grisms mounted in the Utility wheel to be used for single beam spectropolarimeter. Imaging Data Reduction Introduction CASPIR produces images of the infrared sky in one passband at a time. These observations normally consist of a number of object and sky frames acquired through the execution of a DO file. Typical observing sequences would: 1) Record a few frames of one object with small spatial offsets between frames to counter ghosts and bad pixels, and to improve spatial sampling of the images. 2) Record many frames of the same object with a dither pattern of offsets to build up long exposures. 3) Record spatial mosaics of dithered sets of images with limited overlap between frames to cover large regions of sky. These observing sequences naturally lead to the definition of a dataset as the set of related observations of a given object in one filter. In the extreme case, a dataset may contain only a single exposure. Different datasets may require different reduction strategies, depending on the nature of the observing sequence employed. The reduction of most datasets will follow the path: 1) Create BIAS and DARK frames, linearize DARK, object and sky frames, and dark subtract objects and sky frames. 2) Create sky or dome FLAT frames, and remove pixel-to-pixel sensitivity variations. 3) Create background SKY frames, and subtract sky background from object frames. 4) Define relative spatial offsets between each object frame in the dataset. 5) Combine all object frames in a dataset into a single image suitable for analysis, using bad pixel masks to exclude bad pixels. Users are cautioned that infrared imaging datasets often present a greater data reduction challenge than optical CCD images both due to the superior performance of optical CCD detectors (lower dark current, read noise, and pixel-to-pixel sensitivity variations) and especially due to the extreme background-limited nature of most infrared imaging observations. The results at each step in the reduction process should be carefully examined and problems understood before proceeding. Many problems can be solved by the exclusion of bad images from the data sets. The reduction procedures described here use the local MSSSO CASPIR package running in IRAF. The procedures (and this description) are based heavily on the SQIID package and its documentation (written by Mike Merrill at NOAO), but have been adapted at MSSSO for CASPIR reductions. The CASPIR package is available via ftp to merlin.anu.edu.au. You can retrieve it by typing: ftp merlin.anu.edu.au log in as 'anonymous' use your email address as password cd pub/peter/caspir mget * bye Define the IRAF variable caspirdir to point to your CASPIR package directory, and put the following lines in your loginuser.cl file. task $caspir = "caspirdir$caspir.cl" caspir The SQIID package is available via ftp to mira.tuc.noao.edu (140.252.3.85). The SQIID package is contained within a tar file called sqiidd.tar within the ftp-anonymous area. You can retrieve it by typing: ftp mira.tuc.noao.edu log in as 'anonymous' use your last name as password cd pub/sqiid get sqiid.tar bye Restore the package using tar sqiid.tar within your IRAF login directory and follow the README directions for installing the package. Preparations CASPIR writes FITS format data files at the telescope which must be converted to IRAF .imh files before reduction can begin. After restoring all the data files for one night to a disk directory and starting IRAF type: files *%.fits%% > allfiles rfits @allfiles//.fits * @allfiles delete @allfiles//.fits to convert FITS files to IRAF .imh files and remove the FITS files from disk. List processing is fundamental to efficient data reduction, and will be used extensively in what follows. The csplist task is a convenient list generation utility for the reduction of CASPIR datasets. The csplist task has the following parameters. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = csplist keyword = list List key type value = kn List key value images = @ifiles List of images to search (first_i= ir001) First image in list (number = 21) Number of images in output list (delta = 1) File number increment (suffix = ) File name suffix (mode = q) The following examples illustrate the capabilities of this task: To generate a file containing a sequence of five filenames starting with irl53 incremented by two, and having the suffix t, type: csplist list first=irl53 num=10 delta=2 suffix=t > tfiles Then tfiles contains the names: irl53t ir155t irl57t irl59t ir161t To select all frames in the list file allfiles obtained in method 2, and write the filenames to a new list file m2files with a t appended, type: csplist method 2 images=@allfiles suffix=t > m2files To select all frames in the list file allfiles obtained with the Kn filter, and write the new filenames to a new list file knfiles with a t appended, type: csplist filter kn images=@allfiles suffix=t > knfiles To select all frames in the list file allfiles obtained with the HK grism, and write the filenames to a new list file hkfiles with a t appended, type: csplist grism HK-grism images=@allfiles suffix=t > hkfiles To select all frames in the list file allfiles with a header exposure time string of 5.0, and write the filename to a new list file 5files with a t appended, type: csplist time 5.0 images=@allfiles suffix=t > 5files To select all dark frames in the list file allfiles, and write the filenames to a new list file dfiles with a t appended type: csplist dark images=@allfiles suffix=t > dfiles Two more general purpose tasks are mentioned before we begin the reduction: cspdisp sequentially displays a list of images and is useful for quickly gaining a feel for the quality of a dataset. csppeek sequentially displays a list of images after a specified dark frame has been subtracted from each frame. This task is useful for quickly assessing raw data that are dominated by the pedestal pattern until a dark frame has been subtracted. These tasks have the following parameters: I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = cspdisp images = List of input images (zscale = yes) Autoscale display? (z1 = ) Minimum level to be displayed (z2 = ) Maximum level to be displayed next_ima= yes Next image? (verbose= yes) Verbose output? (imglist= ) (mode = q1) I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = csppeek images = ir168 List of input images dark = ir366 Dark frame to use (zscale = yes) Autoscale display? (z1 = 0.) Minimum level to be displayed (z2 = 6550.) Maximum level to be displayed next_ima= yes Next image? (verbose= yes) Verbose output? (imglist= ) (node = q1) Forming BIAS and DARK Frames Proper data reduction requires accurate correction for the electrical offsets introduced by the data acquisition system (BIAS frames), the small additive effects of internal illumination, charge generation and charge leakage (DARK frames), the large additive effects of sky illumination (SKY frames), and the multiplicative effects of position dependent pixel sensitivity (FLAT frames). Implicit in this is the need to correct for non-linearities in the responsivity of the detector array. BIAS and DARK calibration frames are required for the linearity correction, so their creation is the first step in the data reduction process. BIAS Frames BIAS frames are defined to be dark exposures of the minimum duration possible for a given read-out method. The nature of the CASPIR array prevents us obtaining zero length exposures. BIAS frame will normally have been recorded with the CASPIR/BIAS command. The stability of BIAS frames over the duration of a night is questionable, so caution dictates that sets of BIAS frames be recorded at the beginning and end of each night. Intermittent problems with reading out the array make it advisable to record several bias frames. The quality of these should be checked visually using the IRAF display task and acceptable frames combined using the cspcomb task. For example, cspcomb ir001,ir002,ir003 bias average averages the three raw bias frames ir00[1-3], and writes the result to the file bias. The cspcomb task has the parameters listed below. The comb_opt parameter defines how the frames are combined. This should be an average if the number of bias exposures is less than about 5 and a median if greater than about 5. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = cspcomb images = ir001, ir002, ir003 List of raw input images output = bias Combined ouput image (comb_op= average) Type of combine operations (verbose= yes) Verbose output? (imglist= ) (mode = q1) DARK Frames DARK frames should be obtained for each exposure time used during a night. These are used in the linearity correction routine to remove dark current and exposure time dependent electrical offset effects. The same stability concerns associated with BIAS frames also apply to DARK frames. Cautious observers may intersperse DARK frame measurements with object frames. Sets of DARK frames of the same exposure time can also be combined with the cspcomb task, for example, cspcomb ir004,ir005,ir006 dark5 average could be used to average four 5 sec DARK frames and write the result to the file dark5. Linearity Correction The response of the CASPIR detector array to light has a quadratic form which must be allowed for before accurate correction of other additive and multiplicative effects can be achieved. CASPIR data are linearized by the csplin task which subtracts a BIAS frame from the data, applies the quadratic correction, converts the data units from ADUs to electrons, divides by the exposure time to produce signal rate in electrons/sec, and flips the image vertically to match the orientation on the data acquisition displays. csplin also optionally linearizes and subtracts a DARK frame from the data. This should be done for imaging data where the photo-generated dark current is a relatively minor contributor to the total signal, but the detector leakage current can be significant in specific hot pixels. The csplin task has the parameters listed below. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = csplin image = List of raw input images bias = Bias frame to use (dark = "") Dark frame to use (verbose= yes) Verbose output? (imglist= ) (mode = q1) The csplin task can be called directly by typing, csplin @ifiles bias dark=dark5 where ifiles is a list file containing the input filenames, and bias and dark5 are the BIAS and DARK frames which are applied to all images in the input list. All images in the input list, should have the same exposure time as the nominated DARK frame. The input images are overwritten with the linearized data. A more convenient environment in which to conduct this and the remainder of the basic imaging reduction is provided by the redimage task. The redimage task is a multifunction procedure which operates along the lines of the ccdproc task in noao.imred.ccdred. For example, to linearize a set of images obtained with a 5 sec exposure time, first form a list file of the image filenames using the csplist task by typing: csplist list first=ir030 num=12 > ifiles redimage overwrites the input images, so copy the input images to temporary files and work on these. This can be done by typing: files @ifiles//t > tfiles imcopy @ifiles @tfiles Now use epar to set the redimage parameters as listed below. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = redimage images = @tfiles List of CASPIR images to reduce (linear = yes) Linearize data? (flatten= no) Divide by flatfield? (skysub = no) Sky subtract? (fixbad = no) Fix bad pixels? (mosaic = no) Mosaic image set? (coord = no) Add coordinate grid? (display= no) Display result? (qphot = no) Measure photometry? (bias = bias) Bias frame to use (dark = dark5) Dark frame to use (flatfil= sflat_kn) Dome flat file (statsec= [50:200,50:200]) Image section for computing statistics (obstype= objskyobj) Type of sky observation made (subtype= running) Type of sky subraction to use (nrun = 4) Number of frames for running sky subtr. (destrip= no) Subtract column pattern after sky subtr. (badtype= mosaic) Type of bad pixel correction (badfile= caspirdir$caspir) Bad pixel file (mosfile= ) Mosaic output file (mostype= blind) Type of mosaic to make (cboxsiz= 9) Size of automatic mode centroiding box (apertur= 5,10) List of photometry apertures (annulus= 10.) Inner radius of sky annulus in pixels (dannulu= 10.) Width of the sky annulus in pixels (verbose= yes) Verbose output? (imglist= ) (skylist= ) (mode = q1) The flags linear, flatten, skysub, fixbad, mosaic, coord, display, and qphot define the reduction steps that will be performed. The remainder of the parameters are used in the execution of these basic functions. The parameters relevant to the linearization of a particular dataset are dark which specifies the single DARK frame to be used for the entire dataset, and bias which specifies the single BIAS frame to be used for the entire dataset. Note that the DARK frame must have the same exposure time as the objects, so observations of different exposure times must be linearized separately using multiple calls to the redimage task. Run redimage by exiting epar using the :g command. After each successful reduction stage, it is advisable to save a copy of the processed files. This can be done in the case of our example by typing: files @ifiles//l > lfiles imcopy @tfiles @lfiles Once you have gained confidence in the quality of your calibration frames, it is possible to simultaneously set all flags in redimage and perform all steps of a dataset reduction in one pass of redimage. This is particularly useful for assessing data quality at the telescope. In the following, we will treat each step individually. It is most convenient to linearize all observations obtained during a given night at this stage of the reduction. A typical sequence might be: csplist method 2 images=@allfiles > m2files csplist time 0.3 images=@m2files > 03files imcopy @03files @03files//t redimage @03files//t linear+ bias=bias dark=dark03 imcopy @03files//t @03files//l csplist time 5.0 images=@m2files > 5files imcopy @5files @5files//t redimage @5files//t linear+ bias=bias dark=dark5 imcopy @5files//t @5files//l delete 03files, 5files Flatfielding The creation of suitable FLAT and SKY frames is more difficult than for BIAS and DARK frames, but they are crucial to the quality of the final images. Different approaches may be necessary for different types of CASPIR imaging data. Compared to optical band CCD observations, most broadband observations with CASPIR are extremely background limited. Furthermore, the background in the near-infrared is variable at many temporal and spatial scales. Since infrared sources are often much fainter than the broadband sky background, very precise removal of the sky signal is required. Narrowband CASPIR images with both pixel scales and J images with the 0.25 arcsec pixels have lower sky backgrounds and present different data reduction challenges. The primary goal in flatfielding images is to correct for pixel-to-pixel sensitivity variations across the array, so that the relative intensities of objects imaged in different parts of the array are accurately recorded. Flattening the sky background is a secondary effect, although this should also be achieved if the array responds similarly to stellar continuum light, and sky emission. Two flatfielding strategies are possible: A set of sky images can be combined to form a sky FLAT frame, or images of an illuminated screen within the dome can be combined to form a dome FLAT frame. It is still not clear whether better photometric accuracy is achieved for CASPIR data by the use of sky flats or dome flats. Dome flats have the advantages that telescope thermal emission can be removed by differencing lamp on and lamp off pairs, and the energy distribution of the lamp matches that of stars better than the sky background, which is dominated by line emission shortward of about 2.2 micrometers. Sky flats have the advantage that the sky is a uniform source of emission viewed in a similar manner to the objects. However, it may be difficult to obtain sufficiently high S/N sky flats for low background configurations such as narrowband filter imaging and at J with the 0.25 arcsec pixels. Some experimentation is required in deciding on the best approach. Dome flats should be measured in sets of lamp on and lamp off pairs for each filter and image scale required. Dome FLAT frames can be created from linearized data with the cspflat task. The inputs required are a list of lamp on, frames, a list of lamp off frames, and an output filename. For example, cspflat ir007l,ir009l,ir011l ir008l,ir010l,ir012l dflat_kn average averages the lamp on frames ir007l-11l and the lamp off frames ir008l-10l, takes their difference, then normalizes the median pixel value of the difference to unity, and writes the resulting dome FLAT frame to the file dflat_kn. The cspflat task has the parameters listed below: I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = cspflat ons = ir007l,ir008l,ir009l List of lamp ON images offs = ir008l,ir010l,ir012l List of lamp OFF images flat = dflat_kn Output flatfield frame (comb_op= average) Type of combine operation (statsec= [50:200,50:200]) Image section for calculating statistics (verbose= yes) Verbose output? (imglist= ) (mode = q1) The comb_opt parameter should be an average if the number of on or off exposures is less than about 5 and a median if greater than about 5. The statsec default should generally be satisfactory. Sky FLAT frames are formed by taking the median of a reasonable number of empty sky fields (or object fields lacking extended emission) obtained at different times during the night, at different locations on the sky, and preferably with the same exposure time. To create a sky FLAT frame for a particular filter and image scale, all suitable sky observations should first be linearized using the redimage task as described above. Then a filename list should be prepared and the cspflat task used with the offs parameter set to the null string to form a sky FLAT frame for the night. For example, delete sfiles csplist list first=ir020 num=12 suffix=t >> sfiles csplist list first=ir057 num=32 suffix=t >> sfiles csplist list first=ir099 num=48 suffix=t >> sfiles cspflat @sfiles "" sflat_kn median forms the median of the sky files listed in sfiles, normalizes the result to a median pixel value of unity, and writes the resulting sky FLAT to the file sflat_kn. Some cautionary notes are in order regarding sky flats: Observations near bright sources (such as the Moon), which have atypical illumination, should not be used to determine global flatfields. Twilight sky measurements may also have illumination atypical of nighttime observations, and it is difficult to obtain the same signal on the twilight sky as obtained in object images (15 minutes into astronomical twilight the sky background is doubling in intensity every five minutes). Twilight sky measurements are not recommended as sky flats. On the positive side, pixel-to-pixel sensitivity variations are expected to be stable over one night, and probably over several nights. This should be checked in your data. All frames obtained with a particular filter and image scale can be flattened together by unsetting the linear flag in redimage, setting the flatten flag, and setting the flatfile parameter to the appropriat FLAT frame filename. Run redimage from the command line or by exiting epar via the :g command. A typical sequence might be: csplist filter j images=@allfiles suffix=t > jfiles redimage @jfiles linear- flatten+ flatfile=sflat_j csplist filter h images=@allfiles suffix=t > hfiles redimage @hfiles linear- flatten+ flatfile=sflat_h csplist filter kn images=@allfiles suffix=t > knfiles redimage @knfiles linear- flatten+ flatfile=sflat_kn delete jfiles,hfiles,knfiles Sky Subtraction The strong and variable near-infrared background has contributions from OH airglow in the J, H, and K bands, moonlight (either directly or reflected off clouds) especially in the J band, and from thermal emission from the telescope and sky in the K and L bands which varies with temperature and humidity. Although the 10-30% variations in background caused by these factors do not strongly limit the S/N of observations (except at K and L for large changes in temperature), they greatly complicate both the creation of mosaics of large regions and accurate surface photometry of objects with extents comparable to CASPIR's field of view. For such observing programs, it is best to obtain sufficient object exposures (and intermixed sky exposures if necessary) to create a SKY frame for each dataset. For programs with single or a few observations of many objects, a sky calibration based on observations of several objects, possibly combined with subtracting a fitted surface from the final image, is the best that can be accomplished. These grouped observations could be treated as one dataset for the purposes of sky subtraction. It is useful to remember that variable airglow can cause the sky background to vary at H by a factor of 2 and at J by 40% on hour timescales. SKY frames for a dataset are created using the redimage task by setting the subsky flag, and supplying values to the obstype, subtype, nrun, and destripe parameters, obstype defines the type of sky observations in the dataset. obstype=all indicates that all images in the dataset are to be included in the creation of SKY frames. obstype=objskyobj indicates that the first image in the dataset is an object image, and this is followed by a sequence of an off-source sky image and an object image, ending with an object image. Only the off-source sky images will be included in the creation of SKY frames. obstype@skyobjsky indicates that the first image in the dataset is an off-source sky image, and this is followed by object and off-source sky image pairs, ending with a sky image. Only the off-source sky images will be included in the creation of SKY frames. obstype=skyobjobjsky indicates that the dataset consists of sequences of sky, object, object, sky frames. Only the off-source sky images will be included in the creation of SKY frames. obstypc=radio is a special pattern used for nbL band observations, initially of radio galaxies. Sets of ten dithered nbL frames are separated by Kn sky and object frames that (hopefully) allow drifts in telescope pointing to be corrected in the final mosaicing. It is likely that, these patterns will include most observing sequences in user defined DO files. redimage can be extended to include other sky types if this proves necessary. Standard star measurements recorded in pairs with the star displaced on the array can be processed by selecting obstype=standard. This is a special type requiring exactly two input images. An output image is formed by subtracting the second (sky) image of the standard star from the first (object) image of the standard star. A permanent output file is produced with the name stdnnn_mmm where nnn is the number of the first standard star image and mmm is the number of the second standard star image. This sky-subtracted standard star image can be automatically processed in each the following steps except that mosaicing and creating a coordinate grid will be ignored. These steps are not applicable to this image. It is most likely that users will fix bad pixels and then measure aperture photometry on the standard star image after sky subtracting. The subtype parameter in redimage defines the type of sky subtraction that is performed. subtype=all defines that all sky images in the dataset will be included in the creation of a single SKY frame, which is then scaled to the median pixel value of each object image and subtracted from them. This is adequate for small datasets where the total time span of the observation is less than about 20 minutes. Larger datasets need to be subdivided into smaller units, with individual SKY frames. This is achieved by setting subtype=running. This causes a SKY frame to be formed for each object image in the dataset from the median of nrun sky images taken immediately before and after the object image. The object image itself is not included in the running median. The SKY frame created for each object image is then scaled to the median pixel value of the object image and subtracted from it. The destripe parameter in redimage determines whether a residual column bias pattern is to be defined and subtracted from each image after normal sky subtraction. Usually this will not be necessary. However, nbL images obtained with readout method 1 suffer from DC drifts in the bias levels of the four output amplifiers between the object and sky frames that are manifest as a residual column bias pattern with four pixel period that is often not removed by normal sky subtraction. When the destripe parameter is set, redimage determines the shape of this bias pattern by projecting the image in the column direction to a 1D spectrum, and then subtracting this spectrum off each row in the image. A typical redimage parameter list for sky subtracting a single objskyobj dataset in the list file tfiles using a running median SKY frame subtraction is shown below. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = redimage images = @tfiles List of CASPIR images to reduce (linear = no) Linearize data? (flatten= no) Divide by flatfield? (skysub = yes) Sky subtract? (fixbad = no) Fix bad pixels? (mosaic = no) Mosaic image set? (coord = no) Add coordinate grid? (display= no) Display result? (qphot = no) Measure photometry? (bias = bias) Bias frame to use (dark = dark5) Dark frame to use (flatfil= sflat_kn) Dome flat file (statsec= [50:200,50:200]) Image section for computing statistics (obstype= objskyobj) Type of sky observation made (subtype= running) Type of sky subtractions to use (nrun = 4) Number of frames for running sky subtr. (destrip= no) Subtract column pattern after sky subtr. (badtype= mosaic) Type of bad pixel correction (badfile= caspirdir$caspir) Bad pixel file (mosfile= ) Mosaic output file (mostype= blind) Type of mosaic to make (cboxsiz= 9) Size of automatic mode centroiding box (ctype = tan) Type of RA/DEC coordinate projection (apertur= 5,10) List of photometry apertures (annulus= 10.) Inner radius of sky annulus in pixels (dannulu= 10.) Width of the sky annulus in pixels (verbose= yes) Verbose output? (imglist= ) (skylist= ) (mode = ql) Fixing Bad Pixels For single observation data sets, or minimally overlapped mosaics, it is necessary to correct bad pixel by interpolation. In heavily overlapped mosaics, bad pixels can be allowed for when these mosaics are combined. However, in both cases it is necessary to attach a bad pixel file to each image before correction can be achieved. Bad pixels can be interpolated using the noao.proto.fixpix or imedit tasks. Bad pixels are specified to these routines in an ascii bad pixel file which is described in the help instruments man pages. The file consists of lines describing rectangular regions of the image. The regions are specified by four numbers giving the starting and ending columns followed by the starting and ending rows, for example, # CASPIR - untrimmed 25 25 111 111 108 108 87 113 256 256 1 256 1 256 1 1 180 190 240 245 If there is a comment line in the file containing the word untrimmed, the coordinates of the bad pixel regions apply to the original image, rather than a sub-section. The file caspirdir$caspir.bad contains the standard CASPIR bad pixel list in this format. It is possible that users will wish to add other bad pixels to their own versions of this list. Mosaics are combined with the powerful imcombine task which uses more sophisticated bad pixel mask images. These are associated with an image through the 'BPM' header entry for the image. A bad pixel mask image is a pixel list file (.pl extension). It is treated like an image file and can be viewed with display and altered with imedit etc. A bad pixel mask image is created from an ascii bad pixel file using noao.imred.ccdred.badpiximage. The following example shows how to form the bad pixel mas image caspir.pl from the bad pixel file caspir.bad. cp caspirdir$caspir.bad . badpiximage caspir.bad ir001 caspir imcopy caspir caspir.pl imdelete caspir The file ir00l can be any CASPIR data file. It is used as a template to define the size of the bad pixel mask image. Good pixels have a value of 1 and bad pixels have a value of 0 in the bad pixel mask image. The cspmask task can also be used to create a bad pixel list file and a bad pixel mask image directly from a CASPIR image (typically a FLAT frame) by defining upper and lower rejection thresholds. The epar listing for cspmask is shown below. I R A F Image Reduction and Analysis Facility PACKAGE = caspir TASK = cspmask input = sflat_kn Input images output = newcaspir.pl Clipped output images lower_li= 0.8 Lower limit for in/exclusion upper_li= 1.2 Upper limit for in/exclusion (in_valu= 1.) Replacement value inside range (out_val= 0.) Replacement value outside range (section= [*,*]) Image section for replacement (trimlim= [0:0,0:0]) Trim limits around edge (verbose= yes) Verbose output? (outlist= ) (mode = q1) To apply the appropriate bad pixel correction using the redimage task, set the fixbad flag and nominate the type of correction and the bad pixel filename using the badtype and badfile parameters. badtype=interpolate causes fixpix to be used to interpolate over bad pixels. badtype=mosaic causes the bad pixel mask image filename to be associated with each object image, but actual correction of bad pixels is deferred until the mosaic is combined. The badfile parameter should not include the file extension (.bad or pl. redimage will append this depending on the type of bad pixel correction selected. Consequently, it is advisable to maintain a .bad and a .pl copy of each bad pixel file used. If all else fails, use imedit to interactively 'fix' bad pixels by defining a circular aperture and replacing the pixel values within the circle, e.g.: imedit input output radius=5 width=5 Preliminary Mosaicing Mosaicing is the most complex part of infrared imaging data reduction. Two crude levels of mosaicing are provided in the redimage task. Discussion of full interactive mosaicing is deferred to a latter section. To mosaic a dataset using the redimage task, set the mosaic flag, enter the mosaic output filename in the mosfile parameter, and define the mosaicing type using the mostype parameter. Selecting mostype=blind causes the object images in the dataset to be combined at their nominal offsets from the base position, as specified in the DO file used to acquire the data and as recorded in the image header entries 'offra' and 'offdec'. This type of mosaic is generally useful for a first look, or for minimally overlapped mosaics where blind offsetting is all that can be achieved. Selecting mostype=manual causes the object images to be display at their nominal offsets so that the user can mark the location of a suitable reference point with the image display cursor. The reference point should be located within each of the images in the dataset, but need not correspond to a particular object. No automatic centroiding is performed on the marked position, so this option is most suitable for noisy images where the centering determination is subjective. Selecting mostype=auto is useful only if there is a moderately unresolved source at the base position of the mosaic, and this source is in each object image of the dataset. This will often be the case for dithered observations of a single object. When mostype=auto is selected, the nominal offsets are corrected by centroiding on the object at the base position in each frame using the proto.imcntr task. The redimage parameter cboxsize defines the size of the centroiding box used. This option produces excellent results for suitable datasets with a moderately unresolved object at the base position. Some care should be exercised in deciding whether centroiding has been successful. This can usually be gauged from the appearance of off-center stars in the mosaic. Observations obtained with the radio.do pattern, described above, require special treatment to estimate the nbL image offsets from offsets determined from interspersed Kn images of the same object. This is achieved by selecting mostype=radio. These options provide a convenient way of assessing mosaiced data at the telescope and of gauging the result of the data reduction steps performed so far, before committing significant effort to the more involved full mosaicing. Coordinate Overlays A world coordinate (RA and DEC) system can be defined for a mosaic image produced with the redimage task by setting the coord flag. The coordinate system is defined from the base position of the mosaic stored in the image header entries meanra and meandec, and the maximum RA offset and the minimum DEC offset used in combining the mosaic and stored in the mosaic header entries moffra and moffdec when redimage forms the mosaic. If the coord and display flags are set when redimage is run, an RA and DEC coordinate grid is overlaid on the mosaic image when it is displayed. This may be helpful for identifying objects in large mosaics and for determining the scale of a mosaic. The images.tv.wcsiab task can be used to overlay the world coordinate grid at any time when the mosaic image is redisplayed. Note that this does not work with SAOIMAGE under Solaris 2, and the wcslab command must be issued twice when using the XIMTOOL display tinder Solaris 2. Once the world coordinate system grid has been defined, RA and DEC positions of selected objects can be obtained by typing rimcursor wcs=world Use the image display cursor to select objects and type any key to print the coordinates. Exit rimcursor by typing d with the cursor in the image display. Appendices SYSTEM PERFORMANCE The best empirical estimate of the system performance is the observation that for imaging with 0.5 arcsec pixels after 6 hr of on-source integration, objects spread over ~2 arcsec with Kn ~19.5 mag are detectable with a signal- to-noise ratio of ~5, depending on how they are measured. This is confirmed by similar observations with 4 min on-source integration times reaching Kn ~17.0 mag with similar signal-to-noise ratio. The following describes measurements of basic system parameters, and then calculations of the theoretical performance based on these parameters. These calculations can be used to estimate system performance in other configurations, but should be normalized by comparison with the above observed sensitivities. System zero point offsets are based on the total ADUs in a sky-subtracted stellar image after correction for airmass effects and are defined by the equation ZP = M_std + 2.5*log(ADU/sec). Typical values of the zero point offsets for each filter (mainly with the 0.5 arcsec pixel scale) are listed in Table 3. These can be used to calculate the total signal expected on an object of a given brightness or the signal/pixel on an object of a given surface brightness. Table 3: Typical Zero Points Filter Zero Point -------------------------------------- J 21.6 H 21.6 K 20.8 K' 20.6 Kn 20.5 M 11.7 MSO [Fe II] 18.6 H_2 1-0 S(l) 18.0 H I Br_gamma 17.6 2.21 micrometer Continuum 19.1 CO (delta_v=2) 18.5 3.28 micrometer Dust 16.6 3.60 micrometer Continuum 17.3 4.00 micrometer Continuum 16.6 H I Br_alpha 16.3 Typical background brightnesses measured with CASPIR are listed in Table 4. The expected background photon fluxes can be calculated from the tabulated background brightnesses and the system zero point offsets. Table 4: Background Brightnesses (mag/arcsec^2) Filter IRPS IRIS CASPIR CASPIR PICNIC 1984 1993 Nov 1993 Mar 1993 Aug 1994 --------------------------------------------------------------------------- J 15.5 15.0 18.7 .... .... H 14.5 13.7 14.6 .... .... K 11.5 12.5 11.6 11.7 .... K' .... 13.7 12.5 12.6 .... Kn .... 13.2 12.4 12.5 13.1-13.9 2.21 micrometer Continuum .... .... 12.9 .... .... CO (delta_v=2) .... .... 10.7 .... .... 3.28 micrometer Dust .... .... 3.1 .... .... 3.60 micrometer Continuum .... .... 3.8 .... .... 4.00 micrometer Continuum .... .... 2.6 .... .... H I Br_alpha .... .... 2.0 .... .... The noise in an image is a combination of photon shot noise from the sky and telescope, photon shot noise from the object, shot noise from the dark current, read noise, and other systematic noise sources that are difficult to quantify. For small signals, the noise per pixel can be estimated from the equation: Noise = [RN^2 + T*(i_b + i_d)]^1/2 where RN is the read noise in e-, T is the integration time in sec, i_b is the background signal in e-/sec, and i_d is the dark current in e-/sec. The read noise for the double sample readout methods (methods 2-4, and method 5 with FNDR=1) is ~50 e-. The dark current is typically 10 e-/s/pixel for most of the array, but there are a significant number of detectors with dark currents of 100 e-/s/pixel, and some with 1000 e-/s/pixel. With these data, theoretical performance figures for CASPIR can be calculated from the measured background brightness and the camera throughput as quantified by the system zero point. For example, for a Kn background brightness of 12.4 mag/arcsec^2 and a Kn zero point offset of 20.5 mag, the Kn background flux for 0.5 arcsec pixels is (10^[(Z*P-m_std)/2.5])/4 = 434 ADU/sec/pixel In a 5 sec integration, the background count is ~2170 ADU/pixel, or 19550 e-/pixel (1 ADU = 9 e-). The shot noise of this background signal is (19550)^(1/2) = 140 e- which dominates the typical readout noise of 40-60 e-. Consequently, Kn images with 0.5 arcsec pixels and an integration time of 5 sec should be background limited. The total noise per cycle is ~(50^2 + 19550)^.5 ~148 e-/pixel or 148/9 = 16.5 ADU/pixel which is reduced to 4.8 ADU/pixel when 12 cycles are averaged. The measured value is ~5 ADU/pixel. We assume here that sufficient sky frames are averaged so that sky subtraction is essentially noiseless. For a 5*sigma detection of an object spread over n x n pixels, we require an average signal in each of these n^2 pixels of five times the noise per pixel. The total object signal is then n^2 x 5 x 4.8 ADU. This can be converted to a Kn magnitude after dividing by the integration time of 5 sec and using the Kn zero point offset of 20.5 mag. In this way, we can estimate limiting magnitudes at Kn for a range of seeing or object sizes. The results of these calculations are shown in detail in Table 5. Table 5: Performance at Kn (0.5 arcsec/pixel) Image Size 5*sigma Total Signal Mag Time to 20 mag (arcsec) (pixels) (ADU/5 sec) (1 min, 5:1) (min) ------------------------------------------------------------------------- 1 x 1 2 x 2 96 17.3 145 1.5 x 1.5 3 x 3 215 16.4 760 2 x 2 4 x 4 382 15.8 2290 5 x 5 10 x 10 2384 13.8 91200 Predicted performance figures for 5*sigma detections in 1 min of on-source integration in different seeing conditions for various filters are listed in Table 6 for the 0.5 arcsec/pixel scale and in Table 7 for the 0.25 arcsec/ pixel scale. The 1 min integration time does not include time for sky measurements and the dead time between frames of ~20 sec. It is recommended to limit individual frames to 1 min exposures, effectively making the elapsed time ~80 sec, so that an adequate number of sky frames can be obtained in the time scale of 15 min on which the sky level is observed to change significantly. If off-source sky measurements are necessary, it is recommended that equal time be spent on the object, and sky positions. Relative performance figures for each filter are of interest in deciding which passband is most sensitive for a particular observation. These calculations for the 0.5 arcsec/pixel scale in 1.5 arcsec seeing and an integration time of 5 sec with 12 cycles are listed in Table 8 for a 15.0 mag star with zero color (S/N_HotStar), a typical unreddened late-type star with K = 15.0 mag, J - K = 1.0, and H - K = 0.2 (S/N_CoolStar), and a typical AGN power law spectrum with S proportional to v^-1.5 and K = 15.0 mag (S/N_AGN). Performance figures for the grisms can be estimated assuming a slit transmission Tau_Slit = 0.5 and a grism transmission Tau_Grism = 0.5. For example, consider an observation using the HK grism and a 1 arcsec (2 pixel) wide slit of an object spread over 2 arcsec (n_y = 4 pixels) along the slit and recorded in two frames. Table 6: System Performance (0.5 arcsec/pixel) Filter Time Cycles RN^2 Background Dark 5*sigma, 1 min Magnitude Signal Signal 1 x 1 1.5 x 1.5 2 x 2 5 x 5 (sec) (e-)^2 (e-) (e-) ---------------------------------------------------------------------------- J 5.0 12 2500 3100 50 19.1 18.3 17.6 15.6 H 5.0 12 2500 7100 50 18.8 18.0 17.3 15.4 K 5.0 12 2500 53850 50 17.1 16.2 15.6 13.6 K' 5.0 12 2500 19550 50 17.4 16.5 15.9 13.9 Kn 5.0 12 2500 19550 50 17.3 16.4 15.8 13.8 [Fe 11] 10.0 6 2500 900 100 16.8 15.9 15.3 13.3 H20 10.0 6 2500 3810 100 15.8 15.0 14.3 12.3 Cont2.2 10.0 6 2500 6800 100 16.7 15.9 15.2 13.2 CO 10.0 6 2500 29660 100 15.5 14.6 14.0 12.0 Table 7: System Performance (0.25 arcsec/pixel) Filter Time Cycles RN^2 Background Dark 5*sigma, 1 min Magnitude Signal Signal 1 x 1 1.5 x 1.5 2 x 2 5 x 5 (sec) (e-)^2 (e-) (e-) (arcsec) ---------------------------------------------------------------------------- J 5.0 12 2500 775 50 17.9 17.0 16.4 14.4 H 5.0 12 2500 1775 50 17.8 16.9 16.3 14.3 K 5.0 12 2500 13460 50 16.3 15.4 14.8 12.8 KI 5.0 12 2500 4890 50 16.5 15.6 15.0 13.0 Kn 5.0 12 2500 4890 50 16.4 15.5 14.9 12.9 [Fe 11] 10.0 6 2500 224 100 15.4 14.5 13.9 11.9 H20 10.0 6 2500 978 100 14.7 13.8 13.2 11.2 Cont2.2 10.0 6 2500 1700 100 15.7 14.8 14.2 12.2 CO 10.0 6 2500 7420 100 14.6 13.7 13.1 11.1 Cont3.28 0.3 200 2500 706500 3 10.0 9.1 8.5 6.5 nbL 0.3 200 2500 706500 3 10.7 9.8 9.2 7.2 (n_frames = 2) each with an integration time T = 180 sec and with the object placed at different positions along the slit in both images. The spectral resolution at 2.2 micrometers is 2.2/2200 = 0.001 micrometers/pixel, or a factor of Lambda = 0.33/0.001 = 330 lower than for Kn. In the spectral direction, each pixel sees signal from an area of sky A_Sky = 1 arcsec x 0.5 arcsec = 0.5 arcsec^2, dispersed by a factor of Lambda more than for Kn, but reduced in flux by Tau_Grism. The background current per pixel (averaged over features in the sky emission spectrum) is therefore: [10^((Z.P.-m_Bkg)/2.5) * A_sky * Tau_Grism * Gain/Lambda = 11.8 e-/sec/pixel Here we have used the observed Kn background brightness, m_Bkg, of 12.4 mag/arcsec^2 , and zero point offset, Z.P., of 20.5 mag, and conversion between electrons and ADU, Gain, of 9 e-/ADU. The noise per pixel is then Noise = (50^2 + 180*(11.8+10))^1/2 = (2500 + 3933)^1/2 = 80 e-/pixel We see from this that shot noise from the dark current makes a significant contribution to the total noise, and that integration times of order 180 sec are required to minimize the read noise contribution. Integration times significantly longer than this are not recommended because sky intensity variation make accurate sky subtraction increasingly difficult, and significant numbers of hot pixels saturate as the integration time is increased further. Normally, spectra are recorded as a nodded pair to allow sky subtraction, and preferably as an ABBA sequence. In the case where a single sky frame is subtracted from an object frame to perform the sky subtraction, the total noise per pixel is increased by (2)^1/2 because the sky frame has the same noise per pixel as the object frame. For our example, the final noise becomes 113 e-/pixel. We define a 5*sigma detection per pixel by requiring a signal-to-noise ratio of 5 per pixel after averaging n_y pixels along the slit and averaging the n_frames object frames. The average object signal per pixel in the dispersion direction is then Signal = 5 * Noise / (n_y)^1/2 /(n_frames)^1/2 = 200 e-/pixel The equivalent Kn imaging signal would be Kn Signal = 200 / Gain / Tau_Grism / Tau_Slit * Lambda * n_y^2 = 470000 ADU which corresponds to a Kn brightness of Kn = 20.5 - 2.5 * log (470000/180) = 12.0 mag From this we predict that a signal-to-noise ratio of 5:1 can be achieved with the HK grism on a Kn 12.0 mag object in the continuum in 6 min of on-source integration. The 5*sigma line detection sensitivity in the same time can be estimated by assuming the line occupies two pixels in the dispersion direction. The line flux is then 3.84 * 10^-14 * 10^(-12.0/2.5) * 0.001 * 2 = 1.2 * 10^-21 W cm^-2 MISCELLANEOUS NEAR-INFRARED DATA Typical atmospheric extinction corrections for SSO are listed in Table 9, and values for the standard interstellar extinction law from Rieke & Lebofsky (1985, ApJ, 288, 618) are listed for reference in Table 10. Typical values for the near-infrared broadband zero magnitude flux calibration are listed in Table 11. Also tabulated are values obtained from the fit equation: F_nu = 5 * (10^3 / lambda^3) * [exp (1.2/lambda) - 1] Here F_nu is flux density in Jy and lambda is wavelength in micrometers. Table 9: Atmospheric Extinction Filter mag/airmass ----------------------------------- J 0.12 H 0.08 K 0.10 L 0.19 M 0.20 H20 2.00 micrometers 0.11 Continuum 2.21 micrometers 0.11 CO 2.34 micrometers 0.19 Table 10: Standard Interstellar Extinction Law Filter A_lambda/A_v ---------------------------------------------- U 1.531 B 1.324 V 1.000 R 0.748 I 0.482 J 0.282 H 0.175 K 0.112 L 0.058 M 0.023 N 0.052 Table 11: Typical Fluxes for a Zero Magnitude Star Filter lambda_c F_nu F_nu (fit) (micrometers) (Jy) (Jy) -------------------------------------------------- J 1.25 1520 1588 H 1.65 980 1041 K 2.20 620 647 L 3.60 280 270 M 4.80 153 159 WHEEL CONTENTS The contents of the five wheels are listed below. Note that as of Oct 1994, it is necessary to set the Utility Wheel to increasing numbers in order to ensure that it is latched correctly. The Lens Wheel should also be set to increasing numbers. These details will be fixed as soon as convenient. Table 12: Aperture Wheel Contents Position Keyword Content ----------------------------------------------------------------------------- 1 Blank Blank 2 Lslit1 1.0 arcsec x 128 arcsec slit 3 Disk5 5.0 arcsec occulting disk 4 Lslitl.5 1.5 arcsec x 128 arcsec slit 5 Lslit2 2.0 arcsec x 128 arcsec slit 6 SlowClr 0.25 arcsec/pixel baffle 7 Lslit5 5.0 arcsec x 128 arcsec slit 8 Lslitl0 10.0 arcsec x 128 arcsec slit 9 Polar Three 20 arcsec x 120 arcsec slits (for polarimetry) 10 Sslit10 10.0 arcsec x 15 arcsec slit 11 Sslit5 5.0 arcsec x 15 arcsec slit 12 FastClr 0.5 arcsec/pixel baffle 13 Sslit2 2.0 arcsec x 15 arcsec slit, 14 Sslit1.5 1.5 arcsec x 15 arcsec slit 15 Disk2 2.0 arcsec occulting disk 16 Sslit1 1.0 arcsecx 15 arcsec slit Table 13: Upper Filter Wheel Contents Position Keyword Filter lambda_c delta lambda (micrometers) (micrometers) -------------------------------------------------------------------------- 1 Blank Blank ... ... 2 Clear Clear ... ... 3 Helium He I 10830 1.082 0.011 4 PGamma H I P_gamma 1.093 0.010 5 PBeta H I P_beta 1.282 0.015 6 FeII MSO [Fe II] + PK50 1.647 0.018 7 AAOFeII AAO [Fe II] 1.650 0.015 8 H2O H2O 1.996 0.050 9 H2_1_0 H2 1-0 S(1) 2.120 0.027 10 BrGamma H I Br_gamma 2.170 0.022 11 Cont2.2 Continuum 2.210 0.094 12 H2_2_1 H2 2-1 S(1) 2.249 0.024 13 CO CO (delta v = 2) 2.343 0.088 14 Cont1.6 Continuum 1.580 0.012 15 Grid Wire Grid Analyser ... ... 16 Mirror IR85 ... ... Table 14: Utility Wheel Contents Position Keyword Content -------------------------------------------------------- 1 Blank Blank 2 Clear Direct imaging cold stop 3 IJ-grism IJ cross-dispersed grism 4 Blank4 Blank 5 Blank5 Blank 6 H-grism H grism 7 SlowCam Slow camera lens (0.25 arcsec/pixel) 8 JH-grism JH cross-dispersed grism 9 Blank9 Blank 10 Blankl0 Blank 11 J-grism J grism 12 Mask Coronograph pupil mask 13 Blankl3 Blank 14 HK-grism HK cross-dispersed grism 15 Wollaston Wollaston polarimeter analyser 16 K-grism K grism Table 15: Lower Filter Wheel Contents Position Keyword Filter lambda_c delta lambda (micrometers) (micrometers) ------------------------------------------------------------------------------ 1 Blank Blank ... ... 2 Clear Clear ... ... 3 J J 1.275 0.282 4 H H 1.672 0.274 5 KP K' + PK50 2.124 0.337 6 KN Kn + PK50 2.165 0.330 7 K K 2.224 0.394 8 L L' 3.821 0.602 9 Ice 3.08 micrometers H20 ice 3.077 0.102 10 Dust3.28 3.28 micrometers Dust emission 3.299 0.074 11 Dust3.4 3.4 micrometers Dust emission 3.415 0.072 12 Cont3.6 3.6 micrometers Continuum 3.592 0.078 13 Cont4.0 4.0 micrometers Continuum 3.990 0.052 14 BrAlpha H I Br_alpha 4.051 0.054 15 M M 4.777 0.650 16 PK50 2 mm PK50 ... ... Table 16: Lens Wheel Contents Position Keyword Content --------------------------------------------------- 1 Blank Blank 2 FastCam Fast Camera Lens (0.5 arcsec/pixel) 3 Blank3 Blank 4 Clear Clear K Filter Transmission Functions (see digitized TIF files)