PDS_VERSION_ID = PDS3 RECORD_TYPE = STREAM LABEL_REVISION_NOTE = "NULL" OBJECT = INSTRUMENT INSTRUMENT_HOST_ID = "IRTF" INSTRUMENT_ID = "NSFCAM" OBJECT = INSTRUMENT_INFORMATION INSTRUMENT_NAME = "NSF CAMERA" INSTRUMENT_TYPE = CAMERA INSTRUMENT_DESC = " Instrument Overview =================== NSFCAM is a 1-5 um imager with a 256x256 InSb detector. Simultaneous optical imaging is possible via an optical port. Three different magnifications can be selected by rotating different cold lens assemblies into the beam. These give plate scales of 0.3arcsec/pixel, 0.15arcsec/pixel and 0.06arcsec/pixel, with corresponding fields of view of 76.8 arcmin, 37.9 arcmin and 14.1 arcmin respectively. Images can be obtained through a variety of broad- and narrow-band filters, and low-resolution spectroscopy can also be carried out using the circular variable filters (CVFs) or grisms. Polarizing waveplates or grids can be inserted into the beam. Two cold coronograph masks are also available. Images are recorded onto a Sun workstation in FITS format. These are taken in basic stare mode using the main (object) and offset (sky) telescope beam positions; software is available to shift-and-add series of images together to study fine image structure. It is also possible to take frames rapidly in movie mode, storing them directly to the instrument computer and saving overheads in time-critical observations. Electronics =========== Instrument start-up ------------------- The camera is run from a PC which is located out at the telescope. You can communicate with this PC, called NSFPC, via a monitor and keyboard in the control room (to the right of the Planck workstation), or by logging on to NSFPC from a workstation. Often both the computer and the software which runs the camera are referred to as the 'IC', for Instrument Computer. The default Voltage (DETector) of -3.4 implies a 300mV bias across the array (-3.4 - -3.7). This works well in the non-thermal regime and where the background is not high, i.e. through narrow filters or for the CVF. Increasing the bias across the array, i.e. increasing VDET, will increase the well depth allowing you to reach higher counts before entering the non-linear regime, but it will also rapidly increase the number of bad pixels and the dark current. Hence the best value of VDET is a compromise between well depth and increasing bad pixels, and it may vary from night to night; getting the correct value will probably require some experimentation at the telescope. You can reduce the number of counts per pixel and the non-linearity problem by using a smaller pixel scale, if your program will allow this. A non-destructive read consists of reading each pixel without resetting its value. Increasing the number of reads per image reduces the read noise and for this reason the default value of NDR is set to 8. For the CSHELL array (also a SBRC 256x256 InSb) it was found that the read noise was 55 electrons (or 5 DN) for NDR=1, and 22 electrons (2 DN) for NDR=6. For NSFCAM the gain has been measured to be 10 electrons/DN. The higher the number of NDR the longer the minimum integration time, as each read requires about 0.077 seconds. The number of NDR will be set automatically to the maximum number possible (up to 8) for your requested integration time (see following Section). The maximum possible value of NDR (currently 8) as well as the number used on the last 'Go' is displayed on the top User Interface window. Data Modes ========== Readout Modes ------------- 'ARC_D' stands for 'array reset clocking, double sample'. The entire array is readout once (called pedestal sampling) and then after the requested integration time has elapsed the array is readout again (sampled). The difference between these two reads is displayed in VF and saved to disk if Autosave is selected. The minimum integration time is a function of array size. If more than one NDR is done, the pedestal value is read out many times (set by the value of NDR) and the final sample value is read many times. The integration time is the time from the first read of the pedestal to the first read of the sample. The frame that is stored is the difference between the sum of the pedestal read and the sum of the sample reads. Hence the number of counts in the frame will be increased by a factor equal to the number of NDR. If more than one coadd has been requested then this entire process is repeated for the number of coadds, and the frame that is stored is the sum of the frames. Hence when reducing your data the counts in each frame need to be divided by the number of NDR and the number of coadds. This value is given in the FITS header as 'Divisor'. The quick-look utility VF will automatically divide the counts in the image by this value. As described above, read noise decreases with increasing NDR. Hence the maximum number of NDR possible within your requested integration time (up to 8) will be carried out. For full arrays this means for example that the value of NDR will be equal to the integer value of: (integration time) / 0.0768. The other readout modes are not operational however briefly they work as follows. ARC_S (Array Reset Clocking Single Sample) follows the same procedure as ARC_D however no pedestal value is subtracted. PRC (Pixel Reset Clocking) mode is similar to ARC_S however instead of reading the entire array every integration interval, the pixels are read out on an individual basis. Similarly CDS (Correlated Double Sample) reads the pixels on an individual basis but subtracts a pedestal value for each pixel. These last two modes effectively increase the well depth of the pixels as each pixel is not left integrating up in counts while the entire array is being sampled. They will be useful modes for observing in the thermal regime when they are operational. Optics ====== NSFCAM -- A U. Hawaii 256x256 Infrared Array Camera with an In-Sb based chip, with each pixel individually addressable. The camera can be used in many modes including imaging and grism-based, low-resolution spectroscopy. The Observing Parameters are array, filter, lens, dichroic and waveplate. Waveplate --------- A half-wave plate is available for polarizing work with NSFCAM. Polarizing grids are also available in both the CVF and filter wheels. Dichroic -------- The first cold optical element in the light path is the dichroic assembly. There are four turrets currently containing: an open aperture (default position), a visual/infrared dichroic, and two lenses for the cold coronograph instrument CoCo. Lens (& Grism Slit & Coronograph Masks) --------------------------------------- The optical element below the dichroic is the lens turret assembly. There are three plate scales available chosen by selecting the appropriate lenses; the focus position for each scale is different. The three scales are: 0.055arcsec/pixel, 0.148arcsec/pixel and 0.300arcsec/pixel, giving fields of view of 13 arcsec, 35 arcsec and 72 arcsec (allowing for bad edges to the array). There are coronograph masks available for the two smaller scales and a grism slit available with the largest plate scale. The coronograph masks are centrally located on the array and have diameters of 4 arcsec (for the 0.055 arcsec scale) and 6 arcsec (for the 0.148 arcsec scale). Filter and CVF Wheels --------------------- The last optical elements before the detector are the filter and CVF wheels. The software will automatically select the blocking filter or open position in the CVF wheel as appropriate for the filter wheel selection. Similarly the blocker in the filter wheel will be automatically selected for the CVF. (The blockers are necessary in some cases to prevent long-wavelength light leaks). All you need do is select your filter or the CVF and type in the required CVF wavelength with the latter. After making your selection from the filter and CVF menus you must click on the 'move' button before the wheels will be moved. The table below lists the filters in this project. Index Name Step Deg Central Wavelngth FWHM microns 1 J 2500 125 1.26 0.28 3 K' 2020 101 2.12 0.34 8 Spencer 2.3 740 37 2.28 0.17 10 VIS ND=20. 7180 359 16 Brackett ki 2.166 5620 281 2.16 0.02 17 H2 v=2-1 s1 2.248 5380 269 2.25 0.02 18 CO headband 2.295 5140 257 2.30 0.03 19 CO continuum K 4900 245 2.26 0.05 23 .945 micron 3940 197 24 IR ND=2.0 3580 179 27 Open 2740 137 Array (and Subarrays) --------------------- You are not constrained to reading out the full array of the camera - smaller subarrays can be used. You can also store an image made up of up to three non-overlapping subarrays. There are various ways of defining which regions of the array you want read out. In the Observing Parameters window you can use the Array 1/2/3 lines to explicitly define the subarrays. You would enter the x,y coordinates of the top left of the region (the top left corner of the full array is 0,0), followed by the width and height of the region in pixels. The width and height must be divisible by 8, and the software should round off your numbers to the nearest permissible value. In this window 'num array' is the value of the total number of regions you want read out in a single frame. Operational Considerations ========================== Integration Times, Coadds, Cycles --------------------------------- The integration time is specified in seconds on the 'itime' line in the window. The minimum integration time is about 0.077 seconds for a full array. Software will automatically go to the smallest possible value if your value is below this limit. The actual integration time will be displayed and written in the FITS header. The longest possible integration time will be set by the requirements of: not getting the counts up to the non-linear regime and not sitting on any science target for longer than the sky stability period. Setting the number of coadds greater than one will result in repeated reads with your selected integration time, added together, with the sum written out as a single image. This was also described earlier where the connection between integration time, number of NDRs, and coadds was discussed. VF automatically divides the counts in your image by the number of NDR and coadds, so that you can easily check for non-linearity. Typically you will set your integration time to the maximum value possible to still remain safely in the linear regime, using either a bright science object in the field or the sky background if your objects are faint. Then you will set your coadds so that the total time is about equal to sky stability time (one or two minutes at JHK, 20 seconds at longer wavelengths). Keep your counts greater than ~100 to avoid being limited by read noise. Selecting more than one 'cycle' just means that the entire process will be repeated many times with many images being written out (as opposed to a single image if coadds >1). If 'pair' is selected as the 'Obs Mode' then cycles will repeat in a ABBA... pattern. System Performance ================== The focus of the telescope is controlled from the operator's console so that the observer needs to request any focus changes (this will be changed in the future). The focus of the telescope gets more negative with decreasing temperature, and so has to be reduced during the night, usually stabilising by the early hours of the morning. The NSFCAM plate scales have different focus positions. The values depend on the ambient temperature and can change by 20 units (0.20) but typical values (for the non-tiptilt top end) are: Instrument Sensitivity Assuming a 2 arcsec x 2 arcsec Image is as follows: Filter Background per square arcsec 3 sigma mag in min on target with gain of 10 e/ADU and 1 min on sky J 6.9 x 10**3 e/s 15.9 mag 20.6 K' 1.2 x 10**4 14.0 18.9 K 1.7 x 10**4 13.7 18.8 Other magnitudes or signal-to-noise values can be calculated assuming background limited performance: S/N = signal x (sqrt time) So that for example to increase signal-to-noise from 3[[sigma]] to 30[[sigma]], the signal would have to be increased by a factor of 10, implying that the object would have to be 2.5 magnitudes brighter than the values given above, or the integration time would have to be increased by a factor of 100. For the narrow band filters or CVFs S/N can be estimated using the above numbers and the difference in bandwidth. For example, the signal through the CVF is down by about a factor of 10 compared to the corresponding broadband filter and the noise is down by the square root of 10; therefore objects would have to be 1.25 magnitudes brighter than the values listed above for the same signal-to-noise. Note that often systematic effects will limit the accuracy achieved. In particular sky variations will be the limiting factor unless you are careful to limit the exposure time for each image to less than the likely sky variation time. This translates into maximum integration times of about one or two minutes at JHK, and about 20 seconds at longer wavelengths. Keep your counts greater than ~100 to avoid being read noise limited. The minimum number of counts to be background limited is equal to the square of the read noise in electrons divided by the gain. Calibrations and Photometric Color Transformations ================================================== Non-Linearity ------------- The array behaves in a similar manner to the CSHELL SBRC InSb array. The plots shown in the back of the CSHELL manual of counts versus integration time, as a function of bias voltage, can be used as a guide for the camera. Observations of photometric standards with a 300mV bias show that counts around the 2500 level result in a 5% non-linear effect. Standard Stars -------------- For JHK work the UKIRT Faint Standards list provides an excellent source of standard stars. This list is posted on paper in the control room and is also available to the operators and observers via the xstarcat utility. The natural system of NSFCAM is extremely similar to the UKIRT photometric system. The systems appear identical at H and K with a small difference at J described (provisionally) by the color relation: (J-K)nsfcam = 0.95 (J-K)ukirt Using the UKIRT to CIT relationships from Krisciunas (private communication) this implies that the relationship between the NSFCAM natural system and the CIT system (as defined by the Elias standards) is: Kcit = Knsfcam (H-K)cit = 0.854(H-K)nsfcam (J-K)cit = 0.974(J-K)nsfcam Photometric zeropoints for NSFCAM (i.e. the magnitude that is added to the instrumental magnitude to obtain true magnitude) are typically: Filter J H K L' Plate Scale(arcsec) 0.30 0.15 0.30 0.15 0.30 0.15 0.15 Zero Point 22.54 22.44 22.06 21.94 21.43 21.32 20.0 These values are for an airmass of 1.0 and were measured over the time period December 1994 to November 1995. They are a measure of instrument throughput and should not vary by more than about 0.10mag from the above values. Atmospheric extinction is typically in the range 0.05 to 0.20 magnitudes per airmass at JHKL'. When reducing your data don't forget to divide your FITS images by the number of coadds and NDR (DIVISOR parameter value in the header); this is done automatically in VF. The maximum time between successive images, i.e. the number of coadds, should be constrained by sky variability. At JHK sitting on one position for a minute or two should be OK, but in the thermal regime shorter observations are recommended, such as 200 coadds at 0.08 seconds. Flat Fields ----------- At J,H,K good flats can be obtained either from the dome or from the sky. Night to night variations at the level of 2% have been seen so taking flats each night is recommended (this situation may have been improved with the new array temperature controller). All the lights should be on in the dome and the telescope pointed at the usual flat position (which is not the white spot). Taking two sets of data with the lights on, where one set has half the integration time of the other, and subtracting the results, works well. In the thermal regime dome flats require a long integration to get sufficient counts and usually sky flats are used, or flat fielding is not done and the object and calibrator are placed in the same region of the array. The array is flat to about 7%. The usual structure is a gradient across the array with the right brighter than the left, and what Mark Shure refers to as the Starship Enterprise pattern in the center. If you do not see this kind of pattern then it is possible that there is a problem with the array and you should report this to the operator and your support astronomer." END_OBJECT = INSTRUMENT_INFORMATION OBJECT = INSTRUMENT_REFERENCE_INFO REFERENCE_KEY_ID = "LEGGETT&DENAULT1996" END_OBJECT = INSTRUMENT_REFERENCE_INFO END_OBJECT = INSTRUMENT END