This is the text of the NSFCAM article from the March 1994 IRTF Newsletter. ***************************************************************** NSFCAM Commissioned The IRTF's new 256x256 InSb array camera is named NSFCAM. It was commissioned as a "shared-risk" IRTF facility instrument in early September 1993 with a newly installed science-grade array. The camera was used successfully by several observers through early January, after which it was shipped back to Manoa to be upgraded to its final configuration. At the telescope, the camera system provided a total system throughput of 40% (with standard K filter) and a system read noise of approximately 50 electrons (using single reads at the standard readout: 11 Hz full frame). Multiple non-destructive readout (i.e. sampling "up-the-ramp") is also available, reducing the read noise to roughly 20 electrons with 8 samples. Up to three subarrays can be read out per integration. Because of the reduced number of pixels, subarrays can be read out significantly faster than full frames. Background flux measured at the telescope is summarized in the following chart. These values were used to estimate point source sensitivity: Filters Observed Estimated (central wavelength Background Flux Point-Source and FWHM in microns) (see units in text) Sensitivity J+ 1.26 (0.31) 6.9x10^3 (15.9) 20.6 mag H+ 1.62 (0.28) 2.9x10^4 (13.4) 18.9 K 2.21 (0.39) 1.7x10^4 (13.7) 18.8 K'+ 2.12 (0.34) 1.2x10^4 (14.0) 18.9 shK+ 2.15 (0.32) 1.0x10^4 (14.1) 18.9 L 3.50 (0.61) 1.8x10^7 ( 4.9) 14.0 L' 3.78 (0.59) 2.5x10^7 ( 4.5) 13.6 M' 4.78 (0.22) 1.5x10^8 ( 0.3) 10.7 M 4.85 (0.62) 5.3x10^8 (-0.7) 10.7 Background units are electrons/s/arcsec^2 (and mag/arcsec^2). The quoted backgrounds were measured after the secondary was recoated in July 1993 and so reflect the subsequent drop in thermal flux. The presence of a "+" in a filter name denotes the use of a blocking filter (PK50 glass) used to suppress long wavelength leaks. Sensitivities were estimated by assuming 60 second integrations on source and 60 seconds on nearby sky and refer to 3-sigma in the sky-subtracted image. The system throughput (electrons detected per photon incident on the telescope) was assumed to be 40% at all wavelengths. We have also assumed that star images are uniformly spread over a 2.1x2.1 arcsec square (i.e. 7x7 box at 0.3 arcsec/pixel). The flux within all of these pixels is summed and the noise values added in quadrature. The same sensitivity would be obtained if only a single pixel were considered (i.e. 3-sigma in one pixel), assuming that 1/7-th of the star's light falls on that pixel. It must be understood that these estimates assume background-limited performance (easily attained) and perfect flat-fielding (see below for a discussion of flat field results). The backgrounds at J and H are dominated by OH sky emission lines and can therefore vary by more than 50%. If observers are interested in using the narrowband filters or CVFs, we can estimate sensitivities based on OH lines within the passbands. Please contact us for this information. A list of available narrowband (R=100) filters is available upon request or through IRTF Online. Also, recall that thermal band backgrounds will vary with water vapor content of the atmosphere. Unlike the case with ProtoCAM, we often vary the detector bias voltages in NSFCAM in order to vary the well depth. Deeper wells at higher bias must be weighed against an increase in dark current. We use higher bias for thermal wavelengths, where the increased dark current is not a problem. We have typically been using two bias settings, yielding useful well depths of roughly 50,000 and 100,000 electrons. The narrow M' filter was specially designed by a group at the Royal Observatory, Edinburgh to take advantage of the lower background in a clear portion of the earth's atmosphere emission near 5 microns. Note that its sensitivity matches that of the wider bandpass M filter. Because of its high background, the M filter can only be used with the 0.06 arcsec/pixel platescale. We therefore recommend that observers use the M' filter (which can sometimes be used at 0.15 arcsec/pixel), unless they need to maintain continuity with previous standard M-band photometry. Photometry obtained with the M' filter is also less subject to photometric variations from atmospheric water vapor. Preliminary results show that K images can be very accurately flat field corrected. Using a 6x6 grid of star images on the array, we have found that dome flats at K will reduce photometric variation to maximum deviations of +/-1% (rms deviation is much lower). We expect that similar accuracy will be possible at shorter wavelengths. We are having trouble establishing a method to obtain good flat fields at thermal wavelengths. For instance, simply using sky images does not affect the original star grid photometric deviations of +/-5% at L'. Of course, multiple dithered images will give point-source photometry more accurate than this. We are continuing to work on this problem. We have improved the "shift & add" (S&A) mode since its description in the last Newsletter. Though limited to 128x128 subarrays, this mode now will subtract a blank sky image and flat field each image in real-time. The subtraction of a sky image allows us to use S&A mode for much fainter objects than were previously feasible. However, this mode is far from fully tested. We are asking that observers who propose to use this mode demonstrate that they could also use post-processing shift and add techniques to attain their goals. The main advantage of the real-time S&A mode is that it significantly reduces the overhead needed to write each image to disk. Please contact us before proposing to use this mode. Coronagraphic masks have been installed in the camera and are available in real time for all three platescales (0.056, 0.153 and 0.310 arcsec/pixel). The diameters of the focal plane masks are 4.0, 6.0, and 2.0 arcsec for these platscales, respectively. The masks were provided by Bob Brown (STScI), who has kindly agreed to make them available for others to use. We have only begun preliminary testing of these masks and do not have sensitivity estimates. They are not yet available for general use. In April, after the successful reopening of the observatory, the camera should have the following added capabilities: 1) circular variable filters (CVFs) from 1.5-5.5 microns; and 2) an infrared grism. Polarization imaging capability (1.0-2.5 microns) will not be available until the following semester (August 1994 - January 1995). The CVFs will be cut to cover the following ranges: Wavelength range Resolution (microns) 1.51-2.52 90 2.70-4.25 110 4.45-5.50 60 These filters are just about to be cut, so the final wavelengths may differ slightly from these values. Also, because the filters cannot be used all the way up to an edge, the actual wavelength ranges will be slightly truncated. Please contact a support astronomer if you plan to operate within 0.1 micron of one of the above CVF limits. The grism will be diamond-ruled from ZnSe with a resolution of roughly 300. In the absence of test results with the grism, we can only estimate its efficiency, roughly 30% in the L band and 20% in the K band. Estimated continuum sensitivities in the K and L bands are roughly 7x10^-15 W/m^2/micron (K=11.5) at 2.2 micron and 2x10^-14 W/m^2/micron (L=9.0) at 3.45 micron (100-sigma in 60 seconds total integration time). For more information about the grism, please contact John Rayner. We would like to take this opportunity to offer our heartfelt thanks to the technical staff who helped build the camera and make it work. They include Doug Toomey, Peter Onaka, Tony Denault, Darryl Watanabe, Vern Stahlberger, Kevin Criez, Lou Robertson and Danny Cook. In addition, we thank the IRTF daycrew for providing the constant help on the summit to keep the camera working when we're not there to break it. Mark Shure (shure@amber.ifa.hawaii.edu) John Rayner (rayner@galileo.ifa.hawaii.edu)