|Home > Astronomy > ISIS > Red Arm|
ISIS Red Arm
The default chip for the ISIS red arm is Red+. It is a red-sensitive array of 2048×4096 15.0µ pixels with almost no fringing.
The Red+ detector bias frames can be characterised as B(x,y,t) = f(x,y) + g(t), where f(x,y) represents the quasi-fixed pattern of pixel-to-pixel variation, and g(t) represents the variation of the mean (over x,y) bias level with time. The time variability typically amounts to up to ~10-20 counts on the timescale of hours and days, and depends on environmental factors such as CCD and controller-electronics temperatures. The quasi-fixed pattern function f(x,y) is measured from the median of the usual afternoon bias frames, and the time-variation function g(t) is measured from the overscan region of the detector from each science observation. The overscan strip of the Red+ detector extends over rows 4110 to 4190, and is defined in the FITS headers by the usual BIASSEC keyword.
Biases and the overscan region
Wavelength coverage and spectral resolution
The dispersions of the red-arm gratings in Å/mm are 121 (R158R), 62 (R316R), 33 (R600R) and 17 (R1200R). The pixel size of the Red+ detector is 15µ, and the corresponding grating dispersions in Å/pixel are listed in the table, as are the slit widths that project to four pixels (60µ) with the gratings set at blaze. The spectral resolution elements, Δλ, in Å for a 1-arcsec slit are also listed. The corresponding nominal spectral resolutions, λ/Δλ, at 7000Å with a 1-arcec slit are approximately 909 (R158R), 1842 (R316R), 3867 (R600R) and 9333 (R1200R).
Note that the slit width projecting to four pixels increases with grating resolution. This is due to grating anamorphic de-magnification, which is parameterised by cos(θ+φ/2)/cos(θ-φ/2), where θ is the grating tilt with respect to zero order and φ is the camera-collimator angle. This relation demonstrates why the slit size projecting to four pixels is larger for the higher resolution gratings; they are set to larger tilt angles for a given central wavelength, and so their de-magnification is greater. See here for a discussion of this effect.
According to sampling theory a line as recorded on the detector is fully sampled if it has at least two dispersion elements across its FWHM, and so the detector oversamples the resolution element when the slit width is set to 1-arcsec (e.g., to approximately match seeing). In this configuration the detector can be binned ×2 spectrally to increase the signal-to-noise in each wavelength bin, without loss of spectral resolution. This can also be done at the reduction stage, but binning on-chip reduces the readout noise contribution to the resolution element.
It is also possible of course to improve spectral resolution by reducing the slit width. This increases slit losses especially in moderate seeing, and in any case the slit should project to at least two pixels so that the resolution element remains fully sampled by the detector.
Fringing and cosmetic defects
Bad pixel masks for Red+ with different binning levels were created using noao.imred.ccdred task in IRAF. All masks are created for the default CCD window [585:1550,1:4200].
Bad pixel masks
An atlas of arc lines for a range of central wavelengths of the red arm gratings is available here.
Atlas of arc lines in the red arm
|Top | Back|