ISIS Blue Arm

The default CCD on the blue arm of ISIS is a thinned, blue-sensitive EEV12, an array of 4096×2048 (13.5 micron) pixels. EEV12 can only be replaced by a TEK CCD if it is absolutely essential for the observers' program (e.g. drift-scan mode is only implemented with a TEK due to limitations on buffer read-out capacity).

1. General information on the device
2. Spectral resolution and wavelength coverage
3. Spatial scale
4. Fringing and cosmetic defects
5. Linearity measurements
6. Shutter effect
7. Charge spreading variations and effects on spectral resolutions
8. Flux standard data and empirical through-puts
9. Quality control history
10. Bad pixel masks

The table below gives the dispersion provided by each grating when mounted blaze to collimator (see the ISIS manual for more detailed information on gratings and their properties), and the spectral range covered by the EEV12 CCD. The EEV12 is mounted on the blue arm with its 4096 pixel axis along the dispersion direction, giving maximum use of the beam width leaving the cameras. However, the camera optics vignette the outer regions of the dispersed light beam such that approximately 600 pixels at either end of the CCD are vignetted. A plot of a CCD vignetting function across the chip is shown below to illustrate this effect. This function was measured in May 2011 from a flat-field exposure corrected by CCD quantum efficiency, grating efficiency and tungsten lamp spectral emissivity functions. The unvignetted region is from pixel 665 to 3485, which is essentially the central 2820 pixels.

ISIS wavelength coverage and resolution with EEV12
Grating	Blaze	Dispersion (Å/mm)	Dispersion (Å/pix)	Total Spectral range (Å)	Unvignetted range 2820 pixels (Å)	50% unvignetted range 3670 pixels (Å)	Slit-width for 54 mu at detector (in arcsecs)
R158B	3600	120	1.62	6635	4568	5944	0.83
R300B	4000	64	0.86	3539	2436	3170	0.84
R600B	3900	33	0.45	1825	1256	1636	0.89
R1200B	4000	17	0.23	940	646	842	1.08
H2400B	Holo	8	0.11	442	304	396	1.24

The pixel size of the EEV12 (13.5 microns compared to 24 microns of a TEK CCD) means that a slit-width of approximately 1" will project to about 4 pixels Full-Width-Half-Maximum (FWHM) on the detector, for all available gratings. According to sampling theory a line can be considered resolved if it has 2 dispersion elements at its FWHM, hence we are oversampling the best resolution that we can achieve with this spectrograph setup. As a way of increasing the signal in each wavelength bin one can then combine the signals of each pair of adjacent pixels without losing important resolution information. This can be done at the reduction stage, but to decrease the readout noise contribution one can bin the CCD at readout so that two pixel elements are combined to produce one 'dispersion element' which has the readout noise of approximately one pixel.  The above table includes a column for the slit-width which projects a FWHM of 4 pixels at the detector (54 microns), and excludes the diffusion of charges between pixels mentioned here. The observer can choose to bin the chip at readout in the spectral direction by 2 to maximise signal detection, and reduce readout noise contributions.

It is theoretically possible to make use of the smaller pixels of the EEV12 to improve the actual spectral resolution. This of course requires the observer to use quite a narrow slit-width, and results in significant light loss (assuming moderate 1" seeing conditions).

The blue (and red) camera is a folded Schmidt design of 500mm and gives a scale of 14.9 arcsec/mm along the slit. Hence a spatial scale of 0.2 arcsec/pixel is achieved with the EEV12. It is possible to bin in the spatial direction if one is not concerned with high spatial resolution observations, indeed the seeing conditions need to be excellent to allow full advantage to be taken of using an unbinned chip with this pixel scale. The maximum unvignetted slit-length usable with ISIS is 3.7 arcmin (corresponding to a 1100 detector pixels, spanning ~[400:1500, 1:4096]). These thinned chips suffer severely from fringing in the red part of the spectrum, which limits their usefulness in this region despite their continued good QE down to 8000 Å. Click here to see an illustrative flat field spectra.  The following figures are illustrative:

 
Wavelength     Peak-to-Peak Amplitude

6500Å                 5%
7000Å                 15%
7500Å                 30%
8000Å                 50%   
8500Å                 60%
9000Å                 60-70%

There are a few cosmetic defects on the surface of the chip, but nothing particularly severe. A linearity test in standard readout mode with no binning showed the chip to be linear to better than 1% up to just over 60,000 ADU. The Prontor shutter opens the aperture radially symmetrically. The count levels on a series of exposures with requested integration times 2, 1, 0.5, 0.25, 0.13, 0.06, and 0.03 sec suggest that the overhead for opening and closing the shutter is ~ 0.04 - 0.05 sec. An exposure of 0.03 s yields fewer counts, probably due to the shutter not being fully opened. If an exposure time of 0.01 sec is requested, the exposure fails. To clear the error state, type in the instrument control system

    SYS@taurus> dasreset blue

Remember to check the CCD readout-speed and binning after the dasreset. For observations of spectrophotometric standards we recommend a minimum exposure of 2 sec. The diffusion of charges between pixels during integrations causes a degrading of the spatial and spectral resolution. For a long-slit spectrograph like ISIS, at the WHT's mean seeing spatial degradation is not a significant concern with the pixel size of the EEV12, but it is something to consider in the spectral direction. For a back illuminated CCD this charge diffusion (often referred to as the Modulation Transfer Function; MTF) becomes progressively worse for shorter wavelength incident light. For example, using a slit-width projecting 2 pixels on the detector results in a FWHM measured of 2.4 pixels (measured at ~4000Å) when the spectrograph is at best focus. Similarly a slit-width projecting to 4 detector pixels will produce a FWHM of ~4.4 pixels (again at ~4000Å). This effect becomes less severe towards redder wavelengths and is negligible at around 6000Å. The figure below shows the results from throughput measurements of flux standards. The Y-axis is the apparent AB magnitude of star observed at zenith which gives one detected photon per second per Angstrom. The lowest resolution grating was used (R158B) in the blue arm (without a dichroic) with a wide slit (10 arcsec). The conditions were photometric, with negligible dust levels present. This figure shows the response of the whole ISIS blue channel (i.e. WHT primary + secondary + ISIS blue optics + detector response) from ~3200 to ~8300Å.

The figure below shows the results from several throughput measurements of flux standards in both the blue and the red arms and with different detectors. The Y-axis is the apparent AB magnitude of star observed at zenith which gives one detected photon per second per Angstrom. In each case the lowest resolution grating was used (R158B) in the blue arm (without a dichroic) with a wide slit (10 arcsec). In each case, the conditions were photometric, with negligible dust levels present. Bad pixel masks for EEV12 with different binning were created using noao.imred.ccdred task in IRAF. All masks are created for the default CCD window [585:1550,1:4200].

bad pixel mask bin 1 1           bad pixel mask bin 1 2           bad pixel mask bin 2 1

bad pixel mask bin 2 2           bad pixel mask bin 3 1           bad pixel mask bin 3 2

bad pixel mask bin 4 1           bad pixel mask bin 4 2