Goodman Overview

At least read this!

Optics: The Goodman optics are designed to transmit down to the atmospheric cutoff at 320 nm, and include lenses made of CaF2 and of NaCl. The latter are the center elements of fluid-coupled triplets. None of the multiplets are over 4" in diameter which reduced the difficulty compared to spectrographs with larger pupil sizes. Each of the multiplets is sealed on one end with a face-mounted o-ring that imposes a known axial load, and on the other end with a rim-mounted o-ring that imposes a radial load, and finally held captive axially with a retaining ring that incorporates a third o-ring. This last o-ring does not participate in the sealing of fluid, but avoids a metal glass interface that would be undesirable for the CaF2 lenses. The salt lenses are held by the other optics and are never in contact with a seal.

Filters: There are two independent six position filter wheels that hold 4 inch diameter filters. One of these will be used for imaging, the other for spectroscopic order sorting filters. The initial compliment of imaging filters includes U, B, V, and R on the Kron-Cousins system. The spectroscopic order sorting filters includes GG-385, GG-455, GG-495, and OG-570 filters. The filters are in the collimated beam (tilted to avoid ghosts). Installing different filters is straight forward, but should be considered a day-time operation.

Imaging Mode: In imaging mode the plate scale is 0.15 arcsec/pixel and the field of view is 7.2 arcmin in diameter (3096 x 3096 unbinned pixels).

Spectroscopic Mode:

Slits Masks: In Spectroscopic mode the Goodman Spectrograph will be able to obtain spectra of multiple objects simultaneously over a field of 3.0 x 5.0 arcminutes using multi-slit masks. A carousel style mask changer, holding up to 36 masks will allow the slit plates to be accurately and reproducibly located at the instruments entrance aperture.

The instrument currently deployed with a compliment of 7 fixed long slits with widths of 0.45, 0.84, 1.03, 1.35, 1.68,  3.0, and 10.0 arcsec. These are each 3.9 arcminutes long, but can be fitted with optional decker plates to mask the upper and lower portions.

A cutting machine for the fabrication of custom multi-slit masks, has been purchased during 2007 using Brazilian funding, and will be shared between SOAR and Gemini South. We anticipate that the software to design masks from either images or coordinate lists, and to tweak the alignment of the masks at the telescope, will be developed on the same timescale.

NOTE: At present, only the 7 fixed long slits are in use for spectroscopic mode.

Gratings and Preset Observing Modes: Up to three gratings can be installed in the spectrograph at a time, in a linear stage which allows the rapid interchange of gratings. Installing different gratings is straight forward, but should be considered a day time operation.

The grating complement includes 300, 400, 600, 900, 1200, and 2100 l/mm transmission VPH gratings.  The table below shows the dispersion and the wavelength coverage for observations in our set spectroscopic modes.  Please note that the 2100 l/mm grating is operated in Custom mode where the observer selects the central wavelength for their observations. Because of limits in the camera rotation stage, it is not possible to use the 2100 l/mm grating beyond 600nm.

Because VPH gratings operate via Bragg scattering, efficient operation requires Littrow or near-Littrow operation of the spectrograph. A grating rotation stage sets the incident angle to the desired value, which depends upon the line density of the grating and the wavelength of interest. A concentric camera rotation stage must then be set to nearly twice this angle to intercept the diffracted beam. A set of fixed observing modes for each grating are given below.

Calibration Lamps: For the calibration of Goodman spectroscopic data we have a quartz lamp for spectral flats and HgAr, Ne, Ar, and  CuHeAr lamps for wavelength calibration. Plots of these spectra with the lines identified in each of our standard spectroscopic modes will will be posted here as they become available.

CCD: The Goodman focal plane is imaged onto a Fairchild 4k x 4k CCD with the following properties:

Detector Type	Fairchild CCD 486 Backside
Image Size	4096 x 4096 @ 16bits/pixel plus overscan and header 32 Mb per image
Pixel Size	15 microns/pixel = 0.15 arcsec/pixel
DQE	Wavelength (nm)    QE %    300  45  350  70  400  90  500  88  600  85  700  80  800  65  900  42  950  27  1000  14
Dark Current	0.0003 e/pixel/sec
Single Pixel Full Well	139.8 ke
Linearity	0 to 80% Full Well
Cosmetics	Trap: Column 2706, Row 1435; Cluster Defect: Column 2048-2065, Row 129-145
Read Out	Read Rate	Analog ATTN	Gain (e/ADU)	Noise (e)	Read Out Time (sec) Imaging/Spectroscopic
	400 kHz	0	5.67	8.62	24/20
	200 kHz	0 2	1.40 2.67	4.74 5.12	48/40
	100 kHz	0 2 3	0.56 1.06 2.06	3.69 3.72 3.99	96/79
	50 kHz	0 2 3	0.25 0.47 0.91	3.33 3.35 3.41	192/157
Charge transfer efficiency	>99.999

The default image size for imaging mode (1x1 binning) is 3096 x 3096 pixels and the default image size for spectroscopic mode is 4142 x 1896 pixels with 1x1 binning. These values were used calculate the read out times given above using one amplifier readout. Users should also expect 1 or 2 seconds of overhead on every exposure. 

Throughput:  The imaging mode throughput has been measured relative to the SOAR Optical Imager (SOI) for which we have good zero-points. As the table below shows, the throughput relative to SOI is better in the R, comparable in the V and lower in the B and U. The U band throughput should improve when the camera optics are rebuilt. The B band is somewhat more mysterious and we are investigating why it is only 80% of the SOI value.

We have also measured the spectroscopic throughput of Goodman by taking spectra of spectrophotometric standards. These curves show the overall system efficiency for the telescope+instrument+detector combination. That is, they show the fraction of photons striking the primary mirror of SOAR which are eventually detected by the CCD. The measurements were made using a very wide (>10 arcsec slit); slit losses will reduce the efficiency obtained when using a narrower slit by an amount which depends on the seeing. The measurements are corrected for atmospheric extinction; the efficiency obtained in an actual observation will be reduced by the atmospheric extinction for the airmass of observation.

 Figure 1.  Upper Left: Overall system throughput (OST) for (telescope + instrument + detector) the Goodman Spectrograph after new collimator optics were installed in Feb 2011 for the 300l/mm, 600l/mm, and 2100l/mm grating.  Upper Right: OST for Goodman after new collimator optics were installed in Feb 2011 for the 1200l/mm grating.  Middle Left: A comparison of the OST for the 300l/mm, 600l/mm, and 2100l/mm gratings before and after the new collimator optics were installed in Feb 2011.  Middle Right: A comparison of the OST for the 1200l/mm grating before and after the new collimator optics were installed in Feb 2011.  Lower Left: A comparison of the OST for the 400l/mm grating and the 300l/mm grating.  Lower Right: The OST curves for the 930l/mm grating.  Please note that the OST curves for the before and after comparison were taken without blocking filters, so there is a significant blue leak redward of 600nm in both the before and after measurements which somewhat complicates the comparison.

Use of blocking filters: Those taking spectra to the red of ~600nm should be aware that, depending on the spectrum of their target, there may be significant contamination from second order blue light superposed on the first order red spectrum (a blue leak). The blue leak will change the apparent shape of the red continuum, "fill in absorption features in the red, and may "imprint" emission or absorption features occurring in the blue spectrum at roughly twice their wavelength. This second order contamination can be eliminated by use of an appropriate blocking filter. However, this does entail a loss of efficiency in the red since the "in-band" transmission of the available blocking filters is not 100%.

To help users make this trade off, the graphs in Figure 1 for the red set-ups show three curves for each set-up:

1. The dotted curves show the values measured with no blocking filter for a particular standard star (LTT3218, a fairly blue white dwarf). To the red of 600nm the spectra becomes increasingly contaminated by the 2nd order blue spectrum of the star ("blue leak")

2. The dashed curves show the values measured for the same star when an appropriate blocking filter is used. There is now no "blue leak", but the efficiency is significantly reduced by the absorption of the blocking filter even "in-band"

3. The solid lines show the result of dividing the dashed curves by the measured transmission of the blocking filter. They are thus approximately the efficiency that would be obtained if an object with no flux in the blue was observed without a blocking filter. The difference between the solid and dotted curves gives an idea of the importance of the blue leak for that particular standard star. The difference between the solid and dashed curves gives an idea of the efficiency loss due to the blocking filters.

Those needing spectrophotometric calibration should note that essentially all spectrophotometric standards are quite blue, so there will be a significant blue leak if they are measured without a blocking filter, which will invalidate the calibration of science target spectra, even if the targets themselves have no blue flux. A possible approach would be to measure the science targets without a blocking filter, the standards with one, and then correct the standards for the blocking filter transmission. However, we currently do not have measurements of the blocking filter transmission which we consider sufficiently reliable for these purposes (we are working on this). In the meantime observers who plan to do this should measure there own by observing a red star with and without blocking filter.

In principle contamination by the blue leak will also effect arc lamp calibrations (superposing blue lines on the red spectrum at roughly twice their wavelength) and flat fields. However, with the exception of the HgAr lamp the blue lines in all the calibration sources are very weak compared to the red lines, and similarly the flat field sources are much brighter in the red than the blue.

Scattered (and Stray) Light:

Measured to be small in imaging mode by comparing imaging FOV and surroundings with a bright star illuminating the pupil. Some scattered and stray light is seen in the Goodman spectroscopic mode.  We currently see 0.06e/s of stray and/or scattered light in the spectroscopic mode with 1x1 pixel binning and the 100kHz ATTN2 readout.  Efforts are underway to replace the tent that covers Goodman with a light-tight box.  We are also looking for low-level emitting LEDs on all of the Goodman motors and covering them with metallic tape as we find them.

Calibration Issues:

Obtaining good quartz spectra over the entire wavelength range with the 300 l/mm grating is difficult because of the different spectral response at the red and blue ends. The most noticeable effect is that to obtain sufficient counts in the blue end, the red end becomes saturated. A blocking filter is on order so that composite quartz flats can be made. This effect is not as great with the other gratings because the wavelength range isn't as large.

There also exists contamination of the flats by scattering from the back of the second filter wheel. Use of the blocking filter mentioned above mitigates this contamination.

Finally, fringing appears in the spectra to the red of the Hα line. Data are being obtained to measure the fringing and data reduction schemes are being developed to remove this contamination.

Flexure:

The instrument has active flexure compensation based on the Nasmyth rotator angle. The corrections for the old camera were succesful at the fraction of a pixel level for the full range of rotator angles. Calibration needs to be tweaked for the new camera once the adjustment of the counterweight is completed. We also need to check for flexure driven by other effects, such as temperature.