FIGURE 1: The MOS/OSIS double spectrograph, section view
The MOS/OSIS instrument is attached to the Cassegrain bonnette (acquisition/guiding/rotator unit) at the f/8 focus of the CFH 3.6 m telescope. OSIS and MOS are located on 2 opposite sides of the large, stiff, octagonal structure with 4 openings, or side ports, attached to this bonnette (see Figure 1). The two remaining ports are presently unused.
The present optics of OSIS were redesigned for adding infra-red capabilities to the original SIS configuration. They operate between 0.36 µm and 1.8 µm and are optimized to provide good images over this wavelength region and over the entire 3.6' x 3.6' useful field. The initial SIS optical scheme is shown in Morbey (1992, Applied Optics, 31, 2291). The final OSIS scale is 139 µm/", or 0.15"/pix for CCDs with 21 µm pixels.
With this optical design, all rays, coming either from the edge or from the centre of the field, are constrained to be coincident at the pupil and rays from each object point are parallel as they pass through the grism so that the same region of the grism disperses the light from any point of the field in the same way.
From the variation of the rms spot size in the field of view in the visible and in the infrared, it is clear that the image quality lies well within 20 µm in diameter (rms), and that the quality is constant across the field.
At 80% encircled energy, the spot sizes are inside 20 µm in the red, and still inside 30 µm diameter circles in the least favourable configuration (blue spectral range and at the outer edge of the field). The results is that the intrinsic image quality of the instrument is always better than the spatial resolution given by the seeing (~56 µm for 0.4" seeing).
In spectroscopic mode, the image quality is well within two pixels; this has been demonstrated from 300 to 700 nm.
The OSIS optics are mainly coated with MgF2 on various glasses. The flat which folds the beam from the telescope for feeding OSIS is coated to give an average reflectance of 97% over the entire wavelength range of the instrument. The total transmission is then predicted to be more than 80% at 0.7 µm and still higher than 60% at 1.8 µm. In near UV, OSIS has somehat better throughput than SIS.
The OSIS train is shown on the right side of Figure 1. When mounted on a specially designed storage cart, the whole OSIS assembly can be inserted precisely into the ports of the octagon or withdrawn for maintenance.
The main mechanical components in OSIS that the observer can control are the following:
Several grism and filter wheels exist. They are easily interchangeable cassettes (Grundmann et al., 1988, in ``ESO conf. on Very Large Telescopes'', M.-H. Ulrich ed., II, 1173) and can be inserted either in OSIS or in MOS (except for the filter wheels with OSIS-R). The design allows these cassettes to be interchanged quickly and precisely, with minimal danger to their contents; they are encoded to allow the control system to recognize them. All driving and encoding components are kept inside the body of the spectrograph, external to the cassette so that no attached electronics can be damaged during handling. After each wheel is inserted during the set-up of the instrument, the control system maps the wheel positions for further reference. Both the filter and the grism cassettes are located near the pupil plane. Each grism cassette has 8 positions for 65 mm diameter grisms. The grism mountings allow precise alignment of each individual grism, and a locking pin is used to ensure that the grism cassettes can be located and maintained in position to an accuracy of ~20". Each OSIS-V filter cassette has 8 positions for 75 mm diameter filters (but adaptors also permit use of square 2-inch filters, for instance).
Extensive flexure measurements have been made using OSIS in its previous SIS configuration. Internal SIS flexures, from the entrance focal plane to the CCD focal plane, were measured with MOS/SIS installed at the Cassegrain bonnette. A focal plane mask with 25 µm apertures was installed in the mask slide and illuminated with the continuum lamp of the calibration system. The telescope was moved to hour angles between -4h and +4h and declinations between -45 and +65.
At each telescope position, an image of the focal plane mask was recorded on the CCD, and the centroids of the apertures measured. One has to note that this measurement also includes flexures of the CCD in its mechanical housing (LICK2 was used for these tests).
The measured flexures are plotted in Figure 2. The internal flexures of OSIS for the telescope tracking a star are therefore expected to be a maximum of 25 µm at the detector focal plane for a motion of 4 hr in hour angle or equivalently an average of 0.036 arcsec/hr. The analysis of direct images obtained with SIS active guiding are in good agreement with this figure, which shows that the stiffness of the guide probe system is excellent.
No measurement has been done without active guiding. For this mode of observation, flexures are expected to be similar to those observed with MOS, i.e. ~40 µm, or 0.06"/hr.
and 
(OSIS-Visible)
