Focal enlarger
In front of the device, at the Cassegrain focus, is mounted a small enlarger, transforming F/8 into
F/14.5; that is why each 100 µm fibre subtends 0.4" on the sky. This allows to take into account the focal ratio degradation within the fibres. At the exit, the beam is collected by the F/8 MOS collimator without additional optics. So, the spectral resolution for a given grism is about the same as could be obtained with a 100 µm width slit (0.7" on the sky) in classical mode. Since the sampling of the fibre's images by the CCD is correct, we did not attempt to conserve the "étendue" of the beam by adding more optics. The focal enlarger is made of 3 achromatic doublets from Melles-Griot, references LAO 028, LAO 068 and LAO 034 (see drawings). They received anti-reflection coatings.
Fibre bundle
This is the "heart" of the device, the part that changes the geometry of the focal image into a pseudo-slit.The fibres are FHP 100/110/125 from Polymicro. That means a 100 µm diameter core of silica, doped silica cladding giving a 110 microns diameter and further polymid cladding pushing the total diameter to 125 microns.
Transmission of silica is very good down to 2800 Å. transmission curve is given in section 5.
The geometry at the entrance side is an hexagonal arrangement of the fibres within an almost complete hexagonal area: 25 rows, with 30 fibres in the median row and 18 fibres at the edge. This makes a total of 594 active fibres. The useful field of view is 12.8" x 7.8" and the sampling 0.40" per fibre (fibre core).
At the "slit" end, successive rows are just aligned one after another, while keeping the internal order within each row, to form a "long slit". Spacers (1 fibre diameter) are inserted between each row in this pseudo-slit. The geometry is similar to an image slicer. The spacers are useful for the data reduction; they allow to correct for the spectrograph distortion along the slit.
At the entrance, the polymid cladding has been chemically removed to minimize geometrical losses. The packing factor is thus better than 75\%. The distance between fibre centres is 110 µm.
At the "slit" end, it was kept to better separate the successive spectra. The distance between fibre centres is 125 µm, which is a little less than 3 pixels with the MOS optics for a detector with 15 µm pixels.
The fastidious labor of fibre handling for building this device was done at SOVIS (B.P.1, 77 260, La Ferté-sous-Jouarre, France).
MOS/ARGUS slide
On the MOS/ARGUS slide, encoding by micro-switches is similar to those of other MOS slides, with 4 positions: