The spectrograph design and the observing procedure

The choice of the spectrograph design (Fig.2) is determined by the quantity R, the overall dimensions of the PF cage and the restrictions on the rigidity of the structure of the device.

**Figure:** A simplified scheme of the prime focus echelle spectrograph.
Designations: 1 - calibration spectrum sources, 2 - fiber optics, 3 - dekker, 4 - refracting mirror, 5 - collimator, 6 - echelle, 7 - grating of cross dispersion
$\begin{figure} \centerline{ \psfig {figure=fig2.ps,width=8cm} }\end{figure}$

**Figure:** Fragments of the three orders of echelle spectrum for the quasar S5 0014+81
$\begin{figure} \centerline{ \psfig {figure=f3.ps,width=8cm,bbllx=42pt,bblly=90pt,bburx=293pt,bbury=443pt,clip=} }\end{figure}$

We have chosen $\rm D_{coll}=50$ mm, which at the PF relative aperture of 1:4 yields $\rm F_{coll}=200$ mm. The collimator is assembled according to Newton scheme, the dimensions of the flat refracting mirror are such that it falls within the central shaded part of the collimated beam. The diameter of the shaded part is determined by the relation between the main mirror diameter and the PF cage diameter. As the camera's objective a fast 4-lens aplanat is used having a focal distance F_cam=140 mm and a relative aperture of 1:1.8, the resolution at the centre is no less than 80 gr/mm, that over the field at a distance of 3 mm from the edge along the frame ( $\rm 24\times 36\,mm^2$ ) diagonal is 35 gr/mm. These parameters allow a two-pixel resolution over the whole area of the CCD used ( $\rm 17\times 19\,mm^2$ ) to be realized. The light transmission is 0.9, the drop in illumination over the frame ( $\rm 24\times 36\,mm^2$ ) is no more than 20%. An echelle grating of 75 gr/mm, $\Theta_b=63.5\degr$ , is used, the size of the shaded area is $\rm 120\times 60\,mm^2$ , the concentration in the working order is 70% in fractions of the reflected light. As a cross-dispersion element a grating of 300gr/mm, working in the first order, is used, the size of the shaded area is $\rm 90 \times 90\,mm^2$ , the concentration in the working order is 90% in fractions of the reflected light. The instrument is equipped with additional cross-dispersion gratings 600 and 1200 gr/mm. The orientation of the echelle grating conforms to case ``C'' according Schroeder and Hillard (1980), i.e. $\alpha=\beta$ , and $\gamma=6\degr$ . Such a configuration permits the centres of all optical elements to be arranged in one plane. The slit part is a turrel mounting with several deckers of fixed sizes. The sizes of the working rectangular entrance diaphragms are $4\times0.5$ arcsec and $4\times 1$ arcsec. The inclination of monochromatic slit images due to the non-zero $\gamma$ angle value is made up for by changing the orientation of the decker. In the preslit part are mounted the optics of the TV guide and the optics of the calibration channel, the latter being thick fiber-optics passing radiation to the spectrograph entrance. All the optical and mechanical units are mounted on a single frame, on the opposite side of which (i.e. outside the optical part volume) the electronic units of the calibration and control systems are fixed (design and execution of V.I. Fateev). The proposed design of the spectrograph made it possible to minimize deviations of the centre of gravity of the structure, reduce its weight and use such an orientation of the CCD cryostat that enables observations at the prime focus of over 15 hours without replenishing liquid nitrogen.

The CCD system incorporates the cryostat camera and the controller unit, connected by a cable 3 m long. The controller is housed in a commercial 19-inch casing and is connected to the computer (two coaxial cables 200 m in length). The system ensures control by the CCD in the frame mode. The read-out of a frame fragment and the binning over the two coordinates are possible. The read-out time of a frame of $1050\times 1170$ pixels is 101 s with a reduction rate of 12 kpixels/s. The signal reduction channel is characterized by the following parameters:
16-digit analog-to-digital conversion;
charge-code conversion coefficient -- 2.3e^-/ADU;
read-out noise -- 6.5e^-;
bias level -- 1600 ADU;
stability of the bias level -- 0.5 ADU;
integral nonlinearity of the channel -- no higher than 0.05%;
differential nonlinearity -- no more than 0.5 ADU.

The image is aligned on the CCD frame as follows: 1160 work pixels are oriented along the spectral order, 1040 pixels are oriented in cross-dispersion direction. The pixel size is $16\times 16$ microns. The CCD angular scale is 11.45''/mm or 0.37''/pix. With 2.5'' seeings the spectral order is 14 pixels in height, i.e. one can work with a 4-pixel binning across the main dispersion, reducing twice the caused impact by read-out noise. The spectral resolution estimated from spectra is R=20000.

Observations are performed according to usual way (calibration by continuous and line spectra, registration of the spectrum of a fast-rotating star, dark-frames) that can be modified if it is needed for a particular programme. With a properly organized sequence of observations, about 90-95% of night time is used for signal registration. The registration of observational data is performed in the environment Nice (Knyazev and Shergin, 1994). The reduction of images is made in the MIDAS system.