3 Fibre optics and telescope module

  The coupling of telescope and remote spectrograph by optical fibres is a relatively new technique. Not all the problems and their consequences for scientific applications are by now fully understood; therefore it is necessary to test the fibre characteristics such as transmission and degradation as well as their imaging properties and their response to mechanical and thermal conditions. Provided that the latter two items are well under control it is transmission and degradation which must be optimized with respect to the optical and astronomical requirements.

3.1 Properties of optical fibres

Although the basic properties of optical fibres used in astronomy have been examined repeatedly (e.g. Ramsey 1988), there seem to co-exist diverging results about the quantitative performance of different fibre types. We have therefore measured degradation and transmission properties of various fibres from different distributors, among them Laser Components, Mitsubishi, Polymicro, and Schott. Fibres for use in astronomy are mostly of the multimode type with core diameters between 50 and a few hundred $\mu$m. The FOCES optical design is built essentially on the assumption that the exit pupil of the microlens serves as the spectrograph entrance slit. This limits the fibre diameter to values $\le 100~\mu$m. For a resolution of $R \approx 40000$ a larger diameter would require aperture slicing. For better comparison we have examined fibres with core diameters between 50 and $200~\mu$m; the data of which are listed in Table 2.

  

Table 2: Fibres tested with respect to degradation. HCG-MO is from Laser Components, ST and STu are from Mitsubishi, FVP, FHP, and FLP are from Polymicro, and the last three fibres are from Schott. The final column provides the response of the exit light cone to an input of f/5.0

The experimental arrangement is nearly identical to that of Craig et al. (1988), and it therefore differs somewhat from most of the setups found in the literature (e.g. Avila 1988; Ramsey 1988), however, it serves well to at least provide a differential comparison of the fibres. We have taken care that the fibre input is identical to that used in the spectrograph. The fibres are cut, inserted into a holder and polished; the customized fibre ends are then plugged into a precision stage using an adapter. The stage is mounted on a rotating table such that the fibre end is always found on the axis of rotation. A stationary He-Ne laser mounted outside the table illuminates the fibre end with the angle of incidence defined by the angular position of the rotating table. The light cone at the fibre exit is not filled uniformly (except for very small angles of incidence); but its outer edge is well-defined, so the f-ratio can be determined from measurements on an exit screen without ambiguity. From comparison of the results for positive and negative input angles we estimate that our data refer to approximately 80% of encircled light.

The corresponding output light cones are represented in the last column of Table 2 for an f/5 input from which we conclude that quite generally the 100 $\mu$m fibres perform much better than the 200 $\mu$m ones. This does not seem to depend on manufacturer or cladding with the exception that the Mitsubishi 50 $\mu$m fibres have a degradation even stronger than the 200 $\mu$m ones. While similar results have been reported by Avila & D'Odorico (1988) and Guérin & Felenbok (1988) these data are at variance with the fibre properties published by Ramsey (1988). The outstanding performance of the 100 $\mu$m fibres is therefore not easy to understand. The Polymicro fibres seem to have the least degradation and were selected for further transmission studies.

  

Figure 4: Spectral transmission of selected 100 $\mu$m fibres interpolated to a length of 20 m. The fibre types are (from top to bottom): Polymicro FVP, FHP, and FLP. For better comparison they have been offset by 0.2 units

Transmission of monochromatic light was measured with a monochromator using the exit slit as spectral resolution element of $\approx$8 nm. Fibres of 25 cm and 100 m length, respectively, were homogeneously illuminated using a halogen lamp to measure the relative transmission with a photodiode. Typical results interpolated for 20 m fibre length are shown in Fig. 4 for the three Polymicro batches. Whereas the well-known OH^- absorption features in the near IR are clearly seen in the FVP and FHP fibres the red-sensitive FLP did not show it. This surprising result once more emphasizes the necessity to individually examine the batches.

3.2 Fibre confectioning

The degradation tests show that typical fibres react sensitively on strain and stresses. Since optical fibres are usually delivered with only a polyamide coat to avoid light losses, long fibre connections suffer from stresses under heavy bending. To avoid extreme degradation it is therefore necessary to put the fibre in a tube that takes most of the strain and stresses from the fibre itself. The FOCES fibres have been confectioned by wrapping the original fibre in a smoothly fitting teflon mantle, and then again wrapping the resulting tube in a steel spiral. As a result the fibres should not suffer very much from focal ratio degradation. However, first observations (Sect. 4) show a residual non-grey noise level of 0.5% when moving the fibre position with the telescope. Thus at present the spectral properties of the light transfer through the optical fibres must be considered as experimental only.

In spite of the careful confectioning the insertion of the fibre produces differential forces with respect to the tube due to expansion under gravity that can lead to disruption of the fibre. The fibre is therefore carried through a strain compensation box that is attached to the spectrograph immediately in front of the spectrograph slit. In this box the fibre runs in a few uncoated loops which allows for a compensation of the interacting forces. It also makes it possible to thread the fibres into the plugs which otherwise would be a topological problem. Thus the confectioning of the fibres turned out to be one of the most complicated tasks of the project.

3.3 Telescope module and fibre entry

The interface between the telescope and the fibre optic input at the focal plane consists of a telescope module mounted at the Cassegrain flange. It is shown in Fig. 5a.

  

Figure 5: a) Telescope module with fibre head and calibration lamps. b) Fibre positioning unit

The telescope module carries the the fibre head with a reflecting tilted circular aperture that determines the angular field accepted by the fibre, and simultaneously allows guiding on the entrance aperture using the telescope guiding facilities. It also provides the calibration lamps for flatfield and wavelength comparison exposures.

3.3.1 Entrance aperture

  Light entering the échelle spectrograph through the fibre passes through a circular diaphragm in the small tilted mirror just atop the fibre head. Three such entrance apertures can be exchanged, with corresponding input light cone (fibre f ratio) and resolution as given in Table 3. The light reflected from the aperture edges can be observed with the telescope guiding system.

  

Table 3: Single-fibre mode entrance apertures and corresponding resolution

In case of the dual-fibre mode the entrance aperture consists of 2 diaphragms with the same diameters as given above. In the telescope focal plane the holes are separated by 3 mm corresponding to 35 arcsec (2.2 m) and 18 arcsec (3.5 m), respectively. The orientation of the dual-fibre aperture is fixed to E-W direction; it can be changed only by rotating the telescope guide system.

3.3.2 Calibration lamps

At present the telescope module provides two calibration light sources that are housed in tubes attached to the module at $90^{\circ}$ distance. Their beams are designed to have roughly the same f-ratio as the telescope beam in order to fill the entrance aperture with the same light cone. The actual light source is located at the diffusor the diameter of which also determines the light cone. A field mirror tilted by $45^{\circ}$ with respect to the beam reflects the light onto the diaphragm. It can be moved into and out of the field by a sledge, and it can be rotated by $90^{\circ}$ steps to accept light from either input source. Calibration light sources are a halogen lamp for flatfield exposures and a Th-Ar lamp for wavelength comparison standards. Both lamps and the field mirror are operated remotely from the console.

3.3.3 Fibre head

  

Figure 6: Fibre feed and exit optics

As shown in Fig. 5a the fibre head is attached to the telescope module immediately below the telescope focal plane. Its detailed configuration is shown in Fig. 5b. It consists of a fibre head plug that is plugged into a positioning unit, so that the fibre head plug can be adjusted to its proper position by corresponding slides. The fibre head plug holds the head of the optical fibre and the microlens that is used to re-image the telescope pupil onto the fibre head. The imaging principle is shown in Fig. 6a, from which it is obvious that the diameter of the focal plane aperture determines the f-ratio of the light cone entering the fibre. As described in Sect. 4 the available entrance apertures should be chosen with respect to the seeing disk. This particular re-imaging construction of the light path avoids the disadvantages of a direct fibre feed that lead to imperfect scrambling of the object light depending on seeing or, more generally, on the distribution of light in the focal plane (cf. Watson & Terry 1995).

3.3.4 Fibre exit and slit assembly

The fibre exit is optically very similar to the fibre head. It is mounted into the same type of plug and fibre positioner, and it produces a scrambled image of the entrance aperture exactly at the entrance slit of the spectrograph using a microlens glued to the fibre exit surface. The corresponding principle is shown in Fig. 6b. The symmetric optical interfaces at entrance and exit of the fibre also have the advantage that the degradation of the light cone in the fibre does not overly affect the transfer efficiency. Whereas the resolution changes with the slit width, the collimator always produces an image of the full exit pupil on the échelle grating provided that the spectrograph entrance slit is fully open.

This arrangement provides a simple but efficient way to increase the resolution by narrowing the spectrograph entrance slit. The corresponding light losses are substantially smaller than those obtained with a fibre of appropriately reduced diameter or a correspondingly smaller diaphragm at the fibre head. Assuming a Gaussian seeing of full halfwidth $\sigma$, an entrance diaphragm of diameter $\theta$ will pass a fraction of the light of a point source according to

$$F(\theta)/F(\infty) = 1 - \exp\left\{- (\ln 2) (\theta/\sigma)^2\right\}$$

with light losses $\Delta$mag as follows:

  

Table 4: Light losses at the FOCES entrance diaphragm

The spectrograph slit instead cuts out a rectangular cross-section of width $\phi$ of the entrance aperture diameter $\theta$, where now

$$F(\phi)/F(\theta) = 1 - \frac{2}{\pi} \left\{\arccos x - x \sqrt{1 - x^2}\right\},$$

and $x = \phi/\theta$. Since $\phi$ determines the spectral resolution the transfer of photons through the fibre system is controlled by the product of the two formulae above. Thus for a given spectral resolution there are three ranges of seeing that require a particular choice of an optimal entrance aperture as shown in Fig. 7 for R = 40600. Note that the use of the diaphragm with 130 $\mu$m diameter -- although fitting exactly to the resolution -- is only suggested for a seeing below 1 arcsec. The larger diaphragms are considerably better for medium or bad seeing. Of course, no light is lost even with bad seeing if the resolution is adjusted accordingly by opening the spectrograph slit.

  

Figure 7: Light losses at the 2.2 m telescope at R = 40600 with different entrance apertures. The choice of the appropriate aperture is indicated by the continuous part of the curves