next up previous contents
Next: Detailed FLAIR-II Plateholder Up: Flair User Manual - Previous: FLAIR overview

3. Planning observations with FLAIR

3.1. Fibre-feed options

There are three FLAIR fibre feeds, one mounted in plateholder 14/5 and two (with differing core-diameters) in 14/6. Details are given in Table 6. Note that approximate spectral widths and inter-fibre spacings are given in pixels by simple geometrical considerations (e.g. camera magnification factor of 0.39 coupled with the spacing of fibres on the slit with edge-edge separation of 110um and CCD pixel size of 22um). No account is made of scattered light contributions or other shortcomings of the optics which introduce signal in the inter-fibre areas. Empirical results indicate that ~ 4% of the flux from a given fibre will contaminate an adjacent fibres signal. This is not usually a problem in most cases except when very bright and faint objects occur next to each other on the slit or if very strong emission lines are involved. (as with planetary nebulae). In such cases use of alternate fibres is recommended.

If required, the two 14/6 feeds can both be fibred-up on the same plate, but they can only feed the spectrograph one at a time. Normally, only one feed is fibred up and the ferrules of the other are parked on the plate surface with double-sided adhesive tape to avoid risk of damage. However interchanging slit-vanes in 14/6 can enable up to 144 objects to be observed without recourse to changing plateholders and may thus be of benefit for some projects (e.g. galaxy redshift surveys).

Table 6: FLAIR fibre feeds

The larger fibres are generally used for galaxies (B < 17.5) and brighter stars (B < 16). The smaller diameter fibres were originally intended for use with faint point source observations since they accept ~ 3.5 x less sky background. This was expected to lead to improved S/N for faint objects. However, in practice it has been found that this advantage is more than offset by other problems inherent with the smaller diameter fibres. These problems include reduced tolerance to positioning errors, focus errors, temperature variations, CCD pixel undersampling and a greater susceptibility to atmospheric refraction and dispersion which reduces the range of hour-angles that can be worked (see section 3.5). There are also additional complications as a result of an unforeseen problem with the removal of the protective nylon outer-jacketing of the 55um fibres. This resulted in an unwanted staggering of the fibres on the output slit, necessary if the expected fibre numbers were to be retained. The staggering leads to alternately shifted wavelength information which complicates the data reduction. The fibre spacing is also non-uniform with an additional `bunching' of fibre pairs along the slit leading to increased cross-contamination of signal between adjacent fibres. Full details can be found in the AAO internal report on 55um vs 100um fibre performance by Taylor & Parker (copies available on request). The staggering, bunching and contamination problems can be circumvented by using alternate fibres on the slit at the expense of reduced fibre numbers. This technique is generally useful as a means of reducing adjacent fibre spectral contamination when both bright and faint objects are being observed or where the full multiplex advantage is not required. The recommended use of the 55um fibre bundle is now for those occasions where both slit vanes in 14/6 can be effectively used for bright objects where the target number density is high.

In 14/5, the 152-fibre capacity of the plateholder is under-used, so some of the fibres can utilize the full length of their storage chambers, rather than stopping half-way across to accommodate the fibre loop coming in from the other end of the plateholder (see Section 2.4). The effect of this is that they can be pulled out to the full width of the plateholder, whereas the others reach little more than half-way across. There are 42 ``full-width'' fibres, together with the five fiducial fibres. The half-width fibres are ferrule nos. 8-19, 27-39, 56-67 and 73-85; they are highlighted on the plateholder itself. In 14/6, the full capacity is used and all the fibres (including the eight fiducials) are half-width.

All the feeds are made with Polymicro-FH fibre, a ``wet'' ( i.e. high OH-) type fibre giving good UV/blue performance at the expense of increased absorption in the far-red OH bands. The spectral transmission of a 25-metre length is given in Figure 8. (The actual length used in the FLAIR feeds is approximately 11.5 metres.) The 100um fibres are jacketed with a relatively tough polyamide/acrylate combination, giving them a gold colour, while the 55um ones have a strong colourless acrylate/nylon jacket. (All the fiducials are coloured blue, and are the most delicate of the three types.)

Figure 8: Spectral transmission of FLAIR Polymicro-FH fibre.

The fibre ferrules are numbered in a regular sequence in the plateholders (the 100um fibres in 14/6 being differentiated from the 55um by an accompanying dash). However, the fibres themselves are numbered in order at the slit end, and the two numbering systems are different. Look-up tables in the AutoFred software relate the fibre-number along the slit (which corresponds to the spectrum number in a data-frame) to the ferrule number for each feed; they are also given here in Table 7. The fibres are not perfectly straight along the slit and, in 14/6-55um, are grossly staggered. In 14/6-100um, eight fibres at each end of the slit are double-spaced.

Table 7: Fibre/ferrule data

3.2. Spectrograph options

Apart from the two dedicated FLAIR gratings (300B & 600V) that will always be provided, the FISCH spectrograph can utilise any standard AAO grating so a wide range of dispersions and blaze wavelengths are possible. Gratings are identified by reciprocal groove-spacing and blaze wavelength, and the most useful for FLAIR use are 250B, 300B, 270R, 600V, 600R, 1200B, 1200V, 1200R and 1200I. The codes B, V, R and I denote blaze wavelengths (in Littrow configuration) of 4300Å, 5000Å, 7500Å and 10000Å respectively. Details of the performance of these gratings can be found in Efficiencies of the AAO Diffraction Gratings (AAO UM 19) by H.Johnston, P.Gray, R.Stathakis and R.Robinson.

Straightforward derivations from the grating equation give useful formulæ describing the various configurations of the gratings. The grating angle, (angular separation of the grating normal and the camera axis), for a given central wavelength, (in Å), is given by

x ,

where m is the diffracted order, is the grating constant in grooves/mm and (=45 deg) is the fixed angular separation of collimator and camera axes. If is the angle of incidence on the grating, then , and is positive when gratings are used ``blaze-to-collimator'' and negative if ``blaze-to-camera''.

The reciprocal linear dispersion in Å/mm is


where (=125 mm) is the effective focal length of the spectrograph camera. The anamorphic or slit projection factor is


and the projected width of a fibre of diameter D on the detector is


where (=316 mm) is the collimator focal length. Multiplying the projected width by the reciprocal dispersion, , gives the instrumental resolution in Å. The CCD resolution and spectral range follow from the pixel-size (22um) and array width (578 pixels) of the chip.

With FLAIR the gratings are normally used only in first-order. However a second order filter is available for use with grating 270R to eliminate 2nd order blue light contaminating the red end of the spectrum. There is also provision within the CCD system for binning data on the chip in the spectral direction. This is occasionally useful for modifying the effective spectrograph resolution whilst also reducing the read-out noise. Note, though, that binning should be used only with 100 um fibres and with gratings used ``blaze-to-camera'', otherwise the instrumental resolution will be effectively undersampled. Likewise, the 55 um fibres require ``blaze-to-camera'' and no binning. In Table 8, calculated parameters are given for a representative selection of permissible grating configurations. There is, of course, a wide range of other possibilities.

Table 8: Some typical first-order grating configurations

Changing the grating angle for the appropriate wavelength range is carried out using groove settings on the grating table for initial angle selection followed by fine motion adjustment with a micrometer for more precise wavelength range selection. Plots of the groove and micrometer settings for various gratings and desired wavelength ranges are held at the telescope.

3.3. Wavelength calibration

Table 9 gives wavelengths of the principal calibration lines for the main lamps used with FLAIR. (Weak lines are shown in parentheses.) Reference spectra for these lamps are held at the telescope.

Table 9: Calibration sources-Wavelengths

3.4. Sensitivity

With the commissioning of the new thinned CCD impressive performance gains have been obtained. In the area of galaxy observations the previously illusive CaH&K absorption features are now easily visible in the blue. Figure 9 gives a co-added FLAIR frame of 5 x 2000sec exposures of ~ 80 galaxy spectra from a magnitude limited survey to Bj =17.5 of Parker. These data were taken in July 1995 with the new CCD and grating 300B (5.1Å/pix). Wavelength increases from left to right over the range ~ 4000-7000Å. Note the night-sky emission lines of [OI]5577Å, NaD and [OI]6300Å & 6362Å occurring as vertical stripes at the same wavelengths. Reduction of a single such 2000sec FLAIR exposure gave >95% redshift success (includes random morphologies). In general, the few indeterminate velocities are for late-type objects with no strong features. With grating 250B and a resolution of ~ 5Å, absorption-line velocities to ~ 140 km/s and emission-line velocities to ~ 60 km/s can be obtained. These new results compares with the typical 5 x 3000sec exposures needed with the old CCD to achieve just ~ 80% completeness for galaxies at Bj ~ 17!

Figure 9: Cleaned image of approximately 100 galaxy spectra obtained with the new CCD in July 1995 for a redshift survey of Parker. The data represent 5 x 2000sec exposures.

Figure 10 gives some individual reduced galaxy spectra from these data demonstrating the excellent S/N, good blue response and strong absorption features now evident. The Y axes gives average counts per 2000sec exposure whilst the spectra themselves result from combining 5 x 2000sec exposures.

Figure 10: Galaxy spectra obtained with new FLAIR CCD. Average counts from a single 2000sec exposure

Figure 11 gives a plot of average S/N as a function of COSMOS  Bj magnitude for ~ 80 galaxies taken from a single 2000s FLAIR exposure over the region 5000-5500Å. This plot illustrates the good S/N values that can now be obtained in 2000sec with the new CCD. The error bars represent the standard deviation from each average at the given magnitude bin. Although a decreasing trend is evident one can see that average S/N is not a strong function of magnitude over the range  Bj ~ 15.8-17.5. This is because at  Bj ~ 15.8 not all of the galaxies visible disk is typically being sampled by the large 6.7arcsec diameter fibres. As one goes fainter the situation is compensated by a larger fraction of a galaxy's projected diameter being sampled by the fibre. Consequently the average object surface brightness sampled by each fibre remains relatively constant over this magnitude range. Equivalent results for stellar spectra at different resolutions are currently unavailable but similar impressive gains in S/N are anticipated. Details should be found in issues of the AAO newsletter and UKST staff can also be consulted.

Figure 11: Average galaxy S/N over the wavelength range 5000-5500Å as a function of COSMOS  Bj magnitude

The signal-to-noise ratio in each pixel of an extracted FLAIR spectrum is given by


where the F's are fluxes/pixel in the final spectrum (in adu) and r is the read-out noise of the chip ( ~ 10e corresponds to 10 ADU, since the CCD gain ~ 1). Typically, for a galaxy with B ~ 17, an exposure of 2000sec will produce ~ 1100 adu and ~ 720 adu at 5000Å, and hence a signal-to-noise ratio of ~ 25. With the new CCD a minimum exposure-time per field for galaxy redshift work to B ~ 17-17.5 is around 1hour (this allows for 2 x 2000sec exposures to facilitate automatic cosmic-ray removal).

Emission-lines in faint point-sources (quasar candidates) have been observed successfully to B ~ 18.5 using 100um fibres and the lowest available resolution (12.5Å/pixel). These observations were sky-limited, and were obtained in clear skies with good seeing. The new thinned CCD should permit more satisfactory observations of such sources in less optimum conditions.

3.5. Limits to observing

A number of effects conspire to produce a mis-match between the images of the target objects formed in the telescope focal-surface and the pre-set array of fibre input-faces. Generally, they are exacerbated by the wide field of the UKST, and introduce some limits to efficient observing. Most of the effects also become worse when prolonged observations over many hours on the same field are required. This situation is now greatly improved since the new thinned CCD can produce the same S/N in 2000-4000sec that previously required 18000s!

Temperature effects manifest themselves in FLAIR in a change of plate-scale caused by thermal expansion of the copy-plate supporting the fibres. The important temperature-difference is that between the exposure of the original master plate and the use of the resultant copy-plate in the telescope, since the expansion coefficients of both plates and the intermediate glass positive are identical. The expansion coefficient is small (8.1e-6 /°C), but a temperature difference of 20 °C would move a fibre in the plate corner by 40um relative to the plate centre, an unacceptable amount. Fortunately, FLAIR copy-plates tend to be in use at about the same time of year that the original master plates were taken, and temperature differences of this magnitude are unlikely to occur; however, a check should be made when observing.

Differential atmospheric refraction causes an apparent distortion of the telescope field while it is tracked across the sky. Reference has already been made to the elimination of refractive field-rotation by elevation of the UKST polar axis, but the distortion itself remains. Its magnitude can be estimated from the diagrams given in Figure 12, which show the refraction-induced trail rates at four positions of the telescope field plotted as contours over the accessible sky. (They assume the field centre is tracked on the refracted image of a central star, as is the case when the autoguider is used.) Over much of the sky, the trail rates are greatest along the E (and W) edges of the field (b, c, d); they are not in simple proportion to off-axis distance. The position-angles indicate that the nature of the distortion varies across the sky. Near the meridian it is a diagonal shear.

Figure 12: Contours of equal trail-rate (in arcsec/hr) over the southern sky, showing the rates and instantaneous position-angles of the trail suffered by an image in each of four positions in the field of the UKST (assumed to be 6 deg x 6 deg). The positions are (a) N edge mid-point; (b) NE corner; (c) E edge mid-point; (d) SE corner. The quantity e is the polar-axis elevation.

It is possible to estimate the actual trail suffered by a point image during a given exposure by integrating along the appropriate line of constant declination. For 55 um fibres, assuming a 1 arcsec seeing disc, a trail of some 2.7 arcsec is permissible before light is lost to the cladding (ignoring all other possible sources of error). Thus, even for objects along the E and W edges of the field where the trail is greatest, there is a workable range of hour-angles of about 2.5 hours on either side of the meridian. (Note that the smaller fiducial fibres would lose light over less than this range, but they are not usually positioned at the edges of the field.) For the 100um fibres, this range extends to nearly five hours on either side of the meridian (particularly since they are generally used with extended objects, where the positioning is less critical), and this is the normally-accepted range of observation for these fibres.

Atmospheric dispersion presents some difficulties with the 55 um fibres at zenith-distances greater than 45 deg. (At lower zenith distances, the effects are comparable with those of seeing.) A complicating factor is the red-response of the acquisition TV camera (S25, peaking at ~ 6500Å), and it may be necessary to select early-type fiducial stars (or introduce a blue TV filter) when blue spectral-range observations at high zenith-distances are being made with the smaller fibres. The 100um fibres are immune to this effect.

Instrumental effects These include fibre positioning errors, fibre-ferrules which do not lie quite flat on the plate and therefore intercept the fibre beam at an angle (can occur during fibre-ing unless care is taken) and focus problems introduced when too much UV curing cement is placed under the ferrules teflon pads (the glue thickness was not taken into account in the focus calculations for the fibre-prism assemblies). Again the 100um fibres are more tolerant to these effects than the smaller diameter fibres. Further details can be found in the AAO internal report by Taylor & Parker referred to earlier.

Finally, though it is not a field-dependent effect, proper-motion of the fiducial stars may introduce significant acquisition errors. In principle, high proper-motion stars can be eliminated by consulting the SAO Star Catalog, but the most fool-proof method is to use a master plate (for the FLAIR copy) taken within the previous two or three years. Typically, FLAIR copy plates are currently made from Southern J Survey plates, on average about 15 years old, but the more recent Second Epoch Survey red plates are used if available.

3.6. Applying for time

FLAIR time is awarded by the Australian and British time-allocation subcommittees of the Schmidt Telescope Panel. Australian applicants may obtain forms and up-to-date technical information from:

UK Schmidt Telescope
Anglo-Australian Observatory
Coonabarabran, NSW 2357
Tel: 068 426291; Fax: 068 842298

In Britain, the contact address is:

UK Schmidt Telescope Unit
Royal Observatory
Blackford Hill
Edinburgh, EH9 3HJ, UK
Tel: 031 668-8325; Fax: 031 668-8264

LaTeX versions of these forms can also be obtained on request or by anonymous ftp to the AAO vax where it can be found in AAODOCS. Completed applications should be returned to the appropriate contact address. Applicants from countries other than Australia or Britain may apply to either panel. Deadlines for applications in both countries follow the current ATAC dates for the August-January and February-July Semesters. General information about FLAIR, including technical updates and scheduling information, appears regularly in the AAO Newsletter, and on the AAO's UKST WWW pages. Interested parties are encouraged to contact UKST staff at the AAO or UKSTU staff at ROE at any time for advice and news of latest developments.

next up previous contents
Next: Detailed FLAIR-II Plateholder Up: Flair User Manual - Previous: FLAIR overview

Chris Tinney
Mon Aug 19 14:56:26 EST 1996