next up previous contents
Next: Planning observations with Up: Flair User Manual - Previous: Introduction

2. FLAIR overview

 2.1. The U.K. Schmidt Telescope

A full description of the telescope can be found in the UKST Handbook. The main optical parameters are summarised here in Table 1. Mechanically, the telescope is arranged on a classical equatorial fork mounting, with the unusual addition of a remotely-controlled polar-axis jack to allow the instrumental pole to be offset from the true pole. This is to introduce a field-rotation to compensate for the gross effects of atmospheric refraction over the telescope's wide field. A full account of this and other atmospheric effects in relation to multi-object spectroscopy with the Schmidt is given in Watson (1984 and Phd thesis, 1987); see also Section 3.5.

Table 1: U.K. 1.2-metre Schmidt Telescope

The telescope was built for the specific purpose of carrying out deep photographic surveys of the southern sky. When it entered service in 1973, it was not envisaged that it would perform any function other than taking photographic plates, so access to the focal surface can only be obtained through the use of the 397 mm x 397 mm x 56 mm plateholders. These are loaded by means of a mechanical elevator whose lower station is at handling height when the telescope is in its Access Park position.

The telescope is operated via an electromechanical control system from a console enclosure in the dome. The acquisition controls and monitors for FLAIR are also located here. There is access to the dome for heavy (but not large) items such as the FLAIR plateholders by a dumb-waiter, which communicates with the ground and first floors of the building.

From the point of view of optical-fibre coupling, the Schmidt telescope has both advantages and drawbacks. Among the former are the perfect telecentricity of the optical system, which greatly simplifies the beam coupling into the fibres. Likewise, the fast focal ratio is well-matched to all-silica fibres (whose numerical aperture is typically 0.22). Against these are the fine plate scale (requiring high positional accuracy), the steep focal curvature, and the inaccessibility of the focal surface.

2.2. Development of the FLAIR system

The suggestion of a multi-object spectroscopy system for the UKST was made as long ago as 1982. Work began the following year, first on quantifying the scientific usefulness of such an instrument (Dawe & Watson, 1984) and then on building it (Watson & Dawe, 1984). In August, 1985, a 39-fibre prototype FLAIR system saw first light, recording spectra photographically on 35 mm film (Watson, 1986). An arrangement with the Physics Department of Durham University led to a cryogenic slow-scan CCD camera system (an ancestor of the present CCD) being delivered in 1986. March, 1987 brought the start of a limited common-user service, but the poor optical efficiency of the prototype restricted its usefulness until an improved version known as PANACHE (PANoramic Area Coverage with High Efficiency) was introduced a year later (Watson, 1988).

Though still experimental, this demonstrated more convincingly the potential of multi-fibre spectroscopy with the Schmidt. SERC funding was obtained to develop a more fully-engineered version, based on PANACHE but with the capacity for simultaneous observation of up to ~ 100 objects. Known as FLAIR II, the new version was begun in 1989 and completed in 1992, the old PANACHE system maintaining the spectroscopy service throughout this period (Watson, Oates & Gray, 1990). With the commissioning of FLAIR II in March 1992 the system at last began to realise its potential. Commissioning of a new thinned CCD in July 1995 has further improved performance (see later). The current system can now be considered as 12 x more efficient for galaxy redshift projects. FLAIR now offers 3 x the old fibre numbers with consecutive nightly operation, a new spectrograph, semi-automatic fibre positioner, two FLAIR plateholders (which can both be utilised on the same night) and the new CCD (e.g. Parker & Watson, 1994, 1995).

FLAIR's first major science papers were Parker & Watson (1990) Watson, et al. (1991) for galaxy redshift work, and Morgan, Watson & Parker (1992) for stellar observations. FLAIR II itself and subsequent developments are described in detail by Watson et al. (1993a,b), Bedding, Gray & Watson (1993), Watson & Parker (1994) and Parker & Watson (1995).

The following people have been involved with the building and commissioning of the FLAIR II system together with subsequent enhancements and developments: John Barton, Ian Bates, Robert Dean, Frank Freeman, Peter Gray, Allan Lankshear, Paul Lindner, Don Mayfield, André Porteners, Doug Pos, Neal Schirmer, Greg Smith, and Denis Whittard (AAO), Eric Coyte, Bill Green and Michael Kanonczuk (Australian National University), Paddy Oates (RGO), Tim Bedding (Sydney University) and Quentin Parker (AAO/ROE, FLAIR Instrument Scientist 1992-) and Fred Watson (AAO/ROE, Project Scientist 1985- early 1992). Support from the Schmidt Telescope Units in Edinburgh and Coonabarabran is gratefully acknowledged. For brevity, FLAIR II (the current version) is referred to simply as FLAIR in the remainder of this Handbook.

2.3. Fibre positioning method

At the heart of FLAIR is a fibre-positioning technique that is unique to the Schmidt Telescope. Within the 356 x 356mm ( field of view, the fibres need to be positioned to an accuracy approaching 10 um ( ~ 0.7arcsec) if the fibres are to effectively intercept the light from the target objects due to the fine plate-scale of the telescope (67.14arcsec/mm). This is achieved by back-illuminating the fibres, and cementing them onto a standard 1-mm thick, 14-inch square glass copy-plate of the target field, in exact alignment with the selected objects visible on the plate. Each fibre is terminated with a protective ferrule and a 2-mm aperture input-end right-angled prism, so that the fibres themselves lie along the front surface of the glass. The ferrules are merely `tacked' in position with a rapid UV-curing cement and, after observing, they can be removed for re-use with another target field.

2.4. Plateholders

The plate and fibres are supported by specially-designed plateholders that are loaded into the telescope in much the same way as a normal photographic plateholder (see Figure 1). They are backed with solid aluminium mandrels machined to the telescope's focal curvature. The glass field-plate with the affixed fibres is bent over this when the plateholder is tensioned prior to loading in the telescope.

Figure 1: Cross-section of a FLAIR plateholder, showing fibre storage chambers behind the field plate. Uppermost is the curved mandrel supporting the plate in the focal curvature (convex to the telescope mirror). Dimensions are in mm.

Each fibre passes from the front of the plateholder through a length of teflon tube around the edge of the glass field-plate and into its own 25 mm high x 0.7 mm wide storage chamber, where it forms a single loop as indicated in the diagram. Thus the fibres can be retracted into the body of the plateholder and only sufficient needs to be drawn out to reach the target object. From the storage loops, the fibres are gathered together and emerge from the plateholder in an Ultraflex cable which is anchored to the plateholder by the restraint shown at left in Figure 1. The fibres are not self-retracting but can be easily pushed back after use; their maximum extension in use is limited by the rigid 25-mm diameter ring shown in the diagram. Because the top clamp (uppermost section of the diagram) has to be lifted to allow insertion and removal of the glass copy plates, the teflon tubes themselves must slide within the plateholder. Fibre positioning is carried out with the plate supported flat; the top clamp is then lowered to bend the plate to the focal curvature.

An additional complication of the plateholders is that they need to rotate (through ~ ±15 deg) for field acquisition; the complete assembly, including storage chambers, is mounted on a centre-bearing in the plateholder's main frame and driven by an integral motor-micrometer with position read-out (controlled from the telescope console-room). Like the photographic plateholders, the FLAIR versions are fitted with dark-slides to facilitate carrying and loading in the telescope. Before commissioning, focus plates are taken to determine the focus for the new FLAIR plateholders, a difficult operation if the fibre feeds are fitted.

The plateholders contain 76 storage chambers and, since corresponding arrays of teflon tubes are fitted along both the N and S edges (so that two fibres can share a chamber), up to 152 fibres can be accommodated. There are currently two FLAIR plateholders (though a third is currently planned), designated 14/5 and 14/6, each equipped with their own fibre feeds. These are permanently mounted, because the individual fibres have to be threaded through their storage chambers before the ferrules and microprisms are fitted. Details can be found in Section 3.1.

2.5. Fibre feeds

The fibre input-end ferrules are machined from disposable hypodermic needles; the 5 mm x 3 mm base of each ferrule carries a teflon pad to ensure a smooth action on the plate when moving into position and a clean release from the UV-curing cement holding it on the copy plate. The 2-mm aperture 90-deg micro-prisms at the input ends have a rearward extension so that the virtual images of the fibre ends in the prisms (onto which the incoming light is focused) lie in the same plane as the plate surface, i.e. in focus. This avoids a change of scale when the plate is tensioned. The glass prisms are made from SF5 glass which have a higher refractive index than the old discontinued BK7 variety and so are able to accept the entire f/2.5 beam without loss but require a somewhat larger rearward extension (3.55mm c.f. 3.25mm). These new prisms also have an anti-reflection coating on their top surface to further reduce losses. Empirical results from the commissioning of these new SF5 prisms indicate a significant performance benefit in overall system throughput of ~ 30%. Figure 2 gives a side view of a FLAIR ferrule. Note the extended SF5 prism

Figure 2: Side view of a FLAIR ferrule. Note the extended prism.

The Ultraflex cable carrying each of the main fibre feeds is about 10 metres long so as to reach the floor mounted spectrograph at all attitudes of the telescope. At its output end, each feed carries a compact slit unit in which the output ends are aligned to form a 20-mm long (approx.) fibre `slit'. This is mounted on a vane that locates in the focus of the spectrograph collimator.

Details of the three available main fibre feeds can be found in Section 3.1; they have all been made as nearly as possible confocal within the spectrograph collimator.

2.6. Acquisition system

Figure 3: FLAIR acquisition and guidance system.

Although the UKST has relatively good setting accuracy ( ~ 10 arcsec), there is no fixed mechanical relationship between the telescope and the glass field-plate, so an optical acquisition system is used to register the real image formed in the telescope with the array of fibres. This consists of an acquisition image-guide (coherent fibre bundle), and a number of fiducial star fibres, all of which feed a telescope-mounted low light-level intensified CCD camera via a microscope. Once the field is acquired, it is maintained in registration by the telescope's autoguider.

The system is shown schematically in Figure 3, and details are given in Table 2. Note that the output slits of the fiducial bundles for the two plateholders (14/5 and 14/6) are not confocal in the microscope, so that refocusing is necessary to maintain coincidence of focus with the image-guide output face when changing from one plateholder to the other.

Table 2: FLAIR acquisition system

At the plateholder end, the 33 um fiducial fibres are retractable in the same way as the main fibres, and terminate in identical ferrules and micro-prisms. The 1.1-mm diameter acquisition image-guide plugs into a 14 mm x 7 mm x 5 mm socket equipped with a micro-prism carrying a fiducial mark on its exit face. During fibre set-up, the socket is cemented to the copy plate with the image of the fiducial mark in alignment with an acquisition star. (The geometry of the arrangement is such that the micro-prism requires additional optical thickness over the standard micro-prisms and also an image reversal; thus it has the unusual form of a single Porro-prism of the second class with an exit-face extension.)

In use, the acquisition star is centred on its fiducial mark, and the plateholder rotated until the fiducial fibres light up. This is a two-stage process, since each fiducial will illuminate faintly when light falls on the large-diameter optical cladding, before lighting up fully when the fiducial star is centred (see Table 2). Full details can be found in Part.2 of this handbook.

2.7. Fibre positioner

Bedding, Gray and Watson (1992) describe in detail the ``AutoFred'' machine-assisted positioning system. The device comprises a computer-controlled  (x, y, z)  fibre-positioner with a CCD TV microscope and frame-grabber to sense the position of the back-illuminated fibres. A schematic of the control system is shown in Figure 4. AutoFred is used in conjunction with the fibre-positioning table, whose (x, y) carriage supports the plateholder and copy-plate, allowing all areas of the plate to be brought manually within the field of the TV microscope.

Figure 4: AutoFred control system. Arrows show the flow of control and data.

In use, the operator moves the plateholder carriage to bring the target object within the TV field, and marks its position with a cursor. A fibre ferrule is loaded into the gripper, and AutoFred then lowers the ferrule to plate level and adjusts its position until the centre of the illuminated fibre coincides with the marked position to within a pre-determined error (usually 1 pixel, or ~ 10 um). A UV shutter is manually opened to cure the cement whereupon the system releases the gripper and writes the fibre and object identifications to a log file before proceeding to the next fibre. Full details are given in Section 4, while the principal files used by the AutoFred PC in fibre positioning are summarized in Table 3.

Table 3: AutoFred system files

2.8. Spectrograph

A fibre-coupled spectrograph for the UKST demands a fast collimator to accept the ~ f/2 beam (degraded from f/2.5 by focal-ratio degradation within the fibres). It also needs a wide field to accept a large number of fibres and a large collimator/camera focal-length ratio to fit the fibres across the detector and maximize camera speed and resolution. The spectrograph for FLAIR borrows from the design of the telescope itself in utilizing Schmidt-type optics, and satisfies all three of these requirements (see Figure 5). Christened FISCH (for FIbre-SCHmidt) by its designer, Peter Gray, the spectrograph has an f/2.1 collimator feeding an f/0.9 camera via a 150 mm x 200 mm plane reflectance grating of the type used by the AAT's RGO and 2dF spectrographs. Two FLAIR specific gratings are to be purchased (probably the most commonly used types 300B and 600V) and so will be always available. However the full range of AAO 2dF gratings are potentially available for use at the UKST in a wide range of blaze-angles and groove-spacings. The Camera and collimator optics are on indefinite loan from the Royal Greenwich Observatory. Some details of the spectrograph are given in Table 4, while gratings, resolution, etc. are specified in Section 3.2.

Figure 5: FISCH spectrograph-general assembly - click on the image to see a larger version.

Table 4: FISCH spectrograph

The fibre slit is mounted at the internal focus of the collimator; the slit is polished flat (rather than to the focal curvature of the collimator) but produces no serious loss of image quality. Likewise, the fibres are mounted parallel in the slit rather than fanned, producing only a minor increase in vignetting. The camera mirror is enclosed in an evacuated chamber forming part of the CCD cryostat, while the camera Schmidt-corrector provides its front window. A liquid nitrogen dewar is mounted alongside and cools the CCD chip through a cold-finger of braided copper wire enclosed in a connecting vacuum-bellows. The camera design is such that the central obstruction has been kept to an absolute minimum, amounting to ~ 10%. External micrometer adjustments are provided for CCD focus and tilt, and the entire camera body can be rotated through a few degrees to render the detector columns parallel with the spectra. A field-flattening lens is required in the camera; like the two Schmidt corrector-plates it is anti-reflection coated, while the collimator and camera mirrors are, respectively, aluminized and silvered.

FISCH's camera/cryostat is built as a single assembly, but the remaining components are simply mounted on an optical breadboard supported by the vibrationally-isolated table in the dome. This arrangement results in very high stability. A sheet-metal cover with a large access door keeps the spectrograph reasonably light-tight, while the camera itself has a clip-on dark-cover. The spectrograph slit shutter is a small ``flag'' operated by a rotary solenoid, positioned immediately downstream of the fibre slit. Both the camera cover and slit shutter have hard-wired status circuits attached to dual colour LED's situated by the FLAIR console control area. These are independent of the CCD control software logic.

2.9. CCD system

A new thinned, back-illuminated CCD for FLAIR was successfully commissioned in June 1995 to address the poor blue performance of the previous chip which hampered a fuller exploitation of FLAIR. The new EEV CCD-02-06 device is a generic derivative of the previous P8603 variety though with a 20 instead of 30 pin out. However it has the same basic format (578 x 400) and pixel-size (22um) for ease of implementation and is still compatible with the existing electronics. Hence a rapid and cost effective upgrade was possible without recourse to expensive redesign of the CCD camera or electronics. The new CCD is the most significant development of the system since FLAIR-II came on-line in early 1992 and has had an immediate and dramatic effect on system capability. The DQE curves for the old and new CCD's are given in Figure 6 after being characterised in the laboratory at RGO by A.P.Oates. A depth gain of ~ 1 magnitude in the blue was thus expected with ~ 0.5 magnitudes in the red.

Figure 6: DQE plots of the old and new EEV CCD's for FLAIR

Figure 7 gives an empirical sensitivity curve derived by comparing data from the new CCD with the old CCD as a function of wavelength from real FLAIR exposures on the telescope. The Y scale indicates the factor improvement over the old CCD. This curve has been obtained by dividing an average dome flat field spectrum from all fibres obtained with the new CCD from an equivalent average from the old CCD with the same wavelength range, grating and exposure times. These dome flats obtained from a quartz-halogen lamp and flat-field screen have previously been shown to be very reproducible in overall shape though signal repeatability depends on the screen always being illuminated in exactly the same way. Hence although the curve shape is accurate there is some uncertainty in the exact values of the Y scale though constant screen illumination is always attempted. Note the dramatic improvement in sensitivity in the blue. Also note that the sensitivity ratio drops back to ~ 1.5 at the extreme edges of the data. This is expected since there is some framing of the back thinned area at the edges of the CCD so that performance here should be similar to the old CCD. This also gives us confidence that the sensitivity comparison curve Y-axis is good to ~ 50%. The sensitivity comparison also includes the effects of benefits from a new AR coating on the CCD camera corrector window and the new SF5 prisms.

Figure 7: Comparison of the sensitivity of the new and old CCD as a function of wavelength. The Y scale factor is unlikely to be in error by more than about 50%

The CCD camera is operated in slow-scan mode (see Table 5) and is controlled by a 40 MHz 80386 PC running GEM windows, communicating with the CCD sequencer via a 48-bit I/O card and an opto-isolated CCD interface (see Oates 1990).

Table 5: CCD camera

The WIMP (Windows, Icons, Mice and Pull-down menus) environment of the GEM desktop makes an ideal user interface for the CCD system, and for the image-processing facilities available. The PC is connected by Ethernet to the Sun workstation and AAO Vaxcluster, enabling files to be transferred to (or from) the bigger machines for reduction or output. Control of the CCD is either performed from a dome terminal or, more conveniently, from the FLAIR console control desk in the UKST's common-room, where a duplicate FLAIR acquisition TV display and other facilities are situated.

Full details of the system are given in the FLAIR CCD System User Guide together with details of data transfer to the Sun or Vaxcluster.

2.10. Calibration

There is no separate calibration facility within the FISCH spectrograph. Instead, all calibration exposures are made via the telescope and main fibre feeds-a better arrangement as it means the light path is identical to that of the target objects. The excellent stability of the table-mounted spectrograph while it remains undisturbed means that far fewer calibration exposures are required than with a conventional spectrographic instrument.

A set of wavelength calibration arcs are permanently mounted as an array in a specially designed box on the telescope access platform from where they can simply illuminate the inside of the dome or flat-field screen. The lamps are conveniently controlled from the FLAIR console desk in the UKST common room. The range of lamps includes He, Hg-Cd, Ne, Na and Rb; line wavelengths for some of these are given in Section 3.3. Operational procedures for obtaining arc exposures are given in part.2 of this handbook.

Fibre flat-fields can be obtained by reflected white-light off the specially positioned dome flat-field screen (``dome flats'') which is obliquely illuminated. The light source is a quartz-halogen lamp placed on the dome floor which approximates a black-body curve at ~ 5000 K to provide a featureless spectrum. Flat-fields should also be obtained from the zenith sky during twilight if sufficient time is available. Empirical comparisons of the normalised transmission variations of the same fibres from carefully positioned dome flat-fields and equivalent twilight sky exposures give agreement to ~ 2%. Flat-field exposures are essential to the observing process as they enable normalisation of the inherent fibre-fibre transmission variations and thus the ability to perform satisfactory sky-subtraction from the dedicated sky-fibres. Again, procedures are given in Part.2.

Occasionally, it is necessary to obtain a spectrograph flat-field in which the CCD chip is illuminated with a uniform or slowly-varying field (e.g. when checking the read-out noise and gain values). This can best be performed by replacing the grating with a sheet of white card and illuminating it via an accurately flux controlled red LED positioned at the internal focus of the main collimating mirror.

Flux-calibration of target objects or observations of standard objects can also be performed with the FLAIR system by use of a dedicated offset fibre and a rather novel acquisition technique which relies on use of the Image Guide Socket. Such flux or radial velocity standards do not need to be in the current FLAIR field. The FLAIR support astronomer will take care of this procedure should standards be required. Full details are given in part.2.

2.11. Future Developments

With the commissioning of the new thinned CCD FLAIR is now an even more effective means of performing statistical wide-field surveys for a variety of astrophysically interesting objects to moderately faint limits. The largest overheads now rest with the laborious off-telescope fibering-up system. Currently between 4 and 6 hours are required to fibre-up a typical field with ~ 100 fibres. With a fully-engineered pick-place magnetic button system observations of 6 or more fields a night become possible, potentially providing more than 600 target spectra. Consequently investigations into the provision of a much more automated fibre-positioning system based on derivative 2dF technology are underway. Such a system, dubbed FLAIR-III, would enable rapid field reconfiguration. Coupled with the provision of dedicated FLAIR gratings, new plateholders and possible further detector and spectrograph upgrade options there is plenty of scope to develop FLAIR into a system which can maximise the scientific productivity of the UKST for performing multi-object spectroscopy.

next up previous contents
Next: Planning observations with Up: Flair User Manual - Previous: Introduction

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