|Home > Astronomy > AutoFib2 > Target acquisition|
Target acquisitionTarget acquisition is one of the trickiest aspects of multi-fibre spectroscopy. With careful attention to astrometry, the AF2 automatic-acquisition tool commissioned in 2013A delivers the required accuracy. Commissioning tests showed acquisition errors below 0.4 arcsec for 90% of the fibres in the central 20-arcmin radius of the AF2 field of view, when using 7 to 10 fiducial stars well distributed over the field of view. This accuracy is adequate given the fibre diameters and the typical seeing conditions on La Palma. For reference, an offset of 0.5 arcsec leads to a 20% flux loss when the seeing is 0.9 arcsec FWHM.
The sections below provide further information about the causes and effects of these acquisition errors.
Dependence of throughput on seeing and target-acquisition error
In AF2, fibre diameters project to 1.6 arcsec on the sky, which is just above twice the median seeing at ORM. Such small fibres provide a S/N advantage at source flux levels fainter than the sky, as they minimise the amount of sky light that makes it into the fibre. However, whenever seeing is worse than median, aperture losses start to be significant. Additionally, aperture losses are easily aggravated by positional errors: the requirements on positional accuracy are more stringent with smaller fibres.
These effects are quantified for point sources in Figure 1, which shows, for each of several values of seeing, the flux fraction inside the fibre as a function of fibre offset from the target position. The plots assume a point source with PSF following a Moffat profile with β = 3.0, which is appropriate to the ORM sky. Figure 1 shows that, for a seeing of 0.9 arcsec FWHM, a perfectly-centered fibre captures only 69% of the stellar flux, while, with an off-centering of 0.5 arcsec, the in-fibre flux fraction drops to 54%, ~20% below that of the centred fibre. When the seeing is worse, the flux loss due to off-centering is smaller but, at least for offsets <~ 1 arcsec, less light gets into the fibre.
1. Astrometry errors
2. Differential refraction
3. Acquisition + guiding errors
4. Errors in the field distortion map
5. Mechanical robot precision
We discuss each in turn in the following subsections, and end with a summary of results from an on-sky test.
AF2 has 1.6-arcsec fibres over a 1-degree field, which makes it imperative that the astrometry is of the highest accuracy to avoid important positioning errors.
Astrometric catalogues such as UCAC3 and SDSS have proved to provide good results. However, when doing spectroscopy of faint targets such as distant galaxies, the coordinate system of the target catalog may present offsets from that of the fiducial stars. It is the observer's responsibility to ensure that the coordinate system of targets and fiducial stars is on the same reference.
Astrometry errors affect field acquisition in often unpredictable ways, and, especially when using few fiducial stars, astrometry errors in the fiducials can entirely bias the acquisition solution. The astrometry should account for proper motions of the fiducial and science target stars. Even those stars with small proper motions, e.g. 20 mas/year, will have coordinate differences of 0.3 arcsec after 13 years. In a 1.6 arcsec diameter fiber, a 0.3 arcsec offset yields a flux drop of ~5% when the seeing is 0.9 arcsec. This drop is small when considered alone, but users should recall that such error adds to other sources of errors listed above.
Due to atmospheric refraction the positions of objects in the field can slowly shift, as a function of zenith distance, with respect to the field centre. The effect is a differential plate scale stretch in the vertical direction.
This effect can be minimized by reconfiguring the AF2 fibres to new apparent positions, according to the hour angle of the field. Over the 1 degree AF2 field, this is not a serious problem for objects close to the zenith, but it should be considered when observing at elevations below 50 degrees. A field at an elevation above 60 degrees can be tracked for 3-4 hours without needing reconfiguration, if the fibres have been configured for the hour angle in the middle point of the range (e.g. if the field HA = -1h at the start and HA = 3h at the end, we should configure the fibres for HA = 1h; in the case of AF2 we provide the sidereal time instead of the hour angle).
For lower elevations, the reconfigurations should be more frequent depending on the zenithal distance.
Acquisition + guiding errors
The process of field acquisition is another important source of error. Traditionally the acquisition was carried out by the telescope operator, who estimated visually the best offsets to apply to the telescope, by checking the fiducial stars. This is a very subjective, non-repetitive and time-consuming task that can contribute significantly to the acquisition error.
From semester 2013B on, an Acquisition Tool is available, which automates the acquisition process by finding, and applying to the telescope control system (TCS), the optimum translation and rotation offsets to centre the guide stars on the fiducial fibres. It also provides the possibility of guiding with all the fiducial stars, or with the selected ones, which reduces the guiding errors throughout the AF2 field of view.
With the new procedure, the acquisition errors can be reduced to ~0.15 arcsec in translation, and are negligible in rotation, as measured on the fiducial stars.
As well as optimising the acquisition, the tool greatly reduces the acquisition overheads, from typically ~15 min when doing it manually to ~5 min with the tool.
As well as the field-acquisition process, which has been improved, the accuracy of the target-acquisition strongly depends on other considerations that should be taken into account by the observers when preparing the configuration files, namely:
Field distortion errors
The current AF2 focal-plane distortion map is described by the equation:
rtrue = k1 rfp (1 + q rfpr2)
k1 = 17.643 arcsec/mm
q = 320 rad-2
where rtrue is the distance of an object from the optical axis, as measured on the sky (in arcsec), and rfp and rfpr are its distance from the optical axis in millimetres, and radians, respectively.
k1 is the linear scale (in arcsec/mm) and q corresponds to the pincushion coefficient, which accounts for the radial distortions.
The focal plane geometry parameters above were determined empirically in 2013A using rastered observations, and are consistent with the Zemax model of the WHT PF optical corrector. The focal plane geometry is accurate to better than 0.25 arcsec at 20 arcmin radius field of view and to better than 0.05 arcsec over the central 10 arcmin radius.
Mechanical robot precision
The mechanical precision of the AF2 robot is still under investigation. Repeatability is excellent, with a 95% percentile error of 0.17 arcsec. Work on determining the absolute accuracy is ongoing.
We note that the good on-sky positioning results results shown in Figure 2 suggest that the contribution of mechanical errors is small.
Putting it all together
Acquisition accuracy tests were carried out in commissioning nights during 2013A. We selected targets from different stellar astrometric fields, using configuration with 7-10 stars as fiducials and 70-90 as targets. Focal plane geometry was assumed as described above, with a pincushion coefficient q=320. The automatic acquisition tool was used for acquisition and guiding. To characterise the acquisition accuracy, a spiral raster was performed around the initial acquisition, with steps of 0.4 arcsec. Atmospheric seeing and transmission was controlled and deemed to be stable. The acquisition error for each fibre were derived from measurements of the raw flux at each raster position.
Figure 2 shows the fibre offset distribution over the central 20 arcmin radius field of view for the above experiment. The rms of the error distribution is 0.25 arcsec, and the 90 percentile is 0.4 arcsec.
This result indicates that AF2 is capable of acquiring fields to a high degree of accuracy. Users should note that the conditions for acquisition were close to ideal in this experiment. Acquisition errors increase when decreasing the number of fiducial stars, when the distribution of fiducials does not uniformly cover the field of view, or due to astrometric errors.
|Top | Back|