Techniques & tactics of pointing & focussing

To state the very obvious: the telescope must be accurately pointed and focussed before any astronomical data are acquired. In the context of a detector with a single-beam (single-pixel, single-detector element) the goal is then to get that beam pointed in the right direction and to optimize the Z-focus and the X- and Y- table positions of the SMU in order to maximize the telescope gain or efficiency. Pointing with detector arrays may seem less necessary for short observations of bright sources, but remains important for accurate registration of many observations of faint sources.

A pointing observation with a detector array is similar in many ways to regular 'imaging' observations, with the resulting digital image comprising, ideally, a single, small bright source roughly centred on a uniform dark background. The centroid of the bright region of the image yields the location of the source and the pointing of the telescope is usually adjusted so that it coincides (next time) with the centre of the detector array. Z-focus adjustment keeps the image size as small and as concentrated as possible, and X- and Y- adjustments minimize coma, but need proportionate pointing adjustments in elevation & azimuth, respectively, to maintain the source in the centre of the array.

Pointing & focussing - certainly the first lot of the night - is therefore an iterative process.

A pointing observation with a single-beam instrument requires (usually) 5 separate measures (of brightness, or 'flux') with the pointing of the telescope at each measure adjusted up, down, left, and right from the original best-guess position. These 5 flux measures can be analyzed for the centroid as above and the pointing adjusted. The offset distance is optimally the FWHM of the beam, and this process may have to be repeated if the original pointing error is large.

Z-focus and X- and Y- table adjustments are similarly multi-(5-,7-)part observations (with different table settings, this time); again allowing centroid determinations of optimal flux to correct the focus/table-settings.

Since fluxes are to be compared from one measure to another over the space of a few minutes, the sky conditions must be such as to allow this to be meaningful: i.e. photometric. This is true whether the receiver is a dedicated photometric device or a heterodyne receiver.

The observations are most meaningfully and rapidly executed using a bright astronomical point-source such as a small planet (Uranus, Mars - sometimes) or a blazar, although, when desperate, other more morphologically complex targets may be useful. The observations rely on measuring brightness above sky, so need to be done using beam-switching. Once upon a time the bmsw parameters would need providing to the operator, but these days the OMP/OT provides useful defaults for chop throws (of 60arcsec), chop frequency (7.8125 Hz - or whatever new value becomes standard with ACSIS) and Cycle reversal (`true'). Talk to your support scientist for more information.

Ideally, all observing periods should begin by pointing and focusing the telescope with the frontend you propose to use, although for the high-er-frequency receivers (with poorer sensitivity) the number of useful pointing targets is often restrictive. In such cases it is necessary to establish - at some other time of the night, when a bright source is available - the relative pointing between that receiver and one with greater capacity in this regard. The CBE is the continuum back-end used with the single-beam heterodyne instruments.

Integration times are dictated by source brightness, but should not exceed a couple of minutes in total duration if the component measures are to be meaningfully compared within the centroid algorithm. Useful sources are those with flux densities at 850um greater than about 2Jy (see the pointing catalogue), although the brightest thermal sources and planets may be needed for the likes of RxW.

Pointing accuracy depends upon getting good S/N on each component measure, with 1" accuracy requiring S/N>10. Such S/N's should also yield focus accuracies of about 0.1mm in Z, 0.3mm in X and Y; thus step sizes of 0.3mm in Z and 1.0mm in X & Y are used by default in the focus routines.

Repeat the pointing on a source as near as possible to your programme source for best science results. This may necessitate using a fainter pointing target than as described above, with relatively longer integration times : but it's worth spending this time getting the pointing right before taking real data.