This is the default option.
Unlike a regular hexagon, a slightly perturbed hexagonal arrangement can provide uniform spacing in the direction perpendicular to a specific scan direction. See the attached plot
. Assuming a minimum feed spacing of 3.0 beamwidths, with this arrangement one would obtain a set of scans spaced by 1.0 beamwidth in a single sweep across the target field. By oserving two consecutive scans, shifted by 0.5 beam from one another, one would obtain 0.5 beamwidth sampling (i.e. Nyquist) of the field. If there is no box rotator, then one would presumably align the pattern with the azimuth/elevation directions, and perform scans in that coordinate system. With a box rotator, the pattern could be rotated to provide uniform sampling in a different coordinate frame, such as RA/Dec.
- 15 Nov 2007
Columns offset 1/2 BW
Carrying Todd's point a bit further, perhaps an arrangement in which the beams are arranged in columns offset by 1/2 BW in the direction orthogonal to the scan direction. See attached plot
. A sweep of this pattern would fully sample a strip about 20 BW wide, and since starting, stopping, or reversing the GBT is a significant overhead, this might be a an efficient way to map OTF.
- 27 Nov 2007
Feed layout and observing strategies
With a feed rotator there are a number of possible arrangements of the feeds which can give efficient coverage
of a particular swath of the sky. Without a feed rotator any given arrangement is optimum for scans in just a few
angles of azimuth and elevation (the scan directions are fixed with respect to the dish). I think that there are several
criteria whose evaluation will allow us to determine the desired feed configuration:
1) The resulting maps should be fully sampled;
2) The noise should be approximately constant across the desired area;
3) Minimal time should be spent outside the desired area (or in overhead).
I'm of the current opinion
that we need to evaluate feed configurations against these criteria by simulating the mapping of elongated
areas which are oriented at different angles on the sky. The simulations should be run at a range of
declination and hour angle, and should give us the information we need to understand how the array would be
used in practice.
- 28 Nov 2007