[alma-config] Size of model sources

[mel wright 456]

Hello Min,

	Here are some thoughts on what the configuration group could
study.

The ALMA configurations should not be determined from models of single
field images. Most images will be mosaics. This has implications for
the choice of array configurations.

The fidelity of a single field image for a 2' source is irrelevant for
ALMA, and should not be used as a criteria for deciding between two
competing ideologies of array configurations unless one observes at
wavelengths longer than 1cm most of the time.

The ALMA array with 12 m antennas at a wavelength of 1.2 mm has a field
of view of only 20''; angular scales larger than this must be obtained
from mosaic observations.

Mosaics with thousands of pointings requiring on-the-fly mosaicing may
be common (Holdaway 1994, MMA memo 122).  Good snapshot uv-sampling
with similar synthesized beam patterns for each pointing is desirable
in array configurations which do not provide complete snapshot uv
sampling.

For mosaic images, Cornwell, Holdaway \& Uson (1993) show that an image
fidelity of 20, measured as the on-source signal to noise ratio
requires an RMS surface accuracy of lambda/40 and an antenna pointing
accuracy of about 6% of the primary beam FWHM.  This has become the
basis of the antenna specification for the ALMA telescope. Whether the
fidelity of a 2' single field image is 100 or 120 is not relevant.

One might ask whether the choice of array configuration can do anything
to improve the fidelity of mosaiced images.  Certainly, adding a
compact array of smaller antennas is important, and the placement of
these with respect to the 12m antennas should be an integral part of
the array configurations, taking into account the effects of pointing
errors and spatial frequencies which could be determined from cross
correlations between different sized antennas.

Pointing errors will be partially correlated on different antennas.
Design and mechanical features, as well as a part of the atmospheric
refraction will have common pointing errors. Random pointing errors
include wind and atmospheric refraction fluctuations which could be
very different across the array. Redundant uv-sampling could be used to
mitigate the effect of pointing errors.


		Best wishes,

					Melvyn.