An image made from untapered uniformly-weighted 
hour
tracks in a standard VLA configuration at declinations above
, where foreshortening of the array is relatively unimportant,
has a
synthesized beam whose half-power beamwidth is given approximately by
where 
is the observing frequency in GHz and n=1, 2, 3, or 4 for
the A, B, C, or D configuration, respectively.
The exact beamwidth depends on declination, and on details of the u-v
coverage. But this estimate will do for planning purposes.
It is crucial to choose a suitable combination of and
n when planning VLA observations. For example, suppose you want to
study a smooth two-dimensional emission region 30'' across whose
peak apparent brightness 
would be 1 mJy per CLEAN beam
area on an untapered 20 cm image in the B configuration
(resolution 
4''). It could be detected at the 
level in about 10 min of integration at 50 MHz bandwidth in the
B configuration (using the sensitivity data given in
Table 5 and natural weighting). This is an
easy experiment!
But now suppose you try to image the same region using the
A configuration, keeping everything else--observing frequency,
tapering, u-v weighting, bandwidth--the same. The region will have
a peak apparent brightness of only 
0.1 mJy per beam area (the
synthesized HPBW will now be 
). A
detection would therefore require about 14 hours of on-source
integration! This shows why it is extremely important
not to use a wider VLA configuration (i.e., smaller beam area 
)
than is strictly necessary, when studying extended emission.
The choice is even more delicate if you are picking the
observing frequency at which to image steep-spectrum extended radio
emission using a given VLA configuration. The combined effects
of a steep spectrum and changing angular resolution can make such
emission much harder to detect with a given VLA configuration at
the higher frequencies. For example, suppose that an extended
emission region has a peak intensity of 1 mJy per CLEAN beam
area in the VLA's A configuration at 20 cm--a
detection would be made in 10 minutes. If the region has a 
spectrum, the peak intensity in the A configuration at 6 cm
would be 0.029 mJy per CLEAN beam area and a
detection would need a week of integration!
For sources with compact flat-spectrum components and
extended steep-spectrum emission, the dynamic range needed to image
the extended structure increases rapidly with increasing frequency.
Suppose that the extended emission referred to in the previous example
surrounded a 5 mJy point source with a 
spectrum. The dynamic
range required for detection of the extended structure
would be 50:1 in the A configuration at 20 cm. This is
easy to obtain. The dynamic range required in the A
configuration at 6 cm would be 1900:1, a non-trivial goal
without self-calibration. At shorter wavelengths, the dynamic range
requirements would be still greater but the atmospheric coherence
times would likely be shorter and the self-calibration correspondingly
more difficult.
