Options other than the standard (A, B, C, D) configurations of the VLA are available. You will need to combine observations made in more than one VLA configuration if you require a range of baselines that exceeds the range provided by a standard configuration.
As the ratio of the longest to the shortest baseline in a
standard configuration of the VLA is about 40:1, each standard
configuration can be used to image reliably up to 
where 
is given by Equation 5
at the specified frequency. If the values of 
and
needed for your experiment do not both fall between and
40 calculated from Equation 5 for a
given standard configuration and frequency, consider using more than
one VLA configuration--unless you can extend the projected baseline
range by observing at extreme hour-angles (where the u-v baselines may
be significantly foreshortened). Without help from foreshortened
baselines, observations requiring 
40:1 will need to
be done in more than one configuration. Note however that some
observations with 
40:1 may also need more than one
configuration--for example, if the optimum falls mid-way
between two resolutions given by allowed values of n and 
in
Equation 5.
When mixing standard-configuration observations to extend the range of angular scales that can be imaged, compute the bandwidths and total integration times for the different configurations separately. You will not need to spend as long a total integration time in compact configurations as in the more scattered ones, especially if you can use wider bandwidths in the compact configurations. A reasonable approach is to aim for constant surface-brightness sensitivity in observations in different configurations that will be combined to make a single image. But if possible you should also distribute the observing in the more compact configurations over at least 4 hours in hour-angle, to minimize gaps in the u-v coverage. If you leave gaps in the u-v plane by doing all the compact-configuration observing in one ``scan'', your images may be limited by ``rumble'' from poor sampling on large angular scales, rather than by sensitivity. Spreading a small amount of observing time around in hour-angle to improve the u-v coverage may be impossible if you are observing only one source, however. It is much easier to optimize a project in which several sources with similar imaging requirements can be time-shared in the more compact configurations. Otherwise, it may be necessary to collect more data in the compact configurations than sensitivity alone requires, just to sample the inner u-v plane adequately. If you do this, you should expect to reweight the short-baseline data to optimize the final synthesized beamshape and signal-to-noise together--the optimum weighting can be found empirically once the data are in hand.