Array Configurations

by Gustaaf Van Moorsel last modified May 20, 2020 by Emmanuel Momjian


The VLA is reconfigurable and uses all antennas in a single array in four principal array configurations, A through D; the antennas can also observe in multiple sub-arrays as discussed below. The A-configuration provides the longest baselines and thus the highest angular resolution for a given frequency, but yields very limited sensitivity to surface brightness. The D-configuration provides the shortest baselines, translating to a high surface brightness sensitivity at the cost of angular resolution. See the configuration schedule for details for each call for proposals and the highest angular resolution as function of frequency in the OSS. In general, as the baseline length expressed in wavelengths gets longer, the phase stability gets worse which impacts the observing strategy and observing overhead.

It is generally important to consider the following:

  • What angular resolution is required for the proposed science at the desired observing frequency?
  • For resolved sources, how does the desired angular resolution compare to the required surface brightness sensitivity?
  • For the array configuration that gives the desired angular resolution, how much of the flux density is actually in compact components, and will not be resolved out?

The Observational Status Summary section on Resolution, in conjunction with the Exposure Calculator, can help answer these questions.

Please note that low declination sources risk being subject to antenna shadowing at certain azimuths for the C and D configurations. These targets can still be observed. Observing at low declination implies, however, smaller windows of no-shadowing (one on either side of the north-south arm), which effectively makes the setup of the scheduling blocks harder or the observation less sensitive than expected.

Note on hybrid configurations: For very southern (and very northern) declinations, the VLA used to offer hybrid array configurations where the north arm was extended compared to the east and west arms. This provided a more circular synthesized beam at declinations south of −15 deg and north of 75 deg, at the cost of limited scheduling opportunities due to the short duration of the hybrids. Semester 2016A was the last semester to offer the hybrid configurations. See the following section for alternatives to the hybrids.

Alternatives to Hybrid Configurations

The approach required to substitute for the lack of hybrid VLA configurations depends on the science goal and observing mode of a proposal as described below. We assume projects that would have proposed to use the hybrid configurations are requesting them because they wish to observe sources at southern (<−15 deg) or northern (>75 deg) declinations.

Point Sources

Proposers should request the next largest principal configuration and ask for the same amount of observing time that would have been requested in a hybrid.

Extended Structure

Good surface brightness sensitivity is needed for projects aiming to image an extended structure. The surface brightness sensitivity of a hybrid configuration can be reproduced by:

- either doubling the on-source integration time of the next largest principal configuration, or;
- combining 1.0 × (larger configuration) + 0.4 × (smaller configuration).

The choice of which to use depends on the science goal, declination, and observing mode, as described below. An exception is the DnC hybrid, for which shadowing at very low declinations makes the use of the D configuration very inefficient. To substitute for the DnC hybrid, proposers with targets at δ<-25 deg should always request double the amount of DnC observing time in C configuration.

  • Imaging extended structure with on-source integration times of around a minute or more:
  1. CnB/BnA substitute: proposers should combine 1.0 × (larger configuration) + 0.4 × (smaller configuration).
    • Note: if the field contains variable sources, but the science goal is imaging of extended structure, proposers should either request double the time in the next largest principal configuration (if this can be accommodated in a single scheduling block), or be prepared to model and subtract variable sources from individual datasets prior to combining, as needed.
  2. DnC substitute: proposers with targets at δ=-25 deg or higher should combine C+0.4D, as for CnB and BnA; proposers with targets at δ<-25 deg should request double the amount of DnC time but in the C configuration.
  • Imaging extended structure with very short on source integration times (e.g., large mosaics): very short scans can result in large slewing overheads, so to optimize observing efficiency proposers should request double the on-source time that would have been requested in the associated hybrid, for the next largest principal configuration.

For more details on how to optimize the science on the VLA without the hybrid configurations, as well as technical details regarding the above noted recommendations, we refer to the EVLA memo 193. Also, proposers should direct any questions about which configuration they should use to the NRAO Helpdesk. For an example on how to combine data from several array configuration, we refer to the VLA data combination guide.

Solar Observations

The Sun is a very extended source and has a complex brightness distribution that varies in time due to solar rotation and intrinsic source variation. For this reason, only a single configuration should be used to observe the Sun rather than combining different configurations. Furthermore, time variable phenomena such as flares require the use of the instantaneous, or snapshot, uv coverage provided by a configuration. The most useful configurations for observations of the Sun at frequencies above 1 GHz are the C and the D configurations. The A and the B configurations provide coverage that is too dilute and that over-resolves the Sun.

Sub-array Observations

Instead of using all (27) antennas to observe one source using one resource at a time, the VLA can also be split in groups of antennas observing different sources or with different resources simultaneously. The drawback is of course that the fewer antennas in a subarray are less sensitive per unit time, and provide fewer baselines, fewer u,v data points, and thus a sparser sampling of spatial frequencies for image reconstruction. However, many science programs are quite suitable for using sub-arrays, for example bright point sources for which one desires exactly simultaneous observations at different frequency bands, possibly with similar angular resolution, or when surveying a large sample of sources using only half or one third of the observing time.

There are some restrictions in how to set up the correlator or divide the antennas over the sub-arrays as explained in the Observational Status Summary and requesting sub-arrays in the Proposal Submission Tool (PST) is not exactly straightforward. For the latter we provide some guidelines here:

In the PST, there are five items to pay attention to: a check-box on the cover sheet, the list of sources, the resources, the sessions and the fill-out box in the technical justification.

  • Cover sheet: under the "General" item in the proposal there is a subsection for "Observing Type(s)". Tick the relevant items, including the check-box in front of "VLA Subarrays"
  • Sources: for each of the sub-arrays you would want to define a separate source group, unless you observe exactly the same sources in each sub-array.
  • Resources: for each of the sub-arrays you would want to define or select (from the defaults) a separate resource, unless you want to use exactly the same resource in each sub-array. Note that if you use the same sources you typically want a different resource per sub-array and vice-versa.
  • Sessions: this may get a bit tricky as it is impossible to get the total time right with your intended use of sub-arrays. Currently the PST simply sums up the times requested per sub-array, which means it over-estimates the requested total time in the proposal by a factor of Nsub, where Nsub is the number of sub-arrays. For General Observing, Nsub can be up to 3. The easiest way to overcome this limitation in the PST is to divide your total requested time (Ttot) by the number of sub-arrays (Nsub) and use that number as follows: 
    1. Make a session with a source group and a resource,
    2. Fill in the time per session as Ttot/Nsub
    3. Save the session,
    4. Click "Add" to the right of  "Time/Session (hrs)" - you do not click on "New Session" at the top right,
    5. Define a new component to that session with a source and resource, and again a session time of Ttot/Nsub, save, click "Add", until you have defined Nsub components in that session, which now should add to the total requested time Ttot on the cover sheet. 
    Claiming 1/Nsub for the session time per sub-array is in effect equivalent to asking for the total time Ttot per sub-array but only for 1/Nsub part of the full array. It is because the latter is not implemented in the PST, the steps listed above are the work around. If, however, these steps are not followed and for each sub-array Ttot time is requested, then the total time request on the cover sheet will be reflected as Ttot*Nsub. This will be wrong, and will in turn raise questions/concerns during the review process.
  • Technical Justification: in the dedicated text box for sub-arrays, you will need to explain the use of sub-arrays and what has been done to the sessions to get to the total requested time. Also, add any other relevant technical details as needed.