VLBI at the VLA

by Gustaaf Van Moorsel last modified Aug 22, 2016 by Amy Mioduszewski


These pages describe using the VLA as a VLBI station.  The collecting area, receiver suite, and geographical location of the NRAO's Karl G. Jansky Very Large Array (VLA) make it a valuable addition to a VLBI array. The VLA supports standard VLBI observations at frequencies of 1.7, 3.0, 5.0, 8.4, 15, 22, 33, and 43 GHz. Note that currently P-band cannot be phased. The VLA can only be used in phased array mode; observing with a single VLA antenna, or "Y1", is only available through the VLBA Resident Shared Risk Observing program. In phased array mode it offers the equivalent sensitivity, including sampling losses, of a single 115-m antenna. The VLA writes at up to 2 Gbps to a Mark5C recorder. The time and frequency standard is a hydrogen maser. The VLA participates in High Sensitivity Array (HSA) and Global programs. Its participation is arranged by the VLA/VLBA scheduler who can be contacted through schedsoc@nrao.edu. The phased VLA must be used with the VLBA (other antennas may be included). Unless for specific reasons, the data should preferably be correlated at the DiFX correlator located at the Science Operations Center in Socorro, NM.

A well phased VLA, with all 27 antennas, when added to the 10 antennas of the VLBA, will improve the sensitivity in a naturally-weighted image by a factor of about 2.4. Also, baselines between the phased array and any VLBA antenna should be about 4.6 times more sensitive than baselines between any two VLBA antennas.

Questions and concerns should be directed to the NRAO Helpdesk.

Phasing the VLA

TelCal, a real-time program run at the VLA during the observations, derives the delay & phase corrections for each antenna/polarization/subband. The antenna signals are corrected in the correlator, summed up, re-quantized to 2-bits, and finally recorded in VDIF format on the Mark5C recorder at the VLA site.

TelCal does not determine the correction until the end of a scan. In practice, there must be at least 3 good subscans, on a sufficiently strong source, to determine the corrections. The user should allow a scan of about 1 minute for phasing (software run by the NRAO analysts will automatically generate subscans), and the corrections will be determined and stored after the 3rd or 4th subscan. Subsequent scans can apply the stored corrections, e.g. on a target which is too weak to determine the correction.

Autophasing should be done on a calibrator which is a point source to the VLA's synthesized beam. The strength required depends on the frequency, weather and elevation. A good rule of thumb is >100 mJy for 1-12 GHz and >350 mJy for 12-45 GHz. Higher flux densities are required for low elevations particularly at high frequencies.  VLA phase calibrators that rated as "secondary" or "primary" for the frequency/configuration being used are a good choice for an autophase source.

Autophase corrections are valid for a duration that depends on the VLA array configuration, observing band, weather, elevation, and e.g., activity level of the sun. The weather and elevation are particularly important for higher frequencies, and solar activity at lower frequencies. Unfortunately the weather and solar activity cannot really be predicted for these fixed date observations. Our advice is to be conservative because if an observation does not contain frequent enough autophasing, this cannot be fixed in post-processing for the VLBI data, and sensitivity will be lost. When anticipated at the proposal stage, proper planning can mitigate the effects of weather. Consider observing at night and/or in the winter when the atmosphere tends to be calmer and (at night) solar activity is not an issue and avoiding observing at sunrise and sunset. Very broad rules of thumb for frequency of determining and applying new autophase corrections are:

  • C & D config: 20-30 minutes at low frequencies; 10-20 minutes at high frequencies
  • A & B config: 5-10 minutes at low frequencies; 2-5 minutes at high frequencies. May want to avoid observing at 45 GHz in these configurations, also because of the very small synthesized beams.

Restrictions on phasing

  • Phasing uses all but the edge channels, i.e., a continuum source is assumed.
  • Subarrays are not allowed.
  • No transfer of phasing between subbands.
  • No transfer of phasing across different subband setups. That is, there must be no change in subband setup between determining the phase corrections and applying the phase corrections. Changes in setups include: change observing band, tuning, bandwidth, polarization etc. For example:
    1. You cannot have a set of scans like this:
      • scans 1-6: C-band determine autophase
      • scans 7-12: X-band determine autophase
      • scan 13: C-band apply autophase
      • scan 14: X-band apply autophase
      • scan 15: C-band apply autophase
      • scan 16: X-band apply autophase
      • etc...
    2. Instead you should have a set of scans like this:
      • scans 1-6: C-band determine autophase
      • scan 7: C-band apply autophase
      • scan 8: C-band apply autophase
      • scans 9-14: X-band determine autophase
      • scan 15: X-band apply autophase
      • scan 16: X-band apply autophase
      • etc...
  • No transfer of phasing across a reference pointing scan, i.e., bracket your target on each side of the pointing scan with its own calibration scans.

Basebands and Subbands for Phased VLA

For the VLA, "baseband" refers to the frequency band that comes out of the samplers at the antenna electronics racks (its meaning is different from the traditional VLBA baseband). Only 8-bit samplers are used, i.e. there are two 1 GHz basebands, however the entire 1 GHz will not be available for phasing. See the VLA Observational Status Summary for a description of available frequencies and tuning restrictions, but note that they are less restrictive than VLBA tuning limitations.

For the VLA, "subbands" are the continuous blocks of frequency which are correlated by WIDAR in the control building and written to the Mark5C unit for VLBI. Two subband pairs (RCP and LCP) may be phased up. Each pair is a different baseband/IF pair AC or BD. So each pair is independently tunable in frequency.  Setups also matching the VLBA PFB and DDC (4- and 8-channel) observing systems are available on the VLA.  See the section on the RDBE in the VLBA OSS for more details on the DDC-8 and PFB modes and their restriction. The subband(s):

  1. The VLA is always dual polarization, even in shared risk modes. A & C (i.e. RCP and LCP) must be the same frequency and B & D (again, RCP and LCP) must be the same frequency.
  2. Must have the same bandwidth.
  3. Bandwidths of 16, 32, 64 and 128 MHz are allowed on a non-shared risk basis. Bandwidths of 1, 2, 4 and 8 MHz have not been tested and are only allowed as shared risk.
  4. Must align, in frequency and width, with the VLBA IF pairs.
  5. The restrictions are fewer for the VLA than for the VLBA or other HSA stations, so please follow the HSA guidelines.
  6. The VLA must be set up to match the VLBA.

Given the above restrictions, the maximum bandwidth is 256 MHz in 2 polarizations, which matches the maximum bandwidth on the VLBA. Given 2 bit sampling as on the VLBA, this gives a maximum data rate of 2 Gbps. Observing with the VLA that does not exactly mimic the VLBA in frequency setup is only available under the VLBA Resident Shared Risk Program. Examples of such RSRO projects would be single polarization observing and observing with the full VLA bandwidth but only recording the smaller bandwidth to be compatible with the VLBA.


All phased array observations will be fixed date. Please see VLBI @ the VLA: Scheduling Hints.

Phased array observations will be scheduled in the SCHED program, which is available via anonymous ftp, as described in the SCHED User Manual. Please see VLBI @ the VLA: Scheduling Hints. A keyin file is used to describe the observation and SCHED processes this keyin file and produces files to run the participating telescopes and correlator. One of the files created by SCHED is the VEX (VLBI Experiment) file which describes the entire observation. There is a program, vex2opt, which converts the VEX file into files that can be read in by the VLA Observation Preparation Tool (OPT). Vex2opt will be run by NRAO staff once the schedule is submitted. The OPT will then write the observing script for the VLA.

Standard practice is for the user to email the SCHED keyin file to vlbiobs@nrao.edu, NRAO staff will then run SCHED, and distribute any control files to the telescopes participating in the observation. They will run vex2opt and submit the VLA script. However, the user may edit the VLA schedule after it is loaded into the OPT. If the observer adjusts the VLA schedule and the observation fails because of that, then the fault lies solely on the observer and there is no requirement of the NRAO to offer a remedy. Note that SCHED can schedule VLA specific items like pointing, flux calibration, etc., so there is little reason to modify the VLA schedule in the OPT.

The observation will also produce standard VLA visibility data, so the user will probably want to do standard VLA flux calibration, and other calibration required to use the VLA data by itself.

If you have problems scheduling or anything else please use the NRAO helpdesk.

Log files

After the observation is over the observer will receive by e-mail a log from the VLA operator and the VLBA operator.


Please see the VLA and VLBA Observational Status Summary (OSS) for specifics on frequency ranges and tuning limitations. Generally the VLBA is more restricted than the VLA, so it would be best to start with the VLBA. The VLA and VLBA have similar frequency bands, but the VLA receivers generally have a wider tuning range. The common frequency bands are: L (1.35 - 1.75 GHz), S (2.15 - 2.35), C (3.9 - 7.9), X (8.0 - 8.8), Ku (12.0 - 15.4), K (21.7 - 24.1) and Q (41.0 - 45.0). Currently P-band (0.23-0.47 GHz) cannot be phased.

VLA modes

Only (27 antenna) phased array mode is allowed, single dish is only offered as part of the VLBA Resident Shared Risk Observing program. For situations where the observer may only want the inner antennas, this can be handled as a comment to the operator. All antennas in the subarray will be used for phasing, and all will be included in the phased sum. For instance, you cannot obtain WIDAR correlations for all antennas but use only a subset of those antennas in phasing or in forming the phased sum for VLBI recording.

There are two basic modes when scheduling the phased array: determine autophasing and apply autophasing. These modes can be used separately or together.

Data from the VLA

The VLA will produce two sets of data: 1) VDIF format data written to Mark5C recorder intended to be correlated with VLBA; and 2) standard WIDAR (VLA) correlator output.

The standard WIDAR (VLA) correlator output will be available from the NRAO data archive and can be accessed through the observer's my.nrao.edu account. This data will be 64 channels per subband per polarization product and have a 1 second integration time (regardless of configuration).


When preparing VLA schedule files, the following facts and guidelines should be noted:

  1. Please see VLBI @ the VLA: Scheduling Hints.
  2. The observer should follow the VLA general observing restrictions and advice, such as the one minute setup scan for each correlator configuration (i.e. band) at the beginning of the observation and the amount of overhead needed for reference pointing at the start of an observation.
  3. As a rule of thumb, the source on which you autophase should be a point source (to the VLA's synthesized beam) with >100 mJy for 1-12 GHz and >350 mJy for 12-45 GHz. Note that stronger sources may be required in bad weather and/or low elevation, particularly for higher frequencies.
  4. Assume about 1 minute to determine autophase.
  5. The frequency of determining/applying a new autophase depends on the VLA array configuration, elevation, day or night observations, the observing band and the weather. Please see the Phasing the VLA section for more details.
  6. Subarrays are not allowed.
  7. No pulse calibration system is available at the VLA. If you plan to use more than one subband, then you should observe a strong and compact source to serve as a manual pulse calibrator; see the VLBA Observational Status Summary.
  8. The minimum VLA elevation is 8o and zenith angles less than about 10o, source tracking can be difficult.  Over-the-top is not allowed by the VLA Observation Preparation Tool which is used to schedule the VLA.
  9. Very accurate positions (<0.1'') are required for phased-array observations.
  10. Those using the VLA at frequencies higher than 15 GHz should be aware that antenna pointing can be poor at these wavelengths. Therefore at these frequencies reference pointing for the VLA should be used.
  11. If you want to derive source flux densities and/or produce images from the standard VLA data, then your VLA observe file should include at least one scan of a primary flux density calibrator for the VLA.
  12. The VLA slews slower than the VLBA.
  13. If you want to do polarimetry with the standard VLA's data please consult the polarimetry section of the VLA Observing Guide.

Questions and concerns should be directed to the NRAO Helpdesk.