Frequently Asked Questions
What bandwidth should I use for noise calculations?
If you are interested in the noise per channel, as you would be in most spectral-line projects, use the bandwidth of one channel to give you the noise per channel. If you are are doing a continuum detection experiment, in which you want to cover as much spectral range as possible, use the total (RFI free) bandwidth. When in doubt, use the total bandwidth you plan to use in making an image.
How does the number of polarization products affect the noise?
Choices are single polarization (RR or LL), dual polarization (RR and LL), and full polarization (RR, RL, LR, and LL). If the emission is unpolarized, dual polarization gives twice the amount of independent data as single polarization does, and therefore decreases the rms noise by a factor √2, as the sensitivity calculator will show. Full polarization does not improve this any further since RL and LR do not contribute to Stokes I.
Can I decrease the noise by overlapping the basebands?
No! For instance, in the case of the 8-bit samplers there are two independent baseband pairs. These are simply two windows in frequency space on the same data stream. Having the two windows look at the same data does not affect the measured rms noise.
When to Use Overlapping Basebands
So there is never any reason to use overlapping basebands?
Actually, there is. Often basebands are divided up in a number of subbands (e.g. 8), each of which has reduced sensitivity at the edges. So in the above example there are a number of frequency ranges with reduced sensitivity. By overlapping the basebands by one-half of one subband width, ranges with reduced sensitivity in one baseband will have uncompromised data in the other. For more details, see the Spectral Line section within the Guide to Observing with the VLA.
What overhead should I count on for slewing, calibration, etc?
How to tune Basebands
How should I tune the two baseband pairs of the 8-bit samplers? Similar considerations may apply to the 3-bit samplers that provide four baseband pairs.
For the 8-bit samplers, the WIDAR correlator processes two independently tunable baseband pairs. There are a number of common applications; which one to choose is determined by the science you want to achieve.
- Tune end-to-end. This is often done if one wants to obtain as wide a contiguous bandwidth as possible. For instance, one 1 GHz baseband pair is tuned to 4.5 GHz, and the second to 5.5 GHz, for a complete 4–6 GHz coverage.
- Tune as far apart as possible within one band. Often used for spectral index determinations; for instance one 1 GHz baseband pair is tuned to 4.5 GHz and the second to 7.5 GHz, to obtain 4–5 GHz and 7–8 GHz coverage.
- Tune to target different spectral lines; for instance, one 8 MHz subband from a given baseband pair is tuned to the NH3(1,1) line at 23.695 GHz and a second 8 MHz subband from the other baseband pair to the NH3(5,5) line at 24.533 GHz. The basebands are not contiguous, but each covers the line of interest with the desired frequency resolution.
- Tune the two baseband pairs such that they are shifted by a fraction of one subband (i.e., largely overlapping). Each baseband pair consists of a number of subbands, and there is sharply decreased sensitivity at subband boundaries. When tuning the second baseband pair one-half of a subband width away from the first baseband pair, compromised data at a subband boundary in one baseband pair can be replaced by good data in the second baseband pair.
Note that the two baseband pairs offer two 'windows' on the available spectrum. If the baseband pairs overlap, the data in the overlapping part of the spectrum will be essentially identical in either baseband pair. This is why tuning both baseband pairs to the same frequency does not increase the S/N by √2.
How do current observing setup restrictions impact my proposal?
With the WIDAR correlator, there are a number of. Most of these become important when preparing the observing script, but the following are of potential interest to proposers as well.
- Ka band (26.5-40 GHz) has the most restrictive tuning restrictions. Only half of the available basebands can be tuned below 32 GHz, and tuning around 32 GHz might be problematic. Further information on Ka band tuning can be found in the current Observational Status Summary section on VLA Frequency Bands and Tunability.
- Scheduling blocks (SBs) must use the appropriate setup scans for 8-bit and 3-bit resources. This will add to the overhead.
- If a target is observed at multiple frequencies there is increased calibration overhead since each target source scan at a certain frequency has to be bracketed by gain calibrator scans at that frequency.
- The SB Validation checklist (located in the Guide to Observing with the VLA) is important to follow before and after scheduling block creation.
GOST does not start
GOST does not start up when I click its link
This can happen if you don't have the correct webstart installed. In that case we recommend you start GOST using our alternative route.
I used fast switching with the old VLA; how do I do it now?
Fast switching in the old VLA was a way to reduce the slewing and setup overhead compared to traditional iterating between source and calibrator scans. In the new online system this overhead is sufficiently reduced that no special fast switching mode is necessary: its role is now taken over by a regular source - calibrator loop.