As of April 9 2008, Doppler tracking can be
used again but for VLA - VLA or EVLA - EVLA baselines only
Our original recommendation was not to use Doppler tracking only
when crossed (VLA - EVLA) baselines were being used. This was
because the Fluke synthesizers which control the last frequency
conversion for VLA antennas do not preserve phase when changing
frequency. This is not a problem for VLA-VLA baselines, since any
change in phase from the Flukes affect all VLA antennas identically.
However, the Fluke synthesizers are not used for the EVLA antennas,
so any phase change does not propagate to the EVLA antennas, and
thus there is a (random) phase jump on EVLA-VLA baselines when the
Flukes change frequencies. This problem will be with us for the
entire time of the transition period, when we are using both VLA and
EVLA antennas in the array.
The workaround is to 1) not ask for online Doppler tracking and 2)
treat a change in the sky frequency (i.e. manual Doppler tracking)
as any band change, i.e. make sure and bracket those changes with a
calibrator scan so that the phases can be tracked, and, when
calibrating those data, make sure that the time scale for
calibration solutions does not extend over these phase changes.
We have identified another problem, in that the Jobserve program
(and indeed the older Observe program) do not make the same Doppler
calculation as the online system. Depending on the sky frequency
and the sky position, it appears that these two calculations can
differ by at most a few tenths km/sec. Thus an observer should not
use the frequencies computed by Jobserve (or Observe) for the
absolute sky frequency unless such an error is not important (for
example, if the channel widths are significantly larger than this
error).
We have therefore implemented a web-based version of the
"gold-standard" Dopset program to calculate the sky frequency, given
a sky position, rest frequency, velocity, rest frame and velocity
type, and an LST day and time. This program has been carefully
compared to both the on-line system of the VLA and that of the EVLA
antennas. The velocities thus derived with all three of these
systems agree to within 0.01 km/s for all the cases we have tested.
This web-based version, implemented by Bryan Butler, can be found here.
We recommend that observers use this web tool to calculate absolute
sky frequencies for their spectral line observations, and to copy
these frequencies directly into the Frequency Calculations page in
Jobserve. There are two possible ways to observe a spectral line in
fixed frequency mode:
- replicate online Doppler tracking by calculating a new observing
frequency for the target source periodically throughout the
observing run. Bear in mind, however, that we have observed phase
jumps on VLA-EVLA baselines on frequency changes (caused by the VLA
system), so we strongly recommend including calibration scans at
every change in observing frequency. These scans should be at the
same observing frequency as the source scan they refer to.
- if the spectral line is narrow compared with the total bandwidth
the entire observation can be made at a single fixed frequency, and
the Doppler corrections applied in post-processing using the task
CVEL in AIPS. If the line is well-resolved spectrally, this will
not result in any degradation of spectral resolution; if the line
is very narrow and strong relative to the channel width, some
spectral smoothing may be needed to eliminate the Gibbs phenomenon.
For dynamically scheduled projects, we advise users to submit a
valid observe file for the start of the configuration during which
the project is scheduled. As soon as the frequency shift due to
the Earth's orbital motion affects the position of the line of
interest in the band significantly, we advise users to submit an
updated version, and to continue to do so until the project has
been executed.
For further assistance in planning these kind of spectral line
experiments, please contact one of our
support scientists.
|