Users Guide for the VLA-PT Link
Mark Claussen
October 10, 2000
Background
The VLA-Pie Town connection
is a real-time link between the Pie Town VLBA antenna and the VLA, effectively
doubling the longest baseline of the VLA in its highest resolution configuration,
while maintaining the total sensitivity of the VLA. A history of this project
can be found on the www by pointing a web browser to VLA-PT
Link History.
In June 2000, the NRAO accepted
proposals for users to use the VLA-PT link in the upcoming VLA A
configuration, scheduled for October 2000 - January 2001. The current document
is intended to be a guide for those users who have been granted observing
time for the VLA-PT link.
Brief
Description of the System
At PT the sky signal travels
the standard VLBA local oscillator chain until it arrives at the 500 ---
1000 MHz IF. At this point the signal is taken to the VLA-PT link rack
(located at PT), where the four VLA IFs (edges at 1300, 1400, 1550, 1650
MHz) are constituted by appropriate mixing with further local oscillators.
These four analog signals, in a 1200 - 1800 MHz band, are sent back to
the VLA over the optical fiber. The hydrogen maser at PT serves as the
frequency standard for all the local oscillators at PT.
At the VLA, these signals
are processed just as if they were from a VLA antenna, and sent to the
VLA correlator. Local oscillators at the VLA used to process the signal
are locked to the hydrogen maser at the VLA. Digital monitor and control
signals from the VLA control computer are sent to PT via the optical fiber.
N.B. These digital signals only send commands to control the rack of equipment
specialized for VLA-PT operations, not for the standard VLBA equipment,
such as the first LO or antenna pointing. A VLBA control file is required
to control the standard VLBA equipment at PT.
When the signals get to the
VLA correlator, large delays appropriate to the PT delay (geometric as
well as path delay) are inserted into the delay lines. This causes a problem
in which some invalid data (i.e. not
valid data from the antennas)
is pushed over timing signals in the correlator which distinguish good
from bad data. The invalid data is not random, and thus causes systematic
problems which have a distinct effect on the data. In particular, this
effect causes an increase of more than a factor 10 over the expected thermal
noise. We have re-arranged the timing by delaying the "data-invalid" signal
as well as turning off the "self-test" program in the correlator. This
recovers the proper timing for the correlator as regards data invalid,
and the noise from blank sky tests appears to integrate down as one would
expect.
A "round-trip" phase is measured
from PT to the VLA. "Round-trip" is in quotation marks because it is actually
only a one-way measurement. This measurement is made at 1200 MHz. This
measurement is received by the VLA on-line system, and is applied, assuming
it is completely characterized by the difference in the rates of the two
masers.
Performance
of the VLA-PT Link
Over the past year, we have
carried out a number of tests of the VLA-PT link in order to verify that
all hardware and software is
performing optimally. We
have found and overcome a number of problems; there are still some lingering
questions. In particular,
we believe there are some
geometric or tropospheric corrections which have yet to be calculated or
applied correctly. At the time
of this writing, this can
cause phase offsets between two sources that are close together on the
sky. We are currently trying to
untangle these effects.
The phase difference between a pair of such sources can vary over time.
Thus, unless we discover the cause of this problem and correct it, we will
recommend trying to remove this effect by a particular calibration method
(see
Section
5).
As mentioned in Section
2, a "round-trip" phase is measured between the VLA and PT, and a phase
correction based on this measurement is applied in real time. When the
hydrogen masers at the VLA and PT are set so that their drift rates are
very close together, this correction works very well, and phase drifts,
especially at low sky frequencies are quite well-behaved and slow (<1
turn over several hours).
The VLA technique of "fast
switching" (see C.
Carilli's memo), which is used at high frequencies to quickly nod the
antennas between two sources which are close to each other on the sky,
has been tested, and appears to work with the VLA-PT link with a few caveats:
-
Before each fast-switching "scan",
a setup scan of short duration should be observed in order to ensure that
that both the VLA and PT are on-source for the source which is first in
the nodding scan.
-
The fast-switching durations
should be multiples of 10 seconds.
Reference pointing as implemented
on the VLA is not implemented at PT. This is not to say that one cannot
use reference pointing on the VLA --- it just won't have any effect on
the antenna at PT. That is to say, the reference pointing scan will just
send the PT antenna to the pointing source at whatever frequency is being
observed, and the VLA online system will not try to solve the pointing
offsets for PT.
Scheduling the VLA-PT Link
As mentioned above, the monitor
and control signals from the VLA do not control the antenna pointing
or the first local oscillator
at PT. Thus a VLBA control
file must be created to control such things at PT.
We have arranged a method
to do this which also allows the VLA on-line system to calculate the required
local oscillator settings in the VLA-PT link rack at PT, so that those
LOs are set to the correct frequencies. A field in the //LO card image
of the VLA observe file is set to the sum of the first LOs at PT. This
is done automatically from the latest version of Jobserve, the new
Java-based VLA scheduling tool (selection of "Use Pie Town Antenna" from
the "Starting Conditions" screen is required). Barry Clark has written
a new program which reads the VLA observe file, and creates a VLBA control
file (the so-called .crd file). The observer only has to worry about the
front end of this process. Using the latest version of Jobserve
to make his/her observing file, the hooks for the translation program are
already put in place. As an operational procedure, we require that observe
files be submitted to the data analysts (analysts@nrao.edu)
two weeks
in advance of the observing date. This will allow time for the analysts
to check, create and distribute to the telescopes the required observing
control files.
For fast-switching observations,
as mentioned above, you should include a short regular scan on the
source which is the "primary" in the fast-switching mode, to make sure
that both telescopes are on-source at the start of the fast-switching scan.
Calibration
and Observing Strategies
In general, calibration of the
VLA-PT link is not much different than that used to normally calibrate
the VLA by itself. Care should be taken to select calibration sources that
are not resolved by the two times better resolution provided by the PT
link. We haven't had the time to determine which are the best calibrators
with regard to their structure, and so the best advice we can give is to
select those calibrators from the VLA calibrator manual which have "P"
designations for the A configuration. As a rule of thumb, the closer
a calibrator is to your target, the better (in particular with the effects
we described above on the VLA-PT link). For lower frequencies (8.4 GHz
and below) the sensitivity of the VLA allows weaker calibrator to be usefully
used (0.3 to 0.5 Jy) as compared with the higher frequencies (where we
recommend brighter calibrators (> 0.5 Jy).
Due to the current situation
with phase offsets between two sources close together on the sky (see Section
3), we recommend observing two calibration sources that are close together
(one of which should be the calibrator you plan to use for phase calibration).
Observing these two calibrators for a total of about 5 minutes at least
three times during a several hour run should allow the calibration of the
phase difference between the two sources, and the change in this phase
difference over time. This method need only be used if one cannot use self-calibration
on the target source itself.
Travel to the VLA
We strongly urge at least one
member from each observing team come to Socorro and visit the VLA during
their observations. The purpose of this request is for us to obtain direct
feedback on how you view the operation of the VLA-PT link. We also would
like your feedback on the quality of your data, and how well the use of
the VLA-PT link has enhanced your scientific return.
Summary
In order to obtain the best
data possible from the VLA-PT link:
-
Use nearby "P" (for A configuration)
calibrators, nearer is better than very strong.
-
Use a pair of calibrators as
a calibrator/target pair two or three times within your observing run,
if self-calibration on your target is not feasible.
-
Return to your calibrator at
least once every 6 or 7 minutes for observing wavelengths longer than or
equal to 3.6 cm. For shorter wavelengths, the VLA "fast-switching" mode
is recommended.
-
For fast-switching observations,
insert a brief regular observing scan before each fast-switching scan,
in order to guarantee that both telescopes (VLA and PT) are on the "primary
source" for the fast-switching scan.
-
For scheduling purposes, we
require you use the Jobserve scheduling tool, and we need an observing
file two weeks in advance of your observing run.
-
Finally, we strongly urge a
member of your team to come to Socorro, visit the VLA during the observations,
and stay for a few days afterwards to at least take a first look at the
data from the VLA-PT link observations.
If you have further questions
please contact me at mclausse@nrao.edu
or
by telephone at (505) 835-4071. I will expect to be in touch with each
of the observing teams to closely work with you on your observing files.