April 15, 1994 Original
February 10, 1995 Revision 1
July 12, 1996 Revision 2
K. Sowinski
This document is intended to be an introduction to the issues
that must be considered when planning Q band observing. My intended
audience is the experienced VLA user, VLA staff and the VLA operators.
I will consider mostly practical issues; astronomical problems do not
fall within my domain. These issues fall into two categories:
problems associated with using the VLA at short wavelengths and
constraints between subarrays. Subarrays are important in this
context because only 13 antennas will be outfitted to work at Q band,
leaving the remaining 14 to be used for other, independent, purposes.
The only difficulty with short wavelengths that need be considered
here is pointing.
MULTIPLE SUBARRAYS
The 27 VLA antennas may be divided into up to five almost
independent subarrays. I will explain here how this is accomplished
and what is meant by "almost". Subarrays are created by the operator
by editting entries in a file. Every time that the assignment of
antennas to subarrays changes the operator must modify this file.
Because this is tedious and error prone, we discourage frequent
shuffling of antennas between subarrays. An Observe file must be
produced for each subarray; a note in the header of the file listing
the antennas that are to be controlled by this file is a sufficient clue
for the operator to properly configure the system.
While the antennas are all independent, there are global
resources that must be shared in a compatible way by all subarrays.
These include the correlator, the Fluke synthesizers that supply the
LO for the final conversion of the IF to baseband, the integration
time, and a few bits of software. All these topics will be addressed
below.
REFERENCED POINTING
The open loop pointing accuracy of the VLA is very poor when
compared to the beam size at Q band, or even at K band. To make
the VLA usable at these frequencies, we have invented a "referenced
pointing mode" of observing. The simple explanation is that a pointing
scan is done to determine and record the pointing offsets for each
antenna, then these offsets may be used to correct the open loop
pointing model for future scans. Unless observing near the zenith
(say, 75 or 80 degrees) updating the pointing corrections once an
hour is probably adequate. Putting the pointing calibrator and the
observed source on opposite sides of the meridian is clearly not a
good thing to do.
How referenced pointing observations are made will be
described in some detail in the belief that an understanding of a
process will help one to use it effectively. There is a program (PTG)
that is run whenever a subarray is in pointing mode, whose purpose is
to estimate the current pointing offsets for each antenna. It records
the offsets, along with the current azimuth and elevation in three
places; in a file which may be processed off line to determine updates
to the pointing model, on paper for inspection in real time, and in
memory so that the offsets may be used to correct the pointing model
for current conditions. The offsets will be considered valid and
recorded only if a number of conditions are met. There must be enough
flux, as reported for each antenna by ANTSOL, to be able to determine
a statistically significant solution; the offset and beamwidth that
are estimated must be reasonable; and finally, there must be valid
solutions for both polarizations because the offsets stored are the
average of the offsets determined for each polarization. The last
condition is important if the pointing scan is done in spectral line;
a correlator mode must be chosen that will provide at least one IF of
each polarization. The pointing scan must have a dwell time long
enough to complete at least one five-point pointing cycle (no less
than two minutes when using ten second integration time) and must
specify an integration time of ten seconds or a multiple of ten seconds.
If the source used for a pointing scan is not a calibrator (the "calcode"
is blank) we will force it to be a calibrator (it will be given calcode
'P') so that ANTSOL will run, and hope for the best.
The pointing offsets that have been determined and so laboriously
described, will be applied if, you ask for it, and if they exist, and if
they are not more than twelve hours old. These tests are applied
independently for each antenna in the subarray. A bit is recorded on
the Archive tape which describes, for each antenna, whether or not the
pointing correction was applied. As an observer, you ask for previously
determined pointing offsets to be applied by putting a 'T' in cc69 of the
source card. These things are understood beginning with Observe version
3.2.0. If it is not clear to you how to get this version of Observe,
speak to Wes Young (wyoung@nrao.edu).
In order to account for the small but inconstant collimation
differences between bands, a second order referenced pointing technique
has been developed. Using this technique is complex and, probably, not
always warranted. A description is given in a document called "Second
Order Referenced Pointing", May 1, 1996.
CORRELATOR USAGE
The strongest constraint on how subarrays may interact is imposed
by the correlator. The first and simplest rule is that all subarrays
must be in continuum, or all subarrays must be in spectral line modes.
If an attempt is made to use one subarray in continuum and the other
in spectral line, no data at all will be produced and the system will
get quite annoyed. A special dispensation is granted to single dish
subarrays for VLBI; such a subarray may be in continuum even if the
others are in spectral line. The operator is able to deal with this
without your having to get involved.
The constraints when all subarrays are in spectral line are
more difficult to describe. The rule here is that each IF must be set
to the same bandwidth in each subarray in which it is used. Further,
the correlator requires that, within each subarray, all IFs actually
used be set to 50MHz bandwidth or that all be less than 50MHz
bandwidth. For example:
Sub1 BW Sub2 BW Verdict
2AC 12.5 12.5 2BD 3.125 3.125 ok
1A 1.5625 2AC 6.25 6.25 NO
1A 1.5625 2AC 1.5625 1.5625 ok
1A 50 1B 12.5 ok
2AC 50 12.5 2BD 6.25 6.25 NO
PA 3.125 3.125 PB 25 25 ok
1D 6.25 1D 6.25 ok
The second example is illegal because IF A is used at 1.5MHz in subarray
1 and 6MHz in subarray 2. The fifth example is illegal because the two
IFs used by subarray 1 are not both at 50MHz or both less than 50MHz.
With these examples as a guide and the rules given above you should be
able to decide what combinations of correlator mode and bandwidth are
allowed. If all subarrays are in continuum, there are no restrictions
on bandwidth.
FLUKES
The Fluke synthesizers are located in the control building,
and supply the LO used to mix each IF to baseband. To allow
flexibility in the use of subarrays, there are two sets of four
Flukes, one for each of the four IFs. The hardware allows each
antenna to choose which Fluke set is to be used, but software limits
the choice to each subarray. All four IFs of a given antenna are
required to use the same Fluke set. In general, when doing spectral
line work with two subarrays, you will need to request that each
subarray use its own Fluke set. Problems will arise if there is a
third subarray doing single dish VLBI. Depending on the requirements
of a given observing program, it may be possible to divide control of
the Flukes among three or more subarrays. The operator controls,
perhaps in consultation with the observer, the allocation of Flukes to
subarrays. To avoid confusion, one should explicitly call out in
Observe file comments the need for independent Fluke settings.
INTEGRATION TIME
All subarrays must be using the same integration time. By
default, only subarray one may control the integration time for the
array. All other subarrays are forced to use the integration time
requested by subarray one. Dividing the array into subarrays reduces
the number of baselines. In the case of 512 channel spectral line,
using one subarray of 10 antennas and another of 17 reduces the amount
of data sufficiently to allow an integration time of ten seconds.
TIPPING SCANS
I am aware of no interaction between subarrays when any one
subarray is doing a tipping scan. However, if the subarray not doing
a tipping scan is in spectral line mode, then care must be taken to
be sure that the correlator mode and bandwidth for the tipping
subarray is consistent. If tipping scans are done in two subarrays
simultaneously, their outputs will be intermixed on the printer,
however separate disk files will be written for each subarray. The
mechanics of doing a tipping curve have been made more robust. Wrap
cards and "dummy" scans are not necessary, and selecting the azimuth
to be used now works reliably.
LOSER
There is a version of LOSER in /arana/u/ksowinsk that understands
Q band. I do not know if it has been propagated anyplace else. Any
difficulties with this LOSER should be reported to me.
THE OPERATORS'S VIEW
The cooperation and understanding of the VLA operator is key
to carrying out these heretofore very un-standard kinds of observations.
This paragraph can serve as a checklist for what the operator ought
to know. Reading this will help the observer to know what the operator
is up against when carrying out complex observing schedules.
The operator oversees the grouping of the array into subarrays.
Somehow he must learn which antennas will be in which subarray. The
observer may specify this by attribute (all Q band antennas), by
location (the outer nine antennas), or by enumeration (antennas 2,6,12,...).
Further the operator may be aware of things that the observer is not,
such as the need for a VLBI subarray. Whenever subarray affiliations
change, the correlator must be reset and all subarrays will suffer a
source change.
Through fields in the first card of the ARRAY file the
operator is able to tell the system which subarray is allowed to
control various global array resources.
cc1 contains the number of the subarray that has priority if
there is a conflict in the use of the correlator.
cc2 contains the number of the subarray that is allowed to set
the integration time for the array.
cc3-10 contain, for each of the eight Flukes, the number of the
subarray that is allowed to set it.
"1111112222" is the normal contents of these fields.
By manipulating the contents of these fields, the operator is able to
configure the array appropriately for less standard observing
situations. Normally it will only be necessary to alter the fields
that describe the control of the Flukes. For a subarray to actually
use a particular Fluke, or set of Flukes, it must declare which Fluke
set it intends to use. If this subarry has been granted permission to
use these Flukes, then it will set the frequencies it requires,
otherwise it will be forced to use the frequencies set by the subarray
that does control these Flukes. A subarry declares which Fluke set it
wishes to use on the //DS card. It is customary for the operator to
decide upon the allocation of Flukes to subarrays and to declare his
decsions using the //DS cards in the SUB? files. If it is important
that each of your subarrays have an independent Fluke setting, be
sure to communicate this fact to the operator in the header of your
Observe file.
VLBI
The canonical VLBI antenna is one of the three stationary
antennas. In general, this may also be a Q band antenna; we will
have to be sure we know how to deal with non-standard VLBI antennas.
Care will be needed to deal with control of the Flukes when there are
three subarrays.
FUTURE
In the future we plan some changes to relax some of the
constraints described here. This paragraph is a preview of coming
attractions.
If we can show that most of the pointing error that we correct
for by doing referenced pointing can be explained as a tilt, it might
be advantageous to express it this way. This could make the corrections
more accurate near the zenith when the azimuth is moving quickly.
In the more distant future we are planning an upgrade to the
correlator control system that will allow observing in line and
continuum simultaneously. This may take either of two forms. There
may be two subarrays, one in line and the other in continuum; or,
using one subarray, each antenna may be split so that, for example, AC
is in continuum and BD is in spectral line.
HISTORY OF THIS DOCUMENT
Original writing: April 15, 1994
Revision 1: February 10, 1995
Corrected and clarified the paragraph describing bandwidth
restrictions in spectral line modes. Explained that tipping
curve results are now written to disk and that the process is
improved. Various typos were repaired.
Revision 2: July 12, 1996
Simultaneous pointing is now allowed in multiple subarrays.
Second order referenced pointing is described. The list of
things to do has been shortened.
|
National Radio
