Minutes - CA-forum #12
Friday 15 september, 1995.
1. Tied-array.
The september trials were entirely satisfactroy, and the present VBL
run (15-20 sept) is run with antenna 6 recorded in single antenna
mode, antennas 1-5 in tied-array mode.
The circular polarisation reconstruction, tested against some OH maser
observations, is working well.
The hardware is completed on one frequency channel - the second is
under construction, and will be completed by the next observing run.
There was a S/W problem limiting the bandwidth to 16 MHz, but this
has now been cured.
The control software is largely complete: SCHED has been modified
(M.Wieringa) to accept "Tied-array" as key word (see appendix A);
CAOBS recognises this key, and advises the correlator.
At present the 90 degree phase shift (needed for the circular poln
formation) is done in CACAL.
>>> This should be shifted to CAOBS.
There is no return path at present which would provide the user
with status information.
>>> MJK and WEW will organise this as far
as OBSCOM; narrabri will take it from there.
Some fast footwork may be needed by the correlator group to provide
full 2-frequency, 6-antenna operation.
Some additional work will be required in SCHED and CAOBS to provide
the two-frequency, MX-mode of operation, whereby two separate targets
are provided for VLB phase referencing.
2. Data quality observations.
R.Sault and R.Wark are now working on this.
A summary of the most recent run is attached (Appendix B).
- In its basic mode the analysis is automatic (following a script file),
but produces copious amounts of data. This has ben refined to just a
few pages of tables, along with a variable amount of images.
- Some initial findings :
The blank field is not empty -- it contains two sources of
about 15 mJy. A fresh field should be chosen.
Several ~40 degrees jumps have been observed between the A and B
polarisations. The immediate result was that ATLOD rejected about
25% of the data.
The M.Wieringa auto-flagger rejected another 10-15% of the data.
>>> This will need to be pursued.
The Offset field has the source 1 arcmin from the main axis. The
4 stokes images are reasonable at L and C band; at S-band (2.378 GHz)
and at X-band there is distinct off-axis polarisation (~1%); this
is probably reasonable at X-band where the source is a fair fraction
of a beamwidth offset; at S-band where the offset is a small fraction
we are possibly seeing further evidence of the polarised beam pattern
problem (see agenda item 4, below).
>>> It was agreed to use a larger offset at 2.378 GHz (say 6 arcmin).
3. Reference pointing.
R.Sault (see appendix C) argued that the reference pointing should be
done at X-band, rather than at C-band as at present.
No experiment yet has found a squint between the frequencies, so
this is probably safe. It is likely that there is little need for
reference point at L and S-bands; the cost of rotating the turret,
in time and phase stability may be higher than the possible gain from
a pointing tweak.
Reference pointing is now available for use by friendly experts -
indeed, their use is solicited.
A reference to the scheme is available on the web -
http://wwwnar.atnf.csiro.au/www/operations/pointing.html;
The narrabri home page is a useful place to start searches for this type of
information.
http://wwwnar.atnf.csiro.au - home page is a useful place to start)
A user-robust version for general consumption will not be available
for a while.
4. S-band polarised beam patterns.
Two notes (AT/39.3/052 and AT/39.3/053) describing the problem are available.
Recent observations by C.Granat and G.James have shown that the problem
originates with the fin profiles in the quad-ridge OMT.
In principle the C/X and L/S systems are scaled versions of the same design;
however, the fin profiles are different.
The cure is simple in principle - replace the fins. The cost is not -
time and $.
A detailed case for the replacement will be prepared.
5. 128 MHz birdies.
B.Beresford has carried out a number of tests in attempt to moderate the
interference from harmonics of the 128 MHz sampler clock. These
have all been negative.
(The problem is that the sampler clock's phase is adjusted to follow
the fringe rate - as such it is an exact scaled copy of a signal's
fringe phase; a harmonic of the sampler clock which falls within
the observer's band will therefore mimic exactly a genuine astronomical
frequency from the tracking centre. It will correlate perfectly).
- There is no simple hardware solution.
Some options:
a. hardware (expensive variant) : place the samplers in an RFI-tight
container. This would require substantial reworking of the turret
rack.
b. software - a phase switching scheme may have been found which should remove
the birdie; there would be a small loss of signal/noise at large
bandwidths; in line observations the cost would be insignificant.
c. palliatives: miriad ATLOD has a "de-birdie" option to blank the
offending channel.. This can only work well at 128 MHz (triangular
lag spectrum) and 4 MHz (rectangular lag spectrum); at other bandwidths
the birdie's influence is messy and does not lend itself to a simple
deletion algorithm.
6. Pulsar ephemerides logging.
In pulsar mode the correlator is provided with a file describing the
pulsar ephemerides. These files should be written to the RPFITS
data file - WEW will attend to this.
A note describing the capabilities of the pulsar observing mode is
available :
7. Narrabri planning meeting exploder.
The minutes of the regular narrabri planning meetings will be e-mailed
to anyone who wishes to receive them. In initial list of recipients
has been drawn up, and a test message has been sent to them.
If you received the test message you can choose to remain on the list
or to be deleted.
If you did not receive the test message you should contact G.Baines
-------------------------------------------------------------
Appendix A
SCHED modifications. M. Wieringa
Sched changes per Sept 6 1995
------------------------------
1. Screen format changed (now 3 columns),
options page (spectral line) no longer exists
2. New keyword: CALCODE - this is used to specify whether a source is to be
regarded as a calibrator. Specifying 'C' will cause assistance to do
additional checks on the visibilities. (a 'B' here is used for the baseline
solution)
3. New keyword: MODE
The MODE specifies special observing modes. At present the valid choices
are: 'STANDARD', i.e., the usual interferometer mode,
'TIEDCIRC', tied-array, with circular polarizations,
'TIEDLIN', tied-array, with linear polarizations.
'PSR', pulsar binning mode
The PSR mode can be combined with tied array modes, e.g., 'PSR,TIEDCIRC'
Only PSR is presently interpreted by CACOR, TIEDCIRC/LIN will at some point
be used by CAOBS to change the way ampl and phase calibration are applied.
4. New keyword: POINTING
The POINTING flag allows you to specify the pointing mode of the antennas.
'GLOBAL' is the standard, it uses a global (all sky) pointing model determined
at each reconfiguration.
'OFFSET' pointing uses the offsets determined on a nearby calibrator to
improve the pointing locally (near the calibrator) and for a limited time
(offsets are time variable).
'UPDATE' allows you to determine new offsets by executing a POINT pattern on
a calibrator. Selecting 'UPDATE' will set the scan type to POINT. Doing an
update while not on a calibrator is likely to stuff up your OFFSET pointing..
(no sanity check build in yet)
5. New keyword: PtCtrOffset
Pointing Center Offset
Allows you to offset the antenna pointing center (peak of primary beam)
from the array phase center (as specified with RA and Dec). This can be
useful in detection experiments: artefacts like dc-offsets are located
at the phase center, maximum sensitivity is obtained at the pointing
center. The PtCtrOffset parameter takes two values, separated by a comma,
specifying delta-RA and delta-Dec in (whole) seconds and arcseconds
respectively (ie, Pointing-RA = RA + offset-RA, no cos(dec) term).
Note that the off-line software (AIPS/miriad) will assume that the
pointing center coincides with the phase center and will thus get the
primary beam correction wrong.
Time and bandwidth smearing increases linearly with distance from the
phase center, so making the offset very large degrades your image.
6. New keyword: Timecode
Specify the time system for the schedule (this is a global parameter):
'REL' - relative times, lst - caobs will use the scanlengths to determine
when to stop and start scans. This is the default.
'UTC' - absolute times, utc - caobs will wait for the specified start-time
to arrive before starting the scan. This is used for VLBI
observations which need to be synchronized between stations and
for observations of solar system objects. If you specify UTC then
the source names will be checked against the following list:
SUN, MOON, MERCURY, VENUS, MARS, JUPITER, SATURN, URANUS, NEPTUNE
and PLUTO (case insensitive). For any matching source names the
correct position and proper motion parameters will be filled in
when the schedule is written out. A separate log file with
positions and times will be written.
Note that the listing produced using LIST will always be in lst.
This capability replaces the separate programs sched_vlbi(3) and
sched_planet.
7. New keyword: UT start
Specify the absolute start time for the schedule. You need to have TimeCode
set to UTC to see the UT start field. This is a global parameter specifying
the start of the first scan. See TimeCode for more details.
8. Changed keyword Sctype
This is now used to specify either DWELL or a pattern (MOSAIC, FREQSW,
POINT..). 'CDWELL' is no longer recognized, the 'C' goes in CALCODE now.
9. New keywords: Dual Freq, 1Line Mode, 2Line Mode
These take the place of items on the former options page. They will
change to ON automatically if an item that needs them ON is accessed.
(E.g., 2Freq 8640 will switch Dual to ON).
Most of the above descriptions can be found in the internal help for sched.
Appendix B
A Progress Report on the Data Quality Observations
==================================================
Bob Sault, Robin Wark
15 September 1995
Observation Problems (specifically for 27/6/95 and 22/8/95)
===========================================================
* The XY phases on one IF on the second set of frequencies was far from zero.
Possibly CACAL is not getting run on the second set of frequencies (??) but
then we might expect bad delays (which we are not experiencing).
The resultant step in XY phase (~40 degrees) was interpreted
as a system glitch by the "SYSCAL" flagging algorithm in atlod. 25% of the
data for the first IF was consequently flagged.
We are going to check this at the first opportunity (after VLBI finishes).
* There was an error in the sched file, where the blank field declination
was misentered by 20 degrees (!) for the 20/13 cm observation.
* Probably the observations should use no on-line time averaging (these
observations used 30 sec averaging), and a 13 cm frequency of 2368 MHz
(2382 was used).
* Overall data quality was far from as good as the ATCA can produce.
Reduction Status
================
* MHW has implemented automatic flagging of the data. Reduction now takes
~1 hour of elapsed time for 12 hours observing (3 fields x 4 pol'n x 4 freq).
* The automated flagging still leaves some bad data. We probably need to
do better here (new flagging tasks needed).
* The reduction produces a superabundance of plots and numbers (100s pages).
Some effort has gone into condensing this, so that problems will get noticed
in the mass of data. Still we can do better here.
Some results
============
* The 1934 field clearly shows confusing sources at 20 cm at the 10's mJy
level (agrees with NEBK's previous images).
* The blank field is not very blank. There are two ~14 mJy sources at
6 cm, with one of the sources still ~6 mJy at 3 cm.
* The thermal noise limit has been reached in some of the polarisation
images (6 and 3 cm). Residual sources limited the "blank field" I images.
* For XX and YY, theoretical closure phase is reached at 4800 for both
the on- and off-axis 1934 fields. 1344 and 2382 are not bad, but
undoubtedly limited by confusion. 8640 is about 1.5 times theoretical,
possibly as a result of decorrelation within the 30 sec effective
integration time.
* Off-axis polarimetric response is quite apparent. The 13 and 3 cm responses
are grossly similar, and differ from 20 and 6 cm.
Questions
=========
* What are we trying to achieve? Possibilities include
- Detect gross errors (e.g. is the array working?)
- Understand more subtle effects (off-axis polarimetry?)
- Produce a deep image of particular fields (science?)
Are we willing to put the manpower into it?
* Do we really want to do the analysis with no flagging.
* Do we need a better blank field?