ca-forum#9 - minutes
(22/5/95)
1. Digital Synch Demods.
a. The installation work by G.Graves is complete.
b. The units do not work as well as they should, and are
noisy - a factor 2 or so worse than the analogue units.
There appear to be several problems. If a constant (DC) signal
is input to the DSD we expect to see a constant output, with IN and OFF
equal. At Narrabri we find an rms of around 6 counts (with mean
counts of 26000); further, the ON and OFF are anti-correlated; this
effect is not seen in the lab. at epping, and suggests a problem with
the Narrabri clocking.
--> Ron Beresford will set up an experiment, replacing the
Voltage to Freq unit with a precision oscillator (500 KHz); the
switching waveform will continue to be derived from the ACC.
This should help localise this aspect of the problem.
If we sum the ON and OFF we find a lower level rms (2-3 counts),
with a further disconcerting pattern: a steady ramp upwards for
5 or 6 counts, followed by an abrupt fall.
The epping tests show a low level noise in the units which
is higher than expected.
--> This will be pursued by Mark Leach/Warwick Wilson.
2. (UVW) modifications. The task is to ensure that the (UVW)
assigned to a visibility are correct for all samples in a mosaic
observation. Previously the (UVW) were calculated for the mid-point
of the clock cycle (the nominal integration cycle); in mosaicing
some fraction of the start of the cycle is blanked while the antenna
is driving; the (UVW) calculation should reflect this shift in the
mid-point of the actual integration interval.
Changes were made to CAOBS and CACOR. Numerical checks have
been done. (The (UVW) derivatives have been checked; WEW has checked
that the end results are correct).
3. Sampler Birdies.
G.Baines and W.Wilson carried out a series of tests in an attempt to
understand the extent of the problem.
a. Antenna 6 was worse than the others because the OSP connectors
had been incorrectly assembled (fingers missing). Once replaced, the
birdies in antenna 6 are at the same level as all the other antennas.
The question of whether other antennas also lack these fingers was raised.
The Narrabri array was checked a yera or so ago; Mopra should be revisited.
b. The prognosis for further isolation with better hardware is poor:
the OSP connectors are not much worse than the SMA connectors.
Attempts to prevent the signal getting to the feed are viewed
unfavourably -- we'd need to close the gap between the turret and the
upper vertex room floor, thought to be difficult and unreliable.
Absorber is also seen as unlikely to help greatly.
See reports by G.Graves and R.Gough (AT/31.6.2/025, AT/31.6.3/05
and AT/39.2/055).
c. W.Wilson suggests moderating the harmonics in the signals :
---> REB will place some low-pass filters in the 128 and 512 MHz lines.
This will be trialled early in the coming month.
4. Mosaicing enhancements. The problem that needed attention was
the difficulty the system had in accomodating to large swings in
system temperature in mosaicing objects such as CenA.
D.McC has enlarged the environment space in the ACC.
MJK modified CAOBS so that in mosaicing mode the user can ask to
have environments set automatically - a different number for each field
in the mosaic. This means that sampler levels and attenuator settings
will be reset separately for each field. Trials on 8 May indicated that
the system works OK. The procedure is documented on the web.
--> NEBK will subject the system with exahustive CenA observations in june.
5. S-band beam patern assymetries.
D.McKay discovered last week that the S-band beam patterns are
asymmetrical - basically elliptical, different for the two polarisations,
with an apparent squint between the two polarisations - The original
symptom was an offset between the two half-power mid-points.
Derek then produced some beam patterns which show the ellipticity of
the patterns; they also show that the sidelobe structure is bad.
Some of the trouble can be traced to the out-of-focus nature of the
S-band installation.
Lowering the subreflector to its lowest point improved the sidelobes,
but not the ellipticity; nor did it appear to cure the squint.
It is worth noting that at present the subreflector cannot be
driven under computer control. This imposes a number of costs:
a. There is a loss of about 15% at S-band with the subreflector
locked at the (C/X) position.
b. There is a loss of about 5% at C and X while the subreflector
is at a compromise position between the optimum setting for X
and the optimum for C-band.
c. The beam ellipticities, and the different shapes at the two
polarisations will have disturbing effects for wide field observations
at S-band; imaging and polarimetry will be affected.
Recommendations:
a. That computer control of the subreflector be implemented.
(Note that the pace of the subreflector movement is painful -
3-4 minutes from C/X to the S-band position).
The electronics to control the subreflector are complete and installed -
the problem is software, ACC and CAOBS.
b. S-band never be used for pointing checks.
c. The consequences for mapping and polarimetry be estimated,
and the general community be advised.
d. It would be useful to examine the Mopra S-band performance
while it has the Seti feed. There is apparently barely any time for this
test - G.Baines wants to dismantle the feed as soon as the cooperative
program is finished. The experiment would be brief -- 1 hour of observations,
on a strong source (Virgo ... late afternoon/evening); there is some
setup .. linear polarisation would need to be restored; the pointing
dataset would need to be returned.
--> Mal Sinclair/G. Baines/MJK will examine the options.
5. X-Y amplitudes. WEW had found that the amplitudes of the x-y
calibration correlations were unreliable in mosaicing mode. The
first cycle (when HOLD was applied) seemed correct; the subsequent
cycles on the same field (when HOLD was zero) were incorrect -
the ON amplitudes decreased, and the OFF amplitudes increased).
(The phases are reported to be OK, although this proposition may
be suspect at L-band).
--> The effect seems to have been traced to the ACC. DMcC will follow this
up.
6. Data Quality observations.
( I note that these were an ATUC request from circa 1987; we're
slow, but we do listen).
a. What to observe?
1934-638 or 0823-500 depending n the season.
Two fields will be established for each case - one with the source
a modest offset from the field centre (say 1 arcminute); and one
at several primary beamwidths away (say 1 degree).
The large offset field should be empty.
--> John Reynolds and Neil Killeen will look for suitable positions.
b. Frequencies?
The observations should alternate between the two fields, and between
the two major bands:
5 minutes, 1934, at L/S bands
5 minutes, empty field, at L/S
5 minutes, 1934, at C/X bands
5 minutes, empty field, at C/X
The observations should run overnight - 12 hours if possible.
c. Who should observe? - The duty astronomer.
The plan is that these observations should
be scheduled for the last night of the reconfiguration period. If
the calibration obserevations (baseline and pointing) have failed,
then these data quality observations will be pre-empted.
Since these will be fully scheduled observations, the duty astronomer
will be expected to attend and to carry out the observations.
d. Who should analyse the data? - The narrabri staff. Neil Killeen
has volonteered himself and M.Wieringa to prepare a miriad procedure
to process the data. The load on the observatory staff should then be
tolerable: they'll have results to review and compare with previous
runs. There is a complication in that the calibration source will
not be at the phase centre .. N.E.B.K. undertakes to provide the necessary
option in miriad for this.
-----------------------------------------------------------------------
I suggest that the next ca-forum be on Tuesday, June 27, 11:00 am.
A number of figures were shown during the meeting - these are available
on application to M. Kesteven.