Minutes, CA-forum #13
Friday, 3 November 1995
1. Data Quality Observations.
- A summary of the observations is provided after each run. The
summaries are available on the web, at URL
http://wwwnar.atnf.csiro.au/www/operations/dqo/dqo.html
- The procedure is now running smoothly.
- RPN would like to able to recommend that our blank fields are suitable
candidates for the HST southern sky blank field.
>>> D.McKay will check.
2. Off-axis polarisation.
R.Sault made a C-band holography survey of the antennas, with 1934-638
as target. The cross-polarisation image confirms the good performance
of the CA at this frequency. The result - averaged over 4 antennas -
shows a weak "clover-leaf" pattern, with the peak at a half-beamwidth
from the main axis, at a level of 2% of I. This pattern means that we
can correct for instrumental polarisation, to first order, over the
entire primary beam. The image, in postscript form is appended
(as an attachement to this e-mail)
This will become an option in miriad - see RJS for details.
The other frequency bands are less likely to be as accomodating
(see earlier ca-fora for the S-band problems); cross-polarisation
holography will be undertaken as a first step.
M.Sinclair has prepared a brief on the question of replacing the fins
in the L-band OMTs. He suggests that we upgrade the LNAs at the
same time. see AT/39.3/054
3. 128 MHz birdies.
Harmonics of the 128 MHz sampler signal have been a long-standing problem.
Briefly:
a. very narrow band signals can be observed at all multiuples of 128 MHz,
on all baselines.
b. the phase of the signal corresponds to a source at the phase centre.
c. the signals originate in the samplers, most probably leakage from the
connectors, and enter the feed.
d. the problem has occasionally been very pronounced when the connectors
were faulty.
e. even when the connectors are performing up to their specifications
(and most are now thought to be in this category), the phenomenon
persists.
f. no additional hardware palliatives have yet been found.
g. some software schemes have now surfaced which can reduce the
magnitude, without entirely eliminating the effect.
BUT: at this stage it is not clear just how serious is the problem.
i. SET DE_BIRDIE_ON in CAOBS will activate a phase switching scheme
which causes the birdie to alternate phase (from +90 to -90 degrees)
while leaving the astronomical signal largely unaffected. MJK's
test on a single baseline showed a reduction from 600 Jy to 15 Jy
in a one-hour observation. A 15-minute observation by R.Sault
showed a modest overall average (5.5 Jy) only marginally improved (to 1.5 Jy).
The scheme deliberately offsets the delay by 0.2 nsecs. RJS has shown
in simulations that this will degrade the achievable dynamic range
(100 microJy noise, compared to 70 microJy noise after 12 hours).
R.Gough has suggested an alternative scheme whereby the sampler tracking
phase rate is offset by a small amount, so that some cancellation
during each cycle will be achieved.
>>> RGG will flesh out the details.
j. Further tests:
- examine the recent observations of T.Oosterloo which was half/half
birdie ON and OFF.
- some further extended blank field tests would be wise.
recommendations:
* For continuum work, the scheme should not be used, and the 128 MHz harmonic
should be placed at the edge of the band.
* For spectral-line observations the DE_BIRDIE_ON option would probably be
beneficial if the harmonic is within (or close to) the observing band.
* Where possible, the phase centre should be offset from the region of
interest.
4. On-line problems.
- 4-second jumps.
This was a particularly insidious problem - ARRAY was being suspended for
4 seconds, randomly throughout the day. Could bring about a CAOBS
siezure. An evil combination of circumstances (changing computer),
hardware (a faulty connection to the weather station), and software
(two alternative ways of compiling ARRAY) made the problem hard to trace.
Now fixed.
- Scan data corruption after a CAOBS seizure.
CACOR was unable to detect the changing circumstances, which lead
to the first taken after a CAOBS restart being lost.
CAOBS now assumes the worst on start-up, and ensures that CACOR tidies up.
5. X-Y phase.
R.Sault has been worrying about the phase noise in the XY measurements.
This is small, but still significantly larger than theory.
a. The "XY" phase noise derived from astronomical observations
is highly correlated with the XY phase noise due to the noise
source. Thus the phase jitter is systematic, and not due to the
source.
b. It does not originate in the conversion chain - a common UHF
synthesiser was tried.
c. It does not originate in the receiver: both conversion chains
were set to the same frequency so that A and C were identical, as were \
B and D. But the XY phase from AB was uncorrelated with CD.
(Antenna 4 was an exception here, showing correlated phase jitter of
5 degrees .. an additional problem, perhaps).
These plots are appended to this e-mail as attachments.
d. It does not originate in the samplers themselves - tests by WEW
showed that the sampler noise contributes less than 0.1 degree.
e. The XY phase and delay are correlated, which indicates that the
problem is in the sampler machinery.
A further test will arrange for two samplers to have the same clock.
The problem is low level (adding 1 degree or so phase noise), but
it would be wise to locate it. If we can remove it we should.
6. CA-forum on the web -
The forum minutes will be available on the web in the near future.
the technical details (eg, the Data quality observation summaries, or
the 128 MHZ de-birdie algorithms) can be found in the narrabri page.
7. Next meeting :
Early next year - date yet to be decided.