Minutes, CA-forum #5
Thursday, 8 December, 1994
---------------------------------------
Alan Young and Gerry McCulloch were present - their patience, advice
and wisdom were much appreciated.
----------------------------------------
a. Digital Sync. Demods. (G. Graves) :
The printed circuit boards have arrived; the front panels are finished.
But - George has been overwhelmed by the Seti filter task, and has been
unable to complete the construction of the units; nor has he been able
to complete the conversion backplane wiring modifications required.
These are now put off till the next major Narrabri shutdown - April ?
b. The new samplers.
- The modifications are largely completed (REB).
- The additional event lines will be complete by friday.
- The ACC software requires some further effort, but DMcC is confident.
- WEW and ED will be in Narrabri next week to complete the correlator
side of the operation.
There was some discussion of the question of whether the variable
HOLD should synchronize the Cal. source switching. This seems
straightforward as far as the ACC is concerned; it is desirable
for the digital sync. demods; it will require some attention by
WEW to have the X-Y phase units recognize HOLD.
A CAOBS -> CACOR message will need modifying in order to get the
dynamic HOLD to the correlator prior to every cycle.
The ACC drive algorithm still needs some further attention - the last
mosaic run (in november) showed an occasional problem, with all antennas
failing to reach the target on time. DMcC will attend to this during
the present shutdown.
CAOBS needs some work to compute the HOLD time, exploiting the ACC
drive algorithm.
It looks as if the new samplers will not be able to remedy the
two-cycle adjustment problem .. (that it takes two cycles before
the ACC can complete a resetting of the sampler levels).
c. The LO system
[When I drew up the agenda for this meeting I was somewhat pessimistic:-
I feared that the conventional wisdom viewed the LO system as suspect.
My impression is now that the system is reasonably satisfactory and
workable]
- G. Baines described in detail the problems encountered over the
past two years in the course of keeping the LO system operational.
.. amongst which, cleaning the many connectors;
repairing connectors (dry joints);
fixing a bug in the PC software (a 50 KHz phase wrap problem).
The array now operates with just 5 directional couplers (only
at the active station posts), rather than with 35 units daisy-chained.
This has improved the reliability as well as reducing the amplitude
of the round-trip excursions by a factor of ~ 5.
The couplers have a substantial temperature coefficient (~ 1deg/deg C),
so reducing the number in the line must help.
The LO co-ax is now terminated at the last station post in use
(ie at antennas 1 and 5) rather than at station posts 1 and 35).
This can reduce the length of line in service and seems to help.
The current practice increases the number of connections
which are made/unmade, but this cost seems modest in face of the
gains in reliability.
The LO system design requires the transport of 2 frequencies (160 MHz
and 50 KHz); some problems have been experienced with the 50 KHz.
Since antenna 6 operates very satisfactorily with a locally
generated 50 KHz, this option should be examined for the other antennas.
Since one of the problems is mechanical stress in the connectors,
some effort should be made to investigate the temperature excursions
encountered in the station posts.
At the present time Graham monitors the LO round-trip plots every
day. These provide a good index of the health of the system.
It is the case that the astronomical phases do not follow the
long-term large scale variations shown in LOPLOT - in that sense
the LO system is clearly working well. However, it is also the case
that the LO system is not entirely quiet .. the LOPLOTs
do show a number of disconcerting episodes.
It is not clear whether the astronomical phases are immune from these
episodes. Astronomers are not complaining, so perhaps all
is well. Earlier this year (9 april) it was clear that antenna 1
had visibility phases which correlated well with the LOPLOT; Graham
has argued that a number of lines of evidence indicate that antenna 1
is a special case - possibly particularly vulnerable when at the extreme
East stations.
Conclusions and recommendations :
A. The system now seems reliable, but at some cost: perhaps 4 hours
every reconfiguration, plus a half-hour every day to scrutinise the
previous day's LOPLOTs.
B. The astronomers are not complaining.
C. The system is working: calibrator phases remain close to zero for
days.
D. But since the LOPLOTs do show occasional serious episodes,
some effort should be made to look for correlated visibility
phase episodes, whenever a simple (strong) source is being observed.
One suggestion is to syphon off all calibrator data, for subsequent
comparison with the LOPLOT data.
E. Look to generating the 50 KHz at the antenna rather than transporting
it from the central site on the co-ax.
F. Investigate the temperature cycling in the station posts, to see
whether a more benign environment might help.
-------------
There was some discussion about a future LO system.
A star-configuration, possibly in fibre, would be better than the present
single co-ax. It would also be expensive.
Whether the present system will suffice for mm operation is less clear.
The design criterion was 1 degree/GHz. The evidence is that we do better
than that, both in the long-term operation as well as the short-term rms.
A high-quality "mm subset" array could be envisaged, if desperate.
********************
next meeting: first thursday in february, 1995.