ca-forum #7 -- minutes
meeting held on 1 March 1995.
a. Digital Synch Demods.
George Graves is still on target for the April shutdown.
Since ant. 3 is due for repainting during much of the
shutdown, George will modify the back-plane in this antenna
during a reconfiguration period prior to the shutdown.
Time requested during the shutdown: 1 day/antenna.
ACC implementation. An event is required to control the
synch. demods. This could well be the same as the sampler
hold event.
--> Dave McConnell .. ?
Additional Mosaicing matters.
----------------------------
Jim Caswell reported that the dynamic hold calculation was
failing a large fraction of the time.
Lister found that the calculation was too conservative, and that
observing time was lost.
-- some mosaicing tests seem in order.
Lister found that antenna 1 was having servo problems, with
poorly damped oscillations ("hunting") at the end of the travel.
This problem was worst after transit.
(R.Beresford attended to this on 6/3/95)
b. (UVW) derivatives.
MRC has modified EPHMRS; MJK will modify CAOBS and advise
WEW about the revised message format.
c. The channel 33 problem.
(i). There is indeed a 128 MHz signal getting into the samplers
and producing a ~1 Jy signal in channel 33. This is present in
most sampler units; WEW's group will investigate.
HOWEVER, this is not the source of the difficulties first reported
by NEBK.
(ii). Birdies - and 1408MHz in particular.
-- This appears to be causing an increasingly high proportion
of the user unhappiness, and deserves more attention.
The prospects for a simple permanent cure are poor. This is
likely to be a drawn out affair, similar to the LO problem.
Phase switching would not help.
* There is a suggestion that the connectors are a potential
source of difficulty - G.Baines will make some exploratory
investigations - perhaps in antenna 4 which is now one of the
worst offenders.
Although the connectors are supposed to be as good as SMA
connectors experience suggests otherwise; the semi-rigid
loop to the connectors is very small, possibly aggravating
mis-alignment problems.
* Some palliatives:
SCHED will warn schedulers if a 128 MHz harmonic is within
the band.
CACOR will excise the birdie channel in the delay calculations.
ASSISTANCE will also warn the user, once/scan.
An ASSISTANCE log will be provided.
The recommended L-band band will be revised to avoid the 1408
birdie.
d. Positioning cameras.
It was felt that the cost-benefit analysis was against continuing
the project. The gains (better starting point for the pointing
and baseline solutions) did not warrant the effort/cost.
e. Data quality matters.
(i). What are the current "worst problems" ?
- the 128 MHz birdies.
- ACC drives.
- Power outages at ant 6.
- caobs "freeze" when an ACC dies.
- ?
NEBK/DMCK will scan the fault reports to look for a other
problems which need attention.
The questionnaire may ask for astronomer feedback.
(ii). Test observations.
The idea is that reguilar observations be made of some
standard field in order to provide some measure of the
array's performance.
* After each reconfiguration, make use of the second
night, if the pointing/baseline checks were ok the previous
night.
* Two fields, such as 1934 near boresight and outside the
primary beam; with a pattern:
each field would be observed at each band, in continuum and
spectral line mode - 4 scans/field.
* It is important that the data be reduced expeditiously.
* Continuity in the analysis is desirable
------------------------------------------
follow-up.
NEBK has looked at the sampler servoing - it is working well;
But mosaicing problems remain .. there is still a problem
if the Tsys changes dramatically from one field to the next.
(The Centaurus A problem): the sampler levels will be wrong
in the first cycle after any change. It looks as if the environment
matter needs to be revisited. Yet to be checked: whether the
algorithm for coping with incorrect sampler statistics is
adequate.
NEBK has looked at the recorded integration times mosaicing mode
(ie, is HOLD correctly allowed for ?)- this seems ok.
DMcK confirms that integration periods out to 30 secs. are now
possible.