CA - problem forum #3
1 September 1994
1. In future the meetings will be held on the first thursday of each month.
-- next meeting, thurs, 6 OCT.
A. A number of issues, principally related to mosaicing were discussed:
1. Samplers
Warwick described a proposed modification to the samplers which would speed
up the sampler servoing.
At present the sampler statistics are based on all the samples of a cycle, from the
SYNC at the start, up to the End of cycle; the new proposal is to define a separate
window within the cycle - in effect, to wait until the antenna is on source before
starting the statistics collection. A stop before the end of the cycle will allow the
ACC to compute the new sampler levels in time to install them for the next cycle.
To implement this scheme:
a. A new event is required: ON to start the sampler collection; OFF to end it.
(ACC effort - thought to be straightforward).
b. A hardware modification to the samplers (a small "piggy-back" board).
(Effort: perhaps 1 hour/sampler, once the design and testing are complete).
c. Warwick suggests that the samplers be switchable between two modes - the current
(entire cycle) and the new (with the sampler window). That is, a signal from a
dataset would be required to enable the new mode.
d. Some backplane wiring, to bring the event to the sampler; and to bring the MODE
signal from the dataset.
Ron beresford suggests this would require several hours/antenna
The concensus was that the DECEMBER shutdown was a realistic target; with the changeover
being completed within 1-2 weeks.
It would be desirable to implement the digital sync. demods in this same period.
2. Digital Sync Demods.
a. The test unit is in place, apparently working. Derek McKay will look to collecting
data comparing the original analog units with the digital model.
b. Full implementation of the new set (once the tests are complete) hinges on G.Graves
completing the multiplexor design -- George Graves/Mal Sinclair/Warwick Wilson to
negotiate. December shutdown is the plan.
3. Dynamic Hold:
This needs the antenna drive time algorithm. Dave McConnell is testing the latest
version. Once this is done the algorithm will be installed in CAOBS.
Warwick said that the correlator implermentation of the dynamic hold will be
straightforward.
4. ACC Environments
64 different environments can be defined. CAOBS recognises up to 512 different
mosaic fields. CACOR apparently has a limit of 100.
Much discussion centred on whether the environment flag should be linked automatically
to the mosaic ID. Lister argued against the link - ie, he felt that the observer
should have the option of specifying the environment.
The extreme cases: with a uniform field it could be desirable to follow closely
the Tsys variations with elevation --- ie, one flag for the entire mosaic;
with a confused field (CEN A) it could be better to have a separate flag for each field.
Warwick/Neil argued that present evidence is that the "zero" statistic seems to remain
honest even under severe provocation (CEN A), so that the on-line corrections for
erroneous statistics should cope satisfactorily in most conditions. (With
the new sampler statistics machinery we could expect the statistics to be very close
to correct for all but the first cycle of a new field. And the on-line correction should fix
the first cycle.
5. Strobed attenuator.
This waits for George Graves ..??????
6. Sched and Mosaicing.
Some thought has been given, but at the moment we wait for the user interface question to be
resolved.
"Hanning mode" is in sched. CAOBS sends the request to the correlator in the scan header.
On-line diagnostics for mosaicing: Dave plans a new utility which would give some
indication on what is happening - where the fields are (in ra/dec), etc.
Some utility in sched would also be useful.
B, xy phase.
Andrew Bish showed some plots of current xy phase performance -- they are quite impressive.
Some questions:
- do we make best use of the xy phase?
- should the CACAL philosophy change?
*************************************
coordinates -
Practical problems with synthesis coordinates.
a. NCP/SIN in aips.
AIPS contains a set of subroutines to convert between image
pixel coordinates and actual coordinates on the sky.
A "--SIN" label indicates that the data was obtained from
a synthesis array, using (u,v,w) coordinates. The algorithm
is correct only at the field centre; its error increases
with radial distance.
An "--NCP" label indicates that the data was obtained with an
East-West array. The transformation is exact (for reasonable
frequencies) over the entire primary beam.
It should be noted that AT visibilities are labelled with (u,v,w);
these are not the "SIN projection onto the equatorial plane"
coordinates; they are however, the correct set to use if
we want the image to be locally equi-angular (provided that in the
subsequentt processing we discard "w").
Every coplanar array has an exact transformation. The World Coordinate
System will implement a scheme to include a description of the plane
in the header in order that the exact transformation can be exploited.
Some fudging will be necessary in aips (this is not to say that
aips can handle all coplanar arrays).
A coplanar array remains coplanar even after allowing for precession.
b. some further thoughts.
An east-west array has the further property that the array normal
(the polar axis) is common to every field. All these fields can
therefore share the same beam (but in terms of angles on the sky
the beams become increasingly elongated towards the equator).
Thus mosaicing would be simplified if equatorial projection
coordinates rather than (u,v,w) were used.
In computing the tracking phase we allow for all sorts of effects (aberration,
refraction, etc); most of these are not included in the computation of the
the visibility coordinates (u,v,w); aberration is probably the worst
case - there is a small error at the edge of large fields ... in the
very worst case it amounts to a position error of 0.3 arcsecs at the edge
of a 1 degree field. This is a "plate defect" which could be allowed for
in the pixel coordinate <-> sky coordinate algorithm.