[mmaimcal] Re: your spiral configuration

[John Conway]

> Hi John,
>
> Thank you very much for sending me the position of your spiral
> configuration's antennas. It gave me a chance to compare your
> configuration with some my configuration. You are right your
> configuration has rather small sidelobes (<12%) inside of the circle od
> radius 40 lambda/D. It is because your configuration is very
> center condensed. I am attaching the PS file (CONWAY_48_CONF.PS)
> showing the configuration and UV coverage. As a result of the central
> condension your synthesised beam has rather wide wings. Compare the
> slice of your beam (48.CONWAY.SLICE2.PS) with the slice of one of my
> compact configuration (48.EQ3000.SLICE2.PS).  To be honestly I do not
> know if the wingth are so dengerous. But of cource the configuration
> with such a strong central condension has a disadvantage for the
> relevant class of the sources which provide a lot of information in the
> large baselines.

The 48 element spiral in memo 260 instantaneously had antennas
distributed over a range in radius  from the centre of the array of 10
(and ratio of longest  to shortest of baselines of
order 40) while the 64 element one in 216 had a range of radius  8. The 64 
element design was thus a little less centrally concentrated and so gave a
more gaussian main lobe (though not such  a good peak sidelobe - but in
the 64 element case I havn't yet  tried to optimised delta phi to minimise
sidelobe). 

In fact I am a bit worried that most of the filled designs we have
recently been considering, both yours and mine, (in
contrast to the  old single rings or Reuleaux triangles) don't have enough 
short spacings anyway, in this regard most are far worse than a 64 element
VLA 'Y' shape covering the same range of radius, but of course the
snapshot coverage is infinitely better than a Y.  Imaging simulations
are needed here.


> Another point: What about the compact configuration? It seems to me
> you'll have a problem in zooming your spirals to the size of 150m with
> limiting spacing of ~13m. Will you?

Havn't really thought to seriously about this, at present the zoom 
portion stops for the array in memo 216 stop at peak baseline 300m
if one assumes 2pads/antenna in the zoom portion. Perhaps it can slowly
evolve into a tightly wound spiral where the spiral tracks are 20m apart 
(one type of spiral is called Archimedian the other is called... I've
forgotten what its called.. but you probably know what I mean, ones 
equi-angular the other equi-spaced), but perhaps pseudo-random
designs would be better from the point of sidelobes
- the compact configuration needs a lot more attention in general, and
then  as you say there is the question of interfacing to the larger
configurations. 

> You wrote in one of your last message:

>>Finally I would add an important point - my original reason for 
>>proposing self-similar designs in memo 216 was not really to allow a
>>continuous variation, that was a secondary issue(!), rather it was to 
>>give the maximum saving in infrastructure costs (pads, roads and
>>reconfiguration time). As an example my 63 element design in memo
>>216 was such that it required the reconfiguration of only 1/3 of the
>antennas
>>to change the array resolution by a factor of 2 (and give an exactly
>>self-similar beam).

> I think that it is important point. I offered the concentric circles
> configurations (memo 226) with common circle and common pads. These
> configuration required moving only 2/3 of antennas changing
> configuration in scale 3 times.  You idea gives even more effect.
> BUt at that time I was told that the
> pads price is small in comparison with antenna price.....

I suspect that when you add everything ups, pads themselves, conduits to 
connect to them, 'roads' the cost all adds up.

BUT even more important: the sharing of pads between configurations 
obviously significantly reduces the amount of moves of telescopes neeeded
between configurations. This is why I was so interested to find out 
in your memo 265 discussing the cost-benefit of different numbers of 
configurations  about the value of N_{o) assumed
(number of telescopes left stationary when changing configuration) 
As you say for your 2 ring designs N_{o} is one third of the  antennas 
i.e. 21 so it already significantly effects the calculation
of configuration efficency and the  move cost (Holdaway  Memo 199, equ
15). As the number of different configurations becomes large N_{o} 
approaches N_{a}, with  further savings in array efficeny and cost.
Hence the tendency to increased efficency as the number of configurations 
already  noted in 265 is reinforced, furthermore the total move cost
must saturate because for a self-simalar array the total number of 
telescope moves needed  is always the value given by: (numbers of pads -
numbers of antenna), since every pad is moved to once except the
pads occupied at the begining.

Of course there are all sort of practical reasons why a very large 
number of configurations might not be possible and moving small 
numbers of antennas becomes impractical (as Min has well argued), 
so one expects the 'zoom' concept to break down at some point but I think 
there are important forces arguing  for more arrays rather than fewer. 

Again as I noted above my original motivation was not really to add
yet another 'nice feature' of zooming to ALMA and damm the cost or 
the practicality(!); rather quite the opposite, to try to reduce resources 
and ease reconfiguration  by sharing as much as possible betwen
configurations.

Anyway I have to sign off now and go home(!) .. I didn't really mean 
to write another long email which is too long for anyone to read.....
tomorrow I have  to help calibrating headstacks on our VLBI tape drive; since there 
is a an EVN session coming up, and record quality is not as good as it
should be....

 Good to correspond

 Cheers

 John