[mmaimcal]comments on Y+ vs. ring configuration notes from stephane

[Mark Holdaway]

Comments on Bryans comments:

[ Stephane's text deleted (we've read it twice already now ]

[ Bryan's text in ">" ]


> 1) this may or may not be an advantage.  while there are more
>    hybrid configurations, it is not clear that there will be
>    significant demand for them.  will there be significant demand
>    for an, e.g., 10 km hybrid?  i suspect not, as the demand will
>    be for either the highest resolution or for the intermediates
>    (5 km and less).  in that case, we have a bunch of pads that
>    may end up not being needed, in these hybrids.  

All pads in the "hybrid" configurations are used in the penultimate
Y++ configuration, so there is NO waste in pads.

The main increase in flexibility is in the reconfiguration.
A half populated 14km ring array is essentially useless for
imaging experiments, not so with the Y+, which is useful for
imaging at every step.  This is particularly important given
the slowness of reconfiguration of a large array and the likelihood
of reconfiguration delays due to high winds.

> now, the extra
>    flexibility *is* there, so if it turned out that there was
>    demand for the hybrids, it could be accomodated much easier
>    than for the ring, admittedly.

When we compare the gap between the 4.5km and 14 km configurations
with what the rest of the array gives us, this is a HUGE gap.
Maybe not by VLA standards, where one doesn't have any other options.

These other arrays in between 4.5 and the BIGGEST, available only with the
Y+ Array (trademark)  will be used, though perhaps not as often as the
spiral incremental arrays. 


> 2) & 3).  there is actually not a large difference in the road
>    length, at least as i understand the latest ring and Y+ 
>    designs.  the road cost is relatively minor, as well, so these
>    are pretty minor advantages, i think.
> 4) i don't think that the phase stability in the fiber is an issue.
>    mark commented yesterday that bill shillue echoed this sentiment
>    (essentially, the residual fiber phase is << the residual
>    atmospheric phase, so is unimportant in the residual phase budget).
> 5) this is true, but isn't this beyond the current project?  while
>    we shouldn't cast a blind eye to the future, we also need to
>    consider what is in the best interests of the present.
> 6) i don't have a good handle on how big an issue this is.  mark
>    h. says he can quantify it, and i await such quantification.

I don't expect I'll have time to quantify this, but it is the same
excercise as I went through in memo 199 and Min Yun/Kogan went through in
memo 265.  What makes things very different from those memos is that
nearly ALL of the antennas are shared between two adjacent configurations,
which favors making MANY incremental configurations in the case where
there is a distribution of desired resolutions.

> 7) i thought we had decided at the configuration CDR to do away
>    with the hybrid configuration?  this also relates to my comment
>    on 1) above.

This is not because it is not desirable, but because it is too costly
for the 14 km ring (ie, the Y+ hybridization IS an advantage).

The savings of some 35 pads quoted in our memo does not include ANY pads
in a hybrid between the 14 km ring and the 4.5 km array.  A large part of
the savings is because there is a 4.5 km Reloux triangle which surrounds
the Conway outermost spiral configuration, a highest resolution array
SHORT of the 14 km configuration.  For the Y+ concept, this 4.5 km Reloux
triangle is totally unnecesary.  In addition, there are essentially zero
shared pads between the 14 km ring and the spiral/4.5_km_triangle, while
the Y+ array is able to share many pads with the outermost spiral
configuration. In the current design I am working on for the Y+ array, we
actually save about 45 pads over the spiral + 4.5km_triangle + 14km_ring,
which is significant. 

> now, i think there is one big cost issue that stephane missed that
> is an advantage for the Y+.  namely that the fact that the fiber
> lengths are significantly shorter for the Y+ config means that the
> total fiber conduit (and power conduit) length should be significantly
> shorter for the Y+.  this might be a real cost saver, depending on
> what the final cost of these conduits ends up being.  mark and john,
> can you quantify the difference in conduit length for the 2 configs?

If this becomes important, Angel is the man.

> i know that we don't have a good final estimate for conduit cost, but
> at least if we know the relative difference between the 2 configs, this
> is a good starting point.
> 
> stephane then lists as the advantages of the ring:
> 
> A) Slightly better resolution
> B) Better astrometric performance 
> 
> let me again comment on each.
> 
> A) is this true?  i thought that the effective resolution of the
>    2 was essentially the same now?  mark and john, can you comment?

Yes, should be the same.  If it isn't, we just change the Y+ array
until it is the same.

For uniform weighting, which could be used for bright sources, the Y+
will have significantly better resolution (at a loss in sensitivity, of
course).

> B) this is a real key, in my opinion.  one of the primary uses of
>    the most spread out configuration will be astrometry (a good example
>    is the astrometry of stars to see if they have detectable wobble
>    from orbiting planets).

Ed Fomalont, an expert on astrometry, doesn't see any significant
difference in these two arrays for astrometry.

> i think another advantage that is not listed is the ability to
> discriminate between two closely spaced sources, which is *not*
> the same as resolution (there is an extensive literature in the
> electrical engineering and signal processing fields on the difference
> between these two features of any signal processing system).
> it comes about for the same reason as the improved astrometry - the
> distribution of baselines is different, and hence allows for better
> discrimination between closely spaced sources, even though the
> 'resolution' is similar.  it would be nice to quantify both B)
> above, and this other feature.  can the simulation tools that john
> and mark and stephane's group (and maybe koh-ichiro or dave woody or
> mel wright) have do this with any reliability?

This is an old argument of Frazer's for the ring arrays.  I had attempted
such simulations many years ago, and was not able to show anything really
convincing then (we were having the exact same argument with the 
uniform (u,v) distribution and the centrally condensed (u,v)
distribution).  However, given that the ring array has "too many" long
baselines (they get washed out by convolving with the Gaussian restoring
beam), I anticipate that Bryan's point is likely correct, though at a
10-15% improvement over what the Y+ array will do.  For this to be
relevant, we need to do something like restore with a non-Gaussian beam,
or do model fitting (ie, not imaging).

> one final possible advantage of the ring configuration is that it
> may be true that it is easier to negotiate for the land access.
> the ring fits on the current science preserve, and hence no 'extra'
> land is needed.  this is not the case for the Y+, where we know that
> we will need extra land to the west.  i don't know if this is a big
> issue or not, but as long as we're setting these things down, i think
> it should be listed as a possible advantage of the ring, at least.
> 
> i hope these comments aren't taken as an endorsement of either the ring
> or the Y+ on my part.  i don't have any axe to grind here, nor any
> particular current preference for either of them.  i would, however,
> before dismissing one or the other, like to have a more detailed and
> thought out comparison of them, and have it in writing.  i suppose
> this should fall to john, as leader of the configuration group.  it
> is an issue that needs immediate attention, i guess, since the land
> acquisition discussion is going on even as we speak (virtually).

I imagine that John has more useful things to attend to at this moment.

	-Mark