[mmaimcal] Re: [Almasci] Two antennas

[Richard Hills]

Dear Al,

Thanks for sending this round.  I realise that you had a difficult job 
to do in summarizing a lot of different views and that you had an 
unreasonably short time to do it.  I am somewhat unclear as to what this 
document is going to be used for and why there is such a rush.  As far 
as I am aware the only decision that needs to be made very urgently is 
whether or not to ask AEC whether to rotate the quadruped by 45 degrees 
or not.  As far as I can see from this document the conclusion is that 
this is not justified, but it does not actually seem to say that.  Would 
it not be helpful to give the Antenna IPT a clear statement on this?  
Perhaps that has already been done separately and I missed it.

If so then the concern is over the impact on costs, which is indeed the 
title, and in reality the likely loss of quality in the scientific 
results. I don't see why either of these needs to be decided by today, 
but in an attempt to meet your deadline here are my immediate thoughts.

The first item is 1-2 additional FTE (and implicitly more computer power 
or less output) as a result of having to use different primary beam 
shapes for the two designs. I don't think that the facts presented here 
make this case.  The note explains that this will only be necessary for 
"super high fidelity / dynamic range" mosaicing.  .  What is not 
demonstrated is that the differences between the beams of the two types 
will be larger than those that will occur between individual antennas of 
the same type anyway.  Mark's simulations show that the maximum 
differences due to the blocking are at the 0.3% level.  The patterns of 
the individual antennas will differ due to things like the details of 
the feeds, the receiver optics and the alignment.  I haven't done 
calculations, but I would expect the differences in the patterns due to 
these effects to be greater than 0.3% of peak.  I am not sure what the 
6% specification that Mark mentions actually refers to, but if it is 
what has been used to set the tolerances on things like the receiver 
alignment then it is very unlikely that we will actually have maximum 
differences within a single antenna type as  low as 0.3%.  The logic 
therefore is that, if we need to do this because of the different 
antenna types, we would have needed to do it anyway to get to this level 
of fidelity.  In fact we will have to do the full case of carrying 
individual primary beam maps for each of the antennas.  The extra effort 
and complexity cannot therefore be "blamed" on the antenna purchasing 
choices, so this should be removed.

The second item is a further 1 FTE for "commissioning complexity".  I 
certainly agree that there will be additional effort required in 
commissioning the two antenna types and my own guess is that this will 
be a good deal more that the 1 person year that you suggest.  What is 
not clear is how much of this falls in the Science area and how much in 
the Antenna IPT.  I would have thought mostly the latter and I assume 
that this is being accounted for elsewhere.  What does this additional 
Science IPT person actually do?  Is it assumed that they actually carry 
out the characterization of things like path and primary beam shape as a 
function of elevation, temperature and wind?  If so then 1 FTE extra 
seems low just for the Science IPT.

 From the "COmmon Mode Errors" section on I found it a bit difficult to 
find what some of the the conclusions are.

On gravitational sag, it says that it should be OK but no figures are 
given.  In fact this is one of the places where there is a big 
difference between the two designs.  I suspect it will be OK because we 
assume that we always phase reference to a nearby source but I do think 
this needs to be checked quantitatively.

Conversely on thermal effects figures are given and the implication 
seems to be that it will be OK if a correction is made based on 
temperature monitoring, but it doesn't actually say so.

On pointing there is a clear conclusion, but on fibre length we are 
still in the realm of wishing that we didn't have different designs and 
the same applies to most of the mechanical and "other" sections.  It 
seems to me that we need to decide whether there are things that we 
actually need to do here - such as trying to make sure that the fibres 
do actually see nearly identical conditions in the different designs - 
or that, given for example that we do have line length compensation in 
the system, we do not need to take any special action in which case we 
should conclude that there is no cost or science penalty.

Best Richard