[asac] Message from Neal

[Al Wootten]

Neal sent the following message from his Leiden account, where he is not a member
of the asac list (only his Texan identity is a member)...Al
------- start of forwarded message (RFC 934 encapsulation) -------
From: "Neal J. Evans II" <nje at strw.leidenuniv.nl>
Message-Id: <200006211442.QAA10522 at strw.LeidenUniv.nl>
To: asac at nrao.edu
Subject: ALMA Receivers


Dear colleagues,

I hesitate to make a mass mailing, but I offer these thoughts
on the receiver specs for consideration, particularly the idea
of a figure of merit that weights the importance of T_rx and
bandwidth. If you like the idea, send your votes on X and Y to 
Karl...

Cheers,
Neal


Reactions to the Receiver Specs Document

I agree with the points made by Stephane and Al. Where they
differ, I agree with Al's formulation. 

In particular, the receiver specs should change with frequency, as noted
by Al. The current spec is too easy at the low frequencies, and perhaps too
hard at the high end. For clarity, I understand Al's A values to apply
to SSB T_rx.

I don't like the idea of relaxing the bandwidth spec so easily, but I
recognize that there are tradeoffs. It would
be better to formulate a figure of merit that balances T_rx with BW.
If one sacrifices too much on T_rx to meet a spec on BW, one may 
quickly lose more than one gains.
If we adopt Al's formulation, with R_rx = A hnu/k + 4 K and IF bandwidth
symbolized with BW, the figure of merit for line observations is (I leave
out the constant 4 K for clarity, but it could be worked in)
 
           F_line = T_rx/A 

For continuum, it is   

           F_cont = (T_rx/sqrt(BW))/(A/sqrt(8GHz))


Note that this formulation relates to what you can achieve in
a fixed integration time, which is a more realistic approach in
my view than asking what integration time is required for a given
sensitivity level, which would lead to squaring these figures.

The overall figure of merit depends on your evaluation of the
overall importance of line and continuum observations. Call the
overall figure of merit F_total. Then

         F_total = X F_line + Y F_cont    ; X + Y = 1

The ASAC could vote on the values of X and Y. My vote would be
X= 0.6, Y = 0.4.  The reason for this choice is not that I think
line observations are more important than continuum, but because
they are harder-- the continuum sensitivity of ALMA will be very 
good, but for some line problems, the sensitivity will be more marginal.

One could make X and Y a function of band, but I think there won't
be large differences. The idea is that one would relax the BW
spec, replacing it with the spec that F_total be less than or equal
to unity.


On other items, I agree that selecting and tuning to a new band should
take less than 15 min, unless I misunderstand what this means. Presumably,
there will be look up tables and fast iterative procedures to optmize
at a particular frequency.


Cheers,
Neal

------- end -------