[asac] NOTES ON PROPOSED EXTENSION OF BAND 3 TO LOWER FREQUENCIES

[Al Wootten]

Dear ASAC Members:

Larry D'Addario has submitted the following notes, which I propose that
we discuss on the 13 Nov teleconference.  -Al
======================================================================
NOTES ON PROPOSED EXTENSION OF BAND 3 TO LOWER FREQUENCIES
L. D'Addario, 2000-11-07


BACKGROUND

In the report of its March 2000 meeting, the ASAC recommended a
priority ordering for the 10 ALMA receiving bands.  They placed Band 3
(89-116 GHz) in the first priority group and Band 2 (67-90 GHz) in the
second priority group.  However, they also recommended a study of the
feasibility of extending the low end of Band 3 to 86 GHz to cover the
SiO line at 86.2 GHz; if that is done, they said, Band 2 would drop to
third priority.

To date, the recommended study has not been carried out, although
informal notes (emails) giving opinions about the feasibility of some
components have been circulated.  Nevertheless, the JRDG has proposed
that the official specification for Band 3 be changed so that it
covers 84-116 GHz (with no explanation of the further extension beyond
that suggested by the ASAC).  It is reported (in an email from
J. Payne on Oct-18) that the AEC would like the Systems Engineering
Division to comment on the system design implications before approving
the change, but the Systems Group has not received any direct request
from the AEC.

In the report of its October 2000 meeting, the ASAC assumes that Band
3 has been extended to 84 GHz, and it changes the priorities for
several bands from those in its March report, all without explanation.
This is premature and unjustified.

The following notes are presented on behalf of Systems Engineering.
It should be understood that this is done in somewhat of a vacuum,
there being no other available analysis and no explanation of the
proposal beyond two sentences in the ASAC's March report.


ANALYSIS

The present frequency range for ALMA Band 3 is 89.0 to 116.0 GHz.
This is a frequency ratio of 1.303, which is quite similar to the
frequency ratio for most other ALMA bands.  For reference, these are:

1	31.3-45   1.438  (HFET band, compromise optics)
2	67-90	  1.343  (HFET band, compromise optics)

3	89-116    1.303  (SIS or HFET, TBD)
4	125-163   1.304
5	163-211   1.294
6	211-275   1.303
7	275-370   1.345
8	385-500   1.299
9	602-720   1.196
10	787-950   1.207

Possible extensions would result in

3+	86-116	  1.349	  Includes SiO at 86.2 GHz
3++	84-116	  1.381   Includes methanol at 84.4 GHz
3+++	80-116	  1.450	  Includes VLBA coverage of 80-96 GHz

Thus, it would become the widest band except perhaps for band 1, where
compromises have already been made to accomodate this.  Difficulties
caused by having to operate over a wide fractional bandwidth include:

-- Windows: Coverage of Band 6 already requires a 5-layer design, and
   satisfactory performance has been demonstrated only in simulations.
   The current Band 3 could be scaled from this, but Bands 3+ or wider
   would be difficult to accomodate without poor performance (>5%
   loss?) in the outermost ~20%.

-- SIS mixer fixed tuning leads to degraded performance 

-- LO range (for DSB or sideband-separation mixing)
   101-104 currently		1.029
    96-104 for 3++ with 4-12 IF 1.083
    92-108 for 3++ with 4-8 IF	1.174
    84-108 for 3+++ with 4-8 IF 1.286

   It appears that most of these can be supported within the baseline
   LO architecture, except perhaps the last (worst) case (which would
   require multiplier bandwidth beyond the SOA for tunerless devices).

-- LO range (for SSB conversion, HFET option)
   1st conversion to 14-22 GHz IF:
    75-94 currently	   1.253	WR10
    70-94 for 3++	   1.343	WR10-
    66-94 for 3+++	   1.424	WR12+

   The extension would be hard to accomodate within the present
   architecture, especially for Band 3+++.  Of all the >50GHz
   frequency multipliers required for the array, the one for the
   present band 3 must support the largest frequency ratio, and no
   power amplifier for this range is currently planned.  For any of
   the extended ranges, the feasibility of the required components is
   in doubt.  Even if suitable components were available, the need to
   span two standard waveguide bands makes implementation messy.
   Thus, the extension may preclude the SSB HFET option.  (An HFET
   front end with DSB conversion is still feasible, however.)

-- Optics
   The relatively long wavelenth of the extended low end would require
   larger optical components, including lenses, mirrors, window, and
   horn.  The window for 89 GHz was already large enough to force the
   use of room-temperature refocusing optics.

(Band 7 may also be too wide for similar reasons.)


CONCLUSIONS

The system architecture does not contain any "show stoppers" for the
band extension.  Even the extreme case of "band 3+++", 80-116 GHz,
might be accomodated.  However:

- Some options are excluded, especially that of having SSB conversion
  following an HFET front end (with low side LO and an extra frequency
  conversion), due to the required LO range.  An HFET amplifier
  followed by a DSB or sideband-separating mixer would be feasible,
  since it would require the same LO range as an SIS receiver.

- Performance will suffer in several ways:
  . SIS mixer noise temperature will be relatively poor over a
  significant portion of the extended band.  It could be optimized for
  the upper, middle, or lower part of the range.
  . SIS mixer performance may be poor over the upper part of the 4-12
  GHz IF band, for all input frequencies.
  . Receiver temperature and aperture efficiency may suffer over a
  signifcant part of the band due to performance of windows and other
  optics.  Again, performance can be optimized over a pre-selected
  part of the extended range.

- For an HFET amplifier, performance is not significantly compromised
  by the extended frequency range.  Thus, requiring the extended range
  favors selection of the HFET option.

Every one of the proposed extensions, and more, is covered by Band 2.
There is thus no scientific advantage of any extension to Band 3
unless Band 2 is deleted.  Furthermore, the Band 2 and Band 1 front
end assemblies ("cartridges") are the simplest and least expensive of
any band (since they use HFET amplifiers, simplified optics, and
cooling only to 15K).  Including Band 2 in the initial complement of
front end assemblies incurs only marginal cost, since all the
necessary infrastructure (dewar, most of LO, support electronics) will
already be available.

It is therefore recommended that:

a.  No change should be made in the range of Band 3, keeping it at
89-116 GHz, so as to achieve the best receiver performance and
aperture efficiency.

b.  Band 2 should be retained as no lower than second priority, and
its construction (along with Band 1) should be undertaken relatively
early.  Building the Band 2 and Band 1 cartridges should be among the
first uses of any available contingency funds or cost reductions
relative to present estimates.

c.  Should it be decided that the lower part of Band 2 (67-84 GHz) is
of so little interest that it can be permanently deleted from the
array (not just reduced in priority), and only in that case, the above
recommendations might be reversed.  Band 3 should then be extended by
the minimum amount necessary (since performance degradation increases
monotonically), and Band 2 should be deleted.  Elimination of Band 2
has advantages in optics simplification, allocation of focal plane
space, dewar space, crogenics load, and cost.  Only if these
advantages can be realized are the design compromises of the extension
to Band 3 justified.