The LSA/MMA antenna study group met at IRAM in Grenoble during 19-21 August, 1997. Study group members present were:
John Lugten (NRAO) Torben Andersen (ESO)
Peter Napier (NRAO) Dietmar Plathner (IRAM)
David Woody (Caltech)
Also in attendance for parts of the meeting as "observers" were Richard Hills from the UK and from IRAM Albert Greve, Alain Perrigouard, Jean Delannoy and Stephane Guilloteau. Most of the technical issues that were discussed are mentioned in the attached report prepared by Richard Hills for his management. Here we summarize the decisions and action items that were made at the meeting. Copies of all overhead transparencies used at the meeting were distributed at the meeting.
1. Goals of the Study Group
The primary goal of the study group at this time is to provide the antenna cost and performance information that the LSA/MMA Science Study Group needs to determine which arrays are possible and acceptable to the US and European scientific communities. Specifically, the study group will produce cost estimates for antennas in the diameter range 8 m to 15 m. The surface accuracy requirement for the purposes of this costing exercise will be 25 m rms for all diameters. Since it is anticipated that the pointing requirement will be difficult to achieve we will try to determine the cost for two different pointing requirements: 1/30 beamwidth at 300 GHz and 1/15 beamwidth at 300 Ghz. For an 8 m diameter reflector these two requirements are 1 arcsec and 2 arcsec respectively and for a 15 m reflector they are 0.5 arcsec and 1 arcsec respectively. In the report that we prepare, as well as giving the cost information, we will give a written description of each antenna design so that the complexity of the technology required for the various antenna diameters can be assessed.
2. Costing Basis
It is important that all members of the study group cost their designs on the same basis so that the cost comparison as a function of diameter is meaningful. All cost estimates will be for production quantities of antennas as supplied by a commercial prime contractor. Cost estimates will be made using a common cost template in which the antenna is broken down into its many component parts. Entries in the template will be insufficient detail so that the cost basis is evident. (eg. x kg of fabricated steel @ y $/kg). A first draft of the cost template based on D. Plathner's 15 m budget was discussed. P. Napier is responsible for producing a final draft of the cost template.
3. Specification Definitions
Detailed discussions lead to first draft definitions for the computation of surface accuracy, pointing accuracy and phase stability. Responsibility for producing final drafts of these definitions is as follows:
Surface accuracy - P. Napier
Pointing accuracy - J. Lugten
Phase stability - D. Woody
4. Differences in Antenna Performance Requirements
During a discussion of the antenna requirements as currently planned by the European and US groups the following differences became evident and will need to be negotiated:
(i) Low elevation limit: The European design has 15 degrees as a lower limit and the US design has approximately 5 degrees. The US desire for a lower limit comes from the need to use a mountain-top beacon for holography and the belief that with the exceptional atmospheric conditions of the high site useful astronomy can be done at elevations as low as 10 degrees.
(ii) Solar observing: The US design allows direct observations of the sun. This requirement is not currently included in the European design. It is believed that the only impact of this requirement is to roughen the machined aluminum reflector panels by leaving machining grooves of a few microns depth. This has little impact on cost or performance.
(iii) Single dish (total power) observing mode: The US design allows for the possibility of including a subreflector nutator. This requirement is not included in the European design. The impact of this requirement is to increase the load carrying ability of the tripod or quadruped to handle the additional mass at the apex. It is possible that a fast scanning total power observing mode will make the nutator unnecessary.
(iv) Subreflector deformations. Experience at IRAM indicates that distortions of the subreflector due to differential solar heating can have a significant impact on total surface accuracy. This effect is included in the error budget of the European design but is not in the US error budget. The US will include the effect in their error budget in future.
5. Design Responsibilities
The US will study designs for 8 and 10 m diameter. ESO will study a 12 m diameter design. IRAM will study a 12.9 m and a 15 m design.
6. Study Group Communications and Next Meeting
The group will communicate primarily by e-mail and arrange teleconferences only on an as-needed basis. If we do have any teleconferences they will be arranged at 0900 US Mountain Time, 1700 European Time. So that we can have a frank exchange of views the group agreed that received e-mail messages will not be forwarded outside of the group (tanderse@eso.org, plathner@iram.fr, richard@mrao.cam.ac.uk, dwoody@caltech.edu, jlugten@nrao.edu, pnapier@nrao.edu). As a minimum, each group member will send out a brief e-mail progress report each Friday summarizing work done during that week. The next face-to-face meeting of the study group will be in Socorro with the following agenda:
Monday 29 Sept - arrive in Socorro
Tuesday 30 Sept - meet in Socorro
Wednesday 1 Oct - am meet in Socorro, pm visit VLA site
Thursday 2 Oct - meet in Socorro
Friday 3 Oct - depart.
The goal will be to produce a draft report by the end of this meeting.
Notes on a meeting to discuss Antenna Designs for Large Millimetre
Arrays Held at IRAM, Grenoble, Aug 20/22 1997
Richard Hills
This meeting was the first in a series planned to study the practicalities of forming a close collaboration between the US Milli-Meter Array (MMA) and the European Large Southern Array (LSA) Projects. It was attended by:
John Lugten (now NRAO, recently BIMA) Torben Andersen (ESO)
Peter Napier (NRAO) Dietmar Plathner (IRAM)
Dave Woody (Caltech - OVRO)
I attended as a "UK Observer" and there were a number of other IRAM people in attendance including Albert Greve and Alain Perrigouard. A separate meeting on Systems Design aspects was run in parallel. I joined that group for some discussion of the atmospheric phase correction problem.
Background
The situation is that each group has already put considerable effort into preparing antenna designs, but have been working with different sets of specifications and have started with different assumptions as a result of their experiences in previous projects.
The US requirements have placed emphasis on mapping large fields, which
tends to favour large numbers of small dishes (forty 8-metre antennas is
the nominal MMA choice) and is claimed to need extremely high pointing
accuracy (1/30th of a beamwidth). They have recently extended the frequency
coverage planned to 850 GHz, so high surface accuracy (better than 25 microns
rms) is also required. Their design is relatively conservative with
a conventional (but very stiff) yoke mount, a steel backing structure and
aluminium panels machined from castings. This is clearly based on the successful
aspects of the BIMA and OVRO designs (which are now at least 20 years old)
although there is a good deal of innovation in points of detail - e.g.
an elevation bearing and encoder system that prevents deflections of the
yoke from upsetting the pointing. The use of carbon-fibre in the backing
structure has been avoided on the grounds of cost and because of bad experience
with this material on the SMA project. This means that the weakest point
of the US design is probably the thermal deformations, where they have
relied on the BIMA experience that low temperature differentials can be
maintained in a steel structure by covering it and blowing sufficient ambient
air through it. The Americans feel that this design is sufficiently complete
that, if the MMA were proceeding on its own, they would go out for bids
for two prototype antennas early next year.
The European ideas have focussed more on making high-resolution images
of compact objects (e.g. proto-galaxies at very high red-shift) which means
that the main requirement is to have large collecting area at frequencies
between 100 and 300 GHz. Their nominal scheme is for fifty 15-metre dishes.
The antenna design is developed from that of the IRAM interferometer, with
a carbon-fibre backing structure and aluminium panels machined from solid
billets. A variety of telescope mounts have been explored, including many
with unconventional features such as non-crossing axes, in order to meet
the difficult simultaneous requirements of stiffness, transportability
and pointing accuracy. It is suggested that laser interferometers and other
optical devices be used to measure the pointing instead of relying on the
mechanical and thermal stability of the mount. Given the larger diameter
of the European dishes and their relatively low weight (which is what tends
to happen with carbon-fibre designs) the most difficult problem for this
design appears to be obtaining good pointing accuracy in fluctuating winds.
My impression is that, while some parts of this design can be regarded
as rather secure, the more innovative aspects are at a relatively early
stage of
development with detailed R & D now needed.
The question that everybody is focussing on in the collaboration is whether to go for a heterogenous array - i.e. a combination of large and small antennas - or a homogeneous one with just one antenna design. (This may not really be the most important question but it has the merit of being one where a clear choice can be made!) As I see it the advantages of the heterogeneous approach are:
(i) it should be possible to obtain better overall performance - that is to say, to come closer to meeting the conflicting requirements of large effective aperture at the lower frequencies, high efficiency in large scale mapping and good performance at the very high frequencies - for a given total cost.
(ii) it should be possible to use the different antenna sizes to fill in the uv-plane better than with a single antenna size: in particular the problem of missing short spacings should be reduced and earth-rotation synthesis with the the larger antennas on the longer baselines and the smaller ones on the shorter baselines should work better than with a homogenous array.
(iii) it would enable the US and European sides to proceed independently,
at least during the next development phase, without having to bring their
timescales and approval processes into line. (There would still have to
be a great deal of coordination so that common control systems, receiver
packages, etc. could be used. A common foundation design would be highly
desirable
but might not be practical.)
The obvious disadvantages are:
(a) the processing and calibration of data is more complex, and many terms which tend to cancel out with a homogeneous array (e.g. all the antennas bend by the same amount when exposed to the same solar heating) would no longer do so.
(b) even if overall manufacturing costs were lower, more money would be spent on developing two independent designs and all the problems of maintenance, staff training, spares, etc., would be increased. In addition, the antenna transporters would either be more elaborate or there would have to be two different transporter designs.
(c) by concentrating on a single antenna design, it ought to be possible to combine the best ideas and experience from both communities. (Even with a single design, it ought to be possible for the two partners to proceed independently with the procurement of dishes at the speed dictated by the
availability of funds, although this approach would still produce some inefficiencies.)
It is planned that a conclusion will be reached on this question, probably in October, by bringing together the results of studies by this antenna group with those of other groups working on the scientific and managerial aspects. The main task of the antenna group was therefore seen as providing figures for antenna costs as a function of diameter and performance (with pointing
accuracy seen as the most important parameter, as it appears to be the hardest requirement to meet, although surface accuracy obviously has some effect on cost too). I note that attention will also need to be paid to the cost implications of (b) above.
Discussions and Visit
The bulk of the time was spent in detailed discussion of technical points, both on exactly how to define and interpret the specifications for these antennas and on how well the various proposed approaches could meet such specifications. The second day was spent at the IRAM interferometer on Plateau de Bures so we could have a close look at the antennas and systems
there. It would take too long to report on all the issues covered, but here
are some points that I felt were significant:
1) for dishes larger than 8 metres it seems likely that the use of carbon fibre structures would be almost essential to overcome thermal problems. (A structure with CFRP tubes and steel end-fitting has a CTE of about 3ppm/C which is about 4 times lower than steel.) It was however pointed out that cost is still an issue and that it would be possible to install a good deal of temperature control equipment for the same amount of money. As an extreme case one could consider wrapping every bar in thermal insulation and heating it to a constant temperature. (Note that a recent study of similar issues for the LMT project suggested that it would be very difficult to achieve a uniformity of better than about 0.3 C, which is probably not good enough. Power consumption might also be a problem.)The carbon-fibre tubes for the IRAM dishes are now manufactured by a Finnish firm (using fibre from the UK) and quality is reportedly excellent. Plathner stated that the predicted very high Young's modulus and low CTE are achieved and that effects like moisture absorption are not a problem. They cost about 160k pounds per antenna, of which over 100k is for the steel end-fittings rather than the CFRP itself. Note however that only about half the tubes on the current IRAM antennas are CFRP.
2) although the carbon-fibre structures of the IRAM dishes seem to have been highly successful, some of the original panels are now in poor shape. These have a carbon-fibre-sandwich construction with aluminised teflon reflecting surfaces. Small holes in the teflon made by wide-blown particles and ice have allowed water to penetrate and destroy the aluminium coating. IRAM are planning to remove the teflon and are hoping to replace it with a combination of conducting and protective paints. This will probably be possible but they expect to pay a significant penalty in higher emissivity. The secondary mirrors, which have similar construction, are also becoming damaged and are being replaced by machined aluminium. For their new antennas (numbers 5 and 6) they have switched to machined aluminium panels, which look very good. They are light in weight but robust and should have good thermal properties. The cost was quoted at about 550k pounds for one 15-m antenna, including 180k for the aluminium blanks. This figure (nearly $US 5000/sq m) is still rather high if one is hoping to achieve a total collecting are approaching 10,000 sq m, but there is clearly scope for further cost savings. (I note that this is already an interesting price as regards replacing JCMT's panels. These panels are however not quite accurate enough for that application and require five support points rather than the three provided on JCMT. If additional subframes were included it would probably be difficult to get the weight low enough to match the present ones on JCMT.)It seems to me that one possible avenue for development for the LargeMilli-metre Array is the use of subframes which support the panels in an optimised fashion at a larger number of points (perhaps six or nine?) and connect to the backing structure at just 3 or 4 adjusters. This would enable the machined panels themselves to be much thinner. The question is whether the savings in machining and material costs out-weigh the additional complexity.
3) Gravitational errors in the structure are not likely to be a major problem provided the structure is supported by a ring which is at a relatively large radius. (This was done on JCMT using a set of 12 "cone-bars" - copied from the 100 metre. Dietmar Plathner has taken this further with his 15 m design and produced an elegant hexagonal supporting structure that connects thering to the mount.) The backing strucutures of all the designs presently being considered use the Radial/circumferential scheme. Dave Woody pointed out that we should at least have a look at the space-frame approach used on the Leighton dishes (and now favoured for the LMT) with a basically triangular symmetry and hexagonal panels. This should produce simpler joints and have less congestion in the central region.
4) Difficult problems are however caused by wind loading, especially because it is very variable in form and some fraction of the force fluctuates quite rapidly. It is not clear how much can be done to optimise structures to resist these forces apart from simply making them intrinsically stiff -e.g.by making the structure as deep aand well-supported as possible and then simply adding to the thickness of the members. It is important to note that one cannot treat the wind deformations in the same way as is done with gravitational errors and subtract the best-fitting paraboloid unless one imagines that one is going to measure the wind forces and adjust both the pointing and the positioning of the secondary in real time according to some model of the dish behaviour. Some passive compensation may be obtained through chance or good design - i.e. the wind deformations may turn out to move the secondary to roughly the right position - and I feel that active pointing corrections could also be considered, but I would have thought that active control of the secondary in response to wind forces is probably going too far for this project.
5) The pointing requirements do look severe and it is not at all clear that the traditional approach, which relies on the stiffness and stability of the mechanical structure together with encoders on the two axes, will be good enough. Optical devices on the market have certainly advanced in the last few years and it is clearly worth investigating these to see what could be done. Note however that the requirements for positioning these and providing clear paths through the structure means that they need to be incorporated in the design from an early stage.
6) For an interferometer there is, in addition to pointing, a path-length stability problem. Fluctuations in path result from deformations of the dish and (axial) movements of the secondary mirror as well as from flexing of the telescope mount. It is again reasonable to assume that slow variations due to thermal effects and, to some extent, steady winds, can be calibrated out, but the rapidly fluctuating components due to wind buffetting create a problem that definitely needs to be addressed. The structures must either be stiff enough to resist these or a way of measuring the effects will havetobe provided and real-time corrections applied. I estimate that for 10%loss(18.5 degrees rms phase error on each antenna) at 850 GHz the path errors need to be less that about 19 microns rms.
Conclusions:
I was not able to stay for the final session on the morning of the 22nd but I believe the intention is that the two teams will prepare cost estimates(and error budgets?) for their existing 8 and 15 metre designs and will then both develop and cost schemes for intermediate sizes (between 10 and 12.8m). A second meeting will be held in September (23rd to 25th?) in Socorro, NM,to attempt to reconcile the resulting figures and produce a report. I do not expect to be able to attend that, but if it is felt desirable for thereto be a continuing UK presence we could see who was willing and available.
Richard Hills 27th Aug '97