Ron Ekers (ATNF),
Lloyd Higgs (DRAO),
Wu Shengyin (Beijing Obs.),
Francois Viallefond (Obs. Meudon),
Wolfgang Reich (MPIfR),
Gowind Swarup (TIFR),
Yuri Parijskij (SAO),
Peter Wilkinson (NRAL),
Dick Thompson (NRAO),
Robert Braun (NFRA)
URSI Large Telescope Working Group
First Meeting
Jodrell Bank, March 21 and 22, 1994
AGENDA
March 21
14:00 - 18:00 1. Opening of meeting and adoption of an agenda.
2. Review of terms of reference and goals of the
working group. "What do we hope to achieve?"
- basically, we need to plan for a concrete proposal
for a fully specified instrument with reasonable
cost estimates within the next 2 - 3 years
- in addition, we need to develop a funding strategy
3. The international climate and national priorities.
"How can each of us help make it happen?"
- short contributions from each of the participating
countries which provide a background for the
discussion, specifically:
a) which, if any, national priorities have been set
b) which, if any, resources might become available
in the short-term for R&D and in the long-term
4. Elucidating the major scientific drivers.
"What science can we envision with a 100-fold
improvement in performance and how does this
influence the instrumental specifications?"
- successful funding will depend on a convincing
scientific case. We need to critically review
the major drivers (to the extent this is possible)
on the basis of their impact and uniqueness and then
determine which specifications (sensitivity, resolution,
frequency coverage, etc.) these demand?
March 22
9:00 - 13:00 5. Identification of critical technical developments.
"How can we achieve the desired performance?"
- at the outset we need to review the various methods
of EM concentration and detection at radio frequencies
since our endeavour will almost certainly require
a novel solution to the general problem
- a number of concepts will be discussed and analyzed
for strengths and weaknesses
6. Definition of research and development priorities
for the next half year.
"Which concrete steps should we be pursuing?"
- a novel telescope design will require substantial
R&D to assess performance and cost of new technologies
- we will attempt to identify the most promising
concepts and match them to the R&D interests and
resources of the participants
7. Planning of next meeting and closing.
URSI Large Telescope Working Group
First Meeting Jodrell Bank, March 21 and 22, 1994
Minutes prepared 7 April 94 by Robert Braun (NFRA)
March 21, 14:00 - 18:00
I. Opening of meeting and adoption of the agenda.
The distributed agenda was adopted without amendments.
The meeting was attended by Higgs (DRAO), Viallefond (Obs. Meudon), Reich (MPIfR), Wilkinson (NRAL), Wu (Beijing Obs.) and Braun (NFRA). Written contributions were received from Swarup (TIFR) and Thompson (NRAO). About 15 other interested individuals attended including Lovell, Davies, Pedlar, Cohen and Ponsonby of NRAL, Wilson (MPIfR) and Butcher (NFRA).
II. Review of terms of reference and goals of the working group. "What do we hope to achieve?"
The terms of reference and working group membership are attached. The hope is to produce a concrete proposal supported by a well-defined scientific case for a fully specified instrument including reasonable cost estimates within the next 2 - 3 years. At the same time a funding strategy will need to be developed. The planned course of action is to hold a series of meetings spaced by about 6 months to develop the scientific case and coordinate research and development efforts. Meetings are completely open to any interested persons.
III. The international climate and national priorities. "How can each of us help make it happen?"
The working group members were asked to summarize which, if any, national priorities had been set in their communities and which, if any, resources might become available in the short-term for R&D and in the long-term for capitol investment.
Higgs (DRAO) reported that the radio astronomy community in Canada consisted of about 40 individuals with rather diverse interests and that radio astronomy accounted for about 20% of the roughly 25 M$ per year budget for astronomy. The highest current priority in Canada is the commitment to the Gemini project, which implies no new capitol investments before about 2000. The Canadian radio astronomy community will meet in August of this year to discuss options for the future. An obvious area of interest is the proposed "Radio Schmidt Telescope", a compact array of 100 antennas of 12 m diameter with operating frequencies between 400 MHz and 22 GHz. That proposal corresponds well to a subset of the "Large Telescope" project. Other ideas involve further development of capabilities in the mm/sub-mm. There is the hope that about 20 M$ of capitol might become available for a new start if the community can agree on priorities. For the time being, the technical and scientific support staff are fully-/over- commited in their support of the JCMT and the Penticton array.
Butcher (NFRA) described the recent discussions about scientific priorities for instrumentation in the Netherlands. Two possible directions are being studied for a substantial involvement from about 1998 onwards: participation in a major (sub-)mm array or in a next generation cm/dm array. Recent developments at the WSRT are laying some of the ground work for a new cm/dm system. The 92 cm band has been broadened to extend from about 305 to 385 MHz. While intermittent RFI is certainly present, high quality (receiver noise limited) observations have been possible much of the time over most of the band. The new receiver series which will be placed in the 14 WSRT telescopes during the next few years will provide almost continuous frequency coverage between 250 and 1800 MHz with much-improved system temperatures. A new correlator is being built for the WSRT which will accomodate up to 16 telescopes. It is hoped that a promising concept for a cm/dm array can be proto-typed as two elements forming a North-South extension to the current East-West array. On a short time-scale a few 100 kfl per year will be available to support relevant R&D. From about 1997, ten new scientific or technical staff positions may become available to support a new instrument.
Wilkinson (NRAL) outlined the situation in the UK. In the mid-1980's a ground-based astronomy plan was formulated which led to substantial improvement of MERLIN. Much of the upgrade is now in place and MERLIN now has "National Instrument" status implying a high degree of support for off-site users although exact financial aspects for providing this user support remain murky. The time is rapidly nearing when a new ground-based astronomy plan will be formulated. This provides an obvious and necessary vehicle for obtaining funding for new instrumentation starts. At least one, relatively small local project is already well underway in terms of technical R&D. This is the VSA (Very Small Array) which enjoys a high level of local support and together with MERLIN accounts for the entire R&D resources of the NRAL. Capitol expenditures through 2000 will be totally consumed with support for the Gemini project. Pedlar (NRAL) made the suggestion that grant requests may offer a reasonable approach for funding R&D at a modest level once fairly specific questions had been formulated.
Viallefond (Obs. Meudon) related that the cm/dm radio astronomy community in France is small and has been declining in numbers, while the mm radio astronomy community has been enjoying rapid growth fueled by the investments in IRAM. A modest upgrade of the Nancay telescope has recently been approved (2 - 3 MECU and 2 - 3 years of the technical manpower) which should extend the instrument's useful lifetime for about 10 years. Another modest investment (800 kECU) has been made for a new correlator of the Solar Heliograph. The highest current priority is for continued expansion of the Plateau de Bure interferometer. A 5th antenna has now been approved and efforts are continuing to obtain approval for the 6th (the maximum number which the site can accomodate). The direct coupling of all ground-based astronomy expenditures, including those of ESO, has led to a bad climate for new funding given the VLT cost over-runs, which will almost certainly preclude any substantial investment before 2000. There is a strong feeling that the next instrumentation initiative must be international. The Canadian "Radio Schmidt" proposal has received considerable interest in France, particularly from the Solar astronomy community. A number of people are actively working on new ideas for future instrumentation.
Wu (Beijing Obs.) reported that since the summer of 1993 the position of the Chinese government towards support for astronomy has improved substantially. In the first instance, one cannot think of large capitol outlays. A number of proposals are now competing for a maximum funding of about 1 M$. There is considerable interest in participation in a "Large Telescope" project by the observatory directors. Although there is not much hope for major capitol investment, substantial assistence with an appropriate site as well as mass produced local manufacturing are suggested. The results of a preliminary site survey as well as more details of the situation are given in the report by Wu appended to the minutes. An additional contribution discussing telescope siting by Nan R., Cai Z. and Tian W. is also appended.
Reich (MPIfR) outlined the situation in Germany. Radio astronomy represents a fairly small group relative to all astronomy in the country. Current resources are divided between support for IRAM, the newly completed SMO (on Mount Graham, AZ) and the Effelsberg 100 m telescope. In the case of Effelsberg the major push is still to higher frequencies. Upgrade plans for the outer telescope panels and azimuth bearing maintenance are high on the list of things to do. The multi-feed project has suffered a slow-down, implying a completion date still a few years in the future. The current over-subscription of Effelsberg is between about 2 and 3. Many projects require extremely long integrations (20 hours) so the need for higher sensitivity is appreciated. There is clearly interest in a next generation cm/dm facility, but there are not many suitable technical staff for R&D. Expertise at the MPI is primarily in the area of total power antennas. The desirability was voiced for some type of "International NRAO" which would coordinate and carry out plans for new instrumentation in radio astronomy. This desire was echoed by Viallefond for the community in France.
Thompson (NRAO) related (by e-mail) how the national priorities in the USA were shifting to more "targeted" rather than "curiosity-driven" research; a trend that is also apparent to a greater or lesser extent in many other countries. In this difficult climate, he suggests that perhaps one should consider a "growing array" concept with a modest nuclear beggining around which the complete instrument could grow over time.
Swarup (TIFR) contributed (by e-mail) the following. "International climate may not be right at present but it will depend upon importance of science and perhaps situation will improve in two or three years. India has already committed large resources for GMRT compared to its science budget. Our contribution in future would be in terms of small R&D teams to support the international mega telescope (IMT : I like the name!). Of course, we would be extremely interested in such a venture. We have been canvassing the importance of a telescope with a large collecting area for three decades from Ooty to GERT to GMRT. In the long run, India's contribution could be important in terms of well trained manpower for building low cost radio telescopes and also many electronic engineers who are being trained now at TIFR, RRI etc."
In summary, the situation is remarkably similar in many countries. Large capitol investments are now committed, particularly for new optical astronomy instrumentation as well as mm and sub-mm instrumentation for the JCMT, IRAM and the SMO. The next opening for new instrumentation starts will be between about 1998 and 2000, making this the ideal time to begin serious preparations. A solid proposal needs to be ready by 1996 - 1997 to take advantage of the next potential funding slot.
IV. Elucidating the major scientific drivers. "What science can we envision with a 100-fold improvement in performance and how does this influence the instrumental specifications?"
Scientific Objectives: Extragalactic Research
a) Proto-galaxy and proto-cluster evolution at redshifts of 1 to 10 via neutral hydogen from gas masses M_HI <10^10 M_Sun h^-2 - requires microJy sensitivity in a 100 kHz bandwidth at frequencies 200 to 700 MHz
While this is in some sense an obvious and critical observation, the simple statement of the problem doesn't come across as the compelling argument it should be. The recent detection in CO emission of at least one high red-shift galaxy (F10214+4724) to date has led to the question of whether this type of research is not better done with the optically thick CO lines. The enclosed Fig. 1, adapted from Arnault et al. (1988, A&A 205, 41) indicates how the observed flux ratio (in Jy-km/s) of CO (1-0) to HI varies as a function of metallicity in a range of objects. It can be seen how the flux ratio varies from semi-primordial systems like the SMC with 100 times brighter HI than CO to star-bursting systems in which the CO lines have as much as 100 times the brightness of HI. The flux ratio is extended to high luminosity infra-red sources in Fig. 2 based on the data of Young et al. (1989, ApJS 70, 699). The mean flux ratio, even at the highest locally observed infra-red luminosities is about 50. If F10214+4724 follows these trends, then the 15 mJy times 230 km/s of the CO lines would correspond to 300 microJy times 230 km/s of HI or 6 times 10^10 M_sun of optically thin HI. These HI flux levels are within the reach of an instrument perhaps 3 times as sensitive as the GMRT, or only about one tenth of the scope of the instrument we are contemplating. What these considerations illustrate more than anything is the complimentarity of the various tracers and methods.
*** Action Item: Braun has been assigned the task of re-stating this critical application in a more accessible way. A question which arose was whether a CO flux existed for Malin 1. Anyone having the answer to this is requested to pass it on to Braun.
Butcher raised the question of: "What portion of parameter space is left over for the existence of large concentrations of neutral hydrogen in the early universe given the current statistics of QSO absorption lines at high red-shift?"
*** Action Item: Carilli (RUL) who has recently been working on similar issues will be asked to look into this question
b) Primordial recomb. lines of H_n-alpha for n = 20 - 40 from z = 1500. - requires mK or microK sensitivity in a 1 MHz bandwidth at frequencies 100 to 1000 MHz with arcmin resolution. Dubrovich (SAO) has been working for many years on the question of the observability of recombination line emission from the epoch of recombination at about z = 1500. Since the epoch of recombination is spread over some finite red-shift interval, the resulting spectrum will not have the line contrast of a nearby galaxy or individual HII region. In fact, calculations of the emitted spectrum have shown that the distance between subsequent lines only becomes greater than the expected linewidth for transitions having n greater than about 20, which are red-shifted to a wavelength of about 35 cm. The isotropic component of this line emission is expected to have a line strength of a few micro-Kelvin, while non-isotropic emission may have brightnesses of as much as a milli-Kelvin. Calculated line contrasts of the isotropic component are shown in Fig. 3 from Dubrovich and Stolyarov (1994 preprint). Since current efforts to detect anisotropies in the continuum component of the CMB are beggining to approach 10's of micro-Kelvin sensitivity, both the isotropic and non-isotropic components of the line emission may be accessible to a well-designed instrument. Such a detection would have a tremendous impact on our understanding of the physical conditions during, as well as the exact epoch of, recombination. This topic was felt to be sufficiently interesting that confirmation of the calculations was deemed essential.
*** Action Item: Pedlar has agreed to critically read the material in hand (preprint 1994 of Dubrovich and Stolyarov). Braun will attempt to track down, and have translated, previous Russian language literature and motivate further calculations.
c) Galaxy evolution, dark matter and large-scale structure studies between z = 0 and 5.0 using continuum, neutral hydrogen and mega-maser emission in OH, H2CO, CH3OH and H2O - requires microJy sensitivity at frequencies 300 to 5000 MHz combined with sub-arcsecond resolution.
The excellent correlation of non-thermal radio continuum emission with far-infrared thermal emission from dust (shown in Fig. 4 from Knapp 1989, in "The ISM in Galaxies", ed. Thronson and Shull, p. 3) implies that luminous star-bursting systems, even in the absence of (circum-)nuclear "monster" emission are easily observable out to cosmological distances. Normal and star-burst galaxies have a Flux(1GHz) = 5 X 10^-3 Flux(100 micron). For example, the galaxy F10214+4724 at red-shift 2.3, with about 350 mJy of 100 micron flux will have a corresponding 1 GHz continuum flux of about 2 mJy. Systems should as this make excellent candidates for the detection of mega- and giga-maser emission since both the pumping and background emission are available to amplify emission lines of OH, H2CO, CH3OH and H2O. In the case of F10214+4724, OH line fluxes between about 15 and 150 mJy are expected based on the locally observed ratios of OH to IR radiated power (shown in Fig. 5 adapted from Henkel and Wilson 1990, A&A 229, 431). Recent VLBI imaging of OH mega-maser sources (Diamond, private communication) has shown that the maser emission is not diffuse but concentrated into a relatively small number of extremely compact (mas) spots. This circumstance makes it possible to image the kinematics of such maser sources with a resolution which is un-rivaled by any other emission process. Besides OH emission, HI absorption also offers the possibility of probing gas kinematics and opacity on mas angular scales. The two phenomena often go hand-in-hand as seen in Fig. 6 (from Mirabel and Sanders 1987, ApJ 322, 688).
Neutral hydrogen emission remains the most sensitive tracer of galaxy kinematics out to large radii. Only by sampling well beyond the optical and molecular disks do the effects of dark matter begin to dominate the kinematics of at least some local systems. However, imaging studies in HI are severely sensitivity limited so that we are currently restricted to studying only the brightest, nearest systems. Well-imaged systems have recession velocities less than about 5000 km/s, while the most distant galaxies currently in reach have recession velocities less than about 15000 km/s. ***
***Action Item: Braun will look into a survey mode observation with the "Large Telescope" to assess the depth and impact of a blind or directed HI red-shift survey.
d) Power source of quasars and active galactic nuclei - requires 10's of microarcsec res. and microJy sens. at 5000 MHz.
*** Action Item: Wilkinson will flesh out the scientific case for continuum observations generally, both in a local as well as a VLBI mode. One question which can be worked out for example is: "Is there a Balck Hole in every galaxy?"
Scientific Objectives: Interstellar Medium Research
a) Deuterium emission imaging in the Galaxy and nearby galaxies. DI/HI is the most sensitive and direct probe of the baryon content of the universe. - requires sub-milliKelvin sensitivity at arcmin resolution in a 20 kHz bandwidth at a frequency of 300 MHz.
It is clear from Fig. 7 taken from Boesgaard and Steigman (1985, ARAA 23, 319) that D offers the largest gradient of fractional abundance with the baryon density of any nucleosynthesis product. In principle it should be possible to observe DI in emission at about 92 cm just as we observe HI at 21 cm but at a line strength reduced by about 10^-5. Sufficient sensitivity would allow imaging of the entire galactic plane as well as the disks of nearby galaxies in the DI line. Wilson raised the question whether ISM chemistry wouldn't severely distort the primordial abundance of DI by preferentially tying up D into HD under conditions of moderate density and shielding columns.
*** Action Item: Wilson and Braun will look at the literature on this question and consult with Van Dishoeck (RUL).
b) Magnetic field measurements throughout the Galaxy and other galaxies using the Zeeman splitting of right and left circularly polarized HI emission or absorption (shift of 2.8 Hz per microG) - requires 10 to 100 mK sensitivity at arcmin resolution in 5 kHz bandwidth at frequency of 1400 MHz with minimal polarization cross-talk
One of the few methods for obtaining spot measurements of the line-of-sight magnetic field strength in the ISM is that of Zeeman splitting. With sufficient sensitivity and dynamic range the entire Galaxy and even nearby external galaxies become accessible to this method. Assuming good polarization characteristics of the instrument, an interferometric measurement has substantial benefits over total power measurements which are more subject to uncertainties and differences in the polarized beams. An example of the application of this technique in HI absorption towards Orion A is shown in Fig. 8 from Troland et al. (1989, ApJ 337, 342). Field strengths in excess of 100 microG are observed which are well-correlated with the HI opacity indicating the critical role that the magnetic field plays in ISM dynamics.
*** Action Item: Cohen (NRAL) will look into the possibility of observing Zeeman splitting in the faint OH emission lines from quiescent dark clouds which would make it possible to probe magnetic field strengths under these conditions. This is in contrast to the observation of Zeeman splitting in OH maser line emission which probes the rather more extreme physical conditions within maser sources.
c) Recombination line imaging in H, He, C and S of ionized and partly ionized zones around massive stars in the Galaxy and in star-burst galaxies out to high red-shifts - requires microJy sensitivity at frequencies 200 to 5000 MHz, in a 20 kHz bandwidth combined with arcsecond resolution.
Ionization conditions and kinematics of HII regions are effectively probed by recombination line observations at cm wavelengths. Some of the outstanding study areas in this field are zones of highly enhanced Helium ion abundance and the extent and kinematics of the partially ionized CII zones surrounding the Hydrogen Stromgren sphere. These topics are illustrated in Fig. 9 taken from Roelfsema and Goss (1987, PhD thesis, RUG).
An exciting area of research that has only recently opened up is the high resolution, kinematic imaging of highly obscured star-burst galaxies utilizing stimulated recombination lines of H. One of the first examples of this technique is shown in Fig. 10, also from Roelfsema and Goss (1987, PhD thesis, RUG) for M82. Line strengths of 0.1 - 1 % of the (thermal plus non-thermal) continuum have now been observed in a handful of galaxies. Applications are severely sensitivity limited. In principle, physical properties and kinematics of even high red-shift systems can be observed at high resolution. For example, F10214+4724, should have line-strengths of between a few to some tens of microJy per line. With a broad-band correlator it is possible to enhance detectability by co-adding multiple lines.
*** Action Item: Viallefond has been involved in recombination line imaging of a number of external galaxies. He will flesh out the case for this research direction.
Scientific Objectives: Stellar Research
1) Pulsar id and timing throughout the Galaxy and nearby galaxies:
a) dynamics of globular clusters and galactic bulges,
b) gravity wave propagation and other GR effects,
c) pulsar-blackhole binaries,
d) planetary companions. - requires microJy sensitivity at frequencies
200 to 2000 MHz.
Pulsars offer a wealth of possibilities to better understand the universe through, for example, tests of General Relativity (as illustrated in Fig. 11 from Taylor and Weisberg 1989, ApJ 345, 434) and the detection of planetary mass objects (as illustrated in Fig. 12 from Wolszcan and Frail 1992, Nature 355, 145). A wide variety of other applications for these naturally occuring precision clocks awaits us. The pulsar population in nearby external galaxies will also come into reach if we can extend our sensitivity down to the microJy level as illustrated in Fig. 13 (from a presentation by Pogrebenko 1993, NFRA).
*** Action Item: Lyne and Wilkinson will work through more specific examples of the possibilities for pulsar research.
2) Normal star detection to study
a) mass loss via stellar winds,
b) resolved surface and circumstellar features,
c) planetary companions by orbital wobble,
d) protostellar disks and jets,
e) solar flares and active regions,
f) kinematic evolution of associations. - requires microJy sensitivity
at freq. 200 to 5000 MHz, spatial res. from sub-arcsec on the Sun to 10's
of microas, time sampling (variability) from microsec to years.
Besides the many applications relating to specific objects, it is increasingly important to link the optical and radio astrometric frames. The best objects for linking these frames are relatively normal stars, for which there should be minimal ambiguity about the co-extension of the radio and optical emitting zones.
*** Action Item: Taylor (U. Calgary) will be requested to supplement
the scientific case for stellar research.
Viallefond is requested to consult with the French group considering
solar research to bring that case up to date.
March 22, 9:00 - 13:00
V. Identification of critical technical developments. "How can we achieve the desired performance?"
The desired performance which follows from the scientific objectives can be summarized as follows:
a) Continuous frequency coverage between about 200 to 2000 MHz, secondary coverage at 5000 MHz.
b) Full polarimetry capability for both high sensitivity and intrinsic polarization properties.
c) Maximum possible sky coverage, tracking and FOV to obtain: source access, high sensitivity, observing efficiency and allow useful VLBI. A useful specification might be a minimum of 45 deg. isotropic steering with respect to some reference direction with better than 50% sensitivity.
d) A range of angular resolutions for particular applications of order 1 arcmin, 1 arcsec and 0.1 arcsec and with space VLBI to 1 mas at 1000 MHz and 30 microas at 5000 MHz. (Interstellar scattering places these limits on the VLBI resolutions.)
e) Good imaging properties in a modest integration time.
f) microJy flux sensitivity giving brightness sensitivity of: sub-mK at arcmin, 1 K at arcsec and 100 K at sub-arcsec angular scales.
g) No sensitivity limitations induced by source side-lobe confusion.
These performance requirements dictate to a large extent what the global design of the instrument must be:
a) microJy flux sensitivity implies a total collecting area approaching about 1 square km (with conservative assumptions regarding system temperature and aperture efficiency)
b) sub-mK brightness sensitivity at arcmin resolution implies approximately filled (synthetic) aperture(s) of 300 - 400 m diameter
c) sub-arcsec resolution demands at least some baselines out to several 100 km
d) 1 K sensitivity at 1 arcsec resolution demands that "most" of the collecting area be distributed in a region of about 40 km diameter
e) good, fast imaging properties demand that the collecting area be distributed into more than about 30 sub-concentrations in a two dimensional configuration
f) avoiding source side-lobe confusion demands a limited size of (synthesized) primary beam with a low sidelobe level. Since the current generation of 25 m element size synthesis arrays is running into this problem now, a factor of 100 greater sensitivity will probably be realizable with a factor of 100 smaller primary beam area, ie. an "element" size of at least 250 m. A sufficiently low sidelobe level in the primary beam argues for either a single filled aperture or a "many" element compact array.
Together these attributes translate into the schematic configuration shown in Fig. 14. The total collecting area is split into some 30 units, most of which are distributed in an elliptical region of some 40 km diameter with several outriggers extending to radii of a few hundred km. Each unit will have an approximately filled aperture of 300 - 400 meter diameter. The basic design question then becomes:
"What form should the basic unit take?"
Before considering various concepts for the unit design it is useful to review the current possibilities offered by various basic feed and antenna designs.
Feeds and feed arrays with a net forward gain:
1) The simplest type of feed is the half wavelength dipole located at a height of a quarter wavelength above a reflective ground plane. A pair of crossed dipoles can be used to measure both polarizations. The FWHM of the antenna beam is about, H=90 deg. Maximum bandwidths (defined here as upper operating frequency divided by lower operating frequency) are about B=1.3, although with special efforts at broadband matching B=2 can be realized. The effective collecting area of a single ideal dipole at wavelength, W, is Ae=W**2/8. Variants of the simple dipole include placing the dipole probes inside a conducting canister and adding choke rings about the canister opening for modifying the illumination. Complete wavefront sampling can be obtained with an array of dipoles spaced by slightly more than about W/2.
2) A more extreme variant on the simple dipole is the so-called "patch antenna" consisting typically of a thin conducting square or circle of size D=W/2, placed on a dielectric of thickness, T=0.07W, and backed by another layer of conductor. Both polarizations should be accessible with appropriate tapping and combining networks. The typical H=70 deg., maximum bandwidth, B=1.2 and Ae=W**2/4. These types of feeds lend themselves well to mass produced feed arrays with deposition or etching techniques, in particular at relatively high frequencies. Complete wavefront sampling should be possible with a feed array spaced by about 0.7W.
3) Log-Periodic Dipole Arrays are basically christmas tree-like or pyramid-like collections of dipoles which have an active zone where the dipole length is a good match to the observing wavelength. The region of shorter dipoles above the active zone is basically transparent at a given frequency, while the region of longer dipoles acts as the ground plane for that frequency. The natural feed-point of the LPD array is at the apex of the structure. Both polarizations can be obtained. The opening angle of the structure determines it's beamwidth yielding H=60 to 120 deg., which stays constant over an extremely wide bandwidth of B=10 to 20. The effective collecting area is comparable to that of the corresponding simple dipole, ie. Ae=W**2/8 for a single polarization. Since the active zone of the LPD is frequency dependent, the best performance as a feed for a parabola is obtained by placement of the active zone in the point focus. If movement of the LPD is not allowed, the focus can be optimized for some central frequency or the LPD can be placed in the diverging beam just behind the point focus, which leads to a small, constant degradation of performance across the band. An array of LPD antennas with a common orientation can only achieve complete wavefront sampling for the longest operating wavelength given the physical dimensions. However, a converging cluster of LPD's can be arranged to have approximately half wavelength spacing of subsequent elements over the entire frequency range. Hemi-spherical clusters of LPD's can be constructed to provide complete hemi-spherical sampling of, for example, the diverging wavefront near the focus of a parabolic or spherical reflector.
4) Quad-ridged Horns offer another means of obtaining extremely broad bandwidths, B=10 to 40, with two polarizations, although the beam scales as H = W/D for a opening D comparable to Wmax, and a typical length, L=2*Wmax. The effective collecting area is independent of frequency, Ae=D**2. Horn feeds are currently favored for most applications since they generally offer the best noise performance (especially when the entire structure is refrigerated) and the most flexibility regarding illumination over a relatively narrow (B=1.5) frequency band, although they are relatively expensive in terms of both material and fabrication. Complete wavefront sampling is not possible with horn antennas. Spatial undersampling is at least a factor of two in each dimension even when close-packed, since D=Wmax.
Reflector Types:
1) The parabola has traditionally been the reflector of choice for radio astronomy given the obvious benefits of a point focus for illumination of the collecting area with a single feed. Since the frequency range under consideration corresponds to a maximum wavelength of about 1.5 meter, the smallest paraboloid that can be considered for prime focus use is about 9 m in diameter, due to the substantial diffraction losses which would otherwise be incurred. If, on the other hand, an additional reflection were desired for illumination shaping or some other reason, this size limitation would apply to the secondary, implying a primary of perhaps 100 m in diameter. A feed array can be placed near the focus of a parabola to (a) obtain an extended field-of-view (to a maximum of about 10 - 20 beamwidths), (b) compensate for low order surface errors (c) enhance illumination efficiency or (d) all of the above. Given the restricted FOV of a parabola (even with a feed array), some provision must be made for mechanical pointing in two dimensions to obtain the specified sky coverage.
2) The spherical reflector has found quite successful application at the Arecibo observatory. The reflected ray geometry near the focus of a spherical primary is illustrated in Fig. 15 together with the simplest (single reflection) Gregorian corrector geometry for obtaining an on-axis point focus. The Arecibo upgrade plan calls for both a secondary and tertiary reflector (as seen in Fig. 16) to obtain efficient illumination of an offset spherical segment of about 65 degree opening angle (as seen from the focus). In this case the secondary has a diameter of about 0.08R (for a radius of curvature, R) and the tertiary about 0.025R. For Arecibo (R=315 m) these dimensions correspond to about 25 and 8 m respectively, implying a lowest useful frequency of about 300 MHz. If only a single reflection were employed, or the diverging beam above the caustic were intercepted with a feed array (like a LPD cluster), smaller primary dimensions could be used. The inherent spherical symmetry of the primary reflector implies that pointing can be accomplished be moving only the feed structure. Unfortunately, the natural feed point stays close to the surface at R/2 from the center of curvature, which implies fairly large moves of the feed structure. For a non-movable primary, the FOV spec. implies a 125 deg. opening angle as seen from the center of curvature, corresponding to a diameter of 1.78R and a height of 0.48R. Small spherical primaries could conceivably be pointed mechanically, although it would have to be demonstrated that there were some advantage (eg. mount simplicity) over the parabola in this case.
3) The parabolic cylinder has been considered for application to large collecting areas. This geometry produces a line focus, which requires complete sampling in one dimension with an array feed to effectively utilize the physical collecting area. Only the simple dipoles and patches offer this capability in a linear geometry, with a maximum bandwidth, B=1.2-2. Mechanical placement of 5 or 6 different feed arrays (possibly mounted on different faces of a shaft) would therefore be necessary to achieve the required frequency coverage. Electrical pointing could then be employed along the feed array direction (at the expense of reduced sensitivity relative to on-axis) with mechanical pointing only necessary in the other dimension. Full, direction-independent sensitivity of the telescope could be maintained by employing mechanical rather than electrical pointing in azimuth with a turntable arrangement.
4) Naturally ocurring reflective surfaces at the relevant radio frequencies with some focusing power are not common. While pure water has a very low conductivity (and fairly substantial transmission attenuation) it already has a 65% reflectivity below 20 GHz. it can be transformed into a "perfect reflector" below a few GHz with ion concentrations corresponding to a 2.5% solution by weight of HCl or about 5% of H2SO4, HNO3 or NaOH. Unfortunately, these concentrations correspond to about 10,000 times what is found in a typical swimming pool and are rather lethal. The lower ion concentrations found in sea water already lead to an enhanced degree of reflectivity (more than 78% below 1 GHz) but naturally occuring bodies of water are typically "flat" (in calm weather) and concave on large scales. The most reflective naturally occurring liquid surface on the planet is probably that of the Dead Sea or the Great Salt Lake in Utah. Moist, salty sands and clays provide the highest conductivity of naturally occurring solid surfaces. Reflectivities as high as 50% or more at 1 GHz are possible in contrast to the 10% reflectivity typical of dry rocky surfaces. Unfortunately, moist and sandy are usually not found together naturally except below the surface where they don't serve our purpose of reflector well. One possible exception to this dilemma may apply at the underside of glaciers. Pure water ice is very transparent at radio frequencies (with an exponential attenuation pathlength of 2 km), while the glacier slides on a liquid layer of wet, ground, rock which may have a significant reflectivity. If the lower ice surface were sufficiently smooth relative to the observing wavelength, some efficiency as a concentrator may exist. The ice could also provide a low cost method of placing and tracking a feed array (like a hemi-spherical LPD cluster) near the "focus" of an appropriate glaciated valley. Another variant on this theme is the iceberg telescope, where again the disparate transmission and reflection properties of pure water ice and sea-water are utilized. Although the prospect of using a naturally occurring "telescope" is attractive, it seems unlikely that this will be practical.
System Performance:
In order to usefully compare various concepts we need to consider the relative contributions to system performance that apply to each design. The sensitivity of a particular system is directly proportional to both the effective collecting area and the system temperature. The total system temperature is composed of the sum of the following contributions:
Tsys = T_cmb + T_sky + T_gal + T_far + T_rec
in which T_cmb is the 2.7 K background, T_sky is the contribution of atmospheric absorption and emission which has a cosecant(zenith-angle) dependance and zenith values of about 2.5 K for clear sky observing. In wet conditions T_sky will remain the same below about 600 MHz but will double at 5 GHz. T_gal is the Galactic foreground contribution which has a -0.6 spectral index and is strongly dependent on pointing direction, with minimum values of about 35 K at 300 MHz and 1 K at 1400 MHz. T_far is the far sidelobe contribution which is strongly orientation dependent and consists primarily of spill-over and diffraction pick-up of the warm ground. For example, the prime focus WSRT system has minimum values of T_far = 20 K at 300 MHz and 14 K at 1400 MHz, while projections for the GBT call for only 5 K at 1400 MHz (although scattering has not been included in the GBT estimates). T_far contributions from even non-optimally illuminated reflectors can be cut dramatically by the use of reflective ground screens to redirect far sidelobes back to the cold sky. T_rec represents the losses due to the particular feed and LNA design. Narrow-band (B<1.5) systems with cryogenic cooling to less than 20 K can achieve T_rec of 10 K (case CR) or slightly better, while simple ambient temperature systems with B=2 can currently achieve T_rec=20 K (case NB). Broad-band ambient temperature systems with B=10 - 20 may achieve T_rec=40 K (case BB) but this is an area where much development needs to be done and there is room for substantial optimization. Some form of simple electrical cooling may also allow improved performance from non-cryo systems. Total system temperatures at 300 MHz will be about 70, 80 and 100 K for cases CR, NB and BB respectively, while at 1400 MHz they would be about 25, 35 and 55 K. The conclusion from the preceeding discussion is that of order 50% more effective collecting area will be required to achieve a given sensitivity if narrow-band, cryo-cooled systems can not be employed in a particular concept.
Concepts for the "basic unit":
1) A single 500 m diameter spherical cap (R=300 m) of which some 200 m diameter patch is efficiently illuminated for a given pointing direction. This is similar to the proposed "Large Southern Radio Telescope" illustrated in Fig. 17, although in that case only 15 deg. East-West coverage is allowed for rather than the 90 deg. specified here. The basic dimensions of the structure are impressive with a height of about 150 m at the edges. The only practical way to build such a structure is to make extensive use of existing terrain. Appropriate geologic terrain is found in "Karst formations", limestone formations in regions of heavy rainfall in which subterranean cavities are formed which eventually collapse leaving hemi-spherical depressions. Thompson (NRAO) reports by email that: " ... the name comes from the original example on the Dalmation coast of Yugoslavia. Other examples occur in Kentucky, Yucatan Peninsula, Cuba, Puerto Rico, and the Nullabar Plain of Australia. Examples are also found in parts of Southern China, Southeast Asia, Malay Peninsula, Java, Celebes, the Moluccas." A suspended wire mesh would form the reflective surface, while the feed structure would probably be supported by cables. A price indication can be obtained from the "Large Southern Radio Telescope" proposal, although these are based on high performance through at least 12 GHz and possibly 25 GHz. Cost estimates in that case were 25 M$ for the main reflector and 45 M$ for an Arecibo style Gregorian plus a novel focal structure suspension system. By adopting a high frequency cutoff which is lower by about a factor of 10, and possibly a feed array for real-time electrical re-focusing, greatly reduced performance specs. on the mechanical structure may be realized. These in turn might lead to very substantial reductions in cost. Actual estimates based on the required specs. still need to be made. Survival of a low-cost primary surface would be most easily realized at a site with no snowfall or ice build-up.
All of the subsequent concepts call for a "basic unit" built up of a close-packed array of one form or other. This immediately implies many more LNA's (at least a factor of 16) for which cryogenic cooling is no longer felt to be a viable option out of maintenance considerations. If feed arrays are adopted for the monolithic spheres, this conclusion also applies to that concept. The implication, as discussed above, is a factor of 1.5 or so more effective collecting area for a given required sensitivity. An important part of the compact array concept is the assumption that the compact array will be co-phased for one or more directions with a local phasing network. Only the N beams (a minimum of N=1) of the compact arrays will be correlated at the central correlator rather than all of the individual elements. In this way the central correlator requirements remain manageable even for large numbers of small sub-elements.
2) A compact array of 16 x 50 m paraboloids (or 125 m spheroids). The most cost-effective design in this size class to date is that of the GMRT (as illustrated in Fig. 18) for a fully steerable (15 deg. elevation limit) paraboloid. Cost reduction relative to more conventional designs has been realized by the use of a wire mesh surface with low wind loading (only 7%) and a simplified tension-shaped backup structure. The mount is a fairly conventional alt-az type. Swarup estimates a cost of 200 US$/m^2 for the complete mechanical and electrical system (excluding receivers). The design is rated for survival in wind speeds of 130 km/h, although snow or ice build-up would be a major survival problem. Further cost reductions might be realized by limiting the sky coverage to the specified minimum of 45 deg. elevation. This would lead directly to a height reduction and may make a non-conventional but still effective mount possible. Efficient illumination and compensation of low order surface errors could be carried out with a feed array, possibly reducing the mechanical tolerances. The non-movable spheroid possibility would still demand a lot of assistence from the terrain, which may be hard to realize for a close-packed array. The relatively small number of elements which are phased together to form the instantaneous synthesized primary beam may lead to an unacceptably high sidelobe level.
3) A compact array of 64 x 25 m paraboloids. This size class has been "traditionally" employed in the current generation of synthesis arrays. Traditional fabrication techniques in welded steel are far too expensive to be considered for application to the instrument under consideration. It is possible that some type of geodetic space-frame or segmented fiber-reinforced plastic technique (see below) may be applied in this size class. Together with a restricted sky coverage, and therefore lower height and cost mount system, a solution in this direction might be feasible. While the number of elements in this concept is larger than in the case of 50 m paraboloids, the sidelobe level in the instantaneous synthesized primary beam may still be only marginally acceptable.
4) A compact array of 256 x 12.5 m paraboloids. Paraboloids in this size class may be near the upper limit on size for fabrication via molded, fiber-reinforced plastics. This technology offers good performance at very low cost since minimal antenna assembly is required. Retail prices for severe weather rated paraboloids with good performance at 4 GHz in the 6 m class are 50 US$/m^2 including a feed support and fixed mount. (An upper limit to the price of 7 m class dishes including tracking capability of 78 US$/m^2 was obtained from the Beijing Asian Satellite Communication Technique Company Ltd. in the enclosed contribution of Shengyin Wu.) Since the required numbers of elements are comparable to complete production runs, substantial cost savings over these retail prices are expected. Larger paraboloids can be constructed by bolting together a small number of sub-elements. The actual upper limit to paraboloid size accessible to this technology must still be ascertained. Although simple LPD feeds and LNA's could be mass produced for such antennas at low cost, a low maintenance steerable mount is required. In addition, provision must be made for data transmission from each LNA. Both of these problems may be addressed by ganging, say 16, of the small paraboloids into groups. In the case of a steerable mount, rigid rods might be used to interconnect groups of, say, 4 by 4 telescopes at appropriate pivot points to reduce the number of drive motors required. In the case of data transmission, (steerable) co-phasing may already be carried out within such sub-groups to minimize the cabling requirements within the "basic unit". Some idea of the appearance of such a unit is given in Fig. 19.
*** Action Item: Braun will attempt to determine actual performance figures for a broad-band LNA and LPD feeding a small paraboloid.
*** Action Item: Viallefond will look into the mechanical solution adopted in tracking and steering of the French solar furnace which employs some hundreds of independent mirrors.
5) A phased array of 30,000 x 1 m^2 "tiles" or horns. The basic unit of collecting area might be a "tile" something like that illustrated in Fig. 20, where 1 low frequency printed circuit dipole is surrounded by 4 higher frequency dipoles, each of which is surrounded by 4 higher frequency dipoles, ... and so on until the entire frequency coverage is obtained with a set of 5 different dipole sizes. An integrated LNA is used for each dipole and the tile is equipped with all necessary interconnections and local co-phasing networks for each dipole set. For good performance at all frequencies, the tile would probably need to be fabricated in five layers, in order to provide dipole ground planes at appropriate depths. The total tile thickness would probably be 0.07Wmax with an appropriate dielectric substrate. Electrical steering could be employed to scan the approximately 90 deg. beam of the individual dipoles without the need for moving parts (although at the cost of reduced off-axis sensitivity). An alternate fabrication method would employ only a single printed dipole and electronics layer. This could be suspended at about Wmax/4 above a conducting scatterer in order to disperse the backward directed power of the dipoles and so give diminished, although broadband, performance. Pyramidal or quad-ridged horns could also be used as the mass-produced elements of a phased array. A single broad-band LNA would be used for each horn (although current LNA designs with B=10 have poorer noise performance than those with B=2). The limited FOV of each horn, makes it necessary that some provision be made for mechanical pointing and tracking. As discussed for the small paraboloids, some system of ganging is probably indicated for both mechanical aspects and data transmission.
*** Action Item: Noordam (NFRA) and Braun will do more work on tile arrays to establish probable cost and performance.
*** Action Item: Pedlar will work on further defining a horn array concept.
*** Action Item: Bos (NFRA) will be requested to provide some crude estimates of cost for a correlator and (sub-)phasing networks.
*** Action Item: Braun will look into power consumption for various types of components (integrated LNA's, drive motors) to assess infrastructure requirements of compact arrays.
Telescope Site
A critical consideration in choosing the best telescope site is the interference environment. The entire planet already suffers from irradiation by the GPS and GLONAS systems (near 1200 and 1800 MHz) and many other applications are competing for further space-based allocations. Given the desire for high transmission bandwidths and the wide variety of current ground-based allocations, we may be relatively safe from further incursions at the lowish frequencies under consideration here. Between about 860 and 1400 MHz air-borne, shipping and ground-based radar and communications are the major users, while ground-based television transmissions make use of the 470 to 860 MHz band. Over this range of frequencies site remoteness, terrain shielding and local shielding with "fencing" can be effective at reducing the impact on observations. The same applies to the range 250 to 470 MHz where military and civilian mobile communications are active. Besides passive, mechanical methods for interference rejection, active methods will probably also need to be employed. Analog methods of partial interference cancellation by anti-phase injection at the feeds may be an option. A digital concept calls for using a portion of the correlator capacity to do real-time delay and fringe tracking of a number of the most powerful interfering sources in the observing band so that they can be accurately subtracted from the on-source data. Extensive monitoring of the interference environment at the WSRT site in the northern Netherlands has led to the time-averaged spectrum shown in Fig. 21. Even with the substantial (although well controlled) level of spectrum use, receiver noise limited integrations of 10's of hours duration in the band from 305 to 385 MHz have been possible with the WSRT during night time observing with only a few percent data loss to RFI. Site testing and monitoring at other locations with calibrated receiver equipment is highly desirable to quantify the range of conditions we can expect from different sites.
*** Action Item: Viallefond will report on the "Interference Fence" which has been designed for Nancay.
*** Action Item: Wu is exploring acquisition of a receiver and antenna system to conduct further site testing.
Besides the issue of interference are a number of other important considerations effecting the choice of site. Large stationary spherical reflectors are dependent on the availability of suitable terrain to have a chance of being a practical concept. All of the large reflector concepts (including dishes down to about 25 m) will have substantially higher cost if they would need to survive snow and ice conditions. Site remoteness, while potentially of major benefit to the RFI environment, will very likely lead to inflated construction, maintenance and operations costs. An initial survey of potential telescope sites within China is appended in the contribution by Wu, with some further remarks in a contribution by Nan et al.
VI. Definition of research and development priorities for the next half year. "Which concrete steps should we be pursuing?"
A number of specific action items were identified which are clearly indicated within the preceeding text. Reports on action items will be made in the next meeting of the working group.
VII. Planning of next meeting and closing.
The next meeting of the working group is planned for the weekend of 20 and 21 August in the Hague, so as to allow easy attendance for IAU General Assembly attendees. Specific times, places and a meeting agenda will be distributed well in advance.
Robert Braun
NFRA
Postbus 2
7990 AA Dwingeloo
The Netherlands
email "rbraun@nfra.nl"
Phone: (31)-5219-7244