Memo 67 SKA Demonstrators, 2005 Assessment by the Engineering Working Group

P. J. Hall (EWG Chair), On Behalf of the Group 01/12/05

www.skatelescope.org/pages/page_memos.htm 2

Table of Contents

Section Page

Summary Remarks 3

Appendix 1 – Reviews of 2005 SKA Demonstrator Updates Chinese Demonstrator (FAST) 10 Canadian Demonstrator (CLAR) 13 US Demonstrators (ATA, DSAN) 16 European Demonstrator (EMBRACE) 19 South African Demonstrator (KAT) 22

Appendix 2 – Guidelines for Reviewers 29

Appendix 3 – SKA Demonstrator Updates for 2005 33 (PDF documents Appended as Submitted)

SKA Demonstrators – Summary Remarks

Peter Hall, 1 December 2005

1. Introduction

The EWG has considered the five SKA demonstrator submissions received. In four cases these submissions were brief updates of previous (2004) detailed expositions. However, an initial outline of the Karoo Array Telescope (KAT) was received from the South African SKA Consortium and the EWG devoted extra effort to a more thorough assessment of this project.

No demonstrator updates were received from Australia or India although the Australian SKA Consortium submitted a background report in the form of the statutory annual report for their Major National Research Facility program (available at http://www.atnf.csiro.au/projects/mnrf2001/Astronomy_MNRF_0405.pdf). While this report does not contain detailed schedule information for the proposed Extended New Technology Demonstrator (xNTD) project, it does indicate that the SKA Molonglo Prototype (SKAMP) program is still largely on track. However, additional enquiries indicate that SKAMP development beyond mid-2007 still depends on as yet uncertain funding. It should also be mentioned that KAT and xNTD are very similar instruments and much of the EWG commentary is likely to apply to both projects.

As in 2004, detailed individual demonstrator reviews have been produced by individual EWG members and moderated during subsequent e-mail discussions. Individual reviews, together with original review guidelines, are contained in Appendices 1 and 2 of this document while, for reference, the 2005 demonstrator submissions are included as Appendix 3.

2. Basis of Assessment

Following the 2005 reporting format, the EWG has also reviewed the numerical scores awarded to various projects. While a detailed explanation of the scoring is contained in the 2004 report the essence of the evaluation relates to the ability of the demonstrator project, as defined in the written submission, to reach a critical milestone by the end of 2008. That milestone is taken as the point at which the SKA project might, with good judgement, initiate large-scale technology production, at least on a scale sufficient for a 5-10% Phase 1 SKA.

Table 1 summarizes the 2005 scoring. Grey shaded columns show demonstrators for which no detailed 2005 update was received. Positive changes to 2004 scores are indicated by blue cells while changes to lower scores, or to uncertain values, are indicated by red cells.

Table 1 – EWG 2005 Assessment of SKA Demonstrators FAST LAR SKAMP PPD USSKA SKADS/ KAT (2004) (2004) EMBRACE Frequency range < 5 GHz < 1.8 < 1.4 GHz < 5 0.1 – 25 < 1.4 GHz 0.7-1.75 of demonstrator GHz GHz GHz GHz Demonstration of 4 3 2-3 3 (d) 3 3-4 (e) 5 pivotal and/or high-risk technology Demonstration of 4 2 1-2 3-4 2 5 4 cost reduction strategies Demonstration of 3 1 1 2 3 3-4 3 realistic risk management for concept or system Likelihood of 0, 4 (a) 0, 2-3 (b) 0-1(c) 3-4 ? 0 0(f) completion by end of 2008 Likelihood of 4 4 4 4 3 4 4 substantial added knowledge by end of 2006 Realism of 2, 4(a) 3 3 4 ? 4 3(f) project plan (milestones & timescale vs budget and manpower) Definition of 2 5 3 4 ? 5 2 appropriate milestones (suitable for ISPO monitoring) Security of ?, 4(a) ? 3-4 4 ? 5 2-3(g) funding for the project as defined Quality of 3 3 NR NR NR 3 NA responses to EWG 2004 comments

Key:

0 Very poor; or very low 1 Poor; or low 2 Average 3 Good; or high 4 Very good; or very high 5 Outstanding; exceptionally high ? Indicates critical funding outcome still unknown NR = no response received; NA = not applicable

Notes:

a. Higher score refers 30 m ‘demonstrator of demonstrator’ b. Higher score refers to structure unit ‘demonstrator of demonstrator’ c. See Section 1 comments regarding post-2007 funding d. Score is for single FOV feed; would be higher with FPA e. Score for single polarization array, assuming limited-area dual polarization demonstration elsewhere in SKADS f. Based on amended goal of early 2009 “first light” g. Higher score will be given if current funding negotiations are successful

3. Main Outcomes for 2005

There have been several major demonstrator programmatic developments over the past year. These include:

• Funding of the European SKA Design Study (SKADS) program, allowing the Aperture Array concept to be taken to the point of astronomical demonstration via the EMBRACE project. A major part of EMBRACE will focus on manufacturing issues and cost reduction techniques.

• Extended delay in funding the US Technical Development Project (TDP), a project aimed at demonstrating production readiness of the small dishes and associated components required for the Large N – Small D (LN-SD) concept. With the TDP funding uncertain, demonstration of the SKA LN-SD approach will be done largely via the Allen Telescope Array (ATA) and Deep Space Array Network (DSAN) projects; ratings in Table 1 reflect this change.

• Appearance of project plans and first-stage funding for astronomically-capable demonstrators based on the Small Dish – Focal Plane Array (SD-FPA) concept. These instruments are the Karoo Array Telescope (KAT) in South Africa and the Extended New Technology Demonstrator (xNTD) in Australia. Both instruments will complement the ATA and LOFAR as science pathfinders, as well as being 6

key SKA engineering demonstrators. (A funding application for APERTIF, a Westerbork FPA demonstrator project is currently pending).

There have been major technical achievements in the last twelve months, including substantial progress towards a 42-dish intermediate stage of the ATA, preparatory electromagnetic design and analysis work for both phased aperture array and focal plane array instruments, and digital signal processing development for stage-1 FPA demonstrators likely to be complete in 2006, Quarter 1. Notwithstanding this progress, the EWG again stresses the need to keep up the pace of phased array demonstrations as this technology remains central to the SKA, figuring prominently in the new Reference Design. For example, in the next year it will be essential to have at least initial astronomical results from the early stages of instruments like KAT and xNTD.

Latest timelines for SKA demonstrators, derived from submitted reports, are summarized in Figure 1. It is clear that completion of all major demonstrators will now extend beyond the original completion target of 2008, Q4. In future EWG evaluations the completion milestone will be taken as 2009, Q4, an apparently tractable change bearing in mind that the newly-published SKA project plan has SKA Phase 1 construction starting in 2011. However, the slippage will exacerbate an already worrying problem for the international project. Phase 1 preparatory work needs to start in 2007 and delayed regional demonstrators will restrict the number of experienced engineers available to the international project in its early stages. It is obviously in the interests of the international SKA for the regional demonstrators to be completed as early as possible and for engineering staff to be made available for SKA Phase 1 soon.

Two issues concerning developments in the USA and Europe should also be mentioned. First, the EWG appreciates the huge intellectual contribution made by US engineers to the international SKA engineering efforts and the Group underlines the very high quality of engineering science and practice flowing from projects such as the ATA and DSAN. However, both of these projects, while aligned closely with SKA, have other primary aims. Furthermore, the long-term funding situation for both – and for the SKA-specific TDP – is unclear at present. Taken together, these factors account for the EWG’s uncertainty in rating key aspects of the 2006 US demonstrator plan.

In the case of Europe, the EWG appreciates the importance of the Aperture Array demonstration. However, with the SKA Reference Design now incorporating the Small Dish – Focal Plane Array concept and indeed the SD-FPA concept being a major part of SKA Phase 1, the EWG sees an imperative for European engineers to link more closely the AA and SD-FPA demonstration programs. While the AA may figure more prominently in a slightly longer-term evolution model for the SKA, it is clearly in the interests of the international project if the substantial European expertise in phased array construction is also available to SKA Phase 1.

The EWG notes the international collaboration emerging in the area of SD-FPA concept exposition and demonstration. The Group urges proponents to complete the whitepaper now in draft form (see http://www.jb.man.ac.uk/ska/SD-FPA/index.html) and to 7 strengthen engineering ties between projects such as KAT, xNTD and APERTIF. The EWG looks forward to seeing more detailed specifications for these instruments following initial design and measurement work over the coming months.

4. Concept Risk Assessment

As part of the 2004 demonstrator review the EWG also rated the risk involved in developing various concepts for SKA application and, in particular, for possible hybrid SKAs. For this early evaluation the assigned risk rating was “high”, “medium” or “low”. The risk was split into two parts: that associated with reaching the cost goals outlined in the whitepapers, and that attached to reaching stated SKA performance goals. For completeness, the EWG in 2005 considered the small dish – focal plane array SKA concept. The extended tabular summary of risk ratings is shown below.

Concept Frequency Cost Risk Performance Risk KARST < 2 GHz High Medium-High(2) LAR < 22 GHz High High (2,3) CR < 2 GHz Medium Medium (2) PPD < 5 GHz Low (1) Medium (2) LNSD > 1 GHz Low (1) Low AA < 2 GHz High Medium SD-FPA < 3 GHz Medium-High Medium(2)

Notes:

1. Computing cost uncertainty may increase rating 2. FPA unproven (no capable technical or astronomy prototypes) 3. Cooled FPA yet to be developed 8

Figure 1 – Timelines for SKA Demonstrators 9

Appendix 1 – EWG Reviews of 2005 SKA Demonstrator Updates 10

EWG Demonstrator Review - Chinese FAST Demonstrator

S. Ananthakrishnan (with comments by S. Weinreb) 21 November 2005

The present review is an update on the last year's review and hence will carry many of the points in the previous comments and state how they have been answered.

The main purpose of the project remains unchanged: to produce an SKA technology demonstrator as well as a stand alone telescope (FAST) at the low frequency end of up to 2000 MHz, with possible extension to 5 GHz or even 8 GHz. It is clearly not meant for use above these frequency limits. FAST is estimated to cost about 60 million Euro.

In engineering terms, it is a solid and innovative mechanical engineering design on a large reflector concept which should produce a single-dish telescope (FAST) and which will demonstrate facets of the KARST SKA concept, excluding synthesising of apertures, cross-correlator backends, etc. In view of its modern design, the KARST demonstrator (FAST), which is modelled after the Arecibo telescope, will have a larger collecting area and much better declination coverage: from -35 deg. to +90 deg.

The scale of the FAST project has two components. While the reflector overall dimension is 500 m in diameter, the actual area used at any one time is 200-300 m. However, the sophisticated engineering involved in the feed design and the adaptive reflector tiles entails considerable labour and effort. If one combines the two, the scale appears to be ~5% of SKA project effort. The Chinese astronomers had planned to make a 1:10 scale model of FAST by Q3 2005. This seems to have been delayed slightly to Q2 2006.

It is clear from the current report that Chinese engineers have been experimenting with new ideas for the active main reflector, including the active cable-mesh reflector in the place of a solid panel actuator design. While the latter had been approved as a feasible scheme by their engineers, it is a factor of 3 more expensive than their newly-proposed cable-mesh reflector, which uses pre-stressed concepts, somewhat like one of the Indian designs, the PPD dish of 12 m. They state that the new "design replaces solid network of the elements by preloaded steel wires without welding techniques. From the experiment, we have learnt how to distribute and measure the tension forces in the pre-stressed wires, and how to control the energy-loss". By making extensive studies on the new concept of the active cable-mesh reflector (similar to the Arecibo design) and by using triangular segmentation of the surface, the proponents claim that they can “control the parabolic shape to within a few mm”. They have also optimised the stress distribution in the supporting cables and the driving cables via an iterative process. Different groups have worked on these simulations and there seems to be convergence amongst them leading to encouraging results.

The feed support system and its secondary stabilisation mechanism (Stewart Platform) is yet to be fully developed and is a fairly difficult area due to the effect of wind forces on the “platformless feed support structure”. While the weight of the structure is likely to be reduced to a few tons, compared to 10,000 tons of Arecibo and the tracking is planned to be done using opto-mechatronics (including laser ranging), it requires a great deal of attention and the proponents are well aware of this. Their cable- car trolley and platform is quite ingenious, but the issue of survival wind load analysis is not addressed. This area needs particular attention.

In the previous review it was stated that “serious efforts have been made by the proposers to study the working of the focal plane assembly and much experience has been gained by them. The FoV is also being increased from a few minutes of arc to half a degree by adopting the Aperture Array Technology and multi beam concepts are being explored”. The proponents have made significant progress on these studies.

The term “Demonstrator of the Demonstrator” (DoD) was coined in the previous report to describe their 1/10 scale model of FAST being built in the Miyun Observatory. The DoD is now called MyFAST. It was pointed out earlier that “although the DoD will clarify a number of concepts involved in the FAST project, it will not be sufficient from the EWG point of view for scaling the DoD to the level of SKA, or to be able to take any final decision regarding the use of such reflectors for the SKA configuration, unless dramatic progress is made on the FAST project between the years 2006 and 2008”. While it is claimed that “the first light of the scaled model is expected by the spring of 2006”, the problem of time scale remains. As per the milestone table given, the full scale demonstrator FAST, with completion in 2014 (and evaluation probably taking to 2016) is too late for SKA concept selection.

Further, the limitation of the baselines from 30 fixed 200 m antennas does not match the current multiple science requirements of SKA. However, FAST could serve as a powerful instrument which augments SKA within the angular and frequency range in which it operates.

Summary:

1. Referring to the previous report, the main strengths of the FAST project are that of (i) large reflector size giving high sensitivity and hence easy total flux measurements; (ii) low RFI in the KARST region and (iii) low labour costs in China. These strengths remain.

2. The weaknesses mentioned in the earlier reviews remain too, namely that KARST concept has too few antennas for the SKA configuration; frequency coverage may be limited to <5 GHz, etc. Further, while the building of a FAST prototype will be a great achievement for NAOC, it is clear from the milestones that it will be very difficult to speed up the project to less than 6 years from its start.

However, it is to the credit of the proponents that they are on much firmer ground in this report than the previous one, the experience gained is of high quality and they exude a great deal of confidence as far as FAST realisation is concerned.

3. Again, as pointed out earlier, the software issues related to the FAST and later KARST have yet to be worked out (even if they are not major issues); the detailed costing remains to be done; and the software industry involvement has not yet been spelled out in any detail.

4. While the completion of FAST by the year 2014 is too late for considering it as a SKA antenna design, it will be a great achievement for China, will strengthen Chinese radio astronomy enormously, and will act as a supporting instrument to SKA.

Table 1 – FAST Demonstrator: EWG Assessment (on the basis of proposal as submitted, and assuming operation at frequencies < 5 GHz) ------Criterion Rating Demonstration of pivotal and/or high-risk technology 4 Demonstration of cost reduction strategies 4 Demonstration of realistic risk management for concept or system 3 Likelihood of completion by end of 2008 0,4(1) Likelihood of substantial added knowledge by end of 2006 4 Realism of project plan (milestones & timescale vs budget and 2,4(1) manpower) Definition of appropriate milestones (suitable for ISPO monitoring) 2 Security of funding for the project as defined ?,4(1) Quality of responses to IEMT Oct 2003 supplementary questions 3 ------

Notes: 1. Higher score refers to 30 m MyFAST "demonstrator of demonstrator"

Key: 0 Very poor; or very low 1 Poor; or low 2 Average 3 Good; or high 4 Very good; or very high 5 Outstanding; exceptionally high ? Indicates critical funding outcome still unknown NR = no response received ------13

EWG Demonstrator Review – Canadian LAR Demonstrator Project

A. R. Thompson and J. W. Dreher September 30 2005

The LAR project is currently aimed at providing the technology development and design for a 300-350 m LAR demonstrator. Construction of the demonstrator is not included in the current four-year plan which extends to 2009. The LAR comprises five subsystems: (1) the tethered aerostat, (2) the confluence point system at the reflector focus, (3) the phased-array, prime-focus feed, (4) the reflector, and (5) the beamformer. Progress in the first four of these areas is briefly outlined below.

Aerostat A 1/3 scale1 prototype aerostat has been under test for about two years. A three-tether control system has been implemented using position feedback from differential GPS measurements. Motions have been reduced to 2 cm (horizontal) and 5 cm (vertical). A further three tethers are to be added.

Confluence Point Mechanism The aerostat leash and the tethers that control the position of the focal equipment are connected to a space frame from which the feed array is supported. A Mock-up of the system has been built at Laval University and a scaled prototype will be used for testing with the 1/3 scale aerostat.

Focal Plane Phased Array The plan is to develop first an engineering demonstrator of a phased-array feed, then develop uncooled monolithic LNAs and optical outputs for the proposed Vivaldi elements, followed by testing of the feed system on existing antennas including a 26 m telescope at DRAO. The target time for this project is two years.

Main Reflector The reflector panels are composed of triangular plane surfaces (facets), several of which are mounted as a single reflector unit with shape closely approximating a parabolic surface of focal ratio ~2.5. One such panel unit has been built and tested at DRAO. The actuator on which the panel is mounted, and which controls its position, has been demonstrated. It is believed that actuators with throws of ~15 m can be built for relatively low cost. A cost of $400 US per square meter is estimated for the reflector surface, including the supporting actuators and a metrology system for position measurement. Secondary actuators for the individual triangular facets will be required in the later stages of development.

1 The 1/3 linear-scale, i.e. 1/27 of the lifting force of a full-size aerostat.

Answers to Questions2

(1) The main purpose of the LAR project is to demonstrate the LAR technology as a possible concept for the SKA. A single LAR telescope would also be a significant instrument for science. The purpose of the current activity is the development of the required technologies leading to a detailed design for a LAR telescope, and does not include demonstrator construction.

(2) The project aims at demonstrating the full technology of a LAR telescope. Current activity is aimed at design of major subsystems and critical components.

(3) There is a four-year (05/06 to 08/09) development plan, in which the manpower level of full-time employees amounts to a total of 55.1 man-years. This does not include supplementary effort by university staff and students. The proposed 4-year budget is $9.8M (Canadian).

(5) The principal institution involved is the Herzberg Institute for Astrophysics. Others are the University of British Columbia, McGill University, and Laval University.

(7) The technology development and detailed design are to be completed at the end of the present four-year plan in 2009.

(8) The completion of the tasks in (7) within the time outlined seems possible if no major unforeseen problems arise.

(9) It is likely that some design progress will be available in 2006.

(13) The demonstrator that is planned to result from the current design activity would represent a single station of the SKA concept. Data transfer from the feed system to the ground would be involved, but connectivity and data transfer from stations to a distant correlator would presumably not be included.

(14) The budget plan for the current four-year period [see (3) above] includes 30% contingency. Details of risk assessment are not given.

(15) The current emphasis is largely on hardware, with computers involved in such things as winch control and development of beam-forming strategy.

(16) The current plan includes development and design of the Confluence Point Mechanism which is supported by the aerostat and controls the orientation of the phased- array feed plate. Similarly, design of the reflector surface and positioning mechanisms are included.

2 See list of questions in “SKA Demonstrator Evaluation by the IEMT – Guidelines for Reviewers”, Peter Hall, 23 April 2004 (Appendix 2 of this EWG report)

(17) Development of SKA infrastructure is not an important part of the current plan.

(18) As a demonstrator, a single 300-350 m LAR telescope will have a collecting area of about 5% of a square kilometer and should be capable of important scientific research. However, construction of such a demonstrator will not begin until after 2008.

(19) The phased array feed system will use Vivaldi elements, which are also planned to be used in several other concept demonstrators. The beamforming, and signal processing including amplification, digitization, and transmission on optical fibers, will have some commonality with other concepts.

(21) AMEC Dynamic Structure Limited and Bosch-Rexroth Canada are involved in work on the reflector.

Concluding Comments The work follows the plan outlined in 2004 document Project Plans for the Development of the Large Adaptive Reflector – Development-to-Demonstrator, dated April 28, 2004. No important changes appear to have been made and progress is about as expected. As the 2005 report states, the development of the focal-plane phased-array feed system is a key enabling technology and considerable attention is being given to the work in this area. We stress that the problem of maintaining a low system temperature with un-cooled amplifiers is critical and is common to several other SKA concepts such as those using small dishes with focal-plane phased arrays.

One important attribute of the new Canadian PHAD phased array demonstrator project is that, despite its modest bandwidth and processing goals, it is the only current dual polarization phased array feed demonstrator. The EWG suggests that the PHAD experimental program could capitalize on this and could usefully be coordinated with single polarization demonstrator programs such as NTD and EMBRACE. 16

EWG Demonstrator Review — SKA Developments in the US

Bruce Veidt (DRAO) and Ralph Spencer (JBO) 21 November 2005

1. Summary of the US report The US SKA Committee has submitted a short report3 summarizing three major developments in the US. The first is significant progress in the Allen Telescope Array (ATA) with 30 antennas in place and working towards a 42 element complement by the end of the year. In the report they present an initial image of M31 obtained with a 4- antenna subset of the array using fpga processing. Good progress is being made, and they can emplace roughly one antenna element per day. The second development is continuing support of the Deep Space Network (DSN) array by NASA, envisaged as 400 x 12-m antennas at 3 locations. A small prototype array of 12-m antennas will be built at Goldstone by 2009. They have acquired two 6-metre hydro-formed antennas and are in the process of obtaining a paneled 12-m dish. The third development is the failure to obtain funding to the Technology Development Project (TDP). As this was to be a major source of development funds, this is a significant blow to the US effort. However there is continuing lower level NSF funded work on antenna and mount concepts, broadband feeds, RFI mitigation and array configuration studies.

2. Changes to 2004 Review In light of these changes, we make the following adjustments to our 2004 review4 of the US SKA effort. Many of the questions that we answered simply do not apply anymore since the TDP has not been funded. Now US development work depends upon the ATA and the DSN array until new funding is obtained. The problem with this is that these projects have different goals, both to each other and to those of SKA. All of our responses up to Question 9 in the 2004 review no longer apply. Question 9 (substantial knowledge by 2006): the likelihood that "the project will add substantial knowledge prior to the 2006 external review" is poor. Question 10 (outcomes by 2008): the US SKA effort is unlikely to "deliver its major outcomes in time for selection of SKA technologies in 2008". Question 11 and 12 refer to 2003 report and are not relevant. Question 13 (scalability and cost reduction): these are not being addressed in the currently funded programme Question 14 (risk assessment and management): no information given though the ATA and DSN projects are likely to be well managed in these aspects. Question 15(computing issues): many of the software issues being addressed by the

3http://www.skatelescope.org/documents/Engineering/US_update05.pdf 4http://www.skatelescope.org/PDF/review_us_dem.pdf 17

ATA will have relevance to LNSD. Question 16 (mechanical engineering development): the DSN effort is addressing many of these issues (hydro-forming, coolers, etc.). Question 17 (infrastructure): ATA and DSN normal operations cover these aspects, though the application to SKA is unclear since site conditions may be very different. Question 18 (basic astronomy): the ATA has begun basic astronomical measurements with a small number of elements (still not large-n) and should have a large number on- line by 2008. This will be important for demonstrating large-n imaging capabilities. Question 19 (applicability to other concepts): the US large-n small-diameter concept has some compatibility with phased-array fed telescopes such as KAT and xNTD: they could both share the same infrastructure (small dishes) and compliment each other by extending the frequency range of the telescope. Question 20 (links with other groups): Links with other international groups may actually be eroding. For example, there was no US presence at the 2005 FPA Workshop in Dwingeloo, however we note that there is no mention of direct NSF SKA oriented funding on FPAs in the report.

3. Assessment Table Due to the loss of TDP funding, the assessment table has changed considerably. Clearly earlier timelines and plans have been severely disrupted and this is reflected by the question marks in the table.

Table 1: US SKA Assessment

Criterion Rating

Demonstration of pivotal and/or high-risk 3 technology

Demonstration of cost reduction strategies 2

Demonstration of realistic risk management for 3 concept or system

Likelihood of completion by end of 2008 ?

Likelihood of substantial added knowledge by 3 mid-2007

Realism of project plan (milestones & ? timescale vs budget and manpower)

Definition of appropriate milestones (suitable ? for ISPO monitoring)

Security of funding for the project as defined ?

Key

1 Very poor or very low

2 Poor or low

3 Average

4 Good or high

5 Outstanding or exceptionally high

? Highly uncertain

EWG Demonstrator Review – European EMBRACE Demonstrator

Peter Hall 21 November 2005

The 2004 EWG review of the EMBRACE demonstrator project underlined in detail the importance of this ambitious project as a demonstrator of key SKA technology, namely dense, broadband, phased arrays. The project aim is to produce a 2-FOV, astronomically functional, instrument by 2009. Most significantly, the project is aimed squarely at demonstrating a significant reduction in the manufacturing cost of phased array technology and at assessing the viability of the aperture array concept for the SKA. The EMBRACE project deliverable has been altered slightly since 2004 to be a single 300 m2 patch (at Westerbork), abandoning the previous small patches at other European locations. This change is consistent with demonstration priorities suggested by the 2004 EWG report.

In the last year the European SKADS project, of which EMBRACE is a major part, has been successfully funded, albeit with some delay. While the delay in funding was close to a year, the current EMBRACE project plan still shows a completed demonstrator by mid-2009, with significant information about critical technology elements being available in time for a 2007 SKA technology review. It is clear from the exposition of possible manufacturing and assembly methods that the EMBRACE team has been able to sustain at least some parts of the initial design work during the funding hiatus; they are to be congratulated for this.

The 2005 update gives a much more comprehensive exposition of the EMBRACE project plan and, although one still has to go to SKADS proposals for overviews of manpower, it is clear that the project is realistically phased and has a well thought-out management structure.

The EMBRACE team, and associated SKADS partners, have responded constructively to comments and suggestions made in the 2004 EWG review. In particular, the following areas have been addressed.

(a) Polarization. While EMBRACE is still single polarization, SKADS now incorporates a dual polarization tile demonstrator capable of being sky tested. The 2005 EMBRACE exposition also shows a tile manufacturing method extendable to a dual polarization array. Following on from the EWG commentary last year, I note that building a 150 m2 patch of dual polarization station, as opposed to a 300 m2 single polarization station, could still produce a system well- capable of showing cost reduction methodologies. The advantage would be that a full demonstration of the dual polarization system (as required for SKA) would be undertaken. It has to be said though that, with the reduction in built area relative to the 2004 proposal, a dual polarization system of 150 m2 (equivalent to ~14 m 20

dish) is getting marginal in terms sensitivity for key astronomical demonstrations. Regardless of the merits of the single and dual polarization approaches for EMBRACE itself, the important point is that SKADS now incorporates a workable mechanism for supporting the dual polarization demonstration. (b) Instantaneous bandwidth. Since the bandwidth of the RF stages (including beam- formers) and the receiver chain is apparently very wide, the 40 MHz instantaneous processing bandwidth is not unduly restrictive on the effectiveness of the SKA demonstrator. It would be useful if the EMBRACE team could sketch out how bandwidth extensions might be made in the future and how wideband array performance in (e.g. beam squint, scan blindness) might be characterized within the current proposal. (c) LOFAR back-end. We commend the EMBRACE team for effective use of this powerful infrastructure. (d) Risk and contingency. The dense aperture array project is inherently high risk, as the proponents themselves understand. Within the risk envelope though, the EMBRACE team are maximizing the likelihood of success by collaborating with expert technology groups across Europe. This is a practical approach to risk mitigation within the budget of a radio science project. The spread of collaborative work extends to industry and, as the collaborations become functional, the EMBRACE team could, as a service to the global SKA community, summarize the industry links and report on their outcomes. (e) Software. The proponents have given more details for EMBRACE software plans, noting the important links to LOFAR. Capitalizing on LOFAR experience is sensible and future project updates might expand on EMBRACE software subprojects as they become more defined.

In EWG discussions of the 2005 EMBRACE update, a few additional points arose. These are mainly in the vein of recording information useful to the SKA project as a whole. First, EWG members are keen to see updated cost and performance projections for LNAs, and to hear more of plans for receiver integration. Second, the new emphasis on RF-on-fibre is interesting, and more information about the EMBRACE team’s technical approach to this aspect would be of great interest. Finally, concern about the susceptibility of an all-electronic telescope to lightning was raised again, with a recommendation that static immunity be considered as a major design goal for EMBRACE. (In light of RFI site testing experience in South Africa, China and Australia, the potential for damage is obviously real).

Based on the change in funding status and the evolution of SKADS directions, the EWG updates the EMBRACE demonstrator score table as follows.

Table 1 – EMBRACE: EWG 2005 Assessment

Criterion Rating

Demonstration of pivotal and/or high-risk technology 3-4(a) Demonstration of cost reduction strategies 5 Demonstration of realistic risk management for concept or system 3-4 Likelihood of completion by end of 2008 0(b) Likelihood of substantial added knowledge by end of 2007 4 Realism of project plan (milestones & timescale vs budget and 4 manpower) Definition of appropriate milestones (suitable for ISPO monitoring) 5 Security of funding for the project as defined 5 Quality of responses to IEMT Oct 2003 supplementary questions N/A

Notes:

a. Score is for single polarization instrument, assuming limited-area dual polarization demonstration elsewhere in SKADS program b. Commissioning and assessment of EMBRACE is currently scheduled to extend to mid-2009

Key:

0 Very poor; or very low 1 Poor; or low 2 Average 3 Good; or high 4 Very good; or very high 5 Outstanding; exceptionally high

EWG Demonstrator Review – Karoo Array Telescope (KAT)

Peter Hall and Sander Weinreb 21 November 2005

1. Introduction

The Karoo Array Telescope (KAT) is based on small to medium (~15 m) diameter dishes equipped with focal plane arrays (FPAs). It is currently a partly-funded project and is scheduled for completion in 2009. The KAT will be built in the Northern Cape Province of the Republic of South Africa (RSA). A current representative, or reference, design has the instrument consisting of 20 x 15 m dishes, each equipped with a 10 x 10 dual polarization FPA at the prime focus. The target operating band is 0.7 to 1.75 GHz, with an almost frequency-independent field of view (FOV) of 50 deg2. No details of the array configuration are given in the demonstrator plan but enquiries have established that a centrally condensed array, with only a few outlying elements on modest baselines (<10 km), is being considered. The current cost of the project is estimated at around USD 50 million, and about USD 15 million has now been secured to cover the initial development and demonstration phases. Negotiations with the RSA Government for the remaining funds are at an advanced stage.

The EWG commends the KAT team for initiating a project which will be an important demonstrator of the small dish – focal plane array (SD-FPA) SKA concept. The Group notes that its review would have been helped by a concise statement of the KAT’s specifications or goals, including projections for key parameters such as receiver and system temperatures, and antenna efficiencies. A statement of what the RSA group themselves see the KAT demonstrating in the SKA context would also have been useful. For example: is there potential for extending the KAT upper frequency limit?; do they see a KAT-like concept as being part of a hybrid SKA solution?; or do they see the most important SKA science goals being achieved with a ~1.8 GHz instrument? We anticipate that the KAT team will contribute their thinking in these areas to the forthcoming SD- FPA whitepaper.

The EWG feels that the KAT project timescale is very ambitious and probably too optimistic. Availability of complete project funding would help enormously in engaging industry and harnessing their expertise, especially during the lead-up to construction. However, there are formidable radio science and engineering issues still to be resolved before a large-scale construction effort can begin. The EWG is conscious of the lead times being encountered by other radio astronomy groups in developing new technologies. Crucially, the RSA group has recognized its relative inexperience in constructing and commissioning an aperture synthesis telescope and has sensibly established strong international links to accelerate the KAT as much as possible. The EWG sees these links as being vital if the project is to be delivered on a timescale close to that forecast by the KAT team. It should of course be added that such collaboration is 23 also likely to be essential to the timely delivery of most other SKA demonstrators, especially with funding delays for major programs now being ubiquitous.

After consideration of the draft EWG report the KAT team compiled a response (included with the initial KAT demonstrator outline in Appendix 2). In the response the major KAT milestone is now “first light” in early 2009. This revision has been taken into account in the EWG’s rating of the project plan (Table 1).

The EWG understands the importance of the KAT to science and engineering education in the RSA and very strongly commends the KAT team for their well-directed efforts in this area.

2. Answers to Pro-Forma Review Questions (Ref. SKA Memo 55)

1. The KAT is designed to be both an SKA engineering demonstrator and a science pathfinder. While the demonstrator submission focuses on technology, other expositions of the project have stressed the need for the instrument to do world- class science. From the SKA engineering viewpoint, the instrument will be a major contributor to international programs. We assess the science and engineering motivations as being equally strong for the project.

2. The KAT is a complete demonstrator, being a functional instrument based on the SD-FPA concept for mid-band SKA. Furthermore, the instrument will be built at a remote site, allowing aspects of SKA infrastructure engineering to be demonstrated.

3. The project is currently costed at USD 51 million, will have a team of about forty FTEs per year (at full strength), and is scheduled for completion in early 2009. Substantial fractions of the project will be completed under sub-contracts but the submitted plan does not indicate the likely nature of the contractors (e.g. industry, universities, other international radio astronomy institutions, etc.).

4. The project has been recognized by the South African Government and initial funding of USD 15 million has been received. A Ministerial request from the Science and Technology Ministry to Treasury for a large portion of the remaining USD 35 million has been submitted, and additional negotiations are currently underway to obtain further funds via the Trade and Industry Ministry. Enquiries to the RSA SKA Consortium have determined that the first USD 15 million is expected to cover all project costs up to and including the pre-production model; that is, to the point of installation of the first dish in the Karoo.

5. The demonstrator plan does not identify in detail the South African participants but it is clear that bodies such as universities, private consultancies and government agencies are involved. The project team apparently has independent standing and no single external body has the major role. 24

6. The project has an identified manager and established project offices in Johannesburg and Cape Town. Management roles are split largely across functional areas. While not discussed in the demonstrator submission, other KAT presentations show a moderately complex management structure. The point is often made that the geographical diversity and the need to coordinate across many contributing bodies has led to management challenges. On the system engineering front, the KAT team is using a formal design approach not previously used in astronomy. This approach was chosen partly to alleviate the effects of a dispersed design team and is a methodology of interest to the SKA community.

7. The submitted project plan has only a small number of milestones involving key system elements. From an EWG and ISPO perspective a larger number of intermediate milestones are needed for effective project tracking. A production ready version of a single telescope is scheduled to be completed by mid-2008. Expansion to the full KAT is then the scheduled to occur over the following six months. It is clear that to maintain production schedules on this timescale, a very large on-site effort will be required (see Introduction).

8. The KAT is an ambitious project with a tight timescale. However with a ~USD 50 million budget request and over 40 FTEs at full strength, the project is potentially well-resourced. Successful delivery of the project will be critically dependent on international links designed to speed development of key components, such as the FPAs and digital signal processing elements. Despite the probability of a substantial industry involvement in manufacturing and deploying telescopes, the EWG is concerned about the feasibility of highly compressed timescales (see Section 3, note 2). The challenge of commissioning the instrument should not be under-estimated and experience with other telescopes shows that this phase is heavily reliant on the availability of system specialists. The EWG recommends that the KAT team consider the construction and commissioning phases in more detail and outline a proposed model in the next project update.

9. With key FPA and dish milestones set at mid-2007, the project will add substantial knowledge about the SD-FPA concept in time for the 2007 external review of SKA engineering progress.

10. On current projections the KAT project will deliver major outcomes in time for a 2009 SKA technology selection.

11. This is a new SKA demonstrator and there are no previous IEMT or EWG questions pertaining directly to the plan. The document submitted, and other KAT summaries, indicate that the RSA group is aware of the key challenges already flagged for allied SKA concepts involving dishes and focal plane arrays.

12. There are no applicable previously-raised supplementary questions. 25

13. The SD FPA concept is inherently scalable from the collecting area and FOV viewpoints, the maximum FOV being of course determined by the dish diameter. While the focus is on delivering a functional instrument within an ambitious timeframe, the KAT program, by virtue of its strong international links, will be leading-edge in SKA subsystem technology terms. Thus, scalability is inherent in much of the sub-system thinking. At the moment it appears that SKA and KAT cost reduction aims are convergent but the degree to which this is maintainable will depend on the successful demonstration of key components, such as dishes. The KAT timescale is so aggressive that there will be little opportunity to iterate and, ultimately, decisions in favor of more conservative and expensive technology may have to be made. Unless such departures occur across the entire system, the value of the KAT as an SKA engineering demonstrator will remain high.

14. The project plan has a well-developed approach to risk management. It identifies (at least conceptually) a process whereby new (and potentially disruptive) technologies can be incorporated up to the time of large-scale manufacture. Contingency options in some areas involve substituting higher cost subsystems (see above). In areas where this approach is not feasible (e.g. in the FPA), a degree of risk mitigation is achieved by building strong links with international groups undertaking leading-edge development. An additional FPA risk mitigation approach might be to investigate the viability of multi-feed arrays (one feed per pixel) in the context of the KAT science and engineering cases.

15. Software and computing are identified as major parts of the project, with an allocation of 12 to 15 FTEs being shown. Only broad reference to SD-FPA calibration and imaging issues are made in the demonstrator plan but specific details of a relevant RSA - Australian collaboration are given. Presentations by KAT personnel at recent gathering such as the Dwingeloo wide field imaging workshop show that the project group is aware that they face substantial software challenges.

16. Dish fabrication methods are clearly a large part of the KAT project. The plan identifies two options for early phase dish-FPA testing, one of which involves making a "traditionally engineered" prototype antenna in which very low-cost fabrication is not a major goal. Advancing from this point to a production prototype will necessarily involve intensive mechanical engineering reviews by manufacturing specialists. The demonstrator plan shows an understanding of the mechanical engineering cost imperatives for both the KAT and SKA. No details are given in the submission regarding the antenna and control system development plans.

17. While no details of site development are given, other KAT presentations show the proposed site to be quite isolated, with little existing infrastructure. The KAT will therefore yield valuable lessons for SKA construction and operation, including 26

lessons on array infrastructure (roads, buildings, power, communications etc) development.

18. The demonstrator will do astronomical measurements and will demonstrate, perhaps for the first time, aperture synthesis using FPA receptors. The time between having the first antenna on site in mid-2008 and the project practical completion in early 2009 is short. It is therefore hard to see much instrument characterization being possible before the KAT completion.

19. The KAT project is linked to all other SKA concepts, involving as it does dishes and phased arrays. There are very strong KAT links to the Australian xNTD and the Dutch APERTIF projects (also designed to produce SD-FPA demonstrators). The UK 2-PAD digital array tile development is also of great relevance to all three of these demonstrators. There are undoubtedly additional strong links with the Canadian LAR concept (wideband focal plane arrays) and the US LN-SD concept (small dishes and associated mechanical systems).

20. The project has particularly strong links to the Australian xNTD program. Links to European and US partners are also referred to in the demonstrator plan. As mentioned, very effective collaboration will be necessary if the KAT project milestones are to be met.

21. There appear to be growing links with industry in the antenna, FPA, systems engineering and infrastructure development areas. At this point in the project it is not clear what proportion of the total work package will be completed by industry but, with a very substantial construction phase required to be completed in a short time, great industry leverage will be needed. Early engagement with companies competent in key areas will therefore be essential.

3. Quantitative Scoring – KAT (0.7 – 1.75 GHz)

Referring to SKA Memo 55, numerical evaluation outcomes are:

0 Very poor; or very low 1 Poor; or low 2 Average 3 Good; or high 4 Very good; or very high 5 Outstanding; exceptionally high

The EWG assessment of the KAT project is tabulated below.

Table 1 – Karoo Array Telescope: EWG Assessment Criterion Rating

Demonstration of pivotal and/or high-risk technology 5 Demonstration of cost reduction strategies 4 Demonstration of realistic risk management for concept or system 3 Likelihood of completion by end of 2008 1(1) Likelihood of substantial added knowledge by mid-2007 4 Realism of project plan (milestones & timescale vs budget and 3(2) manpower) Definition of appropriate milestones (suitable for ISPO monitoring) 2 Security of funding for the project as defined 2(3) Quality of responses to IEMT Oct 2003 supplementary questions Not applicable

Table 1 - Notes 1. The EWG stresses that strong, functional, international links will be vital in ensuring the KAT’s completion on a 2009 timescale. The score in the table is based on a revised “first light” in early 2009 (Section 1). 2. The Group is concerned that some phases of the project appear quite compressed (e.g. XDMÆ ADMÆPPMÆKAT in six-month steps) and that instrument commissioning plans are not outlined. However, a revised goal of “first light” in early 2009 (Section 1) appears more realistic to the EWG. 3. A higher score will be possible if current funding negotiations with the RSA Government are successful.

4. Additional Comments and Questions

1. It would be useful to have a concise specification table and block diagrams for the KAT, including the A/T goals. 2. It would be useful to include a further level of budget detail, perhaps extracted from the detailed budget already tabled in other presentations. A breakdown into non-recurring and implementation costs would be informative. 3. The KAT appears to fall within the classification of a wide-field survey instrument and, as such, it is not surprising that signal processing and electronics account for a substantial fraction of the total cost. In addition, all the KAT beamforming is done digitally, adding to the DSP fractional cost. It would be interesting to see a more detailed breakdown of the large DSP investment, especially for readers unfamiliar with the digital beam-forming and related DSP requirements. 4. What does the ~USD 500 cost per receiver implied by Figure 2 (p4) of the submission cover? Does it include installation, testing and environmental protection? 5. An outline of the CORE approach to systems engineering would be of interest. 28

6. On p9 of the submission, is the required cost per m2 related to physical or effective area? If for physical area, it implies that a 12 m antenna should cost about USD 270,000 – comparable to estimates being received for 32 GHz antennas. An antenna operating at <4 GHz, perhaps with a mesh surface, should cost less. 7. The selected f/D range of ~0.5 appears to be a good choice based on work presented at the June 2005 FPA workshop in Dwingeloo. With this f/D what is the potential of a multi-feed array (one feed per pixel)? Could such a system form a viable risk mitigation option if the FPA approach proves technically or economically unfeasible? 8. It would be useful to ensure that Patriot Inc (of Albion, Michigan) is included in the list of suppliers providing cost estimates. 9. On p11, the cost benchmark figure of USD 3000/ m2 implies that the cost is proportional to diameter squared. The middle paragraph on p11 states that in SKA Memo1 the antenna cost is linear with antenna diameter. This is incorrect - the memo assumes an antenna cost of USD 100 x D3 in large quantities for an antenna good to 25 GHz. This is USD 337k for a 15 m antenna or USD 173k for a 12 m antenna. 10. On p11, the point in the Schultz cost estimate about no allowance for tooling costs is over-emphasized. The estimate does not include tooling costs but this is small, of the order of USD 5 million or USD 5k per antenna in a quantity of 1000. Note then that the Schultz estimate, based upon a physical design with material weights, structural analysis, and a cost breakdown per component, is USD 200k for a 16 m antenna good to 1.7 GHz and a 12 m antenna good to 34 GHz, in “quantities smaller than a thousand”. The KAT project might consider this design and cost estimate carefully. 11. On p11, the (cheap or free) 7.6 m f/D = 0.35 antenna may not be worth the labor effort to install and experiment with it. The FPA performance depends strongly on f/D and it may be better to either accelerate the production of a KAT traditionally-engineered prototype or, perhaps more simply, to buy a small antenna with the same f/D as the final antenna.

Appendix: SD-FPA Concept Risk Assessment

As explained in SKA Memo 55, the EWG rating of risk for an SKA concept is split into two parts: that associated with reaching the cost goals outlined in the concept whitepaper and that attached to reaching stated SKA performance goals. Assigned ratings are “high”, “medium” or “low”.

For the SD-FPA concept, to be outlined in the forthcoming 2005 exposition, the cost risk is currently rated as “medium to high” and the performance risk is rated as “medium”, the uncertainty in the latter coming largely from the as yet unknown calibration and computing requirements of telescopes using FPAs. 29

Appendix 2 – EWG Review Guidelines

Appendix 2

SKA Demonstrator Evaluation by the IEMT – Guidelines for Reviewers

Peter Hall, 23 April 2004

1. Introduction

This document sets out some guidelines for IEMT (now EWG) members reviewing the SKA demonstrator plans due for submission to the Director on 30 April, 2004. Reviewers should feel free to add additional commentary. I anticipate editing and combining the reports in a format similar to the IEMT’s 2003 report; the composite document would then be made public via the Director and the ISSC.

The second part of this document contains a pro-forma for scoring demonstrator projects against various criteria. This is an exercise internal to the IEMT although the Committee may, after discussion, choose to make the results available more widely.

2. Questions for Reviewers

1. What is the main purpose of the project? SKA technology demonstrator, SKA pathfinder science instrument, or other? Some projects will have several purposes and it would be appropriate to give the reviewer’s feeling on primary and secondary motivations. 2. What does the project demonstrate in engineering terms? A whole concept, major subsystem(s), critical component(s), infrastructure, other? 3. What is the approximate scale and duration of the project? Financial and manpower summaries, together with an overall plan, should have been provided by authors. 4. How much of the project is actually funded? If there are outstanding funding applications, when will the results be known? 5. Which institutions are key players? 6. How is the project co-ordinated? Is the management structure straightforward or complex? 7. What are the major milestones? Are the milestones tightly defined? Will it be possible for the IEMT and ISPO to track progress over the coming years? 8. Are the overall goals and milestones plausible given the resources available? 9. Is it likely that the project will add substantial knowledge (e.g. about a concept) prior to the 2006 external review of SKA engineering progress? 10. Is it likely that the project will deliver its major outcomes in time for selection of SKA technologies in 2008? 11. How well does the project address the key technology questions raised by the IEMT in its 2003 report? 31

12. How well answered are the supplementary questions raised by the IEMT in its 2003 report? Are there major outstanding issues? 13. How well does the demonstrator establish scalability to SKA (including areas such as connectivity and data transfer, in the case of concept demonstrators)? Is there a clear plan in place to demonstrate the cost reductions needed to make SKA feasible? 14. Does the project outline a systematic approach to risk assessment and risk management? What contingency options are discussed and what effect does this have on the concept or system in the SKA context? 15. Is there any software and/or computing component to the project? If so, what are the main issues addressed? 16. Does the plan incorporate activities for further development and review of relevant mechanical engineering and mechatronics? 17. Does the project include activities which demonstrate aspects of SKA infrastructure development? If so, what areas are covered (e.g. remote area construction, passive climate conditioning, remote power provision, wideband data links to remote sites, …..). 18. Will the demonstrator do at least basic astronomical measurements and interferometry (perhaps in conjunction with an existing telescope)? Are there plans to characterize important performance aspects prior to 2008? (Such aspects might include sensitivity, bandwidth, polarization purity, pointing, beamshape stability, …..). 19. Does the project have applicability to other SKA concepts or research efforts? 20. Are there identified links with other international SKA groups, or other major astronomy projects? 21. Are there identified links with industry participants? If so, what is the form of the industry involvement?

3. Quantitative Scoring

In the table below, rate the project on the following numerical scale:

0 Very poor; or very low 1 Poor; or low 2 Average 3 Good; or high 4 Very good; or very high 5 Outstanding; exceptionally high

IEMT Evaluation Table for Demonstrator Project

Project:

Criterion Rating

Demonstration of pivotal and/or high-risk technology Demonstration of cost reduction strategies Demonstration of realistic risk management for concept or system Likelihood of completion by end of 2008 Likelihood of substantial added knowledge by end of 2006 Realism of project plan (milestones & timescale vs budget and manpower) Definition of appropriate milestones (suitable for ISPO monitoring) Security of funding for the project as defined Quality of responses to IEMT Oct 2003 supplementary questions

Appendix 3 – SKA Demonstrator Updates for 2005 Demonstrator Review for KARST

FAST Group

National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China Email: [email protected], or [email protected] Fax: +86 10 64852055

1. Introduction

In 1993, a large radio telescope (LT, now referred as the SKA) was proposed by astronomers from 10 countries at the 24th General Assembly of URSI. The SKA would be a telescope array with a total effective collecting area of about one square kilometer. There are various concepts, worldwide, for realizing the SKA project. Extensive efforts have been made, e.g, by project teams in The Netherlands, Australia, Canada, China, the United States, India etc., for details, see http://www.skatelescope.org. Chinese astronomers are going to build a set of large spherical reflectors by making use of the extensively existing karst landforms in south China [1], which are bowl-shaped limestone sinkholes named after Karst, a Yugoslavian geologist. Now we refer to such efforts for the SKA as the Kilometer-square Area Radio Synthesis Telescope project, i.e., KARST [2]. The Chinese SKA, KARST, consists of about 30 individual elements, each roughly 200 m in diameter. The total cost of the KARST would be estimated of ~ 850 M Euro. As an engineering demonstrator for the KARST, a Five-hundred-meter Aperture Spherical Telescope, the FAST, is proposed, with an estimated cost of ~60 M Euro. The FAST will be over twice as large as the Arecibo radio telescope coupled with much wider sky coverage [3]. Technically, the FAST is not simply a copy of the existing Arecibo telescope but has rather a number of innovations. Firstly, the proposed active main spherical reflector [4], by deforming the illuminated area to a paraboloid of revolution in real time, enables the realization of both wide bandwidth and full polarization capability while using conventional feed design. Secondly, a cable support system for feed, which integrates optical, mechanical and electronic technologies, will effectively reduce the cost of the support structure [5]. In this document we will summarize modeling experiments for the Chinese concept.

1.1 Overview of FAST

Some basic parameters of the FAST are illustrated in figure 1-1. It will have a main spherical reflector radius of R=300 m, a total projected diameter of up to 500 m, and an effective aperture of about 300 m. Since the focal length of FAST is to be set less than R/2, a portion of the parabola (to which the illuminated area of spherical surface is deformed, shown by the dashed line in figure.1-1) lies above the sphere. The geometrical configuration and off-set illumination will enable the FAST to have larger sky coverage of ~ 60° zenith angle than that of the Arecibo telescope (fig.1-2) . The simplified feed system will continuously cover most of the frequency range between 150 and 2000 MHz, with possible capability up to 5 or even 8 GHz or more depending upon the cost. Frequency range of FAST does not meet the expanded SKA requirements in 2003 at the high frequency band which is up to 23 GHz.

Figure1-1 FAST concept and its geometrical configuration

Figure. 1-2 Comparison of sky coverage between Arecibo (upper plot) and the FAST or KARST (the lower one)

2 1.2 Overall design philosophies

The SKA (LT) project was initially proposed in 1993 with a main scientific driver of neutral hydrogen observations at high redshifts. The science case also included transient phenomena of short time scales like pulsars and variable stars. Radio astronomers expect that the new tool will expand astronomical observations from the non-thermal emissions to weak thermal emissions. These needed requires improving sensitivity by constructing huge collecting area of about one square kilometer, which is a big jump with the next generation telescope. Our project team would regard the KARST as a most sensitive low frequency telescope. The SKA could operate at a ‘limited’ bandwidth, although wider frequency range covers wider scientific fields. It may be unpractical and uneconomical to combine very low and very high frequencies into one array. The ALMA and other high frequency facilities will be better options for astronomers working at short wavelengths where the objects (e.g. stellar objects) have naturally stronger emissions. Guizhou is a very moist place, the sites do not host telescope operated more than 12 GHz. We would strongly recommend multi-beam receivers at L-band (or up to C-band) and below to increase the efficiency for surveying. Ideally the FAST would be equipped with 13-beam receiver at L-band, and less number of beams at lower bands to match the sky area beamed. And we also would like to adopt the AAT technology at the focus to enlarge the FoV. We would suggest that the KARST consists of ~ 30 unit elements with a well optimized UV-coverage [6]. The FAST itself is a way to realize the SKA by using small number of elements. The existing large radio arrays like the VLA, MERLIN, WSRT and all VLBI networks are also configured in that way. These powerful telescopes have shown how to reconstruct complicated structures of radio sources. Large number of elements may lead to signal losses in combining and transforming, design and construction complexity, possible high cost of correlation, operation and maintenance. This is also a trade off between cost and flexibility.

As a single dish telescope, the FAST will achieve the largest collecting area in the world, and it is proposed to start construction as a National Mega-science Project of China around the year 2008. Pre-research on the FAST has become a key project in the Chinese Academy of Sciences with funds of ~2 M Euro. Great progresses in active reflector, feed support and focus package have been achieved in recent years.

2. Active Main Reflector of FAST

2.1 Solid panel-actuator design

To reform the reflector, it is necessary to divide it into small elements. Each element is a small part of a spherical surface whose curvature optimized carefully for FAST is R=335 m, which is slightly different from the one of the neutral curve of the main reflector. One proposed way of dividing it is to have ~1800 hexagons on the reflector as shown in figure 2-1. Each element has three actuators to fix its position and connect it with adjacent ones, and there would be an average of one actuator per element. When FAST is tracking and forming a real-time parabola, the

3 actuators move along the radial direction with a maximum range of 67 cm [4,7,8]. Four pieces of elements at moderate altitude of the spherical cap as marked in figure 2-1 were selected for scale modeling. Figure 2-2 shows the fieldwork of the experiment while the last element was being mounted. The up-to-date field bus technology, LonWorks, was employed to serve as the actuator control network [9]. This photo also illustrates that future FAST main reflector will be a buildup of 4 layers – surface and its back structure, self–adaptive connector, actuator, and civil engineering constructions in depression. This model experiment in 2001 has basically approved the feasibility of the engineering concept of the FAST main reflector on the whole.

Figure 2-1. The FAST main reflector is divided into about 1800 hexagons.

Figure2-2. The scaled model of the FAST active reflector. The marked numbers indicate the 4 layers of the main reflector construction: 1-surface and its back structure, 2-self-adaptive connector, 3-actuator, 4-civil engineering between actuators and depressions(here, experiment mount).

4 The previous design of the FAST reflector consists of ~1800 hexagon-shaped elements whose back structures are made of a large amount of steel rods and spherical joints. Recent years, element of tensegrity back structure of FAST (see the more transparent element in figure2-3.) has been modeled and evaluated in August of 2003 as a feasible solution to future FAST reflector, reducing its total weight by a factor of 3. This design replaces solid network of the elements by preloaded steel wires without welding techniques. From the experiment, we have learnt how to distribute and measure the tension forces in the pre-stressed wires, and how to control the energy-loss.

Figure.2-3. Scaled element of FAST reflector with tensegrity back structure.

Many questions, as many as those positive results, were raised from the experiment. Investigations addressed these questions have made notable progresses. These include a new scheme of segmentation to largely reduce the species number of hexagons for mass production, kinematics study on the adaptive connector to modify its controllability, innovated tense grid design for the reflector element to lighten it and to reduce bearing force of the actuators at the tangent direction, test on reliability and life-time, layout of the civil engineering structure in depression, and etc. In this report, a completely different realization for the active main reflector of FAST is proposed based on our learning and understanding of the FAST from prophase feasibility study.

2.2 Active cable-mesh reflector As a piece of massless and ideally stretch rope is slightly pulled away from the tense position at its middle, the total length varies almost invisibly. The increment in length of 500-meter-long rope, for an Figure2- 4. The increment of a long rope. example, is about 4 mm, less than 10-5 of the total, while the central offset is 1 m as shown in figure2-4. This hints us that an antenna surface supported by cable net as Arecibo telescope could be activated in some extend without

5 extra-servos controlling the lengths of those supporting ropes. The central part of a spherical surface is very close to a rotated paraboloid when a proper focal length is chosen. The focal length of the FAST could be accurately determined as f=0.4665R according the formulae given by Li in 1959 [10], which minimizes the peak deviation of the spherical surface from the rotated paraboloid across the ~300 m of illuminated aperture. The integrated length of the dashed line (fig.1-1) indicating the deformed parabola is only 0.36 m shorter than the spherical curve, about one in a thousand of the total length. This small difference required by deforming could be easily achieved within the elasticity of ordinary wire ropes, although the mechanical analysis is much more intricate. Instead of elastic deformation, adjusting the meshes outside the illustrated area can also compensate this small difference in length. Hence, the active cable-mesh reflector [11], a new realization of FAST reflector has been proposed and investigated intensively. Ignoring the active control, the newly proposed FAST surface adopts a similar structure of Arecibo telescope. Since 2002, three groups from Tsinghua University, Tongji University and Harbin Industrial University coordinated by the NAOC have been involved into the relevant R&D. The reflector surface is supported by cable network underneath like Arecibo and deformed by adjusting the length and tension of down-tied cables as the source is tracked. The simulations show that the errors of the fit to paraboloid shape could easily be controlled within a few millimeters. Most rapid progresses have been made in proposed solution by Harbin Industrial University for the cable supported reflector of the FAST. Triangular segmentation of the surface is employed. All the sides of the triangles are small sects from the great circles on the neutral spherical cap (fig2-5). In the simulating, the lengths of the cable sects are assumed to be 11 to 12 meter, which results in 2289 nodes, therefore, 2289 driving cables. The average loads of the supporting cable network and surface element material are 5.4 ㎏/m2 and 2.0 ㎏/m2 respectively. The lengths of driving cables are estimated to be 10 – 50 meters.

Figure 2-5 Triangular segmentation of the surface for FAST

Figure2-6 Deformation of the surface with the displacements enlarged by 20. One high light of this study is the optimization for cable section to achieve required strength by less material to be used in construction. Firstly, the sections of all supporting and driving cables are assumed to be identical - 28mm in diameter, and preconceive stress of 500Mpa in supporting cable and 100Mpa in driving cable are initially postulated for late optimization. Initial stress distribution in the total network was analyzed based on assumptions above. According to the distribution, the stresses of individual sects are inspected, and some of them are replaced by new ones with different sections, holding the pulls in the network constant. This process is iterative. The calculated stresses of cables after optimization are 393.349－466.901Mpa, falling into a narrow and rational range. Six illuminated potions at different altitudes at the cap were simulated dynamically. The final fit accuracy of the supporting mesh could reach 0.001mm without considering the fit of the element surface and real manufacture error. The stress distribution is 122.000 － 601.237Mpa in supporting cable, and 151.979 － 535.019Mpa in driving cable respectively. No looseness is found in the network. Figure 2-6 shows the deformation of the surface with the displacements enlarged by 20. The results in these critical characteristics of the cable mesh network, e.g. the tension distribution in the illuminated area, diameters of cable and the power required to deform the mesh show convergence considering difference in schemes employed by different groups. In 2004,The prototyping models of a ratio 1:10 was designed for the new FAST model at Miyun radio station, including the segmentation of the cap, material and techniques of the surface, attachment between surface and adaptive cable-mesh, mechanical and electronic control of the down-tied cables etc. . The preliminary results from the simulating and experiments are encouraging. If the feasibility of adaptive cable-mesh reflector for FAST, is finally approved, it would benefit future construction of the telescope in several aspects: (1) Simplify the structure of FAST main reflector, reducing the number of layers from 4 to 2; (2) Largely cut down the number of moveable parts, e.g. bearings, screws and joints, easing machinery work and improving reliability; (3) May need not to divide the surface into solid elements with fixed curvature, relaxing appeals to fabrication accuracy and leaving space for further telescope upgrading; (4) Almost avoid the civil engineering between actuators and ground. These possible advantages mentioned above may be potentially beneficial to FAST in construction price, project time, future operating reliability and telescope maintenance.

7 3. Feed Support The FAST is "pointed" by moving the feed cabin, while the reflector surface is deformed in synchronism with the feed pointing motion [12]. One of the pivotal technology of the Chinese SKA concept is the platformless feed support system, which has the potential to reduce construction expenditure, provided the vibration level of the feed can be controlled to what required.

3.1 Cable support system A new design for the feed-support structure (fig.3-1) for the FAST/KARST has been proposed by using six suspended cables connected to mechanical servo-control systems [5]. The main idea is to move the feed along the focal locus dynamically by adjusting lengths of the suspension cables according to the position of feeds. However, the wind-excited vibrations of feed are estimated to be around 0.5 m in RMS, so a secondary feed-stabilizer has to be employed. At present, the Stewart parallel mechanism [13] is chosen as the stabilizer and under development. Compared with the Arecibo 305 m radio telescope, the total weight of the feed support system could be evidently reduced by two orders or more in such a design, probably from nearly 10,000 (if Arecibo-like support taken) to a few tens of tons. The tracking will be by means of integrated mechanical, electronic and optical technologies, i.e., optomechatronics. The whole system will mainly consist of three parts: firstly, the six cables will be driven by six sets of servo-mechanisms controlled by a central computer, so that the movement of the focus cabin along its caustic trajectory can be realized. Given the difference between the apparent and required positions, where the feed (cabin) should point, the central computer will drive each servo-mechanism to adjust the position of the feed. Secondly, a group of receivers with multi-beam feeds will be mounted on a stabilizer in the focus cabin. This is to provide a second adjustment, since the cabin driven by cables alone may not achieve the pointing accuracy required. A laser ranging system, being the third part, will be adopted to accurately measure the position of the feed in real time. The information will be fed back to the central computer for global loop control.

Figure.3-1 Cable support system without a platform Numerical simulations have been carried out for such a design by both nonlinear response analysis of the cabin-cable system with respect to random wind, and precision study of the fine-tuning stabilizer using inverse positional computation, kinematics and singularity analysis.

8 Detailed deductions have shown that the optomechatronics design is capable of satisfying the pointing accuracy required [14,15,16,17].

To demonstrate the design, a 50 m model has been built with success in Xidian University in 2002. For the FAST the focus cabin can be positioned to some tens of centimeters accuracy by cable-driven only, and then further to millimeter accuracy with a fine-tuning stabilizer, i.e., the Stewart platform. A typical Stewart platform consists of six variable-length actuators connecting a mobile plate to a base. As the lengths of the actuators change, the mobile platform is able to move in all six degrees of freedom with respect to the base.

3.2 Cable car configuration In line with the platformless conception, researchers in Tsinghua University proposed a cable-car feed support system in 1998[18]. A small cable car, to serve as the focus cabin housing the feed and receivers, is to be driven by eight cables. Two pairs of parallel supporting cables will be suspended from two pairs of opposite towers, while another four downward cables are securely fastened to four anchors which are symmetrically arranged about the main spherical reflector, to increase the stiffness of the system. The lengths of the connecting cables would be adjusted appropriately as the cabin location changes. Such a design aims to increase the stiffness of the platformless cable support structure. Positioning of the cabin would be achieved by driving the car on two cross sets of supporting cables, which is like a trolley on the cable-way in mountains. The car can move in two directions with the two sets of suspension cables as tracks. And the cabin has two rotational degrees of freedom relative to the cable car, which allows the feed to be arbitrarily pointed, irrespective of what the orientation of the cable car is. Rotation of the feed can be realized by a special mounting in the car, the axis of which should intersect the center of gravity of the cabin. Rotating the feed about its phase center is a way to gain a significant increase in scan range. Beam pointing is unaffected and the net result is that the aperture is fed in an offset manner. The only penalty incurred is the appearance of cross-polarized lobes in the plane orthogonal to the plane of scan. For circular polarization this becomes a small beam squint in that plane. Actuators are to be employed for actively controlling any oscillations of the cabin induced by the motion.

Suspension Tracking Cable Driving cable

Pre-tension

Tower

Position measuring Control Console

Figure3-2. The general assembly of the cable car feed support configuration

9 The cable-car configuration is demonstrated in figure.3-2. The pre-tension cables are introduced to adjust the stiffness of the feed support structure. The effect of the pre-tension cable for suppressing unwanted vibration can be obtained by finite element dynamic analysis with the excitations generated according to the measured wind conditions of candidate sites. Though a precision of about 0.5 m can be expected for reasonable tension level in the stabilizing cable, it is wise to have a secondary feed stabilizing device instead of increasing the stiffness of the whole structure to an unrealistic level. Trim masses can be used to balance the static load of the suspension cable for energy efficiency during operation of the telescope. The cable-car configuration separates the positioning and pointing of the feed, but uses 7 degrees of freedom to control 6 degrees of freedom of the feed. Characteristics and vibration reduction effects of the pre-tension cables have been discussed in the cable-car configuration [18,19,20]. A 1:30 scale physical model for the cable-car system has been constructed in 2001, in terms of the model similarities, the performance of the prototype can be predicted. The predicted position precision for the first support system is below 0.5 m in RMS under wind conditions of 8.2 m/s at the candidate sites. A scaled model (1:5) for the Stewart stabilizer (to stabilize its lower platform by compensating any vibration of its upper one with variable-length actuators) is constructed in 2001, for feasibility study with the available and proposed technologies. The control effects are also encouraging. A new 50-meter model of feed supporting system, combined with the cable car and Stewart stabilizer, was established in Tsinghua University in 2003, see fig.3-3 and.fig3-4. Experiments were carried out to test the controllability and performance of the mechanism under various conditions as well as the reliability and repeatability of the control laws.

Figure.3-3. Field model of cabel car Figure. 3-4. Trolley with Stewart platform

Three Laser Automatic Total Stations are employed to measure three points on the upper platform, see fig.3-5. Two Laser Trackers with positioning precision 10 ppm and sampling rate 75Hz are employed to measure the centers of the upper platform and lower platform respectively. With rotation angles of the two orientating axes and lengths of the six actuators acquired from servo feedback, positions and attitudes of all the platforms are obtained. It is worthy to emphasize in this measurement scheme that the positions of the upper platform and lower platform, being critical to the feedback control, are directly acquired with high-performance instruments.

Figure.3-5. Location and attitude data acquisition The trolley traces trajectories about a spherical surface 12m in radius at velocity of 3mm/s, The workspace is 10m×10m×5m around the center on the ground. Of a typical experiment, time history of positioning error and orientating error are shown in fig.3-6. In the experiments, winds with average velocities below 2m/s have little effect on the control. Dynamical coupling of the two platforms of the Stewart platform is prominent only when improper PID coefficients and/or control time intervals are applied. As a summary of the experiments of this series, the precision of the mechanism, namely that of the lower platform, is 0.4-0.6mm in position and 1.0-2.0×10-3 radian in orientation. According to the similarity law, the positional accuracy of 4~7 mm is predicted for FAST focus tracking.

Figure. 3-6 . Positioning error(left) and orientating error(right) of the upper and lower platforms

4. Focus package One of the great advantages with the KARST is that conventional technology can be implemented. Due to the small number of receivers in the FAST, the total cost will be low. A practical collaboration on the FAST was established between Beijing Astronomical Observatory (Now National Astronomical Observatories, the NAOC) and the University of Manchester's Jodrell Bank Observatory (JBO) by a memorandum of understanding (MoU) signed in July 1999. The joint discussion of the low noise receivers for the FAST [21] is based on the use of existing, proven technologies, to minimize the technical risk for the project. Moreover, some new technologies are going to be applied to improve the efficiency, sensitivity and stability of the system. In order to enlarge the FOV and the sky coverage of the FAST, we have set-up a project to investigate our own focus array at the FAST focus, adopting the AAT technology [22]. This combined solution for large dish will enlarge the FoV of the 300m

11 illuminated area of the FAST from ~3 arcmin to half degree and form at least 100 simultaneous beams within it. The phased array technology also enables us to form an asymmetric illumination pattern as the focus goes to the edge of the active reflector by dynamically weighting the Vivaldi-elements in the array. There are three groups working on the layout design of the AAT type feed from Tsinghua University, Beijing Astronautics University and NAOC. The array will include 1300 Vivaldi antennas on a plate of diameter 2.5 m by a very rough estimation. The electromagnetic field analysis near focus has been completed now, and the results appear close to the ones of FARADAY project.

Figure4-1 Different type of Vivaldi antenna According the requirement of FAST feed system, we have carried out theoretic study in the computer simulation on the feasibility of applying AAT technology to FAST focus. We investigate electromagnetic property of different type of Vivaldi antenna (fig.4-1), achieve the optimized design of frequency band and calculate their field pattern (fig.4-2 ), using software package of ANSOFT HFSS. In order to synthesize the desirable illumination pattern by the Vivaldi array, we use the Woodward-Lawson method to simulate the field pattern by varying the configuration of the 2-degree array and the weights of its elements. The fig.4-3a shows the pattern of the array feed at different zenith angle as the feed go its edge and the 2-D pattern (fig.4-3b) of the array feed as the zenith angle less than 30degree. We also roughly lay out the low noise amplifier, filter, power splitter and vector modulator.

90 90

-0.5 -0.5 135 -1 45 135 45 -1.5 -1 -2 -1.5 -2.5

-3 -2 -3.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 -2.5 -2 -1.5 -1 -0.5 180 0 180 0

225 315 225 315

270 270

(a) (b) Figure4-2 635MHz field pattern of E plane(a) and H plane(b)

FAST telescope has rather small focus ratio, the size of focus spot increase rapidly as the beam leave from the center, we studied the several beams from center overlapped –3dB, which

12 gives us critical information in constructing the network of beam forming.

(a) (b) Figure4-3 pattern of the array feed at different zenith angle

5. Complete FAST prototyping model

We have started a complete FAST prototyping model at Miyun radio astronomical station of the NAOC since Jan. 2004, i.e., Miyun FAST scaled model, hereafter MyFAST. The reflector of MyFAST adopts the structure of cable-mesh rather than the traditional one, as shown in Fig.5-1. The total weight of the reflector would be much lightened and the construction cost of the telescope would be reduced by making use of such a structure. This model of a scale ~1:10 will integrate all results from partial models of key technologies of FAST in the past ten years except the Karst site and the compact array feed, to demonstrate the engineering concept of the FAST by observing few bright astronomical objects. We would see this model as a demonstrator of the FAST. The civil-construction of MyFAST have been accomplished by July 2005, as shown in Fig5-2. The first light of the model is expected around the spring of 2006.

Fig.5-1 Design charter of the cable-mesh reflector of the FAST demonstrator

Fig.5-2 Overview of the MyFAST by July 2005

6. Conclusion remarks Most critical and risky technologies have been successfully analyzed, simulated, and tested by scaled models. Most questions of how to realize the telescope have been answered, and future optimization for cheaper and more reliable realization is forwarding to the right direction. We believe that the critical components of the FAST have passed their feasibility study phase, including science case, Karst site surveying, active reflector, telescope tracking, key techniques required by control and measurement, layout design for the focus package. List of Major Milestones Date Milestone 1998 Start a key project on FAST pre-research in CAS 1999 Sign a MOU on the low noise receivers between NAOC and JBO 2001 Establish a 4-unit scaled model of active reflector in Shanghai 2001-2002 Construct a field scaled model of feed supporting in Xi’an 2002 Start the research on active cable-mesh reflector 2002-2003 Operate the cable-car model in Tsinghua University in Beijing. 2003 FAST demonstrator design finished 2004 Submit FAST funding proposal as a National Mega-science Project 2004-2005 Construct FAST demonstrator at Miyun station 2006 Operate the FAST demonstrator 2007 FAST design study and its reviews 2008 FAST construction for 6 years 2014 First light and operation of the FAST

The project studies are coordinated by the NAOC, having least 20 Chinese institutes and two foreign institutes involved. Totally more than 80 senior scientist and engineer join the cooperation

14 research. About 60 graduate students take relevant R&D of FAST as their thesis topics. We have received funds in total amount of ~2 M Euro from NAOC, CAS, ministry of S&T of China and National Science Foundation of China. The capital budget of ~60 M Euro for the FAST, the real demonstrator of the KARST, would be available in the near future.

REFERENCES

1. Nan R., Nie Y., Peng B., et al., 1996, 'Site surveying for the LT in Guizhou province of China', in Proc. of the LTWG-3 &W-SRT, eds. R. G. Strom, B. Peng & R. Nan, p. 59, IAP. 2. Peng B., & Nan R., 1997, 'Kilometer-square Area Radio Synthesis Telescope KARST project', IAU Symp. 179, 1997,p. 93-94 3. Nan R., Peng B., Zhu W., et al., 2000, 'The FAST project in China', ASP 213, 523 4. Qiu Y., 1998, 'A novel design for a giant Arecibo-type spherical radio telescope with an active main reflector', MNRAS, 301, p. 827-830 5. Duan B., Zhao Y., Wang J., Xu G., 1996, 'Study of the feed system for a large radio telescope from the viewpoint of mechanical and structural engineering', in Proc. of the LTWG-3 &W-SRT, eds. R. G. Strom, B. Peng & R. Nan, p. 85-102, IAP 6. Su Y, Nan R, Peng B, Roddis N, Zhou J ‘Optimization of interferometric array configurations by sieving u-v points’, , 2004, A&A 414, 389 7. Su Y., Zheng Y., & Peng B., 2000, 'Schemes for segmenting the main reflector of the FAST', in Proc. of the 4th EAMA, ASP, 94 8. Zheng Y., 1999, 'Suggestions and comments on segmenting FAST reflector', MEMO to the FAST 9. Qiu Y. & Zhu L., 2001, 'The control system of the active main reflector for FAST', ApSS, 278(1), 249 10. Li T.Y, IEEE Trans. 1959, AP-7 No.3, 223-226 11. Nan R, Ren G, Zhu W, Lu Y, ‘Adaptive cable-mesh reflector for the FAST’2003, ACTA ASTRONOMICA SINICA. Vol.44. Suppl. 13-18 12. Zhu W., 1999, 'Description on tracking traces of FAST feed', MEMO to the FAST 13. Stewart D, “A platform with six degrees of freedom”, Proc. Inst. Mech. Eng., vol. 180, no. 5, pp. 371-386, 1965. 14. Su Y.X. and Duan B., 2000, 'The mechanical design and kinematics accuracy analysis of a fine-tuning stable platform for the large spherical radio telescope', Mechatronics, 10 (7), p. 819. 15. Su Y.X., Duan B.Y., Peng B. and Nan R., 2001, 'A real-coded genetic optimal kinematic design of a Stewart fine tuning platform for a large radio telescope', Robotic Systems, 18 (9), 507 16. Su Y. X, Duan B. Y.,. Wei Q,. Nan R. D, Peng B., ‘The wind-induced vibration control of feed supporting system for large spherical radio telescope using electrorheological damper’, Mechatronics, 2003, 13 (2): 95-110 17. Duan B., 1999, 'A new design project of the line feed structure for large spherical radio telescope and its nonlinear dynamic analysis', Mechatronics, 9(1), p. 53-64. 18. Ren G., Lu Q. & Zhou Z., 2001, ‘On the cable car feed support configuration for FAST’, ApSS, 278(1), 243 19. Cheng Y,. Ren G, and. Dai S, “Vibration Control of Gough-Stewart Platform on Flexible Suspension”, IEEE transactions on robotics and automation, vol. 19, no. 3, pp. 489-493, 2003. 20. Ren G,. Lu Q,. Hu N,. Nan R,. Peng B, “On vibration control with Stewart parallel mechanism”, Mechatronics, 14, pp 1-13, 2004 21. Baines C., Battilana J.A., Kitching G.J., Roddis N., 2001, 'Receiver systems for FAST', internal technical report to the FAST. 22. Ardenne A.van, Smolders B, Hampson G, ’Active adaptive antennas for Radio Astronomy; results for the R&D program toward the Square Kilometer Array’ Proceeding of SPIE vol.4015(2000) pp420-433 23 Peng B, “First half year of 2005 on SKA efforts in China”, SKA newsletter 2005.

15 07/09/2005

Update on LAR Development-to-Demonstrator project

The LAR group Herzberg Institute of Astrophysics (HIA) Dominion Radio Astrophysical Observatory (DRAO) National Research Council Canada (NRC) Penticton, British Columbia, Canada.

Introduction

This document provides an update to the activities on-going in Canada in the R&D required to build an LAR demonstrator, a 300-350m diameter telescope with capability up to 1800 MHz, and with an instantaneous bandwidth of at least 2:1 (potentially as high as 5:1), sky coverage down to 30 degrees elevation, resolution of 2.4 arcmin at 21cm, and a field-of-view diameter of 0.62 degree. From a technology viewpoint, such a telescope will demonstrate all of the required features of the LAR needed for the SKA, except for high frequency capability.

As stated in the original “development-to-demonstrator” document, our planning is currently focussed on development of the required technologies leading to a detailed design of an LAR telescope, and does not include demonstrator construction.

LAR subsystem development

The LAR comprises five subsystems: the tethered aerostat system, the confluence point mechanism at the focus, the phased array feed at the focus, the reflector, and the beam former. The current emphasis of the project is the development of the telescope systems, as opposed to the back-end processing systems. In this section the progress over the past year is provided, along with a brief description of the current plan for continued development.

Aerostat

Overview: With the large focal length for the LAR reflector of several hundred metres, the focus apparatus will be supported by a helium-filled aerostat, whose position will be controlled by a minimum of six computer-controlled winches. Potentially, these tethers will also play a role in the orientation of the focal apparatus. System evaluation is being evaluated using a 1/3-scale prototype to verify a dynamical model of the system. Once verified via the scale prototype, the model becomes an extremely powerful tool for the design of the full-scale system.

Progress: Over the past year, closed-loop control of a 3-tether system has been implemented using feedback from differential GPS measurement. This system has

1 07/09/2005 reduced the horizontal motions of the focus point by 2-orders of magnitude to less than 2cm, within the scatter of the differential GPS measurements. In the vertical direction, position is maintained to within 5cm.

Planning: The cause of the relatively large vertical motion is the lift generated by the aerostat. To reduce the vertical motion further, an investigation of Aerostats with attitude control has been initiated. Further improvements in horizontal control will be achieved with the addition of three more tether/winch systems.

Collaboration: This work is being carried out in collaboration with the dynamics group of Prof. Meyer Nahon at McGill University. It is supported in part by a NSERC Strategic grant ($480k).

Confluence Point Mechanism

Overview: This is the mechanism at the focus of the LAR reflector, supported by the tethered aerostat system. The Confluence Point Mechanism (CPM) controls the orientation of the phased-array feed plate relative to the reflector. The tethers that control the position of the focus are attached to the CPM.

Progress: An investigation of a number of different mechanisms has been completed. The necessary degrees of freedom, the workspace volume and the mass of the mechanism rule out a number of well-known mechanisms, and point to a cable mechanism for the control of the feed plate. In this system, the feed plate is supported by cables from a space frame that is attached to the winch-controlled tethers located on the ground, and to the aerostat leash. A mock-up of this system has been built at Laval University to demonstrate its feasibility.

Planning: Optimization of the size of the space frame and the desired workspace volume is currently underway. Once completed, a scaled prototype will be designed for flight- testing on the 1/3-scale aerostat.

Collaboration: This work is in collaboration with the robotics and mechanisms group at Laval University, led by Prof. Clement Gosselin, supported in part by the NSERC Strategic Grant of $480k.

Focal Plane Phased-Array

A key technology for enabling the LAR concept is a large phased-array feed at the focus of the LAR reflector. Given that to date no astronomically capable phased array system has ever been built, certainly not capable of attaining the current specification goals of the SKA, we plan currently a three-fold process for attaining a phased array feed array for the C-LAR: 1) an engineering demonstrator of a phased-array feed (PHAD); 2) development of uncooled, monolithic, integrated LNAs for the proposed Vivaldi antennas; development of RF-to-Optical out modules for transmission of data from the phased-array to the beam former; 3) the development of an astronomically capable

2 07/09/2005 phased-array for a currently operational radio telescope as a means of proto-typing for the LAR, followed by the design of the array for the LAR.

Progress: PHAD is an experimental apparatus designed to develop a fundamental understanding of the capabilities and limitations of phased-array feeds on reflector antennas. It is a modest-sized, engineering demonstrator consisting of a 200-element Vivaldi array designed for flexibility and quick turn-around of results. Each element will be equipped with a simple, low cost, narrow band (1 MHz) receiver system built from readily available components, with a commercially achievable system temperature of ~100K. The RF output of the receivers will be digitized immediately by a commercially available data acquisition system, with on-board Field-Programmable Gate Arrays (FPGA’s) that make the system highly flexible. Initially the digitized data will be stored in a general-purpose computer (PC) for off-line beam-former processing, but once the optimum beam-forming strategy is formed, real-time beam forming will be possible, using the on-board FPGA’s.

Much groundwork for PHAD has already been completed – a first Vivaldi-array has been fabricated, a data acquisition system has been identified and partly purchased, and much of the remaining hardware has been specified. We envisage complete fabrication of PHAD, and initial experiments within the year. The target project time is two years, overlapped with planning for subsequent development stages.

In addition to the RF aspects of the phased array, we have now established a mechanical design for the large array of Vivaldi antennas required for the LAR. The current design calls for the large array to be built from up to 12 segments, supported by a carbon-fibre frame. An extensive engineering analysis of the design has been completed.

Plans: The PHAD antenna-array will be initially tested using a spherical near-field scanner and on the 26-m telescope at DRAO. These tests will use local transmitters. Tests on the telescope will be carried out using satellites or bright astronomical sources. Collaborators are developing a full theoretical electromagnetic model of phased arrays. PHAD results will be used to “ground-truth” those models.

After achieving the goals outlined for PHAD, the following steps are envisaged: Firstly, identify and develop critical components needed for practical realization of phased-array systems, including antenna feeds. As understood now, there are two components needed: • Integrated Low-noise Amplifiers (LNAs). The specific goal of this work is to design uncooled, monolithic amplifiers, whose characteristics are well matched to those of the antenna elements. The initial goal is to achieve receiver noise levels on the antenna of 15-20 K, a bandwidth ratio of 2:1, a gain of 20-30 dB, and power consumption of <100 mW. • RF-in/Optical-out Receiver Modules (RORM): These modules follow the LNAs – they amplify the signal by 80-100 dB, digitize the high-level signal, and modulate the digital signal on to optical fibre. The optical fibre signal is passed to the beamformer in a production system, via multiplexing steps needed to pack the maximum bandwidth on a single fibre.

3 07/09/2005

Collaboration: Research into the LNA development is being done at the University of Calgary, led by the group of Prof. Jim Hazlett. The RORMs are being investigated in a series of investigative design contracts to BreconRidge Manufacturing Solutions of Ottawa. The PHAD project was recently awarded $870k from a Major Initiatives grant from NRC.

Reflector

Overview: The LAR reflector is a large diameter, faceted approximation to a parabolic reflector with a long focal length (f/D~2.5). Each facet is a triangular section, and several facets make up a structure unit. Each structure unit is actuated at each corner to maintain the parabolic shape as the pointing direction is changed. The reflector is very flat, permitting the weight of the structure to be supported at many locations, rather than the typical single or double mount point.

Progress: An Actuated Structure Unit (ASU) has been built and tested at DRAO. The major achievement of this construction was the success of the low-cost primary actuators. The cost of these actuators scales with the desired throw, and we believe that actuators with throws ~15m can be built for relatively low cost. A preliminary control system has been developed for the ASU. It is now possible to identify a cost estimate for the LAR reflector concept, complete with actuation and metrology system to be ~$400 US per square metre.

Plans: To ensure that the relative position of each of the facets is maintained for optimum reflector performance, secondary actuators for each of the facets are required. A metrology system will also be required to measure the location of the panels, and provide part of the feedback system for control of both the primary and secondary actuators. In the latter stages of the reflector R&D, an investigation of high-frequency panels and concomitant control system is envisaged.

Collaboration: The reflector work is being carried out in conjunction with Prof. Sigi Stiemer at the University of British Columbia, AMEC Dynamic Structures Limited (ADSL), and with Bosch-Rexroth Canada. In addition, Prof. Benoit Boulet of McGill University is collaborating on the design of the reflector control system, supported in part by NSERC.

Personnel

Table 1 summarizes the NRC staffing allocations for the work outlined above. This amount of effort is supplemented by university staff and students, and a the participation of a number of industrial partners. The entries are integrated FTE’s over a 4-year project, to complete in 2009. In the past year, a mechanical engineer has been hired, and two experience electrical engineers have recently become available to the project.

4 07/09/2005

Table 1: Estimated Personnel Requirement at HIA (Totals for a 4-yr Project) ME RF CE MT ET DE Misc. FTE's

Aerostat 4 4 0.3 8.3 Focal Apparatus 1 0.3 2 1 4.3 Reflector 1 1 4 1.5 7.5 Feed 4 8 3 3 18 Beam Former 3 3 Correlator 1 1 Control Engineering 2.5 2.5 Project Management 3 3 Science Support 1.5 1.5 SKA Support 6 6 Total FTE's 10 8 3.8 10 3 4 16.3 55.1 Average Staffing Level 2.5 2 0.95 2.5 0.75 1 4.1 11.0 Existing Staffing Level 2 1 1 1 0 1 1.5 7.5 Average FTE Deficit 0.5 1 -0.05 1.5 0.75 0 2.6 6.3 Priority Staffing requirement 0 1 1 2 ME Mechanical Engineering RF RF Electrical Engineering CE Control Engineering MT Mechanical Technology ET Electronics Technology DE Digital Electronics Engineering Misc. Science, Management, Project Management FTE Full Time Equivalent or Person Year (PY)

Budget

The proposed budget profile for the planned 4-year LAR R&D project is given in Table 2. In addition to the sub-projects outlined above, beam-former and correlator design work, and the anticipated cost of salaries are included. With a 30% contingency, we estimate a total budget of $9.8M CDN.

Table 2 :LAR Development Sub-Projects 05/06 06/07 07/08 08/09 4-yr total Aerostat 352 230 85 0 666 Focal Apparatus 87 545 400 50 1,081 Reflector 212 195 320 50 776 Phased-Array Feed 505 500 900 100 2,005 Beam-Former 0 200 100 0 300 Correlator 0 100 0 0 100 Project Management 110 110 110 110 440 Total Development Operations 1,266 1,879 1,915 310 5,369 Continuing Salaries 300 300 300 300 1,200 New Salaries 100 300 300 300 1,000

Totals w/o Contingency 1,666 2,479 2,515 910 7,569 Totals including 30%Contingency 2,165 3,222 3,270 1,183 9,840

5 SKA Developments in the US A Report from the US SKA Consortium

October 4, 2005

Summary Signiﬁcant work on the SKA project continues in the US, including work on the Allen Telescope Array; work funded by the National Science Foundation on feeds, receivers and RFI mitigation; concept development for replacement of the Deep Space Network with large-array facilities; and signiﬁcant work on the science, engineering, and siting working groups. The large Technology Development Project submitted to the NSF in 2004 March is, as of this writing, still unfunded.

The Allen Telescope Array The ATA is progressing towards a working 42-element complement, with 30 elements in place and the remainder to be emplaced before the end of the year. Element ”kits” are sent to the Observatory and the staﬀ can then emplace about one per day. Production versions for nearly all signal path and electronics are being assembled and those remaining are rapidly being ﬁnalized. Figure 1 shows a fairly recent picture of the site and Figure 2 shows the vehicle used for moving antennas from the construction tent to its waiting pedestal.

Fig. 1.— The ATA site.

Fig. 2.— An ATA antenna and transporter. 2

Currently, four elements are being used for testing various aspects of control and processing. Figure 3 shows an image of M31 from these four antennas and the prototype, fpga-based correlator. Only a single pointing was used in this image of this very large source. Production versions of the feed are being tested in our new feed testing chamber in the Bay Area (Figure 4). Final feeds await a new version of a small quartz circuit board that was required due to a process failure at the manufacturer.

Fig. 3.— An image of M31 along with three spectra made with the ATA.

Fig. 4.— Testing of the ATA feed. 3 The Deep Space Network Array The concept of large arrays for deep space communications (DSN) has developed a consensus at JPL and growing recognition within NASA as an important infrastructure required for the future space program. The DSN receiver array is presently envisioned as 400 x 12m antennas at each of 3 longitudes to be constructed in the 2010-2015 period. It will also include a smaller transmitter array and allow retirement of the present DSN with 40 times improved data rate and, lower operating cost through automation and the inherent soft-failure property of an array. There will be extensive project reviews, and optimization in the 2005-2009 period. A small array with 12m antennas for lunar communications will be implemented at Goldstone by 2009 and provides a prototype for the DSN array. At present the DSN array is in a development phase with design and testing of receivers and antennas. A breadboard array consisting of two hydroformed 6m antennas and a paneled 12m antenna is near completion and will be tested at JPL in 2006 along with digtal array combining equipment and monitor and control software. The receiver covers 8.0-8.8 GHz and 31-38 GHz with a cryogenically-cooled, dual-frequency feed and InP MMIC low noise ampliﬁers providing system temperatures of 20K and 40K measured on a 6m antenna.

NSF Funded Technology Development Members of the US SKA Consortium have been conducting work through funding from a National Science Foundation grant from 2002-2005 (extended to mid-2006). This work includes antenna and mount concept development (Weinreb, CIT/JPL; Cor`tes, Cornell); broadband feed development (Weinreb, CIT/JPL); work on RFI mitigation and excision algorithms (Bhat, Haystack; Cordes, Cornell; Ellingson, VT); and array conﬁguration studies (C. Lonsdale et al., Haystack Observatory). The NSF grant that has supported this work also has contributed to US funding of the International SKA Project Oﬃce.

The Technology Development Project (TDP) The US SKA Consortium’s TDP was submitted as a 10-institution proposal to the NSF in 2004 March and it was reviewed by a panel with whom we had a “reverse site visit” in 2004 October. The written reviews of our proposal were quite positive, although there were signiﬁcant questions about how the timeline for the TDP would ﬁt in to the proposed international time line for the SKA and for presentation of the project to the next “decadal survey” in the US, anticipated in 2009-2010. The funding of the TDP is severely impacted by the Senior Review (SR) that is now taking place to assess the status of NSF-funded astronomy facilities. The SR is aimed at addressing the severe budgetary pressure on the Astronomy Division of the NSF, which derives from (a) the strong demand on new funds for new optical telescopes and for the SKA project; and (b) the need to identify operations funds for ALMA. At this time, funding for the TDP does not appear forthcoming. Consequently, the US SKA Consortium will be proposing smaller subprojects to individual programs in the NSF that target astronomy instrumentation and instrumentation for universities. In this way we can expect to continue progress on subsystems of the large-N/small-D concept over and above those now being achieved with the ATA and the DSN array.

EMBRACE (Electronic Multi-Beam Radio Astronomy ConcEpt) An update 2005

Arnold van Ardenne and Marco de Vos ASTRON, NL 1 Executive Summary:

At Penticton, Canada meeting, we presented the four year plan for the European Demonstrator Embrace. In this paper we present an updated view of the project plan, report on the technical progress and respond to some of the questions raised during the previous meeting.

2 Background:

The EMBRACE is denoted as the Electronic Multi-Beam Radio Astronomy ConcEpt which is planned as a 300 square metre aperture array with multiple independent Field of View (FoV) capabilities. The main objectives of EMBRACE are to demonstrate the technical and scientific potential of the aperture array concept using a low cost phased array station with the essential Square Kilometre Array (SKA) functionality in combination with the Westerbork Synthesis array.

The European SKA Consortium has adopted the aperture array concept [1-5] as most promising to cover the science goals associated with observations to be conducted at low frequencies (0.1 – 2.0 GHz). In this concept beam pointing to the radio-source is electronically controlled over large scan-angles while interferers can be suppressed adaptively. By multiplying parts of the electronic subsystems such as station receiver chains and array correlator by, say M, simultaneous and independent reception of widely separated directions is possible, without the need to multiply the basic mechanical structure. This in effect, corresponds to creating multiple Fields of View. The principle of forming simultaneous and independent beams for Radio Astronomy has already been demonstrated with the completion of the previous project known as THEA, where four square metre tiles were designed, manufactured and measured with two independent multiple beams. As THEA was a prototype design, the cost associated with each tile was high and therefore to extrapolate this tile cost to a SKA station would be unrealistically high. Therefore a larger demonstrator with capability of providing independent multiple beams was proposed with several objectives:

• To reduce the manufacturing costs associated with the production of the tiles. • To understand the issues associated with the design, manufacture, assembly and operational aspects of a large, multiple beam phased arrays. • To understand the observational issues by integration of the phased array telescope with the 14 telescopes of the Westerbork Synthesis Radio Telescope (WSRT). 3 Technical progress on EMBRACE Significant technical progress has been made on several fronts with an important objective of building an inexpensive EMBRACE tile. A European team has been established to share some of the work and benefit from the collective experience of many groups which resides within various institutes. Some of this progress is described below.

3.1 EMBRACE SYSTEM CONCEPT

The EMBRACE system will be built on the similar principle as THEA [2]. A large number of antenna tiles, each of area ~1 m2, will form the collecting area. The signals from the 64

1 elementary radiating elements from each tile will be amplified and initial RF (i.e. analogue) beam-forming will be applied. It is not intended to perform any digital beam forming in the tiles but, due to the limitations imposed by the use of phase shift control with large instantaneous bandwidth requirements, a tile may be split into quadrants. The outputs of these quadrants can then be combined with time-delay lines directly, or be sent to the backend directly for digital processing.

A system level block diagram is shown in Figure 1, indicating the formation of two independent fields-of-view at the tile level and at the aperture array “station” level. Note that all the processing is done at the back-end using an optical analogue link from the aperture array. For the analogue link we are pursuing an RF-on-fibre approach where the RF signal modulates a laser directly. The analogue link from the tiles to the back-end processing simplifies the tile design and totally decouples the analogue antenna from the receiver thus reducing, for example, EMC related problems in general, since the local oscillator and clock signals now have to be distributed in the back-end cabinet only. The down-conversion and digitisation is also carried out in this back-end cabinet.

Figure.1: A system level block diagram of EMBRACE showing the formation of independent multiple fields-of-view at both the tile and station level, with processing at the back-end.

3.2 EMBRACE SPECIFICATION

In Table 1 the specifications for the key parameters of the EMBRACE system are given. From the collecting area and the λ0/2 spacing at 1 GHz, it is estimated that a total of 20,000 antenna elements will be required. The system will be built with tiles approximately of 1x1 metre size although larger tiles will also be considered.

Table 1. EMBRACE Specifications Frequency range of receiver chain: 400 MHz - 1550 MHz. Element separation: λ0/2 at 1GHz Polarization: Single polarisation Physical collecting area ≈ 300 m2 Aperture efficiency: >0.80 Electronic scan range: Full hemispherical Tsys: <100 K @ 1GHz (aim for 50 K) Antenna element phase control accuracy: 3 or 4 bit Instantaneous bandwidth: 40 MHz Dynamic range A/D converter: 60 dB Number of independently tuned FOVs (RF 2 beams): Number of digital beams: 8, of 20 MHz per FoV

3.3 EMBRACE TILE CONFIGURATIONS

The tile provides the physical area to receive the incident electromagnetic waves in the required frequency range with a single polarisation. It consists of the receiving elements, low noise amplifiers and phase shifters as well as the power combiner/splitter. Figure 2 shows how the 4-elemental radiators are combined with 1:2 way RF power splitter, two phase shifters per radiator and further 4:1 power combiners to produce a 2 beam configuration (RFIC 1).

Figure.2: 4-radiator configuration for producing two RF beams Figure.3: Four of the 4- radiator configuration

The 4-radiator configuration shown above now forms a building block for combining rest of the radiators within the tile. The four of the 4-radiator configurations are now used with a further 4:1 (RFIC 2) combiner and amplifiers, to combine 16-elements within the tile as shown in Figure.3.

The sixteen-element arrangement is then further combined to connect all the 64 radiators within a tile as shown in Figure 4, producing two RF beams. There is no loss of generality in producing only 2 RF beams as the concept can easily be extended to produce 4 or more beams.

Figure 4: Four of the sixteen-element radiator configuration

3.4 Embrace low cost design approach

The analyses of the production cost of THEA revealed that for a number of sub-systems a lower cost design must be used, in order to make an array with the size of EMBRACE feasible. A few of these sub-systems will be discussed in this paragraph. Although the list is not complete, a more extensive technology study will be performed in SKADS (SKA Design Studies), “enabling technologies” section, in which all critical components of an aperture phased array system will be addressed.

Printed circuit boards, as used in THEA for antenna, low noise amplifier and first beam-former stage, costs around €25 per antenna element. The element cost can be reduced if we separate the large antenna sheet from the beam-former and integrate the beam-forming electronics. Figure 5 illustrates a low cost production method for the antenna boards where antenna elements are stamped out of rolls of Copper, producing antenna elements for a Euro or less.

Figure 5: Production methods for low cost antennas

The analogue beam-forming in the phased array system, can be performed with time delay sections, a phase-shifter or a vector modulator. Time delay sections are usually expensive and cannot be easily integrated in an IC. Phase-shifters and vector modulators can be made very compact, e.g. four phase-shifters can be designed in one IC. A first prototype phase- shifter has been designed and processed in GaAs technology (Figure. 6), with excellent results. Further cost reductions are expected when the design can be made in Silicon BiCMOS.

Housing of the antenna and beam-forming components is a critical element not only in terms of cost, manufacturability and reliability but also from a system performance perspective (temperature). Limiting the number of connections and parts is an important issue in the

4 design of the EMBRACE tile. In Figure 7, an exploded view of the tile concept including radome is given. These tiles can be placed side by side, creating a large collecting area required for EMBRACE or even a larger station made of phased array tiles.

Figure 6: Phase-shifter layout

Figure 7 Exploded view of the EMBRACE tile concept

3.5 Extension of EMBRACE to a dual polarisation design

We have designed EMBRACE for a single polarisation only and our rationale for this decision is given in section 4.1. We appreciate that a dual polarisation antenna is a requirement for SKA specification. Therefore we will design the radiating element in such a way that it is easy to extrapolate to a dual polarisation case. Some of our ideas for extending the single polarisation to dual polarisation are given below. Figures 8.1 to 8.3 show the linear polarised case and Figures 8.4 to 8.7 shows the dual polarised case. Note that the concepts presented below are novel and will be realised in the near future to determine their suitability.

Figure 8.1: Radiator stand Figure 8.2: Inserting single radiator Figure 8.3: Single polarised ele ment

Figure 8.4: Radiator stand Figure 8.5: Inserting two LP radiators Figure 8.6: Dual polarised elements

Figure 8.7: Completed dual polarised tile

3.6 EMBRACE Work Breakdown Structure

The Work Breakdown Structure for the entire EMBRACE project is given below:

The entire project is divided into three separate phases, namely design and development, production and the testing of Embrace. Some initial engineering testing will also be carried out in T1 and T2 followed by the scientific testing of the entire Embrace station in T3.

The Work Breakdown Structure for the entire project is shown in Figure 9, using different a colour for each phase.

Figure 9. Work Breakdown Structure for the realization of Embrace

The initial phase shown in green (T1) is the design phase where a single tile will be built and tested. Iteration may be required but within 18 months at least 10 tiles will be built and tested. The design will then be frozen and the production phase will commence as shown in beige (T2). In this phase the further production of tiles will be initiated in collaborations with our industrial partners. The development of the required infrastructure at WSRT site will also be completed in time for the delivery of the tiles, in preparation for assembling Embrace.

The final phase of the work requires the engineering testing of Embrace (T3). The testing will involve measuring the antenna related parameters such as gain and the noise level to provide a figure of merit for Aeff/Tsys. Further tests will involve the testing of the antenna pointing accuracy, beam shape stability and its calibration process.

An important aspect is the EMBRACE station beam and how its shape and polarization properties change due to projection effects when an object field is tracked on the sky. Although EMBRACE has only a single polarisation output, it is cross-correlated with the two orthogonal polarization signals of the other WSRT dishes and provides full polarisation characterisation of the single EMBRACE channel. We will verify how well the implemented correction procedures perform relative to the well known behaviour of the WSRT beams. Special attention is needed for the side lobe structure that rotates on the sky in contrast to the patterns of the equatorially mounted WSRT dishes and how that affects the corrections for sources outside the main beam.

3.7 Technical management of EMBRACE

EMBRACE will be built by a small subset of the SKADS consortium members. ASTRON will manage the project as well as take the lead in the Systems design. Nançay will be responsible for the design of the back end processing based on existing ASTRON LOFAR technology. IRA will be responsible for the design of the receiver chain. MPifR will be responsible for the analogue fibre link between the tile and the receiver, to be housed in an air

7 conditioned environment. The management structure for the EMBRACE project is given below in Figure 10.

Figure 10: The Organisational Structure for EMBRACE

A Gantt chart showing the activities for the design, development and testing of Embrace is shown below in Figure 11.

Figure 11. Gantt chart for the design, development and testing of Embrace

3.8 Tile costing estimation

A unique feature of the phased array concept is that FOV can be limited while maintaining the ability to probe the sky at independent places. Limiting FOV is essential, since with a square kilometre aperture it is easy to generate data streams for which no processor is big enough in the time period considered for SKA and far beyond. Having accepted that FOV has to be limited in the instrument, it is clear that this should be done as close as possible to the radiating elements.

The EMBRACE tile architecture uses a full RF tile approach, resulting in a well balanced cost design. Cabling running to the tiles forms a major cost component. However with the EMBRACE concept, the tile cabling is kept to a minimum.

8 The current cost estimate for the prototype tile is around EUR 2000. This includes the housing, radome and antenna elements (EUR 500), and all the tile electronics (EUR 1500). These numbers are based on the volume required for EMBRACE.

3.9 Further cost reduction strategies.

The question of low cost production methods and the use of inexpensive material have been addressed above. However, we have ongoing discussion with major semiconductor houses, who are also interested in the low cost production of the active devices such as the LNA, phase shifters, receiver chain etc. for SKA . The technology push from these companies is very likely to lead to an extremely low cost solution for these active components, which will provide useful input data for the technology assessment of Embrace, to be carried out by the ISSC for the selection of the SKA concept.

4 Further questions from IEMT’s July 2004 report. 4.1 Dual Polarisation

We re-iterate and defend our view regarding the decision to build Embrace with a single polarisation. One of the main objectives of Embrace is to learn about designing a very large phased array and the operational issues related to it by building the largest array within budgetary constraints. The other equally important objective is to obtain an estimate of the likely costs associated with such a large phased array for radio astronomy and to be able to extrapolate these costs to 2012 or so time period. The design, manufacture and testing of Embrace will provide valuable data in terms of performance and cost, which is necessary in order to extrapolate the cost for a SKA.

We appreciate that the SKA antenna requires dual polarisation, however within the SKADS proposal, there is a task activity which is specifically aimed at designing a dual polarised tile. By the time the Embrace demonstrator is completed, we will have also completed, several dual polarisation tiles, with sufficient area to do some sky (or satellite) measurements. It will also provide useful engineering data for the evaluators to extrapolate from the single polarised Embrace to a dual polarised case, required for the SKA programme.

4.2 Instantaneous Bandwidth

It is accepted that in the design of Embrace, every efforts should be made to push the technological boundaries already achieved with THEA. However, it is also accepted that the main objective of Embrace is cost reduction. The cost reduction requirement suggests using existing hardware and software and the use of ‘off the shelf’ equipment, where possible but at the same time it should not limit or restrict the performance of Embrace. Furthermore, the design of Embrace should be such that it must be able to be extrapolated in both performan ce and costs in the 2012 time period.

The instantaneous bandwidth assumed for Embrace is 40MHz, which is twice as much as available for THEA but not as broad as required by the SKA specifications. The choice of Embra ce instantaneous bandwidth does not in any way restrict or limit its extrapolation to the full bandwidth requirements of the SKA specification.

4.3 LOFAR Backend Signal Processing

For Embrace, the LOFAR backend will be used for signal processing. The input bandwidth for the LOFAR backend is 100 MHz whereas the assumed bandwidth for Embrace is 40MHz. The design of Embrace therefore is not being restricted by the capabilities of the LOFAR backend. Also, there is no restriction on the number of beams.

4.4 Risk and contingency Mitigations

The technical merits of phased arrays for radio astronomy with multiple and independently steerable beams have been demonstrated rather convincingly with the completion of the

9 THEA project, yet there remain risks within this project. We address some of these concerns and our mitigation strategies.

Developing a technologically advanced system such as EMBRACE inherently has risks assoc iated with it. By establishing the EMBRACE project as a European collaboration with the active participation of several national institutes in the different phases of the project, we will have access to the experience of these institutes in key technology areas and an extensive knowledge base. This experience will be enabling us to minimize the risk during the project.

Due to the sheer number of units, the costs of the antenna tiles are the most sensitive to changes in the estimated costs. There are three major cost drivers in the antenna tile which are critical: The first is the antenna elements itself, although it will be based on the proven technology from the THEA project. The second is the analogue beam-former chips which will be developed and produced in cooperation with OMMIC, an industrial partner within the SKADS project, to minimise the development and production risks. The third is the mechanical aspects of the tile, where the experience of ASTRON will play a major role in order to reducing the costs and the risks.

With the EU funding secured, the focus for all involved parties within EMBRACE moves towards securing national funding. All national funding has not been secured to date, this remains a slight risk to the EMBRACE project.

The international cooperation provides significant benefits in terms of risk reduction for the EMBRACE project as it introduces the problem of managing a multi-site, multi-language project. From ASTRON this is handled by appointing an experienced project manager to the EMBRACE project, supported by a system group with participation from the different institutes. Also communication links are setup at all levels throughout the project between the persons who need to be working closely together.

4.5 EMBRACE Software

The software implementation is divided in embedded software in the FPGAs performing the digital processing of the data streams and Control software running on a standard computer.

The embedded software will largely be based on the tested and proven VHDL code from the LOFAR system. The adaptations needed for EMBRACE will be specified and implemented by OPAR with the support from ASTRON. The embedded software performs the filtering, sub- band separation, beam-forming and applies the calibration coefficients to the received data. The software also provides for the direct control of the associated hardware as an intermediary for the EMBRACE control unit.

The EMBRACE Control Unit will be based on the LOFAR LCU. The LOFAR LCU is a Linux based computer system which is designed for the control of one remote station. The LCU is also used to support activities as calibration, local management procedures and provides a clear interface to a higher layer management system which manages the complete LOFAR system. As the functionality of the LCU is close to what is needed for EMBRACE, it will be used as a starting point for the development of the EMBRACE control unit. The design will focus on the local management of the EMBRACE system, however interfaces will be provided to the central management systems at Nançay and Westerbork to setup common measurements. 5 Acknowledgement: The material presented in this update paper includes work of the whole EMBRACE team presently located at Astron, Dwingeloo, The Netherlands. The authors would like to acknowledge their considerable contribution. 6 References: 1. A. van Ardene,.”The European Aperture Array SKA Demonstrator., presented at SKA 2004, Penticton, Canada .(http://www.skatelescope.org/PDF/EMBRACE_Final_IEMT)

10 2. A.B. Smolders and G.W. Kant, .Thousand Element Array (THEA)., IEEE, Antenna and Propagation, Symposium 2000. 3. A. van Ardenne, “Active Adaptive antennas for radio astronomy; results of the R&D program toward the Square Kilometer Array”, Proc. SPIE Conf. 4015 Radio Telescopes, München, Germany, March 2000 4. A. B. Smolders, G. A. Hampson, “Deterministic RF nulling in phased arrays for the next generation of radio telescopes”, IEEE Trans. on Antennas & Propagation, September 1999 5. G. A. Hampson, J. G. Bij de Vaate, ‘Verification of THEA Tile Calibration and Beamforming Results using a Near Field Scanner’, London, 31st European Microwave Conference, Proceedings Volume III Pg. 141-144, September 24-28, 2001.

THE KAROO ARRAY TELESCOPE (A SQUARE KILOMETER ARRAY DEMONSTRATOR) PROJECT UPDATE – AUGUST 2005

PREPARED FOR PETER HALL (INTERNATIONAL SKA PROJECT ENGINEER)

Abstract

This document gives an update of current status pertaining to the Karoo Array Telescope, South Africa’s demonstrator project for SKA. TABLE OF CONTENTS

1. KAT REFERENCE DESIGN AND OPTIMISATION ...... 3

1.1 OPTIMISING THE ANTENNA SIZE AND RELATED KAT SYSTEM ISSUES...... 3

2. KAT COSTING AND FUNDING ...... 4

3. COLLABORATION ...... 5

4. CAPACITY BUILDING, BURSARIES AND GRANTS...... 5

5. PROGRESS ON KAT SYSTEM ENGINEERING AND SUB-SYSTEMS WORK...... 6

5.1 NEW CONCEPT EVALUATION ...... 6 5.2 SYSTEMS ENGINEERING ...... 6 5.2.1 Systems engineering approach formulated...... 6 5.2.2 First baseline documentation completed...... 7 5.2.3 First KAT system model captured in CORE...... 7 5.2.4 KAT prototype (FPA digital beamformer) system model captured in CORE...... 7 5.3 ANTENNA STRUCTURES ...... 9 5.3.1 Initial costing and strategy ...... 9 5.3.2 Antenna optical design study...... 9 5.3.3 KAT (traditionally engineered) prototype antenna...... 10 5.3.4 “Fast track” prototype antenna ...... 11 5.3.5 Antenna cost models ...... 11 5.4 FOCAL PLANE ARRAY ...... 12 5.4.1 FPA design study...... 12 5.4.2 Collaboration with ASTRON and CSIRO ...... 13 5.5 DIGITAL SIGNAL PROCESSING ...... 13 5.5.1 CSIRO / ATNF ...... 13 5.5.2 Berkeley ...... 13 5.5.3 Digital beamformer FPA prototype ...... 14 5.6 COMPUTING / SOFTWARE...... 14 5.6.1 Collaboration...... 14 5.6.2 Prototyping...... 15

2 1. KAT REFERENCE DESIGN AND OPTIMISATION

The South African SKA demonstrator project is an array radio telescope planned for construction in the Northern Cape Province. The reference design is 20 parabolic antennas, each of (15) fifteen meters diameter, each equipped with a 10x10 dual-polarization focal plane array in the prime focus.

KAT will be designed so as to have a modular and scalable upgrade path.

The KAT team works from two project offices – the systems engineers, FPA sub-systems manager and project scientists are based in the SKA (Johannesburg) office whilst the software and computing, antenna structures and digital signal processing team members are based in Cape Town. The newly established Cape Town office is close to the major roads and freeways, the University of Cape Town and the South African Astronomical Observatory (SAAO).

1.1 Optimising the antenna size and related KAT system issues Whilst KAT (reference) design and development activities are making good progress, the KAT team is simultaneously involved in investigations to decide on the optimum size for the antennas. The investigation into optimum antenna size arose initially from the issues Tim Cornwell raised (the “Big Gulp” concept), but has since gone beyond his initial concerns with regards to the 15m diameter antenna choice.

Filling the UV plane

Wide-field Dish grating rings efficiency & calibration (blockage is problems an issue)

Antenna size Cost of Total electronics for KAT collecting and software area

Construc- Lifetime tion cost for costs structure and drives

Figure 1: Factors influencing an optimized antenna size

The size of the antennas is affected by the following factors: • Efficiency requires antennas ≥ 3m diameter; • Filling the UV plane might require at least 40 antennas (based on other studies); • The wide-field imaging and calibration problem (arising from strong sources which are on the edge of the beam) requires a wide primary beam and so argues for a smaller antenna1; • The total collecting area should not decrease;

1 This statement needs to be checked for accuracy – is this true in the case where a small antenna size is chosen to have a comparable FOV as 15m antennas? Final 3 • An increase in the number of antennas means more electronics and a much bigger correlator. The question we seek to answer: “Can we build a correlator to handle ≥ 1,000,000 baselines?” • Mechanical parts for many antennas may imply higher maintenance costs than mechanical parts for fewer antennas (but this is unclear if some are COTS devices (smaller antennas) and others are customized (larger antennas)). The manufacturing costs for smaller antennas should be lower, but this remains to be studied in more detail.

2. KAT COSTING AND FUNDING

The SKA and KAT projects continue to enjoy a high profile in the South African Government and are being supported by a number of departments.

The prototyping phase for KAT is fully funded as per budget. The funding for KAT has been requested from the Treasury by the Minister for Science and Technology. Additional funding is being negotiated with the DTI.

An international review of the budget for KAT is currently under way.

Figure 2: Sub-Systems costs as % of total KAT costs (excluding infrastructure)

SE 7% Comput ing 16% FPA 4%

DSP

Comput ing

FPA

Dishes 24% Dishes

Science

Proj ect

Total = USD51.843m

DSP 43% Science 3%

Proj ect 3%

Final 4 3. COLLABORATION

From its inception, the KAT has been an international project. As a result, peer reviews and collaboration on technology development are ongoing. The table below lists key collaborations and objectives thereof.

Table 1: Summary of most significant collaborative activities for KAT KAT sub-system Key international collaborator(s) and objective(s) Systems engineering o International peer review sought out on a continuous basis to ensure learning and input from areas where South Africa may be “thin on the ground” o We have found that there seems to be a lack of international expertise in systems engineering for radio astronomy instruments and are seeking to strengthen this. o Objectives: Ensuring a high quality systems design for a world class instrument; ensuring transfer of our learning to the international radio astronomy community. Antenna structures o Collaborative partners to be identified as investigations continue. Initial collaborations have been with MAN Technology (Germany). o Objective: The US$1,000/m2 SKA antenna FPA o Collaborative partnerships being built with ASTRON (for the 10x10 dual-polarised antenna systems with characterization). We plan to second an antenna engineer to ASTRON in the Netherlands to work with their designers and ensure transfer of knowledge. o Collaboration with the 2-PAD team is also planned. o Objective: Transfer of knowledge and skills from ASTRON and CSIRO; Providing human resources in areas where the international effort perhaps does not have sufficient funds to employ teams; Cost reduction through in-kind contributions from world experts. Computing / Software o Extensive collaboration. We have seconded a radio astronomer with skill in software engineering to CSIRO to work alongside Tim Cornwell for a period of 1-2 months. o Objective: Building a common software platform with CSIRO for both xNTD and KAT; sharing human resources and thereby ensuring technology and knowledge transfer; cost reduction through shared resources and in-kind contributions. Digital Signal o Acquire CSIRO beamformer and integrate with FPA to perform digital beamforming. Work Processing with CSIRO (John Bunton’s group) to develop full digital system (beamformer). The first stage would be the development of a test board to investigating various high speed communication options for implementing the beamformer. o Objective: Acquire hardware in most cost effective way, form KAT DSP team with significant capabilities to complement the CSIRO team. Science case and user o International workshop planned for 19-23 September 2005, requirements o Several other ongoing international peer reviews

4. CAPACITY BUILDING, BURSARIES AND GRANTS

• An extensive recruitment process is under way to strengthen the KAT team in its entirety. In particular, a strong software team has been recruited (and will be grown) and recruitment has started to strengthen the DSP team. In addition, several contracts have been placed with South African industry, research councils and academia to speed up delivery. • To build future capacity in radio astronomy and all related disciplines (engineering and science) a bursary programme has been introduced during 2004, with student supervisors both from South Africa and abroad. Following a first call during which 11 post-graduate bursaries were awarded, a second call was issued during August 2005. The bursary programme is expected to grow over the next three years, ensuring that South Africa has suitable capacity to develop, construct and maintain the SKA should the South African bid be successful. • Academics and other practitioners in areas related to the SKA may apply for travel grants – to date, 6 travel grants have been awarded. Grantees feed back their learning in a structured way, ensuring that money is well spent.

Final 5 5. PROGRESS ON KAT SYSTEM ENGINEERING AND SUB-SYSTEMS WORK

The current 20x15m antenna KAT is the reference design. All systems engineering work and technology development to build this reference design by the end of 2008 / start of 2009 continue with great enthusiasm.

5.1 New concept evaluation The KAT team has developed and implemented a process through which to evaluate new (and potentially disruptive) concepts for an SKA demonstrator telescope which may ultimately replace the current KAT reference design. An overview of such process is shown below.

Reference design uncertainty, risk Design freedom, Time

Revolutionary concepts Manufacture (which survive a rigorous evaluation go / no-go process – based on cost, risk, decision timelines, science impact) Figure 3: Process for evaluating alternative system concepts without disrupting the reference design work. 5.2 Systems engineering

5.2.1 Systems engineering approach formulated

A tailored systems engineering approach has been formulated to satisfy the unique requirements of a project like KAT. Key aspects of this approach include: • A KAT life cycle plan allowing for evolutionary concept development until the pre-production design freeze. • A systems engineering modeling framework for capturing the key system design aspects in an interactive database available to the KAT team.

Final 6 • Converging reference design methodology with intermediate milestones to focus the KAT team.

Validation

Opera- User User Requirements definition tional spec testing Verification

System Design System System & Modeling spec int. & test Verification Sub- Subsystem Design Sub-sys system n & Prototyping Verification int. & test o spec iven i r at -d r n eg o t ti n Component Manufac- Component a i ic p spec ture int. & test if -u m Ver o tt o B Manufacture Go / no-go Risk-driven Concurrent design

Figure 4: KAT life cycle plan overview

5.2.2 First baseline documentation completed

A first draft of the two highest level systems engineering documents have been completed as a first draft baseline: • The Stakeholder Requirements Definition document captures the first draft user requirement. This document will be refined over the next three months to become the first formal stakeholder requirements baseline. • The System Concept Design captures the first draft physical and functional design of the system. This document will be refined over the next year to become the final system design baseline. The purpose of publishing these “first draft” documents was firstly to serve as a draft blueprint to focus the team effort and secondly as documents which can be sent out to system concept reviewers.

5.2.3 First KAT system model captured in CORE

The first draft KAT system design has been captured in a CORE database model. This model includes the functional flow, functional allocation, data definitions, data flow, control flow, physical breakdown, physical interfaces and technical requirements of the system. An example of a functional and data flow diagram from CORE is shown in Figure 5. This model will be updated over the next months and will form the core of all future KAT system design work.

5.2.4 KAT prototype (FPA digital beamformer) system model captured in CORE

The first KAT prototype, called the “FPA digital beamformer” forms the next major milestone for the KAT team. A CORE model has been developed which captures the design of this prototype.

Final 7 This model forms the central design reference for the team and for all software and hardware development work over the next few months.

1.5.1.2

Wide field imaging

EM RF Widefield imaging Complex visibilities trigger

1.5.1.3

Pointed observations IntegratedSampledDa ta

Pointed observations trigger

External Transient Transient event detected trigger 1.5.1.4 ChannelisedStationBe Monitor transients amData

1.5.1.1

Monitor transients LP Schedule resource AND AND LP command usage

1.5.1.5

Implement Schedule

Experiment list

Schedule plan

VLBI recording trigger

1.5.1.6

VLBI recording Point antennas command

VLBI data product

1.5.1.7

Point antennas

Figure 5. Example of KAT functional and data flow model in CORE

Final 8 5.3 Antenna structures

5.3.1 Initial costing and strategy

The antenna costing which was done for the purpose of the KAT budget (worst case scenario) has resulted in a cost of about US$3,000/m².

Our original intention was to invite universities and industries in RSA and other countries to compete to provide the best low cost design for an antenna which would comply with our KAT specifications and which would also allow major cost reduction in the large scale manufacturing for the SKA. The objective was to design and prototype an antenna for KAT which would cost ≤ US$1,000/m², including the pedestal, gears, motors, bearings, encoders etc. Time constraints meant that we could not go for the innovative design immediately - due to the unavailability of an antenna that the KAT team could use as a test bed for the FPA and beam former, the cost optimization studies had to be temporarily delayed in order to ensure a prototype antenna at the HartRAO.

5.3.2 Antenna optical design study

From the outset, the KAT antenna design studies (both optical and mechanical) were aimed at a 15 m diameter antenna with a scanning focal plane array to cover the desired field of view. For KAT the FOV was set at 50 sq deg or nominally 3.5 deg of scan from the boresight position. The target frequency band was 0.7 to 1.75 GHz with the low frequency limit less critical than the upper limit. At 1.75 GHz a 15m antenna has a nominal 0.8 deg 3 dB beamwidth which means that there must be nearly three beamwidths of scan to achieve the field of view.

The antenna optical analysis concentrated on the focal region fields of 15m diameter prime focus and offset antennas with various focal length to diameter ratios (f/D). The conjugate feed matching technique was used to search for a reflector f/D which requires the feed of smallest diameter to collect the maximum power out of the Airy pattern. Reflectors of small f/D (say 0.35) have relatively poor focusing requiring a larger and more complex feed, rapid gain loss with scan and poor cross polarization. A large f/D (say 0.8 or more) overcomes these problems but requires a large feed leading to increased aperture blockage and problems with supporting the FPA relatively far away from the back supporting structure and elevation axis.

Based on on-axis and off-axis scan analyses, a prime focus reflector with f/D of 0.5 to 0.55 was identified as the optimum. Offset reflectors were considered but to achieve equivalent scan pattern and cross polarization performance the f/D must be increased by around 20%. The offset has the advantage that the blockage is eliminated and the feed spillover points away from the ground. However, costing studies have shown that the offset will be significantly more expensive to build for KAT.

From the optical study, the detailed KAT Antenna Structure Subsystem Specification was developed, which identified the functional, operational and performance requirements. Cost

Final 9 drivers in the requirements were identified and effort is being put into a model to trade off performance requirements against cost.

Among the issues have been the rms surface accuracy and pointing stability of the antennas (including stiffness of the antenna and support of the FPA). Using Ruze’s formula at 1.75 GHz, the random surface error efficiency Esurf for 2 mm rms is 0.979 and at 4 mm rms this reduces to 0.918. For 4 mm at 5.5 GHz Esurf is only 0.428. Ruze’s formula is a little pessimistic and a more detailed analysis can be done once the distribution of errors is known or simulated by finite element analysis. The scanned beams at 1.75 GHz nominally cross at the 3 dB points (ways of reducing this crossover loss are being investigated) which means that observations in adjacent beams are not made at the peaks of the beams but on the sloping sides. A pointing stability of +- 0.01 deg gives amplitude fluctuations of +- 0.2 dB while +- 0.05 deg gives +- 0.8 dB. The aspect of pointing stability and its impact on the interferometer correlations of a large array must be investigated in depth. The KAT antenna designs will focus on investigating the “all-up” price tradeoffs for the 15m antennas with f/D of 0.5.

5.3.3 KAT (traditionally engineered) prototype antenna

A prototype antenna is needed to serve as test bed for the research and development work of the FPA and related digital signal processing and software / computing efforts. Due to time constraints, no significant effort was made to drive down the prototype antenna cost – the focus was rather on getting a prototype antenna as quickly as possible and gathering information with regards to existing South African and international contributors and partners in future processes. o A detail specification based on the KAT user requirement document was prepared and sent to MAN technologies in Germany, requesting a conceptual design and costing. The proposal received from MAN technologies came in at US$3,300/m². o The specification was also sent to 17 potential suppliers (as well as international reviewers) to gather information on existing South African and international industry capability and price (for contracting the prototype). o Finally, the specification was sent to potential Chinese and the Russian suppliers. None quoted costs that represented cost advances in antenna design and manufacture. o A conceptual design was done for the prototype antenna and initial structural analyses performed on it. (See Figure 6). Aluminum and steel was considered for the antenna and only steel for the general structure. Figure 6. Results of initial structural analysis on concept antenna design

Final 10 5.3.4 “Fast track” prototype antenna

Investigations to secure a cheap (or free / donated) antenna with appropriate optical qualities and of appropriate diameter which can be used in technology development activities instead of the previously mentioned properly engineered new prototype antenna have been ongoing.

Negotiations are currently under way with Telkom SA (Pty) Ltd for such an antenna (diameter 7.6m and f/d = 0.35). Should these negotiations be successful, the prototype antenna (5.3.2) will not be manufactured, but the development the “$1,000/m²” antenna and of cost models will be fast-tracked and given the highest priority.

5.3.5 Antenna cost models

We performed desktop studies and confirmed that, for dishes from 6–15m, the cost benchmark within the antenna manufacturing industry is approximately US$3,000/m², independent of antenna size. This reconfirmed our perceptions that it will be unlikely for a traditional antenna design and manufacture company to deliver KAT or SKA antennas at the target cost of US$1,000/m².

We revisited and analysed existing cost models (in particular the SKA cost curves produced by Weinreb and D’Addario (SKA Memorandum #1, July 2001)) and noticed a linear relationship between cost and antenna diameter. We believe that these previous cost curves did not reflect the implications of different manufacturing techniques for different antenna sizes and materials in adequate detail and suspect that the various manufacturing techniques must be reflected in cost models as discontinuities / step functions. (For instance, as the antenna size and weight decreases, it must eventually mean that the manufacturing method changes (e.g. to hydroforming) and it becomes possible to use COTS gears and peripherals. This is not reflected in Weinreb et al’s cost curves.

Caltech commissioned a study by Schultz Associates (see “Status Report - USSKA Antenna Development” of July 2005 on the SKA web site) on the design and costing of the USA SKA antenna. They conceptualised a 12m hydroformed antenna which they claim will work well to 11GHz, with a 2m low frequency netting skirt. Schultz concluded that the price for a 16m antenna (12m plus 4m skirt) will range from US$138,000 (for > 1,000 antennas) to ~ US$200,000 for a small number of antennas. At first glance, this appears to be within the SKA cost goal of ~ $1,000/m². Upon reading the finer detail in this report, it is clear that no allowance for tooling has been made. Also, for quantities smaller than a thousand, Schultz suggests a 60% increase in unit costs and recommends a 15% contingency - the US SKA antenna proposed is therefore approximately US$2,000/m² (excluding tooling costs).

Final 11 5.4 Focal Plane Array

5.4.1 FPA design study

Our FPA design studies have concentrated on the use of Vivaldi elements nested in a crossed regular array to achieve the dual polarization. Figure 7 shows a 4X3X2 dual polarized Vivaldi array implemented on relatively low performance FR4 printed circuit board. The dielectric loss is quite high but the board is suitable for rapid prototyping and experimentation. There are 24 SMA(F) connectors which allow direct RF access to each Vivaldi element.

Figure 7: First iteration 4x3 Vivaldi Array investigated

All elements can be terminated in 50 ohm loads and the mutual coupling between neighbouring elements in the array can be measured.

Another 24 element (4X3X2) Vivaldi array is being developed on a GIL 1000 RF substrate which has superior loss performance compared to FR4. This substrate is also suitable for incorporating a LNA directly in the Vivaldi element feedline, thereby reducing losses, so that the LNA noise figure is not degraded. We plan to use this prototype in our first tests on the prototype antenna.

Feed pattern optimisation in the FPA is crucial to achieve high efficiency beams. A synthesis technique is being developed which will allow the feed pattern to be optimised based on the actual measured Vivaldi FPA patterns. The crossed Vivaldi array has a fixed geometry and the element spacing in wavelengths is frequency dependent. This fixed array geometry and finite number of elements in the array will influence the scanning performance of the FPA installed in the focal plane of the dish. Phase centre movement in the Vivadi array as a function of frequency must be quantified since this must be compensated in the beamforming network

The Vivaldi elements were selected for the FPA since they have been studied extensively for more than 15 years for radar applications. To achieve short-term results the Vivaldis remain the

Final 12 elements of choice since they are the lowest risk. Studies should be conducted to investigate alternative elements to Vivaldis particularly in terms of mutual coupling and cross polarization.

5.4.2 Collaboration with ASTRON and CSIRO

Progress with KAT can be speeded up if there is much more formal collaboration with ASTRON and CSIRO. The KAT project will negotiate to buy a 10X10X2 (with appropriate dummy elements) FPA from ASTRON without an analogue beamformer but optimised for the KAT 15m antenna with f/D of 0.5 and scanning FOV. The KAT would need direct access to the RF outputs of the individual elements in the FPA to interface with the rest of the KAT system. (The negotiations with ASTRON may include LNAs to the individual elements on the Vivaldi PCBs).

In addition to the FPA hardware, we will work with ASTRON on the mutual coupling analysis, Tsys and the weights for the beamformer for off-axis beams. We will negotiate with ASTRON to second an antenna engineer to work with/at ASTRON to ensure that the KAT requirements are met. The development of the dual polarization, fully digital FPA (2-PAD) as part of the SKADS programme will be an important contribution to the development of the KAT. It is not possible at present to specify the timelines for that project, so specific collaboration with the SKADS team will be structured following the confirmation of the UK's PPARC funding.

5.5 Digital signal processing The initial focus for the digital signal processing subsystem was to establish international collaboration and to define the digital beamformer prototype. International visits to Berkeley (Dan Wertheimer) and CSIRO ATNF (NTD team) helped to establish links and identify areas for collaboration. Both groups have well- established teams developing hardware and firmware for high performance radio telescope back-ends and it is essential that KAT work together with these resources to assist in moving SKA forward.

5.5.1 CSIRO / ATNF

We met with the NTD team in early May. It was agreed at the meeting that the KAT and x/NTD teams would work together, in particular in the field of DSP. It was further proposed that CSIRO would take the lead due to the extensive expertise, with the KAT team making available human resources to ensure adequate capacity and knowledge transfer.

5.5.2 Berkeley

We met with Dan Wertheimer from Berkeley Space Science Institute, who leads the team developing the BEE2 hardware, which includes hardware to sample and communicate data back to the BEE2 boards. The set of boards (ADC module, iBOB and BEE2) provide a flexible tool chain to Figure 8: The Berkeley approach Final 13 investigate various radio astronomy problems. In addition to the hardware, Berkeley has developed a set of powerful firmware tools based on the Matlab/Simulink/Systerm Generator environment. These tools allow for fast prototyping of various radio astronomy applications. The group has already developed a 128M bin Spectrometer for Seti measurements and is currently developing an 8-antenna prototype correlator for ATA.

5.5.3 Digital beamformer FPA prototype

The goal of the FPA prototype is to build a small bandwidth (20-40MHz) 100 element (dual polarised) fully digital FPA. The initial design considered using the Berkeley ADC and iBOB boards configured in a ring using the iBOB dual XAUI CX4 links. (This prototype could easily be extended to a full bandwidth system using a 10GBASE switch and BEE2 hardware.) After deliberations with CSIRO (Colin Jacka and John Bunton) it was decided to work closer with CSIRO who were focusing efforts on cost optimized hardware. This is more in line with the SKA objectives. CSIRO will be providing KAT with the prototype 24-element beamformer. The KAT DSP team will work with CSIRO to produce the full xNTD/KAT digital hardware, providing human resources and ensuring knowledge transfer. BEE2 hardware will still be used as an initial thin element prototype to be used as a Figure 9: The CSIRO approach digital backend for HartRAO. In addition the Berkeley firmware design flow will be used as a development environment.

5.6 Computing / Software The KAT team’s main objective since the last progress report to the ISSC was to acquaint ourselves with the scope of the computing effort that will be required for KAT, making contacts, developing collaborative approaches and understanding the real technical issues through a strong systems engineering approach. Of late, the focus has shifted towards gearing up on the operational side so that the new SA KAT computing team is established and rapidly starts contributing at the international SKA level.

5.6.1 Collaboration

We have established links with co-workers in Australia, the USA and the Netherlands. In particular, weekly Skype calls are held with Tim Cornwell as we explore working closely together on the software and computing side of KAT and xNTD. One of the new SA KAT computing team members, Simon Ratcliffe, will be spending 1-2 months at the ATNF from September working with Tim Cornwell on simulation work relating to KAT/xNTD and the Big Gulp concept. In addition, SA is contributing to the SKA Software whitepaper.

The KAT computing and software team consists of both scientists and engineers. Engineers have had to acquaint themselves with the basics of two of the main packages used in radio astronomy (AIPS++ and MIRIAD). Tim Cornwell has provided us with an xNTD simulation script which has

Final 14 been run locally on a recently installed AIPS++ installation and which we are beginning to analyse. To run these simulations in an effective way, we are in the process of purchasing the first of our high performance servers. KAT, based upon the SDFPA approach to the SKA, has many “still to be worked out” challenges. Extensive detailed simulation and algorithm development work is needed to address these challenges, which will be the responsibility of a newly appointed core computing and software team for KAT. The team we recruited consists of 8 engineers and scientists (radio astronomy and other disciplines).

5.6.2 Prototyping

We have started the development of a digital IF processor as a high priority to allow us to get a headstart on wideband DSP systems well before we have a working KAT prototype. The outcome of this project will be used to develop algorithms for digital baseband conversion, pre-correlation RFI excision, pulsar de-dispersion, searching and timing, and polar-spectroscopy. A spin-off is that this system will provide us with a digital BBC for the Mk5 VLBI terminal.

Final 15

Final 16 KAT Team response to EWG Draft Review Report - October 2005

General Comments on the EWG review (from the KAT team) 1. Introduction

In the second paragraph, we agree that we must still develop a model for using Ae/Tsys as an optimisation parameter. This has been done at ASTRON and CSIRO and it seems an unnecessary duplication of effort to do this in South Africa from scratch, hence our desire to formalise collaboration between KAT and these 1 organisations . We are, however, targeting an Ae/Tsys = (0.5 x Parkes). Figure 1, shown later in this document, also list the key specifications of the reference design (on which the costing was based).

We are indeed considering an upgrade path to the reference design. Once the reference design and related science case have been signed off (planned for end of November 2005), the implications of the upgrade path on design, cost, risk and timescales will have to be understood before approval by the KAT and SKA Steering committees. All variations to the reference design will be handled according to agreed procedures. Issues considered for such upgrade path include: ° increasing the upper frequency limit, o An upgrade path to higher frequencies will be possible and will require us to tighten up the rms surface accuracy. 2mm rms or better will have to be achieved and systematic errors eliminated during assembly. The back-up structure may have to be stiffened to accommodate gravitational and wind loading effects. We do not have a cost model which trades surface accuracy etc against price of the surface and backing structure and the pointing accuracy. We are also not aware of any such cost models available from other teams. Guidance will be appreciated. We do, however, plan to investigate this in future. ° increasing the maximum baselines, ° re-configuring the layout, or ° substantially increasing the sensitivity (Ae/Tsys).

We certainly see the KAT-like solution as part of a possible hybrid SKA solution. KAT aims to demonstrate key technological but also procedural innovations critical to the SKA, which includes: ° Adapting classical systems engineering methodologies and approaches and life cycle planning analyses for high risk high technology projects (like KAT and SKA) to ensure cost effective, on time construction of the SKA ° Understanding the key issues relating to project management for SKA ° Demonstrating innovative antenna technology (in particular FPAs for radio astronomy) ° Understanding the computing requirements (software) and key issues with regards to architecture for the digital signal processing components of SKA (use of FPGAs versus innovative cell technology (Blue Gene, new Sony PlayStation technology for example) ° Investigating innovative dish design possibilities to ensure cost reduction ° Understanding and refining cost models on infrastructure (power, telecommunications, roads and other access to the site, etc) for SKA

We will contribute to the science discussions at Pune and will focus on SKA science that will be possible with KAT. We have also introduced discussions with CSIRO and ASTRON to work towards a joint science case for KAT, xNTD and Westerbork.

In paragraph 3, we agree that the timescales are very ambitious. We aim at first light experiments at the end of 2009 (and not in early 2009). The international links must be formalised to have the engineering and science teams working together to solve problems of mutual importance, but more importantly to ensure that deliverables which are dependent on international contributors will be met against the deadlines (we already have instances where partner countries and delivery of hardware from these are on the critical paths of key components for the prototypes and KAT.

1 ASTRON requires that we sign a collaboration agreement before giving us information on recent results. We are addressing this and hope to be able to conclude these formalities at Pune. 2. Comments on “Answers to pro-forma review questions”

Comments to “1”. Your assessment is correct.

Comments to “3”. Our team is now 24 FTEs strong. Industry involvement in the research and development stage has been difficult to get going, but in-kind contributions from a number of companies, especially those focussing on digital signal processing issues, have been steadily increasing. We have also been able to negotiate good agreements with suppliers of high-end specialist software for digital signal processing / hardware design. The likely nature of collaboration on the different components of the system are as shown below: Sub-system Likely nature of Notes contractors Systems ° Industry Also negotiating for off-set contributions from BAE engineering ° Science Council Systems Dishes ° Industry Also negotiating for off-set contributions from Ferrostaal / MAN Technology FPAs ° Industry Also contributions from Universities for non-critical ° ASTRON path items Software ° CSIRO It seems likely that the software for xNTD and KAT will be the same or very similar DSP ° Industry Also possibly Berkeley University ° CSIRO .

Comments to “5”. Your assessment is correct.

Comments to “6”. A dispersed team does indeed present unique management challenges for a multi- disciplinary project. It is not easy to have informal technical discussions or to get an experienced group together quickly to discuss key issues. We have implemented processes and systems for communication and sharing of key information and design documents, etc. Regular telephone conferences are conducted and 2- day team meetings held on average 2x per month. The overall project manager and lead systems engineer are key to the overall interaction and communication on both managerial,, administrative and technical issues. These individuals travel frequently between the two offices.

The required management effort will be significant particularly during the integration phase. We believe that it is critical to understand this issue in more detail, especially since the SKA will most likely include teams from around the globe. The SALT optical telescope recently completed in South Africa dealt with similar issues, from which we can learn and improve. A number of the SALT engineers have joined our team and we have regular discussion sessions with the SALT project manager and systems engineer, who are based in Cape Town.

Comments to “8”. The FPA and beam former with associated digital signal processing is key. To reduce the risks, the international collaboration as mentioned is vital to achieve the compressed timescales. We will address the issues you indicate in more detail in the next update, but would like to stress that the “first light”experiments with KAT (end of 2009) are unlikely to have a fully functionally, fully commissioned system. Full system roll-out will happen during 2010, with regular upgrades thereafter during operation.

Comments to “13”. The all-up per square metre cost of the dishes remains an issue. With aggressive timescales the design will always be conservative, and it is unlikely that significant cost savings on the “traditional” design will be achieved for KAT.

Comments to “14”. A plan B to have a substitute for FPAs2 must be developed. This could be smaller dishes requiring fewer effective beam widths of scan angle or compact cluster feeds capable of achieving acceptable

2 Our most recent work on FPAs delivered a 4x3x2 antipodal dual polarized FPA that we will use for prototyping and testing in an anechoic chamber efficiency and inter-beam spacing. As a team, we have decided to defer this decision until after Pune, where debates on the SD-FPA concepts and alternatives for it, will be discussed.

Comment on “15”. Calibration strategy and implementation as well as imaging are major technical issues, cross polarisation calibration seems to be very important. We are working towards a tighter definition of the scope of work and a detailed project plan for the software components of the project. We will have drafts ready for discussion at Pune.

Comments on “16”. With very tight timescales “traditionally engineered” dishes will be our current choice, but with pressure on potential contractors to drive the price down to USD2,000/m2. Developing a 15m dish for prototyping of our 10x10x2 FPA is on the critical path of the project, which leaves little time for design optimisation and new concept definition and refinement. A parallel process will be run to investigate innovative dish designs for SKA, where both quantity and time scales will determine the options.

Comment on “18”. The evaluators have correctly assessed the short time for instrument characterization, this will be a first and the system is expected to be VERY complex.

Comment on “19”. The evaluators see the collaboration as key, we need to get agreements in place to get the engineering teams together on technical issues and more importantly on how the parties will address the IP and prior-art conundrums. We have engaged with a CSIR (South Africa) Intellectual Property and Commercialisation Unit to advise in this matter.

Comments on “21”. We are identifying industries with the key expertise and are approaching these companies / institutions in a systematic way. A number of key government departments will also be critical for successful infrastructure development. These relationships are being built.

KAT team comments on EWG scoring

Table 1 – Karoo Array Telescope: EWG Assessment Criterion Rating Demonstration of pivotal and/or high-risk technology 5 Demonstration of cost reduction strategies 4 Demonstration of realistic risk management for concept or system 3 Likelihood of completion by end of 2008 2(1) Likelihood of substantial added knowledge by mid-2007 4 Realism of project plan (milestones & timescale vs budget and manpower) 2-3(2) Definition of appropriate milestones (suitable for ISPO monitoring) 2 Security of funding for the project as defined 2(3) Quality of responses to IEMT Oct 2003 supplementary questions Not applicable

Table 1 - Notes 1. The EWG stresses that strong, functional, international links will be vital in ensuring the KAT’s completion on a 2009 timescale. 2. The Group is concerned that some phases of the project appear quite compressed (e.g. XDM, ADM, PPM, KAT in six-month steps) and that instrument commissioning plans are not outlined. 3. A higher score will be possible if current funding negotiations with the RSA Government are successful.

Comments on additional comments and questions by the EWG

General comment: We think the scoring is fair and leaves the opportunity to improve.

1. It would be useful to have a concise specification table and block diagrams for the KAT, including the A/T goals.

The comment on Ae/Tsys goal is valid and as discussed elsewhere we need to work with others who have developed this model (especially ASTRON).We are aiming for a sensitivity target of half of Parkes. We hope to finalise the collaboration agreement with ASTRON at Pune in order to allow them to release information to our team.

Location Physical breakdown & Station array key specifications

Station array: - 20 Antennas - Max baseline 1500m

Dish: - 15m diameter dish - F/D = 0.5 - Surface accuracy at least 4mm (target = 2mm)

Positioner: Front-end receiver - Pointing accuracy at least 0.04deg FPA (target = 0.02deg) - Elevation range = 0 to 95deg

Dish FPA: - 10x10, dual pol Vivaldi elements

Positioner Front-end receiver: - 200 up-converter RF modules - Frequency range 750-1750MHz - Noise temperature = 50K

On-Site (Karoo) Digital receiver - Bandwidth = 250MHz

Digital reciever: Beamformer Downconverter: - 200 channels downconversion

A/D converter: - 200 A/D converters - at least 600Msps - at least 6 bits

Beamformer: Station Station - At least 10 beams. Target = 50. Controller Processor

Station processor: - Peformance specs TBD

Station controller: - Resource scheduling - Station control (e.g. beam position) - Timing distribution - Power distribution Operations centre (Cape Off-site Town?) Operations centre: - Experiment scheduling - HPC facility for data processing - Instrument monitoring - Proposal handling User (Scientist) Remote

2. It would be useful to include a further level of budget detail, perhaps extracted from the detailed budget already tabled in other presentations. A breakdown into non-recurring and implementation costs would be informative.

We are planning a budget review over December. More details will be provided afterwards.

3. The KAT appears to fall within the classification of a wide-field survey instrument and, as such, it is not surprising that signal processing and electronics account for a substantial fraction of the total cost. In addition, all the KAT beam forming is done digitally, adding to the DSP fractional cost. It would be interesting to see a more detailed breakdown of the large DSP investment, especially for readers unfamiliar with the digital beam-forming and related DSP requirements.

The KAT DSP hardware budget was based on preliminary KAT user requirement of: ° 512MHz instantaneous bandwidth, ° 10 simultaneous beams, ° utilizing a 10x10 dual polarized FPA ° with 20 dishes. The initial budget also included provision for two prototype stages, including a full single KAT antenna system as well manpower costs. This response to the EWG report addresses the base cost of the (proposed) final KAT DSP system, to provide some more detail on the system architecture and cost breakdown. The KAT DSP hardware consists of two key components, the digital receiver/beam former and the correlator Most of the system costs lies in the first component (beam former)t. The major reason for this is that one requires a full bandwidth digital receiver for each of the antenna elements. Each receiver is required to digitise the data as well as channelise it into small enough “blocks” to perform beam forming. For our 20 dish reference system there are 4,000 receivers, currently costing approximately USD1,000 per receiver. The beam forming algorithm requires selecting a subset of the antenna elements and then performing complex multiplication and summing to form the beams. The initial beam forming costing was based on the Berkeley BEE2 architecture, which uses commodity switches and BEE2 boards to perform this function. Four BEE2 boards are required to process 10 fully polarised beams at the required bandwidth. For each dish, the cost of the beam former is similar to the cost of the digital receivers. The KAT correlator is expected to be an FX correlator also based on the Berkeley BEE2 architecture (or a hybrid of the CSIRO / Berkeley designs). Since KAT is only a 20 dish system the cost is only USD2million, only 15% of the total USD15million estimated for the rest of the system.

4. What does the ~USD 500 cost per receiver implied by Figure 2 (p4) of the submission cover? Does it include installation, testing and environmental protection?

It is the cost for the installed dishes on the KAT site. It includes all hardware costs (excluding antenna software, software integration and some commissioning) associated with the dishes – including project management, CAD and analyses software and tools as well as travel. Since software is not included but is required for feedback control on the drives/controls there will be some commissioning work that is not covered in the budget for the dishes – most of the software and commissioning procedure should be developed and de-bugged for the dish prototype, but this is something that has not been addressed by the team in enough detail. We undertake to investigate this in more detail. Dishes Budget Breakdown: 15m Prototype (1 off) Array (20 off, on site in the Karoo) Antenna structures USD1.456million USD 8.88 million Project Management USD 0.103million USD 0.298million Sub-Total USD 10.744million Contingency (15%) USD 1.612million TOTAL USD 12.356million

5. An outline of the CORE approach to systems engineering would be of interest.

As per request from Peter Hall, we will be able to supply more information at the Pune conference if needed. Basic information is provided below.

The CORE tool: CORE is one of a number of database products that have recently emerged on the systems engineering tools market.

Database systems engineering tools such as CORE are much more powerful than traditional paper-based approaches: • Multi-dimensional design viewing: The database structure allows for a multi-dimensional view on the same design data (as shown in the figure below). This allows the design of a system from multiple points of view, switching between the views as required and also checking for consistency and completeness from different points of view. • Multi-dimensional mapping between design parameters (e.g. data X flows over interface Y in system mode Z) becomes much easier in a database tool. • Traceability: Tracing design decisions through the design process (from user requirements to system design features to subsystem design features, etc.) becomes much easier with a database tool. • Communication: Many of the design data is presented in graphical format which is easy to understand and communicate.

Functional flow view

Physical decomposition Intefaces view view CORE System design database Performance Test & integration allocation view design view

Data & control flow view

KAT CORE-based systems engineering approach: An overview of the CORE-based systems engineering approach is shown in the figure below. The diagram shows a simplified process diagram (not necessarily sequentially executed) and the CORE models that are developed along the process. Each CORE model is a multi-dimensional design representation. The diagram also shows the traceability of design decisions.

CORE models Define how each sub- Traceability Design system architecture system will be tested. - functional flow - functional verification - data flow - performance verification - control flow - interface verification - physical decomposition - interfaces - performance budgets Subsystem 1 Subsystem 1 requirements test plan

System System System Subsystem 2 Subsystem 2 integration & requirements design requirements test plan test plan

Subsystem 3 Subsystem 3 Capture system black box requirements test plan requirements: - System use cases - User inputs & outputs Allocate functions & Define integration - External interfaces plan and how sub- - Measures of effectivenss performance requirements to subsystems assemblies and system will be tested.

Comments to 6, 8, 9 and 10. The all-up per square metre cost has been an issue for several months and until we get firm quotes from manufacturers and build a reliable cost model it will remain so. Mesh dishes will certainly limit the expansion to higher frequencies. Dirk Baker is currently writing a technical note on the mesh size vs frequency and the surface leakage efficiency. To date, we have seen no cost model which says that a mesh surface will reduce the surface cost by X% and the all-up cost by Y%. Unless these savings are substantial (around 20% or more relative to a solid dish) there seems to be very little merit in pursuing this. Some solid dish concepts use the dish surface as part of the structural elements and by so doing reduce the mass and complexity of the back structure (the Indian dish is also an example). Open, light weight meshes need to go on a back structure which has the correct reflector profile or adjustments and this may negate the perceived price advantage. In all of our discussions to date, it was assumed that the USD1000/m2 was related to physical area π ( × D2 ). Your comments brought to our attention that perhaps the USD1,000/m2 relates to effective area – 4 assuming an aperture efficiency of 0.7, it means that the target is USD700/m2 for physical area. We are not sure where your USD270k figure comes from – working with USD1000/m^2 for physical area, we calculate the cost for a 12 meter antenna as USD113k and for a 15 meter antenna it as USD177k.

7. The selected f/D range of ~0.5 appears to be a good choice based on work presented at the June 2005 FPA workshop in Dwingeloo. With this f/D what is the potential of a multi-feed array (one feed per pixel)? Could such a system form a viable risk mitigation option if the FPA approach proves technically or economically unfeasible?

The F/D of 0.5 gives a larger focal spot and larger displacement per off-axis beam than smaller F/Ds and a cluster feed may be possible as a Plan B. Such compact dual polarised feeds will have to be investigated.

8. It would be useful to ensure that Patriot Inc (of Albion, Michigan) is included in the list of suppliers providing cost estimates.

We have contacted them and have sent them the specification - waiting for a response.

9. On p11, the cost benchmark figure of USD 3000/ m2 implies that the cost is proportional to diameter squared. The middle paragraph on p11 states that in SKA Memo1 the antenna cost is linear with antenna diameter. This is incorrect - the memo assumes an antenna cost of USD 100 x D3 in large quantities for an antenna good to 25 GHz. This is USD 337k for a 15 m antenna or USD 173k for a 12m antenna

The budget was developed in detail for both the 12 and 15 meter dish. No relationship was assumed. The resulting cost/m2 just happen to come out more or less the same (difference less than errors in budget due to estimation). Obviously some items will increase by more than D2 (pedestal for example) and some less (software, project management).

10. On p11, the point in the Schultz cost estimate about no allowance for tooling costs is over- emphasized. The estimate does not include tooling costs but this is small, of the order of USD 5 million or USD 5k per antenna in a quantity of 1000. Note then that the Schultz estimate, based upon a physical design with material weights, structural analysis, and a cost breakdown per component, is USD 200k for a 16 m antenna good to 1.7 GHz and a 12 m antenna good to 34 GHz, in “quantities smaller than a thousand”. The KAT project might consider this design and cost estimate carefully.

We believe that tooling costs is not negligible for an array of 20 though – USD5million is a very big portion of the dishes budget for KAT. Also, the manufacturing technology used for 1,000’s (as opposed to 20) dishes is unlikely to be the same. If we compare our antenna cost to that of Robert Schultz for similar amounts of dishes and no tooling, the cost is comparable. We will however be revisiting this during our budget review in December 2005.We are also currently discussing the risk reduction for the full KAT system when using polar mounts and not Alt-Az. Polar mounts, of course, have additional cost implications.

11. On p11, the (cheap or free) 7.6 m f/D = 0.35 antenna may not be worth the labor effort to install and experiment with it. The FPA performance depends strongly on f/D and it may be better to either accelerate the production of a KAT traditionally-engineered prototype or, perhaps more simply, to buy a small antenna with the same f/D as the final antenna.

Agreed, ideally we want the same F/D as for KAT. The Telkom dish pedestals can almost certainly handle a dish up to 10m if we can lay our hands on one. The Telkom dishes have application on other collaborative projects (non-SKA) we are currently setting up and if we can piggy back on this a F/D of 0.35 will still teach us a lot about integration in what is quite a friendly environment. It is still not clear whether we will in fact be able to secure one or more of these dishes.