US MMA/LSA Science Committee Report on Antenna Diameter Frazer Owen, Jack Welch, John Carlstrom, Jean Turner, Neal Evans, Geoff Blake, and Bryan Butler On September 24th, the US MMA/LSA science committee met in Socorro to consider its position on the antenna diameter(s) and numbers for a combined MMA/LSA. It considered the strawman proposals of 1) 40 8m's and 30 15m's and 2) 50 12m's which had been described to the community. It also considered a new European proposal for a hetrogeneous array of 20 8m's and 50 15m's. It concluded unanimously, that it did not like the 20/50 proposal. While it clearly is possible to image with a heterogeneous array, the committee did not like the additional complexity of the imaging process. We were impressed by the simulations which suggest that the fidelity of a mosaic is dominated by pointing, supposedly worse for bigger antennas. We also noted the BIMA experience with mosaicking real sources where dynamic range, limited by pointing, has proven to be a major problem. In addition, we note that mosaicking observations are becoming more and more important to US science, making up 30% of the observations at BIMA in the last year and growing at 10% a year. Furthermore, we think the short spacing problem, described in the European proposal, was not adequately solved by adding 20 8m antennas. Finally we would prefer to avoid paying the one-time development and tooling costs for two different antenna sizes. The committee was split on the merits of the two strawman proposals; however, it could agree, unanimously, that it liked the idea of a homogeneous array of antennas no bigger than 10m's best. Nonetheless, it wanted to wait for the MMA/LSA antenna committee to meet to decide on a detailed proposal. We did not consider the more recent idea from Dennis Downes of two homogeneous arrays of 40 8m's and 40 15m's, although we have discussed it in email since the meeting. On September 30 - October 2, the MMA/LSA antenna committee met in Socorro to converge on antenna cost curves and performance specifications as a function of diameter. Four designs were presented and costed for 8m, 10m, 12m, and 15m telescopes. All antennas pointed to 1/30 of a beamwidth or better at 300 GHz (ignoring anomalous refraction) had surface accuracies of 25 microns or better and could move between sources at a degree/second or better. However, as the antenna size increased the designs had to make compromises in other antenna performance to reach these goals. The 8m design has an elevation limit of 0 degrees, has the feed legs connected to the edge of the dish for minimum blockage/spillover, and has a surface error of 16 microns above 25 degrees elevation. Also by relaxing one or more of these constraints one could do better in other telescope parameters, like pointing and cost. As the antenna size increased, the designers had to make more compromises to achieve the design goals. By 12m the design was forced to a 25 micron surface, a 15 degree elevation limit and increased blockage/ spillover by moving the feedlegs inward to achieve the pointing specification passively. By this diameter all of the flexibility of the smaller designs had been used up to meet the pointing spec. At 15m, one requires active pointing to meet the design goals. The antenna requires the same 15 degree elevation limit, a 25 micron surface and increased blockage/spillover from mounting the feedlegs in the interior of the dish structure. The pointing system described seemed to the committee somewhat risky, although it was judged that some improvement in pointing from such a system was likely. The telescope is designed to make use of the active pointing and thus used a foundation too light to achieve the pointing requirement passively and also would require an extremely precise azimuth track as part of each foundation to point well passively. Thus cost and technical detail argue for active pointing for a 15m which was judged to be risky. All the designs seemed likely to benefit from an active pointing system but the US committee is reluctant to depend on this unproven technology for the MMA/LSA to work well. Our decision argues either for an array of smaller antennas or an array which at least contains enough small telescopes to carry out the high fidelity and submillimeter work if the bigger antennas do not perform well. After looking at the cost curves, we believe the best solution for MMA/LSA is a homogeneous array of antennas not bigger than 10m's in diameter. We propose to build as many of these antennas as will fit into the total budget without giving up the other other performance goals we have set for the MMA (frequency coverage, bandwidth etc). Assuming a total of $400M and the cost curves for the other parts of the array extrapolated from the MMA equations, two interesting examples can be given which the US committee could agree to: 1) 128 8m antennas (at 1.05M$ per antenna) 6434 sq meters 2) 90 10m antennas (at 1.9M$ per antenna) 7069 sq meters The committee is split on which of these it would prefer but the committee can live with this range of options. If the European committee agrees, we would like them to indicate which one they prefer. Based on extensive discussions with the NRAO correlator engineers, we conclude that the correlator is not a technical problem for arrays of this size. There are several reasons for our conclusion in favor of 10m or smaller antennas. 1) The antenna requirements for high quality imaging and acceptable submillimeter performance become more and more difficult to achieve as the diameter increases. With diameters of 10m or less, we should have the ability to fine tune the design to achieve the best compromise between pointing, surface accuracy, elevation limit, and blockage/spillover to reach our overall goals. Above 10m in diameter, we are forced to an unacceptable set of parameters with almost no room to optimize performance unless we assume active pointing. We want a high quality antenna which meets the minimum requirements passively but which might be improved by an active pointing system. 2) As pointed out in the European proposal, the larger the antenna the longer the baselines in the smallest array. This means that, for example, an array of 15m's is only sensitive to subarcsecond structure in the submillimeter. We want to minimize this problem by having dishes as small as possible. Even so, mosaicking and total power observations will be important for many of the observations we are interested in. We expect these same small dishes will make these observations easier by pointing more accurately than the larger antennas (all things being equal). 3) These designs deliver almost as much collecting area as the arrays of larger antennas, which will probably suffer losses from pointing, surface accuracy and blockage/spillover which will offset the small gains in total area under many circumstances. For example, 64 12m's would be 7,238 sq meters without taking into account these losses. 4) The 8m to 10m designs allow the US (or Europe) a reasonable fallback position if the European (or US) money does not materialize: 40 8m's or 36 10m's would fit in 200M$ (which we believe is the limit for a US project). A larger antenna would produce a smaller number of antennas than we would be happy with and an array of antennas too big for much of our science. 5) Anomalous refraction becomes more of a problem for pointing as one goes to bigger antenna sizes. For example, for median conditions on the Chajnantor site at 50 degrees elevation (supposedly a typical elevation), the anomalous refraction rms pointing error is 0.55 arcsec for a 12m. These rapidly changing pointing errors should not be too important for mosaicking but mean that, including the other errors, the 12m design will only point to about 1/7 of a beam rms at 850 microns which we consider unacceptable. Even at 10m, these errors will make it difficult to meet our pointing spec. (see MMA memo 186).