The : Concept for a 30 m GSMT Roger Angel, Warren Davison, Keith Hege, Phil Hinz, Buddy Martin, Steve Miller, Jose Sasian & Neville Woolf University of Arizona 1
The : combining the best of filled aperture and interferometric imaging Perhaps because of the compromises made in existing interferometers, and because none is yet operational and producing scientific results, plans for the next generation of bigger telescopes have so far focused on round, single aperture telescopes corrected with adaptive optics, such as California s 30 m CELT and Europe s 100 m OWL. What would be ideal, though, is a telescope that combines the best features of interferometry and filled-aperture, diffraction-limited imaging. The is aimed at this ideal. It has two 21 m telescopes on a 100 m diameter track. They can be held stationary and operated independently as alt-azimuth telescopes when desired, with correction to the (21 m) diffraction limit made with adaptive secondary mirrors. When higher resolution is needed, the telescopes are moved to the desired spacing and oriented so the baseline is normal to the source. The local azimuth motion is then stopped, and tracking in azimuth is taken over by moving both telescopes around the 100 m track, with the light interfered at a moving station in between. In this way the baseline remains constant in length and perpendicular to the source. The residual effect of Earth s rotation is to turn the baseline about the line of sight, as needed to sample the u-v plane. For wide-field imaging the beams will be combined directly as a Fizeau interferometer at a moving station midway between the telescopes. For the direct imaging and spectroscopy of extra-solar planets, the beams are combined in a Bracewell nulling interferometer. 2
21 m telescope design Short focal length primaries are used for the individual 21 m telescopes. The resulting compact structure minimizes wind buffeting, simplifies control of track motion and reduces enclosure cost. An f/0.7 21 m telescope will fit inside a Keck-sized dome 3
21 m telescope design: optical implementation 21 m primary mirror made from seven 8.4 m segments 2.1 m adaptive secondary is similarly segmented at 1/10 scale The 21 m primary mirror will be made from just seven large segments, so as to minimize contrast-reducing scatter from gaps and phase steps at mirror boundaries. There will be a central hexagonal segment surrounded by six petals. In this way we can take advantage of the established technology developed at the Mirror Lab for the 6.5 m Magellan and MMT mirrors, and the two 8.4 m LBT mirrors. These mirrors are of honeycomb sandwich construction, nearly a meter thick and extremely stiff. They also are ventilated for short thermal time constant, in order to eliminate mirror seeing. The supports and thermal stabilization will be essentially identical to those in the LBT. 4
21 m telescope design: diffraction at the scalloped primary edge The diffraction pattern corresponding to the scalloped edge of the 7 petal design is only slightly more extended than for a circular boundary. Diffraction patterns are shown in normal gray scale (left) and clipped at 1% and 0.01% of the central peak. The bottom sequence is for modest apodization, with smooth shading over 2 m of the outer and inner boundaries. λ = 760 nm. 5
21 m telescope design: apertures and PSFs 52 m (closest spacing with dome enclosures) 0.1 (at 760 nm) Central fringe with FWHM 2.5 mas contains 50% of the total energy 6
21 m telescope design: PSF with apodization showing effect of residual wavefront errors PSF in blue is cut across apodized PSF in previous slide, and assumes perfect optics. Residual wavefront errors after adaptive optics correction, taken to be 20 nm rms on scale of 10 cm, cause degradation shown by red curve. Such PSF is expected for 8th magnitude stars. PSF 1.E+00 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 1.E-06 20 nm rms wavefront perfect wavefront λ=760 nm Residual flux is dominated by diffraction out to 0.1 arcsec radius, where it is 10-5 of the peak. At larger radii it is dominated by scattering, 5x10-7 at 0.5 arcsec radius. A Jupiter in a solar system twin at 10 pc (0.5 arcsec and 10-9 of the star) would be detectable against background photon noise in 5 minutes with single telescope. 1.E-07 1.E-08 1.E-09 1.E-10-1 -0.8-0.6-0.4-0.2 0 0.2 0.4 0.6 0.8 1 displacement (arcsec) Pupil apodized with 2 m cosine band. Wavefront error has 10 cm coherence length. 7
Primary mirror manufacture Constraints 1 At f/0.7 it is extremely aspheric. 2. At 21 m it cannot practically be made and transported as one piece of glass. 3. Segment boundaries are undesirable, and should ideally be minimized by using the smallest number of large segments. Solution 1. Requires very little new mirror technology. Use the method developed over 20 years to cast, polish and support big, rigid, highly aspheric mirrors. These are now demonstrated to have very high quality. 2. Seven 8.4 m segments when fitted together make a 21 m telescope 3. The proven method and facilities are ideally suited to making 8 m off-axis segments of an f/0.7 parent. The stressed-lap polishing system now in use can easily handle the 20% difference in curvature in orthogonal directions. 8
Primary mirror manufacture: 20/20 segments compared to LBT mirrors The segments are the same size as the 8.4 m mirrors already cast at the Mirror Lab for the LBT. 20/20 Outer petal Inner hexagon 9
Primary mirror manufacture: Adobe Photo Shop makes it easy! An LBT mirror blank, as cast, with 20/20 segment boundary marked with tape. LBT mirror photo modified to show a 20/20 segment. The segment may be cast as shown and figured. Alternatively, it may be cast and figured while still circular, and the 3 edges then removed by water jet cutter. The figure would be refinished as required to take care of any warping from strain release. 10
Primary mirror manufacture: figuring the off-axis aspheric segments The off-axis, aspheric polishing will be made with the same stressed lap tool used to make all large mirrors in the Mirror Lab. In operation the lap is continuously distorted as an off-axis segment, to always fit the parent no matter how far off axis it is moved by the polishing motion. The method is thus ideally suited to figuring rigid, off-axis segments. Quantitatively, the amount of lap bending required to accommodate the 20/20 segments is within the range already used to figure the shorter radius Magellan and MMT 6.5 m mirrors. Representation of Magellan I 6.5 m mirror as 633 nm interferogram, and image from telescope. 11
Primary mirror manufacture: measurement of off-axis segments Accurate testing is the key to figuring the off-axis segments Center of curvature test shown was devised by Jose Sasian of UA Optical Sciences Center Detail of lens and mirror null configuration. All surfaces spherical. Wavefront error in single pass of Sasian null lens. Vertical scale, +/- 5 waves @ 633 nm. Residue would be corrected by combination of CGH and data reduction. 12
Primary mirror manufacture: Sasian test fits in the Mirror Lab test tower (sort of) * existing tower shown in red * requires 2.5 m fold flat at the top 13
Adaptive secondary manufacture Each 0.8 m segment will be made of 2 mm thick glass with 800 voice-coil supports, very similar to the adaptive secondaries for the 6.5 m MMT and the LBT. The difference is that the reference body underlying the 7 segments will be a single, rigid, petal-edged piece of glass 2.3 m in diameter. 64 cm adaptive secondary with 336 actuators for the MMT under test in Arizona. The mirror is not yet silvered, and the magnet actuators are visible through the 2 mm facesheet. The measured response time is less than 1 msec 14
Telescope mechanical design: individual telescope mounts A preliminary design for the steel supporting structure has been made by Warren Davison. It is based on the LBT structure, an alt-azimuth concept pioneered by Davison in which the optical assembly is supported by two large diameter (14 m) wheels that turn on hydrostatic bearings to provide the elevation motion. The azimuth motion is provided by a simple, squat carriage that transmits the load from the rings directly to a 14 m diameter azimuth track. LBT and drawn to the same scale The two LBT 8.4 m mirrors mirrors and their cells, which include flotation support and thermal systems, weigh 100 tons; the entire moving mass including all other optics and instruments is 400 tons. The optical assembly of a will be similarly supported by two elevation wheels, in fact the same 14 m diameter suffices. 15
Telescope mechanical design: exploded view of 21 m telescope 16
Telescope mechanical structure: lowest 3 modes Mode 1-5.3 Hz primarily cross axis motion of the secondary Mode 2-6.2 Hz primarily elevation motion of the secondary Mode 3-7.3 Hz. primarily the elevation structure rocking across the elevation axis 17
Interferometer and enclosure design: enclosures and beam combining station The cryogenic beamcombiner moves in sync. with the two telescopes, holding a position midway between. We show it in brown mounted on a 50 m radial beam, shown in blue. The telescope enclosures could be moved on a separate track (brown) or they could remain fixed with the telescopes operating in the open or with separately driven wind shields. 18
Interferometer and enclosure design: enclosed 20/20 compared to the Keck Interferometer Telescopes drawn to same scale Domes about the same size (40 m diameter) 21 m telescopes (4.4 x larger area) fit inside because of faster focal ratio ( f/0.7 vs f/1.75) s set at Keck separation Keck Interferometer with fixed 85 m spacing 19
The big question for segment fabrication Cast and figure mirror as half-hex, half circle, OR cast and figure round mirror then cut to half-hex, half circle? Reasons to cast and figure round mirror Figuring process is mature for round apertures. Minimize local figuring around irregular edges. Reasons to cast and figure mirror in final shape Avoid risky cutting operation. Mirror will almost certainly require some local figuring after cutting, so why not do it from the start? An orbital spindle with small stressed lap or passive laps should be able to control figure near irregular edges. 20
Demonstration of off-axis figuring and testing Goals: Demonstrate that polishing system works equally well with asymmetric aspheric surfaces. Develop methods of testing off-axis segments and aligning them to each other. Plan: Make a 1.8 m off-axis segment of a 7 m f/0.7 parent paraboloid. Use an existing 2 m stressed-lap polishing machine and 1.8 m lightweight mirror blank. The 30 cm diameter polishing tool easily bends to follow the mirror s 4 mm peak-to-valley aspheric departure. Use new testing and alignment techniques developed by Jim Burge and based on CGHs. Verify accuracy with full-aperture autocollimation flat. 21
Manufacturing schedule for mirror segments Existing Mirror Lab, with new polishing machine to be installed early 2003, can produce 8.4 m segments at 9 month intervals, with first segment complete 3 years after start of funding. Entire set of 15 mirrors (including one spare) would require 13-14 years. It would be better to divide the manufacture between two facilities, using a second existing facility or a duplicate Mirror Lab. We hope to produce a demonstration 8.4 m segment for 20/20 starting in 2004. This would accelerate the manufacturing schedule. 22