Figure actuator MMT unit at the Steward Observatory for laboratory tests (left) and at the 6.5m MMT telescope during the June 2002 run. ffl the

Transcription

1 Adaptive secondary mirrors for the Large Binocular Telescope A. Riccardi a,g.brusa a,d, P. Salinari a, D. Gallieni b, R. Biasi c,m.andrighettoni c,h.m. Martin d a Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, Firenze, Italy e ADS International s.r.l., Corso Promessi Sposi 23/d, Lecco, Italy c Microgate s.r.l., Via Kravogl 8, Bolzano, Italy d Center for Astronomical Adaptive Optics - University of Arizona, 933 North Cherry Avenue, Tucson (AZ), U.S.A. ABSTRACT The two adaptive secondary (AS) mirrors for LBT (LBT672) represent the new generation of the AS technology. Their design is based on the experience earned during the extensive tests of the previous generation unit (the MMT AS mirror). Both the mechanics and the electronics have been revised, improving the stability, reliability, maintenance and computational power of the system. The deformable mirror of each unit consists of a 1.6mmthick Zerodur shell having a diameter of 911mm. The front surface is concave to match the Gregorian design of the telescope. Its figure is controlled by 672 electro-magnetic force actuators that are supported and cooled by an aluminum plate. The actuator forces are controlled using a combination of feed-forward and de-centralized closed loop compensation, thanks to the feedback signals from the 672 co-located capacitive position sensors. The surface reference for the capacitive sensors is a 50mm-thick Zerodur shell faced to the back surface of the thin mirror and rigidly connected to the support plate of the actuators. Digital real-time control and unit monitoring is obtained using new custom-made on-board electronics based on new generation 32bit floating-point DSPs. The total computational power (121 Gflop/s) of the LBT672 units allows using the control electronics as wave-front computer without any reduction of the actuator control capability. We report the details of the new features introduced in the LBT672 design and the preliminary laboratory results obtained on a prototype used to test them. Finally the facility in Arcetri to test the final LBT672 units is presented. Keywords: Adaptive optics, wave-front correctors, adaptive secondary mirrors 1. INTRODUCTION The 2x8.4m Large Binocular Telescope 1, 2 (LBT) has two projects of adaptive optics (AO) in advanced phase of design and prototyping. One is the first-light (mid 2004) system, 3 that will provide two units for an independent single-conjugate correction for each 8.4m primary mirror. The other is an interferometric multi-conjugate AO system called NIRVANA 4 that will be operative for the interferometric first-light (early 2006). Both systems rely on adaptive secondary (AS) mirrors (one for each 8.4m primary dish) as wave-front corrector of the atmospheric turbulence and wind-shaking effects. The possibility to use the secondary mirror as deformable mirror (DM) for an AO system has several advantages, especially in our practical implementation with electromagnetic force actuators and capacitive position sensors 5 : ffl relay optics is no more needed to re-image the pupil on the DM. It avoids the introduction of extra warm reflections that reduce the sensitivity of the science instrument in the infrared 6 ; ffl the same DM can serve all the focal stations of the telescope; Further author information: (Send correspondence to A.R.) A.R.: riccardi@arcetri.astro.it

2 Figure actuator MMT unit at the Steward Observatory for laboratory tests (left) and at the 6.5m MMT telescope during the June 2002 run. ffl the large stroke of the electromagnetic actuators (ο 0:1 mm) allows to use the AS not only as high-order corrector, but also as low-order (atmospheric and wind-shaking) corrector with chopping capabilities 7 ; ffl the electromagnetic actuators do not suffer from hysteresis. position sensors permits open loop operations; In particular the presence of capacitive ffl the electromagnetic actuators do not have any physical contact with the deformable shell, simplifying the replacement of a dead actuator. Moreover, in case of failure while the AS is operative, the actuator can be deactivated without producing any local warping of the shell. The two 672-actuator AS mirrors for LBT (LBT672) represent the new generation of the current AS technology. Their design is based on the experience gained during the extensive tests in lab (Fig. 1a) 8, 9 and at the telescope (Fig. 1) 7, 10 of the previous generation unit: the 336-actuator AS mirror for the 6.5m conversion of the MMT. 11 Both the mechanics and the electronics have been revised, improving performances, stability, reliability, maintenance and computational power of the system. Figure 2. a) on left, P1 prototype; b) on right, design of the P45 prototype.

3 Figure 3. a) on the left and b) on the center: exploded and 3-D view of the 672-actuator adaptive secondary unit for LBT. The diameter of the thin mirror is 911mm. See the text for explanation. c) on the right: actuator design. The partners involved in the project are almost the same team of companies and institutes of the MMT336 project. Microgate (Italy) is in charge of the design and production of the electronics and the related microcode. ADS International (Italy) manages the final mechanical design, mechanical production and assembly. Mirror Lab (Steward Observatory, AZ, USA) produces all the optical elements of the units (thin shells and reference plates). Osservatorio di Arcetri (Italy) contributes to the conceptual design and is in charge for the macrocoding, the laboratory characterization and the optical test of the AS units. The LBT672 units are now in advanced 12, 13 phase of design. The large number of modifications introduced in the LBT672 design with respect to MMT336 requires the test of electronic and mechanical subsystems before starting the production of the final units. Fig. 2a shows the single actuator prototype (P1) that has been successfully used to tests in laboratory the new mechanical and electronic solutions. Fig. 2b shows the prototype (P45) that is under construction in order to test the details of the mechanics and electronics at system level. It has 45 actuators (the first three actuator rings of the LBT672 unit) with a 240mm-diameter and 1.6-thick spherical shell. P45 will be operative starting from October The assembly of the final units will start at the end of the test of the P45 prototype (spring 2003). From late 2003 thru spring 2004 the final units will at Osservatorio di Arcetri for the optical tests. The units will be at the telescope at mid A brief description of LBT672 is reported in Sect. 2. Sect. 3, 4 and 5 report more details about mechanics, electronics, and control strategy, emphasizing the differences with respect to the MMT336 unit. Finally, Sect. 6 briefly describes the facility at Osservatorio di Arcetri that will be used for the thermal and optical tests of the LBT AS units. 2. LBT672 ADAPTIVE SECONDARY UNIT In Fig. 3 the 3-D views of LBT672 show the main six components of the AS unit. From the top (or back) to the bottom (or front) we have:

4 Figure 4. a) top-right. Actuator cold-finger (108mm long) with the space for winding the coil on the left tip. b) top-left. A winded coil head. Diameter 11 mm. c) bottom. Capacitive sensor board (95mm long). It is placed on the actuator cold-finger. On the left there are two contacts for the input capacitive signal. 1. an intermediate flange bolted to the M2 mobile hexapod of the telescope that provides a mechanical interface to the unit; 2. three cooled boxes for the electronics. Each box contains 28 DSP boards with 4 DSPs each for the control and the diagnostics of the 672 electromagnetic force actuators. The boards inside each cooled box are organized in two internal crates (14 board per crate). Each crate has an independent bus and contains also a communication and a reference signal generation board. All the communication boards are in dasychain and the reference signal boards refers to the same digital signal to assure synchronization. The total amount of DSPs is 336 for a total computation power of 121 Gflop/s in 32-bit floating-point arithmetic. Each DSP manages two actuators. The huge computational power allows to use the AS electronics as real-time wave-front reconstructor in the LBT AO system; 3. an aluminum plate (cold plate) that provides support and cooling for the actuators. This plate is connected to the intermediate flange via a fixed hexapod. The cooling distribution is provided thru copper pipes inside internal grooves of the cold plate; 4. the 672 electromagnetic force actuators. A coil is placed on the aluminum cold finger tips (Fig. 4a and b) that are faced to the corresponding magnets bonded on the back of the thin mirror. On each actuator there is a board providing the contacts to pick the capacitive sensor signal and the related pre-amplification and de-modulation electronics (shown in Fig. 4c). The analog signals are converted to digital at a 80 khz rate on the DSP boards; 5. a thick (50 mm) Zerodur glass plate (reference plate) with bored holes, attached to the cold plate through a second fixed hexapod and a set of three astatic lever. This plate is used as a position reference for the thin deformable mirror. The coil cold-fingers, supported by the cold plate, pass through the bored holes on the reference plate to reach the deformable mirror. The mirror position is sensed using the co-located capacitive sensors; 6. a deformable Zerodur concave shell (thin mirror) of 911 mm diameter and 1.6 mm thickness with 672 magnets glued to its back surface. The front surface is ellipsoidal to match the Gregorian optical design of LBT. The back surface is spherical to match the front surface of the reference plate. This shell has a central hole to which a central membrane is attached to provide lateral and in-plane rotational constraint. When the mirror is not active its axial constraint is provided by a set of stops located at the inner and outer edge of the mirror. Tab. 1 summarizes the main differences between the LBT and MMT AS units.

5 Table 1. Comparison between the MMT336 and LBT672 units. For the power dissipation and fitting error calculation, 14, 15 we supposed a r0 = 15 cm at =0:5 μm seeing case scaling the results from papers by DelVecchio et al.. The dissipation at actuator level considers also N/act for the flattening and 78 mw dissipated by the capacitive sensor board for LBT. The corresponding values for MMT are N/act and 64 mw respectively. units MMT336 LBT672 Shell diameter [mm] Shell thickness [mm] Shell shape Convex/Hyperboloid Concave/Ellipsoid Shell+magnet mass [kg] No. actuators Optical compression Ave. act. spacing on M1 [mm] Act. accessibility Low. Act bolted, High. Act clamped, inserted from back inserted from front Total mass [kg] No. of act. per DSP 2 2 Total no. of DSP DSP model ADSP-2181 ADSP DSP architecture 16bit integer 32bit floating-point 1xALU, 16bit DMA 2xALU, 32bit DMA DSP comp. power [Mop/s] Total DSP comp. power [Gop/s] On-board WF real-time recontructor No Yes Communication link Fiber Link 2x2Gbit Fiber Channel Effective transfer rate [Mbit/s] 160 (Full duplex) 2900 (Full duplex) Separated diagnostic link No Gbit Ethernet Effective transfer rate [Mbit/s] 400 Capacitive sensor ref. signal [khz] Electronic damping Possibly (digital) Yes (Analog and digital) Position digital sampling [khz] Cap sensor BW [khz](-3db) Coil current driver BW [khz](-3db) (possibly 74) Actuator efficiency [N/ p W] Power dissipated in the crates [W/act] Power dissipated at actuator level [W/act] Fitting error WF [nm rms] 72 64

6 3. NEW SOLUTIONS FOR THE LBT672 MECHANICS AND ELECTRO-MECHANICS The mechanical and electro-mechanical design of the MMT336 unit has shown several weak points: ffl leaks from the cold plate due to corrosion of the aluminum surface at the interface with the glue layer (used as sealant of the cooling grooves) and some micro cracks of the glue layer itself; ffl difficulty in having access to the central actuators because of the crowd of the cables connecting the actuators to the crates; ffl variability in the value of the contact resistance of the capacitive sensor contacts causing occasional jumps in the capacitive sensing reading; ffl contamination of the gap between the thin shell and the reference plate due to particle of dust falling from the upper part of the unit, preventing to set the gap to the nominal thickness of 50 μm. One of the major efforts in the LBT672 mechanical design has been devoted in the modifications aimed at increasing robustness, reliability, accessibility of actuators, simplification of the cabling and maintenance. The details of the improving solutions are shown in the work of Gallieni et al. 12 The design of the cold plate has been modified in order to use safer copper pipes running inside the aluminum cold plate instead of carving the cooling lines directly in the cold plate and sealing them with glue. The consequence is a lower efficiency in the heat removal that will be quantified on the P45 prototype. In order to avoid cable crowd we decided to collect power and capacitive sensor signal lines of each DSP board (8 actuators) in a single cable. Each one of those cables connects one DSP board to a distribution board that is bolted on the back of the support plate of the actuators (i.e. the cold plate). Tracks on the boards bring signals and power to the corresponding actuators. The LBT672 actuators are inserted from the front of the reference plate together with its capacitive sensor board thru the bare holes in the glass, and then they are clamped at the cold plate level. The capacitive sensor board is bolted on the coil side and it is firmly inserted at the cold plate side in a connector on the distribution board (see Fig. 3c). This solution avoids any movement of the board and the corresponding capacitive sensor contacts, decouples the actuator from its cable and allows the same accessibility to all the actuators (from the front). The only drawback is the necessity to remove the thin shell to access the actuators. It can be done at the telescope using a tool similar to the shell extractor" that has been tested for MMT336 during the telescope run. 10 Simulations and test on the P1 prototype showed that the response of the capacitive sensor electronics has a sensitivity on the contact resistance variation of about 0.03 nm/ohm. Disomogeneity of contact resistivity and contact wandering are the cause of the occasional jumps in the capacitive sensor reading of some MMT336 actuators. We addressed the contact resistivity problem substituting the MMT336 chromium vacuum deposition (» 100 nm) with an electrolytic deposition of a thick(ο μm) layer of gold. A layer of chemically deposited silver is used as interface between the glass and the gold to assure gold adhesion. The test on the P1 prototype showed a uniform contact resistivity always below 1 Ohm, even when the contacts move. In the MMT336 case the contact resistance is 1-3 kohm, with occasional variation of contact resistivity of hundreds of Ohms during contact wandering. 4. NEW SOLUTIONS FOR THE LBT672 ELECTRONICS The LBT672 electronics design takes advantage of the more recent technological developments with respect to MMT336. A detailed description of the LBT672 electronics can be found in Biasi et al. 13 The most relevant upgrades are the followings:

7 Figure 5. Scheme of the LBT capacitive sensor electronics. See the text for details. ffl new DSP boards design based on ADSP devices. They are 32-bit floating-point DSP with 4Mbit internal memory and they are able to sustain a computational power of 360 Mflop/s. For comparison the MMT336 unit uses 16-bit integer DSP running 80 Mop/s. In both the cases one DSP controls two actuators. The jump in performance allows using the LBT672 electronics as real-time wave-front recontructor. The reconstruction matrix multiplication introduces a 16 μs extra delay when a 30x30 wave-front sensor is used ( ß/4=1414 slopes). Finally the LBT672 DSP boards implements selfdiagnostics of on-board temperature and current driver functionality; ffl new communication board based on Altera Stratix chip as main system logic, 4 bi-directional 2 Gbit/s Fiber Channel units for the fast-communication and one 1 Gbit/s Ethernet line for configuration and diagnostic communication. The four Fiber Channel units can operate in pairs in order to assure the two logical communication lines (4 Gbit/s each) that are needed in the daisy-chain communication structure of the crates. The crate bus and internal routing of data allows an effective throughput of 2.9 Gbit/s and ο 400 Mbit/s for the fast and diagnostic communication lines respectively. For comparison the MMT336 unit implements acommunications line with a 160 Mbit/s throughput without any dedicated configuration/diagnostic line; ffl new design of the capacitive sensor electronics, as shown in Fig. 5. In the MMT336 case the electronics on the capacitive sensor board contain just the pre-amplification stage in the old-capsens" side. The pre-amplified 40 khz square wave is sent to the MMT DSP board where an 80 khz ADC samples the high and low level to compute the peak-to-peak value (/ 1=gap). In the LBT672 scheme the capacitive sensor board contains also two sample-and-hold circuits giving as output the high-level and the low-level voltage. The signals are sent to the DSP boards where a differential amplifier computes analogically the peak-to-peak value. In the P1 test the sample-and-hold circuits run at 156 khz permitting to use a reference square wave at the same frequency. With the new scheme we obtain a larger effective bandwidth in the LBT case (90 khz) with respect to MMT336 (27 khz); ffl implementation of a large-band electronic damping. The analog output from the LBT capacitive sensor can be filtered to obtain an estimation of the local velocity. It is used as analog feedback for the current driver of the coil. The analog nature of the velocity loop is no subjected to delays introduced by the limitation of the sampling frequency of the ADC electronics (80 khz). In the MMT case, for instance, a velocity loop can be implemented in digital form (changing the DSP software) but it's effectiveness is limited by the 40 khz digital sampling of the position.

8 Figure 6. Scheme of the position-velocity loop control implemented in the MMT336 unit (to be tested) and in the P36 prototype. The Feed-forward loop is not shown here. c is the command received by the wave-front reconstructor. It is updated at 550 Hz. p is the sensed position, n is the noise injected in the capacitive sensor and s is the Laplace frequency. G: proportional gain of the position error loop (40 khz); D: loop delay (6.3μs); CD: coil current driver (-3dB at 56 khz); M: mirror shell; S: capacitive sensor (-3dB at 27 khz); LIN: capacitive sensor output linearization. The black and the gray arrows correspond to the 40 khz digital loop and the analog loop respectively. Figure 7. Position-velocity control loop test on the P45 prototype. On the left: step responce of an actuator with large gap (100 μm, low air damping) and no electronic damping. On the right: step responce with the same gap and 50 Ns/m elecronic damping. The introduction of the electronic damping allowed to increase the gain to 0.17 N/μm and to obtain a settling time of 0.5ms.

9 Figure 8. Scheme of the position-velocity loop control to be implemented in the LBT672 unit. The Feed-forward loop is not shown here. c is the command received by the wave-front reconstructor. It is updated at 1 khz. p is the sensed position, n is the noise injected in the capacitive sensor and s is the Laplace frequency. G: proportional gain of the position error loop (80 khz); D: loop delay; CD: coil current driver (-3dB at 56 khz); K: derivative loop gain, M: mirror shell; S: capacitive sensor (-3dB at 90 khz); LIN: capacitive sensor output linearization. The black and the gray arrows correspond to the 80 khz digital loop and the analog loop respectively. 5. NEW SOLUTIONS FOR LBT672 CONTROL STRATEGY The control strategy concept for MMT/LBT-like adaptive mirrors (i.e. based on electro-magnetic force actuator and internal de-centralized position control loop) has a solid theoretical 16 7, 8, and experimental background. The two key-points in the control strategy are the use of the feed-forward (FF) force and the role of damping. The FF is an open-loop component of the actuator force pattern that is equal to the static forces that are needed to obtain the requested deformation of the mirror. It is used to compensate the static error of the high spatial frequency mode that have a stiffness larger then the static gain of the position control loop. It requires the calibration of the stiffness (also called FF) matrix that is very stable in time. For instance, in the MMT336 unit we have used the same FF matrix for more then 6 months, without noticing any calibration problem. Because there are no modifications in the FF technique between the MMT and LBT units, more details can be found in the cited literature. The other key-point in the control strategy is the role of the damping that is the source of the stabilization of the control loop for the mirror modes having the resonance frequency inside the desired control bandwidth. In particular the achieved bandwidth is almost proportional to the amount of damping per actuator that can be introduced in the system. In the MMT336 unit the damping is generated by the viscous friction of the air that is trapped in the small gap between the thin mirror and the reference plate. MMT336 routinely works with a 50 μm gap in order to push the proportional loop gain to 0:2 N=μm and perform a settling time of 1.7 ms. The gap size has the side effect to set a limit to the stroke of the actuators. Because the capacitive sensor signal saturates for gap less then 24 μm with the current electronics, in the previous case the practical actuator stroke is 26 μm. The damping has a steep dependence on the gap size 19 (/ 1=gap 3 ), then increasing the gap (i.e. the stroke) causes a fast reduction of the dynamical performances of the unit (badwidth / 1=gap 3 ). In order to decouple the stroke from the bandwidth requirements, we decided to substitute the air damping with electronic damping, implementing a position-velocity (PV) control in the actuator loop. In the MMT336 unit the PV control loop can be implemented, as shown in Fig. 6, with a simple modification of the DSP code (digital computation of the velocity signal). Fig. 7 reports the first test results of this implementation on the 36-actuator prototype (P36) 17, 18 of the MMT unit. It shows that in conditions of very low air damping (gap = 100 μm, settling time ο 10 ms) the introduction of an electronic damping can boost the performances to a 0.5 ms settling time level.

10 Figure 9. Telescope Simulator Tower (TST). The 13.5m tube of the TST will be installed in the Solar Tower at Osservatorio di Arcetri. It is shown the design of the tube, the optical layout and the main components of the system (LBT672 with the hexapod (1), the collimator-and-flat optics (2) and the Fisba interferometer (3)). See text for details. The gain of the velocity loop is limited by the delay introduced by the ADC sampling frequency (80 khz) and by the propagation of the noise from the capacitive sensor that is amplified by the derivative filter at high frequency. In the LBT672 case the first problem is strongly reduced increasing the -3 db bandwidth of the capacitive sensor electronics from 27 khz to 90 khz and implementing an analog (instead of digital) feedback loop for the velocity signal (see Sect. 4 for more details). Fig. 8 shows the scheme of the implementation of the PV control loop in the LBT672 unit. This solutions showld allow LBT AS units to achieve actuator stiffness of 1N/μm and 0.5 ms settling time with stroke of mm. 6. THE OPTICAL TEST TOWER FACILITY IN ARCETRI Before shipping the two LBT672 units to the telescope site, we want to run all the tests that are needed to characterize the AS systems under any aspect: thermal behavior, electro-mechanical response and calibration, optical quality and flattening. Moreover we consider as mandatory an extensive test of the full-speed optical closed loop using the complete first-light AO system. It will assure not only a fine tuning of the hardware, but also a good test for the software and all the telescope procedures that will be used for the calibration of the AS units and the whole AO system. Osservatorio di Arcetri is in charge of the tests described above and is working to provide the needed facilities. In Arcetri is already available a 10,000-class clean room for the assembly of the AS units. Moreover we are completing the executive design for the Telescope Simulator Tower (TST) that will be installed in Arcetri during winter 2003 for the optical and thermal tests. As shown in Fig. 9, the TST is a 13.5m-high 1.1m-diameter steel tube enclosing the optical path from the AS to the first-light AO unit. The topside of the tube contains the LBT672 and its hexapod for fine alignment.

11 The first-light AO unit is rigidly connected to the bottom side of the tube that is closed by an optical window. From a mechanical point of view the tube is divided in six sections that are connected together inside the test tower. Each section is made of two concentric steel cylinders (the tube wall) with a 40mm gap between them. The steel is surrounded by an outer polyurethane skin of 2 cm for thermal isolation. When all the sections are bolted together the gap generate a vacuum-proof cavity as long as the tube. The whole structure has a weightof 5,400 kg including the AS and AO units. It is supported by three heavy-load pneumatic isolators located close the bottom of the tube. Four more pneumatic isolators are placed at the top to stabilize and damp the tube tilt. The isolators (by TMC) decouple the system from the ground vibrations and damp the rigid pendulum-like oscillations induced by the wind. The TST tube has a 6 kw cooling system in order to test the AS unit in mountain-like conditions (temperature 20 C). There is also a thermal stabilization system to prevent temperature gradients ( T < 0:2 C) inside the tube and the resulting air convection along the vertical optical path. The temperature stabilization is obtained by the condensation/vaporization of a liquid that is in equilibrium with its vapor in the cavity of the tube wall. Using acetone the pressure in the cavity has to be as low as 20 mmhg for the tests at -20 C. The request of temperature stability with time is driven by the thermal expansion of the materials of the structure. To assure a focus error less then =10 in a differential measurement (about 7 s using our interferometer) a thermal stability of 0.8 C/min is requested. In order to perform the optical tests, the AO unit has a compact interferometer (Fisba, CH) on-board. It produces a F/10 beam that is focused at the Gregorian F/15.0 focal point of the elliptical adaptive shell (13.71 m from its vertex). A beam-splitter folds the beam toward the AS mirror that re-focuses the beam on its F/1.22 primary focus (1.065 m from the shell vertex). A 20mm-diameter F/1.0 collimator produces an image of the deformable shell on a flat mirror that reflects back the beam on the same path. The beam-splitter in front of the interferometer divides the light coming from the adaptive shell in two beams. One re-enter the interferometer, the other feeds the wave-front sensor of the first-light AO unit for a simultaneous measurement of the wave-front aberrations. The collimator-and-flat optics is held in place by a removable Invar structure that is rigidly connected to the fixed side of the hexapod. The Invar structure can be mounted at the telescope for AO/AS re-calibration purposes when needed. The request on its alignment is not very tight (1 mm) because the fine alignment is performed moving the AS unit with the hexapod. 7. CONCLUSIONS The paper presented the design of the two 672-actuator 911mm-diameter AS units for LBT. In particular we emphasized the modifications aimed at improving some of the weak points of the MMT AS unit. The upgrades in the mechanics and electro-mechanics increase robustness, reliability, accessibility of the actuators, and simplification of maintenance procedures. The use of the latest DSP technology for the control boards boosts the computational power of the unit to 121 Gflop/s, allowing on-board wave-front reconstruction. Moreover the introduction of an electronic damping breaks the inverse relation between the actuator stroke and the bandwidth, allowing a 0.5ms settling-time and 0.1mm stroke. Finally the Arcetri optical test facility will allow a full test (optical and electro-mechanical) of the new LBT672 features. Moreover the same facility will be used to perform an extensive test of the optical loop with the first-light AO module, before shipping the whole system to the telescope site (mid 2004). REFERENCES 1. J. Hill and P. Salinari, Large binocular telescope project," Proc. SPIE 4837, L. Miglietta, G. Castelli, C. Galli, R. Bianchi, B. Sampaoli, P. Tassan Din, V. Veri, G. Marchiori, A. Zanon, D. Gallieni, E. Anaclerio, and P. Lazzarini, Large binocular telescope: the pre-erection experience of the largest telescope in the world," Proc. SPIE 4837, S. Esposito, A. Riccardi, J. Storm, M. Accardo, C. Baffa, R. Biasi, P. Biliotti, G. Brusa, M. Carbillet, D. Ferruzzi, L. Fini, I. Foppiani, D. Gallieni, A. Puglisi, R. Ragazzoni, P. Ranfagni, P. Salinari, W. Seifert, P. Stefanini, A. Tozzi, and C. Verinaud, First light ao system for lbt," Proc. SPIE 4839, 2002.

12 4. R. Ragazzoni, T. Herbst, D. Andersen, P. Bizenberger, H. W. Rix, R.-R. Rohlof, C. Arcidiacono, E. Diolaiti, S. Esposito, J. Farinato, A. Riccardi, E. Vernet-Viard, and P. Salinari, Nirvana: a visible mcao channel for lbt," Proc. SPIE 4839, P. Salinari, C. Del Vecchio, and V. Biliotti, A study of an adaptive secondary mirror," in Active and adaptive optics, F. Merkle, ed., Proc. ICO-16 Satellite Conference 48, pp , International Commission for Optics (ICO), (Garching, Germany), M. Lloyd-Hart, Thermal Performance Enhancement of Adaptive Optics by Use of a Deformable Secondary Mirror," PASP 112, pp , Feb G. Brusa, A. Riccardi, P. Salinari, F. Wildi, M. Lloyd-Hart, H. M. Martin, P. Mategazza, R. Biasi, D. Gallieni, and F. Zocchi, Mmt adaptive secondary: performance evaluation and field testing," Proc. SPIE 4839, A. Riccardi, G. Brusa, C. Del Vecchio, P. Salinari, R. Biasi, M. Andrighettoni, D. Gallieni, F. Zocchi, M. Lloyd-Hart, F. Wildi, and H. M. Martin, The adaptive secondary mirror for the 6.5 conversion of the Multiple Mirror Telescope," in Beyond Conventional Adaptive Optics, ESO Proc., in press. 9. F. Wildi, G. Brusa, A. Riccardi, R. Allen, M. Lloyd-Hart, D. Miller, H. Martin, R. Buasi, and D. Gallieni, Progress of the MMT adaptive optics program," Proc. SPIE 4494, pp , July F. Wildi, G. Brusa, A. Riccardi, D. Miller, M. Lloyd-Hart, D. Fisher, and H. M. Martin, 1st light ofthr 6.5m mmt adaptive optics system with deformable secondary," Proc. SPIE 4839, S. C. West, S. P. Callahan, D. L. Clark, B. Cuerden, D. G. Fabricant, D. Fisher, C. B. Foltz, J. D. Gibson, W. L. J. Kindred, H. M. Martin, D. L. McKenna, B. A. McLeod, T. E. Pickering, P. F. Spencer, T. J. Trebisky, K. Van Horn, and T. Williams, J, Status of the 6.5m convertion of the multiple mirror telescope since 2000 may," Proc. SPIE 4837, D. Gallieni, E. Anaclerio, P. Lazzarini, A. Ripamonti, S. Spairani, C. Del vecchio, P. Salinari, A. Riccardi, P. Stefanini, and R. Biasi, Lbt adaptive secondary units final design and construction," Proc. SPIE 4839, R. Biasi, M. Andrighettoni, D. Veronese, V. Biliotti, L. Fini, A. Riccardi, and D. Gallieni, Lbt adaptive secondary electronics," Proc. SPIE 4839, C. Del Vecchio, G. Brusa, D. Gallieni, M. Lloyd-Hart, and W. B. Davison, Static and dynamic responses of an ultra thin adaptive secondary mirror," Proc. SPIE 3762, pp , Sept C. Del Vecchio, G. Gallieni, H. M. Martin, A. Riccardi, G. Brusa, and R. Biasi, Design improvements of the lbt adaptive secondary," in Beyond Conventional Adaptive Optics, in press. 16. G. Brusa, A. Riccardi, M. Accardo, V. Biliotti, M. Carbillet, C. Del Vecchio, S. Esposito, B. Femen a, O. Feeney, L. Fini, S. Gennari, L. Miglietta, P. Salinari, and P. Stefanini, From adaptive secondary mirrors to extra-thin extra-large adaptive primary mirrors," in Proceedings of the Backaskog workshop on extremely large telescopes, T. Andersen, A. Ardeberg, and R. Gilmozzi, eds., ESO Proc. 57, pp , Lund Obs. and ESO, A. Riccardi, G. Brusa, V. Biliotti, C. Del Vecchio, P. Salinari, P. Stefanini, P. Mantegazza, R. Biasi, M. Andrighettoni, C. Franchini, D. Gallieni, M. Lloyd-Hart, P. C. McGuire, S. M. Miller, and H. M. Martin, Adaptive secondary mirror for the 6.5-m conversion of the Multiple Mirror Telescope: latest laboratory test results of the P36 prototype," Proc. SPIE 4007, pp , July G. Brusa, A. Riccardi, V. Biliotti, C. Del Vecchio, P. Salinari, P. Stefanini, P. Mantegazza, R. Biasi, M. Andrighettoni, C. Franchini, and D. Gallieni, Adaptive secondary mirror for the 6.5-m conversion of the multiple mirror telescope: first laboratory testing results," Proc. SPIE 3762, pp , Sept G. Brusa, A. Riccardi, S. Ragland, S. Esposito, C. Del Vecchio, L. Fini, P. Stefanini, V. Biliotti, P. Ranfagni, P. Salinari, D. Gallieni, R. Biasi, P. Mantegazza, G. Sciocco, G. Noviello, and S. Invernizzi, Adaptive secondary P30 prototype: laboratory results," Proc. SPIE 3353, pp , Sept