Adaptive Optics Facility The DSM
	HOME INDEX SEARCH HELP NEWS

The Deformable Secondary Mirror (DSM)

Contents

• Introduction to large deformable mirror technology
• MMT-LBT-VLT large Deformable mirror concepts
• Hexapod
• Thin Glass Shells for large DMs
• Large DMs System Aspects
  
• Correction Capability
  

Introduction to the large deformable mirror technology

The concept of thin shell & force actuators is one of the most promising in the field of large deformable mirrors; the largest deformable mirror have been built/designed with this technology.  A 642mm diameter convex secondary mirror with 336 actuators has been developed and is being used by the MMT (Mt Hopkins, Arizona), while the two 911mm diameter and 672 actuators concave secondary mirrors of the LBT (Mt Graham, Arizona) are being integrated (at the time of this writing). A similar design is being studied for feasibility, for one of the VLT Unit Telescope; the deformable secondary design is 1120mm in diameter and offers 1170 actuators for adaptive correction Figure 1 and Figure 2.

Figure 1: Conceptual design for the 1120mm diameter, 1170 actuators VLT deformable secondary mirror. The M2-Unit contains the electronics (top) an Hexapod for centering and focusing (middle) and the deformable thin shell (bottom).

MMT-LBT-VLT large Deformable mirror concepts

These mirrors are composed of 3 basic elements: a back-plate, holding the voice-coil force actuators, a reference body and the thin shell. Each voice coil applies a force to a corresponding magnet glued onto the back face of the thin shell. A ring of conductive material (chrome, aluminium, gold…) is deposited around each magnet and is mirrored on the reference body. These two opposite coatings constitute a capacitance used as space sensor. The reference body being a calibrated optical surface, an equal spacing for all capacitive sensors insures a basic (not optimal) optical quality on the shell. 

A typical gap of ~50 mm is proposed for the VLT and provides air damping between the shell and the reference body. An internal control loop at 80 kHz insures that the force applied maintains the capacitive sensor to a constant gap. Note also that the derivatives of the capacitive sensor positions provide a measure of the velocity of the shell displacement which in turn is used by the system to define an electronic damping matched to mode stiffness. This feature allows reaching high bandwidth for the system even if some control modes have low resonance frequency.

Figure 2: Three basic components of a force/thin shell mirror (VLT design): Cold plate, Reference body and thin shell.  In this design a Hexapod is attached to the cold plate and used to provide the assembly fine focusing and centering motions.

The back plate has two functions: holding the voice coil actuators and evacuating heat dissipated by the coils with the help of an integrated cooling fluid circuit.

The reference body can be a conventional, thick, ULE or Zerodur optical component, with the exception of the numerous cylindrical openings allowing passage for the actuators. More recent designs (VLT DSM) explored with industrial partner a light-weighting scheme (50-60% light-weighted Zerodur or SiC) to reduce the weight of the complete assembly (realistic without being a huge cost driver). SiC offer the added advantage of being extremely rigid compared to ULE or Zerodur. Note that the front surface of the reference body can be “rough”; the requirement is not strictly speaking the one of an optical surface. Image quality can be specified from the largest linear scale down to a fraction of the inter-actuator spacing.

Figure 3: Lightweighting scheme of the VLT design Reference Body; estimated weight (zerodur) is 35 kg while a monolithic design would lead 130 kg.

The baseline design uses Zerodur for the reference body (and the thin shell) but other materials can also be considered as SiC (Boostec, ECM CeSiC) providing high stiffness and proper optical quality polishing possibility. Price have been seeked from industries.

     Hexapod

A Hexapods is used to produce fine motions on the back-plate; hexapods are particulary well suited for this type of applications allowing fine accurate repeatable motions and high rigidity. Any motion applied to the back-plate is passed on to the optical surface of the thin shell. It can be typically used for small focusing motion and centering correction.

Thin glass shells for large DMs

The thin shell remains a high-technology, costly, high risk venture. Up to now only the Steward Observatory Mirror Lab has been producing thin shells from Zerodur blanks:

• MMT: Convex aspheric R_c=1795mm, k=-1.409,  f=642 mm, 2mm thick, Zerodur 336 actuators
• LBT (2 units):  Concave aspheric R_c=1974.2 mm, k=-0.7328,  f=911 mm, 1.6mm thick, Zerodur 672 actuators

A Call for Tender for the manufacturing of the Zerodur thin shell for the VLT DSM has been launched recently (mid-June) and offers have been received.

• VLT: Convex aspheric, R_c=4553mm, k=-1.66926,  f=1120 mm, 1.8-2.0 mm thick, Zerodur 1170 actuators

The production of an aspheric convex thin shell adds substantial complexity to the manufacturing process. However, a few optical suppliers in Europe have been known to develop expertise in the field of large optical thin plate polishing, even with power. Here is a list of known promising technologies in this field:

• Orbital double side polishing (flat plates 3mm thick, 680 mm diam. Have been produced)
• Stress polishing (spherical polishing on aspheric mandrel)
• Slumping (melting of a thin flat shell on a aspheric mandrel)
• Related to this problematic, and due to the mass production of large LCD display and numerical TV, Corning and Schott are known to produce in large series and cost effectively large flat shell (0.7mm up to 2 m in size).

Large deformable mirror systems aspects

Actuators and magnets

Actual technology imposes magnet sizes of the order of 12mm in diameter and this is what drives the minimum inter-actuator spacing to ~28mm. Reducing this size further brings also complications at the level of the voice coil, but also would reduce the actuator stroke (force).

 Power Dissipation

• 311 W in Coils, 1123 W in racks and 43 W in cables

The VLT DSM design is based on the following distribution of control per channels:

• 13 control boards per rack, 2 DSP per boards, 8 channels per DSP = 208 channels per racks (6 racks are required)

Correction Capability

The following figures extracted from the Fitting Error Analysis report summarize the correction capability of the DSM.

Figure 4: PtV actuator stroke in median seeing conditions (left-hand) and Actuator position rms in median seeing conditions (right-hand). Result of 10000 uncorrelated wave-fronts.

Figure 5 : Peak actuator force in median seeing conditions (left-hand) and Rms actuator force in median seeing conditions (right-hand). Result of fitting 10000 uncorrelated wave-fronts.

Table 1 shows that the residual error with all modes corrected is 70 nm RMS, fulfilling the specifications.

Table 1: Summary of the simulation results in the median seeing case. Results of fitting 10000 uncorrelated wave-fronts.

More details...

Send comments to <rarsenau@eso.org> Last update: July 15th, 2006