OPTICAL SYSTEM WITH WAVEFRONT SENSOR

RELATION BACK:

This document -Ls based upon and claims priority of U . S . Provisional Patent Application 60/643 , 869 , filed January 13 , 2005.

FIELD QF THE INVENTION:

This invention relates generally to optical sensor systems , and more specifically to systems for detecting, localizing and sometimes responding to incident beams of light. The invention encompasses but is not limited to system-level architecture and performance strategies for an adaptive-optias ("AO") system based on microelect-tromechanical- system ("MEMS") mirror arrays , and particularly centered on afocal MEMS beam-steering ("AMBS' ^' '' ) and AO technology.

BACKGROUND :

Historically, a deformable mirror ("DM") consisted of a monolith- ic reflective face-sh.eet that had an array of actuators which deformed the face-sheet through the application of force normal to the face-sheet, or torque , or both, in order to correct for wavefront error. Whether located, at a primary mirror, as in two proj ects known as "Large Active Mirror Program" ("LAMP") and "Active Large Optic TeIe- scope" ("ALOT") , or a~t another pupil location as in the "Active Imaging" ("AI") telescope on Haleakala or the Senior Chevron program, conventional DMs have these limitations :

■ low actuator stroke, limiting the magnitude of a particular Zernike error that can be corrected (2 to 4 μm wavefront) ,

■ presence of broad actuator-to-actuator influence functions limiting- the order of the Zernike error that can be corrected (2 to 3 actuator diameters) ,

■ low temporal bandwidth, when located at the primary mirror ("PM") , due to the large PM mass that must be moved,

■ large mass when located at the PM due to the mass of the actuator needed to deform a PM (the ALOT actuator mass was approximately 1 kg) , ancl

■ minimal ability to direct the sensor field of view ("FOV") within a large FOR .

DM technology based on a spatial light modulator ("SLM") , leveraging electronics lithographic-processing techniques , has been developed recently to address problems associated with the broad, influence function, size , mass , power and cost of the traditional DM approach .

While certain-Ly an advance over the traditional technology, SLMs fit a wavefront witti only a linear-displacement fit.

Deformable U±rrrox Systems One of t-he earliest AO systems was built by Itek Optical Systems for the Air Force Maui Optical and Supercomputing Site (AMOS) . It was used to correc t: for atmospheric turbulence in order to observe satellites . tXsing a DM with a monolithic face-sheet supported by an array of lead magnesium niobate (PMN) at a pupil location conjugate with the prima-try mirror (PM) , these DMs were early precursors to the systems currently produced by Xinetics .

In order to correct the wavefront, the actuators on these mirrors typically provd.de piston displacement to deform the face-sheet, although there are some configurations that can also provide moment to the surface . Relatively high actuator densities (1000' s of actuators) and high tempo-ral bandwidth (ones to tens of kilohertz) can be achieved with this approach .

A limiting problem is that they draw considerable current at high voltage, have modest stroke on the order of ones of micrometers , and are relatively volume intensive if all of the drive electronics is included in the calculation . They are expensive too, due to manually in-

tensive nature of the fabrication process . This topology is not readily amenable to fabrication using semiconductor processing techniques .

Active optical systems such as ALOT, Keck Telescope and LAMP, were segmented deformable primary mirror sys tems , where the actuators and. wavefront sensors were used to control phase between the segments and. wavefront slope within them . In the case of spectrally broadband sys -terns such as ALOT and Keck, absolute phas -e between the segments had to be maintained, requiring even larger strolke in the actuators when compared to narrow-band systems requiring on.-ly modulo 2 phasing.

Controlling wavefront error at the primary mirror, where the mirror stiffness is several orders of magnitude higher than a Xinetics- type DM, requires significant force output, snd the influence functions are multiple actuator diameters broader than membrane-based DMs . This results in a need for either higher actuator density or acceptance of largjer residual error .

The actuator mass for both LAMP and ALOT was on the order of 1 kg, not including the multiple racks of drive electronics associated witli powering them. In general , it is signifricantly more mass efficient to have a minimally active PM, and correct for the error at a pup±.1 location where the magnification allows for smaller mirrors .

Wavefront Sensors

A wavefront sensor ("WFS") providing cl osed-loop feedback for an AMBS -AO system must measure modulo 2%/λ phase . Four wavefront-sensing methodologies are evident candidates .

Shack Hartmann Wavefront Sensor — Companies such as Adaptive Opti cs Associates have commercially developed, the Shack-Hartmann wavefront sensor . In this configuration, the* wavefront is directed to a mi crolens array that focuses the wavefront on a focal-plane array.

Each microlens senses a portion of the wavefront and, to the extent a slope is present, the focus location is displaced in the X and/or Y direction . The derivative of the wa-vefront as a function of X or Y is proportional to the displacement rela -tive to the local optic axis of the microlens .

A wavefront reconstructor then integrates the derivatives , creating a. wavefront map that in turn is used to create a DM-actuator com-

mand set . It is not clear that tliis WFS approach can measure phase between MEMS mirrors .

Stochastic-Based (SB) Image Quality Sensor (IQS) — A stochastic- based IQS , such as the one develop>ed by Weyrauch et al. , provides a promising alternate approach to piroviding high-bandwidth closed-loop* sensing for an AO system. ¹ The IQS is not a wavefront sensor, but evaluates the system image quality- directly.

While only applicable to "simple" images , such as that of a guide star, a raster-scan I-M)AR, or lasex communications system in which the image consists of a single point, -the stochastic-based IQS approach continually trains itself to achie-ve the desired image metric, such as minimizing the radius of the image point spread function (PSF) . The SB-IQS redirects a portion of the incoming image with a beam splitteor , focusing it onto a high-speed VLSI detector array. For each time step, each element in the DM is perturbed and tho resulting influence upon the image metric quality is recorded. Given the resulting "influence function" , a command vector is given that optimizes the resulting quality metric .

Pixelated Phase-Mask. Dynamic Interferometer — A new type of spa- tial phase-shifting, dynamic interferometer that can acquire phase- shifted interferograms in a single camera frame is another candidate component . ² The interferometer is constructed with a pixelated phases- mask aligned to a detector array.

The phase-mask encodes a high. —frequency spatial interference pattern on two colinear and orthogonally polarized reference and test: beams . The phase difference between the two beams can be calculated using conventional N-bucket algorithms or by spatial convolution . The wide spectral response of the mask and true common-path design permits operation with a wide variety of interferometer front ends , and with virtually any light source including white light . The>

¹ T . Weyrauch, M. A. Vorontsov, T. G. Bifano, J. A. Hammer, M. Cohen, and. G . Cauwenberghs, "Microscale Adaptive Optics : Wavefront Control with a Micro-Mirror Array and a VLSI Stochas tic Gradient Descent Controller, " App. Opt . , Vol . 40, 4243-4253, (2001

² This paragraph is closely paraphrased from: James Millerd, Neal Brock, John Hayes, Michael North-Morris, Mat t Novak and James Wyant . "Pixelated Phase-Mask Dynamic Interferometer", 4 D Technology Corporation, 3280 E . Hemisphere Loop, Suite 146, Tucson, A-Z 85706.

potential for high temporal update and ability to work over a broad spectral band make this approach attractive .

Shearing Wavβfront Sensor A shearing wavefront sensor was utilized on both the LAMP and ALOT AO systems for closed-loop control feedback for both segment phasing and slope correction . More detail on the sensor implementation appears below, but a shearing WFS consists of a grating in the image plane where the DM located at a pupil is re- imaged.

The intensities of the 0 and. ±1 orders are then measured with a photodetector, where the phase between the reference detector and detectors viewing other subaperturres are compared — providing slope information over the segments and phase between them. Primary benefits of the shearing WFS are its relatively high bandwidth (given adequate signal-to-noise ratio) , ability to be utilized on "simple" and extended complex scenes , and the ability to measure both slope and phase between segments .

Components not previously associated with highest-precision DM technology: MEMS Mirror Array — The first significant commercial use of MEMS mirrors was in the Texas Instruments Digital Light Projector (DLP) MEMS array. Formed in an array of Ik x Ik 10 μm two-axis mirrors , the bistable mirrors were controlled ojpen-loop, with the mirrors stepped from ±10 ^° locations at rates on the order of 10 μs per step . To the best of my knowledge , the use of MEMS mirrors as DMs has not been suggested heretofore . Even if it had been, the available MEMS mirrors would not have been adequate to the task .

The DLP mirrors , for example, could only operate in the tip and tilt directions . They were not capable of "piston" movements — i . e . , linear motion in or out relative to the overall mirror-array backplane.

Tip and tilt adjustments are desirable for matching wavefront slopes , but piston adjustment too i_s usually needed to compensate the rectilinear, stepwise profile of a wavefront. In a MEMS array, furthermore, wavefronts are disturbed by offsets between the planes of adjacent individual mirrors — especially where the offsets are significant in comparison with the wavelength of the incident light .

Furthermore the DLP mirrors were not anslog or even multistate binary — i . e . , each mirror could take on on ly one of two positions about either axis . Wavefront correction typi cally calls for adjustment to rather small fractions of a wavelength; th erefore either analog or fine-granularity multilevel binary operation is usually needed.

A more closely related development in MEMS scan mirror arrays was in the area of optical switching; here the mixrors could be controlled open-loop about one or two axes over the entiare range of mirror travel , and thus were "analog" in the sense of being sibIe to point the mirror . Lucent in its "Waverunner" optical switch , and Calient Networks , with its "3-D" MEMS-mirror optical switch, are good examples of this technology. These arrays are typically large.tr, hundreds of micrometers to ones of millimeters — but have millisecond-level step response characteristics because they are controlled o]pen loop . Areal densities of these arrays are also low, less than 50% .

Therefore significant modifications to their architecture are required to obtain an adequate DM for an AO system.

Spatial Light Modulators (SLMs) — SLMs come in two basic config- urations , correcting wavefront error by fitting the error through discrete pieαewise phase steps . Some SLM arrays consist of mirrors that are adjustable in-and-out, i . e . in piston .

An SLM is not the same thing as a MEMS airay, and the two should not be confused. Among several important differences , an SLM moves in piston only, and so can only be adjusted to compensate the stepwise profile, not the continuous slopes , of an incoming light-beam wavefront to be corrected.

Lucent Technology is being funded by DARE ¹ A as part of a program to develop a several-million-element SLM device , and Boston Micro Ma- chines has produced smaller SLMs . While SLMs .are certainly part of the DM field, they cannot be associated with DMs oz£ the highest precision .

This is because, as noted above , they opsrate only in piston . Lacking tip and tilt capability, they do a fai-ar job of compensating for wavefront offsets only, in sawtooth fashion (analogous to the so-called "aliasing" in computer images) , but not in matching wavefront slopes .

Ironically this deficiency is precisely opposi-fce to the above-described

limitation of MEMS arrays heretofore — i . β . tip and tilt only, with no piston .

Furthermore even the SLM piston excursion that is available, per actuator, is not high enough to provide the necessary offsets , without an inordinately high actuator density. As will later be seen , the density of actuators required for fully satisfactory waveffront matching, in piston , is almost impractically extreme .

Some SLMs are not mechanical at all — they are liquid-crystal phase modulators (LCPM) , which correct the phase through transmission . Unfortunately they have a significant transmission loss (50% is typical) and are slow, with a several-millisecond response tiiαe .

Conclusion

From the foregoing it can be seen that no prior art provides mechanisms capable of economical , practical (e . σ. lightweight and low- power) , very rapid and accurate compensation of wavefront errors in incident beams of light. Thus important aspects of the technology used in the field of the invention remain amenable to useful refinement .

BRIEF SUMMARY OF THE INVENTION:

The present invention introduces just such refinement . In pre- ferred embodiments the invention has several independent aspects or facets , which are advantageously used in conjunction togethier, although they are capable of practice independently.

In preferred embodiments of its first major independent facet or aspect, the invention is apparatus for imaging a scene . The apparatus includes a deformable mirror for receiving, and forwarding to a detector, optical radiation from the scene .

(In the bodies of certain of the appended claims the "word "such" is used — in place of "the" or "said" — when referring ba_ck to terms introduced in preamble that are not part of the claimed inventive αom- bination but rather are parts of the environment or context of the invention . The purpose of this convention is to make particu larly clear which recited elements are within the claimed invention and which are

not, thereby more particularly pointing out and more distinctly claiming the invention . For example the phrase (in the claims) "such scene" emphasizes that the scene is not part of the invention but only something in the environment to which the elements of the invention are referred or referenced . )

The deformable mirror includes a MEMS mirror array. The invention also includes some means for evaluating wavefront quality of radiation forwarded by the MEMS array . For purposes of generality and breadth in discussing the invention, these means can be called simply the "evaluating means" .

It further includes some means , responsive to the s ensor, for adjusting the MEMS array to optimize the wavefront quality. Again for breadth and generality, these means will be called simply the "adjusting means" . The foregoing may represent a description or defini-fcion of the first aspect or facet of the invention in its broadest or" most general form. Even as couched in these broad terms , however, it can be seen that this facet of the invention importantly advances the art .

In particular, by using a MEMS array as a deformable mirror, for the first time, the invention exploits the remarkable comJbination of very low per-unit cost inertia, weight, and size — and extremely high speed — of MEMS apparatus . In addition, once production engineering is in place, the invention enjoys the benefits of rapid, easy mass- manufacturing techniques , in contrast to the essentially liandmade char- acter of most previous deformable-mirror systems .

Although the first major aspect of the invention thtαs significantly advances the art, nevertheless to optimize enjoyment of its benefits preferably the invention is practiced in conjunction with certain additional features or characteristics . In particular, preferably the MEMS array includes mirrors adjustable in the tip, tilt and piston directions . In this case, further preferably the mirror adjustments are within a adjustment range of roughly ±5° in tip and tilt, or greater, or roughly ±10 μm in piston, or greater — or most preferably both .

Also preferably the evaluating means include a wavefront sensor . In this case , further preferably the adjusting means include electron-

ics for reading output signals from the wavefront sensor and in response generating adjustment signals for the MEMS array — and ideally the wavef-cont sensor includes a shearing wavefront sensor.

Another preference is that the apparatus include an afocal opti- cal element for receiving from the scene , and forwarding to the MEMS array, the optical radiation . In this case preferably the adjusting means include electronics for reading output signals from the wavefront sensor and in response generating adjustment signals for the MEMS array. This latter preference is actually not limited to the afocal-op- tic case, but is a basic preference that applies more generally to all forms of the first aspect of the invention .

Yet another preference is that the detector be an imaging detector for forming signals representative of an image of the scene . If this preference is observed, then a further subpreference is that the apparatus further include an imaging optical element, between the MEMS array and the detector, for focusing the optical radiation on the detector .

A still-further preference is that the apparatus further include some means for dividing the optical radiation from the MEMS array, to send respective parts of the radiation to the detector and to the wave- front-quality sensor . Again for purpose of generality and breadth, these means may be called simply the "dividing means" .

In preferred embodiments of its second major independent facet or aspect, the invention is a method for imaging a scene . The method includes the step of receiving, at a miniature mirror array, optical radiation from the scene .

It also includes the steps of forwarding the radiation from the mirror array to a detector, and operating a sensor to evaluate wavefront quality of radiation forwarded by the mirror array. The method also includes the step of — in response to the sensor — adjusting mirrors of the array in tip, tilt and piston directions to optimize the wavefront quality. The foregoing may represent a description or definition of the second aspect or facet of the invention in its broadest or most general

form. Even as couched, in these broad terms , however, it can be seen that this facet of the invention importantly advances the art.

In particular, this aspect of the invention is first to provide independent adjustment , in all three critical degrees of freedom, for a small-size apparatus .

Although the second major aspect of the invention thus significantly advances the art, nevertheless to optimize enjoyment of its benefits preferably the invention is practiced in conjunction with certain additional features or characteristics . In particular, pref- erably the adjusting step includes operating MEMS mirrors in the tip, tilt and piston directions .

Another preference is that the mirror-operating step include mirror adjustments that are :

■ made within an operating range of roughly +5° in tip and tilt, or greater; or

■ made within an operating range of roughly ±10 μm in piston, or greater; or

■ preferably, both .

The foregoing features and benefits of the invention will be more fully appreciated from the following detailed description of preferred embodiments — with reference to the appended drawings , of which :

BRIEF DESCRIPTION OF THE DRAWINGS :

Fig . 1 is an isometric or perspective view of a three-degree-of- freedom prototype small MEMS-mirror array;

Fig . 2 is a comparison of DM Zernike corrections comparison for three types of DMs — conventional membrane or SLM, and MEMS-type as described herein;

Fig . 3 is an optical-system diagram of an AMBS-AO system with a wavefront sensor, according to the present invention;

Fig . 4 is a like diagram but showing only a portion of the Fig . 3 system, particularly concerning diffraction-limited spot size for off- axis field locations ;

Fig . 5 is a diagram like Fig . 3 but more complete , showing also the attached processor for controlling the MEMS array in response to the wavefront sensor;

Fig . 6 is a diagram, highly conceptual , of a top-level AMBS-AO diffraction-limited error budget, Rx (residual after closed-loop wave- front correction at specified bandwidth) ; Fig . 7 is a diagram like Fig . 4 , but showing the shearing wave- front sensor, for phase measurement ;

Fig . 8 is a related diagram, -very highly conceptual , of a representative MEMS-array sheared image on detector plane, for piston measurement between mirrors ; Fig . 9 is like diagram, but for slope measurement on a representative single MEMS mirror;

Fig. 10 is a hybrid diagram, the upper view showing in a conceptual manner the MEMS-array physical architecture, and the lower view being a partial system block diagram showing the MEMS-array functions ; Fig . 11 is a pair of perspective or isometric views of prototype MEMS mirrors having available movement in three directions : tip, tilt and piston (substantially linear motion in and out of the plane of the mirror or mirrors) — the left-hand view being a single mirror, 800 μm square , and the right-hand view being a two-by-two-mirror array; Fig. 12 is a schematic block diagram of AMBS-AO closed-loop control architecture for the present invention;

Fig. 13 is a graph of step response in a 3 kHz closed-loop MEMS control demonstration, 700 μm mirror — more specifically, the percentage of full step achieved, plotted against time in microseconds ; and Fig. 14 is a general view of a. wavefront-error-σorrection (WEC) demonstration breadboard.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS :

Deformable mirrors ("DMs") using MEMS mirrors , for adaptive optical ("AO") systems , hold the potential to perform two critical func-

tions — revolutionizing the performance , size , mass and power required for active and passive sensor systems . The first such function is the correction, of d . c . -to-kilohertz temporal-frequency and low-to-high- order Zernike spatial-frequency wavefront error due -to these sources : ■ propagation of light through extended atmospheric turbulence,

■ platform dynamic disturbance causing wavefront and line-of-sight

(LOS ) error,

■ thermal variation causing wavefront and line-of-sight error ,

■ phase and wavefront error due to the deployment of large -aperture segmented and monolithic primary mirror systems ,

■ gravity release after launch , and

■ fabrication error a Ia Hubble .

A second maj or function is possible with an AO system of the present invention , which in this document may be called afocal-MEMS- beam-steering adaptive optics or "AMBS-AO" . That second function is achievement of wide f ield-of-regard ( "FOR" ) , high-speed line-of-sight ( "LOS") pointing , and stabilization without a gimbal or angular- momentum control system . AMBS systems having 120 ° FOR have been demon s tra ted .

Central to the AMBS-AO system is a MEMS mirror array in which each mirror is controlled in three substantially orthogonal degrees of freedom. Thus a MEMS array for use as a DM, according to the present invention is capable of greater degrees of freedom th.an either the original DLP array, which operated in tip and tilt di-siensions only, or the more recent SLM devices that can operate in piston only.

A working prototype two-by-two-mirror array 241 (Fig . 1) , prepared for me by a vendor, operates in all three adjus tment dimensions : tip, tilt ( θ _x and θ _γ ) and piston (Z) . The reflective elements 241 , each 0.4 mm on a. side , are manipulated by circuitry 243 an<d other actuator components ±iαmediately below the extremely thin, low-inertia mirror pads .

Even MEMS arrays of present interest, operating in all three dimensions , L . e . tip, tilt and piston ("TTP") , can be inadequate if they do not have sufficient angular range for tip and tilt , and thus cannot quite match wavefront slope . Likewise they can be inadequate if they

3.3 do not have sufficient throw in piston , and thus cannot quite compensate for offset between adjacent miirrors .

My prototype mirrors operate to ±5° in tip and tilt, and ±10 μm in piston , and these ranges are adequate . Previously, however , the Ξ same vendor offered only very rough3_y half this angular variation in tip/tilt .

Those earlier offerings were also capable of some piston motion, although actually the piston excursion happened to be available only as an incidental result of the mechanics employed to implement the tip and O tilt motions . This incidentally ari sing piston operation was likewise only very roughly half of the range needed for the present invention .

Measured against needs and capabilities of my afocal MEMS beam- steering (AMBS) configurations, SLMs therefore have these limitations : 5 ■ significantly higher actuator density required to obtain the same residual error for a given Zernike (more than thirty times the number of actuators/diameter, or 1 , 000 times the number of actuators , are required to obtain the same residual error) ; ■ more closed-loop control loops , given higher count of actuators 0 to fit a wavefront result, making the system more complex —

■ higher wavefront sensor ( ^" WFS) density

■ more control commands

■ higher control processor arates ; and

■ minimal ability to direct the sensor FOV within a large FOR .

AMBS coupled with wavefront sensing (WFS) for modulo 2π/λ phase and slope control utilizing TTP MEMS mirror arrays , provides a revolutionary advance to AO systems used to transmit or receive laser through atmospheric turbulence or limited-band multispectral imaging systems . 0 AMBS-AO has many significant advantages over gimbaled systems and provides the following system-level benefits over other approaches :

■ AMBS replaces gimbal to provide :

■ 10 kHz versus tens of Hz closed-loop control ,

■ 10 to 100 μs step response for beam steering, ■ LOS stabilization,

■ 1/10 to 1/100 reduction in size, mass and power for beam- steering function ,

■ large FOR , up to 2 steradians ; and

■ AMBS with wavefaront sensing provides the following benefits over DM or SLM-piston-only:

■ tip/tilt capability of > ±20° wavefront (demonstarated) , ■ piston capability of ±20 m wavefront, and

■ fewer actuators/diameter, given the ability to piston and tilt about two axes for matching wavefront .

The TTP MEMS "DM" of the present invention, as a part of the AMBS-AO system, provides a significant advantage over piston-only devices such as SLMs , and the more traditional membrane-overr-piston- actuator configurations . My invention yields a residual errror ratio 155 for Zernike aberration (or expansion-term coefficient) tf=5 (Fig. 2) , and ratio 154 for Zernike #19. These error ratios 155 , 154 are as defined in Applied Optics and Optical Engineering XX . They can be contrasted with fairly comparable values 153 , 152 obtained when fitting the Zernike polynomial to performance of an SLM piston-only array.

They can also be contrasted with corresponding values 151 ob- tained for a subset of ^" membrane DMs — where residuals were reported for the DM noted in the footnotes . ^{3 4} Residual error ratio JL s defined as the ratio of residual error after correction to the initisl error in an RMS sense, and the plot is normalized in terms of actuatoars per diameter . The MEMS TTP approach provides two orders of magnitude 156 less residual wavefront error ,for a given number of actuators per diameter . From another perspective, the SLM solution requires more than a thou-

³ Keith Bush, Dee German, Beverly Kleπvme, Anthony Marrs, Michael Schoen, "Electrostatic Membrane Deformable Mirror Wavefront Control Systems : Design and Analysis" Advanced Wavefront Control : Methods, Devices, and Applications II, edited by John D. Gonglewski, , Mark T . Gruneisen, Michael K. Giles, Proceedings of SPIE Vol . 5553 (SPIE, Bellingham, WA, 2004 )

1) 37 hexagonal actuators

2) 112 rectangular actuator pads

3) 384 rectangular actuator pads

4 ) 632 rectangular actuator pads

⁴ William H. Lowreya, John L. Wyniaa, and Mark A Ealey, "Characterization of three advanced deformable mirrors", Part of the SPIE Conference on Propagation and Imaging through the Atmosphere II . San Diego, California • July 1998, SPIE Vol . 3433

5) 349-channel DM

sand times the number of control-loop channels to obtain the same residual error (at 10 ^"3 ) .

The residual error ratio values discussed here, for both the TTP and SLM, were calculated by modeling the best fit in a least-squares sense (tip/tilt/piston for the TTP array and piston-only for the SLM) to the Zernike surface and computing the residual difference . No noise or other error was assumed in the model .

The membrane DM performance is plotted for the discrete points where information is available . The mirrors identified by the numbered bubbles 151 can be identified based on the same numbering in footnotes 1 and 2.

AMBS-AO DM systems pro-vide significant operational scalability in terms of the resulting mass -required for system operation , size of systems that can be supported, and ability to support both narrow- spectral-band applications allowing for large FORs or broad-spectral- band absolute phasing narrow FOR configurations . These benefits are due to the large piston and tip/tilt strokes that are possible, and now available — in response to my manufacturing specifications . Given that the MEMS architecture is based on silicon-wafer semiconductor fabrication technology developed for the telecommunications optical-switching industry, the ability to mass-produce , versus hand- make a DM, will ultimately result in considerable cost savings and reliability. MEMS mirrors consume fractions of microwatts , and again, their silicon wafers make them ideal candidates for placing all support electronics on-chip — thus removing the racks of off-DM electronics currently required. Maximum AMBS-AO size is limited by the silicon wafer . Currently, the largest wafer is 300 mm. Therefore relatively modest magnifications could support systems with large entrance pupils (meter-class) , if required.

Of the four candidate wavefront sensors mentioned earlier, the shearing type is particularly advantageous . It has relatively high bandwidth (given adequate sig'nal-to-noise ratio) , can be used on so-

called "simple" and extended complex scenes , and can measure both phase between mirrors , in a MEMS array, and wavefront slope .

The IQS type, however, for purposes of the present invention is advantageous in that it can evaluate the impact of both slope influence and piston influence — and supports the control of both in an AMBS-AO configuration . Also somewhat attractive is the pixelated phase-mask dynamic interferometer — based on its potential for high temporal update, in addition to the ability to work over a broad spectral band.

Closed-Loop Control Architecture

Closed-loop, versus open-loop, control of the DM in an AO system is important for two principal reasons . First, a closed-loop system utilizing some variation of position, integration and derivative (PID) control , can attain bandwidths 10 to IOO times the fundamental mechan- ical natural frequency of the DM, whereas an open-loop approach, in most implementations , is limited to the first mode .

The second major reason for closed—loop control is that, regardless of the DM approach used, the gains of the individual actuator elements in the DM vary by as much as 10% , resulting in several wave- lengths of error . Moreover, some elements can have hysteresis .

Part of the variation in gain between elements is constant and could be compensated for in a calibration lookup table (CLUT) . Much of the variation, however, is temperature sensitive — and time sensitive, a function of the aging of the electronics , where a CLUT is not practi- σal . The integration portion of a PID closed-loop control system significantly reduces the effect of gain variation, for, regardless of the gain, the DM will be driven to take on the desired wavefront error ("WFE") correction shape .

The only impact of gain variation is then upon the speed with which the DM acquires the shape . A time—response variation of 10% typically has no impact on the system performance .

Current and envisioned AO closed-loop control systems have thousands to hundreds of thousands of elements in the DM, with more degrees of freedom being sensed than controlled. As a result, a least-squares solution minimizing the errors being observed is traditionally used.

This approach also allows the designer to use weighting functions . These allo-w assignment of more importance or "weight" to some observed values than to others .

Minimizing the observed slope and phase erroar is more important than keeping DM actuator elements exactly at midrange . Therefore phase and slope minimization are advantageously given higher weight than actuator stroke . The level of weight given the actuator-stroke term must be balanced, however, or the system runs out of stroke .

AMBS-AO Approach

AMBS basics The AMBS-AO system of the present invention is based on AMBS technology currently under development . A laser-communications ("LASERCOM") transmit/receive ("TX/RX" ) implementation of the AMBS-AO elements is exemplary, although the approach is relevant for both active and passive single-detector and detector-array-based systerns .

In an active lout not necessarily LASERCOM sys tem, radiation 12 (Fig . 3) from a polarized laser source 10 is fed to a collimating beam expander 13 that provides control of the laser beam divergence and ultimately spot siz e on the target . The collimated. beam 18 proceeds to a polarized beam splitter 14 , which reflects the beam to the MEMS scan- mirror array 6 , 15.

The tip/tilt/]?iston MEMS array directs the beam 18 through an afocal lens 1 to the desired field position outside the optical system, and the MEMS array at the same time corrects for wavefront error . The afocal lens 1 provi des angular magnification , enabl ing the MEMS scan mirrors to address the desired full field of regard ("FOR") .

Individual mirrors 6 in the two-axis MEMS scan-mirror array ro- tate in tandem about θ _x and θ _y , pointing the external outbound beam 18 to the desired field point within the FOR . This enables a laser source , in an active system as described above , to project to that field point, and a detector 4 , 34 to see radiance from that field point, simultaneously .

The AMBS-AO optics of my invention are also operable as a passive system, i . e . simply receiving radiation 8 from external sources with-

out excitation 10 , 12 , 13 , 18 from my system. In either event, the incoming laser or other light beam 8 is transmitted through the afocal lens 1 , reflected by the MEMS array 6 , 15.

In an active system, the beam then passes back thorough the pola- rized beam-splitter 14 — which may be omitted for a passive system. A second beam-splitter 16 divides the incoming beam 8 between a LASERCOM detector 3 , 4 and a shearing wavefront sensor 36.

The ene-cgy in the former fraction of the beam is focused 9 by a lens 3 onto the LASERCOM detector 4. The latter beam f.taction , direc- ted to the shearing wavefront sensor ("WFS") , is focused 17 , 28 onto a diffraction g-rating 31 , and a resulting compound image 32 of the MEMS array is created by the MEMS reimaging lens 33 on the WFS detector array 34.

Further details in the implementation of this WFS appear below. Basic optical relationships for the major optical parameters of the AMBS-AO system are outlined here :

■ M, afocal lens magnification , M: l

■ Θ _x Θy/ MEMS-mirror mechanical full-scan angle ;

1. FOR = 2 M x θ _γ , sensor-system field-of-regard

■ Φ _MEMS , MEMS scan-mirror array diameter;

2. IFOV = M x 2.44 λ/φ^^, system diffraction-limited instantaneous field-of-view for wavelength

■ φ _{L ^} LASERCOM detector diameter

■ f _L , LASERCOM reimaging-lens focal length;

3. FOV S= Mφ _L /f _L sensor system field-of-view.

The MEMS scan mirrors move essentially in tandem, allowing the focal-plane array ("FPA") detector 4 to address the entire FOR within tens to hundreds of microseconds . Closed-loop control of the mirrors , so that they all point in the same direction (except for wavefront- error trimming" ) , is a notable part of the invention .

The total size Φ _MEMS °f the mirror array 15 , not the size D of the individual mirrors 6 , determines the number of photons collected — as

well as the diffraction limit of the optical system. This lattear point is true, however, only if phase errors that occur from row to row of the MEMS array are neutralized.

As the mirrors in the array rotate to address different locations within the FOR, a phase error Δ (Fig . 4) between rows of MEMS miirrors arises — leading to an even larger wavefront phase error 2Δ at the re- imaging lens 3. Unless this wavefront error 2Δ due to phase differences Δ between the rows of mirrors is mitigated, the diffraction .Limit is determined (Fig . 4) by the individual MEMS-mirror width D , not. by the MEMS array size Φ _MEMS .

The focused spot 11 , representing the instantaneous field odz view ("IFOV") , is Mf _L 2.44λ/D . The overall angle subtended by the corresponding field of view ("FOV") , outside tlie optical system, is 2Mθ _v .

A copending PCT patent application — number PCT/US2005/028"777 of Kane and Potter, which is hereby who J-Iy incorporated by reference herein — addresses the wavefront-phase- error issue . The approach of the present invention corrects for the wavefront phase error by maintaining the Z or so-called "piston" position of the mirror so that the op-fcical phase difference ("OPD") is maintained modulo 2π/λ as a function of MEMS scan angle, resulting in a corrected wavefront. As noted earlier, an even finer wavefront match is achieved by trimming the mirror singles to closely approximate the wavefront slopes .

Functional Organization Functional elements and concept design described below are g-en- erally applicable to all classes of ΑO systems . The AMBS-AO system approach is applicable to any active sensor system such as a LADAR. , or a passive intelligence, surveillance and reconnaissance ("ISR") sensor device utilizing a focal-plane array _r ^{or a} spectrometer , etc . An AMBS-AO LASERCOM system, however, is exemplary . Elements 1 ,

3 , 4 , 15 , 16 , 36 (Figs . 3 and 5) of the AMBS-AO system common to a.11 AO systems include the front-end optical, system, (the AMBS-AO TX/RX Lens 1 in this example) , deformable mirror 3- 5 located at a pupil , beam-splitter 16 , detector 3-4 , and wavefront sensor 36 — and a control pro ces- sor 123 (Fig . 5) .

In the operating paradigm of jny invention, the processor 123 receives two basic data sets for each MEMS mirror of index i, j ^:

SO

■ instantaneous sensed wavefronfc. slopes and offsets 121 (in each of the two orthogonal directions of mirror movement) δw/δx(i ,j.) , δW/δY (i ,i) , δw(i.,i) — corresponding to tip, tilt and piston respectively; and ■ instantaneous existing adjustment settings θ _x (i.,j.) , θ _γ (i,_j_) and

2 (i,i) of the mirrors — also corresponding to tip, tilt and pi ston respectively.

In response the processor 123 operates a least-squares analysis to find — and return to the mirrors — ideally a best available set of tip, tilt and piston commands θ _Xc θ _Yc (5. ,j.) , θ _x θ _γ (i. ,i) and Z (i,i) to bring the wavefront into internal coherence at the detector .

The active sensor elements, particularly the LADAR transmitter 35 , (as well as the receive detector- assembly 3 , 4) require essentiaLly straightforward optical design . Bas ed upon the teachings in this docu- ment, a person of ordinary skill in the field of optical-system desig"n will be able to perform that work .

In summary, central to the AO closed-loop control approach is ttie acquisition of wavefront slope and pliase error provided by the WFS fox- each of the (i,j) mirrors in the MEMS array, as well as MEMS-mirror- array tip/tilt and piston relative to a local datum. The latter are provided by embedded capacitance sensors .

MEMS scan-mirror actuator commands for tip/tilt and piston are then created by the control processo-tr based on a least-squares minimum- error solution, and transmitted to the MEMS DM. Local proportional , integral and derivative ("PID") control is provided locally on-chip, based on the control-processor position commands and embedded capaci- tive sensors .

Top-Level System Requirements The AMBS-AO system is capable o>f diffraction-limited λ/10 1 σ operation — but is also amenable to lower-performance operation at lowe^ar cost. Thus for example a preliminary wavefront-error ("WFE") budget may allow an overall system-level total 61 (Fig . 6) of 0.1 wavelength _r with confidence 1 σ, for the receiver (Rx) . The system total 61 is readily separable into manufacturing and. integration error residuals 62 , environmental error 63 and atmospherics correction error 64 residuals after correction by the AO system. Wave—

front error associated with cJLosed-loop AO control is captured by a "wavefront control" error 65 entry. Each of these four top-tier components is the residual after closed-loop wavefront correction at a specified bandwidth .

It is assumed that these error components are uncorrelated and random, and thus can be added in a root-sum-squared ("RSS") sense — and also that half the budget is required for the atmospheric-correction residual , and the remainder equally shared by the remaining three components . A similar but separate transmitter (Tx) budget would account for the transmitter.

The manufacturing and integration error 62 encompasses errors 72 from the afocal front end and beam splitters , errors 72 from the MEMS DM, errors 73 from the IiADAR receiver, and optical-alignment errors 7-4. Environmental error 63 includes thermal 75 , dynamic 76 and gravity-release 77 components . Wavefronfc-control error 65 may be broken down into MEMS DM 66, wavefront-sensor ("WFS") 67 and control-processor 68 constituents .

Preliminary top-level recfuirements (Table 1) for the major AMBS- AO Rx elements provide a starti-ng point for the system concept . The basis for each of these preliminary requirements is provided in the comment column of the table.

Table 1 : top-level AMBS-AO system requirements

* Residual after closed-loop WF correction at specified ban<dwidth

Wavefront Sensor

A shearing wavefront sensor is considered the nominally most-appropriate candidate component, for it has the ability to measure wavefront slope and piston between the MEMS mirrors , and can be used against point as well as extended scenes , as was the case for the Active Large Optic Telescope ("A LOT") program built by Itek ⁵ . The shearing wavefront sensor measures phase between two mirrors 6 (Fig . 7) in a MEMS array .

To accomplish this , first a monochromatic point source 81 illumi- nates the MEMS mirror array, whose piston error is defined as Δ . Part of the light from the source passes through a beam sputter 14 to the main imaging channel 38 , 3 , 9 , 4. The remainder is reflected 28 to a reimaging lens 83 that focuses the beam onto a diffraction grating 31. In the case of measuring piston , one of the diffraction orders is blocked (the -1 order in this example) and the resultdLng shear between the +1 diffraction order 81 and 0 (zero) diffraction order 82 is shown . Both these sheared orders 81 , 82 pass through a lens 33 that images the diffraction pattern from the MEMS array onto a deitector plane 34.

The result is two mutually displaced images . 0n«≥ image of the two segments is formed by the 0 order 82 , 92 (Fig . 8) , and another image by the +1 order 81 , 91. (The -1 order 93 , again , is blocked. )

A lower portion 96 of the diffraction pattern is due to a first of two mirrors , and a lower portion 94 is due to a second mirror . One overlapping area of the two patterns is received by a signal detector 95 ; and another part is received by a reference detector 97. More specifically, the signal detector D ₃ sees light from both mirrors 94 , 96 of the array : light from the 0 order via the second mirror 94 ; and that from the +1 order, via the first mirror 96. The reference detector, D _R , sees light from only the 0 and +1 orders of the first mirror 96. If the point-source object 81 moves with upward -Lmage velocity V (Fig . 7) , then the +1 order is Doppler shifted upward (as drawn) by an angular frequency ω, which is defined by V and the diffraction grating period P :

ω = 2πV/P (1)

⁵ M. E . Furber, C. D. Cox, "4-meter diameter Adaptive Optic al System Technology Demonstration" SPIE Vol . 2807/ 141 (1996)

Piston Δ between the two mirrors is determined bjf measuring the phase shift, θ , between power received at reference detector D _R , P _R , as defined by Equation 2 , and power received at signal detector D _S/ P ₃ as defined by Equation 3. Piston Δ is then defined by Equation 4.

P _R = P ₀ + P ₁ + 2 (P ₁ P ₀ ) ¹ ^ cos (ωt) (2)

P ₅ = P ₀ + P ₁ + 2 (P ₁ P ₀ ) ^1/2 cos fωt + 4πΔ/λ> (3)

Δ = λθ/4π (4)

The same technique used to determine piston between mirrors can be used to measure wavefront error over each MEMS mirroar in the array (Fig . 9) . In this configuration both the reference and. signal detectors 195 , 197 (D _R and D _s ) are located to respond to the same mirror 194 , and light from the zero, +1 and -1 diffraction orders 192 , 191 , 193 respectively are received by both detectors .

Assuming a wavefront error, W(x,y) over the surface of the MEMS mirror, the change in slope δW(x,y) /δx at the reference— and signal-de- tector locations D _R and D _s on the mirror can be found by measuring the phase difference in the a . c . signal as defined by Equations 5 and 6.

In a similar manner, change in slope along the Y axis can be measured with a second channel possessing a grating in an orientation orthogonal to the Y axis . Development of a WFS implementation concept is ones advisable starting point . Another is a performance analysis of wavefront error measurement accuracy as a function of signal-to-noise and measurement bandwidth .

P _R = P ₀ + 2P ₁ + 2 (P ₁ P ₀ ) ^1/2 cos fωt + 2πf/PδW(X _R ,-_T _R ) /δx) (5)

P ₃ = P ₀ + 2P ₁ + 2 (P ₁ P ₀ ) ^1/2 cos (ωt + 2πf/PδW(X _s ,ϊT _s ) /δx) (6)

Operation of a shearing wavefront sensor is dependent on the ob- ject point source 81 (Fig. 7) moving perpendicularly to the sensor system line of sight ("LOS") 38 with relative velocity V. In the case of

AIJOT , the object moved relative to the telescope assembly, something not amenable to several applications including laser communications or staring sensor systems .

Two options do not require any such relative motion of the object to the telescope . The first involves rotation about θ _y of all scan mirrors sinusoidally, thus "dithering" the LOS . V in Equation 1 is res tated in Equation 7 as a function of δ θ _γ /δt, afocal lens magnification M, and range to the point source R.

ω = (2πRM δθ _γ /δt) /P (7)

The other option involves driving the grating along the X axis , perpendicular to the LOS (Fig . 7) , where V in Equation 1 is restated in Equation 8 as a function of δx/δt :

ω = (2π δx/δt) /P (8)

MEMS Array

The MEMS scan-mirror array in the AMBS-AO system is based on a two—axis array, with piston control capability, currently under development . In an operational system, all drive electronics , high-voltage amplifiers and inner-loop PID controllers for each mirror in the array should be on-chip, as opposed to separate boxes of electronics off- chip . Additionally, each mirror in the array should be addressable through a serial interface . Requirements and goals for the MEMS sσan- miraror array are outlined in Table 2.

Commands to the MEMS scan-mirror array should be through a digital serial interface (Fig. 10) , with its elements 51-59 — resulting in output mirror motion 47. A final operational array should have the following on-chip functions :

■ multiplexing (within the MEMS beam-steering controller 44 , Fig . 5 lower view)

■ demultiplexing 51 ■ calibration look-up table 53

■ D/A converter 55

■ propor-fcional , integral , derivative (PID) controller 57 for each mirror

■ high-voltage MEMS actuator driver 58

■ embedded rotation and piston sensors 59. Immediately below the array of mirror pads 41 is a physical layer 42 , which may be called the "MEMS actuators and embedded rotation sensors" layer . This layer 42 includes the aotuators-S-sensors block 59.

Within the interface 46 , mirror motion is also fed back 47' to the PID cont-trollers . This return serves particularly to implement the integral and differential aspects of the control — as is generally understood in the related field of electronic control systems , and accordingly is not further detailed here .

Below the actuator/sensor layer 42 is another physical layer 43 , which includes in particular a CMOS mixed-signal PID controller 57 and high-voltage circuits 58. Remaining circuit blocks 44 , 51-56 may be distributed as between the lower two physical layers 42 , 43 ; or the main-logic controller 44 may be elsewhere in the chip .

The controller 44 sends multiplexed commands , for all the mirrors , to the submirror layers 42 , 43. After demultiplexing 51 , the system carries control data 52 for each mirror independently.

Each mirror, furthermore , has been calibrated independently . The calibration, stored in and applied from a lookup table ("LUT") 53 , considers not only mechanical variations within the mirror actuators and sensors 59 , but also optical nonlinearities and variations elsewhere in the system, particularly in the afocal lens 1. The many individual mirror-control signals from the LUT 53 accordingly are corrected for all known peirturbations from ideal operation .

The remainder of the system 54-59 , 47 , 47' too — although illustrated as unitary — is multiple, i . e . provides a separate, independent control-signal channel for each mirror . Following the digital-to-analog converter block 55 , analog mirror-control signals 56 flow to the individual PID control blocks 57.

These analog signals 56 control electrical signals from the earlier-mentioned high-voltage block 58 , which in turn produce mechanical signals from the previously mentioned actuators-and-sensors block 59. These mechanical signals physically move 47 the mirrors .

Table 2 : operational MEMS scan-mirror- array requirements

Scan-mirror arrays with tip/tilt/piston capability have been produced and demonstrated, as seen in fabricated prototypes 141 , 241 (Fig . 11) . A single 0.8 mm x 0.8 mm element (left-hand view) , fully covered with a low-inertia iaicromirror, has extended pads 142-144 for electri- σal characterization.. (The mirrors in my now-preferred array [not shown] are 1 mm square . )

A two-by-two array of actuators 0.4 mm on a side (lower view) carries a two-by-two array 241 of miσromirrors , batch transferred. The present implementation of the actuators is based on preengaged vertical comb drives in siliσon-on-insulator ("SOI") format, and a gimballess design demonstrated previously in large tip-tilt devices .

The fabrication process is a derivative of the multilevel-beam SOI-MEMS process . For small arrays of elements , back-side etching can be eliminated. Three masks are used in deep-reactive-ion etching ("DRIE") , to form shapes into the device layer of an SOI wafer and achieve isolated sets of vertical comb drives with "up" or "down" actuation . Low-inertia micromirrors are fabricated in a separate SOI wafer in a three-level selective DRIE process . Transfer and bonding of individually thinned micromirror plates onto the actuators , stiffened by a backbone of thicker trusses , was achieved by using custom-fabricated miσrotweezers . One such truss is the edge 144 (upper -view) of each mirror . Batch bonding and alignment of multiple mirrors for large-scale , high-fill-factor arrays is an on- going effort .

Closed-Loop Control

The AMBS-AO control architecture (Fig . 12) includes a 10 kHz inner proportional , integral and derivative ("PID") loop 43 , 115 , 122 , 124 , 131 utilizing th.e embedded capacitive sensors 115 of the MEMS array for feedback, and. a 2 kHz-to-3 kHz outer loop 36 , 121 , 8 that generates tip , tilt and piston commands for each mirror — based on a least-squares error-minimization solution. A summing junction 131 combines signals 124 from the least-squares controller 123 with signals 8 from the MEMS array.

The PID controller 43 , 57 drives 58 , 59 the MEMS array 115 , whose outputs include the physical output motion θ _x , θ _y , Z . Other outputs of

the MEMS array 115 are measured adjustment values 122 and the wave- front-error signals 28.

High-bandwidth (3 kHz) PID control of the two-axis mirror has been demonstrated, with closed-loop step input response 139 (Fig . 13) . A similar , although lower-bandwidtln , least-squares controller utilizing wavefront sensor and mirror-to—substrate location was used in the ALOT program .

Noπtraditional-Technology Assessment Skilled people in this field are advised to consider the following departures from established, convent±onal levels of industrial-production technique — or, in other words , recognized "risks" — associated with the major elements illustrated in the AMBS-AO functional blocks (Fig . 5) of my invention . ■ AMBS-AO TX/RX head (low risk)

■ Afocal-lens and beam-splitter assemblies have been demonstrated, and in some cases are commercial , off-the- shelf (COTS) modules .

■ Front-end concept designs should be developed . ■ MEMS DM (high risk)

■ Elements of the MEMS mirror have been demonstrated and a tip-tilt MEMS array should be demonstrated soon . A full- size array with all electronics on-chip is yet to be developed and is estimated as a $2 million to $4 million effort . ■ Operational MEMS scan-mirror assembly architecture , and a sσaleable tip/tilt/piston MEMS array concept with off-chip electronics , are to be developed . ■ Wavefront sensor (high risk)

■ While the ALOT program demonstrated a low-bandwidth shearing wavefront sensor , a high-bandwidth sensor has yet to be demonstrated in terms of provi-ding tip/tilt/piston control for a MEMS scan-mirror array.

■ The present invention contempL ates trading wavefront sensor approaches and developing an operation-traceable concept design .

■ Least-squares control processor (medium risk)

■ A leaεs t-squares control processor has been demonstrated on several adaptive-optics programs , including AHJOT . The moderate risk lies in identifying a processor with adequate I/O data rate and processing rate that can support AMBS-AO desired closed-loop control bandwidths .

■ LADAR receiver/transmitter (low risk)

■ Detection and transmission elements should recguire only straig-litforward optical design .

Other implementation recommendations

It is advisable to directly confront the fact that implementing this invention involves establishing several systems that are each independently major efforts , including at least these : ■ systems util izing large, lightweight, primary mirrors ;

■ systems involving propagation of laser light through, extended atmospheric turbulence;

■ systems that operate with large dynamic aberrations associated with varying; field angles or object distances; and ■ multiconjugate adaptive optics .

AMBS-AQ Concept Tirades , Design and Performance Analysis

After tentative establishment of requirements has been derived for the major elements , the next advisable step is concept tradeoffs , designs and performance analyses . Specifically, these components of such planning are advisable :

■ Optical System Design

This should include a paraxial layout of all elements in the AMBS AO system. System-level imaging performance, including wavefront error over the FOR, optical distortion and optical transmission , should be estimated.

■ Control System Design

A control-system concept should be developed and an associated performance analysis completed. Detail in the following control- system elements should be recognized:

1. WFS

Initially, a trade on the type of wavefront s ensor to be utilized should be performed (Shack-Hartman, Stochastic or Shearing are among the best candidates) . Once a WFS ap- proach has been selected, a concept design should be created and a performance model developed — based on the signal-to- noise ratio (SNR) , given radiometric assumptions of the particular final application .

2. MEMS DM In accordance with this invention , a tip/tilt./piston MEMS scan-mirror array DM is the recommended baseline approach for this system. A DM concept should be developed and a dynamic performance model created.

3. Control Processor S Algorithms A concept and algorithmic approach for a least-squares control processor should be developed. Estimates of performance, given the resulting processor and I/O requirements , should be made . ■ System Performance model A system-level performance model , including imaging performance as a function of SNR, temporal bandwidth and MEMS actuators/diameter, should be developed — utilizing the element performance models noted earlier.

Scalability and Manufacturing Issues

While major elements of my AMBS-AO system invention have been demonstrated, significant scalability issues will arise iin moving forward to an operational system . Some advancements , but within the reach of skilled senior people in this field, may be required in the

WFS , MEMS-array DM, and control-processor areas . During progress it is advisable to update the risk assessment .

It is recommended that one particular major element of the overall system, namely shearing wavefront sensing, be demonstrated as early as manageable , since it is felt that the closed-loojp control of a

MEMS array is a major risk area as noted earlier . Included in the

wavefront sensing demonstration should be a MEMS a-trray that is being viewed by tooth the WFS and a Twyman-Green interferometer .

Known wavefront errors should be induced on the MEMS array and verified b^ the interferometer . This is essential to assess the fidel- ity of the WFS measurement .

My OWΉ layout for the interferometer included a helium-neon laser 141 (Fig. 3.4) , beam expander 142 , afocal lens 143 and test mirror 144 , beam-splitter and reference-beam fold mirror 145 , SOO-mm lens 146 , wavefront-error camera 147 , and a point-spread function camera 148. The resulting interferograms make clear when a field is split between two different phase patterns 151 , 152 , or instead an entire field 149 is in-phase .

The foregoing disclosure is not to be understood as limiting or exhaustive . Rather, it is only exemplary of the invention, ^" whose scope is to be determined from the appended claims .