This is G o o g l e's cache of http://ulua.mhpcc.af.mil/AMOS/1999_AMOSTechnicalConference/Berger_paper/Berger.html.
G o o g l e's cache is the snapshot that we took of the page as we crawled the web.
The page may have changed since that time. Click here for the current page without highlighting.
To link to or bookmark this page, use the following url: http://www.google.com/search?q=cache:SVOLehvEpJ8C:ulua.mhpcc.af.mil/AMOS/1999_AMOSTechnicalConference/Berger_paper/Berger.html+AEOS+completed&hl=en&ie=UTF-8&client=googlet


Google is not affiliated with the authors of this page nor responsible for its content.
These search terms have been highlighted: aeos completed 

Back to   Final Program
Home Page

AEOS ADAPTIVE-OPTICS SYSTEM AND VISIBLE IMAGER

Paul J. Berger, Daniel V. Murphy
Lincoln Laboratory, Massachusetts Institute of Technology, 244 Wood Street, Lexington, MA 02420

John R. Kenemuth, Michael L. Vigil
Air Force Research Laboratory, 535 Lipoa Parkway, Kihei HI 96753
and 3550 Aberdeen Drive, SE, Kirtland Air Force Base, NM 87117

David Witte
Pantera Consulting, 1700 Wells Drive NE, Albuquerque, NM 87112

Rene Abreu, David R. Dean, Charles D. Delp, Michael E. Meline, William M. Rappoport, William P. Zmek, Robert C. Allen, David A. Hansen, Sarma N. Gullapalli, Michael F. Marchionna, Louis S. Mendyk,
Christopher J. Musial, Conrad Neufeld, Ralph Pringle, Andrea M. Sarnik, William J. Swanson
Raytheon Systems Company, 100 Wooster Heights Road, Danbury, CT 06810

Mark A. Ealey, Thomas R. Price
Xinetics Inc., 37 MacArthur Avenue, Devens, MA 01432

David Briscoe
Logicon R&D Associates, 2600 Yale Boulevard SE, Albuquerque, NM 87119


ABSTRACT

This paper reviews the design and performance characteristics of the AEOS adaptive-optics system, consisting of a 2520-Hz frame-rate, low-noise CCD-based wavefront sensor, modified least-squares reconstructor, 941-actuator deformable mirror, tilt-control subsystem, and supporting optical assembly. Images are observed with a wide dynamic range, radiometrically calibrated, 512 x 512 pixel CCD sensor system, operating over the 0.7- 1.1 mm band and equipped with an image derotator and dispersion corrector. Currently installed in the AEOS Coudé Room, the system is being integrated with the 3.67-m diameter AEOS telescope and the Observatory Control System. Initial images were obtained on 29 July 1999.

Keywords: Adaptive Optics, Tracking, Imaging


INTRODUCTION

The purpose of the Adaptive-Optics System and Visible Imager is to provide precise tracking and compensated imaging to enhance the Air Force’s capabilities with the Maui Space Surveillance System (MSSS). The primary mode of operation is terminator mode, but some capability is desirable up to one hour after sunrise and one hour before sunset. The primary operational requirement is for highly automated operation, with the goal of being able to operate all the systems on the telescope with a single operator.

Analysis of the requirements by the Air Force Research Laboratory determined that compensation is possible over a wide range of parameter space of interest to the Air Force and astronomy community [1]. The studies recommended that a Baseline System should be developed, having ~ 900 actuators (30 actuators across the 3.67-m diameter), a bandwidth of 200 Hz, and a capability to compensate objects to visual magnitude Mv = 7. This was consistent with the then-current state of the art in sensor technology and deformable mirrors. It was also recommended that partial compensation be supplemented with image post-processing, using image enhancement algorithms, and that the baseline system be designed to accommodate future growth options. Future growth options include a 1500-actuator deformable mirror, a 350-Hz system bandwidth, a fast deformable mirror figure sensor, and the addition of a laser guide star system for compensating dimmer objects [2].

Raytheon Systems Company, formerly Hughes Danbury Optical Systems Inc., was awarded a contract for the design, fabrication, and integration of the system. Xinetics built the deformable mirror. Lincoln Laboratory supplied the CCD cameras for the wavefront sensor and tracker as Government-furnished equipment. The system Factory Acceptance Test was completed in February 1999. The system was then disassembled and delivered to Maui on 30 March. Installation in the AEOS facility began in April, and the system is currently undergoing integration and checkout in preparation for Site Acceptance Test in the fall. The first set of images, shown in this paper, was obtained on 29 July.


SYSTEM DESCRIPTION

System Block Diagram
Figure 1 is a block diagram showing the major components of the system and interfaces to other elements of the AEOS facility. The optical components are mounted on an optical bench in the Coudé Room. The light from the telescope is directed onto the optical bench by fold flat FF1, through the three pupil relays, PR1, PR2, and PR3, to the input pupils of the Tilt Sensor (TS), the Wavefront Sensor (WFS), and the Visible Imager (VisIm). The signals from the TS are processed by the Local Processing Electronics (LPE) to drive the Tilt Control Mirror (TCM). Signals from the WFS are processed by the LPE and Real-Time Reconstructor (RTR) to drive the Deformable Mirror (DM) via the Mirror Drive Electronics (MDE). The Optical Source Simulator (OSS) contains sources for alignment and calibration, which can be inserted by mirror RM1 into the path between PR1 and PR2.

Seven Optics/Experiment rooms around the Coudé Room are available for visiting experimenters. Beams can be switched to these rooms via the remotely controlled mirror M7. Experimenters have two choices for the type of beam they receive: (1) fold flats FF1 and FF2 can be rolled back to provide an uncompensated f/200 beam, with an angular magnification of - 14.5 and a focus at a height of 1.17 m located in the Experiment Room 11.5 m from the inside wall of the Coudé Room, (2) RM2 can be inserted in front of the VisIm, directing a compensated beam through pupil relay PR4 to provide a 10-cm collimated beam, with the primary mirror re-imaged in the Optics Room 3 m from the inside wall of the Coudé Room. PR4 is scheduled for installation in the spring of 2000.

The AEOS Adaptive Optics Control System (AAOCS) controls the adaptive-optics subsystems and records quick-look data during a test. The Silicon Graphics Inc (SGI) computer controls, and records data from, the VisIm. The Local Control Electronics (LCE) contain the logic and controllers for actuating the electro-mechanical mechanisms in the system. The Observatory Control System (OCS) maintains executive control of the telescope, adaptive-optics system, and other sensors. Data recorded by the adaptive-optics system and Visible Imager are transferred to the OCS for processing and archiving.

Optical Assembly
A photograph of the Optical Assembly is shown in Figure 2. The three pupil relays re-image the primary mirror of the telescope to the TCM, DM, and the input pupils of the TS, WFS, and VisIm. The beam diameters at the TCM, DM, and input pupils of the sensors are 4 cm, 27 cm, and 2 cm, respectively. The dichroic splitter (DS1) reflects wavelengths shorter than 700 nm to the TS and WFS and transmits longer wavelengths to the VisIm. The W/T selector has a dichroic splitter that reflects wavelengths shorter than 540 nm to the tracker and transmits longer wavelengths to the wavefront sensor. [When the system is used for tilt-only correction, the W/T selector has a mirror that reflects all radiation to the tracker.] The optical system is all-reflective, except for a BK7 window at the top of the Coudé room and the two dichroic splitters. All mirrors are coated with a protected, enhanced silver coating, Denton FSS-99. The optical quality of the path from the PR1 to the input pupil of the VisIm, measured in double-pass with an interferometer, is 0.09 waves at 0.6328 mm, with a flat mirror substituted for the deformable mirror. The band-averaged transmissions from PR1 to the input pupils of the Tilt Sensor and Wavefront Sensor are 0.72 and 0.86, respectively, and the transmission at 750 nm to the Visible Imager is 0.83.

Optical Source Simulator
The Optical Source Simulator (OSS) contains a broadband Xenon lamp and a diode laser, which can be switched into the optical path for alignment, calibration, and system checkout. The angular size and brightness of the broadband simulated objects can be selected remotely. Source sizes are 5, 25, and 50 mrad. Visual magnitude can be varied from 1 to 19 in unit steps. A uniform background with a radiance of 1.5 x 1014 photons/sec/cm2/ster over the 500 to 700 nm band can be added to the object to simulate conditions at dawn or dusk. A back-illuminated Air Force standard resolution target can be inserted into the path. The diode laser source provides a nearly diffraction-limited 0.85-mm, 0.02-mrad source for scoring with the Visible Imager. A static phase plate with an average wavefront error of ~ 0.3 waves, rms, can be remotely inserted in the beams to check system performance.

Wavefront Sensor
The wavefront sensor, shown in Figure 3, includes front-end optics, a 32 x 32 lenslet array, a 128 x 128 pixel CCD detector, and read-out electronics. The front-end optics include an input tilt mirror and a pair of Risley prisms for system alignment, a magnification group for re-imaging the input pupil on the lenslet array, and a moveable mirror for inserting an internal calibration source. Transmission from the input pupil to the focal plane is 85%. The input tilt mirror, Risley prisms, and the focus of the magnification group can be remotely controlled to optimize the registration of the wavefront sensor and deformable mirror.

Raytheon fabricated the 32 x 32 lenslet array by a proprietary etching technique. The f/10.8, 84-mm diameter lenses have a 100% fill-factor and produce diffraction-limited spots.

The MIT Lincoln Laboratory supplied CCD detector has 21-mm square pixels, a high quantum-efficiency (> 92% at 650 nm), high pixel-to-pixel uniformity, high transfer-efficiency (> 98.7%), and low read noise (< 9 electrons at a 2-MHz rate). Dark current is less than 2 nA/cm-2 at 23 oC. The device is cooled by a thermo-electric cooler, and normally operated at 0oC.

The camera can be run at frame rates from 500 to 5000 Hz. There are 16 output ports. The analog output is digitized to 12-bits at a gain setting of 3.2 electrons per digital count. The camera can be run in a co-add mode, with 2 x 2 pixels binned as they are read out to give "quad" subapertures, allowing the camera to be operated at a 2520-Hz frame rate with the 2-MHz digitizer. A dark-current map is recorded before a run, stored in memory, and subtracted pixel-by-pixel as the device is read out. A threshold of 2 digital counts is applied before calculating the centroid.

Real-Time Reconstructor and Servo Control
Phase reconstruction is accomplished through a matrix multiply of the 32 x 32 x 2 slope vector at rates up to 2520 Hz. The subsystem is designed to provide 16 different reconstruction matrices and 16 different servo coefficient sets, which can be changed on command within one frame time. A modified least-squares reconstruction is currently implemented, with the matrix coefficients generated by performing an array of poke tests. This minimizes sensitivity to non-uniformity in subaperture gain and actuator stroke. The slaving of the inner and outer actuators is optimized to reduce sensitivity to waffle build-up, an inherent instability with the Hartmann-Fried cross-gradient geometry. The 16 servo coefficient sets are selected to optimize the gain and phase margin for each bandwidth and frame rate combination. The servo includes a "leaky" integrator, with an ~0.3-sec time constant, to reduce waffle and piston modes from building up. Other features include full-aperture tilt removal on a frame-by-frame basis and focus offload to the secondary mirror of the telescope.

Deformable Mirror
The deformable mirror (Figure 4), manufactured by Xinetics Inc., has a 28.8-cm diameter, 2-mm thick ULE facesheet, supported by 941 lead magnesium niobate (PMN) actuators, spaced 9 mm on a square grid. The facesheet was polished to a figure of 0.045 waves, rms, and coated with Denton FSS-99. The 810 control actuators operate over ± 30 volts around a 70-volt bias and produce a stroke of ± 2.4 mm. Hysteresis is < 1% at 20oC, and the coupling at nearest neighbors is ~ 9%. The Mirror Drive Electronics (MDE) consists of 16, 6U chassis, each containing a DSP-based controller board and eight amplifier boards. The amplifier boards utilize an Apex PB-58 power booster amplifier, capable of driving eight channels. Driver bandwidth is 10 kHz. Each channel is fused for protection and has individual Zener diode protection to limit interactuator stroke to 2 mm.

Tilt Control Subsystem
The Tilt Sensor consists of a focusing lens and a 64 x 64 pixel silicon CCD detector. A f/4.7 or f/1.7 lens can be selected, giving fields-of-view (FOV) of 79 and 216 mrad, respectively. The CCD detector has 21-mm square pixels, a high quantum-efficiency (> 80% over the 500 to 700 nm band), high pixel-to-pixel uniformity, high charge transfer efficiency (> 99.99%), and low read noise (~ 13 electrons at a 2.5 or 4.22-MHz rate). Dark current is less than 4 nA/cm-2 at 25oC. The device is cooled by a thermo-electric cooler, and normally operated at 10oC.

The camera can be run at frame rates from 500 to 10,000 Hz. There are 4 output ports. The analog output is digitized to 12-bits at a gain setting of ~ 16 electrons per digital count. A selectable threshold is applied before calculating the centroid.

To achieve the maximum possible bandwidth for a given target size, the height of the track gate (in rows) is set to target size (number of rows with pixels above threshold) plus four rows. A burst read is made through the rows above and below the track gate. Only the pixel data within the track gate are digitized. For small targets only the pixel data in the two center ports are used for the centroid.

Figure 5 is a photograph of the Tilt Control Mirror. The faceplate is a 5.3-cm diameter Beryllium substrate, polished to 0.01 waves, rms, and coated with Denton FSS-99. The mirror has a center flexure and is driven in push-pull by four linear voice-coil actuators. In the 4-cm beam, the mirror angular dynamic range is ± 10 mrad per axis and the cage-mode pointing stability is < 1.9 mrad, rms. These correspond to ± 220 mrad LOS range and < 42 nrad, rms, LOS pointing stability output space. The mirror is designed to minimize reaction torque to the optical bench.

The Tilt Control subsystem includes an automatic spiral scan acquisition mode, used for initial acquisition of the target if it is not seen within the sensor’s FOV and for subsequent re-acquisition if track is broken. The scans can cover the entire ± 150 mrad telescope field-of-view, in 4.36 sec in the narrow field of view (NFOV) and 0.67 sec in the wide field of view (WFOV).

Visible Imager
Figure 6 is a schematic of the Visible Imager. The camera is a Pentamax 512 x 512 pixel CCD cooled detector array. The maximum frame rate is slightly greater than 4 Hz. The spectral range is 0.7 to 1.1 mm. The field of view can be remotely selected, from a choice of 51, 120, or 300 mrad. The optical system includes two six-position spectral filter and neutral density (ND) filter wheels, a remotely controlled dove prism for derotation, a dispersion corrector, and space for the addition of another camera. A NIST-traceable calibration source can be translated in front of the assembly for radiometric calibration of the camera.


SYSTEM PERFORMANCE

Each subsystem was individually tested to demonstrate that it met specifications. This was followed by system tests in January and February 1999. The analysis and test results presented below describe some of the important performance characteristics of the system.

Radiometry
Radiometry is the starting point in calculating the noise-equivalent angle of the tracker and wavefront sensor. The number of photoelectrons per second at the focal plane is equal to Io 10-0.4Mv Fl,Dl Ar Ttel Topt h / (hn), where Io = 2.98 x 10-12 W cm-2, Fl,Dl is the fraction of the black body spectrum within the band from l to l + Dl, Ar is the collecting area of the sensor, Ttel is the transmissions of the telescope to PR1, Topt is the transmission from PR1 to the sensor’s focal plane, h is the detector’s quantum efficiency, and (hn) is the photon energy. Table 1 lists the values for the tilt sensor and wavefront sensor.

Table 1. Tilt Sensor and Wavefront Sensor Radiometry

Sensor Band(nm) "Fl,Dl" Ar (cm2) Ttel Topt h (hn) N’pe (Npe per sec)
TS 500-540 0.053 1.02 x 105 0.8 0.65 0.8 3.85 x 10-19 (1.74 x 1010) 10-0.4Mv
WFS 540-700 0.189 1.44 x 102 0.8 0.73 0.9 3.21 x 10-19 (1.33 x 108) 10-0.4Mv


Tilt Correction
Figure 7 shows the calculated tracker NEA for a 100-Hz bandwidth as a function of visual magnitude for several target sizes. The following is a brief description of the calculation. The number of photoelectrons per pixel per frame time is given by N’pe (1/fRO – tx) / (St / IFOV)2, where fRO is the frame rate, tx is the image to storage transfer time (14 msec), St is the target size, and IFOV is the pixel IFOV. An IFOV of 1.24 mrad was used for the 2.5 and 5-mrad targets and 3.38 mrad for the larger targets. The frame rates were 8880 for the small targets and 8230 for the larger targets. The intensity-weighted centroid is calculated for all pixels within the track gate for each frame. The frame-by-frame values are then filtered by the servo system to determine the rms output error signal, the NEA in the absence of any input disturbance. Figure 7 shows that a NEA less than 50 nrad can be achieved for a 2.5-mrad target for Mv < 9 and for a 20-mrad target for Mv < 6.5.

The tilt-control subsystem provides a selection of 32 Type-2 control laws, spanning the required bandwidth range from 50 to 300 Hz, to optimize tracking performance under varying conditions. Several lower bandwidths selections (5 to 20 Hz) are also available for tracking dim targets (to Mv = 14). Bode tests were performed to determine bandwidth, phase margin, and gain margin of the loops. Table 2 summarizes the results for several bandwidths. Except for the 300-Hz bandwidth, the performance of the track loops meets the requirements for gain margin greater than 7dB and phase margin greater than 45 degrees. The last three columns of the table show the predicted performance of the loops in correcting for the telescope mount disturbances and the line-of-sight (LOS) jitter due to atmospheric turbulence. The atmospheric calculations are for propagation to zenith with a LOS slew rate of 14 mrad/sec (satellite in circular orbit at 550 km). The table does not include the noise-equivalent angle of the sensor, described above. For long-exposure imaging, it is desirable to keep the jitter below 0.25 (l/D) ~ 50 nrad. This criterion can be met with bandwidths above 75 Hz for correcting the atmosphere, but not for correcting mount disturbances. Mount disturbances are not only large, they persist to high frequencies. Additional work on the AEOS system will be needed to achieve lower jitter levels.

Table 2. Summary of Tilt Control System Performance

Selected Bandwidth (Hz)
Measured Bandwidth Az/El (Hz)
Phase Margin Az/El (deg)
Gain Margin Az/El (dB)
Calculated Atmospheric Jitter (nrad)
Calculated Telescope Jitter (nrad)
Atmospheric + Telescope Jitter (nrad)
5
3.8
3.3
162
641
661
50
50
50
60
45
13.5
7.8
68
212
223
100
101
96
48
50
8.1
8.6
35
123
128
125
124
120
48
50
8
8.5
28
107
111
150
154
150
44.3
45.5
7.2
7.4
23
95
98
175
172
173
45.6
45.5
7.6
7.9
20
86
88
205
207
204
45.4
46
7.6
7.6
16
82
84
300
301
293
27
28
4
4.2
11
84
85


Adaptive Optics Loop.
Figure 8 shows the noise-limited performance of the wavefront sensor. The following is a brief description of the calculations. The number of photoelectrons per subaperture per frame is (1.33 x 108) (10-0.4 Mv) (1/fRO – tx), where fRO is the frame rate and tx is the image to storage frame time (~ 44 msec). The rms noise is equal to sqrt(Npe + Nro sR2 + Npix (7600/fRO)), where Nro is the number of pixel clusters that are read out, sR is the read noise, Npix is the total number of pixels per subaperture, and (7600/fRO) is the dark current. In the co-add mode, Nro = 4 and Npix = 16. Figure 8a shows the signal-to-noise ratio (SNR) as a function of visual magnitude for several frame rates. For imaging a star or slow moving object, a system bandwidth of 50 Hz is sufficient to compensate for atmospheric turbulence and the wavefront sensor can be operated at a frame rate of 840 Hz, giving a SNR of 10 with a Mv of 7. For faster objects requiring a 200-Hz bandwidth, the same SNR can be achieved with a Mv of 5.7. The wavefront sensor’s noise-equivalent-angle, NEA, in waves is given by dsa/(4 ls Ns SNR), where ls is the sensing wavelength (0.62-mm average), dsa is the subaperture diameter (84 mm), and Ns is the null slope. The null slope is a function of image size on the focal plane, and can be fit by: Ns = 44 – 1.39 si, for si < 17 mrad, Ns = 20.37 – 0.5 si for (17 < si < 57 mrad). The noise-limited Strehl ratio is given by exp(- jwfs2), where jwfs = 2p NPr (ls/li) (pfB/fRO)1/2 NEA, where ls is the sensing wavelength, li is the imaging wavelength (0.85 mm), fB is the bandwidth, and fRO is the frame rate. NPr , the noise propagator in the reconstructor, is given by (2/3)(1 + 0.25 ln(Nsa)) = 1.76 for Nsa = 720 [3]. Figure 8b shows the Strehl vs. SNR. The solid curve is calculated for a 5-mrad target using the above analysis. The data, obtained during system tests with a 5-mrad target from the OSS, are consistently below the calculated curve and do not reach unit Strehl even for the highest SNR. It appears that there are some non-common-path errors that are not removed by the calibration procedures. There may also be some image spreading due to diffraction, fabrication errors in the wavefront sensor optics, and turbulence in the test enclosure. The dashed curve, the product of a non-common-path Strehl of 0.8 and a calculated wavefront sensor Strehl for a 7-mrad image size, is consistent with the measured points. A non-common-path Strehl of 0.8 is quite respectable at an early stage of system integration. Additional work is needed with the system to identify the source of the limitations, mitigate their effects, and optimize performance.

The dynamic response of the adaptive-optics loops was measured by a Bode step test procedure. In these tests, a step disturbance with a sinusoidal spatial pattern is applied at the deformable mirror summing-junction and the system reacts to correct the disturbance. The loop transfer function is computed from the normalized correlation of the deformable mirror drive vector on subsequent frames to the applied disturbance. The spatial frequency is expressed as a percentage of the Nyquist frequency, the inverse of the actuator spacing. The results of the Bode step tests for several system bandwidths and applied spatial frequencies, as well as the results of a frequency domain simulation, are summarized in Table 3. The measurements and simulation results agree, except for the gain margin, where more margin was achieved than predicted by the simulation. Other measurements show that the system latency is 855 msec, in agreement with design estimates.



Table 3. Summary of Adaptive-Optics Loop Performance

Measurements
Frequency Domain Simulation
Nominal Bandwidth (Hz)
Spatial Sinusoid % Nyquist
0-dB Open-Loop Crossover
Phase Margin (degrees)
Gain Margin (dB)
0-dB Open-Loop Crossover
Phase Margin (dB)
Gain Margin (dB)
50
25
45
57
19.3
100
12.5
97
61
15
98
64
10.7
150
12.5
150
48
13.6
145
54
8.4
200
12.5
195
36
10.9
186
44
5.9
100
25
81
60
19
96
65
10.8
150
25
128
51
15
140
57
8.6
200
25
154
49
19.5
179
48
6.3


Visible Imager
The Visible Imager was calibrated by inserting a high-quality reference beam into the 2-cm collimated beam space between the last element of PR3 and the dichroic splitter. The modulation transfer function (MTF) identified in Figure 9 as "uncompensated" was calculated from the measured point spread function. This MTF falls below the "perfect" MTF for the 2-cm aperture. Errors contributing to the uncompensated MTF include errors inherent to the Visible Imager and errors local to the source. Contributing factors were determined through a combination of measurements and analysis. A MatLab model, developed for this analysis, reproduces the measured "uncompensated" curve, showing that the analysis takes account of all important contributors. The calculated MTF identified as "compensated" includes only factors inherent to the Visible Imager. The compensated MTF corresponds to a camera-limited Strehl ratio of 0.93.


INITIAL IMAGES

The images were taken during an earlier-than-planned integration with the telescope. The primary objectives were to verify alignment with the telescope, obtain data for checking the slaving of the inner and outer ring of actuators in the reconstructor, and assessing long-exposure imaging with the Visible Imager. There was little time available for optimizing system performance. Unfortunately, no atmospheric measurements were made at the site during the measurement intervals, so quantitative comparisons with performance models are not possible.

Figures 10 shows a set of images taken on star number HR 6526, visual magnitude 4.4, on 29 July 1999 at 20:51 local time. The telescope azimuth and elevation angles were 57.4 and 79.7 degrees, respectively. Image exposure time was 0.4 s. A neutral density filter of 4 was inserted for the open-loop and closed-loop exposures. The open-loop and closed-loop Strehl ratios are 1% and 34%, respectively. The tilt and adaptive-optics bandwidths were 120 Hz. The measured jitters were 0.09 and 0.07 mrad, respectively, in azimuth and elevation.

Figures 11 shows a set of images taken on HR 5842, a binary star with visual magnitudes of 4.51 and 4.57, taken on 29 July 1999 at 20:30 local time. The telescope azimuth and elevation angles were –93.1 and 79.8 degrees, respectively. Image exposure time was 1.5 s. A neutral density filter of 4 was inserted for the open-loop and closed-loop exposures. The tilt and adaptive-optics bandwidths were 80 Hz and 105 Hz, respectively. The measured jitters were 0.072 and 0.078 mrad, respectively, in azimuth and elevation.

The measured separation of the two stars in HR 5842 is 0.66 (± 0.1) mrad, smaller than the 0.3 arc second (1.44 mrad) separation listed in the star catalogs. Otto Franz and Larry Wasserman of the Lowell Observatory calculated a separation of 0.135 arc second (0.655 mrad), based on recently published work on binary stars [4], in agreement with the measured separation.

Using a scaling-law model, summarized below, we calculate a Strehl of 0.28 to 0.39 for the single star images, in good agreement with the measured Strehl of 0.34. The calculated Strehl is the product of the following five terms: (1) Wavefront sensor noise (see Figure 8). The Strehl for a Mv 4.4 object is 0.78. (2) Deformable mirror fitting error. The estimated Strehl is exp(-0.35 (da/ro)5/3), where da is the actuator spacing in output space (12 cm) and ro is the atmospheric coherence length. In the absence of measurements, we estimate ro to be 12 cm, the average value for seeing at the time of day the star images were taken. This gives a ftting error Strehl of 0.70. (3) Residual atmospheric wavefront error. The estimated Strehl is exp (-0.95 (fG/fB)5/3), where fB and fG are the bandwidth and the Greenwood frequency. Using the Maui turbulence model, we estimate the Greenwood frequency to be 35 Hz for a zenith angle of 10 degrees and a zero slew rate. This gives an atmospheric residual Strehl of 0.89 for a 120-Hz bandwidth. (4) Jitter. For a long exposure image, the peak is spread by a factor [1 + 4.2(sD/l)2]-1, where s is the one-sigma, one axis jitter, D is the diameter, and l is the wavelength. For a jitter of 0.08 mrad and an imaging wavelength of 0.75 mm, the estimated jitter Strehl is 0.61. (5) Camera limitation. The measured Strehl is 0.93. Multiplying the five terms gives 0.28. Combining the terms by area addition gives 0.39.


SUMMARY

The AEOS Adaptive Optics System and Visible Imager were designed to provide precise tracking and compensated imaging over a wide range of operating conditions. The system Factory Acceptance Test was completed in February 1999. The system is currently undergoing integration and checkout in preparation for Site Acceptance Test in the fall. The first set of images, obtained on 29 July during an earlier-than-planned integration with the 3.67-m diameter telescope, demonstrate the potential of the system. More testing is planned to optimize system performance.


ACKNOWLEDGMENTS

This work was sponsored by the Air Force Research Laboratory under contracts F19628-95-C-0002 and F29601-94-C-0111. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the United States Air Force.

The authors thank Brent L. Ellerbroek of the Air Force Research Laboratory for technical advice and guidance, Bernard B. Kosicki and William H. McGonagle of Lincoln Laboratory for implementing many unplanned improvements to the CCD cameras, LtCol David L. Richards and Maj F. Joseph Bishop for their strong leadership of the AEOS program, LtCol (ret) James J. McNally for advancing his AEOS vision, 1Lt H. John Busque and Maj James R. Passaro for their energetic leadership of the adaptive-optics project, TSgt David Covey for his assistance in shipping the system to Maui, Valerie B. Skarupa and Joseph V. Tisi for their expert program control, Ira M. Schmidt for his skillful program management, Louis R. Damis for his resourceful scheduling, Geralyn M. Fischer for her many contributions and facilitating communications across the time zones, and James F. Iacovacci and Kenneth P. Kearney for their skills and patience in aligning and calibrating the optical system.


REFERENCES

1. M. C. Roggemann and B. L. Ellerbroek, Science Advisory Board Briefing, 1993 (unpublished)
2. R. Q. Fugate, B. L. Ellerbroek, C. H. Higgins, M. P. Jelonek, W. J. Lange, A. C. Slavin, W. J. Wild, D. M. Winker, J. M. Wynia, J. M. Spinhirne, B. R. Boeke, R. E. Ruane, J. F. Moroney, M. D. Oliker, D. W. Swindle, and R. A. Cleis, Two generations of laser guide star adaptive-optics experiments at the Starfire Optical Range, J. Opt. Soc. Am. A 11, 310 - 324 (1994)
3. D. L. Fried, Least-square fitting a wavefront distortion estimate to an array of phase difference measurements, J. Opt. Soc. Am. 567, 370 – 375 (1977)
4. S. Söderhjelm, Visual binary orbits and masses post Hipparcos, Astron. Astrophys. 341, 121 - 140 (1999)