| Resolution: | Diffraction limited for  > 1.6 µm |
| Strehl ratio: | > 30% |
| Sky coverage: | > 80% |
Center for Astronomical Adaptive Optics, University of Arizona, Tucson, AZ 85721
Proceedings of SPIE conference on Optical Telescopes of Today and Tomorrow, 2871, 1996
ABSTRACT
1. INTRODUCTION
2. CLOSED-LOOP RESULTS WITH THE SODIUM LASER
3. DESIGN OF THE 6.5 m SYSTEM
4. EXPECTED PERFORMANCE AT THE 6.5 m TELESCOPE
5. CONCLUSIONS
6. ACKNOWLEDGMENTS
REFERENCES
ABSTRACT
The first images of astronomical objects have been obtained with a telescope exploiting wavefront compensation with
adaptive optics where the reference beacon was generated by laser excitation of mesospheric sodium. This was done using the
FASTTRAC II low-order adaptive optics system at the Multiple Mirror Telescope (MMT). FASTTRAC II is a prototype for a
full-scale adaptive optics system under construction for the 6.5 m telescope that will replace the MMT in late 1997. The 6.5 m
system is designed to provide correction to the diffraction limit of resolution in the near infrared (1-5 µm) with high Strehl
ratio and excellent sky coverage. This paper describes the new system and its expected performance in view of the achieved
performance of FASTTRAC II.
Keywords: telescopes, adaptive optics, laser guide stars, control systems
1. INTRODUCTION
In late 1997, the existing six mirrors of the Multiple Mirror Telescope (MMT) will be replaced by a single 6.5 m mirror,
which is now being polished at the Steward Observatory Mirror Lab.1 Earlier work 2, 3 has shown that an adaptive optics sys
tem using a single sodium laser projected co-axially with the telescope can provide imaging at the diffraction limit in the H
and K photometric bands over most of the sky. Such a system is under construction by the Center for Astronomical Adaptive
Optics (CAAO) in Tucson and San Diego.
The system is designed to operate in the wavelength region from 1 to 5 µm, where scientific return can be maximised for a reasonable expenditure of resources. A major design goal has therefore been the minimisation of thermal background radia tion. To that end, the system will incorporate an adaptive secondary mirror as the wavefront compensator, since such a design requires no reimaging optics at all between the telescope and the science camera. Constraints imposed by the system's limiting magnitude and the size of the isoplanatic patch require the use of an artificial beacon in place of a natural guide star to achieve significant sky coverage. The system has therefore been designed to operate with a sodium resonance beacon, generated by a laser tuned to 589.0 nm illuminating the mesospheric sodium layer. The laser and its beam-projecting telescope comprise the second major component of the system. The final major piece is the instrument package, mounted at the Cassegrain focus, which contains the science imaging array, and the wavefront sensing optics and cameras.
In order to test a number of components and system concepts for the 6.5 m design, a prototype system called FASTTRAC II has been built, and is now in operation at the MMT.4 In brief, FASTTRAC II corrects twelve parameters of the wavefront, namely tip and tilt over each of the MMT's six primary mirrors. Phase errors between the mirrors remain uncorrected. In the near infrared, where D/r0 ~ 2 to 3 for the individual 1.8 m primary mirrors, this level of correction is sufficient to restore imaging to about 0.3 arc sec, close to the diffraction limit of a single primary. Typical seeing at the site yields images of 0.75 arc sec, so FASTTRAC II provides a significant enhancement of image quality both in terms of resolution, and the Strehl ratio, which is typically increased by a factor of 3 to 4.
Over the past three years, work has also been undertaken at the CAAO to develop a sodium laser for FASTTRAC II and the 6.5 m system.3, 5 The laser has recently undergone a series of tests which have culminated in the first astronomical images taken with a sodium laser guided adaptive optics system. FASTTRAC II has been used with the laser to sharpen images by the same principle as will be used in the 6.5 m telescope. In both cases, the higher order wavefront errors are corrected from sig- nals derived from the bright sodium beacon. Correction of these errors alone results in a sharpened image with residual global image motion, which is not sensed by the laser beacon. As in all laser-guided systems, global tilt must be corrected from sig- nals derived from a field star. The important gain from the laser beacon is in sky coverage - since only global tilt needs to be sensed, very faint field stars suffice.
We describe these results more fully in Section 2, and then go on to discuss their implications for the design of the 6.5 m system, which is described in Section 3. Finally, we illustrate the expected performance of the 6.5 m system with some com- puter simulations based on results from the MMT and FASTTRAC II.
2. CLOSED-LOOP RESULTS WITH THE SODIUM LASER
FASTTRAC II has been used in closed loop with a continuous-wave dye laser to demonstrate the feasibility of using a
low-power sodium laser as an artificial reference beacon. Figure 1 shows the instrument in its current configuration, as it is
mounted at the pseudo-Cassegrain focus of the MMT. We have adopted the philosophy of the 6.5 m system in that light is cor-
rected within the telescope optics, and is then brought directly to the infrared science camera to minimise thermal background
contamination. Visible light is reflected off the dichroic beamsplitter at the entrance window to the dewar, and is sent into the
wavefront sensing portion of the instrument. A notch reflector in the beam separates out laser light at 589 nm and sends it to
the wavefront sensor itself; the remaining light is transmitted to a pupil-plane mirror which can be steered to select a star in
the field to be used as an indicator of the global tilt. In addition to the WFS and tilt sensor cameras, each channel has an acqui-
sition camera with a field of view of ~4 arc min.
For the present experiment, the laser was the 4 W dye laser described in Section 3.3, which will be used with the 6.5 m, with the beam projected close to the optical axis of the MMT array. Figure 2 shows the contour plot of a typical short-expo- sure image of the beacon as seen by the WFS camera; the beam profile has a full width at half maximum (FWHM) of 1.1 arc sec. In this configuration, the divergence of the laser beam is about 0.8 arc sec. The beacon has been compared to photometric standard stars, and appears as bright as a star with V magnitude ~10.
Since measurements of the laser return are insensitive to global tilt, only the relative wavefront tilt errors between the six apertures can be determined from the artificial beacon. The laser is therefore used to correct ten degrees of freedom of the per- turbed wavefront, while the remaining two, global tip and tilt, are corrected on the basis of measurements of image motion from a natural star, using the telescope's full aperture to collect the light.
Figure 3 shows sample images of the core of M13 taken in the K band with and without adaptive compensation by the laser beacon and natural tilt star. Corrections were applied at 30 Hz, a factor of 3 slower than the normal operating speed of FASTTRAC II. The reduction in speed is entirely attributable to an inadequacy in the tilt channel communication bandwidth in the computer hardware, which will in future be rectified. The imaging camera was a 256 x 256 NICMOS3 array with 0.093 arc sec pixels, giving a field of view of 24 arc seconds, and the exposure time in each case is 15 s. At left is the raw uncompen- sated image; the point-spread function has a FWHM of 0.74 arc sec, which indicates that seeing conditions at the time were very close to median for the site. The compensated image is shown after post-processing with the iterative blind deconvolu- tion algorithm6 in the right-hand panel. For this exposure, the laser was aimed at the center of the field of view (the beam is of course invisible to the infrared camera), and global tilt information was derived from a natural star separated from the laser by about 40 arc sec, outside the field of view to the upper left. In this image, the adaptive optics system succeeded in reducing the FWHM of the star images to 0.53 arc sec, which has been further reduced to about 0.35 arc sec by the deconvolution.
The reduction in image size is accompanied by an improvement in the Strehl ratio of more than a factor of 3, which leads to a significant increase in detectability. Sources separated by only 0.5 arc sec are detectable across at least 9 magnitudes of brightness.
A remarkable feature of this image is that it shows no detectable anisoplanatism across the field, for stars of more than 25 arc sec separation, even though the tilt star was more than 40 arc sec away from the centre of the field. This observation bears out previous results demonstrating that tip and tilt on the 7 m scale of the telescope are better correlated than the standard atmospheric theory predicts.3
3. DESIGN OF THE 6.5 m SYSTEM
The successful closing of the adaptive loop around the sodium laser signal has a number of important consequences for
the design of the 6.5 m system. Of prime importance is the enormously improved limiting magnitude of the system, which is
achieved because only two degrees of freedom must be measured with natural starlight, and the entire telescope aperture can
be used to collect photons for the measurement. In addition, because the timescale of evolution of the global tilt is longer than
for the high-frequency components of the aberration, the tilt sensor need not run at the same high bandwidth as the WFS cam-
era, which allows longer integration times to be used. We expect therefore to use guide stars six magnitudes fainter than could
be used in the purely natural star mode.
The large size of the isoplanatic patch compared to the prediction of the standard theory is very encouraging. This trans- lates directly into sky coverage, which is particularly important in the J and H bands, where the theory predicts only 10 - 50%. In this low limit, the amount of sky available for diffraction-limited imaging goes almost as the square of the isoplanatic radius.
Use of the laser to achieve the desired sky coverage means that an additional source of residual wavefront error is intro- duced, namely focus anisoplanatism. This error is fundamental in the sense that it cannot be reduced by technological improvements to the laser or the adaptive instrument. Multiple guide stars7 can in principle be used to reduce the effect, but that is beyond the scope of the present project. With that in mind, we have adopted the criterion that other sources of residual error, which can be controlled, should not contribute to the error budget more than focus anisoplanatism.
Guided by the above considerations, we can determine the design parameters of the adaptive optics system based on the requirements for final image quality and the characteristics of seeing at the site. These are set out in Tables 1 and 2. In addi- tion, as a design goal, we would like to use the secondary in a low-frequency chopping mode with a throw of about 10 arc sec. We are lead then to the results summarised in Table 3 for a number of crucial parameters.
| Resolution: | Diffraction limited for > 1.6 µm |
| Strehl ratio: | > 30% |
| Sky coverage: | > 80% |
Table 1: Requirements for image quaility and field of view for the 6.5 m adaptive optics system. Except where shown, numbers are quoted for a wavelength of 2.2 µm, roughly the middle of the range of wavelengths for which the 6.5 m adaptive system is being designed.
| Parameter | r0 (m) |  0 (ms) |  0 (arc sec) | d0 (m) |
| Measured value | 0.9 | 35 | 15 | 15 |
Table 2: Typical measured values of parameters of the seeing at the MMT. Values shown are again for a wavelength of 2.2 µ m.
| Number of subapertures: | 150 |
| Actuator stroke: | 100 µm (chopping mode) 10 µm (non-chopping) |
| System cycle time: | 1 ms |
| WFS CCD read rate: | 10 6 pixels per second |
| WFS read noise: | < 3 e` rms per read |
| Tilt sensor read noise: | < 5 e` rms per read |
Table 3: Critical parameters of the 6.5 m adaptive optics system, based on data from Tables 1 and 2.
Undertaking the manufacture of the adaptive secondary mirror which forms the heart of the 6.5 m system is a formidable challenge. The deformable surface is a piece of glass 2 mm thick and 64 cm in diameter, with a highly aspheric hyperboloidal surface which must in its unstressed state be extremely smooth on all scales. The considerable investment of effort required to make such a piece is only worthwhile if the rest of the system can exploit the advantages offered by this approach, namely high optical throughput and low background emission. In the rest of this section, we outline a design which meets these requirements. A schematic view of the system is presented in Figure 4.
The actuators are implemented by gluing small rare earth magnets to the back surface of the thin shell, and suspending a coil above each magnet. The back surface of the shell is also aluminized for two purposes - firstly to form a common plate for the 324 capacitors, and secondly as a reflective heat shield to minimise the transport of radiant heat from the coils into the facesheet.
The second layer of the sandwich is a thick spherical reference surface made of Zerodur, pierced by 324 holes which accept the cold fingers supporting the voice coils that drive the mirror. Around each hole is a metallized ring which forms the second plate of the capacitive sensor. Heat from the coils is transported back up the cold fingers to the third layer, a thick alu- minium plate to which the cold fingers are rigidly attached. Channels machined into the plate carry circulating glycol which removes waste heat from the telescope through pipes hidden behind the arms of the secondary support spider. Miniaturized electronic circuit boards carrying the voice-coil drivers and capacitive sensor electronics are also attached to the aluminium plate.
Within the science dewar is a second, small-format, infrared array which is operated as a quadrant detector to provide global wavefront tilt information. This detector will be an off-the-shelf 256 x 256 HgCdTe array, operating in broad band from the short wavelength limit of the device around 1.2 µm to 2.3 mm, where background thermal emission would start to contaminate the signal. Use of an infrared detector rather than a CCD to measure global tilt has the primary advantage that the image of the field star is much sharper because of the high-order correction provided by the laser, and this sharpening extends over a large isoplanatic area. Since the accuracy of a centroid measurement is proportional to image width, the uncertainty in the tilt measurement is correspondingly reduced. An additional advantage is that the vast majority of stars have considerably greater photon fluxes in the near infrared than in the visible, particularly in regions of the sky heavily obscured by dust.
Visible light is sent to the WFS, a Shack-Hartmann type with 150 subapertures feeding light to a 4-port thinned, back- side illuminated CCD with 80 x 80 pixel format, manufactured by EEV. Pixels are binned by a factor of 2, and the chip is operated with a customized readout scheme at 250 kHz per port pixel rate, giving a frame rate of 1 kHz. The frame speed of the WFS sets the fundamental cycle time for the whole optical servo loop, and thus determines the rate at which updates are applied to the figure of the secondary mirror.
As in FASTTRAC II, the whole WFS optical assembly is mounted on a precision translation table to accommodate changes in the distance to the sodium layer with zenith angle, and the distinct focal plane required when using a natural refer- ence source. It is also anticipated that changes in the mean height of the sodium layer on time scales of an hour must be accounted for. Regardless of any shift in the sodium layer, we must take steps to ensure that the InSb array and the WFS CCD remain confocal under changes in the gravity vector and ambient temperature. To do this, we will take advantage of the rigid frame to which both cameras are mounted. Relative shifts in the focal planes will be highly repeatable, as we have found in the case of FASTTRAC II, and can be measured initially by reference to natural stars and on future runs removed explicitly via a lookup table. After removal of local effects, we can then monitor the focus term measured by the WFS with a very low- pass filter, again on natural stars, and compile a second lookup table of the DC focus term. When using the laser, any depar- ture from the expected DC focus which emerges must be due to a mean height change in the sodium layer, and can be cor- rected by an appropriate translation of the WFS optical assembly.
The laser itself is mounted on a table attached to the telescope's yoke. From there, the beam is brought up to the elevation bearing, and directed down the axis to ensure that the beam does not wander as the elevation angle is changed. The beam is then reflected up the side of the telescope tube to a reimaged pupil plane opposite the secondary support hub, where, a piezo- driven fast steering mirror removes jitter from the beam on the basis of global tilt measurements from the WFS. Laser light is reflected across the top of the telescope to the secondary hub, where it is expanded and sent out through a launch telescope with 48-cm diameter aperture. The beam is focused to produce the smallest obtainable illuminated spot at the sodium layer.
Our plans call for the laser to be upgraded shortly after first light. At that time, the dye laser will be replaced with a new solid state laser, expected to deliver 10 W of frequency-stabilised light, currently under development in the Optical Sciences Center at the University of Arizona. The projector system will remain exactly the same. Even with the increase by a factor of 2.5, the power density at the sodium layer will remain well below saturation, so we will see the beacon brightness increase to V magnitude of 9 or brighter. The improvement in returned photon flux will allow us at least to double the system bandwidth, which will significantly improve the image quality at the shorter wavelengths of the J and H bands.
Once a set of actuator commands has been sent to the adaptive secondary, local feedback provided by the capacitive sen- sors ensures that the mirror surface responds in the shortest possible time. To defeat unwanted resonances in the thin facesheet, the local control will be implemented digitally. Each actuator has its own DSP built into the local circuitry, and communicates with the DSPs of neighbouring actuators. The algorithm to filter the actuator signal can then readily be opti- mized.
The third main area of digital control is the laser and its beam projector. A separate VME machine is responsible for maintaining the optical alignment of the beam as it leaves the telescope, and for stabilising the frequency, polarisation, and power output from the dye laser. This computer receives input from both pupil- and focal-plane cameras at the top of the tele- scope tube, and a power meter and a spectrograph located next to the laser.
We have adopted the approach of progressively automating the system with a number of ongoing software upgrades. The architecture of the software is designed from the outset to approach as closely as possible the ideal of one-touch operation, where almost all the decision-making about, for instance, the optical configuration of the system and the choice of guide star is left to the computer.
4. EXPECTED PERFORMANCE AT THE 6.5 m TELESCOPE
We have demonstrated that a single sodium laser of very modest power is quite adequate to achieve excellent correction
with a low-order adaptive optics system. Achieving the goal of diffraction-limited imaging with a high-order system will
require that a number of improvements in photon efficiency be made to deal with the smaller subapertures and faster band-
width of the new system. Emphasis is currently being placed on the wavefront sensing detectors; the EEV CCDs identified
above will improve the signal-to-noise ratio (SNR) of the wavefront and global tilt sensors by a factor of 12 compared to those
in FASTTRAC II. Significant improvements also need to be made to the beam projector, which currently operates with an
underfilled exit pupil, no jitter compensation, and some uncoated optics. We predict that filling the pupil and correcting the
outward beam on the basis of global tilt measurements from the WFS will reduce the spot size as seen by the WFS by a factor
of 2. Since the accuracy of a centroid measurement is proportional to image size, this improvement will be realised directly as
an additional improvement in the WFS SNR. Increasing the laser power, and coating the projector optics will provide a further
factor of 3 in beacon signal. Most of these improvements are now being implemented for testing on FASTTRAC II.
Computer simulations have been run to evaluate the performance of the 6.5 m system with the laser. A sample result is shown in Figure 6, which is based on an actual HST image of a portion of the Cygnus loop. The images show a 20 arc sec square of sky, representing the field of view of the adaptive optics science camera. The top view shows the image at 0.7 arc sec resolution, as it would be seen under good seeing conditions from the ground. The view as seen at 1.6 mm by the 6.5 m with correction to the diffraction limit by the laser is shown in the simulated image in the lower left. Here, the laser was aimed at the centre of the field with the tilt star 10 arc sec from the centre. In the lower right panel is the corresponding visible image that would be seen by the WFPC2 camera on the Hubble Space Telescope.
The two lower images look remarkably similar. In fact, the resolution of the HST, with a diameter of 2.4 m, at 0.5 mm is very comparable to the diffraction-limited resolution of the 6.5 m in the H band. In the centre of the field, where anisoplanatic effects are minimised, the final Strehl ratio for the simulated corrected image is 50%. Towards the edge of the frame the Strehl ratio drops to 25%, but without any significant degradation in resolution. It is rather the image contrast which is reduced.
5. CONCLUSIONS
The key features of the adaptive optics system for the 6.5 m MMT allow diffraction-limited imaging across a broad range
of wavelengths, and most of the observable sky, with very low light loss and contamination by thermal background radiation.
This will be achieved through the use of a deformable secondary mirror, and a single sodium resonance laser projected along
the telescope's axis. By using the laser, the system's limiting magnitude will be improved by six magnitudes, which represents
an enormous increase in the fraction of the sky open to high-resolution imaging.
6. ACKNOWLEDGMENTS
This work has been supported by the Air Force Office of Scientific Research under grant number F49620-94-100437.
Many thanks to Keith Hege for the deconvolution of the FASTTRAC II data. We are very grateful to the staff of the MMT
Observatory for their support in this most demanding work.
