Adaptive Optics for the Large Bi by fjhuangjun


									Header for SPIE use

                      Adaptive Optics for the Large Binocular Telescope
                                                   Piero Salinari
                                         Osservatorio Astrofisico di Arcetri
                                        Largo E. Fermi, 5, 50125 Firenze (I)


The Adaptive Optics system of LBT will be based on Adaptive Secondary mirrors and on other components, such as
wavefront sensors and wavefront computers, which are currently under development. I report the status of these
developments, the AO configuration foreseen for first light and discuss possible future upgrades and extensions of the first
light AO system.

Keywords: adaptive optics, adaptive secondary mirrors, wavefront sensors

                                                 1. INTRODUCTION

The adaptive optics system is an essential part of LBT, which is probably the first telescope designed from the beginning for
an extensive use of adaptive optics. One of the main features of LBT is that the interferometric combination of the two
beams can be obtained with a minimum number of mirrors, four in total, obtaining high efficiency and low emissivity. To
maintain this level of performances even with adaptive correction there is no alternative to using an Adaptive Secondary
(AS) in each of the twin telescopes. An AS also provides the possibility of an efficient adaptive correction at all the other
focal stations, except, of course, at Prime Focus.

The decision to design the telescope for using not yet existing “adaptive secondary mirrors” was bravely taken in 1992, and
had important consequences on telescope design, such as the choice of a Gregorian configuration. That decision was
followed by a long development program, aimed to transforming a concept 1 into a high performance and reliable device. We
are now close to the conclusion of that development and of course one of the main points of the present report will be, in
Section 2, the description of the status of the construction of the first adaptive secondary mirror ever made, built for the
new MMT. In Section 3 I will discuss more in detail the current ideas on the specifications and expected performances of
the two LBT adaptive secondary mirrors.

An AO system is not only made of a corrector, such as the AS; in fact its performances will be probably dominated by the
way in which the wave front error is measured and by other details of the AO control loop. Some recent developments show
that it is possible to improve performances in wavefront sensing: Avalanche Photo Diode quad-cells can significantly
improve tip-tilt sensing, while a novel type of pupil plane wave front sensor, the Pyramid Sensor2 (PS), can provide, in
closed loop, better performances than the well known focal plane Shack-Hartman sensor. In section 4 I will briefly report on
the comparison of the wavefront sensor options and try to provide initial estimates of the performance improvement that can
derive from adopting new sensors. In Section 5 I will briefly report on current ideas concerning the rest of the AO system, in
particular the general architecture of the control loop and its most critical component, the wave front computer.

In section 6 I will try to give a picture of the LBT AO system in its first implementation and I will attempt to estimate what
kind of performance we can expect. Section 7 will be devoted to describe improvements that are already foreseen, and to
report ideas on other possible extension of the AO capabilities of LBT.

                              2. TEST OF THE MMT ADAPTIVE SECONDARY

After the successful test6 performed at Arcetri on a prototype AS, called P30, with 30 actuators in two rings controlling a 20
cm diameter mirror, the MMT Observatory decided in 1998 to build an AS for the new 6.5 m MMT telescope. A contract
for all the electromechanical components and for the control electronics was issued to a consortium formed by the same
companies who had produced the P30 prototype for Arcetri: Media Lario, Microgate and ADS International. The Zerodur
back plate and the thin mirrors (one spherical for testing the system, one aspherical for use at the telescope) were fabricated
at the Mirror Lab of University of Arizona where also all the coatings and electrical contacts were applied. The personnel
of the Arcetri Observatory assisted the companies during all the phases of the work, from the initial design, to the solution
of technical problems, prepared the test software and tested the hardware with the companies.

The design of the MMT system was based on 336 actuators controlling the shape of a 64 cm diameter mirror 1.8 mm thick.
As a certain number of subsystems were improved versions of those already tested with P30, a new prototype, P36, was also
made, with 36 actuators in three rings on a 24 cm mirror 2 mm thick. The new hardware, tested on P36 7, provided, as
expected, better performances than the first prototype. The acceptance test of the electro-mechanical system, equipped with
the spherical thin mirror, was performed at the end of July 2000, during the Schloss Ringberg conference. The smiles on the
faces of the Italian-American team performing the acceptance test at Microgate in Fig 1 tell about the success of that test.
                                                                             The acceptance test consisted in running the
                                                                             system with commands for all the actuators
                                                                             corresponding to the correction of a turbulent
                                                                             layer moving at high speed (~50 m/s) carrying
                                                                             Kolmogorov turbulence corresponding to a
                                                                             Fried coherence length r0 ~ 13 cm at 0.55 m
                                                                             wavelength. The residual error in this
                                                                             simulation of the real operation of the AS was
                                                                             computed by comparing the input commands
                                                                             with the actual positions of the thin mirror read
                                                                             by the capacitive sensors. The total atmospheric
                                                                             wavefront error, 730 nm RMS was reduced to
                                                                             70 nm RMS, a value that takes into account the
                                                                             residual “fitting” error, due to the finite density
                                                                             of actuators, the time delay error, due to the
                                                                             finite bandwidth of the system and the noise in
                                                                             the control loop. The system passed the severe
                                                                             test in spite of the fact that 11 actuators out of
                                                                             336 where not working and, even more
  Fig.1: Happy end of the acceptance test. The Adaptive Secondary is         important, the spacing between thin mirror and
  barely visible under the bench. People from left to right: F. Wildi, G.    reference plate was around 60 m rather than at
  Brusa, M. Lloyd-Hart, R. Biasi, M. Andrighettoni, A. Riccardi              its nominal value of 40 m.

As this paper was written several months after the Ringberg Workshop, I can go a bit further: although the system had
already passed the acceptance test, there where a number of problems to be corrected. Among them: the cables of all the
actuators were cut too long, as visible in Fig. 1, some actuators were, as said, not working properly, some minor software
and communications problems were also detected. All the above problems were solved without much effort, while the
problem of the gap, which could not be reduced to less than about 60 m remained unsolved, although probably understood.
The rest of the summer was entirely spent in characterizing the system and in understanding the gap problem, which
resulted to be probably due to the gold contacts glued in the holes of the back plate, which needed to be re-done. The
possibility that there is a problem with the thin mirror shape remained open, although not very probable.

It must be noticed that the thickness the air gap between the reference back plate and the rear face of the thin mirror, t, (both
surfaces are optically polished to essentially the same radius of curvature) determines the amount of viscous damping, , in
the system (  t -3 ) and is therefore a crucial parameter for the control system. A gap of 60m causes a reduction by more
than a factor of three of the damping, compared to the nominal design value of 40 m. This affects the control loop gain and
therefore the bandwidth in an approximately proportional way. The system met the specifications already using a ~ 1 kHz
control bandwidth, corresponding to ~ 1 ms settling time, but would be capable of better performances if the gap could be
reduced to its design value.

The unit was disassembled at the end of the summer and the glass pieces sent to Steward Observatory to eliminate the gold
contacts, to be replaced with vacuum deposited Chromium contacts, and to proceed with optical testing on the
Shimmulator8, a special optical bench to test the MMT AS. At the time when this paper is being written, the optical tests are
foreseen for early 2001.

What conclusion can we draw from the initial phase of testing of the MMT AS? The main result, apart from the detection
of small technical problems, is that it works exactly as expected in the conditions in which it is. In other terms the basic
physics and the complex aspects of the massive parallel control system seems to be fully understood. The results obtained
are those predicted on the basis of theory and of the P30/P36 results for the gap thickness obtained in practice by the MMT
AS, in spite of the tenfold increase of number of actuators. Debugging of the hardware and of the software will continue,
optical testing will tell us about more (still undetected) mistakes and problems, but the basic result is already in hand.

                             3. DESIGNING THE LBT ADAPTIVE SECONDARY

The preliminary design of the LBT secondary mirrors is described in detail in a recent paper10. Although conceptually very
similar to the one of the MMT AS, the LBT units will be different in many important aspects, because the LBT mirrors are
concave instead of convex, larger and with more actuators, with different geometry of the supports and so on. Of course we
are trying to use the largest possible number of technical solutions already tested in the MMT design, of which we already
understand advantages and disadvantages, but we will have to modify something and also to invent something new. I list
here a number of aspects currently under study:

Actuator density and pattern: The basic input parameter for an AS is the density of actuators9, which determines the AO
performances (fitting error, time response) but also important technical parameters (mirror thickness) and system
complexity and cost. Most of our current studies are devoted to optimize the actuator physical separation in a range between
~32 mm, typical of the MMT AS, and ~25 mm (total number between ~600 and ~900 actuators per mirror). Not a large
                                                          range indeed, but corresponding to significantly different
                                                          performances and to different technical solutions, due to the
                                                          highly non-linear interrelation among parameters.

                                                          A highly non-linear behavior is present not only in the AS itself,
                                                          but also in the whole adaptive system. The performances of the
                                                          secondary in terms of fitting error and settling time are crucial if
                                                          one wants to use the advantages of the Pyramid Sensor. As
                                                          discussed later, in closed loop operation a small difference in the
                                                          total residual error after correction changes very significantly the
                                                          magnitude of the star on which the loop can be closed with a
                                                          Pyramid Sensor. This pushes for a system with a high density of
                                                          actuators, to give small residual errors at short wavelengths,
                                                          where the wavefront sensing will be done. The equally strong
                                                          non-linearity of the AS local control system pushes for a lower
                                                          density of actuators, to increase the damping and therefore the
                                                          stability of the system and its speed. The current work is largely
                                                          devoted to find the correct equilibrium point for the above
                                                          contrasting requirements on the basic input parameters.

                                                          Another aspect, which will probably be modified with respect to
                                                          the MMT AS, is the actuator pattern, optimized to increase
                                                          actuator density in the central area, rather than at the edge of the
                                                          mirror. This helps in reducing the thermal gradients due to the
                                                          correction of axially symmetric aberrations such as focus and
                                                          spherical, and helps in maintaining a higher viscous damping
                                                          close to the edge, where it is naturally lower due to the opening at
                                                          the edge.
  Figure 2: An exploded view of the LBT secondary
  mirror unit                                             Actuators: The main effort here is in improving the thermal
                                                          aspects. Although the temperature differences between actuators
during adaptive correction is within the specifications in the MMT AS, there is not much margin for correction of static
errors. If we want to correct possible large scale figure errors of the thin mirror itself, or if we need to correct at the
secondary for primary focal length or conic constant, we will exceed the maximum allowed temperature difference on the
mirror very easily. The primary focal length and conic constant are very tightly specified in LBT, due to the requirements of
interferometry with wide field, and therefore very difficult and expensive to obtain in the polishing of the primary.

Digital control boards: A new board layout is necessary for reasons of physical space. Some important components, in
particular the Digital Signal Processors (DSP) need updating due to the heavier computational work needed for the larger
number of actuators, 600-900 versus 336. The computational work of each DSP, whose number is one half of the number of
actuators, increases linearly with the number of actuators. Fortunately DSP technology is improving faster than AS
development, and new low cost DSP not only can provide the required speed increase of about a factor of three, but also
operate with more bits (32 rather than 16 for the MMT system) and in floating point instead of integer mode. Although the
study of the new boards is not yet completed, the perspective is that the new boards will be considerably faster than those of
the MMT AS, a positive feature not only for the control system of the secondary, but also because it allows to perform
wavefront reconstruction directly in the AS controller. The new boards in fact will be able to perform about 50 GFLOPS,
therefore will be much faster than any other affordable wavefront computer. Of course also the data link needs an update,
and will be replaced by a more recent model of the one used in the MMT AS.

Analog current regulators: The increase of the number of actuators pushes for a better power efficiency of the current
regulators, which are the devices requiring the highest current and delivering most of the heat on board the secondary unit.
The study of an equally fast but more efficient device is in progress.

In March 2001 a Preliminary Design Review will evaluate the results of the current study activities and will define all the
basic parameters. An entire prototype, probably similar to P36, will be built to test all the new or modified components. We
expect this prototyping phase to need about 6 months, after which the production of the complete set of components and the
assembly of the first unit should take about one year. The first LBT AS should therefore be ready for full system laboratory
test in the second half of 2002, and should be installed on the telescope in spring 2003, while the expected date of first light
is in mid 2003. The assembly of the second unit will start after delivery from the factory of the first one.

                                             4. WAVEFRONT SENSORS

                                                                     The choice of the wavefront sensors for the basic
                                                                     configuration of the LBT AO system is currently
                                                                     undergoing significant evolution with respect to the
                                                                     initial baseline of a couple of years ago, which was
                                                                     based on a Shack-Hartmann (SH) sensor for the high
                                                                     orders and on a CCD based Tip-Tilt sensor. The much
                                                                     better performances achieved in practice at the telescope
                                                                     by Tip-Tilt sensors using Avalanche Photo Diodes
                                                                     (APD) on one hand and the equally important sensitivity
                                                                     gain that theory and laboratory measurements predict for
                                                                     Pyramid Sensors (PS) in comparison with SH sensors
                                                                     are driving the change of approach. This change, aimed
                                                                     to better sensitivity and therefore to better sky coverage,
                                                                     is of course not without a penalty, consisting in a
                                                                     considerable re-design of the Acquisition, Guiding and
                                                                     Wavefront sensing units (AGW units) whose design,
                                                                     schematically shown in Fig. 3, is in progress at the
                                                                     Observatory of Potsdam14.

                                                                     Tip-Tilt sensor: A direct comparison at the telescope
                                                                     between the two types of sensors, based on a CCD or on
                                                                     four APD has been performed at the 3.5 m telescope at
Figure 3: The layout of the Acquisition, Wavefront sensing           Calar Alto and, more recently at the Galileo telescope.
and Guiding unit (see reference 14 for details)                      At Calar Alto the test was done with the APD sensor
developed at ESO13, while at the Galileo an APD sensor originally developed at Arcetri and modified by the Padova group
was used. In both cases the APD sensors could work on stars of about two magnitudes fainter than previous limits obtained
with CCD based Tip-Tilt sensors. Not only the APD systems have essentially “zero” readout noise, but they offer much
faster response and they have now a higher quantum efficiency, typically 0.7 in the red region, than in the past. There are
still a number of technical aspects that need further examination, in particular the optics of the APD quad cell, but the
difference in performances is such to deserve further efforts.

The High Order sensor: Recent numerical simulations11 and also laboratory measurements4, 5 have confirmed the original
indication3 that a PS can achieve much higher sensitivity that a SH in closed loop AO operation. The most important
progress in the recent understanding of the performances of the PS is in the fact that the advantage, in closed loop, remains
very significant even in conditions of partial adaptive correction. In this case the wavefront sensor measures the partially
corrected wavefront, with a phase variance reduced from the uncorrected atmospheric value to the partially corrected value
c2, and allows a reconstruction of the wavefront with a reconstruction error of rec2. While c2 depends on the adaptive
system performances, including therefore the reconstruction error and the effects of corrector fitting error, loop time delay,
and other terms, the reconstruction error term depends on sensor response to small local slopes, on photon shot noise and on
error propagation in the reconstruction algorithm. In a “perfect” AO system rec2 is therefore the limit to the achievable
correction, depending essentially on photon statistics. In Fig. 3 one can see an example of the different behavior of a PS and
of a SH in identical conditions. The PS can achieve the same reconstruction errors with lower photon fluxes than the SH,
provided the total residual error c2 is small enough. The average gain in star magnitude for the case of Fig.3 is about two
magnitudes, but it can be higher with different assumptions.

  Figure 3: Comparison of Pyramid Sensor and Shack-Hartmann sensors in closed loop. The three plots report the
  results of the simulations11, based on diffractive calculations, showing the magnitude of the reference star needed to
  obtain a value of the reconstruction error, rec2, as a function of the total variance of the adaptively corrected
  wavefront measured by the wavefront sensor, c2. The three plots refer to different values of the reconstruction
  error, respectively 0.5, 1 and 2 rad 2 at 700 nm. The curves labeled PS 0,1,3 in each plot refer to three different
  values of the modulation, with radius respectively 0, 1 or 3 times the diffraction limited image “size” /D. The
  modulation is circular, so that the image is sweeping the four facets of the pyramid. The curves labeled SHS refer to
  the Shack-Hartmann sensor, for which modulation is of course absent. The parameters for this particular simulations
  are: telescope diameter 6.5 m, wavefront sensor with 16x16 subapertures, 0=700 nm, r0(0)=37 cm, sampling time
  1 ms. The photon fluxes and magnitudes do not take into account optical losses in the atmosphere, telescope and
  optics. Detector readout noise is not considered.

A direct comparison of performances at the telescope of the two types of wavefront sensors will be possible in the near
future at the Galileo telescope, where both types of sensors are already included in the AO module12, currently in
commissioning. It is also possible, in principle, to test the Pyramid sensor at the MMT telescope, where the presence of the
adaptive secondary will provide conditions extremely similar to those planned for LBT.
                                        5. THE WAVEFRONT COMPUTER

We have discussed the basic ingredients of the LBT AO system, but equally crucial is implementing a fast and stable
control loop for a high order adaptive system such as that of LBT. Here the main problem is that of performing the massive
calculation needed for the wavefront reconstruction in a time short enough not to add significantly to the loop cycle time,
which is generally determined by the need of accumulating a sufficient number of photons for adequate wavefront sensing.
In the previous paragraph we have emphasized the necessity of working on faint reference stars, but this is not the only
priority. A lot of important astrophysical work, for instance searches of extra-solar planets, will be done near very bright
stars, therefore with the possibility of fully exploiting the high order correction provided by the adaptive secondary mirrors.
In this case control loop cycle times of less tan one millisecond are in principle possible and help significantly in achieving
the extremely high Strehl ratio desired, for instance, for efficient nulling interferometry. It is therefore important to try to
limit the computational delay to something like 0.1 ms, during which the wavefront slope vector, S, measured by the
wavefront sensor, must be multiplied by the reconstruction matrix R. For high order work the rank of S and R is ~103 and
the number of operations needed are ~10 6 multiply-and-add. More complex reconstruction schemes can use more than a
single wavefront slope vector, to apply time filtering, and require a correspondingly higher number of operations. The
computational requirement is therefore in the range 10-50 x 109 multiply-and-add operations per second, well above the
performances of currently available commercial wavefront reconstruction computers.

We have explored different possible solutions, from multi-processor computers to special digital cards, but we seem to have
a powerful solution at no added cost in the multi-DSP unit used for the control of the actuators in the Adaptive Secondary.
This unit is made of a large number of DSP (one DSP controls two actuators) which can receive simultaneously the vector
S, and which have enough memory to hold two lines of the matrix R and multiple versions of the vector S, received at
different times, if a more elaborate scheme will be used. Each DSP can therefore make a small number of scalar products
between vectors and then make the results available to all the others. By suitable low level programming this operation can
be done, without perturbing the other functions of the DSP, at a speed of > 50 GFLOPS, therefore satisfying our most
restrictive requirements. We are currently studying the details of this approach in the frame of the mentioned refurbishment
of the MMT secondary control unit.

The much simpler calculation of the vector S from the wavefront sensor detector output (several thousands of operations per
cycle at most) can easily be done in a few tens of microseconds by a computer of the class of a modern PC. The same PC
can also manage a number of other functions operating under a real time operative program. Other functions of this PC are:
sending the vector of computed slopes to the Adaptive Secondary control unit, refreshing the matrix R in the DSP array,
storing returning housekeeping and wavefront information, communicating with the Telescope Control System and with the
service computer that will be used for elaboration of real time diagnostics and optimization of the AO loop.

                                  6. THE FIRST LIGHT AO SYSTEM: GOALS

In first light the LBT adaptive system will only use natural stars as reference sources and will be therefore limited in
performances and in sky coverage by the sensitivity of its wavefront sensors. As discussed briefly in previous paragraphs,
the wavefront sensor performances depend strongly on global system performances, and the result is that, when working
with natural stars, the performances of the entire system have to be pushed towards the limits, although, of course, with a
grain of salt.

Currently we consider that the technically safe limit is represented by an AO system based on the components discussed in
the previous paragraphs:
- A high order adaptive secondary.
- A tip-tilt sensor based on an APD quad-cell.
- A Pyramid wavefront sensor.
- A hyper-fast secondary controller used also as wavefront computer.
- A powerful multiprocessor PC (or a small PC cluster) for the control of the AO system.
An AO system of this type can in principle provide not only a better sky coverage in the near IR than most natural star AO
systems of the current generation, but also excellent correction even at visible wavelengths.
Our current goals are to reach excellent Tip-Tilt correction on stars of about V magnitude 18 and, in good seeing conditions,
full adaptive correction in the NIR near reference stars of V magnitude between 15 and 16. This requires of course that
really low noise CCD detectors (~ 1 electron readout noise) will be available. It must be noted that, although the correction
will obtain only a modest Strehl ratio at the sensing wavelength on stars that faint, the Strehl ratio in the observing band (J,
H, K) will be very high. If the above performances will be obtained in practice, the sky coverage at low galactic latitude in
the K band will be very significant, and even at high galactic latitude a non negligible sky coverage (a few percent) would
be granted.

An important consequence of pushing the LBT AO system to the best possible use of natural reference stars with sensing at
short wavelength, is that it will be capable of an higher level of correction than most existing or planned systems at visual
wavelength. This offers the possibility of performing types of observations not possible on most other large telescopes at the
date of LBT first light.

It is not easy to go beyond these qualitative considerations at the current level of definition of the AO system, but realistic
expectations on performances of the various components should be available already towards the end of 2001. By about the
same date we should also have collected sufficient data on the statistics of the turbulence above Mt. Graham with the
Generalized Scidar currently in construction and scheduled to start measurements at the Vatican Observatory in spring
2001. By then more accurate performance simulation will become available.

                                           7. WHAT WILL COME NEXT?

The answer is simple: the next important step will be the use of artificial reference stars. This is the first development
already foreseen in the current design, and one of the aspects on which most of the LBT partners are actively at work.

The high order wavefront sensor (Pyramid or Shack-Hartmann) used for the natural stars needs only a modest addition,
already considered in the design of the AGW units, to be able to use an artificial reference star produced by resonant
scattering on the mesospheric Sodium layer at ~ 90 km height. The telescope is already designed to support the optics for
the launch of the laser beacon. The problem that is still unsolved is that of a powerful (~ 10 W), reliable and affordable
laser at the sodium D doublet wavelength (589 nm). Various groups in the world, including some in the LBT consortium,
are currently working on developing the relevant technology and there is hope that this will become available not far from
the LBT first light. A single Sodium reference star per pupil will greatly improve the sky coverage of the LBT AO system in
the NIR, but would not help much at the shorter wavelengths where the intrinsic error in the wavefront reconstruction (cone
effect) would not allow to reach a reasonable Strehl ratio.

A different approach to artificial reference stars is that of using the Rayleigh scattering on air produced by the propagation
of a strong laser beam. In this case of course the wavelength of the laser is not constrained and it is possible to use lasers
already developed for industrial applications, in particular UV laser at a wavelength around 350 nm. This approach is under
active study at Steward and at Arcetri, where a variety of different sensing techniques is under investigation, and appears to
be promising from the technical and from the financial point of view, although there may be limitations for some of the AO

The main problem of the Rayleigh artificial stars is that they cannot be produced at heights above ~ 30 km, as the air density
becomes too low for a reasonably efficient Rayleigh scattering. It is therefore crucial to understand in detail the statistical
properties of the vertical distribution of the turbulence above the LBT site (this was the main reason to start the Scidar
measuring campaign at the Vatican Telescope). By the end of 2001 we should have sufficient statistics to start evaluating
which applications can make use of Rayleigh artificial stars for what percentage of the time. In practical terms it is the
frequency of occurrence of high (> 12 km) turbulent layers and the associated power that will determine if Rayleigh
artificial stars can replace or not Sodium stars altogether. This is clearly possible only if the number of nights affected by
high turbulence is moderate, or if the turbulence power in the high layers is small enough to allow for good correction at
long wavelength (e.g. > 2 m) for a large percentage of time.

Even with artificial reference stars (Sodium or Rayleigh, or both) it will be necessary to use natural stars to correct tip-tilt
only or, more probably, a few low order modes that are impossible or difficult to measure on artificial stars. The effort
placed on the natural star wavefront sensing remains therefore important even if laser guide stars become available.
Rayleigh lasers can certainly be used to correct low level turbulence, a fact that opens in any case interesting perspectives.
The simplest application could be to correct only very low turbulence, say only the dome turbulence and that in the
mountain boundary layer. In this case only part of the image degradation would be removed, but this would be done over a
relatively wide field, a few minutes of arc in diameter, over which the seeing disk would be reduced significantly, by a
factor close to two. This wide-field partial-correction mode, associated with tip-tilt correction, would greatly improve
performances in the seeing limited modes of MODS and Lucifer, both in imaging and in spectroscopy.

Alone (with the above mentioned limitation on turbulence height) or, better, in combination with a single Sodium star for
each pupil, Rayleigh stars can be used to produce full tomography of the atmospheric phase error. The information obtained
in this way can be used to completely remove the intrinsic limit of the single Sodium star, the cone effect, and therefore
makes it possible to extend to the short wavelengths the sky coverage advantage that the Sodium star can provide directly
for longer wavelengths. As the Adaptive Secondary will have very good performances even in the V band, this seems to be
a quite natural evolution, if a Sodium laser per pupil becomes part of the “basic” LBT AO system.

A further way of exploiting the atmospheric tomography obtained with Rayleigh stars, alone or combined with Sodium
artificial stars, is of course multiconjugate AO. In this case further wavefront correctors are needed in addition to the AS.
The further correctors (one or two per beam) would have to be placed at the conjugate planes of peaks in the average
turbulence vertical profile. In this way not only correction at all wavelengths becomes possible as in the case above, but also
the corrected field would become substantially wider. Clearly a multiconjugate AO system is very attractive for
interferometry, which could so exploit the wide coherent field achievable with LBT at all wavelengths. The insertion of
additional correctors in the beam combination is possible in various ways, for which some preliminary designs are already
available, but, again, a more detailed analysis requires more knowledge of the vertical profile of the turbulence.

In conclusion, the design of the first light AO system for LBT is in progress, with the aim, first of all, of providing the best
possible performances using natural reference star in its initial implementation, but also keeping in mind much more
ambitious and possibly diversified evolutionary stages. The progress on the basic design can be considered very promising
until now, even if much work remains to be done, while the key element of the further AO system evolution is clearly an
efficient and flexible way of producing artificial reference stars.

                                                    8. REFERENCES

1 Salinari, P., Del Vecchio, C., Biliotti, V.
 “A study of an adaptive secondary mirror”
 in “Active and Adaptive Optics”, F. Merkle, Ed.,
ESO Conference Proceedings Series, Vol. 48, pp.247-253 (1993)

2 Ragazzoni, R.,
“Pupil plane wavefront sensing with an oscillating prism”
J. of Modern Optics 43, pp.289-293, 1996

3 Ragazzoni, R., Farinato, J,
“Sensitivity of a pyramidic wavefront sensor in closed loop adaptive optics”
Astronomy and Astrophysics 350, p. L23, 1999

4 Esposito, E., Feeney, O., Riccardi, A.
“Laboratory test of a Pyramid Wavefront Sensor”
In “Adaptive Optical System Technology”, Peter Wizinowich, Ed.,
Proceedings of SPIE Vol 4007, p. 416 (2000)

5 Esposito, S., Riccardi, A., and Feeney, O.,
"Closed-loop performances of pyramid wavefront sensor,"
in “Laser Weapons Technology”, Paul H. Merritt and Todd D. Steiner Eds.,
Proc. SPIE Vol. 4034, (in press)
6 Brusa, G., Riccardi, A., Ragland, S. , Esposito, S. , Del Vecchio, C. , Fini, L. , Stefanini, P. , Biliotti, V. , Ranfagni, P. ,
Salinari, P., Gallieni, D. , Biasi, R. , Mantegazza, P. , Sciocco, G., Noviello, G. and Invernizzi, S.,
"Adaptive secondary P30 prototype: laboratory results,"
In “Adaptive Optical System Technologies”, Domenico Bonaccini; Robert K.Tyson; Eds.,
Proc. SPIE Vol. 3353, p. 764-775, (1998)

7 Riccardi, A., Brusa, G., Biliotti, V., Del Vecchio, C., Salinari, P., Stefanini, P., Mantegazza, P., Biasi, R., Andrighettoni,
M., Franchini, C., Gallieni, D., Lloyd-Hart, M., McGuire, P. C., Miller, S. M., and Martin, H. M.,
"The adaptive secondary mirror for the 6.5m conversion of the Multiple Mirror Telescope: latest laboratory test results
of the P36 prototype,"
In “Adaptive Optical Systems Technology”, Peter L. Wizinowich Ed.,
Proc. SPIE Vol. 4007, p. 524 (2000)

8 P.C. McGuire, M. Lloyd-Hart, J.R.P. Angel, G.Z. Angeli, R.L. Johnson, B.C. Fitzpatrick, W.B. Davison, R.J. Sarlot, C.J.
Bresloff, J.M. Hughes, S.M. Miller, P. Schaller, F.P. Wildi, M.A. Kenworthy, R.M. Cordova, M.L.Rademacher, M.H.
Rascon, J.H. Burge, B.L. Stamper, C.Zhao, P. Salinari, C.Del Vecchio, A. Riccardi, G.Brusa, R. Biasi, M. Andrighettoni, D.
Gallieni, C. Franchini, D.G. Sandler, T.K. Barrett
“Full-System Laboratory Testing of the F/15 Deformable Secondary Mirror for the New MMT Adaptive Optics System”
In “SPIE Conference on Adaptive Optics Systems and Technology”, R.Q. Fugate and R. K. Tyson Eds.,
Proc. SPIE Vol. 3762, 28-37, (1999)

9 Brusa, G., Riccardi, A., Accardo, A., Biliotti, V., Carbillet, M., Del Vecchio, C., Esposito, S., Femenia, B., Feeney, O.,
Fini, L., Gennari, S., Miglietta, L., and Salinari, P.,
"From adaptive secondary mirrors to extra-thin extra-large adaptive primary mirrors,"
In “Workshop on Extremely Large Telescopes”, Torben Anderson; Arne Ardelberg; Roberto Gilmozzi; Eds.,
ESO Proc. No. 57, p. 181-201, (1999)

10 D.Gallieni, d., Del Vecchio, C., Anaclerio, Lazzarini,E. P.
“LBT adaptive secondary preliminary design”
In “Adaptive Optical Systems Technology”, Peter L. Wizinowich; Ed.,
Proceedings of SPIE Vol 4007-p. 508 (2000)

11 Esposito, S. and Riccardi, A.
“Pyramid Wavefront Sensor behavior in partial correction Adaptive Optics Systems”
Submitted to A&AL.

12 Ragazzoni, R., Ghedina, A., Baruffolo, A., Marchetti, E., Farinato, J., Niero, T., Crimi, G., Ghigo, M.
“Testing the pyramid wavefront sensor on the sky”
In “Adaptive Optical Systems Technology”, Peter L. Wizinowich; Ed.,
Proceedings of SPIE vol 4007 p. 423 (2000)

13 Bonaccini, D., Farinato, J., Comin, M., Silber, A., Dupuy, C., Biasi, R., Andrighettoni, M.
“ESO-S.T.R.A.P. units”
In “Adaptive Optical Systems Technology”, Peter L. Wizinowich; Ed.,
Proceedings of SPIE vol 4007 p.431 (2000)

14 Storm, J., Seifert, W,. Bauer, S.M., Dionies, F., Hanschur, U., Hill, J.M., Möstl, G., Salinari, P., Varava, W.,
Zinneker, H.
“Wavefront sensing and guiding units for the Large Binocular Telescope”
In “Adaptive Optical Systems Technology”, Peter L. Wizinowich; Ed.,
Proceedings of SPIE vol 4007 p.461 (2000)

To top