Fig. 2 (10 K)
P. Salinari
Osservatorio Astrofisico di Arcetri
Largo E. Fermi 5, 50125 Firenze, Italy
Key words: Large Binocular Telescope, astronomical interferometry, adaptive optics
3. PROVISIONS FOR ADAPTIVE OPTICS
Due to the small separation (6 m) between its two 8.4 m mirrors and to the fact that they are on a common mount, the LBT interferometer operates in a way that is very different from other large telescope interferometers and more similar to that of a telescope with an elongated pupil. Important features of this optical configuration are the simple resulting PSF, the wide coherent field, the high optical efficiency and the low emissivity of the combined focus. In order to exploit fully these characteristics the telescope was designed so that it can use an advanced Adaptive Optics system, based on adaptive secondary mirrors and artificial reference stars. This AO system is presently being developed and the first results obtained with a laboratory prototype of the adaptive secondary mirror are very encouraging. Optics for beam combination that provides a coherent field larger than the isoplanatic field has been designed for various wavelengths between visible and mid IR. The entire system was evaluated more in detail in the near IR range, considered to be of the highest priority because in this range LBT should be capable of achieving at the same time an extremely high sensitivity and an angular resolution of about 0.01 arcsec.
When, more than ten years ago, the LBT Project was conceived as a single-mount two-mirror telescope, the potential advantages for interferometry of this configuration were well known and confirmed in practice by the experience of the Multiple Mirror Telescope. Although limited in angular resolution by the maximum available baseline of 22.8m (see Fig.1 for the pupil geometry), this configuration is optically very efficient and can provide a relatively large combined field of view.
Only in recent years the adaptive correction of the effects of atmospheric turbulence in astronomical observations was tested in practice and made possible, at least in principle, a full exploitation of the potentials of interferometers using very large pupils. As this happened while LBT was still in the design phase, it was possible to incorporate in the telescope design a number of features that take into account the developments of adaptive optics. In particular we 1) totally redesigned the beam combining optics, 2) adopted a Gregorian configuration, 3) provided the means for projecting artificial reference stars and 4) started the development of adaptive secondary mirrors, as better discussed in the following.
Although any type of large telescope interferometer will greatly profit from the addition of adaptive wave front correction, the binocular approach not only maintains its efficiency advantage, but, due to its wide coherent field, allows one to reach the ultimate sensitivity by integrating on the source while a field star is used for cophasing the two phased beams. In most cases this star can simply be the same used to phase the individual pupils.
The main purpose of this paper is to summarise the solutions adopted in the telescope design phase and the general approach of the current work on the aspects that refer to the interferometric use of LBT. The system optical layout is briefly discussed in section 2 and a brief description of the work in progress on adaptive secondary mirrors is reported in section 3. Section 4 reports a qualitative evaluation of the performance attainable by the telescope in its interferometric mode.
Fig. 1: top view of LBT showing the geometry of the pupils and of the bent Gregorian focal stations that can be used for beam combination.
Many different schemes have been considered for the beam combination [2] and for a long time one of them, using an "ad hoc" F/33 secondary mirror, a flat tertiary and a flat beam combiner mirror on each side of the telescope was considered to be the baseline configuration. Neglecting the problems connected with the cost and the structural impact of this configuration, when we considered ways to implement adaptive correction of the two telescope beams it became clear that it would have been complex to reimage the pupils and then again the field at the combined focal plane.
A new "reimaged beam combiner" (RBC) was then designed. On each side the RBC uses the F/15 beam of the Gregorian secondary, bent by the tertiary and forming a focal plane in the central section of the telescope close to the edge of the primary mirror. In fact, as shown in Fig. 1, there are in total six bent Gregorian focal stations in the central part of the telescope, and therefore three different options for beam combination. The central pair of bent Gregorian foci is on the line joining the centers of the pupils, while the others are offset by about 25 degrees. The bent Gregorian foci are equipped with derotators and guiding probes. The beam combining optics, schematically shown in Fig. 2, consists essentially of a collimator and a flat mirror (BCM) on each side that illuminate a camera forming the final combined focal plane. Each collimator forms a pupil image where a wave front corrector could be located, in alternative or in addition to the secondary. Other optical components can be placed in the parallel beam before, at and after the pupil image as required. The one shown in Fig. 2 is a configuration where the adaptive corrector element is the secondary mirror, therefore the beamsplitters used for the artificial and for the natural reference star can be located anywhere and are shown before the pupil image with the Atmospheric Dispersion Compensator (ADC). Other options can be considered including multiconjugate adaptive systems with a second corrector and wave front sensor located after the pupil image.
Fig. 2 Scheme of the beam combining optics. See text for discussion.
In order to exploit the great flexibility of this configuration, the requirements for the beam combining optics are very severe: excellent image quality has to be provided not only over a fairly large field but also over a very wide wavelength range. For example for near IR interferometry the optics must cover the range from about 0.5 to 2.5 micron to be used in the configuration of Fig. 2, as it is used by the Sodium artificial star at 589 nm, by the natural reference star used for adaptive correction of the low orders (say in R and/or I but, depending on star spectral type, also in the Near IR wavelengths) by the source beam (in J,H,or K) and by that part of the light of the natural reference star that is used to co-phase the two beams (say, again, R or I, or even one of J,H,K). For this reason reflecting optics was studied first and shown to be possible. The best results were however obtained by relatively simple transmission optics, provided refocusing is allowed for the different bands [3].
Fig. 3 The broad-band Point Spread Function in the J band at the centre of the interferometric field.
Fig. 3 shows the interferometer Point Spread Function in the J band (full band) in the field center that is obtained by using two dioptric collimators (achromatic doublet) and a dioptric camera (achromatic doublet). Fig, 4 shows a map of the Strehl Ratio obtained in the same conditions over a field of 1x1 arcmin. One can see that the Strehl Ratio remains larger than about 80% over a field of about 2/3 arcmin diameter, wider than the isoplanatic field expected in J even in excellent seeing conditions. In the H and K bands the performance is even better at the field centre and the high Strehl Ratio field increases at about the same rate as the isoplanatic field. Similar dioptric systems forming excellent images in visible, near IR and also at longer wavelengths, including the entire N band, have been investigated and are feasible, although it becomes difficult to obtain fields of many arcmin diameter corresponding to the much larger isoplanatic angle at the longest wavelengths. High Strehl ratio over a field of up to several arcmin can be obtained at L,M and N.
Fig. 4 Values of the Strehl Ratio in the J band (broad-band) as a function of position in the interferometric field.
The PSF of Fig.3 is very close to the theoretical one obtained with the pupil geometry of LBT, except that some energy is moved from the central peak to the side lobes, that in theory should be 37% of the peak. The data of Fig. 3 and Fig .4 are directly applicable to the central beam combiners, because the retardation due to reflection on the flat mirrors is completely compensated in that case. When the two off-set combined foci are used, retardation is only partly compensated, and, for Aluminium mirrors without any special coating, in the J band about 10% of the energy is lost from the final interferometric image. The loss is smaller at longer wavelengths and larger at shorter ones, although it can be reduced by a proper selection of mirror coating and introducing suitable retarders. In any case the central beam combiner makes life much easier when working at short wavelength, while the off-set combiners can be used very effectively at near IR and longer wavelengths.
It is worth mentioning that the telescope design allows for massive instruments with 'T' shape whose horizontal arms can be up to about 2 m diameter and 5 m long and whose vertical arms can be several meters high. The large available volume can be used for installing in each beam combiner/instrument interchangeable optical components optimized for different wavelength bands. Up to three of these "instruments" can be mounted on the telescope at the same time, and they can be installed and removed in a single piece for maintenance using the telescope service crane. It is therefore possible to enclose the entire "instrument", i.e. beam combinig optics, wave front sensors, detectors etc. in a single vacuum tank where the temperature is controlled at a low enough value to reduce the thermal emission from the beam combiner optics well below the sky+telescope value. In this way it is possible to reduce the thermal background at wavelengths longer than 2 m to that of a very "clean" three mirror telescope.
As briefly discussed in the previous section, the present final beam combination scheme was adopted mainly because it provides a large flexibility and a high efficiency for adaptive correction of the two interfering wave fronts. The wave front sensors and correctors can easily be located close to the two intermediate F/15 foci and become part of the interferomertic "instrument". Other parts of the telescope were also modified at the same time in view of expected developments of the adaptive techniques.
The adoption of a Gregorian configuration for the F/15, also useful for other reasons, was considered an advantage in case the secondary mirror became the correcting element. The conjugate of the Gregorian secondary is real and in the atmosphere, but still very close to the pupil (~100m), therefore there is not a very significant advantage compared to a Cassegrain secondary from this point of view. More important is that the ellipsoidal secondary can be tested on the telescope independently from the primary. If a suitable light source (pin-hole, fibre) is placed near to the primary focus, measurements and calibrations can be done with the wave front sensors available at the focal plane. The fact that a Gregorian secondary is larger than a Cassegrain one providing the same final F/N is normally a disadvantage, but for an adaptive secondary where the physical size of the actuators represents the limit to the maximum number of correcting elements, a larger secondary implies a finer wave front correction.
Thanks to the work of many people in various groups ( [4], [5], [6], [7]), the development of adaptive secondary mirrors along the lines proposed in [1], has now progressed to the point that its technical feasibility seems to be established. Fig. 5 and Fig. 6 show respectively a section and a 3D view of the preliminary design for the Cassegrain adaptive secondary presently studied for the MMT Conversion telescope. Table 1 reports preliminary design parameters for the MMT Conversion and the corresponding (tentative) values for the LBT adaptive secondary mirrors.
Fig. 5 Scheme of the Adaptive Secondary Unit of the MMT Conversion Telescope. See Table 1 for further information.
Fig. 6 3D view of the Adaptive Secondary Unit shown in Fig. 5
It is worth noting that the adaptive secondary mirrors of the type under development can correct at the same time tip-tilt and higher orders. The actuator spacing corresponds approximately to the value assumed in very good seeing conditions by the Fried coherence length in the R band (about 30 cm.). At near IR wavelengths the adaptive secondary is therefore able to provide an excellent level of correction, only limited by wave front sensing accuracy in all practical circumstances. As mentioned above, the adaptive secondary can also be used in conjunction with a second stage of correction that could be placed at the conjugate of high atmospheric turbulence layers.
Table 1 : basic adaptive secondary parameters
Fig 7 shows the provisions adopted on the telescope that allow the projection of an artificial reference stars from behind the secondary mirrors. The laser units, that use a large amount of power and require a stable environment, are located in the telescope pier. The laser beams go up through the central hole in the azimuth platform and are deviated by a mirror (A) placed at the intersection of the azimuth and elevation axis. The mirror is actively rotated about the elevation axis to send the laser beams to the fixed optics that is on the elevation platform. Here we have kept sufficient free space to place collimating optics (B) and diverting flat mirrors in order to reach the top of the secondary support, where the collimated laser beam is projected to the sky on the optical axis. The laser system and its collimating optics are not yet defined in detail.
Fig. 7 Path on the telescope of the laser beams for the Laser Artificial Stars.
Although the maximum angular resolution can be achieved on bright sources with a variety of techniques even without adaptive correction of the wave fronts, this is necessary to obtain the full resolution together with high sensitivity. The performances of the LBT interferometric mode then depend on a large number of parameters affecting the quality of the wave front correction and of the cophasing of the beams. A detailed discussion of the influence of the various parameters is beyond the scope of this section. I will simply illustrate the logic and identify the orders of magnitudes.
Let us assume that a good wave front correction can be made on a particular field, i.e., that at least a sufficiently bright natural star is available for tip-tilt correction and that the laser reference star allows for high order correction. It is then very likely that the LBT interferometer can be cophased using the same natural star in a different band, because the coherent field of LBT is larger than the isoplanatic field and the requirement on the star magnitude for cophasing is not more severe than for tip-tilt correction. It is worth noting
that the natural star itself is largely corrected (depending on the band used), as high order correction is provided by the artificial star. This increases the accuracy of both tip-tilt sensing and cophasing. Therefore, in most cases, if a good adaptive correction is possible, also cophasing becomes possible and one is able to integrate on that field as with a normal telescope.
How long can the integration be? The interferomertric PSF rotates with respect to the sky due to the Alt-Az mount, but a look at Fig. 3 shows that a rotation of several degrees hardly affects the resolution in the direction where it is higher. The PSF rotation rate is a function of the position in the sky, but for any field in the sky it is possible to plan observations in order to obtain a time of 1000 s or more before the rotation affects significantly the final angular resolution.
In the J band (with angular pixel size around 5 mas, a natural sky background of about 104 ph s-1 m-2 and a system transmission of about 20%) in an integration of ~103 s several thousand photo-electrons per pixel could be collected (and more in the H and K bands, due to higher background and, possibly, larger pixel). Therefore in J,H,K an integration time of typically a few minutes could be necessary to work in background limited conditions with current detectors. Once cophased the interferometer can therefore reach in the near IR the sensitivity limit set by the natural background. Many such integrations at different position angles can then be processed to obtain at the same time the maximum angular resolution in two dimensions and a better signal to noise ratio. A minimum of three exposures with a total baseline rotation of about 90 degrees are necessary to reconstruct an image with full resolution in both axes.
In the above conditions and with a total exposure time of 1000 seconds, S/N ~1 (per pixel) would be obtained on a signal of about J mag 30 (a total efficiency of about 20% was assumed for both, signal and background). An unresolved source around J mag 25 could therefore be easily measured, as most of the energy falls on a dozen of pixels.
Even if the adaptive correction is partial and only a fraction of the energy ends up in the final interference pattern, while the rest is dispersed over a fairly large area, if the telescope is kept well cophased and the final image is well stabilised, the sensitivity remains high, due to the possibility of integrating for a long time, although of course it is reduced by the light loss in the interference pattern. A sensitivity loss of only one or two magnitudes with respect to the above ideal case can be expected in good observing conditions with existing adaptive and laser star technology. Introducing a further correction stage in a multiconjugate scheme would not only increase somewhat the sensitivity, but would be particularly important for expanding the area of sky over which high sensitivity observation can be performed. A wider isoplanatic angle would in fact allow a larger angular separation between the observed source and the natural reference star.
What I reported here is of course the result of the collective work of many members of the LBT Project and of large number of discussions with others who contributed ideas and stimulation. I thank Dr Alberto Caruso for the figures and the data from his thesis.
[1] P. Salinari, C. Del Vecchio, V. Biliotti
"A Study of an Adaptive Secondary Mirror"
Proc. ICO-16 Conference on Active and Adaptive Optics, Garching August 2-5,1993.
[2] J. M. Hill
"Strategy for interferometry with the Large Binocular Telescope"
Proc. SPIE Conference on Amplitude and Intensity Spatial Interferometry II, Kona, Hawaii, 1994
[3] P. Byard and D. Bonaccini
"Optical design for interferometry with the Large Binocular Telescope"
Proc. SPIE Conference on Amplitude and Intensity Spatial Interferometry II, Kona, Hawaii, 1994
[4] R. Biasi, D. Gallieni, P. Mamtegazza
"Simulation of Adaptive Secondary Mirror Dynamic Response"
Proc. OSA Adaptive Optics Conference, Munich, October 2-6, 1995
[5] V Biliotti, R. Biasi, G. Brusa. D. Gallieni, R. Spairani, R. Aiello
"High Accuracy Capacitive Displacement Transducer for the Position Local Control Loops at the Adaptive Secondary"
Proc. OSA Adaptive Optics Conference, Munich, October 2-6, 1995
[6] D.G. Bruns, T.K. Barret, D.G. Sandler, H.M. Martin, G. Brusa, D. Modisett, J.R.P. Angel, R. Biasi, D. Gallieni, P. Salinari
"Force-Actuated Adaptive Secondary Mirror Prototype"
Proc. OSA Adaptive Optics Conference, Munich, October 2-6, 1995
[7] C. Del Vecchio, W. Gallieni, P. Salinari, P.M. Gray
"Preliminary mechanical design of an adaptive secondary unit for the MMT Conversion telescope"
Proc. OSA Adaptive Optics Conference, Munich, October 2-6, 1995
Fig.3 (5 K)
Fig.4 (10 K)
Fig.5 (12 K)
Fig.6 (11K)
Fig.7 (11 K)
