To appear in High Resolution Solar Physics: Theory, Observations and Interpretation, Proc. 19th NSO/SP Summer Workshop, (eds. T. Rimmele, R. Radick & K. S. Balasubramaniam), 1999.
The New Swedish Solar Telescope
G.B. Scharmer
Royal Swedish Academy of Sciences
Stockholm Observatory,
SE-13336 Saltsjöbaden, Sweden
M. Owner-Petersen
Lund Observatory, Box 43, SE-22100 Lund, Sweden
T. Korhonen
Tuorla Observatory, FIN-21500 Piikkiö, Finland
A. Title
Lockheed Martin Solar and Astrophysics Laboratory, B/252, O/L9-41
3251 Hanover Street, Palo Alto Ca. 94304, U.S.A.
The design of the Swedish Vacuum Solar Telescope, SVST, was based on the firm conviction, that by using a small number of optical components of excellent optical quality, it would be possible to construct a telescope which performed better than existing even larger solar telescopes (Scharmer et al. 1985). This concept was validated in December 1985, when the SVST saw first light. The SVST has produced diffraction-limited images since the first day of its operation and other solar telescopes world-wide have not been able to match that image quality consistently, in spite of significant efforts. By using real-time frame selection in combination with wavefront sensing and post-processing techniques, this telescope can produce long uninterrupted time sequences of dynamic phenomena in the solar atmosphere at a resolution which is strictly limited by the diameter of the telescope, which is 47.5 cm.
In this paper, we outline plans for a new solar telescope, to replace the existing SVST. The goal of this telescope is imaging at a resolution close to its diffraction limit of 0.1 arcsec in the blue and polarimetry and spectroscopy at approximately 0.2 arcsec resolution. To achieve this goal, a simple but excellent optical system including adaptive optics is proposed.
The site of the SVST on La Palma is the best known site for solar telescopes in the world. By installing the new telescope on the existing tower, by keeping development and design costs down and by re-using some of the instrumentation developed for the SVST, it is possible to build an excellent meter-class telescope within a budget of 2 M$.
The SVST uses an evacuated turret, similar to that of the R.B. Dunn Telescope at the Sacramento Peak Observatory (Dunn 1964). The main advantage of this pioneering design is that the telescope needs no dome which could seriously degrade image quality by heating or cooling air in the vicinity of the telescope. Also, all mirrors are within vacuum, excluding the possibility of mirror seeing. The turret of the SVST has been on the Swedish tower for 12 years and has been exposed to the most severe weather conditions, including more than 50 m/s winds, heavy rains, complete ice coverage and dust storms. Yet it shows no signs of damage, corrosion or wear. The turret mirrors which are in the vacuum system are cleaner after ten years than a mirror exposed even to the favourable conditions in our optics labs during a few weeks. We have seen no evidence of image movement due to vibrations from either the turret or the tower in reasonable wind speeds. The mirrors in the turret are always used at fixed angles of incidence, which makes calibration and modeling of telescope polarization feasible. Significant progress at the Instituto de Astrofisica de Canarias has been made in such modeling of the SVST telescope polarization during the last two years (Pillet et al. 1999, these proceedings). The turret concept has proven itself at the SVST and this concept will be used also for the NSST.
We have considered modifying this turret design, such that the mirrors are used to deflect the beam by 60-70 degrees instead of 90 degrees as for the SVST. The advantages would be lower telescope polarization and reduced cost and weight for the mirrors. The disadvantage is that the turret becomes significantly less compact, leading to greater area exposed to sunlight and wind, which could degrade performance. It was finally agreed, that the advantages of modifying the turret design were less important than the possibility that e.g the telescope loses its pointing stability in moderate wind speeds. We have therefore adopted a design for the NSST which is similar to that of the SVST.
The SVST uses an achromatic doublet in the turret to form an image in the optics lab. The advantage of a doublet objective is that it gives a shallow focus curve over some reasonable wavelength range, which allows broad-band observations. However for several important applications, such as narrow-band imaging in H-alpha and Ca K, or polarimetry using narrowband filters or the spectrograph, a shallow focus curve is not needed. A doublet of 1 meter diameter is difficult to manufacture with high optical quality and would be diffraction limited with the NSST over only 1/4 of the wavelength range compared to the doublet of the SVST. For example, a doublet of 1 meter aperture would prohibit spectroscopy involving the broad wings of the Ca H or K lines with a single focus position. We therefore decided to use a singlet lens as primary image forming element and a corrector, described below, for achromatic imaging or spectroscopy. The primary optical system of the NSST consists of only one singlet objective, acting also as vacuum window, and two flat folding mirrors.
The choice of glass for the singlet objecive and other transmitting lenses in the optics was finally fused silica. The main argument for this is its low coefficient of thermal expansion, which gives small stresses from temperature gradients. Stresses in the glass lead to birefringence which e.g. introduces spatial cross-talk from Stokes I to linear polarization. Compensating for such effects will need deconvolution. Polarimetry of small-scale structures will require accurate determination of both the ordinary wavefront errors and the spatial variation of the orientation and magnitude of the retardation across the pupil. We believe that this will be very difficult if the retardation varies with the temperature gradient in the singlet objective. For example, inserting large polarizers in front of the singlet objective will change the illumination and thus temperature (gradient) of the objective, making such calibrations of telescope polarization inaccurate. By using fused silica instead of BK-7, we reduce this temperature sensitivity by approximately a factor 15. It should also be pointed out that birefringence leads to a reduction in the Strehl ratio which cannot be compensated by adaptive optics. We have estimated that e.g. a 2-degree temperature difference between the edge and the center of of a lens made of BK-7 would reduce the Strehl ratio to 0.8, thus imposing a fundamental limitation on image quality.
The singlet objective has a diameter of 1070 mm, a clear aperture of 970 mm, a center thickness of 80 mm and an edge thickness of 65 mm. It has a focal length of 20.8 m at a wavelength of 460 nm, is corrected for coma and has a small aspherical correction applied to its first surface. Finite element analysis has been used to calculate the stresses from pressure, gravity and temperature gradients and also to calculate the distortions of the surfaces. To evaluate the effects of the deformations of the surfaces of the lens, the co-ordinates of the node points are fitted to sixth-order polynomials. The corresponding coefficients are inserted in a ZEMAX ray-trace program to evaluate optical performance. The analysis shows that the pressure adds fourth- and sixth-order terms to both surfaces which are large but that the resulting spherical aberration from the vacuum load is negligible when combining the effects of the two surfaces.
In order to minimize temperature gradients in the singlet objective, it will be mounted such that its edges are exposed to air and with a shield to prevent the cell from being heated by direct sunlight, in a way similar to SVST. We believe that the spherical aberration from temperature gradients will be small enough to be corrected by the adaptive mirror. Should this not be the case, we will install a system to control the temperature of the holder of the singlet objective.
We have verified that the birefringence from stresses lead to a negligible (3%) loss of Strehl ratio. The maximum tensile stress is 4.0 MPa which is below the recommended design stress of 6.8 MPa for fused silica.
The diffraction limited field of the singlet objective is nearly 1/2 degree or 18 cm at the primary focal plane. At this field radius, the only significant aberration is a small amount of astigmatism which is easily corrected by an adaptive optics system. It is unlikely that adaptive optics will correct atmospheric aberrations over much more than 1 arcmin field. The NSST therefore uses four exit ports close to the optical axis of the singlet, each with its own secondary optical system. This is possible by restricting the field-of view for three of these secondary optical systems to approximately 2 arcmin. This allows the secondary optical systems to be used in parallell and even simultaneously depending on the choosen target on the Sun. Switching between the different secondary optical systems is done simply by changing the pointing of the telescope slightly.
Narrow-band imaging in e.g H-alpha, Ca K or other strong lines as
well as polarimetry using either filters or a spectrograph involves
pass-bands of less than 0.1 nm and is possible without any chromatic
correction of the singlet objective. This allows a very simple
secondary optical system which is shown schematically in
Fig. ![[*]](cross_ref_motif.gif)
. The singlet lens and the two turret
mirrors form the primary focal plane which at a wavelength of 460 nm
is located at the bottom of the vacuum system. A singlet lens, acting
as vacuum window re-images the 97 cm singlet objective on the adaptive
mirror which also folds the beam horizonthally. A re-imaging lens
produces the final image at the desired image scale.
The adaptability of the optics of the NSST is high-lighted by the
secondary optical system shown in
Fig. . This system uses a field
mirror, also located close to the focal plane, to deflect the beam
upwards and away from the optical axis of the telescope. This beam is
maintained within the vacuum system because it is comparatively long,
5.25 meters, and because it re-images the one-meter singlet objective
on a nearly 24 cm large corrector. This corrector consists of a lens
and a mirror. The effect of the lens is to cancel out the chromatic
aberration (variation of focus with wavelength) of the singlet
objective, the effect of the mirror is to create a perfectly
achromatic image at the secondary focus. This concept was proposed by
Ludwig Schupmann 100 years ago.
The idea of using a singlet lens in combination with a Schupmann corrector for solar telescopes has been proposed earlier by Baker (1954), Rush and Schnable (1964) and also by Dunn as an early concept for LEST. It was later considered by Roesch (1983, private communication) for the Pic du Midi refractor, but was rejected because of strong off-axis aberrations. The Schupmann design has been implemented on several telescopes during the last 40 years with mixed results. Of these can be mentioned the two coronagraphs at Climax and at the Sacramento Peak observatories, both of which had serious optical problems. These designs used correctors with a single component, consisting of a meniscus lens silvered on its backside. For solar telescopes, a corrector consisting of a meniscus lens followed by a Zerodur mirror appears a better choice which also adds one more degree of freedom in the optimization. Ray-tracing calculations do not indicate any particular problems in aligning the corrector lens with respect to the mirror, however, it is clear that the image of the singlet needs to be accurately centered on the corrector in order to not give rise to dispersion. We believe that an arrangement as proposed for the NSST, with the Schupmann system at fixed orientation in the stable environment of the tower, with a corrector consisting of a separate lens and Zerodur mirror and combined with a wavefront sensor is a more promising concept than those of earlier designs.
Initially, we planned to replace the curved mirror with a curved adaptive mirror, but finally it was decided that using a flat adaptive mirror acting also as folding mirror, below the vacuum system was a less expensive and simpler solution. Because a folding mirror was needed anyhow, using this mirror as an adaptive mirror does not increase the number of optical surfaces. The cell of the adaptive mirror will be mounted with two piezos to allow tip-tilt correction with large stroke.
In order to give diffraction-limited image quality, either the
corrector lens or the mirror needs approximately six waves PV
correction for spherical aberration. The excellent performance of the
telescope after correction by the Schupmann system is shown in
Table . The design was optimized using
Zemax with tilt-angles for the field mirror and the Schupmann mirror
to allow straightforward implementation of baffling to reduce
straylight. Early designs showed excellent NIR performance but
somewhat poor performance at 400 nm. By adjusting the curvatures of
the singlet objective such that its focal plane agrees with the
location of the field mirror at 460 nm, a better balance between NIR
and blue performance is obtained. The present design allows
diffraction limited performance (Strehl ratio 0.96 or higher) over the
whole 400-900 nm wavelength range at a single focus position.
An important feature of the Schupmann system is its ability to correct
also for atmospheric dispersion. The Schupmann lens cancels the
chromatic variations of the singlet objective only when the image of
the singlet is centered on the corrector lens. Displacing the the
image of the singlet either by tilting the field mirror or by
translating the Schupmann lens in the plane perpendicular to the
optical axis, causes dispersion in the image plane. The amount of
dispersion corresponds to 180 micron, or 1.8 arcsec, separation
between the 400 and 900 nm wavelengths per mm decenter.
Table shows the atmospheric
dispersion at 14.9 deg elevation for an altitude of 2000 m as
calculated by Martin (1997) for wavelengths in the range 400-900 nm
and the residual dispersion after optimum compensation with the
Schupmann system. The Schupmann corrector allows the effects of
atmosperic dispersion to be reduced by approximately a factor 25.
|
The simplest way of correcting the atmospheric dispersion is by tilting the field mirror, but this moves the pupil image which is undesirable for the following adaptive mirror and wavefront sensor. We have therefore choosen to install a fixed pupil stop corresponding to a diameter of the singlet of 950 mm. When compensating atmospheric dispersion at elevations below approximately 15 degrees, this causes the clear aperture of the singlet objective to be displaced slightly outside the pupil stop. The NSST will have a broadband filter which can be switched electrically between a blue and red wavelength without moving optical components. This allows the dispersion to be measured accurately, using either the correlation tracker, the adaptive optics wavefront sensor or any other CCD, and tilting the Schupmann field mirror to compensate the dispersion.
The Schupmann design allows an unobscured pupil and perfect chromatic correction by using an off-axis system and is as a result of this limited to good correction over a fairly limited field. Ray-trace calculations indicate diffraction limited performance over an approximately 3 arcmin field. In recognition of the constraints in the useful field imposed by adaptive wavefront correction, we have restricted the field further to 2 arcmin by reducing the diameter of the field mirror to 36 mm. This reduces the heat load on the Schupmann corrector and should also lead to reduction of stray-light. By locating the field mirror at the focal plane of the singlet at 460 nm, there will be vignetting of part of this field at the 900 nm wavelengths, but this has been considered an acceptable trade-off in order to enhance the performance at blue wavelengths. The unvignetted field at the 900 nm wavelength is 70 arcsec.
The Schupmann design, by requiring a large singlet lens, is of no interest to future larger solar telescopes but appears attractive for an evacuated meter-class solar telescope with adaptive optics.
It is sometimes argued that telescopes equipped with adaptive optics require only low optical quality. We believe that this is not correct. Adaptive optics systems use wavefront sensors which have a limited spatial resolution at the pupil. Telescope aberrations at spatial frequencies higher than resolved by the wavefront sensor will, because of aliasing, be interpreted as low-order aberrations by the wavefront sensor. An adaptive optics system operating in closed loop will thus introduce fixed low-order aberrations in addition to the high-order aberrations which cannot be corrected by the adaptive mirror.
Whereas we feel confident that a low-order adaptive optics system (15-36 Zernike correction) can efficiently compensate seeing in good-excellent seeing conditions (r0 > 12 cm), we feel that the efficient operation of high-order AO systems to correct poorer seeing is presently not realistic.
In order not to be forced to use a high-order adaptive optics system for the NSST and in order that the accumulated telescope aberrations, including those also of the secondary optical systems and adaptive mirror, shall be negligibly small compared to those of the atmosphere in very good seeing after correction by a low-order adaptive optics system, high optical quality is needed. Accordingly, the specifications for the two turret mirrors are 20 nm surface quality PV over any subaperture with 300 mm diameter and 50 nm PV over the entire surface. The wavefront quality for the singlet objective is similar. In order to achieve such high quality for the singlet lens, high homogeneity in the refractive index is required. The fused silica blank manufactured by Corning has refractive index variations within +/- 1.5e-6. These variations are not small enough to ensure required wavefront quality and local polishing will be required.
Optimization of the Schupmann system shows that the second surface of the corrector lens can be flat without degrading performance. This makes testing during polishing easier.
The optics will be polished by Opteon Oy in Finland. During polishing, low frequency deviations from flatness for the flat turret mirrors will be tested with the pentaprism method (Korhonen 1994). High-frequency errors will be tested using interference fringes against a reference flat. The lens will be tested both in autocollimation using the flat mirrors and by using the lens surfaces as a concave mirror. A single elements null lens is used in these tests.
The absence of plane-parallell surfaces in the optical system should eliminate the possibility of fringes.
The goal of the NSST is diffraction limited imaging and near-diffraction limited spectroscopy and polarimetry in excellent seeing, corresponding to r0 > 20 cm. Measurements made during the LEST site testing campaign (Brandt et al 1989), as well as experience with the SVST, indicate that such seeing conditions should occur at least 5% of the time. Simulations indicate that near diffraction-limited imaging, corresponding a wavefront rms of less than lambda/10, should be achievable under such conditions with an AO system correcting the first 10 Zernike modes. In order to allow for a 50% efficiency of the AO system, at least 20 Zernike modes need to be corrected.
We are presently developing an adaptive optics system for the SVST based on a 19 electrode bimorph mirror from Laplacian Optics, capable of perfect correction of the first 15 Zernike coefficients. Recent improvements in the manufacturing process should allow this mirror to be flattened to within 1/10 wave PV, therefore allowing accurate wavefront sensing with little aliasing from high-order aberrations in the adaptive mirror. For the NSST, we expect to use an AO mirror correcting up to 36 Zernikes.
The optical arrangement is shown in Fig. .
In order to minimize the number of optical surfaces, a lens, acting as
exit vacuum window, is used as field lens. The AO mirror is located
approximately 700 mm below the field lens and is used at 45 degree
angle of incidence, thus folding the beam horizonthally above the
optical tables. The bimorph mirror has a stretched geometry to match
the elliptic pupil image. Simulations made for the CFHT by Graves
(1998, private communication) show that this type of arrangement
allows excellent performance of the AO system. The diameter of the
adaptive mirror will be as small as possible to allow adequate stroke
with a 36 mode correction on a one-meter telescope in order to
minimize the risk of telescope seeing. The distance between the field
lens and AO mirror will be matched to that pupil diameter.
The wavefront sensor for these adaptive optics systems is likely to be a Shack-Hartmann wavefront sensor, matched to the geometry of the adaptive mirror. We are also investigating the possibility of using phase-diversity as a real-time wavefront sensor (Scharmer 1999, these proceedings). The computer system which will be used for the wavefront sensor and adaptive mirror is described by Shand et al. (1999, these proceedings).
For future developments, it seems more important to increase the angle over which the wavefront can be corrected rather than extending the number of corrected Zernike modes. Experience with the AO system developed for the SVST and the first year of operation of the NSST will be crucial in making priorities for further AO developments.
The main instruments for the NSST will be the following: imaging CCD:s, including three 2000 x 2000 Kodak CCD:s, an H-alpha filter, a new Michelson Solar Polarimeter (MSP, designed by Lockheed-Martin and funded by NASA) and a short Littrow spectrograph. In addition, we will use broad-band G-band and Ca K filters.
The MSP system uses a Michelson-Lyot filter, will allow 0.14 arcsecond resolution and obtain high signal-to-noise, vector magnetograms of the solar photosphere over a 85 arcsecond field-of-view on a 10 second cadence.
For H-alpha, plans are to re-build the engineering model of the SOUP filter and to use this filter exclusively for H-alpha. This is done in order to avoid using a single tunable filter (and CCD) for too many wavelengths, which would give unacceptably long cycling times.
The Littrow spectrograph will be optimized for modest dispersion and high spatial resolution at a single wavelength. Plans are also for a spectrograph which will allow spectra to be recorded at more than one wavelength simultaneously.
For financial reasons, the instrumentation of the NSST will be restricted but significantly more powerful than that of the SVST.
Acknowledgements
Construction of the NSST is funded by the following private foundations in Sweden: Knut och Alice Wallenbergs Stiftelse, Marianne och Marcus Wallenbergs Stiftelse, Stiftelsen Marcus och Amalia Wallenbergs Minnesfond and the Royal Swedish Academy of Sciences.
The Swedish Vacuum Solar Telescope is operated by the Royal Swedish Academy of Sciences within the Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias on the island of La Palma, Spain.
Darrel Torgerson is thanked for ray-tracing the Schupmann system and providing valuable information on alignment tolerances.
Mats Löfdahl is thanked for valuable help in preparing this manuscript.
We finally want to thank all the observers from the international solar community which made the SVST a successful scientific instrument and thereby contributed to making the NSST possible.
References
Baker, J. G., 1954, A.J. Vol. 59, p.74
Brandt, P., Erasmus, D. A., Kusoffsky, U., Righini, S., Rodriguez, A., & Engvold, O., 1989, LEST Foundation Techn. Rpt 38
Dunn, R. B., 1964, Appl. Opt. Vol. 3, No. 12, p. 1353
Korhonen T, K., Lappalainen, T., Sillanpaa, A. K., 1994, in Fabrication and Testing of Optics and Large Optics, Proc. SPIE 1994, (ed. V. J. Doherty), p. 225
Martin, G., 1997,
http://www.eso.org/lasilla/Telescopes/NEWNTT/emmi/refr.html
Pillet, V. M., Collados, M., Sanchez Almeida, J., Gonzalez, V., Cruz-Lopez, A., Manescau, A., Joven, E., Paez, E., Diaz, J. J., Feeney, O., Sanchez, V., Scharmer, G. B., & Soltau, D. 1999, in High Resolution Solar Physics: Theory, Observations and Interpretation, Proc. 19th NSO/SP Summer Workshop, (eds. T. Rimmele, R. Radick & K. S. Balasubramaniam), pp. ?? Rush, J. H., & Schnable, G. H., 1964, Appl. Opt. Vol. 3, No. 12, p. 1347
Scharmer, G. B., Brown, D. S., Pettersson, L., & Rehn, J., 1985, Appl. Opt. Vol. 24, No. 16, p. 2558
Scharmer G. B. 1999, in High Resolution Solar Physics: Theory, Observations and Interpretation, Proc. 19th NSO/SP Summer Workshop, (eds. T. Rimmele, R. Radick & K. S. Balasubramaniam), pp. ?? Shand, M., Scharmer, G. B., & Wei, W. 1999, in High Resolution Solar Physics: Theory, Observations and Interpretation, Proc. 19th NSO/SP Summer Workshop, (eds. T. Rimmele, R. Radick & K. S. Balasubramaniam), pp. ??
 |
 |
This document was generated using the LaTeX2HTML translator Version 98.1 release (February 19th, 1998)
Copyright © 1993, 1994, 1995, 1996, 1997, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html -split 0 -font_size 12pt -local_icons scharmer_NSST_l2h.
The translation was initiated by Dan Kiselman on 1999-06-10