The SVST AO System: First Results

Göran Scharmer, Wang Wei, Mats Löfdahl
Royal Swedish Academy of Sciences
Stockholm Observatory
SE-133 36 Saltsjöbaden, Sweden

Mark Shand
Compaq Computer Corporation
130 Lytton Av.
Palo Alto, CA 94301, USA

Abstract We give an overview of the adaptive optics system installed at the Swedish Vacuum Solar Telescope in May 1999 and we show the first images recorded with the system.

Introduction

We describe here shortly the main features of the adaptive optics (AO) system installed at the Swedish Vacuum Solar Telescope (SVST) in May 1999 as well as some data that demonstrate that it works. The installation took less than a week, and then another week was used for testing and modifications. The system is currently in use by visiting observers, the first of which arrived less than two weeks after the installation was started.

The main parts of the AO system are a Shack-Hartmann (SH) wavefront sensor, a computer and a deformable mirror. The SH sensor consists of a microlens-array with 19 hexagonal microlenses within an image of the pupil plane, making images of a secondary, corrected focus on a 256×256-pixels Dalsa CCD with a read-out rate of 1 kHz. The software is implemented in C code running on a Compaq alpha professional workstation XP 1000, which also does the SH cross-correlation calculations. The calculations are done in 0.6 ms. The system switches off automatically when the seeing is bad enough that the subimage motions are out of bounds of the cross-correlation. It switches back on automatically after 10 frames that are within bounds.

The 19-electrode bimorph deformable mirror was constructed by Laplacian Optics, Inc. in Hawaii. The electrode pattern shown in Fig. 1 is deposited upon each of two thin, sandwiched plates of piezo electric material that are glued to the mirror. Voltage applied to an electrode expands one of the plates and contracts the other, which makes the bimorph bend. The thicker ring between the inner and outer electrodes represents the 34 mm pupil. The seven inner electrodes introduce wavefront curvature terms within the pupil, while the outer twelve electrodes produce the correct boundary conditions. The mirror is designed to correct for Zernike modes 2-15, but is capable of also correcting Zernike mode 19.

Optical Setup

The test setup used at the SVST on 14 May 1999 is shown in Fig. 2.

The field lens, L1, reimages the telescope pupil onto the pupil stop near the deformable mirror, M1. The primary focus is one focal length (1524 mm) from the collimator lens, L2. A field stop, FS2, in the primary focus protects M2 from excessive heat. When measuring the control matrix of the adaptive mirror, FS2 is replaced by a large pinhole. L2 puts the M1 in collimated space. Both L2 and M1 are slightly tilted, so that the light is reflected at an angle and hits a flat mirror, M2, that deflects the beam towards the sensor part.

A field stop, FS3, at the secondary focus defines the field of view, so that the microlens images do not overlap. FS3 is replaced by a large pinhole for zero-point calibration of the wavefront. The field lens, L3, re-images the M1 pupil stop onto the microlens array. The microlenses re-image the secondary focus onto the detector, a Dalsa CCD.

A beam splitter deflects part of the light towards the imaging camera, a Megaplus 1.6 with a KAF 1600 CCD.

An On/Off Experiment

On 14 May 1999, we collected 10-ms exposure image data at 500 nm with the megaplus camera. We used a regular cadence of about 5 seconds -- no frame selection was used. The AO system was operated in a mode where it is (allowed to be) on for 60 seconds and then off for 60 seconds, and so on. The seeing was variable, causing the AO system to switch on and off frequently during the AO on periods.

We present here a data set covering a little less than one hour, with two breaks for refocusing the table. The images were divided into two sets, one for AO on and one for AO off. Fig. 3 shows histograms for the RMS contrast of each set. Note that while the AO off set has an almost gaussian distribution, the AO on distribution is markedly skewed towards higher RMS with a tail of low contrast values, probably recorded while the AO system switched off itself.

We also show the data as two quasi-simultaneous movies in Fig. 4. The upper tile is AO on and the bottom one is AO off. Note that the image quality is better and more uniform when the AO system is on. Note also the occasional bad frame, when the AO system probably turned off automatically due to too bad seeing.

Further improvements

After these data were recorded, the system has been improved in the following ways:

Further work will be aimed at improving the bandwidth and the servo as well as parts of the cross-correlation algorithm.

Acknowledgments

The multichannel D/A-converter that connects the computer to the high-voltage amplifier was built in the workshop at the Kiepenheuer-Institut für Sonnenphysik in Freiburg, Germany.
Mats Löfdahl <mats@astro.su.se>
Last modified: Wed Jun 9 13:35:49 MET DST 1999