Authors

Niek Doelman (1) Rufus Fraanje (2) Remco den Breeje (1)

Affiliations

(1) TNO Delft (2) Delft University of Technology

Abstract

In recent years various researchers have concentrated on control performance improvement for adaptive optics systems by using more sophisticated design methods. These approaches account for the inherent spatial and temporal correlations in the wavefront sensor data. Several control schemes have been proposed, of which the common essence is the minimization of a criterion function, yielding so-called ‘optimal’ or LQG control solutions. These are in some cases also referred to as ‘predictive control’. Following the H2-optimal control design approach proposed by Hinnen [JOSA A Vol. 24, 2007], a real-sky experiment has been carried out on the McMath-Pierce solar telescope on Kitt Peak, Arizona. The purpose of the experiment was to validate the favourable results of optimal control, as obtained in simulations and laboratory experiments, on a real-time AO system on a telescope with real-sky turbulence. During the experimental week, it appeared that the deformable mirror did not have sufficient stroke to cope with the strong wavefront aberrations as measured by the AO wavefront sensor. Therefore, it was decided to focus on optimal control of the lower aberration modes tip and tilt only (using the separate TT-mirror). The control experiments demonstrate that for the particular AO system and seeing conditions (Nov 14, 2010) real-time optimal control can reduce the tip and tilt amplitudes by an additional factor of about 2 (RMS), compared to the common integrator control of the tip and tilt modes. For the low frequency band the improvement ranges from 10 to 20 dB. This performance agrees reasonably well with the predicted performance which is based on off-line analysis of the WFS data. The paper will discuss the experimental results in detail and also address important aspects like the non-stationarity of the wavefront aberrations, coupled versus decoupled tip-tilt control and measures to increase the robustness of the controller.