Characterization of the T/T conditions at Gemini Using AO data

1. Characterization of the T/T conditions at Gemini Using AO data Jean-Pierre Véran Lisa Poyneer AO4ELT Conference - Paris June 22 - 26, 2009

2. Motivation • Good AO correction requires T/T to be corrected to a residual << than the size of the diffraction limited spot: • For ELT, this corresponds to a few mas rms • This level of correction has never been achieved so far • High contrast imagers on 8-meter class telescopes have similar requirements (coronograph) • GPI: requirement 5 mas rms (tip+tilt); goal: 3 mas rms (tip+tilt) • T/T can have different sources, not very well characterized • Atmospheric turbulence • Telescope windshake • Vibrations originating from telescope systems and/or instruments • Can we learn anything on the actual T/T conditions at Gemini from the various wave-front sensors currently in operation on the telescope ? • Do we need to implement mitigation strategies: • Improved control algorithm ? • Dedicated T/T WFS at higher sampling rate ?

3. Total tilt: 116 mas rms Total residual tilt: 0.9 mas rms (simple integral controller) T/T from atmospheric turbulence: easy, in theory

4. WFS telemetry data from Gemini • P2 and OIWFS data (200 Hz, open loop) • 7 CBs from GN and 5 CBs from GS (March 2007) • Altair data (1 kHz, closed loop) • 61 CBs from Gemini North (2007-2008) • Acquired during M1 tuning • NICI data (1.3 kHz, closed loop) • 46 usable CBs, mostly from one run (November 09) • Closed loop data have to be turned into open-loop data • Use model of the AO rejection transfer function • Accurate knowledge of system calibration is required

6. Typical NICI PSD Dashed line: measured residual T/T Solid line: reconstructed incoming T/T Dotted lines: power -3 fit + noise level Reconstructed incoming T/T always decreases as a ~-3 power law, much slower than -17/3 Residuals after rejection of a GPI type system are still very high (> 10 mas rms) due to slow drop-off Is the -3 drop-off real or is it a Measurement artifact ?

7. Aliasing effect revealed by simulations White curves: • Original incoming tilt (solid) • -6 power law fit (dotted) Green curves: • Incoming tilt reconstructed from telemetry (solid) • power law fit (dotted): • -4 (top) • -6 (bottom) • Measured residuals (dashed)

8. Modeling the aliasing effect • At least 97% of the estimated GPI residual power is an artifact of spatial aliasing on NICI WFS • True GPI residual rms is no more 15% than GPI residual computed from telemetry data

9. If all low frequency T/T is atmospheric turbulence T/T… • Fit an inverse polynomial to reconstructed T/T in [10,50] Hz range. • Apply GPI rejection transfer function and discount residual rms by 85% Average atmospheric T/T residual @ r0 = 0.145 cm is ~ 1.7 mas rms at 2 kHz CAVEAT: Aliasing might hide windshake Windshake is not affected by aliasing and Should not be discounted !!!

10. High frequency vibration background 279 Hz vibration gets amplified by a factor 2.16 (rms) by a 2 kHz GPI rejection transfer function (simple integral controller) 279 Hz vibration changes in amplitude: 2.1 mas rms +/- 0.8 mas rms Total GPI residual @ 2 kHz, r0=14.5cm 4.8 mas rms +/- 0.8 mas rms Total GPI residual @ 3 kHz, r0=14.5cm 1.6 mas rms +/- 0.4 mas rms

11. Common path vibrations Non-common path vibration Vibration mitigation strategies • Kalman filter, using the framework already in place in GPI for predictive control of the Fourier modes (T/T excluded) • L. Poyneer, B. Macintosh and J.-P. Véran, Fourier transform wavefront control with adaptive prediction of the atmosphere, J. Opt. Soc. Am. A, Vol. 24, pp. 2645, 2007. • Formalism can be readily extended to handle vibrations • Vibrations considered as colored noise: L. Poyneer, et al., in preparation • Similar approach to: C. Petit, J. M. Conan, C. Kulcsar, H. F. Raynaud, and T. Fusco, First laboratory validation of vibration filtering with LQG control law for adaptive optics, Opt. Exp.16, 87–97, 2008. • Effect is to tailor the rejection transfer function to: • Notch out vibrations if in science path • Ignore vibrations if in WFS path • Other possible vibration mitigation strategies include: • Explicit notch filter built into the AO controller • Using local pure oscillators whose frequencies and phases are continuously updated by a phase-locked loop: • Di Lieto et al., SPIE Marseille, 2008

12. Excess low/mid frequency T/T mitigation strategies • Increase rejection by using a type II controller • Double integrator with a lead filter • Suitable for woofer-tweeter control. Current baseline for NFIRAOS • See Wang et al., this conference • See Véran et al., OSA San Jose conference Oct 09

13. Conclusions • Spatial aliasing severely limits what we can learn about T/T conditions from AO WFS data • Also affects Shack-Hartmann WFS, although possibly less so • Subtraction of a simple model of the aliasing is possible, but would hide windshake • NICI telemetry data reveal several high frequency vibration lines that would severely limit GPI at 2 kHz • Mostly a ~ 2 mas rms vibration at 279 Hz • GPI would barely meet its 5 mas rms residual requirement • Such high frequency vibration is likely to come from within NICI • Mitigation strategies exist: • Kalman filtering for vibrations • Type II controller for windshake