Laboratory Demonstration of Real-Time Focal Plane Wavefront Control of Residual Atmospheric Speckles

Abstract

Current and future high contrast imaging instruments aim to detect exoplanets at closer orbital separations, lower masses, and/or older ages than their predecessors. However, continually evolving speckles in the coronagraphic science image limit contrasts of state-of-the-art ground-based exoplanet imaging instruments. For ground-based adaptive optics (AO) instruments it remains challenging for most speckle suppression techniques to attenuate both the dynamic atmospheric as well as quasi-static instrumental speckles on-sky. We have proposed a focal plane wavefront sensing and control algorithm to address this challenge, called the Fast Atmospheric Self-coherent camera (SCC) Technique (FAST), which in theory enables the SCC to operate down to millisecond timescales even when only a few photons are detected per speckle. Here we present the first experimental results of FAST on the Santa Cruz Extreme AO Laboratory (SEAL) testbed. In particular, we illustrate the benefit of "second stage" AO-based focal plane wavefront control, demonstrating up to 5x contrast improvement with FAST closed-loop compensation of evolving residual atmospheric turbulence-both for low and high order spatial modes-down to 20 millisecond-timescales.

High Speed Focal Plane Wavefront Sensing with an Optical Chopper

“…Our laboratory setup, mimicking Fig. 1a, uses SEAL-the Santa Cruz Extreme AO Laboratory (Gerard et al 2022a, Jensen-Clem et al 2021. We use a Thorlabs KLS635 laser (set at 0.15 mW unless otherwise noted), ALPAO 97 actuator DM (which also serves as the system pupil stop, Andor Zyla 5.5 sCMOS detector, Thorlabs WFS-20 SHWFS (for this paper used only to flatten the ALPAO DM), a Thorlabs MC2000B optical chopper with blade MC1F10, and a Stanford Instruments DG535 controller and an in-house electrical doubler so that the Andor detector readout is synchronized with the optical chopper (with the chopper as the leader and the Andor camera as the follower 1 ) to produce a chopped and un-chopped pupil every other frame, respectively.…”

“…We use a Thorlabs KLS635 laser (set at 0.15 mW unless otherwise noted), ALPAO 97 actuator DM (which also serves as the system pupil stop, Andor Zyla 5.5 sCMOS detector, Thorlabs WFS-20 SHWFS (for this paper used only to flatten the ALPAO DM), a Thorlabs MC2000B optical chopper with blade MC1F10, and a Stanford Instruments DG535 controller and an in-house electrical doubler so that the Andor detector readout is synchronized with the optical chopper (with the chopper as the leader and the Andor camera as the follower 1 ) to produce a chopped and un-chopped pupil every other frame, respectively. As in Gerard et al (2022a), ALPAO DM units are converted to WFE units via a single scalar multiplication, calibrated via open tip/tilt measurements and corresponding PSF images, but for which higher order conversions become increasingly wrong due to the decreasing stroke limits of the DM as a function of mode order.…”

“…Unless otherwise noted, our standard laboratory configuration for chopper WFS image acquisition is with the chopper blade running at 100 Hz, using the doubler so the Andor camera follower runs at 200 Hz with a 0.1 ms exposure per frame (i.e., a 2 % duty cycle), and also implementing a 2.5ms phase delay for the doubled chopper signal with the Stanford controller. As in Gerard et al (2022a), we also implement a serial 20 ms pause in between acquiring a chopper pair of images and sending any DM commands (both for calibration and real-time control software) due to additional latency from our Python interface to hardware components. Two random chopper imager pair acquisitions are not necessarily in the same order (i.e., one could have the chopped frame first while the other could have the un-chopped frame first), but for consistency we order every chopper sequence with the chopped frame first, as in Equation 1, determined as the frame with less cumulative flux.…”

“…6b is 1.2), illustrating the potential of this technique to stabilize quasistatic temporal disturbances on-sky (e.g., thermal variations, flexure, and additional sources of slow beam wander). Closed-loop control was manually tuned for optimal performance using a integrator controller with gains of 0.01, 0.003, and 0.008 for the TTF, low order Zernike, and high order Zernike modal groups, respectively; such low gains were necessary for loop stability due to SEAL's highly stabilized ∼nm-level on-air WFE (Gerard et al 2022a), which as shown is mostly below the noise floor of individual frames of Fig. 6.…”