Enlarging the Control Space of the Pyramid Wavefront Sensor: Numerical Simulations and Bench Validation

We investigate the currently experienced limitations regarding the use of the Pyramid Wavefront Sensor (PWFS) in the coming generation of very high order adaptive optics (AO) systems for ELTs. As a focal plane wavefront sensor, the PWFS concept paves way for a telescope diffraction limited AO regime, hence potential 80 times or more sensitivity improvement over a comparable Shack-Hartmann (SH) on large telescopes. Although some theoretical similarities exist with quad-cell SHs, the focal/pupil planes inversion brings up significantly different modelizations. Diffractive effects between PWFS pupil images (quadrants), nonlinear behavior –limited dynamic range, optical gain fluctuations– are PWFS features that require AO systems and controllers be specifically engineered. Demonstration of physical validity and robustness of PWFS models on sky-ready, ELT-scaled AO systems remains an open topic to date. Obtaining a high quality pyramidal prism and a model-consistent WFS assembly is noticeably a critical factor. We demonstrate that the traditional gradient sensing controller is extremely sensitive to prism shape defects and common assembly misalignments: we generically analyze the effect of sensor-plane quadrants individual translations, which are induced by a variety of mechanical defects. Quadrant misregistrations displace wavefront information to terms not included in the conventional slopes maps, and high spatial frequencies become invisible to the WFS. Expanded Space Control (ESC) for quad-cell signal incorporates additional WFS signals in addition to the Xand Y-axis differences, and compensates misalignmentinduced information loss, therefore dramatically relaxing machining and alignment constraints for PWFS engineering. We here present validating results of ESC performance and robustness for quadrant misregistration cases, both in end-to-end simulations with COMPASS and on the PYRCADO PWFS demonstrator bench at LESIA.