Deformable mirrors with very high order correction generally have smaller dynamic range of motion than what is required to correct seeing over large aperture telescopes. As a result, systems will need to have an architecture that employs two deformable mirrors in series, one for the low-order but large excursion parts of the wavefront and one for the finer and smaller excursion components. The closed-loop control challenge is to a) keep the overall system stable, b) avoid the two mirrors using control energy to cancel each other's correction, c) resolve actuator saturations stably, d) assure that on average the mirrors are each correcting their assigned region of spatial frequency space. We present the control architecture and techniques for assuring that it is linear and stable according to the above criteria. We derived the analytic forms for stability and performance and show results from simulations and on-sky testing using the new ShaneAO system on the Lick 3-meter telescope.
11 pages, 4 figures, Presented SPIE Astronomical Telescopes and Instrumentation Conference, paper #9148-174