Implement more robust DM response recording scheme with Hadamard modes
We've been doing it wrong all the time ... again. Recording a response using individual actuator pokes is "the worst possible way" to do it as the signal to noise ratio is very bad and the temporal evolution of the wavefront during recording is not cancelled out very efficient. Turns out that acting on all the DM channels all the time in an orthogonal fashion is very beneficial and will lead to a much cleaner response, which increases stability and convergence speed of the loop. Using Hadamard modes as channel offsets seems to be an excellent approach.
Here's a paper - you can skip to section 5. https://www.researchgate.net/publication/241763342_Calibration_of_a_curvature_sensor_bimorph_mirror_AO_system_Interactiocn_matrix_measurement_on_MACAO_systems