Adaptive mirror control for an optical resonator cavity

An adaptive mirror control strategy for tuning an optical resonator is described. The control approach is model-free and uses a small-amplitude perturbation signal to enable real-time estimation of the cost function gradient. The strategy has similarities with the approaches of `stochastic approximation' and `extremum seeking control', which are well-known methods in the control community. Particular fine-tuning of the algorithm, such as perturbation filtering, cost function shaping and disturbance feedforward are discussed. The control strategy has been validated on a (quasi-optical) millimetre wave resonator system. A significant disturbance acting on this set-up is the non-stationary frequency of the wave source (gyrotron). In various experiments it has been demonstrated that the control approach is well capable of keeping the optical cavity in resonance, to track wavelength variations and to reject disturbances in the low and mid-frequency range. The only performance limitation of the approach occurs in cases which require fast disturbance tracking. This is due to the bandwidth restrictions in the underlying mirror motion system.