A Novel Microprocessor-Controlled High-Voltage Driver for Deformable Mirrors

Genetic algorithms have found widespread use in adaptive optics. One important advantage compared to the matrix inversion method is that it is not necessary to measure the wavefront of the optical beam on which the deformable mirror acts. Instead, any signal, that is monotonic increasing with the quality of the desired wavefront, is sufficient as a feedback signal. Often, this signal derives from a power-in-the-bucket measurement; in the far-field of the beam. In coherent control of chemical reactions with temporally shaped femtosecond laser pulses the signal derives from the rate at which the desired chemical is produced. In our adaptive optics experiments we use micro-machined silicon membrane mirrors with up to 59 electrostatic actuators. We have developed a microprocessor-control led high-voltage driver for up to 60 channels because we could not find a suitable driver on the market. The driver is a fully self-contained unit that only needs input power and communicates with a personal computer via a USB interface. It is especially suited for controlling adaptive mirrors with a genetic algorithm. The driver can store up to 100 voltage patterns, apply them sequentially to the mirror, store up to 4 feedback signals for each voltage pattern, and relay these feedback signals back to the host computer. We will discuss performance issues and tradeoffs like speed, bit resolution, and number of electrodes in our presentation.