Investigations of incoherent beam combining using stochastic parallel gradient descent with retroreflector target

Incoherent beam combining of laser beams can increase the available power on target when the output power of a single laser source is limited, and it may also reduce turbulence effects by averaging of scintillations. We have investigated the optimization performance of different variations of the stochastic parallel gradient descent (SPGD) algorithm in a setup where five low power laser beams illuminate a cat's eye retroreflector, and detectors next to the lasers are used to provide feedback for optimization. Angular adjustments of the laser beams are provided by displacement of fiber tips behind collimating lenses. This setup is representative of a dazzling application. Findings include that the optimization demands that there is some initial signal from all laser beams to provide rapid and dependable optimization, which means that the initial pointing errors cannot be much larger than the divergence of the individual beams. Parameter variations show that the sensitivity to settings is relatively low, often a factor two interval of parameter values give an acceptable performance.