A time domain method using MAFIA code has been developed to calculate narrow band beam impedance in RF cavities over a wide range of frequency spectrum. The impedance is obtained through Fast Fourier Transformation (FFT) of computed wakefield by the MAFIA. Analysis of the calculated impedance spectrum will be presented. Application of the method to a known RF cavity design (PEP-II cavity) has shown good agreements with bench and beam measurements. The method has been applied to the RF cavity design of Damping Rings for the Next Linear Collider (NLC).

A high-yield neutron source to screen sea-land cargo containers for shielded Special Nuclear Materials (SNM) has been designed at LBNL [1,2]. The Accelerator-Driven Neutron Source (ADNS) uses the D(d,n)3He reaction to create a forward directed neutron beam. Key components are a high-current radio-frequency quadrupole (RFQ) accelerator and a high-power target capable of producing a neutron flux of >107 n/(cm2 cdot s) at a distance of 2.5 m. The mechanical design and analysis of the four-module, bolt-together RFQ will be presented here. Operating at 200 MHz, the 5.1 m long RFQ will accelerate a 40 mA deuteron beam to 6 MeV. At a 5 percent duty factor, the time-average d+ beam current on target is 1.5 mA. Each of the 1.27 m long RFQ modules will consist of four solid OFHC copper vanes. A specially designed 3-D O-ring will provide vacuum sealing between both the vanes and the modules. RF connections are made with canted coil spring contacts. A series of 60 water-cooled pi-mode rods provides quadrupole mode stabilization. A set of 80 evenly spaced fixed slug tuners is used for final frequency adjustment and local field perturbation correction.

We develop a rapidly converging algorithm for stabilizing a large channel-count diffractive optical coherent beam combination. An 81-beam combiner is controlled by a novel, machine-learning based, iterative method to correct the optical phases, operating on an experimentally calibrated numerical model. A neural-network is trained to detect phase errors based on interference pattern recognition of uncombined beams adjacent to the combined one. Due to the non-uniqueness of solutions in the full space of possible phases, the network is trained within a limited phase perturbation/error range. This also reduces the number of samples needed for training. Simulations have proven that the network can converge in one step for small phase perturbations. When the trained neural-network is applied to a realistic case of 360 degree full range, an iterative scheme exploits random walking at the beginning, with the accuracy of prediction on phase feedback direction, to allow the neural-network to step into the training range for fast convergence. This neural-network-based iterative method of phase detection works tens of times faster than the commonly used stochastic parallel gradient descent approach (SPGD) using a single-detector and random dither when both are tested with random phase perturbations.

We have generated 81 independently controllable beams using a spatial light modulator and combined them on a diffractive combiner, to characterize the combiner and develop a fast phase error detection scheme. A key parameter of the diffractive combiner is measured in a new way, enabling an efficient combination when programming calibrated phases of each beam. This testbed provides a platform for development of advanced feedback phase control of high channel-count beam combination.