Bulk solid optics have helped circumvent the need to align multiple free-space optics within communications systems by substituting multiple mountings of discrete optical components with a single integrated optical unit. The same goal spurred interest at NEC Research Institute in Princeton, N.J., in a bulk optic filtering device called an X-cube.Four identical prisms inside this package form an X-cube. Depending on what coatings are applied to the interior interfaces, the cube can serve as a lossless 4 x 4 beamsplitter, a star coupler or a wavelength router for coarse wavelength division multiplexing applications. Courtesy of Jan Popelek.Developed by Jan Popelek and Yao Li, who have since moved on to Phaethon Communications Inc. in Fremont, Calif., the assembly bonds four identical right-angle rooftop prisms that touch at the center. This forms a cube with two mutually orthogonal and intersecting internal planes that look like an "x." Each prism has a 5-s angular precision for the 90° angle and a 15-s precision for both baseline angles. Interferometric measurement of all three optical planes showed that their surface flatness was within one optical fringe.Depending on how the interior surfaces are treated, the cube could serve as a lossless 4 x 4 beamsplitter, a star coupler or a wavelength router for coarse wavelength division multiplexing applications. In preliminary experiments, Popelek and Li applied a dielectric coating on one rooftop plane of each prism. The coatings transmitted 50 percent of the 1300-nm light. They attached two fiber collimators to each side of the prism housing to act as an optical input and output.The cube was made and mounted by hand. Its dimensions were 35 mm2 without the housing. At those dimensions, it was possible to align the collimators with a screwdriver, but Popelek noted that cube size depends on the size of the collimator.With added polarization controls, the cube demonstrated a 2.1-dB insertion loss and a uniformity variance of 0.279 between channels.The most difficult part of the cube's manufacture -- optimizing angular alignment -- was also its most crucial. Most of the 2.1-dB loss came from angular misalignment between collimators, said Popelek, who added that precision was most important at the rooftop angle of each prism."If you can't make it perfectly 90°, then you can't glue them together effectively, and your losses multiply," he explained. Part of this problem presumably could be worked out in a manufacturing process.NEC Research has not pursued development of the X-cube since Popelek and Li's departure, according to a company spokesman.