2510332 - February 2009
or attaching a thick plate to the exit face, thus displacing the dust to an out-of-focus position, or by sealing the space between the exit-face and the first element downstream.
Hollow mirrored tunnels increasingly are used in smaller systems for several reasons. First, they produce more integration per unit length because they are working in air (index = 1) instead of glass. Hollow tunnels can be shorter than solid rods for the same application. Second, there is no output face to collect dust or imperfections to image to the screen. Third, there are no faces to AR coat, which makes the overall efficiency about the same for short lengths. Fourth, they are much simpler to mount because there are no TIR failure points. The only drawback is the tendency to operate at higher temperatures due to the absorption losses in the mirror coatings, which can weaken adhesives if used to assemble the mirror sections, or create localized heating issues. Also, sizing adjustments are easier and faster to implement with a tunnel.
Design considerations. Typically, TI recommends an integrator length that will produce an approximately 4 x 5 array of arc images for acceptable uniformity, depending on application. Some applications require up to 8 x 10 array images or more, especially if there is arc flicker or arc jump to mitigate. It is recommended that uniformity be modeled in optical design software to determine the optimum length based on the arc profile, far-field distribution, lamp-focus position, and length of the integrator. As the rod length is decreased, the sensitivity of uniformity to lamp-focus tolerances becomes more critical.
An image is created in the array each time the marginal ray crosses the optical axis. The array can be observed in any pupil of the illumination relay, where there are multiple images of the cross section of the rod in a rectangular array, each image containing an image of the arc as viewed by the angles subtended to the arc from that image. The outer array images represent the highest angles of light from the lamp reflector; the interior array images are the shallowest angles. In the center, typically there is a dark spot representing the innermost angles shadowed by the lamp electrodes.
Cross-section sizing. The size of the cross section is determined by optical performance of the illumination relay, the assembly tolerances, and the size and tolerance of the device. At a minimum, the size should prevent any chromatic artifacts or vignetting in the image at the device (screen). Tolerance stackups in the size, position, and magnification of the image of the integrator through the optical system, and the mechanical tolerances from the integrator mount to the device die in the DMD package itself, must all be accounted for. This can be done by oversizing the integrator cross section to always allow the device to be in the image of the integrator, but it results in large overfill lumen loss (efficiency) because the area outside the device contains light that otherwise could be useable. The loss is even more significant for highly uniform profiles created by longer integrator lengths, because the outer areas are nearly as bright as the central areas. A better, more common approach is to size the cross section for minimum overfill and adjust the position of the image to the device at some step in projector assembly. This is a relatively simple process (see Figure 4), and it eliminates many tolerance stackups to allow maximum brightness.
May not be reproduced without permission from Texas Instruments
Copyright 2009 Texas Instruments Incorporated