White paper on the Advances in Contrast Enhancement for DLP Projection

[Guest Kutaragi]

A White Paper From Texas Instruments Incorporated, DLP TM Products Division, Plano, TX, USA

D. Scott Dewald, D.J. Segler, and Steven M. Penn

 

 

Abstract

 

Increased contrast ratios are essential to the continued success of DLP products in the marketplace, from rear projection televisions and home theatre systems to large venue displays and DLP Cinema. The contrast of DLP projection systems has increased steadily over the past 5 years due to improvements in illumination and projection optics, as well as changes to the DMD itself. The authors present an overview of contrast performance of DLP projectors over time as well as describe illumination and projection techniques that optimize the performance of the DMD.

 

1. Introduction

 

A key metric of image quality for a projected or directly-viewed image is the contrast ratio. Contrast ratio itself consists of two measurements, “on/off contrast”, or full-screen contrast, and ANSI contrast which uses a field of 16 black and white rectangles. ANSI contrast can be used to describe the influence of light scattering on the display, and ANSI contrast cannot exceed the value of full-screen contrast. For computer graphics displays, where images have large areas of white or other bright colors displayed, the ANSI contrast value is a very useful metric. An ANSI contrast value of 300:1 is usually considered sufficient due to the dynamic range limitations of the human eye [1].

 

For video applications where the average luminance level of a scene can vary substantially, many times approaching values of 5% of full white, the full-screen contrast ratio becomes the important metric. As the average luminance drops, the black level perceived by the viewer increases due to adaptation of the eye to the lower scene brightness. For this reason, full-screen contrast ratios of more than 1000:1 are needed to reproduce movie content in a pleasing manner. The author’s experience with DLP Cinema has shown that a contrast ratio exceeding 1300:1 is required to give the viewer an experience comparable to film projection. With regard to television, measurements have revealed that the perceived contrast ratio for images that have an average luminance level of 5% and above are often considered to be nearly equivalent between a rear projection DLP.... television (that is capable of 1500:1) and a direct view CRT television (that is capable of >9000:1)[2]

 

For this reason, Texas Instruments’ DLP Products division has engaged in a program to enhance contrast of both DLP Cinema and home entertainment products, specifically front and rear projection home theater products. In each case the incumbent technology, film and the CRT, respectively, can achieve contrast ratios in excess of 5000:1.

 

2. Contrast Improvements 1996-2001

 

Since DLP technology does not use polarized light, the mechanisms limiting contrast are completely different from those of liquid crystal technologies, e.g. birefringence, skew rays, compensation films, etc. Instead, DLP contrast is limited by the geometrical aspects of the projection lens pupil, the pupil light distribution, scattering, and diffraction effects. Earlier efforts to improve contrast, up until 2001, were concentrated on the device itself rather than the optical system in which it was used [3].

 

Contrast of displays using the digital micromirror device (DMD) have improved steadily since the first products were brought to market in 1996 as seen in Figure 1. The first commercial product used the “hidden hinge” (HH) design. Full screen contrast for these products was in the 200:1 to 250:1 range.

 

Following closely were two improvements to the pixel mirror itself, small rotated via (SRV) and “small mirror gap” (SMG). The via is a hollow post which joins the mirror to the structure below. This presents as a small hole at the mirror surface. Small rotated via was simply the rotation of the via geometry 45 degrees from the original design, along with reducing its dimensions from 4x3 microns to approximately 2x3 microns. Analysis had shown that diffraction and scattering from the via was a significant contributor to lowered contrast, and SRV allowed for a 50% increase in full-screen contrast. Similarly, small mirror gap reduced the gap between pixels, thereby increasing the fill factor of the pixel itself. By reducing the amount of light that can pass between pixels and impinge on the structure under the mirror, scattered light, and therefore the black level, is reduced. SMG allowed for another 30% improvement in contrast ratio.

 

 

With the development of DLP CinemaTM, a development program was initiated to reduce the reflectivity of the structure under the mirror itself (Figure 2). Ray tracing simulation identified the “Metal 3 layer” as a major contributor to this scattered light.

 

When the Metal 3 layer is coated with an inorganic layer, its reflectivity is reduced by 60-70%, thereby reducing scattered light significantly. Use of the “DM3” (dark metal 3) has allowed the DMD to achieve 900:1 to 1000:1 contrast ratios in the same optics that achieved 220:1 contrast with early HH devices.

 

[Guest Kutaragi]

 

3. Optical methods for contrast improvement

 

3.1 Theory of operation of the DMD

 

The DMD modulates input light by changing the angle of reflected light on a pixel by pixel basis. Input light is focused onto the DMD with a certain numerical aperture or f/#. For a typical design, maximum NA or minimum f/# is determined by the tilt angle of the mirrors, either 10 or 12 degrees. Light reflected from “on” pixels is reflected generally normal to the DMD plane toward the projection lens, whereas “off” state light is reflected at a higher angle (40 degrees for a 10 degree device) toward a light dump. There also exists a “flat” state, which corresponds to reflection from the window surfaces, and the mirrors themselves when they are flat. Figure 3 shows a schematic of the light cones. [3]

 

 

3.2 Diffraction and scattering effects on contrast

 

Understanding the mechanisms that limit contrast in DLP systems requires visualizing the light that enters the projection lens pupil in both the on state and the off state. This can be done by imaging the projection lens aperture stop onto a screen. For a telecentric illumination/projection system, the image of the projection lens stop is a mapping of the angular content of the light leaving the DMD. Figure 4 shows such an image for a telecentric DLP system using a 10 degree device at f/3.0. Note the image of the lamp reflector and the periodic structure caused by multiple images of the rod integrator entrance. Also note that the image of the lamp is not centered in the pupil. This is because the center of the illumination bundle is incident on the DMD at 22 degrees, or 2 degrees more than twice the mirror tilt angle. Also, because the DMD hinge axis is orientated 45 degrees to the row and column axes, this displacement of the lamp image is up and to the right along this 45 degree angle.

 

 

Figure 5 shows the same situation with the DMD pixels off. Note the concentration of light in the upper right corner. This corner represents angles that are very close to the flat state. The authors have identified several sources of this light, including:

 

1. Diffraction from the mirror active area

2. Light diffracted or scattered from gaps between mirrors and the vias.

3. Wide-angle scattering due to micro-roughness of the Al surface of the DMD mirror.

 

Diffracted light is contained in orders with maxima corresponding to integral values of m below [4]:

 

sin(è )-sin(è o) = m /d

 

where ë is the light wavelength, d is the period of the periodic structure, and è is the angle of the diffracted light relative to the zero order, that is, the undiffracted light corresponding to the flat state. For a 17 micron pitch DMD, the value for è is approximately 1.8 degrees. Referring to Figure 5, if the diameter of the stop represents an f/3 cone of light, then the radius of the stop represents 9.5 degrees of angle, and the diffracted orders appear to be slightly less than 2 degrees apart in the horizontal and vertical directions.

 

 

Because the light source is incoherent and spread over a large continuous range of angles and wavelengths, the diffraction orders blend together to form an somewhat uniform illumination of the pupil at angles far from the 0 order. This can be seen in Figure 5 as a semi-uniform light level in the lower-left areas of the pupil.

 

Scattering of light trapped between mirrors remains difficult to both measure and model accurately. While it is difficult to measure the contribution of directional scattering in angular space using images of the pupil, locations from which light is scattered can be seen in magnified images of the off pixel which will be described in the next section.

 

Wide-angle scattering is theorized to be caused by the micro-roughness

of the Al alloy that makes up the DMD mirror, which is deposited using a physical evaporation process. At this time, the effect of wide-angle scattering is thought to be negligible compared to scattering and diffraction effects.

 

3.3 Evaluation of the off-state pixel structure

 

To determine which locations of the pixel structure contribute to lowered contrast, the authors imaged off- state pixels onto a CCD at high magnification. A false color image of such is found in Figure 6. Arrows indicate the locations of the pixel corners and the via. It is interesting to note that the via and the corner of the pixel are significant contributors to the black level. Therefore, due to the illumination’s 45-degree angular component, diffraction and scattered light from the pixel edges seem to be negligible compared that from other places. The via, though about 2x2 microns in size, causes a significant diffraction energy even though it covers a fraction of the area of the gap between pixels. This data can be used to modify the structure of future DMD pixels for increased contrast.

 

 

3.4 Control of diffracted light by pupil shaping

 

Use of a shaped projection lens aperture to improve contrast has been proposed [5], but TI has found by experiment that, ideally, both the illumination ray bundle and the projection lens must have shaped pertures. In addition, it was found that the optimal shape and orientation of the illumination and projection apertures is somewhat different than that proposed in [5].

 

In general any blocking of the projection lens pupil will result in a reduction in efficiency, but this can be minimized by using a shift in the illumination light bundle toward a higher angle and blocking the part of the pupil that is not filled with light from the lamp. As the illumination angle is increased, the flat state, which also includes the zero-order of the diffracted light, increases in angle accordingly. This causes more angular separation of the flat state from the projection lens pupil, so fewer of the diffracted orders enter the pupil. Depending on the distribution of light in the illumination bundle, some fraction of the light reflected from the DMD in the on state is allowed to miss the projection lens pupil, causing a brightness decrease.

 

The projection lens pupil is now illuminated non-symmetrically (Figure 4) with a section of the pupil not illuminated at all. This region of the pupil still contributes to the black level, or off state, without contributing to the on state. Therefore this region can be blocked in the projection lens without having a large effect on brightness. After experimenting with several geometrical shapes, it was found that the “cat-eye” shape of Figure 8 resulted in the best compromise between light loss and contrast increase.

 

 

[Guest Kutaragi]

4. Experimental Results

 

Experiments were performed on several DLP projection systems, both prototypes and production models, to quantify the contrast increase vs. brightness decrease caused by incorporating the combination of increased illumination angle and shaped apertures in the optics.

 

The results of the first experiment are shown in Figure 9. The optics platform was a prototype f/3 telecentric RPTV engine using a Philips 120W UHP lamp in a parabolic reflector. The arc was focused onto a rectangular rod integrator using an aspheric condenser. The intent of this experiment was to quantify the increase in contrast relative to brightness loss as the illumination cone central angle is varied from the theoretical value of 2X the DMD tilt angle, leaving all other aspects of the optics constant. This causes a shift of the pupil illumination similar to that shown in Figure 4. There is a significant increase in contrast with a small 46.1 / D. S. Dewald

decrease in brightness as the illumination angle is increased 2-3

degrees. This effect was observed with both 10 degree and 12 degree DMD’s.

 

 

A second experiment used a combination of illumination angle increase and projection lens and illumination optics cat-eye apertures. The optics platform was a production model RPTV engine using a fly-eye (lens array) integrator and parabolic lamp. The engine was retrofitted with a 12 degree DMD. Results are shown in Figure 10. Note the dramatic increase in contrast when both the apertures are used, and that the addition of an aperture that matches the shape and orientation of the aperture used in the projection lens does not have a significant impact on light throughput.

 

It is clear from the experiments that the use of both illumination and projection pupil-shaping apertures is recommended for highest contrast. The goal of the projection lens pupil aperture is to control light that is diffracted and scattered into angles very close to the zero-order. Illumination light is then apertured such that all of the zero-order passes through the projection lens pupil (projection and illumination pupils have the same shape and size in angular space). This creates the optimum condition for obtaining maximum contrast improvement with lowest brightness loss.

 

 

5. Discussion

An image of the off-state pixel was taken after the contrast enhancing apertures were in place, and is shown in Figure 11. Note that the relative brightness of the via to the pixel corners Is lower when the shaped apertures are used. This implies that a majority of the light blocked by the apertures is diffracted by the via rather than the pixel edges or corners. Analysis of this image shows that approximately 50% of the off-state light is diffracted by the via with apertures, versus >70% when no aperutres are used. Clearly the via is a significant contributor to black level and its removal could lead to a significant increase in the contrast ratio.

 

DLP Products has embarked on a program to modify the pixel structure to take advantage of the knowledge gained from this contrast enhancement effort. Based on the above data, the contrast ratio of future projectors using DLP technology is expected to increase significantly

 

6. Acknowledgements

 

The authors would like to thank Erik Wilson of DLP Products for his thorough experiments and concise recording and presentation of the data. They would also like to thank Terry Bartlett, Danny Pyles, and Dan Cahill for their support.

[Guest joamonte]

:o mnnnnn.....so...chin!

 

quite intersesting.....but i think at least take a few day to slightly understand what it said....and maybe by then,i can make my own DLP projector! ;D ;D