DMD Field Reliability: A Comparison of Competing Technologies Used In Data Projectors

To see apdf showing picture reliability results, clickhere. To view it, you must have the Adobe Acrobat Reader, which can be downloaded forfree.


Texas Instruments (TI) launched a study in May of 2002 toinvestigate Digital Micromirror Device (DMD) field reliability andlearn how Digital Light Processing™ technology compared tocompeting technologies. The goal was to obtain objective data on bothDMD and other data projector light modulators in a simulated end useenvironment as a gauge for comparisons in future marketing andreliability activities. A sample of projectors using DMD and LCD lightmodulators were placed in a “head to head” life test. Thepurpose of the test was to gain insight on optical performance overtime. Parametric data measurements and image quality evaluations wereperformed at periodic intervals to gauge useable lifetime, or time tounacceptable performance (based on picture quality), for each projectortested.

Munsell Color Science Laboratory (MCSL) at Rochester Institute ofTechnology carried out this study under the direction of TexasInstruments. MCSL was responsible for data collection as well asprojector use and maintenance. TI was responsible for evaluation andinterpretation of the data.

Our end goal was not only to validate the reliable performance ofthe DMD when compared to LCD, but to learn how DLP™ technologyperforms over time when assembled within its final projectorconfiguration. Based on internal lab tests, we knew the DMD performancedid not degrade for many thousands of hours. We lacked the end-use datathat confirmed what we saw in the lab. We also suspected that LCD couldnot make that claim. Therefore, we rolled this technology into thisstudy to validate DLP™ technology has a competitive advantage inpicture reliability. The QRA group within TI volunteered to perform ahigh level study, to see how data projectors perform over time, and todetermine how competing technologies compared.

Though intended as a preliminary investigation, the followingassumptions were targeted in the study:

  • That optical performance and image quality of LCD technology willdegrade over time.

  • That optical performance and image quality failure modes of LCDtechnology are unrecoverable.

  • That there is an optical degradation in LCD technology thattranslates into an unacceptable image quality defect.

  • That both optical performance and image quality of DLP™technology is only affected by lamp performance, and is thus recoveredas lamps are replaced.


TI selected a sample of the latest data projectors incorporating thefollowing light modulators: five LCD and two DMD. All units were placedin a darkened lab at MCSL and run in a continuous operation mode.

Using the standard, ANSI/NAMP IT7.228-1997 (Audiovisual Systems– Electronic Projection - Fixed Resolution Projectors), MCSLcollected the following parametric data at periodic intervals:

  • Luminance

  • Full On/Full Off (FOFO) Contrast

  • ANSI Contrast

  • Uniformity

  • FOFO and ANSI Contrast for red, green, and blue colors

  • Color Chromaticity for white, red, green, and blue

All units received their input signals from the same sourcematerial. Lamp replacement was determined by either an on-board system,or by 1/2 starting lumens (ANSI standard) where no on-board systemexisted.


After approximately 4700 hours of operation, the followingobservations and trends were noted in the data:

Full On/Full Off (FOFO) and ANSI Contrast degraded over time forLCD, but remained steady for DMD, as expected (see Figure 2). Theoptical degradation seen in LCD washed the screen image out and raisedthe dark levels. Trends in color contrast supported conclusions drawnin FOFO and ANSI for white screen. Graphing just the Dark Levels forall projectors also supported the conclusions drawn from the contrastdata.

Significant changes in Color Chromaticity values supported imagequality degradation seen on the screen for LCD (see figures 3, 4, 5,and 6). Color Chromaticity remained stable for DMD.

The term Picture Reliability was selected to refer to picturequality over time. Within this study we defined Picture Reliability as"The time to unacceptable image quality attributed to the lightmodulator subassembly". A graphical representation of PictureReliability was included in this study (see figure 1). As you can see,the first blemish attributed to optical degradation, occurred at 1368hours for LCD (see figures 7 for sample baseline, and 8 for sampleimage defect at 4700 hours). Most LCD projectors exhibited a subtleyellowing of the screen image, then later developed a blue blemish (seefigure 9 for sample of “yellowing”). Picture qualityremained unchanged for DMD (see figures 10 for sample baseline, and 11for sample image at approximately 4700 hours of use).


Data shows definite signs of supporting our base assumptions. Thereis a marked difference in optical performance and picture reliabilityover time between LCD and DMD projectors.

Image quality defects, caused by the optical degradation were asspectacular as anticipated and occurred quite early in operation, butslower than we hypothesized. Assuming as much as 100 hours of operationper month, we would expect to see the earliest occurrence of anunacceptable picture for LCD at just over one year. The average time toan unacceptable picture for LCD would be at about two years.

Continued downward trends in optical performance for LCD are showingin all parametric data.

Changes in visual "on screen" performance for LCD was not gradual.Its occurrence was sudden and unacceptable under normal viewingconditions. Even the subtle yellowing of the image was judgedunacceptable at the point it became apparent to the naked eye.

Data collected for both on-screen lumens and uniformity did notyield any solid conclusions, as far as the two light modulators testedwere concerned. These two parameters seem driven more by the lamp thanthe modulator.

In Closing

The results of this study proved an eye opener, not only where ourcompetition is concerned, but in overall projector performance as well.It pointed out both that our competitors have areas of improvement toexplore, and that DMD picture reliability remains steady beyond theexpected serviceable life of a data projector. This one fact alone isimportant for any end user requiring a data projector to last for morethan a couple of years. LCD light modulators appear to have a similarlife expectancy as the lamps they employ, therefore, preventing userserviceability of projectors based on this technology. One must returnthe projector to the manufacturer to restore it to its originalperformance levels. DLP™ technology allows user serviceable lampreplacement as the only life limited item in the projector.