Texas Instruments' Test Results: DLP vs. LCD
June 26, 2003,
There has been a lot of email asking for comments on a test recently published by Texas Instruments that examined the reliability of the DLP and LCD technologies. In essence, the test results seemed to indicate that DLP-based projectors deliver stable picture quality over their expected usable life, whereas LCD projectors may be expected to degrade over time. This has prompted two obvious questions: Was TI's test valid? And should the results of the test be taken into consideration by consumers who are about to purchase a projector? This article will attempt to address these questions for readers who are concerned about it.
Last year Texas Instruments commissioned a lab test to compare the relative stability and longevity of the DLP and LCD technologies. The test commenced in May, 2002. Two DLP projectors and five LCD projectors were run 24 hours a day, 7 days a week for five months, with breaks only to change lamps as needed. During this time each projector was periodically measured for lumen output, contrast, uniformity, and color chromaticity for white, red, green, and blue.
Though sponsored by TI, the test itself was conducted at the Munsell Color Science Laboratory (MCSL) ( www.cis.rit.edu/research/mcsl/), Rochester Institute of Technology. The technical measurements were taken by MCSL personnel. The selection of the models to be included in the test was done by TI. The interpretation of the results and the publishing of the conclusions was done by TI.
Texas Instruments released the results of the test to the public in March, 2003. In summary, the test results indicated that the two DLP projectors used in the evaluation delivered stable contrast and color balance that remained relatively unchanged for over 4,000 hours of continuous operation. Meanwhile, the five LCD projector test units tended to shift color balance and lose contrast over time. Based on the judgment of TI personnel overseeing the test, the image quality of the LCD projectors eroded fairly rapidly, eventually degrading to the point of becoming subjectively "unacceptable." TI defined this unacceptable condition as the point at which TI believed the picture quality was sufficiently degraded that an average user would not be satisfied with it. The first LCD projector was judged to reach this unacceptable condition in just 1368 hours of operation. The remaining four units were said to have degraded to an unacceptable state in 2160, 2352, 3456, and 3456 hours respectively.
Other than to say that the five LCD models selected for the test were popular products in the marketplace at the time the test commenced in May, 2002, TI has not disclosed either the models or manufacturers of the LCD units. They did release some limited spec information as noted below. However some of the salient technical specifications that might reasonably be suspected to have a bearing on the outcome including specific size and weight of the units, and lamp type and wattage, was not disclosed.
Texas Instruments claims that the results of this test are evidence that DLP technology is superior to LCD when it comes to "picture reliability" over the projector's anticipated lifetime. Picture reliability is defined by TI as the ability to maintain consistent image quality throughout the life of the projector.
Description of the Lab Test
The test was conducted at the facilities of the Munsell Color Science Laboratory in Rochester, NY. It was carried out in a dedicated 10 x 18 foot room. Eight projectors were operated essentially 24 hours per day, 7 days per week. These included the two DLP projectors and five LCD projectors already noted, plus an LCOS-based projector. The technical performance data pertaining to the LCOS machine was not included in the final report issued by TI since a sample of one is not sufficient to support any conclusions about the technology.
The two DLP and five LCD machines were portable-class machines. Six of the seven units (both DLPs and all but one of the LCDs) were XGA resolution. The fifth LCD unit was 16:9 format of unspecified resolution. The five LCD projectors consisted of three with 0.9" panels and two with 0.7" panels. The DLP units represented one each of 0.9" and 0.7" chips. Both DLP projectors were rated at 2000 ANSI lumens. The five LCDs had brightness ratings of 800, 1000, 1100, 2000, and 2000 ANSI lumens.
The eight units in the test were placed in fairly close proximity, from a minimum of 4 to 5 inches, to as much as a foot or more apart. They were placed on three shelves one above another, with several units on each shelf. They were arranged in a manner to prevent the hot exhaust of one unit feeding the intake vent of another. At the end of each shelf a fan was installed to blow air across all units on that shelf. The objective of these fans was to distribute cool air from the air conditioning vents as evenly as possible over all units.
The room was cooled by a central air conditioning unit operated by a wall-mounted thermostat located about 10 to 12 feet from the projectors. Average ambient temperature in the room during the test was 25 degrees C, or 77 degrees F. The actual temperature variance range around the average any given point in time was about ten degrees F, from just under 75 degrees to the mid-80's. Temperatures rose and fell in this range with the cycling of the air conditioning system.
The projectors were run round the clock seven days a week, with downtime for the changing of lamps and filter cleaning/replacement as necessary. They were all fed the same computer data signal with rotating graphic images to prevent burn-in. Technical performance measurements were taken at days 0, 1, 2, and 4; weeks 1, 2, and 4; and months 2, 3, 4, and 5.
Results as reported by Texas Instruments
At the end of about 4700 hours of operation, TI summarized the results as follows:
1. Full On/Off, and ANSI contrast degraded over time on all five LCD units, but remained relatively constant on the two DLPs.
Based on these test results, TI suggests that a fundamental flaw exists in LCD technology that causes the picture quality it delivers to deteriorate well before the end of life of the projector itself might be expected. Because DLP technology is allegedly immune to degradation, it is purported to offer a lower cost-of-ownership since DLP projectors do not need to be replaced as often as LCD-based products.
While the test conducted by TI clearly highlights a failure mode in LCD panels and polarizers, the public must take care not to jump to the conclusion that the failure rates produced in the lab are in any way indicative of expected failure rates in the field. To the contrary, the statistical results in this test do not match up well with general marketplace experience.
To illustrate, let's assume a typical three-hour per day usage on a portable projector. If it fails in just 1368 hours as one of the units did in this test, that translates to a usable lifespan of about 15 months. Taking it one step further, TI's test results indicate that three of the five units (60% of the test population) failed in 2352 hours or less. If we were to use this data to predict the lifespan of LCD projectors in general, we would conclude that with a typical 3-hour per day usage, 60% of all LCD projectors sold would degrade into a condition unacceptable to the user in just a bit over two years.
This extrapolation is not credible. LCD is by far the most popular projector technology in the marketplace, outselling DLP worldwide by a factor of about three-to-one in the last 18 months. Major brand manufacturers including Epson, Panasonic, Sharp, and Sony have projector product lines either exclusively based upon LCD technology or heavily oriented toward it. These companies have outstanding reputations for quality and are not known for marketing products that routinely degrade to failure within a year or two.
Furthermore, the major LCD manufacturers do not have truckloads of bad units coming back from users and dealers. If they did, they would quickly discontinue those LCD product lines since the cost of warranty repair claims would be unbearable. Dealers would stop carrying them to avoid customer complaints, the cost of handling returned items, and the damage to their own reputations for having recommended faulty merchandise. None of this is happening. So there is no evidence that the failure rates documented in the test correspond in any realistic way to typical LCD performance in the field.
So what happened in the test to skew the results?
There is no doubt that a common failure mode appeared in all five LCD products in the test. We are confident that the measurements taken by Munsell Color Lab are reliable. We are also confident in TI's representation that image quality on the LCD units in question degraded significantly in unusually short periods of time. The question is what could have caused this to occur? We believe some factor or combination of factors not present under typical usage was acting upon the units in the lab test, and that these factors accelerated failure rates to an alarming degree. We need to examine what those influences might have been. Possibilities include the following:
1. 24x7 operation. Clearly an obvious difference between the test environment and typical usage was 24x7 operation. The vast majority of LCD projectors sold are not deployed in 24x7 duty cycles. Constant round the clock operation is a high stress factor that contributes to a breakdown of compounds in the polarizers and panels if they are not properly cooled. Because of this projector manufacturers who build products intended for 24x7 operation typically use larger LCD panels, include comprehensive UV protection, and build them into larger form factors with substantial cooling systems that move large volumes of air. Units with this type of design were not included in the test.
Manufacturers recognize that the organic compounds in LCD panels and polarizers are susceptible to high heat and light energy stress, and will eventually break down if deployed in high stress environments-in particular 24x7 operation with higher than normal ambient temperatures. Compact portable LCD projectors are in general not recommended for 24x7 duty cycles because of this.
On the other hand, DLP technology does not use organic compounds. Thus the elements which can be expected to degrade over time under high stress in an LCD projector do not exist in a DLP projector. Therefore when these two technologies are placed side-by-side in an unusually high stress environment as they were in this test, the DLP-based products should be more resistant to image shift over time. TI's test demonstrated this in no uncertain terms.
We agree with TI's assessment that high intensity blue and UV light in the blue channel contributes to accelerated breakdown of the organic compounds in that channel. However degradation due to high intensity light is not normally expected to occur at the rates documented in the test unless the components are subjected simultaneously to abnormally high heat stress. Therefore we suspect that 24x7 operation, higher than normal ambient temperatures, and the close proximity of the test units to one another may have combined to create abnormal conditions that led to a more rapid and severe degradation of the components than users would typically experience.
Thus the generalized inference that many observers have drawn from the test data, which is that LCD technology itself may be expected to routinely break down under normal usage before the expected lifespan of the projector, is difficult to sustain based upon the limited sample size and the abnormal conditions we believe may have existed in the lab.
The test at Munsell Color Science Lab clearly draws attention to the fact that LCD technology has a failure mode that does not exist with DLP, and that this failure mode becomes readily apparent in an unusually high stress environment. What the test does not tell us is how much of a problem this really is in real life. In addition to the test results there is ancedotal evidence of LCDs eventually breaking down in extended use applications. When this occurs the problem is usually fixed by replacing the LCD panel and polarizer in the blue channel. But for the most part typical users of LCD projectors do not seem to experience either the severity of degradation or the rapidity with which it occurred in this particular lab test.
To explore this further a more comprehensive set of lab tests must be run. In order to generate test results that would more reliably predict the long term reliability of LCD and DLP technologies in the field, we would need to incorporate the following elements:
1. A wider array of both LCD and DLP products needs to be included in the sample. LCD machines featuring 1.3" and 1.8" glass must be included along with the smaller formats. SVGA and widescreen resolutions in both WSVGA and WXGA should be included. Furthermore a sample of two DLP units is not adequate to reveal potential long term failure modes that might exist with that technology as well.
Clearly such a test is beyond the scope of what Texas Instruments originally had in mind. Their interest was in the behavior of portable units, and their test was never intended to address the reliability of larger machines built for fixed installation 24x7 operation, LCOS technology, and so on.
Such a test would take at least a year, and would require the cooperation and funding support of the major DLP and LCD manufacturers in the industry. If there is general interest among manufacturers, ProjectorCentral would be willing to manage it. Those who may wish to participate and/or discuss this further should drop an email to us at firstname.lastname@example.org.