Tuesday, December 19, 2006

A New Bedroom HDTV for Me, Part IV

In A New Bedroom HDTV for Me, Part III I described further the "mura" problem of my new Sony Bravia KDL-40XBR2 1080p LCD HDTV, originally mentioned in A New Bedroom HDTV for Me, Part II.

Since I discovered the problem, I've been hanging out in the Official Sony 46" XBR LCD Uneven Backlight/Cloudy Thread. It represents the problems of hundreds of owners of flawed 46" and 40" XBR LCDs who have responded to a poll. The poll asks about Sonys with "clouds" of extra brightness on portions of the screen.

At present writing, 283 responders have said, "YES, clouds can been seen when dark colors are displayed or when switching inputs." Only 133 say, "NO, my screen has a perfect, fully even, backlight."

Any problem with the picture uniformity of an LCD panel can be called a mura defect. Picture uniformity is what happens when the panel can display full-field test patterns of various shades of gray, ranging from reference black to peak white, with no blemishes or signs of unevenness. Mura is different from dead or stuck pixels, problems which can also beset LCD panels.

Most mura problems involve variations in brightness, not color, but sometimes it is the case that color is not uniform over the entire screen as well. The variations, whether of brightness or color, are often subtle. A brightness variation, for instance, will typically be of such low contrast that one simply cannot see it with ordinary program material. That's when full-field test patterns come in handy.

A poor man's test pattern for the kind of "cloudy backlighting" mura most posters to the thread complain of happens when the Sony is tuned to an unused video input. That turns the screen black, and any clouds can readily be seen. Another way to get the same result is to cue up program material with a black background, such as the end credits of a movie.

backlight2To the right is a photo of one poster's problem "clouds." It comes from this album which contains several other shots posted to the thread by various Sony customers.

Notice that the "clouds" problem, which is called "cluster mura," looks different than the "vertical-band mura" my set has (see image at right). My problem shows up on light-gray fields. The "clouds" show up on dark fields.

I have also seen reports of horizontal banding and of light "bleeding" at the very edges of the screen.

The Sonys aren't the only LCDs with "picture uniformity" problems. I just got my January 2007 issue of Sound & Vision. They test the Toshiba Regza 47-inch 1080p LCD HDTV, saying:

... I noticed that the sky on either side of the screen seemed a little lighter than the center. Earlier, I had observed that on dark gray test patterns, the Toshiba's image was slightly brighter in its far left and right zones than in the middle, and now here it was in program material. This kind of imperfection is not unknown to LCDs. (JVC's 46-inch LT-46FN97, reviewed in December, suffered the same fault.) It may have been endemic only to my sample ...


Full-field gray patterns below 50 IRE were slightly darker in the middle than on the sides.

From the review of the Sharp LC-52D62U 52-inch 1080p LCD TV:

... when looking at solid gray test patterns, I noted a distinct pattern of horizontal dark bands across the TV's screen. The bands weren't as visible with regular programs, althought they did show up pretty clearly on occasion. ... Fortunately, a second review sample from Sharp had markedly better picture uniformity, with almost no banding visible on regular programs or test patterns. (Sharp is aware of the issue and said anyone encountering this problem should call 800 BE SHARP [800-237-4277].)

The Sony VPL-VW50 1080p SXRD front projector, using the LCD variant known as Liquid Crystal on Silicon, also had uniformity problems:

Picture uniformity was poor, with pink and green tinting visible on both gray full-field patterns and program material, although the amounts varied on my two test samples.


... the top and bottom edges of the frame had a distinct pink tint and the center looked comparatively green — a problem that was even more apparent on test patterns ... I saw this same issue on the black-and-white stills shown in a PBS documentary ... although it was harder to detect on full-color programs. A second sample from Sony showed a reduced level of pink discoloration, and no greenish tint.

Eventually, Sony blamed this on sample variation that falls within range of its manufacturing tolerances, and said "an experienced technician" can enter the service mode and adjust for uniformity using the projector's multipoint gamma control. I was able to remedy the problem using this adjustment, and chances are you can hire an ISF technician to do the same. However, this calibration would not be covered under warranty.

There are, as I say, many types of mura. And they are hard to measure objectively.

This PDF gives some technical insight into the phenomenon, as do this one and this one.

The first PDF shows pictures of the different mura types on page 2. The thrust of the article is: what kind of algorithm can be used to evaluate muras? Among just "cluster muras," which the "clouds" complained of here in this thread represent, there are "round-type" and "line-type." How can the algorithm evaluate both?

The other two PDFs also make it clear that automated mura detection and measurement is not easy — yet the human eye picks picture nonuniformity up in a flash.

These PDFs are aimed at improving quality control — weeding out bad LCD panels before the customer sees them. The panels made for our problem Sonys as well as for a number of other LCD TV brands weren't duly weeded out, obviously.

Muras aren't binary, on-off, yes-no things like dead pixels. A pixel is either dead or it isn't. But a "cloud" can be a puffy cumulus or a faint haze. Maybe if the participants in this thread turn them all into thunderclouds, Sony will get struck by lightning and change its QC ways!

It begins to look to me as if whatever QC worked for lesser LCD panels is failing with the newer, larger, 1080p panels.

Here's an analogy: In the world of DLP-based 1080p rear projectors, there are two kinds. One uses a 1,920 x 1,080-pixel chip, the other a 960 x 1,080-pixel chip. The latter uses "wobulation" — a swiveling mirror in the light path — to double the effective number of horizontal pixels.

The chips are "digital micromirror devices" or DMDs, not computer chips with transistors, but the principle is the same. The more micromirrors or transistors on a chip of a given size, the denser the chip. The denser the chip, the lower the yield of good chips in the manufacturing process. The lower the yield, the more chips have to be thrown away — and the greater the unit cost of good chips, since the cost of making the bad chips gets piggybacked onto the good ones.

Using the lower-density DMDs along with wobulation holds the line on the price of the TV. True 1080p DLPs cost a lot more.

It stands to reason that the same logic applies to display panels. The denser the panel — and 1080p panels are more than twice as dense as 720p panels of equal size — the lower the yield of good panels, whose manufacturing cost accordingly becomes a huge issue. But there's no workaround such as wobulation — what you see is what you get. It appears to me that Sony is lowering standards on what constitutes a "good" panel in order to keep overall costs down and remain competitive.


Anonymous said...

Thank you very much for your great blog and interesting comments. I tried to download the pdf from http://ltswww.epfl.ch/ltsftp/EI2006/DATA/6070_17.PDF but it's not available any longer. Can you upload the pdf or e-mail it to 2006@karpfenteich.net. Thank you very much, Peter

Anonymous said...


I am an attorney based in Los Angeles and I specialize in class action litigation. My office is investigating the “cloud” defect in certain Sony LCD televisions in connection with a lawsuit that my office may file against Sony.

I am interested in speaking with you concerning the cloud defect that you discuss on your blog and getting copies of the pdf files you mentioned. My contact information is as follows:

Daniel E. Sobelsohn, Esq.
The Sobelsohn Law Firm
1901 Avenue of the Stars
2nd Floor
Los Angeles, CA 90067

direct dial: (310) 461-1333
facsimile: (310) 861-5205
email: dsobelsohn@sobelsohnlaw.com

Thank you for your assistance and I look forward to hearing from you.