Mitsubishi: DEFECT in 2011 DLP Sets/fix - Page 2

That is very interesting.
This indicates to me that it must be a specific processing algorithm rather than any physical part defect.
Note that it applies only to the specific shade range of red, but not derivative colors of red (like purple or orange).

That implies that their sharpening algorithm fails in some specific range... but I wonder why the reds.
An interesting test image would be a red-shade gradient with text/lines so we can see exactly where in the gradient the failure starts and stops.
It sort of feels like we're doing Mitsubishi's job for them, though... if it helps get a fix I'm up for it!

Well, sorry for not posting, I have been away. To add my opinion to the thread, obviously I too have the same defect in my 92840 DLP (obvious to any who were following the 2011 Mitsubishi 740/840 thread).

While there is a "workaround" I feel as though this has a significant enough impact on the PQ of the TV to be considered moot;

* The workaround does significantly reduce light output, enough so to make it painfully obvious during day time viewing.

* The workaround adds a slight flicker to the picture, similar to viewing an old CRT monitor.

* Due likely to the lowered brightness, the picture loses some of the "pop" it had in 2D mode, this includes a loss of vividness in relation to colors.

* The picture tends to have a visibly pixelated look to it if viewing too close (such as within 6 feet). While hard to explain, it has to do with the checkerboard pattern itself.

* For those that care (and while myself and others here do not, there are some that I am sure will) geometry correction CANNOT be used while using the workaround, it will blur the picture.

* The last problem this workaround introduces is what some seem to call a visibly softer image being introduced when using the workaround. Unless you have a video processor, this picture cannot be sharpened to eliminate the soft look introduced as the internal sharpness controls are no functional while in 3D mode.


While this workaround is enough to please some, it certainly cannot and should not be considered a fix, it is not. I state this and the points above sternly as we cannot allow Mitsubishi to see this as a reason to not fix the problem. This workaround is not enough to fix the problem and most of the time I do not even use it as it just degrades the PQ in other ways IMHO.


I also wanted to add I have indeed contacted Mitsubishi regarding this problem as well, and plan to continue contacting them until they send a tech out. It took 5 calls to resolve the first problem I had with this TV, and I am sure it will take no fewer to get them to send a tech out yet again...

Lastly I should state, when I have time I will be updating my Amazon review of the TV with my disappointment with this issue. While I won't give the TV 1 star, I will certainly revoke the 5 stars I gave it previously.

The wheel spins at a fixed speed.
Brightness it controlled by the oscillation speed on the DMD.
Brighter? Spend more time on. Dimmer? Spend more time off.

All of these theories are quite interesting, and one of my most revised theories was certainly along the same lines as resent posts, i began to believe the issue was caused by a timing error between the color wheel and the DMD. However, i recently did some testing that disproved that theory, this involved shutting individual colors off through the color control settings; for example, using the service menu to show blue only. The defect was still present, even though red was noblonger being displayed.

For this reason i began to rework my thought process, and most recently began to realize the problem shows itself as long as the input signal is red on white, should you change the hue (such as using a computer) the problem will phase in and out as the hue changes, however when using the built in controls of the TV to change the colors the problem is ever persistant. Unfortunately i have had a bad day today and as such have found myself a little under the influence... I do wish to further test this concept, eli.inating all color from the TV controls, but for some reason i seem to be feeling quite apathetic, can't think of why.... but so far what i seem to have observed is that fhe problem cannot be related to a timing error between the DMD chip and color wheel, though this was certainly the cause i wanted to be accurate... (as it would have been the easiest to fix, a simple firmware update)

Now on the high note, this points to the problem being pre-processing related, which means again fixable through a firmware update. I think i had some other thoughts to back this up, but they seem to be slipping my mind. I will admit this is not one of my prouder moments, and i am not sure why i am even posting in this condition... but at least it will remind me of where i was going with this tomorrow. I woould really like to narrow this problem down to a specific function and cause, nothing would make me happier than to present exactly what needs to be corrected to Mitsubishi.

Hi thhowl

Thanks very much for the comprehensive answer to my request. I did find the 2010 thread and read numerous pages. But your analysis of the 2011 defect is impressive and I value your well informed analysis.This was the statement which made an impression on me :

"By googling just now, I be confident I am not missing to warn you: no downward spiral started after my 2009 model, no serious PQ degrading issue was introduced in that specific 2010 model, etc. (Specific 2010 model, in contast to ALL 2011 models."

I plan to purchase the 2010 model and I am confident I will be happy with it. If not, at least I received good advice from you. Thank you

The quote from your second call mirrors what the first Authorized Tech guy said to me, while he was fumbling for excuses. They obviously teach or send out memos of their key talking points, hoping to shut people up. Even the managers get the memo, apparently.

His other gems were "Look, I'm no genius" and "Mitsubishi has bad reputation for picture" (English isn't his first language).

I'm scheduled for another part swap now, but the item is on back-order for a few weeks.

Random note about the bug:

I was downloading some DLC from XBox today, and was messing around with 3D Checkerboard vs Non 3D mode to kill the time.... There was a very obvious halo (like a sharpening-effect) around the download bar, even with sharpness at 0 on non-3D mode. In 3D mode, that halo was gone.

Howl and I talked about this waaaaay back.... but I had forgotten about it with all the other discussion of the bug. It's actually comparable to the red/green pictures he posted just a few replies above this one.

I don't have much if anythng new to add. I got my meter back and did a full calibration in 2D mode, except for the defect it looks great. I previewed the same calibration in 3D mode and the loss of dynamic range is severe as is a noticable color balance problem. Over the weekend I will see how much if any that can be restored with a checkboard specific calibration.

Realistically, I think you can only perhaps get the color balance right. One the 92840 the loss of dynamic range is likely a non-starter. I will give it up here bythe begining of the week. The furnace is running too much at night and although my meter supposedly has temperature compensation I don't find that works well with the furnace cycling.

I watched tonight with the TV in checkerboard with a my new fresh normal calibration and it was just gloomy if not dismal. There are only so many clicks of contrast to work with and you hit a color clipping wall. What is surpsing is that you clip just as fast in Checkerboard as in normal mode but I am guessing you are down at least 10 ft lamberts at 100 percent stimulus.. An 82840 or any of the 73" sets should make more than enough light to make up for this. I am seeing 33.5 ft lamberts in Normal 2d mode with this calibration...... yes there are a few lclicks of contrast left but you will just drive color clipping below 235.

We'll see. I hate the horrible teeth. I watch a lot of news and interviews. I don't watch much Fox news but they have one of the best color balanced broadcasts on cable.. Dick Morris was on, a former Clinton pollster who turned Republican stategist and pundit. The skin tones were great, perfectly saturated color. Everything looked great but his teeth looked to be rotting out of his head. It was nauseating and to know it was mostly a video processing error makes you angry.

I am still a week or two away from my probably Quixotic battle with Mitsubishl.

I will report back with measurements on the difference in Max White between Checkerboard and 2D mode.

I am not sure I understand why the brightness goes down in checkerboard.

That is my speculation.. but it seems like this would have been more carefully thought out. You would not want not a brightness decrease in 3d mode as the glasses already hit the overall brightness hard. In nomal mode the 92" is significantly brighter than my 2009 82".. how were these things really accomplished?

Red is the longest wavelength, maybe they just run out of time switching the mirror? I don't know just speculating. If if it just a color processing error than that might explain why 3d modes dont exhibit the defect as much less color processing is done in 3d.

I was in the group believing timing error as well, however, go ahead and try turning on blue-only in the service menu while the problem is present. The problem stays persistent despite eliminating the red and green channels. This is quite odd to me, but definitely points more towards processing error than timing error.

Along the same lines, I also noticed the fact that 2D mode seems to present a slight sharpening halo around certain items such as text, despite having both sharpening at 0 and edge enhancement off. I am now in the belief that there is additional processing being done we both don't know about and cannot adjust in 2D mode, while 3D mode disables this additional processing.

A second reason I believe this to be the case is a timing error with the mirrors and colorwheel couldn't possibly affect as great of an area as it sometimes does when the defect is present. There are times when the red jumbles with the white that is over 10 pixels away, if it was a timing error the defect would only stretch 2 or so pixels you would think. Or on the reverse, the bug would be persistent regardless of where white and red is on the screen (certain shades of red would just show the defect regardless of what is around it).

I would very much like to get a hold of a full service manual for the TV to see what service menu options exist we don't know about. As far as that goes, I might just play around in the service menu and hit random buttons until things happen.... What is most likely the case is that there is a way to shut each processing feature off, but require different menu codes (such as there is a menu code to reset the TV, reset the remote, etc. all different from the service menu code)

light travels at c, regardless of wavelength.
The only issue would be slight color convergence through the optical engine... but this is clearly not the case because of how the checkerboard fixes it, and checkerboard has nothing to do with light traveling through stationary lenses.
The aberration probably causes the slight halos though, but certainly not the red-caused blurring.

Correct, I suppose I didn't specify this well enough in my prior post, one of my earlier posts mentioned passing blue only pre-processing by using my computer took the defect away. I then specified I would do my testing by eliminating the color post-processing using the TV controls themselves as I believed the problem was still persistent regardless of altering the color in the TV itself. This was all to debunk the concept of a timing error between the DMD and colorwheel, and show it is indeed an error creating in processing. Thanks for clarifying. ;-)

I have not really looked into the Service Menu on the new model other than to turn off Geometry correction but the older boxes had the color wheel index adjustment right out where it was easily adjusted. If it is still there one could raise or lower the index a bit and see if the defect changed at all. One of things about DLP and color wheels is the tendency toward banding.

New firmware came out today!

At least for the WD-73640 that is.

I just updated my WD-73840 with the new 13.06 firmware. I turned off 3d checkboard mode and retested one of the more popular jpeg's that had been used to illustrate the issue, also tested a red line test image that someone created on a previous post, and lastly tested my iron man blu-ray disc pause/timeline at the bottom. All three render without the defects that I'm use to seeing.

I'll leave it to some of the other more knowledgeable posters and calibrators as to whether or not there are any adverse effects introduced by this change.

I can confirm the problem is gone on my set with the new update without the weird out of focus text with the 3D Checkerboard work around. Good job thhowl and everyone else who contacted Mits about this. Now if I can get used to the occasional rainbow...

That's great to hear. I'll check and see if my set has updated this evening. Thanks for the heads up!

I can confirm this update works as well!!! I've only tested my PS3's main page however, but the problem appears solved!

I had logged into AVS to tell you guys that the Service Tech came by today to swap out the Optical Block, but the problem persisted (as we all assumed it would). But this tech was very helpful, he took pics to email Mits, called field support and told them that the problem persisted and that it was unacceptable (even at default sharpness of 31, to boot), and needed them to call me to figure out the next step. So he was being very helpful.

But I had a PM from Howl waiting for me telling me about the Firmware update, so I tested it before posting. This is great. Hope it works for the bigger screen models as well. I'll do some more test scenes tonight just to be safe.

Good job guys --- we came together and made Mits aware of a problem.... and to their credit, they fixed it. Although, points deducted for being rude to Howl over the phone, and otherwise blowing off all other attempts at contact from everyone else.


Update:

I've tested the PS3 main screen and Transformers 2 BD for the most obvious forms of the defect, and both were defect-free! Then I tested Freddy Vs Jason BD, where in the very beginning, there's a close-up of Freddy (sans burn-makeup) creepily licking a photo of a victim. This shot had previously shown the defect in a different way.... Freddy's mouth/lips created actual picture info outside the widescreen matte bars (the black lines on the top/bottom of some movies). You could actually see many pixels of color info in the black zone of the wide bars (not like a soft color/light bleeding type thing... actual pixel info). That is gone now too! So far, so good!

Gentlemen,
I am happy to report that I have downloaded and installed the new firmware. While I had very mild symptoms of the reported issues, I did have them. It would appear they have ALL been fixed with this update. Many thanks to this forum for the hard work in getting Mits attention to the problems.

I do have one question...how did Mitsubishi fix a problem they would never acknowledge they had????? Things that make you go.."hhhhmmmmmmm"

:-)