As many of you know, I have been messing with the calibration of my 82837 since I bought it. One issue that has driven me nuts since day one I have not discussed on the forum because I was not sure if it was unique to my set, my meter, my environment, or exactly what was causing it.
What I was seeing was that you could calibrate to a high degree of accuracy but within some random amount of time, sometimes within an hour, sometimes a day, the grayscale would drift, or oscillate. I thought it was my meter, as it would seem to have some correlation to the room temperature. I have two hvac vents that blow pretty much straight down on the set. So I would try to calibrate between AC runs, that seemed to help but sometimes the brightness would drop and green levels would shift no matter what I would do.
I had the meter calibrated, tried this and that, all the while I am going through one new beta after another of the DVDO Duo firmware which while continously improving and is now an extraordinary device for the money, always added another wrinkle. My Pioneer player levels did not match the TV set and caused clipping at a different contrast level then Duo. Small issue after issue clouding what was gong on with the set and its tendency to have grayscale and brightness drift. Anyway I have calibrated my set many times and have come to know its quirks pretty well.
It is a good thing I never paid a pro to work my set. He or she might have come in, did a good job, charged me $500 and then went back to their home state only for my TV to go way off a few days later.
I finally decided that it might be something in the thermal management of the set. I run the set with the lamp on bright. The 2009 82" sets are not light cannons, so they need to be cranked up. The lamp temperature might be getting too high and the logic board was dropping the voltage on the lamp perhaps by design or the spectral characteristics of the bulb change at a certain temperature. I believe it is the former. So either my set has faulty thermal management of the lamp as in a bad chip or sensor, or the Mits firmware just takes its toys and goes home if you run the set too bright.
I bought one of these free standing ThermalTake PC USB fans. A lot of people use them to cool AV equipment. It turned out to be an almost exact height match to the fresh air intake grill for the lamp. It was about 20 bucks with overnight shipping from Amazon. Now it is not completely silent so that might be an issue for some. It does have a speed control, I have been running it on full. I can hear it when I am behind the 82837 but my ambient envrionment is rather noisy so I can't hear it from the couch.
The USB fan blowing into the set at the lamp air intake seems to have completely solved the grayscale drift I was seeing before. There is still some tiny variations in peak brightness, but that is too be expected. dEs are staying put, and grayscale tracking is staying dead on.
I use to hate to get to fussy with the adjustments as it was a waste of time, they were going to shift anyway. Now it is a very stead ship. I will finish a fine tune on grayscale and gamut in the next few days that show some very low error numbers. Credit for the low color errors will have to be given to the DVDO Video Processor, the Mits can't adjust that close on its own, but I never bothered to really work it as tightly as possible because I new that the it would drift anway, and on top of that there was always going to another firmware release from DVDO for the Duo coming in a week or so. The Duo will continue to get new firmware, but that is all for the good. The set is stable and so you are not try to hit a constantly moving target.
Clipping is still a problem with the set, some of that is a remaining bug in the Duo and blue is the one color that I still can not get the dE to go much if any below 2. The other colors all have delta errors on CIE 94 of 1.1 or less. Some people say a dE of 3 or less on color is where you want to be. Tom Huffman who is the developer of ChromaPure says the dE needs to be 1.5 or less to be undetectable.
Just another blank signature.