Limitations of PerfectColor On the Mitsubishi xxx13 RPTV Explained
Let me start by stating that I love Mitsubishi RPTVs. I feel they provide the most options and best picture of any television on the market today. It is not my intent to slam Mitsubishi with this discussion. I whole heartedly recommend Mitsubishi televisions to anyone. It is however my intention to explore and document the limitations imposed by the current color decoder system incorporated in the xxx13 Mitsubishi lines. This discussion will likely be applicable to the new Mitsubishi xxx15 models as well.
What follows is a simplified description of a complex situation. As few words as possible will be used to explain the concepts involved. As a result, some details will be left out. However, there are quite a few photos and as they say, “a picture is worth 1,000 words”. If you read this paper you will likely gain a great deal of knowledge about how video is rendered on an analog television and how the Mitsubishi PerfectColor system works.
Explanation of Problem
The usual way to determine how well a color decoder is working is to display a color bars test pattern (Figure 1) and either measure the resultant colors by eye or with some kind of measuring device. There are two primary devices in the industrial video calibration world that are used to make these measurements. They are the waveform monitor and the vectorscope. Both of these instruments are specialized oscilloscopes. An oscilloscope is simply a device that measures voltage with respect to time. Thus it stands to reason that if one knows how to properly set up and interpret an oscilloscope display, then one can actually opt to use an oscilloscope instead of either of these specialized instruments.
All measurements in this discussion were taken with an oscilloscope. Let me begin by explaining what it is you are looking at when you see the photos of the oscilloscope screen. If, while your TV is displaying the color bars test pattern you were to turn color saturation all the way down to zero, what you would see on the TV screen would be a black and white picture of the color bars. Notice that there are seven bars in the color bars test pattern. They are presented in the following order, gray, yellow, cyan, green, magenta, red, and blue. Figure 2 is a photo of an oscilloscope screen connected to the green CRT displaying the color bars test pattern with the color saturation set to zero. I have chosen to use this photo because it easily explains how to interpret the oscilloscope. Notice, that on the oscilloscope there are also seven bars. The lowest bar (the one at the bottom of the screen) is the gray bar and the highest bar is the blue bar. I have also added a color “dot” under the corresponding bar on the oscilloscope display to more easily illustrate the concept.
On a typical CRT RPTV you have three CRT’s, a red, a green, and a blue. On a Mitsubishi, to measure the color output the oscilloscope is connected to each gun at test point 6 (TP6) of the CRT-PCB. Just so you're aware, this is the same test point that is used to adjust the VR cut-offs for gray scale.
Each of the three guns contributes to four of the seven bars in the color bars test pattern. The red gun is used to make gray, yellow, magenta, and red. The green gun contributes to gray, yellow, cyan, and green. The blue gun is used to make gray, cyan, magenta, and of course blue. Note that all three guns contribute to create gray. Thus, one can then use gray as a reference to determine if the colors are balanced and linear.
At this point, I would like to make sure everyone has some idea as to what tint actually is. Tint is also referred to as hue but from a technical standpoint it should be called phase. In very simple terms, this is because tint is derived from a phase difference between a reference signal. This discussion always uses “phase” rather than “tint”. Also, “color” is always called “saturation”.
If a color decoder is balanced and linear, when blue is adjusted for both phase and saturation, all of the colors will align perfectly. However, it is a well known fact that Mitsubishi incorporates quite a bit of red-push into their color decoders. For this reason simply aligning blue is inadequate because even when blue is aligned, red will still be over saturated. On earlier models red-push could be properly compensated for because those TVs had the capability to change both phase and saturation for the three primary colors independently. In other words, both tint and color could be adjusted for the red gun while leaving the other two CRTs unaffected. Incidentally, if the three primary colors are aligned properly for phase and saturation yellow, cyan, and magenta should fall into place and need no further alignment. Unfortunately, on the xxx13 and xxx15 models this is not the case. Mitsubishi has changed the color decoder used and thus the options available. While Mitsubishi has added the ability to independently adjust saturation for all six colors through PerfectColor, there no longer exists independent phase control for each color. Now, instead of being able to change phase on just the red gun, you are forced to use the master phase (tint) resulting in a change of phase for all colors at the same time. So if blue and green phase is correct but red is not, the only way to correct red is to throw green and blue out of alignment.
All testing was conducted with color and tint set to midlevel while “color balance” was set to manual in the user menu. Good form dictates that to begin the experiment the television’s blue should be set to industry standards. This means that tint and color should be adjusted to properly achieve balanced blue.
When displaying a color bars test pattern, a properly adjusted CRT will have equal output in the four color bars it composes in the test pattern. Many of you are already familiar with the AVIA flashing bars test patterns. Essentially, what I have stated above is simply that these test patterns are displaying correctly and that this carries over to all levels of color saturation. Thus, if you set the blue bars test pattern right, when you look at the standard color bars test pattern above with a blue filter or with the other two CRTs muted, all four bars in the color bars test pattern will have the same color (voltage). If you hook an oscilloscope up to the blue PCB-CRT this is what you will see (Figure 3).
Notice again that the oscilloscope displays seven bars. This time however, instead of having saturation set at zero, it is set to perfectly balance blue (circled in yellow). You know blue is balanced correctly because the gray voltage displayed is equal to that of cyan, magenta, and blue.
Now that blue is set, green should be checked. Figure 4 is the oscilloscope display for the green CRT after blue was balanced. Again notice the four bars associated with the output of the green CRT (circled in yellow). While the voltages associated with the green gun’s color are not perfectly aligned (there is slightly too little yellow and not quite enough cyan) it is still pretty close. In fact it is close enough.
So, all that remains is to examine the red CRT. Remember now that both blue and green are acceptable at the current settings for saturation and phase. Figure 5 is a photo of the oscilloscope display while hooked to the red PCB-CRT after blue was balanced.
Notice that the voltages associated with the colors created by the red gun do not equal those of the gray bar. There is too much yellow and magenta as well as far too much red. Figure five also quantitatively illustrates the red-push. It is indicated by the difference in voltage between the gray and red bars. To compensate for the red CRT misalignment PerfectColor must be utilized to make adjusts to red, yellow, and magenta. Figure 6 is a photo of the oscilloscope on the red CRT after these changes were made. Now (circled in yellow) the voltages associated with the colors composed by the red CRT are equal and correct.
So at first it would seem that everything is in good shape now. Unfortunately, the yellow that was just reduced in PerfectColor to balance red has resulted in the green CRT having too little yellow (Figure 7). Figure 7 is a photo of the oscilloscope while hooked to the green CRT after red calibration. Notice that the voltage of the yellow bar no longer is equal to the gray bar. Thus the green CRT is not contributing enough to the yellow bar.
So now it is known that green is inaccurate as a result of the adjustments to red, but what about the blue gun. Magenta is shared by the red and blue CRTs, so as expected when the oscilloscope is hooked to the blue CRT-PCB the result of this is apparent (Figure 8). Magenta was reduced to compensate the red CRT resulting in the blue CRT not contributing enough to magenta.
Clearly green and blue could be corrected in PerfectColor to make them balanced again, but I think you probably already know what will happen. Of course when you correct green and blue they will again become aligned, but this will result in red ending up exactly where it started. In the red CRT, you will again have too much yellow and magenta. So at this point you are stuck. No matter what you do inside PerfectColor somewhere the color decoder will not be balanced.
What is essential to correctly realign a Mitsubishi with red push is the ability to change the phase of the three primary colors independently. If it were possible to adjust the phase angles of each CRT separately from one another the color decoder could be put into perfect alignment for all three guns simultaneously.
So what’s your best bet? To set up these sets the first thing to do is
return your color and tint sliders to the midpoint in the simple user menu.
Then reset PerfectColor. Display a blue bars test pattern with AVIA or a color
bars test pattern. Go into the service menu and adjust
so that the blue hue and saturation are correct. This now means that the blue phase and saturation are at least in the right places. Now exit the service menu. Go into perfect color. The blue CRT should still be balanced. This means you do not need to adjust cyan or blue in PerfectColor. Now put up the red flashing bars test pattern or color bars test pattern. Use the red slider in PerfectColor to adjust red saturation and magenta to adjust red hue. Now put up the green flashing bars test pattern. Adjust the green slider to level out green saturation and yellow to adjust green hue. Finished. The resultant image will no longer have red-push but unfortunately, as stated above, your color decoder is still not balanced.
Last Revision: 11/5/04
Below is a picture illustrating the correct phase relationship between the colors. I have included it because some of you may better understand phase with this illustration. This is an NTSC diagram but conceptually it stands. The angles indicated are with reference to the axis labeled “burst”. This is also very similar to what the screen of a vectorscope looks like with a balanced color decoder.
I hope you gained some knowledge by reading this paper and I look forward to your comments.
Legal: This document is the property of craigr, all rights reserved, and is being shared with the The Home Theater SPot. As such, it is unlawful to reproduce, in part or in full, distribute, or link to this document directly, without the express written consent of the author. Do not quote this document without citing the original thread on which this discussion can be found in the Mitsubishi Tweaks section of The Home Theater SPot. Please direct all questions and answers to the Mits Tweaks in Progress section of the SPot regarding the preceding discussion.
Thanks to The Home Theater SPot for providing a means for us to discuss such topics...
Please click here to support the SPot!