THE SONY SERVICE CODES - Articles, Comments, Discoveries - Page 9

A really good point. My method assumes *no* input is perfect. But if Sony have guaranteed that the derivation of color from the combined "C" signal is near-perfect (quite possible), then calibrating from that input as your first choice would work well. Yes, maybe. I'm still not sure I understand all the implications of the published block diagram from the service manual, which tries to show all of the signal paths. Those who think this is a mostly-digital set will be disappointed! It appears like the ATSC tuner, memory-stick, and HDMI inputs are immediately converted to analog-component video. All of their signal paths are identified as "YCbCr." Inputs 1, 2, and 3 ("YC") are routed to the "Main Color Decoder, CXA2103." Then they exit as YCbCr -- aka component video. Then that is converted to digital at IC3003 and passed to the DRC processor and on to the MID processor.

The CXA2171 chip switches among V5/V6 component inputs, HDMI, and ATSC, a different path from the CXA2103 chip, above. Then that is converted to digital at IC8601 and passed to the main MID processor as "VDO digital," bypassing DRC.

I made the initial choice that I couldn't necessarily understand all of the implications of this diagram, and I have found it useful to model the behavior of the TV in my head, refining it as I learn more. That's what engineers do on the way to designing something, so it is also a helpful skill when trying to understand a complex system that already exists.

The referenced block diagram is attached.

 

XS955|XBR960 Block Diagram.pdf 139.0234375k . file

Here is a good explanation. See "Component Video Description and History" about 1/3 of the way down. I shoulda read this before . . .

http://members.aol.com/ajaynejr/compon.htm

The article says that YCbCr is the digital equivalent of YPbPr. But the block diagram clearly shows *analog* component inputs V5/V6 feeding YCbCr to the next block. Phooey! Another error in the Sony documentation. Either they are using both terms randomly, to mean the same thing, or they have them turned around.

Yesterday I spent about an hour with the schematic diagrams of the set, and it doea, indeed, look like that is what is happening. The block diagram has numerous errors. (The component-input signals are correctly labeled YPrPb on the schematic, and YCrCb doesn't appear until after the A/D converter. Then things get fairly compex with labels and symbols I couldn't identify, so I gave up.)

It's important to know that there are two main video pathways, one with DRC and a separate one that bypasses it. Well, important for techies, anyway.

I don't think you meant this. The former is white balance, the second is grayscale calibration independent of white balance. You could, for example, have a perfectly linear grayscale calibrated for 5400K.
I think you have stated the issue pretty well. Glen also has a variety of TVs to deal with, and, accepting the inherent asumption in his method, it will succeed.
This is where our methods converge. We *both* have to calibrate CBOF and CROF for the inputs we want calibrated, by whatever means.
No, not grayscale color-neutrality. Just black level (which I included with the term "offsets") with UBOF, and then color (amount) and hue differences between video source devices with UCOF and UHOF.
Or let's just say the color and black level on each input signal, because there is nowhere else to compensate for, say, a permanently-connected VCR with a noticeable hue error, as I own, or for a DVD player whose black level doesn't agree with anything else (mine, too). I would use those settings as compensation so I don't have to fiddle with the remote each time I switch from broadcast to a VCR recording of the same thing. That said, you can make the choice to have UBOF-UHOF set just for imbalances inherent to the TV, and leave further tweaking to the user menu. What else could you do if you were going to switch video sources around every now and then? But in my case, there is no way for me to separate the TV from the DVD input in practice, so I adjust for my DVD player as-connected. When I switch to it, it's right.

"White balance" on my digital still and video cameras means *color-temperature* adjustment, and that's how I used the term. It also has the same meaning in Photoshop and its Camera Raw plug, where you use the White Balance slider to "balance" your image to match the color temperature of the picture as shot.

But let me use your terms: Color-temp (white) setting and grayscale linearity (uniformity) are independent, as you have said in your last sentence above. You can set any CT for any grayscale and adjust grayscale uniformity for any CT. That means the parameters of white balance and grayscale uniformity are independent concepts in principle. We agree; this is semantics! (Of course, the *adjustments* we're stuck with for setting WB and grayscale interact -- but that's just an implementation, not a principle.)

ADU's thinking that he has to have 6500K before he has adequate shadow detail confuses the two concepts of the *color* of the white point and the *level* of the darkest grays. To be sure, I am including the black level at the bottom of the gray scale as part of its setup, and that is the one item that determines adequate shadow detail. So maybe the difficulty is there. Again, the terminology . . .

I have no experience with any kind of problem like this. I guess I need to know more by what you mean by "blurred." Does it look like diagonal lines get very jaggy? Like the number of vertical lines is cut in half? (Horizontal detail stays about the same.) Or is horizontal detail affected, too? A soft & mushy blur, like a camera blur? Are the TV's scan lines getting blurred at all, or just the image?

The "character" of this blur might indicate the cause. Or digital before-and-after photos.

Question: What is the purpose of this Jerrold box? Can't you feed the cable directly? Or is this a digital-cable converter? (You said "old," so I wasn't thinking digital-caable.)

These concepts are easy to confuse. No, red "push" is not related to white balance at all. (You likely already understand the first part of the explanation, but I'll state it anyway for others' benefit.)

Setting a perfect white balance and grayscale can be accomplished in the *absence* of any color at all. Just as with TVs of 20-40 years ago, the xCUT controls determine when the screen starts to glow as the video-signal voltage increases from its minimum, on its way from black to white. Ideally, all of the color guns start to glow at the same time, and you get neutral dark gray. But this isn't how CRTs generally behave, and it is tuned by using different settings of R-, G-, and BCUT.

As the video signal reaches the value for "white," ideally all three color guns contribute the right amounts to produce a pure white, which is defined in the video-production world as 6500K -- roughly the color of front-lighted white clouds at mid-day. Whereas the xCUT controls determine at what point the color guns *begin* illuminating the screen, the xDRV controls determine how much "drive" each gun gets as the video goes from "black" to white. You would adjust R-, G-, and BDRV to get that perfect white, using a known white pattern or a high-contrast TV picture with color turned all the way down.

The xCUT and xDRV adjustments interact. (To an engineer they are classic offset-and-gain adjustments.) A 5-point change in RCUT makes a big change in color of dark gray, but very little change in the color of white because those 5 points are now only a small percentage of the total white. Likewise, the xDRV controls slightly affect the color of dark gray. Obviously one goes back and forth between the two control groups until everything looks fine. And generally, if you get the darkest grays and white right, the grays in the middle are fine, too.

Now that you have a perfectly-calibrated grayscale and white point, we introduce color. But here's where "push" comes in: It is the *favoring* of certain colors by the process that extracts the color from the TV signal, whatever it is. Imagine a picture of three kids with red, blue, and green tee-shirts of equal intensity in the real world, standing on a gray concrete driveway. Some TVs will show the red shirt as much more intense than the others, as with all other reds in the picture. (White and grays, by the way, and that driveway are still perfectly neutral.) This is "red push" -- a *favoring* of red in the color-decoding process. Whatever the historical arguments for this regarding NTSC broadcast deficiencies, color push is now considered a no-no, and *accurate* color decoding is preferred. (Matches reality, no?)

Sony's late-model computer-driven TVs have exquisite controls for all of this in their service-mode codes: Grayscale and white can be finely tuned (basic controls: xCUT and xDRV), and color-push can be reduced to virtually zero with 2170P-4/RYR thru GYB plus a good test pattern from, say, the DVE DVD. Plus, the TV will store two different sets of values for RYR-GYB, for the "Default" and "Monitor" user-menu choices.


Different aspects of how the video is processed by the TV and then displayed, yes.
Nope, they're independent.
If you think of TV color-push as analogous to a specific color film's "personality," you may come close. When you worked with color CRTs for photofinishing, color push would have been an unacceptable defect, I would think, as it is with the color-prepress monitors I have set up. The "push" is created by the TV for an "esthetic" purpose.

Unfortunately, TV manufacturers keep throwing these color "improvements" at us, and we have to keep getting rid of them. Lucky it is that, with these Sony sets, it's a relative breeze. With some older sets, there is no way to alter the color decoding.

My experience, and that of several other posters here, has been that the following settings generally kill all color push for the XBR960 and recent XS955 CRT TVs:

2170P-4 / RYR-RYB-GYR-GYB = 13-15-5-3. (Then WRITE the settings.)

But grayscale and color depend heavily on your particular CRT and its characteristics.

When I watch Lost, CSI/Miami, or any other dark HD program, my choice has been to invoke a lower *gamma* that raises the midtone brightness while leaving black and white alone. White is still dazzling, but the blacks now have increased detail, and the mid-tones appear more natural/realistic.

There is no instant built-in control for this in the user menu. On my TV, I have set up the Pro and Movie modes to be *exactly* the same. (See my articles #04 and 05, posts #5 and 6, in this thread.) Except that (1) I have set GAMM = 1 in Movie mode and (2) set GAMR, GAMG, and GAMB each to 3 and GAMS to 0. Now when I switch from Pro (GAMM = 0 with all GAMx at 0) to Movie mode, the gamma is reduced from about 2.45 to 2.2, and that seems much more pleasing for a lot of HD material. But if the picture seems a bit flat for some broadcasts, I make sure it's in Pro mode with its higher gamma. Point is: Now it's *my* choice!

No, you aren't. You can mess up proportions by adjusting either vertical or horizontal overscan independently. If you use a test pattern, however, that has circular patterns in it, there's little chance of that: You'll see it immediately in the circle. You can even measure the circle to make sure you're close.

I knew something was funny with my set out of the box when the round PBS logo started looking squat in HD. Easy to fix. Just keep the circles circular, and adjust for minimum overscan that hides the defects at the raster's edges. I've ended up at about 4% for 4:3 full-screen material.

Then you want to adjust those blanking "shutters" that are supposed keep the electron beam from striking the sides of the tube: 2170D-3/R-, L-, T-, and BBLK. Procedure: reduce shutters so you can see the true, messy edge of the raster, and adjust the overscan for a clean display, as above. Then increase the shutters so you can just see them, then retract them a bit so they're not visible.

Very likely. Recycling power restores previous settings because they are re-read from memory at power-up. Replugging the power cord as well guarantees it. (Yes, there may be a couple of settings -- I know of one in AUDIO -- that survive the Power button, but not unplugging the set.)
No, they're cleared. In fact, if you do nothing for a few seconds after making a goof, the screen will indicate "SERVICE" again, and you are again NOT in WRITE or INI- mode. Note that the most disastrous mistakes take more than two keystrokes (in that few-seconds period), but thet WRITE takes only two.

However -- if you have made some experimental choices in a code or several that are global settings, and you have gone away from them and become distracted, those settings might be written inadvertently when you do your next intended WRITE. If you are ever in doubt, recycle power, going back into service mode, which guarantees you are starting over "safe." I can't tell you how many times I have done this. I tinker and tinker, then have some kind of epiphany about a certain setting and want to WRITE it. I have learned to stop, make a note of what I want to do, then recycle power, go back to those settings, change them, *then* WRITE. Better safe . . .

Let's assume it is not the program material. (PBS-HD material frequently has green/yellow-tinted blacks. Drives me nuts! I have complained.) You can see it across different inputs, right, and when the Color slider is all the way down?

The settings that most affect "black" and dark grays are the xCUT codes, but they contribute a little to the color of white, as well. You can try decreasing 2170P-1/GCUT until dark grays match whites in color. Sometimes it involves yellow, too, and raising BCUT a bit may help. Note that the average xCUT settings affect black level, and you may need to readjust SBRT a little to compensate. Try to keep at least one of the xCUT settings close to its previous value. *WRITE DOWN* your changes before and as you make them!

The goal, as you adjust these things without a color-temperature meter, is (1) to get the color of dark, medium, and light gray to match each other and the color of white; and to *simultaneously* (2) get white to look really *white.* Not pinkish, bluish, reddish, etc. If you do this with Color turned down to zero in a fairly dark, neutrally-lighted room, and you avoid *staring* at the TV, you can come close to something quite pleasing, as your eye compensates for a lot. But the eye is much less forgiving of grayscale nonlinearities. Get that grayscale nice and even: No warm whites plus greenish or bluish shadows!

However, my experience with dark-room, eye-based calibrations are that they turn out a little cool in white balance. As I have already written, front-lighted puffy-white clouds at mid-day are a nice reality-check, usually a little warmer than the "ideal" 6500K temperature, but very good. Most important, they illustrate an absence of green or pink contamiation or excessive blue.

Neither have I. One of the problems of LEDs, as I recall from my electronics experience, is that their light output is a bit temperature-sensitive. If the R, G, and B LEDs differ in this regard, then the temp of the light changes as well. Maybe this is ameliorated by using a constant-current source for energizing each LED. Of course, then you would have to make *that* insensitive to temperature, or maybe even compensate for the LED's characteristics. You would have to integrate the light from the three LEDs.

Another choice would be to combine a theater gel with a white LED. There are some very subtle colors available from, say, Roscolux, and they offer a nice color-swatch book for free. The light pattern of most white LEDs I've seen have a center that differs in color from the rest of the light output. It, too, would have to be integrated (diffused) somehow.

http://www.rosco.com/us/filters/roscolux.asp

I keep thinking: Maybe the best thing would be to buy/scrounge a cheap TV, make it produce a solid, decently-bright raster, then calibrate it. It would help to have some *area* of standard to compare white with, and that might do the trick. You'd have to warm it up a bit first, but who cares?

My question then becomes: Do I have a friend or relative with a junker TV they'll part with? How about yard sales?