THE SONY SERVICE CODES - Articles, Comments, Discoveries - Page 24

Sorry to have been absent from ths forum for so long. There is that need to shoot the engineer and publish the results, no?

What I have been doing is watching tons of movies from my newish Panasonic S97 DVD player, which is nearly perfect over its component outputs at 480i, letting the TV perform 3:2 pulldown when I care about it (not always). As a result I have been revising my settings for image processing and shaping that is, the equalization and/or enhancement of real-world video so it looks like a million dollars. Then Ive been adapting what I have learned there to the other video inputs and sources.

In particular, I have been trying to track down *every* source of edge enhancement and ringing in the entire signal chain, starting with the DVD disk. Fortunately there are some DVDs that simply have no significant edge enhancement or artifacts of their own. How does one know? By playing them on a computer screen digitally and displaying the picture pixel-for-pixel. A good example is The Incredibles, an example of great computer-to-DVD perfection, followed closely by Monsters, Inc. This test can be done with the open-source software VLC from videolan.com, with the display-ratio settings at 720:480 (Mac or PC).

Without getting into many gory details quite yet, I will say the following:

(1) I mentioned sharpness/EQ settings hiding in a couple of places. Right. Here they are:

(a) First, in plain sight: 2170P-3. I have decided that #9-F1LV introduces too much after-edge outlining for me, and I have generally set it to zero. Further, after having lived with SYSM=3 for some time, I now realize that the least harmful equalization can be found in SYSM=2. Set properly, it acts like true unsharp-masking (a term digital-camera fans will understand).

(b) Also, I use the MID5 enhancements *very* conservatively. POP=63 is always set to all-zeroes for nos. 1-18, which pretty much makes MID5 pass-through if I set MIDE=63 in 2170P-3. Then small amounts of very-fine-detail enhancement can be added in other columns by setting MHYL=3 and varying MHYE from 0 (no effect) to 7 (max effect). Were talking skin and cloth texture, here, not big edge-sharpness. Settings of MHYE=2 or 3 are helpful for all fine sources, such as DVDs and HDTV. A heavier setting for SD sources can be had by setting MHYO to 1 (makes it coarser), MHYL=3, and then MHYE to 1 to 3.

(c) The hidden enhancements are real troublemakers! The worst is in 2103-1, nos. 6-8, SHAP-SHF0-PREO. SHAP is yet another sharpness enhancer, but this one is in mid-detail, and it rings, adding coarseness and a video look to fine sources. 2103-1 does NOT affect certain digital inputs (e.g. HDMI, HDTV from tuner) and always affects V5/V6 component 480i, S-video 480i, and RF-tuner and digital-tuner 480i. Recommended: set SHAP to zero, SHF0 to 3, and PREO (less critical) to 0 or 3 (havent decided yet).

(d) For all S-video, composite-video, and RF-tuner 480i sources, the 3D comb filter separate color from luminance, and the 3D-COMB section has significant sharpness tweaks, nos. 14 and 15: VAPG and VAPI for vertical anhancement (generally not used, which means zeros); and nos. 17 and 18: YPFT and YPFG. In this case, YPFG can act as either a detail-reducer or enhancer, with no effect at YPFG=8. (Possibly 7; still experimenting.) Note that the 3D-COMB settings have no effect on all high-quality sources, such as HDTV, HDMI, and component inputs. Recommended: Set VAPG and VAPI to zero, YPFT to 3 and YPFG to 8.

***************

SO . . . Imagine a chart so we can track this stuff. (Attached is a PDF of a useful chart for keeping track of these settings. Print on legal-size paper.)

The settings can be grouped in a table. Reading down the far-left column . . .

3D-COMB: 14-15-17-18 (4 rows) X 4 columns, 1 for each picture mode (Pro, Standard, Movie, Vivid).

2103-1: 6-7-8 (3 rows), only one column for all picture modes.

2170P-3: 0 thru 16, SYSM-VMLV-VMCR- ... - MIDE (17 rows grouped 1-1-4-4-3-3-1, according to function) X 4 columns, 1 for each picture mode.

MID5: 0 thru 18, POP-MHLY-MHLC- ... -MVCE (19 rows grouped 1-4-4-4-3-3, according to finction) and sample columns showing the POP values for possible MIDE values in 2170P-3, above.

The row-orders are the same as in the parameter charts from the Sony service manuals that have been published.

Example: My latest thinking for DVD-480i thru component inputs in Pro mode is:

3D-COMB: 14-15-17-18 = not applicable.
2103-1: 6-7-8 = 0-3-3
2170P-3: 0 thru 16 = 2 n 0-3-2-8 0-1-3-0 0-1-0 5-0-0 60, where n depends on the velocity-modulation setting. UBOF = 5 balances the black level against my other inputs. There, YMMV.

MIDE: 0 thru 18 = 60 0-0-0-0 0-3-3-0 0-0-0-0 0-0-0 0-0-0

This requires the user Sharpness slider to be set at 25-35, depending on taste. Mine is usually near 30.

************

Cryptic and confusing? I guess those were some of the gory details, indeed! I have been through these numbers so many times that I think of them as a series. When I am writing down what I am doing, for example, in the 2170P-3 settings, I just scribble the 17 numbers in a column with lines between the groups. That way I can see patterns developing or changing as I experiment. If the chart is used as a template, then one can publish settings as a number-series, as I have done above. Note that a chart as described above may cover all picture modes, but there has to be a separate chart for each video-input class and scan type.

Final thoughts for now:

(1) I believe that very conservative velocity modulation settings (ClearEdge) are a benefit for non-HD sources, maybe even for HD. I have set it according to what I have previously written: For all picture modes, 2170P-3-VM, VMH, VMM, VML are set to 0-9-6-3, respectively. I typicaly have the user VM control set to Medium, maybe Low or Off for HDTV. Set conservatively, it does NOT suppress fine detail as rumors would have it. (But some of the ringy enhancement filters can do that, as implied earlier.)

I have been experimenting extensively with the settings in 2170P-3, nos. 2-5, the codes that entirely determine the "look" of VM. The current settings of 0-3-2-8 replace earlier settings of 0-3-1-0 and seem to cause less distortion of high-contrast vertical edges while reducing the mush from the scanning spot. The differences are subtle and matter only if there is any VM turned on at all.

(2) I cant fathom Sonys claim of minimum enhancements for the Pro mode, as it clearly invokes certain digital filters that significantly modiofy the image, in some cases to its detriment. Ive tried to ferret out these effects and control them. I am making judgements based on a near-photographic reproduction of DVD material from source to screen, and some enhancement and sharpening *is* necessary but it doesnt have to be ugly! The same settings applied to HDTV prove the result almost shocking photo-like detail, when the broadcast is good.

(3) Summary: The most valuable lesson I have learned here is setting 2170P-3/SYSM to 2, reserving all fine-detail adjustments for the MID5 group of codes, and turning OFF all other sharpening filters for fine sources. The the Sharpness slider is used to adjust overall sharpness to taste. (Exception: for SD cable-broadcast, I turn 2170P-3: F1LV up to 3. Ordinary broadcast can be pretty awful, but even that benefitted from getting the nasty sharpness eliminated from 2103-1. It also made my VHS tapes look gawd-awful and grainy; much better now.)

UPDATE: Attached chart is now revised to include SSMD and PPHA parameters. See post #1040 here.

 

IPChart05tall.pdf 80.22265625k . file

The focus pot is an *overall-focus* adjustment, as is 2170P-4 #3 QPDC. They sort of duplicate each other -- but not quite. You can isolate them only by adding an electrical jumper onto one of the circuit boards, and that's not possible without taking the back off the set. That's why I suggested setting QPDC to either the default or (maybe for you) the middle of its range.

Yes, the whole screen is affected by the adjustment pot, but you should pay attention only to the center and get that to be the best compromise between the pot setting and QPDC. In addition, you are trying to balance any asymmetry in the deflection yoke or its mounting on the tube by doing this step: "Adjust 2170D-4 #8, DQP until vertical line widths at equal distances from center are balanced." I.e. you want the left and right sides to be *equally* out-of-focus during this step. This is how Sony suggests to do it, and it worked fine for me on my 36XS955.

Having now set up the center focus, all you can do is the best you can with the dynamic-focus controls described in steps 5-7. The corners will *always* be crappier than the center -- fact of life for CRTs. Just do the best you can; it's going to be a compromise no matter what. The detailed dynamic-focus settings are designed to minimize the *differences* between the screen's outer reaches and the center.

Focus and convergence are completely different matters. Focus is the attempt to make the scanning dot (of whatever color) as small as possible over as much of the screen as possible. Convergence is the attempt to make the three color dots (red, green, blue) line up perfectly over as much of the screen as possible, no matter the focus.

If the green pattern is well-focused, the other colors are as well, as the three electron guns are designed to track that way. But if you see color fringing on horizital or vertical edges or lines, that means the three (well-focused) colors aren't lining up properly, and that is what convergence is all about. It's usually set up with a fine-dot pattern over all the screen. You can adjust for nearly perfect horizontal convergence (no fringing on vertical lines), but fringing on horizontal lines is tough to cure, as it means fooling with permanent magnets inside the set -- not a pleasant job, and sometimes the best you can achieve, again, is a compromise.

If you have the center 3/4 of the area of your picture well-focused and converged, you likely will not pay much attention to outer aberrations when watching actual video material, especially in the corners -- well, excepting maybe for games and other computer graphics. . . . And I hope this method works well for your HS500, a set with which I have no personal experience.

I'm really suspicious of this "pulsing." If it were an adjustment-fixable problem, it wouldn't be pulsing. The way you describe it, the pulsing sounds like the difference between the scan rate of NTSC-standard TV (59.94Hz or something like that) and the exact 60Hz of the power mains. This means that it could be caused by a low-level interference from a 60Hz outside magnetic field caused by a big motor, transformer, overhead power lines, roof air conditioner, apartment-building wiring, whatever. If you change the orientation of the set in the room, by turning it or relocating it, and the pulsing changes in pattern or degree, I really would suspect this.

(This problem plagues computer-monitor users, too. How about major power lines just outside a third-floor office of a client? Caused an maddening 15Hz wobble of the 75Hz-scan display.)

Oh, Hell, I just thought of something else less pleasant. It could be that there is some kind of leakage in the device that is supposed to turn of *completely* the degaussing current that causes the big "bzzt" when you first turn on your set. If it's a relay, small chance of that; but if the device is what is called a PTC resistor, then it may be defective. Can you have it checked out under warranty? A definitive test would be to open the back of the set and temporarily disconnect (unplug) the degaussing coil from the curcuit board where it normally connects. If the pulsing stops, that's the problem.

Here is the only one I'm aware of, for the 34XBR910, which has only a few differences with the later sets. There are a couple of vatiations of this, but this seems the most informed. Note that some of the "descriptions" are, at best, wild guesses.

But note that this is not as useful as you would believe. The spreadsheets give you *no* clue as to how the code-settings are memorized for different combinations of picture mode, video mode, and input type. Except for an occasional quick look, I have found the spreadsheets essentially useless! They helped at the beginning of my experimentation, but only the charts can make plain what is stored when you WRITE the settings.

Example: The MID5 code-group alone starts with a table of 64 columns X 19 rows, all visible in the charts. The spreadsheets give no clue! Same with the complex 2170P groups. The simple vertical listing in the spreadsheet is useless!

Both are attached>

 

34XBR910 Service Menu vADR.zip 57.740234375k . file

 

XS955:XBR960 Service Data.pdf 405.4189453125k . file

FAIR WARNING: CHANGES DURING WARMUP

It will not come as any surprise to anyone reading my posts that, for me, no tweaking of these XS955/XBR960 sets brings as much benefit as optimizing the image-processing path. My post #707 is an example. (Of course, that presumes you have satisfactory focus and geometry but geometry is a separate issue.)

But I have some shocking news in case some of you have found yourselves chasing your tail at times: Something substantial changes within the set during warmup. Well, it's more complicated than that. Let me explain.

It is known that solid-state devices (all of the chips, transistors, and processors) change electrical characteristics with warmup. Amplifiers improve in gain and ability to handle high frequencies. Memory and switching chips actually slow down slightly. The whole chain of processing in these sets is really crappy when cold, improving substantially in the first ten minutes of operation after a really cold start, say in the morning in a cool house. The tube doesn't look its best, either.

All you have to do to prove the point is to put up one of AVIA's Resolution test patterns. The second pattern in the Resolution group, 200 TVL, is good. Turn on a cold TV, and put up this pattern in Pro mode as soon as you can from your DVD player. Then watch the vertical wedges and the lower-right circle change over half an hour! But that's only the half of it.

Example of what got me wondering: I once put up a test pattern 5 minutes after startup, reviewed my charts and planned what I was going to do, then looked at the TV and realized I had forgotten to put it into Service mode. 15 minutes had gone by or more. Fine. So I restarted the TV and went through the button sequence on the remote to put it into Service mode. But wait! Now the pattern looked much improved! What was going on here? Only 20 seconds had elapsed.

After getting somewhat confused by this and undergoing some hair-replacement, I have come to this solid conclusion: This TV performs some sort of self-optimizing calibration at startup, likely to compensate for temperature and set-to-set variations in components. It happens during that first 10 seconds or so while the screen is dark. It has nothing to do with Service mode. Important: It performs this calibration at EVERY RESTART.

I'll state it a different way: If you turn on your set from cold, it performs this calibration as best it can, and presents a picture. But it is not at its best yet that takes 15 minutes or more of warmup. If you display high-quality video, say through the component inputs (a 480i test pattern proves the point), then restart the set after 15 minutes or more, the image/pattern improves, and very substantially for the best video (a great DVD, say). And by restart I mean power-off/on for 0.5 - 1 second, that's all. Fine detail seems most affected.

I don't think at present it affects straight-digital inputs, such as HDMI or HD broadcast from the tuner; but it definitely affects any 480i analog video from any input. Which leads me to believe the culprit is the 2130 chip, but I can't prove it, and it doesn't really matter. I can't say whether mine is the only set that does this, either, but I doubt it.

Bottom line: If you are going to experiment with image-processing tweaks, restart the set after 15-30 minutes before working on it. And if you are going to watch any 480i material, warm up the TV first, then restart after 15-30 minutes. It will then be at its best. I now routinely restart the set after warmup, no matter what I am watching, to force recalibration. It may not matter for the evening news or CNN, but it does if the analog broadcast is high-quality and you're awake enough to care.