AVS Forum banner

21 - 40 of 2977 Posts

·
Registered
Joined
·
57 Posts
Ken,

Have you made any changes to your sections (01 -07) since you posted on Agoraquest? I haven't tweaked my 34XS955 yet, still waiting for some burn in time, but I had saved those earlier posts for reference notes. Big thanks for sharing your findings to you and others. - CP
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #23 ·

Quote:
Originally posted by CPetty
Ken,

Have you made any changes to your sections (01 -07) since you posted on Agoraquest?

I re-read those articles and made minor wording changes in my copy before posting here; I don't think there was anything really substantive. And a thorough spellcheck turned up a few embarrassments!


You're welcome!
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #24 ·
08 - SOME USEFUL TEST PATTERNS


Following are sets of test images that are extremely useful in adjusting the displays of both 4:3 and 16:9 sets. They are designed for MEMORY STICK use only, or for display from a computer. (Untested. I could use some feedback on this. Is this resolution okay for computer-to-TV display?)


Why memory stick? Because it is one good way to take nearly complete control over the pixels on-screen, important for test images. The native resolution for memory-stick images is 1080px high by 1440px wide for 4:3 sets, 1920px wide for 16:9 sets. The pixels are square. A same-resolution image is displayed *exactly.*


So if you should create a dynamite custom test image on your computer at exactly 1080h X 1440w or 1920w, you can save it at highest quality as a jpeg image, and put it into the DCIM folder on your memory stick. You can then view it on the TV without any resampling by the TV, in a mode called HD FULL. Perfect!


Unzip the files attached to the following messages, yielding the separate jpeg images. Using an appropriate card reader/writer, copy them from your computer to the DCIM folder on the memory stick. (Create this folder at root level if it's not yet there.) You may also use subfolders inside DCIM to categorize your images. Then write Sony a nasty note after you discover how needlessly convoluted it is to navigate subfolders in the MS-display mode. Sheesh!


Some of these are useful images even if resampled from jpeg-on-CD on a DVD player (inherently 480i). See individual descriptive messages. I wonder how many other ways there are to get custom test images to display accurately. A slideshow on DVD maybe? Suggestions?


**************
Convergence & Geometry Test Patterns. Attached to this message are some convergence and geometry test images that should help in assessing what's going on with your particular TV's display. Remember that the crosshatch patterns will look a lot worse than actual program material! The white-on-90% black are best for convergence. Black crosshatch on light background will help with geometry and focus. All patterns are based on 2-pixel line widths and 2 X 2-pixel dots.

 

converg_geom_test.zip 245.9482421875k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #25 ·
Attached and following are really good focus test patterns using 1-pixel scribbles against light and dark backgrounds. The values are approx 20% and 80% brightness, not black and white, so you can easily see edge-enhancement aberrations and ringing. The two images are posted separately because of the file-size limitations here.

 

focus16by9.zip 458.017578125k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #26 ·
(See previous message.)

 

focus4by3.zip 323.8310546875k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #27 ·
Attached is an image that was a bear to make, but it is valuable in assessing how your TV handles horizontal and vertical detail. How the pattern darkens around the center white circle varies with amount of sharpening, and the MID5 & 2170P-3 settings. I'll also try to make a similar image with true light and dark sectors, not constant-width spokes. That will be tougher! The corner patterns are not very important, as this really isn't a focus test image but more of an sharpness-enhancement display. I didn't do a 16:9 version.

 

spokes.zip 334.40625k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #28 ·
Attached is an image that shows the effect of adjusting the color-sharpness parameters in the 2170P-3 codes and in the MID5 table. Hardly visible in normal program material, color-sharpening and boundary effects are clearly visible in the color boundaries in this chart between the most-opposite colors, such as green-magenta. My settings have been tuned for clear boundaries with minimum bright-line aberrations and ringing. (I'll try to write something about the color parameters another time.)

 

adjacentcolors.zip 79.5068359375k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #29 ·
This excellent chart was lifted from a Photoshop document I found online about ten years ago. It will show you your effective display gamma at a glance. It MUST be displayed at 1080px vertical in Pro mode, since it relies on alternating groups of exact scan lines for its accuracy and NO vertical sharpness enhancement. It can be adapted for different vertical resolution by *trimming* the image to another dimension, NOT by rescaling it. Memory stick works fine.


The principle is this: A pattern of alternating black and white equal scan-line groups has an average brightness of exactly 50% of full-white, inherently, no matter the gamma. The lines are horizontal so phosphor-grid and image-processing factors are eliminated. At a gamma of 1.00, a 50% solid-gray block appears to be the same brightness. (Blurring your eyes helps.) But at a gamma of, say, 2.20, that gray block would appear much darker than 50%.


But one can create a gray block that is lighter by the exact degree that will match the 50% surroundings. Its adjusted brightness can be calculated as raised to the 1/ power, or 0.50 to the 1 / 2.2 power, which is 0.73. So a gray block of 73% brightness embedded within a background of alternating black and white horizontal lines will appear to disappear if you blur your eyes while examining it *only* at a gamma of 2.2. In digital imaging, where black is 0 and white is 255, that square should have a value of 186.


So this chart is made up of a bunch of gray blocks of different brightness set against a black/white-line background, and corresponding gamma figures are indicated below. As you view the chart (not too close), one block will appear as the one that most blends into the background, and the corresponding gamma can be read from the red numbers below.


By this chart, my 36XS955 has a maximum gamma (all settings at zero) of 2.4+ - quite high. Makes a lot of insipid TV look much better. But the standard is 2.2, and I designated the GAMM = 1 column in the table for that. Each of GAMR thru GAMB are set to 3, and that seems to give a perfect display gamma of 2.2. Your outcome may differ, of course.

 

curr_gamma_test.zip 115.45703125k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #30 ·
09 - GETTING GOOD GRAYSCALE


[9-9-05 Note: I no longer use the following "trick" for fudgling grayscale, i.e. setting one of the gamma settings higher than the others. I have reconsidered the whole matter, and I have gotten nearly-perfect grayscale on my 36XS955 by using excellent gray-step patterns from both the AVIA and DVE DVDs. Comparing with the same pattern displayed on a small computer monitor that has been calibrated to 6500K was a big help but probably not necessary. This is best read as documentation of my learning process. See additional notes in brackets.]


[I believe the following is still true.] I believe that the only way to great TV color is by first adjusting for great grayscale. Why? Because if your midtones are, say, cyanish compared to white and dark gray, that tint will attenuate reds everywhere in the midtone areas. Blue skies will have a touch of green. Skin tones will be yellowish not attractive! Color information is added on top of luminance (grayscale) information, and so grayscale reproduction has to be *right* as a fundamental base for building the color image. Setting the Color user-menu slider all the way to the left removes all color, so it's easy to see where you stand. You can just tune random broadcast material and check it out. Your eyes are a very good judge of this. Using a test DVD helps but isn't really necessary for judging this. Aesthetically speaking, errors of the yellowish-greenish type seem most offensive to me; blue or magenta less so.


When I calibrated my 36XS955 for perfect whites (color temp) and neutral dark grays, I found that my midtones were yellowish! White was already correct (the right combo of RDRV, GDRV, and BDRV), and so I thought I could fudge the cutoff settings (RCUT, GCUT, and BCUT) to get neutral midtones. Nope! When I did that, *some* midtones were okay, but now the dark grays were much too blue. [Not any more. Used a different method to get superb results.]


I had been through this before with computer monitors. The problem is that the inherent gamma or electrical response-curve of the CRT is slightly different for the blue electron gun compared to red and green. If the midtones are yellowish, the blue signal channel needs its gamma adjusted downward, pushing the midpoint of the curve upward. [Theoretically true, but may not actually be a determining factor on my TV.]


[The "trick" follows, no longer used.] Well, we can do this in service mode. The adjustment is a bit coarse, but it worked for me. Originally I had set GAMM = 0 and then GAMR, -G, and -B each to 0 for maximum gamma. I had done the same for GAMM = 1 and had set the individual colors to 3. But I now needed to correct the blue. So I tried raising GAMB to 1 where it had been 0, and to 4 where it had been 3, and wrote all data.


Now the grayscale was near perfect! A little twiddling of BCUT, and it now appeared dead-on for normal program material. I looked carefully at the darkest grays and nearly-blacks. Maybe there was a trace of coloration there, but it was totally insignificant compared to the previous yellow-contamination of the midtones. Problem solved.


So, to summarize, I adjusted all settings for the GAMM = 0 thru 3 presets to show GAMB = 1 notch higher than the other colors. This corrected my annoying yellowish midtones in grayscale program material.


[Major reconsideration. Added notes 9-9-05.]
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #31 ·
10 - "PRESETS" AND RELATED SERVICE CODES


Understanding the following is really important to working in service mode: Some codes do not have effects of their own but rather point to a *group* of real parameter-codes that are set elsewhere. For example, GAMM = 1 means whatever the settings for GAMS, GAMR, GAMB, and GAMR define for it. The GAMM settings of 0 thru 3 are presets, represented by four columns in the service-data table for GAMS-GAMB, and these settings dictate each value of GAMM to mean whatever you want it to mean. Likewise, the user-menu choices for Warm and Cool color-temp variations are determined by the values in two columns for codes 2170P-1 #14-22 (WBSW-DCOL). This is hard to discern from the Excel-based service-code listing, and having an appropriate service-data chart is essential.


Here are some preset groups that are likely to be most important to tweaking color and image quality:


(1) User menu: Color Temp > Warm/Cool. Set by the two columns of 2170P-1 #14-22 (WBSW-DCOL). Neutral color temp is set by the fundamental white-point settings of 2170P-1 8-10 (RDRV-BDRV). Warm and Cool are *offsets* from Neutral, and 31 = no offset. See article #03 for more.


(2) 2170P-3 #16, MIDE points to a column in the immense MID5 table, which has 18 unique codes (rows) and 64 columns. MID5 #0, POP is a temporary pointer to any particular column in this table, so you can make changes without schlepping back and forth between 2170P-3 and MID5. Example: If you wanted to make changes in the settings for MIDE = 21, you can go to the MID5 #0, POP, and set it to 21. Then you are in column 21, and you can step thru the codes, make changes, and write them. Now any time you set MIDE in the 2170P-3 tables to 21, those settings you just made will take effect. (Of course, more than one column in the 2170P-3 tables can have MIDE = 21, and it will be the *same* 21.)


(3) User menu: Advanced Video > Program > Color Axis > Default/Monitor Set by two columns in the table for 2170P-4, #13-16 or 7-10, RYR-GYB. For the XS955 table, the columns are labeled Normal and Special Axis for Default and Monitor, respectively. After calibrating the Default color axis with Digital Video Essentials (RYR-GYB = 13-15-6-4), I found that the memory stick color was off using that setting: a real green push was still a problem. So I set the memory-stick axis to Monitor in the user menu, and recalibrated it in service mode (this time making changes in the Special Axis column. You can make changes in the user menu and watch the values change in RYR-GYB.) I used a memory-stick/jpeg custom test pattern that duplicates the DVE pattern. The settings ended up at RYR-GYB = 12-15-14-9. Perfect. Bottom line: You have *two* sets of color-axis adjustments you can apply as you wish with the Default/Monitor setting in the user menu. How you set them up is up to you.


(4) 2170P-4 #11 or 17, GAMM is a table of 17 columns for every possible video/input mode and 4 rows for each picture mode. However, the values of 0-3 simply point to a column in a 4 X 4 table for the four codes following GAMM: GAMS, GAMR, GAMG, and GAMB. (GAMS is a brightness offset and has no effect at GAMS = 7.) To set GAMM = 0 for maximum gamma, for example, set GAMM to 0, then immediately set GAMS = 7, and each of the GAMR, -G, and -B codes to 0. (But see Article #09 for how to fudge the gamma settings to get a perfect grayscale.)


(5) 2170P-4 #22 or 16, BLK controls the Dynamic Image feature advertised by Sony, an automatic modification of contrast and the brightness-response curve that depends on what is being displayed at any moment. It is described by a table just like GAMM, above, whose values of 0-3 point to one of four columns for several related codes following BLK: in 2170P-4, DCTR, APED, DSBO, ABLM, DPSQ; and in 2103-1, #24 and 25, ATPD and DCTR. [BLK is a little tough to figure out, but the factory's BLK = 0 seems to kill all dynamic effects. More another time. This feature is not to my taste.]
 

·
Registered
Joined
·
139 Posts
Hi Ken,


I think that a detailed description for adjusting geometry would be a welcome addition to this thread. I have always had a problem on where to start and how the adjustments affect each other. I did see your earlier attachment from the SM BTW.


thanks,


Joe
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #33 ·
Quote:
Originally posted by jcardani
I think that a detailed description for adjusting geometry would be a welcome addition to this thread.
Well, um, [squirms visibly] aside from reducing overscan a bit for 4:3 material and reducing height slightly for 480i and HD compressed mode with 2170D-1 #13, ASPT, I actually haven't done much for myself. I did some basic recentering, using the SM instructions (attached), but very little was amiss on my set -- compared, say, to the poor folks who have sets that were accidentally reset to factory-newborn status.


I have been writing this stuff as I do it, making sure I walk the talk. I'm really not quite ready to write up anything comprehensive yet for geometry, as other matters have been of greater importance to me -- such as focus, gamma, correct image brightness and color, and image-detail rendition. They contributed the most to my immediate enjoyment of this excelent set.


As I find the time, I will likely get to geometry. But my set has a few static-convergence defects that a Tech will have to resolve first. I'll be calling someone to come out this week. I want *him* to be the first to open the back of the set.

 

xs955|xbr960 geometry.pdf 146.7265625k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #34 ·
While I'm at it . . . for the 30-, 34-, and 36XS955 owners among us, see attached. Now, in living color!

 

xs955 service data.pdf 405.4189453125k . file
 

Attachments

·
Registered
Joined
·
715 Posts
Discussion Starter · #35 ·
And for the 34XBR960 . . .

 

34xbr960 service data.pdf 346.224609375k . file
 

Attachments

·
Registered
Joined
·
19 Posts
Ken,

I haven't made service menu adjustments with a TV for several years, but when I did make them with my Sony 32XBR2, I had to enter the service menu, change whatever values I was interested in changing, exit the service menu and then power up the TV normally to see what effect the changes that I had made had had. This tedious procedure had to be repeated as often as was needed until the parameter being adjusted was satisfactory. However, your posts imply that with current or recent model Sony TVs one can enter the service menu and simultaneously display the parameter that one wants to adjust. Is this inference correct?

Thanks very much.

Ron
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #37 ·

Quote:
Originally posted by rekalil
However, your posts imply that with current or recent model Sony TVs one can enter the service menu and simultaneously display the parameter that one wants to adjust. Is this inference correct?

Yes, you can watch any video input, change channels, switch to memory stick -- all while in service mode, with the SM green character-based readout superimposed over the picture. When you make a change, you see the effect on the picture immediately. You just can't use the number buttons on the remote for channel hopping, as they are used for controlling service mode, as are a few other buttons, too. You can exit SM without saving any changes by simply recycling power. In case any of the settings "stick" anyway (as one Audio setting seems to), pulling the power plug fixes that.
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #38 ·
11 - IMPROVING THE AUDIO


I was very pleased to find that my 36XS955 has a decent subwoofer and not much of a plastic-cabinet sound. Nice! But of course I've been mucking around in service mode on just about everything else, and there *is* a group of codes for AUDIO just waiting to be tweaked. So . . .


On my set I found that voice had an annoying nasal quality, and high-quality piano sounded honky in the same frequency region. Mid-range boost maybe? And the Treble slider in the user menu boosted the annoyance factor lots faster than the extreme highs. So what's up with that?


The 21 codes for audio are in the group called AUDIO. ASYS switches all effects on and off, and only ASYS = 1 (on) sounds anywhere near decent.


I found that #1-3, TRCV, BACV, and MDCV all are controlled by the menu sliders for Bass and Treble, but that they affect treble, bass, and midrange, respectively. If all are set to 3, the tone-control sliders do nothing. It's hard to imagine how this works with the midrange! Seems that settings lower than 3 vary the width or impact of the frequency range affected by the sliders. I set the tone sliders to their midpoint (31) to start, and set the above three codes to 2 for moderate effect. Sony's original settings for these made no logical sense to me.


I discovered that codes #6, 8, and 10, MIDL (midtones), SBAS (lows), and STRE (highs), sort of act like a 3-band equalizer, affecting the tonal balance of the set when the tone controls are set in the middle. My set came with them set at 8 - 7 - 6, respectively. Changing those to 6 - 7 - 9 removed the honkiness and added transparency to the highs.


I noted that the subwoofer-supplied bass was really tubby. Thump - thump on pop music, but no real warmth to voices or other instruments. Warmth in the 300Hz region adds pleasantly to the illusion of size of the presentation (best metaphor I can think of). So I diddled with #7, LOFQ, and found that increasing it to 5 from 0 added back something that was missing. I haven't figured out exactly what #4 and 5, SVHI and SVLO, do, and so for now I have left them each at 4, the original settings. And I haven't a clue about #12, 19, and 20 (PSEF, TRS1, TRS2). So they remain untouched.


I had hoped that the SteadySound feature might miraculously control the nasty-loud commercial breaks. Mine, however, had almost no effect! Then I discovered that, when SteadySound is enabled, #13, AGCL, controls its effect, with AGCL=15 minimum (no effect) ranging to 0 (very substantial compression, obnoxious). My set had 9 as a default - no wonder it had little effect. I reduced it to 4 for now, and am pleased with the results. I generally don't like any compression, as the broadcasters lay it on quite heavily anyway. But how about music for accompanying a memory-stick slideshow? Or afternoon football? I would try settings of 2-4 and then use it with discretion.


Finally there are what appear to be dynamic bass-boost settings with big-time control over the subwoofer. #14, BBE, turns all of it on and off, and off sucks! So BBE will stay at 1. I found it interesting (as opposed to enlightening) to tinker with the four main settings, #15-18, BBEP thru BB2L. But it's really hard to deduce what they do and which of them might be responding dynamically to volume level - like the bass-impact switch on some CD players. So, after playing around a bit, I have left them alone. The other changes detailed above made such a substantial improvement overall, that I am content for now.


So, to summarize:

-- Get one of the service-data charts in front of you. Write down your original settings.
 

·
Registered
Joined
·
637 Posts
Ken, thanks for all the good info. I adjusted my gamma today on my 36XS955 which made a big improvement in black detail. I can actually see the fabric now on items such as black suits which wasn't possible before. I'm going to definatley make some geometry adjustments also as the left side of the picture sorta 'leans" to the right.


A question for you though. Is there a way to adjust the brightness of the picture while recieving a HD station via the atsc tuner and not effect the NTSC tuner brightness? Currently there is a big difference in the two and I have to adjust the brightness everytime I flip back and forth.

Thanks,

Rick
 

·
Registered
Joined
·
715 Posts
Discussion Starter · #40 ·

Quote:
Originally posted by RickE
Ken, thanks for all the good info. . . Is there a way to adjust the brightness of the picture while recieving a HD station via the atsc tuner and not effect the NTSC tuner brightness? Currently there is a big difference in the two and I have to adjust the brightness everytime I flip back and forth.

You're very welcome, Rick.


You'll find the settings you are looking for in 2170P-3 #13-15, UBOF, UCOF, and UHOF. These are brightness (black level), color, and hue offsets on top of the global settings. Each increment here is equivalent to two clicks or so in the user menu.


Good strategy is to find your input source that has the *highest* predictable black level, and assign it a 0 for UBOF. (Touch up 2170P-1 #7 SBRT to recalibrate global black level, if needed.) For me it was my DVD player. Then a little switching around among sources and channels will let you find suitable settings for everything else. Write the change after each input or mode switch or you'll lose it. My favorite VCR has a slight pink shift to skin tones, so I was able to fix that with UHOF without screwing up other inputs.


Note that there are a maximum of about 60-something values that can be stored for *each* of the three offsets: one for each picture mode (4) times one for each input video mode (16 for my 36XS955). That's each of the four input resolutions (480i, 480p, 720p, 1080i) for each of V5/V6 together (component), all S-video, HDMI, and ATSC (tuner, SD and HD), plus individual settings for "RF," memory stick, and TwinView. Whew! Means you can tweak to your heart's content. Note that V5 and V6 component inputs are *not* separate, but the video modes are. Same with all S-video. Two 480i sources connected to V5 and V6 will have to share the same offset; but if one of them is *not* 480i (a set-top HD converter, say) , then they can be different.


SD broadcasts are completely inconsistent in black level, so the best you can do is choose a UBOF setting that causes you the least hassle when changing channels. HD broadcasts seem much more consistent, and your external devices (DVD, etc) may be easy to set & forget. I generally want the "base" setting for Brightness, Color, and Hue to be slider-midpoint = 31. Then I can adjust from there on the fly.


[Some channels (e.g. MTV, FoodTV) broadcast way too much color, and I have to tweak Color down to 25 or so. Is there no respect for standards? Don't they monitor the signal with a vectorscope or something?]


Very soon I will post an article of all of the 2170P-3 settings -- a big project! These settings determine the "look" of the picture: all of the detail rendering, enhancement, and filtering. With a few adjustments, you can do much better than Sony's factory settings!
 
21 - 40 of 2977 Posts
Top