24" Widescreen CRT (FW900) From Ebay arrived,Comments.

spacediver said:
it's most likely to account for the fact that with wide deflection angles, the beam cannot be kept as tight, so the spot size widens. You won't notice it, it's not a problem at all.
Click to expand...

You're probably right. Can you tell me which one would you choose from these 2:
21" GDM F-520
22" HP p1230 - diamondtron

ignore the screen size in your judgement.
 
GDM F520, tho the HP is excellent - it's a rebranded mitsubishi 2070sb. I'm not familiar with the calibration options for the diamondtrons, though I do have the service manual, and I believe there's a service menu. If it allows adjustment of G2 and peak luminance, then that's great.

The F520 has a lower dot pitch, and you can use WinDAS with it.

You might find some useful information in this thread
 
Last edited:
I saw a post on this problem in WinDAS awhile back but I can't find it. I'm in the middle of doing a full writeup on the procedures in WinDAS (I am including all instructions, sliders, what they do, etc.). But when I adjusted the ASC in WinDAS, when I'm asked to confirm the screen size of the ASC, I noticed that it becomes narrower than the 395mm that WinDAS asks for.

It's not a fault of the monitor. When I clicked "READJUST", I noticed that WinDAS actually LOWERED the H_SIZE from 104 (what I had to set it to so that it was 395mm) to 66. I readjusted it to 104 to make it 395. As soon as I clicked OK and went back to the confirmation step, it lowered the setting again!

So I ended up clicking READJUST one more time and set it to 142. Then when I went to the confirmation step, WinDAS shrunk the screen so that it was 395mm and all was good.

What does this mean? This could mean either one of two things:

1. Our "cracked" version of WinDAS is bugged (not likely, the only hack I'm aware of Gregua performing is the removal of the check for the security dongle)
2. WinDAS has a typo and the 395mm should be something else (360mm?).

I believe scenario #2 to be more likely because as I am writing down the instructions, I'm noticing typos all over the place. I find it amazing that Sony meant this for service techs, seeing all the "Engrish" in this software. Shameful. Though, some of the instructions are unintentionally hilarious:

"Adjust G2 to be visible the raster." Pure gold. :D

My parents are in town and are visiting soon, so I'll have a chance to go through this in the afternoon and finish my writeup. So far I have finished VDC and Max Frequency.
 
:) - looking forward to the guide - geometry and convergence in particular are areas I don't have much experience with.
 
about variable dot pitch:

BauEcGF.jpg


they put it as a minus
 
Last edited:
nice find, I haven't noticed any issues, but tbh I haven't looked carefully at focus at edges vs center. With my macro setup, I'll be able to measure sharpness in center vs edge, and I'll also be able to measure the dot pitch at any location on the screen.
 
My understanding of the variable stripe pitch is this. The GDM-FW900 is actually a round tube with extra glass packed on to "flatten" the tube, so to speak. I remember reading in some documents where Sony took that extra step to get a flat tube. The GDM-F520 and GDM-C520K (Artisan) are truly flat tubes, and have a uniform stripe pitch of 0.22mm and 0.24mm respectively.

I think of the GDM-FW900 as this: a "Fine-pitch" version of the GDM-W900. It could very well be that the "F" in "FW900" stands for "FD Trinitron" - which is what Sony's later CRT's were known as. Because on paper, it looks like the FW-900 is a beefed-up W900. Tighter grille pitch, higher bandwidth capabilities, "flat" screen, etc.
 
spacediver said:
:) - looking forward to the guide - geometry and convergence in particular are areas I don't have much experience with.
Click to expand...

I'm already up to 8 pages on MS Word. :eek: I don't know how much I should do on the lower frequency steps though. Because once you do the higher frequencies, it's more or less just a repeat but with lower frequencies.
 
jbltecnicspro said:
My understanding of the variable stripe pitch is this. The GDM-FW900 is actually a round tube with extra glass packed on to "flatten" the tube, so to speak. I remember reading in some documents where Sony took that extra step to get a flat tube. The GDM-F520 and GDM-C520K (Artisan) are truly flat tubes, and have a uniform stripe pitch of 0.22mm and 0.24mm respectively.
Click to expand...

The deflection angle in the 520 series is significantly smaller than in the FW900, so a variable pitch wouldn't be necessary. I'm pretty sure the variable pitch comes down to basic electron optics.
 
jbltecnicspro said:
My understanding of the variable stripe pitch is this. The GDM-FW900 is actually a round tube with extra glass packed on to "flatten" the tube, so to speak. I remember reading in some documents where Sony took that extra step to get a flat tube. The GDM-F520 and GDM-C520K (Artisan) are truly flat tubes, and have a uniform stripe pitch of 0.22mm and 0.24mm respectively.

I think of the GDM-FW900 as this: a "Fine-pitch" version of the GDM-W900. It could very well be that the "F" in "FW900" stands for "FD Trinitron" - which is what Sony's later CRT's were known as. Because on paper, it looks like the FW-900 is a beefed-up W900. Tighter grille pitch, higher bandwidth capabilities, "flat" screen, etc.
Click to expand...

I actually saw such extra glass on a Diamondtron used to get the flat appearance. (Surface of tube flat, but inner aspect could see a curve.) However, F520 and FW900 totally flat from what I can tell.

FD Trinitron -- yes exactly. Flat Trinitrons...achieved with enhanced computer control and such if I recall...
 
jbltecnicspro said:
I saw a post on this problem in WinDAS awhile back but I can't find it. I'm in the middle of doing a full writeup on the procedures in WinDAS (I am including all instructions, sliders, what they do, etc.). But when I adjusted the ASC in WinDAS, when I'm asked to confirm the screen size of the ASC, I noticed that it becomes narrower than the 395mm that WinDAS asks for.

It's not a fault of the monitor. When I clicked "READJUST", I noticed that WinDAS actually LOWERED the H_SIZE from 104 (what I had to set it to so that it was 395mm) to 66. I readjusted it to 104 to make it 395. As soon as I clicked OK and went back to the confirmation step, it lowered the setting again!

So I ended up clicking READJUST one more time and set it to 142. Then when I went to the confirmation step, WinDAS shrunk the screen so that it was 395mm and all was good.

What does this mean? This could mean either one of two things:

1. Our "cracked" version of WinDAS is bugged (not likely, the only hack I'm aware of Gregua performing is the removal of the check for the security dongle)
2. WinDAS has a typo and the 395mm should be something else (360mm?).

I believe scenario #2 to be more likely because as I am writing down the instructions, I'm noticing typos all over the place. I find it amazing that Sony meant this for service techs, seeing all the "Engrish" in this software. Shameful. Though, some of the instructions are unintentionally hilarious:

"Adjust G2 to be visible the raster." Pure gold. :D

My parents are in town and are visiting soon, so I'll have a chance to go through this in the afternoon and finish my writeup. So far I have finished VDC and Max Frequency.
Click to expand...

I had a similar experience as you described there, I ended up clicking READJUST and then increasing the width so that in the next step where WinDAS executes ASC it has the correct size.

Great to hear you're writing a detailed guide! I could sent my current notes, but I'm only at this for about 2 weeks, not sure if that would help :)

----

Today I was busy with Dynamic Convergence, and noticed that most of the points do not lineup with the crosshatch grid pictures from a few pages back (they are horizontally divided by 9 and vertically by 15). So I spend about 1.5hours making my own 'grid' and noticed that it's not linear AT ALL. Check this link for my messed up crosshatch pattern.
So my question is: is the Dynamic Convergence grid just not linear, or is my geometry screwed?
 
Last edited:
That pattern is definitely not linear. It's challenging making these patterns, you have to do a bit of math to figure out the best way to divide the screen, and then carefully construct the pattern using appropriate software and making efforts to ensure no anti-aliasing is introduced when exporting the pattern into a bitmap.
 
spacediver said:
That pattern is definitely not linear. It's challenging making these patterns, you have to do a bit of math to figure out the best way to divide the screen, and then carefully construct the pattern using appropriate software and making efforts to ensure no anti-aliasing is introduced when exporting the pattern into a bitmap.
Click to expand...

It's not the software's problem that it's not linear, nor user error, I made this according to how WinDAS changes the convergence on its 135 points. What I did was purposely misaligned convergence at all points in row A, and all points in column 01 (not all at the same time, but individually). Meanwhile I had open Photoshop in full screen and position the lines according to where the convergence was worst.

So my question is still the same: are the Dynamic Convergence grid points (as provided by WinDAS) just not linear, or is my geometry screwed?
 
Ah, I see, I was wondering how you derived that pattern. Interesting approach. The best way to test your geometry is to load up a pattern that you know is linear (in pixel space), and ensure that it's linear in physical screen space. That way you can eliminate bad geometry as a culprit here.

I may do a similar test as you did when I have time. If I do, I'll definitely report back here. But if your findings are accurate, then it would make sense to use your custom crosshatch pattern to adjust dynamic convergence!

One thing though: it may be the case that your findings may not generalize to all rows and columns. For example, row E may be diff from row A, and column 8 may be different from column 1.


edit: hmm, hang on. How do you know that you misaligned everything equally? Without that knowledge, you can't be sure where the "locus" of the convergence effect is (unless I'm missing something here, which is very likely :p )
 
quick update on my macro project:

The first order of business is to establish a baseline performance of the camera. To do this, I ordered a ronchi ruling, which is basically a high quailty optical target that consists of a square wave grating. For my needs, it serves two purposes:

1: allows me to measure the modulation transfer function of the camera.

2: allows me to measure units of distance in the focal plane. The ruling is 10 line pairs per mm, so I can easily convert pixels to microns once I have acquired a focused image.

Here is an image of the ronchi ruling in action. I have it taped to the display surface (it's transparent, so the light of the CRT conveniently illuminates the ruling).

14jxa8p.png


Also shown is the Canon EOS utility that is extremely useful for this sort of stuff. You can see a zoomed in live view of the ruling. Very handy for focus adjustments.

By taking images of the ruling when it is at a slight angle, I can obtain the modulation transfer function of the camera, and importantly, I can do so at different apertures. This will allow me to find the aperture that is the sweet spot between lens aberration and diffraction.

For anyone interested in modulation transfer functions, and the slanted edge method, see this excellent document by Douglas Kerr. I'm halfway through it, and it's by far the most readable tutorial on this technique that I've found.

Initial measurements indicate each RAW pixel corresponds to about 1.3 microns. Once I subsample the images to extract trichromatic values (of which, luminance is the most important to me), this will increase to 2.6 microns per pixel.
 
I see there are some really great projects in the works! :)

Good luck with your experiments spacediver, I don't know anything about optics, but I'll be really interested by your observations.

About the issue with WinDAS, first of all, it does not come from the patch, I can confirm that, I've looked into it, see how it was done. It's indeed a typo I think, and you have to guess the right values, sadly I can't remember the values I used.

About patches, I worked on a patch for WinCATS, it unlocks it, but I still get a popup saying something weird about the configuration. It asks me to check the configuration everytime I try to do something useful. I never made patches before, so I probably can't do any better.
Anyway, I have no way to test WinCATS, and even if I could, I don't know if I would. This software looks weird, and my monitor is already well calibrated enough.

wincats_check_config.png


About typos in WinDAS, and other stuff, I looked a bit into the way the files are organized in the program folder, and as far as I can see:
[WinDAS\] - the root of the program containing the executabled and libraries
[WinDAS\data] - contains the protocols for each monitor supported (for example G1W is for the FW900, afaik)
[WinDAS\SG] - contains the patterns to generate (format VTG-Ver 5.2.0, I don't know what it is anyway, but it looks interesting...)
[WinDAS\WinDAS] - contains the procedures for each monitor supported (and a file named "temporary.dat" with display modes I don't know what for)
[WinDAS\WinDAS\Msg] - is the interesting folder here, which contains lots of texts, and maybe every each sentence displayed during the different procedures

In the last folder, there are multiple files named "windas_X.msg". For example here is a quote of the beginning of file "winsdas_b.msg".

Code: 
[COLOR="#7F9F7F"][B][I]#-------------------------------------------
# WINDAS_B.MSG
#	Message for geometric distortion adjustment.
#
# Created :
#	Feb/2000	R.Sasaki
#-------------------------------------------
#[/I][/B][/COLOR]
[COLOR="#CEDF99"]$0[/COLOR][COLOR="#9F9D6D"]:[/COLOR]
[COLOR="#CEDF99"]$1[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Increase G2 to be visible the raster.[/COLOR]
[COLOR="#CEDF99"]$2[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust H size and V size to following[COLOR="#8CD0D3"]\n[/COLOR]
spec.[/COLOR]
[COLOR="#CEDF99"]$3[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust H phase to make picture at the[COLOR="#8CD0D3"]\n[/COLOR]
horizontal center.[/COLOR]
[COLOR="#CEDF99"]$4[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust V center to make raster at the[COLOR="#8CD0D3"]\n[/COLOR]
vertical center.[/COLOR]
[COLOR="#CEDF99"]$5[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Press the READJUST button to obtain[COLOR="#8CD0D3"]\n[/COLOR]
tracking, if necessary.[/COLOR]
[COLOR="#CEDF99"]$6[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust H phase and V center to center[COLOR="#8CD0D3"]\n[/COLOR]
the image.[/COLOR]
[COLOR="#CEDF99"]$7[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust V center to make picture at[COLOR="#8CD0D3"]\n[/COLOR]
the vertical center.[/COLOR]
[COLOR="#CEDF99"]$8[/COLOR][COLOR="#9F9D6D"]:[/COLOR][COLOR="#CB9393"]Adjust H_phase_shift to make picture[COLOR="#8CD0D3"]\n[/COLOR]
at the horizontal center.[/COLOR]

You'll see the index $1 matching this now famous scentence. :D
So this just means that, everything can be corrected, and even translated (but I don't see any good reason for a complete translation here anyway).

EDIT: Added the screenshot.
 
Last edited:
Specific question:

I'm using the standard 1920x1200@85hz res. I started playing Bioshock 2 and want to keep using this res, but somehow this game insists that the monitor must run at the highest res of 2304x1440@80hz. When 1200p is selected, it upscales instead of changing the monitor res.

Is the any way to prioritize 1200p or prevent Windows from ever using 2304x1440 without side-effects? I tried CRU and a simple detailed res with the default 1200p setting fixes this problem, but in turn breaks youtube when using hardware acceleration
 
Last edited:
spacediver said:
Ah, I see, I was wondering how you derived that pattern. Interesting approach. The best way to test your geometry is to load up a pattern that you know is linear (in pixel space), and ensure that it's linear in physical screen space. That way you can eliminate bad geometry as a culprit here.

I may do a similar test as you did when I have time. If I do, I'll definitely report back here. But if your findings are accurate, then it would make sense to use your custom crosshatch pattern to adjust dynamic convergence!

One thing though: it may be the case that your findings may not generalize to all rows and columns. For example, row E may be diff from row A, and column 8 may be different from column 1.


edit: hmm, hang on. How do you know that you misaligned everything equally? Without that knowledge, you can't be sure where the "locus" of the convergence effect is (unless I'm missing something here, which is very likely :p )
Click to expand...

When I show a linear crosshatch pattern (pixel space) and measure it in physical space, I can conclude my monitor's geometry is linear. So after tweaking my Dynamic Convergence non-linear pattern and checking each and every point, I also conclude that the points to adjust convergence in WinDAS are not on a linear grid.

The way I've created the non-linear Dynamic Convergence crosshatch pattern is as follows:
1. Each point where horizontal and vertical lines meet is a grid point in the WinDAS Dynamic Convergence procedure.
2. The lines are placed as such that the convergence has the maximum offset at the point of where they meet.
3. In addition to this is that the left and right adjacent lines of a WinDAS convergence point have an equal offset on both sides. With the one exception for column 02 and column 14, these columns affect column 01 and 15 respectively way more.
4. Vertical lines are placed to the closest 16 pixels, while horizontal lines adhere to the closest 8 pixels.

Note: physical measurements were done by me using a magnifying glass while setting VAMP_xx of one point in WinDAS for vertical lines and HAMP_xx of one point in WinDAS for horizontal lines to either 0 or 255 based on what had the maximum convergence offset visually.

I hope I've explained everything clearly and all this makes sense :)

One thing that would really help to solidify this conclusion is when someone else tests my crosshatch pattern on their FW900 monitor, to see if the points match. You can download the pattern here
Although this could also mean that Dynamic Convergence grids vary from monitor to monitor.
 
Last edited:
Mokkat said:
Specific question:

I'm using the standard 1920x1200@85hz res. I started playing Bioshock 2 and want to keep using this res, but somehow this game insists that the monitor must run at the highest res of 2304x1440@80hz. When 1200p is selected, it upscales instead of changing the monitor res.

Is the any way to prioritize 1200p or prevent Windows from ever using 2304x1440 without side-effects? I tried CRU and a simple detailed res with the default 1200p setting fixes this problem, but in turn breaks youtube when using hardware acceleration
Click to expand...

Sounds like scaling is enabled. What kind of video card you using? I know how to disable scaling on nVidia graphics cards but not AMD.
 
Mokkat said:
Specific question:

I'm using the standard 1920x1200@85hz res. I started playing Bioshock 2 and want to keep using this res, but somehow this game insists that the monitor must run at the highest res of 2304x1440@80hz. When 1200p is selected, it upscales instead of changing the monitor res.

Is the any way to prioritize 1200p or prevent Windows from ever using 2304x1440 without side-effects? I tried CRU and a simple detailed res with the default 1200p setting fixes this problem, but in turn breaks youtube when using hardware acceleration
Click to expand...

Why would you want to use 1920x1200@85 over 2304x1440@80? I can see why you'd use 1920x1200@96, but 1920x1200@85 seems a bit pointless.
 
jbltecnicspro said:
Sounds like scaling is enabled. What kind of video card you using? I know how to disable scaling on nVidia graphics cards but not AMD.
Click to expand...

Scaling is disabled in Catalyst.

I can only get the monitor to properly change resolution by setting it to a 4:3 res too large to fit into 2304x1440.

rabidz7 said:
Why would you want to use 1920x1200@85 over 2304x1440@80? I can see why you'd use 1920x1200@96, but 1920x1200@85 seems a bit pointless.
Click to expand...

Started with 96hz since that's what I run my X-Star at, but I settled on 85hz later because of tube longevity speculation, because it still feels plenty responsive and because I was noticing slight noise in the corner areas at 96hz. So I thought I'd fix the browser hardware acceleration problem while I was at it.

I've settled on full res for now in Bioshock 2 because a 7950 can handle it fine, but I prefer 1200p to 2304x1440 in Windows on a 22.5" display and I'm sure I'll want to run newer games at 1200p for performance reasons.
 
Last edited:
n2k said:
I hope I've explained everything clearly and all this makes sense :)
Click to expand...

Yep, I think I understand now. Good work, indeed will be very interesting to see if other tubes have the same pattern. If so, then it makes sense that we should use this more accurate grid pattern than a standard linear one.
 
rabidz7 said:
Why would you want to use 1920x1200@85 over 2304x1440@80? I can see why you'd use 1920x1200@96, but 1920x1200@85 seems a bit pointless.
Click to expand...

1920x1200@ 85 is the prime mode of this monitor. Not only are the geometry and convergence procedures designed to optimize for this mode, but in general, a lower scanning frequency will result in a higher quality image. Plus, it probably will be healthier for the tube - higher scanning frequencies are likely associated with higher temperatures, and this might increase "wear".
 
etienne51 said:
About patches, I worked on a patch for WinCATS, it unlocks it, but I still get a popup saying something weird about the configuration. It asks me to check the configuration everytime I try to do something useful. I never made patches before, so I probably can't do any better.
Anyway, I have no way to test WinCATS, and even if I could, I don't know if I would. This software looks weird, and my monitor is already well calibrated enough.
Click to expand...

nice detective work :p

yea, everything I understand about WinCATS suggests it's useless beyond what WinDAS can do.
 
spacediver said:
1920x1200@ 85 is the prime mode of this monitor. Not only are the geometry and convergence procedures designed to optimize for this mode, but in general, a lower scanning frequency will result in a higher quality image. Plus, it probably will be healthier for the tube - higher scanning frequencies are likely associated with higher temperatures, and this might increase "wear".
Click to expand...

What do you mean by "lower scanning frequency"? A refresh rate? Or is that different?
 
I meant the horizontal scanning frequency, which is the number of raster lines the gun scans per second. At any given horizontal scanning frequency, the vertical refresh rate depends on how many lines need to be scanned (which is primarily determined by the vertical resolution).

In order for the beam to switch directions, the electric field needs to be modulated across the deflection coils. The idea, I think, is that the more frequently this signal is modulated (i.e. higher scanning frequency), the hotter things run. How much, and whether it's significant, is unknown to me.
 
Mokkat said:
Specific question:

I'm using the standard 1920x1200@85hz res. I started playing Bioshock 2 and want to keep using this res, but somehow this game insists that the monitor must run at the highest res of 2304x1440@80hz. When 1200p is selected, it upscales instead of changing the monitor res.

Is the any way to prioritize 1200p or prevent Windows from ever using 2304x1440 without side-effects? I tried CRU and a simple detailed res with the default 1200p setting fixes this problem, but in turn breaks youtube when using hardware acceleration
Click to expand...

I had the same issue for the longest time. Because windows sees the maximum resolution as the recommended, it will always scale the game to 2304x1440 which is completely pointless.
I always made sure to set the refresh rate to 85 in games, either via ingame menu (sadly, not many games have that feature) or via a config file the game uses.
I did also create a custom resolution which is only different by a few pixels from, e.g. 1920x1200 and has only one refresh rate associated to it (and which isn't 80Hz so it doesn't default to 2304x1440 scaling.

Fun fact: nvidia has a feature in their driver which forces the highest available refresh rate for each resolution which makes this an non-issue. But of course AMD doesn't seem to care about features like that :rolleyes:
 
you could probably edit the display.inf file to remove reference to the 2304 resolution. That way, your system might interpret 1920 as the highest.
 
spacediver said:
you could probably edit the display.inf file to remove reference to the 2304 resolution. That way, your system might interpret 1920 as the highest.
Click to expand...

You can just download CRU and delete/add resolutions as you please. The first one your "detailed resolutions" will be interpreted as your recommended (or native) resolution.
 
Can someone tell me what are the rebrands of GDM F520 ?
I'm finding conflicting information, some saying that it's the IBM P275 and some say that's the CPD-G520 rebrand.
 
dunno if the F520 had any rebrands. The IBM P275 was one of the rebrands of the CPD-G520 model you mentioned (I own a P275 and it's definitely not a GDM class monitor).
 
spacediver said:
dunno if the F520 had any rebrands. The IBM P275 was one of the rebrands of the CPD-G520 model you mentioned (I own a P275 and it's definitely not a GDM class monitor).
Click to expand...

Ah, okay...
BTW I mentioned GDM in my first post too, not CPD.

Which year was FW900 produced till? I know that GDM F520 was in production until 2004.
 
Klarion said:
Ah, okay...
BTW I mentioned GDM in my first post too, not CPD.

Which year was FW900 produced till? I know that GDM F520 was in production until 2004.
Click to expand...

Sony ended CRT computer monitor production in 2004. (Except Artisan, which got another year I think.)

No F520 rebrands I know of.

Fear was in the air back then that Sony would cease production...then Sony stopped accepting orders from their website...then the announcement was made...and queue in over a decade and counting of display frustration...
 
You must log in or register to reply here.
Back
Top