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

Recapping isn't the magic recipe that solves everything you know. With these symptoms I'd think about a problem with a digital component since the color differences come in patches, it's not smooth. Also color related, so probably on the A board. ;)

To answer your question I don't think there's a recap kit for any screen, but the list of components is included in the service manual. I successfully made a recap using mostly:
- Panasonic FR and a few Rubycon YXJ for voltages of 50 and below. Rubycon ZLH for the big two 2200 µF/25V ones
- United chemicon KZN, Rubycon YXF-YXJ for 100V ones
- Some Panasonic EE, Nichicon PW if no other option for voltages of 250 and above. Note that for very low capacitances (1-2.2µF/250V), Panasonic NHG series have a much lover ESR than the Nichi PW with similar endurances. I also replaced one electrolytic 2.2µF/250V on the horizontal deflection circuit with a polypropylene film one. The blueprint didn't specify the capacitor type, there were 2 hole sets with enough room for a film capacitor on the board, and I couldn't find a new electrolytic as good as the old one. In theory can only improve the signal but the difference is difficult to assess.
- United chemicon LXS for the huge 330µF/450V one, but the legs don't fit, you have to bend them carefully. The original one on the board has very specific snap-in legs that can't be found anywhere.
 
Strat_84 said:
And here is some porn for FW900 lovers, I actually got the beast set up and running

View attachment 10479

Messing with the focus knobs wasn't that a good idea as I feared, I don't think it really improved the display but it was a real hassle to reach back the original sharpness after changing it.
The sharpness problems specific to some colors seem to be related to some subsystems not able to suck up as much current as they need, proper settings are much more useful to improve this than the knobs. (I'd say, VDC settings, alignment procedures and White point balance)

Anyway, if you touch the focus pots here are a few tips I discovered:
- don't touch them unless you really need to, normally that must be only when the flyback is replaced or the factory setup of the knobs was disturbed.
- a magnifying glass is really helpful to find the exact spot between "too much" and "not enough"
- turning the knobs to unstick the paint/glue whatever isn't enough, that thing is the devil and will still keep a little pressure on the knob, thus may change the setting with time. Unstick it, and then cut the part remaining in contact with the knob with some cutting pliers, it's safer. (to do without power supply connected and tube discharged of course)
- run the focus procedure in Windas. Despite not being always obvious, it does optimize focus. First with the next/previous screens (I would bet on a slight vertical or horizontal improvement depending of the option chosen), second maybe are there some automatic optimisations behind the scene after this, I think I've seen the focus more even along the entire screen surface after the procedure is done and finalized.
Click to expand...

Nice one Strat. And thanks for sharing your observations on the focus knob and WinDAS.
 
Crap, I stepped into it. :facepalm:

edit: Thinking again about Sprak's problem, surely a digital issue but not necessarily linked to colors. Do you see the difference between the two pictures ? When the green goes away, everything is darker. Including the raster which is green on the 1st picture. Might also have something to do with G2.
I was also assuming the screen was supposed to be a plain blue one, but was I right ? Or is it some kind of testing pattern with different blues and the squares are actually normal ?
 
Last edited:
yea it's a strange test pattern to use. Why not use a a better test pattern and take a pic with and without OSD. Would also be a good idea to have a substantial portion of the image with a white field, so that we can be sure that any artifacts are not due to camera white balancing differently or using a different automatic exposure when the OSD screen is up. So make a test pattern that has left half blue and right half white.
 
Damn I can't believe people are still using these! I remember wanting one badly to replace my Dell P1230 (? I think that was the model? The 21" 2048x1536 trinitron). I just bought the Dell 3216 32" 3840x2160 LCD and am happy that the screen resolution has finally eclipsed these, and the size is glorious.

Brings back memories though ...
 
Strat_84 said:
Crap, I stepped into it. :facepalm:

edit: Thinking again about Sprak's problem, surely a digital issue but not necessarily linked to colors. Do you see the difference between the two pictures ? When the green goes away, everything is darker. Including the raster which is green on the 1st picture. Might also have something to do with G2.
I was also assuming the screen was supposed to be a plain blue one, but was I right ? Or is it some kind of testing pattern with different blues and the squares are actually normal ?
Click to expand...

It might be a G2 issue because I did mess around with G2 with windas. I'll try to load the original file into the FW900 this weekend. The pattern was just a color gradients from one of monitor test application.
 
repoman0 said:
Damn I can't believe people are still using these! I remember wanting one badly to replace my Dell P1230 (? I think that was the model? The 21" 2048x1536 trinitron). I just bought the Dell 3216 32" 3840x2160 LCD and am happy that the screen resolution has finally eclipsed these, and the size is glorious.

Brings back memories though ...
Click to expand...
I had to buy a LCD for a relative recently, a 2560x1440 IPS 144hz screen. Despite being a quite acceptable display, it still has flaws. Motion blur problems, view angles still degrading, IPS glow and a slight input lag. Not counting an annoying bug like the screen understanding 2568x1440 sometimes when asked 2560x1440, leading to display artefacts. Really disappointing regarding the price if you compare to a 10+ years old CRT IMO.

sprak74 said:
It might be a G2 issue because I did mess around with G2 with windas. I'll try to load the original file into the FW900 this weekend. The pattern was just a color gradients from one of monitor test application.
Click to expand...
Ok, so forget about the digital thing, if having patches of different colors is normal. :)

From what I've seen, you can't really change the G2 and get away with it, you absolutely have to go through the entire white point procedure with a colorimeter to ensure all parameters are correct and consistent.
 
Last edited:
Few questions:

1. Is there any news on Analogix? Can't wait till I upgrade to Pascal GPU.
2. To my knowledge WPB should not only set proper white balance, but also RGBCMY coordinates properly. If first step is true and I can clearly and easily switch preset white point in OSD, then after measurement there is always green and magenta hue switch. RCYB are almost spot on, but green and magenta are somewhat missing the point. I have tried it on three different FW900 with AG and without AG and surprisingly they all behave exactly the same.
3. If I understand correctly, pro settings in OSD let us correct RGB channel for gain and bias, where gain is white and bias is neutral gray setting (128,128,128). Is there any option to correct blacks? After adjusting gain and bias I can easily achieve great grayscale accuracy somewhere between 35 to 100% luminance, but darks are quite off.
4. Why is gamma so high after WPB? It's above charts until properly profiled with argryll/displayCAL. My point is - I always used to profile anyway, but lately I am learning more and more about calibration and start to wonder. It's said commonly that gamma 2.2 is natural for CRT displays so why would mine be around 3.5 after WPB? I could be doubtful if that was only my FW900, but I have done WPB on few of them and result is always the same.
 
igsux3 said:
Few questions:

.
Click to expand...

Have you read my WinDAS WPB guide? Some of your questions are addressed there. Also note that you cannot change the chromaticity of the RGB primaries with WinDAS.

The claim that the natural gamma for CRTs is 2.2 is misleading, as the gamma changes dramatically with different G2 levels. If you set the G2 such that the blacks are super dark, you end up with a very high gamma.
 
Yeah I'm holding up on building my new PC till these adapters arrive. I really like this monitor and would rather not get another one till there is something better.

Also how often do you guys suggest we do a WinDas calibration. It's been about a year for me but the monitor still looks good.
 
I think a good general rule might be when the black levels drift to a point where things start to look washed out. For me that's about a year I think.
 
Does anyone know where I can buy a degaussing coil from in the UK? I cant seem to find one online apart from shipping from the US.
 
spacediver said:
Have you read my WinDAS WPB guide? Some of your questions are addressed there. Also note that you cannot change the chromaticity of the RGB primaries with WinDAS.

The claim that the natural gamma for CRTs is 2.2 is misleading, as the gamma changes dramatically with different G2 levels. If you set the G2 such that the blacks are super dark, you end up with a very high gamma.
Click to expand...

Yeah I used it many times to go through WPB procedure but never read surroundings which I can now clearly see was a mistake. Thanks a lot.

So now I understand how things work and have some additional questions:

1. to which gray point BIAS settings are related? Is this 30% gray?
2. I have my personal DTP94 and Spectracal C6 at workplace. Would you recommend the latter with spectral CRT correction?
3. Any tips on near-black scale? I am quite happy with my results, but wonder if it could be done better.

SDVz4R9.png
 
igsux3 said:
1. to which gray point BIAS settings are related? Is this 30% gray?
Click to expand...

When you adjust the controls of a CRT, you are actually changing parameters of the triodes themselves (e.g. the grids, such as G1, G2), or of circuits that modulate current before it reaches the triodes (e.g. preamplifiers).

Take a look at this paper.

Notice that when you adjust Bias(aka offset, cutoff) the luminance function shifts up/down (fig 2), and when you adjust Gain, the steepness of the function changes (fig 1). Further, notice that when you adjust the Bias, the change in luminance is more pronounced in the low end of the function, while the opposite is true for Gain. Thus, if you are fine tuning your overall luminance function (which will be produced by choosing the appropriate combination of Bias and Gain values), it makes sense to use a low luminance pattern when adjusting Bias, and a high luminance function when adjusting Gain. That way, you'll be able to quite clearly see the effects of adjusting these parameters. Traditionally, the 30% pattern is used for the low end, and 80 or 100% is used for the high end.

igsux3 said:
2. I have my personal DTP94 and Spectracal C6 at workplace. Would you recommend the latter with spectral CRT correction?
Click to expand...

I'm not sure if the C6 works with HFCR/Argyll, but I could be wrong. I believe the C6 is the same as the i1 display pro, but with added calibration tables. If you can get the C6 to work, I'd go with that, since it's a faster instrument and can read lower luminance levels than the DTP-94.

igsux3 said:
3. Any tips on near-black scale? I am quite happy with my results, but wonder if it could be done better.
Click to expand...

First off, don't use CCT to report performance. Use Delta-E's. The grayscale tracking in the near black region will largely depend on how you implement the Argyll fine tuning, and how sensitive your instrument is at the low end. Sometimes you can achieve inky blacks by sacrificing a bit of color accuracy at the very low end. This may not be too big of an issue.
 
Last edited:
Picked up a HP p1230 Diamondtron 22' inch monitor the other day and it looks really nice and very comparable to the Sony. Would I be able to calibrate that using the WinDas guide?
 
No, but I believe there's a factory mode that allows you to get "deep inside" the diamondtron (I believe your monitor is a rebranded 2070sb).

Here's a link to the service manual (you need to enter an email address to get a link, sorry, too lazy to find a diff host).

Instructions on how to enter and exit factory mode start at section 1.8.3.

Warning, it's quite a bit more involved than WinDAS. Looks like you have to open up the chassis and measure voltages manually.
 
Here's a silly question:

I have some spare Dell p990 monitors, and my sister is getting rid of a Hi-Scan Trinitron 30" tv. Is it even remotely feasible to take most of the electronics from the p990, and put them in the hi-scan set to make a multi-scan TV? I imagine you'd need to keep the boards that directly control the cathode and deflection circuit, as well as the power supply. But the boards from the P990 that tell the deflection circuit what to do could maybe be modded in?
 
EnhancedInterrogator said:
Here's a silly question:

I have some spare Dell p990 monitors, and my sister is getting rid of a Hi-Scan Trinitron 30" tv. Is it even remotely feasible to take most of the electronics from the p990, and put them in the hi-scan set to make a multi-scan TV? I imagine you'd need to keep the boards that directly control the cathode and deflection circuit, as well as the power supply. But the boards from the P990 that tell the deflection circuit what to do could maybe be modded in?
Click to expand...

I am interested in how this plays out.
 
EnhancedInterrogator said:
Here's a silly question:

I have some spare Dell p990 monitors, and my sister is getting rid of a Hi-Scan Trinitron 30" tv. Is it even remotely feasible to take most of the electronics from the p990, and put them in the hi-scan set to make a multi-scan TV? I imagine you'd need to keep the boards that directly control the cathode and deflection circuit, as well as the power supply. But the boards from the P990 that tell the deflection circuit what to do could maybe be modded in?
Click to expand...
I doubt it's even remotely possible. You can't mix the boards of both devices because the scan ranges are different, the power requirements are different, one component may be well calculated for one device and undersized for the other. I doubt the different functions are even broken up on the same boards or the lines connecting the boards are the same.
I'm seeing it currently with an open GDM5410. Basically it's a 21" 4:3 FW900, same production period, very similar electronics with almost the same functions on the same boards but the components aren't always exactly the same, neither is the actual layout of the boards.

You might try of course, but I'd think it would require both electronic schematics, deep knowledge and quite some engineering before doing anything, for the sake of safety.
 
Yeah, that's basically what I was wondering. If Sony kept a similar board layout across their TV's and monitors, then maybe you could swap in certain boards that handled the same function. But I knew that was highly unlikely.

I was just curious to hear from people that know a little more about the internals than I do.
 
On the adapter topic, I tried this one for a 21" (didn't need more than a 1600x1200@85hz ability): https://www.amazon.com/DisplayPort-Displayport-Connecting-Projectors-Displays/dp/B00LM3GF76

Avoid, this is a piece of shit. Couldn't exceed 1280x1024@85hz, and I saw something looking like interpolation at very low resolution. Characters borders were not as accurately cut as with a VGA-VGA cable.

Also, funny story with the GDM5410. The red color was completely gone. I found a resistor cut on the A board (infinite resistance, but no visual sign of failure), as well as a diode that let the current pass through both ways next to it. I replaced them, now the red is back. Even more, the display is red, even the black is red. it turns out some settings are very high for that color, looks like either at least one of the components was faulty from the start, or the color restoration feature managed to compensate one slowly failing. Another white point balance procedure to do now. :)
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
spacediver said:
I'm not sure if the C6 works with HFCR/Argyll, but I could be wrong. I believe the C6 is the same as the i1 display pro, but with added calibration tables. If you can get the C6 to work, I'd go with that, since it's a faster instrument and can read lower luminance levels than the DTP-94.
Click to expand...
Took it from work and can clearly confitrm - C6 works great and is pretty quick device. At least twice as fast as my DTP-94. Should I always use CRT Profile for HCFR and Agryll? I mean one called "CRT (Hitachi CM2112MET, Diamond View 1772ie)".

spacediver said:
First off, don't use CCT to report performance. Use Delta-E's. The grayscale tracking in the near black region will largely depend on how you implement the Argyll fine tuning, and how sensitive your instrument is at the low end. Sometimes you can achieve inky blacks by sacrificing a bit of color accuracy at the very low end. This may not be too big of an issue.
Click to expand...

Just wondering - HCFR shows proper dE and I can tell that they are real:

wOHuXWA.png


lROD99a.png



But DisplayCal shows some unreal low values:

ljJtuMV.png


Even CIE clearly shows that such low values are unreal:

WB3wxJs.png


I have tried to switch Delta E formula for results to CIE 1976, 1994 and 2000 but still they are all impossibly low.


I am about to go through entire WPB procedure again and report back, but I doubt that it would change DisplayCAL behavior.
 
The delta E's from the HCFR readout look great. I have never used DisplayCal, and I have no idea what those columns indicate - all I see are a bunch of numbers.

As for the CIE chart you showed, I'm now sure how that's relevant - you can see the primaries and the saturation tracking, but as far as I can tell there's no information there about white point balance tracking across the gray scale (and the delta E's we're interested in are the white point balance ones)
 
From what I've read, this is about using one software for profiling and veryfing. So profiling done with DisplayCAL would always show very small dE in verification. That's why one should cross-check their profile with either different software or different device. Not really sure about it, but makes some sense.

Anyway you helped a lot with all your knowledge!

By the way I got old P1130 to experiment with. Someone made huge mess with this unit and probably thought it's useless now, so I got it for free with hope to restore its functionality.
First of all I would like to go through geometry adjustments.

My questions are:
  • do you strictly obey timings when adjusting for max / mid / low frequency? Like in this example:
    1gAa6vX.png
  • does this "alignment for factory preset" fill itself after adjusting max / mid/ low / prime modes?
    aGSDZeH.png
  • do you strictly obey geometry guidelines? Like when it says "set v_size to 291mm" even if screen itself is 300mm tall?
  • what kind of measure do you use? Whatever measure I use I find paralax error just too annoying. I'm thinking about printing crosshatch on foil and putting it on screen

Hope to hear from you spacediver. I'm still learning a lot about mastering CRT beauties.
 
igsux3 said:
From what I've read, this is about using one software for profiling and veryfing. So profiling done with DisplayCAL would always show very small dE in verification. That's why one should cross-check their profile with either different software or different device. Not really sure about it, but makes some sense.
Click to expand...

That doesn't make much sense to me. Why would you want to profile your display? Are you using a 3DLUT or color aware software to transform the color space or across different devices? If not, just stick to a WPB and, possibly, a gamma adjustment. Why are you even using DispCal?

igsux3 said:
My questions are:
  • do you strictly obey timings when adjusting for max / mid / low frequency? Like in this example:
  • do you strictly obey geometry guidelines? Like when it says "set v_size to 291mm" even if screen itself is 300mm tall?
  • what kind of measure do you use? Whatever measure I use I find paralax error just too annoying. I'm thinking about printing crosshatch on foil and putting it on screen
Click to expand...

Haven't done geometry stuff for a while, but JBL had some useful stuff a few pages back. And yes, stick to the sizes it says - you don't want to fill the entire screen in many cases.

Yes, measuring is a pain in the ass due to parallax error. I try to keep my line of sight perfectly perpendicular to the point on the screen I'm measuring. It's tedious though.
 
About geometry, what I did for the "set to x mm size" steps is cutting some rulers in thick cardboard with marks for the said sizes. Then, for example when setting the horizontal size, set the ruler on the screen, and look at the mark from the top of the screen, make sure that this mark is in the extension of the border of the display.
The hardest thing is actually to make the cardboard ruler according to proper standards. I found a tape measure which actually was a couple of centimeters wrong for a length of 50cm, there also could be differences of several millimeters between rigid rulers that were supposed to have the same lengths.

"alignment for factory preset" fills itself, probably according to max / mid / low / prime modes adjustments, but it's not always optimal. I found myself forced to tweak a bit V and H size/position almost everytime.
 
Strat_84 said:
Then, for example when setting the horizontal size, set the ruler on the screen, and look at the mark from the top of the screen, make sure that this mark is in the extension of the border of the display.
Click to expand...

What do you mean by "in the extension of the border of the display"? I'm having trouble visualizing what you mean.


Strat_84 said:
The hardest thing is actually to make the cardboard ruler according to proper standards. I found a tape measure which actually was a couple of centimeters wrong for a length of 50cm, there also could be differences of several millimeters between rigid rulers that were supposed to have the same lengths.
Click to expand...

That's very strange.

Also, be sure to read JBL's recent post on geometry adjustments before you start messing around in WinDAS geometry. There's some important information. Should be within the last month or two in this thread.
 
spacediver said:
What do you mean by "in the extension of the border of the display"? I'm having trouble visualizing what you mean.
Click to expand...
Imagine you have your cardboard ruler positionned on the display. The "regular" way to avoid parallax errors is to be exactly in front of the area of the display your are assessing, i.e. the border of the display area and the side of the ruler. But that's very hard to do accurately and you might see both borders matching while they do not because you're not exactly in front but a bit on right or a bit on left. But you can't know it exactly.

What I did is to do this with the eyes nearly parallel to the display surface instead of looking perpendicularly. Then move the ruler so that one end is aligned with the entire side of the display area. If you're properly placed when looking, the side of the ruler will be matching the entire side of the display from the top to the bottom. If you're not you'll see the ruler's side matching the display's border, but only in the middle, top or bottom of the display, not all 3 areas.
 
Ah, you're using the perspective distortion of the human eye here, if I'm not mistaken. But in order to do that, you need to first ensure that the ruler is parallel to the display border, which I suppose you can do before you align your eyes parallel to the display.

To solve the problem of looking perpendicular to screen, I used a pencil as a "line of sight guide", where i pressed the pencil against the screen such that the flat end (where the eraser usually is), was flush against the screen at the point of interest. That way, I knew that the pencil was perpendicular to the screen, and I just looked down the length of the pencil (taking care not to poke my eye out with the lead tip), thus ensuring that my line of sight was perpendicular.
 
spacediver said:
That doesn't make much sense to me. Why would you want to profile your display? Are you using a 3DLUT or color aware software to transform the color space or across different devices? If not, just stick to a WPB and, possibly, a gamma adjustment. Why are you even using DispCal?
Click to expand...

Just for the same reason you use Argryll - linearizing gamma to 2.2 and adding some grayscale corrections. DisplayCAL is just GUI to Argryll, so why not use it?

spacediver said:
and yes, you should follow the timings exactly as instructed.
Click to expand...

Yeah I wish. Problem is that horizontal values are represented in uSec rather than pixels. I have only found calculators like this: http://www.epanorama.net/faq/vga2rgb/calc.html that convert pixels to uSec, but not the other way.
 
igsux3 said:
Just for the same reason you use Argryll - linearizing gamma to 2.2 and adding some grayscale corrections. DisplayCAL is just GUI to Argryll, so why not use it?
Click to expand...

Ah ok, using it for gamma correction makes sense. But the whole profiling/verification thing through me for a loop.


igsux3 said:
Yeah I wish. Problem is that horizontal values are represented in uSec rather than pixels. I have only found calculators like this: http://www.epanorama.net/faq/vga2rgb/calc.html that convert pixels to uSec, but not the other way.
Click to expand...

This was discussed very recently - look at the posts on Halloween of this year.
 
spacediver said:
This was discussed very recently - look at the posts on Halloween of this year.
Click to expand...

I know I might be tired and unable to understand easy things, but in this post: https://hardforum.com/threads/24-wi...ived-comments.952788/page-252#post-1040440823 you mention that things marked in red are the ones that you inputted.
I understand that they all match when compared to windas guides, but hell, how did you get pixel values?

For example you mention: " Blanking pixels = Total pixels - Active pixels" Okay, it's easy to calculate blanking pixels, but not when they are 32.258 uSec and 25.645 uSec.

-- edit --

okay got total horizontal pixels from proportions
 
Last edited:
I don't remember exactly when but there was an excel sheet with all the values in pixels attached here recently.

Also about the resolutions input, they won't always be recognized as the proper mode as there might be close custom resolutions stored in the monitor's memory. To fix this (most of the time), hold the reset button for several seconds, the screen will blink dark a couple of times. This will erase all the custom settings and custom resolutions stored in memory.
 
igsux3 said:
For example you mention: " Blanking pixels = Total pixels - Active pixels" Okay, it's easy to calculate blanking pixels, but not when they are 32.258 uSec and 25.645 uSec.
Click to expand...


It's been a long time since I immersed myself in this, but you should be able to convert back and forth between pixels and time without guessing.


igsux3 said:
okay got total horizontal pixels from proportions
Click to expand...

What do you mean by this?
 
Sorry for all the hassle, I'm really tired after this work week.

By proportions I mean, if horizontal active is 1770 px = 5.255 uSec and horizontal total is x = 7.886 uSec then x = (1770*7.886)/5.255 and that equals to 2656. This way I got the values I needed.

My first thought was like 'damn I have to use values that are 10^-6, that is not going to be precise enough", like 1770*0.000005255. Importantly I got all P1130 timings required by WinDAS. I guess this is to be contiuned tomorrow, it's 11:30 pm in my time zone.

Oh by the way, I have ordered polarizing foil for my FW900 (2 pcs, glued) and I guess I will try to mount them this Saturday. Will post before / after results for sure.
 
ah gotcha. Post the timings that WinDAS are requesting and that you're having problems syncing with (assuming it's not the same one from the screenshot you posted earlier, the 1770x1440 one). If I have time later this evening, I'll try to do the math and see what values I get.

btw, what are you using to create your custom resolution timings - nvidia control panel?

hopefully u've not already gone to sleep :p
 
You must log in or register to reply here.
Back
Top