Mod a CRT to increase its TVL

The place for all discussion on gaming hardware
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

ElBartoMe wrote: You can see the that the amplifier is just not able to follow the signal after about 384 lines. It should be a nice square wave anyway but what we get here is sinusoidal.

Then I put on the TVL card and compared them with and without the mod. As I already expected after having measured the output of the amplifier there is no change to be seen in the TVL before and after the mod.


I think what you guys have done here is great, and commendable. Nothing but respect for the OP for the idea and research and for those who have implemented/tested and have seen improved pictures. That said, this goes back to what we were arguing on the first couple pages, specifically, the original SOTN and full monochome comparisons. If you compare the pictures, there is no visible increase in horizontal granularity/precision, thus no increase in TVL. Vertical granularity however, is an entirely different story. Its very clear that the spot size has decreased and the the larger and more defined blank lines make it easy to see.

As you are hypothesizing, it seems that the HO circuitry simply can not resolve greater horizontal precision than what we are seeing, and even if it could, you are left with the question of if the shadow mask/AG and phosphor layout will allow for it.

https://imgbb.com/pnZBPGQ
https://imgbb.com/Rb4qF3C
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

At first, spot size/beam focus was considered the limitation for TVL, when LukeEvansSimon stepped on the stage with his confident demeanor. The OP now talks about other limitations, but all of that was edited in later.

Then, phosphor pitch was considered the limitation, for example:
mikejmoffitt wrote:Improving beam focus will never overcome the phosphor pitch's limitations.
mikejmoffitt wrote:I think the confusion surrounding the description of the mod as making it "high TVL" might be alleviated a bit if it's phrased more specifically "decreasing spot size to increase TVL such that the pitch of the tube becomes the limiting factor".
And now, we're at RGB signal amplifiers being the limitation.

Remarkable, it emphasizes how much TVL are an inherent quality of the device's configuration. There's a reason why CRTs were deemed the most complex piece of tech until the invention of the IC.
One would expect at least, that, by and large, engineers/manufacturers would have matched these various factors well already, and spot size was probably deliberately not minimized, for example to reduce flicker with interlaced footage.
Josh128 wrote:That said, this goes back to what we were arguing on the first couple pages, specifically, the original SOTN and full monochome comparisons. If you compare the pictures, there is no visible increase in horizontal granularity/precision, thus no increase in TVL.
Two things; first: you couldn't have counted TVL based on these pics. The PSX output there is not suitable at all to do that for various reasons. All you can really infer based on the SOTN photos is that the CRT at that time had a TVL of at least around 192 before and after the mod (256 dots per scanline during gameplay)—"at least" being the important bit. Gratz.

Second, "Granularity (also called graininess), the condition of existing in granules or grains, refers to the extent to which a material or system is composed of distinguishable pieces.". Not an adequate term to use when talking analog signals and TVL. Everything moves on a continuous plane here.
Josh128 wrote:Vertical granularity however, is an entirely different story. Its very clear that the spot size has decreased and the the larger and more defined blank lines make it easy to see.
Correct, if anything, the proposed mod, when spot size is decreased, actually seems to be able to increase vertical resolution of the CRT, not horizontal resolution (TVL). Because on the horizontal axis, the beam is simply doing one continuous streak, so how should a decreased spot size even increase resolution there?


IIRC I'm not the first to express this view, but if this had been advertised simply as "improving the picture" or "mod to increase the scanline effect" or similar instead of "increasing TVL"/variations thereof, there would have been much less confusion.
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

Blacksheep wrote:
Josh128 wrote:That said, this goes back to what we were arguing on the first couple pages, specifically, the original SOTN and full monochome comparisons. If you compare the pictures, there is no visible increase in horizontal granularity/precision, thus no increase in TVL.
Two things; first: you couldn't have counted TVL based on these pics. The PSX output there is not suitable at all to do that for various reasons. All you can really infer based on the SOTN photos is that the CRT at that time had a TVL of at least around 192 before and after the mod (256 dots per scanline during gameplay)—"at least" being the important bit. Gratz.

Second, "Granularity (also called graininess), the condition of existing in granules or grains, refers to the extent to which a material or system is composed of distinguishable pieces.". Not an adequate term to use when talking analog signals and TVL. Everything moves on a continuous plane here.
I didnt count "TVL" on those pics, what I counted was the number of spots / dots per a given distance, and there is no difference. The limitation appears to be in the mask/grille. No difference in number of visible dots means no difference in TVL, which is what I said way back. Whether its a PSX or a PC, the number of those dots on a horizontal line will not change on that TV.

Again, the granularity I speak of doesnt depend on the source, and the fact that the signal is analog has no bearing at all on my use of the term. Its "granular" because you can easily count the horizontal dots and see the spaces between them, unlike higher TVL monitors such as SVGA/XGA CRTs, HD CRTs, and BVM style monitors.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

Josh128 wrote:I didnt count "TVL" on those pics, what I counted was the number of spots / dots per a given distance, and there is no difference. The limitation appears to be in the mask/grille. No difference in number of visible dots means no difference in TVL, which is what I said way back.
And I didn't claim that you counted TVL, I wrote that you couldn't have done that (the latter excludes the former anyway, so in this sense there was no need for you to explicitly state that you didn't count TVL). Obviously the mod does not change mask/grille pitch. But what you counted does not allow you to make any meaningful statement about the TVL of the CRT—"no difference in TVL" is simply impermissible at that point, we don't know based on the available information (the pics you wrote you looked at).
If you make the left half of your screen white and the right half of your screen black, does your screen then have 2 TVL? Or count the visible triads within the white area and let that be your TVL count? No. (for TVL, the black stripes also contain partially lit/invisible (parts of) phosphor triads) And if you then increase the TVL of your CRT by whatever means, then your "TVL count" will still be the same. You need a signal based on a proper test chart, with a high enough number of dots per line or generated in an analog way with sufficient bandwidth, and the CRT to be calibrated accordingly, to determine TVL. Otherwise you choose not to measure because you say that another factor makes that unnesessary, but that could be a mistake, and in that case a measurement may give surprising results. For example: someone presenting the prospect of increased TVL, someone measuring, seeing that TVL are not meaningfully changed. Happened here.
Josh128 wrote:Its "granular" because you can easily count the horizontal dots and see the spaces between them, unlike higher TVL monitors such as SVGA/XGA CRTs, HD CRTs, and BVM style monitors.
It's not that simple. The phosphor triads can be lit partially with decreasing luminosity towards both the horizontal and the vertical direction. The phosphor triads are not like pixels, they are not granular. I advise you to take a close look at some of ElBartoME's close-ups from awhile back for example. The "spaces" between them do not matter at all, although they also exist with higher TVL monitors such as SVGA/XGA CRTs, HD CRTs, and BVM style monitors.
User avatar
maxtherabbit
Posts: 1763
Joined: Mon Mar 05, 2018 4:03 pm

Re: Mod a CRT to increase its TVL

Post by maxtherabbit »

Josh128 wrote: I counted was the number of spots / dots per a given distance, and there is no difference.
that's not relevant, you could could the dots with the monitor powered off and get the same thing

that's not the limiting factor in horizontal resolution here
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

maxtherabbit wrote:
Josh128 wrote: I counted was the number of spots / dots per a given distance, and there is no difference.
that's not the limiting factor in horizontal resolution here
I didnt say it was the limiting factor here. I said, to ElBartoME---
"As you are hypothesizing, it seems that the HO circuitry simply can not resolve greater horizontal precision than what we are seeing, and even if it could, you are left with the question of if the shadow mask/AG and phosphor layout will allow for it.'
So let's not pretend I said anything about the second part of the above quote not first depending upon the first part, OK?
Horizontal output amplifier/ transistor /circuitry, in agreement with ElBartoME is what I said. Its bandwidth, ie, the speed at which it can switch on and off, IS the limiting factor as long as the TVL is below the theoretical limit of the mask. This mod, from its technical description down to the pictures shown here, does not affect that bandwidth at all. The granularity or pitch of that mask (which is perfectly matched with the phosphor coating ) is absolutely the ultimate horizontal resolution limiting factor of any shadow masked tube. You could build a monitor with electronics capable of 1600x1200 but a mask and pitch of a 27" consumer grade SDTV and you are not going to surpass what the mask allows to strike the tube. The mask and phosphors will act as a low pass filter and will not allow for TVL beyond the they are designed for. I was the first "idiot" on this thread to point that out, and mikejmoffit, who knows a "little bit" about CRTs, later said the same thing.

It takes the 3 electron beams each striking R,G,B phosphors simultaneously to create a white dot. TVL determination only allows for alternating white and black lines, not primary colors, so ts impossible for TVL to surpass the number of RGB triad pairs (because the white and black lines must be equal in width and can together minimally only constitute 6 primary elements/2 triad pairs) available across the width of the screen.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

The triads (and/or three individual luminous red, green and blue openings to the phosphor coating they're composed of) still neither are granules, nor are they granular themselves. Suggesting that is just wrong. =) Take a look at these two close-up photos and it becomes clear:

https://commons.wikimedia.org/wiki/File ... cursor.jpg
https://commons.wikimedia.org/wiki/File ... letext.jpg
Josh128 wrote:I was the first "idiot" on this thread to point that out, and mikejmoffit, who knows a "little bit" about CRTs, later said the same thing.
mikejmoffit was first to note that as being the ultimate limit. See here. Your comments on that topic came later. It takes 5 seconds to verify that.
Josh128 wrote:because the white and black lines must be equal in width
No, not at all. You just pulled that out of the hat—that is not part of the definition of TVL. Please provide a proper source for this.
Derf
Posts: 26
Joined: Sun Sep 13, 2020 12:15 am

Re: Mod a CRT to increase its TVL

Post by Derf »

NewSchoolBoxer wrote:I read the entire thread before jumping in. I stay out of mod threads being a non-modder but I didn't appreciate the Reddit post in crtgaming boasting about boosting the TVL. Reddit knowledge level is far lower than here. That's also disingenuous to compare the spot size on Mega Man using non-PVM to PVM and imply your mod makes the image look like the PVM's with its higher quality phosphors.
Assuming you're talking about my linking thread, I only linked it to point out this interesting thread to give visibility - i didn't "boast" about it boosting TVL, I copied the thread title word for word and gave a summary of the thread. I can't vouch for its effectiveness personally.

I think the rest of your post was addressed more or less, but I don't think OP ever implied this would make your TV look exactly like a PVM. Obviously, there are a lot more factors than just spot size that makes a PVM, but upgrading a consumer TV to have PVM-like circuitry and getting an image closer to a PVM isn't disingenuous at all.

Blacksheep wrote:The triads (and/or three individual luminous red, green and blue openings to the phosphor coating they're composed of) still neither are granules, nor are they granular themselves. Suggesting that is just wrong. =)
I think you're hung up on pedantry. It's clear what he meant with his context clue of "precision". You can call the number of illuminated vertical white lines "granules" as laid out in LukeEvansSimon's post at the bottom of page 2. Less white lines for the same image = higher precision = sharper image = higher granularity.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

Derf wrote:Assuming you're talking about my linking thread, I only linked it to point out this interesting thread to give visibility - i didn't "boast" about it boosting TVL, I copied the thread title word for word and gave a summary of the thread. I can't vouch for its effectiveness personally.
Maybe NewSchoolBoxer meant this reddit thread:
https://www.reddit.com/r/crtgaming/comm ... g_crt_mod/
It was posted by LukeEvansSimon. He boasts about increasing the TVL even before this has been proven.
Derf wrote:I don't think OP ever implied this would make your TV look exactly like a PVM.
Maybe he didn't imply it would make your TV look exactly like a PVM, but he did repeatedly say things like "now limited to 1000. This is a 2x increase in resolvable TVL!", "gamers want that high-TVL look" and "A higher TVL CRT makes the blank scan lines thicker and the illuminated scan lines thinner.".
Derf wrote:It's clear what he meant with his context clue of "precision". You can call the number of illuminated vertical white lines "granules" as laid out in LukeEvansSimon's post at the bottom of page 2. Less white lines for the same image = higher precision = sharper image = higher granularity.
I don't agree, but I'm not going to argue further. It's no use. There's a number of problems with that post, but aside from that, FWIW, LukeEvansSimon does not use the terms "granular", "granularity" or "granule" in it. He even replied to Josh128 by saying: "I think you are thinking of slots in the mask as being analogous to pixels, which is wrong.".


Anyway, if the mod makes the picture look prettier in people's eyes without wearing down the CRT faster, then it is a good mod, and that is commendable.
Last edited by Blacksheep on Fri Feb 26, 2021 12:36 pm, edited 1 time in total.
LukeEvansSimon
Posts: 128
Joined: Mon Aug 17, 2020 7:28 pm

Re: Mod a CRT to increase its TVL

Post by LukeEvansSimon »

ElBartoME wrote:
ElBartoME wrote:I am now leaning towards using the 2SC3782 NPN transistor
The 2SC3782 looks very interesting, but you linked the KSC3503DS instead?
ElBartoME wrote:since the 2SC3782 is sufficiently high voltage and high bandwidth
The bandwidth is very impressive but it seems the breakdown voltage is 200V. The transistors used in my E3431D have a breakdown voltage of 300V. I haven't measured any voltage near that but I assume there must be a reason why they chose that number. I assume it depends on G2. If I crank up G2 the voltage to achieve black levels change too. So I guess in my operation point I have at the moment I don't need that high voltage to achieve black. But still I would feel better if the breakdown voltage of a new transistor is the same. Do you have any suggestion?
Sorry, I copy pasted the wrong part number. I meant to say the 2SC3503, which is what I linked to on mouser.com. It definitely has a breakdown voltage of 300V, and should be a drop in replacement for most consumer grade CRT TV cathode amplifiers. BTW, the chroma signal path may have a bandwidth bottleneck at the pre-amp stage. If your oscope probing for the chroma signal right before the cathode amplifier shows the same high frequency distortion, then the pre-amp is also a bottleneck and would need to be upgraded. Pre-amplifiers are not very demanding as they deal with amplifying a roughly 1 volt peak to peak signal to at most 5 volts peak to peak, which is easy for transistor amps to achieve have a very high bandwidth. However, consumer TVs may have used very low grade pre-amps, so best to be sure and check with the oscope.

If the pre-amp signal is not distorted at high TVL, then just upgrade those cathode transistors!

I am not sure if the 2SC3503 can achieve the bandwidth target needed for high TVL, but since the part is a drop in replacement for most TVs and the part is still produced... it is compelling. You could try the much higher bandwidth 2SC3782, as it was used in the last generation of HDTV CRTs that were capable of outputting 1080i (which is 810 TVL), but as you mentioned, the breakdown voltage is only 200 volts, so you'd need to: (i) use a voltage divider to lower the amplifier circuit's input voltage rail, and (ii) make G1 voltage more negative to restore cutoff. Here again, we see the benefit of negatively charging the G1 anode, as it lets us operate the cathodes at a lower voltage, which reduces the burden on the cathode amps.

Looking at the E3431D's schematic, it looks like its neckboard uses a 4 transistor amplifier design for each color for a total of 12 transistors. For a single color there are 2 stages to the amplifier design. First there is a cascode amplifier input stage, which is then followed by a push-pull amplifier output stage. Please double check my work though because I am not familiar with the TV and only briefly looked at its schematic on Google just now.
Blacksheep wrote:At first, spot size/beam focus was considered the limitation for TVL, when LukeEvansSimon stepped on the stage with his confident demeanor. The OP now talks about other limitations, but all of that was edited in later.
What you are observing is real life engineering, as opposed to somebody that advertised a finished product. Nothing works perfectly the first time, and so two things are needed for success: (i) a vision grounded in research of the technology landscape, and (ii) persistence to not give up, because iterating on prototype revisions is always necessary before the vision is achieved. This thread started as a technical vision for a mod. Many naysayers said it was a bad idea and wouldn't work. Some people in this thread, such as ElBartoME, decided to start experimenting variations of the idea, adding their own expertise and ideas in to refine the idea, and these are the people that have made real positive contributions and made the CRT gaming scene better because of their experimentation and their constructive improvements to the mod. As opposed to just saying it won't work, don't try it, these people try it, find issues, and propose solutions.
Blacksheep wrote:
Josh128 wrote:Vertical granularity however, is an entirely different story. Its very clear that the spot size has decreased and the the larger and more defined blank lines make it easy to see.
Correct, if anything, the proposed mod, when spot size is decreased, actually seems to be able to increase vertical resolution of the CRT, not horizontal resolution (TVL). Because on the horizontal axis, the beam is simply doing one continuous streak, so how should a decreased spot size even increase resolution there?
It is visually obvious that vertical resolution is improved. The reason why vertical resolution is improved and horizontal resolution is not, is due to a limitation in chroma signal amplifier rise and fall times (that is, amplifier bandwidth). The screen is drawn as a series of horizontal rows of varying color. The electron gun draws each row at the same speed, regardless of the TVL. So a higher TVL image needs to have the colors varied faster, and the speed that the colors can be varied is currently limiting the horizontal resolution improvement from being achieved by the mod.

ElBartoME's oscope images show exactly that occurring. The chroma signals for alternating vertical white and black lines are 3 square waves of around 1 volt peak to peak, corresponding to red, green, and blue alternating from black to peak color illumination. These square waves need to be amplified to around 200 volts peak to peak! The physics involved in swinging voltage up and down by 200 volts fast enough to draw, say 800 TVL, is a big challenge. Most amplifiers have rise and fall times that are too slow.

In this thread, we are engineering towards a vision. We have demonstrable progress towards realizing that vision, but we found an issue. Do we listen to the naysayers and give up, or do we determine the root cause that is limiting success and design the next prototype to test? Removing the chroma signal bandwidth bottleneck could be as simple as upgrading the cathode amplifiers to higher bandwidth parts such as the 2SC3503 or the even higher bandwidth 2SC3782. However, if the pre-amp stage is also a bottleneck, then that will have to be upgraded too. If the pre-amp stage occurs in the jungle chip, then upgrading pre-amp bandwidth will be challenging. If the pre-amp is made up of discrete NPN and PNP transistors, then common parts can be used to upgrade the bandwidth of the pre-amp.
Last edited by LukeEvansSimon on Thu Feb 25, 2021 12:49 am, edited 2 times in total.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

LukeEvansSimon wrote:What you are observing is real life engineering, as opposed to somebody that advertised a finished product. Nothing works perfectly the first time, and so two things are needed for success: (i) a vision grounded in research of the technology landscape, and (ii) persistence to not give up, because iterating on prototype revisions is always necessary before the vision is achieved. This thread started as a technical vision for a mod. Many naysayers said it was a bad idea and wouldn't work. Some people in this thread, such as ElBartoME, decided to start experimenting variations of the idea, adding their own expertise and ideas in to refine the idea, and these are the people that have made real positive contributions and made the CRT gaming scene better because of their experimentation and their constructive improvements to the mod. As opposed to just saying it won't work, don't try it, these people try it, find issues, and propose solutions.
Perfectly understood, 100% with you here. I never said that it wouldn't work by the way. I'd just like proof for the TVL claim, which bothers me a bit, as does some of the (past) reasoning and bizarre usage of terminology. You definitely did some bold advertising, there's no denying that, but maybe don't advertise like you did before there's a finished product, although it is understandable that you want to inspire people and spark interest in this, which is a good thing. But sometimes it can be better to let the audience do the bulk of labelling and judgement instead of trying to pre-empt everything yourself while not even hitting the mark.
LukeEvansSimon wrote:In this thread, we are engineering towards a vision. We have demonstrable progress towards realizing that vision, but we found an issue. Do we listen to the naysayers and give up, or do we determine the root cause that is limiting success and design the next prototype to test?
No, don't give up. Thank you for the explanations. Reality is not impressed by all the discussion anyway, empiricism will always have the last word.
Last edited by Blacksheep on Thu Feb 25, 2021 1:02 am, edited 1 time in total.
LukeEvansSimon
Posts: 128
Joined: Mon Aug 17, 2020 7:28 pm

Re: Mod a CRT to increase its TVL

Post by LukeEvansSimon »

Blacksheep wrote:
LukeEvansSimon wrote:What you are observing is real life engineering, as opposed to somebody that advertised a finished product. Nothing works perfectly the first time, and so two things are needed for success: (i) a vision grounded in research of the technology landscape, and (ii) persistence to not give up, because iterating on prototype revisions is always necessary before the vision is achieved. This thread started as a technical vision for a mod. Many naysayers said it was a bad idea and wouldn't work. Some people in this thread, such as ElBartoME, decided to start experimenting variations of the idea, adding their own expertise and ideas in to refine the idea, and these are the people that have made real positive contributions and made the CRT gaming scene better because of their experimentation and their constructive improvements to the mod. As opposed to just saying it won't work, don't try it, these people try it, find issues, and propose solutions.
Perfectly understood, 100% with you here. I never said that it wouldn't work by the way. I'd just like proof for the TVL claim, which bothers me a bit, as does some of the (past) reasoning and bizarre usage of terminology. You definitely did some bold advertising, there's no denying that, but it is understandable that you want to inspire people and spark interest in this. This is a good thing. But sometimes it can be better to let the audience do the bulk of labelling and judgement instead of trying to pre-empt everything yourself while not even hitting the mark.
LukeEvansSimon wrote:In this thread, we are engineering towards a vision. We have demonstrable progress towards realizing that vision, but we found an issue. Do we listen to the naysayers and give up, or do we determine the root cause that is limiting success and design the next prototype to test?
No, don't give up. Thank you for the explanations. Reality is not impressed by all the discussion anyway, empiricism will always have the last word.
I am using the terminology ("addressibility" versus "resolution") that was used during the height of CRT R&D. I will not use gamer terminology as it is not precise enough for developing this mod, and I purposely referenced decades old CRT research articles to make clear the definitions of terms I was using. Let's please try to take ElBartoMe's philosophy to this mod and just try to keep experimenting with things until it works.

The timeline of events so far: I painted a ambitious vision and quoted sources to try to convince people it was possible. ElBartoMe was the first person to try to implement it and saw no improvement to the picture. Then I tried my own prototype at a much higher voltage and showed the Castlevania SoTN (fact:best game ever) side by sides as proof that the mod seems to be making the scanlines more pronounced. My prototype used expensive lab bench power supplies, so it was far from a finished product. I tested the mod with a Geiger counter, to ensure that the mod doesn't cause CRTs to emit x-rays (a big concern that turned out to be unwarranted). All small steps towards the vision.

ElBartoMe then tried multiple prototypes and then achieved a level of success that cannot be disputed, as his mod looks much better in his side-by-sides for Super Mario World and Super Metroid. Vertical resolution is clearly improved. Your eyes are not lying. He also did it without expensive lab power supplies, but instead used the DiY power supply idea he proposed early in the thread. ElBartoMe went on to prove that his DiY power supply did not put any stress on the flyback. Then he identified that chroma signal amplifier bandwidth is a bottleneck for realizing horizontal resolution improvements. Now we are discussing a practical, low cost addition to the mod that may unlock the mod's full potential. These NPN transistors are $3.65 for 10 transistors. I won't be shocked if ElBartoMe takes this mod all the way to perfection, and the gaming community needs to give kudos to people like ElBartoMe.

Progress is being made, and it is important not to discourage the community from coming together to see how far we can push the envelop here.
ElBartoME
Posts: 82
Joined: Tue Aug 07, 2018 5:55 pm
Location: Germany

Re: Mod a CRT to increase its TVL

Post by ElBartoME »

LukeEvansSimon wrote:2SC3503
That looks like a drop in replacement for the BF871 that is currently in my E3431D. And they have, at least according to the datasheet, a higher bandwith. I'm gonna order them and change the transistors on my neckboard.
LukeEvansSimon wrote:BTW, the chroma signal path may have a bandwidth bottleneck at the pre-amp stage.
Yes, I had the same idea and I'm going to follow the signal back to the preamp and check which stage is limiting the bandwith the most.
LukeEvansSimon wrote:Looking at the E3431D's schematic, it looks like its neckboard uses a 4 transistor amplifier design for each color for a total of 12 transistors. For a single color there are 2 stages to the amplifier design. First there is a cascode amplifier input stage, which is then followed by a push-pull amplifier output stage.
That is correct, it uses a cascode for the input stage and then a push-pull design. I'll follow the signal path and check where the distortion is introduced. It seems the PNP transistor in the push-pull part is also pretty slow. I think 2SA1381 might be a suitable replacement. The low voltage transistor in the cascode amplifier seems fast enough. Unfortunately the input to the cascode comes directly from the jungle chip so if the signal is not fast enough there it then it might be game over.
LukeEvansSimon wrote:I won't be shocked if ElBartoMe takes this mod all the way to perfection, and the gaming community needs to give kudos to people like ElBartoMe.
Thanks, I plan to keep working on it. :mrgreen:
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

Blacksheep wrote:The triads (and/or three individual luminous red, green and blue openings to the phosphor coating they're composed of) still neither are granules, nor are they granular themselves. Suggesting that is just wrong. =) Take a look at these two close-up photos and it becomes clear:

https://commons.wikimedia.org/wiki/File ... cursor.jpg
https://commons.wikimedia.org/wiki/File ... letext.jpg
Josh128 wrote:I was the first "idiot" on this thread to point that out, and mikejmoffit, who knows a "little bit" about CRTs, later said the same thing.
mikejmoffit was first to note that as being the ultimate limit. See here. Your comments on that topic came later. It takes 5 seconds to verify that.
Josh128 wrote:because the white and black lines must be equal in width
No, not at all. You just pulled that out of the hat—that is not part of the definition of TVL. Please provide a proper source for this.

Mike indeed was the first to mention phosphor pitch (missed it), I was the first to mention the mask /grille. Essentially the same as they go together, so I'll give you that.

As far as the TVL being equal width black and white, have you ever seen a chart for measuring TVL that doesnt reflect that?
LukeEvansSimon
Posts: 128
Joined: Mon Aug 17, 2020 7:28 pm

Re: Mod a CRT to increase its TVL

Post by LukeEvansSimon »

ElBartoME wrote:Unfortunately the input to the cascode comes directly from the jungle chip so if the signal is not fast enough there it then it might be game over.
It is not game over, but would require circumventing the jungle chip for the pre-amp. The generic jungle chips used in high resolution multi-sync arcade CRTs such as the LM1203 or LM1279 would work. The LM1279 would even allow for keeping the TV's OSD, but the LM1203 would lose the OSD and would have amplifier bias and gain (aka brightness and contrast) tuned using pots. The LM1279 is, inexpensive, readily available, and definitely high bandwidth enough and has inputs for an OSD's RGB signal.

My Weiya M3138F tri-sync arcade monitor uses the LM1279, and the Weiya M3138F chassis even has a built-in negative voltage rail coming off the flyback, through a rectifier, decoupling cap, and potentiometer for tuning G1 voltage. For the crowd that doesn't understand why CRT chassis designs didn't allow for controlling G1 voltage... well, many of them did! Just look at the tri-sync arcade CRTs that could do 240p games and even 800x600 resolution arcade games, all without changing the CRT tube. In fact, the Weiya M3138F uses the same model tube as the 36-inch JVC D-Series. I promise the naysayers here that the Weiya M3138F does not magically change the phosphor dot pitch when switching between low and high resolutions.
User avatar
treminaor
Posts: 16
Joined: Thu Sep 10, 2020 3:05 am
Location: San Francisco, CA
Contact:

Re: Mod a CRT to increase its TVL

Post by treminaor »

LukeEvansSimon wrote:In fact, the Weiya M3138F uses the same model tube as the 36-inch JVC D-Series. I promise the naysayers here that the Weiya M3138F does not magically change the phosphor dot pitch when switching between low and high resolutions.
Does this mean we can turn the D Series (and other 15Khz CRTs) into tri-sync monitors?
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

LukeEvansSimon wrote:I promise the naysayers here that the Weiya M3138F does not magically change the phosphor dot pitch when switching between low and high resolutions.
Luke, I have an M3129. It has a dot pitch of .79mm which is in the ballpark of what HD CRT sets and of course it doesnt change dot pitch when switching resolutions-- because its designed to be able to support 31 KHz right off the bat. Using a fairly fine pitch mask and phosphor and stepping it down to a lower resolution is not a problem. Broadcast monitors and multisync VGA CRTs have done that forever. Starting with a course dot pitch tube and trying to resolve higher resolutions however, does present a problem. It will limit TVL at higher resolutions and create moire patterns/artifacts if you try to display resolutions higher than the dot pitch is designed for.

Its possible, for sure, to get some percentage of higher TVL if your horizontal output bandwidth can be increased, but at some point, the mask/phosphor pitch will definitely become the limiting factor. Most CRT monitors scanning bandwidths are fairly well matched to the mask and phosphor pitch, but of course any amount of mismatch could allow for improvement if the scanning bandwidth is increased.
LukeEvansSimon
Posts: 128
Joined: Mon Aug 17, 2020 7:28 pm

Re: Mod a CRT to increase its TVL

Post by LukeEvansSimon »

treminaor wrote:
LukeEvansSimon wrote:In fact, the Weiya M3138F uses the same model tube as the 36-inch JVC D-Series. I promise the naysayers here that the Weiya M3138F does not magically change the phosphor dot pitch when switching between low and high resolutions.
Does this mean we can turn the D Series (and other 15Khz CRTs) into tri-sync monitors?
Deflection circuits are my weakness. It is possible to mod a 15 khz CRT to make it multisync, but my skill level isn’t there yet. Deflection circuits are cyclic and feedback into themselves. I struggle to debug deflection circuits.

The electron gun control circuits that this mod is experimenting with, are much easier to understand because they are sequential staged circuits. The most complex stuff is the amplifier design. There are still great tutorials available online for analogue video amplifier design, which explain the basic cathode amp designs you see in CRTs: common emitter, cascode, and cascode followed by a push pull. Most consumer CRTs use the very simple single transistor common emitter amplifier. Sony being Sony used the more expensive 4 transistor amplifier design that is seen in PVMs and multi-sync arcade CRTs. This 4 transistor cascode with push pull output is higher bandwidth than a common emitter amp, if the NPN transistors are the same. As NPN transistor tech improved, the common emitter amplifier design became more popular because it was more simple, less expensive, and since the NPN transistor was much higher bandwidth in late generation CRTs, the cascode and push pull design were not necessary for the bandwidth target consumer CRT designers had.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

Josh128 wrote:I was the first to mention the mask
Clearly not.

Josh128 wrote:As far as the TVL being equal width black and white, have you ever seen a chart for measuring TVL that doesnt reflect that?
You're moving the goalposts instead of responding to the original issue. The definition of TVL is what counts. Furthermore, absence of evidence is not evidence of absence. I can make you a TVL measuring chart where the black and white strips are not equal in width. Now what?
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

Blacksheep wrote:
Josh128 wrote:As far as the TVL being equal width black and white, have you ever seen a chart for measuring TVL that doesnt reflect that?
You're moving the goalposts instead of responding to the original issue. The definition of TVL is what counts. Furthermore, absence of evidence is not evidence of absence. I can make you a TVL measuring chart where the black and white strips are not equal in width. Now what?
No, I believe you are the one "moving goalposts" by attempting to change the definition of TVL to suit your argument that a CRT can somehow resolve more TVL than sets of phosphor triads it contains in the horizontal measurement area.

The definition of TVL is:
the maximum number of alternating light and dark vertical lines that can be resolved per picture height.
True alternation of those lines means they are going to be equal and opposite, not varying in width. This assertion of yours that they can vary in width is complete horseshit, and you (and everyone reading this) know(s) it. You cant just "make your own" chart for measuring TVL-- there is an accepted standard for measuring it, the EIA 1956, which was designed specifically for TV resolution measurement. And what have we here? The lines in the patterns it uses are exactly equal in width at any point.

Image
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

* You made the assertion of something new, so the Burden of Proof is on you. I've never read anywhere that for TVL measurement, the black and white strips must be of equal width.
* Note that I did not claim that the black and white strips must not be of equal width. They can.
* Note that the EIA Resolution Chart 1956 is not the standard for measuring TVL. It is one standard for measuring TVL. It is one realization of the definition.
* "True alternation"? It's elementary school now, but there is no "true alternation" or "false alternation". Either something alternates, or it doesn't. If something alternates, then that does not mean that it must alternate at a constant frequency. The definition of "to alternate" is already met if two conditions take turns, regardless of frequency at which this happens, or variation thereof.
* Your redundant posting of the definition of TVL and the EIA Resolution Chart 1956 does not change the bullet points above.
* I am making this argument in order to inquire proof for your statement; because as this is a public forum, I feel I have some responsibility. I am making this argument for this reason, and for this reason alone.
* I never claimed or implied that a CRT can resolve more TVL than sets of phosphor triads it contains in the horizontal measurement area. Please point out where you think I did this.
User avatar
matt
Posts: 567
Joined: Mon May 09, 2005 5:46 am
Location: Honolulu, Hawaii

Re: Mod a CRT to increase its TVL

Post by matt »

IMO, modding CRT displays to increase their functionality is far more interesting than arguing over the semantics of TVL.

To be honest, manufacturers seem to fudge the TVL of their products to a point where I'm not sure how relevant it is. I've tried pulling up the EIA 1956 chart on a couple of my displays that do have published TVL ratings, and they're not able to resolve the number of vertical lines claimed in the manual. The picture devolves into a blurry moire mess before the supposed max TVL is reached.
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

Blacksheep, Im tired. Lets be friends bro. :mrgreen:
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

Ah, when you notice the opponent is winning, suddenly the point of contention is not as important anymore. Reminding me of Aesop's fable of The Fox and the Grapes.
matt wrote:IMO, modding CRT displays to increase their functionality is far more interesting than arguing over the semantics of TVL.
I completely agree in principle. If only the OP didn't call his thread "Mod a CRT to increase its TVL" and boast about its "TVL boosting" qualities at various places.
matt wrote:To be honest, manufacturers seem to fudge the TVL of their products to a point where I'm not sure how relevant it is. I've tried pulling up the EIA 1956 chart on a couple of my displays that do have published TVL ratings, and they're not able to resolve the number of vertical lines claimed in the manual. The picture devolves into a blurry moire mess before the supposed max TVL is reached.
Did you make sure the displays you checked were not just worn and that it was not caused by the bandwidth of the video signal being the bottleneck? What does "tried pulling up the EIA 1956 chart" mean, were there any obstacles? Anyway, TVL is sound as a way of measurement—but as you say, false advertising/fraudulent labeling is still possible.
Josh128 wrote:Blacksheep, Im tired. Lets be friends bro. :mrgreen:
*shakes hands* Just be aware that that does not grant you a charm against being challenged by me. :mrgreen:
User avatar
matt
Posts: 567
Joined: Mon May 09, 2005 5:46 am
Location: Honolulu, Hawaii

Re: Mod a CRT to increase its TVL

Post by matt »

Blacksheep wrote:Did you make sure the displays you checked were not just worn and that it was not caused by the bandwidth of the video signal being the bottleneck? What does "tried pulling up the EIA 1956 chart" mean, were there any obstacles? Anyway, TVL is sound as a way of measurement—but as you say, false advertising/fraudulent labeling is still possible.
Worn out? Probably. Although I think mostly the phosphor pitch of the tube was too coarse. In practical terms I'm not sure the manufacturers' TVL numbers translate well into something that's perceptible to the end user.

I just hook up my monitors through my PC's built in VGA port. The signal is fine.
User avatar
Josh128
Posts: 2123
Joined: Thu Jan 16, 2014 9:01 am

Re: Mod a CRT to increase its TVL

Post by Josh128 »

Ah, when you notice the opponent is winning, suddenly the point of contention is not as important anymore. Reminding me of Aesop's fable of The Fox and the Grapes.
A bit presumptuous of you to declare that the reason I tire of arguing with you is because "you are winning". I should clarify, I dont think you are. Regardless, I'm I'll leave it up to whoever reads this thread to decide that. 8)
ElBartoME
Posts: 82
Joined: Tue Aug 07, 2018 5:55 pm
Location: Germany

Re: Mod a CRT to increase its TVL

Post by ElBartoME »

While whatever happened in this thread I ordered higher bandwidth transistors for the amplifier on the neckboard and gonna check if I can see a difference in the signal and then hopefully in the picture.

I hope it will be a drop in replacement without changing any passive components with it.
Blacksheep
Banned User
Posts: 78
Joined: Thu Nov 28, 2019 10:49 pm

Re: Mod a CRT to increase its TVL

Post by Blacksheep »

Josh128, it would be, but I did not have you(r post) in mind specifically. It was more like a feeling that hit me. "I should clarify, I dont think you are": of course that's not an argument at all if you are not going to elaborate.
LukeEvansSimon
Posts: 128
Joined: Mon Aug 17, 2020 7:28 pm

Re: Mod a CRT to increase its TVL

Post by LukeEvansSimon »

ElBartoME wrote:While whatever happened in this thread I ordered higher bandwidth transistors for the amplifier on the neckboard and gonna check if I can see a difference in the signal and then hopefully in the picture.

I hope it will be a drop in replacement without changing any passive components with it.
Ignore the naysayers. Let's treat this as a fun engineering experiment, and see how far we get. Tremendous progress has already been made, and the drama around this is because we are pushing the limits of what many believe is possible. That makes this all the more exciting.

Did you find higher bandwidth transistors for all 4 transistors for each of the 3 colors? The bandwidth of each transistor matters. The second NPN in the cascode tends to be the most common bandwidth bottleneck because it has larger voltage swings than all of the other 3 transistors, and the voltage rise and fall times of these transistors is the limiting factor for achieving a high bandwidth at a few hundred volts peak to peak.

For a single color the second NPN transistor in the cascode performs most of the voltage amplification of the multistage amplifier, and it is used in a common base amplifier configuration, so it does work for the entire waveform. This configuration makes this transistor have high bandwidth (good) but low input impedance (bad). To correct for the low input impedance, the cascode adds an initial NPN transistor in a common emitter amplifier configuration. Together this makes the 2 transistor cascode amplifier have a medium input impedance (good) and high bandwidth (good).

However, the cascode has a high output impedance (bad), and this is why the push-pull amplifier stage is added, the push pull amp can be thought of as a current amplifier that ensures a low output impedance for the amplifier. The push pull amplifier is a pair of common collector amplifiers, one NPN and the other PNP. They each amplify half of the waveform, so they handle half the voltage displacement compared to the common base amp in the cascode. I bet these push pull transistors are the second most likely bottleneck for achieving high enough bandwidth for an increase in horizontal resolution. So getting higher bandwidth replacements for the push pull transistor pair is a good idea.

Upgrading all 4 transistors to higher bandwidth parts is the safest route to higher bandwidth cathode amplifiers. You can also add a 5th transistor before the cascode to further increase bandwidth. This is done by using a PNP transistor in common collector amplifier configuration to act as a voltage buffer that further increases input impedance.

Below is a table of the transistors in your CRT's amplifier for a single cathode, with suggested upgrade parts. Note that the specified bandwidth is measured at a specified voltage. At higher peak to peak voltages, the bandwidth is lower. I may be wrong, but from the schematic and specs, the common emitter in the cascode probably does not need to be upgraded. The first transistor in the table is the 5th transistor that I mentioned before, and this 5th transistor may not be necessary. So only try it if upgrading the existing amps is not enough.
R=Stage&Type&Configuration&Purpose&Original Part&Original Bandwidth&Upgraded Part&Upgraded Bandwidth R=Input&PNP&Common Collector&Buffering&-&-&2SA1381&150 mhz R=Cascode&NPN&Common Emitter&Increase input impedance&2SC2412&180 mhz&-&- R=Cascode&NPN&Common Base&Voltage amplification&BF871&60 mhz&2SC3503&150 mhz R=Push Pull&NPN&Common Collector&Decrease output impedance for half of the signal&BF871&60 mhz&2SC3503&150 mhz R=Push Pull&PNP&Common Collector&Decrease output impedance for other half of the signal&2SA1091&60 mhz&2SA1381&150 mhz

Random trivia for CRT enthusiasts. ElBartoMe's Sony TV uses Philips transistors for video amplification. This is typical for a CRT TV from the later era of CRTs. In the early years of CRTs, each brand made almost every part in the CRTs they sold, even the capacitors! In the later years, the major CRT brands would specialize in making only a few of the parts themselves (in Sony's case they specialized in the CRT tubes), and most other parts in the CRT were purchased from a competing brand! This allowed each company to focus their engineering on improving certain parts of the tech. In the late 1990s, Philips was one of the leaders in analog video amplifiers. A late generation Sony TV will have capacitors from Panasonic, transistor amps from Philips, ICs from Sanyo and Toshiba. Similar for late generation Panasonic CRTs, Sanyo CRTs, Philips CRTs, etc.
DiY analog audio amplifiers are cool, but DiY analog video amplifiers are coolest.
ElBartoME
Posts: 82
Joined: Tue Aug 07, 2018 5:55 pm
Location: Germany

Re: Mod a CRT to increase its TVL

Post by ElBartoME »

LukeEvansSimon wrote:Did you find higher bandwidth transistors for all 4 transistors for each of the 3 colors? The bandwidth of each transistor matters. The second NPN in the cascode tends to be the most common bandwidth bottleneck because it has larger voltage swings than all of the other 3 transistors, and the voltage rise and fall times of these transistors is the limiting factor for achieving a high bandwidth at a few hundred volts peak to peak.
I ordered 12 2SC3503 and 6 2SA1381. Double what I actually need but it's good to have some as backup.
Using the current bandwith product to compare transistors is questionable because the manufacturers also measure them at different conditions. We will see if there will be any noticable change in the bandwith...
LukeEvansSimon wrote:However, the cascode has a high output impedance (bad), and this is why the push-pull amplifier stage is added, the push pull amp can be thought of as a current amplifier that ensures a low output impedance for the amplifier. The push pull amplifier is a pair of common collector amplifiers, one NPN and the other PNP. They each amplify half of the waveform, so they handle half the voltage displacement compared to the common base amp in the cascode. I bet these push pull transistors are the second most likely bottleneck for achieving high enough bandwidth for an increase in horizontal resolution. So getting higher bandwidth replacements for the push pull transistor pair is a good idea.

Upgrading all 4 transistors to higher bandwidth parts is the safest route to higher bandwidth cathode amplifiers. You can also add a 5th transistor before the cascode to further increase bandwidth. This is done by using a PNP transistor in common collector amplifier configuration to act as a voltage buffer that further increases input impedance.
Thanks for the explanation. My expertise is more in digital amplifiers so learning more about analog stuff is very interesting to me! If after exchanging all transistors the bandwith is still bottlenecked at the input I'll add the additional PNP transistor at the input.
LukeEvansSimon wrote:Below is a table of the transistors in your CRT's amplifier for a single cathode, with suggested upgrade parts. Note that the specified bandwidth is measured at a specified voltage. At higher peak to peak voltages, the bandwidth is lower. I may be wrong, but from the schematic and specs, the common emitter in the cascode probably does not need to be upgraded. The first transistor in the table is the 5th transistor that I mentioned before, and this 5th transistor may not be necessary. So only try it if upgrading the existing amps is not enough.
Wow, thanks again for your help. I also thought the 2SC2412 is fast enough so I didn't search for a replacement. I'll change the other transistors and measure every stage and document my findings.
Post Reply