Mod a CRT to increase its TVL

[ElBartoME]

Before I got distracted by family, work, other hobbies, and the US presidential election, I was working on designing and testing a from scratch new CRT neck board that uses a cathode amplifier design that has around 30 mhz chroma signal bandwidth for amplification of the RGB chroma signals up to 200 volts peak to peak. The design uses a "cascode" amplifier design, as well as Sanyo's discrete NPN transistor 2SC3782 that were used in 1080i HDTV CRTs. For generating the negative voltage rail for the G1, my "super neck board" design uses an inexpensive pulse transformer. This allows for tapping the unrectified cathode amplifier rail coming off of the flyback, to create a separate negative voltage rail that is 2x the voltage of the cathode amplifier rail (which is usually around 200 volts).

The idea is for the "super neck board" to take 5 minutes to install for somebody with beginner soldering skills, and by significantly increasing cathode amplifier bandwidth, further resolution improvements will be realized, in addition to those realized by increasing K to G1 and G2 to G1 voltage differentials. I've done all of the LTSpice simulations, ordered all of the parts, and I just need to breadboard prototype it now. I found a HUGE new old stock supply of 2SC3782 NPN transistors. Higher bandwidth transistors have been made since the 2SC3782, but they are not capable of the 200 volts peak to peak amplification that the 2SC3782 is capable of. The 2SC3782 is insane!

Your custom neckboard sounds very interesting! Don't you need to adapt it to different sockets as not every tube uses the same pinout? How are you going to do that? I'm eager to see the results.

Also I'm debating on removing R709 so my G2 can go a bit higher. Not sure if this will cause any problems...

Theoretically I could use a voltage doubler and tap into the G2 voltage before it gets rectified but my concern is will this cause problems with the potentiometer for the screen voltage? It is not designed for that voltage and may arc. Also I'm not sure what the clearance is between G2 and G1. If I double G2 can this cause arcing between the two grids?

I have a Sony TV with a very worn tube. I might try doubling G2 on that TV before I accidentally kill my precious 34 inch Sony.

[zebius]

Here you can see the A3411D on the right side. On the left side is a KV-29X5D (29 inch) as a size comparison.

Nice work! Are you going to do this on the 29 inch too? I have the same model (KV-29X5D) with a blurry screen and would be very interested in the how-to/results.

[LukeEvansSimon]

Your custom neckboard sounds very interesting! Don't you need to adapt it to different sockets as not every tube uses the same pinout? How are you going to do that? I'm eager to see the results.

I need to find the time to get back to my CRT projects, but so many other things in life are taking up my time. The super neckboard PCB will come in two versions: one that uses the B10-277 socket, which will support most shadow mask tubes, and another that will use a socket to cover most aperture grille tubes. Because I mostly collect slot mask tubes, I am starting with the B10-277 version.

The super neckboard will also include a 70mhz bandwidth jungle chip: Texas Instrument's LM1203. So there will be two ways to wire up the super neckboard. You can wire it up to use the jungle chip already in your chassis, but if your chassis's jungle chip is not high bandwidth, then max resolution will be bottlenecked by it. The other way to wire it up will be to just have the input RGB signal feed straight to the LM1203 on the super neckboard. I found new old stock of LM1203s, so it will be low cost to include this feature. The disadvantage of using the LM1203 is that there is no on screen display, so pots on the neckboard have to be used to tune RGB bias (aka brightness) and gain (aka contrast). However, tuning these pots is only required once, during installation of the neckboard.

The biggest challenge for a CRT neckboard are the cathode amplifiers. Amplifying a 5 volt peak to peak signal up to a 200 volt peak to peak signal at 30 mhz bandwidth is impressive. Professional CRTs use smaller electron guns that run at a lower voltage than consumer CRTs that use electron guns that run at higher voltage. A PC CRT or BVM have cathodes that are run at around 50 volts peak to peak, whereas your consumer CRT TV has cathodes that are run at 200 volts peak to peak. The cathode amplifiers used in the large consumer HDTV CRTs are by far, more capable than the amplifiers used in BVMs. However, every CRT gamer knows that HDTV CRTs have horrible laggy circuits that are terrible for gaming. So my super neckboard is backporting the HDTV cathode amplifiers to 15khz SDTVs so we can have the best of both worlds: zero lag and yet ultra high bandwidth video amplication at consumer CRT high voltage.

Also I'm debating on removing R709 so my G2 can go a bit higher. Not sure if this will cause any problems...

Theoretically I could use a voltage doubler and tap into the G2 voltage before it gets rectified but my concern is will this cause problems with the potentiometer for the screen voltage? It is not designed for that voltage and may arc. Also I'm not sure what the clearance is between G2 and G1. If I double G2 can this cause arcing between the two grids?

I have a Sony TV with a very worn tube. I might try doubling G2 on that TV before I accidentally kill my precious 34 inch Sony.

It is important to preserve large curved consumer CRTs. The small PC CRTs and RGB monitors came into popularity very late in the era of CRTs, and yet many CRT gamers are trashing large curved consumer CRTs at a very high rate compared to the small PC CRTs and RGB monitors. As I mentioned above, the electron gun design and video amplifier design in pro CRTs are different than consumer CRTs. One is not superior to the other, just as a PS5 is not superior to a PS1.

The pot that is in series with R709 needs both a current rating and voltage rating to handle the increased current and voltage that it will need to endure if R709 is removed. So you can upgrade the pot too, in order to be safe.

Another model consumer CRT that would likely benefit greatly from this mod is the Mitsubishi CS-40809. It is a 40-inch bubble curved slot mask consumer CRT TV with component! However, the neckboard is designed to have the G1 voltage near zero and so the blank scanlines in 240p content are very thin. You can see a CRT gamer demonstrate the scanlines in this YouTube video. Compared to the smaller 36-inch JVC D-Series, which is also a bubble curved slot mask tube, the Mitsubishi CS-40809's blank scanlines are no where near what they could achieve with a more negative G1 voltage.

[buttersoft]

Ok, that neckboard sounds totally hawt. If you can keep the price down, and shipping is reasonable, I'm sure thumptech and i will put in an order for a few each.

EDIT: any chance the little transformer is going to be easily adjustable? It would be nice to be able to tweak the effect for different sizes/CRTs/games/viewing distances

[ElBartoME]

Does the LM1203 include automatic cut-off control? I noticed that the automatic cut-off control of my TV made the process so much easier. My TV uses the TDA4580 (https://datasheetspdf.com/pdf-file/378437/NXP/TDA4580/1) that implements the automatic cut-off. Unfortunately it seems to only have a bandwith of 10 MHz.


Also I decided to move R709 after the pot. That should increase G2 at the tube but the current through the pot should still remain the same. I'm gonna try that the next time I have the TV open.

Here you can see the A3411D on the right side. On the left side is a KV-29X5D (29 inch) as a size comparison.

Nice work! Are you going to do this on the 29 inch too? I have the same model (KV-29X5D) with a blurry screen and would be very interested in the how-to/results.

Yes, my 29X5D suffers also from a bit of blur. If I find the time I will mod it too.

[LukeEvansSimon]

Ok, that neckboard sounds totally hawt. If you can keep the price down, and shipping is reasonable, I'm sure thumptech and i will put in an order for a few each.

EDIT: any chance the little transformer is going to be easily adjustable? It would be nice to be able to tweak the effect for different sizes/CRTs/games/viewing distances

The super neckboard will have a pot for tuning G1 voltage, which means you can tune for the cathode ray spot size that matches your tastes. The smaller the spot size, the thicker the blank scanlines, and the illuminated scanlines will be sharper. As mentioned in this thread, tuning G1 voltage requires a second step: tuning the G2 voltage. This is the price to pay for adjustable thickness real scanlines.

[SuperSpongo]

[...] So my super neckboard is backporting the HDTV cathode amplifiers to 15khz SDTVs so we can have the best of both worlds: zero lag and yet ultra high bandwidth video amplication at consumer CRT high voltage. [...]

This is a super interesting topic that deserves its own thread entirely, imo.

I was hoping of one day finding a way of displaying 480p on the plethora of 100Hz TVs that are available in Europe, but I lack the knowledge on how to do so. Would you be willing to give a couple of pointers about what can still be used and what can't? I was always under the impression that 100Hz TVs essentially run on a 31kHz signal ("240p120") and indeed, some manufacturers like Grundig and Metz offer a VGA Port that supports 480p60, but my understanding does not go deep enough. Do you think it's possible to tap into a standard 100Hz TV and display a VGA signal with some kind of custom board?

And on the other hand: how are you planning on displaying 15kHz signals on an HD CRT, what is the limiting factor usually that VGA displays and HD CRTs cannot sync as low as 15kHz? (Kind of a noob question )

[LukeEvansSimon]

[...] So my super neckboard is backporting the HDTV cathode amplifiers to 15khz SDTVs so we can have the best of both worlds: zero lag and yet ultra high bandwidth video amplication at consumer CRT high voltage. [...]

This is a super interesting topic that deserves its own thread entirely, imo.

I was hoping of one day finding a way of displaying 480p on the plethora of 100Hz TVs that are available in Europe, but I lack the knowledge on how to do so. Would you be willing to give a couple of pointers about what can still be used and what can't? I was always under the impression that 100Hz TVs essentially run on a 31kHz signal ("240p120") and indeed, some manufacturers like Grundig and Metz offer a VGA Port that supports 480p60, but my understanding does not go deep enough. Do you think it's possible to tap into a standard 100Hz TV and display a VGA signal with some kind of custom board?

And on the other hand: how are you planning on displaying 15kHz signals on an HD CRT, what is the limiting factor usually that VGA displays and HD CRTs cannot sync as low as 15kHz? (Kind of a noob question )

CRT displays have 4 main parts to their circuits: (i) power supply for converting AC to DC, (ii) deflection for scanning the 3 cathode rays across the screen, (iii) chroma signal amplification for changing the intensity of the 3 cathode rays as they are scanned across the screen, and (iv) CRT anode charging for shaping the cathode ray and accelerating it towards the screen. The super neckboard modifies (iii) and (iv). So the HDTV chroma amplifiers are being backported to SDTVs for very high bandwidth amplification. Deflection is not being modified at all. So if your CRT only does 15khz, then super neckboard won’t change that.

Cathode amplifier bandwidth is one of the 3 bottlenecks for TVL. The other 2 are cathode ray spot size and phosphor pitch. I can’t modify phosphor pitch for a reasonable cost, and so super neckboard is focusing on the other 2, which bottle neck TVL on most consumer CRTs.

[SuperSpongo]

Thanks for the additional insights

[thumptech]

Do you think it's possible to tap into a standard 100Hz TV and display a VGA signal with some kind of custom board?

I have achieved this on a 100hz Sony set. The Jungle IC accepted 480p RGBHV VGA signal from a scaler chip named the "Digital Reality Creation". I cut these traces and attached my own VGA port. It works very well with my Sega Dreamcast. I was also able to jack into the RGB lines for the teletext board - these were internally upscaled from 15khz to 31khz in the jungle IC and could be used with older consoles.

I am fairly sure this approach will work with most "100hz" sets.

I'll leave it at that as it is off topic.

[yoZe]

Do you think it's possible to tap into a standard 100Hz TV and display a VGA signal with some kind of custom board?

I have achieved this on a 100hz Sony set. The Jungle IC accepted 480p RGBHV VGA signal from a scaler chip named the "Digital Reality Creation". I cut these traces and attached my own VGA port. It works very well with my Sega Dreamcast. I was also able to jack into the RGB lines for the teletext board - these were internally upscaled from 15khz to 31khz in the jungle IC and could be used with older consoles.

I am fairly sure this approach will work with most "100hz" sets.

I'll leave it at that as it is off topic.

If you could detail this by doing your own tutorial on another subject it would be a real revolution!
Indeed in europe we may have rgb on the SCART but our TV had no component input so it was impossible for us to benefit from 480p.

[SuperSpongo]

If the interest is there we could open up another thread.

[Kdeizy]

While I agree that the final results create a sharper image and have also read through the arguments for and against whether it’s raising the tvl, has anyone actually posted a tvl test pattern with the results?Being a visual form of measurement, it’s the only way to determine if it’s actually being effected. It cannot be determined by close up images of character sprites. I still believe that tvl is more determined by the mask dot pitch and screen size (for instance, the tvl of my vga monitor (aperture grille) remains the same whether it’s displaying in 640x480 or 1600x1200). Pics look good and arguments for it being sharper are justifiable, but changing the tvl hasnt been officially determined as far as ive seen (unless I missed it somewhere).

[Derf]

Such a crazy awesome mod! Thanks for the excellent write-up in the OP. Can't wait for the neckboard. Any ETA on it? I was also curious if these are facts:
- Already high PVMs/BVMs generally will always have a 1:1 TVL to spot ratio already, so this mod will only make blank scanlines thicker
- Consumer CRTs <24" are generally lower than a 2:1 ratio, but aren't 1:1 so they stand to gain some TVL but not double

So a line of 4 consecutive slots in the mask will have no black slots between two white spots because the spot size takes up 2 slots, and after the mod, a spot only illuminates a single slot:
R=Before Mod&After Mod R=White, White, White, White&White, Black, White, Black

Can you get a good picture of this? ElBartoME's high res shots are on small PVMs so I don't think it really demonstrates increased TVL if you zoom in and count illuminated triads.

[retrozar]

Would this be feasible to try on the FV310?

[ElBartoME]

I did the mod on a KV-E3431D this time. The procedure was besically the same as I described with my last post but minor changes.

I had to scrape some traces on the neckboard as on this chassis G1 is on GND. Also all G1 connections (G1-1, G1-2 and G1-3) were used instead of only one like on the A3411D.

I was able to go up to -120V for G1 before I ran out of G2 voltage.

Also this chassis has a weird quirk: When I slowly went more negative with G1 I expected the image to get dimmer. That didn't happen though. Instead above a certain G1 voltage the picture just completely vanished. Only after I raised G2 the picture would come back. I assume on this chassis the beam current is measured and if it is too low it will blank the picture completely. This made it a bit difficult to find a sweet spot. I settled for -120V. I probably could have got a bit higher but I wanted to have a bit of a safety margin.








Click here for high res: https://i.imgur.com/Qlp8JLQ.jpg


Click here for high res: https://i.imgur.com/kf9envt.jpg


Click here for high res: https://i.imgur.com/65JluQJ.jpg

In the second to last picture if you look at the circle you can see the black outline is thinner with -120V on G1. This would speak for an increase in TVL.

[Derf]

In the second to last picture if you look at the circle you can see the black outline is thinner with -120V on G1. This would speak for an increase in TVL.

Thanks for the pictures! I might do it to one of my own sets soon. Is there a sure fire way I can be absolutely certain that my set is using more than 1 TVL for a spot? It'd be nice if 240p Test Suite had a TVL test card like the below so it's more apparent and so we can find the TVL of any set that doesn't have a manual / advertise it. I wonder if it could be compiled with an image swapped out for the card?

https://www.arcdyn.com/articles/wp-cont ... 24x791.png

[LukeEvansSimon]

ElBartoME is the GOAT.

[ElBartoME]

In the second to last picture if you look at the circle you can see the black outline is thinner with -120V on G1. This would speak for an increase in TVL.

Thanks for the pictures! I might do it to one of my own sets soon. Is there a sure fire way I can be absolutely certain that my set is using more than 1 TVL for a spot? It'd be nice if 240p Test Suite had a TVL test card like the below so it's more apparent and so we can find the TVL of any set that doesn't have a manual / advertise it. I wonder if it could be compiled with an image swapped out for the card?

https://www.arcdyn.com/articles/wp-cont ... 24x791.png

I have a computer running CRT emudriver so I can put that picture on the TV. But that will take a while because the TV is not where the computer is so I need to move one of them.

ElBartoME is the GOAT.

Maaaaa

[SuperSpongo]

This is so cool!
I especially like the pics of the sharpness test. The difference is obvious!

[matt]

I like the "vertical lines" screen on the PS1 version of the 240p test suite. It can switch between several different horizontal resolutions on the fly, including 384x240 and 640x240.

[Kdeizy]

In the second to last picture if you look at the circle you can see the black outline is thinner with -120V on G1. This would speak for an increase in TVL.

One thing to keep in mind when discussing tvl increase is that we are dealing with consumer sets that have a much less finer pitch than a higher end monitor. We don't really have any accurate tvl documentation of consumer sets as that information wasn't really provided, but we can look at something like the pvm 2530 in a similar way to a consumer set. A pvm 2530 at an ag pitch of .73 mm calculates to almost exactly 560 tvl for which it is listed. With a more coarse pitch, phosphor separation can be seen with the naked eye and therefore spot size has less of a chance to effect the overall tvl. However, as proven in the before and after photos provided, it can still have an effect on overall image clarity. The results do look awesome.

In contrast, something like the fine pitch of a bvm 20f1u at .30mm has a tvl much more susceptible to variation in spot size. Its size vs pitch works out to approximately 960 tvl which is obviously in excess of its listed tvl rating. With a pitch that low the individual phosphors are much harder to discern with the naked eye and any loss of clarity can reduce overall tvl.

I still think its worth trying to ensure that ur getting the max out of a consumer set though. Its good work all around.

[Blacksheep]

has anyone actually posted a tvl test pattern with the results?Being a visual form of measurement, it’s the only way to determine if it’s actually being effected. It cannot be determined by close up images of character sprites.

Right, it has not been proven here, as of yet, whether the mod results in a TVL increase. Only conjecture. Although the results do look interesting. mikejmoffitt's writeup is missing a before-after comparison unfortunately.

Shouldn't be too hard to grab a measuring tape and fire up some vertical line patterns using some 240p test suite(s). May be imprecise as that only provides a selection of some specific, fixed resolutions but may make it possible to objectively decide TVL increase itself. A proper TVL test chart would be a better gauge no doubt.

[SuperSpongo]

Ah, but now you're in the domain of what LukeEvansSimon calls "Addressability" again You'd have to test this with a special image and some kind of resolution akin to the super resolution settings on retroarch for example.
The 240p test suites are limited to the maximum horizontal resolution of their respective platforms.

[Derf]

Ah, but now you're in the domain of what LukeEvansSimon calls "Addressability" again You'd have to test this with a special image and some kind of resolution akin to the super resolution settings on retroarch for example.
The 240p test suites are limited to the maximum horizontal resolution of their respective platforms.

PS1 and beyond can do 640 horizontal (Saturn can do 704 but I don't think 240pTS has been ported there yet). 640*.75 = 480TVL, so I guess if normal consumer CRTs are ~300TVL like people say, you could detect a bit of an increase but not double. So to go beyond that, you'd need the CRT emudriver to generate a higher res signal for the test cards until you're limited by the CRTs native addressibility. Does my logic check out?

I did end up requesting TVL test cards (or similar) to be added to 240pTS with some responses from Artemio.

[Blacksheep]

Ah, but now you're in the domain of what LukeEvansSimon calls "Addressability" again

There is no such thing as far as TVL are concerned: TVL is an inherent quality of the display device and can be measured without a signal generator that outputs some specific number of dots per line. What repeatedly has been called "addressability" in this thread is irrelevant for the TVL of a display device—it is a feature of a signal (generator) and only says something about the extent to which that is able to exploit the inherent qualities of a display device.

What I meant was: say for example you have a TV with a TVL of around 300, and you display a vertical lines pattern on that using the PS1 240p suite in 640x240 mode. That's 480 dots within the width of a line equivalent to the height of the screen. The pattern will look blurry/gray. Then you do the mod and after that you display the same pattern again. If the lines are then clearly visible/countable, then we would at least know that the mod was able to increase the TVL of the TV by at least around 180 TVL. With an appropriate resolution chart, the (increase in) TVL could be determined more precisely. Maybe such a signal could be generated with CRT emudriver using some sufficiently high number of dots per line.

[ElBartoME]

Alright I setup my PC with a resolution of 1024 x 480 in order to have enough resolution in the horizontal.

I made a quick test with my PVM-14L4 and I think the setup is fine to see any difference in TVL before and after the mod. I will put G1 back to 0V and do a before and after. The TV I modified is not at my home though so I'll do it on Saturday.

https://i.imgur.com/PnYIexw.jpg

https://i.imgur.com/2X4jV5P.jpg

[Derf]

Ah, but now you're in the domain of what LukeEvansSimon calls "Addressability" again

There is no such thing as far as TVL are concerned...

I believe Spongo was referring to exactly what you explained - the TVL test would be limited to the addressability of the console running the 240p test suite. You're on the same page.

El's method would be the best way to tell, especially since consumer TVs are usually an unknown TVL value to start with and it'd be much clearer to see a true doubling which wouldn't be possible with 240p test suite since the doubled TVL would be very likely over 480.

I will put G1 back to 0V and do a before and after. The TV I modified is not at my home though so I'll do it on Saturday.

So excited to see how it turns out! Please upload the final pictures as PNG instead of JPG so that it's a clearer image.

[Blacksheep]

Derf: ok.

So excited to see how it turns out! Please upload the final pictures as PNG instead of JPG so that it's a clearer image.

Seconded, thanks for doing that ElBartoME. Are you able to display the pattern in the correct aspect ratio? Might warp the result otherwise.

[Derf]

Seconded, thanks for doing that ElBartoME. Are you able to display the pattern in the correct aspect ratio? Might warp the result otherwise.

It does look like the squares aren't square and circles aren't circle. He's using CRT emudriver so he can plug in any custom resolution. The picture needs to be cropped to that resolution too I believe.