Mod a CRT to increase its TVL

[LukeEvansSimon]

At 640x480 resolution, a PC CRT's resolution is 480 TVL, and when it switches to 1024x768 resolution, the resolution is increased to 768 TVL.

You know the tech, as you have shown in this thread, but at the same time this is the 2nd time you’ve related TVL to the vertical resolution. This is confusing.

As a bunch have already reminded, TVL relates to the horizontal resolution. I.e how many vertical lines contained in a displayed image can be resolved on the horizontal axis (to be absolutely clear, not talking about scanlines here, which define the limit of vertical resolution, i.e how many horizontal lines (in a displayed image) can be resolved on the vertical axis).

In your example, 640x480 corresponds to 640 TVL, 1024x768 1024 TVL.

I am not confusing TVL with vertical resolution. You are missing the fact that TVL only measures horizontal resolution over the width of the screen equal to the height of the screen. So TVL on a 4:3 aspect ratio CRT is measured over the first 3/4 of the horizontal area of the screen, so for 640x480 the math is 640*3/4 = 480 TVL. For 1024x768, the math is 1024*3/4 = 768 TVL.



As I stated before, TVL is a stupid metric. Citation for proof that metric is stupid in the ways I describe is here:
https://en.m.wikipedia.org/wiki/Television_lines

[Xer Xian]

Congrats on the progress made up to now. I hope you don't mind if I ask - do you you have some reference to back this bit here:

PC CRTs can change spot size on the fly, as they switch between different resolutions

I thought CRTs changed res simply by changing the horizontal scan rate (and the line spacing ofc). I've never heard of varying the spot size on the fly.

[Unseen]

So TVL on a 4:3 aspect ratio CRT is measured over the first 3/4 of the horizontal area of the screen, so for 640x480 the math is 640*3/4 = 480 TVL.

But you are still assuming some amount of horizontal pixels, which does not really apply to an analog signal. The source could instead generate an analog signal from a 2880x480 frame buffer or the source could be a signal generator that sends an analog frequency sweep in each video line.

[6t8k]

Thank you for presenting this interesting mod and digging out those resources!

At 640x480 resolution, a PC CRT's resolution is 480 TVL, and when it switches to 1024x768 resolution, the resolution is increased to 768 TVL. [...] TVL on a 4:3 aspect ratio CRT is measured over the first 3/4 of the horizontal area of the screen, so 640*3/4 = 480 TVL.

It's fallacious to link a discrete resolution, like 640x480, to TVL. The former concept does not exist in the realm of analog signals – as Unseen wrote, the concept of a pixel does not apply. TVL might be discrete in the sense that the vertical lines are countable by a person, but the trick is that TVL is purely a visual and subjective metric. This is why TVL measures a horizontal resolution over a horizontal span equal to the height(=vertical extent) of the picture by means of counting visibly distinct vertically oriented lines across said horizontal span, and does not measure an integer number of pixels a given scanline consists of. In addition, display aspect ratio is decoupled from storage aspect ratio, the latter of which is furthermore not applicable in analog video.

In light of this I think that from the perspective of TVL, there is no dichotomy between 'resolvability' and 'addressability', something like an 'addressable TVL' (as you wrote here for example) does not exist. I've read the document you linked in the OP that gives a meaning to the terms 'resolvability' and 'addressability'; it determines the former to be primarily a function of spot size and the latter to be dictated by the CRT TV set's internal circuitry that drives the electron beam. But the difference doesn't matter for TVL, as it is influenced by both simultaneously, which can be understood through the trivial observation that if one part was absent, there would be no picture. The article gives a good overview, but the way it refers to pixels in the "resolution versus addressability" paragraph is technically not correct.

To counter confusion, it would help to take analog video sources that generate their signal using a digital circuit out of the equation. With analog video signals generated by a digital circuit, one can speak of 'dots'. The information these dots carry can for example overlap in various ways – the signal is still analog though, and therefore continuous in both the time domain and magnitude, as opposed to digital signals, which consist of discrete values that are situated at discrete instants of time (pixels are merely 'blown up in size' for display). During earlier chapters of the history of television, rotating perforated discs over a photodetector or video camera tubes were used to convert light to an analog electrical video signal. A signal generated this way is just as continuous as a signal generated by an old video game system, however, because the analog video signal is generated by an analog circuit, it isn't even appropriate anymore to speak of dots. Of course, you can still measure TVL in this scenario. Take for example the Indian-head test pattern, which was the basis for generating an electrical analog video signal by means of a monoscope tube. The wedge consisting of vertical stripes right below the indian head was used for checking the CRT TV set's horizontal resolution capability, i.e. TVL, meaning whether it could display the following broadcast in sufficient detail. There were also test cards with labelled wedges that let you read off the TVL value of your TV set easily, like the EIA 1956 resolution chart.

On this occasion, I'm also not quite convinced that TVL is "a stupid metric that was clearly designed for marketing purposes as opposed to properly informing people about a display's properties" :p It's a useful metric informing the interested party about the quality of definition and clarity of the picture they can expect of a given CRT television set - independent of the device that's generating the video signal that goes into said TV set. If TVL are measured carelessly, products are labelled wrong, or if the meaning of the term is not understood correctly, then that is not caused by the inventors of the metric, unless the very same people also carelessly counted the TVL and/or labelled products wrong.

A higher TVL CRT makes the blank scan lines thicker and the illuminated scan lines thinner.

That's a correlation, not a causation: the factors TVL depend on form a superset of the factors scanline and skipped line thicknesses depend on.

[maxtherabbit]

snip

thank you for adding some much needed clarity to this TVL discussion

[Harrumph]

You are missing the fact that TVL only measures horizontal resolution over the width of the screen equal to the height of the screen. So TVL on a 4:3 aspect ratio CRT is measured over the first 3/4 of the horizontal area of the screen

Ah indeed, that was something I had missed, thanks for the clarification.

[LukeEvansSimon]

snip

Thanks for being the first and only person to read my citations.

My comment about 640x480 resolution is referring to resolution of 640 by 480 lines. I am not talking about addressability or digital displays, though many can easily address that many lines. When I say “resolution”, I refer to just resolution as defined in:

https://link.springer.com/content/pdf/1 ... 168105.pdf


The better metric for resolution is simply number of horizontal lines by number of vertical lines. TVL is awkward. TVL only talks about horizontal and for a restricted portion of the screen. Vertical resolution matters too as does the entire screen. I understand the complex history of TV that gave rise to the metric, but it is simply antiquated. The distinction between “addressability” and “resolution” is also antiquated... but it is very useful for 240p gaming on high TVL CRTs.

You are right that an analog chroma signal could encode a very high number of vertical lines, even for a 240p signal... but as described in the above reference, the width of the electron beam as it hits the phosphor aka “spot size” limits both horizontal resolution and vertical resolution of what can be drawn on the CRT without the lines bleeding together. The chroma signal feeding into the cathodes may encode 1000 TVL, but unless the spot size is approximately 1/1333th the width of the screen, then the CRT is not 1000 TVL. Similarly the phosphor triad/mask pitch also needs to be at least 1/1333th the width (as stated in OP, this is commonly the case).

The term “TVL” on its own is ambiguous in a general AV context. It could be referring to addressability or to resolution. But for 240p gaming, the ambiguity does not exist because the game systems cannot address anything above 480 TVL (in fact they are much lower). So 240p gaming on an 800 TVL CRT means one thing: it means TVL is referring to the resolution of the CRT and not the addressability.

This is why the antiquated distinction between addressability and resolution is important for 240p gaming.

[LukeEvansSimon]

I'd love to see somebody try this mod on a 600 TVL aperture grill PVM. I bet a PVM G1 charged to -400 volts, the tube would draw at a 1000 TVL fidelity and create razor thin illuminated scanlines. I know that look is an acquired taste, but if you can have a potentiometer dial that lets you achieve both... why not? I mean, the fact that I got any visible results on a shitty, cheapy low end consumer curved slot mask CRT TV suggests the results on a 600 TVL aperture grill will be rockstar.

So as I said I got this 14N2 here that is 500 TVL and it's not turning on.



I got it out of storage today and it actually turned on for a very brief moment.



The image stayed on for a short while, got bigger and then it shut itself off again. If somebody has a pointer in the right direction as to what could cause this it would be very much appreciated. Not exactly sure what causes this.

My idea would be to take my variac (which also is galvanically isolated) and use a voltage doubler to get around 650V rectified. With that I could experiment with different G1 voltages.

Using a variac is the right way to experiment fast. 650v is going to be a bit high. Start lower and slowly dial up. At G1 voltage of approximately -250v, your PVM should achieve a spot size equivalent to approximately 1000 TVL. Do you know the phosphor triad width as a fraction of screen width for that PVM? It needs to be 1/1333th the width of the screen to achieve 1000 TVL. For vertical resolution, the continuous phosphor strips of the aperture grille will be a huge advantage for this mod compared to slot mask.

Regarding fixing the PVM. It looks like a voltage drop issue for the RGB drive amps. A wild guess though. Can you oscope the cathode pins while displaying HD Retrovision’s SMPTE color bars test ROM? If the approximately 200 volt rail for RGB drive amps has a voltage dip, it will cause cathode voltage to drop, preventing the amps from holding cathode voltage high, which it requires for cutting to black during horizontal rescan. So the right side of the picture smears white. If cathode voltage is allowed to drop to zero, then the cathode ray current skyrockets and may then cause a spike in current draw on the chassis that causes some protection circuit to trip.

Have you recapped the chassis yet? Dry caps are a common cause of voltage drops.

[LukeEvansSimon]

People might find this interesting for comparison. I think I found the very spot LukeEvansSimon had Alucard standing, although I put a different cape on him. It's right outside the last save point before the final boss.

Thanks. These are great for comparison. Clearly a smaller dot pitch helps the image look better, even if TVL is the same, but CRTs with the same dot pitch can have different TVLs. Also, your CRTs are dot mask, whereas my modded consumer CRT is a slot mask, so the comparison is not an exact match, but still a useful comparison.

In order for a CRT to reach 1000 TVL for 240p content, the following minimum specs are needed:
R=Attribute&Value&Explanation R=RGB Drive Amp Bandwidth&9.6Mhz&RGB cathodes need to be varied 666 times per line from peak voltage for black to trough voltage for white, corresponding to 666 black and white dot pairs, 240 lines per frame, 60 frames per second R=Phosphor triads + mask/grille width&1/1333th width of screen&if every other triad is illuminated, this is 1333 alternating white and black vertical lines and 1000 alternating white and black lines over the first 3/4 of the width of the screen R=Cathode ray spot size width&1/1333th width of screen&the cathode ray sweeps across the phosphor triads so it cannot illuminate every other triad unless the ray is thin enough

Looking at my consumer CRT collection, most of the TVs above 27-inch meet these requirements, except for the cathode ray spot size (which can be modded as described). Some of my consumer CRTs cut it VERY close to the RGB drive bandwidth, having a bandwidth of 8 to 9mhz. My pro CRTs easily surpass these requirements with a large margin, except cathode ray spot size (again moddable). If these requirements are not met, the mod will still increase TVL for 240p gaming, but will fall short of full 1000 TVL. For example if the RGB drive amps only have 8mhz bandwidth, then TVL will only be able to reach about 833 TVL, because the amps will not be able to increase cathode voltage from peak to trough fast enough to draw the vertical alternating lines without the colors bleeding into each other.

RGB drive amps are moddable. OnSemi in California still manufactures and sells high voltage NPN transistors, such as this one, that surpass the specs required for 1000 TVL CRTs. They cost about 30 cents per amp.

[SamIAm]

R=Attribute&Value&Explanation R=RGB Drive Amp Bandwidth&9.6Mhz&RGB cathodes need to be varied 666 times per line from peak voltage for black to trough voltage for white, corresponding to 666 black and white dot pairs, 240 lines per frame, 60 frames per second

This does not take into account blanking time. One horizontal line of NTSC is 63.5uS, but only 52.6uS of that is active video. Furthermore, the vertical overscan and blanking area is quite large: 525/2 - 240 = 22.5 lines per 60Hz refresh.

52.6 x 0.75 = 39.45uS
1000 / 39.45 = ~25MHz minimum frequency response for 1000 TVL at 15KHz horizontal refresh. This adds up, by the way: Common pixel clock frequencies in 240p game systems are 5-7MHz. Saturn high-res mode is 14.3MHz, which is the highest clock pre-Dreamcast and is 704 pixels across per line.

Depending on the TV, you may encounter bottlenecks to this well in advance of the main drive transistors. Some may even employ lowpass filtering to get rid of unwanted high frequency noise. Heck, if the AC coupling capacitors in your transmission lines are only aluminum electrolytics without anything else in parallel, they probably won't pass 25MHz undistorted.


But this is all pretty academic.


LukeEvensSimon, I do have to give you credit. You saw potential, formed a theory, tested a design, and got results.

I'll also say that you were 100% right when you said that TVL is a function of the whole system and not just the size and density of phosphor triads on the face of the display. I think I can say that many of us, including myself, had been mistakenly assuming that dot-pitch was virtually the exclusive factor in determining TVL, while we can now understand more fully how big a role spot size plays.

If a person has a display with the abundance of phosphor triads in the CRT necessary to make this mod work, then it could be worth a look for them. If they prefer the result, good for them, and you've done them a service. For all of this, we owe you thanks.

Having said that, I think that part of the reason why some people still balk at this whole thing is that a phrase like "double the TVL" sets certain expectations. Not being able to see your modded display in person, I can only imagine how it really comes across, but your SOTN before/after shot doesn't look anything like the Mega Man X3 shots from your original OP that you used to illustrate the mod's potential. Upon closer inspection, the high-TVL MMX3 shot still has the beam hitting multiple phosphor triads horizontally and vertically for every pixel. This goes against your premise that a game image can be pleasingly resolved with only one triad per pixel.

When I look at the shots I posted compared to your before/after, the overall impression I get is that dot-pitch is still the dominant factor determining how a CRT looks, assuming all bandwidth requirements are being met. You might succeed in doubling TVL in some strict technical sense, but the result almost certainly isn't going to look like how a quality high-TVL screen would look, and that's why some people find the title of your thread misleading. At the moment, it looks like a more realistic expectation to set, in terms of the upgrade in appearance to the user, would be something like +100 TVL. That and skinnier scanlines for those who are into that sort of thing.

As you've observed, it's a bit silly to fixate on the highest TVL for its own sake when the vast majority of 240p gaming falls within 480 TVL (I'd say even 384 TVL, or 512 pixels per line). Much to the contrary, my personal takeaway from this is that I ought to turn down the focus on my 900 TVL monitor, which was seemingly designed with small spot-size for its HD capabilities (which I hardly ever use). Anyway, I think you'll see more enthusiasm for this mod if you recast it as potentially removing critical bottlenecks, and set aside hitting sky-high technical benchmarks. 350 and 450 TVL sets sit right before the boundaries needed to resolve certain "high-res" 240p games; you've got a mod that might convincingly get them beyond those boundaries. That, to me, is exciting.

Just my two cents.

[ElBartoME]

Using a variac is the right way to experiment fast. 650v is going to be a bit high. Start lower and slowly dial up. At G1 voltage of approximately -250v, your PVM should achieve a spot size equivalent to approximately 1000 TVL. Do you know the phosphor triad width as a fraction of screen width for that PVM? It needs to be 1/1333th the width of the screen to achieve 1000 TVL. For vertical resolution, the continuous phosphor strips of the aperture grille will be a huge advantage for this mod compared to slot mask.

Regarding fixing the PVM. It looks like a voltage drop issue for the RGB drive amps. A wild guess though. Can you oscope the cathode pins while displaying HD Retrovision’s SMPTE color bars test ROM? If the approximately 200 volt rail for RGB drive amps has a voltage dip, it will cause cathode voltage to drop, preventing the amps from holding cathode voltage high, which it requires for cutting to black during horizontal rescan. So the right side of the picture smears white. If cathode voltage is allowed to drop to zero, then the cathode ray current skyrockets and may then cause a spike in current draw on the chassis that causes some protection circuit to trip.

Have you recapped the chassis yet? Dry caps are a common cause of voltage drops.

Yes of course 650V is a lot to start with. I'll start at 0 and work my way to -250V and observe the effects. At least the 14N2 allows me to change drive and cutoff of every colour seperately. The first experiment with the B&O didn't allow any cutoff change at all.

Unfortunately I do not now the triad width but being a 14 inch screen and having 500TVL shouldn't we be able to calculate the width?

So I can confirm that the overcurrent protection is tripping the shutoff of the PVM. If I disable the overcurrent protection the PVM doesn't shutoff anymore and the picture I posted comes up. I discovered that the horizontal output transistor is getting extremely hot. I'm gonna do a recap of the board (thankfully not that many capacitors). I checked the B+ voltage and it's at around 115V as it should be. I need to measure the voltage at the collector of the HOT.

Yeah I can also put the SMPTE color bars on it and measure the 180V of the RGB voltage rail. I think I already measured that voltage with a black screen and it was at 180V.

I also try to get some replacement HOTs in case it will blow.

[LukeEvansSimon]

...
I also try to get some replacement HOTs in case it will blow.

I am eagerly looking forward to the results you get. It will be interesting to see if you can drive your lower end 600 TVL ap grille Sony PVM at 1000 TVL. They have plenty of RGB drive amp bandwidth, plenty of phosphor triads along each horizontal line, and the continuous strips of phosphor vertically won't cause any distortion issues vertically due to the needle thin beam colliding with the mask.

52.6 x 0.75 = 39.45uS
1000 / 39.45 = ~25MHz minimum frequency response for 1000 TVL at 15KHz horizontal refresh.

Yeah, my math is off, but mainly due to using half the period in my calculation. I will update my OP at the top of this thread with ~25Mhz minimum requirement calculation. Thanks. The Panasonic consumer CRT I am experimenting on has 140mhz RGB drive amps (datasheet). Some consumer CRT TVs have lower bandwidth amps around 8 to 9mhz amps, but this Panny and my JVC D-Series have over 100mhz RGB drive amps. So the amps likely won't be the bottleneck for most CRTs, but some may be bottlenecks by the drive amps.

We are making progress here in this thread. We started with "this is impossible, horrible, rage, rage, dot pitch" to "technically if you use the actual definition of TVL as opposed to thinking of it as blast processing then yes you can increase TVL" and then the predictable "nobody likes the high TVL look anyway" and "your early prototype results have issues, give up". Many CRT collectors own monitors with equal TVL and yet different dot pitches, and we already know that all things being equal, smaller pitch is better. However, it is not a bottleneck preventing most CRTs from increasing their TVL.

The variations in CRT gun triode design are huge. The shape and position of the G1 anode, the size of the G1 aperture, the shape of the cathodes and position relative to the G1... this varies allot between CRTs, and these design differences will make the results of this mod vary considerably from CRT to CRT. Since we lack datasheets, for this mod to progress, we need more people in the community experimenting with this mod, not naysayers pissing on anybody that wants to push the envelop. With more people experimenting with this mod on different CRTs, we will find CRTs that work better than others due to CRT gun design, as well as phosphor/mask design. We will find ways to tweak the mod for better results.

Having played with this mod, my intuition says that an aperture grille CRT holds the most potential due to their vertically continuous strips of phosphor triads. In an aperture grille CRT, there is no mask running in the same horizontal direction as the beam, which can block the beam from displaying when the beam is as thin as a needle for the darker colors and lined up with the mask. This is the bigger issue with large dot pitch slot mask tubes, not the number of triads along the horizontal as people in this thread suggested. There are plenty of triads.

[6t8k]

The better metric for resolution is simply number of horizontal lines by number of vertical lines.

It's impossible. The reason why TVL merely measure across a horizontal span equal to the whole vertical span of the picture, is to account for different aspect ratios, as you always get a square this way. Otherwise, the measured values across CRT TVs with different aspect ratios would be pretty nonsensical: a 16:9 screen isn't sharper just because it's wider than a 4:3 or 5:4 screen!

What you could do, however, is take the square and offset it across the screen in a standardized way, taking multiple measurements until the whole screen is covered, and average the values. This would account for nonuniformities in picture quality across the screen, which CRT TVs inevitably have. You could even weight different areas, as usually, areas you don't look at that often, like the corners, are not as sharp as some area around the center of the screen for example. On the other hand, you'd be sort of patronizing users with such a weighting, who're you to tell them where to look :p There sure are possibilities to overdo it here.

Also note that for vertical resolution of a CRT TV set, you have to take the Kell factor into account (effective vertical resolution = number of visible scanlines per field/frame × Kell factor).

You probably know, but to state it for good measure, the meaning of the term 'resolution' in analog display contexts is quite different from its colloquial meaning in digital display contexts. With the latter, we usually use it to refer to pixel dimensions, while with the former, we refer to a sharpness capability that, when 'translated' to the world of digital displays, would be more close to pixel density. WP calls it a misnomer.

The chroma signal feeding into the cathodes may encode 1000 TVL

TVL by definition only manifest themselves in a logical sense when photons are emitted from a display's surface and a person sees them. If you don't add luma to chroma, you get no picture.

for 240p gaming, [...] the game systems cannot address anything above 480 TVL (in fact they are much lower)

What about for example the Sega Saturn's 704x240 / 704x224, and the Amiga 500's 736×483i / 736×567i modes?

Edit (changed nothing else): I was playing advocatus diaboli with that rhetorical question, mentioning that there are real-world 15kHz video game systems that can output more than 640 dots per scanline. The reason why this irrelevant for TVL is outlined in my previous post.
Note how you generally get a sharper picture by swapping your CRT TV with a higher-TVL one, while sticking with the same video signal, even if the number of dots in the signal per scanline within an electron beam travel distance equal to the height of the screen (let's call that X) is below the TVL value of both CRT TV sets. But it's relevant insofar as if X is larger than your CRT TV set's TVL value, you start being unable to distinguish some dots (might in practice happen sooner because TVL are measured using black and white stripes, i.e. maximum contrast).

LukeEvensSimon, I do have to give you credit. You saw potential, formed a theory, tested a design, and got results.

Yup Please don't get me wrong, this is very cool.

[Downcry]

Everyone stop arguing with OP and let him continue refining this cool mod in peace

[Josh128]

for 240p gaming, [...] the game systems cannot address anything above 480 TVL (in fact they are much lower)

What about for example the Sega Saturn's 704x240 / 704x224, and the Amiga 500's 736×483i / 736×567i modes?

And also the SNES's 512x448 mode. But its irrelevant. In single horizontal line, even at 256x224, there is no space between horizontal pixels sent by the SNES. The spaces you see are determined by the display, not the source.

[LukeEvansSimon]

for 240p gaming, [...] the game systems cannot address anything above 480 TVL (in fact they are much lower)

What about for example the Sega Saturn's 704x240 / 704x224, and the Amiga 500's 736×483i / 736×567i modes?

And also the SNES's 512x448 mode. But its irrelevant. In single horizontal line, even at 256x224, there is no space between horizontal pixels sent by the SNES. The spaces you see are determined by the display, not the source.

Thanks, I wasn't aware of the Saturn's ability to address 528 TVL or Amiga's ability to address 552 TVL. Regardless, my point still stands. For 15khz games, the distinction between "addressibility" vs "resolution" when doing 15khz gaming on CRTs is a valuable distinction. 552 addressible TVL is significantly less than the 750 TVL to 1000 TVL resolution of "high TVL" CRTs.

To make this more clear, when measured in units of TVL, a 15khz CRT’s resolution (as opposed to addressability), can be calculated using the following formula. The mod described in this thread exploits the fact that for many consumer CRTs and even professional CRTs, the "SpotTVL" is considerably lower than "CathodeTVL" and "PhosphorTVL". In fact, SpotTVL is typically 1/4 to 1/2 PhosphorTVL for most CRTs both consumer and professional.

Code: Select all

CRTResolutionTVL = minimum( CathodeTVL, PhosphorTVL, SpotTVL )

...where...

CathodeTVL = (RGB drive amp bandwidth in megahertz) * 39.45

PhosphorTVL = (3/4) * (CRT width) / (phosphor triad width)

SpotTVL = (3/4) * (CRT width) / (cathode ray spot width)

[Ryoandr]

The fact that higher resolutions are not fully resolvable on most TVs was actually used as dither for fake transparency on SNES, with the 512 horizontal res.

Jurassic Park uses it on the HUD
Kirby 3 uses it for foreground plane
Sailormoon S puzzle uses it for menu background animation.

[ElBartoME]

I fixed the 14N2!



It just seemed the HOT was damaged. I replaced it and it works perfectly now. I will finetune the convergence and focus and then start making the voltage supply for the experiment.

[LukeEvansSimon]

I fixed the 14N2!

It just seemed the HOT was damaged. I replaced it and it works perfectly now. I will finetune the convergence and focus and then start making the voltage supply for the experiment.

Impressive work! A friend is going to lend me their 14M2, which is a 600 TVL aperture grille. It has vertical collapse, so I first need to fix the vertical collapse before I try the TVL boosting mod, which may take me a few weekends of work. So you will likely beat me to trying this mod on a "mid-range" aperture grille PVM.

Regardless of people's lack of enthusiasm in this thread, I have high hopes that we are going to find a very convincing implementation of the mod, which will give CRT collectors more options. The super low end consumer slot mask CRT that I tried on may simply not be what convinces people, and maybe the mid-range PVMs end up being the sweet spot that makes this mod popular.

[ElBartoME]

Nice, the more the merrier!

I picked up a PVM-1442QM today as well for the test and I think I might start on it. It's a 600 TVL as well and has a very slight burn in so if it goes boom it's not a big loss.

I have some 800 TVL PVMs here that I'm not yet willing to sacrifice for the test.

[LukeEvansSimon]

Nice, the more the merrier!

I picked up a PVM-1442QM today as well for the test and I think I might start on it. It's a 600 TVL as well and has a very slight burn in so if it goes boom it's not a big loss.

I have some 800 TVL PVMs here that I'm not yet willing to sacrifice for the test.

On the consumer TV I modified, the neckboard circuit design has a thru-hole jumper soldered to connect G1 to ground. So I was able to modify the neckboard by just desoldering a small jumper clip, as opposed to cut a trace. So the mod was totally reversible and not invasive.

The real thing to worry about for this mod is the fact that every triode has a breakdown voltage between its cathode and anodes. As long as the voltage between cathodes and anodes is below the breakdown voltage, this mod will not damage anything. However, if you go above the breakdown voltage, then electrical arcing between cathodes and anodes will occur. This can cause a very short duration high current through the cathodes, which is not good for them as it can burn off their barium layer. The breakdown voltage will vary according to the electron gun's triode design. The G1 anode is a metal tube surrounding the cathodes. The smallest distance between the cathodes and G1 determines the breakdown voltage for K to G1. Similarly for the G2 to G1 breakdown voltage, but the spacing between G2 and G1 is much larger spacing so the breakdown voltage is much higher. This is why you want to start with lower voltages and slowly dial up because you will likely find you can obtain a 50% reduction in spot size at a voltage well under the breakdown voltage.

Starting with an 800 TVL CRT for this mod may look ridiculous if you try to double TVL. The spot size will be thinner than the tip of a sharp needle. However, if you just shoot for boosting from 800 TVL to 1000 TVL, the change will likely be subtle and look great. Regardless, I am eager to see your 600 TVL boosted to 1000 TVL. It will be a big jump in resolution.

I made a collage of my low end 27-inch consumer slot mask CRT versus SamIAm's professional high TVL dot mask CRTs. The consumer TV's modded spot size is somewhere between the 700 TVL and 900 TVL spot size. I am not saying the modded TV looks as good as the pro monitors. I don't care to have a flamewar about that, but an objective comparison of the spot size is possible.
Top Left = unmodded consumer TV
Top Right = modded consumer TV
Bottom Left = 700 TVL pro CRT
Bottom Right = 900 TVL pro CRT

[ElBartoME]

I probably need to cut a trace but I haven't looked at the neckboard of the 1442QM yet.

I'm gonne build the supply today and use a voltage doubler. I think a capacitance of 1µF should be more than enough but I'm thinking of putting a resistor in line to G1 to limit the current in case of a short-circuit. What do you think and do you have a suggestion for a value?

[fandangos]

Can you, please, take pictures of the modding process?
This would help a lot others trying to replicate this.

[LukeEvansSimon]

I probably need to cut a trace but I haven't looked at the neckboard of the 1442QM yet.

I'm gonne build the supply today and use a voltage doubler. I think a capacitance of 1µF should be more than enough but I'm thinking of putting a resistor in line to G1 to limit the current in case of a short-circuit. What do you think and do you have a suggestion for a value?

I used a 3.3uF film cap from G1 to ground, which was way overkill. 1uF should be more than enough enough, but the parasitic elements of the greater tube and circuit are complicated and hard to properly simulate in something like LTSPice. It is very important that the capacitor is very low ESR such as a ceramic or film capacitor. Low ESR liquid electrolytic is not going to work because they are not low ESR enough. You can always just oscope the G1 to ground voltage to see if it is smooth enough with 1uF. Even very short duration voltage ripple will be undesirable because it will cause the spot size to expand and contract for a few "pixels", which will look excessive blooming or even worse it will look like image "jittering".

You can use a fuse between the G1 pin and the capacitor, if you are super paranoid. This is something you can simulate in LTSpice, using a K to G1 capacitance of 5pF and G2 to G1 pF of 5pF. The highest current draw will be during power up. 0.02 amp fuse should be enough to ensure that the fuse doesn't blow during power up, but will blow in the case of a short.

I built a CRT symbol and model for LTSpice, as well as a schematic for the entire neckboard. Simulation showed G1 voltage fluctuate without a low impedance cap. I also simulated cathode and anode shorts, as well as failures of the added power supply and failures of the TV's power supply. The worst edge case is: TV power supply failure, cathode to G1 short, and yet the additional power supply does not fail. This is highly unlikely, but if it does happen, then chassis ground can become a higher voltage than the chassis positive rails, which would damage polar components such as electrolytic capacitors. The safe guard against this is to either add diodes from ground to the cathode pins, or diodes in series with the cathode pins. In my simulation I put diodes from ground to cathode pins. In my real life prototype, I did not bother to safe guard against this edge case. In the final version of this mod, I'd use a secondary winding on the exposed core of the flyback for the added power supply so that if the main supply fails, the added power supply also fails. I'd also use a fuse that blows in case of a K-G1 short, but again, only in the final version of the mod. My bench's current limiter is enough safety for the prototype.

Can you, please, take pictures of the modding process?
This would help a lot others trying to replicate this.

The prototype version of this mod uses a bench power supply and is very simple, and so there isn't really much to show in the circuit mod pictures. The changes are all made on the CRT neckboard. A previous poster in this thread described how you can build a power supply for a few dollars of parts by winding magnet wire around the exposed core of the flyback and use a single diode in series with the lead coming off of that winding. I am going to prototype that over the next few weekends. That will benefit from some pictures. I also will experiment with adding a pot to allow for tuning the cathode ray spot size for the desired TVL. My goal is for the mod to cost under $2 of parts.

Code: Select all

1. Cut trace between ground and the trace to the G1 anode.  My TV had a jumper between ground and G1, so desoldering that jumper was all I needed to change.  Another option is to pull the CRT neck socket pin for G1.

2. Connect an 1uF and approximately 350v film capacitor between ground and the trace to the G1 anode.  This keeps the G1 voltage from jittering.  It is optional if the power supply you use in the next step has a good smoothing capacitor built in.

3. Connect the positive rail of an approximately 250v power supply to the ground leg of the capacitor from step 2.  A higher voltage will make squeeze the cathode ray to make it even thinner and the spot size even smaller, but if you go too high, you will get arcing between the G1 and cathodes.

4. Connect the negative rail of the 250v power supply to the G1 leg of the capacitor from step 2.

5. At this point, the picture will be very dark and all colors crushing into black because a greater force is required to pull the cathode ray through the smaller electromagnetic aperture created by the charged G1.  Using the 240p test patterns (or equivalent), turn clockwise, the screen potentiometer on the flyback to increase the screen voltage until proper black levels are restored.  The screen voltage determines the acceleration force applied to the electrons in the cathode ray.

[ElBartoME]

Thanks for the info!

I used a single diode rectifier with 1µF capacitor afterwards. Not the best solution as I can still see some waviness, but it will do for now.

I'll just quickly show my results. I can do a writedown of what I have been doing specifically later.

Here an image with 0V on G1:

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

Here an image with -220V on G1:
https://i.imgur.com/F32Gusu.jpg

You can see a bit of change but not as much as I thought. The other problem is I can not raise G2 high enough. If I go to -250V the screens is almost completely dark and I only can see image in a dim red.

I tried to adjust the cutoffs to no change. The 1442QM has bias and gain controls in the front but they don't seem to have much of an effect. I'm not sure if I'm overlooking something.

[LukeEvansSimon]

Thanks for the info!

I used a single diode rectifier with 1µF capacitor afterwards. Not the best solution as I can still see some waviness, but it will do for now.

I'll just quickly show my results. I can do a writedown of what I have been doing specifically later.

Here an image with 0V on G1:

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

Here an image with -220V on G1:
https://i.imgur.com/F32Gusu.jpg

You can see a bit of change but not as much as I thought. The other problem is I can not raise G2 high enough. If I go to -250V the screens is almost completely dark and I only can see image in a dim red.

I tried to adjust the cutoffs to no change. The 1442QM has bias and gain controls in the front but they don't seem to have much of an effect. I'm not sure if I'm overlooking something.

Yes, much less change at -250V than I saw on my TV prototype. Are you able to oscope the G1 pin to ensure it is actually at -250V relative to chassis ground? Is your flyback's screen potentiometer already turned up without much room for increasing even before the mod? My consumer TV's flyback pot had lots of room to increase before the mod, about 180 degrees turn before the mod. With G1 at -244v on my consumer TV, have to turn the pot about 90 degrees clockwise to calibrate black levels. So I still have another 90 degrees left on the pot and I can still turn up the flyback's screen pot so high that the screen is almost entirely white, even when displaying a black picture and G1 at -244v!

[ElBartoME]

I had my multimeter connected between G1 and GND to measure the voltage.

Nope, I have to turn the G2 pot almost completely down with G1 at 0V. It seems the PVM doesn't have enough of a range for G2 in order to change G1 significantly.
There is the possibility to boost the G2 voltage with a converter but I think that is too much work.

It's a shame. Maybe I'll take a look at the 14N2 and see if it's better there. I had to scrap traces on the 1442QM but the mod is reversible if I bridge the cut.

I'm curious to see if you have more luck on the 14M2.

[LukeEvansSimon]

I had my multimeter connected between G1 and GND to measure the voltage.

Nope, I have to turn the G2 pot almost completely down with G1 at 0V. It seems the PVM doesn't have enough of a range for G2 in order to change G1 significantly.
There is the possibility to boost the G2 voltage with a converter but I think that is too much work.

It's a shame. Maybe I'll take a look at the 14N2 and see if it's better there. I had to scrap traces on the 1442QM but the mod is reversible if I bridge the cut.

I'm curious to see if you have more luck on the 14M2.

If the screen voltage coming off of the flyback is not yet rectified, you could use a voltage doubler.

[ElBartoME]

900V is coming off the flyback and going to the neckboard. From this voltage the focus and screen voltage is derived.

According to the service manual the voltage range for G2 is between 320 and 440 volts. There is a series resistor to the potentiometer of 680k. I could change that value to squeeze out a bit more voltage like say 550 volts.

[buttersoft]

LukeEvansSimon, thank you for cleaning up the OP!

EDIT: ok, i think my confusion has arisen because i never realised the G1 was connected directly to ground. I knew the guns were operating at negative voltages, but never twigged to that first part. I think i'll just delete the rest of this post...