CurtPalme.com Home Theater Forum

[Francisco]

Progress!

After a failing Geforce 7950 GX videocard I've bought another 7900 GTO for a few bucks and modified with the 5bnc breakout.
This one works pretty well and I've managed to display some interesting resolutions.
My test Barco is a stock 909

It locks onto:

2880x1600P @60hz (progressive) now that's very nice!
3840x1600i @48hz (interlaced) cleary this resolution is overlapping scanlines too much.

It won't lock on 3840x2160i 48Hz or above as the Marquee does. Maybe with customized porch settings it will but it's clearly way out of the comfort zone.

[thewolfman]

Can't you post a few screenshots of all the resolutions you can do? Just interested.

So your saying someone made 3840x2160i 48Hz (+) on a Marquee?

[cmjohnson]

Yeah, I think it's been done. But that doesn't mean it was well resolved.

However, it might be pretty decent if it's interlaced. It might even be watchable,
though 48 Hz is still on the low side and some people will find the flicker to be quite objectionable.

3840x2160x48 would require 597 MHz of bandwidth. Interlaced, call it half that at 298 MHz,
within the reach of the best modded machines or not too far past their flat bandwidth limit, anyway.

But at these resolutions you're going to hit several limits OTHER than mere bandwidth limitations.

For the sake of convenience, you can presume that your CRT raster is five inches wide. Just to make the math easy.

Divide 3840 by 5 inches and you get 768 pixels per inch. Since the "correct" setup makes 'square" pixels, then you'll
have 768 pixels per inch whether it's horizontal or vertical.

That requires an electron beam spot size of..... 1/768 = 0.00130208 inches, 1.303 thousandths of an inch.

CRT and lens resolution is often expressed as line pairs per millimeter. So we convert: 25.4 millimeters per inch.

So, 768 / 25.4 equals....30.236 lines, or 15.118 line pairs per millimeter.

The HD10GT17 lenses are rated to resolve 12 line pairs per millimeter, at some MTF value that I don't know.



I'm totally in favor of pushing resolution to its limits. I'm working as hard to see absurd resolutions working in a
Marquee as anybody, but full 4K (3840 verion, not "true" 4096 pro grade "4k") is going to push virtually everything
in the projector past its range of designed performance.

Jitter and noise in the horizontal and vertical deflection circuits will have a significant resolution reduction effect
at 4K, when it is trivial at 1080p.

To get the best possible performance out of the machine, you're going to have to look at ways to quiet down every rail
on the power supplies, reduce generated noise in all the amplifier circuits, and improve shielding in every place it matters.

Oh, it can be done. It may even be worth doing. But make no mistake, even the limits of the lenses will be pushed by 4K.


I calculate 3048x1714 as being the maximum resolution that can be resolved by the optics at the 12 line pairs per millimeter limit.

I would not complain at all to "only" get 3K.

And it makes the bandwidth requirements rather reasonable. That would only need 230 MHz bandwidth if done as progressive at a 60 Hz scan rate.

So I recommend you try for that as your goal. 3K. I think it's as far as we can go without serious mods to a Marquee chassis.

I have seen for myself what happens when the resolution on the CRT face is beyond the abilities of the lens to resolve on the screen. Because I've seen what
a mess that looks like, it's why I recommend trying to limit your experiments to the resolution capacity of the lenses. It's better to be sharp at 3K than
fuzzy at 4K.

[Francisco]

I did make some screenshots but I have not focused the set optimal yet. Also because of the high res. You can't read desktop anymore. In my opinion 4k on a crt isn't worth the haslle. I rather prefer a well resolved 1080p picture and downscale 4k to it in the future. Also because 4k to Hd is an "even" downscale. That's also the problem with 3k it's an odd and uneven upscale. This could even make picture worse upscalng from 1080p or downscaling from 4k

[xmob135lc]

[cmjohnson]

You can't even get sharp enough focus for 4K without serious modifications. You can't focus it that sharp. Electrically you can't get the spot size small enough, and optically your lenses won't resolve 15 line pairs per millimeter. That's what I was so explicit about in my last post.

4K is a stretch goal for the most highly modified machines, while 3K is at least somewhat plausible without investing a large amount of time in extensive projector mods.

Given that you probably can't even SEE individual 4K pixels unless they're bright spots on a black background, I think you can trust me when I say that 4K downsampled to 3K will still be sharper than 1080p upsampled to 3K. IF the projector can handle 3K well.

[Francisco]

[cmjohnson]

Aliasing artifacts around 3K resolution will be so small that I doubt you will ever see them.

[thewolfman]

[cmjohnson]

At 72 Hz you have to increase bandwidth requirements proportionally. And scan rates go up proportionally as well.

Add 12 percent. You get 258 MHz bandwidth needed for 3K @ 72 Hz.

We can scan that, we may not be able to resolve it well due to issues with beam spot size. Beam focus continues to be a limiting factor.

[thewolfman]

Alright, let's say the hackers come up with a way to remove the copy protection, and digitally remove the black borders, how much bandwidth are you left with then? Assuming the movie was not downscaled, but just nipped and tucked!

[cmjohnson]

Stripping HDCP can be accomplished for under 20 dollars, SUPPOSEDLY, by buying certain HDMI splitter/amps that you can buy on Amazon.

A quick Google search led me to that information.

I bought one.

I don't know if it strips HDCP or not as I haven't been able to test that functionality yet, but for the low price, sure, why not?

Added black bars should add virtually NO additional size to the data stream. In fact, a movie with added letterboxing should have LESS data to transport than one that fills the entire digital frame.

You're asking questions that I am not even qualified to give a halfway educated guess on. For all I know, HDMI could be a constant bandwidth datastream, sending (whatever it is) MHz of data whether there's a movie running or it's in standby.

[xmob135lc]

http://www.epanorama.net/sff/Radio/Transmitters/Hybrid%20RF%20Amplifiers.pdf

Funny, if they made projector tube neck a bit longer it could have housed this stuff , though supposedly they' d burn thru the tube face irregardless

not to mention , historically CRT was seen as an 'industry' , mostly to soak up the bulk of plentyfully mass produced and also mostly mediocre parts .

[Francisco]

We should not care about 72Hz for 3K. You don't want to upscale native 1080P to 3K. Just display 1080p 24Hz content at 1080p@72hz.

4K is shot in 30 - 50 - 60 hz. mostly 60hz I think. You don't have flicker at 60hz and we CRT guys don't have motion blur issues

[xmob135lc]

[noos@xp37+]

That´s some great news, Francisco! My Marquee lacked improved lenses, and certainly some other positive aspects of your machine. My aim was to find resolutions and refresh combos of progressive and interlaced formats for specific projectors - to allow them to display (to necessarily resolve) at a pleasing level above full HD. I focused on even multipliers because of known issues - even if visibility is at question or rarely observable. The total clarity or the tranquility of a picture can be effected.

I liked a 1272 at 2560 horizontally + interlace, or a 8500 Ultra with HD8 at 2880 + interlace (will repeat tests when back, then with HD145 and HD144). If interlaced, please do at 72 or 75 or at least 60 hz.

I my video I do estimate the maximum of our lenses to be at about 3k, based on what I have seen and the following interpolation of the data. Nice that the mathematical approach of cmjohnson is in the same area. The Nikon lenses of the Sony HDID resolve what we would need (007 videos). I did not think about the conversion of my 9500 Ultra to use the Nikons anymore - but now I consider this the best part of returning to CRTs!



Resolution test videos can be found here. Scroll down and watch all with a green preview picture:

https://www.youtube.com/playlist?list=PLhLp7Kp-iLM3iKXCfHT9C39f1Ys7WWnUi

[noos@xp37+]

I quote from a pm I have just sent, to show the motivation behind this:

...
"And, I think 1080p at a smooth refresh rate on a CRT is all humanity needs - and scaling down to it is my goal too! Especially because it is the super multiplier of the factor “2” to 2160p. We (all of us that do it) have to see if increase resolution above 1080p/800p is better than downscaling to 1080p/800p. I am pretty sure there will be both tastes and applications for both."
...

[cmjohnson]

Incidentally, by actually measuring a 9" CRT face rather than guessing at it, it appears that a maximized raster can reach 6 inches
in width before reaching the optical distortion zone at the end of the glass.

Since I think most CRT enthusiasts at least attempt to follow the best practice of maximizing raster width, how this changes the
potential resolution capacity of our machines is worth mentioning.

I based my previous estimates on a 5" wide raster. That's not maximized by any means.

A 6" wide raster could handle 3657.6 by 2057.4 in 16:9 anamorphic squeeze mode, and 3657.6 by 2743.2 in 4:3 mode,
basing this on the lens limit of 12 line pairs per millimeter.


Once upon a time, NTSC at 425 lines was considered "all you'd ever need" and then computers came along, and data grade
projectors, and scalers, and a long climb up the resolution ladder and it kept getting better and better. For a good long while
before I had an HDMI input card I ran at 800x600 via a Faroudja DVP2200 scaler and the picture quality was wonderful, and not
just because of the phenomenal color rendering But then going to 1080p looked like looking out a window.

And now I've seen 4K on OLED flat screen sets running demo material and it shows me just how much room for improvement
there is over 1080p.

I'm all for pushing the limits of CRT technology because I know it can be done and I know the image quality improvements, if done WELL, are well worth the effort.

For 9" projectors that were ALL made to handle 2500x2000 before HD even existed, and with some being rated for 3200x2560 even back then, 1920x1080 really doesn't seem like such a big deal. And in truth, it's not. Not for THESE machines.


The challenges we face in getting the maximum possible peformance of these machines, though, are several: Bandwidth, spot size, beam stability, optical resolution. We can do little about the last one except to attempt to get the best lenses made, but
the rest can all be tweaked and tuned for meaningful improvements.

I am focusing on (if you will pardon the pun) the focusing systems which also are involving the deflection systems due to a perceived need to go to the shorter Barco deflection yoke so the longer Barco focus yoke will sit in the right location on the tube neck.

Mike Parker is working on video chain bandwidth extension.

Who is working on improving the stability and noise reduction in the deflection systems? We're going to need that.

[thewolfman]

How's it going with the Barco coils and deflection mods + MP bandwidth extension?

And btw, last night before going to sleep, I got this idea one should make small marks with a black pencil, on the tube neck, and take a photograph with every 5 mm. And then scrutinize the sharpness with you monitor. I think I'll do that sometime.

[cmjohnson]

I've temporarily suspended that work because I picked up four more Marquees and a lot of 48 P19LUG tubes, of which better than half test as perfect, with zero wear or imperfections of any kind. In total I got three pickup truckloads of stuff and I have been putting all my efforts into refurbishing all four projectors including upgrading the two 8500s to 9500s.

Even though CRT projectors don't sell like they used to, a freshly retubed Ultra will still bring a certain amount of money, so I'm chasing that dime as hard as I can.

After I've got all four projectors rebuilt and fully tested then I can get back to the special HR project.


I thought of doing an interesting little comparison, which I think I am going to do just for the interest of it.

I have green P19LCP, P19LUG, and these "special" 09 P19 tubes. I will take one of the Ultra chassis, equip it with my
MP modded VIM and neck boards, and put one each of each type of tube (all greens) into it for the most direct comparison
of tube sharpness vs. type that I think I can possibly manage. See for myself, side by side, the relative sharpness of an LCP vs a LUG vs the HR tube.

The HR tube is supposed to be the sharpest.

But the part number on the tube is 09 P19LCP. If 09 indicated a tighter spec electron gun, (which is only speculation on my part) then why wouldn't they start with a LUG base electron gun design instead?

I'm no CRT designer, but it would make more sense to me to call the sharpest version the 09 P19LUG if in fact the 09 prefix indicates a higher resolution electron gun.