800XL NTSC Artifacting problems

[_The Doctor__]

I noticed that jingle disk and a host of other software do not artifact in the correct colors on the 800xl. I think I found out how the artifacting colors are generated But want to know about the fixes to date. Most XL' show strange shades or reversed colors.... on the 130XE Jingle disk is red-orange,blue-green,white, or black. I just recently got my 800xl to make it produce it as Green,Red,white,black.. Christmas tree is finally green whoo-- hoo!

 

Any one else know how to go about fixing artifacting problems? aside from changing ctia to gtia.(that is a given)..

Even gtia based machines artifact differently over the years.... so what do about this long standing quandary? post em up!

[_The Doctor__]

We need some experts here, as I am not all knowing, and could sure use the help.

 

Timing for normal ntsc is handled by crystal oscillator etc. and of course theirs a potentiometer and a couple transistors a basic video circuit and our friendly gtia/antic chips.

 

The artifact have a sneaky capacitor resistor network to make the things happen, but it's a part of the existing video circuit mess, it also explains why Atari's engineers made weird choices and messing up the circuit all the time. They forgot that in the 800xl c53 is responsible for detail smoothing as well as artifact colors! So the more they messed about, the worse it got!

 

I selected a 95nF to 1uF (microfarad) cap. (same as on most XE mem chips) in parallel with the existing cap. at c53 on the 800xl. This is part of the r/c network and goes to the transistors from there. A good deal of the video mods do not take this into account and their results vary. The choice must be made so that the artifact colors are close to each other in all output modes as they are hooked up all by themselves, ie rf connected without monitor connected, composite monitor connected, and last but not least if you did the chroma pick off line. with the chroma and luma connected as well. This was just a rough choice to get me what I wanted quickly. If someone with more experience would like to balance the circuitry with other mods It would be great! This modification somehow needs to be incorporated properly with the other mods out there. This looks pretty darn good when in modes that use it, I added it to one of the clear pic mods and they clashed, just a bit too strong! probably duplication in function. So I hope someone can finesse the thing. Looks sharp enough in luma chroma. looks sharp in the modes that use the cap. Looks somewhat muddy for standard gr.0 screens etc when in composite mode.

 

Have at and make %22 more Atari 800XL's happy again!

Edited April 3, 2009 by _The Doctor__

[atariksi]

I noticed that jingle disk and a host of other software do not artifact in the correct colors on the 800xl. I think I found out how the artifacting colors are generated But want to know about the fixes to date. Most XL' show strange shades or reversed colors.... on the 130XE Jingle disk is red-orange,blue-green,white, or black. I just recently got my 800xl to make it produce it as Green,Red,white,black.. Christmas tree is finally green whoo-- hoo!

 

Any one else know how to go about fixing artifacting problems? aside from changing ctia to gtia.(that is a given)..

Even gtia based machines artifact differently over the years.... so what do about this long standing quandary? post em up!

 

Apparently the GTIA chips are not all alike in their hue output; I once substituted a GTIA from Atari 5200 into a 800XL and the colors were way off until I adjusted the Pots inside the machine. Unless Atari knew how different the chips were, it would have been hard to mass produce the computers with consistent color output.

 

I prefer using GPRIOR mode 0 than playing around with artifacting unless you just want a different color and exact color does not really matter.

[Rybags]

Individual chips are probably different, which warrants the adjustable pot in the first place.

 

Irrespective of that, GTIA also supposedly is 1/2 a colour clock offset from CTIA which would give different artifacting.

Although I think the differences go beyond that - if a single hires pixel was the only difference, you could just shift the picture by a pixel and get the same results, which in reality doesn't happen.

[sup8pdct]

It also depends a lot on your screen. some support artifacting much better then others. some don't at all. Yet others will show different colours.

 

James

 

I noticed that jingle disk and a host of other software do not artifact in the correct colors on the 800xl. I think I found out how the artifacting colors are generated But want to know about the fixes to date. Most XL' show strange shades or reversed colors.... on the 130XE Jingle disk is red-orange,blue-green,white, or black. I just recently got my 800xl to make it produce it as Green,Red,white,black.. Christmas tree is finally green whoo-- hoo!

 

Any one else know how to go about fixing artifacting problems? aside from changing ctia to gtia.(that is a given)..

Even gtia based machines artifact differently over the years.... so what do about this long standing quandary? post em up!

[_The Doctor__]

that make sense, as the adjustment pot for color is needed perhaps a variable capacitor should have been used at c53 for artifact. Since the r/c network would need to be flexible for all users with different chips and configurations. Could be we end up with a list of fixed cap values for common models(lower cost) and a variable cap for all.

 

My daughter is pleased with the outcome so far. She loves jingle disk. My wife was completely happy that the tree is now green and the red is now red. In fact I heard her leave a breathe of relief. Some purists would like the tree to have a bluish tinge but that will be when I locate the variable cap and put it in. Unless some experimenter out there come in with another choice.

 

I am starting to remove the clear pic mod from my other XL a piece at a time to see where an overlap occurs within the modification. Once done we will re apply the mod in different order to see if I can isolate what step overcompensates with the new r/c value included.

[atariksi]

Individual chips are probably different, which warrants the adjustable pot in the first place.

 

Irrespective of that, GTIA also supposedly is 1/2 a colour clock offset from CTIA which would give different artifacting.

Although I think the differences go beyond that - if a single hires pixel was the only difference, you could just shift the picture by a pixel and get the same results, which in reality doesn't happen.

 

Well, the TV screen consists of R,G,B phosphers... if we had the CRT perfectly synched so that the resolution was 3X number of triplets and phase locked to first phospher, shifting by one hires pixel would result in a consistent result.

[Rybags]

The problem there is that individual TVs have different dot pitch, and smaller TVs tend to have less definition.

 

e.g. I've got this little 4.5 inch thing (or whatever size it is) and the horizontal resolution is such that it probably doesn't even do 320 pixels across properly.

[_The Doctor__]

it's not so much the the dot pitch, I used 6 different televisions and 3 different monitors, the results were all the same, and thank god for that! I suspect the television have got things pretty well aligned, and they tolerate disturbances well. How else can you watch something from antennae, cable, dish, macrovision encoding, vcr, dvd, computer and kids toys and they are have good color, the guns are align to hit the proper phosper locations. shift when it get triggered and be off by anywhere near 33% and you get a color shift. so 15% 30% 45% 60% 75% of the color clock might be attempted to experiment with, try shifting a single red, single blue, and single green pixel, then try shifting a single white pixel. see what you get with each! strange stuff. tints and colors.

Those tiny televisions with the large lcd pixels use to interlace the frames to get a pictue, that is why they so low def. and were never mega sellers. I can imagine using one for anything... I get a better picture on my cell phone!

[Rybags]

Artifacting is more a product of the luma interfering with the chroma at the signal level, rather than phenonema due to what's happening in relation to the electron beam and what colour phosphor happens to be there at the time.

[_The Doctor__]

yes and no, artifacts can be produced on a color tv without any color clock in the signal at all, a true black and white signal received by the color tv was proven produce a pseudo rainbow when actors wore checkered/striped shirts, ties and jackets. It was annoying enough that the industry avoided those patterns in the wardrobe for quite some time. While all signals affect each other to some degree, some are more pronounced than others. I did the rainbow color effect experiment in technical school. Not as much fun as charging up capacitors and wiring up locker handles, but interesting none the less.

Edited April 5, 2009 by _The Doctor__

[Rybags]

Still, the resultant is because of the luma changes generating high frequencies though. If the same tie had something like thin blue/green stripes of roughly equal luma, you wouldn't get the same effect.

[_The Doctor__]

Still, the resultant is because of the luma changes generating high frequencies though. If the same tie had something like thin blue/green stripes of roughly equal luma, you wouldn't get the same effect.

 

In order to artifact color you have to have a lack at least one of the phosphers lit and one at least one phosper signifigantly lacking intensity or off, this is the whole principle for producing color on any tv, that is why they all use red green and blue. all on is white all off is black. vary the intensity of any one gun and you get colors and tints. they all rely on the distance from the phosphers to give the illusion all the other colors are there. None have bothered to us optics to combine all 3 into a single point save some experimental projection tv's. In order for single point color to work the picture would have to be digitally encoded an pixel point perfect. This is only now starting to be possible with matrixed screens. Old tv relied on the human eye blending the colors at a distance and in a sense we correct the colors we see when the grouping shifts slightly

Edited April 5, 2009 by _The Doctor__

[Bryan]

The color mask and phosphor doesn't have anything to do with how RGB color is derived within a television circuit. The TV generates an analog signal for the 3 color guns. It doesn't care about where the phosphors are or what their resolution is. That is all handled by the mask in the picture tube. It would be just about impossible to aim the guns accurately enough to hit any particular phosphor dot reliably.

 

Artifacting works because alternating on-off pixels on the NTSC Atari creates a waveform with the same frequency as the color carrier. The TV will then interpret this as color by examining the phase and level of the signal in relation to the colorburst. The 7800 has a feature where the colorburst can be disabled so high-resolution graphics won't produce anything but black & while.

 

The artifact colors produced on various models is determined by the relative propagation delays of the luma and chroma circuits because the colorburst comes from chroma, and our 'fake' artifact color signal comes from luma. If you look at the schematics, the luma circuit is very similar from model to model but the chroma circuit was revised several times.

[_The Doctor__]

The color mask and phosphor doesn't have anything to do with how RGB color is derived within a television circuit. The TV generates an analog signal for the 3 color guns. It doesn't care about where the phosphors are or what their resolution is. That is all handled by the mask in the picture tube. It would be just about impossible to aim the guns accurately enough to hit any particular phosphor dot reliably.

 

Artifacting works because alternating on-off pixels on the NTSC Atari creates a waveform with the same frequency as the color carrier. The TV will then interpret this as color by examining the phase and level of the signal in relation to the colorburst. The 7800 has a feature where the colorburst can be disabled so high-resolution graphics won't produce anything but black & while.

 

The artifact colors produced on various models is determined by the relative propagation delays of the luma and chroma circuits because the colorburst comes from chroma, and our 'fake' artifact color signal comes from luma. If you look at the schematics, the luma circuit is very similar from model to model but the chroma circuit was revised several times.

That is kinda the point. On the crt so long as the guns hit 3 phosphors of different color we get what we want, the grouping, alignment and size of the three guns are the same no matter what, the order of the phospor must repeat sequence but so long as alignment of the gun group remain constant it does not matter if it hits an exact spot on the screen, since three different colored phosphors are always hit. Controlling color on an lcd is an entirely different can of worms. It allows pixel specific control down to the element(phospor) level.

All the circuitry is pretty much variations on the same theme, but truly digital control at some point is occurring on lcd or similar output matrices. That distinction was made in the former post. maybe I needed to state crt vs lcd/digital more clearly.

Edited April 5, 2009 by _The Doctor__

[Fröhn]

It's actually pretty simple: The chroma is seperated from the luma by a simple analogue filter with cutoff frequency somewhere near 3 MHz (NTSC) or 4 MHz (PAL). This is the reason why you can use fast enough luma changes to generate an artificial chroma signal.

[Bryan]

That is kinda the point. On the crt so long as the guns hit 3 phosphors of different color we get what we want, the grouping, alignment and size of the three guns are the same no matter what, the order of the phospor must repeat sequence but so long as alignment of the gun group remain constant it does not matter if it hits an exact spot on the screen, since three different colored phosphors are always hit. Controlling color on an lcd is an entirely different can of worms. It allows pixel specific control down to the element(phospor) level.

 

This is kinda true. In reality, a single pixel may hit 2.1 Reds, 1.7 Blues, and 1.5 Greens (depending on tube resolution, focus accuracy, etc..) so the color doesn't become precise until a larger area is lit. The alignment of the guns can actually be quite imprecise. Drawing a thin white grid often shows convergence issues (gun misalignment) in different parts of the screen.

 

All the circuitry is pretty much variations on the same theme, but truly digital control at some point is occurring on lcd or similar output matrices. That distinction was made in the former post. maybe I needed to state crt vs lcd/digital more clearly.

 

Previously it sounded like you were mixing the ideas of artifacting and the way picture tubes display color, as if color came from the gun being turned on and off as it passed over different colored phosphors when in fact the gun is on the whole time due to the 3.58MHz filtering and the mask blocks the beam from hitting the wrong phosphors. On a s-video/chroma-luma monitor, displaying a picture with these patterns simply displays a black and white image with the dot pattern intact.

 

It all has to do with the filter that strips 3.58MHz information off to generate color.

[atariksi]

The problem there is that individual TVs have different dot pitch, and smaller TVs tend to have less definition.

 

e.g. I've got this little 4.5 inch thing (or whatever size it is) and the horizontal resolution is such that it probably doesn't even do 320 pixels across properly.

 

When I was in college, they had this capture board for AT&T PCs that captured 576*486 or something similar and supposedly they were capturing a value for each of the primaries (192 color clocks * 3). So I guess TVs are supposed to have a certain minimum number of RGB phosphers per scanline.

[+poobah]

The problem there is that individual TVs have different dot pitch, and smaller TVs tend to have less definition.

 

e.g. I've got this little 4.5 inch thing (or whatever size it is) and the horizontal resolution is such that it probably doesn't even do 320 pixels across properly.

 

When I was in college, they had this capture board for AT&T PCs that captured 576*486 or something similar and supposedly they were capturing a value for each of the primaries (192 color clocks * 3). So I guess TVs are supposed to have a certain minimum number of RGB phosphers per scanline.

 

You could probably get away with almost anything. Don't forget that the signal to the guns is just an analog voltage sweep. more discrete positions in the mask or grille will give finer resolution, but there's no 'magic number" to get a picture. A CRT control circuit has no concept of pixels.