vertical lines on many XE computers

[flashjazzcat]

This is very weird. I thought I was going mad last week until I realized it was the cable linking the luma and chroma. Obviously not so in your case. From what you say, we can rule out the TV. Something's been missed. I just double-checked with the (excellent) SysInfo program, and this is what you should be seeing:

 

 

Or an approximation thereof. If taking the signal straight from GTIA works better for you, perhaps that's the way to go. I'm out of ideas.

 

By the way - sorry if we seem to ask questions repeatedly, but even the best of us miss things at times, and we're just trying to understand your set up and eliminate things one by one.

Edited January 7, 2010 by flashjazzcat

[Jacques]

Thank you All for trying to help. I'll check the 800XL carefully again this evening, as the mod looks to be done okay, so maybe there is a faulty element or my video circuit or sth... I may also try to use again directly the GTIA's COLOR signal but I'll disconnect this signal completely from the rest of the video circuit. If I'm not wrong, then there should be no artifacting anymore, otherwise the pure CHROMA would have artifacts in itself

 

P.S.

The picture from my other 130XE's S-video on Sysinfo looks perfectly the same (no)artifacts-wise as on your screenshot, so definitely this 800XL has some illness still And my cable doesn't have any connection between LUMA and CHROMA, which would be ideal problem to eliminate

Edited January 7, 2010 by Jacques

[Rybags]

Don't suppose you've got an oscilloscope?

 

I've used mine in diagnosing the video signal - very easy to setup a trigger mechanism via a program to the joystick port so you can do closeup view of various parts of a scanline.

[Jacques]

Unfortunately not. The last time I used one was during my studies at technical university - mid 2000's

[+bob1200xl]

You have to understand that whatever inputs you provide, the end result is a video signal to each of the three 'guns' in your monitor. The proportion of each gun color (the RGB) is dependent on two things - the instantaneous amplitude of the Y component and the point in time that the color signal © changes state. There shouldn't be any amplitude 'noise' on the color signal or it will get decoded into visible screen garbage. By the same token, no color signal should be present on the Y channel. Some low resolution monitors don't have the ability to display some of these defects - others do to a certain extent. Since the color signal is time dependent within any pixel-time, the visible color change will be offset from the Y component change. So, you get color fringing in some circumstances.

 

My point is that a lot of non-obvious effects are involved here and what works for one may not work for another. I remember the 1702 monitor - it has poor trace timing at the beginning of each scan, which I thought was caused by my video hacking. After trying 3 or 4 1702s, I realized that it is a consistent 'feature' of these Commie displays. Most of you folks don't have 3 or 4 monitor samples lying around so you are at the mercy of chance. A lot of technical folks have taken a shot at this... we still don't seem to have a really good solution.

 

I remember firing up one of my 600XLs on the LCD TV and getting a drop-dead beautiful display. Only that 600XL, only that TV. Sure looked good.

 

A lot of luck is involved.

 

 

Bob

 

 

 

 

Thank you All for trying to help. I'll check the 800XL carefully again this evening, as the mod looks to be done okay, so maybe there is a faulty element or my video circuit or sth... I may also try to use again directly the GTIA's COLOR signal but I'll disconnect this signal completely from the rest of the video circuit. If I'm not wrong, then there should be no artifacting anymore, otherwise the pure CHROMA would have artifacts in itself

 

P.S.

The picture from my other 130XE's S-video on Sysinfo looks perfectly the same (no)artifacts-wise as on your screenshot, so definitely this 800XL has some illness still And my cable doesn't have any connection between LUMA and CHROMA, which would be ideal problem to eliminate

[Jacques]

Yes, results may differ but at least there should be no artifacts when LUMA and CHROMA signals are completely separated (which is the case with 130XE on the same cable & monitor combination)

[flashjazzcat]

Yes, results may differ but at least there should be no artifacts when LUMA and CHROMA signals are completely separated (which is the case with 130XE on the same cable & monitor combination)

This is really what has me confounded too. Looks like you have one method of minimising vertical banding, but these artifacts...

 

I can only echo the above, though: every computer and every monitor is different. These kinds of mods are not magic fixes. The trouble I've had getting VBXE to display on a good monitor is a case in point. The point of my 800XL mod was to output as pure a signal as possible. My TV happens to handle it well, and I would think the cleanest possible signal is a good place to start troubleshooting from.

[+poobah]

Yes, results may differ but at least there should be no artifacts when LUMA and CHROMA signals are completely separated (which is the case with 130XE on the same cable & monitor combination)

 

If you have more than one noise source (quite likely inside an 800XL), your mods may remove one, but worsen another.

[MEtalGuy66]

Ok.. I told someone I would scope this crap, so here it is..

 

This was taken on a 130xe that has damn nice svideo output..

Sample was taken somewhere mid-scanline, with the machine powered up to the "Blue Basic READY prompt."

 

 

PHI-01 is the phase 1 signal generated by the CPU..

 

CADJ is the signal that is generated by that circuit that has the color adjustment potentiometer.. It "derives" this crazy assed "AM looking" signal from phase 1 somehow.. This is the signal that is going into pin 17 of GTIA.. This signal had to be really "BLOWN UP" to see it's characteristics.. it runs right at about 4v, and it's entire range of fluctuation is about half a volt.. The others vary by normal TTL voltage range..

 

COLOR-OUT This is the signal comming out of pin 21 of GTIA.. It then goes through the circuits you guys have been tweaking, and is amplified/attenuated to produce the last signal (chroma)..

 

CHROMA This is the actual chroma signal at the monitor jack..

 

 

It's easy to see the apparent relationships between these signals, but the question I have is about CADJ.. When we turn the color adjustment POT, what are we "adjusting"? The pulse-width? The frequency? The amplitude range? The relative voltage offset?

Edited January 12, 2010 by MEtalGuy66

[MEtalGuy66]

Ok.. Answered my own question.. Its the relative voltage level of the signal.. When you turn the color pot, it just moves the whole waveform offset up & down the screen..

[flashjazzcat]

Interesting studies. Thanks for posting those pics.

 

I think my 800XL picture could do with being brighter overall, and TV adjustment just turns surrounding blacks to greys. What modifications would be necessary to boost luminosity?

[dmlloyd]

On the 1400XL, CADJ is a plain DC input. In fact, according to the GTIA data sheet (transcribed):

 

9.2)  D.C. AND OPERATING CHARACTERISTICS: (CONT.)
_______________________________________________________________________________________
|                                                       |        |      |        |      |
|                                                       |  MIN.  | TYP. | MAX.   | UNIT |
|_______________________________________________________|________|______|________|______|
=                                                       =        =      =        =      =
| DEL INPUT: COLOR DELAY LINE ADJUSTMENT                |        |      |        |      |
|                                                       |        |      |        |      |
|   DEL (PIN 17)                                        |        |      |        |      |
|                                                       |        |      |        |      |
| V(ADJUST) INPUT ADJUSTMENT VOLTAGE:                   |  3.0   |      |  8.0   | VOLTS|
|                                                       |        |      |        |      |
| I(LEAKAGE) INPUT LEAKAGE: VIN=7.0 VOLTS               |        |      | 10.0   | uA   |
|                                                       |        |      |        |      |
| C(PIN) PIN CAPACITANCE                                |        |      | 50.0   | pF   |
=                                                       =        =      =        =      =
|_______________________________________________________|________|______|________|______|

 

And nowhere does it say the input should even be an AC waveform of any kind. I think this was a case of Atari engineering error. The XE design is based on the XL design after all. The folks who designed the 1400XL actually must have read the data sheet and done some research and realized that all that crap isn't necessary (they divide down from +5V... pure DC). The Atari 800's input is a DC input as well (though divide down from +12V instead of +5V).

 

Saturation is only going to be a function of the amplitude of the COLOR signal I believe and has nothing to do with this crazy input modulation on CADJ (CADJ affects the *delay* of the COLOR signal).

[dmlloyd]

In fact, I'm going to go out of my way to prove this. Once I finish my color circuit on the 600XL this weekend (I hope), I'll take some before and after screenshots - before and after removing all the AC circuitry and replacing the whole works with the 1400XL circuit.

[MEtalGuy66]

Heh.. That sucks.. I rebuilt the whole crappy assed "modulation" circuit on my 800XL, just to "fix" the color..

 

You mean, I could have just hooked one side of the pot to +5v, and the other side to a resistor, going to ground..... Or what exactly is the 1400XL circuit?

[flashjazzcat]

Watching with interest here, since I'd like to redesign the 130XE colour circuit.

[+bob1200xl]

The CADJ pin is a DC input that varies the phase delay of the CHROMA signal. As you adjust CADJ, the color of the pixels change. The 400/800 had +12v available from their power supplies which was used to feed the adjusting pot. The XLs and XEs do not have +12v so they used a voltage doubler circuit to feed the their pots. But, it is a DC voltage. You can actually lift the CADJ pin and feed it with a signal to change the color dynamically, if you want.

 

If you want a "brighter" screen, increase the value of the output resistor of the driver circuit.

 

Bob

 

 

 

 

On the 1400XL, CADJ is a plain DC input. In fact, according to the GTIA data sheet (transcribed):

 

9.2)  D.C. AND OPERATING CHARACTERISTICS: (CONT.)
_______________________________________________________________________________________
|                                                       |        |      |        |      |
|                                                       |  MIN.  | TYP. | MAX.   | UNIT |
|_______________________________________________________|________|______|________|______|
=                                                       =        =      =        =      =
| DEL INPUT: COLOR DELAY LINE ADJUSTMENT                |        |      |        |      |
|                                                       |        |      |        |      |
|   DEL (PIN 17)                                        |        |      |        |      |
|                                                       |        |      |        |      |
| V(ADJUST) INPUT ADJUSTMENT VOLTAGE:                   |  3.0   |      |  8.0   | VOLTS|
|                                                       |        |      |        |      |
| I(LEAKAGE) INPUT LEAKAGE: VIN=7.0 VOLTS               |        |      | 10.0   | uA   |
|                                                       |        |      |        |      |
| C(PIN) PIN CAPACITANCE                                |        |      | 50.0   | pF   |
=                                                       =        =      =        =      =
|_______________________________________________________|________|______|________|______|

 

And nowhere does it say the input should even be an AC waveform of any kind. I think this was a case of Atari engineering error. The XE design is based on the XL design after all. The folks who designed the 1400XL actually must have read the data sheet and done some research and realized that all that crap isn't necessary (they divide down from +5V... pure DC). The Atari 800's input is a DC input as well (though divide down from +12V instead of +5V).

 

Saturation is only going to be a function of the amplitude of the COLOR signal I believe and has nothing to do with this crazy input modulation on CADJ (CADJ affects the *delay* of the COLOR signal).

[Bryan]

Yup the CADJ circuit is a crude DC-DC upconverter to allow an adjustable range of over 5V if necessary. It uses an unused 6502 clock line which I'm sure someone thought was pretty clever. This voltage is used to drive the color delay line and the voltage affects the amount of delay.

Edited January 12, 2010 by Bryan

[dmlloyd]

Heh.. That sucks.. I rebuilt the whole crappy assed "modulation" circuit on my 800XL, just to "fix" the color..

 

You mean, I could have just hooked one side of the pot to +5v, and the other side to a resistor, going to ground..... Or what exactly is the 1400XL circuit?

 

Yeah, that's exactly what I mean.

 

Look at VR1 here (center, top):

 

And on the 800, the circuit with R308/R309:

[MEtalGuy66]

Yeah those are both 12v sources.. So yeah youd probably still need the voltage doubler to get a useable range of adjustment on a machine that only has +5v sources available..

[Bryan]

Yeah those are both 12v sources.. So yeah youd probably still need the voltage doubler to get a useable range of adjustment on a machine that only has +5v sources available..

Realistically, you need about 7 or 8V max. I don't think I've had to set one higher than that.

[dmlloyd]

Ah, that's what I missed: I didn't notice the 1400XL one was up to 12VDC. Well, at least it explains the mystery behind that circuit. I'll have to measure the voltage that I'm actually using there - maybe I can get away with +5V after all (the PSU actually outputs somewhat higher than 5V). Now if only it were the weekend...

[potatohead]

How do you get the bars on an 800XL?

 

Yes, I want to go the other way to get a more raw video output! I'm hoping I can remove a resistor, or cap and get the signal a bit more opened up, from what it is on the XL.

Edited January 12, 2010 by potatohead

[Ross PK]

I have these on my 800XL.

[goldy/gmg aka lopez453]

I'm also experimenting with 4 800/130 xe's i have (both 1bit and 4bit rams)

 

Be sure you have quality s-video cable (without any diodes etc, and that has two shielded coaxials.) as it the dramatically affect the output. But no need for golden plated monster cables :-)

 

Funny thing is that one of my tv's I experimented with, does not have s-video so I tried to improve the composite output, get rid of the vertical lines etc...

 

I soldered TV input cable to atari PCB luminance output and got colors !!!??????. Aha, I've forgotten my svideo cable plugged in my atari monitor connector, but other end was unconnected. Cable was 1 to 1 without any "diodes". So trying to figure out more I shortened the s-video cable to half.

The colors on the tv were weaker. So i cut another half....

 

It's amazing that the simple induction between chroma and luma in the poor quality s-video unshielded cable gave better results than many mods I tried.

The only difference betweet composite and s-video was the sharpness of edges. however it did not heal the "monochrome signal with vertical lines" plague that is spread in different stages across my atari farm.

Edited January 13, 2010 by goldy/gmg aka lopez453

[+Stephen]

This is very weird. I thought I was going mad last week until I realized it was the cable linking the luma and chroma. Obviously not so in your case. From what you say, we can rule out the TV. Something's been missed. I just double-checked with the (excellent) SysInfo program, and this is what you should be seeing:

 

 

Or an approximation thereof. If taking the signal straight from GTIA works better for you, perhaps that's the way to go. I'm out of ideas.

 

By the way - sorry if we seem to ask questions repeatedly, but even the best of us miss things at times, and we're just trying to understand your set up and eliminate things one by one.

I am having major issues with my machines now. I never tried hooking them up to a modern TV. But I cannot get a decent picture. How are you getting such sharp moire-free images? Even when I have my camera set to 1/60sec exposure (NTSC TV) I get some moire pattern in the image. The camera is a Nikon Coolpix L100. I have not spent any time exploring all the settings yet. Hopefully I can coax a decent image out of it this weekend.

 

Stephen Anderson

 

 

Stephen Anderson