800XL chroma weirdness

[Urchlay]

Not posted here in ages, been lurking a bit lately.

 

TL;DR version: my 800XL's chroma output is connected (stock, not a mod),

but the colors are very faint and completely different from what the

composite output displays.

 

The full story:

 

About 10 years ago, I got this 800XL... it's AFAIK completely stock. All

I did to it was clean the keyboard (someone had spilled what looked like

orange juice in there & let it dry), and replace the RAM (worked when

I got it, one of the MT DRAMs died while it was in storage, replaced them

all with, eh, I dunno who makes these chips but the part number starts

with KM).

 

Using a Commodore 1702 monitor for it, which looks great. I had to

clean the color adjustment pot, which works now although it's missing

the plastic ring/knob that sits on top.

 

The RF shield is missing, and so are the screws that hold the case

together. Probably those are in a box in the garage somewhere...

 

The power supply I'm using is from a Netgear router, modern switching

supply rated at 5V/2A, and it seems rock-solid. All I did was salvage

the 7-pin DIN connector from an Atari "ingot" and solder it to the new

power supply. The next day I found in the closet an Atari "beauty queen"

power supply, which works, but I'm a bit paranoid about using a ~30

year old power supply.

 

Originally I had color stability issues, which I thought might be due to

the power supply, but I fixed it by spraying out the color pot and the

monitor jack with deoxit (or actually the generic 'electronics cleaner'

spray from Wal-Mart, which I've used for guitars/amps/PA stuff for years

without problems). The color is now stable, but read on...

 

This 800XL has a black sticker on bottom, "Made in Hong Kong", and

has Revision C BASIC (mask ROM, not EPROM). It also has the chroma pin

connected (looks like it came that way from the factory, not a mod).

 

Composite video works great, and looks as good as composite ever does

on an 800XL + C1702.

 

"Fake" S-video, with the Atari's composite output plugged into

the monitor's Chroma input, works great and looks nicer than

composite. There's still some artifacting in this mode, but I'm OK

with that.

 

Real S-video, with the Atari's chroma/luma connected to the 1702's

chroma/luma, works... sort-of. The picture is beautiful, crystal clear,

but the color is very weak (looks black & white unless I crank the

monitor's saturation all the way up) and has wrong colors (green becomes

orange, for one thing).

 

I don't think the monitor's at fault here, though I do plan to dig up

another monitor with chroma/luma inputs and see if it does the same thing.

 

My questions:

 

- Is this a known issue with late-model 800XLs, possibly with a known fix?

 

- Shouldn't the color on composite and chroma outputs be the same,

at least as far as the hue?

 

To save time/effort: Please, nobody respond with "You should get a UAV

or VBXE". This 800XL, I'm keeping as stock as possible. I just want to

use it with the 1702's chroma/luma, if it can be made to work. Don't

want to do ClearPic or SuperVideo mods either (not needed, I have a

1200XL with SV if I want that).

[+DrVenkman]

That's a very weird set of symptoms. Have you compared the Chroma output on another display? Or to put it another way, have you done an A/B comparison on the same 1702 Chroma/Luma inputs with your SuperVideo 1200XL?

 

Assuming your monitor and its inputs are fine, have you examined that Chroma line on your 800XL closely? My thought is it might have a solder connection going bad. Alternately, if it picks up the Chroma signal after the 4050 hex buffer, it's possible the 4050 itself (or its socket) might be going bad.

[Urchlay]

That's a very weird set of symptoms. Have you compared the Chroma output on another display? Or to put it another way, have you done an A/B comparison on the same 1702 Chroma/Luma inputs with your SuperVideo 1200XL?

 

I hadn't yet when I posted that... but I just found the 1200XL power

supply and connected it. On the same 1702, with the same video cable,

using chroma/luma inputs, the colors are correct.

 

So the monitor's fine (as I thought).

 

Note: I can't test composite on the 1200XL, I did the 'composite lift' when

I did the supervideo mod, was going to install a switch but never got

around to it...

 

Aside: it's too bad there's not room for the 1200XL and the 1702 on

my desk. By modern standards, both are big beautiful beasts. And the

supervideo's output is gorgeous.

 

Assuming your monitor and its inputs are fine, have you examined that Chroma line on your 800XL closely? My thought is it might have a solder connection going bad. Alternately, if it picks up the Chroma signal after the 4050 hex buffer, it's possible the 4050 itself (or its socket) might be going bad.

I haven't yet traced the chroma pin to see where it goes on the mobo.

 

Would a bad 4050 not also cause the composite output to be bad? Or it's

a "hex buffer", could just one of the 6 buffers be dying? I might have

a spare 4050 in my old box o' parts, will give that a shot. If nothing

else I've got another non-booting 800XL I could steal it from (but who

knows if any of the parts from that 800XL are any good?)

[Rybags]

It's been some time but I'm fairly sure the mod I did to mine included a resistor - there's some descriptions out there that are a wire join only.

[+hunmanik]

Willing to share the serial number of that interesting 800XL? (and/or at http://atariage.com/forums/topic/132201-800-xl-serial-numbers-secampalntsc/ )

 

I thought all 800XL units with chroma were later production units under Tramiel, and from Taiwan. I don't know where one from Hong Kong would fit in.

 

It also sounds very much like this old description: https://comp.sys.atari.8bit.narkive.com/ijf7BxNN/my-stock-800xl-has-chroma-output

[Urchlay]

So I swapped the 4050 from the dead 800XL, but it didn't change anything.

 

The socket looks OK (though my vision isn't what it used to be). I

hesitate to spray contact cleaner into the socket, because the normal way

to clean e.g. an audio jack would be to spray, then plug/unplug a cable

several times to spread it around & make it do its thing... I don't really

want to repeatedly plug/unplug the IC, the sockets Atari used only have

a few insert/remove cycles before they die. Plus I'd likely bend a pin...

 

Time to break out the ohmmeter and start tracing I guess. Also going to

find a magnifying glass (had one, it's gone AWOL).

 

The old c.s.a8 post linked to says the chroma mod is done by connecting

the chroma pin on the jack to the resistor just above Q5. Wonder if that's

where the chroma pin's already going in this 800XL (somehow I doubt it).

[Urchlay]

Willing to share the serial number of that interesting 800XL? (and/or at http://atariage.com/forums/topic/132201-800-xl-serial-numbers-secampalntsc/ )

I thought all 800XL units with chroma were later production units under Tramiel, and from Taiwan. I don't know where one from Hong Kong would fit in.

 

ATARI 800XL

FCC ID: BPA7YJ800

ATARI, INC.

MADE IN HONG KONG

 

Has an Atari Fuji logo plus a Warner logo and tiny text reading "A Warner

Communications Company"

 

Serial number is 7YJ HA 105853 164

 

(16th week of 1984, perhaps?)

 

I'll take a proper photo and post it to the serial numbers thread.

 

It also sounds very much like this old description: https://comp.sys.atari.8bit.narkive.com/ijf7BxNN/my-stock-800xl-has-chroma-output

Hmm, food for thought, thanks for that. I see Rybags posted the details

of his chroma mod way back then

[Urchlay]

So I swapped the 4050 from the dead 800XL, but it didn't change anything.

 

I spake too soon. The replacement 4050 caused color instability (the

hue and/or brightness would drift). So I put the original one back in

and it's back to the colors being stable (even if wrong).

[_The Doctor__]

without photos front and back honed in on the video sections, the article sited in the news archive sound familiar and mostly correct. Though the connection being there both early and late in production sounds about right, I'd say the omitted the connection occurred after realizing it was wrong and knowing most monitors could just use the composite signal and pull chrominance from there, a compromise that works well enough. The only way to be sure is to see it and make sure it's pulled from the correct point in the circuit and the resistor value if any that might be on the line.

[+hunmanik]

Yes, the understanding is that the serial number ending with the 164 stamp should mean it's from the 16th week of 1984.

 

And the "black label" NTSC units with serial numbers starting with 7YJ were manufactured for Atari by Chelco in Hong Kong.

 

I decided to pull out my 3 800XLs and connect them up for video via S-video, and now I'm just further confused. All three of my units are NTSC, one by Chelco, two by Atari Taiwan, and all three have a "weak" nearly black & white signal on the chroma output line. I might describe it the same way you describe yours. Atari's own schematics show Pin 5 on the monitor port to be Not Connected, and so does Sams, but it sure looks like it's connected to something. Do people actually have units with no video on the Chroma pin, corresponding to the schematics? Or do all the units without proper chroma connected actually have this same "weak" chroma on Pin 5? Sorry if this ends up sabotaging this thread, but I sure am curious.

[ChildOfCv]

Are you sure the chroma pin is connected? What you're describing sounds exactly like what you'd expect from no chroma pin: Plug the luma into the luma port, plug the chroma into the chroma port, then crank up the color saturation: Colors become random due to maximum amplification of floating voltage from a disconnected chroma.

 

Comparing the 800XL to the 130XE, there are many differences in how the pins are connected, not just that the chroma is missing.

 

On the 130XE, color splits out from the GTIA immediately on two separate paths. Path 1 is straight to the RF. Path 2 is to a color amplifier and then to the chroma pin (5). Luma goes through the luma amplifier and then to both the RF and the luma output (1). Note that there is no mixing at all on either of those pins. The mixing happens in the RF module, which has an output feed to the composite pin (4).

 

On the 800XL, the color amp has 3 stages for some reason. Then the color and luma output are mixed in 3 paths. Path 1 is a direct mix of color and luma and goes into the RF modulator. Path 2 is the luma, with chroma bleeding in through R56 and going to the composite luma output (1). However, it appears that the luma amplifier Q3 has enough authority over this path to mostly keep color from adding to the output. Path 3 combines luma and chroma into a composite output on the composite video (4) pin. So the 800XL has no direct chroma output. You get either RF, B/W composite, or full composite.

 

So the "fake s-video", the full composite into the chroma input with the B/W composite as the luma, is probably as good as you get without modification. If the monitor/TV is well filtered, it might even be as good as it gets. But if you want a pure luma signal, you'd have to connect pin 5 to... maybe the opposite side of C54 from pin (4). You'd probably have to build another amplifier whose input comes directly off of the last color amplifier stage and whose output has the expected Vp-p and impedance for the chroma.

 

Hmmm... maybe removing R56 would make pin 4 a pure chroma output, for that matter.

 

Okay, final note: Connect a 75 ohm resistor in series with a 1nF capacitor and connect it to pin 5, and connect the other side of this hack to the nexus of Q5, R66, R67, and R68. This should give you pure luma on pin 5 without disrupting the other pins.

Edited April 18, 2019 by ChildOfCv

[Urchlay]

I decided to pull out my 3 800XLs and connect them up for video via S-video, and now I'm just further confused. All three of my units are NTSC, one by Chelco, two by Atari Taiwan, and all three have a "weak" nearly black & white signal on the chroma output line. I might describe it the same way you describe yours. Atari's own schematics show Pin 5 on the monitor port to be Not Connected, and so does Sams, but it sure looks like it's connected to something. Do people actually have units with no video on the Chroma pin, corresponding to the schematics? Or do all the units without proper chroma connected actually have this same "weak" chroma on Pin 5? Sorry if this ends up sabotaging this thread, but I sure am curious.

It's not sabotage, it's a very good question. I've had "800XL lacks

chroma" in my head for so many years... Now that I think of it, I can't

remember if I ever tested it, or just took for granted that it wouldn't

work so never tried it.

 

My other (so far, non-booting) 800XL has a chroma mod that was probably

done by me in the 80s, so I can't tell you what was originally on its

chroma pin.

[Urchlay]

Are you sure the chroma pin is connected? What you're describing sounds exactly like what you'd expect from no chroma pin: Plug the luma into the luma port, plug the chroma into the chroma port, then crank up the color saturation: Colors become random due to maximum amplification of floating voltage from a disconnected chroma.

 

They're not random though, they're wrong but self-consistent. Unplugging

the chroma cable from either end (the monitor or the Atari side) results

in a black & white picture, so I still think the chroma is connected to

*something* inside the Atari.

 

So the "fake s-video", the full composite into the chroma input with the B/W composite as the luma, is probably as good as you get without modification. If the monitor/TV is well filtered, it might even be as good as it gets. But if you want a pure luma signal, you'd have to connect pin 5 to... maybe the opposite side of C54 from pin (4). You'd probably have to build another amplifier whose input comes directly off of the last color amplifier stage and whose output has the expected Vp-p and impedance for the chroma.

 

The "fake s-video" is really nice on the C=1702, there's some red/green

artifacting, but not much compared to using composite. On my LCD TV's

s-video input though, fake-composite results in a pure black & white

picture, as does the 'weak' chroma signal. I kinda do want the LCD

to work, it's inferior for graphics & games, but less likely to give

me a headache when doing 80-column text. Maybe.

 

Hmmm... maybe removing R56 would make pin 4 a pure chroma output, for that matter.

 

Pin 4's composite, I don't want to stop that working... might need it

someday. Might try it for testing purposes though.

 

Okay, final note: Connect a 75 ohm resistor in series with a 1nF capacitor and connect it to pin 5, and connect the other side of this hack to the nexus of Q5, R66, R67, and R68. This should give you pure luma on pin 5 without disrupting the other pins.

Pure luma? I have that already on pin 1... or is this a typo/thinko,

you meant 'pure chroma'?

 

I was thinking to disconnect pin 5 from the motherboard entirely, then

doing the old '1-wire chroma mod' (connect pin 5 to the resistor above

Q5). Actually I think that's basically the same thing you're describing,

minus the 1nF cap (at least, my other 800XL has an extra resistor but

no cap).

 

Unfortunately pin 5 can't be gotten at without desoldering the whole damn

connector (pins 1-3 are reachable from inside the case at the back of

the connector). Plus, before I do anything else, I want to trace this &

see exactly what the heck the chroma pin is already connected to.

[ChildOfCv]

They're not random though, they're wrong but self-consistent. Unplugging

the chroma cable from either end (the monitor or the Atari side) results

in a black & white picture, so I still think the chroma is connected to

*something* inside the Atari.

It may be seeing some RF coupling with the internals of the Atari though, and interpreting that as color information. Since the internals of the Atari run at the color burst frequency, many stray signals could give you a somewhat uniform yet wrong color, especially with the color gain at maximum. Still, it wouldn't hurt to look at the motherboard around the J2 plug to look for traces from pin 5.

 

Pure luma? I have that already on pin 1... or is this a typo/thinko,

you meant 'pure chroma'?

Yep, thinko

 

I was thinking to disconnect pin 5 from the motherboard entirely, then

doing the old '1-wire chroma mod' (connect pin 5 to the resistor above

Q5). Actually I think that's basically the same thing you're describing,

minus the 1nF cap (at least, my other 800XL has an extra resistor but

no cap).

Well, close. That takes the output directly from the final color amp, which will do okay. The composite output uses the 75 ohm resistor from each amplifier to match impedance on the RCA cable, to reduce signal reflection. The capacitor decouples the DC of the chroma amplifier output from the signal out. I used the chroma part of this composite output as a model for the "pure chroma" idea. Edited April 19, 2019 by ChildOfCv

[+DrVenkman]

Unfortunately pin 5 can't be gotten at without desoldering the whole damn

connector (pins 1-3 are reachable from inside the case at the back of

the connector). Plus, before I do anything else, I want to trace this &

see exactly what the heck the chroma pin is already connected to.

The Atari is only a 2-layer board. You can access pin 5 from the bottom of the board. If there’s a trace or a jumper there, it will be evident from a simple visual inspection.

 

And for what it’s worth, I also have a Chelco-made brown label 800XL. It has a rev A PCB, it’s fully socketed (even the 74 series logic) and chroma was most definitely not connected until I ran a wire myself. I also have two layer-rev Taiwan made boards, one fully socketed and the other mostly socketed. Neither of them have factory chroma connections.

[Urchlay]

The Atari is only a 2-layer board. You can access pin 5 from the bottom of the board. If there’s a trace or a jumper there, it will be evident from a simple visual inspection.

 

You're right, I was making it too complicated... and no, there was

nothing connected to the chroma pin. So ChildOfCv is correct, it's the

chroma wire acting as an antenna, picking up RF inside the Atari. "WiFi"

chroma, who knew Atari was so advanced in the 80s?

 

Maybe "ghost chroma" is the right name for this phenomenon. I bet it's pretty

common on 800XLs, just not well known. In fact, from the old newsgroup thread

linked to from this thread, I see this:

 

much to my surprise it's showing a picture when hooked up via

chroma/luma. It's not as bright as the other machine (even a slightly

different hus), but it's certainly just as clear.

 

Sounds like what I was seeing. So there's nothing new here...

 

And for what it’s worth, I also have a Chelco-made brown label 800XL. It has a rev A PCB, it’s fully socketed (even the 74 series logic) and chroma was most definitely not connected until I ran a wire myself. I also have two layer-rev Taiwan made boards, one fully socketed and the other mostly socketed. Neither of them have factory chroma connections.

I went ahead & did the chroma mod, connect 100ohm resistor between the

junction of R67/R68 and pin 5 on the connector. My desire to keep things

stock lost out to my desire to have S-video output...