vertical lines on many XE computers

[Faicuai]

A possible theory on what's happening with the banding:

 

The color carrier is supposed to be a shifting sine wave at the proper frequency for NTSC or PAL, but GTIA puts out a square wave which is run through some combination of filtering components to round it off a bit. Perhaps there's still too many out of band harmonics which leak through the monitor's chroma filters and appear on the screen. Otherwise, unless you're dealing with really strong saturated color (which the A8 isn't able to produce since the carrier is always at the burst reference level) the color carrier should be just about impossible to see.

 

 

...Well, last time I put the composite signal on my Fluke Scopemeter 99B-Series II, I saw voltage levels (per scan-line) that I found quite impressive (much higher than, for instance, AVIA's reference palyed from a decent DVD player, on similar grayscale levels, chroma patters/colors, etc., on composite RCA port as well).

 

What I DID NOTICE, however, is that when plugging BOTH the RCA and S-Video jacks of my cable back into my CRT set (Sony Wega), the banding DIMINISHES sensibly. BTW, the banding seems an analog-level by-product in any case, as it appears on my older CRT or my current Bravia 52" LCD.

 

To me, this problem seems related to actual Chroma, Luma (or both) signal levels. I believe that the LUMA level is too high, and it eventually (somehow) interferes with the Chroma signal (which in my 800XL is NOT fully isolated), thus inducing a Luma-driven pattern that appears especially in solid-colored areas... My 800XL outputs SUPER-RICH, high levels of chroma saturation.

 

Notice that my "JM"-800 DOES SHOW similar vertical banding (although at lesser levels) on s-video output (same custom-made cable I use on my 800XL). The difference is that my JM-800 has a STRATOSPHERIC level of LUMA, and, conversely, much LOWER chroma-signal, when compared to my 800XL.

 

F.

[Bryan]

My XEs can get the same thing depending on what video source I choose.

 

The banding tends to be coincidental with the pixel clock / 1.77 MHz clock (PAL), so maybe too much blame is being devoted to the colour circuitry. Also, like many Ataris, you can get a tendency for the interference to be stronger when a Refresh cycle starts.

Someone posted earlier that the banding went away when chroma was disconnected. Can you verify that? If so, it's got to be caused by whatever's going out on the chroma line.

 

When modifying the video circuits, it's important to pay attention to voltage levels and adhere to standards as much as possible. In other words, just changing values until the clarity improves is not enough. The signal may need to be level-shifted afterwards. This weekend maybe I'll drag the scope back out.

 

When setting final luminance levels, I think matching the original 800 is probably the best bet for achieving what was originally intended. When setting chroma, just get it close to the textbook colorburst level.

[flashjazzcat]

Someone posted earlier that the banding went away when chroma was disconnected. Can you verify that? If so, it's got to be caused by whatever's going out on the chroma line.

I'll verify that.

[+Philsan]

Jon, you have done a lot of experiments.

Does Kjmann's Internal S-Video Upgrade kit solve this problem?

[flashjazzcat]

Jon, you have done a lot of experiments.

Does Kjmann's Internal S-Video Upgrade kit solve this problem?

Indeed. Let me get out of coding-mode for a moment...

 

I never found the 800XL to suffer badly from vertical banding - not to the extent of the XE line. As it happens, though, XLs are where I've done the bulk of testing with Sal's upgrade. Of course results with the already decent 800XLs was absolutely stunning.

 

Clearly, then - I owe it to everyone - including Sal - to find the time to do more testing on the XEs.

 

One thing I can say: I tested Sal's very latest board with a 1200XL. While the shapness of the display was improved, the (awful) vertical banding was not. So we may or may not choose to draw conclusions about how the unit will handle the XE's vertical banding. I imagine we should wait to see how it performs.

 

I'm no electronics expert (far, far from it), but the general gist of the knowledge I've picked up over the years suggests that overpowering chroma causes a host of problems. Not just vertical banding, but artifacting on machines where the composite video has been completely disabled, and chroma/luma isolated as much as possible. This was the case with one of my "UltraVideo" 800XLs which - tuned to perfection for an LCD screen - suffered terrible artifacting on a 1084S until I moderated the chroma. And of course one can't be certain what the internal circuitry of a given display device is doing to the two signals.

Edited September 6, 2011 by flashjazzcat

[sloopy]

eeeewwww vertical banding :'(

(as seen on a scope)

 

another 130XE with rock solid video

 

sloopy.

[+bob1200xl]

The only amplitude signal that should be in the video is the LUMA. Problem is: noise is induced in the ground circuits by refresh, chroma, IDE drives, and just about anything that draws instantaneous current through the power wiring. The 130XE chroma circuits pull a lot of current - try reducing it. Refresh, you can't do much about. You can try filters on IDE drives. You can add non-linear components (diodes) in an attempt to clamp out the ground noise, but you have to be sure that you are not also deleting part of your luma bandwidth.

 

Someone said: start at the digital output of GTIA and build video externally... I hear that.

 

Bob

 

 

My XEs can get the same thing depending on what video source I choose.

 

The banding tends to be coincidental with the pixel clock / 1.77 MHz clock (PAL), so maybe too much blame is being devoted to the colour circuitry. Also, like many Ataris, you can get a tendency for the interference to be stronger when a Refresh cycle starts.

[Bryan]

eeeewwww vertical banding :'(

(as seen on a scope)

 

another 130XE with rock solid video

 

sloopy.

Interesting... I assume we're looking at luminance out? I wonder why the 2nd capture shows DC issues (I noticed they were both captured in AC mode). The question is, what do the spikes coincide with and how are they making the jump onto the signal?

[flashjazzcat]

Someone said: start at the digital output of GTIA and build video externally... I hear that.

This is why - in an ideal world - I'd fit a VBXE to every machine I own. If only the A8 had been RGB in the first place.

[w1k]

eeeewwww vertical banding :'(

(as seen on a scope)

 

another 130XE with rock solid video

 

sloopy.

 

sorry for language, but this is fucking interesting!

[+bob1200xl]

Well, way back then, most TVs didn't have composite inputs, much less s-video, certainly not RGB. I can't hook up to 15KHZ RGB even now. Component is doable...

 

Bob

 

 

 

 

Someone said: start at the digital output of GTIA and build video externally... I hear that.

This is why - in an ideal world - I'd fit a VBXE to every machine I own. If only the A8 had been RGB in the first place.

[flashjazzcat]

Well, way back then, most TVs didn't have composite inputs, much less s-video, certainly not RGB. I can't hook up to 15KHZ RGB even now. Component is doable...

Thank God for SCART. I recall one of the Spectrums having some form of RGB output. But I hear you: when I had an A600, I was using it through a TV via the RF, and that was in the 90s.

[Bryan]

Well, crap...

 

I went to hook up my PAL 800XL and it's gone missing. I must have put it somewhere "safe". Either that, or I've never unpacked it since moving last year, but I'm pretty sure I had it a few months ago...I think....

 

It'll turn up.

[Faicuai]

Since this has certainly remained an elusive problem, here are some specific captures with the Fluke Scopemeter, from two patterns-sets ran directly from the 800XL, and sampled straight at the computer's video jack (Luma Pin).

 

The scope was setup with 10:1 probe, Auto-Set, then Trigger=>TV Mode, ScanLine manual select (my apologies in advance for the gutless, piece-of-shite iPhone pictures, but speed was of essence):

 

1A) "Scan" Line #46 from 256-Color Pattern (still at the top, gray-scale ramp):

 

1B) "Scan" Line #48 from 256-Color Pattern (first color gradient):

 

2A) "Scan" Line #48 from Full-Screen Pattern (Hue:0, Luma 10, still gray):

 

2B) "Scan" Line #48 from Full-Screen Pattern (Hue:1, Luma 10, yellowish, vertical banding visible at this particular combination):

 

3A) "Scan" Line #48 from Full-Screen Pattern, COMPOSITE pin (Hue:0, Luma 10, still gray):

 

3B) "Scan" Line #48 from Full-Screen Pattern, COMPOSITE pin (Hue:1, Luma 10, yellowish, vertical banding visible at this particular combination):

 

 

 

Up to this point, the ONLY modifications I have ever made to the 800XL video circuitry are a) removal of C56, b) Power re-route from C3 to L5, c) enable Chroma Line with NO pick-up resistor (nothing else really seemed relevant or productive).

 

After these tests, it seems that Switch @ R56 is most likely justifiied and I will be testing it. There seems some residual chroma leftovers in what is supposed to be a clean luma signal (Luma Pin samples). Not sure if harmonics, noise, but they ARE definitely present.

 

I will also perform identical tests with AVIA color-fields + svideo port from DVD player, just to make sure that such "residual" signal should (or not) be present on s-video's Luma signal.

 

F.

[Rybags]

If you feel like being adventurous, try it with the COLOR pin bent up on GTIA. That could then let us guage how much interference is due to colour and how much is external from /Halt /Refresh etc.

[sloopy]

on this 130XE i am working on...

screen:

 

and this is the video signal right where it comes out of the 4050 on channel 1, and just before it goes into the video amp Q3...

 

and this is Chroma/Luma, with the RF modulator removed...

 

sloopy.

[goldy/gmg aka lopez453]

I am pretty sure some of the banding is also caused by the ram upgrades.

I have 320XE with socketed extended ram, I was removing 8 exteded ram chips one by one and after all were out, banding dissapeared.

 

However this banding has special pattern and was visible on for example the 1st pixel of every character.

 

As the rams are power hungry (and other upgrades too) I highly recommend checking the voltage quality of 5V and GND at the pins of video circuits, 4050 converter and GTIA, and also the funny COLOR adjust circuit/voltage that is fed into GTIA. Maybe adding some capacitors (both blocking and filtering).

 

I noticed in the headphones when using COVOX only that it had lot of white-noise (with idle computer), when I swapped the switched power supply with regular one, the white noise changed to a "computer noise" from the data/address bus and was also less intense, removing adding extended ram chips also affected this.

 

I don't insist that high power consuption of some components and thin power lines on the board are the only cause, but for sure it is one of them.

Edited September 7, 2011 by goldy/gmg aka lopez453

[Faicuai]

(...)

 

 

I will also perform identical tests with AVIA color-fields + svideo port from DVD player, just to make sure that such "residual" signal should (or not) be present on s-video's Luma signal.

 

F.

 

 

CONFIRMED:

 

A video-signal (sVideo) sampled directly from my Sony DVD video port (Luma Pin + Luma Ground), and using AVIA's Calibration DVD (Advanced Patterns -> Color Fields (full screen) -> Yellow), reports an ALMOST FLAT/clean signal level on the Fluke Scopemeter. In other words, basically NO trace of any disturbance regardless of what's happening with Chroma (the video sample is pretty similar to my prior samples from the Atari, solid/full screen, with yellowish color, although slightly lower in brightness).

 

Therefore, the above finding strongly suggesst that the Luma signal from the Atari IS CONTAMINATED (it is NOT yet clear if such contamination comes from chroma, or from ITSELF, or from somwhere else).

 

Next step is to try to locate at which point or under what conditions the signal is being slightly screwed. More work on the coming days.

 

F.

[Spanner]

I have like colour bands on my 800XL,how do you get rid of them,the more colour you add the more they stands out,I understand now why most people had them on a black and white TV,like my,I have my C64 on a b+w tv and the A800XL at school was on one too.

 

Lode Runner look really bad in colour(the B&w version,you can really see the bands.

Edited September 11, 2016 by Spanner

[+Larry]

Presume you are using composite? Or is this RF to a TV?

 

-Larry

[Atari Nut]

Too bad these vertical lines wern't horizontal. Then I would welcome then as emulated scanlines.

[Atari Nut]

I'm getting these vertical lines with an S-Video cable. I'm not getting any banding. Can someone clarify if I would still get these vertical lines if I switched to a composite video cable?

[_The Doctor__]

that's pretty ugly, I have never seen it that nasty unless there were issues with the video cable.... incorrect impedance... bad connection etc...

[Almerian]

Is this on a LCD screen? Then maybe you could try your 800XL with a CRT. LCD chipsets can really mess up legacy video signals.

Although I've never seen it like this.

[Starblazers1969]

Use a product called Retrotink 2X and it will get rid of the lines if you are using a LCD.