Video Overscan Showing on LCD Monitor?

[+mytek]

OK so let me start out by saying that my system is an XEGS with a UAV board installed. Initially I was sending the S-Video output through an up-converter and viewing on a HDMI LCD wide-screen monitor. But when I started seeing some aberrations I switched over to a DELL monitor with S-Video input to rule out the up-converter as being the cause. What I began to notice was that games without a border were doing some strange things towards the left and right sides of the screen (reference screen shot below). So if you look at the point I have marked as [1] and on the other side [2] you'll see that the graphic of the canoes ends before actually getting over to the actual edge of the screen. And if you look at the place I marked as [3] you'll see a vertical offset occurring. Now keep in mind that I have the DELL set to display in 4:3 mode and no zoom, so you would think that all would be good.

 

So my question is, am I seeing these things because even in 4:3 mode the entire screen output is visible, whereas on an older CRT the picture extends off the screen? And yes I verified on a CRT monitor that this 'extra' area off to the left and right side does not exist.

 

Screen Shot of Dell 20" Square 4:3 LCD Computer Monitor

With XEGS/UAV S-Video Direct Connection - No External Up-Conversion

Game: Preppie!

 

 

BTW, I also see the same result when hooked up through the Dell's composite video input, although overall picture quality is not as good.

 

- Michael

Edited August 31, 2016 by mytekcontrols

[flashjazzcat]

Yes, these look like normally invisible off-screen areas. The canoes are truncated because the standard width playfield does not extend to the screen edges, and the shearing on the colour changes are caused by the background (border) colour being changed just before the horizontal blank. I wondered why I got reports of this kind of shearing on the Ultimate SD Cart menu when the effect is not visible on any of my displays, but it appears some upscalers and NTSC displays display much more of the overscan region.

Edited August 31, 2016 by flashjazzcat

[+mytek]

Yes, these look like normally invisible off-screen areas. The canoes are truncated because the standard width playfield does not extend to the screen edges, and the shearing on the colour changes are caused by the background (border) colour being changed just before the horizontal blank. I wondered why I got reports of this kind of shearing on the Ultimate SD Cart menu when the effect is not visible on any of my displays, but it appears some upscalers and NTSC displays display much more of the overscan region.

 

Thanks for the confirmation

 

Unfortunately if I use the zoom feature on the monitor, it zooms both the vertical and horizontal axis's so you loose part of the vertical display which is not desirable. Be nice if they would let you adjust both axis's independently like on the older CRT displays.

 

- Michael

[Rybags]

My old PCI capture card could display every pixel the Atari could generate except the first colour-clock on the left. Most modern LCDs will chop off somewhat more than that though still show a good deal more than the old 4:3 CRTs.

 

The jagged colour change on the right is usually invisible though. If following the classic programming advice of loading new colour into a register then hit WSYNC and immediately store the new colour value, that's the usual result. We never complained about it with CRTs because we mostly never saw it.

[+mytek]

To put a positive spin on this, at least now I know that my HDMI Up-Converter isn't at fault. If anything, it's just doing too good of a job showing what's actually there to be seen. Even revealing what would normally be wrapped partially around the CRT in an older set. And the Dell monitor also isn't hiding anything as well. Hmm... makes me wonder if I could somehow implement a circuit to gate the video coming out of the S-Video, so that it would stay at black level for a bit at the beginning and end of each scan line. This way the over scan would disappear. Basically creating kind of a shadow box effect on the vertical information. Because I'm seeing this problem on several other games as well.

 

- Michael

[+mytek]

So based on my last line of thought, I've been tinkering around with a possible circuit to eliminate luminance generation during over scan. To make this as easy as possible this circuit would be internal to the A8, intercepting the digital luminance, and then gating it through a quad AND gate. Luminance gate timing would be handled by two 555 timers (or a single 556 dual timer). The first timer 'A' would trigger the 2nd timer 'B' following the first over scan period on the left. Then the 'B' timer would enable the luminance just up to the 2nd over scan period on the right. Csync re-triggers these events on each scan line.

 

Color information is allowed to pass, which I'm hoping won't cause a problem. But if it does,I can gate that as well.

 

The 'A' and 'B' periods will be adjustable to a small degree in order to tweak things in. And off the top of my head I'm not sure if I can have the triggers directly connected to TTL logic, or if they will need an RC single-shot circuit (minor details to work out).

 

Should be interesting to see if this will even work

 

 

 

- Michael

Edited September 1, 2016 by mytekcontrols

[+mytek]

I made a preliminary test to see what would happen with only the chroma line connected (no luminance), and as I feared video was still generated using the color signal alone, although it was obviously missing some information. So that means I'll have to gate the chroma signal as well, but not in the same way as the luma, since I still need to allow the color burst to pass through. So its looking like a 3rd timer with another AND gate, only this time with an OR gate on one of the inputs tied into the 3rd timer output and the Luma Enable line. This way following sync and through the color burst period -OR- during Luma Enable, the chroma signal will get passed through.

 

Some testing to do...

 

- Michael

Edited September 1, 2016 by mytekcontrols

[+mytek]

OK I'm getting pretty close to a working circuit. Had to switch from 555 timers to something a bit faster and made up from fewer IC's. So now I have the color signal also gated, and have allowed for the color burst to make it through while still eliminating the left side over scan.

 

 

 

And here is what it looks like in action. Both the left and right originally visible 'over scan' areas have now been eliminated and replaced with black.

 

 

The motion blur was something my camera did, and is not what it looked like in reality. There's a little bit of noise (Jitter) on the right edge. I think this might either be the result of the circuit being built on a solder-less bread board or maybe attributable to the OR/AND circuit made from diodes and resistors. Although I got to say not bad results from only using two IC packages.

 

- Michael

 

EDIT: I'm really trying to minimize components, hence the diode/resistor logic for the OR/AND gate circuit. Thinking about making a GTIA piggy-back board with this circuit or something similar residing within the GTIA footprint.

Edited September 2, 2016 by mytekcontrols

[Beetle]

Nice idea, Michael! Keep us posted, i consider adding this to my XL laptop.

[+mytek]

Nice idea, Michael! Keep us posted, i consider adding this to my XL laptop.

Hi Beetle,

 

I'll be doing some more fiddling later today (when the sun comes up and hopefully I've slept). Still have 2 left over inverters, so possibly I can utilize those to clean up that right edge. My intention is to get this into a nice reliable form and then lay-out a PCB for it, so you might want to wait on implementing it in your XL laptop until that has been done. As per my normal agenda the PCB Gerber files will be made available for free for the DIY people.

 

- Michael

[Rybags]

With such a device if built into the computer it'd be nice to be able to selectively enable it.

[+mytek]

With such a device if built into the computer it'd be nice to be able to selectively enable it.

Not out of the realm of possibility, but if this works the way I want it to, there really shouldn't be a need to disable it since it won't visibly change a thing on an older monitor. And who would want to see the 'not intended to be seen over scan' on a newer monitor?

 

But with said, I think it won't be too hard to make it selectable (grounding pin 9 on the 74HCT14 would disable it).

 

- Michael

Edited September 2, 2016 by mytekcontrols

[flashjazzcat]

You probably don't want to block out all of the border area to the left and right of the playfield. It's normal for a some border (say slightly narrower than the width of a mode 0 character cell) to be visible. Try changing the colour of the border on the default BASIC screen: the way things are at the moment, it looks like you'll only get border colour at the top and bottom and two black bands at the sides. GTIA pixel shift might also cause issues if you crop tight to the playfield. The truncated canoes in the example game aren't really an overscan issue: that's just the limit of the playfield, and you'd see the same thing on a CRT unless you made adjustments on the monitor so that the playfield is abnormally wide.

Edited September 2, 2016 by flashjazzcat

[+mytek]

You probably don't want to block out all of the border area to the left and right of the playfield. It's normal for a some border (say slightly narrower than the width of a mode 0 character cell) to be visible. Try changing the colour of the border on the default BASIC screen: the way things are at the moment, it looks like you'll only get border colour at the top and bottom and two black bands at the sides. GTIA pixel shift might also cause issues if you crop tight to the playfield. The truncated canoes in the example game aren't really an overscan issue: that's just the limit of the playfield, and you'd see the same thing on a CRT unless you made adjustments on the monitor so that the playfield is abnormally wide.

When I go the shop a bit later I'll check out the border issue you mentioned. I guess the one and only CRT monitor I own must be set wide, because the canoes go off screen on it.

 

No problem on adjusting the cropping, that's what the two trim pots are for.

 

What's GTIA pixel shift? Can you elaborate.

 

- Michael

[flashjazzcat]

What's GTIA pixel shift? Can you elaborate.

 

 

IIRC, it's mode 10 which is shifted half a 4-bit pixel (1 colour clock), resulting in sawtooth playfield edges in some software graphics modes.

[Rybags]

The old Atari800Win+ shows 336 hires pixels. I'm not sure on the exact figure but slightly more than that should avoid unwanted scroll artifacts on the left and the unwanted noise on the right and the colour change jaggie on the right.

 

Altirra emulator showing full display comes in handy, you could just screenshot scenarios you want to investigate with 99% assuredness it'd match what a TV would do, then work out what level of trimming you want.

[+mytek]

Made some tweaks to the design...

 

 

 

And it appears that the noise problems are most definitely coming from the solder-less breadboard being used in the circuit development. If you look at the rats nest that it is (see image below), its a wonder that it isn't much worse. Anyway I was able to tame it a bit by adding a multitude of bypass caps of various values scattered across the power busses.

 

 

 

So using Rybags suggestion, I loaded up the Preppie! game in Atari800 (I use a Linux based computer), and captured the following screen shot.

 

 

 

As can be seen the canoes and the men with lawnmowers do not make it to the edge of the screen before they disappear as FJC mentioned earlier. So now lets take a look at my monitor using the over scan eliminator circuit.

 

 

The borders now look nearly the same, and the noise is no longer present, giving us clean edges on both sides of the screen.

 

 

And to see the adjustments in action here is a short video clip (adjusting the left border first, and then the right border afterwards)...

 

https://www.youtube.com/watch?v=-UM73HU4jZM&feature=youtu.be

 

In this example I was a bit limited in the left border adjustment due to the fact I only had a 1K trim pot and not the 2K as specified in the latest schematic, but it still got the border in the correct location.

 

To sum it up, I think if I take this circuit design into a PCB it'll definitely stabilize things even better. So that will be the next step in the development process for this new device. While I'm at it... since this will piggy-back the GTIA, I think I'll add a header that will break-out the console switch connections (START-SELECT-OPTION), A4, BELL, and a disable line for the over scan circuit as Rybags suggested. This way it'll require virtually no-soldering when also installing any of the piggy-back TK-II products (killing two birds with one stone).

 

- Michael

Edited September 2, 2016 by mytekcontrols

[flashjazzcat]

Looks good - almost like a VGA phase/clock adjustment but for Y/C.

[+mytek]

Looks good - almost like a VGA phase/clock adjustment but for Y/C.

 

Yeah as we migrate into these newer monitors the lack of similar adjustments really dictates that some kind of external solution be implemented. Otherwise it kinda looks like crap when what's suppose to be invisible isn't.

 

Thanks for your help

 

- Michael

[tuf]

Awesome work! Combine this with the UAV and sell it

[+mytek]

Awesome work! Combine this with the UAV and sell it

 

It's in the public domain and will eventually have PCB files available for download, as well as PCB fabrication through OSH Park (I'll post a link here in this thread when they are available). I just like to create stuff and don't sell anything myself. But you, Bryan, or someone else can if you so choose to in the future. BTW, this prototype breadboard was being tested with Bryan's UAV feeding into an S-Video LCD monitor, so we know that works

 

Thanks for the kudos

 

- Michael

[+mytek]

OK here's a look at what this might be like in its final form (as a GTIA PiggyBack PCB for gating the A8 video signal). I also gave it a simple name: V-GATE.

 

 

The V-Gate Disable line when brought high will revert to normal video (no gating of the signal). This can be optionally controlled by the U1MB.

 

- Michael

Edited September 3, 2016 by mytekcontrols

[+mytek]

Jeeze I hope I'm not boring everyone with all my design changes But here is another one

 

 

I didn't really like the loading of the Diode/Resistor OR-AND gate on the color signal, so after thinking about it, I came up with this variation that uses a P-Channel MOSFET switch to gate the color signal. Overall this eliminated one diode, not a big savings, but I feel this circuit will yield much better results, feeding the color signal through relatively undisturbed. I was also able to eliminate one resistor by increasing the value of the Offset Trim Pot and therefore negating the need for a series connected fixed resistor.

 

- Michael

 

EDIT: haven't tested this, but it looks like it should work.

Edited September 3, 2016 by mytekcontrols

[+mytek]

Over-scan compensation board design getting closer to reality

 

Preliminary PCB Component Placement

 

I think this will fit in all variations of Atari 8-Bit computers, including the 400/800 series (but it will be tight). However I can't guarantee that there won't be fit problems with upgrades that also piggyback other chips. I'll be looking for such problems before I actually commit to routing traces.

 

Updated Schematic

 

I abandoned the use of a MOSFET as a Chroma Switch (just didn't work properly), and also eliminated the disable line since if this thing works as planned there really shouldn't be a reason to disable it (who would want to see the unintended over-scan anyway ). Also expanded on the support signal connections for various upgrades. Basically this would allow something like a U1MB and a TK-II-STEREO board to only need two soldered wires for installation, with everything else supported via crimp terminals.

 

If anyone can see a conflict with existing upgrades, please let me know.

 

- Michael

[Bryan]

This is pretty cool. One thing that always bothered me about Atari video is that the border shares one of your playfield colors and can't be blanked. It would have been nice to have a blackout bit.