Screen issues with some demos on 1040STE

[theimann]

Hi all,

I'm running into weird screen issues with some Atari ST/STE demos on my 1040 STE, see attached photo.
Basically, I have moving horizontal bars of black (and a few white) pixels running across the screen.

I did not run into these issues with older software or games AFAICT.

I have TOS 1.06 and 2.06 ROMs and have the same issues with both.
4 MB RAM, no other modifications.

My PSU has been replaced recently so it should be in good shape.
I'm not sure whether I had these issues with the old PSU or not since it's been a while, but I guess so.

Thanks for your help!
Tom

[troed]

Are the demos where this happens:

1) fullscreens
2) high-color ("Spectrum 512") style

or both?

[theimann]

It'll take me a bit to reproduce this with various demos.
What I can say for sure is that it happened with at least two demos I recently ran:

1. Sea of Colour
2. Midnight Sun

IIRC both are full screen, not sure about high-color.
As for the display, I'm using a LCD TV via Scart cable.

[evil]

It's known that MPP (which is used in these two demos) is a bit too tight on timing so these infamous dots can show up.
The way to get rid of it is to switch off the machine, and on again. And hope it will work.

I tried Sea of Colour on a number of machines before releasing it, and it very rarely got these errors.
Then my arch nemesis (hi wietze! found an STe that was very stubborn and would need a lot of retries before the garbage went away.

I don't know if this is due to shifter, glue or something else.

As the images are interlaced, there can be two states of bad pixels in one image. But you mention that the bad pixels are moving around, that confuses me and I've never seen it.


Edit: The intro and end screens are animations, and with these there can be many states of bad pixels (different for each frame), so I understand that it can look like the bad pixels are moving around then. But for the static images, it should only be two states of these bad pixels.

[troed]

Sea of Color indeed isn't "Spectrum 512 black pixels" proof - they show up on my main STE for example The solution is simple and Doug's PhotoChrome supports it, just make sure the code isn't reliant on pixel perfect switching but has two pixel leeway.

Anyway, that's not the cause for the issues shown in the screenshots. That's a much larger problem than regular "Spectrum 512 black pixels". It would seem to be a much larger Shifter desync than is usual.

[theimann]

Here it is in motion (click image for playback): https://photos.app.goo.gl/vitM2g0ajCtnN5BW2

Thanks and regards to Sweden - keep up the great work!

[troed]

I think it would be interesting to hear @Ijor's comments on this. I would ascribe it to a very unusual Shifter timing problem with its internal pixel clock stability.

[ijor]

I'm not very familiar with STE Shifter. But I would assume it is very similar to ST (fm) Shifter in this regard. So probably it is essentially the same issue that I described in other thread some time ago. Why exactly there are here more wrong pixels than usual, I don't really know.

But it should depend on the Shifter wakeup. @OP, is this behavior constant? Or it changes across different power up cycles?