XEGS Weird Behavior Graphic Glitches with Joystick Button Press

[linuxfanman]

Hello all, I just wanted to get some input from everybody before I dive into this repair. I have a XEGS I just purchased (not working) that is doing some weird stuff. It will boot fine into Basic, Missile Command, Self Test. I can run the self test, Salt and super salt with no errors. However, it starts having issues as soon as you press the fire button on the controller, (port 1 and 2). In Missile Command, it will make the all the background graphics jump around on the screen making the game unplayable. However, sprites and color are not messed up. Sprites are located where they should be without glitching. If you run self test and press a joystick button during it, it makes the text disappear and re-appear, over and over again randomly.  If I run a cart like pole position, which you can play without pressing the fire button, it works perfect until you hit fire, and then the games graphics glitch with a bunch of garbage on screen and crashes the whole computer. Seeing that the sprites and color are working ok, I'm ruling out GTIA. I'm thinking it's the ANTIC chip, when you press the joystick button it makes the LP signal go low which it should coming off the trigger lines from GTIA, and going through the 74LS08 AND Gate. I think it has something to do with the VBI (Vertical Blank Interrupt) and when writes to the DLI (Display List Interrupt) in shadow ram $2XX, something is going haywire. The LP signal gets written to the shadow ram at next VBI, that's where I think the problem lies, like it's corrupting the DLI at that time. If pictures or video is needed I can upload some. But could this be caused by the system OS? 

 

Any thoughts??? 

[_The Doctor__]

what joystick is being used?  joystick port and trace back to chip and passive components fine to that point?

keyboard or no keyboard?

check out the fixes on these forums by tf_hh and others to ensure stability.

it is unlikely to be ANTIC.

How is power supply under load? Noise free and strong?

[linuxfanman]

@_The Doctor__  I've tried multiple joysticks, Atari, Jem, Wlco, other 3rd party all known good, all with the same issue. I've traced the line back from the port to the GTIA and the 74LS08, R69, R63, L27, L25, (no shorts, values good) all seems fine. It replicates the issue with the keyboard connected or not. Powered by my bench supply at 5 volts.  Reads 4.97v at the controller ports, GTIA and ANTIC.  Address lines and data lines do look noisy though. Clock signals looks good, no noise. 

[Beeblebrox]

Some videos might help I guess.

 

I had some issues with glitches on an xegs which was down to a bad os. 

 

 

 

Edited February 5 by Beeblebrox

[linuxfanman]

Here's some videos of what it's doing. It does the same thing when you load a cart. Once you hit the fire button, it glitches. Not sure if it does it on tape or disk based games as I have no disk or tape drives for Atari 8bits. But I'm sure it would. 

[Rybags]

Disconnect the LP input from Antic would be a handy test.

I'd think if the OS was bad then it should reflect in a fail of the Rom test since the checksum would be different.

And I take it you're happy with the LS08 AND gate?

 

Address/data noisy - how is this?

[_The Doctor__]

it's odd in that it looks like the display list moved down making text duplication and overwrite. Still stable in that the text is clearly defined, correct letters without mixed symbols etc. So why not swap some chips if you have a known good machine, swap ram, swap OS/GAME chip to start(XEGS ROM fails in woderful ways as it contains the games etc.) and it's supporting selection chips. While it's not a common Antic failure mode do as Rybags outlined.

Edited February 6 by _The Doctor__

[linuxfanman]

@Rybags I'll have to de-solder it and put in a socket. But I was thinking the same thing, just lift pin 4 (LP) to see. 

Yeah the OS passed in Self Test, and SALT tests. But since it handles the display list (I think, could be wrong. I'm still learning the 8bit),  along with ANTIC, it's got to be one of them. 

The LS08 seems to function properly. Logic seems to function as it should according to it's truth table when probing it. PH2/O2 looks good coming out too. 

Seems to me at least, to have a lot of rippling at the top of the wave in the data and address signals.  Here's a video...

ANTIC seems to be the warmest, not hot, but warmest chip on the board also...

[linuxfanman]

@_The Doctor__  Yeah I really need to just make a test rig for Atari 8Bit's like I have for the C64, and few other systems. I really just started getting into Atari 8bit's. So my chip swapping selection is really zilch right now. I have a ton of Amiga, and C64 chips (Denise, Agnus, SID, VICII, etc..) Unless I rip open my 600XL that I just fixed dodgy sockets in...  Hmmm 🤣🤣🤣 I'm still waiting for my chip tester (backbit) to come in, which tests all the 8bit chips (along with a ton of other chips).  Should I open the 600xl or wait? That's the question...  🤣😉🤣

 

 

[linuxfanman]

Well, I de-soldered the ANTIC and put a socket in. I lifted pin 4(LP) which would make it low like the button was pressed permanently and wow, glitchy graphics from the start. So I took the ANTIC out of my 600xl and put the one from the XEGS in it. Now it's producing horrible color (just way off), and the same glitches. Put the ANTIC from the 600XL in the XEGS and bingo, works like new. So it looks like I'm the search for an ANTIC CO21697. Before I head to flea bay, anybody here have one for sale?? 

[Rybags]

A stopgap could be to pull LP high - probably a pullup resistor to a nearby +5 source could do it.

Or possibly just a jumper to the nearby RNMI pin - not used on newer machines so is always high.

 

Also - this is weird behaviour, wonder if it's been witnessed before.

All that light pen processing is supposed to do is latch the VCOUNT and internal horizontal counter positions whenever there's a transition on that pin.

Edited February 6 by Rybags

[linuxfanman]

7 minutes ago, Rybags said:

A stopgap could be to pull LP high - probably a pullup resistor to a nearby +5 source could do it.

Or possibly just a jumper to the nearby RNMI pin - not used on newer machines so is always high.

 

Also - this is weird behaviour, wonder if it's been witnessed before.

All that light pen processing is supposed to do is latch the VCOUNT and internal horizontal counter positions whenever there's a transition on that pin.

I'll put the good ANTIC back in my 600XL and buy a "new" one. It is a weird failure, only thing I can figure its internally shorted to voltage or signal. When LP goes low (ground) it sends voltage/signal through what ever it's shorted to and to DL during the VBI screwing up the whole thing. Idk. Just my guess. Might explain the "noisy" address and data lines. That disappeared too after the ANTIC replacement. 

[_The Doctor__]

trigger diode on that line might stop the flow and noise, just for giggles, messing around with fix hacks can be fun and rewarding, replacing it of course would be best. Always rewarding to work around an issue.

[linuxfanman]

It was so late when I finished last night I really didn't get to test it that much. The graphic glitches are gone now. Which is great, but I can't seem to load my AtariMax 8Mbit flash cart without power cycling a few times. It seems to load to a black screen, the Self Test, or to some random game on the cart. So it's still not right. It will load regular carts just fine. So going with what Rybags said... 

22 hours ago, Rybags said:

And I take it you're happy with the LS08 AND gate?

Yeah I'm not happy with it, I think that's the reason it's having issues loading the flash cart. I'm thinking signaling timing is slightly off on PH2 causing issues loading flash carts. The 600xl loads the flash cart fine. So I put the Sally chip and 74ls08 from the 600XL in the XEGS and it still does the same. So that rules out the Sally chip. Looking through my chip stash now, it doesn't look like I have a 74f08 to swap out right now to see. But I'm thinking that's it. The 600XL buffered PH2 line looks much better then the one produced on the XEGS. 

 

Thoughts???

Edited February 6 by linuxfanman

[linuxfanman]

5 hours ago, _The Doctor__ said:

trigger diode on that line might stop the flow and noise, just for giggles, messing around with fix hacks can be fun and rewarding, replacing it of course would be best. Always rewarding to work around an issue.

Looks like a pull up resistor going from pin4 (LP) to +5VB works. At least with the bad ANTIC in the 600XL right now. I'll get a new one ordered here soon for the true fix.