XEGS - hardware problems

[Rybags]

Earlier today, I decided to have another go at the XEGS I messed up when fitting a socket for Antic late last year.

 

I tested all connections from Antic to their relevant pins on the 6502, GTIA (AN0-AN2 fine), and found the only problem with A15, so I jumpered it to the correct pin on the CPU.

 

So now, the machine powers up and functions, although not quite the way it should be:

 

- the screen is always a brown background, and the foreground colours seem to be shades of that colour.

- the pixels are "fat" as if the machine is always in a GTIA graphics mode. But, I plugged in a joystick and in Ballblazer the horizontal scrolling seems to be smooth.

- powering up without cartridge or keyboard, it goes to Basic instead of running Missile Command. Select/Option seem not to change that.

 

I also tried Computer Chess (cart) and it comes up with the same "fat" pixels.

 

It's as if something is up with the GTIA chip, although despite all I put this machine through, I never actually touched that chip.

 

I've not tried any further diagnostics such as plugging in the keyboard or booting from disk - it's getting late so I've packed it up for now.

 

Any ideas? I suppose my best option might be to do a bit of programming, and see if I can narrow down what exactly is wrong.

I don't think there's any "noise" or other corruption happening on the address/data buses, because I had Ballblazer running otherwise quite OK for a few minutes, aside from the graphics being totally wrong.

[carmel_andrews]

what about swapping out the bad chip for a known good chip

[Rybags]

There isn't necessarily a "bad chip" at this stage.

 

And, it's an XEGS. Antic is the only chip socketed, and that's because I added one.

[candle]

possible short on ANx pins?

but then again... not likely...

check also HALT line - i remember that i had all kinds of strange artefacts (sprite data corrupted, gr.0 letters as if it would 5 colour font mode etc) when antic had problems fetching data during DMA cycles

[Rybags]

Could be... I only went so far as to check for shorts between adjacent lines on Antic.

 

Later on, I'll try again with keyboard and SIO attached so I can investigate more in-depth.

[Rybags]

OK. Some progress after another quick look.

 

DOS boots fine.

 

Keyclick not coming through - Pokey sound and SIO is fine though. The machine cold-starts whenever you press Reset, so possibly it's thinking there's a cartridge when in fact there isn't one.

 

Possible EUREKA moment here...

 

All of GTIA's address space is reading $FF. Obviously, Chip-Select to GTIA aint working, so everything on the chip is sitting on power-up defaults. Probably explains the GTIA mode happening too... if PRIOR defaults to $FF, then you'd get the Gr. 11 type colour only mode (which is what it's stuck in).

[Rybags]

OK, I think I've found the problem. Luckily, I've got a fully working XEGS, so I can use it to help out.

 

I've traced the Chip-Select logic for GTIA and it looks like it's just a case of the 74LS138 gone bad.

Pins 15 and 16 are closed circuit in the bad machine, and open on the good machine.

 

Since /CS is active-low, it's never able to be activated on the GTIA.

 

So I guess it's $3 down at the local for a socket and another '138.

[Rybags]

Finally got it fixed - the ridiculous thing is that the actual problem wasn't as I thought and took all of 10 seconds to fix.

 

The LS138 was fine, but it's got a new socketed one now anyway. While testing for shorts among the pins in the IC socket before inserting the chip, I found that Pins 15/16 were still shorted.

 

Turns out that the earlier work I'd done putting a socket in for Antic was at fault - I'd inadvertantly melted the laquer off the top of an adjacent trace and the solder blob was connecting it to the +5 Volt going into Antic, which in turn was disabling the Chip Select for GTIA, and also causing a short between those pins up the line on the '138.

 

But... all working now, good to have one less faulty bit of hardware back in the cupboard.

Edited April 6, 2010 by Rybags

[mellis]

Finally got it fixed - the ridiculous thing is that the actual problem wasn't as I thought and took all of 10 seconds to fix.

 

The LS138 was fine, but it's got a new socketed one now anyway. While testing for shorts among the pins in the IC socket before inserting the chip, I found that Pins 15/16 were still shorted.

 

Turns out that the earlier work I'd done putting a socket in for Antic was at fault - I'd inadvertantly melted the laquer off the top of an adjacent trace and the solder blob was connecting it to the +5 Volt going into Antic, which in turn was disabling the Chip Select for GTIA, and also causing a short between those pins up the line on the '138.

 

But... all working now, good to have one less faulty bit of hardware back in the cupboard.

 

I'm glad to hear you had success fixing that XEGS. I destroyed an Atari logic board while attempting an upgrade in the early 1990s, and I was never able to resurrect it.

 

Looking on the bright side, at least that taught me to be more careful with the soldering iron in the years that followed.