800XL Boots Right To Memory Test

[simbalion]

Yesterday I picked up a very clean 800XL along with manuals and 1010 Program Recorder at a flea market. It was untested as there was now power supply with it, but the price was right and like I said, very clean with no yellowing to the case. Only blemish is the missing 800XL tag. Anyways, I booted it up with one of my good power supplies and it went right into the memory test. I did, briefly get it to boot fully and it has a beautiful, clear picture and sound when it's behaving. Sadly, it now just goes back into the memory test again! From what I can tell, all the memory is testing fine and all the boxes are coming up. Pressing the Home button does nothing and reset causes it to go to a green-blue screen and freeze up. Any suggestions?

[SoundGammon]

Pick up a basic cart and see if that works.

 

It is possible to hack the 800Xl to keep the internal basic from booting.

 

I did that years ago for a guy in our old computer club.

[+DrVenkman]

First thing I would do is open it up and see if it's socketed. If it is, I'd remove and reseat all the chip, paying especially close attention to the GTIA (since it reads the console keys [OPTION, START and SELECT] and the PIA (which affects switching to the built in Self-Test mode]. If the MMU is also socketed, so much the better. Remove and reseat that one as well.

 

If reseating all the chips doesn't make a difference, you've got some work ahead. If you have a working socketed XL or XE computer, remove the main IC's one at a time and swap them into the working machine, not vice versa. You don't want to risk damaging a working chip by putting into a non-working machine. With luck you'll find a flawed chip right off - my first guess is GTIA but could be MMU, PIA and/or CPU as well. If it's not one of those, it might also be OS or BASIC roms.

​Good luck!

[Tezz]

This occured with the XL I worked on yesterday, with that one it was just that the BASIC ROM wasn't making good contact with the socket. If your machine has been badly stored for many years it'll be worth reseating the socketed ICs as advised above and if no change, swapping them in a working machine as also advised above.

Edited April 8, 2018 by Tezz

[simbalion]

Thanks! I'll pop the unit back open this week and get that cursed RF shield off to check the chips. It is also very clean inside the unit as well. I'd say it was generally well taken care of and, the fact I got it to boot fully once after moving it about, hope it might just be a lose chip. If my rust bucket 800XL worked when I put it away, this one should.

[santosp]

Have you tried to choose the keyboard option in Diagnostics Menu, and after to check there if the "Option" key have stuck?

[simbalion]

It won't let me go to the Option menu. I turn the computer on and it goes RIGHT to the memory test mode and stays there.

[shoestring]

Unplug the keyboard ribbon from the main PCB and see if it still does it.

[simbalion]

Tried that, still does the same thing.

[shoestring]

Obviously not a keyboard issue then.

 

The option key is read by the GTIA via the resistor @ R136 so check the value of it, it should be around 220 ohms.

 

To do this, unsolder one of the ends of the resistor and measure it with your DMM.

 

If the resistor is bad then this may act like a switch where the key is permanently on.

Edited April 12, 2018 by shoestring

[+tf_hh]

The option key is read by the GTIA via the resistor @ R136 so check the value of it, it should be around 220 ohms.

If the resistor is bad then this may act like a switch where the key is permanently on.

 

The resistor is in the path of OPTION key to GTIA - so if it´s bad, always no connection would make, i.e. OPTION never could pressed - the S0...S3 inputs @ GTIA are low-active and have internal pull-up resistors. So without keyboard it would have no impact, as long the GTIA internal works fine.

 

But, when the RAM test immediately starts, then - when - START would be the problem, not OPTION.

 

I´ve seen this some times. In all cases the RAM has a defect. When computer starts, the coldstart subroutine will test the first 1024 bytes of memory. if any error, the self test is started.

 

But, the selftest has some "dummy" tests. Not every byte & range is really tested. You find a delay loop in the code, which fakes the real time needed for the test-time of one square. Also the RAM test didn´t detect RAM errors below $0400 and in the range the self-test was mirrored in ($5000-$57FF).

 

So changing RAM is the way forward IMHO.

[shoestring]

 

I´ve seen this some times. In all cases the RAM has a defect. When computer starts, the coldstart subroutine will test the first 1024 bytes of memory. if any error, the self test is started.

 

But, the selftest has some "dummy" tests. Not every byte & range is really tested. You find a delay loop in the code, which fakes the real time needed for the test-time of one square. Also the RAM test didn´t detect RAM errors below $0400 and in the range the self-test was mirrored in ($5000-$57FF).

 

So changing RAM is the way forward IMHO.

 

 

Yeah that makes perfect sense especially if the stack or zero page is effected, that can make programs do strange things too.

Edited April 13, 2018 by shoestring

[simbalion]

I finally cracked the 800XL back open tonight and, after fighting with that RF shield and losing a tiny nut under my Red Baron cockpit, I finally got into the main board. My 800XL has all socketed chips and it has '2 March 84' stamped on the board. I don't have the right tool to wiggle out the small RAM chips, but two had definitely worked up out of their sockets. One really popped back down in when I pressed on it and the chip beside it also popped down some. I went over the whole board and found two other main chips that had also worked up a bit out of their sockets. I put it back together, hooked it to the test TV, and it works! I have a faint bit of 'jail barring' on the BASIC screen now, but the graphics on Pac-Man are sharp and after a few games and tests, the unit seems stable. Might look even better on a better TV. So, I am cautiously optimistic it is repaired. Oh, that nut is still under my Red Baron...

[shoestring]

Ahh. Chip Creep, who would have thought.

 

The faint jail bars are likely normal if you're using an LCD TV. Try an old CRT TV or monitor if you have one laying around.

[+DrVenkman]

As I said above:

 

 

 

First thing I would do is open it up and see if it's socketed. If it is, I'd remove and reseat all the chips ...

 

​

 

​It's amazing how often that can fix a glitchy system. Chips work loose or - more commonly - develop oxidation on chip legs and the surfaces of the sockets. Moving them a bit breaks the layer of surface corrosion and restores electrical connectivity.