Random RAM test failures on 800XL

[itripn]

Hello folks -- I have an odd issue with an 800XL I am attempting to return to working order. When I boot to the test screen and run each of the tests, everything works well (keyboard, sound, etc) except the RAM test. ROM's check out, but RAM chips randomly report bad. Usually around 30-50% of them, and it changes each time I run it. 

 

If I boot to BASIC, it gets even stranger. I can type in a short program, but as I do, occasionally random characters show up at random locations on the screen. When I try and list back the program, nothing except the ESC glyph shows up. Running the program (even a quick print hello) does not work.

 

So, I ordered a bunch of HM4864P-2 RAM chips, and a Retro Chip Pro tester (because, why not?). Short story there, original RAM chips all pass tests in the tester (if you're curious, 2 of the 24 new ones fail). I also tested the LS logic chips, all good. Everything is socketed on this board (a miracle), so I've reseated everything, and for the RAM chips even used contact cleaner (I have not used the cleaner yet on the other chips, that's next I think).

 

Still same issue. I have an original Atari Diagnostics cart for the 800, and so it's a little confused sitting in an XL. It reports ROM bad (obviously), but when I use it to run RAM tests, it asks for number of 8K blocks to test (1 - 5). If I type in 1, it says the first 8K passes, and I've never had it say otherwise in the 5 or so times I've tried it. But any parameter 2 or more it immediately fails. That could be due to the XL mismatch or something, so I am not trusting this cart too much.

 

After cleaning the contacts on all the sockets, my next step was to try and swap in POKEY, ANTIC and GTIA chips (and maybe the ROM chips?) from another working system I have, but not all of those are socketed on the other system sadly. 

 

Any other tips from the experts here?

 

Cheers,

Ron

Edited May 26, 2023 by itripn

[+DrVenkman]

I wouldn’t start swapping major chips willy-nilly. If all individual DRAMs and glue logic chips pass, you should probably verify continuity on all the address, data and control lines between all the chips (including all the DRAM chip sockets) to verify there are no damaged sockets or traces. You should also verify the MMU (swap that if you have a second working machine). 

[TZJB]

15 minutes ago, itripn said:

Hello folks -- I have an odd issue with an 800XL I am attempting to return to working order. When I boot to the test screen and run each of the tests, everything works well (keyboard, sound, etc) except the RAM test. ROM's check out, but RAM chips randomly report bad. Usually around 30-50% of them, and it changes each time I run it. 

 

If I boot to BASIC, it gets even stranger. I can type in a short program, but as I do, occasionally random characters show up at random locations on the screen. When I try and list back the program, nothing except the ESC glyph shows up. Running the program (even a quick print hello) does not work.

 

So, I ordered a bunch of HM4864P-2 RAM chips, and a Retro Chip Pro tester (because, why not?). Short story there, original RAM chips all pass tests in the tester (if you're curious, 2 of the 24 new ones fail). I also tested the LS logic chips, all good. Everything is socketed on this board (a miracle), so I've reseated everything, and for the RAM chips even used contact cleaner (I have not used the cleaner yet on the other chips, that's next I think).

 

Still same issue. I have an original Atari Diagnostics cart for the 800, and so it's a little confused sitting in an XL. It reports ROM bad (obviously), but when I use it to run RAM tests, it asks for number of 8K blocks to test (1 - 5). If I type in 1, it says the first 8K passes, and I've never had it say otherwise in the 5 or so times I've tried it. But any parameter 2 or more it immediately fails. That could be do the XL mismatch or something, so I am not trusting this cart too much.

 

After cleaning the contacts on all the sockets, my next step was to try and swap in POKEY, ANTIC and GTIA chips (and maybe the ROM chips?) from another working system I have, but not all of those are socketed on the other system sadly. 

 

Any other tips from the experts here?

 

Cheers,

Ron

 

There are better utilities available. You would need the Salt ROM for the XL series http://www.atarimania.com/list_utilities_atari_search_115.97.108.116._8_U.html

 

Also Simcheck and Simtest could be useful to you.

 

 

[_The Doctor__]

Memory, Antic, and MMU, rarely CPU and OSRom

 

memory tester checkers and other utilities litter the forums and sites. (ah I see someone is busy listing some for you)

Edited May 26, 2023 by _The Doctor__

[itripn]

Thanks to all for pointing out the additional utilities I can use -- I guess I was under the (false) impression that I would need a physical cart for these tests if the system is in an unstable state. I have both a FujiNet and an SIO2SD, but I would guess those aren't good options given this machine's state. Let me know if I am mistaken (I can try anyway).

 

Would a SIDE2 or SIDE 3.1 Cart be a better choice here?

 

Thanks for the patient help here, much appreciated.

 

Ron

 

[_The Doctor__]

AVG, Ultimate Cart etc.

The AVG might be able to be set up as a cart on one machine and then used in another more easily with it's presets and such.

A syscheck from @tf_hh would be a great option for this specific case and then can be used for a rom swapper/ extended memory after the diagnosis.

[TZJB]

Fujinet or SIO2SD will be able to load the Simtest software from the XEX files. For SALT you would need a Cart emulator. I went for AVGCart with SIO and PBI/ECI cables as it gives you more options.

 

Looking at the 800XL Service Manual https://archive.org/search?query=800XL+service

 

RAM test failures list the following ICs to check/change. I show the definitions underneath.

 

[itripn]

Thanks @TZJB that's helpful!

 

I was able to get SIMTEST and SIMCHECK to load (intermittently) from Fujinet. The check doesn't indicate anything amiss that I could tell.

 

SIMTEST is a bit of a crapshoot -- it runs sometimes, and indicates no RAM issues. Other times it lists a bunch of blocks which I assume it found issues with. MOST of the time, it runs and characters go random all over the screen and it usually freezes up. Sometimes it completes, but the screen is so full of garbage you can't see, but pressing 1 starts the test again.

 

I've ordered an AVG cart and hopefully between that and the list of chips above I can make progress. I will if I can at least swap out ANTIC as that seems like a possibility. 

 

I'll also break out the multi-meter and do some continuity tests to the best of my ability (low skillset here).

 

Thanks,

Ron

 

[Panther]

26 minutes ago, itripn said:

SIMTEST is a bit of a crapshoot -- it runs sometimes, and indicates no RAM issues. Other times it lists a bunch of blocks which I assume it found issues with. MOST of the time, it runs and characters go random all over the screen and it usually freezes up.

 

Sounds like CPU, MMU, or ANTIC to me, in that order of most likely suspects.

 

[_The Doctor__]

If errors only occur in BASIC programs, replace BASIC ROM with Rev C

Edited May 26, 2023 by _The Doctor__

[itripn]

Swapped ANTIC out with one from a working machine, no change. 

 

I have 3 spare 6502AD chips, but not tested, and each of them result in a black boot screen. Unfortunately, the 6052 in the working machine is soldered to the board. I have a couple of working Apple II machines, but not sure what vintage the 6502 is in those. Might check if I am feeling energetic.

 

Will keep at it, thanks again for everyone's suggestions!

 

Ron

[_The Doctor__]

not exactly, You need a Sally not a standard 6502

or a 6502 halt adapter socket combo.

or a Chinese remarked Sally that looks like a 6502 but is really our heroine Sally with different ink.

Edited May 26, 2023 by _The Doctor__

[itripn]

Right. That knowledge was somewhere in the back of my mind, thanks for bringing it forward @_The Doctor__

 

Cheers,

Ron

[Levas]

Are sure your 5V rail is clean and in specs?

[Forrest]

Time to replace the caps? https://console5.com/store/atari-600xl-800xl-computer-cap-kit.html

[+x=usr(1536)]

6 hours ago, itripn said:

Would a SIDE2 or SIDE 3.1 Cart be a better choice here?

Don't get a SIDE2.  Nothing wrong with them, but the SIDE3.1 is current.  If you ever decide to install a U1MB, the SIDE3.1 is what you'll want from an integration standpoint.

 

From a diagnostic point of view, the SIDE* isn't a diagnostic tool as such, but rather one that permits cartridge images to be loaded.  It can be used for diagnosis, but @tf_hh's Syscheck is probably what you want.

[+Stephen]

2 hours ago, Forrest said:

Time to replace the caps? https://console5.com/store/atari-600xl-800xl-computer-cap-kit.html

NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO.  It's NEVER time to shotgun replace caps on any machine.  Computer, amplifier, etc.  Please god - for the love of all humanity, can we stop this damn myth.

[Forrest]

Leaking capacitors has caused many computer not to work and dissolved solder joints on nearby components. It’s not a myth!

[Beeblebrox]

AFAIK generally the A8's rarely have cap issues. I believe this is more prominent with other 8 and 16 bit computers of the era. 

I've never had to recap any A8. Most common issues are usually down to failing chips, (and where present old, sockets, solder joints, etc), then of course other passive components. Of course caps can fail, but generally as mentioned a whole recapping exercise is rarely required for the A8's.

[Panther]

I've seen capacitor issues on Atari computers, but it depends on where they've been.  I think it might just come down to the temperatures they've spent their lives in.  Cool climates, no problem.  Hot climate systems seem to have more problems.

 

[+tf_hh]

4 hours ago, Forrest said:

Leaking capacitors has caused many computer not to work and dissolved solder joints on nearby components. It’s not a myth!

Well... after more than 30 years of repairing and hundreds of 8-Bit systems in my hand I would say: It´s a myth - *when* talking about Atari 8-bit machines.

 

Leaking / dry capacitors are mostly a problem of newer machines, and also then not every model. For example, the Atari ST series is mostly unaffected by capacitor problems, except their power supplies, of course. The newer the machine, the chance of cap issues went higher. Also Commdore Amiga. While Amiga 500 and 1000 machines mostly have caps working until today flawless, the Amiga 600 and 1200 are well-known for leaking SMD electrolytic caps and sometimes heavy damages to the board.

 

Talking about Atari 8-bit IMHO the only two mainboards with cap issues are some XE series and the 800 XLF (with freddy). At those boards the 470 uF capacitor(s) sometimes are bulged and should be replaced. A leaking cap I never saw at any Atari 8-Bit machine except the 1050. The 1050 has four 47uF caps for motor power control, these one often are bulged and sometimes leaking.

 

Just my 2 cents...

 

[flashjazzcat]

1 hour ago, tf_hh said:

Talking about Atari 8-bit IMHO the only two mainboards with cap issues are some XE series and the 800 XLF (with freddy). At those boards the 470 uF capacitor(s) sometimes are bulged and should be replaced.

Confirmed. 800XLF featured in recent video had two obviously bulging 470uf radial caps. Very common.

[+DrVenkman]

But in any case, bulging or out of spec caps on bog-standard 800XLs are not likely to cause intermittent RAM issues. There’s something wrong with ANTIC, MMU or one of he logic chips probably. The OP has said the logic chips test good, so perhaps a socket issue on one or more chips. I would spend the time to do basic pressure/wiggle tests on the running board, pressing and nudging the socketed chips to see if that induces a failure. If so, inspect that socket very closely for cracked or broken solder joints and damaged traces, then remove the chip and look for bent, corroded or damaged wipes. Repeat as necessary. If you don’t find issues, it’s time to grab schematics and pin outs and start measuring signal continuity methodically all around the main signals tedious as that can be, it WILL find your problem eventually. 

[Forrest]

All,

There was a ? At the end of my suggestion. Then Stephen makes his outrageous statement it’s all a myth.

I have had multiple computers made in the 90’s die due to electrolytic capacitors - it’s not a myth.

[Beeblebrox]

I think it can happen to other 8 bit computers like c64s, etc a lot morr but generally as mentioned by those on this thread the majority of the A8 models don't suffer from cap problems and don't need recapping. Of course there are exceptions but generally recapping an A8 isn't usually something that is required.