Repairing an Atari 600 XL upgraded to 64 KB RAM

[mst]

Greetings mates,

Currently I am trying to fix this rare behaviour in an Atari 600 XL (NTSC),
upgraded to 64 KB RAM,

These are not my screenshots, but the problem is the same;
the machine crashes after few seconds after being power up,
showing corruption at screen.

Initially it appears a RAM problem, but I exchanged the two chips,
with another working ones, same result.

Look at the example screens:

<This is the RAM upgrade method>

64k_upgrade_zpsdyogevna.jpg

< Games are crashed with random garbage pixels and colour changes, in few seconds. 
You are lucky if you manage to load a game entirely >

DigDug_random_02_zpsayedy81g.jpg

< This usually happens at self-test screen after 10 - 50 seconds >

7MemoryBlocksIn_zpsfr4bv9ew.jpg

Researching over the web, I found this topic, talking about Freddie chip; but
this 600 Xl doesn´t have Freddie chip at all:

https://forums.atariage.com/topic/18502 ... h-freddie/

The motherboard is exactly as this, without the RF video output, replaced with the composite video mod
(as curiosity, the composite video mod looks awesome in screen, better than my 65XE):

Atari 600XL PAL placa 02.jpg

All chips are socketed in this mobo; as well I cleaned both chip legs and holders, and reseated all ICs twice.
Same result,

Are there any troubleshooting walkthrough, in order to fix and check step by step for this machine model?
Such, observe this pin signal in the oscilloscope... if the signal perform strange things
you can try this and this...

Thanks in advance,

[stephen_usher]

RAS/CAS issue with the RAM not getting refreshed?

[mst]

RAS/CAS issue with the RAM not getting refreshed?

Thanks for the appointment

Yes, it appears like a non refreshing problem,
this can explain the random nature of the computer's crashing

Where are located the RAS/CAS signal generation in this machine?

[stephen_usher]

I'm afraid I don't know. @rubber_jonnie is far more knowledgeable about the Atari 8 bits.

[foft]

The antic (pin 8) generates the refresh. There are usually, iirc, about 9 refreshes per scan line.

[rubber_jonnie]

OK, so first off, the 64K upgrade looks correctly wired, so that's good, but I'd also use a multimeter to ensure there is continuity from point to point just to be 100% sure all the soldering is good.

Next step, is the PSU good? Is it known good and working with another machine, if not then you need to verify with a known good machine and be sure the PSU is good and 5v is getting to where it should on the board. 74LS ICs tend to have a standard layout when it comes to power, with the highest numbered pin usually being 5v. So something like a 74LS04 which has 14 pins, pin 14 is VCC and should be 5v, for 16 pin ICs it would be pin 16 etc. You can pick off GND anywhere, but 74LS ICs usually have the last pin on the left hand side as GND when viewing from the top and pin 1 at the top left.

What happens if you boot with option held down? Do you get into the self test menu, and can you run the RAM test? If so what are the results as red squares mean bad RAM.

Can I ask, do you have a Star Raiders cartridge? If you do, then I'd try plugging in and seeing if it runs at all, as it cuts out the O/S IIRC. My own 600XL (It had bad RAM, a stuck bit) just gave a reddy brown screen, but would run Star Raiders. It might help with diagnosing if the Antic/GTIA/CPU are running ok if it will actually run.

If it won't run with Star Raiders, do you have a spare machine you could test the GTIA/ANTIC/CPU in? It could even be an 800XL or XE, it doesn't have to be a 600XL as they share the ICs.

If you try the above and get nothing and have a scope I'd suggest check pin 5 of the RAM for the RAS signal and pin 14 for CAS. Also check all the address lines to check they are are not stuck at an odd value. On mine one was stuck at at a level that meant it wasn't high, nor was it low and never changed. I pulled out one chip and saw it start to behave normally and then realised it was a single RAM IC that was bad.

Another problem I believe happens is that the delay line chip, U13 fails. If you can test in another machine you can verify one way or the other. I expect you could scope the delay line and compare to a working machine, but you'd need to look at the schematics to figure out the inputs/outputs in order to measure effectively.

Lastly there is an MMU IC at U2 which could fail, another one to test in another machine if you can. I know I've said test in another machine a lot, but it really is helpful to have a 2nd known good machine for testing. Doing it without really does make life a lot harder.

One last thing. Have you tried spraying the sockets with DeOxit or electrolube at all? The factory sockets are single wipe and IMHO terrible, and I replaced all of my 600XL's sockets with new double wipe types to ensure good contact with the ICs. A bit drastic I know but the original sockets aren't great.

Good luck, keep us posted on how you get on.

[mst]

Thanks @rubber_jonnie,

I provide the answers in the quoted message,

OK, so first off, the 64K upgrade looks correctly wired, so that's good, but I'd also use a multimeter to ensure there is continuity from point to point just to be 100% sure all the soldering is good.

There are continuity between the airwires and the corresponding points,

Next step, is the PSU good? Is it known good and working with another machine, if not then you need to verify with a known good machine and be sure the PSU is good and 5v is getting to where it should on the board. 74LS ICs tend to have a standard layout when it comes to power, with the highest numbered pin usually being 5v. So something like a 74LS04 which has 14 pins, pin 14 is VCC and should be 5v, for 16 pin ICs it would be pin 16 etc. You can pick off GND anywhere, but 74LS ICs usually have the last pin on the left hand side as GND when viewing from the top and pin 1 at the top left.

Yes, I have two modern replacement for Atari 8 bits PSU, these are 5 volts DC output, that provides 5.23 Volts and 5.26 volts;
observed in the osciloscope, the voltage is very stable; if you zoomed it, the DC signal is not completely straight, there is a random
white noise (frequency is not always the same) of less than 0.1 volts amplitude, aorund 0.08 and 0.05 volts amplitude.

I have replaced all electrolitic capacitors with the same value as originals,

What happens if you boot with option held down? Do you get into the self test menu, and can you run the RAM test? If so what are the results as red squares mean bad RAM.

The self test is opened, I can select Memory of Audio video and others...
audio video performs ok, as well as memory test... but after 20 or 30 minutes of use,
the machine started to perform very bad, very unstable, such the screenshots.
If you powered it off, and restarted again, the machien remains unstable.

You need to power off, and wait at least 5 to 10 minutes,
in order to operate the machine another dozens of minutes
with no apparent errors.

Curiosly, when the machine will be turned unstable, there
are some sync problems in the video output. These sync problems
can appears (few ones) when the machine is turned unstable.


Can I ask, do you have a Star Raiders cartridge? If you do, then I'd try plugging in and seeing if it runs at all, as it cuts out the O/S IIRC. My own 600XL (It had bad RAM, a stuck bit) just gave a reddy brown screen, but would run Star Raiders. It might help with diagnosing if the Antic/GTIA/CPU are running ok if it will actually run.

I don´t have any cartridge, only a Fujinet device.

When Fujinet is connected, the machine is being turned unstable
quicker than without it, you can use the Fujinet, load and play games during 5 minutes
prior to turn unstable.


If it won't run with Star Raiders, do you have a spare machine you could test the GTIA/ANTIC/CPU in? It could even be an 800XL or XE, it doesn't have to be a 600XL as they share the ICs.

I have a 65XE and a 130XE, but these ones are PAL ones instead NTSC such the 600 XL;
is this important for IC exchange?

If you try the above and get nothing and have a scope I'd suggest check pin 5 of the RAM for the RAS signal and pin 14 for CAS. Also check all the address lines to check they are are not stuck at an odd value. On mine one was stuck at at a level that meant it wasn't high, nor was it low and never changed. I pulled out one chip and saw it start to behave normally and then realised it was a single RAM IC that was bad.

I have 4 64x4 bits RAM chips to test; the machine uses two of them;
as well these four chips were tested in a ZX Spectrum +2A, loading
128 K based games such Batman the movie and La Abadía del Crimen
and played them without problems.

Another problem I believe happens is that the delay line chip, U13 fails. If you can test in another machine you can verify one way or the other. I expect you could scope the delay line and compare to a working machine, but you'd need to look at the schematics to figure out the inputs/outputs in order to measure effectively.

I need to check this, due to I have no experience or knowledge regarding this,

Lastly there is an MMU IC at U2 which could fail, another one to test in another machine if you can. I know I've said test in another machine a lot, but it really is helpful to have a 2nd known good machine for testing. Doing it without really does make life a lot harder.

I have the european 65XE and 130 XE ones

One last thing. Have you tried spraying the sockets with DeOxit or electrolube at all? The factory sockets are single wipe and IMHO terrible, and I replaced all of my 600XL's sockets with new double wipe types to ensure good contact with the ICs. A bit drastic I know but the original sockets aren't great.

I used a product called Lubri-Limp that is manufactured and intended as electronics contact cleaner.
Clean process performed when all ICs were extracted and reseated,

Good luck, keep us posted on how you get on.

Thanks a lot mate,

[tzok]

Do you have "canned air", a can with gas used to dust off electronics? If you flip it upside down, you will have a freezer. So try to freeze Antic chip (you may also try with Sally), and check if the computer is running correctly as long as you are dripping liquified gas onto it.

[rubber_jonnie]

Ok, so it sounds like you have covered a few things here, in that power is good, as is the RAM and there is good continuity with the wiring of the 64KB mod.

I note you've replaced the electrolytic capacitors, however there are a couple at least that are bi-polar, however I don't recall which ones they are. IIRC they are up near the top left end of the board and I think are axial, but it may make a difference. IIRC they have BP written on them.

I think what @tzok is valid too, you need to check for overheating chips.

[tzok]

Not even overheating. I've recently encountered a few Antics and GTIAs, and one Sally, which were not overheating, but were crashing over 25-30°C, yet while kept below 10°C they were working fine.