Atari 520ST with thermal induced screen flicker

[beatkamp]

I have been restoring my 520ST that was fitted with 512kb expansion that pokes into the PLCC adapter and terminates
at the screen shifter IC. The external power supply has been fully recapped, all electrolytics (save the two non-polar
ones) are replaced on the motherboard. The socket for the shifter was replaced with a machined socket.

initially, before any restoration, the machine would get hot and then the screen would begin to dance around
and flicker. This would occur after some ten minutes of operation time. The monitor is the monochrome SM124.

This computer is not fitted with an RF modulator.

The test program is a game called "Targhan" which is one of the lovelier games for monochrome Atari ST.

After the recap, re-seating of the PLCC shunt for the memory expansion, the machined socket for the video shifter,
the problem got less severe. The machine is now useable.

Next I added a pair of stick on heat sinks to the CPU. They get hot!! Also I removed the top RF shield completely.
Better still.

The issue now is that what I have so far done should mean I have a perfect machine.

I do not. There is a slight flicker on the screen after Targhan is running for about 15 to 20 minutes.

There is yet a problem.

I suspect that its either the 2n2904's in the video circuit or the crystal. I am going to go ahead and replace all of the 2n3904's
and hope thats the end of it. The crystal is an oddball frequency and will not be easy to source.

The machine itself has been updated to British TOS 1.04. Its a German rev 1 motherboard onto which I have installed
the British keyboard. The memory expansion was installed in Germany some time in the 80s. It was well done, perhaps
even by Atari as a 520ST+? I don't know.

Will report back to this thread on progress. If this thermal issue stems from the 2n3904 transistors, this may be of wider interest.

[1024MAK]

As it appears to be heat related, have you considered using freezer spray or using a hair dryer?

By either cooling suspect parts or by heating them, it should be possible to narrow down the problem.

The crystal oscillator can be checked if you have a suitable oscilloscope.

Mark

[Ronin47]

You mentioned removing the PLCC part of the RAM expansion, did you use anything like IPA to clean the contacts then DeOxit or Electrolube when reassembling? The PLCC sockets in STs can be quite fickle and can benefit from a clean, plus the contacts can be problematic as a result of the push in adapter.

With regards to the Shifter adapter, you should at least apply DeOxit or Electrolube when reinserting both the Shifter and the adapter.

As 1024MAK stated, freeze spray would be useful in this scenario since it appears when warm. Given it's a video issue you might want to focus on the Shifter and that adapter first, since Shifter drives the display. You may have a dry joint on that adapter board too.

Have you considered removing the RAM upgrade altogether to see if the problem goes away with the ST in a stock configuration?

[beatkamp]

I am ashamed to admit, I did not clean the contacts when I replaced the shifter socket. I will do that now.

Yes, I have a scope so I will check the shifter clock when heated. Thats a good idea.

As a last resort, I will remove the memory expander if necessary. Its a matter of pride
that I would really like to have it running.

Its clear I have more work to do before I start replacing 2n3904's. After thinking about it
some more, my basis for suspecting them is faulty. There is more work to be done.

Will try blow drying the area to see if I can make it worse.

Thanks everyone!

[beatkamp]

[Ronin47]

You might get more mileage with freeze spray than a hair dryer.

A hair dryer will heat a quite large area, and whilst it might trigger the issue, it might not be obvious where. Freeze spray can target individual chips much more easily making diagnosis simpler.

I had a bad MMU in my Apple IIe recently, and if I'd tried to trigger it with a hair dryer, it could have easily heated several other chips in the vicinity, whereas I was able to target the MMU directly.

If you have a hot air station, then that is certainly an option, because you can control the airflow better and use a nozzle to direct the heat onto individual chips more easily.

Also, don't be afraid to remove the memory upgrade, it could help target your problem more easily. It's much easier to troubleshoot a stock machine than something upgraded with other variables to consider.

[beatkamp]

Contacts cleaned. Problem persists.
Memory expansion removed. Problem persists.
The PLCC socket under the memory expansion looks rough.
I wonder if this could be it.

[beatkamp]

Ok, I will try freeze spray but I have to wait until after the holiday to get it.

Is the MMU something that fails often? I mean if the PLCC socket is ok and everything else is,
could this failure mode (screen flicker / instability) be a symptom of a failing MMU?

[Ronin47]

Contacts cleaned. Problem persists.
Memory expansion removed. Problem persists.
The PLCC socket under the memory expansion looks rough.
I wonder if this could be it.

20231223_143308.jpg

It's possible, but before you go removing the socket you should consider that the PLCC sockets used in the ST have a non-standard pin layout, so unless you have a dead ST you can recover one from, you will struggle to find a replacement.

In addition, the sockets are very difficult to remove, believe me I have removed a few.

You can do this: https://www.exxosforum.co.uk/forum/view ... hilit=plcc and as long as you are patient, you can get the contacts back to some semblance of normality.

However don't assume it is the MMU, there are still other possibilites, so just work through what you find.

As for failing MMUs, I've rarely found them to fail. That's not saying they don't, but in my experience they are generally reliable.

[beatkamp]

Ok, I have just run a test of the machine while manually pressing the MMU into its socket. This makes it work.

When I stop pressing, the screen flickers.

I gather I may safely conclude that the socket is pooched?

[beatkamp]

Ok, I will follow that Exxos tip and adjust the socket manually.

Great tip! Thank you so much for the advice.

[Ronin47]

Ok, I have just run a test of the machine while manually pressing the MMU into its socket. This makes it work.

When I stop pressing, the screen flickers.

I gather I may safely conclude that the socket is pooched?

The socket just needs a bit of TLC, it not unusual for them to have problems like this after having a long term MMU adapter installed.

In this scenario, the simplest solution is to adjust the contacts as per the guide. Just do it carefully as they are quite easy to adjust too far.

You can just focus on the worst looking ones to begin, and when you put the MMU back in make sure you use some contact cleaner as it helps the IC push into the socket a bit more easily and improves contact no end.

[Ronin47]

Ok, I will follow that Exxos tip and adjust the socket manually.

Great tip! Thank you so much for the advice.

No problem, I wrote the article, I'm an admin over on exxos forum

[beatkamp]

Update:

I manually adjusted all 68 pins on the PLCC socket for the MMU and then, just to be thorough I had a look at the bottom. Lo! There were two solder joints that appeared fractured or cold. I resoldered the entire row.

The display is now rock solid and has passed the time threshold where it would begin to shimmer and shake before.

I really should stop here but I am going to put the memory expansion back in. This expansion is to blame for the bad pin alignment in the PLCC socket and, no doubt, means I will be doing this again in a few years. I cannot explain the bad solder joints. Perhaps this machine was always problematic from day one or it was dropped or left in a freezer?

Thanks again for the help. This thread has been extremely useful.

[1024MAK]

Well done

The poor solder joints may have been there since it was manufactured. It may have worked fine when new. But over the years, the air and movement may have caused the connections to have got worse.

Mark

[Ronin47]

Well done, and good job finding the cracked joints too.

There can be several explanations for the joints:

1. It's been there from the factory and has got worse over the years.
2. It's happened due a drop of some sort. A '2 inch drop' was a recommended way to reseat chips without dismantling ST's back in the day. I wouldn't recommend it now.
3. It's a result of board flex caused when the RAM expansion was installed.
4. It's a result of the pressure caused by the MMU adapter on a possibly already weak solder joint.
5. It's a result of manual handling of the board, possibly people pushing down on the MMU when the RAM expansion stopped working.

Of course there may be other reasons, but these are the common ones I've encountered. Manual handling is quite a common cause, I had a Mega ST that died after I plugged in an IDE adapter to the expansion slot and the cause was a via that failed due to board flex. Fixed easily enough with a bodge wire, but a sod to track down.

If you are going to re-fit the RAM expansion, my advice is to do it with the mainboard out of the machine and on a completely flat surface. As you're probably already aware, the MMU adapter takes quite a bit of pressure to install, so it's best done on a completely flat surface to avoid board flex.

[beatkamp]

This machine remains working. Thank you all for the amazing help.