Intermittent memory failures on 800XL

[Sid Pokey]

In my project of bringing my childhood 800XL back to usable life to feed my insatiable nostalgia pangs, I have run up against a big problem.

 

The short version:

 

For a short time, it worked. And then it didn't. Reset to memory test which mostly failed. Then it fixed itself. Then it failed again and seemingly stayed failed.

 

The longer version:

 

Initially all appeared good. That is apart from the 1050 wasn't working. The 800XL could not access it. Giving a short rasp rather than the usual deep rumble bootup sound. Assumed the 1050 was busted. So I got an SDrive-Mini. But that didn't work either. Thereby suggesting something amiss with the SIO port perhaps?

 

So took the 800XL apart to see if there's anything obvious. Bit of a cleanup and hey presto, both the 1050 and the SDrive are now accessible. Victory… was short lived.

 

After a quick go on Elektraglide, while sitting on the title screen, the screen corrupted and it crashed. Reset drops straight into memory test where half the RAM is red. Another reset later though and memory passes. Phew!

 

Some days later with an SD card now full of nostalgic goodies, I'm back for a proper go. SDrive boots, but 10 seconds later, it happens again. This time, the ROM is red too and most of the RAM. And this time, even after another reset and everything disconnected, it has not recovered since.

 

The fact that it's obviously flakey and previously intermittent and it's both RAM & ROM suggests to me the memory itself is probably not the real problem. Something else is wrong. But what?

 

I don't have the skills, knowledge or equipment to do anything much. So can anyone offer any thoughts, suggestions or recommendations?

[TZJB]

9 minutes ago, Sid Pokey said:

In my project of bringing my childhood 800XL back to usable life to feed my insatiable nostalgia pangs, I have run up against a big problem.

 

The short version:

 

For a short time, it worked. And then it didn't. Reset to memory test which mostly failed. Then it fixed itself. Then it failed again and seemingly stayed failed.

 

The longer version:

 

Initially all appeared good. That is apart from the 1050 wasn't working. The 800XL could not access it. Giving a short rasp rather than the usual deep rumble bootup sound. Assumed the 1050 was busted. So I got an SDrive-Mini. But that didn't work either. Thereby suggesting something amiss with the SIO port perhaps?

 

So took the 800XL apart to see if there's anything obvious. Bit of a cleanup and hey presto, both the 1050 and the SDrive are now accessible. Victory… was short lived.

 

After a quick go on Elektraglide, while sitting on the title screen, the screen corrupted and it crashed. Reset drops straight into memory test where half the RAM is red. Another reset later though and memory passes. Phew!

 

Some days later with an SD card now full of nostalgic goodies, I'm back for a proper go. SDrive boots, but 10 seconds later, it happens again. This time, the ROM is red too and most of the RAM. And this time, even after another reset and everything disconnected, it has not recovered since.

 

The fact that it's obviously flakey and previously intermittent and it's both RAM & ROM suggests to me the memory itself is probably not the real problem. Something else is wrong. But what?

 

I don't have the skills, knowledge or equipment to do anything much. So can anyone offer any thoughts, suggestions or recommendations?

 

Sorry that your Atari 800XL has failed. We can surely help but would like more details.

 

Is your power supply reliable? Is it the original or a modern switching type? Can you check the output voltage with a voltmeter?

 

It sounds like from what you have said that the chips may be socketed and that your original cleanup improved connectivity, but I am only guessing.

 

More than likely the DRAM memory has failed and will need to be replaced. It is also most likely to be made by Micron Technology.

 

Can you upload photos of your board here to see what we are dealing with please?

[Sid Pokey]

I'm using a 3A USB power supply which appears stable. There was a saga behind that with the cable which was eventually solved.

 

4 of the RAM chips are Oki and 4 are Fujitsu. Neither the RAM nor the ROM are socketed.

 

What made me doubt the memory chips themselves failing is the fact that they were initially intermittent and that the ROMs apparently failed afterwards, which would seem a little too coincidental.

[_The Doctor__]

Power supply and power switch is a good place to start.

You are having and intermittent failure and those usually equate to dirty patina oxidised legs, swipes, and switches.

Make sure there are no cold/cracked solder joints or bad caps. Sockets can get weak and do fail. clean legs with slight splaying to ensure good contact can help.

 

If the problem only occurred after a specific heating up period and would go away when freeze sprayed I'd blame an IC and would located it with the spray.

 

Failing ram is a possibility but it normally stays failed.

Edited May 6 by _The Doctor__

[351cougar]

other thread said problem with cable wiring. you used all the wall pucks on that cable correct? maybe they all got damaged. without a scope to monitor input, DMM may show ok but have lots of noise getting thru. 

Did you try and re seat all chips that CAN be , pic seems to show several socketed. 

Since you said this was your long time system, and mem chips are still soldered, interesting they are mixed type. 

[flashjazzcat]

15 hours ago, Sid Pokey said:

I'm using a 3A USB power supply which appears stable. There was a saga behind that with the cable which was eventually solved.

 

4 of the RAM chips are Oki and 4 are Fujitsu. Neither the RAM nor the ROM are socketed.

 

What made me doubt the memory chips themselves failing is the fact that they were initially intermittent and that the ROMs apparently failed afterwards, which would seem a little too coincidental.

Are you actually running the machine with that PBI bracket lying loose across the contacts of the bus connector at the back of the machine?

[TZJB]

15 hours ago, Sid Pokey said:

I'm using a 3A USB power supply which appears stable. There was a saga behind that with the cable which was eventually solved.

 

4 of the RAM chips are Oki and 4 are Fujitsu. Neither the RAM nor the ROM are socketed.

 

What made me doubt the memory chips themselves failing is the fact that they were initially intermittent and that the ROMs apparently failed afterwards, which would seem a little too coincidental.

 

 

The original Atari UK power supply should be providing around 5.2V at 1.5A and is fine to use as the potting compound tends to preserve the internal components.

 

34 minutes ago, flashjazzcat said:

Are you actually running the machine with that PBI bracket lying loose across the contacts of the bus connector at the back of the machine?

 

@flashjazzcat is right. Don't do it! Dont apply power with the PBI shield flopping about like that. It needs to be removed or secured in place with screws.

 

I was wrong about the DRAM manufacture though, but it is strange that you have 4 each of two different manufacturers.

 

Examining the motherboard, all 40 pin chips except the PIA are socketed so may cause intermittent problems as they are single wipe sockets i.e. have contacts on only one side of the pins.

 

I notice there has been fluid under the shielding at points around the ground plane. At the top and the right hand side. Do you know if this was purely from condensation or was there an accident?

 

Although there is corrosion I can't see any apparent damage to the motherboard at this point, although C76 and C77 look a funny colour?

 

There is also a foreign object just above U7 C021698 pins 26 & 27 which is probably not relevant, but shouldn't be there.

 

When you remove the PBI shield could you take and upload a photo of the underside of the board please?

 

Then please apply power and see if the self test is still showing bad RAM?

 

 

[flashjazzcat]

23 minutes ago, TZJB said:

There is also a foreign object just above U7 C021698 pins 26 & 27 which is probably not relevant, but shouldn't be there.

Looks like a dead bug, although I can't be sure of its nationality.

[_The Doctor__]

Sure does need a scrubbing, and you have got to be kidding... that metal is in contact with the PBI- holy sh*t. When I read dead bug I though oh cool let's look for a memory upgrade or mod- I was not expecting a actual dead bug. Isn't cleaning the machine up the first order of business and to look for shorts, breaks, and mistakes?

 

FJC is on the jazz with photo examination. Mr. T and Murdoche need to visit SID and get a little Pokey!

[Sid Pokey]

Thank you for all the input. In answer to questions arising:

 

The photo was taken after the top shield was removed before I extracted it. It was only powered when outside of its case, by which time the metal thingy wasn't in place. I’m not that stupid. The memory failures only started happening some time after it was all back together again.

 

While out of its case, I gave it a small cleanup. Rest assured the conductive bug will not be there anymore.

 

My concern at that time was the SIO port, so it was on that side of the board where my focus of attention was. Plus I used contact cleaner on all the ports and plugs.

 

It has never had any liquid spilt on it and it’s never been stored anywhere where I would expect any dampness to get to it. So I can’t explain why there’d be any signs of such.

 

Attached is underside of circuit board. Looks okay to my eyes. Other than looking a bit messy around the RF modulator which I'm guessing was part of a previous repair in 1989.

 

Anyway, next thing to try then is lift the socketed ICs and contact cleaner on them too I guess. I’ll update when had the chance (and bravery) to try.

[flashjazzcat]

So have we completely eliminated the USB power supply (which I assume wasn't powering the machine the last time it worked) as a potential source of issues?

[_The Doctor__]

I also see liquid was into the cartridge guide, it is unmistakable. The power switch legs are completely black with moisture patina corrosion. This Atari has seen liquid no matter if you were aware of it or not. As also evidenced by spotting and and such on the landings.

So a full on de oxit party is indeed in order. Tuner cleaner for switch and pot. Check those scratched tracks and basically every preliminary thing has been covered in the thread.

Edited May 7 by _The Doctor__

[Sid Pokey]

40 minutes ago, flashjazzcat said:

So have we completely eliminated the USB power supply (which I assume wasn't powering the machine the last time it worked) as a potential source of issues?

Supplies I'd previously tried during the cable saga:

• Supply for a wireless charger - 3A, 5.4v
• LiteOn supply with a USB adapter - 3A, 5.3v
• iPad charger - 2.1A, 5.2v
• Samsung phone charger - 2A, 5.1v
• Power block with USB outputs - 2.1A, 4.9v
• Original brick - 5.3v

All tested without load, all has voltage within an normal range, and all still power other devices without any problem. Top 3 been tried since current issues began.

[flashjazzcat]

So no modern, non-USB power supply has been tried. I just distrust USB PSUs since they have been and continue to be the source of many otherwise inexplicable issues.

[Sid Pokey]

1 hour ago, _The Doctor__ said:

I also see liquid was into the cartridge guide, it is unmistakable. The power switch legs are completely black with corrosion. This Atari has seen liquid no matter if you were aware of it or not

Firstly, I don't remember what it was that plopped into the cartridge port - some foodstuff perhaps - but it's been there for decades.

 

And secondly, it's not corrosion on the power switch legs. It seems to be an insulation coating. It's supposed to be black.

 

And as for the residue on the seam to the shield on the right hand side, note how it ends abruptly. Liquid damage wouldn't do that. My assumption is there used to be some sort of glue that's long since dried up.

 

One issue is though as TZJB pointed out, something looks amiss around C74 to C77. Plus also to a lesser degree further down (C102+?). Assume they're capacitors?

 

All this though is on the right hand side which I'm assuming has little to do with the RAM. Maybe that's why I had the original problem with SIO port connectivity though?

[Sid Pokey]

13 minutes ago, flashjazzcat said:

So no modern, non-USB power supply has been tried. I just distrust USB PSUs since they have been and continue to be the source of many otherwise inexplicable issues.

The LiteOn supply is not a USB supply. I just had a USB adapter plug attached to it. I think it used to be the supply from an old TiVo box.

[_The Doctor__]

We are trying to help you, FJC acknowledged no usb supply was involved, the spots are there, moisture is moisture, does not matter if it was spaghetti sauce or mouse pee, there is something tacky left behind be it near the caps identified or the joystick ports or the components near the keycon. Rather than take issue I'd just clean it and touch anything suspect up. It does not matter if something happened yesterday or today, just one more thing to cross off your list.

[R.Cade]

It's anecdotal, but I had a very similar problem recently with my 800xl. It turned out to the the CPU. Random failures in the self test. Sometimes worked fine, sometimes not. It never ran more than 5 minutes without crashing.

I was able to get a new one from Best. It's pretty inexpensive to just get a spare set of all the main custom chips in the XL while you can.

 

Edited May 7 by R.Cade

[aeberbach]

Both the GTIA and the LM386 appear to be labelled U1 in that picture... ?

[_The Doctor__]

C42 looks like it might be cracked, like someone used it to pry up the chip next to it, it might be just a mark but maybe not.

[Sid Pokey]

1 hour ago, _The Doctor__ said:

We are trying to help you, FJC acknowledged no usb supply was involved, the spots are there, moisture is moisture, does not matter if it was spaghetti sauce or mouse pee, there is something tacky left behind be it near the caps identified or the joystick ports or the components near the keycon. Rather than take issue I'd just clean it and touch anything suspect up. It does not matter if something happened yesterday or today, just one more thing to cross off your list.

I'm not taking issue. I too am trying to cross things off the list by offering clarifications so as to minimise going off down rabbit holes chasing things that probably aren't contributing factors. I'm grateful to everyone who'd contributing.

[Sid Pokey]

56 minutes ago, aeberbach said:

Both the GTIA and the LM386 appear to be labelled U1 in that picture... ?

Camera perspective. The GTIA is U17. 😁

[_The Doctor__]

C83 has that 'protectant' on it.

[Sid Pokey]

1 hour ago, R.Cade said:

It's anecdotal, but I had a very similar problem recently with my 800xl. It turned out to the the CPU. Random failures in the self test. Sometimes worked fine, sometimes not. It never ran more than 5 minutes without crashing.

I was able to get a new one from Best. It's pretty inexpensive to just get a spare set of all the main custom chips in the XL while you can.

 

Certainly something to consider, thanks.

 

I've watched a lot of vintage computer repair videos on YouTube. And tracking down fault almost always seems to come down to having a spare one then swapping components over until the dodgy one reveals itself. Intermittent faults always offers an extra later of trouble though. 

[R.Cade]

57 minutes ago, Sid Pokey said:

Certainly something to consider, thanks.

 

I've watched a lot of vintage computer repair videos on YouTube. And tracking down fault almost always seems to come down to having a spare one then swapping components over until the dodgy one reveals itself. Intermittent faults always offers an extra later of trouble though. 

You nailed it!