800xl black screen - basketball cart works?!

[YSG2020]

Sorry for the usual black screen question but I just replaced a bunch of ram and cpu in my 800xl, and it was finally working tested and great 100%.  New tested power supply also. 
Let it sit a week and now black screen.  Will not go into self test mode. Cannot hear keyboard clicks. 
But… it WILL play Atari basketball cart just fine!  I tried a bunch of other carts including star raiders - nothing. 
Can any of you experienced wizards guess which of the chip(s) fried this time?

I’m going to go in and reseat them all and check the sockets, but any help possibly narrowing this down a bit quicker would be appreciated!

Thanks so much! Cheers.
 

[xrbrevin]

if the basketball cart is the same as the star raiders cart then it bypasses the OS ROM. this could indicate flaky OS ROM so check it and the MMU (it latches the OS ROM on boot)

the OS can be replaced by a standard 27xx series EPROM

MMU can be replaced with a programmable GAL chip

Edited January 27, 2023 by xrbrevin
syntax error

[gilsaluki]

Not sure it is related, but the last time my 130XE was finicky about certain carts it was:  Took out the 74LS08, installed a 74F08. Now all carts load.

[YSG2020]

It’s not a cart issue.  It’s definitely another bad IC. Powers to a black screen. Won’t load any cart except basketball (I tried all the early release carts which I own) which I’m guessing since very early release cart it has very simple demands on the system utilization.  It’s crazy and sad how fast these old machines drop ICs now even with a very good tested clean power supply hooked up.  Soon we will be running on pure emulation only rather than original hardware.

Edited January 27, 2023 by YSG2020

[gilsaluki]

3 minutes ago, YSG2020 said:

It’s not a cart issue.  It’s definitely another bad IC. Powers to a black screen. Won’t load any cart except basketball (I tried all the early release carts which I own) which I’m guessing since very early release cart it has very simple demands on the system utilization.  It’s crazy and sad how fast these old machines drop ICs now even with a very good tested clean power supply hooked up.  Soon we will be running on pure emulation only rather than original hardware.

I didn't mean that is was a cart issue.  The 74LS08 is an IC on the board.   Still, it may not be your particular cause, just throwing it out there. 

[Beeblebrox]

Open up the machine. Leave it powered on for a while and feel the ICs. Any running too hot to touch? If so those are the ones I'd focus on. Maybe some of the ram new chips are failing. Did you install notorious Mt branded ram by any chance? Maybe check ram is seated in their sockets, (where I assume you did socket the ram before replacing with the new chips?)

 

Also is the machine fully socketed generally? If you have another A8 which is socketed and is working you could move some of the chips from this 800XL to it one at a time to test them. 

 

Upload a hi res image of the board, (always a good move to help the community diagnose things).

Edited January 27, 2023 by Beeblebrox
Adding extra info

[Rybags]

Fairly sure Basketball is a diag mode cart like many other early ones (Star Raiders, Asteroids are others)

 

Additionally it (and some others) probably only uses the bottom 8K.

As such, issues that can cause a machine to never let you past Self-Test won't necessarily stop these carts running.

They might even be totally normal if the issue (normally failed initial Ram test) isn't too bad.

 

Such games are handy to have around since they can assist with diagnosing a faulty computer.

[YSG2020]

I’ve 3 800xl’s and as chance would have it, a rev 1, a rev 2, and a rev 3 board. Lol. One is not socketed at all, one is partially socketed, and one is fully socketed. I’m using the fully socketed board as my primary unit, and the other 2 as IC donors.  It’s like playing musical chairs swapping all the chips around trying to troubleshoot and get one fully working board. At least I’m getting good at desoldering and socket installation on the donor boards. I’m running out of working donor chips fast.   You know what’s weird? My 1200xl, 800, and XE machines run solid, no issues, reliably. But the 800xl’s keep crapping out consistently. I wonder if it’s because they all were used more back in the day and have more hours on them? 

[Rybags]

800XLs should be reliable.

Ram failure due to MT chips seems to be mostly an XE thing.

But on the other hand the XE has Freddie which means you get rid of a whole bunch of other logic and the delay line which can all be failure points.

Then back to the negatives they seem more prone to CPU failure and getting the later bad GTIA.

[YSG2020]

4 minutes ago, Rybags said:

800XLs should be reliable.

 

Only thing that’s been reliable lately on them has been their keyboards. Had to do the painted Mylar fix on all the others. And I must say, the others all have much nicer native un modded video/color circuitry also than the 800xl. 

[_The Doctor__]

Ruining 2 machines to fix 1 sounds strange to me, running out of donor chips also sounds terrible. If you take good chips and put them in a bad machine, and then the chip is bad, why would you stick another one in?

Please consider using a basic digital multi-meter, logic probe, and do the normal checks first. Voltages correct and in the right places etc. Orientation of components shorted caps, diodes etc.

Grab some field service manuals and schematics from AtariMania, Archive.org, and jerry's schematics littered across AA and other sites.

[YSG2020]

3 hours ago, _The Doctor__ said:

Ruining 2 machines to fix 1 sounds strange to me, running out of donor chips also sounds terrible. If you take good chips and put them in a bad machine, and then the chip is bad, why would you stick another one in?

Please consider using a basic digital multi-meter, logic probe, and do the normal checks first. Voltages correct and in the right places etc. Orientation of components shorted caps, diodes etc.

Grab some field service manuals and schematics from AtariMania, Archive.org, and jerry's schematics littered across AA and other sites.

That’s exactly what I’m doing.  I don’t ruin the other boards or machines - I socket and eventually replace the ICs on those also when I can find them.  I haven’t replaced any chips twice on any of the boards yet, so none of the boards are known bad boards.  The IC’s are just getting old and failing on all of the boards that have higher hours with age.  I only posted this originally to see if there were any more obvious culprits to save myself some time in more tedious troubleshooting as you mentioned above.  Looks like it’s very likely RAM, OS, or MMU though based on the feedback, so I’ll start there and hopefully save myself some time. Thanks gang!  

[_The Doctor__]

The point is IC failure rate isn't as high as your saying, so there has to be a reason for your high failure rate. Having machines running since 1988, no failures. Some near 24/7 as they were BBS machines. Please consider checking chips in a known good working machine, then you will know you have good components to start with. Just saying they're old and prone to failure (which is not the case) is not a good way to look at this.

If you are saying the chips haven't failed but are 'high mileage' and you are swapping them out for others you eventually find, how does one know the hours of use on those replacements. New old stock can fail too.

 

I would definitely not be binning chips and LSI's because they are old or have many of hours of use on them. Most solid state components of the era are rock solid. The chips sometimes fail due to micro fractures at the connection points inside the package and a decent baking sometimes reflows those connections restoring them.

 

Most common failure of chips are outlined on the forums, there are the mt ram chips, certain mexico cpu chips, and defective but work around available antics. Atari didn't have such poor quality chips as some other bread bin computers so you don't see them stacked like cord wood.

 

If it isn't broke, I wouldn't fix it. I might mod it or perform specific improvements for support chips. The LSI's are best left as they are.

Edited January 27, 2023 by _The Doctor__

[YSG2020]

This brings up another interesting question… What is the general consensus on continually using old original hardware vs. storing it properly but unused over time, for what would be more likely to cause eventual IC component failure?  Perhaps neither, and it all comes down to the quality of manufacturing of the original ICs?  Perhaps somewhere along the line, a slowly failing power supply brick was attached to all of my 800xls supplying a slowly creeping up higher voltage before I caught it and replaced it?  The damage may have been done years ago, and is only being realized now as they age further and are thus more sensitive to failure?  As mentioned, none of my other machines or disk drives, etc have failed yet (1200xl, 800, 130XE) and they all get used similarly.  I’m really scratching my head as to why only my 3 800xls keep blowing IC’s. ??  Updated Best Power brick is near new and outputs 100% correct voltage. 

Edited January 27, 2023 by YSG2020

[TZJB]

23 hours ago, YSG2020 said:

Sorry for the usual black screen question but I just replaced a bunch of ram and cpu in my 800xl, and it was finally working tested and great 100%.  New tested power supply also. 
Let it sit a week and now black screen.  Will not go into self test mode. Cannot hear keyboard clicks. 
But… it WILL play Atari basketball cart just fine!  I tried a bunch of other carts including star raiders - nothing. 
Can any of you experienced wizards guess which of the chip(s) fried this time?

I’m going to go in and reseat them all and check the sockets, but any help possibly narrowing this down a bit quicker would be appreciated!

Thanks so much! Cheers.
 

I think that I would first change the DRAM again for known tested DRAM. Could you send us some high res pics of the board in question please? We may be able to see something.

Regarding 800XL reliability, I have several and they are all working fine, except for the 800XLF which had bad RAM and Sally. A couple I have upgraded to Rambo 256K which comes in handy. Compared to the 130XE design, there is no doubt that the 800XL is more robust.

It's just a thought, but are you taking anti-static precautions while working on the boards as static electricity can kill the ICs you are handling?

[YSG2020]

3 hours ago, TZJB said:

 

It's just a thought, but are you taking anti-static precautions while working on the boards as static electricity can kill the ICs you are handling?

Fully. It was also completely reassembled, tested, used extensively and then let sit untouched. Maybe I’ve got some kind of power-on line surge issue or grounding/static problem in my house??  If so, why only the 3 800xls and none of the other stuff?  They all are plugged into the same house circuit and multi outlet surge protector too.  Use the exact same power brick on the 130xe too. No issue. 

[_The Doctor__]

You tried different monitor cords and connections as well?

 

[woj]

Unplug the power cord from the XL, all peripherals, and measure resistance between +5V and ground. On mine it is somewhere upwards of 300 ohm (my XL works fine, but does have some pecularties, like not accepting one of my OS roms that is generally checked to be good). My 130XE used to play up a lot, and also had 300+. After one of severeal chip reseatings and cleaning it went up to 700+ (the other thing was also that the reading got stable, it was not so before) and suddenly got rock stable. I am not an electrical engineer, but that should tell you if you have any current leak, like a failing capacitor or invisible dirt on the PCB. And these are definitely not good for the lifespan of your ICs.

 

Alternatively / additionally, if you have the equipment and know how to (I am only saying this because the first time I did it I almost fried my Atari, but I am generally stupid with those things) I'd measure the current draw of the running machine. Just don't remember what the healthy figure for 800XL is, for the 130XE it was a bit more than 0.6A IIRC.