Installed 74F08 in 800XL, machine no longer boots

[+x=usr(1536)]

59 minutes ago, mickster said:

O-scopes can now be bought for $140 from amazon.

Are you volunteering to shoulder that purchase on my behalf?  Because if you aren't, it's not relevant in the least.

59 minutes ago, mickster said:

There is also plenty of tutorials on youtube (adrian's digital basement is a personal favorite).

Yes.  It's also a channel I've been subscribed to for around 3 years at this point.

59 minutes ago, mickster said:

Best bet is to unsolder any sockets you have installed and check the board with a magnifiers headset (also can be bought from amazon very cheaply).

Both of these things have been done more than once.

59 minutes ago, mickster said:

I know you said you have checked your work, but most likely you have not found your mistake.

And I'm entirely willing to accept that that may still be the case.  What is tiresome is that people aren't bothering to find out what has been done before giving opinions, and there are only so many 'I've already done that' responses I can write before it grows old.

59 minutes ago, mickster said:

Also the fact that changing the Sally changed your video output doesn't really mean that Sally was bad.  If the data/address buss is messed up, different chips can cause a different 'locked up' state.

I never said that it was bad.  I said that I swapped it and noticed a change.  Not the same thing.

59 minutes ago, mickster said:

I cannot stress enough that this is not a component failure, but circuit board damage.  It is fixable since its such a simple pcb. 

If it is PCB damage, then why has continuity consistently been OK on the PCB?

59 minutes ago, mickster said:

Once again, if any of this seems negative, I really don't mean it that way.  Have fun trying to figure out.

It doesn't come across as negative, but it is coming across as armchair quarterbacking as opposed to assistance.  If that's not the intention, then fine, I'll admit that I was wrong, and my apologies for misreading the intent.

[+x=usr(1536)]

Deleted.

[_The Doctor__]

it was working, you installed a socket and put in a chip, it no longer worked. I'd start at the socket, traces in around and under it. Make certain of chip orientation. Socket could be bad, damage from desolder, prying etc. Hopefully the PCB wasn't plugged in during desoldering, soldering.

[+x=usr(1536)]

52 minutes ago, _The Doctor__ said:

It's unclear what tools and diagnostic equipment you have at your disposal,

Diagnostic equipment: nothing beyond human senses and thought.

 

Tools: soldering iron.  Screwdrivers, etc.

52 minutes ago, _The Doctor__ said:

the number of dead components does not make sense and is staggering.

I agree, but that's what is being measured.

53 minutes ago, _The Doctor__ said:

Usually when there is a catastrophic failure of shorts etc, there is evidence of too much voltage and current passing through the components and or circuits, not always mind you but usually at least one thing or another shows that stress.

Also agreed.

54 minutes ago, _The Doctor__ said:

I have a bad feeling the patient was on life support and the surgeon is committing murder. I hope to help you, but don't know how.

I'll be perfectly honest: telling me that you didn't read any of the earlier posts detailing what had been done yet offering opinions based on who-knows-what came across as unbelievably arrogant.  While that may not have been the intention, it certainly set a tone and not a particularly amicable one.

56 minutes ago, _The Doctor__ said:

Others have commented but all are met with rebuttals.

You're right.  It's so much better to not put forward an idea in response to someone and just sit there, blindly doing whatever they say, rather than offering an idea that they may critique as they see fit.

[_The Doctor__]

We read the thread before posting, got lost in the unfolding disarray, and didn't want to go back to sort it all out again. This didn't mean we didn't read it in the first place as you assert. Went back even though the thread's getting longand read it again anyway. Noticed that rather then acknowledging thoughts and responding with results/or doing something again and double checking a result you make rebuttals. You appear frustrated and respondents are here to help, not quibble.

If all you have is a soldering iron and screw drivers, you don't have what you need to perform repairs, I suspect you have a meter or some other equipment but chose to be kurt in you response, how else could you test items? Please let people know what you have. They only want to save your machine.

[mickster]

9 minutes ago, x=usr(1536) said:

Are you volunteering to shoulder that purchase on my behalf?  Because if you aren't, it's not relevant in the least.

Yes.  It's also a channel I've been subscribed to for around 3 years at this point.

Both of these things have been done more than once.

And I'm entirely willing to accept that that may still be the case.  What is tiresome is that people aren't bothering to find out what has been done before giving opinions, and there are only so many 'I've already done that' responses I can write before it grows old.

I never said that it was bad.  I said that I swapped it and noticed a change.  Not the same thing.

If it is PCB damage, then why has continuity consistently been OK on the PCB?

It doesn't come across as negative, but it is coming across as armchair quarterbacking as opposed to assistance.  If that's not the intention, then fine, I'll admit that I was wrong, and my apologies for misreading the intent.

 

An O-scope is very relevant at this point. Unless you are done with the hobby.

 

Unless you never plan on working on these computers again,  This is the perfect problem to start using one.  If you go back 30 years, this is when it would have been suggested you go to Radio Shack and buy a logic probe.  Scopes are so cheap now,  it doesn't even make sense to use those anymore.  You need to be able to see your problem.  An o-scope will do that for you.  

 

As far as if it is pcb damage, you say that you have continuity, that doesn't mean everything is ok.  First of all, are you using ohms or continuity tester.  In this situation make sure you use ohms.  Everywhere you have touched with an iron, can break a trace or bridge a trace.  Magnifiers is what you need to use here, not just reading glasses or if you are lucky enough to have good vison the naked eye.   Also, its possible that you have a bad through hole at this point, that would be very easy to miss.

Also the smallest fleck of solder can bridge two lines together and the machine will not work.  

 

If you want to continue to use the shotgun technique , the best bet would be set another 800xl that works  next to the broken one and swap every component one chip/component  at a time until you fix it.  If its a component, then you will find it.  I would bet against it though.

 

Its not armchair quarterbacking  because I don't care if you made a mistake( or continue to do so).  I am sure you are doing this for fun of it, so it really doesn't matter in the end if you fix it or not.   It is assistance because I am just trying to point you in the right direction.  The atari 800 is an extremely simple computer with very robust components.  With the exception of bad ram occasionally and a bad cpu here and there, they usually just run forever.  The problem is, when a person starts to flex a 40 year old pcb and then take a hot iron to it, new problems can arise.  The fragile part of the computer is not the components, its the PCB (especially when not using hot air station or a desoldering tool). 

 

Once again,  good luck on your repair.  It doesn't really matter if you fix it, if you have fun trying.

 

 

 

 

 

 

[+scorpio_ny]

I wanted to relay a similar experience. My main to go 800XL had stopped working suddenly after just opening it it examine the PCB to get some information. The 800XL was heavily modded (Sophia, Pokeymax, ultimate 1MB) and was my go to machine. After some initial trial and error and reverting back to stock, it still had the issue (with stock GTIA, I was getting a red screen). Swapped components with known good ones and still no dice. I finally had some time to  dedicate to it and after a couple of hours, I discovered what was wrong. Some of the wiring that was used to connect the Ultimate 1MB to the PCB had frayed ever so slightly at the solder points. It had caused a short circuit it seems when I moved the wiring. It was so innocuous, I missed it after looking at two or three times. I was able to find it by painstakingly using my multimeter to look for shorts all over the motherboard. This made me change the way I interface with the motherboard. When I can, I use single posts so I can use DuPont connectors to connect to PCB. I see that there was extensive wiring done on the PCB. I take a closer look at the custom wiring done with a meter to make sure that no shorts occurred when you opened the computer.

[+x=usr(1536)]

19 hours ago, mickster said:

 

An O-scope is very relevant at this point. Unless you are done with the hobby.

Note: I do not disagree with you on this point.  More:

19 hours ago, mickster said:

Unless you never plan on working on these computers again,  This is the perfect problem to start using one.  If you go back 30 years, this is when it would have been suggested you go to Radio Shack and buy a logic probe.

OK, I can actually say that I've slacked on replacing my last logic probe, which bit the dust about 5 years ago.

19 hours ago, mickster said:

Scopes are so cheap now,  it doesn't even make sense to use those anymore.  You need to be able to see your problem.  An o-scope will do that for you.  

It doesn't matter how cheap they may or may not be if it's not financially viable to purchase one.

19 hours ago, mickster said:

As far as if it is pcb damage, you say that you have continuity, that doesn't mean everything is ok.

I concur.

19 hours ago, mickster said:

First of all, are you using ohms or continuity tester.  In this situation make sure you use ohms.

Are you referring to testing traces?  If so, continuity.  Why would measuring resistance be preferable to continuity testing?  This is something that I'm not understanding, so would appreciate clarification.

19 hours ago, mickster said:

Everywhere you have touched with an iron, can break a trace or bridge a trace.  Magnifiers is what you need to use here, not just reading glasses or if you are lucky enough to have good vison the naked eye.   Also, its possible that you have a bad through hole at this point, that would be very easy to miss.

Also the smallest fleck of solder can bridge two lines together and the machine will not work.  

I agree with you 100% on the above.  I also agree that whatever the problem(s) is/are, they haven't been found.  But these are things I've been over time and time and time again with no problems becoming apparent.  Now, that doesn't mean that there isn't a problem there - but it does beg the question of how many times I should repeat the same thing before moving on.

19 hours ago, mickster said:

If you want to continue to use the shotgun technique , the best bet would be set another 800xl that works  next to the broken one and swap every component one chip/component  at a time until you fix it.  If its a component, then you will find it.  I would bet against it though.

Already done using a 600XL, at least for chip testing.  No, they are not exactly the same machines, but as I don't have a spare 800XL to work with it's my closest equivalent and will at least show if the component(s) in question are passing or failing.  So far, no failures.

19 hours ago, mickster said:

Its not armchair quarterbacking  because I don't care if you made a mistake( or continue to do so).  I am sure you are doing this for fun of it, so it really doesn't matter in the end if you fix it or not.   It is assistance because I am just trying to point you in the right direction.

I am more than happy to be assisted; it's why this thread was started.  What I am not happy about is that what I have done up to this point is seemingly being ignored by those who say they want to help, leading to frustration on my end when I try to explain what has already been done and am met with responses that basically come down to, "you're doing it wrong."

 

If I'm doing something wrong, fine.  I have no problem with being told that, and am absolutely receptive to better ways of doing things.  But when the advice is given in ways that come across as having an air of, "because I say so," to them, we end up with situations like this.

19 hours ago, mickster said:

The atari 800 is an extremely simple computer with very robust components.  With the exception of bad ram occasionally and a bad cpu here and there, they usually just run forever.

I agree: in my experience of using these computers over the last nearly 40 years, they're remarkably solid.  

19 hours ago, mickster said:

The problem is, when a person starts to flex a 40 year old pcb and then take a hot iron to it, new problems can arise.  The fragile part of the computer is not the components, its the PCB (especially when not using hot air station or a desoldering tool).

Thing is, my experience has been contrary to that: component failures have far exceeded PCB failures for me.  I'm not saying that they can't happen - they absolutely can -  but that's not something I'm seeing in this instance.  If the failure is in the PCB, I've been unable to replicate it.  If it's in the socket, the third socket is now in place and all of them (or the work around them) has been bad, but I can't find the evidence of where that may be happening, either.

 

The two solid things that I can point to at this time is that the transistors are shot and there's continuity where it shouldn't be.  Why the transistors are dead, I don't know, and as much as I would like to know the root cause, there is nothing that has become evident in testing.

19 hours ago, mickster said:

Once again,  good luck on your repair.  It doesn't really matter if you fix it, if you have fun trying.

I truly do appreciate the sentiment, and thanks.

[+x=usr(1536)]

19 hours ago, _The Doctor__ said:

Noticed that rather then acknowledging thoughts and responding with results/or doing something again and double checking a result you make rebuttals. You appear frustrated and respondents are here to help, not quibble.

I have double-checked results.  Repeatedly.

 

This is what's feeding into a large part of my frustration: either I have failed to mention that in places where it should have been (in which case, it's my fault for not explaining that), or that the places where this would have been described have been skipped entirely by the people who haven't made the effort to go back and find out (not my fault).

 

Either way, what you refer to as a 'rebuttal' is, I suspect, the statement that I've already tried something with no results.  That's not a rebuttal: it's a response.  If you have found a statement that seems otherwise, please quote it so that I may clarify the meaning around it.

19 hours ago, _The Doctor__ said:

If all you have is a soldering iron and screw drivers, you don't have what you need to perform repairs, I suspect you have a meter or some other equipment but chose to be kurt in you response, how else could you test items?

Yes, I have a meter.  Asking the question of what I have at my disposal is very open-ended, however.  I could say that there's a three-prong bearing puller sitting in the next room over, but that's not really relevant in this case.  A basic, "do you have any of the following," question asking about specific tools would be helpful so that I can give a response that actually applies to the question being asked.

19 hours ago, _The Doctor__ said:

Please let people know what you have. They only want to save your machine.

OK, I can list off a DMM, soldering iron, and screwdrivers to start.  I do not have an oscilloscope, powered desoldering pump, or logic probe.  Is there anything else you may specifically want to know about?  I'm more than happy to answer the question, but need to be able to do so accurately.

[flashjazzcat]

Seven months after a single IC was replaced on the board, this thread reminds me of this scene:

 

Edited February 2, 2023 by flashjazzcat

[+mytek]

On 6/30/2022 at 3:11 PM, x=usr(1536) said:

This morning, I decided to replace the 74LS08 at U18 in my 800XL (256K RAM, UAV, NTSC, Rev. C PCB) with a socket and 74F08.  After doing this, the machine booted to a red screen.  Swapped the 74LS08 back in; no change.

This is how it all started, unless I'm missing something.

 

So if I were to assume that the computer was fine before this change occurred, then it can only be one thing that has caused the problem you've been chasing your tail for half a year now, and that is damage was done to the computer motherboard by removing the original 74LS08 and replacing it with a socket.

 

I like so many others are not going to go back through every post in this thread to retrace all that has happened since that unfortunate day in June. But I just can't fathom how this has not been resolved by this time, since the area of concern should have been limited to a relatively small area where that new socket got installed. Something either got shorted, or a trace and/or pad got lifted, or the socket itself has a problem. BTW was that socket ever removed so that the area underneath could be inspected and then if nothing was found amiss, was an entirely new socket installed?

[Beeblebrox]

5 minutes ago, flashjazzcat said:

Seven months after a single IC was replaced on the board, this thread reminds me of this scene:

 

That is one of my all time fav scenes from FT. Get's me everytime, along side the "these (cows) are small..." scene. Love it heh heh!! 

 

[mickster]

30 minutes ago, x=usr(1536) said:

Note: I do not disagree with you on this point.  More:

OK, I can actually say that I've slacked on replacing my last logic probe, which bit the dust about 5 years ago.

It doesn't matter how cheap they may or may not be if it's not financially viable to purchase one.

I concur.

Are you referring to testing traces?  If so, continuity.  Why would measuring resistance be preferable to continuity testing?  This is something that I'm not understanding, so would appreciate clarification.

I agree with you 100% on the above.  I also agree that whatever the problem(s) is/are, they haven't been found.  But these are things I've been over time and time and time again with no problems becoming apparent.  Now, that doesn't mean that there isn't a problem there - but it does beg the question of how many times I should repeat the same thing before moving on.

Already done using a 600XL, at least for chip testing.  No, they are not exactly the same machines, but as I don't have a spare 800XL to work with it's my closest equivalent and will at least show if the component(s) in question are passing or failing.  So far, no failures.

I am more than happy to be assisted; it's why this thread was started.  What I am not happy about is that what I have done up to this point is seemingly being ignored by those who say they want to help, leading to frustration on my end when I try to explain what has already been done and am met with responses that basically come down to, "you're doing it wrong."

 

If I'm doing something wrong, fine.  I have no problem with being told that, and am absolutely receptive to better ways of doing things.  But when the advice is given in ways that come across as having an air of, "because I say so," to them, we end up with situations like this.

I agree: in my experience of using these computers over the last nearly 40 years, they're remarkably solid.  

Thing is, my experience has been contrary to that: component failures have far exceeded PCB failures for me.  I'm not saying that they can't happen - they absolutely can -  but that's not something I'm seeing in this instance.  If the failure is in the PCB, I've been unable to replicate it.  If it's in the socket, the third socket is now in place and all of them (or the work around them) has been bad, but I can't find the evidence of where that may be happening, either.

 

The two solid things that I can point to at this time is that the transistors are shot and there's continuity where it shouldn't be.  Why the transistors are dead, I don't know, and as much as I would like to know the root cause, there is nothing that has become evident in testing.

I truly do appreciate the sentiment, and thanks.

 

The reason for using ohms over continuity is that to prove traces are good, you want a dead short (0 ohms),  continuity tests for current flow. On a  pcb its very easy for trace to 'ring continuity' but be flowing through other components.  Another way to put it is that continuity will get you close, but the ohms check is the 'true' check that needs to be done, especially when having problems.  For example  there should be some form of continuity flowing from +5volts to ground. Otherwise the pcb wouldn't use electricity at all and would not function. IF you take a meter and do a resistance reading from +5 to ground it cannot show 0 ohms or there would be a dead short (also would not function).   

 

Also when testing transistors, you should be using diode check and observing voltage drops,  verses continuity or ohms.  I am sure youtube can explain this better than me.   This is a situation where you may want to compare readings with the 600xl.

 

As far as how many times should you look at the pcb for failures before you go back to swapping components really comes down procedure.

 

Doing the same thing you probably won't catch the problem since you have checked multiple times, so what needs to be done now is it needs to be checked in a different way.  IF it was my pcb and you handed it to me, what I would do is:

 

1) remove your work (not that its bad, but that's the only way to see).  Sometimes this is when it pays to not use a socket because its easier to see traces on the top side of the pcb.

 

2) Once everything is removed, the pcb is visually inspected in multiple ways (with and without magnification).  Many times I will see a broken trace by holding the pcb up to a bright light and look at the pcb with the light shining through the opposite side.

 

3) Use another machine or schematics to ohm out all the traces checking for both shorts and breaks

 

3) replace the components without sockets since my gut is telling me its a pcb problem (many many people would disagree with me on this)

 

4) fire it and do a quick check of video, but plan on it not working.

 

5) Use a scope and start with the normal procedure of checking cpu for clock,reset, address lines, data lines.

 

6) Use a scope on affected area and check for signals flowing in and out.

 

 

I know you don't want the scope to be part of the solution, but at this point its going to be very very very hard to fix the problem without it.

 

For example.

 

You have a flat tire and are stranded.

 

The only tool you have is a screwdriver.

 

You still  need to change your tire.

 

You ask people for advice, but what do you tell someone that needs to change their tire with a screwdriver??

 

If you swap a chip out of a pcb, the only thing a person should need is soldering iron, solder, wick or bulb, and a meter to check your work.

 

To fix a non working pcb requires more tools, experience, magnification and light.  

 

 

Also , it should be noted, if you are trying to fix this for fun, that's great.  Pour in all the tools and parts and do your best to enjoy the journey.

 

If you just want a working 800xl, it would probably be much easier at this point and cheaper to buy another one.  

 

 

 

 

 

 

 

[reifsnyderb]

Well, I've kept my mouth shut.  But, I'll make a very generous offer as I like a challenge.  I'll fix this 800XL free of charge.  Just ship it to me and I'll have it working within days.  Part of the deal involves that I make sure it works before shipping it back.  So, I will gladly test it thoroughly via an extensive multi-year testing process.   🙂

 

Edited February 2, 2023 by reifsnyderb

[reifsnyderb]

Some months ago, I picked up an Atari 800XL that was described, on ebay, as being DOA since new.  The guy got it at a going out of business sale and tried it some months later to find out it didn't work.  So, he put it in storage.  I found it on ebay and hit the buy it now before even reading the whole description as it's in new condition.  (It all does appear brand new, too.  It's beautiful!)  The long story short is that I literally removed every chip, from the board, and tested it on another machine.  The only chip I didn't test was the delay line as it was soldered in.  Finally, I de-soldered and tested the delay line and it was the problem.  

 

With your computer, and at this stage, we really don't know where the problem is as so much has been done.  So, I'd take another 800XL board and test each chip one at a time to make sure all the chips are good.  (Yes, I'd test them all again.)  Next, I'd remove each and every upgrade and change to your original board and restore it back to it's original Atari condition.  Next, I'd remove any sockets I added, inspect the board thoroughly, and replace those sockets with brand new sockets.  (I had a brand new socket that didn't work and caused me grief once.)  Another thing I'd do is, without any chips on the board, check continuity between +5 and ground to make sure there aren't any shorts.  Use a meter for this to make sure the resistance is very low.

 

Something else that can be done is to do a "stare and compare" between a working board and your board.  Have both boards configured the same.  Then use a meter and check resistances all over both boards to make sure they are close.

[+x=usr(1536)]

2 hours ago, mytek said:

This is how it all started, unless I'm missing something.

This is correct.

2 hours ago, mytek said:

But I just can't fathom how this has not been resolved by this time, since the area of concern should have been limited to a relatively small area where that new socket got installed.

Because I haven't been working on this problem non-stop for seven months.  It's been picked up, put down, and picked up again as time permits.

2 hours ago, mytek said:

BTW was that socket ever removed so that the area underneath could be inspected and then if nothing was found amiss, was an entirely new socket installed?

Yes.  There have been three different sockets in that location at this point.  Symptoms did not change when any of them were replaced, nor did any measurements that were taken when it was in or out of the machine.

[+x=usr(1536)]

2 hours ago, flashjazzcat said:

Seven months after a single IC was replaced on the board, this thread reminds me of this scene:

Believe me, it's not just you.  This has been picked up and put down so many times that I'm starting to lose track at this point.

[+x=usr(1536)]

1 hour ago, mickster said:

The reason for using ohms over continuity is that to prove traces are good, you want a dead short (0 ohms),  continuity tests for current flow. On a  pcb its very easy for trace to 'ring continuity' but be flowing through other components.  Another way to put it is that continuity will get you close, but the ohms check is the 'true' check that needs to be done, especially when having problems.  For example  there should be some form of continuity flowing from +5volts to ground. Otherwise the pcb wouldn't use electricity at all and would not function. IF you take a meter and do a resistance reading from +5 to ground it cannot show 0 ohms or there would be a dead short (also would not function).

Makes sense.  I'll re-measure using resistance rather than continuity.

1 hour ago, mickster said:

Also when testing transistors, you should be using diode check and observing voltage drops,  verses continuity or ohms.

Correct, and this is what I've been doing - base to emitter, base to collector, then collector to emitter.  Continuity has only been used on them as a secondary check.

1 hour ago, mickster said:

Doing the same thing you probably won't catch the problem since you have checked multiple times, so what needs to be done now is it needs to be checked in a different way.  IF it was my pcb and you handed it to me, what I would do is:

 

1) remove your work (not that its bad, but that's the only way to see).  Sometimes this is when it pays to not use a socket because its easier to see traces on the top side of the pcb.

Note that I am saying this not to negate the idea, but rather to make sure that everyone is on the same page.  This has been done several times, both with and without the socket in place.

1 hour ago, mickster said:

2) Once everything is removed, the pcb is visually inspected in multiple ways (with and without magnification).  Many times I will see a broken trace by holding the pcb up to a bright light and look at the pcb with the light shining through the opposite side.

I have tried this, and it has not shown anything obvious.

1 hour ago, mickster said:

3) Use another machine or schematics to ohm out all the traces checking for both shorts and breaks

That one I'll redo from schematics.  This is why I normally have a second of each machine on hand, totally stock, that I can compare to.  Unfortunately, with the 800XL, I don't.  My concern with using the 600XL for this specific task is that it's not the same PCB, so won't return the same results.

1 hour ago, mickster said:

3) replace the components without sockets since my gut is telling me its a pcb problem (many many people would disagree with me on this)

I'm open to it being a PCB problem.  I'm open to it being any problem, really.

1 hour ago, mickster said:

4) fire it and do a quick check of video, but plan on it not working.

Doable.

1 hour ago, mickster said:

5) Use a scope and start with the normal procedure of checking cpu for clock,reset, address lines, data lines.

 

6) Use a scope on affected area and check for signals flowing in and out.

Related to 5) and 6) above:

1 hour ago, mickster said:

I know you don't want the scope to be part of the solution, but at this point its going to be very very very hard to fix the problem without it.

It's not that I don't want the scope to be part of the solution, but rather that buying one is out of the question at this time for reasons I do not wish to get into publically.  Your point is understood, but there's just nothing I can do about it right now.

1 hour ago, mickster said:

Also , it should be noted, if you are trying to fix this for fun, that's great.  Pour in all the tools and parts and do your best to enjoy the journey.

This is also part of the reason that it's taken seven months to get to even this point: it's a hobby, not grandma's dialysis machine.  I spent a year getting a 5200 running at one point largely because it wasn't essential to go through it in any particular timeframe.  That people are getting pissy over how long this has taken when it's not their project is frankly baffling to me.

1 hour ago, mickster said:

If you just want a working 800xl, it would probably be much easier at this point and cheaper to buy another one.  

I don't disagree with you, but there is sentimental value to this particular machine for me.  Whenever I (eventually) acquire another one, it'll be the kept-stock reference machine.

[+x=usr(1536)]

30 minutes ago, reifsnyderb said:

With your computer, and at this stage, we really don't know where the problem is as so much has been done.

Which is a fair point.  Frankly, I'm losing track in a lot of ways.  That's a large part of the reason why this thread was started: to give me a reference for what has been done so that I can step away from it, come back, and refresh my memory.

30 minutes ago, reifsnyderb said:

So, I'd take another 800XL board and test each chip one at a time to make sure all the chips are good.  (Yes, I'd test them all again.)

Nearest equivalent I have to do that with is the 600XL, but I can do that.

30 minutes ago, reifsnyderb said:

Next, I'd remove each and every upgrade and change to your original board and restore it back to it's original Atari condition.

With the exception of the sockets, this has been done.  More:

30 minutes ago, reifsnyderb said:

Next, I'd remove any sockets I added, inspect the board thoroughly, and replace those sockets with brand new sockets.  (I had a brand new socket that didn't work and caused me grief once.)

To reiterate: this has been done three times.  I can do it again, but please understand that if the previous three replacements didn't work, I'm doubtful that a fourth will.  That's not to dismiss your suggestion, which I fundamentally agree with, but it does give me pause to wonder if it'll be another replacement without result.

30 minutes ago, reifsnyderb said:

Another thing I'd do is, without any chips on the board, check continuity between +5 and ground to make sure there aren't any shorts.  Use a meter for this to make sure the resistance is very low.

OK, that I have not done (only socketed ICs have been removed when testing), but can make happen.

[reifsnyderb]

Another thought regarding the sockets.  With the chip removed, use a meter and check every pin socket to make sure there is no low resistance connections between pins that should not be connected.  Also, make sure each pin will have a good connection from the socket to the board.

[_The Doctor__]

After the empty sockets tests, consider flipping the board over and making sure nothing get picked up or bent on the bottom side of the PCB. I've had boards come in with legs bent or touching, as well ass solder / work debri or flakes hiding in plain but well disguised sight.

 

I sure hope you are taking anti static precautions for you and your iron, placing a pcb on a slightly conductive anti static bag is not adequate and can prove bad news if powered up with a legs / pins etc touching it. If there is a differential between the iron tip and said circuits that's a bad thing also. Delay lines aren't the only things susceptible to esd.

 

Last hail Mary things are resistivity in switch contacts etc.

Flux can be clear and not seen, imagine looking at a beautifully cleaned, smooth tinned IC leg and finding it's not conductive, hence a mention of deoxit, high IPA, and fiber pens to clean such chip legs etc.

Edited February 2, 2023 by _The Doctor__

[magnusfalkirk]

x=usr(1536) I sent you a PM

[woj]

I will add a bit to the uncertainties. I used to have tons of problems because of invisible flux, that's one. What I discovered is good to "visualize" it is to use acetone, it then matts. But surely working with acetone is totally nasty and I use it as last resort.

 

The other is about swapping chips around - one of my OS roms is clearly OK when (a) read with an eprom burner, (b) installed in my 130XE. But (c) it is not working in my 800XL, while any other OS rom, original or home made, works fine. That's a mystery I will probably never solve, luckily I don't need to.

[+x=usr(1536)]

2 hours ago, reifsnyderb said:

Another thought regarding the sockets.  With the chip removed, use a meter and check every pin socket to make sure there is no low resistance connections between pins that should not be connected.  Also, make sure each pin will have a good connection from the socket to the board.

Both of these have been done as part of past troubleshooting, but I'll re-do it as part of pulling the socket again.

2 hours ago, _The Doctor__ said:

After the empty sockets tests, consider flipping the board over and making sure nothing get picked up or bent on the bottom side of the PCB. I've had boards come in with legs bent or touching, as well ass solder / work debri or flakes hiding in plain but well disguised sight.

This has been done after each round of work and has always come up clean.  However, it will get re-checked when the socket is pulled again.

2 hours ago, _The Doctor__ said:

I sure hope you are taking anti static precautions for you and your iron, placing a pcb on a slightly conductive anti static bag is not adequate and can prove bad news if powered up with a legs / pins etc touching it. If there is a differential between the iron tip and said circuits that's a bad thing also. Delay lines aren't the only things susceptible to esd.

Anti-static mat, working on a tile floor with a laminate countertop, wrist grounded to a pipe.  In other words, in the kitchen at one of the counters.  Space is tight here right now.

2 hours ago, _The Doctor__ said:

Last hail Mary things are resistivity in switch contacts etc.

I have cleaned the power switch with contact cleaner, but that doesn't mean it would be a bad idea to do it again.

2 hours ago, _The Doctor__ said:

Flux can be clear and not seen, imagine looking at a beautifully cleaned, smooth tinned IC leg and finding it's not conductive, hence a mention of deoxit, high IPA, and fiber pens to clean such chip legs etc.

Good point.  To the best of my knowledge, this hasn't been an issue - at least when testing continuity from the IC's leg to wherever the trace it feeds through terminates is checked.  If there's a better way of doing that, I'm all ears.

 

FWIW, everything has been cleaned with 91% (or higher) IPA after any desoldering or soldering.

1 hour ago, woj said:

The other is about swapping chips around - one of my OS roms is clearly OK when (a) read with an eprom burner, (b) installed in my 130XE. But (c) it is not working in my 800XL, while any other OS rom, original or home made, works fine. That's a mystery I will probably never solve, luckily I don't need to.

I did have to change the OS ROM at one point as I discovered that the one that was in there had lost a leg.  I have absolutely no idea how that would have happened as I did not remove it as part of this up to the point where it was found, but it was replaced.  The replacement checks out in the 600XL.

1 hour ago, woj said:

I will add a bit to the uncertainties. I used to have tons of problems because of invisible flux, that's one. What I discovered is good to "visualize" it is to use acetone, it then matts. But surely working with acetone is totally nasty and I use it as last resort.

Agreed, especially indoors.

[+x=usr(1536)]

There has been significant progress; more so than in quite some time.  Before I get into that, however:

 

My unreserved apologies to everyone I've been short with over the past couple of days.  For background: the last two weeks of my life have been made hell from arthritis in my hands, fingers, and now wrists to the point where I am largely unable to sleep due to pain.  This doesn't help either my disposition or ability to relate to people.  As a result, the longer I've been awake and/or hurting for the less rational I may be about things that would normally just roll off of my back.  By no means is any of that an excuse for being difficult, and the apology remains offered without reservation.

 

Moving on: in a fit of wanting to make at least some progress, I sat down and took another look at the 800XL tonight.  The socket at U18 was removed again, and the area inspected for damage, bridged solder joints, etc.

 

What I found was what can only be described as a number of very fine metal slivers - not component legs, they were smaller in diameter than that - that had caused tiny bridges between a number of joints.  A trace had also lifted itself from the board at the pad, but it wasn't visible without removing the entire socket.  To everyone who said to go back and look again, I'm glad that you did: had I not heard it from other people I probably would have continued to take it as read that it couldn't be the problem given the number of times it had been gone over previously.

 

Work has stopped for tonight.  Tomorrow or this weekend I'll fix the trace, reinstall the socket, pop the 74F08 back in, and see what happens.

 

Thanks to everyone for the help - it really is appreciated, even if it didn't look like it at times