PSU question on early Model A 5150

[pkhoury]

Is it a good or bad idea to replace caps on an otherwise working PSU for an early Model A 5150? By early, my serial is 0100485. I think I'll finally have a chance this summer to replace bank 0 on the motherboard with DIP sockets, as well as bad tantalums, and was wondering if it's a good idea replacing stuff in the PSU as well.

Or, should I just make sure it's outputting the correct voltage? With a few exceptions, I want to keep this system as original as possible, including the first revision BIOS it has.

 

[Malc]

Generally speaking the rule is " If it ain't broke don't fix it ", Why replace bank 0 with sockets and bad tant's, Have you got any bad ram or tant's ?, If not leave them alone.

 

[mR_Slug]

Have to agree, it's best to leave it alone. If you mod it, you can never go back.

 

[SomeGuy]

You really don't want to risk damaging the motherboard by socking RAM unless the RAM is faulty. You would then have to install non-original RAM chips. Really no need to mess with tantalums on the motherboard either unless they blow up. I don't think I would risk touching anything, except perhaps RFI capacitors, unless there was a known problem somewhere.

 

[VERAULT]

yes replace the tantalums.. no need to socket ram until it breaks.

 

[pkhoury]

Have to agree, it's best to leave it alone. If you mod it, you can never go back.

I've thought about that, believe me. I just can't see myself soldering in new 4116's (well NOS anyways), given they're not the most reliable memory out there.
Now the CPU is NOT socketed, and I'm keeping that the way it is.

Generally speaking the rule is " If it ain't broke don't fix it ", Why replace bank 0 with sockets and bad tant's, Have you got any bad ram or tant's ?, If not leave them alone.

I actually do have bad tantalum caps. I know that more than one were bad, and I think we just snipped a lead on some. Also ran the Landmark BIOS, which confirms the board works, but Bank 0 is apparently bad.

If anything, I'll replace the tantalums first, and then if I need to, out comes Bank 0. I'll probably try it out first on a somewhat newer Model A motherboard.

 

[modem7]

Also ran the Landmark BIOS, which confirms the board works, but Bank 0 is apparently bad.

Your early 5150 will have the 16KB-64KB version of IBM 5150 motherboard.
Are you aware of the first bug shown at [here] ?

 

[pkhoury]

Your early 5150 will have the 16KB-64KB version of IBM 5150 motherboard.
Are you aware of the first bug shown at [here] ?

Interesting. I read that before and forgot about those bugs. That being said, would the logical thing to do at this point be to replace the memory bank by bank (except for Bank 0 for the time being), until it boots?

And yes, it does have the 16KB-64KB motherboard. I found some pictures I took, not long after I bought it from Computer Reset in 2019. This project to refurbish has been on the back burner for almost 3 years now.

I forgot the tantalums were blue. No special significance, since every other 5150 board I've seen has yellow tantalums? **edit** I just remembered the ones that went bad - they're right beneath each bank's parity chip. Are those still tantalums? Anything with replacing all of them, but leaving the blue tantalums intact for originality's sake? Lastly, for RAM - does speed matter? I ordered some NOS 4116, but they're 150ns. I would assume faster wouldn't be a deal breaker, since I think they're normally 200-300ns, right?

 

[modem7]

I forgot the tantalums were blue. No special significance, since every other 5150 board I've seen has yellow tantalums?

The colour has no special significance.

**edit** I just remembered the ones that went bad - they're right beneath each bank's parity chip. Are those still tantalums?

On a 16KB-64KB motherboard, two per bank, located between each bank's parity RAM chip and the word 'BANK'.



Tantalums, of 10μF/16V, appear to be what was fitted by IBM. (An early circuit diagram suggests that 8.2μF may have been used at some point.)
Looking at the second photo that you posted, you substituted all eight, but it is unclear what you substituted them with.
What did you use?

Anything with replacing all of them, but leaving the blue tantalums intact for originality's sake?

In that sentence, I think the "blue tantalums" that you are referring to are the blue capacitors in and about the RAM chips. Those are ceramic, and of much lower value.

Lastly, for RAM - does speed matter? I ordered some NOS 4116, but they're 150ns. I would assume faster wouldn't be a deal breaker, since I think they're normally 200-300ns, right?

Correct. Faster speed (maximum speed) is not a problem.

Interesting. I read that before and forgot about those bugs. That being said, would the logical thing to do at this point be to replace the memory bank by bank (except for Bank 0 for the time being), until it boots?

I would use the word 'efficient'.

If the only problem is in bank 1, or in bank 2, then I am expecting that when the IBM BIOS ROM is fitted, you would at least be seeing something on-screen. Whilst waiting for the new RAM chips to arrive, you could see what Ruud's diagnostic ROM thinks of the RAM - see [here], a web page that is specific to the 16KB-64KB motherboard.

But I wonder about the eight substituted capacitors that I wrote of earlier. If they were not correctly substituted, RAM operation could be compromised.

And because the Supersoft/Landmark diagnostic can be very misleading regarding RAM on a 16KB-64KB motherboard, maybe the problem is not RAM. What symptom is there when the IBM BIOS ROM is fitted? Speaker connected up?

 

[Andrettigto]

That's the earliest build 5150 I've seen. I do agree with how you feel about preserving the blue caps for originality. I have s/n109596 with a mix of the blue and yellow caps with later 16-64 boards having no blue at all.
Good to see you've moved it from the "back burner" as it's a great early system.

 

[pkhoury]

The colour has no special significance.


On a 16KB-64KB motherboard, two per bank, located between each bank's parity RAM chip and the word 'BANK'.

View attachment 1242742

Tantalums, of 10μF/16V, appear to be what was fitted by IBM. (An early circuit diagram suggests that 8.2μF may have been used at some point.)
Looking at the second photo that you posted, you substituted all eight, but it is unclear what you substituted them with.
What did you use?


In that sentence, I think the "blue tantalums" that you are referring to are the blue capacitors in and about the RAM chips. Those are ceramic, and of much lower value.


Correct. Faster speed (maximum speed) is not a problem.


I would use the word 'efficient'.

If the only problem is in bank 1, or in bank 2, then I am expecting that when the IBM BIOS ROM is fitted, you would at least be seeing something on-screen. Whilst waiting for the new RAM chips to arrive, you could see what Ruud's diagnostic ROM thinks of the RAM - see [here], a web page that is specific to the 16KB-64KB motherboard.

But I wonder about the eight substituted capacitors that I wrote of earlier. If they were not correctly substituted, RAM operation could be compromised.

And because the Supersoft/Landmark diagnostic can be very misleading regarding RAM on a 16KB-64KB motherboard, maybe the problem is not RAM. What symptom is there when the IBM BIOS ROM is fitted? Speaker connected up?

I actually haven't replaced anything on this motherboard. Everything as you see it is original to this board, AFAIK. The speaker is connected up, but nothing happens when the board is powered up. Initially, the fans wouldn't spin, until one of my friends in our retro computer club disconnected some of the tantalums.

 

[pkhoury]

That's the earliest build 5150 I've seen. I do agree with how you feel about preserving the blue caps for originality. I have s/n109596 with a mix of the blue and yellow caps with later 16-64 boards having no blue at all.
Good to see you've moved it from the "back burner" as it's a great early system.

My "practice" system I want to says is in the 109xxx range, but I have to double check. Yup, this system is super early. Maybe that's why it ended up at Computer Reset, but who knows. I want to ultimately show it off at one of our club GTGs, and if I need to run anything big, I'd just run interlink with something on another machine (since there's no ROM support in the first and second BIOS revision).

 

[modem7]

I actually do have bad tantalum caps. I know that more than one were bad, and I think we just snipped a lead on some. ...

I actually haven't replaced anything on this motherboard. Everything as you see it is original to this board, AFAIK. The speaker is connected up, but nothing happens when the board is powered up. Initially, the fans wouldn't spin, until one of my friends in our retro computer club disconnected some of the tantalums.

I got caught out by the fact that I have never seen rectangular shaped tantalums on a 5150 motherboard.

So, are any of those "disconnected" tantalums down near the RAM chips? If so, proper RAM operation is compromised.

 

[pkhoury]

I got caught out by the fact that I have never seen rectangular shaped tantalums on a 5150 motherboard.

So, are any of those "disconnected" tantalums down near the RAM chips? If so, proper RAM operation is compromised.

They are disconnected, because they were blown/shorted. So I suppose it won't hurt replacing those on the bottom? Despite the spirit of authenticity, I'd rather have reliable caps that'll be good another 40 years!

 

[modem7]

So I suppose it won't hurt replacing those on the bottom?

For those there that are disconnected, I don't think you have a choice.
They support the motherboard RAM, and being disconnected, that may be the reason for the Supersoft/Landmark diagnostic reporting a RAM problem.
And if any of the disconnected are against RAM bank 0, that might explain why the motherboard appears dead when the IBM BIOS ROM is fitted (i.e. POST halting on failure of first 16K).

16KB-64KB version of IBM 5150 motherboard uses two-legged tantalums.
64KB-256KB version of IBM 5150 motherboard uses three-legged tantalums.

The norm for a 16KB-64KB version of motherboard (i.e. the version you have) is shown in the photo at [here]. From the photo that you provided, things appear to be different - not using the outer holes.

BTW. Per post #9, the use of Ruud's diagnostic ROM is expected to show whether bank 0 is presently good or bad.

 

[pkhoury]

I would use the word 'efficient'.

If the only problem is in bank 1, or in bank 2, then I am expecting that when the IBM BIOS ROM is fitted, you would at least be seeing something on-screen. Whilst waiting for the new RAM chips to arrive, you could see what Ruud's diagnostic ROM thinks of the RAM - see [here], a web page that is specific to the 16KB-64KB motherboard.

But I wonder about the eight substituted capacitors that I wrote of earlier. If they were not correctly substituted, RAM operation could be compromised.

And because the Supersoft/Landmark diagnostic can be very misleading regarding RAM on a 16KB-64KB motherboard, maybe the problem is not RAM. What symptom is there when the IBM BIOS ROM is fitted? Speaker connected up?

Please don't chew me out, but when I tried the Supersoft/Landmark ROM (which was about late 2019/early 2020), I don't remember if it definitely said Bank 0 had a problem. There were other weird errors, like suggesting that the rest of my ROM (BASIC) was bad, so who knows, but the board did fire up and gave me video output. IIRC, I think I used a mono or Hercules card and a TTL monitor I amazingly happened to have here (now I have the TTL/CGA/EGA to VGA adapter, which is a bit more convenient).

One of the things that has bothered me is all the mixed brands of 4116. I might just opt to start over and replace it all with the NOS stuff, assuming of course that's all good.

When the IBM ROM is fitted, nothing happens. Initially, I think the fan spun briefly and stopped, indicating a short to ground, and that's when some of my friends gave some input, of which one of them is a computer engineer by trade (but we collect a lot of the same things).

So AFAIK, those are the original capacitors at the bottom. But I don't know the full history of this board/system, so I could be mistaken. After the short to ground was discovered, the board booted with the Landmark ROM, but again, not the IBM ROM. On a side note - do the IBM ROM chips ever go bad?

 

[pkhoury]

If the only problem is in bank 1, or in bank 2, then I am expecting that when the IBM BIOS ROM is fitted, you would at least be seeing something on-screen. Whilst waiting for the new RAM chips to arrive, you could see what Ruud's diagnostic ROM thinks of the RAM - see [here], a web page that is specific to the 16KB-64KB motherboard.

I didn't see the note about Ruud's diags until today (actually while composing my previous post). I don't suppose you have any suggestions for anyone in the states who can burn Ruud's diags onto an EPROM with adapter for me? I don't remember ATM where I put the Landmark diag ROM, nor do I know where my old school EPROM burner is (parallel port, only works up to Windows XP).

 

[modem7]

One of the things that has bothered me is all the mixed brands of 4116.

Note that in my various 51xx motherboards, some of the banks have different brands of chips within the bank. And it is never a consideration when I am swapping chips between banks as part of diagnosis.

I might just opt to start over and replace it all with the NOS stuff, assuming of course that's all good.

Bank 0 is soldered in, and unless you have good soldering skills, motherboard damage may result. So perhaps hold off on bank 0 until good evidence is seen of a bank 0 problem, or other simpler things have been eliminated first.

When the IBM ROM is fitted, nothing happens. Initially, I think the fan spun briefly and stopped, indicating a short to ground, and that's when some of my friends gave some input, of which one of them is a computer engineer by trade (but we collect a lot of the same things).

So there was an assumption that the tantalums near the parity RAM chips could be disconnected without compromising motherboard RAM operation. I'm not saying that is the cause of the present symptoms, I'm saying that it is a possibility.

So AFAIK, those are the original capacitors at the bottom.

Just like you, I have been a member of these forums for many many years. We are still discovering things about these old computers. Those rectangular shaped tantalums are all over your early 5150 motherboard. If the soldering indicates that they haven't been replaced, then we can deduce that on some very early 16KB-64KB motherboards, IBM fitted rectangular shaped tantalums.

After the short to ground was discovered, the board booted with the Landmark ROM, but again, not the IBM ROM. On a side note - do the IBM ROM chips ever go bad?

They certainly do fail.

Images to go into a replacement ROM are at [here]. Per that page, note that there are replacement options that do not require an adapter.

I didn't see the note about Ruud's diags until today (actually while composing my previous post). I don't suppose you have any suggestions for anyone in the states who can burn Ruud's diags onto an EPROM with adapter for me?

See [here] for people.
Again, note that there are replacement options that do not require an adapter.

GOING FORWARD

Your first course of action should be to replace-with-new the disconnected (faulty) tantalums, using the same holes, noting polarity. Then see if all now good.

Then you are trying things like:
* See what Ruud's diagnostic thinks of the RAM (i.e. is bank 0 good, something required for the IBM BIOS ROM).
* Try another IBM BIOS ROM.
* Note the information in the early entries at [here].

If unsuccessful, come back here with what information you have:
* Symptoms.
* What did you try.
* Per the web page I pointed to, what did Ruud's diagnostic think of bank 0 (preferably with screenshot)
* A screenshot of what the Supersoft/Landmark ROM is showing.

If successful, let us know what the problem was.

 

[pkhoury]

Bank 0 is soldered in, and unless you have good soldering skills, motherboard damage may result. So perhaps hold off on bank 0 until good evidence is seen of a bank 0 problem, or other simpler things have been eliminated first.

Apologies for not clarifying. Yes, I'll be replacing the faulty tantalums, but nothing else. When I said replacing with NOS, I was referring to Banks 1-3, but I should've been more explicit. I think for now, I'll replace the tantalums solely and see what I can do at getting an EEPROM. Or just bite the bullet and get a modern programmer and some compatible chips.

Images to go into a replacement ROM are at [here]. Per that page, note that there are replacement options that do not require an adapter.


See [here] for people.
Again, note that there are replacement options that do not require an adapter.

GOING FORWARD

Your first course of action should be to replace-with-new the disconnected (faulty) tantalums, using the same holes, noting polarity. Then see if all now good.

I'll keep y'all updated for sure!

On a segue - did any other early 5150 boards have non-socketed CPUs like mine? That was the first thing that caught my eye when I got this system (had the case, PSU but no top cover nor floppies).

 

[modem7]

On a segue - did any other early 5150 boards have non-socketed CPUs like mine?

Myself, I haven't seen enough of a sample size to comment.