5150 Mainboard Troubleshooting

[IBM]

I have been troubleshooting a “64-256k” 5150 mainboard and I just cannot figure out what’s wrong with it. I’ve never had a 5150 that I couldn’t fix, but I think this is going to be the first one. There must be something wrong with one of the non-removable components on the board because every other attempt to replace a part with a known working one has given no results.

None of the components have any obvious external signs of problems either.

When the power is turned on, the board simply does nothing. No post, no beep codes, no blinking cursor on-screen - nothing. The processor gets hot after awhile so things are getting power. A way someone could directly emulate what is happening would be to pull your BIOS ROM chip from your motherboard – this would give behavior identical to the problem I am having here. At least if there were beep-codes I would know there was something I could do, but there isn’t.

I have done/tried the following:

-Have the minimal 64k RAM configured and installed (the soldered in bank).
-Have no add-on cards or other drives hooked up.
-Have all DIP switches set right.
-Switched BIOS with known-good one.
-Switched the processor (regular AMD 8088) with a known-good one.
-Switched power-supply with known-good one.
-Switched PC-Speaker with known-good one.
-Have tried known-good MDA, CGA and even no video cards.
-Tested capacitors and resistors with a multimeter as well as possible.
-Very carefully examined all components on the board to try to find corrosion, damage, broken connections, shorts, etc.

Yet I have made no progress. In addition to the processor, if I install a coprocessor it will also get warm (hot) so I am assuming there is no problem relating to the power-good signal.

If anyone has any idea what it could be or what else I could try, I’d like to hear it. This computer has sat for 15+ years without use but worked when it was put away, so it really could be anything.

I just about out of ideas here. I'd personally be very grateful for some feedback.

Thank you,

-Isaac

 

[Lorne]

I don't how early RAM comes into the start up, but have you tried piggy-backing RAM on that first bank, in case it's one of those soldered in RAM ICs that's bad?

 

[GottaLottaStuff]

The DIP switches can go bad. If they sit in one position for years, oxidation builds up on the contacts, and then when you switch them, they don't work. Back 20 years ago we would drip a good contact cleaner into them and switch them back and forth a few times. Just make sure it's contact cleaner from an electronics dealer, the automotive kind will dissolve plastics.

 

[nestor]

My 5150 had the same symptoms and the problem was one of the soldered RAM ICs in the first bank, it was dead.

 

[reenigne]

I think if it was the RAM or the DIP switches you'd still get a beep code. My guess is the timer chip or some of the bus logic, though I think you'd really need an oscilloscope to see what's going on. Or a makeshift oscilloscope.

 

[RJBJR]

Modem7 has a web site that has great info for troubleshooting the 5150. You might want to read this page http://www.minuszerodegrees.net/5150/5150_ram_64_256.htm toward the bottom has your symptoms. Also might want to follow the Supersoft/Landmark link in that page.



Edit: The suggestion from Lorne in reply #2 may be a viable way to isolate the chip but I have never tried piggybacking ram.

 

[tezza]

Edit: The suggestion from Lorne in reply #2 may be a viable way to isolate the chip but I have never tried piggybacking ram.

It's certainly worth trying. Bear in mind that it's not a definitive test. If there is no improvement it still could be that chip. Whether the test succeeds or not depends on exactly what's wrong with the IC. However, if things fire back to life, then you've diagnosed the problem.

Just make sure you have oriented the piggybacked chip correctly.

Tez

 

[reenigne]

Now that I look at the BIOS source code I see that indeed there isn't a beep code for a failure in the first 16K, but it does output diagnostic information to port 60 for that failure, so if you have (or can obtain or build) a port 60 diagnostic card that would tell you if it is the RAM and (if so) which chip is failing.

 

[modem7]

It's certainly worth trying. Bear in mind that it's not a definitive test. If there is no improvement it still could be that chip. Whether the test succeeds or not depends on exactly what's wrong with the IC. However, if things fire back to life, then you've diagnosed the problem.

Just make sure you have oriented the piggybacked chip correctly.

As well, the technique won't work if multiple RAM chips in the bank have failed. A couple of years ago, I repaired a 'dead' 5150 motherboard that had been in storage for many years (definately more than 10 years). There were 3 faulty RAM chips in the soldered-in bank.

 

[tezza]

As well, the technique won't work if multiple RAM chips in the bank have failed. A couple of years ago, I repaired a 'dead' 5150 motherboard that had been in storage for many years (definately more than 10 years). There were 3 faulty RAM chips in the soldered-in bank.

Yep, I guess it can happen. I experienced this on an Apple II+ plus board, where a handful of RAM had gone. Obviously the RAM chips didn't like mice urine. (-:

Tez

 

[Trixter]

Just make sure you have oriented the piggybacked chip correctly.

I hate to ask the obvious, but a 5-minute google search didn't turn up exactly the info I wanted: Is "piggybacking" a DRAM as simple as it sounds? Just fit each leg over the soldered-in chip beneath it? Or is something else necessary, like cutting a pin or avoiding contact with another pin, etc.?

 

[Druid6900]

I hate to ask the obvious, but a 5-minute google search didn't turn up exactly the info I wanted: Is "piggybacking" a DRAM as simple as it sounds? Just fit each leg over the soldered-in chip beneath it? Or is something else necessary, like cutting a pin or avoiding contact with another pin, etc.?

Yup, that's about it. I first learned the technique from one of my techs when I was running my first company.

You can piggy-back an entire row (in case there is more than one bad chip) and if it works with them all piggy-backed, then you just shut it down, take one off and see what you get on power-up. If it still works, then that chip is good, if it doesn't, that chip is bad, piggy-back it again and you just keep going in that manner.

At the end, whatever chips are still piggy-backed are the bad ones.

 

[modem7]

You can piggy-back an entire row (in case there is more than one bad chip) and if it works with them all piggy-backed, then you just shut it down, take one off and see what you get on power-up. If it still works, then that chip is good, if it doesn't, that chip is bad, piggy-back it again and you just keep going in that manner.

At the end, whatever chips are still piggy-backed are the bad ones.

Whether one chip or multiple chips are used, just make sure that known good ones are used.
Obviously, the use of chips from other banks is risky (they could be bad too).

 

[MikeS]

Whether one chip or multiple chips are used, just make sure that known good ones are used.
Obviously, the use of chips from other banks is risky (they could be bad too).

... And of course it's not definitive; if it works with piggy-backed RAM then it most likely is one or more bad RAM chips, but if it still doesn't work then it could still be a bad RAM chip (or anything else of course). And be sure to check the RAM voltages first so you don't burn out the ones you're piggybacking with.

 

[IBM]

Thanks for all the replies. Sorry I didn’t respond sooner – been very sick and sleeping a good deal. I am surprised at what I read.

Really, I thought that any RAM failure in the first bank would be reported with "00xx 201" error codes. Not having this would make troubleshooting and fixing the first bank even more difficult, as the parts are all soldered on as is. I did try the piggybacking on other boards in the past with great success, but I hadn’t tried that here under the assumption that there would be 201 codes if there was a problem. So I tried it and still nothing. You know, Tezza’s page was where I looked once before http://www.classic-computers.org.nz/blog/2009-03-18-ibm-pc-ram-repair.htm to learn about dealing with RAM issues. Good stuff on there – I salute you!

I was reading “Hardware debugging with no oscilloscope” – very interesting. I’ll have to look into that when I’m felling better (I don’t have an oscilloscope, sadly). As is, I’m still stumped.

I enjoy reading Modem7s page – I don’t think he has ever posted anything that isn’t useful.  I’ve been wanting to burn that diagnostic ROM and try it for some time – have to wait to get my new programmer all the way from China so who knows how long that will be. 

I’ve never had RAM fail in the first bank – knock on wood. I do know how to solder (been silver soldering since age 10) but desoldering is always a challenge for me.

One thing I have always wished someone would make is a diagram of the 5150 motherboard that shows every single chip and resistor,etc (a list too, maybe). All I have ever seen was the official IBM diagram, or a colorized revision of it, that shows the more important components but not everything (no, not a circuit diagram – I’m sure those do exist). For example, what is the empty space labeled U100 for? No socket there, just a space and solder spots, but yet I’ve never seen a board with a chip there either – IBM left a space for it so it must have been for something. I have also always wanted to put something useful in U28 (the empty ROM socket) but I’m still trying to figure out what I could put in there.

Well hey, thanks for all the feedback! I’ll keep working at it.

Cheers,

-Isaac

 

[tezza]

I hate to ask the obvious, but a 5-minute google search didn't turn up exactly the info I wanted: Is "piggybacking" a DRAM as simple as it sounds? Just fit each leg over the soldered-in chip beneath it? Or is something else necessary, like cutting a pin or avoiding contact with another pin, etc.?

Trixter there is a picture of a piggy back in action in this article (figure 3). It's exactly what it says. just slip another RAM chip over the suspected one making sure all the legs are gripping on tighly by friction. As Druid says, if you have spare chips, you can do more than one.

Tez

 

[modem7]

For example, what is the empty space labeled U100 for? No socket there, just a space and solder spots, but yet I’ve never seen a board with a chip there either – IBM left a space for it so it must have been for something.

In case an extra chip is needed down the track. One example. Let's say that before release, or even after release, a flaw is found in the motherboard design. Let's suppose that the flaw can be fixed by delaying a particular signal slightly, and that the amount of delay caused by passing the signal through two inverters is enough of a delay. If there aren't two unused inverters on the motherboard, then IBM could solder in a 7404 chip (contains six inverters) into the space at U100 and then do the associated wiring/rewiring.

 

[IBM]

Modem7: never short on useful information. Guess that makes sense - I see all these small things and wonder sometimes. For years I never knew what the "trimmer" for the color signal did until I was reading a post about CGA and got the answer to that.

What is the small, round, black plastic pop-out cap on the back of the 5150 case for? I always assumed it was to let a cord for something (lightpen?) thru the case.

On another note, I did piggyback all nine chips in the first bank but still no results. I think I will just desolder the whole bank and solder in sockets and see if I can get anything to work that way.

We shall see if I don't mess up the whole board trying!

-Isaac

 

[modem7]

Modem7: never short on useful information.

Well, there are quite a few of us here who could have answered that question. I got in first.

What is the small, round, black plastic pop-out cap on the back of the 5150 case for? I always assumed it was to let a cord for something (lightpen?) thru the case.

I don't know, but yes, cable pass-through seems the most likely, for the reason that you wrote of - light pen cable to the light pen Berg connector on the IBM CGA card. And one of the purposes that IBM intended for the second CGA Berg connector was "driving a user-supplied RF modulator". Cables for custom or third-party cards.

 

[Chromedome45]

I kind of had the same problem with a PCjr I got from a freind. Was still sealed in the bag and never opened. I opened it up and powered it on & just like your 5150 nothing. I tried swapping all socketed chips. No go. even removed and replaced all the on board RAM. Still no go. Well to keep this short turned out to be a bad ROM chip. I was floored! Brand new out of the box in sealed bad and had a bad ROM. Who would have guessed. Maybe you could try swapping ROM's if you have anothher 5150 or burning your own once you get a programmer. I hope this helps in some small way.