IBM 5150 Dead Motherboard - Memory issues

[Chris L]

Hi everyone,

I have an IBM PC 5150 I have had since new. It is a 64kb-256kb motherboard, also has a memory expansion board. It ran fine about 7 years ago, but recently I tried to boot it and it seemed like a dead motherboard, no boot at all. I went through troubleshooting and the power supply is OK and none of the components seem to be dragging it down, so the troubleshooter suggested a memory issue. I purchased a SuperSoft Diagnostic ROM to help diagnose memory issues. When I run it I seem to have three memory issues (tests that fail):

Critical 16K Memory test: Address 05040 bit 5

System Memory to 90000 test: Address 4C000 bits 7-6-2-0

Slow Refresh test: Address 48000 bits 7-6-2-0

Confused about where I go from here. It seems the 16K Critical memory should be in bank 0, and the SuperSoft manual indicated first two numbers indicate bank but if that's the case what is "05" showing in the 16K memory fail. I have also read some of the notes about Supersoft sometimes giving some flakey results, but maybe I am just not understanding addresses?

Since the first (0) bank of memory is soldered, I am wondering if I should pull these and solder in sockets?

Any suggestions, clarification much appreciated. Thanks!

I have attached a photo of the screen just before 7 passes were completed.

Chris

 

[modem7]

Hi everyone,

Welcome to these forums.

I have also read some of the notes about Supersoft sometimes giving some flakey results, but maybe I am just not understanding addresses?

It is a fact that the RAM tests of the Supersoft/Landmark diagnostics can sometimes be wrong or misleading.

Critical 16K Memory test: Address 05040 bit 5

If you were to search these forums for posts where people saw the 05040 address when using the Supersoft/Landmark diagnostics, you would find many posts.
The address of 05040 is wrong and needs to be ignored.
If 05040 appears, it often appears on the first pass only.

System Memory to 90000 test: Address 4C000 bits 7-6-2-0

Address 90000 hex = 589,824 dec = 576 K

If your 5150 has less than 576 KB total of conventional memory fitted, then errors are expected when non-existent RAM is tested. In that situation, many bits are usually shown as faulty. (I do not remember ever seeing all bits in error.)

Slow Refresh test:

No point in paying any attention to the slow refresh test (a kind of stress test) if the earlier RAM tests fail.

Since the first (0) bank of memory is soldered, I am wondering if I should pull these and solder in sockets?

The 5150 motherboard can be easily damaged via soldering. It is not like modern PCB's. And so I recommend that you do not socket all of bank 0.

I think there is high probability that bit 5 in bank 0 is faulty.

Because you had to purchase a Supersoft/Landmark ROM and adapter, I know that you cannot create your own ROM's. If you could, I would have suggested that you try Ruud's diagnostic ROM to see what its test of RAM reveals.

Something you can try is to temporarily piggyback a known-good RAM chip on the bit 5 chip in bank 0. If the RAM test then passes, bit 5 failure in bank 0 is confirmed. See [here] for more information. This is a hit-or-miss thing; if the RAM test still fails, the bit 5 chip is still a suspect.

Sometime today (I need to do something before the approaching rain hits), I will pull out a 64KB-256KB 5150 motherboard, simulate a failure of bit 5 in bank 0, and then see what the Supersoft/Landmark diagnostics show.

 

[Chris L]

Thanks modem for the welcome, and the valuable info. I don't know anything about creating ROM's but sounds interesting.

Thank again,
Chris

 

[maxtherabbit]

The 5150 motherboard can be easily damaged via soldering. It is not like modern PCB's.

Ok you've piqued my interest - what exactly is different about it? Is it any more fragile than any of the other dozens of 1980s PCBs I've worked on?

 

[modem7]

The 5150 motherboard can be easily damaged via soldering. It is not like modern PCB's.

Ok you've piqued my interest - what exactly is different about it? Is it any more fragile than any of the other dozens of 1980s PCBs I've worked on?

My statement was general and aimed at the OP.
Perhaps my statement should have been something like, "The 5150 motherboard can be easily damaged via soldering by someone with little soldering experience."

 

[modem7]

Sometime today (I need to do something before the approaching rain hits), I will pull out a 64KB-256KB 5150 motherboard, simulate a failure of bit 5 in bank 0, and then see what the Supersoft/Landmark diagnostics show.

So I brought out a 64KB-256KB 5150 motherboard. I plugged in an IBM CGA card. No other cards.
I ran the Supersoft/Landmark diagnostics. As expected, the RAM checks passed.

I then simulated a failure of the bit 5 chip in RAM bank 0.
The Supersoft/Landmark diagnostics then reported:


TEST: 16K CRITICAL MEMORY REGION
------------------------------------

Soon after this test started, the bottom-right pane showed bit 5 failure at address 05040.
It stayed displaying that for the remainder of the test.

At the end of the test, the relay clicked once.

( Same result for subsequent iterations of this test. )


TEST: SYSTEM MEMORY AT ADDRESS xxxxx
-------------------------------------------

Soon after this test started, the bottom-right pane showed failure of bit 5 only.

When the test address advanced past bank 0 (i.e. address 10000), bit 5 continued to show in error, but that is probably because the Supersoft/Landmark diagnostics decided to keep a record of what had failed earlier.

When the test address advanced to address 40000 (256K), the bottom-right pane then showed failure of multiple bits. Expected, because motherboard has 256K fitted (i.e. non-existent RAM being tested).

At the end of the test, the relay clicked three times.


TEST: SLOW REFRESH TO xxxxx
-------------------------------

Same behaviour as the SYSTEM MEMORY AT ADDRESS test.

 

[Chris L]

Thanks, great info. Much appreciated and thanks for the prompt effort!

Chris

 

[Chris L]

Next Step on Memory Issue

Next Step on Memory Issue

So I piggybacked a good chip on the bank 0, bit 5 chip and ran the test again and had the same results as initially.

As I understand it this is not unexpected, and it sounds like I should move forward on this chip replacement. Kind of busy that last couple weeks, but will do the chip replacement and see what things look like then.

Thanks for the assistance, will update after the re-check.

Chris