5150 16K Ram Failure

[cowrack]

I recently picked up an IBM 5150 64k-256k version that was in some seemingly rough shape. It had a no power condition. No blown tantalum caps, but would not power on. I installed the diagnostics ROM, powered it up and it started but it failed the 16K ram test. I have since replaced all of the RAM chips and the parity ram chip and still get the error. It indicates memory error at address 040000 and failing bits 1-7 (bit 0 passes). I have tested the 4164's with a ram tester and they all pass. I believe it was the same errors before I desoldered and replaced those chips. I don't have other memory populated in the unit (actually removed most of the sockets to replace them, fun...). This leads me to believe some RAM support circuitry is the issue. I have found lots of issues similar to this on here, but most of them turn out as bad ram. It might be there is just too much damage on this unit, but I think it is worth checking. if anyone has a path to go down to troubleshoot, it would be appreciated. My skills are at a novice level, but I can figure out most things, just need some assistance in getting started (where to start). I am not using the factory power supply, using a later power supply and there seems to be no issues on the power rails (correct voltages).

thank you

Jamie

 

[tradde]

The address 0x40000 seems suspect as that is 262,144 in decimal. Maybe it thinks it has another bank of memory? That's one byte beyond 256k.

 

[modem7]

Hello Jamie. Welcome to these forums.

I installed the diagnostics ROM,

One has to be careful about the behaviour and result interpretation of the SuperSoft diagnostic ROM, particularly in regard to RAM related problems.
For example, I do not like the fact that diagnostic does not stop at the first RAM error.

It indicates memory error at address 040000 and failing bits 1-7 (bit 0 passes).

You have the 64KB-256KB version of IBM 5150 motherboard.
Important information about using the the SuperSoft diagnostic ROM on that particular motherboard is at [here].
Reading that, irrespective of the motherboard switch settings, RAM errors at address 40000 and beyond are expected if the 64KB-256KB motherboard only has 256KB total fitted (total: motherboard + cards).

... but it failed the 16K ram test ....

Failure of the 16K CRITICAL MEMORY REGION test.
Had you the earlier version of the motherboard, one would not trust that reported error, but you have the 64KB-256KB version of IBM 5150 motherboard, so yes, most likely there is problem writing to and reading from bank 0.

... I have since replaced all of the RAM chips and the parity ram chip and still get the error.

How confident are you of your soldering?

This leads me to believe some RAM support circuitry is the issue.

If you are tempted to include RAM refresh circuitry in that, then a read of [this] may lead you to exclude it.

Out of curiosity:

1. At the instant that FAILED is reported for the 16K CRITICAL MEMORY REGION test, what is shown in DIAGNOSTICS SUMMARY window (address, failing bits) ?

2. What does Ruud's diagnostic ROM report? Expected is what is shown at [here].

 

[cowrack]

My apologies on not responding sooner, been a very busy weekend! Thank you for welcoming me, I really enjoy these machines are hope to get this one going. I have a soft spot for machines that are probably well beyond repair...I wish I could say this is the only one, but there are more (not PC's though ) I should have mentioned a few things. First, the only populated bank on this board is bank 0. As a matter of fact, I removed all of the other sockets (someone had pilfered the ram already) as most had some form of corrosion on them. Removing all of those sockets was not fun. I have also replaced the parity ram chips. I need to finish cleaning the board and resolder the sockets (and repopulate with ram). Generally I feel like I do a decent job soldering. I would be more concerned if this was surface mount, that I wouldn't trust me with. I am attaching a picture of the diagnostic screen. Once this error sets, I don't get anything else different (it never changes) although I am going to try it again to see if maybe something flashes fast. (I snapped this photo after it ran a few times). The board does have a lot of corrosion on it and I have been working to get it cleaned off. This is one of the reason I wanted to consider it could be something other than the ram (although I will work on checking common lines on the ram chips to verify, its always possible as mentioned there is a bad solder joint or some other issue. When I mentioned support circuits, I was also thinking any chips which support the ram itself, various 74L chips, etc. Basically, anything along the bus.

Thanks
jamie

​