• Please review our updated Terms and Rules here

Any ideas about what I could do with this intermittent memory issue on IBM PC 300GL?

RootHouston

Member
Joined
Sep 12, 2020
Messages
20
Location
Houston, Texas, USA
Just in case anyone has ever experienced this, I wanted to reach out and check here. I bought a 1999 IBM Personal Computer 300GL (Type 6275-90U). It's a Slot 1 Pentium III, Intel 440BX chipset, with a riser board for adding PCI and ISA cards and 3 PC100 DIMM memory sockets.

About 50% of the time, when I try to boot, it immediately gives either a 1-3-1 or 1-3-4 POST error beep codes, which are RAM-related as per the user manual. If I turn it off, then wait a few moments, I can usually get it to POST and boot just fine. However, sometimes, it will still tell me a lower capacity number when looking at installed RAM in the CMOS settings.
Here's what I've done to troubleshoot:
  1. Used only a single stick
  2. Swapped the stick into each DIMM slot
  3. Swapped with a known working stick and did the same
  4. Reseated the CPU
  5. Swapped the entire CPU with a known working one
  6. Reseated the power connector on the riser board
  7. Swapped the entire PSU with a known working one
  8. Reset CMOS settings to default
  9. Upgraded BIOS to what I believe was the last version
It's beginning to feel more like a motherboard issue. Since it's difficult to source a replacement (IBM FRU 61H2347), I'm guessing that I may have no real recourse to repair this machine. Any thoughts on this?
 
Just a suggestion, since you appear to have tried nearly everything. Sometimes on the mother board there are switching 3.3V sub-power supplies, for the CPU. I'm not sure about yours, or a number of power supply filter electrolytics. But if there are a number of electrolytic capacitors on the mobo, it would be worth checking their ESR with an ESR meter and if any are suspicious, replace them.

There can also be issues with intermittent connections if they are Gold plated. It depends though on how the surfaces make contact. If the contact surfaces slide across each other there is usually no problem and they "self clean", but if they are touching, like say a ZIF socket, they can sometimes still not connect properly, even after a contact cleaner wash, because a clear insulating surface film prevents it. I'm not sure what sort of socket your CPU uses. On one computer I had the only way I could solve it (the intermittent behavior on booting) was to partially close the CPU's ZIF socket, just a little, before the CPU was fully seated, to gain some surface friction on the contacts, and push the CPU home a few times so as to clean off the surface insulating film that was preventing a good connection. Similar problems crop up in switches with touching contacts (like DIP switches and those in Tek scope attenuator assemblies) which result in intermittent operation.
 
Thanks for your help. Unfortunately, I don't own an ESR meter, and by the looks of it, I might be better off buying another machine entirely for the prices of those. The CPU is actually using a Slot 1, and not a socket. There are some DIP switches on the board, but messing with those hasn't done anything for me thus far.
 
1999 was about the time the whole bad capacitor issue started. That went on until around 2003-4ish iirc. I don't know how badly IBM systems were affected but it may be something to consider as well.

The Tech/Service manual should be available from IBM, as is the one for my IBM PC-300GL mini tower from around the same time period.
 
Last edited:
Wow, ya know, I removed some components to get some better lighting, and I noticed some of these "Chhsi"-branded caps that appeared bad. One Google search, and it comes up as a known bad brand from back then. I will try to recap these, and report back. I'm a bad solderer, but will give it a go.

IMG_20220510_202728659.jpg
 
Those capacitors have dried out and the gas pressure inside the canister has domed the tops, they are definitely defective now and you don't need the ESR meter to tell in this case.If you did have the meters you would find they had a low uF value and high ESR. I would replace those with 105 deg C or better 125 deg C rated Nichicon types.
 
Those capacitors have dried out and the gas pressure inside the canister has domed the tops, they are definitely defective now and you don't need the ESR meter to tell in this case.If you did have the meters you would find they had a low uF value and high ESR. I would replace those with 105 deg C or better 125 deg C rated Nichicon types.

I got some of these Panasonics. They're rated for 105 degrees C. I'll let you know how it works out.
 
I got some of these Panasonics. They're rated for 105 degrees C. I'll let you know how it works out.

Those Panasonics look fine. Hopefully this will solve your issues.

If you want to get an ESR meter a good one is the Anatek blue (Designed by Bob Parker), they work better than sine wave based ESR meters, they use a pulse method:


A good entry level but very well made capacitance meter is the lovely little YF-150, and it is only $49 amazingly, I have been using one for over 20 years and it is still fine:


My advice is to get these two meters because they will pay for themselves hundreds of times over as the years pass by, especially if you are likely to have to make repairs in the future.

( I would avoid combination meters and cheap kit meter projects with unverified firmware. They never seem to be as good as an individual meter made for the specific task and some of them even barely fall into the range of what could rightly be called a scientific measuring instrument).

There is a lower cost version of the Bob Parker design which will be just as good, in many ways it is better because it has the original tables of ESR values on it (rather than a graph) which was Bob's original design, $68.95 is a very good price:



(these ESR meters are also very useful for measuring low values of resistance, for this reason their calibration is super easy to check with low value resistors. They have a system that zero's out the test lead resistance)

So for just around $120 in total (excluding shipping) you could get yourself two wonderful meters.
 
Last edited:
1999 was about the time the whole bad capacitor issue started. That went on until around 2003-4ish iirc. I don't know how badly IBM systems were affected but it may be something to consider as well.

The Tech/Service manual should be available from IBM, as is the one for my IBM PC-300GL mini tower from around the same time period.
Somehow I missed this thread in May.

I was told that the story behind these bad capacitors was industrial espionage. But in this case the American factory was aware of it and changed something in the formula. Because this Chinese firm had less costs, they were able to sell their capacitors cheaper and and a lot of firms bought their capacitors. The American firm warned these firms but to no avail.

I ran into this problem as well. When my company had to move in 2012 (?), I found an old storage room that was clearly forgotten (I still wonder how such a thing is possible) and it contained three of these 300GLs, brand new! Two worked fine, the third one booted fine but when going to the graphical mode of Windows 98, it crashed. This one had capacitors of another brand than the other two.
I gave it to the person who told me about the capacitors, he replaced them and it worked fine again.
 
We had 200 of these, labs full of them, and they all started dying, acting flaky, it was a disaster. We finally pulled them all because it was so bad and piled them up and replaced them with new machines. Months later IBM issued a bulletin for them and we replaced the mother boards on all of them, but by then they were getting long in the tooth - so the repaired ones were donated to local schools and charities. By the times we got around to doing the repairs, about 30% of them would fail POST and had visibly leaking / bulging caps. Ironically, when this all started, IBM spend a lot of time blaming everything and everyone on our end. We switched over to HP shortly after that fiasco.
 
Back
Top