1-3-1 beep is memory code for many IBM and phoenix bios. 0807 means it passed 07 and either hung on 08 or halted on next code 09 (likely base 64k test)...
Is a 6381 SX board based on jumpers...
Are you using original IBM ram ? Or did you add some ?

I got the computer as not working. When starting it up the HDD made a horrible noise. So my initial assumption was that this was the original point of failure (i.e. the reason why it was discarted.) and everything else was just left unchanged. The RAM is probably not original IBM (It is labeled with a SAMSUNG sticker but it also has a FRU number suggesting its IBM "approved" ?)

but I would assume it’s the RAM the machine came with.

Regarding the jumpers – there are some jumper options and description in the manual (attached) but I cant think of anything that is obviously wrong ?!

I just saw that jumper J15 - it allows "write protect enabled / disabled " Im not sure what that does.

Ok – done some more testing but no luck. I put the RAM in another computer and it worked without problems. I tried yet another set of RAM in the IBM board – no luck. The CPU and chipset etc. get slightly warm and I also tested another known good power supply – no luck. I feel stuck.

I do have a working DX33 CPU. Would it be worth putting this into the upgrade socket (in case the SX33 CPU is faulty) Would that even work ? The socket sems to have an additional pin – or do I need a 487 to test that ? (I’m scared to damage the DX33)

any help is much appreciated

Aui wrote on 2025-05-09, 03:10:

Would that even work ?

AFAIK you need a 487SX or an ODP, however I am none too sure. Out of interest what is the S-Spec of the SX? I think it starts SX68, but can't make out the last digit.

Here is a better image of the CPU

Aui wrote on 2025-05-12, 04:50:

Here is a better image of the CPU

Thank you, it's a SX684, AFAIK it's CPUID is 0428, but it does not support the CPUID instruction which seems strange when CPUID 0423 CPUs do. If you ever get Windows running on it then it would be interesting to know what my SIV utility reports.

I have another question - Jumper J11 is described as flash memory - enabled / disabled - (see atached pdf). Does anyone know what flash memory means in the context of an IBM 486? Its obviously way before the term became commonplace for modern storage ?

thanks

Aui wrote on 2025-05-12, 06:15:

Does anyone know what flash memory means in the context of an IBM 486?

I suspect the BIOS is in flash memory and it's enable/disable BIOS updates.

If you ever get Windows running on it ...

Ok - lets hope that this will happen one day. Now I need some suggestions as to how to proceed.
As noted before the board looks very good - the RAM sticks are ok - What is wrong here..???

When I get post issues with my systems I plug in my PCI Post Card so I can see how far the post gets, so were it my system I would get a LPT Post Card so I could do the same and take things from there.

There are lots of PCI/PCIe Post Cards on eBay, but very few LPT ones, maybe there will be one close to you.

Aui wrote on 2025-05-12, 07:26:

As noted before the board looks very good - the RAM sticks are ok - What is wrong here..???

It's an IBM, that's what. Most PC are clones and these don't care about the SIMM configuration jumpers/resistors, but this IBM just might. Make a close-up photo of your stick config please.
Also on 486 system one 72-pin stick should be enough to run, as these are 32-bit. So try one in all the slots, then the other one in case the first turns out to be bad. Look for some cheap IBM branded or approved stick on ebay maybe, preferably 4M size as these work on pretty much anything. Different size sticks are a lottery with early 486 mobos.

EDIT: Additional note, this is a SiS chipset and the sticks you have have 9 chips, so it's a parity (aka 36-bit) RAM. You want parity SIMMs for SiS, just in case it's very picky about that.

Ok - tried a few more things:

- microscope inspctions of chips / traces / components - (everything looks like new)
- tried single RAM sticks in all 4 slots - nothing (always get - 1 long - 3 short - one long beep)
- tried additional different parity and non parity RAM chips -nothing
- the original IBM sticks I curently see for sale are huge and too expensive (I will keep looking)
- I ad a few better quality images of the RAM area and the board

Aui wrote on 2025-05-14, 01:16:

- tried single RAM sticks in all 4 slots - nothing (always get - 1 long - 3 short - one long beep)
- tried additional different parity and non parity RAM chips -nothing
- I ad a few better quality images of the RAM area and the board

The DRAM chips on your sticks are Samsung's KM44C1000 which is 1Mx4b, so each stick is 4MiB in total. Assuming the other side of the stick is not populated (I think it's not based on previous photo). The KM44C1003 is also 1Mx4b except quad CAS for parity. These should work fine for the chipset, but then there are the config/presence detect resistors (sometimes solder jumpers). On your sticks it's those 3 marked 000 (0 ohms) and one empty space on the right side of the stick.

Quite a few original IBM machines require these to be set correctly for what the mobo requires. Which BTW also limits what you can install (both capacity and speed-wise), although it's possible to cheat and just resolder these resistors to different locations. I'm familiar with other machines that require such procedure but I've never had original IBM. I just know it's a thing. Consult this page: https://www.ardent-tool.com/memory/Identifica … on.html#SIMM_PD

Note that the resistors/jumpers might be in different location or different order on other sticks, so you always need to either trace the connections or better yet check with a meter, as the page suggest. So I'm not going to guess what config you have set now but 4MiB/70ns is 3 out of 4 closed, except in a specific order which might not match what you have. And again you first need to figure out what the mobo needs (do you have a manual?), perhaps it doesn't actually care about these pins but somehow doesn't like these sticks, or is otherwise broken.

Yes, I do have a manual, according to which I may not have the original RAMs (although, I think these were the RAM sticks the PC was originally delivered with)
I have compiled some information related to memory from the document (full document also atached)

Just is general - related to the system board - which components could fail before the memory ? For example, if the CPU is the cause of the problem, would I still get the memory error ?
(I am still thinking about getting a 487 FPU/CPU and trying that - is this a bad idea ? What about recapping - could that help ?

thanks for the help

Aui wrote on 2025-05-15, 01:37:

Just is general - related to the system board - which components could fail before the memory ? For example, if the CPU is the cause of the problem, would I still get the memory error ?
(I am still thinking about getting a 487 FPU/CPU and trying that - is this a bad idea ? What about recapping - could that help ?

To get the mobo to beep any codes, and always the same ones in timely fashion rather than random noise, the following have to work:
1) Clock generators (CPU/FSB at the very least, RTC is another story)
2) CPU itself (except maybe some higer order address lines, but A31 must work)
3) BIOS ROM chip (it usually has a checksum it'll test, and that's a different error)
4) mobo chipset (though possibly not fully)
5) a lot of helper chips like bus amplifiers/muxes

So the BIOS ROM is 8-bit chip on these mobos but the way 486 CPU works all its data lines must be working, and be connected to mobo chipset, to fetch the code properly. Address lines (some of them) and the connection between RAMs and chipset is a separate thing, but there isn't much on your mobo and it looks clean and nice. Doesn't mean there can't be a micro-cracked via or a trace but that's not what I would expect. I don't think getting a rather rare 487 chip, or maybe DX2 ODP, will help you out (but if you want one, why not). Recapping shouldn't really affect anything either. These caps do dry out but rarely die completly, and I would expect the mobo to sometimes start or in general behave more randomly with dry caps, not always stopping on the same error. And you don't have L2 cache here so that's one less thing to worry about (if costing you some RAM performance).

Since you mentioned recapping I assume you do have a soldering kit and know how to use it? Why not change these 0 ohm resistors on one of the sticks? That's what I would do, way cheaper then getting an approved IBM RAM apparently. Looking at the page it might just need one of these moved to the now empty spot. Again, do check it with a meter on the stick pins which resistor is which.

BTW did you check the mobo without any RAM? Does it beep in the same way?