It should connect to the regulator output, against a nearby ground. But if that did nothing there's something actually wrong somewhere but diagnosing what it may be can be tricky.

Only thing I can think of is to use a fat power resistor with few ohm value on the CPU power rail and then find out what heats up somewhere on the board, that is where the issue is going to be...

kagura1050 wrote on 2024-11-14, 03:05:

I looked at the photos of both sides and traced the wiring. I couldn't find anything suspicious.

Thanks for checking. 😀

By the way, what I find very strange is that even though you have shorted the 5V line and 3.3V line many times, nothing has caught fire/smoke (not even the wiring or the regulator). . I think this indicates that there is a very strong short circuit or that the regulator is not working for some reason (GND is floating, etc.).

I did keep the power on tests very brief, usually just a second or two. I was trying to avoid anything heating up and potentially exploding. 😅

Huh, so I was looking at the datasheet for your UMC L2 cache chips (attached below) and it seems they use a 5V power supply but they can interface with a 3.3V bus. So it seems a different design when comparing with my Soyo 029C2 which uses Alliance AS7C3256 SRAM chips that use a 3.3V supply.

Good news is that there probably isn't anything on your board other than the CPU that would be damaged by 5V on Vcc. As I mentioned above, some 486 motherboards that use voltage autodetection (via VOLDET pin) have a circuit with a 7407 and a MOSFET that shorts the regulator input and output when VOLDET is not grounded, resulting in a 5V Vcc to the CPU. I don't remember exact details about how the circuit was implemented, but in any case I think there would be no point having such a thing here.

Just mentioning this to note that bypassing this type of regulator by shorting input and output does not necessarily lead to disaster, it has been used on some motherboards even though it looks like it could be a waste of power.

Tiido wrote on 2024-11-14, 03:06:

Only thing I can think of is to use a fat power resistor with few ohm value on the CPU power rail and then find out what heats up somewhere on the board, that is where the issue is going to be...

I think so too.

Shponglefan wrote on 2024-11-14, 00:04:

In testing those, they do appear shorted. It's not a dead short, as I get a 4 ohm reading.

5V / 4ohm = 1.25A, 5V * 1.25A = 6W...
If 4 ohms is true, even if you don't have a cement resistor or anything like that on hand, you might be able to test it by simply shorting Vout to GND. (However, please keep the test under a few seconds)

I did a bunch more probing to see if I could find any components connected to both 5V and 3.3V. So far the only components I have found are the main three chipset ICs. Per the datasheets, this is by design for the 430NX chipset.

My current working theory is that the 5V and 3.3V is potentially bridged via one of these chips. To determine this for sure, I would remove each chip one at a time to see if the bridge disappears. Before I do that, I plan to systematically recheck everything else on the board just to make sure I didn't miss anything.

I also have another board that uses the Neptune chipset (Intel Premiere/PCI II). It's currently being used in a build, but I might disassemble that PC just to test that board to see if anything is either the same or different.