Do you know what sort of heatsinc? Did it cover the part?
If you're lucky you might be able to find an image of that exact board and read type type of regulator.
(I googled/images the "P54CNPCI" and what images I saw didn't seem to be the same board. Perhaps there are others IDs on it you
can use to refine a search)

Failing that, I'd try to figure out what it's powering to see what voltage it needs to produce.
Also looking at LT1084 specs you might find a difference that would help identity it
(not very familiar with that one, but traditionally fixed regs have a control signal that it grounded,
while var-regs usually have some sort of adjusted control - perhaps you can see how it's connected
with power-off (ground .vs. not should be pretty easy)

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

The three holes (from left to right in your picture) should be Vin, Vout and GND (in case of a fixed reg.) or ADJ (if it's the adjustable version). So it should be easy to check if the third pin is connected to ground or not. If it is intended for and adjustable regulator that pin should be connected between two resistors that work as a voltage divider.

Yes If pin 1 (closest to board edge is Vadjust) goes to center of those two 1/4 watt resistors R182 and R183 then the variable Vreg is what you want.
Doing some simple math off this datasheet https://www.analog.com/media/en/technical-doc … ts/108345fh.pdf
The Volts out be about 3.5v output which would work for both standard and VRE cpu's.
Those resistors are 124ohm 1% and 315ohm 1% which tells me they should be for a Vreg or else 10% would be good enough....

Thanks for the information. I checked the continuity and the ADJ pin is not connected to ground. The regulator is connected to R182, R183.

Based on the markings, those look like they are supposed to be 124 ohm and 215 ohm. I measured them and they reported ~119 ohm and ~202 ohm respectively. Based on the datasheet, this should yield a voltage of ~3.3V.

I installed an LT1084CT regulator. But when measuring it, it's outputting a perfect 5V. I happened to have a second LT1084CT , so I tried swapping it just to make sure, but the second one outputted the same 5V.

I even tested using a 390ohm resistor on a VCC pin in the CPU socket just in case it needed some load, but it still was reading 5V.

I'm a bit confused as to what is happening here. I'm also concerned that I just shot 5V through VCC that should only have 3.3V. Not sure if I might have just damaged some components in the process, though it was only powered up a couple seconds in both tests.

Is it possible these are fixed 5V regulators that are somehow mislabelled? Is there something else going on with the circuit that is forcing it to 5V?

Weird... only thing I can think of is the regulator output being shorted to the +5V input on the motherboard itself. Some 486 boards with voltage autodetection do this to switch between +5V and +3.3V depending on the VOLDET CPU pin, but it doesn't make much sense on a Socket 5 board.

If it's not shorted on the board, it must be a problem with the regulators. Where did you get them from? I've heard about some fakes being sold on eBay or AliExpress.

TheMobRules wrote on 2024-11-13, 19:14:

If it's not shorted on the board, it must be a problem with the regulators. Where did you get them from? I've heard about some fakes being sold on eBay or AliExpress.

Digikey. I would hope they would be legitimate.

edited: I just did the acetone test on the printing on the regulator and it didn't rub off. And it doesn't show any signs of being sanded / relabelled.

Yeah, if it's from DigiKey it should be legit.

If you have some resistors and a couple of caps you can test the regulator on its own, there's a simple example circuit on the datasheet.

I would try to remove the regulator and check if pins Vin and Vout are shorted on the board. Also, there could be a cap (tantalum probably) between ADJ and ground, if that one is shorted it will cause problems with the regulation. Also check the caps between Vin/Vout and ground for shorts.

TheMobRules wrote on 2024-11-13, 19:33:

If you have some resistors and a couple of caps you can test the regulator on its own, there's a simple example circuit on the datasheet.

That's a good plan. I'll try building a basic circuit and see how these regulators actually work.

I would try to remove the regulator and check if pins Vin and Vout are shorted on the board. Also, there could be a cap (tantalum probably) between ADJ and ground, if that one is shorted it will cause problems with the regulation. Also check the caps between Vin/Vout and ground for shorts.

I had tested the pins prior to installing it. None of them are shorted. And when I measure resistance across the pins, it's what I would expect in relation to the two resistors.

I checked the surrounding caps and didn't find any shorts either. I'll keep probing and see what I find. Worst case, I might try swapping some different resistors in just to see what happens.

I also acquired a 3.3V fixed regulator, so I'm wondering if I could modify the circuit to use that instead.

Shponglefan wrote on 2024-11-13, 21:41:

I also acquired a 3.3V fixed regulator, so I'm wondering if I could modify the circuit to use that instead.

Yes you can! Of the resistors that adjust the output voltage (R182, R183), remove the one that connects to Vout and replace the one that connects to ground with a 0 ohm resistor (aka just a jumper wire). That way, pin 1 of the fixed regulator will be connected to GND as expected.

The mystery deepens!

I went ahead and removed the resistors, soldered a jumper wire, and added the 3.3V fixed regulator.

Still 5V!

I'm highly confused by this point. So I removed the regulator and re-tested the pads.

This is where I realized my mistake. In testing the pads, I had tested each of them to ground and confirmed that none of them are shorted to ground. But I didn't test to see if Vin and Vout were shorted to each other.

In testing those, they do appear shorted. It's not a dead short, as I get a 4 ohm reading. Evidently voltage from the 5V rail is passing through somewhere on the board to Vcc. In measuring both Vin and Vout on the regulator, Vin is slightly higher voltage than Vout. So this seems to make sense.

Now the question is what is connected to both the 5V rail and Vcc? If I can find the culprit, I can remove it and see if that short disappears.

TheMobRules wrote on 2024-11-13, 19:14:

Weird... only thing I can think of is the regulator output being shorted to the +5V input on the motherboard itself. Some 486 boards with voltage autodetection do this to switch between +5V and +3.3V depending on the VOLDET CPU pin, but it doesn't make much sense on a Socket 5 board.

After doing a bunch more testing (including trying the 3.3V regulator), it appears that 5V is being fed from the 5V rail into Vout.

Could it be the case that it's somehow switching to 5V in absence of a CPU? Is there a way to determine if this board has some sort of voltage auto detection?

Can you take a photo of the back of the board? That might give us some clues.

kagura1050 wrote on 2024-11-14, 00:30:

Can you take a photo of the back of the board? That might give us some clues.

Sure, here you go:

I did inspect it for things like solder bridges or other weirdness. But there is nothing odd that I have come across.

Can you try removing the cache chips and see if the short from Vcc to 5V persists? At least on my Neptune boards these are 3.3V chips that are powered by the same rail as the CPU. Maybe one of those went bad and is causing the problem.

I believe the 430NX chipset does use 3.3V for some signals but the power supply is 5V and it should be 5V tolerant on the signal pins as well.

Other things to look for: resistors/diodes gone short, accidental bridges.

TheMobRules wrote on 2024-11-14, 01:30:

Can you try removing the cache chips and see if the short from Vcc to 5V persists? At least on my Neptune boards these are 3.3V chips that are powered by the same rail as the CPU. Maybe one of those went bad and is causing the problem.

I tried removing all ICs (cache, BIOS, clock chip) and the issue persists.

Other things to look for: resistors/diodes gone short, accidental bridges.

I've been probing various components (resistors, caps, etc.) and haven't found anything yet. Nor I have I come across any bridges, though I probably need to do a detailed look under the microscope.

How risky would it be to insert a Pentium CPU to test? Maybe this is just normal for this board and that it might drop the voltage with a CPU inserted, while for some reason outputting 5V in absence of one...

Or is there a way to spoof CPU detection in lieu of an actual processor?

If I had a dead Pentium to test, I'd do that, but as far as I know all of mine are functional and I don't really want to sacrifice one on this board.

you can try just a LED or a small value resistor (22...100ohm) as a dummy load. Some regulators need some load before they start regulating, although you measured a low resistance connection between input and output voltages of the regulator which seems unusual...

Tiido wrote on 2024-11-14, 02:24:

you can try just a LED or a small value resistor (22...100ohm) as a dummy load. Some regulators need some load before they start regulating, although you measured a low resistance connection between input and output voltages of the regulator which seems unusual...

Does it matter which pin it goes on? I tried a resistor previously one of the VCC pins, and it didn't change the output voltage.

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

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.).