You are most likely exceeding the current limit of the VRM which will cause a failure sooner or later... either the Mosfets/circuit design are 15amp limit or their cooling is.
According to the last Prod Spec update 011, the following P3 processors are supported: see picture.
The P3 800Mhz @ 100mhz fsb has TDP of 20.8W or about 12amps and is supported.
Attached picture of the VRM area, those coils have minimum winds which can be a partial clue to a limit (circuit design).
Can you post us the 3 Mosfet numbers ?
That would give a better clue if it is their max current or the circuits design. Thanks

Horun wrote on 2024-04-30, 02:19:

You are most likely exceeding the current limit of the VRM which will cause a failure sooner or later... either the Mosfets/circ […]
Show full quote

You are most likely exceeding the current limit of the VRM which will cause a failure sooner or later... either the Mosfets/circuit design are 15amp limit or their cooling is.
According to the last Prod Spec update 011, the following P3 processors are supported: see picture.
The P3 800Mhz @ 100mhz fsb has TDP of 20.8W or about 12amps and is supported.
Attached picture of the VRM area, those coils have minimum winds which can be a partial clue to a limit (circuit design).
Can you post us the 3 Mosfet numbers ?
That would give a better clue if it is their max current or the circuits design. Thanks

The one by the bottom choke is:

The two on either side of the top choke are both:

Ok it is a design issue, those FET's can handle 30+amps even at 100C temps. The MTD3302 datasheet gives some info on heatsink or board mount, and board mount w/o heatsink they have very limited current capability of about 8A each. So in a bridge design be 16A. Curious what value are those 8 caps ? Comparing to a simple Asus P2B-F with triple+ the coil windings and 15 x 1000mf caps the SU810 cpu vrm looks very wimpy in my opinion.

Horun wrote on 2024-05-01, 02:52:

Ok it is a design issue, those FET's can handle 30+amps even at 100C temps. The MTD3302 datasheet gives some info on heatsink or board mount, and board mount w/o heatsink they have very limited current capability of about 8A each. So in a bridge design be 16A. Curious what value are those 8 caps ? Comparing to a simple Asus P2B-F with triple+ the coil windings and 15 x 1000mf caps the SU810 cpu vrm looks very wimpy in my opinion.

Brown caps are 1000uF (10V) and blue are 820uF (4V).

Edit: although I would think insufficient inductance and/or capacitance world result in ripple out of spec or vDroop out of spec (at high current draw). I would think a hard limit would probably be more related to the cooling on the MOSFETs?

aspiringnobody wrote on 2024-04-30, 01:14:

Hello all, […]
Show full quote

Hello all,

The documentation for this board says:

CAUTION The board supports processors that draw a maximum of 15.2 amps. Using a processor that draws more than 15.2 amps can dam […]
Show full quote

CAUTION
The board supports processors that draw a maximum of 15.2 amps. Using a processor that draws
more than 15.2 amps can damage the processor, the board, and the power supply. See the
processor’s data sheet for current usage requirements.

I've got, on hand: a PIII 1.0GHz (7.5x133), a Celeron 1.1GHz (11x100) and a Celeron 1.0GHz (10x100)
This board only does 100MHz FSB so the PIII runs at 750MHz (and thus I'm sure is well under the limit). The Celerons, however, seem to greatly exceed the current capacity of the motherboard: 18.8A for the 1.1GHz and 16.6A for the 1.0GHz. What do you think the actual risk is to exceed the board's current limit? It seems to run fine with the 1.1GHz chip in -- but obviously I've not left it running long term.
- Evan

Your current estimates for the Celerons seem on the low side. Celerons have the same current draw as Pentiums as far as I know. Here a mini sheet:

PARKE wrote on 2024-05-01, 11:58:

aspiringnobody wrote on 2024-04-30, 01:14:

Hello all, […]
Show full quote

Hello all,

The documentation for this board says:

CAUTION The board supports processors that draw a maximum of 15.2 amps. Using a processor that draws more than 15.2 amps can dam […]
Show full quote

CAUTION
The board supports processors that draw a maximum of 15.2 amps. Using a processor that draws
more than 15.2 amps can damage the processor, the board, and the power supply. See the
processor’s data sheet for current usage requirements.

I've got, on hand: a PIII 1.0GHz (7.5x133), a Celeron 1.1GHz (11x100) and a Celeron 1.0GHz (10x100)
This board only does 100MHz FSB so the PIII runs at 750MHz (and thus I'm sure is well under the limit). The Celerons, however, seem to greatly exceed the current capacity of the motherboard: 18.8A for the 1.1GHz and 16.6A for the 1.0GHz. What do you think the actual risk is to exceed the board's current limit? It seems to run fine with the 1.1GHz chip in -- but obviously I've not left it running long term.
- Evan

Your current estimates for the Celerons seem on the low side. Celerons have the same current draw as Pentiums as far as I know. Here a mini sheet:Copper.JPG

I just did TDP/Voltage (1.1GHz = 33/1.75 = 18.85A) -- intel probably is calculating peak current draw or something like that?

- Evan

I'm not sure. This here is an interesting article on the subject:
https://www.roberthancock.com/dell/xpsrproc.htm

PS I have the same cpu's in a couple of systems but these are "VRM 8.4 ready" and are supposed to be able handle the current.

PARKE wrote on 2024-05-01, 16:41:

I'm not sure. This here is an interesting article on the subject:
https://www.roberthancock.com/dell/xpsrproc.htm

PS I have the same cpu's in a couple of systems but these are "VRM 8.4 ready" and are supposed to be able handle the current.

Unfortunately lower voltage creates less heat per clock but also increases the relative current draw at a given TDP.

I've ordered an Asus P2B I'm just going to replace the board with someone better. although I'm pretty sure the 1.1GHz Celeron exceeds the current capacity of the P2B also.

Yikes, the 1.4GHz Tualatin in my D815EEA is seriously full on double the 18.2A maximum of that board. I (think) all of my other computers should be okay, though. Would an Asus CUSL2 be an acceptable replacement for the D815EEA? It's got the pin mod done. The VRM certainly looks more robust. ASUS doesn't seem to publish a current limit like Intel does, so I'm not sure what the real limit is. I would just buy the TUSL2 that's on eBay right now with bad caps -- but I'm not sure if you can put a modified Tualatin in a board that actually supports the Tualatin.

The Asus CUSL2-C i815 based is a very good soc 370 board with very well built VRM (Mine has 11 x 1500mf caps) and heavy wound coils, running a 1.1Ghz currently w/o issues . https://theretroweb.com/motherboards/s/asus-cusl2-c
Compare to Intel D815EEA with 6 caps and again very limited coil windings, both those indicate a weak amperage VRM. https://theretroweb.com/motherboards/s/intel-d815eea-easton
Just because it is Intel does not mean they made every product good, as Intel made both good retail and low end/less expensive boards back then for OEM usage (Gateway, etc)....

Horun wrote on 2024-05-02, 02:27:

The Asus CUSL2-C i815 based is a very good soc 370 board with very well built VRM (Mine has 11 x 1500mf caps) and heavy wound coils, running a 1.1Ghz currently w/o issues . https://theretroweb.com/motherboards/s/asus-cusl2-c
Compare to Intel D815EEA with 6 caps and again very limited coil windings, both those indicate a weak amperage VRM. https://theretroweb.com/motherboards/s/intel-d815eea-easton
Just because it is Intel does not mean they made every product good, as Intel made both good retail and low end/less expensive boards back then for OEM usage (Gateway, etc)....

I've got a gateway FX510s with a Pentium D 945 SL9QQ but I'm assuming that's okay since the 920 that it came with was also 95W? the FX510SXL seems to have the same board and that came with a Pentium D 940.

My other PCs I think are all okay.

• D975XWT w/ Pentium D 945
• CUSL2 w/ Pentium III 1400
• P2B-N w/ Pentium III 1000/100 (Slot 1)
• MP440BX w/ Pentium II 450
• S7-MVP3 w/ K6-2+ 500
• Intel Advanced/ZP w/ PMMX 266*
• BAT4IP3e w/ Evergreen 5x86 133
• SY-024B1 / SY-024B2 w/ 486DLC-40