Just bumping this one - put a temp probe on there today and they're hitting 84°C....

Still keen to hear from anyone else using this board to find out what the deal is and if there is any way to drop the heat on these.

OK - so it seems tied to the AGP slot. I swapped out my 4200 Ti for a PCI FX5200 and the VRMs dropped to 42°C...

So is this down to trying to use a card that pulls too much power from the slot? I'm going to swap in my Voodoo 3 3000 which is a 15W card I think (vs the 4200 Ti which is 33W)

Lostdotfish wrote on 2025-03-27, 16:58:

Just wondering if anyone is using one of these boards regularly and can comment on the super hot voltage regulators near the fro […]
Show full quote

Just wondering if anyone is using one of these boards regularly and can comment on the super hot voltage regulators near the front I/O headers...?

These puppies...

The attachment Screenshot 2025-03-27 165408.png is no longer available

They are almost too hot to touch and I can see some heat discolouration on the reverse of the board under them. Is this normal for this board?

I was looking at pictures on TRW and I think I can see the same heat discolouration on the board photos there.

Maybe worth checking this... Super Socket 7 Board not going POST

It looks like a linear regulator (probably making 3.3V because so many motherboard manufacturers didn't want to use the 3.3V rail from the PSU for some reason). They used two FETs in parallel to try to spread the heat between multiple parts, but it still didn't work out for them.

I can see that the FETs are directly in parallel, which means that they won't share the current perfectly. There's a possibility of thermal runaway here.

lti wrote on 2025-06-15, 16:08:

It looks like a linear regulator (probably making 3.3V because so many motherboard manufacturers didn't want to use the 3.3V rail from the PSU for some reason). They used two FETs in parallel to try to spread the heat between multiple parts, but it still didn't work out for them.

I can see that the FETs are directly in parallel, which means that they won't share the current perfectly. There's a possibility of thermal runaway here.

I found this on another thread.

From falconfly: […]
Show full quote

From falconfly:

AGP voltage regulator notes: There is a problem with the combination of older motherboards and newer AGP graphic cards. This is due to the fact that some of the early AGP motherboards used a linear regulator to control the power supply to the AGP slot. Linear regulators can supply at most 2.5A of current, while the switching regulator used on newer motherboards can supply up to 6A at 3.3V. When these motherboards with linear regulators were produced, the average consumption of a graphics card was 1 to 2A, making a supply of 2.5A more than sufficient. Since then, graphic cards have developed and currently the average consumption is between 4 and 5A. This is no problem for motherboards using a switching regulator, but older boards with a linear regulator may run into issues such as overheating, burnouts, and lockups.

So try to find out what type of regulator is on that board. If it's a late revision board it should be fine i think.

Safe cards for the older linear regulators should be: Geforce2 MX, MatroxG200, TNT1/TNT2 M64, Voodoo Banshee. Voodoo3 3000 will be too much, don't know about the 2000 but i wouldn't risk it.
Banshee is a perfect match for a fast k6-2 anyway, v3 comes to life with p3.

In my experience Nvidia cards are a hit and miss on ss7 boards when it comes to stability. I have a GTS in my asus p5b but a gf2Ti or MX will not run stable. So matrox and 3dfx are your best choices.

If you want absolute trouble-free operation and take no chances with voltage regulators: Use a PCI card.

Pretty much sums this up I guess.

In my experience the Ti4200 is overkill for any super 7 board any way. The mx440 wil be bottlenecked by the k6-3 already. Advantages: less current, gf2 GTS performance and less noise. And a lot better than the fx5200 as well (which in turn is a real dud).
Next get a lower driver version and strain on the CPU is low as well.

Some video cards, like a radeon 9800, get their main power from a separate molex directly on the video card. Wouldn't this take care of the problem? Or if you have a power supply with 3.3v output, you could patch this into the motherboard instead of the weak regulator.

I would glue a radiator there.

nickles rust wrote on 2025-06-15, 19:36:

Some video cards, like a radeon 9800, get their main power from a separate molex directly on the video card. Wouldn't this take care of the problem? Or if you have a power supply with 3.3v output, you could patch this into the motherboard instead of the weak regulator.

https://github.com/TeHSiGGi/agp-power-monitor … RENCE_VALUES.md
11W
Good idea.
The video card and motherboard burned out - the problem solved itself.

I had a 462 board that got super hot.
Replaced the caps and it was fine.

Could be the same issue. Loss of capacitance would put more load on them

myne wrote on 2025-06-16, 05:42:

Could be the same issue. Loss of capacitance would put more load on them

This issue is a stupid copy-paste from AT motherboards, on which +3.3V had to be done locally.
For some S3trio this was enough.
It is always solved simply - transistors are removed, a jumper is made from the ATX connector to the regulator output.