Anyone have a clue if my board is affected by this? I could only find 1 voltage regulator near the ATX connector which is supposedly for the cpu voltage (LX8384).
Does this mean that the AGP port is supplied directly from the ATX-connector?

A post on the old alt.comp.periphs.mainboard.giga-byte support board re. Voodoo 3 compatibility with Gigabyte would suggest you may be OK (note - F2 is now the newest BIOS rather than 1.21)

Rainer Schumacher wrote in message <38669959...@news.ndh.net>...
>On Thu, 23 Dec 1999 19:56:32 GMT, ske...@spamoff.hotmail.com (Kubiak)
>wrote:

>
>>
>> Has anyone out there found a gigabyte board that will work
>>fine with the Voodoo 3 chipset <either 2000 or 3000 model>? My BXC
>>has a fit with it <I know what the problem is> but I'm wondering if
>>there are any models out there that work fine. BX2000 perhaps?
>

>I'm using a GA 686LX with newest BIOS (1.21 from 12/98) and a V3 3K
>AGP. Never had any problems. Hope it stays that way.
>
>Rainer

Can you post a picture of your board?
In order to find out wether +3.3V is supplied directly from the PSU to your AGP slot you can use a multimeter to check for continuity between the +3.3V pins on the PSU connector and AGP slot. If there's continuity you're 99% safe.
In the area between the CPU slot and RAM slots there are two larger transistors. One of them is likely generating +3.3V. If there was no continuity between the ATX and AGP connector you need to find out which of the two transistors supplies the AGP slot with current and check wether it gets too hot in operation. Since the transistors both probably get some air movement from the CPU fan I'd wager you're going to be fine with the roughly 15W the Voodoo3 cards use.

I think I had various revisions of the GA-686LX before and a later board had a large coil next to one of the transistors. That would indicate that Gigabyte changed the LDO VRM to a switching design, possibly in reaction to higher power requirements?

I will post pictures of the board this evening (CEST time).

There was another transistor between the cpu and memory slots. This was just a NEC K2941 which seems to be a field effect transistor.

I also searched the net for a good AGP pinout from the solder side but couldn't find it

Storm82 wrote on 2021-07-14, 08:07:

I will post pictures of the board this evening (CEST time).

There was another transistor between the cpu and memory slots. This was just a NEC K2941 which seems to be a field effect transistor.

I also searched the net for a good AGP pinout from the solder side but couldn't find it

I quickly sketched one up.

Thanks alot! That helps.

I will check it out when I'm off work this evening

Alright I checked some connections and there is no continuity directly between the 3.3V of the ATX connector and the specific AGP-3.3V-Pins 🙁

Attached are some images from the board

I managed to pull off the plastic from the ATX connector without damaging it, and somehow it seems that only a few pins are really connected to anything 😁

I cannot imagine that the ATX connector is connected through any inner layers of the pcb.

Am I wrong?

Sorry for triple post

Sorry it took so long to respond.
Many of the ATX connector's pins are connected to internal layers. The ground pins are for sure, as are the main power rails.
Did you check wether any of the larger transistors between CPU slot and RAM slots connect to the +3.3V pins on the AGP connector?

No worries, I did a different project in the meantime 😀

I will check them in the afternoon