I have a thread from a few years ago with this issue. The VRM to be bypassed is the one next to the two capacitors. The official instructions from Gigabyte were erroneous. Oddly enough, after 4yrs(I think!) I have just revisited this project (life got in the way, and it got stored). I have replaced the VRM and made a couple more mods. I will finally get arround to finish off the thread when I have finished and tested! Running a wire from the ATX 3.3v lines to the drain of the 3.3v VRM will suffice for you needs. Check the output of the VRM with a digi multimeter to double check you have the right one.

Holler if you have any more q's

Mark

Macsbig wrote on 2025-07-31, 22:47:

I have a thread from a few years ago with this issue. The VRM to be bypassed is the one next to the two capacitors. The official instructions from Gigabyte were erroneous. Oddly enough, after 4yrs(I think!) I have just revisited this project (life got in the way, and it got stored). I have replaced the VRM and made a couple more mods. I will finally get arround to finish off the thread when I have finished and tested! Running a wire from the ATX 3.3v lines to the drain of the 3.3v VRM will suffice for you needs. Check the output of the VRM with a digi multimeter to double check you have the right one.

Holler if you have any more q's

Mark

In my testing the one near PCI "LX8383A" is 3.3v 7.5A, the one near the AGP "LX8384" is 3.5v 5A. this is why I am so confused, if the LX8383A drives the AGP it should be more than capable to drive 1.5v and 3.3v signaling cards. Also did you run the wire with the vrm in question still attached?

here are spec sheets for each VRM.

LX8383-00CP:
https://www.alldatasheet.com/datasheet-pdf/vi … EMI/LX8383.html

LX8384A-00CP:
https://www.alldatasheet.com/datasheet-pdf/vi … X8384-00CP.html

Also Asus P2L97 had this very issue.

Hi, I would think the LX 8383 is a replacement. The original VRMs were rated at 5A, this was part of the original issue. They also get VERY hot. I am currently trialing 7A rated VRM (LT1584CT) with a bespoke heatsink ( being cooled by my easily removeable twin fan mod on my Voodoo3 3000, which I designed to also cool this VRM). So far results look promising, at boot screen I am seeing 68 deg C from a hole drilled in the heatsink after 30 mins, where I have my multimeter temp probe poked in. I need to trial Unreal now with Win95 installed.

It odd your seeing 3.5v on the other VRM, are you sure that's right, I would have thought it's the supply for AGP 2, which can be 1.5v, as I'm sure you know anyway? There's no 3.5v on any mobos I'm aware of. Maybe if someone has modded the board (the LX8383) then maybe a mistake was made if, changing a resistor in the voltage divider of the 8384A adj pin. I would deffo check one or both of the resistors in the voltage divider are not short! This could cause an over volt. As Adj voltage through ground and output pin compare may bias high.

I am not running the wire mod any longer.

Cheers, Mark

I've also dug my one of these boards (6BXE 1.9) out recently, and am having problems.
Firstly, the PSU I have does not offer -5v. The (very annoying) alarm goes off constantly, but the system does boot OK, and the initial screen actually complains to check the system health because the vCore is 2.0... Well, as it's got a PIII (Katmai) 550 in it, that's the correct voltage...
I have ordered a different 500W PSU which does have -5v, so will try that, in case it is that issue, but the bios is just complaining about the wrong voltage.
Otherwise, it may just be disconnect the speaker and live with it...

Which brings me to the next bit - I would like to use a Coppermine based CPU in this, but I cannot get it to boot with either my Slot 1 PIII-650, or my PIII-1GHz Socket370 (100MHz FSB) on a slocket. Both these CPUs work fine in my P2B-B.
If I reset the BIOS, the board powers up for a couple of seconds (fans spin etc) then shuts down again.
Are either of you successfully using coppermine based CPUs?
What BIOS are you using? (I've got the F3a Beta BIOS from Gigabytes website)

I'm not looking at modding the AGP voltage yet, as I'm hoping the Savage4 GPU I'm planning to use with it is low enough current, but I'm thinking I may find a heatsink for that regulator...

Thanks for any help/advice,

F1etch.

Oh, and I second the 8383 being a replacement, my board has an LX8384-00CP in that position.

Indeed you should be OK with the higher amperage rated regulator on there.

If you still want to modify to mitigate the heat factor that will always be present with those types of regulators you could consider an adjustable (or fixed voltage really) "buck converter" board. These are now cheap as borscht, often cheaper then buying a drop in regulator IC alone. A lot more efficient/ less heat.

Macsbig wrote on 2025-08-04, 15:30:

Hi, I would think the LX 8383 is a replacement. The original VRMs were rated at 5A, this was part of the original issue. They al […]
Show full quote

Hi, I would think the LX 8383 is a replacement. The original VRMs were rated at 5A, this was part of the original issue. They also get VERY hot. I am currently trialing 7A rated VRM (LT1584CT) with a bespoke heatsink ( being cooled by my easily removeable twin fan mod on my Voodoo3 3000, which I designed to also cool this VRM). So far results look promising, at boot screen I am seeing 68 deg C from a hole drilled in the heatsink after 30 mins, where I have my multimeter temp probe poked in. I need to trial Unreal now with Win95 installed.

It odd your seeing 3.5v on the other VRM, are you sure that's right, I would have thought it's the supply for AGP 2, which can be 1.5v, as I'm sure you know anyway? There's no 3.5v on any mobos I'm aware of. Maybe if someone has modded the board (the LX8383) then maybe a mistake was made if, changing a resistor in the voltage divider of the 8384A adj pin. I would deffo check one or both of the resistors in the voltage divider are not short! This could cause an over volt. As Adj voltage through ground and output pin compare may bias high.

I am not running the wire mod any longer.

Cheers, Mark

Oh ok that's cool, previous owner had the same idea back in the day and made it so. Now I just need to recap the board.

f1etch wrote on 2025-08-04, 16:56:

I've also dug my one of these boards (6BXE 1.9) out recently, and am having problems. Firstly, the PSU I have does not offer -5v […]
Show full quote

I've also dug my one of these boards (6BXE 1.9) out recently, and am having problems.
Firstly, the PSU I have does not offer -5v. The (very annoying) alarm goes off constantly, but the system does boot OK, and the initial screen actually complains to check the system health because the vCore is 2.0... Well, as it's got a PIII (Katmai) 550 in it, that's the correct voltage...
I have ordered a different 500W PSU which does have -5v, so will try that, in case it is that issue, but the bios is just complaining about the wrong voltage.
Otherwise, it may just be disconnect the speaker and live with it...

Which brings me to the next bit - I would like to use a Coppermine based CPU in this, but I cannot get it to boot with either my Slot 1 PIII-650, or my PIII-1GHz Socket370 (100MHz FSB) on a slocket. Both these CPUs work fine in my P2B-B.
If I reset the BIOS, the board powers up for a couple of seconds (fans spin etc) then shuts down again.
Are either of you successfully using coppermine based CPUs?
What BIOS are you using? (I've got the F3a Beta BIOS from Gigabytes website)

I'm not looking at modding the AGP voltage yet, as I'm hoping the Savage4 GPU I'm planning to use with it is low enough current, but I'm thinking I may find a heatsink for that regulator...

Thanks for any help/advice,

F1etch.

This board only supports up to 100mhz fsb, there is a jumper ( J11) on the board as well for normal mode/ turbo mode fsb. Also are the dip switches in the correct positions? For 6.5x multi dip switches 1,2,3,4 are off, 5,6,7,8 ( FSB ) in off position as well.

xXmobiusXx1 wrote on 2025-08-04, 21:05:

Macsbig wrote on 2025-08-04, 15:30:

Hi, I would think the LX 8383 is a replacement. The original VRMs were rated at 5A, this was part of the original issue. They al […]
Show full quote

Hi, I would think the LX 8383 is a replacement. The original VRMs were rated at 5A, this was part of the original issue. They also get VERY hot. I am currently trialing 7A rated VRM (LT1584CT) with a bespoke heatsink ( being cooled by my easily removeable twin fan mod on my Voodoo3 3000, which I designed to also cool this VRM). So far results look promising, at boot screen I am seeing 68 deg C from a hole drilled in the heatsink after 30 mins, where I have my multimeter temp probe poked in. I need to trial Unreal now with Win95 installed.

It odd your seeing 3.5v on the other VRM, are you sure that's right, I would have thought it's the supply for AGP 2, which can be 1.5v, as I'm sure you know anyway? There's no 3.5v on any mobos I'm aware of. Maybe if someone has modded the board (the LX8383) then maybe a mistake was made if, changing a resistor in the voltage divider of the 8384A adj pin. I would deffo check one or both of the resistors in the voltage divider are not short! This could cause an over volt. As Adj voltage through ground and output pin compare may bias high.

I am not running the wire mod any longer.

Cheers, Mark

Oh ok that's cool, previous owner had the same idea back in the day and made it so. Now I just need to recap the board.

Have you sorted the 3.5v issue?

This may kill a GPU if its supposed to be the AGP 1.5v line

Macsbig wrote on 2025-08-05, 21:34:

Have you sorted the 3.5v issue?

This may kill a GPU if its supposed to be the AGP 1.5v line

And how are you going to insert an AGP 3x only card?
This will be problematic even for a power user. 😀

For reference.
The 1.5V regulator on 440BX motherboards is the VTT voltage for the chipset and CPU.
It has nothing to do with the AGP at all.
Also, the AGP voltage for the 440BX is not switchable and is always 3.3V. After all, this is the AGP 2.0 specification.

OK, Did not expect this result:
PSU (With -5v) turned up today. No more alarm about Vc0re, even though that was not the one "missing" before.
... and no more problems with (100MHz) Coppermine based CPU's either, just booted with my PIII-650 with no issues at all.
I'll do some tests with the Savage4 GPU (For S3TC and DOS compatibility) and see if it's stable. I believe it's a lower power draw card than a Voodoo3 etc, but will check the regulator, and likely add a "memory" heatsink to it anyway.
The board was given to me with a RAGE128 card in it, which I think it had run for years before the owner upgraded their PC, so may also try that.

shevalier wrote on 2025-08-06, 06:41:

And how are you going to insert an AGP 3x only card? This will be problematic even for a power user. :-) […]
Show full quote

Macsbig wrote on 2025-08-05, 21:34:

Have you sorted the 3.5v issue?

This may kill a GPU if its supposed to be the AGP 1.5v line

And how are you going to insert an AGP 3x only card?
This will be problematic even for a power user. 😀

For reference.
The 1.5V regulator on 440BX motherboards is the VTT voltage for the chipset and CPU.
It has nothing to do with the AGP at all.
Also, the AGP voltage for the 440BX is not switchable and is always 3.3V. After all, this is the AGP 2.0 specification.

Even more important to check it's correct then, no?

It's been alot of years, and I admit to being a little rusty on my AGP specs. I've never heard of an AGP 3x card before, so that's new to me.

Point still stands, I'm not aware off the top of my head of any ATX standard using 3.5v, maybe you have some info on that?

Cheers

Macsbig wrote on 2025-08-06, 16:46:

Point still stands, I'm not aware off the top of my head of any ATX standard using 3.5v, maybe you have some info on that?

This is the "famous brands" idea - let's increase the 3.3V supply voltage, and then our hardware will overclock by 0.5% more.
This "idea" at least somehow worked on the ATI RAGE/TNT2, which does not have a any voltage regulator on the board, and all components are powered by AGP.

Why they stupidly bothered to release a motherboard with an AGP voltage regulator in 1999, I don't know.
The Intel reference design didn't suffer from this.
The GeForce 256 and Radeon 7000 (from the end of the same 1999) already had regulators on the video card itself, and, frankly speaking, they did not care about the increased AGP voltage.

P.S.
I would be happy to read how someone tries to run the FX5700 on such a board. 😀
Which LDO to use, and why does it smoke despite the fan being screwed on?
After all, on a normally implemented 440BX, the FX work great.

Haha, interesting, guess it makes sense. But I thought the OP had 3.3v and 3.5v. This is where I'm getting a little confused. Somewhere he needs linear VRM to get his 1.5v. Your post is interesting none the less.
Also he seems to have sorted his issues which is puzzling. Maybe an error in testing?

Macsbig wrote on 2025-08-06, 18:21:

Haha, interesting, guess it makes sense. But I thought the OP had 3.3v and 3.5v. This is where I'm getting a little confused. Somewhere he needs linear VRM to get his 1.5v. Your post is interesting none the less.
Also he seems to have sorted his issues which is puzzling. Maybe an error in testing?

https://elektrotanya.com/intel_440bx_rev_1.0_ … f/download.html
Here are all the answers

1.5V is VTT at S370/Slot 1 CPU
1.5V is for Mendocino/Coppermine. 1.25V is for Tualatin.
His looked in the wrong place and thought the wrong thing.