oh also forgot to mention, i am going for high end oc.

Also i obviously don't mind soldering

The amount of time and effort to do that would seem to make it more expensive, and risky, than just buying one. Unless you want the challenge of course.

RetroPCCupboard wrote on 2025-12-04, 20:18:

The amount of time and effort to do that would seem to make it more expensive, and risky, than just buying one. Unless you want the challenge of course.

of course i do lel

I can take high-res photos and do toning out of pins on my PowerLeap iP3-T if required, but I’ve never done any reverse engineering before.

See sig.
I don't have a ready design, but I do have ways to convert boardviews into kicad and there are plenty of leaked boards to choose slot 1 and socket 370 from.

You'll also do well to acquaint yourself with the bx schematic, and various cpu/chipset datasheets.

One free tip : as far as I know, NONE of the slotkets ever bothered changing the main agtl voltages down to 1.25v. 1.5 was fine.

This might help.
slot1 (Klamath), and s370 symbols/footprints should be in there, but I haven't looked in a while.

Just remember there are 2 versions of slot1, and 3 versions of s370.

myne wrote on 2025-12-05, 07:39:

One free tip : as far as I know, NONE of the slotkets ever bothered changing the main agtl voltages down to 1.25v. 1.5 was fine.

Don't some of them have voltage clamps on agtl voltages?

H3nrik V! wrote on 2025-12-05, 09:01:

myne wrote on 2025-12-05, 07:39:

One free tip : as far as I know, NONE of the slotkets ever bothered changing the main agtl voltages down to 1.25v. 1.5 was fine.

Don't some of them have voltage clamps on agtl voltages?

Thousands of motherboards have been working with Tualatin for two decades without any problems.
However, their insufficient speed causes problems - there are corresponding threads ‘works at 100 MHz, does not work at 133.’
Unnecessary entity.

So if I understand correctly, you want to use a Tualatin with a Slot1 adapter, here for me you can get a normal Slot1-370 adapter for Coppermine, and you can make a simple change to the adapter, to make the Tualatin CPUs work.
If instead you don't want to change the adapter, you can modify the CPU, personally I wouldn't do it, but it's possible to do it without cutting any PIN, just cover the PIN to isolate it.
A fairly cheap solution is the Korean PCB, which you find in the auction portal, you can start from that, and get an interposer, so you can keep the CPU without modifying it, and it should work on most 370 motherboards, and via normal adapter also on Slot1.

H3nrik V! wrote on 2025-12-05, 09:01:

myne wrote on 2025-12-05, 07:39:

One free tip : as far as I know, NONE of the slotkets ever bothered changing the main agtl voltages down to 1.25v. 1.5 was fine.

Don't some of them have voltage clamps on agtl voltages?

Some signals definitely changed from 2or 3v to 1.5 iirc and they definitely have to be reduced, but the main bus is fine at 1.5.

It's a lot of fun going through datasheets playing spot the difference - but there is one datasheet that explains how to make a "universal s370"

Btw, in theory, an adapter can be made for s775 cpus. The main signals are all the same, and some early p4 laptop mobos had a 4:1234 clock divider for one of the cpu/chipset to allow p3 and p4 chips on board variants.

Yep, in theory a core 2 could function on a bx at 1/4 of its speed.

myne wrote on 2025-12-06, 01:46:

H3nrik V! wrote on 2025-12-05, 09:01:

myne wrote on 2025-12-05, 07:39:

One free tip : as far as I know, NONE of the slotkets ever bothered changing the main agtl voltages down to 1.25v. 1.5 was fine.

Don't some of them have voltage clamps on agtl voltages?

Some signals definitely changed from 2or 3v to 1.5 iirc and they definitely have to be reduced, but the main bus is fine at 1.5.

There is a 2.5V power input for Pentium 2, which is not used on Pentium 3 Coppermine/Tualatin (1.5V= Vtt).
It's just a jumper or processor PCB routing.
There is no need to shift anything in terms of level or limit anything.
Otherwise, there would be no first revision Celeron Coppermine 533A, which ran on unmodified Mendocino-only boards.

liverygone wrote on 2025-12-04, 17:14:

an ABIT BE6 (non-II) motherboard

As far as I remember, some early revisions of these motherboards, despite supporting voltages starting at 1.25V, have problems starting up.
Most likely, there are nuances with the CPU VID`s overdrive circuits.
There are two logic chips near the monitoring IC that do this.

If you figure them out, then most likely the entire modification of the board for starting Tualatin will boil down to cutting the wire, soldering the wire,

There are a few that matter if the universal 370 board datasheet is to be believed. I read most of it at year or 2 ago.

Tualatins especially changed quite a few levels.

myne wrote on 2025-12-06, 06:37:

There are a few that matter if the universal 370 board datasheet is to be believed. I read most of it at year or 2 ago. […]
Show full quote

Copy code to clipboard

There are a few that matter if the universal 370 board datasheet is to be believed. I read most of it at year or 2 ago.

The author of the topic will tell us about this when he finishes reading the datasheet to the right place. 😀
The 5th harmonic of 133 (150 for OC) MHz is already half a gigahertz, which is impressive.
Here, impedance matching and line length matching are already needed.
So he will still have to read the datasheets.

Tualatins especially changed quite a few levels.

Officially, yes.
But an officialy Tuallatin with 440BX should never work under any circumstances.

Slightly off-topic.
Twenty years ago, there were discussions about the VTT at 440BX. For some, reducing it to 1.3V helped ensure stable operation.
https://www.tipperlinne.com/p2b-dsvtt.htm

Then there was a thread here, not verbatim.
Dudes, it turns out that ASUS regulates the VTT voltage with a BJT transistor.
If the transistor is with low H21 gain, there is not enough output current.
The VTT starts to drop significantly under load.
But the theory that Tualatin with 1.5V VTT does not work stably has already spread, alas.

"Change Q8 from 2sd1802 to RFD3055
Best Regards,
ASUSTeK Technical Support Division"
https://www.eecis.udel.edu/~portnoi/asusreworkguide.html

There's more than VTT.
A bunch of the lesser signals changed at some point.

EG the 5 or so signals from the southbridge/pci bus.

myne wrote on 2025-12-06, 07:31:

There's more than VTT.
A bunch of the lesser signals changed at some point.

EG the 5 or so signals from the southbridge/pci bus.

P2B Pentium® II/Celeron™ Motherboard
and
search.php?keywords=p2b++Tualatin

Moreover, this has been working since the release of Tualatin.

This is an unnecessary entity
- increasing the cost of BOM
- leading to development and assembly errors

To begin with, just simply need to copy a simple, bug-free slotket.

MS-6905 slotket V2.3 works fine, V2 only at 100MHz

https://www.google.com/search?sca_esv=fc456dd … ih=705&dpr=2.63

You'll note ALL the major brands had a chip on their Tualatin slotkets

But ok, you know it all

Doesn't the ABIT BE6 have the "BIOS Soft Menu"?

As long as the VRM can output the required voltage, you could probably get away with doing the classic 4 pin mod (remove AN3, AK4 and AJ3 pins and cut AK4 trace to edge connector + bridge AN11- AK4) on any cheap generic Coppermine slotket and setting Vcore from menu.

DIYing a slotket from scratch would be a quite the enterprise anyways...

kaputnik wrote on 2025-12-06, 10:16:

Doesn't the ABIT BE6 have the "BIOS Soft Menu"?

In order to access the Soft menu, you need to get voltage on the CPU.
But there isn't any. 😀

The first revisions of these boards did not set the CPU start-up voltage correctly at launch.
They sounded a siren and shut down after 10 seconds.

PS. They feature an advanced monitoring microchip from TI, which compares the CPU supply voltage with the VIDs.
If the difference is more than 5%, the system shutdown.
And the VIDs are overridden by some two discrete logic IC.
At startup, the VIDs and the resulting voltage do not match.