I think it's all dumb glue logic, so a remake shouldn't be too hard 😀
Just follow the teachers and make sure you take quality components to solder on.

For the less skilled (among them myself) you could build your own PCB with only thru hole stuff. Three PCB will be a bit larger, but soldering will be easier.

Just take a multimeter and see what's connected to what and note the detail of the components.

The datasheet for RC5042M is online and even gives example circuit and component values for a P.Pro VRM.

Looks like a pretty easy build. JLCPCB circuit boards would be $2 for 5 of them. Probably $15-20 in parts for the rest.

This is actually a project that I want to tackle one of these days.

It is nothing more than a 5v to 3.3v regulator. The voltage is adjustable via the cpu directly. there are 4 pins on the p-pro that connect to the VRM and "program" it for different voltages. Except I dont recall any PPRO chips that didnt use 3.3v. Even the overdrives will set the regulator to 3.3v. The pentium 2 chips WILL set it lower. but slot-1 motherboards with removeable vregs are rather rare.

I think there is an option for 12v input as well, so if you were running an old board on a new power supply, you can install a 12v-3.3v regulator and take some load off the 5V rail.

The only thing the chip on the board does, is "read" the jumpers from the cpu, and set output voltage accordingly.

Also is there a reference datasheet to recreate pentium VRM to supply 3.3V and any split plane at any voltage like 2.0V through 2.8V?

Cheers,

Honestly, If I were designing one of these boards, i'd make 2 versions. one "normal" one that works 100% as the original chips do. and a "upgraded" version that doesn't even draw power from the motherboard, but directly from a EPS 12v tap. (the 4-pin 12v connector from the P4 era) *edited*

PPRo chips dont run at 3.3v, but 2.8 sorry. (the data lines are still 3.3v)

pentiumspeed wrote on 2021-07-25, 01:34:

Also is there a reference datasheet to recreate pentium VRM to supply 3.3V and any split plane at any voltage like 2.0V through 2.8V?

Cheers,

https://static5.arrow.com/pdfs/2014/11/20/18/ … o/144an9664.pdf

Perfect. I'd rather use shielded inductors to keep magnetic noise contained. Torus inductor and open bobbin are noisy. And update the design.

Cheers,

It looks like the HIP6008 is fully compatible and pin-for-pin identical to the HIP6003. Luckily the HIP6008 is still stocked most places, so a new run of VRMs should be do-able.

Setup the schematic in KiCAD with all the footprints and most of the 3D models including the pin socket connector. Had to make a few custom components and footprints. Converted the inductors to SMD for better shielding. It just needs the PCB designed. I'll have to grab one of my Pentium Pro VRM's out of storage to measure. It might be a little bit of a difficult build for a 2-layer board, but should be possible since it could just be made slightly taller to deviate from the standard. They are using a 4-layer board on the original design. But it might make sense to use a 4-layer so all the grounds can be in one large plane to dissipate heat better -- plus there are a lot of Vss pins that need to connect.

Still have to make the cutouts for the socket clips in the PCB, and have to look more into the 12V and 5V feeds. I believe you are supposed to select which feed being used, but I currently have them both hooked up. The application note also has C23 missing, but shows it in the board layout. I think it is supposed to tie into pin 15 of the HIP6008, but maybe they left it out of the schematic for a reason. grrr..

But yea, it fits nice on a 2-layer board. I left room for heatsinks as well -- the clip on type should work fine.

nice!

is it possible to disconnect the 12/5v supply from the motherboard and use a 4-pin molex? (the standard big floppy interface?)

Anything is possible. The VRM socket already has 5V and 12V pins though. Look at the lower left of my KiCAD schematic. Disconnecting those pins and putting a power connector on the VRM really won't gain you anything.

The thought was to take the load off the motherboard connector. I'd also put on jumpers to select your own voltage for overclocking

The motherboard is designed to feed the amount of current needed. It would just be wasted money and space on the PCB to add a separate power connector.

As for jumpers to select voltage, you could re-route the 4 VID lines to make a jumper block. CR2 would need to move to the left below the bank of 8 capacitors, then a 2x4 jumper block or DIP switch could be added in its place. "Open circuit" (1) means using a 1k pull-up resistor to 5V for each line. Shorted (0) means going to Vss/ground. See table:

Anyway, here's the "finished" design. Untested of course, but follows the application note pretty well. There were some part substitutions, but I conservatively gave them margin over the requirements. I made some copper fills to carry the high current. This is only a 2-layer board, so it might help to order the boards with a heavier copper pour. I did calculate for 2oz copper and it has margin, but I've never really made high power parts like this. Some of the resistors needed should be 1% or better tolerance. Please view the HIP6003/6008 datasheets as well for the specific tolerance. Heatsinks recommended on Q1/Q2/CR2. Test at your own risk.

KiCAD, BOM, Schematics, Gerbers included. See attached in a later post.

Order sent to pcbway, and I'm making a mouser shopping list right now.

Well hold on now.. please go over the whole thing and confirm what I did was correct. Don't want any PPro boards/CPU's to smoke.

and..

Switchable VID version:

Hold up.. cancel order -- not sure if I tied in the V output.. 1 sec.

There we go.. files attached.