With Mr BIOS I got POST code sequence: 02 03 09 0A
0A - Base 64K pattern test failure
and beep code LH-LLLL that has same meaning as POST code 0A.

I foud there was bent 2 springs in one SIMM socket that caused a short. My bad that I did't look carefully, so focused on hydroxide leakage over board. I straight the springs and now it boots fine with my fresh assembled S3 765VL 😀 It passed common tests and a few games.

Now I want to focus on VRM. I already ordered Am486DX5-133 TQFP208 from aliexpress, because it's significantly cheaper than ceramic PGA and this MB has a footprint left for TQFP. I just need a 3.3V VRM to feed it.

Have you checked if all the DX4 WB / Am5x86 specific landings are connected at the TQFP footprint? (It just says "DX / SX" there.)
And are it´s Vcc landings connected to the VRM output or just to +5V?

I didn't checked the TQFP footprint but PGA. I have a list of pins that are floating on socket side (not now with me, I'll post later) and there were also some control pins probably related to cache WB - from a pinout table they seems to be new defined on DX4 so this MB doesn't know them. So it means I probably cannot use WB, just WT? Or will it not work at all?

List of floating pins at CPU socket:
A3 - TCK (JTAG)
S4 - VOLTDET -> JP43
A10 - INV (only for DX4)
B10 - SMI (SMM)
C10 - SRESET (SMM)
C11 - UP (RESERVED#)
A12 - HITM# (only for DX4)
B12 - CACHE# (only for DX4)
C12 - SMIACT (SMM)
A13 - N.C.
C13 - N.C.
A14 - TDI (JTAG)
B14 - TMS (JTAG)
L15 - PWT (Page Write-Through, output)
N15 - LOCK (Bus Lock, output)
Q15 - BREQ -> JP38 - CLKMUL
B16 - TDO (JTAG)
Q16 - PLOCK (Pseudo-Lock, output)
J17 - PCD (Page Cache Disable, output)
P17 - PCHK (Parity Status, output)

Ou, bloody hell, I missed that the footprint on board is not SQFP208 but PQFP196 (for 486SX) and the pinout is totally different so I cannot solder the DX5 there 🙁

Sigh, another interposer created...

Just sent to JLCPCB...
I was suggessted that used LDO might be Sharp PQ30RV2 in isolated TO220-4. I'll have to check pinout on MB if match the datasheet.

Well, I inspected those 3 empty footprints of VRM onboard. I discovered following pinouts:

Q1: 1 - ADJ, 2 - Vin, 3 - Vout; I cannot find any LDO with such pinout, all common 3-terminal LDOs has output at pin 2 and input at pin3. Maybe this Q1 should be really a NPN transistor? Then it would make a sense as common pinout is: 1-B, 2-C, 3-E. But then there should be some opamp or at least zener diode nearby...

Q1B: 1 - Vin, 2 - Vout, 3 - GND, 4 - ADJ; this match with mentioned Sharp PQ30RV2 but I couldn't find a spare but I found another one KA278R33 - fixed 3,3V/2A that has on/off pin instead of ADJ so just bend it and connect with Vin...

Q1A: 1, 2 - Vin, 3 - GND, 4 - Vout, 5 - ADJ; this math with e.g. SPX29302. I could have one somewhere...

Feedback resistor network was measured as 274 Ω between Vout and ADJ, 430 Ω between ADJ and GND - this gives some strange low Vout value with common 1,25V reference...

I found one spare KA278R33 LDO and soldered it to empty Q1B position. I bent on/off# pin4 to Vin pin 1. I tested the MB with ST 486DX4V100 CPU and it works. I just checked Vcc on PQFP196 footprint and it's still 5V even if 3.3V is selected (3.3V is applied only on ZIF PGA socket). The LDO runs a bit hot but still can be hold by a finger. I'll add a small heatsink later...
I got 30FPS in Doom with S3 765VL VGA. I hope it will works with that Am486DX5 too.

The Am5x86/DX5 just arrived. Sigh, a whole one pin row is bent sideways, I have to fix it before soldering for proper alignment.
I did a simple beep test of VCC & GND pins and found that 4 of 8 corner GND pins are floating - strange. But the rest seems to be OK so there's some silicon inside, not just a fake dummy...

RayeR wrote on 2024-10-02, 14:37:

I found one spare KA278R33 LDO and soldered it to empty Q1B position. I bent on/off# pin4 to Vin pin 1. I tested the MB with ST 486DX4V100 CPU and it works. I just checked Vcc on PQFP196 footprint and it's still 5V even if 3.3V is selected (3.3V is applied only on ZIF PGA socket). The LDO runs a bit hot but still can be hold by a finger. I'll add a small heatsink later...
I got 30FPS in Doom with S3 765VL VGA. I hope it will works with that Am486DX5 too.

Hi there. I have the same board. EXP4044 V1.0 and i'm also trying to add a 3,3v vrm to install a dx4-100 processor.

So, reading what you did, the transistor dindn't work out did he ?

nuno14272 wrote on 2024-10-14, 08:05:

So, reading what you did, the transistor dindn't work out did he ?

I didn't try it as I foud suitable LDO. You can try it with transistor if you want. I hoped that someone who has full featured MB would write the part numbers here...

RayeR wrote on 2024-10-14, 11:10:

nuno14272 wrote on 2024-10-14, 08:05:

So, reading what you did, the transistor dindn't work out did he ?

I didn't try it as I foud suitable LDO. You can try it with transistor if you want. I hoped that someone who has full featured MB would write the part numbers here...

LM1084-3.3

LM1084 3,3v applied

Yes, LM1084 is quite common, but cannot be fitted in Q1 position direcly. You had to do some legs gymnastics to fit it in Q1A. No problem, just I like rather to have heatsink ribs directed to outer side.
If you don't mind I would use your MB photo with pins desc. on my webpage dedicated to this MB (with your credit).

Not my work. Found this online and post it here.
https://www.phantom.sannata.org/viewtopic.php?t=18603

Intend to try out myself soon, after the transistor arrived.

By the way, whats your website ?

OK,
my web is at http://rayer.g6.cz/hardware/retropc2.htm#EXP4044_VRM
I translated only some parts, no time, hope google tran.helps...

nuno14272 wrote on 2024-10-14, 15:20:

LM1084 3,3v applied

Out of curiosity, is that just a fixed 3.3V LM1084, or did you use one of the adjustable ones and make a feedback circuit elsewhere?