(This is a draft for the time being, sorry)

Hi!

Years ago I bought a motherboard, lately I acquired another one and in between one perfectly 100% working.

I have 3 x QDI 5/133E (https://theretroweb.com/motherboards/s/qdi-p6 … -advance-5-133e).

1) The first one I purchased had a faulty watchdog chip (LM80 https://www.alldatasheet.com/datasheet-pdf/vi … 3/NSC/LM80.html).
Cpu + Chassis fan wouldn't spin and some other stuff too. After replacing it looked ok.
But then I had issues with the CPUs I was putting in: below a certain Vcore voltage it wouldn't POST at all.
LM80 re-replaced: still nothing.

2) bought a working one;
couldn't find the fault of the first board by only comparing stuff...
Total waste of time.

3) got a 3rd one: seller sold it as faulty as he couldn't get it to POST.
POST card had -- -- with a P-II 350 2.0V, I tried lower voltages down to a Celeron 600 1.5V: bingo! POST successful + stable.
Other CPUs halt at C1 00 or 01 C1 or C5 C3.

What happened?

VTT voltage!

I didn't bother to check that one because I didn't know how relevant it is.

This time I went straight to it and found a few interesting things:
VTT should be 1.5V at all times, no connection to Vcore (at least that's what happens on the good board + GTL specifications).
VTT is generated via a circuit of a LM358M + SOT-223 (prob a LM317 or like a 1117 - you get the idea)
(I'll post a scheme later): LM358 works as a comparator, if the output of the 2 compared 1.5V inputs is 2.2V --> LM317 Adj pin generated output 1.5V which is VTT.
A loop I don't understand where the 1.5V "not VTT" is generated. (no clue yet)

Board 1) VTT is higher than, based on Vcore, if it gets out of the range of the LM80, which has a 2.7V tolerance, won't allow POST.
No codes, just -- -- or C1 00.
If it gets close to tolerance it'll stop at C5 C3. Otherwise POST + stable.

If I probe the Ground pin of the LM80, it'll go to target VTT value (1.5V) and all CPUs work.
Why?
The LM358 (https://www.alldatasheet.com/datasheet-pdf/vi … NSC/LM358M.html) is tasked to compare VTT generated by a SOT-223 (like a LM317 - my guess as it comes with no markings),
starts issuing the right voltage to In (-) and output goes to 2.2V.
Is it the LM358 or the LM317 getting corrected somehow by the (+) probe of my mulimeter???

Board 3)

Here the VTT voltage is low, like 1.0V to 0.60V.
The other 1.5 input on LM358 is spot on (1.49-47 within my lame error margin)
Trick of touching probe doesn't work.
VTT is sinked by something in the loop. I have to determine what by desoldering LM358 to see where the trace goes to.

There are 2 caps (solid + electrolytic) and a res (solid) between the 2 but they look fine by measurements. Will desolder and replace all for good measure.

If anyone has any idea on this feel compelled to hop in 😀
Thanks!