Board 1)
I reflowed the Ground pin after noticing that the lead wasn't actually making a perfect connection.
Board now works 100% with all cpus.

Board 3)
After replacing the LM358 and the connected electro cap (6.3v 1000µF), VTT is still lower than expected but got a little higher.
Now cpus that didn't work get detected (from -- -- to C1 00) but still no POST.

Board 3)
I sincerely can't get what's not working in the VTT generation circuit.

I tried desoldering the G714 but it changed nothing, ADJ voltage is always 3.8V (from LM358).
I don't think it is the culprit, it's not a broken non-LDO.

It has 3 leads:
1=ADJ // 2=Vcc // 3=Output (Vtt)

ADJ voltage is always 3.8V no matter what.
ADJ is generated by the LM358 output1 (InA- and InA+ --> outputA). On a working circuit (InA-) and (InA+) are =1.49xxV (not equal values as they are generated by two different circuits) and outputA is 2.1V
From LM358 to Adj there are 2 Resistances, that should regulate this Adj voltage.

I replaced the LM358 but still 3.8V on Adj. LM358 isn't the culprit either.
There is something in between that isn't allowing a proper voltage to LM358.

It's driving me crazy.
---------

If I follow the path of Vtt to the watchdog, at some point it goes through a resistance and instead of losing voltage it goes up a little, say from 1.0 to 1.05 (happens on good board too).
That's probably my ignorance, but shouldn't it go down?

If I don't install a cpu, Vtt = 0.04V <-------- a Vtt = to ~1.5 should always be present, cpu in or not;
This could be because of Voutput = 3.8V;

Vtt sink may step in if a cpu is inserted? causing a difference and a Vtt> than 0.04V?

If I install a cpu, Vtt is generated but below 1.5, varying according to the Vcore of the cpu.

Vcore on mosfets is always correct (within the tolerance)

--
Next I'll lift the lead from the pad of the LM358 (#1 output) and see if it outputs the same or if it goes down, maybe there's something wrong (but nodes theory doesn't suggest that: a raise in V... unless I'm wrong).

I could add a wire with a [res] to ground to pull down the 3.8 to a target 2.1 (trial and error here).

--

AI doesn't help much 😀

This board works even without the G714, Vtt is present at the low 0.90 the Celeron 600 Vcore 1.55V needs.
I have no clue how that can be.
My guess is that I need to try another non-LDO, like the LM1117.
The G714 is not working.

Something is messed.
I'll post drawings later today.

This is a better pic of the area.

---

The Vtt trace goes to this network and exits to the watchdog (LM80).

This is the area without LM358.

The heck is the sot-223???

Edit: if supply is 3.3V from ATX; Adj is 2.1V; Out is 1.49xx - I can test the component by soldering a wire to atx and add a 2.1v from a dc generator.
If the component works, should it output 1.49, if not I can safely presume it is dead.

When I desoldered it the first time, I resoldered it and at some point (by probing) the excess solder touched the supply pin, OUT was 3.4v and 2.0v cpus would post.
The OUT become the input (-) on the LM358. The loop is needed to regulate Vtt...

IDK if this kind of Vtt generator is a good one.

Nice story, well know digging this rabbit holes when spending endless hours at night leading to nowhere.. Sorry I didn't save the G714 chip when dismantling this faulty MB. But AI tried foul me it's just a generic transistor NZT6714. It may be. It's common that VTT regulators are diffetent than common LDO as they have source and SINK capability, so they have some totem pole output stage to achieve this (maybe internal or external transistors). If sink path is broken it may still source some voltage but will not work properly in real use when sink part also needed...
So at MB1 it was a bad solder joint on that leg? At least one success, good...

RayeR wrote on Yesterday, 12:43:

Nice story, well know digging this rabbit holes when spending endless hours at night leading to nowhere.. Sorry I didn't save the G714 chip when dismantling this faulty MB. But AI tried foul me it's just a generic transistor NZT6714. It may be. It's common that VTT regulators are diffetent than common LDO as they have source and SINK capability, so they have some totem pole output stage to achieve this (maybe internal or external transistors). If sink path is broken it may still source some voltage but will not work properly in real use when sink part also needed...
So at MB1 it was a bad solder joint on that leg? At least one success, good...

AI pointed to this NZT6714 and after adding one parameter it answered with "sorry, I made a mistake; it isn't the NZT..." 🤣
I'm baffled, mostly by my lack of knowledge. I do improvise but real life applications are a bitch.

Yep, mobo 1 was solved. Ground acts as negative pole by design and without it the the LM358 doesn't work correctly.
At least I have that.

Mobo3 is stable under load with a Celeron 600.
This is what makes it a mystery. Hiren works, SIV and others work too.
I still have to install an OS and test some graphics card.
IT MAKES NO SENSE! 🤣 🤣

QDI 5/133, 133/E and BrillianX all share the same design with those LDOs.

Edit: on mobo #2 they come with different markings:
f011 (f of fairchild)
G51

Well, let's hail mary this 😀

At this point, I could generate the Vtt by using a 1085/other, like on other motherboards.
ATX +5V or +3.3V
I have a heatsink to attach to.

A hack that could give the much desired stability.

I don't see anything against it.

Edit: the sot-223 has a short between supply and output: it's dead.
Hence all other readings are in accordance to it.
I need to replace the sot-223.

After trying hard with AI on alldatasheets, NZT6714 looks really a good candidate.
I hope this nightmare ends fast.

To verify if G714 is really a NPN transistor you can just try multimeter diode test and check if there are measurable 0,6V drops on BE and BC junctions. To check it more - desolder it and meausere H21E (current amplification) via multimeter transistor test or some simple test circuits with few resistors, pwr supply and ampermeter...

I found these:

https://www.bp6.com/board/viewtopic.php?t=1484

https://www.pchardwarelinks.com/mixed.htm

RayeR wrote on Yesterday, 22:46:

To verify if G714 is really a NPN transistor you can just try multimeter diode test and check if there are measurable 0,6V drops on BE and BC junctions. To check it more - desolder it and meausere H21E (current amplification) via multimeter transistor test or some simple test circuits with few resistors, pwr supply and ampermeter...

I have this tester.

Uhm, diodes may be also in this package. As Uf is higher than normal it looks like zener diode. Or the transistor is damaged. Can you compare with your working MB? For first check don't desolder it, just use diode test to check if pin 3 has some junction to other pins or is really NC.

RayeR wrote on Today, 13:47:

Uhm, diodes may be also in this package. As Uf is higher than normal it looks like zener diode. Or the transistor is damaged. Can you compare with your working MB? For first check don't desolder it, just use diode test to check if pin 3 has some junction to other pins or is really NC.

Considering that pin 2 = supply 3.3V ATX
on soldered chip on good board:

1-2 red/com = 6.1M ohms
1-2 com/red = 5.3M ohms
1-3 red/com = 5.29M ohms
1-3 com/red = 6.8M ohms
2-3 red/com = 250 ohms
2-3 com/red = 250 ohms

desoldered chip on bad board:

1-2 red/com = no reading
1-2 com/red = 6.4 M ohms
1-3 red/com = 6.5 M ohms
1-3 com/red = no reading
2-3 com/red = no reading; measuring pads = 250 ohms
2-3 red/com = no reading; same

Pads have their reading, independent from G714.

Soldered chip may be influenced by other stuff.
I'm waiting an answer from another guy, as this way of generating the Vtt has been used by other boards besides QDI.
On one there are markings in full but pic is blurry. Hope he sees my email.

I'm thinking to use a DC-DC converter, the ones you solder + and - and there's an adjustable screw.
I have one with a 3.0A rating, for a test I think it's enough.

With dc-dc 1.4987V it POSTs with all cpus.

Can't get what's wrong on the other circuit but if I don't find a solution I'll use this tiny 2A max MP1482DS
https://www.alldatasheet.com/datasheet-pdf/vi … P/MP1482DS.html

I can't see Vtt asking for more than 1.5A.
This would be temporary as I'll probably wait for a higher amperage to have headroom.

If anyone knows anything about this...