ASUS P5GC-MX - No memory, no boot

[socketa]

Re: ASUS P5GC-MX - No memory, no boot

While looking on the backside of the board for a "shortpin" POJP1 (what's a shortpin?), which doesn't appear to exist after lining up the board with the image on Openboardview, to test it's continuity,

i noticed burned (and possibly shorted) traces
i haven't cleaned them up yet, just in case there is some way to test if they are creating a fault condition

On the front side of the board, in the same vicinity, there is the upper and lower mosfets (directly above "0310" on the barcode sticker in the attached photo) who's gates are connected to the POU1 chip.
These mosfets are responsible for supplying +1.8V_DUAL

Any thoughts/suggestions as what could be happening here, and what to do?

[socketa]

Re: ASUS P5GC-MX - No memory, no boot

Cleaned up the burned trace
It's the 3.3V rail - which measures 108 ohms to ground, with the CPU and CMOS battery removed
There was no short to the adjacent traces - the blackness was the burned lacquer (since the copper is now exposed)
So the initial fault still remains.

When i turn on the power, there is 3.3V on the 3.3V rail
so i noticed this circuitry that changes the 3.3V to 2.5V DAC (whatever that is) and tested there.
There is 3.3V there but on the other side of the component there is not 2.5V but rather some cycling from what appears to be 100mV to 400mV
Those two resistors add up to the reading that my meter gives of 180 ohms so they look to be good
So maybe that "switching diode", D1191, is faulty?
Or am i barking up the wrong tree?

[momaka]

Re: ASUS P5GC-MX - No memory, no boot

Cleaned up the burned trace
It's the 3.3V rail - which measures 108 ohms to ground, with the CPU and CMOS battery removed

So you get less than 2 Ohms continuity between the 3.3V rail on the ATX power connector and this trace?
Reason I ask is because I was looking at my P5GC-MX, and my traces are a little different in that area. The one trace that seems like yours appears to be for the 12V rail and goes to the CPU fan.
Of course, your board is the P5GC-MX/1333, so there can be some differences.

That said, I checked resistance between the 3.3V rail and ground on my P5GC-MX, and I get about 196 Ohms - this is with RAM, CPU, and CMOS battery, though (but PSU disconnected, of course.)

When i turn on the power, there is 3.3V on the 3.3V rail
so i noticed this circuitry that changes the 3.3V to 2.5V DAC (whatever that is) and tested there.
There is 3.3V there but on the other side of the component there is not 2.5V but rather some cycling from what appears to be 100mV to 400mV
Those two resistors add up to the reading that my meter gives of 180 ohms so they look to be good

Wait, you first said you get 108 Ohms to ground above. I'm guessing that was a typo, and you meant to say 180 Ohms? If so, that also appears to line up better with the resistance on the 3.3V rail on my board (~196 Ohms.)

Anyways, this missing 2.5V_DAC voltage may or may not be normal. Although.... see below.

So maybe that "switching diode", D1191, is faulty?

By the looks of the schematic you posted, if there is 3.3V coming to anode of diode D1191, then on the cathode side of D1191 where 2.5V_DAC is, you should get 2.5V. If D1191 was bad (shorted), you should have only seen 82.5 Ohms to ground. Moreover, if D1191 is shorted, 2.5V_DAC should be at 3.3V.

But I guess this diode could also be leaky and/or just partially-shorted to some random resistance (which I recently found is actually possible while troubleshooting an ATX PSU and finding a leaky/partially-shorted 1N4148 diode.)

So maybe remove it and check it. That could be the reason why that trace burned up on the 3.3V rail.

Also, I re-read some of the posts back on page 1 again. In particular, I was interested why there is no 1.8V for the RAM VDDQ rail. As you found correctly, POU1 (an APW7120 PWM controller) is what drives those MOSFETs for the RAM VDDQ. Going by your voltages in post #14, I think pin 7 (OCSET) on the APW7120 controller being at 2.55V may be another flag here.

In particular, going by the APW7120 datasheet:
"The OCSET is a dual-function input pin for over-current protection and shutdown control. Connect a resistor (ROCSET) from this pin to the Drain of the low-side MOSFET. This resistor, an internal 40µA current source (IOCSET), and the MOSFET’s on-resistance (RDSON) set the converter over-current trip level (IPEAK) according to the following formula:
...
Pulling and holding this pin below 0.15V with an open drain device, with very low parasitic capacitor, shuts down the IC with floating output and also resets the over-current counter. Releasing OCSET pin initiates a
new soft-start and the converter works again."

So if pin 7 (OCSET) is held below 0.15V, the IC should be in shut-down state. Your voltage is at 2.55V, so it seems like this is a pass. However, I noticed something on page 4 of the datasheet that I'm not 100% sure about: the parameter for Over-Current Reference. The datasheet says 0.4V is typical (and allowed to vary between 0.37V and 0.43V with temperature changes.) This brings the question: if the voltage drops below 0.4V (but is more than 0.15V), is that how the over-current protection is triggered? ... or is it the other way around? Looking at the application circuit on page 6 of the datasheet, it seems that R5 should be the resistor corresponding to the R_OCSET resistor they talk about in the formula on page 5. And if that's the case, the only logical way I can see how the OCSET circuit works is that an over-current event will cause a larger voltage drop through the lower MOSFET's R_DSON, thus pulling the voltage on one side of R_OCSET lower... and when it goes too low (below 0.4V??), that's when over-current protection should trigger.

But if that's the case above, then the 2.55V on the OCSET pin you measured should allow the APW7120 to start switching and the 1.8V RAM VDDQ should be present. Or am I misunderstanding something here?

If I am correct, then that brings the question whether the APW7120 is faulty or not. With that said, perhaps investigate the 2.5V_DAC circuit and see if D1191 is faulty first, as I think that may be easier than trying to source/replace the APW7120. My guess is that you would just have to take out D1191 out of the circuit and check it in both directions to see if it's faulty or not. (And maybe even build a test circuit and see if it functions correctly as a diode by trying to pass 5V through it in both directions to make sure it's not "leaky" in the reverse direction.)

[socketa]

Re: ASUS P5GC-MX - No memory, no boot

Thanks for the reply,

So you get less than 2 Ohms continuity between the 3.3V rail on the ATX power connector and this trace?

Yes - 0.1 ohms to be be precise

Wait, you first said you get 108 Ohms to ground above. I'm guessing that was a typo, and you meant to say 180

Ohms?

No, there is actually 108 ohms to ground - still an error by me though, as i added up the two resistor values (which are actually not even installed!) which came close to 180 and i dyslexically equated that with 108.

So, even with those two resistors missing, there is 108 ohms from the 3.3V rail to ground, (or equivalently across the diode, since it's in parallel with the 3.3V rail)

So i've now removed the switching diode and it appears to test good?
On diode mode:
+ve of diode to +ve meter, -ve of diode to -ve of meter = 0.580V for either -ve legs of diode
Reversing the meter leads gives no reading
+ve of diode to +ve meter, both -ve legs of diode to -ve of meter = 0.555V
Now, with meter on voltage selection, connecting 5V to +ve leg of diode, then both -ve legs to meter, and other meter lead to earth, gives 4.85V
And swapping the wires going to PSU around, gives 0V

i added a photo of the two omitted resistors (circled in red), the switching diode (orange), and the two capacitors (yellow)

i'll contemplate the rest of what you said about the PWM Driver
Maybe something will click, but i doubt it, especially if you are not even sure what's happening there.

[socketa]

Re: ASUS P5GC-MX - No memory, no boot

Finally got a replacement for the switching diode, D1191, and now have the 2.5V on the +2.5V_DAC rail (see schematic on post #22)
So it really seems like the original switching diode was faulty
But i still only have "0000" displayed on a PCI post card, and no VCore for the CPU
And the only voltage that i'm getting on the RAM slots is 3.3V
Is it necessary to have all of the RAM voltages present before the board starts to POST?

[socketa]

Re: ASUS P5GC-MX - No memory, no boot

Fixed!
Finally replaced the APW7120 PWM buck controller, and it now boots.
Thankyou to those that chipped in.

[momaka]

Re: ASUS P5GC-MX - No memory, no boot

Cool!

Good to see another motherboard saved from the dump, even if it took 3 years to get there (but hey, you did it on your own, so applause there. )

I'm surprised to see the PWM controller dead. Those things usually don't die unless a MOSFET does. Very strange. Maybe the previous owner(s) ran it with a really poor quality PSU?