Zero Ohms on CPU power connector after overclock

[col_sanders]

I've not seen what they look like on that board but they should be able to be tested individually. You'll probably have to remove the heatsink (if any) covering them. They tend to still have their pins available for probing even when surface mounted, unless they've started to put the pins on the underside.

 

[Errata]

I've not seen what they look like on that board but they should be able to be tested individually. You'll probably have to remove the heatsink (if any) covering them. They tend to still have their pins available for probing even when surface mounted, unless they've started to put the pins on the underside.

Yeah access to the pins isn't a problem, but it's the testing them I don't understand. For example if we have multiple transistors that have all three pins (base, collector, emitter) tied together so that they're in parallel, how do I know which one is faulty since by testing base->collector or base->emitter or collector->emitter resistance on one I'll actually be testing on all at once (and if I read an unusual resistance of X through the base all I'll know is that at least one is fried but not which one)? Or is this not the case with PWM because of additional elements in the circuit (like something that would introduce resistance between the transistors' individual base pins)? Of course if I can determine that at least one is toast then to your point probably all of them need to be replaced.

Along the same lines I also don't understand how I can test these ICs while they're still in the motherboard, connected to I don't know what else (everything?) on the motherboard. Seems like it would be impossible to know if the resistance I'm measuring between two points on something complex like a motherboard is the resistance between those two points the short way (through only the gate / IC I'm interested in) or the long way (through a lot of other stuff in the motherboard). Maybe some schematics for this motherboard would shed some light on it for me, but I feel like I'm missing something

 

[Bobnova]

There are two mosfets per phase, a high side and a low side.
All the high sides are connected to the 12v rail on one end, and each mosfet's output side has it's own slab of copper connecting it to an inductor, and then to a capacitor and the CPU power plane. The rest state of these mosfets is open.

The low side mosfets have one side all connected to ground, and the other side all connected to the inductor at the same spot (electrically) as the high side mosfet is connected to the inductor. The rest state of these mosfets is closed (connected).

How you would test them is beyond me.

 

[Errata]

There are two mosfets per phase, a high side and a low side. [...] How you would test them is beyond me.

Actually with this particular motherboad (an MSI 790FX-GD70) the high and low side MOSFETs are combined with a "driver" (not really sure what that is atm) into one package: the Renesas R2J20604 http://documentation.renesas.com/eng/products/transistor/rej03g1605_r2j20604npds.pdf. Then they brand it as "DrMOS Technology" The inductors and capacitors and whatnot outside of these ICs are in their usual places though.

I haven't taken the time to dig into that datasheet but I think this aggravates testing difficulties But that doesn't mean I won't try once I get my hands on another multimeter

 

[Bobnova]

Same answer though really, you'll find that the same external connections are present in the multitude of pins on the driver mosfet package.
The only thing changed is that the gate pin of the mosfets is entirely internal and goes straight to the driver, which receives its signal from outside the IC.
This may complicate testing, depending on what sort of signal the driver is expecting, but the connections are identical.

 

[col_sanders]

Hmm. I think you could check the high side and low side individually by checking GL to VSWH and GH to VIN.
Check page 5 in the PDF for pinout
I think that'll work if they died shorting gate to drain. That could also mean the driver is bad too. Not really sure how to test that kind of thing though. Looks like I've got some reading to do too.