Zero Ohms on CPU power connector after overclock

[Errata]

Hi all,

I first posted here http://www.tomshardware.com/forum/301555-30-zero-ohms-power-plug-overclock-790fx but with the number of folks on this site I probably should have posted here first.

The very short version is after exploring overclocking with my 965BE on my MSI 790FX mobo I think I toasted the vregs because an ohm-meter shows 0 ohms between the +12V and ground pins on the motherboard's CPU power connection, and the PSU will only turn on when that 8-pin connector isn't connected. A new motherboard is on the way, but what else might be wrong? Is it abnormal for there to be 0 resistance between those pins?

The link tells the full story. Thoughts? Questions?

 

[Automata]

Wow, you certainly got a rude welcoming over there.

I only have an EVGA 680i to test on, but I get about 1 ohm resistance. Otherwise, it is infinite. It will probably vary from board to board though. Since you have a multi-meter, verify the voltages on the 8 pin are in range. If you have a separate 4 pin connector, try just that. Otherwise, if you have a backup power supply, I'd throw that in.

 

[Errata]

BTW my system specs are as follows:

CPU: Phenom II X4 965BE C3
Motherboard: MSI 790FX-GD70
RAM: 8GB DDR3 Assorted G-Skill @1600
PSU: 1kW Silverstone ST1000-P
Hard drive: A few WD 640's in RAID, SSD for boot
Case: Cooler Master ATCS 840
Monitor, OS, blah blah blah

 

[Bobnova]

If the PSU won't turn on with the 8pin power connector plugged in, the motherboard and/or CPU has an issue.
Step1 IMO is to take the CPU out and see if the PSU will start with the connector plugged in then.

If it still won't start, try that PSU in a different mobo.

 

[Errata]

Wow, you certainly got a rude welcoming over there.

lol yeah kind of. It's much better here so far

Since you have a multi-meter, verify the voltages on the 8 pin are in range. If you have a separate 4 pin connector, try just that. Otherwise, if you have a backup power supply, I'd throw that in.

I checked all the pins in the PSU's 24 and 8-pin connectors and they were all spot on under a load of a few hard drives and case fans spinning, but that was with the 24-pin and 8-pin removed from the motherboard so I could do the paperclip trick.

I'm fairly sure my PSU is solid, unless it's surging. I was using it just the other day to troubleshoot my parents' PC since I'm home for Thanksgiving. But since I've got the time I'll try theirs in my PC!

 

[Errata]

Step1 IMO is to take the CPU out and see if the PSU will start with the connector plugged in then.

Sorry I forgot to include that in my uber-short version. I tried it with everything unplugged from the mobo including CPU and same thing.

 

[johan851]

lol yeah kind of. It's much better here so far


I checked all the pins in the PSU's 24 and 8-pin connectors and they were all spot on under a load of a few hard drives and case fans spinning, but that was with the 24-pin and 8-pin removed from the motherboard so I could do the paperclip trick.

I'm fairly sure my PSU is solid, unless it's surging. I was using it just the other day to troubleshoot my parents' PC since I'm home for Thanksgiving. But since I've got the time I'll try theirs in my PC!

Hmm... not sure if I would want to try the second power supply. If yours is acting normally except when it's plugged into the board it sounds like the board is at fault. What kind of overclock were you running? Pretty moderate, I'm guessing?

 

[Errata]

So I tried my parents' 550W Rosewill and same thing, just a high-pitched chirp and brief moment of the LED fan in the PSU lighting up.

What kind of overclock were you running? Pretty moderate, I'm guessing?

Yeah it was pretty moderate, and not good for long-term use but I wanted to see how far my 1.5 year-old equipment could go that I paid so much for back in the day lol.

I'm a little confused about the voltage settings I was on because there are two on the MSI board that sound identical, one called CPU VID (which I assumed was the traditional CPU core voltage), and another I don't remember the name of but was the only one that made any difference in stability as I nudged the multiplier up. The CPU VID was 1.4something and I'm pretty sure I had the other one over 1.5, maybe even 1.6. All other voltages were stock and I wasn't overclocking NB or HT or RAM or anything else.

In the end (before I went to bed with Prime95 running) I had it 4GHz, 20x multiplier and those voltages. CPU temp was pretty much maxed (68 degrees Celcius) but I figured no harm done for an overnight test. Actually I got it up to 4.3 or something with only one core enabled with lower temps, but that's useless for anything so I went back to OC'ing all four cores.

 

[johan851]

PLL voltage, maybe? Crank that up too high and it's a good way to fry the power stage. I think that's what caused a MOSFET blowout on my P5K after a year or so of high voltage.

Doesn't sound like you pushed the CPU super hard - I hope it's alright. But you'll probably need a new board or potentially a new board and CPU.

 

[Errata]

Hmm... not sure if I would want to try the second power supply.

Actually I'm pretty sure ATX spec says the PSU has to non-destructively handle a permanent short, so no worries even if you put paperclips from all the +12's +5's and +3.3's to ground But to be safe it would be a good idea to keep your fingers out of that mess when you do

 

[johan851]

You're totally right - it *should*. But there are a lot of real crappy power supplies out there with short protection I wouldn't trust.

 

[Errata]

PLL voltage, maybe?

After looking in the mobo manual I'm pretty sure I had CPU VDD (there's no such thing as VID in the BIOS) at one step below 1.5 (which would put it at 1.48 or 1.45 or something), and I kept it there the entire time. The second value is just called CPU Voltage and I was adjusting that because it was finer grained (and apparently the voltage the CPU actually gets? since it's just an offset of CPU VDD). I don't think I ever touched the PLL voltage.

 

[johan851]

So you might have been pushing more than 1.5v including the offset? I've never owned a 1156 rig, but that seems pretty high.

 

[Errata]

So you might have been pushing more than 1.5v including the offset? I've never owned a 1156 rig, but that seems pretty high.

Yes, including the offset it was more than 1.6 but I'm pretty positive less than 1.8. And yeah in hindsight it was too high for my 965BE rig but the heatsink on that motherboard's vregs and NB looks like it could dissipate 100W!

 

[johan851]

1.6v is way too high for that CPU. I haven't pushed 1.8v since silicon was at 130nm.

If you were running 1.6v-1.8v through the processor (someone chime in if I'm wrong about this) then it's not surprising that something went pop. Bummer.

 

[Bobnova]

Motherboard sounds dead.

Officially PSUs should have short circuit protection, reality is a different story though.
Possibly more importantly there is the question of what the PSU considers to be a short circuit, a 1kw unit will have SCP that is a lot harder to trigger than a 400w unit.

 

[Errata]

So I finally had a chance to put my PC back together with a new motherboard and for the record it works fine, so it was the old motherboard. I'll just have to remember not to break 1.6V on the 965

I don't have an ohm-meter handy but I'll try to remember to measure the resistance across the new mobo's 8 pin 12V connector when I get a chance.

 

[Automata]

Glad you figured it out!

You may also want to fix your signature so it doesn't violate our rules, before a mod sees it.

 

[col_sanders]

If you can you can test the mosfets individually on the vrm circuit. Be sure to test each one as if you nuke one more also go with it. I figured that out the hard way when repairing a gtx460 i over volted.

If your digital multimeter has the 3v test mode use that. If they are bad you'll either see no voltage or a short between the gate and drain

It probably would not be worth ordering replacement fets though because they'll all probably be bad if they are in parallel and it would take an annoying level of surface soldering to replace them all(especially if they used lead free solder). Not to mention other stuff is probably damaged as a result.

 

[Errata]

You may also want to fix your signature so it doesn't violate our rules, before a mod sees it.

Lol just got that email :| Thanks though!

If you can you can test the mosfets individually on the vrm circuit.

If your digital multimeter has the 3v test mode use that. If they are bad you'll either see no voltage or a short between the gate and drain

I'm not really sure how to test them individually when they're in parallel besides unsoldering as you said, which would be an even bigger pain than regular MOSFETs because that mobo uses some ICs made by Renesas with at least a dozen pins each. Not to mention I don't have a clue what that circuit looks like inside (nothing some schematics from Renesas won't fix though lol).

It probably would not be worth ordering replacement fets though because they'll all probably be bad if they are in parallel and it would take an annoying level of surface soldering to replace them all(especially if they used lead free solder). Not to mention other stuff is probably damaged as a result.

Good point. Though it kind of sounds like a fun challenge, especially if I can get a working motherboard out of it Perhaps I'll save that for a rainy day, then I'll do both that and replace the capacitors in an LCD monitor's power supply I have laying around.