• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Power issues

B

baldrik

Gawd
Joined
Jan 2, 2005
Messages
817
I've been having an intermittent problem for the past few weeks and I need some help. Randomly, my pc will just shut down. Usually while stress testing or playing a game, but it's not regular enough that I can pin-point the cause. This past week I logged 20 hrs of Borderlands without a single issue and then, suddenly, sunday morning, my pc shut down just from launching Steam. I can run IBT 5 times in a row on maximum one night and then the next day it won't do a standard run. When it decides to shut down, it'll reboot itself and sometimes make it back into windows and sometimes not, before it shuts down again. It's almost as if the power just got unplugged. It'll even do it at stock speeds.

At first, I thought that perhaps I was just drawing too much power from the wall. I had (2) 23" monitors, the overclocked pc, 2.0 speakers & an old 30" crt television running on a power strip. I've since removed everything but the pc from that one outlet. When the machine shut down on me 3 times yesterday, I plugged it into the outlet in the bathroom. I played for about games for about 4 hrs yesterday afternoon without an issue. This morning, it did it again, shortly after startup, with no stress on the pc at all - literally all it did was boot up and I heard it reboot while I was eating breakfast. The plugs in the bedroom do not have ground plugs, so I have to use an adapter that screws on. The plugs in the bathroom do have a ground plg though. The house is 60 years old with the original wiring, so most outlets do not have ground plugs. It's been within the past 5 years that the fuse box was replaced with a breaker.

I also noticed last night, in HWinfo that my 3.3v was hovering at 3.088v and that it had dipped at some point to 2.34v.

So what gives? Is my power supply crapping out? Am I getting bad power from the wall? What can I do to fix that? Can I get a UPS to clean that power up, if that's the case?

The weirdest thing is that I've been running this setup like this since May, with no issues until now.

edit: PC is the one in my sig.
 
Last edited:
HWinfo is notoriously inaccurate, you need to use a multimeter.

Set everything back to stock and see what happens.
 
I don't have a multimeter. I could pick one up, but I wouldn't know where to stick it.

The shutdowns still happen at stock speeds.

Edit: Whatever the name of the ASUS sensor monitor program is shows the same info about the 3.3v as HWinfo. There was a warning displayed when I stopped home for lunch that my 5v had bottomg out to 0.0v.
 
To test 12v, get one of the molex connectors and put the red on the multimeter to the yellow and the black to any of the black. For 5v, put the red to the red on the molex and the black to any black. 3.3v can really only be tested by the SATA connectors. The 3 pins closest to the orange wire carry 3.3v, and once again, black to any black (I would do this on a molex).
 
Well, I thought I was in luck upon finding this in some of grandfathers old stuff.

IMG_0468_zps168da94f.jpg


I changed the battery in it, but it looks busted, won't read anything. Oh darn, now I still have to go buy myself a new toy. I mean tool. Yeah, that's the ticket.
 
Did you have it on DC Volts? Analog meters are typically pretty hard to kill, and dont require a battery.

That being said, I'm not positive thats going o be accurate enough for our uses.
 
Yeah, I had it set to DC volts. I tried all of the settings when that didn't show anything. It did have a battery in it - the contacts in the battery compartment were corroded. I cleaned those off, but still no dice. The leads on the red wires are actually melted a bit on both ends.
 
UPS probably not going to help. Have you checked to see if their are any bulging caps. I would test memory and hard drive with some kind of diagnostic too. Those types of things can cause intermittent issues.
 
I have looked for buldging caps and did not see any. I have not tested the RAM. The boot drive is an SSD. SMART checks out fine on all 4 drives.

I picked up a multimeter at lunch, so I'll see about checking voltages tonight or tomorrow.
 
So my voltages seemed pretty normal. 12.15v, 5.05v, 3.54v. So I pulled the psu out again and and pulled out the surefire flashlight to really give it a looking over. This time I saw something that looked bad =\

IMG_0469_zpsf31c6c40.jpg


That white goop probably belongs on the inside of the capacitor. I don't know how I missed that before. Looks like it's RMA time. Damn thing is only a few months old.
 
After looking a little closer, I think the shadow of the glue makes it look like a gash. I can't really tell without opening it.

So if it's not a blusted capacitor, and all of the voltages are in the normal range, what am I looking for now? I checked the voltage at the wall as well, 121.0v.
 
This is probably where I should have started first.

1_zpsabaa1432.png


I'm testing (1) 8gb stick by itself now. If that tests fine, I'll test the second stick in the same slot.

I'm guessing that since it is such a large amount of ram and I'm actually using so little of it most of the time, this is why my shut down issues are so hit and miss?
 
Most likely yes, the bad section is probably in a little used part of your RAM.
 
I tested both 8gb sticks, individually, in the same slot last night, no errors. I also tested a known 'good' 4gb stick in that slot over night, no errors. I put one of the 8gb sticks into the other slot I was using and left the HCI memtest running this morning when I left for work.

I also stuck both 8gb sticks, individually, into an MSI P67 board that has a memtester in the bios. Both sticks checked out fine with that. My MVG has a button on the board to run a memtester as well, but I've been unable to get that to actually work.

It seems like the HCI memtester has a major draw back in that it can't possibly test all of the ram at once, since windows will be using some. I may have to run the test multiple times with each stick in each slot before I figure out if it's a certain stick or a certain slot that's bad.
 
Yeah, that is one of the drawbacks, but sometimes it can find errors faster than Memtest86+.
 
I went home at lunch to check the system, no errors again. I swapped in the other 8gb stick and set her running again. I also left windows installing on my backup machine so I can test there too. I'll find the bad part eventually!
 
It might be you need to raise the RAM voltage slightly if you're running two sticks. I know it doesn't sound like you should need to, but sometimes quirks like that can happen.
 
As soon as I can pinpoint which part is causing it, I'll see about tweeking the voltages on that part.

I'm starting wonder if perhaps its the IMC on the processor is causing the issue. If all of my sticks test fine by themselves in multiple slots, on multiple machines, that might be the only culprit left.
 
Stick 2 in dimm 2 throws out errors. It tested fine in dimm 4, but it's still possible that the memstester just could not get to the bad spots. I've got it running in the other machine now, running the memtester. it's going to be slow going though, since it's just a dual core pentium G620. If it throws up errors there, I've at least got the problem located. I'll try a voltage bump and if that doesn't work, I guess I'll have to RMA it back to Mushkin.

I'm about to start testing another 4gb stick in dimm 2, just to rule out that dimm as well. Friggin 8gb sticks just take so damn long to check.
 
Yeah, make sure it's the motherboard, not the RAM. If both tested fine in slot 1, chances are it's not the RAM.
 
I've tested 3 sticks in both dimms 2 and 4. The only error encountered was with stick 2 in dimm 2 and I think that if I ram it a few more times in dimm 4, catching different addresses, it'd probably error again. I'm currently running the test with the 2 good 4gb sticks in dimms 2 and 4.

My backup pc actually BSOD'd while running the memtester with stick 2, so I think that stick is the most likely culprit. I'm going to run the tester on that stick a few more times in the backup machine to make sure.

Shit is just time consuming. 1.5-2 hrs to get a "100%" run through on all unused ram on just 8gb.
 
Last edited:
So I left the machine testing for the last 24 hrs on (2) 4gb sticks, to rule out the mobo. No errors whatsoever.

I left the backup machine testing the suspect stick for 24 hrs, to confirm the bad stick. Also no errors whatsoever.

I have no idea what's going on.
 
Interesting, try making sure that the RAM settings are the same across both computers?
 
I did. I was using the XMP profile, which sets the timings to 9-9-9-24 and 1.5v @ 1600mhz.

I stuck the (2) 8gb, suspect sticks back into the main pc this morning before the LSU game and left the tester running. I set them to 1.55v - came back to 4 error boxes. I think I'm just going to RMA these 2 sticks back to Mushkin. The (2) 4gb sticks run fine, so I'll just use those for now.

Also, it wasn't pretty, but GEAUX TIGERS :D
 
You must log in or register to reply here.
Back
Top