Dead PSU, then sudden shutdown's...

[Raiker]

Today at 6Pm i installed a new HD into my machine, a WD 320GiB sata drive, it all worked fine. at 8 pm i decided to plug an WD 80GiB sata drive from a client to see if it worked, and yes it worked, i copied some info and left the machine on. about 1 hour later i find the machine shutted down, with the PSU hot enough to fry eggs but the standby led was still on, so i tried to start it again, but i wouldn't. so i replaced the PSU with a surtech and tried again, it didn't boot, i switched off the PSU and on again and this time it started. all went well for less than 1 hour i come and find it shutted off again. standby led on again. i try to start but the only way is cutting power and plugging back in. i start to take items out like fan's and front panel. i end up with only the HD's and it boots, but less than 20 minutes it shutdowns again, the PSU is hot, but since this is a new PSU i don't know how hot it usually is. Now i booted the machine with only the 200GiB disk and i have 15 minutes of uptime. the only things that are running right now are: Hard disk, Video card, video capturer and Tornado @12V. i checked on the control panel if any of the devices was "not found" in case anything was burned. When i was taking components out there where times when it would shutdown 1 second after starting. if i had an adapter i would try to see the 5V on atx signal, but right now i don't have oscope nor atx adapters. i will double post 'cause i don't know how much time this machine will be on, and my other is being used. thx in advance for any advice (so much for Electronic's Technician title...)

[Raiker]

i forgot, temp's are all ok, 39°C cpu, 44°c NB. hmonitor logs reflect no problems, the only curious thing is that all the logs end when it's writting the I/O voltage. i have disabled Hmonitor this time and also unplugged everything but the primary HD. si far 24 minutes. it would be madness if hmonitor checking a voltage was the cause. right now i'm starting ´rime95 on full power consuption to see if overclock is stable (it was more than 8 hours stable when i first got it.)

 

Update, seconds after starting prime95 the system shutted down again. i used genie to load a config of 200 * 10.5 with timmings 2-3-3-11. starting prime ot see if the proc was damaged or what.

 

yet another edit:

prime has run for a few minutes and i started hmonitor out of curiosity, i found that since the Vcore is back to 1.7V (my default even though cpu is a mobile) the 12V rail i down to 11.86~11.80, which is the same a with my old PSU when i had 1.55~1.65 Vcore. if it stays an hour i will try and start hooking up stuff.

 

More edits!

After 38 minutes at 200 * 10.5 with prime95 i'm starting to think that either the cpu, the mother, the memory or something else was damaged when the vtb psu overheated that may have affected the stability of my overclock... or it could be that the difference in psu doesn't allow me to do that overclock...

[kitfit1]

Hi Raiker, just done a search on both the psu's listed in your sig and nothing comes up. Does that mean that both of them are generic psu's of some sort? If they are, can you get hold of a known quality branded psu at all and try that out?

[Clay]

Welcome To DFI-street: Smaller psu's make more heat as they have to work harder. My Ezmedia480 Would blow out some awefully hot air from the DFI/6800gt combo.A hot psu is a sign of bein' overworked.(Or a short somewhere) 11.80 12v+ doesn't seem that low but still it's slightly outta spec.(Is that at idle or underload?) What is the -12v sensor sayin' that may reveal more.

Can you try what Kitfit suggested? Cause some psu's are overrated.

 

 

Edit: Yeah could be that too... the really sad thing is when some psu's go they take the board with them(Sometimes board.cpu and memory too.) Looks like a nice rig I sure hope you get it ironed out. And it turns out to be a weak psu or somrthing fixable.. What are the 3.3v and 5v rails reporting?

[Raiker]

wow, there is people on this board

i just ran memtest for about 1.5 hours, with only the yamaha crw f1 plugged and booted from an ubuntu cd. then i booted the ubuntu live cd and played around for maybe 30 minutes. now i booted with both the yamaha and the maxtor 200GiB. running prime to see if power regulators are damaged... downloading cpuburn to test... i can't get my hands on a more decent power supply since they cost more than 90USD here, and that represents like $280... the Vitsuba psu handled my machine with all the overclock and with up to 3 hd and 2 opticals at the time (didn't have a gf4ti had a gf4se). so i doubt the psu blew due to overcharge. i repeat this surtech psu isn't even warm so it's not a "too much rig for the psu" case. i though of the following: "the psu has shortcircuit protection, so it may be that after X time or when X temp is reached by some mosfet the said mosfet shortcircuits shutting down the psu, but since the +5VSB remains the chipset thinks the machine is still on and thus won't acknowledge the power button when i try to restart"

 

Edit: yes they are generic in that they have no "big name sticker" but all the machines assembled in a friends store used these vitsubas and now surtech (ok they're not high end rigs but they are p4's and stuff like that) so i trust them a little.

 

Editing Spree! wee!

1: over 25 minutes of uptime in windows with 1 hd and 1 crw. listening to music and running cpu burn. cpu at 43° NB at 47° (tornado at 12V north is passive now)

[Raiker]

I cannot explaing how (i simply plugged today and voila, but it works. i'm even upping my overclock to 250 * 10 (i want a full fledged toaster)... thanks for all ppl assume solved...

[Raiker]

i can't believe i'm raising the dead, but i do it so the context is here.

My rig is again suffering from the sudden shutdown's, since it magically fixed, it now magically appeared again.

overclock has been reduced to 250*8 a long time ago (since i couldn't get 250*10 prime stable)

i have an abit nf7 2.0, an AXP 2600 and a supertalent 256Mb mem i could try swapping, but i don't think it's proc or mem (since i really push the machine). The shutdowns are, for now, completly aleatory, just now i was using opera (about 300Mb of ram) and playing some .psf on winamp and it shutted down. I don't think i mentioned it but the RAM Voltage led stays on after the shutdown, and i can't restart the system unless i pull the plug, let the psu drain the +5vsb and then pluggin in...

[medianOCer]

How much voltage are you running through the OCZ EL Gold RAM? Are those Winbond UTT

chips on those sticks?

 

Have you checked to see how hot the SB (south bridge) of the chipset is getting? The

SB on LPB/UI boards can tend to get rather warm and usually need some type of change

for better cooling on them.

 

Have you looked over the board to see if any physical damage took place with the

initial PSU outage?

 

Corrupted BIOS? Have you tried disconnecting everything, pulling the CMOS jumper,

pulling the CMOS battery and leaving the board for a day or two. Then going back

to it and only connecting the minimum amount of components needed to boot and

access the BIOS, loading opti-defaults and then setting up your BIOS a bit more

agressively than opti-defaults and running Memtest, for example?

[Raiker]

no, i haven't really gotten around to heavy duty testing (i'm in the middle of entrance exams right now). i have 1 80mm blowing the OCZ's and another blowing the SB. a 60mm handles de NB. according to my case's thermocouple the SB is at: 43.5°C (110F) but the thermocouple is right between the heatsink not under. I haven't even tried lowering the OC to something like 166*10 (atm 250*8). I don't know the chip on the OCZ but the pn es: OCZ466512ELCGE-K since they are DDR466 i have them slighty OC'ed. Vdimm is 2.9 if i remember correctly. I think i'll buy a wizard cap checker, if i do i'll test every cap on the board, i don't see obvious damage on the board. but then again it ran fine for over 20 days

I'll memtest when i go to the U (in 3 hours) meanwhile i'll run prime. but the last time i primed and the machine would shutdown with or without prime, while running prime it was the same as not running it, it just shutted down.

 

Like i said the last time, the fact that the Vdimm led stays on hints at what may be the problem (a failing device causing a shot circuit, the psu's protection kicks in and shutdowns everything but +5vsb but the bios doesn't acknowlegde so the led is still on and i cant reboot without unplugging). last time i tested with only 1 mem stick vga and 1 hd, and got the same, swapped hds and the same.

[Raiker]

Another update:

 

I just replaced the GF4 ti4600 with a GF4 MX440SE (crappy card if i've seen one) and the system has not shutted down in 1:38 hours. This doesn't mean anything, since the last time it stopped doing it. but it means i can keep my 600 bucks and i don't need to buy that nf4 + opty... go DFI!

[medianOCer]

Is your system still running okay?

[Raiker]

don't really know, it shutted down 1 with the gf4se and today i swapped back the ti4600, and only 1 incident. seems like i'll never know what's wrong... i mean, it ran prime95 more than 8 hours (unless it shutted down, but no errors till that point..) and memtest too so i can't hold the proc or mem responsible (i'll get around to test them on the ABIT later) and swapping the vid card didn't help much (reduced the incidents but didn't solve)...