View Full Version : Tales Of Computer Woe Episode 24,597: There goes the power supply... (I hope!)

Liger Zero
04-14-2011, 11:17 AM
I'm playing Star Trek Online with my Epic Gaming Rig... Blasting Klingons and running Borg out of town... Good times! :D

Well all of a sudden the screen went funny and naturally everything stopped. I had to hard-reboot... got to Windows and everything froze.

Had to reboot AGAIN... this time got a CMOS CHECKSUM ERROR on boot... and it hung.

Walked away for a few hours... came back and tried it again... it booted to Windows and after about 5 min of "work" it locked up hard again and wouldn't reboot.

Half hour later I was able to get it to reboot to windows... and it crashed after about 5 min.

Instinct says this is the power supply.

That's not a big deal. What WOULD be a big deal is if the Motherboard is fried. This is a "special" motherboard that is hard to get and it is very expensive.

So... is there a chance that the MB is fried? Also... what about my memory? I have... quite a bit of Fancy Ram on this rig.

Power supply is on it's way will be here Saturday. And I can return it if the MB is damaged. Just wondering how these suckers fail... do they "always" fry the attached components or what?

04-14-2011, 11:31 AM
Experience tells me that it is probably not the power supply. That isn't a typical failure mode of a power supply. Check that by looking at the hardware momitor in the BIOS setup. If the voltages are within 10% the PSU is probably fine.

First suspect is overheating of any of the high power chips. Number 1 is the CPU, next is the video GPU, next the bridge chips, then the ram. First make sure that all the heat sinks are nice and clean and the fans are running at the proper rpms. Ensure that the CPU heatsink is properly seated on the CPU. Then take off the side panel and place a high volume room fan so that it blows directly on the MB. Check the CPU temp while running at 100% load for a few minutes. It should stay below 60C.

If that makes no difference then check for bulging capacitors. The top of all the larger electrolytic caps should be dead flat. If any are domed they are toast. If none of that helps then try replacing the power supply.

Liger Zero
04-14-2011, 11:35 AM
In-tres-ting... I'll pull and re-grease all the heat-sinks this evening.

04-14-2011, 11:45 AM
Heatsink grease is actually a heat insulator. The only thing it does is to conduct better than air. You want as close to perfect contact as possible between the sink and the chip case. When you remove the sinks lap the bottom dead flat by carefully stroking them across 400, then 600 and then 1200 sandpaper placed on a surface plate or equivalent. Use some WD-40 as a cutting lube. Float glass will do as a plate if placed on a reasonably flat surface. Then clean off any grease on the chips and put a very thin smear of grease on the heat sink. Replace the sinks and secure.

04-14-2011, 02:38 PM
Evan, I thought the high silver content cpu goo was OK as its high conductivity, so they claim!


04-14-2011, 03:27 PM
You can quickly elimininate the ps:

Here's a site with the how to:


here's the widget to test it:

04-14-2011, 03:37 PM
They make a noise like a rattle snake when they are on their way out ..

i would find a program that tells you all the voltages in real time..see what they are doing when it fails .

have one on my computer ..its called Asus probe

have a look at all the caps ..make sure non are bulging .

all the best.markj

04-14-2011, 03:50 PM
Sounds like motherboard caps to me.

04-14-2011, 03:58 PM
I was running into a similar problem with my Intel Quad Core processor.
The guy at the computer store told me.....wait for it....my power supply was going, and probably wasn't big enough to power the processor and the RAM, and the killer video card. Well, I went home, and checked out hte power supply, and discovered it was only a 300 watt unit, and quite old. My documentation with the Quad Core processor said the minimum power supply was supposed to be 500 watts....Ah Ha! So I went to the big box store, and bought a 750 watt power supply...and still had the problem.

A peek at the hardware monitor in the bios showed I was having repeated "thermal events" which caused a forced shut-down. Exploring alternatives, I found a CPU cooler that used a copper heat sink and a small radiator with twin cooling fans, and a miniture pump to push coolant from the heat sink to the radiator and back (similar to an automobile's cooling system)

After carefully cleaning the "heat transfer grease" from the processor and installing the cooling system, following the instructions. I booted the system up, and opened the bios. The hardware monitor showed a significant drop in CPU temperature, and even after several hours of hard use, the CPU temperature never rose above that mark.

Before I installed the CPU cooler, I had about an hour before the computer would go offline. With the cooler in place, the computer can be left on indefinately with no damaging heat build-up or "thermal events" popping up.

04-14-2011, 04:14 PM
Evan, I thought the high silver content cpu goo was OK as its high conductivity, so they claim!

Compared to air, yes. For some perspective the thermal conductivity of aluminimum is 250 watts per metre per degree Kelvin. Copper is around 400.

Dow Corning 340 Heat Sink Compound is 0.55 watts per m/k

Arctic silver in real world tests scores the same. Toothpaste scored slightly better and Vegemite was about the same. Both tend to dry out quickly while proper thermal compound doesn't. It doesn't much matter what liquid or gel you put between the heat sink and the cpu, the less the gap the better it works. Minimizing the gap is what lapping heat sinks is all about and you will be surprised at how poor the flatness is on many commercial brands.

See here for real test results.


Also take his advice and find a copy of "How to Lie with Statistics" before you visit the Arctic Silver web site.

04-14-2011, 04:55 PM
Power supply fitness is not something you can check with a $9 dongle, Or with your motherboards bios voltage monitoring, Or with even a multimeter.

Sure, if your PSU is fryed, it *might* show a wrong voltage. But more then likey if the computer still runs at all, but unstable as hell, Its likey not the *average* voltage thats the problem, but the ripple voltage.. At 30khz and higher (outside of the usable freqency range of the typical DMM on AC mode)

You basicly need a scope, and need to scope it under load (the real computer load that is).. And then need to have a clue whats 'OK' and whats 'Fail'

Whenever capacitors fail, they tend to allow more ripple to pass through.. this can cause.. just about anything. As an example I once had a bad cap in my 3.3v rail of my psu.. Just *1* bad cap. I could see it bulgeing when I found it. This was the symptons:

Basicly everything I did was stable. No problem at all. Except *1* game that would crash to desktop every 30~60 mins. I assumed it was just the game being unstable as some games are.. Untill I got to one level.. Where it would crash within about 5 mins of going into one area that was unavoidable. Reguardless of graphics settings, resolution, audio settings (even disabled), or where I looked (I tryed a few playthroughs just looking at the floor and walking around)

Eventualy I tryed pulling parts from my PC.. pulled a ram stick. Game stable in that area.. Swaped pulled stick for other stick. Game stable.. Replaced stick. Game unstable.

Sounds like bad ram, or maybe even bad motherboard. Except when I opened my PSU I could clearly see a bulged cap. Replaced it, put all the ram back... And game runs for 2~6 hours beween crashs now (Game was never known to be very stable on ANYONES system)
Note, I played this game (Deus ex 1), a very old game, Shortly after playing HL2 (a rather new game at the time) at highest graphics settings.. And only ever had 1 crash in 20 hours of playtime in HL2!

When I did PC repair, We tested power supplys by... Replacing them. Right off the bat. If problem is fixed, its clearly the PSU.
If a computer came in with 'random' behavor, often crashs, etc, the PSU was the FIRST thing to be checked. And by checked, I mean we'd stick the PC on the bench, disconnect all the psu cables, and just plug a PSU siting outside the case in to the motherboard/hard drives/etc. About 90%* of the time that was the problem.

About 5% of the remaining time, we could'nt reproduce the crashs at all. And the other 5% of the time it was something that actualy took some work. (Hardware conflict, software problems, bad motherboard, etc, etc)

*90% of the time after the great 'motherboard capacitor failure' years where over (or before them), as then it was more like 60% of the time the psu, 30% of the time the bulged motherboard caps, and the remaining 5% was unreproducable with another 5% being something else.

Jaakko Fagerlund
04-14-2011, 05:50 PM
I've had 3 PSU failures in the past on this same machine, the first one was the thing that teached me as it ate my HDD until I noticed anything wrong with it :( The failure mode has always been in the output line electrolytic caps, especially the 12 V lines. I know regularly check them as I clean my machine from dust, they are easy to check just by unscrewing the cover of the PSU. The caps are located just under the bundle of power leads.

The OP's situation sounds either a heat issue or PSU failure. I would check heats first if possible and then open the PSU and if those check out, check the motherboard electrolytic caps that are near the power supply line.

Memory error just barely might perhaps not-so-likely be the issue, if nothing else is wrong. I have had a bad memory module (one 512 MB stick out of two) that works in normal use but very rarely in heavy use might crash the running program or my Ubuntu restarts. The fun thing is, the memory module fails only in MemTest86+ test #8 that does random XOR, but the other tests check out :D Haven't had the money or interest of ordering a new 1 GB stick of DDR400 memory, so I'm going with this until it flips out totally and then I just go with one memory stick :P

04-14-2011, 06:33 PM
Yea memory can be problematic to test because it may only error out under heavy load (See: power supply ripple) or other rare conditions, Usally only found in intense applications and video games. Iv had memory that was bad once (Swaped stick, crashs stoped) and for the life of me I could never get it to 'test bad', even with memtestx86+ But when I stuck it in another PC, that PC found it bad on just the boot up ram check! Brand new ram too.

04-14-2011, 09:18 PM
I've had problems before where the 'cure' was supposed to be the new power supply. When that didn't fix it, it was supposed to be a bad video card, then it was supposed to be a bad motherboard. They even went as far as blaming the modem. I took it to my friend, a network engineer, and she said bad power supply, and we swapped it out- same problem.

The culprit? Bad cmos battery. Other times it was lack of cooling. The computer would seem to be running fairly quietly- hey that's nice- then out it would go. Frozen cpu fan that time.

All the comments about good heat sink contact are right on. There's no substitute for chip-to-sink contact (and of course the air gap eliminator, heat sink grease). I haven't seen this much in computers, but in consumer electronics you can find that when a heat-producing part is screwed to a heat sink, the screw raises the material around the hole and the part is tightly held- just off the surface of the heat sink. I haven't seen this much with tapped holes and bolts, but with screws it was pretty common.

I did a repair once that required replacing some power devices, and I had no heat sink grease on hand at the time. I thought about it, then figured what the heck, I'll use some silicon rubber. That worked fine, but I'm sure it would have been a pain later on to replace that part.

Liger Zero
04-15-2011, 10:08 AM
Ok the computer is at the shop being repaired. It has two issues. One is the power supply the other appears to be bad Ram. As this is "Fancy Ram" it has to be replaced in sets or matched as close as possible... Computer guy has a big pile of ram he can Easter-Egg back and forth until the problem goes away.

Thanks for the advice everyone. The information on heat-sink grease will come in handy too.

04-15-2011, 12:27 PM
I have had several power supply failures over the years, and they have all been 100% failure, no ifs ands or buts. On the other hand, vacuuming and blowing out the dust can work wonders on a neglected PC.

04-15-2011, 12:56 PM
Vacuums and electronics ... very bad due to static buildup. Use compressed air only.

I have had several power supply failures over the years, and they have all been 100% failure, no ifs ands or buts. On the other hand, vacuuming and blowing out the dust can work wonders on a neglected PC.