i920 clocked at 2.8GHz
ASRock X58 Deluxe
6GB Corsair XMS3
2xGTX 260 in SLI
Unbranded 900W supply (Hi-Power ATX900-???? from Ebuyer - no longer in stock )
2x Hitachi 1TB HDD
Cooler Master Scout case
The problem is random shutdowns.
These are full shutdowns so no error, BSOD freezes etc. It just as if the plug is pulled from the supply.
This used to happen if I went overboard with over-clocking so I put this down to overheating. In this case the system would shutdown and resart just like a hardware reset.
Just over the last week a new problem has started.
At first the above shutdown and restart would happen but this time under normal load like email, internet, not just hardcore games.
Over the last two days when the system shutsdown the supply has to be switched off and on before it will power up.
This is what I've done so far
Removed and reseated all the components and cables.
Re-compounded all the heatsink inlcuding NB, MOSFET and SB.
Replaced the tower cooler with the stock intel cooler to get some airflow around the NB and MOSFET heatsinks. These were getting hot.
Done a stress test with Furmark and Prime95 to fully load the system supply wise and thermally. This can run for at least an hour.
Disassembled the PSU to check for shorts, solder blobs floating around and general sloppy soldering. The soldering BTW doesn't look great and neither does the rest of the PSU.
Opened up the case with a desk fan blowing in. This still can shut down.
I've had a look around the PSU and motherboard for the capacitor problem but found nothing unusual. The MB has solid caps anyway so I guess this is a long shot.
I'm currently in the process of removing each component one at a time to see if this helps.
The system now tends to shutdown after a few hours as far as I can tell. I leave it idling overnight doing nothing and in the morning it has shutdown requiring the switch off - switch on to get it going again.
My suspicion is the PSU but I don't have a spare to test it.
The behaviour is like the PSU going into overload or short circuit protection.
The fan works and it doesn't appear to be overheating. The problem will still occur with the PSU out of the case with ambient air.
I would like to know if this is a common pattern of a PSU on it's way out.
What other diagnostics could I do to narrow the supsects down?
Having done some looking around the forums and reviews I'm going to replace the PSU regardless.
The one I'm looking at is the Corsair HX850. It's rated on SLI zone for dual SLI and the reviews seem to suggest that it's a true 850W supply.
I guess I should have done the research on the PSU at the time of the build ( at the time this was for one GTX 260 only).
As I said earlier I was not impressed by the build quality of the PSU, sloppy soldering, cheap caps and just an alround mess intenally.
Corsairs, throughout the model line, will deliver 15% - 20% over their power rating. So what you will have is a very conservatively rated 850 watt PSU or a PSU that practically everyone else would rate at 1000 watts.
I like Corsairs. I own three of them - a 400, a 550, and a 750.
It sounds like a thermal issue of some kind. The First thing I would do is get rid of that PSU. I personaaly like the Corsair and UltraX4. Use this site to determine what size PSU you need. http://extreme.outervision.com/psucalculatorlite.jsp The next thing I would suggest is run the system at a barebomes config. 1 stick of RAM at a time. Usually, you place the RAM in the furthest colored slot away from the CPU. But, check your MOBO. Some board actualy will post on the closest colored slot to the CPU. If that checks out. You may want to run memtest86x+ to check the RAM. If the RAM checks out OK and the system still crashes be sure you set the BIOS back to factory spec. If the system still crashed then You may have and issue with the memory controller on the CPU. This is no way to really test this except by replacing the CPU. If you rule out the RAM, PSU, and BIOS set up, there really is not much else left besided the CPU.