PSU gone bad? computer keeps shutting off
- Thread starter Moloch
- Start date
Deathlike2
Regular
MatiasZ said:I have seen lot's of people advising NOT to use a 20pin power supply with a 24 pin MB, even if the MB support's it or through an adapter. I'm not saying this is your case, but it seems exactly like the kind of problem you could have from bad-not clean-not enough power to the mainboard circuitry (<-- that was tricky to spell)
I think certain mobos (particular brands) have stated that clearly on their website... there definately is some merit to that point.
radeonic2 said:
You are better off using a PSU that has a 24-pin connector that either can be a 20-pin connector on the fly (there are these "universal connectors") or a 24-pin to 20-pin connector.
You may want to go to the provided link for help:
http://forum.msi.com.tw/
Here's some sample links on this topic for instance:
http://forums.anandtech.com/messageview.aspx?catid=27&threadid=1478987&enterthread=y
http://www.hardwareanalysis.com/content/topic/33202/
http://www.hardwareanalysis.com/content/topic/53923/
Last edited by a moderator:
Well since I was failing the blended prime95 test.. the one that uses lots of ram I thought I'd try something.. so i removed a ram stick.
First I took out my cheapy PNY and used my corsair xms 3200C2 2.2 but it erm.. shut off a few times trying that and when I got into windows I got a bluescreen.
So now I've got the PNY running a P95 blend mode test.
Will run test untill I get tomorrow to be sure.
I was erm.. overvolting my ram a bit and I did hear the corsair XMS was a bit sensitive to overvolting.
In my defense MSI says you may have to use higher voltage than memory specs in this mobo because of the erm.. high performance design.
edit- running memtest86+ in dual channal mode.. got it to load using their ISO.
I just failed test 5
So how does corsair do with rmas?
I bought the ram stick 2 years ago in march...
First I took out my cheapy PNY and used my corsair xms 3200C2 2.2 but it erm.. shut off a few times trying that and when I got into windows I got a bluescreen.
So now I've got the PNY running a P95 blend mode test.
Will run test untill I get tomorrow to be sure.
I was erm.. overvolting my ram a bit and I did hear the corsair XMS was a bit sensitive to overvolting.
In my defense MSI says you may have to use higher voltage than memory specs in this mobo because of the erm.. high performance design.
edit- running memtest86+ in dual channal mode.. got it to load using their ISO.
I just failed test 5
So how does corsair do with rmas?
I bought the ram stick 2 years ago in march...
Last edited by a moderator:
well that's not the problem._xxx_ said:
I ran memtest86+ with just the corsair 3200C2 and it's failing tests 5 and 7.
Ram guy (works for corsar) from the corsair warrantly forums thinks it's not a ram issue for those tests, but my PNY 3200 is running memtest86 fine and dandy.
Oh and no offence but I said in this thread I've been able to play serious sam 2 (albeit with some crashing every once in a while) for hours and also ran 3dmark06 for a few hours.
if heat was a problem I wouldn't be able to run 3dmark 06 without a hitch and also it's prime95 stable for the two tests that don't use lots of ram.
You could probably run a game without issue if it doesn't use enough RAM to get onto the Corsair stick.
Most RAM has a lifetime warranty. I'd get in touch with Corsair and ask them what other tests you should run to warrant an RMA. Maybe you could run both version of Memtest as well as the GoldMem demo.
Dunno about the voltage.
Most RAM has a lifetime warranty. I'd get in touch with Corsair and ask them what other tests you should run to warrant an RMA. Maybe you could run both version of Memtest as well as the GoldMem demo.
Dunno about the voltage.
Deathlike2
Regular
radeonic2 said:
Most RAM work fine if you let the mobo autodetect the amount of voltage needed. If there happens to be a problem with the default, then you should adjust by starting to force the default voltage as suggested by the memory manufacturer (this info is generally available on their website). Most DDR400 memory seem to range from 2.5v to 2.8 where 2.8 is used for the higher quality RAM (either by being more overclockable or it is able to use tighter timings for greater performance). You probably want to increase the voltage .1 at a time and see how stable it is.
well actually corsair tested the ram from 2.8V to 2.9V.. and I was running it at 2.8.Deathlike2 said:
I just assumed 2.6 was standard voltage
I upped it to 2.8V and still fails.The Pny 3200 however I have no idea what voltage they tested it at, just that it's memtest86 stable at 2.6V cas 2.5,3,3,1T.
It does have the same SPD timings besides command rate as the corsair though.
I have reason to believe my computert shutting off and the corsair failing memtest maybe be two different things.
It just shutoff again with the pny ram
hmm.
It ran for the whole night last night and now it's shutting off within 10 minutes of testing.
Last edited by a moderator:
radeonic2 said:well actually corsair tested the ram from 2.8V to 2.9V.. and I was running it at 2.8.
I just assumed 2.6 was standard voltage
The Pny 3200 however I have no idea what voltage they tested it at, just that it's memtest86 stable at 2.6V cas 2.5,3,3,1T.
It does have the same SPD timings besides command rate as the corsair though.
I have reason to believe my computert shutting off and the corsair failing memtest maybe be two different things.
It just shutoff again with the pny ram
It does come back on though on it's own and my mobo is set to leave the computer off if there is a power failure
see how long you can survive in safemode with networking
Deathlike2
Regular
radeonic2 said:
I'll just go back to the power supply argument though... you need a power supply that has a native 24-pin connector. That probably accounts for the shutdown.. it will be a worthwhile investment to get an ATX 2.0+ PSU.
The RAM issue seems definately be a separate issue altogether... you need to test the Corsair memory at 2.9/3.0 (maybe even 2.7) and see if it makes a difference. If it doesn't work, it is probably a problem with the memory (perhaps RMA and see if the memory you get back will work)... or you are just better off getting a matching pair of PNY memory (mixing memory types/brands can only lead to more problems). Getting another PNY memory stick would probably be easier if your mobo doesn't have a very flexible memory voltage selection.
Last edited by a moderator:
Well if I RMA my PSU with enermax perhaps I'll get the atx 2.03 version backDeathlike2 said:
The sticks are rated for the same timings and worked well in my barton config in dual channal mode.
I just noticed my PSU fans which stay on for while after you normally shut down continue to stay on when it does a hard shutdown.
wonder if that means anything.
Last edited by a moderator:
Deathlike2 said:
absolutly no way i agree with this. My fathers computer is a power devouring overclocked prescott with an X800XT and the entire computer is running off a generic POS 400W PSU with a 20-pin connector with the top 4 empty on the board, which is fine and noted many times by motherboard manufacturers that you can operate normally with a 20-pin connection. Its not the best thing to do stability wise, but its not going to cause a computer to go flooky like that, especially not in 2D.
radeonic2 said:Well if I RMA my PSU with enermax perhaps I'll get the atx 2.03 version back
The sticks are rated for the same timings and worked well in my barton config in dual channal mode.
run in safemode with networking for awhile to make sure its not a driver or software issue which is what it sounds like to me if it randomly started happening. I hate to say it but i had some strange experiances with my short stay with 7800GTs as well with reboots and random instability but most of that was solved by a bios revision on the cards....do the safemode thing first.
Well it does it during memtest86, which runs off a bootable floopy/cd.SugarCoat said:
The system was stable for a good few days minus one of my HDDs losing a whole folder and not during a random shutdown.,
Just plugged the bloody thing in and windows said that drive needed to be tested with errors during bootup and it came back with some errors and I watched in horror as it deleted corrupted files
Deathlike2
Regular
SugarCoat said:
Well, they have said that the proper placement of the 20-pin connector is crucial on a mobo that uses 24-pins.. they have said to move the connector primarily to the right... but I'm not sure.
radeonic2 said:
This is normal for Enermax PSUs... to ensure that your PSU "lives longer because of generated heat", they leave the fans on to cool the PSU down and turn the fans off when it reaches an acceptable temperature.
Well I know that but I was thinkin if maybe I tripped a protection thing it would shut down the whole PSU.Deathlike2 said:
I did submit a ticket to enermax so maybe they'll have me do some things to be sure its the PSU.
Or maybe they're tell me to fuck off cuz I have an atx 1.3 psu in an atx 2.03 system
You'll want to disable drive caching immediately to prevent further loss of info: Device Manager > [HD name] > Policies.radeonic2 said:Well it does it during memtest86, which runs off a bootable floopy/cd.
The system was stable for a good few days minus one of my HDDs losing a whole folder and not during a random shutdown.,
Just plugged the bloody thing in and windows said that drive needed to be tested with errors during bootup and it came back with some errors and I watched in horror as it deleted corrupted files
The data loss didn't occur doing a power loss, it occured when I simply plugged it in.Pete said:
It was unplugged, I shut off the system and plugged it in.
then windows says it needs to check drive D (160GB drive) and it finds corrupt files.. result= whole folder gone and a few mp3 files.
This occured actually before the computer started acting up.. by a few hours
I did lose data on another drive though (just a partition) of it, I had just installed windows and I shut system off plugged some of my things in and when I powered it up i had an ntldr missing error.
Last edited by a moderator: