Is my PSU on it way out? 3.3v rail running at 2.89v to 3.12v, lots of random reboots!

KillBoY_UK

KillBoY_UK

KillBoY_UK

Soldato
Joined
20 Apr 2004
Posts
4,383
Location
Oxford
Short version: Lots of random restarts, eps when GPU and CPU is loaded, system restarts after less than 5mins of memtest in dos with a single stick of ram. Noticed in windows and the bios the 3.3v rail is idling at 3.12v drooping under 3v seeing as low as 2.89v

Psu is a 5 year old just out of warranty (as of November 4th) Seasonic based Coolermaster V1000 purchased from the forest

Long story: Came home last night after a long day to play some BF4, but system random rebooted , oh I though, same happened again within 5 mins, then a thrid time booting into windows (BSOD), then when idling at windows.

unpluged the PSU, pressed the power button reset my bios so i was at stock as my hand me down 8pack X99 OC Formula is a quirky beast.

When I got into windows I did a aida64 stress test with everything but the GPU, seemed fine for 5mins, switched on the gpu and the rebooted in under 3mins. I was starting to think it was the GPU more than anything else. Then I did abit of mining on just the GPU, was solid for 30mins+... odd. Played some BF4 and again it rebooted within 10mins no BSOD, maybe it was the ram i though.

Temps where fine (custom water) check HWmoniter and the asrock software 12v/5v rails looked fine but the 3.3v rail was abit low at 3.12v ish at idle. I would also see it drop as low as 2.9/2.85v for more than a second before returning to the low 3v/3.1v hmmm.

booted the UBCD on a usb stick with a single stick of ram and BOOM restarted within 5mins, the next time with more all 4 sticks with in 10min.

As the 3.3v rail feed the ram I'm starting to think its not the ram or the gpu but the PSU
 
I got to that stage at about 12:30 last night so tonight I will try and dig out the multimeter and test the ram with a spare 750ti installed to rule out the GPU when I get home tonight.

Also is there a cooler master rep on the forum, as if it is the PSU as it sucks it looks like its failed just a month out of warranty :(
 
Do you have/remember any older 3.3V voltage readings as reference?
Software readings can be unreliable and reading could have been always so and so.
Jumpstarting PSU and using multimeter to check 3.3V would certainly be starting point.

Not sure if DRAM VRM is still powered from 3.3V.
Originally quite a few PC parts were powered more or less directly from lower voltage rails.
But with decrease in their operating voltage and introduction of step-down VRMs, power might be drawn from 12V to keep current lower allowing easier power routing on mobo.
 
I don't ever remembering the seeing the 3.3v drop that low. Having seeing other people with random restarts with low 3.3v rails is what making me think its the PSU. If I was seeing memtest errors before the restart I would think it be more the ram.

ill be testing the 24pin connector with a multimeter tonight .

Come to think about it I did see the RGB strips go on and off a couple of times last night, at they are powered by a corsair commander pro via a sata contector/3.3v rail

If you have any other opinions on what it could be I like to hear them
 
Memory is still 3.3V on all boards as far as I know.

From the technical teardown of the 26 core xeon motherboard https://www.youtube.com/watch?v=eBmrlHi4s4g

an IR35204 is typically used, which accepts 3.3V supply voltage
https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462576f347501579c95e21172b9


As per the ATX spec +3.135 V to +3.465 is the acceptable range for the 3.3V rail, but from the above spec sheet 2.9V is the minimum accepted voltage for the memory VRMs.


KillBoY_UK said:
Come to think about it I did see the RGB strips go on and off a couple of times last night, at they are powered by a corsair commander pro via a sata contector/3.3v rail
Click to expand...

Assume you've tried unplugging it then (just in case the draw of that is pulling the rail down for some reason)?


As @EsaT mentions though don't necessarily rely on software measurements (especially if you don't have any known "good" figures), it's worth getting a multimeter and checking via any test points on the board, or via the pins on a SATA connector.
 
Armageus said:
Memory is still 3.3V on all boards as far as I know.

From the technical teardown of the 26 core xeon motherboard https://www.youtube.com/watch?v=eBmrlHi4s4g

an IR35204 is typically used, which accepts 3.3V supply voltage
https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d462576f347501579c95e21172b9


As per the ATX spec +3.135 V to +3.465 is the acceptable range for the 3.3V rail, but from the above spec sheet 2.9V is the minimum accepted voltage for the memory VRMs.




Assume you've tried unplugging it then (just in case the draw of that is pulling the rail down for some reason)?


As @EsaT mentions though don't necessarily rely on software measurements (especially if you don't have any known "good" figures), it's worth getting a multimeter and checking via any test points on the board, or via the pins on a SATA connector.
Click to expand...

I though so, Ill be removing the strips tonight to see, I only occurred to me an hour ago that that is related.

Am I right that this out of spec 3.3v rail might be the issue ?

Ps annoyingly while my Asrock X99 OC formula has probe points next to the 24pin but not for the 3.3 line :( the 3 cpu voltages and 12v only
 
Last edited:
Armageus said:
It sounds like it, but it's not necessarily the PSU at fault - it could be the Corsair Commander causing the 3.3V rail to be out of spec
Click to expand...

well I have a 4tb drive, the commander and one other RGB contorller (CC don't take the 4pin RBG from res and waterblock) on that 3.3v line as well as the ram. Are the M.2 ports on that line as well as I have a 970pro in one of them
 
4tb hard drive is unlikely to be drawing via 3.3V, and tbh I'd be surprised if the RGB controllers etc are either. Not sure on where M.2 is powered from.

If you've got some of the old Molex->SATA power adapters lying around, you could use them on your hard drive and RGB controllers which will remove any possibility of 3.3V draw - if the rail still shows low and you still get crashes then I would say the PSU failing is a safe bet.
 
Armageus said:
4tb hard drive is unlikely to be drawing via 3.3V, and tbh I'd be surprised if the RGB controllers etc are either. Not sure on where M.2 is powered from.

If you've got some of the old Molex->SATA power adapters lying around, you could use them on your hard drive and RGB controllers which will remove any possibility of 3.3V draw - if the rail still shows low and you still get crashes then I would say the PSU failing is a safe bet.
Click to expand...

naa I don't but ill unplug everything one by one when I do my tests tonight, if it is the PSU I hope CM will be nice and still do a RMA 1 month out of the 5 year warranty

Would my D5 pump of the lone 4pin Molex be 5.5v or 3.3v rail ?
 
Last edited:
Armageus said:
4tb hard drive is unlikely to be drawing via 3.3V, and tbh I'd be surprised if the RGB controllers etc are either. Not sure on where M.2 is powered from.
Click to expand...
M.2 actually uses 3.3V
http://pinoutguide.com/HD/M.2_NGFF_connector_pinout.shtml

Doubt any 2.5" or 3.5" drive uses 3.3V for anything.
Never seen 3.3V power draw mentioned in precise drive specificiations.
And otherwise problem reports from use of Molex to SATA adapters should have been frequent.

KillBoY_UK said:
Would my D5 pump of the lone 4pin Molex be 5.5v or 3.3v rail ?
Click to expand...
D5 is 12V pump and don't see any reason why any single pump would work from lower voltage.
That would only result unnecessary high current.
 
EsaT said:
M.2 actually uses 3.3V
http://pinoutguide.com/HD/M.2_NGFF_connector_pinout.shtml

Doubt any 2.5" or 3.5" drive uses 3.3V for anything.
Never seen 3.3V power draw mentioned in precise drive specificiations.
And otherwise problem reports from use of Molex to SATA adapters should have been frequent.

D5 is 12V pump and don't see any reason why any single pump would work from lower voltage.
That would only result unnecessary high current.
Click to expand...

well least I wont have to worry about disconnecting my pump for a test tonight
 
Well the PSU or something else has fired hardware

PC no longer posts
Memory error codes (1 of 2) depending on what Ram slot and or stick is used
1 of my sticks if I plug it in gets burning hot on one side

So looks like the 1 of my 4 sticks of skill ram is dead (the hot ) prob the others too.
CPU memory controller/motherboard is fired going by the error code behavior depending on the slot.

God i hope the GPU and SSD are ok,

I could test the m2 ssd on my laptop and game on it untill the next load of sales.

Looks like this is going to be expensive ......
 
Only reason for why single particular DIMM can get hot is somekind short circuit in it.
All DIMM slots get same voltage, so VRM failure damaging DIMM, or currently overvoltage, should cause same for other DIMMs.
And four DIMMs should account for possibility of DIMM slot groups having own VRM circuitry.

Have you yet jumpstarted PSU and used multimeter to check voltages?
It's possibility that memory system was starting to draw such amount of power from 3.3V to cause voltage losses on some motherboard wiring.
Though certainly would want to check that PSU with scope while having some load on it before attaching it to new PC parts.

As for M.2 drive at least we don't have yet reason to suspect it having gotten overvoltage, because that would be certainly more risky than some undervoltage.
 
You must log in or register to reply here.
Back
Top Bottom