• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Is my PSU overloaded?

E

ElGhorkieBar

n00b
Joined
Mar 6, 2011
Messages
2
I put together an always-on server in my parents' house back in 2008, over the years I've added to it. Prior to two weeks ago the specs were:

AMD Opteron 1210
8GB (4x2GB DDR2 sticks)
ASUS M2N-LR/SATA motherboard
Three 7200RPM HDDs and a SATA DVD-RW drive
A 5-port PCI-X network adapter (unused, I put it in to test it, but never got round to removing it)

But the important part, the power supply... I forget what model it was nor its wattage but everything was running fine until recently.

...because I replaced one of the hard-drives and added two more, so now the computer has five HDDs in it in addition to everything else.

After putting the new drives in, I thought I heard the PSU's fan blowing louder than usual, but I had no way of knowing for sure, so I ignored it and put the case back on the box.

Two weeks later, accessing the file-share on one of the drives over a VPN connection (I live about 90 miles away) I started getting errors and timeouts. I accessed the server over RDP and found it locking up all the time and generally unusable. I phoned up my parents and instructed them to power it down and disconnect two of the drives I didn't need. The server returned to regular performance. I then opened up the Event Log and found a load of events:

Log name: System
Event ID: 129
Source: nvstor

"Reset to drive \Device\RaidPort2, was issued"

There was also a few entries saying RaidPort1 instead of 2 as well.

(FWIW, I'm not using Raid, I understand the nVidia storage driver just refers to all the SATA ports as RaidPort).

Now that my server seems to be working okay now, I took a look with CPUID's Hardware Monitor. Here's the readings (parenthesised readings are the historical min and max from the last 15 minutes):

VIN1: 1.78V (1.76 to 1.79)
+3.3V: 1.74V (1.73V to 1.74V)
+5V: 5.01V (4.96V to 50.1V)
+12V: 11.31V (11.25V to 11.37V)

All of the readings look fine except for the +3.3V reading. It says 1.74V, that can't be right. FWIW the SATA drives were all using the 5V rail, so I don't know why 3.3 would be affected if those drives were responsible for a power drain that caused the system to keel over.

I ran through a power-supply calculator and it suggested the server needed a 300W PSU (based on an estimated 275W average usage). I'm not able to check on the current model of PSU. That said, I just looked through my order history on my usual component shopping site and it looks like it's an "eBuyer.com Extra Value 650W modular PSU", yes, this is the piece: http://www.ebuyer.com/product/128675 (now discontinued).

I'll need to see it in-person to ensure the supply is actually 650W, but if it is then it was under-used unless something bad happened to it.

Any comments?
 
I am not an expert on power supplies, but I do have an AS in Electrical Engineering. I am not sure how reliable the software voltage detection is. The only way to get hard evidence would be to measure the voltage physically with a multimeter. If those numbers are true that power supply is not very healthy. I dont know the tolerances of the ATX stardard off hand, but all the rails should be within +-5% or less of their specified voltage. The 3.3V looks like its toast, the 5V looks fine, and the 12 volt seems like it is too low. Actually 5% seems even too much for sensitive electronics. 2% deviation may be adequate.
 
CPUID is not very accurate. In my system, the 3.3v is at 1v, 5v at 5, and 12v at 11.5. Once I get into the bios though, all the voltages are spot on (3.3 at 3.38, 5 at 5.1, and 12 at 12.06). So don't trust CPUID. 5% deviation is the ATX standard.

I think you're actually having issues with the drives themselves. Probably more like the drives are having issues communicating with the motherboard. Or the operating system is having issues with the new drives. That doesn't seem like what would happen when a power supply is overloaded, when it is, it would shut down the entire system.
 
That PSU is probably a shitty unit and not nearly capable of outputting 650W. With that said, I have no way of knowing that for certain. Based on the symptoms you described, it could be a PSU issue, but it could also be something else. The only way to isolate the problem is to actually do some troubleshooting. It could be a faulty drive controller on the motherboard, a bad PSU, a driver issue, etc.

As for the software voltage readings, keep in mind that they are not very accurate at all, so I suggest you simply ignore them. If you really want to know what the voltages are, the only way to truly find out would be to measure them directly from the PSU using a multimeter.
 
I would go with what both Tsumi, and Zero82z said. Its really going to be hard to troubleshoot a hardware problem remotely.
 
Haha. "Extra Value". I wonder where they find these units. Personally, I'd just replace it as a rule of thumb. If it can't be identified, it's questionable and it's most likely going to fry something at some point.
 
First, look at the sata spec. HDD's run mostly on 12v. There is a 5v connection. 3.3v is the new voltage to sata devices that the molex -> converter cables do not supply.

If your 12v reading is accurate, you're over 5% out of spec(less than 11.4v). 5% is the tolerance that 12v devices are supposed to endure before having issues. So adding additional HDD's could totally have dropped you so out of spec that you have stability issues.

Remember, your PSU rating is 650W total. While it would be abnormal and unheard of to have the majority of the power elsewhere on the psu, it is possible that most of your power is on the 5 and 3.3v rails, so that your psu simply can't handle the load of your CPU as well as all those HDD's. (e.g. 500w split between 5v and 3.3v, and only 150w on the 12v rail)

The only real way to rule the above out is hook up the hardware to a new PSU that you know works, or hook the suspect PSU up to another set of hardware that you know works.

I used to have problems like this on older PSU's. Specifically the enermax 430w and antec pp412x PSU's. Both had adjustable potentiometers to assist with the issue, but as always if you crack open your PSU to try it... You might:
1. kill yourself if you're not cafeful
2. kill the psu
3. turn the pot too much and kill your hardware.
 
Thanks for the tips guys. Next time I'm round at my parents' house I'll try a multimeter on the PSU's ATX connector and molex connectors, I've no intention to open the thing up though, I need it to at least work until I can get a replacement.

That said, I had a similar issue to this a few years ago and the problem turned out to just be a loose SATA data connector, I hate it when the thin connector edges get bent and become loose.
 
That being said, your problem might just be faulty SATA connections. Once again, software voltage readings are very unreliable with the exception of CPU, chipset, and RAM voltages. And even then, those can have a margin of error of nearly 10%.
 
You must log in or register to reply here.
Back
Top