System Randomly Loses Power (even though connected to UPS)

[djotaku]

OK, I've got a strange issue here, and unfortunately it seems to be semi-random right now so it's proving hard to debug. I have a system connected to a UPS. It's been on there for ~10 months but the problems only started happening recently.

My computer will suddenly lose power. Then the power button won't work to turn it back on. I have to unplug the system from the UPS, rock the switch on the power supply (usually wait a few seconds), then plug it back in, turn it on and everything works. If I just rock the switch, but don't unplug from UPS, it won't work. If I unplug it but don't rock the switch then an orange MOBO light comes on, but the system won't power up until I rock the switch. So I feel like whatever is causing the problem is tripping some sort of safety mechanism on the power supply.

The system might lose power right after it loads Windows or it might go for hours and hours before losing power.

Some details that may or may not be related to this issue:
-when I plug the computer back into the UPS, I see sparks - like static electricity going from the prongs on the plug to the holes in the outlet in the UPS

-When I first built this computer (ten months ago), after a few days the power supply I originally used just up and died. I sent it in for warranty work and put in a new, different brand power supply. Haven't had issues from then until now

-once, a year ago, I thought my 5.1 computer speakers, also plugged into this UPS, were broken. One day it just stopped turning on. I tried changing out the fuse on the speakers and nothing I tried worked. I unplugged it from the UPS and plugged it into a different outlet on the UPS and it started working again.

It LOOKS like it might be a faulty UPS - although I'm confused as to why these problems are only happening intermittently. The UPS is not making any beeping noises as though it's lost mains power. However, I guess it's also possible the power supply is at fault. So, although my first instinct is to replace the UPS, I'd like to see what the tom's hardware crowd thinks first.

(edit: Forgot to provide this before)System Compnents:[\b]
1 x Thermaltake Element V Black Edition Steel Full Tower Computer Case Colorshift Fans VL200K1W2Z
1 x GIGABYTE GA-X58A-UD3R LGA 1366 Intel X58 SATA 6Gb/s USB 3.0 ATX Intel Motherboard
1 x MSI N250GTS Twin Frozr 1G GeForce GTS 250 1GB 256-bit GDDR3 PCI Express 2.0 x16 HDCP Ready SLI Support Video Card
1 x LG 24X Super Multi DVD Rewriter - Bulk Black SATA Model GH24LS50 LightScribe Support - OEM
1 x Microsoft Windows 7 Home Premium 64-bit 1-Pack for System Builders - OEM
1 x Western Digital Caviar Black WD1002FAEX 1TB 7200 RPM SATA 6.0Gb/s 3.5" Internal Hard Drive -Bare Drive
1 x CORSAIR Enthusiast Series CMPSU-650TX 650W ATX12V / EPS12V SLI Ready CrossFire Ready 80 PLUS Certified Active PFC Compatible with Core i7 Power Supply
1 x LG Black 10X Blu-ray Burner - Bulk SATA WH10LS30 LightScribe Support - OEM
1 x Intel Core i7-930 Bloomfield 2.8GHz LGA 1366 130W Quad-Core Desktop Processor BX80601930
1 x Crucial 6GB (3 x 2GB) 240-Pin DDR3 SDRAM DDR3 1066 (PC3 8500) Triple Channel Kit Desktop Memory Model CT3KIT25664BA1067
1 x Crucial RealSSD C300 CTFDDAC064MAG-1G1 2.5" 64GB SATA III MLC Internal Solid State Drive (SSD)

 

[djotaku]

First of all, the power supply I have listed above is the one I ordered with the system that had to be sent for refurb work after a few days. I'll have to take a look later to see what the current power supply in there is.

Second, I've read through a bunch of other posts on here where this happened. It appears to be one of the most common issues users face on here. Unfortunately, it also appears to be one of the toughest to diagnose. Most of the time a power supply swap is called for. Every once in a while, it turns out to be a quirky BIOS. For one guy, it was dependent on what location he plugged it into. (It worked fine at 2/3 places)

I'm still curious if anyone thinks it might be the UPS due to the fact that this is the second power supply to start having problems on this system and the previous problem with my speakers. Maybe someone here can tell me if that makes sense.

 

[RetiredChief]

(1) Arcing when plugging PSU in. Yes this is possible if the PSU rocker switch is on as Inrush current is present briefly. If Rocker switch is off - NO should not have a arc would indicate a wiring problem.
(2) Ups that do NOT provide a "NEAR" sinewave output can damage many of the newer PSUs witch have a PFC circuit. The output of some UPS have a really terrible looking output waveform when on battery power.
(3) have you tried just plugging it into one of the surge protector strips (No UPS). If still have problem, then the Current PSU may have also been damaged by the UPS.

To check the battery on the ups, plug in some lamps (old style filiment) (total of 3 60 watt bulbs or 2 100 Watt bulbs). Unplug UPS and see how long battery last. This will NOT check the waveform output.

 

[djotaku]

Just to double-check with you - if I plug the computer into a power strip and don't have an issue, then it's definitely the UPS. But if it still happens - we don't know if the UPS is causing the failures or not. We just know at the point either the PSU or possibly another component - most often tends to be an overheating CPU - is busted.

Is it enough to plug the power strip into one of the non-battery protected slots on the UPS or does it need to be straight to the wall?

 

[RetiredChief]

(1) Just plug the power strip directly in to the wall, Normally the are surge protected.

(2) No issues, then yes it is the UPS

(3) still have problems, Yes could be the UPS already damaged the CPU. Yes, it could also be a system problem. But you can verify your +5/+12 V rails and temperatures. I normally use HWMonitor. I then run Prime 95 and monitor CPU temp and +5/+12 V rails. Then I run furmark to monitor GPU temps and again monitor the +12V.

 

[djotaku]

Sounds like a plan. What's the best trustworthy site to get those programs:

HWMonitor, Prime 95, and furmark?

 

[RetiredChief]

HWMonitor: http://www.cpuid.com/softwares/hwmonitor.html
Prime 95: http://www.mersenne.org/freesoft/
Furmark: http://www.ozone3d.net/benchmarks/fur/

Note: if you run HWMonitor and see voltage that are reall low, or reall high then the program is not giving correct values - programer screwed up (ie +5 = 3 volts or +12 V = 8volts or 16 volts Reason it is invalid is computer would not be on if they were real)

 

[djotaku]

Here's the important info from hwmonitor:

Hardware Monitors
-------------------------------------------------------------------------

Hardware monitor ITE IT87
Voltage 0 1.20 Volts [0x4B] (CPU VCORE)
Voltage 1 1.52 Volts [0x5F] (VIN1)
Voltage 2 3.33 Volts [0xD0] (+3.3V)
Voltage 3 4.95 Volts [0xB8] (+5V)
Voltage 4 5.12 Volts [0x50] (+12V)
Voltage 5 -12.48 Volts [0xC3] (-12V)
Voltage 6 -1.60 Volts [0x19] (-5V)
Voltage 7 3.66 Volts [0x88] (+5V VCCH)
Voltage 8 3.15 Volts [0xC5] (VBAT)
Temperature 0 36°C (96°F) [0x24] (TMPIN0)
Temperature 1 47°C (116°F) [0x2F] (TMPIN1)
Temperature 2 42°C (107°F) [0x2A] (TMPIN2)
Fan 0 1541 RPM [0x1B6] (FANIN0)
Fan 2 247 RPM [0xAB1] (FANIN2)
Fan PWM 0 0 pc [0x0] (FANPWM0)
Fan PWM 1 0 pc [0x0] (FANPWM1)
Fan PWM 2 0 pc [0x0] (FANPWM2)


Hardware monitor Gigabyte DES
Voltage 0 1.20 Volts [0x4B] (CPU)
Register space LPC, base address = 0x0290

Hardware monitor NVIDIA NVAPI
Voltage 0 0.95 Volts [0x3B6] (VIN0)
Temperature 0 36°C (96°F) [0x24] (TMPIN0)
Fan PWM 0 35 pc [0x23] (FANPWMIN0)

The only strange data there is the +12V is at 5V and -5V is at -1.6V which I guess you said was probably programmer error?

 

[mickey21]

5.12 volts on positive 12V rail, that doesnt look good. Do you have a PSU tester to confirm?

 

[djotaku]

I don't. I'll have to see about getting one. Any other good voltage programs that I might use to verify?

 

[djotaku]

I don't know if SIW is using the same libraries or not, but it also reports 5V on the +12 rail.

 

[djotaku]

My younger brother recently had PSU issues and, although he unfortunately lives too far to borrow his PSU tester, I'm going to ask him which one he got so I can do that test.

I *did* see somewhere online that someone was saying that software voltage was garbage and that you really do have to use a PSU tester to be sure.

If I have some free time this weekend I can also try popping in my spare PSU and seeing if that gives better values or not. And I'll plug it straight into the power strip (not the UPS).

I guess I don't really have a way of knowing if the UPS is causing this without letting it potentially destroy another PSU, so I'm not sure exactly what to do about that.

 

[djotaku]

In terms of continuing to diagnose things - and I still have to figure out how to do a true test on the UPS because I don't really have enough lamps or regular bulbs. Still.....it may be interesting (and possibly unrelated) to note that on my other computer I have on a UPS with Linux and Nut. Periodically - maybe once to twice a week, it comes up and tells me it's gone to battery and then a couple seconds later it goes back to mains power. Maybe I'm drawing too much power on this circuit?

May or may not be related.

 

[djotaku]

Well, it sucks that this is random because so far I've gone 1 hr + without any issues. The first time this happened it was after my system had just booted up and I was trying to launch some programs. The one other time it's happened was after many hours while I was away from the machine.

The only thing I can think of right now is that it happens when it's under high load. According to the small amount of research I've done so far about the low rails voltage - maybe that's an issue when the PC is under high load.

But I guess it's all moot until I get the hardware PSU tester

 

[RetiredChief]

The reading on the +12 volt rail in not incorrect it is INVALID meaning no idea what it is. As indicated you could use a voltmeter. Might try a different program such as speedfan.

A PSU tester is more a go-no-go tester. Will tell you if PSU is OK under a light load, but can not load the +12 V to be a valid test at high load. If software is not indicating then the other two options are (1) use a digital voltment (black meter lead to black molex pin (center two) and red meter lead to the yellow molex pin (one end). Note the other end (red wire is the +5 V). While reading the voltage run furmark to load the +12 V rail.

 

[djotaku]

how do I run furmark while also putting the voltmeter into the system? Wouldn't that require me to have it unplugged?

 

[RetiredChief]

With Power off, connect meter leads to the molex connect - they should just plug in. If a little loose just put a piece of tape to hold. Power up, get reading at idle and then run furmark, get read for under load.

Myself, I power on. plug meter leads in get reading, disconnect meter. run furmark and insert meter leads and get load readins.

PS a DVM runs about 15-20 bucks at walmart. There are some good tutorials on how to use, Just google "how to use a voltmeter"

 

[djotaku]

Yeah, I have a voltmeter and I'm pretty sure it's good up to tens of volts. (I'll double-check. So you are saying to plug it in to a fan plug? Or can I stick it into the top of the 24 pinout?

 

[RetiredChief]

No use the 4 pin Molex connector (goes to old HDDs and DVD drive.
Has red wire (+5V) two center black wires (return/Gnd) and a Yellow (or orange) wire at the other end (+12V)

 

[djotaku]

OK, since it's not my everyday computer and I am busy with my grad school work, I'll get to it this weekend and report in with my results.

 

[djotaku]

Just a bit of an update. Have not yet taken a DMM to the system. I wanted to do a BIOS update to see if that had anything to do with the weird numbers for the volatages as well as to see if it fixes the problem. Just a note - both before and after the BIOS update, the BIOS showed that my positive 12V, 5, and 3.3V rails were spot on. Both before and after the BIOS update the hw program showed my +12V with ridiculous values. 2V as I write this. I'm pretty sure if that were true that the system wouldn't even boot. So I'm definitely going to have to break out the voltmeter.

For completion's sake, the BIOS also listed
VCore 1.22V

 

[djotaku]

To check the battery on the ups, plug in some lamps (old style filiment) (total of 3 60 watt bulbs or 2 100 Watt bulbs). Unplug UPS and see how long battery last. This will NOT check the waveform output.

Retired Chief: this may explain it. Got around to this test. I put 3 60 watt bulbs in and got 8 minutes of power. Obviously that's just 120W vs the potential 650W my computer could be drawing. So do you think it's probably that the mains power was hiccuping and the UPS just didn't have enough power to keep the computer going long enough to shut off? (it had a USB cable that was supposed to shut off the computer if it lost mains power) As I mentioned, I periodically see my Linux computer notify me that it's going to battery power and then within 1-5 seconds going back to mains power.

I'd be strongly interested in your thoughts on this. I can pick up a new UPS today.

Thanks!

 

[RetiredChief]

My I5-750 W/5770 GPU draws about 300 Watts max, My I5-2500k w/6850 GPU draws about 350 Watts max. Both are under 200 Watts at Idle.
I'm using a 540Watt/950VA ratted UPS.

Your system @ idle the battery should last 5 -> 6 mins (when gaming probably 2 - > 4 minutes.

I'm thinking it is more to do with the output waveform than wattage rating. So if you decide to pick-up a new one make SURE it indicates that the output is "NEAR" Sinewave. Only expensive ones will provide a true sinewave

 

[djotaku]

Oh, maybe my UPS is too low of a rating? I have it on a 550 VA UPS.

 

[RetiredChief]

True wattage is lower than V/A rating. Myself I only look at the wattage and don't really bother with the V/A rating.
V/A value only is of imporatance when voltage and current are out of phase by a fair amopunt. Most New pSUs use a Power Correction Factor (PCF) circuit that reduces out of phase conditions - that is also why you need "close to" real sinewave output as the PFC circuit is based on a sinewave.

For Your system I would recommend at least a 500 Watt system. Smaller one will work (ie 400->450) but run time will be shorter and the internal batteries (they are somewhat expensive) to die early.