2 PSU failures in 4 days

[ywycbt]

First, my setup:
AMD Phenom II X4 BE 955 @stock
MSI 790FX-GD70 mobo
HSI HD Radeon 6950 2gb
Intel X-25m SSD 80gig
Samsung Spinpoint F3 500gb hdd
Sony DVD-RW drive
Antec 300 Illusion Case
4gb ripjaw series RAM
First PSU that failed: Corsair TX750w
Second PSU that failed: Corsair TX850w professional series
LG 23 inch monitor
Samsung 24 inch monitor


I moved into a school apartment last thursday, hooked up my computer friday night, and everything seemed to be fine. Sunday after playing some dotas, my computer lost signal to the monitors and I turned it off. On reboot it would post but not load windows, and on third reboot there was no power to anything inside the computer. The green light on the mobo power switch was on, but that was it. I did the paperclip test on the tx750 and the fans would not spin so I sent it in on RMA to corsair (they still have it)

I purchased the 850 professional and figured i would use the RMA as a backup. Part arrived 8/25 installed it, and computer worked fine for 6+ hours of gaming. When I installed the new PSU there was absolutely no signs of anything wrong with it. Computer worked fine and operated at normal temperatures all night. I turned it off and went to bed, then this morning 8/26 I woke up to find the computer without any power and the EXACT same symptoms as the first PSU failure, green mobo light on, no power, failed paper clip test.

Now, I figured there has to either be something inside the computer shorting out the PSU or an electrical problem in the building. I had an electrician come out on 8/24 after the first PSU failure and inspect my apartment and he found nothing unusual at all in the apartment. So that leaves me with an internal problem in the computer that is killing my PSUs. The kicker is, that the second time that this happened, the computer was not even powered on. The computer is hooked up to an industrial rated surge protector, and the other devices on the same surge protector are fine (23 inch monitor and 24 inch monitor.)

Also, when these failures occured the first time, I used an old gaming laptop during 8/22-8/25 and there were no problems with that computer over 15+ hours of gaming. No other problems have occured in any other electrical device in the apartment (Tvs, microwaves, refrigerators)

Please, if anyone has any ideas for what could be causing two extremely highly rated PSUs to die within a span of 4 days let me know, because I am at my wits end here.

Thank you in advance.

 

[mindless1]

Is it fairly warm where you are located? Best/random guess is there's an air conditioner causing power surges when it turns off, or maybe the fridge, and your surge protector simply isn't doing well enough or has failed parts and needs replaced. Random bad luck can also cause repetitive surges below a level a surge protector is designed to handle (so the manufacturer doesn't suffer undue warranty replacement costs) which one PSU design may be more susceptible to than another. I suspect the two PSU you had fail are essentially the same design besides a select few component changes and would suggest trying a different design if you can't find another resolution.

If something in the system is shorting out the PSU that shouldn't kill the PSU, either the PSU would shut off sensing undervoltage condition, or whatever is shorting would have an intense heat buildup, blackening, smoky smell, etc. then a subsystem fault a new PSU won't fix.

 

[ywycbt]

I live in Monterey, CA, it hasnt been above 65 degrees all week, however I do have the heater set to turn on at 70 degrees so it comes on every few hours, and come to think of it the refrigerator in this place makes an loud noise every time it turns off. I know literally nothing about electricity and wiring in houses, but wouldn't those mini surges affect other appliances in my house such as tv, monitors, and other things? How would I test for these mini surges?

 

[Karaktu]

Based on the above thoughts, I would definitely get myself a 600va (or better) battery backup/UPS, preferably with LCD display (so you can see what's going on) and AVR (automatic voltage regulation). They are lifesavers.

http://www.newegg.com/Product/Produc...82E16842102069

 

[Meghan54]

Based on the above thoughts, I would definitely get myself a 600va (or better) battery backup/UPS, preferably with LCD display (so you can see what's going on) and AVR (automatic voltage regulation). They are lifesavers.

http://www.newegg.com/Product/Produc...82E16842102069

Truer words have never been spoken. I won't run a computer in the house without being run off a UPS with AVR. Of course, I live in the country and our power isn't the most stable of things, which is why I use UPS's. Heck, even have one for the TV.

 

[ywycbt]

any suggestions for which UPS to get or what features to look for? I am unfamiliar with these all together, looking at the apc site I don't understand what makes one better than another outside of the wattage.

 

[mrSHEiK124]

any suggestions for which UPS to get or what features to look for? I am unfamiliar with these all together, looking at the apc site I don't understand what makes one better than another outside of the wattage.

http://www.newegg.com/Product/Produc...82E16842101381

There's a good start. Not sure of price (value) but you want AVR! I live in the opposite of the country, and the power surges in my apartment complex are bad enough that they reset my cable box and clocks on my appliances [probably a crap HVAC unit somewhere - seriously, power surges like lightning struck every time my neighbor's unit kicks on]. If I didn't have an APC unit with AVR I'd have already fried my PC.

http://www.newegg.com/Product/Product.aspx?Item=N82E16842101393

 

[westom]

any suggestions for which UPS to get or what features to look for?

A UPS connects your computer directly to AC mains when not in battery backup mode. That is when power is 'cleanest'. This 120 volt sine wave output UPS outputs 200 volt square waves with a spike of up to 270 volts when in battery backup mode. Because protection inside every computer supply is so robust as to make that 'dirtiest' UPS power irrelevant. Computer power supplies are as or more robust than the PSU inside a UPS. So why does that UPS do protection? It doesn't.

Reality is not what advertising says - subjectively. Any subjective recommendation is best treated as a scam. If a UPS does more than battery backup, then UPS manufacturer spec numbers say so. No numbers provided because a UPS does not claim any such functions - except in subjective advertising where lying is legal.

The paper click test does not report anything that could identify a failure. Informed techs would take one minute with a multimeter to have numbers from six wires. Then the actual problem is identified. A solution fixes both defective parts and reason for that damage. Impossible when numbers are not provided.

The OP describes a classic intermittent. A problem that often causes perfectly good parts to be replaced when using a subjective diagnostic technique heavily based in speculation. In CSI, they say, "Follow the evidence." That means numbers.

For example, a perfectly good supply can fail in an otherwise perfectly good computer. A failed power supply can still boot and run a computer for months or years. Both example of why a subjective diagnostic procedure creates confusion.

Same subjective assumptions are why a UPS was recommended. Also unknown to most is how a protector too far from earth ground and too close to computers can make computer damage easier. That anomaly does not apply here. But is another example of how popular myths do not know what actually happens.

That computer's power supply system is more than just a PSU. Numbers from six wires using one minute of labor and a tool from K-mart would have said immediately what of many power components are or are not defective. Meter sells in K-mart because even 12 years olds can use it. Also sells in Harbor Freight, Wal-Mart, or most stores that sell hammers for $5 or $17. About as difficult to use as a hammer or that paper clip. And actually says what is wrong.

Replies are only as useful as facts first provided. Since no numbers are provided, then those who knew this stuff (even long before the IBM PC existed) cannot provide an answer without any doubt. Such definitive answers would exist had the meter reported voltages on six wires. Replies are only as useful as facts provided by the OP. One would be amazed how much information is in numbers from one minute of labor.

 

[ywycbt]

so what tool is this that I need, and do you know how to test it and read it to see why the PSU failed, or is there a site you can steer me towards so I can learn how to do this on my own.

Than you.

 

[mindless1]

^^ hi Tom!

ywycbt, the tool implied was a multimeter. The component most likely fried is the primary side switching transistor(s) or controller. You'd desolder them, get the part # off to make an ID, based on type of transistor then google search for a basic short/open test. If you don't have a multimeter or soldering experience this is probably beyond your skill level and an ineffective use of time and money. I'm all for people learning electronics but it's best to start out with low voltage DC circuit projects, not ~ 170V+ coming from mains AC.

Certain electronics designs have their own weaknesses. Instead of only your PSU(s) failing it could have been only your monitor, or telephone, etc. FWIW, APFC PSU with certain designs are more susceptible to surges, IMO... but I don't have hard data to back that up.

An UPS can be more robust in power surge situations because it (often) has a minimal surge protection circuit in it, and their designs are a little lower tech and time tested/matured - very few PC PSU have any surge protection at all.

Usually power fluctuations that reset gear is a voltage depression or interruption rather than spike, but certainly both conditions could be present in any period of time given the right circumstances.

"Sometimes" you can hear voltage spikes from motors, compressors in a fridge or A/C, etc turning off if you have a poorly filtered speaker/amp set turned on, it'll make a pop sound. Otherwise, you'd need equipment that can log power fluctuations as even a multimeter being watched when a surge happens, may not display peak voltage and be readable by a human in such a short period during which it occurs. "Sometimes" an UPS will log it, depends on the duration and magnitude. I vaguely recall an APC 500W (cheapo) UPS I have, logs almost nothing except complete power loss where it ran for a few seconds or longer... power blimps long enough to reset certain PCs and clocks aren't logged as occurring.

 

[ywycbt]

I figured what he was talking about was way over my head, I would like to meet the 12 year old that does that stuff!

I have ruled out surges, and the electrician thought that it might be undercurrent when the heater was turning on, and he suggested a UPS as well to help stabilize the current. I just get such mixed messages about the UPSes some people saying they work well others saying they dont. To be honest it might even be a short somewhere in the computer, but the UPS is the cheaper of the two solutions for now. If that doesn't work then I will be buying a new mobo/processor/psu/case anyway.

 

[westom]

so what tool is this that I need, and do you know how to test it and read it to see why the PSU failed,

A 3.5 digit multimeter is sold in Lowes, Radio Shack, Wal-Mart, Home Depot, Harbor Freight, and most any store that sells mechanics tools or hammers. Typically $17 in Wal-Mart.

Set the meter to 20 VDC. Attach its black probe to the chassis. Touch its red probe to each wire. Read a three digit number.

Touch the purple wire from the PSU where it enters a nylon connector attached to the motherboard. That should read about 5 volts. Record every numbers to three digits.

Repeat that measurement the green and gray wires; this time both before and when the power button is pressed. Record both numbers and note its power on behavior.

And finally monitor any one red, orange, and yellow wire as the power switch is pressed.

What those numbers report and how a power system works will follow. As I said, even 12 years olds were left to do this.

If have no idea what all that is about driver transistors. None of it is relevant to what I posted or for solving problems.

A properly constructed supply should never suffer component failure. For example, all power supply outputs can be shorted together when power is applied. And that must never cause supply failure. Intel's standard even defined how thick the shorting wire must be - a perfectly normal test for all PSUs without any damage. One example of how robust all supplies must be and why driver transistor failure must not happen.

All supplies are supposed to do be that robust. Show me any UPS is not harmed by any load. The power system in a computer must never by harmed by its load - or harm its load.

Meter numbers should answer what your supply and other system components are actually doing. Even a 12 year old measure as described in the first five paragraphs. Provided is to how to have answers without speculation. Anyone can take those measurements in about one minute.

 

[westom]

... the electrician thought that it might be undercurrent when the heater was turning on,

If undercurrent was causing a problem, then an incandescent bulb on the same circuit must dim to 50% intensity.

If a heater causes any bulb to dim, then that building may have a wiring defect. In rare cases, that wiring defect can also be a human safety threat. IOW fix the defect. Never cure symptoms with a UPS. Especially when that dimming might be reporting a possible human safety threat.

Are your incandescent bulbs dimming? If not, then undercurrent does not exist.

 

[Meghan54]

If undercurrent was causing a problem, then an incandescent bulb on the same circuit must dim to 50% intensity.

If a heater causes any bulb to dim, then that building may have a wiring defect. In rare cases, that wiring defect can also be a human safety threat. IOW fix the defect. Never cure symptoms with a UPS. Especially when that dimming might be reporting a possible human safety threat.

Are your incandescent bulbs dimming? If not, then undercurrent does not exist.

But in some cases where undercurrent exists, and it's not from a heater/AC/etc., but is via the mains, a UPS with AVR can help. I can watch my lights dim, cut off, etc., almost daily. The computer never falters.

 

[ywycbt]

ok, so its not a surge and its not undercurrent...what causes the power supply unit to fail when the computer is off? Everyone else seemed to think it was the current of some sort because, at least to the best of my computer related knowledge, the computer cannot fry a PSU when it is off.

 

[westom]

ok, so its not a surge and its not undercurrent...what causes the power supply unit to fail when the computer is off?

Many reasons. Far more than what others have suggested. Unfortunately, every reply is only speculation. Without hard facts, then nobody can say better.

Apparently your incandescent bulbs are not dimming. A hard fact that eliminates many suspects. Provided was how to have best hard facts. One minute of labor results in three digit numbers from a digital multimeter.

You have assumed the supply is off. Those numbers will expose that speculation. But again, long before identifying any suspects, one must first learn basic concepts. Just another reason why the multimeter was so important. Facts that would have put a 'driver transistor' post into perspective.


Meghan54 - if your lights change intensity, they a wiring problem exists. Either inside the building. Or that is a symptom of a utility problem that might later cause serious damage. In a rare case, even caused a house to explode (because another human safety component inside the house was disconnected).

One never cures those voltage variations with a UPS. Solve the problem. Lights should never vary intensity when any appliance power cycles.

 

[ywycbt]

well I will swing by wal mart tomorrow and get one of the devices. I'll see what results I can get.

 

[ywycbt]

ok I am getting some results.

the red wire from the molex connector is showing .32v
Orange is showing .44v
Yellow is showing .14 off and .22 on
There is no change at all in any of the three wires when the power button is pushed.

purple is reading 5.01v

 

[westom]

ok I am getting some results.
... purple is reading 5.01v

Purple wire is measured when the computer is powered off. If that is what you did, then notice the purple wire is pumping 5 volts (from AC mains) into the motherboard when the computer is off. PSU has two separate supplies. Purple wire is power for a power controller. Apparently that half of the PSU is OK.

Please reread the directions. For example, asked for are voltages from green and gray wires. Both before and when the front panel power switch is pressed.

Same applies to numbers from any red, orange, and yellow wire. For example, does voltage spike up and then drop to a power on number? Does it rise immediately or take more than a second to respond?

 

[ywycbt]

there was no change in any of the readings when the power button is pushed, but again, the computer doesn't power on at all anymore, ill grab the green and grey ones right now, but no wires display any change at all when power button is pushed.

 

[ywycbt]

green wire reads 3.0 then when i press power button goes to 0

grey wire reads 0 in all instances

 

[PreferLinux]

PSU is definitely bad. It is never sending the POWER_OK signal. (If my memory and logic is correct...)

 

[ywycbt]

do any of these measurements help indicate whether the fault was caused my the electricity from the apartment or an internal component in the computer?

 

[ywycbt]

I was messing around with my voltmeter and checking the outlets in my house, and when I plug in a heater to the wall, it will drop the current from 121 volts to 115-118 volts. Is this enough to contribute to the problem?

 

[PreferLinux]

No. Normal is anywhere between 110 and 120 V. But it still shouldn't happen. Did it spike when you disconnected the heater (would be shown by a higher reading, which would immediately drop, but as you probably have a digital meter (a number readout) it wouldn't actually be very readable), and if so, what to?