Solved Is my PSU Dying?

[FerchogtX]

Well... I had an interesting evening today.

Mom was using her PC as she normally does (formerly mine), when suddenly, a savage BSOD appeared, it pointed to atikumg (IIRC), so I assumed the drivers were faulty, so, I unisnstalled drivers, ran AMD driver cleanup, ran Ccleaner and upgraded to the latest drivers that VGA can have. No more BSODS, but system tends to hang completely at random times, which is weird.

This PC was working nicely since 2006... until just today, when it's exhbiting this symptoms just suddenly.

I've cleaned the system, removing any dust, replacing thermal paste on even the VGA itself, I need to test if problem persists tomorrow. System runs cool so far... I'm suspecting the PSU is dying somehow (I don't know how, probably a big ray that impacted nearby two days ago, even with the PC off?)

The VGA works, doesn't smell like burned... nor the PSU, all the rest of the components look good, no bad capacitors on the MoBo or such...

I'm leaving more details here if this helps somehow:
- CPU Temp Idle: 30 - 35 C°
- CPU Temp Full Throttle: 50 C°
- VGA Idle: 38 C°
- VGA Stressed/Full Throttle: 55 C°

The VGA is a Radeon HD 4350... not a gaming card, not even a card that draws a lot of power, the PSU is a PiXXo Transformer 800W 80+ Certification. The rest can be found on my System Spec section.

Is the PSU dying or the VGA? This setup worked like a charm 8+ years or so... this just started to happen suddenly. Ah! This tends to happen whenever she browses the web, keeping Firefox and Thunderbird open at the same time while using Winamp to hear some online music, both have hardware acceleration enabled, Firefox has e10s enabled and H264 decoding available. Checked Even Viewer for some extra info, but nothing was there to help me...

Thanks in advance for your help, guys

 

[Megahertz07]

Lightning discharges can damage a PC even if it's shut ed down. On ATX PSU you have the +5 V SB that is always on.
This power in stand by is also present on TV, Micro wave, washing machine, etc and on any remote controlled device.
To protect these equipment s I have surge protectors with a hardware switch. I have one for the computer +printer +monitor +scanner and another to TV +decoder +DVD +router. When not in use, the switch remains off. Also important is to have the ground grounded.
On your computer the MB could be damage not only by the PS but also by the LAN. I've replaced a MB that was damage by a lighting that came through the cable modem.
It's very difficult to find a intermittent problem. Try to install another PSU. If you have a wired LAN connection to a router or modem, try to disable the LAN on BIOS.

 

[FerchogtX]

Thanks a lot for your answer

Weird stuff though... I've made some tests to stress the GPU and the CPU at the same time, but nothing happened, not even a little BSOD... even recreated all the process, opening winap, Thurderbird, browsing with Firefox... no hangs...

It's... weird... The PC is connected to a multicontact bar, that is connected to a surge protector, so it's never all the time on (meaning the 5v+ SB)... when the lightning happened, all was off... probably the LAN cable is a main suspect in this matter.

I'll see if the symptoms persist... I don't know how to feel about this... just yesterday I just saw BSODs, crashes... I resist to believe that changing thermal paste and removing dust did the miracle?

I'm not closing this thread until I see this system really works as spected...

Thanks for the answer... if there are any ideas, feel free to drop them

 

[ThrashZone]

Hi,
Most people believe hardware acceleration should be disabled where ever possible in browsers.

If you have onboard graphic's have you removed the gpu and tried using it for a little while ?

 

[ignatzatsonic]

You have a no-name PSU that is at least 10 years old.

Is that correct? If it is, the PSU would be pretty high on my suspect list.

That's a low power system and I'd think a PSU of 400 or 500 watts would be plenty.

Do you show any voltage aberrations when using a monitoring program such as HWInfo?

I'm guessing you don't have any sophisticated tools to check the power supply.

Do you have another PSU you can swap in briefly?

 

[ICIT2LOL]

My two cents worth try my ditty to check the PSU volts for example.
One word of warning though don't go trying to to do anything with the PSU (dismantling to look) because it can have lethal voltages stored in the large inverter caps even after an extended length of time of power off.
Using HW Info
PART A:
You can test the volts on the PSU with HW Info HWiNFO, HWiNFO32/64 - Download < download the right bit version and close the right hand window select Sensors and scroll down to the power section where you will see what the volts are doing see my pic. In my pic the section (Nuvoton) with VBATT as a dead give away you are in the section for the rail voltages. There are other section titles and one that pops up often is ITE (sometimes the usual one for Gigabyte boards)
Now the voltage on the different rails have to be within 5% =+/- of what is required or the machine will not work properly if at all.
Limits +/-
12v = 11.4 – 12.6v
5v = 4.74 – 5.25v
3.3v = 3.135 – 3.465v
The Power good signal voltage at pin 8 on the 24 pin plug (grey cable) should be the same as the 5v rail reading/s
See this for the rail voltage info
PSUs 101: A Detailed Look Into Power Supplies (Section 2.)
The original right hand window shows the machine running and is handy for that but for looking at the components in some detail close it and use the main left hand side panel
FOR OTHER COMPONENTS
PART B:
Open each small square with + in it on the section the components are in and then click on the individual component/s (it will highlight in blue) - in the right hand side will appear all sorts of details including brands speeds and other essential info that particular device. See pic for example.

Use PART A:

 

[westom]

It's... weird... The PC is connected to a multicontact bar, that is connected to a surge protector, so it's never all the time on (meaning the 5v+ SB)... when the lightning happened, all was off... probably the LAN cable is a main suspect in this matter.

None of those items even claim to protect from a surge that might do damage. Furthermore, if a surge was both incoming and outgoing from that computer, then a failure is obvious - not speculated. And yes, a surge can be incoming to everything. But only damages items that also have an outgoing path to earth. IOW damage would be on the outgoing path - not a video controller.

Second, you had numbers in that BSOD. Those provided the most important fact. Why are you suspecting all kind of irrelevant suspects (ie thermal compound that never needs changing, DC voltages, dust) when no facts even point in those directions?

Third, better computer manufacturer provide comprehensive hardware diagnostics for free. Those made available so that strange intermittents can be defined. Obviously you do not even have that best diagnostic tool.

Four, heat is a diagnostic tool - not something to be feared. Best way to find an intermittent is to put a computer in a 100 degree F room and run comprehensive diagnostics. A completely defective part will work fine in 70 degrees F and fail at above 90. That same part may start failing month or years later at 70 degrees. Heat at those near zero temperatures does not harm hardware. But does find defective parts.

But again, no useful answers are possible when you did not even record four critical numbers on that BSOD. Maybe use heat to recreate a problem. And please ignore so many ill informed naysayers who foolishly assume thermal compound must be replaced anytime in 30 years.

 

[ICIT2LOL]

Did you run the HW Info surges are irrelevant if the problem lies within any of the internal components of the PSU ie especially caps for example which if they become non functional will cause problems and in most cases failures.

 

[FerchogtX]

When I said I was resisting to believe that only changng the thermal paste and removing dust fixed the problem like magic, I was trying to mean that, unless you have a very OLD thing (kinda... 6 years of no changing thermal compound or cleaning dust), it's like a magical solution taking in mind the syptoms I was experiencing in that PC XD...

This obviously doesn't mean by any chance that not changing thermal compund is a good practice AT ALL... In fact, I have a Cooler Master compound (400 something, I forgot the name) and I like to change thermal paste at least twice a year (ever 6 months if you like) and removing dust, since those two are part of the potential enemies a computer have and CAN lead to big problems.

Going back to HWinfo, I'll get the chance to run it on that PC as soon as possible... I've run AIDA64 and gives similar numbers in 12+ and 5+ rails, but probably it's way better to check with another software to be sure

I appreciate all the help you are giving me guys, thansk a lot.

I hope I can get back to you soon

 

[Megahertz07]

Sorry to say, but you are wasting your thermal compound. As air, or any other gas, is a terrible thermal conductor, thermal compound are used to remove all air between the CPU and the heat sink. The thin it is, the better is the heat conduction between CPU and the heat sink. Once installed some fluids of the compound will evaporate and any space between the CPU and the heat sink will be filled with an dry compound (silver oxide). If you don't remove the heat sink, there is no way to air get between them.
As I wrote before, as the power supply is always on to the MB (+5 V SB (Stand By)), the only effective way to protect your computer from discharges is by disconnecting using a hardware switch, that almost all suppressors has.

 

[ICIT2LOL]

Sorry to say, but you are wasting your thermal compound. As air, or any other gas, is a terrible thermal conductor, thermal compound are used to remove all air between the CPU and the heat sink. The thin it is, the better is the heat conduction between CPU and the heat sink

So true the compound is intended to fill in the microsocopic depressions and scratches left by the manufacturing surface.

My own personal way of applying it is to clean the two surfaces then prepare with the Arctic Silver products here (a cleaner and a preparation solution) Arctic Silver Incorporated - ArctiClean then apply the Arctic Silver compound Arctic Silver Incorporated - Arctic Silver 5 with my index finger covered with a vinyl glove so that the compound is slightly warmed and thinly and evenly spread onto each surface and into the microscopic faults using a circular motion this avoids contaminating the surfaces with anything from a spreader or bare fingers - then apply a rice grain sized spot of Arctic Silver compound in the centre of the CPU surface and assemble the cooling system. The heat generated from the CPU powering up then "bleeds" that spot of compound across the two surfaces.

Have used this method for years and have had no issues. The Arctic Silver stuff may not be the cheapest but I believe it to be the best because of the silver content which is the most efficient metal at conducting thermal energy.

 

[westom]

Let's discuss what thermal compound does. Most heat transfers through the lowest thermal resistance - direct 'CPU to heatsink' interface. Some microscopic holes (air) exist. Thermal compound fills microscopic holes and must not interfere with a 'CPU to heatsink' interface. That means using least amount of thermal compound. Less is always better than more.

Thermal compound has higher thermal resistance. But is more conductive than air. Once that compound is inside those holes, then heat transfers through thermal compound - not air. That compound (and its thermal resistance) does not change even decades later.

Meanwhile, if thermal compound leaks out to edges, then it can get on electrical conductors. That can cause electrical problems; intermittent electrical failures. Just another reason why replacing thermal compound is ill advised. And why thermal compound should never spread out to the edges of a CPU or heatsink.

The naive fear dry thermal compound. Wet or dry, that compound still is making the same connection from CPU to heatsink in microscopic holes. Wet or dry makes no difference.

If a heatsink is properly designed and selected using a 'degree C per watt' parameter, then heatsink alone (no thermal compound) will conduct sufficient heat. Adding thermal compound will lower CPU temperatures by single digit degrees. Just another reason why so much 'hype' associated with thermal compound is bogus.

If anyone says something different, then he must include parameters - numbers. No specifications is a first indication that a recommendation has no credibility.

BTW Arctic Silver is no better and may be worse than thermal compounds from heatsink manufacturers. Arctic Silver is only recommended because feelings created by advertising causes consumers to not first learn spec numbers. Best thermal compound comes from other respected manufacturers who make heatsinks, have been doing this stuff for generations, and also make same even for military products. These include names such as Aavid and Wakefield.

So what is the CPU? AMD or Intel?

 

[Megahertz07]

When I get a new heat sink, I polish it till it's reflecting as a mirror. I fist use a 600 water sand paper and then a polishing paste. That insure that all the microsocopic depressions and scratches left by the manufacturing surface are eliminated.

 

[ICIT2LOL]

Yes they call that lapping but it still will not get the tiniest of imperfections out of any metal surface. On the issue of Arctic Silver it is as good as any other and from numerous discussions on this site about compounds the result is there is no ideal one. There has never been any agreement on this topic since I have been a member and I have read of some truly peculiar ways of applying compound. There are those really expensive metallic compounds that can be used only once but out of my price range I am afraid.

One has to remember that the heat produced by the silicon chip is transferred through a much smaller surface area to the heat spreader and thereby it is down to the transferring properties of that heat spreader to dissipate that heat to the cooler surface. I do have many older CPU's I have taken out that have pure copper heat spreaders but can you imagine for one minute they will revert back to that material when they can produce heat spreaders that have a far less efficient transfer rated material because of the cost??

Again this is a very well worn topic and there will never be any agreement on what is best and what is not.

 

[westom]

Lapping, as noted, will not get the tiniest imperfections out. Worse, if you flatten that surface, then thermal resistance across a 'CPU to heatsink' interface will increase. Most all heat transfers at the center of the CPU.

There will never be agreement because so many need subjective (ambiguous) answers. So many know by ignoring all spec numbers. Separate unreliable answers from the informed using one simple rule. The only inforrmed (honest) answers provide perspective - provide numbers. For example the conductivity of that thermal compound must be provided. Thermal resistance is discussed. The heatsink's degree C per watt number is known. Heatsink must have been machined so that thermal compound only results in single digit degree temperature reduction. Without numbers, then most anything can be claimed.

Is it an AMD or Intel CPU?

 

[Layback Bear]

Of course their are many compounds used and many methods used also. Their are all kinds of test done and charts and graphs all over the net.
Thorough the years many of us have tried many compounds and methods.

Myself, I have never found much temp difference in compounds if they are applied properly.
All lapping did for me was get my 4 cores to run temps closer to each other. Which is good.
No I did not keep all the specs on the changes so I could quote them here.

For me and I believe for 99.9% of users this will would as good as any and is what I use.

Arctic Silver® 5

This little tutorial method will work just great.
Just pick what brand of cpu you have and then which exact cpu you are using.

Arctic Silver Incorporated - Route to Product Instructions for Arctic Silver 5

If done correctly and with the proper mounting of a proper working cooler you still have a heat problem, it lies somewhere else.

The system in (My System Specs) get the compound changed about once a year because that is when I flush the cooling system. Can't hurt.

My older system with a AIO liquid cooler I change the compound about ever two years.
But not because it is running with a temp problem. It's more like a what the hell, why not. Can't hurt.

If I found a new compound that worked a lot better than Arctic Silver® 5 I would use it. The hell with the cost.
With 14 fans, 2 radiators and 2 pumps what's another few dollars for a better compound. Compound is about the cheapest thing one can buy to improve cooling.

I also use Arctic Silver 5 on my air cooled video card in my older system.

 

[ICIT2LOL]

Yep Layback as we both know this topic has been a hotly contested issue for many years and frankly is as monotonous as it always has been.

Like you I I have stuck with Arctic Silver using my method with no problems and in fact I have used just about all the brands there are except the melt in metallic one and they have included such things as diamond based and ceramic based compounds and as you and I know a lot of problems arise when people get sloppy when applying compound or simply have no idea of just what to do.

The other thing that comes to mind is of course a lot of the cooling can depend on how well the silicon chip is in contact inside the lid of the CPU (heatsink) and again there have been a lot of those who have tried de -lidding the CPU to renew the compound between the chip and the underside of that heatsink the manufacturer Intel or AMD have installed themselves and I have yet to hear that that method has made any appreciable improvement in the cooling of the CPU.

In fact gain the cooling has more to do with than just the compound such as the type of liquid used in many those liquid systems and the efficiency of the cooling rad in air cooled systems which usually being made of aluminium means that the heat transfer is just not that good. See this chart for silver copper and aluminium for example.
Thermal Conductivity of Metals

 

[FerchogtX]

Well... checked voltage numbers in HWinfo... everything seems normal... no abnormal numbers or anything according to the program and the examples you give me.

Also, there has been 6 days since the PC was cleaned and changed the thermal compund... nothing, nada... not even a tiny hiccup... I swear to God te weirdest things happen to me XD... I'll continue montoring the computer...

Thanks a lot for all your help, at least, it seems the PSU is not dying

 

[westom]

Well... checked voltage numbers in HWinfo... everything seems normal... no abnormal numbers or anything according to the program and the examples you give me.

Never short your help of information. What number do you call normal. HWinfo does not typically measure all voltages that can cause your symptoms. HWInfo is also dependent on a voltmeter that is not yet calibrated. All three are why those voltages must be measured with a meter. And why useful replies are only possible when actual numbers are posted.

Your numbers can meet what Icit2lol posted and still be reporting an obvious defect. He also did not list all relevant voltage numbers. Behavior of those voltages on power up is also relevant.

 

[ICIT2LOL]

Yes I know westom but there is more info in that link with the HW Info stuff and yes I always recheck with a digital meter if I am still suspect of what HW is showing me - I just find it a really quick and useful tool and I think if you go through the components listings there is even more info than your average rocket scientist could understand - I know it is not the be all and end all.