PDA

View Full Version : Power surge troubleshooting



JMoore
16-10-2012, 07:48 PM
Right, so briefly (but foolishly) removed the surge protector from my PC setup and as Murphy's Law would have it, a power cut occurred that night. The PC now "turns on" when you flick the PSU switch (doesn't require the case switch to be pressed) and runs all fans etc but there is no display output to monitor and I suspect there may be an issue with the motherboard as it isn't beeping when powered up.

Any suggestions on how to troubleshoot this?

Thanks in advance.

wainuitech
16-10-2012, 08:46 PM
Try this --- Unplug the power cord completely from either the wall or from back of the PC (Don't just turn it off) leave it out for a while, say 5-10 minutes, then plug it back in, see if it boots up or not.

Terry Porritt
16-10-2012, 09:01 PM
Be prepared for the 'westom' troll to turn up, the words 'power surge' act like a magnet for his rantings.

wainuitech
16-10-2012, 09:42 PM
Be prepared for the 'westom' troll to turn up, the words 'power surge' act like a magnet for his rantings. LOL I Was thinking that same thing Terry -- have the fly swat at the ready :D

JMoore
16-10-2012, 09:46 PM
No luck with disconnecting the PSU. Still turns on fans and some lights but no beep from motherboard and no video output...
Have tested PSU voltages and 5V and 12V rails are fine.

Billy T
16-10-2012, 10:06 PM
Right, so briefly (but foolishly) removed the surge protector from my PC setup and as Murphy's Law would have it, a power cut occurred that night.

Don't sweat it, a surge protector is pretty useless for power cuts because the time duration of a surge is many milliseconds long, in fact they are more of a spike suppressor and use just one or two MOVs (metal oxide varistors). It would be the disturbance that accompanied the outage that caused the problem and that may have consisted of several large and wide 'spikes' in quick succession complete with major voltage variations.

Best way to check the value of a surge suppressor is to drop it on your foot. If you end up hopping on one leg and telling God all about it, you have a useful device, but if all you get is a tap on the toe it is not worth plugging it back in. I custom-build special spike/surge suppressors for industry that will knock several hundred volts down to a few 10's of millivolts and do that for years. Drop one of those on your foot and you'll need to sit down pretty quickly.

Cheers

Billy 8-{)

JMoore
17-10-2012, 12:08 AM
Haha cheers for that, maybe Murphy wasn't in play after all - but still doesn't help my lifeless computer unfortunately!

Any other suggestions would be greatly appreciated!

westom
17-10-2012, 02:57 AM
Have tested PSU voltages and 5V and 12V rails are fine.
Tested the 5 and 12 volt with what? Important are voltages on six wires using a multimeter. Do you have or can borrow that meter?

CliveM
17-10-2012, 07:36 AM
He is back!!!

wainuitech
17-10-2012, 08:00 AM
JMoore just to eliminate components, do you have another Power Supply to plug in and see if the same thing happens.

dugimodo
17-10-2012, 09:17 AM
Did it beep before this happened? not all cases have speakers these days.
Normally with a power cut my first suspect is corrupted files on the hard drive as that's the most common result, however with it failing to boot it's likely that it is hardware related this time. Perhaps try Wainui's first suggestion again but remove the BIOS battery as well, going for as hard a reset as possible.

If that fails unplug all your Drives and see if it will try to boot, if that makes no difference you are down to PSU, RAM, Graphics, or Motherboard as the next suspects.
If it works plug them back in one by one to Identify the culprit.
If you have onboard graphics you can remove the graphics card (if you have one of course) to see if it will boot just using onboard.

Unfortunately it's almost impossible to diagnose these faults without spare components to swap with or some good instruments and a working knowledge of the electronics involved. You can end up buying new parts one by one trying to fix it, it's probably best after the obvious tests are done to take it in to someone to look at.

It's actually unusual for a power cut to cause damage to components, they normally survive it just fine with the exception of possible data corruption. Your surge arrestor is unlikely to have helped, they have a very specific set of circumstances in which they help and in my opinion are of not much value anywhere that has fairly reliable power. If you want to protect from power cuts you need a UPS, surges are actually pretty uncommon.

Just a further comment, irrelevant if you wish to skip it. Lightning is one source of surges, your average domestic surge arrestor doesn't have a chance in hell of protecting from that. In roadside telecommunications cabinets they use quality surge arrestors to protect from lightning, mainly induced voltages because a direct strike is basically unstoppable. The earth wire connected to ground has to run as straight and vertical as possible beacuse lightning has so much energy some of it will take the shortest path to ground and leap out of any corners or horizontal runs of cable and arc directly to ground. These devices have to be replaced after a decent strike and are much better than cheapo home arrestors.

westom
17-10-2012, 10:03 AM
Unfortunately it's almost impossible to diagnose these faults without spare components to swap with or some good instruments and a working knowledge of the electronics involved. One minute labor (touching wires and reading a number) with a multimeter means others who know this stuff can say what is or is not defective.

A power 'system' involves many components. PSU is only one. If nothing is disconnected and the meter provides 3 digit numbers, then the next reply identifies or exonerates every suspect - without doubt. No more 'it could be' replies. Numbers from six wires are critically important. Otherwise, a reply can only be speculation - it might be this or try that.

Best is to not even swap the PSU - do not disconnect anything - to have a reliable answer.

pctek
17-10-2012, 10:14 AM
Best is to not even swap the PSU

Nonsense, swapping a PSU is fast and easy. Quick way to see if it is the issue.

Chilling_Silence
17-10-2012, 10:55 AM
Nonsense, swapping a PSU is fast and easy. Quick way to see if it is the issue.

I agree, things like this are infinitely easier to test by "Swapping out with good known parts".

In cases like this, I'd hedge my bets on it either being the motherboard or the PSU as pctek mentioned.

dugimodo
17-10-2012, 10:57 AM
After this one I've decided to not respond to Westom, his replies are generally not helpful and take the thread off on a tangent.

For a start a multimeter has been used and the voltages appear normal.
Secondly a multimeter does not tell you if a PSU is functioning correctly, it doesn't reliably show up ripple or fluctuating voltages or show what happens when power is apllied under load. "the next reply identifies or exonerates every suspect" - utter bollox, a multimeter reading tells you exactly what component is preventing the system from booting? on what planet ? it doesn't even prove the PSU other than the fact it does have an output.
Finally swapping out components is long established as the quickest and easiest way to eliminate possible causes of a fault, any other method is indeterminate and can take far too long.

Never mind Opinions, make useful suggestions about how to diagnose the problem and not just vague statements about 3 digit numbers or stay out of it. People are here asking for help, not for you to advance your theories on PC power systems.

Alex B
17-10-2012, 11:11 AM
A Multimeter is a good test but does not load test so doesn't tell the whole story. Still sounds like a rail is out on the PSU to me.

westom
17-10-2012, 11:42 AM
For a start a multimeter has been used and the voltages appear normal.
Secondly a multimeter does not tell you if a PSU is functioning correctly, it doesn't reliably show up ripple or fluctuating voltages or show what happens when power is apllied under load. It does as long as the PSU and nothing else is disconnected. Once disconnected, then even the meter cannot report useful numbers.

Numbers from that meter would identify ripple or fluctuating voltages. It would not say which is the problem. But any of those defects would appear as defective numbers. Many numbers, within ATX specs, may also indicate a defect. Numbers from four other wires are also important.

The meter can even identify reasons for intermittents. But this is a hard failure. Easily identified by a meter.

We (who even designed power supplies when switchers were a new technology) know how to use a meter to identify so many defects including ripple. Simply measure those six relevant wires - one minute of labor. Apparently you will not. But others are invited to learn. How to even identify a defective supply in a computer that has been booting for months. Numbers from a meter may even do that.

Statements are vague because it would take many posts or days to describe what those numbers are reporting. I am not going to teach engineering concepts. Provide numbers to have an answer without speculation in the next reply. You don't know why those three digit numbers are so informative. I do.

Alex B described one possible reason. A defective rail could even be the 5 or 12 volts that was assumed good. Cannot say without those three digit numbers.

Chilling_Silence
17-10-2012, 11:51 AM
Statements are vague because it would take many posts or days to describe what those numbers are reporting. I am not going to teach engineering concepts.
Cool so we'll leave the multimeter and other things to those who have engineering degrees. The rest of us can settle for swapping out the power supply / motherboard.

Sorted.

westom
17-10-2012, 12:16 PM
Cool so we'll leave the multimeter and other things to those who have engineering degrees. We give the meter to techs. They learn by using the meter. Also learn why failures happen. And how to avert a failure before it might happen.

Two choices exist. Get numbers to have a solution using less money and labor. And learn. Or shotgun. Keep replacing good parts until something works. And learn nothing from the experience. Those were always the two choices.

Even auto mechanics who do shotgunning can find themselves quickly unemployeed.

JMoore asked "Any other suggestions would be greatly appreciated!" Provided is the other solution. Start by ignoring many naysayers who offer no solutions. Who really do not understand the underlying concepts. Who only suggest "it might be this or try that". One minute with a multimeter, a simple procedure, and numbers from six wires is the other and more often successful solution. And the only solution that also provides knowledge from the experience.

1101
17-10-2012, 12:32 PM
Wouldnt any "ripple' be at the switching frequency & therefore in no way measurable with std Multimeters ?
Standard Multimeters arnt really usefull for checking for noise & spikes on the rails . Thats what scopes are for.

The easiest/quickest test for a PSU is to swap it for a good one. That is real world PC repair.
Complete waste of time trying to fault test a PSU, unless you intend to pull it apart & repair it.

Also a very high possibilty that its a motherboard fault , not the PSU at all.
Some old faulty motherboards simply wont POST (turn on) after being turned off (power cut), nothing to do with PSU's
Look for bulging,leaking caps .
Pull the ram out & reseat it. Reseat the vid car (if applicable)
Disconnect all cables such as IDE/SATA/USB etc
When you turn the PC on, do any of the keyboard lights flash ?? does the numlock button work(LED will go on & off)

wainuitech
17-10-2012, 12:33 PM
AND of course you forget Westom, that the average person who comes on these forums asking for help is of course going to have all the equipment, Multimeter's, and knowledge/skills to diagnose the problems.The reasons people come here is to ask for help when they are unsure or dont know.

Remember where you're reading and making comments - you're not in some technicians workshop or class teaching.Maybe if a faulty PC was taken to a Computer repairer, they would do as you suggest, then again maybe not.

westom
17-10-2012, 12:59 PM
... the average person who comes on these forums asking for help is of course going to have all the equipment, A multimeter costs about $12 or 7. Anyone who cannot use a meter has no business even trying to use a cell phone. Meter is a perfect example of a tool only for layman.

Denials can only exist when the naysayer is without basic knowledge. If you cannot use a meter, then give it to a 13 year old. Any science student would (should) use one in science lab. That cheap and that simple. A layman's tool sometimes even sold in grocery stores.

Some would, instead, spend $60 on another PSU using only speculation.

OP asked for another solution: the meter. More complex is a microwave oven. Even auto mechanics must use one in their first year of training. Why do some find this so vague? Many who do not even know what a meter does have invented fears. OP can have a useful answer immediately after asking for directions and posting the resulting numbers.

A widespread fear of what so many do not know of is telling.

Terry Porritt
17-10-2012, 01:10 PM
I did warn you all..........do not feed the troll............next step:ban

Mind you, maybe my 6 digit HP 3468A multimeter may give twice the diagnostics of a crummy old $12 3 digit job :thumbs:

1101
17-10-2012, 01:30 PM
A multimeter costs about $12 or 7. Anyone who cannot use a meter has no business even trying to use a cell phone. Meter is a perfect example of a tool only for layman.



"Anyone who cannot use a meter has no business even trying to use a cell phone"
You are a conceited idiot to even suggest this

anyone who doesnt know the limitations of cheap meters shouldnt be preaching their virtues .
MOST laymen DONT know how to use a meter, simple fact. They really dont know the difference between volts, amps & resistance

It seems you have no real world experience in PC repair.

wainuitech
17-10-2012, 01:43 PM
Denials can only exist when the naysayer is without basic knowledge.

Even auto mechanics must use one in their first year of training. Why do some find this so vague? Many who do not even know what a meter does have invented fears.
A widespread fear of what so many do not know of is telling. Ever thought the average Joe in the street is NOT an expert or have egocentric views at everything power related.

Said it yourself very first line.

Chilling_Silence
17-10-2012, 02:43 PM
@westom,
Users don't *care* about that kind of thing. In fact I'd argue over 99% of people in the world don't care about using a multimeter. Why should they care, just because it's the industry you specialize in?
I could say the same about the internet, that you shouldn't be posting online via ADSL without a knowledge of how NAT works, TCP vs UDP, and how to do a packet capture. It's madness!

With that in mind if you don't stop the trolling of this thread, I'll be forced to bring out the ban-hammer. Please don't make me do it.

linw
17-10-2012, 02:44 PM
Why, oh why, do you respond? You know you can't reason him out of his views.

Terry Porritt
17-10-2012, 03:01 PM
Just for the record, if wes-troll had used a fraction of his verbiage to actually explain how to measure the PSU voltage in situ, then his post would have been worthwhile.
For example he could have referenced 'Upgrading and Repairing Pc's' by Scott Mueller

http://books.google.co.nz/books?id=E1p2FDL7P5QC&printsec=frontcover&dq=upgrading+and+repairing+pc%27s&source=bl&ots=M2pgB981iy&sig=RttlrGxKxkQ17esPguB1KR7R_MI&hl=en&sa=X&ei=YAp-UILPL5GhiAfnzoHYAg&ved=0CDwQ6AEwAQ#v=onepage&q=measuring%20voltage&f=false and go to page 1168 on using multimeters and page 1170....Measuring Voltage.

Although this ebook is a little dated it would have given the neccessary info on how to do the job.

I've skimmed a lot of his posts on lots of forums but he never actually gives proper answers, like he hasn't here.

PENTIUM
17-10-2012, 05:53 PM
I happen to be an engineer, but I can fix P Cs too, my money is on a stuffed motherboard, .

JMoore
17-10-2012, 11:45 PM
OK here's an update that hopefully answers some of the questions.

5V and 12V were tested within tolerance with a multimeter as these were the two I could remember (red/yellow) and I made a bit of an assumption that PSU was (reasonably) OK when they measured OK.
I also briefly swapped out the power to the board with another PSU and got the same result which is why I think it's the mobo.

Can't remember whether it beeped before, it's actually my parents computer. Although the mobo (GA-K8NF-9) manual suggests there should be one beep for successful POST. Haven't tried BIOS battery but from what I read I would at least get a boot if this was the problem.

Have swapped drives but no luck.

Have reseated and swapped ram - no luck.

No onboard graphics unfortunately.

We do actually have a UPS and surge protector but both were removed from the system when it happened.

Am currently sourcing a replacement mobo unless anyone can think of anything else? Really appreciate the feedback - I can use a multimeter ;)

westom
18-10-2012, 02:47 AM
Really appreciate the feedback - I can use a multimeter Important are numbers from six wires - not just the red and yellow. And important are numbers to three significant digits. For example, specs say the 5 volts must be above 4.75. But specs do not say everything. With other facts, 4.8 volts would be reporting an obviously defective PSU. What are three digit numbers on purple, red, orange, yellow, green, and gray wires? Also what are voltages on green and gray wires both before and as the switch is pressed?

If concerned about a CMOS battery, then also measure it with the meter. Again, a number that must be three digits.

wainuitech
18-10-2012, 08:14 AM
Am currently sourcing a replacement mobo unless anyone can think of anything else? Really appreciate the feedback There has been the odd time that for some reason the Motherboard is shorting on the case somewhere.

You could try pulling the board out and firing it up out of the case, see if it still wont go.

I've got Two new older Boards here that do the exact thing, on the bench they work perfectly, but put them in a case and they wont boot. I know where the problem is, its actually the face plate at the back shorting somewhere, because if I leave that out of the case they once again boot fine.