Start, ya bastard!

January 15, 2011 3:53 PM

Sometimes my PC turns on, sometimes it doesn't, and sometimes it turns itself off with no rhyme or reason. Why, and how do I fix it?

When I turn it on for the first time in the morning, I hit the power button, it makes the familiar 'whirring' noise, the lights on the front of the case come on... then it all stops. No beep to say that it's starting up. Just dead.

If I then unplug the cord from the wall, plug it in again and push the start button, it usually starts up fine. This morning, however, I had to unplug and re-plug 3 times before it would fire up.

Googling has given me the idea that it's either the power supply unit or motherboard. After a previous Askme question taught me the basics of computer 'guts', I'm confident to replace either part.

How do I determine which part needs replacing, and how do I choose a reliable part without breaking the bank? Ebay's power supply options, for example, are never-ending, and I don't know what to look for and what to avoid.

posted by malibustacey9999 to computers & internet (28 answers total)

Note that I'm not the authority on this.

I would try a new power supply unit first (probably free if under warranty) and see how it works during the next week.

If everything is fine, then that's great.

If not, then I would find out the cost of a new motherboard. If it's still under warranty, awesome! Go get it replaced. If not then it's a decision you have to make depending on the price as well as the age of the laptop.
posted by antgly at 4:04 PM on January 15, 2011

a bad memory chip can do this too
posted by patnok at 4:11 PM on January 15, 2011 [1 favorite]

This sounds like it may be an auto-shutdown trigger caused by the processor overheating. Is the tower located somewhere that the fan can't get air fast/well enough? Is it particularly dusty in your tower? Do you smoke in the same room as your PC?
posted by griphus at 4:18 PM on January 15, 2011

Have you added anything new to your computer recently (video card, hard drive, etc).?
If you haven't, a power supply problem doesn't seem too likely unless your power supply is pretty old.

Have you tried cleaning out/dusting the inside of your case? It could be an overheating/ventilation problem.
posted by Elachim at 4:20 PM on January 15, 2011

Oh, and the dead power supply option is valid as well but do not replace it on your own if you don't know what you are doing.

If you do decide to replace it, I suggest you burn an AskMe on what you should replace it with.
posted by griphus at 4:21 PM on January 15, 2011

Power supply or overheating would be my guesses; bad memory would usually involved beeping and complaining.

I would grab a good, logging hardware monitor and check to see if your voltages or temperatures are way out of whack. I haven't used HWMonitor, but I have used SpeedFan in the past to diagnose and resolve this sort of problem.

If it is a power supply the main things you'll want to look for are to compare the power requirements for your motherboard, CPU, and graphics card and compare them to what the PSU provides; the graphics card is particularly critical, because they can have quite specific requirements - e.g. not just "minimum of 500W", but "minimum of 500W and 3 amps via +5V..." and so on.
posted by rodgerd at 4:25 PM on January 15, 2011

Yeah, SpeedFan is a good idea. If you install it and tell us what temperatures your PC is running at (each component will have its own separate temperature) we can narrow down the problem further.
posted by griphus at 4:28 PM on January 15, 2011

Do you have an Asus Motherboard? Look here.
posted by dougrayrankin at 4:32 PM on January 15, 2011

I had a PC for 9 years (finally just passed away last month) and it went through phases like this.
1st time, it was the power supply. Which was not expensive and pretty easy to replace.
2nd time, it was the fan. Opened up and took canned air to it. Everything was clogged with dust.
3rd time, slightly different - I was getting the blue screen and it was shutting off and restarting by itself. Ended up being a bad network card. Replaced it, worked fine afterward.

I would suggest opening the computer, cleaning out any dust.
If that doesn't change anything, order a new power supply and seeing if that fixes the problem.

I believe I probably ordered my power supply from newegg.com or amazon.com.
Also in regards to figuring out which one to purchase, there should be a sticker on your power supply with that information. Or, you can just google the make and model + power supply and see what comes up.
posted by KogeLiz at 5:12 PM on January 15, 2011

If it's not the power supply, check for leaking capacitors on the motherboard. How old is the computer? This was a common problem with computers from 5 or 6 years ago.
posted by MiG at 5:38 PM on January 15, 2011

Check this out:

http://en.wikipedia.org/wiki/Capacitor_plague

Especially the part about the symptoms:

• Not turning on all the time; having to hit reset or try turning the computer on again
• Instabilities (hangs, BSODs, kernel panics, etc.), especially when symptoms get progressively more frequent over time
• Memory errors, especially ones that get more frequent with time
• Spontaneous reboots
• In case of on-board video cards, unstable image in some video modes
• Failing to complete the POST, or rebooting before it is completed
• Never starting the POST; fans spin but the system appears dead

posted by MiG at 5:40 PM on January 15, 2011

Ran Speedfan, got these readings:

GPU 64C
System 39C
CPU 66C
AUX 51C
HDO 43C
Core 57C

The System reading had a downwards blue arrow next to it, the HDO reading had a tick... but the other 4 had fire icons. Am I correct to assume that's a sign of overheating?

Oh, and the CPUO Fan reading is 2909RPM.

I'm a lousy housekeeper so will remove the case, and will clean/dust as required. The outsides of the vents are vacuumed regularly (well, as regularly as the floor is vacuumed).

It shut down about 10 minutes ago while my daughter was IM chatting to a friend (no high intensity gaming or anything going on), but with the added fun of an loud siren-sounding beep. High tone, low tone, high tone, low tone, over and over with perhaps less than half a second gap between each beep. The screen froze, the beeping started, and I had to unplug it to shut it down. Started up normally 5 minutes later.

Extra info that may or may not be relevant: it's hot here today, and the room with the PC is starting to heat up noticeably. There has been a red light on inside the case for, oh, months but we haven't had problems til the last week or so. It's out of warranty, I bought it second-hand maybe 18 months ago from a play-online-games-on-our-premises-for-$X-an-hour business that was closing down. It was WAY cheap, came with a tonne of games installed (and heaps of other un-needed software which I removed). I bought it just as much as an opportunity for me to learn about how they work as well as a functional PC.

Your help is very much appreciated.
posted by malibustacey9999 at 5:53 PM on January 15, 2011

Your CPU is running a bit warm, but not enough for it to trigger a shutdown, which is what it sounds like is happening anyway. Now, I've found that SpeedFan tends to be a bit of a worrywort give you overheat warnings way too soon; unless my AC is on, no component my PC has never run under ~55C, and I get good mileage out of them.

My guess? Either your CPU has a suspiciously low auto-shutdown trigger or you're having a temperature spikes for some reason. Did you open it up to get the dust out?

Also, I rarely suggest this, but operate your computer at eye-level where you can watch the fan. See if it shuts off at any point, or if anything else odd happens before you get the beeps and so on.
posted by griphus at 6:02 PM on January 15, 2011

...has ever run...
posted by griphus at 6:02 PM on January 15, 2011

Sigh. Okay, didn't read the part where you didn't dust it out yet. Do that. I suggest you use compressed air, but if it is really dusty in there, give it a very careful vacuuming first, then dust it with compressed air. Pay special attention to the fans, making sure they can turn fine. Also, if there's a spiky metal thing on top of the processor, give that a good dusting as well.

And when I said "operate at eye-level," I meant do it with the case open.
posted by griphus at 6:05 PM on January 15, 2011

Very gentle vacuuming done (it's Sunday afternoon in a small country town, I don't think I could buy compressed air anywhere today if my life depended on it). It wasn't too bad, I'd vacuumed a while ago when Askme taught me how to add an old hard drive. A fair bit of dust did fly out as the fans rotated when I held the vacuum in front of them.

And I must confess I found a small air vent on the right-hand side which has always been up against the wall of the computer desk. I had forgotten that vent was there. Doh. I've moved the machine so that vent too can get fresh air.

Reassembled, started up fine, and I suspect the fan noise is a little more quiet than it was before.

Fingers crossed the dust (and perhaps the blocked vent) was the problem. If symptoms persist, I'll be following more of your advice. It'll be a few days til I mark this resolved, it's a case of 'wait and see', I reckon.

Thank you all, you rock!
posted by malibustacey9999 at 6:31 PM on January 15, 2011

Happened to us a lot when we lived in Queensland, and even in Canberra to friends without aircon. Dust gets into fans, summer hits, fans do overtime, triggers get pulled, PC crashes or shuts down. Lifting the case and using a cotton bud dipped in a little soapy water to clean the fans does the trick. If you can, consider running the PC without its case on during hot weather - pointing a desk fan at the internals can make a big difference too.

Quick question - is it a Compaq? Coz I've notice they've got really, really low alarm thresholds.
posted by obiwanwasabi at 6:35 PM on January 15, 2011

Modern computers should not be run with their cases open; the cooling solution depends on the case being closed for proper airflow.

Doubtful it is overheating. An overheating computer doesn't not turn on. And doesn't get fixed by plugging and unplugging. Probably power supply or motherboard.
posted by gjc at 6:39 PM on January 15, 2011

That beeping sound you mentioned could point towards a problem with the RAM. At simplest of course, it could just need to be reseated, it may have come slightly out of place. Snap it out, put it back in and make sure it's locked well into position, and each of the "golden" ends is well inside the slot.

If it's a processor overheating thing, look up how to install a heatsink on your processor on google (Intel and AMD both have different styles on this). It's possible that the heatsink wasn't installed properly, or even that it came off. Take off the heatsink-and-fan, put it back on and lock it properly into position. You might need to apply some thermal paste as well.

Failing that, I'd say that it's probably a failing power unit. And of course, dust the place out, and try to dust out even the hard-to-reach places like the inside of the fans (PSU and CPU), and anywhere else you might find it.
posted by Senza Volto at 6:41 PM on January 15, 2011

Obiwanwasabi, it's an Antec... or at least that's what the label on the case says. (For some reason though, I have a totally unfounded feeling that the gamer guys at the shop built it themselves.)

I'll see how it behaves. If it shuts down again today or refuses to fire up tomorrow morning, I'll probably try the power supply then motherboard route.
posted by malibustacey9999 at 6:53 PM on January 15, 2011

Didn't preview: thanks, Senza Volto, I'll check the RAM and heatsink too if I have any more drama.
posted by malibustacey9999 at 6:55 PM on January 15, 2011

Be careful around fans with canned air, use someting to keep the fan from turning, the bearings can be screwed when blowing things out.

I have built around 300 computers and worked on who knows how many. These things are absolute dust magnets!

Look at PS, it has a fan, also.....
posted by raildr at 7:39 PM on January 15, 2011

I'm going to guess RAM or PSU. If the RAM is doing something bad---erroring, overheating, something, then it would cause sudden random shutdown. If its errored or overheated, it could be misreporting itself to the system board, making the machine not recognize it and thus not make it to POST, as though the machine had no RAM at all. I would run MEMTEST on it after I triple checked that it was seated.

If you don't want to do that, (takes as long as all night) and you have more than 1 stick of ram, play the "pull one swap one" game. First remove the stick in slot 1, restart. If it works, replace it and pull slot 2. If it works, put slot 2 ram in slot 1 hole and slot 1 ram in slot 2 hole, and restart. If it works, do each one by itself.

This lets you know if it's a stick (I bet it's a bad stick) or a slot, or both.

I would do all this, I would vacuum all hell out of my PSU (or canned air it), I'd triple check that my CPU heatsink fins were clear, and I'd report back.

Feel free to MeMail.
posted by TomMelee at 5:47 AM on January 16, 2011

If the case is Antec, then yes, this is a custom built computer. If it really was built by gamers, then it is possible that they have set up the BIOS for over-clocking (running faster than regular specifications) which would reduce stability.

You could consider re-flashing the BIOS but that may be way deeper than you want to go.
posted by JackFlash at 9:15 AM on January 16, 2011

So far, so good...
posted by malibustacey9999 at 3:38 PM on January 16, 2011

But will check RAM as per TomMelee's instructions later in the day (when I haven't got bored kids on school holidays begging to IM their friends).
posted by malibustacey9999 at 3:42 PM on January 16, 2011

The thot plickens.

Yesterday it refused to completely start up. I turned it on, I got the black screen with white gobblydook (is that BIOS?) for a half second (always stayed on screen for a second or two before), then I got the black screen with Abit emblazoned for a half second (always stayed on screen for a second or two before), then I got a half second of the Windows startup thingy. Then... nothing. Blank.

I tried the RAM stick swap (there's two sticks). No difference.

I found a shitload of dust caught in the fan under whatever-the-hell a Galaxy N11360 is (graphics card?). Cleaned it out. No difference (not that I expected that would be the problem but at this point I'll give anything a go).

The motherboard's digital display shows FF, googling seems to mostly say that's normal.

So... what next, my personal oracles of PC death? Is RAM replacement the most logical next step?

I am googling all this stuff before I ask you, but every site seems to have conflicting advice and has technical words of more than one syllable which just confuse me.
posted by malibustacey9999 at 7:02 PM on January 19, 2011

That's...strange. No beeping? The system is staying powered ON at this point? It's not shutting off?

Inside the case, on your MoBo, will be a large watch battery. Yank it out for a good 60 seconds, put it back in, restart. Tell us if that does anything.

Your system should not be overheating that fast, and if it's modern it really should continue to run fine overheated, it'll just shut itself down and run really slowly.

The FF code probably means that it's completed POST, which is "Power on Self Test", which is good. That's the stuff you see when the computer first starts.

You could have a lot of things going on here, anything from a failed MoBo to bad ram to a trashed hard drive.
posted by TomMelee at 8:34 PM on January 19, 2011