PDA

View Full Version : Home Built PC problems.



AlgernonCecil
26-01-2012, 12:49 AM
Here's the situation as it stands. A few months back I decided to have a go at building my own gaming rig for the first time. The assembly was pretty straightforward and in short order I had it booted and Windows installed. I was pretty chuffed, but then the problems began.
Intermittently my PC would completely power down and then after a few seconds restart. First thoughts after talking to some tech minded mates was to check the RAM and sure enough when I took one stick (2GB) out my problems stopped so ,for a while, I ran my rig on just 2 GB RAM.

More recently I decided to fix this problem once and for all by cutting my losses and buying 8 GB of new RAM to replace the faulty old set, but as soon as I had slotted it in the power down problem came back (but slightly different as I now get about 30 seconds of black screen and glitchy sound before complete power down). After consulting my tech mates again it was agreed that it sounded like a graphics problem and since I was looking to upgrade my graphics card anyway that's exactly what I did but the problem still persisted exactly as before.

I also had access to a free replacement motherboard and, since it was free, I thought I may as well try it but once again the same problem still remains. If anyone can help me with this total balls up it would be much appreciated as I have replaced enough supposedly-faulty-but-acctually-still-working parts that I'm not far off owning two complete gaming rigs.

System Spec:
-2.80 gigahertz Intel Core i5 760
-AMD Radeon HD 6800
-8GB DDR3 Crucial Ballistic RAM
-SAMSUNG HD103SI 1TB
-OCZ Stealth Stream 2 600W PSU
-Gigabyte S-series GA-H55M-S2H motherboard
-Windows 7 Home Premium (x64)

Many thanks

Duckee
26-01-2012, 12:50 AM
At a blind whim I would like you to consider checking out your temperatures.

AlgernonCecil
26-01-2012, 12:59 AM
Blimey, that was a quick reply. I did do temp checks on all components when I still had the original set of RAM, graphics card and mother board and everything checked out fine. I will do another check on this lot of components too but I suspect that temperature is not the problem.

Cheers mate.

FuriKuri!
26-01-2012, 08:52 AM
I'd put my money on temps but you can give your CPU and RAM a good work out with this program - http://www.mersenne.org/freesoft/

If it detects incorrect results you have a RAM problem. It'll either be overheating, have the wrong clock speeds in your BIOS or not be seated properly in your slots. Keep an eye on your temps with something like this - http://www.almico.com/speedfan.php

Your CPU should be around 50 celsius.

JohnnyK
26-01-2012, 10:06 AM
Are you sure you have the PSU connected properly to everything, eg. if you have a 6870 it requires two 6-pin connectors (the 6850 has only one I believe)?

As for temp checks - did you do those under load? Is your heatsink on the CPU seated properly? Have you tried resetting your BIOS to failsafe defaults and checked if the problems persist?

Duckee
26-01-2012, 02:06 PM
The new heatsink montages that come with Intels newest CPUs are very easy to mess up if you are not used to them. Also remember to redo the paste if you take it off. (Im guessing you already know this, but better safe than sorry)

CMaster
26-01-2012, 02:52 PM
Just agreeing with the others that it sounds a lot like an overheating issue.

SMiD
26-01-2012, 02:59 PM
Aye. Overheating or PSU issue most likely. Good luck!

KilgoreTrout_XL
26-01-2012, 04:40 PM
I seriously doubt that you're having heating issues, actually. Even if you were, your PC would boot anyway, and then it would blue screen when you put it under load. When you set up your new mobo, you removed the old thermal paste from your CPU with isopropyl alcohol and a q-tip, and reapplied it, correct? If so, then there's no reason that the stock intel fan can't keep that cool enough, esp. if you aren't overclocking.

What I think is more likely is either (#1) your ram is bad, or (#2) your motherboard is trying to run a set of DDR3-1333 speed memory at DDR3-1600 (which one do you have, by the way?)_. I just ran into a bunch of problems with that recently.

But it's probably that the ram cards are bad. Get Memtest86+ (http://www.memtest.org/), burn the image to a CD, and enter your BIOS and change the boot preference to your optical drive.

Before you boot up, take one of the dimms out, and make sure your MB doesn't have a weird preference over which dimm socket you have to use if you're only using one dimm. Figure that out, stick the CD in, boot and and let memtest86+ run through two tests at least twice on each card separately. If there's an error, then you know which card is messed up.

KilgoreTrout_XL
26-01-2012, 04:54 PM
My bad, I didn't notice that you were able to boot.

DL coretemp and list your CPU temps: http://www.alcpu.com/CoreTemp/

AlgernonCecil
27-01-2012, 12:28 AM
Cheers for the feedback, here's where things stand now:

@FuriKuri! - I am currently running the Prime95 self test program from http://www.mersenne.org/freesoft/. So far it's been going for approx. six hours and hasn't detected any problems yet. I'll leave it running overnight and see if it picks anything up.

@JohnnyK - I have checked and double checked that everything is connected correctly and the heatsink id definitely seated properly. I will try resetting my BIOS to failsafe defaults but as far as I am aware it already at default settings.

@Duckee - I probably should have mentioned that I'm not using the heatsink that came with the Intel CPU but rather one from Scythe which was really easy to fit. Also, yes, I did clean off and redo the thermal paste when I switched motherboards.

@KilgoreTrout_XL - I've installed Coretemp and temps in all cores range from 55 - 59 Celsius when I'm not doing anything particularly resource hungry (is that a little high?) and goes up to 60 - 63 Celsius when I play Arkham City with most settings on high. Sorry, I should have mentioned that my RAM is DDR3-1600. What's the best way to check my MB is running at the correct speed? I'm not really clued up on messing about with RAM.

Cheers for the response everyone, given me plenty of things to think about.

Overheating seems to be top of the list but I'm not totally convinced as a pattern has started emerge. It used to be that my system would power down pretty much at random (when playing games, browsing the net, playing video files or even when booting up) but now it's only when I'm playing certain games. Orcs Must Die, Call of Pripyat, Fallout: New Vegas and Skyrim all cause my system to power down at random. Other games which have just as high or even higher specs (Dead Island, L.A. Noire, Arkham City) will play indefinitely at high settings with absolutely no issues.

Gold Stars to those who have taken the time to read to the end this quite large block of text : )

KilgoreTrout_XL
27-01-2012, 01:00 AM
@KilgoreTrout_XL - I've installed Coretemp and temps in all cores range from 55 - 59 Celsius when I'm not doing anything particularly resource hungry (is that a little high?) and goes up to 60 - 63 Celsius when I play Arkham City with most settings on high. Sorry, I should have mentioned that my RAM is DDR3-1600. What's the best way to check my MB is running at the correct speed? I'm not really clued up on messing about with RAM.



Hmm, yeah your idle temps should probably be a good 20c less than that, especially with an aftermarket CPU cooler. My bet is that something went amiss when you installed the CPU cooler, or you have a really crowded and dusty case. And if you're already getting 55c temps while idling, I would hold off on running prime95. After it gets going a while it can add a good 25c to your CPU. You want to stay 15 degrees below that "Tj. Max" figure on coretemp.

Also, CPU-Z tells you how fast your RAM is running. Go memory tab -> DRAM Frequency. Double that value and that is your ram frequency (DDR = Double Data Rate). I've had nothing but trouble setting it any higher so mine is staying at 668.9; if yours is running at max speed it'll say 800 or so. I would go into the BIOS and try to roll that back to 1333 until you get everything else in order.

trjp
27-01-2012, 01:05 AM
I don't think this is overheating (definately not the CPU, almost certainly not the mobo or memory) - I think it's either

a - a power problem (either with the powersupply or the motherboard itself)
b - a problem with your mobo memory controller (which may be fixable with a BIOS upgrade or may be a hardware fault)

What you've described is a system which becomes more unstable the more things you put in it. With 2 DIMMs it crashed a bit - with 1 DIMM it crashed less - when you put in new DIMMs it started crashing more. That's either a power-related issue or the mobo doesn't like multiple DIMMs (I've met motherboards which had exact issue).

First up - I'd want to take a look at the power levels you're getting from the PSU - something like Speedfan will show the actual voltages (your BIOS screen may do this too) and they must - absolutely must - be within 5% of their intended value constantly (Speedfan will log this stuff or fire alarms so you can see how it changes over time and even - maybe - at the time of a crash).

PSUs are tricky things to trace problems with tho (because current drops can be TINY and intermittent) so if you have access to another PSU, put that in and see if things improve - that's the No1 way to diagnose a PSU fault.

If that isn't the source then I'm going to say your mobo is faulty - that's always a hard one because of the work involved replacing it but that's what I'd be worrying about next.

trjp
27-01-2012, 01:07 AM
p.s. - on the temps front, nothing really wrong with a 50-60 idle if the case/fans are a bit dirty (cleaning fans can easily drop temps by 10-15 degrees) - if it's clean and there's good airflow tho, 60 is high for idle.

If his thermal solution was poor, he'd be seeing a MUCH greater rise in active temps tho - not just a few degrees - more like 20-30 and climbing...

When a CPU overheats, a PC will usually refuse to restart (until the temp drops) and then repeatedly crash (as the temp rises again) and he's not describing that - hence I don't think he's overheating.

DigitalSignalX
27-01-2012, 04:01 AM
Was reading some of the comments on OCZ PSU's on newegg and there's a good sized bunch (about 15%) who are also getting weird power-down issues. If you have another PSU at your disposal, I'd suggest a swap out and see if that fixes it.

Odeon
27-01-2012, 07:16 PM
I have to agree that the PSU is your next point of interest. The fact that you've already replaced the RAM, GPU, and mobo and have good (though not the best) CPU temperature with an aftermarket fan means you're down to two things: PSU and hard drive. The PSU is the easiest thing to test by exchanging it with a new one or a known-good one from another system - just be sure that it can handle your setup. The best option would be the same or bigger PSU since 600W seems to be plenty, but the recommended PSU for the 6850 and 6870 is at least 500W in single mode. Which one do you have, by the way?

Replacing the hard drive would require either a complete re-install of Win7 or at least a Ghost-like imaging from the current drive to another drive. Re-installing would rule out Windows problems, but those don't typically result in complete power downs like you're having.

By the way, was the replacement motherboard the same model as the previous one? If not, you probably already did a fresh install of Win7, right? If you did, did you use the same hard drive? And are there any other details you can share?

AlgernonCecil
28-01-2012, 12:54 AM
Right, first of all I posted inaccurate temps in my last post. When I said CPU temp was 55 - 60 Celsius idle, I had forgotten that I had been Prime95 in the background for approx 6.5 hours (derp) so idle CPU temp is actually a much lower 29 Celsius.

I've been running Prime95 for 24 hours + and still no failure so I'm assuming my RAM is probably fine.

I agree, it's beginning to look like a PSU problem but is there any way of testing my PSU without getting a new one (I can't see a way of getting hold of a spare or borrowing one from somewhere). Also, if it turns out that I do need to replace my PSU, can anyone recommend a decent one?

trjp
28-01-2012, 03:25 AM
There's no decent way to test a PSU - as someone who's fixed PCs for years, my solution is always to try another PSU and if that solves it - bin the old PSU.

The most common problems are simply TINY drops in the power which cause all manner of weirdness - and you can't really monitor for those easily (without some expensive kit!!)

Holding a spare PSU isn't a bad idea anyway - they do let go without warning and you don't want to be sitting for a couple of days without a PC whilst the new one is en-route.

As for brands - Corsair and Antec are my first choices usually. I've had problems with Coolermaster, Arctic Power and a few other brands (probably one-offs but you only get burned once) and I'm about to risk a 'Be Quiet' unit on a friend's recommendation (unless someone here says otherwise!!)

p.s. one other occured to me - have you checked that the motherboard is mounted properly in the case and isn't shorting/earthing wrongly? It's a rare things these days (most mobos are designed so that it's not a likely scenario) but earthing can cause power fluctuations which can crash a PC almost 'at random' (and it's one of those things that just moving the PC or even trying to install a DIMM can cure/cause entirely at random)

Mistabashi
28-01-2012, 11:22 AM
Since we're on the subjects of PSUs: Obligatory link to database of PSU OEMs. (http://www.tomshardware.com/reviews/power-supply-oem-manufacturer,2913-4.html) Aside from one or two companies (Seasonic for example), none of the PSU brands you will see in the shops manufacture the units themselves, and most will use several different OEMs for different products, so figuring out who actually makes a particular PSU can be very useful in trying to figure out whether it's a good buy. Obviously, checking out reviews is always preferred, there's several sites that do regular detailled PSU reviews (johnnyguru in particular).

For my money, anything made by Seasonic or Channel Well tends to be a good buy (that includes a lot of Antec Corsair and XFX units, steer clear of the Corsair CX500 & 600 though as they're over-rated and over-priced). The "Be Quiet!" PSUs are either made by Topower or FSP, the latter being much better as far as I know.

corbain
31-01-2012, 03:42 PM
I've built plenty of PCs over the years, and every time i've had a problem like this, it's been the PSU

Beg, borrow or steal a different PSU and see if it works, that's the only way to test it out i'm afraid

corbain
31-01-2012, 03:43 PM
And by the way, best make of PSU is Seasonic. 90% of other brands just use rebranded Seasonics, so just go with the best from the start.

AlgernonCecil
31-01-2012, 10:33 PM
Right, just bought myself a brand new, shiny Corsair TX750 (Seasonic) PSU, connected it up and NOTHING has changed. Still getting the same blank screen then power down problem as before. This is getting tiresome and expensive, the only two original components left from my initial build are CPU and HD. Should I flip a coin?

@trjp - Motherboard has been replaced once and is definitely mounted properly

@Odeon - I have reinstalled Windows a couple of times (using the same HD) and it it has made no difference. The replacement motherboard was an identical model. As for any other details all I can really tell you is that while some games I can play as long as I like with no issue (L.A. Noire, Deus Ex:HR, Arkham City) other games are much more unstable and problems can occur at any time BUT going in to V.A.T.S in Fallout: New Vegas is guaranteed to cause a power down. (and for some reason I can't turn the underline function off but that's probably an unrelated issue.

Somebody help me. I just want to play Skyrim :(

FuriKuri!
31-01-2012, 10:49 PM
Try underclocking your CPU by a few hundred mhz. If the problems go away you've either got a duff cpu or your motherboard hasn't quite got the autodetect right. I can't imagine a duff hard drive causing a full power down.

Odeon
01-02-2012, 12:00 AM
Yeah, sorry to say it but that's a REALLY strange issue considering all of the parts you've exchanged. For the problem to be consistent to only a few programs/games makes it even stranger.

Try FuriKuri!'s under-clocking suggestion first since it'll be the easier option. If that doesn't work, make sure that you've not only updated your graphics drivers and DirectX, but also your .NET assemblies and similar MS support software. I'm not sure what might affect F:NV exactly, so it's best to update all optional MS software through Microsoft/Windows Update.

If either of those don't work, find yourself any hard drive cloning program (Ghost, DriveImage XML, Acronis, others can be read about here (http://www.makeuseof.com/tag/5-ways-to-clone-and-copy-your-hard-drive/) and here (http://downloadsquad.switched.com/2008/09/05/5-free-apps-to-clone-your-hard-drive/)) and clone your hard drive to any spare drive of the same or larger size. Some programs will allow you to use a smaller drive based on the amount of used data on your current drive, but it's easier just to go with a similar or larger-sized drive.

As an alternate, you might try starting a new topic at the F:NV PC (http://forums.bethsoft.com/forum/112-fallout-new-vegas-pc-issues/) Issues board.

AlgernonCecil
01-02-2012, 02:38 AM
I have very little experience or knowledge when it comes over/underclocking (I avoided it mostly so I don't balls things up and make my system unstable. Oh! The irony!). From looking at my rig spec can anyone tell me what my BIOS settings for CPU and RAM (timings ect.) should be or, failing that, link me to a decent and easy-to-understand guide, so I can work it out myself. I'm thinking that maybe my BIOS auto detect has got things wrong?

I'm REALLY hoping that it's not a duff CPU as this project already has turned in to a money pit.

corbain
01-02-2012, 10:30 AM
have a look in the BIOS power saving settings.

I had a strange issue where disabling some of the Intel power saving settings stopped my PSU from making a terrible transistor squeal. (don't ask me what the settings are called, no idea now)

Just fishing around in the BIOS might help.

AlgernonCecil
01-02-2012, 07:34 PM
Just spent the afternoon turning things on and off in BIOS none of which seemed to have any effect. I also successfully underclocked my CPU but this also solved nothing. I'm thinking that if my CPU was duff then I probably wouldn't be able to run any games at all so it's got to be something specifically graphics related, but given that I've already replaced my RAM and graphics card I'm totally stumped.

Suggestions?

KilgoreTrout_XL
01-02-2012, 09:54 PM
Wow, man, I'm sorry to see that you're still having problems.

I was looking over the thread and tried to find confirmation that you created a memtest86+ boot disc and ran each ram card through at least two tests to see if they passed. Did you do that?

Have you run chkdsk on your 1TB drive? I suppose that that could turn something up. And a quick google of your drive turned up this article: http://datacent.com/datarecovery/hdd/samsung/HD103SI+1TB.
Note I have never seen or heard of that site before, but figured it might be something you could check out.

This is a trick that has worked for me in the past: try to to install your video card in a different PCIe slot. You'll probably need to move your sound card and a few cables.

I guess I'm also wondering if you were grounded when you touched the CPU, and whether there's any arctic cooler (or whatever you used) on any of the pins on the bottom of the chip. Also, if you're in an old house with crappy wiring, and if you're using a surge protector.

corbain
02-02-2012, 09:42 AM
Would also definitely suggest a surge protector (even without this problem, you shouldn't run that much expensive hardware straight off the mains without one)

If that's not possible, try moving your PC to a different house and see if the problem persists. Could be dodgy wiring, is it an old building?

These are the absolute worst kind of problems to troubleshoot i'm afraid, i had a very similar one (but not as hard to narrow down) the last PC i built, and I swore i would never self-build again.

One thing to check is the windows error log, buried somewhere in Window 7. It's normally less than useless for troubleshooting a HW issue, but it may throw some light on what's happening just before the power down.

AlgernonCecil
02-02-2012, 03:34 PM
The saga continues...

@KilgoreTrout_XL - OK I spent yesterday evening and overnight running memtest86+ and everything passed just fine. I actually ran chkdsk a couple of weeks ago and everything checked out. As for moving my video card to another PCIe slot, I can't see it helping since I've already moved it from one motherboard to another. Also I was VERY careful when installing the CPU (both times)

@corbain - My house is not particularly old and I am indeed using a surge protector. Also I don't have (legal) access to a second house :) Looking at the event viewer I get an EventID 41 kernal-power , which is not an uncommon problem if google is anything to go by, but, it seems to be the symptom of a huge variety of totally unrelated problems. Finding a single easy fix is impossible.

Also, I decided to check my RAM settings on the Corsair website and it said my latency should be 9-10-9-27 (instead of the 9-9-9-24 stated on the side of the RAM itself) and having changed it in BIOS my system does seem to boot and load quicker but it has done nothing solve the main problem.

Thanks for your suggestions so far, but keep them coming.

Odeon
02-02-2012, 07:06 PM
I have very little experience or knowledge when it comes over/underclocking (I avoided it mostly so I don't balls things up and make my system unstable. Oh! The irony!). From looking at my rig spec can anyone tell me what my BIOS settings for CPU and RAM (timings ect.) should be or, failing that, link me to a decent and easy-to-understand guide, so I can work it out myself. I'm thinking that maybe my BIOS auto detect has got things wrong?

I'm REALLY hoping that it's not a duff CPU as this project already has turned in to a money pit.
I think the only way that the Auto timing settings could be messed up is if your motherboard is messed up. But since you've already replaced it, I think it's safe to say that that's not the issue. Now that I think of it though, have you tried looking for a BIOS update for your motherboard (http://www.gigabyte.com/products/product-page.aspx?pid=3310#bios)? What version is your current BIOS? You can find that by entering the BIOS, but it's usually displayed within a few seconds of bootup unless there is a Gigabyte logo that shows up instead. If you decide to try a BIOS update, be very careful to follow the directions exactly because it could cause the PC not to boot if something goes wrong.

One way to test your CPU is with Prime95, which is a free program. It's not exactly straight-forward, but it will do a good job of putting your CPU through the wringer and can tell you if there's a problem. Make sure you have overclocking turned off before your run it though so you can be sure that any problems are not a result of too much overclocking.

Do you have a spare hard drive that has as much space as what is used on your C drive? If you'd rather not try a BIOS update or you have the latest one already and Prime95 doesn't show any problems, the next step is to try a different hard drive. It's not likely to be the cause of the problem, but there's no much else to try.

AlgernonCecil
06-02-2012, 04:57 PM
Just checked and my BIOS is fully up to date (Version F7). As for testing my CPU I've already tried running Prime95 for approx 30 hours and it turned up nothing. Is there any other way of testing the HD without replacing it? Prices are ridiculous at the moment and I've already wasted enough money replacing parts that had nothing wrong with them.

Odeon
06-02-2012, 06:27 PM
Here's (http://pcsupport.about.com/od/toolsofthetrade/f/testharddrive.htm) a pretty good article on testing hard drives for errors. I would skip the paid testing programs since the freeware options are good enough for almost every type of problem. Give one or more of those a run and if it gives you any errors, you'll want to get your hands on another drive stat.

I understand completely about not wanting to have to buy a new hard drive right now with prices being what they are these days. You don't have a single spare hard drive laying around or know of anyone that does? The size you'll need depends on the amount of used space on your current drive. How much is in use on your C drive?