I recently installed a new graphics card (see specs)and two new case fans, after a few minutes(or a few seconds) my computer will turn off completely as if I had turned off the power supply. I was using a 520W PSU when this first occured, I decided that may be the problem, so I installed a Thermaltake 600W PSU. Alas, no change. I checked the thermal paste on my cpu cooler, it was fine. I tried manually setting the voltages in my BIOS to default, no change. I managed to get it to the desktop a couple of times and looked at SpeedFan to see if anything was overheating, but that was also a dead end. Usually, the sooner it powers off, the shorter time it will stay on for the next boot so it's pretty tough to work with. I haven't been able to get it to stay on long enough to run memtest or any other diagnostic tools(Who Crashed?, action center, event viewer, etc) I'm sure I was grounded during the upgrade. I have the latest catalyst control center for my new card and the latest driver for my motherboard installed. Any advice?

CPU: AMD Phenom II X6 1090T
Mobo: ASUS M4A78T-E
GPU: XFX Double D Black Edition Radeon 7700 Ghz Edition 1GB(was Sapphire Radeon 4650)
PSU: Thermaltake TR2 TR-600 600W(was cheap 520W)
RAM: 2x2GB DDR3 1066, 1x4GB DDR 1066
HDD: 1TB WD Caviar Black 7200 rpm
CPU cooler: Corsair Hydro Series H60
OS: Windows 7 Ultimate 64-Bit

I guess 1st question is, did you isolate the graphic card as the source of the issue? Can you replace it with your original card and see if the lockups continue?

Yes, I tried my old card and the problem persisted.

Ow!! This is not a good sign. If the original card won't run either, then the possibility of a damaged Motherboard unfortunately crops up. None of this is really easy to diagnose. However, the fact that the CPU actually gets through boot up and all the way to the GUI tends to favor something other than hardware. Does the power-down occur if you bring up the BIOS and browse around in the BIOS? If not, then it strongly suggests something other than the Motherboard. Memory is still a possibility (remember that the BIOS runs in system ROM, not in the installed RAM), but memory diagnostics aren't too great at finding problems. The best test for bad memory is to swap in some known-good memory, but for most people, that is not an option. If you have an old disk drive around, you can try installing another OS as a test of the hardware. You could also try a live CD (Easier and Faster), which boots from the CD and lets you run a test copy without installing anything. If either of these tests pass (new OS or Live-CD) you can pretty much blame the problem on software.

By the way, I once had a Motherboard failure resulting from card swapping. Flexing the circuitboard broke an internal trace and the motherboard was trashed. However, in that case it didn't start up at all. On the otherhand, I have several times had machines which would either freeze-up, become un-responsive, or simply power down after a random time interval due to bad memory.

Best of luck, and hope for the best.

It hasn't shut off while I was in the BIOS yet. So I suppose that's a good sign. I also have a spare hard drive I can boot from. I've yet to try reseating my RAM or testing it beyond attempting to run memtest.

Edit: Is there a chance that I just need a better power supply? I was thinking of returning the 600W I bought and trying to see if a 700-800W would change things.

Power supplies are a more complex subject than one might think. First, the ratings are very misleading because there is no real standard about them. Some manufacturers cite power ratings in terms of "Peak" power and others in terms of continuous. Even among these there is variance. What peak and continuous mean to one manufacturer will not mean the same thing to another. Once, for a design project with a small single board computer, I actually measured current draw in real time and found out some really interesting things. The most interesting of these was that instantaneous current drain varies almost on and instruction by instruction basis. I suspect that this is why "Peak" ratings are so common. The high current moments only last for microseconds at a time so it really isn't necessary to sustain these levels continuously and a power supply that did so would be real overkill. Nevertheless, the other thing I discovered was that even very brief excursions above the peak rated power would push the supply into shutdown mode, with the obvious consequences. I suspect that this is why so many seemingly crazy problems are solved with a power supply that has a higher peak rating, even though it's continuous rating is virtually identical.

So, will the bigger supply help? Well, it's hard to say, but it is true that the type of problem you are having is not inconsistent with momentary overload leading to power supply shutdown. And, why did changing the video card get you into trouble? It could be simply that the new card is much more power hungry, pushing the power supply closer to the overload point. Then when some random sequence of instructions comes along and pushes current drain over the shutdown threshhold - BAM!, you shutdown. A bigger supply should fix this (so would a big capacitor across the supply, but that's a real Ruben Goldberg!), but it still begs the question of why the problem now persists with the older card swapped in. Can you maybe borrow a bigger supply to test with? It could save you a few bucks.

Also, try the Live CD trick. If Linux is stable with the new card installed, then it looks more and more like something mysterious creamed the Windows install. If that is the case, trying to find and fix the problem will probably be daunting in the extreme. You should probably simply re-install, but first use the LiveCD to save all your files to a safe place (CD, DVD, Second Disk Drive, etc). Linux will mount your Windows partition by default and allow you to use the Linux GUI to drag and drop your files into safe storage. It may take a few minutes to get things going, but the interface is quite intuitive and should be familiar to any Windows user. Just make sure that the format of the destination drive is Windows compatible - Fat32 or NTFS for disks and standard ".iso" for a data CD. DO NOT USE ANY LINUX FILESYSTEM for this. Windows simply cannot read any non-Windows format. All would not be lost, but you would have to re-read everything with Linux and then re-write in a Windows format.

Good luck!

Wow thanks for all the handy info. I have access to a larger supply so ill try that. i just tried to boot a kubuntu disk, everything started fine, but it suddenly shut off before even loading the desktop.

Just tried a corsair 700w power supply. no dice. im out of ideas.

I had another idea. i put my hard drive into a buddy's computer. when i tried to boot from it, it just said missing o.s. on the screen. when i booted from his drive to see if i could access it that way, his bios nor win7 detected my drive. so could all of this been caused by some kind of hard drive failure?

Just tried a fresh hdd, copy of win8, still shutting off.

Moving HDs around to different machines for booting is usually hit or miss. If it's a clean HD and a new install of 8, you've tried 2 grpahic cards, and a new power supply but you see the same issues, then it's looking more like the MB or Mem.

It look slike you have 3 mem chips. I would remove 1x4 and try running to see if you are stable. Do the same for the others.

Just reset the CMOS battery, didn't make a difference.
Luckily my mobo is still under warranty and I'll be able to get another one to test sometime next week.
Thanks CimmerianX, I'll try out your suggestion and post results shortly.
Edit: On recent boots(CMOS and Cimmerian's suggested RAM test), it hasn't been able to get past the "Starting Windows" splash/loading screen. The graphic stops "vibrating", and it just kinda sits there until it powers off. I think I'm going to try to see if I can boot from my Win8 drive and install it over my old Win7.

Instead of windows, try any CD with memtest+. It's a bootable CD with a memory diagnostic. It will start massive reads/writes to the memory. If your mem is 'flakey' this might identify it.

Also, I don't think we've asked if you are doing any kind of overclocking. If you are, of course set everything back to the default levels.

sound like PSU issue

My linux CD has memtest I'll try that out. My Win8 test failed, removing different memory sticks had no effect either. I'm not doing any overclocking, completely default all around.
henydiah-I've tried multiple larger PSU's and they didn't change anything.

Powered off during the memtest.

So if all chips have been swapped out, and the memtest results in a power off, MB or Proc sounds like a good candidate.

Something else... Does the proc have adequate cooling? Does it have an installed heatsink/fan with heat compound between the proc and the heatsink?

I'm using a Corsair Hydro Series H60 CPU cooler, it tends to stay between 50-60c. I removed the cooler from the cpu fairly easily, the compound seems fine. The cooler is a pretty recent addition. I suppose I could reapply it though.
I just called ASUS and got an RMA number for my MB, going to send it off Monday. Sent my 7770 back for replacement earlier today.

ack. Good luck and report back.

  1. First problem with new card - too much power required.
  2. Second problem. First problem caused low voltage (hence too much current) to motherboard, causing component failure.

At this point, you have only one option - replace system. You might be able to recover the hard drives and new video card, but that is not a given.

Don't bother to try and fix the old system. The video card may be ok (or replaceable under warranty), and the hard drives may be usable, but I would suspect that the motherboard, power supply, etc to be fubar.

I think that sounds about right. should i stick with 600w or should i go bigger?

Well, you've certainly narrowed the field. Not everyone will agree with me, but I'd bet the ranch that this is NOT a power supply issue. It's either memory or something on the motherboard, If the Kubuntu live CD won't start and run in a stable fashion, you've clearly eliminated a software issue. Reinstalling the OS a million times will not help. Neither will a new disk drive. Neither the OS nor the disk drive are involved in running the live CD. I have seen problems like this before that were related to memory. However, since you cannot even run the memory diagnostic, the score for a bad motherboard, or indeed a bad CPU chip, goes up. The problem occurred when you tried the new video card, which also tends to point towards a loose connection somewhere which was worsened by plugging in the new card. There are a few more things to try before moving on to a new motherboard. Try to remove the CPU itself and reseat it. I have had to repair at least one failure due to a very slightly misaligned CPU. I had one other friend who had a similar problem and it turned out to be a toasted North Bridge chip. Also remove all the memory and reseat it. It's a long shot, but you never know. Also, if there are any other bus cards, remove all of them and see if the system boots correctly. If the motherboard has built in video, you should even remove the video card and use only on-board hardware. As a final desperation move, disable ALL onboard hardware except the disk controller and the on board video. If the machine is then stable, re-enable one thing at a time till the board fails. Any failure when re-enabline on-board hardware is a definative test: the motherboard is defective. Good Luck!

First problem with new card - too much power required.
Second problem. First problem caused low voltage (hence too much current) to motherboard, causing component failure.

It's not so much that I disagree with the first item as that I think it's unlikley to have that much significance. First of all, power supply overload of any real significance will simply shut down the power supply. Second, the kind of power supply overload caused by added bus hardware typically manifests as long term instability - that is, the system fails unpredictably when under stress, not when just idling along.

However, I do have to completely disagree with the second item. CMOS logic is extremely low voltage tolerant. Even the older TTL logic families, which were somewhat intolerant of low voltage, only lost logic threshhold margins and did not fail catastrophically. The only real consequence of low voltage is a greater liklihood of logic errors. If you can make an argument for over-voltage, then I would agree. Almost all modern logic is rather intolerant of overvoltage.

Thanks for the insight sbesch. i tried reseating the memory before i sent off my mobo but that didnt have an effect. i didnt reseat my cpu though. I had disabled the onboard video to avoid some potential confusion wity the new card. i would try everything you suggested but ive already sent my mobo and new gpu in for rma.

eighty4prcnt - You've probably done the best thing. These problems can drain off an incredible amount of time and getting new stuff and starting over can get you up and running much more quickly and reliably. Sometimes you have no choice, but when you do, new is usually better. When you consider that most failures of electronics are DOA or "infant mortalities", you are really far better off.

I thought so too. I try to avoid the headache. Ill post an update once I get everything back and rebuilt. Thanks again to everyone who shared their awesome knowledge.

Just got my Mb in. Everything started fine but a couple minutes in it died again. Used the 4650 fr this

This is where you strip the setup to absolute basics. One mem module, cpu, mb with on-board graphics, and a monitor. Next add a system hdd if it can stay on for a few minutes, try to boot. Add more mem... and so on. Finally, the vid card.

I just figured out the problem. Apparently, my CPU cooler malfunctioning. I took off the cooler and put on the heatsink my CPU came with, and to my surprise it stayed on. When I checked the temps before the problem began they weren't particularly high. I only wish I had tried this from the start lol. Thank you all for your suggestions.

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.