Could my PSU be the problem?

Hi all, I've been having problems with random computer shut down while playing games (usually graphic intensive but not always). The screen just turns black and the sound continually loops so it sounds like a buzzing noise and I have to switch off and reboot.

I don't have all the information with me at this second as I am at work, but I know my PSU is 450W ( I didn't have time to check the brand or anything as it was quite late when I checked last night). I know I probably won't get a definitive answer until I can check the brand as the better brands are so much better.

At idle my GPU temp is around 60-65, when playing high intensity games it starts pushing about 90 before shutdown. I also use V-Sync where possible as I know runaway FPS can sometimes cause high temperature, and this has solved the problem in the less graphically intensive games. Before the crash there is no suggestion that anything in my computer is struggling, fan speeds sound normal (for gameplay) and performance is still absolutely fine.

Anyway my specs are:

ATi Radeon 4850 512MB
Intel Core 2 Duo E8600 @3.3GHz
3582MB RAM
500GB Hard Drive
and like I say my PSU is 450W but I didn't have time to check the brand.

I know that's not a lot to go on but its all I had time to jot down before I went to bed last night. Any thoughts? And if the PSU is likely to be the problem, which one should I look at?
14 answers Last reply
More about could problem
  1. Yes, it looks like your PSU is the problem.
    500W is a minimum for 4850 - you need about 32A on +12V rail for your system.
    As far as I remember for reference 4850 90-95 degrees during load was normal.

    I would recommend buying a quality 550-600W PSU. It would be sufficient for your current platform and new one if you plan to upgrade in future (non-SLI). Have a look at products from SeaSonic or Corsair.

    Maybe you could find a cheap aftermarket cooler for your HD4850, for example a used accelero twin turbo. It is always a good idea to keep components as cool as possible.
  2. This sounds like your GPU is overheating. So instead of a new PSU, you should consider going for a new GPU, like a 7770. Faster, much less power draw, less heat, less noise. Your PSU will still be fine for such a card.

    Just to elaborate further, your problem is caused by the GPU. Especially heat won't change a bit with a new PSU. HD 48xx cards are notoriously power hungry > a lot of heat.
    A 7770 would match your other components and be an upgrade at a very reasonable price. Especially since you don't need a new PSU.
  3. Conflicting answers there!! One the one hand I'd rather get a PSU because it would work out cheaper even for a good brand, but if it is the GPU which is causing the issue I guess I'll have to look into that instead/as well. When I can later I'll get rid of any dust as well, it looked fairly dusty inside last night so I'm sure I'll knock a few degrees off the temperatures once thats sorted.

    The thing that initially made me think it was PSU rather than GPU related was the fact that I've never seen my GPU temp get above 95 (It quite probably has but not while I've been monitoring). Also there have been no performance issues or other graphical glitches consistent with overheating. Also, why won't upgrading the PSU help the GPU run cooler? Surely if the PSU is struggling its going to be putting more strain on the components as well as producing more heat? I don't know, clarification would be appreciated :)
  4. Ahhh I also forgot to mention in either post, another reason I thought it was PSU related was that occasionally the PC just fails to start at all. Sometimes I'll press the button and nothing will turn on. Turning off and on at the wall or unplugging the plug and reconnecting it normally sorts it out, but it is a bit of a pain. Never had it fail to start long term, normally just have to fiddle around with it a bit.
  5. First - Quote Power Consumption Results: Looking at total system power consumption levels, the GeForce 9800 GTX+ and Radeon HD 4850 use roughly the same amount power at load. At idle the Sapphire Radeon HD 4850 was a little more energy friendly as it should be. Remember the Sapphire Radeon HD 4850 consumes 110W at most.
    End quote, Ref:
    Based on this 450 Watt PSU is fine (Your Power consumpsion should be < 300Watts under load- HOWEVER this is predicated on the PSU being a QUALITY PSU. If it fales in the "Junk" catagory then YOU should replace it as Junk PSU will only deliever about 60 % of rated Power after they warm up.

    Yes High GPU temp could be the culprit also. Sounds like you 1) need to clean the dust out or 2) improve case air flow.

    Might try running Prime 95 to test Memory/CPU for stability. This will also check CPU temp and Your +12 V rail. While running monitor both, if +12 V is 11.6 or less, Ditch PSU!!

    To check GPU run Furmark in a window. Monitor GPU temp and again the +12 V rail.
    You can try removing the side panel of the case and test, If GPU/CPU temps are significantly lower then problem is Case air flow.

    ADDED - PSU in of itself should NOT significanly effect GPU temp as long as voltage on +12 V rail is above +11.6 and Less than 12.6 V.
    If +12 V is say 11.6V then this will cause the GPU current to increase, However Power = Voltage x Current. With Voltage decreasing and Current Increasing the Power consumption should be very close to Nominal when PSU +12 V is 12.0 V.

    Personnaly I ditch a PSU if +12 V drops below 11.8V (the spec is Lower, +11.4V)
  6. Vetinari said:
    Ahhh I also forgot to mention in either post, another reason I thought it was PSU related was that occasionally the PC just fails to start at all. Sometimes I'll press the button and nothing will turn on. Turning off and on at the wall or unplugging the plug and reconnecting it normally sorts it out, but it is a bit of a pain. Never had it fail to start long term, normally just have to fiddle around with it a bit.

    That probably means for you that you also have an issue with your PSU.

    However, the crashing during gaming occurs because your GPU is overheating. You could try to clean it from dust and/or improve air flow in your case, like it has been suggested. Even if that helps to get temps down enough to stop the system from crashing, a new GPU would be a good thing to consider.
  7. Right thanks for the replies everyone, I'll try cleaning tonight and seeing how much that affects temperature. I'll also check the exact PSU I've got and take it from there as to what, if anything, I upgrade/buy. I think perhaps neither my PSU nor GPU are running ideally anymore! Thanks for the help!!!

    Quick update, the above is the model of PSU in my machine at the moment. Thoughts?

    Not been able to find much on it, but the few places where people have posted reviews they tend to be positive.

    Unfortunately cleaning out all the dust hasn't made much difference to the running temp of the GPU, which is actually pushing closer to 100 than I realised before under very graphically intense conditions.

    While in there, checked all the fans were running, including those on the components and the two in the case, all were running fine. So is upgrading my GPU the best bet? Or just trying to introduce more cooling (though the two case fans should both be cooling the GPU as well)?
  9. I'd go with a GPU and replace the PSU as soon as possible after that, to fix that "PC just fails to start at all" issue.

    If that PSU really can't keep up with the power draw of your 4850 and therefore fails to start sometimes, getting that GPU, might even fix your "fails to start" issue. But I'd keep my fingers crossed on this one.

    Here's a table with power draw per GPU. Like I said, a 7770 would match your system, be an performance upgrade. It would definitely not put any strain on your PSU and doesn't cost much. You can go with a faster card, ofc. But that's up to your budget and you'll run into a CPU bottleneck at some point up the road.

    NOTE: This is sorted by power drawn only, not performance! Newer cards give more performance and draw less power in the process.

  10. OK, thanks for the advice. So on that basis, cross referencing with what RetiredChief said earlier the watts needed would decrease by about 24W? Or is the above chart just the power drawn at idle?

    Done a bit more digging around and CIT don't seem to have that many people raving about them. As long as there's no danger of it going bang at any time soon I'll leave off the PSU for now then.
  11. That's load, but the real TDP of a 4850 is somewhere around 148W, instead of 110W. According to an article I found on

    edit: Maybe not the load on your PSU is the issue, but simply the heat generated by your GPU. Which causes your PSU to overheat? Just an idea. Don't know how/where your PSU is mounted, which direction the fan is facing, what case you got, how many fans and in which direction they are facing ... 90+ degrees makes up for a whole bunch of very hot air that has to go somewhere.
  12. Looking at your PSU, although it is listed @ 450 Watts, it is weak on the +12 V rails.
    12V1 is listed @ 10 Amps (120 Watts). This probably is applied to the MB and provides Power for the CPU, GPU (via the pci-e slot, and possibly the Ram ( Not sure if the Ram voltage is derived from the +12V, could be derived from the +5 or the +3.3.

    +12V2 is a little higher rated @ 13 Amps (156 Watts)

    Combined power on the two +12 V rails is only 256 Watts
    In Most modern Computers, the +12 V provides the majority of power, The other rails ( +5 and +3.3 ) normally do not approach their listed ratings.

    Not sure what the Deregulation vs temperature for your PSU, But have a feeling that @ normal operating temps you will lose 10 -> 20 % which would bring that down to arround 200 -> 220 Watts.
    Junk PSUs can lose upto to 40% of rated output.

    My original comments were based on a 450 Watt PSU with two 18 to 20 Amp +12V rails.
  13. Hmmm I had seen that the amps were a little lower than what was recommended for my card, is it just luck that nothing catastrophic has happened yet then!?

    Although having said that, I remembered last night that this is actually a replacement PSU for the one I originally got in the PC, as the first one went bang (made me jump!) a few weeks after purchase. But they thought it was probably just a faulty PSU and sent exactly the same kind as a replacement. This one has been in the PC for over three years so I guess I need to be looking at replacing both the PSU and GPU then ideally.

    @whatsthatnoise: I don't think the PSU is overheating, its not close to the GPU (with the heatsink on the GPU facing away from the PSU entirely towards the case fans) plus there is a great big fan on the CPU which is between the GPU and PSU.
  14. PSU 101 lesson.
    Your Gpu is only drawing a portion of the +12 V current, Your CPU draws a portion of the +12 V current, The PSU has 100 % of the current thru it's voltage regulator.
    A lot depends on the Temperature (ambient) at which the manuf rates the the load capabilties. Lower end rate at about 22C, "Good" PSUs rate there output at about 40 C. For the Lower end PSU as internal temps go up the Current capabilities go down, called temperature deregulation.

    The Normal intake for cooling a PSU is the air from inside the Computer case which will be Higher than your room temp, How much depends on case airflow, So not just a question of distance the GPU/CPU are frm the PSU intake.

    The Corsair CX430 (lower total Wattage) has a slightly Higher +12V rating @ 28 Amps total (336 Watts). The CX500 has a +12 V rating of 34 Amps (408 Watts). The Corsair is what I consider a mid range (quality) PSU.
Ask a new question

Read More

Power Supplies Components