Symptoms of a Weak Power Supply

Technology and Gaming Help and Support
#1

So, I’m having a bit of trouble with my PC:

I bought a case that came with a PSU, a Coolermaster Extreme Power Plus 550 watt PSU, which seemed like enough for my build at the time. I’ve been using the rig for about a month now, and it work pretty well for the most part. The only thing I’ve been having trouble with is the occasional hang while playing intensive games. I have an Asus 5770 CuCore, and on the back of the box it reads that a minimum 500 watt PSU is required (got it covered), but it also reads that it should deliver 40 Amps on the rails. So, did some quick reading, and my PSU only puts out 32 Amps on the rails. Now, obviously I think this could be a potential problem, but I was able to deal with it by, oddly enough, turning up the fan on the GPU to keep the temps down. But, it still hangs every now and then.

Fastforward to real problem: I’ve worked up the courage to begin overclocking things a little bit. I managed to push the GPU clock on the 5770 from 850Mhz to 910Mhz in Overdrive, and it passed the test. At first, I had the memory from 1200 to about 1330 and it passed as well. So, I proceeded to subject my card to the FurTest. With these settings, it crashed within 3 seconds. So I backed off the memory clock a little, still crashed. Backed it right off to 1200 default, and it went about 30 seconds on FurTest. Now, one thing I really noticed was the temperature. My card only got up to about 64C before it crashed, and I know for sure that it has gotten up to around 80C playing Crysis flawlessly. The system doesn’t lag or anything, it works fine, and then just freezes up. I backed the GPU down to 900, and the card made it up to 75C before crashing.

Anyways, long story short, I’m planning on getting a Coolermaster 750 GX PSU. Does anyone have any idea if this will help with anything at all. Also planning on getting a second 5770 to Crossfire to justify the PSU :retard:

I’ve returned all my clocks to default, I’m really not sure what the problem is. I know that my PSU for sure is a weak link, but I don’t know if it is causing these problems. At first it seemed like a temperature problem, and that I had it solved, but now I’m not so sure, after I’ve started this overclocking business…

#2

The only time I knew there was a problem with my PSU was when my PC instantly turned off with a loud BANG scaring me shitless followed by a prominent smell of burnt wires.

I hope that helps.

…OK, OK, I know it doesn’t.

#3

i know this proberly wont help much, but find a program that can log your psu voltages and post them here.

btw ive got a 460watt coolermaster extreme power plus psu, running a c2q q9650 @3ghz + a msi radeon 4890 cyclone oc edition + xfx 780i sli mobo. i get about +12.10V idle, +12.05v load. I could put your gfx card in my system and it should run fine.

If your voltages are fine then start eyeing up other components. this means logging and checking if your temps are under safe values. btw, chipset heats up with the gpu. (it could still just be either your gpu or psu)

O yeah, im not an expert or anything but i woulda thought increasing your fanspeed would use MORE power, thus increasing psu load on a potentially faulty psu. So i doubt its your psu, especially if increasing fan speed lowers how often it occurs.

#4

Well, at first I thought that it was a temperature problem, seeing as my card idles around 50C, most other 5770 cards I’ve read about idle at a lower temp. than that. The thing that I’m the most concerned with is the fact that my card draws 40amps and the PSU only supplies 32amps. I’ve read in more than one place that the wattage isn’t really important, it’s how many amps you have on the 12v+ rails.

Edit: Oh, and the other thing I forgot to mention is that it doesn’t just hang when the GPU is under load. Lots of times, my games just crash when they are loading. Sometimes, the loading screen never comes up, it just hangs. Then, once the game exits, there is no “crash data” screen. I know, playing SCII and it crashes ingame, a debug screen pops up that allows you to send your data to Blizzard, but when it hangs loading a map or something on SCII, there is no debug screen, it just exits. Another thing that is definitely not temp. related, making me lean more towards the PSU…

Edit2: The last thing I want to mention is that there is no artifacting whatsoever in FurMark or anything that I’ve been doing that is GPU intensive, telling me further that the GPU isn’t overheating…

#5

I seriously doubt that there is 40amps required.
Those requirement numbers are exaggerated to compensate for low quality PSUs that do not supply the advertised wattage/voltage/amperage/efficiency.
I think those 5770’s suggest 400W, while then entire system with a 5770 generally doesn’t get past 250W on full load. It would take a lot of peripherals to get past 250W and a 5770.

That being said, your problems could be related to that power supply. I’ve had two of them die (exact same model) on me within the past 6 months. What you are describing could be cause by regulation or ripple problems. I heard moderately good reviews for it when it came out ( for the price at least ), and was disappointed by the failures.

If you can, try hooking it up to a known working power supply to see if there is any change in what you can do.

While your temperatures are higher than average for that card. airflow / ambient temperatures could influence this. More aggressive fan rules is an easy fix. Many of the 5770’s don’t ever get over 45% fan speeds. There is a lot of room for cooler temps ( depending on how much fan noise you care for ).

#6

Unstable voltages.

#7

Try setting the BIOS to factory settings. My machine used to crash all the time, and I couldn’t for the life of me work out what it was that was causing it to do so, as a guess I reset the BIOS, and I stopped crashing all the time. I think it was related to overheating not power problems but oh well.

Just to check, the crashes were pretty random things that happened a semi-random period of time after I turn my computer on. Is this anything like what your getting?

#8

Nope, only as I described before. However, I did have a faulty PSU on an earlier PC, and weird shit definitely happened. Unexplainable crashes, etc. Anyways, I’m thinking I’ll go ahead and get a new PSU and second 5770 and a couple new fans, since I was planning on doing that anyways. If it keeps doin shit, I’ll have to figure out my temps. Later on. I currently have my fan rules set so that it stays on about 30% (inaudible) under 55C, and it ramps up to full in a series of 5C steps after that. I don’t know how cool I can get it down to… Automatic fan settings seem to leave it idling around 55C. I think I just need better case cooling really.

[Currently Fixable]Root of the problem (possibly): Temperatures

Anyone have any suggestions for fan settings? What’s some good software for controlling the GPU fan and some good settings for it. As well, if it helps, I’ve got an NZXT Sentry LX running an 80mm Antec Tricool, a 120mm tricool, and a 120mm Scythe Kaze fan (Bigass mofo of a fan), and the coolermaster 120mm that came with the case. I’ve got a bottom, a side, and a back vent for placing the fans. Best idea wins and I use it :slight_smile: At this point, how loud it is isn’t an issue, although I’ll likely tweak it to make it quiet a possible.

#9

do you not understand? A motherboard also does very wierd shit when the chipset is dying, if a psu was dying i bet you would have erratic voltages or voltage drops. So before you go blowing $500, go and investigate all your other components.
Currently, you have very little reason to suspect a faulty psu. What evidence do you currently have? None or very little.

Faulty chipset = crashing, Faulty GPU = crashing, Faulty RAM = crashing.
Faulty PSU = crashing.

#10

correction, faulty GPU = black screens, faulty GPU memory = artifacts/BSOD, faulty HDD = hangs, faulty RAM = hangs/BSOD, faulty CPU = BSOD, faulty PSU = random shutdowns/reboots, faulty user = computer over (I doubt anyone except the old time lurkers will get that one)

#11

Add:

faulty motherboard = random shutdowns/reboots
bad memory timings = BSOD
unstable overclock (CPU) = BSOD
unstable overclock (GPU) = greyscreens, blackscreens, freezes, looping sound of death
unstable overclock (memory) = massive artifacts, shutdowns/reboots, BSOD
B2B CAS latency on zero (ASUS boards) = completely random total lock-ups with looping sound of death

Overheating symptoms:

motherboard (north/southbridge) = shutdowns/reboots - freezing on several ASUS boards (Striker, Rampage)
GPU (memory) = artifacts, texture stretching
CPU = automatic throttle slows down computer dramatically eventually shutting it down when temperatures are way beyond threshold
memory = artifacts (rarely) usually a BSOD
PSU = no idea, probably complete shutdown of the computer, if you have a cheap model it will probably BLOW UP AND MELT TO THE CORE OF THE EARTH

Artifacts (videocard related):

#12

Sounds like you needed to clean the gunk out of your mouse.

#13

Final thought for you guys tonight:

I just ran FurMark for 10 minutes straight at the stock clocks without a hitch. No artifacts, no hanging, nothing. It averaged 50fps, getting up to about 75. The card only hit 88C at full load and hovered around 86C for the most part. So it’s way hotter than it was when I overclocked it. Thus, further making me think that it’s a voltage problem rather than temperature. I let the card cool off for about 2 minutes after the FurMark, and it got down to around 47C idling with the fan on 100%. It seems to stay under 50C with the fan on 50%. I think the CuCores have a pretty crappy heatsink, I’d better invest in a different brand with my next one…

#14

Yea like I posted earlier, unstable voltages are really annoying. A good quality brand powersupply can deliver very stable voltages whether it’s idleing or under full stress.

Also, you might want to crank those fans up to the max. when gaming - it’s loud but if you got headphones it should be alright, lower temperature means longer lifespan. I wouldn’t say 88 C is a very safe temperature for prolonged exposure, but it’s up to you.

#15

Nah, I jiggled the cord and it was all good again.

#16

So you’re suggesting that he replace the whole motherboard to test for a problem with the mobo? Yes, it could be the motherboard, or BIOS settings. But if BIOS is up to date and at default settings, the hardware involved doesn’t have known issues leaving little options for troubleshooting. Also, when budget power supplies cause problems, it is more often attributed to either full failure, or voltage regulation issues that cause ripple. This ripple is not handled by the motherboard component capacitors well and will cause corruption and issues as described. You would not detect voltage changes.

I wouldn’t suggest starting with buying a new anything. I’d say first step is see if you can get a hold of a working power supply ( don’t even need to mount it ) and hook it up to the system and run a test. If your problems persist then you have eliminated a source without spending any money or even fussing about unhooking all that much. Even if you have a friend with a good power supply just bring his case over leave it in and just run the wires to your machine. I do this often. Unless you have an actual stress unit that has voltage monitoring this is the best place to start.

I might suggest also stressing the system in other ways. CPU stress or blend tests while at stock frequencies should help diagnose if the problem is limited to the graphically intensive applications.

By the way the 4770 has a max operating temperature of 105C. While 80 is high, it shouldn’t shorten the life of the card.

#17

Actually, if you have old (but compatible) parts lying around, you SHOULD test your hardware with those to see if various problems go away, rather than jumping the gun and buying a new piece of hardware that may not actually have a problem.

The annoying part is even GOOD power supplies can still just kinda die on you for no good reason. A good power supply will be more reliable and probably take heavier loads better, but it still may just decide it doesn’t enjoy living. Then again, I’m the king of getting the defective units for everything I ever own.

#18

The chip itself can handle much higher temperatures, it’s the capacitors and voltage regulators that are located around the chip that have a maximum operating temperature of 105 C, after that they can start to leak which will kill your videocard.

Yea if you’re unlucky enough, but a good quality power supply atleast gives you some safety knowing that it has protection against various problems and has been tested and benchmarked.

#19

In any other situation, I would go the isolate and diagnose route. I do have an ATX PSU in my old PC, not sure if it has a six pin pci-e though… But my plan from the very start was to upgrade the PSU and add a 5770 afterwards, I put together my current build just to tide me over until I could get the money to finish it. So I’m not really jumping the gun and buying a new one only because I suspect that my current one is defective. If it’s fine, then good, I have a spare PSU lying around. If not, hey I already got a new one! But otherwise, I completely respect the proper procedure, and have had to do it a few times in the past, but in this case, I’m going to be lazy. :stuck_out_tongue: If I’m still having problems, there’s a 99% chance that it isn’t the PSU, so that’ll be eliminated. Next will be a CPU stress test, and some form of RAM testing procedure. After that, I reset the BIOS and possibly try the MEM-OK feature on my mobo. And if I’m still having problems, I’ll just suck it up because I am not buying a new motherboard. Unless I can return this one.

Edit: So, I ate my shoe and did some stress testing on the proc like I said I wasn’t going to. I ran Prime95 and it completed 9 tests in 14 minutes no problemo. All 4 CPU cores were at 100% for 14 minutes straight, and the RAM filled up to about 88%, so I’m reasonably sure that it isn’t the CPU or RAM… And with my 5770 performing flawlessly in FurMark and 3DMark Vantage, I’m really, really thinking it’s the PSU that is my problem at this point. I may try hooking up a different PSU and playing SCII for a bit, see what happens…

#20

Double post, I know.

I came to the realization this morning that 14 minutes means diddly, so I put the CPU and RAM through a torture test today. It ran Prime95 at 100% for 13 hours and 30 minutes flawlessly. Completed 537 tests. I’d say my mobo, proc, and memory are stable. They are not the problem. And since it’s limited to graphic intensive situations, that points to either the GPU or PSU. The GPU performs perfect under extreme stress. Leaving only the PSU.

Founded in 2004, Leakfree.org became one of the first online communities dedicated to Valve’s Source engine development. It is more famously known for the formation of Black Mesa: Source under the 'Leakfree Modification Team' handle in September 2004.