[SuperSquashMan]

Hello all-
a couple years ago I got a Compaq CQ5320F for myself. After realizing that I couldn't use the onboard graphics to play any of the games I wanted to, I went and got an XFX nVidia GeForce 9800 GT with 512 MB DDR3 RAM off of eBay. After upgrading my PSU, it worked fine for a little bit, but recently my system's been getting unstable. Sometimes when it boots, it'll go into the 'system restore', and I almost always have to switch to the onboard graphics to got into the contol panel and switch displays. But worst of all, after about 30 minutes of use, the screen will freeze, and the display will become purple and pixely, for lack of a better term. Is there anything I can to to fix this? And if not, are there any affordable graphics cards I can get that have similar performance?

[waldojim]

What kind of power supply did you get? Have you performed any testing yet? Do you know what the temperature of the video card, CPU, or over-all system is?

[SuperSquashMan]

-I have a power supply that says the max output is 480W, while the average is 330W.
-My CPU temp was at about 42ºC, while the GPU temp (while running minecraft on highish settings) was about 85ºC.
as I said, this configuration has been working for some months now, with the screen freezes first occurring about a month ago and getting more frequent.

This post has been edited by SuperSquashMan: 26 December 2012 - 05:28 PM

[waldojim]

Well, sorry to say it, but at first thought, that power supply doesn't sound like it is really up to the task of taking on an 8800GT. Those odd PSU's will take the load for a while, then flake out. If you could, post the full information on the PSU, then run a couple tests.
Namely, get your hands on HW Monitor, and Furmark. Bring up HW monitor and then let Furmark punish the video card, watch the 12V readings on HW Monitor while this is going on. Very likely, you will see it drop below 11.5V - too low to remain stable.

[SuperSquashMan]

The PSU in question is a Logisys PS480D2, which I got on clearance a while back.

I ran the Furmark test, on which it scored a 509. However, there was no 12V display in HWmonitor, as far as I can see.

Attached thumbnail(s)

• 

[LiveBrianD]

Yep, those units are not known to be particularly good. You can get a Corsair CX430 for $25 after rebate (they always seem to have that for it), which is a LOT more stable.

[waldojim]

SuperSquashMan, on 26 December 2012 - 06:36 PM, said:

The PSU in question is a Logisys PS480D2, which I got on clearance a while back.

I ran the Furmark test, on which it scored a 509. However, there was no 12V display in HWmonitor, as far as I can see.

Well, there are rare cases when that software doesn't pick up the right readings... sorry to hear it won't work on that one.

In any event, Logisys is actually very well known. For going up in smoke. I don't think they have actually made a single 400watt or larger power supply in the companies history. When stressed past 250 watts, most of them simply explode. Example: http://www.hardwares...y-Review/1293/7

Quote

The Logisys PS350MA can’t deliver 350 W or even 250 W. The maximum we could pull from it was 220 W. During our test seven, when we tried to pull around 250 W from it, the power supply exploded after a few seconds. The two switching transistors were the components that fried. To give Logisys the benefit of the doubt, we tested two different samples and got the same exact results.

The 600watt model actually had similar results: IE, it burned itself out at a 360watt load.
In the future, I would suggest buying brands known for higher stability PSU's. Antec Earthwatt/High Current series, Seasonic, Corsair, PC Power and Cooling, or OCZ ZT/ZS series. All of those are known for better power supply designs, and better build quality.

[SuperSquashMan]

So I went to Microcenter earlier today and found this. I have it running in my computer now and so far it seems to be doing fine.

[SuperSquashMan]

The screen froze up again. Possibly it's the graphics card itself; I read somewhere on the internet that microcracks in the pin connecters of the GPU could be a problem, and baking it could be a solution. Would that help?

[LiveBrianD]

<Smacks self on head> That's no better. Seriously, take a look at the ones waldojim suggested, and return the Diablotek.

[waldojim]

SuperSquashMan, on 27 December 2012 - 01:40 PM, said:

So I went to Microcenter earlier today and found this. I have it running in my computer now and so far it seems to be doing fine.

You will end up with the same problem with that PSU. Please understand this - power supplies are one of the very few parts that can take everything out when they die. Buying cheap may seem like a great deal walking out of the store, but bring in their own headaches.

[waldojim]

SuperSquashMan, on 27 December 2012 - 02:07 PM, said:

The screen froze up again. Possibly it's the graphics card itself; I read somewhere on the internet that microcracks in the pin connecters of the GPU could be a problem, and baking it could be a solution. Would that help?

Right now, NO. And this isn't microcracks... whoever said that needs ignored! PRONTO. I have baked SEVERAL parts in the past - but I KNEW what the problem was and why that fixed it. The 8800GT was VERY well known for being one of the best made cards on the market. The problem that plagued OTHER video cards from that era does NOT plague the 8800GT. These problems were actually cause by soft silicone that would weaken under high temperature loads, and cause the GPU itself to shift. When it shifted, the ball array in-between the layers shifted out of place. Your problem seriously sounds like very poor choices in POWER SUPPLIES.

[SuperSquashMan]

Okay, I'll order the one you suggested. I suppose this is what I get for not doing research before buying thanks a lot for your help!

[LiveBrianD]

waldojim, on 27 December 2012 - 04:58 PM, said:

SuperSquashMan, on 27 December 2012 - 02:07 PM, said:

The screen froze up again. Possibly it's the graphics card itself; I read somewhere on the internet that microcracks in the pin connecters of the GPU could be a problem, and baking it could be a solution. Would that help?

Right now, NO. And this isn't microcracks... whoever said that needs ignored! PRONTO. I have baked SEVERAL parts in the past - but I KNEW what the problem was and why that fixed it. The 8800GT was VERY well known for being one of the best made cards on the market. The problem that plagued OTHER video cards from that era does NOT plague the 8800GT. These problems were actually cause by soft silicone that would weaken under high temperature loads, and cause the GPU itself to shift. When it shifted, the ball array in-between the layers shifted out of place. Your problem seriously sounds like very poor choices in POWER SUPPLIES.

He said he had a 9800GT btw, not an 8800GT.

[waldojim]

LiveBrianD, on 27 December 2012 - 06:01 PM, said:

waldojim, on 27 December 2012 - 04:58 PM, said:

SuperSquashMan, on 27 December 2012 - 02:07 PM, said:

The screen froze up again. Possibly it's the graphics card itself; I read somewhere on the internet that microcracks in the pin connecters of the GPU could be a problem, and baking it could be a solution. Would that help?

Right now, NO. And this isn't microcracks... whoever said that needs ignored! PRONTO. I have baked SEVERAL parts in the past - but I KNEW what the problem was and why that fixed it. The 8800GT was VERY well known for being one of the best made cards on the market. The problem that plagued OTHER video cards from that era does NOT plague the 8800GT. These problems were actually cause by soft silicone that would weaken under high temperature loads, and cause the GPU itself to shift. When it shifted, the ball array in-between the layers shifted out of place. Your problem seriously sounds like very poor choices in POWER SUPPLIES.

He said he had a 9800GT btw, not an 8800GT.

It is the EXACT same chip. It really doesn't matter which model number you use there, as they are BOTH using the G92 GPU, with 512MB DDR 3 ram on a 256Bit wide memory path. The only difference was the video bios revision.

[waldojim]

SuperSquashMan, on 27 December 2012 - 05:06 PM, said:

Okay, I'll order the one you suggested. I suppose this is what I get for not doing research before buying thanks a lot for your help!

You are quite welcome, please let us know if you have any other troubles.

[SuperSquashMan]

So, I got the Corsair today. Seems like a nice quality power supply. But 20 minutes after starting, the same thing happened again. I've attached a picture of my setup; maybe I'm doing something wrong?

[waldojim]

For the sake of asking, have you tried reinstalling the OS once you got the new PSU?

I am not suggesting that there would be any changes here, only that the instability in the past may easily corrupt necessary files. I just want to make sure that is out of the way before we continue trouble shooting. Now that you have a stable PSU, one that can handle that video card, it will be easier to determine any other problems.

[SuperSquashMan]

So I fixed it, kind of.
I really didn't want to reinstall the OS (shared computer + years worth of files), so I started with installing the latest drivers. After that, it didn't recognize the card, so I disabled the on board graphics in my BIOS and it ran fine. I ran the graphics and the computer still crashed, so I restarted it and ran it with GPU-Z in the background. I noticed that the temperature of the graphics card was really hot - over 100 degrees Celcius in some cases. I took the side off the case, and that seems to help somewhat. Now it doesn't crash when playing the average game, but it crashes when I run Furmark (it seems to crash when the GPU temp reaches about 120 degrees celcius).

[waldojim]

Ok, that card should not exceed 95C to remain stable!

First, take out the card, and take off the heat shroud. Then clean the heatsink out completely, and make sure the fan is clean as well. Apply new thermal paste between the heatsink and the GPU, then put everything back together. If your card is the single slot design, this should get your temps back into the 80~85C range. If yours is a dual slot design, you should be down in the 75~80C range. Make sure the fan is running when testing as well.