Power supply
in [Hardware]
|
Prev: 939 m/b advice recommends
Next: Can't read A drive
From: LeeG on 11 Jan 2010 18:29 Hi paul. On all cards the fans were/are running okay and no temp issue, (this was one of my first concerns - checked the whole system). MY mobo can be overclocked but not doing that right now. Need to stabilise the graphics card before attempting anything like that. I can't find any tech data on the psu I have. With the graphics card drawing its power from the mobo instead of having a seperate molex power supply, is this what the advice concerning a new power supply is based on. Too much power being drawn from rail one. "Paul" wrote: > LeeG wrote: > > Hi Paul. > > > > A brief resume of my problem. Built new system. Graphics card (inno3d > > GT220 1GB)went BSOD on me. RMA'd replacement which BSOD'd. Rma'd new > > card(as stated). No BSOD as yet but some graphics corruption after a short > > time of running 3D games. PSU is WIN POWER 450W. Been told I need 32A on > > 12V for this system on a single rail. > > > > When problem first happened I ran Prime95 and memtest to see if the new > > build was the cause of the BSOD. All tests where okay. I can run my 7300GT > > with no problems. I do not want to damage this new card and want to > > hopefully find the problem before this happens. > > I checked the reviews on Newegg, for various GT220 cards, and I don't see > a lot of problems mentioned there. > > Is the fan on the graphics card running OK ? > > What about the overclocking situation. Is the card clocked higher than the > competition ? Does the card have a temperature sensor ? (I can't tell here, > because the other tabs in GPU-z aren't shown.) > > http://www.xtremecomputing.co.uk/images/review/Inno3D_GT220/inno3d_gt220_12.jpg > > You can get GPU-z here. It is under 500KB. > > http://www.techpowerup.com/downloads/1709/TechPowerUp_GPU-Z_v0.3.8.html > > I don't know if SpeedFan has a working GPU temperature readout or not. > My video card is too old for such things :-( > > http://www.almico.com/speedfan440.exe > > There is some mention here of support for Nvidia video cards. > > http://www.almico.com/sfhistory.php > > Since you've tried two cards, I assume the video card cooler was attached well > on at least one of them. Maybe if you can manage to read out the temperature > of the GPU on the video card, you'll have a better idea whether there is a > temperature issue. > > Paul > > > > > "Paul" wrote: > > > >> LeeG wrote: > >>> I have been advised to get a better PSU because the ampage on the 12V is too > >>> low (20A) for my system. This came about due to corruption of my graphics > >>> when running 3D games. > >>> > >>> Where can I verify this? > >>> > >>> Phenom II x2 550 black > >>> MSI 770-c45 Bios v1.4 > >>> PNY 1GB gt220 PCIe graphics card (no 12v power socket) > >>> 2 x 2gb crucial 10600 memory > >>> 1 x usb printer > >>> 1 x usb All in one > >>> 1 x usb joystick > >>> Microsoft deskset 700 usb > >>> 1 x SATA combo optical drive > >>> 1 x SATA WD 320gb HDD blue > >>> 1 x CPU fan > >>> 2 x system fans > >>> > >>> > >> Phenom II x2 550 black 80W, 80W/12V * (1/0.90) = 7.4A from 12V2 > >> (Assumes 90% Vcore conversion efficiency.) > >> > >> LOL :-) GT220 = 22.9W max! 22.9W/12V = 1.9A from 12V1 > >> Overload. Um, sure... There are ten year old video cards that draw more power. > >> This isn't even going to stress the pins in the video card slot. > >> > >> http://www.xbitlabs.com/articles/video/display/gf-210-gt220_5.html#sect0 > >> > >> HDD = 12V @ 0.6A and 5V @ 1A. 12V would be from 12V1 rail > >> ODD = 12V @ 1.5A and 5V @ 1.5A. 12V would be from 12V1 rail > >> > >> Estimate of all fans - 12V @ 0.5A (check what is printed on the hub of each fan) > >> The fans run from 12V1. > >> > >> Now, assuming we used an older supply, where all the current came from one > >> rail, 7.4+1.9+0.6+1.5+0.5 = 11.9 amps total from 12V rail > >> > >> If the power supply was split into 12V1 and 12V2, and had current > >> limiters inserted in each rail, I suppose you could come up with > >> a way to make the box fall over, but somehow I doubt this is the problem. > >> Post a picture of the label of the power supply (find one on the Internet and > >> post a link) if you want more info. > >> > >> ******* > >> > >> The corruption of the graphics, could be a bad graphics card. > >> > >> My advice would normally be, to set up an alternative testing environment. > >> Either compare your existing Windows install with its graphics problems, > >> to a fresh install on a second hard drive (to eliminate an obscure problem > >> with the files currently on the machine). Or, use a separate OS and > >> test there. For some kinds of problems, I use Linux, but graphics is > >> not one of its stronger suits. It is pretty hard to make a stressful > >> test in Linux, where you can be certain the GPU is getting a workout. > >> > >> If from one OS boot, you get corrupted graphics, and in the other > >> environment, the corruption is gone, you'd suspect the original OS > >> as having a problem of some sort (driver or whatever). If both environments > >> (say comparing Windows to Linux) had corruption problems, you'd know immediately > >> to return the video card to the retailer. I had one computer here, where > >> even sitting idle in a Linux desktop, I could see a graphics problem, > >> so sometimes you get lucky. But if the video card problem is only > >> present under stress, Linux has a hard time duplicating the stress of > >> a Windows setup. > >> > >> At 22.9W, it is hard to blame the power supply. > >> > >> If you want another test, try running Prime95 and getting that 80W processor > >> to run flat out. Your processor can draw more power than the video card, > >> so if it was a power supply problem, you might see a problem while the > >> CPU was loaded up. See if the box crashes with that test running. > >> > >> http://www.mersenne.org/freesoft (p95v259.zip) > >> > >> If you keep getting errors in Prime95, you have issues other than the > >> video card. > >> > >> Paul > >> . > >> > . >
From: Paul on 11 Jan 2010 19:31
LeeG wrote: > Hi paul. > > On all cards the fans were/are running okay and no temp issue, (this was one > of my first concerns - checked the whole system). MY mobo can be overclocked > but not doing that right now. Need to stabilise the graphics card before > attempting anything like that. I can't find any tech data on the psu I have. > With the graphics card drawing its power from the mobo instead of having a > seperate molex power supply, is this what the advice concerning a new power > supply is based on. Too much power being drawn from rail one. Your 12V2 is 7.4A (that is the 2x2 cable powering the processor). Your 12V1 is the rest, which is 4.5 amps. Only 2.4 amps of that (video card plus fans) comes through the main power supply cable. The rest of the current is via Molex hard drive power connectors and the like. The currents involved are pretty small. If the power supply wasn't working properly, and was delivering abnormally low voltage, or was placing electrical noise on a voltage rail, that could cause problems. But the thing is, you note video corruption, without the computer crashing. Which suggests a problem local to the video card itself. You would expect a wider variety of symptoms, like other things complaining about power. (A hard drive, for example, will spin down and spin up again, if it detects even a slight drop in the 12V feeding it.) But without any other symptoms other than glitches on the video card output, I'd have to assume it is some problem with just the video card. And drawing 1.9 amps through motherboard copper tracks, shouldn't be an issue - I don't think this is going to be the result of a motherboard design issue. You could use a multimeter to verify the voltages from the supply if you want, but most people end up unsatisfied after doing those measurements. (Problems are seldom detected with just a multimeter.) I have a clamp-on DC ammeter, and that is handy for detecting accidental shorts or overloads. It is a nice instrument, but at about $300, isn't a good investment for a home repair situation. A $20 multimeter from the hardware store is what most people can afford, and making only voltage measurements with one, doesn't usually lead to an "aha!" moment. My clamp-on ammeter is contactless - you clamp it around a bundle of wires, and it tells you how much current is flowing. In about 30 seconds of measurements, I have a pretty good picture of current flow from 3.3V, 5V, 12V and so on. Measuring the voltage may sometimes show a power supply with a problem, but I wouldn't expect it to be that easy most of the time. Do you have any other symptoms to add ? Are the BSODs ? Anything showing up in the Event log ? Perhaps it boils down to you getting two bad cards in a row. Another question for you - did you remember to uninstall the 7300GT driver, while the 7300GT was still in the computer ? Then, install the new video card, and install a driver for it ? If you've been flipping cards in and out, without changing drivers, I can't imagine the software being very happy. I learned that the hard way. I kept changing video cards, leaving the old drivers in, and just installing over top when the next video card came along. Then one day, I couldn't get one of the acceleration settings to work on my video card any more. I tried driver cleaners, I tried uninstalling drivers and reinstalling them, and nothing fixed it. I had to reinstall the OS to clean up the mess. Now, I'm a bit more careful, to remove the old driver, before installing the new hardware. Paul |