PDA

View Full Version : Bad idea?


MooseMuffin
07-18-2007, 04:46 PM
Hi all. I have a 500w antec (http://www.newegg.com/Product/Product.asp?Item=N82E16817103937) that I purchased close to 2 years ago now. My relelant system components are:

athlon 64 3700+ (socket 939)
1gb ddr
2 hds
2 optical drives

I think my system has one last upgrade left in it before its time to dump it, and an a64 x2 and a 2nd gig of ram wouldn't cost me very much at all. My question to you guys is how high up the graphics card totem pole can I go? Would a 8800gts kill the poor psu?

Super Nade
07-18-2007, 04:53 PM
SP series = Fujhhyu caps
Fujhhyu caps + heat = magic smoke.

You do the math.

:D

jonnyGURU
07-18-2007, 05:38 PM
haha... ok.. well....

To put it more clearly... a PSU like that that's two years old might have some damage that could effect it's performance. If you don't want to open it and inspect it, it's probably best to just assume that something may go wrong in the near future.

MooseMuffin
07-18-2007, 05:38 PM
Despite my limited PSU knowledge, I think that post means no go on the 8800. So what can I put in there then?

Edit: To clarify, I'm not looking to overhaul this thing. The idea was that a cheap cpu/memory upgrade gets me real close to a respectable rig. My psu knowledge has grown considerably since I purchased the thing and I realize its not a quality unit. Pushing the psu too hard is likely gonna kill my rig so I'm looking for the upper limit of video card I can stick in there with minimal risk. If it lasts me a year before blowing up, I'll be fairly satisfied.

Super Nade
07-18-2007, 05:45 PM
A new PSU? :D

Per Hansson
07-18-2007, 05:59 PM
The PSU is more than sufficient for the best (sans SLI) card you can find

However, as the others have said the capacitors most be changed first. And that is no matter if you buy a new GFX card or not. If you do not fix the caps the rest of your computer will have to deal with all the ripple current that PSU will output. Causing failures down the line (especially bad for HDD's for example which have no regulation at all!)

But eventually the VRM filtering caps on your mainboard will fail too, because they become overloaded... So do yourself a favor and fix the issue now rather than later

And to recap (haha) the PSU is of excellent quality, it's just the choice of capacitors in it that makes it a poor PSU, but that can be fixed quite cheaply if you know how or lend it to someone that does

Super Nade
07-18-2007, 06:20 PM
Another alternative would be to send it back to Antec and they will replace it with a Neo series unit.

MooseMuffin
07-18-2007, 07:47 PM
Send it back on what grounds? Its 2 years old and not broken. I get the impression that these Fujhhyu capacitors are crap, but both Jonny and Per Hansson have mentioned damage and fixing. Were these things recalled or something? Like I said, I haven't experienced any problems.

clsA
07-18-2007, 07:56 PM
Send it back on what grounds? Its 2 years old and not broken. I get the impression that these Fujhhyu capacitors are crap, but both Jonny and Per Hansson have mentioned damage and fixing. Were these things recalled or something? Like I said, I haven't experienced any problems.

There was no recall their just know bad units..if your's hasn't been abused and overheated it will last a while longer, but it's still a Time Bomb.

if you want to get a decent Video card that needs little voltage try a Geforce 7600GT with 256mb DDR3

signmeuptoo
07-18-2007, 09:51 PM
To understand some of what they are talking about:

Ripple: In electronics it is the equivalent of a type of "noise", usually a higher frequency low voltage flucuation, not from unstable RAILS, but from signals getting past the electronic filtration that is supposed to protect your hardware from stress. Think of ripple like Chinese water torture, but possibly faster to damage.

You could always sell the supply, but you might have too much of a conscience for that, I know I do. I would try telling them that your system is getting phantom reboots and you have changed your other hardware, that might be one way to get it replaced. Not honest, but that is a call you have to make. Considering that they probably should have sold you a supply with better caps, maybe it is fair, I don't know, personally I have never lied to get a part replaced, but...

Capacitors are used to filter, isolate, and to perform "capacitive" effects. Caps are, in some ways, like a short store battery, or a buffer/ballast like thing, or certain other things, they can be used in circuits to "decouple" things, to "smooth out signals", and to cause certain electrical affects that are needed for a circuit to work in the way needed.

Your supply has some very big caps in it.

I have an Antec that I believe the primary cap blew in.

It's your call, but these fellows are giving you very valuable advice!

Super Nade
07-18-2007, 11:08 PM
This is what a one year old unit in a poorly ventilated case looked like:
http://www.ocforums.com/showthread.php?t=495380

MooseMuffin
07-19-2007, 09:13 AM
Interesting stuff. Thanks for all the advice everyone. I don't really feel right returning it, but I guess I won't be adding any more stress to the poor thing either. Everything in this box is between 2 and 5 years old anyway, so I'll probably just build a new one within the next few months.

Per Hansson
07-19-2007, 08:01 PM
I'd for shure return it

Antec needs to understand that they can't sell crap like this because it will come back and bite them

It's not your fault they used crap components for the PSU, it's the beancounters at Antec...

MooseMuffin
07-25-2007, 11:37 PM
Hey all. Well, I went ahead with the cpu/ram upgrade and left the video card alone and wouldn't you know it, I'm running into issues. Put the new parts, powered on and....no boot. Try again, nothing. Cross my fingers and try a 3rd time...and it comes up! Don't like that very much, but sure enough it boots into windows, I see 2 cores, 2gbs of ram and confirm my temps are ok. Then it freezes during some browsing.

To make a long story short(ish), I discover that I can't boot unless my ram is running at ddr 333 or lower. ddr 400 results in absolutely nothing, and while 333 boots, its unstable and freezy. I think about it for a bit and conclude that maybe my 2nd gig of ram is getting along with the first. They had the same specs, but they're different brands and all so I pull the new ram out. Nada at 400. Boots at 333 and eventually freezes.

Obviously the next step is to put the old cpu in, but swapping cpus is a pain in the ass with the heatsink removal/cleaning/paste re-application, so before I do that I want to ask for some opinions. Can this be a symptom of my crappy psu? The dual core proc pulls enough extra power to prevent running ram at speed? Is there some software I can use to monitor voltage, and if so what am I looking for?

signmeuptoo
07-25-2007, 11:46 PM
Hey moose:

Get ahold of the free self booting application memtest86+

http://www.memtest.org/

Get the ISO version.

Next, using a program such as NERO, choose to "burn IMAGE" to disk, on a blank CD.

Next, unplug computer power and pull all but one stick, check to see which slot should be used when running one stick, it will be in your manual which can be downloaded if you don't have it.

Boot while going into your BIOS set BIOS to boot off of Optical drive and put the burned disk in the drive in question.

Reboot to the disk. Memtest will come up instead of Windows and run the full battery of tests for at least 6 hours per stick, testing only one stick at a time.

NOTE: Make sure you set your BIOS to Auto settings.

I hope you didn't get sold different RAM than what you think, that could be one cause...

MooseMuffin
07-26-2007, 12:00 AM
Already ran memtest actually. Ram seems fine. And like I said above, I removed the new ram and got the same crappy result with my old ram that I know is good. The only variables left are the new cpu being dual core, and the odds that my PSU decided to crap the bed.

Super Nade
07-26-2007, 12:09 AM
Moose, what MB are you using? If it is a board that allows tweaking of drive strength's, I'd go ahead and tweak it. RAM settings vary from CPU to CPU.

MooseMuffin
07-26-2007, 12:16 AM
asrock dual939-sata2. I read the (f-ing) manual to see if there were some jumpers or settings that needed to be changed and found nothing. Theres a setting for dram voltage that doesn't seem to affect my problem. Whats drive strength? That one is new to me.

signmeuptoo
07-26-2007, 12:32 AM
Here is an example:

http://adriansrojakpot.com/Speed_Demonz/New_BIOS_Guide/CPU_Drive_Strength.htm

I don't know much about it myself, I think it is the strength of the bus itself?

Super Nade
07-26-2007, 12:38 AM
Use A64Tweaker and A64Info to tweak the RAM a bit. ;)
http://avala.yubc.net/~lukija/A64Info-beta.exe

For example this is what I'm at:-
http://img528.imageshack.us/img528/7035/tccdlf4.th.png (http://img528.imageshack.us/my.php?image=tccdlf4.png)

Per Hansson
07-26-2007, 08:03 AM
It is very likley that your PSU is giving you troubles

Either send it back to Antec for RMA or have it recapped

MooseMuffin
07-26-2007, 12:24 PM
It very well could be the PSU, but theres a few things that make me think it isn't. If I can't boot with the ram running at 400 but I can at 333, I would think that means I'm pushing the limits of the PSU. But at the 333 configuration I've been able to run a pair of prime95 instances for about 10 minutes without error, only to have it crash later on during some webbrowsing.

Crashing at load would support the PSU theory, crashing randomly is...something else?

Super Nade
07-26-2007, 03:00 PM
It very well could be the PSU, but theres a few things that make me think it isn't. If I can't boot with the ram running at 400 but I can at 333, I would think that means I'm pushing the limits of the PSU. But at the 333 configuration I've been able to run a pair of prime95 instances for about 10 minutes without error, only to have it crash later on during some webbrowsing.

Crashing at load would support the PSU theory, crashing randomly is...something else?

Crash is a crash no matter how you see it. Either 10 min of browsing or 10 min of 3DMark...all the same. There are just too many variables at this point. Perhaps your Windows install is trashed?

MooseMuffin
07-26-2007, 06:23 PM
I was at work all day and unable to mess around with this more, but now that I've had some more time I've pin pointed the problem. It wasn't something that I even realized that happened but here it is. I realize some of this clashes with what I said last night, but I can't argue with the current evidence.

So I bought 2 sticks of new memory (new1 and new2 lets say) to go with my old stuff.

2x old stuff = works
2x new stuff = works
2x old stuff + 2x new stuff = broken
Any combination of old stuff and new1 = works
Any combination of old stuff and new2 = broken

So it turns out I have one module that works alone, works with its partner, but is incompatible with the old stuff. Its identical partner gets along with the old stuff just fine. Like I said, I didn't even know that happened. My question now is that if I return it, how do I lower the chances of this happening again? I was sold 2 working sticks of ram and one of them doesn't play nice with others.

MooseMuffin
07-26-2007, 07:04 PM
Last question before I leave you folks alone. Are speedfan's voltage readings accurate? And if so whats the acceptable voltage range for the 12v rail?

signmeuptoo
07-26-2007, 07:16 PM
Ignore the negative voltages, they are usually not accurate in my experience. This is because they often aren't used on motherboard let alone not even there.

LC_STiK
07-26-2007, 07:17 PM
The readings are perfectly accurate. But the reading is only as good as the voltage sensor, so it might end up not being all that good.

It should 11.4-13.6 if I remember correctly.

DMM will do the job for real though.

If you're running 4 sticks of RAM and it's giving you trouble, make sure you've got it set to 2T or it could have troubles.

1Tx4 sticks usually isn't great for stability. I would doubt the PSU b/c RAM basically doesn't take much power at all compared with the gfx card or the CPU. It's more or less insignificant.

Kvar
07-26-2007, 08:26 PM
Ignore the motherboard voltage readout. They're often wrong.

Poke the DMM probes directly into the back of the 20/24 pin connector and measure each voltage.
black lead = black wire.
Red lead = red wire for 5v, orange for 3.3v, yellow for 12v. -12 and -5v aren't really relevant so you don't need to measure them

If there's nothing wrong with the readings, there's no reason to change it, blow out the dust, then put the case back together. I had my computer on Antec PP-253X 250W.

I was running 3HDD, Athlon XP 2400/2GHz, 3 sticks ram (1.5GB) and basic graphics card. I changed it with 350W that has more Molex connectors when I needed another connector when I added a DVD+RW drive.
Considering the system pulls 150W from the plug, 200W for the first few seconds while all the HDDs are spinning up, that power supply was more than enough for this computer.