Antec SP-500 and 8800GT

[panfist]

To make a long story short I just called Antec to ask about the supposed "dual rails" and the gentleman on the phone told me that my power supply is inadequate and the system would not even boot. I told him it's working fine (it's rock solid at stock clock speeds), and he told me that I'm pushing the power supply too far and I need at least a 550W or 600W. Is he just trying to make a sale? Here are the details...

Newest edit: Here are pics of my power supply. It looks like it's in pretty good shape, right?
http://picasaweb.google.com/ses1984/Psu

Edit: Considering that many high-end PSUs don't supply all of the required amps that the 8800GT needs on one rail, I'm pretty certain the Antec rep was at least partially full of shit; however, my FUD remains.

----------------------------

I have an e8400, 2x1GB of RAM, 8800GT, two hard disks, and an 8800GT running off of an Antec SP-500. The specs for this thing are 2x12V rails at 17 and 19 amps (since there aren't actually two rails can you just add the amps?). All of this is plugged into an ASUS p5n32-e SLI plus motherboard. I also have a 120mm front intake fan and 140mm rear exhaust fan. The CPU cooler has a 120mm fan in between the two towers, and the exhaust fan is also about an inch away from the tower cooler.

The CPU cooler feels very cool to the touch, Like barely above ambient when idle, and just a little bit warmer at load. On the other hand the NB is HOT, like you can't hold your finger on it for more than 1 second. Nvidia monitor and my BIOS both report that my CPU is at 50C and my system temp is at 40C. Load temps are reported as 63C but it still feels cool. This doesn't make any sense. I don't trust the CPU temp reading because it's a new processor. The last BIOS I was using was reporting 35C at load, now all of a sudden I'm getting 50C. Could this be because the cooler is not making good contact with the CPU anymore?

Everything seems to be working fine, but I'm having a little trouble overclocking my system. I can't get the system stable anywhere past 3.4GHz and I'm not sure what's causing the bottleneck. Anything past 3.4GHz and it's stable in Orthos for about an hour until it just freezes. It could be...

[*]The "FSB hole" of my motherboard (asus p5n32-e SLI plus).[/li]
[*]I have a tower cooler that doesn't cool the motherboard's NB and SB heatpipes effectively like a downward blowing heatsink would. There is very little airflow over the motherboard heatpipes and fins.[/li]
[*]The cooler isn't making good contact with the CPU? If this was the case wouldn't my load temps just keep climbing until I crashed?[/li]

Or maybe it's the power supply. I called Antec to ask how I could find out which components are on which rails and the answer I got was "if you use more than is provided on one rail you will just suck power from the other one." This sounds ridiculous; the information from the sticky posts here sounded far more credible. But even after reading that, my doubt still remains.

So my main question is...is my power supply adequate? If it's not, I have a ton of research to do because I'd like to get one that could theoretically allow me to put 2x GTX cards in SLI.

My second question is...if my power supply is adequate, how can I push my overclock higher? Do I need to buy anything new? This all started when I thought myself, maybe a downward blowing cooler would help -> maybe my motherboard hits a brick wall at 380MHz+ -> maybe my power supply is pooping. Well, any input would be appreciated.

 

[jonnyGURU]

Originally posted by: panfist
To make a long story short I just called Antec to ask about the supposed "dual rails" and the gentleman on the phone told me that my power supply is inadequate and the system would not even boot. I told him it's working fine (it's rock solid at stock clock speeds), and he told me that I'm pushing the power supply too far and I need at least a 550W or 600W. Is he just trying to make a sale? Here are the details...

Edit: Considering that many high-end PSUs don't supply all of the required amps that the 8800GT needs on one rail, I'm pretty certain the Antec rep was at least partially full of shit; however, my FUD remains.

Really? Many high-end PSU's don't supply all of the required amps that the 8800GT needs on one rail? What PSU's are these? High end 200W power supplies?

Whatever you think Nvidia or whatever graphics card supplier tells you a "card needs", they're telling you what THE WHOLE SYSTEM using that card needs. And whatever they're telling you it needs on the +12V doesn't have to be on one +12V rail because it's not just for the graphics card, it's for the WHOLE SYSTEM. The graphics card alone doesn't even need 6A on the +12V to be perfectly honest.

Can an SP500 support an 8800GT? Sure. But because of the way it's built, I wouldn't give it long.

Originally posted by: panfist
I have an e8400, 2x1GB of RAM, 8800GT, two hard disks, and an 8800GT running off of an Antec SP-500. The specs for this thing are 2x12V rails at 17 and 19 amps (since there aren't actually two rails can you just add the amps?).

Actually, that unit does have two +12V rails and you wouldn't add them up. There should be a "+12V1, +12V2 combined" rating somewhere on the label. I believe it's underneath the DC output table.

Originally posted by: panfist
All of this is plugged into an ASUS p5n32-e SLI plus motherboard. I also have a 120mm front intake fan and 140mm rear exhaust fan. The CPU cooler has a 120mm fan in between the two towers, and the exhaust fan is also about an inch away from the tower cooler.

The CPU cooler feels very cool to the touch, Like barely above ambient when idle, and just a little bit warmer at load. On the other hand the NB is HOT, like you can't hold your finger on it for more than 1 second. Nvidia monitor and my BIOS both report that my CPU is at 50C and my system temp is at 40C. Load temps are reported as 63C but it still feels cool. This doesn't make any sense. I don't trust the CPU temp reading because it's a new processor. The last BIOS I was using was reporting 35C at load, now all of a sudden I'm getting 50C. Could this be because the cooler is not making good contact with the CPU anymore?

Everything seems to be working fine, but I'm having a little trouble overclocking my system. I can't get the system stable anywhere past 3.4GHz and I'm not sure what's causing the bottleneck. Anything past 3.4GHz and it's stable in Orthos for about an hour until it just freezes. It could be...

[*]The "FSB hole" of my motherboard (asus p5n32-e SLI plus).[/li]
[*]I have a tower cooler that doesn't cool the motherboard's NB and SB heatpipes effectively like a downward blowing heatsink would. There is very little airflow over the motherboard heatpipes and fins.[/li]
[*]The cooler isn't making good contact with the CPU? If this was the case wouldn't my load temps just keep climbing until I crashed?[/li]

Or maybe it's the power supply. I called Antec to ask how I could find out which components are on which rails and the answer I got was "if you use more than is provided on one rail you will just suck power from the other one." This sounds ridiculous; the information from the sticky posts here sounded far more credible. But even after reading that, my doubt still remains.

Yeah... he's full of crap. If you pull too much power from one rail... the PSU shuts off.

Seasonic makes a number of PSU's that claim to have split +12V but only really has one, and the lie they perpetuate that many of their partners (Antec, etc.) also use is "if one rail is overloaded, it gets power from the other." That's probably the mantra the guy you talked to is used to regurgitating because he may not know any better. What's funny about that whole concept is it would defeat the purpose of splitting the +12V rails up in the first place. Seriously... why spit the +12V if one rail can just borrow power from the other? That's like building a dam with a huge hole in it.

The SmartPower isn't and never was made by Seasonic. It has two +12V rails with one rail (17A) going to the CPU and the other rail (19A) going to EVERYTHING else.

Originally posted by: panfist
So my main question is...is my power supply adequate? If it's not, I have a ton of research to do because I'd like to get one that could theoretically allow me to put 2x GTX cards in SLI.

Your power supply should be adequate. But it's old. They're known to have problems with overheating and buldging caps. If a PSU runs hot, it's ability to put out power diminishes. And if a PSU has failing caps in it, then it can have poor voltage regulation and/or high ripple. Either is possible in your case.

Ironic really, you have all these new expensive parts, 6 month old graphics card and a $300 CPU, and a cheap, old PSU.

You can remove all doubt and open up the PSU and look for swollen caps. You can also test the PSU's leads with a multi-meter and see if you're getting voltages within spec (some basic troubleshooting here.)

Originally posted by: panfistMy second question is...if my power supply is adequate, how can I push my overclock higher? Do I need to buy anything new? This all started when I thought myself, maybe a downward blowing cooler would help -> maybe my motherboard hits a brick wall at 380MHz+ -> maybe my power supply is pooping. Well, any input would be appreciated.

Like I said, it's not enough that your PSU is "enough" if it can potentially be failing. Troubleshoot the PSU to the best of your ability first. Or pull the trigger and just get something half way decent in there and call it a day.

 

[panfist]

Originally posted by: jonnyGURU
The SmartPower isn't and never was made by Seasonic. It has two +12V rails with one rail (17A) going to the CPU and the other rail (19A) going to EVERYTHING else.

...

Ironic really, you have all these new expensive parts, 6 month old graphics card and a $300 CPU, and a cheap, old PSU.

...

Like I said, it's not enough that your PSU is "enough" if it can potentially be failing. Troubleshoot the PSU to the best of your ability first. Or pull the trigger and just get something half way decent in there and call it a day.

I edited this section because I re-read the sticky about split rails. When I saw the amperage rating required by the 8800GT I thought it meant just how much the graphics card draws by itself.

When I said even high end power supplies don't have enough on a single rail, I did not have the above fact in mind. It all makes sense now.

---------------

I have a cheap-o digital multimeter from radioshack that I can use to troubleshoot my PSU, but what do I do if I check it out and everything looks good according to the multimeter? Then can I safely assume my PSU is good for a while longer?

I don't mean to get defensive but can you point me to some evidence that the build quality on an SP-500 is inferior? I have a 2 year old $50 PSU that has served me well through an overclocked x1900XT, overclocked pentium D 930, then overclocked e8400 and overclocked 8800GT. If it lasted me through an x1900XT and pentium D 930...why would it die now? It's quiet, it's not hot. I also put one in my girlfriend's computer and it's working well for her. So...what's so bad about it? Even if the PSU is really dieing on me wouldn't pretty much any PSU be in bad shape after 2 years of hard use? Maybe mine has lasted so long because I have really good case airflow. The only place I have bad airflow is right under the tower, where the motherboard heatpipes/fins are.

--------------

(2)
On a side note, I never really had the perception that I had expensive parts thrown in with a crap PSU. My motherboard was $80. I also go the e8400 for $240 shipped with next day service the day it was available. I could have paid $210 if I wanted to wait a week for shipping...so that's not really a $300 processor. My RAM was $40 after MIR. For my CPU heatsink I have a thermalright sonic tower for $35 I bought at least a year ago, which is basically a Tuniq Tower clone. Load temps on my 3.8GHz pentium D were under 45C...unless those were being reported inaccurately, too, but it always felt cool. Anyway, I was always very proud for how far I stretched my dollar. I saved money on all the components in my system so I didn't see anything wrong in choosing to save on my PSU, too. Did I just get lucky that flames haven't shot out of my case yet? lol

--------------

Lastly I would like to say thank you for the very informative reply. If it sounds like I'm arguing, I guess I'm just trying to converse or get you to strengthen your position. One question I would like to ask is what power supplies are halfway decent? I think i can probably get away with a good 400W supply, right? If I do have to spend any dough on a new PSU, I want to get one that will last me at least the next two years like this one did. I also would like to get one that's not just adequate for the parts I have now, but "SLI proof" i.e. would let me install two fat video cards in SLI.

Oh and one more question...Is there any way I can reliably measure the temperature of my CPU?

 

[jonnyGURU]

If you're seeing good numbers with your DMM under load, I would still open up the PSU for a visual inspection. These SmartPowers are KNOWN to blow caps. If it looks good with a visual, I'd assume the PSU is good for the time being and not the reason behind your limited OC.

 

[panfist]

Here are some pics of my power supply. Upon visual inspection, at least to my eye, the caps look fine. I'm no electrical engineer but I have taken a few courses on electronics in college. I guess the next step is to apply a multimeter to some pins at load and idle, according to the troubleshooting guide you linked earlier. I'll save that for the weekend.

http://picasaweb.google.com/ses1984/Psu

Edit: the pics are 4 megapixel if you want to zoom in.

Edit 2: After reading the HardOCP review I'm really interested in the ES-800. Where can I get my hands on one? I was considering the OCZ 700 that's featured on the hot deals forum but it doesn't seem to be beefy enough should I want to SLI the upcoming GT200. But then again the ES-800 probably can't handle 2x GT200. I also probably won't be able to afford 2x GT200 until something better is out. I'm just rambling now...

 

[scheibler1]

I have an 8800gt OC, 1hardrive, 1 dvd/rw, and an e2180 @ 3ghz with stock voltage runnign fine on an ANTEC Earthwatts 380w!

It has been running solid for two months without one problem. I am replacing it with a 650w corsair tx650 tomorrow when it arrives though. I wouldn't worry about it at all if you aren't going to SLI

 

[jonnyGURU]

Originally posted by: scheibler1
I have an 8800gt OC, 1hardrive, 1 dvd/rw, and an e2180 @ 3ghz with stock voltage runnign fine on an ANTEC Earthwatts 380w!

I wouldn't worry about it at all if you aren't going to SLI

Right.

YOU have an EarthWatts.

HE has a SmartPower.

You don't know the difference?

Also, it's a good thing you're replacing the unit now because I'd give that 380W another year running that system like that and it'll probably fail.

Many undersized power supplies will work right out of the box and will continue to work for over year. People don't seem to take into consideration that pushing these units, running then at higher temps, etc. shortens their life. All day long we can tell people "a 380W will run an 8800GT just fine" but at the end of the day, is it wise when it's likely the PSU won't last very long?

 

[panfist]

Ok I checked this out with a multimeter and voltage regulation seems fine, except the base idle voltage is pretty high, close to the end of the spec. I guess Antec is not very confident in the voltage regulation?

At idle, my 12V line is steady at 12.49 - 12.50. This is pretty high and almost out of spec, right? When I run Orthos blend test it drops to 12.46 then stabilizes at 12.48. Hmmmm but there are two fans attached to that line. I'm stopping Orthos, unplugging the fans, and repeating.

Without the fans connected I'm at 12.52V idle. When I run Orthos it drops to as little as 12.48 while paging before stabilizing to 12.50. Now I'm about to stop Orthos blend test, start Orthos CPU test and also Supreme Commander performance test @ 1680x1050 max settings with 2x AA to max out the video card too.

Interesting...I just began Orthos CPU test and my voltage dropped to 12.45V, now 12.46V. I get better voltage regulation when my RAM is full? My CPU temp is steady at 68C without a fan (according to whatever fucked up scale exists between the penryn and my motherboard).

So when SupCom started to load game data it dropped to as low as 12.44V and as high as 12.49V while the hard disk was chugging. As soon as the 3d engine initialized it dropped to 12.42V and then began to fluctuate between 12.42V and 12.45V, I think it hit 12.46V once throughout the benchmark. Again this is while running Orthos CPU test.

So...how do my numbers look? Is there going to be smoke and sparks shooting out of my case any time soon? lol

 

[jonnyGURU]

Those numbers are fine. Upper end of spec is 12.6V so you're no where close.

 

[panfist]

I feel way better that my PSU has passed the basic diagnostics. Thank you so much for all your help.

I thought 12.5 was close to 12.6 last time I checked. I know from an electrical engineering point of view it's well under 12.6 but considering the wide range, 11.4-12.6, it seems to be sitting a little near the high side. I would be much more happy with voltage regulation on the order of 12.2V idle 12.1V load.

Should I check out any other lines, like the 5v or the 3.3?

I feel pretty safe about my PSU at least for the time being. I'm over in the cases and cooling forum now talking to some people trying to figure out what's bottlenecking my overclock. Right now I'm stable@ 6x multi, 400MHz FSB, just trying to see if my motherboard can hit 400MHz when my CPU is far from stressed. I'm reading very similar voltage regulation at this clock configuration. Do I need to bother testing my voltage regulation when I'm underclocking my CPU or can I leave that until after I see how high I can max my northbridge. After I max my northbridge at 6x multi I'm going to try to build up to that same NB speed slowly.

http://forums.anandtech.com/me...id=2173539&STARTPAGE=1

 

[jonnyGURU]

Originally posted by: panfist

Should I check out any other lines, like the 5v or the 3.3?

Absofreakinlutely!!

 

[panfist]

Ugh...ok...it'll be a little harder to test the 5V and 3.3V line.

Since I moved my cables around I only gave myself free access to an extra PCIe connector...so I only can get to the 12V line via a cable. Can I test the 5v and 3.3V line anywhere else? Maybe by sticking leads directly into one of the modular connectors on the PSU? My motherboard 24pin connector does not detach to 20+4.