SPCR Power Supply Test Rig, v.4
This is beginning to look more and more hi-tech.
But I still have to say it. The results from SPCRs test bench are with few exceptions the only results that I trust, both regarding efficiency and noise, and now performance (ripple) too.
I was wondering one thing though. Is it possible that you could maximize the 12V outputs to see whether they will cut of at their rated outputs (i.e. "separate" 12V rails) or if they will go even further to their combined maximum?
This could be important for some people, like those using SLI or CF platforms. One "rail" is used entierly for the CPU, while TWO GPUs (with some being even more power hungry alone than any CPU today) and the rest of the system will share one rail. If a PSUs cut of at say 18A, it would be impossible to use a CF configuration using two X1900XTX, since they alone consume 20A of 12V power.
But I still have to say it. The results from SPCRs test bench are with few exceptions the only results that I trust, both regarding efficiency and noise, and now performance (ripple) too.
I was wondering one thing though. Is it possible that you could maximize the 12V outputs to see whether they will cut of at their rated outputs (i.e. "separate" 12V rails) or if they will go even further to their combined maximum?
This could be important for some people, like those using SLI or CF platforms. One "rail" is used entierly for the CPU, while TWO GPUs (with some being even more power hungry alone than any CPU today) and the rest of the system will share one rail. If a PSUs cut of at say 18A, it would be impossible to use a CF configuration using two X1900XTX, since they alone consume 20A of 12V power.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Thanks for the vote ofconfidence.Ackelind wrote:But I still have to say it. The results from SPCRs test bench are with few exceptions the only results that I trust, both regarding efficiency and noise, and now performance (ripple) too.
see devon's post below.I was wondering one thing though. Is it possible that you could maximize the 12V outputs to see whether they will cut of at their rated outputs (i.e. "separate" 12V rails) or if they will go even further to their combined maximum?
Last edited by MikeC on Tue Oct 31, 2006 8:48 am, edited 1 time in total.
-
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
I suspect that you may end up seeing those tests done implicitly on higher capacity power supplies, since we can't split the load properly (and power supplies that boast >2 rails rarely document how the rails are split anyway). You may have noticed that our highest power test of the M12 pushed ~21A on +12V1, which was certainly beyond what it was rated for. This wasn't a problem, since there's only one rail there anyway...
We have run into this problem before... an earlier revision of the SilverStone Element that we saw had shutdown problems under high load that were traced to overly sensitive OCP on the +12V lines. SilverStone fixed the problem and sent us a newer unit, which became our official review sample.
We have run into this problem before... an earlier revision of the SilverStone Element that we saw had shutdown problems under high load that were traced to overly sensitive OCP on the +12V lines. SilverStone fixed the problem and sent us a newer unit, which became our official review sample.
-
- Posts: 107
- Joined: Sat Feb 05, 2005 3:45 am
- Location: Moscow, Russia
- Contact:
-
- Posts: 5
- Joined: Sat Sep 02, 2006 6:47 pm
- Location: Connecticut
- Contact:
Uh, reducing the voltage via a DC-DC converter isn't done by simple subtraction of 10.5 volts (which is what the quoted sentence implies), but rather by something resembling a switching power supply, involving an additional layer of voltage regulation. Thus the accuracy of the 12V supply is probably less critical than the accuracy of the other supplies, which are fed directly to chips, with only some capacitors to smooth out their voltage levels.However, DC-to-DC conversion on motherboards to drop the voltage from 12V down to the <1.5V level required by current processors means that the same 120mV of ripple can become proportionately quite high, nearly 10%.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Thanks for the clarification.Norman Yarvin wrote:Uh, reducing the voltage via a DC-DC converter isn't done by simple subtraction of 10.5 volts (which is what the quoted sentence implies), but rather by something resembling a switching power supply, involving an additional layer of voltage regulation. Thus the accuracy of the 12V supply is probably less critical than the accuracy of the other supplies, which are fed directly to chips, with only some capacitors to smooth out their voltage levels.However, DC-to-DC conversion on motherboards to drop the voltage from 12V down to the <1.5V level required by current processors means that the same 120mV of ripple can become proportionately quite high, nearly 10%.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Just for fun.... PSU tester V4.1 or Bonefish edition.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Well, he was taking a hammer to a motherboard he hated and in his frenzy, managed to nail a Neo in the head.... RIP.qviri wrote:We want the full story!We happened to have a NeoHE that Devon inadvertently killed some time ago.
uh.... just kidding.
IRRC, he somehow managed to hit the power switch on the PSU (in a system) multiple times in a few seconds -- on/off/on/off/on/off....... -- and it stopped working after that. I had a look at the underside of the PCB: No obvious scorch marks. It could even be just a fuse; there's a self-resetting one in there. Maybe it burned out and replacing it could bring the PSU back to life. Dunno, don't really care any more.
-
- Posts: 2198
- Joined: Thu Feb 10, 2005 11:20 am
- Location: TN, USA
I love it.MikeC wrote:We happened to have a NeoHE that Devon inadvertently killed some time ago.
IRRC, he somehow managed to hit the power switch on the PSU (in a system) multiple times in a few seconds -- on/off/on/off/on/off....... -- and it stopped working after that. I had a look at the underside of the PCB: No obvious scorch marks. It could even be just a fuse; there's a self-resetting one in there. Maybe it burned out and replacing it could bring the PSU back to life. Dunno, don't really care any more.
You know, I really appreciate the time, effort, honesty, etcetera that you guys throw into your work. If you weren't several days drive and a border crossing away I'd be up there at least once a year to lend a helping hand if you wanted the help.
Heck I'd love to just be able to borrow time on the test rig (with your acoustic tools) for a day to test my own stuff once every two years or so.
Alas it's no less than a 4 day drive each way 2750 miles/4420 kilometers by road(Southeast US to Vancouver CA and back) and that isn't even considering cost.
Even by Air it'd be a full days travel (11-14 hours in the air or airports) plus the ground travel at each end.
I wonder if SPCR should be testing such high-capacity PSUs at all? After all, if you are pumping almost a kilowatt into an average ATX case (or even a P180) getting <30dBA @1m is going to be almost impossible without exotic cooling. And even with AMD 4x4 + SLI 8800GTX + lots of hard drives 850W is the realistic maximum.And that's our PSU tester, Bonefish edition. The end result is load capability of nearly 1.1 kilowatt. Hopefully, we won't be testing such loads often.
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Yeah, I had such thoughts too. Certainly it's not about noise. It's about promotional value, mostly. If/when we test a kilowatt psu, we'll have new readers come check us out, and then some of them might stick around for the rest of our quality site.jaganath wrote:I wonder if SPCR should be testing such high-capacity PSUs at all? After all, if you are pumping almost a kilowatt into an average ATX case (or even a P180) getting <30dBA @1m is going to be almost impossible without exotic cooling. And even with AMD 4x4 + SLI 8800GTX + lots of hard drives 850W is the realistic maximum.
-
- SPCR Reviewer
- Posts: 1850
- Joined: Sun Sep 21, 2003 11:23 am
- Location: Vancouver, BC, Canada
Actually, it wasn't the switch on the PSU, it was the switch on my power bar, which was rapidly cycled when I plugged in a Power Angel on top of it. For the record, we tried replacing the internal fuse, and the new one blew as soon as we plugged it in. I definitely damaged something.MikeC wrote:IRRC, he somehow managed to hit the power switch on the PSU (in a system) multiple times in a few seconds -- on/off/on/off/on/off....... -- and it stopped working after that. I had a look at the underside of the PCB: No obvious scorch marks. It could even be just a fuse; there's a self-resetting one in there. Maybe it burned out and replacing it could bring the PSU back to life. Dunno, don't really care any more.
-
- Posts: 24
- Joined: Tue Nov 01, 2005 12:40 am
- Location: Sydney, Australia
Yes, and even for the more or less regulars, a high-power-system might stoved away some place in the house as a server or something else. SPCR has very high quality tests on the remaining aspects of a PSU also. And even if you have it in another room, a "more quiet" PUS might just be enough to kepp it inaudible in the neighbouring room.MikeC wrote:Yeah, I had such thoughts too. Certainly it's not about noise. It's about promotional value, mostly. If/when we test a kilowatt psu, we'll have new readers come check us out, and then some of them might stick around for the rest of our quality site.jaganath wrote:I wonder if SPCR should be testing such high-capacity PSUs at all? After all, if you are pumping almost a kilowatt into an average ATX case (or even a P180) getting <30dBA @1m is going to be almost impossible without exotic cooling. And even with AMD 4x4 + SLI 8800GTX + lots of hard drives 850W is the realistic maximum.
AtW