Desktop computer PSU specs; possible to DIY?
May 18, 2010 at 9:04 PM Thread Starter Post #1 of 26

ShinyFalcon

500+ Head-Fier
Joined
Dec 29, 2006
Posts
877
Likes
11
Back when I RMA'd my desktop power supply, I got back a sheet showing some load voltages/current/etc. I thought I share with everyone...
 
Code:
 PC Power & Cooling Silencer PPCS500 500W Rated for +3.3V@24A,+5V@24A,+12V@35A,-12V@0.8A,+5VSB@2.5A +12V Load +12V -12V +3.3V +5V +5VSB Load Current 34.00 A 0.50 A 8.000 A 10.000 A 2.000 A Upper range 12.60 V -13.00 V 3.465 V 5.500 V 5.250 V Lower range 11.40 V -11.00 V 3.135 V 4.750 V 4.750 V Outcome 11.91 V -11.71 V 3.260 V 4.900 V 4.850 V Wattage 404.9 W 5.86 W 26.100 W 49.000 W 9.700 W Total Wattage 495.6 W +5V Load +12V -12V +3.3V +5V +5VSB Load Current 30.00 A 0.50 A 9.000 A 20.000 A 2.000 A Upper range 12.60 V -13.00 V 3.465 V 5.500 V 5.250 V Lower range 11.40 V -11.00 V 3.135 V 4.750 V 4.750 V Outcome 12.19 V -11.91 V 3.250 V 4.770 V 4.840 V Wattage 365.7 W 5.96 W 29.300 W 95.400 W 9.680 W Total Wattage 506.0 W +3V Load +12V -12V +3.3V +5V +5VSB Load Current 30.00 A 0.50 A 20.000 A 12.000 A 2.000 A Upper range 12.60 V -13.00 V 3.465 V 5.500 V 5.250 V Lower range 11.40 V -11.00 V 3.135 V 4.750 V 4.750 V Outcome 12.07 V -11.81 V 3.240 V 4.840 V 4.840 V Wattage 362.1 W 5.91 W 64.800 W 58.100 W 9.680 W Total Wattage 500.6 W Burn in +12V -12V +3.3V +5V +5VSB Load Current 30.00 A 0.50 A 12.000 A 15.000 A 2.000 A Upper range 12.60 V -13.00 V 3.465 V 5.500 V 5.250 V Lower range 11.40 V -11.00 V 3.135 V 4.750 V 4.750 V Outcome 12.03 V -11.85 V 3.260 V 4.840 V 4.840 V Wattage 372.9 W 5.92 W 39.100 W 72.600 W 9.680 W Total Wattage 500.3 W
 
For a PSU that can supply this much current, linear PSU's are out of the question. There are linear PSU's out there like the PicoPSU, but it probably won't be able to power high-end systems.
 
My criteria is a PSU that is silent (which is why I RMA'd my PSU in the first place... piercing coil whine) and is able to supply relatively high current in the +12V rail and +5V rail. I don't really know what uses the 3.3V rail.
 
Any starters on how we can design an audiophile desktop PSU?
evil_smiley.gif

 
May 19, 2010 at 2:25 PM Post #3 of 26
Quote:
I believe the CPU runs off the 3.3V rail (regulated down), as well as the RAM, seeing as how both are usually under 3.3V. Possibly more.
Not a speculation without merit, but the CPU runs off of the 12V rail. Not sure about RAM.
 
If OP wanted something quiet, a Seasonic or Seasonic-built unit would have done the job. Or one of the fanless designs floating around.
 
If you want to build your own, I hope you know a source for big honkin heatsinks. Other than that, I can't help you. If you ever do get it to work well, send it in to Jonnyguru. That'll get you famous, quick like (if it's good, anyways).
 
(You get bonus points if you use Nichicon KZs for everything)
 
 
May 19, 2010 at 2:43 PM Post #4 of 26
You will probably find it easier to modify a commercial unit that to build your own thing, especially if you are looking for that much power and something that will fit in a standard slot.
 
A popular tweak for combating inductor noise (the whine you describe) is to DIY-pot the inductors, although most people dont get the rest of their systems quiet enough to benefit from this: you should be proud that you heard it over the fans and HDD's
 
May 19, 2010 at 4:43 PM Post #8 of 26


Quote:
Power is still drawn from the 12V rail, and then downed to 1-1.5v, or whatever the processor wants.
 


That's quite a big drop, especially considering the CPU in a desktop can draw a lot of current (hence my assumption that it was drawn from 3.3V). Out of curiosity, why do they regulate from 12V, and what is the 3.3V used for then?
 
May 19, 2010 at 4:59 PM Post #9 of 26
PCI and PCI Express cards can use all three rails.
 
And 12.x V to 1.x V for the CPU is indeed a very big drop. The ATX connector also includes all three rails, and I still feel that the 3.3V is probably regulated down for the memory and for the CPU.
 
Edit: OK, looks like the CPU does gets its power from the 12V rail. See schematics: http://download.intel.com/design/chipsets/schematics/25281302.pdf page 84
 
May 19, 2010 at 7:58 PM Post #11 of 26
Quote:
http://www.cicsmemoryplayer.com/uploads/B_Linear_PSU.jpg
 
http://www.cicsmemoryplayer.com/index.php?n=CMP.ApdxBAdvanced
 
It has already been done.
 
 
Oh my.
 
Quote:
PCI and PCI Express cards can use all three rails.
 
And 12.x V to 1.x V for the CPU is indeed a very big drop. The ATX connector also includes all three rails, and I still feel that the 3.3V is probably regulated down for the memory and for the CPU.
 
Edit: OK, looks like the CPU does gets its power from the 12V rail. See schematics: http://download.intel.com/design/chipsets/schematics/25281302.pdf page 84

Told you so.
 
May 20, 2010 at 6:50 AM Post #13 of 26
Maybe its Google's fault?  It has been reported that their PCs have only 12V going to the board.  They regulate everything else on board as it makes for a cheaper overall unit.
 
May 20, 2010 at 7:40 AM Post #14 of 26


Quote:
So no theories as to why they drop the 12V down so much instead of just using the 3.3V, which seems to have plenty of amperage?

 
 
It is more efficient, in a nut shell. The drop doesn't matter at all for SMPS's, they have very good efficiencies where it is usually more efficient to run from higher voltage inputs. The wire connecting your PSU to the board don't need to be as heavy either, as there is less current flowing for the same power delivered, and if you look at how many separate 12v wires they put in there already these days, then you need to multiply that by 4 for what would be required for 3.3v supply.
 
Don't think that if you put a linear power supply in a computer that you'll make it electrically quieter, everything in your computer is made of digital gates switching at all sorts of frequencies spewing out more garbage back into the power supply that what a good SMPS is doing in the first place. If you really care, just filter the sensitive circuits that you want like everyone else
 
May 20, 2010 at 8:44 PM Post #15 of 26


Quote:
It is more efficient, in a nut shell. The drop doesn't matter at all for SMPS's, they have very good efficiencies where it is usually more efficient to run from higher voltage inputs. The wire connecting your PSU to the board don't need to be as heavy either, as there is less current flowing for the same power delivered, and if you look at how many separate 12v wires they put in there already these days, then you need to multiply that by 4 for what would be required for 3.3v supply.


Thanks for the answer. So the motherboard has an onboard SMPS to convert 12V to 3.3V? I figured it would be a simple linear regulator, in which case you'd be wasting a ton of heat. It does make sense not to run the extra 3.3V cables if it can be avoided, but why supply 3.3V in the first place then?
 

Users who are viewing this thread

Back
Top