Arctic Feezer 7x
Results 1 to 3 of 3

Thread: Power Supply Improvement

  1. #1
    New Member

    brett2014 is offline

    Last Online
    08-07-2011 @ 04:29
    Join Date
    Dec 2010
    Ashfield - NSW - Australia
    CPU: i7 2600K
    M/B: Fatal1ty B3 Prof
    RAM: G.Skil 2 x 4Gb Ripjaws
    GPU: Dual Sparkle GTX 460
    • brett2014's Full Spec's
      • Case:
      • CoolMaster HAF-912 +extra 6" fan
      • PSU:
      • CoolerMaster GX series 750W
      • Cooling:
      • Thermaltake SpinQ VT cooling tower; G.Skil Turbulance II RAM cooler; Extra Push fan; extra rubber mounts & gaskets for quiet;
      • Sound:
      • TBA after testing
      • Monitor:
      • Samsung Syncmaster 21" B2230
      • OS:
      • 7/64b & Ubuntu later
      • Misc:
      • Logitech G110 Gaming Keybd; dual RS-232 for specialist legacy peripheral support. Backup: LiteOn BlueRay writer with LightScribe
    Thanked 2 Times in 2 Posts
    Points: 699, Level: 6
    Points: 699, Level: 6
    Level completed: 75%,
    Points required for next Level: 51
    Level completed: 75%, Points required for next Level: 51
    Overall activity: 48.0%
    Overall activity: 48.0%

    Default Power Supply Improvement

    I have a question based on my electronics background, not my IT one (which is years old, and applicable to mainframes and big mini's).

    Todays Gaming PC's power supplies derive a single 12V rail at a huge number of amps - my PSU delivers 62 amps (continuous), which is a huge current. Hell, that is almost what the starter motor of a small car uses! And it is far from being the biggest PSU available.

    Now, from the 12V rail, other voltages are derived - 5V and 3.3V for instance - so you really want your 12V rail to stay at 12V even when a sudden load hits it. Older styles of supply would derive all the required voltages separately and this newer concept is superior by miles. Evidence of a sag affecting the system would probably manifest itself by hanging or unexpectedly rebooting - that's an educated guess based on PSU's in large computers, and their behaviour when this happens.

    Also, car audio enthusiasts use 1F caps for the identical purpose. If their car sound systems drag a sudden load to power every speaker when at full volume, a small car battery may not deliver enough current, and the voltage will drop momentarily. Or so it is claimed - and it is theoretically correct (I'm not going to argue whether it happens or not). This is remedied by drawing power from one or two 1 Farad capacitors. The power supplied by them does not need to last long, only for the duration of the drop, and the cap(s) are recharged immediately upon a brief removal of the excess load.

    My question here is whether I will benefit by using a 1F (@24V) capacitor across the 12V rail of a gaming PSU. Would it help - ever? If the 12V rail sags, the cap will hold it up for a short time; and short times are all we are talking about. Should something drag it to (say) 9.2V for 6 seconds, you will probably be stuffed with or without a large capacitor. But if that sag to 9.2V lasts 640mS, it is a whole different story - the 12V rail will remain at 12V. And no, I have not sat and calculated this yet, these numbers are off the top of my head but are in the right territory.

    Also, you do NOT place a capacitor like this directly across the 12V rail. For a short period at turn-on it is seen as a short circuit by the PSU. Naughty!! Instead you use a low value series resistor which limits that initial inrush of current - use a low value resistor to limit that initial current - I have not worked out the ideal value but a 1.2 ohm resistor will limit the inrush current to the cap to 10A. You might even consider 1.5 or 1.8 ohms. 12W will be dissipated by a 1.2 ohm resistor so a large wattage device is needed - say 20W. 1.2 ohms is also low enough to allow significant current flow-back to the 12V rail if it droops.

    In addition, my design would have a fast diode in parallel with the resistor. It is forward biased when the 12V rail droops so current flows OUT (of the cap) via the diode, to the 12V rail; but otherwise reverse biased (so you do not overload the PSU at turn-on). A drawn circuit diagram would show the arrow part of the diode symbol facing the +12V rail, NOT the cap, and (resistor and diode in parallel) would be in series with the positive terminal of the cap - going to the +12V rail. You will sacrifice 0.7V to the diode but this will be lessened by the resistor to a much lower amount. In fact, the diode will be a seriously heavy duty one, and you may lose more than 0.7V when forward biased. No worries - the resistor will overcome much of this loss.

    So, there is the thinking, theory, and some rough calculations.

    Is there ever any evidence that a sudden unexpected current draw compromises the power supply? This is CERTAINLY NOT INTENDED as a solution to overcoming under-rated PSU's. Wrong answer for the wrong reasons. This question is aimed at correctly rated PSU's that may not have a lot of reserve. In my rig, video cards dissipate 100W (ea) and a 2600K chip takes another 95W. As to what else is used, and where, isn't added up, but a 750W supply runs it fine. This is a question that could help when a very brief drop-out or sag occurs. It is not to overcome inherent design faults or under-rated PSU's.

    If in any doubt about how important manufacturers are about the output capacitance, look at the CoolerMaster GX range of PSU's where the wording in their 'Special Features' list includes <quote> "Huge bulk capacitor for hold up time >17ms at full loading". Now, 17mS is not terribly long, and a 1F capacitor will extend this greatly, although I am unsure by what exactly - at least 20 times I imagine. I have yet to calculate it, and everyones individual rig would be different. But note how CoolerMaster make an big deal of this exact issue (and rightly so). My suggestion is merely a bigger version of what they are already doing.

    So, in summary, is there a possibility that once in a while such a cap would aid the system, or is it completely unnecessary? CoolerMaster certainly makes a big thing about the size of the output capacitor. So, the need for a big output cap isn't in doubt, simply the added benefit. I have a spare 1F 24V cap, so is it worth using this way, or a waste of $90?

    Cheers, Brett
    Last edited by brett2014; 01-04-2011 at 21:53.

  2. #2
    Folding Team Leader

    grumpydaddy is offline

    Last Online
    23-03-2019 @ 20:29
    Join Date
    Jan 2008
    UK (Plymouth)
    CPU: E5-26xx @2.54
    M/B: MSI X79 GD45 8d
    RAM: 16GB (2*8) G-Skill Trident X 2400 10,12,12 @1866
    GPU: sapphire hd7950
    • grumpydaddy's Full Spec's
      • Case:
      • Lian Li PC90B
      • PSU:
      • PC P&amp;C 910
      • Cooling:
      • C4 NIC
      • Sound:
      • with mb
      • Monitor:
      • dell 24inch
      • OS:
      • Win7 Ult
      • Misc:
      • CPU is an ES processor with 8 cores/16 threads
    Thanked 509 Times in 395 Posts
    Points: 51,457, Level: 70
    Points: 51,457, Level: 70
    Level completed: 22%,
    Points required for next Level: 1,093
    Level completed: 22%, Points required for next Level: 1,093
    Overall activity: 0%
    Overall activity: 0%


    Hi Brett, Just a couple of points... Not all PSU's derive their voltages from the 12v rail(s) and if there were no size limitations for the case of a psu then I expect you would see maanufacturers develop a similar idea to yours But, I may be wrong in my thinking when I consider Caps to resolve problems of a transient nature whereas the power drawn by a rig tends to be more constant under load.

    I think if I had $90 to spend I would just buy a better PSU in the first place but if you really want to pose this question and get some truly meaningful answers then do so on jonnyguru forums where there are a couple of specialists for whom this Idea will be something to get their teeth into.
    Rig2: 2*e5-26xx (16c/32t @2.4), Asus z9 pe-d8 ws, 32gb ripjawsz under water, rig3 2*e5-26xx (16c/32t @2.4), Supermicro 7047A-T 16GB 1600 ecc reg, on air, Rig4: 2*e5-26xx (16c/32t @3.1), Supermicro x9 DAi 16GB 1600 ecc reg, on air, Rig5: 2*e5-2660 (16c/32t @2.7), Supermicro x9 DRW iF 32GB 1600 ecc reg, on air

  3. The following user thanks grumpydaddy For this above post:

    brett2014 (03-04-2011)

  4. #3
    OC Jedi Master

    Deton is offline

    Last Online
    27-02-2014 @ 22:13
    Join Date
    Dec 2006
    CPU: i7 920, 4.2Ghz HT
    M/B: P6X58D Premium
    RAM: OCZ Flex EX PC3-17000 12GB
    GPU: Asus EAH5870
    • Deton's Full Spec's
      • Case:
      • Lian Li A70F
      • PSU:
      • Enermax Galaxy Evo 1250W
      • Cooling:
      • Custom H20 setup.
      • Sound:
      • Onboard
      • Monitor:
      • ViewSonic VX2435
      • OS:
      • Win 7 - 64bit
    Thanked 1,179 Times in 1,026 Posts
    Points: 50,246, Level: 69
    Points: 50,246, Level: 69
    Level completed: 36%,
    Points required for next Level: 904
    Level completed: 36%, Points required for next Level: 904
    Overall activity: 7.0%
    Overall activity: 7.0%


    yup have a couple electrical specialists.

Remove Ads

Sponsored Links

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts

Similar Threads

  1. Power Supply Issue
    By Nyarlathotep in forum General Hardware
    Replies: 19
    Last Post: 07-03-2011, 16:08
  2. Power Supply's
    By atomicvai in forum General Hardware
    Replies: 8
    Last Post: 14-03-2010, 17:54
  3. Sharkoon Out With New Rush Power Power Supply Series
    By News Team in forum Hardware News
    Replies: 0
    Last Post: 19-12-2009, 05:40
  4. Power supply suggestions
    By grumpydaddy in forum General Hardware
    Replies: 5
    Last Post: 05-09-2008, 03:34
  5. Power supply voltages
    By grumpydaddy in forum General Hardware
    Replies: 4
    Last Post: 31-07-2008, 22:46

Search tags for this page

There are currently no search tags.
Click on a term to search for related topics.

Tags for this Thread