QUESTION Can i use capacitor bank in parallel to PSU

[rakesh_sharma23]

Can i use capacitor bank in parallel to PSU.
This type of capacitor bank helps in performance in sound amplifiers how comsume a lot of power.
Will this mod help me.

http://img10.imageshack.us/img10/3019/powerot2.jpg

any sugg.

[cirthix]

what do you expect to gain by doing so?

adding more caps to stuff does not magically make it overclock higher :/

[rakesh_sharma23]

As in sound Amplifiers Capacitor Bank help in Power handling and Filtering.

Dose it do the same in PC
I am not thinking about overclocking.

[Nanometer]

that's not really going to do much of anything, just buy a better power supply.. you really shouldnt still be using the one that came with your raid max case from fry's electronics.

[NapalmV5]

As in sound Amplifiers Capacitor Bank help in Power handling and Filtering.

Dose it do the same in PC
I am not thinking about overclocking.

pc is no different :)

pc psus are very dirty.. the nastiest things around :yepp:

the cleaner the rails the better :up:

and caps do magically make stuff oc higher :D

[caps_buster]

Well, the cleaner the voltage, the better.
One thing - capacity does nothing to do with cleaning out the voltage - it has everything to do with the capacitor ESR and ripple ratings. The lower the ESR, the better. The higher the ripple ratings, the better.

Now what ratings do your 40 000uF caps bank have? :p: What caps you planing to use?

I would go for Panny FM, Rubycon MCZ, Nichicon MN/M/Z or Samxon GC bank, on the start/end finetuned by nice polymer cap (Samxon URL, Nichicon LE) and insert a ceramic SMD cap (say 4.7uF) between each of the caps as they are in the bank :D

But hey, I'm a bit extreme there :)

It is not better to look first for a GOOD brand PSU, then replace PSU caps, then add coils and THEN add caps? :)

Anyway, it will help stabilize your voltages, gain longetivity and stability and overclock things better - even if you are not going to overclock.

In fact, it was measured that even CPUs running on good caps running faster!
(SuperPI 1M difference go up 1sec when replacing Ruby MBZ with Samxon polymers, IIRC)

[rakesh_sharma23]

Thanks Caps Buster..
The diagram was just a thought, but after your reply, i think i should give it a try.

I am having few Nichicon KS Serier 10000 uf caps (Rates rippel 2.95, and Tan = 0.3). These are few of best caps for Audio amplifiers.
I will join them in Parallel to get 40000 uf for each voltage rail of PSU.
Further add few Non-polar 10uf polymer caps.

May be this cap-bank will help get better OC (when voltage-mod is done on MB and GPU).

Currently having:-
Cooler Master RS-650-ACAAA1 Real Power Pro 650W Power Supply
GIGABYTE GA-MA78G-DS3H M/B,
AMD Phenom 9550,
XFX Nvidia 9600 GT,
2x 2GB 1066Mhz DDR2 RAM

[caps_buster]

Well, I just hope you are going to connect these caps banks ON or after the power socket on the mainboard, and not directly to the PSU itself. The resistance of wires helps and prevent the mosfets in PSU fry on power-on, or PSU over-current protection to kick-in ;)

Also, what is best for audio applications (eg. Rubycon BlackGate caps, or Elna RSF Silmic caps, or polyester caps...) is nowhere near optimal, nor best, for voltage cleaning purposes.
You gotta understand that caps are not "just caps", but there is many differences, variations and so on of caps to best suit them for DIFFERENT purposes. And you pick the best caps based on what are you going to use them for.

Only deaf person waste a Nichicon HZ or Samxon GA cap for Audio signal separation.
Only a fool use audio grade capacitor for voltage filtering purposes.

The fact that such capacitor missuses are common does not justify them.

Once again, for example Panasonic FM 4700uF 6.3V give around 3600mA ripple - eg. notably more that 2950mA of Nichicon KS series, witch are big-size caps, but that it is. If you talk about Nichicon, then better look out for HZ, HN or HM ones. Use more caps to got your target capacity, but they will still have at least twice the ripple rating in the end.
From other brands consider Rubycon MCZ, MBZ and Samxon GA and GC caps. Maybe even Chemi-con KZJ, if you can get hold of them...

I don't know about any nonpolar polymer caps, but Murata or Panny or Taiyo Yuden ceramics will do fine.
Murata X7R - 10uF 16V size 1206, Digi-key order number 490-3911-1-ND

I would measure SuperPi times before and after this mod as well. Not only overclock increase, but using good caps make the machine also run faster (less buss errors on CPU = more speed)...
I seen difference over 1sec on SuperPI 1m test when exchanging mediocre KZG caps in Vcore output to Samxon URL polymers. I got mysels about 0.7sec faster SuperPi using GA Samxons in the Vcore output.


PS. Nichicon KS caps are snap-in huge audio caps, entierly unsuitable for computers and switching power voltage filtering job. Lovest voltage they are on is 25V and 10 000uF give only 2450mA ripple - but at 25V! Your 2950mA are for 35V, not for 12, 5 and 3.3V.
Using high-voltage caps on low voltages does not help things, that is another very common mistake. In fact, even some companies do it, but that again did not justify stupidity.
The catch is the rated voltage for the ripple ratings. It is 35V, so, on 10x lower voltage (3.3V) be ready for about 10x lower ripple!
(not quite 10x, but the differences are huge, depending on caps type and frequency of discharge and it is also nonlinear...)
Hence using as close voltage rated caps to their working voltage is desirable. 16V caps for 12V rail, 6.3V caps for 5V and 3.3V rails - because 4V caps are not common in elytes. They are common in polymers, but making so huge capacity out of polymers would be kinda expensive.

Back for the "voltage overhoot" thing. Like I said, it is justified by the "higher ripple values" in datasheets, however on notably lower voltages are the ripple values far worser, even worser that the very same cap type designed for lower voltages already and having lower ripple in datasheets.
But there is one more argument for the BIG voltage caps. Heat dissipiation. Bigger cap (for bigger voltage) dissipiate heat better. However the catch is, that the bigger cap has also (because it is for higher voltage) notably higher internal resistance WHEN working on low voltages, so it actually not only dissipiate heat better - it produce notably more heat by itself! :)
So, it is almost laughable to see a 16V caps used on Vcore input filters on many old mainboards taking the Vcore from 5V rail... :)

Actually, it would be laughable, if it was not sadly copyied by MANY brain-dead persons.

[Boogerlad]

too much capacitance is bad. this is because too much current required at bootup, no turn on. also, if you're not ocing. it won't result in any performance increase.