Enermax 431W

[trodas]

As you can check his specs there:
http://www.pcstats.com/articleview.cfm?articleID=409
It is a oldie but good made PSU. Sturdy, I using two of them for very very long time (probably over 3 years now) and since they use the infamous Vent cap on the 12V rail as well, as everything else by the inferior Jpcon caps, I felt that is time for a change. This one was smooth, since the design used nonstandard d 13,5mm caps - so I can use d 12,5mm Pannyes easily:





Beauty, aren't they?

The 3300uF 16V one is a BIG suxxka, tough:





The only real problem arise with the imput filtering caps. None of these Panny TS-ED caps in 1000uF 200V could fit, so I used 820uF 200V ones instead:



It should not have a negative impact.
Also I replaced the back 80mm fan, witch is not spinning up properly anymore at 5V - I use both 80 and 92mm fans there at 5V to make the PSU quiet

The results.

Ripple before recapping:
12.13V - 54.4mV
5.13V - 11.8mV
3.349V - 8.6mV

Ripple after recapping:
12.14V - 17mV
5.14V - 17.9mV
3.361V - 4.6mV

Ripple with added bridging 10uF 16V ceramics SMD caps on the caps:
12.14V - 15.2mV
5.14V - 19.2mV
3.361V - 6.2mV

Ripple after removing these SMD caps:
12.14V - 18.1mV
5.14V - 13.4mV
3.361V - 5.4mV


Conclusion.
Ripple get better w/o ceramics bridging caps. The measuring margin of error is very high and indeed changes in few mV hardly could be mentioned as positive change, however at least these new cap's does not have the smell and the ripple in 12V get notably better still From 54mV to 17-18mV. Hmmm

I wonder if I remove these SMD caps from some of my recapped boards, if I could overclock them higher...!

[yanz]

Re: Enermax 431W

such a details is always nice.

we know that the ripple showing on the output is probably 50khz - 60khz. probably the effect of adding smt caps is adding the esr/impedance of the al caps? but it againts all our assumption before about the goodness of smt caps. also, is it at full load when you test the line?

[trodas]

Re: Enermax 431W

yanz - well, no screenie this time, and I did not remember the frequency, but I think it was about or higher what I get on the Eurocase - about 1,6kHz maybe? I don't know.

And yes, it clearly show that blindly adding ceramics to any cap is not good thing to do. It is clear to me now, that ceramics should be as far from the voltage regulator sense wire, as possible.
Obviously they do trick the regulation, causing much frequent switches and that is far from desirable...

As load was, as in the Eurocase case used my testing switch/load device (1A on 3.3, 5 and 12V lines) with the two HDDs connected.

Maybe I should check the ripple compared to the actuall mainboard? I did not fancy dummy loads too much, as they have constant current, whie current to computer mainboard is far from constant, same for HDDs and CD/DVD devices...

But is true, that my load thingie should utilize at least 5A per rail. Maybe if I get bored sometimes I could create 2nd revision by just adding another 4 pieces of resistance wires to each already used ones to get to the 5A stuff