Replacing caps: how close does it have to be?

[Termina]

The bad caps I'm replacing are 1200 uf 10v. I had some 1200 uf 16v caps, but ran out near the end.

If I replaced the bad 1200 uf 10v caps with 2200 uf 6.3V caps, would this cause any problems?

[willawake]

Re: Replacing caps: how close does it have to be?

it will be ok.

if you see 16v caps on the board dont put caps less than that because it is on the 12v line. otherwise 6.3v is fine everywhere else. increasing capacitance is fine.

[Termina]

Re: Replacing caps: how close does it have to be?

Sorry, I've been misunderstood.

Board originally had 1200 uf 10v caps
I replaced many with new 1200 uf 16v caps

I ran out, and would like to replace the rest with 2200 uf 6.3V caps.

Is this ok?

[willawake]

Re: Replacing caps: how close does it have to be?

yes it is ok

6.3v is fine everywhere on the board except where there is 16v caps there already. In your case there is not 16v caps there already, there is 10v caps so it is ok to use 6.3v caps.

increasing capacitance is ok.

[Per Hansson]

Re: Replacing caps: how close does it have to be?

I want to ask a question about this too, is there any level where it actually becomes unsafe to put higher farad rated capacitors in the place of lower?

I.e. is there a limit where you theoretically could run into problems? Take for example a PSU, would it not actually be better, in this case, to use as high rated caps you can find? For example I see that Elfa carries 3900uF 6.3v caps, should only be better to use than for example a 2200uF cap which is what you oftenly find in PSU's right?

But how does this work on a mobo? Say on the 12v side of the VRM you have already 3x 3300uF 16v caps like on my 9NDA3+ and then 7x 3300uF 6.3v caps on the secondary side, what if I placed 3900uF on the secondary side only, would there be any risk that I create some sort of imbalance in the VRM and actually manage to increase the ripple current?

[gonzo0815]

Re: Replacing caps: how close does it have to be?

The only thing which is concern, should be the maximum capacitiv load during system start. IMHO the ATX spec limmite this value to a certain value, eg. inrush current should not exceed a spezified value. But most motherboard desings have already managed that in putting an inductor in series with the vrm unit of the motherboard. So i think on input side, you should be safe to double the capacitance, on the output side i think this woun`t do mutch harm, as in my opinion more bulk capacitance will always a good thing. And most motherboard manufcturer are going with far less o/p capacitance as in most vrm controler datasheets is suggested in using things like droop compensation funktions etc.
So o think that it will at least no problem to doulbe the o/p capazity. However, increasing the capacitance could have a large impact on the duty cycle of the vrm circuit. But in most cases the duty cycle will decrease and this is in most vrm aplikations save, as switching loses will decrease.

On PSU`S i think, that above said is aplikable to, but keep in mind that switchmode psu`s are not that easy to desing, as the whole circuit has to be tuned to surpress self oszilation. And if you change some values this could have bad impact on a particular psu. But ok, i think this isn`t a problem if you double the capacitance....but in any case, mileage will vary.

[willawake]

Re: Replacing caps: how close does it have to be?

here's an interesting thread.
http://forums.procooling.com/vbb/showthread.php?t=10529

[Per Hansson]

Re: Replacing caps: how close does it have to be?

Very very interesting read Will...

[gonzo0815]

Re: Replacing caps: how close does it have to be?

Very very interesting, but as far as in most cases the mainboards are far away from maximum Capazitance and lowest ESR the VRM controler ist suited, i think adding more Capacitance woun`t hurth. I haven`t expected any other aproch from the manufacturers, as removing caps untill it breakes.We all should remmeber, that even this may be an kind of high tech, it is consumer high tech, not military tech. So all the sophisticated things like computer simulation, large testing is far too expensive for low cost elektronics. With an fast to market time like IT consumer stuff. Only Intel makes the thing right, which could be the case for it`S good reputaion on stability. Amd has IMHO never produced ther own boards.....

But anyway, this statements about ESL will add something to our knowledge why OSCON caps are in some ways beter than elektrolytic ones. As they are mutch smaler, i think they have a realy low ESL, compared with electrolytic ones.

And as we have senn by trodas work, it isn`t that dificult to improve a VRM unit without the whole understanding and testing equipment.

[Elitist]

Re: Replacing caps: how close does it have to be?

When discussing the rated capacitance of electrolytics it's worth remembering that higher values used for smoothing/reservoir functions mostly have a +50% tolerance anyway! In practice, since they are very much active electrochemical cells when the power is ON, it's anyone's guess what the exact value is at any one time. gonzo's advice is much closer to reality, perhaps than even he had imagined. Termina's 1200uF units may already be more than halfway towards 2200uF.
Voltage ratings are different, but not wildly so. They may initially, during design, be somewhat reliant on calculation of material thickness, etc, etc, but, again in practice, are heavily based on testing and field experience. Sometimes a 10V can survive as a "16V" replacement (implying use on the 12V line plus a little allowance for surge); it shouldn't be relied upon but worth a try for emergencies and temporary repairs. Always think (guess?!) about the role and position of a cap in the circuit. Running a 10V unit at, say, 50V, however, could invalidate your house insurance. More important, as alluded to by gonzo, voltage excursions could take the rating well beyond design limits, especially if the supply is not adequately choked.

[Topcat]

Re: Replacing caps: how close does it have to be?

If I replaced the bad 1200 uf 10v caps with 2200 uf 6.3V caps, would this cause any problems?

No, that os NOT ok!

Do not decrease the voltage like that without knowing what voltages are present on the caps. Doing so might result in a nice fireworks show.

[Elitist]

Re: Replacing caps: how close does it have to be?

What willa is saying is that there is only 3,5 & 12V present on m/boards. Therefore, if you see a rating of 6.3, it must be on either the 3 or 5V line, so a rating of 6.3WV should be adequate. So it should be safe. Any higher rating means that the board manufacturer is allowing a greater margin for surge or got a better deal on a standard stock item - usually the latter!

[gonzo0815]

Re: Replacing caps: how close does it have to be?

In gernerall, i do not endors using caps with lower voltage rating, but on mainboards and other uControler stuff, there are in my opinion no need for 10v caps, at least with low esr. Exacly like Elitist explained.
Therefore i replace all caps with 6,3v or 16v ones.
But ok, i alway do a quick chek with an dvm too.
I doubt, that a manufacturer, who is using the cheapest crap caps on the market, would use an 10v cap to have some headroom for surges.
If they would have wanted headroom or something else, they probably had bought good quality caps, which have that kind of headroom build in already.
And this is exacly the opinion of

Regarding the capazitance, i am sure, that nobody did realy know, what kind of prozesses will start in an al-cap.
The problem is, that you can`t make an cocktail with such agressive chemicals, a big al surface and then specifie that to any akurate value.

[willawake]

Re: Replacing caps: how close does it have to be?

it is not only price/availability/convenience but if you look at some spec sheets occasionally you can understand why they used some strange values of caps, their specs stand out as being better than the correctly specified part.

what TC is saying though is the correct procedure is to check first the voltage at that point in the circuit using a DMM before deciding to use a lower rated cap.

[Elitist]

Re: Replacing caps: how close does it have to be?

Time for a reality check. In this cruel world, there is the spec sheet and the 'best price' offer from the supplier! What you get is the best deal the builder could cut. If the mass production lines are busy churning out 1500uF rated 10V, this could be what you get, regardless of what the designer intended. That is why it is necessary to THINK!!! OK - and use the DMM, although that won't tell you about likely excursions; for that, look for 'L' circuit elements.
Furthermore, the oxide/electrolyte interface, more often than not, is capable of growing some extra thickness in a dynamic electrolytic cell - an electrolytic capacitor is NOT a passive component. Therefore - entirely at one's own risk and in emergencies (read exactly what I said - we are not talking to inexperienced neophytes in this Forum) it may be worth trying to uprate a 10V unit to 12V. I suggested this because:
a) emergencies sometimes call for emergency expedients!
b) the tolerance on voltage rating should already be ~+10% (up to 11WGV)
c)there may be scope for additional film growth to accomodate the additional 10% field, especially bearing in mind that the field gradient at the interphase is already measured in MeV!! (possibly up to 12WGV).
On the other hand, it may blow up - but that was always a possibilty with low quality and 'bad' caps, anyway......

[gonzo0815]

Re: Replacing caps: how close does it have to be?

That is right Elitist, but it is well known, that cpas of good manufaktures have greater resistancwe to more voltage than cheap caps. This could be easily be testet at home ;-)

[kc8adu]

Re: Replacing caps: how close does it have to be?

i did the testing like you describe.
rubycon yxg 3300@10v can be reformed to 16v.unsure of long term effects but i did it on a hunch that high quality parts have a bigger safety margin.kinda like overclocking.
tried the same on some samxon caps rated ay 6.3v.they looked like they would reform a bit too.but at 7.5v after the initial leakage dropped it rebounded and 2 hours later it had gotten bad enough to bulge.

[Elitist]

Re: Replacing caps: how close does it have to be?

This kind of experimenting is even more exciting than o'cing! Could end in a big bang.....

[Per Hansson]

Re: Replacing caps: how close does it have to be?

This kind of experimenting is even more exciting than o'cing! Could end in a big bang.....

Haha, then you have not overclocked enough my friend :P

[tux_topo]

Re: Replacing caps: how close does it have to be?

I have a problem here, with an MSI K8N Neo4-F, with a vary fluctuating vcore. I'm gone to recap. But i cannot find good caps. That's why, i'm gone re-use the caps of a dead Abit NF7-S. The problem is that the MSI have 4 caps of 1500uF 16v, and the Abit have 4 caps of 1200uF 16v. Changing the capacitance in 300uF will damage something?