Cap specs & rated voltage

[trodas]

I wonder, how much the caps specs differ when the cap is running on other that rated voltage. Let me say this - there is 16V cap. Let's assume that it run at 8V - half of the rated voltage.

How much capacitance it have, compared to the original specs? (half...???)
How does the ESR change? (probably only very slight, right?)
How does the ESL change? (probably also very slight...)
And the lifetime?

And it is the change linear or rather logaritmic?
What about running 16V cap at 5V?

And since most Vcore voltages are about 1.2 - 2.0V, how much suitable is to use 6.3V caps there? Sure, the PWM pulses are notably higher that the resulting Vcore after caps, as oscilloscope showed me, so 6.3V are probably necessary there - but it really is?

Some caps manufacturers made even 4V ones. Suitable for Vcore?

[Rainbow]

Re: Cap specs & rated voltage

Capacitance does not depend on voltage.

[yanz]

Re: Cap specs & rated voltage

yup. i think if it is 1000uF, it is 1000uf wether it on 5v or 12v. that's the meaning of "capacitance". cmiiw. geeks over here can explain much better i think..

about ESR & ESL, i have the same question as you, trodas let's see what they'll explain about this..

[PeteS in CA]

Re: Cap specs & rated voltage

With lytics, the C doesn't change with applied DC voltage, nor does the ESR or ESL. OTOH, ceramic caps' C does change with applied DC voltage (except for "COG" ceramic), how drastically depends on the type of ceramic. I think Z5U and Y5V are the worst, but even X7R caps are affected some.

[Spacedye69]

Re: Cap specs & rated voltage

Is there any need for 10v at all on pc equiptment, or is 6.3v too close on 5v line?

[willawake]

Re: Cap specs & rated voltage

is there any chance that the dielectric is not being maintained well with low applied voltage to much higher voltage caps. say 3v going through 16v caps for instance?

[VILE DISRUPTION]

Re: Cap specs & rated voltage

is there any chance that the dielectric is not being maintained well with low applied voltage to much higher voltage caps. say 3v going through 16v caps for instance?

Possibly. it depends on how long the low voltage is there but then when it breaks down just use a variable PSU to bring it back up if it is needed for something else. But maybe it will lower the lifespan on the cap(s). So it is best to use the right voltage cap. when i build a circuit and the instructions say a 16v is needed i measure the coltage first and see if i need a 16. but for another part where the voltage is 9 i just use a 10. now i hope this answers that. but sometimes people go overkill w/ 1000v where only a 4.5v is called for. how would you rehabilitate that? my desk psu is only 0-30v

[yanz]

Re: Cap specs & rated voltage

so we should be carefull if we attempt to reuse used caps and we should check the original voltage that was applyed to them, right?

that is, if someone find 16v or 10v caps, he should be carefull if he want to use it on 12v or 5v line. that caps probably was used for 3.3v lines (or ~2v -> cpu vrm output) for years..

[yanz]

Re: Cap specs & rated voltage

i guess, that is a 'memory effect' phenomenon? like what was common to the antique NI-Cd batteries..

[gonzo0815]

Re: Cap specs & rated voltage

As i remmeber, i have read something about those issues on rubycons hp. They stated that the dielectric barrier will decrease if lower voltages are apllied to them, but the oxid layer will rebuild itself when higher voltages are applied. But they recomend the usuall formating thing with some rsistor to slwoly rise the voltage (as at the beginning the leakage current will be fairly high). And they stated that an cap will only at the rated voltage will archive full capacitance. For ESL and ESR i can`t remeber that those where affeckted.
The rebuild process will always consume some of the elektrolyt, thus the lifespann could be reduced somhow.

[trodas]

Re: Cap specs & rated voltage

they stated that an cap will only at the rated voltage will archive full capacitance.

Yep, that is exactly what I remember sometime ago...

Novadays they seems to use better electrolytes, witch could possibly mostly eliminate these abovementioned problems, however I still think that the caps parameters (not only capacitance, but ESR and ESL as well as lifetime - but lifetime probably only increase?) are affected somehow.

Maybe less that before, but who knows for sure - let him speak.

Also it possibly depend on cap type/producer.


And now about the measuring... DMM measure the average voltage and not spikes - in CPU VRM there is about max. 2V average, but on the caps I saw at least 4V high spikes, so 6.3V there have some sense... At least on Epox 8RDA+ and DFI LP B voltage regulators
And to that time I thought we can use 2 - 4V solid polymer caps there...

[gonzo0815]

Re: Cap specs & rated voltage

You can use 4v polymeres, as they have more headroom for spikes.
But the problem with thinning oxid at lower voltages is IMHO not fixable due to elektroly improvment. But anyhow, i won`t give to mutch on those issues.

[999999999]

Re: Cap specs & rated voltage

Practically speaking, you can use the higher rated caps, even reuse them, at lower voltages. The change in capacitance (if any) will be trivial for most common purposes we see on this 'site, and the ESR and ESL depend most on the physical can size, not charged voltage vs rated voltage. This is considering same family of cap for comparison, and modest changes in rated voltage, not using a 400V cap where a 16V used to be, as this will start having more significant capacitance loss.

Thus, most often the ideal is the largest cap you can fit in the allotted space, and whether you pick a higher voltage than it sees or a higher uF value instead, depends most on whether you have a supply with slow sensing or frequency (favoring higher uF) or lower turn-on surge capability (favoring the lower uF but still retaining the low ESR from the larger can size).

As for what is actually required or most cost-effective, that may depend as much on how long you want the part to work properly, if the caps are among the earlier failure points. It is much easier and not much of a cost factor for one-off build or repair to just fudge a little on the safe size with lower ESR or a little more capacitance than necessary, but still if you had a 10V 1500uF cap and could fit it where same family of 6.3V 1500uF could go instead, in general it's not going to be a problem.