Capacitor Endurance Rating vs. Lifetime

[mockingbird]

momaka made an interesting statement in a previous thred:

That 2000 hour rating is Endurance, not Life Time. It simply states for how long you can "abuse" the cap with maximum ripple, voltage, and at max temperature (105C in this case) before it goes out of spec or vents.

Let's say I'm ordering batches of small value electrolytics for PSU re-caps in price-break quantities of around 100+, but I want ones that will have a long lifetime and won't be the weak link in the PSU, is the endurance rating of a particular series something I should take into consideration?

For example, most 5mm caps endurance rating is below that of wider caps even within the same series. Is a 5mm cap with an 1000 hour endurance rating less reliable than a cap of the same series with say a 2000 hour or more endurance rating, or do they last just as long as their counterparts as long as they are not 'abused'?

Should I value higher ripple ratings in 5mm caps over endurance ratings?

[BigTroll]

Re: Capacitor Endurance Rating vs. Lifetime

Put in a SU'SCON!

[momaka]

Re: Capacitor Endurance Rating vs. Lifetime

I want ones that will have a long lifetime and won't be the weak link in the PSU, is the endurance rating of a particular series something I should take into consideration?

If you are running the cap at or near it's maximum specified ripple current, voltage, and temperature rating then yes.

Basically, the endurance is the life time only when you are running the capacitor at it's maximum allowed specs. Now, for every 10 degrees C drop in temperature below the maximum rating, the lifetime more or less doubles. So if you have some 105C caps rated for 2000 hours endurance, and you are running them at maximum allowable ripple current, maximum voltage, but only at 95C, then their life time will be approximately twice that of the endurance - i.e. about 4000 hours. If you lower the temperature another 10C, the life time will now be 2 x 2 x 2000 = 8000 hours. So basically temperature is what relates the endurance and the life time. The maximum ripple current and voltage probably also have an effect on the life time, but Im not sure how exactly. Well, the ripple current has a pretty big effect on the temperature of the capacitor, so obviously the lower the ripple current in the circuit, the less the cap has to filter and the less it will heat up.
(By the way, all of the above info is from what I remember from posts made by PCBONEZ. I think he said he read some data sheets on the matter, so all credit goes to him and whoever else may have posted but I forgot to mention.)

So to answer your question - if you want a long life time from a cap and the cap will be ran in a hot environment, then yes, the endurance would matter.

Is a 5mm cap with an 1000 hour endurance rating less reliable than a cap of the same series with say a 2000 hour or more endurance rating, or do they last just as long as their counterparts as long as they are not 'abused'?

Technically, they should be "less reliable". Again, however, if the temperature is significantly lower than the maximum, then the lifetime of both the 1000 hour and 2000 hour endurance-rated caps will be so long that you wouldn't notice any difference - at least not in the useful lifetime of the electronic device these caps are in.

Should I value higher ripple ratings in 5mm caps over endurance ratings?

Depends on the application. In PSUs, I would say no. On motherboards - yes, in most cases (unless it's for a linear regulator, in which case ripple current and ESR are hardly relevant).

[PeteS in CA]

Re: Capacitor Endurance Rating vs. Lifetime

What momaka said. I'd just add that the "Endurance" or "Life" is what the manufacturer is willing to guarantee (how much does that mean if the cap company is a fly-by-night in another country? to an end user?). IOW, ignoring crap-cap-producers/excreters, that number applies across the entire production population, and many/most of that population would last longer. It's just that if you rely on most caps lasting longer than spec'ed, you've assumed the risk of early failures of caps that only just meet the Endurance spec. And of any consequence from that realized risk.

[Pentium4]

Re: Capacitor Endurance Rating vs. Lifetime

And if you're using cheap caps and just one of them dies, doesn't it create a chain effect because then the other caps on the circuit have to work harder to make up for the dead cap, killing them faster?

[PeteS in CA]

Re: Capacitor Endurance Rating vs. Lifetime

To at least some degree the AC current, the ripple current, is a constant current source. So unless the dying cap dies dead short, each remaining cap in the parallel network "sees" an increase in its share of the total ripple current. So, yes the remaining caps are stressed more, hastening their demise, whether premature or just sooner than intended.

[Wester547]

Re: Capacitor Endurance Rating vs. Lifetime

What PCBONEZ also said that is that cap manufactures expressly state those endurance ratings with 60% confidence in the datasheet (given the failure rate)... I think mariushm said that too... so a capacitor that is rated at 6,000 hours @ 105*C could be in actuality anywhere from 3,600 hours or 10,000 hours... of course, it's still better to use capacitors with higher endurance ratings so the worst case scenario isn't a terrible one. As for linear regulation - I always thought that wasted lots of power, lots through the capacitors, so while that sort of DC-DC converter usually provides a very clean output noise and ripple wise, I would still think it's tough on the capacitors.

The main problem with Taiwanese brands is that the quality of aluminum foil is so poor (too much copper or other impurities) that it isn't difficult for those impurities to leech into the electrolyte and you know the rest. When that happens, it won't matter if the capacitor is rated at 20,000 hours @ 105*C, it's done. Internal heating, or ripple current, or external sources of heat do exactly that, even in the most minimal manner (if you can keep the ripple and heat down, though, it may never happen). Some of these "better" brands probably do better in PSUs because it's easier to give them ample airflow in such a design and a good design will result in little ripple current. Also note that the end ripple voltage (what you see measured in millivolts on each rail) and the actual ripple current going through the capacitors isn't the same thing. While the end ripple (going to the motherboard) could be low the actual ripple going through the capacitors is much higher, often thousands of milliamps (it has to be "dumped" to the ground through low ESR capacitors so it doesn't "escape" to your system), so ripple rating is very important, IMO.

[mariushm]

Re: Capacitor Endurance Rating vs. Lifetime

Linear regulators waste a lot on the pass transistors inside the regulators themselves, capacitors don't get too much stress.

However, it's important to use adequate capacitors so as to not get the regulator unstable, some regulators need capacitors to have esr within a particular range, or lower than a value, otherwise they'll oscillate.

See page 18 and onwards, but the whole app note pdf has good info.

I don't remember the 60% value and honestly it seems kinda low. A value of about 80% wouldn't surprise me.
Either way, everyone recommends using capacitors rated for at least 50% larger voltage than the voltage the capacitors will work at and so on.

With ceramic capacitors - which may be tempting due to very low esr - it's even more critical to use proper capacitor as the capacitance of ceramic capacitors decreases A LOT with the voltage.. a 10uF 16v could have only 0.5-1uF at 10v. Usually, it's recommended to use 5-6x times bigger voltage rated caps when using ceramic capacitors

[bruce14120]

Re: Capacitor Endurance Rating vs. Lifetime

So, If I cool the power supply caps in my LCD HDTV with a cooling fan I should be able to reduce the ambient temperature significantly and extend the life of all power supply components. Likewise, if I need to replace caps I should be able to use high quality 95 degree caps. Makes sense to me!