are poly caps really better?

[jack black]

this is what i thought, but looking up specs on DigiKey shows 10,000 hrs 105C Al electrolytes available from reputable makers vs only 2,000 hrs 105C Al polymer caps of similar size/voltage. BTW, i also checked tantalum ones and it was funny how expensive they were in the same size.

[mariushm]

Re: are poly caps really better?

It really depends on what you're going to use the capacitors for.

Polymer capacitors are ultra/very low esr capacitors, which are great for dc-dc converters. Electrolytic capacitors are very low / low esr capacitors, to get the same performance as a polymer capacitor in some applications you may need to either put several electrolytic capacitors in parallel (to reduce esr) or use capacitors rated for higher voltage and higher capacitance, which means larger size capacitors.

For example, if you design a dc-dc converter for a processor or something that requires 1.8v at 5A, it may be enough to use a single 820uF 2.5v or 4v polymer instead of using two 1200-1500uF 6.3v electrolytic capacitors.

Also note that some series of capacitors are advertised as up to 10k hours @ 105c, like Panasonic FR series for example, but this lifetime rating varies with the diameter and height of the capacitor. Maybe a 3300uF 16v capacitor is rated for 10k hours, but a 470uF 16v one has a much lower lifetime rating.

There may also be issues with height, besides the diameter of capacitors... for example a motherboard manufacturer may prefer to use polymer capacitors because they're not so tall, which allows buyers to use larger/wider heatsinks that basically could even cover those polymer capacitors. At the very least, they make room for heatpipes and other crap.

See for example these boards, look how close the capacitors and inductors are to the cpu socket: http://www.anandtech.com/show/8950/a...er-motherboard



Could you imagine electrolytics 2-3 times taller there?

Electrolytic capacitors still have their place, for example in computer power supplies which are standardized to a certain height and don't usually require ultra low esr, therefore tall electrolytics (lots of capacitance) can be used.


Tantalum capacitors are really not recommended to be used, unless you're really forced. A lot of the raw materials are mined in countries with constant wars and slave labor (think black diamonds).
Then, they're not so tolerant to high voltages, and sometimes they can explode... for example where you'd use a 16v rated capacitor (for 12v), you'd need to use a 25-35v rated tantalum to be safe.

[jack black]

Re: are poly caps really better?

mariush,
thanks for that comprehensive explanation. i agree with everything you said about low ESR, rip current rating, size, etc. the part i focused in my post is the reliability/longevity of the caps. at least on the paper, most poly caps i could find on DigiKey had lower tested life than quality electrolyte caps that are cheaper. so, i'm thinking there is no need to switch to poly caps for the average recap fix.

[mockingbird]

Re: are poly caps really better?

The answer isn't so clear cut.

Let's hypothesize that there are two scenarios of 'cap longevity' so-to-speak:

1) The cap is operating within or siginifactly lower than its ripple spec, and far below 105C. Considering the self healing nature of electrolytic capacitors, and the fact that 40+ year old high quality electrolytics have been found to measure within spec, this cap might last up to 100 years.

2) The cap is still operating well below 105C but still well above room temperature. In addition to this, it is operating near its maximum ripple spec. A good example of this would be a motherboard VRM in a Pentium 4/Pentium D class system. An automobile is another good example, especially in colder climates where it will be subjected to extreme cold in the winter and extreme heat in the summer.

Additionally, I'd refer you to this:

If you overstress an electrolytic, hydrogen gas may form. With that proviso, the general answer to your Q is, "Yes." At a finer level of detail, there are something like three different solvents used in lytic caps, ethylene glycol, gamma butyro lacrone and one whose name I forget. Additionally, the ultra-low impedance types have elevated water content (e.g.: Nichicon HD, HE, HM, HN, HZ; Panasonic FM; Rubycon YXG, YXH, ZL, ZLH, MCZ; UCC KY, KZG, KZH, KZE). The latter technology seems to be finicky and more sensitive to ambient temperature. I'd be very wary of ultra-low impedance caps from vendors other than those four (and possibly Samxon); I'm not sure whether Sanyo has parts with that electrolyte technology. As some may point out, there have been problems with Nichicon HM and HN (mfg. process related) and possibly with UCC KZG.

And Pete also has a post where he talks about testing various series while working at a company, but I can't find it right now.

The earliest polymer capacitors AFAIK are the purple Sanyo OS-Con. To my knowledge, they've been in use from at least the mid to late 90s, so I submit that you would have to collect information on how well they've held up over the years, to compare them against electrolytics. Yes, they look more impressive on paper, but they've not been in service long enough to accurately measure their reliability.

[Wester547]

Re: are poly caps really better?

Pete's posts:

https://www.badcaps.net/forum/showpo...9&postcount=30

https://www.badcaps.net/forum/showpo...4&postcount=23

The purple Sanyo OS-CONs (SP on motherboards) aren't actually polymer capacitors. They are solid electrolytics but don't use conductive functional polymer material for the cathode. Their electrolyte is formed of TCNQ complex salt (organic semiconductor material) which has a lower melting point (and lower conductivity, though all solid capacitors do achieve conductivity by way of electrons as opposed to wet lytics which do so via ions) than polymers and the ripple rating requires temperature multipliers, unlike with polymers. Sanyo OS-CON SPs are sealed with resin, unlike most other polymers which have rubber bungs, that's another way to tell them apart.

There was another series of capacitors which uses the same technology as Sanyo OS-CON SP, Chemi-con OS-CON FP (I believe Chemi-con actually acquired that line of solid capacitors from Marcon), also sealed with epoxy. Went out of production a long time ago but is about a direct cross to Sanyo OS-CON SP.

Sanyo OS-CON SP datasheet:



Chemi-con OS-CON FP datasheet:



As for electrolytics lasting 40 years, usually those are the ones sealed with something better than rubber. The rubber bungs are usually compromised before then. And as an addendum to what mariushm said about tantalums and voltages, the reason that tantalums don't handle surge voltages or even continuous voltages as well (or rather, slight voltage spikes and dips) as lytics has to do with the dielectric material used. Tantalums use different dielectric material (on the cathode and on the anode, IIRC) than lytics (a pentoxide tantalum material rather than an aluminum oxide film). Unfortunately, solid capacitors, polymer or otherwise, suffer from this issue as well, at least to some extent, as they too use different dielectric material (a form of polymer insulators with a lesser dielectric constant and strength). Look at the Fujitsu RE datasheet and notice the lower surge voltages:

https://cdn.badcaps-static.com/pdfs/...eac5c87a98.pdf

Compared to the KMG series, for an example:



This may be why you sometimes find 16V polymers failed on the +12V input of the VRM converter for the CPU, or just for the filtering otherwise. Polymers and solid capacitors don't deal with peak and surge currents and voltages as well as electrolytics (more sensitive to overvolting), though with their very high ripple rating peak currents may not be an issue. I think that issue (voltage sensitivity) is worse with solid tantalums (either made of manganese dioxide or polypyrrole), though, where the voltage may need to be derated as much as 50% or thereabouts.

[montemcguire]

Re: are poly caps really better?

I agree that Tantalum should be avoided. The only situation where it must be used is on the output of certain linear voltage regulators. A solid Ta cap has a specific, and somewhat high output impedance compared to modern caps. This high impedance is needed to assure stability of the regulator's control loop, so modern low impedance caps can't be used.

Fortunately, there are more and more alternatives to these ancient but very popular linear regulators, so hopefully, Tantalum's days are numbered. The regulators that require solid Ta are the LM317, LM337, 7800 and 7900 series, and their variants.

If a solid Ta cap must be used, it should be derated to 50% or less of its voltage rating, and measures should be taken to limit peak currents. Rapidly discharge a solid Ta and it'll usually catch fire - no fun!

[budm]

Re: are poly caps really better?

"Tantalum's days are numbered" a long way to go yet.

[mockingbird]

Re: are poly caps really better?

Wester547 -

Brilliant post, thanks for clearing things up. And I've seen those UCC "OS-CONs", I have an old Dell board here with a few of them.

Let me ask you a hypothetical -

You're an electrical engineer, you're under contract by a company to engineer and produce a certain circuit, that means to say - you deliver the end product to them under your own brand (I've noticed this is common for lighting engineers). Your device is going to be placed in a very hot environment.

Now you've got two choices of caps. One is a relatively new but untested long life electrolytic series by a trusted manufacturer. The other choice is polymer caps. You only have these two choices because you need the caps to be miniaturized to fit your application.

The caveat here is that this specialized eletrolytic series is extremely expensive, and the polymer caps are one third the price. If you take the polymer caps, you stand to make significantly more profit.

You obviously can't make any guarantees to the company regarding absolute longevity of the product, but there is an unwritten understanding between you and your client that they expect it to last a very long time, or at the very least, they don't want the capacitors to be the weak point in the product.

Which do you pick?

[Wester547]

Re: are poly caps really better?

Wester547 -

Brilliant post, thanks for clearing things up. And I've seen those UCC "OS-CONs", I have an old Dell board here with a few of them.

Thank you.

Let me ask you a hypothetical -

You're an electrical engineer, you're under contract by a company to engineer and produce a certain circuit, that means to say - you deliver the end product to them under your own brand (I've noticed this is common for lighting engineers). Your device is going to be placed in a very hot environment.

Now you've got two choices of caps. One is a relatively new but untested long life electrolytic series by a trusted manufacturer. The other choice is polymer caps. You only have these two choices because you need the caps to be miniaturized to fit your application.

The caveat here is that this specialized eletrolytic series is extremely expensive, and the polymer caps are one third the price. If you take the polymer caps, you stand to make significantly more profit.

You obviously can't make any guarantees to the company regarding absolute longevity of the product, but there is an unwritten understanding between you and your client that they expect it to last a very long time, or at the very least, they don't want the capacitors to be the weak point in the product.

Which do you pick?

The polymers, just because the wet electrolytics would likely dry up faster under the same conditions. I would overspec the polymers, though - IE, I would be sure to use 4V polymers or 6.3V polymers on the VRM output and 25V polymers on the +12V VRM input.

[stj]

Re: are poly caps really better?

you pick both,
you keep it to yourself and make it relate to the serial number.

then you can track the lifespan / failure-rate of the 2 types.