United Chemi-Con PSG with 20,000h endurance!

[Per Hansson]

Was browsing Mouser looking to get some new 1500µF 10v caps in 10x16mm size to replace UCC KZG and Nichicon HM rated at 0.018Ω ESR and 2000mA ripple.
No such caps in electrolytic versions really exist from reputable brands anymore.

So I looked at the solid polymer range and came across the updated version of the UCC PSG.
It's rated for 0.008Ω ESR and 7700mA ripple with an endurance of 20000h, that is 10 times the endurance of the KZG & HM electrolytics!
Have not seen anything like that before!
I have been using Sanyo SEPC 820µF 4v solid polymers in place, they are rated for 0.007Ω ESR and 6640mA ripple, but "just" 5000h endurance.
Might actually get the 1000µF version of the UCC PSG next time I order, as it's ESR is slightly higher at 0.012Ω putting it closer to the originals

If Z height is no objection then the electrolytic Panasonic FR 1500µF 16v at 10x20mm with 0.020Ω ESR and 2180mA ripple is a given choice.
Being rated at 10000h it's a pretty much identical replacement.
But as can be seen in the picture that can not always be used everywhere

[PeteS in CA]

You're a Mouser Browser?

More seriously, back when I was looking at polymers for Curtis Inst. they were usually rated for 2000 hours. They've come a long way!

[mockingbird]

Out of curiosity, I opened up two polymers last night. The first, a Sanyo OSCON, probably around 15 or so years old. The second, a Fujitsu bumblebee.

These are both polymer caps. People have speculated that the bumblebees were "hybrid polymer", but I'm not sure what they meant by it, because it looked like an ordinary polymer to me.

Now the surprising part was that BOTH the polymer caps had dry, crusty insulative polymer layers.

It got me thinking: Was this just a result of early polymer technology, or will modern polymers also degrade to this degree over time. Seeing your thread title gets me thinking again whether polymers are indeed better than good electrolytics over a long period of time.

The 20,000 hour load multiple (the rule that longevity increases with lower stress), may not apply to polymer capacitors. Yes, they may be able to withstand a harsh environment in the longer run versus an electrolytic cap, but they probably can't compete with an electrolytic cap in the long run under low-stress conditions.

[stj]

oscon's do age badly.
i cant say if anything else does because i have only seen really old oscon's

[Wester547]

Yeah, Chemi-con has been rating their polymers for 15,000 hours - 20,000 hours at 105ºC at the rated voltage for a while now. That’s not at the rated ripple current, however, which would increase the core temperature (in the case of polymers since they are rated for very high ripple ratings) an additional 20ºC (105ºC polymers have a maximum core temperature of 125ºC), thus decreasing the lifespan to 1,500 - 2,000 hours.

Nichicon confirms this much, if you use Nichicon’s online capacitor calculator and put in the maximum rated ripple current in both ripple current fields, then calculate the results. In accordance to this document by Chemi-con, the formula for calculating life expectancy for polymers is different to that of electrolytics (pages 11 and 12). A polymer’s lifespan will increase 10x per 20ºC drop (although polymers that are sealed with rubber are still only rated for 15 years max, in accordance to that calculator and Chemi-con).

Originally Posted by mockingbird

Out of curiosity, I opened up two polymers last night. The first, a Sanyo OSCON, probably around 15 or so years old. The second, a Fujitsu bumblebee.

These are both polymer caps. People have speculated that the bumblebees were "hybrid polymer", but I'm not sure what they meant by it, because it looked like an ordinary polymer to me.

Now the surprising part was that BOTH the polymer caps had dry, crusty insulative polymer layers.

Yep, willawake dissected a bumblebee Fujitsu polymer many years ago and found the same thing.

Rubycon’s PZA series also has a vent stamp, but there is no mention on their site of PZA being a “hybrid” in any sense of the word. PCBONEZ was somehow (may still be somehow) convinced that the older Fujitsu polymers are hybrids because they are “functional” polymers, but looking at page 3 of that Chemi-con document, “functional polymer” is just another term for solid polymer. Hybrids do exist, of course, but their specs are not as good as regular polymers.

Those early Fujitsu polymers were inconsistent insofar as reliability. Sometimes they would slightly bloat and go open after a couple years, other times they’d last much longer... quite a few 16+ year old Fujitsu polymers are still good.

It got me thinking: Was this just a result of early polymer technology, or will modern polymers also degrade to this degree over time. Seeing your thread title gets me thinking again whether polymers are indeed better than good electrolytics over a long period of time.

The 20,000 hour load multiple (the rule that longevity increases with lower stress), may not apply to polymer capacitors. Yes, they may be able to withstand a harsh environment in the longer run versus an electrolytic cap, but they probably can't compete with an electrolytic cap in the long run under low-stress conditions.

The one advantage liquid electrolytics have over polymers is a genuine self-healing mechanism: so long as the pH balance of the electrolyte hasn’t risen and the electrolyte hasn’t dried up, they are very capable of correcting any defects or issues in the anodic oxide layer, as long as there is bias applied to the plates. Polymers are capable of isolating faults but not correcting them. So it’s possible for a polymer to fail by way of a random short-circuit after many years, especially since leakage current goes up.

However, so long as that doesn’t happen, I would peg polymers as more reliable than even good electrolytics. The reason being that good electrolytics will eventually dry out (or possibly leak) even with a good rubber bung. Electrolytics also rely upon hydrogen scavengers, hydrogen absorbers, depolarizers, neutralizers, oxidizers, and other additives to preclude the generation of hydrogen gas, a process that is always occurring even minimally in all electrolytics. Polymers do not have a liquid solution that is bound to eventually decompose (although they may experience field crystallization over time, decreasing the conductivity of the polymer electrolyte).

As for Sanyo OS-CON, there are two different types of “OS-CON” capacitors: those that use a conductive functional polymer (PEDOT) as the cathode layer material (those unsleeved) and those that use TCNQ complex salt (those with the lilac sleeve, which were discontinued 7 years back). Those have a much lower pyrolysis temperature than polymers, and also rely on temperature multipliers for the ripple rating (unlike polymers). I’m guessing the latter are the ones that age badly.

[stj]

the degraded poly's i saw where the purple sleeve ones - the blue-sleeve ones seem less degraded with age.
both pulled from the same intel motherboards - so had identical hours on them.

[mockingbird]

Originally Posted by Wester547

Those early Fujitsu polymers were inconsistent insofar as reliability. Sometimes they would slightly bloat and go open after a couple years, other times they’d last much longer... quite a few 16+ year old Fujitsu polymers are still good.

Even the crusty ones probably still work ok. It would be interesting to see exactly what effect there is when the polymer dries out some. I'm in the process of constructing my ESR meter so I'll be able to check them in the future.

Regarding the bloating, it was also mentioned at the time IIRC that bloat on bumblebees was harmless and that the top should just be manually pushed back in.

However, so long as that doesn’t happen, I would peg polymers as more reliable than even good electrolytics. The reason being that good electrolytics will eventually dry out (or possibly leak) even with a good rubber bung. Electrolytics also rely upon hydrogen scavengers, hydrogen absorbers, depolarizers, neutralizers, oxidizers, and other additives to preclude the generation of hydrogen gas, a process that is always occurring even minimally in all electrolytics. Polymers do not have a liquid solution that is bound to eventually decompose (although they may experience field crystallization over time, decreasing the conductivity of the polymer electrolyte).

What do you reckon then caused the polymer to harden up like that? Plastic shouldn't degrade like that when it's not exposed to UV, right?

As for Sanyo OS-CON, there are two different types of “OS-CON” capacitors: those that use a conductive functional polymer (PEDOT) as the cathode layer material (those unsleeved) and those that use TCNQ complex salt (those with the lilac sleeve, which were discontinued 7 years back). Those have a much lower pyrolysis temperature than polymers, and also rely on temperature multipliers for the ripple rating (unlike polymers). I’m guessing the latter are the ones that age badly.

Yes, correct... They also have an epoxy bunge (as in, the epoxy was filled in to form and stiffen during manufacture).

Originally Posted by stj

the degraded poly's i saw where the purple sleeve ones - the blue-sleeve ones seem less degraded with age.
both pulled from the same intel motherboards - so had identical hours on them.

You're referring to the UCC OS-CONs. I have some of those too I should really have a peek inside one of them.

[Wester547]

Regarding the bloating, it was also mentioned at the time IIRC that bloat on bumblebees was harmless and that the top should just be manually pushed back in.

They are known to fail when they bloat too.

What do you reckon then caused the polymer to harden up like that? Plastic shouldn't degrade like that when it's not exposed to UV, right?

Were the OS-CON with crusty layers of polymer sleeved or unsleeved? It’s possible the Fujitsu failed simply due to inadequate rubber bungs which allowed for moisture ingress (that would eventually volatize to steam) and the degradation of the functional polymer layer.

Yes, correct... They also have an epoxy bunge (as in, the epoxy was filled in to form and stiffen during manufacture).

Yes, the TCNQ OS-CON capacitors use epoxy-coated rubber bungs. Although I’m not sure they helped much as stj observed they would still degrade over time (at least leakage current wise).

You're referring to the UCC OS-CONs. I have some of those too I should really have a peek inside one of them.

The NCC OS-CONs had dark blue sleeves, but the Sanyo OS-CONs came with both blue and purple sleeves.

[stj]

sanyo ones - light sky blue sleeves.
they are distinct series - but i dont have the datasheets handy.

[mockingbird]

Originally Posted by Wester547

Were the OS-CON with crusty layers of polymer sleeved or unsleeved? It’s possible the Fujitsu failed simply due to inadequate rubber bungs which allowed for moisture ingress (that would eventually volatize to steam) and the degradation of the functional polymer layer.

They were the sleeved ones... You know... the really, really early ones.

Yes, the TCNQ OS-CON capacitors use epoxy-coated rubber bungs. Although I’m not sure they helped much as stj observed they would still degrade over time (at least leakage current wise).

Are you sure it's a rubber bunge coated in epoxy? I had to use a large strong pair of grip pliers to crack the epoxy before I could remove it. Once it was cracked it was pretty easy to remove and it fell right out (and the can kept its form relatively well too). I did not see any rubber in there. It looked like it was just simply black epoxy. In fact, some of it was still stuck to the aluminum/polymer coil bottom.

[stj]

i doubt there was a bung too, the can has no crimp-ridge

[Wester547]

Originally Posted by mockingbird

They were the sleeved ones... You know... the really, really early ones.

They weren’t technically polymer then, although a solid capacitor all the same. It’s intriguing that they both had crusty layers.

Are you sure it's a rubber bunge coated in epoxy? I had to use a large strong pair of grip pliers to crack the epoxy before I could remove it. Once it was cracked it was pretty easy to remove and it fell right out (and the can kept its form relatively well too). I did not see any rubber in there. It looked like it was just simply black epoxy. In fact, some of it was still stuck to the aluminum/polymer coil bottom.

You are right, I just poked around the epoxy of a Sanyo OS-CON SP and could not find any rubber. I think there were some old American capacitors from the 80s that did put epoxy over the rubber bung so as to prevent them from leaking. I think the confusion on my part also stemmed from the fact that the manufacturers refer to the epoxy seals as epoxy *end* seals, or epoxy coated end seals. Furthermore, I just looked at a document which makes no mention of rubber bungs on the sleeved OS-CON capacitors (at the bottom of the second page).

[Per Hansson]

Originally Posted by Wester547

Yeah, Chemi-con has been rating their polymers for 15,000 hours - 20,000 hours at 105ºC at the rated voltage for a while now. That’s not at the rated ripple current, however, which would increase the core temperature (in the case of polymers since they are rated for very high ripple ratings) an additional 20ºC (105ºC polymers have a maximum core temperature of 125ºC), thus decreasing the lifespan to 1,500 - 2,000 hours.

That's very good information as always, didn't know that!
Of course I can understand why they did it, in one way it's unfair.
But it depends how you look at it.

Since the PSG caps in my example are rated for 4x the ripple of the KZG or HM caps they would be replacing of course they will not be exposed to that kind of ripple current.
So in all likelyhood they will last longer than I will

[Behemot]

There are still the Suncons WG, I do want to get them, but since I am not heavy on cash ATM, it will have to wait. Payment up front with this distributor. Considering how well other crowfunds ever happened here, guess that it will be no ultra-low ESR caps for yo guys anytime soon.

Maybe if you guys were browsing less through all these Mousers and crap, could have had all the necessary caps already…I use them too ya know. Just could not afford to stock everything I like cause most jobs usually don't even pay the full batch if I only need few pieces of it. Only occasionally, like now, I also need some bipolars or larger HV (100V) caps, like for this terrible Cambridge/Creative sub with all that shit inside.* There is work enough so it will maybe pay signifficantly more caps than I need. But there is still the thing with owing for rent for like year and a half now
___________
*It seems that during the first, warranty repair, they changed the original caps for another crap (Teapo SEK). They managed to get bad again, together with the 100V caps. Does anybody really need more evidence to believe only ANIMALS use crap caps for repairs?

[Per Hansson]

Yea Suncon WG really is a perfect replacement too.
I have a list of caps that are equivalent replacements
It's Samxon GC that I've used in the past.
And Rubycon MBZ and MCZ.
But as we know they are all EOL.

Anyway replacing them with what I wrote in the OP is really the best.
Sanyo WG don't exist in 10000h lifetime version, let alone 20000h

[Behemot]

Yep, that's why I want to det those when there is pretty much no alternative

None of the ultra-low ESR caps ever got over 2000 hours IIRC. Though I noticed they updated these and now the sheet lists 2000-4000 hours. Not THAT bad I guess.

[Clutchbox]

Wow! Those nice fat PSG series polymer makes perfect replacements for some larger 16v KZG-class caps (8x16 to 10x20).

[Behemot]

Well if what Wester says is true, the actual endurance is NOT listed at temperature AND ripple, and by recalculating it, you get the same 2000 hours…is it a much difference than?

[Wester547]

At temperatures below 105ºC, yes. Because again, the temperature coefficient for electrolytics is 2x the lifespan per 10ºC drop, and for polymers, it’s 10x the lifespan per 20ºC drop.

[Behemot]

Well according to my math, it makes 15 years at 55 °C for Suncon WG 10×23 and 22831 years for the PSGs at 2000 hours (real). So I think pretty much the same, 15+ years for MoBo? Like srsly?