(If there are chemistry differences, FYI the part is WE's 870025574001.)
Incidentally, not the thing I was expecting to fail. I was testing my new current limiter, exercising the TVS diodes. With a too-heavy load connected, pressing "START" delivers a ~150ms pulse. Spamming "START" makes the unit run ~continuously. The TVS diodes handle this with aplomb, rising
thermal limit brings that to a halt (until it falls back below ~70C).
I tested some polymers that, when you increased the voltage, the first sign of anything happening was a lethal short at about 1.7x rated voltage.
And I tested some others (posted here a while back) that leaked like a regular aluminum, more and more current as the voltage increased, but no dramatic failure. So polymers aren't all the same.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
The prototype has some, uh what where they, Rubycon 68uF 100V something or other, I think it was? Around 0.1 ohm ESR, at the inlet of a constant-off-time controller dumping 20A at a gulp. So the ripple is pretty high, high enough that they get warm after a few seconds of run time. Warmer than anything else in the design actually, aside from the shunt resistor (which is just a hunk of metal, who cares).
These polymers are lower ESR and higher ripple, but I suspect they're rated like EPCOS film caps: at the bleeding edge. They'll do what they say on the tin, but not a smidge more.
I know polyester has a runaway loss tempco problem; I wonder if some (most?) polymers do?
I have a pair of 10's on there already, they take the edge off (the ripple is nice and smooth) but don't do squat for the bulk of it.
I'd need maybe a dozen of those in parallel, or half as many 1210s, but now the cost skyrockets (these polymers are quite cheap -- granted that they aren't enough as-is!), and as others have noted, ceramic is a short-circuit-in-waiting, and I'd rather not have very low ESR, piezoelectric ceramics on a hot-plugging node. (I could also put TVS on the input and output terminals, to deal with the hot plugging voltage surge, but that's using up even more precious space..)
Maybe I should double down searching for low-ESR electrolytics... hmm...
It's intended for 150ms pulses, with a maximum of maybe a dozen chained together at a time. I can afford some pro-rating of components here... just not quite this much :-)
Tim, no offence, but I think that something from your rich collection of past tenses would be much more appropriate here. You used to have four elcos on your board, now you just have a scent of them.
And all this bad luck has just happened to a brand new part. Imagine their condition after, say, 10 years of continuous operation. This circuit has a serious reliability flaw. For exactly this reason I'll stick to my bulky linear-mode inductors and delcare all the parts based on electrochemical principles verboten. The non-existent parts have the unique trait of infinite MTBF.
Well, scent of one. The other three seem to be fine.
Insides looked like this,
formatting link
which doesn't really tell much. Left to right: pins (they look like normal electrolytic pins, crimped to the foil), bottom rubber bung, enclosure, and winding. I assume the winding isn't normally this crusty and black (and smelly).
Reliability?
This is a benchtop device -- if it last a few years in continuous service, that's fine!
It only needs to perform its current limiting duties for a few seconds at a time. If the capacitor lifetime is measured in total hours of current limiting run time, that's more than enough.
I could even argue it's the customer's fault for running it too hard, defining "normal operation" as operation below the current limit threshold. Though personally, I do want a better level of reliability than that.
Don't assume the requirements of your domain apply to every product. Remember, engineering is multidisciplinary. Cost and reliability are factors, just as volts and amps are!
Why, was someone vying for it? Sounds pretty sucky, prophets always being rejected from society, burned at the stake, or suchlike...
Trump prophesyed that he'd make America great again, and he hasn't been burnt at the stake yet, though so far he's mainly made America grate on everybody else.
The world's climate scientists are prophesying anthropogenic global warming, and they haven't been burnt at the stake yet either (though Trump may have some ideas about that).
I usualy place them double voltage rated of max operating voltage.
The rating of tantalum or Polymers for discharge dV/dt might be more problematic, thus it is recommended to limit the max inrush or discharge amps by a series resistor =:-(
Therefore I totally kicked them out of all designs since many years.
Ceramics are placed widely in automotive applications, like I said it is mostly a package problem. If a bigger part is selected, special soldering technique for prototypes must be considered.
Piezo can lead to some acoustic noise problem. This is a combination of PCB, which can be solved by layout. I.E. place two parts, top bottom to minimise PCB bending.
You can bridge a resistor by small FET with very low RDSON after the cap is charged, requires maybe a mcu, which is not a cost driver anymore and capable of saving few analogue chips as well.
The actual performance of the different caps you may scope out by a high speed current probe. Cost nowadays much more than scopes, but if you exceed limits, what you have proven - it shows you the full background and guides your choice.
Blowing up an Elco by heat of internal losses is not impressive - try melting of the cap wires by HF ripple. Maybe this type of open failure is the one you are looking for ;-)
--
mit freundlichen Gruessen/ best regards Joerg Niggemeyer Dipl.Physiker
WEB: http://www.nucon.de https://www.led-temperature-protection.com
Steinbecker Muehlenweg 95, 21244 Buchholz idN, Germany
UST-IDNR.: DE 231373311, phone: +49 4181 290913, fax: +49 4181 350504
Indeed, voltage rating is utterly meaningless, just like current rating of ferrite beads. The thing saturates wherever it does.
Package size is no gurantee, either, on account of packages not always being maximally full of electrodes. The biggest (C*V) values in a given package should be fullest of electrodes, but who knows. And you don't always need max-size caps, so you're screwed outside of there anyway.
So, you have to pore over all the datasheets and databases (if the mfgs provide them at all) to find something that's actually useful. Ugh.
Long story short, double rated voltage isn't good enough. In smaller values (0.1uF or less at 0805, say) it's rare if they poop out even at rated voltage. Something like a 10uF 25V 1206 might be lucky to retain 50% at half rated...
Yeah, tant is straight out for me because of surge. Maybe not wet tant then, except...cost. :)
Polymer I guess should be good on surge, but I haven't seen anything for or against it.
Yeah, you either get flex termination, or use two in series (or both). I'm not that worried here, but ceramics just aren't big and cheap enough for what I need here.
FWIW, the failure method is probably overvoltage. I'm doing about 40V in full swing, kind of a lot for these 25V caps. Some types will put up with that (film caps are usually /tested/ at 1.5 x Vrated), others not so much. Offhand, I don't see a surge voltage rating on these caps at all.
Anything can fail shorted, but it's extremely rare for a ceramic cap to do that. It's usually caused by mechanical damage. Ceramics can typically stand many times rated voltage.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.