Very durable electrolytics or similar?

Right, that's one reason why even selenium rectifiers in old radios and TV set were replaced with a diode and a series resistor, not just the diode itself. And then the capacitors would still last years.

I believe that they simply made better quality caps back then. Definitely not oversized ones because almost every one of these companies had a few people like Earl Muntz. Yet people expected their radios to last pretty much forever. Else the reputation of a company would be toast (and that happened).

--
Regards, Joerg 

http://www.analogconsultants.com/

"Joerg"

** Have you got a PhD in trolling yet ???

.... Phil

Those look good although the ones Martin suggested seem to be even better. The 200V Nichicon caps can be had in 150C/2000h which is amazing. At 85C they'd have a theoretical 20 year life span. Since it is

85C at most 30% of the time it's more like 30+ years.

Much of it is confidential, like with any engineering project. But I believe I gave enough info. I do not yet know the exact ripple current but that doesn't matter, one simply takes more caps if higher. It all scales.

20-30 years is considered indefinitely in most circumstances, except with aircraft, cars and so on. There we try to design for more (in my case aerospace, I don't do much automotive yet).

--
Regards, Joerg 

http://www.analogconsultants.com/

Ahh Stefan-Boltzmann again. (I hope I don't f-up the algebra this time.) So 300 K radiates at ~500W/m^2. (onto a body at 0K.) For 1 kW/m^2 I need a temp that's 2^1/4 higher... I get about 356K = ~84C and that assumes that the environment that it "sees" is at 0K. It'll be much worse than that.

100 C doesn't seem impossible.

George H.

"George Herold"

** FFS - go f*ck yourself.

You pathetic, asinine, ASD f***ed TROLL.

.... Phil

High temperature operation does not imply long or extended life. And it wou ld not be wise to operate the cap at maximum WVDC, that is a big failure ac celeration stress, same goes for ripple current. Then the handling and boar d manufacture has to be specified, there are solvent/material incompatibili ties that may induce nascent corrosion pathways over the long term.

My 1924 set has 2uF psu caps. Evidently it ran on dc mains.

NT

As outlined in your first link, lifetime generally scales up by a factor of 2 for every 10C lowering in temperature. Coming down from 150C/2000h seems like the better deal when compared to 105C/20000h. But the NIC Components caps are certainly also good quality products.

I have never done that and never will.

Yup, electrolytics have to be dealt with at a similar level of care as batteries.

[...]

--
Regards, Joerg 

http://www.analogconsultants.com/

Transformer resistance is quite significant, at least with 50Hz lumps in do mestic appliances. A 10% regulation transformer would get you 14x mean curr ent if you shorted its output at V_peak, but in practice the reservoir cap is already fairly much charged so it seldom sees 10x. Higher R transformers give lower peaks of course.

NT

Hmm Well if the box is in an area at 300K then it's getting 500W/m^2 from the radiation around it and another 1kW from the sun. So it needs to radiate three times as much (1.5kW/m^2) 3^1/4 * 300 = 395 K ~120 C. But I expect you'll just yell some more.

George H.

See this :

If you think all you have to do is plop that thing on a board and then get record setting longevity, you're dreaming. The spec sheet should be viewed as just a start, but after that there are a hundred ways to destroy the ultimate performance.

I wonder why no-one here seems to be designing corner cutting goods. There's no shortage of them out there.

I once made a stuffing mistake, and a 16v cap ended up running at 21v the wrong way round. I never found out about it until a very long time later, it was still working perfectly.

NT

You do understand that a full one half the capacitors under test failed before 2000h at 150oC, by definition? Don't go around jumping up and down with glee until you can cite the distribution. Like I said before, that 50% MTBF is a weak benchmark.

On Mon, 31 Mar 2014 08:04:41 -0700 (PDT) snipped-for-privacy@care2.com wrote in Message id: :

That would be sci.electronics.design.ch

CDE flatpack series

I kinda realised that after I posted :)

NT

Which definition?

See page 7.

I believe its actually 10% (not 7%) and the standard is IEC60384-14.

I am never designing anything for 50%.

--
Regards, Joerg 

http://www.analogconsultants.com/

stimation_AAL.pdf

That's a good reference for those acceleration "pi" factors the MIL-HDBK us ed to call them.

I thought they trained you people better at the wissenschaften? Pathetic! G et a clue between a test that gives you an accurate /estimate/ of MTBF and one that actually measures it. The "M" in MTBF is mean (is average) lifetim e of the lot, now how would you get that mean with only 7% of sample failin g?

This guy is slow, but his formulas are right:

Using a simplified exponential, the probability of any single component act ually making it to MTBF is less than 37%.

_Estimation_AAL.pdf

used to call them.

Get a clue between a test that gives you an accurate /estimate/ of MTBF an d one that actually measures it. The "M" in MTBF is mean (is average) lifet ime of the lot, now how would you get that mean with only 7% of sample fail ing?

ctually making it to MTBF is less than 37%.

Reversing this, you need an 850,000 hour MTBF to claim 90% chance of making it to the 90,000 hour mark. Think you're doing pretty good so far?