On Sun, 09 Dec 2012 14:35:59 -0500, Spehro Pefhany
wrote:
Here's yet another formula:
http://www.low-esr.com/endurance.asp
L2 = L1 * 2^(T1-T2/10) = 2000 * 2^((105-30/10)) =
L2 = 362,000 hrs = 41.3 years
I assumed 55 C temperature, so you'd get 7 years with that formula.
Oops. I assumed that ambient meant room temperature or ambient for
the entire thermal system, not the maximum capacitor operating
temperature. It would have been helpful if the various web pages had
specified operating temperature, not ambient temperature. Using 55C
results in more realistic lifetimes. Thanks.
The transformer core and copper loses and capacitor ripple current
both heat the cap (plus a bit from the diodes and thermal fuse), and
the adapter will heat the air around it to some extent. If the ambient
is 30°C (ambient for the adapter), the cap will be warmer when the
adapter is plugged in and much warmer when it is doing something.
I've seen the cheap phenol PCB's under the diodes appear somewhat
scorched from overheating. I haven't seen that under the capacitor.
Using my IR gun type thermometer, the warmest wall wart I could find
in the house was 35C, with an (ambient) room temperature of 18C.
Obviously, the insides will be warmer, especially if the wall wart is
running into transformer core saturation. All of my switchers are the
same as ambient.
Some of the crappier ones even use 85°C caps.
Yep. The electrolyte in 85C caps boils at about 125C, while the 105C
caps boil at about 190C. Other than that, there's not much difference
between the two types. If you can keep the temperature low, 85C works
just as well as 105C. Above 100C(?), the 85C cap develop increased
leakage current, while the 105C doesn't leak.
On a previous cost cutting expedition for a former employer, I
recommended switching to 85C caps. Total parts cost savings would
have been about $0.40/radio or maybe $5,000 over the life of the
product. The reaction was predictable as I was immediately accused of
being a heretic, enemy of the state, and in league with the devil.
Walmart quality was thrown in several times. Although it was decided
that the possible gain was not worth the risk, I took the time to
replace all the 105C caps with 85C caps in the radio, attach
thermistors to all the major electrolytic capacitors, and run a 30C to
70C temp test in the env chamber. (The reason for the 30C is that we
were low on CO2 for cooling and I didn't care about low temperatures).
Since I had to do this anyway to run a preliminary test on some of the
other cost cutting measures, it was an easy test. Since there were no
ripple current, there was no self heating, so the capacitor
temperatures matched that of the oven. I ran a few basic tests and
found nothing that would suggest performance deterioration or failure.
Upon presenting my results to the committee, I was again labeled a
heretic, etc...
Some of them use crappy
lams so they run quite warm to the touch even without load. I'll bet
some Chinese caps are 85C marked as 105C.
Of course. Why counterfeit a marginal product when the better quality
product is more sellable. Extra credit for fake mil-spec markings.
The ones I spec'd I had reasonable confidence they will last 5 years
under normal use (so far, so good), but I don't think it would be
harder to consistently kill them faster than that.
Capacitor brands that work are a favorite topic in the numerous "Bad
Caps" forums. What most people seem to ignore is that all the
capacitor manufacturers offer different technologies and formulations.
Some work, some don't. If someone designs a board which requires high
quality low-ESR caps in order to keep the self heating down, and
someone later substitutes a different grade from the same
manufacturer, of course it's going to fail. It's like buying anything
solely by brand name. Every brand has its winners and losers.
There are extended life capacitors, with up to 10,000 hrs rated life
for commodity components. I don't have a clue on pricing. Example:
http://www.niccomp.com/help/presentations/AlumE-CapExtendedOperation0809-revA.pdf
http://www.low-esr.com/endurance.asp
From the prices I've seen, it would cost ~10x as much to use the
cheapest such 10K-hour caps (maybe doubling the cost of the adapter).
Ouch.
They're also physically larger, pushing up the size of the case,
pushing up the size of the packaging, and allowing fewer units to fit
in a TEU, so increasing sea shipping costs.
If you can get, say, 7+ years typical life, and 99% will last more
than 5, meet all technical specs and safety requirements, as well as
meet the price point marketing wants, what would you choose?
It's usually not my decision. However, the cycle seems to be to
initially specify the best quality parts. Contrary to what common
sense would suggest, Rev 1.0 usually works fairly well. However, just
before the product hits volume production, the cost cutting committee
arrives and substitutes the absolute cheapest parts that can be found.
When the failures and returns start piling up, an over priced
consultant discovers that none of the production units were actually
built according to the original parts list. Someone mumbles something
about depleting the current stock, so the failures and returns
continue to pile up. Better quality parts are eventually introduced
just before the product is labeled end of life.
--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060
http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558