On 06/29/2013 9:40 PM, Jeff Liebermann wrote:
On Sun, 30 Jun 2013 11:34:29 +1000, "Phil Allison"
wrote:
"Jeff Liebermann
No, No, and no. They're the absolute maximum operating temperature,
at which the maximum safe applied voltage hits zero.
85C, 105C and 125C are max usable temperatures at the rated DC voltage.
http://www.navsea.navy.mil/nswc/crane/sd18/Images1/Capacitors/CapacitorsDerating14b2.gif
Do you see where it says "absolute maximum rating"? Notice the
recommended operating voltage at the absolute maximum rating point for
all 3 caps. It's zero because it's not recommended running the
capacitor at the absolute maximum rating. That also includes
semiconductors, automobiles, and thermionic valves, all of which to
disgusting things when operating at their absolute maximum rating.
Wrong. A capacitor only draws current when the voltage across the
leads changes.
** What about leakage ?
What about it? At the typical computer and solid state LCD monitor
voltages, its negligible. It might a problem in hi voltage caps, as
in your tube stuff, solar power inverters, or in electric vehicles,
but not low voltage computah and LCD panels.
The capacitor only dissipated power, and converts it
to heat, when the voltage changes. Pure DC across a capacitor does
nothing to produce heat.
** What about leakage ??
2mA of leakage times 500 volts = 1 watt.
Show me where I can find 500 volts in an LCD monitor or TV?
Incidentally, todays monitors and TV's use LED backlighting, not CCFL
tubes.
Ignoring frequency dependent effects, the power dissipated is:
Power = Ripple_voltage^2 / ESR
or
Power = Ripple_current^2 * ESR
where ESR = equivalent series resistance. For example, the CPU filter
caps are typically 1000uF/6.3V electrolytics (0.12 ohms ESR). With a
current probe, I can usually see at least 4A average ripple current on
the CPU power leads. While trying to keep the voltage constant over
such large current variations, each cap would smoke:
P = 4^2 * 0.12 = 2 watts each.
** ESR is not a fixed number - it varies dramatically with temperature.
In your world, nothing happens without drama.
The ESR measured at room temp is typically 5 to 10 times higher than when
the cap at say 80C. Check this out with any electro, an ESR meter and hot
air any time you like - electrolytes become way more conductive when HOT.
This makes nonsense of your calculation.
Ummm... yeah. See:
http://www.edn.com/design/components-and-packaging/4396282/Power-Tip--50--Avoid-these-common-aluminum-electrolytic-capacitor-pitfalls
Look at Fig 2 and:
http://urrg.eng.usm.my/index.php?option=com_content&view=article&id=249:l inear-and-switching-power-supply-fundamentals-part-14-&catid=31:articles&Itemid=70
At 25C the ESR of their electrolytic is unity (at 100KHz). When the
temp climbs to 85C, it's 0.3 or 0.4 (as best as I can read the
graphs). That makes the room temp value about 3 or 4 times the ESR at
85C, not your 5 to 10 times value. The internal dissipation change
between room temp and 85C is 1/3. My 2 watts of smoke becomes 0.67
watts.
The problem is that nobody runs electrolytics at 85C. Even with 40C
maximum ambient for the monitor, I doubt if anything gets hotter than
maybe 55C. It might be hotter in a computer, where the caps are
wrapped around a hot CPU, but those are usually polymer caps, not
electrolytics.
My guess(tm) is that designers take advantage of the drop in ESR, and
run the ripple current even higher, thus negating any alleged benefits
to running the caps hot.
The capacitance of an electrolytic increases about 5% from 25C to 85C,
which would reduce the ripple voltage by about the same percentage.
That also helps keep the cap cool, but the effect is not large
compared to the change in ESR with temperature.
(Hint: Always design for the worst case, which always seems to happen
at inconvenient times).
Look at Fig 13 in the URL I cited. At 85C, the maximum working
voltage is zero.
All electros are speced for full voltage at max rated temp.
Nothing is ever specified simultaneously at maximum voltage and
maximum temperature. If you're ever tried to run a semiconductor at
more than one of the maximum ratings at the same time, you will likely
have a pile of smoking silicon. Same with an electrolytic capacitor.
All that the max rating really mean is that you can possibly hit *ONE*
of those ratings, and not destroy the part.
For fun, and when it cools down somewhat, I'll make some boiling water
(for tea) and drop in an electrolytic while measuring the ESR with my
Bob Parker ESR meter. It should be interesting to see if practice
follows theory.
But at max temp, the rated life is typically only a few thousand hours -
before the electrolyte vanishes.
Yep. That's because the rated lifetime of a capacitor is specified at
the rated maximum temperature. At 85C, you'll get about 2000 hrs.
Drop the temp 10C, and the expected life will double.
You can get a fair idea of how it works if you ignore ripple current
heating for now:
http://www.illinoiscapacitor.com/tech-center/life-calculators.aspx
Plugging in:
Rated life = 2000 hrs
Rated max voltage = 6.3 VDC
Operating voltage = 5.0 VDC
Max temp rating = 85 C
Ambient = 40 C
Projected Life = 57,000 hrs = 2,375 days = 6.5 years
which is about what I'm seeing with commodity capacitors in ATX power
supplies. If I add 10C to the ambient for self heating by ripple
current (which was somehow left out of the calculator), I get:
Projected Life = 28,511 hrs = 1,188 days = 3.3 years
which is close to what I see with computers running in a burn-in room.
___________________________
/| /| | |
||__|| | Please don't |
/ O O\__ feed |
/ \ the trolls |
/ \ \ |
/ _ \ \ ----------------------
/ |\____\ \ ||
/ | | | |\____/ ||
/ \|_|_|/ | __||
/ / \ |____| ||
/ | | /| | --|
| | |// |____ --|
* _ | |_|_|_| | \-/
*-- _--\ _ \ // |
/ _ \\ _ // | /
* / \_ /- | - | |
* ___ c_c_c_C/ \C_c_c_c____________
http://jni.sdf-eu.org/trolls.html
While Phil may actually know what he is talking about - the way he
presents the information - not to mention the personal attacks, places
him in the troll category and until his manners improve he should
simply be ignored.
I don't read his posts, while I do read yours...
John :-#)#
--
(Please post followups or tech enquiries to the newsgroup)
John's Jukes Ltd. 2343 Main St., Vancouver, BC, Canada V5T 3C9
Call (604)872-5757 or Fax 872-2010 (Pinballs, Jukes, Video Games)
www.flippers.com
"Old pinballers never die, they just flip out."