Radial Electrolytic vs SPA capacitors

  • Thread starter Thread starter johns
  • Start date Start date
J

johns

I have a CAD lab with about 15 boxes now using
P35 chipset mobos that have them. From experience
electrolytics have a shelf life of maybe 3 years, and
then they begin to lose their rated capacitance.
It drops to about half its value in 3 years. As buffers
for circuits needing power in surges, that will work
but it can generate flakey operation. The other
failure mode is sudden shorts that will cause a
reboot for no obvious reason. That is fine for most
mobos, because they are badly out of date after
3 years on the job. However, I think the most
recent chipsets are so fast, that I would like for
a business PC to last maybe 5 years or more.
The technology is really pretty good now, and
a long-life box is a good investment. And I would
especially like to see some long-life power supplies
with no electrolytics in them.

johns
 
Also, one company that got hit bad was Antec. Their SmartPower
power supplies had failed electrolytics in the stand-by circuit.
The failure mode was that after the box went into standby, it
could not recover. That was more a heat problem than a
slurry problem in the cap itself. Electrolytics cannot take much
heat ... even the best of them. The symptom there is a very
short life time with capacitance dropping to 50% in a year.
For that reason, electrolytics do not belong in modern high
power systems that run hot. I make every effort to build
systems using cases with 120 mm quiet fans that can run
fast enough to keep the ambient close to room temp.
If you put a P35 chipset mobo in an old case with small
fans ... good luck !!!

johns
 
Here's what a pair of the cheap ones look like: (one on left is blown)
These are radial electrolytic capacitors

http://www.dwarfsoft.com/GameDev/blowncap.jpg


And here's some Solid Polymer Aluminum (SPA) capacitors:

http://www.pcstats.com/articleimages/200604/asusp5n32esliplus_m1.jpg

http://www.gigabyte.com.tw/FileList/NewTech/2006_motherboard_newtech/article_02_all_solid.htm

Info on the capacitor plague
http://en.wikipedia.org/wiki/Capacitor_plague

Why you should not leave your computer on all the time (par. starting with
'Pentium IV')
http://www.dashdist.com/1u2u/company/capacitor.html



I'm curious - the benefits of SPA capacitors seem difficult to argue
against, but how many of you have boards that exclusively use them?
 
OhioGuy said:
Why you should not leave your computer on all the time (par. starting with
'Pentium IV')
http://www.dashdist.com/1u2u/company/capacitor.html

That one seems to be more of an ad for Jetway.
I'm curious - the benefits of SPA capacitors seem difficult to argue
against, but how many of you have boards that exclusively use them?

My video card seems to use them, but that's of little consequence since the POS
bushing fan is going out. :(
 
johns said:
I have a CAD lab with about 15 boxes now using
P35 chipset mobos that have them. From experience
electrolytics have a shelf life of maybe 3 years, and
then they begin to lose their rated capacitance.
It drops to about half its value in 3 years. As buffers
for circuits needing power in surges, that will work
but it can generate flakey operation. The other
failure mode is sudden shorts that will cause a
reboot for no obvious reason. That is fine for most
mobos, because they are badly out of date after
3 years on the job. However, I think the most
recent chipsets are so fast, that I would like for
a business PC to last maybe 5 years or more.
The technology is really pretty good now, and
a long-life box is a good investment. And I would
especially like to see some long-life power supplies
with no electrolytics in them.

johns

You can read about why there were increased failures
in aluminum electrolytic capacitors here. It doesn't
mean the products have to be bad. The bad capacitors
lack a stabilizer component in the electrolyte formula,
which causes the higher than normal failure rate.

http://web.archive.org/web/20030219...rum.ieee.org/WEBONLY/resource/feb03/ncap.html

Many capacitor companies, offer basic technical information
about electrolytic capacitors. For example, this article
breaks the issues down into two pieces. An operating
temperature related, reliability term, and an end of life
related to physical construction (limiting life to about 15
years in their estimation).

http://www.illinoiscapacitor.com/uploads/papers_application/85560DAA867C4AE2871F2EFA1749A6C7.pdf

The temperature based component works like this. The cap
may be rated at 3000 hours at 105C (a value you might find in
the catalog). If the air around the cap is 95C, it lasts for
6000 hours. If the air is 85C, the cap lasts for 12000 hours.
By the time we get the air cooled to 45C, the life is 192000 hours.
Now, that number is past the 15 year point, so now the seal is
becoming more important, than the temperature related
reliability effect.

As long as the design remains reasonably cool, and as long
as the capacitor does not have excessive ripple current
(measured in amps) flowing through it, to the point that
there is significant "ESR heating", the cap should last
a long time. The reason they are used in clumps of five or six
on the motherboard, is to spread the ripple current over more
caps. One datasheet I was reading for a Vcore converter,
mentioned that the absolute capacitance value, wasn't that
critical, as the main reason for the number of caps, was
to meet the ripple rating needed.

BTW - what brand and model of capacitance meter did you use,
to make those capacitance measurements ?

Paul
 
Electrolytics cannot take much heat ... even the best of >them.

Considering that there are a lot of overclockers out there, and
overclocking generates serious amounts of heat (& some of the capacitors are
right near the CPU), do overclockers tend to buy motherboards with the SPA
capacitors?
 
Considering that there are a lot of overclockers out there, and
overclocking generates serious amounts of heat (& some of the capacitors are
right near the CPU), do overclockers tend to buy motherboards with the SPA
capacitors?

There're overclockers and then there're overclockers. Can mean as a
little edge on "free" performance, all the way to balls-to-the-wall
water or reference refrigerator-cased overclocking. Premature failure
can be a result. On the "free" OC side, basically it's running
slightly out of MB/CPU specs, at times seemingly a manufacturer
oversight stipulated on the safe side of chipset ratings --
differences between CPU assembly dies, notably, as word quickly
spreads what revisions and make lend to an easy OC. Practically no
difference whatsoever from a lesser-rated chip when not excessive.
Even unscrupulous vendors have been known to overclock MB/CPUs for
clip-shot markup prices from the uninformed.
 
Back
Top