Endurance of tantalum caps

[Klaus Kragelund]

Hi

If we ignore the discussion about tantalums and dv/dt susceptibility, and focus on endurance. I have looked at different caps

http://www.vishay.com/docs/40004/592d.pdf

Goes into FIT numbers in this page:

http://www.vishay.com/docs/40110/faq.pdf
(last page)

Stating a FIT of 5, which is good

Another one from Nichecon:

http://www.nichicon.co.jp/english/products/pdfs/e-f72_75.pdf

States endurance of 2000 hours at 85 degrees dropping 10% rated capacitance. But, will this curve stay one the same trend, so about 20% rated capacitance reduction at 4000 hours? (seems to me this is a very bad cap and can be used for next to nothing)

Anyone has experience of tantalums and capacitance versus lifetime in general?

Thanks

Klaus

 

[Robert Macy]

Hi

If we ignore the discussion about tantalums and dv/dt susceptibility, andfocus on endurance. I have looked at different caps

http://www.vishay.com/docs/40004/592d.pdf

Goes into FIT numbers in this page:

http://www.vishay.com/docs/40110/faq.pdf
(last page)

Stating a FIT of 5, which is good

Another one from Nichecon:

http://www.nichicon.co.jp/english/products/pdfs/e-f72_75.pdf

States endurance of 2000 hours at 85 degrees dropping 10% rated capacitance. But, will this curve stay one the same trend, so about 20% rated capacitance reduction at 4000 hours? (seems to me this is a very bad cap and can be used for next to nothing)

Anyone has experience of tantalums and capacitance versus lifetime in general?

Thanks

Klaus

What mechanism 'ages' a tantalum cap?

Low esr?! wow, that esr is really 'case' sensitive.

The Vishay data sheet, pg 9, has the 'fishiest' looking plots I've
ever seen. Like to extend your question to ask if esr indeed acts so
perversely? And, what mechanism causes the esr to increase so much as
the frequency approaches DC?

 

[Tim Williams]

And, what mechanism causes the esr to increase so much as
the frequency approaches DC?

Damping factor (dielectric loss). Which is usually modeled as a parallel
R+C or something like that (at a given frequency), but then, it's
*equivalent* series resistance, so that's okay.

Tim