Joe, you asked:
>Hi. I saw some of your notes on Tantulum caps. Do you folks have any
>experience with tantulum cased (instead of silver cased) tantulum caps?
>
>In order to prevent failure shorts from burning holes into our backplanes, we
>are looking at using tantulum cased caps, to avoid the dendritic growth
>shorting problem.
>
>Just wondering.
>
>Joe Dzekevich
>3Com ISD Reliability Engineering
I am not sure. I thought that tantalums all had slugs (capacitive elements)
with an interior made of tantalum, tantalum penoxide, and manganese
dioxide. The coating is the counterelectrode, layered with manganese
dioxide, graphite, and silver.
If this is what you refer to as the "case", then I'm unaware of the
distinction regarding damage from shorting. When you say dendritic growth
shorting, do you refer to problems with multilayer ceramic caps (maybe not,
but I thought that was where the thread left off...)?
IMO, Ta caps are more reliable and stable than ceramic, and pack more
Farads per footprint. But they will sure as heck make a big loud pow if you
put them on backwards (they are polarized), which is not a concern with
ceramics. You also have random failures in properly polarized devices due
to DC leakage. And, although this is not in the vendor literature, I have
seen parts marked backwards, so that they blow up even when placed properly
(and we had positive proof!).
AVX (and probably others) make an internally fused Ta cap, they call them
the TAS series. If your objective is reducing the possibility of circuit
damage, these might make a good subsitution, but these are 1-100 uF,
typically for bulk decoupling. They are supposed to be able to react to
even leakage, and go open quickly enough to prevent damage.
In general, the ceramics are much lower capacitance, and are used for low
impedance device decoupling in the 0.001 - 0.1 uF range. They have to be
connected with short traces to the VCC and GND right next to the part,
that's why there are so many.
MLC's are the salt and pepper of digital circuit design. I think one of the
big discrete device suppliers paid some EE textbook people to recommend two
caps per active device maybe 20-30 years ago, and doomed me to a career of
stuffing little 0.01's by the thousand. ;-)
In my experience, both tantalums and ceramics will fail. Ta, due to
misplacement. Ceramics, due mainly to thermal stress, with an occasional
suspected weak lot from the supplier. Over the years, I guess the problem
of misplacing the Ta part is a more dominant failure mode, even considering
the lower population density. If you have that mode controlled, then the
ceramic could be a greater concern. Here at Interphase, I have seen more
small MLC's fail than tantalums.
I have a paper here from AVX/Kyocera called "Reliablility Management of
Tantalum Capacitors" by Chris Reynolds, 1991. Call AVX at 803-448-9411,
they would no doubt send you a copy.
regards,
Jerry Cupples
Interphase Corporation
Dallas, TX USA
http://www.iphase.com
|
