Random thought on caps

[bronson]

Looking over the past few designs on my bench (some by me, some by others, some open source) I'm really impressed that:

There are almost no tants or lytics anywhere!

And it's not just batteries/mobile.  A lot of wall wart powered stuff is so efficient that it just takes a few chip capacitors to bypass everything.  And chippies basically last forever (right?  I've never seen one spontaneously fail)

So, I think I'll rework this design to get rid of the single tantalum.  Yay to living in the future.

Alas, the next generation may not understand the joy of scrobbling all over your scope trying to find the shorted cap and then replacing all the damage it did.   Ah well.

[T3sl4co1l]

Ceramics age, broadly, something like half capacity over... a decade or a few I think, for like X7R or X5R.  The crappier ones (Z5U, etc.) of course are worse.  I forget if this occurs under bias, by age alone, or both.  Capacity can be partially restored by annealing past the Curie temperature.

Kinda-sorta-on-topic plot: Z5U C vs. V plot.  Think this was a "203" part (20nF nominal), 8mm disk, random junk bin stock; test amplitude maybe 1V AC, maybe less.  Annealing was holding the soldering iron (~350C) on each lead for 20 seconds, enough to get it bubbling hot.


Personally, I can't escape using big films (well, they're more-or-less unlimited life too), electrolytics or the like, because when you're doing more than a few watts like I'm usually doing, you don't have much choice there.  It's very difficult to make something more than, "implantable" sized shall we say, that uses only chip and film caps while remaining economical.

Tim

[Kjelt]

Also the ceramic caps have less capacity with higher voltage, lets say a 4u7/16V ceramic has only 4u7 till 5V then it drops to 25% of its capacity 
If you need more uF's, lets say a few hundred uF or more, ceramics are no longer price competitive.

[ConKbot]

I'm putting together a capacitive discharge spot welder, with plans to put a FET front end on it later for a sort of EDM front end, 1KW DC supply to power it all, but I'm putting together a few DC-DC converters for cap charging/powering it during EDM/spark erosion use.  Ive been trying to avoid electrolytics on the converters as I'm sure the buck converter input caps would have short, tortured life, (10A ripple @ 500 KHz    )  and yeesh its a pain trying to pack enough capacitance in with ceramics, and not totally break the bank. 

Got my first revision done, and I'm not even going to get the boards made, because I think I want to ditch most the ceramics and go with big chunky film caps for bulk capacitance and a few high ripple ceramics very close to the switching FET.  Plus no voltage coefficient of ceramics to screw stuff up with the films. 

That being said, electros have their place... You just have to use them well, no reason one cant last for 10-20 years if youre not beating the snot out of it with excess ripple current and temperature, and a marginal value which makes the circuit go to crap once it does degrade.