Caps caps caps...

Fred Bartoli a écrit :

Should have said : No wear out mechanism and _ultra low FIT_ : product life = 20 years,

Polymer aluminums, e.g.

You can get 100 uF MLCCs, e.g.

A couple of those in parallel might get you there.

Cheers

Phil Hobbs

On a sunny day (Sun, 16 Dec 2012 18:21:08 +0100) it happened Fred Bartoli wrote in :

Sounds like a ceramic cap.

That's a lot of uF for a uW supply though. 100uA = 1V/S from a 100uF cap., which is only 1mV ripple if you switch at 1kHz.

Frequent, short pulses into a smaller filter cap might be worth considering.

I've designed discrete microwatt boost converters. That's probably easier cap-wise, 0.5 x CV**2 and all.

Jan Panteltje a écrit :

LOL!

That one seems the right model for you:

At the qty level I guess it's a no-no, but I'll have to check what the customer say...

Thanks for the hint. I didn't look because I thought the 100uF were only of the Y5V-Z5U ilk...

I'll have to check what they're worth though. On a previous design I've found some 1uF/10V 0402 X7R loose their capacitance with voltage rather quickly, almost like the Y5V :-(

A bit nervous with chip cracks too... (but the board is smallish)

Maybe you can rout out a C-shaped strain relief around it, as with voltage references.

Cheers

Phil Hobbs

Oh, I'm way higher in power, at the mW level :-) Switching at 100kHz, but it's not a bypass cap, rather a reservoir cap here. Sorry can't say too much...

Indeed, but I'm bucking it down later and efficiency is a real big concern...

Excellent idea! Thanks.

On a sunny day (Sun, 16 Dec 2012 19:53:37 +0100) it happened Fred Bartoli wrote in :

Yes, I have considred ordering one of those, but I need a very powerful motor for liftoff...

e)

he

...

uF

mW? Monster!

Naturally. Who uses 330uF bypass caps?

Oh yeah, 'twas a current source, right? So that's current ripple.

I used discrete components expressly for high efficiency. (All the commercial controller chips would've drawn several times the intended load power--no one made efficient controllers for a 20uW supply.)

I'd think a good CMOS totempole would be an awfully good head-start on a synchronous mW buck, but I haven't actually tried.

e)

he

her

The loss of C with V goes with volume--bigger caps used at a fraction of rated voltage are far less affected. Still, I'm surprised to hear that of X7R--I've not seen that. But then I've not personally seen a

Sweet.

Oh no, this is an application where there are electrons and those can go berserk. So you need one of these:

But be careful. This can also cause a cantilever vibration at the end of a fall, and hair-crack the ceramic.

If you need a lot of reservoir cap, and you need to step down, maybe you could consider an all-capacitor charge-pump type converter, with some CMOS analog switches and a bunch of ceramic caps. They can be very efficient at low power levels. It would be interesting to design one that put all the caps in parallel after they were pumped up, or rather pumped down.

le)

the

,...

could

og

ower

rallel

Lots of switches, lots of switches to drive. But that might work, since he's really making a current source.

As an alternate, he might be able to dispense current 'packets' via a cap[*], and smooth the doses with an inductor. That would be very efficient.

If it's a reservoir you're looking for, you might try storing at a higher voltage, and only use the part and energy stored, when needed. The storage could be performed over the long term, not drawing significantly on power consumption, and not effecting efficiency, once charged.

Switch into buck reg when needed.....

RL

On a sunny day (Sun, 16 Dec 2012 13:36:07 -0800) it happened Joerg wrote in :

wrote in :

Yes I did read that one, the modern NSA brain control transmitters are in underground spaces, those beam upward, and then the tinfoil cap focusses the beam IN the brain. Have you felt pro US govermnment lately? Make sure there is no tin foil in your cap! Some Shampoos also reflect..

Not to answer your question, but I do have some thoughts....... (at least on the days when I'm not in senile mode).

My thought today is to change all circuit boards. In the old days, tubes had sockets because they failed often. These says the caps are so lousy, that they need to be replaced often. Therefore, all caps should have sockets, rather than being soldered in. Then the corner drug store should have a cap tester where you take your caps to be tested, and buy new replacements for those that check out bad.