Using Many Ceramic Chip Capacitors Series vs Parallel

"D*****ad from BC"

You need far more than 50uF to filter 3 amps at 120Hz.

..... Phil

Or 10,000uF.

Vr ~= (1/120) * 3A / 1000uF = 25V ... 10,000uF = 2.5V

Best regards, Spehro Pefhany

(that's peak-to-peak ripple, of course)

Best regards, Spehro Pefhany

Ooops...The ripple current is inaccurate.. However, I still have the same problem

Here's a more accurate description if it were a single capacitor.

My app can tolerate 84Vpkpk ripple on the cap.

Capacitor Voltage

___ 170V \\/\\/\\/\\/\\/\\/\\/\\ 84Vpkpk _______________ 0V

The load on the capacitor is an active 800mA constant current load. So the capacitor current looks like this:

Capacitor Current ___ 2.4A _|\\_|\\_|\\_|\\_|\\ ___ -0.8A

That's about 1.2Arms.

I'm trying to figure out the best configuration (series or parallel ) if given a ridiculous amount of ceramic chip capacitors to do this..

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

"D*****ad from BC"

?????????????????????????????????

...... Phil

In general I always prefer parallel. Series connections of caps are dicey. First, you'd need bleeder resistors to make sure there is no DC runaway in a few of them. Then, 47uF will have lots of tolerance so that alone with cause uneven voltage levels. Phssst ... POOF ... *BANG*.

Here's a rewrite for that snippit.

"Any commercially available ceramic chip capacitors that when in series approximately totals to 50uF +/-5uF and has a V rating not much greater than 200V."

Or just change that 50 to say 52.

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

Yup.. I'll need 52 bleeder resistors if I'm using 52 chip capacitors in series... That's extra work and space. Just the chip caps alone are ludicrous. However, series might still be the cheaper solution.

Pricing ~$17.00 for chip caps in series

"D*****ad from BC"

Forgotten to take the usual ones ?

Got no idea how caps in series behave?

You need 50 *2500uF* caps in series to get 50uF !!

IMBECILE !!

...... Phil

"D*****ad from BC"

.... Phil

Oh crap...That's right it's 1/((1/C)* n). Not C/n. for identical capacitors. Well..that means I put the chip capacitors in parallel. Problem solved.

How do I can erase this whole post? :)

D from BC myrealaddress(at)comic(dot)com British Columbia Canada

They'd have to be 4v each to get 200v total.

It might be more plausible and practical to use, say,

--+--C--+--C--+--C--+--C--+---- | | | | | +--C--+--C--+--C--+--C--+ | | | | | +--C--+--C--+--C--+--C--+ | | | | | +--C--+--C--+--C--+--C--+ | | | | | '--R--+--R--+--R--+--R--'

Statistically that cuts down on the potential mismatch between capacitors--the paralleled banks, that is--and uses fewer resistors.

Or you could do what everyone else does and just pick a nice electrolytic.

Cheers, James Arthur

At low temperatures, aluminums crap out. Tantalums explode. Polymer alums only go up to 16 (25?) volts. That leaves film caps, I guess.

John

These are 10uF / 50v, $0.162 :

I figure a 6x30 array would do it, 6 series banks of 30 paralleled caps, costing about $30.

That's 6 in series rather than 4 to cut the voltage across each, otherwise you're just making the world's largest varactor. Might need more than 6, for that matter.

I also chose a 1210-cased part over the 1206 parts, for the same reason.

James Arthur

Uh uh, Y5V...

What capacitance is left at rated voltage?

-- Thanks, Fred.

You snipped this part from my post:

"That's 6 in series rather than 4 to cut the voltage across each, otherwise you're just making the world's largest varactor. Might need more than 6, for that matter.

I also chose a 1210-cased part over the 1206 parts, for the same reason."

All things being equal, bigger cased-parts suffer less capacitance loss with applied voltage, and I seem to remember higher-voltage parts suffer less (we had a thread here in s.e.d. a while ago).

How much capacitance is left? The makers almost never say!

But, never wonder when you can measure:

2) 08056D106MAT2A 8.8uF/0v, 6.6uF/3.6v

4) Taiyo-Yuden JMK105F474ZV-F, 470nF/6.3V, Y5V, 0402 0v 440nF (520nF on first measurement, 440nF after voltage had been applied then removed.) 1v 238nF 2v 134nF 3v 82nF 4v 64nF 6.3v 40nF

Hmmm, Y5V makes *excellent* varactors. I actually have an application for this: a VCO.

But for DC from BC? Probably should use film caps, like John said.

Cheers, James Arthur

They also work as temperature sensors, strain gauges and vibration sensors :)

I have a paper somewhere that uses ceramic caps as the active elements in a high-voltage nonlinear transmission line (shock line) pulse sharpener/booster. That suggests the nonlinearity holds up at high frequencies, at least in some caps. Unfortunately, the most nonlinear caps have the worst TCs. Varicaps have rotten TCs, too.

I wish I had a low-TC varicap just now. Varicap TC changes with bias:C, so any compensation must itself be nonlinear. How tedious.

I suppose you could make an audio power amp using ceramic caps as the gain elements. Or a lamp dimmer, stuff like that.

John

Also as microphones. But you have to strum the twang box real hard ;-)