Voltage variable-capacitance capacitors

For some time now I've been using caps with voltage ratings 3 to 5x higher than what they'll see. And I've been avoiding small high-cap parts. No wonder I didn't discover the scandal on my own.

--
 Thanks, 
    - Win
Reply to
Winfield Hill
Loading thread data ...

Hey, that's a 1959 paper. Sure it's of any use now?

--
 Thanks, 
    - Win
Reply to
Winfield Hill

DC

all

oss

t have

o it.

While you can find data published on the voltage dependency of ceramic caps , this info is not always published by any given manufacturer. Then while some put this in their data sheets, others publish is on their web site in special documents separately. I kinda figure everything important will be in the data sheet, but obviously that is not the case.

The voltage dependency not only varies with the ceramic type, it varies wit h the voltage rating, with some having significant impact on the capacitanc e at different percentages of the rated voltage. Of course, this entirely depends on manufacturer as well. I suspect there is an element of the moon phase, season and election year as well. That's why they only publish "ty pical" curves and never max/min specs that I've seen. So caveat emptor.

--
  Rick C. 

  - Get a 1,000 miles of free Supercharging 
 Click to see the full signature
Reply to
gnuarm.deletethisbit

You could put a power resistor next to the capacitor on the PCB, and re-anneal the capacitor from time to time!

Reply to
Chris Jones

It'd look like the old Analog Devices 310 and 311, one supposes.

Reply to
whit3rd

Or parametric *oscillator*? Nope, the response is far too symmetrical, +- wise.

Ah, from your IEEE link it appears that this resembles the "half- frequency soliton effect." Whack your wineglass, and the Faraday surface waves on the wine will be at half the frequency of the moving glass "piston." Or, listen during a dental cleaning, and the ultrasound signal will become audible, as similar nonlinear- ities create 1/2 F and 1/4 F harmonics rather than higher har- monics.

Or, pump a swinging pendulum by pulling upwards on the string (yanking twice per cycle, so your 2F drive will amplify the 1F pendulum frequency.)

Huh, thinking on *rubbing* wet wine glasses ...I wonder if adding a diode could create an oscillator, if it was part of a tank circuit using the right value of 0603 ceramic cap? Where diode losses were smaller than capacitor gain during every other cycle? Must think more on this while less awake!

Other thing: these bad capacitors probably are exhibiting ferroelectric saturation, analogous to ferromagnetic saturation.

In that case the scandal is just like having some ferrite supply filters rated for one amp, but finding that the inductance falls by 10X when actually run at 1A! (And it only happens for the physically smallest one-amp inductors, rather than all of them. And the sellers never mention that your 1-amp rated filters must never actually be run at more than 50mA!)

--
((((((((((((((((((((((( (  (    (o)    )  ) ))))))))))))))))))))))) 
William J. Beaty              Research Engineer 
 Click to see the full signature
Reply to
Bill Beaty

Oh, one more also: the bad capacitors show hysterisis. Maybe a couple of percent, but I didn't measure it. After they've once been exposed to full WVDC, their capacitance is less at zero volts. And, if you reverse the polarity and slowly raise the voltage again, first the capacitance INCREASES briefly, before dropping as before! Very expected, since the previous "high" voltage had semi-permanently "poled" the PZT ceramic, at least a bit.

Capacitors as permanent memory elements. Giant electrostatic core-memory bits. But how did the capacitors start out without an initial stored effect? Thermal annealing, as mentioned?

So, build a nonvolitile flip-flop using a bad capacitor as a memory element, then either write a one or zero using pos and neg pulses, or if no neg supply, instead use your hot resistor to erased the stored poling. Stored "Electrification?" "Electri- tization?"

If a reel full of bad capacitors are identical enough, then one could build an array, add a lens, and take thermal image snapshots.

Hmmm. If the year was ~1820, I could get everyone to describe the dielectric poling process with the term "electri-fi-kating," and maybe that term would take off and become common. Your capacitors stop working? Way too electri-ficated! (Analogous to "magnetized." Or perhaps convince everyone to describe a PM bar as having been "magnet-tificated.) Far too late to have it work now though.

--
((((((((((((((((((((((( (  (    (o)    )  ) ))))))))))))))))))))))) 
William J. Beaty              Research Engineer 
 Click to see the full signature
Reply to
Bill Beaty

(I would only use cog/npo ceramics in a filter.) But it does go to show that ~energy storage goes as the volume. (hmm, not sure that's true. Well at least that's ~true for film caps.

Energy density is proportional to k, so you can figure it out pretty easily for type 2 dielectrics assuming a practical threshold for saturation (say

30%).

You can still keep going to higher voltages -- energy keeps going up, even if not as quickly as you would've liked.

As it happens, saturation tends to follow a curve such that the energy is approximately linear with voltage, after a curved toe-in (from the linear range).

It turns out, the breakdown of type 2 dielectric is significantly worse, too. So despite C0G being so much lower k, you can stack that many more, thinner, layers before it breaks down, mostly recoup the value, and better the energy density.

The real downside is, those layers are, I don't know, probably nanometers, if you wanted say a "6.3V" cap. You can't make that with a ceramic tape process. So, C0G is only economical for energy storage at high voltages, over 200V or so. (Meanwhile, 100 and 250V X7Rs are just _atrocious_ in C(V), so you need so much more than advertised, to get what you need.)

It may follow that, despite its fairly low k, diamond makes capacitors of the highest energy density. I haven't looked at the properties though.

One very different thing you can do: polarize the dielectric, making an electret. This shifts the voltage where the capacitance peak is placed. And, because maximum capacitance is seen at real voltage, it absorbs massive energy! (The ~2015 Google Little Box inverter contest winner used this as the solution for energy storage to carry supply ripple.) Downside? Type 2 dielectrics anneal at soldering temperatures. So... you can either polarize it in situ (during or after soldering) -- good luck with that; or, you very carefully hand-solder a pre-polarized part.

They're also not cheap, but that's to be expected from niche components.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

Right, pumping a swing by standing on it and going at twice the freq. can be dangerous.

So maybe a parametric oscillator first. Something like this?

+-----------+--- Vout | | V_bias.DC---R_b---+ ___ ) | ___ C1 )L1 | | | | ) +-----| |----+-----+ | V_in.AC | | | | ___ | both input ___ C1 | Voltage sources | | tied to Gnd +-----------+ not shown | Gnd

Where C1's are big voltage coef caps.

I want small high voltage caps for a big voltage coef, digikey.. George H.

Reply to
George Herold

Yeah I don't know much about the k coef. I am replacing all film caps up to about 0.1 uF with cog ceramics, cheaper and better, mostly.

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.