Ceramic chip peak current capacity

Dunno if it'll hold together, but I'm guessing that if it does crack it's from uneven heating, not piezo stresses.

No idea, but I doubt the current is 'huge' - will be limited by the ESR of the cap, plus any external resistance in the charge/discharge path.

Piezo effects are voltage related, not current related, so again the stress, if any, is limited by the dV/dT.

uge

of the cap, plus any external resistance in the charge/discharge path.

ess, if any, is limited by the dV/dT.

Strictly speaking, it's strain - rather than stress - which breaks crystals .

A high voltage can impose a lot of stress, but it it's transient, the mass of the crystal won't get the chance to move far enough under that stress to develop enough stain to break.

dV/dt won't come into it - except by feeding a lot of current into or out o f the capacitor, and warming it up, producing thermal stress and eventual m echanical strain. From a piezo-electric point of view it's the length of ti me that the high voltage is sustained that matters, so you'd be more intere sted in the integral of the high voltage impulse than it's derivative.

The dielectric layer will probably have a resonant frequency - set by it's thickness and the speed of sound in the layer - which might come into it, b ut my guess would be that the Q of any mechanical resonance wouldn't be tha t high, mainly because the material won't be all that uniform across the ar ea of the dielectric layer(s).

What's the cap size? How huge the currents?

I've never seen a ceramic cap that was bothered by being hard discharged, like with a screwdriver.

Say 0.1uF, 100V, 1206 size (so it still has useful capacitance up there).

That's the general idea, actually: how much reliability can I predict based on "shorting" type surge conditions -- 1kA/us sort of rates I think. Or in terms of surge, how long will it last until the MOV (I didn't mention there's a MOV in the circuit also) kicks in. Of course the MOV sets maximum terminal voltage, which controls breakdown, so no worries there.

Speaking of, looked at an article that claims most 250V ceramics fail in the

Tim

Yes, MLCC crack or arc with extremely high dV/dt. Some give vague warnings about being designed for low ripple voltages while others have have specifics.

Tim Williams prodded the keyboard

It gets bloody hot and I've seen them simply vanish ! Either its fallen off or disintegrated.

CW I understand, but from pulses?

Tim

Not low duty cycle pulses. Ceramic caps can unsolder themselves in high power RF circuits. There are super low ESR caps made for high power RF, porcelain and ceramic.

It's not clear: if the ceramic/metal MLCC structure has voltage-caused reshaping of the ceramic, that might deform the metal (which then can accumulate internal strains). When/if the applied voltage on the ceramic changes, the NEW (low-voltage) shape then causes more deformation of the metal. You might see a MLCC (multilayer chip capacitor) change value over time, and return to original state when/if overheated (and the tin annealed).

As I understand it, there might be some acoustic resonances, with ceramic layers at antinode and metal layers at nodes; the 'self-resonance' is something inductor manufacturers often specify, but I don't see that much in chip capacitors. Probably the piezoelectric materials used are NOT oriented crystals (haven't been polled) during manufacture, though stress in the field might create orientation. Conventional wisdom is to sometimes use several different bypass capacitors. I've thought that was old lore, but there may be some new validity.

Tim Williams prodded the keyboard

Half a kilowatt at 432/434 Mhz. So yes CW.

John Larkin prodded the keyboard

Yes high power RF. I had a pair of "American Ceramics" caps in an output filter simply disapear. Just the tape leads remained. I've still got the pair of replacements that they sent me somewhere. I just assume that like the others they got really hot and fell off.

They do change value over time, which is partially or fully restored on annealing above the Curie temp. But as far as I know, it's a crystal thing, and develops regardless of bias(?).

Anything to do with strain or deformation would have to be fractional percent change, and thus fractional percent change in value, no? Or at best, twice the change (increase in area / decrease in thickness), but still more or less fractional.

Such a change would be observable in C0Gs, but they aren't piezoelectric...

Presumably you could use one as a ceramic resonator, but I guess they're too mechanically lossy to show any evidence of it in the impedance curve. Probably, because the structure is extremely strongly coupled between layers, it either cancels out (they are alternating, after all), or causes so much degeneracy in the resonances that they all turn to mush, or because the stack is so dense (and end/side capped), the mechanical impedance is too low to matter (even despite the mechanical advantage of many layers).

Quartz crystals aren't much different in size, and place in the MHz; you'd expect to see an impedance dip in the same region if this were important. While you usually *do*, that's easily explained by ESL, not by mechanical forces...

And, it follows: if it's mechanically well damped, it probably isn't going to care about a transient, because if nothing else, you know it's not going to resonate and exhibit any more strain than the static voltage would produce.

Tim

Yes, of course. What's unclear, though, is whether the strain is more than the elastic limits (of the metal, usually tin). I could be chasing a completely solved issue...