Specifying SMD footprints and parts

[R.G.]

OK, got it - thanks.

My quick educate-myself on this turned up some things I missed the last time I dug deep into ceramics a decade or so ago. Mostly I'm writing this out to set it firmly in my own head.

MLCCs in particular seem to have issues, and the smaller the package the worse it gets, according to a couple of makers. That is, 0603 is worse than 0805 and that's worse than 1206 or 1210; apparently this is due to the thinner and thinner layers of insulator as the caps get smaller.

And it's a characteristic of all "class 2" ceramic cap materials, not just X7R. Pretty much all of the high-C*V ceramic caps do it, as it's a characteristic of the high-permitivity insulator, largely BaTiO3. How much the variation is depends on whatever secret sauce the maker uses, but if I'm reading correctly, all the high-CV ceramics based on BaTiO3 do it. C0G and NP0, and other temperature-stable types don't do it to a significant degree.

A couple of makers/users affirmed my immediate suspicion - just like with voltage coefficient of resistance causing distortion in carbon comp resistors, voltage coefficient of capacitance in BaTiO3 multilayer ceramic caps causes distortion in signals passed through them.

To avoid this mine field:
1. Don't use MLCCs other than the temperature stable Class 1 materials for passing audio unless you just can't avoid it.
2. Use film. If you can't afford or find SMD film types (and there are good reasons for that) use through hole films.
3. If you can't avoid it, use the biggest packages and highest voltage rating parts you can shoehorn in, to minimize the effect. Highest voltage rating parts used at lower DC voltages means you're minimizing the relative shift as a percentage.
4. It's not just SMD ceramics. Low-voltage-rated, high C*V through hole parts do the same, as this is a property of the material.

But now I'm thinking - D@#$^!! a voltage variable cap that varies a whole lot. There ought to be an effects use for that. Finding a high-change type might make for a voltage tunable something.

Again, thanks. I appreciate education.

[merlinb]

By what I've understood, X7R capacitors exhibit strong capacitance loss when subjected to DC

I didn't know that, thanks!
https://www.vishay.com/docs/45263/timedepcapdrix7rmlccexptoconstdcbiasvolt.pdf

[fryingpan]

So, they should work in a tonestack I suppose? Since there is no DC, I mean... (or I just go with C0G or film caps, since they're small anyway...).

[R.G.]

So, they should work in a tonestack I suppose? Since there is no DC, I mean... (or I just go with C0G or film caps, since they're small anyway...).

Having DC wiped off by other caps would eliminate the voltage coefficient of capacitance. It would not change the drift with temperature. Digikey has a good page on temperature drift in ceramics: https://forum.digikey.com/t/understanding-ceramic-capacitor-temp-coefficients/727
The temperature drift over the temperature range is noted on the type number. X7R, for instance, has a drift indicator of "R", which the industry says is a +/- 15% drift over its temp range.  The X says low temp of -55 C, the 7 says high temp of 125C, so it drifts +/- 15% over -55 to +125.
It's highly unlikely that a pedal PCB would be used below 0C or above 40C, so the temp drift would be lower; but still there.
So yeah - they would work in a tone stack, but would have some temperature drift. It's probably minor, as 40C is small part of 180C( 125C - (-55C)), but it's still there, and unless the maker specifies the temp drift curve or direction, you don't know which way it drifts. But in a tone stack, probably good enough.

You're actually describing a crossover, I believe. There will be some overlap in the middle, with phase shifts. See Elliot Sound Products' article on crossovers.