Hacker News Reader

Comment by Workaccount2 | Hacker News Reader

My approach to decoupling my circuits is to take out my 100nF salt shaker and to sprinkle liberally over the board.

When an 0805 is 2 cents in bulk, I can't be bothered to do it any other way.

0805 ! Get thee some 0201s and a good microscope.... ;-) If you are working at DC / audio, sure, 0805 probably fine, tho. Truth be told, I generally use leaded parts because they're easier to work with. If you understand what is going on (a leaded Cap is actually a cap in series with some noticeable inductance), and if the self-resonant freq is 'high enough' --- it's fine.

Yeah, and this is a case where a well cultivated sense of "good enough" is actually quite important because optimizing decoupling capacitors for high capacity, small size, and low price is a recipe for picking an exotic ceramic with a wacky tempco and then having a device fail intermittently when hot or cold:

https://www.allaboutcircuits.com/technical-articles/x7r-x5r-...

I find that 0402s are plenty fine, thanks, and I can still see them with the naked eye to grab them with tweezers.

In my experience, 0402s are the breakpoint on the curve.

0603s are large enough that there are a couple of cases where they're kinda marginal due to size (inductance) and the size gain isn't quite enough for things like QFN packages. 0402s almost never have the issue. 0201s aren't enough better that they are worth the extra grief to deal with unless I'm on a really constrained board, and I'm only letting an assembler do the board (nothing by hand from me).

For most of the embedded designs today the power planes are the bulk of low-ESR/ESL decoupling (even on 4 layer boards with powerplanes in the middle layers), it is a cap of ridiculously small capacity, but with essentially zero ESL/ESR.

That's how you get a capacitor shortage :)

Reply on news.ycombinator.com