# Sprinkle switching circuits liberally with small capacitors?

I have now read enough applied circuit diagrams that I believe I'm missing something that is obvious to those skilled in the art: Where an abstract diagram has a few components, a complete wiring diagram of the same circuit is absolutely littered with capacitors in the microfarad range.

(To do: Find and insert some example images here.)

I can sort of understand how these capacitors would smooth out voltage drops, (and I guess there are usually a few diodes to deal with voltage spikes)?

But how do engineers come up with these applied circuits? Are there unwritten rules they just know, like, "Yep, gotta put a microfarad on this side of a diode, that side of a transistor, and three of them in parallel by this sort of switch..."? Or do they wire up a prototype, start scoping every junction, and add them based on actual signal noise they observe? Or is more prophylactic, e.g., "I wouldn't want a lot of voltage noise here, so why not add some capacitance just to be sure?"

• Does What is a decoupling capacitor and how do I know if I need one? answer your question? Commented May 14, 2016 at 21:08
• @AndrewMorton - Wow, that is very illuminating. It might be the whole answer. I see, "Put bypass caps everywhere if space and cost allow," "on every power input" (though why power supplies don't provide adequate capacitance isn't addressed), and "decoupling caps as specified by ICs." Is that all accurate and complete? Commented May 14, 2016 at 21:18
• The power supply cannot compensate for the induction of the current path between it and the component. The induction acts to slow down the recovery of the voltage to its required value. A capacitor immediately next to the component has less inductance in the path and so can reduce the change in voltage at a higher rate. (As far as I understand the operation.) Commented May 14, 2016 at 21:33
• Capacitors at the supply are not adequate because there is trace inductance and resistance between the supply and the load. Commented May 14, 2016 at 21:37
• Yes, in most cases it's not even a thumb rules, just a habit. Actually, not many people bother with the question "why". By the way, same happens even where calculation may be used, like in switching supplies. Then you ask the designer, why the capacitor is so big, can't he take a cheaper one, amd he says "no idea, i always use 100uF". We all hope that IC designers recommend decoupling because they know something we don't. But frankly, i lost my trust in them too. Their recommendations for hard components like FPGA are barely realistic, so it always looks they just escape responsibility.
– user76844
Commented May 14, 2016 at 21:47