Bypassing Caps - what is it really about?

Thanks for the answers! I would not have intuitively thought about this because the caps I’ve used (Audio Note, VCap, Jupiter, etc) all seem to have fantastic high frequency sound, even if their ESR measures this way or that. (I kind of wonder what kind of person uses VCap teflon or Deuland silver and still doesn't hear enough high frequencies detail) 

I do have a 350uF industrial filter cap on the preamp plate, and that seems like a target for bypassing to experiment with.

Also, from reading articles from various "tone" guys. they seem to bypass based more on blending the sound of their caps. For example I’ve blended electrolytics on each wave of the rectifier, it seems strange each get a different brand of cap, but it did "split the difference" in sound...one Elna was too noisy, the other Kaisei was too intense.

The 10uF film capacitors we use behave like near ideal capacitors almost to 1MHz. Bypassing that capacitor with a smaller one is not going to have any benefit.

I am not an electronics expert, but I know speakers well. For crossovers, this bypassing of a large capacitor with a small one, from what I have read, often very small, sounds like an urban myth that became real.

The effect of a capacitor in a crossover is related to its impedance at the frequencies you need it to work. Others have mentioned the capacitive element and inherent resistance and inductance. A 0.1uF capacitor is going to have lower inductance than a 10uF capacitor, but the resistance may be high, and at the same frequency, the effective of capacitance is 1/100. That 0.1uF capacitor is not going to fix any perceived issue with the 10uF capacitor. Maybe a 1uF and a 10uF, but then you have a much different capacitor and perhaps a 10% change in crossover frequency. Maybe if you are using first order crossovers, bypassing an electrolytic with a film capacitor if the film capacitor is a significant portion of the electrolytics values could be beneficial in a shunt location, by preventing frequencies at say 10x the crossover frequency. I am clutching at falling straws at this point.

Maybe in an amplifier or pre-amp bypassing an electrolytic audio coupling capacitor with a small film capacitor will have a benefit over the full frequency range, but I will leave that answer to the EEs.

. For crossovers, this bypassing of a large capacitor with a small one, from what I have read, often very small, sounds like an urban myth that became real.

That's your problem right there.  Research without experience.  $40 in parts and a weekend is worth more than all of the Internet.

See that’s your problem right there. You confuse $40 in parts and a weekend with experience. I have been designing speakers for professional applications for the better part of 2 decades. When you lack experience, you may be able to convince yourself with $40 in parts and a weekend that bypassing a capacitor that is ideal up to nearly a MHz provides some benefit other than just changing the total value. I don’t have such a luxury.

Pick a large respectable high end speaker manufacturer. Magico, B&W (high end), Wilson. Do they bypass their large capacitors with much smaller ones? No. Could they afford it? At those margins, of course, small capacitors are inexpensive. Why don’t they? It will have no benefit. Do you honestly believe these companies don’t know any better?

My comment is specifically about crossovers. I will leave power supplies to those that know better, and the same for capacitors passing audio signals in amplifiers and other electronics.

That’s your problem right there. Research without experience. $40 in parts and a weekend is worth more than all of the Internet.

I can see a stuffy sounding vintage cap with poor high frequency transparency being bypassed by one that does to create some layering (?).

But that could create some weird situations - imagine caps which image poorly with a very mono sound, being bypassed in the high freq with a high fidelity cap. So the high frequencies image 3d dimensionally on top of an old low fi midrange? Or is that effect the goal? It could be interesting....??