High frequency ripple and decoupling capacitors

DIY STUFF
1

Whenever I search for anything about decoupling and ripple, there seem to be two kinds of results:

  1. Incredibly direct and simple answers like “just stick a big electrolytic on the power rails and a ceramic on each IC”. Very easy to follow, but doesn’t really help me to advance my understanding beyond that

  2. Incredibly long and detailed discussions of ESR, ESL and resonant frequencies that seem to be written for professional engineers. I have a very loose idea of what those things are, but I can’t just intuitively apply them to my circuits, nor can I do the maths

The other thing is that, much of this discussion also seems to not be about audio but radio or computing or something.

Anyway, here’s what I’m stuck on

  1. Is ripple above a certain frequency ever just too far from the audio band to really matter for synth diy?

  2. Do we even have to worry about self resonant frequencies in synth diy? Or are they too far away to be important? Does this differ by material?

  3. Will a larger value ceramic across the pins always do more to clean up noise and ripple than a smaller one? Or is there some kind of practical trade off?

  4. Are higher switching frequencies on a power supply generally better because the amplitude of the ripple is smaller and the decoupling caps can soak it up better…

1 Like
2
6 Likes
3

also: https://www.allaboutcircuits.com/technical-articles/clean-power-for-every-ic-part-1-understanding-bypass-capacitors/

2 Likes
4

Thanks. This is a really good video but it doesn’t really say anything about ripple or audio

5

I take that to mean bypass capacitors have little to do with ripple or audio.

6

No, it does not matter, if you do it correctly.

That depends mostly on the ICs you are using, but the oscillation can be prevented by adding capacitors to feedback loops in op-amps, for example.

Yes, but there are trade-offs in efficiency, see above.

4 Likes
7

No, it means that it doesn’t touch on the specific things that I was asking.

8

This is awesome! Thanks dude.

1 Like
9

I mean I think it doesn’t touch on the specific things that you were asking because bypass capacitors are not especially important to the specific things you were asking about (ripple and audio distortion).

1 Like
10

Geez louise! Why would you have me watch a half hour video then?

11

Sorry, I thought you wanted to learn about bypass capacitors.

3 Likes
12

Ok, so it’s looking like I was very confused on a few points.

  1. I think, because I was searching for how to clean up ripple and noise at a specific switching frequency, Google saw the phrase “switching frequency” and directed me to discussions around circuits in digital computing. Those guys were really into their ceramics because they’re kicking noise back into the power rails at 100MHz and more. These just aren’t going to make a dent in the switching frequency at tens or hundreds of kHz.

  2. I was right about how capacitance and ESL interact to create a self resonant frequency, but I totally got wrong what to do with it. I’ve normally understood a resonance to be a range of frequencies that get louder, whereas here I want to make them go away. But actually, the self resonant frequency of a decoupling capacitor is where it is at its most useful, where the impedance is lowest, sending more of the AC content to ground. So I want to match that frequency as close to the ripple as is practical.

  3. I’m still a bit confused as to how to actually select a cap for resonant frequency. The capacitance is clearly labeled but the ESL isn’t, and maybe it depends on the board too. It sounds like 10nH is a very vague ballpark for a medium sized electrolytic filter cap?

  4. I probably need an oscilloscope if I want to get deeper into this

Thanks for your patience

13

Yes, matching the resonant frequency of the filter network to the switching frequency makes fiktering most efficient.

Unfortunately, the self inductance of an electrolytic is not easy to control or specify, because it depends on a lot of factors (eg. Error tolerance, pcb layout, even the way it’s mounted)

You can however, put multiple capacitors in parallel to lower their effective inductance. This will of course raise the capacitance, but using a lot of smaller caps in parallel can give you the same capacitance but at a waaay lower inductance. It’s why some devices have big banks of electrolytics, instead of one large reservoir cap. Bypassing with a 100nf or so ceramic is good practice.

Hope this helps a bit, i’m no expert on this subject either.

1 Like