Efficient Use of Decoupling Capacitors

Thread Starter


Joined Feb 21, 2018
Hello all!

I could not figure out yet why it is recommended to use different values of capacitors for decoupling in any (analogue/digital/mixed-signal) system; be it the decoupling of power pins, or the reference pins or any other pin.

For suppressing transients, capacitors are needed, bypassing that pin to ground. Why is it recommended to use different values of capacitors, some small some large, for the decoupling? Isn't it so that the higher the decoupling capacitance, the more powerfully it suppresses the transients, along with RF and other noise? Wouldn't it be better that instead of instead of using different values of decoupling capacitors, we use all the capacitors of the same (highest) value, which would have a higher equivalent capacitance on the same amount of (precious) board space, which would ultimately lead to better transient performance per unit of board area?

It would be really appreciable if someone could explain why these different values are used for decoupling, and what should be these values in a typical system?

Thanks in advance.


Joined Feb 20, 2016
Some caps work best at various frequencies so a range of caps will bypass a wider range of frequencies.
For instance, electros, because of their construction, have pretty poor high frequency response so an electro with a parallel ceramic that has a high frequency response is a better option. I generally use Electrolytics in the power supply and then Tantalum caps and ceramics in the works.


Joined Feb 8, 2018
Parts off Table 2 in the "article" link at #5 are pretty much nonsense.

Ceramic, plastic film and aluminum electrolytics are all listed as having "medium" leakage, while tanatalum and Oscon are listed as having "low" leakage. Plastic film capacitor leakage is typically orders of magnitude less than that of aluminum electrolytics and tantalum. The same generally applies to ceramics.

"Plastic Film" is parenthetically elaborated as "polyethylene polystyrene" beside a photo of a surface mount part. Neither of those dielectrics would survive reflow soldering.. Polyethylene is not used as a capacitor dielectric. All manufacturers of polystyrene film ceased production before that ap note was written and in consequence polystyrene capacitors are no longer made (unfortunate, they were excellent in some ways).

Tantalum are described as "nonlinear (bad for audio)" but ceramic get no such comment. Low dielectric constant (K) ceramics (C0G types) are generally very good. High K ceramics are absolutely dreadful, grossly non-linear and typically have very high negative voltage coefficent of capacitance. The voltage coefficient of some ceramics is so bad that you can be down to 20% or less of nominal capacitance if you operate them at anywhere near their rated DC voltage.


Joined Feb 8, 2018
Unfortunately none of the figures display in the article.

Regards, Dana.
I don't know whether EDN ever got its act together and figured out how to manage figures in articles on its website. I used to subscribe to the print version, then went to the on-line version and promptly quite reading it at all because of its incompetence. It sort of reminded me of the joke about the discovery of some old scientific book in German - it turned out to be unintelligible because it had been published as two volumes, they only found volume 1 and all the verbs were in volume 2.