In the datasheet of TB6612FNG DC motor controller, Typical Application Diagram section.

Both DC sources are smoothed by caps. But I don't understand why there are two caps in each position.

The polarized cap is 10uF and the non-polarized cap is only 0.1uf. Which is a hundred time smaller than the other, so I guess it is not for making a specific capacitance.

Is the small cap for shielding off back emf spike? But wouldn't the electrolytic cap break in such case?

application circuit


The two capacitors complement each others’ strengths and weaknesses.

Large electrolytic capacitors work better at lower frequency than they do at high. This is due to their internal resistance, referred to as Equivalent Series Resistance, or ESR. They also have relatively high series inductance, or ESL. Both of these limit the electrolytic capacitors’ ability to filter high frequency.

Ceramic caps don’t have the high storage capacity of electrolytic. They also have an issue called bias effect, where their value diminishes as the working voltage increases. But they do have something that electrolytics don’t: very low ESR and ESL, making them much more effective at high frequency.

Used together the two capacitor types can cover a wider range of frequencies, and thus, do a better job of filtering noise.

However... Anti-Resonance Is A Thing. Mix Values With Care

Different value capacitors can interact in unexpected ways that can lead to peaks and valleys in the overall response, in some cases leading to worse performance than if you’d only stuck with one value. This phenomenon is called anti-resonance, and it needs to be considered when mixing cap values to improve frequency response.

In the diagram shown, the use of the two types is kind of a ‘Hail Mary’ - the designer has placed them without really analyzing about how the caps interact with each other.

As it so happens, 10uF and 0.1uF together have an anti-resonance peak at about 3.6MHz of over 3 ohms (per KSIM below.) If you had, say a switching regulator that worked at 600kHz, its sixth harmonic would be right in the middle of that peak and thus not be well-filtered. Not what you want.

To overcome this problem I would increase the ceramic to 1uF or even 2.2uF, or maybe chuck the whole thing and start over using an all MLCC design.

Some Online Resources For Capacitor Selection

Kemet has an online tool called KSIM that allows you to experiment with different values and combinations of values to identify these issues. Link: http://ksim.kemet.com/

Murata has a tool called SimSurfing that allows you to view different capacitance characteristics. Unfortunately they don't support combining impedances like KSIM, but it's still useful. Link: https://www.murata.com/en-us/tool/simsurfing

Murata also has this very useful appnote about bypassing caps: https://www.murata.com/~/media/webrenewal/support/library/catalog/products/emc/emifil/c39e.ashx

|improve this answer|||||
  • \$\begingroup\$ Thanks, if I have some 10uF ceramic cap, do I still need the 0.1uF one? \$\endgroup\$ – somebody4 Aug 13 '19 at 15:51
  • \$\begingroup\$ If you choose an electrolytic with a low ESR/ESL, probably not. But see more about anti-resonance above. \$\endgroup\$ – hacktastical Aug 13 '19 at 16:03

Okay, I find this from quora:

answer from Rafal Muszynski

Capacitors are connected on input and output side to secure stable (oscillation-free) operation and improve step response (behavior during sudden changes of the load or input voltage). Most of the times the required capacity is in order of 1uF-100uF and such values can only be practicaly achieved with electrolythic capacitors (the one you quite correctly call “polarized”). However, this kind of capacitors has rather large equivalent series resistance, which means they cannot cope with fast transients. Therefore they are connected (in parallel) with other type of caps, such as ceramic, most often 10nF-100nF, in order to improve fast transient response (block high frequencies). These small caps are non-polarized.

Typically, these caps are supposed to be connected (soldered) physicaly as close as possible to the regulator IC input and output terminals, and the ground connected in a star, that is single common point for all caps and regulator ground (or reference) terminal. It’s a common practice to even draw such a star grounding and a direct close connection of the caps on the schematic diagram just to highlight the importance of such arrangement for the best transient performance and stability of the IC.


|improve this answer|||||
  • \$\begingroup\$ Jep, it explains it pretty well. It's generally a good idea to spend some time (more than 30 min) doing your own research before asking a question on the forum. especially if you don't want people getting annoyed with you ;) \$\endgroup\$ – Vinzent Aug 13 '19 at 11:49
  • 2
    \$\begingroup\$ This answer is only a link to some other site, that's not appreciated much on this site, this answer might get removed because of that. You can answer your own question but we then do expect a "proper" answer, not just a link. \$\endgroup\$ – Bimpelrekkie Aug 13 '19 at 11:52
  • \$\begingroup\$ @Bimpelrekkie Thanks, edited \$\endgroup\$ – somebody4 Aug 13 '19 at 12:21
  • \$\begingroup\$ @Vinzent, thanks, maybe I missed that on early search. Most results are just comparing polarized & non-polarized caps. \$\endgroup\$ – somebody4 Aug 13 '19 at 12:23
  • 1
    \$\begingroup\$ You mean copy-pasted. I would not make a habit out of this (asking question, copy-pasting answer from elsewhere) if I were you. \$\endgroup\$ – Bimpelrekkie Aug 13 '19 at 12:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.