1
\$\begingroup\$

I want to choose decoupling capacitors for some microcontroller peripherals, which in their data sheets only specify some capacitans. Since you have quite a choice of passive components and the data sheets not always specify which type to use (ceramic, some electrolytic, aluminum or tantalum) I feel insecure.

The data sheet says to use a 1uF and a 0.1uF cap. I feel sure that the 0.1uF one could be a ceramic. But could the 1uF be one to? I want to operate at 3.3V will a 1uF ceramic cap lose to much capacity to make the recommended 1uF. And is a 0402 cap as good as a 0603 for the same capacity?

Thank you for any help!

\$\endgroup\$
0

3 Answers 3

4
\$\begingroup\$

Decoupling capacitors are usually of the ceramic kind, sometimes combined with an electrolytic capacitor. Ceramic capacitors are widely used for most applications, the electrolytic ones are useful if you need higher capacitance values in a relatively small package. In most cases ceramic capacitors will fit your need just fine

I usually use the biggest footprint I can fit on to the board (0805-1206) unless I don't have the space. Capacitors with bigger footprints usually have a higher max voltage rating without a lot of added costs. Also they tend to have improved temperature stability. So the 0402 cap will probably have a reduced max rating, and slightly reduced temperature stability in comparison with the 0603. But if you are not planning on using the product in extreme environments and the Vmax of the 0402 fits your needs (including some extra tolerance as buffer) it should be fine.

Another thing to watch out for is the type of capacitor X5R, X7R and C0G/NP0 are less susceptible to a broader temperature range as Y5V.

This is I think pretty much the basics, other things that can be taken into account are the ESR/Q needed for example. Which can be of importance when designing filters of switching power supplies.

ADDITIONAL EDIT:

I just briefly want to add that if you are working with high frequency, where the performance is determined mostly on the inductance. That a 1206 SMT capacitor will have approximately twice the inductance of a 0603 capacitor. Therefor for high frequencies it is recommended to use the smallest practical size for the application. And find the largest capacitor value available in this size. (A smaller value will have no improvement in high frequency performance but will degrade low frequency effectiveness)

\$\endgroup\$
0
4
\$\begingroup\$

I think you're over thinking this.

The datasheet says that you need to use 1 uF in parallel with 0.1 uF and that is because the combination of those will give you proper decoupling over a wide (enough) frequency range. See this video from the EEVBlog.

Since the 0.1 uF will (should) indeed be a ceramic type the type of the 1 uF does not matter that much. You can use the type which you prefer. My guess is that the 1 uF capacitor will also be a ceramic type as there are quite common.

I want to operate at 3.3V will a 1uF ceramic cap lose to much capacity to make the recommended 1uF

Not sure I understand you here, since you want to use the decoupling at 3.3 V you need to make sure the capacitors can handle that. Any decent 1 uF capacitor will have a value of 1 uF at 3.3 V. Even if the value of the capacitor deviates somewhat over DC voltage that is not a problem for supply decoupling. The actual value of the capacitors does not matter that much.

The size of the capacitors 0402 or 0603 does not matter (much) either, use whatever you prefer. Either size will decouple the supply lines equally well.

You might also want to watch: Are bypass capacitors really needed?

\$\endgroup\$
2
  • 1
    \$\begingroup\$ About the capacitance point, I thought, that ceramic caps can lose a dramatic share of their capacitance over increasing voltage. To cite this article: "eramic capacitors are DC voltage sensitive, [...] And the ones with high capacity in a very small package are often the worst" \$\endgroup\$ Commented Aug 13, 2018 at 10:47
  • \$\begingroup\$ Sure, that's all true, but the actual value of the capacitors used for decoupling isn't that relevant. So a 1 uF becomes 0.5 uF, then use two or just get a 2.2 uF cap. What is relevant for decoupling is that there are some caps present to short spikes on the supplies. The actual value of the capacitors is not so relevant. The "golden rule" for supply decoupling is that there should be a cap close to each supply pin on all ICs. If you want add extra caps, feel free to do so. If you follow this rule then you did what is needed and your design should be OK. \$\endgroup\$ Commented Aug 13, 2018 at 10:57
1
\$\begingroup\$

Ceramic capacitors are fine in bypass applications, but you're right to worry about the voltage coefficient. Along with temperature and aging, it can affect the actual capacitance--in fact, larger capacitors are often at the low end of their target value even at nominal test conditions. It's an issue for all class II ceramics, and as package sizes grow smaller, it becomes more prevalent. App notes for bypass caps are typically fairly generic, so you don't need to be concerned about matching the value exactly, but for the larger caps (>1uf) you'll get better results from a larger package.

This is a reference I find indisposable. I don't work for MuRata, but it's a great tool, and most caps from other manufacturers (given the same size and specs) will behave similarly.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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