2
\$\begingroup\$

I'm trying to choose the right size of smd mlcc ceramic capacitor.

Two of the biggest values of my capacitors are, 4.7uF and 10uF which I first chose to be 0603 because my pcb space is very limited.

Those capacitors are output capacitor in power manager IC (battery charger with power path), and decoupling capacitors, for GPS module, MCU, etc.

The voltage on their terminals does not exceed beyond 4.2V in worst case.

I was wondering if there is any significant reason (Apart from cost) to choose larger footprints for an smd MLCC especially in my case.

EDIT: What is the main difference between two sizes considering same capacity, voltage and dielectric?

\$\endgroup\$
2
  • 1
    \$\begingroup\$ The dielectric type is a very important consideration. C0G and NP0 are the best. But only available in small capacitors (pF and maybe nF). For capacitors in the 4.7uF range I would normally use X5R dielectric. When you compare ceramic capacitors, you should be comparing different packages with the same dielectric and voltage rating. If you are deciding between an X5R 10uF in 0805 and an Y5V dielectric 10uF in 0603, it is very likely that the most important difference is the dielectric type. Y5V capacitors are weird, very non-linear devices with a very wide capacitance tolerance. \$\endgroup\$
    – user57037
    Commented Mar 25, 2018 at 16:30
  • 1
    \$\begingroup\$ You are right, I was referring between two capcitors with the same capacity and dielectric! \$\endgroup\$
    – MrBit
    Commented Mar 25, 2018 at 18:08

1 Answer 1

18
\$\begingroup\$

I was wondering if there is any significant reason (Apart from cost) to choose larger footprints for an smd MLCC especially in my case.

High-K ceramic dielectrics (X7R, Z5U etc) have voltage-dependent capacitance, ie the capacitance drops as voltage increases.

This does not actually depend on the voltage: it depends on the E field which is voltage divided by thickness of dielectric (V/m), and of course the dielectric type (those that cram the most capacitance into the tiniest space for the cheapest price tend to have worst capacitance variation).

A physically larger capacitor (more volume) will have thicker dielectric, thus lower E-field inside the dielectric, thus it will lose less capacitance as voltage is applied. You can find the info in the datasheets. This graph shows X7R vs X5R, the lower-grade dielectrics like Z5U are worse, so watch out.

enter image description here

See also this answer.

This will bite you if you forget to account for it, as your 10µF cap will actually have less µF when voltage is applied to it. If you designed for 10µF and the cap actually gives you 4µF then you may not meet your specs. But if you do account for it, then there is no problem.

Note that as capacitance loss depends on the E-field and thus the volume of the cap, two caps of the same footprint (like 0805) and same dielectric (like X7R) then the thicker cap should lose less capacitance at working voltage.

There could be an issue with ripple current if it's high enough (you don't want to overheat the capacitor with RI^2 losses). A large ceramic cap will allow a higher ripple current, due to having lower ESR and lower thermal resistance to the board. Ceramics are surprisingly good at this, ESR is very low, and a tiny cap can take 1 amp ripple without problems, but maybe not 10 amps due to losses being in I^2... good thing they're cheap and easy to parallel. Try googling "ceramic capacitor ripple current" for a bunch of useful links, like this one.

Inductance is also slightly higher for larger packages, but this also depends a lot on mounting inductance (ie, pcb, stackup and vias). In your case, you're not decoupling a huge CPU with 50 caps around, so layout is a lot more important than the difference between 0603 and 0805.

EDIT --

Peter says: I find Kemet K-Sim very useful: ksim.kemet.com

See also Murata SimSurfing

\$\endgroup\$
6
  • 3
    \$\begingroup\$ I find Kemet K-Sim very useful: ksim.kemet.com \$\endgroup\$ Commented Mar 25, 2018 at 13:15
  • \$\begingroup\$ @PeterSmith excellent! Murata has a similar tool -- simsurfing -- quite useful to be able to download spice models, esp. for ferrite beads \$\endgroup\$
    – bobflux
    Commented Mar 25, 2018 at 16:07
  • \$\begingroup\$ Thank you very much for your valuable post! Although, I made an edit in my question because my concerned was based on comparing only size (same capacity, dielectric, and rated voltage) \$\endgroup\$
    – MrBit
    Commented Mar 25, 2018 at 18:10
  • \$\begingroup\$ @MrBit I think this answer handles that... basically larger cap volume means less E-field and thus less capacitance loss (if the dielectric is the same). \$\endgroup\$
    – bobflux
    Commented Mar 25, 2018 at 18:22
  • \$\begingroup\$ @peufeu, I already replaced those capacitors with 0805 package because there's significant amount of ripple current and it'll be better with larger package. I'm struggling to make some space though... \$\endgroup\$
    – MrBit
    Commented Mar 25, 2018 at 18:29

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