0
\$\begingroup\$

I have noticed there are different sized(SMT foot print size) ceramics capacitor with same voltage and capacitance ratings. I have been told that smaller the footprint, smaller the parasitic inductance and parasitic resistance a capacitor would have.. does this means I should select the smallest possible capacitor available? I'm gonna use this capacitors as bypass capacitors for MCUs. Available foot print sizes {C0402, C0603, C0805, C1206, C1812} capacitance needed - 100nf, 27pf.

and also, should I use a 1uf ceramic cap, parallel to 100nf bypass capacitors, just in case?

\$\endgroup\$

3 Answers 3

2
\$\begingroup\$

The difference in inductance between a 0402 and 0603 is miniscule. Your layout is probably contributing an order or magnitude more inductance than the difference between 0402 and 0603

If you are comfortable with rework of a 0402, then by all means go for it, but there are no technical reason to do so.

A slight note is that 0402 is more likley to tombstone during assembly, but as long as you use a PCBA manufacturer that is not really an issue anymore. (Assuming you don't use a cheap backalley manufacturer with outdated equipment)

\$\endgroup\$
2
\$\begingroup\$

The parasitic inductance will probably not be an issue for you for simple micro controller decoupling.

Choosing a package size depends on several factors: how you will be assembling the boards, how many and what other components are you using.

I typically use 0603 as they are widely available and easily soldered by and or on an assembly line.

\$\endgroup\$
4
  • \$\begingroup\$ so if I got options, I'm better off with a 0402? \$\endgroup\$ Commented Mar 27, 2021 at 7:35
  • \$\begingroup\$ @JohnCortex Are you able to solder something as small as a 0402? Also note, that the TENDENCY (not the rule) is for parasitic inductance to decrease with package size, the tendency is also that capacitance characteristics degrade the more capacitance you stuff into a smaller package (i.e. DC bias). \$\endgroup\$
    – DKNguyen
    Commented Mar 27, 2021 at 8:51
  • \$\begingroup\$ @DKNguyen I'm gonna use a SMT assembly service and they support 0402. And yes, I have heard of such issue you're mentioning, Capacitance tend to degrade drastically in such small package sized caps, but that's pretty much all I know. if you do know more, could you write an answer or share a link? I couldn't find reliable sources on that matter, only empirical evidences and small talks in some forums. \$\endgroup\$ Commented Mar 27, 2021 at 10:05
  • \$\begingroup\$ @JohnCortex You're pretty much going to have to go on empirical data anyways. Like looking at capacitor graphed data which isn't provide by all manufacturers. That's why I use Murata. They have graphs for all their capacitors. Go compare at and stare at a bunch of graphs for two or three hours from their website for impedance vs frequency and capacitance vs DC bias voltage all versus package size. \$\endgroup\$
    – DKNguyen
    Commented Mar 27, 2021 at 20:46
0
\$\begingroup\$

Footprint selection: Basically yes, choose the smallest but also consider cost and manufacturability.

Bigger footprints with the same rating also are more stable under DC bias usually.

Which caps to choose for decoupling is an involved question. The datasheet recommendations should work usually. If there is no additional 1uF in there, you shouldn't need it. I am not a friend of stacking different values of ceramic caps for decoupling. If both values are low ESR types, you can get antiresonances with strongly worsened power supply impedance even at low frequencies between 5-20 MHz that are typical for MCU. Instead I usually trust one 0.1 uF - 1.0 uF for low powered things for the intermediate frequency power impedance (up to ~30 MHz) and rely on the plane capacitance between power and ground for the range beyond.

\$\endgroup\$
1
  • \$\begingroup\$ Yes, it was a typo, let me correct it. \$\endgroup\$ Commented Mar 27, 2021 at 7:18

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