decoupling capacitance, series impedance !!!

[taofeng]

Normally the larger the better, but actually I think decoupling capacitor should also work with the series impedance on the power supply, otherwise it makes no sense for EMI or EME. Is that right ?

Can anyone educate me on this ?


jef

 

[karesz]

Hi jef,
So is that! It makes no sens to EMi if you have enough low parasitics as ESR & ESL, but in the full badwidth(...
This is the reason for "splited capacitances", exactly you must have such high cap (electrolythic as 100/1000uF), depend of your consumed current volumen, 100/or 10 nF pro CMOS or high-speed analog IC & if you have such circuits for paar hundred MHz-or higher- Frequenciec; of course a 100pF , alls parallel , tight to your IC/circuit...
In all case dont forget a GND-plane and if possible power plans too, than the good selected capacitors are only efficient in full bamndwidth with a very low impedance supply system. All in full bandwith of Dc upto x hundred MHz or maybe a couple of GHz...
K....

 

[dick_freebird]

Larger capacitors tend to have larger parasitic inductances
and resistances than smaller sizes, let alone the benefits of
paralleling.

The DC series impedance of the supply is not really the issue,
it's the effective impedance of the noise sources (which are
not always inside the supply, such as self-generated switching
noise which is on the wrong side of PCB trace inductance,
from the power source).

Your bulk capacitance is for the ripple baseband and its first few harmonics.

Your distributed capacitance is for AC-shorting the local
current loops as close-in as possible.

 

[karesz]

Hi,
I belief to sayd the same-maximum with such other words, but you are welcome
K.