How to select decoupling capacitors

Thread Starter

sharath_412

Joined May 7, 2007
32
Hi Friends,
Can anyone help me out?

1. How a decoupling capacitor is selected? Is there any formula to calculate? I have seen some circuits have .1 uF, 1 uF in parallel, how are they selected?


thanks
sharath
 

bloguetronica

Joined Apr 27, 2007
1,541
Are you talking about decoupling capacitors in power supplies, that is, filter capacitors?

For that: C = I / (Vripple x 100)

But I am not sure if is that what you want, since there are all sorts of situations using decoupling capacitors. I only mentioned because it is a common application. It depends on the situation where do you want to use it. Please, be more specific. Post a circuit diagram, so we can help.
 

niftydog

Joined Jun 13, 2007
95
If I'm guessing right, the parallel 0.1µF and 1µF caps are often found decoupling logic ICs and such. The way I understand it, the small cap is better at filling in the very fast transients, while the larger one can deal with a more sustained voltage droop. They work in tandem to improve the effectiveness of decoupling.

With values like 0.1µF, 1µF and 10µF it's often a case that it doesn't really matter what value the caps are and it's just easy to pick round numbers.
 

Distort10n

Joined Dec 25, 2006
429
You can choose the value of a bypass capacitor based on the noise that you expect that could affect your system. So if you are worried about HF noise > 1 MHz then you pick a value of capactior that will short at frequencies equal or above that range. Of course, non-ideal caps will have ESR and parasitic inductances that you may or may not have to consider.

http://www.edn.com/index.asp?layout=article&articleid=CA454638

This is a good article about the subject.
 

Thread Starter

sharath_412

Joined May 7, 2007
32
I wanted to know about decoupling capacitors used in power supply section.
Also what are the sites where I get detailed information on decoupling capacitors.


Thanks
sharath
 

Skeebopstop

Joined Jan 9, 2009
358
Shouldn't we first clarify whether decoupling or bypass is wanted? All of the answers were directed towards bypass caps, where decoupling caps are generally for decouping AC from DC, or am I missing something here?

For decouping I would just say what my desired impedance is at a given frequency. So for audio, at 200Hz I would like my series input impedance to be 10, such that any next transistor stages are more than dominant with their input impedance being much higher. As such I would say 10 = 1/(2*pi*200*C) and solve for C. So in this case about 79uF should do the trick to let my signal through.

If the intention really was bypass, 1uF and 10nF work well together, 10uF and 100nF for higher power chips work well together. I would say low ESL/ESR caps are only needed if the chip is performing a higher power switching function, otherwise general purpose should be fine. Just don't go using tantalums for this as their ESL/ESR are quite poor, even in comparison to general purpose electrolytic etc..
 

KL7AJ

Joined Nov 4, 2008
2,229
Hi Sharath:

This whole matter is a bit of an art form...especially when working with mixed R.F. and digital signals. I wish there were a simple formula, but there isn't one. There are some principles though.

The more capacitance the better.... HOWEVER.....

LARGE capacitors also have large parasitic inductance...because of the way they're constructed....electrolyte foil wrapped in a coil...a nice inductor!

This inductance is enough to make the capacitor ineffective at high frequencies. So, putting a smaller capacitor in parallel...or even several values in parallel...allows you to bypass very short transients with a low-inductance capacitor, and lower frequency garbage with capacitors where the stray inductance doesn't matter too much.

At VHF/UHF, it's standard practice to use the lead inductance and the "intentional" capacitance to form a SERIES resonant circuit, which is a very effective bypass. However, generally speaking, only OLD GEEZERS like me have much experience actually making this work! When we're all dead, I don't know WHO's going to design R.F. circuits!

:D:eek:

Eric
 
Top