[sdiy] Decoupling caps x3?

[Harry Bissell Jr]

The idea is to keep the impedance as low as possible
ober the widest range.

The 1uF is for decoupling down to near DC... self
inductance of the part will limit the high frequency
response...

the .1 is too small for DC (usually)... and eventually
it will have a high frequency limit as well.

The .01 will be good only at high frequencies.

This distributed capacitor array will likely damp a
single self-resonance that would occure from stray
inductance, a single cap... and high Q values
(low ESR in the caps).

The best idea is to decouple with the absolute minimum
loop area... and to use capacitors with the lowest
possible self-inductance... namely monoltyhic ceramic
SMT caps for the highest frequencies.

Ken Stone did some interesting things in the
"MorphLag"
layout he did from my design. He put the decoupling
caps on the back of the board and used SMT... it is a
very effective design.

If you use the Bissell / Patchell method of .1uF
decoupling caps, on every power pin of every IC you
can probably get away with just using .1uF caps (at
audio frequencies) there will be enough total
capacitance you will not miss the 1uF caps.  You may
also need to soft-start your designs  (we get into
several hundred
decoupling caps !!!)

H^) harry

--- Seb Francis <seb at burnit.co.uk> wrote:

> Hi,
> 
> I'm noticing on quite a lot of datasheets for mixed
> signal devices 
> (e.g.CODECs, PICs) there is recommended 3
> simultaneous values of 
> decoupling caps: 1uF, 0.1uF, 0.01uF.
> 
> I've always just used a combination of a few uF
> electrolytic here and 
> there and plenty of 0.1uF multilayer ceramic
> (usually X7R) very near 
> power pins and voltage references.
> 
> I thought 0.1uF multilayer ceramic were good enough
> for damping high 
> frequency stuff as they have very low ESR and
> inductance.  Is there 
> really an advantage to adding some 0.01uF caps as
> well?  And if so what 
> type to use .. the datasheets don't specify - more
> ceramics, or maybe 
> polyester?
> 
> Seb
> 
> P.S. This is a very good article on the general
> subject of mixed signal 
> decoupling and grounding ...
>
http://www.analog.com/en/content/0,2886,760%255F788%255F97529,00.html
> (but it didn't answer my question above...)
> 
> 
> 
>