RE: High speed capacitors

["Ingraham, Andrew" <Andrew.Ingraham@compaq.com>]

>  The
>parameter that effects this issue is inductance.  Certain capacitor
>materials/structures are highly inductive at high frequencies or fast
signal
>transitions.
 
Primarily it is the structure (physical shape of capacitor and whether it
has leads) that affects inductance, and that is what determines whether a
capacitor is good as a "high speed" capacitor.  The ceramic material has
little if any effect (except for how it indirectly affects the body
size/shape).  The internal structure can affect it as well, but most
ceramics use a similar internal (multi-layer) structure.

Capacitors with leads are generally poorer than SMT types.

>  An older
>measurement is dissipation factor which is given as a percentage (lower the
>better).
 
Dissipation and DF are caused by ESR, not inductance.

>    Still another way of specifying it is ESR (equivalent series
> resistance)
> which is usually applied to larger caps that are used as power filters in
> switching power supplies.
 
Again, this is resistance, not inductance.

I am told that ESR is rarely given for ceramics, because it is only "felt"
close to the resonant frequency where the reactance approaches zero.
Otherwise, reactance, whether capacitive or inductive, dominates over ESR
(for a ceramic).

While inductance is the big concern, in truth there is no such thing as "the
inductance of a capacitor."  Inductance requires a closed loop path.  Often
the total inductance (of a mounted capacitor) is dominated by the traces and
vias on your circuit board!  Short, wide SMT capacitors are better (the
'sideways' 0612 body is one of them), but you can ruin the low inductance
potential of a good capacitor by connecting it to traces that aren't kept
short.  Tangential vias, or vias-in-pads, are "better".  Long vias also can
add to the inductance and make bypass caps less effective at high
speeds/frequencies.

Regards,
Andy