Hi,
It is common to use a bypass capacitor connected in parallel from VCC
to GND to reduce noise on the power line. I have seen also inductors
connected in series to do the same thing. I have seen also both (cap &
inductor) being used at the same circuit to reduce noise on the power
line. In my opinion, the caps are smaller and more cheaper than
inductors, so why use an inductor?
Keep in mind that the voltage across a capacitor has to
change, in order for it to deliver current to a load. In
the case of a bypass, this voltage can be reduced to an
arbitrarily low value by making the capacitor arbitrarily
large. But for any practical capacitor, the voltage never
gets all the way to zero change.
Let me give you an example of where a combination of
capacitor and inductor worked well. I have a magnetometer
that requires a fairly large pulse of current (80 mA peak)
to activate it, 65000 times a second. I want to phase lock
it to a crystal oscillator that runs on the same supply.
Any supply bounce caused by the magnetometer pulse tends to
influence the crystal oscillator, so that it is not an
independent frequency reference.
No matter how large a capacitor I tried as bypass at the
magnetometer, I still saw influences at the oscillator when
they got near the same frequency. Very large capacitors had
increasing series inductance that kept them from ideally
lowering the bounce at the edges of the pulse. But once I
got up to a 10 uF ceramic at the magnetometer, even if the
supply line was long and inductive the +5 supply, the
magnetometer had a stable enough supply to work fine. So I
picked that value of bypass but added a small inductor in
series with the +5 line to that bypass. There is a
measurable but harmless bounce at the magnetometer, but the
inductor isolates that small bounce from a separate bypass
capacitor at the crystal oscillator, so that it now operates
essentially independently from the magnetometer. In effect,
the inductor in the +5 line forces all the voltage noise
generated by the magnetometer pulse to occur on the +5 line
(gives that line elasticity), and the ground referenced
output has a clean ground line to be compared against (looks
more rigid by comparison), while the rest of the +5 system
is nice and quiet.
