> Anyway, my point was that bypass caps are primarily for holding the power
> voltage at the part, not for protecting the rest of the power net from hi=
gh
> frequency glitches caused by the part (they have that effect, but that is
> not their primary purpose).
>=20
>=20

I would say that their primary purpose is to operate as a reservoir for sho=
rt=20
term current surges in the parts they are connected to. If they are connect=
ed=20
in the correct way (as close to the part as possible with short, low impeda=
nce=20
tracks (thick)) and if they have low ESR for high frequencies they will hav=
e=20
the following effects:

* The power voltage of the part is kept at a steady, stable level.

* The current loop caused by the surge is kept very close to the IC responi=
sble=20
for the surge and by doing so the voltage on the rest of the power net is a=
lso=20
kept at a steady, stable level.=20

* Since the current loop is kept very small, the resulting conducted and=20
radiated EMI can also be kept at a very low level (if rating these effects,=
 I=20
would say that this is the primary one).

When laying out bypas caps for an IC, I try to visualize how the current su=
rges=20
from the IC will flow in the circuit and keep those currents as close to th=
e IC=20
as possible. This always includes a very good ground plane and perhaps more=
=20
than one via from the negative side of the cap to that ground plane. While=
=20
keeping the ground plane as big and homogeneous as possible, I also try to=
=20
connect the IC and the cap to the power net in a way that the current for t=
he=20
surge will most likely be drawn from the cap than from the rest of the net=
=20
(impedance much lower to the cap than the rest of the power net). This can =
mean=20
that the positive side of the cap will be connected to the power net with a=
 Y=20
(or V) like connection instead of a T with the traces between the IC and th=
e=20
cap much thicker than the traces for the rest of the power net.

Another thing to keep in mind here is that the frequency domain in the curr=
ent=20
surges from digital circuits (especially high frequency microprocessors whe=
re=20
thousands of transistors switch states in each clock cycle) can have very h=
igh=20
frequency components which is why it is important to keep the impedance bet=
ween=20
the IC and the cap very low at those high frequencies.

I have been doing circuit boards for low to medium frequncy (up to around=20
50MHz) microprocessors for a very long time and following those simple rule=
s=20
mentioned above will almost always produce boards with very low emitted EMI=
=20
which is very important since all board that will be used and sold=20
professionally will have to be compliant against conducted and radiated EMI=
=20
standards. Since measuring the radiated EMI is one of the most expensive EM=
C=20
tests, it is good to keep the test time as short as possible.

/Ruben
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D
Ruben J=F6nsson
AB Liros Electronic
Box 9124, 200 39 Malm=F6, Sweden
TEL INT +46 40142078
FAX INT +46 40947388
ruben@pp.sbbs.se
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.