Hi - thanks for the description.  This is very interesting.  Do you put=20
power and ground parallel to each other on the same pcb side, or have=20
them follow each other on opposite sides of the pcb?  It sounds like=20
they run parallel with your description of connecting the decouple cap=20
then the chip.

Do you have a pic showing this?  I want to make sure I understand the=20
"T" vs. the "V".

I'm doing a bluetooth LE design in the coming weeks, and if I can make=20
it on a 2-layer (vs. 4-layer reference design) it would be a big win.  I=20
realize there's a lot more to successful RF than power routing and=20
decoupling caps, but it would certainly help.

Thanks!
J


David VanHorn wrote:
> I use a 3d connexion space pilot with Altium.  Many buttons, and the
> knob thing is really nice for 3d views.
>
>
> Routing:
> Place your unmoveable components, then the others, high priority to
> bypassing and high frequency paths.
>
> Place crystals and their caps as close as possible to the chip, and
> return the crystal caps directly, and only to the nearest ground pin
> on the chip that is using the crystal.  From that point flood an
> isolated island on the other side of the board.
>
> Select bypass caps according to the frequencies of interest, not 0.1uF fo=
r all.
>
> I diverge from the conventional wisdom in that I don't use planes to
> return power, I route explicit paths that mirror as closely as
> possible the sourcing path.  The power and ground route to the bypass
> cap, and then to the chip. Never "T" into a bypass, rather make it a
> "V" with the cap at the point of the "V". FORCE the current to travel
> to the cap, don't give it any other options.
>
> The result of all this is a board that is two layer only, spends
> usually $0.00 on emi supression, and passes radiated and conducted
> emissions by being so quiet that it is almost completely unobservable
> in a GTEM.
>
> RF immunity on our products only needs to be 3V/M, but the last one I
> tested in extremis was running as a bare board (no enclosure of any
> kind) in the GTEM at a field>  192V/M.  The field measurement
> instrument started giving erratic readings above that point, so we
> couldn't trust it for higher readings.  We were putting>50W into the
> chamber.
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.