[sdiy] wall warts--wishful thinking

[Tim Parkhurst]

On 8/16/05, John Blacet <blacet at blacet.com> wrote:
> I did some very basic external filtering work with a small switcher and
> got nowhere! Obviously, this not trivial.
> 
> If anyone comes up with a decent filter or some technical info of
> interest, let us know.
> 
> It would be great to get rid of heavy transformers and get more
> efficient energy wise.
> 
> --
> Regards,
> John Blacet
> 
> Blacet Research
> http://www.blacet.com
> 

Hey John,

Have you tried the CLC "Pi filter" suggested by FMC (Fabio)? It was
suggested in the "switched mode PSU" thread, and looks to be
promising. I've included an excerpt below.


Tim (Pi R round, the last time I looked) Servo
-- 
"Imagination is more important than knowledge." - Albert Einstein

*************************
Old-fashion pi filters (C-L-C) can do miracles... don't know why they
aren't used anymore. (cost? space?..) Try a dual pi scheme, ie:

_____C______   __L1__   _____C_____    __L2__   _____C______
100pF + 1nF ->  50uH -> 1nF + 0.1uF -> 150mH -> 0.1uF + 470uF

L1 can be done with some copper and one 1/2W resistor, L2 can be done
from the secondary of a dead wall-wart trafo (choose one with an open
primary, discard those whose primary is burnt (melt))

I've seen some designs using Rs instead Ls for the pi. This is bad.
The inductor on the pi keeps Rsource low while having a hi_z for the ac
content (including transients), something that doesn't happens with
resistors. A Pi(c-r-c) and a Pi(c-l-c) are two completely diferent
things.

Of course, a good emi shield is another completely diferent thing too.

As a side note: I always include L-C filtering in power rails of
every pcb I make. It takes a bit more time and space (just a bit)
but it extend noise isolation not only between the card and the PS
but among the cards themselves. (and it doesn't hurt)

Fabio.
***************