RE: [SI-LIST] : Bead-choking the power supply--everywhere!

About this list Date view Thread view Subject view Author view

From: S. Weir (weirsp@atdial.net)
Date: Sun Sep 17 2000 - 00:50:58 PDT


Marque,

At 10:02 AM 9/17/00 -0700, you wrote:
> >
> > Subject: RE: [SI-LIST] : Bead-choking the power supply--everywhere!
> >
> > I'll 'see' Mike - overkill, and raise him one. Just use the ferrites. The
> > caps bypass the noise into the ground planes making them noisy, polluting
> > everything else... <grin> and besides, you can't get enough caps anyway.
> >

I think you missed the <grin>. I am quite certain that reply was
tongue-in-cheek.

>Maybe someone could correct me, but I thought that if power and ground planes
>were used, the idea was to maximise the capacitance between the planes to get
>a good decoupling supply and to minimise lead inductance from the IC package
>to the planes. Won't adding ferrites between the IC packages and the
>planes just
>act to increase inductance between the package and the supply leading to
>nasties
>within the package itself (ie exhibit symptoms of the supply to the
>package itself
>dipping when switching events happen)
Very much so. It has been beaten to death on this list many times.

>I could be wrong as Im fairly new to this game, but I thought that from
>the work
>that was done at Sun and by others showed that boards with contiguous planes
>didn't actually need the traditional one cap per IC, but rather needed a
>strategy
>that was applied to the board as a whole for decoupling purposes.

Yes, carpet bombing with caps is a rule of thumb, not the result of
specific science.

>And that the
>location of the bypassing caps on the board was not important.

Not entirely. Electrolytics, and tantalums have sufficiently low SRF's
that propagation is not an issue for any practical board size, although DC
resistance of the planes may come into play for large boards, or
backplanes. Placement of the higher freq. ceramics does make a difference,
particularly on large boards. However the types of dimensions that we
would worry about are in the 2 inch radius range, so there is considerably
leeway.

>The power system needs to be adequately bypassed at all frequencies up to
>(and past)
>the fastest switching edge on the board. This would mean that the decoupling
>system as a WHOLE as frequency increases must remain capacitive and not become
>inductive as various caps go through resonance.

The power system needs to have a sufficiently low impedance at all
frequency components of interest. It is the magnitude of the impedance that
matters, not the sign. At the high-end, component physics, mounting
limitations, and board dimensions conspire against the shunt impedance
remaining capacitive.

The effects of parallel resonance are to raise the impedance, and must be
managed.

Regards,

Steve.

>Could people comment - is my assumption correct?
>
>Marque Crozman
>Innotech Control Systems
>Brisbane, Australia
>(swamped by the olympics)
>
>
>**** To unsubscribe from si-list or si-list-digest: send e-mail to
>majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
>si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
>si-list archives are accessible at http://www.qsl.net/wb6tpu
>****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:29:30 PDT