Re: [SI-LIST] : 20-H Rule for Power Planes

Vinu Arumugham (vinu@cisco.com)
Sun, 30 May 1999 15:46:51 -0700

Grading the dielectric constant is one possibility. The other is to grade the dielectric thickness. I think a 20-H rule implementation achieves a gradual increase in the effective dielectric thickness at the edge of the board.

Vinu

S. Weir wrote:

> I don't think so. The geometry of the planes themselves still has the
> abrupt edge. All we have done is to move the center point of the fringing
> pattern so that the field gradients across any susceptor antennae, (
> traces, metalwork ) near the board edge is less. To do what you suggest, I
> believe that we would need a material where we could grade the dielectric
> constant. Does anyone feel differently?
>
> Steve
> At 11:23 PM 5/29/99 -0700, you wrote:
> >Can the implementation of the 20-H rule not be viewed as replacing an abrupt
> >impedance discontinuity at the edge of the board, with a gradual
> >discontinuity? If so, it will have the effect of "smearing" the resonant
> >peaks and reduce
> >the radiated power at any given frequency.
> >
> >Vinu
> >
> >Michael E Vrbanac wrote:
> >
> >> I second all that, John and Todd.
> >>
> >> I think you have said it much better than I!
> >>
> >> The radiation problem seems to require two necessary conditions:
> >> 1. a fringing field that is set up between two planes
> >> 2. a conductor placed in the "near field" to that fringing field
> >> which has access to a susceptible circuit or can carry the energy
> >> outside the system to "radiate".
> >>
> >> Without those two things together.... "nothing happens" (and could
> >> be a valid reason why some folks haven't seen it before).
> >>
> >> As Todd noted (and others and I agree with), in the "two plane model",
> >> the 20H rule does not stop the "power" plane's ability to radiate.
> >>
> >> Additional "ground" planes provide better field capture to the problem
> >> presented by the "two plane model" (when the "power" plane is between
> >> the "ground planes" and inset by 20H the distance to the nearest
> >> plane). This reduces the probability for significant near-field
> >> coupling to ANY nearby conductors. This is easily implemented in
> >> multi-layer stackups.
> >>
> >> As far as the "un-balanced" question somebody raised earlier, I
> >> haven't thought about that much but perhaps the additional ground
> >> planes would help "rebalance" the things compared to the two plane
> >> model.... any thoughts?
> >>
> >> To me, the 20H rule is a "field manipulation technique", not a "source
> >> suppresion technique" .... as the field not particularly "reduced" but
> >> more or less moved or reoriented. I usually do the source suppression
> >> stuff first to try to eliminate the problem but I won't overlook an
> >> opportunity to use field manipulation if I need to.
> >>
> >> Michael E. Vrbanac
> >>
> >> >
> >> > Thanks Todd,
> >> >
> >> > I think you may have put your finger on the source of the confusion here.
> >> > There was a lot of talk of reducing emissions by shrinking the power
> planes
> >> > slightly. That may have created the impression that emissions directly
> from
> >> > the power planes were reduced rather than reducing the power plane
> coupling
> >> > to something else that was radiating. In the products where I have had
> >> > problems that caused me to cut back the power planes it was always
> coupling
> >> > to a trace near the edge of the power plane, or the power plane at the
> >> > board edge coupling to part of the chassis or an air vent. These cases
> >> > always involved small distances (fractions of inches).
> >> >
> >> > John Lockwood
> >> > Juniper Networks
> >> >
> >> > At 09:09 AM 5/28/99 -0700, you wrote:
> >> > >Wow! What an interesting discussion! Since we have recently been
> >> > >investigating this issue, I can't resist adding my own 2 cents worth.
> >> > >
> >> > >First of all, the 20-H rule was developed years ago, before radiation
> >> > >directly from the power planes was a common problem. As a couple of
> >> > >people
> >> > >have pointed out, pulling the power plane in away from the edge of the
> >> > >board
> >> > >reduces near-field coupling to other boards, cables, or the enclosure.
> >> > >This
> >> > >can be a very good thing, because it keeps energy from coupling to the
> >> > >things that may be good antennas.
> >> > >
> >> > >However, in a board with only 1 return plane, pulling the power plane in
> >> > >away from the edge of the board does not reduce the power bus
> >> > >structure's
> >> > >ability to radiate. In fact, slightly more power can be radiated when
> >> > >the
> >> > >power and ground planes are not of equal size. (I liked Larry Smith's
> >> > >intuitive remarks regarding the loss of balance.)
> >> > >
> >> > >Placing a ring of return trace around the perimeter of a board and
> >> > >stitching
> >> > >it to the return plane also does not reduce radiation directly from the
> >> > >power bus. This is something we have experimented with in our lab. The
> >> > >gap
> >> > >between the power plane and the return ring becomes the new "edge" and
> >> > >radiates just as effectively.
> >> > >
> >> > >I am not saying the 20-H rule is not a good idea. It can be very
> >> > >effective
> >> > >at eliminating EMI problems resulting from near field coupling off the
> >> > >edge
> >> > >of the board. However, it does not generally reduce EMI at power bus
> >> > >resonant frequencies by making the power bus a less efficient radiator.
> >> > >
> >> > >Todd Hubing
> >> > >University of Missouri-Rolla
> >> > >
> >> > >**** To unsubscribe from si-list: send e-mail to
> >> > majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> >> > si-list, for more help, put HELP. si-list archives are accessible at
> >> > http://www.qsl.net/wb6tpu/si-list ****
> >> > >
> >> > >
> >> > >
> >> >
> >> > **** To unsubscribe from si-list: send e-mail to
> >majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> >si-list, for more help, put HELP. si-list archives are accessible at
> >http://www.qsl.net/wb6tpu/si-list ****
> >> >
> >> >
> >>
> >> **** To unsubscribe from si-list: send e-mail to
> >majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> >si-list, for more help, put HELP. si-list archives are accessible at
> >http://www.qsl.net/wb6tpu/si-list ****
> >
> >
> >
> >
> >
> >**** To unsubscribe from si-list: send e-mail to
> >majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
> >si-list, for more help, put HELP. si-list archives are accessible at
> >http://www.qsl.net/wb6tpu/si-list ****
> >
>
> **** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP. si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list ****

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